Impact Of Response Shift On Time To Quality-Of-Life Scores Deterioration In Patients With Breast Cancer

1584 Background: Time to quality of life (QoL) score deterioration (TD) is a method of longitudinal QoL data analysis that has been proposed for breast cancer (BC) patients (Hamidou et al Oncologist 2011). As for RECIST criteria, the optimal definitions dealing with reference should be explored. This study aims to study the impact of changes in internal standards (CIS) of response-shift (RS) and the influence of baseline QoL expectancies on TD. Methods: A prospective multicenter study including all women hospitalized for a primary BC was conducted. The EORTC-QLQ-C30 and BR-23 questionnaires were used to assess the QoL at baseline, at the end of 1st hospitalization, and 3 and 6 months after. CIS was investigated by the then-test method. QoL expectancy was assessed at baseline using Likert scale. Deterioration was defined as a decrease in QoL scores reaching at least the mean difference identified as minimal clinically important difference (MCID) using Jaeschke’s transition question. Sensitivity analyses were done using the then-test score as reference score, and considering 5 and 10 points as MCID. TD was estimated using Kaplan-Meier method. Cox regression analyses were used to identify factors influencing TD. Results: From February 2006 to February 2008, 381 women were included. For role functioning dimension, the median TD increased from 3.2 months [95% CI: 3.1-3.36] to 4.76 months [3.3-6.2] when adjusting on CIS. For body image when adjusting on CIS, sentinel lymph node biopsy became significantly associated with longer TD (HR: 0.64[0.43-0.94]) as compared to axillary lymph node dissection, radiotherapy to a shorter TD (HR: 0.63[0.42-0.95] and the type of surgery had no effect on TD. For global health, cognitive and social functioning dimensions, patients expecting deterioration in their QoL had a significantly shorter TD. For fatigue and breast symptom scales, patients expecting no change had a significantly shorter TD, as compared to patients expecting an improvement. Sensitivity analyses using a MDCS of 5 or 10 confirmed these results. Conclusions: Our results suggest that it would be more accurate to take into account CIS component of RS as well as QoL expectancies to estimate TD of QoL scores in patient with BC.