Detection of ovarian cancer (± neo-adjuvant chemotherapy effects) via ATR-FTIR spectroscopy: comparative analysis of blood and urine biofluids in a large patient cohort

Abstract

Ovarian cancer remains the most lethal gynaecological malignancy, as its timely detection at early stages remains elusive. Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy of biofluids has been previously applied in pilot studies for ovarian cancer diagnosis, with promising results. Herein, these initial findings were further investigated by application of ATR-FTIR spectroscopy in a large patient cohort. Spectra were obtained by measurements of blood plasma and serum, as well as urine, from 116 patients with ovarian cancer and 307 patients with benign gynaecological conditions. A preliminary chemometric analysis revealed significant spectral differences in ovarian cancer patients without previous chemotherapy (n = 71) and those who had received neo-adjuvant chemotherapy-NACT (n = 45), so these groups were compared separately with benign controls. Classification algorithms with blind predictive model validation demonstrated that serum was the best biofluid, achieving 76% sensitivity and 98% specificity for ovarian cancer detection, whereas urine exhibited poor performance. A drop in sensitivities for the NACT ovarian cancer group in plasma and serum indicates the potential of ATR-FTIR spectroscopy to identify chemotherapy-related spectral changes. Comparisons of regression coefficient plots for identification of biomarkers suggest that glycoproteins (such as CA125) are the main classifiers for ovarian cancer detection and responsible for smaller differences in spectra between NACT patients and benign controls. This study confirms the capacity of biofluids' ATR-FTIR spectroscopy (mainly blood serum) to diagnose ovarian cancer with high accuracy and demonstrates its potential in monitoring response to chemotherapy, which is reported for the first time. ATR-FTIR spectroscopy of blood serum achieves good segregation of ovarian cancers from benign controls, with attenuation of differences following neo-adjuvant chemotherapy.

Keywords: ATR-FTIR spectroscopy; Biofluids; Chemotherapy; Liquid biopsies; Ovarian cancer; Spectroscopy.

ATR-FTIR spectroscopy of blood serum achieves good segregation of ovarian cancers from benign controls, with attenuation of differences following neo-adjuvant chemotherapy.

Fig. 1

PCA score plots with P-values for intra-class comparisons in plasma, serum and urine. Top graphs: non-endometriosis (benign) versus endometriosis benign controls. Bottom graphs: non-chemotherapy (OC no chemo) versus NACT (OC chemo) ovarian cancer patients. OC, ovarian cancers; chemo, chemotherapy; NACT, neo-adjuvant chemotherapy

Fig. 2

PCA score plots with P-values for inter-class comparisons in plasma, serum and urine. Top graphs: non-chemotherapy ovarian cancers (OC no chemo) versus all benign controls (controls). Bottom graphs: NACT ovarian cancers (OC chemo) versus all benign controls (controls). OC, ovarian cancers; chemo, chemotherapy; NACT, neo-adjuvant chemotherapy

Fig. 3

PLS-DA discriminant function plots for plasma, serum and urine. Top graphs: non-chemotherapy ovarian cancers (OC no chemo) versus all benign controls (controls). Bottom graphs: NACT ovarian cancers (OC chemo) versus all benign controls (controls). OC, ovarian cancers; chemo, chemotherapy; NACT, neo-adjuvant chemotherapy. o = training samples; + = test samples

Fig. 4

PLS-DA regression coefficient plots for identification of spectral biomarkers in plasma, serum and urine. Key wavenumber regions have been marked with different colours. Top graphs: non-chemotherapy ovarian cancers (OC no chemo) versus all benign controls (controls). Bottom graphs: NACT ovarian cancers (OC chemo) versus all benign controls (controls). OC, ovarian cancers; chemo, chemotherapy; NACT, neo-adjuvant chemotherapy