Phase II Trial of Reduced-Intensity Fludarabine, Melphalan, and Total Body Irradiation Conditioning With Haploidentical Donor Peripheral Blood Stem Cell Transplant

Abstract

Human leukocyte antigen (HLA) haploidentical (haplo) hematopoietic cell transplantation (HCT) with post-transplant cyclophosphamide (PTCy) has expanded access to this curative therapy with rates of graft-versus-host disease (GVHD) that are similar to HLA matched donor HCT [1]. The original Johns Hopkins PTCy platform uses a reduced intensity conditioning (RIC) regimen consisting of Fludarabine (Flu) 150 mg/m², Cy 14.5 mg/kg on Day-6 and-5, total body irradiation (TBI) 200 cGy on Day-1 (Flu/Cy/TBI) and a bone marrow graft (BMT). Though this regimen yields favorable toxicity including rates of non-relapse mortality (NRM) ~10%–15%, severe acute GVHD ~10%, and chronic GVHD ~10%, outcomes have been marred by high rates of disease relapse > 50% [1].

While myeloablative conditioning (MAC) regimens can reduce the risk of relapse, older and frail patients cannot tolerate MAC regimens. To improve relapse after RIC haplo HCT with PTCy, investigators at the MD Anderson Cancer Center attempted a conditioning regimen consisting of Flu 160 mg/m², melphalan (Mel) 100–140 mg/m², and either Thiotepa 5 mg/kg or TBI 200 cGy [2]. While they reported low 1-year relapse rates of 19%, NRM was high at 21%, resulting in 1-year disease-free survival (DFS) of 60%.

As haplo HCT with PTCy is increasingly used, strategies to improve outcomes are essential. We hypothesized that reducing the Mel dose to 70 mg/m² in a Flu/Mel70/TBI regimen followed by haplo peripheral blood stem cell transplant (PBSCT) with PTCy would successfully improve DFS by reducing relapse without a significant increase in NRM compared to Flu/Cy/TBI haplo HCT in patients unfit for MAC.

This trial, NCT04191187, is single-institution phase II study of Flu/Mel70/TBI haplo PBSCT with PTCy/sirolimus/mycophenolate mofetil (MMF) in patients with high-risk hematologic malignancies. It was approved by the Advarra Institutional Review Board and is compliant with the Declaration of Helsinki.

Eligible patients were required to be ≥ 55 years or < 55 years with significant comorbidities defined as an HCT comorbidity index (HCT-CI) ≥ 3, and receiving an HLA haplo donor HCT for a hematologic malignancy.

All patients received conditioning with Flu 30 mg/m²/day from −6 to −2 based on actual body weight, Mel 70 mg/mg² on day-6 based on actual body weight, and TBI 200 cGy on day-1. On day 0, patients received the PBSCT (Figure 1A). The target CD34+ cell dose was 5 × 10⁶ CD34+ cells/kg, with a maximum allowable dose of 7 × 10⁶ CD34+ cells/kg. GVHD prophylaxis consisted of PTCy/tacrolimus/MMF for the first five subjects. The protocol was subsequently modified to follow the PTCy/sirolimus/MMF platform published separately by our group [3].

FIGURE 1

Open in figure viewerPowerPoint

(A) Treatment schema. Outcomes over time for (B) grade II–IV acute graft-versus-host disease (GVHD), (C) moderate to severe chronic graft-versus-host disease (D) relapse, (E) non-relapse mortality, (F) disease free survival, and (G) overall survival.

The primary endpoint of this study was DFS. Comparing to a historic DFS at 18 months of 40%, we assumed that the DFS in patients treated with Flu/Mel70/TBI would be 60%, corresponding to a hazard ratio (HR) of 0.557 [1, 2]. A total sample size of 34 subjects provided 90% power to detect a HR of 0.557 using the one-sample log-rank test at a one-sided significance level of 0.1. Cumulative incidence was used to estimate the probabilities of GVHD, relapse, infection, and neutrophil and platelet recovery, treating deaths as a competing risk. For GVHD and NRM, relapse also served as a competing risk. The Gray test was used accordingly. Ninety-five percent confidence intervals were estimated from respective standard errors and the complementary log–log transformation. Analyses were performed and plots generated using SAS 9.4 (SAS Institute, Cary, NC) and/or R 3.0.2.

Detailed baseline characteristics of the 34 subjects treated are shown in Table 1. Median time to neutrophil engraftment was 18 days (range: 14–42), and the incidence by day 30 was 88%. There were no cases of primary graft failure among evaluable patients who survived past day 30. The median time to platelet engraftment was 29 days (range: 16–67), and engraftment by day 60 was 82%.

TABLE 1. Baseline characteristics.

Baseline characteristics
Follow-up time in months, median (range)	18 months (range: 18–20)
Patient age in years, median (range)	66 (31–74)
Donor age in years, median (range) CD34+ cells/kg, median (range)	34 (range: 21–49) 5.72 × 106 (range: 4.45–7.0 × 106)

Variables		n = 34	%
Patient sex	Female	17	50%
	Male	17	50%
Donor relation	Son	19	56%
	Daughter	14	41%
	Brother	1	3%
Karnofsky Performance Scale	100	2	6%
	90	23	68%
	80	8	24%
	≤ 70	1	3%
Hematopoietic Cell Transplant Comorbidity Index	≥ 3	14	41%
	0–2	20	59%
Race/ethnicity	White/Non-Hispanic	25	74%
	Black/Non-Hispanic	5	15%
	Hispanic	2	6%
	Asian	2	6%
Disease	Acute myeloid leukemia	15	44%
	Myelodysplastic syndrome	10	29%
	Chronic myelomonocytic leukemia	4	12%
	Myeloproliferative neoplasm	2	6%
	Lymphoma	2	6%
	Acute lymphoblastic leukemia	1	3%
Disease status at transplant	Complete remission/MLFS	13	38%
	Partial remission	1	3%
	Stable disease	20	59%
GVHD prophylaxis	PTCy/sirolimus/MMF	29	85%
	PTCy/tacrolimus/MMF	5	15%
ELN for acute myeloid leukemia (n = 15)	Adverse	7	47%
	Intermediate	8	53%
IPSS-M for MDS and CMML (n = 14)	Very high	10	71%
	High	4	29%
CMV serology (donor/recipient)	Donor+/recipient+	10	29%
	Donor−/recipient+	17	50%
	Donor+/recipient−	2	6%
	Donor−/recipient−	5	15%
Donor/recipient sex	Male to female	8	24%
	Female to male	5	15%
	Same	21	61%

• Abbreviations: CMML, chronic myelomonocytic leukemia; CMV, cytomegalovirus; ELN, European LeukemiaNet; GVHD, graft-versus-host disease; IPSS-M, Molecular International Prognostic Scoring System; MDS, myelodysplastic syndrome; MLFS, morphologic leukemia free state; MMF, mycophenolate mofetil; PTCy, post-transplant cyclophosphamide.

Of 32 subjects with unsorted marrow chimerism samples, 30 (93.8%) had > 95% donor chimerism (range: 35% to 100%) at day 30. One of the two patients with low donor chimerism reached 100% donor chimerism at day 90, while the other relapsed and ultimately succumbed to his disease. Similarly, 30 of 31 (96.8%) with available peripheral blood chimerism samples reached > 95% donor in both the myeloid (range: 73% to 100%) and lymphoid compartments (range: 4% to 100%) by day 30.

A total of four subjects experienced grade II–IV acute GVHD, resulting in a cumulative incidence of 11.8% (95% C.I.: 3.6% to 25.1%) by day 100 (Figure 1B). Of these, one subject developed grade IV acute GVHD and there were no cases of grade III acute GVHD. At 18 months, the cumulative incidence of moderate to severe chronic GVHD was 8.8% (95% C.I.: 2.2% to 21.4%, Figure 1C).

With median follow up of 18 months for survivors (range 18–20), the cumulative incidence of relapse at 18 months was 11.8% (95% C.I.: 3.6% to 25.2%, Figure 1D). The diagnoses for the four subjects with disease relapse were MDS (n = 3) and MPN (n = 1). The cumulative incidence of NRM at 18 months was 17.6% (95% C.I.: 7.0% to 32.2%, Figure 1E). Causes of death for these six subjects with NRM events were organ failure (n = 2), sepsis (n = 2), Stevens-Johnson Syndrome (n = 1), and acute kidney injury (n = 1).

The observed 18-month DFS was 70.6% (one-sided 90% C.I.: ≥ 59.2%, Figure 1F), thus meeting the primary endpoint of the trial of 18-month DFS of 40% (p = 0.0017). The median DFS was not reached. Kaplan–Meier estimate for OS at 18 months was 73.3% (95% C.I.: 54.9% to 85.1%, Figure 1G). At 18 months, GRFS was 61.8% (95% C.I.: 43.4% to 75.7%).

Grade 3–5 adverse events are listed in Table S1. CRS and viral infections are adverse events of interest in haplo HCT. CRS was observed in 28 subjects (82.4%), including 23 subjects (67.6%) with grade 1 CRS and five subjects (14.7%) with grade 2 CRS. Two of these subjects received tocilizumab for treatment of CRS. There were no cases of grade 3–5 CRS. Grade 3 CMV infection was identified in five subjects (14.7%), including one case of CMV pneumonitis. Notably, letermovir for CMV prophylaxis was adopted at our institution during the latter portion of the trial and was administered to 15 subjects, of whom 6 (40%) developed CMV infections. Grade 3 HHV6 reactivation occurred in five cases (14.7%) and resolved with foscarnet, including two cases with confirmed HHV6 in the cerebrospinal fluid. There were no grade 4–5 viral infections.

Currently, the role of MMF in the PTCy platform is debated. We collected blood samples on post-transplant day 7 at pre-MMF dose, end of MMF infusion, and 1, 2, 4, and 8 h after the end of MMF infusion to study the pharmacokinetics of MMF and its active metabolite mycophenolic acid (MPA). Concentrations were determined by the Cancer Pharmacokinetics and Pharmacodynamics Core at the Moffitt Cancer Center using LC–MS/MS methods that have been validated according to ICH/FDA guidelines for bioanalytical analysis. MMF and MPA concentrations were evaluated by non-compartmental analysis (NCA) on Day 5 in 33 subjects to determine if steady-state pharmacokinetics of MMF were affected by the coadministration of cyclophosphamide and were consistent with single-agent exposure, highlighting that there does not appear to be a significant MMF drug–drug interaction with Cy exposure (Figure S1, Table S2). A higher volume of distribution of MPA was correlated with a lower risk of grade II-IV acute GVHD (HR = 0.06, 95% CI: 0.01 to 0.54, p = 0.01) suggesting MMF does contribute to acute GVHD prevention in this regimen (Table S3).

To date, no studies have evaluated the effects of genetic polymorphisms on patients receiving PTCy. Prior to starting therapy, blood samples were collected from 24 subjects to assess the presence of single nucleotide polymorphisms (SNPs) in genes of interest that influence drug metabolism (Table S4). The association with the incidence of acute and chronic GVHD was explored by the Fine-Gray regression model. Nine genes with differential expression within the cohort were evaluated for associations with clinical outcomes: GSTP1, SLCO2B1, LST3, CYP1A2, SULT1A1, SLC15A2, UGT2B15, UGT2B7, and ABCG2. No significant interactions with GVHD, relapse, or survival were identified (Table S5). However, the sample size in this trial is not adequately powered to evaluate such differences, and future studies are warranted as this may impact outcomes in different ethnicities or inform dosing.

The Flu/Mel70/TBI regimen reported here is a promising approach to RIC haplo PBSCT with PTCy, meeting the primary endpoint of improved DFS compared with historical benchmarks in a patient cohort unfit for MAC. While Mel-based conditioning has been successful for matched donor HCT, incorporating Mel with haplo HCT has previously demonstrated high rates of CRS, GVHD, and NRM, even in younger cohorts [4]. We reduced the dose of Mel and moved it earlier in the regimen to day-6 in hopes of decreasing tissue injury and inflammation that drives these complications after transplant [5]. We also hypothesized that the inclusion of TBI after the Flu/Mel may deplete recipient T cells that would contribute to CRS after cell infusion [6]. With these modifications, toxicity and engraftment were similar to prior studies and, encouragingly, relapse rates were low. This regimen offers a favorable approach for patients unfit for MAC regimens and merits further investigation in larger, multicenter prospective trials.