Haploidentical donor transplant is associated with secondary poor graft function after allogeneic stem cell transplantation: A single‐center retrospective study

Abstract Background Secondary poor graft function (sPGF) is a serious complication after allogeneic hematopoietic stem cell transplantation (allo‐HSCT) related to poor outcome. We aimed to retrospectively evaluate the morbidity and hazard elements of sPGF after allo‐HSCT. Methods Eight hundred and sixty‐three patients who achieved initial engraftment of both neutrophils and platelets were retrospectively reviewed in this study. Results Fifty‐two patients developed sPGF within 180 days post‐transplants, with the median onset time was 62 days (range, 34–121 days) post‐transplants. The overall cumulative incidence of sPGF within 180 days post‐transplantation was 6.0%, with 3.4%, 3.4%, and 10.1%, respectively, in matched sibling donor (MSD), matched unrelated donor (MUD), and haploidentical donor (HID) transplant (p < 0.0001). Multivariable analysis showed that HID (HID vs. MSD: hazard ratio [HR] 2.525, p = 0.004; HID vs. MUD: [HR] 3.531, p = 0.017), acute graft versus host disease (aGVHD) within +30 days ([HR] 2.323, p = 0.003), and cytomegalovirus (CMV) reactivation ([HR] 8.915, p < 0.0001) within +30 days post‐transplants were hazard elements of sPGF. The patients with sPGF had poorer survival than good graft function (51.7±8.1% vs. 62.9±1.9%, p < 0.0001). Our results also showed that only CMV reactivation was the hazard element for the development of PGF in HID transplant ([HR] 12.521 p < 0.0001). Conclusion HID transplant is also an independent hazard element of sPGF except for aGVHD and CMV reactivation.


| INTRODUCTION
Complete and stable hematopoietic reconstitution is a key element of allogeneic hematopoietic stem cell transplantation (allo-HSCT) success. 1,2 The occurrence of initial hematopoiesis engraftment is generally within 4 weeks post-transplant. 3 The recipients who fail to achieve the initial engraftment or lose their initial hematopoietic reconstitution are defined as graft failure, which can be classified into poor graft function (PGF) and graft rejection. [4][5][6] PGF is a serious complicating disease after allo-HSCT, leading to high mortality. [7][8][9][10] Generally, PGF is divided into primary PGF, which fails to achieve initial hematopoietic reconstitution, and secondary PGF (sPGF), which loses initial hematopoietic reconstitution. 11,12 In clinic, sPGF is more frequent than primary PGF, with the incidence of 5%-27%. [7][8][9][10] Many factors have been demonstrated to be related to sPGF development, such as graft versus host disease (GVHD), virus infections, and so on. [5][6][7] However, whether haploidentical donor (HID) transplant is a hazard element of sPGF remains unclear.
In our research, we retrospectively analyzed the morbidity, hazard elements, and outcome in patients with sPGF after allo-HSCT. Our result suggested that transplantation from HID was a hazard element of sPGF.

| PATIENTS AND METHODS
Adult patients with hematological malignancies who received their initial transplantation were retrospectively reviewed between 1 January 2014 and 30 June 2019 at our institution. The patients who obtained initial neutrophils (NEUs) engraftment and platelets (PLTs) engraftment as well as a fully chimeric state by the +30 day after receiving transplantation were reviewed, and the patients undergoing non-myeloablative transplantation were excluded from this study. Protocol of our research was performed according to the Declaration of Helsinki. Our institution also approved this research according to Review Board.

| Evaluation points and definitions
Our research mainly explored the morbidity and hazard elements of sPGF. Reconstruction of NEUs was defined as in the absence of stimulating by granulocyte colonystimulating factor (G-CSF) at the first 3 successive days post-transplantation, the absolute NEU number could achieve 0.5 × 10 9 /L. Reconstitution of PLTs definition was that PLT number was ≥20 × 10 9 /L without PLT infusion at the first 7 successive days after receiving transplantation. The recipients who achieved consistent reconstitution of both NEUs and PLTs with no need for transfusion were defined as good graft function (GGF). sPGF definition was that sustained neutropenia (NEU count ≤0.5 × 10 9 /L), thrombocytopenia (PLT count ≤20 × 10 9 /L), and/or hemoglobin (Hb) ≤70 g/L for a minimum of 3 successive days with complete donor chimerism, or depending on requirements of G-CSF support and/or blood transfusion after day +30 post-HSCT. 11,12,17 In addition, patients with serious GVHD or hematological recurrence were removed from sPGF diagnosis. 11,12,17 Hematological recurrence definition is that the tumor cells appeared again in patients' peripheral blood and the rate of recurrence blasts in BM was greater than 5%. In addition, appearance of extramedullary infiltration at any time also belonged to relapse. 12,13 Overall survival definition was that time from transplantation to death or date of last follow-up in alive patients. The response criteria of sPGF were defined as follows: (1) CR: NEUs >1.5 × 10 9 /L and PLTs >50 × 10 9 /L for 3 continuous days posttreatment; (2) Partial response (PR): NEUs >0.5 × 10 9 /L and PLTs >20 × 10 9 /L for 3 continuous days post-treatment but failed to achieve the diagnostic criteria of CR; (3) NR: did not reach the above two standards; and (4) Overall response: including both CR and PR.

| Statistical analysis
Patient follow-up was up to 30 April 2020. Continuous variables were stated as the median and categorical variables were stated as a percentage (%). One-way ANOVA was performed for comparison of continuous variables. The Chi-squared test or Fisher's exact test was performed for comparison of percentage. Survival rate was analyzed by life table method. The Kaplan-Meier method was performed to analyze survival, nonrecurrent mortality, and cumulative incidence of CMV reactivation within +30 days. For comparison between different groups, the log-rank test (Mantel-Haenszel) was applied. Multivariate analysis used Cox regression to further evaluate the hazard elements. Cumulative incidence was applied to calculate the incidence of sPGF and the death was seen as a risk of competition. Cumulative incidence was also performed to calculate NEUs and PLTs response the death before response was seen as a risk of competition. Twosided p values were applied. A p < 0.05 was regarded as statistical significance. SPSS Version 19.0 was applied to analyze the statistical data. Competitive risk model in R method (R version 3.4.3) was used to analyze the cumulative incidence.

| Patient demographics and transplants characteristics
In all, 998 patients with malignant hematological diseases were reviewed in current retrospective research. Finally, 863 patients were reviewed while 125 were excluded due to early death or failing to achieve initial NEUs and PLTs engraftment by the day +30 post-transplants. Of the 863 patients, 413 underwent MSD, 114 MUD, and 336 HID transplants. Five hundred and fifteen males and 348 females were reviewed in our research and their median age was 32 years (range: 16-60). Primary diseases included AML (N = 406), ALL (N = 327), myelodysplastic syndromes (N = 89), and other hematological malignancies (N = 41). Based on the development of sPGF posttransplants, the reviewed patients were divided into sPGF group and GGF group (Table 1). Transplant characteristics between the two groups are shown in Table 1. Donor source (p < 0.0001), HLA disparity (p = 0.001), use of ATG (p = 0.009), acute graft versus host disease (aGVHD) within +30 days post-transplants (grades 1-4, p = 0.009), CMV reactivation within +30 days post-transplants (p < 0.0001), and EBV reactivation (p = 0.032) within +30 days post-transplants were significantly different between the two groups.
To explore the reason why HID transplant was a hazard element for PGF, we further analyzed the relation between the development of PGF and CMV reactivation in patients undergoing HID transplantation. Our results showed that only CMV reactivation was the hazard element for the development of PGF in HID transplant (p < 0.0001, [HR] 12.521 [95% CI: 5.982-26.209]) ( Table 3). In addition, we analyzed the incidence of CMV reactivation within +30 days, CMV serostatus of recipients and donors before transplantation, and levels of maximum viral loads among MRD, MUD, and HID. Our results showed that the cumulative incidence of CMV reactivation within +30 days of HID (19.0±2.1%) was higher than MSD (14.5±1.7%, p = 0.023) and MUD (12.3±3.1%, p = 0.035), while there T A B L E 1 (Continued) was no significant difference between MSD and MUD (p = 0.367). CMV serostatus (CMV IgG) of recipients and donors before transplantation was not significantly different among MSD, MUD, and HID (data not shown). The levels of median viral loads within +30 days in MSD, MUD, and HID were 8.02 × 10 4 copies/ml (range, 0.12-17.60 × 10 4 copies/ml), 12.48 × 10 4 copies/ml (range, 0.23-22.31 × 10 4 copies/ml), and 17.75 × 10 5 copies/ml (range, 0.07-85.60 × 10 4 copies/ml), respectively. The viral loads within +30 days in HID were higher than MSD (p < 0.0001) and MUD (p = 0.003), while there was no significant difference between MSD and MUD (p = 0.601).

| DISCUSSION
PGF is a severe complication that can threaten patients' life, and the occurrence of sPGF is more frequent than primary PGF. 9,11,12,17 In our retrospective study, our results proved that the overall cumulative incidence of sPGF within 180 days post-transplants was 6.0%, with 3.4%, 3.4%, and 10.1%, respectively, in MSD, MUD, and HID transplant. The multivariable analysis showed that hazard elements of sPGF included HID transplant, aGVHD, and CMV reactivation. The patients with sPGF had poorer survival than GGF. The incidence of sPGF varied from 5% to 27% after allo-HSCT, depending on the number of hazard elements. [7][8][9][10] Our results were consistent with the incidence of sPGF reported by Nakamae et al. 5 and Sun et al., 11 in which the incidence of sPGF within the first 100 days post-transplants was 7.0% and 5.7%, respectively. A report from Korean revealed that 12.7% patients developed sPGF in the recipients within 60 days after allo-HSCT. 6 Many factors may be associated with PGF development, such as prior alloimmunization, conditioning regimen, HLA matching, donor type, GVHD, and infections. 9,17 Our results showed that aGVHD and CMV reactivation were hazard elements of sPGF, which were consistent with literatures. 11,12,[18][19][20] More importantly, HID transplant was also demonstrated as an independent hazard element of sPGF in our research, which was not consistent with Sun et al. reported. 11 Emerging experimental and clinical evidence suggests that CMV infection is a major cause of sPGF. 11,12,[21][22][23][24] CMV might directly inhibit hematopoiesis by infecting hematopoietic stem cells and BM stromal cells [21][22][23][24] or indirectly inhibiting hematopoiesis through antiviral drug toxicities. 5 Some studies suggest that recipients undergoing HID transplant have a higher incidence of CMV reactivation, 25,26 which were consistent with our result. In addition, our result showed that CMV reactivation was the only hazard element of sPGF development in the patients with HID transplant. Several groups confirmed the clear association between DSA and primary graft failure as well as PGF in HSCT with HLA-mismatched donors. 27,28 Regretfully, the data of DSA in our study were incomplete so that DSA were not involved in our analysis. We will try to improve our DSA data in the future. Based on these, we speculated that the high risk of sPGF after HID transplantation might be associated with high incidence of CMV reactivation in HID. But further exploration is needed. The prognosis of sPGF is very poor. Limited available therapy options for patients with PGF are found, including hematopoietic growth factors and stem cell reinfusion as well as second transplantation. Hematopoietic growth factors are often effective only for short periods of time. Stem cell reinfusion or second transplantation is related to a high rate of risk of transplant-related mortality. 17,[29][30][31][32] Response rate of PGF reported in the literatures was 35%-85%. 9,33-35 Our previous study 9 suggested that BM-derived MSCs from a third-party donor combined with donor stem cell or cord blood were effective to PGF. In this study, we obtained the similar result to our previous results. 9 In addition, Han et al. reported that low-dose decitabine was effective in patients with isolated thrombocytopenia post-HSCT. 36 Of the seven patients with sPGF who received decitabine administration in our study, six had response and significant PLT recovery, which agreed with the good efficacy reported by Han et al. 36 Retrospective single-center analysis is the main inadequacy of our research. The multicenter studies are required to verify our observations.
In summary, except for aGVHD and CMV reactivation, HID transplant is also an independent hazard element of sPGF. The high risk of sPGF in HID transplant might be associated with their high incidence of CMV reactivation.