Radiation-induced changes in brain tissue can be divided into acute, early delayed and late effects.1, 2, 3, 4 Several publications suggest that vascular damage is the primary cause of acute and early delayed effects.3, 5 Both acute and early delayed effects are considered reversible, manifesting as dilation and thickening of blood vessels, decrease in vessel density, endothelial cell damage and disruption of the blood-brain-barrier in normal appearing brain tissue.1, 3, 4, 6
Few studies have assessed radiation-induced changes in brain perfusion in normal appearing brain tissue after fractionated radiotherapy (FRT) or stereotactic radiosurgery.5, 7, 8, 9, 10, 11, 12, 13, 14 Contradictory results have been published, however, perfusion techniques and post-processing methods differ extensively between studies. Overall, the published data show a reduction of both cerebral blood volume (CBV) and cerebral blood flow (CBF) after completion of FRT or single fraction stereotactic radiosurgery, with an inverse dose-response relationship.
Perfusion MRI is useful in the diagnostic evaluation of gliomas as well as for longitudinal response assessment and prognostication, with dynamic susceptibility contrast (DSC)-MRI being the most widely applied perfusion MRI technique in clinical practice.15, 16, 17, 18, 19, 20 DSC-MRI is also one of several physiological imaging techniques that has the potential to be incorporated into the Response Assessment in Neuro-Oncology (RANO) criteria as proposed by the RANO working group.21, 22 DSC-MRI can assess perfusion parameters like CBV and CBF but has several limitations, leading to both quantification and reproducibility issues.18 These limitations are related to acquisition and post-processing of the data.15, 16, 18, 23, 24, 25, 26 Generally, only relative measurements,
Seventeen patients, 18 years or older, with newly detected glioma WHO grade III-IV proven by histopathology and scheduled for FRT and chemotherapy were included prospectively. This study was done in accordance with the declaration of Helsinki and was approved by the local ethical committee (
Exclusion criteria were inconsistent or missing MRI examinations and/or deviation from a prescribed total radiation dose of 60 Gray (Gy).
All MR examinations were performed with a consistent imaging protocol on a 1.5 T scanner (Avanto Fit, Siemens Healthcare, Erlangen, Germany) and included DSC perfusion and contrast-enhanced 3D-T1-weighted (3D-T1w) images.
DSC-MRI (2D-EPI, gradient-echo; Repetition time/Echo time/Flip angle = 1340/30/90; 128 × 128 matrix; 1.8×1.8×5 mm3; time resolution = 1.34 s; 18 slices). A bolus of 5 ml gadolinium-based contrast agent (GBCA) (Gadovist, Bayer AG, Berlin, Germany) was administered for a DCE-MRI and was also regarded as a pre-bolus to diminish the effects of contrast agent extravasation15, 18 for the following DSC-MRI. For the actual DSC-MRI, a second standard dose bolus of 5 ml GBCA was administered. The contrast agent was administered using a power-injector at a rate of 2 ml/s for the first injection and 5 ml/s for the second injection. 3D-T1w image (3D-gradient Echo; Repetition time/Echo Time/Inversion time/Flip angle = 1170/4.17/600/15; 256 × 256 matrix: 1×1×1 mm3: 208 slices)
CT imaging for radiotherapy planning was acquired with a Philips, Brilliance Big Bore (Philips Healthcare, Best, the Netherlands) with a voxel size of 0.525 × 0.525 × 2 mm3.
Signal to concentration time curves conversion has been described previously.29, 30 Concentration time curves were visually inspected before analysis. CBV was determined as the ratio of areas under the tissue and arterial concentration time curves. CBF was determined through deconvolution as the initial height of the tissue impulse function.24, 29 Deconvolution was carried out using standard singular value decomposition (sSVD) with Tikhonov regularisation with an iterative threshold.29, 31, 32, 33 A patient-specific arterial input function (AIF) was defined in the middle cerebral artery branches in the hemisphere contralateral to the tumour19 in the pre-FRT examination, the same AIF was then applied to the patient’s following post-FRT examinations. Contrast agent leakage correction was applied according the method described by Boxerman
CBV and CBF maps were co-registered to the pre-FRT 3D-T1w images for each patient and examination using the SPM12 toolbox (Wellcome Trust Centre for Neuroimaging, London, UK). Planned radiation dose levels for each region were acquired by rigidly transforming the dose-planning CT (including related radiation dose plans) to the pre-FRT 3D-T1w images using the standard Elastix registration toolbox.36
Grey matter (GM) and WM probability maps were segmented from the 3D-T1w images, for each examination, using the segmentation tool in the SPM12 toolbox. WM and GM maps were defined as partial volume fraction above 70%. Contrast-enhancing tissue, oedema, resection cavity, tumour progression and recurrence, if present, were excluded, reviewed by an experienced neuroradiologist. Registered radiation dose plans were divided as follows: 0–5 Gy, 5–10 Gy, 10–20 Gy, 20–30 Gy, 30–40 Gy, 40–50 Gy and 50–60 Gy, creating a total of seven binary dose regions for each tissue type, (mean volume and standard deviation for each dose region is presented in S1 Table).
Mean CBV and CBF were calculated in each dose region and normalised (nCBV and nCBF) to the mean CBV and CBF in 0–5 Gy WM and GM regions, respectively. Super- and subscripts are used to distinguish between tissue type from which measurements were derived (superscript) and reference tissue type used for normalisation (subscript),
Mean relative change and standard deviation (SD) were calculated and presented as a part of the descriptive analysis. A linear regression model was applied to assess a possible dose-response relationship between relative change and received radiation dose. Graphpad Prism 7 for Mac (Graphpad Software, La Jolla California USA) was used for statistical analysis and graph design.
Seven patients were excluded due to inconsistent or missing pre-FRT examinations (
A representative dose region distribution map with corresponding pre-FRT 3D-T1w image are displayed in Figure 1. Global nCBV and nCBF with 95% CI and derived
and
decreased at FRTPost-1, decreased further at FRTPost-2, and recovered at FRTPost-3, however, still below corresponding values at pre-FRT. Significant differences were found for all values except
at FRTPost-2;
and
showed the same tendency. Only small variations between pre-FRT and post-FRT examinations were present in
and
implying no change after FRT. Comprehensive figures over regional nCBV and nCBF and derived
Mean, standard error of mean (SEM) and change relative pre-fractionated radiotherapy (pre-FRT) for global nCBV and nCBF. Global normalised cerebral blood volume (nCBV) and normalised cerebral blood flow (nCBF) (mean and SEM), change in percentage relative pre-FRT (mean and SD) and derived
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
ΔCBV
1.63±0.03
1.53±0.04
1.49±0.05
1.53±0.03
ΔnCBV (%)
-6.7±7.7
-4.6±11.7
-6.0±9.3
p-value
<
0,0535
<
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
nCBF
1.60±0.03
1.53±0.03
1.37±0.04
1.46±0.03
ΔnCBF (%)
-5.1±11.7
-12.5±11.4
-7.5±8.2
p-value
<
<
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
nCBV
1.10±0.03
1.06±0.03
1.03±0.03
1.05±0.03
ΔnCBV (%)
-4.3±7.6
0.7±11.7
-3.6±11.9
p-value
0,3818
0,1197
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
nCBF
1.09±0.02
1.08±0.03
0.96±0.03
1.05±0.03
ΔnCBF (%)
-3.1±7.7
-7.4±10.5
-3.4±10.3
p-value
0,2267
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
nCBV
1.00±0.01
0.99±0.01
1.02±0.02
1.0±0.02
ΔnCBV (%)
-0.9±4.0
1.7±5.7
-0.4±6.8
p-value
0,1629
0,184
0,6445
Pre-FRT
FRTPost-1
FRTPost-2
FRTPost-3
nCBF
1.02±0.01
1.01±0.01
1.00±0.01
1.02±0.01
ΔnCBF (%)
-0.1±5.1
-2.8±4.7
0.0±4.5
p-value
0,4922
0,9252
Using a linear regression model, both
and
demonstrated an inverse response to radiation dose,
and
demonstrated a varied response to radiation dose. Furthermore, the same tendency could be seen for
and
In Figure 3, relative change and derived linear regression curve and equation for regional nCBV is illustrated (corresponding for nCBF can be found in S3 Figure).
In this study, we found decreasing perfusion values indicating acute and early delayed effects in normal appearing brain tissue after FRT. We also observed a dose-response relationship in WM but not in GM.
Petr
A number of contradictory findings have been reported in the literature. Jakubovic
Radiation-induced vascular structural changes, such as dilation and thickening of vessels, decreased vessel density, blood-brain-barrier disruption, endothelial cell damages may introduce thrombosis, tortuosity and occlusion. This could affect the perfusion and together with decreased vessel density and may partly explain our results.3, 4, 9 Furthermore, the recovery in nCBV and nCBF seen in our results agrees with the theory of acute and early delayed effects being reversible.1, 3, 4 However, radiation-induced changes in brain tissue is a complex process involving several tissue elements. Moreover, histopathology has mainly been studied in rodent models or single dose experiments.3, 5, 45 Interpreting findings from animal models and applying them to humans should be done with caution.
Our findings suggest that the GM response to the administrated treatment is independent of the radiation dose received; however, there is still an apparent reaction to radiation. This suggestion is based on two preliminary findings; first, a linear regression of relative difference in regional
and
demonstrated both positive and negative slopes, with small β values compared to WM tissue. Second, the resulting linear regression for
and
is also small and close to zero. This is to be expected if no dose-response relationship exists in GM, and furthermore, the use of radiation-induced changes in low-dose WM as reference tissue can be rejected as a confounding factor in this specific case because GM was used as the reference tissue. To the best of our knowledge, we are the first to report a perfusion decrease independent of radiation dose in GM using DSC. Several publications have shown that grey matter volume decreases after fractionated radiotherapy increasing with radiation dose46, 47, 48, 49, since both CBV and CBF are tissue volume dependent parameters, we believe that the dose-independency found in grey matter is a result of decreased grey matter volume instead of an actual response in CBV and CBF independent on radiation dose given.
Despite encouraging results, some potential limitations need to be addressed. First, the patient number is small and for the evaluation of perfusion on examination FRTPost-2 only five data sets were analysed. The severity of the disease significantly contributed to the high number of excluded patients through drop-outs and terminating examinations, which was beyond our control. Our efforts to keep a consistent FRT protocol and imaging time frame also contributed to exclusions. However, since we investigated response to radiation dose over time, it was necessary to keep both radiation dose and imaging time point consistent in the patient material. Secondly, concomitant and adjuvant chemotherapy was given to all patients. While there are no reports of temozolomide or PCV affecting brain perfusion, a recent publication reported that bevacizumab may decrease CBF in contralateral normal appearing GM.50 However, only one patient was given bevacizumab during the examinations analysed, it is therefore unlikely that our data are affected by the adjuvant chemotherapy given.
The limitations of DSC-MRI are, in the present study, considered by several post-processing selections. The use of patient-specific AIF in DSC measurements has been shown to increase the reproducibility between examinations, minimising the effects on reproducibility inherent in partial volume effects and noise.51 This approach also confronts the concern regarding misleading results due to radiation-induced changes in pixels defined as the AIF.9 Vessel segmentation was performed to eliminate macro vessel signal contributions causing overestimation of CBV and CBF. We used pre-bolus administration and contrast agent leakage correction as suggested.15, 18 Furthermore, post-FRT effects such as oedema were excluded from the dose regions during segmentation, and can thereby not influence our results.
In summary, significant decrease of global nCBV and nCBF between pre-FRT and post-FRT examinations was found in our study. As proposed by Petr
Our findings suggest that radiation-induced perfusion changes occur in normal-appearing brain tissue after FRT. This can cause an overestimation of relative tumour perfusion using DSC-MRI, and thus, can confound tumour treatment evaluation.