Fax: (011) 81-27-220-8942
Usefulness of liquid-based cytology specimens for the immunocytochemical study of p16 expression and human papillomavirus testing
A comparative study using simultaneously sampled histology materials
Version of Record online: 12 FEB 2004
Copyright © 2004 American Cancer Society
Volume 102, Issue 2, pages 100–108, 25 April 2004
How to Cite
Yoshida, T., Fukuda, T., Sano, T., Kanuma, T., Owada, N. and Nakajima, T. (2004), Usefulness of liquid-based cytology specimens for the immunocytochemical study of p16 expression and human papillomavirus testing. Cancer, 102: 100–108. doi: 10.1002/cncr.20046
- Issue online: 12 APR 2004
- Version of Record online: 12 FEB 2004
- Manuscript Accepted: 9 DEC 2003
- Manuscript Revised: 1 DEC 2003
- Manuscript Received: 25 AUG 2003
- cervical lesions;
- liquid-based Papanicolaou smear;
- p16 immunostaining;
- human papillomavirus (HPV)
In cervical lesions, the overexpression of p16 is reported to be closely associated with high-risk human papillomavirus (HPV) infection. The objective of the current study was to confirm the usefulness of liquid-based cervical specimens for p16 staining as well as tissue sections.
A total of 98 patients with cervical lesions were entered into the current study. After the cytologic examination using liquid-based cervical smears, the same slides were immunostained for p16 and were compared with slides of simultaneously obtained, immunohistologically stained tissue sections. Moreover, the status of the HPV infection was examined by polymerase chain reaction using residual cytologic samples.
Using liquid-based Pap smears, 98 cases were diagnosed as atypical squamous cells of undetermined significance (38 cases), low-grade squamous intraepithelial lesion (12 cases), high-grade squamous intraepithelial lesion (HSIL) (33 cases), and invasive carcinoma (15 cases). The concordance rate between the cytologic and histologic diagnoses was found to be higher in high-grade lesions compared with low-grade lesions. Immunohistochemistry revealed that all HSIL and invasive carcinoma cases contained p16-positive cells in the liquid-based Pap smears and diffuse p16 staining was observed in all high-grade lesions with greater than CIN Grade 3 cervical intraepithelial neoplasia except for two adenocarcinoma cases. Of the 98 cases, 60 were found to be positive for high-risk HPV and 55 of these 60 HPV-positive cases were found to be p16 positive on cytologic examination. There were 16 cases that demonstrated marked discrepancies between the cytologic and histologic diagnoses.
The results of the current study confirmed that the immunohistochemical detection of p16 was more sensitive and specific than HPV status in cervical lesions using a liquid-based method as well as tissue samples, suggesting that p16 should be used as a satisfactory biomarker for the primary screening of cervical cytology. Cancer (Cancer Cytopathol) 2004. © 2004 American Cancer Society.
Cervical carcinoma is a representative model of multistep carcinogenesis that develops into an invasive tumor through squamous metaplasia, dysplasia, and carcinoma in situ.1 Mass screening using cytology samples is a useful and effective method with which to identify patients at risk of developing cervical carcinoma. In developed countries, cervical carcinoma mortality and morbidity rates have been reported to have decreased drastically because of the introduction of mass screening programs.2 The Papanicolaou (Pap) test, which is a cytologic staining technique, is reported to play a key role in the mass screening of cervical carcinoma.3 However, the Pap test is limited with respect to its sensitivity and specificity.4 The false-negative rate for cervical premalignant lesions and cervical carcinoma lies between 15–50% and false-positive rates of approximately 30% have been reported to date.5 These errors in conventional Pap smear testing occur as a result of variations in sampling, preparation, screening, and interpretation. To improve the Pap test, the Bethesda system and liquid-based technology have been introduced during the last 15 years.6 The introduction of the Bethesda system has made the interpretation of findings more uniform and liquid-based technology has decreased the false-negative results found in cervical cytology.7 In gynecologic cytology, many studies have shown that the liquid-based Pap test significantly increases the detection of squamous intraepithelial lesions compared with the conventional Pap test.8
Conversely, it is well known that human papillomavirus (HPV) infection plays a crucial role in the development of various preneoplastic and neoplastic cervical lesions and the progression of these lesions largely depends on the infection of high-risk HPV subtypes such as HPV-16, HPV-18, HPV-31,etc.9, 10 Therefore, the detection of HPV subtypes in these lesions is clinically useful for evaluating a patient's prognosis.11 However, to date, > 80 subtypes of HPV have reportedly been identified, and 18 of these subtypes have been reported to be high-risk subtypes.12, 13
Recently, the CDKN2A gene product, p16 protein (p16), was found to be overexpressed in cervical preneoplastic and neoplastic lesions in which high-risk HPV subtypes exist.14 In cervical carcinogenesis, the E6 and E7 oncogenes of HPV are reported to cause inactivation of the tumor suppressor gene protein products p53 and Rb, resulting in disruption of the p53 and Rb pathways at the G1 checkpoint, respectively.15 In cervical lesions, the overexpression of p16 is believed to result from increased levels of the transcription factor E2F-1, which is released from Rb protein after binding to HPV's oncogenic E7 protein rather than Rb protein phosphorylation by cyclin-dependent kinases.16–18 Therefore, p16 is up-regulated according to the potential of Rb inactivation by oncogenes of HPV. In fact, the overexpression of p16 differs greatly among cervical lesions infected with low-risk and high-risk types of HPV.
The objective of the current study was to confirm the usefulness of liquid-based cytology specimens for the immunohistochemical detection of abnormal cervical cells with the p16 antibody. For this purpose, we investigated the relation between cytologic and histologic findings, p16 overexpression, and HPV infection status in various cervical lesions, including carcinoma.
MATERIALS AND METHODS
Of the patients who visited the Department of Obstetrics and Gynecology at Gunma University Hospital over a 1-year period starting in January 2002, 98 patients with abnormal cytology were selected randomly for the current study and underwent both cytologic and histologic examinations for diagnosis. All patients underwent cervical punch biopsy immediately after the cytologic examination at the time of the first visit. For treatment and final diagnosis, 43 patients underwent conization. The age of the patients ranged from 20–80 years (mean age, 30 years) at the time of the cytologic examination. Prior to the initiation of the current study, the protocol was approved by the Institutional Review Board of Gunma University Hospital. A diagram of specimen retrieval, staining, and HPV detection is illustrated in Figure 1.
Cytology and Tissue Samples
Cytologic specimens were obtained from the patients using a Cervex-Brush® (Rovers Medical Devices B.V.; Oss, The Netherlands) and immediately immersed in fixative solution (SurePath™ preservative fluid; TriPath Imaging Inc., Burlington, NC); the cells then were washed out from the Cervex-Brush into the fixative (Fig. 1). Liquid-based cytologic specimens were processed according to the manufacturer's protocol (TriPath Imaging, Inc.). Briefly, cell debris and inflammatory cells were partially removed from the sample by centrifugal sedimentation through density reagent. After centrifugation, cell pellets were transferred into small plastic chambers, mounted on microscope slides, and fixed with 95% ethanol. Fixed specimens were stained using the Pap method. Tissue samples were fixed with 15% formalin, embedded in paraffin, and processed for routine pathologic examination. The Bethesda system and the cervical intraepithelial neoplasia (CIN) classification were used for the diagnosis of cytology and histology, respectively. Two pathologists examined the cases, and discrepancies between the cytologic and histologic diagnoses were reviewed again in conjunction with the results of immunostaining for p16.
Immunohistochemistry for p16 Using Liquid-Based Pap Smears and Paraffin Sections
Pap-stained slides were examined microscopically for atypical cells, and all atypical cells were photographed with a digital camera. The coverslip then was removed with xylene and the slides were destained with an alcohol series and 1% periodic acid solution (Fig. 1). For the immunohistochemistry for p16, endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 30 minutes at room temperature and antigen retrieval was performed for 20 minutes by boiling in 0.01 M of citric acid phosphate buffer (pH 8.0). Nonspecific staining was eliminated by incubating the tissue with normal bovine serum for 30 minutes at room temperature. The specimens were reacted with mouse monoclonal antibody against p16 (JC8 at a dilution of 1:50; Neomarkers, Fremont, CA) for 30 minutes at room temperature in a humidified chamber, then washed thoroughly and incubated with ENVISION reagent (Dako Cytomation Co. Ltd., Kyoto, Japan) for 30 minutes at room temperature. The specimens were visualized by benzidine reaction and counterstained lightly with Mayer hematoxylin.
For the immunohistochemistry using tissue sections, 3-μm thick paraffin sections were cut from a paraffin block, deparaffinized with xylene, and immunohistochemically stained for p16 in a manner similar to that used for the liquid-based Pap smears.
Evaluation of p16 Immunohistochemistry
The liquid-based Pap smears were evaluated as positive if they included atypical cells showing specific immunoreactivity for p16 in both the nucleus and cytoplasm. In the tissue sections, the results of p16 immunohistochemistry were evaluated as follows: negative if there were no p16-positive cells in the lesion, sporadic positive if there were < 30% p16-positive cells present in the lesion, and diffuse positive if there were > 30% p16-positive cells present in the lesion. The result was statistically analyzed using the Student t test distribution.
HPV-DNA Detection and Typing
Residual cytologic specimens were used for HPV-DNA detection by polymerase chain reaction (PCR) (Fig. 1). Cell DNA for PCR amplification was obtained from the collected cytology specimens by one-step DNA extraction using Chelex® 100 (Promega, Madison, WI). HPV-DNA was amplified by PCR using consensus primers (LICl/LIC2) for the L1 open reading frame. The L1C1/L1C2 primers were as follows: LICl: 5′-CGTAAACGTTTTCCCTATTTTTTT-3′ and LIC2: 5′-TACCCTAAATACTCTGTATTG-3′. These primers allowed for the amplification and identification of at least nine types of genital HPV-DNA. PCR was performed with 40 amplification cycles as described previously.19 The PCR conditions were denaturation at 94 °C for 30 seconds, annealing at 48 °C for 30 seconds, and extension at 72 °C for 45 seconds. As a control for PCR, HPV-negative cervical squamous cell carcinoma was used simultaneously. The amplified HPV-DNA fragments were typed on the basis of restriction fragment length polymorphism (RFLP) among HPV using three restriction enzymes: RsaI, DdeI, and HaeIII. With these RFLP analyses, HPV types 6, 11, 16, 18, 31, 33, 42, 52, and 58 could be detected and other HPV types were classified as unknown. For reconfirmation of HPV-16 and HPV-18 status, PCR using another type-specific primer was performed as described in a previous study.20
The liquid-based slides from the 98 patients were diagnosed cytologically as atypical squamous cells of undetermined significance (ASCUS) (38 cases), low-grade squamous intraepithelial lesion (LSIL) (12 cases), high-grade intraepithelial lesion (HSIL) (33 cases), squamous cell carcinoma (12 cases), and adenocarcinoma (3 cases). Using the Cervex brush for collecting cytologic materials, cervical surface epithelia usually were observed in small clusters on the liquid-based Pap smears (Fig. 2A, E, I, and M). Conversely, the 98 cases were diagnosed histologically to be squamous metaplasia (38 cases), cervical intraepithelial neoplasia (CIN) of type 1 (8 cases), CIN2 (8 cases), CIN3 (29 cases), squamous cell carcinoma (13 cases), endocervical adenocarcinoma (1 case), and endometrioid adenocarcinoma (1 case) (Fig. 2D, H, L, and P). Table 1 shows the comparison between the cytologic and histologic diagnoses. In the low-grade cervical lesions, the concordance rate between the cytologic and histologic diagnoses was low, but high-grade cervical lesions showed a high concordance rate. Histologic diagnosis revealed that ASCUS lesions contained various cervical lesions, including two cases of CIN2, three cases of CIN3, and one invasive carcinoma case. The total concordance rate between the cytologic and histologic diagnoses was 60% in the current study.
|Cytologic diagnosis||No.||Histologic diagnosis|
|HSIL||33||4||2||3 (9)||21 (64)||3|
|Invasive carcinomab||15||2||0||0||2||11 (73)|
Using immunohistochemistry, specific immunoreactivity for p16 was found to be diffusely present both in the cytoplasm and nuclei of positive cells. Table 2 shows that p16-positive cells were found in all cases of HSIL and invasive carcinoma (100%), but the majority of ASCUS cases did not contain any p16-positive cells (13%) (Fig. 2B, J, and N). In the LSIL cases, p16-positive cells were present in 7 of 12 LSIL cases (58%) (Fig. 2F). The immunostaining pattern of p16 in LSIL cases tended to be weak compared with that of the HSIL cases. The immunohistochemistry for p16 in tissue sections demonstrated that low-grade cervical lesions such as squamous metaplasia and CIN1 had a low frequency (8% and 38%, respectively) of p16 positivity as well as a sporadic staining pattern in CIN1 (Table 3) (Fig. 2C, G). In high-grade cervical lesions > CIN3, all lesions were found to be positive for p16 (100%) and the majority demonstrated a diffuse staining pattern for p16 (Fig. 2K, O). In 7 of the 8 CIN2 cases (88%), diffuse, strong staining of dysplastic cells in the lower half of the epithelium was observed. All CIN3 cases were found to be positive for p16; diffuse and sporadic staining was noted in 25 cases and 4 cases, respectively, of the 29 CIN3 cases (86% and 14%, respectively) (Table 3). Diffuse, intense staining for p16 was observed in the 13 invasive squamous cell carcinoma cases. However, two adenocarcinomas showed a sporadic staining pattern.
|HPV total positive (%)||HPV-16||HPV-18||HPV-31||HPV-33||HPV-52||Unknown|
|ASCUS||38||5 (13)||3 (8)||1||1||0||0||1||0|
|LSIL||12||7 (58)||11 (92)||1||1||1||1||1||6|
|HSIL||33||33 (100)||31 (94)||11||0||3||2||2||13|
|Invasive carcinoma||12||12 (100)||12 (100)||5||1||1||0||3||2|
|Adenocarcinoma||3||3 (100)||3 (100)||0||2||0||0||0||1|
|Total||98||60 (61)||60 (61)||18||5||5||3||7||22|
|No.||Sporadic||Diffuse||Total positive (%)|
|Squamous metaplasia||38||2||1||3 (8)|
|Invasive carcinoma||13||0||13||13 (100)|
Using combined PCR and RFLP analyses, HPV-DNA was detected in 60 of the 98 cytology samples investigated (61%) (Table 2). HPV-16 was the most frequently detected HPV subtype, followed by HPV-52 and HPV-31. The majority of HPV subtypes with the exception of HPV-18 had a tendency to be present to a greater extent in high-grade cervical lesions than in HSIL (Table 2). In the current study, none of the low-risk HPV subtypes such as HPV-6 and HPV-11 were detected. Reconfirmation using another set of primers showed the same result. Compared with p16 immunohistochemistry, 55 of the 60 HPV-positive cases (92%) were revealed to be positive for p16 (Table 4). In contrast, 38 HPV-negative cases were comprised of 5 p16-positive cases (13%) and 33 p16-negative cases (87%). The mean age of the patients with the HPV-positive and HPV-negative lesions was 43.7 years and 46.4 years, respectively. Similarly, the mean age of the p16-positive patients (42.2 years) was significantly younger than that of p16-negative patients (51.4 years) (P < 0.01).
|Mean age (yrs)||42.2||50.1|
|HPV+||All types||60||55 (92)||5 (8)|
|HPV-16||18||17 (94)||1 (6)|
|Unknown||22||18 (82)||4 (18)|
|HPV−||38||5 (13)||33 (87)|
|Total||98||60 (61)||38 (39)|
In the low-grade cervical lesions, especially ASCUS, there were several cases for which there were discrepancies between the cytologic and histologic diagnoses (Fig. 3). One ASCUS lesion showed a diffuse staining pattern for p16 immunohistochemistry, and this case was revealed to be positive for HPV-52 and was diagnosed histologically as marked reserve cell hyperplasia of the epithelium (Fig. 3A–D). Another case of ASCUS was diagnosed histologically as CIN2 and did not demonstrate any p16 immunoreactivity in the cytologic and histologic specimens, in which HPV-DNA could not be demonstrated by PCR (Fig. 3E–H). Table 5 summarizes the results of p16 and HPV studies by cytologic diagnosis. In the current study, there were 16 cases with marked discrepancies, defined as differing by > 2 categories between the cytologic and histologic diagnoses (Table 6). These cases were comprised of nine cases of ASCUS, one case of LSIL, four cases of HSIL, and two cases of invasive carcinoma diagnosed cytologically. Paired biopsy samples of the nine ASCUS cases showed three cases of CIN2, three cases of CIN3, and one case of invasive carcinoma by histologic diagnosis. One case of LSIL was found to be CIN3. Four cases of HSIL and two cases of invasive carcinoma were diagnosed histologically as squamous metaplasia (Table 6). Of these 16 cases, 10 cases and 6 cases, respectively, were of a higher diagnostic category and a lower diagnostic category on cytologic diagnosis. It was interesting to note that 8 of the 10 cases with a cytologically higher diagnostic category also were found to have discrepancies with regard to p16 immunoreactivity in the liquid-based Pap smear-biopsy pairs (namely, they were p16 negative in the smear but positive in the biopsy) and that all 6 cases with a cytologically lower diagnostic category were found to be immunohistochemically positive for p16 in the liquid-based Pap smear but were negative on the biopsy.
|Case no.||HPV||Cytologic diagnosis||p16||Histologic diagnosis||p16|
|15||16||Invasive carcinoma||+||Squamous metaplasia||−|
|16||18||Invasive carcinoma||+||Squamous metaplasia||−|
In the current study, cytologic diagnosis using the liquid-based Pap smear test was compared with histologic diagnosis in biopsy samples taken from the same patients. Although case selection was biased, there were discrepancies noted between the cytologic and histologic diagnoses, especially in low-grade cervical lesions. Previous studies have revealed that interobserver reproducibility of cytologic and histologic interpretations is lower in low-grade cervical lesions than in high-grade lesions.21, 22 Recently, a study on interobserver reproducibility in the histologic diagnosis of CIN reported that significant discrepancies occur in the diagnostic interpretation of low-grade lesions.23 Therefore, both cytologic and histologic diagnoses are limited with regard to an exact diagnosis in low-grade CIN lesions, and a new biomarker for dysplastic or abnormal cervical cells is required to aid in the exact diagnosis and to evaluate a patient's prognosis. Conversely, HPV is well known to be a cause of cervical carcinoma.1 In particular, persistent detection of high-risk HPV types is a strong predictor of the development of high-grade cervical precancerous lesions and invasive cervical carcinoma.9, 10 Therefore, in primary cervical carcinoma screening, HPV-DNA testing has been used and the sensitivity and specificity are estimated to be higher and lower, respectively, than that of the conventional Pap smear test.4
Recently the liquid-based Pap test has been introduced to overcome the drawbacks of the conventional Pap smear and has resulted in the reduction of limited and unsatisfactory specimens and improvement in the adequacy and detection rates for squamous intraepithelial lesions.7 The liquid-based Pap test also has an advantage for HPV-DNA testing and immunocytochemistry in that residual liquid specimens that remain after the cytology slide has been made can be used. In the current study, the detection of HPV subtypes was performed with PCR using DNA taken from the residual cells of liquid samples used for the liquid-based Pap smear.
HPV-DNA was detected in 60 of the 98 cytology samples (61%) and the frequency of HPV-DNA detection increased according to the grade of malignancy, with the exception of LSIL lesions. All HPV subtypes detected in the current study belonged to the high-risk group and these results were found to be in keeping with previous findings showing that HPV-16 was the most frequent HPV subtype detected in cervical lesions, followed by the HPV-52 subtype.24
Because the advantages of HPV testing have been established in cervical carcinoma screening, known biomarkers such as Ki-6725 and MN26 antigen have been found to be unsatisfactory in terms of their specificity and sensitivity in primary screening for cervical carcinoma. In 1998, Sano et al. clearly demonstrated the overexpression of p16 in CIN lesions and nearly all cervical carcinomas.16, 30 Moreover, because p16 overexpression is observed in many CIN lesions with high-risk and intermediate-risk HPV infections, p16 is believed to be a suitable biomarker for screening CIN lesions and cervical carcinoma during routine pathologic diagnosis. Several studies using p16 have corroborated the above results and confirmed that nearly all high-grade CIN lesions and invasive carcinomas immunohistochemically express very high levels of p16, whereas normal and hyperplastic cervical epithelia usually do not express p16.21, 30, 31 Moreover, these findings also have been confirmed in cytologic specimens using liquid-based Pap smears. However, to our knowledge, there are no systematic reports published to date examining immunohistochemical p16 expression and HPV status in both cytology and tissue samples obtained simultaneously from the same patient. In the current study, the percentage diffuse staining pattern or positivity for p16 was found to increase according to the severity of the cervical lesions in both the cervical tissue and liquid-based Pap smear samples. Moreover, high-grade cervical lesions > CIN3 and HSIL demonstrated a diffuse staining pattern for p16 and the presence of p16-positive cells, respectively.
As shown in Table 5, p16-positive cells were noted in all HSIL and cervical carcinoma samples and the rate of HPV detection was found to be very high in these lesions. Moreover, the results of the current study indicate that the immunohistochemical detection of p16 might be more advantageous in terms of specificity and sensitivity for cervical carcinoma screening than HPV testing. A previous immunohistochemical study of p16 using ThinPrep® smears (Cytyc Corporation, Boxborough, MA) has revealed that p16 marks dysplastic squamous and glandular cells of the cervix with a sensitivity of 99.9% and a specificity of 100%, values that are far beyond those of HPV.30 Even in adenocarcinoma of the cervix, it has been reported that p16 diffuse expression is correlated with high-risk HPV infection.31 In LSIL cases, the frequency of p16 expression was found to be relatively low in the current study compared with previously published data.21, 29 This may be attributed most likely to the different primary antibody used for the study.
Previous long-term follow-up of women with a diagnosis of ASCUS has revealed that 20% of these patients develop LSIL and 10% develop HSIL.27 As the definition of ASCUS has emphasized exclusion criteria rather than what criteria should be included in the category, it is inevitable that poorly sampled HSIL is interpreted as ASCUS. These facts have forced the introduction of new diagnostic criteria that can distinguish atypical squamous cells from HSIL (ASC-H) in the 2001 Bethesda system.28 In the current study, there were five p16-positive ASCUS cases, three of which were positive for HPV. Therefore, p16 immunohistochemical analysis might help to differentiate ASC-H from ASCUS in cytologic smears as well as improve histologic interobserver disagreement with regard to the pathologic diagnosis of cervical specimens. In the current study, there were a significant number of cases with marked discrepancies in the interpretation between the cytologic and histologic diagnoses of the simultaneously sampled Pap smears and biopsy specimens. These discrepancies between the cytologic and histologic diagnoses are likely to be attributed to sampling error rather than interpretation error because the p16 staining results in the discrepant cases also showed discordance between the liquid-based Pap smears and biopsy specimens. These findings indicate that p16 immunostaining of liquid-based Pap smears and simultaneously obtained biopsy specimens is helpful in arbitrating some diagnostic disagreements, although obtaining additional biopsies or Pap smears might eliminate these discrepancies.
- 2National Cancer Institute. Surveillance, Epidemiology, and End Results. Washington, DC: Department of Health and Human Services, 2000.