Sirolimus versus cyclosporine in haploidentical stem cell transplantation with posttransplant cyclophosphamide and mycophenolate mofetil as graft‐versus‐host disease prophylaxis

Abstract Sirolimus has emerged as an alternative to calcineurin inhibitors‐based (CNI) graft‐versus‐host disease (GVHD) prophylaxis. This retrospective study compares the outcome of 133 consecutive adult patients with haematological malignancies undergoing haploidentical stem cell transplantation with posttransplant cyclophosphamide (PTCy) and mycophenolate mofetil (MMF), combined with cyclosporine A (PTCy–CsA–MMF, n = 67) or sirolimus (PTCy–Sir–MMF, n = 66) as GVHD prophylaxis strategy. The median follow‐up was 48 (range 22–83) and 13 (range 3–33) months, respectively. PTCy–CsA–MMF was associated in multivariate analyses with a higher risk of acute kidney injury (HR 2.1, 95% CI, 1.21–3.57, p = .008) and thrombotic microangiopathy (HR 12.5, 95% CI, 1.66–93.5, p = .014), whereas PTCy–Sir–MMF was associated with a higher risk of hepatic sinusoidal obstruction syndrome (SOS) (HR 10.8, 95% CI, 1.52–77, p = .018), especially late‐onset forms, which totally resolved and none of the patients needed discontinuation of sirolimus. Two SOS‐related deaths were detected, both in the PTCy–CsA–MMF subgroup. Both GVHD prophylaxis strategies were otherwise comparable in terms of engraftment, GVHD incidence and survival.

In recent years, sirolimus has been explored as an alternative to CsA-based prophylaxis and some advantages have been suggested, including a more favourable toxicity profile, especially in terms of renal toxicity [3], better immune reconstitution [4,5], more powerful graft-versus-lymphoma effect [6,7], and reduced incidence of Cytomegalovirus (CMV) DNAemia [8].
Encouraging results have recently been reported using CNI-free GVHD prophylaxis with PTCy and sirolimus in patients undergoing SCT from both HLA-matched related and unrelated donors [7,9], as well as haploidentical donors [10][11][12]. However, a formal comparison of CNIfree with CNI-based GVHD prophylaxis is still missing.
In the present study, we aimed to compare outcomes and toxicities of patients undergoing haplo-SCT at two institutions using PTCy, sirolimus, and mycophenolate mofetil (PTCy-Sir-MMF) with a historical cohort of haplo-SCT with PTCy, CsA and MMF (PTCy-CsA-MMF).

Patients and eligibility criteria
In this retrospective study, we included two consecutive cohorts of

Donor selection and source of stem cells
Selection of a haploidentical donor did not vary between periods and it has previously been described in detail [12]. Briefly, a haploidentical family donor was selected according to the following considerations

Key Points
• Both prophylactic regimens provide comparable efficacy preventing GVHD. • PTCy-CsA-MMF was associated with a higher risk of acute kidney injury and thrombotic microangiopathy.
• PTCy-Sir-MMF was associated with a greater probability of sinusoidal obstruction syndrome at expense of lateonset forms, which totally resolved with treatment and no associated mortality was detected.
(in order of preference): absence of recipient HLA antibodies against donor antigens, male sex, younger age, matched CMV serostatus and matched ABO group. Initially, peripheral blood mobilised CD34+ stem cells were used at both transplant centres. In mid-2019, bone marrow source was prioritised, when feasible, at both centres in elderly donors without comorbid conditions, in order to mitigate the development of severe GVHD.

Conditioning regimen
The conditioning regimens used during the study period are detailed in

Graft-versus-host disease prophylaxis
Both GVHD prophylaxis schemas included cyclophosphamide (50 mg/kg/day) intravenously on days +3 and +4 and MMF at In the absence of GVHD or relapse, CsA or sirolimus were gradually tapered from day +90 to day +150.

Supportive care
Patients were nursed in HEPA-filtered rooms.

Definitions
Myeloid recovery was defined as the first day of an absolute neutrophil count of 0.5 × 109/L lasting for 3 consecutive days. Platelet recovery was defined as the first day of a platelet count of 20 × 109/L or higher, without transfusion support for 7 consecutive days. Patients who survived more than 28 days after transplantation and failed to achieve myeloid engraftment were considered to have primary graft failure.

Statistical analysis
Primary end point was overall survival (OS) and the cumulative inci-

Platelet engraftment
Nine

Acute GVHD
In

F I G U R E 2 Cumulative incidence of grades II-IV acute graft-versus-host disease (aGVHD) according to the GVHD prophylaxis (A). Cumulative incidence of grades III-IV aGVHD according to the GVHD prophylaxis (B)
F I G U R E 3 Cumulative incidence of chronic graft-versus-host disease (cGVHD) according to the GVHD prophylaxis (A). Cumulative incidence of moderate to severe cGVHD according to the GVHD prophylaxis (B)  Table 3.

Infection, no. 4 a 4 b
Sinusoidal obstruction syndrome, no. and conditioning intensity, with no significant differences between subgroups.  Figure 5).

Multivariate analyses for different outcomes and posttransplant events
In multivariate analysis, PTCy-CsA-MMF was associated with a higher probability of AKI (HR 2.1, 95% CI, 1.21-3.57, p = .008) and larger risk of TMA (HR 12.5, 95% CI, 1.66-93.5, p = .014), whereas PTCy-Sir-MMF was associated with a greater probability of SOS (HR 10.8, 95% CI, 1.52-77, p = .018). Details of statistical analyses are described in Table S3. None of these variables, including subgroup analyses of AKI grades, had a statistically significant impact on OS or EFS when analysed by Cox proportional-hazards regression model.

DISCUSSION
In this study, we found that a CNI-free GVHD prophylaxis with PTCy- In contrast with prior studies [7,9,11], in which the cumulative incidence of NRM at 1 year ranged from 13% to 17%, in our series it was 24%, with no significant differences between the two prophylactic approaches. The apparent decrease in the cumulative incidence of relapse achieved in the sirolimus-treated group (10% at 2 years), not only compared to the group treated with CsA in our study, but also to that reported in other studies using a similar CNI-free prophylaxis [7,[9][10][11], should be interpreted cautiously. The higher proportion of patients who received myeloablative conditioning in the CNI-free prophylaxis group, which may lead to a lower relapse rate [32], could explain, at least in part, these encouraging results. A larger sample and longer follow-up could establish the relative role of GVHD conditioning and prophylaxis in these and other survival outcomes.
Conflicting results have been reported regarding an increased risk of CMV infection in haplo-SCT [12,33,34], as well as an increased incidence of CMV-related disease in this setting [33,35]. We have confirmed that approximately half of the patients in our study had CMV infection requiring PET, but the rate of CMV-related disease in the PTCy-Sir-MMF group remained very low (3%), despite the significant proportion of older patients in this group, which has been previously described as an independent risk factor for CMV infection [34].
Regarding toxicity, it should be noted that the cohort with PTCy-Sir-MMF prophylaxis had a lower incidence of AKI and TMA, but that of SOS increased, especially late-onset forms (seven out of nine patients), which typically manifested as fluid retention and hepatitis without hyperbilirubinemia. All of them resolved and none of the patients needed discontinuation of sirolimus. Of note, the two SOSrelated deaths were detected, both in the PTCy-CsA-MMF subgroup.
The use of sirolimus has been associated with larger risk of SOS in the setting of CNI-based GVHD prophylaxis [36], although previous experience in the context of CNI-free regimens does not support this association [7,10,11]. The higher incidence of SOS in our study could be related to our target of sirolimus levels, which are higher than previous studies [7,11], following prior evidence of a lower incidence of CMV [8]. To note, Bejanyan et al. used similar target of sirolimus, showing a lower rate of SOS. This study involves a younger cohort of patients with a lower proportion of lymphoproliferative diseases. The use of pharmacokinetically targeted Bu could also explain this phenomenon [10]. The analysis of sirolimus through levels and its role in SOS is extremely challenging. Sirolimus is largely cleared by the liver via the cytochrome P450 system (CYP3A4), and when liver dysfunction occurs sirolimus levels increase as its clearance is significantly reduced [37]. As previously suggested [12], replacing busulfan with treosulfan or establishing busulfan levels monitoring is now under consideration in our institutions in order to decrease this complication [38,39].
In conclusion, haplo-SCT with PTCy-Sir-MMF as GVHD prophylaxis provides similar outcomes to those with PTCy-CsA-MMF. In terms of toxicity, while CNI-free prophylaxis produced less renal toxicity and microangiopathy, an increase in late-onset SOS was also observed, all of which resolved with treatment without sirolimus discontinuation. Prospective randomised studies comparing GVHD prophylaxis schemas are warranted in order to confirm these findings.