The Effect of the Extent of Neuroendocrine Differentiation on Cytopathological Findings in Primary Neuroendocrine Neoplasms of the Breast.

OBJECTIVE: This study aimed to describe the cytological features of neuroendocrine breast tumors and to show the effect of the extent of neuroendocrine differentiation on cytological features.
METHODS: Breast tumor excision materials showing immunostaining with neuroendocrine markers (Synaptophysin or Chromogranin A) were determined and divided into two groups: cases with focal (10%-50% of tumor cells) staining and cases with diffuse (>50% of tumor cells) staining. A group of cases without neuroendocrine features/staining was used as control group. Fine needle aspiration biopsy specimens of the tumor mass or metastatic lymph nodes were examined and compared.
RESULTS: Twenty cases with neuroendocrine differentiation were included. Eleven cases were in the diffuse group, nine cases were in the focal group. Clean background, high cellularity, loosely cohesive cell groups with monotonous appearance, and naked nuclei were more common in the diffuse group. On the contrary, tight cohesive cell groups, the proportion of large cells, nuclear pleomorphism, and nucleolar prominence were higher in the group with focal staining. Plasmocytoid appearance, isolated cell groups, and binucleation were in similar distribution in both groups. Although round-oval nuclei were dominant in both groups, round nuclei were observed to be slightly more in the diffuse group. Only two cases in diffuse group showed cytoplasmic granularity and one case in focal group showed necrosis and mitosis. In the control group, tight cohesive groups, large cell size, pleomorphism, prominent nucleoli, and coarse chromatin were more commonly encountered.
CONCLUSIONS: Clean background, hypercellularity, loss of cohesion, naked nuclei, monotonous cells with round nucleus, and granular cytoplasm were more prominent in cases showing diffuse staining with neuroendocrine markers. Suspecting neuroendocrine differentiation in tumors that show focal staining with neuroendocrine markers can be challenging in cytological preparations. Copyright:

INTRODUCTION

Primary breast carcinoma with neuroendocrine features was first described in 1963 by Feyrter and Hartmann, with the report of two cases showing “carcinoid” growth pattern.[1] In 1977, Cubilla and Woodruff described eight cases of carcinoid tumor with a solid/alveolar pattern.[2] Today, primary neoplasms of the breast with neuroendocrine features include a heterogeneous group of tumors with different biological behavior and prognosis.[3] Its incidence ranges between 0.5% and 18%.[456] Neuroendocrine breast carcinomas are usually seen in older women, clinical presentation is not different from other breast tumors.[4]

It is important to differentiate metastatic neuroendocrine neoplasia from primary neuroendocrine neoplasia of the breast. In a study of 18 NET cases that metastasized to the breast, it was revealed that 62% of these tumors originated from the gastrointestinal tract and 28% from the lung.[7] It has been reported that 44% of these tumors were incorrectly diagnosed as primary breast carcinoma.[7] In another study, it was shown that up to 3% of gastrointestinal neuroendocrine tumors can metastasize to the breast.[8] Cases of medullary thyroid carcinoma metastasizing to the breast have also been reported.[91011] The presence of an intraductal component in the tumor is strong evidence showing that neuroendocrine carcinoma is of breast origin.[12] Immunohistochemical studies also can help in the diagnosis. The presence of a neuroendocrine tumor in a different organ should be excluded clinically before a primary breast neuroendocrine neoplasia is diagnosed.[13]

There are different findings regarding the prognosis of neuroendocrine breast tumors. Some studies have found that these tumors have a better prognosis than other invasive breast carcinomas,[1415] whereas other studies state more aggressive behavior.[161718] Some studies show that neuroendocrine marker expression does not affect prognosis.[1920]

Neuroendocrine tumors of the breast were first included in the World Health Organization (WHO) Classification of Breast Tumors in 2003 (3rd Edition, 2003), based on the study by Sapino et al.[421] In WHO 2003 classification, at least 50% of the tumor cells were expected to have immunoreactivity with neuroendocrine markers for the diagnosis of neuroendocrine carcinoma.[4] According to the criteria in the following WHO 2012 classification, invasive carcinomas showing neuroendocrine differentiation were included in neuroendocrine carcinomas regardless of the proportion of tumor cells expressing neuroendocrine marker.[22] In the WHO 2012 classification, breast tumors with neuroendocrine differentiation were divided into three categories based on their morphological features: (a) well-differentiated (carcinoid-like) neuroendocrine tumor, (b) poorly differentiated/small cell neuroendocrine carcinoma, and (c) invasive breast carcinoma with neuroendocrine differentiation.[22] In the latest edition of the WHO classification (5th Edition, 2019), attention was paid to the percentage of neuroendocrine pattern in the tumor. According to this, tumors with less than 10% neuroendocrine pattern were classified as invasive-no specific type or other types, tumors with 10%–90% neuroendocrine pattern as mixed neuroendocrine neoplasia, and tumors with more than 90% neuroendocrine pattern were classified as NET or NEC.[23]

Synaptophysin and Chromogranin A are neuroendocrine markers with very good sensitivity and specificity.[2425] Neuron-specific enolase positivity can be observed in most of breast carcinomas. However, this is not always an indicator of neuroendocrine differentiation.[26] The sensitivity and specificity of CD56 and other neuroendocrine immunohistochemical markers are lower.[27] High molecular weight cytokeratins stain non-neuroendocrine tumors, thus their negativity supports neuroendocrine tumors.[2728]

Estrogen receptor (ER) and progesterone receptor (PR) are positive in almost all neuroendocrine breast carcinomas, whereas HER2 immunostaining is negative in almost all cases.[29] The molecular subtype in this group of tumors is mostly the luminal phenotype.

This study aimed (i) to describe the cytological features of this group of tumors in the breast and (ii) to reveal the effect of the extent of neuroendocrine differentiation on the cytological features.

MATERIALS AND METHODS

Excision materials diagnosed as primary breast carcinoma between 2013 and 2017 were retrospectively scanned from the hospital electronic database. The study includes only primary breast carcinoma cases; metastatic lesions were excluded from the study. Cases showing 10% or more staining with neuroendocrine markers were selected for the study. Among these cases, those who underwent breast fine-needle aspiration biopsy or axillary lymph node aspiration biopsy and whose cytological material was suspicious for malignancy or diagnosed as malignant were included in the study. Patient age, demographic data, clinical characteristics, cytological and histological diagnoses were obtained from medical records and pathology reports.

Cases were grouped into two as cases with focal (>10% and <50% in tumor cells) staining with Synaptophysin or Chromogranin A and cases with diffuse (>50% in tumor cells) staining with Synaptophysin or Chromogranin A. A group of primary breast carcinoma cases without neuroendocrine features was used as control group. Immunohistochemically, all the cases in the control group were negative for Synaptophysin or Chromogranin A. Fine needle aspiration biopsy specimens of the primary breast carcinoma or metastatic axillary lymph nodes of these two groups and the control group were examined and compared.

Smears were evaluated for the following features: hypercellularity (hypocellular, moderately cellular, hypercellular), background features (clean, hemorrhagic, inflammatory), architecture (tight cohesive, loose cohesive, discohesive), isolated (single) cells, presence of naked nuclei, and binucleation, cell size (small, medium, large), cell shape (round, round-oval), nuclear location (central, eccentric), monotony, pleomorphism, chromatin distribution (fair, coarse), the presence of nucleoli and cytoplasmic granularity, and mitosis.

All of the immunohistochemical studies were applied in breast excision materials containing the tumor. ER, PR, Her2, Ki-67, Chromogranin A and Synaptophysin were applied using the avidin-biotin method (Synaptophysin [antibody polyclonal, dilution: 1:500, CellMarque Sigma-Aldrich Co, Rocklin, California] and Chromogranin A [antibody LK2H10, dilution: 1:500, CellMarque Sigma-Aldrich Co, Rocklin, California]). Ten-fifty percent staining in the tumor with at least one of Synaptophysin and Chromogranin A was classified as “Focal” and ≥50% staining was classified as “Diffuse”. Cases showing focal and diffuse staining were compared in terms of cytological features.

RESULTS

Clinical findings

Detailed clinicopathological information of all cases with neuroendocrine features (n = 20) is given in Table 1a and the control group (n = 15) in Table 1b.

Table 1a

The clinicopathological features of all cases with neuroendocrine features

Case no	Age	Tumor localization	Tumor size (cm)	FNA site	HG	NG	Axillary lymph nodes	Excision type	Synaptophysin	Chromogranin	ER	PR	cerbB2 (score)	Ki67 (%)	Molecular subtype	Months follow-up	Status at last follow-up
1	71	right	2,5	breast	2	3	0/15	BCS + axilla	focal positive	negative	90% +++	60% +++	2	40	luminal B	77	NED
2	86	left	1,5	breast	3	3	1/8	MRM	focal positive	negative	negative	negative	0	40	triple negative	73	NED
3	36	left	1,5	lymph node	2	2	2/22	BCS + axilla	focal positive	negative	90% +++	90% +++	0	15	luminal B	95	NED
4	58	left	2,5	breast	2	2	7/8	MRM	focal positive	negative	100% +++	100% +++	0	25	luminal B	82	AWD (liver metastasis)
5	69	right	2	lymph node	2	2	1/10	MRM	diffuse positive	negative	100% +++	100% +++	0	14	luminal A	53	NED
6	59	right	3	lymph node	2	2	6/13	BCS + axilla	diffuse positive	negative	60% ++	<10% +	0	25	luminal B	54	DOD (bone metastasis)
7	63	left	2,2	lymph node	2	2	6/11	BCS + axilla	focal positive	negative	100% +++	70% ++	0	16	luminal B	60	NED
8	69	right	2	breast	2	2	1/10	MRM	diffuse positive	negative	100% +++	100% +++	0	14	luminal A	62	NED
9	76	right	6	lymph node	2	2	2/4	MRM	diffuse positive	negative	100% +++	20% +++	0	38	luminal B	45	DOD (bone metastasis)
10	56	right	2,2	lymph node	2	2	12/19	BCS + axilla	focal positive	negative	90% +++	60% +++	0	13	luminal A	87	AWD (liver metastasis)
11	64	left	2	breast	2	2	0/1	BCS + SLNB	diffuse positive	negative	100% +++	80% +++	0	8	luminal A	65	NED
12	45	right	2,7	breast	2	2	0/1	mastectomy + SLNB	diffuse positive	negative	90% +++	10% +	3	70	luminal B	79	AWD (bone metastasis)
13	57	left	4,5	lymph node	1	2	0/9	mastectomy + SLNB	diffuse positive	negative	50% ++	90% +++	1	60	luminal B	76	NED
14	78	left	4,5	breast	2	2	0/1	BCS + SLNB	focal positive	diffuse positive	100% +++	100% +++	0	9	luminal A	51	NED
15	75	right	1	breast	1	1	0/1	BCS + SLNB	diffuse positive	diffuse positive	100% +++	100% +++	0	5	luminal A	40	NED
16	76	left	3,5	lymph node	3	3	2/23	MRM	focal positive	negative	90% +++	<10% ++	0	70	luminal B	87	NED
17	47	left	2,5	lymph node	2	2	1/11	BCS + axilla	diffuse positive	focal positive	70% ++	70% +++	0	13	luminal A	N/A	N/A
18	47	right	4	lymph node	2	2	1/11	BCS + axilla	diffuse positive	negative	95% +++	90% +++	0	10	luminal A	48	NED
19	62	left	2,2	breast	2	2	3/13	BCS + axilla	focal positive	negative	80% +++	1-5% +	0	35	luminal B	82	NED
20	32	right	1,5	breast	2	2	1/2	BCS + axilla	focal positive	negative	90% ++	10% ++	1	25	luminal B	76	NED

Table 1b

The clinicopathological features of all cases in the control group

Case no	Age	Tumor localization	Tumor size (cm)	FNA site	HG	NG	Axillary lymph nodes	Excision type	Tumor type	ER	PR	cerbB2 (score)	Ki67 (%)	Molecular subtype
1	77	right	6	lymph node	3	3	11/21	MRM	IDC	90% +++	10% ++	2 (SISH -)	38	luminal B
2	69	left	3	lymph node	2	2	13/13	MRM	IDC	60% ++	90% +++	1	20	luminal B
3	65	right	3,5	lymph node	3	3	5/21	BCS + axilla	IDC	100% +++	70% +++	0	40	luminal B
4	57	left	2,5	lymph node	2	2	24/24	BCS + axilla	IDC	100% +++	70% ++	0	18	luminal B
5	48	left	1,8	breast	2	2	0/1	BCS + SLNB	IDC	85% +++	70% +++	0	10	luminal A
6	64	right	2,8	breast	2	2	4/13	BCS + axilla	IDC	100% +++	5% +++	0	20	luminal B
7	72	left	1,4	breast	2	2	0/2	BCS + SLNB	IDC + mucinous ca	100% +++	100% +++	0	22	luminal B
8	40	left	1,5	breast	2	2	0/2	BCS + SLNB	IDC	100% +++	100% +++	1	48	luminal B
9	43	right	1,5	lymph node	2	2	3/22	mastectomy + axilla	IDC	100% +++	95% +++	2 (SISH -)	33	luminal B
10	54	right	3,6	lymph node	3	3	4/8	BCS + SLNB	IDC	100% +++	10% +++	0	80	luminal B
11	62	right	1,8	lymph node	2	2	4/21	mastectomy + axilla	IDC	98% +++	98% +++	0	22	luminal B
12	41	left	2	lymph node	2	2	2/13	MRM	IDC	100% +++	90% +++	0	28	luminal B
13	73	left	3,2	lymph node	2	2	4/7	BCS + axilla	IDC	100% +++	25% ++	0	15	luminal B
14	52	right	2,3	lymph node	1	2	1/15	mastectomy + axilla	IDC	95% +++	5% ++	1	25	luminal B
15	84	right	2,5	lymph node	3	2	2/6	MRM	IDC	negative	negative	1	80	triple negative

A total of 20 patients who had 10% or more staining with neuroendocrine markers on the excision materials and who were diagnosed as suspicious for malignancy or malignant on breast fine-needle aspiration biopsy or axillary lymph node aspiration biopsy were included in the study. All cases were female. Age ranged from 32 to 86 years (mean: 61.3, median: 62). Ten of the lesions were located on the right breast and 10 on the left breast. FNAB was performed from the breast mass in 10 patients and from the axillary lymph node in 10 patients. Among 20 patients, 19 had clinical follow-up, ranging between 40 and 95 months (mean: 68 months). In the diffuse staining group, seven patients showed no evidence of disease, whereas one patient was alive with disease and two patients were dead of disease. All of these three patients had bone metastasis. In the focal staining group, seven patients showed no evidence of disease and two patients were alive with disease, both with liver metastasis.

Pathological findings

Findings in excision specimens

Histological diagnoses were made on breast-conserving surgery material in 12 cases, modified radical mastectomy material in 6 cases, and simple mastectomy material in 2 cases. Tumor size ranged from 1 cm to 6 cm (average 2.7 cm).

Histological examination revealed 10 cases of invasive ductal carcinoma, 2 cases of neuroendocrine carcinoma of breast, 2 cases of invasive breast carcinoma with neuroendocrine differentiation, and 2 cases of mucinous carcinoma. The coexistence of invasive ductal and invasive micropapillary carcinoma was seen in 2 cases, the coexistence of invasive ductal and invasive papillary carcinoma was seen in 1 case, and invasive ductal and invasive mucinous carcinoma also in 1 case. The histological grade (Modified Bloom Richardson) was found to be Grade 1 in 1 case, Grade 2 in 16 cases, Grade 3 in 3 cases.

The number of axillary lymph nodes excised was between 1 and 23 (mean 9.6). Axillary lymph node metastasis was observed in 14 of 20 cases. The number of metastatic lymph nodes in these cases ranged from 1 to 12 (average 3.3).

Immunohistochemical studies applied to the excision materials are summarized in Table 1a. When the distribution of the cases according to molecular subtypes was examined, it was seen that eight cases had luminal A and 11 cases had luminal B phenotype. One case was triple-negative, but no staining was observed with basal-type immune markers. The Ki-67 proliferation index averaged 27.3% (range: 5%–70%). The cut-off value for the Ki-67 proliferation index for molecular subtyping was accepted as 14%.[30] It was determined that luminal A subtype was observed more frequently (7/11) in the group staining diffusely with neuroendocrine markers, and luminal B type more frequent (7/9) in the group with focal staining.

Synaptophysin showed positivity in all cases; it was stained diffuse (≥50%) in 10 cases and focal (10%–50%) in 10 cases. Chromogranin A showed diffuse staining in two cases and focal staining in one case; it was negative in all of the other cases.

Findings in fine-needle aspiration biopsies

All cytological findings and the comparison of these findings between the two groups are shown in Table 2.

Table 2

The cytological findings observed in FNAB materials of the cases with neuroendocrine features and the control group; and the distribution of these findings in groups with diffuse and focal neuroendocrine expression

Cytological features	Cases with diffuse neuroendocrine expression (n=11)	Cases with focal neuroendocrine expression (n=9)	Total (n=20)	Cases without neuroendocrine expression (control group) (n=15)
Cellularity
Hypercellular	5 (45%)	3 (33%)	8 (40%)	7 (47%)
Moderately cellular	4 (36%)	4 (44%)	8 (40%)	7 (47%)
Hypocellular	2 (18%)	2 (22%)	4 (20%)	1 (6%)
Background
Clean	6 (55%)	2 (22%)	8 (40%)	14 (93%)
Inflammatory	1 (9%)	4 (44%)	5 (25%)	0 (0%)
Hemorrhagic	4 (36%)	3 (33%)	7 (35%)	1 (7%)
Architecture
Tight cohesive groups	3 (27%)	7 (78%)	9 (45%)	11 (73%)
Loose cohesive groups	7 (64%)	1 (11%)	9 (45%)	4 (27%)
Cohesion loss	1 (9%)	1 (11%)	2 (10%)	0 (0%)
Isolated cells	8 (73%)	8 (89%)	16 (80%)	7 (47%)
Naked nuclei	8 (73%)	4 (44%)	12 (60%)	3 (20%)
Cell size
Large	0 (0%)	2 (22%)	2 (10%)	4 (27%)
Medium-large	4 (36%)	4 (44%)	8 (40%)	10 (67%)
Medium	7 (64%)	3 (33%)	10 (50%)	1 (6%)
Pleomorphism	2 (18%)	9 (100%)	11 (55%)	14 (93%)
Nuclear shape
Round-oval	7 (64%)	7 (78%)	14 (70%)	15 (100%)
Round	4 (36%)	2 (22%)	6 (30%)	0 (0%)
Binucleation	2 (18%)	2 (22%)	4 (20%)	0 (0%)
Chromatin
Coarse	1 (9%)	1 (11%)	2 (10%)	8 (53%)
Fine	10 (91%)	8 (89%)	18 (90%)	7 (47%)
Prominent nucleoli	1 (9%)	3 (33%)	4 (20%)	6 (40%)
Cytoplasmic granularity	2 (18%)	0 (0%)	2 (10%)	0 (0%)
Necrosis	0 (0%)	1 (11%)	1 (5%)	0 (0%)
Mitosis	0 (0%)	1 (11%)	1 (5%)	1 (7%)

All lesions were sampled with a 22-gauge needle. Ten of the cytological diagnoses were given by FNAB material taken from the breast, and 10 cases were given by FNAB of the lymph nodes. All of the cytological samples were reported as malignant cytology or suspicious for malignancy (5 cases suspicious for malignancy, 15 cases were malignant). The diagnosis of “suspicious for malignancy” was made due to hipocellularity. Preparations fixed with 96% alcohol were stained with Papanicolaou stain (PAP), and air-dried smears were stained with May–Grunwald–Giemsa stain (MGG).

Smears of breast and axillary lymph node FNABs were evaluated in terms of cellularity, background features, architecture, isolated cell presence, presence of naked nucleus and binucleation, cell size, cell shape, nucleus location, monotony, pleomorphism, chromatin distribution, presence of nucleolus, presence of cytoplasmic granularity, and mitosis.

In our study, 40% of the cases had smears with clean background, whereas 35% had hemorrhagic and 25% had inflammatory background. The ratio of clean background was more common in the group with diffuse staining (diffuse group 6/11, focal group 2/9). In terms of cellularity, 40% of the cases were hypercellular, 40% were moderately cellular, and 20% were hypocellular. Cellularity was higher in the group with diffuse staining (d5/11, f3/9). In half of the cases, loose cohesive groups of tumor cells were seen. Tight cohesive cell groups were more frequent in the focal staining group (d3/11, f7/9) [Figure 1a–b], and loose cohesive cell groups were more frequent in the diffuse staining group (d7/11, f1/9) [Figure 1c]. In 60%, naked nuclei were present. Bare nuclei were evaluated in preparations in which the nuclei were intact and did not contain pressure-related crush artifacts. Ratio of naked nuclei was higher in the diffuse staining group (d8/11, f4/9) [Figure 1d]. Medium-sized cells were seen in 10 cases (50%), medium-large-sized cells in 8 cases (40%), and large cells in 2 cases (10%). The proportion of large cells was higher in the group with focal staining (d0/11, f2/9) [Figure 2a–c]. Pleomorphism was evident in tumor cells in all cases in the focal stained group (f9/9) [Figure 2a–d], and in 2 cases in the diffuse stained group (d2/11). Binucleation was slightly more common in the focal stained group (d2/11 f2/9) [Figure 1d and Figure 3d]. Six cases had a distinct round nucleus, whereas 14 cases had a round-oval nucleus. Round-oval nuclei were dominant in both groups. The nucleus was located eccentrically in all of our cases [Figure 3a–3b]. Fine chromatin was observed in 18 cases (90%), whereas coarse chromatin was present in 2 cases (10%). The fine chromatin ratio was similar in both groups (d10/11, f8/9). Prominent nucleoli were noted in 4 cases (f3/9 d1/11) [Figure 2d]. Presence of granularity was seen in only 2 cases, both of which were in the group showing diffuse staining with neuroendocrine markers (d2/11, f0/9) [Figure 3b–Figure 3c]. Necrosis and mitosis were observed in one case each, both in tumors showing focal staining with neuroendocrine markers.

Figure 1

(a) Focal staining group- Cohesive groups of tumor cells on the inflammatory background (MGGx200) (b) Focal staining group- Cohesive groups of tumor cells with marked pleomorphism on the inflammatory background (MGGx400) (c) Diffuse staining group- Hypercellular smear with clean background and single tumor cells showing loss of cohesion (MGGx400) (d) Diffuse staining group- Monotonous tumor cells with round nuclei, occasional binucleation and naked nuclei (MGGx1000)

Figure 2

(a) Focal staining group- Cohesive groups of medium-large tumor cells with marked pleomorphism on the inflammatory background (MGGx400) (b and c) Focal staining group- Cohesive groups of medium-large tumor cells with marked pleomorphism (MGGx1000) (d) Focal staining group- Tumor cells with marked pleomorphism and prominent nucleoli, naked nuclei (arrow) and accompanying inflammatory cells (MGGx1000)

Figure 3

(a) Diffuse staining group- Tumor cells with eccentric nuclei and cohesion loss (PAPx200) (b) Diffuse staining group- Tumor cells with eccentric nuclei and marked cohesion loss seen as single cells, some of them showing cytoplasmic granularity (arrow) (PAPx400) (c) Diffuse staining group- Monotonous tumor cells with round nuclei and rare cytoplasmic granules (MGGx1000) (d) Diffuse staining group- Monotonous tumor cells with round nuclei and marked binucleation (MGGx1000)

When cytological features of the cases with neuroendocrine features are compared with the control group, similar rates of cellularity are noticed. Inflammatory and hemorrhagic background is more common in cases with neuroendocrine features than the control group. Architecturally, cases in the control group showed higher rates of tight cohesive groups, whereas cohesion loss was more common in the cases with neuroendocrine features. Isolated cells and naked nuclei were more frequent in the neuroendocrine group. Regarding cellular features, cells in the control group had larger size, more pleomorphism, and prominent nucleoli than neuroendocrine group. Chromatin tends to be fine in cases with neuroendocrine features, whereas coarse chromatin was commonly seen in the control group.

DISCUSSION

In this study, we aimed to reveal how the extent of NE differentiation in tumor tissue is reflected in cytology samples by reviewing the FNAB samples of breast carcinomas with neuroendocrine features. Hence, the cases were divided into two groups according to the NE differentiation extent in breast excision materials detected by immunohistochemical staining (Synaptophysin and/or Chromogranin A), those with 10-50% NE differentiation in tumor tissue (focal) and those with ≥50% NE differentiation (diffuse). Cytological findings of these groups were compared.

Forty percent of the cases in our study had smears with clean background, the ratio of clean background was more common in the group with diffuse staining. As the extent of neuroendocrine differentiation decreased, inflammation and hemorrhage increased. Kawasaki et al.[31] detected clean background in 59% of the 26 FNAB samples in their in-situ neuroendocrine ductal carcinoma series consisting of 32 cases. Similarly, in Thiha et al.'s[32] series of seven cases of breast carcinoma with neuroendocrine features, the background was clean in four cases. In the study by Kawanishi et al.,[33] hemorrhagic background was observed in the smears of a solid neuroendocrine carcinoma case.

In our study, 40% of the cases were hypercellular, 40% were moderately cellular, and 20% were hypocellular. Cellularity was found to be higher in the group with diffuse staining. This finding was consistent with most of the literature. High cellularity was reported in 95% of the cases by Kawasaki et al.[31] in their series. However, in Ohashi et al.'s[34] series of mucinous breast carcinoma with neuroendocrine differentiation consisting of 37 cases, there was no significant difference in cellularity between the group with a high rate of neuroendocrine differentiation (staining ≥15% with neuroendocrine markers) and the group with a low rate of neuroendocrine differentiation (staining <15% with neuroendocrine markers). In different case series and case reports, cellularity is reported to be high or moderate.[5323536373839]

In half of our cases, the tumor cells consisted of loose cohesive groups. Although tight cohesive cell groups were more frequent in the focal staining group, loose cohesive cell groups were more frequent in the diffuse staining group. One case in both groups showed significant cohesion loss. There were isolated cells in the background in 80% of the cases. Isolated cells were observed at a similar rate in both groups, but slightly more frequently in the focal group. In the study by Tang et al.,[39] significant cohesion loss and single-cell pattern were observed in 12 of 135 malignant breast cytology materials, and 5 of these cases were found to be positive with neuroendocrine markers. It was concluded that loss of cellular cohesion combined with low cytological atypia is characteristic of neuroendocrine breast carcinoma. Ohashi et al.[34] observed moderate-to-severe cohesion loss in 94.4% of mucinous carcinomas with a high rate of neuroendocrine differentiation, whereas they did not detect significant discohesion in the group with low neuroendocrine differentiation. In different case series and case reports, it was observed that cohesion loss was significant in cases with neuroendocrine differentiation.[31353637] However, in Thiha et al.'s[32] series of seven cases of neuroendocrine breast carcinoma, four cases had cohesive groups, and three cases had loose cohesive groups. Cohesion loss also can be seen in different types of breast tumors other than neuroendocrine tumors, such as lobular breast carcinoma. The minimal rate of anisonucleosis also may cause lobular breast carcinoma to be confused with neuroendocrine breast tumors.[38] Neuroendocrine markers can help differentiate these two tumors.[35] Also, the loss of e-cadherin observed in lobular carcinoma is rarely seen in neuroendocrine breast carcinomas. In the series by Tang et al.,[39] e-cadherin was found to be positive in 11 of 12 cases with cohesion loss, and e-cadherin loss was not observed in any of the five cases staining with neuroendocrine markers.[40] Similarly, loss of cohesion and single-cell pattern can be seen in lymphomas, but nuclear irregularity and coarse granular chromatin structure are more prominent in lymphomas.[38]

In 60% of our cases, naked nuclei were present in the background. Ng et al.[35] stated that although naked nuclei are common in neuroendocrine tumors of other organs and systems, it is rare in breast neuroendocrine tumors, and they attributed this to the more intact cell cytoplasm in breast-derived neuroendocrine tumors. In our case series, however, a significant number of naked nuclei were observed in the majority of cases, especially in the diffuse staining group.

In the evaluation of tumor cell sizes, medium-sized cells were seen in 10 cases (50%), medium-large-sized cells in 8 cases (40%), and large cells in 2 cases (10%). The proportion of large cells was higher in the group with focal staining, and it was not seen in the group with diffuse staining. Although most of the group with diffuse staining consisted of monotonous cells, focal staining group did not have a monotone cell population. Pleomorphism was evident in tumor cells in all cases in the focal stained group. Binucleation, which was observed in 20% of all cases, was observed to be at a similar rate in both groups. Although these findings are compatible with some of the existing literature,[3340] in the case series of Thiha et al.,[32] small cell size was observed in 6 of 7 cases and pleomorphism was not detected. In the study of Ohashi et al.,[34] no significant nuclear size difference was observed between the group with high neuroendocrine differentiation and the group with low neuroendocrine differentiation. However, large cells with an atypical, round-elongated nucleus and the presence of molding should be warning for poorly differentiated breast neuroendocrine carcinoma.[37]

In the evaluation of nuclear shape, 6 cases had a distinct round nucleus, whereas 14 cases had a round-oval nucleus. Although round-oval nuclei were dominant in both groups, round nuclei were observed to be slightly more frequent in the diffuse stained group. Ohashi et al.[34] also found that the ratio of round nuclei was higher in patients with high neuroendocrine differentiation (77.8%) and that the ratio of oval nuclei was higher (57.8%) in patients with low neuroendocrine differentiation. The nucleus was located eccentrically in all of our cases. This feature, which gives the cell a plasmacytoid appearance, has been reported as a frequently observed finding in neuroendocrine breast tumors.[333540] In addition to the round-oval nuclear structure, spindle-like cells also can be seen. It has been reported that the presence of spindle and polygonal cells is an important finding in the diagnosis of in-situ neuroendocrine ductal carcinomas.[31]

In our study, fine chromatin structure was observed in 18 cases (90%), whereas coarse chromatin structure was present in 2 cases (10%). The fine chromatin ratio was similar in both groups. The rate of nucleolar prominence was higher in the group showing focal staining with neuroendocrine markers. These findings were consistent with the literature.[31323335]

Considering the cytoplasmic features, the presence of granularity was seen in only 2 cases, both of which were in diffuse staining group. Thiha et al.[32] detected cytoplasmic granules in all cases in their series. Some studies have reported that cytoplasmic granules can be evident in some of the cases.[3540] In the study by Kawanishi et al.,[33] however, cytoplasmic granularity was not detected. Similarly, Ohashi et al.[34] reported that cytoplasmic granules are not evident in breast carcinoma cases with high and low rates of neuroendocrine differentiation. The fact that cytoplasmic granularity is common in neuroendocrine tumors originating from other organs, but not observed in cases of breast origin, was attributed to the accumulation of intracytoplasmic mucin.[34] As our study did not consist of mucinous carcinoma cases alone, it is not possible to explain the occurrence of granularity in a small number of cases only by intracytoplasmic mucin accumulation.

In our study, necrosis and mitosis were observed in one case each, both in the focal staining group. Rare mitosis has been reported in a small number of cases in the literature, and necrosis is not an expected finding.[313235]

In the comparison of the neuroendocrine group and the control group in terms of cytological features, it was noticed that inflammatory and hemorrhagic background, cohesion loss in tumoral cells, isolated cells and naked nuclei are more common in cases with neuroendocrine features than the control group. Cells in the nonneuroendocrine cases had larger size, more pleomorphism, more frequent coarse chromatin and prominent nucleoli than neuroendocrine cases.

The cytological features of breast neuroendocrine tumors have been described in various studies and case reports up to now. The common features of these tumors defined in these studies can be listed as: cellular smears with small groups of cells showing loose cohesion and single-cell pattern, monomorphic plasmacytoid appearance, eccentrically located nucleus, fine and/or granular chromatin, inconspicuous nucleoli, and granular cytoplasm.[33383941]

In this study, it was observed that clean background, hypercellularity, loss of cohesion, naked nuclei and monotone-medium sized cells with round nucleus, granular cytoplasm, features which are expected to be seen in neuroendocrine breast tumors, were more prominent in cases showing diffuse staining with Synaptophysin and Chromogranin A. As most of the studies on these tumors are based on the WHO 2003 classification, in other words, tumors staining >50% with neuroendocrine markers are included in these studies, it is not surprising to see the features reported in the previous studies in our diffuse staining group. The cytological features observed in the cases in this group reflect the neuroendocrine characteristics of the tumor. In the following WHO 2012 classification, tumors with low neuroendocrine characteristics also were classified as neuroendocrine tumor class. These tumors show less of the cytological features expected in neuroendocrine tumors. Therefore, cytologically, it may be more difficult to suspect neuroendocrine differentiation in tumors that show focal staining with neuroendocrine markers.

In regards of cytological-histological correlation of the cases in this study; there was no discrepancy between cytological and histological diagnoses, as all of the study cases were selected from cases with a cytological diagnosis of 'malignant' or 'suspicious for malignancy'. However, we re-examined the slides of seven cases diagnosed as 'suspicious for malignancy' and tried to explain why these cases remained at the suspicious category. Five of these cases were cytological samples taken from lymph nodes and all had low number of tumoral cells. The other two cases were cytological samples taken from the breast and it was thought that the diagnosis of definitive malignancy was avoided because tumor cells had low level of nuclear atypia.

The limitations of our study are small number of cases and retrospective nature of the study. Larger studies on this subject provided with prognostic data, which may prove importance in the follow-up and treatment of the patients with this tumor, may be useful.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.