Rapid evaporative ionisation mass spectrometry for breast tissue characterisation towards intraoperative margin assessment in breast surgery
File(s)
Author(s)
St John, Edward
Type
Thesis
Abstract
Background:
A fifth of patients undergoing breast conserving surgery (BCS) require re-operative intervention for positive margins. Re-operation is costly, negatively affects cosmesis and increases morbidity. Rapid Evaporative Ionisation Mass Spectrometry (REIMS) uses chemical analysis of lipid profiles from electrosurgical aerosol to classify breast tissues in real-time.
Methods:
Electrosurgical aerosol derived from breast tissues was aspirated into a mass spectrometer (MS). Spectral databases were built from malignant and non-malignant breast tissues with classification models generated using statistical methods. Significantly different peak intensities between tissue classes were identified using univariate statistics and tandem MS aided identification of lipid species. REIMS was performed during breast surgery and classification models were used for identification of cancer at the margin.
Results:
REIMS was optimised for analysis of heterogeneous breast tissue. An ex-vivo classification model built from n=359 fresh ex-vivo samples allowed for recognition of n=260 new breast tissue samples, achieving 90.9% sensitivity and 98.8% specificity. 112 possible glycerophospholipids were identified following univariate analysis and tandem MS of significant peaks. A combined ex-vivo model of fresh and frozen, cut and coag samples (n=369 tumour, n=467 normal) provided good diagnostic accuracy [sensitivity=94.9% and specificity=94.4%] from which intraoperative recognition models were built. Intraoperatively, continuous high intensity spectra were obtained (n=85 patients, 511 margins) with rapid onscreen results (1.57 seconds, SD +/-0.26), and invasive tumour margins were identified [sensitivity=77.7% (7/9), specificity=92.9% (460/495)].
Conclusions:
REIMS has been optimised for analysis of heterogeneous breast tissues based on alterations in lipid metabolism. Multivariate statistical recognition models accurately classify ex-vivo breast tissue. REIMS has potential to be further developed as an intelligent knife (iKnife), capable of rapid data collection and analysis. Pilot data suggests the system shows promise as an intraoperative, real-time, margin assessment tool for breast cancer surgery. A multicentre clinical trial [REI-EXCISE, CRUK/16/021] of REIMS during BCS will provide further validation.
A fifth of patients undergoing breast conserving surgery (BCS) require re-operative intervention for positive margins. Re-operation is costly, negatively affects cosmesis and increases morbidity. Rapid Evaporative Ionisation Mass Spectrometry (REIMS) uses chemical analysis of lipid profiles from electrosurgical aerosol to classify breast tissues in real-time.
Methods:
Electrosurgical aerosol derived from breast tissues was aspirated into a mass spectrometer (MS). Spectral databases were built from malignant and non-malignant breast tissues with classification models generated using statistical methods. Significantly different peak intensities between tissue classes were identified using univariate statistics and tandem MS aided identification of lipid species. REIMS was performed during breast surgery and classification models were used for identification of cancer at the margin.
Results:
REIMS was optimised for analysis of heterogeneous breast tissue. An ex-vivo classification model built from n=359 fresh ex-vivo samples allowed for recognition of n=260 new breast tissue samples, achieving 90.9% sensitivity and 98.8% specificity. 112 possible glycerophospholipids were identified following univariate analysis and tandem MS of significant peaks. A combined ex-vivo model of fresh and frozen, cut and coag samples (n=369 tumour, n=467 normal) provided good diagnostic accuracy [sensitivity=94.9% and specificity=94.4%] from which intraoperative recognition models were built. Intraoperatively, continuous high intensity spectra were obtained (n=85 patients, 511 margins) with rapid onscreen results (1.57 seconds, SD +/-0.26), and invasive tumour margins were identified [sensitivity=77.7% (7/9), specificity=92.9% (460/495)].
Conclusions:
REIMS has been optimised for analysis of heterogeneous breast tissues based on alterations in lipid metabolism. Multivariate statistical recognition models accurately classify ex-vivo breast tissue. REIMS has potential to be further developed as an intelligent knife (iKnife), capable of rapid data collection and analysis. Pilot data suggests the system shows promise as an intraoperative, real-time, margin assessment tool for breast cancer surgery. A multicentre clinical trial [REI-EXCISE, CRUK/16/021] of REIMS during BCS will provide further validation.
Version
Open Access
Date Issued
2017-12
Date Awarded
2018-03
Advisor
Darzi, Ara
Leff, Daniel
Takats, Zoltan
Sponsor
Waters Corporation (Firm)
Grant Number
Waters Corporation - Micromass Grant.
RCSeng one year research fellowship
European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013) / ERC grant agreement n° 617896
Publisher Department
Department of Surgery & Cancer
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)