SF3B1 mutated MDS: Blast count, genetic co-abnormalities and their impact on classification and prognosis

Recently, MDS with mutated SF3B1 and blast count <5% was proposed as distinct entity with favorable prognosis by the international working group for the prognosis of MDS (IWG-PM), the 5th edition of the WHO classification and the International Consensus Classification. To further characterize this entity with respect to the genomic landscape, AML transformation rate and clinical outcome, we analyzed 734 MDS patients by whole genome sequencing. SF3B1 mutations were identified in 31% (n = 231), most frequently accompanied by TET2 mutations (29%). 144/231 (62%) SF3B1mut samples fulfilled entity criteria proposed by IWG-PM (SF3B1ent). These cases were associated with longer survival, lower AML transformation rate, normal karyotypes and harbored less accompanying mutations compared to SF3B1mut samples not falling into the proposed SF3B1 entity (SF3B1nent). Of SF3B1mut cases 7% (15/231; SF3B1ent: 3/144 [2%]; SF3B1nent: 12/87 [14%]) progressed to AML compared to 15% SF3B1 wild-type patients (75/503). Of these 15 SF3B1mut cases, 10 (67%) showed RUNX1 mutations at MDS or AML stage. Multivariate analysis revealed that del(5q) and RUNX1 mutations were independent negative prognostic factors for overall survival, while blast count >5% was not. In conclusion, SF3B1mut MDS has a favorable prognosis independent of blast count if karyotype and RUNX1 mutations are considered.


INTRODUCTION
Myelodysplastic neoplasms (MDS) are clonal disorders characterized by peripheral cytopenias, morphologic dysplasia in hematopoietic cells and ineffective hematopoiesis [1,2]. The currently used revised 4th edition of the WHO classification (WHO 2017) in MDS is mainly based on the number of cytopenias, dysplastic lineages, and the percentage of ring sideroblasts (RS) and blasts detected in bone marrow and peripheral blood samples [3]. Within the last years, the use of next generation sequencing (NGS) enabled the identification of driver genes in MDS providing insights into the underlying heterogeneous genetic landscape [4][5][6]. In this line, about half of MDS patients harbor somatic mutations in splicing pathway genes. Of these, SF3B1 is the most commonly mutated gene and if mutated shown to be associated with RS, higher white blood cell counts and lower bone marrow blasts [6][7][8][9]. Moreover, SF3B1 mutations define a distinct MDS subset showing favorable prognosis and indolent disease course [10]. Thus, in the WHO 2017 the SF3B1 mutation is integrated into the diagnosis of MDS-RS (diagnostic criteria: RS ≥ 15% or RS ≥ 5% if SF3B1 mut ) [1].
Following up on this, the international working group for the prognosis of MDS (IWG-PM) proposed MDS with mutated SF3B1 as a distinct entity if certain criteria are fulfilled (Supplementary Table  S1) [11]. These criteria included: (1) cytopenia as defined by standard hematologic values, (2)  The upcoming 5th edition of the WHO Classification (WHO 2022) emphasizes a genetic basis for defining diseases and has now categorized MDS into morphologically defined MDS and MDS with defining genetic abnormalities (DGA) while largely abandoning the blast cut-off between MDS and AML if AML DGA are present [2]. It has further incorporated many of the proposed IWG-PM criteria into the newly introduced entity "MDS with low blasts and SF3B1 mutation" [2]. However, according to WHO 2022 only biallelic TP53 inactivations are excluded besides certain cytogenetic abnormalities (Supplementary Table S1). In contrast to the WHO 2022, the International Consensus Classification (ICC) requires an SF3B1 variant allelic frequency (VAF) ≥ 10% in the absence of certain cytogenetic abnormalities, RUNX1 and multi-hit TP53 (Supplementary Table S1) [12]. It further sets the blast cut-off for AML-DGA to 10%, while cases with 10-19% blasts without DGA are assigned as a new category MDS/AML. In this study, we defined the SF3B1 entity (SF3B1ent) based on the first publication proposed by the IWG-PM, but also discuss the changes in classification according to 5th edition of the WHO classification and ICC.
The aim of the study was to analyze the SF3B1 mutation and the proposed SF3B1 entity in a large cohort of 734 MDS patients with respect to the incidence, genomic landscape, AML transformation rate and clinical outcome.

MATERIAL AND METHODS Patients cohort and samples
For this analysis, we selected 734 MDS samples with material available to perform whole genome sequencing sent to the MLL Munich Leukemia Laboratory between 09/2005 and 12/2019. Diagnoses (from peripheral blood and bone marrow) were made based on cytomorphology, cytogenetics and molecular genetics as previously published [13][14][15]. All cases were classified into specific subgroups according to WHO 2017 [16]. For abbreviations of entities, see Table 1. Therapy-related MDS were excluded from this study. The MDS cohort comprised 310 (42%) female and 424 (58%) male cases with a median age of 73 years (range: 23-93 years) and a median follow-up of 9.3 years. All patients gave their written informed consent for genetic analyses and to the use of laboratory results as well as clinical data for research purposes according to the Declaration of Helsinki. The study was further approved by the laboratory´s institutional review board.
Whole genome sequencing (WGS) and variant filtering WGS analysis was performed for all patients. For this, total genomic DNA was extracted from lysed cell pellet of bone marrow or peripheral blood using the MagNA Pure 96 with DNA and Viral Nucleic Acid Large Volume Kit and Cellular RNA Large Volume Kit (Roche, Basel, Switzerland). Library preparation and sequencing as well as calling and filtering of single nucleotide variants, structural variants and somatic copy number variations (CNVs) were performed as previously described [17,18]. Copy neutral loss of heterozygosity (CN-LOH) was assessed using HadoopCNV.

Mutational analysis
In this study, we evaluated mutations in 73 genes associated with myeloid neoplasms for all patients from WGS data only or from combined WGS and targeted NGS panels (see supplementary material). Out of all 734 cases, 605 samples were additionally analyzed by targeted sequencing within a recent study [6] and 87 cases were analyzed by targeted NGS during routine diagnostics [19]. WGS data confirmed all mutations detected by targeted NGS panels and was further consulted for completing the mutational analysis of the 73 genes. The presence of FLT3-ITD and KMT2A-PTD were retrieved from WGS data only.

Statistical analysis
Statistical analyses were performed using SPSS version 19.0 (IBM Corporation, Armonk, NY). Analyses for overall survival (OS) and cumulative incidence (CI) of disease progression were performed according to Kaplan-Meier and compared using two-sided log rank tests. The OS was calculated as time from diagnosis to death or last follow-up. For the CI of disease progression death was considered as a competing event. Between different groups numerical variables were compared using the Mann-Whitney-U-Test, and dichotomous variables using chi-square test. Cox proportional hazards regression model was used to identify the impact of different variables on OS or AML transformation. All results were considered significant at p < 0.05.  (Table 1; Fig. 1A-C). In addition, 13% (37/300) of MDS with excess blasts (MDS-EB-1/2) and 20% (21/107) of MDS 5qharbored SF3B1 mutations together accounting for 25% (58/231) of all SF3B1 mut cases (Table 1; Fig. 1B, C). The remaining 1% of SF3B1 mut cases were an MDS-SLD and an MDS-MLD sample. Of note, SF3B1 mutations were most frequently found in patients with blast count <5% (192/419; 46%).
Notably, cytogenetic risk groups poor and very poor according to the Revised International Prognostic Scoring System (IPSS-R) were found in 11 SF3B1nent but in none of SF3B1ent cases ( Supplementary Fig. S5B). Within SF3B1ent 47% (67/144) did not harbor any additional mutation in 73 analyzed genes resulting in an average of 1.8 mutations (including SF3B1) in this group (Fig. 4), while 53% (77/ 144) harbored one to four additional mutations. SF3B1nent patients showed on average 2.6 mutations (MDS with isolated del(5q): 1.9; MDS-EB: 2.7; MDS-RS: 3.1; Fig. 4). Although SF3B1ent samples showed in total few mutations, additional mutations (if present) were detected in 27 different genes (Supplementary Fig.  S6A). Additional mutations in SF3B1nent samples were found in 9 to 20 different genes depending on the respective entity ( Supplementary Fig. S6B-F).
Prognostic impact of additional aberrations in SF3B1 mut MDS Next, we analyzed the prognostic contribution of additional gene mutations and other risk factors to OS in SF3B1 mut patients. Within all SF3B1 mut patients the number of mutations showed a significant impact on OS (Supplementary Fig. S9A; p = 0.040). However, within SF3B1ent cases, OS was not affected by the presence of additional mutations ( Supplementary Fig. S9B). In univariate analyses bone marrow blasts <5% was a good prognostic marker (hazard ratio HR: 0.616; p = 0.033), while RUNX1 mutations (HR: 4.347; p < 0.001), ASXL1 mutations (HR: 1.836, p = 0.023) and del(5q) (HR: 1.977; p = 0.008) were poor prognostic markers ( Table 2). Of note, the poor prognostic impact on OS of complex karyotypes did not reach statistical significance (p = 0.063) presumably due to the small samples size (n = 7). Further, del(5q) was not restricted to isolated del(5q) cases but comprised all cases with deletions on chromosome 5, and thus included also cases with complex karyotypes. In multivariate analysis only RUNX1 mutations (HR: 3.581; p < 0.001) and del(5q) (HR: 2.146; p = 0.003) were independent prognostic factors. Patients with SF3B1 mut having either del(5q) or RUNX1 mutations (n = 31) showed shorter OS compared to SF3B1 mut patients not having these abnormalities (n = 200; median OS: 43 vs. 88 months; p < 0.001; Supplementary Fig. S9C).

Molecular genetics of SF3B1 mut patients transforming to AML
Of SF3B1 mut patients 7% (15/231) progressed to AML compared to 15% (75/503) of SF3B1 wild-type patients (median follow-up: 9.3 years; Fig. 5A). In addition, time to AML was shorter in SF3B1 wildtype compared to SF3B1 mut patients (median: 14 vs. 27 months, p = 0.046; Fig. 5B). Notably, an AML transformation rate of 14% (12/87) was seen in SF3B1nent and 2% (3/144) in SF3B1ent (median follow-up: 122 and 112 months; Fig. 5A). A trend for a  longer time to AML transformation was observed for SF3B1ent compared to SF3B1nent, however not reaching statistical significance (71 vs. 17 months, p = 0.0825). Eleven of 15 SF3B1 mut MDS cases were also analyzed for the presence of molecular mutations at their diagnosis of AML (Fig. 5A, further details are provided in the supplement). Regarding the prognostic contribution of additional gene mutations and other risk factors to AML transformation in SF3B1 mut patients univariate analyses revealed bone marrow blasts <5% to be associated with lower risk (hazard ratio HR: 0.097; p = 0.021) and RUNX1 mutations (HR: 3.518; p = 0.05) with higher risk for AML transformation (Supplementary  Table S3).

DISCUSSION
SF3B1 mutations are frequently detected within MDS and associated with favorable prognosis [5][6][7]. In our WGS-based cohort of 734 MDS patients we identified 231/734 (31%) cases with SF3B1 mutations verifying known hotspots in K700, K666 and H662 [4,7,8,20,21] and confirming a heterozygous SF3B1 mutation status with high median VAFs (35%) across all entities. VAFs >30% were observed in 77% of SF3B1 mut samples. SF3B1 mutations persisted over the entire disease courses in many AML-transforming patients supporting that SF3B1 mutations are early events in MDS. However, SF3B1 mut samples with low VAFs (<15%) were mainly found in SF3B1 non-entity cases showing excess blasts, del(5q) or TP53 mutations but also in two SF3B1ent samples harboring other spliceosome mutations. In line with previous reports, SF3B1 mutations were predominantly found in MDS-RS-SLD/MLD supporting the association of SF3B1 with RS [7,10]. Moreover, we confirmed that SF3B1 mutations in MDS were favorable with regard to OS and AML transformation [7,10,11].
Recently, the IWG-PM suggested MDS with mutated SF3B1 as a distinct entity [11]. In this study, we evaluated the IWG-PM proposed SF3B1 entity criteria. We confirmed the favorable clinical  Fig. 4 Molecular characterization of SF3B1 mut MDS patients. Illustration of all 231 samples, each column represents one patient. Genes (gray: wild-type; red: mutated) as well as the WHO entity (incl. SF3B1ent) are given for each patient. Light green: SF3B1ent patients with isolated SF3B1 mutation; VAF variant allelic frequency. outcome of SF3B1 entity similar to recently published studies [21,22]. Additionally, in line with Komrokji et al. we observed a significantly longer OS of SF3B1 entity patients compared to SF3B1 non-entity patients, in contrast to Venable et al. who did not observe significant differences in OS between SF3B1ent and SF3B1nent presumably due to the small cohort size [21]. In contrast to Malcovati et al. [11], we observed that SF3B1 mutations were associated with significantly shorter OS within MDS 5q-concordant with previous reports [19,22,23] highlighting the adverse prognostic impact of mutated SF3B1 within this entity. Within our SF3B1 mut cohort, only 5% (11/231; all SF3B1nent) had poor or very poor cytogenetic risk groups concordant with a previous report [11] adding to the reasons for the favorable prognosis of SF3B1 mutations. In this line, the lately published IPSS-M, a unique risk score, improves the risk stratification of MDS patients by including molecular genetics into their model [24], in contrast to the IPSS-R, which considers only morphological features and cytogenetics [25]. The IPSS-M model further incorporates SF3B1 mutations with different weights depending on co-abnormalities (i.e. isolated del(5q) or BCOR, BCORL1, RUNX1, NRAS, STAG2, SRSF2 mutations).
With regard to the mutational landscape of SF3B1 mut cases the most frequent additional mutations were DNMT3A, TET2, and ASXL1 (DTA) similar to previous reports showing that epigenetic and histone modifiers are commonly mutated in MDS, but also in aging individuals [5,21,[26][27][28]. The number of additional mutations significantly impacted on OS in all SF3B1 mut patients. Within SF3B1ent the number of co-mutations did not affect OS as shown in IWG-PM results [11], however the number of comutations was low compared to SF3B1nent cases. In line with previous studies [5,10], we confirmed that progression to AML occurs at a relatively low frequency in SF3B1 mut patients (7%; 15/231). Furthermore, AML transformation was less frequent in SF3B1ent compared to SF3B1nent (2% vs. 14%). Progression of MDS to AML is suggested to be driven by cooperating genetic lesions [28,29]. In this regard, we found that AML-transforming patients harbored on average more mutations than non-progressing patients (3.2 vs. 2.0). We further demonstrated that at MDS diagnosis 47% (7/ 15) of AML-transforming patients showed RUNX1 mutations, significantly more frequent in AML-transforming compared to non-transforming patients. Moreover, during disease progression chromosomal aberrations were gained in two cases whereas most frequently RUNX1 mutations were acquired (n = 3) highlighting the role as potential driver gene and confirming the strong adverse prognostic value of RUNX1 mutations [11] associated with worse OS and a higher AML transformation rate within SF3B1 mut patients as also shown by Komrokji et al. [22].
Recently, the 5th edition of WHO and the ICC introduced MDS with mutated SF3B1 as a new entity [2,12]. The entity criteria proposed by the WHO 2022 and ICC mainly follow those suggested by the IWG-PM but differ in excluding mutations. Further, it is stated that the diagnostic criteria of MDS 5q-remain and that an SF3B1 mutation does not per se override this diagnosis. This is supported by our data as SF3B1 mutations show a negative impact on OS in MDS 5q-and do not seem to be the defining mutation in this setting, as suggested by the frequently low SF3B1 VAF. In contrast to the IWG-PM proposal, both WHO 2022 and ICC guidelines exclude biallelic TP53 inactivations from the SF3B1 entity. In our SF3B1 mut cohort, only two samples (2/231; <1%) harbored biallelic TP53 inactivations (both with blast count >5%). Concordant with the IWG-PM, ICC also excludes RUNX1 mutations from the SF3B1 entity supported by our data showing RUNX1 mutations as independent negative prognostic factors for OS and AML transformation.
In our univariate analysis, we confirmed the negative prognostic impact of del(5q) on OS of SF3B1 mut cases and additionally found a negative impact of blast count >5% as well as RUNX1 and ASXL1 mutations. However, our multivariate analysis could not confirm the independent prognostic impact of blast count >5%, but showed del(5q) and RUNX1 mutations as independent prognostic markers. Thus, based on our data the threshold of <5%, which is used by IWG-PM, ICC and WHO 2022, is not required if presence of del(5q) and RUNX1 mutation are exclusion criteria for the SF3B1 entity. Of note, studies from Malcovati et al. showed a significant impact of excess blasts on the survival of SF3B1 mut patients [10,11], however, RUNX1 mutations were not included in their multivariate analysis.
In conclusion, SF3B1 mutations are associated with good clinical outcome. Patients fulfilling the criteria of the SF3B1 entity proposed by the IWG-PM show an even better prognosis (longer OS, lower AML transformation rate). Our data suggest that the identification of the good prognostic subset within SF3B1 mut patients can be achieved by excluding only cases with del(5q) and/or RUNX1 mutations, however completely independent of blast count.

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