Effusion cytology of metastatic carcinosarcoma

INTRODUCTION

Carcinosarcomas (CSs) are aggressive, biphasic tumors demonstrating malignant epithelial/ carcinomatous and mesenchymal/sarcomatous components.^[1] These occur mainly in the elderly age group.^[2,3] These tumors commonly involve the head-and-neck region, respiratory tract, female genital tract, and rarely the gastrointestinal tract.^[4] Within the spectrum of gynecological malignancies, uterine CS constitutes less than 5% of all gynecological carcinomas.^[5] The carcinomatous component comprises endometrioid or serous carcinoma, clear cell carcinoma, or undifferentiated carcinoma. The sarcomatous component is usually a high-grade sarcoma, not otherwise specified. This sarcomatous component is thought to be derived from the carcinomatous component through epithelial-mesenchymal transition.^[1]

Effusion cytology forms an essential aspect of the evaluation of gynecological neoplasms. There are conflicting data regarding the exact role of effusion cytology in CS and other gynecological malignancies. Some studies mention that positive cytology correlates with a poorer prognosis, while others do not substantiate this.^[6] Effusions, when present, are always sampled for cytological evaluation, and the results are recommended to be reported.^[7] There are limited data on the effusion cytology of CS in particular. An unequivocal metastasis to the serous cavities denotes malignant effusion (category 5) as per the recent international system for reporting serous fluid cytopathology.^[8] Most cases of CS on cytology are reported as metastatic carcinoma, and the actual diagnosis is almost always made on the surgical specimens. This study evaluates the cytomorphological features of CS in effusion samples and their diagnostic pitfalls.

MATERIAL AND METHODS

The present study is retrospective and ethically approved by the Departmental Ethical Committee. The proper consent was taken from the patients. The Helsinki accord and strict ethics were followed. The study was conducted in the Department of Cytology and Gynecological Pathology, Post Graduate Institute of Medical Education and Research, Chandigarh. All histopathologically proven cases of gynecological CS from January 2015 to December 2021 were retrieved from the archives. Among these, cases of CS presenting with effusion were selected.

For each effusion sample, a minimum of two smears, one air-dried conventional smear stained by May–Grünwald– Giemsa stain and other Papanicolaou stained liquid-based preparation (SurePath^™), were prepared. These effusion cytology smears were then reviewed (RK, MS, and PD), and those demonstrating metastatic malignancies were selected. The cytomorphological features were then analyzed in light of the primary histopathological diagnosis of CS. The features included cellularity, cytoarchitectural pattern, presence of epithelial and mesenchymal cells, cellular pleomorphism, cytoplasmic and nuclear features of malignant cells, and the background. The histopathology slides provided information on the presence and relative proportion of epithelial and mesenchymal components, presence of heterologous elements, lymphovascular space invasion, myometrial invasion, cervical involvement, and the status of the omentum and lymph nodes.

RESULTS

A total of 20,986 effusion fluid samples were analyzed during the said period, out of which 2240 cases were reported as malignant. Among all the malignant effusion cases, only 10 cases (0.45%) were of gynecological CS. Overall, 112 cases of gynecological CS were retrieved. Fifty-four (48.2%) cases had corresponding effusion samples available. Of these 54 cases, the 10 cases with malignant effusions constituted 18.5% of gynecological CS with effusions. Considering total 112 points of gynecological CS, these comprised 8.9%. The age of the patients ranged from 20 to 74 years (mean: 55.3 years). The most common site of origin for CS was the uterus (7/10, 70% of cases) followed by the ovary (3/10, 30% of cases) [Table 1]. Nine (90%) of these were pre-operative samples, and diagnosis on the primary surgical specimen was rendered afterward. The post-operative sample was submitted for cytology only in 1 case (10%).

Cytomorphological findings

The smears were cellular in seven cases, paracellular in two cases, and highly cellular in one [Table 2]. Cells were arranged predominantly in clusters (9/10, 90% of cases; all cases except case 2). In addition, ball-like clusters were found in 4/10, 40% of cases (cases 4, 7, 8, and 10) and singly scattered cells were present in 3/10 (30%) cases (cases 1, 5, and 9) [Figure 1a and b]. An occasional case showed discrete spindle cells [Figure 1c]. In most cases, 7/10 (70%) (cases 1, 3, 4, 5, 7, 8, and 10), the tumor cells were moderately pleomorphic, while they were mildly pleomorphic in 2/10 (20%) (cases 2 and 6) and markedly pleomorphic with bizarre forms in 1/10 (10%) (case 9). The cytoplasm was scant to moderate with vacuolations in 2/10 (20%) (cases 8 and 9) [Figure 1d]. Nuclear chromatin was coarse in 9/10 (90%) (all cases except case 8) and vesicular in 1/10 (10%) (case 8). Nucleoli were conspicuous in all cases except one. The background consisted mostly of mesothelial cells, inflammatory cells, and blood. Based on these cytomorphological features, 7/10 (70%) cases were diagnosed as metastatic adenocarcinoma, while the rest were reported as positive for malignancy. None of the cases was diagnosed as CS. Cell blocks were available in 3/10 cases. The sections from the cells block showed multiple clusters and gland-like arrangements [Figure 2a and b]. The cells had vacuolated cytoplasm and moderately pleomorphic nuclei. The tumor cells were positive for vimentin and cytokeratin (CK) [Figure 2c and d].

Table 2: Cytomorphological features of the effusion samples in patients with primary carcinosarcoma of FGT (n=10).

Case No.	Cellularity	Cytoarchitectural pattern	Cellular pleomorphism	Amount of cytoplasm	Nuclear characteristics		Background	Cytological diagnosis
					Chromatin	Nucleoli
1	Cellular	Clusters, singly scattered	Moderate	Moderate	Coarse	Inconspicuous	Inflammatory cells, mesothelial cells, blood	Positive for malignant cells
2	Pauci- cellular	Scattered	Mild	Scant	Coarse	Conspicuous	Inflammatory cells, mesothelial cells	Metastatic adenocarcinoma
3	Cellular	Clusters	Moderate	Moderate	Coarse	Conspicuous	Inflammatory cells, mesothelial cells	Metastatic adenocarcinoma
4	Cellular	Ball-like clusters	Moderate	Scant	Coarse	Conspicuous	Inflammatory cells, mesothelial cells	Metastatic adenocarcinoma
5	Cellular	Clusters, singly scattered	Moderate	Scant to moderate	Coarse	Conspicuous	Mesothelial cells, blood	Metastatic adenocarcinoma
6	Pauci cellular	Clusters	Mild	Moderate	Coarse	Conspicuous	Mesothelial cells, blood	Positive for malignant cells
7	Highly cellular	Clusters, ball-like clusters	Moderate	Moderate	Coarse	Conspicuous	Inflammatory cells, mesothelial cells, blood	Metastatic adenocarcinoma
8	Cellular	Clusters, ball-like clusters	Moderate	Moderate, vacuolated	Vesicular	Conspicuous	Mesothelial cells, blood	Metastatic adenocarcinoma
9	Cellular	Clusters, singly scattered	Marked; Bizarre cells	Moderate, vacuolated	Coarse	Conspicuous	Blood	Positive for malignant cells
10	Cellular	Clusters, ball-like clusters	Moderate	Moderate	Coarse	Conspicuous	Inflammatory cells	Metastatic adenocarcinoma

FGT: Female genital tract

Close

Figure 1:

(a) Multiple tight ball-like clusters of cells (Papanicolaou stain ×200). (b) Cluster of tumor cells with background mesothelial cells and lymphocytes (May–Grünwald–Giemsa stain ×400). (c) Occasional scattered spindle cells present (see arrow) (May– Grünwald–Giemsa stain ×400). (d) Individual cells show moderate nuclear enlargement and pleomorphism (May–Grünwald–Giemsa stain ×400).

Export to PPT

Close

Figure 2:

(a) multiple gland-like structures (Hematoxylin and eosin stain ×200), (b) higher magnification shows the glands lined by moderately pleomorphic nuclei. (Hematoxylin and eosin stain ×400), (c) histopathology section showing strong cytoplasmic positivity of vimentin (×200), and (d) membranous positivity of cytokeratin (×400).

Export to PPT

Histopathological findings

The pertinent histopathological findings are shown in [Table 3]. There were both adenocarcinoma and sarcoma components [Figure 3a and b]. Most of the cases were diagnosed on morphology, and immunohistochemistry (IHC) was performed in one case. In this, the carcinomatous component was a high-grade serous carcinoma, staining positive for CK, epithelial membrane antigen, PAX-8, and p53 (mutant). The sarcomatous component stained positive for vimentin and smooth muscle actin. The microsatellite instability panel was negative by IHC.

Close

Figure 3:

(a) Histopathology section of the uterine carcinosarcoma shows both adenocarcinoma and adjacent sarcomatous area (hematoxylin and eosin stain ×200). Insert showing the higher magnification figure of the rhabdomyoblastic differentiation, (b) higher magnification of the sarcomatous area shows moderately pleomorphic spindle cells (hematoxylin and eosin stain ×400). Insert showing the higher magnification of the pleomorphism of the spindle cells.

Export to PPT

Table 3: Histopathological details of the carcinosarcomas of the female genital tract with malignant effusion.

Case No.	Specimen	Primary site	Tumor size (cm)	Heterologous elements	Myometrial invasion	LVSI	Status of additional structures
							Other FGT structures	Omentum	Lymph node	Bowel/others
1	TAH+ BSO+ Om+ App	Ovary	14×8 × 8 (multiple fragments)	Absent	NA	Absent	TAH-uninvolved, Fallopian tubes- involved	Involved	NA	Appendix-uninvolved
2	EB	Uterus	NA	Absent	NA	NA	NA	NA	NA	NA
3	TAH+ BSO+ Om+ PLN	Uterus	5×5 × 4	Absent	>50%	Present	Bilateral ovaries and fallopian tubes-uninvolved Cervix-involved	Uninvolved	Uninvolved	NA
4	TAH+ BSO+ Om	Uterus	5×5 × 3	Chondrosarcomatous differentiation	>50%	Present	Bilateral ovaries and fallopian tubes - involved	Involved	NA	NA
5	TAH+ BSO+ Om	Uterus	10×9 × 5	Rhabdomyoblastic differentiation	>50%	Present	Cervix-uninvolved, Bilateral tubo-ovarian masses-involved	Involved	NA	NA
6	TAH+ BSO+ Om+ REILN	Uterus	6.5×6	Absent	>50%	Present	Bilateral ovaries and fallopian tubes-uninvolved	Uninvolved	Involved	NA
7	TAH+ BSO+ Om	Ovary	15.5×13×11.5 and 10×10×5.4 (Bilateral ovaries)	Rhabdomyoblastic differentiation	NA	Present	TAH and right fallopian tube- uninvolved, left fallopian tube- involved	Involved	NA	NA
8	TAH+ BSO+ Om+ Segment of LI+ CC	Ovary	10×8.5×3.5 and 5×3 × 1.5 (Bilateral ovaries)	Chondrosarcomatous and rhabdomyoblastic differentiation	NA	Absent	TAH-uninvolved, fallopian tubes- involved	Involved	NA	Serosal involvement of bowel, CC-uninvolved
9	TAH+ BS+ Om biopsy+ CC	Uterus	10	Rhabdomyoblastic differentiation	>50%	Present	Cervix and bilateral fallopian tubes-uninvolved	Uninvolved	NA	CC-uninvolved
10	TAH+ BSO+ Om	Uterus	10×7 × 3	Chondrosarcomatous differentiation	>50%	Not seen	Bilateral ovaries- involved, fallopian tubes-uninvolved.	Involved	NA	NA

App: Appendicectomy, BSO: Bilateral salpingo-oophorectomy, Bx: Biopsy, CC: Cholecystectomy, EB: Endometrial biopsy, LI: Large intestine, NA: Not available, Om: Omentectomy, PLN: Pelvic lymphadenectomy, REILN: Right external iliac lymph nodes, TAH: Total abdominal hysterectomy, FGT: Female genital tract, Om: Omentum

DISCUSSION

In the present study, we described the cytomorphological features of 10 cases of CS in effusion samples. The cases showed adenocarcinoma components predominantly. The exact diagnosis of CS was not possible due to the absence of recognizable sarcomatous component.

CSs, previously known as malignant mixed Müllerian tumors, are thought to originate from the cells of epithelial origin, which undergo mesenchymal differentiation.^[9] The epithelial/carcinomatous component is usually composed of endometrioid adenocarcinoma.^[10] The mesenchymal/sarcomatous component can be classified as homologous or heterologous.^[5] The heterologous elements, seen in almost 40% of the uterine CS, correlate with poor survival compared to those with homologous elements.^[1] The most commonly encountered heterologous elements are rhabdomyosarcoma and chondrosarcoma, followed by osteosarcoma, liposarcoma, and angiosarcoma.^[11] In the present study, on histopathology, rhabdomyosarcomatous and chondrosarcomatous areas were noted in more than half of the cases. However, these components were not apparent in cytology. The inability to appreciate heterologous elements in cytology has been documented.^[6,10,12-14]

The incidence of malignant effusion in CS is highly variable. Lew et al.^[15] estimated a low incidence of malignant peritoneal fluid (0.7%) and malignant pleural effusion (0.2%) across all cases of malignant effusions. On the other hand, nearly 50% of patients with gynecological CS with malignant effusions diagnosed on cytology have been reported.^[10] In the present study, a small proportion (8.9%) of all CS patients had malignant effusion. There are no precise data regarding the sensitivity and specificity of cytomorphological evaluation of effusions in CS.

Few studies have described the cytology of effusion samples in CS.^{[6,10,13,16,17]} Smears are usually highly cellular.^[6] An ideal smear should show carcinoma and sarcoma’s typical biphasic cell pattern.^[16] However, mostly the carcinomatous component is readily identified, represented almost exclusively as adenocarcinoma, and only a minority of effusion samples show additional sarcomatous component.^[10,18,19] The sarcomatous cells usually appear more degenerated, arranged as loose aggregates with many dissociated cells, which could easily be mistaken for inflammatory cells. Atypical sarcomatous cells with oval to spindly stromal configuration with hyperchromatic to pyknotic nuclei may be seen.^[20] Motoyama and Watanabe^[21] found both adenocarcinomatous and sarcomatous cells in the ascitic fluid samples, with some of the sarcomatous cells seen as giant cells. Cytopathologists face many difficulties diagnosing such cases due to their overall rarity.^[6] The predominance of the carcinomatous component in positive effusion samples has been attributed to poor cohesion and increased shedding of the cells into fluid samples. On the other hand, the sarcomatous components are either very minimal or too obscured and inconspicuous to be readily identified, especially when the epithelial component is overwhelming. In such scenarios, it is often tough to detect the sarcomatous component unless there are frank heterologous mesenchymal elements.^[10,12]

Immunocytochemistry (ICC) on cell block is an excellent way to overcome confounding factors such as inflammatory and mesothelial cells populating the smears in malignant effusions.^[22-25] Identification of a “second-foreign” population of cells other than mesothelial and inflammatory cells is quintessential for diagnosing metastatic effusions. Subtractive coordinate immunoreactivity pattern approach on serial cell block sections is recommended for objectivity in the interpretation of the immunoreactivity pattern of different immunocytochemical markers in the same cells.^[26] A proposed algorithmic approach for serous effusion cytology is depicted in [Figure 4]. Applying ICC on cell block could confirm the two coexistent cell populations in CS. The carcinomatous component is highlighted by AE1/AE3, BerEp4, and CD-15, while the sarcomatous component is lit up by desmin, HHF-35, S-100, and SMA.^[10,19] TP53 mutation analysis has also been attempted in such metastatic effusion samples; however, the results have not been promising.^[27]

Close

Figure 4:

Algorithmic approach of carcinosarcoma in effusion.

Export to PPT

There is a need to emphasize effusion cytology in CS as a positive peritoneal fluid cytology at an early stage of the disease is predictive of early intraperitoneal dissemination and a poorer prognosis.^[6] The follow-up histomorphological correlation almost always reveals some preserved mesenchymal/sarcomatous elements likely to have been missed in the initial cytological assessment calling for a stringent evaluation of the effusion samples. The availability of cell block and ICC makes possible the assessment of architectural patterns and exact categorization of the elements with the identification of typical heterologous elements.

CONCLUSION

Malignant effusion is a rarity in CSs. These effusion samples are often erroneously reported due to the failure to delineate the different elements on morphology alone. Knowledge of subtle morphological features, awareness of the possible diagnostic pitfalls, and ICC can improve diagnostic accuracy.

AUTHORS’ CONTRIBUTIONS

Reetu Kundu and Malvika Shastri: Collection of data, analyzing and drafting of the manuscript

Parikshaa Gupta, Nalini Gupta, and Radhika Srinivasan: Reporting the cases

Pranab Dey: Conception of the idea, analyzing and drafting of the manuscript, and execution of the study.

DATA SHARING

The authors refuse to share the data of the patients for privacy issues.

ETHICAL APPROVAL

The present study is retrospective and ethically approved by the Departmental Ethical committee. The proper consent was taken from the patients. The Helsinki accord and strict ethics were followed.