Research Presentation Session
08:27J. DeBevits, Farmington / US
Purpose:
To investigate for MRI evidence of calvarial gadolinium (Gd) deposition during monthly high-dose contrast administration over two years and at 10 years, and to interrogate whether osseous Gd deposition correlates with hypophosphataemia.Methods and materials:The study cohort consisted of a retrospective analysis of 67 patients with MS or CIS who had participated in the BECOME trial and who received monthly off-label triple-dose gadopentetate dimeglumine (0.3mmol/kg) enhanced MRIs for up to 26 months. Pre-MRI blood samples were collected. An ROI mask was created in the diploic space (DS) using manual segmentation and co-registered to track signal intensity (SI) changes on fat-suppressed non-Gd T1-weighted. S/N ratios (SI Bone/SI Air) established an average SI change over time. To evaluate the association between SI changes and hypophosphataemia, we performed linear mixed regression modelling with a random intercept to test the linear trend in the SI change between the groups with and without hypophosphataemia.
Results:The monthly rate of T1-SI change in the diploic space during the first 14 months of triple-dose Gd administration was S/N=0.039 (S.E. 0.008; p<0.0001). ~10 years after the study termination, diploic S/N decreased to pre-study levels (N=28). Patients who developed hypophosphataemia (<2.5 mg/dl) at least once experienced a significantly slower increase of T1-hyperintensity compared to patients without hypophosphataemia (mean monthly S/N difference=0.034; p=0.037). Those who experienced >=1 episode of moderate hypophosphataemia (<2.0 mg/dl) did not experience any significant change in monthly diploic S/N.
Conclusion:Monthly administration of triple-dose Gd is associated with progressive T1 hyperintensity in the cranial diploic space, suggesting Gd deposition in bone. This change was lower in patients who developed hypophosphataemia and washed out after 10 years.
Limitations:The volume averaging, segmentation technique, and updates in MRI technique.
Ethics committee approvaln/a
Funding:Guerbet.
07:25S. Rozenblatt, Tel Aviv / IL
Purpose:
To compare changes in the signal intensity (SI) measured at the dentate nucleus (DN) and globus pallidus (GP) on unenhanced T1-weighted magnetic resonance (MR) images in patients who received gadobutrol (gadavist) to those who received gadoterate meglumine (dotarem).
Methods and materials:47 patients with CNS tumours who had undergone at least 4 MRI exams were divided into two groups. Group 1=25 patients who underwent gadoterate meglumine-enhanced MR imaging. Group 2=22 patients who underwent gadobutrol-enhanced MR imaging. Region of interest measurements were applied to unenhanced T1-weighted images by two radiologists. GP to the thalamus (TH), DN to middle cerebellar peduncle (MCP) SI ratio, and relative change (Rchange) between the initial and final examinations were calculated for each patient. The differences in the mean Rchange were analysed using a non-parametric regression test. The relationship between Rchange and the number of enhanced MR imaging examinations was evaluated using a Pearson correlation coefficient.
Results:Gadobutrol-enhanced MR imaging showed a significant increase in Rchange for DN-MCP (P<0.004). Contrarily, gadoterate meglumine-enhanced MR imaging showed no significant increase in Rchange for DN-MCP. There was a significant linear association between the number of MRI examinations and an increase in Rchange for DN-MCP (r=0.56, P=0.001) and for GP-TH (r=0.53, P=0.007) in patients who received gadobutrol.
Conclusion:A statistically significant dose-dependent T1-weighted signal increment observed in DN and GP is associated with multiple gadobutrol administrations but not with gadoterate meglumine administrations.
Limitations:A retrospective design consisting of patients with CNS tumours and damaged blood brain barrier, lacking a control group of patients who never received GBCAs.
Ethics committee approvalThe institutional review board approved this study.
Funding:No funding was received for this work.
05:02A. Meloni, Pisa / IT
Purpose:
We evaluated signal changes in the dentate nucleus, globus pallidus, pons, and thalamus (normalised to the deep cerebellum white matter) in T1-weighted magnetic resonance (MR) images after serial injections of gadobutrol in patients with thalassemia without neurological lesions.
Methods and materials:Three study groups were scanned at both 1.5T and 3T: 15 thalassemia patients who were transfused and chelated with ≥4 gadobutrol administrations at a high dose (0.2 mmol/Kg per scan) for cardiovascular MR, 8 thalassemia patients, and 13 healthy subjects who never received gadolinium-based contrast agents (GBCA). The iron overload was assessed by the T2* technique.
Results:Demographics were comparable among the groups.
Signal intensity (SI) ratios and T2* values at 1.5T in all regions were comparable among the three groups. No correlation was detected between SI ratios and T2* values.
In patients with more than 4 GBCA administrations, the SI ratios were not associated with the total cumulative gadolinium dose and the total number of contrast-enhanced examinations.
The SI ratios at 1.5T were significantly higher than those obtained at 3T. Moreover, no correlation was found between SI ratios at 1.5T and 3T.
Conclusion:Our study describes the lack of increased SI in T1-weighted MR images after repeated administrations of gadobutrol for cardiovascular MR studies in a high-risk population (high dose per scan, iron overload that can facilitate the transmetallation of gadolinium). A potential role of chelation therapy cannot be excluded. Moreover, SI ratios in the sampled anatomical areas differ between 1.5T and 3T machine, which is to be taken into account in trials or group analysis studies.
Limitations:The limited sample size.
Ethics committee approvalEthics committee approval obtained.
Funding:E-MIOT project: “no-profit support” from industrial sponsorships (Chiesi Farmaceutici, ApoPharma, Bayer).
05:32B. Dissaux, Brest / FR
Purpose:
The value of 4D-MRA for the follow-up of treated dural arteriovenous fistulas (DAVF) has rarely been evaluated.
We hypothesized that this technique could be valuable for the follow-up and post-therapeutic assessment of DAVF. The purpose of this study was to evaluate the performance of 4D-MRA at 3T for the follow-up of patients with treated DAVF.
Methods and materials:Patients treated for a DAVF in two centres from 2008-2019 were included if they met the following criteria: DAVF treated by embolisation or surgery and imaging during the follow-up with both 4D-MRA at 3T and DSA performed within 6 months. One reader analysed DSA and two readers analysed 4D-MRA.
For both modalities, 4D-MRA and DSA, readers first evaluated in a binary fashion whether a residual/recurrent DAVF was present or not. Then, when present, each residual/recurrent DAVF was classified according to the criteria of Cognard et al. The intertechnic agreement was assessed for detection of residual/recurrent DAVF and bleeding risk grading using Cohen’s kappa.
Results:We recorded 59 couples of examinations for 52 patients with a median age of 64 years (range 24-86 years). Sensibility, specificity, predictive positive value, and predictive negative values of 4D-MRA for the detection of a residual/recurrent DAVF were 58.3% (95% CI: 41.8% to 74.8%) and 91.4% (95% CI: 83.9% to 98.8%), respectively. Predictive positive value and predictive negative values of 4D-MRA for the detection of a residual/recurrent DAVF were 82% (95% CI: 71.6% to 93%) and 76% (95% CI: 63.7% to 88.6%). The intertechnic agreement was moderate with k=0.5 (95% CI: 0.3 to 0.7).
Conclusion:4D-MRA at 3T could be interesting for the follow-up of treated DAVF.
Limitations:A retrospective study.
Ethics committee approvaln/a
Funding:No funding was received for this work.
05:33S. Koshino, Tokyo / JP
Purpose:
To demonstrate the effects of deep learning-based computer-aided detection (CAD) on the reader’s performance for the detection of cerebral aneurysms in MR angiography (MRA).
Methods and materials:200 brain MRA datasets (93 men and 107 women; mean age, 64.7 years±15.3 [standard deviation]; mean aneurysm size, 3.3 mm±1.74), 50 with unruptured cerebral aneurysms and 150 without, were acquired on scanners from 3 different vendors at 1.5 or 3T and were retrospectively evaluated. 20 physicians (10 radiologists including 5 certificated radiologists and 5 non-certificated radiologists, and 10 neurosurgeons including 4 certificated neurosurgeons and 6 non-certificated neurosurgeons) were asked to detect cerebral aneurysms on MRA scans and associated projection images without and with CAD using ResNet–18, a convolutional neural network. The observers' performance was evaluated using a jackknife free-response receiver operating characteristics (JAFROC) analysis.
Results:The figure-of-merit (FOM) values computed using the JAFROC program significantly improved from 0.717 without CAD to 0.751 with CAD for all readers (P<0.001). The average sensitivity increased from 68.2% to 77.2% for all readers, whereas the average specificity decreased from 79.4% to 72.1%. On subgroup analyses, the FOM values increased from 0.647 to 0.697 for non-certificated radiologists, from 0.799 to 0.822 for certificated radiologists, from 0.645 to 0.681 for non-certificated neurosurgeons, and from 0.810 to 0.836 for certificated neurosurgeons. There was no statistical significance in the reading performances between radiologists and neurosurgeons both without CAD (P=0.826) and with CAD (P=0.753). However, the performances between non-certificated neurosurgeons and certificated neurosurgeons with CAD showed a significant difference (P=0.0479).
Conclusion:The diagnostic accuracy of cerebral aneurysms improved for radiologists and neurosurgeons with CAD.
Limitations:n/a
Ethics committee approvalThis study was approved by the institutional review board.
Funding:This work was funded by LPixel Inc.