Evaluation of unfavorable histologic subtypes in endometrial adenocarcinoma Timothy O. Wilson, MD, Karl C. Podratz, MD, PhD, Thomas A. Gaffey, MD, George D. Malkasian, Jr., MD, Peter C. O'Brien, PhD, and James M. Naessens, MPH Rochester, Minnesota A retrospective review of 388 patients who presented to the Mayo Clinic for treatment of endometrial carcinoma between 1979 and 1983 was performed and the surgical and pathologic observations were documented. An uncommon histologic subtype was detected in 52 patients (13%): 20 adenosquamous, 14 serous papillary, 11 clear cell, 7 undifferentiated. In contrast to the survival of patients with endometrioid lesions (92%), the overall survival in these patients was only 33%. Each of the individual abnormal histologic subtypes exhibited a survival of less than 50%. At the time of surgical staging, 62% of patients with unfavorable histologic subtypes had extrauterine spread of disease. Despite liberal utilization of postoperative adjuvant therapy in 42 of the 52 patients (81%), only 10% of these patients survived 5 years. Fifty-five percent had a component of recurrence outside of the abdominal/pelvic cavity. Subsequent treatment considerations should incorporate regimens addressing systemic and local tissue control. (AM J OBSTET GVNECOL 1990;162:418-26.)
Key words: Unfavorable histologic subtypes, endometrial carcinoma, adenosquamous, serous papillary, clear cell, undifferentiated
Carcinoma of the endometrium is the most common invasive gynecologic malignancy. The American Cancer Society estimates that 40,000 cases of endometrial carcinoma will be diagnosed during 1988 and 3000 deaths will result from this disease. Because of the low death rate, endometrial carcinoma is often considered a prognostically favorable disease, but several subsets of patients can be identified who respond poorly to standard treatment. Various morphologically different histologic subtypes make up the group of malignancies classified as adenocarcinomas of the endometrium. Each of these subtypes, including adenocarcinoma with or without squamous differentiation, clear cell carcinoma, papillary serous carcinoma, adenosquamous carcinoma, mucinous adenocarcinoma, secretory carcinoma, and undifferentiated carcinoma, exhibits its own behavior pattern and metastatic potential. This report describes our experience with these histologically different variants of endometrial carcinoma.
Methods and material A retrospective review was performed of all patients presenting to the Mayo Clinic for definitive therapy of endometrial carcinoma from 1979 to 1983. During From the Department of Obstetncs and Gynecology, the SectIOn of Surgical Pathology, and the Department of Health SCIences Research, Mayo Clinic and Mayo Foundation. Presented at the Fifty-sixth Annual M eetmg of the Central ASSOCIatIOn of Obstetricians and Gynecologtsts, Salt Lake City, Utah, September
22-24, 1988.
Repnnt requests: Timothy O. WIlson, MD, Mayo Climc, 200 FIrSt St. SW, Rochester, MN 55905 .
418
this interval, 388 patients received primary institutionally standardized therapy that consisted of surgical evaluation and resection, without routine preoperative intracavitary or external beam radiation therapy. Surgical management included cytologic evaluation of peritoneal fluid or washings, total abdominal hysterectomy, bilateral salpingo-oophorectomy, and selective lymph node sampling. If extrauterine spread was detected, histologic confirmation was requested and cytoreductive surgery was performed. Although Broders' classification I was used for determination of tissue differentiation, the clinical FIGO (International Federation of Gynecology and Obstetrics) system was used for surgical staging. All pathologic slides were reviewed by one of the authors (T. A. G.). The reviewer had access to the previous pathology report but no knowledge of any clinical data, including adjuvant therapy or patient status. Histologic assessment included determination of the grade of differentiation, the depth of myometrial invasion, and the histologic subtype of the lesion. If additional specimens were needed, the paraffin blocks were recut or the archival gross specimens were retrieved for additional processing as indicated. Under circumstances when diagnostic accuracy was potentially compromised because of no or limited identifiable tumor, the original diagnostic dilatation and curettage specimen was obtained (if performed elsewhere) and again reviewed. Patients with precancerous lesions were excluded as were patients with noncarcinomatous or mixed miillerian lesions of the uterus. Histologic subtypes included were endometrioid or
Volume 162 ]\;umber 2
"ordinary" endometrial carcinoma, endometrioid with squamous differentiation, clear cell, serous papillary, adenosquamous, and undifferentiated carcinoma. Tumors that comprised two or more histologic subtypes were grouped according to the predominant type at the discretion of the reviewing pathologist. Secretory carcinoma of the endometrium is thought to be a type of endometrioid carcinoma. Previous studies"-4 included these under clear cell carcinomas but these patients fare much better than those with clear celliesions 5 and therefore are included with the endometrioid malignancies. Mucinous adenocarcinomas were infrequently encountered and were invariably found in combination with lesions of another histologic type. As their clinical course is similar to that of endometrioid lesions,b they were classified as such unless the other histologic type predominated, dictating a specific classification. In our review endometrioid lesions with squamous differentiation (adenoacanthoma) were assumed to contain squamous elements that were benign. In accordance with the literature, adenosquamous lesions contained both malignant-appearing glandular and squamous elements. 7- 1O The classification of papillary serous carcinoma was rigorous; it required the presence of a papillary architectural configuration and marked nuclear atypia. These nuclear changes included pleomorphism, hyperchromatism, and evidence of abnormal chromatin characteristics such as giant nuclei and abnormal nucleoli. Patients whose tumors did not exhibit these nuclear changes, even though morphologic features were papillary, were not considered to have a serous papillary lesion. Clear cell carcinomas were categorized as per the description detailed by Kurman and Scully.!! Undifferentiated carcinomas were high-grade lesions whose appearance consisted of sheets of tumor without glandular architecture, but they were thought to arise from the endometrium. Patient follow-up data were obtained by record review, by letter, or by telephone. If a recurrence or death was noted, records from the local physician or a death certificate was obtained whenever possible. Statistical analyses used Kaplan-Meier methods!2 for determining survival estimates and Cox regression analysis n for assessing the influence of multiple covariates. Results Of 388 patients, 285 (73%) had "ordinary" endometrioid adenocarcinoma without squamous differentiation; 50 (13%) additional patients had "ordinary" endometrioid adenocarcinoma with squamous differentiation. There was no significant difference in disease-free survival in the above groups; 92% of those without and 91 % of those with squamous differentiation were alive without evidence of recurrence at a mean follow-up of 3 years. Hence these subtypes were
Histologic subtypes in endometrial adenocarcinoma
419
Table I. Distribution of histologic subtypes in endometrial carcinoma (n = 388) Patients H 1Stologzc subtype
No.
Ordinary type without squamous differentiation Ordinary type with squamous differentiation Adenosquamous Serous papillary Clear cell Undifferentiated Unknown (postradiation)
285
73
50
13
20 14
5 4
II
7 I
%
3
2 0.3
considered to be favorable histologic lesions. One patient received preoperative radiation therapy before referral and the histologic subtype could not be accurately determined. An uncommon histologic subtype was detected in 52 patients (13%) (Table I). In contrast to the survival in patients with endometrioid lesions (92%), the 5-year survival in these patients was only 33%. Attesting to the grave prognosis associated with each of the identified subtypes were the corresponding 5-year survival rates: adenosquamous, 40%; serous papillary, 25%; clear cell, 33%; and undifferentiated, 20% (Table II). Thirty-five patients were followed up until death (mean survival, 17 months; range, 2 to 71 months). Only I patient who died of disease survived more than 40 months (71 months). Thus the overall survival is 33%. The remaining 17 patients who are alive have a mean survival of 75 months (range, 48 to 107), and only 3 patients have a posttreatment observation interval of %
Through serosa
I
Unfavorable hzstologtc subtype (n = 52)
I
%
No.
101 197 34 4
30 59 10 1
1 13 24 14
2 25 46 27
57 204 37 29 9
17 61 11 9 3
1 16 6 6 23
2 31 12 12 44
No.
%
but had stage I at operation. In the remaining 16 patients with incorrect staging, understaging was done in all. The frequency of poor tissue differentiation and deep myometrial penetration was skewed adversely within the subgroup consisting of unusual histologic subtypes. Whereas 298 of the 335 patients (89%) with endometrioid lesions exhibited Broders' grade 1 or 2 lesions, 73% (38 of 52) of the unfavorable histologic subtypes had grade 3 or 4 lesions (Table IV). These patients also experienced deeper myometrial invasion; 23 patients (44%) had tumor penetration through the uterine serosa whereas only 17 (33%) had tumor confined to the inner third of the myometrium. In contrast, 261 (78%) of the endometrioid group had tumor confined to the inner third of the myometrium, whereas only 3% of patients had penetration of the uterine serosa (Table IV). Lymph node status was assessed histologically in 23 patients (44%) with unfavorable histologic subtypes; only 9 patients had pelvic and paraaortic lymph nodes sampled. Metastatic involvement of pelvic or paraaortic lymph nodes was documented by pathologic exam i-
Histologic subtypes in endometrial adenocarcinoma 421
Volume 162 Number 2
Table V. Postoperative adjuvant therapy according to surgical stage Patients Stage I (No.)
Therapy
Progestational agents Pelvic radiation*
Stage II (No.)
Stage III (No.)
2 9t
I
9 2 I I 5 19
32p
Cytotoxic chemotherapy Extended field radiation:j: None TOTAL
It
3 2 16
Stage IV (No.)
6 2 3 2 3 16
Total No.
I
% 17 40 4 8 12 20
9 21 2 4 6 10 52
*One patient received progestational agents. tIncomplete course. :j:Whole abdominal pelvic irradiation 5 patients; paraaortic boost I patient.
Table VI. Survival at 5 years after adjuvant therapy in patients with unfavorable histologic subtypes according to stage Patients Therapy
Progestational agents Pelvic radiation 32p
Stage I (No.)
011
Stage II (No.)
011*
5/9
0/2 011 011
Cytotoxic chemotherapy Extended field radiation None
4/5
TOTAL
5114 (36)
(80)
011
Stage III (No.)
Stage IV (No.)
112
0/6
112
119 (II) 12/21 (57)
0/2
113
0/3 0/2 0/3
116 (17) 4110 (40)
8114 (57)
1113 (8)
6*/9
Total (No.)
012 0/4
14/42 (33)
Numbers in parentheses are percentages. *Incomplete course in I patient.
nation III 10 patients; 4 patients had both lymphatic regions involved. Only positive pelvic lymph nodes were observed in 4 patients and 2 patients had positive paraaortic lymph nodes without pathologically proved pelvic nodal disease. Cytologic evaluation of peritoneal fluid was performed in 43 of the 52 patients. Peritoneal cytologic results were positive in 14 (27%) and negative in 29 (56%) patients; 9 patients (17%) did not have cytologic evaluation. This is in contrast to the endometrioid lesions, in which peritoneal cytologic results were positive in 18 of335 patients (5%) and negative in 219 patients (65%), and cytologic evaluation was not performed in 98 patients (29%). The advanced surgical stage is primarily responsible for the positive peritoneal cytologic results, as 11 of the 14 positive samples were obtained in surgical stage III or IV disease and three positive samples were obtained in 19 patients with surgical stage I disease, with one patient not having samples obtained for peritoneal cytologic testing. Postoperative adjuvant therapy was used in 42 of the 52 patients (81 %) with unfavorable histologic subtypes in contrast to 32% of patients with endometrioid subtypes. The decision regarding adjuvant therapy was
based on the stage of disease, tumor grade, depth of myometrial invasion, and peritoneal cytologic test results. Several patients refused adjuvant therapy. This therapy consisted of the following: progestational agents in 9 patients, pelvic radiation in 21, radioactive colloid chromic phosphate in 2, cytotoxic chemotherapy in 4, and extended field radiation in 6 (Table V). Notwithstanding relatively aggressive adjuvant management with progestational agents, cytotoxic chemotherapy, intraperitoneal radioisotopes, and extended field radiation, the 5-year survival in these patients was poor (10%). Only 2 patients were salvaged among the 21 treated, which, in part, may reflect selection biases; only 5 patients (24%) had surgical stage I disease (Table VI). One of these 2 had a recurrence and died of disease at 71 months. Of the 21 patients treated with pelvic radiation, 12 (57%) survived. In contrast to the former group, 10 patients (48%) did not exhibit extrauterine spread of disease. Surprisingly, 4 of 10 patients who did not receive adjuvant therapy survived, but cautious interpretation is necessary in that all survivors were patients with stage I disease, which made up 50% of this subgroup. Evaluation of the failure sites allo~s additional in-
422 Wilson et al.
February 1990 Am J Obstet Gynecol
Table VII. Distribution according to sites of recurrence (n = 36) Szte
Patients (No.)
Unknown* Pelvic alone Abdomen alone Abdomen and pelvis Distant site only Distant site and abdominal-pelvic component
7 4 2 7 9 7
*Includes 5 patients who possibly died of endometrial carcinoma.
sight into biologic characteristics of the tumor and treatment efficacy. Seven patients failing progestational therapy for unfavorable histologic subtypes had an identifiable site of recurrence. Each of these 7 had an abdominal or pelvic component of the failure and 4 also had detectable disease present outside the abdominal cavity. Likewise, 3 of the 4 patients who received cytotoxic chemotherapy had an identifiable site of recurrence. Of these 3, all had an abdominal or pelvic component of the failure, and 1 had an extraabdominal component as well. Of the 2 patients treated with 32p, 1 had failure in the abdomen and the other in the chest. In addition, 6 patients were treated with extended field radiation, with four known sites of recurrence, including three recurrences within the field of radiation. Of the 21 patients treated with pelvic radiation, 2 did not complete therapy. The mean dose was 45.1 Gy to the mid pelvis, with a range of 20 to 54 Gy. If the 2 patients who did not complete therapy are excluded, the mean dose was 47.5 Gy (range, 43 to 54). It is interesting that 1 of these 2 patients is alive without evidence of disease. Overall, 12 of these patients (57%) survived, and 1 died without documented recurrence. One patient had recurrence in the pelvis and was treated with a total pelvic exenteration; she is surviving without evidence of disease 18 months after recurrence. Therefore 9 known recurrences were documented in this group, with 5 outside the abdominalpelvic cavity, 3 in the pelvis alone, and a combination of both pelvic and distant sites in 1. Of the 10 patients who did not receive adjuvant therapy, 4 are alive without known recurrence and 2 died without confirmed recurrence of endometrial carcinoma. The sites of recurrence in the remaining 4 patients include the abdominal-pelvic cavity in 1, the abdominal-pelvic cavity and distant sites in 2, and an isolated distant site in 1. Eleven patients who had surgical stage I disease had recurrence: in the abdominal-pelvic cavity in 4, in distant sites in 5, and in both the abdominal-pelvic cavity and a distant site in 1; the site of recurrence was unknown in 1 patient.
Overall, there were 31 recurrences resulting in 30 deaths, but in 2 the sites of recurrence were unknown. Thus, of the 29 known sites of recurrence, 16 (55%) had a component of failure outside the abdominalpelvic cavity, and in 9 (31 %) of these recurrence was outside the abdomen and pelvis. Only 4 patients (14%) experienced an isolated pelvic recurrence and 9 (31 %) had recurrence in either the abdomen or pelvis or in the abdomen alone (Table VII). Comment
Patients with endometrial carcinoma of unfavorable histologic SUbtypes experience a 5-year survival that is lower than that of their cohorts with endometrioid lesions. Independent and multivariate analyses (adjusting for surgical stage, grade, lymph node involvement, estrogen supplementation, and lower uterine segment involvement) revealed that the histologic subtype had a significant correlation with survival (p < 0.001 and
Key words: Unfavorable histologic subtypes, endometrial carcinoma, adenosquamous, serous papillary, clear cell, undifferentiated
Carcinoma of the endometrium is the most common invasive gynecologic malignancy. The American Cancer Society estimates that 40,000 cases of endometrial carcinoma will be diagnosed during 1988 and 3000 deaths will result from this disease. Because of the low death rate, endometrial carcinoma is often considered a prognostically favorable disease, but several subsets of patients can be identified who respond poorly to standard treatment. Various morphologically different histologic subtypes make up the group of malignancies classified as adenocarcinomas of the endometrium. Each of these subtypes, including adenocarcinoma with or without squamous differentiation, clear cell carcinoma, papillary serous carcinoma, adenosquamous carcinoma, mucinous adenocarcinoma, secretory carcinoma, and undifferentiated carcinoma, exhibits its own behavior pattern and metastatic potential. This report describes our experience with these histologically different variants of endometrial carcinoma.
Methods and material A retrospective review was performed of all patients presenting to the Mayo Clinic for definitive therapy of endometrial carcinoma from 1979 to 1983. During From the Department of Obstetncs and Gynecology, the SectIOn of Surgical Pathology, and the Department of Health SCIences Research, Mayo Clinic and Mayo Foundation. Presented at the Fifty-sixth Annual M eetmg of the Central ASSOCIatIOn of Obstetricians and Gynecologtsts, Salt Lake City, Utah, September
22-24, 1988.
Repnnt requests: Timothy O. WIlson, MD, Mayo Climc, 200 FIrSt St. SW, Rochester, MN 55905 .
418
this interval, 388 patients received primary institutionally standardized therapy that consisted of surgical evaluation and resection, without routine preoperative intracavitary or external beam radiation therapy. Surgical management included cytologic evaluation of peritoneal fluid or washings, total abdominal hysterectomy, bilateral salpingo-oophorectomy, and selective lymph node sampling. If extrauterine spread was detected, histologic confirmation was requested and cytoreductive surgery was performed. Although Broders' classification I was used for determination of tissue differentiation, the clinical FIGO (International Federation of Gynecology and Obstetrics) system was used for surgical staging. All pathologic slides were reviewed by one of the authors (T. A. G.). The reviewer had access to the previous pathology report but no knowledge of any clinical data, including adjuvant therapy or patient status. Histologic assessment included determination of the grade of differentiation, the depth of myometrial invasion, and the histologic subtype of the lesion. If additional specimens were needed, the paraffin blocks were recut or the archival gross specimens were retrieved for additional processing as indicated. Under circumstances when diagnostic accuracy was potentially compromised because of no or limited identifiable tumor, the original diagnostic dilatation and curettage specimen was obtained (if performed elsewhere) and again reviewed. Patients with precancerous lesions were excluded as were patients with noncarcinomatous or mixed miillerian lesions of the uterus. Histologic subtypes included were endometrioid or
Volume 162 ]\;umber 2
"ordinary" endometrial carcinoma, endometrioid with squamous differentiation, clear cell, serous papillary, adenosquamous, and undifferentiated carcinoma. Tumors that comprised two or more histologic subtypes were grouped according to the predominant type at the discretion of the reviewing pathologist. Secretory carcinoma of the endometrium is thought to be a type of endometrioid carcinoma. Previous studies"-4 included these under clear cell carcinomas but these patients fare much better than those with clear celliesions 5 and therefore are included with the endometrioid malignancies. Mucinous adenocarcinomas were infrequently encountered and were invariably found in combination with lesions of another histologic type. As their clinical course is similar to that of endometrioid lesions,b they were classified as such unless the other histologic type predominated, dictating a specific classification. In our review endometrioid lesions with squamous differentiation (adenoacanthoma) were assumed to contain squamous elements that were benign. In accordance with the literature, adenosquamous lesions contained both malignant-appearing glandular and squamous elements. 7- 1O The classification of papillary serous carcinoma was rigorous; it required the presence of a papillary architectural configuration and marked nuclear atypia. These nuclear changes included pleomorphism, hyperchromatism, and evidence of abnormal chromatin characteristics such as giant nuclei and abnormal nucleoli. Patients whose tumors did not exhibit these nuclear changes, even though morphologic features were papillary, were not considered to have a serous papillary lesion. Clear cell carcinomas were categorized as per the description detailed by Kurman and Scully.!! Undifferentiated carcinomas were high-grade lesions whose appearance consisted of sheets of tumor without glandular architecture, but they were thought to arise from the endometrium. Patient follow-up data were obtained by record review, by letter, or by telephone. If a recurrence or death was noted, records from the local physician or a death certificate was obtained whenever possible. Statistical analyses used Kaplan-Meier methods!2 for determining survival estimates and Cox regression analysis n for assessing the influence of multiple covariates. Results Of 388 patients, 285 (73%) had "ordinary" endometrioid adenocarcinoma without squamous differentiation; 50 (13%) additional patients had "ordinary" endometrioid adenocarcinoma with squamous differentiation. There was no significant difference in disease-free survival in the above groups; 92% of those without and 91 % of those with squamous differentiation were alive without evidence of recurrence at a mean follow-up of 3 years. Hence these subtypes were
Histologic subtypes in endometrial adenocarcinoma
419
Table I. Distribution of histologic subtypes in endometrial carcinoma (n = 388) Patients H 1Stologzc subtype
No.
Ordinary type without squamous differentiation Ordinary type with squamous differentiation Adenosquamous Serous papillary Clear cell Undifferentiated Unknown (postradiation)
285
73
50
13
20 14
5 4
II
7 I
%
3
2 0.3
considered to be favorable histologic lesions. One patient received preoperative radiation therapy before referral and the histologic subtype could not be accurately determined. An uncommon histologic subtype was detected in 52 patients (13%) (Table I). In contrast to the survival in patients with endometrioid lesions (92%), the 5-year survival in these patients was only 33%. Attesting to the grave prognosis associated with each of the identified subtypes were the corresponding 5-year survival rates: adenosquamous, 40%; serous papillary, 25%; clear cell, 33%; and undifferentiated, 20% (Table II). Thirty-five patients were followed up until death (mean survival, 17 months; range, 2 to 71 months). Only I patient who died of disease survived more than 40 months (71 months). Thus the overall survival is 33%. The remaining 17 patients who are alive have a mean survival of 75 months (range, 48 to 107), and only 3 patients have a posttreatment observation interval of %
Through serosa
I
Unfavorable hzstologtc subtype (n = 52)
I
%
No.
101 197 34 4
30 59 10 1
1 13 24 14
2 25 46 27
57 204 37 29 9
17 61 11 9 3
1 16 6 6 23
2 31 12 12 44
No.
%
but had stage I at operation. In the remaining 16 patients with incorrect staging, understaging was done in all. The frequency of poor tissue differentiation and deep myometrial penetration was skewed adversely within the subgroup consisting of unusual histologic subtypes. Whereas 298 of the 335 patients (89%) with endometrioid lesions exhibited Broders' grade 1 or 2 lesions, 73% (38 of 52) of the unfavorable histologic subtypes had grade 3 or 4 lesions (Table IV). These patients also experienced deeper myometrial invasion; 23 patients (44%) had tumor penetration through the uterine serosa whereas only 17 (33%) had tumor confined to the inner third of the myometrium. In contrast, 261 (78%) of the endometrioid group had tumor confined to the inner third of the myometrium, whereas only 3% of patients had penetration of the uterine serosa (Table IV). Lymph node status was assessed histologically in 23 patients (44%) with unfavorable histologic subtypes; only 9 patients had pelvic and paraaortic lymph nodes sampled. Metastatic involvement of pelvic or paraaortic lymph nodes was documented by pathologic exam i-
Histologic subtypes in endometrial adenocarcinoma 421
Volume 162 Number 2
Table V. Postoperative adjuvant therapy according to surgical stage Patients Stage I (No.)
Therapy
Progestational agents Pelvic radiation*
Stage II (No.)
Stage III (No.)
2 9t
I
9 2 I I 5 19
32p
Cytotoxic chemotherapy Extended field radiation:j: None TOTAL
It
3 2 16
Stage IV (No.)
6 2 3 2 3 16
Total No.
I
% 17 40 4 8 12 20
9 21 2 4 6 10 52
*One patient received progestational agents. tIncomplete course. :j:Whole abdominal pelvic irradiation 5 patients; paraaortic boost I patient.
Table VI. Survival at 5 years after adjuvant therapy in patients with unfavorable histologic subtypes according to stage Patients Therapy
Progestational agents Pelvic radiation 32p
Stage I (No.)
011
Stage II (No.)
011*
5/9
0/2 011 011
Cytotoxic chemotherapy Extended field radiation None
4/5
TOTAL
5114 (36)
(80)
011
Stage III (No.)
Stage IV (No.)
112
0/6
112
119 (II) 12/21 (57)
0/2
113
0/3 0/2 0/3
116 (17) 4110 (40)
8114 (57)
1113 (8)
6*/9
Total (No.)
012 0/4
14/42 (33)
Numbers in parentheses are percentages. *Incomplete course in I patient.
nation III 10 patients; 4 patients had both lymphatic regions involved. Only positive pelvic lymph nodes were observed in 4 patients and 2 patients had positive paraaortic lymph nodes without pathologically proved pelvic nodal disease. Cytologic evaluation of peritoneal fluid was performed in 43 of the 52 patients. Peritoneal cytologic results were positive in 14 (27%) and negative in 29 (56%) patients; 9 patients (17%) did not have cytologic evaluation. This is in contrast to the endometrioid lesions, in which peritoneal cytologic results were positive in 18 of335 patients (5%) and negative in 219 patients (65%), and cytologic evaluation was not performed in 98 patients (29%). The advanced surgical stage is primarily responsible for the positive peritoneal cytologic results, as 11 of the 14 positive samples were obtained in surgical stage III or IV disease and three positive samples were obtained in 19 patients with surgical stage I disease, with one patient not having samples obtained for peritoneal cytologic testing. Postoperative adjuvant therapy was used in 42 of the 52 patients (81 %) with unfavorable histologic subtypes in contrast to 32% of patients with endometrioid subtypes. The decision regarding adjuvant therapy was
based on the stage of disease, tumor grade, depth of myometrial invasion, and peritoneal cytologic test results. Several patients refused adjuvant therapy. This therapy consisted of the following: progestational agents in 9 patients, pelvic radiation in 21, radioactive colloid chromic phosphate in 2, cytotoxic chemotherapy in 4, and extended field radiation in 6 (Table V). Notwithstanding relatively aggressive adjuvant management with progestational agents, cytotoxic chemotherapy, intraperitoneal radioisotopes, and extended field radiation, the 5-year survival in these patients was poor (10%). Only 2 patients were salvaged among the 21 treated, which, in part, may reflect selection biases; only 5 patients (24%) had surgical stage I disease (Table VI). One of these 2 had a recurrence and died of disease at 71 months. Of the 21 patients treated with pelvic radiation, 12 (57%) survived. In contrast to the former group, 10 patients (48%) did not exhibit extrauterine spread of disease. Surprisingly, 4 of 10 patients who did not receive adjuvant therapy survived, but cautious interpretation is necessary in that all survivors were patients with stage I disease, which made up 50% of this subgroup. Evaluation of the failure sites allo~s additional in-
422 Wilson et al.
February 1990 Am J Obstet Gynecol
Table VII. Distribution according to sites of recurrence (n = 36) Szte
Patients (No.)
Unknown* Pelvic alone Abdomen alone Abdomen and pelvis Distant site only Distant site and abdominal-pelvic component
7 4 2 7 9 7
*Includes 5 patients who possibly died of endometrial carcinoma.
sight into biologic characteristics of the tumor and treatment efficacy. Seven patients failing progestational therapy for unfavorable histologic subtypes had an identifiable site of recurrence. Each of these 7 had an abdominal or pelvic component of the failure and 4 also had detectable disease present outside the abdominal cavity. Likewise, 3 of the 4 patients who received cytotoxic chemotherapy had an identifiable site of recurrence. Of these 3, all had an abdominal or pelvic component of the failure, and 1 had an extraabdominal component as well. Of the 2 patients treated with 32p, 1 had failure in the abdomen and the other in the chest. In addition, 6 patients were treated with extended field radiation, with four known sites of recurrence, including three recurrences within the field of radiation. Of the 21 patients treated with pelvic radiation, 2 did not complete therapy. The mean dose was 45.1 Gy to the mid pelvis, with a range of 20 to 54 Gy. If the 2 patients who did not complete therapy are excluded, the mean dose was 47.5 Gy (range, 43 to 54). It is interesting that 1 of these 2 patients is alive without evidence of disease. Overall, 12 of these patients (57%) survived, and 1 died without documented recurrence. One patient had recurrence in the pelvis and was treated with a total pelvic exenteration; she is surviving without evidence of disease 18 months after recurrence. Therefore 9 known recurrences were documented in this group, with 5 outside the abdominalpelvic cavity, 3 in the pelvis alone, and a combination of both pelvic and distant sites in 1. Of the 10 patients who did not receive adjuvant therapy, 4 are alive without known recurrence and 2 died without confirmed recurrence of endometrial carcinoma. The sites of recurrence in the remaining 4 patients include the abdominal-pelvic cavity in 1, the abdominal-pelvic cavity and distant sites in 2, and an isolated distant site in 1. Eleven patients who had surgical stage I disease had recurrence: in the abdominal-pelvic cavity in 4, in distant sites in 5, and in both the abdominal-pelvic cavity and a distant site in 1; the site of recurrence was unknown in 1 patient.
Overall, there were 31 recurrences resulting in 30 deaths, but in 2 the sites of recurrence were unknown. Thus, of the 29 known sites of recurrence, 16 (55%) had a component of failure outside the abdominalpelvic cavity, and in 9 (31 %) of these recurrence was outside the abdomen and pelvis. Only 4 patients (14%) experienced an isolated pelvic recurrence and 9 (31 %) had recurrence in either the abdomen or pelvis or in the abdomen alone (Table VII). Comment
Patients with endometrial carcinoma of unfavorable histologic SUbtypes experience a 5-year survival that is lower than that of their cohorts with endometrioid lesions. Independent and multivariate analyses (adjusting for surgical stage, grade, lymph node involvement, estrogen supplementation, and lower uterine segment involvement) revealed that the histologic subtype had a significant correlation with survival (p < 0.001 and
