Evaluation of cervical cone biopsies for coexpression of p16INK4a and Ki-67 in epithelial cells

Authors

  • Miriam Reuschenbach,

    Corresponding author
    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    • Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Im Neuenheimer Feld 220, 69120 Heidelberg, Germany
    Search for more papers by this author
    • Tel.: +49-6221-565210, Fax: +49-6221-565981

    • M.R. and M.S. contributed equally to this work

  • Mirjam Seiz,

    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author
    • M.R. and M.S. contributed equally to this work

  • Christina von Knebel Doeberitz,

    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author
  • Svetlana Vinokurova,

    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author
  • Alexander Duwe,

    1. mtm Laboratories, Heidelberg, Germany
    Search for more papers by this author
  • Ruediger Ridder,

    1. mtm Laboratories, Heidelberg, Germany
    Search for more papers by this author
  • Heike Sartor,

    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author
  • Friedrich Kommoss,

    1. Institute of Pathology, Mannheim, Germany
    Search for more papers by this author
  • Dietmar Schmidt,

    1. Institute of Pathology, Mannheim, Germany
    Search for more papers by this author
  • Magnus von Knebel Doeberitz

    1. Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany and Clinical Cooperation Unit Applied Tumor Biology, German Cancer Research Center, Heidelberg, Germany
    Search for more papers by this author

  • Conflict of interest: AD and RR are employees of mtm Laboratories that develops and markets products related to this study. DS has been a temporary clinical advisor and speaker for mtm Laboratories, and MvKD is a member of the supervisory board of mtm Laboratories. RR, CvKD and MvKD declare a financial interest in mtm Laboratories. All other authors disclose no potential conflict of interest.

Abstract

Diffuse overexpression of p16INK4a in basal and parabasal cells of cervical epithelium is a hallmark of human papillomavirus-mediated transformation. Focal p16INK4a expression is occasionally observed in nondysplastic epithelium. In normal cells, expression of p16INK4a triggers cell cycle arrest. However, cells undergoing transformation in intraepithelial lesions actively proliferate. To prove that the different expression patterns of p16INK4a, i.e., focal versus diffuse, reflect biologically different entities, we hypothesized that p16INK4a-positive cells in epithelia displaying focal p16INK4a expression pattern do not coexpress proliferation-associated Ki-67 protein, while p16INK4a-positive cells in lesions with diffuse p16INK4a expression may do. A total of 138 cervical cone biopsies were stained for the expression of p16INK4a and Ki-67 using a primary antibody cocktail. All metaplastic lesions (n = 21) displayed focal staining for p16INK4a, and in all of these lesions p16INK4a-positive cells were found to be negative for Ki-67 expression. Diffuse expression of p16INK4a was observed in 12/21 (57.1%) cervical intraepithelial neoplasia (CIN) 1 lesions, all of them simultaneously showed Ki-67 immunoreactivity in a large proportion of p16INK4a-positive cells. Seventeen of 23 (73.9%) CIN2 lesions and all 27 (100%) CIN3/carcinoma in situ (CIS) as well as all 46 (100%) carcinoma cases displayed diffuse and combined expression of p16INK4a and Ki-67. Coexpression of Ki-67 and p16INK4a in the same cell is entirely restricted to cervical lesions displaying diffuse p16INK4a expression, whereas in lesions with focal p16INK4a expression, p16INK4a-expressing cells are negative for Ki-67. Thus, diffuse expression of p16INK4a reflects lesions with proliferation-competent cells, while p16INK4a-expressing cells associated with focal expression patterns are cell cycle arrested.

Cervical cancer represents the second most common cancer in women worldwide, with an increased incidence in low resource countries.1 Virtually all cervical cancers are caused by persistent infections with high-risk human papillomavirus (HR-HPV) types which may cause cervical intraepithelial neoplasia (CIN) and invasive cancer.2 An essential prerequisite for the shift from a clinically unapparent transient HPV infection to initiation of transformation and maintenance of neoplastic growth of the cell is the continuous expression of the viral oncogenes E6 and E7 in basal and parabasal epithelial cells.3 This transforming stage of HR-HPV infections is highlighted by strong and diffuse overexpression of the cyclin-dependent kinase inhibitor p16INK4a.4, 5 p16INK4a is a valuable surrogate marker for deregulated HPV oncogene expression and is used as a biomarker for high-grade cervical precursor lesions and cancer.6, 7 Diffuse immunohistochemical p16INK4a staining, arising from basal cells and variably reaching superficial epithelial layers is characteristic for high-grade cervical lesions (CIN2, 3) and is also observed in a proportion of CIN1 lesions. Its use in conjunction with hematoxylin and eosin (H&E)-stained sections has substantially improved the interobserver agreement and diagnostic accuracy in the histopathologic evaluation of cervical biopsies.8–10 Furthermore, there is evidence that detection of p16INK4a overexpression may be a prognostic marker for the fate of CIN1 lesions, as those that display a diffuse overexpression of p16INK4a were shown in several studies to have a higher tendency to progress to high-grade cervical precancerous lesions as compared to CIN1 lesions that do not show diffuse p16INK4a expression and thus more likely reflect transient, permissive HPV infections.11–14

The mechanism by which p16INK4a is upregulated in cells undergoing HR-HPV-mediated transformation has not finally been elucidated. There is evidence suggesting that a negative feedback mechanism controlling p16INK4a levels in normal cells is disrupted by a reduction of pRB activity in proliferating squamous epithelial cells expressing HR-HPV E7.15 This hypothesis is in line with the observation that also in some other cancer entities an inverse correlation between pRB and p16INK4a expression levels has been observed, irrespective of the HPV status.16

The p16INK4a protein inhibits the activity of cyclin-dependent kinases and thus prevents the phosphorylation and functional inactivation of pRB.17 Expression of p16INK4a is observed in aged cells or those that are no longer capable of completing the full normal differentiation pathway.18 In line with its function as a cyclin-dependent kinase inhibitor, p16INK4a provides strong cell cycle arresting signals. When transferred into cancer cells devoid of endogenous p16INK4a activity it may trigger inhibition of proliferation of the respective cancer cells.19

Nondysplastic squamous cervical epithelia are commonly either completely negative for p16INK4a expression or show a focal p16INK4a expression pattern, i.e., a sporadic or patchy p16INK4a immunoreactivity restricted to the more differentiated intermediate or superficial squamous epithelial cell layers. This clearly differentiates them from the diffuse staining pattern observed in transforming HPV infections which is characterized by the continuous staining beginning in basal and parabasal cell layers.10 Only the diffuse staining pattern is considered as p16INK4a positive in cervical pathology interpretation, and any focal or sporadic staining is considered as p16INK4 negative. This typical histopathologic pattern allows for the differentiation between p16INK4a expression in nondysplastic epithelium as observed in squamous metaplasia from CIN lesions undergoing HPV-mediated transformation and has led to the routine implementation of p16INK4a immunohistochemistry in the diagnostic work up of cervical lesions.9, 10

We hypothesized that it is likely that markers associated with proliferation can differentiate p16INK4a-positive cells in cervical squamous cell dysplasia from p16INK4a-positive cells in nondysplastic cervical epithelia, as dysplastic cells should be transformed by the activated expression of HPV oncogenes that trigger uncontrolled proliferation. In contrast, nondysplastic cells should not be transformed, retain pRB functions and should thus not proliferate. Confirmation of this hypothesis would indicate that the diffuse and focal or sporadic expression patterns of p16INK4a reflect a different biological behavior of cervical squamous epithelial cells. To test this hypothesis, the proliferation status of p16INK4a-positive cells was analyzed in cervical cone biopsies diagnosed as intraepithelial neoplasia, invasive cancer or squamous cell metaplasia, by applying combined immunohistochemical staining for p16INK4a and the proliferation marker Ki-67. The results were subsequently compared to the different expression patterns of p16INK4a within the epithelium (diffuse staining, vs. focal/negative patterns).

Material and Methods

Cervical lesions and p16INK4a/Ki-67 dual staining immunohistochemistry

Formalin-fixed, paraffin-embedded cervical cone biopsy specimens diagnosed as CIN1 (n = 21), CIN2 (n = 23), CIN3/carcinoma in situ (CIS) (n = 27), invasive cervical carcinomas (n = 46) or squamous cell metaplasia (n = 21) were included in this analysis. All H&E sections were reviewed by two pathologists (DS/FK), and discordant results were discussed and a consensus diagnosis was obtained. For immunohistochemical combined staining of p16INK4a and Ki-67, 2 μm sections were stained using the CINtec PLUS kit (mtm Laboratories, Heidelberg, Germany). The staining is based on two monoclonal antibodies, a mouse monoclonal antibody directed against human p16INK4a (clone E6H4) and a rabbit monoclonal antibody directed against human Ki-67, respectively. p16INK4a expression is visualized by a brown 3,3′-diaminobenzidine and abbreviation (DAB) reaction and Ki-67 expression by a FastRed reaction, each obtained after incubation with species-specific secondary antibody polymers conjugated to horseradish peroxidase or alkaline phosphatase, respectively. The staining was done according to the instructions given by the manufacturer for cytological specimens with some minor modifications to optimize the protocol for use in histology. In brief, slides were deparaffinized in xylene and rehydrated by passage through alcohol. After antigen retrieval in the supplied solution, the primary antibody cocktail was incubated for 30 min. DAB chromogen was incubated for 8 min, and FastRed chromogen was incubated for 15 min twice. Slides were counterstained with alcohol-free hematoxylin (Dako, Glostrup, Denmark).

Evaluation of p16INK4a and Ki-67 expression

The p16INK4a/Ki-67 dual staining was analyzed independently by two observers (MR/MS) blinded to the H&E-based diagnosis. For cases with discrepant results, a consensus result was obtained by including a third observer (CvKD). The slides were assessed for p16INK4a expression only, for Ki-67 expression only and for coexpression of both proteins. p16INK4a expression was categorized into two groups as described in Ref.10: (i) a positive rating was assigned to a diffuse p16INK4a expression, i.e., nuclear and/or cytoplasmic, continuous staining beginning in basal and parabasal cells and variably reaching intermediate and superficial cell layers and (ii) a negative rating for p16INK4aexpression, which comprises lesions showing either a focal or sporadic staining of epithelial cells, or that are completely negative for p16INK4a expression.

The extent of Ki-67 expression in the lesions was categorized into four groups based on the percentage of epithelial cells showing nuclear staining. (i) <10% of the cells, restricted to the parabasal cell layers, (ii) 10 to <30% of the cells, restricted to the lower third of the epithelium, (iii) 30 to <70% of the cells, reaching the upper third of the epithelium and (iv) 70% or more of the epithelial cells including full thickness expression of Ki-67.

The extent of cells coexpressing both proteins was determined as either negative or positive for at least one cell expressing p16INK4a and Ki-67. Dual stain-positive lesions were further categorized for the percentage of dual-stained cells visible in a high power field at 200-fold magnification. The same four quantitative categories as used for the interpretation of the Ki-67 expression alone were applied.

Results

Expression of p16INK4a in cervical metaplasia, dysplasia and carcinoma

All 138 specimens that were subjected to p16INK4a/Ki-67 dual staining were evaluated for the expression of the two proteins separately and for the extent of the presence of cells coexpressing both markers. The frequency of lesions showing a diffuse expression pattern of p16INK4a increased with the severity of the dysplasia (Table 1). Among the CIN1 lesions, 12/21 (57.1%) showed a diffuse p16INK4a expression pattern and therefore are considered as p16INK4a-positive. Two of 21 (9.5%) CIN1 cases had a focal p16INK4a expression, and the remaining cases (7/21; 33.3%) were completely negative. Among the CIN2 lesions, a diffuse p16INK4a expression was seen in 17/23 (73.9%), focal expression was observed in 2/23 (8.7%) and no expression in 4/23 (17.4%). All of the CIN3/carcinoma in situ (n = 27) and carcinoma (n = 46) cases showed a diffuse p16INK4a expression. All lesions categorized as squamous cell metaplasia (n = 21) showed focal p16INK4a expression.

Table 1. Frequency of lesions expressing p16INK4a, Ki-67 and coexpressing both proteins within the same cells
inline image

Extent of Ki-67 expression in cervical metaplasia, dysplasia and carcinoma

Ki-67 expression was categorized according to the fraction of stained cells in the lesion. Differing from what was observed for p16INK4a expression, which is completely negative in a proportion of CIN1 and CIN2 lesions, Ki-67 was found to be expressed at least in the parabasal cell layers in all analyzed lesions. We observed an increase in Ki-67 expression with the severity of the lesion (Table 1). Among the CIN1 lesions, 8/21 (38.1%) expressed Ki-67 in 10 to <30% of the cells. One out of 21 (4.8%) of the CIN1 cases showed Ki-67 expression in 70% or more of the cells. In CIN2 lesions, 6/23 (26.1%) had expression of Ki-67 in 70% or more of the cells. With 17/27 (63.0%) most CIN3/carcinoma in situ (CIS) were highly proliferative with an expression of Ki-67 in 70% or more of the cells. The proportion of lesions with Ki-67 expression in 70% or more of the cells was somewhat lower in the carcinomas (21/46, 45.7%) than in the CIN3/carcinoma in situ (CIS) group of cases. In the metaplastic lesions, Ki-67 expression was found predominantly (17/21; 81.0%) in less than 10% of the cells in the parabasal layers. The remaining metaplastic cases showed Ki-67 expression in 10 to <30% (3/21; 14.3%) or 30 to <70% (1/21; 4.8%) of the cells.

Frequency of cells coexpressing p16INK4a and Ki-67 and correlation to p16INK4a expression patterns

Next, we assessed whether and to what extent the two proteins were expressed within the same cells, and whether coexpression on a single cell level correlates with the p16INK4a expression pattern. Lesions showing at least one cell coexpressing p16INK4a and Ki-67 within the same cell were considered as dual stain positive. Comparing the presence of individual cells simultaneously coexpressing p16INK4a and Ki-67 to the various p16INK4a expression patterns revealed that the detection of dual stain-positive cells was entirely restricted to lesions showing a diffuse p16INK4a expression pattern. All diffusely p16INK4a-positive cases (n = 102) displayed cells coexpressing p16INK4a and Ki-67, while in none of the cases with focal p16INK4a expression (n = 25) dual stain-positive cells were detected (Table 2).

Table 2. Correlation of p16INK4a expression patterns to the presence of p16INK4a/Ki-67 dual-stained cells
inline image

The fraction of dual stain-positive cells increased with the severity of the dysplasia (Table 1). In CIN1, 12/21 (57.1%) of the lesions showed cells coexpressing both proteins, mostly restricted to the parabasal cell layers. Among the CIN2 samples, 17/23 (73.9%) comprised dual-stained cells. CIN3/carcinoma in situ and carcinoma cases showed both markers expressed in the same cell in 100% of the cases of the study cohort (27/27 and 46/46, respectively; Table 1 and Fig. 1). Importantly, in none (0/21; 0.0%) of the metaplastic lesions squamous epithelial cells were identified that coexpressed both markers simultaneously (Table 1 and Fig. 2).

Figure 1.

Hematoxylin & eosin staining [(a) and (c)] and p16INK4a/Ki-67 dual stain [(b) and (d)] of sections from serial cuts from cervical cone biopsies (×20 objective and ×40 objective detail). (b) p16INK4a-negative and dual stain negative CIN1 with parabasal Ki-67 expression (red nuclei). (d) Diffuse p16INK4a expression (brown) in a dual stain-positive (red nuclei in brown cells) CIN3.

Figure 2.

Hematoxylin & eosin staining (a) and p16INK4a/Ki-67 dual stain (b) of sections from serial cuts from a cervical cone biopsy (×40 objective). (b) Focal p16INK4a expression (brown cells) in a dual stain-negative metaplasia (no red nuclei in brown cells) with isolated Ki-67 expression (red nuclei). Black arrow indicates p16INK4a-positive cell (brown), empty arrow indicates Ki-67-positive cell (red nucleus).

Discussion

The high prevalence of HPV infections without concomitant cervical dysplasia and thus low specificity of HPV DNA testing for high-grade cervical dysplasia as well as the known low sensitivity of Pap cytology warrant the evaluation of biomarkers that specifically highlight HPV-transformed dysplastic cells and thereby help to identify relevant cervical disease.20, 21 The detection of p16INK4a overexpression has been evaluated in numerous studies and has been shown to provide a promising and beneficial tool in the diagnostic evaluation of cervical biopsies. Its immunohistochemical detection in tissues is widely used and accepted.9, 22 Histopathology interpretation of slides stained with H&E is the gold standard for diagnosis of CIN. However, diagnostic variability has been observed and documented in several studies among pathologists.23 This may depend on an equivocal morphological grade of the lesion, and atrophy, reactive changes and some squamous metaplasia are mimics of high-grade intraepithelial neoplasia and might cause misinterpretations.24, 25 Immunohistochemical staining for p16INK4a is applied in cervical pathology as a marker for HPV-transformed lesions and it has been demonstrated in various studies that p16INK4a staining improves the interobserver agreement and accuracy in the diagnosis of high-grade cervical neoplasia.8, 10, 26

The observation of occasionally p16INK4a-expressing epithelial changes without dysplastic alterations is well known and documented.27 In histology, the tissue context allows to distinguish diffuse p16INK4a-positive staining patterns defined as continuous staining beginning in the basal and parabasal squamous epithelial cell layers (Fig. 1) from lesions showing focal or sporadic staining characterized by few p16INK4a-positive cells in intermediate or superficial squamous epithelial cell layers. Based on the concept that HPV-mediated transformation is triggered by deregulated expression of the viral oncogenes E6 and E7 in basal and parabasal cells, p16INK4a immunohistochemistry was hypothesized to distinguish between transforming and nontransforming HPV infections, and only the diffuse p16INK4a expression pattern was defined as hallmark of HPV-dependent transformation and thus considered as p16INK4a-positive.3, 6, 10

Ki-67 expression alone has been found in previous studies to be associated with the grade of dysplasia.28, 29 There are few studies that evaluated p16INK4a and Ki-67 immunohistochemistry in HPV-induced dysplastic lesions and showed correlation with progression grade of the lesion.30, 31 We used a cocktail of both antibodies in one staining procedure simultaneously, thus allowing to evaluate whether p16INK4a and Ki-67 are coexpressed within exactly the same cell and to link these findings to the grade of histopathologic abnormality of the cervical squamous epithelium.

We found that the frequency of lesions with p16INK4a/Ki-67 coexpressing cells increased from CIN1 to invasive carcinoma. In all lesions displaying a diffuse p16INK4a expression pattern, dual stain-positive cells were detected, while in none of the lesions with focal p16INK4a expression dual stain-positive cells were identified. The data gathered in our study therefore add further evidence that verifies this hypothesis, as all lesions displaying the diffuse p16INK4a staining pattern indeed coexpressed Ki-67 pointing to deregulated cell cycle control. Cells overexpressing p16INK4a in lesions displaying focal or sporadic staining patterns in contrast were consistently devoid of Ki-67 expression, strongly suggesting that in these cells cell cycle control was retained.

In conclusion, our data show that cells coexpressing the markers p16INK4a and Ki-67 are found exclusively in cervical epithelia which show diffuse p16INK4a expression, while p16INK4a-positive cells in cervical epithelia with focal or sporadic p16INK4a expression do not coexpress Ki-67. This finding confirms the association of the diffuse p16INK4a expression pattern with deregulated cell cycle in cervical epithelia.

Recently, data from several studies evaluating the utility of p16INK4a detection in cervical cytology specimens were published and suggest that p16INK4a detection might be more specific than HPV DNA testing in the triage of equivocal or mildly abnormal Pap cytology results.32, 33 However, the evaluation of p16INK4a expression in cytological specimens requires additional criteria than simply the presence of one or more p16INK4a-positive cells to consider a sample as positive, as not only dysplastic cells may express p16INK4a. Additional criteria in the past were such as considering only samples as positive extending a certain number of p16INK4a-positive cells or applying morphological interpretation criteria to p16INK4a-positive cells to consider a slide as positive.34 Based on our results, it is very well conceivable that the detection of p16INK4a-positive cells that simultaneously coexpress Ki-67 within the same cell in cervical cytology preparations may identify cervical epithelial cells with deregulated cell cycle control mediated by HR-HPV oncogene expression and thus may be indicative of cervical dysplasia.

The presented data strongly suggest that the immunohistochemical staining for p16INK4a alone allows distinguishing transformed from nontransformed cervical epithelial lesions based on the expression pattern of p16INK4a immunoreactive cells in the squamous epithelium. However, in cytology, where it is unclear whether a distinct p16INK4a-positive cell was derived from focally or diffusely p16INK4a expressing epithelia, the combined use of p16INK4a and Ki-67 antibodies in a single staining kit may substantially facilitate the distinction of transformed and nontransformed cells and thus allow for substantially more accurate interpretation of cytology results. The latter hypothesis is currently being investigated in large clinical studies.

Ancillary