Genome sequencing in myelodysplastic syndromes: can molecular mutations predict benefit from hypomethylating agent therapy?

Abstract

Evaluation of: Bejar R, Lord A, Stevenson K, et al. TET2 mutations predict response to hypomethylating agents in myelodysplastic syndrome patients. Blood 2014 Oct 23;124(17):2705-12.

Patients with myelodysplastic syndromes (MDS) have clinically variable courses even within the same prognostic subgroups. Although hypomethylating agents (HMAs) have been shown to improve outcomes in patients with high-risk MDS, many patients do not derive benefit. There is an urgent clinical need to identify patients with low probability of benefiting from HMAs but no reliable clinical predictors or biomarkers have been discovered to date. Although some recurrent molecular mutations in MDS carry independent prognostic value, their ability to predict benefit from HMAs is not clear. Here, we discuss an important article in which sequencing from samples of 213 patients identified recurrent mutations associated with response to HMAs. Although an important step in the right direction, the clinical implications of these findings are far from optimal and identification of biomarkers that can reliably predict benefit from HMAs and other therapies in patients with MDS remains a top clinical and a research priority.

Keywords:

• biomarkers
• genome sequencing
• hypomethylating agents
• myelodysplastic syndromes
• prognostication
• TET2 mutation

Myelodysplastic syndromes (MDS) are a group of clonal neoplasms of the hematopoietic stem cell characterized by ineffective hematopoiesis leading to variable degrees of peripheral blood cytopenias and risk of transformation to acute myeloid leukemia [1]. Currently, the most widely used prognosticating system for patients is the International Prognostic Scoring System (IPSS) which divides patients into four risk groups (low, intermediate-1, intermediate-2, and high) based on number of cytopenias, bone marrow (BM) blast percentage and recurrent prognostic karyotypic abnormalities to predict survival and evolution to acute myeloid leukemia [2]. Those with higher risk MDS (intermediate-2, high) have a median survival of <1 year, whereas those with lower risk MDS (intermediate-1, low) have a median survival of 3–8 years [3]. Prediction of outcomes is important in management as this impacts risk/benefit estimations for initiation of treatment, including hypomethylating agents (HMAs) and allogeneic hematopoietic cell transplantation (AlloHCT) [4]. Although HMAs improved outcomes for patients with higher risk MDS, not all patients benefit from this therapy and no reliable methods have been discovered to predict differential likelihood of benefit from HMAs. The IPSS, as well as other newer and improved scoring systems such as the WHO classification-based Prognostic Scoring System and the revised IPSS (IPSS-R), were not designed to predict response to specific treatments and do not include the recently discovered recurrent molecular mutations [4].

Technological advances have allowed for substantial decreases in time and cost of DNA sequencing, thus enabling investigation to explore the role of novel mutations in this disease [5]. Gene mutations are considered among the key steps in development of MDS and the diversity of clinical phenotypes is thought in part to be due to the variety of genetic mutations [6]. In addition, these genetic alterations were found to provide additional prognostic information [7,8]. For example, mutations in TP53, EZH2, ETV6, RUNX1 and ASXL1 were independently found to be associated with shorter overall survival (OS) [9]. However, how the presence of the recurrent molecular mutations correlates with the achievement of clinical benefit from HMA therapy is not clear.

Summary of methods & results

In a multicenter prospective study, Bejar et al. [10] collected tumor DNA from peripheral blood or BM samples of 213 patients with MDS before treatment with HMA to evaluate the predictive potential of 40 genes that are recurrently mutated in MDS. Of the patients, 5, 40, 36, 17 and 1% had IPSS low, intermediate-1, intermediate-2, high risk, and unknown; 14, 11, 59, 3, 10 and 3% had refractory anemia (RA), RA with ringed sideroblasts, RA with excess blasts, RA with excess blasts in transformation, chronic myelomonocytic leukemia, and other, respectively. Whole genome amplification was performed on genomic DNA with generation of a genotype fingerprint of 22 common single nucleotide polymorphisms for each sample. Target enrichment was performed, followed by sequencing using Illumina Hi Seq 2000. Logistic regression models were used to predict response to therapy, adjusting for age, sex, IPSS risk group and treatment. OS was calculated from the time of treatment to the time of death from any cause or date last known alive.

A mutation in at least one of the targeted 40 genes was found in 94% of patients with the following being the most commonly mutated: ASXL1 (46%), TET2 (27%), RUNX1 (20%), TP53 (18%), DNMT3A (16%), SRSF2 (16%), SF3B1 (15%) and U2AF1 (14%). As expected, mutations of SF3B1 were found mainly in patients with RA with ringed sideroblasts. The overall response rate (ORR, defined as complete response + partial response + hematologic improvement) was 47 with 31% achieving complete response. Of the patients, 20, 68 and 13% were treated with azacitidine alone, decitabine alone, and decitabine along with an unspecified agent, respectively, and there was no significant difference in response based on treatment regimen. No mutation was associated with significantly improved ORR but TET2 mutant patients had a trend toward increased ORR compared with wild type (55 vs 44%; p = 0.14).

In a revised analysis wherein mutations with a variant allele fraction of ≥10% only were considered mutated, TET2 mutations were associated with a significantly higher ORR compared with wild type (60 vs 43%, odds ratio [OR]: 1.99, p = 0.036; adjusted OR 1.98, p = 0.044). Using logistic regression analysis to look at association of gene mutations with ORR, those with mutated TET2 and unmutated ASXL1 had an increased ORR compared with all other combinations of mutations. Of the 146 patients with survival data, 119 died in follow-up, with median follow-up being 3.8 years. TET2 mutation status was not associated with OS, whereas TP53 and PTPN11 mutations were associated with lower OS. Those with complex karyotype and TP53 mutation had a median survival of 0.9 years compared with an OS of 1.3 years for patients with complex karyotype and no TP53 mutation.

In an in vivo model of murine competitive BM transplantation, donor Tet2-null and wild-type mice were treated with plpC to induce deletion of exon 3 of Tet2. BM was then harvested and combined in a 1:2 ratio with marrow from wild type donors for transplantation into a murine recipient. At 2.5 weeks after transplant, there was greater engraftment of Tet2-null cells. Seven of the transplanted mice were treated with azacitidine and seven were treated with control. Those treated with azacitidine had significant decreases in white blood cell counts and hematocrit as well as significantly decreased presence of Tet2 null cells in the blood [10].

Discussion & five-year view

An important focus for current research in MDS revolves around predicting benefit (or lack thereof) to HMA therapy to avoid exposing patients with low probability of benefit to prolonged and potentially toxic ineffective treatment, increased costs, and to allow earlier consideration of clinical trials that can alter the natural history of the disease with potentially more time for benefit [11]. Unfortunately, all efforts to date have failed in discovering clinical and/or biomarkers that allow reliable prediction of clinical benefit from HMA therapy [4,11]. A French prediction tool was developed to separate higher risk MDS patients based on OS advantage achieved with azacitidine using readily available clinicopathologic parameters [12], but it was subsequently found to offer no prognostic utility beyond that of the more commonly used IPSS-R [13]. After the disappointment of methylation predictive studies, the recent discoveries of >40 genes with recurrent somatic mutations in the vast majority of patients have renewed hope of identifying that evasive biomarker(s) [7,8,10]. The recurrent mutations were observed most commonly in genes involved in important biologic pathways that contribute to the pathogenesis of MDS such as DNA methylation (DNMT3A, TET2, IDH1 and IDH2), chromatin modification (EZH2 and ASXL1), transcriptional regulation (TP53, RUNX1 and GATA2), RNA splicing (SF3B1, U2AF1, SRSF2 and ZRSR2) and signal transduction (JAK2, KRAS and CBL) [14]. Several of these mutations were found to have an independent prognostic value and, interestingly, there appears to be no significant difference in survival impact of a specific mutation if present in a dominant clone versus a subclone [8]. Despite the growing literature supporting the independent significance of molecular mutations in MDS, none have yet been incorporated formally into validated prognostic scoring systems (in contrast to chromosomal lesions) nor used to make treatment decisions or predict responses to specific therapies [8,9,15].

In contrast to prior large studies, the report by Bejar et al. [10] clarifies the impact of recurrent mutations on OS among HMA-treated MDS patients as well as their potential impact on response to treatment. Interestingly, the only two genes whose mutations affected OS (TP53 and PTPN11) had no association with response to HMAs [10]. Importantly, among the genes previously reported to be associated with OS in largely untreated patients [9], only TP53 mutations remained independently prognostic of OS among HMA-treated patients, further demonstrating the challenges of incorporating molecular alterations into clinical decision-making. Several groups have reported on the negative impact of TP53 mutations, its association with 5q- and complex cytogenetic profiles, and that TP53 appears to be responsible for the dismal outcomes in these patients with complex karyotype [9,16]. Patients with 5q-MDS who had TP53 mutations (∼ 20%) had poor outcomes despite lenalidomide therapy [17,18]. Recent data suggest that in 5q-MDS haploinsufficiency for casein kinase 1A1 gene (CSNK1A1) might increase sensitivity to lenalidomide which induces the ubiquitination and consequent degradation of the kinase by activating the CRBN-CRL4 E3 ubiquitin ligase [19,20]. TP53 mutations might confer resistance to lenalidomide by reducing the sensitivity of casein kinase 1A1 to the drug effects [19,20]. In a recent study of MDS patients who underwent alloHCT, TP53 mutations were the most important independent predictor for mortality after alloHCT and patients with these mutations had a strikingly poor median OS of 4.6 months [21]. Taken together, these findings suggest that neither HMA therapy nor alloHCT abrogate the negative prognostic impact of TP53 mutations and that such patients represent a subgroup with a particularly dismal prognosis in whom novel approaches and clinical trials should be strongly considered [18].

Although the current report by Bejar et al. [10] represents the largest and most comprehensive published experience correlating molecular mutations with benefit from HMAs, this issue has been addressed by other investigators. The French group used Sanger sequencing specifically for TET2 on samples from 86 patients with MDS or oligoblastic acute myeloid leukemia and similarly found that TET2 mutations predicted objective responses to azacitidine therapy compared with wild-type TET2 (85 vs 52%) without affecting OS [22]. In another retrospective study of 92 HMA-treated patients, mutations in TET2 and DNMT3A in multivariate analysis predicted improved ORR to HMAs [23]. Interestingly, mutations in TET2 (along with TP53 and DNMT3A) were also associated with shorter OS after alloHCT [21]. As the authors pointed out, TET2 mutations might increase susceptibility to HMAs via increasing the percentage of actively cycling cells rather than by methylation-mediated mechanisms [10]; therefore potentially explaining the observed partial correlation.

Despite the importance of this study, several important limitations are noted. First, from a clinical standpoint, although the 10% of patients with TET2 mutations (variant allele fraction ≥10%) and wild-type ASXL1 had the highest ORR to HMAs (74%), the authors did not identify any pattern of mutations defining subset of patients with low enough ORR to HMAs that could potentially justify withholding upfront therapy with the only class of agents proven to prolong OS in higher risk MDS. Second, mutation status was assessed only before initiation of HMA therapy, whereas serial analysis can potentially discover mutations associated with evolving resistance to HMA therapy [10]. Third, the authors did not account for the potential confounding effects of the inter-patient differential expression of nucleoside-metabolizing enzymes which might affect ORR to HMAs [24]. Fourth, logistical challenges such as standardization of the conduction and the interpretation of assays, and determination of the importance of variant allele fraction levels, mutational burden of clones and subclones, and intra-tumor genomic heterogeneity need further research [8,15,18]. Fifth, the complete response rate observed with HMAs in the cohort reported by Bejar et al. [10] was higher than the 10–20% reported in clinical trials and retrospective analyses [4] which might affect the generalizability of the results. Finally, although TET2 mutation status predicted response rates to HMAs, it failed to predict OS, further complicating the application of the results to the clinic.

Although challenging to conduct, serial and comprehensive prospective evaluation of a large cohort of HMA-treated patients to evaluate genetic, epigenetic, immunologic, and clinical parameters might be the only way to define reliable predictors of clinical benefit to HMAs in MDS patients.

Key issues

• Despite the growing literature on molecular mutations in MDS, they have not yet been formally incorporated into prognosticating systems or used to predict benefit from specific therapies.

• Patients with TET2 mutations and wild-type ASXL1 appear to have the highest response rate to treatment with hypomethylating agents among MDS patients, but only mutations in TP53 and PTPN11 were associated with survival.

• No pattern of recurrent somatic mutations in patients with higher risk MDS currently justifies withholding upfront therapy with hypomethylating agents.

Financial & competing interests disclosure

A Zeidan is on an advisory board for Boehringer Ingelheim. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.