Key words
prostate - PI-RADS - prostate cancer - uroradiology - MR imaging
Introduction
After lung cancer, prostate cancer is globally the second most common malignant non-cutaneous
tumor disease in men. With 1,276,105 new cases worldwide in 2018, it was responsible
for 3.8 % of all cancer-related deaths in men [1]. In recent years, multiparametric magnetic resonance imaging (mpMRI) of the prostate
has become significantly more important for the diagnosis of this disease and has
become an important part of current European guidelines [2]. mpMRI of the prostate not only improves diagnosis and local staging but also facilitates
prognosis prediction and treatment individualization [3]
[4].
In 2012, the guidelines of the European Society of Urogenital Radiology (ESUR) provided
for the first time consensus-based recommendations for standardizing the acquisition,
evaluation, and reporting of prostate MRI [5]. These guidelines, which are named after the Prostate Imaging – Reporting and Data
System (PI-RADS), were first revised in 2015 (PI-RADS 2.0) [6].
PI-RADS 2.0 simplified the concept of the dominant sequence for the first time (DWI
in the peripheral zone, T2 in the transition zone). On the whole, the new version
was well received. However, studies continue to have a relatively low detection rate
for carcinomas in the transition zone compared to the peripheral zone and a relatively
high interreader variability, which can be attributed in part to ambiguous formulations
in PI-RADS version 2.0 [7]
[8]
[9]
[10]
[11]. Under consideration of these issues and to incorporate technical advances, the
AdMeTech Foundation, the ESUR, and the American College of Radiology (ACR) as the
PI-RADS Steering Committee published PI-RADS version 2.1 in the spring of 2019 [12].
The updated version implements changes to minimize uncertainties in scoring and to
reduce interreader variability without fundamentally changing the established application
algorithm [13]. This article provides an overview of the most important changes in PI-RADS version
2.1 and discusses possible clinical implications.
Changes regarding image acquisition
Changes regarding image acquisition
As in version 2.0, revised PI-RADS version 2.1 also recommends the use of multiparametric
MRI (mpMRI) from T2-weighted (T2w), diffusion-weighted imaging (DWI), and dynamic
contrast-enhanced (DCE) sequences. The advantages and disadvantages of biparametric
MRI (bpMRI), which does not use contrast-enhanced sequences, are discussed in the
guidelines, but the method is not recommended as the standard.
Revised version 2.1 also still recommends the use of MRI with a field strength of
1.5 and 3 Tesla. PI-RADS version 2.1 also does not provide a general recommendation
regarding the use of endorectal coils, particularly since current scanners at both
field strengths can ensure an adequate signal-to-noise ratio even without their use.
T2-weighted imaging
T2w remains the dominant sequence for evaluating the transition zone. While PI-RADS
2.0 still recommended the acquisition of T2-weighted sequences on all three standard
planes (axial, sagittal, coronal), T2w on the axial plane as well as at least one
additional orthogonal plane is sufficient according to PI-RADS 2.1. Even though PI-RADS
2.1 does not explicitly favor the sagittal or the coronal plane, the combination of
at least transverse and sagittal T2w is indirectly advocated based on the recommended
determination of the prostate volume using the sagittal plane.
Due to possible partial volume effects, the authors feel that acquisition of T2w in
all three standard planes is advantageous as long as time allows. This is particularly
true with respect to the differentiation between typical and atypical BPH nodules
based on the completeness of the T2 hypointense encapsulation of the nodule in the
transition zone (see below) as newly added to PI-RADS 2.1. In addition to the two-dimensional
T2-weighted sequences, 3D-T2w images can be useful for obtaining a better anatomical
overview. However, the soft-tissue contrast is inferior to 2D-T2w images in some cases.
In contrast to PI-RADS 2.0, version 2.1 provides a recommendation regarding the orientation
of the axial (T2w) sequence. The orientation should either be straight axial to the
patient (regardless of the position of the prostate) or in an oblique axial plane
perpendicular to the long axis of the prostate (perpendicular to the max. extension
of the prostate between the base and the apex on the sagittal plane). The latter approach
has the advantage that it typically facilitates fusion with ultrasound images during
transrectal biopsy.
Diffusion-weighted imaging
Diffusion-weighted imaging (DWI) remains the dominant sequence for evaluating the
peripheral zone.
Since ADC values can be subject to certain fluctuations as a function of the underlying
b-values, PI-RADS 2.0 provided recommendations to keep these variations as minimal
as possible. Therefore, PI-RADS 2.0 recommended the acquisition of low b-values between
50 and 100 s/mm2 to avoid pseudoperfusion effects at a b-value of 0 s/mm2
[14]
[15]. Since the technical requirements of PI-RADS 2.0 are more difficult to meet in some
cases, PI-RADS 2.1 allows a low b-value of 0 s/mm2. To avoid kurtosis effects, 1000 s/mm2 should be the maximum b-value used to calculate ADC [16]
[17].
Apart from DWI with at least two b-values between 0 and 1000 s/mm2 for the ADC calculation, which should always be performed, it has been able to be
shown many times that “ultra-high” b-values can additionally reduce the T2 shine-through
effect and increase the contrast between tumor and normal prostate tissue [18]
[19] ([Fig. 1]). Therefore, PI-RADS version 2.1 requires an ultra-high b-value of at least 1400 s/mm2 (a) either extrapolated from the acquired lower b-value data (b0 to b1000) used to
create the ADC map or (b) additionally measured separately. Since the detection of
prostate cancer has improved significantly [20]
[21]
[22]
[23], the introduction of mandatory ultra-high b-values in PI-RADS 2.1 is definitely
a useful update.
Fig. 1 Different b-values in a biopsy-naïve 80-year-old man with a PSA of 6.01 ng/ml. Low
T2w signal a at the base in the anterior transition zone on the left with restricted diffusion
on ADC maps b. The b1000 image c shows only a slight signal elevation in the corresponding area, which can be more
clearly delimited from the surroundings in the b1400 image d. The calculated b2000 image e shows the lesion most clearly. The 1.7 cm lesion corresponds to a PI-RADS 5 finding.
The biopsy revealed a carcinoma with a Gleason score of 3 + 4 = 7a.
It is important when selecting the ultra-high b-value to take the field strength of
the MRI system into consideration. As a rule, higher b-values are possible at 3.0 T
than 1.5 T under otherwise identical conditions [24]. Clinical practice in the coming years will show whether the introduction of the
mandatory ultra-high b-value also proves to be technically feasible in the case of
less powerful MRI systems.
Dynamic contrast-enhanced sequence
For image acquisition after i. v. contrast administration, PI-RADS 2.1 recommends
the use of three-dimensional T1 sequences and highlights the advantages of a high
spatial resolution compared to 2D-T1 sequences. DCE can help to better differentiate
periprostatic veins from small carcinomas near the pseudocapsule [25].
Since temporal resolution during acquisition always entails compromises regarding
spatial resolution, version 2.1 recommends a temporal resolution of up to 15 seconds
(PI-RADS 2.0: ≤ 10s; < 7 s preferred), since the possible risk of missing early contrast
enhancement of lesions is negligible here [26]. To simplify the evaluation of dynamic contrast-enhanced sequences as much as possible,
PI-RADS recommends a simple qualitative visual evaluation of DCE and does not require
a quantitative evaluation. As in the previous version, PI-RADS 2.1 recommends a minimum
examination time of 2 minutes after the administration of contrast agent.
Multiparametric or biparametric MRI?
In recent years there has been increasing interest in biparametric MRI (bpMRI) of
the prostate, i. e., limitation to T2 and DWI sequences. Studies show that bpMRI has
yielded good results in large centers [27]
[28]. bpMRI is also faster and more cost-effective than mpMRI, a fact that must be taken
into consideration in light of the significantly increasing number of cases [29]. Moreover, since contrast agents containing gadolinium are not used, the possible
risks associated with these contrast agents are also eliminated.
In PI-RADS 2.1 the PI-RADS steering committee voices an opinion regarding the role
of bpMRI for the first time but does not recommend it for general use. One reason
for this is studies showing a higher sensitivity of mpMRI as a result of DCE-MRI [30]
[31]
[32]
[33]
[34]. This is particularly true in settings with less experienced evaluators. In addition,
bpMRI relies completely on high-quality T2 and DWI thereby losing DCE-MRI as an important
“back-up sequence” in cases of suboptimal DWI ([Fig. 2]). The elimination of DCE-MRI can also result in more lesions being classified as
PI-RADS 3 than PI-RADS 4 [35]. A high number of PI-RADS 3 lesions is problematic since this categorization does
not include a standardized approach for a further course of action. In the absence
of DCE-MRI, proponents of bpMRI are considering making the clinical management of
category 3 lesions dependent on lesion volume. Patients with a PI-RADS 3 lesion with
a volume of < 0.5 cm3 would undergo PSA follow-up and annual bpMRI examination and those with a lesion
volume of > 0.5 cm3 would undergo targeted biopsy [36]
[37].
Fig. 2 DCE-MRI as a “backup sequence” for DWI of low quality. A 77-year-old male with a
PSA of 4.7 ng/ml after two negative biopsies. Focal blurred low T2w signal a. Artifacts due to abundant air in the rectum distort diffusion imaging. While focally
reduced signal on ADC maps in this area b, the b1000 image c cannot be evaluated diagnostically for this region. In this case, a contrast-enhanced
sequence d shows corresponding focal early arterial enhancement (arrow) – the lesion is upgraded
to a PI-RADS category 4. The biopsy revealed prostate carcinoma with a Gleason score
of 3 + 4 = 7a. Note: This example clearly shows that rectal voiding prior to examination
contributes decisively to technical success and high diagnostic accuracy.
The authors agree with the recommendation of the PI-RADS steering committee that mpMRI
should be given preference when the priority of the examination is to make sure that
no clinically significant prostate cancer is missed. bpMRI should initially be reserved
for qualified centers and specialized radiologists with extensive experience and its
role should be further evaluated in studies.
Summary of the most important changes regarding image acquisition according to PI-RADS
2.1
-
T2w images should be acquired in axial orientation and at least one other orthogonal
plane.
-
Axial T2w images should be obtained either straight axial to the patient or in an
oblique axial plane perpendicular to the long axis of the prostate.
-
Acquisition of low b-values between 0–100 s/mm2 (50–100 s/mm2 is preferred) is now also possible.
-
The maximum high b-value for ADC calculation is ≤ 1000 s/mm2.
-
A high b-value of at least 1400 s/mm2 is either to be extrapolated or to be additionally measured separately.
-
After i. v. administration of contrast agent, 3D-T1w sequences are to be given preference
over 2D-T1w sequences.
-
In the case of acquisition of dynamic contrast-enhanced sequences, a temporal resolution
of up to 15 seconds is possible.
-
bpMRI is currently not yet recommended for general use.
Changes regarding interpretation of findings
Changes regarding interpretation of findings
While the interpretation algorithm established in PI-RADS 2.0 relates exclusively
to lesions in the peripheral zone and the transition zone, the anterior fibromuscular
stroma (AFMS) and the central zone are additionally included as relevant regions in
PI-RADS 2.1. However, these regions do not have their own scoring system and unremarkable
findings do not require separate mention in the radiology report. In addition, there
are small but clinically relevant changes regarding the scoring of lesions in the
transition zone and the interpretation of DWI and DCE-MRI.
Central zone
The central zone (CZ) is histologically very similar to the seminal vesicles and encircles
the ejaculatory duct while extending from the base of the prostate dorsal to the TZ
caudally in the direction of the prostate apex to the seminal colliculus (verumontanum).
Due to its V-shaped configuration, the CZ can often be better identified on the coronal
plane in up to 93 % of mpMRI examinations [38]
[39].
The normal CZ is a region with bilaterally symmetric low signal intensity in T2w and
ADC and is mildly hyperintense on DWI. Due to the histological similarity of the CZ
with the seminal vesicles and the low incidence of cancer of the seminal vesicles,
cancer in the CZ is also extremely rare and accounts for less than 5 % of all cases
of prostate cancer. Its etiological and locoregional connection with the seminal vesicles
explains why the seminal vesicles are often infiltrated by tumors of the CZ [40]. Tumors of the CZ are often seen as asymmetrical masses with a hypointense signal
on T2w, a hyperintense signal on DWI, and early enhancement on DCE-MRI. By discussing
this anatomical region of the prostate and its physiological signal behavior, PI-RADS
2.1 aims to avoid possible misinterpretation of the normal CZ as cancer and to prevent
misinterpretation of tumors near the base at the junction between the peripheral zone
and the transition zone as the CZ.
Anterior fibromuscular stroma
Like the CZ, the normal anterior fibromuscular stroma (AFMS) is specifically discussed
in PI-RADS 2.1 since it can be difficult to differentiate from ventral tumors [41]. It is comprised of vertically running muscle bundles and connective tissue and
forms the ventral margin to the prostate as non-glandular tissue. The bilaterally
symmetrical AFMS therefore has a low signal intensity on T2w, DWI, and ADC and delayed
contrast enhancement (“low on all sequences”) [13]. Cancers do not originate in the AFMS but can infiltrate the AFMS from the transition
zone or the peripheral zone [42]. The evaluation criteria for lesions in the AFMS are therefore to be selected according
to the region of origin (PZ or TZ). However, since this cannot always be definitively
determined, there is a certain degree of diagnostic uncertainty in such cases which
is recognized by the PI-RADS steering committee as a limitation of the assessment
system.
Transition zone
Approximately 30 % of all prostate cancers develop in the transition zone (TZ) [42]. It is known that the sensitivity and specificity of mpMRI for cancers in the TZ
are lower than in the PZ [43]. BPH nodules of various morphology and levels of signal intensity, cystic changes,
and stromal changes with an inhomogeneous T2 signal are regularly seen in the TZ [44]. Since “typical” BPH nodules (see below) are normally seen in the TZ in patients
with suspicion of prostate cancer, PI-RADS version 2.1 reclassified “typical” BPH
nodules in the TZ from a PI-RADS score of 2 (in version 2.0) to a PI-RADS score of
1 (in version 2.1) (see diagram in [Fig. 3]).
Fig. 3 Reporting scheme according to the current PI-RADS version 2.1 The changes to PI-RADS
version 2.0 are highlighted by a red frame.
Typical BPH nodules are lesions that are fully encapsulated on T2w ([Fig. 4], [5a]). Experience in the clinical routine with PI-RADS 2.1 and studies will show whether
a more precise definition regarding the capsule is necessary, e. g. definition of
the number of planes on which a complete capsule must be visible.
Fig. 4 “Typical nodule” in transition zone (TZ) in a 60-year-old man with PSA increase to
6.38 ng/ml and negative biopsy. Oval, fully encapsulated (arrows) BPH-nodule in T2w
within the left TZ. ADC maps and the high b-value DWI images (not shown) revealed
no findings. According to PI-RADS 2.1, this is a PI-RADS 1 finding, which therefore
does not usually need to be mentioned separately in the findings.
Fig. 5 Schematic illustration of different types of atypical BPH-nodules in comparison to
the typical BPH-nodule.
In contrast, atypical BPH nodules are assigned a PI-RADS score of 2 and are defined
as nodules with the following characteristics:
-
Mostly (but not completely) encapsulated ([Fig. 5b])
-
Homogeneous, hypointense, circumscribed, without encapsulation ([Fig. 5c])
-
Homogeneous mildly hypointense area between nodules ([Fig. 5d])
The presence of microcysts in the BPH nodule is also discussed in PI-RADS 2.1, is
classified as a benign change [45], and is assigned a PI-RADS score of 1 (completely encapsulated nodule) or 2 (mostly
but not completely encapsulated nodule).
PI-RADS 2.1 uses or more detailed and more differentiated description of changes in
the T2 categories PI-RADS 1 and 2 to take into account the many different nodular
changes in these categories (see [Fig. 3]).
In addition to the revised T2 scoring in the TZ, PI-RADS 2.1 assigned greater value
to diffusion imaging in the TZ: Atypical nodules in the TZ (PI-RADS score 2 see above)
can now to be upgraded via DWI to a total PI-RADS score of 3 in the case of diffusion
restriction with a score ≥ 4 (see [Fig. 3], [6]). An upgrade from a T2w score of 3 to a total PI-RADS score of 4 can still only
be achieved in the case of a DWI score of 5 (see [Fig. 3]).
Fig. 6 Upgrade from a PI-RADS-2 to a PI-RADS-3 lesion in a 62-year-old man with a PSA of
6.08 ng/ml. Circumscribed low T2w signal a without a capsule at the anterior margin of the TZ paramedian left, corresponding
to a PI-RADS 2 in the T2 image (“atypical node”). In addition to a significant signal
drop in the ADC maps b, a corresponding significant diffusion restriction is shown in the b1000 image c, in the b1400 image d and in the b2000 image e, corresponding to a PI-RADS score 4 in the DWI. According to PI-RADS 2.1, this results
in an overall PI-RADS score of 3.
The greater inclusion of DWI in PI-RADS 2.1 is intended to increase the sensitivity
of the detection of lesions in the TZ. However, this will only be the case if corresponding
PI-RADS 3 lesions are also biopsied in a targeted manner. Initial results indicate
that prostate cancers in the TZ can be detected significantly better using PI-RADS
2.1 than PI-RADS 2.0 with simultaneously reduced interreader variability [44].
Interpretation of DWI
Categories 2 and 3 of DWI were not clearly defined in version 2.0 and have occasionally
resulted in difficulties regarding image interpretation. PI-RADS 2.1 revised both
of these categories (see [Fig. 3]). [Table 1] provides a comparison of the definitions of DWI categories 2 and 3 between PI-RADS
versions 2 and 2.1.
Table 1
Changes in definitions of PI-RADS categories 2 and 3 in diffusion-weighted imaging
in PI-RADS v2.1 compared to PI-RADS v2.0.
|
PI-RADS 2.1
|
PI-RADS 2.0
|
|
DWI PI-RADS category 2
|
linear/wedge-shaped hypointense on ADC and/or linear/wedge-shaped hyperintense on
high b-value images (no focal lesions) ([Fig. 7])
|
indistinct hypointense on ADC
|
|
DWI PI-RADS category 3
|
focal (discrete and different from the background) hypointense on ADC and/or focal
hyperintense on high b-value DWI; may be markedly hypointense on ADC or markedly hyperintense
on high b-value DWI, but not both
|
focal mildly/moderately hypointense on ADC and isointense/mildly hyperintense on high
b-value DWI images
|
Fig. 7 Typical PI-RADS-2 lesion in DWI in a 50-year-old biopsy-naïve man with an increase
in PSA from 3.76 ng/ml to 4.17 ng/ml within 5 months. According to PI-RADS 2.1, this
is a linear, band/wedge-shaped, non-focal signal reduction in the posteromedial right
PZ (arrow) in T2w a with corresponding diffusion restriction in ADC maps b and corresponding signal increase in both the b1400 image c and the b2000 image d.
The term “marked” indicates a greater signal deviation than at any other focus in
the same zone. In the case of multiple lesions, according to PI-RADS 2.1 (analogous
to version 2.0) up to four lesions in categories 3–5 can be specified with the index
lesion corresponding to the one with the highest PI-RADS score or, in the case of
equivalent lesions, to the one with extraprostatic extension (if not applicable, the
largest lesion).
Interpretation of DCE
While the definition of positive contrast enhancement in DCE remains unchanged, PI-RADS
v2.1 states with respect to negative contrast enhancement that diffuse or multifocal
enhancement is not a typical sign of cancer but rather is an expression of inflammatory
changes. Focal enhancement of a lesion demonstrating typical features of BPH on T2 W
is still not considered a positive finding.
Prostate volume measurement
Every mpMRI finding in the prostate should include a volume measurement of the organ,
in order to use this measurement to calculate the individual PSA density (e. g., cut-off
> 0.15 ng/ml/cm3) as a clinical biomarker of increasing importance for clinical decision processes
[46]
[47]. A new recommendation in PI-RADS 2.1 is to measure the largest anterior-posterior
(ap) extension of the prostate on a mid-sagittal image (not axial as previously recommended).
The goal here is to measure the ap extension independent of the orientation of the
axial plane perpendicular to the long axis of the prostate (perpendicular to the max.
extension of the prostate between the base and apex on the sagittal plane). Methods
for the automatic segmentation of the prostate are also considered as an alternative
[17].
Prostate sector map
The prostate sector map is intended to make it easier to allocate described and classified
lesions as precisely as possible to an anatomical region of the prostate. 27 prostate
sectors were defined for this purpose in the first PI-RADS version and 36 in the second
version. PI-RADS 2.1 has introduced two additional sectors so that the sector map
of the prostate includes 38 prostate sectors, two sectors for the seminal vesicles,
and one sector for the membranous urethra for a total of 41 sectors. The posteromedial
zones of the peripheral zone (PZpm) to the left and right of the base of the prostate
are a new addition to PI-RADS 2.1.
Summary
PI-RADS version 2.1 addresses various small changes regarding examination technique
and interpretation of findings and is an evolution of PI-RADS 2.0 with the basic concept
still being zone-related dominant sequences.
The greater importance placed on diffusion imaging in the transition zone with the
goal of increasing the still comparably low sensitivity of mpMRI in this zone is particularly
relevant. The newly introduced differentiation between “typical” and “atypical” BPH
nodules based on T2 sequences in the transition zone places an even greater focus
on this zone. The introduction of the mandatory ultra-high b-value in DWI with which
many centers have already had good experiences is also a welcome addition. Given the
increase in discussions regarding biparametric MRI, it is fitting for PI-RADS version
2.1 to address this option. However, multiparametric MRI remains the method of choice,
particularly at facilities with a small number of cases.
Moreover, simplification of the scoring algorithm, detailed definition of specific
signal changes, and further standardization of image acquisition are intended to help
lower interreader variability and to further strengthen the value of mpMRI in the
detection of clinically significant prostate cancer.
Outlook
PI-RADS is not a fixed concept but rather is subject to constant change. Future versions
must address clinical and technical advances. Due to the good experiences regarding
the detection and localization of prostate cancer, there has been increasing interest
in the applicability of PI-RADS or other scoring systems for active surveillance and
in the post-therapeutic setting. As a result of rapid advances, artificial intelligence
and deep learning algorithms will probably support the interpretation of findings
in the future [48]. It may be possible in the future with the help of new qualitative and quantitative
factors still to be evaluated to link the PI-RADS system, like the BI-RADS method,
to concrete instructions regarding biopsy to verify findings with the incorporation
of additional clinical and primarily laboratory parameters (e. g. PCA-3, etc.) [49].
