Venetoclax abrogates the prognostic impact of splicing factor gene mutations in newly diagnosed acute myeloid leukemia

Pathogenic mutations in splicing factor (SF) genes leads to abnormal functioning of the spliceosome complex and promotes oncogenesis through exon skipping and intron retention secondary to alternate splicing.¹ In The Cancer Genome Atlas project, acute myeloid leukemia (AML) was associated with the largest burden of alternate spliced events.² The 4 most commonly studied SF genes in myeloid malignancies are SRSF2, U2AF1, SF3B1, and ZRSR2. In myelodysplastic syndrome (MDS), SF3B1 mutation is associated with a characteristic increase in ringed sideroblasts and improved response to specific therapies (ie, luspatercept),³ whereas the other SF mutations (SF^mut) are associated with inferior outcomes.^4,5 

In AML, SF^mut are enriched in patients with secondary AML (s-AML) or de novo myelodysplasia-related AML.^1,6,7 Recent data have shown that these mutations are associated with an inferior outcome in patients with AML.⁷ In a cohort of 500 patients with de novo AML in which 363 (73%) received intensive (INT) therapy, Hou et al demonstrated that SF^mut had an independent negative prognostic impact.⁸ In another study, van der Werf et al showed that SF^mut tend to group with RUNX1 mutations and led to outcomes comparable with patients with historically adverse risk (AR) AML.⁹ Based on results from these and other studies, the European LeukemiaNet (ELN) 2022 recommendations on diagnosis and management of AML included SF^mut in the AR category, unless co-occurring with other favorable-risk factors (ie, NPM1 or bZip CEBPA mutations), which entails important therapeutic interventions, including the consideration of allogeneic stem cell transplantation (SCT) in first remission.¹⁰ However, a majority of the patients with SF^mut AML in the above studies were younger and received INT chemotherapy–based treatment, raising a question on whether these adverse outcomes are seen in all patients and with all treatment regimens. The ELN prognostic risk scoring is largely derived from data on the outcomes of younger patients treated with INT therapy and might not always be applicable to older patients treated with low-intensity (LI) therapy.

Venetoclax in combination with LI therapy is now approved as frontline therapy for patients aged ≥75 years or young unfit patients with newly diagnosed (ND) AML, based on results from the VIALE-A and VIALE-C trials.^11,12 In a previous analysis from our group, Lachowiez et al showed that in patients with ND AML treated with a hypomethylating agent (HMA) and venetoclax, outcomes in patients who had an SF^mut (n = 39) were similar to those with wild type (wt) SF (SF^wt) (n = 80).¹³ The majority of patients with ND AML treated with LI therapy now receive it in combination with venetoclax, and venetoclax based combinations are also being increasingly studied in combination with INT chemotherapy.¹⁴ In view of these ongoing advances in treatment options, and to better understand the impact of SF^mut in reference to their association with AML cytogenetics (CTG) and other concurrent myeloid mutations, we evaluated the impact of SF^mut in patients with ND AML and compared their outcomes with patients without these mutations. We factored in the patient age, baseline AML genomics, history of previous myeloid hematologic disorder, intensity of frontline treatment regimens, and the use of venetoclax among other prognostic factors to evaluate the independent prognostic significance of SF^mut.

This is a retrospective study including adult patients (≥18 years) with ND AML, treated at The University of Texas MD Anderson Cancer Center, Houston, from January 2017 to April 2022. Clinical data were captured from our methodically curated computerized patient database. Patients were stratified into young (≥18-59 years) and old (≥60 years) age at AML diagnosis. Baseline performance status based on the Eastern Cooperative Oncology Group was captured for all patients. Patients with previous nonmyeloid disorders who had received chemotherapy or radiotherapy were considered as therapy-related AML (t-AML). Patients with previous myeloid disorders were considered as s-AML, and they were further stratified as untreated secondary, if exposed to only erythropoiesis stimulating agents or lenalidomide, and treated secondary, if exposed to HMA or other therapy.

All frontline treatment data were captured and stratified into INT and LI therapy. Patients who received HMA based doublets or triplets and low-dose cytarabine–based doublets or triplets were considered in the LI therapy arm, and all therapies with higher intensity (ie, cytarabine at a dose of ≥1000 mg/m²) were included in the INT arm. Further details of the treatment regimens received are in supplemental Tables 1 and 2, available on the Blood website. The use of venetoclax as part of the frontline therapy was additionally documented.

All patients had an 81-gene myeloid next generation sequencing (NGS) mutation assessment through high throughput sequencing performed within our in-house Clinical Laboratory Improvement Amendments–certified laboratory. Details of the methodology are in the supplemental data. CTG was performed via conventional G-banded karyotyping, and AML directed fluorescent in situ hybridization was conducted as appropriate.

The SF^mut included in our analysis were SRSF2, SF3B1, U2AF1, and ZRSR2 (supplemental Table 3). A cutoff of ≥2% variant allele fraction on NGS was considered as positive for mutation, and germ line variants and variants of undetermined significance were discounted. Given the inclusion of SF^mut in ELN 2022 AML risk stratification, we used the ELN 2017 recommendations for risk stratification.¹⁵ 

Treatment response to the induction regimen was tabulated as per the ELN 2017 guidelines.¹⁵ Response was documented as overall response rate (ORR) and included a combination of complete remission, complete remission with incomplete count recovery, and morphological leukemia free state; this was the best response anytime on frontline therapy and before proceeding to SCT or any maintenance therapy. Relapse-free survival (RFS) was tabulated from the time of best response to relapse or death and overall survival (OS) from the time of therapy initiation to death; RFS was not censored for SCT. SCT conducted only in first remission was included.

Baseline demographics and disease parameters were documented and compared between patients with and without SF^mut. Continuous variables were compared using a Mann-Whitney U test and categorical variables using a 2-sided Fisher t test with a significance of <.05.

Kaplan-Meier estimates were used for RFS and OS, and a Mantel-Cox log-rank test was used for comparison. Logistic regression analysis was performed to identify factors associated with response. For analysis of factors affecting survival, a Cox proportional hazard model was used for univariate analysis (UVA) and multivariate analysis (MVA). Analysis was done using GraphPad Prism version 9.3.1 (GraphPad Software, Boston, MA) and the Lumivero (New York, NY) (2023) XLSTAT statistical solution.

From January 2017 to April 2022, 994 patients with ND AML were treated at our center, of whom 456 (46%) were females, with a median age of 67 years (range, 19-95 years). Baseline characteristics are detailed in Table 1. A total of 66 patients (9%) had core-binding factor (CBF) AML, of whom none had a SF^mut. Patients with CBF-AML were excluded from further comparative analysis for response and survival outcomes. In total, 266 patients (27%), with a median age of 72 years (range, 27-95) had an SF^mut, of whom 76 (29%) were females. The median age of the patients without SF^mut was 65 years (range, 19-64); Patients with SF^mut were older (225/266 [80%]) than patients with SF^wt (463/728 [64%]; P < .0001). There was no difference in the frequency of the patients in different Eastern Co-operative Oncology Group performance status strata based on the status of SF^mut and whether they received therapy with venetoclax (supplemental Table 4).

The most common SF^mut was SRSF2, present in 140 patients with SF^mut (53%), followed by U2AF1 in 74 patients (28%), SF3B1 in 49 patients (18%), and ZRSR2 in 11 patients (4%). Only 8 patients (3%) had >1 SF^mut consistent with previous reports.^8,9 Overall, 88 of 258 (34%) patients with available CTG data had ELN 2017 AR CTG (54/258 patients [21%] with a complex karyotype). In the SF^wt group (n = 662), CTG data were available for 644 patients, of whom 309 patients (48%) had AR CTG (254/644 [39%] with a complex karyotype). Using a combination of the available CTG data and myeloid mutation data, all patients could be stratified by ELN 2017 risk stratification. Significantly more patients in the SF^mut group had AR disease (n = 186 [70%]) than the SF^wt group (n = 386 [58%]; P = .001). This was driven by a higher frequency of ASXL1 (32% vs 10%, P < .0001) and RUNX1 mutations (29% vs 9%, P < .0001) in the SF^mut group. There were 4 (0.6%) patients with biallelic CEBPA in the SF^wt group compared with 1 (0.4%) in the SF^mut group; 2 patients with SF^mut (0.8%) and 16 patients with SF^wt (2.4%) had a bZip in frame mutation in CEBPA in the absence of concurrent AR CTG or TP53 mutation. Twenty-six (9.8%) SF^mut and 123 SF^wt (18.6%) patients had an NPM1 mutation in the absence of AR CTG, TP53 mutation or FLT3-ITD high allelic ratio (NPM1 favorable risk), P = .0007.

Supplemental Figure 1 highlights the association of SF^mut with other relevant myeloid mutations, and Figure 1A-B shows the heatmap of the relevant myeloid mutations for the SF^mut and SF^wt groups, respectively. The overall frequency of TP53 mutations was lower in patients with SF^mut than those with SF^wt (14% vs 31%, P < .0001). Among the patients with SF^mut specifically, TP53 mutation(s) were highest in the SF3B1 mutated group (12/49 [24%] vs 26/217[12%]) in the remaining patients with SF^mut (P = .039) (Table 2; supplemental Figure 2). Patients with SRSF2 mutations had the lowest frequency of AR CTG (24%), compared with 43% with U2AF1 (P = .009), 47% with SF3B1 (P = .005), and 62% in the ZRSR2 (P = .03) groups. In the SF3B1 group, 12 of 48 (25%) patients with available CTG data had a chromosome 3 aberration, of whom 9 had inv(3)(q21;q26.2) or t(3;21)(q26.2;q11.2). The overall frequency of ELN AR disease was not significantly different among the different SF^mut groups; representing 67%, 70%, 79%, and 88% patients with SRSF2, U2AF1, SF3B1, or ZRSR2 mutations, respectively.

A significantly higher number of patients with SF^mut had s/t- AML than those in the SF^wt group; 114 (43%) with s-AML (74 [28%] treated secondary) and 21 (8%) with t-AML in the SF^mut group, compared with 195 (29%) s-AML (131 [20%] treated secondary) (P = .002) and 98 (15%) t-AML (P = .05) in the SF^wt group. In both de novo and s/t-AML groups, more patients in the SF^mut group were older (79% vs 58% in the de novo group and 90% vs 75% in the s/t-AML group for SF^mut and SF^wt respectively) (supplemental Table 5; supplemental Figure 3).

Of the 266 patients in the SF^mut group, 135 (51%) had s/t-AML, including 114 with s-AML. MDS was the most common myeloid disorder, seen in 72 patients (63%), followed by myeloproliferative neoplasm in 22 patients (20%), chronic myelomonocytic leukemia in 17 patients (15%), and chronic myeloid leukemia and blastic plasmacytic dendritic cell neoplasm in 1 patient each. Overall, 62 of 114 patients (54%) had exposure to HMA any time before their diagnosis of AML; 12 (11%) had exposure to venetoclax, and 40 patients (35%) had received no therapy at all (n = 30) or had exposure to only luspatercept, lenalidomide, ruxolitinib, imatinib, or tagraxofusp (n = 9), before therapy for AML. Only 3 patients (2.6%) with s-AML had a previous SCT.

A significantly higher percentage of patients in the SF^mut group received LI therapy (80% vs 61%, P < .0001) than those in the SF^wt group, possibly attributed to the higher number of older patients in the former (supplemental Table 6). Equal proportion of patients with SF^mut and those with SF^wt received venetoclax as part of their frontline AML therapy, with either LI or INT regimens. There was no difference in the ORR after LI or INT therapy in the 2 groups (LI = 68% vs 67% and INT = 79% vs 81% for SF^mut and SF^wt groups, respectively).

The median follow-up of the entire cohort of patients was 26 months (range, 2-59). The median RFS was 9.6 vs 10.6 months (P = .04) and the OS was 13.1 vs 11.9 months (P = .27) for patients with SF^mut and those with SF^wt, respectively. Although RFS and OS were expectedly superior in younger patients compared with older patients, there was no significant difference in RFS and OS in individual age groups based on SF^mut (supplemental Figure 4). Patients with SF^mut with ELN AR disease had similar RFS (8.3 vs 6.2 months, P = .14) but superior OS (10.1 vs 7.8 months, P = .02) compared with those with SF^wt (supplemental Figure 5). There was no difference in survival outcomes in patients with the NPM1–mutated favorable-risk AML based on SF^mut (supplemental Figure 6).

Given the significant heterogeneity in patient age and intensity of therapy received, we further stratified survival outcomes and compared patients based on the presence or absence of SF^mut and the above characteristics.

More patients with SF^mut than those with SF^wt received LI therapy as mentioned before. We compared the RFS and OS of the patients in these 2 arms by selecting patients aged ≥60 years who received LI therapy. Among the 200 older patients in the SF^mut arm who received LI therapy, 137 (69%) had an ORR with a median RFS of 9.3 months compared with an RFS of 7.7 months in the SF^wt arm, in which 253 of 366 older patients (69%) attained an ORR (P = .35); the median OS was 12.3 and 8.5 months, respectively, in the 2 groups (P = .14) (Figure 2A and B). Comparatively, in the above-mentioned treatment groups, the frequency of ELN AR AML, AR CTG, and TP53 mutation was 140 (70%), 65 of 194 (34%), and 28 (14%) in the patients with SF^mut and 239 (65%), 196 of 358 (55%) and 144 (39%) in the patients with SF^wt, respectively.

In the SF^mut group, 129 of 200 older patients (65%) received venetoclax along with LI therapy, whereas 254 of 366 patients (69%) received venetoclax along with LI therapy in the SF^wt group. When stratified by exposure to venetoclax, the median RFS in the venetoclax + LI treated patients with SF^mut was 9.8 months and OS was 14.1 months, compared with a RFS of 9.1 months (P = .5) and OS of 9.6 months (P = .11) in patients with SF^wt (Figure 2C-D). The median RFS and OS were similarly inferior in both mutational groups not treated with venetoclax (RFS 5.3 vs 5.1 months and OS 6.8 vs 7.1 months for patients with SF^mut and SF^wt, respectively).

Patients with s/t-AML have inferior outcomes compared with patients with de novo AML. Among the older patients treated with LI therapy, 104 of 200 (52%) patients with SF^mut and 195 of 366 patients with SF^wt (53%) had s/t- AML; the RFS was 6 vs 5 months (P = .64), whereas the OS was 9.1 vs 6.6 months (P = .08) for the 2 groups, respectively (Figure 2E-F). In the de novo AML subset, the median RFS was 13.7 vs 9.8 months (P = .28) and OS was 19.1 vs 13.6 months (P = .44) for patients with SF^mut and SF^wt, respectively (Figure 2G-H). To better understand the impact of venetoclax addition to LI therapy, we stratified the survival of patients who received venetoclax in the s/t-AML and de novo groups. Cumulatively, in the s/t-AML arm, 60 of 104 patients (58%) with SF^mut and 128 of 195 patients (66%) with SF^wt had received venetoclax; the RFS and OS in these patients were 6.1 vs 5 months (P = .99) and 9.8 vs 7 months (P = .28) for patients with SF^mut and wt, respectively. In the de novo group, 70 of 96 patients (73%) with SF^mut and 126 of 171 patients (74%) with SF^wt had received venetoclax with LI therapy; RFS and OS was 19.1 vs 12 months (P = .19) and 24.8 vs 14.9 months (P = .19) in the 2 patient groups, respectively.

A minority of younger patients received LI therapy (SF^mut = 13/266 and SF^wt = 38/662); the median RFS (18 vs 6.5 months) and OS (13.8 vs 7.9 months) were numerically higher in patients with SF^mut than in those with SF^wt, however, this was not statistically significant.

In the SF^mut group, a smaller percentage of patients (53/266 [20%]) received INT therapy than those in the SF^wt group (258/662 [39%], P < .0001). Among them, a higher proportion of patients in the SF^mut group were ELN AR (66% vs 44%, P = .006) and had s/t-AML (45% vs 28%, P = .02) than those with SF^wt (supplemental Table 7). In the INT therapy cohort, the median RFS and OS in the SF^mut group (9.6 and 15.9 months) were significantly shorter than the median RFS of 21.4 months (P = .04) and OS of 26.7 months (P = .06) in the SF^wt group (Figure 3A-B), consistent with the perceived AR biology of SF^mut AML treated with standard INT chemotherapy. Because venetoclax is now administered to a significant percentage of patients in combination with INT therapy at our center, we compared the outcomes of patients treated in the INT arm who received venetoclax as part of their frontline AML therapy. A total of 29 of 53 patients (55%) with SF^mut and 131 of 258 patients (51%) with SF^wt received venetoclax with their INT therapy, and ORR was attained in 25 (86%) and 108 (82%) patients, respectively. The median RFS was 15.4 vs 20.3 months (P = .36) and median OS 19.6 vs 30.7 months (P = .98) in the SF^mut and SF^wt groups (Figure 3C-D).

We subsequently selected only younger patients (aged <60 years) who received INT therapy (SF^mut = 27 and SF^wt = 191) with or without venetoclax; the median RFS was 13.5 vs 25.4 months (P = .20) and median OS was 39.5 vs 58.7 months (P = .99) in the SF^mut and SF^wt groups, respectively. Considering the use of venetoclax in this younger population treated with INT therapy (SF^mut = 18 and SF^wt = 107), the median RFS and OS were 18.3 vs 20.3 months (P = .82) and the median OS was not-reached vs 30.7 months (P = .22) in the SF^mut and SF^wt groups, respectively.

A slightly higher proportion of patients with SF^wt proceeded to SCT after frontline LI/INT therapy (44% vs 32%, P = .13), possibly because the SF^mut group was enriched with older patients (47% vs 26%). On analyzing the impact of SCT in younger patients treated with INT therapy, SCT equivocally improved RFS and OS in both patients with SF^mut and those with SF^wt (supplemental Figure 7).

We analyzed the survival outcomes of patients with SF^mut based on the individual SF^mut. Given the small number of patients in the ZRSR2 mutated group, we discounted them from the comparative survival outcomes, along with 8 patients who had ≥1 SF gene mutation. The median RFS was 9.6, 9, and 6 months for patients with SRSF2, U2AF1, or SF3B1 mutations, respectively, and the median OS was 15.2, 10.1, and 8 months, respectively (P = .01) (supplemental Figure 8). Given the frequent association of patients with SF^mut with ASXL1 and RUNX1 mutations, we stratified the survival outcomes of patients with SF^mut based on those who had AR CTG ± TP53 mutation and those who had ASXL1 ±, RUNX1 mutation; the median RFS was 5.6 vs 12 months (P = .07) and the OS was 6.9 vs 15.7 months (P = .003) between the former and the latter groups, respectively. Importantly the RFS and OS of the ASXL1 ± RUNX1 mutated group was similar to that of patients with SF^mut without ELN AR AML (RFS of 12.9 months and OS of 20.6 months).

Understanding the significant heterogeneity in the patient population, we next evaluated additional factors that could affect response and survival. On UVA, SF^mut did not affect the attainment of ORR; de novo AML and use of venetoclax was associated with higher odds of ORR, whereas age ≥60 years, ELN AR, and use of LI therapy was associated with lower odds of ORR (Table 3). On MVA, SF and age were not significant, whereas all the other variables remained independently significant.

We used Cox regression analysis to understand factors predicting survival. To reduce the wide patient age and treatment heterogeneity we ran the model in prestratified groups. In the first analysis, we looked at factors significant in older patients treated with LI regimen. We included SF^mut status, ELN 2017 risk (adverse or not), de novo vs s/t-AML, and use of venetoclax as variables. On UVA, SF^mut did not affect the hazard of relapse, whereas ELN AR increased the hazards, and conversely, de novo AML and use of venetoclax were associated with lower hazards of relapse. On MVA, the same attributes were maintained except for SF^mut which favorably reduced the hazards of relapse. SCT reduced the hazard of relapse on both UVA and MVA. For assessing hazards of death (OS), the same variables were evaluated; on UVA, SF^mut did not have any effect, whereas ELN AR disease increased the hazards of death, and de novo AML, use of venetoclax, and SCT was associated with lower hazards of death. In addition, on MVA for OS, SF^mut appeared to favorably reduce hazards of death whereas the other variables remained consistent (Table 3).

Next, we analyzed factors predicting RFS and OS in younger patients who received INT therapy and included the same variables as the previous analysis. On UVA for RFS, SF^mut, venetoclax, and de novo AML were not significant and ELN AR increased the hazards, whereas SCT was associated with lower hazards for relapse; on MVA, ELN AR remained a significant (negative) prognostic factor for RFS and SCT was favorable. For OS, on UVA, de novo AML and SCT were protective, whereas ELN AR increased hazards of death, and on MVA, only ELN AR and SCT remained significant; SF^mut had no impact on hazards of relapse or death (Table 3).

We analyzed the impact of the SF^mut using another Cox regression model replacing the ELN2017 AR variable with the individual myeloid mutational data, including NPM1, ASXL1, RUNX1, TP53, and AR CTG, while including the other variables mentioned in the previous model (supplemental Table 8). The frequency of TP53 mutation was lower in the SF^mut group (14%) than in the SF^wt group (31%) (Table 1). In both older patients treated with LI therapy and younger patients treated with INT therapy, SF^mut did not have any independent prognostic impact, whereas TP53 mutation was independently associated with increased hazards of relapse (hazard ratio [HR], 1.36; 95% confidence interval [CI], 1.00-1.86 for older patients treated with LI and HR, 5.37; 95% CI, 2.47-11.67 for younger patients treated with INT) and death (HR, 1.38; 95% CI, 1.07-1.79 for older patients treated with LI and HR, 3.64; 95% CI, 1.86-7.11 for younger patients treated with INT). ASXL1 and RUNX1 mutations did not show any independent prognostic impact on the hazards of relapse and death in both treatment groups.

Among the patients with SF^mut, we ran Cox regression to understand the impact of specific SF^mut on relapse and death. Although no individual SF^mut had any significant impact on RFS, SRSF2 mutation reduced the hazards of death on UVA but not on MVA. The use of venetoclax remained protective for relapse and death on UVA and MVA (supplemental Table 9).

We present here a large retrospective analysis of patients with ND AML and describe the impact of SF^mut on responses and survival in these patients. In this cohort, SF^mut did not impact the rates of response to frontline therapy or survival outcomes, irrespective of the patients age and AML risk stratification. Overall, 266 (29%) of patients with non–CBF-AML (n = 928) in our cohort had an SF^mut, and in concordance with other published data, only 8 patients (3%) had >1 SF^mut, and SF^mut occurred more commonly in male patients (71% of patients with SF^mut).⁹ 

SF^mut have significant relevance in myeloid neoplasms, and have been associated with distinct phenotype and outcomes in patients with MDS.¹ In AML, the recent inclusion of SF^mut as an adverse prognostic factor in the ELN 2022 risk stratification has added increased clinical relevance.¹⁰ Our analysis suggests the negative prognostic impact of SF^mut emanates primarily from studies involving younger patients receiving INT therapy that did not include venetoclax.^8,9 This negative prognostic impact of SF^mut in younger patients receiving standard INT therapy was also seen in our analysis, with a shorter RFS and OS in the unsupervised SF^mut group. However, when we selected among them those patients who had received venetoclax along with the INT therapy, there was no difference in survival outcomes; on UVA and MVA, SF^mut did not affect the hazard of relapse or death in young patients treated with INT therapy. In the group of patients treated with LI therapy, SF^mut led to lower hazards of relapse and death independently.

SF^mut are more common in patients with s/t-AML and tend to associate more often with ELN 2017 AR strata.¹⁷ Patients with s/t-AML have inferior outcomes compared with patients with de novo AML, especially if they have received previous therapy.^18,19 In our study, 51% of patients with SF^mut had s/t-AML (28% treated secondary). Furthermore, more patients with SF^mut had ELN 2017 AR disease, fueled by a higher frequency of RUNX1 and ASXL1 mutations in these patients, although the frequency of AR CTG and TP53^mut was lower in patients with SF^mut than in patients with SF^wt. In our study, on MVA, when adjusting for baseline patient and disease factors including other AR mutations, SF^mut did not negatively affect response rates and survival outcomes.

The retrospective nature of our study along with the heterogeneity in patient age and treatment regimens are relevant limitations. In our analysis, across both age groups and treatment intensity, we show that SCT was associated with lower hazards of relapse/death. However, this benefit could be confounded by the depth of response to frontline therapy, pre-SCT performance status, and comorbidities, and possibly cannot be attributed to SCT alone. Within these limitations, to our knowledge, we showed for the first time in a large cohort of patients, that SF^mut do not have independent negative prognostic significance, especially when patients are treated with venetoclax containing regimens. Stratified MVA shows that even in the SF^mut group, the use of venetoclax was associated with lower hazards of relapse and death. Newer biological evidence demonstrates that altered splicing may lead to increased susceptibility to venetoclax and warrants focused clinical studies and trials to evaluate the responses and outcomes with the use of venetoclax in patients with SF^mut.^20,21 

In our study, we used an age cutoff of 60 years at AML diagnosis to stratify the study population. We acknowledge this is an arbitrary cutoff, and the intention to proceed to INT or LI therapy is often determined by the AML genomics, institutional guidelines, patient performance status and comorbidities, and physician preference. Venetoclax is approved for treatment of older and unfit patients with ND AML along with HMA or low-dose cytarabine, and clinical trials are evaluating venetoclax in combination with INT therapy.^14,16 Secondly, SF^mut are enriched in older patients with AML, who are more often treated with LI therapy regimens containing venetoclax. Given these important considerations, the independent prognostic significance of SF^mut in ND AML should be revisited based on treatment approach and intensity. Efforts to develop risk stratification models considering venetoclax as part of the treatment regimen are much needed and ongoing²²; for example, a recent analysis from the VIALE-A study has shown that the ELN 2017 risk stratification does not optimally prognosticate patients treated with HMA + venetoclax, and SF^mut do not confer a negative prognostic impact.²³ Further validation of the ELN 2022 prognostic risk stratification in large patient cohorts treated with venetoclax based combination regimens is warranted.

In conclusion, SF^mut are associated with older patients with AML, enriched in patients with s/t-AML, and occur most commonly in the ELN 2017 AR strata. In our cohort of patients, a majority received venetoclax along with LI or INT therapy, and in this treatment milieu, SF^mut failed to show an independent prognostic impact. Newer prognostic tools for AML that incorporates history of prior myeloid disorders and therapy exposure, use of venetoclax for AML therapy, along with the AML genomic landscape are increasingly important for precise risk stratification of these patients.

The authors are grateful to the patients who were treated at The University of Texas MD Anderson Cancer Center, Houston, and were included in this study.

The visual abstract was created with BioRender.com.

Contribution: H.M.K. and C.D.D. conceptualized the study; J.S. and C.D.D. designed and wrote the manuscript; J.S. analyzed the data and made the figures; J.S. and S.L. risk stratified the patients; S.L. and K.P. analyzed the laboratory data; N.J.S., G.C.I., A.M., H.A.A., N.G.D., N.P., K.S.C., K.S., T.M.K., D.E.H., G.B., F.R., H.M.K., G.G.-M., and C.D.D. treated patients; S.P. curated the data; C.D.D. supervised the study; J.S. and C.D.D. reviewed the data; and all authors reviewed the manuscript and approved the final version.

Conflict-of-interest disclosure: N.G.D. has received research funding from Daiichi Sankyo, Bristol Myers Squibb, Pfizer, Gilead, Sevier, Genentech, Astellas, Daiichi Sankyo, AbbVie, Hanmi, Trovagene, FATE Therapeutics, Amgen, Novimmune, Glycomimetics, Trillium and ImmunoGen; and has served in a consulting or advisory role for Daiichi Sankyo, Bristol Myers Squibb, Arog, Pfizer, Novartis, Jazz, Celgene, AbbVie, Astellas, Genentech, ImmunoGen, Servier, Syndax, Trillium, Gilead, Amgen, Shattuck Labs, and Agios. N.P. serves on the board of directors for Dan’s House of Hope; is a consultant for AbbVie, Aptitude Health, Astellas Pharma US, Inc, Blueprint Medicines, Bristol Myers Squibb, Celgene Corp, Cimeio Therapeutics AG, ClearView Healthcare Partners, CTI BioPharma, Dava Oncology, Immunogen, Incyte, Intellisphere, LLC, Novartis AG, Novartis Pharmaceuticals Corp, OncLive (owned by Intellisphere, LLC), Patient Power, PharmaEssentia, Protagonist Therapeutics, Sanofi-Aventis, Stemline Therapeutics, Inc and Total CME; has financial relationships (eg, stock, royalty, gift, employment or business ownership) with Karger Publishers; is a scientific/advisory committee member for Cancer.Net, CareDx, CTI BioPharma, EUSA Pharma, Inc, Novartis Pharmaceuticals Corp, Pacylex, and PharmaEssentia; and receives a speaker/preceptorship from AbbVie, Aplastic Anemia & MDS International Foundation, Curio Science LLC, Dava Oncology, Imedex, Magdalen Medical Publishing, Medscape, Neopharm, PeerView Institute for Medical Education, Physician Education Resource, Postgraduate Institute for Medicine, and Stemline Therapeutics, Inc. T.M.K. discloses consulting or advisory roles for Novartis, Jazz Pharmaceuticals, Pfizer, AbbVie/Genentech, Agios, Daiichi Sankyo/UCB Japan, Liberum, Sanofi, Servier, and Pinot Bio and research funding from Bristol Myers Squibb, Celgene, Amgen, BiolineRx, Incyte, Genentech/AbbVie, Pfizer, Jazz Pharmaceuticals, AstraZeneca, Astellas Pharma, Ascentage Pharma, Genfleet Therapeutics, Cyclacel, and Pulmotech. H.M.K. discloses honoraria, advisory board participation, and/or consultancy from AbbVie, Amgen, Amphista, Ascentage, Astellas, Biologix, Curis, Ipsen Biopharmaceuticals, KAHR Medical, Novartis, Pfizer, Precision Biosciences, Shenzhen Target Rx, and Takeda and research grants from AbbVie, Amgen, Ascentage, BMS, Daiichi-Sankyo, ImmunoGen, Jazz, Novartis. Cellenkos, Glycomimetics, Astex Pharmaceuticals, Iterion Therapeutics, and Delta-Fly Pharma. C.D.D. receives research funding or honoraria or consultancy fee from Cleave, Astex, LOXO, ImmuneOnc, Takeda, Gilead, Genmab, GSK, Kura, Foghorn, Astellas, Forma, bluebird bio, Servier, Novartis, Bristol Myers Squibb, AbbVie, and Jazz and previously held a membership on an entity’s board of directors or advisory committees or holds stock options in a privately-held company for Notable Labs. The remaining authors declare no competing financial interests.

Correspondence: Courtney D. DiNardo, Department of Leukemia, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Unit 428, Houston, TX 77030; e-mail: cdinardo@mdanderson.org.