Missing Data and Censoring in the Analysis of Progression-Free Survival in Oncology Clinical Trials

Association of differential censoring with survival and suboptimal control arms among oncology clinical trials

Differential censoring, which refers to censoring imbalance between treatment arms, may bias the interpretation of survival outcomes in clinical trials. In 146 phase III oncology trials with statistically significant time-to-event surrogate primary endpoints, we evaluated the association between differential censoring in the surrogate primary endpoints, control arm adequacy, and the subsequent statistical significance of overall survival results. Twenty-four (16%) trials exhibited differential censoring that favored the control arm, whereas 15 (10%) exhibited differential censoring that favored the experimental arm. Positive overall survival was more common in control arm differential censoring trials (63%) than in trials without differential censoring (37%) or with experimental arm differential censoring (47%; odds ratio = 2.64, 95% confidence interval = 1.10 to 7.20; P = .04). Control arm differential censoring trials more frequently used suboptimal control arms at 46% compared with 20% without differential censoring and 13% with experimental arm differential censoring (odds ratio = 3.60, 95% confidence interval = 1.29 to 10.0; P = .007). The presence of control arm differential censoring in trials with surrogate primary endpoints, especially in those with overall survival conversion, may indicate an inadequate control arm and should be examined and explained.

Clinical Trial Endpoints in Metastatic Cancer: Using Individual Participant Data to Inform Future Trials Methodology

Meta-analysis based on individual participant data (IPD) is a powerful methodology for synthesizing evidence by combining information drawn from multiple trials. Hitherto, its principal application has been in questions of clinical management, but an increasingly important use is in clarifying trials methodology, for instance in the selection of endpoints, as discussed in this review. In oncology, the Aide et Recherche en Cancérologie Digestive (ARCAD) Metastatic Colorectal Cancer Database is a leader in the use of IPD-based meta-analysis in methodological research. The ARCAD database contains IPD from more than 38 000 patients enrolled in 46 studies and continues to collect phase III trial data. Here, we review the principal findings of the ARCAD project in respect of endpoint selection and examine their implications for cancer trials. Analysis of the database has confirmed that progression-free survival (PFS) is no longer a valid surrogate endpoint predictive of overall survival in the first-line treatment of colorectal cancer. Nonetheless, PFS remains an endpoint of choice for most first-line trials in metastatic colorectal cancer and other solid tumors. Only substantial PFS effects are likely to translate into clinically meaningful benefits, and accordingly, we advocate an oncology research model designed to identify highly effective treatments in carefully defined patient groups. We also review the use of the ARCAD database in assessing clinical response including novel response metrics and prognostic markers. These studies demonstrate the value of IPD as a tool for methodological studies and provide a reference point for the expansion of this approach within clinical cancer research.

Impact of COVID-19 pandemic on oncology clinical trial design, data collection and analysis

BACKGROUND

To identify and assess via simulation the impact of COVID-19 pandemic on oncology trials and discuss potential mitigation strategies for study design, data collection, endpoints and analyses.

METHODS

We simulated clinical trials to evaluate the COVID-19 impact on overall survival and progression-free survival. We evaluated survival in single-region trials with different proportions of impacted patients across treatment arms, and in multi-region randomized trials with different proportions of impacted patients across regions. We also assessed the impact on PFS when the missingness of disease assessment and censoring rules vary. Impact on the trial success and robustness of statistical inference was summarized.

RESULTS

Without regional impact, the impact on OS analysis is minimal if proportions of impacted patients are similar across arms, however, if a larger proportion of treatment arm patients are impacted, trials may suffer substantial power loss and underestimate treatment effect size. For multi-region trials, if more treatment arm patients are enrolled from more severely impacted regions, trials also have poorer performance. For PFS analysis, the intent-to-treat rule performs well even when the treatment arm patients are more likely to miss disease assessments, while the consecutive-missing censoring rule may lead to poorer performance.

CONCLUSION

COVID-19 affects oncology trials. Simulations would be highly informative to Data Monitoring Committee in understanding the impact and making appropriate recommendations, upon which the sponsor could start planning potential remedies. We also recommend a decision tree for choosing the appropriate methods for PFS evaluation in the presence of missing disease assessments due to COVID-19.

Quantifying Withdrawal of Consent, Loss to Follow-Up, Early Drug Discontinuation, and Censoring in Oncology Trials

BACKGROUND

Censoring due to early drug discontinuation (EDD) or withdrawal of consent or loss to follow-up (WCLFU) can result in postrandomization bias. In oncology, censoring rules vary with no defined standards. In this study, we sought to describe the planned handling and transparency of censoring data in oncology trials supporting FDA approval and to compare EDD and WCLFU in experimental and control arms.

METHODS

We searched FDA archives to identify solid tumor drug approvals and their associated trials between 2015 and 2019, and extracted the planned handling and reporting of censored data. We compared the proportion of WCLFU and EDD between the experimental and control arms by using generalized estimating equations, and performed logistic regression to identify trial characteristics associated with WCLFU occurring more frequently in the control group.

RESULTS

Censoring rules were defined adequately in 48 (59%) of 81 included studies. Only 14 (17%) reported proportions of censored participants clearly. The proportion of WCLFU was higher in the control group than in the experimental group (mean, 3.9% vs 2.5%; β-coefficient, -2.2; 95% CI, -3.1 to -1.3; P<.001). EDD was numerically higher in the experimental arm in 61% of studies, but there was no statistically significant difference in the proportion of EDD between the experimental and control groups (mean, 21.6% vs 19.9%, respectively; β-coefficient, 0.27; 95% CI, -0.32 to 0.87; P=.37). The proportion of EDD due to adverse effects (AEs) was higher in the experimental group (mean, 13.2% vs 8.5%; β-coefficient, 1.5; 95% CI, 0.57-2.45; P=.002). WCLFU was higher in the control group in studies with an active control group (odds ratio [OR], 10.1; P<.001) and in open label studies (OR, 3.00; P=.08).

CONCLUSIONS

There are significant differences in WCLFU and EDD for AEs between the experimental and control arms in oncology trials. This may introduce postrandomization bias. Trials should improve the reporting and handling of censored data so that clinicians and patients are fully informed regarding the expected benefits of a treatment.

Informative censoring of surrogate end-point data in phase 3 oncology trials

BACKGROUND

Kaplan-Meier (K-M) analysis, the cornerstone of cancer clinical trial interpretation, assumes that censored patients are no more or less likely to experience an event than those followed. We sought to investigate the patterns of censoring in surrogate end-points of oncology randomised controlled trials (RCTs) and examine the relationship between censoring in practice-changing treatments that failed to demonstrate survival gain.

METHODS

In this cross-sectional study of phase III RCTs published in the New England Journal of Medicine, Lancet, and JAMA, between 2010 and 2020, K-M curves of surrogate end-points with statistical significance were extracted. The reverse K-M method (i.e., events and censoring are flipped) was used to examine differential censoring using the analogous reverse hazard ratio and restricted mean survival time. Sensitivity analysis was performed by partially restoring the balance in censoring between study arms.

RESULTS

Of the 73 eligible studies with significant surrogates, 33 (45%) reported significant overall survival benefit (concordant trials), and 40 (55%) did not (discordant trials). The proportion of studies with significant differential censoring in surrogates was 43% (17/40) and 51% (17/33) in discordant and concordant trials, respectively. Trials with a significant censoring imbalance in the experimental arm occurred only in discordant trials (15% vs 0%, odds ratio [OR] = 12.62, P = 0.033), compared to excessive censoring in the control arm which occurred more in concordant trials (28% vs 52%; OR = 0.36, P = 0.036). Although censoring imbalance occurred in both groups, after sensitivity analysis, 50% of the discordant trials lost their statistical significance, compared to 15% of concordant trials (OR = 5.6, P = 0.0018).

CONCLUSION

Censoring imbalance between study arms of RCTs suggests a potential systemic bias and raises uncertainty regarding the validity of the results. Informative censoring may explain the inconsistency between therapies that seem to improve disease outcomes without concomitant survival benefit and should trigger further investigation.

A Practical Overview and Reporting Strategies for Statistical Analysis of Survival Studies

Survival (time-to-event) analysis is commonly used in clinical research. Key features of performing a survival analysis include checking proportional hazards assumptions, reporting CIs for hazards ratios and relative risks, graphically displaying the findings, and analyzing with consideration of competing risks. This article provides a brief overview of important statistical considerations for survival analysis. Censoring schemes, different methods of survival function estimation, and ways to compare survival curves are described. We also explain competing risk and how to model survival data in the presence of it. Different kinds of bias that influence survival estimation and avenues to model the data under these circumstances are also described. Several analysis techniques are accompanied by graphical representations illustrating proper reporting strategies. We provide a list of guiding statements for researchers and reviewers.

Exaggeration of PFS by blinded, independent, central review (BICR)

BACKGROUND

Recent published studies have shown meaningful discrepancies between local investigator and blinded, independent, central review (BICR) assessed median progression-free survival (PFS). When the local review but not BICR shows progression, generally, no further assessments are carried out and patients are censored in the BICR analysis, leading to violation of the statistical assumptions of independence between censoring and outcome used in survival analysis methods.

METHODS

We carried out a simulation study to assess methodological reasons behind these discrepancies and corroborated our findings in a case study of three BRCA-mutated ovarian cancer trials. We briefly outline possible methodological solutions that may lead to improved estimation of the BICR medians.

RESULTS

The Kaplan-Meier (KM) curve for the BICR PFS can often be exaggerated. The degree of bias is largest when there is reasonably strong correlation between BICR and local PFS, especially when PFS is long compared with assessment frequency. This can result in an exaggeration of the medians and their difference; however, the hazard ratio (HR) is much less susceptible to bias. Our simulation shows that when the true BICR median PFS was 19 months, and patients assessed every 12 weeks, the estimated KM curves were materially biased whenever the correlation between BICR and local PFS was 0.4 or greater. This was corroborated by case studies where, in the active arm, the BICR median PFS was between 6 and 11 months greater than the local median PFS. Further research is required to find improved methods for estimating BICR survival curves.

CONCLUSIONS

In general, when there is a difference between local and BICR medians, the true BICR KM curve is likely to be exaggerated and its true median will probably lie somewhere between the observed local and BICR medians. Presentation of data should always include both BICR and local results whenever a BICR is carried out.

Systematic bias between blinded independent central review and local assessment: literature review and analyses of 76 phase III randomised controlled trials in 45 688 patients with advanced solid tumour

OBJECTIVE

Unbiased assessment of tumour response is crucial in randomised controlled trials (RCTs). Blinded independent central review is usually used as a supplemental or monitor to local assessment but is costly. The aim of this study is to investigate whether systematic bias existed in RCTs by comparing the treatment effects of efficacy endpoints between central and local assessments.

DESIGN

Literature review, pooling analysis and correlation analysis.

DATA SOURCES

PubMed, from 1 January 2010 to 30 June 2017.

ELIGIBILITY CRITERIA FOR SELECTING STUDIES

Eligible articles are phase III RCTs comparing anticancer agents for advanced solid tumours. Additionally, the articles should report objective response rate (ORR), disease control rate (DCR), progression-free survival (PFS) or time to progression (TTP); the treatment effect of these endpoints, OR or HR, should be based on central and local assessments.

RESULTS

Of 76 included trials involving 45 688 patients, 17 (22%) trials reported their endpoints with statistically inconsistent inferences (p value lower/higher than the probability of type I error) between central and local assessments; among them, 9 (53%) trials had statistically significant inference based on central assessment. Pooling analysis presented no systematic bias when comparing treatment effects of both assessments (ORR: OR=1.02 (95% CI 0.97 to 1.07), p=0.42, I2=0%; DCR: OR=0.97 (95% CI 0.92 to 1.03), p=0.32, I2=0%); PFS: HR=1.01 (95% CI 0.99 to 1.02), p=0.32, I2=0%; TTP: HR=1.04 (95% CI 0.95 to 1.14), p=0.37, I2=0%), regardless of funding source, mask, region, tumour type, study design, number of enrolled patients, response assessment criteria, primary endpoint and trials with statistically consistent/inconsistent inferences. Correlation analysis also presented no sign of systematic bias between central and local assessments (ORR, DCR, PFS: r>0.90, p<0.01; TTP: r=0.90, p=0.29).

CONCLUSIONS

No systematic bias could be found between local and central assessments in phase III RCTs on solid tumours. However, statistically inconsistent inferences could be made in many trials between both assessments.

The changing face of clinical trials in the personalized medicine and immuno-oncology era: report from the international congress on clinical trials in Oncology & Hemato-Oncology (ICTO 2017)

In the past decade, the oncology community has witnessed major advances in the understanding of cancer biology and major breakthroughs in several different therapeutic areas, from solid tumors to hematological malignancies; moreover, the advent of effective immunotherapy approaches, such as immune-checkpoint blockade, is revolutionizing treatment algorithms in almost all oncology disease areas. As knowledge evolves and new weapons emerge in the “war against cancer”, clinical and translational research need to adapt to a rapidly changing environment to effectively translate novel concepts into sustainable and accessible therapeutic options for cancer patients.

With this in mind, translational cancer researchers, oncology professionals, treatment experts, CRO and industry leaders, as well as patient representatives gathered in London, 16-17 March 2017, for The International Congress on Clinical Trials in Oncology and Hemato-Oncology (ICTO2017), to discuss the changing face of oncology clinical trials in the new era of personalized medicine and immuno-oncology. A wide range of topics, including clinical trial design in immuno-oncology, biomarker-oriented drug development paths, statistical design and endpoint selection, challenges in the design and conduct of personalized medicine clinical trials, risk-based monitoring, financing and reimbursement, as well as best operational practices, were discussed in an open, highly interactive format, favoring networking among all relevant stakeholders. The most relevant data, approaches and issues emerged and discussed during the conference are summarized in this report.

The role of censoring on progression free survival: oncologist discretion advised

Censoring is increasingly appreciated as a potential bias affecting estimates of progression free survival (PFS) in randomised trials. In this commentary, we explore the central assumption of censoring. Censored patients are considered no more or less likely to undergo the event of interest than those who remain in the analysis. Instead however, if one makes alternate assumptions, that censored patients are different than those who remain on the trial, estimates of PFS change. Using the example of the recent BOLERO-2 trial of exemestane and everolimus, we show that by altering the assumptions for censoring, the major conclusions of clinical trials may change. As such, the number of censored patients at each time interval should be routinely reported in randomised trials to better understand the implications of censoring.