Comparison of In Vivo Study of Co-Polarized and Cross-Polarized Fluorescence Spectroscopy to Co-Minus Cross-Polarized Fluorescence Spectroscopy to Diagnose Precancerous Cervical Lesions by Handheld Probe

Abstract

Background

In 2020, the number of new cases of cervix uteri was 604,127, i.e., 3.1% of all cancers, and the number of deaths was 341,831 (3.3%) among both sexes. In vivo fluorescence spectroscopy is an emerging optical technology that offers promise for the diagnosis of disease & has the capability to quickly, noninvasively and quantitatively probe the biochemical and morphological changes that occur as tissue becomes dysplastic.

Materials and Method

A cross-sectional observational study was conducted from December 2019 to September 2021 in the OBGY Department, UISEMH, in collaboration with optical imaging laboratory, BIOPHOTONICS, IIT Kanpur. A fabricated in-house fluorescence spectroscope consisting of a laser diode (405 nm) as light source and a miniature spectrometer is used to detect fluorescence signal from the sample. Patient’s cervix was examined in the OPD, using an optical handheld probe, which functions on the principle of polarized fluorescence spectroscopy. The tissues were examined and classified on the basis of varying patterns of polarized spectroscopy (co-polarized, cross-polarized and co-minus cross-polarized light). The results were compared with that of cytological, colposcopy and histopathological findings and on various demographic variables.

Results and Conclusion

In vivo handheld probe based on polarized fluorescence spectroscopy is an excellent screening technique. Co- and cross- polarized light has shown enhanced accuracy. Accuracy of co-minus cross-polarized light is poor. It is fast, noninvasive and quantitative and, with further developments, has the potential to become a regular screening tool in future.

Background

According to GLOBOCON, in 2020, the number of new cases of cervix uteri was 604,127, i.e., 3.1% of all cancers, and mortality was 341,831 (3.3%) among both sexes in all ages. The prevalence and mortality in South East Asia alone is 17.8 per 100,000 (ASR) and 10.0 per 100,000 (ASR), respectively. Approximately 84% of all cervical cancers and 88% of all deaths occur in developing countries.

Cervical cancer is almost a preventable cancer due to its long natural history, effective treatment of precancerous lesions and HPV vaccination. The latent phase prior to the development of invasive cancer is 12 years (avg), providing the opportunity of screening and prevention. The diagnosis and treatment in the advanced stages is economically challenging. Additionally lack of awareness, poor prognosis and high mortality, makes it important to diagnose the disease in its early stages providing effective treatment, thereby increasing the survival rate [1].

Various screening techniques for cervical cancer have been employed with the advent of intensive research, viz. PAP Smear [2], liquid-based cytology, colposcopy [3–5], VIA, VILI but application of such techniques on a large scale has still not become feasible due to logistic problems, subjectivity, and various shortcomings.

In view of this, a pilot study of the index research was conducted and published in 2005 “NEED FOR MASS SURVEILLANCE PROGRAM ESPECIALLY RURAL AREAS,” that identified the risk factors for cancer cervix in our population [6].

Initially with the use of traditional screening modalities, a cross-sectional study was used to devise an algorithm and develop a screening method which is accurate, easy to use and cheap [7, 8]. A handheld probe, developed by the Biophotonics Lab, IIT Kanpur, is an instrument based on polarized fluorescence spectroscopy that provides a screening method which is minimally invasive and can be done by paramedical staff with basic training [9, 10]. It can be applied even on target population who belong to rural area and do not approach any health center [11].

Materials and Methods

A cross-sectional observational was study conducted in the Department of Obstetrics & Gynecology of GSVM in collaboration with optical imaging laboratory, Department of BIOPHOTONICS, IIT Kanpur, where the design of the handheld probe has been developed, and the Department of Pathology, GSVM Medical College for histopathology from December 2019 to September 2021 [12].

• In the early stages, the neoplastic transformations of cervical tissue induce complex biochemical & structural changes in tissue such as epithelial cell morphology, metabolic activity and differentiation, stromal angiogenesis and epithelial stromal communication.

• Optical techniques can detect earliest biochemical & structural changes, whereas cytology detects late structural changes through autofluorophores in a cell.

• Normal cervical epithelial cells show autofluorescence attributed to [13]:

  • Mitochondrial NADH (peak around 460 nm), NAD and FAD (peak around 510 nm)
  • Collagen in stromal layer (peak around 400 nm).

• At the end of each Kreb's cycle, the cell gain NADH and FADH2 while it loses NAD and FAD; therefore, in a cancerous cell, NADH fluorescence increases but FAD fluorescence decreases.

• In normal cervical tissue, collagen cross-links give rise to high fluorescence in the stroma. Stromal fluorescence is greatly reduced in cervical precancers and cancers (due to breakage of collagen cross-links adjacent to neoplastic epithelium)

• This is consistent with recent studies which show that HPV-immortalized keratinocytes show increased NADH and but reduced FAD fluorescence relative to normal keratinocytes

With the existing tools, the discrimination of normal cells from cancer cells is good, but the results are limited by sensitivity and specificity in discriminating cancerous cells from precancerous lesions due to masking of biochemical information in a turbid medium by scatterers and absorbers.

Principle of fluorescence spectroscopy.

• When a biologic molecule is illuminated at an excitation wavelength, which lies within the absorption spectrum of that molecule, it absorbs this energy and gets activated from its ground state to an excited state.

• The molecule then relaxes back from the excited state to the ground state by generating energy in the form of fluorescence, at emission wavelengths.

The tissues were examined and classified on the basis of varying patterns of polarized spectroscopy, wherein diode laser light of 405 nm wavelength and xenon light (200–2000) nm wavelength was incident on tissue and polarized spectra, first in parallel followed by perpendicular planes was recorded by a miniature spectrometer (HR4000, Ocean Optics, Inc, Dunedin, Florida) [14–16].

Cancer detection based on polarized autofluorescence of FAD present in human cervical tissues showed emission of peak at 510 nm exciting with 405-nm laser diode source.

Methodology

The device consisted of 2 light sources [17, 18]:

Laser diode (405 nm):—measured polarized fluorescence.

Xenon lamp: measured polarized elastic scattering spectra.

Vertically polarized light was incident on a sample through beam splitters and lenses. By rotating the analyzer, co- and cross-polarized spectra were recorded.

• By rotating the analyzer, co- and cross-polarized spectra were recorded

• Source effect: eliminated by high pass filter (450 nm cutoff)

• Signal was directed to a spectrometer through an optical fiber and spectra recorded by it.

Source effect: eliminated by high-pass filter (450 nm cutoff) (Figs. 1, 2).

Fig. 1.

Experimental setup of handheld probe

Fig. 2.

In fluorescence spectroscopy studies reported the a healthy squamous tissue of the ectocervix, b compared to the shift of emission peak towards longer emission wavelengths for precancerous zones (decrease of emission intensity)

Band area around peak intensity of emission spectra was calculated to discriminate different normal and abnormal tissue with polarized (vv) and cross-polarized light (vh) and co-minus cross-polarized light.

The collected signal was transferred to computer, and the data were processed by Windows-based data acquisition program [19, 20].

Difference in cellular chemistry associated with tissue pathology was to be reflected on the fluorescence spectral profile, providing quantitative diagnostic [21–23].

The unwanted tissue fluorophores such as blood, discharge and other substances posed a challenge in creating a scatter free environment. To prevent this scattering, co- and cross-polarized light was subtracted. The benefit of co-minus cross-fluorescence spectroscopy is that it removes the artifacts due to unwanted fluorophores.

In each patient, pelvic examination and vaginal cytology were done. Patients with abnormal pap smear, suspicious looking cervix, post-coital bleeding, postmenopausal bleeding, leukoplakia, persistent vaginal discharge, irregular bleeding were subjected to in vitro co- and cross-polarized fluorescence spectroscopy using handheld probe in the outpatient department itself. Patients also underwent colposcopy and colposcopic findings suspicious of cervical intraepithelial neoplasia/cervical carcinoma and were subjected to colposcopic directed biopsy and the sample sent for histopathological examination for comparative study [24].

Inclusion Criteria

Patients with:

• Persistent abnormal vaginal discharge

• Unhealthy cervix

• Abnormal report of PAP smear

• Complaint of abnormal vaginal bleeding

  • Contact bleeding
  • Postmenopausal bleeding

• Patients who come for regular screening after the age of 25 years and agreed for cervical biopsy on suspicion of any abnormality

Exclusion Criteria

• Pregnant females

• Frank invasive cervical cancer

• Less than 18 years of age

• Active genital bleeding

• Not willing to participate

Sample size has been calculated for estimation of sensitivity and specificity using the following formulae-

$$n>\frac{Z_{1-\alpha /2}^2\times \text{senstivity}\left(1-\text{senstivity}\right)}{d^2\times \text{prevelance of the disease}}$$ 1

$$n>\frac{Z_{1-\alpha /2}^2\times \text{senstivity}\left(1-\text{senstivity}\right)}{d^2\times (1-\text{prevelance of the disease)}}$$ 2

We have used the results of results of our previous studies [11] for preliminary estimates of sensitivity and specificity and incidence of cancer cervix to be 22 per 1,00,000 women. Based on this, we can safely assume a prevalence of 50/1,00,000 women. Z_{1 − α/2} = 1.96, and based on the results of the above study, the minimum sample size of diseased cases came out to be 30 and 44, respectively. For better results, we have taken a sample size of 89.

Result

On correlating our cytology findings with histology, sensitivity was calculated (at 95% confidence interval) to be 89.09% (77.75–95.89%) and specificity was 88.24% (72.55–96.70%) and accuracy was 88.76%. It also shows the correlation of colposcopic finding to histopathology. We used Swede score for scoring of lesions. The sensitivity (at 95% confidence interval) was 60.00% (45.91–72.89%) while specificity was 97.06% (84.67–99.93%) and accuracy was 74.16% (Table 1).

Table 1.

Co-relation of cytological findings with histopathological findings and of colposcopic findings in relation to histopathological diagnosis (swede score) [3–5]

Cytological findings	HPE					Total
	INF	CIN 1	CIN 2	CIN 3	CA
Normal	8	0	0	0	0	8
INF	22	5	2	0	0	29
LSIL	4	14	5	4	0	27
HSIL	0	1	5	7	5	18
CA	0	0	0	0	7	7
Total	34	20	12	11	12	89
Colposcopy grade
< 3	33	16	5	1	0	55
4–5	0	4	6	2	0	12
6–9	1	0	1	8	12	22
Total	34	20	12	11	12	89

Observation of Optical Spectroscopy

ROC curve to determine the diagnostic ability of screening method using co-polarized fluorescence spectroscopy between normal and CIN1/CIN2/CIN3/Carcinoma in situ. AUC was found to be 0.909 (p < 0.001). It was very high and statistically significant, and this implies that handheld probe in vivo trials has proved itself as an excellent discriminator between normal and diseased cases. 95% confident interval of AUC is 0.83–0.96 implying that we were 95% confident that the screening method under study will also move to be an excellent discriminator in the population also. The ideal cut off value for intensity was ≤ 2498.67. At this cut-off point, sensitivity was 98.18% and specificity was 79.41%. High sensitive was required to minimize false negative (Figs. 3, 4).

Fig. 3.

ROC curve for co-polarized

Fig. 4.

ROC curve for cross-polarized light

Figure 5 shows the ROC curve to determine the diagnostic ability of screening method using cross-polarized fluorescence spectroscopy between normal and CIN1/CIN2/CIN3/carcinoma in situ. AUC was found to be 0.883 (p < 0.001). It was very high and is also statistically significant; this implies that handheld probe in vivo trials has proved itself as an excellent discriminator between normal and diseased cases. 95% confident interval of AUC is 0.798–0.942 implying that we are 95% confident that the screening method under study will be an excellent discriminator in the population also. The ideal cut off value for intensity is ≤ 2398.423. At this cutoff point, sensitivity is 96.36% and specificity is 70.59%. High sensitive was required to minimize false negative.

Fig. 5.

ROC curve for co-minus cross-polarized light

Figure shows low AUC of 0.511 and poor sensitivity, specificity and a low accuracy in predicting precancerous cervical lesions.

Discussion

On comparing the sensitivity and specificity for screening of precancerous cervical cancerous lesion by colposcopy and fluorescence spectroscopy, fluorescence spectroscopy has emerged to be a better screening modality.

Previously in vitro and in vivo studies have been done, and from in vitro studies it was analyzed that sensitivity of handheld probe in detecting precancerous lesions using co-polarized light was 92.86% while specificity was 94.87%, while for cross-polarized light, sensitivity was 92.86% and specificity was 89.74%. Kiran Pandey et al.(2020) evaluated from in vivo studies analyzed that sensitivity of handheld probe in detecting precancerous lesions using co-polarized light was 100% while specificity was 88.9%, while for cross-polarized light, sensitivity was 100% and specificity was 83.3%.

The high and satisfactory discriminatory power of co- and cross-polarized fluorescence spectroscopy was established from previous studies but with a smaller data.

The present study was conducted to validate their results in a new dataset. It was also postulated by the IIT team that co-minus cross-polarized light might yield better results.

When the results of ROC curve analysis of co-polarized and cross-polarized fluorescence spectroscopy were compared, though both are excellent discriminators of the disease, co-polarized fluorescence spectroscopy was found to be better than cross-polarized fluorescence spectroscopy as AUC was higher (0.909) in co-polarized fluorescence spectroscopy than cross-polarized (0.883). Thus, the results of our study were in line with the results of previous studies and it is established that both co-polarized and cross-polarized light yield reliable values of intrinsic fluorescence and provide a highly sensitive and specific marker to identify diseased cases. Of the two, co-polarized light is slightly better. The analysis of co-minus cross-values of intrinsic fluorescence did not provide good results as area in this case is 0.51 which indicates that the discriminatory power of the test is not good. The findings have been reported to the IIT team for further work on intrinsic fluorescence technology [23, 25].

The IIT team is working upon to improve the prototype of handheld probe further so that it may be conveniently used by the para-medical staff in the peripheral and rural areas, where screening for cancer cervix is the need of the hour. This will result in detection of cases which have early neoplastic changes (CIN I or CIN II) but have no symptoms. The early detection of neoplastic changes and timely referral to tertiary care centers will reduce female morbidity and mortality.

The handheld probe for fluorescence spectroscopy is an excellent discriminator of normal tissue for all types of precancerous and cancerous lesions and can play a very important role in reducing female morbidity and mortality.

Conclusion

Interest in fluorescence diagnostics has increased considerably in the last few years as a result of several factors: its availability, decreasing cost and clinical suitability of the required technology, development of new exogenous fluorophores, detailed understanding of autofluorescence, the need for better method of early tumor detection and commercial opportunities in a conscious health care environment [23].

The conclusion obtained by analyzing the various socio-demographic characteristics of case and the results of various diagnostic techniques are as follows:

1. Effective screening programme can lead to earlier detection of cancer and its precursors, thus leading to decline in morbidity and mortality.

2. The detection rate of premalignant and malignant lesions by vaginal cytology screening was found to be 58.42%, out of which 23.47% were CIN.

3. Increasing number of cases were found to be lower grade of lesion histologically. About 11.6% cases were of mild dysplasia. Therefore, an effective screening leads to detection and treatment of carcinoma cervix in precancerous stage.

4. Precancerous lesions are rare in nulliparous women. The severity of lesions increases with increasing parity, indicating that trauma and infections due to abortion and child birth predispose to cervical carcinoma.

5. Lower socioeconomic strata have propensity for higher-grade lesions of the cervix due to various factors like ignorance, poverty, poor hygiene, overcrowding, infections, etc.

6. Age at 1st coitus significantly affects the incidence of cervical lesions. Early sexual activity predisposes to precancerous lesions.

7. The grade of dysplasia increases with increasing age mean age for inflammatory lesions was found to be 30.47 ± 7.3 years, whereas for CIN1, CIN2, CIN3, carcinoma was 36.30 ± 8.8, 43.58 ± 6.6, 47.63 ± 10.9, 56.08 ± 6.8 years.

8. In vivo handheld probe based on polarized fluorescence spectroscopy is an excellent emerging screening technique for detection of cervical precancerous lesion and as good as in vitro method. Co- and cross-polarized light has shown enhanced accuracy [22].

9. Accuracy of co-minus cross-polarized light is seen to be poor.

The advantages of fluorescence spectroscopy over the conventional screening modalities are:

• No external dye or chemical is used, and natural fluorophores of the body itself are used

• It is a minimally invasive method of screening

• Real-time analysis by developed stastical tool

• User friendly

• Removes observer based error that is in case of VIA and VILLI_ (objective-based analysis)

• It can be operated by minimally training the existing health workers and has a small learning curve

• This probe can be a boon in a country like India with a huge population where mass screening is a cumbersome process.

• It is most cost-effective as the only cost involved is in production of handheld probe; later on, the costs are negligible.

With further developments, it has the potential to become a regular screening tool in future.

The diagnostic technique based on optical imaging has the potential to link the biochemical and morphologic properties of tissues to individual patient care. In particular, these techniques are fast, noninvasive and quantitative.

Abbreviations

ASR

Age specific ratio

PAP

Papanicolaou

VIA

Visual inspection with acetic acid

VILI

Visual insection with Lugol’s iodine

GSVM

Ganesh Shankar Vidhyarthi Memorial

IIT KANPUR

Indian Institute of Medical College

FAD

Flavin adenine dinucleotide

NAD

Nicotinamide adenine dinucleotide

AUC

Area under curve

INF

Inflammatory

LSIL

Low-grade squamous intraepithelial lesion

HSIL

High-grade squamous intraepithelial lesion

CA

Carcinoma

HPE

Histopathological examination

ROC curve

Receiver operating characteristic curve

Author’s Contribution

All authors meet these criteria: Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; and drafting the work or reviewing it critically for important intellectual content; and final approval of the version to be published; and agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding

This study was not funded by any individual or agency.

Declarations

Conflict of interest

They authors declare that they have no conflict of interest.

Ethical approval

Institutional ethical committee clearance was obtained for study. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Ethical clearance code: EC/178/JUNE/2022.

Informed consent

Informed consent was obtained from all individual participants included in the study.

Human or animal participants

The research involved human participants with due permission and consent.