Anatomic Pathology / CYTOLOGY OF CARCINOSARCOMA OF UTERUS An Abnormal Cervicovaginal Cytology Smear in Uterine Carcinosarcoma Is an Adverse Prognostic Sign Analysis of 25 Cases Matthew J. Snyder, MD, 1 Stanley J. Robboy, MD, 1 Robin T. Vollmer, MD, 1,2 and Leslie G. Dodd, MD 1 Key Words: Carcinosarcoma; Uterus; Malignant mixed mesodermal tumor; Malignant mixed müllerian tumor; MMMT; Papanicolaou smear; Cytology; Prognosis Abstract Carcinosarcoma of the uterus has been poorly characterized on cervicovaginal (Pap) smears, and we examine whether they effectively screen for carcinosarcoma and whether an abnormal Pap smear result has any clinical importance. Twenty-five patients with histologically confirmed carcinosarcoma had a conventional Pap smear shortly before diagnosis. Eleven smears (44%) originally were read as abnormal (malignant or atypical), and 4 additional cases were read as abnormal on retrospective review (15/25 [60%]). All malignant elements were epithelial, and 2 cases (8%) had atypical spindle cells, but no diagnostic sarcoma. Cervical involvement was the only histologic parameter correlating with an abnormal Pap smear result (P =.04). Univariate analysis found stage III or IV disease was an adverse prognostic sign compared with stage I or II disease (mean survival, 8 vs 36 months, respectively; P =.001), and multivariate analysis indicated that an abnormal Pap smear result correlated with worse survival (P =.023). The conventional Pap smear is insensitive (60%) for detecting carcinosarcoma, but when the result is abnormal, the Pap is an important stage-independent adverse prognosticator. Carcinosarcoma, earlier known as malignant mixed mesodermal/müllerian tumor, is an uncommon uterine tumor that occurs principally in older women. The tumor typically appears grossly as a large intrauterine mass that often fills and expands the endometrial cavity and histologically discloses a mixture of malignant epithelial and mesenchymal elements. These tumors usually are associated with a poor prognosis. Our study had 3 aims: (1) test whether the routine Papanicolaou (Pap) smear effectively screens for the detection of carcinosarcoma, (2) search for consistent findings to recognize carcinosarcoma on Pap smears, and (3) examine whether the finding of an abnormal smear has independent clinical significance. Materials and Methods With institutional review board approval, all cases of uterine carcinosarcoma diagnosed from January 1983 through December 2003 were retrieved from the Duke University Medical Center (Durham, NC) Department of Pathology files. Cases were included only if an antecedent Pap smear was obtained within 6 months of the histologic diagnosis and both the Pap and histologic slides were available for review. The conventional Pap smears were reviewed in a blinded manner by two of us (M.J.S. and L.G.D.) and classified as diagnostic of malignancy, suspicious for malignancy, or negative for malignancy. The smears were further evaluated qualitatively by cellular components present (eg, mesenchymal or epithelial elements). The sarcomatous component was classified histologically on the surgical specimen (eg, sarcoma not otherwise specified, rhabdoid, chondroid, or lipomatous 434 Am J Clin Pathol 2004;122:434-439 Downloaded 434 from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428
Anatomic Pathology / ORIGINAL ARTICLE differentiation) and compared quantitatively with the histologic types of carcinoma present (eg, endometrioid, serous, squamous, clear cell, and mucinous). The Pap smear findings then were correlated with the presence and site of metastasis, uterine cervix involvement, presence and extent of myometrial invasion, and lymphatic or vascular invasion. Data about clinical stage, length of time to and site of recurrence, and outcome were analyzed. Statistical Analysis We used χ 2 tests to compare categorical variables. Kaplan- Meier plots and log-rank tests were used to relate single prognostic variables to survival time. The Cox proportional hazards model was used for multivariate analysis of survival. 1 Results Twenty-five conventional Pap smears were identified and reviewed using the aforementioned criteria. All cases had histologic confirmation by endometrial biopsy (5 cases) or hysterectomy (20 cases). Eleven (44%) of the Pap smears originally were read as malignant or atypical, and 4 additional cases were classified as atypical on retrospective review, for a sensitivity of 60% (15/25) Table 1. No cases originally read as negative were reclassified as diagnostic of malignancy. The 4 cases found to be atypical on review had few or small clusters of atypical epithelial cells. The malignant or atypical elements were epithelial in all 15 cases Image 1 and Image 2. Of these, 7 (47%) originally were diagnosed as adenocarcinoma of likely endometrial origin, 2 (13%) as squamous cell carcinomas (no squamous carcinoma present histologically within the carcinosarcoma), and 1 (7%) as poorly differentiated carcinoma. The remaining 5 cases were not otherwise classified. Two Pap smears (8%; cases 2 and 6) disclosed atypical spindle cells (Image 1A), but no smear revealed diagnostic sarcomatous elements. Case 6 with atypical spindle cells had a large sarcomatous component in the surgical specimen (95% was sarcoma), but case 2 did not (35% of the tumor was sarcoma). The 4 other cases (cases 8, 10, 20, and 22; Table 1) with 90% or more sarcoma lacked Table 1 Summary of Papanicolaou Smear and Histologic Data Lymphatic Original/Review Cervix Myometrial or Vascular Diagnosis/Case No. * Diathesis Carcinoma Sarcoma Metastasis Involvement Invasion Invasion Diagnostic/diagnostic 1 No E, 50 U, 40; R, 10 NK 2 Yes E, 60; CC, 4; M, 1 C, 20; U, 15 Yes Yes Yes, O Yes 3 No S, 75; E, 10 U, 14; R, 1 No No Yes, Su No 4 Yes E, 90 U, 10 Yes No Yes, O Yes 5 No E, 70 C, 30 No No Yes, I Yes 6 Yes E, 5 U, 80; R, 15 NK 7 Yes E, 30 U, 50; C, 20 No Yes No No Atypical/diagnostic 8 Yes E, 5 U, 95 NK Atypical/atypical 9 No E, 40 U, 60 NK 10 No E, 5 R, 50; U, 25; L, 20 No Yes Yes, O No 11 No E, 95 U, 5 No No Yes, Su No Negative/atypical 12 No E, 80 U, 20 Yes No Yes, O Yes 13 No S, 50 U, 50 No No No No 14 No E, 20 U, 45; R, 30; U, 5 No Yes Yes, O No 15 No E, 50 U, 50 NK Negative/negative 16 No E, 50 U, 50 Yes No No No 17 No S, 60; CC, 30 U, 9; C, 1 Yes No Yes, Su No 18 No E, 35 U, 60; R, 5 No No Yes, Su No 19 No E, 70 U, 30 Yes No Yes, O No 20 No S, 5 U, 95 No No No No 21 No E, 35; Sq, 20 U, 40; R, 3; C, 2 No No Yes, I No 22 No E, 10 U, 85; C, 5 No No Yes, Su No 23 No E, 65; CC, 5 U, 30 No No Yes, Su Yes 24 No S, 25; Sq, 20; E, 5 U, 40; C, 5; L, 5 No No No No 25 No E, 80 U, 20 Yes Yes Yes, O Yes C, cartilaginous; CC, clear cell; E, endometrioid; I, inner half; L, lipomatous; M, mucinous; NK, not known; O, outer half; R, rhabdoid; S, serous; Sq, squamous; Su, superficial (1-2 mm); U, unspecified sarcoma. * Diagnosis from Papanicolaou smear; diagnostic indicates diagnostic of malignancy; atypical indicates atypical, suspicious for malignancy; negative indicates negative for malignancy. Cases 2 and 6 had atypical spindle cells. Numbers represent the percentages of carcinoma and sarcoma in the tumor. Downloaded from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428 Am J Clin Pathol 2004;122:434-439 435 435 435
Snyder et al / CYTOLOGY OF CARCINOSARCOMA OF UTERUS A B C Image 1 (Case 2) Carcinosarcoma. A, Several scattered atypical spindle cells (arrows) are present in a background of necrosis (Papanicolaou smear, 150; inset, corresponding histologic slide, H&E, 100). The cells alone are not diagnostic of sarcoma because similar cells can be present with an endometrial adenosquamous carcinoma or cervical squamous cell carcinoma. B, The carcinomatous component is poorly differentiated with a high nuclear/cytoplasmic ratio and can be confused with squamous carcinoma in situ (Papanicolaou smear, 150). C, The corresponding histologic slide (H&E, 100) shows the high-grade carcinoma and surrounding sarcoma (arrow). atypical spindle cells in the Pap smear. Of the 8 Pap smears with changes diagnostic of malignancy, 5 (63%) had a readily identifiable tumor diathesis, typified by a dirty necrotic background (Image 2B), often with many macrophages. The histologic review confirmed both a carcinomatous and sarcomatous component in every case (examples in Images 1 and 2). Cervical involvement in the hysterectomy specimen correlated positively with an abnormal Pap smear result (P =.04) but not with clinical stage (P =.11) or survival outcome (P =.70). Having an abnormal Pap smear result did not predict which of the 20 patients who underwent hysterectomy would have metastasis (P = 1.0), myometrial invasion (P =.30), or lymphatic or vascular invasion (P =.14). Nearly every tumor contained endometrioid adenocarcinoma and sarcoma of an unspecified type or as a mixture of other varieties (Table 1). The cases that had a carcinoma component in the carcinosarcoma of 50% or more in the resection or curettage specimen were not more likely to have had an abnormal Pap smear result (P =.93). Almost half of the patients (12/25 [48%]) had a recurrence, and half of the recurrences (6/12) were vaginal. Every vaginal recurrence was discovered on a routine follow-up Pap smear and confirmed by biopsy. The patients who had a vaginal recurrence were more likely to have tumor involving the cervix revealed by hysterectomy (P =.002). Of the 6 cases with a vaginal recurrence, 3 had cervical involvement, 1 did not, and the other 2 had only an endometrial biopsy. This is contrasted with 13 cases that did not have cervical involvement or a vaginal recurrence and only 1 case that had a vaginal recurrence without cervical involvement. Univariate and multivariate statistical analyses identified features related to survival with an average of 31 months of follow-up (range, 1-150 months) Table 2. Patients with stage III or IV disease had a median survival of 8 months in 436 Am J Clin Pathol 2004;122:434-439 Downloaded 436 from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428
Anatomic Pathology / ORIGINAL ARTICLE A B C D Image 2 A (Case 3), This carcinosarcoma had a large serous carcinomatous component with distinctive papillary groups on the Papanicolaou smear (Papanicolaou smear, 150) and histologic slide (inset, H&E, 100). B (Case 4), Clusters of adenocarcinoma on the Papanicolaou smear were immersed in a necrotic tumor diathesis (Papanicolaou smear, 150; inset, corresponding histologic slide, H&E, 100). C (Case 7), Clusters of the adenocarcinoma on the Papanicolaou smear were intermixed with normal squamous cells (Papanicolaou smear, 150). D (Case 7), The corresponding histologic case (H&E, 100) demonstrates a malignant cartilaginous component (left) adjacent to the carcinoma (right). comparison with a median survival of 36 months for those with low-stage disease (P =.006; log-rank test) Figure 1. Whereas an abnormal Pap smear result was not associated with a higher clinical stage (P >.8; χ 2 test), Cox proportional hazards model analysis of survival time demonstrated that the presence of an abnormal Pap smear result added significant prognostic information to that provided by clinical stage Table 3 and Figure 2. In both stage groups, the patients with an abnormal Pap smear result had shorter survival; furthermore, most of the affected patients had stage I or II tumors. The patients with low-stage (I and II) tumor and an abnormal Pap smear result had a decreased median survival from 44 to 36 months, whereas patients with high-stage tumors with an abnormal Pap smear result had a drop in median survival time from 8 months to only 7.5 months. Discussion The results of this study show that the conventional Pap smear is insensitive (60% sensitivity) for detecting uterine carcinosarcoma, even on retrospective review. This confirms Downloaded from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428 Am J Clin Pathol 2004;122:434-439 437 437 437
Snyder et al / CYTOLOGY OF CARCINOSARCOMA OF UTERUS Table 2 Summary of Clinical Data 1.0 Case No. Clinical Outcome Follow-up (mo) Stage 1 D-NED 52 II 2 DOD 17 IIIc 3 DOD 36 Ib 4 DOD 2 IV 5 DOD 29 Ib 6 DOD 7 IIIc 7 DOD 22 IIb 8 DOD 5 IV 9 DOD 1 III 10 DOD 28 Ilb 11 A&W 40 Ib 12 DOD 9 III 13 DOD 8 IIIb 14 DOD 40 II 15 DOD 11 IIIc 16 DOD 44 III 17 DOD 3 IV 18 D-NED 44 Ib 19 DWD 6 III 20 A&W 150 Ia 21 AWD 119 IV 22 DOD 20 Ib 23 A&W 52 Ib 24 DOD 19 IIc 25 DOD 8 IV A&W, alive and well; AWD, alive with disease; D-NED, died with no evidence of disease; DOD, died of disease; DWD, died with disease. Table 3 Cox Model Analysis of Survival Time * Variable Coefficient Standard Error P Stage 1.65 0.542.002 Papanicolaou smear 1.30 0.569.023 * Stage and Papanicolaou smear results are coded as binary variables, ie, 1 for clinical stage greater than II; otherwise, 0; 1 for abnormal Papanicolaou smear result; otherwise, 0. Survival Probability A 1.0 0.8 0.6 0.4 0.2 0.0 Pap Pap+ 0 20 40 60 80 100 120 140 Follow-up (mo) B Survival Probability 1.0 0.8 0.6 0.4 0.2 0.0 Pap+ Pap 0 20 40 60 80 100 120 Follow-up (mo) Figure 2 Survival as affected by an abnormal Papanicolaou (Pap) smear result (Kaplan-Meier) in patients with low clinical stage (A; stage I or II) and high clinical stage (B; stage III or IV) disease. Survival Probability 0.8 0.6 0.4 0.2 Stage <3 Stage >3 0.0 0 20 40 60 80 100 120 140 Follow-up (mo) Figure 1 Survival time as a function of stage (Kaplan-Meier) (P =.006). other reports of 56% to 70% sensitivity. 2-7 Endometrial aspiration is more effective as shown in small numbers of patients, but this method is used uncommonly. 8 When diagnostic, the malignant cells appear as adenocarcinoma in the Pap smear (Images 1 and 2). A diagnostic sarcomatous component was consistently absent, even though the tumor often was large and extensive in the curettage or resection specimen. The Pap smear, even when diagnostic of malignancy, was never diagnostic of carcinosarcoma. Because atypical spindle cells were present in only 8% (2/25) of the Pap smears (Image 1), we found no reliable feature that identified carcinosarcoma specifically on the Pap smear. Heterologous elements usually are identified only on endometrial aspirates. 8 The 2 cases (cases 2 and 6; Table 1) that had atypical spindle cells were diagnostic of epithelial malignancy. While a diagnosis of carcinosarcoma could have been suggested in these 2 cases, the differential diagnosis includes a spindle cell component of an endometrial adenosquamous carcinoma and cervical squamous cell carcinoma. Given both the paucity of the nonspecific atypical cells and the relative infrequency of carcinosarcoma in comparison with conventional endometrial adenocarcinoma, we suspect that most cytopathologists interpreting the Pap smear might not be inclined to raise the possibility of carcinosarcoma in the differential diagnosis. Some investigators have reported a diagnostic sarcomatous component in a very small number of conventional Pap smears, sometimes as rhabdomyoblasts 4 or malignantappearing spindle cells. 2,3,6 These studies also found that the majority of abnormal Pap smears had an adenocarcinoma as a more consistent finding. Because the epithelial component typically is represented on an abnormal Pap smear, a carcinosarcoma that has a large carcinomatous portion might be more likely to yield an abnormal Pap smear result. However, 438 Am J Clin Pathol 2004;122:434-439 Downloaded 438 from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428
Anatomic Pathology / ORIGINAL ARTICLE the cases in our study with a large carcinomatous component were not more likely to have an abnormal Pap smear result. Our analysis indicates that the sarcomatous component does not shed easily, even when the sarcomatous portion was substantial in size on the resection specimen. However, if the cervix was involved, the Pap smear result was more likely to be abnormal (P =.04). This finding, though, did not affect clinical outcome or correlate with a higher clinical stage. Also, an abnormal Pap smear result was not associated with metastasis, presence or depth of myometrial invasion, or lymphatic or vascular invasion, in contrast with cases of pure adenocarcinoma. 9 Even though the Pap smear may not be a sensitive screening tool for detecting carcinosarcoma of the uterus, significant prognostic information is obtained when the Pap smear result is abnormal. Our outcome data indicate that an abnormal Pap smear result is an important adverse prognostic sign that is stage-independent. As expected, clinical stage showed a strong correlation with outcome (Figure 1), but patients with a high-stage tumor were not more likely to have an abnormal Pap smear result. Similar results have been reported with endometrial adenocarcinoma on Pap smears. 10 Multivariate analysis, which controlled for the obvious effect of stage on outcome, showed that patients with an abnormal Pap smear result had statistically worse outcomes (Table 3 and Figure 2). This is the first study, to our knowledge, showing this association. Several speculations about why an abnormal Pap smear result adds information to clinical stage include the following: (1) Current stage definitions are incomplete. (2) Less cell-to-cell adhesion results in an abnormal Pap smear result and implies a more aggressive tumor. If additional studies confirm that an abnormal smear result is an adverse factor and stage-independent, then consideration might be given to incorporating this finding into staging protocols. The past few years have witnessed a major change in cervicovaginal screening with increasing use of the more sensitive liquid-based techniques. Given that about two thirds of patients with carcinosarcoma had abnormal conventional Pap smear results, it is surmised that at least the epithelial component would be detected more often with the liquid-based techniques, given results for detecting the more common forms of endometrial adenocarcinoma. 11-13 Perhaps the sarcomatous cells also would be seen more easily, but this component does not seem to shed as easily as the carcinoma. It will be interesting to test whether an abnormal liquid-based screening test result in women with carcinosarcoma will be associated with a worse prognosis independent of stage, as we have demonstrated with conventional Pap smears. From the Departments of Pathology, 1 Duke University Medical Center and 2 Durham Veterans Affairs Hospital, Durham, NC. Address reprint requests to Dr Snyder: Dept of Pathology, Duke University Medical Center, Box 3712, Durham, NC 27710. Acknowledgment: We thank Steven Conlon for technical assistance with the images and figures. References 1. Cox DR, Oakes D. Analysis of Survival Data. London, England: Chapman and Hall; 1984. Monographs in Statistics and Applied Probability; vol 21. 2. Costa MJ, Tidd C, Willis D. Cervicovaginal cytology in carcinosarcoma (malignant mixed müllerian [mesodermal] tumor) of the uterus. Diagn Cytopathol. 1992;8:33-40. 3. Massoni EA, Hadju SI. Cytology of primary and metastatic uterine sarcomas. Acta Cytol. 1984;28:93-100. 4. Matsuyama S, Tokunaga T, Mimori H, et al. Exfoliative cytology of mixed mesodermal tumors of the uterus: review of cytologic smears sampled prior to therapy [in Japanese]. Nippon Sanka Fujinka Gakkai Zasshi. 1985;37:1125-1134. 5. Parker JE. Cytologic findings associated with primary uterine malignancies of mixed cell types (malignant mixed müllerian tumor). Acta Cytol. 1964;8:316-320. 6. Tenti, P, Babilonti L, La Fainza A, et al. Cytology of malignant mixed mesodermal tumour of the uterus: experience of 10 cases. Eur J Gynaecol Oncol. 1989;10:125-128. 7. White TH, Glover JS, Peete CH, et al. A 34-year clinical study of uterine sarcoma including experience with chemotherapy. Obstet Gynecol. 1965;25:657-663. 8. An-Foraker SH, Kawada CY. Cytodiagnosis of endometrial mixed mesodermal tumor. Acta Cytol. 1983;29:137-141. 9. Larson DM, Johnson KK, Reyes CN Jr, et al. Prognostic significance of malignant cervical cytology in patients with endometrial cancer. Obstet Gynecol. 1994;84:399-403. 10. Gu M, Shi W, Barakat RR, et al. Pap smears in women with endometrial carcinoma. Acta Cytol. 2001;45:555-560. 11. Guidos BJ, Selvaggi SM. Detection of endometrial adenocarcinoma with the ThinPrep Pap Test. Diagn Cytopathol. 2000;23:260-265. 12. Schorge JO, Hossein Saboorian M, Hynan L, et al. ThinPrep detection of cervical and endometrial adenocarcinoma: a retrospective cohort study. Cancer. 2002;96:338-343. 13. Uyar DS, Eltabbakh GH, Mount SL. Positive predictive value of liquid-based and conventional cervical Papanicolaou smears reported as malignant. Gynecol Oncol. 2003;89:227-232. Downloaded from https://academic.oup.com/ajcp/article-abstract/122/3/434/1759428 Am J Clin Pathol 2004;122:434-439 439 439 439