BackgroundIndividual patient data (IPD) meta-analysis is considered to be a gold standard when the results of several randomized trials are combined. Recent initiatives on sharing IPD from clinical trials offer unprecedented opportunities for using such data in IPD meta-analyses.MethodsFirst, we discuss the evidence generated and the benefits obtained by a long-established prospective IPD meta-analysis in early breast cancer. Next, we discuss a data-sharing system that has been adopted by several pharmaceutical sponsors. We review a number of retrospective IPD meta-analyses that have already been proposed using this data-sharing system. Finally, we discuss the role of data sharing in IPD meta-analysis in the future.ResultsTreatment effects can be more reliably estimated in both types of IPD meta-analyses than with summary statistics extracted from published papers. Specifically, with rich covariate information available on each patient, prognostic and predictive factors can be identified or confirmed. Also, when several endpoints are available, surrogate endpoints can be assessed statistically.ConclusionsAlthough there are difficulties in conducting, analyzing, and interpreting retrospective IPD meta-analysis utilizing the currently available data-sharing systems, data sharing will play an important role in IPD meta-analysis in the future.

No abstract

Zolbetuximab is a chimeric monoclonal antibody that targets claudin‐18.2, a candidate biomarker in patients with advanced gastric/gastroesophageal cancer. This nonrandomized phase 1 study (NCT03528629) enrolled previously treated Japanese patients with claudin‐18.2–positive locally advanced/metastatic gastric/gastroesophageal cancer in two parts: Safety (Arms A and B, n = 3 each) and Expansion (n = 12). Patients received intravenous zolbetuximab 800 mg/m2 on cycle 1, day 1 followed by 600 mg/m2 every 3 weeks (Q3W; Safety Part Arm A and Expansion) or 1000 mg/m2 Q3W (Safety Part Arm B). For the Safety Part, the primary endpoint was safety (i.e., dose‐limiting toxicities [DLTs]) and a secondary endpoint was objective response rate (ORR) by investigator. For the Expansion Part, the primary endpoint was ORR by investigator and secondary endpoints included ORR by central review and safety. Additional secondary endpoints for both the Safety and Expansion Parts were disease control rate (DCR), overall survival (OS), progression‐free survival (PFS), duration of response, pharmacokinetics, and immunogenicity. In 18 patients, no DLTs (Safety Part) or drug‐related treatment‐emergent adverse events (TEAEs) grade ≥3 were observed. Most TEAEs were gastrointestinal. In 17 patients with measurable lesions, best overall response was stable disease (64.7%) or progressive disease (35.3%). The DCR was 64.7% (95% confidence interval 38.3–85.8). In Arm A and Expansion combined (n = 15), median OS was 4.4 months (2.6–11.4) and median PFS was 2.6 months (0.9–2.8). In Arm B (n = 3), median OS was 6.4 months (2.9–6.8) and median PFS was 1.7 months (1.2–2.1). Zolbetuximab exhibited no new safety signals with limited single‐agent activity in Japanese patients.

Risk difference is a relevant effect measure in epidemiologic research. Although it is well known that when there are few events per confounder, logistic regression is not suitable for confounding control, it is not clear how many events per confounder are required for valid estimation of risk difference using linear binomial models. Because the maximum likelihood method has a convergence problem, we investigated the number of events per confounder necessary to validly estimate risk difference using modified least-squares regression in a simulation. We simulated 864 scenarios, according to the number of confounders (2-20), the number of events per confounder (2-12), marginal risk (0.5%-40%), exposure proportion (20% and 40%), and 3 sizes of risk difference. Our simulation showed that modified least-squares regression provided unbiased risk difference-regardless of the number of events per confounder-and reliable confidence intervals when more than 5 events were expected in the exposed and in the unexposed, irrespective of the number of events per confounder. We illustrated the modified least-squares regression analysis using perinatal epidemiologic data. Modified least-squares regression is considered to be a useful analytical tool for rare binary outcomes relative to the number of confounders.