Linear regression analysis for complete blood count parameters during radiotherapy.

Abstract

Objective: This study aimed to evaluate the correlations between complete blood count (CBC) during radiotherapy and patient and treatment factors.

Patients and methods: Data of cancer patients, including age, sex, concurrent chemotherapy (CCRT), radiotherapy dose (equivalent dose in 2‑Gy fractions with an alpha/beta value of 10 Gy, EQD2Gy10), radiotherapy location, and baseline CBC were collected. Linear regression was used to determine results during radiation. Validation data comprised 20% of the whole cohort.

Results: A total of 496 radiotherapy courses and 1884 weekly CBC results during treatment were analyzed. Baseline hemoglobin (Hb) was positively associated with subsequent Hb. Each 1 g/dL increase in baseline Hb predicted a 0.73 g/dL increase in Hb during treatment (95% confidence interval [CI] 0.7-0.76). Male sex was associated with a 0.16 g/dL higher Hb (95% CI 0.04-0.29), while female sex showed the opposite trend. CCRT was associated with a 0.18 g/dL reduction in Hb (95% CI -0.33 to -0.03). Radiotherapy to the pelvis, bone, and head and neck regions resulted in Hb reductions of 0.18, 0.34, and 0.94 g/dL, respectively (95% CI -0.33 to -0.03, -0.53 to -0.15, and -1.26 to -0.62, respectively), while brain irradiation increased Hb by 0.22 g/dL (95% CI 0.05-0.38). Age, cumulative dose, and thoracic irradiation did not show a significant correlation with Hb changes. Adjusted R‑squared for the development and validation data were 0.6 and 0.71 for Hb, 0.42 and 0.11 for white blood cell count, 0.36 and 0.32 for neutrophils, 0.42 and 0.06 for absolute neutrophil count, and 0.43 and 0.36 for platelets, respectively.

Conclusion: Hb levels during radiotherapy could be explained using linear regression, although they did not negatively correlate with cumulative dose.