A web-based tool to predict chemotherapy-associated cognitive impairment during survivorship.

Abstract

e21609 Background: Chemotherapy-associated cognitive impairment (CACI) experienced during survivorship can dramatically impair quality of life, particularly among patients receiving treatment with curative intent. We construct a web-based tool to predict the risk of self-perceived CACI at 15 months post-chemotherapy initiation among early stage breast cancer patients, to facilitate screening and early intervention. Methods: We recruited patients receiving chemotherapy for early-stage breast cancer between 2011 and 2014. Subjective cognitive function was assessed using FACT-Cog before, during and at 15 months after initiating chemotherapy. Based on a previously established minimal clinically important difference, an 11-point reduction in FACT-Cog score relative to that at baseline was considered as clinically significant CACI. Factors evaluated as potential predictors of CACI included clinical, laboratory and demographic factors, behavioural symptom and quality of life assessments (EORTC-QLQ-C30). Logistic regression was used to construct a model predicting CACI. Validation was performed on an external cohort recruited between 2015 and 2016. Model discrimination and calibration were evaluated using the receiver operating curve (ROC) and Hosmer-Lemeshow goodness-of-fit, respectively. The model was implemented as a web-based tool. Results: A total of 131 patients were recruited for model development. The model consisted of baseline EORTC-QLQ-C30 fatigue subscale percentage [effect size (95% confidence interval), 5.0% (2.0% to 7.0%), p < 0.001] and the change in FACT-Cog score after 2 cycles of chemotherapy, relative to baseline [-4.0% (-7.0% to -1.0%), p = 0.007]. The model was validated on an external cohort of 33 patients, with an ROC of 0.60, Hosmer-Lemeshow goodness-of-fit statistic of 5.86 (p = 0.44), sensitivity of 57.1% and specificity of 76.9%. The web application is accessible at http://bit.ly/CognitiveRisk Conclusions: A web application to predict CACI risk has been developed and externally validated. The tool facilitates early screening for patients at high risk for CACI and allows timely interventions to prevent or minimize the impact of CACI during survivorship.