TP53 mutations and variant allele frequency in myelodysplastic syndromes with del(5q): A Mayo-Moffitt study of 156 informative cases

Revised diagnostic criteria for myelodysplastic syndromes (MDS) with del(5q) were recently issued by the International Consensus (ICC) and World Health Organization (WHO 5th edition) classification systems.1, 2 Specific diagnostic criteria common to both included presence of (i) del5q, alone or associated with no more than one other cytogenetic abnormality, exclusive of −7/del(7q), and (ii) <5% bone marrow (BM) and <2% circulating blasts. In addition, the ICC requires the absence of “multi-hit” TP53 mutation, defined as the presence of two distinct TP53 mutations, each with variant allele frequency (VAF) ≥10%, a single TP53 mutation with VAF >50%, or a single TP53 mutation (VAF ≥10%) associated with 17p13.1 deletion, copy-neutral loss of heterozygosity (LOH) at the 17p TP53 locus, or complex karyotype in the absence of LOH information.1 Diagnosis according to the WHO system also required the absence of biallelic TP53 inactivation, although VAF cut-off levels were not commented upon.2 The prognostic relevance of TP53 VAF and allelic state were recently highlighted in the context of acute myeloid leukemia3 and MDS4 and previously noted in lower-risk MDS.5 In the current study, we examined the prognostic contribution of TP53 mutation and its VAF in MDS-del(5q), with the objective to provide evidence for the aforementioned considerations in disease classification. The current study was conducted under institutional review board approved minimum risk protocols that allowed retrospective collection and analysis of data from patients with MDS-del(5q) who were seen at the Mayo Clinic or Moffitt Cancer Center, USA. Diagnosis of MDS-del(5q) was according to ICC/WHO criteria, with the exception that patients with biallelic or multi-hit TP53 mutations were included to allow analysis of impact on prognosis; however, assignment to multi-hit TP53 mutation category required strict adherence to ICC criteria.1 Patient selection was based on availability of mutation information from next-generation sequencing (NGS) that covered, at a minimum, TP53 (including VAF), spliceosome genes, ASXL1, JAK2, IDH1, IDH2, TET2, DNMT3A, and RUNX1.6 The latter was assessed at different time points, but all analysis, including correlation with laboratory or clinical variables and estimation of overall and leukemia-free survival, were based on variables collected at the time of initial diagnosis of MDS-del(5q). TP53 mutation VAF threshold of 22% was considered based on previously published evidence of prognostic relevance.4 Patients were managed according to standard treatment strategies at the discretion of the treating physician. Conventional statistical methods were applied using JMP Pro 16.0.0 software (SAS Institute, Cary, NC, USA). A total of 156 patients (median age 72 years; 63% females) who met ICC criteria for MDS-del(5q) [N = 144] or TP53-mutated MDS-del(5q) [N = 12] were recruited from the Mayo Clinic (n = 73) or Moffitt Cancer Center (n = 83). Presenting clinical and laboratory features, stratified by institution, are outlined in Table S1; 30 (19%) patients were qualified as being “therapy-related”. At the time of initial diagnosis of MDS-del(5q), 63 (40%) patients were red cell transfusion-dependent and median (range) leukocyte and platelet counts were 4.1 (0.7–26.4) and 220 (17–1280) × 109/L respectively. An additional cytogenetic abnormality, other than −7/del(7q), was present in 24 (15%) patients, with the most frequent being trisomy 8 (n = 4) and trisomy 21 (n = 3). Overall median follow-up was 3.8 years (range, 0.002–18.5) and similar between the Mayo and Moffitt cohorts (p = 0.21). During this period, 60 (38%) deaths and 26 (17%) leukemic transformations were documented; there were more deaths reported in the Moffitt cohort (p = 0.007) and more leukemic transformations in the Mayo cohort (p = 0.036). The latter differences were believed to be the result of database maturity and record keeping in the two patient cohorts. Next-generation sequencing data were available in all 156 study patients, but testing was performed at different time points: within 3 months of diagnosis (N = 66; group A), 3 months of leukemic transformation (N = 10; group B), or in-between these time periods (N = 80; group C; median interval between diagnosis and testing 27 months, range 4–220). TP53 and SF3B1 mutations were the most prevalent in all A/B/C groups: 20%/50%/21% for TP53 and 18%/50%/14% for SF3B1 (Table S2). Furthermore, the two mutations displayed significant clustering with co-expression in 11 patients (p = 0.02) and both were more prevalent in leukemic phase disease while their frequencies were similar between groups A and C, as signified above. Other mutations, including DNMT3A (11%), TET2 (11%), ASXL1 (10%), and JAK2 (6%) were equally distributed among the three groups. Among TP53-mutated cases (n = 35), VAF was ≥10%, in at least one allele, in 30 (86%) and >22% in 21 (60%); 12 (34%) were categorized as “multihit” per ICC. Frequencies for TP53 ≥ 10% VAF in groups A/B/C were 15%/50%/19% (p = 0.06), for TP53 > 22% VAF 12%/50%/10% (p = 0.01), and for multihit TP53 9%/30%/4% (p = 0.04). Among 21 patients with TP53 VAF >22%, 10 (48%) also qualified for “multihit” status whereas among 12 with multihit status, 10 (83%) displayed VAF >22%, in at least one allele. Among all 26 patients with documented leukemic transformation, 10 (39%) harbored TP53 mutations, all at >22% VAF (p < 0.01), while only 3 (12%) met criteria for multihit TP53 mutation (p = 0.44). Among 10 patients in whom NGS was performed within 3 months of leukemic transformation, three co-expressed TP53 and SF3B1 mutations, two TP53 alone, 2 SF3B1 alone, and three neither. Having established the significant clustering of TP53 mutation with leukemic phase MDS-del(5q), we next examined predictive value for overall and leukemia-free survival, in those patients where mutation information was available, either at diagnosis or at least 3 months prior to leukemic transformation (groups A + C; n = 146). Figures 1 and 2, respectively, illustrate overall and leukemic-free survival data in these patients, stratified by TP53 mutation at different VAF thresholds or allelic state. TP53 VAF >22% was the most predictive of inferior overall (p < 0.01, HR 3.3, 95% CI 1.6–6.7; Figure 1) and leukemia-free (p < 0.01, HR 7.2, 95% CI 2.4–21.8; Figure 2) survival; interestingly, ROC analysis determined optimal TP53 VAF cutoff level in the current study was 23%. Of note, there was no overall or leukemia-free survival difference between patients with wild-type versus ≤22% TP53 VAF patient groups, thus allowing combining the two groups together in the category of TP53 VAF ≤22%, in Figures 1 and 2. Multivariable analysis of patients belonging to group A or C (N = 146), which included other mutations listed in Table S1, confirmed TP53 VAF >22% as the sole predictor of inferior overall and leukemia-free survival. A similar analysis that included clinical variables of interest, such as age, therapy-related disease, additional cytogenetic abnormality, or transfusion need, confirmed the independent prognostic contribution from TP53 VAF >22%; none of the other variables in the multivariable model garnered independent significance; of note, univariate/multivariate “p” values for SF3B1 mutation were 0.19/0.61 for overall and 0.87/0.3 for leukemia-free survival. Finally, we repeated survival analysis restricted to group A (N = 66) and group C (N = 80) patients separately and confirmed the prognostic significance of TP53 VAF ≥22% with p values of 0.03 and 0.004 respectively. The current study confirms TP53 (often with ≥10% VAF) and SF3B1 as the most prevalent mutations in MDS-del(5q), with their significant coexpression suggesting pathogenetic cooperation. Novel observations from the current study included (i) similar frequencies of TP53 and SF3B1 mutations in newly diagnosed and chronic phase cases but significantly higher frequencies in leukemic phase disease and (ii) TP53 VAF rather than allelic state was the strongest predictor of both overall and leukemia-free survival. TP53 VAF of >22%, occurring in approximately 15% of chronic phase disease, was the most significant and sole predictor of both overall and leukemia-free survival and was more likely, as opposed to multi-hit mutational state, to be associated with leukemic phase disease. By contrast, allelic state was found to be prognostically more detrimental in a previously reported study of MDS, regardless of risk group, although an adverse outcome was also associated with monoallelic TP53 VAF >22%.4 Regardless, the current study provides additional evidence to further refine current considerations in both the ICC and WHO classification schemes, regarding the interpretation of TP53 mutations and their use as disease classifiers in myeloid neoplasms, especially in the context of lower-grade MDS. Previous reports on lower-grade MDS have suggested an adverse survival effect from TP53 mutations, per se,7 or at VAF >6%5 while SF3B1 but not TP53 mutation was reported to be detrimental in MDS-del(5q).6 It should also be noted that the ICC definition of multi-hit used in the current report did not include careful copy number and LOH analysis and might have led to incorrect classification in some cases. Regardless, from a practical standpoint, the current study suggests consideration of allogeneic stem cell transplant in MDS-del(5q) associated with TP53 VAF >22%, sooner than later, although we acknowledge the limited value of such treatment in TP53-mutated myeloid neoplasms, in general.8, 9 Ayalew Tefferi, Rami Komrokji, Farah Fleti, Amritpal Singh, Maymona G. Abdelmagid, Najla H. Al Ali, and Onyee Chan designed the study and compiled the data. Ayalew Tefferi and Farah Fleti performed the analyses and co-wrote the paper. Ayalew Tefferi, Rami Komrokji, Onyee Chan, Aref Al-Kali, Michelle A. Elliott, Kebede H. Begna, James M. Foran, Talha Badar, Nandita Khera, Eric Padron, David A. Sallman, Mithun Shah, Animesh Pardanani, and Naseema Gangat contributed patients. Kaaren K. Reichard, Rhett P. Ketterling, Devendra Hiwase, Attilio Orazi, and Daniel A. Arber provided expertise in hematopathology, cytogenetics, and molecular information. All authors reviewed the final draft of the paper. None of the authors declare any conflict of interest in regard to the current work. Please email the corresponding author. Table S1. Clinical features of 156 patients with myelodysplastic syndrome (MDS) with isolated del (5q), stratified by data collected at the Mayo Clinic versus Moffitt Cancer Center. Table S2. Clinical characteristics at time of diagnosis and mutation landscape of 156 patients with myelodysplastic syndromes with del (5q), stratified by time of next-generation sequencing analysis Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.