Correlation Between Colposcopic Patterns and Histological Grade of Vaginal Intraepithelial Neoplasia: A Retrospective Cohort Study

Discussion

Vaginal intraepithelial neoplasia is an uncommon HPV-related disease, accounting for approximately 3% of all gynaecologic malignancies (1). Although VaIN may lead to vaginal cancer through a process, which is similar to the one that leads to cervical cancer starting from CIN (4), it is still an underdiagnosed disease. This is probably due to the fact that VaIN is asymptomatic, and as such, it is usually recognized following a positive cervical smear and/or HPV test, or during the follow-up of previously treated lower genital tract diseases. Furthermore, the colposcopic examination of the vagina may be complicated by several limitations, including a still scant knowledge of vaginal colposcopic patterns, a wider examination field compared to the cervix, the fact that VaIN is often multifocal and that in hysterectomized patients it may be localized in the lateral “dog ears” of the vaginal vault. All together, these factors may reduce the reliability of the colposcopic examination of the vaginal walls (14, 16). Starting from such evidence, the aim of the present study was to evaluate the correlation between colposcopic findings and histologic diagnosis of VaIN. Although the literature regarding the diagnostic accuracy of colposcopy in cervical neoplasia is extensive (17), evidence regarding the accuracy of vaginal colposcopy is still scarce. Despite that, most authors report a poor concordance between colposcopic patterns and histological grade of vaginal intraepithelial lesions (3, 9, 18, 19). Our results seem to confirm such hypothesis, as we observed a poor correlation between colposcopy and histology when comparing women with LG-VaIN with those with HG-VaIN. In addition, our data regarding sensitivity, specificity, PPV, NPV and diagnostic accuracy (56.3%, 64.5%, 69%, 51.2%, 59.7%, respectively) were comparable to those of previous studies.

However, we found a greater concordance between colposcopic patterns and histology, and an increased diagnostic accuracy of colposcopy, when comparing women with a diagnosis of VaIN1 with those with VaIN3. This result may be explained by the ambiguous behavior of VaIN2, which is classified as a high-grade lesion despite being clinically similar to low-grade lesions (20). In fact, the scientific debate regarding the distinction of CIN 2 and CIN 3 in two separate clinical entities, and the need for specific management strategies and treatment types for each kind of high-grade lesion, is also applicable to VaIN2 and VaIN3. VaIN3 lesions are characterized by a greater risk of malignant transformation and may conceal an occult malignant disease more frequently compared to VaIN2 (21). For this reason, surgical excision is the treatment of choice for VaIN3, especially for lesions of the vaginal vault, whereas patients with a diagnosis of VaIN2 may be treated using ablative procedures, which are easier to perform and are more feasible in outpatient settings (21).

Moreover, currently there is no clear definition of microinvasive vaginal carcinoma (14, 22). Cervical microinvasive carcinoma is defined as a tumor in which the depth of cervical invasion is less than 3 mm. As such, its treatment may be conservative, especially in young women who desire a pregnancy in the near future. The fact that microinvasion has not been defined for the vagina makes the distinction between intraepithelial and invasive lesions evermore crucial. Recognition of VaIN3 is also crucial as these lesions entail a high risk of occult cancer and are at greater risk of neoplastic transformation (8, 21).

In our series, the rate of grade I colposcopic anomalies was similar among patients with VaIN1 (64.5%) and those with VaIN2 (60.7%), reinforcing the hypothesis of a greater similarity between VaIN1 and VaIN2 lesions than between VaIN2 and VaIN3 lesions. In fact, we obtained the highest k-value when VaIN2 lesions were excluded and in addition we found higher values of sensitivity (80.5%) and diagnostic accuracy (70.4%) of colposcopic examination, compared to the cases in which these abnormalities were included (56.3% and 59.7%, respectively).

Accordingly, Sopracordevole and co-workers observed grade II abnormal colposcopic patterns more frequently among women with VaIN3 than among those with VaIN1 and VaIN2. Moreover, among women with a grade I abnormal colposcopic pattern, the rate of VaIN3 was significantly higher when a vascular pattern (punctuation or mosaicisms) was observed (62.5 vs. 37.5%) (3). In a series of 467 patients, Qi Zhou and colleagues found a moderate agreement (69.16%; kappa=0.437, p<0.001) between colposcopy (evaluated using the 2011 International Federation of Cervical Pathology and Colposcopy terminology for the vagina) and pathology, with a 23.34% overestimation rate and a 7.49% underestimation rate. Agreement between colposcopy and pathology was lowest (35.71%) in the high-grade VaIN group (p<0.01) (19). These results reveal the need for more extensive clinical research in these areas. What is certain is that an accurate colposcopy should always include an examination of the vaginal walls and that a biopsy of all suspect lesions should be performed.

In Boonlikit and Noinual’s study, the upper third of the vagina was the most common site for VaIN (86.8%) and the most frequently observed colposcopic pattern was acetowhite epithelium (7). However, vaginal anatomy, along with the peculiar characteristics of the vaginal epithelium, may mislead colposcopists, making the establishment of acetowhite epithelium thickness and the detection of vascular lesions challenging (9, 19). For this reason, Indraccolo and Baldoni proposed a novel VaIN pattern classification based on simplified colposcopic terminology with the aim to improve the predictability of VAIN (9).

It is arguable that experience plays a crucial role in increasing the sensibility of vaginal colposcopy. Stuebs and co-workers reported that colposcopy-directed biopsies had an 80% sensitivity in detecting vaginal-HSIL when carried out by practitioners with more than 10 years experience in colposcopic examinations (18). In our center, colposcopic evaluation was performed by three gynecologists with at least 10 years’ experience in the lower genital tract.

Some authors have evaluated the role of Pap smears and HPV DNA tests in the detection of HG-VaIN (23-26). In the study of Sopracordevole et al., diagnosis of VaIN3 was preceded by the detection of major cytological abnormalities in most cases (72.7% vs. 27.3%). Moreover, major cytological abnormalities (HSIL and ASC-H) on referral Pap smears were significantly more frequent in postmenopausal women (64.9% vs. 36.7%) and in women with a previous diagnosis of an HPV-related cervical intraepithelial or invasive lesion (70.5% vs. 39.5%). However, minor lesions (ASCUS or LSIL) did not exclude HG-VaIN, especially VaIN2. The authors concluded that a biopsy of all suspicious areas detected during colposcopy is mandatory (23).

Conversely, when analyzing the accuracy of colposcopy-directed biopsies, Stuebs and co-workers found an 82.5% accuracy for the detection of HG-VaIN, with a 15.8% underdiagnosis rate and a 1.8% overdiagnosis rate. The vast majority of women with HG-VaIN had major cytological abnormalities and an HPV test positive for high-risk HPV. The one case of vaginal cancer also tested positive for high-risk HPV (18). Qing Cong and colleagues reported that the sensitivity of cytology for VaIN2/3 alone, concomitant with cervical or vulvar lesions and post-hysterectomy was 42.1%, 80.0%, and 80.8%, respectively. When comparing high-risk HPV testing with cytology/high-risk HPV co-testing for the detection of VaIN2/3 in patients with an intact uterus versus those who had previously undergone hysterectomy, sensitivity was 93.5% versus 92.3% and 92.0% versus 100%, respectively, meaning the sensitivity of cytology and high-risk HPV testing for VaIN was non-inferior to that of CIN2+ (24).

Alemany and co-workers specifically analyzed the prevalence of HPV in vaginal lesions and detected a74% prevalence in invasive cancers and a 96% positivity in high grade intraepithelial lesions. HPV16 was the most frequently type detected in both precancerous and cancerous lesions (59%) (25). Such evidence suggests that a reduction of the burden of vaginal neoplastic lesions may be observed in the coming years among vaccinated cohorts.

In our study, the proportion of women with VaIN and a positive referral Pap smear was significantly higher than that of women with a negative Pap smear (87.9% vs. 2.1%). Moreover, in most cases of HG-VaIN, the diagnosis was preceded by major cytological abnormalities.

Unfortunately, in our series, data about HPV status was not available in most of the cases. Together with its retrospective design, this is one of the main limitations of our study. The small sample size constitutes a further limitation, although it should be considered that HG-VaIN is a rare condition. Strengths include the fact that all patients were evaluated by the same three experienced colposcopists, reducing inter-observer variability.