Effect of HLA mismatch on post-transplant infections in allogeneic hematopoietic stem cell transplantation with PTCy-based GvHD prophylaxis

TO THE EDITOR: Post-transplant cyclophosphamide (PTCy) has increased the use of haploidentical (HAPLO) donors [1] Early data associated HAPLO transplants with increased infections, though only HAPLO patients received PTCy [2]. In a registry-based study comparing HAPLO vs. MATCHED grafts with PTCy [3], there were increased bacterial infections and fungal infections in HAPLO transplants with myeloablative and reduced-intensity conditioning, respectively [3]. In the present study, we compared post-transplant infections in a cohort of patients with HAPLO or MATCHED donor and uniform PTCy-based immunosuppression.


Infections
Follow-up for infections set at 100 days post-transplant.
In scoring BSI, common contaminants were excluded unless isolated from consecutive cultures. Positive cultures from central line paired with negative cultures from peripheral vein were considered CVC colonization and excluded.
IFI was classified as proven, probable, or possible based on international guidelines [5]. CMV or EBV infection/reactivation was defined as DNA copies >1000 ml -1 or >5000 ml -1 , respectively.

Statistical analysis
We performed competing risk analysis to compare risk of infections in MATCHED vs. HAPLO grafts [6] with non-infection mortality as the competing risk.
A confirmatory analysis with propensity matching was performed, limited to 68 MATCHED and 68 HAPLO transplants. Criteria for matching included donor and recipient age, disease status, and conditioning.
Categorical data were compared using chi-square test.
During the first 20 days, there were 75 BSI events, of which 8/75 were CVC-related. The most frequent pathogens were gramnegative bacteria. During days +21 to +100, BSI were less frequent, and the most frequent were gram-positives.
To account for GvHD and steroid therapy on BSI, we characterized patients according to GvHD. Of 56 HAPLO patients with BSI, 2/56 developed GvHD prior to BSI, compared with 0/14 MATCHED patients. To test for an effect of conditioning, we stratified patients according to days of busulfan: BSI occurred in 45%, 63%, and 41% of HAPLO grafts with TBF1, TBF2, and TBF3, respectively, compared with 20%, 19%, and 12% of MATCHED grafts.
Since prophylactic antifungal in the post-transplant period was limited to patients treated with steroid for GvHD, we stratified patients with IFI according to whether they were receiving steroid and antifungal at the time of IFI. Of 34 HAPLO patients with IFI, 7/34 (20%) were receiving steroid and posaconazole for acute GvHD (aGvHD) (1 grade 1 and 6 grade 2-4), compared with 2/11 (18%) of MATCHED patients with IFI who were receiving for steroid and posaconazole for aGvHD (1 grade 1 and 1 grade 2).
Our data suggest more marked impairment in immune function conferred by HLA mismatch given a uniform PTCy-based immunosuppression. This model would be consistent with data on immune reconstitution showing impaired reconstitution of NK cells and T cells in HAPLO transplants [7]. While our data indicate similar time to leukocyte engraftment, others have shown differences, including CD4 + T cells in HAPLO patients skewed in favor of Tregs and NK cells represented by immature subsets (CD16 -, CD56 bright ) [7].
Our group previously found an association between HAPLO grafts and increased grade 2-4 acute GvHD and moderate-severe chronic GvHD [4]. While steroids for GvHD contributes to immunosuppression in HAPLO grafts, it does not account for increased infections in the present study. Surprisingly, majority of IFI occurred in the absence of GvHD, with increased GvHD and a trend for increased IFI in HAPLO grafts.
Compared with other reports, we observed a relatively high frequency of early BSI in HAPLO grafts (43%), though early BSI in MATCHED grafts were relatively infrequent (21%) [8,9]. Others have reported similar rates of pre-engraftment BSI between HAPLO and MATCHED grafts [8,9]. A relatively high rate of preengraftment BSI in non-HAPLO grafts seen in one study may reflect inclusion of single-antigen-mismatched grafts in the non-HAPLO group [9]. It was also hypothesized that threosulfan-based conditioning lessened the mucosal injury predisposing to BSI [9], of which HAPLO patients are more susceptible [2]; that said, we did not observe an effect of conditioning. Additionally, there are differences in infection protocols, with patients at other centers routinely receiving prophylactic fluoroquinolone and azole [9,10].
Limitations of our study are the retrospective nature and that MATCHED grafts were performed with PB, vs. BM for HAPLO transplants. PB transplants are associated with earlier engraftment. A study on HAPLO BM vs. HAPLO PB grafts, however, showed no difference in neutrophil engraftment or graft failure [11].
In conclusion, patients with HAPLO donors have increased risk of post-transplant infections and IRM, compared with patients with MATCHED donors. Our findings call for diligent monitoring in patients undergoing a HAPLO transplant. Gemelli IRCCS, Roma, Italy. ✉ email: johndmarra@gmail.com

DATA AVAILABILITY
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.