|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 29 (October 10), 2007: pp. 4513-4515 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2007.12.7803
Interpreting P Values in Pharmacogenetic Studies: A Call for Process and PerspectiveSection of Hematology/Oncology, Department of Medicine, University of Chicago, Chicago, IL
Section of Genetic Medicine, Department of Medicine, University of Chicago, Chicago, IL To our knowledge, the cancer pharmacogenetics article by Marsh et al1 in this issue of the Journal of Clinical Oncology is the largest published pharmacogenetic study of carboplatin and taxanes, focusing on the potential association of candidate polymorphisms with treatment toxicity and disease response outcomes. It is a model for how collection of whole blood samples for DNA in a phase III clinical trial (which, by design, prospectively collects important phenotype data with unbiased treatment assignments) enables high quality pharmacogenetic research. Strikingly, but appropriately, it is published with no positive findings to report. Attaining the proposed benefits2 of the Human Genome Project requires an iterative process of hypothesis generation and hypothesis testing. It is not disconcerting that this study failed to confirm previously published associations between various polymorphisms and outcomes from combined carboplatin and docetaxel or paclitaxel regimens. It is a signal that more work might be done, and should be a warning to those investigators who have jumped from small, hypothesis-generating pharmacogenetic studies with borderline P values into large prospective randomized trials to test the utility of the putative pharmacogenetic test, but without first attempting to validate the association. The first two pharmacogenetic tests in oncology, for thiopurine methyltransferase (TPMT) and UGT1A1 deficiencies, were developed through a candidate gene/polymorphism approach. In each case, clinical observations led to multiple preclinical studies, including testing of the relationship between specific functional polymorphisms of TPMT and UGT1A1 and enzyme function. Ultimately, small prospective clinical trials were conducted where the primary objective was to confirm the relationship between the genetic variant and excess exposure and consequent toxicities from 6-mercaptopurine and irinotecan, respectively.3 Use of this approach does not ensure successful development of a pharmacogenetic test. While dihydropyrimidine dehydrogenase deficiency is heritable and leads to excess exposure to fluorouracil and consequent toxicity, it has been associated with numerous polymorphisms in the DPYD gene, but with each single nucleotide polymorphism (SNP) found at different frequencies in different populations. Hence, any single genotype has insufficient negative predictive value to be clinically useful. At the same time, adhering to the principles of the candidate gene/polymorphism approach has instead led investigators to focus efforts on developing phenotype-based assays.4 Marsh et al utilized an approach likely to be of increasing importance in pharmacogenetics. They were able to test specific hypotheses regarding candidate polymorphisms in a large clinical trial data set because DNA had been collected and was available for analysis. By removing biases of treatment assignment, completing data collection within a specified timeframe, and by careful phenotyping for features of immediate clinical relevance, large clinical trials of cancer therapy provide important data sets for developing pharmacogenetic tests. Since preclinical testing does not assess the magnitude of effect for a candidate gene/polymorphism in the clinical setting,5 genotyping for associations with previously collected phenotypic data provides a crucial opportunity for validating the candidate polymorphism, without incurring the expense of a new prospective study. Investigators could change the development plan based on these tests and increase the likelihood of success in the prospective setting, for example by increasing the number of polymorphisms typed to capture better the relevant genetic variants affecting a treatment phenotype or by switching to a phenotype test altogether. The study by Marsh et al is also a lesson in how to assess appropriately the relationship between candidate polymorphisms and clinical outcomes. One critical obstacle to successful development of a genotype-based diagnostic test is the high number of spurious associations. Each patient has 109 base pairs (bp) in his/her genome. With SNPs occurring every 100 to 300 bp, taken to the extreme, each patient has 107 potential hypotheses to test for association with each clinically relevant phenotype of interest. With a P value cutoff of .05, 500,000 SNPs will demonstrate a significant association with one phenotype by chance alone. As increasing numbers of phenotypes or combinations of polymorphisms are considered, the multiple-hypothesis testing in the same data set increases the number of false-positive associations. With so many expected false-positive associations, a nominally significant P value merely suggests, but does not confirm, a hypothesis that would warrant additional testing. Marsh et al addressed this problem for their association of 27 candidate polymorphisms with six clinical phenotypes first by predetermining an acceptable false-discovery rate and screening these SNPs in 609 randomly selected patients with univariate models. They then incorporated the two SNPs passing the false-discovery rate criteria into a multivariate model that was then tested in 305 patients, and the two SNPs proved to be false-positives by this statistical assessment. Fundamentally, their study served as a filter for determining which SNPs had sufficiently robust associations with relevant phenotypes to proceed with further development, and none passed the test. As in any iterative process, the results are to be interpreted with care. Some other important phenotype that was not tested might be tightly associated with one of these SNPs or one of these SNPs might inform treatment decisions with these agents in a different disease setting. However, none of these SNPs demonstrates sufficient evidence for association to proceed with further development as a predictive marker in this treatment setting. The DNA and clinical data collected in this clinical trial, the results of which were published in 2004,6 remain useful for further investigation as a resource for new methods of genotype-phenotype association study discovery and as a reference set in which candidate genes/polymorphisms can continue to be tested. Before completion of the Human Genome Project, the International HapMap Project began to catalog and analyze the interrelationships among the SNPs identified during various sequencing studies. At any site in the genome, SNP genotypes may be locally correlated such that knowing the genotype at one SNP provides significant ability to predict the genotypes at many other SNPs. Such a SNP is a "tagging SNP" and the block of coinherited SNPs and intervening DNA sequence is a haplotype block. By typing a subset of carefully determined tagging SNPs, on average, large segments of genome sequence can be determined just by typing one SNP. The latest genotyping techniques enable determination of approximately 900,000 to 1,000,000 SNPs dispersed across the genome at reasonable cost. The first set of major successes for genome-wide association includes identification of genes conferring risk for incidence of breast cancer,7,8 type 2 diabetes mellitus,9-11 myocardial infarction,12,13 and prostate cancer.14-16 The contribution of any single gene variant to these complex phenotypes is small. To achieve results with suitable false discovery rates within the genome-wide association approach required replication in multiple population samples, but some of these associations were identified with fewer than 2,500 subjects; newer methods will continue to reduce this number. The typical enrollment for larger cancer therapy trials approaches the scale of these successful genome-wide association studies. Due to the emphasis on statistical association, this approach is sensitive to the biases generated when investigators collect data retrospectively by single institution chart reviews because generating sufficient numbers often cluster patients with various related diagnoses across time periods of diagnosis and variable treatment regimens. The prospective clinical trial plays a unique role in our efforts to advance personalized cancer medicine, as it not only provides the scale necessary for successful genotype-phenotype association-based discovery, but also as a data set in which candidate gene/polymorphisms identified with the genotype-phenotype association approach in other trials or iterative laboratory studies can be tested. Given the potential returns and relatively limited expense, it is unclear why we have not more aggressively committed to collecting DNA in phase III oncology clinical trials. In this new field of diagnostic development, oncology investigators are challenged with reconciling three conflicting interests: (1) delivering a clinically useful diagnostic test that will better individualize cancer patient care; (2) achieving this first goal rapidly and accurately enough to provide valuable return to the community of practitioners and patients who would use the test; and (3) publishing results with sufficient frequency and merit to maintain grant funding, encourage prospective clinical collaborators, and garner sufficient recognition for investigators' professional development and promotion. Our growing knowledge of SNPs and plummeting cost of genotyping, as well as the hundreds of new agents in clinical trials have pressured interests 2 and 3. Focus on these at the expense of the primary goal jeopardizes this entire field of research. It is only 6 years since the first publication of a human genome sequence, so optimal methods for resolving these competing interests are not established. The publication of this article provides important recognition from a high impact journal for rigorous efforts to personalize cancer medicine even in the face of negative results. Perhaps other oncology editors and reviewers will take note and in the future, rather than focus on positive findings with marginally attractive P values, attend to the plausibility of the generated hypothesis and the appropriateness and rigor of the efforts used to substantiate the investigators' claims. AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a "U" are those for which no compensation was received; those relationships marked with a "C" were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Employment or Leadership Position: None Consultant or Advisory Role: Mark J. Ratain, Bristol-Myers Squibb (C), OSI Pharmaceuticals (C) Stock Ownership: Mark J. Ratain, Applera, Illumina Honoraria: None Research Funding: Michael L. Maitland, Bristol Myers-Squibb Expert Testimony: None Other Remuneration: Mark J. Ratain AUTHOR CONTRIBUTIONS Conception and design: Michael L. Maitland, Nancy J. Cox, Mark J. Ratain Manuscript writing: Michael L. Maitland, Nancy J. Cox, Mark J. Ratain Final approval of manuscript: Michael L. Maitland, Nancy J. Cox, Mark J. Ratain REFERENCES
1. Marsh S, Paul J, King CR, et al: Pharmacogenetic assessment of toxicity and outcome following platinum/taxane chemotherapy in ovarian cancer (SCOTROC1). J Clin Oncol 25:4528-4535, 2007 2. Collins FS: Shattuck lecture–medical and societal consequences of the Human Genome Project. N Engl J Med 341:28-37, 1999 3. Maitland ML, Vasisht K, Ratain MJ: TPMT, UGT1A1 and DPYD: Genotyping to ensure safer cancer therapy? Trends Pharmacol Sci 27:432-437, 2006[CrossRef][Medline] 4. Mattison LK, Fourie J, Hirao Y, et al: The uracil breath test in the assessment of dihydropyrimidine dehydrogenase activity: Pharmacokinetic relationship between expired 13CO2 and plasma [2-13C]dihydrouracil. Clin Cancer Res 12:549-555, 2006 5. Rebbeck TR, Spitz M, Wu X: Assessing the function of genetic variants in candidate gene association studies. Nat Rev Genet 5:589-597, 2004[Medline] 6. Vasey PA, Jayson GC, Gordon A, et al: Phase III randomized trial of docetaxel-carboplatin versus paclitaxel-carboplatin as first-line chemotherapy for ovarian carcinoma. J Natl Cancer Inst 96:1682-1691, 2004 7. Easton DF, Pooley KA, Dunning AM, et al: Genome-wide association study identifies novel breast cancer susceptibility loci. Nature 447:1087-1093, 2007[CrossRef][Medline] 8. Hunter DJ, Kraft P, Jacobs KB, et al: A genome-wide association study identifies alleles in FGFR2 associated with risk of sporadic postmenopausal breast cancer. Nat Genet 39:870-874, 2007[CrossRef][Medline] 9. Saxena R, Voight BF, Lyssenko V, et al: Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science 316:1331-1336, 2007 10. Scott LJ, Mohlke KL, Bonnycastle LL, et al: A genome-wide association study of type 2 diabetes in finns detects multiple susceptibility variants. Science 316:1341-1345, 2007 11. Zeggini E, Weedon MN, Lindgren CM, et al: Replication of genome-wide association signals in UK samples reveals risk loci for type 2 diabetes. Science 316:1336-1341, 2007 12. Helgadottir A, Thorleifsson G, Manolescu A, et al: A common variant on chromosome 9p21 affects the risk of myocardial infarction. Science 316:1491-1493, 2007 13. McPherson R, Pertsemlidis A, Kavaslar N, et al: A common allele on chromosome 9 associated with coronary heart disease. Science 316:1488-1491, 2007 14. Gudmundsson J, Sulem P, Manolescu A, et al: Genome-wide association study identifies a second prostate cancer susceptibility variant at 8q24. Nat Genet 39:631-637, 2007[CrossRef][Medline] 15. Haiman CA, Patterson N, Freedman ML, et al: Multiple regions within 8q24 independently affect risk for prostate cancer. Nat Genet 39:638-644, 2007[CrossRef][Medline] 16. Yeager M, Orr N, Hayes RB, et al: Genome-wide association study of prostate cancer identifies a second risk locus at 8q24. Nat Genet 39:645-649, 2007[CrossRef][Medline]
Related Article
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
