(179) Validation of claims algorithm to identify HR+/HER2- metastatic breast cancer patients

Background: Hormone-receptor-positive/human-epidermal-growth-factor-receptor-2-negative (HR+/HER2-) is the most common breast cancer (BC) subtype and accounts for 75% of BC cases in the US population. However, the lack of a subtype-specific diagnosis code in the health care delivery system prevents readily available cohort identification for use in real-world evidence generation.

Objectives: This study aimed to validate a claims-based algorithm used in literature to identify patients with HR+/HER2- metastatic breast cancer (mBC).

Methods: This retrospective cohort study used administrative data and claims from the Komodo Research Dataset linked to genomics data from NeoGenomics to validate HR+/HER2-mBC patients identified via a code-based algorithm based on prior studies. Eligible patients were required to have evidence in claims of: a) 2+ diagnoses for BC between January 2020 and August 2023 (≥30 days apart), b) 2+ diagnoses for secondary neoplasms 30 days before or any time after the first BC claim (mBC diagnosis), and c) 1+ HR+ treatment and no HER2+ treatment received during the study period. Patients with mBC diagnosis dates before January 2020 were excluded to reflect genomics data availability. Demographics were evaluated on the first indicated HR+/HER2- mBC therapy start date after mBC diagnosis. Cases that met the algorithm were then validated against genomics data. Specifically, immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) test results prior to mBC diagnosis were used to determine HER2 status; estrogen receptor (ER)/progesterone receptor (PR) test data were used to determine HR status (positive, negative, or unclear). Patients with unclear HR/HER2 status were excluded. Overall and subgroup positive predictive values (PPV) and 95% confidence intervals (CI) were calculated.

Results: Genomics data was obtained and reviewed for 2,215 HR+/HER2- mBC patients identified in claims. Of these patients, 30% were over 64 (mean 57) years of age and 41% were white. Majority of patients were from the South (62%), followed by the West (22%), Midwest (12%), and Northeast (4%). Overall PPV of the HR+/HER2- mBC identification algorithm was 90% (CI 89–92%). PPV was also high (>83%) across demographic subgroups. The algorithm resulted in higher PPV for patients who were white (92% [CI 90–94%]), ≥65 years of age (92% [90–94%]), and ≥75 years of age (94% [90–97%]); and lower PPV among patients who were black or African American (84% [79–88%]).

Conclusions: The results demonstrate that the claims-based identification algorithm was highly accurate for identifying HR+/HER2 - mBC patients.