Phase I/II multisite trial of optimally dosed clofarabine and low-dose TBI for hematopoietic cell transplantation in acute myeloid leukemia

Abstract

Clofarabine is an immunosuppressive purine nucleoside analog that may have better anti-leukemic activity than fludarabine. We performed a prospective phase I/II multisite trial of clofarabine with 2 Gy total body irradiation as non-myeloablative conditioning for allogeneic hematopoietic cell transplantation in adults with acute myeloid leukemia who were unfit for more intense regimens. Our main objective was to improve the 6-month relapse rate following nonmyeloablative conditioning while maintaining historic rates of non-relapse mortality (NRM) and engraftment. Forty-four patients, 53 to 74 (median: 69) years, were treated with clofarabine 150 to 250 mg/m², of whom 36 were treated at the maximum protocol-specified dose. One patient developed multifactorial acute kidney injury, but no other grade 3–5 non-hematologic toxicities were observed. All patients fully engrafted. The 6-month relapse rate was 16% (95% CI, 5%–27%) among all patients and 14% (95% CI, 3%–26%) among high-risk patients treated at the maximum dose, meeting the pre-specified primary efficacy endpoint. Overall survival was 55% (95% CI, 40%–70%) and leukemia-free survival was 52% (95% CI, 37%–67%) at 2 years. Compared to a historical high-risk cohort treated with the combination of fludarabine at 90 mg/m² and 2 Gy TBI, protocol patients treated with the clofarabine-TBI regimen had lower rates of overall mortality (HR of 0.50, 95% CI, 0.28–0.91), disease progression or death (HR 0.48, 95% CI, 0.27–0.85), and morphologic relapse (HR 0.30, 95% CI, 0.13–0.0.69), and comparable NRM (HR 0.85, 95% CI 0.36–2.00). The combination of clofarabine with TBI warrants further investigation in patients with high-risk AML.

1. INTRODUCTION

The median age of patients diagnosed with acute myeloid leukemia (AML) is 68 years, and more than 30% of patients with AML are over 75 years.¹ Retrospective and registry studies suggest that allogeneic hematopoietic cell transplantation (HCT) offers the best chance of durable remissions compared to non-transplant strategies, even in older adults.^2–4 For many patients with AML, including those with high risk genetic features, HCT is the only potentially curative option. Minimally toxic non-myeloablative fludarabine-containing preparative regimens have been developed to permit donor engraftment in older and unfit patients, but the ensuing graft-vs-leukemia effect is often too slow to develop or insufficiently robust, such that reported relapse rates generally range from ~30% to 60% within 2 years of HCT.⁵

Clofarabine (2-chloro-9-[deoxy-2′-fluoro-8-D-arabinofuranosyl] adenine; CI-F-ara-A; CAFdA) is a rationally designed, second generation purine nucleoside analogue. It was designed as a hybrid molecule to overcome the limitations and incorporate the best antileukemic and immunosuppressive qualities of both fludarabine (F-ara-A) and cladribine (2-CdA).⁶ Clofarabine is superior to fludarabine in inhibiting ribonucleotide reductase⁶ and superior to cladribine in inhibiting DNA polymerase.⁷ It enhances tumor radiosensitivity to the same degree as 5-fluorouracil and gemcitabine by preventing the repair of DNA damage.^8–10 In addition, unlike fludarabine, clofarabine can directly induce apoptosis by disrupting the integrity of mitochondria,¹¹ which may account for its potent cytotoxic effects on nondividing lymphocytes.¹² These properties may be particularly effective not only in enhancing specific anti-leukemia effects, but also in promoting allograft engraftment by suppressing host T-lymphocytes in the non-myeloablative HCT setting. We hypothesized that substituting clofarabine for fludarabine in the conditioning regimen with low-dose (2 Gy) total body irradiation (TBI) would reduce the relapse rate without compromising the favorable toxicity profile of this regimen. The efficacy part of the study was preceded by a dose-finding part, aimed at determining the optimal clofarabine dose when added to low-dose TBI.¹⁰

2. METHODS

2.1. Study design and objectives

The primary objective was to improve the 6-month relapse rate following non-myeloablative conditioning among patients with AML considered to be at high risk for serious toxicities if given a high-dose preparative regimen. The first aim was to determine the maximum tolerated dose (MTD) of clofarabine in combination with 2 Gy TBI. The second aim was to determine the efficacy of the clofarabine MTD combined with 2 Gy TBI; a satisfactory improvement was a priori defined as reducing the 6-month relapse rate to ≤20% among high-risk patients and to ≤5% among low-risk patients. For the high-risk group, 36 patients treated at the MTD were estimated to provide 83% power to detect an absolute 15% improvement in the relapse/progression rate (35% to 20%) with a one-sided significance level of 0.13. For the low-risk group, 30 patients treated at the MTD were estimated to provide 81% power to detect an absolute 10% improvement in the relapse-progression rate (from 15% to 5%) with a one-sided significance level of 0.15.

To stratify patients with respect to risk of relapse, we used a risk score developed on a historical cohort of 274 patients with AML transplanted at our Center from related or unrelated donors after conditioning with 2 Gy TBI, with or without fludarabine (90 mg/m²) (supplementary Table S1); a calcineurin inhibitor and mycophenolate mofetil (MMF) were used for immunosuppression.¹³ Patients were then stratified according to whether the total risk score was low (≤0) or high (>0), with historical 6-month relapse rates of 14% and 36%, and 2-year relapse rates of 17% and 49%, respectively.

Only patients at high risk for relapse (score >0) were enrolled in the dose-finding Part 1. Dose limiting toxicities (DLTs) were defined by NCI Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 grade 4 toxicity of the lungs, heart, liver, or kidney (not resolving within 48 hours), gastrointestinal tract, or central nervous system. The 3+3 design entailed a starting dose of clofarabine 30 mg/m²/day on days −6 to −2 (total dose: 150 mg/m²) and escalation to 40 mg/m²/day (total dose: 200 mg/m²) and 50 mg/m²/day (total dose: 250 mg/m²) in the absence of a DLT (assessed when the last subject of a cohort reached 21 days after clofarabine initiation). In Part 2, the MTD of clofarabine determined from Part 1 was planned to be administered to 33 additional high-risk patients and 30 additional low-risk patients.

Secondary objectives included assessing leukemia-free and overall survivals (LFS, OS), day-100 non-relapse mortality (NRM, goal: <5%), engraftment rate (goal: ≥95%), and pharmacokinetic analysis.

The initial TBI dose was 2 Gy. The protocol made provisions to increase the TBI dose by 1 Gy (up to a maximum of 4 Gy) if the rate of graft rejection exceeds the historical rate of 5%. Graft rejection was defined as the inability to detect or loss of detection of >5% donor T cells as a proportion of the total T-cell population.

2.2. Eligibility

Patients with AML were eligible to enroll if they were ≥55 years of age or 1–55 years of age with comorbidities or a history of prior therapy considered to substantially increase the risk for serious toxicities if given a conventional, high-dose preparation regimen. They were required to have an HLA-identical related donor or unrelated donor matched at HLA-A, -B, -C, -DRB1 and -DQB1 loci or mismatched for a single HLA allele defined by high resolution typing at HLA-A, -B, or -C. They had to be in a morphologic leukemia-free state (marrow blasts <5% without extramedullary disease or morphologically detected circulating blasts, and any central nervous system disease must have cleared with intrathecal chemotherapy) as assessed within 21 days before conditioning. Key exclusion criteria included a left ventricular ejection fraction <35%; a corrected DLCO or total lung capacity or FEV1 <40%; supplementary continuous oxygen; undiagnosed/unstable pulmonary nodule(s); cirrhosis; chronic viral hepatitis with serum bilirubin >3 mg/dL; below-normal creatinine clearance; poorly controlled hypertension despite multiple antihypertensives; non-melanoma solid cancer requiring therapy or treated into remission but with an estimated 5-year relapse-rate >20%; HIV positivity; active bacterial infection or fungal infection not responsive to therapy; and Karnofsky score <60. Patients with FAB M3 AML were eligible if in second (or greater) complete remission (CR2). Only patients with a high relapse risk score¹³ (>0) were eligible for Part 1.

2.3. Treatment plan: Conditioning, stem cell source, GvHD prophylaxis, supportive care

Patients were treated with clofarabine as a 2-hour intravenous infusion daily for 5 days (days −6 to −2) according to their dose-level cohort, followed by TBI on day 0. Then unmodified G-CSF-mobilized peripheral blood mononuclear cells were infused. Hepato- or nephrotoxic drugs (e.g., voriconazole, vancomycin) were contraindicated on days of clofarabine administration. Graft-vs-host disease (GvHD) prophylaxis was with cyclosporine (CSP) and MMF. No serotherapy was given.

CSP was started on day −3, and in the absence of GvHD, CSP was tapered starting on day +56 for related donor grafts and on day +100 for unrelated donor grafts, and discontinued on day +180. MMF was started at 15 mg/kg on day 0 (maximum 45 mg/kg/day). For related donor grafts, MMF was continued q12h until day +28 and stopped without tapering. For unrelated donor grafts, MMF was continued q8h through day +40, and, in the absence of GVHD, tapered starting on day +41 and discontinued +96. (See Supplementary Figure S1.)

2.4. Post-HCT assessments

Patients underwent routine bone marrow examinations on days +28, 56, 84, 180, 365, 1.5 years, annually in years 2–5, and when clinically indicated to assess donor cell engraftment and disease status. Disease response assessments followed international criteria.¹³ Donor chimerism was evaluated in peripheral blood T cells (CD3+), granulocytes (CD33+), and natural killer cells (CD56+), and unsorted marrow.

Disease progression was defined as persistence or appearance of aberrant blasts on a bone marrow aspirate by flow cytometry or of a cytogenetic or molecular abnormality previously associated with the patient’s AML. Relapse was defined as >5% blasts by morphology on a marrow aspirate or morphologically-apparent circulating blasts. Graft rejection was defined as the inability to detect or loss of detection of >5% donor CD3+ T cells as a proportion of the total T-cell population. Graft failure was defined as grade 4 thrombocytopenia and neutropenia after day +28 lasting >2 weeks and refractory to growth factor.

2.5. Statistical analysis

Descriptive statistics were used to summarize patient characteristics, toxicities, engraftment, incidence of GvHD and immune reconstitution. Comparisons between protocol and historic patients were by chi-squared test or Wilcoxon rank sum test. Death was considered a competing risk for acute and chronic GvHD. Death in remission was considered a competing risk for relapse. Overall survival (OS) and leukemia-free survival (LFS) were estimated using the Kaplan-Meier method. Event rates for endpoints subject to competing risks were estimated by cumulative incidence methods. Cox regression was used for time-to-event analyses comparing protocol and historic patients. All computations were performed in SAS, version 8.

2.6. Pharmacokinetic study

All patients enrolled in Part 1 were offered participation in the pharmacokinetics (PK) sub-study. Blood was drawn on day −6 before clofarabine dosing, at the end of the infusion, and at 3, 4, 5, 6 and 24 hours after the start of infusion. Only pre-dose samples were drawn on days −5, −4, and −3. For clofarabine quantitation, plasma samples were analyzed using tandem liquid chromatography – mass spectrometry according to the methods of Berg et al.,¹⁴ the limit of quantitation is 0.864 ng/ml with a coefficient of variation of 8.06%. Areas under the concentration-time curves (AUC, ug×h/L) for day −6 were calculated using the trapezoidal rule. Clearance (CL, L/h) was calculated as CL = does/AUC_0→∞ CL was normalized to actual weight.

2.7. Data sharing

This study is a therapeutic clinical trial. There is no present intention to seek an IRB waiver of consent to share individual patient-level data.

3. RESULTS

3.1. Patient characteristics

Forty-four patients enrolled in the trial, with 9 in Part 1, 33 in the Part 2 high-risk cohort, and 2 in the Part 2 low-risk cohort. Median age was 69 years (range, 53–74). Recruitment to the low-risk cohort was halted in 08/2014 due to low accrual. The two low-risk patients are included in the toxicity and PK analyses, but the efficacy analyses were restricted to the high-risk patients treatment at the MTD. The patient characteristics are summarized in Table 1. The median follow-up of surviving patients was 31 (range 4 to 83) months. All patients received 2 Gy TBI.

TABLE 1.

Patient and disease characteristics (n=44)

Age, years, median (range)	69 (53 – 74)
Male sex, n (%)	28 (64)
Prior HCT, n (%)	1 (2)
None	40 (91)
Allogeneic	1 (2)
Autologous	3 (7)
HCT-CI, n (%)15
0–1	3 (7)
2–3	19 (43)
≥4	22 (50)
Performance status, n (%)
90–100	24 (55)
80–85	15 (34)
60–70	5 (11)
Donor type, n (%)
Matched related	8 (18)
Matched unrelated	35 (80)
Allele-mismatched unrelated	1 (2)
CD34+ cell dose, 106/kg, median (range)	8.0 (3.8–23.6)
Therapy-related AML, n (%)	10 (23)
Evolved from prior MDS/MPN, n (%)	10 (23)
AML stage, n (%)
CR 1	35 (80)
CR 2	8 (18)
CR 3	1 (2)
MRD positivity pre-HCT, n (%)	18 (41)
ELN 2017 classification, n (%)
Favorable	6 (14)
Intermediate	17 (39)
Unfavorable	21 (48)

3.2. Toxicity and non-relapse mortality

No DLTs were observed at any dose level during Part 1. In all, 42 of 44 patients completed their planned infusions. One patient’s clofarabine dose was reduced from 50 mg/m² to 25 mg/m² on days −4, −3, and −2 due to grade 2 elevation in liver function tests, but he was transplanted successfully. The transaminitis never reached grade 3/4. Another patient received only 4 doses of clofarabine prior to HCT due to development of hyperbilirubinemia at 1.5 mg/dL. After cell infusion, this patient deteriorated with septic shock and multiorgan failure, including liver failure and acute renal failure, and died on day +18. A third patient developed acute kidney injury prior to the 5^th dose of clofarabine. He completed all planned chemotherapy infusions. Transplant was aborted due to renal failure and he subsequently died of multiorgan failure. No other grade 3–5 non-hematologic toxicities were observed.

Including all patients who started conditioning, day 100 NRM was 7% (95% CI, 0% to 14%); at 2 years NRM was 28% (95% CI, 14%–42%). Causes of NRM included GvHD and infection (n = 11), cardiac arrest (n = 1), early death due to acute kidney injury and multiorgan failure (n=2, described above), and unknown (n = 1).

3.3. Engraftment and chimerism

At day 28, 25 of 41 evaluable patients (61%) had full donor (≥95%) CD3 chimerism; the remaining 16 patients were all >50%. All but one of 38 evaluated patients (97%) had full donor CD56 chimerism at day 28. At day 84, 27 of 35 evaluable patients (77%) had full donor CD3 chimerism, 7 of the remaining patients (20%) had chimerism > 50%, and 1 was at 44%. Thirty of 35 evaluable patients (86%) had full donor CD33 chimerism at day 84, 4 of the remaining patients had chimerism >50%, and 1 was at 3%. This patient had converted to 100% CD33 chimerism when assessed at one year.

3.4. Graft-vs-host disease

The cumulative incidence of grade II-IV acute GvHD by 100 days was 65% (95% CI, 51%–79%). The cumulative incidence of severe grade III-IV acute GvHD by 100 days was 9% (95% CI, 1%–18%). The cumulative incidence of chronic extensive GvHD by 2 years was 49% (95% CI, 33%–64%).

3.5. Relapse and survival

The cumulative incidence of morphologic relapse at 6 months was 19% (95% CI, 5%–27%) among all study patients and 14% (95% CI, 3%–26%) among high-risk patients treated at the MTD. At 2 years the cumulative incidences of relapse were 22% (95% CI, 9%–39%) and 21%, (95% CI, 7%–35%), respectively (Figure 1).

FIGURE 1. Outcomes of patients with high-risk AML conditioned with clofarabine-low dose TBI compared to historical high-risk and low-risk patients conditioned with fludarabine-low dose TBI.

(A) Overall survival for patients transplanted for AML. (B) Non-relapse mortality. (C) Relapse. (D) Leukemia-free survival. Red line: High-risk patients treated on study with clofarabine 250 mg/m² (n = 35). Black line = Low-risk historical patients treated with fludarabine-TBI. Blue line = High-risk historical patients treated with fludarabine-TBI. Use of clofarabine conditioning appears to overcome clinical indicators of high relapse risk, resulting in survival outcomes comparable to historical low-risk patients treated with fludarabine-TBI.

Eight patients experienced persistence or re-appearance of measurable residual disease (MRD) detected by flow cytometry or cytogenetic abnormalities with <5% blasts. They received azacitidine as per our institutional practice in an effort to delay or prevent morphologic relapse. Among this subgroup, three patients progressed to morphologic relapse within 2 months, one additional patient progressed at 16 months, and four achieved and maintained MRD-negative CR through the end of study follow-up. One other patient who received 6 cycles of prophylactic azacitidine after bone marrow aspirates at days 28, 56, and 83 post-transplant confirmed MRD-negative CR.

Overall survival was 55% (95% CI, 40%–70%) among all patients and 59% (95% CI, 42%–75%) among high risk patients treated at the MTD at 2 years. Leukemia-free survival at 2 years was 52% (95% CI, 37%–67%) among all patients and 56% (95% CI, 39%–72%) among high-risk patients (Figure 1).

3.6. Comparison to historical controls

Table 2 compares study patients treated with the clofarabine 250 mg/m²-TBI regimen (excluding the one patient whose transplant was aborted) to those transplanted after receiving fludarabine 90 mg/m²-TBI. Patients treated with clofarabine were older and had higher comorbidity scores (HCT-CI¹⁵), higher prevalence of unfavorable cytogenetics, and were more likely to receive unrelated donor grafts. Patients treated with fludarabine were more often beyond first remission, more likely to have received an HLA-mismatched graft, and less likely to have recovered their blood counts before transplant. On balance, the groups seem to have a similar overall disease and toxicity risk profile.

TABLE 2.

Comparison of protocol and historic high-risk patients

	Clofarabine-TBI (n=35)	Fludarabine-TBI (n=140)	P
Age, years, median (range)	69 (53–74)	60 (5–74)	<0.0001
Male sex, n (%)	22 (63)	82 (59)	0.64
Prior HCT, n (%)
None	32 (91)	116 (83)
Allogeneic	1 (3)	6 (4)
Autologous	2 (6)	18 (13)	0.44
HCT comorbidity index,15 n (%)
0–1	3 (9)	50 (36)
2–3	14 (40)	51 (36)
≥4	18 (51)	39 (28)	0.003
Performance status*, n (%)
90–100	19 (54)	48 (54)
80–85	13 (37)	25 (28)
60–70	3 (9)	16 (18)	0.35
Donor type, n (%)
Matched related	7 (20)	53 (38)
Matched unrelated	27 (77)	70 (50)
Mismatched	1 (3)	17 (12)	0.01
CD34+ cell dose, 106/kg, median (range)	8.2 (3.8–23.6)	6.9 (1.5–29.9)	0.12
Therapy-related AML, n (%)	7 (20)	17 (12)	0.23
Evolved from prior MDS/MPN, n (%)	7 (20)	42 (30)	0.24
AML stage, n (%)
CR 1	29 (83)	80 (57)
CR 2	6 (17)	27 (38)
CR 3	0	22 (16)	0.008
MRD positivity pre-HCT, n (%)	14 (40)	37 (26)	0.13
SWOG unfavorable, n (%)	30 (86)	74 (53)	0.0004
Failure to recover counts pre-HCT, n (%)	16 (46)	87 (62)	0.08
≥ 18 months from diagnosis to HCT, n (%)	1 (3)	20 (14)	0.06
AML risk score13
0.5	4 (11)	44 (31)
1.0	17 (49)	44 (31)
1.5	11 (31)	35 (25)
≥2.0	3 (9)	17 (12)	0.07

^*Lansky if < 16, Karnofsky if ≥ 16.

Abbreviations not defined in the text: HCT-CI – hematopoietic cell transplantation comorbidity index; MDS/MPN – myelodysplastic syndrome/myeloproliferative neoplasm.

Figure 1 shows that clofarabine-TBI conditioning appears to overcome clinical indicators of high relapse risk, resulting in survival outcomes comparable to historical low-risk patients treated with fludarabine-TBI.

Clofarabine-TBI substantially reduced the risk of mortality, disease progression, and morphologic relapse compared to fludarabine-TBI (Table 3). Even in multivariable analysis accounting for patient age, HCT-CI score, HLA mismatch, pre-transplant MRD status, and relapse risk score, the beneficial effects of clofarabine-TBI remained substantial, with a HR of 0.50 (95% CI, 0.28–0.91) for OS, 0.48 (95% CI, 0.27–0.85) for PFS, and 0.30 (95% CI, 0.13–0.69) for relapse. This benefit was observed without any detrimental effect on NRM (HR 0.85, 95% CI 0.36–2.00).

TABLE 3.

Hazard ratio analysis of clofarabine-TBI vs. fludarabine-TBI conditioning for high-risk AML

	Unadjusted		Adjusted*
	HR (95% CI)	P	HR (95% CI)	P
OS	0.59 (0.35–1.01)	0.05	0.50 (0.28–0.91)	0.02
PFS	0.55 (0.33–0.93)	0.02	0.48 (0.27–0.85)	0.01
Relapse	0.34 (0.16–0.73)	0.006	0.30 (0.13–0.69)	0.004
NRM	1.02 (0.50–2.09)	0.95	0.85 (0.36–2.00)	0.70

^*Adjusted for age (<60, ≥60), HCT-CI (0–1, 2–3, ≥4, missing), HLA mismatch (no, yes), relapse risk score (0.5, 1, 1.5, ≥2), MRD pre-transplant (no, yes)

3.7. Pharmacokinetics

After 40 mg/m² (n = 2) or 50 mg/m² (n = 3) of IV clofarabine, the AUC ranged from 2509–3719 mg×h/L with 17%–24.5% of the AUC extrapolated from the last concentration-time point to infinity. These results are similar to that reported by Büttner et al., which had a median (range) AUC of 2886 (1342) after clofarabine 40 mg/m.¹⁶ Clofarabine clearance ranged from 0.29–0.53 L/h/kg of actual body weight (supplementary Table S2), which is lower than previously reported in pediatric HCT recipients (0.25–14.9 years).¹⁷ Age has been associated with clofarabine pharmacokinetics. No consistent trend was observed in trough (pre-dose) clofarabine samples in the limited dataset (n=4).

4. DISCUSSION

This prospective, phase I/II trial is the first to report the safety and efficacy of clofarabine in combination with low-dose TBI as non-myeloablative HCT conditioning for patients with AML in morphologic remission. The regimen was well-tolerated despite the high level of comorbidities and advanced age of the patients. Moreover, this trial enrolled nearly exclusively patients at high risk for relapse and achieved the primary endpoint of reducing the 6-month relapse rate to ≤20%. The relapse rates observed in patients with high-risk AML treated with clofarabine-TBI were comparable to those observed after fludarabine-TBI among patients at low risk for relapse. This salutary effect on relapse rates did not appear to come at the cost of NRM or GvHD when compared to historical rates with dual CSP/MMF immunosuppression.⁵ However, these assertions are all limited by the use of a historical control.

All evaluable patients achieved full donor engraftment and hematologic recovery in an acceptable timeframe. This compares well to the fludarabine-2 Gy TBI regimen, where 4% of patients transplanted for AML experienced graft rejection.5

This trial employed CSP/MMF immunosuppression. In a separate multicenter, phase III randomized trial using non-myeloablative fludarabine-TBI, we demonstrated that the addition of sirolimus to standard CSP/MMF substantially lowers the incidence of GvHD-related death.¹⁸ Using the triple regimen of CSP/MMF/sirolimus may further improve the NRM we observed with clofarabine-TBI, since 8 of 15 non-relapse deaths were GvHD-related.

We sought to evaluate clofarabine pharmacokinetics, but few patients agreed to stay in the clinic for outpatient pharmacokinetic sampling. Within fludarabine-TBI-conditioned patients, we demonstrated that patients do comply with pharmacokinetic sampling after fludarabine¹⁹ or MMF²⁰ dosing if a population pharmacokinetic-based limited sampling schedule is used. Unfortunately, at the time of study conduct, we did not have a clofarabine population PK model-based limited sampling schedule to improve subject compliance. Understanding the association of clofarabine plasma AUC or maximum plasma concentration are of interest. In relapsed or refractory AML patients treated with clofarabine monotherapy (i.e., without HCT), a higher clofarabine AUC (but not Cmax) was associated with higher aspartate aminotransferase (AST) concentrations after adjusting for age, sex, cumulated clofarabine dose, baseline AST, and glomerular filtration rate (GFR). Clofarabine AUC and Cmax were not associated with skin toxicity and the response to re-induction chemotherapy.¹⁶

Two of 36 patients treated at the 250 mg/m² clofarabine dose suffered probable conditioning-related mortality. Early hepatic and renal toxicity attributed to clofarabine-based conditioning has been observed by others,^21,22 but real-time PK-guided dose adjustment is not currently standard-of-care. Our trial included only patients with normal creatinine clearance, and nonetheless, stochastic acute kidney injury did occur. Similarly, hepatotoxicity was not easily foreseen: The patient who developed grade 2 transaminitis had a pre-conditioning AST slightly higher than twice the upper limit of normal and an alkaline phosphatase level 1.3 times the upper limit. The patient who developed liver failure had a history of substantial alcohol consumption, but he had normal liver function tests and no signs of liver fibrosis or cirrhosis were found during the pre-transplant evaluation or at autopsy.

Apart from the non-negligible incidence of early toxicity, we found clofarabine-TBI to be well-tolerated, as did a recent multicenter phase I trial of clofarabine-2 Gy TBI as conditioning for pediatric patients (median 9 years old, range 1–21) with acute lymphoblastic leukemia (n = 11) or AML (n = 7).²³ Other studies^24–30 found that clofarabine in cumulative doses ranging from 120 to 300 mg/m² could be incorporated into alkylator-based HCT conditioning regimens for a variety of diseases without jeopardizing engraftment or incurring excessive treatment-related mortality; the median age of patients in these studies was substantially lower than those included here (34–62 years). Among 24 older patients with AML (median age 59–60 years), Chevallier et al. retrospectively observed similar NRM (12.5% vs. 20.8%, P = .58) but lower 2-year relapse incidence (16.7% vs 41.2%, P = .05) and better 2-year LFS (70.8% vs. 38%, P = .03) and OS (79.2% vs. 38%, P = .01) with clofarabine-2-day busulfan HCT conditioning compared to 248 who received fludarabine-2-day busulfan.³¹ Other trials found that clofarabine as part of bridging chemotherapy or combined with myeloablative busulfan or melphalan conditioning could salvage a proportion of patients with primary induction failure and those with chemosensitive relapse.^32–34 Median ages of patients those trials were 47–61 years. Trials combining clofarabine with high-dose TBI have been restricted to children and young adults.³⁵ Because of its immunosuppressive effects, clofarabine has also been studied in haploidentical pediatric HCT^21,36 and in adult haplotransplant using a modified Baltimore regimen with 2 Gy TBI and post-transplant cyclophosphamide.^37,38

We do not attribute the low hematologic relapse rate and the 2-year overall survival observed in this trial to the pre-emptive use of azacitidine because azacitidine generally delays overt relapse but does not avoid it. In a prospective phase II trial of preemptive azacitidine at our Center (median bone marrow leukemic blasts 1.38%), only 3 of the 39 patients attained a complete remission and 2-year OS was only 25%.³⁹ Others found that pre-emptive azacitidine could delay hematologic relapse until a median of 231 days after detecting initial evidence of residual disease, but hematologic relapse nonetheless occurred in 65% of post-transplant patients.⁴⁰ A follow-up study that included non-transplanted and HCT patients treated with pre-emptive azacitidine reported a cumulative incidence of relapse of 49% a median of 422 days after initial MRD detection.⁴¹ Similarly, after initial intensive chemotherapy for older patients with AML or MDS, maintenance azacitidine improved 12-month disease-free survival but not overall survival.⁴²

In conclusion, this phase I/II trial demonstrates the tolerability and efficacy of non-myeloablative clofarabine-TBI conditioning in older adults at high risk for post-HCT relapse. To reduce relapse rates and improve overall survival, it is important to consider the particular conditioning and immunosuppressive regimen in addition to pre-HCT comorbidities.⁴³ For example, regimens like fludarabine-treosulfan-2 Gy TBI seem promising for older adults with AML and MDS.⁴⁴ The different outcomes across trials devoted to patients unsuitable for myeloablative conditioning highlight the need to develop conditioning regimens that increase anti-leukemic effects while maintaining tolerability. For those with high-risk AML, a randomized comparison of clofarabine-TBI to other reduced-intensity regimens, such as fludarabine-treosulfan-TBI or fludarabine-melphalan, is warranted. To further reduce relapse rates, studies that aim to modulate the post-grafting immunologic landscape, e.g., with prophylactic or pre-emptive cellular therapies, and studies of maintenance therapy such as tyrosine kinase inhibitors are also welcome. Clofarabine 250 mg/m² with 2 Gy TBI, as tested here, is a useful platform for such approaches.