Understanding the comorbidome and its effect on survival in colorectal cancer.

210 Background: Effectively managing comorbidities is an essential component of high-quality cancer care. Evidence suggests colorectal cancer (CRC) patients with multiple comorbid conditions are less likely to complete standard treatments and can have lower rates of survival. In order to provide personalized care, it is critical to understand how comorbid conditions cluster within CRC patients.

METHODS: We identified Kaiser Permanente Southern California CRC patients diagnosed with first malignancy between 01/01/2008 - 12/31/2013. We used latent class analysis to identify clinically useful phenotypes defined by combinations of comorbidities at diagnosis, and compared survival using the Kaplan-Meier method.

RESULTS: The cohort included 7803 patients: 52% male; average age at diagnosis 66 years (SD: 13); 22% Hispanic, 15% Black, 9% Asian, 52% White; 42% Stage I, 22% Stage II, 22% Stage III, and 14% Stage IV. One-fifth of patients had a Charlson comorbidity index score of ≥ 4. We found 4 distinct classes (Lo-Mendell-Rubin p<0.001). Class 1 was relatively healthy with few comorbidities (Table). Class 2 included individuals with cardiovascular diseases; those in Class 3 had complicated diabetes. Class 4 members had multiple chronic conditions, including diabetes with micro- and macrovascular complications. Kaplan-Meier estimates revealed a statistically significant difference in overall survival by class (log rank p<0.001).

CONCLUSIONS: We identified 4 clinically distinct classes of comorbid conditions in CRC patients. These data can be used to inform personalized care for CRC patients throughout the cancer care continuum. [Table: see text].