Utility of Scrape Cytology in Management of Ovarian Neoplasms: a Cross-sectional Study in a Tertiary Care Hospital of Eastern India

Abstract

Ovarian tumors are one of the most common gynecological neoplasms worldwide affecting women of all ages. As preoperative diagnosis is often inaccurate, intraoperative diagnostic modalities play indispensable role in management of ovarian tumors to avoid unnecessary extensive surgery. Although reliable, frozen section is highly expensive for resource-poor developing countries. Intraoperative cytological methods can help to overcome these limitations. The aim of this study was to evaluate the accuracy of scrape cytology in diagnosis of ovarian tumors. The freshly removed specimens of ovarian tumors sent without any fixative during the study period were included in this observational prospective cross-sectional study. Scraping was done from the most representative area of the cut surface of the tumor. Cytological smears were stained by hematoxylin–eosin method. All the cytological smears and corresponding histological sections were evaluated separately. Cytological interpretation was then compared with histological diagnosis to assess the diagnostic accuracy of scrape cytology. On histology, 66 cases were diagnosed as benign (60%), 5 cases as borderline (4.55%), and 39 as malignant (35.45%). Of these, 64 cases (58%) were of epithelial origin, 25 cases (22.7%) were of germ cell origin and 6 (5.45%) cases were categorized as sex cord-stromal tumor. Accuracy of scrape cytology in diagnosis of benign, borderline, and malignant ovarian tumors was found to be 95.45%, 40%, and 89.74%, respectively. Overall accuracy rate was 90.91%. Scrape cytology is a rapid, inexpensive yet effective tool to diagnose benign and malignant ovarian tumors with high diagnostic accuracy.

Introduction

Ovarian tumors are one of the most common gynecological neoplasms worldwide affecting women of all ages [1]. They are of diverse origin including epithelial, germ cell, and sex cord-stromal. Ovarian cancer is the eighth most common cause of malignancy diagnosed in women leading to 4.4% of all cancer deaths in females [2]. Incidence of ovarian cancer varies in different parts of the world, and maximum number of cases are detected in the developed countries like Eastern and central Europe (> 8 per 100,000), and the lowest rate is noted in Asia and Africa (≤ 3 per 100,000). Migration from lower to higher prevalence zone increases the risk, subsequently indicating the role of environmental factors in the etiology [3]. In India, ovarian cancer encompasses 8.7% of all cancers among women [4]. Mortality rate in developed countries has decreased recently, whereas the developing countries are showing a recent trend of increasing incidence and mortality rate of ovarian carcinoma due to economic growth and changes in lifestyle pattern [3]. Preoperative determination of the nature of ovarian neoplasm may not always be possible because clinical, radiological, serological, and even intraoperative findings of different ovarian tumors are not much specific, and they are often overlapping. In some instances, even advanced malignant ovarian neoplasms may mimic inflammatory processes [5]. Hence, intraoperative diagnostic modalities play indispensable role in management of ovarian tumors. Intraoperative diagnostic methods help surgeon to determine the optimum extent of operative procedure required for the patient and thus avoids unnecessary extensive surgery. It is especially important when fertility is concerned [6]. Intraoperative diagnostic methods include frozen section, imprint cytology, and scrape cytology. Frozen section can provide fairly accurate result with an accuracy of 86–97% [5–8]. But it requires expensive instruments like cryostat machine and qualified technicians. So, it is always not available in resource-poor developing countries. Freezing artifact is another unavoidable limitation of frozen section. Intraoperative cytological methods can help to overcome these limitations. They are inexpensive, rapid yet effective method for intraoperative diagnosis. Scrape smear cytology was introduced by Leonard S. Dudgeon and Vincent Patrick in 1927 at University of London [8]. Scrape smear cytology is a modification of imprint cytology in which cells are harvested by scraping the cut surface of the freshly removed surgical specimen [9]. It can provide better cellular yield than imprint smear [10]. Thus, it is a better diagnostic method compared to imprint cytology [11]. But it requires knowledge about the cytology of various types of ovarian tumors. Previous studies were mostly limited to assess the role of imprint cytology and frozen section in intraoperative diagnosis of ovarian neoplasms. Only few literatures are available to evaluate the role of scrape cytology in diagnosis of ovarian tumors which necessitates further studies to make both clinicians and cytopathologists aware of the utility of scrape cytology. With the background of above knowledge, the present study was conducted to study the distribution of morphological pattern of benign and malignant ovarian neoplasms in different age groups in Eastern India and also to evaluate the accuracy of scrape cytology in diagnosis of ovarian tumors by comparing the cytological interpretations with histopathological diagnosis since histopathology is considered to be the gold standard for diagnosis.

Materials and Methods

The present study was conducted after obtaining Institutional Ethical Clearance and informed consents from all the participants. Patient identity was kept confidential. It was an observational cross-sectional study done over a period of 12 months in the Department of Pathology of a tertiary care hospital of West Bengal. All the freshly removed surgical specimens of ovarian tumors sent in air-tight properly labeled containers without any fixative during the study period were included in this study. Random sampling was done. Cases with known inflammatory and/or infectious conditions or specimens sent in either formalin or normal saline or in unlabelled containers were excluded from this study. Immediately after receiving the specimen, gross examination of the tumor was done to look for solid and cystic areas. The specimen was then cut with a sharp knife into two halves. The cut surface was first wiped off to remove excess blood, if present, with the help of a filter paper. Then, scraping was done from the most representative area of the cut surface of the tumor with the help of a sharp scalpel or the end of a glass slide, depending upon the type of tissue. Areas of necrosis or hemorrhages were avoided. Cytological smears were prepared from the cellular material obtained by scraping and immediately fixed into 95% ethyl alcohol for subsequent staining with hematoxylin–eosin (H&E). One to four smears per case were prepared from different representative areas of the tumor. After obtaining the scrape smear the specimens were fixed in formalin. After 12 h of fixation, sections were taken from the same area from where scrapings were taken. Paraffin blocks of the sections were processed in the routine way and sections were stained with H&E stain. All the cytological smears and histological sections were evaluated separately and diagnosis was made.

After both cytological and histological interpretations were made, the results were categorized as follows to assess the efficacy of scrape cytology:

• Concordant: The cases where cytological interpretations were similar to the histological diagnosis with regard to major tumor categories (benign, borderline, or malignant) and cell type of origin (epithelial, sex cord-stromal, or germ cell) were considered as concordant. In cases of malignant tumors the identification of origin of tumor (primary or secondary) was also taken into consideration for determining concordance. However, inability to identify specific subtype of benign, borderline, or malignant lesions on scrape cytology were not considered as discordant, because this distinction generally does not influence intraoperative management [12].

• Discordant: The cases where equivocal interpretations were made on cytology or failure to categorize the tumor correctly as benign, borderline, or malignant or cases where correct interpretations regarding origin of the tumor (epithelial, sex cord-stromal, or germ cell tumor) could not be made were considered as discordant. Inability to differentiate primary from secondary malignancy were also considered under this category. These interpretations were made because failure to identify aforesaid categories may interfere with the optimal intraoperative management [12].

Thus, the cytological interpretation was compared with the histological diagnosis to assess the efficacy of scrape cytology in diagnosis of ovarian neoplasms.

Results

Total 110 cases were included in the present study. Most common age group was of 3rd decade with range from 12 to 70 years. Left ovary alone was involved in 48 cases, right ovary affected in 50 cases while in rest 12 cases bilateral ovaries were involved.

On final histopathology examination, 66 cases were diagnosed as benign (60%), 5 cases as borderline (4.55%), and 39 as malignant (35.45%). Of these, 64 cases (58%) were of epithelial origin whereas 25 cases (22.7%) were of germ cell origin and 6 (5.45%) cases were categorized as sex cord-stromal tumor. Among the benign neoplasms, the most common was serous cystadenoma, followed by mature cystic teratoma and mucinous cystadenoma comprising 39.39%, 21.21%, and 18.18% of all benign tumors, respectively. The most common malignant tumor was serous cystadenocarcinoma comprising 23.08% of all malignant cases of ovary (Table 1).

Table 1.

Distribution of ovarian masses according to histological diagnosis and correlation with cytological interpretation on scrape cytology

Histological diagnosis		Total no. of cases (N = 110)	Cytological interpretation
			Concordant (no. of cases)	Discordant (no. of cases)	Diagnostic accuracy (%)
Epithelial tumors	Mucinous cystadenoma	12	12	0	100
	Mucinous tumor (borderline)	3	1	2	33.33
	Mucinous cystadenocarcinoma	6	6	0	100
	Serous cystadenoma	26	26	0	100
	Serous tumor (borderline)	2	1	1	50
	Serous cystadenocarcinoma	9	9	0	100
	Clear cell adenocarcinoma	2	2	0	100
	Endometrioid carcinoma	2	2	0	100
	Brenner tumor	2	1	1	50
Germ cell tumors	Mature cystic teratoma	14	14	0	100
	Immature teratoma	2	0	2	00
	Yolk sac tumor	1	1	0	100
	Dysgerminoma	5	5	0	100
	Mixed germ cell tumor	3	3	0	100
Sex cord-stromal tumors	Granulosa cell tumor-adult type	3	3	0	100
	Fibrothecoma	2	2	0	100
	Sertoli-Leydig cell tumor	1	0	1	00
Other	Endometriotic cyst	10	8	2	80
	Metastatic tumor	4	3	1	75
	Lymphoma	1	1	0	100
Total		110	100	10	90.91

Twelve cases of mucinous cystadenoma were diagnosed cytologically by presence of monolayered sheets of columnar epithelial cells with honeycomb appearance and peripheral palisading (Fig. 1A and B). Cells were monomorphic with apical mucin, basally located nuclei, and there was no nuclear atypia or stratification. Background revealed some inflammatory cells and mucinous material. Six cases of mucinous cystadenocarcinomas were differentiated from benign ones by presence of nuclear stratification with some degree of nuclear atypia. All the benign and malignant mucinous tumors diagnosed on scrape cytology were corroborative with the histological diagnosis. Twenty-six cases of benign serous tumors showed clusters of cuboidal epithelial cells mostly with occasional ciliated epithelial cells. There was no nuclear atypia whereas 9 cases of malignant serous tumors showed marked cytological atypia and presence of papillae on scrape smears. Cytologoical interpretations of both benign and malignant serous neoplasms were in concordance with the histological diagnosis. Smears from 2 cases of clear cell carcinoma showed clusters of atypical cells with clear cytoplasm while 2 cases of endometriod carcinomas presented with clusters of highly atypical cells, lack of papillae, and occasional cells with squamous metaplasia (Fig. 1C and D). Both of the histological types of aforesaid carcinomas were correctly diagnosed on cytology. Out of total 64 epithelial tumors, 60 cases were diagnosed correctly on cytology, whereas in the remaining 4 cases including 2 cases of borderline mucinous tumor, 1 case of borderline serous tumor, and 1 case of Brenner tumor, cytological diagnosis was discordant with the histological diagnosis. All 3 borderline epithelial tumors were incorrectly diagnosed as benign tumors on scrape cytology, and cytological interpretation was inconclusive in case of Brenner tumor due to inadequate cellularity of the smear. There were 5 cases of dysgerminoma. Scrape smears from them were cellular smears containing dispersed population of cells with large vesicular nuclei, prominent nucleoli and abundant pale fragile cytoplasm. Background was tigroid in appearance with numerous scattered lymphocytes (Fig. 2A and B). There was a single case of yolk sac tumor diagnosed by identification of glandular arrangement of cuboidal epithelial cells, vacuolated cytoplasm, and occasional hyaline globules in the smear. All 14 cases of mature cystic teratomas diagnosed by plenty of aneucleated squames and keratin debris in the scrape smears (Fig. 2C and D). Out of total 25 cases of germ cell tumors evaluated, only two cases of immature teratoma could not be diagnosed correctly on cytology. They were misinterpreted as mature teratoma because the cytology smears did not reveal the immature neuro-epithelial element. Cytological preparation from 3 cases of granulosa cell tumors showed dyscohesive aggregates of cells with monomorphic nuclei, granular chromatin pattern, and occasional nuclear groove producing characteristic coffee-bean appearance (Fig. 3A and B) Cells had pale cytoplasm with indistinct cell border. Occasional Call-Exner body was also helpful for making the cytological diagnosis in cases where it was available. Two cases of fibrothecoma were encountered in this study. Both of them showed predominantly spindle cells with bland nuclear features, and occasional plump cells with vacuolated cytoplasm resembling theca cells were also present in the smears (Fig. 3C and D). Out of total 6 cases of sex cord-stromal tumors examined in the present study, only 1 case of Sertoli-Leydig cell tumor was incorrectly diagnosed as epithelial tumor on scrape cytology due to presence of extensive tubular component. Out of 10 cases of endometriotic cyst, 8 were diagnosed correctly on cytology. Remaining 2 cases showed presence of blood elements only in the smear making the cytological interpretation inconclusive. One case of lymphoma and 3 cases of metastatic adenocarcinoma were diagnosed correctly on cytology. However, 1 case of metastatic adenocarcinoma was wrongly interpretated as primary mucinous adenocarcinoma of ovary as there was extensive involvement of the ovary by the tumor and no clinical history was available. Table 1 shows the diagnostic accuracy of scrape cytology with respect to different histological type of ovarian tumors.

Fig. 1.

Mucinous cystadenoma showing columnar mucinous epithelium arranged in single row (A scrape cytology, B histology). Endometrioid carcinoma showing pleomorphic tumour cells and squamous metaplasia (C scrape cytology, D histology) (H&E, × 400)

Fig. 2.

Dysgerminoma showing large cells and lymphocytes (A scrape cytology, B histology) (H&E, × 400). Mature cystic teratoma: C scrape cytology showing aneucleated squames (H&E, × 400); D histology showing mature squamous epithelium (H&E, × 100)

Fig. 3.

Granulosa cell tumor showing longitudinal nuclear groove (A scrape cytology, B histology). Fibrothecoma showing bland spindle cells (C scrape cytology, D histology) (H&E, × 400)

Accuracy of scrape cytology in diagnosis of benign, borderline, and malignant ovarian tumors was found to be 95.45%, 40%, and 89.74%, respectively. Overall accuracy rate was 90.91% which is quite remarkable (Table 2).

Table 2.

Accuracy of scrape cytology in diagnosis of major categories of ovarian masses

Major histological subtype of tumor	No. of cases	Cytological interpretation
		Concordant	Discordant	Diagnostic accuracy(%)
Benign	66 (60%)	63	3	95.45
Borderline	5 (4.55%)	2	3	40
Malignant	39 (35.45%)	35	4	89.74
Total	110	100	10	90.91

Discussion

Intraoperative diagnostic methods including frozen sections, imprint cytology, and scrape cytology are important aspects of surgical pathology which guides the surgeon to decide the optimal surgical procedures required for the patients. Frozen sections though considered as gold standard, but it has some limitations like cost and expertise in the field. Intraoperative cytology has some advantages over frozen section including rapidity without compromising quality of diagnosis, simplicity, low-cost, absence of freezing artifact, and preservation of the tissue for further evaluation [13–15]. Again, scrape cytology shows better cellular yield and preservation of cellular morphology than conventional imprint smear. Sharma et al. studied the accuracy of scrape cytology in neoplastic lesions of different organs. They recorded 96% accuracy of scrape cytology in their study [16]. The role of scrape cytology in diagnosis of tumors of various organs other than ovary has been evaluated in other studies also. But till date, only few literatures performed on limited number of cases are available to evaluate the role of scrape cytology in diagnosis of ovarian tumors which necessitates further study. Total 110 cases of ovarian masses were evaluated in the present study. The age group ranged from 12 to 70 years, and 10.91% of the ovarian tumors were bilateral. The incidence of bilaterality was less in this study compared to previous study done by Sodha et al. In their study, the incidence of bilaterality was as high as 24.56% of the cases [17]. Tumors of epithelial origin was the most common type of ovarian neoplasm found in this study comprising 58% of the total cases followed by germ cell tumors (22.7%). Among the epithelial tumors 46.88% were benign, 7.8% were borderline, and rest 45.31% cases were malignant. Distribution of epithelial tumors found in this study was comparable to the previous study done by Shahid et al. [18] Serous cystadenoma was the most common type of tumor overall, and serous cystadenocarcinoma was the most common malignant tumor type. Previous studies also found serous tumors as the major tumor type like this study [19, 20]. In the current study, most of the benign and malignant epithelial tumors were correctly diagnosed on scrape cytology with overall sensitivity of 98.31%. A case of benign Brenner tumor could not be diagnosed on scrape cytology due to inadequate cellularity and lack of transitional epithelial component in the smear indicating inadequate sampling. Elaheh et al. recorded the overall accuracy of scrape cytology and frozen section as 91.86% and 92.68%, respectively, in diagnosis of ovarian neoplasms [21]. A meta-analysis done by Geomini et al. found sensitivity of frozen sections in diagnosis of benign and malignant adnexal masses varied from 65 to 97% and 97 to 100%, respectively [22]. So, the accuracy of scrape cytology in diagnosis of benign and malignant epithelial tumors was found to be comparable to frozen section in this study. However, Sodha et al. observed lower sensitivity of scrape cytology (89.47%) in diagnosis of benign and malignant tumors [17]. This may be due to higher number of mucinous tumors included in their study compared to the present one. Ganjei et al., Ali et al., and Gupta et al. also found difficulties in diagnosis of mucinous tumors on cytology because of huge size, heterogenous appearances, presence of both benign and malignant components in same tumor, and inflammatory component [23–25]. Low accuracy in diagnosis of borderline tumors were noted in present study like previous studies. Out of 5 cases of borderline epithelial tumors encountered in this study, only 2 cases could be diagnosed correctly on scrape cytology with calculated accuracy of 40% only. Remaining 3 cases were misdiagnosed as benign tumors. Sodha et al. and Shahid et al. also misinterpreted cases of borderline serous tumors as benign ones on cytology in their study [17, 18]. Previous studies noted difficulties in diagnosis of borderline tumors in frozen section also. Only 30 to 69% of borderline neoplasms could be correctly diagnosed on frozen section [26, 27]. The difficulty in diagnosis of borderline tumors often arises from lack of cellular atypia and inability to demonstrate stromal invasion on cytological smears. The calculated accuracy of scrape cytology in diagnosing germ cell tumor and sex cord stromal tumor was found to be 92% and 83.33%, respectively. Two cases of immature teratomas were diagnosed as mature cystic teratoma due to inability to identify immature neuro epithelial element in the smear and 1 case of Sertoli-Leydig cell tumor was misinterpreted as epithelial tumor on cytology. Sodha et al. also noted difficulty in diagnosing Sertoli-Leydig cell tumor similar to this study; however, they found abundant neuroepithelial element in the cytology smear of immature teratoma and they could make the diagnosis correctly [17].

To summarize, overall accuracy of scrape cytology in diagnosis of ovarian tumors was found to be as high as 90.91% in the present study which is remarkable and comparable to the previous studies (Table 3). Pitfall of the scrape cytology was noted in diagnosis of borderline ovarian epithelial neoplasms and immature teratomas. Borderline tumors specially having focal proliferation and atypia are often misinterpreted as benign on scrape smear due to lack of identification of nuclear stratification and cytological atypia in representative smears. Immature teratomas are frequently diagnosed as mature teratoma due to lack of immature neuro-epithelial elements in the cytological preparation. Inadequate sampling is mostly responsible for this misinterpretation. Hence, an extensive sampling from representative areas may help to overcome this limitation of scrape cytology.

Table 3.

Comparison of accuracy of cytology in diagnosis of ovarian tumors between present and previous studies

Study done by	Year of study	Place of study	No. of cases studied	Diagnostic accuracy (%)
Suen KC [28]	1978	Canada	57	87.72
Colin J.R. et al. [12]	2010	Perth, Australia	402	93.8
Shahid et al. [18]	2012	Aligarh	50	95.5
Shubha. H.V et al. [29]	2018	Karnataka	21	90.5
Bhardwaj S et al. [30]	2019	New Delhi	60	96.0
Sodha et al. [17]	2021	Gujrat	57	96.49
Present study	2021	West Bengal	110	90.91

Conclusion

Scrape cytology is a rapid, inexpensive yet effective tool to diagnose benign and malignant ovarian tumors with high diagnostic accuracy. So, it can be utilized as an intraoperative method of diagnosis. However, the efficacy of scrape cytology is limited by inability to identify borderline ovarian neoplasm and immature teratoma correctly.

Limitations of the Study

Comparative role of imprint cytology and frozen section in diagnosis of ovarian tumors could not be correlated with scrape cytology in this study.

Declarations

Conflict of Interest

The authors declare no competing interests.