TP53 mutation variant allele frequency of ≥10% is associated with poor prognosis in therapy-related myeloid neoplasms

Abstract

Revised diagnostic criteria for myeloid neoplasms (MN) issued by the International Consensus Classification (ICC) and the World Health Organization (WHO) recommended major change pertaining to TP53-mutated (TP53^mut) MN. However, these assertions have not been specifically examined in therapy-related myeloid neoplasm (t-MN), a subset enriched with TP53^mut. We analyzed 488 t-MN patients for TP53^mut. At least one TP53^mut with variant allele frequency (VAF) ≥ 2% with or without loss of TP53 locus was noted in 182 (37.3%) patients and 88.2% of TP53^mut t-MN had a VAF ≥10%. TP53^mut t-MN with VAF ≥ 10% had a distinct clinical and biological profile compared to both TP53^mut VAF < 10% and wild-type TP53 (TP53^wt) cases. Notably, TP53^mut VAF ≥ 10% had a significantly shorter survival compared to TP53^wt (8.3 vs. 21.6 months; P < 0.001), while the survival of TP53^mut VAF < 10% was comparable to TP53^wt. Within TP53^mut VAF ≥ 10% cohort, the inferior outcomes persisted irrespective of the single- or multi-hit status, co-mutation pattern, or treatments received. Finally, survival of TP53^mut patients was poor across all the blast categories and MDS patients with >10% blasts had inferior survival compared to <5%. In summary, TP53^mut VAF ≥10% signified a clinically and molecularly homogenous cohort regardless of the allelic status.

Introduction

Tumor protein p53 (TP53) located on chromosome 17p13 is frequently mutated in cancer, including myeloid neoplasms (MN). Approximately 7–11% of de novo myelodysplastic syndromes (MDS) and 10–13% of acute myeloid leukemia (AML) patients harbor TP53 mutations (TP53^mut) [1–15]. TP53^mut MN are often associated with the features of genomic instability such as complex and monosomal karyotype (CK and MK, respectively) and are universally associated with adverse outcomes [1, 2, 4–11, 13, 15]. Therapy-related myeloid neoplasms (t-MN) are rare, but often fatal MN that develop following exposure to cytotoxic therapies [7, 12, 16] and are highly enriched in TP53^mut, CK, and MK [17].

The recently published 5th edition of the World Health Organization classification of MN (WHO-5) [18], and the International Consensus Classification (ICC) [19] recommended major reorganization of the MN. Among the congruent changes, both WHO-5 and ICC recognize the poor prognostic impact of biallelic TP53^mut defined by the presence of ≥2 mutations or 1 mutation with the loss of residual wild-type TP53 (TP53^wt). Both classifications consider variant allele frequency (VAF) ≥ 50% as presumptive evidence of biallelic/multi-hit TP53^mut, and biallelic TP53^mut MDS to be AML equivalent for therapeutic purposes. However, there are critical differences between the two classifications about allelic-status, VAF cut-off and blast categories [18, 19].

Another major change highlights the importance of genetic driver(s) and reduces the importance of the antecedent history and/or therapy. For example, ICC [19] removed t-MN as a distinct category and replaced it with a diagnostic qualifier, whereas WHO-5 [18] grouped t-MN with secondary MN as MN-post cytotoxic therapy (AML-pCT and MDS-pCT).

The underlying assumption of these changes is that TP53^mut MN are characterized by similar characteristics and outcomes. However, the studies driving these changes were highly enriched in de novo MN [2], excluded patients with <10% blasts [3] or >20% blasts [4], or only included MN with CK [14]. For example, in a cohort of predominantly of de novo MDS, single-hit TP53^mut had outcomes similar to TP53^wt, whereas the association with CK, high risk of AML transformation, and poor survival were limited to multi-hit TP53^mut patients [2]. Meanwhile, TP53^mut AML and MDS with excess blasts (MDS-EB) had equally poor survival irrespective of single or multi-hit TP53^mut status [3]. Furthermore, in MDS and AML with CK, the single- or multi-hit TP53^mut was the only disease-related factor predicting survival [14]. We recently demonstrated that TP53^mut t-MN is associated with poor survival irrespective of single or multi-hit status [20], suggesting that the prognostic impact of allelic loss of TP53^mut MN is context dependent.

Hence, we performed a comprehensive analysis of a TP53^mut t-MN cohort to: (i) define the genomic landscape of TP53^mut t-MN; (ii) study the interaction of TP53^mut with BM blast % and structural chromosomal changes; (iii) study the impact of 17p loss in the absence of a concurrent TP53^mut, and (iv) identify the optimal TP53^mut VAF threshold in t-MN.

Methods

This retrospective multi-center study was conducted by Mayo Clinic, Rochester (USA) and the South Australia MDS Registry (SA-MDS, Australia) and includes all t-MN patients who had conventional G-banding chromosome analysis (CBA) and mutation testing using targeted sequencing of the most recurrently mutated genes in MN (please refer to supplementary section for details) [12, 21]. The respective databases captured patient-level information that included diagnostic characteristics, treatments including the use of allogeneic stem cell transplant, the response to therapies, and long-term follow-up.

Integrated genomic analysis that includes acquired copy-number abnormalities (CNA) analysis based on NGS data, SNP-array and FISH was performed in a subset of patients.

Statistical methods

Comparisons were performed using Mann-Whitney U-test for non-normally distributed variables. Fisher’s exact test was used to determine associations between categorical variables. Overall survival (OS) was calculated from date of t-MN diagnosis to the last follow-up or the date of death. Post-transplant survival for patients who underwent allogeneic stem cell transplantation (SCT) was assessed from day of transplantation. Kaplan-Meier estimations were used with comparisons using log-rank tests. Cox regression multivariable analysis with backward selection was undertaken. P values < 0.05 were considered statistically significant. Further details of statistical analysis are provided in the Supplementary Methods section.

Data Sharing Statement

Additional methods and data can be found in the Supplementary Methods section. For original data, please contact devendra.hiwase@sa.gov.au or Shah.Mithun@mayo.edu.

Results

Patient cohort

This international cohort of 488 t-MN patients included 318 with t-MDS (65.2%) and 170 with t-AML (34.8%). The median age at t-MN diagnosis was 68 (IQR 60, 74) years. The most common primary cancers were lymphoproliferative disorders (n = 142, 29.1%), plasma cell neoplasms (n = 64, 13.1%), breast cancer (n = 61, 12.5%), and prostate cancer (n = 39, 8%). Most common DNA-damaging therapies used for treating the primary disease were chemotherapy (n = 230, 47.1%), chemotherapy plus radiotherapy (n = 160, 32.8%), autologous SCT (n = 95, 19.5%), or radiation therapy alone (n = 79, 16.2%, Table S1).

The median latency from the time of primary disease to t-MN diagnosis, was 81 (IQR 40, 149) months. Following t-MN diagnosis, most patients were treated with disease-modifying therapies (DMT) including hypomethylating agents (HMA) (n = 160, 33.8%), intensive chemotherapy (n = 100, 20.5%), venetoclax-based therapies (n = 70, 14.3%), and 91 (18.6%) patients underwent allogeneic SCT (Table S1).

We first analyzed genome-wide allelic imbalances to include arm-level alterations detected by CBA. In agreement with prior studies [22, 23], 365 (76%) had at least one chromosomal aberration including CK (n = 190, 39.3%), MK (n = 183, 37.90%), deletion 7q or monosomy 7 (n = 148, 30.6%), deletion 5q or monosomy 5 (n = 108, 22.4%), and deletion 17p across TP53 locus (referred to as 17p loss hereafter, n = 58, 12%) (Figure S1A). We next analyzed the somatic mutation landscape of TP53^mut in t-MN.

Genomic landscape of TP53^mut t-MN

The next generation sequencing (NGS) identified 253 putative oncogenic mutations in TP53 at VAF ≥ 2% in 182 (37.29%) patients (Fig. 1A, B). A diverse spectrum of TP53^mut including highly frequent missense mutations followed by frameshift insertion-deletions, splice-site, and nonsense mutations was observed (Fig. 1C, D and Figure S1B, C). Nearly all missense mutations occurred in the TP53 DNA-binding domain (Fig. 1C and Figure S1B).

Fig. 1. Landscape of TP53 mutation (TP53^mut) in therapy-related myeloid neoplasms (t-MN).

A Oncoplot showing cytogenetic and mutational landscape of TP53^mut and wild-type TP53 (TP53^wt) t-MN. Patient-related factors (t-MN phenotype, abnormal karyotype, primary disease, treatment status, and gender) are shown in the upper panel, and the distribution of somatic gene mutations (including TP53 mutation status) in the lower panel. Each column represents an individual patient, and the presence of the aberration or mutation is colored as indicated above; B Consort diagram of the mutant TP53 status of 488 t-MN patients analyzed by integrated analysis employing next gene sequencing (NGS), conventional cytogenetics, FISH, SNP-array and CNA analysis. *Of the 48 patients with single TP53^mut VAF 10–50% LOH information was available in 33 patients. Importantly, 80% of the remaining TP53^mut patients (n = 15) without LOH information has complex karyotype and are considered equivalent to multi-hit by ICC; C Distribution of TP53^mut along the gene. Mutations from single-hit patients are shown at the bottom and those from multi-hit patients are shown at the top. Missense mutations are shown as blue circles, truncated mutations corresponding to nonsense mutations as orange circles, frameshift deletions or insertions as red circles, and splice site variants are shown as purple circles. Other types of mutations are shown as green circles. Functional protein domains are indicated in yellow (transactivation motif), brown (transactivation domain 2), blue (DNA binding domain), and green (tetramerization motif); D Summary of TP53^mut separated by mutation type and frequency of the mutations.

In the majority of the TP53^mut t-MN with available information (n = 178; 97.8%), TP53^mut was a dominant driver clone with VAF ≥ 40% (n = 124; 69.7%), while 20 (11.2%), 14 (7.8%) and 20 (11.2%) patients had VAF ≥ 22–40%, 10–22% and <10%, respectively (Fig. 2A).

Fig. 2. TP53^mut drive genomic instability and was associated with poor overall survival in therapy-related myeloid neoplasms (t-MN).

A Distribution of cases according to TP53^mut VAF; B Volcano plot comparing cytogenetic aberration and somatic mutations in TP53^mut and TP53^wt t-MN. Chromosomal aberrancies highly prevalent in TP53^mut (red) and somatic mutations enriched in TP53^wt cohort (green). Genomic changes that are not differentially expressed between the two groups are shown in gray color; C Frequency of cytogenetic aberrations or driver oncogenic gene mutations in TP53^wt and TP53^mut t-MN; D Number of co-mutations in TP53^wt and TP53^mut t-MN; E Overall survival (OS) of TP53^mut with VAF ≥10% or loss of TP53 locus was significantly poor compared to wild-type TP53 (TP53^wt) and TP53^mut with VAF < 10% t-MN; F Multivariate Cox-regression analysis of factors predicting overall survival in t-MN; G OS of TP53^mut t-MN according to VAF cut-offs; H Frequency of loss of heterozygosity (LOH) and copy neutral LOH (cnLOH) according to number of TP53^mut; I Density estimation of VAF of single-hit and multi-hit TP53^mut; J OS is equally poor in single- and multi-hit in t-MN.

Next, we determined the TP53^mut VAF threshold associated with poor prognosis in t-MN. The receiver operating characteristics analysis identified 8% as the optimal VAF threshold (Figure S2A–C). As only 5 (2.3%) patients had VAF 8–10%, we adapted the ICC VAF threshold of 10% for further analyses (Figure S2C, D).

TP53^mut VAF ≥ 10% is associated with distinct presentation, features of genomic instability, and outcomes in t-MN

The majority of TP53^mut t-MN had VAF ≥10% (n = 158; 88.8%) (Fig. 1B). The loss of 17p across TP53 locus (LOH) or copy neutral LOH (cnLOH) was detected in 73 (38.4%) patients including TP53^mut VAF ≥10% (n = 60), VAF < 10% (n = 5), and in the absence of TP53^mut (n = 8). Collectively, 171 (35%) had TP53^mut VAF ≥ 10% or LOH/cnLOH across the TP53 locus.

We next compared cytogenetics abnormalities and somatic mutations in 30 genes analyzed in both (South Australian and Mayo) cohorts. Genomic instability was highly evident in TP53^mut with VAF ≥ 10% and/or loss of TP53 locus compared to TP53^wt t-MN. CK, MK, chromosome 5 aberrancies, and marker chromosomes were enriched in TP53^mut t-MN (Fig. 2B, C, Table S2). In contrast, recurrent oncogenic mutations such as ASXL1, DNMT3A, FLT3-ITD, IDH1, IDH2, NPM1, PTPN11, RAS, RUNX1 and TET2 were less frequent in TP53^mut t-MN (Figs. 1A, 2B, C, Table S2). Moreover, total number of co-mutations were significantly less in TP53^mut cases compared to TP53^wt (P < 0.001) (Table S2 and Fig. 2D). Enrichment of chromosomal aberrancies and lower frequency of somatic mutation was observed in TP53^mut complex karyotype MDS [4] and multi-hit TP53^mut compared to single-hit TP53^mut and TP53^wt de novo MDS [2]. TP53^mut not only influenced the genomic instability but also dictated the clinical presentation. TP53^mut t-MN had more severe anemia (P < 0.001), leukopenia (P < 0.001), and thrombocytopenia (P < 0.001) (Table S2).

Importantly, patients with TP53^mut VAF ≥ 10% with or without loss of TP53 locus had significantly shorter survival compared to TP53^wt (8.3 vs. 21.6 months; P < 0.001) (Fig. 2E). The three-year overall survival was 7% in patients with TP53^mut compared to 34% in TP53^wt patients (P < 0.0001). Poor survival of TP53^mut was observed in t-MDS (9.9 vs. 24.1 months; P < 0.001) and t-AML (3.6 vs. 13.2; P < 0.001) (Figure S3A, B). The inferior outcomes of TP53^mut persisted across all the t-MN treatment types, including supportive care (3.9 vs. 19.1; P < 0.0001), intensive chemotherapy (7.3 vs. 23.1; P < 0.0001), hypomethylating agents (10.9 vs. 20.5; P = 0.001), venetoclax-based combination therapies (8.1 vs. 23.3; P = 0.01) and allogenic SCT (20.6 vs. not reached; P = 0.01) (Figure S4A–F). Univariate Cox-regression analysis suggested that TP53^mut, specific chromosomal abnormalities, complex karyotype, bone marrow blasts, age and allogeneic SCT predicted OS of t-MN (Table S3). Furthermore, multivariable multivariate Cox regression analyses validated inferior survival of TP53^mut (HR 2.18, 95% CI 1.47–3.25; P < 0.001), independent of BM blast percentage (P = 0.003), chromosome 19 (P < 0.001) and allogeneic SCT (P < 0.001) (Fig. 2F). Poor prognosis with associated with TP53^mut and higher blast counts is known in AML and MDS, however prognostic impact of abnormalities in chromosome 19 are not well known. In contrast to de novo AML [24] and MDS [25], chromosome 19 was associated with poor survival in t-MN.

In contrast to TP53^mut with VAF ≥ 10%, majority of the clinical, cytogenetic, and mutation profile were similar between TP53^mut VAF < 10% and TP53^wt t-MN, except lower BM blasts and higher frequency of del 5q, CK, and ring chromosome. While somatic mutations were more prevalent in TP53^wt t-MN (Table S4). Importantly, OS was not significantly different between the two groups (26.8 vs. 21.6 months, P = 0.69; Fig. 2E). In contrast, chromosomal aberrancies and OS were significantly different in TP53^mut patients when categorized according to VAF < 10% vs. ≥10% (Fig. 2E and Table S5). Surprisingly, OS was equally poor in TP53^mut when VAF 10–22%, 22–40%, and ≥40% (6.15 vs. 7.96 vs. 8.78 months, Fig. 2G) cut-offs were used.

Majority of the TP53^mut t-MN harbor biallelic loss of TP53

Among the 158 patients with TP53^mut VAF ≥ 10%, 70% had TP53^mut plus LOH/cnLOH of TP53 locus (n = 52; 32.9%) or ≥2 TP53^mut (n = 59; 37.3%), while 29.1% (n = 47) had single TP53^mut (Fig. 2H). Of the 47 patients with single TP53^mut, 26 (55.3%) and 21 (44.7%) patients had VAF > 50% and 10–50%, respectively (Fig. 1B). Additional 13 patients had loss of the TP53 locus without evidence of TP53^mut (n = 8) or with TP53^mut VAF < 10% (n = 5) (Fig. 1B). Frequency of LOH/cnLOH was significantly higher in cases with single TP53^mut compared to cases with ≥2 TP53^mut (33% vs. 5%, P = 0.0001; Fig. 2H).

Next, we compared the clinical features, profiles of genome stability and patterns of co-mutation for each TP53 allelic state. Integrated cytogenetic, copy number and somatic mutation analysis classified TP53^mut as multi-hit if there is: (1) presence of ≥2 distinct TP53^mut, each with VAF ≥ 10%, or (2) a single TP53^mut associated with either: (i) cytogenetic deletion of 17p13 involving the TP53 locus; (ii) a VAF of >50%; or (iii) copy-neutral loss of heterozygosity (cnLOH) at the TP53 locus. Single TP53^mut with VAF 10%-50% or loss of 17p13 involving TP53 locus without TP53^mut were defined as single-hit. In total, 34 (19.9%) of the 171 patients with TP53^mut and/or loss of TP53 locus were considered single-hit and 137 (80.1%) were multi-hit (Fig. 1B). In single-hit TP53^mut cases, the median VAF was significantly lower compared to multi-hit TP53^mut (34% vs. 38.2%, P = 0.006) (Table S6 and Fig. 2I). Overall, the spectrum of TP53^mut was shared among single- and multi-hit states (Fig. 1C).

Unlike de novo MDS [2], there was no significant difference in CK, MK, CK plus MK, chromosome 5 aberrancy, or co-mutations between single- and multi-hit TP53^mut (Table S6). We did not observe significant differences in clinical features, age, latency, blood counts, BM blast percentage and cytogenetics when stratified by the allelic status except that the multi-hit TP53^mut were enriched for marker chromosome. In contrast to previous publications [2, 14], the distribution of single- and multi-hit TP53^mut was not different across the t-MN phenotype and the BM blast categories (Figure S5A, B).

Consistent with our previous observation [20], the OS was not significantly different between the single- and multi-hit TP53^mut t-MN (Fig. 2J). Similarly, there was no survival difference between single- and multi-hit TP53^mut when stratified by t-MDS vs. t-AML, according to the blast cut-off proposed by ICC (Figure S6A–D), or the type of treatment received (Figure S7A–C). Finally, there was no difference in the incidence of progression to AML in single versus multi-hit TP53^mut t-MDS (Figure S7D).

TP53^mut burden increases with number of chromosomal aberrancies

The proportion of TP53^mut increased from 4.5% in normal karyotype cases to 17.3% in cases with two chromosomal aberrancies (P = 0.019; Fig. 3A) and 76.8% in cases with CK (P < 0.0001). Even within the CK group, enrichment of TP53^mut was observed with the increasing number of cytogenetic abnormalities: from 26.3% in cases with three chromosomal abnormalities to 75%, 96.6%, and 94% in cases with 4–6, 7–9, and >9 chromosomal aberrancies (Fig. 3A and Fig. S8). More than 80% of TP53^mut were segregated in t-MN with >4 cytogenetic abnormalities (Figure S8). Furthermore, TP53^mut were significantly high in typical- compared to atypical-CK (Fig. 3B). Typical CK is defined as CK with ≥3 abnormalities that include 5q, 7q, and/or 17p loss and atypical-CK as CK with ≥3 abnormalities without these specific abnormalities [26]. Conversely, TP53^mut were enriched for CK compared to TP53^wt (84.8% vs. 12.0%, P < 0.0001; Fig. 3C).

Fig. 3. TP53^mut was enriched in complex karyotype (CK) and is associated with significantly poor outcome of CK t-MN.

A Frequency of TP53^mut according to number of structural cytogenetic abnormalities; B High frequency of TP53^mut in typical-CK compared to atypical-CK t-MN; C TP53^mut cases were enriched for CK; D In the absence of CK, TP53^mut were prevalent in cases with 5q loss compared to 7q loss (+, present; -; absent; +/-, present or absent). Loss of 5q was defined as monosomy 5 or del 5q, while loss of 7q was defined as monosomy 7 or del 7q; E CK is associated with significantly poor survival in t-MN; F TP53^mut status further stratify CK with very poor outcome.

The enrichment of TP53^mut was also noted in cases with del 5q without CK. However, such enrichment of TP53^mut was not observed in del 7q without CK (Fig. 3D). Together these findings indicate that TP53^mut burden increases not only with number but with type of chromosomal aberrancies.

Complex karyotype is associated with poor survival (10.1 vs. 19.5 months; P < 0.001) (Fig. 3E). We assessed if TP53^mut can further stratify by CK t-MN. Even within CK, TP53^mut is associated with a higher structural genomic instability. In particular, MK, marker chromosomes, ring chromosome, 5q del/monosomy 5, chromosome 12, and 18 abnormalities were highly prevalent in TP53^mut-CK compared to TP53^wt-CK (Table S7). While number of somatic mutations, including RAS, ASXL1, and RUNX1 were enriched in TP53^wt-CK (Table S7). Importantly, TP53^mut further stratified outcomes for CK, with inferior survival of TP53^mut-CK compared to the TP53^wt-CK (8.3 vs. 19.7 months; P < 0.001, Fig. 3F). Conversely, CK-status did not influence the poor outcome of TP53^mut t-MN (8.3 vs. 7.7 months, P = 0.29; Fig. 3F). Furthermore, OS of TP53^wt CK was not significantly different than non-CK TP53^wt. Together these findings suggest that poor prognosis of CK is driven by its association with prognostically adverse TP53^mut (Fig. 3F). Similar association was previously reported in CK-MDS [4].

TP53^mut t-MDS can be stratified according to ICC BM blast categories

Next, we evaluated the frequency of TP53^mut according to the disease phenotype. TP53^mut burden was significantly higher in t-MDS compared to t-AML (40.3% vs. 29.3%, P = 0.021; Fig. 4A). However, the frequency of TP53^mut in t-MDS was similar across the BM blast categories: 37.3 vs. 38.2 vs. 38.1% in <5%, 5–9% and 10–19% blasts, respectively (Fig. 4B and Table S8). There was no significant difference in age at t-MN diagnosis, latency, the type of the primary disease, or the degree of cytopenia at t-MN diagnosis across the blast % categories (Table S8). Moreover, genomic instability as evidenced by CK, MK, and number of cytogenetic abnormalities were also similar across all four categories (Table S8, Fig. 4C). Finally, the proportion of patients with single- vs. multi-hit, TP53 VAF, and co-mutations were comparable across the four blast categories (Table S8).

Fig. 4. Interaction between TP53^mut and bone marrow (BM) blast percentage.

A Distribution of t-AML and t-MDS according to TP53^mut status; B TP53^mut frequency in t-MDS was similar across blast categories at diagnosis; C Chromosomal aberrancies and TP53^mut allelic status across BM blast %; D Multivariate Cox-regression analysis showing BM blast >10%, chromosomal 18/19 abnormalities and disease modifying therapies were independent predictors of TP53^mut OS; E In TP53^mut t-MN, BM blast 10–19% and ≥20% were associated with poor OS compared to BM blast <5%; (F) OS of TP53^mut t-MN according to ICC blast categories (0–9% vs. 10–19% vs. ≥20%).

Due to the molecular homogeneity of TP53^mut across the blast categories, we investigated prognostic impact of BM blast %, variants in cancer-related genes, cytogenetics, and the severity of cytopenia in TP53^mut t-MN. BM blasts 10–19%, ≥20%, hemoglobin, chromosome 19 abnormalities, and the first line therapies including allogeneic SCT were independent predictors of outcome in TP53^mut t-MN (Fig. 4D; Table S9). Importantly, OS of TP53^mut patients was poor across the blast categories with statistically significant difference between the BM blasts <5% vs. 10–19% and <5% vs. ≥20% categories (Fig. 4E). The OS of TP53^wt t-MN significantly worsened with increasing blast % (P = 0.001) (Figure S9).

Recently, ICC proposed three categories of TP53^mut MN according to BM blast % [19]. OS was significantly different in t-MDS with blast 0–9% vs. 10–19% blasts (10.9 vs. 8.2 months; P = 0.01). Similarly, OS was significantly different in TP53^mut t-MN with BM blast 0–9% vs. ≥20% (10.9 vs. 4.6 months, P = 0.0008) (Fig. 4F). Together these findings suggest that TP53^mut t-MN with >10% is uniformly associated with poor survival.

Discussion

The results above collectively demonstrate that the classification of t-MN based on TP53 status is clinically and biological relevant. This conclusion is supported by our findings that (1) TP53^mut t-MN with ≥10% is uniformly associated with poor survival; (2) the presence of TP53^mut was an independent risk factor for poor survival and was associated with inferior outcome, even within the traditionally known high-risk subsets such as CK; (3) the inferior outcomes persisted irrespective of the single- or multi-hit status, the co-mutation pattern, or the treatments received; (4) poor survival of single and muti-hit TP53^mut was observed across the blast categories of t-MN.

TP53^mut influenced the clinical presentation and survival. Although severe cytopenia is reported across multiple studies [2, 4, 14], the association with BM blast percentage is debated [2, 4, 10, 27]. TP53^mut de novo MDS present with high BM blast [4, 27], especially those with the multi-hit alterations [2]. However, MDS with TP53^mut/CK present with low BM blast burden [14]. We observed that TP53^mut patients were more likely present as t-MDS. Furthermore, within the t-MDS cohort, BM blast burden was lower in TP53^mut cases. In contrast to de novo MDS [2], frequency of multi-hit TP53^mut remained similar across the BM blast categories. Secondly, poor prognosis of multi-hit TP53^mut MDS/AML was reported irrespective of the BM blast and therapy-relatedness [14]. In our cohort, OS of TP53^mut t-MN was poor across all the blast categories irrespective single- or multi-hit status and type of disease modifying therapies. However, OS of TP53^mut MDS BM blasts >10% and AML was significantly inferior compared to TP53^mut BM blasts <10%. Thus, the interaction between BM blast and TP53^mut appears to be disease ontogeny specific. Overall, our findings support the ICC stratification of TP53^mut MN using three blast cut-offs.

Though poor survival of TP53^mut is well known, the prognostic implication TP53^mut VAF remains an active area of research. Multiple studies reported VAF > 40% is associated with poor survival in high-risk MDS [4, 10, 28] while one study suggested that poor survival is a direct function of increasing VAF as a continuous variable [7]. In contrast to these findings, other studies showed an inferior OS irrespective of TP53^mut VAF [3, 14, 29, 30]. Furthermore, prognostic implication of TP53^mut VAF also depend upon TP53^mut allelic status. Single-hit TP53^mut MDS with VAF > 22% had poor survival and the favorable survival comparable to the TP53^wt was restricted to the single-hit cases with VAF ≤ 22%. Conversely multi-hit patients had poor outcome across the range of TP53 VAF [2]. In t-MN, OS was significantly poor in TP53^mut with VAF ≥ 10% compared to VAF < 10%. There was no survival difference in cases with VAF 10–22% vs. >22–40% vs. >40%. Together these findings suggest prognostic implication of TP53^mut VAF is context dependent and varies significantly between de novo and therapy-related MN.

As expected, TP53^mut and CK/MK were highly enriched in t-MN compared to de novo MDS [2] and AML [3, 5, 6, 31]. Within t-MN, CK was more frequent with TP53^mut than TP53^wt. Conversely, increasing genomic instability was associated with enrichment of TP53^mut: 75% and ~90% of patients with ≥4–6 and ≥7 chromosomal abnormalities harbored TP53^mut, respectively, compared to only 26% of patients with 3 chromosomal abnormalities. In addition to number, type of chromosomal abnormalities also influences enrichment of TP53^mut. Critical understanding of the relationship between TP53^mut and chromosomal aberrancies can be harnessed for prioritization of TP53^mut testing in limited resources, and screening/counselling appropriate patients for clinical trials of novel therapies as waiting time for mutation results can be up to 3–4 weeks. Importantly, within CK t-MN, patients with co-existent TP53^mut had evidence of profound genomic complexities and structural aberrancies [14, 32], and had poor outcomes compared to CK-TP53^wt in our t-MN cohort and other MN [10], emphasizing the importance of assessing TP53^mut alongside complex karyotype for an accurate risk estimation. Conversely, poor prognosis of biallelic TP53 loss compared to single allelic loss was evident only in non-CK, while presence of CK was invariably associated with poor survival irrespective of TP53^mut allelic status [33].

In contrast to the findings in predominantly de novo MDS, we did not observe a difference in the frequency of structural chromosomal aberrancies including CK, MK, or co-mutation pattern between single- and multi-hit TP53^mut in t-MDS and as such in the whole t-MN cohort. Moreover, there was no difference in the transformation rate to AML and the OS between the multi- and the single-hit TP53^mut t-MDS. These findings have significant impact on the classification and management of t-MN patients especially considering the other recent changes in the WHO classification. The WHO has grouped t-MN with secondary MN and renamed it as “myeloid neoplasm post cytotoxic therapy”, with the assertion that a majority of MDS and AML occurring post-cytotoxic therapy have TP53^mut and that only multi-hit TP53^mut had a poorer outcome compared to single-hit [2, 18], thus undermining the poor prognosis of single-hit TP53^mut t-MN. Exclusion of single-hit TP53^mut t-MDS from the TP53 mutated MDS have huge impact on management such as consideration for allogeneic stem cell transplantation [34], and exclusion from clinical trials targeted toward TP53^mut MDS. For example, allogeneic SCT may not be offered to fit single-hit TP53^mut t-MDS with BM blast 5–9% (according to ICC) and <20% (according to WHO) as they are considered to have OS similar to TP53^wt MDS. Similarly, these patients would be excluded from enrolment in clinical trials.

The apparent discrepancies in the conclusions among prior studies can be on the account of the characteristics of the study cohort, technical aspects, as well as true biological differences. For example, t-MN constituted a small subset of the patients in some large studies [2, 4]. Others excluded patients with low blasts [3], or only included patients with CK [4, 14]. Secondly, variable VAF thresholds have been used: 1% [3], 2% [2, 14], or 10% in the ICC guidelines. Thirdly, and likely the most significant difference is the criteria used to designate single vs. multi hit status. For example, in the absence of a detailed analysis of the TP53 locus (using CBA, FISH, or chromosomal microarray), both WHO and ICC consider >50% VAF as presumptive evidence multi-hit, though the evidence suggests that VAF estimation is a poor surrogate for the allelic status [2]. In the absence of LOH information, the presence of a single TP53^mut in the context of CK is considered equivalent to a multi-hit TP53^mut by ICC, but not in WHO-5. Key studies driving recent classifications variably used CBA, FISH, NGS, and SNP array to determine the hit status [2, 3]. In contrast, a recent study used WGS to confer hit status [35]. Therefore, what LOH assessment is considered minimal or optimal is unclear at this time. In summary, there is an urgent need to define a uniform diagnostic genetic tools and criteria for determining allelic status of TP53^mut and VAF in all types of MN including t-MN, secondary and de novo MN.

Finally, the universally poor outcome of TP53^mut t-MN highlights the great unmet need for these patients and the necessity for novel therapies. Unfortunately, none of the routinely offered leukemia-directed therapies were able to overcome the impact of TP53^mut. TP53^mut are noted to be present long before the eventual development of t-MN—sometimes even before the original exposure to DNA-damaging therapies. Recent evidence suggests a deterministic order of genetic and genomic changes following TP53 mutation/loss [36, 37]. A comprehensive characterization of the genomic changes, and its correlation with the resultant morphological changes, may help identify patients at the risk of imminent leukemic transformation and devise effective preventive strategies. Hence, the identification of individuals at high-risk of developing t-MN and employment of preventative approaches may improvement outcomes for this difficult-to-treat group of patients [16, 38–40].

Author contributions

MVS designed the study, contributed the patient data, and edited the manuscript; ET and RC collated the data, analyzed the data, and edited the manuscript; SS, AB, and CS abstracted patient data; DL contributed to cytogenetic analysis; DC reviewed pathology; PG contributed to cytogenetics and chromosomal array analyses; CK performed statistical analysis and edited the manuscript; CH, ET, ALB, and HS contributed to variant annotation and edited the manuscript; DT edited manuscript; SK contributed DDR expertise and edited the manuscript; AA, HA, NG, MP, and AT contributed patients and edited manuscript. DH designed the study, contributed the patient data, analyzed the data, and wrote the manuscript. All authors agree to the final version of the manuscript.

Competing interests

MVS—Research funding to the institution from AbbVie, Celgene, MRKR Therapeutics, and Astellas; AA—Research support to institution (Novartis, Astex); MP—Membership on an entity’s Board of Directors or advisory committees (Stemline Therapeutics) and Research funding (Kura Oncology); DH—Membership on an entity’s Board of Directors or advisory committees (AbbVie, Novartis). All other authors disclose no conflicts of interest.

Supplementary information

The online version contains supplementary material available at 10.1038/s41408-023-00821-x.