Analysis of an Online Decision Support Tool for Chronic Lymphocytic Leukemia: Disparities in Treatment Selection Between Experts and Community Practitioners

Background. Rapid advances in clinical discovery and availability of new treatment options have increased the complexity of treatment decisions for patients with CLL. Guidelines list multiple agents and combinations as recommended therapeutic options for CLL but often do not provide specific treatment recommendations for individual patients. We developed an online treatment decision tool that provides treatment recommendations from CLL experts for specific patient cases. We hypothesized that these individualized recommendations from recognized experts would affect treatment plans. Here we report on an analysis of data entered into this CLL decision support tool, including variance between intended treatment of tool users and the recommendations made by the experts and the impact of the tool on subsequent therapy decisions.

Methods. In December 2015, 5 experts provided treatment recommendations for 1380 case variations based on key factors that guide treatment choice. Expert-selected factors for newly diagnosed CLL included age, fitness (based on ECOG PS, CIRS, and renal function), and cytogenetic abnormalities (del[17p], del[11q], or other). Additional variables for patients with relapsed/refractory (R/R) disease after first-line treatment included previous treatment, response duration, and burden of comorbidities. To use the tool, drop-down menus allowed users to select from choices for each variable and their intended treatment for that patient. The corresponding treatment selection from 5 experts was then displayed and users were asked about the tool's impact on their planned treatment.

Results. An analysis of 883 patient scenarios (67% treatment naive and 33% with R/R CLL) entered into the tool from February 2016 through July 2016 found substantial variation between the intended therapy choice among tool users and the recommendations from the experts.For example, in every patient case with del(17p), all 5 of the experts recommended ibrutinib as first-line therapy whereas only 49% of tool users planned to use ibrutinib for these patients. Of those users whose intended first-line therapy for del(17p) CLL did not match the experts' recommendation, 54% indicated that this tool would change their original treatment plan and 17% indicated a barrier to implementing this treatment. For either elderly or unfit patients without del(17p), 4 of 5 experts recommended obinutuzumab plus chlorambucil, but only 41% of tool users planned to use this regimen with 50% citing barriers to this treatment approach.

For patients with del(17p) CLL and disease relapse or recurrence after chemoimmunotherapy, all 5 experts recommended ibrutinib for these cases with the exception of patients with a history of atrial fibrillation, anticoagulation, or difficult-to-control hypertension where 4 of 5 experts recommended idelalisib/rituximab. Again, the intended treatment plan of approximately 50% of tool users failed to match the experts' recommendation for these cases, and half of these users indicated that this tool would change their original treatment plan. At the time of tool development, all experts recommended either idelalisib/rituximab or clinical trial for patients with R/R CLL and del(17p) who previously received ibrutinib, but 61% of users indicated that they were unsure of the next appropriate treatment. All users who answered the impact question indicated that they now intended to use the expert-recommended treatment for these patients.

For patients without del(17p) cytogenetics, treatment selection was more variable among experts and users and changed based on age, fitness, and previous therapy. For patients with del(11q) or other cytogenetics, approximately 20% of tool users were unsure of the appropriate treatment after progression on first-line therapy but 71% of those who answered the impact questions indicated that they remained unsure of their treatment approach despite viewing expert recommendations.

Conclusions.Our analysis demonstrates that this interactive online therapy decision tool providing expert recommendations for specific case scenarios in CLL can support optimal decision making and change intended treatment for a majority of cases in which the planned therapy differed from the experts. Detailed comparisons of expert and user responses from the online tool will be presented.