Dear Editor,

Chronic myelomonocytic leukemia (CMML) is a clonal hematopoietic stem cell disorder with overlapping myelodysplastic and myeloproliferative features [1]. TET2 is the most frequently mutated gene in CMML, present in ~60% of patients [2]. In a 2016 study in CMML [3], we reported a TET2 mutational frequency of 43% with 58 (52%) out of the 113 TET2-mutated cases harboring ≥2 TET2MUT; the presence of TET2MUT was associated with superior overall survival (OS) but significance was lost during multivariable analysis that was adjusted for adverse karyotype or the Mayo Molecular risk Model (MMM) [4]; furthermore, the results were not affected by the type or number of TET2MUT [3]. In a subsequent multicenter study [5], we observed a significantly longer OS in the presence of ≥2 TET2MUT, compared to that seen with one or no TET2MUT. In a more recent study [6], we showed that the survival difference in patients harboring 0 vs. 1 vs. ≥2 TET2MUT was apparent in both myelodysplastic (CMML-MD) and myeloproliferative (CMML-MP) CMML variants, with significance sustained during multivariable analysis that accounted for both the MMM [4] and CMML-specific molecular (CPSS-mol) [7] risk models. Other investigators have also associated ≥2 TET2MUT with longer OS in CMML [8]. The objective of the current study was to obtain additional information on the relationship between the number of TET2MUT and prognosis in CMML, in the context of contemporary risk models, including BLAST, BLAST-mol, and CPSS-mol [7, 9, 10].

The current retrospective study was conducted under an institutional review board approved minimum risk protocol that allowed retrospective collection and analysis of data from patient records. The study population consisted of 536 patients with CMML who underwent CLIA-approved Next-Generation Sequencing (NGS) and were seen at the Mayo Clinic, Minnesota, Florida, and Arizona, USA. TET2 mutations were annotated by location in either the proximal N-terminal or distal catalytic (C-terminal) domains and also according to mutation types: truncating (frameshift, nonsense, splice site) and non-truncating (missense, in-frame deletions/insertions). In patients with ≥2 TET2MUT, mutation type designation was based on the dominant variant. Diagnostic criteria were according to the International Consensus Classification [1]. Cytogenetic results were reported according to the International System for Human Cytogenetic Nomenclature [11]. Anemia was defined as “severe” (transfusion requiring or hemoglobin <8 g/dL in women or <9 g/dL in men) or “moderate” (hemoglobin 8 to <10 g/dL in women or 9 to <11 g/dL in men) [12]. Correlative analyses considered clinical and laboratory data collected at the time of initial diagnosis/referral. The Kaplan–Meier method was used to construct time-to-event curves, which were compared by the log-rank test. Both OS and blast transformation (BT)-free survival (BTFS) were calculated from the time of diagnosis as well as time of NGS to the time of death, for OS, or BT or death, for BTFS; both calculations were censored for allogeneic stem cell transplantation (ASCT). Statistical analyses were conducted using JMP Pro 17.0.0 software (SAS Institute, Cary, NC, USA).

A total of 536 Mayo Clinic patients with CMML (71% male; median age 72 years) were classified according to the number of TET2MUT (Table 1): 221 (41%) wild-type, 148 (28%) with one, 153 (29%) with 2, and 14 (2%) with ≥3 TET2MUT. Baseline clinical and laboratory characteristics across TET2 mutational categories (0 vs. 1 vs. 2 vs. ≥3) included CMML-2 (16% vs. 8% vs. 7% vs. 7%; p < 0.01), red blood cell transfusion need (24% vs. 12% vs. 9% vs. 14%; p < 0.01), platelets <100 ×109/L (38% vs. 59% vs. 54% vs. 50%; p < 0.01), leukocytes ≥13 × 109/L (50% vs. 38% vs. 35% vs. 36%; p = 0.02), circulating blast ≥2% (25% vs. 8% vs. 7% vs. 0%; p < 0.01), bone marrow blast ≥10% (13% vs. 5% vs. 5% vs. 7%; p = 0.01), abnormal karyotype (30% vs. 20% vs. 13% vs. 14%; p < 0.01), presence of del(4q) (0% vs. 3% vs. 0% vs. 0%; p = 0.02), and severe anemia (24% vs. 14% vs. 9% vs. 14%; p < 0.01) [Table 1]. Of note, most of these differences were influenced by wild-type TET2 with hardly any differences in patients with one vs two TET2MUT. Table 1 also includes information on variant allele frequency (VAF) and mutation type; TET2 VAF was significantly higher in patients with one TET2MUT with 30% displaying >50% VAF vs. 10% with two and 0% with 3 TET2MUT (p < 0.01). Among patients with 1 TET2MUT, 3% had an accompanying del(4q) and 24% had VAF > 55%. Patients with two TET2MUT were more likely to harbor N-terminal, as opposed to C-terminal mutations, compared to those with one or ≥3 TET2MUT (59% vs. 40% vs. 43%; p < 0.01). There was no difference in the distribution of truncating vs. non-truncating TET2MUT (p = 0.6). Co-mutations across TET2 mutational subcategories (0 vs. 1 vs. 2 vs. ≥3 TET2MUT) included ASXL1 (58% vs. 46% vs. 29% vs. 36%; p < 0.01), SRSF2 (39% vs. 58% vs. 53% vs. 36%; p < 0.01), KRAS (16% vs. 7% vs. 14% vs. 21%; p = 0.03), PHF6 (3% vs. 8% vs. 12% vs. 21%; p < 0.01), SETBP1 (14% vs. 5% vs. 0.6% vs. 0%; p < 0.01), IDH2 (11% vs 0.7% vs. 0% vs. 0%; p < 0.01) and BCOR (6% vs. 0.7% vs. 0.6% vs. 0%; p < 0.01); upon exclusive analysis of one vs. two TET2MUT, differences were observed for ASXL1 (46% vs. 29%; p < 0.01), SETBP1 (5% vs. 0.6%; p = 0.02), and KRAS (7% vs. 14%; p = 0.04) mutations.

Table 1 Baseline clinical and laboratory characteristics, and co-mutational patterns of TET2 in wild-type and mutated configurations in 536 informative patients with chronic myelomonocytic leukemia.
Full size table

The median transplant-censored OS (TCOS) in patients with 0 vs. 1 vs. 2 vs. ≥3 TET2MUT was 26 vs. 39 vs. 51 vs. 22 months from the time of mutation detection (Fig. 1), and 30 vs. 42 vs. 59 vs. 31 months from the time of diagnosis (Supplementary Fig. 1), respectively. In age-adjusted univariate analysis, TCOS, calculated from the time of mutation detection, was superior in patients with one (p < 0.01; HR 0.7) or 2 (p < 0.01; HR 0.4) but not ≥3 (p = 0.4) TET2MUT, compared to those with wild-type TET2 (Fig. 1). TCOS was also superior with 2 vs. 1 (p = 0.02; HR 0.7) but not with ≥3 vs. 1 (p = 0.7) TET2MUT (Fig. 1). 2 TET2MUT showed superior TCOS vs. 1 TET2MUT + VAF ≥ 55% and/or 4q21 deletion (p < 0.01), while ≥3 TET2MUT did not (p = 0.55); there was no survival difference between sum of individual VAFs (≥55% vs. <55) within 2 TET2MUT (p = 0.52); all aforementioned results were similar when TCOS was calculated from the time of diagnosis, as opposed to the time of mutation detection (Supplementary Fig. 1). The survival advantage of CMML patients with two TET2MUT, compared to those with wild-type TET2 (p < 0.01; HR 0.5), one TET2MUT (p < 0.01; HR 0.6), or ≥3 TET2MUT (p < 0.01; HR 0.4) was confirmed in multivariable analysis that included age, sex, sex- and severity-adjusted anemia, circulating blasts ≥2%, leukocyte count ≥13 × 109/L, and presence of del (4q); in the particular analysis, there was no significant difference between wild-type TET2 and one TET2MUT (p = 0.06) or between wild-type TET2 and ≥3 TET2MUT (p = 0.7). Figure 1 also depicts a similar favorable impact of exactly 2 TET2MUT in CMML-MD and CMML-MP.

Fig. 1
Fig. 1
Full size image

Overall survival data in 536 Mayo Clinic patients with chronic myelomonocytic leukemia, stratified by the number of TET2 mutations—calculated from the time of mutation detection and censored for allogeneic stem cell transplantation.

Additional multivariable analyses considered the presence or absence of two TET2MUT against TET2 VAF (i.e., sum of individual VAFs ≥55% vs. <55%; dominant allele frequency ≥55% vs. <55%), type of mutation (i.e., truncating vs. non-truncating; C-terminal vs. N-terminal), karyotype, and other mutations previously identified as being prognostically relevant in CMML, including TP53, BCOR, RUNX1, SETBP1, NRAS, ASXL1, PTPN11, U2AF1 and DNMT3A; the analysis confirmed the independent association of exactly two TET2MUT and significantly longer TCOS (p < 0.01); no additional prognostic contribution was noted for either TET2 VAF, modeled both as a summed or dominant metric, or type of mutation. However, while the prognostic relevance of two TET2MUT was sustained in the context of the BLAST (p < 0.01; HR 0.5) and CPSS-mol (p = 0.02; HR 0.7) risk models, it lost significance when accounting for the BLAST-mol risk model (p = 0.12). Our observations were largely similar in regard to BTFS (Supplementary Figs. 2 and 3).

The observations from the current study illustrate the complexity of molecular prognostication in CMML. The long-standing impression of a survival advantage in CMML patients with TET2MUT might be influenced by a number of additional factors, including TET2 multihit status and its prognostic interaction with other mutations. The latter scenario is consistent with the observed loss of prognostic significance in the context of the BLAST-mol risk model, which accounts for prognostic interaction between mutations in CMML [10]. In other words, it is conceivable that the co-mutation pattern, rather than mutational structure or multihit status, dictates prognostic relevance associated with TET2MUT. Notwithstanding this caveat, we were able to illustrate the apparent prognostic benefit of exactly two, as opposed to one or ≥3 TET2MUT; this particular association persisted even after accounting for TET2MUT VAF and accompanying del(4q); variables effectively used in a similar study to operationalize “multi-hit” (biallelic) TET2 involvement [13].

A mechanistic explanation for this novel observation includes the possibility that TET2 mutational categories based on mutation count reflect distinct biological states rather than having a simple linear effect. In other words, different numbers of TET2MUT may place the disease into more or less stable biological states. Accordingly, patients with exactly two TET2MUT likely represent a relatively stable, functionally biallelic TET2-deficient state. In contrast, a single TET2 mutation may be biologically insufficient to exert this effect and, in our cohort, is more often accompanied by adverse co-mutations such as ASXL1 and SETBP1, which may dilute or override any favorable impact and potentially exclude the emergence of additional TET2MUT. At the other extreme, the presence of ≥3 TET2MUT may not reflect greater TET2 functional loss but rather increased clonal complexity. In this setting, multiple TET2 hits may serve as a marker of clonal instability or ongoing disease evolution, offsetting the favorable biology associated with a stable biallelic TET2 configuration. The persistence of this signal despite adjustment for established molecular and genetic risk factors supports its consideration in the development of future CMML prognostic models.