Are outcomes of locally advanced cervical cancer associated with prebrachytherapy hemoglobin values and transfusion practice? An observational study comparing two large academic centres with divergent clinical guidelines

BACKGROUND AND PURPOSE: Anemia is common in locally advanced cervical cancer. Clinical practice varies greatly for management of anemia during brachytherapy, with some centres providing red cell transfusion to increase hemoglobin levels above 100 g/L. MATERIALS AND METHODS: This is a retrospective observational cohort study of adult patients with cervical cancer treated with brachytherapy at two academic hospitals. One hospital (H1) uses a liberal transfusion strategy with hemoglobin threshold of 100 g/L during brachytherapy and the other uses a restrictive target of 70 g/L (H2). RESULTS: Overall, 336 patients met inclusion criteria (H1: 150 patients, H2: 186 patients). 11 patients were excluded (2 at H1, 9 at H2). Demographics at both sites were comparable, except for cancer stage and smoking history. External beam radiation and chemotherapy provided was similar. Hemoglobin values were compared at baseline (within 4 weeks of oncology consult), and prior to the first and second brachytherapy treatments. In total, 101red blood cell (RBC) units were transfused to patients at H1 and 19 units to patients at H2. Patients were followed for a median of 37.0 months (0.6-80.5) at H1, and 33.3 months (1.6-82.0) at H2. There was no significant difference in progression-free or overall survival. Multivariable logistic regression analysis showed that FIGO stage was a predictor for both overall survival and cancer progression. Age, tumor size, chemotherapy, and hemoglobin levels were not predictors of disease progression or mortality. CONCLUSIONS: The practice of liberal transfusion should be re-evaluated in the absence of robust data to support its use. Crown Copyright (c) 2024 Published by Elsevier Inc. on behalf of American Brachytherapy Society. All rights are reserved, including those for text and data mining, AI training, and similar technologies.