Frailty assessment in patients with advanced/metastatic renal cell carcinoma using routine data collected during treatment with tyrosine kinase inhibitors in the STAR trial.

Abstract

4531 Background: The STAR trial (ISRCTN06473203) was a phase 2/3 randomised controlled non-inferiority trial (N=920) which assessed whether treatment breaks from tyrosine kinase inhibitors (TKIs, oral sunitinib/pazopanib) could be safely used in locally advanced/metastatic renal cell carcinoma (RCC). We aimed to determine if it was possible to obtain a frailty index (FI) which could be linked to toxicity and clinical outcomes from routinely collected large trial data, using the STAR trial as an example. Methods: We developed the FI using Rockwood’s accumulation of deficits methodology. STAR data was initially searched for variables measuring baseline health problems and functional limitations. Variables were excluded if there was a significant proportion of missing values (>10%); if too rare or too common (<1%, >80%); or if significantly correlated with another variable ( r >0.95). Variables were coded from 0 (no deficit) to 1 (full deficit); these were summed then divided by the number of variables assessed to give the FI. Frailty thresholds were adopted in line with the literature: Not Frail (FI≤0.08); Pre-Frail (FI 0.08-0.24); Frail (FI≥0.25). Results: Of 57 variables screened, 35 variables remained for inclusion in the FI. 50 participants missing >20% of FI variables were excluded, leaving n=870. FI scores ranged from 0 to 0.43 (median 0.15, IQR 0.10-0.22). Kaplan-Meier survival analysis showed a statistically significant difference by FI for overall survival (OS) (FI≥0.25: HR 2.48, p<0.001, 95% CI 1.89-3.26). In multivariate Cox proportional hazards regression, FI remained a statistically significant risk factor for OS (HR 1.41, P=0.047, 95% CI 1.00-1.99); other statistically significant variables included age group, sites of metastatic disease, IMDC and MOTZER scores. Toxicity was assessed over the first 6 months, while all participants were treated with continuous TKIs. The most common severe toxicities (G3+) were hypertension (200/870), hepatobiliary disorders (97/870) and fatigue (63/870). Time free of severe toxicity was statistically significantly shorter for frail participants. Conclusions: Amongst STAR trial participants with advanced/metastatic RCC treated with TKIs, frailty was a statistically significant risk factor for poorer survival and for shorter time to severe toxicity. It is noted that the eligibility criteria for STAR included a baseline PS of ECOG 0-1 which will have excluded many frailer patients. This data shows that it is feasible to use routinely collected trial data to create a clinically meaningful FI and this approach could be applied to other trial datasets. Clinical trial information: ISRCTN06473203 . Toxicity-free survival, by frailty group. Toxicity FI group HR p-value 95% CI Severe toxicity (G3+) Intermediate 1.00 0.960 0.83-1.21 Frail 1.28 0.025 1.03-1.59 All toxicity (G1+) Intermediate 1.11 0.031 1.01-1.22 Frail 1.08 0.201 0.96-1.20