Evolutionary action score identifies a subset of TP53 mutated myelodysplastic syndrome with favorable prognosis

Introduction
 TP53 mutations (TP53MT) are seen in ~10% of myelodysplastic syndromes (MDS). TP53MT are distributed across the entire coding region, with less than a third occurring in focal hotspots. In vitro and in silico studies suggest that different types of TP53MT lead to distinct functional consequences that include oncogenic gain-of-function and protein loss-of-function with dominant-negative effect. The functional effects of these mutations likely influence disease biology and outcome, either independently or by influencing known variables such as VAF and karyotype. While studies have shown that TP53MT MDS with complex/monosomal karyotype (CK/MK) and multiallelic TP53 alterations have a worse prognosis, however, the impact of different types of TP53MT on phenotype, prognosis and outcome of MDS is not known. To assess this, we quantified the deleterious effects of missense TP53MT using the computationally-derived evolutionary action score (EAp53, range, 0-100, higher score indicates worse impact), followed by 3D protein mapping to identify prognostic subsets.
 Methods
 We selected 270 consecutive newly-diagnosed MDS and oligoblastic AML with at least 1 missense TP53MT by NGS. EA53 scores were determined using evolutionary trace approach. TP53 immunohistochemistry (IHC) was performed on selected cases. The optimal EAp53 cutoff was determined using recursive partitioning and regression trees (RPART) based on Classification & Regression Trees. TP53 protein structural analysis was conducted using the PyMOL molecular visualization system and the crystal structure of the TP53 core domain in complex with DNA (PDB ID of 4HJE).
 Results
 The median age was 68 (18-90)]. Majority (81%) had a CK. The median EAp53 score was 79 (4-98) [Fig 1A, 1B]. Using RPART, we identified an EAp53 score >52 predicted for worse OS (median, 10 vs. 48 months; HR: 2.6 [1.22-5.56]; p=0.01) [Fig 1C]. We divided our cohort into low-EAp53 (≤52; n=17, 6%) and high-EAp53 (n=253, 94%). Low-EA MDS had fewer cytogenetic abnormalities [median, 3 vs. 7; p=0.019], lower frequency of CK/MK (p=0.02), lower number of additional TP53MT (6% vs. 32%, p=0.027), lower frequency of multiallelic TP53 alterations (29% vs. 63%, p=0.009), and higher number of additional gene mutations (63% vs. 33%; p=0.05) involving NRAS and RUNX1 genes (p=0.02). There was no difference in median TP53 VAF or R-IPSS scores (Table 1). By TP53 immunohistochemistry, TP53 protein expression was significantly different between wild-type (median H-score, 6), low EAp53 (48) and high EAp53 (158) [wild-type vs. low EA, p=0.04; low vs. high EA, p=0.0014]. Low EAp53 showed clearance of TP53 mutation in 67% (vs. 45%) in those that achieved partial/ complete response.
 Due to observed differences in outcome despite similar EAp53 score, we correlated survival with the mutant location within 3D protein structure. Majority of mutations mapped to the evolutionarily important sites of the TP53 core domain, residues near the DNA binding site or within the protein structural core and solvent inaccessible (Fig 1D). We divided our cohort into two groups based on a survival cut-off of 10 months. TP53MT associated with OS <10 months formed two clusters: a large cluster interfacing with the DNA binding site and a small cluster formed by residues V157, Y220, L257 and E258 (Fig 1E).
 By univariate analysis, the following associated with worse OS: higher TP53 VAF (as a continuous variable), higher number of TP53MT, high-risk EAp53 (>52) group, higher IPSS-R score, presence of CK/MK, higher serum LDH and creatinine levels, lower platelet, hemoglobin, and serum albumin. TP53 allele state and del(17p) did not associate with OS. By multivariable analysis (CK excluded due to due to a strong association with EAp53), high-EAp53 was associated with worse OS independent of R-IPSS score [HR 5.1; CI 1.5-17.2; p=0.009]. Neither TP53 VAF nor the number of TP53MT was independently prognostic.
 Conclusions
 A subset of TP53MT MDS patients with low-EAp53 (≤52) showed improved outcomes independent of TP53 VAF or IPSS-R scores in HMA treated MDS, and associated with specific clinico-pathologic features. High-EAp53 associated with worse OS independent of R-IPSS score. Mutational mapping using 3D protein model showed clustering of poor-outcome mutations, suggesting that structural location further influences the outcome. Combination of EAp53 and 3D mapping can be help identify prognostic subsets.
 Figure 1
 
 
 Sasaki: Novartis: Consultancy, Research Funding; Pfizer Japan: Consultancy; Daiichi Sankyo: Consultancy; Otsuka: Honoraria. Jabbour:AbbVie: Other: Advisory role, Research Funding; Adaptive Biotechnologies: Other: Advisory role, Research Funding; Amgen: Other: Advisory role, Research Funding; Takeda: Other: Advisory role, Research Funding; Pfizer: Other: Advisory role, Research Funding; Genentech: Other: Advisory role, Research Funding; BMS: Other: Advisory role, Research Funding. Kadia:Celgene: Research Funding; Cyclacel: Research Funding; BMS: Honoraria, Research Funding; Pfizer: Honoraria, Research Funding; Novartis: Honoraria; Ascentage: Research Funding; JAZZ: Honoraria, Research Funding; Incyte: Research Funding; Amgen: Research Funding; Genentech: Honoraria, Research Funding; Abbvie: Honoraria, Research Funding; Pulmotec: Research Funding; Astellas: Research Funding; Cellenkos: Research Funding; Astra Zeneca: Research Funding. Andreeff:Centre for Drug Research & Development; Cancer UK; NCI-CTEP; German Research Council; Leukemia Lymphoma Foundation (LLS); NCI-RDCRN (Rare Disease Clin Network); CLL Founcdation; BioLineRx; SentiBio; Aptose Biosciences, Inc: Membership on an entity's Board of Directors or advisory committees; Daiichi-Sankyo; Breast Cancer Research Foundation; CPRIT; NIH/NCI; Amgen; AstraZeneca: Research Funding; Daiichi-Sankyo; Jazz Pharmaceuticals; Celgene; Amgen; AstraZeneca; 6 Dimensions Capital: Consultancy; Amgen: Research Funding. Short:Takeda Oncology: Consultancy, Honoraria, Research Funding; AstraZeneca: Consultancy; Astellas: Research Funding; Amgen: Honoraria. Daver:Pfizer: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Karyopharm: Research Funding; Servier: Research Funding; Genentech: Research Funding; AbbVie: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Astellas: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Novimmune: Research Funding; Gilead: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Amgen: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Trovagene: Research Funding; Fate Therapeutics: Research Funding; ImmunoGen: Research Funding; Novartis: Consultancy, Membership on an entity's Board of Directors or advisory committees; Celgene: Consultancy, Membership on an entity's Board of Directors or advisory committees; Jazz: Consultancy, Membership on an entity's Board of Directors or advisory committees; Trillium: Consultancy, Membership on an entity's Board of Directors or advisory committees; Syndax: Consultancy, Membership on an entity's Board of Directors or advisory committees; Amgen: Consultancy, Membership on an entity's Board of Directors or advisory committees; KITE: Consultancy, Membership on an entity's Board of Directors or advisory committees; Agios: Consultancy, Membership on an entity's Board of Directors or advisory committees; Daiichi Sankyo: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Bristol-Myers Squibb: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding. Borthakur:BioLine Rx: Consultancy; BioTherix: Consultancy; Nkarta Therapeutics: Consultancy; Treadwell Therapeutics: Consultancy; PTC Therapeutics: Consultancy; Argenx: Consultancy; FTC Therapeutics: Consultancy; Curio Science LLC: Consultancy; Oncoceutics: Research Funding; Xbiotech USA: Research Funding; Polaris: Research Funding; AstraZeneca: Research Funding; BMS: Research Funding; BioLine Rx: Research Funding; Cyclacel: Research Funding; GSK: Research Funding; Jannsen: Research Funding; Abbvie: Research Funding; Novartis: Research Funding; Incyte: Research Funding; PTC Therapeutics: Research Funding. Ravandi:Amgen: Consultancy, Honoraria, Research Funding; Astellas: Consultancy, Honoraria, Research Funding; BMS: Consultancy, Honoraria, Research Funding; AstraZeneca: Consultancy, Honoraria; Xencor: Consultancy, Honoraria, Research Funding; Jazz Pharmaceuticals: Consultancy, Honoraria, Research Funding; Orsenix: Consultancy, Honoraria, Research Funding; Macrogenics: Research Funding; Abbvie: Consultancy, Honoraria, Research Funding; Celgene: Consultancy, Honoraria. Kantarjian:Immunogen: Research Funding; Jazz: Research Funding; Novartis: Honoraria, Research Funding; Aptitute Health: Honoraria; Actinium: Honoraria, Membership on an entity's Board of Directors or advisory committees; Adaptive biotechnologies: Honoraria; Oxford Biomedical: Honoraria; Delta Fly: Honoraria; BioAscend: Honoraria; Daiichi-Sankyo: Honoraria, Research Funding; Pfizer: Honoraria, Research Funding; Sanofi: Research Funding; Janssen: Honoraria; Abbvie: Honoraria, Research Funding; Amgen: Honoraria, Research Funding; Ascentage: Research Funding; BMS: Research Funding. Garcia-Manero:Genentech: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; H3 Biomedicine: Research Funding; AbbVie: Honoraria, Research Funding; Acceleron Pharmaceuticals: Consultancy, Honoraria; Celgene: Consultancy, Honoraria, Research Funding; Jazz Pharmaceuticals: Consultancy; Helsinn Therapeutics: Consultancy, Honoraria, Research Funding; Astex Pharmaceuticals: Consultancy, Honoraria, Research Funding; Bristol-Myers Squibb: Consultancy, Research Funding; Amphivena Therapeutics: Research Funding; Merck: Research Funding; Novartis: Research Funding; Onconova: Research Funding.



Dear Editor,
The prognosis of TP53-mutated myelodysplastic syndromes (MDS) can be heterogeneous. TP53-mutated MDS with low variant allele frequency (VAF), without complex karyotype (CK), and those with mono-allelic TP53 alterations have significantly improved outcomes [1][2][3] . TP53 mutations are diverse and distributed across the codons of the entire coding region 4 . Different types of TP53 mutations lead to distinct functional consequences (such as oncogenic gain-of-function, protein loss-offunction with dominant-negative effect etc [5][6][7], that likely influence disease biology and outcome, either independently or by influencing known variables such as VAF and allelic state 2,3 . Until now, the relationship between various TP53 mutations and genomic/phenotypic features including outcomes is not wellcharacterized. This knowledge is important to assess the efficacy of novel therapeutic strategies that restore TP53 function 8 . Evolutionary Action score (EAp53) is a computationally-derived score to quantify the deleterious impact of different missense TP53 mutations based on (A) phylogenetic divergence of the mutated sequence position [evolutionary trace (ET)] and (B) perturbation due to amino acid (AA) substitution 9 . EAp53 score ranges between 0 and 100, a higher score indicates a worse impact, and 0 indicates wild-type function. EAp53 score has been shown to be an objective, reliable prognostic biomarker in patients with head and neck (H&N) and colorectal cancers [10][11][12][13] . Here, we used the EAp53 scoring system to evaluate the impact of different types of missense TP53 mutations on clinico-pathologic and genomic features in MDS.
We identified 270 patients with newly-diagnosed MDS or oligoblastic AML (<30% blasts) with ≥1 missense TP53 mutation(s) at baseline detected by next-generation sequencing (Fig. 1A). The median TP53 VAF was 33.9 (1-94.4); 165 (61%) had multi-allelic TP53 alterations. Majority were treated with hypomethylating agents (HMA). Informed consent was obtained, the study was performed per institutional-approved protocols in accordance with the Declaration of Helsinki. See Supplementary Materials for detailed methodology.
10 months (p = 0.01; HR: 2.6 [1.22-5.56]). EAp53 score of 75, previously described in TP53-mutated H&N squamous cell carcinoma, did not show a survival difference in MDS. By univariate analysis, high-EAp53 (>52), TP53 VAF, number of TP53 mutations, IPSS-R score, CK/monosomal karyotype (MK), higher serum LDH and creatinine, lower platelet, hemoglobin, and serum albumin associated with worse OS. Neither TP53 allele state nor del(17p) associated with OS. By multivariable analysis, the EAp53 risk retained the independent predictive value for OS along with IPSS-R score and serum albumin, but not TP53 VAF or the number of TP53 mutations (CK excluded due to a strong association with EAp53 score; Fig. 1D; Table S2). EAp53 risk was the only independent predictor of AML transformation. EAp53 risk did not affect transformation-free survival, relapse-free survival, overall response, and complete remission rates (Table S3). showing the frequency distribution of missense TP53 mutations and associated concurrent non-missense mutations. B Spectrum of EAp53 scores of the TP53 mutations noted within our MDS cohort: the majority had a high (>52) EAp53 score. C Using RPART, an EAp53 score of 52 provided an optimal cut-off based on overall survival in MDS patients. D The multivariate model identified EAp53 score, R-IPSS risk score, and serum bilirubin to be an independent predictor for worse overall survival. E Mutational frequencies of genes in the cohort separated by EAp53 risk category. Low-risk EAp53 MDS patients had a significantly higher frequency of mutations in NRAS and RUNX1, and a trend for higher frequencies in NPM1, WT1, and ASXL1 mutations.
The observed distinctive clinical, cytogenetic, and mutation characteristics provide support to the clinical validity of EAp53 scoring and confirm that low and high-EAp53 do not reflect different positions on the early to late disease trajectory. The presence of at least 1 additional gene mutation, frequently NRAS, in low-EA-MDS corroborates the leukemogenic role of RAS. These additional hits potentially modify the phenotype and outcome of low-EA-MDS. The need for additional hits in high-EA-MDS is abrogated by chromosomal aneuploidies, involving chromosomes 17 and 5, that harbor negative regulators of the RAS pathway 14 .
Since protein function is further modulated by the 3D location of the residue, we performed protein structural analysis using the crystal structure of the TP53-coredomain in complex with DNA (PDB ID: 4HJE; PyMOL molecular visualization). We hypothesized that this may explain the variable survival rates noted in some high-EA-MDS patients with similar EAp53 scores. All the TP53 mutations of this cohort mapped to the evolutionarily important sites of the TP53-core-domain. When segregated based on survival of 10 months, TP53 variants with poor-survival (OS < 10 months) formed two clusters: a large cluster interfacing the DNA-binding site and a small cluster formed by residues V157, Y220, L257, and E258, showing that structure location further stratifies the outcome (Fig. 2E, F). Analysis with different survival cut-offs yielded the same results.
Finally, we verified the biological relevance of EAp53 scoring using other independent computational methods. CADD and REVEL segregated the same prognostic subgroups (but not DANN, Polyphen 2, MutPred, PROVEAN, SIFT; Fig. S3). To validate the EAp53 cut-off of 52, we used an independent single-center cohort of 62 MDS patients, selected using the same criteria and treated using HMAs. There were 3 (5%) low-EA-MDS patients [p. Y220H, p.F134L, p.R209W] with a longer median OS (112 vs. 32 months, p = 0.25) compared to high-EA-MDS (Fig.  S4). CADD and REVEL could not separate these patients, suggesting that the EAp53 method was superior. When study and validation cohorts were merged, all 3 methods were concordant [EAp53, p = 0.0103; REVEL, p = 0.03; CADD p = 0.006; Fig. S5].
The study has a few limitations. Although this is a large retrospective study, the inherent low frequency of low-EAp53 MDS (~6%) warrants validation in multi-center cohorts. While the possibility that some of the low-EAp53 variants represent rare single nucleotide polymorphisms (SNP) cannot be completely excluded, to the best of our knowledge, all low-EAp53 variants were clinically reported by the laboratory after extensive curation using literature, online databases including COSMIC, dbSNP, 1000 genome, EXAC, ClinVar and in-silico prediction tools. Repeat NGS on 9 (53%) patients showed clearance or significant variations in the TP53 VAFs, strongly suggesting somatic origin. TP53 VAF was not independently prognostic in this study. Unlike other reports 1,2 , we note that this cohort is unique because it excluded patients with nonsense/frameshift TP53 mutations that are likely to have higher VAF and multi-allelic TP53 alterations due to a loss-of-function phenotype. Further, VAFs were not normalized based on copy number. The study did not assess copy-neutral loss-of-heterozygosity that could explain the lack of association with TP53 allele status.
In conclusion, this is the first study to show the independent prognostic value of the EAp53 score, thereby expanding the previously established genomic attributes impacting the outcomes of TP53-mutated MDS 1-3 . While VAF and karyotype are dependent on the aspirate quality (often compromised by fibrosis in TP53-mutated MDS) and vary with disease evolution and therapy, EAp53 score is mutation-dependent, stable predictive biomarker, not influenced by therapy or time for baseline riskstratification. These findings are important in lieu of novel mutation type-specific therapeutic strategies 7,15 . Low-EAp53 mutants may benefit from strategies that utilize residual TP53 function while small molecules, such as APR-246 and COTI-2, which restore TP53 function may be appropriate for high-EAp53 mutants 8 . Together with structural mapping, the EAp53 score can guide treatment. Overall, the study shows that the EAp53 score can identify prognostic subsets within TP53-mutated MDS and facilitate a personalized therapeutic approach.