AUTHORS
Wajeeha Aiman, MD, Muhammad Ashar Ali, MD2, Naheed Alam, MD3, Muhammad Salman Faisal, MBBS, MD4, Salman Fazal, MD5
AFFILIATIONS
1 Roswell Park Comprehensive Cancer Center, Buffalo, New York, United States
2 Allegheny General Health, Pittsburgh, Pennsylvania, United States
3 Erie County Medical Center, Buffalo, New York, United States
4 University of Oklahoma Stephenson Cancer Center, Oklahoma City, New York, United States
5 UPMC Hillman Cancer Center, Harrisburg, New York, United States
ABSTRACT
Background
Estrogen receptor (ER) signaling has emerged as a targetable axis in myeloid neoplasm. Tamoxifen induces apoptosis of hematopoietic stem cells and multipotent progenitors, blocks development of JAK2 V617F–driven myeloproliferative neoplasm in vivo and sensitizes MLL-AF9 leukemias to chemotherapy while sparing steady-state hematopoiesis. ERβ is selectively toxic to leukemia stem cells when targeted pharmacologically, and in NPM1-mutated acute myeloid leukemia, E2/ERα signaling transcriptionally upregulates hepatocyte growth factor to promote blast survival, with ER blockade plus cytarabine yielding synergistic antileukemic activity in vitro and in vivo. The TAMARIN phase II trial met its A'Hern success criterion for mutant allele burden reduction in myeloproliferative neoplasm but noted thrombotic events. Despite the preclinical rationale, no clinical data addresses whether patients incidentally exposed to selective estrogen receptor modulators (SERMs) around the time of a myeloid malignancy diagnosis experience different outcome.
Methods
We reviewed the TriNetX US Collaborative Network. The inclusion criteria had adults with myeloid neoplasms classified as SERM-exposed (tamoxifen, toremifene, raloxifene, or clomiphene) who had exposure within last year, excluding hematopoietic cell transplant and the comparator cohort was myeloid neoplasms without any SERM exposure (non-SERM). Index was the first myeloid neoplasm diagnosis. Cohorts were 1:1 propensity-score matched on all recorded baseline characteristics including age, sex, ethnicity, comorbidity index (Heart Failure, Chronic kidney disease, Diabetes), BMI, diagnosis of breast cancer, exposure of antineoplastic agents, and smoking status. All-cause mortality was assessed at 1, 3, and 5 years. Secondary outcomes over days 1–365 were hospitalization/ED encounter, relapse of myeloid neoplasm, red cell transfusion, platelet transfusion, granulocyte colony-stimulating factor (G-CSF) use, and venous thromboembolism (VTE). Risk ratios, Kaplan-Meier estimates with log-rank tests, and Cox hazard ratios with proportionality testing were computed; patients with the outcome before the window were excluded.
Results
Unmatched cohorts comprised of 1,907 SERM and 216,949 non-SERM patients: after propensity matching 1,734 patients per arm were selected. Mean ages at diagnosis were 63 ± 16.8 and 57 ± 15.8 in SERM vs non-SERM cohort, respectively. The ratio of female was 86.3% and 45.5% in SERM and non-SERM cohort, respectively.
Mortality was lower with SERM exposure at all three intervals, with stable effect size and no interval reaching significance. At 1 year: 15.8% vs 17.4%; RR 0.910 (0.784–1.058); HR 0.901 (0.765–1.062); p=0.215; overall survival was 82.9% vs 80.8% in SERM vs non-SERM. At 3 years: risk was 22.4% vs 24.3%; RR 0.920 (0.815–1.039); HR 0.899 (0.783–1.033); p=0.133; overall survival 73.9% vs 70.9%. At 5 years: risk was 25.4% vs 28.0%; risk difference −2.6% (−5.5% to +0.4%); RR 0.908 (0.813–1.015); p=0.089. Median survival was not reached.
Secondary 1-year outcomes favored the SERM cohort. Hospitalization/ED visits were less in the SERM cohort 14.2% vs non-SERM cohort 23.0%; RR 0.617, HR 0.563 (0.427–0.742; p=0.001). Among those hospitalized, the mean number of encounters was lower with SERM exposure (4.34 ± 5.79 vs 5.60 ± 9.59; p=0.003). Red cell transfusion was also reduced in SERM cohort (12.2% vs 17.1%); RR 0.714, HR 0.698 (0.572–0.851; p=0.001). G-CSF use was significantly reduced as well in SERM cohort 6.2% vs 9.8%; RR 0.632, HR 0.616 (0.475–0.799; p=0.001). VTE risk was also less: 3.4% vs 5.0% in SERM vs no SERM cohort; RR 0.685, HR 0.678 (0.478–0.961; p=0.028). 0.563 (0.427–0.742); p<0.001.It is significant but there is limited evidence available on the usage of anticoagulation and other factors like active healthier lifestyle in these cohorts.
Disease Relapse (5.1% vs 6.4%; HR 0.785, 0.583–1.059) and platelet transfusion (4.9% vs 5.0%; HR 0.980, 0.720–1.333) did not differ.
Conclusions
SERM exposure was associated with significantly lower 1-year hospitalization, red cell transfusion, and VTE, and a consistent but non-significant 9–10% relative mortality reduction through 5 years. Reduced transfusion requirements without a definitive survival benefit are more consistent with preserved hematopoiesis. This abstract is hypothesis generating for further studies.
DISCLOSURES
Wajeeha Aiman, MD: No relevant conflicts of interest to declare.