Introduction: Severe gastrointestinal (GI) toxicity following 5-FU-based treatment is highly prevalent and negatively affects therapy. No effective predictors for severe GI toxicity risk have been identified. Our pilot study (n=34) reported TLR2 and TNF genetic variants and cancer type (colorectal and gastric) predicted severe toxicity risk (receiver operator characteristic (ROC) area under the curve (AUC) 87.3%).
Aims: To validate our pilot study observations in an independent patient cohort.
Methods: One hundred and three participants whom previously received 5-FU-based treatment were recruited. GI toxicity data (toxicity: symptoms Grade ≥ 3 NCI’s CTCAE v4, or requiring treatment cessation or reduction), demographics and treatment parameters were mined from clinical records. Genetic variability in the following genes was determined using a customised Sequenom MassArray: IL1B, IL2, IL6, IL6R, IL10, TNF, TGFB, TLR2, TLR4, MD2, MYD88, CRP, ICE and OPRM1. Multivariate logistic regression produced a prediction model of severe GI toxicity risk, with ROC AUC assessing model performance.
Results: Twenty-two participants experienced severe GI toxicity. Colorectal and gastric cancer and IL1B variants (rs16944 and rs1143634) were significantly associated with severe GI toxicity risk with a ROC AUC of 82% (P=3.4x10-5).
Conclusions: Compared to the pilot study, different genetic variants were identified as predictive of severe GI toxicity risk. This may reflect the shift in cancer type between the pilot and current study that favored breast cancer regimens. Nonetheless, we have identified genetic markers related to inflammatory responses as being predictive. Further patients are being recruited to expand study cohort to 150 to meet a-priori power calculations.