Modified Frailty Index (mFI) as a Survival and Functional Outcome Predictor in Older Patients Undergoing Surgery for Brain Metastases

Abstract BACKGROUND Surgical decision making in older patients with brain metastases (BM) is challenging. Frailty and comorbidities are often thought to limit therapeutic options as well as functional and survival outcomes. MATERIAL AND METHODS 139 consecutive patients undergoing surgery 2017-2021 for BM were analyzed retrospectively. Cut-off for age was ≥60 yrs. in accordance with the graded prognostic assessment (GPA) index. Frailty was investigated using the mFI index, a compound measure assessing comorbidities and the patients’ functional health status. RESULTS Median age was 69.0 (IQR: 63.0-74.5) yrs. 55.4% had multiple metastases. We recorded 9.4% major (CTCAE III-V) surgical and 10.1% medical complications. 7.9% of patients had major new focal neurological deficits ≥30 days. Median overall survival (mOS) was 7.8 (IQR: 5.0-10.6) months. Frailer patients lived shorter (mOS for “least-frail”/mFl 0: 13.6 [IQR: 8.6-18.6], “moderately frail”/mFl 1-2: 9.0 [IQR: 4.9-13.0] and “frailest”/mFl ≥3: 2.7 [IQR: 0.9-4.5] months; p<0.001). Multivariate analysis with all parameters correlating significantly with mOS in the univariate analysis (mFI, higher preop. KPS, younger age [<69 yrs.], female sex) revealed a higher preop. KPS to be the only independent significant survival predictor (70 vs. 80-100; HR 2.1, p<0.001). Frailer patients had poorer functional outcomes (patients w/ postop. KPS ≤70 vs. 80-100; mFI 0: 9/36 [25.0%], mFI 1-2: 20/65 [30.8%], mFI ≥3: 25/38 [65.8%]; p<0.001). Multivariate analysis with all parameters correlating significantly with better functional outcome in the univariate analysis (mFI, higher preop. KPS, younger age, female sex and single metastasis) showed that a higher preop. KPS was the strongest prognostic parameter (RR 17.2, p<0.001). The mFI was not an independent outcome predictor. In a ROC analysis the preoperative KPS proved a much better predictor of functional outcome than the mFI (AUC 92 vs. 71%; sensitivity/specificity optimum 80 vs. 2, resp.). mFI was not a predictor of complications (p=NS). CONCLUSION The mFI proved a strong survival and functional outcome predictor in older patients. However, this effect was no longer seen in multivariate analyses when the KPS was included as a covariate. This may suggest that surgical outcomes depend on the patients’ presurgical functional health status rather than comorbidities. The presence of comorbidities (if well treated) should therefore not deter from BM surgery in older patients.