Radiation Therapy And Risk Of Herpes Zoster In General Cancer Patients: A Propensity Score Matched Analysis

Herpes zoster (HZ) is a common comorbidity among cancer patients. The role and impact of radiation therapy (RT) on the development of HZ is unclear. In this study, we investigated the association between the two by statistical analysis and assessment of anatomical correlation. Institutional database for National Cancer Registry (2011–2015) was the main data source for this study. Data pertaining to HZ and other comorbidities were sourced from the diagnosis database for social insurance claims. Propensity score matched analysis was performed to ensure comparability of the RT and non-RT cohorts. Propensity score was calculated using logistic regression based on age, sex, primary cancer site, existence of metastatic lesion, use of systemic therapy, use of surgical intervention, and status of coexisting disease, including diabetes mellitus, autoimmune diseases and HIV infection. The HZ event rate in two groups was compared using the long-rank test and Cox proportional hazard model; index day was adjusted to align the lead period from the date of cancer registration to the initiation of RT using new-user cohort design. The index day for the analysis was the initial day of the first course of RT for the RT cohort; an identical duration after the date of cancer registration was used to determine the index day for the respective non-RT matched counterpart. Individual case records of patients who developed HZ after RT were used to assess the association between RT field and anatomical distribution of HZ lesions; the expected number of HZ events within the radiation field was calculated on the basis of the RT field extent in each case and contrasted with the observed number of events. Of the 17,655 cancer patients, propensity score matching yielded 4,350 pairs; of these, 3,891 pairs were eligible for comparison. Distribution of baseline covariates was adequately balanced in matched dataset. RT group showed significantly higher incidence of HZ at 2 years (3.0 %) than that in the non-RT group (1.0 %) (log-rank p < 0.001; hazard ratio: 2.59, 95% confidence interval: 1.84 to 3.66). Of the 120 patients who developed HZ after RT, the median number of involved dermatomal segments within the radiation field was 20 (first quartile–third quartile: 12–26). The expected and the observed numbers of HZ events within the radiation field were 39.9 and 75, respectively (p < 0.001). In this retrospective database study, general cancer patients who received RT showed a significantly higher risk of HZ. Post-RT HZ events occurred significantly more frequently within the radiation field. These findings suggest a relationship between RT and the development of HZ.