Cytohistological correlation of endocervical gland involvement with high-grade squamous intraepithelial lesions.

BACKGROUND: Diagnosis of endocervical glandular involvement by high-grade squamous intraepithelial lesion (HSIL-EGI) on Papanicolaou (Pap) smears can affect the clinical management of patients. AIM: The cytological criteria for the diagnosis of HSIL-EGI are described and the accuracy of this diagnosis was investigated.
MATERIALS AND METHODS: Seventeen patients diagnosed with HSIL-EGI and 40 patients with diagnosis of HSIL on Pap smears with follow-up cone or loop electrocautery excision procedure (LEEP) biopsies were included in the study. The following criteria were evaluated for the cytological diagnosis of HSIL-EGI: atypical cells with definite features of HSIL, three-dimensional atypical squamous cell clusters (TDCs) with attached benign endocervical epithelium, finger-like TDCs covered with intact epithelium on most sides, which represent the finger-like invaginations of the endocervical glandular area involved by HSIL, and the absence of cytological findings of in situ adenocarcinoma of the cervix.
RESULTS: On subsequent histopathological evaluation, 16 of 17 (94.1%) patients with a cytological diagnosis of HSIL-EGI and 17 of 40 (42.5%) patients with HSIL exhibited endocervical glandular involvement (P < 0.001, sensitivity: 48.5%, specificity: 95.8%, positive predictive value: 94.1%, negative predictive value: 57.5% and accuracy: 68.4%).
CONCLUSION: Diagnosis of HSIL-EGI may be possible on Pap smears with a high positive predictive value and specificity but low sensitivity, possibly due to cytological sampling limitations. To clarify the results of the present study, more extensive studies with a particular emphasis on the sampling of the endocervical glandular area for cytological evaluation of the cervix are needed.

Introduction

Very few reports have investigated the accuracy of the cytological diagnosis of endocervical glandular involvement by high-grade squamous intraepithelial lesion (HSIL-EGI) on Papanicolaou (Pap) smears. The Bethesda system does not include the diagnosis of HSIL-EGI on Pap smears. However, it is obvious that the cytological diagnosis of HSIL-EGI is very important in terms of clinical management of the patients. It has been demonstrated that there is an association between EGI and recurrent dysplasia, even independent of the margin status in loop electrocautery excision procedure (LEEP) cones.[1-3] Several authors reported an increased failure rate using cryotherapy when the endocervical glands were involved in cervical intraepithelial neoplasia (CIN).[3] The information concerning EGI on Pap smears may influence the decision to perform deeper conization, LEEP biopsy or cryotherapy to ensure negative margins.[1-8]

A few authors described the criteria for the cytological diagnosis of HSIL-EGI in conventional Pap smear and liquid-based cervical cytology samples, and they revealed the effectiveness of their criteria.[9-12] However, the results of two authors differed from those of the others; they did not demonstrate the increased frequency of EGI on subsequent cone biopsies of patients with cytological HSIL-EGI diagnosis.[1314] In our study, we describe simple criteria for the cytological diagnosis of HSIL-EGI and investigated the accuracy of this diagnosis on conventional Pap smears compared with that of subsequent LEEP/cone biopsies (SCBs).

Materials and Methods

Seventeen patients with cytological diagnoses of HSIL-EGI and 40 patients with cytological diagnoses of HSIL and satisfactory follow-up LEEP or cone biopsies (within 3 months of cytological diagnosis) were included in the study groups. All patients were identified from Umraniye Research and Education Hospital pathology computer data files over a period of 4 years (January 2007–December 2010). All cervical samples were obtained with a cytobrush, smeared on to one glass slide for each case, alcohol-fixed, and stained according to the Pap method.

All cases were new patients without a prior history of disease. For the cytological diagnosis of HSIL-EGI, the following criteria were evaluated: atypical squamous cells with definite features of HSIL,[15] three-dimensional atypical squamous cell clusters (TDCs) with attached benign endocervical epithelium [Figure 1], finger-like TDCs covered with intact epithelium on most sides, which represent the finger-like invaginations of the endocervical glandular area involved by HSIL [Figure 2], and the absence of cytological features of in situ adenocarcinoma of the cervix[15] (AIS). The 40 cases with cytological diagnosis of HSIL were reviewed retrospectively and they did not include either of the two types of TDCs.

Figure 1

Three-dimensional atypical squamous cell cluster with attached benign endocervical epithelium (Pap, ×100)

Figure 2

Finger-like three-dimensional atypical squamous cell cluster that belongs to the invagination of the endocervical glandular area (Pap, ×100)

The majority of biopsy specimens were oriented by a single suture designating the 12 o’clock position. All specimens were fixed in formaldehyde and processed in a standardized fashion. Surgical margins of the whole specimens were inked and then divided into four quadrants starting from the 12 o’clock position. All quadrants were radially sectioned, submitted clockwise, embedded in paraffin, and stained with hematoxylin and eosin. All specimens were examined by one surgical pathologist. Only the specimens that represented both ectocervical squamous and endocervical glandular tissue were included in the study. The highest degree of cervical dysplasia and EGI were recorded.

Statistical analysis

The McNemar test was used to assess the homogeneity between paired categorical variables. P values (two-sided test) less than 0.05 were considered significant. All statistical analyses were performed using Number Cruncher Statistical System 2007, Power Analysis and Sample Size 2008 statistical software (Kaysville, Utah, USA).

Results

Table 1 lists the cytological criteria of HSIL-EGI. Of these criteria mentioned above, features required for the diagnosis of HSIL-EGI were as follows: atypical squamous cells with definite features of HSIL, the absence of cytological features of AIS, at least one TDC with attached benign endocervical epithelium, and/or finger-like TDCs. On histological evaluation of SCBs, 16 of 17 (94.1%) patients with cytological diagnoses of HSIL-EGI and 17 of 40 (42.5%) patients with cytological diagnoses of HSIL exhibited EGI (P < 0.001, sensitivity: 48.5%, specificity: 95.8%, positive predictive value: 94.1%, negative predictive value: 57.5% and accuracy: 68.4%). For the patient with a cytological diagnosis of HSIL-EGI but without histological confirmation, the transformation zone was included partially in the SCB. Histologically, none of the 57 cases exhibited AIS.

Table 1

Cytological features of HSIL-EGI on Pap smears

Discussion

It has been demonstrated that EGI was a significant factor correlating with a positive endocervical margin and residual/recurrent disease in women with CIN who did not undergo hysterectomy. The reported success rates in treating dysplasia by LEEP biopsy range from 60 to 95%.[16-20] It has been demonstrated that after LEEP, EGI was predictive of residual/recurrent dysplasia.[1-3] Livacy et al.[3] found an association between EGI and recurrent dysplasia independent of margin status in LEEP cones. Patients with negative margins and positive EGI had approximately twofold higher recurrence rates than patients with negative margins and negative gland involvement. Some authors report increased failure rates in treatment with LEEP attributed to the smaller size of LEEP versus conventional cold-knife conization specimens.[2122] However, Sun et al.[23] concluded that EGI and the depth of conization were significant factors that correlated with positive cone margins. In addition, several authors report an increased failure rate using cryotherapy when the endocervical glands were involved in CIN.[3] Therefore, information concerning gland involvement should influence patient management; the option of deeper conization may have advantages over LEEP or cryotherapy to ensure negative margins for suitable patients.[1-8]

Some authors have described the criteria for the cytological diagnosis of HSIL-EGI and confirmed the effectiveness of their criteria.[9-12] However, the results of two of these authors differed from those of the others; they did not confirm the increased frequency of EGI on SCBs of patients with cytological HSIL-EGI diagnoses.[1314] Selvaggi[911] defined the cytological criteria of HSIL-EGI with a particular emphasis on patterns A and B. Round to oval “syncytial” clusters with smooth cellular borders, central whorling and peripheral flattening of small dysplastic cells were designated as “pattern A,” and sheets of abnormal columnar vacuolated cells with peripheral palisades and nuclear pseudostratification were designated as “pattern B.” We actually experienced difficulty in understanding and recognizing the cell clusters of patterns A and B. Instead, we used two more simple TDCs, which were mentioned by Drikoningen et al.[10] and Levine et al.[13] as follows: the first was a TDC with attached benign endocervical cells[10] and the second one was a finger-like TDC covered with intact epithelium on most sides, which represents the finger-like invaginations of the endocervical glandular area involved by HSIL.[13] In the present study, the required and consistent findings for the cytological diagnosis of HSIL-EGI were atypical squamous cells with definite HSIL features, at least one of two simple-type TDCs and the absence of cytological findings of AIS. As the HSIL descends into the crypt, it displaces the benign endocervical epithelium, which may be seen attached to the atypical squamous cell clusters on the cytological examination. Finger-like TDCs with smooth contours might be observed in the Pap smears when the cytobrush removes fragments of tissue including endocervical crypts and accompanying dysplastic tissue. Differential diagnosis from AIS was possible in most cases if diagnostic criteria were strictly applied. None of these TDCs showed cell polarity and/or glandular openings. Cell patterns including those of strips, cell rosettes, feathering and cigar-shaped nuclei, which are unique to AIS, were not observed. Furthermore, normal endocervical cells were predominant, while in AIS, normal endocervical cells are usually scant. When in doubt, one has to consider that AIS occurs less frequently than HSIL.[10] On histological evaluation of SCBs, 16 of 17 (94.1%) patients with cytological diagnosis of HSIL-EGI exhibited EGI. However, 17 of 40 (42.5%) patients with cytological diagnosis of HSIL exhibited EGI on histological evaluation of SCBs (P < 0.001, sensitivity: 48.5%, specificity: 95.8%, positive predictive value: 94.1%, negative predictive value: 57.5% and accuracy: 68.4%). According to our study results, the cytological diagnosis of HSIL-EGI has disadvantages of low sensitivity and negative predictive value, with a high frequency of false-negative diagnoses. We feel that the limited and difficult sampling of endocervical crypts for cytological examination was responsible for these results. All cervical samples were obtained with a cytobrush. The highest yield of endocervical cells has been found for the cytobrush among various cell collection devices.[24-27] However, during cytological sampling without special attention to deep endocervical sampling, endocervical glandular invaginations may not be available for the cytological evaluation.

In conclusion, diagnosis of HSIL-EGI may be possible on Pap smears with a high positive predictive value and specificity but with low sensitivity, possibly due to cytological sampling limitations. To clarify the results of the present study, more extensive studies with a particular emphasis on the sampling of the endocervical glandular area for cytological evaluation of the cervix are needed.