Missing tumor measurement (TM) data in the search for alternative TM-based endpoints in cancer clinical trials

Abstract

Purpose: Missing data commonly occur in cancer clinical trials (CCT) and may hinder the search for alternative trial endpoints. We consider reasons for missing tumor measurement (TM) data in CCT and how missing TM data are typically handled. We explore the potential impact of missing TM data on predictive ability of a set of TM-based endpoints.

Methods: Literature review identifies reasons for and approaches to handling missing TM data. Data from 3 actual clinical trials were used for illustration. A sensitivity analysis of the potential impact of missing TM data was performed by comparing overall survival (OS) predictive ability of alternative endpoints using observed and imputed data.

Results: Reasons for missing TM data in CCT are presented, based on the literature review and the three trials. Although missing TM data impacted individual objective status (e.g. 12-week status changed for 53% of patients in one imputation set), it surprisingly only minimally impacted endpoint predictive ability (e.g. median c-indices of 500 imputed datasets ranged from 0.566 to 0.570 for N9741, 0.592-0.616 for N9841, and 0.542-0.624 for N0026).

Conclusion: By understanding the reasons for missingness, we can better anticipate them and minimize their occurrence. Our preliminary analysis suggests missing TM data may not impact endpoint predictive ability, but could impact objective response status classification; however these findings require further validation. With response status accepted as an important phase II endpoint in the development of new cancer therapies (including immunotherapy), we urge that in CCT complete TM data collection and adherence to protocol-defined disease evaluation as closely as possible be a priority.

Keywords: Cancer trials; Missing data; Phase II; Tumor measurement-based endpoints.

Fig. 1

CONSORT Diagram for observed data in the case study across all 3 trials¹. ¹Above horizontal dotted line: exclusions due to missing data for reasons as stated. Below horizontal dotted line: additional exclusion due to no assessments after 12 weeks. See Appendix A3 for trial-specific CONSORT diagrams for categorical endpoint and continuous endpoint analyses using observed and imputed datasets.

Fig. 2

Observed lesion size (black solid circles) versus the imputed lesion size (red crosses) for a sample of patients, across 5 imputation sets. In the analysis, any observed lesion measurements were retained and not replaced by imputed values; any imputed values appearing below (for which an observed measurement was available) are solely for illustration purposes to facilitate comparison between imputed and observed values. The imputed values and the observed measurements are similar. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Distributions of statistical measures of overall survival (OS) predictive ability across 500 imputed datasets, by study and by endpoint. Statistical measures include the c-index, Hosmer-Lemeshow statistic, and the BIC. Endpoints include dichotomous (CR/PR vs. SD/PD), trichotomous (CR/PR vs. PD vs. SD), disease control rate (PD vs. CR/PR/SD), absolute change, and relative change.

Fig. 4

Statistical measures of overall survival (OS) predictive ability for categorical endpoints (observed vs. imputed; only 5 imputed datasets are shown, for illustrative purposes), by study. Endpoints include dichotomous (CR/PR vs. SD/PD), trichotomous (CR/PR vs. PD vs. SD), and disease control rate (PD vs. CR/PR/SD). The important observation is that, for a given endpoint and study, discriminatory ability (measured by the c-index) is similar across imputed datasets. Although the imputed datasets initially have the same sample size (“Total”), after responses are calculated based on imputed measurements, some patients are found to progress before 12 weeks and so are excluded from the 12-week landmark analysis.

Fig. 4

Statistical measures of overall survival (OS) predictive ability for categorical endpoints (observed vs. imputed; only 5 imputed datasets are shown, for illustrative purposes), by study. Endpoints include dichotomous (CR/PR vs. SD/PD), trichotomous (CR/PR vs. PD vs. SD), and disease control rate (PD vs. CR/PR/SD). The important observation is that, for a given endpoint and study, discriminatory ability (measured by the c-index) is similar across imputed datasets. Although the imputed datasets initially have the same sample size (“Total”), after responses are calculated based on imputed measurements, some patients are found to progress before 12 weeks and so are excluded from the 12-week landmark analysis.

Fig. 5

Statistical measures of overall survival (OS) predictive ability for continuous endpoints using different datasets (observed vs. imputed; only 5 imputed datasets are shown, for illustrative purposes), by study. Endpoints include RECIST response (CR/PR vs. SD/PD), absolute change (change in tumor size from 0 to 6 and 6–12 weeks), and relative change (relative change in tumor size from 0 to 6 and 6–12 weeks) endpoints. The important observation is that, for a given endpoint and study, discriminatory ability (measured by the c-index) is similar across imputed datasets. Although the imputed datasets initially have the same sample size (“Total”), after responses are calculated based on imputed measurements, some patients are found to progress before 12 weeks and so are excluded from the 12-week landmark analysis.

Fig. 5

Statistical measures of overall survival (OS) predictive ability for continuous endpoints using different datasets (observed vs. imputed; only 5 imputed datasets are shown, for illustrative purposes), by study. Endpoints include RECIST response (CR/PR vs. SD/PD), absolute change (change in tumor size from 0 to 6 and 6–12 weeks), and relative change (relative change in tumor size from 0 to 6 and 6–12 weeks) endpoints. The important observation is that, for a given endpoint and study, discriminatory ability (measured by the c-index) is similar across imputed datasets. Although the imputed datasets initially have the same sample size (“Total”), after responses are calculated based on imputed measurements, some patients are found to progress before 12 weeks and so are excluded from the 12-week landmark analysis.