Abstract
In pediatric hematopoietic cell transplantation (HCT) recipients, transplanted donor cells may need to function far beyond normal human lifespan. In this study, we investigated the risk of clonal hematopoiesis (CH) in 144 pediatric long-term HCT survivors and 258 nontransplanted controls. CH was detected in 16% of HCT recipients and 8% of controls at variant allele frequencies of 0.01 to 0.31. Mutations were predominantly in DNMT3A (80%) and TET2 (20%). Older hematopoietic age (OR: 1.07; P < 0.001) and the HCT procedure (OR: 2.53; P = 0.02) independently increased the risk of CH, indicating both aging- and transplantation-induced effects. Large clones (variant allele frequency >0.10) were found exclusively in HCT recipients. Notably, CH was also detected within 15 years after a cord blood HCT. Inflammatory processes around graft infusion were associated with CH presence. Future studies are required to track the evolution of posttransplant CH and its impact on future cardiovascular diseases, second malignancies, and overall survival.
As survival of HCT recipients continues to improve, late treatment effects gain importance. We demonstrate that pediatric HCT recipients show increased risk of CH compared with age-matched controls. Prospective studies are crucial to understand the clinical implications of posttransplant CH in this young population.
Introduction
Clonal hematopoiesis (CH) refers to the detectable presence of a leukemia-associated mutation without overt hematologic malignancy (1, 2). In the general population, CH is strongly age-dependent and associated with an increased risk of developing myeloid malignancies, cardiovascular diseases, and overall mortality (1–3). Previously, we showed that clonal expansion rates depend on the mutated gene (4). Still, large variation exists in expansion rates, even in individuals with the same mutation (4, 5). Which mutation-independent processes drive the expansion of mutant clones remains incompletely understood.
Allogeneic hematopoietic cell transplantation (allo-HCT) provides a unique, controlled setting to study the impact of forced proliferation, inflammation, and drug therapy on CH. Previous studies have shown that (i) hematopoietic stem cell (HSC) clones with CH driver mutations can be present in hematopoietic cell transplantation (HCT) donors, even at very young age (6, 7); (ii) donor-derived HSCs with CH mutations can engraft in the recipient (6–11); (iii) clone size is generally increased in recipients compared with donors (8, 10); (iv) the extent to which donor-derived mutant clones expand upon transplantation varies largely between individuals (6, 7, 9–11); and (v) forced proliferation of HSCs during hematopoietic regeneration results in DNA damage and telomere shortening (8), which may reduce long-term clonal diversity. Whereas long-term effects are most relevant to HCT recipients transplanted at pediatric age given their long life expectancy, most studies on posttransplant CH are limited to adults. Moreover, which transplant-related exposures facilitate the preferential outgrowth of CH mutant over wild-type clones remains unstudied.
In this study, using a well-annotated cohort of 144 long-term survivors of pediatric HCT and 258 nontransplanted controls, we investigated whether HCT at a young age increases the risk of developing CH and which extrinsic factors may confer a proliferative advantage to CH mutant clones.
Results
CH Prevalence Is Higher in Long-term Survivors of Pediatric HCT Compared with Healthy Controls
A total of 144 HCT recipients were enrolled at a median calendar age of 20.5 years [interquartile range (IQR): 16.2–27.7; Fig. 1A; Table 1; Supplementary Fig. S1]. In line with previous research (11), we defined “hematopoietic age” as the sum of the donor’s age at the time of stem cell harvest (median: 13.5 years, IQR: 5.1–26.8) and the number of years between HCT and CH assessment (“follow-up time,” median: 10.4 years, IQR: 8–17.9). This resulted in a median hematopoietic age of 32.1 years (IQR: 15.6–43.7), which represents the age of the donor-derived blood cells in the recipient. In total, we identified 30 CH mutations with variant allele frequencies (VAF) ≥0.01 in 23 HCT recipients (16%; 95% confidence interval: 10%–23%; Fig. 1B; Supplementary Table S1 and S2). Mutations were exclusively found in DNMT3A (24 mutations, 80%) or TET2 (six mutations, 20%). Most were single-base substitutions (23 mutations, 77%), of which 12 were cytosine-to-thymine transitions (Supplementary Fig. S2A and S2B). VAFs ranged from 0.01 to 0.31, with a median of 0.02. Chimerism analysis confirmed that posttransplant CH originated from donor cells in 20 of 23 cases. In the remaining three cases, a recipient origin of CH could not be excluded due to residual patient chimerism levels exceeding the size of the CH clone.
CH is common after HCT at pediatric age. A, Study design. B and C, Oncoprint of all HCT recipients (B) and controls (C), sorted by identified CH mutation and VAF (orange bars). A darker shade indicates multiple mutations in the same gene. *Additional genes in the CH panel without identified mutations: BRAF, CALR, CBL, CSF3R, ETNK1, EZH2, FLT3, IDH1, IDH2, JAK2, KIT, MPL, MYD88, NOTCH1, NPM1, NRAS, RUNX1, SETBP1, TP53, and WT1. NGS, next-generation sequencing. (A, Created in BioRender. Belderbos, M. (2024) https://BioRender.com/a16z966)
CH is common after HCT at pediatric age. A, Study design. B and C, Oncoprint of all HCT recipients (B) and controls (C), sorted by identified CH mutation and VAF (orange bars). A darker shade indicates multiple mutations in the same gene. *Additional genes in the CH panel without identified mutations: BRAF, CALR, CBL, CSF3R, ETNK1, EZH2, FLT3, IDH1, IDH2, JAK2, KIT, MPL, MYD88, NOTCH1, NPM1, NRAS, RUNX1, SETBP1, TP53, and WT1. NGS, next-generation sequencing. (A, Created in BioRender. Belderbos, M. (2024) https://BioRender.com/a16z966)
HCT characteristics for the study cohort.
Determinant . | Total (n = 144) . | CH (n = 23) . | No CH (n = 121) . | P value . |
---|---|---|---|---|
Recipient sex | ||||
Male | 83 (58%) | 13 (57%) | 70 (58%) | 1 |
Recipient age at inclusion; (years, median, IQR) | 20.5 (16.2–27.7) | 28.3 (19.4–33.2) | 20.2 (15.9–25.7) | 0.03 |
Recipient age at HCT; (years, median, IQR) | 8.0 (4.0–12.3) | 7.3 (3.6–15.3) | 8.0 (4.1–6.7) | 0.59 |
Hematopoietic age; (years, median, IQR) | 32.1 (15.6–43.7) | 44.0 (34.7–51.4) | 29.6 (14.2–42.0) | <0.01 |
Donor age at graft donation; (years, median, IQR) | 13.5 (5.1–26.8) | 30.2 (19.6–36.4) | 11.6 (4.3–25.1) | <0.01 |
Follow-up time; (years, median, IQR) | 12.1 (8.0–17.9) | 18.4 (8.5–22.8) | 11.6 (7.9–16.1) | 0.07 |
Underlying disease | 0.33 | |||
Hemato-oncology | 116 (81%) | 21 (91%) | 95 (79%) | |
Bone marrow failure syndromes | 24 (17%) | 2 (9%) | 22 (18%) | |
Immunodeficiency | 4 (3%) | 0 (0%) | 4 (3%) | |
Stem cell source | 0.01 | |||
Bone marrow | 101 (70%) | 18 (78%) | 83 (69%) | |
Cord blood | 33 (23%) | 1 (4%) | 32 (26%) | |
Peripheral blood | 10 (7%) | 4 (17%) | 6 (5%) | |
Relation to donor | 0.21 | |||
Unrelated | 84 (58%) | 16 (70%) | 68 (56%) | |
Family | 60 (42%) | 7 (30%) | 53 (44%) | |
Sibling | 53 (37%) | 5 (22%) | 48 (40%) | |
Parent | 7 (5%) | 2 (9%) | 5 (4%) | |
Conditioning | ||||
Chemotherapy-based | 112 (78%) | 18 (78%) | 99 (78%) | 1 |
TBI-based | 32 (22%) | 5 (22%) | 27 (22%) | |
Myeloablative | 134 (93%) | 20 (87%) | 114 (94%) | 1 |
Nonmyeloablative | 10 (7%) | 3 (13%) | 7 (6%) | |
Serotherapy | 0.01 | |||
Yes | 87 (60%) | 20 (87%) | 68 (56%) | |
No | 54 (38%) | 3 (13%) | 53 (44%) |
Determinant . | Total (n = 144) . | CH (n = 23) . | No CH (n = 121) . | P value . |
---|---|---|---|---|
Recipient sex | ||||
Male | 83 (58%) | 13 (57%) | 70 (58%) | 1 |
Recipient age at inclusion; (years, median, IQR) | 20.5 (16.2–27.7) | 28.3 (19.4–33.2) | 20.2 (15.9–25.7) | 0.03 |
Recipient age at HCT; (years, median, IQR) | 8.0 (4.0–12.3) | 7.3 (3.6–15.3) | 8.0 (4.1–6.7) | 0.59 |
Hematopoietic age; (years, median, IQR) | 32.1 (15.6–43.7) | 44.0 (34.7–51.4) | 29.6 (14.2–42.0) | <0.01 |
Donor age at graft donation; (years, median, IQR) | 13.5 (5.1–26.8) | 30.2 (19.6–36.4) | 11.6 (4.3–25.1) | <0.01 |
Follow-up time; (years, median, IQR) | 12.1 (8.0–17.9) | 18.4 (8.5–22.8) | 11.6 (7.9–16.1) | 0.07 |
Underlying disease | 0.33 | |||
Hemato-oncology | 116 (81%) | 21 (91%) | 95 (79%) | |
Bone marrow failure syndromes | 24 (17%) | 2 (9%) | 22 (18%) | |
Immunodeficiency | 4 (3%) | 0 (0%) | 4 (3%) | |
Stem cell source | 0.01 | |||
Bone marrow | 101 (70%) | 18 (78%) | 83 (69%) | |
Cord blood | 33 (23%) | 1 (4%) | 32 (26%) | |
Peripheral blood | 10 (7%) | 4 (17%) | 6 (5%) | |
Relation to donor | 0.21 | |||
Unrelated | 84 (58%) | 16 (70%) | 68 (56%) | |
Family | 60 (42%) | 7 (30%) | 53 (44%) | |
Sibling | 53 (37%) | 5 (22%) | 48 (40%) | |
Parent | 7 (5%) | 2 (9%) | 5 (4%) | |
Conditioning | ||||
Chemotherapy-based | 112 (78%) | 18 (78%) | 99 (78%) | 1 |
TBI-based | 32 (22%) | 5 (22%) | 27 (22%) | |
Myeloablative | 134 (93%) | 20 (87%) | 114 (94%) | 1 |
Nonmyeloablative | 10 (7%) | 3 (13%) | 7 (6%) | |
Serotherapy | 0.01 | |||
Yes | 87 (60%) | 20 (87%) | 68 (56%) | |
No | 54 (38%) | 3 (13%) | 53 (44%) |
P values were calculated from the χ2 test or Fisher exact test if expected count <5 (categorical variables) or the Wilcoxon rank-sum test (continuous variables). Serotherapy consisted of either antithymocyte globulin or alemtuzumab.
Abbreviations: IQR, interquartile range; TBI, total body irradiation.
To determine whether HCT affects the prevalence and characteristics of CH, we compared our HCT cohort to 258 nontransplanted controls, including 244 individuals from a diverse age range (median age: 36.5 years, IQR: 19.8–50.0; Supplementary Table S3), and 14 cord blood controls. We identified 22 CH mutations in 21 controls (Fig. 1C; Supplementary Fig. S2C and S2D; Supplementary Table S2). In line with existing large population cohorts (1, 2, 12, 13), CH was virtually absent in control individuals younger than 40 years and increased in prevalence at older ages (Supplementary Fig. S2E). No CH mutations were detected in cord blood controls (Supplementary Fig. S2E).
In the HCT setting, the age of the transplanted HSCs can differ from that of the HCT recipient by several decades. This discrepancy raises two key questions (Fig. 2A): (i) are HCT recipients at an increased risk of CH compared with age-matched nontransplanted controls? and (ii) does the HCT procedure itself increase the risk of CH in transplanted HSCs compared with age-matched nontransplanted HSCs? To study whether HCT recipients are at an increased risk of CH, we first matched each HCT recipient to a nontransplanted control in a 1:1 ratio based on the recipient’s age at study inclusion (Supplementary Tables S1 and S3). HCT recipients showed a significantly higher prevalence of CH compared with age-matched controls (OR: 6.3; P < 0.001; Fig. 2B; Supplementary Fig. S3A). This difference was especially visible in younger age groups. Whereas CH was absent in controls younger than 32 years of age, we detected CH in HCT recipients as young as 8 years. Given the long life expectancy of these young HCT recipients, these results highlight the importance of understanding the prevalence and potential clinical consequences of CH after pediatric HCT.
Comparison of CH in HCT recipients and controls. A, Schematic representation of recipient age and hematopoietic age concepts, used for matching of HCT recipients and controls. B and C, Cumulative prevalence of CH plotted against either HCT recipient/control age (B) or hematopoietic age (C), split by cohort. P values are derived from the multivariable logistic regression models shown in Supplementary Fig. S3. D, Violin plot of VAFs of CH clones in all controls and HCT recipients. P value is calculated using the Wilcoxon rank-sum test. E, Scatterplot of VAFs of the largest CH clone per individual vs. hematopoietic age. Linear regression is calculated for each cohort separately, excluding the outliers with VAF >0.30 in the HCT recipient cohort. (A, Created in BioRender. Belderbos, M. (2025) https://BioRender.com/a70q960)
Comparison of CH in HCT recipients and controls. A, Schematic representation of recipient age and hematopoietic age concepts, used for matching of HCT recipients and controls. B and C, Cumulative prevalence of CH plotted against either HCT recipient/control age (B) or hematopoietic age (C), split by cohort. P values are derived from the multivariable logistic regression models shown in Supplementary Fig. S3. D, Violin plot of VAFs of CH clones in all controls and HCT recipients. P value is calculated using the Wilcoxon rank-sum test. E, Scatterplot of VAFs of the largest CH clone per individual vs. hematopoietic age. Linear regression is calculated for each cohort separately, excluding the outliers with VAF >0.30 in the HCT recipient cohort. (A, Created in BioRender. Belderbos, M. (2025) https://BioRender.com/a70q960)
The prevalence of CH after transplant is likely (in part) dependent on hematopoietic age. Indeed, in multivariable regression analysis, both donor age at stem cell donation and follow-up time, together constituting hematopoietic age, were significant predictors of posttransplant CH, whereas recipient age at HCT was not (P = 0.69; Supplementary Fig. S3B). To investigate whether the HCT procedure itself increases the risk of CH beyond that associated with aging, we matched HCT recipients to nontransplanted controls based on hematopoietic age. Both older hematopoietic age (OR per year increase 1.07; P < 0.001) and the HCT procedure (OR 2.53; P = 0.02) independently increased the risk of CH (Fig. 2C; Supplementary Fig. S3C), indicating that HCT-related processes may play a significant role in driving CH.
Next, we compared the size of the CH clones. Most CH clones were small, both in HCT recipients and controls (median VAFs 0.023 and 0.015, respectively; P = 0.12; Fig. 2D). Notably, very large clones with VAF >0.10 were exclusively detected in HCT recipients. For both HCT recipients and controls, older hematopoietic age was associated with larger clone size (Fig. 2E). However, this only explained part of the observed variation in clone sizes. Specifically, two HCT recipients carried DNMT3A-mutated clones with VAFs of 0.30 and 0.31 at hematopoietic ages of 14.5 and 30.0 years, respectively. Strikingly, one of these recipients received a cord blood transplant, suggesting that even cord blood grafts may contain HSCs with CH driver mutations. Collectively, these data demonstrate that HCT recipients are at increased risk of CH at younger hematopoietic age compared with controls, likely originating from mutant donor HSCs which preferentially expand in the context of HCT and persist into adulthood.
Systemic Inflammation Is Associated with Posttransplant CH
To identify HCT-related exposures that may contribute to the risk of CH, we compared clinical characteristics between HCT recipients with and without CH. No differences were observed in underlying disease, stem cell source, CD34 dose, or donor–recipient relation (Fig. 3A). Remarkably, we identified various posttransplant exposures associated with the presence of CH. First, CH prevalence was higher in recipients who had received serotherapy as part of their conditioning regimen. After serotherapy administration (usually day −9 to −5 before graft infusion), T-cell lysis can result in an inflammatory response marked by fever and increased C-reactive protein (CRP; ref. 14). To investigate whether HCT recipients with long-term CH had experienced such an inflammatory response, we investigated CRP values around the time of graft infusion. Hereto, we extracted CRP values, obtained during preemptive monitoring, for all HCT recipients transplanted at the University Medical Center (UMC) Utrecht/Princess Máxima Center for Pediatric Oncology (PMC) (n = 80; Fig. 3B and C; Supplementary Fig. S4). Within this subset, all HCT recipients who later developed CH had received serotherapy. As expected, serotherapy was associated with a transient CRP rise (Supplementary Fig. S4A and S4B). Strikingly, CRP levels before and shortly after graft infusion were higher for HCT recipients with subsequent CH compared with those without both in the overall cohort and within those who had received serotherapy (Fig. 3B and C; Supplementary Fig. S4C). This difference was not observed at later time points. In addition, posttransplant viral reactivations, defined as Cytomegalovirus, Epstein–Barr virus, and/or adenovirus loads >1,000 copies/mL during preemptive screening, were associated with CH in univariable analysis (Fig. 3D). Whether this is the consequence of serotherapy or an independent inflammatory event could not be established. We did not observe a significant association between acute graft-versus-host disease (aGvHD) and CH (Fig. 3E). None of the HCT recipients in our cohort had developed a secondary hematologic malignancy at the time of enrollment.
Clinical factors associated with the prevalence of CH. A, Forest plot showing the effect of clinical characteristics on CH prevalence of univariable and multivariable logistic regression models. Stem cell source was corrected for donor age at the time of graft donation, CD34 dose was corrected for stem cell source, and serotherapy was corrected for relation to the donor. B, CRP measurements around HCT in individuals with and without CH. Lines indicate LOESS regression with 95% confidence interval (CI). C, Box plots demonstrating the highest CRP measured within 5-day intervals around graft infusion, split by serotherapy (yes/no) and CH status. P values are calculated using the Wilcoxon rank-sum test. D and E, Prevalence of CH in individuals with and without viral reactivations, including CMV, EBV, and ADV reactivation (D), and with and without severe aGvHD (E). P values are calculated using the χ2 test. ADV, adenovirus; BM, bone marrow; CB, cord blood; CMV, Cytomegalovirus; EBV, Epstein–Barr virus; MUD, matched unrelated donor; PBSC, peripheral blood stem cells; TBI, total body irradiation.
Clinical factors associated with the prevalence of CH. A, Forest plot showing the effect of clinical characteristics on CH prevalence of univariable and multivariable logistic regression models. Stem cell source was corrected for donor age at the time of graft donation, CD34 dose was corrected for stem cell source, and serotherapy was corrected for relation to the donor. B, CRP measurements around HCT in individuals with and without CH. Lines indicate LOESS regression with 95% confidence interval (CI). C, Box plots demonstrating the highest CRP measured within 5-day intervals around graft infusion, split by serotherapy (yes/no) and CH status. P values are calculated using the Wilcoxon rank-sum test. D and E, Prevalence of CH in individuals with and without viral reactivations, including CMV, EBV, and ADV reactivation (D), and with and without severe aGvHD (E). P values are calculated using the χ2 test. ADV, adenovirus; BM, bone marrow; CB, cord blood; CMV, Cytomegalovirus; EBV, Epstein–Barr virus; MUD, matched unrelated donor; PBSC, peripheral blood stem cells; TBI, total body irradiation.
Discussion
Due to the increasing use of HCT as a treatment and improvements in supportive care protocols, the number of long-term HCT survivors continues to increase. Although much research has focused on the nonhematopoietic late effects after HCT (15), the long-term effects on the hematopoietic system—the only organ not directly exposed to chemotherapy—remain largely unclear. Long-term complications are particularly relevant for pediatric recipients, who have a life expectancy of several decades after successful HCT. In this study, we demonstrate that long-term survivors of pediatric HCT are at increased risk of CH compared with the general population. Importantly, older hematopoietic age and the HCT procedure itself were independently associated with posttransplant CH, suggesting the involvement of both aging- and transplantation-induced effects. Furthermore, consistent with existing preclinical data indicating that inflammation promotes clonal expansion (16), we found an association between posttransplant CH and (serotherapy-induced) inflammation around graft infusion. This supports the hypothesis that systemic inflammation might promote the selection and/or preferential expansion of mutant HSCs, resulting in detectable CH years later.
The incidence and health consequences of CH are of substantial scientific and clinical interest (1, 2, 5, 17). Whereas previous studies have addressed CH in nontransplanted cancer survivors, in whom CH primarily arises from chemotherapy-induced toxicity (17), our study is the first to study CH in a large cohort of pediatric HCT survivors whose HSCs have not been exposed to chemotherapy. Our findings are in line with studies in adult HCT cohorts, demonstrating that the prevalence of CH is higher in HCT recipients compared with their donors (18). Using sensitive backtracking studies, most cases of adult posttransplant CH could be traced back to the donor graft, even in very young donors (6). Furthermore, donor-derived CH clones were more likely to engraft and/or expand after HCT, resulting in an increased clone size in recipients compared with donors (8, 10). Whereas residual graft material was not available for our study cohort, most transplant donors would be very unlikely to harbor detectable CH clones given their relatively young age. In addition, it is important to note that the detection of CH is highly dependent on the sequencing thresholds used and that CH-negative individuals may still harbor small CH clones below the lower limit of detection. Accordingly, the absence of detectable CH in a HCT donor does not rule out the possibility of mutated HSCs already being present in the graft. In line with this notion, we found a positive association between the age of the donor at stem cell donation and the risk of posttransplant CH, suggesting that most cases of posttransplant CH are donor-derived. Notably, we also observed CH within 15 years after cord blood HCT, which is amongst the youngest cases of CH reported thus far. This is consistent with previous single-cell lineage tracing (19) and twin studies (20), suggesting that HSCs carrying leukemia-associated mutations might already be present at birth and only grow out to become detectable clones upon HCT, which creates an environment favoring the selective expansion of mutant HSCs.
The clinical impact of posttransplant CH remains, to a large part, unknown. Long-term HCT survivors are confronted with various potential late effects, including second malignancies and cardiovascular disease (21, 22), which may be affected by or related to the presence of CH (23, 24). In our cohort, we did not observe any individual with donor-derived myelodysplasia or secondary hematologic malignancies. However, the follow-up duration of our study might still be too short to detect these outcomes. Notably, we have recently demonstrated that the risk of malignant transformation of CH depends on both the specific mutated gene and its growth dynamics (4). The observed prevalence of posttransplant CH at very young hematopoietic age suggests that transplantation causes either a transient or persistent increase in clonal growth. Distinguishing between these two scenarios could help identify clones with different risks of malignant progression. Future research should focus on prospective, longitudinal monitoring of CH in allo-HCT recipients to investigate growth dynamics and better assess the long-term clinical consequences, including cardiovascular disease, second malignancies, and overall survival.
Our study has several limitations, including heterogeneity in the patient population and time of sampling, and the unavailability of pretransplant graft material. Additionally, by focusing on very long–term survivors, HCT recipients with severe post-HCT complications may be underrepresented in our study cohort, as these complications may affect their long-term survival and/or ability to enroll. Accordingly, the lack of association between CH and potentially lethal HCT complications, such as aGvHD, needs to be interpreted with caution. Yet, from a clinical perspective, our findings remain relevant, as long-term HCT-related effects are relevant only for those who survive. Finally, although we observed an association between CH, serotherapy, CRP, and post-HCT viral reactivations, these findings do not indicate causality. In fact, many of these processes are linked, for example, serotherapy-induced T-cell depletion is associated with an increased risk of viral reactivations. In addition, the presence of CH itself may predispose to post-HCT inflammatory processes. As an exploratory study, we cannot distinguish between these possibilities or establish causality. Future studies, including longitudinal sampling of both the donor and the recipient, as well as mechanistic studies, may help understand how different aspects of HCT (e.g., mobilization, engraftment, forced proliferation, and inflammation) affect the prevalence and dynamics of posttransplant CH.
In summary, pediatric HCT recipients are at increased risk of CH at a young age, emphasizing the importance of prospective screening. Insight into the longitudinal dynamics of CH after HCT will be crucial to understanding the underlying mechanisms and to determine its potential impact on posttransplant health.
Methods
Cohort Characteristics
A total of 144 long-term survivors of allo-HCT (>5 years after transplantation), who underwent HCT during childhood (≤18 years) between 1981 and 2018, were enrolled during post-HCT follow-up at the PMC from January 2022 to October 2023 (Supplementary Fig. S1). Inclusion criterion was having received an allo-HCT at least 5 years prior. Exclusion criterion was the inability to understand the patient informed consent forms. Enrolled patients had been transplanted at the UMC Utrecht/PMC (n = 80) or Leiden University Medical Center (n = 64). The prevalence of CH in these HCT recipients was compared with nontransplanted controls. Adult controls were allo-HCT or blood donors unrelated to our HCT recipient cohort. Pediatric controls were children undergoing diagnostic procedures at the PMC who were not diagnosed with malignancies. Cord blood controls were obtained from leftover graft material for allo-HCT, unrelated to our HCT recipient cohort. Detailed information per individual is provided in Supplementary Table S1. The study was approved by the local ethics committees (NedMec no. NL77721.041.21, Radboud UMC CMO 2013/064) in accordance with the Declaration of Helsinki and institutional guidelines and regulations. All patients or their caretakers provided written informed consent.
Cell and DNA Isolation
From each HCT recipient, 5.0 to 8.5 mL of peripheral blood was collected into PAXgene Blood DNA tubes [Qiagen, catalog number (Cat. No.) 761115] and stored at 20°C until further processing. DNA was extracted using PAXgene Blood DNA Purification Kit (Qiagen, Cat. No.: 761133) by following the manufacturer’s instructions. For control samples, DNA was isolated from mononuclear cells separated by density-gradient centrifugation for peripheral blood samples or from cord blood mononuclear cells for cord blood grafts.
Panel Design, Library Preparation, and Sequencing
CH was assessed by error-correcting sequencing using single-molecule inversion probes (smMIP), targeting coding exons or specific regions of the 27 myeloid and lymphoid driver genes (4). smMIPS were designed in a tiling manner, preferentially covering all target nucleotides with two smMIPs targeting both DNA strands independently (Supplementary Table S4). This method enables detection of low-frequency clones and accurate quantification of the VAF. For library preparation, isolated DNA was first sonicated to 400-bp fragments using a Bioruptor Pico sonication device (Diagenode, Cat. No. B01080010). Library preparation and sequencing were performed according to the adapted manufacturer’s protocols [for library preparation: Protocol A: Pool and denature libraries for sequencing (standard loading), document # 1000000106351 v05; for sequencing: NovaSeq 6000 System Guide, document # 1000000019358 v17, Illumina]. The following adaptations were used: Per smMIP capture, a volume of 7 μL containing 100 ng of genomic DNA was used as input. PCR was performed in a total volume of 50 μL using 20 μL from the exonuclease-treated capture mixture (New England Biolabs; exonuclease I: Cat. No. M0293L, exonuclease III: Cat. No. M0293L). To obtain high-quality blood-derived genomic DNA (gDNA), the gDNA:smMIP ratio was set to 800:1. Finally, smMIP libraries were pooled, denaturated, and diluted to a concentration of 1.4 nmol/L. Sequencing was performed on NovaSeq 6000 (Illumina, Cat. No. 20012850, 300 cycles NovaSeq 6000 S1 reagent kit, Cat. No. 20028317), resulting in 2 to 150 bp paired-end reads.
Data Processing and Variant Calling
Data processing, including BCL to FASTQ conversion, demultiplexing of barcoded reads, alignment and generation of consensus reads, and variant calling were performed using commercial analysis software (Sequence Pilot version 5.4.1, JSI medical systems). The threshold for variant calling was set at VAF ≥0.01 and ≥10 consensus variant reads. To prevent detection of germline variants, variants at VAF >0.40 were excluded. All variants were manually inspected and curated to exclude possible false positives, using an extensive in-house database for recurrent errors, sequencing artifacts, and polymorphisms, based on 5+ years of experience with the panel in a diagnostic setting. In addition, run-specific false-positive calls were excluded by evaluating background noise at positions with multiple variant calls within each sequencing run. This extensive pipeline enabled sensitive and harmonized variant calling while improving the scientific value of called variants.
Clinical Data
Patients were transplanted following (inter)national disease-specific guidelines. Clinical characteristics and demographics were extracted from medical charts. Conditioning regimens were classified into total body irradiation–based or chemotherapy-based according to the most active component in the regimen. In addition, regimens were classified as myeloablative versus nonablative based on the conditioning intensity, taking into account the underlying disease. Serotherapy included antithymocyte globulin or alemtuzumab administered within 14 days prior to HCT with the goal of lymphodepletion. aGvHD was defined according to the modified Glucksberg criteria. Posttransplant viral reactivations were defined as Cytomegalovirus, Epstein–Barr virus, and/or adenovirus loads >1,000 copies/mL, detected during biweekly preemptive screening. For patients transplanted in the UMC Utrecht/PMC, historic laboratory data, including weekly preemptive CRP measurements around graft infusion, were available and extracted via the Utrecht Patient Oriented Database (25).
Statistical Analysis
For comparison of CH prevalence in HCT recipients and controls, HCT recipients and controls were matched using the MatchIt (RRID: SCR_025618) and optmatch (RRID: SCR_026185) packages in R (method = “optimal,” ratio = 1, replace = FALSE), based on either HCT recipient age or hematopoietic age [i.e., the age of the donor at the time of stem cell donation and the time elapsed between HCT and follow-up (11)] and excluding cord blood controls. Logistic and linear regression models were used to identify risk factors for the prevalence of CH and its VAF, respectively, correcting for potential confounders. Within HCT recipients, associations between clinical characteristics and CH were tested using Wilcoxon, χ2, or Fisher exact tests as appropriate.
Data availability
Raw sequencing data for this study were generated at the Genome Technology Center of the Radboud UMC. Derived data supporting the findings of this study are available in Supplementary Table S2. Key clinical information is available in Supplementary Table S1. In-depth clinical data are not publicly available due to patient privacy requirements but are available upon reasonable request from the corresponding author.
Authors’ Disclosures
D.S. Neuberg reports other support from Madrigal Pharmaceuticals outside the submitted work. M.E. Belderbos reports grants from Nederlandse Organisatie voor Wetenschappelijk Onderzoek, the European Hematology Association, DKMS Foundation, and Stichting Kinderen Kankervrij during the conduct of the study. No disclosures were reported by the other authors.
Authors’ Contributions
K.F. Müskens: Conceptualization, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. N. Wieringa: Investigation, writing–review and editing. M.G.J.M. van Bergen: Formal analysis, writing–review and editing. J.E. Bense: Investigation, writing–review and editing. B.M. te Pas: Investigation, writing–review and editing. A.P.J. de Pagter: Investigation, writing–review and editing. A.C. Lankester: Writing–review and editing. M.B. Bierings: Writing–review and editing. D.S. Neuberg: Formal analysis, writing–review and editing. S. Haitjema: Writing–review and editing. L.C.M. Kremer: Writing–review and editing. G.A. Huls: Writing–review and editing. S. Nierkens: Supervision, investigation, writing–review and editing. J.H. Jansen: Investigation, writing–review and editing. C.A. Lindemans: Supervision, investigation, writing–review and editing. A.O. de Graaf: Formal analysis, investigation, writing–review and editing. M.E. Belderbos: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, writing–original draft, writing–review and editing.
Acknowledgments
This study was financially supported by a VENI grant of the Netherlands Organization for Scientific Research (grant number VI.Veni.202.021 to M.E. Belderbos), a physician scientist grant of the European Hematology Association (to M.E. Belderbos), a John Hansen research grant from the DKMS Foundation, and a Stichting Kinderen Kankervrij project grant (project number 418). The Stichting Kinderen Kankervrij project grant supported the salary of K.F. Müskens. The funders had no role in the design, execution, or writing of this work, nor in the decision to submit results. We would like to thank M.C.H. de Groot for extracting historic laboratory data from the Utrecht Patient Oriented Database and the Genome Technology Center, Radboud UMC, for performing NovaSeq sequencing. Finally, we would like to thank the members of the Trial and Data Center of the PMC and the LTHIT study team for their help in patient inclusion and sample collection.
Note: Supplementary data for this article are available at Blood Cancer Discovery Online (https://bloodcancerdiscov.aacrjournals.org/).
References
Supplementary data
Cohort information
Sequencing data