Issues and Experience in Analyzing Transgenic Mouse Carcinogenicity Studies: An Industry Perspective Ronald Menton Wyeth Research 2005 FDA/Industry Statistics.

Slides:



Advertisements
Similar presentations
High Resolution studies
Advertisements

Planning Reports and Proposals
Multistage Sampling.
HydroQual, Inc. Application of the SWMM Storage-Treatment Block for Analysis/Design of Extended-Detention Ponds Thomas L. Newman II, P.E. Tarig A. Omer.
Chapter 6 Flowcharting.
STATISTICS HYPOTHESES TEST (I)
Effective Change Detection Using Sampling Junghoo John Cho Alexandros Ntoulas UCLA.
ISSUES THAT PLAGUE NON- INFERIORITY TRIALS PAST AND FUTURE RALPH B. DAGOSTINO, SR. BOSTON UNIVERSITY HARVARD CLINICAL RESEARCH INSTITUTE.
Evaluation of Live Phase Results from Carcinogenicity Studies Wherly Hoffman, Ph.D. Statistics and Information Sciences Lilly Research Laboratories.
By: Saad Rais, Statistics Canada Zdenek Patak, Statistics Canada
Gatekeeping Testing Strategies in Clinical Trials Alex Dmitrienko, Ph.D. Eli Lilly and Company FDA/Industry Statistics Workshop September 2004.
Increasing Survey Cooperation: Motivating Chronic Late Responders to an Annual Survey National Science Foundation Division of Science Resources Statistics.
Katherine Jenny Thompson
The Application of Propensity Score Analysis to Non-randomized Medical Device Clinical Studies: A Regulatory Perspective Lilly Yue, Ph.D.* CDRH, FDA,
FDA/Industry Workshop September, 19, 2003 Johnson & Johnson Pharmaceutical Research and Development L.L.C. 1 Uses and Abuses of (Adaptive) Randomization:
1 ESTIMATION IN THE PRESENCE OF TAX DATA IN BUSINESS SURVEYS David Haziza, Gordon Kuromi and Joana Bérubé Université de Montréal & Statistics Canada ICESIII.
NANPA Oversight Working Group Status Report to the NANC April 17, 2001 Chair Pat Caldwell.
Incorporating Historical Control Data when Comparing Tumor Incidence Rates in Rodent Cancer Bioassay Shyamal D. Peddada Biostatistics Branch National Inst.
ActionDescription 1Decisions about planning and managing the coast are governed by general legal instruments. 2Sectoral stakeholders meet on an ad hoc.
EPAA Annual conference November Regulatory acceptance of alternative approaches for pharmaceuticals Jean-Marc Vidal Safety & Efficacy of Human Medicines.
Jeopardy Q 1 Q 6 Q 11 Q 16 Q 21 Q 2 Q 7 Q 12 Q 17 Q 22 Q 3 Q 8 Q 13
Jeopardy Q 1 Q 6 Q 11 Q 16 Q 21 Q 2 Q 7 Q 12 Q 17 Q 22 Q 3 Q 8 Q 13
Lecture 2 ANALYSIS OF VARIANCE: AN INTRODUCTION
1 Contact details Colin Gray Room S16 (occasionally) address: Telephone: (27) 2233 Dont hesitate to get in touch.
Estimates of Survival and Mortality from Successive, Cross-Sectional Surveys David W. Smith, Ph.D., M.P.H. Consultant Stephanie L. McFall, Ph.D. Institute.
Evaluating Provider Reliability in Risk-aware Grid Brokering Iain Gourlay.
Understanding Multiyear Estimates from the American Community Survey Updated February 2013.
SURVIVAL AND LIFE TABLES
1 Understanding Multiyear Estimates from the American Community Survey.
1 Implementing Internet Web Sites in Counseling and Career Development James P. Sampson, Jr. Florida State University Copyright 2003 by James P. Sampson,
Internet Survey Method in the 2010 Census and Challenges to the 2015 Census in Japan Population Census Division, Statistics Bureau of Japan Hideki Koizumi.
Multilevel Event History Modelling of Birth Intervals
Fact-finding Techniques Transparencies
You will need Your text Your calculator
Department of Engineering Management, Information and Systems
On Comparing Classifiers : Pitfalls to Avoid and Recommended Approach
Company Confidential © 2012 Eli Lilly and Company Beyond ICH Q1E Opening Remarks Rebecca Elliott Senior Research Scientist Eli Lilly and Company MBSW 2013.
Stability Studies - Evaluation of Outcomes and Development of Documentation For Regulatory Submissions Bob Seevers.
Chapter 7 Hypothesis Testing
6. Statistical Inference: Example: Anorexia study Weight measured before and after period of treatment y i = weight at end – weight at beginning For n=17.
1 Project 2: Stock Option Pricing. 2 Business Background Bonds & Stocks – to raise Capital When a company sell a Bond - borrows money from the investor.
CHAPTER 16 Life Tables.
1 Slides revised The overwhelming majority of samples of n from a population of N can stand-in for the population.
USE OF EVIDENCE IN DECISION MODELS: An appraisal of health technology assessments in the UK Nicola Cooper Centre for Biostatistics & Genetic Epidemiology,
Determining the Significant Aspects
Phase II/III Design: Case Study
The world leader in serving science TQ ANALYST SOFTWARE Putting your applications on target.
Statistical Analysis SC504/HS927 Spring Term 2008
Chapter 15 ANOVA.
Science as a Process Chapter 1 Section 2.
Issues of Simultaneous Tests for Non-Inferiority and Superiority Tie-Hua Ng*, Ph. D. U.S. Food and Drug Administration Presented at MCP.
1 Workshop on inventories of greenhouse gas emissions from aviation and navigation May 2004, Copenhagen EU greenhouse gas emission trends and projections.
Statistical Inferences Based on Two Samples
© The McGraw-Hill Companies, Inc., Chapter 10 Testing the Difference between Means and Variances.
©2006 Prentice Hall Business Publishing, Auditing 11/e, Arens/Beasley/Elder Audit Sampling for Tests of Controls and Substantive Tests of Transactions.
Overall Audit Plan and Audit Program
How do we delay disease progress once it has started?
Multiple Regression and Model Building
January Structure of the book Section 1 (Ch 1 – 10) Basic concepts and techniques Section 2 (Ch 11 – 15): Inference for quantitative outcomes Section.
Patient Survey Results 2013 Nicki Mott. Patient Survey 2013 Patient Survey conducted by IPOS Mori by posting questionnaires to random patients in the.
UK Renal Registry 17th Annual Report Figure 5.1. Trend in one year after 90 day incident patient survival by first modality, 2003–2012 cohorts (adjusted.
4/4/2015Slide 1 SOLVING THE PROBLEM A one-sample t-test of a population mean requires that the variable be quantitative. A one-sample test of a population.
Hypothesis Testing. To define a statistical Test we 1.Choose a statistic (called the test statistic) 2.Divide the range of possible values for the test.
1 Volume measures and Rebasing of National Accounts Training Workshop on System of National Accounts for ECO Member Countries October 2012, Tehran,
Shibing Deng Pfizer, Inc. Efficient Outlier Identification in Lung Cancer Study.
Regulatory Toxicology James Swenberg, D.V.M., Ph.D.
False Discovery Rates for Discrete Data Joseph F. Heyse Merck Research Laboratories Graybill Conference June 13, 2008.
Adaptive randomization
The Future of Chemical Toxicity Testing in the U.S.
Acute Toxicity Studies Single dose - rat, mouse (5/sex/dose), dog, monkey (1/sex/dose) 14 day observation In-life observations (body wt., food consumption,
Presentation transcript:

Issues and Experience in Analyzing Transgenic Mouse Carcinogenicity Studies: An Industry Perspective Ronald Menton Wyeth Research 2005 FDA/Industry Statistics Workshop Washington, DC, Sep 2005

Outline Some statistical questions for 2-year studies Transgenic models Some thoughts on the questions for transgenic models Final Comments

Study Design Questions? Are two control groups needed? How many animals per group? What groups are needed? Statistical methods?

Some In-Life Questions? When should we terminate group x? When should we terminate the study? Do we have a valid study?

Questions at End of Study? DO WE HAVE A VALID STUDY? ARE ANY FINDINGS STATISTICALLY SIGNIFICANT?

Transgenic Mouse Models Mouse model more susceptible to drug-induced tumors due to –Knocking out gene associated with tumor suppression (eg., p53 +/-, XPA - ) –Insertion of multiple copies of human gene associated with tumor promotion (eg., TgrasH2,TG.AC) The increased signal permits shorter study duration and smaller group sizes

Transgenic Models Current Regulations (ICH S1B) permit sponsors to conduct the traditional 2 year rat study plus a short- or medium- term rodent study in lieu of 2 year studies in both rats and mice The Committee for Medicinal Products for Human Use stated that the TgrasH2 and p53 +/- mouse models are acceptable alternatives to the 2-year mouse study. CPMP (2004)

Why Conduct Transgenic Study? Faster –In-life: 6-months vs 2 years –Study completion: 1 year vs > 3 years Less Resources –Fewer animals –People –Space Increased Flexibility for Drug Development

Typical Study Design for 2-Year Rodent Study

Are Two Control Groups Needed? Many companies routinely use two vehicle control groups for 2-year carcinogenicity studies. Why? –Permits an assessment of variation in tumor rates between groups –Poor survival in control group is problematic See Haseman (1990) for discussion

Multiple Control Groups in 2-Year Studies Eight of 14 companies indicated that multiple control groups are employed for at least 75 % of their studies. What type of multiple control group designs are routinely used? 9 Two vehicle control groups 2 Vehicle control and water control 2 Vehicle control and untreated control Survey of 14 PhRMA Companies on Statistical Methods Used for 2-year Rodent Carcinogenicity Studies. Menton R (2003)

Are Two Control Groups Needed? Are Two Vehicle Control Groups Needed in Short-term Carcinogenicity Studies? Not for most models –Low spontaneous rate of tumors –Survival rate usually high for at least 6 months

Survival for P53 Mouse from 6 NTP Studies NTP Web Site

Mortality in TgrasH2 Mice

Spontaneous Tumors in P53 Mice Adapted from NTP Website

Spontaneous Tumors in TgrasH2 Mice Usui (2001) summarized tumor incidence and time of first tumor for common spontaneous tumors (incidence > 1%) in 12 ILSI ACT studies. 180 male and 178 female mice (15 per study/sex) Male tumor incidence: 0 – 1.8% Female tumor incidence: 0 – 2.3% In most cases, the incidence of these common tumors was only marginally greater than 1.0%

How Many Animals Per Group? 2-year mouse studies typically use between animals per group. Study duration was typically 24 months for both rat and mouse studies. The number of animals per group per sex was evenly divided between 50, 60, and 65.

How Many Animals Per Group? Original ILSI protocols recommended 15 animals per group for transgenic studies Recent papers and presentations have recommended per group –Morton (2002) –Lin (2004) –CPMP (2004)

Sample Size Recommend 20 to 25 mice/sex/group for carcinogenicity assessment studies in TgrasH2 mice. (Morton 2002) Group size of 15 animals in the original transgenic mouse study protocol is too small. To have a level of power between 80 and 90% in detecting a true 15% difference, animals per group are needed. (Lin 2004) The number of animals per group in the ILSI/HESI studies is too small. An increase in group size to animals per group is recommended. (CPMP 2004)

Power to Detect Selected Increases in Tumor Rate Assuming Background Tumor Rate Near 0

What Groups to Include? Typical 2-year carcinogenicity study includes 5 groups: C1, C2, L, M, H All but one respondent indicated that a typical study includes three dose groups, with one stating that they usually employ four dose groups.

Study Design for TgrasH2 Study

What Groups to Include? Original ILSI Protocol recommended 7 Groups C, L, M, H, Positive Control, WT-C, WT-H WT groups are now considered optional Two questions: –Is the PC control group needed? –If PC group included, then how many animals are needed in this group?

Positive Controls in Short-term Studies Storer (2001) summarized results for 19 ILSI ACT studies that used p-cresidine as the positive control group N=15 per sex Males –P-cresidine was considered positive for 18 of 19 studies –Bladder tumor incidence ranged from 0 to 86.7% Females –P-cresidine was considered positive for 15 of 19 studies –Bladder tumor incidence ranged from 0 to 60%

Positive Controls in Short-term Studies

Power for Comparing Tumor Incidence Between Positive Control and Vehicle Control Group

Possible Design for 6-Month P53+/- or TgrasH2 Study

Statistical Methods? Eleven of 13 respondents familiar with the procedures detailed in the draft FDA guidance document, Statistical Aspects of Design, Analysis, and Interpretation of Animal Carcinogenicity Studies. Twelve companies stated that they are using Peto type tests for the analysis of tumor data. Petos test is commonly used for the statistical analysis of tumor data for 2-year carcinogenicity studies

Options for Statistical Methodology for P53 and TgrasH2 Studies Cochran-Armitage Trend test and Fishers Exact test Exclude animals that die with short survival times. Definition of sufficient survival based on time of tumor observation in sponsors historical data and literature Peto Methods Poly-K methods

Cochran-Armitage and Fisher Exact Tests Advantages Simple, well known test Exact tests available Easy to block or stratify for other covariates Appropriate if there are few fatal tumors and intercurrent mortality is similar among groups Disadvantages Requires specification of survival time for excluding animals Does not account for time of tumor onset or cause of death

Peto Methods Advantages FDA may use Petos method Accounts for time of tumor onset and cause of death Software available Exact tests available Scientists familiar w/ methods Disadvantages Requires specification of incidental intervals Specification of incidental intervals is complicated due to small number of deaths in vehicle control groups Complexity makes stratification/blocking more difficult

Poly-K Methods Advantages Adjusts for mortality Does not require cause of death determination Do not have to specify time intervals Easy to block or stratify for the two studies Fairly simple method Disadvantages Not much experiece for 6- month study Biologists not familiar with method Application of exact tests for poly-k method is a research topic

Statistical Methods? Incidence of mortality, neoplasms/select non-neoplasms will be compared among dosage groups using the Cochran- Armitage trend test and Fisher's exact test between each dosage group and the vehicle-control group. If excessive intercurrent mortality is observed then the trend and pairwise tests of tumor data will be conducted using Peto's method. What constitutes excessive mortality? Number of early deaths: > 5? > 10? Employ Poly-k Method?

Questions During In Life Ten of 13 companies indicated that at least one dose group was terminated early or the top dose lowered for at least one study in the past five years. Mortality and/or differential intercurrent mortality raises statistical questions during conduct of 2-year studies –Should the high dose be lowered? –Should one or more groups be terminated early? –Should the study be terminated early?

Mortality Guidelines for 2-year Studies animals per group should be alive during weeks –FDA Draft Guidance (May 2001) High-Dose group could be terminated early when the survival of the group is reduced to animals –Fairweather et al (1998). Drug Information Journal A study could be terminated if survival of the control group goes below after weeks –FDA Draft Guidance (May 2001)

Mortality Issues for Short-term Studies Survival is usually very high in short-term studies However, what do we do if it isnt? What are the criteria for evaluating if study is acceptable, terminating a study, or terminating a dosage group?

Mortality Issues for Short-term Studies We (scientific community) do not currently know how many animals are needed at the end of a 26-week carcinogenicity study We also do not know how many weeks represents sufficient exposure We do know that the more animals per group the more sensitive the statistical tests will be for detecting compound related tumor increases of a specified magnitude

Power for Reduced Survival

Description of Power Calculations Simulations were conducted to estimate the probability of detecting differences of % in tumor rates between the treated groups and control group –Power calculations assume that tumor incidence is compared among 4 dosage groups using a one-sided Cochran-Armitage trend test conducted at the 5% significance level –Background tumor incidence ranged from 0.1% to 3% –Tumor incidence in L and M dosage groups ranged from background rates to 2/3 of that in H dosage group –Power was computed via simulation (1000 runs per simulation) –Calculations performed for two sets of samples sizes: 25, 24, 22, and 15 in the C, L, M, and H dosage groups, 25, 24, 22, and 10 in the C, L, M, and H dosage groups,

Some Thoughts On Mortality Guidelines for Short-term Studies xx-yy animals per group should be alive during weeks ww-zz –xx - yy = 15 – 20? –ww-zz likely species dependent High-dose group could be terminated early when the survival of the group is reduced to (?) animals before weeks ww-zz. A study could be terminated if survival of the control group goes below 20 (assuming n = 25) before weeks ww - zz

Are Any Findings Statistically Significant? Six of 13 companies employ the decision rule in FDA s draft guidance document of for rare tumors and for common tumors. What significance levels are used for the evaluation of rare/common tumors? Rare/Common /0.05 with no adjustments for multiple tumors /0.05 with an adjustment for multiple tumors /0.01 i.e., Haseman Rule /0.005 i.e., FDA Decision Rule What is Considered Statistically Significant? Different approaches are utilized to adjust for the multiple statistical tests performed in 2-year carcinogenicity studies.

Decision Rule in FDAs Draft Guidance Adapted from US FDA (May 2001)

What is Considered Statistically Significant? Is a multiplicity adjustment needed for short-term studies? No –Only a handful of tumor types observed in a study –Probability of a false positive is low due to low spontaneous rate

Final Comments Alternative mouse models provide additional flexibility in drug development While 25 animals per sex/group is reasonable for the control and treated transgenic groups, smaller sample sizes make sense for the positive control group Simple statistical methods work well when survival is high More research and/or guidance is needed on defining adequate survival

Some References CPMP Safety Working Party. CHMP SWP conclusions and recommendations on the use of genetically modified animal models for carcinogenicity assessment. London, 23 June Haseman JK, Hajian G, Crump KS, Selwyn MR, and Peace KE, Dual controls in rodent carcinogenicity studies. In: Statistical issues in drug research and development, Ed by KE Peace. Marcel Dekker, New York Lin K. Statistical Issues in Review of Carcinogenicity Studies of Pharmaceuticals, Drug Information Association 40th Annual Meeting, June 16, 2004, Washington, DC MacDonald J, et al. The utility of genetically modified mouse assays for identifying human carcinogens: a basic understanding and path forward. Toxicol Sci. 2004: Menton R. and R Perry. Statistical Methods for 2-Year Rodent Carcinogenicity Studies. Midwest Biopharmaceutical Workshop, Muncie, In, Morton D. The Tg rasH2 Mouse in Cancer Hazard Identification, Toxicol Pathol, 2002: NTP web pages on Histoical Controls for P53 Mice. Study Results & Research Projects >> Study Data Searches >> Historical Controls >> NTP Historical Control for Genetically-Modified Models Storer R, et al. p53+/- Hemizygous Knockout Mouse: Overview of Available Data. Toxicol. Pathol.,2001, 29 Suppl: Takaoka M, et al. Interlaboratory comparison of short-term carcinogenicity studies using CB6F1- rasH2 transgenic mice. Toxicol Pathol, 2003: US Food and Drug Administration, Statistical Aspects of Design, Analysis, and Interpretation of Animal Carcinogenicity Studies, Draft Guidance for Industry, May Usui T, et al., CB6F1-rasH2 mouse: Overview of Available Data. Toxicol Pathol, Suppl: