Diagnostic accuracy of MRI without gadolinium for follow-up of pilocytic astrocytoma in the paediatric population

Abstract

Purpose

Pilocytic astrocytoma is the most common glial tumour in the paediatric population with a long-term life expectancy after surgery. Long-term radiological follow-up with magnetic resonance imaging (MRI) is necessary to detect recurrence of tumour or growth of residual tumour. Established MRI protocols typically dictate post-gadolinium sequences despite concerns around the side-effect and safety concerns of gadolinium-based contrast agents. This study aims to investigate whether omission of gadolinium-enhanced sequences for the long term follow-up MRI for paediatric patients with pilocytic astrocytoma, maintains diagnostic accuracy assessing potential recurrence or growth of residual tumour.

Methods

A retrospective review of follow-up MRI for 47 patients with histopathologically proven pilocytic astrocytoma was performed. Patients with optic pathway or suprasellar glioma were excluded from this study. All patients underwent surgery and had a minimum of 2 years of postoperative imaging for review. MRIs were chosen from most recent report of stability or at a time when growth/progression had been diagnosed. Two neuroradiologists and two paediatric neurosurgeons were randomly allocated a series of MRIs with gadolinium enhanced sequences removed, reviewers were blinded to the original report and subsequent treatment decisions. In addition, 30 paired MRI studies were randomly allocated to second review to test interobserver reliability. The reviewer responses were recorded and compared with the original report and analysed with respect to preserved diagnostic accuracy.

Results

170 MRI scans were subject to review across 66 episodes of care for 47 patients. 22.7 % of patients experienced growth of residual tumour during the period of follow-up. The sensitivity of non-enhanced MRI for detection of growth was 82 % (95 %CI 64.40–92.12) with a specificity of 97.10 % (95 % CI 90.03–99.20). Accuracy was similar for both neuroradiologists and neurosurgeons (91.49 % vs. 94 %). Interobserver reliability was calculated using Cohen’s Kappa co-efficient with a result of 0.792 showing substantial agreement. We also confirmed a statistically significant difference between gross total resection and sub-total resection and correlation with future growth (41 % vs. 0 %, n = 64, p = 0.001).

Conclusion

In paediatric patients who have undergone surgery for pilocytic astrocytoma, long term MRI follow-up without gadolinium-enhanced sequences maintains diagnostic accuracy compared with enhanced sequences. Omission of gadolinium-enhanced sequences may lead to decreased costs, duration of scans and anxiety around follow-up procedures.