Cytological Features of Cervical Smears in Serous Adenocarcinoma of the Endometrium

Jpn J Clin Oncol 2003;33(12)636 641 Cytological Features of Cervical Smears in Serous Adenocarcinoma of the Endometrium Yukiharu Todo, Shinichirou Minobe, Kazuhira Okamoto, Mahito Takeda, Yasuhiko Ebina, Hidemichi Watari, Mizue Terashima, Masanori Kaneuchi, Ritsu Yamamoto and Noriaki Sakuragi Department of Gynecology, Hokkaido University Graduate School of Medicine, Sapporo, Japan Received June 16, 2003; accepted November 3, 2003 INTRODUCTION Objective: Serous adenocarcinoma (SAC) of the endometrium has a poor prognosis compared with that of typical endometrioid adenocarcinoma (EAC). The objective of this study was to determine whether SAC can be distinguished from EAC preoperatively by cervical or endometrial cytology. Study design: Cervical smears and endometrial smears obtained from 128 patients with endometrial carcinoma were reviewed. Histological types included 117 cases of EAC and 11 cases of SAC. The positive rates of cervical smears and those of endometrial smears in SAC and EAC cases were compared. Papillary clusters and bare nuclei of malignant cells in positive cervical smears were also investigated for their diagnostic significance in discriminating between EAC and SAC. Results: The positive rate of cervical smears in SAC was significantly higher than that in EAC (72.7 vs 27.4%, P 0.05). Among cases with positive cervical smears, there were significantly more cases with papillary clusters and/or bare nuclei in cases of SAC than in cases of EAC. Conclusion: When endometrial carcinoma is clinically suspected and a cervical smear is positive, the predominance of either papillary clusters or features of bare nuclei of malignant cells in the smear may indicate the presence of SAC. Key words: papillary serous carcinoma endometrial carcinoma cervical smears Many studies have shown that the prognosis of serous adenocarcinoma (SAC) of the endometrium is poorer than that of typical endometrioid adenocarcinoma (EAC) (1 6). Although earlier studies suggested that the prognosis for patients with SAC was poor even at an early stage of the cancer, it has recently been shown that the survival rates of patients with early-stage SAC who had undergone surgery involving lymph node dissection were relatively good (7,8). Since patients with SAC of clinical stage I or II are more frequently found to be in surgically advanced stages than those with EAC (6), it is important to identify SAC before surgery and to determine the disease stage by thorough surgical staging, including lymphadenectomy. According to an earlier report, malignant cells are often found in cervical smears of patients with SAC (9). It has been suggested that the presence of papillary clusters or bare nuclei in cervical smears might be helpful for identifying SAC (10). For reprints and all correspondence: Noriaki Sakuragi, Department of Gynecology, Hokkaido University Graduate School of Medicine, Kita-15, Nishi-7, Kita-Ku, Sapporo 060-8638, Japan. E-mail: nkuma@aol.com We reviewed cervical smears of patients with endometrial carcinoma in whom thorough surgical staging had been performed to determine whether it is possible to distinguish SAC from EAC using the cytological features of cervical and endometrial smears. PATIENTS AND METHODS PATIENTS AND CYTOLOGY SLIDES We reviewed cervical cytology slides from 128 patients with endometrial carcinoma who had been treated at the Department of Obstetrics and Gynecology, Hokkaido University Hospital, Sapporo, during the period from 1990 to 1999. These patients included 117 patients with EAC and 11 patients with SAC. Of the patients with EAC, 73 were in FIGO (International Federation of Gynecology and Obstetrics) stage I, 12 were in stage II, 25 were in stage III and seven were in stage IV. Of the patients with SAC, five were in stage I, none were in stage II, one was in stage III and five were in stage IV. Operations were not performed in one patient with EAC and one patient with SAC. Both of the patients were in stage IV. Systematic lymphadenectomy was performed in 112 patients with EAC and 10 patients with SAC (Table 1). 2003 Foundation for Promotion of Cancer Research

Jpn J Clin Oncol 2003;33(12) 637 Table 1. Clinicopathologic characteristics of patients with endometrioid adenocarcinoma and serous adenocarcinoma of endometrium FIGO stage (1988) Number of patients (%) EAC (n = 117) SAC (n = 11) P-value I/II 85 5 0.083 III/IV 32 6 Architectural grade G1/G2 99 9 NS G3 18 2 Nuclear grade G1/G2 107 7 0.019 G3 10 4 Myometrial invasion 1/2 74 5 NS 1/2 42 5 Cervical invasion Negative 93 7 NS Positive 23 3 Lymph-vascular space invasion Negative/minimal 91 4 0.014 Moderate/prominent 25 6 Lymph node metastasis Negative 98 5 0.0083 Positive 14 5 Peritoneal washing cytology Negative 106 8 NS Positive 10 2 Hysterectomy was performed in 116 patients with EAC and 10 patients with SAC. Systematic lymphadenectomy was performed in 112 patients with EAC and 10 patients with SAC. Cases with positive cervical smears were investigated to determine the ratio of papillary clusters to malignant cell clusters and the number of malignant cells with bare nuclei. In evaluating cell clusters, normal glandular cell clusters (i.e. normal endometrial cells and normal endocervical cells) and stroma cell clusters were excluded and cell clusters suspected of being malignant cell clusters were counted as malignant cell clusters. When the ratio of papillary clusters to malignant cell clusters was 50%, papillary clusters were regarded as being predominant. Malignant cells with bare nuclei were classified into four groups based on the maximum number of bare nuclei in a single high-power field ( 400): ( ) for 0, (+) for 1 3, (++) for 4 6 and (+++) for 7. Other than degenerative cells, only cells with bare nuclei that could be definitely judged as being malignant cells according to their size and intranuclear structure were counted. Table 2. Cytologic diagnosis of cervical and endometrial Papanicolaou smears of typical endometrioid adenocarcinoma and serous adenocarcinoma Cytology EAC, endometrioid adenocarcinoma; SAC, serous adenocarcinoma. STATISTICAL ANALYSIS Fisher s exact test was used to determine statistically significant differences between the categorical values. P values 0.05 were considered to indicate statistical significance. RESULTS Number of suspicious or positive cytodiagnosis EAC (n = 117) SAC (n = 11) P-value Cervix (suspicious/positive) 46 (39.3%) 9 (81.8%) 0.01 Cervix (positive) 32 (27.4%) 8 (72.7%) 0.05 Endometrium (suspicious/positive) 99 (84.6%) 11 (100%) NS Endometrium (positive) 74 (63.2%) 11 (100%) NS Table 3. Cytologic diagnosis of cervical Papanicolaou smears of cases without cervical invasion Cervical cytology EAC without CI (n = 93) SAC without CI (n = 7) P-value Suspicious/positive 32 (34.4%) 5 (71.4%) 0.0504 Positive 20 (21.5%) 5 (71.4%) 0.05 EAC, endometrioid adenocarcinoma; SAC, serous adenocarcinoma; CI, cervical invasion. Table 2 shows that there were significantly more suspicious/ positive cervical smears in SAC patients than in EAC patients (P 0.01) and more positive cervical smears in SAC cases than in EAC cases (P 0.05). On the other hand, there was no significant difference between the proportions of suspicious/positive endometrial smears or between the proportions of positive endometrial smears in EAC cases and SAC cases. All of the 18 patients with negative endometrial smears were with EAC. These included 16 cases in FIGO stage I, one case in stage II and one case in stage III. Table 3 shows that in patients without cervical invasion: there seemed to be a trend that cervical smear was suspicious/positive in more patients (71.4%) of SAC compared with those with EAC (34.4%); while the proportion of positive cervical smear was significantly higher in SAC patients (71.4%) than in EAC patients (21.5%). Tables 4 and 5 show the results of investigation of papillary clusters and bare nuclei in 55 cases of suspicious/positive cervical smears and 40 cases of positive cervical smears, respectively. In both suspicious/positive and positive cervical smears, there were significantly more papillary clusters comprising 50% malignant cell clusters in cases of SAC than in cases of EAC (Table 4). There were significantly more malignant cells with bare nuclei in cases of SAC than in cases of EAC. In

638 Cervical smears in endometrial cancer a b Figure 1. Papillary cell clusters in (a) typical endometrioid adenocarcinoma and in (b) serous adenocarcinoma in cervical smears. Bar indicates 25 m. suspicious/positive cervical smears, 13 and 56% of malignant cells with bare nuclei were found (++/+++) in EAC and SAC cases, respectively. In positive smears, 19 and 63% of malignant cells with bare nuclei were found (++/+++) in EAC and SAC cases, respectively (Table 5). Fig. 1a shows a malignant cell cluster that was not judged to be a papillary cell cluster, while Fig. 1b shows a malignant cell cluster that was judged to be a papillary cell cluster. Fig. 2 shows bare nuclei judged to be (++). When we used predominance of papillary clusters ( 50% of clusters) and bare nuclei cell count (++/+++) as indices for differential diagnosis of histological subtypes of suspicious/ Figure 2. Bare nuclei cells in serous adenocarcinoma in cervical smears. Bar indicates 25 m. positive or positive cervical smears, there were significantly more cases with either of these indices in SAC patients than in EAC patients. Papillary clusters ( 50%) and/or bare nuclei (++/+++) were found in 17% of EAC cases and 78% of SAC cases (P 0.01) in suspicious/positive cervical smears and in 25% of EAC cases and 88% of SAC cases in positive cervical smears (P 0.005). Table 6 shows eight cases of EAC exhibiting either papillary clusters or bare nuclei. They included six patients with advanced cancer. Two cases of early cancer were of the histological grade 3. Table 7 shows 11 cases of SAC. There were two cases with negative cervical smears and nine with suspicious/positive smears. Two cases exhibiting neither papillary clusters nor bare nuclei were cases of early cancer. Although the follow-up period was not long enough, no recurrence/relapse occurred in the cases with negative cervical smears or in those exhibiting neither papillary clusters nor bare nuclei. To confirm these observations, we performed logistic regression analysis of clinicopathological factors related to the appearance of papillary clusters and/or bare nuclei in cervical cytology. It was found that only the histology of SAC was related to these cytological features in cervical cytology (Table 8). Table 4. Papillary cell clusters in suspicious/positive and positive cervical smears of typical endometrioid adenocarcinoma and serous adenocarcinoma Proportion of papillary cell clusters Suspicious/positive Cytodiagnosis of cervical smears Positive EAC (n = 46) SAC (n = 9) EAC (n = 32) SAC (n = 8) 50% of malignant cell clusters 41 (89%) 4 (44%) 27 (84%) 3 (37%) 50% of malignant cell clusters 5 (11%) 5 (56%) 5 (16%) 5 (63%) P-value 0.01 0.05 EAC, endometrioid adenocarcinoma; SAC, serous adenocarcinoma.

Jpn J Clin Oncol 2003;33(12) 639 DISCUSSION If endometrial cells are observed in cervical smears during the second half of the menstrual cycle or at any time in the postmenopausal period, it suggests abnormal endometrial desquamation. Ng et al. reported that adenocarcinoma was found in 3.2% of 662 patients exhibiting such cells (11) and Cherkis et al. reported that adenocarcinoma was found in 11.2% of 179 patients exhibiting such cells (12). Moreover, it has been reported that this feature becomes more significant with age and with increase in the degree of cytological atypia. Schneider et al. reported the rates of positive cervical smears from patients with endometrial carcinoma in a retrospective study of 220 patients and in a prospective study of 123 patients (13). In the former study, the rate of positive smears and that of suspicious/positive smears were 33.2 and 58.7%, respectively, and in the latter study 36.6 and 67.5%, respectively. Hence the rate of positive cervical smears from patients with endometrial carcinoma may be ~35% and the rate of suspicious/positive cervical smears may be ~60%. SAC has recently attracted much attention because of its high tendency towards advanced surgical stages and its poor prognosis. Regarding the surgical stages, Chambers et al. Table 6. FIGO surgical stage, cervical involvement, and histologic grade of eight cases of typical endometrioid adenocarcinoma exhibiting either papillary clusters or bare-nuclei featured cells in cervical smears Case (age) Cytologic features of cervical cytology FIGO Stage (1988) Papillary clusters Bare nuclei Cervical invasion Histologic grade Outcome (months) 1 (57) ( ) (++) Ic ( ) G3 NED (67) 2 (54) ( ) (++) IIb (+) G3 NED (51) 3 (66) (+) (++) IIIa (+) G3 NED (22) 4 (64) ( ) (++) IIIc (+) G1 NED (60) 5 (53) ( ) (+++) IVb (+) G3 DOD (13) 6 (50) (+) (++) IVb (+) G2 DOD (38) 7 (60) (+) (+++) IVb (+) G2 DOD (7) 8 (40) (+) (+) IVb (+) G3 AWD (24) Papillary clusters: ( ) for 40% and (+) for 50% of malignant cell clusters; Bare nuclei cells, per high power field ( 400): ( ) for nil, (+) for 1 3, (++) for 4 6, (+++) for 6; NED, no evidence of disease; AWD, alive with disease; DOD, dead of disease. Table 7. Cytologic features in cervical smears and clinicopathologic characteristics of 11 cases of serous adenocarcinoma Case (age) Cervical cytology Cytologic features of cervical cytology FIGO Stage (1988) Outcome (months) Papillary clusters Table 5. Bare nuclei cell count in suspicious/positive and positive cervical smears of typical endometrioid adenocarcinoma and serous adenocarcinoma Bare nuclei cell count EAC, endometrioid adenocarcinoma; SAC, serous adenocarcinoma. Bare nuclei Cytodiagnosis of cervical smears Suspicious/positive 1 (67) Negative Ia NED (27) 2 (55) Positive (+) ( ) Ia NED (43) 3 (60) Positive ( ) ( ) Ib NED (12) 4 (57) Suspicious ( ) ( ) Ib NED (11) 5 (63) Positive (+) (+) Ic NED (44) 6 (68) Negative IIIc NED (24) 7 (67) Positive (+) (+++) IVb DOD (1) 8 (67) Positive ( ) (++) IVb NED (15) 9 (72) Positive (+) (++) IVb DOD (16) 10 (54) Positive ( ) (++) IVb DOD (7) 11 (44) Positive (+) (++) IVb DOD (3) Positive EAC (n = 46) SAC (n = 9) EAC (n = 32) SAC (n = 8), + 40 (87%) 4 (44%) 26 (81%) 3 (37%) ++, +++ 6 (13%) 5 (56%) 6 (19%) 5 (63%) P-value 0.01 0.05 Papillary clusters: ( ) for 50% and (+) for 50% of malignant cell clusters; Bare nuclei cells, per high power field ( 400): ( ) for nil, (+) for 1 3, (++) for 4 6, (+++) for 6; NED, no evidence of disease; DOD: dead of disease.

640 Cervical smears in endometrial cancer Table 8. Logistic regression analysis of factors relating to papillary cell clusters and/or bare nuclei in cervical cytology Histopathologic factor Univariate analysis Multivariate analysis 2 P-value 2 P-value FIGO surgical stage (1988) 11.03 0.0009 1.13 0.29 Histologic subtype 18.06 0.0001 11.21 0.0008 Architectural grade 5.91 0.015 1.95 0.16 Nuclear grade 10.52 0.0012 0.12 0.13 Lymph-vascular space invasion 13.89 0.0002 1.32 0.25 Myometrial invasion 6.75 0.0094 1.52 0.22 FIGO stage, I/II vs III/IV; Histologic subtype, endometrioid vs serous; Architectural grade, 1/2 vs 3; Nuclear grade, 1/2 vs 3; Lymph-vascular space invasion, /+ vs ++/+++; Myometrial invasion, 1/2 vs 1/2. reported that the 5-year survival rate of patients was 45% when SAC was in surgical stage I or II (4). Ward et al. reported that the 3-year survival rates were 44% in surgical stage I and 33% in stage II (3). Thus, SAC has often been reported as a disease with a poor prognosis even if it is found in stage I or II. However, lymph node dissection or even sampling was not performed in any of those studies. In studies where lymph node sampling was performed, Grice et al. (7) and Gitsch et al. (8) found that the 5-year survival rate was 80% for patients with surgical stage I or II SAC. Goff et al. reported that lymph node metastasis was found in 36% of cases of SAC without myometrial invasion (14). Hence it should be borne in mind that lymph node dissection is an important factor that may greatly affect prognosis according to surgical stage. It has been reported that the prognosis of endometrial carcinomas with a papillary architecture and serous differentiation, which were found in 25% of the tumor, was as poor as that of pure papillary serous carcinoma (15). In the light of these findings, complete surgical staging, including lymph node dissection, is necessary for the treatment of SAC and cytological examination before a surgical operation with suspicion of SAC provides important information regarding the necessity for this procedure. Hendrickson et al. reported that SAC had a papillary architecture supported by a fibrovascular framework identical with that encountered in ovarian serous carcinomas and that the stroma supporting the papillae was abundant or edematous (16). They also reported that the diagnosis of SAC was not based solely on assessment of architectural features but also on the cytological finding of anaplasia. Psammoma bodies were found in one-third of the patients in the series examined by Hendrickson et al. However, none of these reports suggested the possibility of differentiation of papillary serous carcinoma from typical endometrial carcinoma by means of endometrial smears. In 1989, Kuebler et al. reported that the possibility of SAC should be considered whenever a cervical smear contains numerous papillary groups of large tumor cells with macronucleoli (9). In 1999, Wright et al. reported that SAC and EAC may be distinguished by investigating some parameters in cervical smears, such as hypercellular smears, background tumor diathesis, papillae, bare nuclei and cells with large pleomorphic nuclei and bulky dense cytoplasm (10). Schneider et al. reported that the rate of positive cervical smears increases with advancement of histological grade, cytological grade and postsurgical stage (13). Since SAC is associated with a high cytological grade and tends to be in more advanced post-surgical stages, spontaneous exfoliation may occur and most smears may therefore be judged as positive rather than suspicious. It was only Kuebler et al. who compared cervical smears obtained from patients with EAC and SAC and they reported that 35% of the cervical smears from EAC cases were positive and 55% were suspicious/positive, whereas 71% of the smears from SAC cases were positive and 78% suspicious/positive (9). This indicates that the rate of positive cervical smears is higher in patients with SAC. Our results are consistent with their findings: 27% of the cervical smears in our EAC cases were positive and 39% suspicious/positive, whereas 73% of the cervical smears in our SAC cases were positive and 82% suspicious/positive. We also compared the ratios of papillary clusters with malignant cell clusters and the numbers of malignant cells featuring bare nuclei in positive cervical smears in SAC and EAC patients. We found that SAC is more frequently associated than EAC with papillary clusters comprising 50% of malignant cell clusters and with malignant cells with a score of (++/+++) for the bare nuclei cell count. However, it is not clear whether these features could be indices for differentiation of histological subtypes, since they are not very disease-specific. Patients exhibiting neither of the two indices were exclusively with EAC. However, EAC in advanced post-surgical stages and at a higher histological grade tends to exhibit these features. At present, it is at least justifiable to suspect SAC when a cervical smear is positive or when either papillary clusters or bare nuclei are found in cervical smears. When SAC cannot be proven histologically but one of these cytological features is found, surgery involving lymph node dissection should be considered, taking into account the possibility of a mixedserous tumor. References 1. Abeler VM, Kjorstad KE. Serous papillary carcinoma of the endometrium: a histopathological study of 22 cases. Gynecol Oncol 1990;39: 266 71. 2. Sutton GP, Brill L, Michael H, Stehman FB, Ehrlich CE. Malignant papillary lesions of the endometrium. Gynecol Oncol 1987;27:294 304. 3. Ward BG, Wright RG, Free K. Papillary carcinomas of the endometrium. Gynecol Oncol 1990;39:347 51. 4. Chambers JT, Merino M, Kohorn EI, Peschel RE, Schwartz PE. Uterine papillary serous carcinoma. Obstet Gynecol 1987;69:109 13. 5. Carcangiu ML, Chambers JT. Early pathologic stage clear cell carcinoma and uterine papillary serous carcinoma of the endometrium: comparison of clinicopathologic features and survival. Int J Gynecol Pathol 1995;14: 30 8. 6. Sakuragi N, Hareyama H, Todo Y, Yamada H, Yamamoto R, Fujino T, et al. Prognostic significance of serous and clear cell adenocarcinoma in surgically staged endometrial carcinoma. Acta Obstet Gynecol Scand 2000;79:311 6.

Jpn J Clin Oncol 2003;33(12) 641 7. Grice J, Ek M, Greer B, Koh WJ, Muntz HG, Cain J, et al. Uterine papillary serous carcinoma: evaluation of long-term survival in surgically staged patients. Gynecol Oncol 1998;69:69 73. 8. Gitsch G, Friedlander ML, Wain GV, Hacker NF. Uterine papillary serous carcinoma. A clinical study. Cancer 1995;75:2239 43. 9. Kuebler D, Nikrui N, Bell D. Cytologic features of endometrial papillary serous carcinoma. Acta Cytol 1989;33:120 6. 10. Wright CA, Leiman G, Burgess SM. The cytomorphology of papillary serous carcinoma of the endometrium in cervical smears. Cancer (Cancer Cytopathol) 1999;87:12 8. 11. Ng AB, Reagan JW, Hawliczek S, Wentz BW. Significance of endometrial cells in the detection of endometrial carcinoma and its precursors. Acta Cytol 1974;18:356 61. 12. Cherkis RC, Patten SF Jr, Andrews TJ, Dickinson JC, Patten FW. Significance of normal endometrial cells detected by cervical cytology. Obstet Gynecol 1988;71:242 4. 13. Schneider ML, Wortmann M, Weigel A. Influence of the histologic and cytologic grade and the clinical and postsurgical stage on the rate of endometrial carcinoma detection by cervical cytology. Acta Cytol 1986; 30:616 22. 14. Goff BA, Kato D, Schmidt RA, Ek M, Ferry JA, Muntz HG, et al. Uterine papillary serous carcinoma: patterns of metastatic spread. Gynecol Oncol 1994;54:264 8. 15. Sherman ME, Bitterman P, Rosenshein NB, Delqado G, Kurman RJ. Uterine serous carcinoma: a morphologically diverse neoplasm with unifying clinicopathologic features. Am J Surg Pathol 1992;16:600 10. 16. Hendrickson M, Ross J, Eifel P, Martinez A, Kempson R. Uterine papillary serous carcinoma: a highly malignant form of endometrial adenocarcinoma. Am J Surg Pathol 1982;6:93 108.