The Effect of Autologous Stem Cell Transplant (ASCT) on Survival in Californians with Multiple Myeloma (MM) in the Era of Modern Treatment

Background: The use and timing of ASCT on survival after MM in the era of modern therapy remain topics of debate. Using population based data, we evaluated factors associated with the receipt of ASCT and the effect of ASCT on overall survival (OS). 

Methods: Patients diagnosed with MM during 2000 – 2012 were identified in the California Cancer Registry (CCR) (n=12,714). CCR data were linked to the California Patient Discharge Database (PDD). Logistic regression estimated the odds ratio (OR) of having an early (< 1 year from diagnosis) or late (> 1 year) ASCT (vs. no ASCT).  OS was calculated using the Kaplan-Meier (KM) method. To determine the effect of ASCT on OS from diagnosis date, Cox regression models estimated adjusted hazard ratios (aHR) of death treating ASCT as a time dependent covariate. OS time was compared after matching ASCT to no ASCT patients on age, sex, race/ethnicity, neighborhood socioeconomic status (SES), comorbidity at diagnosis, year of diagnosis, and accounting for time to transplant.

Results:  The majority of MM patients were male (54%) and of non-Hispanic white (58%) race/ethnicity; 19% Hispanic, 12% African American, and 9% Asian. Median age at diagnosis was 67 (range 18 – 104). African Americans and Hispanics were younger than non-Hispanic whites (median age 64 and 65 vs 69). Comorbidity data from the PDD was available in 59% of the patients in the 2 years prior to MM diagnosis: 7.5% had 0, 21% had 1-2, and 31% had ≥3 comorbidities.

A total of 2136 (17%) patients underwent ASCT: 1347 < 1 year from and 789 ≥1 year after diagnosis. Time to ASCT did not change over time: among patients diagnosed 2000 – 2003 median time to transplant was 9.2 mo, 10 mo among those diagnosed 2004 – 2007 and 9.7 in those diagnosed 2008 – 2012. Patients who underwent ASCT were younger than those who did not (median age 56 vs 70 respectively). African Americans were less likely to undergo early ASCT (OR 0.7, P<0.001), but not late ASCT (OR 0.8, P=0.07). Patients with ≥3 comorbidities (vs. 0) at diagnosis were less likely to have ASCT (OR 0.42 P<0.001 and OR 0.28 P<0.001 for early and late, respectively), while patients with 1-2 comorbidities were less likely to have late ASCT (OR 0.59 P<0.001). The lowest 2 quintiles of SES was associated with less use of early ASCT (OR 0.62 p<0.001 and 0.65 p<0.001 respectively), but not late ASCT (OR 0.89 p=0.4 and 0.96 p=0.7 respectively). The likelihood of receiving ASCT increased over time: compared to 2000-2003, the ORs for patients diagnosed in 2004 – 2007 were 1.36 for early (P<0.001) and 1.64 (P<0.001) for late ASCT and were 2.64 (P<0.001) for early and 1.80 (P<0.001) for late for those diagnosed in 2008-2012.

The median follow-up was 32 months. Median OS from diagnosis for the entire cohort, unadjusted for age, comorbidities, and SES was 37 months. Adjusting for sex, race/ethnicity, age at diagnosis, SES, comorbidities, insurance status and year of diagnosis, OS improved over time: compared to patients diagnosed in 2008 – 2012, aHR of death of those diagnosed 2000-2003 was 1.58 (P<0.001), and 1.35 (P<0.001) for those diagnosed 2004-2007. ASCT at any point was associated with a 23% reduction in the risk of death from all causes (aHR 0.77 P<0.001). Patients who received early ASCT had a 27% reduction (aHR 0.73 P<0.001), while those receiving late ASCT had an 11% decrease (aHR 0.89 P<0.001) in risk of all cause death. In the matched analysis, the median OS from date of transplant, or matched date in the no ASCT cohort, were: no ASCT = 49 mo, early ASCT = 83 mo, and late ASCT 65 mo (P<0.001 Figure 1). The effect of aSCT on OS differed by date of diagnosis (P for interaction <0.001). Improvements in OS due to ASCT were more pronounced in later time periods: aHR for early and late ASCT in 2000-2003 were 0.9 (P = 0.12) and 0.98 (P 0.86) compared with those in 2004-2007 (0.63 P<0.001 and 0.85 P = 0.06) and in 2008-2012 (0.55 P<0.001 and 0.74 P=0.08).

Conclusions: ASCT was utilized in 17% of Californians with MM during 2000-2012, and its use increased over time. The use of ASCT, whether within a year of diagnosis or later in the disease course, is associated with improved OS and this effect may be more pronounced in the era of novel agents. Despite the inherent limitations of analyses of administrative databases, the large number of patients and established robust nature of CCR and PDD data makes accurate depiction of results in the community probable. These data support the continued role of ASCT in the management of patients with MM.