Development of an oncological-multidimensional prognostic index (Onco-MPI) for mortality prediction in older cancer patients

Purpose: A multidimensional prognostic index (MPI) based on a comprehensive geriatric assessment (CGA) has been developed and validated in independent cohorts of older patients demonstrating good accuracy in predicting one-year mortality. The aim of this study was to develop a cancer-specific modified MPI (Onco-MPI) for mortality prediction in older cancer patients. Methods: We enrolled 658 new cancer subjects ≥70 years (mean age 77.1 years, 433 females, 65.8 %) attending oncological outpatient services from September 2004 to June 2011. The Onco-MPI was calculated according to a validated algorithm as a weighted linear combination of the following CGA domains: age, sex, basal and instrumental activities of daily living, Eastern Cooperative Oncology Group performance status, mini-mental state examination, body mass index, Cumulative Illness Rating Scale, number of drugs and the presence of caregiver. Cancer sites (breast 46.5 %, colorectal 21.3 %, lung 6.4 %, prostate 5.5 %, urinary tract 5.0 %, other 15.3 %) and cancer stages (I 37 %, II 22 %, III 19 %, IV 22 %) were also included in the model. All-cause mortality was recorded. Three grades of severity of the Onco-MPI score (low risk: 0.0–0.46, medium risk: 0.47–0.63, high risk: 0.64–1.0) were calculated using RECPAM method. Discriminatory power and calibration were assessed by estimating survival C-indices, along with 95 % confidence interval (CI) and the survival-based Hosmer–Lemeshow (HL) measures. Results: One-year mortality incidence rate was 17.4 %. A significant difference in mortality rates was observed in Onco-MPI low risk compared to medium- and high-risk patients (2.1 vs. 17.7 vs. 80.8 %, p < 0.0001). The discriminatory power of one-year mortality prediction of the Onco-MPI was very good (survival C-index 0.87, 95 % CI 0.84–0.90) with an excellent calibration (HL p value 0.854). Conclusion: Onco-MPI appears to be a highly accurate and well-calibrated predictive tool for one-year mortality in older cancer patients that can be useful for clinical decision making in this age group.