A new method using Raman spectroscopy for in vivo targeted brain cancer tissue biopsy

Abstract

Modern cancer diagnosis requires histological, molecular, and genomic tumor analyses. Tumor sampling is often achieved using a targeted needle biopsy approach. Targeting errors and cancer heterogeneity causing inaccurate sampling are important limitations of this blind technique leading to non-diagnostic or poor quality samples, and the need for repeated biopsies pose elevated patient risk. An optical technology that can analyze the molecular nature of the tissue prior to harvesting could improve cancer targeting and mitigate patient risk. Here we report on the design, development, and validation of an in situ intraoperative, label-free, cancer detection system based on high wavenumber Raman spectroscopy. This optical detection device was engineered into a commercially available biopsy system allowing tumor analysis prior to tissue harvesting without disrupting workflow. Using a dual validation approach we show that high wavenumber Raman spectroscopy can detect human dense cancer with >60% cancer cells in situ during surgery with a sensitivity and specificity of 80% and 90%, respectively. We also demonstrate for the first time the use of this system in a swine brain biopsy model. These studies set the stage for the clinical translation of this optical molecular imaging method for high yield and safe targeted biopsy.

Figure 1

Schematic representation of the optical core needle biopsy, with a magnified view of the tip showing the biopsy window and the beveled optical fibers used for illumination and detection. The fibers are located opposite to the biopsy window and the needle is rotated by exactly 180° to collect a tissue sample spatially co-localized with a spectral measurement. Figure produced by Dariush Bagheri.

Figure 2

RS measurements in normal brain. (a) Schematic of the in vivo acquisition steps: 1) needle insertion along the planned trajectory, 2) RS collection, 3) 180° rotation of the needle, and 4) tissue collection. (b) In vivo Raman spectra averaged over all measurements (n = 11) and compared with ex vivo spectra for white matter and cortex.

Figure 3

(a) Schematic depiction of in vivo RS measurements taken in the surgical cavity during glioma resection using the handheld contact probe in dense cancer (red), infiltrated brain (yellow) and surrounding normal brain. (b) In vivo high wavenumber Raman spectra of dense cancer, infiltrated brain and normal brain, averaged over all samples. (c) Representative H&E-stained micrographs for each tissue type.

Figure 4

(a) Boxplots of the ratio of the lipid and protein bands (2930 cm⁻¹/2845 cm⁻¹) for normal brain, infiltrated brain and dense cancer tissue in glioma patients. (b) Receiver operating characteristic (ROC) curve computed using the SVM algorithm and leave-one-out cross-validation. The indicated point at the minimal distance from the upper-left corner of the ROC curve was chosen for calculating the sensitivity and specificity values.