Fax: (312) 413-1346
The role of deeper levels and ancillary studies (p16Ink4a and ProExC) in reducing the discordance rate of Papanicolaou findings of high-grade squamous intraepithelial lesion and follow-up cervical biopsies
Version of Record online: 26 FEB 2009
Copyright © 2009 American Cancer Society
Volume 117, Issue 3, pages 157–166, 25 June 2009
How to Cite
David, O., Cabay, R. J., Pasha, S., Dietrich, R., Leach, L., Guo, M. and Mehrotra, S. (2009), The role of deeper levels and ancillary studies (p16Ink4a and ProExC) in reducing the discordance rate of Papanicolaou findings of high-grade squamous intraepithelial lesion and follow-up cervical biopsies. Cancer Cytopathology, 117: 157–166. doi: 10.1002/cncy.20020
- Issue online: 11 JUN 2009
- Version of Record online: 26 FEB 2009
- Manuscript Accepted: 18 DEC 2008
- Manuscript Revised: 16 DEC 2008
- Manuscript Received: 30 JUN 2008
- high-grade squamous intraepithelial lesion;
- cervical biopsy;
- sampling variation
Discordant results of cervical biopsy histology after a cytologic diagnosis of high-grade squamous intraepithelial lesion (HSIL) are often attributed to sampling variation. The purpose of the current study was to determine whether deeper levels and ancillary staining (p16Ink4a and ProExC) reduce the discordant rate.
A total of 246 cases of HSIL were retrieved from the computerized database from 2005 and 2006. Of these cases, 151 were followed by cervical biopsy. There was cytologic-histologic correlation in 87 cases, as defined by the presence of high-grade (2 or 3) cervical intraepithelial neoplasia (HGCIN). For each discordant biopsy (n = 64), 2 deeper levels for hematoxylin and eosin (H&E) were taken at 30-μ and 90-μ depths, and 4 sections for p16Ink4a and ProExC staining were taken at a 60-μ depth. All cytologic and histologic material from these 64 cases was reviewed by 3 cytopathologists. In 2 cases, the original HSIL diagnoses were downgraded and the cases censored from the study.
Fifty-seven of the 62 discordant cases had sufficient tissue for deeper levels and ancillary staining. Two of 57 cases were reclassified to HGCIN. In both of these cases, reclassification was suggested by results of immunostains; however, the H&E sections were necessary for definitive interpretation of the immunostain results.
In the current study, deeper levels and ancillary staining with p16Ink4a and ProExC did not significantly reduce the discordance rate. Although there are many known causes of sampling variation, including factors related to colposcopic technique, regression of infection, and insufficient histologic sectioning, sampling variation remains a valid justification of noncorrelation in women with HSIL followed up by cervical biopsy alone. Cancer (Cancer Cytopathol) 2009. © 2009 American Cancer Society.
There has been extensive discussion on the subject of discordant results of cervical biopsy histology after a cytologic diagnosis of high-grade squamous intraepithelial lesion (HSIL). Some generally accepted causes include: interpretation error, sampling variation, and spontaneous regression.1 Sampling variation is commonly used as an explanation for noncorrelation in women whose Papanicolaou (Pap) tests demonstrate HSIL but whose follow-up cervical biopsies indicate only cervical intraepithelial neoplasia (CIN) of type 1 (CIN 1), koilocytosis, or reactive/inflammatory changes. The purpose of the current study was to determine whether deeper levels and ancillary stains help to reduce the discordance rate between HSIL Pap tests and follow-up biopsies.
MATERIALS AND METHODS
One of the authors (R.D.) searched the Gemini PathNet (Cerner, Kansas City, Mo) computerized database of the University of Illinois Medical Center at Chicago and identified 246 cases of HSIL that were diagnosed between January 1, 2005 and December 31, 2006 by Pap testing. Both liquid-based (ThinPrep; Cytyc Corporation, Marlborough, Mass) and conventional Pap tests were included in this study. As part of routine cytologic-histologic correlation, the computerized medical records were reviewed for follow-up on a quarterly basis. Where procedures were done, procedure types, and pathologic diagnoses were recorded. When records beyond the initial Pap result were unavailable electronically, follow-up information was obtained from the submitting physician's office. Of 246 patients, 151 had follow-up colposcopic-directed cervical biopsies. In the histology laboratory at our institution, once the specimen is processed, the paraffin block is faced and then 1 level is prepared on 1 slide, and a second level 10 μm deeper is prepared on a second slide for review by the pathologist. Of the 151 follow-up cervical biopsies, 87 were concordant with the HSIL cytology. This analysis considered the presence of CIN types 2-3 (HGCIN) as histologic sample-concordant. There were 64 index HSIL Pap slides corresponding to the patients who had follow-up biopsy specimens that did not demonstrate CIN types 2-3. All these slides were reviewed by 3 cytopathologists (O.D., M.G., and S.M.) to confirm that the original diagnosis was HSIL using the 2001 Bethesda System criteria.2 An example of an HSIL cytology specimen that was reviewed and considered to be HSIL is shown in Figure 1. Two cases were reclassified as low-grade squamous intraepithelial lesion (LSIL) and were censored from the study, resulting in a total of 62 cases in which there was HSIL cytology with available follow-up cervical biopsy, not loop electrosurgical excision procedure (LEEP)/cone biopsy, material that was discordant. In none of the 62 cases were there endocervical curettage specimens that demonstrated dysplasia or other pathology. In 17 of the 62 cases, there was >1 biopsy, most commonly 2. Each was subjected to the following protocol. For each discordant biopsy (n = 62), deeper levels were cut. One section for hematoxylin and eosin (H&E) staining was taken at a depth of 30 μm, 4 sections for p16Ink4a (mtm Laboratories AG, Heidelberg, Germany) and ProExC (BD, Franklin Lakes, NJ) staining were taken at a depth of 60 μm, and an additional section for H&E staining was taken at a depth of 90 μm. All histologic material from the discordant cases was reviewed by 4 pathologists (O.D., M.G., S.M., and S.P.). Cervical biopsies were classified as negative for dysplasia, atypical squamous metaplasia, CIN 1, or CIN 2-3. The presence of transformation zone, inflammation, and reactive changes were also noted. Transformation zone was defined as the presence of either coexistent squamous and endocervical mucosa in a single piece of tissue or as the presence of immature squamous metaplasia. Sections subject to p16Ink4a staining were initially reviewed independent of the corresponding H&E-stained sections and were classified as negative for dysplasia, CIN 1, or CIN 2-3 based on interpretation guidelines provided by the manufacturer (CINtec p16INK4a Histology kit; mtm Laboratories, Inc., Westborough, Mass). Specifically, cytoplasmic staining with or without nuclear staining from the basal layer up to the basal third of the epithelium was classified as CIN 1. Staining into the middle or superficial third was considered CIN 2-3. In areas of immature squamous metaplasia, diffuse staining was considered positive, and classification was deferred to the degree of cytologic atypia noted on the H&E sections. ProExC-stained sections were classified as per the p16Ink4a classification system above, except that ProExC is a strictly nuclear stain (BD, Franklin Lakes, NJ). No cytoplasmic staining was observed in any of the cases. All slides were reviewed by consensus of the pathologists. If any slide was considered high grade by any 1 of the 3 methods, all slides from that case were reviewed by consensus, and a determination of the degree of dysplasia was made. Six confirmed cases of HGCIN (2 CIN 2 and 4 CIN 3 cases) were also included as positive controls and demonstrated appropriate staining for both p16INK4a and ProExC.
The interval between the initial Pap and the follow-up cervical biopsy ranged from <1 month to 1 year. Approximately 60% of the discordant cervical biopsies were within 2 months of the initial Pap, 32% were between 2 and 6 months, and 8% were >6 months. Transformation zone was represented in 39 of the 57 biopsies (68%). Information regarding follow-up procedures subsequent to the biopsy was available in 34 (60%) of these cases. Of these, 12 were Pap tests, 7 were biopsies, and 15 were LEEP/cone biopsies. More detailed follow-up information is presented below and in Table 1.
|Original Diagnosis*||Interval†||TZ||Deeper Levels||p16||ProExC||Study Diagnosis||Follow-up||Comment|
|CIN 1||2 mo||+||CIN 1-2||Negative||CIN 2||CIN 2||NA||Fig. 3; physician notified|
|CIN 1||1.5 mo||+||CIN 2||CIN 2||CIN 2||CIN 2||Pap (negative)||Fig. 4; physician notified|
|Atypical immature squamous metaplasia with chronic cervicitis||11 mo||+||Chronic cervicitis with immature squamous metaplasia, indeterminate for dysplasia||Dysplasia of indeterminate grade||Dysplasia of indeterminate grade||Dysplasia of indeterminate grade||LEEP (chronic cervicitis and microglandular hyperplasia)||Fig. 5|
|4 Pap smears (negative)|
|Immature squamous metaplasia with acute and chronic cervicitis||1 mo||+||Immature squamous metaplasia with acute and chronic cervicitis||Negative||Negative||Immature squamous metaplasia with acute and chronic cervicitis||LEEP (HPV)|
|3 Pap smears (negative)|
|CIN 1/HPV||2 wk||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||NA|
|Cervicitis||1 y||+||Negative||Negative||Negative||Negative||Pap (negative)|
|HPV||5 mo||+||CIN 1||CIN 1||CIN 1||CIN 1||Biopsy (CIN 1)|
|Suggestive of HPV||3 wk||−||Negative||Negative||Negative||Negative||Pap (ASCUS)|
|Negative||3.5 mo||−||Negative||Negative||Negative||Negative||Pap (ASC-H)|
|Biopsy (CIN 2)|
|HPV||2 wk||−||Negative||Negative||Negative||HPV||Biopsy (HPV)|
|HPV||1.5 mo||−||Negative||CIN 1?||Negative||HPV||Biopsy (CIN 1/HPV)|
|Suggestive of HPV||2 wk||−||Negative||Negative||Negative||Negative||LEEP (CIN 2/HPV)|
|Negative||2 mo||+||Negative||Negative||Negative||Negative||LEEP (negative)|
|CIN 1/HPV||2 wk||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||LEEP (CIN 1/HPV)|
|Negative||7 mo||−||CIN 1||Negative||CIN 1||CIN 1||NA||Physician notified|
|CIN 1/HPV||1 mo||+||CIN 1/HPV||CIN 1||Negative||CIN 1/HPV||LEEP (CIN 1-2)|
|CIN 1||5.5 mo||+||Negative||Negative||Negative||Negative||Pap (negative)|
|HPV||2 wk||−||Negative||Negative||Negative||Negative||Pap (HSIL)|
|HPV||2 mo||−||Negative||Negative||Negative||Negative||Pap (negative)|
|Koilocytosis||1 mo||+||CIN 1||Negative||Negative||CIN 1||LEEP (CIN 2/HPV)|
|CIN 1/HPV||1 mo||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||2 Pap smears (negative)|
|HPV||11 mo||+||Negative||Negative||Negative||Negative||Pap (ASCUS)|
|Negative||4.5 mo||+||Negative||Negative||Negative||Negative||2 Pap smears (negative)|
|CIN 1/HPV||3 mo||+||CIN 1||CIN 1||Negative||CIN 1/HPV||Pap (ASCUS)|
|HPV||9 mo||+||Negative||Negative||Negative||Negative||Pap (ASCUS)|
|Negative||2 mo||−||Negative||Negative||Negative||Negative||Biopsy (CIN 1)|
|HPV||7 mo||+||Negative||Negative||Negative||Negative||LEEP (HPV)|
|HPV||3 mo||+||Negative||Negative||Negative||Negative||LEEP (CIN 1/HPV)|
|CIN 1/HPV||1 mo||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||LEEP (CIN 3)|
|Suggestive of HPV||4 mo||+||Negative||Negative||Negative||Negative||NA|
|HPV||9 mo||+||Negative||Negative||Negative||Negative||Pap (LSIL)|
|Negative||2 wk||−||Negative||Negative||Negative||Negative||Biopsy (CIN 2-3)|
|CIN 1/HPV||3 wk||+||CIN 1||CIN 1||CIN 1||CIN 1||NA|
|HPV||2 mo||+||Negative||Negative||Negative||Negative||LEEP (CIN 2/HPV)|
|Suggestive of HPV||Concurrent||−||Negative||Negative||Negative||Negative||LEEP (CIN 2/HPV)|
|CIN 1/HPV||2 mo||+||CIN 1||CIN 1||Negative||CIN 1/HPV||NA|
|CIN 1/HPV||2 mo||+||CIN 1||CIN 1||Negative||CIN 1||NA|
|HPV||2 mo||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||LEEP (HPV)|
|CIN 1/HPV||Concurrent||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||NA|
|CIN 1/HPV||Concurrent||+||CIN 1||CIN 1||CIN 1||CIN 1/HPV||LEEP (CIN 1/HPV)|
|Focal CIN 1||1 mo||+||Negative||Negative||Negative||Negative||LEEP (CIN 3)|
|CIN 1||3 mo||+||CIN 1||CIN 1||CIN 1||CIN 1||NA|
|Negative||4 mo||−||Negative||Negative||Negative||Negative||Biopsy (CIN 2)|
Of the 62 cases with discordant follow-up cervical biopsy material and true HSIL cytology, there was sufficient tissue in 57 cases to obtain deeper levels, p16Ink4a, and ProExC staining. Figure 2 illustrates the approach to the study. Results are summarily presented in Table 1.
Two cases were upgraded from CIN 1 to HGCIN. In 1 case, the first (30 μm) H&E level demonstrated CIN 1. The deepest (90 μm) H&E level demonstrated dysplasia involving greater epithelial thickness, equivocal for CIN 2 (Fig. 3, Top). The p16Ink4a-stained section was completely negative. The ProExC-stained section, however, strongly supported a diagnosis of CIN 2 (Fig. 3, Bottom). Given the histology observed in the deepest H&E and the ProExC result, the case was reclassified as HGCIN, concordant with the original cytologic diagnosis of HSIL. This case illustrates the need for the judicious correlation of H&E and ancillary study findings. There was no further follow-up on this patient at the time of publication. In the second case that was upgraded, both the 30-μm and 90-μm H&E-stained sections supported a diagnosis of CIN 2. The p16Ink4a-stained section also supported the diagnosis of high-grade dysplasia (Fig. 4, Top). In the ProExC-stained section, there also were strongly positive nuclei scattered halfway up the epithelium (Fig. 4, Bottom). A follow-up Pap performed 1 year later was negative for intraepithelial lesion or malignancy (NILM). In both of these cases, reclassification was suggested by results of immunostains; however, the H&E sections were necessary for definitive interpretation of the immunostain results for both p16Ink4a and ProExC. In both cases, the physicians who submitted the cervical biopsy specimens were contacted by telephone and directly informed of the diagnostic reclassification.
The ancillary stains also proved particularly useful in distinguishing immature squamous metaplasia from dysplasia at the transformation zone in cases with abundant superimposed chronic inflammation. Two cases, different from the ones presented above, demonstrated immature squamous metaplasia. The first demonstrated cytologically atypical immature squamous metaplasia with marked chronic inflammation. Both of the ancillary-stained sections suggested dysplasia of indeterminate grade (Fig. 5, Top and Bottom). This case was ultimately not upgraded to high-grade dysplasia because the histologic features in the H&E sections did not definitively warrant a diagnosis of HGCIN. A subsequent cone biopsy (LEEP) indicated chronic cervicitis and microglandular hyperplasia. Four subsequent annual Pap tests were NILM. In the second case, the H&E-stained sections demonstrated immature squamous metaplasia against a background of acute and follicular cervicitis with reactive epithelial changes. Both of the ancillary-stained sections were completely negative. A follow-up cone biopsy (LEEP) demonstrated human papillomavirus (HPV) effect without dysplasia. Three subsequent annual Pap tests were NILM.
Of the 55 cases that were not upgraded and therefore remained discordant with the original HSIL Pap, further follow-up was available in 31 cases and consisted of Pap in 11 cases, repeat cervical biopsy in 7 cases, and LEEP/cone in 13 cases. One of the 11 Pap tests demonstrated HSIL; 3 of the 7 biopsies and 7 of the 13 LEEP/cone specimens showed HGCIN (see also Table 1). These results in particular, although too small in number and varied with regard to the type of specimen to permit meaningful statistical analysis, are interesting because they confirm the uncomfortable reality of true HSIL not corroborated by a subsequent cervical biopsy.
Several studies have analyzed the discrepancies between a cervical cytology diagnosis and follow-up biopsies.3-7 The frequency of discrepancies was found to range from 11% to 32%. For each of these studies, sampling error was listed as the major cause, to which 61% to 93% of discrepancies were attributed. Actually, we favor the use of the term “sampling variation” over “sampling error,” because we believe that the former better describes the procedural challenges that clinicians face when attempting to acquire the most diagnostically informative material with a high degree of consistency.
In the current study, we retrospectively performed deeper levels and ancillary stains on follow-up cervical biopsies to determine whether these additional steps would reduce the rate of discordance between HSIL cytology and follow-up cervical biopsies. p16Ink4a is a useful biomarker for cervical dysplasia and a valuable adjunct in the diagnosis of HGCIN lesions. It is a cyclin-dependent kinase inhibitor that functions as a tumor suppressor and has been found to be overexpressed in precancerous and malignant cervical lesions. More specifically, high-risk HPV (hrHPV) E7 oncoprotein-mediated inactivation of retinoblastoma protein results in up-regulation of p16Ink4a by a negative feedback mechanism.8, 9 p16Ink4a has been studied extensively in cervical specimens. One recent study of p16Ink4a as a diagnostic tool in the triage of atypical squamous cells of undetermined significance (ASCUS) Pap tests found that p16Ink4a scoring alone could not be used to recategorize ASCUS to NILM.10 In another study of the value of p16Ink4a immunocytochemistry in sorting discrepant cases (ie, Pap-positive HSIL/biopsy-negative HSIL), this marker was reported to have 100% sensitivity and 100% negative predictive value using cone biopsy as the gold standard. In the same study, adding p16Ink4a immunocytochemistry to the evaluation of ASCUS cytology was shown to improve the specificity of colposcopy.11
ProExC is a cocktail of minichromosome maintenance protein 2 (MCM2) and topoisomerase II alpha protein (TOP2A), which together play an important regulatory role in eukaryotic DNA replication. For example, hrHPV E6 and E7 oncoprotein-mediated bypass of critical cell-cycle checkpoints results in a prolonged and aberrant S-phase induction cycle. During the transcriptional activation of the aberrant cell cycle, levels of MCM2 and TOP2A increase in the proliferating cells. MCM2 and TOP2A have been shown to be overexpressed in a wide variety of dysplastic and malignant processes, including hrHPV-related cervical neoplasia. The overexpression of these proteins in morphologically abnormal cells, as demonstrated by a moderate-to-intense brown nuclear staining pattern using immunohistochemical techniques, is indicative of the presence of aberrant S-phase induction.12 A recent study by Tambouret et al concluded that when ProExC immunocytochemistry results were combined with any level of cytologic atypia, the sensitivity for CIN 2+/HSIL was 92% and the specificity was 84%, and, furthermore, that ProExC could enhance the sensitivity and specificity of cervical cytology for subsequent CIN 2+/HSIL, as well as help identify those lesions most likely to progress.13 A recent study by Badr et al14 corroborates our finding that ProExC is reliable as a marker for high-grade CIN, can be applied to tissue sections, and can be used to triage atypical squamous metaplasia cases.
One other noteworthy recent study directly compared p16Ink4a and ProExC in histologic sections of cervical tissue. Shi et al15 evaluated 62 cervical biopsy specimens and found p16 to be a more sensitive and specific marker for HSIL, ProExC to be better for the detection of LSIL, and p16/ProExC to be best for detecting both HSIL and LSIL.
It is generally recognized that factors beyond the control of the pathology laboratory also significantly impact rates of discordance between cytologic, biopsy, and excisional tissue samples. The sensitivity of the colposcopic detection of HGCIN may be highly variable.16 One study of atypical squamous cells, in which a high-grade lesion could not be ruled out (ASC-H) concluded that a substantial percentage of cases with biopsy-proven CIN 2 or higher after cytologic interpretations of ASC-H required >1 colposcopy for a definitive diagnosis of a high-grade dysplastic lesion. Also noted was that if a CIN 2 or higher lesion was present, it usually could be diagnosed in a biopsy performed within 1 year of the ASC-H interpretation.17
Regression must also be considered as a factor that might contribute to the discordance rate. Regression has been defined by at least 1 author as significant when the interval of time between the Pap test and the follow-up biopsy is >6 months.18 By this metric, we do not consider regression to be a significant factor in the current investigation because the majority of our patients (66%) had follow-up cervical biopsies within 2 months of the initial HSIL cytology. However, we recognize that it is impossible to accurately address the question of regression in this study.
In our study of a 2-year period, 149 women with HSIL by Pap test were followed by colposcopic cervical biopsy. At the end of the study, only 60% (89 of 149) of the women with follow-up biopsies had concordant biopsies. Of the women with discordant biopsies, 57 had sufficient tissue to permit the deeper levels and ancillary stains described earlier. Only 2 cases were ultimately upgraded to HGCIN.
Therefore, what should be done to decrease the discordance rate between HSIL cytology and follow-up cervical biopsies? We have addressed and putatively disputed error related to the histologic preparation and interpretation of cervical biopsy specimens. In the current study, the use of deeper levels and ancillary stains did not permit us to decrease the discordant biopsy rate. There are some studies that offer other ways to address this problem. For example, standardizing colposcopy examination might be expected to improve the discordance rate. One large study in particular demonstrated that the International Federation for Cervical Pathology and Colposcopy (IFCPC) classification method is an effective colposcopic screening method used in correlation with a histologic sample.19 This study was conducted on a cohort of 3040 women who were screened by means of Pap test, unaided visual inspection, and high-risk HPV testing. All colposcopic examination results with abnormal findings and with biopsy confirmation (n = 468) were recorded, reviewed by 2 blinded colposcopists according to the IFCPC nomenclature, and included in the analysis. The IFCPC terminology was easily reproduced by the 2 observers. Colposcopy had a sensitivity of 86% and a specificity of 30.3% in distinguishing a healthy cervix from that with CIN/carcinoma, and a sensitivity of 61.1% and a specificity of 94.4% in distinguishing a healthy cervix/low-grade lesions (CIN 1) from high-grade lesions (CIN 2 of 3)/carcinoma. Colposcopic abnormalities within the transformation zone and large lesions were found to be more closely related to high-grade lesion/carcinoma, whereas a sharp outer border, multiple colposcopic abnormalities, and iodine negativity were not reported to be statistically correlated with severe lesions.
Despite current and emerging techniques in the area of cervical neoplasia and cervical cancer screening, there appears to be no consistently reliable gold standard for cytology. We conclude from the results of the current study that the discordance rate is not significantly lowered by histopathologic modalities alone. Although ancillary stains appear to help in equivocal cases (ie, atypical squamous metaplasia vs high-grade CIN at the transformation zone), they cannot be used independently without H&E correlation. Unless there are changes in the approach to the management of women with HSIL cytology, sampling variation will likely remain a common justification of discordance in the quality control/improvement documents of cytology laboratories nationwide.
Conflict of Interest Disclosures
The authors made no disclosures.
- 2SolomonD, NayarR, eds. The Bethesda System for Reporting Cervical Cytology.2nd ed. New York, NY: Springer-Verlag; 2004.
- 3An analysis of discrepancies between the cervical cytologic diagnosis and subsequent histopathologic diagnosis in 1260 cases [abstract]. Acta Cytol. 1994; 39: 812., .
- 12TriPath Molecular Oncology. Available at: http://www.bd.com/tripath/products/proexc/sphase.asp#a Accessed on June 1, 2008.
- 17Papanicolaou test interpretations of “atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion”: an investigation of requisite duration and number of colposcopic procedures to a definitive diagnosis of high-grade dysplasia in routine practice. Cancer (Cancer Cytopathol). 2007; 111: 477-481., , .