Clinical Study Data Request Registered Users, Please Login

Proposal 1433

Title of the Proposed Research

Identification of patient subgroups with enhanced treatment response in RCTs with time-to-event outcomes: use of counterfactual restricted survival times.

Lead Researcher

Ludovic Trinquart


INSERM U1153 METHODS team, Paris, France, Cochrane France

Funding Source


Potential Conflicts of Interest


Data Sharing Agreement Date

17 June 2016

Lay Summary


Therapeutic evaluation through RCTs is commonly focusing on difference in the average outcome between the experimental and control groups. However, a treatment is unlikely to be equally effective for all patients. As a consequence, there is an increased interest in identifying subgroups that receive additional benefit from the experimental treatment. Identifying characteristics associated with improved response is critical to subsequently allow making treatment decisions using patient profiles. This paradigm - stratified medicine - is especially promising for cancer treatment.

The counterfactual model is gaining in importance as a way to identify those subgroups. The underlying principle would be to estimate the outcome of a patient under the experimental treatment, when the patient was actually allocated to the control group, and the other way round. The use of counterfactual models for treatment selection has been proposed for binary and continuous outcomes (1-4).

In cancer trials, time-to-event outcomes (eg, time to progression, relapse, failure, or death) are common. For a given patient, a causal effect would be a measure of the discrepancy between the times to event under the experimental and control treatments (only one being observed). In absence of censoring, simple sample averages could be used to estimate the mean time to event of the two groups. However, in presence of right-censoring, the average causal effect estimate may not be identifiable.

An alternative is to use the restricted mean survival times (RMST), i.e. the expectation of survival time restricted to some time horizon. If the time horizon is 12 months, the RMST measures the average number of months survived over one year. Thus, the difference in RMST measures the gain in one-year life expectancy associated with the experimental group. Restricted survival times have been advocated as a relevant survival outcome for RCTs (5,6).


This proposal involves a methodological study done by our research team. We develop a statistical method to identify and characterize patient subgroups with an improved treatment response in RCTs with time-to-event outcomes. This method relies on a counterfactual model for the restricted survival time of a given patient under each treatment; it will allow estimating the benefit of the experimental treatment as the difference in counterfactual restricted survival times between the experimental and the control treatment for this particular patient. Data from the selected studies for which we request access will allow illustrating the application of this novel method and will provide additional information for the methodological study we are conducting.

Expected results

The potential benefit of this methodological development for public health could be substantial. If responsive subpopulations are identified, it would inform the treatment selection process and influence clinical decision-making and clinical practice.

Study Data Provided

Study ROCHE-ML18147: A randomized, open-label phase III Intergroup study: Effect of adding Bevacizumab to cross over fluoropyrimidine based chemotherapy in patients with mCRC and disease progression under first-line standard CTx/Bevacizumab combination
Study ROCHE-ML17102: A randomized, open-label study of the effect of MabThera, with and without fludarabine/cyclophosphamide, on progression-free survival in patients with chronic lymphocytic leukemia
Study ROCHE-NO25026 (BRIM3): A randomized, open-label, controlled, multicenter, global study on progression-free and overall survival in previously untreated patients with unresectable stage IIIC or stage IV melanoma with V600E BRAF mutation receiving RO5185426 or dacarbazine
Study BI-1200.23: BIBW 2992 and BSC Versus Placebo and BSC in Non-small Cell Lung Cancer Patients Failing Erlotinib or Gefitinib (LUX-LUNG 1)
Study LILLY-H3E-MC-JMHD: A Randomized Phase III Study of Pemetrexed plus Carboplatin and Bevacizumab Followed by Maintenance Pemetrexed and Bevacizumab versus Paclitaxel plus Carboplatin and Bevacizumab Followed by Maintenance Bevacizumab in Patients with Stage IIIB or IV Nonsquamous Non-small Cell Lung Cancer
Study LILLY-H3E-EW-S124: A Phase 3, Double-Blind, Placebo-Controlled Study of Maintenance Pemetrexed Plus Best Supportive Care Versus Best Supportive Care Immediately Following Induction Treatment With Pemetrexed + Cisplatin for Advanced Non-squamous Non-Small Cell Lung Cancer.
Study LILLY-H3E-MC-S103: A Randomized Phase 2 Study Comparing Erlotinib-Pemetrexed, Pemetrexed alone, and Erlotinib alone, as Second-Line Treatment for Non-Smoker Patients with Locally Advanced or Metastatic Nonsquamous Non-Small Cell Lung Cancer

Statistical Analysis Plan

The statistical analysis plan will be added after the research is published.

Publication Citation

The publication citation will be added after the research is published.

Summary Results

Results summary or link will be posted when available.