TP53 Mutation Status in Myelodysplastic Neoplasm and Acute Myeloid Leukemia: Impact of Reclassification Based on the 5th WHO and International Consensus Classification Criteria: A Korean Multicenter Study
2025; 45(2): 160-169
Ann Lab Med 2025; 45(2): 170-177
Published online December 16, 2024 https://doi.org/10.3343/alm.2024.0194
Copyright © Korean Society for Laboratory Medicine.
Department of Laboratory Medicine, Korea University Anam Hospital, Korea University College of Medicine, Seoul, Korea
Correspondence to: Jiwon Yun, M.D., Ph.D.
Department of Laboratory Medicine, Korea University Anam Hospital, Korea University College of Medicine, 73 Goryeodae-ro, Seongbuk-gu, Seoul 02841, Korea
E-mail: drjwyun@kumc.or.kr
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background: In 2022, the revised WHO classification and International Consensus Classification (ICC) for myeloid neoplasms were published. We examined the impact of these guidelines on AML diagnoses alongside the 2022 European LeukemiaNet (ELN) recommendations on risk stratification.
Methods: We included 450 adult patients with newly diagnosed AML (non-acute promyelocytic leukemia) from the cBioPortal open-source dataset. Diagnoses and risk stratifications were revised based on the new guidelines and compared with the 2017 WHO classification. Survival analyses were performed using Cox regression.
Results: Among the patients included, 190 (42.2%) had consistent diagnoses across the three classifications, whereas 225 (50.0%) had inconsistent diagnoses. The two major WHO 2017 subtypes, AML not otherwise specified (AML-NOS) and AML with myelodysplasia-related changes (AML-MRC), were further subdivided according to the WHO 2022 and ICC. The ICC had the highest prognostication power among the three classifications. Subgroup analysis according to the different definitions of myelodysplasia-related AML and the introduction of AML with mutated TP53 (AML-TP53) showed that the differentiation of AML-TP53 was beneficial. The update from ELN 2017 to ELN 2022 resulted in significant transitions in a subset of patients. The updated diagnostic classification and ELN risk stratification (i.e., the ICC and ELN 2022) showed a straightforward relationship.
Conclusions: This study presents an integrative comparative analysis of past and current guidelines for AML diagnosis and risk classification based on open-source data. The ICC diagnostic criteria are clinically significant for determining AML prognosis. In line with the changing treatment paradigm for AML, future research is needed to continuously validate diagnostic and risk stratification systems.
Keywords: Acute myeloid leukemia, Classification, European LeukemiaNet, International Consensus Classification, World Health Organization
The classification of and diagnostic criteria for myeloid neoplasms are based on the WHO classification, which has been repeatedly updated, such as in the 3rd (2001), 4th (2008), revised 4th (2017), and 5th (2022) editions [1–4]. Clinical advisory committees have also independently updated the revised 4th edition of the WHO classification, which resulted in the release of the International Consensus Classification (ICC) of myeloid neoplasms and acute leukemias in 2022 [5]. Clinicians and pathologists face confusion in the practical application of the two independent guidelines, which can lead to different diagnoses [6]. Moreover, the European LeukemiaNet (ELN) has updated the 2010, 2017, and 2022 versions of the recommendations for the diagnosis and management of adult AML [7–9]; the diagnosis of AML in the ELN 2017 and 2022 recommendations is based on the WHO 2017 and ICC 2022 guidelines, respectively. The updated WHO classification extended the previous classification of AML with myelodysplasia-related (MR) changes (AML-MRC) by adding MR gene mutations and named it AML, MR (AML-MR). Furthermore, the ICC newly introduced “AML with mutated TP53 (AML-TP53),” which supersede AML with MR gene mutations (AML-MR-G) and AML with MR cytogenetic abnormalities (AML-MR-CG). In contrast to the WHO 2022 classification, the ICC further described the characteristics of MR aberrations, such as gene mutations and cytogenetic abnormalities.
Amidst the changes in the classification systems, the impact of the new AML diagnostic criteria and risk classification has been investigated [6, 10–12]. However, the enrollment of patients requires considerable effort and time. Leveraging existing datasets is a powerful method for analyzing the impact of new classifications. The cBioPortal (https://www.cbioportal.org/) is publicly available and provides clinical profiles and organized genomic data for various types of cancers. We used the cBioPortal to collect open-source data from patients with AML to investigate the impact of the newly introduced WHO and ICC classifications alongside the ELN 2022 risk classification.
AML datasets were searched using the cBioPortal platform. The cohort of patients with AML from a 2022 study performed at the Oregon Health & Science University (OHSU) was selected [13] because it provided WHO 2017 classifications for adult patients. The study included 942 representative AML cases. Exclusion criteria for the present study included non-initial disease status, non-AML, non-specified AML subtype, missing karyotype or mutations, discrepant mutation profiles between data from the same sample, redundant samples from the same patient, pediatric patients (<18 yrs of age), and acute promyelocytic leukemia. In total, 450 patients were included in this study. Five had no survival information and were not included in survival analysis. The patient selection procedure is summarized in Supplemental Data Fig. S1. Diagnoses were assigned according to the WHO 2017 and 2022 classifications as well as the ICC [3–5], and risk classifications were assigned according to the ELN 2017 and 2022 [8, 9]. The study was exempt from approval by the Institutional Review Board of the Korea University College of Medicine because it relied solely on existing open-source data.
Continuous variables are expressed as medians and ranges or interquartile ranges, and categorical variables as numbers and percentages. To analyze the impact of each classification on survival, the Cox proportional hazards model was used; hazard ratios (HRs) were calculated with 95% confidence intervals (CIs), and the Akaike information criterion (AIC) and concordance index (C-index) with SE were derived. The Kaplan–Meier method and log-rank test were used to estimate overall survival and evaluate differences among groups. All statistical analyses were performed using R version 4.2.3. Statistical significance was set at P<0.05.
The demographics of the 450 patients with AML included in this study are presented in Table 1. One hundred and ninety patients (42.2%) had consistent diagnoses among the WHO 2017 and WHO 2022 classifications and ICC, including AML with RUNX1::RUNX1T1 (N=9), AML with CBFB::MYH11 (N=35), AML with KMT2A::MLLT3 (N=9), AML with GATA2, MECOM (N=10), AML with BCR::ABL1 (N=1), and AML with mutated NPM1 (N=126). Thirty-five patients (7.8%) were diagnosed as having therapy-related AML according to the WHO 2017 guidelines and as having AML post cytotoxic therapy according to the WHO 2022 guidelines; their subtypes were clarified by the ICC, which added the diagnostic qualifier “therapy-related,” including AML with RUNX1::RUNX1T1 (N=1), AML with CBFB::MYH11 (N=2), AML with MLLT3::KMT2A (N=6), AML with other KMT2A rearrangement (KMT2A-r) (N=2), AML with DEK::NUP214 (N=1), AML with mutated NPM1 (N=3), AML-TP53 (N=9), AML-MR-G (N=5), AML-MR-CG (N=4), and AML not otherwise specified (AML-NOS) (N=2).
Characteristic | N (%)* |
---|---|
Age, yrs, median (range) | 62 (18–88) |
Female sex | 200 (44.4) |
Race | |
White | 256 (56.9) |
Hispanic | 24 (5.3) |
Asian | 10 (2.2) |
Black | 9 (2.0) |
Admixed white | 1 (0.2) |
Admixed black | 1 (0.2) |
Unknown | 149 (33.1) |
History >2 months before AML diagnosis | |
MDS | 37 (8.2) |
MDS/MPN | 12 (2.7) |
WHO 2017 classification | |
AML with t(8;21)(q22;q22.1); RUNX1::RUNX1T1 | 9 (2.0) |
AML with inv(16)(p13.1q22) or t(16;16)(p13.1;q22); CBFB::MYH11 | 35 (7.8) |
AML with t(9;11)(p21.3;q23.3); MLLT3::KMT2A | 9 (2.0) |
AML with other translocations involving KMT2A | 6 (1.3) |
AML with inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2); GATA2, MECOM | 10 (2.2) |
AML with BCR::ABL1 | 1 (0.2) |
AML with mutated NPM1 | 126 (28.0) |
AML with biallelic mutations of CEBPA | 10 (2.2) |
AML with mutated RUNX1 | 3 (0.7) |
AML with myelodysplasia-related changes | 110 (24.4) |
Therapy-related AML | 35 (7.8) |
AML not otherwise specified | 96 (21.3) |
2017 ELN risk stratification by genetics | |
Favorable | 143 (31.8) |
Intermediate | 97 (21.6) |
Adverse | 196 (43.6) |
Undetermined | 14 (3.1) |
Treatment | |
Intensive chemotherapy | 312 (69.3) |
HMA-based/low intensity chemotherapy | 96 (21.3) |
Other chemotherapy† | 4 (0.9) |
Allogenic HSCT | 131 (29.1) |
None/supportive | 38 (8.4) |
Response to induction treatment in all patients | |
CR/CRi | 212 (47.1) |
Refractory | 116 (25.8) |
Unknown | 122 (27.1) |
*Except age.
†Bevacizumab (N=1), entospletinib (N=1), ruxolitinib and hydroxyurea (N=1), vincristine and dexamethasone (N=1).
Abbreviations: MDS, myelodysplastic syndrome; MDS/MPN, myelodysplastic/myeloproliferative neoplasm; ELN, European LeukemiaNet; HMA, hypomethylating agent; HSCT, hematopoietic stem cell transplantation; CR, complete remission; CRi, CR with incomplete hematologic recovery.
The diagnoses of the remaining 225 patients (50.0%) differed among the three classifications (Fig. 1). Patients classified as the two main subtypes, AML-NOS and AML-MRC, according to the WHO 2017 classification were further subdivided. While a subset of AML-NOSWHO2017 changed to AML, MR (AML-MR)WHO2022 & ICC, most patients with AML-MRCWHO2017 remained AML-MRWHO2022. A fraction of AML-MRWHO2022 was reclassified as AML-TP53 according to the ICC. The heterogeneous definitions of AML with CEBPA mutations among the three classifications played a minor role in the reclassification of patients with AML. In the ICC, diagnostic qualifiers, such as “progressing from myelodysplastic syndrome (MDS)” and “progressing from myelodysplastic/myeloproliferative neoplasm (MDS/MPN),” were added for the diagnosis of AML-MR and AML-TP53.
Kaplan–Meier curves with numbers at risk and the results of Cox proportional hazards modeling based on the WHO 2017 and WHO 2022 classifications and ICC are provided in Supplemental Data Fig. S2. The proportional hazards model included the variables sex, age, prior MDS or MDS/MPN, therapy relation (only for the ICC), AML subtype, and treatment. The ICC had the highest prognostication power (AIC: 2,724; C-index: 0.768 [SE: 0.014]), followed by the WHO 2022 (AIC: 2,747; C-index: 0.755 [SE: 0.014]) and WHO 2017 (AIC: 2,756; C-index: 0.754 [SE: 0.014]) classifications.
The different definitions of AML-MR(C) among the WHO 2017 and WHO 2022 classifications and ICC are summarized in Supplemental Data Table S1. The cytogenetic abnormalities considered in defining AML-MR differed slightly between the WHO and ICC. Based on the different definitions of AML-MR among the three guidelines and the introduction of AML-TP53 in the ICC, we generated four subgroups: (1) AML-NOSWHO2017 & ICC|AML defined by differentiation (AML-DBD)WHO2022 (N=35); (2) AML-NOSWHO2017|AML-MRWHO2022 & ICC (N=45); (3) AML-MR(C)WHO2017 & WHO2022 & ICC (N=72); and (4) AML-MR(C)WHO2017 & WHO2022|AML-TP53ICC (N=34). Subgroup 1 was generated for comparison with subgroup 2. According to the Cox proportional hazards model adjusted for sex, age, prior MDS or MDS/MPN, AML subtype, and treatment, patients diagnosed as having AML-MR(C) by the WHO classification but as AML-TP53 by the ICC (subgroup 4) had the poorest prognosis, whereas the remaining three subgroups (subgroups 1, 2, and 3) exhibited comparable overall survival: subgroup 1 vs. subgroup 4: HR=0.381, 95% CI=0.213–0.683; subgroup 2 vs. subgroup 4: HR=0.424, 95% CI=0.252–0.715; and subgroup 3 vs. subgroup 4: HR= 0.348, 95% CI=0.204–0.591) (Fig. 2).
The ELN 2017 and ELN 2022 are based on the WHO 2017 and ICC, respectively [8, 9]. Fig. 3 shows a Sankey diagram describing the reclassification of ELN risk groups and their relationships with AML classifications. Among 143 patients with favorable risk according to the ELN 2017, 16 (11.2%) and three (2.1%) were reclassified as having intermediate and adverse risk, respectively. Of the 16 patients with favorableELN2017|intermeditateELN2022, 15 had NPM1 mutations and low allelic ratios of internal tandem duplication mutations in FLT3 (FLT3-ITD); one patient had biallelic CEBPA mutations that did not correspond to in-frame mutations in the basic leucine zipper (bZIP) domain. The three patients with favorableELN2017|adverseELN2022 had biallelic CEBPA mutations (not in-frame bZIP mutations) and harbored mutations in MR genes according to the ICC (ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, and ZRSR2). Among 97 patients with intermediate risk according to the ELN 2017, one (1.0%) and 16 (16.5%) were reclassified into the favorable and adverse groups, respectively. A patient with intermediateELN2017|favorableELN2022 harbored a monoallelic in-frame bZIP CEBPA mutation, and 16 patients with intermediateELN2017|adverseELN2022 harbored mutations in MR genes. Among 196 patients with adverse risk according to the ELN 2017, eight (4.1%) were reclassified as intermediate because they harbored wild-type NPM1 with FLT3-ITD. Although the risk groups of 14 patients could not be determined because of the absence of the FLT3-ITD allelic ratio in the ELN 2017, the ELN 2022 categorized them as having intermediate (N=13) and adverse (N=1) risk.
The ELN 2017 risk groups for AML-NOSWHO2017 included intermediate and adverse (Fig. 3); however, the ELN 2022 risk group for AML-NOSICC was only intermediate as MR genes were introduced in both AML-MR in the ICC and the adverse risk group in the ELN 2022 (Fig. 3). The relationship between AML subtype classification and risk stratification was more simplified in the 2022 versions of the guidelines than in the 2017 versions.
AML is a highly heterogeneous hematologic malignancy; specifying AML types is pivotal for treatment and determining prognoses. When updating the WHO 2017 classification, both the WHO 2022 and ICC guidelines emphasized genomic factors over morphology. For example, the WHO 2022 and ICC no longer include multilineage dysplasia as an inclusion criterion for AML-MR.
Given the lack of a consensus between the coexisting WHO 2022 and ICC guidelines, many clinicians and hematopathologists are confused. Therefore, we classified AML according to the WHO 2017, WHO 2022, and ICC guidelines and visualized the transition flows from WHO 2017 to the ICC using a sunburst plot to facilitate a better understanding of AML reclassification. In addition, we summarized the definitions of AML-MR according to the three guidelines to illustrate their differences.
Our findings are largely consistent with previous findings. Huber, et al. [6] and Lee, et al. [12] compared the diagnoses of patients with MDS and AML among the WHO 2017, WHO 2022, and ICC guidelines. Both the WHO 2022 classification and ICC reduced the number of AML-NOS cases, whereas the number of AML-MR cases increased under the WHO 2022 classification; additionally, AML-TP53 was newly classified in the ICC [6, 12]. Sargas, et al. [10] investigated the 2017 and 2022 ELN risk classifications in the PATHEMA cohort in Spain. In line with our findings, they found major shifts from favorable to intermediate and from intermediate to adverse; however, in contrast to our findings, the number of intermediate cases was reduced under the ELN 2022. Wang, et al. [14] compared the 2017 and 2022 ELN guidelines in 536 patients with AML. They found notable shifts from favorable to intermediate, intermediate to adverse, and unclassified to intermediate, consistent with our findings.
Lachowiez et al. [11] compared and validated the ELN 2022 guidelines in AML using the OHSU Beat AML cohort (N=513), which was the same data source as that used in the present study. They demonstrated that the prognostic power of the ELN 2022 in patients who received intensive chemotherapy was slightly superior to that of the ELN 2017 (C-index: 0.68 [SE: 0.02] vs. 0.66 [SE: 0.02]) [11]. In the present study, the prognostic power of the ELN 2022 was similar to that of the ELN 2017 in all patients (C-index: 0.764 [SE: 0.014] vs. 0.763 [SE: 0.015]).
In subgroup analysis according to the different definitions of AML-MR and the introduction of AML-TP53 in the ICC, patients diagnosed as having AML-MR(C)WHO2017 & WHO2022|AML-TP53ICC demonstrated the most unfavorable prognosis. This indicates that the differentiation of AML-TP53ICC is helpful in accurate prognosis prediction, and MR and TP53 gene sequencing is highly recommended as a routine screening in patients with AML. As the relationship between the AML subtype classification (ICC) and risk stratification (ELN) has been straightened in the 2022 guidelines, it is somewhat possible to estimate the ELN 2022 risk group based solely on the ICC diagnosis. Collectively, our findings indicate that the ICC guideline is more beneficial than the WHO 2022 guideline for predicting the outcomes of patients with AML.
Compared with previous studies, this study had three advantages. We used a validated open-source dataset published in an article, ensuring its reliability. Additionally, we presented a sunburst plot and a summary of AML-MR definitions among the three guidelines, facilitating an understanding of their complexities. Finally, by conducting a subgroup analysis using the Cox proportional hazards model and examining the relationship between ELN risk stratification and diagnostic classification, the ICC was found to provide greater value in determining AML prognosis.
This study also had some limitations. We only used open-source datasets that did not require data access requests, resulting in a smaller study population. Other genomic data repositories that provide data access to authorized researchers, such as dbGAP and the European Genome-Phenome Archive, could be used in future studies to enroll larger study populations. Furthermore, treatment approaches for adult patients with AML have evolved: the combination of venetoclax and hypomethylating agents has become the standard treatment for patients ineligible for intensive chemotherapy [15]. As the fraction of patients receiving venetoclax and hypomethylating agents was low in the present study, future studies should include patients who have received these novel treatments.
In conclusion, we conducted an integrative comparative analysis of the previous and current guidelines for the diagnosis and risk classification of AML using open-source data. The ICC diagnostic criteria are the most clinically useful for AML classification and prognosis prediction. As treatments for AML evolve, validating the diagnostic and risk stratification systems with data from more various sources based on recent patient data will be needed.
None.
Yun J designed the study, performed the research, analyzed the data, and wrote the paper.
None declared.
This study was supported by the research fund of the Korean Association of External Quality Assessment Service (Grant No. KEQAS-2024-12).
Supplementary materials can be found via https://doi.org/10.3343/alm.2024.0194