Effective use of electronic medical record (EMR) data in analysis of association between KRAS mutations and depression in colorectal cancer patients.

Abstract

861 Background: Previous studies have demonstrated that KRAS mutations were associated with higher rates of depression in patients with metastatic colorectal cancer (CRC). The objective of this study was to evaluate the feasibility of extracting EMR data to examine the association between KRAS mutations and positive screening test for depression in CRC patients. Methods: Retrospective review of stage I to IV CRC patients seen between 2011 and 2015 at an academic, NCI-Designated Comprehensive Cancer Center was performed. At each clinic visit, depression was assessed using the Patient Health Questionnaire-2 (PHQ-2), which is part of the institution’s universal Distress Screening tool. PHQ-2 score of 2 and above was considered positive screening test for depression. PHQ-2 and KRAS mutation data were extracted from the EMR via the Clinical Data Exchange Network bioinformatics tool and confirmed by retrospective chart review. Chi-square test was used to assess the association between KRAS mutation and depression. Multiple imputation was used to impute the missing values. Results: Of the 484 CRC patients, KRAS status was known in 45 cases: 22 (49%) were KRAS mutated and 23 (51%) were KRAS wild type. PHQ-2 score was 0 in 42 cases (93.3%), 1 in 2 cases (4.4%), and ≥2 in 1 case (2.3%). The rate of positive PHQ-2 for KRAS mutated vs wild type was 4% vs 0% (p = 0.36). The result based on 50 imputed datasets suggests a trend towards an association between KRAS mutation and depression (p = 0.09). Conclusions: This study did not demonstrate an association between KRAS mutation and depression in patients with colorectal cancer, probably due to a high proportion of missing data. Bioinformatics studies that extract and analyze EMR data are a feasible and effective platform to assess the association of genomic data with clinical outcomes. Additional validated algorithms and data are needed to further optimize such studies.