Visual inspection with acetic acid in the early detection of cervical cancer and precursors.

Abstract

Organized cytology screening programs are not feasible in many developing countries where four-fifths of the world burden of cervical cancer occurs. Among other possible approaches to prevention, the performance of unaided visual inspection (to detect lesions of the cervix) has been shown to be unsatisfactory, particularly in detecting pre-invasive lesions (Nene et al.,1997; Wesley et al.,1997; Sankaranarayanan et al.,1997). Unmagnified, naked eye inspection of the cervix, after the application of 3–4% acetic acid [termed visual inspection with acetic acid (VIA) or cervicoscopy] has been reported to detect lesions missed by cytology (Cecchini et al., 1993; Le et al., 1993). Aceto-whitening of cervical epithelium may reflect the presence of an abnormal transformation zone as a result of increased cellular density with increased abnormal nuclei and DNA content. Acetic acid application is routinely used in colposcopic examination to visualize abnormal epithelium. Here, we report a comparison of the performance of VIA and cervical cytology in detecting cervical lesions in a clinic-based study in Kerala, India. The study subjects were 1,351 women aged 22–70 years attending the Early Cancer Detection Centre (ECDC), Ernakulam, Kerala, India, during 1995–1997. These women presented themselves for a routine examination or were referred from elsewhere to rule out cervical pathology. Their mean age was 38.9 years (SD 7.3 years). Their educational levels were as follows: nil: 66 (4.9%), primary school: 305 (22.6%), middle school: 221 (16.4%), high school: 495 (36.6%), college and above: 260 (19.2%), not known: 4 (0.3%). They were subjected to speculum examination by a nurse trained in recognizing cervical abnormalities and aceto-white lesions after acetic acid application. The nurse initially recorded the findings of unaided visual inspection, and then took a cervical smear using Ayre's spatula. This was followed by application of 3–4% freshly prepared acetic acid using a cotton swab and the cervix was examined after 1–2 min under adequate light. If aceto-white areas were detected in the cervix, the test outcome was recorded as being positive. Cervical smears were examined by the cytopathologist (B.S.) of the ECDC, who is also an experienced colposcopist. Cytology was considered to be positive if it revealed any of the following lesions: atypia, mild dysplasia, severe dysplasia, carcinoma in situ and invasive cancer. Subjects with positive VIA or with a positive cervical smear were referred for colposcopy. In addition, women with grossly abnormal looking cervix (bleeding on touch, hypertrophied elongated irregular edematous cervix, stippled cervix, suspected growth/ulcer) but with negative VIA and cytology were also referred for colposcopy/biopsy, in order to increase the number of participants subjected to verification of their true disease status and thereby to reduce “verification bias”. Those with normal looking cervix, features of inflammation and squamous metaplasia on colposcopy were assumed to be false positives and were not subjected to biopsy. Directed biopsies were performed for those with colposcopic diagnosis of atypia or worse lesions. Those with normal findings or inflammation or squamous metaplasia or mild dysplasia on histological examination of biopsy specimens were also considered to be false positives. Those revealing moderate dysplasia or severe dysplasia or carcinoma in situ or invasive cancer on histology were considered to be true positives. Thus, negative colposcopy and the results of histological diagnosis of colposcopically guided biopsy were considered the gold standard to evaluate the performance of the tests. The detection rate of moderate dysplasia or worse lesions (on biopsy): this is calculated by dividing the number of screened women with these lesions by the total number of screened women. The ratio of sensitivities between the 2 tests: this is calculated by dividing the detection rate of VIA by the detection rate of cervical cytology; a ratio of more than unity indicates that the first test is more sensitive. McNemar's test applied to the discordant women was used for significance testing of the difference in sensitivities between VIA and cytology. The approximated specificity of each screening test: this is calculated by dividing the number of women with negative screening tests by the total number of screened subjects less the number of true positive cases detected by the test. McNemar's statistics applied to the discordant women was used to assess the statistical significance of difference in specificities. The index (100-specificity) indicates the proportion of screen positive women recalled unnecessarily for follow-up investigations. The positive predictive value (PPV): this is obtained by dividing the number of women with true positive screening tests by the total number of subjects with positive screening tests. The results of screening tests are given in Table I. There were 509 (37.7%) women (with aceto-white cervix) positive for VIA; 205 (15.2%) were positive on Pap smear. One hundred seven women had an abnormal looking cervix on direct visual inspection but with negative VIA and cytology. The 83 subjects who did not comply with referral for colposcopy were excluded, yielding a total of 1,268 women for further analysis. On colposcopy, 293 women did not reveal abnormality; 162 had inflammation; 44 had features of squamous metaplasia; the remaining 102 revealed colposcopic abnormalities and were subjected to biopsy; of these, 86 (84.3%) women were positive for VIA. On histology, 4 were normal; 2 had inflammation; 9 had squamous metaplasia; 3 had squamous metaplasia with atypia; and 13 had mild dysplasia. The remaining 71 had moderate dysplasia or worse lesions. The distribution of these by results of screening tests is given in Table I; 68/71 (95.8%) with these lesions were detected by VIA as opposed to 44/71 (62.0%) by cytology. VIA resulted in the detection of a higher proportion of mild and moderate dysplasia compared with Pap smear; both tests detected almost all cases of severe dysplasia or worse lesions among those subjects who had biopsy. VIA detected 25 lesions missed by cytology and cytology 1 lesion missed by VIA. One of the concerns may be the extent of true positives missed among subjects who were negative for both screening tests. We feel that it may be negligible, in view of the fact that both tests missed only 2 true positive cases among 107 subjects with negative screening tests but with clinically abnormal cervix who were subjected for reference investigations. The detection rate of moderate dysplasia or worse lesions by VIA was 53.6/1,000 women as opposed to 34.7/1,000 by cytology, yielding a ratio of sensitivities of 1.54 (χ2 = 20.3, p < 0.001). The approximated specificity of cervicoscopy was 68.0% and that of cytology was 89.5% (χ2 = 7.6, p < 0.01). The PPV of VIA was 14.8% (68/452) and that of cytology was 25.4% (44/173). In this clinic-based study, with a relatively high prevalence of cervical abnormalities, VIA demonstrated a higher detection rate of moderate dysplasia or worse lesions than cytology. However, it was significantly less specific than cytology and resulted in referral of more than one-third of women for colposcopy. However, its ability to detect pre-invasive lesions is of interest. Any alternative screening test considered in the developing country setting should be sensitive enough to detect pre-invasive lesions. It is easier to treat these lesions and they result in prevention of invasive cancer. Many developing countries do not have sufficient facilities to treat invasive cervical cancer (e.g., radiotherapy) and the outcome from advanced cervical cancer in these settings is not satisfactory (Nandakumar et al., 1995; Jayant et al., 1996; Sankaranarayanan et al., 1996, 1998). Facilities to manage pre-invasive lesions (colposcopy, LEEP, cryotherapy) can be established at much lower costs than investments for invasive cancer management. The ability of VIA to detect lesions missed by cytology has been reported by others (Cecchini et al., 1993; Le et al., 1993). In one study, cervicoscopy was found to be more sensitive than cytology in detecting lesions, but resulted in a recall of 25.4% of 2,105 subjects for further investigations, as opposed to 3.8% with cytology (Cecchini et al., 1993). In another study, 85 subjects with aceto-white lesions on the cervix and normal Pap smear were subjected to colposcopy; 34 of them had normal colposcopic appearance and the rest were subjected for biopsy and 13 cervical intraepithelial neoplasia (CIN) lesions were detected among those (Le et al., 1993). In a study involving 2,426 women in a suburb of Capetown, those positive on VIA or those with squamous intraepithelial lesion (SIL) on cytology were referred for colposcopy and biopsy (Megevand et al., 1996a). Of these, 61 were positive on VIA plus cytology; 15 were positive on VIA only; 254 were positive for cytology only; and 2,096 were negative for both VIA and cytology. Of the total 31 histologically detected high-grade SIL lesions in this study, 20 were found positive for both tests; 11 were found positive on cytology only. It was concluded that since VIA detected more than 60% of the high-grade SIL, it warrants consideration as an alternative to cytology in low resource settings. In a workshop, a review of preliminary or final results from several studies investigating the performance of VIA, with or without magnification, in detecting cervical neoplasia in low resource settings in Asia (India, Indonesia) and sub-Saharan Africa (Kenya, Zimbabwe, South Africa) suggested that VIA performs comparably to the Pap smear and/or other screening tests being investigated in those settings (Gaffikin et al., 1997). Sensitivity for VIA has consistently been measured at between 60 and 70% and specificity at approximately 70%. Although the high sensitivity of VIA is offset by lower specificity, the increased costs associated with more false positive referrals can be reduced if follow-up colposcopy is performed immediately (during the same visit), even by trained para-medical staff. The feasibility of offering colposcopy and large loop excision of the transformation zone under local anesthesia during the same visit, following a positive screening test, has been well demonstrated in South Africa (Megevand et al., 1996a, b). Furthermore, improved techniques (recognizing artifacts due to glare from the light source, wiping away extraneous mucus and secretions), and referral of those with dull aceto-white areas that cannot be wiped away and not those with faint and suspicious aceto-white lesions, may reduce false positives without compromising sensitivity. Rigorous training of providers in correctly recognizing aceto-white lesions may further improve specificity and reduce false positive referrals. These improvements may open up new opportunities for disease control such as one-stage testing with VIA and treatment with options such as cryotherapy/LEEP, keeping the proportion (false positives) treated unnecessarily as low as possible. In one of our recently concluded studies in another location in India, 10% of the participants were scored as having positive VIA (data not shown). The non-invasive nature and the easy applicability of the test coupled with the immediate availability of results facilitating colposcopy and treatment of pre-invasive lesions at the time of examination make VIA an attractive screening test. The usefulness of VIA, both in screening for cervical cancer in developing countries and as a case-finding tool in actual clinical practice settings, certainly merits further evaluation. R. Sankaranarayanan*, B. Shyamalakumary , R. Wesley , N. Sreedevi Amma , D.M. Parkin*, M. Krishnan Nair , * Unit of Descriptive Epidemiology, International Agency for Research on Cancer, Lyon, France, Early Cancer Detection Centre, Ernakulam, India, Regional Cancer Centre, Trivandrum, India