Repeatability of radiotherapy dose-painting prescriptions derived from a multiparametric magnetic resonance imaging model of glioblastoma infiltration

Abstract

Background and purpose: Glioblastoma (GBM) patients have a dismal prognosis. Tumours typically recur within months of surgical resection and post-operative chemoradiation. Multiparametric magnetic resonance imaging (mpMRI) biomarkers promise to improve GBM outcomes by identifying likely regions of infiltrative tumour in tumour probability (TP) maps. These regions could be treated with escalated dose via dose-painting radiotherapy to achieve higher rates of tumour control. Crucial to the technical validation of dose-painting using imaging biomarkers is the repeatability of the derived dose prescriptions. Here, we quantify repeatability of dose-painting prescriptions derived from mpMRI.

Materials and methods: TP maps were calculated with a clinically validated model that linearly combined apparent diffusion coefficient (ADC) and relative cerebral blood volume (rBV) or ADC and relative cerebral blood flow (rBF) data. Maps were developed for 11 GBM patients who received two mpMRI scans separated by a short interval prior to chemoradiation treatment. A linear dose mapping function was applied to obtain dose-painting prescription (DP) maps for each session. Voxel-wise and group-wise repeatability metrics were calculated for parametric, TP and DP maps within radiotherapy margins.

Results: DP maps derived from mpMRI were repeatable between imaging sessions (ICC > 0.85). ADC maps showed higher repeatability than rBV and rBF maps (Wilcoxon test, p = 0.001). TP maps obtained from the combination of ADC and rBF were the most stable (median ICC: 0.89).

Conclusions: Dose-painting prescriptions derived from a mpMRI model of tumour infiltration have a good level of repeatability and can be used to generate reliable dose-painting plans for GBM patients.

Keywords: ADC, apparent diffusion coefficient; CSF, cerebrospinal fluid; CTV, clinical target volume; CV, coefficient of variation; DP, dose prescription; DSC, dynamic-susceptibility contrast; Dose-painting; EORTC, European Organisation for Research and Treatment of Cancer; FLAIR, fluid-attenuated inverse recovery; GBM, glioblastoma; GTV, gross tumour volume; Glioblastoma; ICC, intraclass correlation coefficient; Multiparametric MRI; PTV, planned target volume; RC, repeatability coefficient; Radiotherapy; Repeatability; SVZ, subventricular zones; T1CE, T1-weighted post-contrast; TP, tumour probability; VOI, volume of interest; mpMRI, multiparametric MRI; rBF, relative cerebral blood flow; rBV, relative cerebral blood volume; ΔTP, difference in tumour probability between timepoint 2 and timepoint 1; σb2, between-subject variance; σw2, within-subject variance.

Fig. 1

Schematic overview of the image analysis pipeline. The pipeline involves five steps: image pre-processing, including registration of the parametric to the anatomical images, perfusion imaging modelling, registration of the parametric images from the two timepoints, image resampling, normalisation and standardisation; tumour probability modelling, according to the linear formula obtained from the regression analysis coefficients; dose prescription mapping, linearly mapping TP values to a dose prescription; volume of interest delineation from the gross tumour volume and the clinical target volume; repeatability analysis, with calculation of repeatability metrics for the parametric, TP and DP maps within the volume of interest. ADC, apparent diffusion coefficient; D_max, maximum dose; D_min, minimum dose; DP, dose prescription; rBF, relative blood flow; rBV, relative blood volume; TP, tumour probability.

Fig. 2

Comparison of MRI-derived parametric, tumour probability and dose prescription maps between two imaging sessions. The figure displays left–right, top–bottom T1CE images from the two timepoints with overlayed a) volume of interest contours shown in purple, b) ADC maps, c) rBV maps, d) rBF maps, e) ADC-rBV TP maps, f) ADC-rBF TP maps, g) ADC-rBV DP maps, h) ADC-rBF DP maps. T1CE, T1-weighted contrast enhanced image; ADC, apparent diffusion coefficient; DP, dose prescription; rBF, relative blood flow; rBV, relative blood volume; TP, tumour probability.

Fig. 3

Example of histograms of voxel-wise tumour probability. The figure displays top–bottom, normalised histograms of the distributions of ADC-rBV (top) and ADC-rBF (bottom) voxel-wise tumour probability at timepoint 1 (left), at timepoint 2 (centre) and difference of voxel-wise tumour probability between timepoint 2 and timepoint 1. ADC, apparent diffusion coefficient; rBF, relative blood flow; rBV, relative blood volume. Bin size: 100.

Fig. 4

Voxel-wise repeatability metrics. The figure displays in a| ICC values and in b| within-voxel CV values of the parametric, TP and DP maps obtained from the voxel-wise repeatability analysis. ADC, apparent diffusion coefficient; CV, coefficient of variation; DP, dose prescription, ICC, intraclass correlation coefficient; rBF, relative blood flow; rBV, relative blood volume; TP, tumour probability. Bars represent the median values from the 11 GBM patients.

Fig. 5

Bland–Altman plots for analysis of repeatability. The figure displays the Bland-Altman plots of the mean ADC values (top); mean rBV, mean ADC-rBV TP and mean ADC-rBV DP (left); mean rBF, mean ADC-rBF TP and mean ADC-rBF DP (right) calculated from the volume of interest-based repeatability analysis. ADC, apparent diffusion coefficient; DP, dose prescription; rBF, relative blood flow; rBV, relative blood volume; TP, tumour probability. Coloured dotted lines represent limits of agreement. Black dotted lines represent bias.