|Home | About | Journals | Submit | Contact Us | Français|
P16INK4a is a biomarker for transforming HPV infections that could act as an adjunct to current cytological and histological assessment of cervical smears and biopsies, allowing the identification of those women with ambiguous results that require referral to colposcopy and potentially treatment.
We conducted a systematic review of all studies that evaluated the use of p16INK4a in cytological or histological specimens from the uterine cervix. We also estimated the mean proportion of samples that were positive for p16INK4a in cytology and histology, stratified by the grade of the lesion.
Sixty-one studies were included. The proportion of cervical smears overexpressing p16INK4a increased with the severity of cytological abnormality. Among normal smears, only 12% (95% CI: 7–17%) were positive for the biomarker compared to 45% of ASCUS and LSIL (95% CI: 35–54% and 37– 57% respectively) and 89% of HSIL smears (95% CI: 84–95%). Similarly, in histology only 2% of normal biopsies (95% CI: 0.4– 30%) and 38% of CIN1 (95% CI: 23– 53%) showed diffuse staining for p16INK4a compared to 68% of CIN2 (95% CI: 44– 92%) and 82% of CIN3 (95% CI: 72– 92%).
Although there is good evidence that p16INK4a immunostaining correlates with the severity of cytological/histological abnormalities, the reproducibility is limited due to insufficiently standardized interpretation of the immunostaining. Therefore, a consensus needs to be reached regarding the evaluation of p16INK4a staining and the biomarker needs to be evaluated in various clinical settings addressing specific clinical questions.
Since the Papanicolaou (Pap-test) cytological screening for cervical precancerous lesions was introduced in the 1940s, there has been a significant reduction in the incidence and mortality from cervical cancer (1). However, the efficacy of the Pap test is hampered by high interobserver variability and high false negative and false positive rates that range between 20–30% (2) and 5–70% (3), respectively. Technical improvements of the Pap test such as the Liquid Based Cytology (LBC) have not been shown to improve sensitivity or specificity for detection of high-grade cervical intraepithelial neoplasia (CIN) compared to the conventional cytology (4).
The introduction of human papillomavirus (HPV) DNA testing in clinical practice raised hopes for further improvements in the efficacy of the primary screening, triage and post-treatment surveillance. Randomized clinical trials published recently have demonstrated that HPV testing can be efficiently integrated into primary screening, either as an adjunct to cytology or as a sole primary test (5, 6). It has also been shown that HPV DNA testing can be used to triage women with equivocal cytological abnormalities (7) and that it has a potential role in identifying women at risk of residual or recurrent disease after treatment for CIN (8, 9). However, it fails in the triage of low-grade lesions (9) and even if implemented as a primary screening test, it would be necessary to have a more disease specific triage marker to identify women that would need to undergo colposcopy. Furthermore, a single HPV DNA test although it could confirm infection by the virus, present in 99% of all cervical cancers (10), it does not discriminate between transient and chronic infection. The discrimination between the two types of infection is crucial as it is the persistent infection that predisposes to progression to cervical neoplasia and not the transient one (11).
Even the histological assessment of cervical biopsies that is often considered as the “gold standard”, can be significantly hampered by intra- and inter-observer variability (12). Novel markers applied on histological specimens could improve the identification of women with ambiguous results that require treatment.
Research nowadays is focused on the development of objective biomarkers that can distinguish transforming from productive HPV infections and predict disease severity. The cellular tumor suppressor protein p16INK4a (p16) has been identified as a biomarker for transforming HPV infections. Physiologically, p16 blocks the activity of cyclin-dependent kinases CDK4/6. In a transforming HPV infection the viral oncogenes E6 and E7 interfere substantially with apoptosis and cell cycle regulation. Most importantly, E7 disrupts the protein of retinoblastoma (pRb) from its binding to E2F transcription factor and thereby promotes cell cycle progression, a molecular switch that is usually activated by CDK4/6. Affected cells strongly express p16 to counteract the irregular cell cycle activation; however, since E2F is not released through CDK4/6 action, but by E7, p16 expression has no effect on cell cycle activation. Over time, p16 accumulates in the nucleus and cytoplasm of affected cells and can be detected by immunostaining (13).
This review represents an attempt to collect, systematically present and analyse the existing evidence on possible clinical applications of p16 in cytological and histological samples from the uterine cervix.
We searched two electronic databases - MEDLINE and EMBASE - targeting reports published between January 1998 and September 2007. The search strategy used terms such as “cancer”, “dysplasia”, “SIL”, “CIN”, “cervix”, “p16” and “cyclin-dependent kinase”. The references of retrieved articles together with the proceedings of relevant conferences were hand-searched in order to identify other potentially eligible studies for inclusion in the analysis missed by the initial search or any unpublished data. Additional cross-searches were performed in MEDLINE using the names of investigators who were the lead authors of at least one eligible study.
The literature search, assessment of inclusion and exclusion criteria, quality of studies and extraction of data were independently undertaken and verified by two investigators (IT, MK). The results were then compared and, in case of discrepancies, a consensus was reached with the involvement of a third investigator (MA). There was no language restriction.
All retrospective or prospective studies that assessed p16 immunostaining in cervical cytological samples, in conventional cytology or in LBC, as well as in histological specimens from the uterine cervix were included in this review. We evaluated all methods and interpretation of p16 immunostaining.
We excluded studies that assessed the expression of the biomarker in glandular or invasive cervical lesions. In cases of overlap or duplicate studies, we retained only the most comprehensive one.
All outcomes were defined prior to the literature search. The primary outcome was the correlation between cytological or histological degree of cervical abnormality and overexpression of p16 identified by immunochemistry. Other parameters assessed were the role of the biomarker in the cervical cancer screening, its role in the triage of equivocal or low-grade cytological abnormalities compared to HPV testing and its efficacy as a marker of progression risk in low grade cervical lesions.
For all included studies we generated descriptive tables for population and study characteristics. We recorded the first author, publication year, country of the investigators, sample size and interventions. Furthermore, we described the method of p16 immunostaining, the type of antibody and medium used and the various interpretations of p16 immunoreactivity as adopted by each author.
We applied the 1991 Bethesda reporting system (TBS91) for the cytological classification (14). Three cytological groups were considered, i.e. atypical squamous cells of undetermined significance (ASCUS), low-grade (LSIL) and high-grade squamous intra-epithelial lesions (HSIL). If the cytological abnormalities were presented in different reporting formats, they were converted into TBS91 using published standard translation tables (15). The CIN nomenclature was applied in order to describe histological outcomes (16).
Statistical analysis was performed using STATA (Stata Corp., College Station, Texas, US). Random effects models were used to pool proportions for pooling (17) and analysis and interstudy heterogeneity was assessed with the Cochran's 𝑄 test (18).
In the meta-analysis of cytological studies we adopted the cut-off of p16 positivity proposed by each author, whilst for the analysis of histological studies we included in the meta-analytic pool only those that either adopted the distribution of staining proposed by Klaes and colleagues (19) or could be converted to the above classification system. According to the Klaes classification the diffuse staining (>25% of cells stained for p16) was considered as the cut-off of positivity.
The electronic search yielded 584 studies that were assessed for inclusion in the review. Of those, 97 were potentially eligible and subsequently scrutinized in full text (Fig. 1).
Amongst the relevant studies, thirty-six failed to meet the inclusion criteria and were excluded from this overview. Nine of the studies assessed p16 immunostaining in glandular (20– 28) and twenty-one in invasive cervical lesions only (29–49). Another six studies represented duplicate reports and were subsequently excluded (50–55) (Fig. 1).
Sixty-one studies qualified for the overview. Twenty-seven assessed the p16 immunoreactivity in cytological specimens (56 –81) and in six amongst them (56, 58, 59, 66, 70, 72) the biomarker was assessed in histological specimens as well, whereas 34 studies assessed staining only in histological samples (19, 83– 115). The characteristics of included studies are presented in Tables 1 and and22 respectively.
In cytology, the immunostaining for p16 was performed in selected series of smears of different degrees of cytological abnormality. Cervical smears were selected at random from a screening population in only one study (64); however only those women who were HPV tested were included so even in this study there was selection and disease enrichment.
Liquid based cytology was the preparation method adopted in 19 studies (56–59, 61, 62, 64–66, 69, 71, 73, 75, 77–82) and conventional cytology in eight (60, 63, 67, 68, 70, 72, 74, 76). Immunostaining was performed in cell block preparations in two reports (61, 79) and in one study the p16 stain was used with a Papanicolaou counterstain (74). Correlation between cytology, histology and p16 immunoreactivity was assessed in eight studies (62, 65, 70, 74, 77, 80–82) (Table 1).
The histological samples included punch, loop large excision of the transformation zone (LLETZ) or cone biopsies and hysterectomy specimens. In four studies (87, 90, 110, 114) women were followed-up in time allowing an evaluation of the longitudinal predictive value of the biomarker (Table 2).
A variety of antibodies were used for immunodetection of p16: CINtec p16INK4a Cytology Kit (clone E6H4, Dako Cytomation, Glostrup, Denmark), clone G175–405 (PharMingin, San Diego, CA, USA), clone 16P07 (Neomarkers, Fermont, CA, USA), Biosource (Camarillo, CA, USA), clone JC8 (Biocare, Boston, MA, USA), clone 6H12 (Novocastra, Newcastle upon Tyne, UK), clone F-12 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), clone #7962 (Abcam, Cambridge, UK), clone SC468 (Santa Cruz, CA, USA) . The method of staining was similar amongst cytological (conventional or LBC) and histological specimens, with small alterations indicated by each manufacturer.
P16 positivity in cytological specimens was determined according to the distribution of staining into the cell compartments (cytoplasm or nucleus) and the number of cells overexpressing the biomarker. The cut-offs of positivity varied substantially across the studies. Quantitative criteria (i.e. the number of stained cells) were evaluated in 12 reports (57, 60, 61, 63, 67, 68, 70, 72, 75, 78, 79, 81), whereas both quantitative and qualitative parameters (i.e. morphological cytological features) were assessed in nine studies (56, 58, 62, 66, 69, 71, 73, 80, 82) (Table 3). The proportion of smears stained positive for p16 per type of cytological abnormality is presented in Table 3.
The majority of histological studies adopted the classification proposed by Klaes and colleagues (19) that scored the distribution of p16 positivity on a semiquantitive scale as follows: negative (<1% of the cells were positive), sporadic (isolated cells were positive, but <5%), focal (small cell clusters, but <25% of the cells were positive), diffuse (>25% of cells stained positive). In order to reduce the heterogeneity of the results, we converted alternative classifications of staining adopted by other studies into the Klaes classification (Table 4). However, this was not feasible in four studies (72, 96, 111, 113). We considered as positive cut-off for p16 immunoreactivity the diffuse stain that includes both basal and parabasal cell layers and indicates hrHPV-induced transformation (13). The proportion of positive p16 biopsies per grade of histological lesion is presented in Table 4.
The biomarker was hardly evaluated in the triage of women with equivocal cytological results. Only four studies (62, 70, 80, 81) allowed assessment of the accuracy of p16 staining and HPV testing with the hybrid capture 2 (HC2) assay Qiagen, city, state, USA) for an underlying CIN2+ in women with ASCUS (70, 80) or LSIL (62, 80) (Table 5). P16 in general showed lower or equal sensitivity but higher specificity compared to HC2. We could not assess the statistical parameters such as sensitivity and specificity of the biomarker in the fourth study (81).
In histology, the immunohistochemical expression of p16 as a marker of progression risk in low-grade dysplastic lesions of the cervix uteri was evaluated in four studies (87, 90, 110, 114). Negri and colleagues (87) assessed the role of p16 in predicting CIN1 lesions that were likely to progress to CIN3 in a four year follow-up. The authors concluded that although p16 may be expressed in low-grade squamous lesions (CIN1) that undergo spontaneous regression, cases with diffuse staining (>25% cells stained) had a significantly higher tendency to progress to a high-grade lesion than p16-negative cases. Overall, 71.4% and 37.8% of p16-negative and diffusely positive CIN1 regressed at follow-up, respectively, whereas 28.6% and 62.2% negative and diffusely positive CIN1 lesions progressed to CIN3, respectively. In the study by Wang and colleagues (90) CIN1 lesions were followed-up for 5 to 7 years. The positive predictive value (PPV) of diffuse staining for p16 for CIN2+ was 39% and the negative predictive value (NPV) was 85%. Another study (110) demonstrated 96% sensitivity and 39% PPV of diffuse p16 immunostaining in detecting those CIN1 biopsies that could progress to CIN2 in a five year follow-up for a considerably high NPV (96%). Finally, higher risk of progression to CIN3 was observed in CIN2 lesions with diffuse staining for p16 in a 2 year follow-up in a study performed by Omori and colleagues (114).
Taking into consideration the aforementioned discrepancies in the interpretation of staining, we estimated and also presented graphically the mean proportion of samples that were positive for p16 in cytology and histology, stratified by the grade of the lesion (Fig. 2, ,33).
The pooled average proportion of p16-positive smears increased with the severity of cytological abnormality (Fig. 2). Among normal smears, only 12% (95% CI: 7– 17%) were positive for the biomarker compared to 45% of ASCUS (95% CI: 35– 54%), 45% of LSIL (95% CI: 37 –57%) and 89% of HSIL (95% CI: 84– 95%) smears (Fig. 2).
Similarly, in histology, on average only 2% of normal biopsies (95% CI: 0.4–30%) and 38% of CIN1 (95% CI: 23–53%) showed diffuse staining for p16 (Klaes classification) compared to 68% of CIN2 (95% CI: 44– 92%) and 82% of CIN3 (95% CI: 72 – 92%) (Fig. 3).
Results were very heterogeneous in all types of specimen and categories of epithelial neoplasia (p for Q test always <0.001). In particular in the intermediate categories, inter-study variation was extreme. The percentage of smears positive for p16 ranged from 10% (61) to 100% (59) for ASCUS and from 10% (71) to 86% (59) for LSIL (Fig. 2). Similarly, in histology the proportion of CIN1 biopsies exhibiting diffuse staining for p16 varied from 0% (66, 93) to 100% (84) (Fig. 3). Only at both ends of the spectrum of lesions (normal specimen or HSIL/CIN3 lesions) the percentages tended to scatter, after exclusion of one or two outlying studies, within a more narrow range near the left and right border of the forest plots.
The meta-analysis was restricted only to the correlation between p16 immunoreactive samples and grade of intra-epithelial abnormality, on cytological or histological specimens, due to lack of adequate number of studies that explored other clinical parameters of p16 immunostaining.
The progress in the understanding of HPV- related cervical carcinogenesis promoted the evaluation of various biomarkers that could potentially improve the current methods of cervical cancer screening. The cyclin-dependent kinase inhibitor protein p16 is one of the most promising and most studied of these biomarkers.
Immunostaining for p16 can be easily applied in both cytology and histology specimens. However, assessment of its clinical applications is seriously hampered by lack of standardized methodology. Until now, despite several proposed evaluation strategies of p16 in both cytology and histology, there is no general consensus for establishing threshold values above which a sample becomes “p16-positive”. This is especially a problem in p16 cytology, since the histological criteria of p16 positivity (negative, focal, diffuse) are now widely accepted. Several features of p16 staining have not been analyzed systematically, e.g. nuclear versus cytoplasmic staining, or the intensity of staining. Currently these features are not considered relevant for the assessment of p16 staining by the majority of the authors.
As some non-dysplastic cells may also display p16 immunoreactivity additional criteria that discriminate p16 staining in abnormal cells from atrophic or metaplastic cells may increase specificity. In a recent study (71) a nuclear scoring (NS) system was introduced for p16 in LBC samples. By assessing the nuclear size and nuclear-cytoplasmic ratio as well as other nucleomorphic features, a four-tiered classification of nuclear abnormalities has been proposed to describe p16 positive squamous cells. By applying the NS on p16 positive cells the authors noticed a significant increase in specificity for the detection of CIN2+ as compared to calling every slide with any p16 positive cells positive (from 50% to 80%) (82). However, the morphological assessment of p16-stained cells may be difficult in some cases, thus training and experience are necessary (80). An ideal biomarker should avoid relying on subjective criteria as much as possible.
Taking into consideration the above described limitations in the interpretation of p16 staining, we attempted a meta-analysis of the available data in order to evaluate the correlation of the p16 immunostaining to the degree of cytological or histological abnormality. P16 is related to the severity of cervical neoplasia in both cytology and histology however, there is extremely wide variation of p16 positivity in equivocal and low grade squamous cervical abnormalities reflecting its lack of reproducibility in these categories (Fig. 2, ,33).
None of the studies assessed the role of p16 in primary screening for cervical cancer. Although in the study by Sahebali and colleagues (64) cervical smears were randomly obtained from women who underwent routine cervical cancer screening, only those who were tested for HPV were included, so even in this study there was selection and disease enrichment.
Furthermore, the limited number of studies assessing the role of p16 in the triage of LSIL and ASCUS smears versus HC2 (62, 70, 80, 81), combined with the lack of consensus over the interpretation of p16 positivity, hampered our attempt to assess whether the biomarker could outperform the viral testing in these diagnostic categories.
Finally, there is some evidence that diffuse p16 immunostaining in histological specimens could be a predictor of disease progression (87, 90, 110, 114) identifying those low grade lesions (CIN1) that need more intensive follow-up. However, the number of studies is limited.
The very large majority of studies on p16 immunostaining focus on the correlation between the biomarker and the degree of cytological or histological abnormality. Only a few address specific clinical questions such as the role of p16 in primary cervical cancer screening, in the triage of equivocal or low-grade smears versus the HPV-DNA test and its role as a marker of progression risk in low-grade dysplastic lesions of the cervix uteri. Furthermore, the discrepancies in the interpretation of p16 immunostaining in both cytology and histology reduce its reproducibility and hamper the interpretation of the data across different studies.
The available evidence does not allow us to formulate any definite recommendations regarding the applications of p16 in clinical practice. The standardization of sample preparation and evaluation is of paramount importance. After a consensus is reached regarding the interpretation of staining, the biomarker needs to be evaluated in various clinical settings addressing specific clinical questions. A good and meaningful study would require a rigorous colposcopy-biopsy workup (with multiple biopsies taken) and at least a two year follow up to detect disease missed in initial colposcopy/biopsy.