Clinical determinants of a positive visual inspection after treatment with acetic acid for cervical cancer screening

Authors


Abstract

Objective

To examine the determinants of a positive visual inspection after acetic acid (VIA), including the relationship of testing positive for high-risk human papillomavirus (HR-HPV), which is the necessary cause of cervical cancer.

Design

A prospective cohort study.

Setting

Three clinical sites in rural China.

Population

A total of 7541 women aged 25–65 years.

Methods

All women underwent VIA, DNA testing, by two DNA tests performed on both clinician- and self-collected specimens, and HPV E6 oncoprotein testing. Those positive by any test underwent colposcopy and four-quadrant biopsy evaluation. A random sample of women with negative screening results also underwent colposcopy and, if colposcopic abnormalities were observed, four-quadrant biopsy evaluation was performed. Women diagnosed with cervical intraepithelial neoplasia grade 2 (CIN2), or more severe grades (CIN2 + ), underwent treatment.

Main outcome measure

Testing positive for VIA.

Results

Overall, 7.6% (95% confidence interval, 95% CI, 7.0–8.2%) had a positive VIA. Women who tested positive for HPV were more likely to have a positive VIA than women who tested negative for HPV (15.0%, 95% CI 12.9–17.2% versus 6.3%, 95% CI 5.7–6.9%; P < 0.001). Older women were less likely to have a positive VIA (Ptrend < 0.001), including women with CIN2 +  (Ptrend < 0.001). A logistic regression model demonstrated that diagnosis (CIN2 +  versus <CIN2; odds ratio, OR, 32; 95% CI 11–100), testing HPV positive with a higher viral load (highest versus lowest; OR 4.3; 95% CI 2.5–7.4), and age (51 years and older versus <38 years; OR 0.22; 95% CI 0.17–0.30) were independent determinants of having a positive VIA. VIA was more likely to be positive for women with CIN2 + having an abnormal colposcopic impression versus women with CIN2 +  regardless of colposcopic impression (71.4 versus 47.2%).

Conclusions

The age of the population and method of disease ascertainment should be considered in the interpretation of any VIA performance.

Introduction

Cervical cancer is one of the most common cancers and causes of cancer-related mortality in females globally. At least 85% of cervical cancer occurs in low- and middle-income countries (LMICs),[1] where effective screening programmes based upon the Papanicolaou (Pap) test have not yet been successfully established.[2, 3] Although prophylactic human papillomavirus (HPV) vaccination may prove to be the ultimate prevention strategy, current HPV vaccines do not treat pre-existing HPV infections and related disease.[4, 5] Therefore, two or three generations of women already exposed to HPV will not benefit greatly from HPV vaccination. For this reason, alternative strategies to Pap test-based screening need to be considered if the current and imminent burden of cervical cancer is to be reduced.

Visual inspection after treatment with acetic acid (VIA) is one method that is being used as an alternative in LMICs because of the minimal equipment required. Data from recent systematic reviews and multicentric trials show VIA to be moderately sensitive for precancerous lesions, as compared with Pap testing.[6-8] Two large trials have been conducted in India. One reported a 25% reduction in cervical cancer incidence and a 35% reduction in cervical cancer-related mortality,[9] whereas another reported no reduction in incidence but a non-significant 14% reduction in cervical cancer-related mortality from a single screen by VIA,[10] both being compared with a control group.

There are limited data on the determinants of testing positive for VIA, which can help to define the optimal population for VIA-based screening, and its relationship with high-risk human papillomavirus (HR-HPV), the necessary cause of cervical cancer.[11] As part of the Screening Technologies to Advance Rapid Testing for Cervical Cancer Prevention – Utility and Program Planning (START–UP) project, we conducted a clinical trial involving 7500 women living in rural China to examine different strategies for low-cost screening and triage of screen positives. All women were screened for HR-HPV DNA and VIA, allowing us to look at the relationship of HR-HPV DNA detection and semiquantitative viral load, the presence or absence of cervical precancer or cancer, and age in relationship to a positive VIA. We chose to focus on these three determinants a priori because: HR-HPV infections cause cervical cancer and its immediate precursor lesions; detecting the presence of precancer and cancer is the purpose of screening with VIA; and age influences where the lesions are located in the cervix. We believed that these three factors would greatly influence the rates of positive VIA. Our goal was to better understand who has a positive VIA and why.

Methods

Population

The population for this study was recruited from October 2010 to August 2011. First, we selected two high-risk communes from each of three counties (Yangcheng, Xinmi, and Tonggu) according to the proposed sample size of 7500. Second, the number of women aged 25–65 years in each commune was collected from the local police's residence registry. Third, we determined the candidate villages for the study considering the size of the village and its transportation conditions. Fourth, all of the women aged 25–65 years and living in the chosen villages were invited to participate in the study if they met the study criteria. The recruitment was stopped when the target sample size was reached. We noted challenges in the recruitment of the oldest and youngest women, as older women were less willing to undergo screening, and many of the younger women were transient and could not be located. Consequently, our study population was biased towards women who were 35–50 years old. The study targeted the enrollment of 7500 women to achieve a diagnosis of 100 cases of CIN3 +  (the actual number was 99). The a priori sample size of 100 CIN3 +  cases was chosen so that the precision over a range of sensitivities for CIN3 + , from 20 to 90%, would be ±10% or less.

Women aged 25–65 years of age were considered to be eligible if they: (1) had not been previously diagnosed with cervical cancer; (2) had a cervix; (3) were not pregnant; (4) were physically able to undergo routine cervical cancer screening; and (5) were able to provide informed consent. We did not exclude women if they had undergone previous cervical cancer screening because we assumed that even if a few women had been screened for cervical cancer, the quality of cytology screening was likely to be very poor. Women were excluded if they were not married and reported never having had sexual intercourse. Local doctors conducted the initial recruitment and eligibility screening. Eligibility was confirmed at the study clinic. Eligible women were then asked to complete the written, informed consent in order to participate in the study.

The study was approved by institutional review boards (IRBs) at PATH, Cancer Institute and Hospital, Chinese Academy of Medical Sciences (CICAMS), and the US National Cancer Institute. All participants provided written informed consent.

Enrollment

Participants were given an education session about cervical cancer prior to the start of the study procedures. First, women were asked to complete a short risk-factor survey administered by study personnel. Then, women were given instructions on how to collect a cervicovaginal specimen and were provided a private room for the self-collection of their specimen. Next, women underwent a routine pelvic exam, during which the VIA was performed and results recorded. Then two cervical specimens were collected: the first into a dry tube for the OncoE6™ (Arbor Vita Corporation) test and the second into dcm buffer (Qiagen) for HPV DNA testing.

The VIA testing was performed by local experienced practitioners from each clinical site. VIA was performed by inspecting the cervix with the naked eye using a vaginal speculum and a bright halogen focus lamp 1 minute after the application of 5% acetic acid. The results were reported by quadrant as per the International Agency for Research on Cancer (IARC) criteria.[12] VIA was considered negative when: (1) no acetowhite lesions were observed; (2) the lesion was ill-defined or scattered; or (3) acetowhite lesions were not in the squamocolumnar junction. VIA was recorded as positive when fixed, dense, acetowhite areas were present in the transformation zone near the squamocolumnar junction. VIA was judged suspicious for invasive cancer when there was a densely white ulceroproliferative growth on the cervix.

Clinical management

Women who tested positive for any of the six screening tests performed [VIA; HPV E6 16, 18, and 45 oncoprotein by Onco E6 Test (Arbor Vita Corporation, Fremont, CA, USA); and DNA testing for HR-HPV by Hybrid Capture 2 (HC2; Qiagen, Gaithersburg, MD, USA) and careHPV™ (Qiagen) on both clinician-collected and self-collected specimens] were referred to colposcopy and underwent a four-quadrant microbiopsy protocol, as previously described.[13] Also, a ~10% random sample of the women who tested negative for all screening tests underwent colposcopy. Colposcopy was performed by different senior practitioners trained by CICAMS. Colposcopists were blinded to the screening results. As dictated by the IRBs, women who had no visible lesions had their screening result revealed and, if they were screen negative by all tests, and had no visible lesions, no biopsies were taken.

Laboratory tests

The careHPV™ test was performed as previously described at the clinical sites by a laboratory technician who had a general level of training comparable with that of the local hospital staff, and was trained to administer careHPV™ by a senior CICAMS technician.[14, 15]

The HC2 was performed according to the manufacturer's instructions, except that 50 μL of the dcm specimen was combined with 25 μL of kit denaturation reagent rather than combining 1 mL of the specimen in specimen transport medium with 500 μL of kit denaturation reagent, as typically suggested in the manufacturer's specifications. Because HC2 cannot be used at local clinical sites, after careHPV™ testing, the residual dcm specimens were sent to the special lab on site and the CICAMS technician performed the HC2 test. HC2 testing using this research-only protocol was 96% sensitive and 87% specific for CIN2 + , which was comparable with previous reports of HC2 performance in China.[16]

Pathology

Two pathologists at CICAMS read the slides and came to an agreement through joint review on the diagnosis when there was disagreement.

Analysis

The Pearson chi-square test was used to test for differences in positive VIA by HR-HPV results and by histologic diagnosis. Binomial exact 95% confidence intervals (95% CIs) were calculated for point estimates of prevalence or positivity. Crude, unadjusted logistic regression was used to calculate the odds ratio (OR) and 95% CI as a measure of association between testing HR-HPV positive by either HR-HPV test and having a positive VIA. The Kruskal–Wallis test was used to test for differences in signal strength of a positive HR-HPV result as a semiquantitative measure of HPV viral load by VIA result.[17] We tested for trends in positive VIA across age groups (<38, 38–43, 44–50, and 51+ years) and careHPV™ status (careHPV™ negative; careHPV™ positive, and signal strength of 1 to <10 relative light units per positive control (rlu/pc); careHPV™ positive, and signal strength of 10 to <100 rlu/pc; and careHPV™ positive, and signal strength of ≥100 rlu/pc).[18] Finally, logistic regression models were used to calculate adjusted ORs and 95% CIs and to predict an estimated percentage of positive VIA by age group, careHPV™ status, and diagnosis (<CIN2 versus CIN2 + ).

Results

A total of 7543 women were recruited at three clinical sites in China, of which 7541 were eligible (two were ineligible because of age). The mean age, median age, and interquartile range of ages were 44.5, 44, and 38–51 years, respectively. The number of women enrolled from the Yangcheng, Xinmi, and Tonggu sites were 3241, 2498, and 1802, respectively. Most women reported that they were married (97.1%) and never smoked (99.6%). Of the 7541 women enrolled, 573 women (7.6%, 95% CI 7.0–8.2%) had a positive VIA. VIA was positive in 505 (6.8%) of the 7398 women with <CIN2. By comparison, 68 (47.2%) of the 144 women diagnosed with CIN2 +  were positive upon VIA (P < 0.001), and of the 84 colposcopically evident CIN2 +  cases, 60 were positive upon VIA (71.4%).

Table 1 shows the relationship of the VIA results to the HR-HPV DNA results by HC2 and careHPV™ tests. The percentage of positive VIA was 6.1% (95% CI 5.5–6.7%) for HC2 negatives and 16.0% (95% CI 13.9–18.2%) for HC2 positives (P < 0.001). Likewise, the percentage of positive VIA was 6.3% (95% CI 5.7–6.9%) for careHPV™ negatives and 15.0% (95% CI 12.9–17.3%) for careHPV™ positives (P < 0.001). Given the similarity in the relationship between either of the HPV DNA tests and positive VIA, and the correlation of signal strengths of the two HPV DNA tests (Spearman ρ = 0.84 among either careHPV™ or HC2 positives), suggesting that like HC2 signal strength, careHPV™ signal strength is a semiquantitative measure of HPV viral load,[17] subsequent analyses used careHPV™ results only.

Table 1. The relationship between positive VIA and the detection of cervical high-risk HPV by careHPV and HC2 for all women, women with <CIN2, and women CIN2 + 
 careHPVHC2
NegativePositiveNegativePositive
All women
n 6434110764071134
n (VIA+)407166392181
%VIA+6.3%15.0%6.1%16.0%
Women with <CIN2
n 64289696401996
n (VIA+)404101388117
%VIA+6.3%10.4%6.1%11.7%
Women with CIN2+
n 61386138
n (VIA+)365464
%VIA+50.0%47.1%66.7%46.4%

Categorising the careHPV™ results as careHPV™ negative or careHPV™ positive, with signal strength 1 to <10, 10 to <100, or ≥100 rlu/pc (‘careHPV™ status’; Figure 1), there was an increasing probability of VIA being positive from careHPV™ negative to careHPV™ with the highest signal strength, but only among women with a diagnosis <CIN2 (Ptrend < 0.001). Among women who were careHPV™-positive with a CIN2 +  diagnosis, the percentage of VIA-positive results actually decreased, albeit non-significantly, with higher signal strength category (Ptrend = 0.073).

Figure 1.

Relationship between positive visual inspections after acetic acid (VIA) and careHPV results, stratified by disease status.

There were also strong inverse relationships between age and the likelihood of VIA being positive for all women, women with less than a CIN2 diagnosis, and women with a CIN2 +  diagnosis (Ptrend < 0.001 for all; Figure 2). For women with CIN2 + , the percentage of positive VIA went from a high of 73.9% in women aged less than 38 years to 25.5% in women aged 51 years and older.

Figure 2.

Relationship between positive visual inspections after acetic acid (VIA) and age.

In a logistic regression model (Table 2), having a CIN2 +  diagnosis (versus <CIN2) and testing careHPV-positive with the highest signal strengths (10 to <100 and ≥100 rlu/pc), versus careHPV™-negative, were positively associated, and older age groups (versus <38 years of age) were negatively associated with positive VIA (model 1). Adding an interaction term between careHPV™ status and diagnosis improved the fit of the model (P = 0.0034; model 2), and we therefore included an interaction term in our models for all subsequent analyses.

Table 2. Logistic regression models as measure of association of careHPV™ status, diagnosis, and age group
  n %VIA+ORLLULORLLULORLLUL
  1. Model 1 did not include interaction terms. Model 2 included an interaction term of careHPV™ status and diagnosis. The P value for the likelihood ratio test for inclusion of the interaction (versus exclusion) was 0.0034. Model 3 only includes CIN2 +  diagnosed with a positive colposcopy (n = 84). Positive VIA (%VIA+) adjusted for the other variables and odds ratios (ORs) with 95% confidence interval (95% CIs) are presented.

Diagnosis
<CIN2 (ref.)73976.83%1.01.01.0
CIN2+14447.22%7.5422925.011.38047254.765425130828.1820
careHPV
careHPV– (ref.)64346.33%1.01.01.0
careHPV+ (1 to <10 rlu/pc)4577.44%1.1300050.781.6465940.940.611.4355471.00.701.6
careHPV+ (10 to <100 rlu/pc)39117.14%2.1512461.5483123.02.11.4712133.02.21.53.1
careHPV+ (≥100 rlu/pc)25925.10%3.1413352.2066474.4719364.02.7675115.7896373.72.65.3
Age group
<38 years (ref.)175811.77%1.01.01.0
38–43 years18278.76%0.690.550.870.690.550.860.700.560.88
44–50 years19387.07%0.520.410.660.520.410.660.530.420.67
51 years and older20183.42%0.210.160.290.220.160.290.220.170.30

In model 2, a positive VIA was also associated with testing careHPV™-positive with a signal strength of ≥100 rlu/pc (OR 4.3, 95% CI 2.5–7.4), and with a signal strength of 10 to <100 rlu/pc (OR 2.2, 95% CI 1.3–3.9), compared with a low signal strength of 1 to <10 rlu/pc, which was not significantly different from careHPV™-negative. Older age groups were increasingly less likely to have a positive VIA compared with women of <38 years of age, with women aged 51 years and older the least likely to have a positive VIA (OR 0.22, 95% CI 0.16–0.29). Having a CIN2 +  diagnosis (versus <CIN2 diagnosis) was most strongly associated with a positive VIA (OR 32, 95% CI 11–100). Diagnosis, careHPV™ status, and age were associated with positive VIA at each clinical site (data not shown).

Finally, we used the model to predict the percentage of positive VIA for the different subgroups defined by each unique combination of careHPV™ status and diagnosis for each age group (Figure 3). Women with CIN2 +  were more likely to be VIA positive than those with <CIN2. The percentage of women who were VIA positive declined with older age for every combination of signal strength and diagnosis, whereas it increased with increasing semiquantitative viral load, but only among those with <CIN2 diagnosis.

Figure 3.

CareHPV™ status- and diagnosis-adjusted VIA-positive results, by age group: <38 years (a), 38–43 years (b), 44–50 years (c), and 51 years and older (d).

Discussion

Main findings

Herein we present an analysis of the clinical determinants for positive VIA from a large population-based study conducted in China. Few studies have looked in detail at the relationships of HR-HPV detection, diagnosis, and age with positive VIA, despite the promotion of its use as a primary screening tool for cervical cancer in lower resource settings.[19, 20] Such investigations can provide important insights into when and why the VIA does and does not work, and can help to define populations in which the VIA test might perform best.

The VIA test was performed relatively accurately in our study, with much higher specificity but lower sensitivity for CIN2 +  than has been reported elsewhere.[6-8] Its lower sensitivity in this study may in part result from the better disease ascertainment in this study, including the referral to colposcopy of women who tested HR-HPV DNA positive by any combination of careHPV™ or HC2 testing on their clinician-collected or self-collected specimens, and rigorous assessment of disease in colposcopically negative women. VIA and colposcopic impression are highly correlated (in this study OR 9.3, 95% CI 7.6–11.0), which can lead to inflated estimates of VIA sensitivity.[21] To this point, the crude sensitivity of VIA for colposcopically evident CIN2 +  was 71.4%, which was more similar to previous reports,[6-8] and as shown in model 3 the association of positive VIA with colposcopically evident CIN2 + was greater than for any CIN2 + .

We found that diagnosis of CIN2 + , detection of HR-HPV DNA at a higher (semiquantitative) viral load, and age were all independent factors in determining whether a woman was scored as being VIA positive. Although the presence of a histologic CIN2 +  lesion increased the likelihood of being VIA positive, as expected and desired, we observed that increasing age decreased the likelihood of being VIA positive, even among those with a CIN2 +  lesion. Thus, we found that VIA was more sensitive and less specific at younger ages, and that the converse true at older ages.

Strengths and limitations

We found a complex relationship between semiquantitative HPV viral load and positive VIA. One limitation of our analysis was that we used a DNA test for a pool of HR-HPV genotypes, which is ideal for screening, but we were unable to determine the number of concurrent infections. Thus, the semiquantitative viral load measure might represent either the viral load contributed by a single infection or the combined contribution of more than one infection, which may in turn cause multiple acetowhite lesions; however, multiple infections are generally in the minority (~20–40% of all infections), compared with single infections. We hypothesise, but cannot address in this study, that high viral load by single HPV infections and multiple HPV infections would both increase the likelihood of being VIA positive.

We conducted our study in a population with a low prevalence of human immunodeficiency virus (HIV) and did not identify women who had HIV co-infection, so we cannot extrapolate our findings to this high-risk subgroup, for whom VIA is being recommended in lower resource settings. HIV-positive populations have a higher prevalence and higher viral loads of HPV than HIV-negative populations, so we would expect that the overall level of positive VIA would likewise be greater.[22, 23] Whether this predicted increase in VIA positivity would be restricted to HIV-positive women with <CIN2 is uncertain.

Finally, we had limited disease ascertainment in women who were negative for all six screening tests. Therefore, the estimates of association between VIA and CIN2 +  are biased towards visible lesions. The bias is likely to be small, as it is very unlikely that cases of CIN2 +  were negative for all six screening tests.

Interpretation

A few previous studies have specifically reported on the effects of age on VIA. One study performed in India found that positive VIA increased with age, whereas a multicentric study of VIA conducted in India[24] and Africa found that positive VIA declined slightly with age until the age of 65 years and older, the age group in which it was the most likely to be positive.[8] The same multicentric study reported that the sensitivity for CIN2 +  decreased with age.[8] Another study found that the fraction of adequate VIA evaluations declined with age.[25]

We cannot explain the differences in the age-specific performance of VIA between what was observed in this study and the aforementioned reports. We noted that there was a similar decrease in sensitivity and increase in specificity for CIN2 +  in colposcopy (based on colposcopic impression) with age, suggesting that this age-specific phenomenon might have been a characteristic of the population. Conversely, VIA was performed with high specificity but lower sensitivity in this study, compared with the other studies, and as a result may have clarified these relationships, obscured in the other studies in which VIA was performed with higher sensitivity but lower specificity. Given the subjective nature of VIA and differences in populations such as age, age-specific HPV prevalence, obscuring inflammation/cervicitis, history of screening, and prevalence of HIV co-infection, it may be difficult to compare between studies.

Conclusion

This is the first report on the relationship of semiquantitative HPV viral load and positive VIA, and we noted a complex relationship between the two, which needs to be replicated. The age of the population should be considered in the interpretation of any VIA results and measures of clinical performance, and age standardisation may need to be used to compare between studies of VIA. Likewise, the method of disease ascertainment will influence the performance characteristics of VIA, and must be taken into account in interpreting the results. Future studies should consider whether careHPV™ or another lower-cost, validated HPV test might be useful as a quality control/quality assurance measure for scaling up VIA, given its subjective nature and the possibility of significant inter-rater variability. Similar strategies have been proposed for quality assurance of cervical cytology.[26-28]

Disclosure of interests

P.E.C. has served as a paid consultant for BD, Roche, Cepheid, GE Healthcare, and Gen-Probe/Hologic. P.E.C. has received commercialised HPV tests and reagents for research at a reduced or at no cost from Qiagen, Roche, Norchip, and MTM. P.E.C. is a paid member of a data and safety monitoring board for clinical trials of HPV vaccines for Merck.

Contribution to authorship

P.E.C., J.J., and Y.L.Q. were responsible for the design of the study, and oversaw its conduct. P.E.C. designed and executed this analysis. Y.L.Q., F.H.Z., W.C., X.Z., L.N.K., P.B., and F.C. implemented the field study and acquired the primary data. M.V., P.B., P.P., and R.P. contributed to the cleaning of the data, the analysis, and the interpretation of the results of the analysis. All authors contributed to the drafting or revising of the article, and read and approved the final version.

Details of ethics approval

The study ran under continuing ethical approval by PATH (HS539) and CICAMS (10-19/391).

Funding

The Screening Technologies to Advance Rapid Testing for Cervical Cancer Prevention – Utility and Program Planning (START-UP) project is entirely funded by a grant from the Bill & Melinda Gates Foundation. The views expressed herein are solely those of the authors, and do not necessarily reflect the views of the Bill & Melinda Gates Foundation.

Acknowledgements

None.

Ancillary