Analysis of Time-to-treatment Discontinuation of Targeted Therapy, Immunotherapy, and Chemotherapy in Clinical Trials of Patients With Non-Small-Cell Lung Cancer

G. M. Blumenthal; Y. Gong; K. Kehl; P. Mishra-Kalyani; K. B. Goldberg; S. Khozin; P. G. Kluetz; G. R. Oxnard; R. Pazdur


Ann Oncol. 2019;30(5):830-838. 

In This Article

Abstract and Introduction


Background: Pragmatic end points, such as time-to-treatment discontinuation (TTD), defined as the date of starting a medication to the date of treatment discontinuation or death has been proposed as a potential efficacy end point for real-world evidence (RWE) trials, where imaging evaluation is less structured and standardized.

Patients and methods: We studied 18 randomized clinical trials of patients with metastatic non-small-cell lung cancer (mNSCLC), initiated after 2007 and submitted to U.S. Food and Drug Administration. TTD was calculated as date of randomization to date of discontinuation or death and compared to progression-free survival (PFS) and overall survival (OS) across all patients, as well as in treatment-defined subgroups [EGFR mutation-positive treated with tyrosine kinase inhibitor (TKI), EGFR wild-type treated with TKI, ALK-positive treated with TKI, immune checkpoint inhibitor (ICI), chemotherapy doublet with maintenance, chemotherapy monotherapy].

Results: Overall across 8947 patients, TTD was more closely associated with PFS (r = 0.87, 95% CI 0.86–0.87) than with OS (0.68, 95% CI 0.67–0.69). Early TTD (PFS—TTD ≥ 3 months) occurred in 7.7% of patients overall, and was more common with chemo monotherapy (15.0%) while late TTD (TTD—PFS ≥ 3 months) occurred in 6.0% of patients overall, and was more common in EGFR-positive and ALK-positive patients (12.4% and 22.9%). In oncogene-targeted subgroups (EGFR positive and ALK positive), median TTDs (13.4 and 14.1 months) exceeded median PFS (11.4 and 11.3 months).

Conclusions: At the patient level, TTD is associated with PFS across therapeutic classes. Median TTD exceeds median PFS for biomarker-selected patients receiving oncogene-targeted therapies. TTD should be prospectively studied further as an end point for pragmatic randomized RWE trials only for continuously administered therapies.


Progression-free survival (PFS) is a common end point in oncology trials to measure the efficacy of a drug. PFS, typically measured by the response evaluation criteria in solid tumors (RECIST), quantifies the time from random assignment until objective progression or death.[1,2] In contrast to overall survival (OS), PFS is not impacted by cross-over or subsequent treatments.[3,4] The RECIST criteria specify a definition of disease progression which is historically consistent, though it may not fully represent the biology of tumor progression and does not account for other clinically relevant reasons for treatment discontinuation (e.g. toxicity).[5]

A new generation of more tolerable targeted therapies, including oral kinase inhibitors and immune checkpoint inhibitors (ICIs), has led to a reconsideration of the clinical significance of objective RECIST progression.[6,7] Because these drugs can be well tolerated, induce dramatic responses and alter tumor growth kinetics, oncologists now commonly maintain patients on treatment beyond RECIST progression, a phenomenon seen both in clinical practice and in clinical trials.[8] In one analysis, continued treatment beyond objective progression was found to be associated with slower tumor growth, a type of disease progression that may portend a better prognosis as compared to rapid tumor growth or progression with new metastatic sites of disease.[9]

Scrutiny of conventional criteria for disease progression has heightened interest in alternate efficacy metrics, including end points that can be generated using real-world data (RWD).[10] RWD refers to data relating to patient health status or healthcare delivery collected from a variety of sources, including the electronic health record (EHR) digital health technology (e.g. wearable devices), or social media. Real-world evidence (RWE) is the clinical evidence regarding the usage, and potential benefits or risks, of a medical product derived from analysis of RWD.[11] Since clinical trial evidence supporting FDA approval is generated from smaller and more homogenous patient populations than the population that will be exposed to a drug once it enters the clinic, interest in using RWE to support regulatory decisions has increased. RWE could address gaps by supporting evidence generation for subsequent indications, optimal dosing, and studying special populations.

One end point that can be calculated using RWD is time-to-treatment discontinuation (TTD) defined as the time from the start of a therapy to the time of treatment discontinuation for any reason. We chose to further explore TTD as an efficacy end point for several reasons. First, data elements for TTD are consistently captured in clinical trials as well as from the EHR during routine care, and in insurance records and claims databases. Second, TTD may track closely with PFS, a generally accepted end point in oncology clinical trials. Finally, since traditional efficacy measures such as objective response rate and PFS per RECIST criteria are rarely if ever measured in routine cancer care, TTD may be a potential pragmatic end point for RWE studies.

These considerations motivated our analysis of the relationship of TTD, PFS, and OS using clinical trial datasets submitted to FDA as part of New Drug Applications (NDA) or Biologic License Applications (BLA). Because of the range of treatment regimens approved for metastatic non-small-cell lung cancer (mNSCLC) as well as the increased practice of treatment beyond progression, we focused on mNSCLC trials for initial proof of principle. We chose TTD rather than a related end point, Time-to-next treatment (TTNT) since the date of subsequent treatments can be heterogeneously captured in trial datasets submitted to FDA. It is acknowledged that TTNT is another potential RWD end point that may capture additional information including potential treatment-free intervals which have been described with immunotherapy.[12]