FOOD AND DRUG ADMINISTRATION




















                          Tuesday, July 27, 2004


                                8:30 a.m.








                           ACS Conference Room

                             5630 Fisher Lane

                           Rockville, Maryland





       Otis W. Brawley, M.D., Acting Chair

       Johanna M. Clifford, M.S., RN, Executive Secretary




       Ronald M. Bukowski, M.D.

       Bruce D. Cheson, M.D.

       James H. Doroshow, M.D.

       Stephen L. George, Ph.D.

       Antonio J. Grillo-Lopez, M.D.

       Pamela J. Haylock, RN

       Maha H.A. Hussain, M.D.

       Alexandra M. Levine, M.D.

       Joanne E. Mortimer, M.D.

       Michael C. Perry, M.D.

       Maria Rodriguez, M.D.


       CONSULTANT (Voting)


       Ralph D'Agostino, Ph.D.




       Sheila Ross




       Martin Cohen, M.D.

       Richard Pazdur, M.D.

       Robert Temple, M.D.

       Yong-Cheng Wang, Ph.D.



                             C O N T E N T S


                                                           Page No.


       Call to Order and Opening Remarks                          5

       Introduction of Committee                                  5

       Otis Brawley, M.D.


       Conflict of Interest Statement                             7

       Johanna Clifford, M.D., RN


                     NDA 21-677, Alimta (pemetrexed)

                           Eli Lilly & Company



       Richard Pazdur, M.D.                                      10


       Sponsor Presentation


       Introduction and Objectives of the Presentation

       Paolo Paoletti, M.D.                                      21


       Background on Non-Small Cell Lung Cancer

       Second-Line Treatment

       Frances Shepherd, M.D.                                    32


       Alimta Development

       Roy Herbst, M.D., Ph.D.                                   40


       Clinical Efficacy from the Pivotal Study JMEI

       Paul Bunn, M.D.                                           46


       Safety Profile from the Pivotal Study JMEI

       Richard Gralia, M.D.                                      64


       Overall Conclusions

       Paul Bunn, M.D.                                           78


       FDA Presentation


       Clinical Review

       Martin H. Cohen, M.D.                                     86

       Statistical Review

       Yong-Cheng Wang, Ph.D.                                   103



                       C O N T E N T S (Continued)


                                                           Page No.


       Open Public Hearing                                      112


       Questions from the Committee                             114


       ODAC Discussion                                          195



                          P R O C E E D I N G S


                    Call to Order and Opening Remarks


                 DR. BRAWLEY:  Good morning.  I am Otis


       Brawley.  I am a professor at Winship Cancer


       Institute of Emory University.  I will be the


       Acting Chair of the Oncologic Drugs Advisory


       Committee for the day.


                 I would like to welcome everyone here and


       like to start out by coming the meeting to order.


                 The first order of business will be to


       introduce the members of the committee and then we


       will have the conflict of interest statement read.


                 So, if we can start off to my left with


       Ms. Sheila Ross, if you would introduce yourself,


       and as members introduce themselves, if they could


       mention what institution they are from.


                        Introduction of Committee


                 MS. ROSS:  Thank you.  Good morning.  My


       name is Sheila Ross.  I am the Washington


       representative for the Alliance for Lung Cancer.  I


       am here as a patient advocate.  I am also a


       two-time survivor of non-small cell lung cancer.



                 DR. GRILLO-LOPEZ:  My name is Antonio


       Grillo-Lopez.  I am a hematologist/oncologist with


       the Neoplastic and Autoimmune Diseases Research




                 MS. HAYLOCK:  I am Pamela Haylock.  I am


       an oncology nurse and a doctoral student at the


       University of Texas Medical Branch in Galveston,


       and I am the consumer representative.


                 DR. D'AGOSTINO:  Ralph D'Agostino from


       Boston University, a biostatistician, consultant to


       the panel.


                 DR. GEORGE:  Stephen George, also in


       biostatistics, Duke University.


                 DR. LEVINE:  Alexandra Levine,


       hematology/oncology at University of Southern


       California in L.A.


                 DR. BUKOWSKI:  Ronald Bukowski, medical


       oncologist, The Cleveland Clinic.


                 DR. DOROSHOW:  Jim Doroshow, medical


       oncology, National Cancer Institute.


                 DR. RODRIGUEZ:  Maria Rodriguez,


       hematology/oncology at M.D. Anderson Cancer Center



       in Houston.


                 MS. CLIFFORD:  Johanna Clifford, Executive


       Secretary to this committee.


                 DR. HUSSAIN:  Maha Hussain, Professor of


       Medicine and Urology, University of Michigan.


                 DR. PERRY:  I am Michael Perry from the


       University of MIssouri, Ellis Fischel Cancer Center


       in Columbia, Missouri, hematology/oncology.


                 DR. CHESON:  Bruce Cheson,


       hematology/oncology, Georgetown University,


       Lombardi Comprehensive Cancer Center.


                 DR. WANG:  Yong-Cheng Wang, FDA,


       statistical reviewer.


                 DR. PAZDUR:  Richard Pazdur, Division


       Director, FDA.


                 DR. BRAWLEY:  Thank you.


                 If Ms. Clifford could read the conflict of


       interest statement.


                      Conflict of Interest Statement


                 MS. CLIFFORD:  Thank you.  The following


       announcement addresses the issue of conflict of


       interest and is made a part of the record to



       preclude even the appearance of such at this




                 Based on the submitted agenda and all


       financial interests reported by the committee


       participants, it has been determined that all


       interests in firms regulated by the Center for Drug


       Evaluation and Research present no potential for


       appearance of a conflict of interest with the


       following exceptions:


                 Dr. Ronald Bukowski has been granted a


       208(b)(3) waiver for consulting with a competitor


       on an unrelated matter.  He receives less than


       10,001 a year.


                 Dr. Maha Hussain has been granted waivers


       under 208(b)(3) and 21 USC 505(n) for owning stock


       in two competitors.  The stocks are valued from


       $25,001 to $50,000, and from $50,001 to $100,000.


                 Sheila Ross has been granted a waiver


       under 21 USC 505(n) for owning stock in a


       competitor, valued between $5,001 to $25,000.


       Because her stock interests falls below the de


       minimis exception allowed under 5 CFR



       2640.202(b)(2), a waiver under 18 USAC 208 is not




                 A copy of the waiver statements may be


       obtained by submitting a written request to the


       agency's Freedom of Information Office, Room 12A-30


       of the Parklawn Building.


                 We would also like to note that Dr.


       Antonio Grillo-Lopez is participating as the acting


       industry representative, acting on behalf of


       regulated industry.  Dr. Grillo-Lopez is employed


       by the Neoplastic and Autoimmune Disease Research




                 In the event that the discussions involve


       any other products or firms not already on the


       agenda for which an FDA participant has a financial


       interest, the participants are aware of the need to


       exclude themselves from such involvement and their


       exclusion will be noted for the record.


                 With respect to all other participants, we


       ask in the interest of fairness that they address


       any current or previous financial involvement with


       any firm whose products they may wish to comment





                 Thank you.


                 DR. BRAWLEY:  Thank you, Ms. Clifford.


                 The committee is gathered today to discuss


       the New Drug Application for Alimta or pemetrexed,


       an Eli Lilly compound proposed as a single agent


       treatment of patients with locally advanced or


       metastatic non-small cell lung cancer after prior




                 I would now like to introduce Dr. Richard


       Pazdur, Director of the Division of Oncology Drug


       Products, Center for Drug Evaluation & Research of


       the FDA to give us an introduction.


                     NDA 21-677, Alimta (pemetrexed)


                           Eli Lilly & Company




                           Richard Pazdur, M.D.


                 DR. PAZDUR:  Thank you, Otis.  It is a


       pleasure to be here, and I welcome the


       participants, the members of ODAC, as well as the


       audience to this most interesting ODAC





                 I have entitled my comments "Inferiorities


       of Non-Inferiority Trials."  I will just start off


       by saying I was listening to the Democratic


       Convention yesterday and Al Gore was talking about


       the 2000 election, and he said, "There are winners,


       there are losers, and then there is this third


       area," and it is kind of this third area, if I


       could take some statistical liberties that we are


       going to be talking about, and that is this whole


       area of non-inferiority, not positive, not


       negative, but some assumption of being equal.


                 I would like to preface today's


       presentation with a few comments really to focus


       your attention on key issues. This NDA highlights


       some unique challenges in developing oncology drugs


       regarding non-inferiority trial design and




                 Survival as an endpoint for regular


       approval has been a well-established endpoint for


       clinical benefit and regular approval.  In oncology


       trials, test drugs have generally demonstrated


       survival improvements compared to active controls.



                 Alternatively, an effect on the survival


       endpoint may be accomplished by demonstrating a


       non-inferior survival effect.  Non-inferiority


       ensures that a survival advantage, the so-called


       "control effect," would not be lost by a new agent.


       To determine the control effect, external


       historical information from multiple control trials


       is generally required.


                 A certain proportion of the control


       effect, known as the margin, should be preserved to


       demonstrate non-inferiority.  The active control in


       a non-inferiority trial should have an effect that


       is of substantial magnitude and that can be


       precisely estimated with estimates relevant to the




                 The ICHE9 guidance states that an


       acceptable active comparator "could be a widely


       used therapy whose efficacy in the relevant


       indication has been clearly established and


       quantified in well-designed and well-documented


       superiority trials"--and I emphasize the plurality


       of that word--"and which can be reliably expected



       to have similar efficacy in the contemplated active


       control trial."


                 The active control, therefore, should be


       preferably derived from multiple studies with a


       large consistent drug effect suitable for a


       convincing meta-analysis to be performed.


                 Constancy assumptions must be addressed in


       designing a non-inferiority trial, ensuring that


       the active control effect should be the same as in


       the historical controls.  These considerations


       ensure that the population enrolled in the


       historical trials is similar to the population in


       the proposed trial with respect to baseline


       characteristics, supportive care, additional


       available therapies, and observational frequencies.


                 The primary objective in the present


       Alimta trial was not achieved.  Neither superiority


       nor non-inferiority to docetaxel were adequately




                 The FDA believed that Alimta's


       non-inferiority for overall survival cannot be


       demonstrated for two reasons.  First, only a single



       small historical study exists to estimate the


       docetaxel treatment effect.  This study randomized


       a total of 104 patients, approximately 50 patients


       in each arm, to receive either docetaxel or best


       supportive care.


                 A second study was used in the docetaxel


       approval consideration.  This study compared


       docetaxel to either ifosfamide or vinorelbine.


       Neither agent had a demonstrated survival effect in


       this setting.


                 This second trial failed to demonstrate an


       overall survival benefit associated with docetaxel,


       however, there was an improvement in one-year


       survival.  Although sufficient data existed to


       approve docetaxel in this setting, the FDA believed


       that there is not a reliable and reproducible


       characterization of the docetaxel effect to use in


       a non-inferiority analysis.  Constancy assumptions


       cannot be verified and interstudy variability is




                 An additional concern is the existence of


       crossover in the present study.  Over 30 percent of



       patient randomized to receive Alimta subsequently


       received docetaxel at disease progression.


       Crossover obscures the differences between


       treatments, hence, in a superiority trial,


       crossover may lead to a false negative conclusion


       potentially denying an active drive a marketing




                 The use of a time to progression endpoint,


       an analysis occurring prior to crossover, may be


       preferred in settings where significant crossover


       is expected.


                 In contrast to superiority trials,


       crossover in non-inferiority trials may lead to a


       false positive conclusion.  This crossover


       confounds our interpretation of survival since the


       observed survival in both arms can theoretically be


       attributed to the control drug, in this case




                 Similarly, data integrity problems, known


       as trial sloppiness, either lack of attention to


       details in data collection or execution may obscure


       the observation of differences leading to false



       positive non-inferiority trials, hence, the agency


       has strongly recommended two trials to support a


       non-inferiority claim in an attempt to ascertain a


       true effect.


                 For regular approval of a drug, the


       sponsor must demonstrate that the drug is safe and


       effective in adequate and well-controlled trials.


       The effectiveness must be demonstrated on an


       endpoint that the agency believes to represent


       clinical benefit, usually survival, disease symptom


       amelioration or established surrogates for these.


                 The sponsor is not obligated to show that


       the drug is safer and/or more effective than an


       approved drug.  Many other therapeutic areas


       conduct placebo-controlled trials, drug A versus


       placebo, ensuring that superiority can be easily


       demonstrated even if a comparator drug is


       commercially available.


                 It is more difficult to demonstrate


       superiority in an active control trial, drug A


       versus drug B.  The test drug must possess the


       entire activity of the active control on the



       endpoint plus an incremental addition effect to


       demonstrate superiority.


                 The agency has frequently recommended


       add-on trials, A plus B versus B.  This design was


       used in the approval of Alimta plus cisplatinum in


       mesothelioma earlier this year.


                 In the add-on design, the test drug plus


       active control combination is compared to the


       active control alone or, alternatively, active


       control plus placebo.  This design ensures that all


       patients receive the active treatment, yet isolates


       the test drug's effect.


                 To demonstrate superiority, the test drug


       must only possess an incremental advantage over the


       active control on the primary endpoint rather than


       the control effect plus an increment.


                 We will be asking the committee to


       consider this application for accelerated approval.


       For accelerated approval, an improvement over


       available therapy must be demonstrated and may


       utilize a surrogate endpoint "reasonably likely to


       predict clinical benefit."



                 A more favorable safety profile could


       constitute a "improvement over available therapy."


       This decision requires considerable clinical


       judgment, and is not merely an exercise in adding


       up Grade 3 and 4 toxicities in two columns and


       declaring a winner.


                 The importance of a selected toxicity in


       patient management, toxicity duration, and


       overlapping toxicity, such as concomitant


       neutropenia plus diarrhea, concomitant neutropenia


       plus stomatitis may direct your clinical opinion.


                 With regards to surrogate endpoints for


       accelerated approval in this application, the


       agency has used response rates of similar magnitude


       and duration as demonstrated in this Alimta trial


       for past accelerated approvals in similar disease




                 In making a regulatory decision, we must


       consider all available data, a comprehensive drug


       evaluation including past approvals and single-arm


       studies.  As noted, Alimta  in combination with


       cisplatinum was approved for a mesothelioma



       indication earlier this year.  An improvement in


       overall survival advantage was demonstrated, the


       first for a drug in this disease.


                 In contrast to other accelerated approval


       applications that commonly use single-arm trials,


       the sponsor has provided a large randomized trial.


       Randomized trials always provide greater




                 We have comparative response rate data, we


       have comparative toxicity data, and we have the


       ability to examine time to event endpoints although


       we believe formal, non-inferiority analysis can


       neither be performed on TTP nor survival.


                 The sponsor is conducting large randomized


       trials in early lung cancer that can serve as


       confirmatory studies for clinical benefit if


       accelerated approval is granted.  The statistical


       analysis and the design of non-inferiority trials


       is an evolving field and represents considerable




                 Non-inferiority trials are difficulty.


       They take considerable resources in planning,



       designing, and executing trials and usually require


       considerable patient resources.


                 In conclusion, winning is always better


       than tieing.  The demonstration of superiority is


       always better than that of non-inferiority.


       Winning moves the field forward by identifying new


       agents and treatments.


                 However, a win may not only be an efficacy


       improvement, but may also be a safety improvement


       especially in a field such as oncology where


       toxicity concerns may dictate treatment choices or


       whether a patient even receives any therapy.


                 However, as we would like you to discuss


       later this morning, this regulatory decision must


       be carefully weighed against the clinical relevance


       of any potential survival loss.


                 I hope these comments will focus your


       attention and deliberations on the essential issues


       presented in this application.


                 Thank you.


                 DR. BRAWLEY:  Thank you, Dr. Pazdur.


                 Our sponsor presentation will now begin



       and last over the next hour.


                 If I can introduce Dr. Paolo Paoletti of


       Eli Lilly, who will give us the introduction


       objectives, and if you would present the presenters


       as we move along.


                 I should add that we are going to hold all


       questions until after the open public hearing.


                           Sponsor Presentation


             Introduction and Objectives of the Presentation


                           Paolo Paoletti, M.D.


                 DR. PAOLETTI:  Good morning.  My name is


       Paolo Paoletti.  I am the Vice President for Lilly


       Oncology Clinical Research and Oncology Products.


       I want to thank the FDA and the members of the


       Advisory Board for allowing Lilly to present the


       data on Alimta for the treatment of second-line


       non-small cell lung cancer.




                 Here is the agenda for the Lilly


       presentation.  I will give a short introduction on


       the objectives of the presentation, the historical


       context, and the rationale for the design of the



       pivotal registration trial.


                 Dr. Frances Shepherd, Professor of


       Medicine at the University of Toronto, and


       President of the International Association for the


       Study of Lung Cancer, and also principal


       investigator for the Phase III pivotal trial,


       Alimta versus docetaxel, will give the ground for


       the treatment of second-line non-small cell lung




                 Dr. Roy Herbst, the Chief of Thoracic


       Oncology at M.D. Anderson, University of Texas,


       will present the development of Alimta after the


       pivotal trial JMEI.


                 Dr. Paul Bunn, Director of the University


       of Colorado Cancer Center, past President of ASCO,


       and principal investigator for the Phase III trial


       Alimta versus docetaxel will present the efficacy


       result of the pivotal trial JMEI.


                 Dr. Richard Gralla, President of the


       Multinational Association of Supportive Care, will


       report the data on safety profile and patient


       reported outcomes for the same trial.



                 Finally, Dr. Bunn will give the






                 Additional experts from other


       international academic institutions are here today


       to answer your questions, and also experts from






                 In this slide, you can see the specific


       expertise are here to answer to your questions.




                 The objective of the presentation is to


       provide evidence that Alimta is effective and safe.


                 We intend to show that given the superior


       safety results, Alimta has a better risk-to-benefit


       profile than docetaxel and provides benefit to


       patients with non-small cell lung cancer.


                 This is supported by, first, Alimta is a


       novel and effective agent in non-small cell lung


       cancer.  Alimta has the same efficacy as docetaxel


       when looking at the variety of efficacy endpoint


       including survival, time to progressive disease,



       response rate in the entire population of patients.


                 In addition, this efficacy is consistently


       present when looking at the large number of


       subgroups.  Alimta is estimated to retain 102


       percent of docetaxel benefit over best supportive




                 Alimta is superior to historical best


       supportive care.  Alimta has an excellent safety


       profile and superior safety results when compared


       to docetaxel.  Therefore, Alimta offers an


       effective and safer second-line treatment option


       for patients with non-small cell lung cancer.




                 We propose the following indication.


       Alimta as a single agent is indicated for the


       treatment of patients with locally advanced or


       metastatic non-small cell lung cancer after prior


       chemotherapy, and at the dose of 500 mg/m2 i.v.


       with a 10-minute infusion at day 1 of each 21-day


       cycle, and to control toxicity, oral folic acid at


       the daily dose of 350-1,000 microgram and vitamin


       B12 at the dose of 1,000 microgram every 3 cycles



       given IM, dexamethasone 4 mg/bid on day minus 1,


       day of the treatment, and day plus 1.




                 In this slide, I summarize the historical


       context and the rationale for the statistical


       design when the pivotal trial JMEI was initiated.


                 Alimta showed consistent activity in


       non-small cell lung cancer in seven Phase II trials


       as a single agent or in combination with platinum


       agents both in first- and second-line.


                 This activity compares well with data from


       other commonly used regimens.  Folic acid and B12


       interventions significantly improve the safety


       profile of Alimta, however, the magnitude of this


       intervention was not completely known at the time


       of the initiation of the Phase III pivotal trial




                 It was decided to proceed with the Phase


       III trial in second-line to offer a better


       alternative treatment.




                 The trial, as Dr. Pazdur was saying,



       presented several design challenges and


       limitations, but we decided to run a head-to-head


       trial Alimta versus docetaxel.


                 We wanted to run a global clinical trial


       to support global registration.  Best supportive


       care in second-line treatment of non-small cell


       lung cancer was considered not practical because of


       the presence of the docetaxel as an approved agent


       in second-line treatment and not feasible in the


       United States and in many countries in Europe.


                 Combination chemotherapy was considered


       not appropriate especially in this second-line


       setting.  Docetaxel was approved in second-line


       non-small cell lung cancer primarily based on the


       result of the trial TAX 317B where superior


       survival over best supportive care was demonstrated


       in 55 patients treated at the dose of 75 mg/m2.


                 Survival was selected as the primary


       endpoint, however, we acknowledge the presence of


       limited historical data on the effect of docetaxel.


       Moreover, a pure equivalency trial would require


       more than 4,000 patients.





                 The JMEI is a global registration trial,


       and we discussed the statistical design with both


       FDA and the European Regulatory Agency.  Sample


       size of 520 patients allows for testing of


       superiority.  With the assumption of superiority,


       this sample would also allow for testing


       non-inferiority.  The hazard ratio was the basis to


       compare treatment arms for survival.


                 The protocol specified superiority


       testing, as well as testing 10 percent fixed margin


       for non-inferiority. This margin was agreed upon


       with the European Agency.  We always believe this


       was a very conservative matching.  Indeed, the


       magnitude of the effect of folic acid


       supplementation on toxicity was not known at the


       time.  Thus, safety advantages of Alimta were not


       considered in the definition of this match.


                 Before unblinding the data, we included


       the percent retention method for non-inferiority in


       the statistical analysis plan.  The FDA suggested


       for the evaluation of Alimta the retention of the



       effect of docetaxel, docetaxel versus best


       supportive care.


                 The FDA used this methodology to approve


       docetaxel in breast cancer and capecitabine in


       colon cancer.  Rothmann and co-authors published


       percent retention method in January 2003, and the


       details of the percent retention analysis were


       included in the statistical analysis plan before


       unblinding the data and before any analysis was






                 This slide shows the Alimta lung cancer


       submission timeline.  The first patient was


       enrolled on March 20, 2001. The last patient was


       enrolled on February 6, 2001.  The Final


       Statistical Analysis Plan was approved on January


       24, 2003.


                 Unblinding of the analysis and the data


       occurred on January 30, 2003.  U.S. fast track


       designation for second-line treatment of non-small


       cell lung cancer was granted on July 23, 2003.


       Non-small cell lung cancer submission was filed in



       November 4, 2003 in the U.S., and in July 2003 for




                 In June 22nd of this year, the European


       CHMP, the regulatory agency, gave a positive


       opinion for both second-line non-small cell lung


       cancer and mesothelioma.




                 Alimta has already shown to be an active


       agent in cancer.  In fact, Alimta, in combination


       with cisplatin, was approved on February 4, 2004


       for the treatment of mesothelioma in the United




                 This slide shows the survival curve.  You


       can see that the combination Alimta plus cisplatin


       has a median survival of 12.1 months, while


       cisplatin alone has a median survival of 9.3


       months.  The difference was statistically


       significant at P of 0.02.




                 Based on the evidence of the next


       presentation, we believe that Alimta merits the


       approval for the treatment of second-line non-small



       cell lung cancer for the following reasons.


                 Seven Alimta Phase II studies in


       first-and second-line non-small cell lung cancer


       show consistent evidence of activity within the


       range of activity of other agents currently




                 From this large Phase III randomized


       clinical trial in second-line non-small cell lung


       cancer, Alimta showed consistent similar clinical


       efficacy when compared to docetaxel in all primary


       and secondary endpoints and in all subgroup




                 Alimta is better than historical best


       supportive care.  Moreover, Alimta is significantly


       better for clinically relevant toxicity when


       compared to docetaxel.


                 Only docetaxel is approved for second-line


       treatment today, and there is a need for more


       second-line treatment option.




                 Alimta is an effective drug for the


       treatment of second-line non-small cell lung



       cancer, and it has a better risk-to-benefit profile


       when compared to docetaxel.


                 As you hear the rest of our presentation


       and that from the FDA today, please keep into


       consideration the following points:


                 Docetaxel at the dose of 75 mg has shown


       activity across several studies in second-line of


       non-small cell lung cancer after the pivotal trial


       TAX 317B, however, its use is limited by its


       toxicity.  The results in 288 patients receiving


       docetaxel in the JMEI pivotal trial confirms


       docetaxel's survival effect.


                 As I mentioned before, docetaxel was


       approved based on limited data, hence, the


       imprecision of the effect of docetaxel made


       non-inferiority design and related analyses very




                 This context, together with the lack of


       feasibility to conduct placebo-controlled trial


       once the drug is approved makes further advancement


       in drug development very difficult.


                 Although post-study treatment, inevitable



       in the United States, may confound survival result,


       the analysis from the pivotal trial JMEI suggest


       that such a confounding effect is unlikely.


                 In conclusion, I respectfully request that


       the members of this advisory board evaluate the


       data in second-line treatment of non-small cell


       lung cancer considering the overall efficacy and


       safety that will be presented.


                 Now, Dr. Frances Shepherd will give the


       background for the second-line treatment for


       non-small cell lung cancer.


                       Background on Non-Small Cell


                    Lung Cancer Second-Line Treatment


                        Frances A. Shepherd, M.D.


                 DR. SHEPHERD:  Thank you very much,


       members and guests.




                 In 1997, the ASCO Guidelines stated that


       "there is no current evidence that either confirms


       or refutes that 2nd-line chemotherapy improves


       survival in non-small cell lung cancer."


                 This conclusion was reached only seven



       years ago because, at that time, only single-arm,


       Phase II trials were available.  However, several


       trials of the third-generation agent docetaxel


       suggested that this agent might be appropriate to


       study further in randomized Phase III trials.




                 In the first trial initiated, the TAX 317


       study, patients previously treated with at least


       one platinum-based regimen were stratified based on


       their ECOG performance status, 0.1 versus 2, and on


       their best response to prior chemotherapy.


                 They were randomized to receive either


       docetaxel 100 mg/m2 or best supportive care.


       Routine safety monitoring revealed 5 or 10 percent


       early toxic deaths in the chemotherapy arm.


       Therefore, after discussion with the principal


       investigators and the FDA, the docetaxel dose was


       reduced to 75 mg/m2 for the second half of the




                 The sample size was maintained at 200


       patients as originally planned due to the


       difficulty in accruing patients to this study



       because of the best supportive care arm.




                 The overall response rates to docetaxel


       100 and 75 mg/m2 were both 6 percent.  Time to


       progressive disease was 2.8 months for patients


       treated with docetaxel 75 mg compared to only 1.6


       months for best supportive care.  Median survival


       was significantly longer for docetaxel 75 mg


       treated patients at 7.5 months compared to only 4.6


       months for best supportive care.  One-year survival


       was 3-fold higher for docetaxel patients.




                 Survival is shown graphically in this


       slide for the second half of the trial at docetaxel


       75 mg/m2, the FDA approved dose.  Survival was


       significantly longer for patients treated with


       docetaxel with a log-rank p-value of 0.01.


       One-year survival was significantly higher with a


       chi-square p-value of 0.003.




                 This is a very important slide to


       concentrate on. In this trial, patients must have



       received one platinum-containing regimen, but could


       have received more than one regimen before entering


       the trial.


                 As you can see from this slide, the


       numbers of patients unfortunately are small, and


       these must be considered exploratory subset


       analyses, however, they suggest that patients


       treated with docetaxel after two or more regimens


       derived absolutely no survival benefit from the


       treatment as compared to best supportive care




                 The entire survival benefit of the trial


       came from the administration of docetaxel in the


       true or strictly defined second-line setting.




                 The second large trial was the TAX 320


       trial and was performed in the United States where


       a best supportive care trial could not be




                 In this trial, patients were stratified by


       their best response to platinum-based therapy and


       performance status, and were randomized to receive



       docetaxel 100 mg/m2 or docetaxel 75 mg/m2, or a


       comparator of vinorelbine or ifosfamide.  This was


       largely vinorelbine.




                 The overall response rate was 11 percent


       for patients treated with docetaxel 100 mg, and 7


       percent for patients in the 75 mg group.  Both of


       these response rates were significantly higher than


       the 1 percent response rate noted in the control


       group with p-values of 0.001 and 0.036.


                 There was no difference in median or


       overall survival among the three treatment arms,


       however, the one-year survival rate of 32 percent


       for patients treated with docetaxel 75 mg, the


       FDA-approved dose, was significantly better than


       the 19 percent one-year survival rate of patients


       treated with vinorelbine or ifosfamide.  Chi square


       p-value for this is 0.05.




                 This is shown graphically on this slide


       where you will see the survival curve separating in


       the latter part.





                 The FDA-approved label for docetaxel 75 mg


       reports Grade 3/4 neutropenia of 65.3 percent,


       Grade 3/4 infection of 10.2 percent, and using a


       very stringent definition, febrile neutropenia rate


       of 6.3 percent.


                 Although Grade 3 and 4 diarrhea and


       neurotoxicity are rare at this dose of docetaxel,


       lesser grades of both of these toxicities may be


       distressing to patients.  Similarly, alopecia,


       although never life-threatening, may have a major


       negative emotional impact on both men and women.




                 Quality of life and symptom control was


       measured in both the TAX 317 and 320 trials.  Pain


       was significantly better controlled in the 317


       trial, and this was not because of increased opioid


       use.  You can see from this slide that opioid use


       was the same at study entry in both arms of the


       trial, however, significantly fewer patients


       treated with docetaxel required additional opioids


       and significantly fewer patients required the



       introduction of new opioids.




                 Weight loss was measured closely, and you


       will see that in the TAX 317B trial, 25 percent of


       patients treated with best supportive care had


       weight loss greater than 10 percent compared to


       only 2 percent of patients treated with docetaxel.


       Weight loss greater than 10 percent was seen in


       only 5 percent of patients treated with docetaxel


       75 mg/m2 in the 320 trial compared to 8 percent for


       vinorelbine patients.




                 Treatment was not at the expense of


       quality of life or performance status.  Indeed,


       performance status improved during the study for


       patients treated with docetaxel whether measured at


       initiation, across the cycles, or at the last






                 In summary, these landmark trials showed


       that second-line chemotherapy prolonged survival in


       non-small cell lung cancer.  It also improved



       symptom control and does not have a negative effect


       on quality of life or performance status.


                 These trials led to the approval of


       docetaxel 75 mg/m2 for the second-line treatment of


       non-small cell lung cancer in 1999.




                 In 2003, the revised ASCO evidence-based


       guidelines recommended docetaxel for patients with


       non-small cell lung cancer who have progressed on


       first-line platinum-based therapy.




                 In summary, the body of evidence shows


       that patients derive benefit from second-line


       treatment of non-small cell lung cancer with


       docetaxel.  However, better tolerated or more


       effective alternatives are needed.


                 Finally, docetaxel is being used more


       frequently in the first-line setting and no options


       are currently available for patients who are


       treated first-line with docetaxel-containing




                 As docetaxel is the only approved agent



       for the second-line treatment of non-small cell


       lung cancer, additional options are required.


                 Dr. Roy Herbst will now discuss the


       development of Alimta.


                            Alimta Development


                         Roy Herbst, M.D., Ph.D.


                 DR. HERBST:  Good morning, panel members


       and guests.  My name is Roy Herbst from the M.D.


       Anderson Cancer Center.  Our group and myself


       personally have worked with this drug both in the


       front and second-line setting in non-small cell


       lung cancer.




                 My purpose this morning is to provide some


       background information regarding this novel


       antifolate and to share supporting evidence that


       Alimta has activity in patients with non-small cell


       lung cancer, as well as providing clinical benefit.




                 First, a word about the structure.  As you


       can see, Alimta is very similar to folic acid, but


       really it is quite a unique and novel compound. 



       You can see two circles areas on the slide.  The


       N-10 nitrogen has been replaced by a methylene


       group, and most importantly, the pyrrolo-pyrimidine


       ring circled makes this structurally different from


       other antifolates.  That is important because it


       gives it some very unique qualities as I will talk


       about in the next slide.




                 Shown here is the mechanism of action of


       this drug, which is a multi-targeted antifolate.


       As shown in the left, the Alimta enters the cell by


       reduced folate carriers. Once inside the cell, it


       is polyglutamated.  This potentially allows it to


       be stored in the cell for higher intracellular




                 You can then see that it blocks three


       different enzymes involved in folate


       metabolism - TS, DHFR, and GARFT. There is also the


       potential for this drug to be active in MTAP [ph]


       efficient cells.  This makes it potentially more


       active, as well, at any cell that might upregulate


       any one of these different enzymes.





                 Activity has been across a wide spectrum


       of tumor models.  Today, we will focus on lung


       cancer.  Here, you can see four non-small cell lung


       cancer cell lines with activity in the nanomolar


       range.  There is also evidence here of a lung


       cancer xenograph, and you can see the drug is quite


       active, as well.




                 What about clinical experience?  First,


       the front-line experience.  Shown here are two


       studies that looked at Alimta before vitamin


       supplementation in patient with non-small cell lung


       cancer, compared to several studies with docetaxel


       also in the front-line setting.


                 The important thing to notice here is that


       the activity, both based on response rates in the


       20 percent range and the median survivals, from 7


       to 9 months, is quite consistent with what one


       would expect for docetaxel or, in fact, most of the


       third-generation chemotherapeutics that we now use


       for non-small cell lung cancer.





                 Activity has also been seen, and quite


       favorable toxicity, in combination with platinum,


       which, of course, is the way we treat lung cancer


       in the front-line setting.


                 Shown here are four studies, two using


       cisplatinum, two using carboplatinum, and again you


       can see in these Phase II studies, response rates


       that are quite similar to other agents in this


       setting, in one case in the 40 percent range,


       median survivals between 8 and 10 months, in fact,


       13.5 months in our M.D. Anderson study, and one-year


       survivals are quite good.  This drug clearly


       has activity with platinum in the front-line


       setting of lung cancer, as well.




                 Going into the randomized trial that you


       are about to hear about, this was the Phase II


       experience, the study from Smit and colleagues, 79


       patients.  This is a refractory group of patients


       with non-small cell lung cancer.  One hundred


       percent of these patients were refractory within



       three months, and importantly, it's an especially


       bad group because 66 percent were refractory within


       one month.


                 You can see that this drug demonstrates a


       clear response rate of 8.9 percent with a median


       survival of 5.7 months, and a one-year survival of


       23 percent.  There is clearly activity based on


       this trial in the second-line setting, and we will


       hear more about this, of course, today.




                 Now, what about safety?  As with most


       antifolates, the primary toxicity of this drug is


       hematologic.  Early data showed that high


       homocysteine levels, a surrogate for functional


       folate or B12 deficiency, correlated with high


       levels of toxicity.


                 So, a decision in development was made


       early on to supplement all patients with folic acid


       and vitamin B12 when they received this drug.  This


       resulted in decreased toxicity with no detrimental


       effect on efficacy.





                 I show one slide here.  This is basically


       showing a group of patients, 246, without vitamin


       B12 and folate supplementations, single agent


       administration, or 220, who did receive




                 Shown on the left are all the toxicities


       lumped together that I am going to show in this


       slide.  You can see in the white before, and in the


       green after, with a significant improvement.


                 Then, breaking that up into the top three,


       you can see Grade 4 neutropenia is significantly


       reduced, Grade 3/4 diarrhea also significantly


       reduced, and at least in this Phase II experience,


       you can see toxic death rate is zero, and then we


       are seeing when the supplementation was given.




                 So, in summary, Alimta has shown activity


       in non-small cell lung cancer as a single agent,


       both in the first- and second-line setting, in


       combination with platinum agents in the first line.


                 The safety has been well characterized.


       The toxicity is significantly reduced after adding



       folic acid and B12, and a very low incidence of


       neutropenia, febrile neutropenia, and other


       non-hematologic toxicities.


                 I can personally say, both for my group


       and myself, this has been our experience, as well.


                 Based on these results, a pivotal Phase


       III study in the treatment of second-line non-small


       cell lung cancer was indicated, and Dr. Paul Bunn


       will now present those data.


                 Thank you.


              Clinical Efficacy from the Pivotal Study JMEI


                             Paul Bunn, M.D.


                 DR. BUNN:  Good morning, Dr. Brawley, ODAC


       members, and guests.




                 As one of the principal investigators, I


       will review the results of the pivotal trial JMEI,


       which was a head-to-head comparison of Alimta to


       docetaxel in the second line treatment of patients


       with advanced non-small cell lung cancer.




                 I will begin this presentation with the



       study design and patient demographics.  The results


       of the primary endpoint survival will be given with


       a detailed discussion of the survival result and


       comparison with docetaxel and historical best


       supportive care with and without adjustment in a


       Cox model.


                 Following this discussion, I will review


       the results of the secondary efficacy endpoints and


       a brief discussion of the effect of third line


       therapy.  Because efficacy cannot be considered in


       the absence of toxicity, I will give a brief


       overview of toxicity, and then Dr. Gralla will


       review the safety results and patient reported


       outcomes in detail.  Then, I will wrap up with a


       few concluding remarks.




                 After stratification for known prognostic


       factors including performance status and stage, as


       well as other possible prognostic factors listed,


       patient were randomized to Alimta 500 mg/m2 I.V.


       day 1 every 21 days or docetaxel 75 mg/m2 day 1


       every 21 days.



                 The 283 patients randomized to Alimta


       received B12 and folic acid supplementation and


       dexamethasone was given to prevent skin rash.


                 The 288 patients randomized to receive


       docetaxel received dexamethasone according to the






                 The primary study endpoint was survival.


       This survival endpoint is expressed as a hazard


       ratio of Alimta to docetaxel with a 95 percent


       confidence interval.


                 Secondary endpoints included


       progression-free survival, time to tumor


       progression, response rate toxicity and patient


       reported outcomes as measured by the Lung Cancer


       Symptom Scale.




                 Of course, these endpoints were assessed


       in one of two populations, intention to treat and


       randomized and treated.  The primary endpoint


       survival, as well as all other time to event


       variables were assessed on an intent to treat



       population.  This population included all


       randomized patients regardless of therapy.


                 The toxicity endpoints were evaluated on


       randomized and treated population.  This group


       included randomized patients who received at least


       one dose of treatment.




                 Important inclusion and exclusion criteria


       included histologic or cytologic diagnosis of Stage


       III or IV non-small cell lung cancer.  All patients


       had progressed after at least one prior


       chemotherapy treatment, but not more than one prior


       chemotherapy treatment for metastatic disease.


       Prior adjuvant and neoadjuvant therapy was allowed.


                 Patients had performance status 0 to 2 and


       adequate organ function.  Active brain metastases,


       severe peripheral neuropathy or significant weight


       loss were not allowed.  Uncontrolled pleural


       effusions and prior docetaxel was not allowed.


       Prior paclitaxel was allowed and prior platinum was


       not required.





                 The most important prognostic variables,


       performance status and stage, were well balanced


       between the arms.  The less important variables,


       such as age and gender, there were minor but


       nonsignificant differences.  Histology and


       pre-treatment homocysteine levels were well






                 There were no differences in the fraction


       of patients responding to initial chemotherapy or


       the fraction with early relapse after prior




                 The two groups had no relevant differences


       in prior chemotherapy in terms of taxane or


       platinum exposure.




                 In both groups, dose intensity was well


       preserved with a similar number of patients


       receiving at least 4 cycles of therapy and a median


       of 4 cycles of therapy in both arms.  The percent


       of the planned dose intensity and dose delays were


       similar.  There was a significant increase in dose



       reductions in the docetaxel arm.




                 Of course, survival is so important in the


       primary endpoint and the unadjusted Kaplan-Meier


       survival curves for Alimta and docetaxel were


       overlapping and crossed several times.  The median


       survival times are 8.3 months and 7.9 months,


       favoring Alimta.


                 The one-year survival rates was 29.7


       percent in both arms.  The unadjusted hazard ratio


       was 0.99 in favor of Alimta.  The 95 percent


       confidence interval was 0.82 to 1.2.  This hazard


       ratio and confidence interval did not show


       superiority, nor rule out a 10 percent margin.




                 In order to more fully understand the


       survival implications of Alimta relative to both


       docetaxel and to best supportive care, the data


       must be put in the context of this and other




                 Percent retention analysis is a means of


       estimating the amount of benefit of docetaxel over



       best supportive care that is retained by Alimta.


       This analysis, as you have heard from Dr. Paoletti,


       was not in the original protocol, but was


       prespecified in the statistical analysis plan prior


       to unblinding and prior to data analysis.


                 The retention analysis was based on the


       results of TAX 317B, which was docetaxel 75 mg/m2


       versus best supportive care.  This analysis takes


       into account variability within the studies and


       allows for comparison of Alimta to best supportive




                 An important assumption of the percent


       retention analysis is comparability of populations


       and results between TAX 317 and JMEI.  This allows


       for the assumption that if the best supportive care


       arm were to be included in JMEI, its survival curve


       would have been similar to that seen in TAX 317.




                 For the most important prognostic factor,


       such as performance status and stage, populations


       in TAX 317 and JMEI were very similar.  There were


       less important factors, such as age and gender,



       there were minor differences, but overall, the


       pre-treatment characteristics  make the populations


       appear comparable.




                 Looking at the outcome of the 75 mg/m2


       docetaxel arms in both TAX 317B and in JMEI, shown


       here, shows the results are very similar.  The


       Kaplan-Meier survival estimate of docetaxel 75


       mg/m2  from TAX 317 is shown in green and JMEI in


       blue.  This outcome confirms the finding of TAX


       317B for docetaxel.




                 Once the populations are shown to be


       comparable, then the percent retention analysis


       allows for comparison of survival between TAX 317B


       and JMEI.


                 Superimposing the Alimta result of JMEI,


       which is the yellow curve I just added, it is


       evident the result is similar to docetaxel 75 mg/m2


       from both 317B, the prior study, and the current


       study JMEI.  This finding shows that Alimta is


       equivalent to docetaxel 75 mg/m2.



                 Now, adding in the best supportive care


       result, in white, strongly suggests the superiority


       of Alimta to best supportive care.  The hazard


       ratio of Alimta to best supportive care is 0.55


       with a 95 percent confidence interval that does not


       overlap 1, 0.33 to 0.9, the p-value is 0.019.




                 Another way to understand the survival


       results, which are real, and the confidence


       interval around these results is to compare hazard


       ratio and confidence interval in the trial results.


                 Shown in yellow are the actual study


       results showing an unadjusted 0.99 hazard ratio, a


       95 percent confidence interval of 0.82 to 1.2.  For


       reference, the percent retention of docetaxel's


       benefit over best supportive care is shown below


       the line.


                 For the actual data, the hazard rate of


       0.99 represents retention of 102 percent of


       docetaxel's benefit over best supportive care.  A


       hazard ratio of 0.82 represents 150 percent


       retention, and so forth.



                 If we want to determine whether Alimta has


       benefit over best supportive care, we can calculate


       the hazard ratio if the percent retention were


       zero, indicating that best supportive care and


       Alimta were the same.  In this case, the hazard


       ratio of Alimta to docetaxel would be 1.33.


                 Since the upper limit of the hazard ratio


       was 1.2, we can be quite confident that Alimta is


       better than best supportive care.


                 If we want to determine the hazard ratio


       if Alimta retained at least 50 percent of the


       benefit of docetaxel, the hazard ratio would need


       to be less than 1.21 for 95 percent confidence, and


       again this criteria was met.


                 If the upper limit of the 95 percent


       confidence interval was less than 1.11, then,


       Alimta would have been within 10 percent of


       docetaxel as originally requested by the European


       Regulatory Group.  As shown, it did not reach this


       value.  However, after reviewing the totality of


       the evidence, the European Authorities have


       recommended approval.



                 Alimta would have been declared superior


       to docetaxel if the upper limit of the 95 percent


       confidence interval had been less than 1.  The


       result did not reach this threshold of superiority.




                 Since not receiving therapy can affect


       non-inferiority analyses, the ICH Guidelines


       recommend that analyses of non-inferiority


       performed percent retention calculations on both an


       ITT, as well as the randomized and treated RT




                 This table shows the calculation from both


       populations.  For the ITT population, Alimta


       retained 52 to 150 percent with a p-value for 50


       percent retention of 0.047.


                 For the RT population, Alimta retained 58


       to 168 percent with a p-value for 50 percent


       retention of 0.036.


                 These data support retention of docetaxel


       survival benefit by Alimta.




                 As a prespecified secondary analysis, a



       Cox multivariate regression analysis was performed


       with these 7 prespecified prognostic factors in the


       model.  These factors included stage, performance


       status, time since last therapy, response to prior


       therapy, prior taxane, prior platinum, and number


       of prior chemotherapies.




                 The results from this model showed that


       three factors predictive for survival - Performance


       Status 2, time since last chemotherapy less than 3


       months, and Stage IV, all predictive for a worse


       survival outcome.




                 This slide shows the adjusted survival


       hazard ratio on a similar number line.  The actual


       data is represented in yellow.  The hazard ratio


       was 0.93 with a 95 percent confidence interval of


       0.76 to 1.13.  The p-value for the 10 percent fixed


       margin was p equals 0.051.


                 The difference between the upper limit of


       the confidence interval 1.13, and the prespecified


       10 percent fixed margin 1.11, translates into



       approximately 3.6 days difference.




                 This slide demonstrates subgroup analyses


       unadjusted and adjusted for known or potentially


       important prognostic factors for JMEI.


                 In most instances, relative subgroups,


       there were no appreciable treatment effect


       differences.  For Performance Status 2 patients,


       the hazard ratio favored Alimta, but in this case,


       the sample was small, and the result was not


       statistically significant.


                 For no prior platinum, the apparent


       differences in the adjusted hazard disappeared when


       imbalances important to other factors were taken


       into account.


                 These data provide confidence that the


       observed results were consistent across all


       subgroups and that the results could not be


       explained by a large benefit within any particular






                 We will now review the secondary endpoints



       of progression-free survival, time to progression,


       tumor response, and toxicity.  In addition, I will


       provide an exploratory data on possible confounding


       effect of post-study chemotherapy.




                 Shown is the Kaplan-Meier estimate for


       progression-free survival in intent to treat


       population.  It difficult to see that there is two


       curves here because they are so overlapping, but


       there are two distinct curves with a median


       progression-free survival of 2.9 months in both




                 The hazard ratio was 0.97, slightly


       favoring Alimta, with a 95 percent confidence


       interval of 0.82 to 1.16.




                 This is the Kaplan-Meier estimate of the


       time to tumor progression for JMEI.  Again, the


       curves overlap considerably with a median time to


       tumor progression of 3.4 and 3.5 months for Alimta


       and docetaxel respectively.


                 The hazard ratio was again 0.97, with



       confidence intervals of 0.8 to 1.17.




                 A review of chemotherapy given after this


       study showed that more patients on Alimta received


       any chemotherapy.  Not surprisingly, docetaxel,


       which is the only approved drug, was given more


       frequently after progression on Alimta despite the


       evidence you have heard that it provides no benefit


       in this setting.


                 Receipt of docetaxel does represent a


       crossover of sorts.  As expected, patients on


       docetaxel received more gemcitabine, more


       vinorelbine, and more gefitinib, as well as more


       other chemotherapy.


                 Of course, you will recall that gefitinib


       is the only agent for which there is any evidence


       for survival effect in third-line non-small cell


       lung cancer.




                 To further understand the post-study


       treatment effect, an analysis was performed to look


       at the type of post-study therapy and its potential



       effect on survival.  Of course, all of these are


       retrospective and are subject to great bias, which


       we can discuss later.


                 Patients who received post-study therapy


       lived longer than those who did not, not


       surprisingly, regardless of the nature of that


       therapy or the study arm.


                 Patients on Alimta who received


       post-therapy docetaxel did numerically worse than


       those who received other post-treatment study, such


       as gemcitabine or vinorelbine.


                 Patients on the docetaxel arm who received


       post-therapy docetaxel actually had numerically


       better survival than those receiving docetaxel


       after Alimta.  This post hoc analysis does not


       suggest any crossover effect or post-study effect


       of docetaxel treatment.




                 In fact, this slide shows the distribution


       of survival after progressive disease by treatment


       arm, Alimta versus docetaxel.  A higher proportion


       of patients on the Alimta arm received docetaxel,



       and a higher proportion of patients on docetaxel


       received other therapies.


                 The median survival was 4.5 months in both


       arm. This comparison suggests there is no


       difference between salvage therapies in the two




                 Assuming that patients with progressive


       disease have similar prognoses in the groups, this


       comparison implies the crossover to docetaxel in


       the Alimta arm did not affect any conclusion


       regarding survival.




                 Investigators determined the best response


       in the study according to South West Oncology Group


       criteria.  The response rate between the arms was


       virtually identical, 9.1 for Alimta and 8.8 for


       docetaxel, respectively.


                 Stable disease was seen in about 46


       percent of patients on each arm.  These data are


       consistent with the previously published data using


       both docetaxel and Alimta in this setting.


                 Because all efficacy parameters were



       equivalent, much of the clinical benefit of Alimta


       relates to toxicity, so the toxicity analysis, of


       course, becomes important.




                 This table provides a brief overview of


       significant toxicity differences regardless of


       causality. Alimta was associated with significantly


       less Grade 3/4 neutropenia, less febrile


       neutropenia, less infection with neutropenia, and


       less diarrhea.


                 There were also significantly less


       clinically relevant alopecia of all grades.


                 Alimta treatment was associated with


       significantly more ALT elevations, 2.6 percent


       versus 0.4 percent.




                 In conclusion, the results of JMEI


       demonstrate that Alimta afford efficacy benefits


       for patients with non-small cell lung cancer


       undergoing treatment after progression with prior




                 The survival result is similar to that of



       docetaxel with a hazard ratio of 0.99.  This hazard


       ratio translates into 102 percent retention of


       docetaxel's benefit over best supportive care.


                 The results are internally consistent


       across subgroups.  In JMEI, there is no evidence of


       an effective crossover or other post-study


       chemotherapy effect.


                 The survival results robustly support


       Alimta's superiority to historical best supportive




                 In addition to the survival endpoint, all


       secondary endpoints, including response, time to


       progression, progression-free survival affirm


       Alimta's activity and benefit to this group of




                 Finally, the safety profile of Alimta,


       which Dr. Gralla will review in detail, is clearly


       superior to docetaxel.


                 Now, I would like to invite Dr. Richard


       Gralla to review symptom and safety results from


       Study JMEI.


                Safety Profile from the Pivotal Study JMEI



                           Richard Gralla, M.D.


                 DR. GRALLA:  Thank you, Dr. Bunn, and good






                 In considering second-line treatment in


       any patient with advanced lung cancer, both


       physicians and patients also regard the safety or


       toxicity of an agent with great concern.


                 At the same time, all wish to preserve the


       efficacy benefits of treatment including symptom


       control and to do so with fewer potential risks


       from treatment.




                 Recognizing that significant patient


       reported outcome advantages, including pain


       control, were seen with the docetaxel when compared


       with supportive care, as Dr. Shepherd discussed


       with TAX 317 trial, it was important to assess


       prospectively this efficacy parameter in the


       current trial.


                 The study was designed to evaluate the


       impact of symptoms as measured by the average



       symptom burden parameter of the LCSS instrument.


       Dr. Bunn outlined briefly the significantly lower


       toxicity profile with Alimta, which I will discuss


       in greater detail in a few minutes, but it is


       crucial to ascertain that the safety advantages


       were not achieved at the expense of the decrease in


       symptom control as expressed by patients.




                 PRO, or patient reported outcome


       evaluations, are best conducted when using


       previously validated instruments. The LCSS has good


       published psychometric properties and was selected


       for prospective use in this trial for several




                 It is demonstrated high patient and


       observer acceptability, it was designed


       specifically for randomized comparative clinical


       trials, and was used in the docetaxel TAX 317 and


       320 trials.


                 Patients completed the instrument weekly,


       allowing 85 percent of the patients to be included


       in the PRO evaluation.



                 Two major questions are associated with


       PRO evaluation.  First, are the quality of life


       instruments used sensitive enough to reflect


       changes that patients experience, and, second, is


       there value in receiving second-line chemotherapy


       in terms of symptom relief and quality of life




                 Does the magnitude of response, major


       response versus stable disease versus progressive


       disease predict the degree of benefit expressed by






                 This slide shows the patient reported


       results displayed by the objective response


       category achieved.  For this analysis, the results


       of both the Alimta and docetaxel arms were




                 As can be seen, major response was


       associated with the greatest patient expressed


       benefit, the green bars, while a lesser impact, but


       still a positive result, was reported by those


       patients in whom stable disease, the magenta bars,



       was their best response.


                 Of note is the fact that over 50 percent


       of patients had either a major response or stable


       disease in this trial, and that these groups


       reported symptomatic benefits as seen on the slide.


                 In light of the PRO benefits overall in


       the trial, and with the significantly lower


       toxicity on the Alimta arm, it is important to see


       that the response related symptomatic benefits were


       preserved with the less toxic Alimta regimen.




                 This slide shows the evaluations for


       patients by randomized treatment arm and examines


       the results seen in those patients with major


       response or stable disease.


                 The bar graphs represent the six general


       and thoracic symptoms evaluated in the LCSS and the


       average symptom burden index, or ASBI.  It is clear


       that these results show similar symptom


       amelioration for each treatment arm in these lung


       cancer related symptom areas.





                 In all new agent evaluation, efficacy and


       safety are the main considerations.  Given the


       similar efficacy endpoints in terms of survival,


       response, and patient reported outcomes found with


       both agents in this large randomized trial, safety


       issues are of marked importance when considering


       therapeutic index differences between the agents.


                 To place the overall safety profiles for


       second-line treatment in context, it is useful to


       review briefly the safety findings of the currently


       available second-line agent docetaxel.




                 The docetaxel arms at 75 mg/m2 from the


       TAX 317 and 320 trials, which Dr. Shepherd outlined


       in her presentation, are seen on this slide.


                 When one concentrates on marked


       toxicities, as expressed as a percentage of


       patients experiencing Grade 3 or 4 levels of


       toxicity, it is clear that neutropenia is the


       primary concern occurring in the majority of


       patients.  In fact, as originally designed, the


       amount of docetaxel given in TAX 317 had to be



       lowered during the study to 75 mg/m2  because of


       undue toxicity.


                 Nonetheless, even at this dose, nearly


       two-thirds of patients still experienced marked or


       severe neutropenia. Physicians remain particularly


       concerned with the high degree of this potentially


       life-threatening toxicity.


                 While patients and physicians appreciate


       the modest benefits of docetaxel, concerns with


       neutropenia and its complications have led to the


       frequent need for growth factor injections and


       alterations of doses and schedules.




                 An overall view of the safety in the JMEI


       trial is seen in this slide.  The table shows the


       incidence of the most serious toxicity, death,


       serious adverse events or SAEs, and finally, any


       adverse event called the treatment emergent adverse


       event, or TEAE.


                 As can be see for any of these parameters,


       a higher rate of adverse events was found in this


       study with the docetaxel arm.



                 When one looks at either the serious


       adverse events affecting a minority of patients, or


       the treatment emergent adverse events affecting


       most patients, significant differences favoring the


       Alimta arm are found when the results are evaluated


       for events that are drug related.




                 Of course, the toxicity outcome of


       greatest concern with any drug is death.  As seen


       in this slide, while the number of deaths during


       the study are relatively similar between the two


       treatment arms, fewer deaths are seen in total on


       the Alimta arm, and in the important categories of


       study drug related deaths and lung cancer related






                 When examining adverse events, any


       toxicity can be relevant, but major toxicity, that


       is, Grade 3 and 4, is of greatest concern and


       deserves our focus.


                 Clearly, an approach that lessens toxicity


       from the marked Grade 3 and 4 categories to Grades



       1 and 2 would have the same overall toxicity


       percentage, but by lessening the severity would be


       a major benefit.  All drugs have side effects, the


       severity of these side effects is a crucial issue


       in patient management and in the assessment of


       toxicity in this trial.




                 This slide is the first of several


       summarizing laboratory-based major toxicities from


       the current Alimta versus docetaxel randomized


       trial as displayed as Grade 3 and 4 level of




                 As expected, the most commonly occurring


       laboratory-measured side effect was neutropenia.


       Of note is the finding that there was a markedly


       different occurrence of this toxicity depending on


       the treatment arm.


                 Not only was there a highly significantly


       different rate of neutropenia, favoring those


       patients randomly assigned to Alimta, but the


       related life-threatening toxicity of febrile


       neutropenia occurred far less often in the



       Alimta-treated patients affecting fewer than 2




                 Not surprisingly, documented infection


       rates were lower in those patients receiving Alimta


       with no occurrences found on this arm of the trial.


                 Now, stepping away from the statistical


       analysis at this point and placing it in a clinical


       context, these results mean that 1 of every 8


       patients in this study, randomized to docetaxel,


       had febrile neutropenia, while this


       life-threatening toxicity occurred in less than 1


       of every 50 patients on Alimta.


                 The only laboratory area in which a


       significantly higher side effect rate was seen with


       the Alimta, was in the hepatic transaminase ALT.


       Fortunately, this degree of elevation was uncommon,


       occurring in fewer than 3 percent of patients.




                 In general, rates of non-laboratory side


       effects were relatively low in this study.


       Nonetheless, the distressing but not


       life-threatening side effect alopecia occurred far



       less often in patients receiving Alimta.


                 Additionally, a significantly different


       rate of serious diarrhea was found again favoring




                 Thus, when considering both laboratory and


       non-laboratory events, threatening overlapping


       toxicities, such as neutropenia and diarrhea, were


       significantly reduced by the use of Alimta.




                 When one looks at the occurrence of all


       serious laboratory toxicities, that is, Grade 3 and


       4, by treatment regimen, it is clear that Grade 3


       toxicities occurred in only about half as many


       patients randomly assigned to the Alimta arm, and


       that Grade 4 toxicities were markedly lower in


       patients on that arm.




                 During the trial, anemia was reported by


       about 7 percent of patients on either arm of the


       study.  This could be related to the chemotherapy


       or to anemia associated with the lung cancer


       itself.  Overall, physicians elected to transfuse



       or to give erythropoietin to between 22 percent and


       24 percent of patients with no significant


       differences between treatment arms.


                 With markedly lower drug-induced


       neutrophil counts on the docetaxel arm, 7 times as


       many of these patients were given


       granulocyte-stimulating growth factors, again a


       highly significant difference.




                 The advantages in non-laboratory


       toxicities are perhaps best illustrated when


       looking at serious toxicities of any cause.  The


       more minor toxicity grades 1 an 2 are similar


       between the treatment arms, however, when one


       reviews the more serious toxicity grades, important


       differences are clear.


                 Grade 3 toxicity rates approach


       statistical significance.  In Grade 4, the most


       marked toxicity category, a third fewer patients on


       the Alimta arm had this rate of serious toxicity a


       statistically significant difference between the


       treatment arms.





                 It can be useful to review briefly


       hospitalization patterns.  As seen in this slide,


       hospitalizations due to adverse events of all


       causes were significantly lower in patients on the


       Alimta arm.


                 The driving factor behind this rate


       involved the significantly fewer hospitalizations


       for the life-threatening complication of febrile


       neutropenia.  Paradoxically, the number of days in


       hospital was modestly greater in the Alimta arm.


       This imbalance was due entirely to non-drug-related


       factors, that is, longer hospitalizations for


       social considerations and for management of


       complications of the metastatic lung cancer, not


       for drug-related issues.


                 In particular, it is the appropriate


       concern with the risk of major toxicity that limits


       the willingness of physicians to advise second-line


       docetaxel despite demonstrated survival and


       symptomatic gains from the TAX 317 study as


       outlined by Dr. Shepherd.



                 Many individuals involved in new agent


       investigation have struggled to display clearly


       this balance between toxicity and benefit, or at


       least ways of showing the overall effect of major


       toxicity rates on survival.




                 This slide demonstrates one attempt to do


       this.  It is interesting to look at the experiences


       of all patients on this large Alimta versus


       docetaxel trial with regard to the time of


       survival, which was free of serious Grade 4




                 As is seen in terms of the remaining


       period of survival, patients randomized to the


       Alimta arm spent two to three times as long without


       this degree of serious toxicity when compared with


       those on docetaxel.


                 This analysis helps to demonstrate the


       impact of the more favorable toxicity profile of


       Alimta when compared with docetaxel.




                 We conclude that this large multi-center



       trial demonstrated several major advantages for the


       group randomized to Alimta with the real but


       limited benefits found in second-line treatment of


       non-small cell lung cancer.  A decrease in the risk


       of treatment is an important advantage for Alimta.


                 These significant benefits were found in


       the key areas of decreased neutropenia and febrile


       neutropenia, less risk of alopecia and diarrhea,


       and few drug-related deaths and serious adverse


       events overall.


                 From a safety and patient reported


       outcomes perspective, Alimta is a useful and safe


       treatment option for patients with non-small cell


       lung cancer who are candidates for second-line




                 The toxicity advantages associated with


       Alimta with similar symptomatic and quality of life


       benefits are of great value to patients.  The PRO


       and toxicity evaluations, coupled with the other


       major endpoints, help to support the finding that


       Alimta treatment is safer without any compromise in


       survival response or palliative outcomes.



                 I would like now to call on Dr. Bunn to


       summarize these results and to put them into the


       context of current treatment.


                           Overall Conclusions


                             Paul Bunn, M.D.


                 DR. BUNN:  In the past three talks, we


       have reviewed the relevant data supporting Alimta


       for the treatment of advanced non-small cell lung


       cancer after prior chemotherapy.  I would like to


       take a few minutes to summarize the salient issues


       in your review.  I also appreciated Dr. Pazdur's


       overview of the issues before you and just make a


       few comments as I go through my presentation.


                 Of course, you are here to provide your


       advice to the agency.  Your advice is largely going


       to depend on how much you think about safety and


       about efficacy, and your confidence in the safety


       and the efficacy relate to survival, they relate to


       patient-reported outcomes, and they relate to


       safety, and we must consider not only the JMEI


       trial, but what is known in the literature, as Dr.


       Pazdur alluded to before and how confident are we



       about what best supportive care does and how


       confident are we about what docetaxel does and how


       many trials are there.




                 From this presentation, Alimta clearly


       provides a new, a safe and clearly an effective


       treatment option for patients with advanced


       non-small cell lung cancer in the second-line




                 This is important as advances in


       treatment, patients with lung cancer are living


       longer and they are living better.  As a result,


       more of these patients are candidates for


       second-line therapy.


                 At present, they have only one approved


       option, docetaxel.  As noted, docetaxel's use is


       limited by its significant toxicities and also its


       use in the first-line setting.




                 What about safety?  Alimta is clearly


       safer than docetaxel with respect to any clinically


       relevant toxicity. Its advantage, of course, is



       most marked in the reduction of febrile


       neutropenia, from 12.6 percent to 1.9 percent.


                 A secondary benefit that results from this


       is a concomitant reduction in the use of G and


       GM-CSF, fewer visits to the clinic for neutropenia,


       fewer hospitalizations for neutropenia.


                 However, not all the benefit is isolated


       to reduction in neutropenia.  There was also a


       significant reduction in Grade 3/4 diarrhea and a


       reduction in alopecia, a side effect particularly


       important to patients.


                 Finally, there was a 3-fold reduction in


       hospitalization for drug-related adverse events.




                 How confident can we be in the safety


       profile of Alimta?  Shown here are the safety


       results of Alimta in JMEI and in the safety


       database of all other Phase II monotherapy of


       Alimta with vitamins.


                 Of note is the consistent results of


       Alimta in febrile neutropenia, in diarrhea and


       alopecia, that were all lower than docetaxel in







                 On looking at the direct pivotal trial


       evidence for survival benefit from JMEI, Alimta has


       comparable activity with a hazard ratio of 0.99.


       Median survivals are essentially the same.


       One-year survival rates were identical and there


       was internal consistency across all groups.


                 When indirectly compared to best


       supportive care, Alimta preserved at least 50


       percent of docetaxel's benefit over best supportive




                 With respect to non-inferiority analyses,


       the 1.11 fixed margin was not met statistically,


       and many p-values can be calculated different


       methods, however, we can be confident that Alimta


       retains docetaxel survival advantage over best


       supportive care, not only from comparison to TAX


       317B, but also comparison to other historical best


       supportive care trials and the consistency of


       Alimta's survival result across all first- and


       second-line trials that you have heard.





                 Reviewing all secondary endpoints, the


       following conclusions can be made from a direct


       comparison to docetaxel from JMAI.


                 The time to progression is identical


       almost.  Progression-free survival was the same,


       and the response rate was very similar.  Over 50


       percent of all patients on each arm showed improved


       or stable symptoms.


                 Indirectly, the response rates of median


       time to progression for Alimta are consistent


       across all trials and show relevant activity in all


       non-small cell lung cancer either in the first line


       or second line, and these endpoints are superior to


       historical best supportive care.  So, this is what


       Dr. Pazdur was talking about.


                 How do clinicians review efficacy of a


       compound, and how can we tell if one seems similar


       to another?  It is helpful if there are multiple


       randomized trials.


                 Fortunately, there are five randomized


       trials of docetaxel in the second-line setting, and



       those five randomized trials are shown here.




                 Obviously, a meta-analysis has not been


       done because some of these are recent, but these


       are the five randomized trials using docetaxel 75


       mg/m2 in one arm.  These consistent results with


       median survivals of 6 to 8 months in all trials


       give us confidence about the effect of docetaxel.


                 In each of these five studies, docetaxel


       75 mg/m2 was numerically superior to the


       comparator.  Note that two of these trials, the


       comparator was docetaxel 100 mg/m2  with the worst


       outcome.  That is the reason there are not A versus


       A + B trials in the second-line setting.  Just a


       little bit of extra neutropenia made survival worst


       in these patients, and it does limit our ability to


       develop new agents, because the A + A + B design is


       very difficult in this setting.


                 If one were to review, then, the best


       supportive care results from available second-line


       randomized trials, once again we see consistent


       results.  Median survival in the best supportive



       care arms was 4.5 and 5.5 months.


                 The BR21 slide results that are shown on


       this slide is limited to those patients who got


       second-line therapy, as that trial also included


       some third-line patients.


                 These survival rates with the best


       supportive care are clearly inferior to docetaxel.


       Finally, when one reviews the median survival for


       Alimta in this context, the similar outcomes of


       docetaxel and the superiority to best supportive


       care is obvious.




                 In summary, Alimta merits full approval as


       a single agent for the treatment of patients with


       locally advanced or metastatic non-small cell lung


       cancer after prior chemotherapy.


                 There are many agents that have received


       full approval that you know about, sometimes based


       only on response rate.  Here, we have data and


       efficacy on response rate, progression-free


       survival, and survival, as well as patient reported





                 Alimta has a superior response rate,


       progression-free survival, and survival compared to


       best supportive care.  Alimta has similar response


       rate, progression-free survival, and survival


       compared to docetaxel.


                 The safety profile of Alimta is clearly


       superior to docetaxel.  There are many second-line


       lung cancer patients. They deserve to be offered


       the safest and most effective treatment that


       physicians have available.


                 Approval of this drug will make a safe and


       effective agent available for patients with this


       devastating disease.


                 Thank you for your attention.


                 DR. BRAWLEY:  Thank you, Drs. Bunn,


       Gralla, Herbst, Shepherd, and Paoletti, and your


       support staffs for preparing the presentation.


                 We would now like to move to the FDA


       presentation, the clinical review and the


       statistical review.


                 The clinical review will be given by Dr.


       Martin Cohen.



                             FDA Presentation


                             Clinical Review


                          Martin H. Cohen, M.D.


                 DR. COHEN:  Good morning.  My name is


       Martin Cohen and I am going to present the FDA


       clinical review of Alimta, also known as pemetrexed


       and LY231514.


                 My review will be followed by the FDA


       statistical review by Dr. Wang.




                 The proposed indication for Alimta is as a


       single agent for the treatment of patients with


       locally advanced or metastatic non-small cell lung


       cancer after prior chemotherapy.




                 A single study was submitted comparing


       treatment with Alimta to treatment with docetaxel.


       The stratification factors were performance status,


       disease stage, number of prior regimens, response


       to the last prior chemotherapy, whether or not the


       patient received prior platinum or paclitaxel


       therapy, homocysteine levels, and treatment site.





                 I would like to comment on the


       determination of baseline homocysteine values.


       Elevated pre-treatment homocysteine values have


       previously been shown to be an excellent predictor


       of Alimta treatment toxicity and that reduction of


       those elevated homocysteine levels with folic acid


       and vitamin B12 was accompanied by a significant


       reduction in Alimta toxicity.


                 Whether vitamin supplementation would also


       decrease docetaxel toxicity is unknown.  There is


       no reason, however, not to expect a toxicity


       reduction similar to that observed with Alimta.




                 Since docetaxel is the comparator in the


       Alimta trial, this slide summarizes the clinical


       materials that were submitted for approval of


       docetaxel as second-line non-small cell lung




                 The first study listed on this slide, as


       previously discussed, was reported by Dr. Shepherd


       and colleagues.  In this study, patients with



       performance status zero to 2, who had failed one or


       more platinum-based chemotherapy regimens, were


       initially randomized to receive docetaxel 100 mg/m2


       or best supportive care.


                 Because of early toxic deaths, the


       protocol was amended to reduce the docetaxel dose


       to 75 mg/m2.  After this amendment, there were 55


       patients who received docetaxel 75 mg/m2 and 49


       patients who received best supportive care.


                 Docetaxel treatment gave a response rate


       of 5.5 percent.  The median survival was 7.5 months


       for docetaxel versus 4.6 months for best supportive


       care.  The difference in overall survival was


       statistically significant at a p-value of 0.01, and


       one-year survival was 37 percent versus 12 percent,


       and that also was statistically significant.


                 The second study on the slide was reported


       by Fosella and colleagues.  This was a randomized


       trial comparing docetaxel 100 mg/m2 or docetaxel 75


       mg/m2 to a physician's choice of either vinorelbine


       or ifosfamide.


                 The study population had a higher percent



       of Stage IV patients and more patients who had


       received two or more prior chemotherapy regimens


       than did the Shepherd study.  The docetaxel 100


       mg/m2 dose was again associated with early toxic


       deaths and will not be discussed further.


                 The 75 mg/m2  docetaxel-treated patients


       had a response rate of 5.7 percent versus 0.8


       percent for the physician's choice arm.  The median


       survivals were 5.7 to 5.6 months, and the one-year


       survivals were 30 percent versus 20 percent.


                 The difference in overall survival between


       the two treatment groups was not statistically


       significant.  The p-value for the one-year survival


       difference was 0.025.




                 Alimta drug administration is shown on


       this slide. Alimta 500 mg/m2 was administered


       intravenously over 10 minutes on day 1 of a 21-day


       treatment cycle.


                 Patients receiving Alimta, as mentioned


       previously, also received folic acid, vitamin B12,


       and dexamethasone at the doses and schedules listed



       on the slide.


                 Folic acid and vitamin B12 were


       administered for the purpose of reducing blood


       homocysteine levels so as to ameliorate Alimta


       toxicity.  Dexamethasone was given to prevent or


       decrease the occurrence of skin rash.




                 Docetaxel drug administration is shown on


       this slide.  Docetaxel 75 mg/m2 was administered


       intravenously over 60 minutes on day 1 of a 21-day


       treatment cycle.


                 Dexamethasone in the doses scheduled


       listed on the slide was given as prophylaxis


       against fluid retention and hypersensitivity






                 There were 135 investigational sites in 23


       countries that participated in this study, and


       approximately 21 percent of the study population


       came from United States institutions.




                 This slide demonstrates selected patient



       characteristics.  As shown the two treatment groups


       were comparable for performance status, prior


       chemotherapy regimens, prior platinum and


       paclitaxel therapy.


                 Approximately 30 percent of patients in


       each treatment group had an elevated baseline


       homocysteine level.




                 This slide shows efficacy endpoints.  The


       primary endpoint was overall survival, and the FDA


       survival analysis will be discussed in the


       following FDA presentation.


                 Secondary efficacy endpoints included


       response rate and duration, time to progression,


       progression free survival, and lung cancer systems


       as measured by the Lung Cancer Symptom Scale.


                 Because progression free survival results


       mirror time to progression, only the former will be


       discussed on the subsequent slide.  Similarly,


       because no differences were identified between the


       two patient groups in any of the Lung Cancer


       Symptom Scales, symptom burden will also not be



       further discussed.




                 Alimta treatment resulted in 1 complete


       response and 23 partial responses, for an overall


       response rate of 9.1 percent.  Docetaxel treatment


       resulted in no complete responses and 24 partial


       responses, for a response rate of 8.8 percent.


                 The overlapping 95 percent confidence


       limits of the two response rates are listed.


       Median response durations were 4.6 months for


       Alimta and 5.3 months for docetaxel.




                 This slide shows time to progression for


       both the intent to treat, or ITT patient


       population, and the randomized treated, or RT


       patient population.


                 As indicated, time to progression was


       similar for Alimta and for docetaxel treatment


       groups whether one compares results for either the


       ITT or RT population groups.  For the ITT


       population, there was a slight advantage of median


       time to progression favoring Alimta, whereas, for



       the RT population, there was a slight advantage


       favoring docetaxel.




                 Now, we get to one of the more


       controversial aspects of this review, the issue of


       post-study chemotherapy.  The patient population


       analyzed in this slide is the randomized and


       treated population.


                 At the time of disease progression,


       patients were allowed to receive post-study


       chemotherapy.  This slide lists the drugs that were


       most frequently used.  As indicated on this slide,


       126 or 48 percent of Alimta-treated patients and


       107 or 39 percent of docetaxel-treated patients


       received post-study chemotherapy.


                 Of possible importance to a


       non-inferiority survival analysis, 85 or 32 percent


       of Alimta-treated patients crossed over to


       docetaxel treatment.  Patients on the docetaxel arm


       were not permitted to cross over to Alimta, and


       they received a variety of other drugs including


       those listed on this slide.





                 This slide shows the median survival of


       randomized treated populations who received or did


       not receive post-study chemotherapy.


                 139 Alimta patients did not receive


       post-study chemotherapy and 169 docetaxel-treated


       patients did not receive post-study chemotherapy.


       The 30 patient difference between the two treatment


       arms might be important, because patients on both


       study arms who did not receive post-study


       chemotherapy had shorter median survivals, 6.2


       months for Alimta patients and 5.0 months for


       docetaxel patients than patients who did receive


       post-study chemotherapy, as summarized in the last


       two lines on this slide.




                 Because this slide demonstrates that


       post-study chemotherapy improved survival, it is


       important to look at patients who did not receive


       post-study chemotherapy.  The presumption might be


       made that these patients were too sick to receive


       treatment, and that is why they had a worse





                 This does not appear to be the case,


       however.  This slide shows the last recorded


       performance status of patients who did not receive


       post-study chemotherapy.  Again, there were 139


       Alimta-treated patients and 169 docetaxel-treated




                 As is evident from this slide, the large


       majority of patients who did not receive post-study


       chemotherapy were performance status zero or 1 at


       their last study visit, and conceivably, could have


       received additional treatment.




                 In our previous look at this slide, we


       were concerned with patient who did not receive


       post-study chemotherapy.  We are now concerned with


       patients who were treated.


                 While it appears that all treatments,


       including post-study docetaxel or post-study other


       chemotherapy, gave comparable survival results, it


       must be remembered that these are not randomized


       patients and that prognostic features of each group



       may be very different.


                 Thus, post-study chemotherapy treatment


       may well have been of more benefit than post-study


       docetaxel treatment may well have been more


       beneficial than other post-study chemotherapy






                 Turning now to safety considerations, this


       slide shows patient exposure to treatment.  The


       median number of cycles we see by patients on each


       treatment arm was 4, and there was no striking


       difference in the percent of planned dose intensity


       received by patients on either treatment arm.




                 This slide summarizes all toxicities


       experienced by study patients regardless of


       causality based on their CTC grade.  As evidence


       from this slide, there was no difference between


       Alimta and docetaxel for Grade 1 and Grade 2


       toxicities.  For Grade 3 toxicity, Grade 4


       toxicity, and Grade 3 or 4 toxicity, Alimta was


       significantly less toxic than docetaxel.



                 Alimta's safety advantage for Grade 3 or 4


       toxicity comes primarily from less neutropenia,


       less febrile neutropenia, and less infection


       accompanying neutropenia.




                 Looking specifically at neutropenia, this


       slide shows Grade 3 to 4 neutropenia accompanied


       with fever or with infection.  Thirty-six or 13


       percent of docetaxel-treated patients had febrile


       neutropenia versus 5 or 2 percent of Alimta-treated




                 Also, indicated on this slide, documented


       infection in the setting of neutropenia occurred in


       5.8 percent versus zero percent of docetaxel and


       Alimta-treated patients, respectively.




                 Therefore, if one now looks at all


       toxicities regardless of causality excluding white


       blood cell events, such as decreased leukocytes and


       lymphocytes, neutrophils, granulocytes, infections,


       febrile neutropenia, or other white blood cell


       related events, there is no longer a significant



       difference between Alimta and docetaxel treatment.


                 For Grade 3 or 4 toxicity, for example,


       the p-value is 0.781.




                 CTC Grade 3 or 4 adverse events regardless


       of causality are listed on this slide.  As


       indicated, alopecia and diarrhea occurred


       significantly more often with docetaxel treatment


       than with Alimta treatment.


                 Grade 3 to 4 diarrhea occurred at 4


       percent of docetaxel-treated patients versus 0.4


       percent of Alimta-treated patients.


                 There was no statistically significant


       difference in the occurrence of the other listed


       toxicities - fatigue, nausea, vomiting, stomatitis,


       pulmonary toxicity, or neurosensory toxicity.




                 Turning now to treatment emergent adverse


       events, of TEAEs, this slide shows all treatment


       emergent adverse events regardless of causality for


       which there was a statistically significant


       difference between treatment groups based on an



       uncorrected p-value of less than 0.001.


                 As shown, nausea, weight loss, increase in


       hepatic enzymes, the alanine and aspartate amino


       transferases, and decrease in creatinine clearance


       were all more frequent in Alimta-treated patients.


       Alopecia was worse in docetaxel-treated patients.




                 This slide shows all treatment emergent


       adverse events regardless of causality for which


       there was a statistically significant difference


       between treatment groups and an uncorrected p less


       than 0.05 value.


                 Myalgias, arthralgias, neurotoxicity, and


       diarrhea were all more common in docetaxel-treated


       patients, while constipation, fatigue, and skin


       rash were more common in Alimta-treated patients.




                 Hospitalizations present a mixed picture.


       Docetaxel-treated patients had somewhat more


       hospital admissions, 364 versus 337, but


       Alimta-treated patients spent somewhat more time in


       the hospital, 1,722 days versus 1,410 days for



       docetaxel-treated patients.




                 As regards efficacy conclusions, you will


       hear the opinion of the FDA statisticians regarding


       survival subsequently.


                 Whatever your views on the relative merits


       of the survival analyses, however, the fact is that


       post-study chemotherapy confounds the survival




                 With regards to post-study chemotherapy,


       there are two issues.  The first issue is the


       crossover of 85 Alimta-treated patients to


       docetaxel treatment.  While median survival of


       these patient is similar to the median survival of


       patients receiving other chemotherapy regimens,


       such survival analyses do not take into account


       possible prognostic differences between the various


       treatment groups.


                 The second issue is that patients who did


       not receive post-study chemotherapy had a shorter


       survival than those who did receive such treatment.


       There were 30 more docetaxel-treated patient than



       Alimta-treated patients who did not receive


       post-study chemotherapy.


                 The large majority of untreated patients


       had a performance status of zero or 1 at the time


       of progression, and could conceivably have received


       additional treatment.


                 Alimta did show evidence of activity,


       however, in that it produced a response rate of 9.1






                 The toxicity spectrum of docetaxel clearly


       differs from that of Alimta, and this slide


       summarizes the differences between the two drugs.


                 Docetaxel produces more neutropenia and


       neutropenic complications, including febrile


       neutropenia, infections, and need for


       colony-stimulating factors.  It also causes more


       neurotoxicity, myalgias, alopecia, and diarrhea.


                 Alimta, on the other hand, produces more


       thrombocytopenia, more skin rash, more nausea and


       vomiting, more elevations of hepatic enzymes, a


       decrease in creatinine clearance, and more weight



       loss than does docetaxel treatment.


                 An important point on this slide is that


       folic acid and vitamin B12 supplements presumably


       by reducing elevated homocysteine levels have been


       shown to ameliorate Alimta toxicity.  Whether such


       supplements, which were not given to


       docetaxel-treated patients, would ameliorate


       docetaxel toxicity is not known.


                 Thank you for your attention.


                 DR. BRAWLEY:  Thank you, Dr. Cohen.


                 Dr. Yong-Cheng Wang.


                            Statistical Review


                          Yong-Cheng Wang, Ph.D.


                 DR. WANG:  Thank you, Dr. Cohen.


                 Good morning.  I am Yong-Cheng Wang, the


       statistical reviewer for the application being


       discussed today.  In this presentation, I will


       present the results of efficacy analysis of Study






                 Here is the outline of my presentation.


       The results of protocol specified primary endpoint



       analyses. Post-hoc  50 percent of retention non-inferiority


       analyses, which was submitted in the




                 The critical issues in Study JMEI.  The


       results of secondary endpoint analyses.  Efficacy


       conclusions will be given at the end of this






                 The protocol specified two study


       objectives, superiority hypothesis and fixed margin


       non-inferiority hypothesis.


                 In the superiority hypothesis, the goal is


       to demonstrate that Alimta is more effective than




                 In the fixed margin non-inferiority


       hypothesis, the goal is to demonstrate that Alimta


       is not worse than docetaxel by 11 percent clinical


       benefit, or in other words, that non-inferiority


       margin is fixed at 1.11.


                 The fixed margin of 1.11 was specified at


       the recommendation of EMEA, and was not based on


       any historical trial data.  However, from our



       calculation, this margin is close to FDA/CBER






                 Here are the results of primary endpoint


       overall survival analysis for the intent to treat


       population.  For the overall survival, the median


       survival is 8.3 months for the Alimta group and 7.9


       months for docetaxel group.


                 The study failed to demonstrate superior


       efficacy of Alimta to docetaxel with a log-rank


       p-value of 0.93.  It also failed to demonstrate


       non-inferiority based on the fixed margin


       non-inferiority test.  The p-value is 0.256.


                 Based on the Cox regression model, the HR


       of Alimta versus docetaxel is 0.99 with 95 percent


       confidence interval 0.82 to 1.2.  The


       non-inferiority margin 1.11 is less than the upper


       limit 1.2.




                 For the randomized and treated population,


       the results are similar to ITT population as


       presented in the previous slide.





                 The sponsor also included a post hoc


       non-inferiority hypothesis of 50 percent of


       retention of docetaxel effect in the NDA




                 In this hypothesis, the goal is to


       demonstrate that at least 50 percent of docetaxel


       effect will be retained by Alimta.  In the current


       study, we have serious reservation about this


       analysis as presented in the next few slides.




                 There are two major critical issues in


       Study JMEI.  First, the docetaxel effect is


       estimated from only one small historical trial,


       therefore, we cannot assure the ability to repeat


       the results.


                 Also, we cannot reliably assess the


       magnitude of the docetaxel effect.


                 Second, the survival results are


       confounded by crossover of Alimta to docetaxel.




                 I will now go over the details of these



       critical issues.  The historical trial which is


       used for the estimation of the docetaxel effect is


       TAX 317.  As presented here, this is a very small


       trial, total of 104 patients were enrolled with 55


       patients in the docetaxel arm and 49 patients in


       the best supportive care arm.


                 So, the estimate of the docetaxel effect


       is not reliable and not robust.  Since this is the


       only one historical trial used for the estimation


       of docetaxel effect, the constancy assumption that


       docetaxel effect in Study JMEI is the same as in


       the historical trail cannot be verified.


                 It should also be noted that these results


       are in the ITT population only, and we do not have


       results based on the randomized and treated






                 This slide shows the critical issue of


       treatment crossover of Alimta to docetaxel.  There


       are more than 30 percent patients who crossed over


       from Alimta group to docetaxel group.  Therefore,


       the survival results are confounded.





                 I will now present the results of


       secondary endpoints analysis.




                 This slide shows the results of survival


       rate analysis.  For the 6 month, Alimta has a


       slightly higher relative risk than docetaxel in the


       survival rate.


                 For the 12, 18, and 24 months, Alimta has


       a slightly lower relative risk than docetaxel for


       the survival rate.




                 This slide shows the results of time to


       progressive disease.  Alimta is not significantly


       superior to docetaxel for the time to progressive


       disease in the ITT population.




                 This slide shows the results of


       progression-free survival.  These results are


       similar to the time to progressive disease.




                 This slide shows the results of response



       rate analysis.  Alimta is not significantly


       superior to docetaxel with respect to tumor


       response.  The results of symptom improvement


       analysis are not present either, as there was


       missing data.  Results were based on a subset of


       patients in this open label study.


                 It should be noted that even though


       p-values have been presented for all the secondary


       endpoint analysis, these values are not


       interpretable, and none of them are adjusted for




                 Efficacy conclusions.  Based on the


       overall survival analysis, a single, randomized,


       open-label, multi-center study JMEI in advanced


       non-small cell lung cancer patients treated with


       Alimta versus docetaxel failed to demonstrate


       superior efficacy of Alimta to docetaxel.


                 It also failed to demonstrate


       non-inferiority compared to docetaxel.




                 The estimate of docetaxel effect based on


       only one small historical trial is not reliable and



       not robust.


                 In the presence of treatment crossover


       from Alimta to docetaxel, the survival results are


       confounded and non-inferiority analysis is very


       difficult to interpret.


                 Therefore, the result of 50 percent


       retention non-inferiority analysis is not




                 Thank you for your attention.


                 DR. BRAWLEY:  Thank you.


                 As we move forward, I would like to ask


       Dr. Pazdur if he wants the current questions,


       Question No. 1 and Question No. 2, and you would


       like a vote on Question No. 1 and Question No. 2.


       Thank you very much.


                 At this point, it is 10:31.  I would


       propose that we go to break until 10:45.  I would


       ask the members to be back in their seats at 10:45.


       I think we can finish a little earlier today than


       is currently posted.




                 DR. BRAWLEY:  As we come to order, this is



       the section for open public discussion.  I


       understand there is one discussant.  I need to say


       the following:


                 Both the Food and Drug Administration and


       the public believe in a transparent process for


       information gathering and decisionmaking.  To


       ensure such transparency at the open public hearing


       session of the Advisory Committee meeting, FDA


       believes that it is important to understand the


       context of an individual's presentation.


                 For this reason, FDA encourages you, the


       open public hearing speaker, at the beginning of


       your written or oral statement to advise the


       committee of any financial relationship that you


       may have with the sponsor, its product, and, if


       known, its direct competitors.


                 For example, this financial information


       may include the sponsor's payment of your travel,


       lodging, or other expenses in connection with your


       attendance at the meeting.


                 Likewise, FDA encourages you at the


       beginning of your statement to advise the committee



       if you do not have any such financial


       relationships.  If you choose not to address this


       issue of financial relationships at the beginning


       of your statement, it will not preclude you from




                 I am sorry.  That is an official sort of


       thing that has to be read into the record.


                 MS. POLLACK:  I understand.


                           Open Public Hearing


                 DR. BRAWLEY:  If you can introduce


       yourself and begin your statement.


                 MS. POLLACK:  Certainly.  Good morning.


       My name is Michelle Pollack and I am the Director


       of Marketing and Development for the Wellness


       Community, an international non-profit organization


       that provides support, education, and hope to


       people affected by cancer.


                 For the record, the Wellness Community


       receives unrestricted educational funding from Eli


       Lilly, however, we received no funding or


       compensation for my presence here today.


                 The Wellness Community offers free



       programs including professionally led support


       groups, educational seminars, nutritional


       workshops, exercise and mind-body programs, among




                 Our mission is to help people living with


       cancer regain a sense of control over their lives,


       feel less isolated, and restore their hope for the


       future regardless of the stage of their disease.


                 Last year, we provided support services to


       more than 30,000 people with cancer including


       people with locally advanced or metastatic


       non-small cell lung cancer.  Through the virtual


       Wellness Community on-line, we were able to reach


       even more people.


                 At the Wellness Community, we have learned


       a great deal from those we support and we believe


       in the importance and value of an educated and


       empowered patient.  Since people with cancer often


       feel stigmatized, alone, and overwhelmed with


       grief, they feel stronger and more hopeful when


       they have more treatment options available to them.


                 With an estimated 174,000 new diagnoses of



       lung cancer in 2004 in the United States alone,


       with 80 percent of those non-small cell lung


       cancer, there is no doubt that we are in need of


       improved treatments, more manageable and tolerable


       side effects, and greater accessibility to those




                 We have the opportunity to expand the


       chances that these families have for a better life


       with new treatment options, and we feel very


       strongly about supporting that opportunity.


                 Today, I ask you to carefully consider the


       plight of people with locally advanced or


       metastatic non-small cell lung cancer and empathize


       with the range of daily physiological and


       psychosocial issues that they face.


                 Please take a leadership role in approving


       a broader range of treatments and then encourage


       patients to be informed, empowered, and optimistic


       about the possibility of longer, healthier lives.


                 Thank you.


                 DR. BRAWLEY:  Thank you, Ms. Pollack.


                 I believe there is no other speakers for



       the open public hearing, am I correct?  Hearing


       none, then, we are going to move on.


                 I would like to ask the committee to


       address any questions to either the sponsor or the




                 Dr. D'Agostino.


                       Questions from the Committee


                 DR. D'AGOSTINO:  If I read correctly the


       way the FDA has put the questions to us, the


       discussion really gets onto secondary events and


       toxicity, and so forth, but there is a couple of


       comments in the front statement of the FDA about


       the ability with the one small historical study to


       actually estimate survival and also the crossovers.


                 I know they were mentioned in the


       discussion of the sponsor, but I think it would be


       useful to hear a response from Lilly in terms of


       those two questions, so that we discuss them and


       put them aside, or discuss them and think they are




                 DR. PAOLETTI:  No crossover is inevitable


       in a situation like that especially in the United



       States.  I will ask Dr. Frances Shepherd to review


       the historical context of third-line treatment in


       lung cancer to answer the question in this way.


       Then, I will ask Dr. Bunn to respond to the


       question in terms of what we have observed in our


       data, and, finally, Dr. Scott Emerson from a


       statistical point of view to address this issue.


                 DR. SHEPHERD:  Yes, we really do not feel


       that there was a significant effect on survival


       from crossover. If we could have the first slide


       projected, please.


                 You may be uncomfortable with the survival


       that was achieved with docetaxel in the TAX 317 or


       the TAX 320 trials.  There have been several


       studies that have followed after that of docetaxel


       75 mg/m2, and as Dr. Bunn showed you, every single


       one of those studies had a median survival in a


       very tight range that was similar to the TAX 317




                 So, we now have at least five randomized


       trials of docetaxel showing where the median


       survival is expected to be in this clinical





                 With respect to the best supportive care


       arm, we have fewer studies, and there has been no


       study in the third-line setting of chemotherapy.


                 Looking at this slide, though, a


       retrospective study was done by the M.D. Anderson


       and Institute Gustaf Ruce [ph] looking at 700


       patients who had had first-line and second-line


       chemotherapy.  Of those, 43 were treated with




                 The response rate was a mere 2.3 percent,


       and the median survival was less than four months.


       When you look on the other side of the slide, this


       is the subset analysis from the TAX 317 study.


       This is the only randomized data that exist that


       compare third-line chemotherapy to best supportive


       care.  We do not underestimate the small sample


       size here.  These are exploratory analyses, but


       there is nothing in this curve that would suggest


       that third-line chemotherapy contributes to




                 Next slide, please.



                 I am going to show you the survival curve


       from the BR21 trial, No. 568.  This is the survival


       curve in the BR21 trial, which was a trial of


       placebo and best supportive care versus erlotinib


       in the second- and third-line setting. Erlotinib


       showed a significant survival advantage.


                 I show this to you for two reasons.  One,


       to show you that the survival of the untreated


       group, the median survival was 4.5 months, almost


       identical to the best supportive care group of the


       TAX 317 trial.


                 So, we have a supporting trial that


       provides a similar survival advantage or


       disadvantage with no treatment.  So, it gives us a


       little bit more confidence that the best supportive


       care group in TAX 317 was exactly what we would


       expect to see in larger populations.


                 Now, in actual fact, if you look


       carefully, more patients on the docetaxel arm


       received Iressa, a drug very similar to Tarceva, in


       the third-line setting.  So, in actual fact, the


       only treatment that has been shown to prolong



       survival in the third-line setting is an EGFR


       inhibitor and more patients on the docetaxel arm,


       four times as many patients on the docetaxel arm


       actually received that kind of treatment.


                 So, if anything, that would have favored


       docetaxel and not Alimta.


                 DR. BUNN:  Not only do we wish that we had


       more treatments in the third-line setting to make


       people live longer, but when we look at the


       analysis, it is not, I don't think, appropriate to


       say that the third-line treatment made people live


       longer in the study.


                 People who got chemotherapy in the


       third-line did live longer, but that is just a


       prognostic group.  That is like saying responders


       live longer than aggressive disease. That doesn't


       mean that the treatment made them live longer.


                 But we looked very hard to try to sort out


       whether there was any evidence that third-line


       treatment did anything here to the best of our




                 So, you see here on the top of this curve



       is the overall survival results, and presumably,


       the third-line therapy is given after some period


       of time, and if it had an effect, the curves might


       look different at the end.


                 I think it is easy to say, in the survival


       curve, there is no difference in the beginning,


       there is also no difference at the end.


                 If there had been a difference in


       progression, the time of progression, it might have


       favored one group, and on the lower left you see


       that the time to progressive disease was identical


       in the two things.


                 Finally, if there was an effect post


       study, the post-study survival is shown in the


       lower right curve, as I showed before, and there


       was absolutely no evidence, not even a hint that


       there was some survival effect in the post-study




                 Obviously, post hoc analyses like this are


       difficult, and there are many statistical issues.


       I am going to ask the statistician to get up, from


       a clinical point, no matter how we looked at this,



       we couldn't find any evidence that there was an


       effect of post-study treatment that was different


       between the groups.


                 DR. PAOLETTI:  Dr. Emerson, please.


                 DR. EMERSON:  Scott Emerson from the


       University of Washington.  Slide 64, please.  This


       is a slide, that this is now the fourth time we


       have seen this in some version, and as Dr. Bunn


       remarked earlier, this is a very biased


       presentation, this is not really a very informative


       presentation at all, and I would just like to point


       out what we can say from this and what we can't say


       from this.


                 We certainly can say that those people who


       survived long enough to get post-study chemo,


       survived longer than those who didn't survive


       longer to get post-study chemo.


                 The grouping is true, that there is longer


       survival among those who got post-study chemo, but


       that is not quite as strong as what Dr. Cohen said


       when he said that the post-study chemo made you


       live longer.



                 So, to address that, if I could have the


       slide 669.  We did an analysis that tried to


       compare apples more with apples.  Let's compare


       those people who got post-study chemotherapy at a


       certain point in time with other people who had


       also survived that long, so we will assign your


       group as to whether you got post-study chemo


       according to the time that you are on the study.


                 So, this time-variant covariate analysis


       allows us to compare Alimta to docetaxel, keeping


       that post-study chemo variable constant across the


       groups being compare.


                 It also allows us to estimate the effect


       of post-study chemotherapy.  Let me qualify what


       that effect is.  It allows me to estimate the


       difference in survival among those who got post-study


       chemotherapy to the survival among those who




                 I am not going to claim that this is truly


       a cause and effect, because, of course, this isn't


       randomized.  There was a lot of physician


       discretion that went into this.



                 But from this analysis, if I look among


       patients who had no post-study chemotherapy at any


       time during the study--and again I would have


       switched them to another group if they had--the


       Alimta to docetaxel hazard ratio is actually 0.84,


       it is looking a stronger effect than we saw when we


       just did the intention to treat or RT analyses.


                 If we look at the effect of post-study


       chemotherapy, now I am just going to look at among


       those patients alive at any given time, on the


       docetaxel arm, who are getting post-study


       chemotherapy compared to those on the docetaxel arm


       that aren't getting post-study chemotherapy at that


       same time, the hazard ratio is 1.12.  This estimate


       suggests there is a 12 percent increased risk of


       death if you get post-study chemotherapy.


                 On the Alimta arm, it is far more


       striking.  There is a 58 percent higher chance of


       death among those subjects on the Alimta arm who


       are getting post-study chemotherapy relative to


       those who don't.


                 So, this nonrandomized comparison, which I



       don't really believe is the effect of post-study


       chemotherapy, but this analysis would suggest quite


       the contrary to what was worried about, was that


       the post-study chemotherapy is responsible for the


       better survival is actually if we took this at face


       value, you would say if we could just write in the


       indication that you don't do any post-study


       chemotherapy, we are doing better than docetaxel.


                 I don't believe that, because I truly


       believe that physicians are pretty smart.


                 Can we go back to slide 64 for a moment.


                 What we see here is that 139 subjects had


       no post-study chemotherapy on Alimta and 169


       subjects on docetaxel. My personal belief would be


       that physicians, faced with a progression or a


       patient who is failing on Alimta, would recognize


       that docetaxel has been approved for second-line


       therapy and those patients should really give that




                 I think that physicians are pretty able to


       recognize when patients are in trouble, that they


       are on a path towards worse and worse conditions,



       and I personally believe that that is the primary


       effect that we are seeing with that greater rate.


                 Patients on docetaxel could not be


       switched to another therapy if, in fact, they were


       already experiencing a fair amount of toxicity.


       You wouldn't want to try them again on that


       chemotherapy.  We may just be seeing physician


       behavior, so again, I am not claiming that that


       higher post-study therapy is there, but I am


       claiming that we don't have any evidence to suggest


       in this data that there is an added benefit of


       post-study chemotherapy to improve survival.


                 Lastly, if I could see slide 20, just to


       make a point again that Dr. Shepherd made, and that


       is this concept that in this study, the patients


       receiving that third-line chemotherapy were not


       randomized, but in TAX 317, they were randomized.


                 It is a subgroup analysis, but when we did


       a randomization based on that, we clearly saw no


       benefit.  That would be presumption, that if we had


       done randomization to third-line therapy, that this


       would likely have been the case and that we



       wouldn't have seen that added risk.


                 DR. PAOLETTI:  Dr. D'Agostino, should we


       answer your second question, or do you want to


       continue on this issue?


                 DR. D'AGOSTINO:  It is up to the Chair.


                 DR. BRAWLEY:  Go ahead with the second




                 DR. D'AGOSTINO:  You don't want to ask


       questions on what they just presented?


                 DR. BRAWLEY:  Does anyone have questions


       on what was just presented?


                 DR. D'AGOSTINO:  I have a question.


                 DR. BRAWLEY:  Oh, go ahead, I am sorry.  I


       misunderstood you.


                 DR. D'AGOSTINO:  What if Alimta was not


       effective at all, and it was just the


       post-chemotherapy of the crossover that gave these


       individuals an increased survival? I don't think


       there is an interpretation that they just gave us,


       but there is another interpretation that is just as


       viable, that the crossover is adding quite a bit to


       the--it's not the third line--it's the second-line





                 The other thing is that I am concerned


       with really Dr. Cohen's presentation where he


       showed that those who didn't get the added


       chemotherapy on the prognosis basis looked pretty


       good, and it is hard for to me to understand that


       the third line isn't helpful, yet, the ones who


       didn't get any added to their line, that some are


       crossovers, some aren't doing as well.


                 I don't really want to make a big


       statistic discussion out of it, because I agree 100


       percent that we are beyond statistics, it is just


       that it does raise a question about how to deal


       with this type of data.


                 DR. EMERSON:  Could I address your second


       question just slightly.  Performance status, we got


       identical results essentially in the time variant


       covariate, if I adjusted for a time variant


       performance status, as well, in this trial.


                 DR. PAOLETTI:  As regards the question


       about efficacy, it's a point like progression-free


       survival, time to progression of disease where



       there is no effect of crossover, the results are


       identical, as well as response rate.


                 Dr. Bunn.


                 DR. BUNN:  I think if Alimta had no effect


       in the early analysis for time to progression, we


       would have seen a difference, and we would have


       seen a survival difference if it didn't have any


       effect.  We probably would have seen a response


       rate different, and we probably would have seen a


       patient reported outcome difference if it didn't


       have any effect.


                 DR. D'AGOSTINO:  I mean that was an


       extreme statement I made.  The point is that it may


       be it is not as effective as, and it is the added


       boost of the chemotherapy, the second- or the


       third-line chemotherapy that makes the difference.


       I don't see how you can sort that out from the




                 DR. BUNN:  I would just like to comment


       about, you know, giving third-line therapy.  You


       know, we are oncologists and we generally like to


       offer therapy where it might be effective, and I



       think that most of our patients would prefer to get


       treatment where it would be effective.


                 There does come a time when neither the


       patient nor the physician is anxious to give


       chemotherapy.  Usually, that is in people who are


       quite ill.  Sometimes they are ill and show up as a


       performance status, but sometimes they have been


       beat up by chemotherapy and they don't have


       sufficient blood counts, or they have neuropathy,


       or they have many other things that would preclude.


                 It is hard to imagine, to me, that the


       physicians would have a bias in the third-line


       setting about treating or not treating patients.


       As a doctor, I find that hard to believe.


                 DR. D'AGOSTINO:  I didn't say anything


       about bias.


                 DR. PAOLETTI:  Dr. Shepherd.


                 DR. SHEPHERD:  Just one further point.  I


       think the point that Dr. Cohen made showing us how


       many good performance status patients do not get


       chemotherapy underlines the belief of the lung


       cancer treating oncologist that third-line



       chemotherapy is not beneficial.


                 When you have no evidence from historical


       data to suggest a survival benefit, when you have a


       response rate less than 3 percent, the potential


       for toxicity is higher than the potential for


       benefit, so clinical practice on the whole is not


       to offer chemotherapy.


                 You don't want to make a performance


       status zero or 1 patient, performance status 3 or 4


       with toxicity, if you don't have a good chance of




                 DR. NGUYEN:  Maybe another clarification


       on this point.  Binh Nguyen, Eli Lilly, Oncology


       Platform Team.


                 I would like to address Dr. D'Agostino's


       questions.  471, please.  Out of those performance


       status that were shown by Dr. Cohen, actually, the


       patient who would perform zero and 1 and alive at


       one month after discontinuation is only half, so


       not all those 139, 169 could receive chemotherapy,


       so you have to take that into consideration and


       look at the difference between the two arms.  A



       drop now is not 30 patients, it is only 12




                 So, it is obvious these patients actually


       die very quickly, that is why they couldn't receive


       post-chemotherapy even thought they had a


       performance status of zero and 1.  I think these


       data are very important.


                 DR. D'AGOSTINO:  I think this is the type


       of discussion I was hoping to hear in terms of


       responses, why are they looking so good, are they


       really dying or not dying.  The group actually


       again, even though there is this discussion that we


       heard, the ones who did not get the second shot out


       at the third-line chemotherapy do not do as well,


       and it is just not clear to me yet that there is an


       obvious reason that one can see on that.


                 DR. BRAWLEY:  Dr. D'Agostino, did you have


       a second question?


                 DR. D'AGOSTINO:  I asked a second


       question.  That was about the sample size.


                 DR. BRAWLEY:  Dr. Mortimer.


                 DR. PAOLETTI:  Actually, you were



       referring to the non-inferiority design, et cetera.


       Again, we acknowledge that the historical data at


       the beginning, when we designed this trial, were


       limited to the TAX 317.


                 However, as Dr. Bunn was showing at the


       conclusion of his presentation, additional


       historical data, additional data were growing


       during all this year, and most importantly, the


       results from our trial in 288 patients are


       confirming the performance of the TAX 317.


                 I would like to ask Dr. Don Berry to


       answer the question from a statistical point of




                 DR. BRAWLEY:  I think we need to move on.


                 DR. MORTIMER:  I have two sort of


       questions.  One relates to a comment Dr. Shepherd


       just made. I mean is it possible to ferret out the


       patients who were on the docetaxel arm who might


       have actually refused therapy because of the risk


       of hospitalization since they were hospitalized


       more often.


                 Secondly, is there a difference in



       patterns of relapse in these arms, specifically,


       brain metastases?


                 DR. PAOLETTI:  Not to my knowledge, but I


       will ask--no, actually.


                 DR. BRAWLEY:  Dr. Perry.


                 DR. PERRY:  Thank you.  I have a question


       for Dr. Pazdur.  Did the study proponents run this


       proposal through the FDA, was it approved by the


       FDA before it was actually set into place?


                 DR. PAZDUR:  I would have to check if it


       had a special protocol assessment.  Obviously, it


       was discussed with the sponsor, the design of the


       trial.  Whether or not there was a special protocol


       assessment, I would have to check on that.


                 DR. PERRY:  The issue to me is there is a


       lot of criticism of the protocol design,


       particularly about the crossover, and if the