An Accelerated CD8+, but Not CD4+, T-Cell Reconstitution Associates with a More Favorable Outcome Following HLA-Haploidentical HSCT: Results from a Retrospective Study of the Cell Therapy and Immunobiology Working Party of the EBMT

Introduction and Aim: HLA-haploidentical hematopoietic stem cell transplantation (haplo-HSCT) is increasingly offered to patients with high-risk acute myeloid (AML) or lymphoid leukemia (ALL). Unfortunately, graft manipulation employed to overcome the HLA barrier significantly delays immune reconstitution, posing the patients at risk of infections. Accordingly, non-relapse mortality after haplo-HSCT clearly extends beyond day 100 post-transplant. Over the years, different approaches have been investigated to speed-up immune reconstitution. In the absence of validated immune biomarkers, it is however difficult to evaluate the clinical impact of accelerated immune reconstitution. The aim of this EBMT retrospective study is to explore immune-cell counts early after haplo-HSCT as predictive of its overall outcome.

Methods and Patients: Among AML and ALL patients in the EBMT database who underwent haplo-HSCT in the period 2001-2012, criteria for study entry were survival beyond day 100 and availability of differential immune-cell counts (CD3+, CD4+, CD8+ T cells, CD19+ B cells, CD16+/CD56+ NK cells) within this period. Accordingly, statistical analysis was landmarked at day 100. Of 259 patients meeting these criteria (age 2-70, median 33), 67 (26%) were children. The underlying disease was AML in 162 cases (63%), while ALL in the remaining (including 5 cases of bi-phenotypic leukemia). Fifty-two percent of patients were transplanted in CR1. The stem-cell source was G-CSF mobilized peripheral blood in all but one patient (>99%) and 171 received TBI (66%). The graft was manipulated in 199 patients (78%), including CD34-selection (50%), ex vivo T-cell depletion (15%) or both (13%). Female-to-male transplants were 68 (26%), while 204 (79%) recipients were CMV seropositive. Sustained hematopoietic engraftment was reached in 246 patients (95%)

Results: The estimated overall survival at 2yrs was 43%. The estimated cumulative incidence of death due to relapse was 33%, while that of death due to other causes was 35% (51% of those were infections) The occurrence of grade III-IV GVHD and of chronic GVHD was 9% and 18% (7% extensive), respectively. As expected, overall survival was better in children (62% vs 36%, P=0.002 by Log-rank), who clearly had a lower incidence of death due to causes other than relapse compared with adults (10% vs 37%, P=0.0001). Negative prognostic factors for overall survival were any disease state other than CR1 at time of transplantation (P=0.002) and CMV seropositivity (P=0.009). Type of leukemia, TBI or graft manipulation had no effect on the outcome. By day 100 post-transplant, patients reached the following median immune-cell counts: 100 CD3+ T cells (range 0-2576), 30 CD4+ T cells (0-1714), 48 CD8+ T cells (0-1880), 276 CD16+/CD56+ NK cells (18-3581), 21 CD19+ B cells (0-790). Importantly, CD3+ counts above the first quartile (1Q) of the entire data set (29 cells per microL) were significantly associated with a better overall survival (P=0.0005 by Log rank) and a lower incidence of death due to causes other than relapse (P=0.002 by Gray test). The same held true for CD8+ counts (1Q: 15 cells per microL; P=0.003 on overall survival; P=0.0004 on death due to other causes). CD4+ counts also showed similar correlations, but at higher values (above the median). None of the other immune-cell counts analyzed correlated with clinical outcome. Strikingly, when challenged in multivariate analysis taking into account age category, CMV seropositivity, graft manipulation and CR1 status at transplant, CD3+ and CD8+ counts above the 1Q adjusted to fit optimal cut-off points were still significantly associated with a better overall survival (P=0.006 and P=0.015, respectively), but only CD8+ values associated with a lesser risk of death due to causes other than relapse (P=0.026). Conversely, similarly adjusted median CD4+ counts failed to show any association.

Conclusions: Contrary to what is generally accepted, these results indicate that an accelerated CD8+, but not CD4+, T cell reconstitution associates with a more favorable clinical outcome after haplo-HSCT, likely due to its protective role against opportunistic viral infections. Moreover, they suggest that yet to be validated CD8+ cut-off points, rather than the commonly used arbitrary value of 200 CD4+ T cells per microL, should be considered as surrogate biomarkers in clinical trials.