UCSF

Abstract

INTRODUCTION

Microscopic residual disease at the tumor margin is the primary barrier to complete resection which impacts outcome. Stimulated Raman scattering microscopy (SRM) has proven effective to classify CNS tumor tissue and detect tumor cells within grossly normal post mortem glioma specimens. In this study, we use SRM for detection of residual glioma cells within the resection cavity margins.

METHODS

Patients with gliomas undergoing surgical resection were included. Margin samples were taken after the primary surgeon determined the resection cavity wall to be grossly normal and were analyzed using (1) true H&E, (2) SRM pseudo H&E, (3) immunohistochemical stains (IHC) for IDH1-R132H or p53. The sections were scored on a scale of 0–3 by a neuropathologist blinded to the corresponding results from each modality (0=no definite tumor cells; 1=rare tumor cells; 2=moderate tumor cells with preserved neuropil; 3 = abundant tumor cells). Positive and negative predictive values and Spearman correlation coefficients were calculated.

RESULTS

Ninety-one margin samples from 19 patients were included. Tumors were WHO grade II (29.7%), III (38.5%), and IV (31.8%). There was a strong correlation between SRM and H&E (Spearman correlation (r) = 0.76), SRM and IHC (r=0.81), and H&E and IHC scores (r=0.91). PPV of SRM score of 3 was 100%. The presence of tumor cells on SRM (scores 1–3) strongly correlated with the presence of tumor cells on H&E (PPV = 95%) and IHC (PPV = 91%). The absence of tumor cells on SRM correlated with absence of tumor on H&E and IHC only 10% and 50% (NPV) of the time, respectively.

CONCLUSIONS

The PPV of utilizing SRM to detect residual glioma cells in grossly normal resection cavity margins is high however the NPV is low. SRM may have utility as a rapid point-of-care intraoperative tool for identification of infiltrative glioma within resection cavity margins.