Predicting Chemotherapy Toxicity And Death In Older Adults With Colon Cancer: Results Of Most (Massilia Oncologic Senior Tests) Study.

10041 Background: Older patients with colon cancer are more vulnerable to chemotherapy toxicity and early death. Establishing simple scores specific for colon cancer (cc) patients able to predict severe chemotoxicity or early death is needed to select the best appropriate treatment. Methods: This prospective multicenter study included cc patients aged ≥ 70 years receiving first-line adjuvant or metastatic chemotherapy. Five frailty markers (FM) : nutrition, physical activity, mobility, energy, grip strength), six domains of CGA (functional status, comorbidities, falls, nutrition, cognition, depression) and laboratory parameters were collected at admission. Logistic or Cox regression was used to examine at 500 days the association between FM, CGA, laboratory parameters and grade 3-4 toxicity or death, respectively. Results: 93 patients (mean age 78.8 years, range 70-90) received adjuvant (32, 34.4 %) or metastatic (61, 65.6 %) chemotherapy. During the first 500 days grade 3-4 toxicity occurred in 49 (53.3 %) and 23 (26.7 %) died. Multivariate logistic regression identified polychemotherapy, albuminemia < 32 g/L, abnormal grip strength and C-reactive protein > 11 mg/L (odds ratio: 6.85, 2.79, 2.40, 1.43, respectively) as independent predictors of toxicity. Cox regression identified metastasis, alkaline phosphatases > 100 IU/ml, age > 82 years, sex (female), and abnormal grip strength (hazard ratio: 7.99, 3.45, 3.31, 3.2, 1.18, respectively) as independent predictors of premature death. The predictive model for grade 3-4 toxicity combined (polychemotherapy x 3) + (albuminemia x 2) + (abnormal grip strength x 1.5) + C-reactive protein, cut-off > 3. The predictive model for premature death combined (metastasis x 5) + (age x 2.5) + alkaline phosphatases + sex + abnormal grip strength, cut-off > 6. Characteristics of both models are : Clinical trial information: NCT02148731.Conclusions: These two simple and efficient scores will help clinicians to better identify cc older patients with increased risk of toxicity and/or premature death. Risk AUC ROC curve (± SD) 95 % CI P Sensitivity (%) Specificity (%) Toxicity 0.747 ± 0.052 0.647-0.832 < 0.0001 71.4 72.1 Death 0.908 ± 0.032 0.826-0.960 < 0.0001 87.0 81.0