TET2 mutation is an independent favorable prognostic factor in myelodysplastic syndromes (MDSs)

Myelodysplastic syndromes (MDSs) are hematopoietic stem cell disorders with impaired differentiation and risk of progression to acute myeloid leukemia (AML).¹  Initially, the French-American-British (FAB) and World Health Organization (WHO) classifications distinguished several subtypes.^2,3  The International Prognostic Scoring System (IPSS) considered the number of cytopenias, the karyotype, and the percentage of bone-marrow blasts.⁴  Prognostic parameters for survival and progression to AML have been proposed, including circulating blasts, lactate dehydrogenase serum levels, and red blood-cell transfusion requirement.^5-7  However, prognostic factors are lacking for the early stages of the disease.

Little is known regarding the pathogenesis of MDS, although approximately 50% of MDS patients display karyotype aberrations in cytogenetic analyses.⁸  Recurrent molecular anomalies have been identified in less than 10% of MDS mainly advanced stages (for a review, see Nimer¹ ). High-resolution single nucleotide polymorphism (SNP) array–based karyotyping recently has contributed to the detection of small genomic imbalance and acquired segmental uniparental disomy in MDS.^9,10 

We previously described 4 MDS and AML patients with acquired interstitial deletions affecting the 4q24 chromosomal region.¹¹  Molecular and cytogenetic approaches allowed the identification of the ten-eleven translocation 2 (TET2) gene in a common 500-kb minimal deleted region.¹²  The TET2 gene comprises 11 exons spread over 150 kb.¹²  The 2002 amino acids predicted TET2 protein exhibits 2 evolutionary conserved regions: one region located from amino acid 1134 until amino acid 1444 and a second region located near the carboxyterminal end from amino acid 1842 until amino acid 1921 that is related to the hydroxylase family depend on iron and 2 oxoglutarate.¹³  We and others^12,14-19  recently have reported the recurrent alteration of the TET2 gene in the stem cells of various myeloid disorders, including 20% of MDS and MDS/myeloproliferative neoplasm (MPN)–related AML and 8% to 15% of MPN. In the present study, we investigated TET2 mutations in a series of MDS to describe the characteristics of TET2-mutated patients and to evaluate the prognostic value of TET2 mutations.

Ninety-six samples of MDS at diagnosis were collected at Cochin Hospital (Paris) and at Paoli-Calmettes Institute (Marseille). Clinical and hematological data were recorded after patients gave their informed consent in accordance with the Declaration of Helsinki. According to the FAB classification, patients were classified as refractory anemia (RA; n = 29), refractory anemia with ringed sideroblasts (RARS; n = 12), refractory anemia with excess blasts (RAEB; n = 47), and RAEB in transformation (RAEBt; n = 8). According to the WHO criteria, patients were classified as RA (n = 16), 5q-syndrome (n = 2), undefined MDS (MDS-U, n = 4), refractory cytopenia with multilineage dysplasia (RCMD, n = 7), RARS/RARS with thrombocytosis (RARS-T, n = 8), refractory cytopenia with multilineage dysplasia (RCMD) with ringed sideroblasts (n = 4), RA with an excess of blasts less than 10% (RAEB1, n = 25), RAEB more than 10% blasts (RAEB2, n = 23), and secondary AML with a bone-marrow blast count between 20% and 29% (n = 7). IPSS was Low in 31, Int-1 in 30, Int-2 in 24, and High in 11 patients. In 12 cases, samples were available before and after disease progression. Among these 96 patients, 88 patients had a follow-up for 15 days or longer and were studied for overall survival (OS), leukemia-free survival with AML transformation defined as 30% bone marrow blasts or greater according to the FAB classification, and event (death or AML transformation)-free survival (for flow chart, see supplemental Figure 1, available on the Blood website; see the Supplemental Materials link at the top of the online article). The follow-up information was updated by means of a telephone call to patients or their doctors or to town halls. Data concerning treatments received during the period of follow-up were recorded for all patients. This study has been approved by the Hopital Cochin (Paris) and Institute Paoli-Calmette (Marseille) ethics committees.

DNA (n = 96) was extracted from frozen samples by the use of commercial kits (QIAGEN). For TET2 gene analysis, polymerase chain reaction (PCR) and direct sequencing were performed starting from 20 ng of genomic DNA, as previously described.¹²  All reported TET2 mutations were scored on both strands by use of ABI 3300 capillary sequencers. Sequence traces were analyzed with Mutation Surveyor (Softgenetics Inc) and reviewed manually. Missense mutations were considered only when located in conserved regions spanning from 1134 to 1444 and 1842 to 1921 amino acids of the TET2 protein. TET2 anomalies were numbered according to European Molecular Biology Laboratory nucleotide sequence reference FM992369. Previously annotated single nucleotide polymorphisms in database (http//www.hapmap.org) were discarded. N- and K-RAS exons 2 and 3 DNA mutation screening were performed by high-resolution melting PCR with the use of LightCycler 480 in High Melting Resolution Master Mix 1X (Roche Applied Science) with 10 ng of genomic DNA, primers 0.1 μmol/L each, and 25 mmol/L MgCl₂. High-resolution melting-PCR cycling conditions were initial denaturation at 95°C for 10 minutes followed by 50 cycles at 95°C for 10 seconds, at 63°C for 15 seconds, and at 72°C for 25 seconds. Melting curve was measured from 72°C to 95°C with 25 acquisitions per degree centigrade. Primers used are listed as follows:

• KRAS_X2_F4 GGCCTGCTGAAAATGACTGAA;

• KRAS_X2_R4 AATTAGCTGTATCGTCAAGGCACTC;

• KRAS_X3_F2 TCCCTTCTCAGGATTCCTACAGG;

• KRAS_X3_R2 ACAGGGATATTACCTACCTCATAAACATT;

• NRAS_X2_F2 CTGATTACTGGTTTCCAACAGGTTCT;

• NRAS_X2_R2 TGGGTAAAGATGATCCGACAAGT;

• NRAS_X3_F1 TTGTTGGACATACTGGATACAGCTG;

• and NRAS_X3_R1 CATTTCCCCATAAAGATTCAGAACAC

For SNP arrays, DNA from 23 cases was analyzed with the use of Affymetrix GeneChip Human Mapping SNP6.0 following the manufacturer's protocol (Affymetrix). In brief, 250 ng of DNA were digested with NspI and StyI, and adapters were ligated to each restriction enzyme digestion, followed by PCR amplification. The 2 PCR products were combined, cleaned-up, and end-labeled to be hybridized to the SNP array containing 1.8 million features with half of them specifically designed to detect copy number variations. SNP array data were analyzed for signal intensity and SNP calls by use of the Gene Chip Genotyping Console 3.0.2 (GTC; Affymetrix). Areas of copy number change (CN) and loss-of-heterozygosity (LOH) were investigated by use of the bioinformatic tools GTC version 3.0.2 (Affymetrix), DChip (http://biosun1.harvard.edu/complab/dchip/), and the Partek Genomics Suite (www.partek.com/). All 23 chips passed Affymetrix QC standards (Contrast QC: minimum 2.4, average 3.1; QC call rate minimum 96%, average 98%, median of the absolute values of all pairwise difference: maximum 0.33, average 0.26). We paid special attention on the 106,287,392-106,420,407 region of chromosome 4, which contains the TET2 gene (http://genome.ucsc.edu).

Continuous variables are reported as medians and interquartile range and were compared by use of the Wilcoxon test. Categorical variables are reported as count and percentage and were compared by the use of the χ² or Fisher exact tests. The prevalence of TET2 mutation was estimated with the 95% confidence interval (95% CI). OS was measured as the time in months from diagnosis to death or censored at last patient's follow-up if no death occurred. The time to AML transformation was defined as the time between diagnosis and documented transformation (≥ 30% bone-marrow blasts). OS and time to AML transformation were studied with Kaplan-Meier estimates, and Log-rank test was used to compare the survival curves between mutated and unmutated patients. Multivariate Cox proportional hazards model was used to study the independent factors of death adjusted for TET2 mutation status, age (by 10-year increments), IPSS, and treatment by red blood-cell transfusions, these variables being statistically significant in univariate analyses. Unadjusted and adjusted hazards ratio and their 95% CI are provided. All statistical analyses were 2-sided, and P values less than .05 were considered to have statistical significance. Statistical analyses were performed with the use of the SAS package version 9.01 (SAS Institute).

We analyzed the TET2 coding sequence (exons 3-11) in a cohort of 96 MDS patients. According to the FAB classification, patients were classified as RA (n = 29), RARS (n = 12), RAEB (n = 47), and RAEBt (n = 8). According to the WHO criteria, patients were classified as RA (n = 16), 5q-syndrome (n = 2), undefined MDS (MDS-U, n = 4), refractory cytopenia with multilineage dysplasia (RCMD, n = 7), RARS/RARS with thrombocytosis (RARS-T, n = 8), RCMD with ringed sideroblasts (n = 4), RA with an excess of blasts less than 10% (RAEB1, n = 25), RAEB more than 10% blasts (RAEB2, n = 23), and secondary AML with a bone-marrow blast count between 20% and 29% (n = 7). The majority of patients had a rather good prognosis according to IPSS, including 63.5% of Low/Int-1 and 36.5% Int-2/High risk.

We observed 27 different anomalies of the TET2 coding sequence in 22 patients, including 13 (48.1%) frameshifts, 8 (29.6%) nonsense mutations, and 4 (14.8%) missense mutations, 1 case with a potential structural abnormality of the gene (UPN 51), and 1 mutation affecting a splice site (UPN 137; Table 1). Mutations were spread all over the gene, mostly affecting exon 3 (10 events) and exon 11 (11 events; supplemental Figure 2). Missense mutation leading to I1873T substitution was absent in nontumoral DNA in one case, suggesting that this substitution was not a polymorphism. The acquired nature of the K1299G or L1872P substitutions could not be checked in 2 other patients, but substitution at K1299 and L1872 has already been reported as acquired.^17-19  Likewise, L1872 and I1873 targeting has been described and lies within the catalytic domain of the TET2 homologue TET1.¹³  It is therefore unlikely that these alterations would be of germinal origin in these specific patients. Interestingly, 2 different anomalies of the TET2 coding sequence were detected in 5 patients, suggesting the alteration of the 2 copies of the TET2 gene (Table 1). The type and localization of mutations were not related to WHO, FAB, or IPSS classes (Table 1 and supplemental Figure 2).

By conventional cytogenetics, deletion of the 4q24 region was detected in 4 cases. A first case (UPN 84) with a deletion of the 4q24 region and a second case (UPN 146) with a complete loss of one chromosome 4 both had a frameshift in the coding sequence of the remaining copy. In the 2 other cases (UPN 210 and 211), the deletion of 4q24 was associated to a missense or a nonsense mutation on the remaining copy (Table 1). High-resolution SNP-arrays were used to compare tumor to normal DNA in 23 tumor samples. We detected 2 cases with LOH without change in copy number (UPN 38, 137). These 2 patients carried mutation in the TET2 coding sequence (Table 1), indicating that the LOH of chromosome 4q resulted in the loss of a wild-type TET2 copy. Overall, genetic defects likely targeting the 2 TET2 copies were detected in 11 (50%) of 22 mutated patients.

We found mutations of the TET2 coding sequence in 22 of 96 MDS patients. Overall prevalence was 22.9% (95% CI, 14.5%-31.3%), including 20.7% in RA, 16.7% in RARS, 27.6% in RAEB, and 12.5% in RAEBt according to the FAB classification (Table 2). The prevalence was 22.7% in RA/5q-syndrome/MDS-U, 16.7% in RARS/RARS-T/RCMD-RS, 14.3% in RCMD, 32% in RAEB1, and 21.7% in RAEB2 according to the WHO classification. The prevalence was 22.6%, 26.7%, 25.0%, and 9.0% in Low, Int-1, Int-2, and High IPSS risk groups, respectively. The differences in the prevalence were not statistically significant within subgroups of WHO, FAB, or IPSS classifications (Table 2).

Among the 96 patients, 12 patients had samples available at diagnosis and after disease progression. In these 12 cases, the TET2 mutation pattern remained unchanged. Mutations detected in 7 patients at diagnosis were always present, and no other change in the TET2 coding sequence was observed after progression. None of the 5 patients lacking TET2 mutations had gained mutations at the time of AML transformation. This finding indicates the stability of TET2 mutations during the course of the disease.

The clinical and hematologic characteristics of mutated and unmutated patients are compared in Table 2. There was no significant difference in age, sex, previous exposure to chemotherapy or radiotherapy, peripheral blood parameters, circulating blast cells, multilineage dysplasia, and IPSS-based karyotype distribution between mutated and unmutated cases. When analyzing separately IPSS Low/Int-1 (n = 61) and Int-2/High (n = 35) risk groups, clinical and hematologic parameters were not different (supplemental Table 1). These parameters were also equivalent between the groups of patients with 1 versus 2 targeted copies of TET2.

Additionally, the mutational status of N- or K-RAS exons 2 and 3 was determined in 84 among the 96 samples. The frequency of N- or K-RAS mutation was 2.3% (2/84) in this group of patients. The 2 patients with N- or K-RAS mutation also had TET2 mutation: one patient with RA of Low IPSS had a mutation of exon 2 in K-RAS, and one patient with RAEB1 of Int-1 IPSS had a mutation of exon 2 in K-RAS. None of the patients devoid of TET2 mutation were mutated in the N- or K-RAS genes. In this cohort, no statistically significant link could be established between these 2 molecular events (χ² test, P = .06).

The prognostic impact of TET2 mutations was evaluated in 88 MDS patients with a follow-up of 15 days or more (supplemental Figure 1). The prevalence of TET2 mutation was 21 (23.8%) in 88 patients. At the time of last follow-up, 4 of 21 patients with TET2 mutation had died compared with 27 of 67 in the unmutated group. Treatments received by TET2 mutated and unmutated patients did not differ significantly (Table 2). Treatments had consisted in supportive care with red blood-cell transfusion in 27.1% of cases, erythropoiesis-stimulating agents with or without granulocyte colony-stimulating factor in 34.4%, thalidomide or lenalidomide in 5.2%, demethylating agents in 5.2%, and low doses or intensive chemotherapy in 7.3%. More than 40% of the patients did not receive any therapy. A Kaplan-Meier curve showed that OS was significantly increased in patients with TET2 mutations (Figure 1A). The 5-year OS was 76.9% (95% CI, 49.2%-91.3%) in mutated patients versus 18.3% (95% CI, 4.2%-41.1%) in unmutated patients (P = .005). Kaplan-Meier analysis also showed that the freedom from AML progression within 3 years was 89.3% (95% CI, 63.1%-97.0%] in the mutated group versus 63.7% [95% CI, 48.2%-75.4%) in the unmutated group (Figure 1B). This difference was significant (P = .035). The event-free survival (death or transformation in AML) within 5 years was 76.9% (95% CI, 49.2%-90.7%) versus 18.1% (95% CI, 3.9%-40.7%) in the mutated and unmutated groups (P = .002; Figure 1C). This finding suggests that a lower rate of AML transformation in mutated patients might account for a longer survival in MDS with TET2 mutation.

Overall survival also was analyzed according to the IPSS in the group of 88 patients. In the group of Low and Int-1 MDS (n = 54), mutated patients (n = 14) had significant longer survival within 5 years than unmutated patients (n = 40; P = .020). However, the difference in OS at 3 years between mutated (n = 7) and unmutated (n = 27) patients within the Int-2/High risk group (n = 34) did not reach the significance (p = 0.092) probably as a result of the small size of this subgroup (supplemental Figure 3).

In univariate analysis, age, IPSS, number of cytopenia, percentage of bone-marrow blasts, IPSS-based unfavorable karytoype, and transfusion requirement were identified as OS prognostic factors. The absence of TET2 mutation was associated with a 4.1-fold (95% CI, 1.4-12.0-fold) increased risk of death (P = .010). In multivariate analysis, including mutational status, age, IPSS, and transfusion requirement, the absence of TET2 mutation is associated with a 5.2-fold (95% CI, 1.6-16.3-fold) increased risk of death (P = .005), suggesting that TET2 mutation was an independent favorable prognostic factor (Table 3).

We recently identified TET2 gene mutation in 19.8% of a series of MDS patients belonging to all WHO or FAB subtypes and in AML after MDS.¹²  The present retrospective study confirms the high frequency (22.9%) of TET2 mutations found in MDS. The prevalence was 24.6% in the Low/Int-1 MDS subgroup and 20% in the Int-2/High MDS subgroup, demonstrating that the frequency of mutations does not increase in advanced diseases. In 12 cases, the mutational status was stable before and after disease progression, suggesting that mutations of TET2 gene are not linked to the onset of leukemia. N- or K-RAS mutations are detected in only 2.3% patients of the present MDS cohort. TET2 mutations that appear as a frequent molecular event are an independent factor of good prognosis in MDS.

In half of the patients, a single alteration of the TET2 coding sequence was observed. Therefore, a haploinsufficiency effect of TET2 could not be excluded, as suggested for several genes of the 5q31 region such as NPM1or RPS14.^20-22  Although our results are compatible with TET2 being haploinsufficient, we have not ruled out the inactivation of the wild-type copy of TET2 gene by other mechanisms such as hypermethylation. With the use of SNP arrays, we detected LOH without change in copy number, including the 4q24 region in 2 of 23 cases, a frequency that might be underestimated but was in line with published results^9,10  and did not identify additional patients with TET2 anomalies with respect to sequence analysis. Overall 2 anomalies of TET2 were observed in the other half of the patients, which appeared preferentially to fall in the Low/Int-1 MDS. This situation contrasts with inactivation of both copies of other tumor suppressor genes, either bi-allelic mutation of TP53, hypermethylation of p15INK4B locus or hypermethylation of CTNNA1 or FZD9 gene together with deletion that is associated with disease progression or therapy-related MDS.^23-27 

MDS patients of this cohort had a rather good prognosis (63.5% Low/Int1 and 36.5% Int-2/High risk), and only few of them received MDS-related therapy. For the first time and despite the small size of the sample, we report that MDS patients with TET2 mutations had a longer OS than unmutated patients. This result has to be confirmed with larger series of patients. The survival advantage resulted at least in part from less AML progression in the mutated patients. Multivariate analysis including age, IPSS, and transfusion requirement demonstrates that TET2 mutation was an independent factor of favorable prognosis. Other mutations in TP53, AML1/RUNX1, and N- or K-RAS genes have been reported with low prevalence in MDS.^23,24,28-30  Some of the molecular abnormalities described in MDS have demonstrated unfavorable prognostic value, including NRAS and TP53 mutations or hypermethylation of p15INK4B.^29-32  Those mutations were generally observed only in advanced stages or therapy-related diseases, and their prognostic value was generally not independent of other parameters, especially of IPSS. In sharp contrast, TET2 mutation is a favorable prognostic factor in MDS that is independent of IPSS. TET2 mutations are observed during the early steps of the disease^12,14,19  and might influence the phenotype and clinical behavior. In MPN, mutations in JAK2 or MPL might cooperate with TET2 mutations to induce the clinical phenotype.^12,15  The oncogenic events specifically cooperating with TET2 abnormalities during MDS pathogenesis remain to be identified that will allow the evaluation of their respective contribution to phenotype and clinical behavior in these neoplasms.

Our results suggest that the determination of TET2 mutational status might assist risk stratification. Prospective studies are required to determine whether response to treatments depends on TET2 mutations.

We thank Virginie Trouplin, Rosa Sapena, Cécile Pierre-Eugène, Edith Mortera, Céline Marie, and Monique Grandjean for recording clinical data, banking, and sequencing; Maryline Delattre for support; and Norbert Ifrah for helpful discussion.

This work was supported by grants from the Direction Régionale de la Recherche Clinique, AP-HP (PHRC MAD06; to M.F.), the Institut National du Cancer (INCa; to M.F., O.A.B., W.V.), Canceropole Ile-de-France (AAP INCA 2008; to M.F.), the Fondation de France, Fondation contre la leucémie (to O.A.B., D.B.), and the Ligue Nationale Contre le Cancer (to O.A.B., W.V.).

Contribution: O.K., V.G.-B., and M.C. performed research and analyzed data; S.G. performed the statistical analyses; V.D.-V., F.P., F.V., and E.D. performed research; B.Q., E.S., O.B.-R., V.M.-D.M., M.H.-B., C.P., P.F., and F.D. provided patient samples; D.B. and M.F. planned research; E.S., W.V., O.A.B., D.B., P.G., and M.F. analyzed data and drafted the manuscript; and E.S., W.V., P.F., D.B., O.A.B., and M.F. finalized the manuscript. All authors reviewed and agreed with the final version of the manuscript

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A list of Groupe Francophone des Myelodysplasies participants can be found in the supplemental Appendix, available on the Blood website.

Correspondence: Michaëla Fontenay, Service d'Hématologie Biologique, Hôpital Cochin, 27 rue du Faubourg Saint-Jacques, F75679 Paris Cedex 14, France; e-mail: michaela.fontenay@inserm.fr.