Transfusion burden across hematologic and solid tumors: A real-world comparative study in AML, multiple myeloma, lung cancer, and melanoma Free

Transfusions are a routine aspect of cancer care; however, their frequency and implications vary by tumor type and treatment. We hypothesized that transfusion burden reflects underlying disease biology and treatment-related toxicity, with distinct patterns across hematologic and solid tumors. Despite their central role in supportive care, few real-world studies compare transfusion use across malignancies, and none to our knowledge do so across both hematologic and solid tumors. In this retrospective multi-cohort study, we used the TriNetX Analytics Network to evaluate short-term transfusion rates and early mortality across four cancers: acute myeloid leukemia (AML), multiple myeloma (MM), lung cancer, and melanoma.

Patients with AML or MM who started venetoclax-based therapy were analyzed over a 14-day follow-up to capture early marrow suppression. In parallel, patients with lung cancer or melanoma receiving immune checkpoint inhibitors (ICIs)—nivolumab or pembrolizumab—were followed for 30 days, reflecting delayed hematologic effects. Transfusions were identified using CPT code 36430. Mortality was assessed from structured EHRs. The primary outcome was transfusion; the secondary outcome was all-cause mortality. Cohorts were unmatched due to divergent treatment indications, follow-up windows, and disease biology. ICI cohorts excluded patients with pre-existing anemia, neutropenia, prior transfusions, or hospice enrollment—potentially underestimating transfusion rates. Identifying tumor-specific transfusion patterns may support blood product planning, anticipate hematologic toxicity, and guide supportive care strategies.

Among 3,806 patients with AML, 72.6% received a transfusion within 14 days of therapy, compared to 27.4% of 5,673 patients with MM (risk difference 45.2%; risk ratio [RR] 2.65, 95% CI 2.55–2.76)—representing more than a twofold increase in early transfusion needs. Among solid tumors, transfusion occurred in 2.2% of 6,445 lung cancer patients and 1.1% of 6,083 melanoma patients (risk difference 1.1%; RR 1.96). Mortality followed a similar pattern: AML (61.3%) and lung cancer (38.5%) had higher early mortality than MM (28.8%) and melanoma (20.5%). The AML transfusion rate exceeds that seen in phase 3 venetoclax trials, suggesting a greater real-world demand for hematologic support. These findings may reflect more advanced disease at presentation, intensified regimens, or lower baseline marrow reserve in routine practice.

These results reveal a biologically consistent gradient of transfusion need. In AML, high utilization likely reflects marrow failure and cytotoxic intensity. In MM, contributors may include chronic anemia, cumulative therapy, and transplant history. The higher rate in lung cancer versus melanoma, despite similar ICI exposure, may reflect comorbidities, osseous involvement, or baseline cytopenias. Understanding these differences may help clinicians tailor surveillance intensity and supportive care planning during the early phases of treatment.

These patterns may inform preemptive planning and patient counseling, particularly in individuals with limited marrow reserve or restricted transfusion access. Transfusions may delay recovery, increase infection risk, and diminish quality of life—especially in under-resourced settings. They can also impact hospitalization rates, clinic workflow, and transfusion service demand—affecting care delivery at the system level. Additionally, transfusion events may be under- or misclassified due to coding variability or EHR lag. As we could not distinguish prophylactic from reactive transfusions, or adjust for lab values, findings warrant cautious interpretation. Follow-up durations were not uniform, and matching was not feasible across such distinct populations. Still, this study provides a practical benchmark for how transfusion requirements vary by cancer type and treatment approach.

These findings support prospective validation using lab-based cytopenia data and assessment of transfusion use as a marker of early toxicity or clinical deterioration. Such tools could inform real-time transfusion protocols, trigger-based monitoring, or risk-adapted care pathways—particularly in settings with constrained supportive care resources. Future directions include integrating transfusion need into clinical risk models, exploring its role in treatment response prediction, and tailoring supportive care strategies accordingly.