Abstract
BACKGROUND: Demographic, health history, and lifestyle factors have been associated with prognosis of colorectal cancer (CRC), but mechanisms underlying these associations remain poorly understood. A compelling mechanism involves changes in expression of tumor markers that influence treatment outcomes, such as secreted protein acidic and rich in cysteine (SPARC), lower levels of which have previously been associated with poorer CRC prognosis. OBJECTIVE: We explored the association of factors that have been previously associated with CRC prognosis with expression of SPARC in tumor tissues. DESIGN: We conducted a prospective evaluation of 50 participants of a longitudinal cohort study that went on to develop CRC. METHODS: Tumor and normal tissue cores were taken from formalin-fixed paraffin-embedded (FFPE) blocks of incident CRC cases and were used to create tissue microarrays (TMAs). Slides created from the TMAs were stained with SPARC antibodies and analyzed to calculate H-scores for both epithelial and non-epithelial components of tumor and normal tissues. H-scores were ln-transformed and analyzed in association with demographic, lifestyle, and health history factors assessed before cancer diagnosis using linear regression models. RESULTS: In CRC tumor epithelium, smoking was associated with a 0.53-fold lower level of SPARC expression (P = .054). Higher income was associated with a 1.33-fold greater level of SPARC expression in tumor non-epithelial tissue (P = .041). Higher cancer stage was associated with a 0.74-fold lower level of non-epithelial tumor SPARC expression (P = .040). In the epithelial component of normal colorectal tissues, higher fruit consumption was associated with a 2.74-fold greater SPARC H-score (P = .002). CONCLUSIONS: The associations we observed for smoking, income, and cancer stage with SPARC in tumor tissue are consistent with previously established associations of these factors with CRC prognosis. Larger studies with prognostic data are needed, but our results suggest that differences in SPARC expression may contribute to previously observed impacts of various factors on CRC prognosis.