Impact of Health Insurance and Access to Autologous Stem Cell Transplantation (ASCT) on Racial Inequity in Overall Survival (OS) Among Patients with Multiple Myeloma (MM)

Background: Survival for patients with MM has improved in the past two decades, largely attributed to the advent of novel therapies and increased use of autologous stem cell transplantation (ASCT). However, questions remain whether African Americans/Blacks (AAs) have benefitted equally from survival improvement compared to whites. Few studies have examined the simultaneous impact of health system inequities ( i.e. insurance, access to ASCT) on differential outcomes by race. We addressed these knowledge gaps using the real-world data of the nationwide Flatiron Health electronic health record (EHR)-derived de-identified database.

Methods: We included 5722 patients newly diagnosed with MM between January 1, 2011 and December 1, 2021. The study was limited to those with known state of residence, diagnosed in the community oncology setting, had available first-line therapy and insurance data. Longitudinal insurance data included the start and end dates of documented insurance coverage which were used to determine the insurance type at time of initial MM diagnosis. Insurance was grouped as Private, Medicare (± private or non-Medicaid government programs), any Medicaid, or Other (non-Medicaid government program ± unknown payer type). Follow-up was from diagnosis until death or last known activity in the EHR, to December 31, 2022. Patients with del(17p), t(14;16), t(4;14), or amp(1q21) were considered to have high cytogenetic risk (CR); standard risk otherwise. Dates of Medicaid expansion adoption by states were obtained from public data. Missing data on International Staging System (ISS), CR, and insurance type at diagnosis (missingness: 40%, 59%, 19%, respectively) were multiply imputed 10 times. Multivariable step-wise Cox regression was used to examine the effects of race, ASCT, and insurance on OS, treating ASCT as time-varying covariate. For access to ASCT as outcome, we used Fine-Gray subdistribution hazard regression taking death as competing risk. Covariates considered in the multivariable model included: ISS, CR, age at diagnosis, sex, and prior history of cancer.

Results: Median age (range) at diagnosis was 70.5y (24 - 85); 32% were <65y at diagnosis; 54% were male; 24% were AA. Median follow-up was 2.9y (0.01 - 12). Compared to whites, AAs were more likely to be diagnosed at a younger age (median, 67y vs 71y), have less aggressive disease (ISS Stage I: 41% vs 31%) and standard CR (41% vs 36%), and more likely to reside in states without Medicaid expansion at time of MM diagnosis (74% vs 58%). Medicare was more common in whites (65.4% vs 50.5%) while Medicaid was more common in AAs (15.3% vs 4.4%). Impact of Insurance: Adjusted analyses showed a higher overall mortality risk in AAs compared to whites (adjusted Hazard Ratio [aHR]=1.10, p=0.042). Racial difference was abrogated (aHR=1.06, p=0.28) when the model was adjusted for insurance type. Compared to patients with Private insurance, patients with Medicare had lower mortality risk (aHR=0.78, p<0.0001) while the risk was higher (aHR=1.29, p=0.004) in patients with Medicaid or Other insurance (Medicaid/Other ). Impact of ASCT: Adjusted for disease, demographics, and insurance, mortality risk was higher without receipt of ASCT (aHR=1.73, p<0.0001). Of note, while ASCT utilization was lower in AAs compared to whites (aHR=0.70, p<0.0001, adjusted for insurance), this difference did not manifest as racial difference in OS within each insurance type (p>0.5). Impact of Insurance + ASCT: Considering the joint effects of ASCT and insurance type, mortality risk was highest for Medicaid/Other insurance without receipt of ASCT (aHR=3.05, 95% CI: 1.94 - 4.80); reference Private/Medicare with receipt of ASCT ( Figure). With the receipt of ASCT, mortality risks were not different between Private/Medicare and Medicaid/Other insurance (p=0.46). Similar trends were observed for AAs and whites.

Conclusions: Adjusted for relevant risk factors, survival was worse in AAs with MM compared to whites. However, racial differences were abrogated after adjusting for insurance type. Overall, OS was worse in patients with Medicaid compared to those who had private insurance or Medicare. OS was also worse without receipt of ASCT, independent of insurance type. The highest mortality risk was seen in patients with Medicaid without receipt of ASCT. These findings provide real-world evidence of how inequity in health insurance and access to ASCT can impact MM survival.