Developmental And Validation Studies

It is important to distinguish the studies that develop parmacogenomic classifiers from those that utilize such classifiers for targeting treatment selection or for evaluating the clinical utility of such classifiers. The vast majority of published prognostic marker studies are developmental. Developmental studies are often based on a convenience sample of patients for whom tissue is available but who are heterogeneous with regard to treatment and stage. Although there is a large literature on prognostic markers, few such factors are used in clinical practice. Prognostic markers are unlikely to be used unless they are therapeutically relevant, and most developmental studies are not based on a cohort medically coherent enough to establish therapeutic relevance.

The patients included in a developmental study of a pharmacogenomic biomarker to be used in drug development should be appropriate to enable identification of patients who are most likely to benefit from the new drug in a pivotal study. For example, suppose that the pivotal study involves advanced disease patients who have failed first-line treatment and involves comparing survivals for patients receiving the new drug to survivals for patients receiving palliative care.

Patients from single arm phase II trials of the new drug can be used to develop a phar-macogenomic biomarker classifier of those patients likely to respond to the new drug. Dobbin and Simon (9) have studied sample size considerations for developmental studies of predictive binary classifiers and have indicated that generally at least 20 cases in each class are required. Consequently, a phase II database containing at least 20 responders and 20 non-responders would be needed for the development of a pharmacogenomic classifier to be used in the subsequent pivotal trials. This may require a larger phase II developmental program than is conventional.

If the pivotal study involves comparison of outcome for patients receiving a standard regimen C versus those receiving C plus the new drug, then development of a gene expression-based classifier is more complex. The classifier can be developed based on phase II studies of patients receiving C plus the new drug, but unless one also studied patients receiving C without the new drug one would not know whether prediction was drug specific or just reflected general responsiveness of the tumors.

It is possible to develop pharmacogenomic predictors of risk of tumor progression rather than tumor response. Even if the patients are receiving the investigational drug as a single agent, however, it may not be clear to what extent the predictor reflects drug effect rather than non-specific disease pace.

As indicated in the previous paragraphs, there are limitations to the adequacy of a conventional phase II database for empirically developing a pharmacogenomic classifier for use in a pivotal study. In many ways the best resource for developing a pharmacoge-nomic biomarker classifier for use in a pivotal trial is a collection of pre-treatment tumor specimens from patients enrolled in such a pivotal trial. For example, archived material from a "failed" pivotal trial of the drug can be used to develop a biomarker classifier of patients most likely to benefit from the drug compared to the control. The classifier can be based on the actual endpoint used in the clinical trial or upon an intermediate endpoint such as progression-free survival for which there may be more events available.

By "failed" pivotal trial, we mean a trial for the same target population of patients that did not establish a statistically significant benefit for the drug for the randomized patients as a whole. The classifier developed based on archived material in a failed pivotal trial should be considered to have the same status as a classifier based on a phase II database. That is, the classifier should be used to design a new pivotal trial that establishes the clinical benefit of the drug in a prospectively specified subset of patients. Using the same pivotal trial to develop a pharmacogenomic classifier and to test treatment effects in subsets determined by the classifier is generally not valid.

Freidlin and Simon (10) have shown, however, how one pivotal trial can be used potentially for both purposes—if the set of patients used to develop the classifier is kept distinct from the set of patients used to evaluate treatment benefit. Generally, however, the studies should be kept separate. Developmental studies are exploratory, though they should result in completely specified binary classifiers. Studies on which claims of drug benefit are based should be non-exploratory, but should instead test prospectively defined hypotheses about treatment effect in a pre-defined patient population.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment