ABSTRACT
OBJECTIVE: To evaluate the association of C-reactive protein/albumin ratio (CAR) with the prognosis of patients with colorectal cancer and compare the prognostic value of CAR with other inflammation-based prognostic scoring systems. METHODS: We retrospectively evaluated 163 newly diagnosed colorectal cancer patients in Nanfang Hospital between January, 2007 and December, 2014. All recommended cutoff values of the clinicopathological factors were defined using receiver- operating characteristic (ROC) curve analyses. We evaluated the prognostic value of CAR in comparison with Glasgow Prognostic Score (GPS) and neutrophil lymphocyte ratio (NLR) with the area under the ROC curve. Univariate and multivariate analyses using the Cox proportional hazards model were performed to identify the factors closely associated with overall survival of the patients. Kaplan-Meier analysis was used to compare overall survival curves between patients with a high CAR and those with a low CAR. RESULTS: The recommended cutoff value of CAR was 0.132. Kaplan-Meier analysis and log rank test demonstrated a significant difference in the overall survival between patients with a low CAR (<0.132) and those with a high CAR (≥0.132) (2157.0∓395.3 vs 1661.0∓136.4 days, P<0.001). The area under the ROC curve of CAR, NLR and GPS was 0.656, 0.550 and 0.642, respectively, indicating a better prognostic value of CAR. Univariate analyses showed that age, C-reactive protein, albumin, CAR, NLR, GPS, platelet, TMN stage, Dukes stage and chemotherapy regimens were associated with the overall survival of the patients (P<0.05). Multivariate analyses showed that TMN stage [HR=1.689 (95%CI: 1.146-2.488), P=0.008] and Dukes stage [HR=2.447 (95%CI: 1.349-4.441), P=0.003] were associated with the overall survival of the patients. CONCLUSIONS: Similar to the previously reported inflammation-based prognostic systems (GPS and NLR), CAR is useful for predicting the survival of patients with colorectal cancer and can be complementary to the two prognostic scoring systems.
