computed tomography for evaluation of ovarian masses.

ORIGINAL ARTICLE

Assessment of Combination of Contrast-Enhanced Magnetic Resonance Imaging and Positron Emission Tomography/Computed Tomography for Evaluation of Ovarian Masses Takahiro Tsuboyama, MD, PhD,* Mitsuaki Tatsumi, MD, PhD,* Hiromitsu Onishi, MD, PhD,* Atsushi Nakamoto, MD, PhD,* Tonsok Kim, MD, PhD,* Masatoshi Hori, MD, PhD,* Makoto Sakane, MD,* Yumiko Hori, MD,Þ Eiichi Morii, MD, PhD,Þ Jun Hatazawa, MD, PhD,þ and Noriyuki Tomiyama, MD, PhD* Objectives: The objectives of this study were to correlate fluorodeoxyglucose uptake in ovarian masses on positron emission tomography/computed tomography (PET/CT) with pathological grades of malignancy and subtypes and to determine the appropriate approach for combining PET/CT and contrast-enhanced magnetic resonance imaging (CE-MRI) to characterize ovarian masses. Materials and Methods: A retrospective study was conducted including 127 patients who underwent surgical resection of an ovarian mass (30 benign, 31 borderline, 66 malignant). Maximum standardized uptake values (SUVmax) obtained with PET/CT were compared between pathological grades of malignancy and subtypes. Two radiologists each independently conducted a blind evaluation of CE-MRI for all lesions and classified them by the grade of malignancy as determinate (benign, borderline, or malignant) or indeterminate and by subtype as mucinous or nonmucinous. The appropriate approach for combining CE-MRI and PET/CT was determined by comparing the combined diagnostic ability with that of CE-MRI alone. Results: The SUVmax of malignant tumors was significantly higher than that of benign and borderline lesions (mean, 7.8, 1.7, 2.4; P G 0.05). Among malignant tumors, SUVmax was significantly lower in mucinous adenocarcinomas compared with nonmucinous malignant tumors (mean, 3.3, 8.4; P G 0.05) and lower in clear cell adenocarcinomas compared with other subtypes of nonmucinous malignant tumors (mean, 6.0, 9.4; P G 0.05). The SUVmax cutoff that best differentiated malignant lesions from benign/borderline lesions was 2.4 for mucinous and 4.0 for nonmucinous tumors. These cutoffs correctly classified lesions as malignant or not in 88.2% of cases (112/127). When PET/CT was combined with CE-MRI, the readers correctly classified 85% (34/40) and 86.5% (32/37) of indeterminate lesions on CE-MRI. However, PET/CT was not useful for classifying determinate lesions on CE-MRI, particularly because PET/CT correctly classified only 70.1% (12/17) of clear cell adenocarcinomas, whereas CE-MRI alone correctly classified 94.1% (1617). Thus, compared with CE-MRI alone, the diagnostic accuracy of CE-MRI + PET/CT when PET/CT was added only for indeterminate lesions on CE-MRI was significantly higher for both readers for differentiating between benign and borderline/malignant (P G 0.05), as well as between benign/borderline and malignant (P G 0.01). Conclusion: Fluorodeoxyglucose uptake in ovarian masses correlates with pathological subtypes as well as the grade of malignancy. Furthermore, the combination of CE-MRI and PET/CT is a highly accurate method for

Received for publication November 15, 2013; and accepted for publication, after revision, January 24, 2014. From the Departments of *Radiology, †Pathology, and ‡Nuclear Medicine, Osaka University Graduate School of Medicine, Osaka, Japan. Conflicts of interest and sources of funding: none declared. Reprints: Takahiro Tsuboyama, MD, PhD, Department of Radiology, Osaka University Graduate School of Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]. Copyright * 2014 by Lippincott Williams & Wilkins ISSN: 0020-9996/14/4908Y0524

524

www.investigativeradiology.com

characterizing ovarian masses because PET/CT can be used as a complement to classify indeterminate lesions as malignant or not based on appropriate cutoff SUVmax for mucinous and nonmucinous tumors. Key Words: Ovarian tumors, Magnetic resonance imaging, Positron emission tomography/computed tomography (Invest Radiol 2014;49: 524Y531)

P

reoperative differential diagnosis of ovarian masses as malignant, borderline, or benign is important because the standard treatment for the first two has been radical surgery.1 Recently, however, it has become acceptable to perform conservative fertility-sparing surgery and to omit lymphadenectomy for borderline tumors because of their favorable prognosis.2 Therefore, it has become increasingly important to differentiate borderline tumors from malignant tumors before surgery. Contrast-enhanced magnetic resonance imaging (CE-MRI) is currently the established method for characterizing ovarian masses, and its usefulness for differentiation of borderline/malignant tumors from benign lesions has been well documented.3Y5 However, it is still challenging to differentiate some benign tumors such as sex cordstromal tumors and germ cell tumors, which present as predominantly solid or solid and cystic masses.3,6 Furthermore, although unique imaging features of serous borderline tumors (SBTs) and Mu¨llerian mucinous borderline tumors (MMBTs) have recently been described,7Y9 it is reportedly difficult to differentiate borderline tumors from malignant tumors because both have similar papillary projections or are multilocular.10Y12 Positron emission tomography and computed tomography (PET/CT) using 18F-fluorodeoxyglucose (FDG) is now accepted as effective for evaluating various kinds of malignancies. However, the role of PET/CT in characterizing ovarian masses has not yet been established.13Y22 One reason for this may be that FDG uptake in borderline tumors is low compared with malignant tumors,18Y20 causing false-negative results when borderline tumors are included in the malignant group.14Y19 Another reason may be that FDG uptake is low even in some malignant tumors with poorly defined characteristics. It has been demonstrated that mucinous adenocarcinomas in many organs show low FDG uptake due to hypocellularity.23Y25 However, FDG uptake in ovarian tumor subtypes has not yet been investigated. The rapid expansion of clinical uses of PET/CT has presented more opportunities to evaluate ovarian masses with both CE-MRI and PET/CT. However, there is no established method for combining CE-MRI with PET/CT. Therefore, the objectives of this study were to correlate FDG uptake in ovarian masses on PET/CT with pathological grades of malignancy and subtypes and to determine the appropriate approach for combining PET/CT and CE-MRI to characterize ovarian masses. Investigative Radiology

&

Volume 49, Number 8, August 2014

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Investigative Radiology

& Volume 49, Number 8, August 2014

were 15 clear cell, 9 endometrioid, 7 mucinous, 3 serous tumors, and 1 mixed epithelial tumor among the stage I tumors.

MATERIALS AND METHODS Patients The review board of our institution approved this retrospective study and waived the requirement to obtain informed consent. Gynecological, pathological, and radiological records of 127 patients aged 20 to 86 years (mean, 53.4 years), including 56 premenopausal patients, who underwent PET/CT and CE-MRI before surgical resection of an ovarian mass at our institution between January 2008 and March 2013 were reviewed. The initial enrollment was 142 patients, but 15 patients who underwent neoadjuvant chemotherapy before surgical resection were excluded. Two patients underwent ovarian cystectomy, 32 patients underwent unilateral or bilateral salpingo-oophorectomy, and 93 patients underwent bilateral salpingo-oophorectomy with hysterectomy. Lymphadenectomy was performed on 60 patients and omentectomy was performed on 93 patients, based on the results of intraoperative inspection and frozen section pathology.

Ovarian Lesions Pathological records, which were reported by experienced pathologists using the 2003 World Health Organization classification,26 classified 30 of the ovarian masses as benign, 31 as borderline, and 66 as malignant (Table 1). Primary borderline or malignant tumors were staged according to the International Federation of Gynecology and Obstetrics (FIGO) system. Sixty-five tumors were stage I and 30 tumors were stage II-IV. All borderline tumors were stage I except for 1 SBT that was stage II. By subtype, there

TABLE 1. Pathological Diagnosis of Ovarian Masses Pathological Diagnosis Benign (n = 30) Surface epithelial-stromal tumors

Sex cord-stromal tumors Germ cell tumors Others

Borderline (n = 31) Surface epithelial-stromal tumors

Sex cord-stromal tumors Malignant (n = 66) Surface epithelial-stromal tumors

Germ cell tumors

n Mucinous Serous Mixed Fibrothecoma Sclerosing stromal tumor Mature teratoma Struma ovarii Endometriotic cyst Simple cyst Cavernous hemangioma

6 5 1 9 1 2 1 3 1 1

Mucinous Serous Clear cell Endometrioid Brenner Granulosa cell tumor

18 9 1 1 1 1

Mucinous Serous Clear cell Endometrioid Mixed Immature teratoma Squamous cell carcinoma

8 18 17 17 2 1 1 2

Secondary tumors

* 2014 Lippincott Williams & Wilkins

Combination of MR and PET/CT for Ovarian Masses

Magnetic Resonance and PET/CT Examination For magnetic resonance (MR) examinations, a 1.5-T MR system (Signa Excite HD 1.5T; GE Healthcare, Milwaukee, WI) was used for 74 patients and a 3-T system (Signa HD 3.0T; GE Healthcare; or Achieva 3.0T X; Philips Medical Systems, Best, the Netherlands) was used for 53 patients. An 8- or 32-channel torso array coil system was used. The sequences for review and their acquisition parameters are shown in Table 2. Contrast-enhanced MRI was obtained with fat-saturated T1-weighted imaging after intravenous injection of 0.1 mmol/kg of gadoteridol (ProHance; Eisai, Tokyo, Japan). An integrated scanner (Gemini GXL; Philips, Cleveland, OH) was used to perform PET/CT imaging. Whole-body images were acquired approximately 60 minutes after intravenous injection of 0.10 mCi/kg of 18F-FDG. The acquisition parameters are listed in Table 2. The median interval between CE-MRI and PET/CT was 13 days (range, 1Y49 days), and that between the first imaging examination and the surgery was 42.5 days (range, 4Y110 days).

CE-MRI Evaluation Imaging features of ovarian and extraovarian diseases on CE-MRI were assessed by 1 radiologist with 11 years of experience in gynecological MRI and compared among benign, borderline, and malignant lesions. Solid components were defined as tissue that showed enhancement and included solid portions, papillary projections, thickened septa larger than 3 mm, and aggregations of fine loculi. Peritoneal diseases were recorded when peritoneal enhancement or peritoneal masses were identified. Swollen lymph nodes were recorded when they were larger than 1 cm in the short axis. Two radiologists with 15 and 10 years of experience in gynecological MRI retrospectively and independently characterized lesions using CE-MRI without knowledge of the pathological and PET/CT findings. They classified all lesions by the grade of malignancy as determinate (benign, borderline, or malignant) or indeterminate (benign/malignant, borderline/malignant, or benign/borderline/ malignant) and by subtype as mucinous or nonmucinous. Diagnostic criteria for the grade of malignancy and subtypes, which were based on previous studies,3Y12 are summarized in Table 3. Both mucinous and nonmucinous tumors can appear as purely cystic masses or as masses with solid components showing high intense papillary architecture on T2-weighted images, and thus, such masses were classified by subtype as ‘‘any cystic lesions’’ and ‘‘SBT/MMBT,’’ respectively. The signal intensity of the solid components on T2-weighted images was considered low when lower than that of the outer myometrium, intermediate when equal to or greater than that of the outer myometrium, and high when similar to that of serous fluid. The sensitivity, specificity, and accuracy for differentiation between borderline/ malignant and benign lesions, as well as for differentiation between malignant and benign/borderline lesions, were calculated and are shown with 95% confidence intervals.

PET/CT Evaluation The standardized uptake value (SUV), which is the decaycorrected tissue activity divided by the injected dose per patient body weight, was used to semiquantitatively evaluate PET/CT. A radiologist with 18 years of experience in nuclear medicine drew regions of interest manually on consecutive axial PET images around the entire ovarian lesion while avoiding adjacent structures by referring to fused PET/CT images.27,28 The maximum pixel value of the SUV within each lesion (SUVmax) was then measured. www.investigativeradiology.com


525


Tsuboyama et al


TABLE 2. MR and PET/CT Examination Protocols MR Examination Pulse sequences Acquisition planes Repetition time, ms Echo time, ms Flip angle, Matrix Field of view, mm Slice thickness, mm

T2WI

T1WI Without Fat Saturation

T1WI With Fat Saturation

2D FSE Axial and sagittal 4000Y5100/4500Y8000 85/80Y100 90 512 256/512Y560 256Y348 240Y280 5Y8

3D GE Axial 6.6/4.1Y5.7 4.2/2.3Y2.4 10Y12 320 192/320 192Y203 280Y320 3Y4

3D GE Axial and sagittal 4.7/3.6Y3.9 2.3/1.7Y1.9 10Y12 320 192/240Y320 192Y242 280Y320 3Y4

PET/CT Examination Scan method Acquisition Reconstruction Slice thickness/interval, mm kVp/mAs

PET

16-Slice CT

3D emission scan 2 min scan/bed position 11 positions OSEM 4/4

Helical scan From the apex of the lungs to the pelvis during 1 breath hold Filter back projection 5/4 120/50

Data in the MR examination section are values for the 1.5/3-T system when values differ between systems. MR indicates magnetic resonance; PET, positron emission tomography; CT, computed tomography; T2WI, T2-weighted imaging; T1WI, T1-weighted imaging; 2D FSE, 2-dimensional fast spin echo; 3D GE, 3-dimensional gradient echo; OSEM, ordered subsets expectation maximization.

Differences in SUVmax between pathological grades of malignancy, FIGO stages, and pathological subtypes were assessed. Respective cutoff values for SUVmax that best differentiated

between benign/borderline and malignant tumors were then determined for mucinous and nonmucinous tumors.

TABLE 3. Diagnostic Criteria for Grade of Malignancy and Subtypes on Contrast-Enhanced Magnetic Resonance Imaging

To determine the appropriate approach for combining CEMRI + PET/CT, the usefulness of adding PET/CT for characterizing lesions classified as determinate or indeterminate on CE-MRI was evaluated. For indeterminate lesions on CE-MRI, PET/CT was used to differentiate malignant from benign/borderline lesions based on the appropriate cutoff SUVmax for the subtype classification, that is, mucinous or nonmucinous. For determinate lesions on CE-MRI, the appropriate SUVmax was used and the diagnosis based on FDG uptake was compared with the diagnosis made by the 2 readers on CE-MRI. According to the diagnostic criteria for subtypes on CE-MRI (Table 3), the cutoff SUVmax for nonmucinous tumors was used for lesions diagnosed as benign (noncystic) and malignant from CE-MRI, and the cutoff for mucinous tumors was used for lesions diagnosed as benign (purely cystic) and borderline.

Grade

Lesion Subtype

Determinate lesions Benign (B) Any cystic lesions Nonmucinous

Borderline (BL)

Serous or Mu¨llerian mucinous tumors

Malignant (M)

Nonmucinous

Indeterminate lesions BL/M Mucinous (intestinal type)

B//M or B/BL/M

526

Nonmucinous

Imaging Findings Purely cystic masses Unilateral masses with solid components showing homogeneous low intensity on T2-weighted imaging (T2WI) and weak enhancement. Unilateral or bilateral masses with solid components showing high intense papillary architecture with or without low intense internal branching on T2WI. Unilateral or bilateral masses with papillary projections or irregular solid portions showing intermediate intensity on T2WI. Unilateral smooth-walled cystic masses with numerous loculi. Presence of focal solid components such as thickened septa or aggregations of fine loculi. Solid masses with smooth outlines that are indeterminate as fibroma because of bilaterality, intermediate intensity on T2WI, or cystic degeneration. Also, masses with focal fatty tissue that are indeterminate as mature teratoma due to the presence of noninvasive solid components.


Lesion Characterization With Combined CE-MRI and PET/CT

TABLE 4. Comparison of Imaging Features on Contrast-Enhanced Magnetic Resonance Imaging Characteristics

Benign (n = 30)

Borderline (n = 31)

Malignant (n = 66)

P

Size, median 8.0 (6.0Y12.0) 13.4 (6.3Y18.0) 9.7 (7.6Y14.0) 0.05 (interquartile range), cm Bilaterality 4 (13.3) 2 (6.5) 14 (21.2) 0.16 Solid and cystic components Purely cystic 1 (3.3) 1 (3.2) 0 0.33 Mixed 22 (73.3) 28 (90.3) 58 (87.9) 0.11 Solid 7 (23.3) 2 (6.5) 8 (12.1) 0.14 Ascites 7 (23.3) 11 (35.5) 29 (43.9) 0.15 Peritoneal disease 0 3 (9.7) 21 (31.8) G0.01 Lymph node 0 0 6 (9.1) 0.05 swelling Unless otherwise indicated, data show n (%) of patients.





a significant difference. Agreement between the 2 readers regarding lesion characterization was determined with the J statistic as poor (J G 0.40), moderate (0.40 e J G 0.70), or excellent (J Q 0.70).29 MedCalc 12.5.0.0 (MedCalc Software, Ostend, Belgium) was used for all statistical analyses.

TABLE 5. Comparison of Diagnostic Ability Between Contrast-Enhanced MR and Combined Contrast-Enhanced MR and PET/CT Parameter

Contrast-Enhanced MRI

Contrast-Enhanced MRI + PET/CT

Reader 1 Differentiation of borderline/malignant tumors from benign lesions Sensitivity 92.8 (90/97) [87.6Y97.9] 96.9 (94/97) [93.5Y100] Specificity 16.7 (5/30) [3.3Y30.0] 40.0 (12/30) [22.5Y57.5] Accuracy 74.8 (95/127) [67.3Y82.4] 83.5 (106/127) [77.0Y89.9] Differentiation of malignant tumors from benign/borderline lesions Sensitivity 81.8 (54/66) [72.5Y91.1] 97.0 (64/66) [92.8Y100] Specificity 44.3 (27/61) [31.8Y56.7] 83.6 (51/61) [74.3Y92.9] Accuracy 63.8 (81/127) [55.4Y72.1] 90.6 (115/127) [85.5Y95.6] Reader 2 Differentiation of borderline/malignant tumors from benign lesions Sensitivity 96.9 (94/97) [93.5Y100] 97.9 (95/97) [95.1Y100] Specificity 20.0 (6/30) [5.7Y34.3] 36.7 (11/30) [19.4Y53.9] Accuracy 78.7 (100/127) [71.6Y85.9] 83.5 (106/127) [77.0Y89.9] Differentiation of malignant tumors from benign/borderline lesions Sensitivity 84.8 (56/66) [76.2Y93.5] 97.0 (64/66) [92.8Y100] Specificity 36.1 (22/61) [24.0Y48.1] 75.4 (46/61) [64.6Y86.2] Accuracy 61.4 (78/127) [53.0Y69.9] 86.6 (110/127) [80.7Y92.5]

P

RESULTS 0.13 0.02 G0.01 G0.01 G0.01 G0.01

1.00 0.06 0.03 G0.01 G0.01 G0.01

Data are shown as % (numbers used to calculate percentages) [95% confidence interval]. MR indicates magnetic resonance; MRI, MR imaging; PET/CT, positron emission tomography/computed tomography.

The diagnostic ability of the combination of CE-MRI + PET/CT when appropriate as determined above to predict the pathological grade of malignancy was compared with that of CE-MRI alone.

Statistical Analysis


The W2 test was used to compare categorical variables. Onefactor analysis of variance was used for multigroup comparisons of continuous variables, and the Student-Newman-Keuls test was used for all pairwise comparisons when there was a significant difference among groups.29 Receiver operating characteristic analysis was used to identify the cutoff values for SUVmax. The McNemar test was used to compare the sensitivity, specificity, and accuracy of CE-MRI and CE-MRI + PET/CT. A P value of 0.05 was considered to indicate

Lesion Characterization With CE-MRI There were no significant differences in lesion size, bilaterality, solid and cystic components, ascites, and lymphadenopathies among benign, borderline, and malignant lesions (Table 4). The diagnostic capability of CE-MRI is summarized for each reader in Table 5. Readers 1 and 2 classified 40 (31.5%) and 37 (29.1%) lesions, respectively, as indeterminate (Figs. 1Y3). Mucinous and sex-cord stromal tumors accounted for 72.5% and 75.7% of these indeterminate lesions according to readers 1 and 2, respectively (Figs. 1 and 2). When characterizing indeterminate lesions by subtype, each reader incorrectly classified 6 nonmucinous lesions as mucinous (reader 1, 3 serous tumors, 1 fibrothecoma [Fig. 2], 1 Brenner, and 1 clear cell tumor; reader 2, 3 endometrioid, 2 serous, and 1 Brenner tumor).

Correlation of FDG Uptake on PET/CT With Pathological Findings There was a significant difference in SUVmax among benign, borderline, and malignant tumors (P G 0.01) (Fig. 5A). Malignant tumors showed significantly higher SUVmax than both benign and borderline lesions did (mean, 7.8, 1.7, 2.4; P G 0.05), whereas there was no significant difference between benign and borderline lesions (Figs. 1Y4). Fluorodeoxyglucose uptake was also related to the FIGO stage. There was a significant difference in SUVmax among benign/ borderline tumors, stage I malignant tumors, and stage II-IV malignant tumors (P G 0.01) (Fig. 5B). Malignant tumors at stage I already showed significantly higher FDG uptake than benign/borderline lesions did (mean, 6.5, 2.0; P G 0.05), although they showed significantly lower uptake than stage II-IV tumors did (mean, 9.2; P G 0.05). There were also significant differences in SUVmax between pathological subtypes (P G 0.01) (Fig. 5C). Significantly lower SUVmax was detected in mucinous adenocarcinomas compared with nonmucinous malignant tumors (mean, 3.3, 8.4; P G 0.05) and also in clear cell adenocarcinomas compared with other subtypes of nonmucinous malignant tumors (mean, 6.0, 9.4; P G 0.05) (Fig. 6). Among malignant tumors, mucinous adenocarcinomas showed comparable SUVmax with benign/borderline lesions. Receiver operating characteristic analysis showed that the cutoff SUVmax that best differentiated malignant from benign/borderline

FIGURE 1. Mucinous borderline tumor in a 54-year-old woman. A, T2-weighted MR image shows a multilocular cystic mass with irregular thickened septa (arrow). The mass was classified as an indeterminate (borderline or malignant) mucinous tumor by both readers. B, PET/CT shows true-negative FDG uptake (SUVmax = 1.5) on the thickened septa (arrow) with the cutoff SUVmax of 2.4 for mucinous tumors. Thus, adding PET/CT to MR resulted in a correct diagnosis of the mass as borderline. * 2014 Lippincott Williams & Wilkins



527


Tsuboyama et al


FIGURE 2. Fibrothecoma with marked cystic change in a 57-year-old woman. A, T2-weighted MR image shows a solid and cystic mass, with the solid component showing low to intermediate intensity (arrow). The mass was classified as indeterminate by both readers but as mucinous (borderline or malignant) by one and as nonmucinous (benign or malignant) by the other. B, PET/CT shows moderate SUVmax of 2.6 for the solid component (arrow), which would result in a true-negative diagnosis if the cutoff SUVmax of 4.0 for nonmucinous tumors were used but a false-positive if the cutoff of 2.4 for mucinous tumors were used. Classification as a mucinous or nonmucinous tumor with MR therefore affected the diagnosis.

lesions was 2.4 for mucinous and 4.0 for nonmucinous tumors. Using these cutoff SUVmax values, 88.2% (112/127) of lesions were correctly classified as malignant or not (Figs. 1Y4, Table 6). False-positive uptake was detected in 8 borderline tumors (6 mucinous, 1 serous, and 1 Brenner tumor), and false-negative uptake, in 7 malignant tumors (1 mucinous tumor, 5 clear cell tumors, and 1 immature teratoma).

Lesion Characterization With Combination of CE-MRI and PET/CT When PET/CT was added to CE-MRI for indeterminate lesions using the appropriate cutoff SUVmax of 2.4 (for mucinous) or 4.0 (for nonmucinous) according to the CE-MRI classification, reader 1 correctly classified 34 of 40 (85%) lesions and reader 2 correctly classified 32 of 37 (86.5%) lesions as malignant or not. Of the 6 nonmucinous tumors incorrectly classified as mucinous on CE-MRI, 1 fibrothecoma with an SUVmax of 2.6 was incorrectly diagnosed as malignant by reader 1 because a cutoff SUVmax of 2.4 rather than 4.0 was used (Fig. 2). When PET/CT was added to CE-MRI for determinate lesions, readers 1 and 2 correctly changed the diagnosis of 5 (2 benign,

3 borderline) and 11 (6 benign, 5 borderline) lesions, respectively, whereas reader 1 incorrectly changed the diagnosis of 10 lesions (5 borderline, 5 malignant), and reader 2, another 10 lesions (4 borderline, 6 malignant). Clear cell adenocarcinomas accounted for a considerable number of cases where the diagnosis was changed from malignant. These were correctly classified as malignant in 94.1% of cases (16/17) by CE-MRI alone but only 70.1% of cases (12/17) using PET/CT. The clear cell adenocarcinoma that both readers misclassified on CE-MRI also showed false-negative FDG uptake (SUVmax = 1.4, 1.8) on PET/CT, and thus, adding PET/CT did not assist either reader in correctly changing the diagnosis. The diagnostic capability of CE-MRI + PET/CT when PET/CT was added only for lesions classified as indeterminate by CE-MRI (Fig. 7) is summarized in Table 5. Adding PET/CT significantly improved the sensitivity for malignant tumors (P G 0.01 for both readers) but not for borderline/malignant tumors (reader 1, P = 0.13; reader 2, P = 1.00). Accuracy was significantly higher with CE-MRI + PET/CT than with CE-MRI alone for differentiation between benign and borderline/malignant lesions (reader 1, P G 0.01; reader 2, P = 0.03) and differentiation between benign/borderline and malignant lesions (P G 0.01 for both readers).

FIGURE 3. Stage I endometrioid adenocarcinoma in a 77-year-old woman. A, T2-weighted MR image shows a smoothly outlined solid mass with intermediate and focal low intensity (arrow). The mass was classified as an indeterminate nonmucinous tumor by both readers. B, PET/CT shows focal true-positive FDG uptake (SUVmax = 4.4) based on the cutoff SUVmax of 4.0 for nonmucinous tumors (arrow). Thus, adding PET/CT to MR resulted in a correct diagnosis of the mass as malignant. Notably, this mass was composed histopathologically mostly of borderline adenofibroma with a focal section composed of adenocarcinoma (not shown). Asterisk points to uterus. 528







FIGURE 4. Serous cystic borderline tumor in a 29-year-old woman. A, T2-weighted MR image shows a cystic ovarian mass with high intense papillary projections with hypointense internal branching (arrow). Although 1 reader correctly classified this mass as borderline, the other classified it as malignant. B, PET/CT shows true-negative SUVmax of 2.0 on the papillary projections. However, under the appropriate approach for combining MR + PET/CT, PET/CT should not be used to characterize lesions classified as determinate with MR. Asterisk points to uterus.

Interobserver agreement was moderate for CE-MRI (J = 0.68) and excellent for CE-MRI + PET/CT (J = 0.76).

DISCUSSION On the basis of our results, we propose the following 3-step approach as the appropriate method of combining CE-MRI and PET/CT for characterizing ovarian masses. First, CE-MRI is used to classify ovarian masses as determinate or indeterminate by the grade of malignancy. Second, indeterminate lesions are classified by subtype as mucinous or nonmucinous. Finally, PET/CT is used to classify lesions that were classified as indeterminate by CE-MRI as malignant or not using the appropriate cutoff SUVmax for the CE-MRI subtype classification. We found that this approach for adding PET/CT to CE-MRI improved the diagnostic accuracy and interobserver agreement for predicting the pathological grade of malignancy. We demonstrated that PET/CT is useful for differentiating between malignant and benign/borderline tumors, especially when the diagnosis with MR was indeterminate, because glucose metabolic activity appears to be low in benign/borderline lesions and high in malignant tumors.15Y17 However, PET/CT has been reported to be

less useful than MR for differentiating between borderline/malignant and benign tumors.14Y19 In our study, PET/CT was only minimally useful or even detrimental for lesions classified as determinate on CE-MRI. Therefore, PET/CT and MRI play complementary roles in characterizing ovarian masses, and it is very important to combine them appropriately. Some malignant tumors reportedly do not show high FDG uptake.14Y18,22 We found that mucinous adenocarcinomas showed significantly lower SUVmax than nonmucinous malignant tumors did, which is why we used a lower cutoff SUVmax of 2.4 for the former compared with 4.0 for the latter. Similarly, a low cutoff SUVmax of 2.5 or less is used for mucinous adenocarcinomas in many other organs.23Y25 Although a low cutoff SUVmax proved effective for avoiding false-negative results caused by mucinous adenocarcinomas, some mucinous borderline tumors caused falsepositive results instead because there was some overlap in FDG uptake between them. In addition, some nonmucinous tumors mimicked mucinous tumors owing to being multilocular, as has been previously reported,30 and such benign/borderline tumors can be incorrectly classified as malignant when a lower cutoff SUVmax is used.

FIGURE 5. Box plots of SUVmax of ovarian masses showing the median (horizontal line), the 75th (top of box) and 25th (bottom of box) percentiles, as well as the smallest and largest values (whiskers). Outside values that are larger than the upper quartile plus 1.5 times the interquartile range are displayed as separate points. A, Comparison of SUVmax for benign, borderline, and malignant lesions demonstrated a significant difference among the 3 groups (P G 0.01). Malignant tumors showed significantly higher SUVmax than both benign and borderline lesions did (*P G 0.05). B, Comparison of SUVmax by pathological grade of malignancy and stage demonstrated a significant difference among the 3 groups (P G 0.01). Stage I malignant tumors showed significantly higher FDG uptake than benign/borderline lesions did but significantly lower uptake than stage II-IV tumors did (*P G 0.05). C, Comparison of SUVmax by pathological grade of malignancy and subtype demonstrated a significant difference among the 4 groups (P G 0.01). Among malignant tumors, mucinous and clear cell adenocarcinomas showed significantly lower SUVmax, whereas the SUVmax of the former was equivalent to that of benign/borderline lesions (*P G 0.05). * 2014 Lippincott Williams & Wilkins



529


Tsuboyama et al


FIGURE 6. Stage I clear cell adenocarcinoma in a 46-year-old woman. A, T2-weighted MR image shows a cystic mass with intermediate intense papillary projections (arrow), with fluid inside the cyst showing high intensity on fat-suppressed T1-weighted image (not shown), suggesting malignant transformation of an endometriotic cyst. Both readers correctly classified it as malignant. B, PET/CT shows a false-negative SUVmax of 3.3 on the papillary projections based on the cutoff SUVmax of 4.0 for nonmucinous tumors. False-negative results can be avoided if the appropriate approach for combining MR and PET/CT, namely, not using PET/CT for determinate lesions, is used.

We unexpectedly found that some clear cell adenocarcinomas showed low SUVmax. Interestingly, clear cell renal cell carcinomas also show inconsistent FDG uptake, although the reason for this remains unknown.31 However, 1 possible reason for the variable FDG uptake could be the variable cellularity observed between tubulocystic, papillary, and solid morphologies.32 The MRI features of clear cell adenocarcinomas are well known,33 and thus, a determinate diagnosis as malignant made with CE-MRI should not be changed by low FDG uptake on PET/CT. Our findings may help to eliminate unnecessary timeconsuming and expensive PET/CT used only to characterize ovarian masses that have already been diagnosed as determinate by CE-MRI. Moreover, our study may also contribute to more accurate evaluation of ovarian lesions with newly developed integrated PET/MR.34 Although optimal image acquisition protocols for PET/MR are still being determined,35 its diagnostic performance is expected to be superior to that of PET/CT for the pelvic region, where MR provides better image quality than CT does.36 There were certain limitations to our study. First, the study population of each tumor subtype was small. Second, the retrospective design of this study made it impossible to avoid sampling bias. Our study included ovarian masses for which PET/CT was indicated

for preoperative staging, and consequently, ovarian masses that are generally identified as malignant on CE-MRI were included. Third, scanning protocols were not standardized across MR examinations using 1.5- and 3-T systems. Fourth, advanced MR techniques such as diffusion weighted imaging (DWI) were not used to evaluate MRI, despite the fact that many authors have reported that DWI is effective for characterizing ovarian masses as benign or malignant.37,38 Diffusion weighted imaging with apparent diffusion coefficient analysis was reported to correlate well with PET/CT in an evaluation of treatment response in patients with malignant lymphoma,39 but we believe that PET/CT is still important for differentiation between borderline and malignant ovarian lesions even when compared with DWI in light that Takeuchi et al37 reported 7 borderline tumors that each showed high intensity on DWI in the same way as malignant tumors. Proton MR spectroscopy is another advanced MR technique that appears promising for differentiating between malignant and benign ovarian tumors, but MR spectroscopy of borderline ovarian tumors has not yet been evaluated.40 Finally, although we determined cutoff SUVmax values retrospectively, they need to be verified prospectively. Furthermore, SUV is affected by many technical and

TABLE 6. FDG Uptake by Pathological Grade and Subtype SUVmax Pathology Benign (n = 30) Mucinous Nonmucinous Borderline (n = 31) Mucinous Nonmucinous Malignant (n = 66) Mucinous Nonmucinous Clear cell Others

G2.4

Q2.4, G4.0

Q4.0

25 (83.3) 6 19 20 (64.5) 12 8 4 (6.0) 1

5 (16.7) 0 5 6 (19.4) 3 3 9 (13.6) 6

0 0 0 5 (16.1) 3 2 53 (80.3) 1

3 0

2 1

12 40

Numbers in parentheses are percentages. FDG indicates fluorodeoxyglucose; SUVmax, maximum standardized uptake value.

530


FIGURE 7. The flowchart illustrates the proposed appropriate approach for combining CE-MRI and PET/CT. PET/CT is used to characterize lesions classified as indeterminate on CE-MRI as malignant or not using different cutoff SUVmax for mucinous and nonmucinous tumors depending on the classification made on CE-MRI. * 2014 Lippincott Williams & Wilkins




physiological factors such as patient preparation procedures, scan acquisition, image reconstruction, and data analysis settings, and thus, our results cannot be directly applied to other institutions.27,28 Multicenter studies with strictly standardized methods will be necessary to obtain more accurate values for cutoff SUVmax. In conclusion, our findings suggest that FDG uptake in ovarian masses correlates with pathological subtypes as well as the grade of malignancy. Furthermore, because PET/CT can be used as a complement to classify indeterminate lesions as malignant or not based on appropriate cutoff SUVmax for mucinous and nonmucinous tumors, the combination of CE-MRI and PET/CT is a highly accurate method for characterizing ovarian masses.

18. Kitajima K, Suzuki K, Senda M, et al. FDG-PET/CT for diagnosis of primary ovarian cancer. Nucl Med Commun. 2011;32:549Y553. 19. Yamamoto Y, Oguri H, Yamada R, et al. Preoperative evaluation of pelvic masses with combined 18F-fluorodeoxyglucose positron emission tomography and computed tomography. Int J Gynaecol Obstet. 2008;102:124Y127. 20. Jung DC, Choi HJ, Ju W, et al. Discordant MRI/FDG-PET imaging for the diagnosis of borderline ovarian tumors. Int J Gynecol Cancer. 2008; 18:637Y641. 21. Risum S, Høgdall C, Loft A, et al. The diagnostic value of PET/CT for primary ovarian cancerVa prospective study. Gynecol Oncol. 2007;105:145Y149. 22. Nam EJ, Yun MJ, Oh YT, et al. Diagnosis and staging of primary ovarian cancer: correlation between PET/CT, Doppler US, and CT or MRI. Gynecol Oncol. 2010;116:389Y394. 23. Yang ZY, Hu SL, Shi W, et al. The clinical value of fluorine-18 fluorodeoxyglucose positron emission tomography/computed tomography in postoperative patients with gastrointestinal mucinous adenocarcinoma. Nucl Med Commun. 2011;32:1018Y1025. 24. Shim SS, Han J. FDG-PET/CT imaging in assessing mucin-producing nonYsmall cell lung cancer with pathologic correlation. Ann Nucl Med. 2010; 24:357Y362. 25. Takanami K, Hiraide T, Tsuda M, et al. Additional value of FDG PET/CT to contrast-enhanced CT in the differentiation between benign and malignant intraductal papillary mucinous neoplasms of the pancreas with mural nodules. Ann Nucl Med. 2011;25:501Y510. 26. Tavassoli FA, Devilee P. World Health Organization Classification of Tumors: Pathology and Genetics of Tumors of the Breast and Female Genital Organs. Lyon, France: IARC Press; 2003. 27. Boellaard R, Oyen WJ, Hoekstra CJ, et al. The Netherlands protocol for standardization and quantification of FDG whole body PET studies in multicenter trials. Eur J Nucl Med Mol Imaging. 2008;35:2320Y2333. 28. Boellaard R. Standards for PET image acquisition and quantitative data analysis. J Nucl Med. 2009;50:11Y20. 29. Morelli JN, Runge VM, Ai F, et al. Magnetic resonance evaluation of renal artery stenosis in a swine model. Invest Radiol. 2012;47:376Y382. 30. Tanaka YO, Nishida M, Kurosaki Y, et al. Differential diagnosis of gynecological ‘‘stained glass’’ tumours on MRI. Br J Radiol. 1999;72:414Y420. 31. Kang DE, White RL Jr, Zuger JH, et al. Clinical use of fluorodeoxyglucose F 18 positron emission tomography for detection of renal cell carcinoma. J Urol. 2004;171:1806Y1809. 32. DeLair D, Oliva E, Ko¨bel M, et al. Morphologic spectrum of immunohistochemically characterized clear cell carcinoma of the ovary: a study of 155 cases. Am J Pathol. 2011;35:36Y44. 33. Matsuoka Y, Ohtomo K, Araki T, et al. MR imaging of clear cell carcinoma of the ovary. Eur Radiol. 2001;11:946Y951. 34. Quick HH, von Gall C, Zeilinger M, et al. Integrated whole-body PET/MR hybrid imaging: clinical experience. Invest Radiol. 2013;48:280Y289. 35. Hartung-Knemeyer V, Beiderwellen KJ, Buchbender C, et al. Optimizing positron emission tomography image acquisition protocols in integrated positron emission tomography/magnetic resonance imaging. Invest Radiol. 2013; 48:290Y294. 36. Nakajo K, Tatsumi M, Inoue A, et al. Diagnostic performance of fluorodeoxyglucose positron emission tomography/magnetic resonance imaging fusion images of gynecological malignant tumors: comparison with positron emission tomography/computed tomography. Jpn J Radiol. 2010;28:95Y100. 37. Takeuchi M, Matsuzaki K, Nishitani H. Diffusion-weighted magnetic resonance imaging of ovarian tumors: differentiation of benign and malignant solid components of ovarian masses. J Comput Assist Tomogr. 2010;34:173Y176. 38. Thomassin-Naggara I, Toussaint I, Perrot N, et al. Characterization of complex adnexal masses: value of adding perfusion- and diffusion-weighted MR imaging to conventional MR imaging. Radiology. 2011;258:793Y803. 39. Lin C, Itti E, Luciani A, et al. Whole-body diffusion-weighted imaging with apparent diffusion coefficient mapping for treatment response assessment in patients with diffuse large B-cell lymphoma: pilot study. Invest Radiol. 2011;46:341Y349. 40. Esseridou A, Di Leo G, Sconfienza LM, et al. In vivo detection of choline in ovarian tumors using 3D magnetic resonance spectroscopy. Invest Radiol. 2011;46:377Y382.


REFERENCES 1. Covens AL, Dodge JE, Lacchetti C, et al. Surgical management of a suspicious adnexal mass: a systemic review. Gynecol Oncol. 2012;126:149Y156. 2. Cadron I, Leunen K, Van Gorp T, et al. Management of borderline ovarian neoplasms. J Clin Oncol. 2007;25:2928Y2937. 3. Imaoka I, Wada A, Kaji Y, et al. Developing an MR imaging strategy for diagnosis of ovarian masses. Radiographics. 2006;26:1431Y1448. 4. Hricak H, Chen M, Coakley FV, et al. Complex adnexal masses: detection and characterization with MR imagingVmultivariate analysis. Radiology. 2000;214:39Y46. 5. Sohaib SA, Sahdev A, Van Trappen P, et al. Characterization of adnexal mass lesions on MR imaging. AJR Am J Roentogenol. 2003;180:1297Y1304. 6. Kim KA, Park CM, Lee JH, et al. Benign ovarian tumors with solid and cystic components that mimic malignancy. AJR Am J Roentogenol. 2004;182: 1259Y1265. 7. Tanaka YO, Okada S, Satoh T, et al. Ovarian serous surface papillary borderline tumors form sea anemone-like masses. J Magn Reson Imaging. 2011;33: 633Y640. 8. Kataoka M, Togashi K, Koyama T, et al. MR imaging of Mullerian mucinous borderline tumors arising from endometriotic cysts. J Comput Assist Tomogr. 2002;26:532Y537. 9. Bazot M, Haouy D, Darai E, et al. Is MRI a useful tool to distinguish between serous and mucinous borderline ovarian tumours? Clin Radiol. 2013;68:e1Ye8. 10. deSouza NM, O’Neill R, Mclndoe GA, et al. Borderline tumors of the ovary: CT and MRI features and tumor markers in differentiation from stage 1 disease. AJR Am J Roentgenol. 2005;184:999Y1003. 11. Okamoto Y, Tanaka YO, Tsunoda H, et al. Malignant or borderline mucinous cystic neoplasms have a larger number of loculi than mucinous cystadenoma: a retrospective study with MR. J Magn Reson Imaging. 2007;26:94Y99. 12. Bent CL, Sahdev A, Rockall AG, et al. MRI appearances of borderline ovarian tumours. Clin Radiol. 2009;64:430Y438. 13. Kubik-Huch RA, Do¨rffler W, von Schulthess GK, et al. Value of (18F)-FDG positron emission tomography, computed tomography, and magnetic resonance imaging in diagnosing primary and recurrent ovarian carcinoma. Eur Radiol. 2000;10:761Y767. 14. Kawahara K, Yoshida Y, Kurokawa T, et al. Evaluation of positron emission tomography with tracer 18-fluorodeoxyglucose in addition to magnetic resonance imaging in the diagnosis of ovarian cancer in selected women after ultrasonography. J Comput Assist Tomogr. 2004;28:505Y516. 15. Fenchel S, Grab D, Nuessle K, et al. Asymptomatic adnexal masses: correlation of FDG PET and histopathologic findings. Radiology. 2002;223:780Y788. 16. Rieber A, Nu¨ssle K, Sto¨hr I, et al. Preoperative diagnosis of ovarian tumors with MR imaging: comparison with transvaginal sonography, positron emission tomography, and histologic findings. AJR Am J Roentgenol. 2001;177: 123Y129. 17. Castellucci P, Perrone AM, Piucchio M, et al. Diagnostic accuracy of 18FFDG PET/CT in characterizing ovarian lesions and staging ovarian cancer: correlation with transvaginal ultrasonography, computed tomography, and histology. Nucl Med Commun. 2007;28:589Y595.




531

Evaluation of sellar and parasellar masses by computed tomography.

Computed tomography of intracerebral leukemic masses.

The role of computed tomography in the evaluation of neck masses.

Prospective comparison of ultrasound and computed tomography in the evaluation of gynecologic pelvic masses.

Variability in computed tomography diameter measurements of solid renal masses.

Ultrasonography and computed tomography in the diagnosis of homogeneous masses.

Computed tomography of localized adipose deposits presenting as tumor masses.

Mediastinal masses: magnetic resonance imaging in comparison with computed tomography.

Diagnostic potential of multidetector computed tomography for characterizing small renal masses.

Spontaneous rupture of ovarian cystadenocarcinoma: pre- and post-rupture computed tomography evaluation.

Evaluation of serum mesothelin in malignant and benign ovarian masses.

Evaluation of computed tomography: achievement and challenge.

Evaluation of pediatric hydrocephalus by computed tomography.

Critical clinical appraisal of the role of computed tomography-guided minimally invasive aspiration cytology in evaluation of retroperitoneal masses.

Evaluation of abdominal abscess with computed tomography.

Evaluation of seizure disorders by computed tomography.

Is postoperative computed tomography evaluation a prognostic indicator in patients with optimally debulked advanced ovarian cancer?

Comparison between computed tomography multislice and high-field magnetic resonance in the diagnostic evaluation of patients with renal masses.

Computed tomography and magnetic resonance imaging features of ovarian fibrothecoma.

Is computed tomography scan the ultimate modality for airway evaluation?

computed tomography in diagnostic evaluation of carcinoma urinary bladder: comparison with computed tomography.

Letter: Ultrasonography of ovarian masses.

Imaging of pediatric ovarian masses.

Role of Computed Tomography (CT) in Localisation and Characterisation of Suprahyoid Neck Masses.