AUTHORS
Ian Bouligny, M.D., Joya Chandra, PhD2, Tapan Kadia I, MD3, Naval Daver, MD4, Gautam Borthakur, MD5, Naveen Pemmaraju, MD6, Ghayas C. Issa, MD7, Maro Ohanian, DO8, Nicholas Short, MD9, Yesid Alvarado Valero, MD10, Guillermo Montalban-Bravo, MD11, Elias Jabbour, MD12, Musa Yilmaz, MD13, Steven Kornblau, MD14, Danielle Hammond, MD15, Fadi G. Haddad, MD16, Koichi Takahashi, MD,PhD17, Lucia Masarova, MD18, Alexandre Bazinet, MD, MSc19, Mahesh Swaminathan, MD20, Eitan Kugler, MD PhD21, Hussein Ali Abbas, MD, PhD22, Wei Ying Jen, BM BCh, FRCPath23, Jayastu Senapati, MBBS, MD, DM24, Alex Bataller, MD, PhD25, Abhishek Maiti, MBBS26, Farhad Ravandi, MD27, Guillermo Garcia-Manero, MD28, Courtney DiNardo, MD, MSCE29
AFFILIATIONS
1 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
2 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
3 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
4 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
5 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
6 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
7 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
8 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
9 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
10 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
11 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
12 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
13 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
14 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
15 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
16 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
17 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
18 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
19 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
20 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
21 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
22 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
23 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
24 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
25 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
26 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
27 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
28 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
29 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
ABSTRACT
Background
Smoking is a dose-dependent risk factor for AML and promotes clonal hematopoiesis by favoring expansion of hematopoietic stem cells with mutations in epigenetic regulators. No study has examined the impact of smoking on genomics and clinical outcomes in the venetoclax era. We conducted a preplanned analysis of four prospective trials to determine whether smoking is associated with distinct clinicopathologic features and inferior survival in newly diagnosed AML treated with frontline venetoclax-based strategies.
Methods
Patients with newly diagnosed AML and documented smoking history enrolled on four prospective venetoclax-based trials at MD Anderson (2017–2025) were analyzed: CLIA-venetoclax (n=107), FLAG-ida-venetoclax (n=53), CLAD-LDAC-venetoclax (n=217), and decitabine-venetoclax (n=118). All endpoints and covariates were prespecified. The primary endpoint was OS between smokers and never-smokers by log-rank and multivariable Cox regression adjusting for ELN 2022 risk, treatment intensity, ECOG, age, and SCT (time-dependent). Secondary endpoints included EFS, CRc, MRD negativity, and 30/60-day mortality. Molecular and cytogenetic associations were assessed using logistic regression. Unsupervised hierarchical clustering was performed on 81-gene NGS panels and cytogenetic data. The study required ≥227 events for 80% power to detect a hazard ratio of 1.45 (α=0.05).
Results
Of 495 patients analyzed (ever-smokers, n=184; never-smokers, n=311), median age was 65.8 years (IQR 55.2–71.8), and 54.9% were male. By ELN 2022, 20.2% had favorable-, 18.8% intermediate-, and 61.0% adverse-risk disease. Common cytogenetic findings included normal karyotype (39.6%), complex karyotype (24.4%), +8 (11.9%), del(5q) (10.1%), and -7 (7.9%). Frequently mutated genes included DNMT3A (26.5%), TET2 (23.4%), NPM1 (20.8%), TP53 (19.8%), and NRAS (19.0%).
At a median follow-up of 3.8 years, 248 deaths (50.1%) were observed. Ever-smokers had inferior OS (median 604 vs. 1,751 days; log-rank P=0.001; univariable HR 1.51, 95% CI 1.17–1.94; 2-year OS 48.9% vs. 60.4%). Ever-smokers were more likely male (68.5% vs. 46.9%; P<0.001), had higher CCI (5 vs. 4; P=0.044), and worse ECOG (≥2: 22.2% vs. 14.8%; P=0.016). After prespecified multivariable adjustment, the HR was 1.23 (95% CI 0.95–1.59; P=0.11); SCT was independently protective (HR 0.47; P<0.001). EFS was shorter among ever-smokers (median 417 vs. 608 days; P=0.045).
CRc (81.0% vs. 81.4%; P=0.91), MRD negativity (66.1% vs. 71.1%; P=0.30), and 30/60-day mortality did not differ between smoke-exposure cohorts, suggesting that inferior survival is not driven by treatment resistance. Instead, ever-smokers harbored significantly more adverse-risk biology: ELN 2022 adverse-risk disease was more prevalent (67.9% vs. 56.9%; OR 1.60; P=0.017), with enrichment of monosomy 7 (OR 3.34; P<0.001), del(12p) (OR 8.75; P=0.004), -17/abn(17p) (OR 2.44; P=0.015), monosomy 5 (OR 4.06; P=0.044), SRSF2 (OR 1.76; P=0.023), and PTPN11 (OR 1.97; P=0.032). A composite of these six smoking-associated adverse features was present in 53.3% of ever-smokers vs. 26.4% of never-smokers (OR 2.95; P<0.001), and causal mediation analysis confirmed this composite as a significant mediator of the smoking-survival association (P=0.034). Within the ELN adverse-risk subgroup (n=302), ever-smokers retained worse OS (HR 1.35, 95% CI 1.02–1.80; P=0.037).
Unsupervised clustering identified three genomic clusters with distinct survival (P<0.001): a NPM1/normal karyotype cluster, a chromatin-spliceosome cluster (enriched with RUNX1, SRSF2, and ASXL1), and a TP53/complex karyotype cluster. The ever-smoker proportion increased progressively across clusters (P=0.05); the TP53mut/complex cluster had the highest ever-smoker proportion (44.2%) and worst median OS (211 days). Among ever-smokers with pack-year data, increasing pack-year burden was associated with SF3B1 mutations (OR per 10 pack-years: 1.38, 95% CI, 1.08–1.75; P=0.009), consistent with a dose-dependent effect of smoke exposure.
Conclusions
In the largest current study of smoking and AML outcomes in the venetoclax era, smoke exposure was significantly associated with adverse cytogenetics, SRSF2, and PTPN11 mutations. Causal mediation analysis identified these recurrent adverse-risk features as a pathway linking smoking to inferior survival. Despite equivalent initial responses, smoking conferred worse OS both overall and within the ELN 2022 adverse-risk group.
DISCLOSURES
Ian Bouligny, M.D.: No relevant conflicts of interest to declare.