Differential diagnosis of high-grade squamous intraepithelial lesions and benign atrophy in older women using p16 immunocytochemistry

1 Introduction

The incidence of cervical cancer in China is increasing rapidly, in contrast to decreasing trends in many Western countries.^1–3 Population-wide cervical cancer screening, which is effective in reducing cervical cancer incidence and mortality, has only been available since 2009, with a lower population coverage than other countries.^3,4 Liquid-based cytology (LBC) and testing for high-risk HPV (HR-HPV) are the most common cervical screening tools. Routine cytology has a high specificity but a lower sensitivity. Testing for high-risk HPV, on the other hand, has a high sensitivity but a low specificity, which can lead to unnecessary colposcopy referral and treatment, increases health costs and causes anxiety for women involved.⁵ Cervical cancer screening could be improved by developing biomarkers of higher specificity and sensitivity for detection of high-grade squamous intraepithelial lesions (HSIL).

Atrophic cytology, which is common in Pap smears from postmenopausal and perimenopausal women, presents a diagnostic challenge.^6–9 Atrophy can mimic HSIL due to cells lacking maturity and with a high nuclear to cytoplasmic ratio and can lead to an over-interpretation of dysplasia. However, sometimes a high grade or malignant smear might be under-diagnosed as atrophy.⁶ Since many women in China are older at their first Pap test than in other countries, atrophy, which may coexist with dysplasia or neoplasia, makes diagnosis particularly challenging. Short-term estrogen treatment, followed by a repeat of the Pap smear can correct false-positive interpretations,¹⁰ but it would be preferable to have biomarkers that can increase the specificity of the initial Pap smear.¹¹

One useful biomarker for diagnosis of HSIL is p16^INK4a (henceforth p16).^12–14 p16 immunohistochemistry is widely used to facilitate the diagnosis of HPV-associated lesions.¹² p16 is a cyclin dependent kinase inhibitor that regulates the cell cycle. It is expressed at very low concentrations in normal cells, but is strongly overexpressed when RB is inactivated by the HPV E7 oncoprotein15. There have been a number of studies evaluating the use of p16 immunocytology in Pap smears,^3,16–22 but few that focus on cases with atrophy.²³ Therefore, our objective was to evaluate the use of p16 immuno-cytology in identification of high-grade dysplasia vs. benign atrophy.

2 Materials and methods

3 Results

The characteristics of the specimens studied are summarized in Table 1. A total of 163 patients were examined, 94 from Wuhan and 69 from Shenzhen, and 116 showed atrophy. Women with atrophy were older than the population screened (51.9 vs. 43.4 years average). There were 116 cases positive for HR-HPV, with those from Wuhan showing a higher proportion positive than the cases from Shenzhen (84% vs. 52%). On routine liquid-based cytology, 103 cases were <ASCUS (NILM or koilocytic), 16 cases were ASCUS, and 44 cases were greater than ASCUS, including 1 AGC, 17 LSIL, 14 ASC-H, and 12 HSIL. On immunocytochemical analysis, 33 cases (20%) were positive for p16. Histological diagnoses included 21 cases with high-grade dysplasia (6 CIN2 and 15 CIN3), of which 13 were from cases with atrophy.

Examples of p16 immunocytology and the corresponding biopsies are shown in Fig. 1. The first is a woman of 51 years old. Her Pap smear diagnosis by routine liquid-based cytology was NILM and atrophy, but p16 immunocytology revealed a cluster of cells showing strong nuclear and cytoplasmic staining of p16 (Fig. 1A). The biopsy result was CIN3 involving endocervical glands (Fig. 1B). The second case is a 49 years old woman, whose Pap smear diagnosis was ASCUS and atrophy. Immunocytology showed some cells with nuclear and cytoplasmic p16 staining (Fig. 1C). The corresponding biopsy finding was CIN2 (Fig. 1D).

The number of cytological specimens, with and without atrophy, positive for HR-HPV, for p16, and for cytology in cases with a histological diagnosis ≤ CIN1 (i.e., false positives) and cases diagnosed as ≥ CIN2 (i.e., true positives) are detailed in Table 2. In the presence or absence of atrophy, both HR-HPV and p16 were positive for a large proportion (89–100%) of high-grade dysplasia cases. For cases without atrophy, most high-grade dysplasia cases (≥CIN2) were diagnosed as HSIL or ASC-H on cytology. For cases with atrophy, however, very few were ≥LSIL. HR-HPV testing had a high proportion of false-positives both for atrophic (67%) and non-atrophic (69%) specimens. For cases with or without atrophy, p16 immunocytochemistry had only 5%–12% false positives, and routine cytology (using a cutoff of HSIL/ASC-H) also had a low (5–13%) proportion of false-positives.

Using a histological diagnosis of high-grade dysplasia (CIN2/CIN3) as end-point, we examined the sensitivity (Fig. 2A) and specificity (Fig. 2B) of HR-HPV vs. p16 immunocytology vs. routine liquid-based cytology in atrophic cases (solid bars) or cases without atrophy (open bars). High-risk HPV had a high sensitivity for both the atrophic cases (92%) and those with no atrophy (100%). However, HR-HPV had a low specificity for detection of high-grade dyslasia for cases with (33%) or without (31%) atrophy. Routine liquid-based cytology, counting HSIL/ASC-H as positive, had a 67% sensitivity for case without atrophy, and a significantly lower sensitivity (17%) for atrophic cases. Unlike routine cytology, the sensitivity of p16 immunocytology was unaffected by the presence of atrophy. p16 immunocytology also had a high specificity for both atrophic cases and for cases without atrophy, significantly higher than HR-HPV testing. These results demonstrate that routine cytology has a low sensitivity, especially in women with atrophy, while HR-HPV testing has a high sensitivity and low specificity but p16 staining has a high sensitivity and a high specificity for cases with or without atrophy.

Combinations of the three tests examined do not perform better than p16 immunocytology alone (Table 3) and can, in some combinations significantly decrease sensitivity and negative predictive value. For all samples (with and without atrophy), examined as a single group, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of p16 immuno-cytology are all better than or equal to HR-HPV testing or routine cytology. Combining p16 immunocytology with HPV testing and/or routine cytology had no added benefit, as compared to p16 staining alone.

4 Discussion

We find that p16 immunocytology compares favorably with routine cytology and HPV testing in the differential diagnosis of HSIL and benign atrophy. It is more sensitive than routine liquid-based cytology, particularly for atrophic specimens, and is more specific than HR-HPV testing. p16 immunocytology may decrease the need for colposcopy referrals and could be a useful tool for early detection of cervical cancer in screening peri- and post-menopausal women, who are more likely to have HSIL coexisting with atrophy.

Since wide-spread Pap screening was begun in China only in 2009, there is a large proportion of older women at first screening. Pap smears from post-menopausal women are often atrophic, which presents a diagnostic challenge.^8,10 It is difficult to appreciate HSIL in a background of atrophy because of the lack of maturation of squamous cells and similarities between atrophic cells and dysplastic cells. Highly atypical squamous cells with a high nucleus to cytoplasmic ration and a degenerative chromatin pattern that are often observed in older patients with few risk factors can elicit morphological concern for HSIL.²⁵ Thus, there is a particular need for improvements in differential diagnosis of benign atrophy and high-grade dysplasia in programs screening older women.

Because neither routine liquid-based cytology nor HR-HPV testing is both sensitive and specific, there is a need for a combination of tests in screening programs. HR-HPV testing is increasingly being used as a high-sensitivity, low specificity primary screening tool for prevention of cervical cancer, necessitating a triage strategy for positive cases. Besides cytology, 16/18 genotyping and evaluation of viral load have been evaluated as triage strategies.^26,27 In agreement with our results, p16 immunocytochemistry has been suggested as a useful biomarker for patients with HPV infection.²⁸

There are several limitations of the present study. First, the cases studied may not represent the general screening population. The selection of cases that were referred for colposcopy was based on known HR-HPV status and/or liquid-based cytology. Thus, women negative for HR-HPV with normal no atypia are not represented. Second, it is assumed, because of the advanced age of the women in this study, that the bulk of the cases with atrophy are from post-menopausal or peri-menopausal women. However, clinical histories confirming post-menopausal status is not available. A few atrophic cases could be due to post-partum status or to contraceptive use. Finally, because a simplified scoring system is used in this study, our results may not be directly compared to previous screening studies that have used morphology-based scoring.^{3,16,17,20,29}

Immunohistochemical staining of cervical biopsy specimens is in widespread use for aiding difficult diagnoses.¹² There have also been several studies that show p16 immunocytochemistry or dual staining for p16/Ki67 can aid in cytologic diagnoses.^3,16–19 Our results agree with previous studies showing the usefulness of p16 immunocytology in cervical screening studies.^21,22

In contrast to routine liquid-based cytology, extensive expertise in cell morphology is not required for interpretation of p16 immunostained cytological specimens. Thus, it may be useful in low-resource environments and in primary hospitals. Moreover, p16 cytology can be performed at a high-throughput with a low cost and may even be adaptable to machine-based reading, suggesting it may be more useful than routine liquid-based cytology in triage of HR-HPV positive cases in large-scale cervical screening programs.

Declaration of competing interest

The authors declare that they have no conflict of interest.