During the reproductive years, endometrial cells are normally detectable in the menstrual and early proliferative phases of the cycle. Beyond day 10–12 of the cycle the finding is considered abnormal, and should be evaluated in the light of clinical details such as the presence of an intrauterine contraceptive device, a recent pregnancy, contraceptive hormone use with break-through bleeding, recent uterine instrumentation or known endometrial pathology. In postmenopausal patients, any shedding of endometrial cells seen in cervico-vaginal cytological samples should be considered abnormal, since it is a potential indicator of endometrial pathology.

Possible endometrial conditions that may cause abnormal endometrial shedding include endometritis, endometrial polyps, submucosal leiomyomas, endometrial hyperplasia and endometrial carcinoma. However, the sensitivity of cervico-vaginal sampling for endometrial pathology is low since significant exfoliation does not occur even in cases of endometrial carcinoma.⁷ Moreover the likelihood of degenerative changes in spontaneously exfoliated endometrial cells makes their interpretation difficult.

Direct abrasive sampling of endometrial tissue at cervical smear taking may be seen in cases of previous cone biopsy, high sampling with an endocervical brush or cervical endometriosis.

Direct endometrial sampling is significantly more reliable than cervico-vaginal sampling. Several procedures have been used to obtain direct samples of endometrium including:

The technique of direct aspiration of the uterine cavity by a rigid cannula was first introduced by Papanicolaou and Cary in 1943 and since then, a number of flexible devices have been used. Endometrial washings may be obtained under positive or negative pressure. However, the use of both endometrial aspiration and washing techniques has long been discontinued, mainly due to the potential risk of spreading malignant cells, particularly in the case of endometrial washings. Additional disadvantages are the lack of simplicity in these procedures and their high cost.

Today, the endometrial brushing method is the technique most frequently used for cytological endometrial sampling. The specimens may be obtained by means of several devices. The majority consist of an outer tube that reduces cervical and vaginal cell contamination and an inner stick which has a sampling tip on the end (Fig. 26.1). Before the introduction of the device into the uterus, the sampling tip is held inside the outer tube. Once inside the uterine cavity, the sampling tip is released and rotated clockwise and anticlockwise several times. After the collection of endometrial material the tip is retracted inside the outer tube and the device removed. Outside of the uterus, the device is cleaned with gauze to remove the cervical and vaginal cells and then the sampling tip is exposed.

Fig. 26.1 Endometrial brushing: three devices with different sampling tips.

The collected cells may be processed in several ways: conventionally by the ‘flicked method’ or by liquid-based processing.

In conventional endometrial cytology, the brush is rolled on to a glass slide, immediately fixed and subsequently stained by the Papanicolaou method.

In 1968, Johnsson and Stormby⁸ reported the use of a cytological brushing technique to obtain cells from endometrial lesions. However, the sensitivity of endometrial sampling depends in part on whether tissue fragments are obtained so that cytoarchitectural criteria can be used in direct endometrial sampling, especially for the diagnosis of endometrial hyperplasia and well-differentiated adenocarcinoma. We currently use the Uterobrush (Cooper Surgical, Trumbull, CT: ASKA Pharmaceutical, Tokyo, Japan) because insertion to the uterine cavity is easy and painless.^9,10

The Uterobrush is pre-sterilised and consists of a sheath of polypropylene 175 mm in length with a thin wire 0.4 mm in diameter attached to a handle. The tip of the wire has a collecting brush 6 mm in length, composed of nylon bristles with an overall width of 20 mm. Graduation marks along the length of the Uterobrush provide guidance to the operator when introducing the instrument into the endometrial cavity. The tip of the instrument has a bulbous end to minimise the risk of perforation of the uterus.

To collect samples, the sheathed brush is inserted to the level of the uterine fundus and the sheath is slid down toward the handle to expose the brush. Endometrial cells are collected by rotating the handle a few times then the sheath is replaced over the brush in order to entrap the sample as described above.

For specimen preparation, the procedure is improved by using the so-called ‘flicked’ method as follows (Fig. 26.2):

Fig. 26.2 The manipulation procedure of the Uterobrush ‘flicked’ method. (A) The stem of a brush is held in one hand, and forceps (or a thin stick) is held in the other. (B) The tip of the brush is placed on a glass slide and is sharply flicked by forceps or stick. (C) The brush is moved little by little along the slide repeating step B several times. (D) Papanicolaou stained slide.

(From Fujihara et al. 2006.¹¹)

Uterobrush samples prepared by the ‘flicked’ method have a much greater quantity of cell clumps than those using the earlier Endocyte sampler,¹¹ although the size of the cell clumps is the same as with the Endocyte brush. The same criteria are applicable to both types of sample. Besides improving the cytodiagnosis of endometrial lesions, the findings are helpful in the standardisation of criteria in direct intrauterine cell samples,¹¹ for example by allowing observation of cell clumps with either a tubular or sheet-like pattern, tube-shaped glands being symmetrical with nearby cohesive endometrial stromal cells but disrupted tube-shaped glands yield a sheet-like configuration (Fig. 26.3).

Fig. 26.3 Cell clumps with tube or sheet-like pattern. (A) Large tissue fragments with tube-shaped glands surrounded by endometrial stroma. (B) Tube-shaped gland. (C) Sheet-like pattern, proliferative phase (PAP).

(From Fujihara et al. 2006.¹¹)

In liquid-based cytology, the brush head of the device is immersed in a vial of fixative solution where it is vigorously rotated several times to ensure release of the cells collected (Fig. 26.4). The device is then removed from the vial. The sample is ready for processing after about 30 minutes in fixative solution and is stable for several weeks thereafter. Excess blood and mucus are eliminated by means of washing through a succession of centrifugation and resuspension of the sample in mucolytic and haemolytic agents. The cells are then placed onto the slide by special automated or semi-automated processor machines which transfer a representative fraction of the collected cells for staining. Subsequently, further slides may be obtained from the remaining sample and these may be stained routinely or used for immunohistochemical or molecular investigations.

Fig. 26.4 (A,B) After removal from the uterus the sampling tip is exposed and then immersed in a vial containing a fixative solution where it is vigorously rotated several times to ensure release of the endometrial cells (C). (Endoflower device; RI-MOS, Mirandola, Modena, Italy.)

When endometrial brushings are suspension-fixed, cell aggregates maintain their three-dimensional pattern comprising, in essence, microbiopsies. Three-dimensional structures such as glands may be observed in cytology preparations and this allows direct correlation between histological and cytological features.¹²

Endometrial cells, both epithelial and stromal, show several different cytoarchitectural features depending on the woman’s age, the menstrual phase and any administration of hormonal therapy. Epithelial cells may be aggregated in three-dimensional cylindrical clusters or in two-dimensional sheets. They are small, only slightly larger than mature lymphocytes. The cytoplasm, apart from during the secretory phase, is scant. Nuclei are small, round and uniform in size and shape. The chromatin is finely granular and small chromocentres may be present (Fig. 26.5).

Fig. 26.5 Tubular endometrial clusters (proliferative endometrium): endometrial cells show scant cytoplasm and small nuclei. LBC direct endometrial sampling.

Stromal cells are present as single elements or as cell groupings. Stromal clusters may show cellular overlapping and irregularity in their outline with protruding bare nuclei. Individual cells may be spindle-shaped with scant cytoplasm or, when modified by decidual change, may be larger and polygonal with obvious cytoplasm. Under the influence of progesterone there is an increase in cytoplasmic volume due to glycogen accumulation. Nuclei are isomorphic and have finely granular chromatin. Nucleoli are generally small or absent, although with progesterone they may be prominent. The distinction between stromal and epithelial endometrial cells might sometimes be difficult to ascertain. In such cases, the immunocytochemistry, which is easily performed on liquid-based specimens, may help in this distinction. Indeed, endometrial stromal cells are CD10 (human membrane-associated neutral peptidase) positive while endometrial epithelial cells are not (Fig. 26.6).¹³ Blood vessels may also be identified in liquid-based samples, as described below.

Fig. 26.6 (A) Stromal cluster on progesterone: obvious cytoplasm and isomorphic nuclei showing finely granular chromatin and small chromocentres. (B) CD10 positive immunostaining. LBC direct endometrial sampling.

Cervical cells, either columnar, squamous, intermediate and metaplastic, may variably contaminate the cytological endometrial specimens. Columnar mucin-secreting cells may be identified as single cells or they may occur as strips or sheets showing a honeycomb pattern. Nuclei are round or oval and, depending on their orientation on the slide, basally or centrally located. Chromatin is finely granular. Nucleoli are generally inconspicuous. Cytoplasm is abundant and clear. Superficial squamous cells are polygonal with eosinophilic or cyanophilic cytoplasm and small pyknotic nuclei. Intermediate cells have more vesicular, larger nuclei than superficial squamous cells. Cervical metaplastic cells are smaller than either of these. They are identified by their cytoplasmic prolongations and also may show prominent chromocentres. Metaplastic cells usually appear as cell clusters or as flat sheets.

Cylindrical ciliated cells, deriving from the isthmus, from endometrioid metaplasia or from the fallopian tube, are also sometimes visible in cytological endometrial samples. Such cells have thin wavy cilia extending from the luminal pole of the cell with its thick eosinophilic terminal bar (Fig. 26.7).

Fig. 26.7 Columnar cells showing thin cilia extending from the pole of the cells where a thick terminal bar is clearly visible. LBC direct endometrial sampling.

Occasional inflammatory cells, either mono- or polymorphonuclear may be found in non-pathological endometrial samples. The number of polymorphs found in samples from women having menstrual cycles is greatly dependent on the phase of the cycle. Indeed, as with cervico-vaginal samples, endometrial samples obtained in the late secretory phase may be accompanied by a large number of granulocyte or ‘K’ cells. Multinucleated histiocytes are frequently found during the menopause, having no association with any endometrial pathology. They are oval to round in shape and may be huge. The cytoplasm is voluminous, the nuclei, often in dozens, are oval with slight variation in size and finely granular chromatin (Fig. 26.8).¹⁴

Fig. 26.8 Multinucleated histiocyte close to atrophic endometrial gland. LBC direct endometrial sampling.

As can be seen from the discussion so far, it is apparent that if optimal benefit is to be obtained in the detection of the endometrial pathology by the cellular approach, the provision of pertinent clinical informations is important. The age, the day of menstrual cycle, or menopausal status of the patient should be provided. Other important information such as any presenting symptoms, the use of exogenous hormones and details of any recent intrauterine instrumentation should be included.

Endometrial cell clumps of various sizes are usually abundant in cytological material collected using the Endocyte or Uterobrush sampler. The cytological characteristics of fragments from normal uterine endometrium are illustrated below.

Endometrial glands appear as virtually straight and tube-shaped (Fig. 26.9). The width of the tube-shaped gland is approximately uniform, and cohesion of the endometrial stromal cells to the margins of the gland are noted. When the tube-shaped gland is disrupted and opened out, it has a sheet-like shape and adhesion of the stromal cells is observed. The surface covering of endometrial cells also appears as a sheet (Figs. 26.10, 26.11).

Fig. 26.9 Cell clumps with a tube or sheet-like pattern. The width of the tube-shaped gland is approximately uniform, and shows adhesion of the endometrial stromal cells to the margins of the gland (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.10 When a tube-shaped gland is disrupted and opened out, it shows a sheet-like form (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.11 Frequent large tissue fragments with tube-shaped glands surrounded by endometrial stroma are seen (proliferative phase) (PAP).

(From Norimatsu et al. 2006.²⁸)

In these conditions, the endometrial glands and their lining cells undergo a number of changes. The cytoarchitectural characteristics of the resulting abnormal endometrial and stromal cell clumps include the following.

Dilated or branched patterns

In these examples, irregular dilation and branching were noted in tubular glands. The maximum width of a gland is more than twice that of its minimum width and shows adhesion of the endometrial stromal cells to the margins of the gland (Figs 26.12–26.15).

Fig. 26.12 Cell clump with dilated pattern. Simple endometrial hyperplasia (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.13 Complex endometrial hyperplasia (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.14 Large tissue fragments with dilated or branched glands are seen in, complex endometrial hyperplasia (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.15 A cell block specimen from the case shown in Figure 26.14. The crowded glands show irregular dilatation with out-pouching into the endometrial stroma and branching (H&E).

(From Norimatsu et al. 2006.²⁸)

Cytological findings: dilated or branched patterns

• Cell clumps of abnormal endometrium

• Tube-shaped gland have irregularly dilated or branched patterns

• The maximum width of a gland is greater than twice that of the minimum width

• Stromal cells adhere to the margins of the gland.

Irregular protrusions patterns

Some irregular small projections were noted at the edges of cell clumps (Fig. 26.16). The margin of the cytoplasm of those small projections is clearly observed.

Fig. 26.16 Cell clump with irregular protrusions. Some irregular small projections are observed from the margins of the cell clumps (PAP).

(From Norimatsu et al. 2006.²⁸)

Papillotubular patterns

The endometrial gland shows a papillary growth pattern with irregular branching and projections as described above (Figs. 26.17, 26.18). Cohesion of the endometrial stromal cells is not noted at the margins of the gland. When the papillary structure is complex and confluent, much glandular space is formed, and back-to-back structures with a cribriform pattern are also recognised (Figs 26.19–26.21). Occasionally, a fibrovascular core may be observed in the epithelial papillae.

Fig. 26.17 Cell clumps with a papillotubular pattern The gland shows a papillary growth pattern and irregular branching (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.18 Large tissue fragment with a papillotubular pattern. A papillary growth pattern was observed in this Grade 1 adenocarcinoma (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.19 Large tissue fragment with a papillotubular pattern. A cribriform structure is recognised (Grade 1 adenocarcinoma) (PAP).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.20 Large tissue fragment with a papillotubular pattern. The back-to-back structure is obvious (Grade 1 adenocarcinoma).

(From Norimatsu et al. 2006.²⁸)

Fig. 26.21 A cell block specimen from the case shown in Figure 26.20. The back-to-back pattern and cribriform structure are apparent (H&E).

(From Norimatsu et al. 2006.²⁸)

Cytological findings: papillotubular patterns

• Cell clumps of abnormal endometrium

• Papillary growth pattern and irregular branching with projections

• Back-to-back structures with a cribriform pattern

• Cohesion of the endometrial stromal cells at the gland margins is not seen.