The WHO Classification of the diagnostic criteria for hematologic malignancies was revised in 2022 [1]. Meanwhile, clinical advisory committees developed the International Consensus Classification (ICC) of myeloid neoplasms (MNs) and acute leukemias [2]. The coexistence of two new diagnostic classification systems has led to confusion among many clinicians [3]. Numerous researchers have investigated the reclassification of diagnostic entities according to the new criteria by recruiting and examining patient cohorts [4^-8]. However, these studies often had a limited sample size.

The cBioPortal for Cancer Genomics is an open-source resource developed at the Memorial Sloan Kettering Cancer Center (New York, NY, USA) and hosted on GitHub (https://www.cbioportal.org/). It contains large-scale cancer genomics data and clinical profiles of various cancer types. The 2022 WHO and 2022 ICC diagnostic criteria for myelodysplastic neoplasms (MDSs, referred to as myelodysplastic syndromes in the previous WHO classification and current ICC) are similar in some aspects but differ in others. Using cBioPortal open-source data, we reclassified patients diagnosed with MDSs using the 2017 WHO criteria based on the 2022 WHO criteria and examined the differences in diagnoses. In addition, we compared the 2022 WHO and 2022 ICC diagnostic criteria in classifying two genetics-based MDS subtypes, SF3B1- and TP53-mutated MDS, which are newly introduced MDS subtypes with different definitions between the 2022 WHO and ICC criteria.

Study population

We retrieved a dataset from a study by Bernard, et al. [9], which includes 2,957 representative MDS samples, because this dataset includes information on copy number variations of the TP53 locus, which is essential for the diagnosis of MDS with biallelic TP53 inactivation (MDS-biTP53) according to the 2022 WHO criteria, from cBioPortal. Exclusion criteria included myelodysplastic/myeloproliferative neoplasms, unspecified diagnosis, and discrepancy between the diagnosis and copy number variation. The dataset does not describe dysplastic lineage. Thirty-nine patients with MDS, unclassifiable (MDS-U), who harbored an MDS-defining abnormality were omitted because, without information on dysplasia, these patients may have been classified as MDS-U based on defining cytogenetic abnormality. MDS-U, based on defining cytogenetic abnormality, is reclassified as clonal cytopenia of undetermined significance (CCUS) in the 2022 WHO criteria. Finally, 2,454 patients with MDS were included in the study. The flow of patient selection is summarized in Supplemental Data Fig. S1.

The study was exempt from approval by the Institutional Review Board of Chung-Ang University College of Medicine (Seoul, Korea) because it relied solely on open-source data. Patients were diagnosed according to the 2017 WHO [10] and 2022 WHO [11] classifications. The ICC criteria [2] were applied in diagnosing SF3B1- and TP53-mutated MDSs. Table 1 summarizes the commonalities and differences between the two criteria.

Table 1 Comparison of the 2022 WHO and 2022 ICC diagnostic criteria in classifying MDS subtypes with SF3B1 or TP53 mutations

MDS subtype	Commonalities	Differences
		2022 WHO	ICC
MDS-SF3B1WHO/ICC	≥1 dysplastic lineage and cytopenia each Blasts: <5% BM, <2% PB Cytogenetics: absence of del(5q), -7/del(7q), or CK Mutations: absence of biallelic TP53	Erythroid lineage dysplasia is required Mutations: ≥5% VAF of SF3B1	Cytogenetics: absence of abn3q26.2 Mutations: ≥10% VAF of SF3B1, absence of RUNX1
MDS-LB-RS with wild-type SF3B1WHO		Satisfied for MDS-SF3B1WHO except for SF3B1mut ≥15% RS	Not defined and included in MDS, NOS
MDS-biTP53WHO/MDS with mutated TP53ICC		≥1 dysplastic lineages and cytopenias Blasts: <20% BM and PB Genetics: ≥2 TP53 mutations, or one mutation with evidence of TP53 copy number loss or cnLOH	Not stated for dysplastic lineage/cytopenia Blasts: 0%–9% BM and PB Genetics: >10% VAF of TP53mut (prerequisite), 2 distinct TP53mut or 1 TP53mut with (1) del(17p) on cytogenetics (2) VAF>50% (3) cnLOH at TP53 locus or (N/A for TP53 locus LOH status) CK often with del(17p)
MDS/AML with mutated TP53ICC		Not defined Cases with 10–19% blasts and biallelic TP53 inactivation are classified as MDS-biTP53	Blasts: 10%–19% BM and PB Any somatic TP53 mutation (VAF>10%)

Abbreviations: ICC, International Consensus Classification; BM, bone marrow; PB, peripheral blood; CK, complex karyotype; del, deletion; VAF, variant allele frequency; abn, abnormality; LB, low blasts; RS, ring sideroblasts; NOS, not otherwise specified; cn, copy-neutral; LOH, loss of heterozygosity; mut, mutation; N/A, not available.

Statistical analyses

Continuous variables are expressed as medians and interquartile ranges, and categorical variables as numbers and percentages. Validation of the 2017 and 2022 WHO classifications and assessment of risk factors were conducted using Cox proportional-hazards models; hazard ratios (HRs) were calculated with 95% confidence intervals (CIs), and the Akaike information criterion (AIC) and concordance index (C-index) with standard errors (SEs) were derived. The Kaplan–Meier method and log-rank test were used to estimate overall survival and to evaluate differences among groups. All statistical analyses were performed using R software, version 4.2.3. Statistical significance was set at P<0.05.

Reclassification of MDSs according to the 2022 WHO classification

The demographics of the 2,454 patients with MDSs are presented in Table 2. The median values for haemoglobin concentration, platelet count, and absolute neutrophil count were 9.6 g/dL, 127×10⁹/L, and 1.8×10⁹/L, respectively. The reclassification of the patients based on the 2022 WHO classification is summarized in Table 3. In total, 35 patients previously classified as having MDSs were reclassified as having AML because they harbored AML-defining genetic abnormalities irrespective of blast count, as follows: DEK::NUP214 (N=1), KMT2A rearrangement (N=3), MECOM rearrangement (N=5), and NPM1 mutation (N=26). As the information on erythroid lineage dysplasia was unavailable, cases in which SF3B1 was mutated and ring sideroblast counts were <5% or not assessed were considered “presumed MDS with low blasts and SF3B1 mutation” (“MDS-SF3B1²⁰²²”).

Table 2 Demographics of the study population

Characteristics	N (%)
Age (yrs)*	72 (63, 78)
Female sex	975 (39.7)
Ontogeny
Primary	2185 (89.0)
Secondary/therapy-related	207 (8.4)
NA	62 (2.5)
WHO 2017 classification
MDS-del(5q)	139 (5.7)
MDS-SLD	191 (7.8)
MDS-MLD	639 (26.0)
MDS-SLD/MLD†	91 (3.7)
MDS-RS-SLD	246 (10.0)
MDS-RS-MLD	212 (8.6)
MDS-RS-SLD/MLD†	3 (0.1)
MDS-EB-1	458 (18.7)
MDS-EB-2	429 (17.5)
MDS-U	46 (1.9)
IPSS-R
Very low	383 (15.6)
Low	917 (37.4)
Intermediate	482 (19.6)
High	312 (12.7)
Very high	253 (10.3)
NA	107 (4.4)
IPSS-M
Very low	302 (12.3)
Low	746 (30.4)
Moderately low	258 (10.5)
Moderately high	244 (9.9)
High	325 (13.2)
Very high	430 (17.5)
NA	149 (6.1)
Disease-modifying treatment
None	1662 (67.7)
Lenalidomide alone	140 (5.7)
HMAs‡	377 (15.4)
Intensive chemotherapy§	28 (1.1)
Transplantation||	247 (10.1)

*Ages are presented as median with interquartile range; data for one patient was missing.

^†“MDS-SLD/MLD” and “MDS-RS-SLD/MLD” indicate that the number of dysplastic lineages was not specified, based on the pre-existing diagnosis assigned by Bernard, et al. [9].

^‡HMAs (plus lenalidomide).

^§Intensive chemotherapy (plus HMAs).

^||Hematopoietic stem cell transplantation (plus lenalidomide, HMA, or intensive chemotherapy).

Abbreviations: NA, not assessed; -del(5q), with isolated del(5q); -SLD, with single lineage dysplasia; -MLD, with multilineage dysplasia; -RS, with ring sideroblasts; -EB, with excess blasts; MDS-U, MDS, unclassifiable; IPSS-R, Revised International Prognostic Scoring System; IPSS-M, Molecular International Prognostic Scoring System; HMAs, hypomethylating agents.

Table 3 Reclassification of patients with MDS using the 2022 WHO diagnostic criteria

MDS subtype	2022 WHO classification
	MDS-5q	MDS-biTP53	MDS-SF3B1	Presumed MDS-SF3B1	MDS-LB	MDS-LB-RS	MDS-IB1	MDS-IB2	AML	N (%)
2017 WHO classification
MDS-del(5q)	134	5	0	0	0	0	0	0	0	139 (5.7)
MDS-SLD	0	4	0	22	165	0	0	0	0	191 (7.8)
MDS-MLD	0	32	0	39	561	0	0	0	7	639 (26.0)
MDS-SLD/MLD*	0	3	0	4	83	0	0	0	1	91 (3.7)
MDS-RS-SLD	0	1	214	0	3	28	0	0	0	246 (10.0)
MDS-RS-MLD	0	14	135	0	14	48	0	0	1	212 (8.6)
MDS-RS-SLD/MLD*	0	1	1	0	0	1	0	0	0	3 (0.1)
MDS-EB-1	0	59	0	0	0	0	392	0	7	458 (18.7)
MDS-EB-2	0	79	0	0	0	0	0	331	19	429 (17.5)
MDS-U	0	1	2	7	36	0	0	0	0	46 (1.9)
N (%)	134 (5.5)	199 (8.2)	352 (14.6)	72 (3.0)	862 (35.7)	77 (3.2)	392 (16.2)	331 (13.7)	35	2,454

*“MDS-SLD/MLD” and “MDS-RS-SLD/MLD” indicate that the number of dysplastic lineages was not specified, based on the pre-existing diagnosis assigned by Bernard, et al. [9]

Abbreviations: -del(5q), with isolated del(5q); -SLD, with single lineage dysplasia; -MLD, with multilineage dysplasia; -RS, with ring sideroblasts; -EB, with excess blasts; MDS-U, MDS, unclassifiable; -5q, with low blasts and isolated 5q deletion, -biTP53, with biallelic TP53 inactivation; -SF3B1, with low blasts and SF3B1 mutation;-LB, with low blasts; -IB, with increased blasts.

Validation of the 2017 and 2022 WHO classifications in patients with MDSs

The results of survival analyses according to the 2017 and 2022 WHO classifications are presented in Supplemental Data Fig. S2A and S2B. Cox proportional-hazards models adjusted for sex, age, ontogeny, subtypes, and treatment revealed that the 2022 WHO criteria stratified patients with MDSs more effectively than the 2017 WHO criteria: AIC, 14,516; C-index, 0.681 (SE, 0.009) for 2017 WHO vs. AIC, 14,152; C-index, 0.705 (SE 0.009) for 2022 WHO (Supplemental Data Fig. S2C and S2D). Patients whose diagnosis was changed from MDS to AML according to the new criteria had a median overall survival of 1.4 yrs, which was the shortest overall survival (excluding patients with MDS-biTP53 or MDS-IB2). This finding suggests that the definition of AML in the 2022 WHO criteria is well established.

Subgroup analyses focusing on SF3B1-mutated MDS

To evaluate factors affecting survival in SF3B1-mutated MDS, a Cox proportional-hazards model adjusted for sex, age, ontogeny, treatment, type of SF3B1 variants, variant allele frequency (VAF) of SF3B1 variants, and RUNX1 co-mutation was used. The distribution of SF3B1 variants in patients with MDS-SF3B1^WHO is plotted in Supplemental Data Fig. S3. Most variants were missense variants, and the SF3B1 K700E variant was the most frequently observed. Compared with K700E alone, the other variants, including non-K700E variants and K700E plus non-K700E variants, did not affect survival (Fig. 1A; HR, 0.85; 95% CI, 0.59–1.21). To evaluate the ICC criteria, we performed subgroup analyses of MDS-SF3B1^WHO. SF3B1 variants with a VAF<5% are excluded in the diagnosis of MDS-SF3B1 by the WHO. Compared with VAF≥10%, 5%≤VAF<10% did not affect survival (Fig. 1B; HR, 0.97; 95% CI, 0.30–3.12). In contrast, RUNX1 co-mutation (including multiple mutations) was associated with a worse prognosis than wild-type RUNX1 (Fig. 1C; HR, 3.63; 95% CI, 1.83–7.19).

Figure 1. Subgroup analyses in MDS-SF3B1^WHO. Kaplan–Meier curves of overall survival were plotted according to (A) type of SF3B1 variants (K700E alone vs. others), (B) VAF of SF3B1 variants (5%≤VAF<10% vs. VAF≥10%), and (C) RUNX1 co-mutation (wild-type vs. mutated). HRs and 95% CIs were calculated using Cox proportional-hazards models adjusted for sex, age, ontogeny, treatment, type of SF3B1 variants, VAF of SF3B1 variants, and RUNX1 co-mutation. The HR is shown with the 95% CI in parentheses.
Abbreviations: VAF, variant allele frequency; HR, hazard ratio; CI, confidence interval.

Subgroup analyses focusing on TP53-mutated MDS

We classified TP53-mutated MDS samples according to the 2022 WHO and ICC criteria. The relationship between the two criteria is shown in Fig. 2. When the ICC criteria were applied, 199 patients with MDS-biTP53^WHO were classified as MDS with mutated TP53^ICC (N=103), MDS/AML with mutated TP53^ICC (N=76), and others (N=20). The last group of 20 patients, referred to as “others,” harbored TP53 mutations with a VAF≤10%, and the majority (17/20, 85%) had <10% blasts. Thus, 103 patients with MDS and mutated TP53^ICC and 76 patients with MDS/AML and mutated TP53^ICC harbored multi-hit TP53 mutations with a VAF>10%. Among 331 patients with MDS-IB2^WHO, 10 patients (3.0%) were classified as having MDS/AML with mutated TP53^ICC because they harbored single TP53 mutations with VAF>10%.

Figure 2. Relationship between the 2022 WHO and ICC classifications focused on MDSs with mutated TP53.
Abbreviations: ICC, International Consensus Classification; VAF, variant allele frequency; -biTP53, with biallelic TP53 inactivation; -IB, with increased blasts; NOS, not otherwise specified; -EB, with excess blasts.

A Cox proportional-hazards model adjusted for sex, age, ontogeny, diagnosis, and treatment was used to evaluate the prognostic impact in patients with MDS with mutated TP53^ICC and patients with MDS/AML with mutated TP53^ICC. Notably, patients with MDS/AML and mutated TP53^ICC (≥10% blasts) had a poorer prognosis than patients with MDS and mutated TP53^ICC (<10% blasts) (Fig. 3A, HR, 1.47; 95% CI, 1.04–2.08). Subgroup analysis of patients with MDS-biTP53^WHO was performed using a Cox proportional-hazards model adjusted for sex, age, ontogeny, treatment, type of biallelic TP53 inactivation, and VAF of TP53 mutation. MDS-biTP53^WHO can be diagnosed in cases harboring two or more mutations in TP53 or one mutation with TP53 locus loss or copy-neutral loss of heterozygosity (cnLOH). Patients with these two mutation types did not show a difference in survival: one TP53 mutation with TP53 locus loss or cnLOH vs. two or more mutations in TP53; HR, 1.30; 95% CI, 0.91–1.86 (Fig. 3B). Among patients with MDS-biTP53^WHO, those with a VAF≤10% did not qualify for MNs with mutated TP53^ICC. Patients with a VAF>10% did not differ in survival compared with patients with VAF≤10% (Fig. 3C; HR, 1.55; 95% CI, 0.84–2.88). Patients with ≥10% blasts were categorized into three subgroups based on the combined 2022 WHO and ICC diagnoses: MDS-IB2^WHO| MDS/AML^ICC (N=321), MDS-IB2^WHO|MDS/AML with mutated TP53^ICC (N=10), and MDS-biTP53^WHO|MDS/AML with mutated TP53^ICC (N=76). MDS-IB2^WHO|MDS/AML with mutated TP53^ICC refers to cases with a single TP53 mutation, which were relatively rare (2.5%) among patients with MDS with ≥10% blasts. Patients diagnosed as having MDS-biTP53^WHO|MDS/AML with mutated TP53^ICC (TP53 multi-hit) had a poorer prognosis than patients with MDS-IB2^WHO|MDS/AML^ICC (TP53 wild-type, HR, 3.92; 95% CI, 2.85–5.39, Fig. 3D). Because of the small number of cases, a comparison with MDS-IB2^WHO|MDS/AML mutated TP53^ICC was not conducted.

Figure 3. Subgroup analyses of MDSs with mutated TP53. (A) Kaplan–Meier curves of overall survival were plotted for MDS with mutated TP53^ICC and MDS/AML with mutated TP53^ICC. The HR and 95% CI were calculated using a multivariate Cox regression model adjusted for sex, age, ontogeny, diagnosis, and treatment. Within MDS-biTP53^WHO, Kaplan–Meier curves of overall survival were plotted according to (B) the type of biallelic TP53 inactivation and (C) VAF of TP53 mutations (whether the ICC diagnoses are MNs with mutated TP53 or not). The HR and 95% CI were calculated using a multivariate Cox regression model adjusted for sex, age, ontogeny, treatment, type of biallelic TP53 inactivation, and VAF of TP53 mutations. (D) Within MDS cases with ≥10% blasts, Kaplan–Meier curves of overall survival were plotted in MDS-IB2^WHO|MDS/AML^ICC, MDS-IB2^WHO|MDS/AML with mutated TP53^ICC, and MDS-biTP53^WHO|MDS/AML with mutated TP53^ICC. The HR and 95% CI were calculated using a multivariate Cox regression model adjusted for sex, age, ontogeny, treatment, and diagnosis. The HR is shown with the 95% CI in parentheses.
Abbreviations: ICC, International Consensus Classification; cnLOH, copy-neutral loss of heterozygosity; TP53_mut, TP53 mutations; VAF, variant allele frequency; MNs, myeloid neoplasms;. -IB2, with increased blasts-2, -biTP53, with biallelic TP53 inactivation; HR, hazard ratio; CI, confidence interval.

Using a large open-source dataset, we reclassified MDS patients diagnosed based on the 2017 WHO criteria using the 2022 WHO criteria, and we focused on MDSs with mutated SF3B1 and mutated TP53 to compare the 2022 WHO and ICC criteria. MDS²⁰¹⁷ changed to AML²⁰²² or CCUS²⁰²² in a subset of patients with MDS. MDS-U²⁰¹⁷ is removed and allocated to specific MDS subtypes in the 2022 WHO classification. Zhang, et al. [4] compared the 2017 and 2022 WHO criteria in a cohort of 856 patients with MDSs and reclassified 30 patients (3.5%) previously diagnosed as having MDSs to having AML because they harbored NPM1 mutations. In addition, among 21 patients with MDS-U²⁰¹⁷, nine patients (42.9%) were reclassified as having CCUS [4]. We validated the prognostic performance of the 2022 WHO criteria, which are superior to the previous criteria, using Cox proportional-hazards models adjusted for clinical variables.

The VAF cut-off for SF3B1 mutation for the diagnosis of MDS-SF3B1 is higher in the 2022 ICC criteria (10%) than in the 2022 WHO criteria (5%), and RUNX1 co-mutation should be absent according to the ICC criteria (Table 1). Our study demonstrated that the VAF cut-off of 5% (2022 WHO criteria) and the absence of RUNX1 co-mutation (2022 ICC criteria) are clinically relevant. However, the SF3B1 variant type does not influence prognosis.

The 2022 WHO classification specifies MDS-biTP53 when the blast count is <20%, and AML with mutated TP53 is not specified as a disease entity (Table 1). In contrast, the ICC defines a category termed “MNs with mutated TP53,” which includes MDS (<10% blasts), MDS/AML (10%–19% blasts), and AML (≥20% blasts). In the ICC criteria, the TP53 mutation status should be biallelic in MDS with mutated TP53, but it can be monoallelic or biallelic in MDS/AML with mutated TP53 and AML with mutated TP53 (Table 1). While the WHO does not specify a threshold for the VAF, the ICC mandates that the VAF for TP53 mutations should be >10% (Table 1). In our study, when classifying TP53-mutated MDSs, using a blast cut-off of 10% (2022 ICC criteria) to distinguish MDS and MDS/AML was prognostically valuable. In MDS cases with blast counts ≥10%, TP53 multi-hit (2022 WHO criteria) was an independent risk factor as compared with wild-type TP53. However, the TP53 variant type within TP53 multi-hit and a TP53 VAF cut-off of 10% within MNs with mutated TP53^ICC were not prognostic indicators.

Our study had some limitations. First, the lack of information on bone marrow cellularity, fibrosis, and dysplastic lineage posed a challenge in the assessment of MDS, hypoplastic²⁰²², MDS with increased blasts and fibrosis²⁰²², and MDS-SF3B1²⁰²², as well as the reclassification of MDS-U²⁰¹⁷. This potentially introduced a selection bias and impacts the generalizability of our findings. Second, because this study relied on a pre-existing dataset, inherent biases were present. For example, the study population was biased toward European ethnicity, limiting the generalization of our results to other ethnicities. In the future, we aim to expand our analyses as more open-source data become available for other ethnicities. Finally, the observational and retrospective nature of this study limited the ability to establish causal relationships between variables.

In conclusion, our findings support the refinements of the 2022 WHO classification of MDS. We comprehensively discussed the newly introduced SF3B1- and TP53-mutated MDSs according to the 2022 WHO and ICC. We advise clinicians to use both the 2022 WHO classification and ICC to appropriately diagnose patients with SF3B1- and TP53-mutated MDSs. Our study used well-validated open-source data and involved a significant number of patients, thereby ensuring both reliability and representativeness.

None.

Yun J and Kim HR designed the study; Yun J performed the research, analyzed the data, and wrote the original draft; Kim HR reviewed and revised the manuscript. The authors have accepted responsibility for the entire content of this manuscript and approved its submission.

None declared.

This study was supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by the Ministry of Education (NRF-2021R1A2C2013359).

Supplementary materials can be found via https://doi.org/10.3343/alm.2024.0079