Context.—The protein p16Ink4a is overexpressed in cervical lesions associated with high-risk human papillomavirus (HPV) subtypes 16 and 18, but not in low-risk HPV subtypes 6 and 11 or non–HPV-associated cervical lesions.
Objective.—To determine whether p16Ink4a expression in equivocal cervical lesions helps distinguish atypical non-HPV changes from HPV-related changes.
Design.—One hundred ninety-one cervical lesions, including 81 cervical intraepithelial neoplasia 1, 52 squamous metaplasia, 33 cellular features suggestive of HPV-related change, 9 reserve cell hyperplasia, 4 microglandular hyperplasia, and 12 inflammatory cervicitis, were randomly selected from archival cervical biopsy specimens. All 191 samples were studied with p16Ink4a (JC8 monoclonal antibody). Reactivity for p16Ink4a was scored on a 3-tier system as follows: negative, 0% to 5% cells reactive; focal/scattered positive, greater than 5% and less than or equal to 80% cells reactive; diffuse positive, greater than 80% cells reactive. Reactivity was based on normal/reactive cervical specimens where anti-p16 antibody was negative/weakly expressed in non–cervical epithelial cells. Cervical intraepithelial neoplasia 1 lesions not reactive for p16Ink4a were investigated for the presence of high-risk HPV by real-time polymerase chain reaction.
Results.—No p16Ink4a reactivity was detected in the cervical lesions associated with atypical non-HPV change. Eleven of the cervical intraepithelial neoplasia 1 lesions showed focal/scattered reactivity expression for p16Ink4a, and 19 of the CIN 1 lesions had diffuse reactivity. Fifty of 51 of the CIN 1 lesions negative for p16Ink4a were real-time polymerase chain reaction negative for the presence of high-risk HPV; 1 was real-time polymerase chain reaction positive for high-risk HPV.
Conclusions.—The data support the routine use of p16Ink4a immunohistochemical evaluation of cervical biopsy specimens for better discrimination of non–HPV-associated lesions from HPV-related lesions.
Current literature suggests that the protein p16Ink4a may be used as a biomarker for the identification of cervical intraepithelial neoplasia (CIN) as well as a test to decrease interobserver variability in making the diagnosis of CIN lesions.1–9 p16Ink4a is a cell-cycle regulatory protein that negatively influences cell proliferation through a reciprocal relationship with another tumor suppressor protein, retinoblastoma gene product (pRb).1,5 The action of pRb is to inhibit cells from entering the S phase of the cell cycle and proliferating.5 It is modified via phosphorylation by cyclin D1 complexed with other cyclin-dependent kinases (cdk 4 through 6).1–3,5 In the presence of the human papillomavirus (HPV) oncoprotein E7, pRb is inactivated.5 This inactivation of pRb allows for the transcription of genes required for DNA replication and the inappropriate shifting of the cell cycle past the G1/S restriction point into the S phase.2,3,5 As pRb is functionally inactivated by HPV oncoprotein E7, there is a reciprocal overexpression of p16Ink4a.1–3,5
Marked overexpression of the p16Ink4a protein can be demonstrated immunohistochemically using monoclonal antibodies. Studies of cervical carcinomas and preneoplastic lesions of the cervix resulting from high-risk human papillomavirus (HR-HPV) subtypes, specifically subtypes 16 and 18, have demonstrated a strong and diffuse nuclear and cytoplasmic immunoreactivity of p16Ink4a in these lesions.2,3,5,8–10 Subtypes 16 and 18 are representative of HR-HPV and are the most clinically important HPV subtypes because infection by these subtypes is associated with an increased risk of cervical carcinoma.2,4 Cervical lesions associated with low-risk HPV subtypes, specifically HPV 6 and 11, express focal and weak p16Ink4a immunohistochemical reactivity and are not associated with an increased risk of cervical carcinoma.2,8,9
The utilization of p16Ink4a immunohistochemistry in the assessment of cervical intraepithelial lesions may be applicable in reducing interobserver variability in the diagnosis of CIN. Studies on interobserver variability in the histological grading and diagnosis of CIN lesions have demonstrated an overall poor interobserver agreement.11–14 Although experienced pathologists did well with invasive lesions and moderately well with high-grade CIN lesions (CIN 2/3), significant disagreement occurred in differentiating CIN 1 lesions from reactive squamous proliferations.11,15,16
It is recognized that certain metaplastic and reactive lesions unrelated to HPV or neoplasia can mimic CIN 1 lesions.7 Some of the histologic findings that may resemble CIN 1 include squamous metaplasia, reserve cell hyperplasia, microglandular hyperplasia, and inflammatory conditions such as cervicitis.17
In order to avoid overinterpretation of CIN 1 in nonneoplastic cervical biopsy specimens, strict criteria for the diagnosis of CIN 1 lesions on cervical biopsy and cone specimens have been proposed.17 In addition to the presence of koilocytes, which has been correlated with the presence of fully assembled intranuclear HPV particles,18 other criteria, including the maintenance of an intact basement membrane with a loss of the usual maturation of the squamous epithelial cells at the surface, a 3-fold increase in nuclear size, conspicuous nuclear hyperchromasia with irregular contour of nuclear membranes, and multinucleation of cells in some cases, have been suggested.15 Despite these suggested criteria, however, discrimination of CIN 1 lesions from nonneoplastic equivocal cervical lesions is still not optimal.15,16,18
A study evaluating the use of strict histologic criteria for the diagnosis of CIN 1 lesions demonstrated a false-positive rate, or overinterpretation of CIN 1, in up to 34% of cervical biopsy specimens that, via molecular methods, were negative for HPV infection.17 This was, however, an improvement over using the presence of koilocytosis alone, where a false-positive or overdiagnosis rate of CIN 1 occurred in 62.3% of the cervical biopsy cases.17 The authors' proposed explanation for this was that inflammatory, environmental, or processing artifacts could be responsible for the look-alike lesions.17 Some authors have suggested the use of ancillary molecular laboratory techniques, such as in situ hybridization, for better discrimination of these CIN 1 lesions from nonneoplastic equivocal cervical lesions as a means of quality control.19 Other authors advocate the use of biomarkers, including p16Ink4a, recognizing that various studies looking at p16Ink4a have shown focal immunohistochemical reactivity for p16Ink4a in normal cervical epithelium, squamous metaplasia, and some inflammatory conditions.6–10,20
The purpose of this study is to evaluate p16Ink4a expression in equivocal cervical squamous lesions that may be overdiagnosed as CIN 1 and determine whether p16Ink4a can assist in distinguishing non–HPV-related changes, such as squamous metaplasia, from HPV-related changes like CIN 1.
MATERIALS AND METHODS
The records of the Department of Pathology at the University of Florida/Shands Hospital at the University of Florida (Gainesville) were searched using SNOMED codes for CIN 1, immature squamous metaplasia, reserve cell hyperplasia, microglandular hyperplasia, cellular features suggestive of HPV-related effect, and inflammatory cervicitis. A random sample of cervical biopsy specimens collected between 2002 and 2004 was used. One hundred ninety-one cervical lesions including 81 CIN 1, 52 squamous metaplasia, 33 cellular features suggestive of HPV-related change, 9 reserve cell hyperplasia, 4 microglandular hyperplasia, and 12 inflammatory cervicitis were evaluated for p16Ink4a reactivity. All CIN 1 lesions were diagnosed using strict histologic criteria for CIN 1 lesions, as previously described.15 The lesion distribution is shown in Table 1. Institutional review board approval was given for this study.
Evaluation for p16Ink4a immunoreactivity was performed as described elsewhere21 using a purified mouse anti-human p16Ink4a monoclonal antibody (clone JC8, Dako, Carpinteria, Calif) at a dilution of 1:25. Paraffin-embedded, formalin-fixed cervical biopsy specimens were depariffinized in xylene and rehydrated through a series of graded alcohol solutions. Endogenous perioxidase was blocked using 3% H2O2 for 4 minutes, and heat-induced epitope retrieval was performed using a Trilogy solution (Cell Marque, Biocare Medical, Walnut Creek, Calif). Application of p16Ink4a primary antibody was applied for 32 minutes, followed by incubation with biotin anti-mouse immunoglobulin G. The bound primary antibody was visualized by streptavidin–horseradish perioxidase detection according to the manufacturer's instructions and using diaminobenzidine chromogen as the color-developing agent. Sections were counterstained with hematoxylin.
Interpretation of p16Ink4a Reactivity
The reaction pattern for p16Ink4a reactivity was scored in a similar manner to that used in the literature.2 Each lesion was graded according to a 3-tier system: brown nuclear and cytoplasmic reactivity was scored from negative to 2+ (negative, representing 0% to 5% reactive cells; +1, focal/scattered positivity, representing greater than 5% and less than or equal to 80% reactivity; +2, diffuse positivity, representing greater than 80% reactivity). p16Ink4a reactivity was based on normal/reactive cervical specimens where anti-p16 antibody was negative or weakly reacted with a subset of proliferating non–cervical epithelial cells, such as fibroblasts, endothelial cells, and histiocytes.2
HR-HPV Typing by Real-Time Polymerase Chain Reaction Analysis
All CIN 1 lesions not reactive for p16Ink4a were investigated for the presence of HR-HPV subtypes 16 and 18.
HPV DNA Extraction
DNA extraction was performed from paraffin-embedded tissue sections, and real-time polymerase chain reaction (RT-PCR) was performed as described elsewhere.21 DNA extraction was performed on paraffinized tissue samples using TaKaRa Biomedicals' DEXPAT Reagent (Otsu, Shiga, Japan). Samples were placed in a 1.5-mL tube, and 0.5 mL of DEXPAT reagent was added. The tubes were incubated in a block heater at 100°C for 10 minutes. The samples were placed in a microcentrifuge and spun at 13 000 rpm for 10 minutes at 4°C. Four layers formed in the microcentrifuge tubes: (starting from the bottom) a layer of resin, a thin layer of tissue debris, the clear supernatant layer containing the DNA, and a thin layer of paraffin. The supernatant layer was carefully removed, making sure not to disturb the other layers. The supernatant was placed into a fresh 1.5-mL tube; 20 μL of sodium acetate was added, followed by 500 μL of 100% ethanol. The tubes were refrigerated overnight at −20°C. The tubes were centrifuged at 13 000 rpm for 10 minutes. The supernatant was removed, and 0.5 mL of 80% ethanol was added. The tubes were centrifuged again at 13 000 rpm for 5 minutes, and the supernatant was removed. Tubes were left open under a ventilation hood until no alcohol was present. The pellet was resuspended in 30 μL of Tris-EDTA buffer.
Real-Time PCR Analysis
Fluorophore-labeled LUX primers and their unlabeled counterparts were obtained from Invitrogen (Carlsbad, Calif). The primers for HPV type 16 were as follows: labeled, GTACGAAAAGCACACACGTAGACATTCGTAC; and unlabeled, TTCCTAGTGTGCCCATTAACAGGT. Primers for glyceraldehyde-3-phosphate dehydrogenase (GADPH) were used as internal controls. Reactions were conducted in a 96-well spectrofluorometric thermal cycler (ABI PRISM 7700 Sequence detector system, Applied Biosystems, Foster City, Calif). Fluorescence was monitored during every PCR cycle at the annealing step. The PCR program was as follows: 95°C for 10 minutes and then 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. Results were analyzed with SDS 2.0 software (Applied Biosystems).
All 191 cervical lesions were evaluated for p16Ink4a expression. Thirty (37%) of the 81 CIN 1 lesions immunohistochemically expressed p16Ink4a. Eleven (37%) of the 30 lesions exhibited focal/scattered nuclear and cytoplasmic reactivity (Figure, A and B) and the remaining 19 lesions (63%) exhibited diffuse nuclear and cytoplasmic reactivity (Figure, C and D). No expression of p16Ink4a was detected in the 110 non–CIN 1 cases, including those with metaplasia and suggestive of HPV-related changes, metaplastic/reactive or inflammatory cervical lesions.
The CIN 1 lesions not reactive for p16Ink4a were investigated for the presence of HR-HPV subtypes 16 and 18 by RT-PCR. Fifty-one (63%) of the 81 CIN 1 lesions were not reactive for p16Ink4a. Ninety-eight percent (50/51) were RT-PCR negative for the presence of HPV subtypes 16 and 18. Only 1 CIN 1 lesion that was not reactive for p16Ink4a was RT-PCR positive for HR-HPV.
The need for biological markers for cervical carcinoma and its precursor lesions is emerging.6 Numerous biomarkers have been evaluated in the literature (Table 2). All are reported to be available for immunohistochemical studies on paraffin-embedded tissue. Some of the markers are nuclear proteins like the DNA polymerase cofactor proliferating cell nuclear antigen, the nonhistone protein Ki-67 (MIB-1), or the ribonucleoprotein enzyme telomerase.6 Others are cell cycle–related proteins such as cyclin E, ProEx C, and p16Ink4a. Others are like Bcl-2 and function as inhibitors of apoptosis.1,3,4,6,22,23 The tumor suppressor protein pRb has also been investigated.3
Except for Bcl-2, for which studies are conflicting,4 overall immunoreactivity for these proteins is seen in invasive squamous cell carcinoma of the cervix and high-grade CINs; however, the reported reactivity of biomarkers in reactive/metaplastic or inflammatory cervical lesions is varied. Proliferating cell nuclear antigen exhibits rare basal cell activity in normal and reactive cervical squamous epithelial cells. Like cyclin E, there is strong immunohistochemical reactivity for proliferating cell nuclear antigen in nonneoplastic inflammatory cervicitis.6 Studies using Ki-67 (MIB-1) are varied. Some literature has demonstrated that there is no significant immunohistochemical expression in reactive or inflammatory cervical lesions.6 A recent study, however, found diffuse Ki-67 immunoreactivity in CIN 1 lesions and in reactive inflammatory lesions, showing that the immunoreactivity is highly dependent upon the amount of nuclear atypia and number of mitoses within the lesion in question.8
Evaluations of the enzyme telomerase as a biomarker are limited owing to its low immunohistochemical expression in lesions and high background activity.6 ProEx C, a new biomarker recently available for immunohistochemical use, is a cocktail of topoisomerase II-α and minichromosome maintenance protein 2.22,23 Studies on ProEx C are few, and further evaluation is needed to determine its value as a biomarker for CIN.
The retinoblastoma gene product (pRb) is detected immunohistochemically in normal cervical squamous epithelium and HPV-related lesions of the cervix. As would be expected, due to the inhibitory effect of HPV E7 oncoprotein on the pRb, the expression of pRb is decreased as the grade of the cervical neoplasia increases.3 Accordingly, studies looking at the concurrent expression of pRb and p16Ink4a in CIN and cervical carcinomas demonstrate that as p16Ink4a reactivity increases, the pRb expression decreases.3 The integral role that HPV plays in the carcinogenesis of CIN has provided many possible biomarkers, with p16Ink4a having significant potential. p16Ink4a immunohistochemical evaluation shows repeated utility in assessing CIN and invasive squamous cell carcinomas of the cervix, and in decreasing interobserver variability of these lesions.2–5,7–9
Despite proposed criteria for low-grade cervical lesions, various factors such as concurrent inflammation may confound the diagnosis.17 Further diagnostic dilemmas may occur with nonneoplastic lesions that histologically may resemble CIN 1 lesions, or occur with and mask CIN 1 lesions. The purpose of the current study was to determine the utility of p16Ink4a expression in nonneoplastic reactive/metaplastic and inflammatory cervical lesions and determine whether this marker can be used to help distinguish these equivocal lesions from CIN 1 and therefore be of value to reduce the risk of overdiagnosis of CIN 1.
In this study none of the nonneoplastic equivocal cervical lesions including squamous metaplasia, reserve cell hyperplasia, cellular features suggestive of HPV-related effect, or inflammatory cervicitis exhibited p16Ink4a reactivity. These findings are similar to previous reports using p16Ink4a immunohistochemical evaluation of nonneoplastic lesions of the cervix.20 Klaes and coauthors,20 in their study of p16Ink4a in cervical tissue biopsy specimens, reported no significant p16Ink4a immunoreactivity in inflammatory lesions and reserve cell hyperplasic changes without concurrent evidence of dysplasia. One of the greatest advantages p16Ink4a may offer is in discriminating reactive inflammatory changes versus the presence of dysplasia in the company of inflammation.17 Frisch24 has suggested that inflammation has long been a reported confounding factor in false-negative screening Papanicolaou smears. In a retrospective review of cervical smears, Frisch24 found up to 34% of cases diagnosed as inflammatory atypia were likely to have a concurrent CIN lesion. The results of this current study support the role of immunohistochemical study with p16Ink4a on equivocal cervical biopsy specimens in order to discriminate between inflammatory conditions or reactive conditions that either mask or mimic dysplasia in cervical biopsy specimens. While p16Ink4a would not be needed in every case, it would be an added tool for pathologists to help improve patient care.
In addition to the presence of p16Ink4a immunoreactivity, the literature suggests that the pattern of p16Ink4a immunoreactivity within CIN lesions may also be of clinical utility. Sano and coauthors put forward that different p16Ink4a immunoreactivity patterns may be seen in cervical epithelial lesions infected by low-risk versus high-risk HPV subtypes, and that these differences may allow for differentiation between CIN lesions.2,7 Furthermore, the literature implies that the presence of p16Ink4a in a biopsy specimen may indicate a higher likelihood that the lesion will progress to a higher-grade neoplasm.2,25 Negri and coauthors report that a CIN 1 lesion with diffuse p16Ink4a immunoreactivity was statistically more likely to progress to a CIN 3 lesion than a CIN 1 lesion that was negative for p16Ink4a.25,26 The CIN 1 lesions not reactive for p16Ink4a were more likely to regress at follow-up.25,26 In a recent study looking at p16Ink4a reactivity in CIN 1 lesions, Hariri and Oster26 reported the negative predictive value of p16Ink4a to be as high as 96%. This difference in p16Ink4a immunoreactivity between CIN 1 lesions may allow for better identification of patients more likely to have progression of their cervical lesions and require aggressive treatment.
In the current study, despite the use of strict histologic criteria for the microscopic diagnosis of CIN 1, 63% (50/ 81) of the CIN 1 lesions evaluated were negative for p16Ink4a. Furthermore, 98% (50/51) of those p16Ink4a-negative lesions were also RT-PCR negative for HR-HPV. This percentage, while it appears high, is within the range of other studies when compared with published literature.27,28 As suggested in previous literature, a possible reason for the lack of p16Ink4a expression in some of these lesions is that they are caused by low-risk HPV types.8,16 These low-risk HPV types can still cause histologic changes that would qualify as a CIN 1 lesion using the diagnostic criteria but would be immunohistochemically nonreactive for p16Ink4a. In addition, in the current study, these lesions would be RT-PCR negative since the lesions were only tested for high-risk HPV subtypes and were not tested for any concurrent low-risk subtypes. Although the current study did not test these specimens for the presence of the low-risk HPV viral subtypes, if the low-risk subtypes caused these lesions, previous studies indicate that these particular lesions should clinically regress.25,26
Overall, the results of this study are comparable to those in the literature evaluating p16Ink4a in low-grade cervical neoplasia as well as using p16Ink4a in discriminating nonneoplastic equivocal cervical lesions from CIN 1. Used in the correct situation, p16Ink4a can provide an important ancillary technique to pathologists in the setting of possible dysplasia with either confounding metaplastic or reactive histologic changes. This ability to better discriminate between CIN and nonneoplastic equivocal lesions will help reduce false-negative interpretations and improve cervical precancer diagnosis as well as reduce false-positive interpretations, thus decreasing unnecessary surgical procedures.
The authors have no relevant financial interest in the products or companies described in this article.
The data have been presented in abstract form at the 2006 Biennial Meeting of the American Society for Colposcopy and Cervical Pathology, Las Vegas, Nev.
Reprints: Rachel Redman, MD, Department of Pathology, College of Medicine, University of Florida, 1600 SW Archer Rd, Gainesville, FL 32610-0275 (firstname.lastname@example.org)