MEMORANDUM           

 

DATE:                                                 April 2218, 2005

 

TO:                                                      FDA Antiviral Advisory Committee Members/Guests

                                                                                                                                               

FROM:                                                Tipranavir Review Team (HFD-530)

                                                           

THROUGH:                                         Mark Goldberger, MD, MPH

                                                            Director, Office of Drug Evaluation IV

                                                            Debra Birnkrant, MD

                                                            Director, Division of Antiviral Products

 

 

This briefing document provides background information for the May 19, 2005 Antiviral Drugs Advisory Committee meeting on tipranavir (TPV). On this day, the committee will be asked to consider efficacy and safety data submitted to support the accelerated approval of TPV administered with low dose ritonavir (RTV, r) and provide comments on the risk-benefit analysis of the use of this drug product given the following challenging issues:

 

1)      Design/analyses of efficacy in studies of “heavily pretreated,” HIV-infected individuals 

2)      Impact of resistance on treatment response 

       3)  Management of known and potential TPV/r drug-drug interactions

       4)  TPV/r safety concerns including liver toxicity, lipid abnormalities, rash (particularly in women) and HIV clinical events and mortality

1)Design/analyses of the efficacy in studies of “heavily pretreated” populations 

1)Impact of resistance information

      3)  Management of known and potential drug-drug interactions

      4)  Safety concerns including liver and lipid monitoring/management, rash and gender   differences, and clinical events on study including mortality

 

TPV is a non-peptidic inhibitor of the HIV protease that inhibits viral replication by preventing the maturation of viral particles.  The Applicant submitted NDA 21-814 (tipranavir) 250 mg Capsules on December 22, 2005 seeking approval for marketing under accelerated approval regulations: 21 CFR 314.510 Subpart H.  Under the current guidance for HIV treatment, the basis for approval will be based upon surrogate endpoint analyses of plasma HIV RNA levels for primary efficacy balanced with safety analyses in controlled studies up to 24 weeks duration. 

 

 

I.          SUMMARY OF EFFICACY AND SAFETY DATA

 

Efficacy: Two open-label, multi-center Phase 3 trials (RESIST 1 and 2) submitted in support of this NDA provide evidence of the antiviral effect of TPV over currently available antiretroviral regimens in a population which are “heavily pretreated” (3 class antiretroviral experienced with a median number of 12 prior antiretroviral drugs), and infected with a high level of resistant virus at baseline (97% of the isolates were resistant to at least one PI, 95% to at least one NRTI, and >75% to at least one NNRTI).  The Applicant submitted 24-week efficacy data on all 620 subjects in RESIST 1 and 539 out of 863 subjects in the RESIST 2.  In both RESIST trials combined,

 87% of the subjects were possibly/definitely resistant to the assigned comparator protease inhibitor (CPI).  Thus, although these pivotal trials are presented as TPV/r + optimized background regimen (OBR) versus CPI/r + OBR, in actuality, the results should be interpreted more as TPV/r versus a partially active control with both arms utilizing a large variety of OBR (n = 161 different drug combinations as per FDA statistical analysis) necessitating a superiority efficacy anaylsis.

Thus, although these pivotal trials are presented as TPV/r + optimized background regimen (OBR) versus CPI/r + OBR, in actuality, the results should be interpreted more as TPV/r versus suboptimal control with both arms utilizing a large variety of OBR (n = 161 different drug combinations as per FDA statistical analysis) necessitating a superiority efficacy analysis. 

 

The primary efficacy endpoint was the proportion of subjects with confirmed 1 log₁₀ RNA drop from baseline at week 24 without evidence of treatment failure.  The trial was designed with an escape clause to allow subjects in the comparator arm with a lack of initial virologic response at week 8 to discontinue the RESIST trials and receive TPV in a rollover safety trial.  Lack of initial virologic response was defined as no drop in viral load > 0.5 log₁₀ and failure to achieve a viral load of <100,000 copies/mL during the first 8 weeks of treatment despite a > 0.5 log₁₀ drop.  Subjects who discontinued treatment due to lack of initial virologic response in the comparator arm were considered as treatment failures at week 24, which largely accounted for the treatment difference between the two arms in the primary efficacy endpoint.  The initial virologic treatment difference (24%) between the two arms  at week 8 explains the virologic treatment difference (20%) between the two arms at week 24. 

These same discontinued subjects in the comparator arm were considered as treatment failures at week 24 largely accounting for the treatment difference in the primary efficacy endpoint.  The initial virologic treatment difference (24%) between the two arms (95% CI for the difference in proportions of 18%, 29%) shown at week 8 explains the virologic treatment difference (20%, 95% CI of 15%, 24%) between the two arms at week 24.   

 

For all-cause mortality the numbers of on-treatment deaths (15 TPV/r versus 13 CPI/r) were similar between the two arms.  The added virologic benefit (as measured by the surrogate of plasma HIV RNA) did not translate into any reduction in mortality at the 24 week time-point.  .  These results may be explained by the fact thatHowever, these studies were not powered for mortality, the 24 week time-point ismay be too premature to see any clinical endpoint differences, and/or the comparator arm’s escape clauseoption option at week 8 may have salvaged subjects prior to prolonged virologic failure.  The relationship of plasma HIV RNA as surrogate endpoints to the actual clinical outcomes may be less well understood in studies of heavily pretreated populations.  In addition, the open-label design of the RESIST trials  as well as the comparator arm’s escape clause for lack of initial virologic response by 8 weeks make it somewhat difficult to discern treatment differences in some efficacy and safety parameters beyond 8 weeks of treatment.  Lastly, In addition, due to the open-label design of these RESIST trials with the inherent bias as well as the built in escape clause for the comparator arm at 8 weeks after lack of initial virologic response, it is difficult to discern meaningful comparative efficacy data (both virologic and clinical) beyond 8 weeks of treatment. AIDS defining or AIDS progression events were captured in RESIST trials as adverse events only and not specifically abstracted or adjudicated. 

 

Resistance:  Genotypes from 1482 isolates and 454 phenotypes from both studies were submitted for review for the combined RESIST 1 and 2 studies.  The FDA analyses of virologic outcome by baseline genotype resistance showed consistently greater response rates for the TPV/r arm over CPI/r arm across multiple sensitivity analyses.  Both the number and type of baseline PI mutations affected response rates to TPV/r in RESIST 1 and 2.  Virologic response rates in TPV/r-treated subjects were reduced when isolates with substitutions at positions I13, V32, M36, I47, Q58, D60 or I84 and substitutions V82S/F/I/L were present at baseline.  Virologic responses to TPV/r at week 24 decreased when the number of baseline PI mutation was 5 or more.  Subjects taking TPV/r with ENF were able to achieve >1.5 log₁₀ reductions in viral load from baseline out to 24 weeks even if they had 5 or more baseline PI mutations.  Virologic responses to TPV/r decreased in Resist 1 and 2 when the baseline phenotype for TPV was >3.  The most common protease mutations that developed in >20% of isolates from treatment- experienced subjects who failed on TPV/r treatment were L10I/V/S, I13V, L33V/I/F, M36V/I/L V82T, V82L, and I84V.  The resistance profile in treatment-naive subjects has not yet been characterized. 

 

Drug-drug interaction: The drug-drug interaction potential of 500 mg of TPV in combination with 200 mg of ritonavir is extensive.  TPV/r can alter plasma exposure of other drugs and other drugs can alter plasma exposure of TPV/r.  The known and potential interactions between TPV/r and other HIV medications are listed in Table 12 on Page 21-23.  The table also describes the potential for interactions with other classes of drugs.

 

·   Administration of TPV/r can increase plasma concentrations of agents that are primarily metabolized by CYP3A, because TPV/r is a net inhibitor of CYP3A.

 

·   The applicant did not evaluate the effect of TPV/r on substrates for enzymes other than CYP3A.  In vitro studies indicate TPV is an inhibitor of CYP1A2, CYP2C9, CYP2C19 and CYP2D6.  Due to the known effect of RTV on CYP2D6, the potential net effect of TPV/r is CYP2D6 is inhibition. The net effect of TPV/r on CYP1A2, CYP2C9 and CYP2C19 is not known.

 

·   In vivo data suggest that the net effect of TPV/r on P-glycoprotein is induction.  Based on current data, it is difficult to predict the net effect of TPV/r on oral bioavailability and plasma exposure of drugs that are dual substrates of CYP3A and P-gp. 

 

·   TPV is a CYP3A substrate as well as a P-gp substrate. Therefore, co-administration of TPV/r and drugs that induce CYP3A and/or P-gp may decrease TPV plasma concentrations and reduce its therapeutic effect. Conversely, co-administration of TPV/r and drugs that inhibit P-gp may increase TPV plasma concentrations and increase or prolong its therapeutic and adverse effects.  Co-administration of TPV/r and drugs that inhibit CYP3A may not further increase TPV plasma concentrations, based on the results of a submitted mass balance study.

Administration of TPV/r can increase plasma concentrations of agents that are primarily metabolized by CYP3A, because TPV/r is a net inhibitor of CYP3A.   The Applicant did not evaluate the effect of TPV/r on substrates for enzymes other than CYP3A.  In vitro studies indicate TPV is also an inhibitor of CYP1A2, CYP2C9, CYP2C19 and CYP2D6.  Due to the known effect of RTV on CYP2D6, the potential net effect of TPV/r is CYP2D6 is inhibition. The net effect of TPV/r on CYP1A2, CYP2C9 and CYP2C19 is not known.  In vivo data suggest that the net effect of TPV/r on P-glycoprotein is induction.  Based on current data, it is difficult to predict the net effect of TPV/r on oral bioavailability and plasma exposure of drugs that are dual substrates of CYP3A and P-gp.  TPV is a CYP3A substrate as well as a P-gp substrate. Therefore, co-administration of TPV/r and drugs that induce CYP3A and/or P-gp may decrease TPV plasma concentrations and reduce its therapeutic effect. Conversely, co-administration of TPV/r and drugs that inhibit P-gp may increase TPV plasma concentrations and increase or prolong its therapeutic and adverse effects.  Co-administration of TPV/r and drugs that inhibit CYP3A may not further increase TPV plasma concentrations, based on the results of a submitted mass balance study.

 

Safety Issues:  A safety concern throughout the TPV drug development program has been hepatotoxicity.  Initial signals were observed throughout the 18 Phase 1 studies in healthy volunteers.  A total of 36 (5.5%) healthy HIV-negative subjects experienced treatment emergent grade 3 or 4 liver abnormalities (rise in ALT) in the Phase 1 studies.   The Phase 2 dose-finding study 1182.52 showed that ALT increases were TPV dose dependent. The proportions of patients who had grade 3/4 ALT increases in three treatment arms, TPV/r 500/100 mg , TPV/r 500/200mg, and TPV/r 750/200mg , were 4%, 11%, and 23%, respectively.  The higher proportion of ALT abnormalities on the TPV/r 750 /200 mg arm compared to the TPV/r 500/200 mg arm probably resulted from increased TPV concentrations because RTV exposure was actually lower in the TPV/r 750/200 mg arm than in the TPV 750/200 mg arm.  In addition, detailed exposure-response analyses on Study 1182.52 indicate that ALT increases are associated with increased TPV exposures.Initial hepatotoxicity signals were observed throughout the 18 Phase 1 studies in healthy volunteers.  A total of 36 (5.5%) healthy HIV-negative subjects experienced treatment emergent grade 3 or 4 liver abnormalities (rise in ALT) in the Phase 1 studies.   Results from the Phase 2 dose-finding study 1182.52 indicated that the ALT increases were TPV dose dependent. The proportions of subjects who had grade 3/4 ALT increases in three treatments, TPV/r 500/100 mg , TPV/r 500/200mg , and TPV/r 750/200mg , were 4.3%, 11.1%, and 23%, respectively. The ALT abnormality comparison between treatment of TPV/r  500/200 mg and TPV/r 750 /200 mg suggested that the increased transaminase elevations in the TPV/r 750/200 mg arm most likely resulted from increased TPV exposures instead of RTV, because RTV exposure was lower in the TPV/r 750/250 mg.  Further exposure-response analyses on study 1182.52 indicated that the ALT increases were associated with increased TPV exposures and not RTV exposures.

 

In the RESIST trials, 10% of subjects on the TPV/r arm compared to 3% on the CPI/r arm developed treatment emergent grade 3 or 4 ALT or AST elevations.  For RESIST 1, time to first DAIDS Grade 3 or 4 ALT elevation (p=0.0028) was significantly different between the two arms with subjects in the TPV/r arm more likely to develop Grade 3 or 4 elevations in ALT and at a significantly faster rate than those in the CPI/r arm.  For RESIST 2, time to first Grade 3 or 4 ALT elevation (p=0.0255) was significantly shorter for subjects in the TPV/r arm compared those for subjects in the CPI/r arm. Very few subjects had documented concurrent symptoms; however, at the time of data submission, a substantial number of subjects (~50%) had not resolved their LFT elevations, and therefore, no conclusions can be made about the acute clinical impact of these laboratory abnormalities.  At this time, FDA exploratory analyses examining the possible baseline risk factors for hepatotoxicity (i.e. baseline CD4 counts, hepatitis co-infection, gender, or race) are ongoing.

 

More subjects in the TPV/r arm developed Grade 3 or 4 laboratory lipid abnormalities than those in the CPI/r arm and at a significantly faster rate.  For combined RESIST 1 and 2 datasets, 21% of subjects developed treatment emergent grade 3 or 4 triglycerides compared to 11% of subjects on the CPI/r arm. Analyses of RESIST 1 laboratory data showed that the time to first Grade 3 or 4 in total cholesterol (p=0.0007) or triglycerides (p=0.0186) were significantly different between the two arms.  Analyses of RESIST 2 laboratory data showed that the time to first Grade 3 or 4 elevation in total cholesterol (p=0.0255) or triglycerides (p<0.0001) were shorter for subjects in the TPV/r arm. 

 

The significant differences in the frequency of Grade 3 or 4 lipid or transaminase elevations  between the TPV/r and CPI/r arms may be due to differences in follow-up between the two arms. The escape clause in these studies resulted in a differential duration of randomized treatment exposure and laboratory monitoring between the two arms. On the other hand, it is important to keep in mind many subjects randomized to the CPI/r arms (13%) already had a long duration of exposure to the CPI drug because they entered the study and continued on their current PI. 

The significant differences in developing DAIDS Grade 3 or 4 elevations in liver or lipid laboratory between TPV/r and CPI/r regimens may be due at least in part to the differences in the lengths of follow-up between the two arms. For example in RESIST 1, a median of 24.1 weeks in laboratory tests for triglycerides was obtained for subjects in the TPV/r arm, significantly greater than a median of 19.8 weeks in the CPI/r arm.  Again, the RESIST trials’ open-label trial design with an escape clause resulted in differential drug exposure duration between TPV/r versus CPI/r study arms. On the other hand, it is important to keep in mind that there were subjects enrolled into the CPI/r arms (13%) who already had a large exposure to the CPI drug because they entered the trial and continued on their current PI. 

 

Cutaneous reaction (adverse event of “rash”) was another safety event of special interest in this review due to a substantial Phase 1 signal from an oral contraceptive study in healthy HIV negative women (Study 1182.22).   Seventeen subjects (33%) developed a rash while receiving TPV.  This high and unexplained incidence of rash in healthy, female volunteers raised the possibility that gender and immune status may have an impact on the frequency and types of adverse events (AEs) observed with TPV/r use. 

Other phase 1 trials in healthy HIV-negative volunteers showed that rash was seen in 14/390 (3.6%) males as compared to 34/265 (13%) females.  In Phase 2 trials of HIV infected subjects, one large study (1182.51) showed a rash rate of 10.2% (32/315).  Rash was only reported in males but the study population was 93% male.  In another large phase 2 study (1182.52), 8.6% (18/216) of subjects in the study developed treatment-emergent rash.  Dose relation was suggested because there were 10 subjects who developed rash in TPV/r 750/200 mg group, including one discontinuation, whereas there were 5 subjects in the TPV/r 500/200 mg group and 3 subjects in the TPV/r 500/100 mg group.  Relationship of the development of rash to an intact immune system (as indicated by preserved CD4 cell counts) could not be examined in these two large Phase 2 studies because these subjects were heavily pretreated and advanced in HIV disease with median CD4 cell count of 133 (1182.51) and 178 (1182.52).  Phase 2 trials enrolled predominantly males: however of the limited data available, females on the TPV/r in phase 2 trials had higher incidence of rash (15/114 or 13.2%) as compared to males (59/745 or 7.9%).

                                                                                                                            

In the phase 3 RESIST trials, the overall incidence of rash was similar in both arms (11% TPV/r versus 10% CPI/r).  The severity and need for treatment were also similar between the two arms.  Since the RESIST trial population was immunologically depleted, adequate exploration of the immune-mediated rash was limited.  An exploratory analysis of females in the RESIST trials (n=118 TPV/r; n=90 CPI/r) showed that the females on the TPV/r arm had a higher incidence of rash (14%) as compared to the females on the CPI/r arm (9%).  However, the small number of women in these trials made it impossible to draw any definitive conclusions.  Although BI is currently conducting a study in antiretroviral naïve subjects, the study is already fully enrolled with only about 20% of female subjects (similar to the RESIST trials) and based on baseline CD4+ count, viral load and AIDS defining illnesses, these naïve subjects have advanced HIV disease.  Therefore, it appears unlikely that the current naïve trial will provide definitive answers to whether or not TPV/r affects women and/or immunocompetent subjects differently than the remainder of the HIV+ population.

 

Mortality: One hundred and two subjects died during the entire TPV clinical development program up through the database lock on June 11, 2004.  In total, 12 subjects died during the pretreatment phase and 90 subjects died after being exposed to at least one dose of drug (post-drug exposure).  For most deaths, subjects had advanced HIV disease and multiple concomitant medications.  Three of the 90 post-drug exposure subject deaths were considered to be possibly TPV/r treatment related by the Applicant.  However, FDA could not rule out relatedness or a possible contribution of the effects of TPV in most death cases.  This unclear ascertainment of study drug’s relationship to mortality (and to morbidity) is due to the nature of the population under study, and in many cases, was due to the lack of available information surrounding the death cases.

 

Overall, there were more deaths in RESIST 1 than in RESIST 2 (22 versus 11), and there were more deaths on the TPV/r arms compared to the CPI/r arms (19 versus 14).  The observed virologic benefit of the TPV/r over CPI/r did not translate to better mortality outcome at the 24 week time-point.  However, the RESIST trials were not designed to assess clinical endpoints.  The escape clause at 8 weeks precluded optimal evaluation of longer term clinical efficacy and safety.

 

In order to place the numbers of deaths in the TPV program in perspective, mortality rates were examined from the in the NDA databases of all “treatment-experienced” trials which led to approval of an antiretrovirals.  The population enrolled in the enfuvirtide (ENF) phase 3 studies most closely approximated the TPV phase 3 studies.   Comparison of the frequency of deaths and mortality rates (MR, #death/100 patient years) between the test and control arms were relatively similar for both the TPV and ENF NDAs at 24 weeks as summarized below:

 

·         TPV vs. CPI:                      2% (4.5 MR)  vs. 1.2% (2.6 MR)

·         ENF vs no treatment:           1.5% (3.3 MR) vs.1.5% (3.3 MR) 

 

Based on the information as summarized above summary, we would like the committee’s feedback on the issues outlined in section II.  The remaining sections of this background document provides greater detail on the efficacy, safety, resistance profile, and clinical pharmacology of TPV/r. 

From the archives of DAVDP, these analyses showed that the population enrolled (http://www.fda.gov/cder/foi/nda/2003/021481_fuzeon_review.htm) in the enfuvirtide (ENF) phase 3 studies most closely approximated the TPV phase 3 studies.   Comparison of % frequency of deaths or mortality rates (MR, #death/100 subject years) between the test and control arms were relatively similar for both the TPV (2% vs. 1.2% or 4.5 MR vs. 2.6 MR) and ENF (1.5% vs. 1.5% or 3.3 MR vs. 3.3 MR) NDAs at 24 weeks. 

 

 

II.        ISSUES FOR COMMITTEE DISCUSSION

 

·         The risk/benefit assessment of TPV/r given the data provided for safety and efficacy in the treatment of “heavily pretreated” HIV-infected individuals. 

 

·         Appropriate safeguards for the use of TPV/r given the limited inclusion criteria of the RESIST trials, TPV/r drug-drug interactions, the impact of resistance on response and the safety considerations outlined above.

 

·         Display of TPV/r resistance data/analyses in the TPV package insert that would be useful to clinicians.

 

·         Monitoring and management of hepatotoxicity during clinical use of TPV/r given the transaminase elevations data in healthy volunteers and HIV-infected patients in the development program.

 

·         Further investigation and characterization of the safety signal of rash in females in the TPV program given the limited available data in HIV-infected females. 

 

·         Lessons learned from the TPV drug development program regarding the study of heavily pretreated HIV-infected individuals including:

 

o        Need for drug-drug interaction and resistance data

o        Use of open-label study designs

o        Use of escape clauses resulting in a diminishing comparator arm

o        Need for better adjudication of clinical events (i.e. treatment-emergent AIDS progression events) and need for comprehensive data collection for serious adverse events including death

o        Increasing female participation in HIV drug trials

II. ISSUES FOR THE COMMITTEE DISCUSSION

 

·The risk/benefit assessment of TPV/r given the data provided for safety and efficacy in the treatment of previously “heavily pretreated” HIV infected population. 

 

·Appropriate safeguards for the use of TPV/r given the limited inclusion criteria of the RESIST trials, the drug-drug interactions, the resistance information and the safety considerations.

 

·Display of TPV/r resistance data/analyses in the TPV package insert that would be useful to the clinician.

 

·Monitoring and management of hepatotoxicity during clinical use of TPV/r given the transaminase elevations data in healthy volunteer studies, dose-response/dose-exposure studies, and both RESIST trials.

 

·Further investigation and characterization of the safety signal of rash in females in the TPV program given the limited available data in HIV-infected females. 

 

·Discussion of increasing female participation in HIV drug trials in general.

 

·Lessons learned from the TPV drug program regarding the study of heavily pretreated HIV population which includes the

oNeed for drug-drug interaction and resistance data

oOpen-label study design with inherent bias

oEscape clause with loss of comparator arm

oNeed for better adjudication of clinical events (i.e. treatment-emergent AIDS progression events) and need for comprehensive data collection for serious adverse events including death

This briefing document provides background information for the May 19, 2005 Antiviral Drugs Advisory committee meeting on tipranavir. On this day, the committee will be asked to consider efficacy and safety data submitted to support the accelerated approval of tipranavir for the treatment of HIV infection in the “heavily pretreated” HIV-infected adult population.

 

Tipranavir (TPV) is a non-peptidic inhibitor of the HIV protease that inhibits viral replication by preventing the maturation of viral particles.  The applicant submitted NDA 21-814 (tipranavir) 250 mg Capsules on December 22, 2005 seeking approval for marketing under accelerated approval regulations: 21 CFR 314.510 Subpart H.  Under the current guidance for HIV treatment, the basis for approval will be based upon surrogate endpoint analyses of plasma HIV RNA levels for primary efficacy balanced with safety analyses in controlled studies up to 24 weeks duration. 

 

I. SUMMARY OF EFFICACY AND SAFETY DATA

 

The FDA analyses of the submitted NDA data thus far are consistent with the applicant’s overall findings.  Two open-label, multi-center Phase 3 trials (RESIST 1 and 2) submitted in support of this NDA provide evidence of the additional antiviral effect of TPV over currently available antiretroviral regimens in a population which are “heavily pretreated” ( 3 class antiretroviral experience with median number of prior therapy at 12 drugs).  Overall at baseline, 97% of the isolates were resistant to at least one PI, 95% of the isolates were resistant to at least one NRTI, and >75% of the isolates were resistant to at least one NNRTI.  It is important to note that close to 90% of comparator protease inhibitors (CPI) exhibited resistance at baseline to the clinical isolates.   Thus, although these pivotal trials are being presented as TPV/r + Optimized background regimen (OBR) versus CPI/r + OBR, in actuality, the results should be interpreted more as TPV/r versus placebo with both arms utilizing a large variety of OBR (n = 161 different drug combinations as per FDA statistical analysis).  TPV/r showed significantly greater treatment effect than CPI/r when subjects were already possibly or definitely resistant to their treatment CPIs.  There was no significant effect of TPV/r over CPI/r if the subjects were sensitive to their CPI. 

 

The added antiviral benefit of the TPV arm over the comparator arm was mainly the effect of the lack of initial virologic response* in the comparator arm measured at week 8.  This measured benefit of the TPV arm over the comparator arm at week 8 was sustained at week 24 based upon the composite endpoint** largely due to those same comparator subjects with initial lack of virologic response being discontinued from study (rolling over to a TPV safety study) and being considered treatment failures at week 24.   The initial virologic treatment difference (24%) between the two arms shown at week 8 explains the virologic treatment difference (20%) between the two arms at week 24.  Again, this virologic treatment difference was only measured over comparator PI regimens which were possibly/definitely resistant.  TPV/r did not offer added antiviral benefit over CPI/r for subjects in the comparator arm who were sensitive to their PIs.  Moreover, using all-cause mortality as a definitive clinical event in these trials (AIDS-defining events were captured in these trials as adverse events only and not separately captured or adjudicated), it is worthy of note that the number of on-treatment deaths (15 TPV/r versus 13 CPI/r) were similar between the two arms.  The added virologic benefit (as measured by the surrogate of plasma HIV RNA) did not translate into any reduction in mortality at the 24 week time-point.  These results may be explained by the fact that these studies were not powered for mortality and the 24 week time-point is too premature to see any clinical endpoint differences.  It is worthy of note however that the use of plasma HIV RNA as a surrogate endpoint in clinical trials of antiretrovirals was examined in populations who were treatment-naïve or early experienced.  The use of viral surrogates in studies of the current heavily pretreated population is an extrapolation with unmeasured harms or benefits not yet well understood.  Moreover, due to the open-label nature of these RESIST trials with all the inherent bias as well as the built in escape clause for the comparator arm at 8 weeks after lack of initial virologic response, it is difficult to discern meaningful comparative efficacy data (both virologic and clinical) beyond 8 weeks of treatment.

 

* defined as Lack of Initial Virologic Response by Week 8: proportion of subjects with

1) Viral load has not dropped 0.5 log₁₀ during the first 8 weeks of treatment  and 2) Failure to achieve a viral load of <100,000 copies/mL during the first 8 weeks of treatment, despite a 0.5 log₁₀ drop after 8 weeks of treatment.

 

**defined as Composite endpoint at 24 weeks:  proportion of subjects with 1) confirmed 1 log RNA drop from baseline and 2) without evidence of treatment failure

 

One important subgroup analyses was virologic response in subjects with concomitant enfurvitide (T-20) use which improved virologic response for both arms. When T-20 was added to TPV/r, the treatment effect was greater than if T-20 was not used (net treatment effect of 29.4% vs 15.6%, respectively, for T-20 users versus non-use of T-20).  The concomitant use of T-20 in the RESIST trials also illustrates an example of how post-randomization bias enters into open-label trials.  For TPV/r randomized subjects, 9 additional subjects who did not have T-20 pre-specified in their OBR received T-20 post-randomization.  Conversely for CPI/r randomized subjects who did have T-20 pre-specified in their OBR, 9 subjects did not ultimately receive their specified T-20.

 

Genotypes from 1482 isolates and 454 phenotypes from both studies were submitted for review for the combined RESIST 1 and 2 studies.  The FDA analyses of virologic outcome by baseline resistance showed consistently greater response rates for TPV/r arm over control across multiple sensitivity analyses. The most common protease mutations that developed in >20% of isolates from treatment- experience subjects who failed on TPV/r treatment were L10I/V/S, I13V, L33V/I/F, M36V/I/L V82T, V82L, and I84V.  The resistance profile in treatment-naive subjects has not yet been characterized.  Both the number and type of baseline PI mutations affected response rates to TPV/r in RESIST 1 and 2.  Virologic response rates in TPV/RTV-treated subjects were reduced when isolates with substitutions at positions I13, V32, M36, I47, Q58, D60 or I84 and substitutions V82S/F/I/L were present at baseline.  Virologic responses to TPV/r at week 24 decreased when the number of baseline PI mutation was 5 or more.  Subjects taking enfuvirtide with TPV/r were able to achieve >1.5 log₁₀ reductions in viral load from baseline out to 24 weeks even if they had 5 or more baseline PI mutations.  Virologic responses to TPV/r decreased in Resist 1 and 2 when the baseline phenotype for TPV was >3. 

 

The drug-drug interaction potential of 500 mg of TPV in combination with 200 mg of ritonavir is extensive.   TPV/r can affect other drugs and other drugs can affect TPV/r. TPV is a CYP 3A inhibitor, as well as a CYP3A inducer. TPV/r is a net inhibitor of the CYP3A.  TPV/r may therefore increase plasma concentrations of agents that are primarily metabolized by CYP3A and could increase or prolong their therapeutic and adverse effects. Studies in human liver microsomes indicated TPV is an inhibitor of CYP1A2, CYP2C9, CYP2C19 and CYP2D6.  The potential net effect of TPV/r is CYP2D6 is inhibition. The net effect of TPV/r on CYP1A2 and CYP2C9 is not known. Data are not available to indicate whether TPV inhibits or induces glucuronosyl transferases.  Tipranavir is a P-glycoprotein (P-gp) substrate, a weak P-gp inhibitor, and likely a potent P-gp inducer as well. Data suggest that the net effect of TPV/r is P-gp induction at steady-state.  Based on the current limited data, it is difficult to predict the net effect of TPV/r on oral bioavailability of drugs that are dual substrates of CYP3A4 and P-gp.   TPV is a CYP3A substrate as well as a P-gp substrate. Therefore, co-administration of TPV/r and drugs that induce CYP3A and/or P-gp may decrease TPV plasma concentrations and reduce its therapeutic effect. Conversely, co-administration of TPV/r and drugs that inhibit P-gp may increase TPV plasma concentrations and increase or prolong its therapeutic and adverse effects.  Co-administration of TPV/r and drugs that inhibit CYP3A may not further increase TPV plasma concentrations based on the results of a submitted mass balance study.

 

TPV/r has established or potential drug-drug interactions with multiple antiretroviral drugs including zidovudine, didanosine, abacavir, delavirdine, amprenavir, lopinavir, and saquinavir as well as the other protease inhibitors (indinavir, nelfinavir, atazanavir).  In addition, antiarrhythmics, antihistamines, antimycobacterials (rifampin), ergot derivatives, GI motility agents (cisapride), herbal products (St. John’s wort), HMG CoA reductase inhibitors (lovastatin, simbastatin), neuroleptics, and sedatives/hypnotics are contraindicated and not recommended for co-administration with TPV/r.  Other drugs  which may be used concomitantly in the HIV population and exhibit established or potential important drug-drug interactions are antacids, antidepressants (SSRIs, atypicals), antifungals (fluconazole, itraconazole, ketoconazole, voriconazole), anticoagulant (warfarin), anti-diabetic agents, antimycobacterials (rifabutin), macrolides (clarithromycin, azithromycin), calcium channel blockers (felodipine, nifedipine, nicardipine), corticosteroid (dexamethasone), HMG-CoA reductase inhibitors (atorvastatin), narcotic analgesics (methadone, meperidine), oral contraceptives/Estrogens (ethinyl-estradiol), despiramine, theophylline, and disulfiram/ methronidazole,

 

A safety concern throughout the TPV drug development program has been hepatotoxicity.  Initial signals were observed throughout the 18 Phase 1 studies in healthy volunteers.  A total of 36 (5.5%) healthy HIV-negative subjects experienced treatment emergent grade 3 or 4 liver abnormalities (rise in SGPT) in the Phase 1 studies.   Results from the Phase 2 dose-finding study 052 indicate that the SGPT abnormality was TPV  dose dependent. The proportion of patients who had grade 3/4 SGPT abnormality in three treatments: 500mg /100mg tipranavir/ritonovir (TPV/RTV), 500mg /200mg TPV/RTV, and 750mg /200mg TPV/RTV, was 4.3%, 11.1%, and 23%, respectively. The SGPT abnormality comparison between treatment of 500mg/200 mg TPV/RTV and 750 mg/200 mg TPV/RTV suggested that the increased liver toxicity in the higher TPV arm most likely resulted from increased TPV exposure instead of RTV, because RTV exposure was lower in the arm with higher liver toxicity.  Logistic regression analysis also suggested that when TPV trough concentration doubles, the odds of having grade 3/4 SGPT abnormality was increased by 96%.  Detailed exposure response analysis on this Study 052 indicated that the SGPT abnormality was associated with TPV exposure. The likelihood that RTV contributes to the SGPT abnormality was small. 

 

In the RESIST trials, 10% of subjects on the TPV/r arm compared to 3% on the CPI/r arm developed treatment emergent grade 3 or 4 ALT or AST elevations. 

For RESIST 1, time to first DAIDS Grade 3 or 4 in ALT (p=0.0028) and Gamma GT (p=0.0002) were significantly different between the two arms with subjects in the TPV/r arm more likely to develop Grade 3 or 4 elevations in ALT and Gamma GT as well as at a significantly faster pace than those in the CPI/r arm.  For RESIST 2, time to first Grade 3 or 4 in ALT (p=0.0255) and Gamma GT (p<0.0001) were significantly shorter for subjects in the TPV/r arm compared those for subjects in the CPI/r arm.  Again, subjects in the TPV/r arm were more likely to develop DAIDS Grade 3 or 4 in liver enzymes and at a faster pace than those in the CPI/r arm. 

 

The significant differences in developing Grade 3 or 4 toxicity and in change from baseline laboratory test measurements between CPI/r and TPV/r regimens may be due at least in part to the significant difference in lengths of follow-up period between the two arms. For example for RESIST 1,  a mean of 21.8 weeks (std=5.7 weeks) and a median of 24.1 weeks in laboratory tests for triglycerides were obtained for subjects in the TPV/r arm, significantly greater than a mean of 18.9 weeks (std=6.8 weeks) and a median of  19.8 weeks in the CPI/r arm.  Again, the current open-label study design with an escape clause for this highly pretreated population resulted in differential drug exposure duration between TPV/r versus CPI/r study arms from the start of the trial to the cut-off time-point of safety comparisons at 24 weeks and beyond.  On the other hand, it is important to keep in mind that there were subjects enrolled into the CPI/r arms (13%) who already had a large exposure to the CPI drug because they entered the study with an already failing regimen.  

 

The relationship (and time-course) of these liver enzyme elevations with symptomatic clinical disease manifestation was difficult to ascertain.  For possible baseline risk factors of outcome, Grade 3 or 4 transaminase elevations on the TPV/r arm were associated with higher baseline median CD4+ counts (238.5 cells/mm³ versus 175 cells/mm³) as compared to the general TPV/r population.  The numbers of subjects were too small to draw any conclusions about the risk factors of viral hepatitis co-infection, gender, or race.

 

In regards to lipid abnormalities measured in the RESIST trials, TPV/r is consistent with what has been generally observed as an important safety concern regarding the PI class.  Analyses of RESIST 1 laboratory data showed that the time to first Grade 3 or 4 in total cholesterol (p=0.0007) and triglycerides (p=0.0186) were significantly different between the two arms.  Analyses of RESIST 2 laboratory data showed that the time to first Grade 3 or 4 in total cholesterol (p=0.0255) and triglycerides (p<0.0001) were  significantly shorter for subjects in the TPV/r arm. More subjects in the TPV/r arm developed Grade 3 or 4 total cholesterol and triglycerides than those in the CPI/r arm and at a significantly faster pace. For combined RESIST 1 and 2 datasets, 21% of subjects developed treatment emergent grade 3 or 4 triglycerides compared to 11% of subjects on the CPI/r arm.  Clinically at the 24 week time-point, none of the subjects with grade 3 or 4 triglycerides on either arm went on to have documented clinical pancreatitis.

 

Cutaneous reactions (adverse event incidence of “rash”) was another safety event of special interest in this review due to a substantial Phase 1 signal from an oral contraceptive study in healthy HIV negative women (study 022).   Seventeen subjects (33%) developed a rash while receiving TPV and 20% had musculoskeletal pain.  Three subjects had both skin and musculoskeletal findings.  An additional three subjects reported symptoms that can be associated with drug hypersensitivity while receiving TPV; one had generalized pruritis and conjunctivitis on day 11, one had conjunctivitis on day 11, and the other had intermittent numbness and tingling in the leg on day 11.  Therefore in the most conservative analysis, 51% of these healthy subjects had possible drug hypersensitivity. FDA’s review of other supportive studies as well as the RESIST studies for this safety signal was focused on examining possible gender differences, immunologically based skin reactions, and/or sulfa-related effect (TPV is a sulfonamide).

Other phase 1 trials in healthy HIV-negative volunteers showed that rash was seen in 14/390 (3.6%) males as compared to 34/265 females (13%).  In Phase 2 trials of HIV infected subjects, one large study (051) showed a rash rate of 10.2% (32/315).  These subjects were all males since the study population was 93% male.  In the subset of subjects identified with a history of sulfa rash (n=58), a higher % of subjects (17%) developed a hypersensitivity-like rash within the first 6 weeks.  In another large phase 2 study (052), 8.6% (18/216) of subjects in the study developed treatment-emergent rash.  Dose relation was suggested because there were 10 subjects who developed rash in 750TPV/200 RTV mg group, including one discontinuation, whereas there were 5 subjects in the 500/200 mg group and 3 subjects in the 500/100 mg group.  Relationship of the development of rash to an intact immune system (as indicated by preserved CD4 cell counts) could not be examined in these two large Phase 2 studies because these subjects were heavily pretreated and advanced in HIV disease with median CD4 cell count of 133 (study 051) and 178 (052).  Phase 2 trials enrolled predominantly males: however of the limited data available, females on the TPV/r in phase 2 trials had higher incidence of rash ( 15/114 or 13.2%) as compared to males (59/745 or 7.9%).

In the phase 3 RESIST trials, the overall incidence of rash was similar on both arms (11% TPV/r versus 10% CPI/r).  The severity and need for treatment were also similar between the two arms.  Three subjects on the TPV/r arm compared to zero on the CPI/r arm ended up discontinuing study treatment due to their rash.  Since the RESIST trial population was a clinically advanced and immunologically depleted, examination of immunologically-mediated rash (or drug hypersensitivity) adverse reactions was limited.  Sulfa-allergic subjects were not excluded in these trials and ------------------------- A subgroup analysis of the females in the Resist trials (n=118 TPV/r; n=90 CPI/r) revealed that the females on the TPV/r arm had a higher incidence of rash (14%) as compared to the females on the CPI/r arm (9%).  Seven of the 17 subjects on the TPV/r had no baseline CD4+ count recorded, so FDA can not make an accurate assessment of the immunologic status of these women. 

 

A total of 103 death cases representing 102 patients died during the entire TPV clinical development program up through the database locking of pivotal studies 1182.12 and 1182.48 on June 11, 2004.  In total,12 subjects died during the pretreatment phase and 90 subjects died after being exposed to at least one dose of drug (post-drug exposure).  Three of the 90 post-drug exposure subject deaths were considered to be possibly TPV/r treatment related by the applicant.  Subject 521394 from the rollover study 1182.17 died of acute renal failure, but the subject had a history of chronic renal disease and was on a number of potentially nephrotoxic agents.  Subject 121025 from the rollover study 1182.17 died of multi-system organ failure including hepatic failure.  The subject had a history of fatty live disease and was taking other potentially hepatotoxic medications at the time of death.  Subject 215 in study 1182.6 died from respiratory failure and brain stem infarction subsequent to developing elevated liver enzymes and lactic acidosis.  For most death cases, subjects had advanced HIV disease and multiple concomitant medications.  Although only these three cases are described here, relatedness or possible contribution of the effects of TPV to the death events could not be ruled out by the FDA reviewers for almost all death cases.  This unclear ascertainment of study drug’s relationship to mortality (and to morbidity) is due to the nature of the population under study, and in many cases, was due to the lack of available information surrounding the death cases.

 

Overall there are more deaths in Resist 1 than in Resist 2 (22 versus 11), and there are more deaths on the TPV/r arms compared to the CPI/r arms (19 versus 14).  In Resist 1 there are two major differences between the two arms: 1. The number of deaths on the TPV/r arm over the CPI/r arm (14 versus 8, p-value = 0.19), and 2. the TPV/r arm had a lower median baseline and last CD4+ count as compared to the CPI/r arm (baseline CD4: 13.75 versus 149; last CD4: 13 versus 158).    Certainly, the observed virologic benefit of the TPV/r over CPI/r did not translate to better mortality outcome at the 24 week time-point.  The importance of examining the relationship between the virologic effect and clinical outcome in this evolving heavily pretreated population is paramount.  Unfortunately for the RESIST trials as currently designed, the comparative efficacy or safety database is less than optimal after 8 weeks of study and the limitations worsen over time due to the large discontinuations of subjects in the comparator arm.

 

In order to place the numbers of deaths in the TPV program in perspective, mortality rates in the NDA database of all “treatment-experienced” trials which led to approval of an antiretroviral from the archives of DAVDP was examined.   This analyses showed that the population enrolled in the T-20 phase 3 studies most closely approximated the TPV phase 3 studies.   Comparison of % frequency of deaths or mortality rates (MR, #death/100 patient years) between the test and control arms were relatively similar for both the TPV (2% vs. 1.2% or 4.5 MR vs. 2.6 MR) and T-20 (1.5% vs. 1.5% of 3.3 MR vs. 3.3 MR) NDAs at 24 weeks.  

 

The Division is convening this meeting to solicit the committee’s comments on the breadth of the proposed treatment indication and the risk-benefit analysis of the use of tipranavir administered with low dose ritonavir given the following challenging issues:

1)Design/analyses of the efficacy in studies of “heavily pretreated” population 

1)Impact of resistance information

      3) Management of known and potential drug-drug interactions

      4) Safety concerns including liver and lipid monitoring/management, rash and gender differences, and clinical events on study including mortality.

 

 

III.       DESIGN/ANALYSES OF THE EFFICACY IN STUDIES OF “HEAVILY             PRETREATED” POPULATION 

 

A.     Study Design of  Phase 3 Trials

 

Please see Appendix I for discussion of dose selection for RESIST trials..

 

RESIST 1 (1182.12) and RESIST 2 (1182.48), were multi-center, multi-national, randomized and controlled, open-label studies in highly treatment-experienced HIV-infected subjects with triple antiretroviral class (NRTI, NNRTI, and PI) experience and with at least two failed PI-based regimens.  The two major differences between the RESIST trials was 1) RESIST 1 was conducted in the United States, Canada and Australia, while RESIST 2 was conducted in Europe and Latin America; and 2) RESIST 1 performed 24 week interim analyses while RESIST 2 performed 16 week interim analyses.  For the accelerated approval application, the Applicant submitted 24-week efficacy data on all 620 subjects in RESIST 1 study and 539 out of 863 subjects in the RESIST 2 study who were able to reach 24 weeks.  The safety and efficacy of TPV/r 500 mg/200 mg was compared through 24 weeks of treatment against a control group of other protease inhibitors boosted with RTV (comparator PI/r or CPI/r) where the control PIs were genotypically determined.  The studies were designed to continue through 96 weeks.  Genotypic resistance testing was done at screening, and as protocol defined, subjects were required to have at least one primary PI mutation(s) at codons 30N, 46I/L, 48V, 50V, 82A/F/L/T, 84V, or 90M and have no more than two protease mutations at codons 33, 82, 84, or 90. 

 

Subjects were randomized 1:1 to either TPV/r or the comparator PI/r group and stratified with respect to pre-selected protease inhibitor (PI) as well as use of ENF.  Both treatment groups (TPV/r versus CPI/r) were designed to receive OBR regimen based on genotypic resistance testing prior to randomization.  Due to the complex comparator treatment group containing various protease inhibitors, the studies had to be designed as open-label trials.  Furthermore, the FDA review team strongly recommended that the studies be designed to test for superiority of efficacy of TPV/r versus CPI/r, since testing for non-inferiority against multiple control groups in such an experienced population would be uninterpretable.  A schematic of the RESIST trials shows the complexity of the study design of these trials (Appendix II). As shown in the schema, the subjects who had a lack of initial virologic response by Week 8 in the control arm of comparator protease inhibitors were allowed to enroll into the roll-over Study 1182.17 where all subjects would receive TPV/r.  This escape clause for subjects in the control group has complicated our ability to interpret the efficacy of TPV/r in a controlled fashion beyond 8 weeks of treatment.

 

 

 

B.     Baseline demographics and disease characteristics in RESIST trials

 

Baseline characteristics of subjects enrolled in these studies are summarized below.

Table 1:  Baseline Characteristics: Studies 1182.12 and 1182.48

	RESIST 1 (012)	RESIST 2 (048)
# of Subjects Randomized	630	880
# of Subjects Treated	620	863
Age (Years)    Mean    Median    Range	45 44 24, 80	43 42 17, 76
Sex (%)    Male    Female	91 9	84 16
Race (%)    Caucasian    Black    Asian    Missing	77 22 1 0	68 5 1 26
Weight (kilograms)    Mean    Median    Range	76 75 35, 151	69 68 32, 118
CD4 Cell Count (cells/mm3)    Mean    Median    Range	164 123 0.5, 1183.5	224 189 1.5, 1893
HIV RNA (log10 copies/mL)    Mean    Median    Range   Proportions w/ HIV RNA (copies/mL) < 10,000 >=10,000 to <100,000 ≥ 100,000	4.7 4.8 2.0, 6.3       16% 43% 41%	4.8 4.8 2.9, 6.8       15% 49% 36%
Stage of HIV Infection (CDC Class)    Class A    Class B    Class C	24% 73% 3%	17% 80% 3%
Protease Inhibitor Stratum    APV    IDV    LPV    SQV	14% 4% 61% 21%	40% 3% 38% 20%
Genotypic Resistance to Pre-selected Protease Inhibitor    Not Resistant    Possible Resistance    Resistant	8% 35% 57%	20% 6% 74%
Actual use of ENF    Yes    No	36% 64%	12% 88%

 

C.     Primary Efficacy Endpoints

 

The primary efficacy endpoint in the RESIST trials is the proportion of subjects with a treatment response at 48 weeks (≥ 1 log₁₀ reduction from baseline HIV RNA in two consecutive measurements without prior evidence of treatment failure).  The efficacy endpoint for the 24-week data submitted in this application is the proportion of subjects with a treatment response at 24 weeks.  Multiple secondary analyses were performed for each study.

 

This efficacy analysis is models the FDA analysis of time to loss of virologic response (TLOVR) analysis which is an intent-to-treat analysis that examines endpoints using the following definitions of treatment response and treatment failure for subjects who have achieved a confirmed 1 log₁₀ drop in HIV RNA from baseline.

 

Treatment response is defined by confirmed virologic response (two consecutive viral load measurements ≥1 log₁₀ below baseline) without prior treatment failure, i.e., occurrence of any of the following events: death, permanent discontinuation of the study drug, loss to follow-up, introduction of a new ARV drug to the regimen for reasons other than toxicity or intolerance to a background ARV drug, and confirmed virologic failure (defined as 1) viral load of <1 log₁₀ below baseline confirmed at two consecutive visits >2 weeks apart, following a confirmed virologic response of two consecutive viral load measurements ≥1 log₁₀ below baseline, or  2) one viral load of <1 log₁₀ below baseline followed by permanent discontinuation of the study drug or loss to follow up, following a confirmed virologic response of two consecutive viral loads ≥1 log₁₀ below baseline.)

 

According to the study design, investigators were allowed to switch subjects in the control arm of boosted CPI/r after 8 weeks of treatment if they had initial lack of virologic response (defined as 1) viral load has not dropped 0.5 log₁₀ during the first 8 weeks of treatment and 2) failure to achieve a viral load of <100,000 copies/mL during the first 8 weeks of treatment, despite a 0.5 log₁₀ drop after 8 weeks of treatment.

 

 

D.     Study Design Issues

 

The open-label design of the RESIST trials was unavoidable because of the choice of various CPIs in the control arm (LPV, IDV, SQV, APV—boosted with low-dose ritonavir).  Additionally, due to the choice of the control group, the studies must be evaluated for superiority of TPV/r over the CPIs to which the majority of the subjects have documented drug resistance at baseline. 

 

The open-label design poses a number of challenges in evaluation of efficacy.  Both RESIST trials were conducted in subjects with very limited treatment options for whom TPV represented a potential and possibly the only option.  Therefore subjects who met the same failure criteria or experienced similar toxicity or safety events may act differently depending on the treatments they received: TPV subjects will be more likely to elect to remain in the same treatment group despite problems whereas control group subjects will be more likely to switch to TPV through the roll-over trial 1182.17.  This creates a potential bias in efficacy assessments if we regard all switches or discontinuations as failures.

 

To address this open-label bias issue, we used the protocol-defined failure criteria—of initial lack of virologic response—at Week 8 to supplement the analysis.  In other words, all subjects who met the failure criteria at Week 8, regardless of whether they switched treatments, were considered failures for the Week 24 evaluation in the FDA analysis. 

 

Another bias that was introduced by the open-label design of RESIST trials was the ability to change the pre-determined OBR.  Subjects were required to have a pre-determined background regimen at the time of randomization based on their genotypic resistance test results and background ARV medication history.  In RESIST 1 and RESIST 2 trials, there were a total of 11% and 14%, respectively of subjects whose pre-determined OBR was different from the actual background regimen received.  One example of this bias is the number of subjects who had ENF predetermined as part of their OBR (TPV/r 165 versus CPI/r 159) differed from the number of subjects who actually took ENF (TPV/r 166 versus CPI/r 134).  The TPV/r arms had a net gain of 1 subject using ENF although it was not predetermined, while the CPI/r arm had a net loss of 25 subjects who did not actually use ENF although it was part of their predetermined background.  The Applicant believes, and DAVDP concurs, that the RESIST Investigators likely wanted to save ENF for use with a known active PI, and therefore, once subjects were randomized to the CPI/r the Investigator changed the OBR to exclude ENF.   In addition, due to the high total number of combinations of ARVs in the OBRs (161), it was also difficult to examine the treatment effect by ARV regimen.  This analysis might have helped determine the clinical effect of TPV drug-drug interactions.

 

The Applicant had difficulty enrolling the RESIST trials as designed to compare TPV/r to an active CPI/r, so they amended the protocol to allow subjects with no available sensitive PI, as per their genotype, to enroll.  This amendment resulted in complete enrollment of the RESIST trials; however, most of the CPI/r subjects entered the trial already genotypically resistant to their assigned PI (92% of subjects in RESIST 1 and 80% of subjects in RESIST 2 had possible or full resistance to the pre-selected PIs).  Therefore, the CPI/r arm is not truly an active control arm, but a suboptimal control arm.  The results of the RESIST studies should be interpreted as TPV/r versus suboptimal control, and the studies must be evaluated for superiority of TPV/r over the CPIs.

 

 

E.     HIV RNA Results

 

Tables 2 and 3 show the primary efficacy results for TPV on the proportion of subjects with treatment response (confirmed 1 log₁₀ reduction in HIV RNA from baseline without prior evidence of treatment failure).  This was based on the time-to-loss of virologic response (TLOVR) algorithm as defined in the primary efficacy endpoint. 

 

In each RESIST trial, the proportion of treatment responders were significantly higher in the TPV/r treated group versus the subjects in the CPI/r treated group (RESIST 1:  36% TPV/r versus 16% CPI/r; RESIST 2:  32% TPV/r versus 13% CPI/r.

 

As explained above, in order to address the bias due to an open-label study design, the FDA analysis treated all subjects who showed an initial lack of virologic response by Week 8 (that is no 0.5 log₁₀ drop in HIV RNA during first 8 weeks of treatment and failure to achieve viral load <100,000 copies/mL) as treatment failures.  We believe that the FDA analysis differs from the Applicant’s results primarily due to this group of subjects who had initial lack of virologic response during first 8 weeks.  These subjects would be most likely to discontinue the study drug later, roll-over to Study 1182.17 to receive TPV, or add additional background ARV drugs.

 

 

 

 

 

 

Table 2: RESIST Outcome at Week 24:  FDA Analysis (TLOVR)

	RESIST 1 Trial 1182.12		RESIST 2 Trial 1182.48		Total
	TPV/r	CPI/r	TPV/r	CPI/r	TPV/r	CPI/r
	n (%)	n (%)	n (%)	n (%)	n (%)	n (%)
Total treated	311 (100)	309 (100)	271 (100)	268 (100)	582 (100)	577 (100)
Treatment response at Week 24	112 (36)	49 (16)	86 (32)	34 (13)	198 (34)	83 (14)
No confirmed 1 log10 drop from baseline	172 (55)	234 (76)	143 (53)	223 (83)	315 (54)	457 (79)
Initial Lack of Virologic Response by Week 8	109 (35)	166 (54)	97 (36)	176 (66)	206 (35)	342 (59)
Rebound	40 (13)	40 (13)	28 (10)	26 (10)	68 (12)	66 (11)
Never suppressed through Week 24	23 (7)	28 (9)	18 (7)	21 (8)	41 (7)	49 (8)
Added ARV drug	20 (6)	21 (7)	35 (13)	8 (3)	55 (9)	29 (5)
Discontinued while suppressed	1 (<1)	2 (1)	4 (1)	1 (<1)	5 (1)	3 (1)
Discontinued due to adverse events	3 (1)	1 (0)	3 (1)	2 (1)	6 (1)	3 (1)
Discontinued due to other reasons	3 (1)	2 (1)	0 (0)	0 (0)	3 (1)	2 (0)
Consent withdrawn	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)	0 (0)
Lost to follow-up	1 (<1)	1 (<1)	0 (0)	0 (0)	1 (<1)	1 (<1)
Non-compliant	0 (0)	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)
Protocol violation	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)	0 (0)
Source:  FDA Statistical Reviewer’s Analysis

 

Table 3:  Proportion of Subjects with Treatment Response

	RESIST 1 – 24 weeks		RESIST 2 – 24 weeks
HIV RNA	TPV/r + OBR n/N (%)	CPI/r + OBR n/N (%)	TPV/r + OBR n/N (%)	CPI/r + OBR n/N (%)
Response Rate (confirmed 1 log10 drop in HIV RNA)	112/311 (36)	49/309 (16)	86/271 (32)	34/268 (13)
Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)	20.2% (13.4%, 26.9%)		19.0% (12.2%, 25.9%)
p-value	<0.001		<0.001
Source:  FDA Statistical Reviewer’s Analysis

 

In the RESIST trials, randomizations were stratified according to the pre-selected protease inhibitors (APV, IDV, LPV, SQV) based on genotypic resistance testing and according to the use of ENF or not.  FDA conducted subgroup analyses based on these stratification factors which are summarized in the tables 4 and 5 below

 

Treatment difference between the TPV/r 500 mg/ 200 mg group and the CPI/r group was statistically significant in both subgroups of the ENF-use strata (used ENF or did not use ENF).  These results were consistent between RESIST 1 and RESIST 2 studies.  In addition, FDA conducted statistical tests to examine interaction between the subgroups on ENF use and treatment group.  A statistically significant treatment interaction was observed for the subgroup of subjects who actually used ENF versus did not use ENF (p-value = 0.02 significant at a=0.15 level). 

 

In other words, in this highly treatment-experienced subject population, the net proportion of subjects with confirmed 1 log₁₀ reduction in HIV-RNA using TPV/r in combination with ENF would be likely to be significantly greater than if TPV/r was used alone without ENF (net treatment effect of 29.4% vs 15.6%, respectively, for ENF users versus non-use of ENF).

Table 4:  Proportion of Subjects with Treatment Response through 24 weeks by ENF use

Both RESIST Trials combined (confirmed 1 log10 drop in HIV RNA from baseline)
Enfuvirtide (ENF) used?	TPV/r N=311	CPI/r N=309	Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)†	Test for treatment effect p-value‡	Test for treatment by subgroup interaction p-value§
Yes (25%)	76/158 (48%)	24/128 (19%)	29.4% (19.0%, 29.7%)	<0.0001	0.02**
No (75%)	122/424 (29%)	59/449 (13%)	15.6% (10.3%, 20.9%)	<0.0001
†      Asymptotic confidence intervals based on normal distribution. ‡      p-value is based on the Mantel-Haenszel chi-square test. §      p-value based on t-test **    Treatment by subgroup interaction is statistically significant at a 0.15 level.
Source:  FDA Statistical Reviewer’s Analysis.

 

 

With regard to the pre-selected comparator protease inhibitor stratum, FDA also conducted analyses to see the treatment effect of TPV/r in the PI strata if subjects were not-resistant to the PI versus possibly/definitely resistant to the comparator PI.  In both RESIST trials combined, only 13% were not resistant to the pre-selected PI stratum, and remaining 87% were possibly/definitely resistant to the comparator PIs.  In the subgroup of subjects for whom the pre-selected PI was not resistant to the HIV, the treatment difference between TPV/r and CPI/r was not consistent between RESIST 1 (US, Canada, Australia) study versus RESIST 2 (the non-US study). The treatment difference between TPV/r and CPI/r (-4.8%) among subjects not resistant to PIs was not statistically significant in RESIST 1 (-4.8%) or in RESIST 2, (15.4%).  Additionally, in RESIST 1, there was a strong treatment by subgroup interaction (p-value = 0.03) between the non-resistant group versus possibly/definitely resistant group, indicating that the treatment effect in non-resistant group was not significant (-4.8%) and in resistant group was significant (~20%).  For both RESIST studies combined, among the subgroup of possibly/definitely resistant comparator PIs, the treatment difference was statistically significant in favor of TPV/r versus CPI/r (treatment effect of ~21%).  The result of this subgroup of subjects with possible/definite resistance to PIs was consistent with the overall results on the primary efficacy endpoint (treatment effect of 19% to 20%). 

 

In summary, TPV/r showed significantly greater treatment effect than CPIs/r only when subjects were possibly or definitely resistant to their CPI/r.  When ENF was added to TPV/r, the treatment effect was even more significantly greater than if ENF was not used.

Table 5: Proportion with Treatment Response through 24 weeks by resistance CPI stratum

RESIST 1
Resistance in PI stratum	TPV/r N=311	CPI/r N=309	Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)†	Test for treatment effect p-value‡	Test for treatment by subgroup interaction p-value§ (Not Resistant versus Possibly or Resistant)
Not Resistant	5/21 (24%)	8/28 (29%)	-4.8% (-29.5%, 19.9%)	0.711	0.03**
Possibly Resistant	47/120 (39%)	18/94 (19%)	20% (8.2%, 31.8%)	0.002
Resistant	60/169 (35%)	23/187 (12%)	23.2% (14.6%, 31.8%)	<0.0001
RESIST 2
Not Resistant	18/55 (33%)	9/52 (17%)	15.4% (-0.1%, 31.5%)	0.0677	0.61
Possibly Resistant	9/15 (60%)	5/18 (28%)	32.2% (-.001%, 64.5%)	0.066
Resistant	59/200 (29%)	20/198 (10%)	19.4% (11.8%, 26.9%)	<0.0001
†      Asymptotic confidence intervals based on normal distribution. ‡      p-value is based on the Mantel-Haenszel chi-square test. §      p-value based on t-test **    Treatment by subgroup interaction is statistically significant at a 0.15 level.
Source:  FDA Statistical Reviewer’s Analysis.

 

 

F.      CD4 Cell Counts

 

At baseline the mean CD4 cell counts in RESIST 1 and RESIST 2 trials were 164 cells/mm³ and 224 cells/mm³, respectively.  FDA conducted an on-treatment analysis to compare the change from baseline in CD4 cell counts between TPV/r and CPI/r groups and determine whether the results would be significantly different if subjects in the CPI/r group were to continue beyond Week 8 rather than discontinue in the CPI/r arm at Week 8.  In general, the CD4 cell counts increased in the TPV/r group through Weeks 2, 4, 8 and 16, and remained stable at Week 24.  The mean increase in CD4 cell counts in the TPV/r group at Weeks 8 and 24 were +50 and +58 cells/ mm³, respectively, for both RESIST studies combined.  The mean increases in CD4 cell counts from baseline in the CPI/r group were modest through Week 8 and were around +20 cells/mm³.  Recall that there were greater numbers of subjects with initial lack of virologic response during the first 8 weeks in the CPI/r group who may have influenced the mean increase in CD4 cell counts. 

 

At Weeks 16 and 24, among the subjects who remained in the RESIST 1 trial with the assigned treatment, the differences between TPV/r group and CPI/r group were no longer statistically significant.  However, in RESIST 2, the difference in mean increase in CD4 cell count at Week 24 was statististically significant, but this difference may not have clinical significance due to the small magnitude of differences.  For both studies combined, the Week 24 mean increase in CD4 cell counts in TPV/r group and CPI/r groups were +58 and +40 cells/mm³, respectively.

 

 

II.                                                                                              DESIGN/ANALYSES OF THE EFFICACY IN STUDIES OF “HEAVILY                                                                                               PRETREATED” POPULATION 

 

A.Study Design of  Phase 3 Trials

 

The two identically designed RESIST trials, namely, RESIST 1 (1182.12) and RESIST 2 (1182.48) were multi-center, multi-national, randomized and controlled, open-label studies in highly treatment-experienced HIV-infected patients with triple antiretroviral class and dual protease inhibitor (dual PI)–drug regimen experience.  The difference between the two studies was that RESIST 1 was conducted in the United States, Canada and Australia, while RESIST 2 was conducted in Europe and Latin America.  Tipranavir boosted with ritonavir (TPV/r  500 mg/200 mg) was compared with respect to safety and efficacy through 24 weeks of treatment against a control group of other protease inhibitors boosted with ritonavir (comparator PI/r or CPI/r) where the control PIs were genotytpically determined.  The studies are designed to continue through 96weeks.

 

Patients were highly antiretroviral treatment-experienced HIV-infected with triple ARV class (NRTI, NNRTI, and PI) experience and dual-PI regimen experience.  Genotypic resistance testing was done at screening in which patients must have at least one primary PI mutation(s) at codons 30N, 46I/L, 48V, 50V, 82A/F/L/T, 84V, or 90M and have no more than two protease mutations 33, 82, 84, or 90. 

 

Patients were randomized equally to either TPV/r or comparator PI/r group and stratified with respect to pre-selected protease inhibitor (PI) as well as use of enfuvirtide (T-20).  Both treatment groups (TPV/r versus CPI/r) were designed to receive optimized background regimen based on genotypic resistance testing prior to randomization.  Due to the complex comparator treatment group containing various protease inhibitors with varying degrees of resistance profiles of the drugs, the studies had to be designed as open-label trials.  Furthermore, the FDA review team strongly recommended the Applicant that the studies be tested for superiority of efficacy of TPV/r versus CPI/r, since testing for non-inferiority against multiple control groups in such an experienced population will be uninterpretable.  A schematic of the RESIST shows the complexity of the study design of these trials. As shown in the schema, the patients who had a lack of initial virologic response by Week 8 (viral load did not drop 0.5 log₁₀ from baseline during the first 8 weeks of treatment and failed to achieve viral load <100,000 copies/mL despite a 0.5 log₁₀ drop) in the control arm of comparator protease inhibitors were allowed to enroll into the roll-over Study 1182.17 where all patients would receive tipranavir (TPV/r).  This escape clause for patients in the control group has complicated our ability to interpret the efficacy of tipranavir beyond 8 weeks of treatment.

 

Figure 1: Schematic of RESIST Trials—Study Design

Figure 1 Continued:

Source:  FDA Statistical Reviewer’s depiction of study design and Protocols 1182.12 (RESIST 1) and 1182.48 (RESIST 2), Volume 1.6 of Module 5

 

 

 

 

 

A.Baseline demographics and disease characteristics in resist trials

 

Baseline characteristics of subjects enrolled in these studies are summarized below.

 

 

Table 1:  Baseline Characteristics: Studies 1182.12 and 1182.48

	RESIST 1 (012)	RESIST 2 (048)
# of Subjects Randomized	630	880
# of Subjects Treated	620	863
Age (Years)    Mean    Median    Range	45 44 24, 80	43 42 17, 76
Sex (%)    Male    Female	91 9	84 16
Race (%)    Caucasian    Black    Asian    Missing	77 22 1 0	68 5 1 26
Weight (kilograms)    Mean    Median    Range	76 75 35, 151	69 68 32, 118
CD4 Cell Count (cells/mm3)    Mean    Median    Range	164 123 0.5, 1183.5	224 189 1.5, 1893
HIV RNA (log10 copies/mL)    Mean    Median    Range   Proportions w/ HIV RNA (copies/mL) < 10,000 >=10,000 to <100,000 ≥ 100,000	4.7 4.8 2.0, 6.3       16% 43% 41%	4.8 4.8 2.9, 6.8       15% 49% 36%
Stage of HIV Infection (CDC Class)    Class A    Class B    Class C	24% 73% 3%	17% 80% 3%
Protease Inhibitor Stratum    …APV    …IDV    …LPV    …SQV	14% 4% 61% 21%	40% 3% 38% 20%
Genotypic Resistance to Pre-selected Protease Inhibitor    …Not Resistant    …Possible Resistance    …Resistant	8% 35% 57%	20% 6% 74%
Actual use of Enfuvirtide (T-20)    …Yes    …No	36% 64%	12% 88%

 

 

 

A.Primary Efficacy Endpoints

 

The primary efficacy endpoint in the RESIST trials is the proportion of patients with a treatment response at 48 weeks (≥ 1 log₁₀ reduction from baseline HIV RNA in two consecutive measurements without prior evidence of treatment failure).  The efficacy endpoint for the 24-week data submitted in this application is the proportion of patients with a treatment response at 24 weeks.  Multiple secondary analyses were performed for each study.

 

This efficacy analysis is designed after the FDA analysis of time to loss of virologic response (TLOVR) analysis which is an intent-to-treat analysis that examines endpoints using the following definitions of treatment response and treatment failure for patients who have achieved a confirmed 1 log₁₀ drop in HIV RNA from baseline.

 

Treatment Response

Treatment response is defined by confirmed virologic response (two consecutive viral load measurements ≥1 log10 below baseline) without prior treatment failure, i.e., occurrence of any of the following events.

1.Death.

1.Permanent discontinuation of the study drug.

1.Loss to follow-up.

1.Introduction of a new ARV drug to the regimen for reasons other than toxicity or intolerance to a background ARV drug.

1.Confirmed virologic failure.

(Confirmed virologic failure is defined as: 

a.Viral load of <1 log10 below baseline confirmed at two consecutive visits >2 weeks apart, following a confirmed virologic response of two consecutive viral load measurements ≥1 log10 below baseline, or

a.One viral load of <1 log10 below baseline followed by permanent discontinuation of the study drug or loss to follow up, following a confirmed virologic response of two consecutive viral loads ≥1 log10 below baseline.)

 

According to the study design, investigators were allowed to switch patients in the control arm of boosted comparator protease inhibitors (CPI/r) after 8 weeks of treatment if they had initial lack of virologic response.  This was defined as follows.

 

Lack of Initial Virologic Response by Week 8

 

1.Viral load has not dropped 0.5 log₁₀ during the first 8 weeks of treatment.

1.Failure to achieve a viral load of <100,000 copies/mL during the first 8 weeks of treatment, despite a 0.5 log₁₀ drop after 8 weeks of treatment.

 

A.Study Design Issues and Data Challenges

 

The open-label design of the RESIST trials was unavoidable because of the choice of various comparator protease inhibitors in the control arm (LPV, IDV, SQV, APV—boosted with low-dose ritonavir).  Additionally, due to the choice of the control group the studies must be evaluated for superiority of TPV/r over the control PIs to which majority of the patients have developed drug resistance. 

 

The open-label design poses a number of challenges in evaluation of efficacy.  Both RESIST trials were conducted in patients with very limited treatment options for many of whom tipranavir represents a potential and possibly the only option.  Therefore patients who are meeting the same failure criteria or experiencing similary toxicity or safety events may act differently depending on the treatments they are receiving: tipranavir patients will be more likely to elect to remain in the same treatment group despite problems whereas control group patients will be more likely to switch to tipranavir through the roll-over trial 1182.17.  This escape clause in the study design creates a potential bias in efficacy assessment if we regard all switches or discontinuations as failures.

 

To address this open-label bias issue, we used the protocol-defined failure criteria—of initial lack of virologic response—at Week 8 to supplement the analysis.  In other words, all patients who met the failure criteria at Week 8, regardless of whether they switched treatments, were considered failures for the Week 24 evaluation in the FDA analysis.

 

Another potential open-label bias may be introduced when patients who were randomized to the control arm wish to be on tipranavir as soon as possible and therefore elect not taking the assigned treatment to meet the failure criteria sooner. We examined the early response pattern of all patients to identify patients who had very little initial improvements. After censoring these patients in the control arm the responses are ...

 

Another bias that could be introduced by the open-label design of RESIST trials was the change in conduct of the study with respect to the use of pre-determined optimized background regimen (OBR).  Patients were required to have a pre-determined background regimen at the time of randomization and based on their genotypic resistance test results and background ARV medication history.  In RESIST 1 and RESIST 2 trials, there were a total of 11% and 14%, respectively, patients whose pre-determined OBR was different from the actual background regimen received.  In addition, patients were changing their background antiretroviral regimen during the so-called optimization period between Week 0 and Week 2 of treatment.  The most commonly used actual background ARV regimen were 3TC+TDF (12%), ddI+TDF (7%), and 3TC+ddI+TDF (7%).  The total number of combinations of actual background antiretroviral drugs in the regimen was 161.

 

In addition to the complexity of the study design and the advanced HIV status/treatment-experience of patients in RESIST trials, the evaluation of electronic data was extremely challenging to the FDA review team.  The NDA submission contained numerous versions and iterations of the raw datasets and analysis datasets for each study with different file structures, ambiguity in naming of variables and/or files, coding of data and little explanation of derived data in the raw data files.  Almost all data files had completely vertical structures with multiple records of different characteristics that made it challenging to discern the meaning of the data transferred from Case Report Forms to Raw Datasets.  For example, numeric and character data were stacked on one another making it uninterpretable and unprogrammable for analysis.  Due to the nature of the highly treatment-experienced patient population and the open-label nature of RESIST trials, it was important for reviewers to examine the conduct of the trial such through a quality check of the pre-determined optimized background regimen and switching of background antiretrovirals during the study.  After numerous weekly communications with the Applicant, FDA reviewers were able to get essential data and obtain clarifications on the data in evaluation of patient disposition, switching of background ARV drugs and protease inhibitors, and primary efficacy endpoint.  Some FDA reviewers had reviewed 5 sets of electronic data submissions on 5 efficacy studies (RESIST 1, RESIST 2,  Studies 1182.52, 1182.51, and 1182.17) with approximately 25 data files per set over a period of 6 months of intense review.  The Applicant assured that the contents of the raw efficacy data in electronic files had not changed but the format of data structure may have changed. 

 

In each submission the Applicant had submitted 24-week efficacy data on all 620 patients in RESIST 1 study and 539 out of 863 patients in the RESIST 2 study.

 

 

A.HIV RNA Results

 

Tables 2 and 3 show the primary efficacy results for tipranavir on the proportion of patients with treatment response (confirmed 1 log₁₀ reduction in HIV RNA from baseline without prior evidence of treatment failure).  This is based on the time-to-loss of virologic response (TLOVR) algorithm as defined in the primary efficacy endpoint.  TLOVR gives an intent-to-treat analysis.

 

In each RESIST trial, the proportion of treatment responders were significantly higher in the TPV/r treated group versus the patients in the CPI/r treated group (RESIST 1:  36% TPV/r versus 16% CPI/r ; RESIST 2:  32% TPV/r versus 13% CPI/r).  It is noteworthy that the comparator protease inhibitor control arm was not a completely active control in this highly treatment-experienced group of patients.  In RESIST 1 92% and in RESIST 2 80% of the patients had possible resistance or full resistance to the pre-selected protease inhibitors.  In comparison, patients in the TPV/r were receiving a new drug that had no resistance to the HIV virus as yet.  Therefore, the control group was an inactive control arm and would be more likely to approximate a placebo group with respect to efficacy.  As such superiority of the TPV/r treatment group over the control arm shows the efficacy and antiviral activity of tipranavir.  It does not necessarily prove that tipranavir is superior to other comparator protease inhibitors if other PIs were not resistant to HIV in a given patient.

 

As explained above, in order to address the bias due to an open-label study design, the FDA analysis treated all patients who showed an initial lack of virologic response by Week 8 (that is no 0.5 log₁₀ drop in HIV RNA during first 8 weeks of treatment and failure to achieve viral load <100,000 copies/mL) as treatment failures.  We believe that the FDA analysis differs from the Applicant’s results primarily due to this group of patients who had initial lack of virologic response during first 8 weeks.  These patients would be most likely to discontinue the study drug later, roll-over to Study 1182.17 to receive tipranavir, or add additional background ARV drugs.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2:
Treatment Outcome at Week 24—RESIST Trials
FDA Analysis (Time to Loss of Virologic Response)

	RESIST 1 Trial 1182.12		RESIST 2 Trial 1182.48		Total
	TPV/r	CPI/r	TPV/r	CPI/r	TPV/r	CPI/r
	nN (%)	nN (%)	nN (%)	nN (%)	nN (%)	nN (%)
Total treated	311 (100)	309 (100)	271 (100)	268 (100)	582 (100)	577 (100)
Treatment response at Week 24	112 (36)	49 (16)	86 (32)	34 (13)	198 (34)	83 (14)
No confirmed 1 log10 drop from baseline	172 (55)	234 (76)	143 (53)	223 (83)	315 (54)	457 (79)
Initial Lack of Virologic Response by Week 8	109 (35)	166 (54)	97 (36)	176 (66)	206 (35)	342 (59)
Rebound	40 (13)	40 (13)	28 (10)	26 (10)	68 (12)	66 (11)
Never suppressed through Week 24	23 (7)	28 (9)	18 (7)	21 (8)	41 (7)	49 (8)
Added ARV drug for any reason	20 (6)	21 (7)	35 (13)	8 (3)	55 (9)	29 (5)
Discontinued while suppressed	1 (<1)	2 (1)	4 (1)	1 (<1)	5 (1)	3 (1)
Discontinued due to adverse events	3 (1)	1 (0)	3 (1)	2 (1)	6 (1)	3 (1)
Discontinued due to other reasons	3 (1)	2 (1)	0 (0)	0 (0)	3 (1)	2 (0)
Consent withdrawn	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)	0 (0)
Lost to follow-up	1 (<1)	1 (<1)	0 (0)	0 (0)	1 (<1)	1 (<1)
Non-compliant	0 (0)	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)
Protocol violation	1 (<1)	0 (0)	0 (0)	0 (0)	1 (<1)	0 (0)
Source:  FDA Statistical Reviewer’s Analysis

 

 

Table 3:
Summary of Efficacy – RESIST Studies
Proportion of Patients with Treatment Response (confirmed 1 log10 drop in HIV RNA from baseline without prior treatment failure)

	RESIST 1 – 24 weeks		RESIST 2 – 24 weeks
HIV RNA	TPV/r + OBR n/N (%)	CPI/r + OBR n/N (%)	TPV/r + OBR n/N (%)	CPI/r + OBR n/N (%)
Response Rate (confirmed 1 log10 drop in HIV RNA)	112/311 (36)	49/309 (16)	86/271 (32)	34/268 (13)
Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)	20.2% (13.4%, 26.9%)		19.0% (12.2%, 25.9%)
p-value	<0.001		<0.001
Source:  FDA Statistical Reviewer’s Analysis

 

 

 

 

HIV RNA Results according to Baseline Characteristics

 

In the RESIST trials, randomizations were stratified according to the pre-selected protease inhibitors (APV, IDV, LPV, SQV) based on genotypic resistance testing and according to the use of enfuvirtide (T-20) or not.  FDA conducted subgroup analyses based on these stratification factors which are summarized in the tables 4 and 5 belows in Appendix.

 

Treatment difference between the TPV/r (500 mg/ 200 mg) group and the low-dose ritonavir boosted comparator protease inhibitor group (CPI/r ) was statistically significant in both subgroups of the enfuvirtide-use strata (used T-20 or did not use T-20).  These results were consistent between RESIST 1 and RESIST 2 studies.  In addition, FDA conducted statistical tests to examine interaction between the subgroups on T-20 use and treatment group.  A statistically significant treatment interaction was observed for the subgroup of patients who actually used T-20 versus did not use T-20 (p-value = 0.02 significant at a=0.15 level). 

 

In other words, in this highly treatment-experienced patient population, the net proportion of patients with confirmed 1 log₁₀ reduction in HIV-RNA using TPV/r in combination with T-20 would be likely to be significantly greater than if TPV/r was used alone without T-20 (net treatment effect of 29.4% vs 15.6%, respectively, for T-20 users versus non-use of T-20).

 

Table 4:
Proportion of Patients with Treatment Response through 24 weeks
(confirmed 1 log10 drop in HIV RNA from baseline)
by enfuvirtide (T-20) use—RESIST 1 and RESIST 2 trials

Both RESIST Trials combined
Enfuvirtide (T-20) used?	TPV/r N=311	CPI/r N=309	Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)†	Test for treatment effect p-value‡	Test for treatment by subgroup interaction p-value§
Yes (25%)	76/158 (48%)	24/128 (19%)	29.4% (19.0%, 29.7%)	<0.0001	0.02**
No (75%)	122/424 (29%)	59/449 (13%)	15.6% (10.3%, 20.9%)	<0.0001
†  Asymptotic confidence intervals based on normal distribution. ‡  p-value is based on the Mantel-Haenszel chi-square test. §  p-value based on t-test **  Treatment by subgroup interaction is statistically significant at a 0.15 level.
Source:  FDA Statistical Reviewer’s Analysis.

 

 

 

With regard to the pre-selected comparator protease inhibitor stratum, FDA also conducted analyses to see the treatment effect of TPV/r in the PI strata if patients were not-resistant to the PI versus possibly/definitely resistant to the comparator PI.

 

In both RESIST trials combined, only 13% were not resistant to the pre-selected PI stratum, and remaining 87% were possibly/definitely resistant to the comparator PIs.  In the subgroup of patients wherefor whom the pre-selected PI was not resistant to the HIV and patients were randomized to either TPV/r or CPI/r, the treatment difference between TPV/r and CPI/r was not statistically significant (p-value=0.199)consistent between RESIST 1 (US, Canada, Australia) study versus RESIST 2 (the non-US study).  In RESIST 1 the treatment difference between TPV/r and CPI/r (-4.8%) among patients not resistant to PIs was not statistically significant and in RESIST 2, the treatment difference  was also not statistically significant but positive (15.4%).  Additionally, in RESIST 1, there was a strong treatment by subgroup interaction (p-value = 0.03) between the non-resistant group versus possibly/definitely resistant group, indicating that the treatment effect in non-resistant group was not significant (-4.8%) and in resistant group was significant (~20%).  For both RESIST studies combined, In among the subgroup of possibly/definitely resistant comparator PIs, the treatment difference was statistically significant in favor of TPV/r versus CPI/r (treatment effect of ~21%).  The result of this subgroup of patients with possible/definite resistance to PIs was consistent with the overall results on the primary efficacy endpoint (treatment effect of 19% to 20%). 

 

In summary TPV/r showed significantly greater treatment effect than other PIs only when patients were possibly or definitely resistant to other comparator protease inhibitors.  When T-20 was added to TPV/r, the treatment effect was even more significantly greater than if T-20 was not used.

 

Table 5:
Proportion of Patients with Treatment Response through 24 weeks
(confirmed 1 log10 drop in HIV RNA from baseline)
by resistance to comparator PI stratum—RESIST 1 and RESIST 2 trials

RESIST 1
Resistance in PI stratum	TPV/r N=311	CPI/r N=309	Difference in proportions (TPV/r – CPI/r) (95% Confidence Interval)†	Test for treatment effect p-value‡	Test for treatment by subgroup interaction p-value§ (Not Resistant versus Possibly or Resistant)
Not Resistant	5/21 (24%)	8/28 (29%)	-4.8% (-29.5%, 19.9%)	0.711	0.03**
Possibly Resistant	47/120 (39%)	18/94 (19%)	20% (8.2%, 31.8%)	0.002
Resistant	60/169 (35%)	23/187 (12%)	23.2% (14.6%, 31.8%)	<0.0001
RESIST 2
Not Resistant	18/55 (33%)	9/52 (17%)	15.4% (-0.1%, 31.5%)	0.0677	0.61
Possibly Resistant	9/15 (60%)	5/18 (28%)	32.2% (-.001%, 64.5%)	0.066
Resistant	59/200 (29%)	20/198 (10%)	19.4% (11.8%, 26.9%)	<0.0001
†  Asymptotic confidence intervals based on normal distribution. ‡  p-value is based on the Mantel-Haenszel chi-square test. §  p-value based on t-test **  Treatment by subgroup interaction is statistically significant at a 0.15 level.
Source:  FDA Statistical Reviewer’s Analysis.

 

 

 

A.CD4 Cell Counts

 

At baseline the mean CD4 cell counts in RESIST 1 and RESIST 2 trials were 164 cells/mm³ and 224 cells/mm³, respectively.  FDA conducted an on-treatment analysis to compare the change from baseline in CD4 cell counts between TPV/r and CPI/r groups and determine whether TPV/r would be significantly different if patients in the CPI/r group were to continue beyond Week 8 rather than discontinue in the CPI/r arm at Week 8.

 

In general, the CD4 cell counts increased in the TPV/r group through Weeks 2, 4, 8 and 16, and remained stable at Week 24.  The mean increase in CD4 cell counts in the TPV/r group at Weeks 8 and 24 were +50 and +58 cells/ mm³, respectively, for both RESIST studies combined.  The mean increases in CD4 cell counts from baseline in the comparator PI group were modest through Week 8 and were around +20 cells/mm³.  Recall that there were greater numbers of patients with initial lack of virologic response during the first 8 weeks in the CPI/r group who may have influenced the mean increase in CD4 cell counts. 

 

At Weeks 16 and 24, among the patients who remained in the RESIST 1 trial with the assigned treatment, the differences between TPV/r group and CPI/r group were no longer statistically significant.  Although in RESIST 2, the difference in mean increase in CD4 cell count at Week 24 was statistically significant between TPV/r and CPI/r group, this difference may not have clinical significance due to the small magnitude of differences.  For both studies combined, the Week 24 mean increase in CD4 cell counts in TPV/r group and CPI/r groups were +58 and +40 cells/mm³, respectively.

 

IV.       Impact of resistance information

 

TPV has 50% inhibitory concentrations (IC₅₀ value) ranging from 40 to 390 nM against laboratory HIV-1 strains grown in vitro in PBMCs and cell lines.  The average IC₅₀ value for multi PI-resistant clinical HIV-l isolates was 240 nM (range 50 to 380 nM). Human plasma binding resulted in a 1.6- to 4-fold shift in the antiviral activity.  Ninety percent (94/105) of HIV-1 isolates resistant to APV, ATV, IDV, LPV, NFV, RTV, or SQV had <3-fold decreased susceptibility to TPV. 

 

Because TPV will be administered to HIV-1 positive subjects in combination with other antiretroviral agents, the activity of TPV in combination with other antiviral drugs was determined in cell culture to assess the impact of potential in vitro drug interactions on overall antiviral activity.  Additive to antagonistic relationships were seen with combinations of TPV with other PIs.   Combinations of TPV and each of the NRTIs were generally additive, but additive to antagonistic for TPV with ddI or 3TC.  Combinations of TPV and DLV or NVP were additive, and TPV with EFV was additive to antagonistic. Activity of TPV with ENF was synergistic. 

 

A.        In Vitro Selection of TPV-Resistant Viruses

TPV-resistant viruses were selected in vitro when wild-type HIV-l_NL4-3 was serially passaged in the presence of increasing concentrations of TPV in tissue culture.  Amino acid substitutions L33F and I84V emerged initially at passage 16 (0.8 mM), producing a 1.7-fold decrease in TPV susceptibility. Viruses with >10-fold decreased TPV susceptibility were selected at drug concentrations of 5 mM with the accumulation of six protease mutations (I13V, V32I, L33F, K45I, V82L, I84V).  After 70 serial passages (9 months), HIV-1 variants with 70-fold decreased susceptibility to TPV were selected and had 10 mutations arising in this order: L33F, I84V, K45I, I13V, V32I, V82L, M36I, A71V, L10F, and I54V.  Mutations in the CA/P2 protease cleavage site and transframe region were also detected by passage 39.  TPV-resistant viruses showed decreased susceptibility to all currently available protease inhibitors except SQV.  SQV had a 2.5-fold reduction in susceptibility to the TPV-resistant virus with 10 protease mutations.

 

B.        Clinical TPV Resistance

The efficacy of TPV/r was examined in treatment-experienced HIV-infected subjects in two pivotal phase III trials, RESIST 1 and 2.  Genotypes from 1482 isolates and 454 phenotypes from both studies were submitted for review.  In the comparator arm (CPI/r), most subjects received LPV/r (n=358) followed by APV/r (n=194), SQV/r (n=162) and IDV/r (n=23).  The subject populations in RESIST 1 and 2 were highly treatment-experienced with a median number of 4 (range 1-7) PIs received prior to study.  In the combined RESIST trials at baseline, 97% of the isolates were resistant to at least one PI, 95% of the isolates were resistant to at least one NRTI, and >75% of the isolates were resistant to at least one NNRTI.  The treatment arms from both studies were balanced with respect to baseline genotypic and phenotypic resistance.  Baseline phenotypic resistance was equivalent between the TPV/r arm (n=745) and the CPI/r arm (n=737) with 30% of the isolates resistant to TPV at baseline and 80-90% of the isolates resistant to the other PIs - APV, ATV, IDV, LPV, NFV, RTV or SQV.  The number of PI-resistance mutations was equivalent between the TPV/r and CPI/r arms in RESIST 1 and 2 and the median number of baseline PI, NRTI and NNRTI mutations was equivalent between arms in both studies (Table 6).

 

 

 

Table 6. Median Number of Mutations at Baseline in RESIST 1 and 2

	FDA PI mut	TPV PI mut	Key PI mut	Primary PI mut	IAS PI mut	NRTI mut	NNRTI mut
TPV/r n = 745	4	3	2	3	9	5	1
CPI/r n = 737	4	3	2	3	9	5	1

FDA PI mut - Number of substitutions at D30, V32, M36, M46, I47, G48, I50, F53, I54, V82, I84, N88, or L90 at baseline

TPV PI mut - Number of tipranavir-specific protease mutations: 10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, or 84V at baseline

Key PI mut - Number of protease mutations at 33, 82, 84, or 90 at baseline

Primary PI mut - Number of primary protease mutations at 30, 33, 46, 48, 50, 82, 84, or 90 at baseline

IAS PI mut - Number of protease mutations at 10, 20, 24, 30, 32, 33, 36, 46, 47, 48, 50, 53, 54, 63, 71, 73, 77, 82, 84, 88, or 90 at baseline

NRTI mut - Number of RT mutations at 41, 44, 65, 67, 69, 70, 74, 115, 118, 184, 210, or 215 at baseline

NNRTI mut - Number of RT mutations at 98, 100, 103, 106, 108, 181, 188, 190, 225, 230, or 236 at baseline

 

C.        Mutations Developing on TPV Treatment

TPV/r-resistant isolates were analyzed from treatment-experienced subjects in Study 1182.52 (n=32) and RESIST 1 and 2 (n =59) who experienced virologic failure.  The most common mutations that developed in greater than 20% of these TPV/r virologic failure isolates were L10I/V/S, I13V, L33V/I/F, M36V/I/L V82T, V82L, and I84V .  Other mutations that developed in 10 to 20% of the TPV/r virologic failure isolates included E34D/R/Q/H, I47V, I54V/A/M, K55R, A71V/I/L/F, and L89V/M/W.  In RESIST 1 and 2, TPV/r resistance developed in the virologic failures (n=59) at an average of 38 weeks with an average decrease of >30-fold in TPV susceptibility from baseline. The resistance profile in treatment-naive subjects has not yet been characterized.

 

D.        Baseline Genotype/Phenotype and Virologic Outcome Analyses

The FDA analyses of virologic outcome by baseline resistance are based on the As-Treated population from studies RESIST 1 and 2.  To assess outcome, several endpoints including the primary endpoint (proportion of responders with confirmed 1 log₁₀ decrease at Week 24), DAVG24, and median change from baseline at weeks 2, 4, 8, 16, and 24 were evaluated.  In addition, because subjects were stratified based on ENF use, we examined virologic outcomes in three separate groups - overall (All), subjects not receiving ENF (No ENF), and subjects receiving ENF (+ENF) as part of the optimized background regimen. We focused on the No ENF group in order to assess baseline resistance predictors of virologic success and failure for TPV/r without the additive effect of ENF use on the overall response.

Both the number and type of baseline PI mutations affected response rates in RESIST 1 and 2.  Virologic responses were analyzed by the presence at baseline of each of 25 different protease amino acids using both the primary endpoint (>1 log₁₀ decrease from baseline) and DAVG24.  Reduced virologic responses were seen in TPV/r-treated subjects when isolates had a baseline substitution at position I13, V32, M36, I47, Q58, D60 or I84 (Table 7).  The reduction in virologic responses for these baseline substitutions was most prominent in the No ENF subgroup. Virologic responses were similar or greater than the overall responses for each subgroup (All, No ENF, +ENF) when these amino acid positions were wild-type. 

 In addition, virologic responses to substitutions at position V82 varied depending on the substitution.  Interestingly, substitutions V82S or F or I or L, but not V82A or T or C, had reduced virologic responses compared to the overall. 

 

Table 7. Effect of Type of Baseline PI Mutation on the Primary Endpoint in Resist 1 and 2.

	TPV/r Arm (n=513)			CPI/r Arm (n=502)
Mutation	All	No ENF	+ENF	All	No ENF	+ENF
Overall	47% (240/513)	40% (147/369)	65% (93/144)	22% (109/502)	19% (75/389)	30% (34/113)
I13V/A/L/S	40% (69/171)	27% (32/119)	69% (37/54)	20% (35/178)	15% (20/133)	33% (15/45)
V32I/L	39% (29/74)	26% (12/46)	61% (17/28)	15% (9/59)	14% (6/43)	19% (3/16)
M36I/A/V/L/N	40% (124/310)	29% (60/208)	63% (64/102)	20% (65/318)	18% (45/345)	27% (20/73)
I47V/A	31% (29/93)	18% (11/62)	58% (18/31)	11% (9/82)	10% (6/63)	16% (3/19)
Q58E	38% (28/74)	27% (14/52)	64% (14/22)	18% (17/93)	18% (14/79)	21% (3/14)
D60E/K/A/N	39% (43/110)	30% (24/79)	61% (19/31)	12% (8/66)	11% (6/53)	15% (2/13)
V82 any change	48% (149/311)	41% (90/222)	66% (59/89)	18% (54/202)	14% (33/236)	32% (21/66)
V82A/T/C	50% (133/264)	45% (85/189)	64% (48/75)	18% (46/259)	13% (27/202)	33% (19/57)
V82S/F/I/L	34% (16/47)	15% (5/33)	79% (11/14)	21% (9/43)	21% (7/34)	22% (2/9)
I84V/A	41% (64/155)	31% (32/103)	62% (32/52)	20% (32/162)	20% (23/115)	19% (9/47)

 

Analyses were also conducted to assess virologic outcome by the number of PI mutations present at baseline.  In these analyses, any changes at protease amino acid positions - D30, V32, M36, M46, I47, G48, I50, I54, F53, V82, I84, N88 and L90 were counted if present at baseline. These PI mutations were used based on their association with reduced susceptibility to currently approved PIs, as reported in various publications. The results of these analyses are shown in Tables 8 and 9.

Regardless of the endpoint used for these analyses, the response rates were greater for the TPV/r treatment arm compared to the CPI/r arm.  In both the TPV/r and CPI/r arms of RESIST 1 and 2, response rates were similar to or greater than the overall response rates for the respective treatment groups for subjects with one to four PI mutations at baseline.  Response rates were reduced if five or more PI-associated mutations were present at baseline.  For subjects who did not use ENF, 28% in the TPV/r arm and 11% in the CPI/r arm had a confirmed 1 log₁₀ decrease at Week 24 if five or more PI mutations were present at baseline (Table 8).  The subjects with five or more PI mutations in their HIV at baseline and not receiving ENF in their OBT achieved a 0.86 log₁₀ median DAVG24 decrease in viral load on TPV/r treatment compared to a 0.23 log₁₀ median DAVG24 decrease in viral load on CPI/r treatment (Table 9).   In general, regardless of the number of baseline PI mutations or ENF use, the TPV/r arm had approximately 20% more responders by the primary endpoint (confirmed 1 log₁₀ decrease at Week 24) (Table 8) and greater declines in viral load by median DAVG24 (Table 9) than the CPI/r arm.

 

 

 

Table 8. Proportion of Responders (confirmed 1 log₁₀ decrease at Week 24) by Number of Baseline PI Mutations

# Baseline FDA PI Mutations	TPV/r N=531			CPI/r N=502
	All	No ENF	+ ENF	All	No ENF	+ ENF
Overall	47% (241/531)	40% (148/369)	65% (93/144)	22% (110/502)	20% (76/389)	30% (34/113)
1-2	70% (30/43)	69% (27/39)	75% (3/4)	44% (19/43)	41% (17/41)	100% (2/2)
3-4	50% (117/236)	44% (78/176)	65% (39/60)	27% (60/221)	23% (39/169)	40% (21/52)
5+	41% (94/231)	28% (43/151)	64% (51/80)	13% (31/236)	11% (20/178)	19% (11/58)

# Any change at positions 30, 32, 36, 46, 47, 48, 50, 53, 54, 82, 84, 88 and 90

 

 

 

Table 9. Median DAVG24 by Number of Baseline PI Mutations

# Baseline FDA PI Mutations	TPV/r N=704			CPI/r N=705
	All	No ENF	+ ENF	All	No ENF	+ ENF
Overall	-1.31 (704)	-1.02 (546)	-1.88 (158)	-0.36 (705)	-0.33 (574)	-0.60 (131)
1-2	-1.43 (76)	-1.44 (69)	-1.42 (7)	-1.13 (65)	-1.01 (63)	-1.90 (2)
3-4	-1.36 (322)	-1.29 (259)	-1.96 (63)	-0.53 (316)	-0.44 (252)	-0.89 (64)
5+	-1.07 (303)	-0.86 (215)	-1.81 (88)	-0.24 (322)	-0.23 (258)	-0.27 (64)

# Any change at positions 30, 32, 36, 46, 47, 48, 50, 53, 54, 82, 84, 88 and 90

 

 

 

An examination of the median change from baseline of HIV RNA at weeks 2, 4, 8, 16 and 24 by number of baseline PI mutations (1-4 and 5+) showed the largest decline in viral load by Week 2 for all groups with the greatest decline observed in the TPV/r arms (Figure 1).  A 1.5 log₁₀ decrease in viral load at Week 2 was observed for subjects receiving TPV/r regardless of the number of baseline PI mutations (1-4 or 5+).  Subjects who had five or more baseline PI mutations and who received TPV/r without ENF began to lose antiviral activity between Weeks 4 and 8 with their HIV RNA trending back toward baseline (Figure 1B).  However, sustained viral load decreases (1.5 – 2 log₁₀) through Week 24 were observed in subjects receiving TPV/r and ENF (Figure 1C).

 

 

 

 

 

 

 

Figure 1. Median Change from Baseline by Number of Baseline PI Mutations

1A. Overall Response

  N@ Week:         0           2                 4               8             16            24

  TPV 1-4        398        378            284           387           365           262

  TPV 5+         303        288            289           297           289           211

  CPI 1-4         381        352            358           363           308           173

  CPI 5+          322        304            312           311           242           110

 

 

 

1B. Response without ENF Use

N@ Week:        0               2                  4               8              16              24

TPV 1-4       328            311             315           318            297            199     

TPV 5+        215            204             201           211            201             136

CPI 1-4        315            291             294           298            254             131    

CPI 5+         258            244             252           249            194               62

 

 

 

 

 

 

1C. Response with ENF Use

N@ Week:       0                2                 4               8              16              24

TPV 1-4        70              67               69             69              68              63

TPV 5+         88              84               85             88              83              75

CPI 1-4         66              61               64             65              54              42

CPI 5+          64              60               60             62              48              28

 

 

E.         Proportion of Responders by Baseline TPV Phenotype

TPV/r response rates were also assessed by baseline TPV phenotype.  Again, we focused on the No ENF group in order to more accurately assess the effect of baseline phenotype on virologic response for TPV/r.  With no ENF use, the proportion of responders was 45% if the fold change in IC₅₀ value from reference of TPV susceptibility was 3-fold or less at baseline (Table 10).  The proportion of responders decreased to 21% when the TPV baseline phenotype values were >3- to 10-fold and 0% when TPV baseline phenotype values were >10-fold.

 

 

Table 10. Proportion of Responders by Baseline TPV phenotype

Baseline TPV Phenotype
	All	No ENF Use	ENF Use
Overall	47% (146/313)	39% (84/218)	65% (62/95)
0-3	54% (120/223)	45% (74/163)	77% (46/60)
>3-10	29% (22/75)	21% (10/47)	43% (12/28)
>10	27% (4/15)	0% (0/8)	57% (4/7)

 

 

 

 

 

 

III.                                                                                                                                                                                           Impact of resistance information

 

Tipranavir (TPV), a protease inhibitor, has 50% inhibitory concentrations (IC₅₀ value) ranging from 40 to 390 nM against laboratory HIV-1 strains grown in vitro in PBMCs and cell lines.  The average IC₅₀ value for multi PI-resistant clinical HIV-l isolates was 240 nM (range 50 to 380 nM). Human plasma binding resulted in a 1.6- to 4-fold shift in the antiviral activity.  Ninety percent (94/105) of HIV-1 isolates resistant to APV, ATV, IDV, LPV, NFV, RTV, or SQV had <3-fold decreased susceptibility to TPV. 

 

Because TPV will be administered to HIV-positive patients as part of a HAART regimen comprising several antiretroviral agents, the activity of TPV in combination with other antiviral drugs was determined in cell culture to assess the impact of potential in vitro drug interactions on overall antiviral activity. Additive to antagonistic relationships were seen with combinations of TPV with other PIs.  Combinations of TPV with the NRTIs were generally additive, but additive to antagonistic for TPV in combination with ddI and 3TC.  Combinations of TPV with DLV and NVP were additive and with EFV were additive to antagonistic. Activity of TPV with enfuvirtide (T20) was synergistic. 

 

A.                                                                                                                                                                                           In Vitro Selection of TPV-Resistant Viruses

TPV-resistant viruses were selected in vitro when wild-type HIV-l_NL4-3 was serially passaged in the presence of increasing concentrations of TPV in tissue culture.  Amino acid substitutions L33F and I84V emerged initially at passage 16 (0.8 mM), producing a 1.7-fold decrease in TPV susceptibility. Viruses with >10-fold decreased TPV susceptibility were selected at drug concentrations of 5 mM with the accumulation of six protease mutations (I13V, V32I, L33F, K45I, V82L, I84V).  After 70 serial passages (9 months), HIV-1 variants with 70-fold decreased susceptibility to TPV were selected and had 10 mutations arising in this order: L33F, I84V, K45I, I13V, V32I, V82L, M36I, A71V, L10F, and I54V.  Mutations in the CA/P2 protease cleavage site and transframe region were also detected by passage 39.  TPV-resistant viruses showed decreased susceptibility to all currently available protease inhibitors except SQV.  SQV had a 2.5-fold change in susceptibility to the TPV-resistant virus with 10 protease mutations.

 

B.                                                                                                                                                                                           Clinical TPV Resistance

The efficacy of TPV/r was examined in treatment-experienced HIV-infected subjects in two pivotal phase III trials, study 012 (RESIST 1) and study 048 (RESIST 2).  Genotypes from 1482 isolates and 454 phenotypes from both studies were submitted for review.

In the comparator arm (CPI), most patients received LPV/RTV (n=358) followed by APV/RTV (n=194), SQV/RTV (n=162) and IDV/RTV (n=23).  The patient populations in RESIST 1 and 2 were highly treatment-experienced with a median number of 4 (range 1-7) PIs received prior to study.  In the combined RESIST trials at baseline, 97% of the isolates were resistant to at least one PI, 95% of the isolates were resistant to at least one NRTI, and >75% of the isolates were resistant to at least one NNRTI.  The treatment arms from both studies were balanced with respect to baseline genotypic and phenotypic resistance.  Baseline phenotypic resistance was equivalent between the TPV/r arm (n=745) and the CPI arm (n=737) with 30% of the isolates resistant to TPV at baseline and 80-90% of the isolates resistant to the other PIs - APV, ATV, IDV, LPV, NFV, RTV or SQV.  The number of PI-resistance mutations was equivalent between the TPV/r and CPI arms in RESIST 1 and 2 and the median number of baseline PI, NRTI and NNRTI mutations was equivalent between arms in both studies (Table A).

 

Table 46. Median Number of Mutations at Baseline in RESIST 1 and 2

	FDA PI mut	TPV PI mut	Key PI mut	Primary PI mut	IAS PI mut	NRTI mut	NNRTI mut
TPV/r n = 745	4	3	2	3	9	5	1
CPI n = 737	4	3	2	3	9	5	1

FDA PI mut - Number of substitutions at D30, V32, M36, M46, I47, G48, I50, F53, I54, V82, I84, N88, or L90 at baseline

TPV PI mut - Number of protease mutations at 10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, or 84V at baseline

Key PI mut - Number of protease mutations at 33, 82, 84, or 90 at baseline

Primary PI mut - Number of primary protease mutations at 30, 33, 46, 48, 50, 82, 84, or 90 at baseline

IAS PI mut - Number of protease mutations at 10, 20, 24, 30, 32, 33, 36, 46, 47, 48, 50, 53, 54, 63, 71, 73, 77, 82, 84, 88, or 90 at baseline

NRTI mut - Number of RT mutations at 41, 44, 65, 67, 69, 70, 74, 115, 118, 184, 210, or 215 at baseline

NNRTI mut - Number of RT mutations at 98, 100, 103, 106, 108, 181, 188, 190, 225, 230, or 236 at baseline

 

 

C.                                                                                                                                                                                           Mutations Developing on TPV Treatment

TPV/r-resistant isolates were analyzed from treatment-experienced patients in Study 052 (n=32) and RESIST 1 and 2 (n =59) who experienced virologic failure.  The most common mutations that developed in greater than 20% of these TPV/r virologic failure isolates were L10I/V/S, I13V, L33V/I/F, M36V/I/L V82T, V82L, and I84V .  Other mutations that developed in 10 to 20% of the TPV/r virologic failure isolates included E34D/R/Q/H, I47V, I54V/A/M, K55R, A71V/I/L/F, and L89V/M/W.  In RESIST 1 and 2, TPV/r resistance developed in the virologic failures (n=59) at an average of 38 weeks with an average decrease of >30-fold in TPV susceptibility from baseline. The resistance profile in treatment-naive subjects has not yet been characterized.

 

D.                                                                                                                                                                                           Baseline Genotype/Phenotype and Virologic Outcome Analyses

The FDA analyses of virologic outcome by baseline resistance are based on the As-Treated population from studies RESIST 1 and 2.  To assess outcome, several endpoints including the primary endpoint (proportion of responders with confirmed 1 log₁₀ decrease at Week 24), DAVG24, and median change from baseline at weeks 2, 4, 8, 16, and 24 were evaluated.  In addition, because subjects were stratified based on enfuvirtide (T20) use, we examined virologic outcomes in three separate groups - overall (All), subjects not receiving T20 (No T20), and subjects receiving T20 (+T20) as part of the optimized background regimen. We focused on the No T20 group in order to assess baseline resistance predictors of virologic success and failure for TPV/r without the additive effect of T20 use on the overall response.

Both the number and type of baseline PI mutations affected response rates in RESIST 1 and 2.  Virologic responses were analyzed by the presence at baseline of each of 25 different protease amino acids using both the primary endpoint (>1log₁₀ decrease from baseline) and DAVG24.  Reduced virologic responses were seen in TPV/r-treated subjects when isolates had a baseline substitution at position I13, V32, M36, I47, Q58, D60 or I84 (Table B).  The reduction in virologic responses for these baseline substitutions was most prominent in the No T20 subgroup. Virologic responses were similar or greater than the overall responses for each subgroup (All, No T20, +T20) when these amino acid positions were wild-type. 

 In addition, virologic responses to substitutions at position V82 varied depending on the substitution.  Interestingly, substitutions V82S or F or I or L, but not V82A or T or C, had reduced virologic responses compared to the overall. 

 

Table 75. Effect of Type of Baseline PI Mutation on the Primary Endpoint in Resist 1 and 2.

	TPV Arm (n=513)			CPI Arm (n=502)
Mutation	All	No T20	+T20	All	No T20	+T20
Overall	47% (240/513)	40% (147/369)	65% (93/144)	22% (109/502)	19% (75/389)	30% (34/113)
I13V/A/L/S	40% (69/171)	27% (32/119)	69% (37/54)	20% (35/178)	15% (20/133)	33% (15/45)
V32I/L	39% (29/74)	26% (12/46)	61% (17/28)	15% (9/59)	14% (6/43)	19% (3/16)
M36I/A/V/L/N	40% (124/310)	29% (60/208)	63% (64/102)	20% (65/318)	18% (45/345)	27% (20/73)
I47V/A	31% (29/93)	18% (11/62)	58% (18/31)	11% (9/82)	10% (6/63)	16% (3/19)
Q58E	38% (28/74)	27% (14/52)	64% (14/22)	18% (17/93)	18% (14/79)	21% (3/14)
D60E/K/A/N	39% (43/110)	30% (24/79)	61% (19/31)	12% (8/66)	11% (6/53)	15% (2/13)
V82 any change	48% (149/311)	41% (90/222)	66% (59/89)	18% (54/202)	14% (33/236)	32% (21/66)
V82A/T/C	50% (133/264)	45% (85/189)	64% (48/75)	18% (46/259)	13% (27/202)	33% (19/57)
V82S/F/I/L	34% (16/47)	15% (5/33)	79% (11/14)	21% (9/43)	21% (7/34)	22% (2/9)
I84V/A	41% (64/155)	31% (32/103)	62% (32/52)	20% (32/162)	20% (23/115)	19% (9/47)

 

Analyses were also conducted to assess virologic outcome by the number of PI mutations present at baseline.  In these analyses, any changes at protease amino acid positions - D30, V32, M36, M46, I47, G48, I50, I54, F53, V82, I84, N88 and L90 were counted if present at baseline. These PI mutations were used based on their association with reduced susceptibility to currently approved PIs, as reported in various publications. The results of these analyses are shown in Tables C and D.

Regardless of the endpoint used for these analyses, the response rates were greater for the TPV/r treatment arm compared to the CPI arm.  In both the TPV/r and CPI arms of RESIST 1 and 2, response rates were similar to or greater than the overall response rates for the respective treatment groups for subjects with one to four PI mutations at baseline.  Response rates were reduced if five or more PI-associated mutations were present at baseline.  For subjects who did not use T20, 28% in the TPV/r arm and 11% in the CPI arm had a confirmed 1 log₁₀ decrease at Week 24 if they had five or more PI mutations in their HIV at baseline (Table C).  The subjects with five or more PI mutations in their HIV at baseline and not receiving T20 in their OBT achieved a 0.86 log₁₀ median DAVG24 decrease in viral load on TPV/r treatment compared to a 0.23 log₁₀ median DAVG24 decrease in viral load on CPI treatment (Table D).   In general, regardless of the number of baseline PI mutations or T20 use, the TPV/r arm had approximately 20% more responders by the primary endpoint (confirmed 1 log₁₀ decrease at Week 24) (Table C) and greater declines in viral load by median DAVG24 (Table D) than the CPI arm.

 

Table 86. Proportion of Responders (confirmed 1 log₁₀ decrease at Week 24) by Number of Baseline PI Mutations

# Baseline FDA PI Mutations	TPV/r N=531			CPI N=502
	All	No T20	+ T20	All	No T20	+ T20
Overall	47% (241/531)	40% (148/369)	65% (93/144)	22% (110/502)	20% (76/389)	30% (34/113)