DOI 10.1007/s00404-012-2569-y GYNECOLOGIC ONCOLOGY Acceptable predictive accuracy of histopathology results by colposcopy done by Gynecology residents using Reid index Hadi Shojaei Fariba Yarandi Leila Ghozati Niloufar Yarandi Narges Izadi-Mood Zahra Eftekhar Received: 20 May 2012 / Accepted: 10 September 2012 Ó Springer-Verlag 2012 Abstract Objective The aim of this study was to determine the strength of the correlation between colposcopic impression according to Reid colposcopic index (RCI) done by Gynecology residents and biopsy histology in a university hospital. Methods Colposcopy was performed on 260 women. According to RCI, the scores zero, one, or two were given to each of four standardized colposcopy patterns (acid staining, iodide staining, margin of lesion, and vascular pattern) and the total score was calculated. In those with multiple lesions, the patterns with the highest score were considered. Then the biopsy was obtained from the lesion and put in formalin for pathological evaluation. Results There was a statistically significant association between colposcopy findings and histopathology findings and the score was increased as parallel as malignancy grade (r = 0.680, P \ 0.05). The highest sensitivity and specificity for diagnosis of each CIN also were related to staining with acetic acid. For high-grade CIN lesions, the highest specificity was related to staining with acetic acid, but the sensitivity was equal for four findings. H. Shojaei (&) F. Yarandi N. Yarandi Z. Eftekhar Gynecology Oncology Department, Mirza-koochak-Khan Hospital, Tehran University of Medical Sciences, Nejatollahi St., Karim Khan Zand Ave., Tehran, Iran e-mail: hadi.shojaee@gmail.com L. Ghozati Gynecology and Obstetrics Resident, Mirza-koochak-Khan Hospital, Tehran University of Medical Sciences, Tehran, Iran N. Izadi-Mood Pathology Department, Mirza-koochak-Khan Hospital, Tehran University of Medical Sciences, Tehran, Iran Conclusion Colposcopy using RCI yields a good correlation with histology results. It also showed that colposcopy done by Gynecology residents using RCI is a feasible and acceptable cervical cancer screening method in a university hospital. Keywords Colposcopy Reid colposcopic index Histology Obstetrics and gynecology resident Introduction Pap smear test has been less successful in identifying those women with the highest risk for premalignant disease, so the patients with equivocal Pap smear would need further evaluation with colposcopy [1, 2]. Performing the colposcopy with more accuracy would result in better diagnosis of malignant and premalignant lesions [3]. This matter also may be achieved better when we have a systematized scoring system [4]. The Reid colposcopic index (RCI) is one of the most recently developed colposcopic grading systems for assessment of preinvasive disease [5, 6]. It considers four colposcopic signs: margin, color, vessels, and iodine staining. Each sign is assigned a score between 0 and 2 reflecting variation in colposcopic appearances. Scores of 0 represent normal variant (immature metaplasia) or human papillomavirus (HPV)-induced changes. Scores of 2 represent features suggestive of cervical intraepithelial neoplasia (CIN) 3, with a score of 1 suggesting intermediate grade pathology. When the four scores are summed, total scores ranging from 0 to 2 indicate immature metaplasia or CIN 1, scores of 3 5 indicate CIN 1 to CIN 2, and scores of 6 8 represent CIN 3. The accuracy of the RCI varies but has been reported to be high in some investigations [7].
However, performing a comparison with directed biopsy and determining the sensitivity and specificity of colposcopy are required to obtain more definite results. Accordingly, this study was conducted to determine the diagnostic accuracy of colposcopy done by Gynecology residents according to RCI in diagnosis of malignant and premalignant cervical lesions in a university hospital in Iran. Materials and methods In this partially blinded cross-sectional study, 260 women attending to Mirza-koochak-Khan Hospital, Tehran, Iran, from 2006 to 2008 were evaluated. The inclusion criterion was abnormal cytological results. Exclusion criteria were the patients with normal transitional zone and those without lesion. The study was approved by ethical committee of Tehran University of Medical Sciences. Therequireddatawerecollectedbyaquestionnaire. The histopathology results were added, subsequently. To determine the accuracy of resident s colposcopic impression, all colposcopies were performed by a third or fourth year gynecology resident under the guidance of an attending Gynecologic oncologist. Patients examined in a lithotomic position and the speculum was inserted, and a Pap smear (in patients with no current Pap smear) was prepared and all the patients underwent colposcopic examination with three solution. First, the cervix was washed with normal saline wool, and then the vascular pattern was washed with 5 % acetic acid for 1 min and the patients with normal transitional zone and those without lesion were excluded in this stage. In patients with cervical lesion and abnormal colposcopy, the vascular pattern was assessed with green filter and the margin and the color were rechecked after use of 5 % acetic acid and potassium iodide solution with 1/4 dilution. According to RCI, the scores zero, one, or two were given to each of four standardized colposcopy patterns (acid staining, iodide staining, margin of lesion, and vascular pattern) and the total score was calculated. In those with multiple lesion that with highest score were considered. Then the biopsy was obtained from the lesion and put in formalin for pathological evaluation. The pathologists were blind about the clinical history and total score of the patients. Data obtained from 260 women were analyzed using SPSS (version 11.5) software (Statistical Procedures for Social Sciences; Chicago, IL, USA). Exact-Fisher and Chi- Square tests were used with a significant level of 0.05. Also the sensitivity, specificity, positive predictive value, and negative predictive value were calculated. Results The mean age (±standard deviation) was 40.9 ± 12.3 years with a range from 18 to 75 years. Totally 47 patients (18.1 %) were in post-menopausal ages. The frequency of those in post-menopausal ages mentioned according to pathological diagnosis (Table 1). The frequency of different Pap smear results according to pathological diagnosis is shown in Table 2. Totally, 98 patients (37.3 %) had normal pathology results, 35 subjects (13.3 %) had HPV, 53 patients (20.4 %) had CIN1, 15 subjects (5.8 %) had CIN2, 24 patients (9.2 %) had CIN3, and 35 subjects (13.5 %) had cancer. Patients with normal pathology results underwent repeated Pap smear after 6 months for follow up, which was normal in all patients. Table 1 The mean age for each pathological diagnosis Biopsy Minimum Maximum Mean ± standard deviation Normal (pathology) 18 65 38.3 ± 10.9 HPV 20 63 36.3 ± 11.1 CIN 1 21 53 36.3 ± 7.8 CIN 2 26 67 41.4 ± 13.2 CIN 3 25 69 43.1 ± 12.3 Cancer 28 75 51.7 ± 12.7 Table 2 The frequency of Pap smear results for each pathological diagnosis Pathology Pap smear Normal ASCUS ASC-H LSIL HSIL AGCS Cancer ASCUS atypical cells of undetermined significance, ASC-H atypical squamous cellscannot exclude HSIL, LSIL lowgrade squamous intraepithelial lesion, HSIL high-grade squamous intraepithelial lesion, AGCS atypical glandular cells Normal 42 27 16 9 0 4 0 HPV 21 3 4 6 0 0 1 CIN 1 28 5 2 14 0 0 0 CIN 2 0 0 0 11 4 0 0 CIN 3 0 0 7 3 12 0 6 Cancer 0 0 3 0 12 2 18 Total 91 35 32 43 28 6 25
Table 3 Odds ratio for each colposcopy finding in CIN lesions Colposcopy findings Frequency CIN lesions Crude odds ratio (95 % CI) Margin 0 132 41 (31.1 %) 1 a 1 99 65 (65.7 %) 4.2 (2.4 7.3) 2 29 21 (72.4 %) 5.8 (2.3 14.2) Acetic acid staining 0 106 11 (10.4 %) 1 1 94 69 (73.4 %) 23.8 (10.9 51.68) 2 60 47 (78.3 %) 31.2 (13 74.9) Vascular pattern 0 100 30 (30 %) 1 1 68 (55.3 %) 2.8 (1.6 5.03) 2 37 29 (78.4 %) 8.4 (3.4 20.6) Iodine staining 0 138 43 (31.2 %) 1 1 83 54 (65.1 %) 4.1 (2.3 7.3) 2 39 30 (76.9 %) 7.3 (3.2 16.8) a In all colposcopy findings the zero score is reference Table 4 Odds ratio for each colposcopy finding in high-grade CIN lesions Colposcopy findings Frequency CIN lesions Crude odds ratio (95 % CI) Margin 0 132 15 (11.4 %) 1 a 1 99 38 (38.4 %) 4.8 (2.4 9.5) 2 29 21 (72.4 %) 20.4 (7.7 54.3) Acetic acid staining 0 106 3 (2.8 %) 1 1 94 26 (27.7 %) 13.12 (3.8 45.07) 2 60 45 (75 %) 103 (28.4 373.4) Vascular pattern 0 100 14 (14 %) 1 1 32 (26 %) 2.1 (1.07 4.3) 2 37 28 (75.7 %) 19.1 (7.4 48.9) Iodine staining 0 138 12 (8.7 %) 1 1 83 33 (39.8 %) 6.9 (3.3 14.4) 2 39 29 (74.4 %) 30.4 (11.9 77.2) a In all colposcopy findings the zero score is reference Polymerase chain reaction diagnosed 35 (13.3 %) patients with high-risk HPV. With regard to case distribution according to type of high-risk HPV diagnosed by PCR, 19 (54.3 %) of patients was positive for HPV 16, while 16 patients (45.7 %) presented HPV 18. The odds ratio for each colposcopy finding is shown in Table 3 for CIN lesions. 52 out of 60 patients (87 %) with total score equal with or higher than five, (high grade according to Reid), had CIN2 pathology result and there was a significant association between clinical scoring and histopathology results in high-grade lesions (OR = 52.2 [22.11 125.05]). Also 18 out of 69 patients with total score 3 4 (intermediate grade according to Reid) had high-grade CIN and 51 subjects (73 %) had low grade CIN. Also 127 out of 131 patients with total score 0 1 had low grade CIN or normal results and 4 subjects (3.1 %) had high-grade CIN and all four cases were CIN2. It means that we had four cases of under diagnosis and eight cases of over diagnosis. The odds ratio for each colposcopy finding is shown in Table 4 for high-grade CIN lesions. This table shows that staining with acetic acid has a more odds ratio for detection of malignancy compared with other colposcopy findings. The highest sensitivity and specificity for diagnosis of each CIN are related to staining with acetic acid. For high-grade CIN lesions, the highest specificity is for staining with acetic acid, but the sensitivity is equal for four findings. The sensitivity, specificity, positive predictive value, and negative predictive value for each colposcopy finding in high grade and any CIN lesions are shown in Table 5. Table 5 Sensitivity, specificity, positive predictive value, and negative predictive value for each colposcopy finding in high grade and any CIN lesions Discussion Sensitivity Specificity PPV NPV Any CIN Total score C3 78.1 78.7 79.3 77.5 Margin C1 68.4 67.7 68.9 67.1 Acetic acid staining C1 71.4 91.3 89.6 75.3 Vascular pattern C1 52.6 76.3 70 60.6 Iodine staining C1 71.4 66.1 68.8 68.8 High-grade CIN Total score C5 95.6 70.2 89 86 Margin C2 95.6 28.3 77.05 72.4 Acetic acid staining C2 91.9 60.8 85.5 75 Vascular pattern C2 95.1 37.8 79.3 75.6 Iodine staining C2 94.6 39.1 79.6 74.3 As seen in this study, there was a statistically significant association between colposcopy findings and histopathology findings and the score was increased as parallel as malignancy grade (r = 0.680, P \ 0.05). Our findings reveal that colposcopy performed in obstetrics and gynecology residency training program is accurate, and the colposcopic impressions of trainees are highly associated with underlying cervical pathology. In our study, staining with acetic acid had higher odds ratio for detection of malignancy compared with other
colposcopy findings. The highest sensitivity and specificity for diagnosis of each CIN also were related to staining with acetic acid. For high-grade CIN lesions, the highest specificity was related to staining with acetic acid, but the sensitivity was equal for four findings. Findings are consistent with others in that the whitening from acetic acid leads us to the high-grade lesions. Colposcopy was carried out using Reid colposcopic index (RCI) scoring system and directed biopsy on 344 women by Mousavi et al. [8]. The association between colposcopic impression and biopsy histology was highly significant (P \ 0.001), both in RCI colposcopy group and general colposcopy group as well as our study. However, the strength of the correlation between colposcopy impression and biopsy histology in RCI colposcopy group was more than the general colposcopy group (0.74 vs. 0.45). The positive predictive value of any colposcopic abnormality for any histologic abnormalities in the RCI group was 92 %. The negative predictive value of a benign colposcopic impression was 70.5 %. The sensitivity was 74 %, and the specificity was 90.7 %. These measurements are similar to our findings. In another study done by Massad et al. [9] colposcopy was performed on 2,112 women. Exact agreement was found in only 893 (37 %) women, but results agreed within one grade in 1,203 (75 %). The association between impression and histology was significant (P \ 0.001), but the strength of the correlation was poor (0.20). However, in our study the correlation was stronger. The positive predictive value of any colposcopic abnormality for any histologic abnormality was 80 % that is relatively similar to our findings. The negative predictive value of a benign colposcopic impression was 68 % that is lower than that found in our study. The sensitivity of colposcopy with a threshold of any lesion detected was 89 %, and the specificity was 52 %; the first was near to our findings, but the latter was much lower than our findings. In a study by Baum et al. [10], agreement within onestep between cervical histology and the colposcopic impression was found in 351 (77 %) of the subjects. The association between cervical biopsy and impression was highly significant (P \ 0.0001) as well as our study. However, the strength of the correlation was only slight (j = 0.197) despite of kappa equaling 0.680 in our study. The positive predictive value for the association of any colposcopically detected abnormality with any histologic abnormality was 64.1 % that is lower than that found in our study. In a study by Hong et al. [11], the sensitivity of high-risk HPV testing for detecting HSIL was 94.4 %, the specificity was 65.0 %, the positive predictive value was 75.5 %, and the negative predictive value was 91.0 %. The sensitivity, specificity, positive predictive value, and negative predictive value of the colposcopic impression for detecting HSIL were 91.3, 92.9, 93.6, and 90.3 %, respectively. We also found similar results in current study. Most of the current literature has discredited the RCI as inaccurate and non-reproducible even when it is applied by expert colposcopists [12]. The only tool that has been found to be effective in increasing the sensitivity of colposcopy is taking more biopsies [13]. Our study found that residents in a university hospital were able to successfully utilize the RCI to accurately predict histology. One potential explanation offered for the currently poor performance of the RCI in the US is that the lesions are smaller and less pronounced and so harder to predict [14]. Our study was done in a referral Gynecology hospital, and maybe our patients cervical lesions were larger and more advanced. Totally, colposcopic assessment by novice colposcopists is still a valuable exercise and can help residents and others to get organized and biopsy the most abnormal areas. Currently not everyone believes it is of value. They just want colposcopists to take more biopsies. Residency training programs are a good setting to evaluate early skills and knowledge. The improved sensitivity of cytological screening, along with the addition of HPV testing, could potentially lead to large numbers of referrals to colposcopy. However, the actual technique of colposcopy has yet to be improved. According to our results, showing a good association between colposcopic results using Reid index and histopathologic results, it is suggested to use this scoring system for better use and interpretation of colposcopy especially in university hospitals. Conflict of interest References We declare that we have no conflict of interest. 1. Dane C, Batmaz G, Dane B, Cetin A (2009) Screening properties of human papillomavirus testing for predicting cervical intraepithelial neoplasia in atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesion smears: a prospective study. Ann Diagn Pathol 13(2):73 77 2. Yarandi F, Izadi MN, Mirashrafi F, Eftekhar Z (2004) Colposcopic and histologic findings in women with a cytologic diagnosis of atypical squamous cells of undetermined significance. Aust N Z J Obstet Gynaecol 44(6):514 516 3. Dexeus S, Cararach M, Dexeus D (2002) The role of colposcopy in modern gynecology. Eur J Gynaecol Oncol 23(4):269 277 4. Hammes LS, Naud P, Passos EP et al (2007) Value of the International Federation for Cervical Pathology and Colposcopy (IFCPC) Terminology in predicting cervical disease. J Low Genit Tract Dis 11(3):158 165 5. Ferris DG, Greenberg MD (1994) Reid s colposcopic index. J Fam Pract 39(1):65 70 6. Reid R, Scalzi P (1985) Genital warts and cervical cancer. VII. An improved colposcopic index for differentiating benign
papillomaviral infections from high-grade cervical intraepithelial neoplasia. Am J Obstet Gynecol 153(6):611 618 7. Ferris DG, Miller MD (1993) Colposcopic accuracy in a residency training program: defining competency and proficiency. J Fam Pract 36(5):515 520 8. Mousavi AS, Fakour F, Gilani MM et al (2007) A prospective study to evaluate the correlation between Reid colposcopic index impression and biopsy histology. J Low Genit Tract Dis 11(3): 147 150 9. Massad LS, Collins YC (2003) Strength of correlations between colposcopic impression and biopsy histology. Gynecol Oncol 89(3):424 428 10. Baum ME, Rader JS, Gibb RK et al (2006) Colposcopic accuracy of obstetrics and gynecology residents. Gynecol Oncol 103(3): 966 970 11. Hong DG, Seong WJ, Kim SY, Lee YS, Cho YL (2010) Prediction of high-grade squamous intraepithelial lesions using the modified Reid index. Int J Clin Oncol 15(1):65 69 12. Massad LS, Jeronimo J, Schiffman M (2008) Interobserver agreement in the assessment of components of colposcopic grading. Obstet Gynecol 111(6):1279 1284 13. Gage JC, Hanson VW, Abbey K, Dippery S, Gardner S, Kubota J et al (2006) Number of cervical biopsies and sensitivity of colposcopy. Obstet Gynecol 108(2):264 272 14. The ASCUS-LSIL Triage Study (ALTS) Group (2003) Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 188(6):1383 1392