Interim monitoring in a treatment strategy trial with a composite primary endpoint

Minhee Kang, Birgit Grund, Sally Hunsberger, David Glidden, Paul Volberding

Research output: Contribution to journalArticlepeer-review


When a clinical trial has a composite endpoint and a comparison of treatment strategies with multiple intervention components, interim data reviews by a data safety and monitoring board (DSMB) can be challenging as the data evolve on multiple fronts. We illustrate with a study in the treatment of Kaposi sarcoma (KS), an HIV-associated cancer with a multi-faceted disease presentation. The study, ACTG-A5264/AMC-067, was a 1:1 randomized trial to compare two strategies: immediate initiation of etoposide with antiretroviral therapy (ART), or ART with delayed etoposide upon disease progression. The outcome was a composite endpoint that included the following events, ordered from worst to best in the following three categories: (1) KS progression at 48 weeks, death, initiation of alternate KS treatment, loss to study follow-up; (2) stable KS; and (3) partial or complete KS response at 48 weeks. We present the interim results on the composite endpoint and the individual components, where components favored different study arms at an interim review. To facilitate interim data monitoring for complex trials, we recommend clear communications between the study team and the DSMB prior to the initiation of the trial on the need for a composite endpoint, the intentions behind the defined strategies, and relative importance of individual components of the composite endpoint. We also recommend flexibility in the timing of data reviews by the DSMB to interpret emerging data in multiple dimensions.

Original languageEnglish (US)
Article number105846
JournalContemporary Clinical Trials
StatePublished - Nov 2019

Bibliographical note

Publisher Copyright:
© 2019 Elsevier Inc.


  • Composite endpoint
  • Futility
  • Interim monitoring
  • Treatment strategy trial


Dive into the research topics of 'Interim monitoring in a treatment strategy trial with a composite primary endpoint'. Together they form a unique fingerprint.

Cite this