Abstract
Ontogeny of acute myeloid leukemia (AML) provides prognostic information, however closer interrogation with respect to AML characteristics, genomics, and various treatments are warranted. We defined untreated clinical secondary (CS) AML as AML with a diagnosis of antecedent myelodysplastic syndrome (MDS) or MDS/myeloproliferative neoplasm (MDS-MPN) without exposure to hypomethylating agents or chemotherapy; genomic secondary (GS) AML included patients with myelodysplasia related cytogenetics (MRC) or myelodysplasia related mutations (MRM) without a known antecedent myeloid neoplasm or prior chemo-radiotherapy. Amongst newly diagnosed AML patients classified as untreated CS-AML (n=133) or GS-AML (n=389), median relapse-free survival (RFS) (11.9 versus 12.4 months, p=0.36) and overall survival (OS) (11.6 vs. 14.4 months, p=0.75) were similar. No difference in RFS and OS between these groups treated with low-intensity therapy (LIT) and venetoclax regimens was seen, but both were inferior to de novo (DN) AML without secondary-type genomics (pure DN-AML). GS-AML defined by the presence of only MRM had superior OS compared with MRM ± MRC with LIT+ venetoclax therapy (RFS 19.5 vs. 6.8 months [p<0.01] and OS 29.6 vs. 8.4 [p<0.01]) and had similar RFS (29.8 months, p=0.48) and OS (32.0 months, p=0.48) to pure DN-AML treated with LIT+ venetoclax. On multivariate analysis in patients treated with LIT+ venetoclax, untreated CS-AML (vs. GS-AML), adverse cytogenetics and ELN 2024 adverse-risk disease (mutated TP53) were associated with higher hazard of death. Adverse cytogenetics was the strongest prognostic variable predicting survival. Mutation-driven genomic ontogeny of newly diagnosed AML with MRM appears less prognostic than cytogenetic-driven ontogeny with venetoclax-based therapy.
Keywords: Secondary AML, Myelodysplasia related genomics, Cytogenetics, Venetoclax, Stem cell transplantation
Introduction
Significant strides have been made in understanding the ontogeny and genetic aberrations in acute myeloid leukemia (AML). These findings have led to the development of several genomically inspired targeted drugs that are improving clinical outcomes for patients with AML1–5. Cytogenetic and mutational profiling has not only enhanced prognostication for patients with newly diagnosed AML but has also enabled the use of specific molecular markers for depth of response assessment and treatment modification6–11. Additionally, genomic interrogation of AML can reveal distinct biological subgroups that represent different clinical phenotypes, history of antecedent myeloid neoplasms, or exposure to prior cytotoxic therapy for non-myeloid malignancies1,12,13. The relevance of these findings to contemporary treatment regimens remains unclear. Although some of these findings add clinical relevance, the true prognostic significance of these genomic markers in relation to other clinical and pathologic characteristics needs more granular assessment, especially in the setting of novel therapeutics.
Secondary AML includes a subgroup of AML clinically defined in patients with known antecedent myeloid neoplasms like myelodysplastic syndrome (MDS), chronic myelomonocytic leukemia (CMML), or MDS/myeloproliferative neoplasms (MPN), with or without exposure to prior therapy for these disorders. Beyond this clinical diagnosis, AML in patients without a known antecedent myeloid neoplasm with specific cytogenetic changes and/or mutations is also considered as myelodysplasia-related AML-by the World Health Organization (WHO) classification of hematolymphoid tumors and the International Consensus Classification (ICC)14,15. Except for minor differences (i.e. trisomy 8, del20(q) and RUNX1 mutation), the defining cytogenetic aberrations and mutations for definition of genomically secondary AML are similar between these two classification systems.8,16. Most patients with AML harboring myelodysplasia-related cytogenetic changes or mutations are older and receive low-intensity therapy (LIT), usually with venetoclax2,17. The prognostic relevance of these myelodysplasia-related genomic aberrations when receiving venetoclax-based therapy is not well understood, particularly when they co-occur with favorable genomics such as NPM1 mutations.
Myelodysplasia-related mutations are now considered as adverse-risk by the European LeukemiaNet (ELN) 2022 recommendations, if not accompanied by favorable defining genomics9. However, we recently showed that the addition of venetoclax to LIT or intensive chemotherapy (IC) mitigated the negative prognostic impact of splicing factor gene mutations in patients with newly diagnosed AML18. Other studies demonstrate a variable impact of concurrent myelodysplasia-related mutations in patients with NPM1 mutated AML but were not analyzed with respect to venetoclax therapy19,20. In this study, we interrogate the impact of untreated secondary AML, defined by clinical history or myelodysplasia related cytogenetics (MRC) and/or myelodysplasia related mutations (MRM), with attention to treatment response and survival outcomes in patients with newly diagnosed AML receiving contemporary AML therapy, with and without venetoclax.
Methods:
Patients, study definitions and treatment
Adult patients (≥18 years) with newly diagnosed AML treated at the University of Texas MD Anderson Cancer Center in Houston, TX were retrospectively analyzed. Patients with core binding factor (CBF) AML, blast-phase MPN, and myeloid blast phase chronic myeloid leukemia were excluded. Clinically defined secondary AML (CS-AML) included patients with a known prior diagnosis of a myeloid neoplasm (MDS, CMML, or MDS/MPN), regardless of genomics (Supplemental [S] Figure 1 and Table S1]. They were considered untreated CS-AML if they did not receive any prior hypomethylating agent (HMA), chemotherapy, or allogeneic hematopoietic stem cell transplantation (HSCT). Treated secondary AML (TS-AML) was a specific subgroup of patients who had received prior HMA, chemotherapy, and/or HSCT. Genomically secondary AML (GS-AML) was defined as AML harboring MRC or MRM in patients without an established diagnosis of an antecedent myeloid neoplasm and no prior exposure to cytotoxic therapy (chemotherapy or radiotherapy) for non-myeloid neoplasms14. Patients with CS-AML who had such myelodysplasia related genomics (MRC and/or MRM) were classified as CS-AML and NOT as GS-AML. Additionally, patients with untreated CS-AML who had received prior cytotoxic therapy for non-myeloid cancers were retained in the untreated CS-AML category. AML was defined as de novo (DN) in patients without an antecedent myeloid disorder or history of prior chemo/radiotherapy for non-myeloid malignancies; pure DN-AML included patients with DN-AML no myelodysplasia related genomics (MRM and/or MRC).
Patient demographics, history of antecedent myeloid neoplasm and treatment, clinical characteristics, and treatment for AML were obtained from the clinical records. Treatment for AML was categorized as intensive if patients received frontline cytarabine ≥1 g/m2 with an anthracycline; LIT included hypomethylating agent or low dose cytarabine based therapy, or rarely single agent drugs on clinical trials. The use of venetoclax with IC or LIT was documented. HSCT in first remission was included for outcomes analysis.
Genomic assessment
Cytogenetic data was based on a combination of conventional karyotyping and fluorescent in situ hybridization (FISH). Next generation sequencing using an 81 gene panel targeting hot-spot or entire exonic regions of genes recurrently mutated in myeloid neoplasms, was performed at initial AML diagnosis for all patients, as described previously decsribed14. Myelodysplasia-related genetics were defined as follows, as per the WHO: MRCs included complex karyotype (≥3 abnormalities including a structural abnormality), del(5q)/−5, del(7q)/−7, del(11q), del(12p)del(13q)/−13, del(17p)/−17, isochromosome 17q, idic(X)(q13) while MRMs included mutations in ASXL1, BCOR, EZH2, SF3B1, SRSF2, STAG2, U2AF1, ZRSR214. Cytogenetics were risk stratified as per ELN 2017 criteria. Given the inclusion of several of the MRMs as adverse risk in ELN 2022, overall AML risk stratification was also done as per ELN 2017 criteria8,9,18,21. Patients receiving LIT were additionally stratified by the ELN 2024 risk stratification for LIT21,22.
Response and survival outcomes
Response to frontline therapy was annotated as per ELN 2017 recommendations8. Composite complete response (CRc) included a combination of CR and CR with incomplete count recovery (CRi) while overall response rate (ORR) also included morphological relapse free state (MLFS). Measurable residual disease (MRD) by flow cytometry at a level of detection of 1 in 104 was assessed from first remission bone marrow (BM) and best MRD response was annotated. Relapse free survival (RFS) was calculated in responders (patients with at least an overall response) from the time of best response to relapse or death. Overall survival (OS) included time of therapy initiation to death and censored at last follow-up. RFS and OS were not censored for HSCT.
Statistical analysis
Mann-Whitney U test and a 2-sided Fisher’s exact test were used to compare continuous and categorical baseline variables. Follow up was estimated using the inverse Kaplan Meier method and RFS/OS were calculated using Kaplan Meier and compared using the Mantel Cox log-rank test. A Cox proportional hazard model with HSCT as time dependent covariate was used to study factors associated with survival. Factors with p<0.05 on univariate analysis (UVA) was used on multivariate analysis (MVA) and model selection was done using backward selection. Classification and regression tree (CRT) was done using a binomial recursive portioning with a training and internal validation set. Analysis was done using GraphPad Prism version 9.3.1 (GraphPad Software, Boston, MA), Lumivero (New York, NY) (2023) XLSTAT statistical solution, and RStudio version 2024.9.0.375 (Boston, MA).
Results
From April 2017- December 2023, 1380 patients with newly diagnosed AML, excluding CBF and blast phase MPN/CML, were treated at our institution. The median age was 68.2 years (range, 18.8–94.8 years). 989 patients (71.7%) were ≥60 years of age, 617 patients (44.7%) were female and 1089/1362 (80.0%) patients with self-declared ethnicity were white. 510 (36.9%) patients had a history of antecedent myeloid neoplasms (CS-AML). Overall, 685 patients (49.6%) had de novo AML (Figure 1). Among 642/685 classifiable DN-AML patients (based on available complete cytogenetic and mutational data) 389 (60.6%) had myelodysplasia-related genetic alterations (MRM [64.5%] and MRC [51.7%]) and were classified as GS-AML (Table S1); the remaining 253 patients were classified as pure DN-AML.
Figure 1. Distribution of patients with newly diagnosed acute myeloid leukemia (N=1380).
Abbreviations and definitions: CS-AML, clinical secondary AML (known antecedent myeloid disorder); US-AML, untreated clinical secondary AML (no previous chemotherapy, or hypomethylating agent, or HSCT for antecedent myeloid disorder); TS-AML, treated clinical secondary AML (exposed to previous chemotherapy, and/or hypomethylating agent, and/or HSCT for antecedent myeloid disorder)
*not classifiable: these patients did not have adequate cytogenetic data for classification into pure de novo AML or genomic secondary AML
A total of 302/1380 (21.9%) patients had received prior cytotoxic therapy for non-myeloid cancers, 117/302 classified in the CS-AML category and the rest 185/302 as therapy-related AML; the latter were not included for further analysis.
Secondary AML and Pure De novo AML: Baseline characteristics
Among 510 patients with CS-AML (median age 70.8 years, range, 20.7–90.9 years), 377 patients (73.9%) patients had received therapy (HMA/chemotherapy or HSCT) for antecedent myeloid disorders and referred to as treated CS-AML (TS-AML) while the remaining 133 (26.1%) were untreated CS-AML (Table 1, Figure 1). A significant proportion of patients with untreated CS-AML (85%, p <0.001) and GS-AML (70.4%, p<0.001) were ≥60 years at AML diagnosis compared to patients with pure DN-AML (49.4%).
Table 1:
Baseline characteristics of patients with untreated secondary AML (untreated clinical secondary and genomic secondary AML)
| Characteristics | Untreated Clinical Secondary | Genomic Secondary | Pure De novo AML | |
|---|---|---|---|---|
| n | 133 | 389 | 253 | |
| Age (years) | 70 [28.9–70.5] | 67.9 [18.8–94.8] | 59.9 [41.7–88.1] | |
| Age ≥60 | 113 (85.0) | 274 (70.4) | 125 (49.4) | |
| Age ≥70 | 67 (50.4) | 167 (42.9) | 65 (25.7) | |
| Gender | Female | 46 | 153 | 140 |
| Race | White | 110 | 300 | 194 |
| Ethnicity | Hispanic | 11/129 | 48/378 | 46/244 |
| Prior therapy for non-myeloid cancer | 31 | 0 | 0 | |
| Antecedent myeloid neoplasm | MDS | 102 | 0 | 0 |
| MDS-MPN/CMML | 31 | 0 | 0 | |
| Baseline CBC | Hb (g/dl) | 8.7 [3.5–12.7] | 8.6 [6.6–14.3] | 8.9 [3.9–14.3] |
| WBC (x109/L) | 3.6 [0.2–50.9] | 2.8 [0.1–336] | 4.3 [0.5–144] | |
| Platelets (x109/L) | 45 [6–689] | 36 [1–574] | 37 [4–1306] | |
| ECOG PS | 0–1 | 98 (74) | 266 (68) | 188 (74) |
| 2–3 | 35 (26) | 120 (31) | 65 (26) | |
| 4 | 0 | 3 (1) | 0 | |
| Cytogenetics | Diploid | 42/125 (34) | 111/375 (30) | 174 (69) |
| Adverse | 59/125 (47) | 201/375 (54) | 17 (7) | |
| • Complex | 44/125 (35) | 153/375 (41) | 0 | |
| Mutations | ASXL1 | 29 (22) | 83 (21) | 0 |
| BCOR | 10 (8) | 29 (7) | 0 | |
| DNMT3A | 28 (21) | 74 (19) | 87 (34) | |
| EZH2 | 2 (1) | 22 (6) | 0 | |
| FLT3 TKD | 3 (2) | 21 (5) | 36 (14) | |
| FLT3 ITD | 9 (7) | 43 (11) | 72 (28) | |
| IDH1 | 6 (5) | 33 (8) | 22 (9) | |
| IDH2 | 8 (6) | 65 (17) | 47 (19) | |
| NPM1 | 16 (12) | 41 (10) | 119 (47) | |
| PTPN11 | 10 (8) | 30 (8) | 29 (11) | |
| RAS | 31 (23) | 75 (19) | 58 (23) | |
| RUNX1 | 27 (20) | 70 (18) | 6 (2) | |
| TET2 | 24 (18) | 67 (17) | 38 (15) | |
| TP53 | 34 (26) | 130 (33) | 6 (2) | |
| WT1 | 10 (8) | 17 (4) | 21 (8) | |
| Cohesin complex mut | 13 (10) | 48 (12) | 11 (4) | |
| Splicing gene mut | 51 (38) | 177 (46) | 0 | |
| Myelodysplasia related genomics | MRC | 55/125 | 194/375 | 0 |
| MRM | 68 | 251/389 | 0 | |
| Overall | 104/129 | 389 (100) | 0 | |
| ELN 2017 risk strata | Favorable | 12/132 (9) | 30/384 (8) | 114 (45) |
| Intermediate | 27/132 (20) | 63/384 (16) | 111 (44) | |
| Adverse | 93/132 (70) | 291/384 (76) | 28 (11) | |
Abbreviations: MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm; CMML, chronic myelomonocytic leukemia; Hb, hemoglobin, WBC, white blood cells; MRC, myelodysplasia related cytogenetics; MRM, myelodysplasia related mutations; ELN, European LeukemiaNet
Based on the ELN 2017 risk stratification scheme, compared to only 28 (11.1%) patients with pure DN-AML, 93 of 132 evaluable (70.4%) patients with untreated CS-AML (p<0.001) and 291 of 384 evaluable (75.8%) patients with GS-AML (p<0.001) had adverse risk disease. Adverse risk cytogenetics was identified in 47.2% of evaluable patients with untreated CS-AML and 53.6% of evaluable patients with GS-AML; only 6.7 patients with pure DN AML had adverse cytogenetics. TP53 mutation(s) were present in 34 (26%) patients with untreated CS-AML (p<0.001) and 130 (33%) patients with GS-AML (p<0.001), compared to 6 (2%) patients with pure DN-AML. (Table 1).
Clinical outcomes in Secondary-type AML and pure DN-AML
Based on the above categorization, a total of 899 patients had secondary-type AML (treated CS-AML =377, untreated CS-AML=133, GS-AML= 389). Stratified by treatment intensity, 214 (23.8%) patients received IC (121 with venetoclax [56.5%]) and 685 (76.2%) received LIT [470 (68.6%) with venetoclax). The ORR was 61.1%, and CRc rate 52.4%. At a median follow-up of 34.4 months (95% CI 29.5–38.8 months), the median RFS was 9.3 months (95% CI 7.8–10.0 months) and OS 8.8 months (95% CI 7.5–9.9 months) (Figure S1) with a 2-year RFS and OS of 30.7% and 25.5%, respectively. The median OS for responding secondary-type AML patients was 16.2 months, and 2-year OS was 38.8%. The median OS of patients with CS-AML was only 5.9 months, compared to 14.4 months for GS-AML (p<0.001) (Figure S2). Patients with treated CS-AML was a particularly high-risk subgroup, with an OS of 4.4 months compared to 11.6 months in patients with untreated CS-AML (p<0.001) (Figure S3). When comparing patients with untreated CS-AML to GS-AML, there was no difference in median RFS (11.9 vs. 12.4 months; p=0.36) or median OS (11.6 vs. 14.4 months, p=0.75) (Figure 2). The median RFS and OS for patients with pure DN-AML were 57.1 (95% CI 39.8-NR) and 49.7 months (95 % CI 32.0–66.2) respectively.
Figure 2:
Relapse Free Survival (RFS) (A) and Overall Survival (OS) of untreated patients with genomically secondary AML (GS-AML) compared to patients with a history of antecedent myeloid neoplasm (untreated CS-AML).
Comparative survival outcomes with venetoclax in untreated CS-AML, GS-AML and Pure DN-AML
Given the established inferior outcomes of patients with treated CS-AML in previously published work23 as well as in the present analysis, we focus here on the outcomes of patients with untreated CS-AML, GS-AML and pure DN-AML with respect to venetoclax-based therapy. Details of the different LIT regimens in these patents are in Table S2. The CRc rate and ORR of these groups based on treatment intensity and venetoclax exposure is summarized in Table 2. Venetoclax trended to significantly improve CRc rates in patients treated with either IC or LIT, across all three ontogeny groups. In patients with untreated CS-AML treated with intensive regimens, the CRc rate was 80% vs 100% (p=0.056) without and with venetoclax, respectively, while for LIT treated patients it was 44% vs. 64% (p=0.076). In patients with GS-AML treated intensively, the CRc rate was 65% and 81% (p=0.048), without and with venetoclax respectively, while in patients treated with LIT, the CRc rate was 49% vs. 74% (p<0.001).
Table 2:
Treatment intensity, venetoclax and response outcomes in untreated secondary AML (untreated clinical secondary and genomic secondary AML)
| Untreated Secondary (n=133) |
Genomic Secondary (n=389) |
Pure De novo AML (n=253) |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Intensive (n=23) |
Low-intensity (n=110) |
Intensive (n=112) |
Low-intensity (n=277) |
Intensive (n=148) |
Low-intensity (n=105) |
|||||||
| Ven+ | Ven- | Ven+ | Ven- | Ven+ | Ven- | Ven+ | Ven- | Ven+ | Ven- | Ven+ | Ven- | |
| n | 18 | 5 | 83 | 27 | 64 | 48 | 211 | 66 | 82 | 66 | 81 | 24 |
| ORR | 18 (100) | 4 (80) | 61 (73) | 14 (52) | 54 (84) | 32 (67) | 174 (82) | 37 (56) | 79 (96) | 59 (89) | 74 (91) | 15 (62) |
| CRc | 18 (100) | 4 (80) | 53 (64) | 12 (44) | 52 (81) | 31 (65) | 156 (74) | 32 (48) | 79 (96) | 59 (89) | 72 (89) | 14 (58) |
| MRD-ve * | 12 (67) | 3 (75) | 26/50 (52) | 4/10 (40) | 41/51 (80) | 21/30 (70) | 99/150 (66) | 10/29 (34) | 66/78 (85) | 48/56 (86) | 59/69 (85) | 4/13 (31) |
| HSCT | 13 (72) | 2 (40) | 23 (28) | 2 (7) | 30 (47) | 15 (31) | 49 (23) | 3 (4) | 43 (52) | 27 (41) | 19 (23) | 4 (17) |
MRD was calculated amongst patients who attained a CRc
Abbreviations: ORR, overall response rate; CRc, composite complete response; MRD -ve, measurable residual disease negative; HSCT, allogeneic hematopoietic stem cell transplantation
Among patients treated with LIT+ venetoclax, the median RFS was 9.7 months in the untreated CS-AML group, 11.1 months in GS-AML and 29.8 months in patients with pure DN-AML (p=0.03). Median OS was 9.6, 14.1 and 32 months for these groups, respectively (p=0.001) (Figures 3A and 3B). When comparing LIT+ venetoclax-treated patients with MRD-negative best response, the median RFS was 14.6, 18.2 and 48.8 months (2-year RFS 40% vs. 44,5% vs. 58.5%; p=0.62) and median OS was 28.2, 29.6 and 49.7 months (2- year OS 54.9% vs. 54.9% vs. 59.2%; p=0.85) for untreated CS-AML, GS-AML, and pure DN-AML respectively (Figure S4). In patients treated with IC (± venetoclax), the median RFS for the three respective groups were not reached (NR) vs. 58 months and 57.1 months (p=0.03) while median OS was NR vs. 21.1 months vs. 58 months, respectively (p<0.001) (Figures 3C and 3D). With IC+ venetoclax the median RFS was NR for all 3 groups (p=0.06) while median OS was NR for untreated CS-AML, 30.3 months for GS-AML and NR for the pure DN-AML group (p=0.007) (Figures 3E and 3F). RFS and OS in IC+ venetoclax treated patients who had an MRD negative best response is shown in Figure S5. Outcomes of patients with NPM1 mutated AML harboring MRMs are in the supplement (Figure S6).
Figure 3.
Clinical outcomes of patients with untreated CS-AML, GS-AML, and pure de novo (DN) AML stratified by treatment intensity. (A) RFS and (B) OS in patients receiving LIT + venetoclax. (C) RFS and (D) OS in patients receiving IC ± venetoclax. (E) RFS and (F) OS in patients receiving IC + venetoclax only.
ELN 2024 LIT risk stratification
Using the recently proposed ELN 2024 risk stratification in the LIT+ venetoclax treated patients24; in the favorable risk group, the median OS was 24.6 months vs. 19.4 months vs. 58 months for patients with untreated CS-AML, GS-AML and pure DN-AML respectively (p=0.003) (Figure 4A). For intermediate risk, the median OS was 20.2 vs. 21.4 vs 45.2 months (p=0.037) (Figure 4B), and for adverse risk the median OS was 5.8 vs 7.2 vs. 10.7 months (p=0.38) for the 3 groups respectively (Figure 4C). There was no difference in OS between patients with favorable and intermediate risk by the ELN 2024 with GS-AML or untreated CS-AML (p=0.30 and 0.41, respectively) (Figure S7). Overall, the outcomes of patients with pure DN-AML were superior to patients with secondary AML in each risk category.
Figure 4:
Overall survival stratified by the ELN 2024 risk classification for LIT in patients comparing patients with untreated CS-AML, GS-AML, and pure DN-AML treated with LIT + venetoclax. (A) Favorable risk by ELN 2024. (B) Intermediate risk by ELN 2024. (C). Adverse risk by ELN 2024.
Impact of myelodysplasia related genomics in untreated CS-AML
Among the 133 patients with untreated CS-AML, only 29 (21.8%) patients did not have any myelodysplasia related genomics (Figure S8 and S9). The median OS was 28.2 vs. 10.7 months for untreated CS-AML without myelodysplasia related genomics who received venetoclax-based combinations (LIT or IC) compared to those with myelodysplasia related genomics, respectively (p=0.03) (Figure S10A). In LIT + venetoclax treated patients, the median OS was 10.7 vs. 9.6 months (p=0.27) when stratified by the presence of myelodysplasia related genomics (Figure S10B). Outcomes with HSCT in the untreated CS-AML group is in the supplement (Figure S11).
Impact of Myelodysplasia Related Mutations and Myelodysplasia Related Cytogenetics on Survival Outcomes in GS-AML
We analyzed outcomes based on the type of myelodysplasia related genomics in patients with GS-AML treated with venetoclax based therapy. Patients with GS-AML (n=375) were stratified into MRC alone (n=138), MRM alone (n=181), or both (n=56). Fourteen patients had incomplete CTG data and were excluded (Table S3). Among patients who had MRC (n=194), 154 patients (79.4%) had a complex karyotype ± 17p.13/TP53 loss, and 123 patients (63.4%) had a TP53 mutation. Responses stratified by treatment intensity are in Figure S12. Patients treated with LIT+ venetoclax (n=201) experienced a median RFS of 6.5, 19.5 and 7.3 months in the MRC, MRM and MRC+MRM groups, respectively (p<0.001), with a median OS of 8.8, 29.6 and 7.8 months (p<0.001) (Figure 5A–B), highlighting an improved outcome in the MRM only cohort. Among patients treated with IC+ venetoclax (n=64), the median RFS was 10.5 months in the MRC group vs. NR in the MRM or MRC+MRM groups (p=0.18) while the median OS was 16.3 vs. NR vs. 21.1 months, respectively (p=0.006) (Figure 5C–D). Therefore, patients with MRM fared better than both MRC and MRC+MRM in patients treated with LIT+ venetoclax, whereas patients with MRM or MRC+MRM did better in those treated with IC+ venetoclax, though patient numbers were small in the second analysis. Focusing on those treated with LIT+ venetoclax, the 2-year RFS amongst patients with GS-AML defined by MRM and pure DN-AML were similar (45.9% vs. 53.4% p=0.48) as was the 2-year OS (57.3% vs. 54.2%, p=0.98) (Figure S13). The median RFS and OS of patients with GS-AML who were treated with LIT without venetoclax (n=66) were 5.3 and 8.6 months; among these patients the median RFS and OS for those who only had MRM (n=25) was 6.5 and 8.9 months respectively (Figure S14).
Figure 5. Clinical outcomes with myelodysplasia related cytogenetics (MRC) and myelodysplasia related mutations (MRM) in patients with genomic secondary AML (GS-AML) treated with venetoclax-based regimens.
Kaplan Meier curves were constructed and stratified on whether patients had MRC alone (blue), MRM alone (yellow), or both MRM and MRC (green). (A) RFS and (B) OS in patients receiving LIT + venetoclax. (C) RFS and (D) OS in patients receiving IC + venetoclax.
Patients with only MRM (n=181) were then stratified by splicing factor gene (SF) mutations (SRSF2, SF3B1, U2AF1, ZRSR2) (n=126) vs. other mutations (n=55) (Table S4 and Figure S15). The median OS of the patients treated with LIT + venetoclax was 29.6 months vs. 32.8 months with SF mutations (n=76) compared to those with non-SF MRMs (n=22) (p=0.84) (Figure S16A). Similarly, overall survival was not different between patients receiving IC + venetoclax who had (n=19) and did not have (n=17) SF mutations (p=0.39) (Figure S16B). Overall, 24/138 (17.4%) of patients with GS-AML and MRC alone underwent HSCT, compared to 58/181 patients with MRM alone (32%) (p=0.003) and 13/56 (23.2%) in patients with MRC+MRM (p=0.24). The median time to HSCT was 2.7 months [IQR 2.7–4.2 months]. In the GS-AML group, on a landmark comparative analysis of transplanted patients to responding patients <70 years who did not undergo HSCT, the median OS was NR vs. 17.5 months (2-year OS 64.4% vs. 41.6%).
Multivariate analysis (MVA) for survival in untreated CS-AML and GS-AML treated with LIT + venetoclax
Using a multivariate Cox analysis, we compared outcomes of patients with untreated CS-AML and GS AML treated with LIT+ venetoclax (n=294), adjusting for age (</≥60 years), ELN 2024 risk stratification, baseline cytogenetics, and HSCT. HSCT (HR=0.33, 95% CI=0.01–0.50) was independently favorable, while untreated CS-AML (vs. GS-AML; HR=1.5, 95% CI=1.1–2.1), adverse cytogenetics (vs. non-adverse cytogenetics; HR=2.02, 95% CI=1.39-NE), and ELN 2024 adverse risk strata (HR=1.90, 95% CI 1.33-NE) were associated with an increased hazard of death (Table S5, Figure S17). Using a classification and regression tree we aimed to predict survival at 2 years (including patients alive ≥2 years or died before 2 years) for patients with untreated CS-AML and GS-AML treated with LIT+ venetoclax (n=233; random learning set=153, validation set=50). The variables included age (<60/≥60 years), cytogenetic risk category (adverse or not), ELN 2024 risk stratification (adverse or not), and HSCT. The primary predictive factor was cytogenetic risk stratification with only 11.4% patients with adverse cytogenetics predicted to be alive at 2 years, compared to 48.7% patients without adverse risk cytogenetics (AUROC 0.82 and 0.80 for learning and validation sets respectively). Subsequent nodes were occupied by HSCT, AML type and ELN 2024 classification (Table S6, Figure S18).
Next, to evaluate the impact of specific mutations and cytogenetics on survival in untreated CS-AML and GS-AML, we used a multivariate Cox model with backward elimination for model selection in patients with untreated CS-AML and GS-AML stratified by treatment intensity. In the untreated CS-AML group treated with LIT+ venetoclax (n=83), TP53 mutation (HR=2.12, 95%CI 1.1–4.1) and adverse risk cytogenetics (HR=2.2, 95% CI 1.14-NR) increased hazards of death, while HSCT was favorable (HR=0.45, 95%CI=0.01–0.88) (Table 3A, S7, and Figure S19). There were not enough patients with untreated CS-AML treated with IC ± venetoclax to conduct an MVA. In patients with GS-AML treated with LIT + venetoclax (n=211), adverse cytogenetics (HR=1.60, 95% CI 1.01-NE), K/NRAS (HR=1.60, 95% CI 1.04–2.46) and TP53 mutations (HR=2.22, 95%CI 1.38–3.35) were hazardous, while HSCT was favorable (HR=0.38, 95% CI= 0.01–0.62) and NPM1 mutation trended to be favorable (HR=0.59, 95%CI=0.31–1.1) (Table 3B and S8). When adding ELN 2024 risk stratification as a variable, ELN 2024 adverse risk, adverse cytogenetics and ASXL1mut were independently unfavorable, while IDH2mut and HSCT were favorable (Table S9). Finally, amongst patients with GS-AML treated with IC + venetoclax (n=64), HSCT was favorable (HR=0.26, 95%CI=0.01–0.63), while TP53 mutations were unfavorable (HR=4.28, 95% CI=1.65–11.08) (Table S10).
Table 3A:
Multivariate Cox proportional hazard analysis for factors affecting hazards of death in patients with untreated CS-AML treated with LIT+ venetoclax (n=83)
| Variable* | Hazard ratio (HR) | HR, 95% C.I. | P Value |
|---|---|---|---|
| IDH1 mut | 0.30 | 0.04–2.24 | 0.24 |
| Splicing factor mutations | 1.49 | 0.81–2.73 | 0.20 |
| TP53 mut | 2.12 | 1.10–4.07 | 0.02 |
| Adverse cytogenetics (vs. non-adverse) | 2.22 | 1.14-NE | 0.02 |
| HSCT-Yes | 0.257 | 0.01–0.53 | <0.01 |
Variables assessed: NPM1, ASXL1, FLT3-ITD, IDH1, IDH2, RAS, RUNX1, TP53, Splicing factor gene mutations, cytogenetics (as per ELN 2017) and HSCT (Cox model details in Table S6)
Table 3B:
Multivariate Cox proportional hazard analysis for factors affecting hazards of death in patients with GS-AML treated with LIT+ venetoclax (n=211)
| Variable# | Hazard ratio (HR) | HR, 95% C.I. | P value |
|---|---|---|---|
| ASXL1 mut | 1.85 | 1.20–2.85 | <0.01 |
| IDH1 mut | 0.60 | 0.29–1.26 | 0.18 |
| IDH2 mut | 0.56 | 0.01–0.93 | 0.02 |
| RAS mut | 1.47 | 0.05–2.27 | 0.08 |
| TP53 mut | 2.49 | 1.54–4.04 | <0.01 |
| Adverse cytogenetics (vs. non-adverse) | 1.82 | 1.14-NE | 0.01 |
| HSCT-Yes | 0.40 | 0.01–0.65 | <0.01 |
Variables assessed: NPM1, ASXL1, FLT3-ITD, IDH1, IDH2, RAS, RUNX1, TP53, Splicing factor gene mutations, cytogenetics (as per ELN 2017) and HSCT (Cox model details in Table S7)
Discussion
We present one of the largest datasets of contemporarily treated patients with secondary AML and extensively evaluate important predictors of clinical outcomes including clinical vs genomic secondary AML, type of genomic aberration, treatment intensity, use of venetoclax, and HSCT. We show that overall patients with untreated CS-AML and GS-AML have similar survival outcomes, particularly when treated with LIT + venetoclax. However, we show that the group of GS-AML is heterogenous with patients having only MRM have survival very comparable to those with pure DN-AML, when treated with venetoclax based therapy, and unlike the inferior outcomes in patients harboring MRC. Furthermore, in MVA in the subgroups of untreated CS-AML and GS-AML, adverse cytogenetics have an independent prognostic impact along with the recently proposed ELN 2024 LIT risk stratification, which warrants further consideration in future prognostication categories.
Secondary AML is a common clinico-genomic subtype of AML and the recent inclusion of genomic entities (MRC and/or MRMs) for classifying otherwise untreated patients (clinical DN AML) as secondary AML is of clinical significance. Several of the mutations included under this category are venetoclax sensitizing mutations18,25–27, while MRC, especially complex karyotype, is associated with inferior response and outcomes with venetoclax based therapy28. This creates significant heterogeneity in the subgroup of GS-AML. In our analysis, patients with GS-AML based on MRC had a median OS of ~ 8 months with LIT + venetoclax, and ~ 20 months with IC + venetoclax. Approximately 80% of GS-AML patients with MRC had complex karyotype. This OS contrasts with the significantly superior median OS of 30 months and NR for patients with GS-AML having only MRM and treated with LIT+ venetoclax and IC+ venetoclax respectively. Therefore, grouping secondary AML with MRC and/or MRM together may lead to significant variability in treatment outcomes, especially with respect to venetoclax based therapy.
The standard ELN risk stratification includes cytogenetics as part of the prognostic risk assessment. The recently proposed ELN 2024 risk stratification for patients treated with LIT relies primarily on the mutations FLT3-ITD, K/NRAS and TP5324. In our analysis of patients with untreated CS and GS-AML, adverse cytogenetics was an independent prognostic factor regardless of whether patients were treated with IC or LIT, with and without venetoclax. Even when accounting for the ELN 2024 LIT risk stratification, adverse cytogenetics remained significant. Additionally on the CRT model including untreated CS-AML and GS-AML, adverse cytogenetics was the primary predictive factor for survival at 2 years amongst patients treated with LIT+ venetoclax.
Recent studies have tried to analyze outcomes of patients with AML and NPM1 mutations when they occur with MRM19,20. In our analysis we show comparable outcomes in patients with NPM1 mutations in the setting of either GS-AML or pure DN-AML, when receiving LIT+ venetoclax or IC+ venetoclax based therapy. Othman and colleagues recently showed that MRD responses in patients with AML are dictated by co-occurring mutations such as FLT3-ITD and DNMT3A11. In their analysis, MRM were not associated with lower odds of MRD negativity after 2 cycles of intensive therapy in NPM1 mutated AML but appeared to increase risk of relapse from an MRD negative state (HR=1.52, 95% CI 0.94–1.46). However, interrogation of MRM with respect to venetoclax based therapy was not a part of that analysis and therefore future large studies should evaluate this effect.
Finally, we demonstrate that the distinctions between the nomenclature of GS-AML defined by MRM and pure de novo AML may have limited clinical/prognostic relevance when patients are treated with venetoclax based therapy, while GS-AML defined by MRC may remain relevant.
The limitations of our study emanate from the retrospective nature of this analysis. However, >80% patients in all the subsets and overall were treated in frontline AML clinical trials with prospectively curated data with respect to baseline parameters, treatment regimens, response, and HSCT. The heterogeneity of treatment may limit wider extrapolation of the data; however, given this allowed us to stratify patients based on treatment intensity and use of venetoclax to increase applicability.
In summary, though clinical outcomes for patients with untreated CS-AML are overall similar to those with GS-AML, GS-AML is a heterogenous groups with survival outcomes contingent upon the present of MRM only or concurrently with MRC. Venetoclax based therapies improves outcomes in patients with GS-AML defined by MRM with outcomes similar to pure DN-AML, but outcomes in those with MRC remain inferior. Adverse cytogenetics remains an independent risk factor along with the ELN 2024 LIT risk stratification and warrants incorporation in future prognostic models.
Supplementary Material
Funding:
The study was supported by University of Texas MD Anderson Cancer Center Grant (CA016672) and University of Texas MD Anderson SPORE (C1100632).
Footnotes
Author COI:
The authors have no relevant conflict of interest
Data availability statement:
Data available from corresponding author on reasonable request
References:
- 1.Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic Classification and Prognosis in Acute Myeloid Leukemia. New England Journal of Medicine 2016;374(23):2209–2221. DOI: doi: 10.1056/NEJMoa1516192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.DiNardo CD, Wei AH. How I treat acute myeloid leukemia in the era of new drugs. Blood 2020;135(2):85–96. DOI: 10.1182/blood.2019001239. [DOI] [PubMed] [Google Scholar]
- 3.Kantarjian H, Kadia T, DiNardo C, et al. Acute myeloid leukemia: current progress and future directions. Blood Cancer Journal 2021;11(2):41. DOI: 10.1038/s41408-021-00425-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.DiNardo CD, Erba HP, Freeman SD, Wei AH. Acute myeloid leukaemia. Lancet 2023;401(10393):2073–2086. (In eng). DOI: 10.1016/s0140-6736(23)00108-3. [DOI] [PubMed] [Google Scholar]
- 5.Senapati J, Daver NG, Pemmaraju N. Antibody-Drug Conjugates in Myeloid Leukemias. Cancer J 2022;28(6):454–461. (In eng). DOI: 10.1097/ppo.0000000000000635. [DOI] [PubMed] [Google Scholar]
- 6.Dillon LW, Gui G, Ravindra N, et al. Measurable Residual FLT3 Internal Tandem Duplication Before Allogeneic Transplant for Acute Myeloid Leukemia. JAMA Oncol 2024. (In eng). DOI: 10.1001/jamaoncol.2024.0985. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Short NJ, Zhou S, Fu C, et al. Association of Measurable Residual Disease With Survival Outcomes in Patients With Acute Myeloid Leukemia: A Systematic Review and Meta-analysis. JAMA Oncol 2020;6(12):1890–1899. (In eng). DOI: 10.1001/jamaoncol.2020.4600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood 2017;129(4):424–447. DOI: 10.1182/blood-2016-08-733196. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Döhner H, Wei AH, Appelbaum FR, et al. Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN. Blood 2022;140(12):1345–1377. DOI: 10.1182/blood.2022016867. [DOI] [PubMed] [Google Scholar]
- 10.Othman J, Tiong IS, O'Nions J, et al. Molecular MRD is strongly prognostic in patients with NPM1-mutated AML receiving venetoclax-based nonintensive therapy. Blood 2024;143(4):336–341. (In eng). DOI: 10.1182/blood.2023021579. [DOI] [PubMed] [Google Scholar]
- 11.Othman J, Potter N, Ivey A, et al. Molecular, clinical, and therapeutic determinants of outcome in NPM1-mutated AML. Blood 2024;144(7):714–728. DOI: 10.1182/blood.2024024310. [DOI] [PubMed] [Google Scholar]
- 12.Lindsley RC, Mar BG, Mazzola E, et al. Acute myeloid leukemia ontogeny is defined by distinct somatic mutations. Blood 2015;125(9):1367–1376. DOI: 10.1182/blood-2014-11-610543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Senapati J, Loghavi S, Garcia-Manero G, et al. Clinical interrogation of TP53 aberrations and its impact on survival in patients with myeloid neoplasms. Haematologica 2020. DOI: 10.3324/haematol.2024.286465. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Khoury JD, Solary E, Abla O, et al. The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Myeloid and Histiocytic/Dendritic Neoplasms. Leukemia 2022;36(7):1703–1719. DOI: 10.1038/s41375-022-01613-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Arber DA, Orazi A, Hasserjian RP, et al. International Consensus Classification of Myeloid Neoplasms and Acute Leukemias: integrating morphologic, clinical, and genomic data. Blood 2022;140(11):1200–1228. DOI: 10.1182/blood.2022015850. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Martínez-Cuadrón D, Megías-Vericat JE, Gil C, et al. Outcomes after intensive chemotherapy for secondary and myeloid-related changes acute myeloid leukemia patients aged 60 to 75 years old: a retrospective analysis from the PETHEMA registry. Haematologica 2024;109(1):115–128. (In eng). DOI: 10.3324/haematol.2022.282506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.DiNardo CD, Jonas BA, Pullarkat V, et al. Azacitidine and Venetoclax in Previously Untreated Acute Myeloid Leukemia. New England Journal of Medicine 2020;383(7):617–629. DOI: doi: 10.1056/NEJMoa2012971. [DOI] [PubMed] [Google Scholar]
- 18.Senapati J, Urrutia S, Loghavi S, et al. Venetoclax abrogates the prognostic impact of splicing factor gene mutations in newly diagnosed acute myeloid leukemia. Blood 2023;142(19):1647–1657. DOI: 10.1182/blood.2023020649. [DOI] [PubMed] [Google Scholar]
- 19.Chan O, Al Ali N, Tashkandi H, et al. Mutations highly specific for secondary AML are associated with poor outcomes in ELN favorable risk NPM1-mutated AML. Blood Advances 2024;8(5):1075–1083. DOI: 10.1182/bloodadvances.2023011173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Eckardt J-N, Bill M, Rausch C, et al. Secondary-type mutations do not impact outcome in NPM1-mutated acute myeloid leukemia – implications for the European LeukemiaNet risk classification. Leukemia 2023;37(11):2282–2285. DOI: 10.1038/s41375-023-02016-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Döhner H, Pratz KW, DiNardo CD, et al. Genetic Risk Stratification and Outcomes Among Treatment-Naive Patients With AML Treated With Venetoclax and Azacitidine. Blood 2024. (In eng). DOI: 10.1182/blood.2024024944. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Döhner H, Pratz K, Dinardo C, et al. ELN Risk Stratification Is Not Predictive of Outcomes for Treatment-Naïve Patients with Acute Myeloid Leukemia Treated with Venetoclax and Azacitidine. Blood 2022;140:1441–1444. DOI: 10.1182/blood-2022-169509. [DOI] [Google Scholar]
- 23.Senapati J, Kantarjian HM, Haddad FG, et al. Outcomes of Patients With Treated Secondary Acute Myeloid Leukemia: A High-Risk Subtype That Warrants an Independent Prognostic Designation. American Journal of Hematology 2025;100(2):249–259. DOI: 10.1002/ajh.27561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Döhner H, DiNardo CD, Wei AH, et al. Genetic risk classification for adults with AML receiving less-intensive therapies: the 2024 ELN recommendations. Blood 2024. DOI: 10.1182/blood.2024025409. [DOI] [PubMed] [Google Scholar]
- 25.DiNardo CD, Tiong IS, Quaglieri A, et al. Molecular patterns of response and treatment failure after frontline venetoclax combinations in older patients with AML. Blood 2020;135(11):791–803. DOI: 10.1182/blood.2019003988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Gangat N, Johnson I, McCullough K, et al. Molecular predictors of response to venetoclax plus hypomethylating agent in treatment-naïve acute myeloid leukemia. Haematologica 2022;107(10):2501–2505. (In eng). DOI: 10.3324/haematol.2022.281214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Senapati J, DiNardo CD, Kadia TM, et al. Cohesin Complex Gene Mutated Acute Myeloid Leukemia (AML) Responds Favorably to Venetoclax Based Frontline Therapy. Blood 2024;144(Supplement 1):4306–4306. DOI: 10.1182/blood-2024-208602. [DOI] [Google Scholar]
- 28.Pollyea DA, Pratz KW, Wei AH, et al. Outcomes in Patients with Poor-Risk Cytogenetics with or without TP53 Mutations Treated with Venetoclax and Azacitidine. Clin Cancer Res 2022;28(24):5272–5279. (In eng). DOI: 10.1158/1078-0432.Ccr-22-1183. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data available from corresponding author on reasonable request