Cervical cytology: Evaluation of atypical and malignant glandular cells

INTRODUCTION — Atypical glandular cells (AGC) on cervical cytology usually originate from the glandular epithelium of the endocervix or endometrium. They are a less common finding than abnormal squamous cells.

Patients with AGC require further evaluation for premalignant conditions of the cervix, uterus, and, rarely, ovary and fallopian tube. The evaluation of patients with cervical cytology demonstrating atypical and malignant glandular cells is reviewed here. Cervical cancer screening strategies and techniques, interpretation of cervical cytology results, follow-up of other abnormal cytology results, and management of cervical neoplasia are discussed separately.

●(See "Screening for cervical cancer in resource-rich settings".)

●(See "Cervical cancer screening tests: Techniques for cervical cytology and human papillomavirus testing".)

●(See "Cervical cancer screening: The cytology and human papillomavirus report".)

●(See "Cervical cancer screening: Risk assessment, evaluation, and management after screening".)

●(See "Cervical intraepithelial neoplasia: Management".)

TERMINOLOGY — Terminology for reporting cervical cytology was standardized by the Bethesda System in 1988 [1]. This system has been revised several times (table 1) [2-4]. The terminology used to classify AGC is:

●Atypical glandular cells (AGC) – Either endocervical (AEC), endometrial, or not otherwise specified (NOS) is noted as a subcategory. This replaces the previous term "atypical glandular cells of undetermined significance (AGUS)." This term should not be confused with terminology for squamous cell abnormalities, which includes atypical squamous cells of undetermined significance (ASC-US). (See "Cervical cancer screening: The cytology and human papillomavirus report", section on 'Intraepithelial cell abnormalities'.)

●Atypical glandular cells, favor neoplastic – Endocervical is the only subcategory. This designation is for specimens that show features suggestive of, but not sufficient for, an interpretation of adenocarcinoma.

●Endocervical adenocarcinoma in situ (AIS).

●Adenocarcinoma – Adenocarcinoma cells may be subcategorized as endocervical, endometrial, extrauterine, or NOS.

INCIDENCE — Glandular abnormalities are found in <1 percent of cervical cytology samples, most commonly in patients ages ≥40 years (eg, 1.1 to 2.4/1000 at ages 40 to 69 versus 0 to 0.8/1000 in younger patients [5]) [6-8]. By contrast, abnormal squamous cytologic findings (atypical squamous cells or a squamous intraepithelial lesion) are more common and are more prevalent in patients ages 15 to 29 years [5].

There are few data regarding AGC during pregnancy or the postpartum period. Based upon available data, the incidence of AGC and of significant neoplasia appears to be similar or higher in pregnant compared with nonpregnant patients of similar age [9,10].

ANATOMIC SITE OF ORIGIN OF AGC — Almost all AGC originate from the endocervix or endometrium, but occasionally, further evaluation results in a diagnosis of disease at another site. Ovarian/fallopian tube cancer is the most common of these extrauterine sites and has been reported in 0.1 to 0.6 percent of patients with AGC [7,11,12]. More rarely, AGC is a sign of an occult glandular malignancy from a nongynecologic organ, such as the colon, breast, or stomach [7,13-15].

Endometrial polyps and tubal metaplasia are examples of benign uterine disorders that have been associated with AGC [11,16].

RISK OF PREMALIGNANT OR MALIGNANT DISEASE — AGC on cervical cytology are associated with premalignant or malignant disease in approximately 30 percent of cases [11,17-25]. Interestingly, most of the lesions found upon histologic evaluation are squamous rather than glandular [7,19].

Histologies in patients with AGC cytology — Histologic evaluation of patients with AGC may identify normal findings or squamous or glandular lesions. The distribution of histology was illustrated in the following studies:

●Two literature reviews including approximately 4300 patients with AGC (or atypical glandular cells of undetermined significance [AGUS] for specimens evaluated using the pre-2001 Bethesda classification (see 'Terminology' above)) reported benign findings in 64 to 71 percent of cases [7,19]. Abnormalities included:

•Squamous cervical intraepithelial neoplasia (CIN) – 20 to 28 percent, including:

-Low-grade CIN (CIN 1) – 9 percent

-High-grade CIN (CIN 2,3) – 11 percent

•Squamous cervical carcinoma – 0.3 to 1.0 percent

•Cervical adenocarcinoma in situ (AIS) – 3 to 4 percent

•Cervical adenocarcinoma – 1 to 2 percent

•Endometrial hyperplasia – 1 percent

•Endometrial adenocarcinoma – 2 to 3 percent

●A Swedish national registry study included 14,625 patients with AGC on their first reported cervical cytology specimen who were followed for up to 15.5 years [26]. In this study, AGC was classified by the Swedish Association for Clinical Cytology, and subcategories of AGC were not used. On follow-up of the AGC cytology specimen within six months, the following histologies were noted:

•Squamous cervical carcinoma – 0.3 percent

•Cervical adenocarcinoma – 0.99 percent

Patients with AGC continued to be at a higher risk of cervical cancer than patients with normal cytology for up to 15.5 years, with the highest risk within 3.5 years, and then decreasing over time. Compared with patients with normal index cytology, the risks were 24-fold at 0.5 to 3.5 years, 10-fold at 3.5 to 6.5 years, and 4.8-fold at 10.5 to 15.5 years. Patients with AGC were at a higher risk of adenocarcinoma than of squamous carcinoma. Interestingly, for up to 10.5 years, patients with AGC were at a higher risk of developing cervical cancer (either adenocarcinoma or squamous) than patients with high-grade squamous intraepithelial lesion, and the risk became comparable from 10.5 to 15.5 years. A limitation is that the study followed patients forward from an index cytology result but did not include information about subsequent cytology results.

●In a retrospective review of 238 patients with endometrial cancer, 50 (21 percent) had abnormal Pap smears, and 17 of the 50 abnormal results were AGC [27]. AGC correlated with cervical stromal invasion of the endometrial cancer.

The lower rate of glandular compared with squamous cervical abnormalities diagnosed in patients with AGC in literature reviews may be secondary to the lower sensitivity of cervical cytology for detecting endocervical glandular dysplasia and malignancy than for detecting squamous malignancy [7,19]. This lower sensitivity has been attributed to several factors: it can be difficult for the cytopathologist to distinguish between high-grade squamous cells and glandular cells; endometrial cells, reactive endocervical cells, tubal metaplasia, and cervical endometriosis may mimic AIS; and lesions may not be seen or sampled because they are small or high in the endocervical canal. Another reason for underdiagnosis is the presence of minimal, poorly preserved endocervical cells, which are difficult to assess [28].

AGC has also been associated with less frequent histologies, including adenosquamous cervical cancer and uterine carcinosarcoma (formerly referred to as malignant mixed müllerian tumors) [29-31].

Correlation between AGC subcategory and subsequent histology — The cytologic subcategory is somewhat predictive of the histologic diagnosis. As an example, in a study of 460 patients with AGC on cytology, the most common histology by subcategory was [32]:

●AGC-not otherwise specified was associated with endometrial adenocarcinoma in 10.2 percent of cases.

●AGC-endocervical (also abbreviated AEC) was associated with invasive cervical adenocarcinoma in 5.9 percent of cases, AIS in 2.4 percent of cases, and CIN 2,3 in 5.3 percent of cases.

●AGC-endometrial was associated with endometrial adenocarcinoma in 27.8 percent of cases and atypical complex endometrial hyperplasia in 22.2 percent of cases.

A limitation of this study was that it combined AGC, favor neoplastic subcategories with other AGC categories (eg, AEC, favor neoplastic was combined with AEC), thus increasing the likelihood of malignancy in each subcategory.

There are few data about the histologic correlation of AGC classified as favor neoplastic. In one study (n = 138), high-grade CIN, AIS, or invasive cervical or endometrial cancer was found more often in patients with AGC, favor neoplastic compared with other patients with AGC (56 versus 8 percent) [23].

Frequency of a coexisting squamous cytologic abnormality — Approximately one-half of patients with AGC have a coexisting squamous cytologic abnormality (atypical squamous cells or a squamous intraepithelial lesion), and these patients appear to be more likely to have a squamous rather than a glandular lesion [19,32,33]. As an example, in a study that included 353 patients with AGC who underwent histologic evaluation, the proportion of squamous histologic lesions was higher in those with AGC plus a squamous cytologic finding (squamous histology: 90 percent; glandular histology: 10 percent) compared with AGC cytology alone (squamous histology: 72 percent; glandular histology: 28 percent) [19].

RISK FACTORS FOR MALIGNANT OR PREMALIGNANT DISEASE

Human papillomavirus infection — For patients with AGC, testing for high-risk human papillomavirus (HPV) is not as reliable for triaging follow-up as it is for atypical squamous cells because AGC may also represent upper tract pathology (see 'Anatomic site of origin of AGC' above) [33,34]. However, it is still used in the evaluation of patients with AGC and a positive high-risk HPV test is associated with an increased risk for a subsequent histologic diagnosis of cervical intraepithelial neoplasia (CIN) [35-39]. (See 'Follow-up of patients with negative or low-grade histologic findings on initial evaluation (colposcopy and biopsy)' below.)

In a cohort study of over 1.5 million patients followed for more than a decade, the immediate risk of CIN 3+ in patients with AGC (all categories) on cytology and an unknown PAP/HPV history (3252 total patients) was increased in those who were HPV-positive compared with those who were HPV-negative (26 versus 1 percent) [40]. The five-year risk of developing CIN 3+ in this cohort was 35 and 1.5 percent, respectively. In another study in which 222 patients had HPV-positive AGC, subsequent evaluation revealed 93 cases of high-grade squamous intraepithelial lesion (HSIL), 33 cases of cervical adenocarcinoma in situ, and 6 cases of invasive cervical adenocarcinoma, whereas among the 170 HPV-negative patients with AGC, there were 4 cases of HSIL, 1 case of invasive cervical squamous cell cancer, 5 cases of endometrial cancer, and 1 case of breast cancer [39].

HPV genotyping can also help stratify risk. In a cohort study of 341 patients with non-16/18 HPV-positive AGC, those with HPV 33 and 31 had higher rates of CIN 3 or greater than patients with HPV 45 or 52 (82.9, 82.1, 49, and 42.5, respectively) [41]. If a patient is HPV positive or negative, when co-tested with cytology, the diagnostic work-up and management are similar [42]. The use of other markers (ie, Dual Stain cytology with p16/Ki-67) for triage of squamous lesions has not been studied in patients with AGC [43]. (See "Cervical cancer screening tests: Techniques for cervical cytology and human papillomavirus testing", section on 'Sample processing'.)

Older age — The risk of malignancy in patients with AGC increases with age [11,17,22,32,44]. In one retrospective study of 662 patients with AGC, the overall rate of malignancy was highest in patients ≥50 years (15 percent) compared with those ages 40 to 49 years (2.8 percent) or <40 years (2.0 percent) [11]. In patients ≥50 years, endometrial malignancy was most common, whereas in patients <40 years, cervical malignancy was most common.

Persistent AGC-NOS — Patients with persistent AGC-NOS on two or more cytology results are at especially high risk of significant glandular disease [17]. In one study, three of five postmenopausal patients with persistence on repeated cytologic testing were subsequently found to have endometrial adenocarcinoma [21].

INITIAL EVALUATION

General principles

●The presence of AGC on cervical cytology is a significant marker for premalignant disease of the cervix or endometrium; thus, evaluation includes testing of both sites. The approach to initial evaluation differs somewhat by AGC subcategory and pregnancy status. (See 'All AGC categories (except endometrial)' below and 'AGC-endometrial' below and 'Modifications for pregnant patients' below.)

●Management of patients with negative or low-grade histologic findings (no cervical intraepithelial neoplasia [CIN] 2 or higher [CIN 2+], no adenocarcinoma in situ [AIS], and no cancer on colposcopy with directed cervical biopsies and sampling of the endocervical canal) is described below. (See 'Follow-up of patients with negative or low-grade histologic findings on initial evaluation (colposcopy and biopsy)' below.)

●Management of patients with positive findings depends on the histologic diagnosis:

•(See "Cervical intraepithelial neoplasia: Management".)

•(See "Invasive cervical cancer: Staging and evaluation of lymph nodes".)

•(See "Cervical adenocarcinoma in situ".)

•(See "Invasive cervical adenocarcinoma".)

•(See "Small cell neuroendocrine carcinoma of the cervix".)

•(See "Endometrial hyperplasia: Clinical features, diagnosis, and differential diagnosis".)

•(See "Endometrial carcinoma: Staging and surgical treatment".)

Procedures for colposcopy and endometrial sampling are discussed in detail separately. (See "Colposcopy" and "Endometrial sampling procedures".)

All AGC categories (except endometrial) — Nonpregnant patients with cervical cytology of AGC-not otherwise specified or AGC-endocervical (including those classified as favor neoplastic), AIS, or adenocarcinoma are evaluated initially with the following (algorithm 1) [45-47]:

●Colposcopy and sampling of the endocervical canal (endocervical curettage).

●Endometrial sampling is performed for patients ≥35 years old and for patients <35 years old at risk for endometrial neoplasia (risk factors or symptoms are present) (table 2 and table 3). It is important to get an adequate endometrial sample.

This evaluation is required regardless of human papillomavirus (HPV) results.

AGC-endometrial — Nonpregnant patients who have AGC-endometrial should be evaluated with endometrial sampling and endocervical sampling; colposcopy is not required but may be performed as part of the initial evaluation (algorithm 1) [45,47]. Patients are then treated, as appropriate, for the histologic abnormality that is detected.

If the cytology report also has abnormal squamous findings (atypical squamous cells or squamous intraepithelial lesion) in combination with AGC-endometrial, colposcopy should be performed. (See "Cervical cancer screening: Risk assessment, evaluation, and management after screening".)

Modifications for pregnant patients — For pregnant patients with AGC, endocervical sampling with a curette and endometrial sampling should not be performed as there is a risk of disturbing the pregnancy; however, the endocervical canal may be sampled gently with a cytobrush. Colposcopic examination may also be performed approximately six weeks later; endocervical lesions initially missed in early pregnancy may be seen as the pregnancy progresses and eversion of the endocervix occurs [48,49].

FOLLOW-UP OF PATIENTS WITH NEGATIVE OR LOW-GRADE HISTOLOGIC FINDINGS ON INITIAL EVALUATION (COLPOSCOPY AND BIOPSY) — The follow-up of patients with negative or low-grade (cervical intraepithelial neoplasia [CIN] 1) histologic findings on initial evaluation depends upon the AGC subcategory. Some patients will require cervical conization because glandular neoplasia of the cervix, in contrast to squamous disease, is often characterized by noncontiguous lesions ("skip lesions") and some glandular disease may be located high in the cervical canal.

AGC-NOS or AEC-NOS cytology — The initial approach to AGC-not otherwise specified (AGC-NOS) or atypical endocervical cells (AEC)-NOS cytology is as follows (algorithm 2) [45,47]:

If the initial evaluation (colposcopy and biopsies) is CIN 2 or less (no CIN 2+, adenocarcinoma in situ [AIS], or cancer is found), then co-testing with cervical cytology and human papillomavirus (HPV) testing at one and two years is performed:

●If cytology and HPV testing are both negative – Repeat co-testing in three years.

●If any abnormality is found at one or two years – Perform colposcopy with endocervical curettage (ECC) and then endometrial sampling if the colposcopic findings are nondiagnostic.

•If CIN 2 or CIN 3, but no glandular abnormalities are identified, management is directed by the specific abnormality. (See "Cervical intraepithelial neoplasia: Management".)

•If colposcopy with ECC and endometrial sampling are nondiagnostic, we suggest conization (cold knife, electrocautery, or laser techniques are acceptable). In addition, vaginal colposcopy should be performed in patients with a history of exposure to diethylstilbestrol.

If the initial evaluation (colposcopy and biopsies) is positive for CIN 2+ but no glandular neoplasia is found, management is directed by the specific abnormality. (See "Cervical intraepithelial neoplasia: Management".)

AGC, favor neoplastic; adenocarcinoma in situ; or adenocarcinoma cytology — Cervical cytology classified as AGC, favor neoplastic; AIS; or adenocarcinoma is associated with a very high risk of underlying invasive disease [23]. If the initial evaluation (colposcopy and biopsy) is negative for AIS or cancer, the following is our approach:

●A diagnostic excisional procedure followed by an ECC of the remaining endocervix is required (algorithm 2) [45,47]. We suggest a cold knife conization rather than a loop electrosurgical excision procedure to avoid thermal artifact at the margins. (See "Cervical intraepithelial neoplasia: Diagnostic excisional procedures".)

●If the results of the diagnostic excisional procedure are negative, endometrial sampling is required. Due to skip lesions and the higher incidence of lower uterine segment involvement, we perform a dilation and curettage at the same time as the cone biopsy.

●If the results of endometrial sampling are negative, a pelvic ultrasound and other testing (eg, cancer antigen 125 [CA 125]) can be performed to look for a fallopian tube or ovarian lesion.

●If pelvic ultrasound and CA 125 are also negative, we obtain cervical cytology, HPV, and ECC every six months. (See 'Patients with persistent AGC and negative colposcopy, endometrial biopsy, and conization' below.)

Positive findings are managed as appropriate for the abnormality identified.

AGC-endometrial cytology — If the initial evaluation (endometrial and endocervical sampling) is negative, then (algorithm 1) [45,47]:

●Colposcopy is performed.

●If colposcopy is negative, we repeat endometrial and endocervical sampling and pelvic ultrasound (to assess the uterine lining) within six months.

•If clinical concern for malignancy is high, our authors have also used pelvic ultrasound and hysteroscopy with dilation and curettage in this setting.

●If repeat endometrial and endocervical sampling are negative, the patient may resume routine screening for cervical cancer. (See "Screening for cervical cancer in resource-rich settings".)

Positive findings are managed as appropriate for the abnormality identified.

PATIENTS WITH PERSISTENT AGC AND NEGATIVE COLPOSCOPY, ENDOMETRIAL BIOPSY, AND CONIZATION — Patients with persistent cytologic findings associated with a high risk of malignancy (persistent AGC-not otherwise specified; AGC-endocervical; AGC-endometrial; AGC, favor neoplastic; adenocarcinoma in situ; adenocarcinoma) and who have negative cervical and endometrial findings despite comprehensive evaluation (ie, colposcopy, endometrial biopsy, conization) should be evaluated for primary or metastatic disease involving the ovary/fallopian tube. Occasionally, AGC is a sign of an occult glandular malignancy from the ovary/fallopian tube or a nongynecologic organ, such as the colon, breast, or stomach [7,13-15]. AGC with a negative human papillomavirus (HPV) test is rare but can be consistent with AGC of noncervical origin. This result does not exclude a uterine or cervical origin of AGC.

The first-line study is a transvaginal ultrasound and cancer antigen 125 (CA 125). Patients with an adnexal mass or an elevated tumor marker should undergo further evaluation for ovarian or fallopian tube cancer. (See "Approach to the patient with an adnexal mass" and "Screening for ovarian cancer", section on 'Lack of benefit of screening strategies'.)

If the above evaluation is negative, the clinician should confirm that age-appropriate breast and colon cancer screening has been performed and order this screening if it has not been performed. Rarely, these patients have a primary tumor at an occult site, particularly the breast or colon.

There is no standard work-up for patients with persistent AGC that is unexplained. Further evaluation, which may include abdominal and pelvic imaging and serum tumor markers, depends on the medical history and physical examination and physician and patient preferences. In a systematic review that included almost 7000 patients with AGC, 203 malignancies were diagnosed after histologic follow-up and included endometrial cancer (57.6 percent), cervical adenocarcinoma (23.6 percent), cervical squamous carcinoma (5.4 percent), ovarian or fallopian tube carcinoma (6.4 percent), and other malignancy (6.9 percent, histologic types not provided) [7].

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Cervical cancer screening, prevention, and management".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Beyond the Basics topics (see "Patient education: Cervical cancer screening (Beyond the Basics)")

SUMMARY AND RECOMMENDATIONS

●Incidence and risk of malignancy – Atypical glandular cells (AGC) are found in <1 percent of cervical cytology samples. AGC are associated with premalignant or malignant lesions of the endocervix or endometrium in approximately 30 percent of cases. Most of the lesions found upon histologic evaluation are squamous rather than glandular. (See 'Incidence' above and 'Risk of premalignant or malignant disease' above.)

●Site of origin – Almost all AGC originate from the endocervix or endometrium, but occasionally, AGC have been associated with ovarian/fallopian tube malignancy or vaginal cancer and, even less commonly, with an occult glandular malignancy from a nongynecologic organ, such as the colon, breast, or stomach. Endometrial polyps and tubal metaplasia are examples of benign uterine disorders that have been associated with AGC. (See 'Anatomic site of origin of AGC' above.)

●Risk factors for malignancy – In patients with AGC, older age (eg, ≥50 years) is a risk factor for malignancy. Testing for high-risk human papillomavirus (HPV), however, is not as reliable for triaging follow-up as it is for atypical squamous cells because AGC may also represent upper tract pathology. (See 'Risk factors for malignant or premalignant disease' above.)

●Initial evaluation

•Nonpregnant patients with AGC-not otherwise specified (AGC-NOS) or AGC-endocervical (AEC; including those classified as favor neoplastic), adenocarcinoma in situ (AIS), or adenocarcinoma are evaluated initially with cervical colposcopy plus directed cervical biopsies and sampling of the endocervical canal (algorithm 1). Endometrial sampling is performed for patients ≥35 years old and for younger patients at risk for endometrial neoplasia (risk factors or symptoms are present) (table 2 and table 3). This evaluation is required regardless of HPV results. (See 'All AGC categories (except endometrial)' above.):

•Nonpregnant patients with AGC-endometrial are evaluated with endometrial sampling and endocervical sampling; colposcopy is not required but may also be performed as part of the initial evaluation (algorithm 1). Patients are treated, as appropriate, for the histologic abnormality that is detected. If the cytology report also has abnormal squamous findings (atypical squamous cells or squamous intraepithelial lesion) in combination with AGC-endometrial, colposcopy should be performed. (See 'AGC-endometrial' above.)

•For pregnant patients with AGC, endocervical sampling with a curette and endometrial sampling should not be performed as there is a risk of disturbing the pregnancy; however, the endocervical canal may be sampled gently with a cytobrush or evaluated with colposcopic examination six weeks later. (See 'Modifications for pregnant patients' above.)

●Follow-up

•Patients with negative or low-grade findings on initial evaluation of AGC-NOS; AGC, favor neoplastic; AIS; or adenocarcinoma require further assessment, which depends upon the AGC subcategory (algorithm 2). (See 'Follow-up of patients with negative or low-grade histologic findings on initial evaluation (colposcopy and biopsy)' above.)

•Patients with negative or low-grade findings on initial evaluation of AGC-endometrial should have colposcopy, and if negative, endometrial and endocervical sampling as well as pelvic ultrasound are repeated within six months. If colposcopy and repeat endometrial and endocervical sampling are negative, the patient may resume routine screening for cervical cancer. (See 'AGC-endometrial cytology' above.)

●Persistent findings – Patients with persistent cytologic findings associated with a high risk of malignancy who have negative cervical and endometrial findings despite comprehensive evaluation (ie, colposcopy, endometrial biopsy, conization) should be evaluated for primary or metastatic disease involving the ovary/fallopian tube. The first-line study is a transvaginal ultrasound and cancer antigen 125 (CA 125). Patients with an adnexal mass or an elevated tumor marker are further evaluated for ovarian or fallopian tube cancer.

If this evaluation is negative, the clinician should confirm that age-appropriate breast and colon cancer screening has been performed and order this screening if it has not been performed. Rarely, these patients have a primary tumor at an occult site, particularly the breast or colon. (See 'Patients with persistent AGC and negative colposcopy, endometrial biopsy, and conization' above.)