Changes in DTI and DSC Parameters as Markers for Assessing Treatment Response in Glioblastomas
Sumei Wang1, Sanjeev Chawla1, Maria Martinez-Lage2, Tianyu Yin1, Gaurav Verma1, Robert A Lustig3, Steven Brem4, Suyash Mohan1, Ronald L Wolf1, Arati Desai5, and Harish Poptani6

1Radiology, University of Pennsylvania, Philadelphia, PA, United States, 2Pathology and Laboratory Medicine, University of Pennsylvania, Philadelphia, PA, United States, 3Radiation Oncology, University of Pennsylvania, Philadelphia, PA, United States, 4Neurosurgery, University of Pennsylvania, Philadelphia, PA, United States, 5Hematology-Oncology, University of Pennsylvania, Philadelphia, PA, United States, 6Cellular and Molecular Physiology, University of Liverpool, Liverpool, United Kingdom


The study was performed to determine whether changes in DTI and DSC parameters can aid in differentiating glioblastomas with pseudo-progression (PsP) from true-progression (TP) and partial response. MRI data from thirty patients with these diagnoses (based on pathological evaluation and clinical follow-up) were included. All patients underwent two MR scans before pathological confirmation. A significant increase in median rCBV and rCBVmax value was noted in TP compared with PsP, while none of the DTI parameters showed significant differences between groups. Our preliminary results indicate that changes in rCBV may be helpful in identifying PsP from TP.


Early assessment of treatment response is critical in patients with glioblastomas as patients exhibiting pseudo-progression (PsP) can be continued on conservative therapy while patients with true progression (TP) or recurrence can be candidates for novel therapeutic regimens or repeat surgery. However, in the absence of a true predictor, routine follow-up with MRI is recommended until the patient shows clinical deterioration or develops a substantial increase in tumor volume, which can substantially delay their clinical management and affect survival. DTI and DSC have been shown to be useful in differentiating PsP from TP. However, most of the studies focused on initial evaluation and surgical confirmation1-3. We hypothesized that changes in physiologically sensitive imaging parameters may also aid in better assessment of treatment response, especially when surgical biopsy/resection is not clinically warranted.


Thirty glioblastoma patients (12M/18F, age 24-78) who had completed standard TMZ and radiation treatment, exhibited new enhancing lesions within six months and with an equivocal clinical diagnosis of PsP, partial (mixed) response or TP were included in the study. As the diagnosis was equivocal, all patients were recommended for a follow-up MRI scan within 4-8 weeks. Both the baseline and follow-up scans included DTI and DSC MRI. Most patients underwent repeat surgery after the second MRI for determination of PsP versus TP. Pathological evaluation of the surgical tissue categorized patients as PsP (n=8) with <25% of tissue exhibiting malignant features and TP (n=11) with >75% recurrent tumor. Cases with 25-75% recurrent tumor characteristics were grouped as mixed or partial response (n=6). Additionally, 5 patients demonstrated decreased contrast enhancement at subsequent follow-up MRI and were thus classified as PsP at the time of the second MRI for this study. All MRI studies were performed on a 3T scanner with a 12-channel phased-array head coil. DTI data were acquired using a single shot spin echo EPI sequence with parallel imaging using GRAPPA (acceleration factor = 2); TR/TE = 5000/86 ms, NEX = 3, FOV = 22 × 22 cm2, b = 1000 s/mm2, number of diffusion weighting directions = 30, in-plane resolution = 1.72 × 1.72 × 3 mm3. DSC T2* weighted gradient-echo echo planar images were obtained using the following parameters: TR/TE = 2000/45 ms, FOV = 22 × 22 cm2, in-plane resolution = 1.72 × 1.72 × 3 mm3, and 20 slices covering the brain. MD and FA maps were computed using in house software. Leakage corrected CBV maps were generated using Nordic ICE (Nordic Imaging Lab). Contrast-enhanced T1 weighted images, FLAIR, CBV and DTI maps were co-registered and a semi-automated segmentation routine was used to segment the contrast-enhancing ROI. The median MD, FA, rCBV values from this ROI were used to analyze the data. The 90th percentile rCBV values were also measured to compute maximum rCBV (rCBVmax). To evaluate changes in MRI parameters between the two studies, percent changes between the first and the second scan were calculated as (2nd – 1st)/1st × 100. A pair-wise comparison between the PsP, TP and mixed response was performed for each parameter using a Mann-Whitney U test. Receiver operative characteristic (ROC) analysis was employed to determine the best predictor.


Representative MD, FA and CBV images from a patient at baseline (1st) and follow-up (2nd) study are shown in Fig.1. The percent changes in MD, FA, median rCBV and rCBVmax between the baseline and follow-up scans are shown in Fig. 2. The difference in the percent change in median rCBV and rCBVmax values was significantly higher in TP compared with PsP (p<0.05). Patients with mixed response showed similar trend as PsP, with a small insignificant decrease in median rCBV and rCBVmax. None of the DTI parameters showed any significant changes between groups. ROC analysis revealed that the change in rCBVmax was the best predictor for both TP vs non-TP (PsP+mixed) with AUC 0.78, a very high specificity (100%) but a low sensitivity of 46%. Using the change in rCBVmax as a parameter to evaluate patients with PsP from TP or mixed response showed an AUC of 0.69 with a low sensitivity of 46% but again with a very high specificity of 94%.


A significant increase in rCBV of up to 25% in TP patients within 4-8 weeks after initial scan, despite no apparent changes in tumor volume or contrast enhancement, indicates a rapid increase in vascular volume suggestive of aggressive tumor phenotype and increased angiogenesis4, 5. On the other hand, relatively unchanged rCBV values or a slight decrease in rCBV reflects vascular trimming reflective of treatment response in patients with PsP or mixed response. While a slight decrease in MD (reflective of higher cell density) was noted in TP patients compared to PsP and mixed response, these changes were not significant indicating limited use of DTI, probably due to tumor heterogeneity.


These findings indicate that monitoring changes in rCBV may aid in differentiating PsP from TP, however, these findings need to be validated in a larger patient cohort.


This work was supported by NIH grant 1R21CA170284.


1. Wang S, et al. AJNR 2015 epub. 2. Prager AJ, et al. AJNR, 2015 epub. 3.Kong DS, et al. AJNR 2011; 32:382. 4.Mangla R, et al. Radiology 2010; 256: 575. 5. Boxerman JL, et al. Am J Clin Oncol. 2014 epub.


Fig. 1: Representative baseline (1st) and follow-up (2nd) MR images from a glioblastoma patient showing very similar tumor size and contrast enhancement.

Fig. 2. Boxplots showing percent changes in DTI and DSC parameters. Most PsP and mixed cases show decreased rCBV, whereas TP patients demonstrate increased rCBV and rCBVmax. MD and FA changes didn't show any significant difference between groups. * indicates p<0.05

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)