Whole body MRI (WB-MRI) has been gaining momentum as a tool in cancer staging as a consequence of recent developments in scanning technology. Prostate cancer staging is a particular area of interest as it is the most common cancer in Western males¹.

The aim of this study is to evaluate the role of multiparametric WB-MRI in the metastatic and nodal staging of prostate cancer at initial presentation, using a 90-minute scanning protocol combining pre and post contrast mDIXON, T2WI and diffusion-weighted-imaging (DWI). The overall diagnostic accuracy of WB-MRI is compared against an 18FCH-PET-CT/expert panel based reference standard.

23 men with biopsy confirmed intermediate or high-risk prostate cancer² were recruited to this prospective study between July 2012 and June 2015 at a single Tertiary referral centre.

WB-MRI was performed using a 3T scanner (Ingenia, Philips Healthcare, Netherlands) with coverage from vertex to feet. 20mls intravenous gadolinium was hand injected prior to acquisition of post-contrast images. A full description of acquisition parameters is provided in Figure 1.

¹⁸F-fluorocholine PET-CT was performed, within 2 weeks of the WB-MRI, using an integrated 64-slice PET/CT scanner (Discovery VCT; GE Healthcare) with coverage from vertex to mid thigh.

Image analysis

Two radiologists independently reviewed each WB-MRI dataset (HS, with 7 years radiology experience; and NR, with 10 years radiology experience). Both radiologists were aware of the PSA level only, and were unaware of other clinical or imaging results. Radiologists recorded the presence or absence of nodes or metastases at soft tissue/bony sites.

The criteria in Figure 2 were used to classify PET-CT and WB-MRI findings. Where discordances between WB-MRI and PET-CT arose, a best value comparator was chosen as it was deemed unethical to biopsy all discordances. This involved review by a multidisciplinary team including an Oncologist, Urologist and specialty Radiologist where biopsy, further imaging tests or imaging follow-up were suggested at clinical discretion.

False negative and false positive findings were classified as caused by either:

(i) disease incorrectly ascribed as present/absent by a reader which on retrospective evaluation can be correctly attributed as present/absent.

(ii) technical error - disease incorrectly ascribed as present/absent by a reader which on retrospective evaluation cannot be correctly attributed as present/absent.

Statistical analysis

Agreement between MRI readers, and between multi-parametric whole-body MRI and PET-CT was assessed using Cohen’s kappa (κ). Sensitivity and specificity of multi-parametric whole-body MRI and 18FCH PET-CT were determined for nodal and bony disease using a patient-based analysis against the reference standard.

23 patients had a median age of 65.6 years (range 51.9 - 81.7) and a median PSA of 23.0 (range 5 – 587).

According to the reference standard, nodal involvement was present in 5 and absent in 18 patients. Bone involvement was present in 2 and absent in 21 patients. There were no metastases to solid organs.

Comparison of WB-MRI and PET-CT against reference standard

For nodal disease, PET-CT was 100% sensitive and 88.88% specific due to 2 technical false positives. WB-MRI was 100% sensitive and 94.44% specific due to 1 technical false positive.

For bony disease, PET-CT was 50% sensitive due to 1 technical false negative, which is shown in Figure 3. PET-CT was 95.23% specific due to one technical false positive. Whole body MRI was 100% sensitive and 90.38% specific due to 2 technical false positives. An example of a technical false positive is given in Figure 4.

Radiologist performance of WB-MRI

The agreement between WB-MRI and PET-CT was κ=0.775 for nodal disease (substantial agreement) and 0.246 for bony disease (fair agreement).

In the nodes, there was a perceptual error rate of 2.2% (1FN in [2x 23] =46 reads) In the bones, there was a perceptual error rate of 13.0% (3FP, 3FN/46) No perceptual errors were made during the PET readings.

There is a lack of data comparing WB-MRI and choline PET-CT in the primary staging of prostate cancer, with a handful of small studies, of which two used DWI^3-4 This is the first study to use four diffusion b-values and a DIXON technique. A very high technical diagnostic accuracy was achieved for both nodal and metastatic disease, with WB-MRI outperforming PET-CT in the detection of bone metastases. However, the higher rates of perceptual errors found with WB-MRI suggest reporting experience is required.Whole body MRI offers improved sensitivity in bone lesion detection when compared with choline PET-CT. Other advantages include the lack of ionising radiation, widespread availability of scanning technology and the ability to perform tests as a ‘one-stop’ staging examination in combination with multiparametric prostate MRI.