IJC International Journal of Cancer

Transcription

1 IJC International Journal of Cancer Accuracy of concurrent visual and cytology screening in detecting cervical cancer precursors in rural India Kedar Deodhar 1, Rengaswamy Sankaranarayanan 2, Kasturi Jayant 3, Jose Jeronimo 4, Ranjit Thorat 3, Sanjay Hingmire 3, Richard Muwonge 2, Aruna Chiwate 3, Rutha Deshpande 3, Dulhan Ajit 1, Rohini Kelkar 1, Bharat Rekhi 1, Irene Ruben 1, Sylla G. Malvi 3, Roshni Chinoy 1, Nirmala Jambhekar 1 and Bhagwan M. Nene 3 1 Tata Memorial Center, Tata Memorial Hospital and Cancer Research Institute, E. Borges Marg, Parel, Mumbai , India 2 Early Detection and Prevention Section and Screening Group, World Health Organization, International Agency for Research on Cancer (WHO-IARC), Lyon, France 3 Nargis Dutt Memorial Cancer Hospital (NDMCH), Barshi, Maharashtra, India 4 Program for Appropriate Technology in Health (PATH), Seattle, WA The high burden of cervical cancer and inadequate/suboptimal cytology screening in developing countries led to the evaluation of visual screening tests, like visual inspection with acetic acid (VIA) and Lugol s iodine (VILI). We describe the performance of VIA, VILI and cytology, carried out in a multinational project called Screening Technologies to Advance Rapid Testing in 5,519 women aged years, in detecting cervical intraepithelial neoplasia (CIN). VIA, VILI and cytology were positive in 16.9%, 15.6% and 6.1% women, respectively. We found 57 cases of CIN2, 55 of CIN3 and 12 of cervical cancer; 90% of CIN3 and 43% CIN2 cases were positive for p16 overexpression and high-risk HPV infection, indicating a high validity of histological diagnosis. The sensitivity of VIA, VILI and cytology to detect high-grade CIN were 64.5%, 64.5% and 67.7%, respectively; specificities were 84.2%, 85.5% and 95.4%. A high proportion of p16 positive CIN 3 (93.8%) and 2 (76.9%) were positive on cytology compared with visual tests (68.8% and 53.8%, respectively) indicating a higher sensitivity of cytology to detect p16 positive high-grade CIN. However, the immediate availability of the results from the visual tests permits diagnosis and/or treatment to be performed in the same sitting, which can potentially reduce loss to follow-up when women must be recalled following positive cytology. Organizing visual screening services in low-resource countries may facilitate the gradual building of an infrastructure committed to screening allowing the eventual introduction of more sensitive, highly objective, reproducible and affordable human papillomavirus screening tests in future. Key words: visual inspection with acetic acid, visual inspection with Lugol s iodine, cytology, cervical intraepithelial neoplasia Abbreviations: AGUS: atypical glandular cells of unknown significance; ASCUS: atypical squamous cells of unknown significance; CIN: cervical intraepithelial neoplasia; DNA: deoxyribonucleic acid; ECC: endocervical curettage; FHW: female health workers; HPV: human papillomavirus; HC: hybrid capture; IARC: International Agency for Research on Cancer; LEEP: loop electrosurgical excision procedure; NDMCH: Nargis Dutt Memorial Cancer Hospital; PCR: polymerase chain reaction; PATH: Program for Appropriate Technology in Health; SCJ: squamocolumnar junction; START: Screening Technologies to Advance Rapid Testing; TMC: Tata Memorial Centre; TZ: transformation zone; VIA: visual inspection with acetic acid; VILI: visual inspection with Lugol s iodine; WHO: World Health Organization Grant sponsor: Program for Appropriate Technology in Health (PATH), Seattle, WA DOI: /ijc History: Received 8 Dec 2011; Accepted 17 Apr 2012; Online 14 May 2012 Correspondence to: Rengaswamy Sankaranarayanan, Head, Early Detection & Prevention Section (EDP) and Head, Screening Group (SCR), International Agency for Research on Cancer (WHO-IARC), 150 cours Albert Thomas, Lyon Cedex 08, France, Tel.: þ , Fax: þ , sankarr@iarc.fr Early detection of high-grade cervical intraepithelial neoplasia (CIN 2 3 lesions) by screening and their effective treatment constitute the most common and widely used strategy to prevent cervical cancer throughout the world. Conventional cervical cytology is the most widely used screening test and the workhorse of large-scale cervical screening programs globally. Cytology is repeated frequently, at 1 5 year intervals, to ensure lesions missed in a given round of screening and incident lesions are detected in subsequent rounds. Collection of cervical cells, smear preparation, processing, reading and reporting requires several labor intensive steps as well as a laboratory infrastructure. Stringent quality assurance is vital to avoid any suboptimal performance of cytology in detecting cervical cancer precursors. The difficulties in ensuring optimal cytology screening with inadequate coverage for both testing and treatment of precursor lesions or lack of screening per se are responsible for the continuing high risk and burden of cervical cancer in developing countries. 1 4 The need for affordable and accurate alternative tests for use in low-resource health services has led to the increasing use and evaluation of naked eye visual inspection methods relying on visible changes on the cervix following the application of 4 5% acetic acid (VIA) or Lugol s iodine (VILI) to screen for cervical lesions. Int. J. Cancer: 131, E954 E962 (2012) VC 2012 UICC

2 Deodhar et al. E955 We used cytology, VIA and VILI to screen women aged years in Solapur district, India, as part of a multinational project called Screening Technologies to Advance Rapid Testing (START) project sponsored by the Program for Appropriate Technology in Health (PATH), Seattle, WA. The goal of the START project was to develop simple, rapid, accurate, safe, acceptable and inexpensive biochemical tests that can be used in cervical cancer screening programs in low-resource settings. Field studies were organized in India and China for the START project to assist in the development and evaluation of the new tests. Specimens from women diagnosed with CIN in the field studies were used to research and develop, verify and validate new biochemical tests that detect the presence of CIN. These efforts resulted in the development of the carehpv test that was found to be a promising primary screening method for cervical cancer prevention in low-resource regions, having very similar performance characteristics as that of hybrid capture (HC) II in detecting cervical neoplasia. 5 We describe the performance of cytology and visual screening in detecting high-grade CIN in the Indian field studies as part of the START project in this manuscript. We investigated the biopsies for p16 INK4a expression and for human papillomavirus (HPV) infection to validate histopathology. Diffuse p16 INK4a overexpression in basal and parabasal cells of cervical epithelium is a hallmark of HPV-mediated transformation and diffuse expression of p16 INK4a reflects lesions with proliferation-competent cells, while p16 INK4a -expressing cells associated with focal expression patterns are cell cycle arrested. 6 The Indian field studies were jointly organized by the Nargis Dutt Memorial Cancer Hospital (NDMCH), Barshi and the Tata Memorial Centre (TMC), Mumbai in collaboration with the International Agency for Research on Cancer (IARC), Lyon, France. The START study proposal was reviewed and approved by the institutional review boards and ethics committees of TMC, IARC and PATH. Materials and Methods Study design We used a cross-sectional study design to evaluate the accuracy of the screening tests. All women participating in the study were tested sequentially with cytology, VIA and VILI by independent providers and investigated by colposcopy, blind to the findings of the screening tests; biopsies were taken in women with abnormal or suspicious findings on colposcopy. Eligible women Eligible women comprised of apparently healthy women with no debilitating disease aged years, ever married, with no clinical suspicion of pregnancy, able to give informed consent, with no previous history of CIN, cervical cancer or hysterectomy, and who were not menstruating at the time of the study visit. They were recruited from 31 villages in two subdistricts of Solapur district, Maharashtra, Western India, between May 29, 2006 and April 1, Female health workers (FHW), employed by the project, visited each household in the villages and explained the study to the potential participants. Recruitment of eligible women was carried out after the informed consent was read to and signed by the participants. Consenting women were given an appointment to visit the study screening clinic. Women complying with the invitation and attending the clinic were given additional information about the study, screening and diagnostic investigations, and possible treatment involved. A structured questionnaire was then used to collect information on sociodemographic, reproductive and behavioral characteristics for each woman before the clinical study procedures. After the interview, screening tests and diagnostic investigations were carried out. Training and quality control The FHWs, VIA and VILI test providers, cytotechnicians involved in processing and reporting of cytology smears, histopathology technicians involved in processing biopsy and excision specimens, colposcopists, pathologists and data entry personnel in this project were re-trained and re-oriented at the beginning of the study and thereafter every 6 months to maintain high standards. They were familiar with all the study procedures, as they were involved in a previous large cluster randomized control trial that involved recruitment and screening of about 130,000 women. 7 IARC manuals on visual inspection, 8 and colposcopy and management of precursor lesions 9 were used to retrain visual inspection providers and colposcopists. Collection of cervical and vaginal specimens for research purposes Before collecting specimens for cytology and doing visual screening, three vaginal and three cervical sample specimens were obtained by a nurse using plastic-shafted swabs/brushes stored within half an hour at 80 C to facilitate the development of new biochemical tests as part of the START project. Specimens of all women with histological CIN and invasive cancer and a random sample from women with normal findings were shipped on dry ice to PATH to facilitate the development of a new rapid batch test and a rapid strip test for HPV screening in low-resource settings. Screening, diagnostic procedures and treatment Following the collection of the specimens for the START project, cervical cells were collected by the project nurse using the Rovers Cervex Brush VR and immediately smeared onto glass slides which were fixed in 95% alcohol. Then VIA was performed (as described in the IARC manual 8 ) by the same nurse who collected the research and cytology specimens and recorded the findings as negative or positive. VIA was scored positive when a well-defined, dense acetowhite

3 E956 Accuracy of screening tests in detecting precancer lesions Table 1. Criteria for taking biopsy as part of reference diagnostic investigations at screening and call back Criteria Type of specimen Colposcopic abnormalities suggestive of CIN or cancer Colposcopy-guided biopsy Clinician in doubt and cannot decide on grade of colposcopy diagnosis Colposcopy-guided biopsy Colposcopy biopsy found to be inadequate Repeat colposcopy-guided biopsy Colposcopy normal, but cytology reported as high-grade lesion or cancer Diagnostic LEEP biopsy Clinical growth Punch Biopsy Colposcopy unsatisfactory ECC and biopsy ECC inconclusive Diagnostic LEEP biopsy Colposcopic abnormalities suggestive of CIN or cancer, but lesion going into the endocervical canal Diagnostic LEEP biopsy Final cytology diagnosis is ASCUS or LSIL and colposcopy is abnormal at 12 month call back Colposcopy-guided biopsy Final cytology diagnosis is high-grade lesion and colposcopy is abnormal at 3-month call back Colposcopy-guided biopsy Abbreviations: CIN: cervical intraepithelial neoplasia; LEEP: loop electrosurgical excision procedure; ECC: endocervical curettage; ASCUS: atypical squamous cells of undetermined significance; LSIL: low-grade squamous intraepithelial neoplasia. area appeared 1 min after application of 5% acetic acid touching the squamocolumnar junction (SCJ) in the transformation zone (TZ). 8 The colposcopist, blinded to VIA test result, then performed colposcopy and recorded the findings after application of acetic acid and Lugol s iodine in terms of normal, inflammation, probable low-grade lesion, probable high-grade lesion, probable invasive cancer, cervical cancer or inconclusive. 9 Eight nurses were involved in providing the visual tests. After the colposcopy procedure, a second nurse, blinded to the findings from VIA and colposcopy, performed VILI and recorded the findings as negative or positive. A positive VILI finding was obtained when dense, thick, bright, mustard- or saffron-yellow areas were found on the cervix close to the SCJ in the TZ. 8 After reporting the VILI findings, women with colposcopic abnormalities suggestive of probable CIN or probable invasive cancer had multiple (two or more) punch biopsies directed. Those with inconclusive and/or unsatisfactory colposcopic findings were offered endocervical curettage (ECC). The different criteria for obtaining biopsy specimens in the study are given in Table 1. A diagnostic loop electrosurgical excision procedure (LEEP) was performed when colposcopy was normal, but cytology was reported as high-grade lesion or cancer; when ECC was reported as inadequate in cases with inconclusive or unsatisfactory colposcopy. A large amount of lesion was excised by LEEP when colposcopic abnormalities suggested high grade CIN or preclinical microinvasive cancer, with the lesion extending into the endocervical canal, to have a large piece of tissue to exclude invasive cancer. Although a more invasive procedure, we preferred a diagnostic LEEP over a four-quadrant biopsy 10 with ECC in the above instances in view of the higher possibility of a definite diagnosis and the associated therapeutic advantage of contributing to treatment in the same time when the lesions have been completely removed. The cytology specimens were stained and read by the study cytotechnicians; all the positive cytology smears and a 5% sample of negative smears were examined by the pathologists at NDMCH and TMH. The Bethesda system 11 was used to categorize the cytology results. A positive cytology finding was considered at the atypical squamous cells of unknown significance (ASCUS) or atypical glandular cells of unknown significance (AGUS) threshold. The directed punch biopsy, ECC and excisional specimens were immediately fixed in 10% neutral buffered formalin, transported, processed and read at NDMCH and TMH. The CIN terminology was used to categorize the results of histopathology. 12 All cytology, biopsy, ECC, excisional specimen slides were read by more than one cytotechnologist and pathologist at NDMCH, Barshi and TMH, Mumbai and discordant specimens were read by a third pathologist at TMH to reach consensus. Pathologists were blinded to the findings from the screening tests and cytology and histology findings from the other pathologists. The final cytology and histology readings were obtained after a rigorous procedure where two readers had to agree on the diagnosis, failing which, a tie breaker s reading from a third Indian pathologist was considered and a consensus decision was taken. Women with findings suggestive of probable high-grade lesions on colposcopy and/or histology were offered treatment with cryotherapy, if the lesion covered less than 75% of the TZ. Women with CIN 2 and 3 lesions, not appropriate for cryotherapy, were offered treatment with LEEP. Women with invasive cancer were referred to NDMCH for further investigations and management. HPV DNA extraction and HPV genotyping HPV deoxyribonucleic acid (DNA) testing by polymerase chain reaction (PCR) and levels of p16 INK4a staining, as a surrogate marker of high-risk HPV infections (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68), were evaluated from biopsy paraffin blocks pertaining to all CIN cases and invasive cancer, as well as a sample of normal cases at the laboratories of the Infections and Cancer Biology Group, IARC, Lyon. Increased immunoexpression of this marker has

4 Deodhar et al. E957 been observed with worsening grades of CIN and invasive cancer. 13 Three sections of 10 lm were cut from each paraffin block. We avoided the risk of crosscontamination between different specimens during the cutting by frequently changing the blades and washing the microtome extensively with DNA away (Dutscher, Brumath). Possible crosscontamination was also monitored by using and blindly analyzing empty paraffin blocks every ten study specimens. No contamination was detected throughout the study. To prepare DNA, paraffin tissue sections were incubated the in digestion buffer (10 mm Tris/HCl ph 7.4, proteinase K 0.5 mg/ml and Tween %) overnight at 37 C. Samples were then incubated for 10 min at 95 C, rapidly centrifuged for 2 min at 13,000 rpm in a bench centrifuge, and chilled on ice, to inactivate the proteinase K and to remove the paraffin. Finally, the aqueous phase was transferred to a new tube. HPV typing was performed using the multiplex PCR/ APEX assay, 14 that detects the DNA of 19 different HPV types, i.e., 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 70, 73 and 82. Two primers for the amplification of b-globin were added to provide a positive control for the quality of the template DNA. 15 Immunohistochemistry for p16 INK4a Immunohistochemistry for p16 was performed according to the protocol provided with the CINtec Histology p16 INK4a Kit (9511, mtmlabs) for the qualitative detection of the p16 INK4a antigen on slides prepared from formalin-fixed, paraffin-embedded cervical biopsies. The kit contains a collection of ready-to-use reagents optimized for immunohistochemical staining of p16 INK4a. Briefly, slides were deparaffinized and then rehydrated before the staining procedure was performed. For the antigen retrieval process, slides were placed in a preheated Epitope Retrieval Solution ph9 (95 99 C) and maintained at this temperature for 10 min followed by a 20-min cool-down period at room temperature. After a brief rinse in distilled water, slides were immersed in Wash Buffer in preparation for immunostaining. Specimens were blocked with a peroxidase-blocking reagent for 5 min, followed by 30 min incubation with p16 INK4a antibody (mouse anti-human). Slides were then incubated for 30 min with the visualization reagent, followed by 5-min incubation with DAB. Slides were subsequently counter-stained with hematoxylin, dehydrated, mounted with permanent mounting medium, and coverslipped. All steps, with the exception of the epitope retrieval, were performed at room temperature. Immunoreactivity was visualized by light microscopy. The staining pattern for p16 INK4a was coded in each lesion as follows: 1, negative staining, no basal p16 INK4a staining; 2, weak and/or rare basal p16 INK4a staining; 3, diffuse and moderately positive with only part of the cervix epithelium stained; 4, strong and full thickness staining of p16 in the cervix epithelium. Codes 3 and 4 were taken as positive results. Definition of the final disease status on histopathology The final disease status on histology of an individual was based on taking the individual s highest histology result within 12 months (with a 3-month window period) after screening, or negative histology result or negative colposcopy result for those missing the previous two criteria. All women with abnormal colposcopy who did not accept that a biopsy be taken (n ¼ 29) were excluded from this analysis. CIN 2 or worse lesions diagnosed by histopathology based on colposcopically directed biopsies or excision specimens constituted the true disease; true negatives were those with no evidence of disease on colposcopy, colposcopically directed biopsy, excision specimens or histologically diagnosed CIN 1 cases. Data management and statistical analysis Double data entry was employed to maintain accuracy of the data entered using Access 2000 software. Statistical analysis was carried out using STATA 11 software. Study participant characteristics and screening test positivity rates were presented as numbers and proportions. Detection rates of CIN were presented as the proportion of screen-positive CIN detected among the screened women. Standard formulas were used to obtain the screening tests accuracy parameters in terms of sensitivity, specificity, positive predictive value and negative predictive value. To compare proportions, tests on equality of proportions using large-sample statistics were carried out. These statistics include the test of the difference between two proportions using a normally distributed test statistic for proportions. 16 Results Participants Number of women identified, screened with different screening tests, excluded from the final analysis and positive on screening tests are given in Figure 1. A total of 8,711 eligible women aged years were identified from 31 villages in two subdistricts of Solapur district, Maharashtra, Western India, by household visits and 5,650 women were screened between May 29, 2006 and April 1, Among the screened women, 3,219 had no formal education, 2,132 had more than three children, 4,976 had postpartum sterilization as a contraceptive method and 891 were post-menopausal women. Screening test results, reference standard investigations The number of women screened negative and positive on the different screening tests and the distribution of final true status are given in Table 2. Of the 5,519 women eligible for final analysis, VIA was positive in 931 (16.9%) and VILI was positive in 859 (15.6%); of the 5,440 women eligible for final analysis, cytology was positive (ASCUS or worse lesions) in 330 (6.1%); of the 5,519 eligible women who had colposcopy, 4,724 had no colposcopically detectable abnormalities and no

5 E958 Accuracy of screening tests in detecting precancer lesions Figure 1. Screening process in the study. Table 2. Final disease status by screening test result Final disease VIA VILI Cytology Status Negative Positive Total Negative Positive Total Negative Positive Total Normal 4, ,348 4, ,348 5, ,269 CIN CIN CIN Invasive cancer Total 4, ,519 4, ,519 5, ,440 Abbreviations: VIA: visual inspection with acetic acid; VILI: visual inspection with Lugol s iodine; CIN: cervical intraepithelial neoplasia. biopsies were directed from them; 795 women had abnormal colposcopy findings and multiple punch biopsies or diagnostic loop excisions were carried out on them to establish the histological results of the tissue samples. Final disease status was classified as normal in 5,348 women, CIN 1 in 47 women, CIN 2 in 57 women, CIN 3 in 55 women and cancer in 12 women. VIA and VILI were associated with higher detection of CIN 1 lesions as compared to cytology; on the other hand, the detection of CIN 2 3 lesions were similar for all three tests (data not shown). Sensitivity and specificity of screening tests to detect CIN 2 or worse lesions The sensitivity, specificity and predictive values of screening tests and their 95% confidence intervals to detect CIN 2 or worse lesions among all the participants and by age group (30 39 and years) are given in Table 3. The sensitivity of VIA, VILI and cytology were 64.5, 64.5 and 67.7%, respectively; the corresponding specificities were 84.2, 85.5 and 95.4%, respectively. All three tests had similar sensitivity (VIA vs. VILI, p ¼ 1.000; VIA vs. cytology, p ¼ 0.591; VILI vs. cytology p ¼ 0.591), but cytology had a significantly higher specificity and positive predictive value than visual tests (p < 0.001). Both VIA and VILI tests had similar test characteristics to detect disease. The visual tests missed more CIN 3 cases than cytology, whereas the reverse was true for the CIN 2 disease. Although the three screening tests had lower sensitivity in detecting lesions among women aged years than in those aged years, the difference did not reach statistical significance (VIA, p ¼ 0.140; VILI, p ¼ 0.140; cytology, p ¼ 0.207); however, they were associated with significantly higher specificity in those aged years (VIA, p < 0.001; VILI, p < 0.001; cytology, p ¼ 0.05). Sensitivity and specificity of screening tests to detect CIN 3 lesions CIN 3, the most histologically reproducible of the CIN categories, is considered to be a more valid and true precursor

6 Deodhar et al. E959 Table 3. Accuracy of screening tests to detect CIN 2 or worse lesions Screening test Sensitivity (%) (95% CI) Specificity (%) (95% CI) PPV % (95% CI) NPV % (95% CI) Overall VIA 64.5 ( ) 84.2 ( ) 8.6 ( ) 99.0 ( ) VILI 64.5 ( ) 85.5 ( ) 9.3 ( ) 99.1 ( ) Cytology 67.7 ( ) 95.4 ( ) 25.5 ( ) 99.2 ( ) Aged VIA 69.1 ( ) 81.1 ( ) 8.3 ( ) 99.1 ( ) VILI 69.1 ( ) 83.5 ( ) 9.4 ( ) 99.1 ( ) Cytology 71.6 ( ) 95.8 ( ) 30.2 ( ) 99.3 ( ) Aged VIA 55.8 ( ) 89.0 ( ) 9.3 ( ) 99.0 ( ) VILI 55.8 ( ) 88.6 ( ) 9.0 ( ) 99.0 ( ) Cytology 60.5 ( ) 94.7 ( ) 18.8 ( ) 99.2 ( ) Abbreviations: CIN: cervical intraepithelial neoplasia; PPV: positive predictive value; NPV: negative predictive value; CI: confidence interval; VIA: visual inspection with acetic acid; VILI: visual inspection with Lugol s iodine. lesion for cervical cancer. 17 The sensitivity of cytology, VIA and VILI were 93.8% (95% CI: %) 68.8% (95% CI: %) and 68.8% (95% CI: %), respectively; the corresponding specificities were 94.8% (95% CI: %), 83.6% (95% CI: %) and 84.9% (95% CI: %), respectively. Sensitivity of screening tests to detect high-risk HPV positive and/or p16 positive CIN The distribution of p16 overexpression among CIN and cancer cases and the number and proportion of p16 positive cases with positive screening test results are given in Table 4. The distribution both high-risk HPV infection and p16 overexpression among the CIN and invasive cancer cases are given in Table 5. Of the 51 women with CIN 3 lesions who had immunohistochemistry, 48 (91.4%) cases were positive for p16 overexpression and 46 (90.2%) were positive both for p16 and for high-risk HPV infection (Table 5), indicating the high validity of the histological diagnosis of CIN 3 in our study. A high proportion of CIN 3 (93.8%) and CIN 2 (76.9%) cases showing p16 overexpression were positive on cytology; half of the CIN 2 cases with p16 overexpression were positive on visual tests, whereas 68.8% of CIN 3 cases with p16 overexpression were VIA or VILI positive (Table 4). VIA and VILI had relatively similar sensitivity, while cytology had a higher sensitivity than the visual tests to detect highgrade CIN that were positive for p16 overexpression. Treatment of CIN and cancer Most of the women (88%) detected with CIN 2 or worse lesions received treatment; of the 112 women with CIN 2 3 lesions, 99 (88%) received treatment (98 received LEEP and 1 received LEEP and hysterectomy). No adverse events, such as bleeding, severe pain, vasovagal manifestations, such as fainting attacks and anaphylactic reactions, and hospitalization, as a consequence of complications of intervention, were observed as a consequence of screening, diagnosis and treatment in the study. Discussion There has been a great deal of discussion and introspection in recent years on how effective and wide spread screening services can be established and coverage improved in lowand medium-income countries with a high incidence of cervical cancer. A major emphasis has been to find out suitable alternative approaches to cytology screening, which has been the workhorse for cervical screening for the last six decades. Although cytology screening has been effective in reducing cervical cancer incidence in many high-income countries, its successful implementation requires a variety of requirements to be fulfilled, which are not feasible in many low-resource settings where a high risk of cervical cancer is experienced. For instance, cytology screening requires a laboratory infrastructure, microscopes, several resource personnel (smear collectors, cytotechnicians and pathologists), consumables (slides, fixative, Pap stain containing five dyes and three solutions) and several steps with inbuilt quality assurance procedures. Moreover, cytology must be repeated at frequent 3- to 5-year intervals to ensure satisfactory sensitivity and optimum detection of cervical cancer precursor lesions and repeat visits are necessary after a positive cytology for diagnosis and treatment, which may lead to drop outs. Several years of cytology screening in low- and middle-income countries have not led to significant reductions in cervical cancer in these countries, possibly due to the difficulties in offering high-quality cytology and inadequate coverage of screen-positive women with diagnosis and treatment. New approaches to cervical cancer screening have been developed and evaluated in a variety of settings in this context and one of the aims of our current study was to evaluate the test

7 E960 Accuracy of screening tests in detecting precancer lesions Table 4. Screening test positivity among the p16 positive cases by histology report Histology No. of cases No. positive for p16 No. positive on VIA (%) No. positive on VILI (%) No. positive on cytology (%) Report Normal (28.6) 2 (28.6) 3 (42.9) CIN (66.7) 7 (77.8) 3 (33.3) CIN (53.8) 14 (53.8) 20 (76.9) CIN (68.8) 33 (68.8) 45 (93.8) Invasive cancer (100.0) 10 (100.0) 10 (100.0) Total (65.0) 66 (66.0) 81 (81.0) Table 5. Distribution of high-risk HPV PCR and p16 results for the different histology reports Histology report Invasive HrHPV PCR and P16 results Normal CIN 1 CIN 2 CIN 3 cancer Total HrHPV PCRþ, p16þ HrHPV PCR, p16þ HrHPV PCRþ, p HrHPV PCR, p HrHPV PCRþ, p16 not done HrHPV PCR not done, p16 not done Total characteristics of concurrent cytology and visual screening tests to detect high-grade CIN diagnosed by high-quality reference diagnostic standards. The screening tests were evaluated under stringent and well-controlled settings with quality assurance for both screening and reference standard investigations and with minimal verification bias, facilitating precise estimation of the diagnostic properties of both a single visual and cytology test in our study. Verification bias has been minimized considerably due to the fact that all women had colposcopy assessment and all those with colposcopic abnormalities had multiple directed biopsies or diagnostic LEEP. Our ethical review did not permit performing biopsies in women with normal colposcopic findings and hence we did not resort to four quadrant biopsies with ECC in all screened women as has been done in cross-sectional studies in China. 5,10 Although we may have missed a few women with lesions among those with normal colposcopy, this is likely to be negligible and we considered all women with normal colposcopic findings without biopsies as true negatives for all pragmatic purposes. Although our results indicate that the visual tests have similar sensitivity to that of cytology, albeit with significantly lower specificity and positive predictive value, for detecting histologically confirmed high-grade CIN and invasive cancer when concurrently evaluated in research settings as observed in cross-sectional studies in the past, 18,19 cytology had a higher sensitivity than the visual tests to detect high-grade CIN that were positive for p16 INK4a overexpression and/or high-risk HPV types. The specificity of VIA and VILI were low, compared to that of cytology. The high false-positive tests associated with visual tests leads to the possibility of unnecessary diagnostic work up (e.g., colposcopy/biopsy), or treatment (e.g., cryotherapy), if all screen-positive women without clinical evidence of invasive cancer are treated without diagnostic triage in a single-visit screen and treat approach. For instance, only 8% of VIA-positive women had CIN 2 or 3 lesions in our study, so a screen and treat policy would entail more than 90% of VIA-positive women receiving unnecessary treatment, which has considerable resource and logistic implication. It is not clear if specificity can be improved without substantial loss in sensitivity, by standardizing reporting categories and training strategies. On the other hand, the immediate availability of result after a visual test permits diagnostic procedures (colposcopy with or without biopsy) and/or treatment to be performed at the time of the screening visit leading to reduced loss to follow-up. Almost all information on the test performance of visual tests come from clinical research settings and there is currently limited information on how these tests will perform when introduced for wide-spread routine use in real-life settings in public health services and in large programmatic contexts. Overexpression of p16 INK4a has been observed when retinoblastoma protein is inactivated by high-risk HPV oncoprotein E7. Although p16 overexpression is not usually found

8 Deodhar et al. E961 in normal cervical epithelium 20,21 and p16 INK4a -negative CINs and carcinomas do exist, its overexpression is highly associated with high-grade dysplastic lesions and carcinoma. The frequency of p16 INK4a overexpression differed significantly between CIN 1, 2 and 3 in our study, suggesting that the disorder of cell cycle regulation by HPV frequently occurred from CIN 2 onward. CIN 3 is considered to be a more valid and true precursor lesion for cervical cancer 17 and is the most reproducible of the CIN categories. It has been well established that a high proportion of CIN 3 lesions are positive for high-risk HPV and p16 INK4a immunoexpression, which may serve as useful biomarkers for the accurate early diagnosis of cervical precancerous lesions. 6,22,23 Our results are consistent with this observation that validates the histological diagnosis of CIN 3 lesions in our study. The fact that cytology was more sensitive and specific than the visual tests in detecting high-grade CIN lesions positive for HPV and/or p16 INK4a overexpression raises the possibility that diagnostic misclassification or overcalling of CIN lesions on histology may partially account for the moderate or poor sensitivity of cytology. However, compared to the resources and challenges associated with implementing cytology screening in low-resource settings, visual screening is less resource-intensive and more feasible to implement in settings where HPV screening is not. The efficacy of VIA screening in reducing the burden of high-grade CIN following VIA screen and treat using cryotherapy has been established in studies in South Africa and Peru. 4,24,25 There was a 35% reduction in the cumulative frequency if CIN 2 3 lesions in women receiving cryotherapy following a positive VIA in screen in South Africa. 4,24 A four-fold lower prevalence of CIN 2 3 and 9-fold lesser frequency invasive cervical cancer were observed in women previously screened by VIA in Peru, as compared with women not previously screened by VIA, implying that a single VIA screening can lower the population risk for cervical cancer. 25 A 25% reduction in cervical cancer incidence and a 35% reduction in cervical cancer mortality following a single round of VIA screening were observed in the context of a randomized trial in South India, 26 although VIA screening References was not followed by significant reduction in cervical cancer burden in another randomized trial in Western India. 7 An almost 50% decline in cervical cancer incidence rates was observed in Dindigul district, Tamil Nadu after wide-spread use of VIA screening in the district. 27 Thus, there is a wide ranging evidence base on accuracy 28 and efficacy indicators 4,24 26 that justify the use of visual screening tests for cervical cancer screening as an alternative approach to cytology in two contexts in settings where HPV testing is neither feasible nor affordable: either as a primary screening test or to triage women who have tested positive for high-risk HPV infection for directing treatment. Although currently available HPV tests are expensive, new more affordable HPV tests (e.g., carehpv test) are likely to be commercially available in the near future in low-resource countries. Studies in China indicate that the carehpv test is a promising primary screening method for cervical cancer prevention in low-resource regions, with high accuracy to detect cervical cancer precursor lesions. 5 Whereas the wide publicity surrounding HPV vaccines has led to a sense that HPV vaccine programs will be soon more widely available in both developed and developing countries, introduction of visual screening in programmatic contexts (where HPV screening is not feasible) as an alternative to cytology has received much less attention. Improving screening coverage with visual screening in low-resource countries would facilitate the gradual building of an infrastructure committed to screening that will facilitate the eventual introduction of more sensitive and highly objective and reproducible HPV screening, when affordable HPV tests are available for use in developing countries. This approach will save the lives of several women who have already been exposed to persistent HPV infection and will develop cervical cancer during the next years. Acknowledgements The authors express their appreciation to Mrs. Mary Renaud for her secretarial and technical assistance in the conduct of the START project; to the women and their families who participated in the project as well as the local administrative and health authorities in India for their support to the field studies. The authors thank Mrs. Evelyn Bayle and Mrs. Krittika Guinot for their help in preparing this manuscript. 1. Sankaranarayanan R, Budukh AM, Rajkumar R. Effective screening programmes for cervical cancer in low- and middle-income developing countries. Bull World Health Organ 2001;79: IARC. IARC handbooks on cancer prevention, Vol. 10. Cervix cancer screening. Lyon: International Agency for Research on Cancer, Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN Int J Cancer 2010;127: Denny L, Kuhn L, Hu CC, et al. Human papillomavirus-based cervical cancer prevention: long-term results of a randomized screening trial. J Natl Cancer Inst 2010;102: Qiao YL, Sellors JW, Eder PS, et al. A new HPV- DNA test for cervical-cancer screening in developing regions: a cross-sectional study of clinical accuracy in rural China. Lancet Oncol 2008;9: Reuschenbach M, Seiz M, von Knebel DC, et al. Evaluation of cervical cone biopsies for coexpression of p16ink4a and Ki-67 in epithelial cells. Int J Cancer 2012;130: Sankaranarayanan R, Nene BM, Shastri SS, et al. HPV screening for cervical cancer in rural India. N Engl J Med 2009;360: Sankaranarayanan R, Wesley RS. A practical manual on visual screening for cervical neoplasia. IARC Technical Publication No. 41. Lyon: International Agency for Research on Cancer, Sellors J, Sankaranarayanan R. Colposcopy and treatment of cervical intraepithelial neoplasia: a beginners manual. Lyon: International Agency for Research on Cancer, Pretorius RG, Zhang WH, Belinson JL, et al. Colposcopically directed biopsy, random cervical biopsy, and endocervical curettage in the diagnosis of cervical intraepithelial neoplasia II or worse. Am J Obstet Gynecol 2004;191:

9 E962 Accuracy of screening tests in detecting precancer lesions 11. Solomon D, Davey D, Kurman R, et al. The 2001 Bethesda System: terminology for reporting results of cervical cytology. JAMA 2002;287: Tavassoeli FA, Devilee P. Pathology and genetics tumours of the breast and female genital organs. WHO classification of tumors, 3rd edn., vol.4. Lyon: International Agency for Research on Cancer, Tsoumpou I, Arbyn M, Kyrgiou M, et al. p16(ink4a) immunostaining in cytological and histological specimens from the uterine cervix: a systematic review and meta-analysis. Cancer Treat Rev 2009;35: Gheit T, Landi S, Gemignani F, et al. Development of a sensitive and specific assay combining multiplex PCR and DNA microarray primer extension to detect high-risk mucosal human papillomavirus types. J Clin Microbiol 2006;44: Saiki RK, Gelfand DH, Stoffel S, et al. Primerdirected enzymatic amplification of DNA with a thermostable DNA polymerase. Science 1988;239: StataCorp LP. Stata base reference manual: Release 9. College Station, TX: StataCorp LP, Schiffman M, Wentzensen N, Wacholder S, et al. Human papillomavirus testing in the prevention of cervical cancer. J Natl Cancer Inst 2011;103: Sankaranarayanan R, Gaffikin L, Jacob M, et al. A critical assessment of screening methods for cervical neoplasia. Int J Gynaecol Obstet 2005; 89(Suppl 2):S4 S Sankaranarayanan R, Thara S, Sharma A, et al. Accuracy of conventional cytology: results from a multicentre screening study in India. JMed Screen 2004;11: Volgareva G, Zavalishina L, Andreeva Y, et al. Protein p16 as a marker of dysplastic and neoplastic alterations in cervical epithelial cells. BMC Cancer 2004;4: Missaoui N, Hmissa S, Sankaranarayanan R, et al. [p16ink4a overexpression is a useful marker for uterine cervix lesions]. Ann Biol Clin (Paris) 2010;68: Ishikawa M, Fujii T, Saito M, et al. Overexpression of p16 INK4a as an indicator for human papillomavirus oncogenic activity in cervical squamous neoplasia. Int J Gynecol Cancer 2006;16: Ekalaksananan T, Pientong C, Kongyingyoes B, et al. Combined p16ink4a and human papillomavirus testing improves the prediction of cervical intraepithelial neoplasia (CIN II-III) in Thai patients with low-grade cytological abnormalities. Asian Pac J Cancer Prev 2011;12: Kuhn L, Wang C, Tsai WY, et al. Efficacy of human papillomavirus-based screen-and-treat for cervical cancer prevention among HIV-infected women. AIDS 2010;24: Luciani S, Munoz S, Gonzales M, et al. Effectiveness of cervical cancer screening using visual inspection with acetic acid in Peru. Int J Gynaecol Obstet 2011;115: Sankaranarayanan R, Esmy PO, Rajkumar R, et al. Effect of visual screening on cervical cancer incidence and mortality in Tamil Nadu, India: a cluster-randomised trial. Lancet 2007;370: Swaminathan R, Selvakumaran R, Esmy PO, et al. Cancer pattern and survival in a rural district in South India. Cancer Epidemiol 2009;33: Sauvaget C, Fayette JM, Muwonge R, et al. Accuracy of visual inspection with acetic acid for cervical cancer screening. Int J Gynaecol Obstet 2011;113:14 24.