The prognostic value of the interaction between ASXL1 and TET2 gene mutations in patients with chronic myelomonocytic leukemia: a meta-analysis

ABSTRACT

Purpose

The prognostic role of TET2 and/or ASXL1 mutations which are common gene mutations in chronic myelomonocytic leukemia (CMML) remains controversial. Therefore, we conducted this meta-analysis to evaluate the prognostic efficacy of ASXL1 and TET2 mutations in CMML population.

Methods

PubMed, Cochrane and Embase for relevant research were employed to identify 16 studies. Overall survival rate (OS) with hazard ratios (HRs) was used for analysis, and each individual HR was applied to calculate the combined HR.

Results

The total HR of OS was 0.74, 95% CI = 0.61 - 0.91, P = 0.005, compared with CMML patients without TET2 mutations (TET2^MT), and the total HR of OS was 1.56, 95% CI = 1.34 - 1.80, P = 0.000, compared with CMML patients without ASXL1 mutation (ASXL1^WT), indicating that TET2^MT and ASXL1^WT were favorable for prognosis of CMML. According to whether the gene is mutated or not, the acute transformation rate of disease and mortality rate were further considered for assessment. Compared with the CMML patients with TET2^MT and ASXL1^WT, the HR of patients with in both TET2^MT and ASXL1^MT was 1.51 (95% CI = 1.14 - 1.99; P = 0.004), the HR of patients with neither TET2^MT nor ASXL1^MT was 1.49 (95%CI = 1.12 - 1.98; P = 0.007), and the HR of TET2^WT and ASXL1^MT patients was 1.88 (95%CI = 1.21 - 2.94; P = 0.005).

Conclusion

Presence of TET2^MT and ASXL1^WT genotype was the most beneficial for the survival of CMML patients.

KEYWORDS:

• Chronic myelomonocytic leukemia
• TET2
• ASXL1
• meta-analysis

1. Introduction

Chronic myelomonocytic leukemia (CMML) is a disease with overlapping features of myelodysplastic syndromes (MDS) and myeloproliferative neoplasm (MPN), which average median survival time for patients is about 30 months. The inherent risk of leukemia transformation is about 15-20% within 3–5 years [1]. In the guideline of World Health Organization (WHO) for myeloid neoplasms classification revised in 2016, CMML is defined as a clonal hematopoietic stem cell disease characterized by the presence of sustained (>3 months) peripheral blood (PB) mononucleosis (>1×10⁹/L, with monocytes≥10% of the white blood cell count) and myelodysplastic (BM) [2]. According to diverse clinicopathology, it is divided into 2 types: CMML type 1 (CMML-1) with less than 10% medullary and less than 5% peripheral blasts, and CMML type 2 (CMML-2) with 10% to 19% medullary and/or 5% to 19% peripheral blasts [3]. Hypomethylating agents (HMA) and therapeutic allogeneic blood or bone marrow transplantation (BMT) are still the most available standard treatment regimens. However, BMT is a rarely feasible treatment option because of the age and/or the complications in CMML patients [4, 5]. And also therapeutic drugs such as HMA cannot significantly change the natural history of CMML, therefore, effective and available treatments or strategies need to be upgraded [6].

Previous studies have shown that there are gene mutations in more than 90% of CMML cases [7]. In recent years, with the development of next-generation sequencing technology, new somatic mutations might play an important role in the pathogenesis and clinical prognosis of CMML. Epigenetic regulators (TET2∼60%), modulators of chromatin (ASXL1∼40%) are often involved in CMML [8]. TET2 (ten-eleven translocation (TET) oncogene family member 2—chromosome 4q24), a member of the TET protein family (TET1-TET3), is the only gene frequently mutated in the TET family in medullary malignancies, especially in CMML. TET2^MT (TET2 mutation) inhibits the hydroxymethyl catalytic activity of epigenetic enzymes and can lead to low levels of 5-hmc in genomic DNA, which will disrupt hematopoietic differentiation [8–11]. Some studies reported there was no significant relationship between TET2 mutation and overall survival (OS) in overall CMML population [12–15]. However, a few studies also discovered that the OS of CMML-1 patients with TET2^MT was shortened, while the OS of TET2^MT patients was significantly better than that of TET2^WT (TET2 wild type) patients [12]. Therefore, the prognostic effect of TET2 mutation in CMML still remains controversial.

The additional sex combs like 1 (ASXL1) gene involved in modulators of chromatin is located on human chromosome 20q11 and encodes a highly conserved protein [13, 16]. Similarly, the ASXL1 gene is also one of the most common mutated genes in malignant myeloid diseases. It was reported that myeloid tumors with the highest frequency of ASXL1 gene mutations were also CMML in MDS/MPN [14]. ASXL1 mutation is also related to the acute transformation of CMML. Patients with ASXL1^MT (ASXL1 mutation) are prone to transform into acute myeloid leukemia (AML) [15, 17]. And this AML caused by CMML is a high-risk subtype, usually associated with poor prognosis and the abnormal changes related to myeloproliferation [18]. There were studies reporting that ASXL1 gene mutation was associated with a lower survival rate and decrease in CMML [19, 20]. It has been reported that 9 patients treated with hypo methylation medication during follow-up period, the prognosis of ASXL1^MT was poor among 79 CMML patients [21]. However, data reported by MM Patnaik et al. showed that ASXL1^MT had no effect on OS or leukemia-free survival in patients with CMML [22]. But there were a few studies indicating that ASXL1 has no relationship with prognosis. So, ASXL1, as an independent prognostic factor of CMML, was to be suspected.

Then, how about TET2 and ASXL1 combinative efficacy? The studies have combined two more representative genes in CMML disease and found that there was no obvious interaction between TET2 and ASXL1 status [21]. However, there were some studies supporting that CMML patients with both ASXL1^WT (ASXL1 wild type) and TET2^MT has a better prognosis. [23–26]. Therefore, we evaluated the efficacies of TET2 and/or ASXL1 mutations for prognosis prediction of CMML, and also accessed whether the potential interaction between TET2 and ASXL1 played as beneficial predictor in CMML. In this study, we conducted the meta-analysis focused on fully understand the optimal strategies for the clinical prognosis of CMML.

2. Methodology

2.1. Study selection

The study was carried out based on the PRISMA statement of systematic reviews and meta-analysis [27]. Before searching the literature, all members had confirmed and approved the research protocol. Until March 18, 2021, we have searched electronic databases including PubMed, Cochrane (Cochrane database of systematic reviews) and Embase for relevant research. We used the following MeSH words and free words which were combined with Boolean operations: (‘TET2’ and/or ‘ASXL1’) and (‘CMML’ or ‘Chronic myelomonocytic leukemia’). The search was limited to human research and there were no language restrictions. Search the qualified articles from the above-mentioned databases and obtain additional articles by manually searching references of related studies.

The search strategy received 1152 literature. Literature was included in this meta-analysis if they met the following criteria: (1) Provide detailed information of patients with ASXL1 and/or TET2 mutations. (2) According to ASXL1 and/or TET2 gene mutations, patients in the original study were clearly grouped. (3) Reported survival results or hazard ratio (HR) of clinicopathological data were based on the expressions of mutated and unmutated ASXL1 and/or TET2. (4) The presented survival curve or other sufficient data could be used to calculate 95% CI of HR. Literature was excluded if one of the following criteria existed: (1) In vitro or laboratory studies; (2) Overlapping studies; (3) Review or case reports; (4) Letters and comments to editor; (5) Studies with unclear groupings; (6) Studies with no obvious prognostic indicators. Two reviewers had screened by reviewing the title and abstract first, then reviewing the full text, and independently evaluated the full text of the selected references according to the inclusion or exclusion criteria. Disagreements were resolved through discussion, or by a third reviewer when needed. 16 articles were eventually included in the study. The specific process is shown in Figure 1.

Figure 1. Flow chart based on PRISMA guidelines.

2.2. Date extraction

Two reviewers used standardized abstract forms to independently extract data from each included study. Data extraction was discussed. If there were disagreements between the observers, the other two reviewers would record the decision. All disagreements were resolved through discussion and consensus (Table 1). The extracted information included: first Author’s name, year of publication, study design, Journal, region, total patients, mean age, gender distribution of patients, mutated gene, number of mutations, criteria for classification of CMML, therapeutic method and outcome (OS).

Table 1. Summary of the data extracted from the 16 studies included.

Study Cohort	Year	Journal	Region	Patients (n)	Mean ages	Male/(%)	Mutated gene	Mutations (n)	Criterion	Therapy	Outcome	NOS score
Veronique Gelsi-Boyer [17]	2010	Br J Haematol	France	53	NR(41-88)	35/(66)	A	25	FAB	C/U	OS	7
MM Patnaik [22]	2013	Leukemia	US	226	71(20-90)	152/(67)	A	87	WHO	–	OS	7
Raphael Itzykson [21]	2013	J Clin Oncol	US	312	74(41-93)	210/(67)	A	125	WHO	C	OS	7
MM Patnaik [19]	2014	Leukemia	US	466	73(20-93)	309/(66)	A	164	WHO	–	OS	7
David A Sallman [20]	2017	Blood Cancer J	US	60	74(48-94)	41/(68)	A	28	WHO	C/S	OS	8
Yajuan Cui [28]	2015	Exp Hematol Oncol	China	145	63(18-85)	98/(68)	A	65	WHO/FAB	C/S	OS	7
Sandrine Niyongere [29]	2018	Leukemia	US	215	74(29-91)	139/(65)	A	49	WHO/FAB	–	OS	7
Hyun-Young Kim, M.D. [30]	2018	Ann Lab Med	Korea	36	71(38-83)	26/(72.2)	A	18	WHO	C	OS	7
MM Patnaik [31]	2016	Blood Cancer J	US	261	70(28-91)	168/(65)	T	109	WHO	–	OS	7
Olivier Kosmider [32]	2009	Haematologica	France	88	76(54-93)	61/(69)	T	44	WHO	C	OS	7
Alexander Kohlmann [3]	2010	J Clin Oncol	Germany	81	72.8(40-85.5)	–	T	36	WHO	–	OS	8
Thorsten Braun [12]	2011	Blood	France	39	71(54-88)	30/(77)	T/A	13/19	WHO	C	OS	8
MM Patnaik [23]	2016	Blood Cancer J	US	175	70(18-90)	116/(66)	T/A	80/82	WHO	–	OS	7
Yajuan Cui [33]	2016	Stem Cell Invest	China	141	63(16-85)	95/(67)	T/A	46/65	WHO	–	OS	8
Matthieu Duchmann [25]	2018	EBioMedicine	France	174	72(66-78)	118/(68)	T/A	66/75	WHO	C	OS	8
Giacomo Coltro [26]	2019	Leukemia	US	1084	72(18-95)	714/(68)	T/A	604/442	WHO/FAB	C	OS	8

Study design is case control study. A: ASXL1, T: TET2, C: Chemotherapy, U: Untreated, S: Surgery, WHO: World Health Organization, FAB: French-American-British, NR: not reported

For the extraction of survival data, HR and 95% CI were used directly from the second priority of multivariate survival analysis or univariate analysis. Otherwise, if the study only provides Kaplan-Meier curves, we would use Engauge Digitizer version 10.7 for extraction. Simultaneously, we also tried to contact the corresponding author to make up for the missing data.

2.3. Quality assessment

The study used Newcastle-Ottawa quality assessment (NOS) to assess the quality of non-randomized studies in the meta-analysis (Ottawa, Canada: Department of Epidemiology and Community Medicine, University of Ottawa). It also independently extracted the necessary information and verified the extracted information. In the NOS case–control study standard, there are 8 items in total, which were divided into three main categories: selection (4 items), comparability (1 item) and exposure (3 items). Among the 10 criteria evaluated, 10 stars can be marked. If these studies meet a standard, they will get a star (one point), otherwise they will not. The total score is 10, and studies with a total score greater than 6 are considered as high quality. The above was all independently evaluated by four reviewers, and any disagreement will be resolved by final consensus. The results are shown in Table 1.

2.4. Statistical analysis

Stata version 14.0 (StataCorp, University Town, Texas, USA) was used for data analysis. Compared HRs and 95% CIs evaluation with wild type, the effects of ASXL1 and/or TET2 mutations on prognosis of CMML patients could be shown. We also performed a subgroup analysis of CMML patients, using I² to evaluate the statistical heterogeneity of the efficacy (The degree of heterogeneity: I²=  0-25%, no heterogeneity. I²=  25-50%, moderate heterogeneity. I²=  50-75%, high heterogeneity. I²=  75-100%, extremely poor heterogeneity). Generally, when the heterogeneity was moderate or below, the fixed effects model would be used to calculate the data. Otherwise, the random effects model would be used. But comprehensive consideration should be done in combination with clinical heterogeneity and methodological heterogeneity. When substantial heterogeneity was detected, our assessment was only based on random effects models. Funnel plots, Begg’s and Mazumdar’s and Egger's s-tests were used to assess possible publication bias. The statistical significance was set at P < 0.05. The reported results conformed to the AMSTAR (the Assessment of Multiple Systematic Reviews) guidelines.

For prognostic-related indicators that cannot be assessed with forest plots directly, the two rates of each result can be combined. The two rates are P_1i = E_1i/n_1i, P₂ = E_2i/n_2i (E: events, the number of people who had the event, n: the total number of people in the group). The combined rate is P_i= (E_1i+E_2i) / (n_1i+n_2i), d_i = P_1i-P_2i. The mean and estimated variance of the mean are: (1) $\overline{d}=\mathrm{\Sigma }{W}_i{d}_i/\mathrm{\Sigma }{W}_i$(1) (2) $S^{2}d=\mathrm{\Sigma }{W}_i{P}_i\left(1-P_i\right)/(\mathrm{\Sigma }{W}_i{)}^2$(2) (3) $95\text{\%}Cl=\overline{d}\pm 1.96S_{\overline{d}}$(3)

According to Cochrane's definition [34], the weights in formulas (1) and (2) are: $W=n_{1i}{n}_{2i}/\left(n_{1i}+n_{2i}\right)$

3. Results

3.1. The effects of TET2 mutation on prognosis in CMML patients

A total of 5 studies involving TET2 gene mutation status reported OS (From random allocation to death due to any cause), 643 patients was included. The results of OS subgroup analysis were shown as below: for CMML patients with TET2^MT, HR = 0.74, 95% CI = 0.61 - 0.91, P = 0.005 (Figure 2), suggesting the better prognosis was found in the CMML patients with TET2^MT, compared with TET2^WT. The heterogeneity test results were P = 0.760, I²= 0%, suggesting that there was no heterogeneity in the result of this part. Therefore, it was speculated that TET2 mutation indicated a better prognosis in CMML patients.

Figure 2. Forest plot of the risk ratio (HR) and 95% confidence interval of the overall survival (OS) of patients with TET2 mutation and non-mutation CMML. The size of the square or square represents the weight, and the length of the line represents the width of the 95% CI.

Sample size and analysis method were also considered in the further subgroup analysis. As shown in Table 2, TET2 mutation did not indicate a correlation with prognosis analyzed with multivariate analysis in the subgroup further study. However, it might be biased for the further analysis due to the insufficient sample size.

Table 2. Subgroup analysis of pooled HRs for OS with or without TET2 mutation.

Categories	Studies (n)	Patients (n)	Pooled HR (95%CI)	Heterogeneity
				I²	P-value
Gene type
TET2 mutation	5	643	0.74(0.61 - 0.91)	0.0%	0.760
Sample size
≥100	2	435	0.75(0.61 - 0.93)	0.0%	0.699
＜100	3	208	0.68(0.33 - 1.40)	0.0%	0.439
Analysis method
Univariate	2	435	0.75(0.61 - 0.93)	0.0%	0.699
Multivariate	3	208	0.68(0.33 - 1.40)	0.0%	0.439

Among studies involving TET2 mutation or not, there were three studies respectively reported the relationship between the acute conversion rate and mortality of the disease and the gene status (Table 3). We performed the combination of the difference between the two rates (Incidence of events with or without mutation) (Table 4) and the consistency test. For the acute conversion rate, the estimated variance S²d is 0.000367, and the calculated weighted mean d is 0.0845 (95% CI = 0.0654 - 0.104). For mortality, the estimated variance S²d is 0.000659, and the calculated weighted mean $\overline{d}$ is 0.133 (95% CI = 0.108 - 0.159). The above suggests that TET2 mutation had advantages.

Table 3. The effect of TET2 mutation status on the acute transformation and death rate of CMML patients.

Event	TET2 Mutation		TET2 Wild Type		Total
	E/M	E Rates	E/W	E Rates	E/Total	E Rates
Acute Transformation
Giacomo Coltro [26]	85/604	0.14	103/480	0.21	188/1084	0.17
MM Patnaik [31]	4/109	0.04	33/152	0.22	37/261	0.14
MM Patnaik [23]	11/80	0.14	14/95	0.15	25/175	0.14
Death
Giacomo Coltro [26]	234/604	0.39	275/480	0.57	509/1084	0.47
Alexander E. Smith [35]	33/55	0.60	137/300	0.46	170/355	0.48
MM Patnaik [23]	62/80	0.78	84/95	0.88	146/175	0.83

E: Events, M: Mutation, W: Wild type

Table 4. Combining Table 3 to calculate the difference between the two rates (Incidence of events with or without mutation) in different events.

	d	p	p(1-p)	wi	wi*d	wi*p(1-p)
Acute Transformation
Giacomo Coltro [26]	0.07	0.17	0.14	267.45	19.76	38.34
MM Patnaik [31]	0.18	0.14	0.12	63.48	11.45	7.72
MM Patnaik [23]	0.01	0.14	0.12	43.43	0.43	5.32
Total				374.36	31.64	51.38
Death
Giacomo [26]	0.19	0.47	0.25	267.45	49.61	66.62
Alexander E. Smith [35]	−0.14	0.48	0.25	46.48	−6.66	11.60
MM Patnaik [23]	0.11	0.83	0.14	43.43	4.74	6.00
Total				357.36	47.69	84.22

d: Difference between two rates, p: Total rate, w_i: Weights

3.2. The effect of ASXL1 mutation on prognosis of CMML patients is different from TET2

10 studies were included by ASXL1 gene status reported OS, 1726 patients were included in the study. The OS subgroup analysis revealed CMML patients with ASXL1 mutation HR = 1.56, 95% CI = 1.34 - 1.80, P = 0.000 (Figure 3), suggesting that the ASXL1 mutation in CMML patients was unfavorable for the prognosis compared with ASXL1 without mutation. The heterogeneity test results were P = 0.197, I² = 26.8%, indicating that there was no heterogeneity among the studies. Therefore, it was suggested that ASXL1 non-mutation could be used as a prognostic indicator in CMML, which had a positive significance for the prognosis.

Figure 3. Forest plot of the risk ratio (HR) and 95% confidence interval of the overall survival (OS) of patients with ASXL1 mutation and non-mutation CMML. The size of the square or square represents the weight, and the length of the line represents the width of the 95% CI.

Next, we conducted subgroup analyses based on gene type, sample size and different analysis methods. As shown in Table 5, among ten studies involving ASXL1 above, the difference in sample size and analysis method did not affect the final results.

Table 5. Results of subgroup analyses of pooled HRs for OS with or without ASXL1 mutation.

Categories	Number of studies	Number of patients	Pooled HR (95%CI)	Heterogeneity
				I²	P-value
Gene type
ASXL1 mutation	10	1726	1.56(1.34 - 1.80)	26.8%	0.197
Sample size
≥100	6	1538	1.58(1.28 - 1.93)	38.6%	0.149
＜100	4	188	1.90(1.10 - 3.26)	12.2%	0.332
Analysis method
Univariate	2	210	1.52(1.02 - 2.26)	10.8%	0.29
Multivariate	8	1516	1.65(1.31 - 2.08)	37.0%	0.134

Among studies involving the gene status of ASXL1, there are four and three respectively reports the relationship between the acute conversion rate and mortality of the disease and the gene status (Table 6). We performed the combination of the difference between the two rates (Table 7) and the consistency test on the data. For the acute conversion rate, the estimated variance S²d is 0.000773, and the calculated weighted mean is −0.0595, (95% CI = −0.0873 – −0.0317). For mortality of the disease, the estimated variance S²d is 0.00135, and the calculated weighted mean $\overline{d}$ is −0.146, (95% CI = −0.183 – −0.109). The above suggests that ASXL1 mutation had no advantage. Although TET2 and ASXL1 had higher mutation rates in CMML, the prognostic results of mutations were opposite.

Table 6. The effect of ASXL1 mutation status on the acute transformation and death rate of CMML patients.

Event	ASXL1 Mutation		ASXL1 Wild Types		Total
	E/M	E Rates	E/W	E Rates	E/Total	E Rates
Acute Transformation
MM Patnaik [23]	13/82	0.16	12/93	0.13	15/175	0.09
MM Patnaik [19]	26/164	0.16	44/256	0.17	70/420	0.17
Veronique Gelsi-Boyer [17]	11/25	0.44	0/26	0.00	11/51	0.22
Hyun-Young Kim, M.D. [30]	8/16	0.50	4/17	0.24	12/33	0.36
Death
MM Patnaik [23]	71/82	86.59	75/93	80.65	146/175	0.83
MM Patnaik[19]	117/164	71.34	137/256	53.51	254/420	0.60
Hyun-Young Kim, M.D. [30]	15/18	83.33	11/18	61.11	26/36	0.72

E: Events, M: Mutation, W: Wild type

Table 7. Combining Table 6 to calculate the difference between the two rates (Incidence of events with or without mutation) in different events.

	d	p	p(1-p)	wi	wi*d	wi*p(1-p)
Acute Transformation
MM Patnaik[23]	−0.03	0.09	0.08	43.58	−1.29	3.42
MM Patnaik [19]	0.01	0.17	0.14	99.96	1.34	13.88
Veronique Gelsi-Boyer [17]	−0.44	0.22	0.17	16.78	−7.38	2.84
Hyun-Young Kim, M.D. [30]	−0.27	0.36	0.23	10.78	−2.86	2.49
Total				171.09	−10.18	22.63
Death
MM Patnaik [23]	−0.06	0.83	0.14	44.98	−2.67	6.22
MM Patnaik[19]	−0.18	0.60	0.24	99.96	−17.82	23.89
Hyun-Young Kim, M.D. [30]	−0.22	0.72	0.20	9.00	−2.00	1.81
Total				153.94	−22.49	31.92

d: Difference between two rates, p: Total rate, w_i: Weights

3.3. Publication bias and sensitivity analysis

A publication bias study was conducted on the included studies. In a study on the effect of TET2 mutation status on CMML patients, the Begg's test result was P = 0.624, P >0.05, indicating that no publication bias was observed in the HR assessment (Figure 4A), which was verified by Egger's test result—— P = 0.841, P >0.05. And in a study on the effect of ASXL1 mutation status on CMML patients, the Begg's test result indicated P = 0.421, P >0.05, indicating that no publication bias was observed in the HR assessment (Figure 4B), which was verified by the result of Egger's test P = 0.209, P >0.05.

Figure 4. Funnel plot analysis of publication bias in research on different genes (A) TET2 gene (B) ASXL1 gene.

In addition, we conducted a sensitivity analysis of the included studies to confirm the stability of the results of the meta-analysis of TET2 or ASXL1 on OS in CMML patients (Figure 5).

Figure 5. Sensitivity analysis of individual research results (A) The effect of TET2 gene mutation status on the overall survival of CMML patients (B) The effect of ASXL1 gene mutation status on the overall survival of CMML patients.

3.4. The effect of the interaction between TET2 and ASXL1 mutation status on the prognosis of CMML patients

According to the results of single gene analysis, TET2^MT or ASXL1^WT is beneficial for the prognosis. A total of 3 studies combined the two genes for analysis, including 1397 patients involving both TET2 and ASXL1 mutation status. In the OS subgroup analysis, with TET2^MT and ASXL1^WT as controls, the HR of both TET2 and ASXL1 mutations (TET2^MT+ASXL1^MT) in CMML patients, was 1.51, 95% CI = 1.14 - 1.99, P = 0.004 (Figure 6A). The heterogeneity test results were P = 0.613, I²= 0%, suggesting that both TET2 and ASXL1 mutations in CMML patients are not good for the prognosis of CMML patients. The HR of neither TET2 nor ASXL1 mutations (TET2^WT+ASXL1^WT) in CMML patients was 1.49, 95% CI = 1.12 - 1.98, P = 0.007 (Figure 6B). The heterogeneity test results were P = 0.414, I²= 0.0%, It is suggested that the mutation status of TET2 and ASXL1 not mutated in CMML patients is not conducive to the prognosis. The HR of TET2 without mutation and ASXL1 mutation (TET2^WT+ASXL1^MT) In CMML patients was 1.88, 95% CI = 1.21 - 2.94, P = 0.005 (Figure 6C). The heterogeneity test results were P = 0.086, I²= 59.2%, suggesting that the genetic status of TET2^WT and ASXL1^MT in CMML patients is not conducive to the prognosis.

Figure 6. Forest plot of the risk ratio (HR) and 95% confidence interval of the overall survival (OS). The size of the square or square represents the weight, and the length of the line represents the width of the 95% CI. (A) Patients with TET2 mutation ASXL1 without mutation and TET2&ASXL1 mutations (B) Patients with TET2 mutation ASXL1 without mutation and TET2&ASXL1 without mutation (C) Patients with TET2 mutation ASXL1 without mutation and TET2 without mutation ASXL1 mutation.

Therefore, for CMML patients, having both TET2^MT and ASXL1^WT are more beneficial for the prognosis than TET2^MT or ASXL1^WT alone, and is a better prognostic indicator of CMML.

4. Discussion

CMML is a rare type of bone marrow tumor in the elderly. There is limited prospective data on the natural history of CMML. And about 25% of patients will progress to AML. However, these secondary AMLs are notoriously difficult to treat [7, 36]. The prognosis of CMML disease is of great significance to the quality of life of patients. In recent years, with the rapid development of next-generation sequencing technology, targeted therapies to identify new somatic mutations may have a potential impact on myeloid malignancies.

Mutations in TET2 and ASXL1 are widespread in a variety of myeloid tumors. Mutations in ASXL1 in medullary carcinoma usually coincide with mutations in the TET2 gene [37]. The mechanism of their occurrence and development in the disease is not yet fully understood. As for their roles in the prognostic evaluation of CMML, there is still no consensus. Our study included 16 studies and conducted a meta-analysis of the prognosis of TET2 and ASXL1 mutations on CMML. In this meta-analysis, single gene studies or two gene combination studies are single-center non-random trials or small sample sizes or lack of high-quality evidence. Therefore, we conducted this meta-analysis to estimate the prognostic value of ASXL1 and TET2 in CMML more accurately.

Our meta-analysis combined the results of CMML patients from 16 independent studies, and showed that TET2^MT had a survival advantage in CMML (HR = 0.74, 95% CI = 0.61 - 0.91, P = 0.005), and ASXL1^MT had a significant negative impact on the prognosis of CMML (HR = 1.56, 95% CI = 1.34 - 1.80, P = 0.000). According to the results of single gene analysis, TET2^MT or ASXL1^WT is beneficial for the prognosis. And the subgroup analyses showed that the sample size of the study and the difference of the analysis method will not affect the final results (Tables 2 and 5). We combine the two genes for further analysis. Our results showed that compared with patients with TET2^MT and ASXL1^WT, the HR of patients with in both TET2^MT and ASXL1^MT was 1.51 (95% CI = 1.14 - 1.99; P = 0.004), the HR of patients with neither TET2^MT nor ASXL1^MT was 1.49 (95%CI = 1.12 - 1.98; P = 0.007) and the HR of TET2^WT and ASXL1^MT patients was 1.88 (95%CI = 1.21 - 2.94; P = 0.005). All showed a poor prognosis, suggesting that the genotype of TET2^MT and ASXL1^WT is the most beneficial for the survival of CMML patients. It is speculated that ASXL1 and TET2 might have interaction and regulatory mechanisms in CMML.

ASXL1 can regulate chromatin by interacting with multiple sets of inhibitory complex proteins (PRC1 and PRC2). Monoubiquitination of lysine 119 on histone H2A (H2AK119Ub) is a histone modification related to the inhibition of chromatin state. H2AK119Ub and histone 3 lysine 27 trimethyl (H327Kme3) play a synergistic role in PRC-mediated gene suppression [38]. And ASXL1^MT causes PRC2 to mediate H3K27 trimethylation [39]. Studies have shown in vitro [40], ASXL1^MT can enhance the activity of the ASXL1-BAP1 (BRCA-related protein 1) complex in bone marrow precursors. This complex leads to the complete elimination of H2AK119Ub and depletion of H327Kme3, thereby promoting transcription disorders [39]. And TET2 is essentially an epigenetic enzyme that can convert DNA 5-methylcytosine (5mc) to 5-hydroxymethylcytosine (5hmc). TET2^MT can damage the hydroxymethyl catalytic activity of epigenetic enzymes and may result in a decrease in 5-hmc levels in genomic DNA [41]. This is one of the mechanisms of CMML pathogenesis caused by TET2^MT. By producing bone marrow chimera, the overactive ASXL1-BAP1 complex and the loss of TET2 function synergistically promote the targeted transformation of hematopoietic cells to the myeloid lineage [42]. Therefore, ASXL1 can interact with H2A ubiquitination and TET2-mediated DNA modification. From this perspective, the role of TET2 in the progression of CMML depends on the presence of mutations in ASXL1.

Although we included the previous research as fully as possible into our analysis, the results still have their limitations. First of all, due to lack of data, we cannot evaluate other clinical parameters except OS. There are only three studies on the two genes of TET2 and ASXL1. Some other statistical methods can only play a supporting role. It causes certain limitations to our analysis. Secondly, all selected studies are observational studies, not prospective randomized controlled studies. Finally, the number of included studies is relatively small, especially for TET2 studies. When the number of studies included in the meta-analysis is less than 10, the Egger’s and Begg’s test are relatively low. So, Egger’s and Begg’s test may not be able to detect publication bias. Therefore, more epidemiological data, many in vitro experimental studies and prospective randomized controlled studies, are still needed for more subgroup analysis and further analysis to verify our conclusions.

In addition, for specific mutation types of genes, only one of the studies involved showed that the ASXL1 frameshift mutation (FS) had the greatest impact on OS compared with other mutation types (HR = 5.87, 95% CI = 1.98 - 7.4, P = 0.001) [20]. We cannot reveal the significance of mutation types on the prognosis of CMML through meta-analysis. Therefore, in the future, further research can be done on the specific types of gene mutations.

This study is the first meta-analysis to compare the effects of two genes (TET2, ASXL1) on patient prognosis. Our analysis observed that TET2 mutation and ASXL1 non-mutation contribute to the prognosis of CMML patients. We provide a basis for evaluating the prognosis of CMML patients, exploring potential therapeutic targets, and predicting the efficacy of treatment. It is possible to create a more accurate new prognostic scoring system for CMML including TET2 and ASXL1 mutations in the future.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by The National Key Research and Development Program of China [grant no 2016YFA0201504], National Natural Science Foundation of China [grant no 81673471, 81102464], The National Mega-project for Innovative Drugs [grant no 2014ZX09201042], The CAMS Initiative for Innovative Medicine [grant no 2016-I2M-2-002] and The Drug Innovation Major Project of China [grant no 2018ZX09711001-007-002].