Abstract
Infection-related mortality (IRM) still represents a major determinant of non-relapse mortality (NRM) after allogeneic hematopoietic stem cell transplantation (allo-HSCT). Despite the curative potential of allo-HSCT, the profound and prolonged status of immune incompetence following transplantation poses patients at risk for lethal opportunistic infections. This is particularly important in transplants from HLA-mismatched donors. Although numerous studies have investigated the role of pre-transplant clinical variables for the prediction of IRM, very few have focused on biological culprits.
This study was aimed at the development of a composite clinico-biological prognostic scoring system for the prediction of early and late IRM after allo-HSCT.
A total of 492 consecutive adult patients receiving allo-HSCT for hematological disorders were studied from January 2009 to May 2015. Second transplants were excluded, as well as patients for which pre-transplant biological data were missing. The Receiver Operating Characteristics (ROC) curve analysis defined the optimal cut-offs of pre-transplant biological variables predicting 100-days IRM. All variables were then challenged in multivariate analysis. Based on the value of the coefficients selected in the final multivariate model, a weighted score predicting IRM was elaborated in the training cohort of patients (n=273, from January 2012 to May 2015), and then tested in an internal validation cohort of patients (n=219, from January 2009 to December 2011).
The median follow-up was of 31 months (range, 1-85). Acute leukemia was the main indication to transplant, accounting for the 61% (n=309) of patients. Fifty-one percent of patients received an HLA-haploidentical donor graft, 27% a fully HLA-matched (10/10) or a single mismatch (9/10) unrelated donor (MUD) graft, 19% a HLA-identical sibling donor and 3% a cord blood unit. Noticeably, 49% (n=248) patients underwent transplant for advanced diseases. To assess uniformity between the training and validation cohorts, we calculated the incidence of NRM as well as overall survival (OS) and progression-free survival (PFS) in each group. OS at 2 yrs was 53% (95% CI: 47% to 60%) in the training cohort and 46% (95% CI: 40% to 53%) in the validation cohort (P=0.050). PFS at 2 yrs was 23% (95% CI: 16% to 23%) in the training cohort and 17% (95% CI: 12% to 27%) in the validation cohort (P=0.440). The 2-yr incidences of NRM and IRM were 28% (95% CI, 23% to 24%) and 22% (95% CI, 18% to 28%), respectively, in the training cohort and 34% (95% CI: 28% to 40%) and 23% (95% CI: 18% to 29%), respectively, in the validation cohort (P=0.371 and P=0.702).
Only four variables were found independent predictors of IRM: age >60 yrs (P=0.003), CMV host/donor serostatus different from negative/negative (P<0.001), pre-transplant levels of IgA <1.11 gr/L (P=0.004) and pre-transplant levels of IgM <0.305 gr/L (P=0.028). Noticeably, these associations were independent from disease type or disease status at transplant, from the type of donor and intensity of conditioning, from the use of in vivo T or B-cell depletion or from previous colonization by multi-drug resistant Gram-negative pathogens. A weighted score using these four factors subsequently devised a 3-tiered prognostic model (low-risk: £ 10.17 points; intermediate-risk: 10.17- 11.11; and high-risk: > 11.11 points). Considering the entire cohort, the 2-yr IRM was 12% (95% CI: 8% to 17%) for low-risk patients (n=196), 23% (95% CI: 17% to 30%) for intermediate-risk patients (n=175) and 37% (95% CI: 28% to 46%) for high-risk patients (n=108) (P<0.0001).
Figure 1 shows IRM according to the 3-tiered prognostic model in the training group (A) and in the validation group (B).
Additionally, the 2-yrs OS was significantly different among the 3 groups, being 59% (95% CI: 52% to 67%), 50% (95% CI: 43% to 59%) and 37% (95% CI: 29% to 48%), for low, intermediate and high-risk patients, respectively (P=0.0001).
The proposed clinico-biological scoring system is able to predict IRM after allo-HSCT and may foster active surveillance and early pre-emptive antimicrobial strategies in patients at higher risk for infectious complications. Moreover, this score pave the way for the prospective investigation of prophylactic infusions of IgM/IgA-enriched immunoglobulins to reduce IRM after allo-HSCT and to improve its overall outcome.
Ciceri:MolMed SpA: Consultancy.
