Real-Time Diagnosis of Ovarian Cancer with the Surgical Intelligent Knife (iKnife)

David Phelps1,Julia Balog1,2,Mona El-Bahrawy1,Abigail Speller1,Zoltan Takats1,Robert Brown1,Sadaf Ghaem-Maghami1

1Imperial College, London, UK,2Waters Corporation, Budapest, Hungary

Presenting date: Wednesday 4 November
Presenting time: 10.30-10.45

Background

Women have a 2% lifetime risk of developing ovarian cancer. Five year survival for stage III or IV disease is 19% and 3% respectively(1). Standard treatment is surgical debulking followed by chemotherapy. Some centres give neo-adjuvant chemotherapy followed by delayed debulking, which renders tumour deposits difficult to identify during surgery. Debulking to zero residual disease improves prognosis for stage IV to 54.6 months compared to 23.9 months with >10mm residual disease(2).


Surgeons rely on frozen tissue sections for histopathological diagnosis during surgery which is time-consuming and expensive. The inability to accurately identify non-descript lesions can lead to more radical, perhaps unnecessary surgery.

Method

A Xevo Q-TOF mass spectrometer was used with a modified handheld diathermy. The technology analyses the diathermy smoke for ionic species. The spectra and histopathological diagnoses were used to populate a library with background subtracted and lock mass corrected spectra in the phospholipid range. Principal component (PCA) and linear discriminant analysis (LDA) were used to find the variance in spectral signatures between chosen tissue groups.

Results

486 sampling points were obtained from 144 fresh frozen tissues (normal ovary n=15, fallopian tube n=15, peritoneum n=15, benign tumours n=14, borderline n=15, malignant n=70). The iKnife differentiated tissue types using unique phospholipid spectral signatures as fingerprints. The leave-one out patient cross-validation showed 100% sensitivity and specificity in the separation of normal ovary and malignant ovary sampling points (n=189). Sensitivity and specificity remained 100% when including normal fallopian tube and peritoneum in the model (n=291). Overall the correct tissue classification rate was 92.7%, reduced from 100% due to some spectral crossover within the normal tissue groups.

Conclusion

The iKnife accurately separates different gynaecological tissues. In surgical gynae-oncology the iKnife has the potential to reduce operating times, improve margin status and identify lesions. Enhanced on-table decision making may result in reduced morbidity and improved survival.