Comparison of the revised 4th (2016) and 5th (2022) editions of the World Health Organization classification of myelodysplastic neoplasms

We used data from 852 consecutive subjects with myelodysplastic neoplasms (MDS) diagnosed according to the 2016 (revised 4th) World Health Organization (WHO) criteria to evaluate the 2022 (5th) edition WHO classification of MDS. 30 subjects previously classified as MDS with an NPM1 mutation were re-classified as acute myeloid leukaemia (AML). 9 subjects previously classified as MDS-U were re-classified to clonal cytopenia of undetermined significance (CCUS). The remaining 813 subjects were diagnosed as: MDS-5q (N = 11 [1%]), MDS-SF3B1 (N = 70 [9%]), MDS-biTP53 (N = 53 [7%]), MDS-LB (N = 293 [36%]), MDS-h (N = 80 [10%]), MDS-IB1 (N = 161 [20%]), MDS-IB2 (N = 103 [13%]) and MDS-f (N = 42 [5%]) and MDS-biTP53 (N = 53 [7%]). 34 of these subjects came from the 53 (64%) MDS-biTP53 previously diagnosed as MDS-EB. Median survival of subjects classified as MDS using the WHO 2022 criteria was 45 months (95% Confidence Interval [CI], 34, 56 months). Subjects re-classified as MDS-biTP53 and MDS-f had significantly briefer median survivals compared with other MDS sub-types (10 months, [8, 12 months] and 15 months [8, 23 months]). In conclusion, our analyses support the refinements made in the WHO 2022 proposal.


INTRODUCTION
A summary of the World Health Organization (WHO) 5th edition (2022) classification of myelodysplastic neoplasms (MDS) was recently published in LEUKEMIA [1]. The WHO 2022 classification reorganize MDS categories by emphasizing histological and genetic co-variates. Diagnostic criteria for MDS with low blasts and isolated del(5q) (MDS-5q) was unchanged. MDS with biallelic TP53 inactivation (MDS-biTP53) is introduced as a new sub-type defined by the presence of multi-hit TP53 mutations and supersedes other MDS sub-types. Presence of a SF3B1 mutation and low blasts is considered consistent with a MDS diagnosis (MDS-SF3B1) and supersedes the prior entity of MDS with ring sideroblasts (MDS-RS). In the sub-types defined by histology the WHO 2022 classification retains cutoffs between MDS with low blasts (MDS-LB) and MDS with increased blasts (MDS-IB). Persons without increased blasts are divided into hypoplastic MDS (MDS-h) and MDS-LB. Persons with increase blasts are divided into MDS-IB1, MDS-IB2 and MDS with fibrosis (MDS-f). MDS-h and MDS-f are considered distinct subtypes underscoring the importance of a trephine bone marrow biopsy. We used a dataset of 852 consecutive subjects with MDS initially diagnosed using the WHO 2016 criteria to compare how these subjects would be classified using the WHO 2022. Our analyses support the refinements made in the WHO 2022 proposal.

SUBJECTS AND METHODS Subjects
852 consecutive subjects ≥18 years with newly-diagnosed MDS according to the 2016 (revised 4th) WHO criteria in our centre from August 30, 2016 to September 22, 2021 were enrolled [2]. Bone marrow aspirate and biopsy samples were obtained from all subjects. Diagnostic procedures were according to recent recommendation [3]. Subjects were re-classified according to the WHO 2022 classification. The prognostic impact was evaluated with the International Prognostic Scoring Systems-Revised (IPSS-R) and International Prognostic Scoring Systems-Molecular (IPSS-M) [4,5]. Baseline co-variates at diagnosis are displayed at Table 1

Bone marrow evaluation
Wright-Giemsa staining was done on bone marrow and blood slides for histological assessment with ≥ , 500 and 200 nucleated cells enumerated. Prussian blue stain was done on bone marrow slides to identify and enumerate ring sideroblasts. Erythroblasts Periodic acid-Schiff (PAS), neutrophil alkaline phosphatase (N-ALP) and CD41 immune staining were done to identify dysplastic lineages as described [6,7]. Bone marrow biopsies were done on all subjects. Routine bone marrow biopsy section thickness was 3 µm. Hematoxylin-eosin (H&E), PAS and Gomori sliver stains were done routinely. Age-adjusted bone marrow cellularity and degree of bone marrow fibrosis were determined using European consensus guidelines [8].

Multi-parameter flow cytometry
Multi-parameter flow cytometry (MPFC) was done within 24 h after collection of EDTA-anti-coagulated bone marrow aspirates. A combination of flow antibody panel (Table S1) was designed to assess MDS associated phenotypic abnormalities according to The International and European Leukemia Net Working Group Guidelines [9].

Cytogenetic analyses
Chromosome analyses were done on unstimulated bone marrow cells after 24 h of culture using G-and/or R-banding techniques. Chromosome identification and cytogenetic descriptors were applied following the International System for Human Cytogenetic Nomenclature [10]. In subjects with <12 metaphases we used fluorescence in situ hybridization (FISH) analyses including probes for -5/-5q, -7/-7q, +8, -20q, 17p-and -Y.
Targeted gene sequencing DNA from diagnosis bone marrow mononuclear cells was used for nextgeneration sequencing (NGS) as described [11]. We sequenced DNA from 592 subjects using a 141-genes panel from August 2016 to March 2020 (Table S2). DNA from 260 subjects was sequenced with a 267-genes panel from April 2020 to September 2021 (Table S3). PRPF8 and GNB1, defined as residual genes in IPSS-M, were not included in the 141-gene panel. TP53 allele state was determined as described [12].

Statistics analyses
Continuous co-variates were described by median and IQR and categorical co-variates were summarized with count and relative frequency. Continuous co-variates (non-normal distribution) were compared using the Mann-Whitney U tests. Categorical co-variates were compared using the Fisher exact test or the Pearson chi-square test. Survival was calculated as the interval from diagnosis to last follow-up or death and analyzed by the Kaplan-Meier method. The log-rank test was used for uni-variable comparisons. All P-values were 2-tailed. Statistical significance was set at P < 0.05. Analyses were conducted using GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA) and SPSS software (IBM, Chicago, IL, USA).

Re-stratification from the WHO 2016 to WHO 2022 classifications
Diagnostic criteria were largely unchanged in the new proposal except for these: (1) persons with KMT2A, MECOM, NUP98 rearrangements and NPM1 mutation are classified as acute myeloid leukaemia (AML) regardless of percentage blasts; and (2) diagnosis of MDS unclassifiable (MDS-U) was eliminated and partly replaced by clonal cytopenia of undetermined significance (CCUS). Applying the WHO 2022 classification 30 subjects with an NPM1 mutation were re-classified as AML, previously classified as MDS with excess blast type2 (MDS-EB2; n = 13), MDS with multilineage dysplasia (MDS-MLD; n = 9), MDS with excess blast type1 (MDS-EB1; n = 6) and MDS-U (n = 2) according to the WHO 2016 criteria. Nine subjects previously classified as MDS-U were reclassified to CCUS.

DISCUSSION
We used data from 852 consecutive subjects with myelodysplastic neoplasms (MDS) diagnosed according to the 2016 (revised 4th) World Health Organization (WHO) criteria to evaluate the 2022 (5th) edition WHO classification of MDS. Overall, our analyses support the refinements made in the WHO 2022 proposal. Below we discuss some differences between the classifications.
NPM1 mutations, common in AML, also occur in persons with MDS, are associated with Auer rods and can rapidly progress to AML [13,16]. Those people were biologically more akin to AML regardless of blast percentage and may benefit from AML therapies [17,18]. Consistent with these findings the WHO 2022 re-classifies these subjects to AML.
The category of MDS-U involves a small subset of subjects who cannot be accurately into any other MDS sub-type. Three categories in the WHO 2016 classification of this sub-type include: (1) 1% blood blasts; (2) pancytopenia and single lineage dysplasia; and (3) absence of significant dysplasia but with MDS-defining cytogenetic abnormalities [2]. People with the 1st 2 categories now re-identified as MDS-LB in the WHO 2022 classification. However, those with 1% blood blasts have a poor prognosis resembling to MDS-EB and should be closely followed [19,20]. People in the 3 rd category do not fulfill current diagnostic criteria for MDS and are considered CCUS in the WHO 2022 classification. Consequently, MDS-U is removed from the WHO 2022 classification.
SF3B1 mutations are common in people with MDS-RS [21]. The ring sideroblasts thought to be caused by impaired iron homeostasis [22,23]. In the WHO 2016 classification persons with SF3B1 mutation and as few as 5% ring sideroblasts without excess blasts are identified as MDS-RS [2]. Recent studies report SF3B1 mutation identifies a homogeneous subgroup regardless of bone marrow sideroblasts or dysplasia lineages [24,25]. For this reason the WHO 2022 classification substitutes MDS-SF3B1 for MDS-RS and incorporates single and multi-lineage dysplasia. This change may qualify more people to receive luspatercept [26].
Subjects with MDS-biTP53 and MDS-f had the briefest survivals in our dataset consistent to prior studies [12,[27][28][29]. Multi-hit TP53 mutations in MDS identify a very-high-risk sub-type independent of IPSS-R and co-mutation patterns. Those persons typically have complex cytogenetics, fewer co-mutations, rapid disease progression and therapy resistance [12]. Moderate to severe bone marrow fibrosis in MDS is an independent adverse risk co-variate for more severe thrombopenia, faster progression to AML and bone marrow failure and is associated with poor survival [27][28][29]. We found a high frequency of U2AF1 mutations in subjects with MDSf. We previously reported U2AF1 mutations were associated with grades-2/-3 bone marrow fibrosis [11]. Our data suggest MDS-biTP53 and MDS-f should be recognized as distinct sub-types as in the WHO 2022 classification.
Persons with MDS-h have more severe cytopenias but a better prognosis compared with those with normal/hyper-cellular MDS [30]. These findings were reproduced in our dataset. Persons with hypoplastic MDS were reported to have specific immunological and genomic features, suggesting a unique pathogenesis of this subset [31][32][33]. In accordance with these findings, the WHO 2022 classification recognize MDS-h as a distinct MDS sub-type. The distinction of MDS-h and other hypoplastic bone marrow failure disorders can be difficult. Careful morphological evaluation is critical [30,34].
Subjects with single lineage dysplasia are a small but heterogeneous cohort with a high prevalence of bi-cytopenias or pancytopenia. Isolated cytopenia is not associated with the same dysplastic lineage [35,36]. Although persons with single lineage dysplasia were reported to have a better prognosis, replicability in identifying single versus multi-lineages is low [14,15,35,36]. The threshold of 10% may explain these discordances and a threshold of 40% dysplastic megakaryocyte has been proposed [37]. The WHO 2022 classification integrates MDS-SLD and MDS-MLD in the WHO 2016 classification into MDS-LB in the WHO 2022 classification. The new sub-type emphasizes low blasts and provides a better description of those persons. Regardless, we suggest persons with MDS-LB-MLD may have a worse prognosis compared with people with MDS-LB-SLD.
Our study has important limitation, such as our data should be confirmed by an independent cohort.
In conclusion, our evaluation supports the refinements made in the WHO 2022 classification of MDS.

DATA AVAILABILITY
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.