Germline Pathogenic Variants in <scp> <i>MUTYH</i> </scp> Are Associated With Inferior Survival After Hematopoietic Cell Transplantation in Patients With Hematologic Malignancies or Disorders

Hematopoietic cell transplantation (HCT) is a curative option for many hematologic diseases but remains associated with high mortality and morbidity risks. Recent efforts have focused on identifying genomic biomarkers beyond the well-established HLA loci to improve patient outcomes [1, 2]. DNA damage response and repair (DDR) genes are of particular interest due to recipients' exposure to DNA damaging agents like ionizing radiation and alkylating agents. While individuals with biallelic DDR variants (e.g., Fanconi anemia [3], ataxia telangiectasia [4, 5]) are known to require modified regimens to mitigate toxicity, the impact of monoallelic germline DDR variants on HCT outcomes remains unknown. In this study, we evaluated the associations between germline DDR variants and post-HCT outcomes and explored whether these associations were affected by the type of conditioning regimen. We used exome sequencing data we generated for 2008 allogeneic HCT recipients with severe aplastic anemia (SAA; N = 683) [6], primary or secondary myelofibrosis (MF; N = 930) [7], or myelodysplastic syndrome (MDS; N = 395; including 79 therapy-related MDS patients) [8]. Pre-HCT blood samples and clinical data were provided by the Center for International Blood and Marrow Transplant Research (CIBMTR). Patients with clinically or genetically diagnosed inherited bone marrow failure syndromes (N = 146) were excluded [6, 8]. We curated germline variants (with variant allele fraction; VAF 40%–60%) in 129 DDR genes involved in seven repair pathways (Table S1), following the American College of Medical Genetics and Genomics/Association of Molecular Pathology (ACMG/AMP) guideline [9]. Only tier-1 pathogenic/likely pathogenic (PLP) variants were included, defined as loss-of-function (LOF) variants (frameshift, nonsense, and splicing) in genes where LOF is a recognized mechanism, or variants previously reported as PLP in ClinVar (Table S2). The primary outcome was overall survival (OS) post-HCT. Survival probabilities were estimated using Kaplan–Meier methods with log-rank tests for group comparisons. Multivariable Cox proportional hazards models were used to assess associations between PLP DDR variants and OS, adjusting for clinical covariates (age, sex, conditioning intensity, donor type, year of transplant, and Karnofsky Performance Status) and stratified on graft source and disease type. Competing risks models were used to evaluate transplant-related mortality (TRM) and disease relapse. The survival analysis included only genes with PLP variants affecting ≥ 10 patients. Details are available in online Supporting Information. The patient median age at HCT was 20.2 years for SAA, 58.4 years for MF, and 60 years for MDS. Fludarabine-Busulfan (Flu-Bu)-based regimen was frequent in MDS (50.6%) and MF (44.2%), while total body irradiation (TBI) was used far more often in SAA (58%) than in MF (12%) or MDS (7%). Reduced intensity or non-myeloablative conditioning (RIC/NMA) regimens were most common in SAA patients (61%) (Table 1). We identified 274 tier-1 PLP variants in 86 DDR genes (Table S2). All were monoallelic, except for two homozygous variants in PMS1 or CHEK2 found in two patients, both died in the first 30 days post-HCT. LOF variants included 105 frameshift, 80 nonsense, and 47 splicing variants. PLP DDR variants were detected in 20.3% of SAA, 23.9% of MF, and 15.4% of MDS patients. Among 423 positive patients, 9.7% had variants in ≥ 2 DDR genes. PLP variants occurred most frequently in CHEK2 (N = 65, 3.2%), enriched in MF patients (5.5% vs. 1.3% in SAA or MDS, p < 0.0001), followed by MUTYH (N = 40, 2%), POLQ (N = 34, 1.7%), and POLG in (N = 24, 1.2%), with comparable frequencies across disease cohorts (Figure S1). MUTYH and BRCA2 were the only DDR genes statistically significantly associated with inferior survival (5-year OS = 39% vs. 58% in MUTYH-positive vs. negative, p = 0.003; and 23% vs. 58% in BRCA2-positive vs. negative, p = 0.03) (Figure 1 & Table S3). Multivariable analysis confirmed the MUTYH association (HR = 2.0, 95% CI = 1.3–2.9, p = 0.0007), but not for BRCA2 (HR = 1.7, 95% CI = 0.9–3.2, p = 0.10). Patient and transplant-related characteristics by MUTYH status are summarized in Table S4. In a disease-stratified analysis comparing MUTYH-positive and MUTYH-negative patients, the risk was highest among patients with SAA (HR = 2.7, 95% CI = 1.4–5.3, p = 0.004), followed by those with MF (HR = 1.8, 95% CI = 1.0–3.3, p = 0.04), then MDS patients, although not statistically significance (HR = 1.5, 95% CI = 0.6–3.7, p = 0.39). Surprisingly, multivariable analysis stratified by conditioning intensity, showed that MUTYH-positive compared to negative patients had a higher mortality with RIC/NMA (HR = 2.8; p = 0.0004) but not with myeloablative conditioning (MAC) (HR = 1.6; p = 0.09; p-heterogeneity = 0.06) (Table S5). Analyses stratified by the type of conditioning regimen, and adjusted for regimen intensity and other clinical factors, showed significantly worse survival in MUTYH-positive vs. negative patients among those who received TBI (HR = 2.42, 95% CI = 1.17–5.0; p = 0.017), while no survival difference was observed in MUTYH-positive vs. negative patients among those who received Flu-Bu regimen (HR = 1.18, 95% CI = 0.58–2.42, p = 0.64; Table 2). In a secondary analysis, worse OS was noted in patients with PLP variants in EXO1 (HR = 5.97, p = 0.01) among those receiving the Flu-Bu regimen, and in FANCD2 (HR = 5.76, p = 0.007) or PNKP (HR = 4.26, p = 0.02) in the TBI regimen category (Table S6). Multivariable cause-specific analysis showed a statistically significant increased risk of TRM among MUTYH-positive compared to MUTYH-negative patients (HR = 2.25, 95% CI = 1.48–3.43, p = 0.0001). This was primarily driven by a high risk of death from organ failure (HR comparing MUTYH-positive vs. negative =3.44, 95% CI = 1.38–8.56, p = 0.008). The frequency of causes of death in patients with and without MUTYH with calculated hazard ratio for each cause are available in Table S7. MUTYH is a critical component of the base excision repair (BER) pathway, responsible for repairing DNA damage from environmental sources such as ionizing radiation and alkylating agents [10]. Pathogenic monoallelic MUTYH variants occur in 1%–2% of populations in Northern Europe, Australia, and the United States [11, 12], potentially affecting ~1500 of the ~84 000 annual HCT recipients globally [13]. Two founder mutations, c.536A>G (p.Y179C) and c.1187G>A (p.G396D), together account for over 80% of MUTYH mutations in individuals of White European ancestry, while other variants are more common in different populations [12]. In this study, 2% of patients carried monoallelic PLP variants in MUTYH, with 80% harboring the common founder mutations and similar frequencies across disease cohorts. Additionally, MUTYH variants are expected in donors, raising questions about whether donor carrier status affects HCT outcomes and if the effect is stronger when both recipient and donor are carriers, which warrants further studies. HCT conditioning involves high-dose alkylating agents or radiation to deplete host hematopoietic stem cells, causing substantial DNA damage that requires intact DDR pathways for repair [14]. Consistent with this, we showed patients with PLP MUTYH variants exhibited increased post-HCT mortality, mainly from organ toxicity/failure. Previous studies have not directly linked rare MUTYH variants to HCT outcomes; one SNP-based study of 27 BER genes suggested associations for noncoding variants in MUTYH, LIG3, and OGG1 [15], but these were not confirmed in larger cohorts [2, 16]. In the current study, potential associations with LIG3 or OGG1 could not be assessed due to very few carriers. Our data suggest that the inconsistency in the literature related to MUTYH is possibly due to differences in used conditioning regimen. We observed the strongest association in SAA patients (HR = 2.7, p = 0.004) who frequently received TBI or cyclophosphamide, whereas the lowest risk was in MDS patients (HR = 1.5, p = 0.39) who mostly received Flu-Bu. Similarly, the differences in conditioning regimens likely accounted for the unexpected excess risk of mortality with MUTYH for RIC/NMA (HR = 2.8, 95% CI = 1.6–4.8, p = 0.0004) where high frequency of TBI-based and Flu-Mel regimens (62%) was observed (Table S8). Compared to MUTYH-negative patients, Flu-Bu regimens did not increase mortality in MUTYH-positive patients, whereas Flu-Mel showed a trend toward higher risk. The increased risk associated with TBI or Flu-Mel is likely attributable to radiation or melphalan inducing oxidative DNA damage that depends on MUTYH-mediated BER activity to repair [17, 18]. In contrast, busulfan induces DNA crosslinks [19] repaired by alternative DDR pathways (e.g., homologous recombination and nonhomologous end joining), less dependent on MUTYH function. Of note, busulfan can indirectly increase oxidative stress by depleting glutathione—the key cellular antioxidant—which may exacerbate the toxicity of other alkylating agents like cyclophosphamide [19]. Cyclophosphamide also showed a non-significant trend toward higher mortality in MUTYH carriers. The study's strengths include a large cohort, the use of exome sequencing that enabled evaluation of a comprehensive set of DDR genes, and high-quality real-world clinical data. Additionally, by focusing on tier-1 PLP variants, mostly LOF variants, the study captured associations often missed by SNP-based studies. Limitations include restriction to single nucleotide variants without evaluating copy number or structural variants, and lack of standardized reporting of transplant related causes of death, which may be misclassified in approximately 20% of cases [20]. Used blood samples could potentially include somatic variants, but this risk is minimal because the analyzed DDR genes are not commonly affected by clonal hematopoiesis, and only variants with germline-level VAFs were considered. The study was limited by the small sample size for some genes with possible impact on HCT survival; this warrants investigation in a larger study. Finally, the study lacked information on HCT subsequent cancers which can be impacted by MUTYH or other DDR genes. In conclusion, we found that carriers of MUTYH PLP variants have inferior OS and may benefit from TBI-free regimens. Flu-Bu regimens may provide an alternative with similar OS in patients with or without MUTYH variants. Our findings highlight the importance of genetic testing to detect clinically relevant variants so that the transplantation approach can be tailored to each patient [6, 8, 21]. Study concept and design: M.R. and S.M.G.; Acquisition of biological samples and clinical data: W.S., M.H., S.R.S., R.A., S.J.L., Y-.T.B.; Genomic analysis and bioinformatics: M.R., K.J., B.H., A.L.K.; Statistical analysis: M.R., Y.W., S.M.G.; Data interpretation: All authors; Manuscript drafting: M.R. and S.M.G.; Critical revision: All authors. The study was approved by the National Marrow Donor Program Institutional Review Board. All patients or their guardians provided written informed consent for the research use of their samples and clinical data. The authors declare no conflicts of interest. Exome sequencing data used in this study are available in dbGaP under accession numbers phs001710.v1.p1. for the SAA cohort, and phs002635.v1.p1 for the myelofibrosis cohort. The dataset used for this analysis will be available at https://cibmtr.org/CIBMTR/Resources/Publicly-Available-Datasets. Data S1: ajh70093-sup-0001-Supinfo.zip. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.