Eur Radiol DOI 10.1007/s00330-013-3083-8
UROGENITAL
Whole-body MRI with diffusion-weighted sequence for staging of patients with suspected ovarian cancer: a clinical feasibility study in comparison to CT and FDG-PET/CT Katrijn Michielsen & Ignace Vergote & Katya Op de beeck & Frederic Amant & Karin Leunen & Philippe Moerman & Christophe Deroose & Geert Souverijns & Steven Dymarkowski & Frederik De Keyzer & Vincent Vandecaveye
Received: 8 August 2013 / Revised: 13 November 2013 / Accepted: 14 November 2013 # European Society of Radiology 2013
Abstract Objectives To evaluate whole-body MRI with diffusionweighted sequence (WB-DWI/MRI) for staging and assessing operability compared with CT and FDG-PET/CT in patients with suspected ovarian cancer. Methods Thirty-two patients underwent 3-T WB-DWI/MRI, 18 F-fluorodeoxyglucose positron emission tomography/ computed tomography (FDG-PET/CT) and CT before diagnostic open laparoscopy (DOL). Imaging findings for tumour characterisation, peritoneal and retroperitoneal staging were Electronic supplementary material The online version of this article (doi:10.1007/s00330-013-3083-8) contains supplementary material, which is available to authorized users. K. Michielsen : K. Op de beeck : S. Dymarkowski : F. De Keyzer : V. Vandecaveye (*) Department of Radiology, Medical Imaging Research Centre, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium e-mail:
[email protected] I. Vergote : F. Amant : K. Leunen Department of Obstetrics and Gynaecology, Leuven Cancer Institute, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium P. Moerman Department of Morphology and Molecular Pathology, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium C. Deroose Department of Nuclear Medicine, Medical Imaging Research Centre, University Hospitals Leuven, Herestraat 49, 3000 Leuven, Belgium G. Souverijns Department of Radiology, Jessa Ziekenhuis – Campus Virga Jessa, Stadsomvaart 11, 3500 Hasselt, Belgium
correlated with histopathology after DOL and/or open surgery. For distant metastases, FDG-PET/CT or image-guided biopsies were the reference standards. For tumour characterisation and peritoneal staging, WB-DWI/MRI was compared with CT and FDG-PET/CT. Interobserver agreement for WBDWI/MRI was determined. Results WB-DWI/MRI showed 94 % accuracy for primary tumour characterisation compared with 88 % for CT and 94 % for FDG-PET/CT. WB-DWI/MRI showed higher accuracy of 91 % for peritoneal staging compared with CT (75 %) and FDGPET/CT (71 %). WB-DWI/MRI and FDG-PET/CT showed higher accuracy of 87 % for detecting retroperitoneal lymphadenopathies compared with CT (71 %). WB-DWI/MRI showed excellent correlation with FDG-PET/CT (κ=1.00) for detecting distant metastases compared with CT (κ=0.34). Interobserver agreement was moderate to almost perfect (κ=0.58–0.91). Conclusions WB-DWI/MRI shows high accuracy for characterising primary tumours, peritoneal and distant staging compared with CT and FDG-PET/CT and may be valuable for assessing operability in ovarian cancer patients. Key Points • Whole-body MRI with diffusion weighting (WB-DWI/MRI) helps to assess the operability of suspected ovarian cancer. • Interobserver agreement is good for primary tumour characterisation, peritoneal and distant staging. • WB -DWI /MRI improves mesenteric /serosal metastatic spread assessment compared with CT and FDG-PET/CT. • Retroperitoneal/cervical-thoracic nodal staging using qualitative DWI criteria was reasonably accurate. • WB-DWI/MRI and FDG-PET /CT showed the highest diagnostic impact for detecting thoracic metastases.
Eur Radiol
Keywords Diffusion-weighted MRI . Whole-body imaging . Ovarian cancer . Tumour staging . Gynaecologic surgical procedures
The aim of this study was to evaluate the diagnostic value of WB-DWI/MRI for staging and assessing operability compared with CT and FDG-PET/CT in patients with suspected ovarian cancer.
Abbreviations and acronyms WB-DWI/MRI whole-body MRI with diffusion-weighted sequence DOL diagnostic open laparoscopy PC peritoneal carcinomatosis FIGO International Federation of Gynaecology and Obstetrics NACT neoadjuvant chemotherapy CA cancer antigen MPR multiplanar reformatting TSE turbo spin-echo SI signal intensity PPV positive predictive value NPV negative predictive value
Materials and methods
Introduction Ovarian cancer is often diagnosed with extensive peritoneal and/or distant metastases [International Federation of Gynaecology and Obstetrics (FIGO) stages IIIc and IV], decreasing the patient’s survival rate to 20–40 % in FIGO stage IIIc disease and to 10 % in FIGO stage IV [1]. Complete macroscopic tumour resection (R0 resection) has been demonstrated to be the single most important prognostic factor [2–9]. If disease extent precludes upfront radical debulking surgery, neoadjuvant chemotherapy (NACT) followed by interval debulking, represents a valid alternative in patients with advanced stage IIIc or IV ovarian cancer [2]. Imaging aims to identify patients unfit for surgery, either by diagnosing a primary tumour other than ovarian cancer, or by depicting disease volume and/or extent beyond the reach of surgery [10, 11]. Although 18 F-fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) surpasses CT in the detection of lymphadenopathies and distant metastases, both may underestimate peritoneal and particularly mesenteric and serosal bowel disease extent, necessitating surgical staging to assess operability [12–17]. Conversely, the combination of contrast-enhanced magnetic resonance imaging (MRI) with diffusion-weighted imaging (DWI) has shown high accuracy for staging of peritoneal carcinomatosis (PC) [18, 19]. Because at least 11 % of patients have distant metastases, including cervical-thoracic metastases, the development of whole-body MRI with diffusionweighted sequence (WB-DWI/MRI) over abdominal DWI/ MRI may be justified [20].
Patients and selection of treatment Approval was obtained from the local Institutional Review Board and all patients provided written informed consent. The inclusion criterion for the study was suspicion of ovarian cancer based on gynaecological ultrasound or CA125. Patients with contraindications to MRI were excluded. Thirty-two women (mean age 61.9, range 20–83 years) were consecutively enrolled in this prospective study between October 2010 and February 2012. Routine diagnostic workup included clinical and CA-125 assessment, gynaecological ultrasound by an expert sonographer, thoraco-abdominal CT and diagnostic open laparoscopy (DOL) [21]. Additionally, all patients underwent WB-DWI/MRI and FDG-PET/CT before DOL. The decision to treat by primary debulking surgery, or by NACT with interval debulking, was determined according to well-defined criteria [22]. In short, the main criteria precluding upfront debulking surgery consisted of medical comorbidities preventing major surgery and disease-related factors: (1) Distant metastases, except for resectable inguinal lymph nodes, solitary resectable retrocrural or paracardiac lymph nodes, and pleural fluid with cytological proof of tumour but without detectable macroscopic tumour deposits on imaging; (2) The presence of hepatic metastases; (3) Tumoral infiltration of the duodenum, stomach, pancreas, large vessels of the coeliac trunk and hepatoduodenal ligament, and metastases behind the portal vein; (4) Diffuse serosal carcinomatosis necessitating multiple intestinal resections; (5) Deep tumoral involvement of the superior mesenteric artery and mesenteric root of the small bowel.
Imaging techniques Computed tomography Breath-hold contrast-enhanced [intravenous injection of 120 ml iodinated contrast agent (Visipaque, GE Healthcare); 320 mg/ml; chest after 17 s, abdomen after 90 s] CT (Sensation 16, Sensation 64, Definition Flash, Siemens Medical Systems, Erlangen, Germany) was obtained 90 min
4.57×4.71
2.19×2.16
1
Reconstructed voxel size (mm)
Number of signal averages (NSA)
Field of view (FOV) (mm)
Acquired voxel size (mm)
0.1
420×329
Intersection gap (mm)
5 Echo time (TE) (ms)
5
67
Repetition time (TR) (ms)
50/station
8,454
Parallel imaging factor
Slice number
2.5
b-Values (s/mm2)
Slice thickness (mm)
STIR (TI=250 ms)
0-1,000
Fat suppression
Free-breathing
DWI, diffusion-weighted imaging; mDIXON, multi-echo 2-point Dixon; eTHRIVE, T1-weighted high-resolution isotropic volume examination; SSTSE, single-shot turbo spin-echo imaging; STIR, short T1 inversion recovery; SPAIR, spectrally adiabatic inversion recovery
1 1 1 1 1
1.49×1.5
0.98×0.97 0.71×0.71
1.49×1.5 1.49×1.5
0.71×0.71 0.93×0.93
1×1 1×1
0.78×0.78
0
375×304 400×352
0 0
375×304 375×447
0.6 0.6
357×339
1.5
148 133
1.5 1.5
90 35/station
6 6 5
41/station
3.2
2
None
1.5 1.25 – 2.20 1.25 – 2.20 87
87
3.6 3.6 3,000 3,000
2 2 4 2
None
SPAIR (mDIXON) SPAIR (mDIXON)
None None
None None
None
Chest (1) Abdominopelvic (2) 20 s breath-hold
Abdominopelvic (2) 3
Respiratory Respiratory
4 MPR MPR 4
Transverse Coronal
Sagittal Transverse
Respiration
CT, WB-DWI/MRI and FDG-PET/CT were separately evaluated, by two abdominal radiologists (12 and 10 years of
Image stations head to mid-thigh
Interpretation of imaging
Table 1 Sequence parameters for WB-DWI/MRI
Patients drank 1 l of pineapple juice 2 h before the MRI to distend the bowel lumen and suppress the signal from bowel contents because of the negative contrast properties resulting from its manganese content [23, 24]. In addition, an antispasmodic agent (butylhyoscine, 20 mg IV) was administered at the start of abdominal DWI. Three-Tesla WB-DWI/MRI (Ingenia, Philips Healthcare, Best, The Netherlands) was performed with parallel radiofrequency transmission and phasedarray head-neck and surface coils. Free-breathing WB-DWI was acquired at four imaging stations, covering the head and neck, chest, upper abdomen and pelvis. Multiplanar reformatted (MPR) coronal and sagittal WB-DWI images were reconstructed from the transverse b1000 images. WB transverse and coronal respiratory triggered T2-weighted turbo spin-echo (TSE) and breath-hold abdominal-pelvic and thoracic contrast-enhanced 3D T1-weighted gradient-echo sequences obtained 80 s after injection of 15 ml gadoliniumDOTA were used for anatomical reference. The total imaging time was 40 min. A detailed overview of the imaging protocol is shown in Table 1.
T2 SSTSE
WB-DWI/MRI
DWI
Coronal
Hybrid FDG-PET/CT (Biograph 40 TruePoint with TrueV, Siemens Medical Solutions, Erlangen, Germany) was performed for 25/32 patients, whereas stand-alone FDG-PET (HR+, Siemens, Knoxville, TN, USA) was performed for 7/32 patients. Patients fasted for 6 h before the examination. For FDG-PET/ CT, whole-body, free-breathing spiral CT was performed first (85 mAs, 120 kV, slice thickness 5 mm, collimation 24×1.2 mm, table feed 23 mm/rotation), with per-oral and intravenous iodinated contrast agent. PET images were acquired (mean 69 min, range 53– 99 min) after intravenous administration of a weightadjusted amount of FDG (average activity of 303 MBq, range 220–388). PET images were corrected for attenuation using the CT data. Because of constraints on imaging time, no attenuation correction was performed for the PET-only acquisitions. All PET images were iteratively reconstructed.
Transverse
Contrast-enhanced 3D T1 gradient-echo
FDG-PET(/CT)
Coronal
Transverse
after per-oral contrast ingestion [30 ml iodinated contrast agent (Telebrix Gastro, Guerbet), 300 mg/ml, in 900 ml water] using the following parameters: pitch 1.2, rotation speed 0.5 s, slice thickness 5 mm, slice gap 1 mm and collimation 0.6 mm. The reference tube voltage was set at 120 kV, and the reference current was 110 mAs for the thorax and 200 mAs for the abdomen using automated Care Dose software.
SPAIR (eTHRIVE)
Eur Radiol
Eur Radiol
experience, respectively) and a nuclear medicine physician (11 years of experience). Observers were blinded to all information regarding the other imaging tests, clinical, laboratory and pathological findings. The interpretation criteria are outlined in Table 2 following the previously published imaging criteria [10, 25–29]. WB-DWI/MRI: interobserver agreement Interobserver variability between an academic and a nonacademic radiologist was assessed. Of the 32 cases, 8 were randomly selected for 1-week training of the second reader. Subsequently, the WB-DWI/MRI findings for the remaining 24 cases were interpreted by the non-academic reader, who was blinded to the results of the academic radiologist. Reference standard Primary tumours and peritoneal and hepatic-hilar metastases were primarily confirmed during DOL and/or surgery with histopathology whenever possible. Retroperitoneal lymphadenopathies were primarily confirmed by histopathology after surgery. For distant metastases, FDG-PET/CT was used as reference standard if biopsy was technically impossible. Endobronchial ultrasound-guided fine-needle aspiration of a mediastinal lymphadenopathy was performed for three patients and cytology was performed after pleural drainage in one patient. In the case of NACT, the findings at interval debulking and histopathology were used for additional correlation with the initial DOL. To facilitate this correlation, lesions recorded at preoperative imaging were topographically correlated and annotated at imaging follow-up before interval debulking. Primary tumours were classified as benign, primary ovarian (borderline-malignant) or non-ovarian in origin. The extent of PC was classified as follows: the bladder peritoneal surface, Douglas pouch, peritoneal left and right pelvic, lateroconal area and diaphragm, subhepatic space/Morrison’s pouch, hepatic surface, splenic surface, omentum, small bowel, and colonic mesentery and serosa. Retroperitoneal nodal extent was classified as infrarenal (aorto-caval, right and left iliac) and suprarenal. Hepatic-hilar metastases encompassed the portal vein, gallbladder and falciform ligament. The extent of abdominal disease at imaging, surgery and histopathology was assessed following higher mentioned classification (Fig. 1). Accordingly, resection specimens were labelled before fixation, and histopathological findings were reported. When surgical and pathological findings were discordant, pathological findings prevailed. The size distribution of metastatic lesions was recorded at surgical evaluation by measuring the largest lesion per region or by correlative anatomical imaging.
Statistical Analysis Statistica 9.1 software (Statsoft Inc., Tulsa, OK, USA) was used for all statistical analyses, with P 0.05). Characterisation of primary lesions WB-DWI/MRI showed 100 % sensitivity, 50 % specificity, 94 % accuracy, NPV of 100 % and PPV of 93 % for the characterisation of primary lesions (malignant versus benign, irrespective of tumour origin) compared with 96 % sensitivity, 25 % specificity, 88 % accuracy, NPV of 50 % and PPV of 90 % for CT and 100 % sensitivity, 33 % specificity, 94 % accuracy, NPV of 100 % and PPV of 94 % for FDG-PET/CT. Peritoneal carcinomatosis Comparative sensitivities, specificities, accuracies, PPVs and NPVs of WB-DWI/MRI, CT and FDG-PET/CT for detecting PC are shown in Table 4 and the supplementary table. Table 4 Per-site accuracy for WB-DWI/MRI
Site Peritoneal cavity Bladder surface Douglas pouch Right pelvic Right lateroconal
TP, true positive; FN, false negative; FP, false positive; TN, true negative; Sens, sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value; Acc, accuracy
Subhepatic+Morrison’s space Right diaphragm Hepatic surface Left diaphragm Surface spleen Left lateroconal Left pelvic Omentum Small bowel serosa Small bowel mesentery Colonic serosa Colonic mesentery Summary Retroperitoneum Infrarenal Suprarenal Summary Liver-hilum
For overall peritoneal staging, the sensitivity and specificity of WB-DWI/MRI were significantly higher than CT and PET/ CT (P