Author: Sophie Bockel, Eric Deutsch, Frederic Dhermain, Ibrahima Diallo, Anh Thu Le, Elaine Limkin, Pauline Maury, Charlotte Robert, Killian Sambourg, Camilla Satragno, Cristina Veres, François de Kermenguy 👨🔬
Affiliation: Gustave Roussy, Département de radiothérapie, Université Paris-Saclay, Gustave Roussy, Inserm U1030, Radiothérapie Moléculaire et Innovation Thérapeutique 🌍
Purpose:
To study the correlation between the dose to circulating lymphocytes as evaluated by the LymphoDose framework and the incidence of severe radiation-induced lymphopenia (sRIL) in patients treated for high-grade gliomas.
Methods:
Patients with high-grade gliomas treated with CRT have been retrospectively identified. For each patient, head-and-neck lymph nodes were automatically contoured, and out-of-field doses delivered to these structures were calculated using an in-house deep learning algorithm. The LymphoDose framework was then applied for each patient to calculate DVH received by peripheral blood (H1) and circulating lymphocytes (H3). The blood/lymphocyte pool irradiated volume V>0Gy, V>0.125Gy, mean dose, D50%, and D2%, in addition to several relevant clinical parameters and treatment characteristics, were used for the statistical analysis. In univariate analysis, each variable was assessed for its association with absolute lymphocyte count (ALC) nadir. A stepwise variable selection using Variance Inflation Factors (VIF ≤ 2) addressed multicollinearity among the selected variables, yielding a refined set for a subsequent multivariable analysis. P-values < 0.05 were considered statistically significant.
Results:
A total of 184 patients were included in the study, with an average of 7.2 ± 2.2 lymphocyte blood samples collected between baseline and one-year post-RT. After univariate analysis, 12 variables were selected, of which 6 were removed through VIF-based selection. In multivariate analysis, only the ALC baseline (p = 0.001) was significantly correlated with the ALC nadir. The KPS (p = 0.056) and the median dose to lymphocytes (p = 0.06) approached the statistical significance.
Conclusion:
The moderate correlations between the dose received by highly radiosensitive circulating lymphocytes and ALC nadir suggest that the direct cytotoxic effect of irradiation does not fully explain the occurrence of sRIL. Other indirect effects could be considered, such as the irradiation-induced myeloid-derived suppressor cells, and mechanistic learning techniques could help to take these phenomena into account.