Molecular characterization and improved risk stratification of myelodysplastic syndromes with bone marrow fibrosis through integration of TP53 and MPL mutations

Abstract

Myelodysplastic syndromes with moderate-to-severe bone marrow fibrosis (MDS-F) represent a distinct entity with poor prognosis, yet their molecular landscape and precise risk stratification remain poorly defined. This study aimed to characterize the genetic landscape of MDS-F and develop an integrated prognostic model. In this study, clinical and genetic data from 102 treatment-naïve MDS patients with moderate to severe marrow fibrosis (MF-2 and MF-3) were analyzed retrospectively. The cohort comprised 63 MF-2 and 39 MF-3 cases showing comparable clinical characteristics and survival outcomes. Targeted next-generation sequencing of 29 myeloid-associated genes demonstrated distinct mutational patterns: 19.6% harbored TP53 mutations, while 19.6% exhibited JAK-STAT pathway mutations (JAK2: 9.8%, MPL:7.8%, CALR: 2%). Notably, patients with JAK-STAT mutations displayed superior clinical outcomes compared to wild-type counterparts, with the favorable prognostic impact appearing to be primarily driven by MPL mutations (median OS: not reached vs. 24 months, p = 0.004). In the univariate analysis, age, IPSS-R score, as well as TP53, IDH1/2 and MPL mutations were evidently linked with overall survival (OS) and leukemia-free survival (LFS). Of these factors, only IPSS-R score, TP53 mutations and MPL mutations retained independent prognostic significance in the multivariate analysis (p < 0.05). By incorporating IPSS-R and genetic mutations, we generated a novel prognostic model that improved risk stratification, delineating four distinct survival trajectories (median OS: not reached vs. 40 vs. 26 vs. 11 months, log-rank test p < 0.001). In conclusion, MDS patients with MF had specific genetic lesions. The integration of mutational profiling, including the potentially favorable impact of MPL mutations, with conventional scoring systems enables refined prognostic assessment and warrants further validation to guide risk-adapted therapeutic strategies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-04962-w.

Introduction

Myelodysplastic syndromes (MDS), an array of heterogeneous clonal myeloid neoplasms, bring about various outcomes [1]. MDS are classified based on cytopenia, morphology of bone marrow (BM) dysplasia, BM blast percentage, and cytogenetic profile [2], rarely on BM histological features. Biopsy analysis shows that about 10–20% of patients with MDS are complicated with myelofibrotic BM [3–5]. MDS with bone marrow fibrosis (MDS-F) manifests distinct and unique clinical features, including multilineage dysplasia, transfusion dependence, profound cytopenia, cytogenetic abnormalities and poor prognosis [6, 7]. In the 2008 WHO classification, MDS-F was in the spotlight and “with myelofibrosis” as modifiers to the WHO subgroup of MDS [8]. Though further revised in 2016 WHO classification, MDS-F has not yet been established as a distinct entity [2]. MDS-F is currently listed as a type of MDS, according to the World Health Organization Classification of Haematolymphoid Tumours (5th edition) [9]. The clinical features and molecular mechanisms of MDS-F need to be further confirmed.

Moreover, the pathophysiology of MDS-F remains obscure. BM fibrosis in MDS involves transforming growth factor-β secreted by megakaryocytes and platelets [10]. It is considered as a reactive manifestation in MDS. In the past several years, myelofibrosis has been well studied in myeloproliferative neoplasms(MPN). Somatic mutations in MPN, including primary myelofibrosis (PMF), are divided into “drivers” (e.g., JAK2, CALR, and MPL) and “others” (e.g., ASXL1, SRSF2, U2AF1). The drivers bring with MPN phenotypes, whereas the “others” contribute to disease progression and leukemic transformation [11]. Mutations of genes TP53, SETBP1 and U2AF1 are frequent in patients with MDS-F [6, 12], but their clinical significance is not clear. This study aimed to describe the genetic mutations related to MDS-F, and their role in the outcomes of MDS-F.

Subjects and methods

Patient samples

The diagnoses of MDS and leukemic transformation were according to the 2016 World Health Organization (WHO) classification system [2]. This retrospective study screened 920 consecutive treatment-naïve MDS patients at our institution between January 2009 and November 2020. A total of 102 patients who met the criteria for moderate-to-severe bone marrow fibrosis (MF-2/MF-3) and were confirmed as “MDS-F” through a rigorous multidisciplinary review process constituted the final analysis cohort. The diagnosis of MDS-F was established by integrated assessment involving clinical hematologists and hematopathologists, based on clinical presentation, bone marrow morphology/biopsy, cytogenetics, molecular data, and longitudinal follow-up to exclude evolution into classical myeloproliferative neoplasms (MPN) or MDS/MPN overlap syndromes. Prognosis-related risks were evaluated in the Revised International Prognostic Scoring System (IPSS-R) [13]. A history of myeloproliferative neoplasm or myelodysplastic/ myeloproliferative syndrome were set as exclusion criteria. As required by the institutional review board–approved protocols, all subjects offered their informed consent to participate in this study performed under the Declaration of Helsinki.

BM fibrosis assessment

BM aspiration and biopsy were performed to make a diagnosis and evaluate disease progression. BM biopsy specimens were stained with reticulin. Bone marrow fibrosis (MF) was graded from 0 to 3 according to the European Bone Marrow Fibrosis Network (EUMNET) consensus criteria [14]. Specimens demonstrating heterogeneous reticulin deposition patterns were evaluated by determining the predominant histomorphometric pattern. To ensure grading accuracy, all cases classified as MF-2 or MF-3 underwent blinded independent re-evaluation by two certified hematopathologists.

Mutation analysis

Genomic DNA was extracted from whole bone marrow nucleated cells using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). All samples were collected at initial diagnosis and subjected to next-generation sequencing (NGS) on the Illumina platform (San Diego, CA) through a certified clinical laboratory (KingMed Diagnostics, Guangzhou, China). A total of 29 genes were targeted (see Supplementary Table 1 for complete gene list), and all their exons were sequenced. Somatic mutations were filtered by Clinvar, 1000 Genomes, ESP6500, PolyPhen, SIFT and COSMIC database. To ensure accuracy, potential somatic mutations were confirmed through an independent secondary bioinformatics analysis of the raw sequencing data to minimize technical false positives.

Statistical analysis

All data were processed in the SPSS software (version 22.0) (SPSS, Inc.). Categorical variables were compared using χ² test or Fisher’s exact test as appropriate. Survival curves were generated via Kaplan-Meier methodology with comparisons performed by stratified log-rank test. Overall survival (OS) was calculated over a period from diagnosis to the most recent follow-up or death, and leukemia-free survival (LFS) over a period from diagnosis to AML evolution or death. Transplant recipients (n = 2) were censored at hematopoietic stem cell transplantation (HSCT) initiation. Univariate Cox proportional hazards regression identified prognostic covariates (p < 0.1 threshold), which were subsequently incorporated into multivariate models. A prognostic scoring system was developed based on β coefficients from the final multivariate Cox model, weighted as: Score = Σ(β_i × X_i). Its predictive performance was evaluated with concordance-index (C-index). A P value was 2-sided and considered significant at 0.05.

Results

Clinical characteristics of patients with MDS-F

We analyzed a consecutive cohort of 102 treatment-naïve MDS patients with bone marrow fibrosis (MF-2/MF-3) (59 males, 43 females; median age 63 years [range 27–80]) (Table 1). Fibrosis distribution spanned all MDS subtypes (Fig. 1A): including 6 (5.9%) MDS with single lineage dysplasia (MDS-MLD), 2 (2.0%) MDS with isolated del (5q), 3 (2.9%) MDS with RS and single lineage dysplasia (MDS-RS-SLD), 4 (3.9%) MDS with RS and multilineage dysplasia (MDS-RS-MLD), 52 (51.0%) MDS with multilineage dysplasia (MDS-MLD), 25 (24.5%) MDS with excess blasts-1 (MDS-EB-1), and 10 (9.8%) MDS with excess blasts-2 (MDS-EB-2). Cytogenetic data were available in 101 patients. The distribution of karyotypes is shown in Fig. 1B. The abnormal karyotype was detected in 52 (51.5%), complex karyotype (≥ 3 alterations) in 20 (19.8%), monosomic karyotype in 11 (10.9%), i17 karyotype in 2 (2.0%), and − 7 (1.0%) in 1 patient. In addition, trisomy 8 and del (20q-) were detected in 6 individuals, respectively.

Table 1.

The characteristics of patients with MDS-F

Variable	Total N = 102	MF-2 N = 63(%)	MF-3 N = 39(%)	P value
Median age	63	64	60	0.160
Sex				0.316
Male	59	34 (54.0)	25 (64.1)
Female	43	29 (46.0)	14 (35.9)
MDS classification				0.630
MDS-SLD/5q	8(7.8)	4 (6.3)	4 (10.3)
MDS-RS	7(6.9)	6 (9.5)	1 ( 2.6)
MDS-MLD	52(51.0)	32 (50.8)	20 (51.3)
MDS-EB-1/2	35(34.3)	21 (33.3)	14 (35.9)
Cytopenia
Hemoglobin (mean*g/L)	75.7	75.8	75.7	0.992
WBC counts (mean*109/L)	4.2	4.19	4.20	0.997
Platelets counts (mean*109/L)	100.8	104.0	95.6	0.761
Median blast	2.4	3.0	1.6	0.084
IPSS-R				0.738
Very low/Low	36(35.3)	20 (31.7)	16 (41.0)
Intermediate	26(25.5)	16 (25.4)	10 (25.6)
High	23(22.5)	15 (23.8)	8 (20.5)
Very high	17(16.7)	12 (19.0)	5 (12.8)

Fig. 1.

Clinical characters of patients with bone MF. A Disease subtypes according to 2016 WHO classification. B Cytogenetic characteristics. C Overall survivals in patients with different grades of MF. D Leukemia-free survivals in patients with different grades of MF

According to IPSS-R scores, the patients were allocated into five risk groups: 3 (2.9%) in very low risk group, 33 (32.4%) in low risk group, 26 (25.5%) in intermediate risk group, 23 (22.5%) in high risk group, and 17 (16.7%) in very high risk group. Given limited very low-risk cases (n = 3), this subgroup was merged with low-risk for subsequent analyses. Fibrosis severity distribution showed MF-2 in 61.8% (63/102) versus MF-3 in 38.2% (39/102). The clinical features, OS and LFS did not differ between patients with MF-2 and MF-3 (Fig. 1C, D).

Mutational landscape of MDS with fibrosis

Targeted sequencing of 29 myeloid-associated genes revealed somatic mutations in 86.3% (88/102) of patients, with a median mutational burden of 2 variants per case (range 0–7). A detailed list of all mutation sites and their corresponding VAFs were shown in Supplementary Table 2. When integrated with cytogenetic abnormalities (detected in 51.5%, 52/101), the combined genetic alteration rate reached 91.2% (93/102). Notably, 83.3% of cytogenetically normal cases harbored ≥ 1 mutation. The mutation was most frequent in seven genes, including TP53 (19.6%), ASXL1 (16.7%), U2AF1 (14.7%), TET2 (12.7%), DNMT3A (12.7%), RUNX1 (12.7%) and SETBP1 (10.7%) (mutation in > 10% of the cases). The mutation was less frequent (5–10%) in JAK2 (9.8%), EZH2 (8.8%), KIF20B (8.8%), IDH1/2 (8.8%), KMT2D (7.8%), MPL (7.8%), SF3B1 (6.9%), BCOR (6.9%), NRAS/KRAS (6.9%) and GATA2 (5.9%) (Fig. 2A left). When comparing the gene mutation profiles of the MF-2 and MF-3 subgroups, it was found that the U2AF1 mutation was more common in MF-2 patients ( 20.6% vs. 5.1%, p = 0.032, see Supplementary Table 3).

Fig. 2.

Mutational landscape of MDS-F in the cohort. A Mutation frequencies in different subtypes shown in indicated colors (left); Mutation frequencies of genes involved in common functional pathways (right). B Heat map showing the mutational profile detected on next-generation sequencing

Functional annotation classified mutations into seven key pathways: DNA methylation (30.4%, 31/102), RNA splicing (27.5%, 28/102), transcriptional regulation (24.5%, 25/102), chromatin modification (21.6%, 22/102), JAK-STAT pathway (19.6%, 20/102), cell cycle (19.6%, 20/102) and others (9.8%, 10/102) (Fig. 2A right). The landscape of genetic mutations in MDS-F is shown in Fig. 2B.

Clinical and molecular features of patients harboring mutations of genes involved in the JAK-STAT pathway

The genes in the JAK-STAT pathway (JAK2, MPL or CALR) showed mutation in 20 cases, with most being non-hotspot mutations. These mutations were mutually exclusive. JAK2 mutation at p.V617F was found in 5 cases. Other mutation sites in JAK2 included p.E987K, p.L925I, p.R923H, p.R839W and p.C282G. Mutation sites of MPL included p.L265F, p.G266S, p.Y591H(2 cases), p.Y591S, p.W515K, p.A58V and p.L627V. CALR mutations at p.Q365fs and p.L367Tfs were detected. The rate of JAK-STAT pathway gene mutations presented no significant difference between patients with MF-2 and MF-3.

We further compared patients with any mutation of JAK2, MPL or CALR (JAK-STAT MT) to those without it (JAK-STAT WT) (Supplementary Table 4). The MDS-EB1/2 type was less frequent in patients with JAK-STAT (MT) (10.0%) than in those with JAK-STAT(WT) (40.2%, p = 0.012, Fig. 3A). The patients with JAK-STAT pathway gene mutations had similar levels of leukocytes, hemoglobin and platelets, compared with those without JAK-STAT mutations. In addition, the patients with JAK-STAT pathway gene mutations had a lower level of BM blast cells (p = 0.006) and and a lower risk of IPSS-R (p = 0.044), compared with JAK-STAT (WT) subjects (Fig. 3B). Moreover, we compared the mutation landscapes in two groups. The mutations of RUNX1, SETBP1, and IDH1/2 were less common in the JAK-STAT (MT) group, but without statistically significant difference (Supplementary Table 5). The patients with JAK-STAT (MT) had a median OS and LFS (not reached), compared with those with JAK-STAT (WT) who had a median OS of 26 months (HR 0.50, 95% CI 0.26–0.96; p = 0.038, Fig. 3C) and a median LFS of 24 months (HR 0.53, 95% CI 0.28–0.99; p = 0.046, Fig. 3D).

Fig. 3.

Clinical and molecular features of patients with mutations of genes in the JAK-STAT pathway. A Disease classification in cases with or without JAK-STAT mutations. B Distribution of the IPSS-R risk group. C Overall survivals in subgroups with and without JAK-STAT mutations. D Leukemia-free survivals in subgroups with and without JAK-STAT mutations

Prognostic impact of somatic mutations

The follow-up had a median of 30 months (range 1-115 months). The median OS was 28 months and the median LFS was 24 months in the overall cohort. TP53 and IDH1/2 mutations were associated with worse OS and LFS, whereas mutations in MPL were associated with better OS and LFS (Fig. 4). The patients carrying TP53 mutations had a median OS of 11 months and a median LFS of 9 months, both significantly shorter than those of patients without TP53 mutations who had a median OS of 40 months (HR 6.80, 95% CI 2.70-17.11; p < 0.001, Fig. 4A) and a median LFS of 30 months (HR 5.92, 95% CI 2.40-14.58; p < 0.001, Fig. 4D). In the 9 cases with mutations in IDH1 or IDH2, the median OS was 17.5 months (Fig. 4B) and the median LFS was 20 months (Fig. 4E). Unexpectedly, the patients with MPL mutations had longer OS (not reached) and LFS (not reached) than those without MPL mutations who had a median OS of 24 months (HR 0.29, 95% CI 0.13–0.68; p = 0.004, Fig. 4C) and a median LFS of 21 months (HR 0.29, 95% CI 0.13–0.66; p = 0.003, Fig. 4F).

Fig. 4.

Overall survivals and leukemia-free survivals associated with mutations of TP53, IDH1/2 and MPL. A–C Overall survivals in patients with mutations of TP53, IDH1/2 and MPL. D–F Leukemia-free survivals in patients with mutations of TP53, IDH1/2 and MPL

Performance of the prognostic model incorporating mutation data

The results of univariable and multivariable analyses are shown in Table 2. In the univariate analysis, age ( < = 60 vs. >60), IPSS-R score, as well as TP53, IDH1/2 and MPL mutations were significantly associated with OS. In the multivariate analysis, only IPSS-R score (HR 1.39, 95% CI 1.16–1.68), TP53 (HR 2.14, 95% CI 1.06–4.30) and MPL (HR 0.13, 95% CI 0.02–0.96) maintained their associations (p < 0.05). IPSS-R score was combined with TP53 and MPL mutations (IPSS-RM) to stratify the patients more accurately. The score of IPSS-RM was calculated based on the weighted coefficients of these factors: IPSS-R*0.33 + TP53*0.76-MPL*2.0. Four risk groups were constructed: low (score < 0; n = 8), intermediate (score 0 ~ 1; n = 33), high (score 1 ~ 2; n = 39) and very high (score > 2; n = 22) risk groups. The median OS was not reached in low risk subgroup, but 40, 26 and 11 months in intermediate, high, and very high risk subgroups; the counterpart values of median LFS were not reached, 40, 19 and 9 months, respectively. Our model had a better C-index (0.62), compared to IPSS-R (0.59). Moreover, the survival curves generated from IPSS-RM were more separated than those from IPSS-R (Fig. 5).

Table 2.

Univariable and multivariable analysis for OS

Variables	Univariate	Multivariate
	P	P	HR (95%CI)
Age	0.004	0.269	1.421(0.763–2.648)
IPSS-R	< 0.001	0.001	1.393(1.155–1.680)
TP53 MT	< 0.001	0.034	2.136(1.060–4.303)
IDH1/2 MT	0.046	0.115	2.030(0.841-4.900)
MPL MT	0.004	0.047	0.134(0.018–0.975)

Fig. 5.

Overall survivals (A) and leukemia-free survivals (B) in the cohort predicted by the IPSS-R scores. Overall survivals (C) and leukemia-free survivals (D) predicted by the new model incorporating IPSS-R scores and genetic mutations

Discussion

In the current study, MF-2 or MF-3 could present in all subtypes of MDS and are associated with high-frequency complex karyotypes and TP53 mutations. Patients with mutated genes in the JAK-STAT pathway emerged favorable clinical features. In addition, we comprehensively evaluated the prognostic factors of MDS-F. By incorporating IPSS-R and mutations of TP53 and MPL, we developed a model that generated more distinct survival curves for MDS-F, highlighting the potential of integrated molecular profiling.

The significance of bone marrow fibrosis in predicting the prognosis of MDS is still controversial. MDS patients with MF-2 or 3 are usually classified into MDS-F and often have a worse OS than patients with MF0-1 [7, 15, 16]. However, in a recent study, only MF-3 was associated with a worse OS independent of IPSS-R risk in MDS patients [6]. In our study, the clinical characteristics, cytogenetic data, gene mutations, and outcomes of MF-2 patients were identical to those of MF-3 patients. Therefore, it is reasonable to combine patients with MF-2 and MF-3 into one category in related research.

Cytogenetic and mutational profiles can help assess the prognosis of MDS. Fu et al., after analyzing 79 MDS-F patients, reported that the cases of MDS with or without bone MF demonstrated similar distribution of cytogenetic risk and frequency of monosomic karyotype [4]. In another study including 107 patients with MF-3, Melody et al. noted high probabilities of cytogenetic and IPSS-R in patients with MF-3, as shown by that the complex karyotype was more common in cases with MF-3 (31% vs. 16%) than in those with MF1-2 [6]; in addition, more frequent mutation of TP53 and SETBP1 appeared in MF-3 patients. Those results are consistent with ours in the present study: the abnormal karyotype was observed in 51.5%, the complex karyotype in 19.8% and the monosomic karyotype in 10.9% of patients. Consistent with cytogenetic results, high-frequency TP53 mutations were detected in patients of MDS-F. It has repeatedly reported the links of TP53 mutations with complex karyotype, elevated blasts, severe thrombocytopenia, rapid transformation to acute myeloid leukemia, and poor prognosis in MDS [17–20]. Therefore, these genetic findings are closely correlated with the poor prognosis of MDS-F.

Other easily mutated genes detected in our cohort included TET2, ASXL1, U2AF1, DNMT3A, RUNX1, SETBP1, JAK2 and MPL. Most of them mutate in MDS patients and have been extensively studied. Surprisingly, JAK2 and MPL mutations were found in 9.8% and 7.8% of MDS-F patients, respectively; only two patients harbored CALR mutations. According to functional pathway analysis, the genes in the JAK-STAT pathway mutated in 19.8% of patients. The JAK-STAT pathway can be activated due to JAK2, CALR or MPL mutations, a process prominently involved in the development myeloproliferative neoplasm. However, mutations in JAK-STAT pathway genes are rare in MDS patients. In the entire cohort of MDS, JAK2 mutation occurs in about 2–5% of patients [21, 22]. Recently, JAK2 was reported to mutate in 2.5% of patients with lower-risk MDS, and associated with shorter OS after adjustment for IPSS [23]. In the subjects with therapy-related MDS, Fu et al. revealed more frequent JAK2 V617F variation in patients with MF (21% vs. 2%) than in those without MF [24]. In another study, JAK2 V617F mutation was identified in 6/28 (21%) of primary MDS patients with bone MF, compared to 1/41 (2%) in those without [4]. The relatively high frequency of JAK-STAT pathway mutations (19.8%) in our rigorously defined MDS-F cohort, compared to general MDS, likely reflects intrinsic biological heterogeneity rather than diagnostic impurity. Notably, the majority of these mutations were non-canonical variants (e.g., 5/10 JAK2 and 7/8 MPL mutations were non-hotspot), a profile distinct from that of classical MPN. This supports the hypothesis of an ‘MPN-like molecular subset’ within MDS-F, where alternative activation of the JAK-STAT pathway, possibly through different mechanisms or with different consequences, contributes to the fibrotic phenotype within a predominantly dysplastic framework.

Clinically, we further found that the patients with JAK-STAT pathway gene mutations tended to have a lower level of bone marrow blasts, a lower frequency of MDS-EB subtype, and more favorable outcomes. Nevertheless, these results should be tested in larger-sample studies, but we still speculate that JAK-STAT pathway gene mutation may be a myeloproliferative feature of MDS-F.

The clinical prognosis varies a lot in patients with MDS. Several somatic mutations are associated with the prognosis of MDS [21, 22]. However, the prognostic value of gene mutations for the clinical outcomes of MDS-F remains unclear. In the current study, we for the first time identified that MPL mutations were associated with favorable OS and LFS in univariate analysis and showed a strong protective association in multivariate modeling. MPL, from the type 1 homodimeric cytokine receptor family, acts on megakaryopoiesis by regulating platelet production in response to TPO. MPL mutations are the least frequent “drivers” in MPN, with approximate incidences of 3% in ET and 10% in PMF [25]. Intriguingly, the association of MPL mutations with favorable prognosis in MDS-F contrasts with observations in classical MPN (e.g., PMF), where MPL is a driver linked to fibrotic progression but not typically an independent poor prognostic factor [26, 27]. This discrepancy suggests that the same driver mutation may exert distinct biological effects and clinical implications in different disease contexts (MDS vs. MPN). In MDS-F, MPL mutation might mark a more indolent subtype with prominent fibrotic reaction but lower leukemic potential. However, given the limited number of MPL-mutant cases (n = 8) in our study, this observation is exploratory and must be validated in larger cohorts.

Integrating gene mutations into IPSS-R can improve the stratification of MDS patients in several prognostic models [28–30]. These models differ in disease subtype and treatment background. We identified TP53 and MPL mutations as most significant in univariate and multivariate analyses. As expected, fusing these mutations with IPSS-R score enhanced its predictive ability. The new model generated a high C-index and a clearer survival curve than IPSS-R. One limitation is that our model is derived from a single institution. A validation should be performed in a larger and multicenter sample of MDS-F patients. Another limitation of this study is the small sample size of the MPL-mutant subgroup. Although it retained independent prognostic significance in multivariable analysis, the observed strong protective effect should be interpreted with caution due to potential overfitting in a small cohort. Therefore, this novel finding is exploratory and hypothesis-generating, requiring validation in larger, prospective, multicentric cohorts.

In conclusion, genetic lesions are common in patients of MDS-F. Our data suggest a potential association of MPL mutations with more favorable clinical outcomes in this rigorously defined MDS-F cohort. The prognostic model combining IPSS-R score and gene mutations, including this potential MPL effect, represents an innovative approach to better stratify patients with MDS-F, although it requires validation in prospective, independent studies.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Y.Z. designed the research, collected the data and wrote the paper; C.C. and J.G. designed the research study and reviewed the paper; S.Z., R.W. and L.S. collected and analyzed the data.

Funding

This study was supported by the Shanghai Science and Technology Committee(21Y31920400) and the Clinical Research Plan of SHDC (NO. SHDC2024CRI057).

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information. Detailed gene mutation information are provided in Supplementary Table 2.

Declarations

Ethics approval

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Shanghai Sixth People’s Hospital (protocol code 2021-033, date of approval 25 February 2021).

Consent to participate

Informed consent was obtained from all individual participants or their legal guardians included in the study.

Consent for publication

All authors agree to publish.

Competing interests

The authors declare no competing interests.