An update of pitfalls in prostate mpMRI: a practical approach through the lens of PI-RADS v. 2 guidelines

High-quality multi-parametric Magnetic Resonance Imaging (mpMRI) has become an important tool in the diagnosis, characterisation and treatment planning of prostate cancer (PCa) [1‐3]. Specifically, mpMRI involves T2-weighted anatomical images combined with functional imaging methods such as dynamic Contrast Enhanced Imaging (DCE), Diffusion-Weighted Imaging (DWI) and/or Spectroscopy, if necessary. All these sequences have a limited accuracy, when considered individually, but their association has shown a greater performance in the assessment of PCa [4].

However, inter-reader variability represents a real limitation for prostate mpMRI, and expertise in reporting is crucial to improve cancer detection and staging accuracy. Therefore, it is important to validate the current protocols [5]. In 2012, the European Society of Urogenital Radiology (ESUR) set up an expert panel to develop a standardised system for prostate mpMRI interpretation and reporting, under the name of Prostate Imaging Reporting and Data System (PI-RADS) [6]. In 2015 a revision of this classification led to the publication of PI-RADS v. 2 [7], with the aim to promote a global standardisation of these guidelines, and to reduce the variability in the acquisition, interpretation and reporting of prostate mpMRI. A detailed explanation of PI-RADS v. 2 is beyond the aim of this report, but it is worth summarising the main points of these guidelines: (1) DWI is the dominant sequence in the peripheral zone (PZ); (2) T2-weighted imaging (T2-WI) is the dominant sequence in the transitional zone (TZ); (3) the role of DCE is secondary to DWI in the PZ; and (4) there is an overall 5-point (from 1 to 5) PI-RADS assessment (1: low probability – 5: very high probability of clinically significant cancer).

Nonetheless, the interpretation of mpMRI of the prostate can be challenging and new radiological skills are needed, especially when potential pitfalls (i.e., normal anatomic structures, benign conditions of the prostate or artefacts due to technical issues) might be erroneously interpreted as pathological conditions. Moreover, the PI-RADS v. 2 guidelines may be subject to some interpretation variability according to the radiologists’ individual experience, lowering the ability to distinguish pitfalls from true malignancy [8]. Awareness of these pitfalls is therefore fundamental.

Currently, there are only three reports (two from the United States and one from Europe) [9‐11] addressing the mpMRI pitfalls in PCa. They have all suggested some strategies to assist the radiologists in avoiding misdiagnosis (and consequently mistreatment), but a systematic approach on how to tackle these aspects applying PI-RADS v. 2 has yet to be reported.

Therefore, the purpose of this article is to provide a practical approach for imaging interpretation in PCa, with a special focus on how to apply PI-RADS v. 2 to discriminate pathological conditions from the most common pitfalls that could be encountered during daily clinical practice. Of note, this report is based on a pragmatic consensus among a panel of different international radiologists highly experienced in mpMRI of the prostate (VP, FG, YXK, FC, GV).

For the sake of completeness, the cases shown in this pictorial report represent a cohort of men aged 47–79 years, with prior suspicion of PCa based on abnormal digital rectal examination, rise of prostate specific antigen (PSA) and/or family history of PCa. The exams have been acquired on a 3.0 T system (Discovery MR750, GE Healthcare, Waukesha, WI), by using a 32 multi-channel (or 8 multi-channel + endorectal coil) surface phased-array body coil.

Awareness of diagnostic pitfalls is important to avoid both false-positive and false-negative interpretations. A systematic approach to mpMRI of the prostate using PI-RADS v. 2 helps in the identification of PCa, but there still remains a wide spectrum of pitfalls. These can be broadly split into two main groups: (1) pitfalls related to clinical indications and (2) technical and physiological artefacts (Table 1).

These pitfalls are related both to the anatomy of the prostate and to certain benign conditions of the gland, which can mimic the presence of cancer. The typical appearance of PCa in the PZ is a focal hypointense area on T2-WI; however, other benign conditions, such as prostatitis, fibro-muscular bands and post-biopsy haemorrhage can mimic this signal change, leading to potential misdiagnoses.

The use of an endorectal coil in addition to the surface coil improves the signal-to-noise ratio and the spatial resolution both at 1.5 and 3 T. On the contrary, patient or bowel movements during image acquisition may cause repetitive circular artefacts along the boundaries of the endorectal coil. These artefacts can be minimised by rectal emptying and by the administration of a spasmolytic drug prior to the examination.

According to PI-RADS v. 2, an area of homogeneous low-signal intensity on T2-WI in the PZ, showing restricted diffusion on DWI and focal enhancement (+) can be scored as 4/5, suggesting the presence of clinically significant PCa. However, the use of a single surface coil could erroneously suggest extracapsular extension (T3 stage) because of the lower resolution. In this case, the application of PI-RADS v. 2 together with the use of the endorectal coil could help to rule out capsular involvement (T2 stage), thanks to the increased resolution.

The endorectal coil should be positioned correctly, in order to avoid the risk of incurring potential diagnostic pitfalls that could mimic the presence of PCa. The correct position of the coil is in a plane perpendicular to the left-right phase encoding direction. If the coil is not positioned properly, it is possible to see a focal area of enhancement (+) with restricted diffusion (scored 4/5 according to PI-RADS v. 2) adjacent to the coil surface. This finding could be erroneously interpreted as a lesion (Fig. 15), but the ADC maps and T2-WI do not show this artefact, and therefore PCa can be ruled out.

To sum up, there are different artefacts that can mimic the presence of PCa and which the radiologist should be aware of. These include the presence of gas in the rectum, and the interpretation of ADC maps with low-value pixels, as these latter can show the same dark signal as fat, creating problems for lesions along the capsule. As far as the use of spasmolytic agents is concerned, the radiologist should bear in mind that many patients undergoing mpMRI will have a large prostate due to BPH (with related urinary problems). Therefore, these medications should be administered carefully and preferably before positioning the patient on the mpMRI table. Glucagon should be chosen in patients with urinary retention.

Since its introduction, mpMRI of the prostate has been changing the management of suspicious PCa, especially in men with non-specific high PSA, where the detection of clinically significant PCa has been shown to be more accurate than standard TRUS biopsy [14]. Despite the standardised attempts to report mpMRI (PI-RADS v 2) [7], some para-physiologic appearances of the prostate gland can mimic cancer. As there are no established guidelines that suggest scoring a TZ lesion that protrudes into the PZ using the dominant sequence from the TZ (rather than PZ), we deem that in this scenario the experience of the radiologist and the knowledge of prostate anatomy and specific morphological features of BPH (e.g., regular capsule and margins) are more important than ever. Moreover, we believe that the radiologist should be aware of clinical data such as the exposure to antiandrogen therapy for BPH, as this could affect the conspicuity of tumours in the TZ as well as on DWI. The radiologist should also keep in mind that other sequences (e.g., DWI) can be of great help while reporting prostate mpMRI, if there are some doubts on T2-WI, as suggested by the PI-RADS v. 2 guidelines [15‐21]. Although these guidelines use DWI to upgrade some PI-RADS 3 lesions in the TZ to PI-RADS 4, DWI alone is nonetheless a sensitive sign for detection of tumours even in the TZ [22‐23]. Rosenkrantz and colleagues [22] reported that the incorporation of DWI and ADC maps (b value:1000 s/mm²) significantly improves the sensitivity for TZ tumours compared with T2-WI alone. The main reason is the diffuse background heterogeneity and the presence of multiple nodules in the TZ, which make tumours harder to identify on T2-WI. As known, DWI investigates the movement of water molecules within tissues and reflects changes in cellularity; thus, it provides complementary information that may help depict lesions not initially visible on T2-WI, leading to improved sensitivity.

In conclusion, we recommend that any radiologist involved in prostate mpMRI be fully aware of the pitfalls mentioned in this pictorial report, in order to avoid underestimation and overestimation of PCa detection. We also deem that this manuscript gives a repertoire of potential solutions for the improvement of the future PI-RADS guidelines.