GYNECOLOGIC
ONCOLOGY
4,
253-259 (1991)
Detection of Endometrial Cancer by Transvaginal Ultrasonography with Color Flow Imaging and Blood Flow Analysis: A Preliminary Report THOMAS H. BOURNE, MRCOG, STUART CAMPBELL, FRCOG, CHRISTOPHER V. STEER, MRCOG, PATRICK ROYSTON, M.Sc.,* MALCOLM I. WHITEHEAD, FRCOG, AND WILLIAM P. COLLINS, D.Sc. Academic
Department *Department
of Obstetrics of Medical
and Gynaecology, King’s College School of Medicine Physics, Royal Postgraduate Medical School, Ducane
and Dentistry, Road, London
Denmark Hill, London SE5 8RX, WI2 ONN, United Kingdom
and
Received June 26, 1990
tage). However,
less than 10% of women with post-
A prospectivestudy wasundertakento assess whether changes menopausal bleeding have endometrial cancer. A less inin uterine blood flow could be usedto detect endometrialcancer vasive technique with a high detection rate and a low in 138selectedpostmenopausal women(34 had uterine bleeding, false-positive rate for this disease would be of value for 17 with endometrial cancer; 104 did not have uterine bleeding; 1 had endometrial cancer). Thirty-five of the asymptomatic selecting those women who require diagnostic surgery. if the technique could detect endometrial womenwere receiving estrogenreplacementtherapy (ERT). The Furthermore, endpointswere endometrial(including tumoral) thicknessand a cancer at an early stage in asymptomatic women (i.e., pulsatility index (PI) derived from flow velocity waveforms re- those who do not have postmenopausal bleeding), then corded from both uterine arteriesand from within a tumor. We surgery might be curative and the mortality rate (about found an overlap in endometrialthicknessbetweenthosewomen 4000 per annum) might be reduced by the implementation with endometrial cancer and those without. The mean arterial of an appropriate screening program. PI value was invariably lower in women with postmenopausal Transabdominal ultrasonography can be used to detect bleedingand endometrial cancer (mean 0.91, range 0.31-1.49) many forms of endometrial pathology including cancer than in thosewith other reasonsfor the blood loss(mean 3.83, range 1.95640). The index was 1.10 in the woman with en- [3,4]. Moreover, the advent of transvaginal ultrasonogdometrial cancerbut no signof postmenopausal bleeding.Blood raphy has enabled the production of more detailed images flow impedancewasinverselyrelated to stageof cancer.PI values of the uterus [5-71 and facilitated the measurement of (or tumoral) thickness. To date, however, in healthy women tended to increaseslightly with age, but de- endometrial features that are unique to malignant creaseduring ERT. The detection rate was 100% within the no morphological limitations of the study design, and the false-positiverate was masses have been identified. Recently the use of trans1% for all women not receiving ERT and 11% for patients re- abdominal or transvaginal pulsed Doppler probes, with ceiving ERT. Malignant tumors showsignsof altered vascular- and without color flow imaging, has enabled an accurate ization and a low PI (mean0.49, range0.29-0.92). We conclude assessment of blood flow impedance in uterine arteries that transvaginal ultrasonography, with or without color flow during the menstrual cycle [g-11]. We now report the imaging,and bloodflow analysiscanbeusedto detectendometrial use of transvaginal ultrasonography and color flow imcancerin womenwith postmenopausal bleeding.A screeningproaging to detect endometrial cancer in women with postcedurefor asymptomaticwomenmustallow for changesin uterine menopausal bleeding. At the same time we have underblood flow during ERT. o WI Academic press, IIIC.
INTRODUCTION About 33,000 new cases of uterine cancer are reported in the United States each year [l]. The incidence of the disease increases markedly about the time of the menopause and reaches a peak between the ages of 55 and 65 years [2]. Uterine bleeding is the most frequent initial sign of the disease in postmenopausal women and demands invasive investigation (e.g., dilation and curet-
taken preliminary studies to identify endpoints and obtain reference values that might be used to detect endometrial cancer in asymptomatic women who are being screened for ovarian cancer [12,13] or undergoing estrogen replacement therapy (ERT). SUBJECTS
AND METHODS
We studied postmenopausal women (i.e., women who had not menstruated spontaneously for at least 1 year) 253 MN-8258/Yl $1.50 Copyright 0 1991 by Academic Press, inc. All rights of reproduction in any form reserved.
254
BOURNE
who were eventually classified into one of five groups. Group 1 consisted of 17 patients who presented with uterine bleeding in whom a histological diagnosis of endometrial cancer was made. Group 2 contained 1 patient who did not have postmenopausal bleeding, but presented with abdominal discomfort. Transvaginal ultrasonography revealed bilateral ovarian masses, and the woman was referred for laparotomy. The surgical and histological diagnosis was primary ovarian cancer (bilateral endometrioid adenocarcinoma, FIG0 stage IC) and primary endometrial adenocarcinoma (FIG0 stage IA). The third group contained 17 patients who presented with uterine bleeding, but dilation and curettage was unremarkable and there was no histological evidence of endometrial cancer. The fourth group consisted of 68 healthy women who had no evidence of postmenopausal bleeding or endometrial cancer and were not receiving ERT. Twenty of these women subsequently underwent dilation and curettage before starting ERT. The fifth group contained 35 women who were already receiving ERT (Estraderm 50 pg estradiol TTS, Ciba-Geigy, Horsham, Sussex, UK; and sequential Micronor, norethisterone acetate 0.70 to 1.05 mg for 12 days each calendar month, Ortho Cilag, High Wycombe, UK), and were thus undergoing regular episodes of induced uterine bleeding. Women in groups 1, 2, and 3 were referred or selected from the Gynecological Outpatients clinic at King’s College Hospital. Twenty-five women had already undergone dilation and curettage 7 to 10 days before being scanned because they were suspected of having endometrial cancer, but the ultrasonographer was unaware of the diagnosis. The remaining women (10) had a biopsy taken after ultrasonography. The women in groups 4 and 5 were recruited from the Ovarian Cancer Screening Clinic or Menopause Clinic.
Scanning Procedure All women were examined at the time of referral or selection without reference to current or planned treatment. They had an empty bladder and were scanned in the lithotomy position with a slight reverse Trendelenberg tilt to localize free fluid in the pouch of Douglas. A Diasonics SPA 1000 Scanner (Sonotron, Bedford, UK) with an endovaginal probe was used initially. The 7.5-MHz transducer produced a 110” sector angle that was offset 15” to allow a good view of the pelvic organs. The probe had a maximum diameter of 1.5 cm. Prior to use the end was covered with a coupling gel, then inserted into a condom, and finally recoated with gel before insertion into the vagina. The uterus and vasculature were assessed systematically [14]. The probe was tilted gently, pushed, pulled, and rotated to change the scanning plane. Both uterine arteries could be detected in the transverse or
ET AL.
longitudinal plane above the supravaginal position of the cervix. Occasionally one or both arteries could be seen as pulsatile echolucent areas. The high-pass filter was set at 100 MHz. Doppler signals in the form of flow velocity waveforms (FVWs) were displayed on line. The thickness of the endometrium was measured, and the size, shape, and echogenicity of abnormal structures were recorded. Signals were taken from both uterine arteries in the transverse plane and from any endometrial abnormalities. Ten women (five with endometrial cancer) were also examined with an Aloka SSD 680 scanner (Aloka Co. Ltd., Tokyo, Japan) with an endovaginal probe producing a 5.0-MHz beam for imaging and a 6.0-MHz system for pulsed and color Doppler blood flow analysis. The shade of color produced was proportional to the Doppler frequency shift. Red and blue indicated blood flow toward or away from the probe, respectively. Color flow images of the uterine arteries were sampled lateral to the cervix in a longitudinal plane to accurately identify a consistent anatomical position. The probe was first directed into the right vaginal fornix and the ascending branch of the right uterine artery was located as a blue or red pulsating area along the lateral border of the uterus. The angle of insonation was changed to obtain maximum color intensity. A range gate was then placed across the vessel. The filter level was set at 100 MHz and the FVWs were displayed on line. The angle of the probe was moved to obtain the maximum waveform amplitude and clarity. The left uterine artery was located and studied in an identical manner. Areas of neovascularization within the endometrium were identified. Details of all scans were recorded on Polaroid film and video tape. Quantitation The maximum endometrial thickness (ET) including any abnormal tissue (i.e., the longest distance between each myometrial/endometrial interface across the lumen excluding free fluid) was measured in the longitudinal plane. The pulsatility index (PI) was found to be a useful way of expressing blood flow impedance distal to the point of sampling, particularly in circumstances when the diastolic shift was below the filter level [15]. The value for each artery, or intratumoral blood vessel, was calculated electronically from a smooth curve fitted to the average waveform over three cardiac cycles according to the formula PI = (A - @/mean where A is the peak Doppler frequency shift, B is the minimum Doppler frequency shift, and mean is the mean maximum Doppler shift frequency over the cardiac cycle. The index will increase if the proximal conditions remain constant and the distal vascular bed constricts. Conversely, a low value for the PI indicates decreased imped-
ENDOMETRIAL
CANCER
TABLE 1 Age and Years Since Menopauseof Women with and without Endometrial Cancer Years since menopause
Age
255
DETECTION
TABLE 3 Mean Impedanceto Uterine Arterial Blood Flow (as Reflected by the Pulsatiliy Index) in Womenwith and without Endometrial Cancer. Pulsatility index ____SD Minimum Maximum
Number of Women
Mean
Range
Mean
Range
Cancer With pmb Without pmb
17 1
68 53
50-83 -
17 4
l-3.5 -
Cancer With pmb Without pmb
17 1
0.91 1.10
0.36 -
0.31 -
1.49
Noncancer With pmb Without pmb With pmb (ERT)
17 68 35
58 56 54
49-71 43-67 43-66
7 6 5
l-31 1-19 1-15
Noncancer With pmb Without pmb With pmb (ERT)
17 68 3s
3.83 4.15 2.53
1.30 1.39 0.87
1.Y5 1.50 1.27
6.40 8.00 4.67
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT.
estrogen
replacement
ante to blood flow in the distal vasculature. A one-way analysis of variance of replicate data (at least triplicate) from 20 women gave a coefficient of variation of 10.3% for the left uterine artery and 9.7% for the right. The corresponding value for intraovarian blood flow for 13 women was 7.6%. Histopathology All 34 women with postmenopausal bleeding underwent dilation and curettage (either 7 to 10 days before or 1 to 5 days after the ultrasound scan), and the tissues were sent for routine histological examination. The diagnosis was based on the report from the histologist. Subsequently the estimated stage of any cancer was reassessed by the surgeon at the time of laparotomy and by the histological state of the excised tissues [16].
Number of women Mean
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT,
estrogen
replacement
women had uterine bleeding, 1 did not). There was no evidence of endometrial cancer in the tissue obtained from the other 17 patients with uterine bleeding. About equal numbers of patients with and without cancer were scanned after curettage. The age and years since menopause of the women with and without endometrial cancer are shown in Table 1. Cancer tends to be associated with increased age and years since menopause. Endometrial Thickness
The thickness of the endometrium and the presence of any abnormalities as reflected by the ET in women in every group are shown in Table 2. There is overlap in the values observed for women with and without cancer.
RESULTS Uterine Arterial Blood Flow
There were 7 cases of stage IA endometrial cancer, 3 stage IB, 4 stage IC, 1 stage IIA, and 3 stage IIIA (17 TABLE 2 Maximum Endometrial (or Tumorai) Thicknessin Women with and without Endometrial Cancer. Endometrial
thickness (cm)
Number of women
Mean
SD
Minimum
Maximum
Cancer With pmb Without pmb
17 1
2.0 0.7
0.9 -
0.6 -
4.1
Noncancer With pmb Without pmb With pmb (ERT)
17 68 35
0.8 0.4 0.7
0.9 0.2 0.5
0.2 0.2 0.3
1.5 1.6 2.1
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT,
estrogen
replacement
The mean impedance to uterine arterial blood flow as reflected by the PI is shown in Table 3. The maximum value (1.49) in patients with endometrial cancer and postmenopausal bleeding is below the minimum value (1.95) in patients with uterine blood loss due to other factors. The patient with endometrial cancer (FIG0 stage IA) but no sign of uterine blood loss had a mean uterine arterial PI of 1.10, which was below the minimum value (1.50) of the reference groups of women not receiving ERT. One-way analysis of variance showed a statistically significant difference in the PI between all five groups (P < .OOOl), but not between the 17 women with uterine bleeding but no evidence of cancer and the 68 who did not have postmenopausal bleeding and were not receiving ERT (P = 0.09). The values for those two groups were combined for further analysis. Similarly, the values for all women with cancer (16 with uterine bleeding, 1 without) were reconsidered as one group.
256
BOURNE n = 18 0 0 0
n = 84
ET AL.
n = 35
0 00 0
0
0
o.oJ Cancer
Non-cancer (no ERT)
Non-cancer (with
FIG. 1. Individual values for endometrial thickness in patients with and without endometrial cancer (note log scale on y axis).
Overall Predictive Value of ET and PI
The individual values for endometrial thickness in women from the three groups (cancer with and without postmenopausal bleeding, noncancer with and without postmenopausal bleeding, and noncancer without irregular uterine blood loss and receiving ERT) are shown in Fig. 1. An arbitrary value of 1.0 cm was selected to discriminate between the groups. The rate of detection [and 95% confidence interval (CI)] of endometrial cancer was 14/17 (82%, CI 54-96%), while the false-positive rate (FPR) was 4/84 (5%, CI l-12%) for asymptomatic
I
n = 18
n = 85 0
IA
EQT)
n = 35
16
IC IIA Stage of cancer
0
116
IIIA
FIG. 3. Relationship between pulsatility index and stage of endometrial cancer.
women not receiving ERT and 6/35 (17%, CI 7-34%) for asymptomatic women receiving ERT. The corresponding values for the PI are shown in Fig. 2. An arbitrary value of 1.50 was selected to distinguish between the groups. The observed detection rate for endometrial cancer was 17/17 (lOO%, 95% CI 81-lOO%), while the FPR was l/85 (1%) for women with and without postmenopausal bleeding but not receiving ERT and 4/35 (11%) for asymptomatic women receiving ERT. The data in the cancer and the noncancer/no-ERT groups were logarithmically transformed to approximate normal distributions and the overlap was estimated using the means and standard deviations. We calculated that a cutoff value of 2.00 would give a detection rate of 99.0% with a falsepositive rate of 2.6%. Consequently, the calculated predictive value of a positive screening result for women with postmenopausal bleeding (assuming a 10% prevalence of endometrial cancer) would be 80.9% and the predictive value of a negative screening result would be 99.9%. The positive and negative predictive values of screening results for postmenopausal women without postmenopausal bleeding attending a screening clinic (assuming a 0.00071% prevalence of endometrial cancer) would be 9.1 and 99.9%, respectively. Factors Affecting PI
Cancer
Non-cancer ho EAT)
Non-cancer (With
EAT)
FIG. 2. Individual values for the mean pulsatile index derived from flow velocity waveforms from the right and left uterine arteries in patients with and without uterine cancer (note log scale on y axis).
The relationship between the PI and the stage of endometrial cancer is shown in Fig. 3. There is a trend toward a reduction in the impedance to arterial blood flow as the disease progresses. The two patients with a low PI at stages IA and IB were aged 83 and 75 years, respectively. The effect on the PI value of the number of years since
ENDOMETRIAL
Non-cancer (no ERT)
Cancer
CANCER
257
DETECTION
the PI for patients with cancer and a slight upward trend in the PI for women with postmenopausal bleeding but no cancer.
0
Intratumoral Blood FLOW
5.0-
.
a a 0
2 -cl $4
0 2.0.
,” 2 .r( +r (0 rn l-l 2
Oil 0 8 II 0 0 0 :
0 0
0
a
0
0
1.0.
0
00 0
0 0
0
OO
The PI was determined within the endometrial cancer in 10 of 18 subjects (5/5 with color flow imaging). The value was similar to the mean uterine arterial PI in 3 cases and lower in 7 subjects (mean 0.49, range 0.29-0.92). It was not possible to accurately identify different FVWs in the other 7 cases using pulsed Doppler without color flow imaging. This technique was available to identify intratumoral areas of neovascularization in five cases and an example is shown in Fig. 5. DISCUSSION
Transvaginal ultrasonography is a relatively new technique that has some important advantages over the use of transabdominal transducers. In particular there is no need for the women being scanned to have a full bladder, Iears2 sin2 mirYopa2u0sd"{ear: sinc5eme'n0opa2u0sF and the use of higher-frequency ultrasound enables the FIG. 4. Relationship between pulsatility index and years since menmorphology of the reproductive tract to be assessed in opause in women with and without endometrial cancer, not receiving greater detail. Descriptions of the uterus by this approach hormone replacement therapy (note log scale on x and y axes). have been shown to correlate well with those observed at surgery [7]. C onsequently transvaginal ultrasonography menopause in patients with cancer and in women without can be used to assess the extent to which cancer of the endometrium has penetrated the myometrium or cervix, postmenopausal bleeding or cancer and not receiving ERT is shown in Fig. 4. There is a downward trend in and hence can be of value preoperatively for staging con-
FIG. 5. Sonogram showing intratumoral
vascularization in a patient with endometrial cancer (FIG0 stage IB).
258
BOURNE
firmed disease [17-201. Furthermore, the results of our study of ET in postmenopausal women show that an ET >l.O cm has a high predictive value for the presence of endometrial cancer. The detection rate for this disease of 82% with a FPR of 5% for postmenopausal women not receiving ERT shows that the technique is clinically useful in the absence of facilities to measure uterine blood flow. The use of transvaginal pulsed Doppler ultrasonography in measuring blood flow impedance in the uterine arteries of nonpregnant women has been described previously [lo]. Moreover, we have reported how transvaginal color flow imaging may be used to identify the uterine arteries and obtain optimum FVWs for the analysis of blood flow during the ovarian and menstrual cycles [ll]. The further application of these techniques shows that there is an apparent distinction in average PI for the two uterine arteries between those postmenopausal patients presenting with uterine bleeding in the presence of endometrial cancer and other women with the same symptom but with no apparent cause [21]. Infection, however, may also affect uterine blood flow (unpublished observations) and the presence of pathogenic bacteria should be considered before a diagnosis is made. A cutoff value of 1.50 was arbitrarily selected for the PI to obtain a high detection rate and a low false-positive rate for endometrial cancer. One case of the disease was detected in a woman without postmenopausal bleeding, and the FPR for all women not receiving ERT was about 1% (l/85). It is of interest that the woman with a false-positive result from the measurement of PI had an ET
ONCOLOGY
4,
253-259 (1991)
Detection of Endometrial Cancer by Transvaginal Ultrasonography with Color Flow Imaging and Blood Flow Analysis: A Preliminary Report THOMAS H. BOURNE, MRCOG, STUART CAMPBELL, FRCOG, CHRISTOPHER V. STEER, MRCOG, PATRICK ROYSTON, M.Sc.,* MALCOLM I. WHITEHEAD, FRCOG, AND WILLIAM P. COLLINS, D.Sc. Academic
Department *Department
of Obstetrics of Medical
and Gynaecology, King’s College School of Medicine Physics, Royal Postgraduate Medical School, Ducane
and Dentistry, Road, London
Denmark Hill, London SE5 8RX, WI2 ONN, United Kingdom
and
Received June 26, 1990
tage). However,
less than 10% of women with post-
A prospectivestudy wasundertakento assess whether changes menopausal bleeding have endometrial cancer. A less inin uterine blood flow could be usedto detect endometrialcancer vasive technique with a high detection rate and a low in 138selectedpostmenopausal women(34 had uterine bleeding, false-positive rate for this disease would be of value for 17 with endometrial cancer; 104 did not have uterine bleeding; 1 had endometrial cancer). Thirty-five of the asymptomatic selecting those women who require diagnostic surgery. if the technique could detect endometrial womenwere receiving estrogenreplacementtherapy (ERT). The Furthermore, endpointswere endometrial(including tumoral) thicknessand a cancer at an early stage in asymptomatic women (i.e., pulsatility index (PI) derived from flow velocity waveforms re- those who do not have postmenopausal bleeding), then corded from both uterine arteriesand from within a tumor. We surgery might be curative and the mortality rate (about found an overlap in endometrialthicknessbetweenthosewomen 4000 per annum) might be reduced by the implementation with endometrial cancer and those without. The mean arterial of an appropriate screening program. PI value was invariably lower in women with postmenopausal Transabdominal ultrasonography can be used to detect bleedingand endometrial cancer (mean 0.91, range 0.31-1.49) many forms of endometrial pathology including cancer than in thosewith other reasonsfor the blood loss(mean 3.83, range 1.95640). The index was 1.10 in the woman with en- [3,4]. Moreover, the advent of transvaginal ultrasonogdometrial cancerbut no signof postmenopausal bleeding.Blood raphy has enabled the production of more detailed images flow impedancewasinverselyrelated to stageof cancer.PI values of the uterus [5-71 and facilitated the measurement of (or tumoral) thickness. To date, however, in healthy women tended to increaseslightly with age, but de- endometrial features that are unique to malignant creaseduring ERT. The detection rate was 100% within the no morphological limitations of the study design, and the false-positiverate was masses have been identified. Recently the use of trans1% for all women not receiving ERT and 11% for patients re- abdominal or transvaginal pulsed Doppler probes, with ceiving ERT. Malignant tumors showsignsof altered vascular- and without color flow imaging, has enabled an accurate ization and a low PI (mean0.49, range0.29-0.92). We conclude assessment of blood flow impedance in uterine arteries that transvaginal ultrasonography, with or without color flow during the menstrual cycle [g-11]. We now report the imaging,and bloodflow analysiscanbeusedto detectendometrial use of transvaginal ultrasonography and color flow imcancerin womenwith postmenopausal bleeding.A screeningproaging to detect endometrial cancer in women with postcedurefor asymptomaticwomenmustallow for changesin uterine menopausal bleeding. At the same time we have underblood flow during ERT. o WI Academic press, IIIC.
INTRODUCTION About 33,000 new cases of uterine cancer are reported in the United States each year [l]. The incidence of the disease increases markedly about the time of the menopause and reaches a peak between the ages of 55 and 65 years [2]. Uterine bleeding is the most frequent initial sign of the disease in postmenopausal women and demands invasive investigation (e.g., dilation and curet-
taken preliminary studies to identify endpoints and obtain reference values that might be used to detect endometrial cancer in asymptomatic women who are being screened for ovarian cancer [12,13] or undergoing estrogen replacement therapy (ERT). SUBJECTS
AND METHODS
We studied postmenopausal women (i.e., women who had not menstruated spontaneously for at least 1 year) 253 MN-8258/Yl $1.50 Copyright 0 1991 by Academic Press, inc. All rights of reproduction in any form reserved.
254
BOURNE
who were eventually classified into one of five groups. Group 1 consisted of 17 patients who presented with uterine bleeding in whom a histological diagnosis of endometrial cancer was made. Group 2 contained 1 patient who did not have postmenopausal bleeding, but presented with abdominal discomfort. Transvaginal ultrasonography revealed bilateral ovarian masses, and the woman was referred for laparotomy. The surgical and histological diagnosis was primary ovarian cancer (bilateral endometrioid adenocarcinoma, FIG0 stage IC) and primary endometrial adenocarcinoma (FIG0 stage IA). The third group contained 17 patients who presented with uterine bleeding, but dilation and curettage was unremarkable and there was no histological evidence of endometrial cancer. The fourth group consisted of 68 healthy women who had no evidence of postmenopausal bleeding or endometrial cancer and were not receiving ERT. Twenty of these women subsequently underwent dilation and curettage before starting ERT. The fifth group contained 35 women who were already receiving ERT (Estraderm 50 pg estradiol TTS, Ciba-Geigy, Horsham, Sussex, UK; and sequential Micronor, norethisterone acetate 0.70 to 1.05 mg for 12 days each calendar month, Ortho Cilag, High Wycombe, UK), and were thus undergoing regular episodes of induced uterine bleeding. Women in groups 1, 2, and 3 were referred or selected from the Gynecological Outpatients clinic at King’s College Hospital. Twenty-five women had already undergone dilation and curettage 7 to 10 days before being scanned because they were suspected of having endometrial cancer, but the ultrasonographer was unaware of the diagnosis. The remaining women (10) had a biopsy taken after ultrasonography. The women in groups 4 and 5 were recruited from the Ovarian Cancer Screening Clinic or Menopause Clinic.
Scanning Procedure All women were examined at the time of referral or selection without reference to current or planned treatment. They had an empty bladder and were scanned in the lithotomy position with a slight reverse Trendelenberg tilt to localize free fluid in the pouch of Douglas. A Diasonics SPA 1000 Scanner (Sonotron, Bedford, UK) with an endovaginal probe was used initially. The 7.5-MHz transducer produced a 110” sector angle that was offset 15” to allow a good view of the pelvic organs. The probe had a maximum diameter of 1.5 cm. Prior to use the end was covered with a coupling gel, then inserted into a condom, and finally recoated with gel before insertion into the vagina. The uterus and vasculature were assessed systematically [14]. The probe was tilted gently, pushed, pulled, and rotated to change the scanning plane. Both uterine arteries could be detected in the transverse or
ET AL.
longitudinal plane above the supravaginal position of the cervix. Occasionally one or both arteries could be seen as pulsatile echolucent areas. The high-pass filter was set at 100 MHz. Doppler signals in the form of flow velocity waveforms (FVWs) were displayed on line. The thickness of the endometrium was measured, and the size, shape, and echogenicity of abnormal structures were recorded. Signals were taken from both uterine arteries in the transverse plane and from any endometrial abnormalities. Ten women (five with endometrial cancer) were also examined with an Aloka SSD 680 scanner (Aloka Co. Ltd., Tokyo, Japan) with an endovaginal probe producing a 5.0-MHz beam for imaging and a 6.0-MHz system for pulsed and color Doppler blood flow analysis. The shade of color produced was proportional to the Doppler frequency shift. Red and blue indicated blood flow toward or away from the probe, respectively. Color flow images of the uterine arteries were sampled lateral to the cervix in a longitudinal plane to accurately identify a consistent anatomical position. The probe was first directed into the right vaginal fornix and the ascending branch of the right uterine artery was located as a blue or red pulsating area along the lateral border of the uterus. The angle of insonation was changed to obtain maximum color intensity. A range gate was then placed across the vessel. The filter level was set at 100 MHz and the FVWs were displayed on line. The angle of the probe was moved to obtain the maximum waveform amplitude and clarity. The left uterine artery was located and studied in an identical manner. Areas of neovascularization within the endometrium were identified. Details of all scans were recorded on Polaroid film and video tape. Quantitation The maximum endometrial thickness (ET) including any abnormal tissue (i.e., the longest distance between each myometrial/endometrial interface across the lumen excluding free fluid) was measured in the longitudinal plane. The pulsatility index (PI) was found to be a useful way of expressing blood flow impedance distal to the point of sampling, particularly in circumstances when the diastolic shift was below the filter level [15]. The value for each artery, or intratumoral blood vessel, was calculated electronically from a smooth curve fitted to the average waveform over three cardiac cycles according to the formula PI = (A - @/mean where A is the peak Doppler frequency shift, B is the minimum Doppler frequency shift, and mean is the mean maximum Doppler shift frequency over the cardiac cycle. The index will increase if the proximal conditions remain constant and the distal vascular bed constricts. Conversely, a low value for the PI indicates decreased imped-
ENDOMETRIAL
CANCER
TABLE 1 Age and Years Since Menopauseof Women with and without Endometrial Cancer Years since menopause
Age
255
DETECTION
TABLE 3 Mean Impedanceto Uterine Arterial Blood Flow (as Reflected by the Pulsatiliy Index) in Womenwith and without Endometrial Cancer. Pulsatility index ____SD Minimum Maximum
Number of Women
Mean
Range
Mean
Range
Cancer With pmb Without pmb
17 1
68 53
50-83 -
17 4
l-3.5 -
Cancer With pmb Without pmb
17 1
0.91 1.10
0.36 -
0.31 -
1.49
Noncancer With pmb Without pmb With pmb (ERT)
17 68 35
58 56 54
49-71 43-67 43-66
7 6 5
l-31 1-19 1-15
Noncancer With pmb Without pmb With pmb (ERT)
17 68 3s
3.83 4.15 2.53
1.30 1.39 0.87
1.Y5 1.50 1.27
6.40 8.00 4.67
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT.
estrogen
replacement
ante to blood flow in the distal vasculature. A one-way analysis of variance of replicate data (at least triplicate) from 20 women gave a coefficient of variation of 10.3% for the left uterine artery and 9.7% for the right. The corresponding value for intraovarian blood flow for 13 women was 7.6%. Histopathology All 34 women with postmenopausal bleeding underwent dilation and curettage (either 7 to 10 days before or 1 to 5 days after the ultrasound scan), and the tissues were sent for routine histological examination. The diagnosis was based on the report from the histologist. Subsequently the estimated stage of any cancer was reassessed by the surgeon at the time of laparotomy and by the histological state of the excised tissues [16].
Number of women Mean
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT,
estrogen
replacement
women had uterine bleeding, 1 did not). There was no evidence of endometrial cancer in the tissue obtained from the other 17 patients with uterine bleeding. About equal numbers of patients with and without cancer were scanned after curettage. The age and years since menopause of the women with and without endometrial cancer are shown in Table 1. Cancer tends to be associated with increased age and years since menopause. Endometrial Thickness
The thickness of the endometrium and the presence of any abnormalities as reflected by the ET in women in every group are shown in Table 2. There is overlap in the values observed for women with and without cancer.
RESULTS Uterine Arterial Blood Flow
There were 7 cases of stage IA endometrial cancer, 3 stage IB, 4 stage IC, 1 stage IIA, and 3 stage IIIA (17 TABLE 2 Maximum Endometrial (or Tumorai) Thicknessin Women with and without Endometrial Cancer. Endometrial
thickness (cm)
Number of women
Mean
SD
Minimum
Maximum
Cancer With pmb Without pmb
17 1
2.0 0.7
0.9 -
0.6 -
4.1
Noncancer With pmb Without pmb With pmb (ERT)
17 68 35
0.8 0.4 0.7
0.9 0.2 0.5
0.2 0.2 0.3
1.5 1.6 2.1
Group
” pmb, therapy.
postmenopausal
bleeding;
ERT,
estrogen
replacement
The mean impedance to uterine arterial blood flow as reflected by the PI is shown in Table 3. The maximum value (1.49) in patients with endometrial cancer and postmenopausal bleeding is below the minimum value (1.95) in patients with uterine blood loss due to other factors. The patient with endometrial cancer (FIG0 stage IA) but no sign of uterine blood loss had a mean uterine arterial PI of 1.10, which was below the minimum value (1.50) of the reference groups of women not receiving ERT. One-way analysis of variance showed a statistically significant difference in the PI between all five groups (P < .OOOl), but not between the 17 women with uterine bleeding but no evidence of cancer and the 68 who did not have postmenopausal bleeding and were not receiving ERT (P = 0.09). The values for those two groups were combined for further analysis. Similarly, the values for all women with cancer (16 with uterine bleeding, 1 without) were reconsidered as one group.
256
BOURNE n = 18 0 0 0
n = 84
ET AL.
n = 35
0 00 0
0
0
o.oJ Cancer
Non-cancer (no ERT)
Non-cancer (with
FIG. 1. Individual values for endometrial thickness in patients with and without endometrial cancer (note log scale on y axis).
Overall Predictive Value of ET and PI
The individual values for endometrial thickness in women from the three groups (cancer with and without postmenopausal bleeding, noncancer with and without postmenopausal bleeding, and noncancer without irregular uterine blood loss and receiving ERT) are shown in Fig. 1. An arbitrary value of 1.0 cm was selected to discriminate between the groups. The rate of detection [and 95% confidence interval (CI)] of endometrial cancer was 14/17 (82%, CI 54-96%), while the false-positive rate (FPR) was 4/84 (5%, CI l-12%) for asymptomatic
I
n = 18
n = 85 0
IA
EQT)
n = 35
16
IC IIA Stage of cancer
0
116
IIIA
FIG. 3. Relationship between pulsatility index and stage of endometrial cancer.
women not receiving ERT and 6/35 (17%, CI 7-34%) for asymptomatic women receiving ERT. The corresponding values for the PI are shown in Fig. 2. An arbitrary value of 1.50 was selected to distinguish between the groups. The observed detection rate for endometrial cancer was 17/17 (lOO%, 95% CI 81-lOO%), while the FPR was l/85 (1%) for women with and without postmenopausal bleeding but not receiving ERT and 4/35 (11%) for asymptomatic women receiving ERT. The data in the cancer and the noncancer/no-ERT groups were logarithmically transformed to approximate normal distributions and the overlap was estimated using the means and standard deviations. We calculated that a cutoff value of 2.00 would give a detection rate of 99.0% with a falsepositive rate of 2.6%. Consequently, the calculated predictive value of a positive screening result for women with postmenopausal bleeding (assuming a 10% prevalence of endometrial cancer) would be 80.9% and the predictive value of a negative screening result would be 99.9%. The positive and negative predictive values of screening results for postmenopausal women without postmenopausal bleeding attending a screening clinic (assuming a 0.00071% prevalence of endometrial cancer) would be 9.1 and 99.9%, respectively. Factors Affecting PI
Cancer
Non-cancer ho EAT)
Non-cancer (With
EAT)
FIG. 2. Individual values for the mean pulsatile index derived from flow velocity waveforms from the right and left uterine arteries in patients with and without uterine cancer (note log scale on y axis).
The relationship between the PI and the stage of endometrial cancer is shown in Fig. 3. There is a trend toward a reduction in the impedance to arterial blood flow as the disease progresses. The two patients with a low PI at stages IA and IB were aged 83 and 75 years, respectively. The effect on the PI value of the number of years since
ENDOMETRIAL
Non-cancer (no ERT)
Cancer
CANCER
257
DETECTION
the PI for patients with cancer and a slight upward trend in the PI for women with postmenopausal bleeding but no cancer.
0
Intratumoral Blood FLOW
5.0-
.
a a 0
2 -cl $4
0 2.0.
,” 2 .r( +r (0 rn l-l 2
Oil 0 8 II 0 0 0 :
0 0
0
a
0
0
1.0.
0
00 0
0 0
0
OO
The PI was determined within the endometrial cancer in 10 of 18 subjects (5/5 with color flow imaging). The value was similar to the mean uterine arterial PI in 3 cases and lower in 7 subjects (mean 0.49, range 0.29-0.92). It was not possible to accurately identify different FVWs in the other 7 cases using pulsed Doppler without color flow imaging. This technique was available to identify intratumoral areas of neovascularization in five cases and an example is shown in Fig. 5. DISCUSSION
Transvaginal ultrasonography is a relatively new technique that has some important advantages over the use of transabdominal transducers. In particular there is no need for the women being scanned to have a full bladder, Iears2 sin2 mirYopa2u0sd"{ear: sinc5eme'n0opa2u0sF and the use of higher-frequency ultrasound enables the FIG. 4. Relationship between pulsatility index and years since menmorphology of the reproductive tract to be assessed in opause in women with and without endometrial cancer, not receiving greater detail. Descriptions of the uterus by this approach hormone replacement therapy (note log scale on x and y axes). have been shown to correlate well with those observed at surgery [7]. C onsequently transvaginal ultrasonography menopause in patients with cancer and in women without can be used to assess the extent to which cancer of the endometrium has penetrated the myometrium or cervix, postmenopausal bleeding or cancer and not receiving ERT is shown in Fig. 4. There is a downward trend in and hence can be of value preoperatively for staging con-
FIG. 5. Sonogram showing intratumoral
vascularization in a patient with endometrial cancer (FIG0 stage IB).
258
BOURNE
firmed disease [17-201. Furthermore, the results of our study of ET in postmenopausal women show that an ET >l.O cm has a high predictive value for the presence of endometrial cancer. The detection rate for this disease of 82% with a FPR of 5% for postmenopausal women not receiving ERT shows that the technique is clinically useful in the absence of facilities to measure uterine blood flow. The use of transvaginal pulsed Doppler ultrasonography in measuring blood flow impedance in the uterine arteries of nonpregnant women has been described previously [lo]. Moreover, we have reported how transvaginal color flow imaging may be used to identify the uterine arteries and obtain optimum FVWs for the analysis of blood flow during the ovarian and menstrual cycles [ll]. The further application of these techniques shows that there is an apparent distinction in average PI for the two uterine arteries between those postmenopausal patients presenting with uterine bleeding in the presence of endometrial cancer and other women with the same symptom but with no apparent cause [21]. Infection, however, may also affect uterine blood flow (unpublished observations) and the presence of pathogenic bacteria should be considered before a diagnosis is made. A cutoff value of 1.50 was arbitrarily selected for the PI to obtain a high detection rate and a low false-positive rate for endometrial cancer. One case of the disease was detected in a woman without postmenopausal bleeding, and the FPR for all women not receiving ERT was about 1% (l/85). It is of interest that the woman with a false-positive result from the measurement of PI had an ET
