A new method to analyze adverse events longitudinally in oncology clinical trials.

Abstract

3622 Background: There are significant limitations to the current method for displaying treatment-related adverse events (AE) in clinical trials. Tables displaying grade 3 and 4 AEs do not capture toxicity profiles that evolve over time and do not provide clinically relevant information on when a given AE could arise, its probable duration and its severity at a given point in therapy. We developed a novel method of analysis to better capture and represent longitudinal AE data. Methods: Graphical and analytical routines were developed to analyze adverse events longitudinally. Mean symptom and toxicity values are plotted over time on a single axis for direct comparison. Butterfly plots compare adverse events by treatment arms longitudinally. Multiple longitudinal techniques (repeated measures modeling, profile analysis, area under the curve analysis, change from baseline, and time to event) are included. Heat maps, event charts and stream plots are produced for individual patient results over time. Results: The methodology was validated in two completed clinical trials in colorectal cancer treatment (N9741) and cancer control (979254). In N9741, a higher incidence and severity of diarrhea, nausea and vomiting in patients on IROX (irinotecan/oxaliplatin) versus FOLFOX (fluorouracil/oxaliplatin) was most pronounced in the first treatment cycle and declined sharply by cycle 2. An increased incidence of alopecia in this cohort was also most apparent upfront, from the first to the second cycle, as compared to a lower overall incidence of alopecia with a gradual rise from cycles 1 through 5 in patients on FOLFOX. In study 979254, a higher incidence of dry mouth in patients using venlafaxine 150 mg/d as compared with placebo was most prominent only later in therapy, at weeks 4 and 5 (incidence of 22% for placebo and 13% for venlafaxine 150 mg/d at week 1, p=0.25, versus, 2.2% and 49%, respectively, at week 5, p=0.0001). Conclusions: A new method of analyzing AEs in clinical trials that incorporates the dimension of time offers a more complete depiction of chemotherapy toxicity than current methods. Longitudinal analyses of toxicity over time can provide enhanced patient education and guide clinical management on a given chemotherapy strategy.