Phase 2 Colorectal Cancer Trial Misses Overall Endpoint, but Plasma DKK1 Offers a Clue for Antibody Drug Stratification

For cancer drugs, a negative primary endpoint does not necessarily mean that every signal has disappeared. Peer-reviewed results from the randomized phase 2 DeFianCe trial showed that adding the anti-DKK1 antibody sirexatamab to chemotherapy and bevacizumab did not significantly prolong progression-free survival in the overall population of patients with metastatic colorectal cancer. However, the treatment effect appeared to strengthen as baseline plasma DKK1 concentrations increased, providing an opening for subsequent precision-stratification research.

The trial enrolled 188 patients whose disease had progressed after first-line systemic treatment and assigned them in a one-to-one ratio. In both groups, investigators selected either FOLFIRI or mFOLFOX6 chemotherapy in combination with bevacizumab, with sirexatamab additionally administered in the experimental group. Median progression-free survival was 9.2 months in the overall population and 8.3 months in the control group, with a hazard ratio of 0.84 and a one-sided p value of 0.1712, failing to meet the prespecified primary endpoint. Objective response rates were 35.1% and 26.6%, respectively, and there was also no significant difference in overall survival.

The turning point came from an analysis of plasma DKK1, which had been prespecified as a candidate biomarker but remained exploratory. Among 88 patients whose baseline concentrations were above the median, the objective response rate was 38.0% in the sirexatamab group and 23.7% in the control group; the hazard ratio was 0.61 for progression-free survival and 0.42 for overall survival. Data previously released by the company also showed that the efficacy gap widened further among the 44 patients in the highest DKK1 quartile, although the sample size also decreased accordingly.

DKK1 is a secreted protein that can be measured in blood, and sirexatamab is designed to bind to and neutralize free DKK1. These results therefore raise an important hypothesis: the drug may not be suitable for a broad, unselected patient population, but may be more likely to act in tumor-biological settings with higher DKK1 levels. If this association can be prospectively reproduced, plasma testing could potentially influence both patient selection and the drug-development pathway.

However, it remains unclear whether high DKK1 predicts the treatment benefit of sirexatamab or primarily reflects a poorer underlying prognosis. The study had an open-label design, and the primary endpoint was assessed by trial physicians. The final analysis accumulated only 119 progression-free survival events, fewer than the originally planned 145, which may also have weakened the statistical power of the overall comparison. More importantly, the subgroups defined by median or quartile cutoffs were smaller, resulting in greater uncertainty in the effect estimates, and cannot replace an adequately powered validation trial with a prespecified cutoff.

Regarding safety, both the paper and company data stated that overall outcomes were broadly similar between the two groups, with no indication that adding sirexatamab clearly altered the adverse-event profile of the existing treatment. This places greater focus on efficacy stratification in the development program: the next step is not merely to conduct another phase 3 trial, but first to establish a reproducible DKK1 testing method and selection threshold, and then use a prospective design to determine whether this blood biomarker truly has predictive value.