(192) Line of therapy algorithm for real-world data analysis in multiple myeloma

Background: Line of therapy (LoT) algorithms are used to incorporate treatment information from RWD and are critical in identifying relevant cohorts for analysis, and interpretation of results. This is especially critical in disease areas where the treatment landscape is rapidly changing due to many recent and coming drug approvals.

Objectives: To develop a new LoT algorithm for Multiple Myeloma (MM) and evaluate its performance in treatment patterns and outcome analysis.

Methods: This retrospective cohort study included newly diagnosed MM patients who received first-line of therapy (1L) between January 2011 and May 2023; from the USA nationwide, electronic health record-derived Flatiron Health de-identified database. Flatiron LoT algorithm is based on an oncologist defined, rule-based lines of therapy, which maps medication administrations and medication orders to lines of therapy. In general, a change in a LoT is defined based on the addition of new MM agents and gap periods in which no treatment was received. A new MM LoT algorithm was created and applied using an additional set of rules to the Flatiron LoT. Rules included exclusion of non-related MM or specific monotherapy drugs (eg. steroids), modification of ongoing maintenance therapy regimens, and merging of lines (eg. when treatment duration was short, a drug or combination of drugs was repeated within lines or dropped in the subsequent). Medical review of new rules was done by MM expert hematologist and LoT performance evaluated during patient selection, treatment patterns and outcome analyses.

Results: A total of 15,316 patients receiving at least 1L treatment were selected using the actual Flatiron LoT. After using the new LoT, a decrease on the number of patients was observed during the cohort selection (n=14,870) and across lines: 8547 and 7215 in 2L, 5007 and 3762 in 3L and 2944 and 2017 in 4L, using the actual Flatiron LoT and the new LoT, respectively. The new algorithm led to at least one of the following changes within the lines of therapy of the patient cohort: exclusion of Non-MM drugs (n=4557), and steroids (n=1172) from the line, modification of maintenance therapy (n=2271), and merging of 963 lines (due to short treatment [n=105], a drop of a drug in the current combination treatment in the subsequent line [n=238], or drug or combination repeated in subsequent lines [n=620]). Details on treatment patterns and outcomes using both MM LoT algorithms will be presented.

Conclusions: Adjusting the LoT algorithm can significantly affect both the selection of patient cohorts and their associated treatment and outcomes. Further validation of this new MM LoT algorithm should consider chart reviews and application to other RWD sources, such as insurance claims.