Running title: Busulfan and Subsequent Malignancy

Text word count: 2479

Abstract word count: 249

Tables: 5

Figures: 1

Abstract:

The incidence of secondary malignancies associated with busulfan exposure is considered low, but has been poorly characterized. Because this alkylating agent is increasingly utilized as conditioning prior to gene therapy in non-malignant hematologic and related disorders, more precise characterization of busulfan’s potential contribution to subsequent malignant risk is warranted. We conducted a literature-based assessment of busulfan and subsequent late effects, with emphasis on secondary malignancies, identifying publications via PubMed searches and selecting those reporting at least 3 years of follow-up. We identified 8 pediatric and 13 adult publications describing long-term follow-up in 570 pediatric and 2,076 adult hematopoietic cell transplant (HCT) recipients. Secondary malignancies were reported in 0.5% of pediatric HCT recipients, with no cases of myelodysplastic syndrome (MDS) or acute myelocytic leukemia (AML). Fatal secondary malignancies were reported in 0.8% of 1887 evaluable adult HCT recipients, and an overall incidence of secondary malignancies of 4.8% was reported in a subset of 389 evaluable adult patients. We also reviewed long-term results from 8 publications evaluating lentiviral- and human promotor-based HSC-targeted gene therapy in 215 patients with non-malignant conditions, in which busulfan/treosulfan monotherapy or busulfan/fludarabine was the only conditioning. Two malignancies were reported in patients with Sickle Cell Disease (SCD), one of which was potentially busulfan-related. No additional malignancies were reported in 173 patients with follow-up of 5-12 years. The incidence of busulfan-related secondary malignancies is low, and likely to be substantially less than 1% in pediatric transplant recipients, especially those receiving busulfan monotherapy for non-malignant conditions other than SCD.

Introduction:

The alkylating agent busulfan has been a core component of conditioning regimens enabling hematopoietic cell transplantation (HCT) prior and subsequent to its approval in 1999-2002. Busulfan has been particularly essential in autologous lentiviral mediated gene therapy for non-malignant genetic disorders, both for investigational programs and approved therapies. Busulfan-related short-term toxicities are well-characterized and include mucositis, seizures and hepatic veno-occlusive disease. Longer term effects include impaired fertility, organ damage including bronchopulmonary dysplasia, and cancer predisposition. The incidence of some indolent effects, particularly regarding fertility in pediatric recipients, has not been precisely characterized, and is confounded because of the co-administration of additional cytotoxic therapies in the majority of HCT settings. For these same reasons, the risk of busulfan-associated secondary malignancy has also been incompletely delineated. The recent identification of non-insertional myeloid malignancy in a 45 year old patient with Sickle Cell Disease (SCD) who had received autologous HSC-directed gene therapy three years previously resulted in additional discussion of busulfan as a potentially carcinogenic and leukemogenic agent. As the number of LV-mediated gene therapy investigational programs and approved therapies continue to expand, often with compelling efficacy, we believed it was imperative to enable a more optimal characterization of malignancy potential in patients receiving this agent, in order to optimally inform patients and clinicians considering gene therapy options for non-malignant hematologic and related disorders. This is of particular importance in pediatric settings. We conducted a literature-based evaluation regarding the incidence and nature of busulfan-related and other late effect post-HCT malignancies.

Methods:

A systematic literature search (SLR) was based on the PICOS strategy (Patient, Intervention, Comparator, Outcome, and Study design) outlined in Table 1.

EMBASE and MEDLINE (via Proquest Dialog) were searched for the terms busulfan (including variants), transplantation, conditioning, and hematopoietic stem cell transplantation (detailed search strategy is described in supplementary methods) encompassing the publications, in English language from 2012 through September 2022.

Literature was extracted to a data extraction template (DET) and duplicates were removed. The entire list was then screened by title/abstract to identify articles for inclusion/exclusion in the SLR. Exclusion criteria consisted of conference abstracts, reviews without patient outcomes, and out of scope publications (conditioning regimens without Busulfan comprising at least one of the components).

The search results, specific criteria employed, and publications selected for detailed review are depicted in Figure 1 , culminating in cohorts of publications describing results in pediatric or young adult transplant studies with at least 5 years of median follow-up, adult transplant studies with at least 5 years of median follow-up, and adult transplant studies with at least 3 years (but less than 5 years) median follow-up.

Results:

From the 455 records identified and extracted, only 4 papers were excluded at the abstract level and 396 when assessing full text. Because alkylating agent-mediated myeloid malignancies predominantly arise 4-7 years following exposure,^, we placed particular emphasis on studies providing long-term results (median follow-up of at least 3 years), in both pediatric (or mixed pediatric/adult) or adult settings. We also evaluated results, with an emphasis on secondary malignancies, from publications detailing autologous LV-based gene therapy clinical trials utilizing self-inactivating LVs incorporating human promotors in hematologic and related disorders, also with emphasis on studies for which at least 5 years of median follow-up is available. Excluded from this evaluation are long-term results from LV-based gene therapy studies in cerebral adrenoleukodystrophy (CALD), a program using a vector containing a viral promotor associated with in vitromoderate transactivating capability, and the only LV-based program to-date associated with insertional mutagenesis.

49 papers were then comprehensively assessed based on the previous inclusion criteria and 25 were excluded for: a) Non-hematologic pediatric condition; b) Case-controlled risk-analysis; c) Pediatric study <3y follow-up; d) Adult study < 3y follow-up and e) Malignancy incidence not evaluable. A list of 24 records were included for comprehensive review, categorized as follows: Pediatric/young adult study ≥3y follow-up, n=8 (≥ 5y: n=8); and Adult study ≥ 3y follow-up, n=16 (≥ 5y: n=8; ≥3y and <5y: n=8).

Eight allogeneic HCT publications were identified describing long-term outcomes in 642 predominantly pediatric patients, including median follow-up of at least 5 years in 602 patients, as detailed inTable 2. No secondary malignancies were reported in 6 of these studies (involving n=504 patients). In one study, 3 of 66 (4.5%) pediatric leukemia patients receiving busulfan-based multi-agent conditioning regimens developed secondary malignancies, specifically thyroid (n=1), basal cell skin cancer (n=1) and meningioma (n=1), in contrast to 20 of 174 (11.5%) of patients receiving TBI-based regimens without busulfan. In one additional study, a single patient developed a secondary melanoma but it was not specified whether this patient had received busulfan-based (n=72 patients) or TBI-based (n=77) conditioning. Thus, for 570 evaluable pediatric patients receiving busulfan-based conditioning, secondary malignancies were reported in 3 patients, representing a 0.5% incidence. There were no reported cases of AML/MDS in the pediatric population. These findings are in contrast to secondary malignancies occurring in 131 of 2721 (4.8%) of pediatric patients receiving TBI-based regimens in the absence of busulfan.

Of note, in one study evaluating long-term effects in very young pediatric patients (age <2y; 82% with non-malignant conditions) receiving multi-agent busulfan-based conditioning between 1993 and 2008 (predominantly busulfan, cyclophosphamide and ATG), late effects included dental complications (92%), short stature (56%), cognitive deficits (54%) and pubertal abnormalities (28%). In one study evaluating HCTs in 99 pediatric β-thalassemia patients receiving transplant between 1985-2012, detailed fertility effects were reported as follows: 12 of 33 (33%) pre-pubertal females experienced normal puberty with younger age at HCT (median 2.5 years) relative to those who had delayed puberty (median age 8.7 years). Pre-pubertal males experienced subsequent normal puberty in 18 of 22 (82%) of instances. Eleven of 27 (40%) of evaluable females had at least one successful pregnancy; and 4 of 21 (19%) of evaluable males fathered at least one child. Information regarding percentages within groups actively seeking pregnancy/fatherhood was not provided. A majority of these patients received either oral busulfan or intravenous therapy prior to the implementation of pharmacokinetically guided therapy.

Thirteen publications in adult hematologic malignancies were identified describing long-term results for busulfan/chemotherapy combination pretransplant conditioning regimens, encompassing 2,076 patients. Three additional publications described long-term results in adults (also with hematologic malignancies) when busulfan was combined with TBI and additional chemotherapy agents, involving 744 patients. Five of the busulfan/chemotherapy publications and each of the three busulfan/TBI/chemotherapy publications involved median follow-up exceeding 5 years, summarized in Table 3 ; the remaining 8 selected studies involved median follow-up of between 3 and 5 years, summarized in Table 4 .

For 11 of the 13 studies evaluating busulfan/chemotherapy conditioning regimens, long-term outcomes were reported as fatal secondary malignancies, involving 1,887 patients. (It is presumed that these malignancies, resulting in patient deaths, are identified in contrast to localized or less aggressive secondary neoplasms such as non-invasive skin cancers, which were reported in >5% of patients in some adult series). Incidences of fatal secondary malignancies ranged from 0 to 5.9% across studies (median incidence 0.7%). Overall there were 16 fatal secondary malignancies reported subsequent to HCT in 1,887 patients, representing a 0.8% incidence. Of note, of the four studies with the highest incidence of fatal secondary malignancies, 3 of these 4 employed reduced intensity (RIC) conditioning, as did 2 of the 4 studies with the lowest incidences.

The overall incidences of secondary malignancies were reported in four of the adult studies, involving 389 patients. Nineteen of these 389 (4.8%) patients encountered secondary malignancies, with the incidence ranging between 0 and 10.7%. Importantly, 16 of the 19 secondary malignancies were reported from a single study in high-risk hematologic malignancies in an older adult population (median age 61 years, 33% above 65 years at time of HCT). Malignancies in this study included one of the previously mentioned fatal cancers, and 15 additional secondary malignancies (7 of which were dermatologic); 14 of these 15 secondary malignancies occurred in patients who had experienced either acute (n=6) or chronic (n=8) GVHD.

The highest incidences of secondary malignancies occurred in the three series in which busulfan and additional chemotherapy agents were administered in conjunction with TBI, as conditioning for hematologic malignancy-directed HCTs in adults. Secondary malignancy incidences in these three studies, involving 744 patients, were 6%, 7.5% and 8% respectively. Specific malignancies were described in two of these publications and were non-invasive skin cancers, occurring in a total of three patients.

The potential contribution of busulfan to secondary malignancies is potentially best delineated by the experience in autologous hematopoietic stem cell directed gene therapy studies, several of which utilize busulfan monotherapy or combination exclusively with fludarabine. In these studies, busulfan dosage ranged from reduced intensity (RIC; as low as AUC 10-19 mg*h/L) to myeloablative (70-85 mg*h/L). Long-term follow-up results are available for several of these trials, with follow-up durations of 5-12 years available across 8 trials encompassing 173 patients, detailed in Table 5. Also described in Table 5 are results from 42 patients enrolled in SCD studies for whom follow-up of at least 3 years is available. Notably, the only malignancies occurring in these studies were in the SCD studies, in two patients receiving therapy as part of the initial Group A process, involving bone marrow harvest, lower CD34+ cell doses, and more limited hemoglobin response relative to the more recently treated 35 patients. Both myeloid malignancies harbored mutations in oncogenes RUNX1 and PTPN11, and monosomy 7 cytogenetic abnormalities. In one case, the malignant cell population was comprised of gene-corrected cells, with an integration in proximity to VAMP4 , a loci not consistent with mutagenic potential. In the second case, no gene markings were present in the malignant population, and the AML was considered potentially secondary to busulfan in the context of underlying SCD. Importantly, for the entirety of LV-based gene therapy studies utilizing vectors with human promotors, no malignancies have been reported secondary to insertional mutagenesis. A recent update indicated no clonal hematopoiesis or associated detectable mutations in 10 ADA-SCID gene therapy recipients who received RIC busulfan with 7-10 years of follow-up.

Discussion:

In summary, the incidence of busulfan-associated secondary malignancies is low, particularly in pediatric settings, with a majority of long-term evaluations indicating an incidence less than 1%, and an overall aggregate incidence across more than 500 patients of 0.5%. Secondary MDS/AML in the setting of pediatric busulfan exposure was not reported and is likely to be rare. The incidence of post-busulfan secondary malignancies was higher in adult transplant studies, but also relatively limited. The incidence of fatal secondary malignancies is likely lower than 1%, and the overall incidence of secondary malignancies is likely lower than 5%, including a substantial proportion of non-invasive dermatologic malignancies. A substantial proportion of the secondary malignancies identified were identified in studies with extensive participation of older adults. It is likely that older age confers a higher susceptibility to conditioning-related malignancies, consistent with this association in more general populations. Results in adult populations are unlikely to precisely inform potential malignant risks for pediatric patients.

Notably, in the adult transplant series, secondary malignancies were reported both in cohorts receiving myeloablative and RIC regimens. It is likely that although RIC regimens involve an overall reduced cytotoxic burden, this effect is potentially countered by survival of hematopoietic progenitors which have been exposed to alkylator and other therapies with potential for mutagenic DNA damage. RIC regimens may also be more likely utilized in older adult cohorts, who are at higher baseline cancer risk.

A limitation of this study is that the overwhelming proportion of both allogeneic and autologous HCTs were performed in settings of hematologic malignancies, in which multi-agent conditioning regimens were likely to have been preceded by induction and additional chemotherapies, such that patients’ overall exposure to cytotoxic therapy was extensive. This limits the extent to which one may ascribe the specific contribution of busulfan to secondary malignancies or other long-term complications. These results also pre-date the implementation of therapeutic drug monitoring (TDM) to enable more precise busulfan dosing; because the pharmacokinetics of oral and intravenous busulfan are highly variable, there was higher potential for overexposure in these earlier studies than in current programs. An additional limitation is that very few publications provided detailed results regarding other long-term effects of interest – including fertility and neurocognitive and overall development in pediatric transplant recipients. These remain substantial risks for children receiving high-dose combination chemotherapy; as with secondary malignancies, the precise risks conferred by busulfan cannot be determined because of the multi-agent nature of the regimens evaluated. Long-term results from an EBMT dataset are likely forthcoming and will hopefully provide more precise information regarding fertility impairment in adult survivors of pediatric transplant.

The most busulfan-specific results are likely to emanate from autologous HSC-directed gene therapy studies, including several in which busulfan was utilized exclusively as a conditioning agent (at exposures ranging from AUC 10-85 mg*h/L). As indicated, long-term results from the earliest LV-mediated studies are now available, with 5-10 years of follow-up for more than 150 patients. In these settings, administration of busulfan conditioning therapy is less confounded by concomitant conditioning agents or prior anti-cancer therapies. The only potential busulfan-related malignancy (AML) developed in a patient with SCD, a condition increasingly recognized as predisposing to myeloid malignancy, albeit with uncertainty regarding the specific contributing factors or overall increase in risk. Follow-up of more than 5 years in patients receiving busulfan-based conditioning for genetic disorders other than SCD indicates that the risk of secondary malignancy is likely to be very rare.