Pearls and Pitfalls in Applying PI-RADS 2.1

Abstract The use of Prostate Imaging Reporting and Data System (PI-RADS) with multiparametric magnetic resonance imaging (MRI) has significantly improved the detection of clinically significant prostate cancer (csPCa), but there are certain challenges that the reader may face. This review provides an overview of the pitfalls associated with the PI-RADS system for multiparametric prostate MRI (mpMRI), with suggestions/pearls to help overcome these pitfalls. PI-RADS assessment is hindered by several causes of false positives (FPs) and false negatives (FNs). In addition, there is wide variability in the positive predictive value (PPV) of the PI-RADS system across different centers, highlighting the need for improvement. While the negative predictive value (NPV) for csPCa is generally high, variations exist. This review discusses the pitfalls contributing to FNs, including MRI artifacts, such as susceptibility and motion artifacts. Techniques to optimize image acquisition, such as switching the phase encoding direction and reducing bowel peristalsis, are suggested to mitigate these artifacts. Improper b -value selection for diffusion-weighted imaging (DWI) is another pitfall, emphasizing the importance of using high b -values (≥1,400 s/mm 2 ) to optimize neoplasm detection. Similarly, optimizing window settings to visualize csPCa, correctly positioning the endorectal coil, awareness of rare variants like mucinous adenocarcinoma and cribriform adenocarcinoma, and distinguishing central zone tumors from normal central zone are discussed. This article highlights the common pitfalls causing FPs, such as benign pathologies like prostatitis, granulomatous prostatitis, prostatic abscess, stromal BPH nodules, extruded BPH nodules, and prostatic calcifications. It also discusses the pitfalls related to normal anatomical structures, including the central zone, anterior fibromuscular stroma, thickened surgical capsule, neurovascular bundle, and periprostatic venous plexus. Techniques for carefully evaluating these entities' morphology and distribution of signal abnormalities are described to avoid overdiagnosing these as PCa. The article also addresses the pitfalls related to postintervention changes, including postbiopsy hemorrhage and artifacts caused by the UroLift procedure, while providing recommendations for managing these challenges. Finally, the pitfalls that may be encountered during staging, including evaluation for extraprostatic extension, and pelvic nodal and osseous metastases, are highlighted.


Introduction/Background
Advancements in prostate magnetic resonance imaging (MRI), along with the Prostate Imaging Reporting and Data System (PI-RADS) scoring system, have changed the management of prostate cancer (PCa).Initial clinical guidelines in 2012 for multiparametric prostate MRI (mpMRI) included the original PI-RADS reporting system. 1 This was followed by an updated PI-RADS version 2 in 2015 and another update, PI-RADS version 2.1, in 2019. 2 These guidelines have profoundly impacted the detection of PCa, particularly clinically significant PCa (csPCa), with a recent meta-analysis showing a 94% accuracy of mpMRI for csPCa. 3owever, the utility of the PI-RADS scoring system is limited by false positives (FPs) and false negatives (FNs), lowering the positive predictive value (PPV) and negative predictive value (NPV).A recent large multicenter study demonstrated PPV of 35% (interquartile range [IQR], 27-44%) for a PI-RADS score 3/5 and 49% (IQR, 27-48%) for a PI-RADS score 4/5. 4 Thus, the PPV was low and varied widely across centers.The NPV of mpMRI is higher in comparison, with a meta-analysis showing an NPV of 88.1% (IQR, 85.7-92.3)for csPCa, 5 with negative defined as PI-RADS categories 1 and 2. Of note, the definition of positive mpMRI has varied in the literature as a score 3/5 or 4/5.PPV and NPV influence the test's utility and affect decisionmaking.Therefore, this review aims to highlight some pitfalls that are the common cause of FP and FN, which, if avoided, could improve the test's utility.

Discussion
PI-RADS version 2.1 has defined the primary sequences used in determining the overall suspicion for csPCa (dominant sequence) for the peripheral and transition zones (TZs).The abnormalities on the dominant sequence for a respective zone carry higher weightage.For the peripheral prostate zone (PZ), apparent diffusion coefficient (ADC) map with diffusion-weighted imaging (DWI) utilizing high b values (1,400 s/mm 2 ) is the dominant sequence.The dominant sequence for the TZ is the T2-weighted imaging (T2WI).A mimic or pitfall affecting a particular zone on its dominant sequence has a higher potential of causing diagnostic confusion within that zone.

FNs Related to MRI Technique
Pitfalls related to the MR technique are primarily seen in association with diffusion imaging.While allowing rapid acquisition, DWI also warrants technique optimization due to extreme sensitivity to susceptibility and motion.It is often performed as an axial-free breathing echo-planar technique combined with spectral fat saturation.
• Susceptibility and motion artifacts: DWI acquisition with echo-planar imaging (EPI) involves rapidly changing strong magnetic field gradients.This can induce super-imposed magnetic fields via the induction of eddy current, which can ultimately cause anatomical distortion of the prostate structures compared to the T2WI.Sources of magnetic field heterogeneity, such as gas, calcium, or metal in the prostate region, worsen this problem (►Fig.1A-F).This can present as stretching and distortion of the prostate boundary, surgical capsule, or focal lesions, along with shape deformation and spatial misregistration.Similar anatomical distortion can occur with motion, often related to rectal peristalsis, affecting the T2WI and DWI sequences.

Pearls
Image acquisition, particularly DWI, can be optimized to mitigate the above-described artifacts.The phase encoding direction can be switched routinely to side to side for DWI to avoid propagation of the artifacts related to rectal stool/gas onto the prostate (►Fig.1G-J).This contrasts the routinely performed anterior to posterior phase encoding in the pelvic MRI based on relatively shorter pelvic dimension than side to side.Patients may also be encouraged to evacuate the rectum or enema may be considered just before the MRI to reduce the artifacts from gas, stool, and peristalsis.Occasionally, placing rectal catheters to remove gas or administering water-based gel can help if susceptibility from stool or gas still degrades the examination.Although bowel peristalsis can be reduced by using antispasmodic agents such as glucagon or hyoscyamine, their use is not essential; it needs to be weighed against the increased costs and potential side effects. 6Utilizing 1.5-T MRI instead of the 3-T MRI can specifically be considered in patients with pelvic metal prostheses (e.g., hip arthroplasty) to reduce the associated susceptibility artifacts.
• Improper b-value selection for DWI: While b-values of up to 1,000 s/mm 2 are often sufficient in most DWI assessments of abdominopelvic organs, the background benign prostate parenchyma is not sufficiently suppressed on such bvalues to allow optimal neoplasm detection.The utility of high b-values in the 1,400 to 2,000 s/mm 2 range has been demonstrated in optimizing the conspicuity of csPCa, 7 most of which restrict diffusion, allowing them to appear sufficiently hyperintense relative to the background parenchyma.

Pearls
A high b-value (1,400 s/mm 2 ) image set is mandatory per the PI-RADS committee recommendations, with a b-value of 1,600 to 2,000 s/mm 2 often preferred.• Suboptimal viewing of DWI/ADC: The PI-RADS criteria for DWI/ADC maps are based on qualitative visual assessment.Optimal assessment requires that these sequences be viewed with consistently similar contrast.Notably, the default window settings on the picture archiving and communication system (PACS) may not allow the tumor to be sufficiently conspicuous from the background parenchyma.

Pearls
Due to the variable DWI acquisition technique, the optimal window settings may need to be determined for each system.The window level and width settings of 1,400/1,400 for the ADC map may be used as a starting point, but this may not be ideal for all systems.The levels should be adjusted to visualize csPCa as markedly hyperintense on high b-value DWI and markedly hypointense on ADC maps.A quantitative ADC threshold of less than 750 to 900 μm 2 /s may help differentiate malignant from benign tissue in PZ (lower value favoring csPCa). 8However, the routine clinical utility of such thresholds is limited by the potential technical variability of ADC and overlap of ADC between benign and malignant lesions.
• Artifacts related to the use of endorectal coil (ERC): Using an ERC is optional based on the PI-RADS recommendations.Surface coils with a sufficiently high number of external phased array coil elements and radiofrequency (RF) channels (16 or more) can provide adequate signal-to-noise ratio (SNR)/image quality without ERC, even with 1.5-T MRI.However, ERC can increase the SNR at any MR field strength.It can be advantageous, such as in some older  10 In some cases, mucinous adenocarcinoma can present as isointense with the normal PZ tissue. 11Diagnosis remains challenging even with biopsy due to the paucity of cells obtained.Therefore, it is crucial to be aware of this rare variant and recognize the appearance of this entity on MRI.

Pearls
The typical appearance of multilocular cystic mass with T2 hyperintensity and dynamic contrast-enhanced (DCE) MRI showing enhancing septations should raise suspicion for this rare subtype, even without diffusion restriction (►Fig. 2).
• Cribriform adenocarcinoma: Typical acinar morphology is observed in over 90% of PCa. 9 However, a predominant cribriform morphology exists among the spectrum of variants of prostatic adenocarcinoma and impairs csPCa detection. 12In a study, predominant cribriform morphology accounted for 22% of patients with csPCa that had negative mpMRI, 5 with another study demonstrating that only 17% of cribriform histologic tumor types were detected on imaging. 9Yet others have suggested a high detection rate of PCa with cribriform patterns. 13Our own experience suggests that these lesions may be difficult to detect, particularly if the technique is not optimal or if the reader has limited experience.This difficult detection is possibly related to sparse cellular structure with higher water content, leading to higher T2 signal and less diffusion restriction (►Fig.3).Detection is important given recent evidence of cribriform morphology to be an aggressive PCa pattern with higher cancer-specific mortality. 9

Pearls
While no distinctive imaging features have been conclusively associated with this histologic subtype, some have suggested lower ADC values than tumors without these patterns. 10owever, this is yet to be confirmed and contrasts the findings mentioned above from others, suggesting less diffusion restriction of these lesions.
• Diffuse high-grade tumor: The appearance of diffuse highgrade cancer can be variable depending on cancer grade but can mimic prostatitis.

Pearls
Diffuse high-grade cancer tends to have lower signal intensity (SI) on T2WI and more marked diffusion restriction, resulting in lower SI on ADC and higher SI on DWI, compared to prostatitis.Another finding suggestive of infiltrative highgrade cancer is the loss of normal zonal anatomy with the obliteration of the surgical capsule (►Fig. 4).

Pitfalls Causing False Positives (FPs)
FPs Related to Benign Pathologies • Acute or chronic prostatitis: This is the most common cause of FP finding for PCa. 14Prostatitis can be acute or chronic and the signal abnormalities may suggest csPCa on any of the mpMRI sequences.They may include an area of hypointensity on T2WI, often in the PZ close to the anatomical capsule, with a corresponding area of hypointensity on the ADC map and early arterial enhancement on DCE MRI (►Fig. 5).

Pearls
Morphology of the signal abnormality is often helpful in classifying an otherwise suspicious signal abnormality as secondary to a benign inflammatory process.Inflammation or postinflammatory changes are usually linear, wedgeshaped, or non-mass-like instead of rounded, defined, masslike areas of signal abnormality exhibited by the PCa.Distribution of the signal abnormality is often also helpful, with diffuse involvement suggesting inflammatory etiology as opposed to a focal signal abnormality typically seen with PCa.
Another feature to keep in mind is the orientation of the abnormal signal.While the linear/bandlike shape of signal abnormality favors benignity, it should bridge the anatomical and surgical capsule rather than paralleling the capsule.Despite these differences, appearances can overlap, and sometimes, a confident differentiation between inflammation-related changes and csPCa may be difficult (►Fig. 5).In such instances, a reasonable approach may involve a targeted biopsy and a repeat MRI after 6 months or greater if the targeted biopsy demonstrates benign pathology.A decision may then be made regarding continued imaging surveillance or repeating the biopsy based on the appearance on follow-up imaging.
• Granulomatous prostatitis: While most cases of granulomatous prostatitis are idiopathic, we most likely see it secondary to intravesical bacillus Calmette-Guerin (BCG) therapy for urinary bladder cancer. 15On mpMRI, this can demonstrate a discrete masslike appearance with  marked signal abnormalities, including hypointensity on T2WI and ADC, hyperintensity on DWI, and possibly periprostatic fat stranding mimicking extracapsular extension, falsely suggesting csPCa.The signal abnormalities are often more pronounced than those with other forms of infectious prostatitis, 16 such that often a biopsy is necessary to rule out malignancy (►Fig.6A-E).

Pearls
A helpful clue that, if present, may suggest this diagnosis and avoid biopsy includes the presence of peripheral enhancement with the central nonenhancing necrotic region (►Fig.6F-I).
While a masslike appearance of granulomatous prostatitis without central necrosis can be exceedingly difficult to differentiate from csPCa and may require histopathologic confirmation, a history of intravesical BCG therapy may be a clue toward diagnosing granulomatous prostatitis by increasing the pretest probability of this benign diagnosis.
• Prostatic abscess: A prostatic abscess may complicate acute bacterial prostatitis with a distinct clinical presentation of fever, dysuria, and digital rectal examination (DRE) eliciting tenderness even with gentle palpation, often with fluctuance.
If imaging is desired based on clinical suspicion of complicated bacterial prostatitis, transabdominal ultrasound is preferred.The appearance is typical of abscesses at other sites, demonstrating a complex collection within the prostate exhibiting peripheral vascularity.If mpMRI is performed, imaging appearance of a typical abscess, including a circumscribed region of marked hyperintensity on high b-value DWI, with corresponding marked hypointensity on ADC, 17 may mimic csPCa based on PI-RADS assessment.

Pearls
The first clue suggesting a diagnosis of a prostatic abscess would be a typical clinical picture as described earlier.The imaging clues to diagnosing an abscess rather than PCa include hyperintense or heterogeneous hypointense signal on T2WI and peripheral enhancement on DCE (►Fig. 7).
• Stromal BPH nodules in TZ: Benign prostatic hyperplasia (BPH) results from benign proliferation of prostatic epithelial and stromal cells in the TZ. 18The SI of BPH nodules on T2WI depends on the proportion of stromal versus glandular elements.A greater percentage of glandular elements leads to higher SI on T2WI related to higher fluid within glandular secretions. 19,20BPH nodules may also often demonstrate a mixed stromal and glandular appearance reflected by a heterogeneous signal on T2WI.
A characteristic feature of stromal BPH nodules is their T2 hypointensity, unlike the T2 hyperintense glandular BPH nodules.This hypointense appearance on T2WI can cause the stromal BPH nodules in the TZ to be spuriously deemed highly suspicious for csPCa.This problem is compounded by the often-heterogeneous multinodular appearance of the TZ in patients with BPH, which generally makes detecting TZ tumors more difficult than PZ tumors. 21Up to 30% of PCas arise from the TZ 22 and are also more frequently missed during standard prostate biopsies, 23 necessitating their accurate identification on mpMRI.

Pearls
An important feature favoring BPH nodules over the TZ PCa is the presence of discrete margins and an oval or spherical shape.Furthermore, a T2 hypointense capsule and T2 hypointensity lower than expected from PCa can provide additional clues (►Fig. 8).Internal foci of T2 hyperintensity may be observed in some stromal BPH nodules, along with only mild diffusion restriction, sometimes only on ADC maps or DWI.Additionally, the enhancement of these nodules is typically similar to other benign nodules within the prostate, rather than appearing earlier and more robust than the rest of the prostate, which is characteristic of PCa.This is usually in contrast to the noncircumscribed or lenticular shape, or irregular margins seen with TZ malignancy. 24,25TZ malignancy can also be diagnosed when a clearly invasive behavior, such as extension into anterior fibromuscular stroma (AFS) or extraprostatic extension (EPE) is noted.
• Extruded BPH nodules in the PZ: Extruded BPH nodules in the PZ present a unique diagnostic challenge.These nodules result from the extension of BPH tissue from the TZ into the PZ and often abut the surgical capsule.They appear as focal nodules, often demonstrating hypointensity on T2WI that can show hyperintensity on high bvalue DWI, hypointensity on ADC, and early enhancement.This appearance, especially that of a diffusionrestricting lesion in the PZ, mimics csPCa based on PI-RADS assessment.

Pearls
The presence of a capsule is extremely helpful in characterizing a typical BPH nodule.Another pearl is to utilize multiplanar T2WI, particularly sagittal and coronal planes.This can aid in identifying the capsule surrounding the nodule and demonstrate the continuity of the suspected signal abnormality with the BPH nodule arising from the TZ.Additionally, hyperintense foci on T2WI corresponding to the dilated acini may suggest extruded BPH nodules over PCa.These findings can increase diagnostic confidence and prevent overdiagnosing ectopic BPH nodules as malignancy (►Fig.9A-F).
A pseudo-capsule can rarely form around PCa related to compressive or reactive peritumoral fibrosis, and can be associated with high-grade and stage disease. 26Therefore, an atypical appearance of a large encapsulated PZ lesion, especially one that does not appear to be contiguous with the TZ, should raise suspicion (►Fig.9G-K).
• Prostatic calcifications: Incidence and degree of calcifications within the prostate gland increase with age, and they are prevalent after 50 years of age.They can be idiopathic or secondary to prostatitis, BPH, trauma (including iatrogenic such urethral stents or radiation therapy), diabetes mellitus, or even PCa. 27They are most often believed to arise from the calcification of the corpora amylacea that, in turn, may be related at least in part to the prostatic secretions.Prostatic calcifications are often more easily visualized on computed tomography (CT) than on MRI.When seen on MRI, they exhibit signal void on all sequences due to the diamagnetic effect of these calcifications.The resulting marked hypointensity on T2WI and ADC can mimic csPCa within the TZ or PZ.

Pearls
In addition to the dark signal on T2WI and ADC, these calcifications appear dark on DWI at all b-values, suggesting their true nature.Moreover, they exhibit a complete absence of enhancement.While these calcifications can be solitary, they occur more often in clusters, which can be an additional clue to their presence.Correlation with CT, if available, can be conclusive in their characterization (►Fig. 10).FPs Related to Normal Anatomical Structures  or TZ (given the hypointensity on T2WI).Assessing this structure can sometimes present a challenge to even experienced radiologists.

Pearls
A typical symmetric appearance of the CZ at the posterior prostate base and mid-gland, with no early arterial enhancement on DCE, helps differentiate this normal structure from PCa (►Fig.11A-D).Also, look for a mustache or dumbbell shape on the axial images, a teardrop or triangular shape on coronal images, and the typical location surrounding the ejaculatory ducts in the posteromedial base to the mid-gland PZ.CZ accounts for less than 5% of PCa cases.Focal early enhancement within the CZ or an asymmetry between the CZ on T2WI, ADC maps, or high b-value DWI that cannot be attributed to BPH should raise suspicion for PCa 29 (►Fig.11E-H).
• Anterior fibromuscular stroma (AFS): This tissue layer in the anterior prostate lacks glandular tissue and thus cannot give rise to PCa.Typically, it exhibits bilateral symmetry with a crescentic shape and appears as a hypointense region on T2WI, similar to pelvic floor musculature.There is no diffusion restriction, exhibiting hypointensity on both DWI and ADC map, with absent early arterial enhancement.In certain cases, AFS may show thickening, particularly in patients lacking significant TZ hypertrophy.

Pearls
Although PCa does not arise from AFS, cancer from PZ or TZ can involve AFS.Suspicious findings may appear hyperintense on T2WI and hypointense on ADC relative to pelvic floor musculature (typically more hypointense than the normal AFS), with hyperintensity on DWI.Other suspicious findings include focal masslike or asymmetric enlargement and early arterial enhancement.If this is suspected, the PI-RADS criteria should be applied utilizing the criteria for the respective zones from which the lesion is suspected to arise (►Fig.12).
• Thickened surgical/pseudo-capsule: The prostatic surgical capsule, or pseudo-capsule, refers to the fibromuscular layer between the prostate's PZ and TZ.While this layer is loose and indistinct in normal prostate gland without BPH, it can be thickened in response to pressure from BPH nodules arising from the TZ. 30 On T2WI, it appears dark, and exhibits decreased ADC values, similar to csPCa. 31arls A noteworthy characteristic is its bandlike or crescentshaped appearance surrounding the TZ.Given the anatomical distortion of the ADC maps, the correlation of the hypointensity on the ADC map to the T2WI showing the typical appearance and location of the surgical capsule can be extremely helpful in preventing mislabeling this anatomical structure as csPCa (►Fig. 13).• Neurovascular bundle: The neurovascular bundles course along the posterolateral aspect of the prostate capsule on both sides, positioned at the 5 o'clock and 7 o'clock positions. 32Due to its proximity to the PZ and imaging characteristics, including hypointensity/signal void on the T2WI and ADC map, the neurovascular bundle might be mistaken for a suspicious lesion within the PZ.It can also induce anatomical distortion of the prostate boundaries, potentially leading to misinterpretation as a mass lesion within the periprostatic space.

Pearls
The typical location along the posterolateral margins, a rounded contour on a single axial slice, tubular morphology across consecutive slices, and bilaterally symmetric bundles assist in resolving the suspicion. 33The neurovascular bundles are often better assessed on T1WI, where they are recognized as oval (when viewed axially) or linear (when viewed in the long axis), hypointense structures surrounded by hyperintense fat.Neurovascular bundle invasion on T1WI may appear as asymmetric enlargement with loss of the intervening periprostatic fat plane or as gross tumor extension. 34Periprostatic venous plexus: The periprostatic venous plexus is located close to the prostate capsule, and distinguishing it from the PZ on mpMRI can occasionally present a challenge due to low SI related to flow void on T2WI.Occasionally, it may also exhibit mildly restricted diffusion on EPI (►Fig.14), and along with focal enhancement indicating abnormal enhancement kinetic, these findings may mimic csPCa.

Pearls
Awareness of the presence and expected location of this normal anatomical structure around the prostate should lead to evaluating the postcontrast sequence.Delayed enhancement would then point toward a periprostatic venous structure instead of a tumor. 35s Related to Postintervention Changes • Postbiopsy hemorrhage: Often, patients with suspected PCa have undergone prostate biopsy before MRI evaluation.This may cause hemorrhage, producing MRI signal abnormalities within the prostate.Hemorrhage-related signal changes include, more commonly, T2 hypointensity, but depending on the age of the hemorrhage, they can also appear to restrict diffusion. 36These changes overlap with the appearance of PCa on T2WI and DWI/ADC.Moreover, the T1 hyperintense appearance of the hemorrhage can interfere with assessing the region on DCE, wherein a tumor would be expected to demonstrate early enhancement.The normal prostate gland produces citrate, which possesses anticoagulant properties.Consequently, the hemorrhage occurring after a biopsy is more extensive and lingers longer in the region of the normal PZ. 37On the other hand, citrate levels are diminished in the areas affected by PCa, 38 resulting in a more rapid resolution of hemorrhage in the tumor due to the absence of the anticoagulant effect. 39

Pearls
It is recommended to perform mpMRI at least 6 to 8 weeks after the biopsy to allow resolution of the hemorrhage, although hemorrhage-related changes can persist for up to 4.5 months. 40csPCa exhibits more discrete masslike abnormality, demonstrating more marked SI alterations on DWI and DCE than postbiopsy hemorrhage.The hemorrhage exclusion sign can be helpful and describes the presence on T1W images of either a well-defined area of hypointensity in the PZ surrounded by an area of hyperintensity on at least one side or diffuse hyperintensity within a PZ region with an abrupt cutoff outlining a lesion.When this sign is present with a corresponding area of hypointensity on T2WI, it can strongly suggest csPCa 36 (►Fig.15).T1WI subtraction postcontrast images or perfusion maps can help differentiate actual enhancing lesion from intrinsically T1 hyperintense hemorrhagic focus.
• Postprostatic urethral lift (UroLift) procedure: One of the newer minimally invasive approaches to treat symptomatic BPH includes a prostatic urethral lift with a UroLift procedure.It involves permanent metallic implants that retract the prostatic tissue, creating a channel for urine passage. 41This MRI conditional device is safe for use with field strengths of up to 3 T. 42 However, it causes artifacts that negatively affect the MRI quality, with a recent study demonstrating obscuration of a third of the gland on DWI. 43

Pearls
The potential to significantly affect the evaluation for PCa with MRI should be a part of the preprocedure discussion with the patient.A baseline MRI prior to the UroLift procedure to assess for PCa could also be considered.On MRI following a prior UroLift procedure, the UroLift device should not be mistaken for other linear metallic prostate implants utilized in PCa management.Brachytherapy seeds are usually more numerous than the UroLift tabs with distribution throughout the prostate, unlike UroLift tabs that are ideally placed anterolaterally at the 10 o'clock and 2 o'clock positions in the prostate (►Fig. 16).Fiducial markers used in external beam radiation therapy for PCa are similar in number to UroLift tabs but are shorter and thicker when compared to UroLift tabs. 43

Pitfalls during Staging
• Assessment of extra-prostatic extension (EPE): EPE assessment is crucial preoperatively due to its association with prognosis after radical prostatectomy and for deciding whether to spare the neurovascular bundles at surgery. 44,45 newly proposed MRI-based EPE grading system showed an accuracy of 77 to 81%. 44,46This is a three-tier system with grade 0 defined as having no suspicion of pathologic EPE; grade 1 with either capsular contact length (CCL) of 15 mm or capsular irregularity/bulge; grade 2 with both CCL of 15 mm and capsular irregularity/bulge; and grade 3 with frank EPE visible at MRI.Other specific findings of EPE include asymmetry/thickening of the neurovascular bundle and blunting of the rectoprostatic angle.However, several pitfalls exist that require careful consideration.

Pearls
A low threshold for suspecting EPE is recommended, particularly at the apex, where the lack of a capsule poses a higher risk.Apical lesions have implications for radiation therapy, and are at risk of positive margins and recurrence after radical prostatectomy.Lesion proximity to the urethra should also be noted since urethral invasion is a concern (PI-RADS 5) and poses a surgical risk of injury to the external urinary sphincter, affecting urinary competence (►Fig. 17).
It should be remembered that MRI underestimates tumor boundaries compared to pathology, 47 with DCE performing better than T2WI and DWI for tumor volume estimation.
T2WI is key for evaluating a seminal vesicle (SV) that typically demonstrates a high signal on T2WI when  distended with seminal fluid.SV invasion, either as tumor growth along the SV or extension from the prostate base, defines the T3b TNM stage.Tumor infiltration or focal tumor deposits exhibit T2 hypointensity. 48However, caution must be exercised as nondistended SV (such as from recent ejaculation) or benign pathologies, including inflammation and related amyloidosis, 49 can falsely mimic invasion on T2WI due to a dark signal (►Fig.18A-E).Corresponding diffusion restriction and focal enhancement can increase confidence in characterizing T2 signal abnormality as pathologic (►Fig.18F-J).
• Lymph node assessment: Pelvic lymph node dissection (PLND) is the most reliable method for documenting nodal invasion. 50Prostate MRI allows pelvic and retroperitoneal lymph node evaluation for metastatic involvement but relies solely on morphological features.A common criterion for identifying metastatic pelvic nodes   is a short-axis diameter greater than 0.8 cm.A metaanalysis utilizing size criteria of short-axis diameter greater than 1 cm for oval and greater than 0.8 cm for round nodes with cross-sectional imaging demonstrated a pooled sensitivity of around 39 to 42% and a specificity of 82% for detecting positive nodes. 51Nevertheless, met-astatic and benign reactive lymph nodes show significant overlap in size (►Fig.19).

Pearls
Other morphological signs of malignancy may be helpful, including irregular borders indicating extranodal extension,   a round or irregular shape, and the absence of a fatty hilum. 52 combination of morphological criteria and a DWI sequence demonstrated a specificity of 94% and sensitivity of 41% for detecting lymph node metastasis. 53While malignant nodes may exhibit lower ADC values than benign nodes, overlap of their ADC values limits clinical utility. 54Bone assessment: Bone is the second most common site of early metastases after lymph nodes. 55Traditional bone scans have limited sensitivity and a high FN rate of up to 40%.Whole-body MRI with DWI has demonstrated greater accuracy than bone scans in detecting bone metastases. 56utine prostate MRI can also effectively detect bone metastases, 57 given that most PCa metastases arise within the pelvis and lumbar spine, often covered on prostate MRI.

Pearls
Bone metastases exhibit hypointensity on T1WI, intermediate-to hypointensity on T2WI, hyperintensity on DWI, and early, robust enhancement on DCE (►Fig. 20).MRI should be interpreted in the context of the PSA, Gleason score, and clinical stage to enhance metastatic detection and avoid unnecessary investigation of benign bone lesions, including hyperplastic hematopoietic bone marrow islands. 58The  presence of intralesional fat is associated with benignity and can be helpful.Prostate MRI assessment of bone metastases is limited by restricted coverage and varying image acquisition protocols compared to musculoskeletal MRI (e.g., noninclusion of short tau inversion recovery [STIR] on mpMRI of the prostate).Depending on the institutional protocol, bone lesions may often need to be primarily addressed in the large field-of-view T1WI, while T2WI, DWI, and DCE sequences concentrate on the prostate and SVs for optimal anatomical detail. 59

Summary
In conclusion, while the PI-RADS scoring system utilizing mpMRI has revolutionized the management of PCa, it is essential to acknowledge the limitations and pitfalls of this technique.Understanding and addressing these challenges and ongoing refinements to this system can improve utility and accuracy in detecting csPca.A summary of common entities causing FN and FP findings is summarized in ►Table 1.

Fig. 1 (
Fig. 1 (A) Axial T2-weighted image (T2WI; arrow) shows a focal lesion exhibiting moderate hypointensity in the right mid-gland prostate zone (PZ).(B) Diffusion-weighted imaging (DWI; arrow) and (C) apparent diffusion coefficient (ADC) map (arrow) show no discrete corresponding DWI hyperintense or ADC hypointense abnormality; however, it can be seen that the susceptibility artifact from rectal gas degrades the assessment of the most posterior aspect of the PZ.(D) On subsequent magnetic resonance imaging (MRI), axial T2WI (arrow) shows a similar focal lesion, which is now clearly noted to exhibit marked diffusion restriction on DWI (arrow) and (F) ADC map (arrow).Biopsy showed Gleason 3 þ 4 prostate cancer (PCa) in the right mid-gland.In another patient, there is a band of artifacts related to rectal stool and gas (white arrows) seen extending horizontally on both (G) DWI and (H) ADC.Since DWI was performed with the phase encoding direction switched side to side as part of routine prostate imaging protocol, this prevented the band of artifact from propagating onto the prostate, which could have otherwise obscured the diffusion-restricting PI-RADS 4 lesion (yellow arrows in G and H) exhibiting (I) focal hypointensity onT2WI with (J) early enhancement.The T2 hypointense area in the left PZ was considered a postinflammatory change considering the well-demarcated margins.

Fig. 5
Fig. 5 Axial (A) T2-weighted image (T2WI; arrows) shows focal moderate hypointensity in the right peripheral zone (PZ) and transitional zone (TZ) with (B) early enhancement (arrows).There is corresponding marked diffusion restriction with (C) hyperintensity on high b-value diffusion-weighted imaging (DWI; arrows) and (D) hypointensity on apparent diffusion coefficient (ADC) map (arrows).Biopsy of this PI-RADS 5 lesion revealed prostatitis.In another patient, (E) a wedge-shaped hypointensity on T2WI (arrow) also exhibits marked diffusion restriction on (F) high b-value DWI (arrow) and (G) ADC map (arrow).A biopsy of this lesion showed Gleason 3 þ 3 adenocarcinoma.Note that although the focal T2 hypointensity (arrow in E) is bandlike, it parallels the capsule rather than bridging anatomical and surgical capsules.

Fig. 6
Fig. 6 Multifocal lesions in bilateral peripheral zone (PZ) and left transitional zone (TZ) exhibiting (A) marked hyperintensity on diffusionweighted imaging (DWI; arrows), (B) marked hypointensity on apparent diffusion coefficient (ADC; arrows), and (C) positive DCE (arrows).(D) The lesions are homogenously hypointense on T2-weighted image (T2WI; arrows), with the left TZ lesion exhibiting moderate lenticular hypointensity.(E) There is broad-based capsular abutment, multifocal capsular bulging, and extracapsular extension cranially as seen on coronal T2WI (arrows), consistent with PI-RADS 5, and suspicious for multifocal clinically significant prostate cancers.Subsequent biopsy findings were consistent with nonspecific granulomatous prostatitis.In a different patient with a history of intravesical Bacillus Calmette-Guérin (BCG) therapy for urinary bladder cancer, (F) axial T2WI (arrow) shows a moderately hypointense lesion in the left PZ, with (G) corresponding marked diffusion restriction on high b-value DWI (arrow) and (H) ADC map (arrow).(I) However, note the absence of enhancement of the lesion (arrow), suggesting necrosis, which suggests BCG-related granulomatous prostatitis in this patient.

Fig. 9 (
Fig. 9 (A-F) A focal lesion in the left apical peripheral zone (PZ) exhibits (A) marked hyperintensity on high b-value diffusion-weighted imaging (DWI; arrow) and (B) marked hypointensity on apparent diffusion coefficient (ADC; arrow), (C) with early dynamic contrast enhancement, meeting Prostate Imaging Reporting and Data System 4 (PI-RADS 4) criteria.(D) Corresponding hypointense signal abnormality on axial T2-weighted image (T2WI (arrow) shows hyperintense foci.Moreover, there is a capsule surrounding the nodule with continuity of the focal signal abnormality with the transitional zone (TZ), better seen on the (E) coronal (arrow) and (F) sagittal T2WI (arrow).These findings suggest extruded benign prostatic hyperplasia (BPH) nodule in this otherwise PI-RADS 4 lesion that showed benign prostatic tissue on a subsequent biopsy.(G) In another patient, axial T2WI shows a large moderately hypointense focal lesion involving most of the right PZ surrounded by a thin rim of low T2 signal intensity, giving the appearance of an extruded BPH nodule.However, some atypical features exist, as the nodule is not completely encapsulated.The lesion also has marked associated diffusion restriction on (H) high b-value DWI (arrow) and (I) ADC map (arrow).(J) Coronal T2WI and (K) axial T2WI show additional extension of the lesion crossing midline to the left posteromedial peripheral zone at the

Fig. 10
Fig. 10 Heterogeneous areas of signal loss on (A) axial (arrows) and (B) coronal (arrows) T2-weighted image (T2WI) within the posterior right greater than the left transition zone, with (C) a complete absence of enhancement on DCE (arrows).A corresponding dark signal on (D) apparent diffusion coefficient (ADC; arrows) and (E) high b-value diffusion-weighted imaging (DWI; arrows) suggest signal void rather than true diffusion restriction.(F) Correlation with computed tomography (CT) confirmed corresponding calcifications (arrows).

Fig. 12 (
Fig. 12 (A) Axial (arrow) and (B) sagittal T2-weighted images (T2WIs; arrow) show focal masslike asymmetric enlargement of the left anterior fibromuscular stroma (AFS) appearing to arise from the left anterior peripheral zone (PZ).The lesion exhibits a T2 intermediate signal that is higher than the normal T2 dark AFS.There are corresponding areas of diffusion restriction with (C) marked hyperintensity on diffusion-weighted imaging (DWI; arrow) and (D) marked hypointensity apparent diffusion coefficient (ADC) map (arrow), with (E) asymmetric early arterial enhancement (arrow).This was suspicious for a Prostate Imaging Reporting and Data System 5 (PI-RADS 5) lesion from the left anterior prostate zone (PZ) invading the AFS.Subsequent radical prostatectomy revealed a Gleason score 4 þ 3 prostatic adenocarcinoma mostly present in the left quadrants.

Fig. 11 (
Fig. 11 (A) Axial T2-weighted image (T2WI; arrow) shows triangular hypointense area in the posterior midline peripheral zone (PZ) at the gland base with diffusion restriction on (B) high b-value diffusion-weighted imaging (DWI; arrow) and (C) apparent diffusion coefficient (ADC) map (arrow), (D) without early enhancement (arrow).This location and appearance are consistent with the normal central zone and should not be mistaken for a suspicious lesion, although it meets the criteria for a Prostate Imaging Reporting and Data System 4 (PI-RADS 4) lesion.Contrast this in another patient with (E) axial T2WI showing focal asymmetry in the left central zone (CZ) with a hypointense signal similar to the remaining CZ (arrow).The lesion also exhibits marked diffusion restriction on (F) high b-value DWI (arrow) and (G) ADC map (arrow).(H) Note the early arterial enhancement (arrow).Biopsy showed left prostate base Gleason score 3 þ 4 adenocarcinoma.

Fig. 13 A
Fig. 13 A focal region in the left peripheral zone (PZ) appears mildly hyperintense on (A) high b-value diffusion-weighted imaging (DWI; arrow) and (B) markedly hypointense on apparent diffusion coefficient (ADC; arrow), raising suspicion for at least a Prostate Imaging Reporting and Data System 3 (PI-RADS 3) lesion.However, correlating with (C) axial and (D) sagittal T2-weighted image (T2WI), the lesion corresponds to the anatomical region of the surgical capsule between the transitional zone (TZ) and the prostate zone (PZ) with a crescent-shaped appearance (arrow), allowing more confident recognition as the thickened surgical capsule.

Fig. 14
Fig. 14 Diffusion restriction is noted along the lateral margins of the bilateral peripheral zone (PZ) on (A) the high b-value diffusion-weighted imaging (DWI; arrows).(B) On correlating with T2-weighted image (T2WI; arrows), this corresponds to the bilateral periprostatic venous plexus.

Fig. 16 (
Fig. 16 (A) Axial T2-weighted image (T2WI; arrows) shows susceptibility artifacts in the transition zone (TZ) due to the stainless-steel inner tabs of the UroLift device; outer nitinol tabs are much less conspicuous with less susceptibility artifacts than inner tabs.(B) Corresponding diffusion-weighted imaging (DWI), (C) apparent diffusion coefficient (ADC) map, and (D) and axial in-phase T1WI show a larger degree of susceptibility artifact, with a large rounded signal void (arrows) almost obscuring the entire TZ, especially on diffusion imaging.(E) Coronal and (F) axial computed tomography (CT) images better demonstrate the device with inner stainless-steel tabs (solid arrows) exhibiting relatively greater attenuation than outer nitinol tabs (dashed arrow).

Fig. 15 A
Fig. 15 A focal lesion in the left peripheral zone (PZ) exhibiting (A) marked hyperintensity on high b-value diffusion-weighted imaging (DWI; arrows) and (B) marked hypointensity on apparent diffusion coefficient (ADC) map (arrows), (C) with patchy areas of homogenous moderate hypointensity on T2-weighted image (T2WI) that are nonspecific in the setting of recent biopsy.(D) On pre-T1WI, a diffuse hyperintense region of hemorrhage is present throughout the right prostate zone (PZ), with an abrupt cutoff to a region of no hemorrhage (arrows, hemorrhage exclusion sign) and corresponding focal moderate hypointensity on T2WI (arrows in C), increasing confidence for cancer detection.Note that DCE assessment of the focal lesion was limited by the abutting regions of (E) intrinsic T1 hyperintensity (arrows), (F) but the perfusion map clearly shows positive DCE corresponding to the focal suspicious region (red areas marked by arrows).The lesion is thus Prostate Imaging Reporting and Data System 5 (PI-RADS 5).Prostate biopsy showed Gleason score 4 þ 4 adenocarcinoma in the left apex and mid-gland with benign tissue in the right gland.

Fig. 17
Fig. 17(A) Axial T2-weighted image (T2WI) shows a moderately hypointense focal lesion (arrow) in the left apical peripheral zone.The lesion is markedly diffusion restricted on high b-value (B) diffusion-weighted imaging (DWI; arrow) and (C) apparent diffusion coefficient (ADC) map (arrow), and (D) dynamic contrast enhancement is positive (arrow).Slightly higher (E) axial T2WI (arrow) shows evidence of extraprostatic extension into the posterior periprostatic fat with asymmetric thickening and (F) enhancement of the left neurovascular bundle (arrow).The lesion involves the prostatic urethra and the external sphincter (arrowheads in A and D).This lesion is thus Prostate Imaging Reporting and Data System 5 (PI-RADS 5).
Fig. 17(A) Axial T2-weighted image (T2WI) shows a moderately hypointense focal lesion (arrow) in the left apical peripheral zone.The lesion is markedly diffusion restricted on high b-value (B) diffusion-weighted imaging (DWI; arrow) and (C) apparent diffusion coefficient (ADC) map (arrow), and (D) dynamic contrast enhancement is positive (arrow).Slightly higher (E) axial T2WI (arrow) shows evidence of extraprostatic extension into the posterior periprostatic fat with asymmetric thickening and (F) enhancement of the left neurovascular bundle (arrow).The lesion involves the prostatic urethra and the external sphincter (arrowheads in A and D).This lesion is thus Prostate Imaging Reporting and Data System 5 (PI-RADS 5).

Fig. 18 (
Fig. 18 (A) Axial (arrows) and (B) coronal T2-weighted images (T2WIs; arrows) demonstrate diffuse T2 hypointensity of the seminal vesicles in this patient with Gleason 3 þ 3 disease in the peripheral zone (PZ; not shown).The appearance of the seminal vesicles is characteristic of amyloidosis, which was noted on a prior prostate biopsy.Note the (C) absence of enhancement (arrows) and (D,E) diffusion restriction (arrows) in the region of seminal vesicle signal abnormality.In contrast, in a different patient, (F) axial (arrow) and (G) coronal T2WI (arrow) show irregular T2 hypointense regions within the right greater than the left seminal vesicles.There are corresponding areas of (H) enhancement (arrow) and (I,J) diffusion restriction (arrow) concerning seminal vesicle extension in this biopsy-proven Gleason 5 þ 4 disease involving the bilateral prostate base (not shown).

Fig. 20 (
Fig.20(A) Axial and (B) coronal T2-weighted (T2W) images show a hypointense focal marrow replacing lesion involving the left acetabulum (arrow).(C) On T1WI (arrow), the lesion exhibits homogenous hypointensity to a greater degree than adjacent skeletal muscles.The lesion also exhibits strong enhancement on (D) the postcontrast image (arrow) and (E) diffusion restriction on diffusion-weighted imaging (DWI; arrow) and (F) apparent diffusion coefficient (ADC) map (arrow), which is highly suspicious for metastasis.
Pearls and Pitfalls in Applying PI-RADS 2.1 Pandey et al.
Pearls and Pitfalls in Applying PI-RADS 2.1 Pandey et al.
1.5-T MR systems, with larger patients where SNR is limited with external coils alone or where particularly high spatial resolution is desired for staging purposes.

Table 1
Pitfalls in using PI-RADS for imaging analysis Swap phase and frequency direction • Good bowel preparation prior to the examination • Antispasmodic agents • Using 1.5 T in patients with pelvic metal prosthesis • PSA values 2. Improper high b-value selection on high DWI images • Use high value 1,400 s/mm 2 • Preferably greater than 1,600 s/mm 2 • Prefer not to use high-value DWI for ADC map generation 3. Suboptimal viewing of DWI/ADC • Manually adjust the optimal viewing window level and width to enhance the contrast-to-noise ratio between normal glandular tissue and tumor 4. Improper use of endorectal coil • Try to perform the examination with optimized parameters without using the endorectal coil whenever possible • When used, proper positioning of the coil is very important 5. Unusual pathologic variants of prostate cancer • The reader should be familiar with the imaging appearance of the pathologic variant of primary adenocarcinoma • These entities should be suspected based on imaging appearance even in light of noncongruent PSA values Look for typical imaging patterns that suggest inflammation; segmental, linear, or wedge-shaped sharply marginated areas of abnormality • No significant change from the prior examination with negative fusion biopsy on the prior examination 2. Granulomatous prostatitis • History of BCG therapy for superficial bladder cancer • Ringlike enhancement on postcontrast images 3. Stromal BPH nodule in TZ • Look for well-demarcated margins and capsules • Symmetric enhancement compared to the rest of the TZ 4. Extruded BPH nodules in PZ • Presence of well-demarcated capsule; maintained continuity with TZ if present Abbreviations: ADC, apparent diffusion coefficient; BCG, bacillus Calmette-Guerin; CT, computed tomography; DWI, diffusion-weighted imaging; PI-RADS, Prostate Imaging Reporting and Data System; PSA, prostate-specific antigen; TZ, transitional zone.Journal of Gastrointestinal and Abdominal Radiology ISGAR © 2024.The Author(s).