From protocol to practice: The gap in obtaining progression disease biopsies across 182 oncology trials

Background: Post-progression (PD) biopsies provide critical insights into treatment resistance and tumor evolution. Despite their growing inclusion in clinical trial designs, the extent to which PD biopsies are successfully obtained and the factors that influence said acquisition are unknown. Methods: Protocols from consecutive oncology clinical trials that integrated at least one research biopsy [at either pre-treatment, on-treatment and/or post-progression timepoint(s)] from 1/2021 to 12/2022 were reviewed and categorized according to tumor type, trial phase, study type, planned biopsy timepoints, mandatory/optional biopsy status, biopsy rationale and endpoint inclusion. All enrolled patients at the University of Texas MD Anderson Cancer Center were reviewed for research biopsy execution. Statistical associations were tested using Fisher's exact or chi-square analyses, with p<0.05 considered significant. Results: A total of 182 clinical trial protocols were analyzed, of which 119 (65%) included a PD research biopsy. Of these 119 protocols, 81% involved solid tumors, 87% were multicenter, 59% were phase 1 trials, 27% specified PD biopsy as mandatory, and 7% included use of PD biopsy as a study endpoint. Incorporating a PD biopsy in a protocol was significantly associated with cancer type (hematologic (81%) vs solid tumors (19%), p = 0.042) and study type (multicenter (87%) vs. single center (13%), p = 0.003). In these 119 protocols a total of 797 patients were enrolled at MDACC (median 3, range 0-66), of whom 465 (58%) were treated (median 2, range 0-25). At least one research biopsy [screening (n=263), on-treatment (n=126), or PD (n=48)] was performed in 74 trials (62%), with a median of 2 biopsies per trial (range 1-46). PD biopsies were obtained in 24 of 119 protocols (20%) with a median of 1 PD biopsy per trial (range 1-12). Phase 1 trial type was the only protocol factor significantly associated with obtaining any research biopsy (p = 0.01). No protocol factors were statistically associated with a patient undergoing a PD biopsy. Among protocols where PD biopsy was mandated, 9/32 (28%) executed a PD biopsy versus 15/87 (17%) in optional protocols (p = 0.204). Similarly, 3/8 (38%) of protocols listing PD biopsy as a study endpoint executed a PD biopsy versus 21/111 (19%) of those without endpoint incorporation (p = 0.201). Conclusions: Despite widespread protocol inclusion, PD biopsies were successfully obtained in only one-fifth of eligible trials, underscoring the gap between design intent and clinical feasibility, which limits the ability to study mechanisms of resistance and tumor evolution. These findings highlight the need for pragmatic trial designs (PD biopsies at time of tumor marker rise or increasing RECIST percentage) and institutional infrastructure that support consistent tissue collection at progression.