San Antonio, Texas USA (UroToday.com) Due to the wide spread use of abdominal section imaging, the number of incidental renal masses has increased. 20% of these masses are estimated to be benign and require no treatment. Furthermore, about 6000 renal masses are resected and found to be benign every year.
The use of renal mass biopsy is still controversial due to its in accuracies. Raman spectroscopy has been previously used for tissue characterization in kidney, bladder and liver neoplasms and was shown to differentiate accurately between benign and malignant tissues however; the use of 785nm or 830 nm lasers in tissues with strong auto-florescence has not been effective. The aim of the present study was to assess the ability of 1064 nm RS system to differentiate malignant and normal kidney tissues.
12 ex vivo samples of renal cell carcinoma and normal kidney were obtained from the Vanderbilt University Cooperative Human Tissue Network. A benchtop RS system, using a 1064 nm laser, was constructed. The pre-processing stage included white light correction and florescence subtraction using polynomial fitting. Then, the data was input into a machine learning algorithm, Sparse Multinomial Logistic Regression (SMLR) that provided the probability of class membership for each of the 12 samples. In order to identify the most important Raman Shift band, a metric called feature importance was used.
A total of 87 measurements were performed using the 1064 RS system. Out of the 355 Raman shift bands, the SMLR algorithm identified 155 significant features. 18 features had a feature importance higher than 0.2. The posterior probability of malignant class assignment separated clearly between the malignant and normal tissues. Lastly the accuracy, sensitivity and specificity of the 1064 RS system were 85%, 82% and 87% respectively.
In conclusion, 1064nm RS system and accurately differentiate between malignant and normal kidney tissues. This suggests the use of the system as an optical biopsy tool. The system’s ability to identify nuclear grade and tumor subtype were not examined and need to be tested in the future.
Authors: Miki Haifler, Isaac J. Pence, Benjamin Ristau, Richard Greenberg, David Chen, Marc Smaldone, Alexander Kutikov, Rosalia Viterbo, Robert Uzzo, Amnon Zisman, Anita Mahadevan-Jansen, Chetan Patil
Written By: Miki Haifler, MD, SUO Fellow, Fox Chase Cancer Center
17th Annual Meeting of the Society of Urologic Oncology - November 30 -December 2, 2016 – San Antonio, Texas USA