Effective cervical neoplasia detection with a novel optical detection system: A randomized trial

Abstract

Objective.

To assess whether the use of a novel optical detection system (ODS) as an adjunct to colposcopy increases the detection of biopsy-confirmed CIN 2,3.

Methods.

This is a multicenter two-arm randomized trial comparing colposcopy alone with colposcopy plus a pre-commercial ODS system that utilizes fluorescence, white light tissue reflectance, and cervical video imaging. Patients were recruited from 13 colposcopy clinics in a variety of practice settings. 2299 women referred for the evaluation of an abnormal cervical cytology were randomized with stratification by cytology; subsequently 113 women were excluded for a variety of reasons. The main study outcomes were differences in true-positive rates (CIN 2,3 and cancer identified) and false-positive rates between the study arms.

Results.

The true-positive (TP) rates were 14.4% vs. 11.4% (p = 0.035, one-sided) for the combined colposcopy and ODS arm compared to colposcopy-only arm, respectively, in women with either an atypical squamous cell (ASC) or low-grade squamous intraepithelial lesion (LSIL) cytology result. TP rates were similar between the two arms among women referred for the evaluation of HSIL. The 26.5% gain in true-positives observed with the use of ODS and colposcopy among women referred for an ASC or LSIL cytology was achieved with only a fractional increase in number of biopsies obtained per patient (0.30) and a modest increase in false-positive rate (4%). In the combined colposcopy and ODS arm among women with ASC or LSIL, the PPV of biopsies indicated by ODS was 15.0% and the PPV of biopsies indicated by colposcopy was 15.2%. Joint hypothesis testing indicates that ODS and colposcopy provides benefit compared to colposcopy alone among women with ASC or LSIL.

Conclusions.

Combining ODS with colposcopy provides a clinically meaningful increase in the detection of CIN 2,3 in women referred for the evaluation of mildly abnormal cytology results.

Introduction

For the last three decades, colposcopic examination with cervical biopsy has been considered the standard of care for evaluating women with abnormal cervical cytology. However, two recent studies have shown that this approach has a much lower sensitivity than previously recognized. In the U.S. multicenter ASCUS/LSIL Triage Study (ALTS), a single colposcopic examination failed to detect 33% to 36% of women who were subsequently identified as having biopsy-confirmed high-grade cervical intraepithelial neoplasia (CIN 2,3) [1], [2]. Similarly, in a study from China colposcopy failed to detect 40% of CIN 2,3 lesions [3].

Quantitative optical spectroscopy and imaging offer considerable promise as an approach to improving the performance of colposcopy [4]. The optical methods currently being developed include intrinsic tissue fluorescence, white light backscatter, and video imaging techniques [4]. These methods exploit the biochemical and structural changes that occur within tissue during the pathogenesis of neoplasia to produce a probabilistic prediction of tissue type. Several studies that have utilized research prototypes have demonstrated that optical methods can be used to classify cervical tissue as being either normal or as having CIN [5], [6], [7], [8], [9], [10], [11]. Therefore, optical methods may offer an objective, reproducible result that may improve the sensitivity of colposcopy.

To assess the impact that an optical detection system (ODS) would have on the performance of colposcopy in clinical practice, we conducted a two-arm randomized controlled clinical trial comparing standard colposcopy with colposcopy assisted by ODS. The ODS evaluated in this study was a pre-commercial device that utilizes intrinsic fluorescence, white light backscattered diffuse reflectance spectroscopy, and cervical video imaging.