Predictive factors for sorafenib-induced hand-foot skin reaction using ordered logistic regression analysis.

Abstract

Predictive factors for sorafenib-induced hand-foot skin reaction (HFSR) using ordered logistic regression analysis were studied. This retrospective analysis evaluated patients admitted to a university hospital in Japan from May 2008 through October 2013. Patients age 20 years or older with relapsed or metastatic renal cell carcinoma, unresectable hepatocellular carcinoma, or gastrointestinal stromal tumor resistant to imatinib and sunitinib were included. Data were manually collected from patients' clinical records and included sex, age, Eastern Cooperative Oncology Group (ECOG) performance status, initial daily dose of sorafenib, duration of sorafenib use, concomitant medications, number of metastases, sites of metastases, physical examination findings, and type of cancer. Laboratory test values related to the patient's medical condition that seemed to influence HFSR or the absorption and pharmacologic effects of sorafenib were also collected. HFSR severity was also assessed. Univariate ordered logistic analysis was performed for HFSR severity outcomes and each candidate independent variable. A multivariate ordered logistic regression model was then constructed using a stepwise forward selection procedure. Data were screened for multicollinearity. Data from 113 patients were evaluated. This analysis identified duration of sorafenib use (odds ratio [OR], 0.0531), use of a proton pump inhibitor (PPI) (OR, 0.351), ECOG performance status (OR, 0.555), C-reactive protein level (OR, 17.74), and male sex (OR, 0.403) as significant factors for the occurrence of HFSR. Multivariate logistic regression analysis revealed that short duration of sorafenib use, avoidance of PPIs, good ECOG performance status, high C-reactive protein level, and female sex were predictive factors for the development of HFSR.