Predicting Severity of Radiation Induced Lymphopenia in Individual Proton Therapy Patients for Varying Dose Rate and Fractionation Using Dynamic 4-Dimensional Blood Flow Simulations

Abstract

Purpose: Radiation-induced lymphopenia has gained attention recently as the result of its correlation with survival in a range of indications, particularly when combining radiation therapy (RT) with immunotherapy. The purpose of this study is to use a dynamic blood circulation model combined with observed lymphocyte depletion in patients to derive the in vivo radiosensitivity of circulating lymphocytes and study the effect of RT delivery parameters.

Methods and materials: We assembled a cohort of 17 patients with hepatocellular carcinoma treated with proton RT alone in 15 fractions (fx) using conventional dose rates (beam-on time [BOT], 120 seconds) for whom weekly absolute lymphocyte counts (ALCs) during RT and follow-up were available. We used HEDOS, a time-dependent, whole-body, blood flow computational framework, in combination with explicit liver blood flow modeling, to calculate the dose volume histograms for circulating lymphocytes for changing BOTs (1 second-300 seconds) and fractionations (5 fx, 15 fx). From this, we used the linear cell survival model and an exponential model to determine patient-specific lymphocyte radiation sensitivity, α, and recovery, σ, respectively.

Results: The in vivo-derived patient-specific α had a median of 0.65 Gy^-1 (range, 0.30-1.38). Decreasing BOT to 1 second led to an increased average end-of-treatment ALC of 27.5%, increasing to 60.3% when combined with the 5-fx regimen. Decreasing to 5 fx at the conventional dose rate led to an increase of 17.0% on average. The benefit of both increasing dose rate and reducing the number of fractions was patient specificࣧpatients with highly sensitive lymphocytes benefited most from decreasing BOT, whereas patients with slow lymphocyte recovery benefited most from the shorter fractionation regimen.

Conclusions: We observed that increasing dose rate at the same fractionation reduced ALC depletion more significantly than reducing the number of fractions. High-dose-rates led to an increased sparing of lymphocytes when shortening the fractionation regimen, particularly for patients with radiosensitive lymphocytes at elevated risk.

Fig. 1.

Overview of methods for blood dose calculation (A, left), estimating lymphocyte radiosensitivity and recovery (B, middle), and patient-specific estimate of lymphocyte depletion across alternative RT parameters (C, right). Abbreviations: ALC = absolute lymphocyte count; RT = radiation therapy.

Fig. 2.

Waterfall plot of a (A, left) and number of days until recovery to 80% of initial ALC (B, right) with overlying median line (dashed) in both figures. *Note: for 4 patients, the recovery was predicted to be longer than the simulated length of 300 days from treatment start. Abbreviation: ALC = absolute lymphocyte count.

Fig. 3.

Comparison of the cumulative per fraction bDVH for a 1-second (A, top left), 120-second (B, top middle), and 300-second (C, top right) beam-on time for a sample patient and quantitative comparison of bDVH mean dose, D_mean (D, bottom left), maximum dose, D_max (D, bottom left), and volume of lymphocytes receiving greater than 0.1 Gy, V_>0.1Gy (E, bottom right) taking the mean across all patients between conventional and hypofractionated regimens. Abbreviation: bDVH = bloodstream dose-volume histograms.

Fig. 4.

The relative changes in ALC nadir compared with the conventional treatment regimen across beam-on times and fractionation schedules. The hollow circles are 5 fractions and the filled circles are 15 fractions whereas the respective median change is represented with a line for each column. Abbreviation: ALC = absolute lymphocyte count.

Fig. 5.

Relationship between the lymphocyte radiosensitivity term, α, and the ALC nadir difference when increasing the dose rate (fractionation kept constant at 15 fractions) (A, left) and the relationship between the number of days to recovery to 80% ALC baseline (in log scale) and the ALC nadir difference when shortening fractionation from 15 to 5 fractions (beam-on time kept constant at 120 seconds) (B, right). The P value was determined from the Pearson correlation coefficient. Crosses denote a negative ALC nadir difference, whereas the circles denote a positive ALC nadir difference. Abbreviation: ALC = absolute lymphocyte count.