CD3/8 T-Cell Responses to CMV Reactivation and Association with Overall Survival in T-Cell Replete Haploidentical Transplants: A Retrospective Analysis

CMV reactivation has been associated with increased non-relapse mortality (NRM) and improved early relapse incidence (RI) post HLA matched allogeneic hematopoietic stem cell transplant (HSCT), however, it's effect has been less extensively studied in T cell replete haploidentical (HI) HSCT. In the 2 step approach to HI HSCT developed at our institution, a large fixed dose of T cells (2 x 10⁸) is administered after conditioning ( Step 1), followed 2 days later by cyclophosphamide (CY)  for T cell tolerization.  In Step 2, a CD34-selected stem cell product is infused 1 day after completing CY. A significant increase in T cell numbers, especially in  CD3/8 counts, was  associated with CMV reactivation in many patients treated with this approach.  We hypothesized that a CMV-associated increase in CD3/8 counts would impact HSCT outcomes.

A retrospective outcomes analysis (OS, NRM and RI) using multivariable proportional hazards regression was performed  on all patients enrolled on a 2 step clinical trial since 2006, who were alive and disease free at D90 (n=106).  High v low CD3/8 count at D90, a history of GVHD treated with steroids and CMV reactivation (defined by > 100 copies/ml by PCR) both by D90, were the factors  of interest. Known predictors of outcomes including  disease at HSCT  and hematopoietic comorbidity index (HCT CI) were included in the analysis. The median CD3/8 count for the group, 125 cells/ul, was  used to  differentiate  CD8H v CD8L levels.

43% patients  reactivated CMV prior to D90. The median CD3/8 count for CMV-R (reactivators) v CMV-NR (nonreactivators) was 308.4 v 53.7 cells/ul (p<0.0001).  For the whole group, CD8H had a significant protective effect for OS and NRM. CMV reactivation, disease at HSCT and higher HCT CI score had a  significant negative impact. Table 1.  No variables were significantly associated with  RI. In a subset analysis, patients with acute GVHD/CD8H had superior OS in both CMV-R and CMV-NR groups. CMV-NR patients with CD8L /no acute GVHD had the poorest OS. Fig 1. CMV-R patients with CD8L had equally poor OS with or without acute GVHD. Fig 2.

Higher CD3/8 counts were significantly associated with improved OS and lower NRM in all patients irrespective  of CMV reactivation. Higher CD3/8 counts in CMV-R patients may mitigate the effects of CMV reactivation while preserving the beneficial effects of GVHD on OS, a finding that requires further investigation. Prospective analyses of CD3/8 and CD3/4 numbers and strategies to increase them such as early  withdrawal of immunosuppression are warranted.

Table 1: Multivariable model for OS

Variable	Comparison	Hazard Ratio (95% CI)	p-value
CMV  D90	R v NR	3.28 (1.19,9.05)	0.022
CD8 D90	CD8L v CD8H	3.87 (1.39,10.8)	0.0096
GVHD/Steroid D90	Y v N	0.71 (0.32,1.57)	0.40
Dz at HSCT	N v Y	0.27 (0.11,0.65)	0.0036
HCTCI	1 unit increase	1.43 (1.08,1.90)	0.013
Age	1 year increase	0.99 (0.97,1.02)	0.60
Conditioning	Myelo v RIC	1.36 (0.63,2.94)	0.44