Citations

Abstract Infection is increasingly recognized as a significant cause of morbidity and mortality in patients with relapsed/refractory (R/R) large B-cell lymphoma (LBCL) receiving CD19 chimeric antigen receptor (CAR) T-cell therapy. The current study analyzed the natural history, risk factors, and outcomes of infection in 3350 patients with R/R LBCL receiving commercial CD19 CAR T-cell therapy (n = 2804 axicabtagene ciloleucel [axi-cel], n = 546 tisagenlecleucel) from December 2017 to June 2022. Infection developed in 834 patients (24.9%) within 100 days after infusion, resulting in an infection density of 0.43 per 100 patient days and a 100-day cumulative incidence of 22%. Bacterial, viral, and fungal infections were recorded in 527 (15.7%), 374 (11.2%), and 108 patients (3.2%), respectively, with corresponding infection densities of 0.23, 0.15, and 0.04 per 100 patient days. After a 24-month median follow-up, 1482 patients (44%) had died, with infection as the primary cause in 173 cases (12%). The 100-day infection-related mortality (IRM) was 1.6% (95% confidence interval, 1.2-2.0). Patients with a Karnofsky performance score of ≤80, infection history before CAR T-cell therapy, axi-cel therapy, severe cytokine release syndrome (grade ≥3), and severe immune effector cell–associated neurotoxicity syndrome (grade ≥3) had increased infection risk. Infections within 100 days were an independent risk factor for inferior overall survival beyond day 100 after CD19 CAR T-cell therapy. In conclusion, study results show a significant incidence of infection and IRM in patients with R/R LBCL treated with CD19 CAR T-cell therapy. Furthermore, results identify patients at a heightened risk of infection, offering insights to guide potential interventions aimed at mitigating infection and improving patient outcomes after CAR T-cell therapy.

e18508 Background: BPDCN is a rare and aggressive hematologic malignancy. Despite the advent of targeted therapies, median overall survival (mOS) remains <1 year (y) in the absence of stem cell transplant (SCT). This study evaluates clinical and molecular predictors of outcomes in a large cohort of BPDCN patients at a tertiary center. Methods: BPDCN patients seen at Mayo Clinic (AZ, FL, MN) between 2000-2024 were identified retrospectively. Analyses considered parameters at diagnosis. Next generation sequencing (NGS) was performed using a 47-gene assay. Standard statistical methods were used; calculations used R. Results: Among 69 patients, the median age was 70y (range 19-94) with male predominance (83%). At presentation, skin (80%), bone marrow (73%), lymph nodes (52%), spleen (27%), and CNS (12%) were involved; 26% had a concurrent hematologic malignancy. Among 20 cases with NGS, the median number of mutations was 3 (range 0-5), most frequently TET2 (n=16), ASXL1 (n=13), SRSF2 (n=7), and NRAS (n=5). Frontline therapy (Table) yielded a complete response (CR) rate of 63% and overall response rate of 76%. CR rates were similar among regimens, but patients who received an ALL (hyper-CVAD based) regimen (8/13) or tagraxofusp (6/9) were more likely to be bridged to SCT (6/29, p<0.01). After median follow up of 38 months (mo), mOS for all was 15 (95% CI 12-21) mo; 44 (64%) were deceased at last follow up. Among 62 evaluable patients, 20 (32%, median age 60 vs 74y, p<0.01) underwent SCT which improved survival (mOS 51 vs 10 mo, 3y OS 52% vs 5%, p<0.01); mOS was not different between allogeneic (n=17) and autologous (n=3) SCT (54 vs 38 mo, p=0.45). When censored for SCT, the 1y OS for patients receiving an ALL regimen, tagraxofusp, or other regimen was 82%, 67%, and 47%, respectively (p=0.15). In univariate analysis, age ≥50y, lymph node involvement, bone marrow involvement, hemoglobin <10 g/dL, leukocytes (WBC) >11 x10 9 /L, abnormal karyotype, and HCT were significantly associated with OS. In multivariate analysis, WBC >11 x10 9 /L (HR 3.7, p<0.03), abnormal karyotype (HR 5.4, p<0.01), and SCT (HR 0.23, p<0.01) remained significant. Separately, SRSF2 mutation was associated with inferior OS (15 vs 51 mo, p=0.02), but significance was lost when adjusted for age, abnormal karyotype, and SCT. Conclusions: Newly-diagnosed BPDCN patients treated with an ALL regimen or tagraxofusp followed by SCT had longer OS compared to those treated with all other regimens. WBC >11 x10 9 /L and abnormal karyotype were independent predictors of inferior survival whereas SCT was beneficial. Regimen N Age (y) ANNOVA CR Chi Sq HCT Chi Sq 1y OS mOS (mo) Log-rank Tagraxofusp 9 68 (40-70) <0.001 7 (78%) 0.548 6 (67%) 0.018 88% 20.8 0.121 ALL 13 43 (20-72) 8 (67%) 8 (62%) 74% 50.7 Hyper-CVAD 11 43 (20-72) 6 (60%) 6 (60%) 67% 35.8 AML 18 70 (48-80) 13 (76%) 5 (28%) 58% 14.2 Intensive 8 66 (48-77) 7 (88%) 3 (38%) 57% 12.4 HMA+Venetoclax 8 69 (65-80) 6 (75%) 2 (25%) 50% 14.0 Lymphoma 8 74 (49-85) 3 (43%) 1 (13%) 50% 14.0 CHOP 6 74 (49-85) 3 (50%) 1 (17%) 50% 17.7 Other 3 66 (63-73) 1 (33%) 0 (0%) 33% 12.0

6558 Background: Allogeneic stem cell transplantation (ASCT) is currently the only curative therapy in chronic myelomonocytic leukemia (CMML). Methods: A Mayo Clinic enterprise-wide database search identified 138 CMML cases who underwent ASCT. Conventional statistical methods were used for analyses. Results: 138 CMML patients (transplanted between 1995-2024) were included (median age 63 years; males 62%);104 (group A) received ASCT before and 34 (group B) after blast transformation (BT). At initial diagnosis, CMML-1/CMML-2 and dysplastic/proliferative representations were 78%/22% and 54%/46%; 31% displayed abnormal karyotype and most frequent mutations were ASXL1 (56%), TET2 (44%) and SRSF2 (38%). At time of initial diagnosis, CPSS-Mol risk categories were low (17%), intermediate-1 (11%), intermediate-2 (40%), and high (32%). Median time from diagnosis to ASCT was 11 months (range 0-201). Median overall survival (OS) from the time of initial diagnosis was 67 months (range 4-239) and from the time of ASCT 54 (range 0-212) months. Occurrence of BT before ASCT was associated lower post-transplant survival (PTS; 16 vs 95 months, P=0.01, HR 1.9, 95%CI 1.2-3.2). Bone marrow (BM) blasts, at the time of ASCT, <5%, 5-9%, and 10-19% correlated with median OS of 171, 81, and 18 months in group A (p=0.01) and 50, 25, and 13 months in group B (P=0.07), respectively. Pre-ASCT hypomethylating agent exposure was associated with lower PTS in group A (P=0.02). PTS was also adversely affected by DNMT3A and SETBP1 mutations. In group A, PTS was the longest with myeloablative busulfan-based (median not reached) and the shortest (median 22 months) with Cy-TBI-based conditioning (p=0.1). Donor type did not impact PTS: matched unrelated (63%), matched sibling (23%), mismatched unrelated (7%), or haplo-identical (7%). Post-transplant cyclophosphamide was associated with a numerically lower median PTS (22 months), compared to other forms of GVHD prophylaxis (107 months; p=0.1) and significantly higher non-relapse mortality (p=0.02). Documentation of morphologic CR at day 100 was associated with significantly longer PTS in both groups A (92%,164 vs. 18 months; p=0.01) and B (84%, 42 vs. 11 months; p=0.01). The presence of abnormal karyotype at day 100 was associated with shorter PTS in group A (p=0.01). Grade ≥3 acute and moderate to severe chronic GVHD occurred in 20% and 34%, respectively, in group A patients and in 26% and 44% in group B. GRFS at 1/3 years were 42/21% in group A and 31/16% in group B. At last follow up, 44% in group A and 65% in group B were dead. Causes of death were relapse/non-relapse related in 36/64% in group A and 39/61% in group B. Conclusions: ASCT is effective in securing long-term survival in CMML, especially when the procedure is performed prior to BT and in the setting of <5% BM blasts/promonocytes. Response assessment at day 100 was highly informative of outcome.

Abstract Allogeneic hematopoietic stem cell transplant (alloHCT) is considered for all patients with myeloid neoplasms (MNs) harboring TP53 mutations (TP53mut). The aim of this international study across 7 transplant centers in the United States and Australia was to identify factors associated with improved post-alloHCT survival. Of 134 TP53mut MN cases who underwent alloHCT, 80% harbored complex karyotype; 94% of TP53 variants were localized to the DNA-binding domain (DBD). Most common comutations were ASXL1 (7%), TET2 (7%), and DNMT3A (6%). Median post-HCT survival was 1.03 years, and overall survival (OS) at 1 year, 2 years, and 3 years was 51.4%, 35.1%, and 25.1%, respectively. A total of 103 cases (76.9%) met the International Consensus Classification (ICC) criteria for MN with mutated TP53 (referred to as ICC-defined TP53mut MN hereafter). The 3-year OS of ICC-defined TP53mut was significantly shorter compared with that of other TP53mut MNs (3-year OS, 16.9% vs 54.9%; P = .002). ICC-defined TP53mut MNs was independently associated with inferior OS (hazard ratio [HR], 2.62; P = .019). The presence of non-DBD TP53mut only (HR, 3.40; P = .005), DNMT3A comutation (HR, 2.64; P = .016), and pre-alloHCT bone marrow blasts ≥5% (HR, 2.76; P = .006) was independently associated with inferior relapse-free survival (RFS), whereas melphalan-based conditioning was associated with superior RFS (HR, 0.52; P = .005). Combining these variables, we constructed a hierarchical model that stratified patients into low-, intermediate-, and high-risk categories with 1-year RFS of 81.3%, 31.3%, and 6.7%, respectively (P &lt; .001). In conclusion, a subset of MN harboring TP53mut who have low blasts pre-alloHCT and received melphalan-based conditioning derived long-term benefit from alloHCT.

766 Background: BASECAMP-1 (NCT04981119) is a pre-screening study to identify patients with tumor-associated human leukocyte antigen (HLA)-A*02 loss of heterozygosity (LOH) for interventional studies, such as EVEREST-2 (NCT06051695), a phase 1/2 study of logic-gated chimeric antigen receptor T-cell (CAR T) therapy for MSLN-expressing cancers. MSLN is a cell surface protein expressed in several cancer types, including mesothelioma (MESO), colorectal (CRC), non-small cell lung (NSCLC), ovarian (OVCA), and pancreatic (PANC) cancer, which can be associated with poor prognosis (1). Longitudinal clinical, genomic, and biomarker data were collected from patients enrolled in BASECAMP-1, providing a large dataset for translational discovery and propensity scoring. The aim of this analysis was to determine whether MSLN expression measured by RNAseq and IHC are correlated among patients with solid tumors. Methods: In BASECAMP-1, tumor tissue from patients with germline heterozygous HLA-A*02 is tested for HLA-A*02 LOH using an investigational next generation sequencing device codeveloped with Tempus Labs (Lozac'hmeur, et al. NPJ Precis Oncol. 2024) that detects somatic alterations, including HLA LOH, and generates RNAseq data (eg, MSLN expression). Gene-level transcripts per million read (TPM) values were determined and the log2 TPM +1 was displayed in a scatter plot (the mean and standard deviation [SD] are provided in the Table). Patients with HLA-A*02 LOH also submit tissue for IHC testing for exploratory biomarkers (eg, MSLN expression using anti-MSLN clone 5B2). Statistical significance was determined with ANOVA or t-tests. Results: As of June 1, 2024, 314 patients had been screened for BASECAMP-1 and had RNAseq results; 34 patients also had MSLN IHC results. MSLN expression by RNAseq was consistently higher in patients with PANC or OVCA vs CRC or NSCLC ( P <0.001). MSLN expression by RNAseq and IHC were correlated overall ( P <0.001), particularly among patients with PANC and OVCA. Among the 8 patients with OVCA or PANC who were IHC+ for MSLN, only 1 had an RNAseq value below 6 log2 TPM+1; this patient had mesonephric-like ovarian adenocarcinoma, which may account for the low value. The one MESO sample with both RNAseq and IHC available was IHC+ and had a high RNAseq value. Conclusions: This analysis demonstrated high correlation between RNAseq and IHC MSLN expression in tumor tissue. 1. Faust, et al. Cancers. 2022. Clinical trial information: NCT04981119 . MSLN expression by RNASeq by tumor type and IHC status. Tumor Overall (N=314) IHC data available (n=34) MSLN Positive MSLN Negative n Mean TPM a (SD) n Mean TPM a (SD) n Mean TPM a (SD) All 314 5.1 (2.3) 20 6.0 (2.1) 14 2.3 (1.7) adNSCLC 33 3.8 (2.4) 2 4.7 (1.7) 7 1.5 (1.3) sqNSCLC 4 3.1 (1.8) 0 0 MESO 4 4.5 (4.2) 1 5.7 0 CRC 148 4.4 (2.0) 8 5.1 (2.1) 7 3.1 (1.6) PANC 100 6.1 (1.8) 3 7.8 (0.6) 0 OVCA 25 7.2 (2.1) 5 6.6 (2.8) 0 a MSLN log2 transcripts per million+1. ad, adenocarcinoma; sq, squamous.

ABSTRACT Patients with newly diagnosed acute myeloid leukemia (ND‐AML) derive variable survival benefit from venetoclax + hypomethylating agent (Ven‐HMA) therapy. The primary objective in the current study was to develop genetic risk models that are predictive of survival and are applicable at the time of diagnosis and after establishing treatment response. Among 400 ND‐AML patients treated with Ven‐HMA at the Mayo Clinic, 247 (62%) achieved complete remission with (CR) or without (CRi) count recovery. Multivariable analysis–derived hazard ratios (HR), including 1.8 for European LeukemiaNet (ELN) adverse karyotype, 4.7 for KMT2Ar , 1.7 for TP53 MUT , 2.6 for KRAS MUT , and 2.1 for IDH2 WT were applied to develop an HR‐weighted risk model: low, intermediate, and high; respective median survival censored for allogeneic stem cell transplant (ASCT) (3‐year survival) were “not reached” (67%), 19.1 (33%), and 7.1 months (0%). In patients achieving CR/CRi, adverse karyotype, KMT2Ar , KRAS MUT , IDH2 WT predicted inferior survival, allowing for a complementary response‐stratified risk model. The model was externally validated and was shown to be superior to the ELN 2024 risk model (AIC 179 vs. 195 and AUC 0.77 vs. 0.69). Survival was inferior with failure to achieve CR/CRi or not receiving ASCT; 3‐year survival for high‐risk with or without ASCT was 42% versus 0% ( p < 0.01); intermediate 72% versus 43% ( p = 0.06); and low‐risk 88% versus 78% ( p = 0.53). The Mayo genetic risk models offer pre‐treatment and response‐based prognostic tools for ND‐AML treated with Ven‐HMA. The current study underscores the prognostically indispensable role of achieving CR/CRi and ASCT.

Background: Therapy-related myeloid neoplasms (t-MN) occur after exposure to DNA-damaging agentsincluding cytotoxic chemo and radiation therapy and are associated with dismal outcomes. TP53 and PPM1D (protein phosphatase Mn2+/Mg2+-dependent 1D) mutations (MT) disrupt the DNA damage response and repair (DDR) pathway, are often co-mutated (25-35%), and although enriched in t-MN (20%), can also be seen in age-related clonal hematopoiesis (CH) (5-10%). While progressive genomic instability with biallelic inactivation has been described as a mechanism of MN progression with TP53 MT, similar mechanisms remain to be elucidated for PPM1D MT. We conducted this study to assess the clinical phenotype and outcomes of TP53 and PPM1D co-mutated (PPM1DTP53m) versus PPM1D MT and wild type TP53 (PPM1DTP53wt) CH and MN. Methods: After approval from the Mayo Clinic IRB, the CH and MN databases were queried and 119 pts with PPM1D MT (CH= 26, clonal cytopenia of unknown significance [CCUS]= 64, myelodysplastic syndrome [MDS]=16, myeloproliferative neoplasm [MPN]=8, acute myeloid leukemia [AML]= 5) were identified. We analyzed baseline characteristics, karyotypic changes, somatic mutations, and survival outcomes between the PPM1DTP53m (n= 22) and PPM1DTP53wt (n= 97) groups (gp). Results: Baseline characteristics The median age in years (yrs) was 68.5 (range [R], 43-89) and 71 (R,30-92) in the PPM1DTP53m and PPM1DTP53wt gp (p= 0.65). 22% vs 18% (p=0.78), 58% vs 32% (p=0.03), 9% vs 32% (p=0.01), 7% vs 4.5% (p=&gt;0.99) and 3% vs 14% (p=0.07) pts met criteria for CH, CCUS, MDS, MPN and AML in the PPM1DTP53wt and PPM1DTP53m gp, respectively. Similarly, 19% vs 18% (p=&gt;0.99), 47% vs 32% (p=0.23), 7% vs 36% (p=0.001), 2% vs 0% (p=&gt;0.99) and 2% vs 14% (p=0.04) of pts met criteria for t-CH, t-CCUS, t-MDS, t-MPN and t-AML in the PPM1DTP53wt and PPM1DTP53m gp, respectively. All PPM1D variants were truncating, located in exon 6 of the gene, with median variant allele frequency (VAF) (8% [R,1-64] vs 5.5% [R,2-24], p= 0.10) and number of PPM1D MT (1 [R,1-5] vs 1 [R, 1-3], p=&gt;0.99) being similar between the two gp. While proportion of pts with t-MN were similar between the two gps (77% vs 86%, p= 0.40), cytogenetic (CG) abnormalities (67% vs 31%, p= 0.006), including monosomies (28% vs 2%, p= &lt;0.001) and complex CG (53% vs 4%, p= &lt;0.001) were more frequent in PPM1DTP53m gp compared to PPM1DTP53wt gp, respectively. The median TP53 MT VAF was 14% (R, 1-14); 50% had multi-hit TP53 MT. Somatic mutations in PPM1D MT myeloid neoplasm with and without TP53 MT Somatic co-mutations excluding TP53 were seen in 71% of pts; 73.5% in PPM1DTP53wt and 59% in PPM1DTP53m gp (p= 0.20). Apart from CBL (0% vs 9%,p=0.03), MT involving DNMT3A (20% vs 14%, p= 0.56), TET2 (15% vs. 14%, p= &gt;0.99), IDH1 (6% vs 19%, p= 0.07), splicing factors (8% vs 14%, p=0.08), ASXL1 (11% vs 4.5%, p=0.46), and JAK2 (9% vs 0%,p=0.20) were comparable between PPM1DTP53wt and PPM1DTP53m gps, respectively. The proportion of pts with ≥ 2 somatic co-mutations excluding TP53 MT were numerically higher in PPM1DTP53wt gp (43%) compared to PPM1DTP53m gp (23%) (p= 0.09). In the PPM1DTP53wt gp, other somatic driver mutations included RUNX1 (n= 7 [7%]; CH [n=1], CCUS [n=2], MDS [n=3], MPN [n=1]), NRAS (n=1 [1%]; CCUS), BCOR (n=1 [1%]; CCUS), and PTPN11 (n=1 [1%]; CCUS). Leukemic transformation and survival With a median follow up of 6.43 months (mo) (R, 0.23-36.4), median overall survival (mOS) in mo among high-risk MDS/AML (16.4 vs 15.0, p= 0.66) and CHIP/CCUS pts (35.79 vs 26.28, p= 0.93) was similar between PPM1DTP53wt and PPM1DTP53m gps, respectively. CHIP/CCUS pts harboring multiple vs single PPM1D MT had an inferior mOS (23.9 vs 36.7 mo, p= 0.05). Three (2.5%) MDS pts underwent AML transformation; 1 (1%) and 2 (10%) in PPM1DTP53wt and PPM1DTP53m gp, respectively (p= 0.07). Conclusions In our study, 18% of pts had concurrent TP53 and PPM1D MT (82% t-MN), with the PPM1DTP53wt genotype being more frequent in CCUS, while the PPM1DTP53m genotype was more frequent in MN. The PPM1DTP53m genotype had a higher frequency of CG abnormalities, including complex CG. Survival outcome was inferior and comparable in PPM1D MT MN with or without TP53 MT. While 73.5 % of the PPM1DTP53wt gp had somatic driver MT and 31% had karyotypic abnormalities, 22% had no additional MT or CG changes (all CH/CCUS), underscoring the need to understand how PPM1D MT contribute to hematopoietic dysfunction and MN.

Background Venetoclax + hypomethylating agent (Ven-HMA) therapy results in favorable response rates (CR/CRi 66%) and overall survival (OS; median 14.7 months) in unfit patients with newly-diagnosed acute myeloid leukemia (ND-AML; N Engl J Med 2020;383). However, there is substantial heterogeneity in individual patient outcomes based on response to treatment. The objective of the current study was to develop a genetic risk model to predict survival outcome in treatment responders and non-responders, analyzed separately, and also account for allogeneic stem cell transplant (ASCT). Methods Treatment-naïve AML patients receiving Ven-HMA, outside of a clinical trial, at the Mayo Clinic, USA (MN, AZ, FL) were retrospectively recruited. Molecular studies (48-gene panel) were performed by next-generation sequencing. Response was assessed according to the 2022 European LeukemiaNet (ELN) criteria (Blood 2022;140). Standard statistical analyses were performed using JMP Pro (Version 17.0.0). Results 400 ND-AML patients (median age 73 years, 64% male, 60% de novo) received Ven-HMA. ELN cytogenetic risk included favorable 2%, intermediate 60%, or adverse 38%. Mutations involved TP53 in 26%, TET2 19%, RUNX1 19%, SRSF2 18%, ASXL1 18%, DNMT3A 15%, K/NRAS 13%, NPM1 12%, IDH2 12%, FLT3-ITD 10%, IDH1 7% and DDX41 in 4% of informative cases. 153 (38%) patients achieved CR, 94 (24%) CR with incomplete count recovery (CRi), resulting in CR/CRi in 247 (62%) patients. CR/CRi rates were higher with NPM1 (86% vs 59%), IDH2 (77% vs 60%), and DDX41 mutations (93% vs 61%) and lower with TP53 (45% vs 67%), FLT3-ITD (41% vs 64%), RUNX1 (49% vs 65%) mutations and adverse karyotype (48% vs 71%), with all retaining significance in multivariable analysis. CR/CRi rate was highest (87%) in 75 patients with one or more favorable mutations (NPM1, IDH2, DDX41) and no unfavorable mutation (TP53, FLT3-ITD, RUNX1). CR/CRi rate was lowest in 171 patients with at least one unfavorable mutation and no favorable mutation (44%; p&lt;0.01). After a median follow up of 10.5 months (0.5-66), 105 relapses (43% of patients in CR/CRi), 248 (62%) deaths and 65 (16%) ASCT were documented. Univariate analysis for transplant-censored OS disclosed superior survival in patients achieving CR/CRi (median 21 vs 3.6 months; HR 5.4, 95% CI 4.1-7.1). Among patients achieving CR/CRi, IDH2 mutation was associated with superior (median not reached vs 20.5 months) and male gender (median 17.6 vs 29.2 months), TP53 mutations (median 11.9 vs 24.3 months) and adverse karyotype (median 12.6 vs 29.4 months) with inferior survival; multivariable analysis resulted in HRs (95% CI) of 2.1 (1.3-3.4) for male gender, 2.1 (1.2-3.8) for adverse karyotype, and 2.5 (1.1-5.7) for absence of IDH2 mutations. Male gender did not show an association with karyotype, or TP53 mutations (p&gt;0.1). Accordingly, an HR-weighted genetic risk model was generated with assignment of one point each to male gender, adverse karyotype, and absence of IDH2 mutation: high risk = 3 risk factors (n=38; median 9.6 months; 3-year survival 0%); intermediate risk = 2 risk factors (n=129; median 22 months; 3-year survival 38%); and low-risk = ≤1 risk factor (n=78; median not reached; 3-year survival 69%; p&lt;0.01). In patients bridged to ASCT (n=60), post-transplant survival was inferior in the presence of ≥ 2 risk factors (n=42; median 46 months) vs. ≤1 risk factor (n=18; median not reached; p=0.1). In patients not achieving CR/CRi, risk factors for OS included male gender (median 3.4 vs 6.1 months), adverse karyotype (median 3.3 vs 4.5 months) and TP53 mutation (median 3.1 vs 4.5 months) while those with IDH2 mutations lived longer (median 8.4 vs 3.4 months). In multivariable analysis, HRs (95% CI) were 1.9 (1.3-2.9) for male gender, 2.0 (1.4-2.9) for adverse karyotype, and 2.1 (1.1-4.3) for absence of IDH2 mutations. OS was shortest for non-responders harboring 3 risk factors (n=45; median 3 months; 1-year survival 0%) and longer for those harboring two (n=80; median 6 months, 1-year OS 15%) or ≤ 1 risk factor (n=25; median 14 months; 1-year OS 56%; p&lt;0.01). Conclusion The current study describes a molecular profile for predicting response to Ven-HMA in ND-AML and also introduces a novel approach to survival prediction that considers treatment response as the starting point. In both responders and non-responders, male gender and adverse karyotype were identified as unfavorable and IDH2 mutation favorable risk factor for OS.

Background Venetoclax + hypomethylating agent (Ven-HMA) therapy has demonstrated improved outcomes in elderly/unfit patients with newly-diagnosed acute myeloid leukemia (ND-AML) (N Engl J Med 2020;383). However, ~50% of patients who respond to Ven-HMA, eventually relapse (Am J Hematol 2024;99). In the current study we examined the prognostic contribution of gene mutations, karyotype, measurable residual disease (MRD) status and clinical risk factors for relapse in patients with ND-AML treated with Ven-HMA therapy. Methods Our study population of ND-AML was retrospectively recruited from Mayo Clinic (MN, AZ, FL), after institutional review board approval and based on documentation of complete remission with/without count recovery (CR/CRi) following Ven-HMA therapy. Relapse-free survival was determined from the time of remission to relapse or last follow-up. Survival analysis was censored for transplant. The 2022 European LeukemiaNet (ELN) criteria was used to assign cytogenetic risk and classify treatment response (Blood 2022;140). Standard statistical analyses were performed using JMP Pro (Version 17.0.0). Results 247 ND-AML patients treated with Ven-HMA (median age 73 years, 62% male, 64% de novo) who achieved CR (n=153), or CRi (n=94) were included. MRD by multiparameter flow cytometry (sensitivity 0.01%) performed at the time of CR/CRi, was negative in 117 (70%) of 166 informative cases. ELN cytogenetic risk categories at diagnosis included favorable 2%, intermediate 69%, or adverse 29%. Mutations involved TET2 21%, SRSF2 21%, TP53 19%, ASXL1 18%, DNMT3A 18%, NPM1 17%, IDH2 15%, RUNX1 15%, K/NRAS 12%, IDH1 8%, FLT3-ITD 7%, and DDX41 in 6% of informative cases. Relapse was documented in a total of 99 (40%) patients who achieved CR (39%) or CRi (41%) after a median remission duration of 6 months (1-31). Patients that relapsed within the first-year vs those that did not, were more likely to be males (75% vs 55%), MRD positive (44% vs 24%), and harbor TP53 mutations (23% vs 16%). Univariate analysis for RFS identified male gender (12 vs 29 months; p=0.02), secondary AML (10 vs 21 months; p&lt;0.01), adverse karyotype (9 vs 22 months; p&lt;0.01), TP53 mutations (6 vs 18 months, p&lt;0.01) and MRD positive remission (13 months vs not reached; p&lt;0.01) as risk factors for relapse; on the other hand, IDH2 mutations (not reached vs 15 months; p&lt;0.01) were associated with a lower risk of relapse. Multivariable analysis confirmed male gender (HR 2.2), TP53 mutations (HR 3.0) and MRD positive remission (HR 1.7) as unfavorable predictors of relapse. Subsequently, in a three-tiered relapse prediction model, RFS was significantly inferior in the presence of three risk factors (male gender, TP53 mutations, MRD positive remission) (n=10, median 4 months), vs two (n=34, median 10 months) vs ≤ 1 risk factor (n=122, median 31 months) (p&lt;0.01); 1-year cumulative incidence of relapse was 100%, 56%, and 26% in the presence of three, two and ≤ 1 risk factors, respectively. 50 of 99 (65%) patients received salvage therapy which included cladribine-cytarabine-Ven (n=19), intensive induction chemotherapy (n=9), FLT3 inhibitors (n=5), Ven-HMA-FLT3 inhibitor (n=1), IDH1/2 inhibitors (n=5), glasdegib-cytarabine (n=4), gemtuzumab (n=3), lenalidomide (n=1), or investigational therapies (n=3); in addition, Ven-HMA was resumed in 14 patients in whom treatment had been discontinued for a median of 5 months (2-12). Overall, 20 (31%) of patients achieved second CR/CRi and response rates were similar with intensive vs less intensive therapy (CR/CRi; 44% vs 29%; p=0.37). Median follow up after relapse was 3.4 months (1-66), during which 73 (74%) deaths, and 4 (5%) ASCT were recorded. Median transplant-censored survival was 4.2 months (1-year; 22%), and inferior in patients with adverse karyotype (median 4.6 vs 3.4 months; p=0.02), and TP53 mutations (3.6 vs 4.4; p=0.11); on the other hand, male gender (3.5 vs 6.8 months; p=0.51), MRD positive remission (4.6 vs 4.2; p=0.56), and IDH2 mutations (6.7 vs 4.2; p=0.28), did not influence post-relapse survival. Conclusions The current study identifies male gender, TP53 mutations, and MRD status as powerful predictors of relapse and proposes a relapse prediction model for ND-AML patients receiving Ven-HMA therapy. Furthermore, relapse following Ven-HMA was associated with uniformly poor outcomes, regardless of salvage treatment strategy.