CH prevalence and mutation profile in ICUS patients.
A total of 281 ICUS patients were included in the study, among which 129 cases were female and 152 cases were male, with a median age of 55 years (range 12–87 years). As for the cytopenia types, the number of patients with SLC, BC, and PC was 35, 85, and 161, respectively. The clinical characteristics and blood routine indices were shown in Table 1. Among the included ICUS patients, 76 cases (27.0%) were found to harbor CH (Fig. 1A). The detected mutations were enriched in the genes of chromatin modifier, DNA methylation, signaling, splicing and transcription. As for the mutation types, missense mutation, frameshift insertion, frameshift deletion, nonsense mutation, in-frame insertion, and in-frame deletion were detected, and missense mutations were the most prevalent, accounting for more than half of the mutation types (Fig. 1B). It has been established that CH accumulates with age, and the incidence often shows a gender disparity in general population [17]. Accordingly, we also found that patients with CH were significantly older than those without CH (P = 0.001; Fig. 1C), and the percentage of CH in male patients was much higher than that in female patients (36.84% vs. 15.50%, P < 0.001; Fig. 1D). Moreover, the cumulative incidence of CH increased progressively with age in ICUS patients (Fig. 1E).
Table 1
Clinical characteristics of ICUS patients.
Variable
|
Total patients
|
CH
|
Non-CH
|
P value
|
Number of patients
|
281
|
76
|
205
|
-
|
Age (years)
|
53 ± 17
|
58 ± 16
|
51 ± 17
|
0.001
|
Gender (F/M)
|
129/152
|
20/56
|
109/96
|
< 0.001
|
WBC count (× 109/L)
|
3.23 ± 2.00
|
3.25 ± 2.12
|
3.22 ± 1.95
|
0.872
|
NEU count (× 109/L)
|
1.71 ± 1.59
|
1.77 ± 1.76
|
1.69 ± 1.52
|
0.587
|
LYM count (× 109/L)
|
1.22 ± 0.63
|
1.09 ± 0.45
|
1.26 ± 0.67
|
0.003
|
HGB (g/L)
|
87.5 ± 28.9
|
86.2 ± 31.1
|
88.01 ± 28.1
|
0.448
|
PLT count (× 109/L)
|
54.02 ± 62.63
|
67.32 ± 65.63
|
49.10 ± 59.72
|
0.005
|
MCV (fL)
|
103.12 ± 10.94
|
102.04 ± 10.60
|
103.52 ± 11.06
|
0.145
|
RET (× 109/L)
|
58.42 ± 33.54
|
52.74 ± 34.97
|
60.49 ± 32.89
|
0.033
|
Abbreviations. ICUS: idiopathic cytopenia of unknown significance; CH: clonal hematopoiesis; WBC: white blood cell; NEU: neutrophil; LYM: lymphocyte; HGB: hemoglobin; PLT: platelet; MCV: mean corpuscular volume; RET, reticulocyte. |
Mutated genes and loci in ICUS patients.
Mutations were observed in 20 genes, including ASXL1, EZH2, PHF6, BCOR, STAG2, TET2, DNMT3A, IDH2, JAK2, U2AF1, SF3B1, SRSF2, CEBPA, ETV6, RUNX1, WT1, CBL, NPM1, SETBP1 and TP53. Among these genes, ASXL1, DNMT3A, U2AF1, and TET2, were the most frequently mutant genes, followed by BCOR, SF3B1, NPM1, RUNX1, SRSF2, PHF6, and IDH2 (Fig. 2A). Among the 76 ICUS patients with CH, 51 patients (67.11%) had 1 mutated gene, 15 cases (19.74%) harbored 2 different mutated genes, 9 cases (11.84%) with 3, and 1 patient (1.32%) with 4 (Fig. 2B). The VAF of mutant genes varied from 2.10–39.99%, and IDH2 mutation had the highest VAF while SF3B1 mutation had the lowest VAF among mutated genes (Fig. 2C). ASXL1, DNMT3A, U2AF1, TET2, BCOR, SF3B1, NPM1 and RUNX1 were the top 8 most frequently mutated genes, and their mutation sites and mutation types were plotted in detail in Fig. 2D. Mutant loci in most mutated genes, such as AXLS1, TET2, and DNMT3A, occurred in the hotspot regions dispersedly distributed across the whole gene. By contrast, U2AF1 mutations in ICUS were concentrated in hotspots of S34 and Q157. These genetic mutations caused the changes in protein structures or functions, which might be related to the development of hematological malignancies.
Ch Patients Had Different Clinical Characteristics Compared With Non-ch Patients
The relationship between blood indices and CH status in treatment-naive ICUS patients was also analyzed. It was observed that the LYM counts in CH group were significantly lower than those in non-CH group (1.09 ± 0.45 ⋅ 109/L vs. 1.26 ± 0.67 ⋅ 109/L, P = 0.003; Fig. 3A). By contrast, the PLT counts in CH group were significantly higher than those in non-CH group (67.3 ± 65.6 ⋅ 109/L vs. 49.1 ± 59.7 ⋅ 109/L, P = 0.005; Fig. 3B). Results also showed that the RET counts in CH group were significantly lower than those in non-CH group (52.7 ± 35.0 ⋅ 109/L vs. 60.5 ± 32.9 ⋅ 109/L, P = 0.033; Fig. 3C). No statistical difference was observed in WBC, NEU, HGB or MCV between CH and non-CH group (all P > 0.05). We further investigated CH rates in the 3 subgroups of ICUS patients. No significant difference in CH incidence was found among patients with SLC, BC and PC (P = 0.347; Fig. 3D). Interestingly, when we focused on the SLC group, patients of ICUS-N had a slightly higher percentage of CH than those of ICUS with thrombocytopenia (ICUS-T), though statistical difference was not reached (75.0% vs. 19.0%, P = 0.053; Fig. 3E).
We also analyzed bone marrow dysplasia and blast percentages in the include patients. There was no statistical difference in CH prevalence between patients with dysplasia and those without dysplasia (28.7% vs. 25.4%, P = 0.532; Fig. 3F). Moreover, VAF of the mutated genes between ICUS patients with dysplasia and these without dysplasia also showed no significant difference (18.6 ± 12.9% vs. 20.7 ± 13.8%, P = 0.507; Fig. 3G). By contrast, percentage of patients with CH was significantly higher in patients with blasts than those without blasts (50.0% vs. 24.9%, P = 0.008; Fig. 3H), while statistical elevation in VAF of the patients with blasts was not achieved compared with those without blasts (22.7 ± 13.5% vs. 18.9 ± 13.2%, P = 0.412; Fig. 3I).
Mutated Genes Were Associated With Abnormal Blood Indices In Ch Patients
For patients with CH, we further assessed the association between mutated genes and blood indices. Using univariate logistic regression analysis, we found that EZH2 mutation was significantly associated with PLT counts (OR = 1.823 (-0.051–3.702), P = 0.050). Moreover, multiple logistic regression analysis showed that BCOR mutation was significantly associated with lowered LYM counts (OR = 23.551 (1.666–332.906), P = 0.019) after adjustment for sex and age. Pearson coefficient also revealed that EZH2 mutation was associated with decreased PLT count (P < 0.001; Fig. 4). Negative correlation was also found between IDH2 and HGB, IDH2 and MCV, SRSF2 and MCV, CBL and MCV, respectively (all P < 0.05).
As most genes function synergistically or counteractively with other genes, we further analyzed the co-mutations and exclusive mutations between different genes. The results showed that DNMT3A mutation and U2AF1 mutation were mutually exclusive (P = 0.031), while SRSF2 was strongly co-mutated with TET2 and WT1 (P = 0.003 and P = 0.024; Table 2).
Table 2
Co-mutations and exclusive mutations between genes.
Gene 1
|
Gene 2
|
WT/WT
|
WT/M
|
M/WT
|
M/M
|
P
|
|
DNMT3A
|
U2AF1
|
45
|
14
|
17
|
0
|
0.031
|
exclusive
|
SRSF2
|
TET2
|
64
|
0
|
9
|
3
|
0.003
|
co-occurring
|
SRSF2
|
WT1
|
41
|
0
|
0
|
1
|
0.024
|
co-occurring
|
Ch Conferred Higher Malignant Transformation And Lower Survival In Icus Patients
It was reported that gene mutations promoted disease progression in hematological malignancies such as myeloproliferative neoplasm (MPN) [18], while their role in ICUS transformation remains unknown. In the present study, we explored the role of CH in the transformation of ICUS. During the follow-up of our cohort, 15 of the 281 ICUS patients (5.34%) transformed to hematological malignancies, including 7 patients in CH group and 8 patients in the non-CH group. The transformation rate in patients with CH was significantly higher compared to that in non-CH patients (46.67% vs. 22.58%, P = 0.039; Fig. 5A). More specifically, 9 patients progressed into MDS, and 6 cases transformed into AML. The transformation rates from ICUS into MDS between CH and non-CH group showed no statistical difference (7.14% vs. 4.69%; P = 0.691); by contrast, CH patients showed a remarkably increased incidence of leukemia transformation compared with non-CH patients. (9.52% vs. 1.56%, P = 0.033; Fig. 5B).
We further analyzed the OS and PFS of included patients using Kaplan-Meier analysis. The results showed that ICUS patients with CH had a shorter OS compared to non-CH patients (P = 0.028; Fig. 5C). Moreover, ICUS patients with CH also presented a shorter PFS in comparison with non-CH patients (P = 0.006; Fig. 5D).
Mutation profile in the malignant transformation of ICUS
Among the 20 mutated genes, ASXL1, SETBP1, DNMT3A, IDH2, U2AF1, SF3B1 and TP53 mutations were present in both transformed and non-transformed patients, while ETV6 mutation was detected only in transformed patients. Missense mutations were the main mutation type in transformed patients (Figure 6A). Other mutation types, such as frameshift insertion and frameshift deletion, were also detected in transformed patients. All of the mutations in patients with leukemic transformation were missense mutations, while frameshift mutations were also present in patients with MDS transformation (Figure 6B). It was notable that the incidence of SETBP1, IDH2, U2AF1, SF3B1, ETV6 and TP53 mutations in transformed patients was higher compared with non-transformed cases (Figure 6C), suggesting the potential predictive value of specific gene mutations in disease transformation.
To further investigate the roles of mutations in transformation, we followed 6 patients who performed NGS after MDS transformation, including 3 patients from the CH group and 3 patients from the non-CH group. The follow-up periods of these 6 patients ranged from 6 months to 26 months. Three patients in non-CH group showed new emergence of TET2, AXSL1, or PDGFRA mutations after MDS transformation (Figure 6D-F). The other 3 patients in CH group also showed VAF changes in already existing mutations or emergence of new mutations after disease transformation. The first patient had a relative stable level of ASXL1 mutation, elevated ETV6 mutation, and emergence of CBL and NRAS mutations during follow-up (Figure 6G). By contrast, U2AF1 and TP53 mutation VAFs of the second patient increased after MDS transformation (Figure 6H). MDS transformation of the third patients was accompanied by a reduction in U2AF1 mutation VAF and new occurrence of RUNX1 mutation (Figure 6I).