Integration of cerebrospinal fluid methylome and proteome can obviate the need for surgical biopsy in primary central nervous system lymphoma

Sunday, May 5, 2024

2:31 PM – 2:33 PM CT

Location: S405a

Presenting Author(s)

Alexander Landry, MD

Neurosurgery Resident
University of Toronto
Toronto, ON, CA

Co-Author(s)

AA

Andrew Ajisebutu, MD

Neurosurgery Resident
University of Toronto, United States
YE

Yosef Ellenbogen, MD

Neurosurgery Resident
University of Toronto, United States
CG

Chloe Gui, MD

Neurosurgery Resident
University of Toronto, United States
Farshad Nassiri, MD PhD

Neurosurgeon
University of Toronto
Toronto, Ontario, United States
VP

Vikas Patil, PhD

Bioinformatician
University of Toronto, United States
MV

Mathew R. Voisin, MD

Neurosurgery Resident
University of Toronto
Toronto, Ontario, United States
Justin Z. Wang, MD

Neurosurgery Resident
University of Toronto
Toronto, ON, CA
Gelareh Zadeh, MD PhD

Chair of Neurosurgery
University of Toronto
TORONTO, Ontario, Canada
JZ

Jeff Zuccato, MD PhD

Neurosurgery Resident
University of Toronto, United States

Disclosure(s):

Alexander Landry, MD: No financial relationships to disclose

Justin Z. Wang, MD: No financial relationships to disclose

Gelareh Zadeh, MD PhD: No financial relationships to disclose

Introduction: The diagnosis of brain tumors currently requires histopathological examination of tissue obtained by neurosurgery, which comes with inherent risks. Non-invasive techniques for diagnosis provide an opportunity to mitigate these risks, particularly in patients with CNS lymphoma who do not derive benefit from tumour debulking/resection. In this study, we sought to determine whether integrating the cerebrospinal fluid (CSF) methylome and proteome from patients with glioblastoma (GBM), brain metastases (BM), and primary CNS lymphoma (PCNSL) can reliably distinguish these entities without surgical biopsy.

Methods: Patients with histopathology confirmed GBM (IDH wild type), BM, and PCNSL with accompanying CSF were included in our study. CSF was sampled during placement of Ommaya reservoirs as part of their clinical care. Elastic net models were trained on proteomic and methylation data separately and together using both early and late integration schemes. Classification performance was assessed with AUC, sensitivity, and specificity.

Results: Our cohort includes 20 patients with GBM, 17 with BM, and 14 with PCNSL, each with matching CSF cell free DNA methylation and shotgun proteomic profiling. We show that these data can be integrated to fully discriminate PCNSL from its major diagnostic counterparts with a perfect median AUC of 1.00 (95% CI 1-1) as well as 100% specificity. Integrated models significantly outperform models trained on methylation or protein data alone, though the same degree of improvement was not demonstrated in GBM or BM, suggesting this observed synergistic biological information is specific to lymphoma.

Conclusion : There is a critical need to diagnose patients with intra-axial tumours, particularly CNS lymphoma, without relying on invasive and costly surgery. We present the most specific and accurate CNS lymphoma classifier to date by integrating the methylome and proteome of CSF. This has the potential for immediate clinical utility, eliminating the need for invasive and costly surgery in an important subset of patients.