SEARCH

SEARCH BY CITATION

Keywords:

  • dual immunocytochemical staining;
  • HSIL;
  • AGC;
  • CIN;
  • diagnostic accuracy

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES

BACKGROUND:

Although previous studies have shown that p16INK4a and Ki-67 are sensitive and specific markers for high-grade lesions (≥CIN2) on cervical biopsies, limited information is available regarding the performance of a dual-staining approach as a diagnostic adjunct in cervical cytology. We evaluated a dual p16INK4a/Ki-67 immunocytochemistry (ICC) assay to determine its sensitivity and specificity versus that of high-risk HPV (HR-HPV) in a US-based pilot cytology study.

METHODS:

ThinPrep specimens from 122 cervical cytology specimens encompassing 23 negative (NILM), 20 ASC-US, 22 LSIL, 17 ASCH, 22 HSIL, and 18 AGC cases were processed for multiplexed ICC staining using a CINtec Plus Kit. Dual-positive assay results were defined based on the detection of 1 or more epithelial cells that were stained for both p16INK4a and Ki-67 without regard to cellular morphology. HR-HPV testing was performed by multiplex PCR with capillary electrophoresis genotyping.

RESULTS:

Dual staining for p16INK4a and Ki-67 was frequently detected in HSIL and AGC but was rarely detected in NILM cases. The HR-HPV assay showed a sensitivity of 76.2% and a specificity of 55.8% for the detection of clinically significant cervical squamous or endometrial lesions. In contrast, the colocalization of p16INK4a plus Ki-67 maintained a high sensitivity of 81.8% and improved specificity to 81.8% for biopsy-confirmed CIN2/3, endocervical adenocarcinoma, or endometrial adenocarcinoma.

CONCLUSIONS:

Dual staining for p16INK4a/Ki-67 immunocytochemistry dramatically increased specificity and maintained high-level sensitivity for the diagnosis of CIN2/3 or glandular lesions compared with PCR-based testing for HR-HPV. Cancer (Cancer Cytopathol) 2012. © 2011 American Cancer Society.

The routine use of cervical cytology as a screening method has significantly decreased the incidence of cervical cancer in the United States. This is because of the excellent correlation between cytology and histology when cervical intraepithelial neoplasia 2/3 (CIN)/high-grade squamous intraepithelial lesion (HSIL), squamous cell carcinoma (SCC), adenocarcinoma in situ (AIS), or adenocarcinoma is diagnosed on routine screening Papanicolaou (Pap) smears. There is poor specificity, however, in identifying underlying high—grade lesions in patients with clinically indeterminate cytologic abnormalities such as atypical squamous cells of unknown significance (ASC-US), atypical squamous cells—cannot exclude high-grade squamous intraepithelial lesion (ASC-H), low-grade squamous intraepithelial lesion (LSIL), or atypical glandular cells (AGC).1 This is a significant problem as there are more than 2 million women per year in the United States who have a diagnosis of ASC-US on Pap smear, and approximately 5% of these patients will have histologically confirmed CIN3+ at colposcopy.2-4 Novel approaches are needed to enhance the specificity of the cytologic diagnosis for clinically significant lesions such as HSIL and AGC in patients with equivocal results.

p16INK4a has been proven to be a sensitive as well as a specific marker of high-grade squamous and glandular neoplasia of the cervix5-7 that can reliably identify underestimated CIN in high-risk human papillomavirus (HR-HPV)–positive patients with a negative cervical biopsy.8 It has also been shown to have similar sensitivity but superior specificity at identifying underlying clinically significant subtypes when compared with polymerase chain reaction (PCR) for the detection of high-risk HPV subtypes.9 Ki-67 is a nuclear antigen and a cellular proliferation marker expressed in all cell-cycle phases except G0. The monoclonal antibody MIB-1, which is directed against Ki-67, is routinely used by pathologists as a method to determine the proliferative activity of cells in histological specimens10, 11 and has been used in the evaluation of equivocal cervical lesions and Pap smears to increase specificity.12, 13

There is substantial interpretative variability of cervical cytology.1 We hypothesized that the specificity of detecting lesions that have a clinical impact would be improved using a dual immunocytochemistry approach when compared with HR-HPV PCR, or p16INK4a staining alone. The goal of this study was to evaluate a dual immunocytochemical assay for Ki-67 and p16INK4a and to determine its impact on detection of clinically significant lesions of the cervical mucosa in a pilot series of cervical cytology specimens.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES

Case Selection

Cytology specimens were collected from the outpatient clinics affiliated with the University of Massachusetts, Worchester, Massachusetts, between January 1, 2010, and July 29, 2010. Over this period, a total of more than 40,000 cases were processed for routine diagnosis, including 35,587 NILM, 2567 ASC-US, 162 ASC-H, 1899 LSIL, 490 HSIL, 56 AGC, 2 SCC, and 8 adenocarcinoma cases. Following routine protocols of the Cytopathology Laboratory of the University of Massachusetts, the Digene HC2 test (Qiagen Inc., Valencia CA) was performed to guide clinical triage of ASC-US cases but was rarely performed on samples from other diagnostic categories. A small number of cases were retrieved from the cytology laboratory based on availability (routine laboratory practice was to routinely discard most vials shortly after diagnosis) and the presence of more than 1 mL of residual volume in the cytology vials after a diagnosis had been rendered. The final test series consisted of the number of cases initially selected from each diagnostic group: 23 NILM, 20 ASC-US, 22 LSIL, 22 HSIL, 17 ASC-H, and 18 AGC cases. The study was performed under institutional review board approval for both institutions.

Cytology Specimen Preparation

Cervical cytology specimens were collected using a broom-like device, were placed into ThinPrep (Hologic, Marlborough, MA) vials containing PreserveCyt transport medium, and were processed for routine staining, screening, and diagnosis. After the final diagnostic reports were released, the residual fluid cytology specimens and corresponding paraffin-embedded tissue blocks of the colposcopic biopsies were deidentified and entered into the study.

p16INK4a and Ki-67 Dual Localization in Cervical Cytology Specimens

An immunocytochemical (ICC) multiplex assay for the simultaneous qualitative detection of p16INK4a and Ki-67 was performed using a CINtec Plus Kit (MTM Laboratories, Westborough, MA) according to the manufacturer's guidelines. The primary mouse monoclonal antibody clone E6H4 to p16INK4a and the primary rabbit monoclonal antibody clone 274-11 AC3 to Ki-67 were used along with polymer reagent conjugated to HRPO and goat antimouse Fab′ and polymer reagent conjugated to alkaline phosphatase and goat antirabbit Fab. The chromogens utilized were 3,3′ diaminobenzidine (DAB; brown stain = p16INK4a) and fast red chromogen (red stain = p16INK4a). A ThinPrep 2000 was used to prepare the slides. Evaluation was performed by 2 individuals who were blinded to the cytology diagnosis. Positive cytology test results were defined as staining in epithelial cells with both p16INK4a brown nuclear and cytoplasmic staining (DAB chromogen) and Ki-67 bright red nuclear staining (fast red chromogen), resulting in cells with brown cytoplasm and red-brown- to maroon-colored nuclei. Immunocytochemical test results were based on the detection of 1 or more dual-stained epithelial cells without regard to the specific cytologic diagnosis of the immunoreactive cells. Negative controls included substitution of the primary antibody with subclass-matched monoclonal immunoglobulins; HeLa cell–spiked cytology specimens served as positive controls.

HR-HPV Detection by Multiplex Polymerase Chain Reaction Using Genotype-Specific Primers

DNA was extracted from cytology samples using proteinase K (Roche Applied Science, Indianapolis, IN; 200 μg/mL final concentration) in 100 μL of digestion buffer (10 mM Tris HCl [pH 8.5], 1 mM ethylenediaminetetraacetic acid, 0.5% Tween 20) at 55°C for 3 hours. Multiplex PCR for HR-HPV detection was performed as described previously by Nishiwaki et al14 with minor modifications. Briefly, PCR was performed using 1 μL of digested sample as the DNA template, 10 μL of 2× Qiagen (Valencia, CA) multiplex PCR master mix, 7 μL of nuclease-free distilled water, and 2 μL of the total primer mix at a final concentration of 0.1 μM. The HPV multiplex primer mix (Eurofins MWG Operon, Huntsville, AL) consisted of primers to detect 16 HPV genotypes (genotypes 6, 11, 16, 18, 30, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 66) along with another primer pair to amplify a 100-bp aminolevulinate synthase 1 internal positive control. Amplification was performed by incubation at 95°C for 15 minutes, followed by 34 cycles of denaturation for 30 seconds at 94°C, 1.5 minutes of annealing at 70°C, and 1 minute of elongation at 72°C in a PTC-225 Peltier Thermal cycler apparatus (MJ Research, Waltham, MA). HeLa and SiHa DNA-positive controls and reagent aqueous control samples were included with each PCR run.

Individual HPV Typing by Capillary Electrophoresis

Ten-microliter aliquots of the PCR products were processed by capillary electrophoresis at the genomics core facility of Stony Brook University Medical Center, using a QIAxcel DNA Screening Kit (2400; Qiagen, Valencia, CA). The HPV type(s) in each sample were analyzed using the amplicon size of each HPV type normalized to the 100-bp internal control in each sample and the 2 positive controls used in each reaction run. In addition, PCR reactions were performed to confirm each of the HPV types using individual primers in at least 3 samples with a single infection for each HPV type. High-risk HPV–positive test results were defined as detection of types 16, 18, 31, 33, 15, 39, 45, 51, 52, 56, 58, 59, 66, or 68. Cases that tested positive only for low-risk HPVs 6, 11, or 30 or that were HPV negative were scored as negative for high-risk HPVs.

Statistical Analysis

Descriptive statistics were used to evaluate diagnostic test performance. To compare diagnostic tests, McNemar's test was used to evaluate the match-paired comparisons of the dual staining for p16INK4a/Ki-67 ICC and HPV test options, holding biopsy findings constant. For statistical comparisons of test performance, cytology cases that were classified as NILM, ASC-US, or LSIL were coded as negative cytology, and cases that were classified as AGC, ASC-H, or HSIL were coded as positive cytology. Cervical biopsies that were classified as negative for malignancy or CIN2/3 were coded negative, and cases that were classified as CIN2, CIN3, AIS, or carcinoma were coded as positive histology, based on consideration of current clinical practices that would prompt surgical intervention (cone biopsy and/or hysterectomy). Negative or low-risk positive HPV test results were coded as HR-HPV test negative, and the detection of 1 or more HR-HPV types was coded as positive HR-HPV test results. For each specimen, the concordance for the each of the HPV, dual staining for p16INK4a/Ki-67 ICC, and cytology tests was compared with the biopsy result to evaluate the test's “match rate” or “percent agreement.” The match rate was calculated based on the proportion of cases that were concordant, where concordance was defined as the test having the same result as the biopsy, whereas nonconcordance was defined as the test having different findings from the biopsy. As another statistical performance metric, kappa statistics were used to compare the actual agreement beyond chance (that was observed) with the potential agreement beyond chance alone (that would be expected). A kappa in the range of 0.4 to 0.6 was deemed “moderate agreement,” whereas a kappa in the range of 0.6 to 0.8 was considered “good agreement.” The null hypothesis tested was that there was no difference in either the positive or negative likelihood ratios for the dual staining for p16INK4a/Ki-67 ICC versus HR-HPV tests in comparison with the biopsy gold-standard assessment. SAS version 9.1 was used for these comparative analyses.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES

p16INK4a/Ki-67 ICC Dual-Stain Cytology

p16INK4a Immunocytochemistry resulted in brown (DAB-based) nuclear and cytoplasmic staining of scattered epithelial cells and epithelial cell clusters in slides prepared from remnant cytologically abnormal cytology specimens. Ki-67 immunocytochemistry resulted in red nuclear staining of squamous and/or glandular cells. Dual-stain-positive cells showed brown cytoplasmic staining (for p16INK4a) and dark red to red-brown or maroon nuclear staining (reflecting colocalization of both p16INK4a and Ki-67). Although p16INK4a only (brown) or Ki-67 only (red) stained cells were detected in some cases for all diagnostic categories, dual-stained cells were found only in cases with abnormal cytologic findings (Figs. 1 and 2). Dual positive p16INK4a/Ki-67 ICC staining was not observed in any NILM cases (0 of 22), was detected in a minority of ASC-US, ASC-H, and LSIL cases, and was detected in a high proportion of AGCs (55.6%) and in most HSILs (86.4%); see Table 1.

thumbnail image

Figure 1. p16INK4a and Ki-67 dual-stain immunocytochemistry, cervical cytology, squamous cells (brown, p16INK4a; red, Ki-67): (A) squamous cell positive for p16INK4a (dual-stain-negative test result); (B) metaplastic squamous cell positive for Ki-67 but negative for p16INK4a (dual-stain-negative test result); (C) ASC-US with dual-stain-positive test result (several cells stained for p16INK4a, including 1 that also expressed K-67); (D) LSIL with rare smaller cells that were dual stain positive (subsequent biopsy on this case confirmed a diagnosis of CIN2/3; (E) HSIL, dual stain positive, biopsy confirmed CIN3; (F) HSIL with intense p16 expression that limited detection of Ki-67 coexpression, biopsy confirmed CIN3.

Download figure to PowerPoint

thumbnail image

Figure 2. p16INK4a and Ki-67 dual-stain immunocytochemistry, AGC (brown, p16INK4a; red, Ki-67). All cases shown were scored dual stain positive. Biopsy results: (A) endometrioid adenocarcinoma; (B) CIN3; (C) endocervical adenocarcinoma; (D) AIS and CIN3.

Download figure to PowerPoint

Table 1. p16INK4a and p16INK4a/Ki-67 Dual-Stain Test Results
Cytologic DiagnosisHigh-Risk HPV Positiveap16/Ki-67 Dual Stain Positive (%)
  • a

    High-risk HPVs defined as types 16, 18, 31, 33, 15, 39, 45, 51, 52, 56, 58, 59, 66, and 68.

  • b

    HPV test results were noninformative in 2 of 21 ASC-US, 1 of 22 HSIL, and in 2 of 18 AGC cases.

  • c

    Dual-stain immunocytochemistry was noninformative in 1 ASC-US case.

NILM (n = 22)8/22 (34.8)0/22 (0)
ASC-US (n = 21)5/19 (26.3)b7/20 (35.0)c
ASC-H (n = 17)13/17 (76.5)8/17 (47.1)
LSIL (n = 22)10/22 (45.5)8/22 (36.4)
HSIL (n = 22)21/21 (100)b19/22 (86.4)
AGC (n = 18)4/16 (25)b10/18 (55.6)
Total (n = 122)61/117 (52.1)b52/122 (42.6)

HR-HPV Typing

The internal control aminolevulinate synthase 1 DNA target was identified in 117 of 117 tested specimens (100%); see Figure 3. HR-HPVs were detected in a high proportion of cases with NILM cytology (34.8%), and positive test results ranged from 25% in AGC to 100% in HSIL cases (Table 1). High-risk HPV types 16 and/or 18 were identified in 33 cases (28.7%), including 4 of 22 NILM, 2 of 21 ASC-US, 6 of 17 ASC-H, 2 of 22 LSILs, 15 of 22 HSILs, and 4 of 18 AGC. Other high-risk types were identified in 28 cases (24.3%). A single HR-HPV type was identified in 57 cases (49.6%), and coinfection by multiple high-risk HPV types was seen in 4 (3.5%) of a total of 61 high-risk positive cases. The most commonly detected HR-HPV types in decreasing order of prevalence were 16, 18, 66, 58, 31, and 39. Test results were concordant in 17 of 19 ASC-US cases that were processed by both PCR and HC2; 2 cases tested positive by HC2 that were negative by PCR, but no cases tested positive by PCR that were not also HC2 positive.

thumbnail image

Figure 3. HR-HPV detection and genotyping by multiplex PCR and capillary gel electrophoresis. The 100-bp internal positive control band represents aminolevulinate delta synthase 1 (W, water negative control; N, a cytology sample negative for human papillomavirus [HPV]; arrow, internal control). The HPV types were determined by the specific base pair size generated for each column.

Download figure to PowerPoint

Relative Performance of Cytology, p16INK4a/Ki-67 Dual Stain, and HR-HPV for Detection of CIN2+

Statistical comparisons of test performance were limited to the subgroup of highly selected cases for which there were biopsy data. NILM cases were included but were assumed to have benign cervical mucosa because they were selected from a normal screening population. Cases with other cytologic test results, however, were excluded from statistical comparison if they did not have biopsy results. Thus, from the initial set of 122 cases, 53 cases with abnormal cytology were excluded from the analysis of test performance because of the lack of biopsy data, and 4 cases were excluded because HPV testing was noninformative, leaving 65 cases with matched paired sets for comparison of HR-HPV, p16INK4a/Ki-67 dual stain, and cytology results with final clinical/histologic diagnosis.

In this limited series of cases with biopsy correlation, dual staining for p16INK4a/Ki67 series was detected in 13 of 16 CIN2/3s, in 1 of 1 AISs, and in 5 of 5 adenocarcinomas, including 1 endocervical and 4 endometrial adenocarcinomas that were diagnosed by endocervical and endometrial biopsies, respectively (Table 2). Dual staining for p16INK4a/Ki67 was also detected in 13 of 16 biopsy-confirmed CIN2/3 cases, all of which were also positive for HR-HPVs. Although 2 HSIL cases had negative biopsies, the HSIL diagnosis was confirmed on review of the cytology slides, and both were positive for both dual-stain immunocytochemistry and for HR-HPVs. Thus, these likely represent false-negative biopsy results as a result of sampling error rather than false-positive test results.

Table 2. p16INK4a/Ki-67 Dual-Stain-Positive Cytology Cases: Correlation with Biopsy
Clinical/Biopsy Diagnosis (n)p16INK4a/Ki-67 Dual-Stain-Positive Test Results (%)
Cytology Diagnosis (NILM Specimens and Cases With Corresponding Biopsy)
 NILMASC-USASC-HLSILHSILAGC
  • a

    NILM cases were from a screening population and assumed to be clinically equivalent to cases with negative cervical biopsy results.

Negativea (27)0/22 (0)0/2 (0)0/3 (0)
CIN 1 (12)2/3 (66.7)1/2 (50)2/4 (50)3/3 (100)
CIN 2/3 (16)2/2 (100)2/2 (100)7/10 (100)2/2 (100)
AIS (1)1/1 (100)
Adenocarcinoma (5)5/5 (100)
Other (1)0/1 (0)

Cytology was the most sensitive test (100%), and p16INK4a/Ki-67 dual stain was the most specific test (81.8%) for underlying CIN2+ (CIN2, CIN 3, AIS, or adenocarcinoma) in this highly selected series of cases (Table 3). For biopsy-matched findings, there was 80% agreement between the findings based on the dual stain and the HR-HPV, or a kappa value of 0.6 (indicating moderate to good agreement). There was not a statistically significantly difference in the overall diagnostic test performance compared with the biopsy gold standard between the dual-stain p16INK4a/Ki-67 and HR-HPV findings. HR-HPV testing was less sensitive (76.2%) and was also less specific (55.8%) than either cytology or dual-stain immunocytochemistry for predicting the subsequent biopsy diagnosis of CIN2/S, AIS, or adenocarcinoma. When cases of endometrial adenocarcinoma (n = 4) and complex hyperplasia with atypia (n = 1) were excluded from the analysis, however, the sensitivity and specificity of HPV testing for cervical lesions ≥ CIN2 were 93.8% and 55.8%, respectively.

Table 3. Cytology, p16INK4a/Ki-67 Dual Stain, and HR-HPV Test Performance
 Sensitivity (%)Specificity (%)Positive Predictive Value (%)Negative Predictive Value (%)
  • Test performance based on review of data from NILM cases and cases with abnormal cytology and corresponding biopsy results. Positive cytology results were defined as ASC-H, AGC, LSIL, or HSIL. Clinical disease was defined as CIN2, CIN3, AIS, or adenocarcinoma.

  • a

    Sensitivity calculations based on HPV test performance in all categories. Excluding cases of endometrial adenocarcinoma (n = 4) and endometrial complex hyperplasia with atypia (n = 1), the sensitivity of HPV testing was 93.8%, and the specificity was 55.8% for underlying CIN2/3 lesions.

Cytology (n = 66)10054.552.4100
p16INK4a/Ki-67 dual stain (n = 66)81.881.869.290.0
HR-HPV (n = 64)76.2a55.845.9a82.8

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES

Data from numerous studies have shown that p16INK4a, a cyclin-dependent kinase inhibitor and G1/S cell-cycle checkpoint regulator, is a surrogate marker of HPV E7–mediated functional inactivation of pRb and is consistently overexpressed in CIN2/3, SCC, and AIS and in endocervical adenocarcinoma.5, 15, 16 Although p16INK4a cannot normally be expressed in cycling cells, it is paradoxically overexpressed in cells that lack functional pRb and in the context of cervical tissue, p16INK4a is a sensitive marker of premalignant and malignant lesions.17, 18 The utility of immunocytochemical staining for p16INK4a, however, is limited by sporadic expression in squamous metaplasia, nonmucinous secretory endocervical cells, tubal endometrial metaplasia, and cervical endometriosis.5, 19 Thus, previous studies have required a morphologic and quantitative assessment of p16INK4a-positive cells to arrive at a final p16INK4a test result.9, 20 These quantitative scoring systems require careful correlation of immunostaining patterns with cell morphology, and this can be challenging, particularly in immunocytochemical preparations. p16INK4a is predicted to be expressed only in cervical epithelial cells that have HPV E7–mediated pRb defects and in sporadic benign squamous and glandular cells that are not dividing. In contrast, Ki-67 is expressed in both benign and premalignant or malignant cycling cells but is not expressed in noncycling senescent cells. Therefore, the coexpression of p16INK4a and Ki-67, in the context of the cervical mucosa, was predicted to be a sensitive and specific marker of premalignant or malignant lesions of the cervical mucosa but was not expected to be found in benign noncycling cell types.

The primary purpose of this pilot study was to confirm that a novel, commercially available immunocytochemical test to colocalize p16INK4a and Ki-67 is technically feasible and practical when applied to remnant ThinPrep fluid-based cervical cytology specimens. However, this study was not intended to predict test performance in routine practice, and the spectrum of cases was not reflective of the proportion of cytologic abnormalities that would be seen in either a high-risk clinic or in a general screening population. Nevertheless, it is interesting to note that our findings are remarkably similar to those of the recent European Equivocal or Mildly Abnormal Papanicolaou Cytology Study (EEMAPS), which showed that dual p16INK4a/Ki-67 staining of ASC-US/LSIL cases was significantly more specific than HPV testing in detecting underlying biopsy-confirmed CIN2+.21 Furthermore, another recent study of Pap-negative/HPV-positive cervical cytology specimens from Germany reported high sensitivity and specificity of dual p16INK4a/Ki-67 staining for the detection of underlying CIN2+ lesions.22

In contrast to previous immunocytochemical studies that targeted only p16INK4a, the colocalization of p16INK4a plus Ki-67 did not require either correlation of stain results with cell morphology or with an assessment of the number of immunoreactive cells. Review of the stained slides also detected isolated cells or groups of cells that were immunoreactive for p16INK4a only (brown cytoplasmic and nuclear stain) or Ki-67 stain only (red nuclear stain), but these cells were usually readily distinguished from dual-stain-positive cells that had dual-brown cytoplasmic staining plus red/brown or maroon nuclear staining. Furthermore, positive test results were scored on cases that had any dual-stain-positive cells, but there was no requirement to count the number of positive cells or to correlate staining with cytomorphology. Dual-stain-positive cells were usually screened for at low power magnification (10× objective), and as a result, the scoring of most cases was relatively easy, straightforward, and rapid. It should be noted, however, that in some cases the intensity of the brown p16INK4a nuclear stain was so great that it was difficult to determine if there was a superimposed red Ki-67 signal. In these cases, however, we invariably found that search of the entire slide identified other cells that were more clearly dual stain positive. Nevertheless, users of this technology should be aware that overwhelmingly intense positive p16 test result could potentially obscure the colocalization of Ki-67 in some cases.

Although the case design of the current study was limited to the analysis of a nonrandomly selected series of cases and only a relatively small number of these had biopsy correlation, it was interesting to observe that all the NILM cases were test negative, that the great majority of HSIL cases were test positive, and that intermediate levels of dual-stain-positive results were found in other diagnostic categories. These preliminary results also suggest that the dual-stain approach may be nearly as sensitive but more specific than HR-HPV detection for underlying CIN2+ lesions, although confirmation of this finding will depend on the results of large-scale prospective clinical trials in both high-risk and low-risk screening populations. Furthermore, it is important to consider the fallibility of histologic diagnosis as a gold standard for CIN. Thus, discordance between cytology-based test results with the corresponding biopsy results should ideally be further evaluated by p16INK4a/Ki-67 staining on the biopsy specimens, but this was not done in the current study. Our study is further notable from the standpoint that there was a very high proportion of AGC cases that had underlying glandular lesions of the endometrial mucosa and that all of these cases were scored p16INK4a/Ki-67 dual stain test positive. The inclusion of the cases with endometrial glandular lesions in the current study decreased the apparent sensitivity of the HR-HPV assay because endometrial lesions are usually not related to HPV infection. Excluding cases with endometrial glandular lesions, the HR-HPV cytology assay detected 94.1% of cases with subsequent positive (CIN2/3) tissue biopsies. Although endometrial adenocarcinomas typically show sporadic rather than uniform expression of p16INK4a, the dual-stain approach detected these cases as well as most cervical mucosa primary lesions. The potential of using a dual-stain p16INK4a plus Ki-67 immunocytochemical assay as a test for endometrial carcinoma in cytologic samples, therefore, also deserves further study.

There was an unusually high proportion of high-risk HPV–positive test results in the NILM cases in the current study, including 1 case classified as HPV positive based on the detection of HPV type 66 that would not be detected by HC2 (because type 66 is not included in the HC2 probe set). The other high-risk HPV–positive NILM cases all had HPV types included in the HC2 assay. Although we could not directly determine if differences in analytic sensitivity of the PCR versus the HC2 assay could have contributed to the unusually high proportion of HPV-positive test results in the NILM group, there was close correlation of the PCR and HC2 test results in ASC-US cases. Although HPV test performance was not a primary focus of the current study and our PCR-based test methods are not intended to be directly compared with HC2 test data, these results suggest that the observed high rate of HPV detection in NILM cases is unlikely to be a result of enhanced sensitivity of the PCR-based method versus the HC2 assay. However, because the PCR-based methods that were used have not been validated for clinical HPV testing, HPV data from this small series of cases should not be directly compared with data from patients who were screened for high-risk HPVs by conventional methods.

In summary, in this pilot series of remnant cytology cases, dual immunocytochemical staining for both p16INK4a and Ki-67 dramatically increased the specificity and maintained high-level sensitivity for diagnosis of HSIL and other high-grade squamous and glandular lesions when compared with cytology or the detection of HR-HPV typing alone. The p16INK4a and Ki-67 dual-staining assay also provides for interpretation that is independent of morphologic or quantitative criteria, which should enable decreased interobserver variability Additional prospective clinical studies are still needed to determine the clinical test performance of this method in large-scale high-risk and screening patient populations.

Note Added in Proof

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES

FUNDING SOURCES

This study was support by academic enrichment funds of the Department of Pathology at Stony Brook University Medical Center.

CONFLICT OF INTEREST DISCLOSURES

Dr. Kenneth R. Shroyer has served as a member of the MTM Laboratories Scientific Advisory review board.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Note Added in Proof
  7. REFERENCES