Computed Tomography in Endometrial
Carcinoma
JOSEPH P. CONNOR, MD, JANET I. ANDREWS, MD, BARRIE ANDERSON, MD, AND
RICHARD E. BULLER, MD, PhD
Objective: To determine the value of computed tomography
(CT) scans for preoperatively detecting extrauterine-nodal
disease and postoperative recurrent disease in patients with
endometrial cancer.
Methods: We reviewed records of 702 women with pri-
mary endometrial carcinoma that was diagnosed between
1979 and 1993. Preoperative CT findings were compared with
pathologic findings to assess nodal disease. The yield of
postoperative CT was reviewed in clinically suspicious and
routine settings.
Results: Among 492 women eligible for analysis, 178 (36%)
had a total 326 CT scans. Among 56 women who had
preoperative CT scans and lymph node samplings, positive
and negative predictive values for nodal involvement were
50% and 94%, respectively, and sensitivity and specificity
were 57% and 92%, respectively. Preoperative CT findings
altered treatment plans in only six patients (8%). Forty-five
asymptomatic women had 73 routine CT scans, and recur-
rence was diagnosed by CT in only two (4.4%). Thirty-seven
women had CT scans for suspicion of recurrence, which was
confirmed in 17 (46%). Kaplan-Meyer analysis showed no
survival advantage in women with subclinical recurrences
diagnosed by CT scan.
Conclusion: Routine preoperative CT scanning rarely al-
ters treatment and is a poor predictor of nodal disease.
Computed tomography in the postoperative period might be
helpful for detection and follow-up of recurrent disease, but
there was no difference in survival when subclinical recur-
rence was found by CT. Thus, CT scanning of any woman
with endometrial cancer should be discouraged unless it is
to evaluate symptoms. (Obstet Gynecol 2000;95:692– 6.
© 2000 by The American College of Obstetricians and
Gynecologists.)
Computed tomography (CT) has been proposed as a
useful imaging tool for pretreatment staging and post-
treatment evaluation of recurrence of gynecologic ma-
From the Department of Obstetrics and Gynecology, Division of
Gynecologic Oncology and the Department of Obstetrics and Gynecol-
ogy, University of Iowa Hospitals and Clinics, Iowa City, Iowa.
692 0029-7844/00/$20.00
PII S0029-7844(99)00626-2
lignancies.1– 8 Some studies concluded that CT might
accurately image large masses or extensive spread of
disease, but that information might not translate into
important alterations in treatment or improve surviv-
al.9,10 Endometrial cancer is the most common gyneco-
logic malignancy, with approximately 35,000 new cases
annually. Preoperative CT is often done to evaluate the
extent of myometrial invasion or to define extrauterine
spread of disease, including lymphatic metastasis. Most
women who present with stage I disease subsequently
have surgical staging and treatment, so the value of
preoperative radiographic staging seems limited. Post-
operative indications for CT imaging vary. It can be
used as a routine adjuvant to physical examination for
detecting recurrence of disease in asymptomatic
women or confirming clinical suspicions of recurrence
that are not clearly evident on examination.
We reviewed records of patients with endometrial
carcinoma during a 15-year period to determine
whether CT imaging provided useful information to aid
preoperative and postoperative treatment. We also
evaluated whether diagnosis of subclinical recurrence
of disease by CT compared with recurrences that were
diagnosed clinically influenced survival.
Materials and Methods
Seven hundred two consecutive women with primary
endometrial carcinoma diagnosed between 1979 and
1993 were identified by the cancer registry at The
University of Iowa Hospitals and Clinics. This time
frame was selected to encompass adequate follow-up
and because CT imaging became available routinely in
1979. We reviewed charts and abstracted age, gravidity
and parity, age at menopause, stage of disease, cell type,
histologic grade, depth of invasion, nodal involvement,
peritoneal cytology, current disease status, recurrence
location, number of CT scans done, and indications for
CT scanning. Patients were excluded from evaluation if
Obstetrics & Gynecology
their follow-up treatment was at another institution, if
they had diagnosis of second malignancy, or if primary
surgery was not planned initially. After exclusions, 492
women were available for analysis. Their disease was
staged, depending on date of diagnosis, either by the
1971 International Federation of Gynecology and Ob-
stetrics classification (FIGO) clinical staging criteria or
the 1988 FIGO surgical classification. Node sampling in
surgically staged cases was done according to the
surgical guidelines of the Gynecologic Oncology
Group’s surgical manual.
Computed tomographic scans were done with and
without contrast at 1- and 2-cm intervals through the
pelvis and abdomen. Pelvic and para-aortic lymph
nodes were considered enlarged if they exceeded
1.5 cm.11 Scans were coded as preoperative or postop-
erative. Indications for postoperative scans were classi-
fied as routine in asymptomatic patients, suspicious of
recurrence in patients with symptoms but no clinically
evident disease, or all other unrelated indications, in-
cluding follow-up of previously abnormal scans. Use of
CT imaging was based on treatment practices of indi-
vidual gynecologic oncologists and no prospective al-
gorithms were used to determine who was scanned
preoperatively or postoperatively. Interpretations of
preoperative CT were compared with histopathologic
findings at laparotomy to determine sensitivity, speci-
ficity, and predictive value of CT evaluation of lymph
node status. The utility of postoperative CT scans was
reported in clinically suspicious and routine settings.
Clinical and pathologic factors of women who were
examined by postoperative CT scans were compared
with those who did not have CT scans by Student t test
for continuous variables and ␹2 analysis for categorical
variables. Survival analysis estimates were calculated
by the Kaplan-Meyer method with the generalized
Wilcoxon test. All statistical analyses were done using
BMDP (BMDP, Los Angeles, CA) statistical software,
with significance at P ⬍ .05.12
Results
The study spanned FIGO’s change from clinical to
surgical staging for endometrial cancer. The distribu-
tion of study population across the staging systems is
depicted in Table 1. Surgical staging increased the
proportion of stage III disease from 5% to 16% primarily
because of node sampling.
One hundred seventy-eight women had a total of 326
CT scans. Fifty-nine women had preoperative scans
only, 103 had postoperative scans, and 16 had preoper-
ative and postoperative scans. The remaining 314
women did not have CT scans. Fifty-three (71%) of the
preoperative CT scans were done in the surgically
VOL. 95, NO. 5, PART 1, MAY 2000
Table 1. Distribution of Clinical and Surgical Staging
Clinical* Surgical† Total
Stage n (%) n (%) n (%)
I 198 (75) 152 (67) 350 (71)
II 46 (17) 27 (12) 73 (15)
III 12 (5) 37 (16) 49 (10)
IV 9 (3) 11 (5) 20 (4)
Total 265 (100) 227 (100) 492 (100)
* International Federation of Gynecology and Obstetrics classifica-
tion 1971.
†
International Federation of Gynecology and Obstetrics classifica-
tion 1988.
staged group and 22 (29%) in the clinically staged
group. That represented the differences in endometrial
cancer treatment strategies of gynecologic oncologists
involved during the different times. In the postopera-
tive setting, CT scans were divided equally between
surgically and clinically staged patients.
Women who had CT scans did not differ in age,
gravidity, parity, or age at menopause compared with
those not scanned (P ⬎ .05 for each factor). Postopera-
tive CT scans were done more often in women with
recurrent disease; 59 of 79 women (75%) with recur-
rence had CT scans whereas 20 (25%) were never
scanned.
Among 75 women who had preoperative CT scans,
treatment was modified based on CT findings in only
six (8%). Three morbidly obese patients (body mass
index 50, 59, and 61) had vaginal hysterectomies be-
cause no adenopathy or evidence of extrauterine dis-
ease was detected by CT. A fourth medically compro-
mised patient had a vaginal hysterectomy after a
negative preoperative scan. All those women were alive
and without evidence of disease at the time of the
analysis. Two women (3%) did not have surgery be-
cause CT detected advanced disease with large para-
aortic adenopathy that was not anticipated at diagnosis.
The first of those women was treated with primary
chemotherapy and died of progressive disease 4
months after diagnosis. The second was treated with
primary radiation therapy. She had residual disease in
the uterus at the time of brachytherapy and was treated
with progestin until she died of progressive disease 11
months later.
Fifty-six women had preoperative CT scans and
lymph node sampling at laparotomy. Eight scans
showed adenopathy suspicious of disease, and 48 scans
showed no suspicious adenopathy. Three of 48 women
with negative CT scans had lymph node metastases on
pathologic evaluation. Four patients with CT-reported
suspicious adenopathy had pathologically negative
lymph nodes. Table 2 shows the correlation of preop-
erative CT and pathologic evaluation of lymph nodes.
Connor et al Computed Tomography 693
Table 2. Correlation of Computed Tomography and
Pathology of Pelvic and Para-aortic Lymph Nodes*
Pathology CT positive CT negative
result adenopathy adenopathy
Positive 4 3
Negative 4 45
Total 8 48
CT ⫽ computed tomography.
* Sensitivity ⫽ 57%, specificity ⫽ 92%, positive predictive value ⫽
50%, negative predictive value ⫽ 94%.
Sensitivity and specificity of preoperative CT for detect-
ing nodal metastasis were 57% and 92%, respectively.
Positive predictive value for nodal disease was only
50%; negative predictive value was 94%.
After initial surgery, 45 asymptomatic women had a
total of 73 CT scans (one to six scans per patient) as
surveillance for recurrent disease. Forty-three women
had negative scans, and two (4.4%) had recurrence
diagnosed by CT. Both patients had advanced disease
at diagnosis and subsequently had CT-diagnosed para-
aortic recurrences. Both were dead of their disease 38
and 46 months after diagnosis of recurrence.
Thirty-seven women had a total of 41 scans to eval-
uate clinical symptoms of recurrence without clinically
evident disease. Gastrointestinal complaints such as
nausea, vomiting, or weight loss were the most com-
monly reported indications for scans in that group.
Other indications included abdominal pain, increase in
CA 125, unexplained jaundice, or lower-extremity deep
venous thrombosis. Computed tomography provided
evidence of recurrence in 17 patients (46%). All recur-
rences were confirmed by biopsy. Among 20 patients
with clinical findings suspicious for recurrence not
confirmed by CT, seven (35%) later had clinical diag-
noses of recurrent disease.
One hundred thirty-seven CT scans were done in 61
women for indications other than suspicion of recur-
rence or routine surveillance. Those indications are
listed in Table 3. Several patients were treated accord-
ing to Gynecologic Oncology Group protocols and were
required to have periodic CT evaluation by protocol.
One woman had 16 CT scans while on that protocol,
then had recurrence diagnosed by elevated CA 125
level between protocol-scheduled CT scans. Recurrence
was confirmed by CT scan and she was subsequently
observed using CA 125 levels. One woman (1.6%) in
that group had an unexpected aortic node recurrence
diagnosed by CT scan that was done as follow-up 6
months after a negative scan when she presented with a
lower extremity deep venous thrombosis (the first scan
was coded as not suspicious for recurrence).
Twenty women from the three groups had recur-
rences diagnosed by CT scan before they were clinically
694 Connor et al Computed Tomography
Table 3. Other Indications for Postoperative Computed
Tomography Scans
No. of No. of
Total Indication patients scans
7 Gynecologic Oncology Group protocol 4 37
49 Evaluation of abdomen or pelvis after 39 73
56 recurrence or follow-up recurrence
Immediate postoperative complication 6 6
Rule out or assess abscess 3 5
Follow-up of prior suspicious CT scan 3 8
Evaluate gallstones 1 2
Evaluate lymphocoele 2 2
Evaluate hip pain 1 1
Evaluate for pancreatitis 1 2
Follow-up after pentoxyphilline treatment 1 1
Total 61 137
CT ⫽ computed tomography.
evident. There were no significant differences in the
distribution of stages (P ⫽ .55) or grades (P ⫽ .08)
between the recurrences diagnosed by CT and those
diagnosed clinically. In both groups, tumors that re-
curred were 40 –50% stage III–IV and 70 – 80% grade 2
or 3 lesions. The median survival of the 20 women with
CT-diagnosed recurrence was 42 months; 16 (80%)
women were dead of disease at the time of analysis. The
remaining 59 women with recurrent disease were diag-
nosed clinically and had a median survival of 32
months. Seventy-five percent of those women have died
of their disease. Kaplan-Meyer survival analysis in
Figure 1 shows no survival advantage for the cases of
subclinical recurrent disease diagnosed by CT scan (P ⫽
.71).
Discussion
Many studies have addressed the effects of CT imaging
on treating gynecologic malignancies.1–10 However,
most of those studies were limited to small cohorts of
patients with endometrial cancer (range 4 – 65 subjects),
and few studies focused on CT imaging as it relates
solely to treating endometrial cancer.4,6 – 8 This study
comprised a large series of 178 patients with endome-
trial cancer who had CT scanning. Because of the
retrospective nature of the data, changes in the imaging
instrumentation and techniques and the medical staff
involved over 15 years brought additional confounders
and some selection bias to the analysis. However,
although the data represent a retrospective review of
medical records, the small number of cases in which CT
imaging changed treatment or outcome minimizes the
utility of this imaging method in treatment of endome-
trial cancer.
Previous reviews suggested that the usefulness of CT
scanning is uncertain in preoperative treatment of en-
Obstetrics & Gynecology
Figure 1. Kaplan-Meyer survival curve for
women with subclinical recurrent disease diag-
nosed by computed tomography scan (diamonds
indicate median survival 42 months) and those
recurrences clinically diagnosed (circles indicate
median survival 32 months). Generalized Wil-
coxon test shows no difference in survival P ⫽
.71.
dometrial cancer. Walsh and Goplerud7 concluded that
there was “confirmed value” in preoperative CT scan-
ning to evaluate women with advanced disease. They
evaluated only 19 women and found that CT findings
agreed with surgical staging in 16 of them, but two
women had microscopic lymph node metastases that
were not detected by CT scans. The sensitivity of 57%
and positive predictive value of 50% for detecting nodal
metastases in the present study suggest that preopera-
tive CT is a poor predictor of nodal disease. The
calculated sensitivity and positive predictive value are
based on only seven cases with positive nodes (12.5%),
so the values could be overestimates or underestimates.
Although a specificity of 92% and negative predictive
value of 94% (based on 49 cases with negative nodes)
are reassuring for noninvolved nodes, CT scans should
not be used in place of surgical sampling except in cases
in which patients are believed to be at excessive risk for
operative morbidity or mortality. Based on our results,
we concluded that routine preoperative scanning in
women who are to have surgical staging rarely alters
the accepted treatment plan, is unreliable, and with rare
exceptions (such as for medically compromised patients
or those with marked obesity) should be discouraged.
A second issue is the utility of CT for detecting and
helping treatment of recurrent disease. For CT scanning
to be beneficial, it should be considered an accurate
method of detecting tumor spread. However, as Franchi
et al8 show, the major disadvantage of CT is the
difficulty distinguishing inflammation and postsurgical
VOL. 95, NO. 5, PART 1, MAY 2000
or radiation therapy fibrosis from recurrent disease in
the pelvis. Conversely, CT might provide supplemen-
tary data for deciding treatment options (surgery, che-
motherapy, or radiation) for patients with known recur-
rent disease.
Our data indicated that routine use of CT scanning of
asymptomatic women in the postoperative period to
detect subclinical recurrence of disease was not effec-
tive. Only two of 45 asymptomatic women had recur-
rent disease diagnosed by CT, and both were expected
to develop recurrent disease based on advanced stage at
diagnosis. Both patients ultimately died of their disease,
and their survival times were within the range of those
for clinically diagnosed recurrences. The majority of
CT-diagnosed recurrences were in women with clinical
symptoms suggesting recurrence before scanning. In
that setting, CT scanning detected subclinical recur-
rence in 46% of cases. When grouped together, the
patients with CT-diagnosed recurrences lived no longer
than those diagnosed clinically. As in Figure 1, the
percentage of women alive at 5 years differs by less than
10% between the two groups. With only 20 cases of
CT-diagnosed recurrences, the power of the Kaplan-
Meyer analysis is limited. The data, as presented, can
detect a 40 – 45% difference with 20% power; however,
to detect the 10% difference seen in Figure 1 with the
same 20% power would require more than 300 patients
per group. Those limits notwithstanding, it appears that
postoperative CT does not benefit patients in terms of
survival.12–14
Connor et al Computed Tomography 695
References
1. Photopulos GJ, McCartney WH, Walton LA, Staab EV. Computer-
ized tomography applied to gynecologic oncology. Am J Obstet
Gynecol 1979;135:381–3.
2. King LA, Talledo OE, Gallup DG, el Gammal TAM. Computed
tomography in evaluation of gynecologic malignancies: A retro-
spective analysis. Am J Obstet Gynecol 1986;155:960 – 4.
3. Chen SS, Kumari S, Lee L. Contribution of abdominal computer
tomography (CT) in the management of gynecologic cancer: Cor-
related study of CT image and gross surgical pathology. Gynecol
Oncol 1980;10:162–72.
4. Hasumi K, Matsuzawa M, Chen HF, Takahashi M, Sakura M.
Computed tomography in the evaluation and treatment of endo-
metrial carcinoma. Cancer 1982;50:904 – 8.
5. Bandy LC, Clarke-Pearson DL, Silverman PM, Creasman WT.
Computed tomography in evaluation of extrapelvic lymphadenop-
athy in carcinoma of the cervix. Obstet Gynecol 1985;65:73– 6.
6. Balfe DM, Van Dyke J, Lee JKT, Weyman PJ, McClennan BL.
Computed tomography in malignant endometrial neoplasms.
J Comput Assisted Tomogr 1983;7:677– 81.
7. Walsh JW, Goplerud DR. Computed tomography of primary,
persistent, and recurrent endometrial malignancy. Am J Roentge-
nol 1982;139:1149 –54.
8. Franchi M, La Fianza A, Babilonti L, Bolis PF, Alerci M, Di Giulio
G, et al. Clinical value of computerized tomography (CT) in
assessment of recurrent uterine cancers. Gynecol Oncol 1989;35:
31–7.
9. Kerr-Wilson RHJ, Shingleton HM, Orr JW Jr, Hatch KD. The use of
ultrasound and computed tomography scanning in the manage-
ment of gynecologic cancer patients. Gynecol Oncol 1984;18:54 – 61.
696 Connor et al Computed Tomography
10. Mann WJ, Baim R, Patsner B, Chalas E, Taylor A, Westermann C,
et al. The value of CT scanning in the management of patients with
gynecologic malignancies. Arch Gynecol Obstet 1989;246:15–25.
11. Friedman AC, Radecki PD, Lev-Toaff AS, Hilport PL, eds. Clinical
pelvic imaging. St Louis: The CV Mosby Company, 1990.
12. Kleinbaum DG, Kupper LL, Muller KE, eds. Applied regression
analysis and other multivariable methods. 2nd ed. Boston: PWS-
KENT Publishing Company, 1988.
13. Hulley SB, Cummings SR, eds. Designing clinical research: An
epidemiologic approach. Baltimore: Williams & Wilkins, 1988.
14. Schlesselman JJ. Case-control studies. New York: Oxford Univer-
sity Press, 1982.
Address reprint requests to:
Joseph P. Connor, MD
Department of Obstetrics and Gynecology
University of Illinois at Chicago
820 South Wood Street, MC 808
Chicago, IL 60612-7313
E-mail: jconnor@uic.edu
Received July 22, 1999.
Received in revised form October 21, 1999.
Accepted November 15, 1999.
Copyright © 2000 by The American College of Obstetricians and
Gynecologists. Published by Elsevier Science Inc.
Obstetrics & Gynecology