Introduction: Signet ring cells (SRCs) may be observed in carcinomas from multiple primary sites. Elucidating unknown primaries from metastases with SRCs represents a diagnostic challenge. This study examined morphologic characteristics of adenocarcinomas with SRCs from stablished primary sites and described objective features, which can aid in identifying the site of origin. Methods: The series encompasses 257 cases of adenocarcinomas with SRCs from gastroesophageal junction (GEJ, n = 38), stomach (n = 48), pancreatobiliary system (n = 16), colorectum (n = 40), appendix (n = 32), breast (n = 41), and lung (n = 42). H&E sections were examined and scored using architectural and cytologic criteria. Morphometric analysis was performed using QuPath software. Results: Extracellular mucin was more abundant in GEJ, colorectal, and appendiceal carcinomas. Poorly cohesive morphology was the most frequent pattern in gastric and breast carcinomas. The cytoplasmic mucin/vacuole was predominantly clear and targetoid in breast carcinomas. Breast and gastric carcinomas showed the highest nuclear to cytoplasmic (N/C) ratio, whereas appendiceal carcinoma the lowest. Conclusion: Morphological evaluation (extracellular mucin, architectural patterns, and the nature of cytoplasmic mucin/vacuole) represents an important step to determine the cancer site of origin in adenocarcinomas with SRCs and guides further ancillary studies. Cytological morphometry may help further refine morphological criteria and facilitate the construction of digital-pathology algorithms.

Signet ring cell (SRC) morphology is a distinct histopathological feature, which can be observed not only in malignancies from the gastrointestinal tract and pancreatobiliary (PB) system but also in lung and breast cancers, as well as tumours from the urinary system and female and male genital organs. The established definition of an SRC carcinoma is based on the presence of an eccentrically placed nucleus and an abundant intracellular mucin or non-mucinous vacuole, pushing the nucleus to the cell periphery [1].

Tumours showing a predominant SRC phenotype can be categorized as SRC carcinomas. The current 5th edition of WHO classification of tumours of the digestive system [2] has recognized SRC carcinoma as a distinct histologic subtype for carcinomas of the stomach, gastroesophageal junction (GEJ), small intestine and ampulla, pancreas, gallbladder and extrahepatic bile ducts (within the broader category of poorly cohesive carcinomas), as well as for appendiceal and colorectal cancers. From the clinical standpoint, it is essential for the pathologist to recognize and report the presence of SRC features, as SRC carcinomas represent a subset of tumours with a distinct biological behaviour, prone to peritoneal spreading pattern [3] and poor prognosis [4‒7], especially for GEJ, gastric, and pancreatic primaries [8].

In the clinicopathological practice, elucidating unknown primary sites from metastases with SRC morphology represents a diagnostic challenge. Data from the literature underline the importance of ancillary studies, namely, immunohistochemistry [9] and molecular tests [10], as a critical component for accurate diagnosis. However, few studies have focused attention on the strength of morphology and histopathological criteria as the first step of workup that can provide preliminary clues and help in refining the ancillary studies for the identification of primary tumours.

In this study, we analysed the histologic patterns and the cytologic characteristics of adenocarcinomas with SRC features from established primary sites. The aim was to identify objective criteria, which can aid in identifying the primary site of malignancies with SRC morphology.

Case Series

Two hundred and fifty-seven cases of primary (n = 226) and metastatic (n = 31) adenocarcinomas with SRC features were retrospectively identified and chosen by random selection from the database of the Department of Pathology of Memorial Sloan Kettering Cancer Center (MSKCC) between 2017 and 2019 with the approval of Institution Research Board (approval number #16-1683A[3]). The selection of cases was based merely on the presence of SRCs, irrespective of the SRC percentage; that is, SRC carcinomas and carcinoma containing SRCs were both included in the series. The primary sites included GEJ (n = 38), stomach (n = 48), PB system (n = 16), colon/rectum (n = 40), appendix (n = 32), breast (n = 41), and lung (n = 42). Cases in which the primary site of origin was not well established were excluded (Table 1). Both biopsies (n = 149, 58.0%) and surgical specimens (n = 108, 42.0%) were included in the series and chosen by random selection. Patients submitted to neoadjuvant therapy were excluded.

Table 1.

Case series description according to tumour origin and primary or metastatic site

Tumour originPrimary tumours, N (%)Metastases, N (%)Total, N (%)
GEJ 36 (15.9) 2 (6.5) 38 (14.8) 
Stomach 47 (20.8) 1 (3.2) 48 (18.7) 
Pancreas/bile ducts 8 (3.5) 8 (25.8) 16 (6.2) 
Colon/rectum 38 (16.8) 2 (6.5) 40 (15.6) 
Appendix 24 (10.6) 8 (25.8) 32 (12.5) 
Breast 32 (14.2) 9 (29) 41 (16.0) 
Lung 41 (18.1) 1 (3.2) 42 (16.3) 
Total 226 31 257 
Tumour originPrimary tumours, N (%)Metastases, N (%)Total, N (%)
GEJ 36 (15.9) 2 (6.5) 38 (14.8) 
Stomach 47 (20.8) 1 (3.2) 48 (18.7) 
Pancreas/bile ducts 8 (3.5) 8 (25.8) 16 (6.2) 
Colon/rectum 38 (16.8) 2 (6.5) 40 (15.6) 
Appendix 24 (10.6) 8 (25.8) 32 (12.5) 
Breast 32 (14.2) 9 (29) 41 (16.0) 
Lung 41 (18.1) 1 (3.2) 42 (16.3) 
Total 226 31 257 

Histopathologic and Cytologic Analysis

One H&E section from each tumour was selected. In heterogenous tumours, the H&E section with the most variable morphology/SRC types was considered. Slides were examined by two pathologists, independently (WD, IG). The histological pattern was determined by analysing two representative fields at ×100 magnification, while the cytologic characteristics and morphometric data were obtained from a field at ×200 magnification. Both pathologists evaluated all the cases. When evaluation was discordant, cases were evaluated a second time and consensus was reached.

Architectural criteria for the analysis included (1) extracellular mucin (present vs. absent); (2) desmoplastic reaction (scant intercellular tumour stroma vs. prominent desmoplasia); (3) cellular cohesion (cohesive vs. poorly cohesive); (4) glandular/acinar architecture (present vs. absent); (5) nested architecture (present vs. absent), defined as the presence of well-defined clusters of three or more cells, with delineated boundaries and separated by stroma or vasculature; and (6) sheeted/solid tumour growth pattern (present vs. absent), when cells were organized in larger and irregular sheets (e.g., >200 cells), with ill-defined boundaries and confluent growth. Moreover, the predominant architectural pattern was chosen for each tumour.

The cytologic analysis included the nature of the cytoplasmic vacuoles: (1) vacuoles with a vesicular mucinous appearance, i.e., multiple mucin vesicles within the cytoplasm (“intestinal type,” “goblet type,” “bluish”); (2) clear (“empty”/white) cytoplasmic vacuoles; and (3) vacuoles with a targetoid appearance (punctuate and centrally located eosinophilic material within a clear or mucinous cytoplasm). Representative examples of architectural patterns and cytological characteristics are depicted in the figures shown in the Results section.

Morphometric Analysis

Morphometric analysis was performed in eight cases for each primary site (GEJ, stomach, PB system, colon/rectum, appendix, breast, lung) for a total of 56 cases. For each tumour, a *.tiff image representative of a field at ×200 magnification was obtained. Each image sized 1920 × 989 pixels and presented a resolution of 96 dots per inch.

Morphometric analysis was performed using the QuPath Open Source Software for digital pathology image analysis [11]. For each image, 50 cells with well-preserved nuclear and cytoplasmic contours were selected. Cells showing mitotic figures were excluded. Accordingly, 400 cells were analysed for each primary site and a totality of 2,800 cells were included in the morphometric analysis.

For each cell, the nuclear and cytoplasmic contours were manually delineated using the QuPath polygon tool. For each annotation obtained, the following data were measured: (1) nucleus area (pixel number); (2) cytoplasmic area (pixel number); (3) nuclear/cytoplasmic (N/C) ratio. A representative example of morphometric analysis is shown in Figure 1.

Fig. 1.

Morphometric analysis was performed selecting 50 cells for each image (top) and delineating the cytoplasmic (red) and nuclear (green) contours (bottom), to define the cytoplasmic area, nuclear area, and N/C ratio.

Fig. 1.

Morphometric analysis was performed selecting 50 cells for each image (top) and delineating the cytoplasmic (red) and nuclear (green) contours (bottom), to define the cytoplasmic area, nuclear area, and N/C ratio.

Close modal

Statistical Analysis

IBM SPSS (Release 20.0.0) and GraphPad Prism 8 were used for statistical analysis. All tests were two sided, and differences were considered significant when p ≤ 0.05. Comparisons of categorical variables (histopathologic and cytologic analysis) were performed by χ2 test or Fisher’s exact test, as appropriate. For multiple categorical variables, post hoc test was used by applying the Bonferroni correction. Comparisons of quantitative variables (morphometric analysis) from multiple groups were performed by Welch and Brown-Forsythe ANOVA as equal variances were not assumed by Levene’s test. The Games Howell post hoc test was applied.

Carcinomas with SRCs from different primary sites showed distinctive architectural patterns and cytologic features. Significant higher frequency of cases with more abundant extracellular mucin was identified in carcinomas with SRCs from GEJ (n = 26/38; 68.4%), colon/rectum (n = 26/40; 65.0%), and appendix (n = 28/32; 87.5%) (Table 2; Fig. 2). By contrast, a lower frequency of cases with less abundant extracellular mucin characterized carcinomas with SRCs from stomach (n = 4/48; 8.3%) and breast (n = 1/41; 2.4%) (Table 2). Breast and lung carcinomas with SRCs exhibited more frequently prominent desmoplasia in the tumour stroma (n = 33/41; 80.5% and n = 33/42; 78.6%, respectively), when compared with cancer from other sites (Table 2; Fig. 2).

Table 2.

Histopathologic patterns and cytologic features of carcinomas with SRC features from different primary sites

Histopathologic patterns and cytologic detailsGEJ, N (%)Stomach, N (%)PB system, N (%)Colon/rectum, N (%)Appendix, N (%)Breast, N (%)Lung, N (%)Total, N (%)
Extracellular mucin 
 Absent 12 (31.6) 44 (91.7) 10 (62.5) 14 (35.0) 4 (12.5) 40 (97.6) 34 (81.0) 158 (61.5) 
 Present 26 (68.4) 4 (8.3) 6 (37.5) 26 (65.0) 28 (87.5) 1 (2.4) 8 (19.0) 99 (38.5) 
p value 0.00004* 0.00000* 0.93091 0.00018* 0.00000* 0.00000* 0.00458 
Desmoplastic reaction 
 Scant 21 (55.3) 26 (54.2) 3 (18.8) 26 (65.0) 21 (65.6) 8 (19.5) 9 (21.4) 114 (44.4) 
 Prominent 17 (44.7) 22 (45.8) 13 (81.2) 14 (35.0) 11 (34.4) 33 (80.5) 33 (78.6) 143 (55.6) 
p value 0.14270 0.12931 0.03324 0.00424 0.00965 0.00048* 0.00107* 
Cohesion 
 Poorly cohesive 22 (57.9) 37 (77.1) 9 (56.2) 20 (50.0) 9 (28.1) 39 (95.1) 4 (9.5) 141 (54.5) 
 Cohesive 16 (42.1) 11 (22.9) 7 (43.8) 20 (50.0) 23 (71.9) 2 (4.9) 38 (90.5) 117 (45.5) 
p value 0.25746 0.00000* 0.57235 0.93596 0.01004 0.00000* 0.00000* 
Glandular/acinar architecture 
 Absent 18 (47.4) 43 (89.6) 7 (43.8) 19 (47.5) 11 (34.4) 38 (92.7) 5 (11.9) 141 (54.9) 
 Present 20 (52.6) 5 (10.4) 9 (56.2) 21 (52.5) 21 (65.6) 3 (7.3) 37 (88.1) 116 (45.1) 
p value 0.28965 0.00000* 0.3562 0.28319 0.01113 0.00000* 0.00000* 
Tumour cells arranged in nests 
 Absent 15 (39.5) 35 (72.9) 5 (31.2) 14 (35.0) 9 (28.1) 17 (41.5) 17 (40.5) 112 (43.6) 
 Present 23 (60.5) 13 (27.1) 11 (68.8) 26 (65.0) 23 (71.9) 24 (58.5) 25 (59.5) 145 (56.4) 
p value 0.5803 0.00001* 0.30437 0.32741 0.08870 0.64462 0.82424 
Tumour cells arranged in sheets or solid architecture 
 Absent 29 (76.3) 35 (72.9) 9 (56.2) 23 (57.5) 17 (53.1) 29 (70.7) 17 (40.5) 159 (61.9) 
 Present 9 (23.7) 13 (27.1) 7 (43.8) 17 (42.5) 15 (46.9) 12 (29.3) 25 (59.5) 98 (38.1) 
p value 0.05915 0.10285 0.58497 0.46406 0.23408 0.20241 0.00117* 
Predominant architecture 
 Glands 13 (34.2) 1 (2.1) 8 (50.0) 13 (32.5) 11 (34.4) 0 (0.0) 26 (61.9) 72 (28.0) 
p value 0.35694 0.00001* 0.04316 0.49188 0.39191 0.00001* 0.00000* 
 Poorly cohesive 14 (31.6) 42 (87.5) 2 (12.5) 9 (22.5) 9 (28.1) 34 (82.9) 0 (0.0) 87 (33.9) 
p value 0.42122 0.00000* 0.01142 0.00474 0.07293 0.00000* 0.00000* 
 Nested 6 (15.8) 0 (0.0) 1 (6.2) 7 (17.5) 4 (12.5) 4 (9.8) 7 (16.7) 52 (20.2) 
p value 0.34165 0.00614 0.51105 0.17631 0.81627 0.73590 0.22805 
 Solid/sheets 5 (13.2) 5 (10.4) 5 (31.2) 11 (27.5) 8 (25) 3 (7.3) 9 (21.4) 46 (17.9) 
p value 0.40888 0.13374 0.15025 0.08473 0.26274 0.05386 0.51413 
Vesicular (mucinous) cytoplasmic vacuoles 
 Absent 1 (2.6) 0 (0.0) 1 (6.2) 1 (2.5) 0 (0.0) 36 (87.8) 2 (4.8) 41 (16.0) 
 Present 37 (97.4) 48 (100) 15 (93.8) 39 (97.5) 32 (100.0) 5 (12.2) 40 (95.2) 216 (84.0) 
p value 0.01640 0.00097* 0.27133 0.01242 0.00932 0.00000* 0.0000* 
Clear cytoplasmic vacuole 
 Absent 13 (34.2) 23 (47.9) 2 (12.5) 22 (55.0) 22 (68.8) 4 (9.8) 20 (47.6) 106 (41.2) 
 Present 25 (65.8) 25 (52.1) 14 (87.5) 18 (45.0) 10 (31.2) 37 (90.2) 22 (52.4) 151 (58.8) 
p value 0.76418 0.00000* 0.08913 0.00097* 0.00000* 0.00000* 0.02145 
Targetoid cytoplasmic vacuole 
 Absent 24 (63.2) 28 (58.3) 11 (68.8) 35 (87.5) 30 (93.8) 13 (31.7) 37 (88.1) 178 (69.3) 
 Present 14 (36.8) 20 (41.7) 5 (31.2) 5 (12.5) 2 (6.2) 28 (68.3) 5 (11.9) 79 (30.7) 
p value 0.76418 0.00000* 0.08913 0.00097* 0.00000* 0.0000* 0.02145 
Variability of vacuole quality 
 Variable (all vacuoles) 9 (23.7) 18 (37.5) 5 (31.2) 3 (7.5) 0 (0.0) 22 (53.7) 2 (4.8) 59 (23.0) 
 One/two vacuole types 29 (76.3) 30 (62.5) 11 (68.8) 37 (92.5) 32 (100.0) 19 (46.3) 40 (95.2) 198 (77.0) 
p value 0.01242 0.00014* 0.31731 0.13361 0.02781 0.04550 0.10960 
Histopathologic patterns and cytologic detailsGEJ, N (%)Stomach, N (%)PB system, N (%)Colon/rectum, N (%)Appendix, N (%)Breast, N (%)Lung, N (%)Total, N (%)
Extracellular mucin 
 Absent 12 (31.6) 44 (91.7) 10 (62.5) 14 (35.0) 4 (12.5) 40 (97.6) 34 (81.0) 158 (61.5) 
 Present 26 (68.4) 4 (8.3) 6 (37.5) 26 (65.0) 28 (87.5) 1 (2.4) 8 (19.0) 99 (38.5) 
p value 0.00004* 0.00000* 0.93091 0.00018* 0.00000* 0.00000* 0.00458 
Desmoplastic reaction 
 Scant 21 (55.3) 26 (54.2) 3 (18.8) 26 (65.0) 21 (65.6) 8 (19.5) 9 (21.4) 114 (44.4) 
 Prominent 17 (44.7) 22 (45.8) 13 (81.2) 14 (35.0) 11 (34.4) 33 (80.5) 33 (78.6) 143 (55.6) 
p value 0.14270 0.12931 0.03324 0.00424 0.00965 0.00048* 0.00107* 
Cohesion 
 Poorly cohesive 22 (57.9) 37 (77.1) 9 (56.2) 20 (50.0) 9 (28.1) 39 (95.1) 4 (9.5) 141 (54.5) 
 Cohesive 16 (42.1) 11 (22.9) 7 (43.8) 20 (50.0) 23 (71.9) 2 (4.9) 38 (90.5) 117 (45.5) 
p value 0.25746 0.00000* 0.57235 0.93596 0.01004 0.00000* 0.00000* 
Glandular/acinar architecture 
 Absent 18 (47.4) 43 (89.6) 7 (43.8) 19 (47.5) 11 (34.4) 38 (92.7) 5 (11.9) 141 (54.9) 
 Present 20 (52.6) 5 (10.4) 9 (56.2) 21 (52.5) 21 (65.6) 3 (7.3) 37 (88.1) 116 (45.1) 
p value 0.28965 0.00000* 0.3562 0.28319 0.01113 0.00000* 0.00000* 
Tumour cells arranged in nests 
 Absent 15 (39.5) 35 (72.9) 5 (31.2) 14 (35.0) 9 (28.1) 17 (41.5) 17 (40.5) 112 (43.6) 
 Present 23 (60.5) 13 (27.1) 11 (68.8) 26 (65.0) 23 (71.9) 24 (58.5) 25 (59.5) 145 (56.4) 
p value 0.5803 0.00001* 0.30437 0.32741 0.08870 0.64462 0.82424 
Tumour cells arranged in sheets or solid architecture 
 Absent 29 (76.3) 35 (72.9) 9 (56.2) 23 (57.5) 17 (53.1) 29 (70.7) 17 (40.5) 159 (61.9) 
 Present 9 (23.7) 13 (27.1) 7 (43.8) 17 (42.5) 15 (46.9) 12 (29.3) 25 (59.5) 98 (38.1) 
p value 0.05915 0.10285 0.58497 0.46406 0.23408 0.20241 0.00117* 
Predominant architecture 
 Glands 13 (34.2) 1 (2.1) 8 (50.0) 13 (32.5) 11 (34.4) 0 (0.0) 26 (61.9) 72 (28.0) 
p value 0.35694 0.00001* 0.04316 0.49188 0.39191 0.00001* 0.00000* 
 Poorly cohesive 14 (31.6) 42 (87.5) 2 (12.5) 9 (22.5) 9 (28.1) 34 (82.9) 0 (0.0) 87 (33.9) 
p value 0.42122 0.00000* 0.01142 0.00474 0.07293 0.00000* 0.00000* 
 Nested 6 (15.8) 0 (0.0) 1 (6.2) 7 (17.5) 4 (12.5) 4 (9.8) 7 (16.7) 52 (20.2) 
p value 0.34165 0.00614 0.51105 0.17631 0.81627 0.73590 0.22805 
 Solid/sheets 5 (13.2) 5 (10.4) 5 (31.2) 11 (27.5) 8 (25) 3 (7.3) 9 (21.4) 46 (17.9) 
p value 0.40888 0.13374 0.15025 0.08473 0.26274 0.05386 0.51413 
Vesicular (mucinous) cytoplasmic vacuoles 
 Absent 1 (2.6) 0 (0.0) 1 (6.2) 1 (2.5) 0 (0.0) 36 (87.8) 2 (4.8) 41 (16.0) 
 Present 37 (97.4) 48 (100) 15 (93.8) 39 (97.5) 32 (100.0) 5 (12.2) 40 (95.2) 216 (84.0) 
p value 0.01640 0.00097* 0.27133 0.01242 0.00932 0.00000* 0.0000* 
Clear cytoplasmic vacuole 
 Absent 13 (34.2) 23 (47.9) 2 (12.5) 22 (55.0) 22 (68.8) 4 (9.8) 20 (47.6) 106 (41.2) 
 Present 25 (65.8) 25 (52.1) 14 (87.5) 18 (45.0) 10 (31.2) 37 (90.2) 22 (52.4) 151 (58.8) 
p value 0.76418 0.00000* 0.08913 0.00097* 0.00000* 0.00000* 0.02145 
Targetoid cytoplasmic vacuole 
 Absent 24 (63.2) 28 (58.3) 11 (68.8) 35 (87.5) 30 (93.8) 13 (31.7) 37 (88.1) 178 (69.3) 
 Present 14 (36.8) 20 (41.7) 5 (31.2) 5 (12.5) 2 (6.2) 28 (68.3) 5 (11.9) 79 (30.7) 
p value 0.76418 0.00000* 0.08913 0.00097* 0.00000* 0.0000* 0.02145 
Variability of vacuole quality 
 Variable (all vacuoles) 9 (23.7) 18 (37.5) 5 (31.2) 3 (7.5) 0 (0.0) 22 (53.7) 2 (4.8) 59 (23.0) 
 One/two vacuole types 29 (76.3) 30 (62.5) 11 (68.8) 37 (92.5) 32 (100.0) 19 (46.3) 40 (95.2) 198 (77.0) 
p value 0.01242 0.00014* 0.31731 0.13361 0.02781 0.04550 0.10960 

GEJ, gastroesophageal junction; PB, pancreatobiliary.

*Significant p value (after Bonferroni correction).

Fig. 2.

Upper panel: Higher amounts of extracellular mucin were found in gastroesophageal, colorectal, and appendiceal carcinomas with SRCs. Middle panel: Poorly cohesive phenotype characterized tumours of breast and gastric origin, whereas lung carcinoma showed more frequently cellular cohesion. Lower panel: Targetoid and clear cytoplasmic vacuoles were more frequent in breast carcinomas, whereas gastric adenocarcinomas showed more frequently a mucinous and clear phenotype, with less prominent targetoid mucin. Appendiceal carcinomas with SRCs (as well as colorectal adenocarcinomas) showed mucinous cytoplasmic vacuoles.

Fig. 2.

Upper panel: Higher amounts of extracellular mucin were found in gastroesophageal, colorectal, and appendiceal carcinomas with SRCs. Middle panel: Poorly cohesive phenotype characterized tumours of breast and gastric origin, whereas lung carcinoma showed more frequently cellular cohesion. Lower panel: Targetoid and clear cytoplasmic vacuoles were more frequent in breast carcinomas, whereas gastric adenocarcinomas showed more frequently a mucinous and clear phenotype, with less prominent targetoid mucin. Appendiceal carcinomas with SRCs (as well as colorectal adenocarcinomas) showed mucinous cytoplasmic vacuoles.

Close modal

Poorly cohesive architectural pattern was the characteristic phenotype for gastric (n = 37/48; 77.1%) and breast (n = 39/41; 95.1%) carcinoma with SRCs (Table 2). Accordingly, these two tumour types were arranged less frequently in glandular/acinar structures (n = 5/48; 10.4% and n = 3/41; 7.3%, respectively). By contrast, cellular cohesion was present in almost the totality of lung adenocarcinomas with SRCs (n = 38/42; 90.5%) with glandular/acinar structures (n = 37/42; 88.1%) and solid sheets (n = 25/42; 59.5%) constituting the predominant architectural phenotypes in this cancer type (Table 2).

Regarding the nature of the cytoplasmic mucin vacuole constituting the SRCs, all gastric adenocarcinomas showed cytoplasmic vesicular mucinous vacuoles (n = 48/48; 100%) combined with less commonly clear (n = 25/48; 52.1%) or targetoid appearance (n = 20/48; 41.7%). In contrast, the targetoid nature of the cytoplasmic vacuole was more frequent in breast cancers (n = 28/41; 68.3%), along with a clear phenotype (n = 37/41; 90.2%) and less prominent vesicular mucinous appearance (n = 5/41; 12.2%).

Colorectal and appendiceal adenocarcinomas presented goblet-type vesicular mucinous vacuoles in most of the cases (n = 39/40; 97.5% and 32/32; 100%, respectively), while clear (n = 18/40; 45% and n = 10/32; 31.2%, respectively) and targetoid (n = 5/40; 12.5% and n = 2/32; 6.2%) cytoplasmic vacuoles were less prominent. When primary tumours and distant metastases were compared, histopathologic patterns and cytologic features showed no statistically significant differences (p > 0.05).

Carcinomas with SRCs from Different Primary Sites Showed Distinctive Morphometric Data

The morphometric analysis of nuclear and cytoplasmic areas from different sites of tumour origin, performed manually by a pathologist (IG) with a dedicated digital software, revealed statistically significant differences and is graphically represented in Figure 3. Carcinomas with SRCs of appendiceal origin were found to exhibit the highest cytoplasmic area, the lowest nuclear area, along with the lowest N/C ratio (p < 0.001). By contrast, carcinomas with SRCs of gastric origin were found to exhibit the lowest cytoplasmic area and the highest N/C ratio (p < 0.001).

Fig. 3.

Data from morphometric analysis, including the comparison of cytoplasmic area (a), nuclear area (b), and N/C ratio (c). SRCs of appendiceal origin (d) showed the highest cytoplasmic area, the lowest nuclear area, and the lowest N/C ratio, while SRCs from gastric primaries (e) showed the lowest cytoplasmic area and the highest N/C ratio.

Fig. 3.

Data from morphometric analysis, including the comparison of cytoplasmic area (a), nuclear area (b), and N/C ratio (c). SRCs of appendiceal origin (d) showed the highest cytoplasmic area, the lowest nuclear area, and the lowest N/C ratio, while SRCs from gastric primaries (e) showed the lowest cytoplasmic area and the highest N/C ratio.

Close modal

Appendiceal and PB tumours clustered together as the primary sites with higher amount of cytoplasm, while tumours from the GEJ, colon/rectum, breast, and stomach clustered together as the primary sites with the lowest amount of cytoplasm (p < 0.001). PB and lung primaries showed the largest nuclear areas, while appendiceal and colorectal tumours the lowest (p < 0.001).

When compared to carcinomas of appendiceal origin, colorectal cancers with SRCs presented higher nuclear areas, lower cytoplasmic areas, and higher N/C ratio. When compared to carcinomas of gastric and GEJ origin, PB carcinomas with SRCs showed higher cytoplasmic areas, higher nuclear areas, and lower N/C ratio. Finally, no differences between gastric and breast primaries with SRCs were identified.

In this study, we demonstrated the differential histologic, cytologic, and morphometric characteristics of adenocarcinomas with SRCs at different primary sites, namely, GEJ, stomach, PB system, colon, appendix, breast, and lung. Data from the literature have underlined the importance of ancillary tests, especially immunohistochemistry, in distinguishing the cancer site of origin [9, 12, 13]. Several immunohistochemical biomarkers have been described as useful discriminating tools in carcinomas with SRC features, including expression of mucins (MUC2, MUC5AC, MUC6) and transcription factors (CDX2, TTF1). The advantage and practicality that high-throughput molecular platforms may provide in this context is still debated. Indeed, distinct molecular profiles may characterize carcinomas from the same primary site, as demonstrated by Kwon et al. [14] in a series of gastric cancers with heterogeneous characteristics of poorly cohesive pattern and SRCs. Despite the recent developments in diagnostic tools, approximately 2–5% of metastatic carcinomas remain as tumours of unknown origin and the most common subtype of carcinoma of unknown primary is adenocarcinoma [15].

In contrast to the findings in the body of literature, in this study we have stressed the importance of morphologic evaluation as an initial and critical tool, which may be used to guide further ancillary studies. Histopathological features helpful in discriminating cancer origin include specific architectural patterns (i.e., poorly cohesive phenotype, glandular, nested or solid architecture), the presence (or absence) of extracellular mucin, and the nature of cytoplasmic mucin/vacuoles (i.e., vesicular mucinous, clear, or “targetoid”).

The analysis of cytologic morphometric data (nuclear area, cytoplasmic area, N/C ratio) may help to refine the recognition of cancer site of origin. The findings of this study revealed that carcinomas with SRCs of appendiceal origin show highest volume of cytoplasmic and lowest nuclear areas (thus the lowest N/C ratio), which distinguished them from SRC carcinomas of colorectal origin, with higher N/C ratio. SRC carcinomas of gastric origin exhibited highest N/C ratio. In addition, we found that carcinomas with SRCs of pancreatic origin exhibited the highest both cytoplasmic and nuclear areas, thus placing them in the lower N/C ratio group. We also observed that the distinction between gastric and breast carcinoma with SRCs was challenging on the morphometric basis alone since both tumours exhibited high N/C ratio. However, the nature of cytoplasmic vacuoles (i.e., vesicular mucinous for gastric origin and clear and targetoid for breast origin, respectively) can facilitate the distinction between the two primaries.

Given the changing in clinicopathological characteristics of gastric cancer in the West, with a decreasing incidence of distal, intestinal-type tumours and the corresponding increasing proportion of tumours with Laurén diffuse or WHO poorly cohesive including SRC histology, Mariette et al. [1] involved a multidisciplinary expert team of the European Chapter of the International Gastric Cancer Association. This research team pinpointed 11 topics on pathological classifications used for poorly cohesive and SRC gastric carcinoma. Each topic was debated followed by the elaboration of consensus statements, which contributed to the most recent WHO edition and the standardization for the definition of gastric SRC carcinomas, including poorly cohesive carcinomas with over 90% poorly cohesive cells exhibiting SRC morphology being categorized as SRCs. Other SRC types were further subdivided [1]. The present study contributes to a further stratification of SRCs with their heterogeneity and characteristics at different primary sites and thereby could potentially minimize the window of carcinomas of unknown origin.

We acknowledge that the morphologic criteria described in this study have some overlap features between carcinomas with SRCs from different primary sites. Thus, in isolation, they should not be considered as pathognomonic for a particular tumour entity. The accurate pathologic interpretation requires combined clinical and pathologic evaluation along with the necessary ancillary studies.

In conclusion, carcinomas with SRC features at different primary sites have recognizable histologic and morphometric features. While immunohistochemistry and molecular tests can facilitate the establishment of the primary site, in this study we stressed the importance of morphologic evaluation as an initial and critical tool to guide further ancillary studies. Histopathological features helpful in discriminating cancer origin include architectural patterns, the presence (or absence) of extracellular mucin, and the nature of cytoplasmic vacuoles/mucin. The analysis of cytologic morphometry may further help in refining the recognition of cancer site of origin. The establishment of conventional histologic criteria can facilitate the learning process for pathology trainees as well as precede forthcoming applications of artificial intelligence in pathology. Although performing a manual selection of nuclear and cytoplasmic areas on daily-based cases is time consuming and not reliable, the construction of digital-pathology algorithms could be feasible in the near future.

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions. Ethics approval was obtained. This study protocol was reviewed and approved by Memorial Sloan Kettering’s Institutional Review Board (approval number #16-1683A[3]). Considered the retrospective nature of the study, the authors were not expected to have consent to participate and written informed consent was not obtained from the participants (Memorial Sloan Kettering’s Institutional Review Board (approval number #16-1683A[3]). No potentially identifiable patient images or data were included in this study.

The authors have no conflicts of interest to declare.

This work has been supported in part by the NCI/NIH Cancer Center Support Grant P30 CA008748 and Mushett Family Foundation to LHT.

Wissam Dahoud and Irene Gullo are first co-authors of this work. Wissam Dahoud carried out the database search and case selection, performed the histopathologic and cytologic analysis, participated in the concept, design, and planning of the study, and wrote the manuscript. Irene Gullo performed the histopathologic and cytologic analysis and the morphometric analysis, carried out the data analysis, participated in the concept, design, and planning of the study, and co-wrote the manuscript. Rami Imam participated in manuscript preparation and review. Laura H. Tang participated in the concept, design, and planning of the study and co-wrote and reviewed the manuscript.

Additional Information

Wissam Dahoud and Irene Gullo contributed equally to this work.

All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author.

1.
Mariette
C
,
Carneiro
F
,
Grabsch
HI
,
van der Post
RS
,
Allum
W
,
de Manzoni
G
.
Consensus on the pathological definition and classification of poorly cohesive gastric carcinoma
.
Gastric Cancer
;
2018
.
2.
WHO classification of tumours
.
Digestive system tumours
. 4h ed.
International Agency for Research on Cancer
;
2019
.
3.
Solomon
D
,
DeNicola
N
,
Feingold
D
,
Liu
PH
,
Aycart
S
,
Golas
BJ
, et al
.
Signet ring cell features with peritoneal carcinomatosis in patients undergoing cytoreductive surgery and hyperthermic intraperitoneal chemotherapy are associated with poor overall survival
.
J Surg Oncol
.
2019
;
119
(
6
):
758
65
. .
4.
Sheng
H
,
Wei
X
,
Mao
M
,
He
J
,
Luo
T
,
Lu
S
, et al
.
Adenocarcinoma with mixed subtypes is a rare but aggressive histologic subtype in colorectal cancer
.
BMC Cancer
.
2019
;
19
(
1
):
1071
. .
5.
Remo
A
,
Fassan
M
,
Vanoli
A
,
Bonetti
LR
,
Barresi
V
,
Tatangelo
F
, et al
.
Morphology and molecular features of rare colorectal carcinoma histotypes
.
Cancers
.
2019
;
11
(
7
):
1036
. .
6.
Tuncel
D
,
Basturk
O
,
Bradley
KT
,
Kim
GE
,
Xue
Y
,
Reid
MD
, et al
.
Poorly cohesive (signet ring cell) carcinoma of the ampulla of vater
.
Int J Surg Pathol
.
2019
;
28
(
3
):
236
44
. .
7.
Legué
LM
,
van Erning
FN
,
Creemers
GJ
,
de Hingh
IHJT
,
Lemmens
VEPP
,
Huysentruyt
CJ
.
The prognostic relevance of histologic subtype in appendiceal adenocarcinoma
.
Eur J Surg Oncol
.
2019
;
46
(
3
):
433
8
. .
8.
Li
H
,
Zong
Z
,
Zhou
T
,
Sun
L
,
Wang
A
,
Zhang
K
, et al
.
Trends of incidence and survival in patients with gastroenteropancreatic signet ring cell carcinoma: an analysis from the Surveillance, Epidemiology, and End Results program
.
J Gastrointest Oncol
.
2019
;
10
(
5
):
979
88
. .
9.
Terada
T
.
An immunohistochemical study of primary signet-ring cell carcinoma of the stomach and colorectum: II. Expression of MUC1, MUC2, MUC5AC, and MUC6 in normal mucosa and in 42 cases
.
Int J Clin Exp Pathol
.
2013
;
6
(
4
):
613
21
.
10.
Moran
S
,
Martínez-Cardús
A
,
Sayols
S
,
Musulén
E
,
Balañá
C
,
Estival-Gonzalez
A
, et al
.
Epigenetic profiling to classify cancer of unknown primary: a multicentre, retrospective analysis
.
Lancet Oncol
.
2016
;
17
(
10
):
1386
95
. .
11.
Bankhead
P
,
Loughrey
MB
,
Fernández
JA
,
Dombrowski
Y
,
McArt
DG
,
Dunne
PD
, et al
.
QuPath: Open source software for digital pathology image analysis
.
Sci Rep
.
2017
;
7
(
1
):
16878
. .
12.
Pozos-Ochoa
LI
,
Lino-Silva
LS
,
Leon-Takahashi
AM
,
Salcedo-Hernandez
RA
.
Prognosis of signet ring cell carcinoma of the colon and rectum and their distinction of mucinous adenocarcinoma with signet ring cells. A comparative study
.
Pathol Oncol Res
.
2018
;
24
(
3
):
609
16
. .
13.
Kerckhoffs
KGP
,
Liu
DHW
,
Saragoni
L
,
van der Post
RS
,
Langer
R
,
Bencivenga
M
, et al
.
Mucin expression in gastric- and gastro-oesophageal signet-ring cell cancer: results from a comprehensive literature review and a large cohort study of Caucasian and Asian gastric cancer
.
Gastric Cancer
.
2020
;
23
(
5
):
765
79
. .
14.
Kwon
CH
,
Kim
YK
,
Lee
S
,
Kim
A
,
Park
HJ
,
Choi
Y
, et al
.
Gastric poorly cohesive carcinoma: a correlative study of mutational signatures and prognostic significance based on histopathological subtypes
.
Histopathology
.
2018
;
72
(
4
):
556
68
. .
15.
Handa
H
,
Gomi
D
,
Fukushima
T
,
Kobayashi
T
,
Sekiguchi
N
,
Sakamoto
A
, et al
.
Signet ring cell carcinoma of unknown primary origin detected incidentally by lymph node purification for thyroid carcinoma
.
Intern Med
.
2018
;
57
(
1
):
31
5
. .