|
Indian Journal of Cancer, Vol. 47, No. 3, July-September, 2010, pp. 308-313 Original Article Evaluation of colposcopy vs cytology as secondary test to triage women found positive on visual inspection test Pimple SA, Amin G1, Goswami S, Shastri SS Department of Preventive Oncology, Tata Memorial Hospital, Mumbai, India, 1 Honarary Colposcopist, Cama Albless Hospital, Mumbai, India Code Number: cn10074 PMID: 20587908 DOI: 10.4103/0019-509X.64726 Abstract Background: Colposcopic evaluation and guided biopsy is an important diagnostic step and standard of management for abnormal cytology smears in developed countries. Aim: The present study evaluates the performance of colposcopy vs conventional cytology in estimating the presence and grade of cervical disease against the reference standard of histopathology as a secondary test modality to triage women found positive on primary screening by visual inspection with 5% acetic acid (VIA). Settings and Design: Colposcopy and directed biopsy were performed after primary screening for cervical cancer in an urban hospital cancer screening clinic between January 2000 and June 2005, in Mumbai, India. Materials and Methods: Healthy women (8863) in the age group of 35-65 years participated in the cervical cancer early detection program in the hospital and community cancer screening clinics in Mumbai. Women found positive on the primary screening test by VIA underwent diagnostic evaluation by Pap smear cytology and colposcopy evaluation with directed biopsies. Statistical Analysis: Accuracy parameters and their 95% confidence intervals were calculated using 2 Χ 2 tables and standard formulae. Results: The test range of sensitivity of colposcopy for the detection of histologically confirmed cervical intraepithelial neoplasia grade 1+ (CIN1+) or CIN2+ was 58.0-74.7% and that of specificity was 57.5-92.9%. The sensitivity and specificity of cytology to detect CIN2+ was 57.4% and 99.4%, respectively. Conclusion: Colposcopy is a good sensitive test for the detection of CIN and can be considered as a secondary testing tool to triage women found positive on VIA. Keywords: Cervix cancer, colposcopy, screening, sensitivity, specificity Introduction India accounts for one fifth of the world burden of cervical cancer with an estimated 126,000 new cases and 71,000 deaths annually due to the disease. [1] Contributing to the cervical cancer health burden in a country like India is the lack of awareness among people and policy makers that it is readily preventable when effective programs are implemented to detect and treat its precursor lesions. The rates are high particularly in the rural areas, and the absolute number of cases is on the rise due to population growth. Conventional cytology-based screening programs currently are not feasible in India where infrastructure and quality assurance requirements are not consistently met. Newer cervical cancer screening strategies, such as visual inspection-based approaches as primary screening tests have been extensively investigated in India. [2],[3],[4],[5],[6],[7],[8],[9],[10],[11] Although visual inspection with 5% acetic acid (VIA) is a test with good sensitivity, low specificity has been its limitation, which would result in excessive referrals and treatment of false-positive lesions subsequently increasing the referral load as well as the cost of unnecessary treatment on the health system. Thus, determining which women with positive VIA-based tests are at risk for significant cervical disease, performing appropriate diagnostic workups, and treating cancer precursors presents a major public health challenge. Several studies have considered human papillomavirus (HPV) testing and repeat cytology as a triage method for women with atypical squamous cells of unknown significance (ASCUS). [12],[13],[14],[15],[16] But these triage modalities are not feasible in a developing country such as India. Cervical cytology screening programs despite its history of success in cancer screening has important limitations, particularly its high false-negative rate, which carries important public health implications. [17] Colposcopic evaluation and guided biopsy remains a critical diagnostic step for women with squamous intraepithelial lesions to identify women who require treatment. [18] Thus, in the context of adopting VIA-based approaches, namely VIA as a primary screening modality in low-resource settings, we tried to evaluate the performance of colposcopy in estimating the presence and grade of cervical disease vs conventional cytology testing as a secondary test modality to triage women found positive on VIA test. Colposcopy may then be used to identify women who are likely to benefit from immediate treatment. Materials and Methods An urban cancer treatment facility conducted cervical cancer screening, diagnostic evaluation, and precancer management of women through its hospital-based and community-based cancer screening clinics in Mumbai. A total of 8863 apparently healthy women in the age group of 30-65 years participated in the cervical cancer early detection program between January 2000 and June 2005. The present study involved the analysis of 2034 women who tested positive on primary screening by VIA and who further underwent diagnostic evaluation by Pap smear cytology and colposcopy and were subsequently subjected to colposcopy-directed biopsies. Women with intact uterus, nonpregnant, and with no past history of cervical neoplasia were selected for the study. Health awareness programs about the prevention of cervical cancer and screening were conducted prior to participation in screening clinics by trained social workers and health workers. The doctors and health workers were trained in an intensive course utilizing training manual prepared by the IARC (International Agency for Research in Cancer). [19] The doctors received an intensive training on colposcopy and precancer management by cryotherapy and loop electrosurgical excision procedure using the manual prepared by the IARC. The technicians involved in cytology testing were trained in collection, processing, and microscopic interpretation. A medical social worker explained in local languages the nature and the purpose of the screening tests before obtaining the women′s written consent. Subject information material and consent forms were read out to the subject participants wherever necessary before obtaining their signature or left thumb impression. A detailed sociodemographic and reproductive history was then obtained in a structured questionnaire before subjecting them to screening. The primary cervical cancer screening tests were administered by high school graduate-trained health workers. Women underwent screening by VIA. Acetic acid (5%) was applied to the cervix using a cotton swab and VIA findings were reported 1 min after the application as negative or positive. The result of VIA test was recorded positive when there were sharp, distinct, well-defined, dense acetowhite areas with or without raised margins, closer to the squamocolumnar junction in the transformation zone and not far away from the cervical os. The women with negative findings were reassured, counseled, and sent back with an appointment for a next screening date. The women with positive findings were referred to a hospital facility for diagnostic evaluation. The field clinics were always supervised by a medical doctor for quality checks. In the screening clinic, conventional cytology testing was obtained by scraping the cervical cells with a thin cotton swab by a trained technician. A smear was prepared by spreading the specimen uniformly across a glass slide, which was immediately fixed in 95% ethyl alcohol contained in a plastic or glass jar and transported to the Tata Memorial Hospital, cytology laboratory. Cytology results were reported according to the Bethesda system.[20] Cytology was considered to be positive at the high-grade squamous intraepithelial lesion (HGSIL) threshold. All those who were found positive on VIA were subjected to colposcopy directed biopsies. The colposcopic diagnosis and grading was done on Reid index, which assigns scores of 1-8 for the colposcopic appearance of margins, lesion color, vascularity, and iodine staining. [21] The colposcopic findings were classified according to the nomenclature into 1. Normal; 2. Squamous metaplasia; 3. Leukoplakia; 4. Condyloma/wart; 5. Probably low-grade lesion: atypia/cervical intraepithelial neoplasia grade 1 (CIN1); 6. Probably high-grade lesion CIN2-3; and 7. Invasive carcinoma. Punch biopsies were obtained from the worst of any abnormal areas under colposcopic guidance. Biopsy specimens obtained were fixed in formalin and were processed and reported using the CIN system at the histopathology laboratory of the Tata Memorial Hospital. The study outcome was defined as CIN2 and worse lesions, and this disease threshold was used to calculate sensitivity, specificity, and predictive values of the screening tests. The final reference diagnosis was based on the histopathology findings. Data were entered in the institution using a standard computer software (EPIINFO software, 3.5.1 developed by the US centers for disease control). Sensitivity, specificity, and predictive values and their 95% confidence intervals for single and combined tests were calculated using 2 Χ 2 tables and standard formulae. The data were retrospectively analyzed by selecting 2034 women who were positive for VIA primary screening tests and had undergone cytology smear and colposcopy evaluation and were subsequently subjected to colposcopically directed biopsies. Results A total of 8863 healthy women in the age group of 35-65 years participated in the cervical cancer early detection program. Out of the 8863 women who participated in the cervical cancer screening clinics, 2034 women were found to be acetowhite-positive on primary screening test (VIA). In total 103 cases were excluded from the study due to inconclusive colposcopies and inadequate cytology or biopsy results, leaving 1931 cases to be analyzed. The colposcopic impression [Table - 1] was benign in 977 (50.6%) women, condylomatous changes in 82 (4.2%), CIN1 changes in 653 (34%), CIN2-3 in 166 (8.6%), and invasive carcinoma in 53 (2.7%). Conventional cytology [Table - 2] was found to be normal in 1762 (91.2%) women, ASCUS were seen in 31 (1.6%), low-grade squamous intraepithelial lesion (LGSIL) in 32 (1.6%), HGSIL in 72 (3.7%), and invasive cancer in 32 (1.6%). Histopathology findings were reported as benign in 1576 (81.6%), atypia or HPV changes in 80 (4.1%), CIN1 in 113 (5.8%), CIN2 in 56 (2.9%), CIN3 in 50 (2.6%), and invasive carcinoma in 56 (2.9%). [Table - 3] shows the test characteristics of colposcopy and cytology at various thresholds for outcomes of CIN2+ on histology. The estimates of sensitivity for low- and high-threshold colposcopy were 58.0% and 74.5%, respectively, and those of specificity were 57.5% and 92.9%, respectively. The positive predictive value (PPV) and the negative predictive value (NPV) for low-threshold colposcopy were 13.8% and 96.1%, respectively, and those for high-threshold colposcopy were 42.9% and 96.0%, respectively. The test characteristics for conventional cytology to detect HGSIL performed against the gold standard of histopathology (CIN2 and above lesions), on all the VIA-positive women showed sensitivity, specificity, PPV, and NPV of 57.4%, 99.4%, 89.4%, and 96.2%, respectively. The sensitivity of cytology at lower thresholds to detect ASCUS+ and LGSIL+ lesions, for outcomes of CIN2+ on histology, was higher at 67.9% and 62.3%, respectively, with a slight dip in the specificity at 96.8% and 98.0%, respectively. The false-negative rate of 42.6% for conventional cytology is almost comparable to that of colposcopy 41.9% at a higher threshold. The PPV of a high-grade colposcopic impression (CIN2 and above) for high-grade histology result was 42.9% compared with 89.4% of conventional cytology. Discussion Organized and frequently repeated cytology screening has resulted in a substantial reduction of cervical cancer burden in developed countries. But in low-resource countries where organized cytology-based cervical cancer screening programs cannot be implemented due to financial, technical, and logistic barriers, low-cost technologies, such as the VIA-based approaches have been successfully tested and proposed to address the need to effectively improve and extend screening services in the country. [2],[3],[4],[5],[6],[7],[8],[9],[10],[11] With the added advantage of the immediate availability of VIA test result, VIA-positive women can be subjected to further investigative procedures to ensure diagnostic and treatment compliance with a "Single Visit" approach. Diagnostic triage of VIA-positive women by cytology or colposcopy directed biopsy are still not very feasible in low-resource country settings where adequate expertise, facility, and infrastructure are still not available for cytology and histopathology confirmation, outside of the city limits. Also, poor patient compliance for further diagnostic or treatment visits and inadequate patient tracking system creates further barriers in the successful implementation of screening programs. Hence a "Single Visit" screen and treat strategy that uses VIA and colposcopy alone that eliminates the need for repeated visits due to delays in diagnostic results, will be highly attractive in terms of cost-effectiveness and compliance to treatment, which is crucial to bring down the incidence and mortality due to cervical cancer. Thus, in the context of adopting VIA-based approach as a primary screening modality in low-resource settings, we tried to evaluate the performance of colposcopy to estimate the presence and grade of cervical disease vs conventional cytology testing to triage VIA-positive women. The performance and accuracy of colposcopy depends largely on the training, experience, and skills of the colposcopist. Hence, accuracy of colposcopy varies widely among studies in different parts of the world. In a meta-analysis, Mitchell and colleagues report studies that distinguished normal cervix from all other diagnosis, for which the individual estimations of sensitivity of diagnostic colposcopy (87-99%) were high, whereas those of specificity (23-87%) were lower. Similarly, among 8 other studies with fully separated disease categories, for distinguishing normal cervix, atypia, and LGSIL from HGSIL and cancer, the estimates of sensitivity of diagnostic colposcopy ranged from 64% to 99% and the specificity from 30% to 93%. [22] Also data from Massad and Collins reported that the sensitivity of colposcopy with a threshold of any lesion detected was 89% but fell to 56% when the threshold was raised to a high-grade result. [23] In the present study, the sensitivity of colposcopy at low thresholds was high (74.5%), but the specificity was lower at 57.5%. Lower specificity for a diagnostic test would always increase further burden on the health system in terms of unnecessary increased health care cost for treatment of the false positives, which offsets the primary goal of making cancer screening services cost-effective for low-resource settings. But at a higher threshold, the sensitivity of colposcopy for the detection of CIN2+ histology lesions drops down to 58.0%, with a consequent rise in specificity at 92.9%. The test characteristics of colposcopy in the current study at high threshold colposcopy impression for detection of CIN2+ histology lesions is almost comparable to that of conventional cytology (sensitivity 57.4%, specificity 99.4%, and NPV 96.2%), wherein cytology smears were tested at a tertiary care institute with internal and external quality control measures adopted in cytology and pathology laboratories. The above comparable estimates have an important implication in adopting colposcopy to triage VIA-positive women, which obviates the necessity of resource-intensive cytology and also because the performance of cytology is known to be suboptimal outside the centers of excellence or beyond the tertiary care centers in less developed countries. Cytology at lower cutoffs (ASCUS+, LGSIL+), however, shows improved sensitivity in our study but requires laboratory services and skilled cytologists. Also it does not provide immediate results, which entails repeated patient visits for further testing and management and has remained one of the biggest challenges in implementing cytology-based cervical cancer screening programs in the developing country settings. The advantage of higher specificity could be considered less primordial given the possibility to treat lesions with cryotherapy. The present study, however, suffers from the limitation of colposcopy being performed by multiple colposcopists at various levels of expertise, many of them recently trained, presumably in their learning curves during the entire phase of the study. Although more recent studies suggest that in expert hands colposcopy can be highly accurate, skill levels vary. [24],[25],[26],[27],[28],[29],[30] Thus in spite of the above limitations, our findings suggest that colposcopy at lower colposcopy cutoffs shows acceptable sensitivity for a histologic outcome of CIN2+ than cytology, but the specificity of colposcopy is then much lower. Thus colposcopy, which gives immediate results, can be considered as a secondary testing tool to triage women found positive on VIA in settings where cytology and histopathology services are logistically and technically not feasible. The disadvantage of colposcopy is its lower specificity at low cutoffs and the need for the colposcope and that of a skilled colposcopist. The lower specificity could still be considered acceptable if the treatment by cryotherapy, which is effective, cheap, and safe is offered and the colposcopists are skilled in recognizing the exclusion criteria of cryotherapy. Thus, alternative approaches to the traditional model that reduce the need for extensive cytology and histopathology services would make cervical cancer screening and treatment possible in developing countries. Acknowledgements The authors wish to acknowledge the contributions of the following persons who were associated with the service program in various capacities at different times during the conduct of the program. Dr. Bhavana Mangal, Mr. Sachin Khedekar, Ms. Madhuri Deshmukh, and Ms. Priya Menon. References
Copyright 2010 - Indian Journal of Cancer The following images related to this document are available:Photo images[cn10074t1.jpg] [cn10074t2.jpg] [cn10074t3.jpg] |
|