Two- and three-dimensional transvaginal ultrasound with power … · Transvaginal ultrasound (TVS) is the most cost-effective ﬁrst-line test in the diagnostic work-up of women with

Ultrasound Obstet Gynecol 2015; 45: 734–743Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.13421

Two- and three-dimensional transvaginal ultrasound withpower Doppler angiography and gel infusion sonographyfor diagnosis of endometrial malignancy

M. DUEHOLM*, J. W. CHRISTENSEN*, S. RYDBJERG*, E. S. HANSEN† and G. ØRTOFT**Department of Obstetrics and Gynecology, Aarhus University Hospital, Aarhus, Denmark; †Department of Pathology, Aarhus UniversityHospital, Aarhus, Denmark

KEYWORDS: color Doppler; endometrial neoplasms; postmenopause; ultrasonography; uterine hemorrhage

ABSTRACT

Objectives To evaluate the diagnostic efficiency oftwo-dimensional (2D) and three-dimensional (3D)transvaginal ultrasonography, power Doppler angiogra-phy (PDA) and gel infusion sonography (GIS) at offlineanalysis for recognition of malignant endometrium com-pared with real-time evaluation during scanning, and todetermine optimal image parameters at 3D analysis.

Methods One hundred and sixty-nine consecutive womenwith postmenopausal bleeding and endometrial thick-ness ≥ 5 mm underwent systematic evaluation of endome-trial pattern on 2D imaging, and 2D videoclips and 3Dvolumes were later analyzed offline. Histopathologicalfindings at hysteroscopy or hysterectomy were used asthe reference standard. The efficiency of the differenttechniques for diagnosis of malignancy was calculatedand compared. 3D image parameters, endometrial vol-ume and 3D vascular indices were assessed. Optimal 3Dimage parameters were transformed by logistic regressioninto a risk of endometrial cancer (REC) score, includ-ing scores for body mass index, endometrial thicknessand endometrial morphology at gray-scale and PDA andGIS.

Results Offline 2D and 3D analysis were equivalent,but had lower diagnostic performance compared withreal-time evaluation during scanning. Their diagnosticperformance was not markedly improved by the additionof PDA or GIS, but their efficiency was comparable withthat of real-time 2D-GIS in offline examinations of goodimage quality. On logistic regression, the 3D parametersfrom the REC-score system had the highest diagnosticefficiency. The area under the curve of the REC-scoresystem at 3D-GIS (0.89) was not improved by inclusionof vascular indices or endometrial volume calculations.

Correspondence to: Dr M. Dueholm, Department of Obstetrics and Gynecology, Aarhus University Hospital, Brendstrupgaardsvej 100,8200 Aarhus N, Denmark (e-mail: [email protected])

Accepted: 16 May 2014

Conclusion Real-time evaluation during scanning is mostefficient, but offline 2D and 3D analysis is useful forprediction of endometrial cancer when good image qualitycan be obtained. The diagnostic efficiency at 3D analysismay be improved by use of REC-scoring systems, withoutthe need for calculation of vascular indices or endometrialvolume. The optimal imaging modality appears to bereal-time 2D-GIS. Copyright © 2014 ISUOG. Publishedby John Wiley & Sons Ltd.

INTRODUCTION

Transvaginal ultrasound (TVS) is the most cost-effectivefirst-line test in the diagnostic work-up of women withpostmenopausal bleeding1–3 and may identify those atvery low risk of endometrial cancer4,5. In patients withincreased endometrial thickness, identification of thoseat high risk for endometrial cancer at the first-line TVSexamination may allow them to be referred for fast-trackinvestigation and staging. This could shorten the timefrom first investigation to treatment. In experiencedhands, evaluation of endometrial pattern on TVS, salinecontrast sonohysterography (SCSH) and Doppler6–9 yieldpromising results for identification of patients at high riskof endometrial cancer6–12. Ideally, first-line investigationsshould be performed by experienced assessors to ensurefast-track referral, but they are available to perform suchprocedures only rarely.

Assessment at two-dimensional (2D) TVS is usu-ally performed in real-time and requires skill in imageoptimization and pathology recognition. Modern ultra-sound machines and software have simplified imageoptimization, but recognition of pathology continuesto require considerable experience. Offline analysis ofthree-dimensional (3D) volumes is reliable for assessing

Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd. ORIGINAL PAPER

3D ultrasound and endometrial cancer 735

adnexal masses13, and offline analysis of 3D volumes or2D video recordings by experts could compensate forthe lack of local experts and could hence be an efficientalternative.

Real-time evaluation of endometrial pattern based on ascore system at 2D-TVS and gel infusion sonography (GIS)is highly efficient for diagnosis of malignancy14. The scoresystem may be improved at 3D-volume assessment byinclusion of 3D vascular indices and endometrial volume(EV) measurements10–12. Thus, we hypothesize that 3Dultrasound may increase the diagnostic performance of2D ultrasound for endometrial malignancy. We proposea rethink of the diagnostic approach, with the currentpractice of local first-line assessment being supplementedby secondary evaluation of 3D volumes by non-localexperienced assessors.

The aim of this study was to assess and compare thediagnostic efficiency of real-time analysis at 2D-TVSand offline analysis of 2D- and 3D-TVS, powerDoppler angiography (PDA) and GIS for discriminationbetween benign and malignant endometrial conditionsin women with postmenopausal bleeding and thickenedendometrium and, furthermore, to evaluate optimal imageparameters for 3D analysis.

METHODS

Two investigators, who were blinded to patients’ identitiesand to the results of all prior microscopic specimensand diagnoses, performed TVS with Doppler and GIS in169 consecutive women (93 unselected and 76 referredfrom other hospitals) with postmenopausal bleeding andendometrial thickness ≥ 5 mm, from October 2010 toFebruary 2012, at Aarhus University Hospital, Denmark.2D clips and 3D volumes (TVS, TVS with Doppler andGIS) were also stored after each examination and lateranalyzed offline. Both examiners had several years ofexperience in the staging of endometrial cancer. Theexclusion criteria used in the present study have beendescribed elsewhere14. The Central Denmark RegionalEthics Committee approved the study protocol. Thereference standard was histopathological findings athysteroscopy and/or hysterectomy; as this was standardprocedure in the department, the committee foundthat no informed consent for this procedure wasneeded.

Ultrasound analysis was performed in four rounds,as follows. In the first round, real-time evaluation atTVS with PDA and at TVS with GIS was performedand 2D clips and 3D volumes were stored. Afterultrasound evaluation, a standard history with respectto clinical variables was obtained. In the second round,stored 2D clips (2D-TVS, 2D-PDA and 2D-GIS) wereassessed. In the third round, 3D volumes (3D-TVS,3D-PDA and 3D-GIS) were evaluated. In the fourthround, data on vascular indices and EV at 3D-PDA wereobtained.

Each round of analysis was followed by systematicevaluation of pattern (SEP), i.e. a systematic evaluation

of the endometrial pattern, performed according to theterms introduced by the International Endometrial TumorAnalysis (IETA) group15. This included completion ofa standard systematic form with IETA parameters; asubjective impression of malignancy was then formedat the end of each examination (real-time 2D-PDA and2D-GIS, and offline 2D-TVS, 2D-PDA, 2D-GIS, 3D-TVS,3D-PDA and 3D-GIS).

Included for offline analysis were all patients for whom2D clips or correctly stored 3D volumes, obtained at anyof the three examinations (TVS, PDA or GIS), could beevaluated (2D-TVS (n = 165), 2D-PDA (n = 153), 2D-GIS(n = 159), 3D-TVS (n = 169), 3D-PDA (n = 155) and3D-GIS (n = 160)). A total of 1142 2D and 3D offlineevaluations were undertaken at TVS, PDA and TVS withGIS. In addition, vascular indices and EVs were calculatedin 155 3D-PDA volumes.

Real-time imaging

Transvaginal ultrasound (TVS) and power Dopplerangiography (PDA)

Examinations were performed, according to a predeter-mined scanning protocol, with a Voluson E8 Expert (GEHealthcare Ultrasound, Milwaukee, WI, USA) ultrasoundmachine, equipped with a multifrequency (5–12 MHz)endovaginal probe.

Endometrial structure and thickness, and analysis ofvessels on PDA, were evaluated using a standard formmeeting the IETA criteria15, as previously described14; theform comprised the following elements: internal endome-trial echo structure (hyperechogenic/hypoechogenic/iso-echogenic, cystic (regular/irregular), homogeneity/hetero-geneity; endomyometrial border, subendometrial halo(endomyometrial junction): visualization and interrup-tion of subendometrial halo, regular/irregular, homogene-ity/heterogeneity, and bright line.

A 2D power Doppler gate was activated to assess vas-cularization from the myometrium and the endometrium.Power Doppler settings were set to achieve maximumsensitivity to detect low-velocity flow without noise, usingpredetermined, standardized settings (frequency, 6 MHz;power Doppler gain, 50; dynamic range, 10 dB; edge,1; persistence, 2; color map, 1; gate, 2; filter, 3; pulserepetition frequency, 0.6).

PDA analysis included visual evaluation of the vessels(Doppler parameters) for15: dominance (single or double);origin (focal or multifocal); number (few or multiple);large vessels (yes/no); branching (yes/no); if branching,regularity (regular/irregular); presence of: color splashes,densely packed vessels, circular flow.

After TVS and Doppler examinations, evaluation ofthe image parameters was entered on the standard form.After SEP, the main diagnosis of malignancy was madesubjectively in the presence of: heterogenic endometriumwith irregular or interrupted endomyometrial junction;indistinct endomyometrial border; multiple vessels onDoppler sonography.

Copyright © 2014 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2015; 45: 734–743.

736 Dueholm et al.

Gel installation sonography (GIS)

GIS was performed as described previously14. A smallflexible sterile catheter (infant-feeding tube, Unometricno: 227581, Abena A/S, Aabenraa, Denmark) mountedwith a 10-mL syringe containing Instillagel® (E.TjellesenA/S, Lynge, Denmark) was introduced into the uterinecavity. During instillation of gel (GIS), the pressure wasadjusted manually until sufficient for expansion of theuterine cavity. The distension was observed continuouslyby TVS, and was continued until the entire uterine cavitywas clearly visible. The uterine cavity was evaluated insagittal and coronal views. The same parameters wereevaluated at GIS as had been at TVS. Additionally,endometrial thickness was measured in the sagittal plane,as the sum of the maximum endometrial thickness at theanterior and the posterior walls.

A standardized coding sheet was used to record theadditional GIS parameters defined by the IETA system15,as previously described14: structure of the endometrialsurface (smooth, polypoid, irregular), size of localizedand/or diffuse lesions (i.e. lesion < or > 25% of surface),and structure of the surface of local lesions (regularor irregular). In order to avoid missing values, a thirdcategory, ‘indefinable’, was used for each classification.

After SEP, the main diagnosis of malignancy at GISwas made subjectively in the presence of: TVS findingsof malignancy and/or irregular surface of a localized ordiffuse lesion at GIS.

3D volume acquisition

A 3D volume acquisition box was placed over the uterus.The patient was asked to remain as still as possible, and atleast two 3D volumes were acquired, with a sweep angle of90–120◦. 3D-PDA of the uterus was then performed withthe previously defined fixed settings. Volume acquisitionwas repeated when artifacts appeared.

Offline analysis

2D clips and 3D volume storage

At each examination (TVS, Doppler, GIS), three 2Dvideoclips were obtained, including two sweeps in thesagittal plane and one in the transverse plane. The systemwas set for retrospective storage of clips of the last 5–10seconds of scanning. Two 3D volumes at each of 3D-TVS,3D-PDA and 3D-GIS were stored for later evaluation on apersonal computer using the virtual organ computer-aidedanalysis (VOCAL) software (GE Healthcare Ultrasound).

Evaluation of stored 2D videoclips

2D clips obtained during the real-time investigations werestored (by S.R.) in a separate file using the patients’ID numbers. All 169 examinations (2D-TVS + 2D-PDA + 2D-GIS) were then given arbitrarily a new identitynumber. Three months after the investigation was

completed, all 2D clips were evaluated (by M.D.) withoutknowledge of the patient’s identity and pathology. Theclips were assessed in random patient order, but for eachpatient in the order 2D-TVS, then 2D-PDA, then 2D-GIS.

On a standard form, the assessor noted the patient’s ageand entered data on the quality of the images (score, 1–5)and the same morphological endometrial parameters asthose obtained at the primary evaluation14,15. Again, afterSEP, a subjective diagnosis of malignancy was given atthe end of each evaluation, first for TVS, then for Dopplerand finally for GIS.

Evaluation of stored 3D volumes

The stored 3D volumes from the 3D-TVS, 3D-PDAand 3D-GIS examinations were evaluated using VOCALsoftware (by M.D.) 3 months after the 2D-videoclipevaluations had been completed. Again, all examinationswere given a new identification number and wereperformed in random patient order, with the examinerblinded to the previous date of examination and thepatient’s identity. Each patient’s examination consisted ofsix to 10 volume determinations in the order 3D-TVS,then 3D-PDA, then 3D-GIS. For each examination,we opened all stored volumes and used the volumewith the highest quality for evaluation, according tothe morphological criteria stated above. A diagnosis ofmalignancy was based on subjective impression after SEPand stated at the end of each examination and the results,including image quality, were entered into a standardform.

Analysis of 3D power Doppler indices

The 3D-PDA volumes were analyzed for 3D powerDoppler indices 4 months later. Using the VOCALprogram, we evaluated the endometrial area manuallyin the coronal or ‘C’-plane. Using a rotational techniquewith a 9◦ step, as described previously11,16, we obtained20 endometrial slices that outlined the endometrium atthe endomyometrial junction, from the fundus to theinternal cervical opening. The VOCAL program wasused to calculate the EV and three 3D power Dopplerindices: the vascularization index (VI), flow index (FI)and vascularization flow index (VFI). VI measures thenumber of color voxels in the volume. It may representthe vessels in the tissue and is expressed as a percentage.FI is the mean color value in the color voxels. It indicatesthe average intensity of blood flow and is expressed as anumber from 0 to 100. VFI is the mean color value in allthe voxels in the volume. It represents both vascularizationand blood flow and is also expressed as a number from 0to 100.

Doppler and risk of endometrial cancer (REC) scores

The most optimal Doppler image parameters at real-time2D evaluation and at offline 2D videoclip evaluation bytwo observers were evaluated, and a Doppler score was



developed. This score was obtained by simple addition ofthe following Doppler parameters: vessels, but no singleor double dominant vessel (1 point); multiple vessels(more than four or five) (1 point); large vessels (1 point);color-splash/densely packed vessels (1 point)14.

A score system for evaluation of the risk of endome-trial cancer (REC score) was designed based on different2D image parameters analyzed by logistic regression andthe most optimal real-time evaluation of 2D imagingin this population14. The REC-score system included:body mass index (BMI) ≥ 30 (1 point); total endometrialthickness ≥ 10 mm (1 point); total endometrial thick-ness ≥ 15 mm (1 point); vascularity, but no single/doubledominant vessel (present = 1 point); multiple (more thanfour or five) vessels (present = 1 point); large vessels(present = 1 point); color-splash/densely packed vessels(present = 1 point); interrupted endomyometrial junction(present = 1 point); irregular surface at GIS (present = 1point). Simple addition of these values constituted theREC score14. A diagram of parameters in the REC-scoresystem and accompanying ultrasound images is shown inFigure 1.

Pathology

All patients underwent hysteroscopy and/or hysterectomy;microscopic pathology was the reference standard forthese examinations14. Removal of all focal changes wasattempted at hysteroscopy; in cases of large, diffuse orlocalized changes, resectoscopic biopsies were taken fromthe area of the endometrium with the largest changes.Three to five biopsies were sampled. In cases of normalhysteroscopic findings, one biopsy was taken from theanterior wall of the uterine cavity, one from the posteriorwall of the uterine cavity, and curettage was performed atthe end of the hysteroscopy. Two pathologists specializedin gynecological oncology evaluated all specimens.

Statistics and analysis

Data were analyzed using STATA (Statistic DataAnalysis, STATA Corp., TX, USA). Continuous data andnormally distributed data are expressed as mean ± SD.Wilcoxon’s signed rank test and McNemar’s test wereused for comparison of the scored quality of theexamination. We evaluated and compared the diagnosticperformance of real-time 2D imaging vs offline 2Dvideoclip analysis vs offline 3D volume analysis. Thediagnostic performance, with regard to discriminationbetween benign and malignant endometrium, of thedifferent techniques was evaluated in terms of the areasunder the receiver–operating characteristics (ROC) curve(AUC), sensitivity, specificity, positive likelihood ratio(LR+) and negative likelihood ratio (LR–). ROC-curveanalysis was used to compare the diagnostic performanceof the different techniques. ROC-curve analysis wasperformed for all parameters measured by 3D imaging(pattern variables and parameters calculated by VOCALprogram (Doppler indices and endometrial volume)). The

sensitivity, specificity, LR+ and LR– of the most optimalcut-off points were calculated.

Stepwise multivariate logistic regression was performedto build a model that could be used to predict malignancyat 3D evaluation. A maximum of four fitting variableswere allowed in the model. Optimal cut-off points werecalculated using a 45◦ tangent line intersection or thesmallest sum of residual sensitivity and specificity. Themost optimal ROC curve calculated by logistic regressionwas compared to the REC score developed at real-time 2Devaluation; χ2 tests were used for discrete data. Statisticaltests were two-tailed; P < 0.05 was considered statisticallysignificant.

RESULTS

The mean age of patients was 64.1 (range, 45–89) years.There was endometrial cancer present in 69 (40.8%) ofthe patients, hyperplasia in 20 (11.8%) and endometrialpolyps in 51 (30.2%).

Overall diagnostic efficiency of real-time 2D imagingand offline 2D videoclip and 3D volume evaluations

Table 1 shows the diagnostic efficiency of real-time 2Dimaging and the offline evaluation of 2D videoclips and3D volumes. Sensitivities ranged from 73% to 89%and specificities were close to 90% at the differentoffline 3D and 2D evaluations. Addition of PDA andGIS did not change markedly the diagnostic efficiencyfor offline 2D and 3D evaluation. Real-time evaluationduring 2D-GIS had better diagnostic efficiency than didanalysis of both 3D-GIS volumes and 2D-GIS videoclips(153 patients, P < 0.05). Figure 2 displays the diagnosticperformance (AUC) obtained by the different real-timeand offline 2D and 3D techniques. The figure includesonly the 136 patients in whom all evaluations withall methods were completed. There was no significantdifference between AUCs for offline 2D-TVS and 3D-TVSor between AUCs for offline 2D-PDA, offline 3D-PDAand real-time 2D-PDA evaluations.

Diagnostic efficiency and quality of offline 2D clipsand 3D evaluations

The scores for quality were significantly better for 2D-GISvideoclips than for 3D-GIS volumes (P < 0.001), with13.2% of the 2D clips being of low quality, comparedwith 22.5% for the 3D volumes.

When we restricted evaluations to cases for which both2D videoclips and 3D volumes of sufficient quality hadbeen obtained (Table 1), the diagnostic efficiency of these2D videoclips or 3D volumes was not significantly lowerthan the diagnostic efficiency of real-time evaluationduring scanning. Cases with consensus between 2Dvideoclips and 3D volumes had a high diagnosticefficiency. The discrepancies (16% between 2D-TVS and3D-TVS, 5% between 2D-PDA and 3D-PDA, and 9%


738 Dueholm et al.

2

3(a)

(e)

6 8

97

1

54

Figure 1 Diagrams and ultrasound images illustrating the nine parameters in the risk of endometrial cancer (REC) score system14.(a) Diagram representing REC-score components for: (1) body mass index ≥ 30 (+1 point), (2) total endometrial thickness ≥ 10 mm(+1 point), (3) total endometrial thickness ≥ 15 mm (+1 point), (4) interrupted endomyometrial junction (+1 point) and (5) irregular surfaceat gel instillation sonography (GIS) (+1 point). Accompanying ultrasound images illustrate interrupted endomyometrial junction (c) andirregular endometrial surface at two-dimensional GIS (b) and three-dimensional (3D) GIS (d). (e) Diagram representing: (6) vascularity withno single/double vessel (+1 point), (7) multiple (more than four or five) vessels (+1 point), (8) large vessels (+1 point) and (9) color-splash/densely packed vessels (+1 point). Accompanying ultrasound images illustrate a single vessel (3D endometrial polyp) (f), denselypacked vessels (g) and multiple and large vessels (h). Simple addition of the REC-score points constituted the REC score.



Tab

le1

Dia

gnos

tic

effic

ienc

yfo

ren

dom

etri

alm

alig

nanc

yof

real

-tim

ean

dof

fline

anal

ysis

oftw

o-(2

D)

and

thre

e-(3

D)

dim

ensi

onal

tran

svag

inal

ultr

asou

nd(T

VS)

,pow

erD

oppl

eran

giog

raph

y(P

DA

)an

dge

linf

usio

nso

nogr

aphy

(GIS

),in

pati

ents

wit

hpo

stm

enop

ausa

lble

edin

gan

den

dom

etri

alth

ickn

ess≥5

mm

Sens

itiv

ity

Spec

ifici

ty

N%

(95%

CI)

n/N

%(9

5%C

I)n/

NL

R+

(95%

CI)

LR

–(9

5%C

I)A

UC

(95%

CI)

AU

C(9

5%C

I):

Pat

ient

sw

hoha

dal

ltes

ts

Rea

l-ti

me

eval

uati

onN

=15

82D

-PD

A16

987

.0(7

7–

94)

60/6

988

.0(8

0–

94)

88/1

007.

3(4

.2–

12.4

)0.

15(0

.08

–0.

27)

0.87

5(0

.82

–0.

93)

0.87

9(0

.83

–0.

93)

2D-G

IS15

888

.6(7

9–

95)

62/7

095

.5(8

9–

99)

84/8

819

.5(7

.5–

50.9

)0.

12(0

.06

–0.

23)

0.92

0(0

.88

–0.

96)

0.92

0(0

.88

–0.

96)

Offl

ine

eval

uati

on2D

vide

oclip

sN

=14

22D

-TV

S16

580

.6(6

9–

89)

54/6

788

.8(8

1–

94)

87/9

87.

2(4

.1–

12.6

)0.

22(0

.13

–0.

36)

0.84

7(0

.79

–0.

90)

0.83

8(0

.78

–0.

90)

2D-P

DA

153

79.7

(67

–89

)47

/59

89.4

(81

–95

)84

/94

7.5

(4.1

–13

.6)

0.23

(0.1

4–

0.38

)0.

845

(0.7

9–

0.91

)0.

838

(0.7

8–

0.90

)2D

-GIS

159

84.3

(74

–92

)59

/70

83.1

(74

–90

)74

/89

5.0

(3.1

–8.

0)0.

19(0

.11

–0.

33)

0.83

7(0

.78

–0.

90)

0.83

9(0

.78

–0.

90)

3Dvo

lum

esN

=14

53D

-TV

S16

969

.6(5

7–

80)

48/6

992

.0(8

5–

97)

92/1

008.

7(4

.4–

17.2

)0.

33(0

.23

–0.

48)

0.80

8(0

.75

–0.

87)

0.80

3(0

.74

–0.

87)

3D-P

DA

155

72.6

(60

–83

)45

/62

89.2

(81

–95

)83

/93

6.8

(3.7

–12

.4)

0.31

(0.2

0–

0.46

)0.

809

(0.7

4–

0.87

)0.

809

(0.7

4–

0.87

)3D

-GIS

160

78.6

(67

–88

)55

/70

87.8

(79

–94

)79

/90

6.4

(3.7

–11

.3)

0.24

(0.1

6–

0.39

)0.

832

(0.7

7–

0.89

)0.

832

(0.7

7–

0.89

)In

vest

igat

ion

wit

hag

reem

ent

at2D

and

3D(o

fflin

ean

alys

is)

N=

120

TV

S13

782

.4(6

9–

92)

42/5

197

.7(9

2–

98)

84/8

635

.4(9

.0–

140)

0.18

(0.1

0–

0.33

)0.

900

(0.8

5–

0.96

)0.

921

(0.8

7–

0.97

)D

oppl

er14

383

.3(7

1–

92)

45/5

494

.4(8

7–

98)

84/8

914

.8(6

.3–

35)

0.18

(0.1

0–

0.32

)0.

889

(0.8

3–

0.94

)0.

925

(0.8

8–

0.98

)G

IS14

088

.1(7

7–

95)

52/5

990

.1(8

2–

96)

73/8

18.

9(4

.6–

17.3

)0.

13(0

.07

–0.

27)

0.89

1(0

.84

–0.

94)

0.91

6(0

.86

–0.

97)

Imag

equ

alit

y1

–3*

(offl

ine

anal

ysis

)N

=11

5

3D-G

IS12

481

.1(6

8–

91)

43/5

390

.1(8

1–

96)

64/7

18.

2(4

.0–

16.8

)0.

21(0

.12

–0.

37)

0.85

6(0

.79

–0.

92)

0.87

9(0

.82

–0.

94)

2D-G

IS13

589

.1(7

8–

96)

49/5

586

.3(7

7–

93)

69/8

06.

5(3

.7–

11.3

)0.

13(0

.06

–0.

27)

0.87

7(0

.82

–0.

93)

0.87

8(0

.82

–0.

94)

Dia

gnos

isw

asba

sed

onsu

bjec

tive

impr

essi

onof

mal

igna

ncy

follo

win

gsy

stem

atic

eval

uati

onof

endo

met

rial

patt

ern.

Ref

eren

cest

anda

rdw

ashi

stop

atho

logy

athy

ster

osco

pyor

hyst

erec

tom

y.*I

mag

equ

alit

y1

–3:

qual

ity

exce

llent

toin

term

edia

te.C

ompa

riso

nof

area

sun

der

the

rece

iver

–op

erat

ing

char

acte

rist

ics

curv

es(A

UC

):3D

-TV

Svs

2D-T

VS

offli

ne(n

=16

3):N

S;3D

-PD

Avs

2D-P

DA

offli

nevs

real

-tim

e2D

-PD

A(n

=15

1):N

S;3D

-GIS

vs2D

-GIS

offli

nevs

real

-tim

e2D

-GIS

(n=

153)

:P<

0.05

.LR

+,po

siti

velik

elih

ood

rati

o;L

R–

,neg

ativ

elik

elih

ood

rati

o;N

S,no

tsi

gnifi

cant

.


740 Dueholm et al.

0.80

0.75

PDAGIS

2D-TVS2D-PDA

2D-GIS3D-TVS

3D-PDA3D-GIS

0.85

AU

C

0.90

0.95

Figure 2 Areas under the receiver–operating characteristics curves,with 95% CIs, of two- and three-dimensional transvaginalultrasonography (2D-TVS, 3D-TVS), power Doppler angiography(2D-PDA, 3D-PDA) and gel infusion sonography (2D-GIS, 3D-GIS)at offline analysis and real-time analysis during imaging (PDA,GIS), for diagnosis of endometrial malignancy in patients withpostmenopausal bleeding and endometrial thickness ≥ 5 mm; onlythe 136 patients in whom all evaluations by all techniques werecompleted are included.

between 2D-GIS and 3D-GIS) revealed a small group ofindefinable evaluations at offline analysis.

3D power Doppler calculations

Table 2 summarizes endometrial thickness, the vascularindices and EV according to different kinds of pathology.ET, EV, VI, FI and VFI were all clearly associated withcancer (P < 0.01).

Table 3 shows the diagnostic efficiency of the differentcriteria used at the different 3D evaluations. EVcalculation at 3D analysis did not have a higher diagnosticefficiency than did measurement of the endometrialthickness. The individual, different Doppler parameters(vessel patterns a–g) at 3D-PDA had large AUCs thatwere comparable to those of VI, VFI and FI. Dopplerscore (comprising patterns c, e, f and g) had an AUCabove 0.80. Addition of or replacement with VI, VFI orFI at multivariate logistic regression did not increase thediagnostic efficiency.

The model which included BMI, endometrial thickness,presence of an interrupted endomyometrial junction andDoppler score (Table 3, Model 3), had a diagnosticefficiency (AUC, 0.879) that was higher (P = 0.01) thanthat of the primary subjective evaluation (SEP) at 3D-PDA(AUC, 0.809; Table 1). Evaluation of 3D-GIS (Model 4,Table 3), with BMI, an interrupted endomyometrialjunction, Doppler score and irregular endometrial surfaceat 3D-GIS, had the highest diagnostic efficiency onmultivariate regression, with an AUC (0.908) that wasclearly higher than that of the subjective SEP at 3D-GIS(AUC, 0.832; Table 1). Application of the REC-scoresystem at 3D-PDA or 3D-GIS had comparable efficiencycompared with their respective models (Models 3 and 4).REC scores also had a higher calculated diagnosticefficiency than that of the subjective SEP at 3D-PDA or3D-GIS (Table 1; P = 0.01). The REC scores for 3D-PDAand 3D-GIS had higher AUCs than had Model 1 (Table 3).

DISCUSSION

At offline analysis of 3D volumes and 2D videoclips,endometrial cancers were identified with sensitivities of73–89%, with very few false-positive cases. However,the diagnostic efficiency of neither 2D videoclips nor 3Dvolumes reached the high diagnostic level of evaluationat real-time imaging in the hands of an experiencedinvestigator using GIS. Addition of vascular indicesand volume analysis did not improve the efficiency at3D analysis. The parameters included in the REC-scoresystem had the highest diagnostic efficiency at both 2D-and 3D-GIS evaluation14.

A problem with the offline evaluation of 3D volumesand 2D videoclips seemed to be loss of image quality; thediagnostic efficiency was acceptable when restricted to thethree-quarters of the images that had high image qualityor evaluations in which there was agreement between 2Dvideoclip and 3D volume determinations. Competencein image optimization is important to obtain images ofsufficient quality for later offline analysis.

Fast-track investigation of all postmenopausal womenwith increased endometrial thickness is very costly.Improved selection at first-line TVS seems rational andwas effective at real-time evaluation during TVS in expe-rienced hands in this population, as it has been in otherstudies6–10,12,17. The main obstacle to implementation

Table 2 Endometrial thickness and volume and the vascular indices in 169 patients with postmenopausal bleeding and endometrialthickness ≥ 5 mm, according to type of endometrial pathology

Vascular index

Pathology NEndometrial

thickness (mm)Endometrialvolume (cm3) VI VFI FI

Endometrial cancer 69 18.4 (15.5–21.3) 20.5 (7.7–33.2) 13.81 (9.0–18.6) 4.09 (2.6–5.6) 27.73 (25.7–29.8)Hyperplasia 20 12.3 (9.5–15.2) 5.47 (3.1–7.8) 4.85 (2.1–7.6) 1.42 (0.58–2.3) 24.28 (20.8–27.8)Benign uterine polyps 51 9.9 (8.8–11.1) 5.88 (3.1–8.6) 5.85 (2.7–9.1) 1.13 (0.65–1.6) 22.39 (20.4–24.3)Other 29 8.1 (6.7–9.6) 5.62 (0.10–11.1) 3.83 (1.5–6.2) 1.02 (0.35–1.7) 21.63 (19.0–24.2)

Data are given as mean (95% CI). P < 0.01 for cancer groups vs all other groups, for endometrial thickness, endometrial volume,vascularization index (VI), vascularization flow index (VFI) and flow index (FI).



Table 3 Diagnostic efficiency of different parameters for prediction of endometrial cancer at three-dimensional (3D) transvaginal ultrasound(TVS), power Doppler angiography (PDA) and gel infusion sonography (GIS), in patients with postmenopausal bleeding and endometrialthickness ≥ 5 mm

Parameter AUC (95% CI)

Correctlyclassified

(%) Cut-offSens.(%)

Spec.(%) LR+ LR–

3D-TVS (N = 169)ET 0.788 (0.71–0.86) 78.7 ≥ 14.9 60.9 91.0 6.8 0.43EV 0.745 (0.67–0.82) 70.4 4.495 72.5 69.0 2.3 0.40Heterogeneous echogenicity 0.573 (0.51–0.63) 52.1 85.5 75.0 1.2 0.50Non-cystic echogenicity 0.625 (0.55–0.70) 61.0 71.0 54.0 1.5 0.54Irregular endomyometrial junction 0.725 (0.66–0.79) 72.2 73.9 71.0 2.6 0.37Interrupted endomyometrial junction 0.774 (0.71–0.84) 80.5 60.9 94.0 10.1 0.42Non-intact endomyometrial junction 0.787 (0.73–0.85) 81.7 62.3 95.0 12.5 0.40

3D-GIS (N = 160)ET 0.807 (0.74–0.88) 77.5 ≥ 12.9 78.6 76.7 3.4 0.28Interrupted endomyometrial junction 0.822 (0.76–0.88) 83.8 70.0 94.4 12.6 0.31Non-intact endomyometrial junction 0.829 (0.77–0.89) 84.4 71.4 94.4 12.9 0.31Irregular endometrial surface 0.785 (0.72–0.85) 78.1 81.4 75.6 3.3 0.25

3D-PDA (N = 155)Presence of endometrial flow (a) 0.596 (0.54–0.65) 53.3 91.9 27.2 1.3 0.30No dominant vessel (b) 0.739 (0.67–0.81) 75.5 66.1 81.7 3.6 0.41Vascularity, but no single or double dominant vessel (c) 0.734 (0.66–0.81) 74.8 66.1 80.6 3.4 0.42Multifocal vessels (d) 0.688 (0.60–0.77) 69.0 68.4 69.2 2.2 0.46Multiple vessels (e) 0.769 (0.70–0.84) 80.0 61.3 92.5 8.1 0.41Large endometrial vessels (f) 0.704 (0.63–0.78) 70.3 71.0 69.9 2.4 0.42Areas with densely packed or color-splash vessels (g) 0.710 (0.64–0.78) 75.5 48.4 93.6 7.5 0.55Doppler score (c,e,f,g) 0.814 (0.74–0.88) 81.3 ≥ 2 67.7 90.3 7.0 0.36Vascularization index 0.709 (0.62–0.79) 68.4 3.67 69.4 67.7 2.2 0.45Vascularization flow index 0.724 (0.64–0.81) 67.7 0.87 69.4 66.7 2.1 0.46Flow index 0.683 (0.60–0.77) 65.8 24.8 64.5 66.7 1.9 0.53

Multivariate logistic regression and REC scoreModel 1: Age, BMI, ET 0.828 (0.77–0.89) 76.3 0.359 76.8 76.0 3.20 0.31Model 2: Age, BMI, ET, interrupted endomyometrial

junction0.861 (0.80–0.92) 80.5 0.294 81.2 80.0 4.06 0.24

Model 3: BMI, ET, interrupted endomyometrial junction,Doppler score

0.879 (0.82–0.94) 82.6 0.305 83.9 81.7 4.59 0.20

Model 4: BMI, interrupted endomyometrial junction,Doppler score, irregular endometrial surface at 3D-GIS

0.908 (0.85–0.96) 87.6 0.459 85.3 89.3 7.96 0.17

REC score 3D-PDA (BMI ≥ 30, ET ≥ 10 mm, ET ≥ 15 mm,interrupted endomyometrial junction, Doppler score)

0.882 (0.82–0.94) 83.5 ≥ 3 86.9 81.0 4.56 0.16

REC score 3D-GIS (BMI ≥ 30, ET ≥ 10 mm, ET ≥ 15 mm,interrupted endomyometrial junction, Doppler score,irregular surface at 3D-GIS)

0.894 (0.84–0.95) 86.2 ≥ 4 85.3 86.9 6.51 0.17

Doppler score (c,e,f,g): vascularity, but no single/double dominant vessels (+1) + multiple vessels (+1) + large vessels (+1) + densely packedor color-splash vessels (+1). REC-score system is presented in Figure 1. Comparison of areas under the receiver–operating characteristicscurves (AUC): Model 1 vs Model 3 vs Model 4 (n = 145, P = 0.02); Model 1 vs REC score 3D-TVS vs REC score 3D-GIS (n = 145,P = 0.01). BMI, body mass index; ET, endometrial thickness; EV, endometrial volume; LR+, positive likelihood ratio; LR–, negativelikelihood ratio; REC score, risk of endometrial cancer score; Sens., sensitivity; Spec., specificity.

of SEP at the first TVS visit is the required presence ofan experienced investigator, which may not be a realisticpossibility in general practice at the present time. Thus, anapproach in which investigators perform real-time evalu-ations and acquire 3D volumes and/or 2D videoclips mayallow for secondary evaluation by experienced investiga-tors at image conferences. Thereby, a limited group ofindefinable evaluations may be identified that requireadditional imaging, such as those with discrepanciesbetween 2D and 3D evaluations or images of poor quality.

At first-line ultrasound, this would split patients intofour groups: 1) those with a thin endometrium, in whom

endometrial cancer is unlikely; 2) a second, high-riskgroup, in whom cancer is very likely; these patients shouldundergo fast-track evaluation; 3) a third, intermediategroup in whom benign pathology is likely; and 4) a fourthgroup in whom additional imaging is needed. It is muchsimpler and cheaper to evaluate images remotely than tosend the patient to a specialized unit. This approachis already used for magnetic resonance imaging; forultrasound, such a system could be time- and cost-efficient,when patients are grouped as suggested.

Prior to this study, association between endometrialcancer and multiple vessel pattern10 and irregular


742 Dueholm et al.

branching8 has been described, but how best to correlatedifferent vascular pattern findings to malignancy remaineduncertain. In our study, at SEP of 2D- and 3D-PDA, onlythe presence-of-multiple-vessels pattern was interpreted asmalignancy. Thus, as supported by others12, TVS based onSEP supplemented with 2D- or 3D-PDA did not improvethe diagnostic efficiency for malignancy.

The diagnostic efficiency of the different endometrialpatterns at 3D imaging was comparable to that reportedin other studies6–8,10. Vascular index parameters were notmore efficient than was the Doppler score in this study.Other authors have not compared the vascular indicesto a vascular pattern score system. The vascular indiceshad high diagnostic efficiency in some studies9,11,18,19, butnot in others12, and, while the indices had the advantageof low observer variation9,18,19, there are shortcomings:their measurement is time-consuming, there is a crucialdependence on image settings and a lack of reproducibilitybetween different categories of ultrasound machines20,21.However, we found that combining the different vascularpatterns into a total vascular pattern score (Doppler score)was more efficient at 3D-PDA than was measurement ofthe vascular indices.

The REC-score system had the highest diagnosticefficiency at 3D evaluation. This system improved thediagnostic efficiency by implementing the Doppler scoreand adding BMI, endometrial thickness, interruptedendomyometrial junction and irregular surface at GIS. Inpatients with increased endometrial thickness and limitedvascularity, the surface at GIS is an important parameterin the REC-score system. This system also had the highestefficiency in a prior 2D analysis14.

A clear strength of the present study was the use of areliable reference standard in all patients. Shortcomingsare combination in the study group of unselected patientswith selected patients referred from other hospitals.Another possible weakness could have been recall bias.We attempted to eliminate this by having gaps ofseveral months between the four rounds of ultrasoundanalyses, analyzing patients in a different order in eachround and blinding observers to the identity of patients.The investigator was also blinded to the pathologicaldiagnosis during all evaluations of 2D videoclips and3D volumes. This design excluded continuous learningduring evaluations, which might have increased thediagnostic performance. The quality of videoclips and 3Dvolumes was clearly a problem. Videoclips were acquiredas sweeps, and several 3D volumes had low quality,especially at GIS. We used gel instead of saline infusionto eliminate motion artifacts due to the constant spillof saline. Although we have several years of experiencewith SCSH, we experienced a gradual improvement inthe quality of the 3D-GIS examinations; for example,the catheter was removed before 3D volume acquisitionduring the later examinations, which reduced the acousticnoise due to the catheter shadow and hence improved theimage quality.

Vascular findings are dependent on tumor charac-teristics22,23 and might also depend on the sample of

cancer cases; they therefore must be evaluated in largerstudies. We did not evaluate color scores, which couldhave improved the evaluation9. Moreover, it might havebeen advantageous to add Doppler analysis to GIS andto evaluate 3D volumes and 2D videoclips concomitantlyrather than in two separate stages; this approach shouldbe evaluated in future studies.

In conclusion, SEP offline, from stored 2D videoclipsand 3D volumes at GIS, does not possess the samediagnostic efficiency as does real-time evaluation at2D-GIS, but such offline analysis may be considered inthose patients in whom the stored images and volumes areof sufficient quality. As for 2D ultrasound, the efficiency of3D ultrasound may be improved by implementation of theREC-score system, but addition of vascular indices andcalculation of EV apparently have no benefit. At first-lineultrasound examination, endometrial pattern evaluationwould split patients into four groups: 1) one group with athin endometrium in whom endometrial cancer is unlikely;2) a second, high-risk group in whom cancer is very likely;these patients should undergo fast-track evaluation; 3) athird, intermediate group in whom benign pathology islikely; and 4) a fourth group in whom additional imagingis needed. The optimal image modality appears to bereal-time 2D-GIS.

ACKNOWLEDGMENT

This study was supported with grants from the DanishCancer Society. We gratefully acknowledge the help ofElisabeth Melin and the staff at the day surgery unitfor conducting the blinding of observers, and CharlotteMøller for obtaining 2D videoclips and 3D volumes.

REFERENCES

1. Clark TJ, Barton PM, Coomarasamy A, Gupta JK, Khan KS. Investigatingpostmenopausal bleeding for endometrial cancer: cost-effectiveness of initialdiagnostic strategies. BJOG 2006; 113: 502–510.

2. Dijkhuizen FP, Mol BWJ, Brolmann HA, Heintz AP. Cost-effectiveness of the useof transvaginal sonography in the evaluation of postmenopausal bleeding. Maturitas2003; 45: 275–282.

3. van HN, Breijer MC, Khan KS, Clark TJ, Burger MP, Mol BW, TimmermansA. Diagnostic evaluation of the endometrium in postmenopausal bleeding: anevidence-based approach. Maturitas 2011; 68: 155–164.

4. Timmermans A, Opmeer BC, Khan KS, Bachmann LM, Epstein E, Clark TJ, GuptaJK, Bakour SH, Van den Bosch T, van Doorn HC, Cameron ST, Giusa MG,Dessole S, Dijkhuizen FP, Ter RG. Endometrial thickness measurement for detectingendometrial cancer in women with postmenopausal bleeding: a systematic reviewand meta-analysis. Obstet Gynecol 2010; 116: 160–167.

5. Smith-Bindman R, Kerlikowske K, Feldstein VA, Subak L, Scheidler J, Segal M,Brand R, Grady D. Endovaginal ultrasound to exclude endometrial cancer and otherendometrial abnormalities. JAMA 1998; 280: 1510–1517.

6. Epstein E, Valentin L. Gray-scale ultrasound morphology in the presence or absenceof intrauterine fluid and vascularity as assessed by color Doppler for discriminationbetween benign and malignant endometrium in women with postmenopausalbleeding. Ultrasound Obstet Gynecol 2006; 28: 89–95.

7. Epstein E, Skoog L, Isberg PE, De Smet F, De Moor B, Olofsson PA, GudmundssonS, Valentin L. An algorithm including results of gray-scale and power Dopplerultrasound examination to predict endometrial malignancy in women withpostmenopausal bleeding. Ultrasound Obstet Gynecol 2002; 20: 370–376.

8. Opolskiene G, Sladkevicius P, Valentin L. Ultrasound assessment of endometrialmorphology and vascularity to predict endometrial malignancy in women withpostmenopausal bleeding and sonographic endometrial thickness > or = 4.5 mm.Ultrasound Obstet Gynecol 2007; 30: 332–340.

9. Opolskiene G, Sladkevicius P, Valentin L. Prediction of endometrial malignancy inwomen with postmenopausal bleeding and sonographic endometrial thickness ≥ 4.5mm. Ultrasound Obstet Gynecol 2011; 37:232–240.



10. Alcazar JL, Castillo G, Minguez JA, Galan MJ. Endometrial blood flow mappingusing transvaginal power Doppler sonography in women with postmenopausalbleeding and thickened endometrium. Ultrasound Obstet Gynecol 2003; 21:583–588.

11. Alcazar JL, Galvan R. Three-dimensional power Doppler ultrasound scanning forthe prediction of endometrial cancer in women with postmenopausal bleeding andthickened endometrium. Am J Obstet Gynecol 2009; 200: 44–46.

12. Opolskiene G, Sladkevicius P, Jokubkiene L, Valentin L. Three-dimensionalultrasound imaging for discrimination between benign and malignant endometriumin women with postmenopausal bleeding and sonographic endometrial thickness ofat least 4.5 mm. Ultrasound Obstet Gynecol 2010; 35: 94–102.

13. Alcazar JL, Iturra A, Sedda F, Auba M, Ajossa S, Guerriero S, Jurado M.Three-dimensional volume off-line analysis as compared to live ultrasound forassessing adnexal masses. Eur J Obstet Gynecol Reprod Biol 2012; 161: 92–95.

14. Dueholm M, Møller C, Rydbjerg S, Hansen ES, Ørtoft G. An ultrasound algorithmfor identification of endometrial cancer. Ultrasound Obstet Gynecol 2014; 43:557–568.

15. Leone FP, Timmerman D, Bourne T, Valentin L, Epstein E, Goldstein SR, MarretH, Parsons AK, Gull B, Istre O, Sepulveda W, Ferrazzi E, Van den Bosch T.Terms, definitions and measurements to describe the sonographic features of theendometrium and intrauterine lesions: a consensus opinion from the InternationalEndometrial Tumor Analysis (IETA) group. Ultrasound Obstet Gynecol 2010; 35:103–112.

16. Raine-Fenning NJ, Campbell BK, Clewes JS, Kendall NR, Johnson IR. The reliabilityof virtual organ computer-aided analysis (VOCAL) for the semiquantification of

ovarian, endometrial and subendometrial perfusion. Ultrasound Obstet Gynecol2003; 22: 633–639.

17. Alcazar JL, Galvan R. Three-dimensional power Doppler ultrasound scanning forthe prediction of endometrial cancer in women with postmenopausal bleeding andthickened endometrium. Am J Obstet Gynecol 2009; 200: 44–46.

18. Alcazar JL, Ajossa S, Floris S, Bargellini R, Gerada M, Guerriero S. Reproducibilityof endometrial vascular patterns in endometrial disease as assessed by transvaginalpower Doppler sonography in women with postmenopausal bleeding. J UltrasoundMed 2006; 25: 159–163.

19. de Kroon CD, Hiemstra E, Trimbos JB, Jansen FW. Power Doppler area in thediagnosis of endometrial cancer. Int J Gynecol Cancer 2010; 20: 1160–1165.

20. Alcazar JL. Three-dimensional power Doppler derived vascular indices: what arewe measuring and how are we doing it? Ultrasound Obstet Gynecol 2008; 32:485–487.

21. Raine-Fenning NJ, Nordin NM, Ramnarine KV, Campbell BK, Clewes JS, PerkinsA, Johnson IR. Evaluation of the effect of machine settings on quantitativethree-dimensional power Doppler angiography: an in-vitro flow phantom experiment.Ultrasound Obstet Gynecol 2008; 32: 551–559.

22. Epstein E, Van HC, Mascilini F, Masback A, Kannisto P, Ameye L, FischerovaD, Zannoni G, Vellone V, Timmerman D, Testa AC. Gray-scale and color Dopplerultrasound characteristics of endometrial cancer in relation to stage, grade and tumorsize. Ultrasound Obstet Gynecol 2011; 38: 586–593.

23. Saarelainen SK, Vuento MH, Kirkinen P, Maenpaa JU. Preoperative assessmentof endometrial carcinoma by three-dimensional power Doppler angiography.Ultrasound Obstet Gynecol 2012; 39: 466–472.


Documents

Two- and three-dimensional transvaginal ultrasound with power … · Transvaginal ultrasound (TVS) is the most cost-effective ﬁrst-line test in the diagnostic work-up of women with