Professional Documents
Culture Documents
Original Article
Xiaowei Chen, M.D., Rebecca Arend, M.D., Diane Hamele-Bena, M.D., Ana I. Tergas, M.D., MPH,
Melanie Hawver, M.D., Guo-Xia Tong, M.D., Thomas C. Wright, M.D.,
and Jason D. Wright, M.D.
From the Departments of Pathology and Cell Biology Carcinosarcomas, otherwise referred to as malignant
(X.C., D.H.-B., T.C.W.); Obstetrics and Gynecology, Division of mixed Mullerian tumors or MMMT, are rare highly
Gynecology Oncology (A.I.T., J.D.W.), Columbia University aggressive malignant tumors of the female genital tract.
Medical Center; Department of Pathology and Laboratory
Medicine, Staten Island University Medical Center, Northwell These tumors are biphasic, containing both malignant
Health (G.-X.T.), New York; Department of Pathology, St. Peter epithelial and mesenchymal components. The prog-
Hospital, Albany (M.H.), New York; and Department of nosis of carcinosarcomas has been reported to depend
Obstetrics and Gynecology, Alabama University of Alabama
School Medicine, Birmingham, Alabama (R.A.). on the histologic type of the carcinomatous compo-
The authors declare no conflicts of interest. nents, depth of myometrial invasion, and presence of
Address correspondence and reprint requests to Xiaowei Chen, lymphovascular and cervical involvement (1).
MD, Columbia University Medical Center, Department of
Pathology, 630 West 168th Street, VC14-215, New York, NY Although several studies have been conducted, the
10032. E-mail: xc2159@cumc.columbia.edu. histogenesis of uterine carcinosarcomas still is not
Supplemental Digital Content is available for this article. Direct completely understood. Two major theories of
URL citations appear in the printed text and are provided in the
HTML and PDF versions of this article on the journal’s Website, histogenesis have been proposed: the ‘‘collision’’ or
www.intjgynpathology.com. ‘‘multiclonal’’ hypothesis, which proposes that the
epithelial and mesenchymal components are distinct, confirm the diagnoses based on established criteria
coexisting populations with different cell origins, and of the tumor containing malignant epithelial and
the alternative ‘‘monoclonal’’ hypothesis, which malignant stromal components (Fig. 1).
suggests that carcinosarcomas arise from multipotent A representative formalin-fixed and paraffin-
stem cells differentiating along epithelial and mesen- embedded block containing both malignant epithelial
chymal pathways. The majority of studies favor the and mesenchymal components was retrieved from
‘‘monoclonal’’ theory supporting a common clonal each case. Immunohistochemical staining for p16
origin of both histologic components (2–5). In the (clone E6H4, prediluted; Ventana, Tucson, AZ), p53
World Health Organization tumor classification (clone DO7, confirm, prediluted; Ventana), and
schema, these tumors have been redesignated as PAX8 (rabbit polyclonal, prediluted; Cell Marque,
endometrial carcinosarcoma to reflect the concept Rockline, CA) were performed with proper positive
that they are primarily epithelial tumors that have and negative controls on a Ventana Medical Systems
developed a mesenchymal component (6). automated stainer according to the manufacturer’s
Although the clinical course of carcinosarcomas is manual. Although p16 is known to stain both the
similar to that of high-grade endometrioid cancers and cytoplasm and nucleus, for all 3 stains, only nuclear
serous cancer, it is important to separate carcinosarco- staining was interpreted as positive, in accordance
mas from other high-grade endometrial adenocarcino- with other reports (8–10). Percentage of positive
mas because carcinosarcomas have a worse prognosis. staining (P) and intensity of staining (I) were
Recently, 2 markers, p16 and p53, have been used in determined separately for epithelial and mesenchymal
the diagnosis of high-grade endometrial adenocarcino- components. Cases were scored as follows: for
ma. However, few studies have described the immu- percentage of tumor cells staining, P = 0 if <5% of
nohistochemical profile of carcinosarcomas. Meis and cells; P = 1 if 5% to 10% of cells; P = 2 if 11% to
Lawrence’s (7) study demonstrated overlapping im- 50% of cells; and P = 3 if >51% of cells. For
munohistochemical findings between carcinosarcomas immunostaining intensity, I = 0 if no staining; I = 1
and endometrioid adenocarcinomas and concluded if mild intensity; I = 2 if moderate intensity; I = 3 if
that immunohistochemistry was often not reliable or strong intensity. A staining index was calculated as
useful in distinguishing between the 2 malignant the product of staining area and staining intensity.
tumors. Some studies (8) have shown that pathologic The tumors were categorized as high-expressing
expression of p53 and p16 in curettage material (staining index >4) or low-expressing (staining index
identified high-risk endometrial carcinoma patients r4), utilizing a previously published method (11).
with poor prognosis. Buza and Tavassoli (9) found
p16 and p53 immunoreactivity to be concordant
between the 2 components of the same carcinosarcoma Statistical Methods
tumor in 90% and 83% of cases, respectively. Staining intensity and percent of the tumor cells
Reviewing the literature, we found very limited were used to summarize the study data. Contingency
information regarding PAX8 expression in carcinosar-
comas (10). The aim of this study is to examine the
possible utility of p16, p53, and PAX8 immunohisto-
chemistry in the diagnosis of carcinosarcomas and their
potential role in the histogenesis of these tumors.
tables and the Fisher exact test were used to test the >50% myometrial invasion (51.6%); included in these
association of epithelial and stromal p16, p53, PAX8 were 6 cases with full thickness myometrial invasion.
expression with demographics and clinical baseline Within the 62 cases, 2 cases had no cervical evaluation
characteristics. because the cervix in one of these was not identified and
Overall survival was calculated as the time from the another was a supracervical hysterectomy. Twenty-three
date of surgery to the date of death of any cause or date of the remaining 60 cases showed cervical involvement
that the patient was last known to be alive. Recurrence- (38.3%). One case was a total hysterectomy; therefore,
free survival was calculated from the date of surgery to only 61 of 62 cases had ovary and tube evaluation, among
the date of recurrence or death, whichever came first. which 17 cases had tumor involvement (27.9%). Thirty-
Patients who died before recurrence were censored at two of the 62 cases had omentectomies or pelvic side wall
the time of death. The Kaplan-Meier plots (12) were biopsies, among which 9 cases had tumor involvement
used to estimate and illustrate the probabilities of (28.1%). Lymphovascular space involvement was identi-
overall survival and recurrence-free survival. In the fied in the majority of the cases (80.6%). Forty-three of
univariate analysis, the log-rank test (12) was used to the 62 cases had lymph node evaluation, among which 32
assess the association of overall survival and recurrence- cases had negative lymph nodes (74.4%), and 11 cases
free survival with age at diagnosis, obesity, nulliparity, had metastatic carcinoma (25.6%). Forty-six of the 62
FIGO (International Federation of Gynecology and cases had pelvic washings, among which 35 cases were
Obstetrics) stage at diagnosis, histologic grade, stromal benign (76.1%) and 11 cases contained malignant cells
type, and depth of invasion. Those factors with P- (23.9%). Clinical information and pathologic data are
values r0.20 in the univariate analysis were included in summarized in Tables 1 and 2, respectively.
the Cox proportional hazards model for multivariable The staining patterns and staining indexes of p16,
analysis (12). p53, and PAX8 in epithelial and stromal components
All statistical tests were 2-sided, and P-values of are shown in Table 3. p16 staining (Fig. 2) showed
0.05 or less were considered to indicate statistical almost equal high-expression in epithelial (74%) and
significance. mesenchymal components (71%), and p53 expression
(Fig. 3) also was similar in epithelial (48%) and
stromal (44%) components. High-expression of
RESULTS
PAX8 (Fig. 4) was more common in the epithelial
The ages of the 62 patients ranged from 34 to 87 yr (73%) than mesenchymal (13%) components, and
old (median: 67.3 yr). The majority of patients were PAX8 stromal positivity was never seen in the
postmenopausal and presented with bleeding. Other absence of PAX8 epithelial positivity.
presentations included pelvic/abdominal pain or There were no significant associations among the
bloating. Almost half (30) of the patients were epithelial and stromal p16, p53, and PAX8 expression
clinically obese and only 8 were nulliparous. with demographics and clinical characteristics (data
All cases were surgical resection specimens. Tumor not shown).
stages were as follows: 23 tumors were stage I (37.1%), 10 Tables 4 and 5 show the associations of demographics
were stage II (16.1%), 19 were stage III (30.7%), and 10 and baseline clinical characteristics with overall survival
were stage IV (16.1%). Of the 62 cases, the carcinomatous
component was endometrioid type in 41 cases (66%) and TABLE 1. Summary of Clinical Information
nonendometrioid type in 21 cases (34%). Nonendome-
Demographics No. Patients % of Patients
trioid cases included serous carcinoma, clear cell carcino-
ma, and mixed types of serous/clear cell carcinoma and Age (yr)
o65 24 38.71
other type of carcinoma. The epithelial component Z65 38 61.29
appeared to be poorly differentiated adenocarcinoma in Race
50 cases (80.7%), moderately differentiated in 11 cases White 27 43.55
Nonwhite 35 56.45
(17.7%), and well differentiated in 1 case (1.6%). The Obesity
sarcomatous component was homologous in 33 cases No 19 38.78
(53.2%) and heterologous in 29 cases (46.8%). Of the Yes 30 61.22
(Unknown) (13)
tumors with heterologous stromal components, most Nulliparity
were rhabdomyosarcomas (44.8%) and chondrosarcomas No 51 86.44
(34.5%). In 30 cases, <50% myometrial invasion was Yes 8 13.56
(Unknown) (3)
identified (48.4%). The remaining 32 cases showed
TABLE 4. Association of Demographics and Baseline Clinical Characteristics With Overall Survival and Recurrence-free
Survival in Uterine Carcinosarcomas (Univariable Analysis)
Overall Survival (mo) Recurrence-free Survival (mo)
Prob. of Overall Prob. of Recurrence-free
Factors N Median (95% CI) Survival at 5 yr ± SE P* Median (95% CI) Survival at 5 yr ± SE P*
Overall 62 20.9 (15.9, 43.2) 0.32 ± 0.07 15.9 (9.4, 22.6) 0.19 ± 0.05
Age (yr)
o 65 24 Not reached 0.57 ± 0.11 0.008 27.4 (7.9, 161.0+) 0.35 ± 0.10 0.017
Z65 38 17.4 (15.7, 24.3) 0.16 ± 0.07 15.9 (8.5, 20.9) 0.10 ± 0.05
Obesity
No 20 17.4 (5.4, 34.1) 0.24 ± 0.11 0.19 9.9 (3.6, 20.9) 0.11 ± 0.07 0.056
Yes 30 36.1 (15.7, 129.8) 0.44 ± 0.10 24.6 (10.6, 43.2) 0.31 ± 0.09
Nulliparity
No 51 20.8 (12.4, 36.1) 0.29 ± 0.07 0.34 15.6 (8.6, 22.6) 0.19 ± 0.06 0.50
Yes 8 129.8 (1.0, 129.8) 0.53 ± 0.20 16.5 (0.4, 41.6) 0.13 ± 0.12
FIGO stage at diagnosis
I/II 33 36.1 (20.4, 107.0) 0.40 ± 0.09 0.033 23.5 (15.9, 43.2) 0.28 ± 0.08 0.005
III/IV 29 13.6 (8.8, 22.6) 0.24 ± 0.09 8.6 (5.7, 11.5) 0.11 ± 0.06
Histologic grade
Well/moderately 12 24.3 (6.1, 117.9) 0.40 ± 0.16 0.71 17.9 (6.1, 24.3) 0.08 ± 0.08 0.51
differentiated
Poorly differentiated 50 20.4 (13.6, 43.2) 0.30 ± 0.07 15.7 (9.4, 27.4) 0.22 ± 0.06
Stromal type
Homologous 33 20.4 (10.8, 107.0) 0.35 ± 0.08 0.91 17.1 (8.6, 27.4) 0.26 ± 0.08 0.27
Heterologous 29 32.9 (16.3, 54.2) 0.28 ± 0.11 15.7 (6.9, 23.5) 0.10 ± 0.06
Depth of invasion (%)
o50 30 43.2 (17.2, 107.0) 0.45 ± 0.10 0.065 24.3 (10.6, 49.6) 0.29 ± 0.09 0.021
Z50 32 16.3 (11.5, 22.6) 0.20 ± 0.08 9.9 (6.9, 18.2) 0.10 ± 0.06
*On the basis of logrank test.
CI indicates confidence interval.
TABLE 5. Association of Demographics and Baseline Clinical Characteristics With Overall Survival and Recurrence-free
Survival in Uterine Carcinosarcomas (Multivariable Analysis)
Overall Survival (mo) Recurrence-free Survival (mo)
Factors N HR (95% CI) P* HR (95% CI) P*
Age (yr)
o65 24 1.00 0.002 1.00 o0.001
Z65 38 4.67 (1.80, 12.15) 4.69 (2.07, 10.64)
Obesity
Yes 30 1.00 0.042 1.00 0.003
No 20 2.38 (1.03, 5.46) 3.19 (1.47, 6.90)
FIGO stage at diagnosis
I/II 33 1.00 0.019 1.00 0.002
III/IV 29 3.38 (1.22, 9.32) 4.42 (1.76, 11.10)
Depth of invasion (%)
o50 30 1.00 0.69 1.00 0.76
Z50 32 0.83 (0.33, 2.09) 0.88 (0.40, 1.96)
*On the basis of Wald w2 test from Cox proportional model including those factors with a P-value r0.20 in the univariate analysis (ie, age
at diagnosis, obesity, FIGO stage, and depth of invasion).
CI indicates confidence interval; HR, hazard ratio.
overall survival, other studies have shown the subtypes some subtypes of endometrial carcinomas, for example,
and grades of the epithelial component in carcinosarco- squamous cells carcinoma (18), serous carcinomas (19),
mas to be significant predictors of survival (1,13). clear cell carcinomas (20), and high-grade endometrioid
However, the prognostic relationship between histologic adenocarcinoma (21). Semczuk et al.’s study (21)
subtypes of the sarcomatous component is controver- suggested that p16 inactivation may play a role in the
sial. Norris et al.’s (14) and Ferguson et al.’s (15) studies tumorigenesis of a subset of sporadic endometrial
suggested that the presence of certain heterologous carcinomas, particularly in cases exhibiting aggressive
sarcomatous components portend a poor prognosis, but clinical behavior. Tsuda et al. (22) proposed that p16
a large GOG study (1) in 1990 found no significant expression may be an early event in the neoplastic
relationship between heterologous elements and surviv- transformation of endometrial cancer. Buza and
al. In our study, we did not identify a significant Tavassoli (9) reported that p16 was diffusely expressed
relationship between the type of stromal differentiation in carcinosarcoma, indicating that it has an important
and overall survival. role in its tumorigenesis. The p16 staining pattern in
Although there have been many studies reporting our study was similar to those previous reports, and
immunohistochemical phenotypes in endometrial demonstrates that p16 high-expression is not only seen
adenocarcinomas, there are limited studies regarding in uterine serous carcinomas, but in carcinosarcomas,
carcinocarcomas. Mayall et al. (16) suggested that as well.
concordant staining between the epithelial and It is well known that p53 plays an important role in
stromal components reflected a monoclonal origin. the tumorigeneses of many neoplasms. Engelsen et al.’s
Chiesa-Vottero et al.’s study (17) demonstrated study (8) revealed that pathologic expression of p53
overexpression of p16 and p53 in high-grade endo- and p16 was associated with poor prognosis in
metrial and serous adenocarcinomas in the uterus endometrial carcinoma. Studies have shown that p53
and ovary; however, only Buza and Tavassoli’s mutation and LOH for TP53 occur frequently in
study (9) focused on p16 and p53 in carcinosarcomas. carcinosarcomas, and this was used for clonal study of
It is widely accepted that p16 is overexpressed in the 2 tumor components (23). Previous studies have
human papilloma virus–related cervical squamous and suggested that mutation of tumor suppressor gene p53
glandular lesions due to functional inactivation of pRb and consequent overexpression may play an important
by human papilloma virus viral oncoproteins and lack role early on in the tumorigenesis of carcino-
of negative feedback on the p16 promoter. Therefore, sarcoma (24–27). Szukala et al. (28) demonstrated
positive p16 immunoreactivity in gynecologic tumors concordant p53 staining in carcinomatous and
has been considered to be an indication of cervical sarcomatous elements of carcinosarcomas in both
origin. Recent immunohistochemical studies, however, primary and metastatic sites and then suggested a
have shown that p16 overexpression may be seen monoclonal origin. Similarly, we found no significant
independent of human papilloma virus infection in difference in p53 staining in both malignant components
20. Reid-Nicholson M, Iyengar P, Hummer AJ, et al. Immuno- 26. Lax SF. Molecular genetic changes in epithelial, stromal, and
phenotypic diversity of endometrial adenocarcinomas: impli- mixed neoplasms of the endometrium. Pathology 2007;39:46–54.
cations for differential diagnosis. Mod Pathol 2006;19: 27. Liu FS, Kohler MF, Marks JR, et al. Mutation and
1091–100. overexpression of the p53 tumor suppressor gene frequently
21. Semczuk A, Boltze C, Marzec B, et al. p16INK4A alterations occurs in uterine and ovarian sarcomas. Obstet Gynecol
are accompanied by aberrant protein immunostaining in 1994;83:118–24.
endometrial carcinomas. J Cancer Res Clin Oncol 2003;129: 28. Szukala SA, Marks JR, Burchette JL, et al. Co-expression of
589–96. p53 by epithelial and stromal elements in carcinosarcoma of
22. Tsuda H, Yamamoto K, Inoue T, et al. The role of p16-cyclin the female genital tract: an immunohistochemical study of 19
D/CDK-pRb pathway in the tumorigenesis of endometrioid- cases. Int J Gynecol Cancer 1999;9:131–6.
type endometrial carcinoma. Br J Cancer 2000;82:675–82. 29. Tong GX, Devaraj K, Hamele-Bena D, et al. PAX8: a marker
23. Boyd J. Molecular genetics of gynecologic cancer. In: Morrow for carcinoma of Müllerian origin in serous effusions. Diagn
CP, Curtin JP, eds. Synopsis of Gynecologic Oncology, 5th ed. Cytopathol 2011;39:567–74.
New York, NY: Churchill Livingstone; 1998:515–31. 30. Mhawech-Fauceglia P, Wang D, Menesses T, et al. PAX8 is a
24. Gallardo A, Matias-Guiu X, Lagarda H, et al. Malignant reliable marker in making the diagnosis in advanced stage
mullerian mixed tumor arising from ovarian serous carcinoma: epithelial ovarian carcinoma and primary peritoneal carcinoma
a clinicopathologic and molecular study of two cases. Int J for neoadjuvant chemotherapy on cell block and biopsy
Gynecol Pathol 2002;21:268–72. specimens. Histopathology 2012;60:1019–20.
25. Taylor NP, Zighelboim I, Huettner PC, et al. Mismatch repair 31. Brunner AH, Riss P, Heinze G, et al. Immunoexpression of
and TP53 defects are early events in uterine carcinosarcoma PAX8 in endometrial cancer: relation to high-grade carcinoma
tumorigenesis. Mod Pathol 2006;19:1333–8. and p53. Int J Gynecol Pathol 2011;30:569–75.