Association between diarrhea and survival in patients with HER2-positive advanced breast cancer treated with pyrotinib-based therapy: A landmark analysis from the real-world PRETTY study

BACKGROUND: Diarrhea is the most common adverse event of pyrotinib. This study analyzed the association between diarrhea and survival in patients with HER2-positive advanced breast cancer treated with pyrotinib-based therapy. METHODS: A secondary analysis was performed using the individual patient data from the nationwide, prospective real-world PRETTY study. Baseline and treatment characteristics were summarized in groups by severity of diarrhea. Multivariable Cox regression analysis was used to analyze the association of diarrhea with real-world progression-free survival (rwPFS) and overall survival (OS), respectively. Immortal-time bias was adjusted using the landmark analysis and time-dependent Cox model. Multiple imputation was performed for missing data, and pooling ln(hazard ratio [HR]) estimations on the association between diarrhea and survival were reported. RESULTS: Of 1129 patients, 826 (73.2%) had any-grade treatment-related diarrhea (including 174 [15.4%] with grade ≥3 diarrhea). After multiple imputation, the multivariable Cox regression analysis showed that diarrhea was independently associated with longer OS (grade 1-2 diarrhea vs. none: HR = 0.57 [95% CI, 0.38-0.86], P = 0.007; grade ≥3 diarrhea vs. none: HR = 0.56 [95% CI, 0.31-0.99], P = 0.047) but was not associated with rwPFS. Significant association between diarrhea and OS was also observed in the 1-month landmark analysis (grade 1-2 diarrhea vs. none: HR = 0.59 [95% CI, 0.39-0.89], P = 0.012; grade ≥3 diarrhea vs. none: HR = 0.54 [95% CI, 0.30-0.98], P = 0.043). Time-dependent Cox model indicated the trend on better OS when diarrhea occurred, but without statistical significance. CONCLUSION: For patients with HER2-positive advanced breast cancer, diarrhea that occurs during pyrotinib-based therapy maybe a prognostic marker of longer OS.