Background:Comorbidities are an important consideration when selecting preparative regimens for allogeneic hematopoietic stem cell transplantation (HSCT). We hypothesized that the impact of individual comorbidities on the risk of non-relapse mortality (NRM) varies between conditioning regimens. Therefore, we retrospectively analyzed the risk associated with pre-transplant comorbidities and NRM across standard preparative HSCT regimens.
Methods: The study included adult patients transplanted for all indications between 2006 and 2015 at a single large academic medical center who were treated with any of the following conditioning regimens: Fludarabine + 2 days Busulfan (Flu/Bu2), Fludarabine + 4 days Busulfan (Flu/Bu4), Fludarabine + Melphalan (Flu/Mel), or Fludarabine + Treosulfan (Flu/Treo). Patient comorbidity status was determined using the guidelines set out in the hematopoietic cell transplantation comorbidity index (HCT-CI, Sorror et. al, 2005) with the following modifications: renal disease was defined as an estimated glomerular filtration rate < 60 mL/min/1.73 m2 instead of a serum creatinine >2 mg/dL; obesity constituted a body-mass index > 30 kg/m2 rather than > 35 kg/m2. Additional comorbidities not included in the HCT-CI, such as hypertension, hyperlipidemia, and HBV or CMV seropositivity, were also assessed.
The hazard associated with individual comorbidities was evaluated in each regimen using cause-specific Cox regressions for 2-year non-relapse mortality (NRM), in a multivariate analysis adjusted for age, disease risk and donor type. Comorbidities were assessed in a given regimen if positive in 10 or more patients treated with that regimen. Finally, the cumulative incidence of NRM was plotted and compared using the Fine and Gray test.
Results: A total of 692 patients was included in the analysis, with a median age of 59 years. Patients underwent conditioning with Flu/Treo (42%), Flu/Bu2 (23%), Flu/Mel (17%), or Flu/Bu4 (18%). Grafts were obtained from matched sibling donors in 45% of cases.
Comorbidities associated with increased risk varied between the four conditioning regimens (Figure 1). While both Flu/Bu4 and Flu/Treo are considered high-intensity regimens, cardiac disease and hypertension were independent risk factors for NRM (Hazard Ratio [HR] 3.9 [1.5-10.3], p = 0.002 and 3.5 [1.2-10.6], p = 0.024, respectively) only in patients conditioned with Flu/Bu4. Cumulative incidence of NRM in patients treated with Flu/Bu4 with and without cardiac disease was 33.3% (95% CI 18.6-59.7), versus 11.4% (6.5-20.0, p = 0.006), respectively. A diagnosis of diabetes also had an increased risk for NRM in Flu/Bu4 (HR 3.3 [1.2-8.7], p = 0.017) but not in Flu/Treo (1.4 [0.7-3.0], p= 0.314), though the confidence interval in the former was rather wide. Additional risk was seen in the Flu/Treo regimen with active infection at time of transplantation (HR 3.6 [1.8-7.2], p < 0.001), and in the Flu/Mel and Flu/Treo regimens with severe pulmonary disease (HR 4.2 [1.7-10.7], p = 0.002). Few patients (9/124 [7%]) with severe pulmonary disease received Flu/Bu4, and therefore the risk for NRM was not calculated in this population.
An HCT-CI score of 3 or greater was associated with increased risk in Flu/Mel (2.4 [1.2-4.5], p=0.010) and Flu/Treo (1.8 [1.1-2.9], p = 0.018) conditioning. However, no clear hazard was observed for the HCT-CI score in the Flu/Bu2 and Flu/Bu4 settings.
Conclusion: It is common practice to group preparative regimens by conditioning intensity, however our analysis demonstrates that individual comorbidities may have effects unique to a particular conditioning regimen. Cardiac disease, for example, imparts additional hazard in the setting of Flu/Bu4 conditioning, whereas it may not increase non-relapse mortality in other reduced-toxicity or reduced-intensity contexts. The understanding of the roles played by individual comorbidities in relationship with different conditioning regimens may be a factor for consideration when selecting a conditioning regimen, allowing for the personalization of therapy.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.