Analyzing Misdiagnosed Metastasis on Imaging Studies Jacqueline Angel

Analyzing Misdiagnosed Metastasis on Imaging Studies Jacqueline Angel, MD Julie Harreld, MD St. Jude Children’s Research Hospital ASNR 2016 Annual Meeting May 23 -26, 2016 Control #: 2288 Poster #: e. P-179

Disclosures Nothing to disclose

Outline Background Purpose Methods Results Conclusion

Background Children are subject to ependymomas and embryonal tumors (primitive neuroectodermal tumor, medulloblastoma, atypical teratoid rhabdoid tumor, and pineoblastoma) of the CNS, which frequently metastasize via the CSF. Ependymoma Post-contrast sagittal T 1 Medulloblastoma Post-contrast axial T 1 Primitive Neuroectodermal Tumor (PNET) Post-contrast axial T 1 Atypical Teratoid Rhabdoid Tumor (ATRT) Post-contrast axial T 1 Pineoblastoma Post-contrast sagittal T 1

Background Accurate detection of leptomeningeal metastasis is critical for appropriate risk stratification and treatment of childhood CNS tumors. MRI findings correlate better with survival than CSF analysis, but leptomeningeal metastases may be missed or misdiagnosed on examinations not optimized for pediatric cancer imaging.

Background: Leptomeningeal Metastasis Dissemination of tumor via CSF along surfaces of brain and spinal cord MRI appearance § T 1: sulci may appear normal or subtly irregular § T 1 +Contrast: may or may not enhance § T 2: most are hyperintense § FLAIR: nodular or confluent hyperintensity in sulci and cisterns § FLAIR +Contrast: increased conspicuity of leptomeningeal metastases § T 2 hyperintense § Gadolium enhancement § CSF suppressed § Non-visualization of vessels in sulci

Leptomeningeal Metastasis: Intracranial T 1 + Contrast T 2 FLAIR + Contrast Metastatic right frontal PNET (outlined). Extensive non-enhancing leptomeningeal metastases with nodular irregularity of sulci on axial T 1 WI +Contrast and axial T 2 WI, and filling of sulci on T 2 FLAIR, are much more obvious on axial FLAIR +Contrast with fat saturation.

Leptomeningeal Metastasis: Spine T 1 WI +Contrast show nodular and plaque-like enhancing medulloblastoma metastases along the spinal cord.

Purpose To investigate the association of missed metastases on MRI with technical imaging factors and radiologist qualifications to promote improvements in pediatric cancer imaging and metastasis detection.

Methods Retrospectively reviewed reports for outside (OSH) and in-house brain and spine MRIs performed within 35 days of each other for 86 children presenting to our institution between 1/1/2011 and 12/31/2012 with leptomeningeal-seeding brain tumors Contemporaneous CSF and/or in-house MRI were considered gold standard for metastasis detection OSH and in-house MRI techniques and radiologist qualifications were compared for cases of missed/misdiagnosed metastases

Methods In-house spine imaging protocol (30 minutes total): § Sagittal T 1 +Contrast § Upper and Lower sections § Two 3 mm (0 gap) acquisitions at each level, offset by one slice § Interleave two acquisitions 3 mm slices with 1. 5 mm OVERLAP § No gap, reducing chance of missing metastasis § Increase confidence of diagnosis § Axial VIBE T 1 +Contrast § 3 D acquisition, 3 mm (0 gap) reconstructions § No CSF pulsation § Sagittal T 2 § Upper and Lower § Detect non-enhancing metastases

Methods In-house brain imaging protocol (42 minutes total): Sag T 1 (4 mm, 0 gap) Ax T 2 (4 mm, 0 gap) Ax T 1 +Contrast (4 mm, 0 gap) T 1 subtractions Cor T 1 +Contrast (4 mm, 0 gap) Sag 3 D T 1 +Contrast (1 mm isotropic) Ax FLAIR +Contrast (4 mm, 0 gap)

Results 31/86 patients (36%) had leptomeningeal metastases (13 brain, 3 spine, 15 both) by initial inhouse MRI ± CSF. Anaplastic Ependymoma ATRT 2 (6%) 8 (26%) Ependymoma Medulloblastoma Pineoblastoma 2 (6%) 16 (52%) 2 (6%) PNET 1 (3%) Of these, 10/31 (32%) had metastases previously undiagnosed by OSH MRI of brain (n=3), spine (n=4), or both (n=3).

Results Contributing factors – Brain MRI: § 2/6 had no DWI § 5/6 had no post-contrast FLAIR

Results: DWI T 1 +Contrast Axial FLAIR at an OSH of a 33 mo male with ATRT does not demonstrate leptomeningeal metastasis. No DWI sequence was performed at the facility. FLAIR +Contrast In-house MRI performed 2 weeks later shows no focus of enhancement to suggest leptomeningeal metastasis. However, DWI images reveal hypercellular metastases with diffusion restriction in the left frontal horn and right Sylvian fissure (arrows). DWI

Results: Post-contrast FLAIR • Leptomeningeal metastases are best visualized on postcontrast FLAIR with fat saturation, given its T 1 and T 2 contributions T 1 post-contrast and FLAIR without contrast, OSH. In this 16 yo male with a PNET, metastases are not identified. • CSF nulling increases conspicuity of metastases in the sulci, when compared to T 2 Axial T 1 post-contrast and FLAIR post-contrast, inhouse. Non-enhancing leptomeningeal metastases clearly fill the sulci.

Results: Reading Radiologist The OSH reading radiologist did not identify an infundibular recess metastasis, which was present, on T 1 post-contrast, DWI, or ADC MR images in this 4 yo female with medulloblastoma. Normal infundibular recess on T 1 postcontrast MRI. In-house sagittal T 1 postcontrast MRI, however, shows an enhancing lesion in the infundibular recess (arrow).

Results • Contributing factors – Spine MRI: § 3/7 had no spine MRI at the outside facility § 1 out of the 4 imaged spines utilized a TSE for axial T 1 post-contrast MR images, which is susceptible to CSF pulsation artifact § 2/4 had no sagittal T 2 sequence

Results: Sagittal T 2 and TSE Sequences In-house T 2 T 1 VIBE +Contrast These metastases of an ATRT are best seen on sagittal T 2 MRI, which was not performed at the OSH. In-house T 1 VIBE +Contrast 3 D acquisition, also, shows the lesions (arrows). OSH In-house T 1 +Contrast In-house and OSH T 1 post-contrast MRI do not demonstrate the metastases. Additionally, axial T 1 post-contrast images at the OSH were performed as a TSE sequence, resulting in CSF pulsation artifact obscuring the metastases (arrow). T 1 TSE +Contrast

Results: Gap Unlike in-house examinations, all outside studies with missed metastases had gaps between slices on one or more sequences. In-house 3 mm slices 0 mm gap § Use of thin axial images (3 mm) with no gap permits lesion visibility on 3 consecutive axial images on an in-house study, increasing confidence of diagnosis compared to OSH study using thick slices and a gap. OSH 6 mm slices 2 mm gap

Results: Gap Slice thickness: 3 mm Gap: 0 mm Overlap: 1. 5 mm Slice thickness: 4 mm Gap: 1. 5 mm In-house OSH Differences in slice thickness and gap between slices can limit evaluation for metastases, as seen in this metastatic pineoblastoma on in-house and OSH sagittal post-contrast T 1 WI.

Results • In-house and OSH false positive rates were identical (2/55 or 4%) • 4 of 11 (36%) of the missed/misdiagnosed cases with identifiable OSH radiologists had a CAQ in neuroradiology, compared to 3 out of 5 (60%) in-house neuroradiologists

Conclusion Standardization and optimization of MRI technique, and interpretation by subspecialtytrained neuroradiologists, could promote more accurate pediatric CNS cancer detection and risk assessment.

References Engelhard HH, Corsten LA. Leptomeningeal metastasis of primary central nervous system (CNS) neoplasms. Cancer Treat Res 2005; 125: 71 -85. Estlin E, Gilbertson RJ, Wynn R. Pediatric hematology and oncology : scientific principles and clinical practice. Chichester, West Sussex, UK ; Hoboken, NJ: Wiley-Blackwell; 2010. Fangusaro J, Van Den Berghe C, Tomita T, et al. Evaluating the incidence and utility of microscopic metastatic dissemination as diagnosed by lumbar cerebro-spinal fluid (CSF) samples in children with newly diagnosed intracranial ependymoma. J Neurooncol 2011; 103: 693 -698. Fukuoka H, Hirai T, Okuda T, et al. Comparison of the added value of contrast-enhanced 3 D fluid-attenuated inversion recovery and magnetization-prepared rapid acquisition of gradient echo sequences in relation to conventional postcontrast T 1 -weighted images for the evaluation of leptomeningeal diseases at 3 T. AJNR Am J Neuroradiol. May 2010; 31(5): 868 -873. Griffiths PD, Coley SC, Romanowski CA, Hodgson T, Wilkinson ID. Contrast-enhanced fluid-attenuated inversion recovery imaging for leptomeningeal disease in children. AJNR Am J Neuroradiol. Apr 2003; 24(4): 719 -723. Kremer S, Abu Eid M, Bierry G, et al. Accuracy of delayed post-contrast FLAIR MR imaging for the diagnosis of leptomeningeal infectious or tumoral diseases. J Neuroradiol. Dec 2006; 33(5): 285 -291. Maroldi R, Ambrosi C, Farina D. Metastatic disease of the brain: extra-axial metastases (skull, dura, leptomeningeal) and tumour spread. Eur Radiol 2005; 15: 617 -626. Osborn AG. Osborn’s Brain: Imaging, Pathology, and Anatomy. Amirsys; 2013. Pang J, Banerjee A, Tihan T. The value of tandem CSF/MRI evaluation for predicting disseminated disease in childhood central nervous system neoplasms. J Neurooncol 2008; 87: 97 -102. Poltinnikov IM, Merchant TE. CSF cytology has limited value in the evaluation of patients with ependymoma who have MRI evidence of metastasis. Pediatric blood & cancer 2006; 47: 169 -173. Terterov S, Krieger MD, Bowen I, et al. Evaluation of intracranial cerebrospinal fluid cytology in staging pediatric medulloblastomas, supratentorial primitive neuroectodermal tumors, and ependymomas. J Neurosurg Pediatr 2010; 6: 131 -136.