Predicting Intracavitary Lesions Based on Stringent Histologic Criteria to Diagnose Endometrial Polyps Amin A. Ramzan, MD 1 ; Paulette Mhawech-Fauceglia, MD 2 ; Brian Kay, MD 2 ; Teodulo Meneses, MD 2 ; Shivani Aggarwal, MS 3 ; Yan Wang, MD 2 ; and Christina Dancz, MD 1 Abstract Objective: To determine whether use of stringent histologic criteria to diagnose endometrial polyps can increase the clinical predictive value for intracavitary lesions and to describe the inter- and intra-observer reliability of these criteria. Methods: A diagnostic study of 104 participants who underwent office hysteroscopy or sonohysterography following an endometrial biopsy to measure sensitivity, specificity, and intra- and inter-observer reliability. Histology slides were reviewed by three pathologists to reach a final diagnosis that would be correlated with the presence of lesions on office imaging as the true positive. The diagnosis of polyp was made when at least 2 of the 3 following histologic criteria were met: (1) thick-walled blood vessels; (2) irregularly shaped and positioned glands; and (3) stroma altered by fibrosis/excessive collagen Results: Using standard biopsy results, 57 participants had polyps compared to 47 with other benign diagnoses. 38 participants had lesions on imaging, resulting in a PPV of 33.3%, a NPV of 59.6%, sensitivity of 50.0% and specificity of 42.4%. Intra-observer reliability of repeat evaluations by the same pathologist to identify polyps (if 2 of the 3 criteria were met) was fair to good for each of the pathologists (Fleiss' Kappa 0.50 to 0.66). When the consensus among all three pathologists was used, the PPV and NPV improved to 43.7% and 66.7%, respectively, while the sensitivity and specificity changed to 36.8% and 72.7%, respectively. Conclusions: Use of stringent histologic criteria to diagnose endometrial polyps can improve the ability to predict the presence of intracavitary lesions on office imaging. Date of Submission: 09/09/15, Revised 10/21/15 1 Department of Obstetrics and Gynecology; 2 Department of Pathology; 3 Department of Preventive Medicine, University of Southern California, Los Angeles, California *Correspondence to Amin Ramzan, MD, Department of Obstetrics and Gynecology, Los Angeles County Medical Center, University of Southern California, 1200 N. State St., IPT C3F107, Los Angeles, CA 90033; Tel: +1-323-409-8898; Fax: +1-323- 441-7219; Email: aramzan@ucla.edu Conflict of Interest Disclosure: The authors listed above have no commercial or financial relationships relevant to the subject of this manuscript. 6.Keywords Endometrial Polyp, Histologic Diagnosis, Intracavitary Lesion E INTRODUCTION ndometrial polyps are common growths of endometrial glands and stroma that protrude into the uterine cavity. Their prevalence has been quoted as high as 20-25% in premenopausal women, although they may be asymptomatic i. Symptomatic patients commonly present with abnormal uterine bleeding, including menorrhagia and postmenopausal bleeding. The majority of polyps are benign, but a 3.1-3.5% rate of malignancy argues for the removal and histologic evaluation of most, if not all diagnosed polyps ii-iii. Women presenting with persistent abnormal uterine bleeding are typically initially evaluated with an endometrial pipelle biopsy and a pelvic ultrasound. If the biopsy is diagnostic of polyp or the pelvic ultrasound is suggestive for an intracavitary lesion, then the patient is triaged to office hysterosocopy or sonohysterography. The sensitivity and specificity of sonohysterography has been reported to be 93.1% and 93.9%, respectively iv. While typically well tolerated by patients, these diagnostic procedures add time and risk for the patient, can be quite anxiety producing and generate a significant cost to the medical system. The sensitivity, specificity and predictive value of a histologic diagnosis of polyp on pipelle endometrial biopsy are unknown. The three histologic criteria most commonly used to diagnose endometrial polyp are: (1) thick-walled blood vessels (Figure 1); (2) irregularly shaped and positioned glands (Figure 2); and (3) stroma altered by fibrosis or excessive collagen (Figure 3) v. Although these histologic criteria are widely acknowledged, adherence to stringent criteria, defined as meeting at least two of the three histologic criteria, varies widely by individual pathologist. There have been no studies to date demonstrating the positive predictive value of stringent histologic criteria to diagnose clinically relevant endometrial polyps. The primary objective of this study is to determine whether use of stringent histologic criteria when diagnosing
endometrial polyps can increase the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) to detect clinically relevant intracavitary lesions. Secondary objectives are to describe the inter- and intra-observer reliability of these stringent histologic criteria. Figure 1. Blood vessels with thickened walls are seen here in an arborizing pattern. Figure 2. Endometrial glands showing irregular ectatic shapes and luminal dilation. participants were identified from a list of consecutive patients referred for sonohysterography or hysteroscopy using the procedural schedule log of the gynecology clinic at LAC+USC Medical Center. The time period of participants included in the study spanned from 8/24/2012 to 10/15/2013. Eligibility criteria were: women over 18 who underwent endometrial pipelle biopsy for any indication and subsequently underwent evaluation with office hysteroscopy or sonohysterography in the gynecology clinic. Participants with biopsies showing malignancy or endometrial hyperplasia were excluded, as were cases where the biopsy was unavailable, or where the biopsy was from a dilation and curettage specimen. Additionally, cases for which imaging findings revealed a distorted cavity secondary to submucosal myomas were excluded. Among the eligible participants identified, medical records were abstracted to collect clinical information for basic demographics, indication for biopsy, referral endometrial biopsy histology and impression at the time of office hysteroscopy or sonohysterography. Cases with an initial histologic diagnosis of polyp or fragments suggestive of polyp were defined as being positive for polyp, while those with a diagnosis of secretory, proliferative, menstrual phase, atrophic, or any other benign, non-hyperplastic endometrium were defined as being negative for polyp. Participant demographics included age, race, ethnicity, body mass index (BMI), and menopausal status. The indications for biopsy were categorized as heavy/prolonged menses, oligomenorrhea, post-menopausal vaginal bleeding, thickened endometrial echo complex (EEC) on ultrasound, or other indication not listed. Endometrial biopsy results were classified as endometrial polyp, proliferative endometrium, secretory endometrium, or benign endometrium not otherwise specified (NOS). Whether the study participant had been on any form of hormonal treatment prior to endometrial biopsy was also recorded. The presence or absence of an intracavitary lesion at the time of office imaging was noted, in addition to the narrative impression of the provider. For sensitivity/specificity analyses, a positive test was defined as office imaging with a finding of polyp necessitating surgical intervention. Office hysteroscopy and sonohysterography were performed by resident obstetrics and gynecology (Ob/Gyn) physicians under the direct supervision of an attending Ob/Gyn physician. Figure 3. Stromal fibrosis with increased collagen deposition is present in association with irregularly dilated glands. METHODS After Institutional Review Board (IRB) approval was obtained at the University of Southern California (USC), 7 For determination of intra- and inter-observer reliability, the endometrial biopsy histology slides for each study participant were retrieved from the pathology archive and reviewed independently by three pathologists. One pathologist was an experienced gynecological pathologist (YW), one was senior gynecological pathologist (PMF), and one was a surgical pathology fellow still in training (TM). Each pathologist was blinded to the initial diagnosis and any clinical information. Specimens were evaluated for the presence of each of the three histologic criteria: 1) thick-walled blood vessels, 2) irregularly shaped and positioned glands, and 3) stroma altered by
fibrosis/excessive collagen. One month later this process was repeated, with pathologists blinded to their initial evaluation. If both evaluations were equivalent for a given specimen, then this counted as the final diagnosis for that particular pathologist. However, if there was intra-observer disagreement, the pathologist reviewed the specimen a third and final time to determine a final diagnosis. If the three pathologists independently reached the same final diagnosis, then this represented the consensus diagnosis. If there was disagreement, then the histology was reviewed at a conference with all three pathologists present, and a consensus diagnosis was reached after a final review of the histology. An initial power calculation revealed that to detect a change in diagnosis from positive to negative in 20% of participants, and a change from negative to positive in 1% of participants, with 80% power and an α of 0.05, a total of 44 participants per diagnosis group (positive polyp and negative polyp) was required. Basic demographic characteristics were assessed for all study participants, by diagnosis status (polyp versus non-polyp), and by lesion status (yes/no) using a t-test for continuous variable, and a Chi-square test or Fisher s exact test for categorical variables. Intra-observer reliability was determined using Fleiss kappa statistic, and inter-observer reliability was determined using Kendall s coefficient of concordance. Sensitivity, specificity, PPV, and NPV were determined using finding of intracavitary lesion on office imaging as the gold standard. A p-value of 0.05 was considered significant for all statistics. A kappa coefficient of over 0.75 was considered excellent, 0.40 to 0.75 was fair to good, and below 0.40 was considered poor. All statistics were performed using SAS version 9.3 vi. There were 38 participants with a finding of intracavitary lesion on office imaging (hysteroscopy or sonohysterography) (Table 2). Of those, 19 had a referral endometrial biopsy of endometrial polyp. The PPV for identifying an intracavitary lesion after a biopsy result of polyp was 33.3% (95% CI: 21.40% - 47.06%). The NPV for absence of an intracavitary lesion after a non-polyp diagnosis on an endometrial pipelle biopsy was 59.6% (95% CI: 44.27% - 73.63%). RESULTS A total of 314 patient records were reviewed in the time period. Of the 104 participants that met inclusion criteria and were included in the analyses, 57 participants had polyp initially diagnosed on endometrial biopsy and 47 participants had a non-polyp histologic diagnosis. The participant demographics are shown in Table 1. Among all participants, the mean age was 44.9 years and the majority was Hispanic (83.6%). The mean BMI was 31.8 kg/m2 and most women were pre-menopausal (87.5%). The most common indication for endometrial biopsy was heavy/prolonged bleeding (87.5%), followed by post-menopausal vaginal bleeding (9.6%). Demographic characteristics were compared based on initial diagnosis of polyp on endometrial biopsy and presence of intracavitary lesion on office imaging. Post-menopausal vaginal bleeding was significantly associated with an initial diagnosis of endometrial polyp on biopsy (p-value < 0.05). Hispanic ethnicity was also associated with an initial diagnosis of polyp, but this should be viewed with caution given the overwhelming Hispanic ethnic make-up of the patient population. Table 3 summarizes the intra-observer reliability for each pathologist. Overall the intra-observer reliability between the first and second review for each pathologist was fair to good, with Fleiss s kappa statistic ranging from 0.50 (95% CI: 0.35 0.64) to 0.66 (95% CI: 0.52 0.80). The inter-observer reliability for the initial review of each specimen was approaching excellent, with a Kendall s coefficient of concordance of 0.74, p-value < 0.0001 (data not shown). 8
On review of the consensus diagnosis based on stringent histologic criteria, the most common consensus result (69.2%) was negative for polyp (0 or 1 criteria met). Only 9.6% of specimens met all three criteria. The remainder of the participants met various combinations of two criteria (21.2%). The consensus diagnosis was then used to determine the sensitivity, specificity, PPV and NPV based on the presence of absence of an intracavitary lesion on office imaging (Table 4). One hundred four participants had histology data and consensus diagnosis data. In comparison to the original biopsy result, use of the stringent criteria results in an increased PPV of 43.8% (95% CI: 26.36% 62.34%) and an increased NPV of 66.7% (95% CI: 54.57% 77.34%), with a sensitivity of 36.8% and a specificity 72.7%. A more rigorous definition of polyp, in which a specimen is positive for all three criteria, further increases the PPV to 80.0% (95% CI: 44.39% 97.48%) with a specificity of 97.0% (95% CI: 89.48% 99.63%), but comes at the cost of a reduced sensitivity of 21.1% (95% CI: 9.55% 37.32%) with a relatively unchanged NPV of 68.1% (95% CI: 57.67% 77.33%). specificity increased to 72.7% while the sensitivity decreased to 36.8%. When stringent criteria were used to diagnose polyp, the intra-observer reliability was fair to good for all three pathologists in our study (Kappa statistic of 0.50 to 0.66). The most experienced pathologist had the highest intra-observer reliability, suggesting that experience may improve intra-observer reliability. Inter-observer reliability was similarly good, with a Kendall s coefficient of concordance of 0.74, indicating that use of stringent criteria is generally reproducible among pathologists. Additional criteria have been advocated for in the diagnosis of endometrial polyp. Kim described parallel arrangement of the endometrial glands long axis to the surface epithelium (PGE) as an additional criterion viii. In his study of 76 endometrial polyps, 51 specimens exhibited PGE, with this finding significantly more common in pre-menopausal women. An additional proposed criterion that is more relevant to polyps removed intact is presence of surface epithelium on three sides ix. Both of these criteria are more likely to be encountered with hysteroscopically removed specimens or dilation and curettage (D&C) specimens versus the fragmented pipelle biopsy specimens. These criteria were so infrequently seen in our pipelle biopsy specimens that we were not able to assess their predictivity. DISCUSSION Our study is the first to assess the use of stringent histologic criteria for endometrial polyps to determine the predictive value for intracavitary lesions at the time of office hysteroscopy or sonohysterography. The criteria utilized in this study (thick-walled blood vessels; irregularly shaped and positioned glands; stroma altered by fibrosis or excessive collagen) are well established vii. However, it is unknown how commonly pathologists use these criteria in a strict and methodical fashion to aid in the diagnosis of endometrial polyp. In our study population, the initial diagnosis of polyp was made in 57/104 (54.8%) of the cases; when stringent criteria were applied, only 32/104 (30.8%) cases qualified as diagnostic of polyp. The initial PPV of a diagnosis of endometrial polyp for a clinically significant intracavitary lesion was 33.3%. When applying stringent criteria defining endometrial polyp, the PPV improved to 43.8% and the NPV improved to 66.7%, the 9 There are obvious potential benefits to both patients and the healthcare system in improving the accuracy of the histologic diagnosis of endometrial polyp. A decrease in false positive diagnoses of polyp, such as that observed in this study with the use of stringent criteria would decrease the number of unnecessary, invasive, office procedures. These procedures are expensive, time consuming, uncomfortable and anxiety provoking for patients. Furthermore, as office imaging is neither 100% sensitive nor specific, some of these patients ultimately end up having unnecessary surgical procedures. Using stringent criteria to define polyp in this study resulted in a slightly increased rate of false negative diagnoses of polyp. In such cases, a false negative diagnosis on histology could potentially delay diagnosis. These women are unlikely to respond to medical management, and are therefore likely to eventually undergo office imaging based on clinical indicationss. The potential clinical implications of such a delay are unknown. Our study has several strengths, including the inclusion of patients that originally had polyp and non-polyp diagnosis on their histologic report. Also, two of the three pathologists that reviewed specimens (YW, PMF) are pathologists with specialty training in gynecological pathology. Finally, the multiple measurements, in which each pathologist reviewed every specimen at least twice, allowed assessment of intra- and inter-observer reliability of diagnosis. Limitations of the study include its retrospective design and the relative ethnic homogeneity of our study population. We are unable to comment on the true clinical implications or cost benefit of
increasing the specificity of polyp diagnosis at the expense of a decrease in sensitivity. Another inherent limitation is the theoretic possibility that a small intracavitary polyp is removed in its entirety by the pipelle biopsy, and subsequently not visualized on office imaging. In conclusion, the use of stringent criteria in the diagnosis of endometrial polyps improves the PPV, NPV and specificity for the presence of intracavitary lesions on office imaging, though at a slight decrease in sensitivity. We advocate limiting the diagnosis of endometrial polyp to specimens that meet at least two of three of the criteria. References i. Van bogaert LJ. Clinicopathologic findings in endometrial polyps. Obstet Gynecol. 1988;71 (5): 771-3. ii. Baiocchi G, Manci N, Pazzaglia M et-al. Malignancy in endometrial polyps: a 12-year experience. Am. J. Obstet. Gynecol. 2009;201 (5): 462.e1-4. iii. AAGL practice report: practice guidelines for the diagnosis and management of endometrial polyps. J Minim Invasive Gynecol. 2012;19(1):3-10. iv. Kamel HS, Darwish AM, Mohamed SA. Comparison of transvaginal ultrasonography and vaginal sonohysterography in the detection of endometrial polyps. Acta Obstet Gynecol Scand. 2000;79 (1): 60-4. v. Kindelberger DW, Nucci MR. Gynecologic Pathology. 1st Ed. Elsevier; 2009. Chapter 6, Benign Endometrium; p. 197-232. vi. Fleiss JL. Statistical methods for rates and proportions. John Wiley & Sons; 1973. vii. Anderson MC, Robboy SJ, Russell P. Endometritis, metaplasias, polyps and miscellaneous disorders. In: Robboy SJ, Anderson M, Russell P, eds. Pathology of the Female Genital Tract. London: Churchill Livingstone, 2002:298 300. viii. Kim KR, Peng R, Ro JY et-al. A diagnostically useful histopathologic feature of endometrial polyp: the long axis of endometrial glands arranged parallel to surface epithelium. Am. J. Surg. Pathol. 2004;28 (8): 1057-62. ix. Silverberg SG. Problems in the differential diagnosis of endometrial hyperplasia and carcinoma. Mod Pathol. 2000;13(3):309-27. 10