Mandrekar, Sumithra J.

Background Biomarkers have many distinct purposes, and depending on their intended use, the validation process varies substantially. Purpose The goal of this article is to provide an introduction to the topic of biomarkers, and then to discuss three specific types of biomarkers, namely, prognostic, predictive, and surrogate. Results A principle challenge for biomarker validation from a statistical perspective is the issue of multiplicity. In general, the solution to this multiplicity challenge is well known to statisticians: pre-specification and replication. Critical requirements for prognostic marker validation include uniform treatment, complete follow-up, unbiased case selection, and complete ascertainment of the many possible confounders that exist in the context of an observational sample. In the case of predictive biomarker validation, observational data are clearly inadequate and randomized controlled trials are mandatory. Within the context of randomization, strategies for predictive marker validation can be grouped into two categories: retrospective versus prospective validation. The critical validation criteria for a surrogate endpoint is to ensure that if a trial uses a surrogate endpoint, the trial will result in the same inferences as if the trial had observed the true endpoint. The field of surrogate endpoint validation has now moved to the multi-trial or meta-analytic setting as the preferred method. Conclusions Biomarkers are a highly active research area. For all biomarker developmental and validation studies, the importance of fundamental statistical concepts remains the following: pre-specification of hypotheses, randomization, and replication. Further statistical methodology research in this area is clearly needed as we move forward.

Abstract Progression-free survival (PFS) is frequently used as the primary efficacy endpoint in the evaluation of cancer treatment that is considered for marketing approval. Missing or incomplete data problems become more acute with a PFS endpoint (compared with overall survival). In a given clinical trial, it is common to observe incomplete data due to premature treatment discontinuation, missed or flawed assessments, change of treatment, lack of follow-up, and unevaluable data. When incomplete data issues are substantial, interpretation of the data becomes tenuous. Plans to prevent, minimize, or properly analyze incomplete data are critical for generalizability of results from the clinical trial. Variability in progressive disease measurement between radiologists further contributes to data problems with a PFS endpoint. The repercussions of this on phase III clinical trials are complex and depend on several factors, including the magnitude of the variability and whether there is a systematic reader evaluation bias favoring one treatment arm particularly in open-label trials. Clin Cancer Res; 19(10); 2613–20. ©2013 AACR.

AbstractSulindac is a prescription‐based non‐steroidal anti‐inflammatory drug (NSAID) that continues to be actively investigated as a candidate cancer chemoprevention agent. To further current understanding of sulindac bioavailability, metabolism, and disposition, we developed a population pharmacokinetic model for the parent compound and its active metabolites, sulindac sulfide, and exisulind. This analysis was based on data from 24 healthy subjects who participated in a bioequivalence study comparing two formulations of sulindac. The complex disposition of sulindac and its metabolites was described by a seven‐compartment model featuring enterohepatic recirculation and is the first reported population pharmacokinetic model for sulindac. The derived model was used to explore effects of clinical variables on sulindac pharmacokinetics and revealed that body weight, creatinine clearance, and gender were significantly correlated with pharmacokinetic parameters. Moreover, the model quantifies the relative bioavailability of the sulindac formulations and illustrates the utility of population pharmacokinetics in bioequivalence assessment. This novel population pharmacokinetic model provides new insights regarding the factors that may affect the pharmacokinetics of sulindac and the exisulind and sulindac sulfide metabolites in generally healthy subjects, which have implications for future chemoprevention trial design for this widely available agent.

The aim of a phase I oncology trial is to identify a dose with an acceptable safety profile. Most phase I designs use the dose‐limiting toxicity, a binary endpoint, to assess the unacceptable level of toxicity. The dose‐limiting toxicity might be incomplete for investigating molecularly targeted therapies as much useful toxicity information is discarded. In this work, we propose a quasi‐continuous toxicity score, the total toxicity profile (TTP), to measure quantitatively and comprehensively the overall severity of multiple toxicities. We define the TTP as the Euclidean norm of the weights of toxicities experienced by a patient, where the weights reflect the relative clinical importance of each grade and toxicity type. We propose a dose‐finding design, the quasi‐likelihood continual reassessment method (CRM), incorporating the TTP score into the CRM, with a logistic model for the dose–toxicity relationship in a frequentist framework. Using simulations, we compared our design with three existing designs for quasi‐continuous toxicity score (the Bayesian quasi‐CRM with an empiric model and two nonparametric designs), all using the TTP score, under eight different scenarios. All designs using the TTP score to identify the recommended dose had good performance characteristics for most scenarios, with good overdosing control. For a sample size of 36, the percentage of correct selection for the quasi‐likelihood CRM ranged from 80% to 90%, with similar results for the quasi‐CRM design. These designs with TTP score present an appealing alternative to the conventional dose‐finding designs, especially in the context of molecularly targeted agents. Copyright © 2013 John Wiley & Sons, Ltd.

Purpose In lung cancer, randomized trials assessing hyperfractionated or accelerated radiotherapy seem to yield conflicting results regarding the effects on overall (OS) or progression-free survival (PFS). The Meta-Analysis of Radiotherapy in Lung Cancer Collaborative Group decided to address the role of modified radiotherapy fractionation. Material and Methods We performed an individual patient data meta-analysis in patients with nonmetastatic lung cancer, which included trials comparing modified radiotherapy with conventional radiotherapy. Results In non–small-cell lung cancer (NSCLC; 10 trials, 2,000 patients), modified fractionation improved OS as compared with conventional schedules (hazard ratio [HR] = 0.88, 95% CI, 0.80 to 0.97; P = .009), resulting in an absolute benefit of 2.5% (8.3% to 10.8%) at 5 years. No evidence of heterogeneity between trials was found. There was no evidence of a benefit on PFS (HR = 0.94; 95% CI, 0.86 to 1.03; P = .19). Modified radiotherapy reduced deaths resulting from lung cancer (HR = 0.89; 95% CI, 0.81 to 0.98; P = .02), and there was a nonsignificant reduction of non–lung cancer deaths (HR = 0.87; 95% CI, 0.66 to 1.15; P = .33). In small-cell lung cancer (SCLC; two trials, 685 patients), similar results were found: OS, HR = 0.87, 95% CI, 0.74 to 1.02, P = .08; PFS, HR = 0.88, 95% CI, 0.75 to 1.03, P = .11. In both NSCLC and SCLC, the use of modified radiotherapy increased the risk of acute esophageal toxicity (odds ratio [OR] = 2.44 in NSCLC and OR = 2.41 in SCLC; P < .001) but did not have an impact on the risk of other acute toxicities. Conclusion Patients with nonmetastatic NSCLC derived a significant OS benefit from accelerated or hyperfractionated radiotherapy; a similar but nonsignificant trend was observed for SCLC. As expected, there was increased acute esophageal toxicity.

Abstract Purpose: Novel therapies are challenging current drug development standards. Agents with specific biologic targets, unknown dose-efficacy curves, limited toxicity profiles, and in combination therapy require novel statistical designs. We compared the operating characteristics of three model based designs incorporating both toxicity and a measure of efficacy to identify the optimal dose region for combination therapies. Methods: Three designs were considered: continuation ratio model with a trinary outcome, bivariate logistic model treating toxicity and efficacy as independent outcomes, and bivariate probit model allowing for dependent toxicity and efficacy outcomes. All designs utilized a model-based methodology to adaptively determine dose escalation/de-escalation, based on the data observed to date in the trial. The model parameters were estimated using Markov chain Monte Carlo approach in a Bayesian framework. The target threshold for toxicity was set at 33%; the minimum and maximum sample sizes were set at 30 and 45, respectively, for the simulated trials; and the cohort size was 3 with at least nine patients needed to be treated at the proposed combination dose level. The performance of the designs was evaluated in simulation studies under six clinically relevant scenarios. Results: As expected, the ability to correctly identify an optimal dose region was sensitive to the changes in the toxicity and/or efficacy across the dose level combinations. Considering 50% as a benchmark for the recommendation rate for the optimal dose or region (dose levels within 10% of the maximum efficacious dose level), at least one of the three models performed well in recommending the optimal dose region except when the dose-efficacy surface was non-monotone or the dose-toxicity surface was close to the target rate of 33%. All three models worked well in not recommending any dose combination when the starting dose level combination was too toxic. The median sample size across all scenarios ranged from 27–33. Conclusions: Despite favorable operating characteristics, the implementation of these new designs in practice is challenging for a variety of reasons. Adoption of novel designs require a multi-pronged approach that provides a) clinically obvious advantages, b) practical design framework with a transparent decision making process, c) timely access to both toxicity and preliminary efficacy data, and d) portable user-friendly software, or very attentive statisticians. Citation Information: Mol Cancer Ther 2009;8(12 Suppl):C53.

Abstract Biomarkers are critical to targeted therapies, as they may identify patients more likely to benefit from a treatment. Several prospective designs for biomarker-directed therapy have been previously proposed, differing primarily in the study population, randomization scheme, or both. Recognizing the need for randomization, yet acknowledging the possibility of promising but inconclusive results after a stage I cohort of randomized patients, we propose a 2-stage phase II design on marker-positive patients that allows for direct assignment in a stage II cohort. In stage I, marker-positive patients are equally randomized to receive experimental treatment or control. Stage II has the option to adopt “direct assignment” whereby all patients receive experimental treatment. Through simulation, we studied the power and type I error rate of our design compared with a balanced randomized two-stage design, and conducted sensitivity analyses to study the effect of timing of stage I analysis, population shift effects, and unbalanced randomization. Our proposed design has minimal loss in power (&lt;1.8%) and increased type I error rate (&lt;2.1%) compared with a balanced randomized design. The maximum increase in type I error rate in the presence of a population shift was between 3.1% and 5%, and the loss in power across possible timings of stage I analysis was less than 1.2%. Our proposed design has desirable statistical properties with potential appeal in practice. The direct assignment option, if adopted, provides for an “extended confirmation phase” as an alternative to stopping the trial early for evidence of efficacy in stage I. Clin Cancer Res; 18(16); 4225–33. ©2012 AACR.

Abstract Purpose: The categorical definition of response assessed via the Response Evaluation Criteria in Solid Tumors has documented limitations. We sought to identify alternative metrics for tumor response that improve prediction of overall survival. Experimental Design: Individual patient data from three North Central Cancer Treatment Group trials (N0026, n = 117; N9741, n = 1,109; and N9841, n = 332) were used. Continuous metrics of tumor size based on longitudinal tumor measurements were considered in addition to a trichotomized response [TriTR: response (complete or partial) vs. stable disease vs. progression). Cox proportional hazards models, adjusted for treatment arm and baseline tumor burden, were used to assess the impact of the metrics on subsequent overall survival, using a landmark analysis approach at 12, 16, and 24 weeks postbaseline. Model discrimination was evaluated by the concordance (c) index. Results: The overall best response rates for the three trials were 26%, 45%, and 25%, respectively. Although nearly all metrics were statistically significantly associated with overall survival at the different landmark time points, the concordance indices (c-index) for the traditional response metrics ranged from 0.59 to 0.65; for the continuous metrics from 0.60 to 0.66; and for the TriTR metrics from 0.64 to 0.69. The c-indices for TriTR at 12 weeks were comparable with those at 16 and 24 weeks. Conclusions: Continuous tumor measurement–based metrics provided no predictive improvement over traditional response–based metrics or TriTR; TriTR had better predictive ability than best TriTR or confirmed response. If confirmed, TriTR represents a promising endpoint for future phase II trials. Clin Cancer Res; 17(20); 6592–9. ©2011 AACR.

Abstract Pomalidomide at doses of 2 or 4 mg/d has demonstrated excellent activity in patients with multiple myeloma (MM). We opened 2 sequential phase 2 trials using the pomalidomide with weekly dexamethasone (Pom/dex) regimen at differing doses to study the efficacy of this regimen in patients who have failed both lenalidomide and bortezomib. Pomalidomide was given orally 2 or 4 mg daily with dexamethasone 40 mg weekly. Thirty-five patients were enrolled in each cohort. Confirmed responses in the 2-mg cohort consisted of very good partial response (VGPR) in 5 (14%), partial response (PR) in 4 (11%), minor response (MR) in 8 (23%) for an overall response rate of 49%. In the 4-mg cohort, confirmed responses consisted of complete response (CR) in 1 (3%), VGPR in 3 (9%), PR in 6 (17%), MR in 5 (14%) for an overall response rate of 43%. Overall survival at 6 months is 78% and 67% in the 2- and 4-mg cohort, respectively. Myelosuppression was the most common toxicity. This nonrandomized data suggests no advantage for 4 mg over the 2 mg daily. Pomalidomide overcomes resistance in myeloma refractory to both lenalidomide and bortezomib. This trial is registered at http://ClinicalTrials.gov, number NCT00558896.