Interpretation
In this study, the sensitivity and specificity of the conventional colposcopy were 54.2% and 60.8%, which was at a relatively low level compared with previous research6, 7. Most of the statistics in previous studies referred to the detection efficiency of CIN 2+ by colposcopy11, 16, 17, and our study included CIN 1 as a positive pathological result. The sensitivity of the conventional colposcopy to detect CIN 1 was only 16.7% (1/6) in our research. CIN 1 lesions were too mild to show the staining reactions of acetic acid and iodine. When CIN 1 was included in the statistics, the efficiency of colposcopy would generally be reduced. In addition, the diagnostic efficiency may also be related to the population we included. Previous studies had shown that HPV 16+ was related to clearer visual acetowhite changes in the epithelium, therefore reaching a better diagnostic efficiency for colposcopy18. Patients infected with any of the 16 high-risk types of HPV could be included in our study. The colposcopic impression also varied between different observers19, 20. The application of dynamic spectral imaging colposcopy and related intelligent algorithms would significantly improve the diagnostic efficiency of colposcopy in the future21, 22.
The main outcome measure for fluorescence colposcope was the level of SBR at the lesions. We set a cut-off value and defined the positive result as the SBR of the cervical sampling site was greater than the cut-off value. For the conventional colposcope, the positive results included the colposcope assessment impression of low grade, high grade and cancer. The two types of colposcope displayed similar efficiencies in identifying cervical precancerous lesions on per-patient basis. However, the fluorescence colposcope possessed higher sensitivity and better NPV on per-site basis. One of the major advantages of the fluorescence colposcope was its high sensitivity, and the minimum detectable concentration of the fluorescent dye was approximately 2.7 μg/mL8. Because it was too sensitive, some inflammatory lesions could also be recognized after being stained with fluorescent dyes, which leads to the prevalence of false positives. In addition, the cut-off value we set was different on the per-patient basis and the per-site basis. At present, this fluorescence colposcope is in the preliminary research stage, and the exact SBR reference value cannot be determined yet. And more detailed and standardized procedures also need continuous improvement.
In addition to the fluorescence signal at the lesions, we could sometimes see the arc-shaped non-specific fluorescence signals at the vaginal vault of the fluorescence colposcopy group, as shown by patient No. 68 in Figure 2A. Before fluorescence colposcopy, unbound fluorescent dyes need to be scrubbed off, and the vaginal vault area was prone to residual fluorescent dyes and resulted in non-specific fluorescent signals. Therefore, this fluorescence colposcope was not suitable for detecting vaginal intraepithelial neoplasia (VIN). Conventional colposcopy-guided biopsy with acetic acid and Lugol’s iodine was crucial in detecting VIN23, 24. In order to avoid missed diagnosis of VIN or vaginal cancer in this study, suspicious vaginal lesions were also sampled for biopsy. However, the SBRs of the vaginal lesions in the fluorescence colposcope group could not be accurately calculated, so the vaginal lesions in both groups were not included in the statistical analysis.
A variety of fluorescent probes targeting tumors at the molecular and cellular levels have been developed for image-guided surgery25-27. However, there are few relevant studies in colposcopy-guided biopsy. The fluorescence colposcope combined with tumor-specific dye TMTP1-PEG4-ICG provides a new idea for the identification of cervical cancer and precancerous lesions.