Clonal Hematopoiesis-Associated Gene Mutations Affect Acute Graft-Versus-Host Disease After Hematopoietic Stem Cell Transplantation in AML Patients

Abstract

Background

The relationship between clonal hematopoiesis (CH)-associated gene mutations and allogeneic hematopoietic stem cell transplantation (allo-HSCT) has been extensively studied since next-generation sequencing (NGS) technology became widely available. However, research has mainly focused on the relationship between donor CH mutations and transplant prognosis, and research into the relationship between CH mutations in the recipient and acute graft-versus-host disease (aGVHD) is lacking.

Material/Methods

We analyzed NGS results and their correlation with aGVHD and prognosis in 196 AML patients undergoing allo-HSCT.

Results

A total of 93 (47.4%) patients had CH mutations. The most frequently mutated genes were DNMT3A (28 of 196; 14.3%), TET2 (22 of 196; 11.2%), IDH1 (15 of 196; 7.7%), IDH2 (14 of 196; 7.1%), and ASXL1 (13 of 196; 6.6%). The incidence of aGVHD was higher in patients older than 45 years old with DTA mutations (DNMT3A, TET2 or ASXL1). DNMT3A mutation but not with TET2 or ASXL1 mutation was an independent risk factor for aGVHD in patients receiving allo-HSCT older than 45 years old. With a median follow-up of 42.7 months, CH mutations were not associated with overall survival and leukemia-free survival.

Conclusions

DNMT3A mutation, but not TET2 or ASXL1 mutation, was associated with higher incidence of aGVHD.

Introduction

Clonal hematopoiesis (CH) is defined as the clonal expansion of single hematopoietic stem cells (HSCs) with driving somatic mutations or chromosomal abnormalities that give them a competitive advantage under positive selection pressure, resulting in clonal dominance [1]. Due to advancement of next-generation sequencing (NGS) technology in the past decade, more and more research groups found that CH is a common condition with aging and mutations in specific genes (eg, DNMT3A, TET2, ASXL1) [2] and can increase the risk of developing hematological malignancies, with a progression rate of ~0.5% per year [3], including bone marrow failure diseases, myeloproliferative neoplasms, and therapy-related myelodysplasia syndrome/acute myeloid leukemia (AML) [3,4]. Various terms have been used to describe states related to CH, such as clonal hematopoiesis of indeterminate potential (CHIP), age-related CH (ARCH), idiopathic cytopenia of unknown significance (ICUS), and clonal cytopenia of unknown significance (CCUS) [5]. Despite many similarities among these terms, we will use the term CH to simplify our discussion.

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) remains the only curative therapy for AML. Unfortunately, successful outcomes of allo-HSCT are limited by graft-versus-host disease (GVHD) and malignant relapse. Transplantation and GVHD have been suggested to accelerate the aging of transplanted donor-derived HSPC, which in turn could foster the emergence of CH [6]. Studies on CH and HSCT have mainly focused on the correlation between donor CH mutations and disease prognosis, with inconsistent results [7–12]. Further research on the relationship between CH and GVHD will help to determine the disease prognosis.

Another noteworthy issue is that allo-HSCT cannot clear all the mutations in recipients [7]. Kim found that chemotherapy and HSCT significantly reduced allelic burden, but 41 mutations (28.9%) were detected at post-HCT day 21, with DNMT3A being the most frequent [12]. Hasserjian also showed the persistence of CH after treatment, which may not reflect residual AML, complicating the determination of measurable residual disease (MRD) [9]. In Heuser’s study, DNMT3A, TET2, and ASXL1 (DTA) were not eliminated in 17.6% of patients with DTA mutations after allo-HSCT, which may represent residual CH of the host [10]. Since CH mutations persist after transplantation, which can affect GVHD as well, more attention should be paid to recipient CH mutations. In the literature on recipient CH and transplantation, there is a lack of studies on GVHD. Similarly, the prognostic role of CH mutations in recipients is still unclear [8,10,11,13,14].

We analyzed NGS results and their association of aGVHD and prognosis in 196 AML patients undergoing allo-HSCT. This work focused on the relationship between recipient CH mutations and aGVHD, providing early warning biomarkers for aGVHD and discussing their possible therapeutic targets.

Material and Methods

Patients

We included 196 AML patients who received allo-HSCT at Chinese People’s Liberation Army General Hospital between January 2010 and May 2021. Before treatment, NGS test was performed on all patients’ peripheral blood or bone marrow samples. This study is based on the 2016 World Health Organization (WHO) standards for diagnosis and classification of AML. NCCN risk stratification is based on 2021 NCCN guidelines. Complex karyotype was defined as ≥3 unrelated chromosome abnormalities in the absence of other class-defining recurring genetic abnormalities, excluding hyperdiploid karyotypes with 3 or more trisomies (or polysomies), without structural abnormalities [15]. Cytomegalovirus (CMV) infection, Epstein-Barr virus (EBV) infection, and fungal infection were defined as infection before aGVHD or in the 100 days after transplantation in patients without aGVHD.

All patients received cyclosporine (CsA), short-term methotrexate (MTX), and mycophenolate mofetil (MMF) as prophylaxis against graft-versus-host disease (GVHD). CsA was administered intravenously at a dose of 3 mg/kg from day 10 and transitioned to oral administration at around day 20–30. CsA was discontinued in patients who did not show signs of aGVHD. MTX was administered intravenously at a dose of 15 mg/m² on day 1 and 10 mg/m² on days 3, 6, and 11. MMF was given orally at a dose of 250 mg every 12 h from day 10 until engraftment.

Genetic Studies

Analysis of mutations present in the genes DNMT3A, TET2, ASXL1, IDH1, IDH2, SRSF2, SF3B1, JAK2, TP53, U2AF1, BCOR, CSF3R, IKZF1, CBL, BCORL1, CREBBP, and GNAS, all of which have previously been shown to be associated with CH, was performed by directional acquisition depth sequencing combined with NGS tests (Acornmed Biotechnology Co., Ltd., Tianjin, China).

Statistical Analysis

SPSS 26.0 software was used for statistical analysis. The primary endpoints were overall survival (OS) and leukemia-free survival (LFS). OS was defined as the time from diagnosis to death from any cause or the last follow-up (July 28, 2021). LFS was defined as the time from the first complete remission to recurrence or death or the last follow-up. OS and LFS were estimated using Kaplan-Meier analysis, and the difference was tested with log-rank tests. Univariate and multivariate analyses were performed on Cox models for LFS and OS. Continuous variables (age, white blood cell count, hemoglobin level, platelet count, proportion of bone marrow blasts) are expressed as median (range), which were compared by the Mann-Whitney U test. Chi-square tests and Fisher’s exact tests were used to compare the classification variables (sex, donor type, FAB classification, NCCN risk stratification, GVHD). Logistic regression was used to explore the risk factors of aGVHD. Factors with P value less than 0.15 in the univariate analysis were included in the multivariate analysis. P<0.05 was regarded as statistically significant.

Results

Mutation Landscape of CH Mutations in AML Patients Received Allo-HSCT

CH mutations were detected in 93 of 196 (47.4%) patients who had NGS performed before treatment. The most frequently mutated genes were DNMT3A (28 of 196; 14.3%), TET2 (22 of 196; 11.2%), IDH1 (15 of 196; 7.7%), IDH2 (14 of 196; 7.1%), and ASXL1 (13 of 196; 6.6%; Figure 1A). Of the 93 patients with CH mutations, 58 (62.4%) had 1, 27 (29.0%) had 2, and 8 (8.6%) patients had 3 different CH mutations (Figure 1B).

Figure 1.

The landscape of CH mutations in 196 AML patients. (A) The most frequently mutated genes were DNMT3A, TET2, IDH1, IDH2 and ASXL1. (B) Co-mutation distribution in 93 patients with CH mutations. (C) The distribution of age on the basis of CH status and mutations in individual genes or groups of genes. Figure was created using GraphPad Prism 10 by Dotmatics.

Clinical and Genetic Characteristics of CH Mutations

CH prevalence increased with advancing age: median age 36 years vs 42 years for patients without and with CH mutations (P =0.003, Table 1). Patients with mutated DNMT3A and TET2 were older than those without CH mutations (Figure 1C). The presence of CH mutations was also associated with higher platelet levels (P=0.048). Aside from the above differences, there were no significant differences in sex, complex karyotypes, NCCN risk stratification, and disease status at transplantation (Table 1).

Table 1.

Characteristics of patients.

Characteristic	Total N=196	No CH mutation N=103	With CH mutations N=93	P value
Age	38 (11–64)	36 (13–62)	42 (11–64)	0.003a
Sex
Female	73 [37.2]	33 [32.0]	40 [43.0]	0.113
Male	123 [62.8]	70 [68.0]	53 [57.0]
AML type
De Novo	182 [92.6]	93 [90.3]	89 [95.7]	0.142
Secondary	14 [7.4]	10 [9.7]	4 [4.3]
WBC, ×109/L	31.1 (3.8–40.7)	36.3 (4.7–49.9)	25.3 (3.0–31.2)	0.094b
Platelet, ×109/L	79.1 (26.0–103.0)	79.9 (25.0–90.0)	84.8 (27.3–105.0)	0.048b
Hemoglobin, g/dL	89.4 (70.0–110.0)	89.5 (70.0–110.0)	89.3 (70.3–111.0)	0.923b
Bone marrow blasts, %	57.9 (38.5–80.0)	58.8 (40.0–80.6)	56.9 (38.1–77.2)	0.556b
Complex karyotypes
Absent	184 [93.9]	98 [95.2]	86 [92.5]	0.436
Present	12 [6.1]	5 [4.8]	7 [7.5]
NCCN risk stratification#
Favorable-risk	80 [40.8]	43 [41.7]	37 [39.8]	0.584
Intermediate-risk	49 [25.0]	28 [27.2]	21 [22.6]
Poor-risk	67 [34.2]	32 [31.1]	35 [37.6]
Disease status at transplantation
De Novo	1 [0.5]	1 [1.0]	0 [0]	0.305*
CR	174 [88.8]	91 [88.3]	83 [89.3]
PR	6 [3.1]	3 [2.9]	3 [3.2]
NR	6 [3.1]	2 [1.9]	4 [4.3]
Relapse	9 [4.5]	6 [5.9]	3 [3.2]
Donor type
Sibling matched	62 [31.6]	37 [36.0]	25 [26.9]	0.397
Unrelated matched	10 [5.1]	5 [4.8]	5 [5.4]
Haploidentical	124 [63.3]	61 [59.2]	63 [67.7]

NOTE. Data are no. [%] unless otherwise specified. P values are from two-sided tests for categoric variables.

^arepresents the result from unpaired t test.

^brepresents the results from Mann-Whitney tests.

^*represents the results from Fisher’s exact tests.

^#NCCN risk stratification is based on NCCN guidelines in 2021.

CH mutation – clonal hematopoiesis-associated mutation; AML – acute myeloid leukemia; WBC – white blood cell count; NCCN – National Comprehensive Cancer Network; CR – complete remission; PR – partial response; NR – no response.

Associations of CH Mutations and aGVHD

AGVHD occurred in 116 (59.2%) patients – 40 patients with grade I aGVHD, 53 patients with grade II aGVHD, 17 patients with grade III aGVHD, and 6 patients with grade IV aGVHD – and data were missing for 1 patient (Table 2).

Table 2.

Incidence of aGVHD.

	Total N=196	No CH mutations N=103	With CH mutations N=93	P
aGVHD grade*				0.210
No	79 [40.5]	47 [46.1]	32 [34.4]
I	40 [20.5]	20 [19.6]	20 [21.5]
II	53 [27.2]	21 [20.6]	32 [34.4]
III	17 [8.7]	9 [8.8]	8 [8.6]
IV	6 [3.1]	5 [4.9]	1 [1.1]

Data are no. [%] unless otherwise specified.

^*A patient’s data is missing.

CH mutation – clonal hematopoiesis-associated mutation; aGVHD – acute graft versus host disease.

Considering the association of CH with older age, we investigated the incidence of aGVHD in patients in various age groups. In haploidentical hematopoietic stem cell transplantation (haplo-HSCT), patients with CH mutations had a higher incidence of aGVHD (79.4% vs 63.3%, P=0.049, Table 3). Then we analyzed which kind of mutations would contribute to this phenomenon, but there were no statistically significant results (not shown). However, when we classified the mutations by their mechanisms, statistically significant correlations were found between DTA (mutated DNMT3A, TET2 or ASXL1) and aGVHD (80.0% vs 40.9%, P=0.010, Table 3) in patients over 45 years old who underwent haplo-HSCT. To further investigate which mutation of the 3 played the most important part, we analyzed the data of patients with DNMT3A mutation but no TET2 or ASXL1 mutations and patients without DTA mutation, and found that patients with DNMT3A mutation but not the other t2 mutations had a higher probability of aGVHD (81.8% vs 40.9%, P=0.026, Table 3).

Table 3.

DNMT3A but not TET2 or ASXL1 mutation was associated with increased risk of aGVHD in patients over 45 years old.

	Allo-HSCT, ≥45 years old			Haplo-HSCT			Haplo-HSCT, ≥45 years old
	No aGVHD	aGVHD	P	No aGVHD	aGVHD	P	No aGVHD	aGVHD	P
CH mutations			0.320			0.049			0.184
No	14	12		22	38		8	6
Yes	17	26		13	50		9	19
DTA mutations			0.048			0.062			0.010
No	22	18		29	58		13	9
Yes	9	20		6	30		4	16
Mutated DNMT3A			0.042			0.139*			0.026
Noa	22	18		29	58		13	9
Yesb	4	12		2	14		2	9
Mutated TET2			0.652*						0.096*
Noa	22	18					13	9
Yesc	2	3					0	3
Mutated ASXL1			0.187*						0.165*
Noa	22	18					13	9
Yesd	1	4					1	4
Mutated TET2 or ASXL1			0.300						0.113*
Noa	22	18					13	9
Yese	5	8					2	7

^*Represents the results from Fisher’s exact tests.

^aPatients without DTA mutation;

^bpatients with DNMT3A mutation but not the other 2 mutations;

^cpatients with TET2 mutation but not the other 2 mutations’

^dpatients with ASXL1 mutation but not the other 2 mutations;

^epatients with TET2 or ASXL1 mutations but no DNMT3A mutation.

Allo-HSCT – allogeneic hematopoietic stem cell transplantation; haplo-HSCT – haploidentical hematopoietic stem cell transplantation; aGVHD – acute graft versus host disease; CH mutations – clonal hematopoiesis-associated mutation; DTA mutations – mutated DNMT3A, TET2, or ASXL1.

The incidences of grade II–IV and grade III–IV aGVHD were 69.0% vs 45.0% (P=0.048) and 20.7% vs 17.5% (P=0.738) for patients over 45 years old undergoing allo-HSCT with and without CH mutations, respectively (Table 4). A similar conclusion was also drawn for patients over 45 years old undergoing allo-HSCT (Tables 3, 4).

Table 4.

The risk of aGVHD.

	Grade II–IV aGVHD			Grade III–IV aGVHD
	No	Yes	P	No	Yes	P
Haplo-HSCT
CH mutations			0.056			0.835
No	22	39		54	7
Yes	13	50		55	8
DTA mutations			0.067			0.367*
No	29	59		79	9
Yes	6	30		30	6
Allo-HSCT
CH mutations			0.143			0.395
No	46	57		89	14
Yes	32	61		84	9
DTA mutations			0.348			0.823
No	59	82		124	17
Yes	19	36		49	6
Haplo-HSCT, ≥45 years old
CH mutations			0.120			0.692*
No	8	6		12	2
Yes	9	19		21	7
DTA mutations			0.010			0.268*
No	13	9		19	3
Yes	4	16		14	6
Allo-HSCT, ≥45-year-old
CH mutations			0.247			0.535*
No	14	12		20	6
Yes	17	26		36	7
DTA mutations			0.048			0.738
No	22	18		33	7
Yes	9	20		23	6

^*Represents the results from Fisher’s exact tests.

Allo-HSCT – allogeneic hematopoietic stem cell transplantation; haplo-HSCT – haploidentical hematopoietic stem cell transplantation; aGVHD – acute graft versus host disease; CH mutations – clonal hematopoiesis-associated mutation; DTA mutations – mutated DNMT3A, TET2, or ASXL1.

In the logistic regression analysis in patients over 45 years old and receiving haplo-HSCT, DNMT3A (with DNMT3A mutation but no TET2 or ASXL1 mutations vs without DTA mutations: P=0.045, OR=25.443, 95% CI: 1.072–603.681, Table 5) was corelated with the higher risk of aGVHD. The same conclusion was also drawn in patients over 45 years old undergoing allo-HSCT (with DNMT3A mutation but no TET2 or ASXL1 mutations vs without DTA mutations: P=0.033, OR=5.274, 95% CI: 1.140–24.393, Table 6).

Table 5.

Univariate and Multivariate analysis of the association of various clinical features with aGVHD in patients receiving haplo-HSCT.

	Univariate analysis			Multivariate analysis
	OR	95% CI	P	OR	95% CI	P
Haplo-HSCT
Age (y)	0.975	0.944–1.007	0.129	0.952	0.912–0.994	0.026
Disease status at transplantation (non-CR vs CR)	1.448	0.491–4.275	0.502
Complex karyotypes (yes vs no)	0.274	0.058–1.292	0.102	0.144	0.020–1.069	0.058
Bone marrow blasts (%)	1.015	0.998–1.034	0.092	1.009	0.988–1.030	0.395
CH mutations (yes vs no)	2.227	0.995–4.981	0.051	1.842	0.443–7.658	0.401
DTA mutations (yes vs no)	2.500	0.935–6.684	0.068	2.287	0.443–11.799	0.323
Mutation of methylation (yes vs no)	2.400	0.943–6.107	0.066	1.413	0.278–7.193	0.677
CMV (yes vs no)	0.484	0.209–1.116	0.089	0.460	0.163–1.302	0.144
EBV (yes vs no)	0.394	0.161–0.963	0.041	0.445	0.147–1.353	0.154
Fungal (yes vs no)	0.180	0.031–1.033	0.054	0.194	0.018–2.101	0.178
Haplo-HSCT, <45 years old
CH mutations (yes vs no)	3.391	1.005–11.439	0.049	2.447	0.682–8.780	0.170
CMV (yes vs no)	0.235	0.078–0.706	0.010	0.288	0.092–0.900	0.032
Fungal (yes vs no)	0.129	0.011–1.514	0.103	0.269	0.017–3.104	0.269
Haplo-HSCT, ≥45 years old
Mutated DNMT3A (yesa vs nob)	6.500	1.127–37.484	0.036	25.443	1.072–603.681	0.045
Disease status at transplantation (non-CR vs CR)	0.458	0.044–4.820	0.516
Platelets (×109/L)	1.011	0.998–1.025	0.103	1.007	0.994–1.019	0.294
Fungal (yes vs no)	0.182	0.030–1.102	0.064	0.046	0.003–0.738	0.030

^aPatients with DNMT3A mutation but no TET2 or ASXL1 mutations;

^bpatients without DTA mutations;

^cpatients with TET2 or ASXL1 mutations but no DNMT3A mutation.

OR – odds ratio; haplo-HSCT – haploidentical hematopoietic stem cell transplantation; CR – complete response; CH mutations – clonal hematopoiesis-associated mutation; DTA mutations – mutated DNMT3A, TET2, or ASXL1; CMV – cytomegalovirus; EBV – Epstein-Barr Virus.

Table 6.

Univariate and Multivariate analysis of the association of various clinical features with aGVHD in patients receiving allo-HSCT.

	Univariate analysis			Multivariate analysis
	OR	95% CI	P	OR	95% CI	P
Allo-HSCT
Age (y)	0.984	0.960–1.008	0.179
Disease status at transplantation (non-CR vs CR)	0.841	0.335–2.110	0.712
NCCN#
Intermediate vs favorable	0.820	0.391–1.717	0.598	0.795	0.359–1.762	0.572
Poor vs favorable	0.568	0.291–1.107	0.097	0.496	0.241–1.018	0.056
Donor type (matched vs haploidentical)	1.950	1.416–2.686	<0.001	3.760	2.005–7.049	<0.001
CH mutations (yes vs no)	1.566	0.878–2.794	0.129	1.497	0.803–2.792	0.205
Fungal (yes vs no)	0.316	0.077–1.302	0.111	0.229	0.052–1.012	0.052
Allo-HSCT, ≥45 years old
Mutated DNMT3A (yesa vs nob)	3.667	1.008–13.343	0.049	5.274	1.140–24.393	0.033
Mutated TET2 or ASXL1 (yesc vs nob)	2.139	0.540–8.480	0.279
Donor type (matched vs haploidentical)	1.584	0.598–4.196	0.355
Disease status at transplantation (non-CR vs CR)	0.283	0.030–2.676	0.271
EBV (yes vs no)	0.219	0.041–1.181	0.077	0.164	0.027–1.005	0.051

^aPatients with DNMT3A mutation but no TET2 or ASXL1 mutations;

^bpatients without DTA mutations;

^cpatients with TET2 or ASXL1 mutations but no DNMT3A mutation;

^#NCCN risk stratification is based on NCCN guidelines in 2021.

OR – odds ratio; Allo-HSCT – allogeneic hematopoietic stem cell transplantation; NCCN – National Comprehensive Cancer Network; CH mutations – clonal hematopoiesis-associated mutation; CMV – cytomegalovirus; EBV – Epstein-Barr Virus.

Analysis of Factors Related to aGVHD

In the logistic regression analysis in patients receiving allo-HSCT, including NCCN risk stratification, donor type and different pathogens infection as covariates, donor type (matched vs haploidentical: p<0.001, OR=3.760, 95% CI: 2.005–7.049, Table 6) was associated with higher risk of aGVHD. Meanwhile, in the logistic regression analysis in patients with haplo-HSCT, including age, bone marrow blasts, DTA mutations, mutation of methylation, and different pathogens infection as covariates, age (P=0.026, OR=0.952, 95% CI: 0.912–0.994, Table 5) was associated with aGVHD. In patients younger than 45 years old and receiving haplo-HSCT, CMV infection (vs no CMV infection: P=0.032, OR=0.288, 95% CI: 0.092–0.900, Table 5) was corelated with the lower risk of aGVHD.

Impact of CH Mutations on the Prognosis of AML Patients

The median follow-up was 42.7 months (4.6–145.8 months) in this study. The 5-year OS rate was 65.3%. Kaplan-Meier survival analysis did not show survival differences according to CH mutations regardless of the type of transplantation (Figure 2). The complication of transplantation-associated thrombotic microangiopathy (TA-TMA) was associated with a significantly shorter OS and LFS (Figures 3, 4) and a poor pretransplant disease status was associated with a significantly shorter LFS in patients undergoing allo-HSCT (Figure 4).

Figure 2.

Kaplan-Meier curves depict the overall survival and leukemia-free survival. (A) For patients undergoing HSCT, CH mutation was not associated with OS. (B) The same conclusion was drawn for patients who had haplo-HSCT. (C, D) The same conclusion was drawn for LFS. CH mutations – clonal hematopoiesis-associated mutation; HSCT – hematopoietic stem cell transplantation; OS – overall survival; haplo-HSCT – haploidentical hematopoietic stem cell transplantation; LFS – leukemia-free survival. Figure was created using GraphPad Prism 10 by Dotmatics.

Figure 3.

Kaplan-Meier curves depict the overall survival. (A) For patients undergoing HSCT, TA-TMA was associated with a significantly shorter OS. (B) The same conclusion was drawn for patients who had haplo-HSCT. HSCT – hematopoietic stem cell transplantation; TA-TMA – transplantation-associated thrombotic microangiopathy; OS – overall survival; haplo-HSCT – haploidentical hematopoietic stem cell transplantation. Figure was created using GraphPad Prism 10 by Dotmatics.

Figure 4.

Kaplan-Meier curves depict the leukemia-free survival. (A, B) For patients undergoing HSCT, TA-TMA, and a poor pretransplant status were associated with a significantly shorter LFS. (C, D) Only TA-TMA was connected to a shorter LFS in patients had haplo-HSCT. HSCT – hematopoietic stem cell transplantation; TA-TMA – transplantation-associated thrombotic microangiopathy; LFS – leukemia-free survival; haplo-HSCT – haploidentical hematopoietic stem cell transplantation; CR – complete remission; PR – partial response; NR – no response. Figure was created using GraphPad Prism 10 by Dotmatics.

Univariate analysis of OS and LFS in patients undergoing allo-HSCT showed that CH mutations, DTA mutations, and mutated DNMT3A were not risk factors for poor prognosis of OS and LFS (Table 7). The risk factors include pretransplant disease status (OS: NR vs CR, P=0.006, HR=3.646, 95% CI: 1.457–9.121; LFS: NR vs CR, p=0.013, HR=3.570, 95% CI: 1.301–9.793, Rel vs CR, P=0.039, HR=2.182, 95% CI: 1.005–4.739; Table 7) and complication of TA-TMA (yes vs no, OS: P=0.010, HR=3.050, 95% CI: 1.313–7.083; LFS: P=0.009, HR=2.803, 95% CI: 1.287–6.101; Table 7). In allo-HSCT patients, multivariate analysis showed that patients with pretransplant NR tended to have worse OS than those with pretransplant CR (p=0.009, HR=3.444, 95% CI: 1.366–8.685; Table 7), and patients with pretransplant recurrence tended to have poorer LFS (Rel vs CR, p=0.023, HR=2.490, 95% CI: 1.134–5.467; Table 7). The risk factors of OS and LFS also include concurrent TA-TMA (yes vs no, OS: P=0.012, HR=3.000, 95% CI: 1.277–7.049; LFS: P=0.013, HR=2.703, 95% CI: 1.230–5.940; Table 7).

Table 7.

Univariate and multivariate analyses for OS and PFS in 196 patients receiving allo-HSCT.

	Univariate analysis			Multivariate analysis
	HR	95% CI	P	HR	95% CI	P
OS
Age	1.007	0.987–1.027	0.481
Donor type (haploidentical vs matched)	1.605	0.970–2.656	0.065	1.361	0.809–2.291	0.246
Disease status at transplantation
PR vs CR	2.501	0.901–6.940	0.078	2.293	0.804–6.544	0.121
NR vs CR	3.646	1.457–9.121	0.006	3.444	1.366–8.685	0.009
Rel vs CR	1.323	0.480–3.644	0.589	1.224	0.443–3.383	0.697
CH mutations (yes vs no)	1.304	0.821–2.071	0.260
DTA mutations (yes vs no)	0.973	0.571–1.660	0.921
Mutated DNMT3A (yes vs no)	0.713	0.327–1.555	0.395
aGVHD (yes vs no)	1.050	0.656–1.680	0.838
cGVHD (yes vs no)	0.770	0.442–1.342	0.356
TA-TMA (yes vs no)	3.050	1.313–7.083	0.010	3.000	1.277–7.049	0.012
LFS
Age	1.006	0.988–1.024	0.513
Donor type (haploidentical vs matched)	1.550	0.990–2.427	0.056	1.333	0.837–2.123	0.226
Disease status at transplantation
PR vs CR	1.435	0.450–4.578	0.541	1.416	0.434–4.616	0.564
NR vs CR	3.570	1.301–9.793	0.013	2.626	0.935–7.380	0.067
Rel vs CR	2.182	1.005–4.739	0.039	2.490	1.134–5.467	0.023
CH mutations (yes vs no)	1.494	0.987–2.261	0.058	1.482	0.963–2.281	0.074
DTA mutations (yes vs no)	1.123	0.711–1.772	0.619
Mutated DNMT3A (yes vs no)	0.898	0.478–1.688	0.738
aGVHD (yes vs no)	1.104	0.721–1.689	0.649
cGVHD (yes vs no)	0.698	0.42–1.158	0.164
TA-TMA (yes vs no)	2.803	1.287–6.101	0.009	2.703	1.230–5.940	0.013

HR – hazard ratio; OS – overall survival; LFS – leukemia-free survival; CR – complete remission; PR – partial response; NR – no response; CH mutations – clonal hematopoiesis-associated mutation; DTA mutations, – mutated DNMT3A, TET2, or ASXL1; aGVHD – acute graft versus host disease; cGVHD – chronic graft versus host disease; TA-TMA – transplantation-associated thrombotic microangiopathy.

Discussion

GVHD results from immune cells transplanted from a non-identical donor recognizing the recipient as foreign, thereby initiating an immune response that causes disease. There has been fierce debate between CH and GVHD in previous studies. Frick found a high cumulative incidence of chronic GVHD (cGVHD) but not aGVHD in recipients allografted with donor CH [11]. Frick’s view was supported by Gibson, who concluded that DNMT3A mutation increased the incidence of cGVHD [16]. The opposite is claimed by Oran, who contends that donor CH is linked to an increased risk of grade II–IV and III–IV aGVHD in AML/MDS patients [17]. There has been no study focusing on the relationship between recipient CH and GVHD.

In our study, DNMT3A mutation but not TET2 or ASXL1 mutations was associated with an increased incidence of aGVHD in AML patients older than 45 years of age who received HSCT. So far, there has been no research focused on the mechanism between recipient CH and GVHD, which needs to be further explored.

The association between CH and aging is receiving increasing attention [5,18,19]. As expected, patients with DNMT3A mutation were older than those without DNMT3A mutation (P<0.001, not shown) in our study. Besides HLA disparity, older age of both the recipient and donor are risk factors for the development of aGVHD [20]. The simultaneous increase in DNMT3A mutations and aGVHD in recipients may be caused by aging of the recipients.

Furthermore, according to relevant studies on the pathogenesis of aGVHD and the function of DNMT3A, DNMT3A mutation may increase the occurrence of aGVHD by affecting proinflammatory cytokines, T cell differentiation, TCR diversity, and other pathways. Inflammatory disorders like GVHD are influenced by the cytokine milieu and subsequent interactions between naive T cells and antigen-presenting cells that increase differentiation and expansion. A growing body of evidence suggests that dysregulated inflammation contributes to clonal expansion and associated comorbidities [21]. Thus, CH, as a driver and consequence of inflammation, is intimately tied to GVHD. DNMT3A mutations are most commonly observed in the development of AML; it has been demonstrated that these mutations interfere with the functionality of DNMT3A [22]. Experiments in mice have demonstrated that DNA methylation plays a significant role in T cell polarization, particularly in terms of regulating cytokine expression [23,24]. Gamper and Cooke’s team reported that mice receiving allo-BMT from DNMT3A KO donors developed severe aGVHD, with increases in inflammatory cytokine levels and organ histopathology scores [25]. Additionally, IL-12p70, which was significantly higher in recipients of DNMT3A-CH and was correlated with other proinflammatory cytokines, has been implicated in the development of GVHD and GVL through its positive effects on Th1 polarization and INF-γ production in CD4 T cells [16,26,27]. These results provided a mechanistic rationale for exploring therapeutic modulation of DNA methyltransferase activity to augment efficacy of cell-based immune therapies. However, Tanja Bozič found that partial hypermethylation could still be observed after silencing DNMT3A using shRNA, which was inconsistent with the expected results. Other methyltransferases, such as DNMT3b, may increase as a compensatory measure, but the regulation of their activity may differ from that of DNMT3A [28]. A recent study by Trowbridge used a DNMT3A mutant mouse model to produce a TNFα-TNFR1 model in which clonal size and mature blood lineages produced could be independently controlled to regulate favorable and unfavorable outcomes [29]. Thus, targeted drugs or signaling pathways for DNA methylation need further investigation. Moreover, Husby found lower T cell clonotype levels in the elderly and patients with clonal hematopoiesis, which may be associated with GVHD [30].

In our study, CH mutations were not associated with OS and LFS. There are different views on the impact of CH mutations on prognosis. In Heuser’s study, DTA mutations had no prognostic significance in cumulative incidence of relapses, relapse-free survival, or overall outcome, excluding donor origin [10]. Grimm concluded that the presence of CH did not negatively impact the outcome of an allo-HSCT [8]. Frick and Newell confirmed that donor CH had no effect on prognosis [11,13]. Conversely, Gibson showed that CH was associated with an adverse prognosis after autologous stem cell transplantation for lymphoma [14]. Upon successful allo-HSCT, the recipient’s hematopoiesis is replaced by the donor’s, thereby eliminating CH mutations. It has therefore been speculated that CH mutations are reliable markers of MRD after allo-HSCT, as they indicate the persistence or relapse of recipient hematopoiesis and the leukemic clone [31]. Unfortunately, there is insufficient evidence to prove that CH mutations can be used as MRD tests [15]. Based on our study, CH mutations are not associated with OS and LFS, which may be due to the following 2 factors. Inflammation as well as increased proliferation of HSPCs, which may promote expansion of clones with a proliferative advantage, are inevitable consequences of HSCT, resulting in poor prognosis [32], but chemotherapy and HSCT can eliminate CH clones and their negative effects [12]. Secondly, the enhanced GVL of the immune response mitigates the adverse outcome of the high inflammatory response [11].

This study has potential limitation. In our research, it was difficult to distinguish CH from AML clones. We could only report this phenomenon, and further exploration of the mechanism and efforts to distinguish CH and AML clones are still needed. When specific mutation is involved, such as only TET2 mutation but not DNMT3A or ASXL1 mutation, the number of cases is relatively small, and more cases are needed for further verification. Additionally, the existence of competing risks can bias estimation of the probability of endpoint events. However, the limited data available to us made it difficult to complete the calculation of the Cox models with competing risk. In future work, it would be advisable to perform on Cox models with competing risks.

Conclusions

DNMT3A mutation but no TET2 or ASXL1 mutation was an independent risk factor associated with aGVHD in HSCT patients older than 45 years old. CH mutations were not associated with OS and LFS. CH likely represents a primitive stage in the development of hematologic neoplasms, whatever the mechanism leading to clonal expansion of hematopoietic progenitors [33]. Further studies to confirm this association, and strategies to decrease the risk of aGVHD including changes in the prevention for specific gene mutated patients are warranted to improve transplant outcomes.