Abstract
Aim
To evaluate the feasibility and value of a modified Papanicolaou counterstain for p16INK4a immunostaining in liquid‐based cervicovaginal samples.
Methods
Immunocytochemical analyses were carried out with p16INK4a and modified Papanicolaou counterstain on 81 liquid‐based samples, including 23 of within normal limits (WNL), 6 of low‐grade squamous intraepithelial lesion (LSIL), 20 of high‐grade squamous intraepithelial lesion (HSIL), 16 of atypical squamous cells of undetermined significance (ASC‐US) and 16 of atypical squamous cells, high‐grade lesion cannot be excluded (ASC‐H). Results were compared with histological or cytological follow‐up. For comparison, samples from 29 more cases (10 of LSIL, 10 of ASC‐H and 9 of HSIL) were immunostained with p16INK4a and conventionally counterstained with haematoxylin. The intensity of immunostaining in cases of squamous intraepithelial lesion (SIL) was assessed using a 0–3 scoring system. Interobserver agreement was calculated by κ statistics.
Results
Expression of p16INK4a was detected in 3 of 23 cases of WNL, 4 of 6 cases of LSIL, all cases of HSIL, 5 of 16 cases of ASC‐US and 13 of 16 cases of ASC‐H. Excluding two cases with no residual dysplastic cells in the immunocytochemistry, all cases of cervical intraepithelial neoplasia (CIN)2 or CIN3 at follow‐up expressed p16INK4a and none of the p16INK4a‐negative cases showed a high‐grade lesion at follow‐up. No evident differences in pattern or intensity of p16INK4a expression were observed between the specimens of the study and control groups. Interobserver agreement was significantly better in the study group than in the group with conventional immunostaining (combined κ 0.773 v 0.549; p<0.05), and still better, albeit statistically not significant, than with conventional immunostaining and cervical smear test together (combined κ 0.773 v 0.642).
Conclusion
Immunocytochemistry with p16INK4a and modified Papanicolaou counterstain may add to the cervicovaginal cytology the full potentiality of p16INK4a without the need of a further slide and the risk of loss of dysplastic cells, yet maintaining the typical morphological features of the smear test.
The value of p16INK4a as a diagnostic marker for cervical dysplasia and carcinomas of the cervix uteri has already been shown. On histological specimens, a diffuse p16INK4a staining is immunohistochemically detected in almost all high‐grade precanceroses and carcinomas of squamous and glandular epithelia of the cervix.1,2,3,4,5 Several studies have already highlighted the possibility of carrying out p16INK4a staining also in liquid‐based cytology samples.6,7,8,9,10,11,12 These samples, however, are typically analysed for immunocytochemistry on a subsequent slide that, although prepared from the same sample, always shows cells different from the original slide. As some non‐neoplastic epithelia may also be stained with p16INK4a,6,7,12 the interpretation of immunocytochemical results requires a careful evaluation of the morphology of the stained cells. Unfortunately, the usual counterstain with haematoxylin produces a poor chromatic contrast, which may result in difficult evaluation of the cytological features. This may be particularly critical in cases of immature metaplasia or of a cytopathic effect induced by human papillomavirus (HPV), in which the chromatic features of the cytoplasm are often a key to distinguishing dysplastic changes from reactive ones. The interpretation of conventional immunostainings may be particularly troublesome for cytologists who are not used to the evaluation of immunohistochemical stains. This may lead to a poor acceptance of new immunocytochemical techniques. Thus, an alternative counterstain may be useful in improving the interpretation of immunostained specimens. However, immunocytochemical analysis with p16INK4a is a sensitive technique in which pre‐treatment and post‐treatment have critical roles with regard to the final result.
In this study, we evaluated the feasibility and usefulness of a modified Papanicolaou counterstain for p16INK4a immunostaining on liquid‐based cervicovaginal cytological samples.
Materials and methods
Immunocytochemical analysis with p16INK4a and modified Papanicolaou counterstaining was carried out on 81 liquid‐based cervicovaginal cytological samples, including 23 of within normal limits (WNL), 6 of low‐grade squamous intraepithelial lesion (LSIL), 20 of high‐grade squamous intraepithelial lesion (HSIL), 16 of atypical squamous cells of undetermined significance (ASC‐US) and 16 of atypical squamous cells, high‐grade lesion cannot be excluded (ASC‐H). For comparison, 29 ThinPrep samples (10 of LSIL, 10 of ASC‐H and 9 of HSIL) were stained with p16 and routinely counterstained with haematoxylin. Histological or cytological follow‐up was available in all cases.
All ThinPrep slides for immunocytochemical analysis were processed with the TP 3000 processor (Cytic Corporation, Boxborough, Massachusetts, USA), according to the manufacturer's instructions.
For the immunocytochemical staining, slides were first fixed with Merckofix spray fixative (Merck, Darmstadt, Germany) and then rinsed in 50% ethanol for 30 min. Immunostaining with p16INK4a was carried out using the CINtec p16INK4a Cytology Kit (clone E6H4, Dako Cytomation, Glostrup, Denmark). Antigen retrieval was first carried out for 40 min at 95–99°C in a water bath. After blocking the endogenous peroxidase activity, the slides were incubated with the primary antibody for 30 min. Secondary antibody and chromogen (diaminobenzidine) were used according to the instructions on the kit. After immunostaining, the slides from the study group were rinsed in distilled water for 90 s and then stained with Harris haematoxylin (Papanicolaou solution 1a, Merck) for 30 s. After further rinsing in distilled water (30 s), a watery solution of 0.05% hydrochloric acid (HCl; 30 s) and again in distilled water (30 s), the slides were dehydrated in ethanol at 80°C and 90°C, and then stained with Orange II solution (Papanicolaou solution 2b, Merck) for 1 min. After further rinsing twice in ethanol at 95°C for 30 s, slides were finally stained with polychromatic solution EA50 (Papanicolaou solution 3b, Merck) for 90 s, rinsed in ethanol and xylol, and mounted with entellan new (Merck).
Specimens from the control group were conventionally counterstained with haematoxylin (ChemMate Hematoxylin, Dako Cytomation) for 2 min.
The intensity of immunostaining for cases of SIL was assessed using a 0–3 score system: 0, no stain; 1, weak; 2, moderate; and 3, strong. For assessing interobserver agreement, three cytologists (GM, EM and CM) first received a set containing immunostainings with the modified Papanicolaou counterstain. Each of them independently reviewed and classified the samples according to the Bethesda system. Subsequently, they received the control immunostainings with the conventional haematoxylin counterstain. Finally, interobserver agreement in the control group was also analysed with the conventional immunostaining together with the respective liquid‐based cytology. To avoid a statistical bias due to the different number of samples in the two groups, only 6 cases of LSIL, 10 cases of ASC‐H and 9 cases of HSIL were randomly selected from each group. Interobserver agreement was assessed by combined κ (Fleiss–Nee–Landis test).13 The null hypothesis of no difference between the combined κ values of the two groups was tested by a χ21 statistic.14
Results
In the study group, p16INK4a expression was observed in all cases of HSIL, 4 of 6 cases of LSIL, 3 of 23 cases of WNL, 5 of 16 cases of ASC‐US and 13 of 16 cases of ASC‐H. The mean staining intensity was 2.3 and 1.2 for cases of HSIL and LSIL, respectively (table 1).
Table 1 Expression of p16INK4a and staining intensity in samples from the study group.
| Cytological diagnosis | Total cases (n) | p16INK4a positive, n (%) | Mean staining intensity of positive SIL cases |
|---|---|---|---|
| WNL | 23 | 3 (13) | NA |
| ASC‐US | 16 | 5 (31.3) | NA |
| ASC‐H | 16 | 13 (81.3) | NA |
| LSIL | 6 | 4 (66.6) | 1.2 |
| HSIL | 20 | 20 (100) | 2.3 |
ASC‐H, abnormal squamous cells, high‐grade lesion cannot be excluded; ASC‐US, abnormal squamous cells of undetermined significance; HSIL, high‐grade squamous intraepithelial lesion; LSIL, low‐grade squamous intraepithelial lesion; NA, not assessed; SIL, squamous intraepithelial lesion; WNL, within normal limits.
In samples from the control group that were counterstained conventionally with haematoxylin, p16INK4a was expressed in 7 of 10 cases of LSIL, 9 of 10 cases of ASC‐H and 9 of 9 cases of HSIL. The mean staining intensity was 2.1 and 1.2 for cases of HSIL and LSIL, respectively (table 2).
Table 2 Expression of p16INK4a and staining intensity in the control group.
| Cytological diagnosis | Total cases (n) | p16INK4a positive, n (%) | Mean staining intensity of positive cases of SIL |
|---|---|---|---|
| ASC‐H | 10 | 9 (90) | NA |
| LSIL | 10 | 7 (70) | 1.2 |
| HSIL | 9 | 9 (100) | 2.1 |
ASC‐H, atypical squamous cells, high‐grade lesion cannot be excluded; HSIL, high‐grade squamous intraepithelial lesion; LSIL, low‐grade squamous intraepithelial lesion, NA, not assessed; SIL, squamous intraepithelial lesion; WNL, within normal limits.
In the study group, all cases of HSIL had a histological follow‐up of CIN3 and all cases of LSIL had either a cytological or a histological follow‐up of ASC‐US (n = 2) or LSIL or CIN1 (n = 4). Follow‐up was negative in eight cases of ASC‐US, two of which expressed p16INK4a. Of the remaining eight cases of ASC‐US, three had a CIN3 and five had an LSIL or CIN1 in the follow‐up. Overall, two of the cases of CIN3 and three of the cases of LSIL or CIN1 at follow‐up were p16INK4a negative. At revision, no residual abnormal cells could be recognised in the two cases of CIN3 with negative p16INK4a immunostaining. Four cases of ASC‐H had a histological follow‐up of CIN2 or CIN3. All of them expressed p16INK4a, as well as the four cases with a cytological follow‐up of ASC‐US or ASC‐H. Of the eight cases of ASC‐H with negative follow‐up, five were p16INK4a positive (table 3).
Table 3 Expression of p16INK4a and follow‐up in the study group.
| Cytological diagnosis | Total cases (n) | p16INK4a positive (n) | Follow‐up | P16INK4a negative (n) | Follow‐up |
|---|---|---|---|---|---|
| WNL | 23 | 3 | 3 WNL* | 20 | 20 WNL* |
| ASC‐US | 16 | 5 | 2 WNL* | 11 | 5 WNL* |
| 2 LSIL* | 1 LSIL* | ||||
| 1 CIN3 | 1 CIN0 | ||||
| 2 CIN1 | |||||
| 2 CIN3† | |||||
| ASC‐H | 16 | 13 | 2 WNL* | 3 | 1 WNL* |
| 1 ASC‐US* | 2 CIN0 | ||||
| 3 ASC‐H* | |||||
| 3 CIN0 | |||||
| 2 CIN2 | |||||
| 2 CIN3 | |||||
| LSIL | 6 | 4 | 1 ASC‐US* | 2 | 1 ASC‐US* |
| 2 LSIL* | 1 CIN1 | ||||
| 1 CIN1 | |||||
| HSIL | 20 | 20 | 20 CIN3 | 0 | – |
ASC‐H, abnormal squamous cells, high‐grade lesion cannot be excluded; ASC‐US, abnormal squamous cells of undetermined significance; CIN, cervical intraepithelial neoplasia; HSIL, high‐grade squamous intraepithelial lesion; LSIL, low‐grade squamous intraepithelial lesion; WNL, within normal limits.
*Cytological follow‐up.
†No residual abnormal cells in the immunocytochemistry samples.
In both groups, dysplastic cells showed a nuclear or nuclear and cytoplasmic p16INK4a staining (fig 1). Most non‐dysplastic cells that expressed p16INK4a were metaplastic or endocervical cells. The overall concordance of study group, conventional immunostaining alone and conventional immunostaining together with the respective liquid‐based cytology was 80.0%, 58.6% and 65.5%, respectively. The statistical analysis confirmed that interobserver agreement was significantly better in the study group than in the control group with immunostaining alone (κ 0.77 v 0.55; p = 0.0237), whereas for conventional immunostaining and the respective liquid‐based cytology together, the interobserver agreement showed an intermediate value (κ 0.64, not significant). Concordance was higher in cases of SIL than in those of ASC in all groups (table 4).
Figure 1 Immunocytochemical analyses with p16INK4a on ThinPrep specimens with the modified Papanicolaou (A–F) and conventional haematoxylin (G,H) counterstain. (A) Low magnification, showing the overall chromatic features of the modified counterstain. (B) High‐grade squamous intraepithelial lesion (HSIL). (C) HSIL with one p16INK4anegative metaplastic cell. The typical cytoplasmic features in metaplasia are preserved. (D) Low‐grade squamous intraepithelial lesion (LSIL), p16INK4anegative koilocytes. (E) LSIL. (F) Abnormal squamous cells, high‐grade lesion cannot be excluded (ASC‐H); this case becomes one of cervical intraepithelial neoplasia (CIN)3 in the follow‐up. (G) HSIL, low magnification, showing the overall chromatic features of the conventional counterstain. (H) LSIL, p16INK4anegative koilocytes.
Table 4 Analysis of concordance between three observers in the study and control groups.
| Diagnosis | Modified counterstain | Conventional counterstain | Conventional counterstain and smear test | ||||||
|---|---|---|---|---|---|---|---|---|---|
| κ* | 95% CI | p Value† | κ* | 95% CI | p Value† | κ* | 95% CI | p Value† | |
| WNL | 0.842 | 0.616 to 1.069 | <0.001 | 0.435 | 0.225 to 0.645 | <0.001 | 0.284 | 0.127 to 0.440 | 0.004 |
| ASC‐US | 0.306 | 0.079 to 0.532 | 0.004 | 0.174 | −0.036 to 0.384 | 0.052 | 0.380 | 0.224 to 0.537 | <0.001 |
| ASC‐H | −0.042 | −0.268 to 0.185 | 0.641 | −0.088 | −0.298 to 0.123 | 0.793 | −0.061 | −0.218 to 0.096 | 0.715 |
| LSIL | 0.781 | 0.554 to 1.007 | <0.001 | 0.731 | 0.521 to 0.942 | <0.001 | 0.839 | 0.682 to 0.995 | <0.001 |
| HSIL | 0.945 | 0.719 to 1.171 | <0.001 | 0.761 | 0.551 to 0.971 | <0.001 | 0.815 | 0.658 to 0.971 | <0.001 |
| Combined‡ | 0.773 | 0.622 to 0.923 | <0.001 | 0.549 | 0.427 to 0.671 | <0.001 | 0.642 | 0.544 to 0.739 | <0.001 |
ASC‐H, abnormal squamous cells, high‐grade lesion cannot be excluded; ASC‐US, abnormal squamous cells of undetermined significance; HSIL, high‐grade squamous intraepithelial lesion; LSIL, low‐grade squamous intraepithelial lesion; WNL, within normal limits.
*Landis–Koch extension.
†p Value for the test z (for κ = 0).
‡Combined κ (Fleiss–Nee–Landis test).
Discussion
The low sensitivity of conventional smear test and the frequent occurrence of indeterminate cell changes in cervicovaginal samples is still an unresolved problem. Markers of dysplasia may therefore be very interesting, potentially permitting an increase in sensitivity and specificity of cytological screening. Among the various proposed ancillary tests, p16INK4a staining is one of the most promising. The specificity of p16INK4a for cervical lesions is a consequence of the oncogenic mechanism of HPV. The oncogenic influence of HPV depends on the interaction of viral gene products, particularly E6 and E7, with specific host proteins. Whereas E6 inactivates p53, E7 binds and thus inactivates the tumour suppressor retinoblastoma protein. As the transcription of the cyclin‐dependent kinase inhibitor p16INK4a underlies a negative feedback control by the retinoblastoma protein, its inactivation due to HPV results in an overexpression of p16INK4a in replication‐competent epithelial cells.1 Several studies have already described the possibility of p16INK4a staining in liquid‐based cytology samples.6,7,8,9,10,11,12 Although p16INK4a immunocytochemistry is characterised by a high sensitivity for high‐grade lesions, some cases of WNL may show a positive staining in non‐dysplastic cells.6,7,12 Thus, the interpretation of immunocytochemical results requires a careful evaluation of the morphology of the stained cells. Unfortunately, the usual counterstain with haematoxylin produces a poor chromatic contrast that may result in difficulty in evaluating the cytological features.
The modified Papanicolaou counterstain that we propose in this study can be helpful in resolving this issue. To enable the detection of even faint immunostains and at the same time warrant good cytological details, we modified the conventional Papanicolaou stain and used it as a counterstain. As shown in fig 1, although the modified counterstain is slightly weaker than a conventional Papanicolaou stain, the usual chromatic pattern is conserved. This is particularly important in cases of immature metaplasia or an HPV‐induced cytopathic effect, in which the chromatic features of the cytoplasm are often a key to distinguishing dysplastic changes from reactive ones. Overall, diagnostic concordance was higher in the study group than in the control group, which used conventional counterstaining (80.0% v 58.6%), with a significantly higher interobserver agreement in the study group (combined κ 0.773 v 0.549; p<0.05). Not surprisingly, as the observers share the same extensive experience and common diagnostic criteria with respect to the use of Papanicolaou stains, adding the respective liquid‐based cytology to conventional immunostaining partially increased the interobserver agreement. In fact, concordance with conventional immunostaining and the respective liquid‐based cytology together showed an intermediate value (κ 0.64) that was still lower than that in the study group, even if not significant.
The relatively low overall concordance may be owing to the high percentage of cases of ASC‐H in our study. Indeed, not surprisingly, in each group, concordance was considerably better in cases of SIL than in those of ASC. Furthermore, the three observers who evaluated the slides for the agreement analysis had little experience with immunostainings. An improvement in interpretation performance may occur over time as the cytologists gain more experienced with p16INK4a.
Although, theoretically, a weak cytoplasmic immunostaining may be masked by the modified counterstain, we found no meaningful difference in p16INK4a staining pattern and intensity between study and control groups. The mean staining intensity for cases of HSIL was 2.3 and 2.1 in the study and control groups, respectively, and 1.2 for LSIL in both groups. As in our study the overall p16INK4a results agree with those of previous studies,6,7,8,9,10,11,12 we conclude that the modified Papanicolaou counterstain does not mask or interfere with the p16INK4a immunostaining. All cases of HSIL were stained with p16INK4a, and even those with relatively weak immunocytochemical expression could be readily recognised. Considering cases of HSIL and ASC together, all but two cases of a CIN2 or CIN3 histological follow‐up were positive for p16INK4a and only two cases with negative p16INK4a expression showed a high‐grade lesion in the follow‐up. At revision, in the two cases of CIN3 with negative p16INK4a immunostaining, no residual dysplastic cells were present. Because subsequent slides always show cells different from those in the original one, the loss of dysplastic cells in immunocytochemical samples may be possible. In particular, cases with a small number of abnormal cells in the original slide may be at risk of losing diagnostic cells. Combining the Papanicolaou stain with immunocytology may avoid this risk, making a further slide unnecessary. This may be particularly interesting in the follow‐up of cases with previous abnormal cytology, in which the high probability of dysplastic cells may make such an alternative approach attractive. Also, the results in cases of LSIL and WNL are in accordance with those from previous studies.6,7,8,9,10,11,12 In the present study, p16INK4a was expressed in 4 of 6 cases of LSIL and 3 of 23 cases of WNL. In cases of LSIL with negative immunostaining, the features of the HPV‐induced cytopathic effect were obvious (fig 1D), particularly in comparison with the conventionally counterstained control group (fig 1H). In cases of WNL, the modified counterstain facilitated the differentiation between dysplastic and metaplastic cells, in which the typical features of the cytoplasm were conserved (fig 1C).
Take‐home messages
The modified Papanicolaou counterstain improves the interobserver agreement of p16INK4a immunocytochemistry on liquid‐based cervicovaginal samples.
Immunocytochemistry with the modified Papanicolaou counterstain may add the full potentiality of p16INK4a without the need of a second slide and the risk of loss of dysplastic cells, yet maintaining the typical morphological features of the test.
In conclusion, the modified Papanicolaou counterstain can help in interpreting p16INK4a immunostainings. This technique may be an alternative approach particularly in high‐risk cases, for instance in the follow‐up of cases of previous abnormal cytology. Immunocytochemistry with p16INK4a and modified Papanicolaou counterstain can replace the routine smear in these cases, adding the full potentiality of p16INK4a without the need of a second slide or the risk of affecting the morphological features of the sample.
Abbreviations
ASC‐H - atypical squamous cells, high‐grade lesion cannot be excluded
ASC‐US - atypical squamous cells of undetermined significance
CIN - cervical intraepithelial neoplasia
HPV - human papillomavirus
HSIL - high‐grade squamous intraepithelial lesion
LSIL - low‐grade squamous intraepithelial lesion
SIL - squamous intraepithelial lesion
WNL - within normal limits
Footnotes
Competing interests: None.
References
- 1.Klaes R, Friedrich T, Spitkowsky D.et al Overexpression of p16INK4a as a specific marker for dysplastic and neoplastic epithelial cells of the cervix uteri. Int J Cancer 200192276–284. [DOI] [PubMed] [Google Scholar]
- 2.Sano T, Oyama T, Kashiwabara K.et al Immunohistochemical overexpression of p16 protein associated with intact retinoblastoma protein expression in cervical cancer and cervical intraepithelial neoplasia. Pathol Int 199848580–585. [DOI] [PubMed] [Google Scholar]
- 3.Negri G, Egarter‐Vigl E, Kasal A.et al p16INK4a is a useful marker for the diagnosis of adenocarcinoma of the cervix uteri and its precursors. An immunohistochemical study with immunocytochemical correlations. Am J Surg Pathol 200327187–193. [DOI] [PubMed] [Google Scholar]
- 4.Agoff N S, Lin P, Morihara J.et al p16INK4a expression correlates with degree of cervical neoplasia: a comparison with Ki‐67 expression and detection of high‐risk HPV types. Mod Pathol 200316665–673. [DOI] [PubMed] [Google Scholar]
- 5.Tringler B, Gup C J, Singh M.et al Evaluation of p16INK4a and pRb expression in cervical squamous and glandular neoplasia. Hum Pathol 200435689–696. [DOI] [PubMed] [Google Scholar]
- 6.Bose S, Evans H, Lantzy L.et al p16(INK4A) is a surrogate biomarker for a subset of human papilloma virus‐associated dysplasias of the uterine cervix as determined on the Pap smear. Diagn Cytopathol 20053221–24. [DOI] [PubMed] [Google Scholar]
- 7.Yoshida T, Fukuda T, Sano T.et al 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. Cancer 2004102100–108. [DOI] [PubMed] [Google Scholar]
- 8.Nieh S, Chen S F, Chu T Y.et al Expression of p16 INK4A in Papanicolaou smears containing atypical squamous cells of undetermined significance from the uterine cervix. Gynecol Oncol 200391201–208. [DOI] [PubMed] [Google Scholar]
- 9.Pientong C, Ekalaksananan T, Swadpanich U.et al Immunocytochemical detection of p16INK4a protein in scraped cervical cells. Acta Cytol 200347616–623. [DOI] [PubMed] [Google Scholar]
- 10.Bibbo M, DeCecco J, Kovatich A J. P16INK4A as an adjunct test in liquid‐based cytology. Anal Quant Cytol Histol 2003258–11. [PubMed] [Google Scholar]
- 11.Murphy N, Ring M, Killalea A G.et al p16INK4A as a marker for cervical dyskaryosis: CIN and cGIN in cervical biopsies and ThinPrep smears. J Clin Pathol 20035656–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Trunk M J, Dallenbach‐Hellweg G, Ridder R.et al Morphologic characteristics of p16INK4a‐positive cells in cervical cytology samples. Acta Cytol 200448771–782. [DOI] [PubMed] [Google Scholar]
- 13.Fleiss J L, Nee C M, Landis J R. Large sample variance of kappa in the case of different sets of raters. Psychol Bull 197986974–977. [Google Scholar]
- 14.Fleiss J L.Statistical methods for rates and proportions. 2nd edn. New York: Wiley, 1981222