• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services

Vaccines, Blood & Biologics

  • Print
  • Share
  • E-mail

FW: SAS Programs/Outputs - Provenge, December 15, 2006

(System Info - 126764 Kim Helen 04/22/2010 19:00:05 TULL)
 

From: Kim, Helen [hkim@Dendreon.com]
Sent: Friday, December 15, 2006 4:23 PM
To: Zhen, Boguang (CBER)
Cc: Smith, Liz; Yuh, Lianng; Harmon, Matt; Tull Lori; Liu, Ke; Bross, Peter F
(CBER)
Subject: FW: SAS programs/outputs

Attachments: D9901TimeToObjectDxProg.pdf; g_KM_TmToObjDxProg.sas

Dr. Zhen -

Please find below our response to your December 14, 2006 Questions regarding SAS
Programs/Outputs for Studies D9901 and D9902A.

Please contact Ms. Elizabeth Smith at (206) 829-1556 should you have any
questions.

Regards,

Helen Kim
Sr. Regulatory Affairs Manager
Dendreon Corporation
(P) 206.829.1464
(F) 206.441.4070

.......................................................................................................................................................................


1. Cox model -- for Study 9901

I conducted several additional sensitivity analyses with different set of
covariates in Cox model. In one of the analyses, I used localization of disease
(bone and soft tissue vs. bone only or soft tissue only), PSA (<20, 20-<100,
>=100), and Gleason score (<=6, 7, >=8) as covariates. Could you check the
program (9901Cox.sas) to make sure that I used the right variables/datasets?

[Note: would like to let you know that this is just one of the many sensitivity
analyses -- although p-value for treatment effect from this particular
sensitivity analysis is 0.0788, p-values from many of my other sensitivity
analyses are below the 0.05 level]

Dendreon Response:
You have used the correct datasets and variables for this specific adjusted
analysis.


2. Time to objective disease progression as measured by imaging -- for Study
9901

I tried to repeat your analysis in 11.4.2 Time to objective disease progression
confirmed by imaging studies (p74 of the study report) and tried to generate
Kaplan-Meier curves for this endpoint as suggested by our medical reviewers.
However, I was unable to do so due to failure in identifying the right
variables/datasets. Could you generate Kaplan-Meier curves for us and tell me
which variables you used for this analysis?

Dendreon Response:
We have attached a .pdf file that contains the Kaplan-Meier curve for the
analysis of Time to Objective Disease Progression Confirmed by Imaging Studies.
We have also attached the SAS program that generated this figure. This program
is derived from a macro included in the original D9901 materials named
?kmtable2.sas.? The kmtable2 macro was used to generate Table 14.2.6, the table
supporting section 11.4.2 of the CSR. Both this program and the original macro
require the use of the ?EFFICACY.xpt? dataset previously provided with the D9901
analysis transport files in the August 21, 2006 BLA submission. The variables
used to assign the censoring values are TTIMAGE (label = ?Time to Progression
(Radpharm)?), and IMAGONLY (label = ?Progression Date-Radpharm(Text)?).


3. Survival analysis in Study 9902a

When I run my program for Kaplan-Meier curves and log-rank test for the overall
survival endpoint, I found that the median survivals, CIs, number of events from
my analysis matched with those in the study report (11.4.2.1, page 69 of the
study report) except p-value. P-value = 0.3482 from my analysis while p=0.331
in the study report.

Again, when I tried to duplicate the results in Table 13: Proportional Hazards
Regression Models of Survival (page 72), I could not match my results with those
in the table. However, I could duplicate the results for Cox model using the
same variables/datasets as you recommended for Study 9901.

Could you please check my program (9902a_SURV.sas) to see if the program is
correct? If not, please show me how to duplicate the results.

Dendreon Response:
The program "9902a_SURV.sas" can be used if you incorporate the stratification
variable "---b(4)---" into the model. According the statistical methods for
Study D9902A, the Cox regression models were stratified by --b(4)----. We
believe this should explain the p-value discrepancy.
 



 

***********************************************************************

This email message including any attachments is for the sole use of the intended
recipient(s) and may contain confidential and privileged information. Any
unauthorized review, use, disclosure or distribution is prohibited. If you are
not the intended recipient, please contact the sender by reply email and destroy
all copies of the original message. If you are the intended recipient, please be
advised that the content of this message is subject to access, review and
disclosure by the sender's Email System Administrator.