Validation of biomarkers to predict response to immunotherapy in cancer: Volume II - clinical validation and regulatory considerations

Abstract

There is growing recognition that immunotherapy is likely to significantly improve health outcomes for cancer patients in the coming years. Currently, while a subset of patients experience substantial clinical benefit in response to different immunotherapeutic approaches, the majority of patients do not but are still exposed to the significant drug toxicities. Therefore, a growing need for the development and clinical use of predictive biomarkers exists in the field of cancer immunotherapy. Predictive cancer biomarkers can be used to identify the patients who are or who are not likely to derive benefit from specific therapeutic approaches. In order to be applicable in a clinical setting, predictive biomarkers must be carefully shepherded through a step-wise, highly regulated developmental process. Volume I of this two-volume document focused on the pre-analytical and analytical phases of the biomarker development process, by providing background, examples and "good practice" recommendations. In the current Volume II, the focus is on the clinical validation, validation of clinical utility and regulatory considerations for biomarker development. Together, this two volume series is meant to provide guidance on the entire biomarker development process, with a particular focus on the unique aspects of developing immune-based biomarkers. Specifically, knowledge about the challenges to clinical validation of predictive biomarkers, which has been gained from numerous successes and failures in other contexts, will be reviewed together with statistical methodological issues related to bias and overfitting. The different trial designs used for the clinical validation of biomarkers will also be discussed, as the selection of clinical metrics and endpoints becomes critical to establish the clinical utility of the biomarker during the clinical validation phase of the biomarker development. Finally, the regulatory aspects of submission of biomarker assays to the U.S. Food and Drug Administration as well as regulatory considerations in the European Union will be covered.

Keywords: Assay; Biomarker; Cancer; Immunotherapy; Regulatory; Validation.

Fig. 1

The biomarker development process can be schematically divided into sequential phases, including preanalytical and analytical validation, clinical validation, regulatory approval, and demonstration of clinical utility

Fig. 2

The clinical sensitivity and specificity of a biomarker assay must be demonstrated through robust receiver operative characteristics (ROC) curves. As illustrated, an ROC curve is a plot that captures true positive rate (TRP) against false positive rate (FRP) at various threshold settings

Fig. 3

The impact of improper resampling shown on an RNASeq dataset [13, 14]. Samples are classified into Group 1 (CEU, n = 69 samples) versus group 2 (YRI, n = 60 samples) using the lasso logistic regression classifier as implemented in the glmnet package [36]. The “No CV” case did not use cross-validation to pick a value for the tuning parameter, instead using a fixed value 4e-9. The “naïve CV” method used naïve, non-nested cross-validation to pick the tuning parameter. The “nested CV” method used nested cross-validation to pick the tuning parameter, so that there was never any overlap between the data used to develop the predictor and the data used to estimate and evaluate the prediction scores. The accuracy estimated from the correct nested CV method is 95 %, and from each of the other methods is 100 %, the difference representing bias due to erroneous resampling

Fig. 4

There are three basic phase III design options for assessing the ability of a biomarker. The enrichment design includes only patients who are positive for the biomarker in a study evaluating the effect of a new therapy (1). In the biomarker stratified design, all patients, independent of biomarker results, are enrolled and randomized to treatment and control groups within each of the biomarker positive and negative groups to ensure balance (2). Finally, in the strategy design, patients are randomized between no use of the biomarker (all patients receive standard therapy on that arm) and a biomarker-based strategy where biomarker-negative patients receive standard therapy and biomarker-positive patients receive the new therapy (3)

Fig. 5

Representative survival curves illustrating the different clinical scenarios involved in the FDA approval of pembrolizumab using the PD-L1 22C3 PharmDx assay (a) vs. nivolumab using the PD-L1 28–8 PharmDx assay (b). For pembrolizumab administered in second-line NSCLC, panel a shows Kaplan–Meier estimates of progression-free survival according to the proportion score of the percentage of neoplastic cells with membranous PD-L1 staining. In this context, the biomarker was used as inclusion criteria to select the patient population in which the clinical activity of pembrolizumab was assessed. For nivolumab, PD-L1 expression was assessed retrospectively in prospectively collected tissue samples. Panel b illustrates Kaplan-Meier estimates of progress-free survival in patients receiving nivolumab or docetaxel by PD-L1 expression level. In this study, the test was not used for patient selection but to evaluate the interaction between PD-L1 expression and clinical benefit. Panel a from The New England Journal of Medicine, 2015, 372, 2018-2028 Edward B. Garon et al., Pembrolizumab for the Treatment of Non–Small-Cell Lung Cancer. Copyright © 2015 Massachusetts Medical Society. Panel b from The New England Journal of Medicine, 2015, 373, 1627-1639 Hossein Borghaei et al., Nivolumab versus Docetaxel in Advanced Nonsquamous Non–Small-Cell Lung Cancer, 373, 1627-1639. Copyright © 2015 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society