Fax: (570) 271-5916
Immunohistochemical detection of p16INK4a in liquid-based cytology specimens on cell block sections†
Article first published online: 28 FEB 2007
Copyright © 2007 American Cancer Society
Volume 111, Issue 2, pages 74–82, 25 April 2007
How to Cite
Liu, H., Shi, J., Wilkerson, M., Huang, Y., Meschter, S., Dupree, W. and Lin, F. (2007), Immunohistochemical detection of p16INK4a in liquid-based cytology specimens on cell block sections. Cancer, 111: 74–82. doi: 10.1002/cncr.22577
Presented in part at the 54th American Society of Cytopathology Annual Scientific Meeting in Toronto, Ontario, Canada, November 3–8, 2006.
- Issue published online: 12 APR 2007
- Article first published online: 28 FEB 2007
- Manuscript Accepted: 4 JAN 2007
- Manuscript Revised: 28 NOV 2006
- Manuscript Received: 31 OCT 2006
- cervical smears;
- liquid-based thin layer;
- cell blocks;
- squamous intraepithelial lesions
Colposcopy biopsy procedure is a standard recommendation for atypical squamous cell cannot exclude high-grade lesion (ASC-H) in abnormal Papanicolaou smears. p16 (p16INK4a), a cell cycle regulator, has been shown to be overexpressed in squamous dysplasia. To further improve the diagnostic accuracy of the ASC-H Papanicolaou smear and to reduce unnecessary procedures, the authors evaluated the utility of immunodetection of p16 in liquid-based cytology specimens on cell blocks.
Seventy-five liquid-based (SurePath; TriPath Imaging, Inc. Burlington, NC) cytology specimens were prepared for cell blocks. Three groups (G1, G2, and G3) of cases were included: G1 comprised 44 cases of ASC-H; G2, 14 cases of high-grade dysplasia; and G3, 17 negative/reactive cases. All cases in G1 were confirmed by cervical biopsy or Digene Hybrid Capture 2 (Digene, Gaithersburg, Md) human papilloma virus (HPV) testing. Immunodetection for p16 was performed on cell blocks.
In G1, 26 of 44 (59%) cases showed squamous dysplasia, with 14 high-grade squamous intraepithelial lesion (HSIL) cases. Twenty-two of 28 (79%) p16-positive cases were confirmed by surgical biopsy or HPV testing, with a diagnostic sensitivity of 85%, specificity of 67%, positive predictive value (PPV) of 79%, and negative predictive value (NPV) of 75%. Four cases with false-negative staining for p16 were identified. All 28 cases of HSIL (14 from G1 and 14 from G2) were positive for p16.
1) p16 is a sensitive marker to confirm the diagnosis of ASC-H on a cell block; 2) Multiple unstained slides with adequate cellularity can be obtained from each cell block; and 3) Additional markers can be used to further increase diagnostic sensitivity and specificity. Cancer (Cancer Cytopathol) 2007. © 2007 American Cancer Society.
The mortality rate for cervical cancer has dramatically declined since the introduction of mass screening by the Papanicolaou (Pap) smear.1 The 1991 and 2001 revisions of the Bethesda System for interpretation of Pap smear findings have established more uniform diagnostic criteria and subsequently reduced interpretational errors. In the past 15 years, the addition of liquid-based technology has further increased the diagnostic accuracy of the Pap smear.2 Unfortunately, despite its great success, the Pap smear is not perfect. A false-negative rate of 15% to 50% and false-positive rate of 30% have been reported.1
In the 2001 version of the Bethesda System for Reporting Cervical Cytology, 5 different categories are recommended for squamous cell abnormalities, including ASCUS (atypical squamous cell of undetermined significance, ASC-H (atypical squamous cells, cannot exclude high-grade squamous lesion), LSIL (low-grade squamous intraepithelial lesion), HSIL (high-grade squamous intraepithelial lesion, and SCC (squamous cell carcinoma). Following the Bethesda 2001 workshop, the American Society for Colposcopy and Cervical Pathology developed consensus guidelines for management of squamous cell abnormalities. Human papilloma virus (HPV) testing was recommended for ASCUC.2 Colposcopy/biopsy was recommended for ASC-H, LSIL, HSIL, and SCC.2
A diagnosis of ASC-H is made in <1% of all abnormal Pap smears.2 Colposcopy/biopsy is the standard recommendation for ASC-H.2 Though associated with a high-positive predictive value for underlying high-grade lesions, other conditions may mimic ASC-H, such as atrophy, inflammation, and artifact.3
It is generally believed that most cervical squamous cell intraepithelial lesions are associated with oncogenic human papillomavirus infection, especially the high-risk types such as HPV 16, 18, 33, 35 45, 56, and 58.4 Recently, p16, a cell cycle regulator, has been shown to be overexpressed in 100% of high-grade squamous dysplasias, invasive squamous cell carcinomas, and adenocarcinomas of the endocervix, and in approximately 70% to 100% of low-grade squamous cell dysplasias.5–9 In contrast, benign squamous epithelium and endocervical epithelium are usually negative or very focally positive for p16.5–9
In this study, we tested the utility of p16 to detect HPV infection in liquid-based cytology specimens by using cell block sections. Correlations and comparisons were made among cytologic, cell block, immunostain results, and surgical diagnoses. The objective of this study is to further improve the diagnostic accuracy of the Pap smear in the diagnosis of ASC-H and to reduce unnecessary colposcopy/biopsy procedures.
MATERIALS AND METHODS
Sample Collection and Preparation
Following the approval of this study protocol by the institutional review board at Geisinger Medical Center, 75 samples were collected from the remaining liquid-based cytology specimens (SurePath, TriPath Imaging, Burlington, NC). Three groups of cases were included in this study: Group 1 (G1) comprised 44 cases of ASC-H; Group 2 (G2), 14 cases of HSIL; and Group 3 (G3), 17 negative/reactive cases. The 75 original Pap smears contained adequate cellularity. Samples that were satisfactory for evaluation but limited by low cellularity were not included in this study.
Preparation of Cell Block
The residual SurePath sample was placed in a 50 mL Falcon tube and spun in Beckman Coulter Allegra 6 Centrifuge (Beckman Coulter, Fullerton, Calif) for 10 minutes at 2000 rpm. The supernatant was poured off, which left the small pellet/button in the bottom of the tube. Approximately 5 drops of plasma were added to the cell button to resuspend it. Then approximately 5 drops of bovine thrombin (Fisher Scientific Item # 23-306291; Thermo Fisher Scientific, Waltham, Mass) were added, and the mixture was allowed to stand for 10 minutes. The cell pellet was fixed in 10% formalin for 10 minutes. The cell block material was transferred to a biopsy bag by pouring the contents of the Falcon tube into a biopsy bag over a funnel and beaker. The cell block material in the biopsy bag was placed into a histology cassette. The cassettes were processed as routine surgical specimens.
Cell block slides from all 3 groups were blindly reviewed. The cytologic diagnoses and p16 staining results were disassociated from each slide. Four categories were assigned to the cell block diagnoses, 1) high-grade squamous intraepithelial lesion; 2) low-grade squamous intraepithelial lesion; 3) atypical squamous cells; and 4) negative/reactive cellular changes.
Immunohistochemical testing for the detection of mouse monoclonal anti-p16 antibody (16P04 also known as JC2; 1:50 dilution; EDTA antigen retrieval; Cell Marque, Hot Springs, Ark) was performed on the formalin-fixed and paraffin-embedded cell block sections using an EnVision-HRP detection kit (Dako, Glostrup, Denmark). The detailed staining protocol was similar to a protocol described in a previously published report.10 The results were recorded as negative (no staining or <3 positively stained cells); 1+ (3–10 positively stained cells); and 2+ (>10 positively stained cells). The staining intensity was recorded as strong (when observable at 4× or 10× magnification power) and weak (when observable at 20× or 40× magnification power). The presence of groups/sheets of positively stained epithelial cells (usually >5 cells) was also recorded. Epithelial cells with nuclear staining and both cytoplasmic and nuclear staining were regarded as positive results. Tissues with a biopsy proven to be positive or negative for squamous cell dysplasia were used as positive and negative controls.
Immunodetection of a cocktail of p16 and ProEx C (TriPath Imaging, Burlington, NC) which is a mixture of MCM2 (minichromosome maintenance protein 2) and DNA topoisomerase II alpha (TriPath Imaging, Burlington, NC), was performed on the 4 cases of LSIL from G1. Those 4 cases were negative for p16 immunostain. The immunostaining protocol was similar to that of p16, except a mixture of an equal amount of p16 (1:25 dilution) and prediluted ProEx C was used. Nuclear staining or both nuclear and cytoplasmic staining was considered to be a positive result. Details are described in the DISCUSSION section below.
Determination of Diagnostic Sensitivity and Specificity
The sensitivity, specificity, and positive and negative predictive values were calculated.
All cases in G1 were confirmed by either cervical biopsy (40 cases) or Digene Hybrid Capture 2 HPV testing (4 cases). All cases in G2 were confirmed by cervical biopsy.
Interpretation of Cell Block Slides
Adequate cellularity was obtained from all 75 cell blocks, with approximately 10% of cases showing low cellularity. Figures 1A, 1B, and 1C show high, intermediate, and low cellularity, respectively. Overall, good correlation (70%) with the cervical biopsy was achieved in G2 (high-grade squamous intraepithelial lesion) and a high-grade squamous lesion from G1, whereas correlation with negative Pap smear (G3) and negative and low-grade lesions from G1 was relatively poor, approximately 50%, 50%, and 60%, respectively.
Interpretation of p16 Immunostained Slides
Four slides with adequate cellularity could be obtained from each cell block. In G1, 26 of 44 (59%) cases, confirmed by cervical biopsy (n = 22) or positive HPV testing (n = 4), showed squamous cell intraepithelial lesions, including 14 cases with HSIL. Twenty-two of 28 (79%) p16-positive cases were confirmed by surgical biopsy (N = 20) and positive HPV testing (n = 2), with a diagnostic sensitivity of 85%, specificity of 67%, positive predictive value (PPV) of 79%, and negative predictive value (NPV) of 75%. Four cases with false-negative staining for p16 were identified. Two cases were low-grade dysplasia confirmed by cervical biopsy, and the other 2 were positive for HPV. One case of biopsy-proven LSIL with positive p16 staining is shown in Figures 2A and 2B. Six false-positive cases were noted; 3 showed that the positively stained cells had glandular morphology, consistent with benign endocervical cells. In addition, positively stained endometrial cells were observed in 1 case. The diagnostic sensitivity, specificity, positive predictive value, and negative predictive value are summarized in Table 1.
|Cervical biopsy and HPV testing||Total cases|
|Positive cases||Negative cases|
|p16||Positive Cases||22* (a)||6 (b)||28 (a+b)|
|Negative Cases||4† (c)||12 (d)||16 (c+d)|
All 28 cases of HSIL (14 from G1 and 14 from G2) were positive for p16. Three different staining distributions were noted: 1) high-grade cells forming a group/sheet as shown in hematoxylin and eosin (H & E)-stained section (Fig. 3A) and positive for p16 staining (Fig. 3B); 2) many single high-grade cells as shown in H & E-stained slide (Fig. 4A) and p16-positive cells in a low magnification (Fig. 4B) and a high magnification (Fig. 4C); and 3) rare single high-grade cells as shown in H & E-stained slide (Fig. 5A) and p16-positive cells (Fig. 5B).
Three of 17 (18%) cases in G3 showed cellular components with immunoreactivity for p16. The positively stained cells were endocervical cells (Fig, 6A), metaplastic squamous cells (Fig, 6B), and endometrial cells (Fig, 6C). Scattered inflammatory cells, fungal elements, and Trichomonas also may be weakly positive for p16.
The overall p16 staining results and the correlation with the cervical biopsy results are shown in Table 2. The distribution of p16 staining in 75 cases is summarized in Table 3. All cases, except 3 cases with weak staining, which showed strong nuclear and cytoplasmic or nuclear staining for p16.
|Marker||HSIL, n = 28||LSIL, n = 12||Negative, n = 35*|
|p16-positive cases||28 (100%)||8 (67%)||9 (25.7%)|
|p16-negative cases||0||4 (33%)||26 (74.3%)|
|HSIL, n = 28||0||5||23||12|
|LSIL, n = 12||4||2||6||1|
|Negative, n = 35||26||2||7||3|
Interpretation of p16/ProEx C Immunostained Slides
Three of 4 p16-negative cases (ASC-H) were positively stained with the mixture of p16 and ProEx C. The positively stained cells showed nuclear or both nuclear and cytoplasmic staining. Two of the 3 p16/ProEx C-positive cases were confirmed by cervical biopsy. One case was negative for the staining.
Only a limited number of reports in the literature delineate the utility of a cell block prepared from a liquid-based cytologic smear in the diagnosis of cervical squamous cell lesions.11–16 The most recent study by Akpolat and coworkers16 demonstrated that 1) a good correlation (73%) between cell block diagnosis and Pap (ThinPrep; Cytyc, Marlborough, Mass) diagnosis was obtained; 2) a relatively high false-negative rate in the diagnosis of high-grade lesion (30%) was reported because of lack of diagnostic cells on cell block sections; and 3) immunostaining for p16 and Ki-67 performed on cell block sections showed a significant association with the presence of significant lesion on both cell blocks and ThinPrep smears. In their study, a cutoff of 10% p16-positive cells in a given slide was used to define a positive versus a negative result (ie, positive when >10% of cells stained with p16). In addition, follow-up biopsy results were not available; instead, diagnoses from Pap smears served as the gold standard for diagnosis.
In contrast to the previous studies, including the 1 from Akpolat et al,16 our current study revealed some similarities and differences. Therefore, the following issues were addressed: 1) specimen adequacy; 2) limitations on cell bock diagnosis; 3) interpretation of p16 staining; and, more importantly, 4) the potential utility of this technique in conjunction with specific biomarkers in the screening and diagnosis of cervical squamous cell lesions.
Our current study showed that an adequate cellularity was obtained from all samples, with 4 unstained slides for each case, which was the important first step to insure successfully applying biomarkers by immunostains on these slides. It should be mentioned that previous studies reported up to 25% of cases with insufficient material in cell blocks,11, 15 and a high percentage of slides lacking diagnostic cells was reported by Akpolat and coworkers.16 It is not clear what criteria were applied to assess specimen adequacy in these previous studies.11–16
To the best of our ken, well-defined criteria on the specimen adequacy of cell block slides are not available. Therefore, several issues concerning an adequate specimen on a cell block need to be discussed here. First, we used residual material from SurePath cytologic samples instead of ThinPrep samples as reported in others' studies. Our samples may have contained more cellular components, as less are used to prepare a SurePath smear than a ThinPrep smear.
Second, as aforementioned, there is no consensus on how many cells a cell block slide should contain to be considered an adequate sample. In fact, to the best of our knowledge, the adequacy of cellularity for a cell block has not been discussed in previously published studies.11–16 The study by Diaz-Rosario and Kabawat showed that sufficient cellularity from cell blocks was obtained from 75 of 95 (79%) cases.15 In contrast, Keyhani-Rofagha and Vesey-Shecket reported a satisfactory cellularity in 95.5% of cases in their study of 131 cases.11 However, the meaning of sufficient and/or satisfactory cellularity was not further defined.11, 15 In the current study, 10 intermediate power fields (magnification, 20×) of cellular components can be easily observed on each cell block slide. Hence, an adequate cell block slide was defined as one that contained at least 10 intermediate power fields (20×) of cellular elements. We further divided the adequate cellularity into 3 categories: high cellularity (a 20× field showing over 50% cellular components), intermediate cellularity (a 20× field containing 25% to 50% cellular components), and low cellularity (a 20× field revealing <25% cellular components). With these criteria, 90% of cases met the standard for intermediate to high cellularity as shown in Figures 1A and 1B, and 10% cases were classified as low cellularity as shown in Figure 1C.
Finally, we recommend that 4 unstained slides should be obtained at the outset to avoid repeatedly trimming the cell block. One slide should be stained for H & E, and the remaining 3 should be kept for immunostaining.
The utility of cell blocks in the diagnosis of squamous cell intraepithelial lesions is relatively controversial. Akpolat et al and Keyhani-Rofagha et al reported 73% and 69% agreement with ThinPrep diagnosis, respectively.11, 16 However, 8 cases of high-grade lesions in Akoplat's study were rendered LSIL or negative, and 8 of 11 cases of ASCUS were interpreted as negative.16 Similar results from Richard et al showed that cell blocks were not helpful in distinguishing ASCUS from LSIL in 43 of 45 cases.14 Furthermore, cell blocks may not be useful in distinguishing HSIL from atrophy and inflammatory atypia.14 Our current study demonstrated that a relatively good correlation (70%) in the diagnosis of HSIL by cell blocks can be obtained. However, a significant number of cases of inflammatory atypia, atrophy, and even endometrial cells may mimic ASCUS and LSIL, and in a few instances, even HSIL. It should be emphasized that all cases in G1 (ASC-H) and G2 (HSIL) of this study were confirmed by cervical biopsy or positive HPV testing. In contrast, some previous studies were based on the ThinPrep diagnosis as the gold standard,16–18 which may raise the concern of the reliability of the correlation between the cell block diagnosis and the final diagnosis documented in the previous studies. It is obviously debatable whether Pap smear or cervical biopsy should serve as the gold standard.16 Our limited experience suggests that cell blocks may play a limited role in diagnosis of ASCUS, ASC-H, and LSIL.
The value of p16 in diagnosis of cervical glandular cell lesions and squamous cell lesions has been well documented.5–9 In surgical biopsy specimens, nearly 100% of HSILs, squamous cell carcinomas, and endocervical adenocarcinomas were positive for p16. Approximately 70% of LSILs were positive for p16 as well.5–9 In liquid-based cytologic smears, similar diagnostic sensitivities have been reported.19–21 Our current study demonstrated that p16 immunoreactivity was observed in 100% of HSILs and 67% of LSILs, which is compatible with previous studies that used liquid-based cytologic smears and cell blocks.16, 19, 21–24 However, false-positive staining was noted in 6 of 18 negative cases in G1 (ASC-H) confirmed by cervical biopsy and in 3 of 17 negative cases in G3 (negative/reactive by Pap smear diagnosis only). In G1, 4 of 6 cases with positively stained cells were either endocervical or endometrial cells. If these 4 cases were considered negative, the positive predictive value (PPV) for G1 would have been higher. Similarly, all 3 positive cases in G3 were endocervical glandular cells, endometrial cells, and probable metaplastic squamous cells, respectively.
It is important to re-emphasize here that HSIL can be easily detected by using cell blocks in conjunction with p16 immunostaining. A high diagnostic sensitivity and relatively good specificity can be obtained. If the positively stained endocervical glandular cells, inflammatory cells, endometrial cells, and metaplastic squamous cells could be excluded on the basis of cytomorphology, the positive predictive value for detection of HSIL would be excellent. However, if one attempts to use this technique, ie, combined cell block and p16 immunostaining, as an initial screening procedure, the diagnostic sensitivity in detection of LSIL would not be satisfactory.
To further improve diagnostic sensitivity in the detection of LSIL, we have tried to use a cocktail of p16 and ProEx C. ProEx C is a mixture of 2 monoclonal antibodies against MCM2 (minichromosome maintenance protein) and DNA topoisomerase II alpha. MCM2, together with other MCM proteins, plays an important role in the early stage of DNA replication. It has been shown that MCM proteins are useful in the detection of cancers, including cervical dysplasias and carcinomas.25–28 Similarly, DNA topoisomerase II alpha plays an important role in the cell cycle by modulating the topological structure of DNA during replication.29 MCM proteins (including MCM2) and DNA topoisomerase II alpha have been shown to be relatively sensitive biomarkers for the detection of squamous cell intraepithelial lesions in both cervical biopsy specimens and liquid-based cytology smears.30–32 The rationale for using a cocktail of 3 antibodies (p16, MCM2, and DNA topoisomerase II alpha) is based on the fact that each of these 3 markers is capable of detecting approximately 70% of LSILs and nearly 100% of HSILs.
Our unpublished data also demonstrate that this mixture (a cocktail of p16 and ProEx C) could detect 100% of HSILs and nearly 100% LSILs on cervical biopsy specimens. Theoretically, this mixture should detect nearly 100% of both HSILs and LSILs on cell blocks. Interestingly, 3 of 4 p16-negative cases (ASC-H) were positively stained with the mixture of p16 and ProEx C. Two of the 3 cases were confirmed by cervical biopsy. The p16 and ProEx C-negative case was positive for HPV, but it is not certain whether it was positive for squamous cell dysplasia. It has been shown that HPV was detected in 27% of Pap smears without an intraepithelial squamous cell lesion.33 It is unknown whether the diagnostic specificity may decline because of a parallel increase in nonspecific staining with ProEx C on other nondysplastic cells. Further study in a large series is warranted to confirm this preliminary finding.
In summary, our preliminary data show that 1) p16 is a sensitive marker to confirm the diagnosis of HSIL and high-grade lesions within ASC-H on a cell block; 2) Multiple unstained slides with adequate cellularity can be obtained from each cell block prepared from SurePath specimens, compared with only 1 direct smear usually obtained from each case; and 3) Additional markers, such as ProEx C, can be used to further increase diagnostic sensitivity and specificity.
- 1The Pap smear. In: DeMayRM, ed. The art and science of cytopathology. Chicago: American Society of Clinical Pathologists Press; 1996: 61–205..
- 15Cell block preparation by inverted filter sedimentation is useful in the differential diagnosis of atypical glandular cells of undetermined significance in ThinPrep specimens. Cancer. 2000; 90: 265–272., .
- 31Correlation of p16, MCM2, and cyclin E as molecular markers for cervical disease in IHC and ICC formats [abstract]. Mod Pathol. 2005; 18( suppl): 280A. Abstract 928., , , , , .
- 32Comparison of nuclear, cytoplasmic and membrane staining of molecular markers in detecting cervical dysplasia in SurePath cytology specimens [abstract]. Acta Cytologica. 2004; 48: 696. Abstract 54., , , et al.
- 33Distribution of human papillomavirus types in ThinPrep Papanicolaou tests classified according to the Bethesda 2001 terminology and correlations with patient age and biopsy outcomes. Cancer. 2006; 106: 1054–1064., , , , , .