In 1988 the National Cancer Institute established «The Bethesda System», a standardized terminology to report cervical cytology. Over the years several reviews to «The Bethesda System» have been made. In 2001 the terminology of atypical glandular cells of undetermined significance (AGUS) was updated to atypical glandular cells (AGC) which was meant to describe morphologic changes in glandular cells that were felt to be beyond normal changes, but not sufficient to be diagnostic of in situ adeno carcinoma (AIS). Since then AGUS was renamed AGC and subdivided into atypical glandular cells-not otherwise specified (AGC-NOS) and atypical glandular cells-favor neoplastic (AGC-FN). This terminology chance in AGUS was motivated by the substantial risk of high grade lesions associated with AGC which for so current guidelines recommend an extensive initial evaluation of women with AGC, in cervical Pap tests, based on age and AGC subtype [1].

Since the introduction of cytological screening in the 20^th century there has been a decreasing incidence of squamous cell carcinoma but an increased relative rate of adenocarcinoma is being perceived throughout European countries, especially among younger women [2,3]. Whereas these used to account for less than 5% of all cervical cancers, they now comprise 20% to 25% [4]. One of the main reasons for the increasing relative rate of adenocarcinomas seems to be the lower Pap sensitivity in diagnosing glandular lesions and a high false negative rate [5,6].

Glandular abnormalities are found in 0.1 to 2.1% of Pap tests being more frequent in women over 40 [7,8]. The most common source of glandular cells on a Pap test is the endocervical canal followed by the endometrium, in most cases by flaking [6,9]. Cells from another location like the tube, ovary and other intraperitoneal sources can also migrated and be reported [6,9].

Following a cytological diagnose of AGC, most women, on the colposcopy-guided biopsies, show no significant pathology [5,10]. The association to premalignant and malignant disease is seen is approximately 30% of the cases [5,11-13].

Once AGC prevalence is low, a review through literature shows few data regarding the histological outcome of AGC-NOS in comparison to AGC-FN and no studies on AGC subtypes histological outcomes, among Portuguese women, were found. For so our retrospective study aimed to investigate the clinical significance of AGC and its correspondent subtypes in a group of Portuguese patients.

From January 2009 to December 2013 a total of 26662 liquid-based Pap tests (ThinPrep^®; Hologic 1996) collected with cytobrush were interpreted at the Pathology Department of our hospital, a tertiary centre in the capital of Portugal. All Pap test were classified according to the 2001 revision of Bethesda classification system.

A total of 84 from 26662 patients had AGC in the cytological interpretation. Medical records of those patients were retrospectively analysed and data was collected in what regards age, onset of sexual life, Human Papilloma Virus (HPV)-polymerase chain reaction (PCR) and histological results from cervical punch biopsy, endometrial hysteroscopy biopsy or curettage, endocervical curettage, excisional procedures of transformation zone or hysterectomy.

The HPV-PCR was preformed according to the guidelines from Portuguese Gynaecology Society which are based on the American Society of Colposcopy and Cervical Pathology guidelines [14]. Samples for HPV-PCR were collected with cytobrush and sent to virology analyses in a ThinPrep^® (Hologic 1996).

In what concerns histological results it was considered the higher grade lesion for each patient. Classification of histologic specimens from cervical biopsies was performed according to the 2001 Bethesda System: CIN1-mild dysplasia, CIN2-moderate dysplasia, CIN3-severe dysplasia/in situ carcinoma) [1].

Statistical analysis was performed using SPSS version 20.0 (IBM, New York, USA). Data were analysed as a whole and by AGC subgroups; results were determined via Pearson chi-square for categorical variables and Student´s t-test for continuous ones. Two-tailed p-value inferior to 0,05 was considered to be statistically significant.

Prevalence and characterization of AGC

Abnormalities of glandular cells had a prevalence of 0.3% (84/26662). The study group had an average age of 46.8 ± 14.5 with a range from 22 to 86 years old, 64.9% more than forty years.

The distribution within different sub-types of AGC is shown in Table 1.

Table 1: Prevalence of AGC sub-groups.

AGC-NOS and AGC-FN groups

Clinical and histological outcomes: Characterization of AGC subgroups is presented in Table 2.

Table 2: Clinical and histological outcomes of AGC-NOS and AGC-FN..

Patients with AGC-NOS were diagnosed at younger ages although the onset of sexual life was identical between groups.

The prevalence of benign cervical pathology was higher in AGCNOS, being endocervical polyps the most common histological finding (63.4%). AGC-NOS and AGC-FN relate to a disease ratio of 75% and 100%.

The subgroup AGC-FN showed an association with AIS and a higher correlation with endometrial carcinoma. Although there was no statistical significant difference between groups in what concerns squamous cervical intraepithelial neoplasia (CIN), the study evidences a tendency towards a higher prevalence of CIN in AGC-FN group. There was no evidence of cervical squamous cell carcinoma.

Correlation of HPV with histological findings: Twenty-eight HPV type-specific testing were performed in AGC-NOS pap tests and HPV-deoxyribonucleic acid (DNA) was identified in 8 cases (28.8%). Table 3 describes HPV-DNA positive finding in AGC-NOS.

Table 3: Type-specific HPV detected in AGC-NOS.

Between the 8 patients that tested positive for HPV-DNA, 4/8 (50%) showed CIN pathology in the follow-up (one case of CIN1 and three cases of CIN 3). All the AGC cytology that tested negative for HPV-DNA (n=20) had benign histological findings during follow-up study.

Coexisting squamous and glandular abnormalities

From the total of 84 AGC interpreted Pap tests, seven patients (8.3%) had dual squamous and glandular abnormalities.

The patient mean age was 44.1±9.7 years old. Follow-up histological results of these patients, presented in Table 4, evidenced lesions predominantly of squamous origin (57.1%) with three reported cases of CIN 3 and one case of squamous cell carcinoma. Glandular lesions, with AIS were detected in two cases (28.6%). There was one identified case of benign pathology and no coexisting cases of squamous and glandular pathology.

Table 4: Outcomes in coexisting squamous and glandular abnormalities.

Discussion

The AGC (including the subgroups: AGC-NOS and AGC-FN) in our population study had a low prevalence of 0.3%, being more commonly found in women over 40. In our 5 years retrospective analyse AGC are predominately associated with benign pathology (57.1%).

Our study also emphasizes the knowledge that although glandular cells originate from a variety of locations and the majority of abnormal cells derives from benign lesions (endocervical and endometrial polyp, endocervicitis, endometriosis, microglandular hyperplasia, tubal metaplasia, endosalpingiosis and Arias-Stella chances of pregnancy cervix) [6,9], the identification of this cells on cervical cytology must be vigorously investigated, to identify other significant pathology, given the fact that low grade lesions, pre-malignancy and malignancy accounted for 20.8% of all clinical reports. The histological follow-up evidenced that most of the pre-malignant cervical pathology found were squamous lesions (5.2% of CIN3 versus 1.3% of AIS). In what concerns malignancy, endometrial adenocarcinoma was the most prevalent (7.8%).

The literature shows that only 41% to 71% of AIS are diagnosed histologically through a cytological AGC report [15-18]. The neoplastic glandular cells can proliferate hidden under normal metaplastic and dysplastic squamous epithelium in 60% of the cases which makes these cells not accessible to cytological sampling [19]. Although cytological screening has a high false negative rate for adenocarcinoma for now there is not a better available screening test.

In the present study, histological examination of AGC found a total of 7.8% of malignancy which agrees with Chhieng et al. that showed a prevalence of 13.3% of malignant disease among 45 diagnosed AGUS [20]. The subgroup of AGC-FN has older women, being all above forty years old (AGC-NOS: 45.7 ± 14.6 versus AGC-FN: 57.0 ± 14.6 years, p-value=0.032) with a higher prevalence of AIS (AGC-NOS: 0% versus AGC-FN: 11.1%, p-value=0.006) and endometrial adenocarcinoma (AGC-NOS: 4.4% versus AGC-FN=33.3%, p-value=0.003). Our reported results highlight the importance of cervical study and pelvic sonography in AGC evaluation, and for so, close follow-up is recommended [14].

The relationship between cervical cancer and HPV is well known particularly in what concerns HPV type 16 and 18 which are identified in about two thirds of cervical cancer cases worldwide [21]. Current guidelines have suggested a role for HPV testing in the management of Pap test that report AGC [14]. In this study we tested 28 AGC liquid-based cytology for HPV DNA with posterior identification of 10 types of HPV DNA, 7 being high risk HPV DNA. The HPV associated disease ratio in our study was 50% which is in agreement with existing literature that reports rates which range from 11 to 53% (average 21%) [11].

In our study the sensitivity of HPV-DNA in detecting cervical lesions was 50% with a specificity of 100%, a positive predictive value (PPV) of 50% and a negative predictive value (NPV) equivalent to 100%. From the sample tested for HPV-DNA, in our study, no conclusion can be taken about the most commonly isolated HPV types. Although there are limitations, due to the retrospective design and the small sample analyses, our results confirm that HPV DNA testing has an important role in the management of AGC Pap results follow-up once HPV-DNA presence is associated with cervical disease, with a PPV of 50% in our study. Based on the evidence of HPV-DNA PPV for cervical pathology, current guidelines recommend, in the absence of significant pathology in the initial evaluation, a re-evaluation of AGCNOS patients, in 6 months if HPV DNA is positive and 12 months if HPV DNA is negative [14].

In the majority of abnormal Pap tests, only a single epithelial cell abnormality is observed and reported. The incidence and relevance of coexisting squamous and glandular epithelial is unknown. Our study shows an 8.3% prevalence of dual cell abnormality being the squamous abnormalities more prevalent (57.1%) in relation to glandular abnormalities (28.6%). Although we report these results, the sample is too small in size and consequently we cannot predict the final histological findings based only in the evidence of dual interpretation.

Our study emphasises the importance of close follow-up when AGC cytology is reported since pre-malignancy and malignant disease was identified in 20.8%. Along this review it becomes evident that AGC-FN subgroup has higher prevalence of malignant pathology. It is clear the importance of cervical and endometrial evaluation when in the presence of AGC cytological diagnosis. In our study group similar rates of cervical and endometrial pathology (Cervix: 11.7%; Endometrium 9.1%; p-value=0.599) were found possibly owing to the fact that the average age of AGC study group is above 40.

Although our HPV-DNA testing sample was small in size we found a ratio of HPV associated cervical disease of 50% which indicates that the concomitant use of HPV-DNA can be very useful on AGC followup when no significant pathology is found on cervix or endometrium since studies report a false negative rate of 4.7% in AGC follow-up [6].

Beyond the retrospective design there are other limitations to this study that should be noted. Mostly because of the low prevalence of AGC cytology, our study sample is small in size, which has implications when it comes to subgroup analyses. In the particular case of AGC-FN, although women are older, and the presence of non-benign pathology is more prevalent we cannot exclude a possible bias effect as a result of the small number of patients. When it comes to coexisting squamous and glandular abnormalities we are not able to predict histological outcomes based on the small sample of dual interpretation.

Although showing some limitations our study compares histological findings between AGC subgroups highlighting the importance of HPVPCR co-test in AGC-NOS follow-up and the concomitant evaluation of the endometrium especially in women with AGC over 35 years old.