Magnetic resonance imaging diagnosis of subependymal giant cell astrocytomas in follow-up of children with tuberous sclerosis complex: should we always use contrast enhancement?

BACKGROUND Subependymal giant cell astrocytomas (SEGAs) arise in 10-26% of tuberous sclerosis complex (TSC) patients. SEGAs cause obstructive hydrocephalus and increase morbi-mortality. It is recommended that TSC patients be followed with contrast enhanced magnetic resonance imaging (CE-MRI), but repetitive use of gadolinium-based contrast-agents (GBCAs) may cause organ deposits. OBJECTIVE To compare the diagnostic performances of non-CE- and CE-MRI to differentiate SEGAs from subependymal nodules in TSC patients during follow-up. MATERIALS AND METHODS Thirty-five TSC patients (median age: 2.4 years) were enrolled in this retrospective single-center study from September 2007 to January 2019. Inclusion criteria were a certain diagnosis of TSC and at least three follow-up brain MRIs with GBCA injection. Two consecutive MRI scans per patient were selected and anonymized. Three radiologists performed a blinded review of non-enhanced and enhanced MRI sequences during different sessions. The diagnostic performances were compared (sensitivity, specificity, positive/negative predictive values, accuracy, inter/intra-observer agreements). RESULTS The accuracies for detecting SEGAs were good and similar between the non-enhanced and enhanced MRI sequences. The sensitivity and specificity of non-CE-MRI to diagnose SEGA ranged from 75% to 100% and from 94% to 100%, respectively. The differences in numbers of false-positive and false-negative patients between non-CE- and CE-MRI never exceeded one case. Nodules size >10 mm, location near the Monro foramen, hydrocephalus and modifications between two consecutive MRI scans were significantly associated with the diagnosis of SEGA for the three readers (all P-values <0.05). Inter- and intra-observer agreements were also excellent for non-enhanced and enhanced MRI sequences (kappa=0.85-1 and 0.81-0.93, respectively). CONCLUSION The performances of non-enhanced and enhanced MRI sequences are comparable for detecting SEGAs, questioning the need for systematic GBCA injections for TSC patients.