- Email: [email protected]
Common Technical and Anatomical Pitfalls in the Evaluation of Multiparametric Prostate Magnetic Resonance Imaging
Author's Accepted Manuscript
Common Technical and Anatomical Pitfalls in the Evaluation of Multiparametric Prostate MRI Xiaozhou Liu, Sadhna Verma MD
www.elsevier.com/locate/enganabound
PII: DOI: Reference:
S0037-198X(15)00042-5 http://dx.doi.org/10.1053/j.ro.2015.08.004 YSROE50519
To appear in:
Seminar in Roentgenology
Cite this article as: Xiaozhou Liu, Sadhna Verma MD, Common Technical and Anatomical Pitfalls in the Evaluation of Multiparametric Prostate MRI, Seminar in Roentgenology, http://dx.doi.org/10.1053/j.ro.2015.08.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Common technical and anatomical pitfalls in the evaluation of multiparametric prostate MRI
!"#$%& ' () ) *( + ( ,( # -( !"#$ %& * &' (. ((/(&& 0 1(2
Common technical and anatomical pitfalls in the evaluation of multiparametric prostate MRI Introduction Multiparametric MRI is the single most versatile imaging assessment of the prostate, useful in delineating local soft-tissue anatomy and in tumor visualization, while providing superior soft-tissue resolution. Multiparametric MRI readily detects extracapsular and seminal vesicle invasion, both of which are important staging and prognosticating factors. [1-3] MRI is also helpful in assisting surgical planning, especially if a neurovascular-bundle-sparing technique is considered. [4] As no sequence alone provides sufficient characterization, a multiparametric approach is the current standard. Despite recent technological advancements, radiologists must be aware of the potential pitfalls and diagnostic challenges in interpreting prostate MR. Below we discuss a few such pitfalls that are commonly seen in clinical practice.
Acquisition related artifacts Diffusion-weighted imaging (DWI) DWI is based on the principle of random motion of water molecules in tissue. In cancer, tissues with tightly packed cells, water diffusion tends to be more restricted. T2 “shine through” artifacts due to the high water content in the peripheral zone can be reduced by calculating apparent diffusion coefficient (ADC). This allows malignancies to be detected as focal restriction on ADC maps (figure 1). DWI sequences with ADC maps are prone to artifacts, especially metal implants such as hip replacement, owing to the relatively low signal-to-noise ratio (figure 2). Although correlation between Gleason score and ADC values have been reported, [5-8] there is significant overlap between benign and malignant
lesions, reducing ADC specificity for malignancy. [9] At our institution, we use multiple b values and an additional high-b-value sequence up to 2000 s/mm2 for prostate DWI acquisition. In our experience, and consistent with recent studies, [10-13] high b values are associated with improved lesion detection (figure 3). However, the high b value may also lead to contour distortions of the gland on DWI. This is caused by random eddy currents in the human tissue produced by the rapidly changing magnetic fields of DWI echo-planar sequences. These eddy currents create irregular magnetic fields that interfere with the external field and degrade the images. [14] As a result, prostatic capsule, focal lesions, and internal boundaries may appear distorted or obscured on high-b-value images (figure 3D). For ADC map to be informative, an appropriate windowing level is critical. Figure 4 demonstrates an example of how optimal window accentuates a focal lesion that would be difficult to visualize on sub-optimal windows. Determining an optimal window can be a technically complicated process as it is affected by the various scanning parameters. A prior study [14] has advocated specific values for the window and level settings (window width of 1.650 x 10-6 mm2/s and level of 1.675 x 10-6 mm2/s). In our experience, standardized window and level settings are useful to ensure consistent imaging quality, after manual adjustments to account for technical differences between scanners. Dynamic contrast enhancement (DCE) In DCE imaging, a bolus of IV contrast is injected, followed by a series of rapid T1-weighted (T1W) image acquisitions. Prostate cancer demonstrates increased early enhancement compared to normal tissue due to increased vascularity and vessel permeability and due to tissue factors secreted by the carcinoma. [15, 16]
Although experience with DCE-MRI is limited, focal early enhancement is suggestive of
malignancy. DCE-MRI combined with T2W images have also shown high sensitivity and specificity (90% and 88%, respectively) for lesions greater than 0.5 cm. [17, 18] In our experience, DCE-MRI is an important sequence as part of a multiparametric exam. However, it has limited value on its own, as some nonmalignant lesions such as prostatitis could also demonstrate early focal enhancement (figure 5).
Interpretive errors Errors in detection when a reader fails to perceive an abnormality
Occasionally, prostate tumors may have unusual or uncommon imaging presentations, leading to diagnostic dilemma or false-negative interpretations. Awareness of these less common presentations is important, since these are also prone to being missed on standard biopsy. Subcapsular crescentic tumors Small subcapsular tumors can be diagnostically challenging, as they tend to be crescentic in shape and may be difficult to differentiate from the crescentic-shaped capsule on T2W images. It has been previously hypothesized [19] that peripheral zone tumors may initially exhibit subcapsular spread during early growth. Subcapsular tumors are important to identify for optimal surgical planning, given the propensity for extraprostatic extension. As demonstrate in figure 6, such a tumor may be indistinct on T2W images from the adjacent capsule. The flattening of the capsule by an endorectal coil, if used, may also hinder identification. DWI and DCE facilitate detection by more clearly showing small subcapsular signal abnormalities. the lesion was revealed on the ADC map. Distal apical tumors A very distal apical tumor can be difficult to detect on a multiparametric MR exam, due to the relatively small size and lack of clear capsule in distal apical region (figure 7). In addition, a tumor in this location is often difficult to biopsy due to anatomical factors, such as need for sampling tissue at an oblique angle from the needle tip. A previous study has reported relatively low negative predictive value of a standard 12-core TRUS biopsy for distal apical tumors, in which among patients who underwent a 12-core biopsy followed by radical prostatectomy, 28% of patients were positive at the apex on biopsy, and 65% were positive for distal apical malignancy on final pathology specimen. [20] As such, it is crucial for a radiologist to inspect this region closely for suspicious lesions, since the distal apical region has been
reported as the most common site of positive margins after surgery. [21] Such a tumor typically demonstrates similar features as other peripheral zone tumors (T2 hypo-intense with focal restriction on ADC map and early enhancement on DCE), therefore correlation with functional sequences such as ADC is helpful (figure 7, note that the abnormality is readily visible on ADC map). Peri-urethral tumors The peri-urethral region is generally not well visualized, partly related to non-specific decreased T2 signals in this area, as well as effacement by surrounding benign prostatic hyperplasia (BPH) nodules. Such a tumor is often not well-visualized on an anatomical sequence, but is sometimes detectable on DWI or DCE images (figure 8). Detection of peri-urethral tumors is important due to staging and surgical planning considerations. Anterior horn tumors While transition zone occupies the bulk of anterior prostate in most men, a portion of the peripheral zone extends along the lateral margin of the transition zone to the anterior prostate margin, described as “anterior” or “anterolateral” horn of peripheral zone. Tumors in this region can be diagnostically challenging – they are difficult to detect on a standard biopsy given the anterior location, as well as on anatomical MR images, due to the variable size and positioning of transition zone at the apex. Also, depending on the extent of BPH, stromal nodules may bulge into the apex and distort anterior anatomy in unpredictable ways. Such a tumor has MR features similar to elsewhere in the peripheral zone, and correlation with functional sequences is helpful (figure 9). Infiltrative tumors A subtype of prostate carcinoma manifests as diffusely infiltrative tumor. Histologically, the tumor is comprised of scattered malignant glands intermixed with normal prostate tissue, rather than forming a mass lesion. These tumors are challenging to diagnose, as the gland may appear completely unremarkable on multiple sequences in a multiparametric exam (figure 10). Radiologists, urologist and patients should
be aware of this known limitation of multiparametric prostate MRI with current technology, and take this into account when considering clinical management following a negative exam. Finally, malignancies in areas with stromal BPH have traditionally posed great diagnostic challenges. This will be discussed in greater details later in this review. Errors of interpretation when an abnormality is identified but attributed to a wrong cause Normal anatomic structures mistaken for malignancy The anatomical central zone was initially described by McNeal [22] in 1981. The central zone surrounds the ejaculatory ducts. Tumors arising within the central zone account for approximately only 2.5% of prostate carcinoma but have a propensity to invade the seminal vesicles and are associated with poor prognoses overall. [23] The traditional school of thought [24] is that as the transition zone expands with aging, the central zone is gradually effaced and can no longer be distinguished from the transition zone on a standard MRI. The transition zone and the central zone are collectively referred to as the central gland. Recent developments in imaging now allows visualization of the central zone as a distinct structure to some extent, with one report suggesting that it may be identified in up to 80% of patients with prostate cancer. [24] On a multiparametric prostate MRI, the central zone is an area near the base of the prostate that is T2 hypo-intense with decreased signal on ADC map, mimicking a peripheral zone lesion. The central zone can be identified as an area of butterfly shape on coronal images (figure 11). The neurovascular bundle (NVB) contains cavernous neural plexus in close proximity to the local vascular structures (including the periprostatic venous plexus) and courses along the posterolateral margin of the gland bilaterally at 5 o’clock and 7 o’clock positions. [25-27] The cavernous neural plexus supplies the corpora cavernosa, and is critical to the preservation of sexual function in a nerve-sparing prostatectomy. [26, 27] Occasionally, the NVB may appear as focal suspicious lesions along the capsule adjacent to the peripheral zone. This pitfall can be avoided by visualizing the structure on multiple images,
as NVBs are typically curvilinear, tubular structures as opposed to a round, mass-like lesion typical for prostate cancer (figure 12). Non-cancerous abnormalities mistaken for malignancy Post-biopsy hemorrhage is hypo-intense on T2W images, similar to prostate carcinoma. In our experience, areas of hemorrhage may present significant diagnostic difficulties by either mimicking or masking tumors (figure 13). This is exacerbated by the fact that the prostate gland actively secretes citrate, which has anti-coagulant properties and contributes to the prolonged presence of hemorrhage in the peripheral zone. [28, 29] As a result, a delay of several weeks prior to multiparametric MR is recommended after biopsy to allow adequate blood product resorption. There is debate about the optimal wait time, but a minimum of 4 to 8 weeks is generally recommended. [30, 31] Several strategies exist to mitigate the effect of residual hemorrhage. T1W images are often helpful in highlighting blood products and delineating its distribution in suspicious areas seen on T2W images (figure 13A). Previously, Barrett et al. have [32] have reported the hemorrhage exclusion sign where post-biopsy hemorrhage spares and outlines the tumor periphery, thus aiding in its visualization. A few prior studies have also shown that DWI remains sensitive to prostate cancer when combined with T2W images in the setting of post-biopsy hemorrhage. [33, 34]
In our experience, the manifestations of post-biopsy hemorrhage are more varied. Areas of
hemorrhage have decreased ADC values in benign tissue, and increased image distortion due to susceptibility artifacts. Incompletely resorbed hemorrhage may mask a tumor from detection (figure 13C and D), rather than outlining it. As such, DWI can have major limitations in the presence of post-biopsy hemorrhage. Stromal BPH nodules in the transition zone significantly increases the difficulty of cancer detection, [35] where 30% of prostate cancer arise. Appearance on T2W images may delineate the etiology of a hypointense transition zone lesion. As a general rule, tumors tend to be lenticular with irregular hypointensities circumscribed by poorly-defined margins, whereas stromal BPH nodules tend to be wellcircumscribed with sharp boundaries. However, imaging characteristics on anatomical sequences may not
always be reliable. For example, an occult tumor in area of BPH can be nearly imperceptible on T2W images (figure 14A). The same tumor can be visualized on ADC map and DWI with high b value (figure 14B and C). Oto et al. [36] have demonstrated that ADC values may help differentiate transition zone prostate carcinoma from stromal BPH nodules. In our experience, we have only had limited success with ADC maps for transition zone lesions. This is primarily because evolving stromal BPH nodules can have ADC values that overlap significantly with tumor. [9] A recent study [37] has shown that prostate MR spectroscopy may be useful in transition zone lesions. In the study, a significant correlation was observed between aggressiveness of a lesion and metabolite ratios (ȡ=0.58 for choline + creatine/citrate, and ȡ=0.60 for choline/creatine ratios). In summary, stromal BPH nodules pose significant challenge in prostate MR interpretation. In challenging cases, imaging alone may not be sufficient to differentiate tumors from BPH nodules. Acute and chronic prostatitis due to an infectious or inflammatory etiology can cause PSA elevations. [38] The imaging presentations can also be similar to malignancy and contribute to false-positive findings. [39, 40]
Both active processes and sequelae of prostate infection or inflammation demonstrate T2 signal
abnormalities. Highlighting the dilemma, a prior study [41] showed that a significant proportion of biopsies yielded prostatitis or inflammatory changes after a positive multiparametric prostate MR. Past observations [42] are consistent with our experience in that inflammatory lesions tend to be curvilinear with ill-defined margins as opposed to tumors which tend to be rounded mass-like lesions. However, exceptions are plenty and imaging characteristics alone are often insufficient to reach definitive diagnoses. Granulomatous prostatitis is a distinct entity that deserves attention. The etiology is diverse, with bacteria, fungi, parasites and viruses all implicated, [43] although up to 69% may have non-specific causes. [44] With an overall low incidence, granulomatous prostatitis is frequently seen in the setting of bladder cancer as a complication of bacillus Calmette-Guerin (BCG) immunotherapy. Clinically, granulomatous prostatitis may be detectable as a firm nodule on digital rectal exam, and may also cause PSA elevation. Granulomatous prostatitis can appear as a large consolidative mass with T2 and DWI signal abnormalities,
and may appear highly suspicious for prostate cancer (figure 15). Biopsy is the only way to distinguish between granulomatous prostatitis and malignancy due to a lack of differentiating characteristics on imaging. [45] In challenging cases, clinical correlation is often helpful. A patient from an area with endemic tuberculosis (parts of Asia and India) should prompt consideration for genitourinary tuberculosis and tuberculosis prostatitis. A history of prior superficial bladder cancer treated with BCG immunotherapy should prompt consideration for granulomatous prostatitis.
# "
,
* &
+ !
# "
!"3%#$ !%& '
() !" % $ *"+%," - . ' / '( ' ( ! %3#. $ .%!01.%" 2
!!" 44 $ ,,%# 3 '
! " # '( ' ( !! %#"$ 3#% 24
! $% ! # " 5 !$% $ ",#% 425
& $ ! ' 5 !$5 $ "&&%+, - .
'( ' ( ! %3 $ #*"%& 62 78
$% (7 !! $ #!%" 9 9
$
: !!*# $ &" %* 9 / 9
(
" 7 7 . !!&6 $ +#%+ ; 9
) * 5 !!+$4$ ",%, 4 4
"
(7 !!&6#$ ,, %
< '
5+
= ! $% $ &&%
, * & + !
# " ,
* & + #! # # ## #" #, #
>79
) ++"6#$ *&%&* 3 =
' (6 ++# # $ "!%+ ( ?(
,-.( / !
5++*4#$ ",%, 4
..
) & = ' ! $4$ !,%" 7=
(@ A
) ) 012 346 !!#%"&%,* : -
) 3* ) 5 ! 6%$ #*%" . ('(
( ) !!* 65$ #&,%+B #&+%+! 7=
(@ " 6 +&$ #,%"+ (
+ " () !!& 6%,$ * %*B ** - '
4
" " ! 5 ! 3#$ &+"%+! ;.@
.
) 5 !!*3%#$ ,!%" =;3
! '( ' ( ++, 3#"$ + #%* 4 5 7
! 5++ 5 $ !*% . 7;
+
! 5++# 5% $ ##+%, 3.
)
! 5++, %$ $ #&,%+! 9 C
' (' !" 3 $ &!%" 0 55 '
5 ! ! '( ' ( !!" 5"$ !*+%&# 4
6 3 ! 5 ! 3#$ *,%* 2 1'
)
* (7 !! $ #&*%+" 4 4
)
! 5 !! $ ,#%+ ;
+ ! "'( ' ( !! 53#$ * +%" A '
) ! 5 !!$6#$ *,% #
#* #& #+ "! "
" "# "" ",
9 4
) ! $ 5 ! 3$ $ ",*%* =
7
() ++, $# 6$ "!*%# 2 7
) ! 2" 2 5 !3 $ "% 1(A
7.8! ! *%3*@ ! "$ *"%,* 24
$ $ / : !!&#!$ &%+" .2 %8 :'
+ ! 3( ! 5 !!" #$ **% " . 5( 9 & 7 6 !!" 6#$ # &%"& D7 =C D : 3 (.A !!#$ #! 7
# / () !!, ,$ "*"%&
A
B
*
Figure 1 45-year-old man with Gleason score 6 (3+3) adenocarcinoma of the prostate. (A) Axial T2W image at the level of apex, which shows non-specific low signal in the right peripheral zone apex (arrow). There is moderate buldging of the prostatic capsule on the right and effacement of rectoprostatic angle (*). (B) ADC map demonstrates focal restriction in right apex (arrow) correlating with the low signal area on T2W images.
A
B
Figure 2 58-year-old male with targeted biopsy-confirmed far-anterior prostate adenocarcinoma of Gleason score 7 (4+3). This patient also has a left hip prosthesis. (A) Axial T2W image at the level of mid-gland, demonstrating an ill-defined far anterior lesion with capsular bulge (arrow). (B) ADC map at the same level demonstrating extensive metal artifact. Except for a small area in the right peripheral zone, the ADC map is un-interpretable and the anterior lesion seen on the T2W image cannot be evaluated.
A
B
C
D
b1000
b1400
Figure 3 72 year-old male with biopsy-proven Gleason score 7 (3+4) prostate adenocarcinoma. (A) Axial T2W image shows subtle ill-defined hypointensity in the left lateral mid gland peripheral zone (arrow). (B) ADC map shows equivocal, if any, diminished signal in the suspicious region. (C) DWI acquired at b=1000 s/mm2 , the lesion with subtle T2 signal abnormality is not apparent (arrow points to the expected location of the lesion). (D) DWI acquired at b=1400 s/mm2 , the lesion is now apparent (arrow). Note that spatial resolution becomes extremely poor at b=1400 s/mm2 with obscuration of the capsule and internal structures, also note the prominent “ghosting” artifact (between the arrowheads).
A
B
C
Figure 4 65 year-old male with targeted biopsy-proven bilateral adenocarcinoma of Gleason score 6 (3+3). (A) Axial T2W image demonstrates non-specific hypointensity within the mid gland peripheral zone bilaterally (arrows). (B) ADC at sub-optimal window (W = 3357, L = 1149), the lesions are not visible (arrows point to the expected location of bilateral lesions in this patient). (C) ADC map reveals focal restrictions after window optimization (arrows, W=1500, L = 1600).
A
B
D
E
C
Figure 5 60 year-old male with targeted biopsy-proven Gleason score 8 (4+4) in the right base to mid gland and prostatitis with calcifications in the left mid-gland. (A) Axial T2W image at the mid-gland level demonstrates T2 hypointense mass, as well as heterogeneous hypo-intense lesions in the left mid gland peripheral zone (arrowhead), which represent calcifications. (B) Coronal T2W image demonstrates T2 hypointense mass in the right superior base (arrow), as well as heterogeneous hypo-intense calcifications (arrowhead). (C) ADC map demonstrates focal restriction (arrow). (D) Axial DCE image at the right base shows focal early enhancement in the region of biopsy-proven malignancy (arrow). (E) Axial DCE image at mid-gland level shows focal early enhancement in the region of prostatitis with calcifications (arrowhead).
A
B
Figure 6 55 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) On this axial T2W image, the subcapsular crescentic tumor is indistinct from the prostatic capsule (arrow). (B) ADC map reveals focal restriction in a crescentic shape along the capsule, consistent with malignancy.
A
B
Figure 7 49 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) Axial T2W image in the distal apical region. An area of low signal (arrow) can be equivocally identified. (B) ADC map at the same level demonstrates focal restriction consistent with malignancy (arrows).
A
B
C
Figure 8 55 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) Axial T2W image at the far apical region demonstrates no obvious signal abnormality. (B) ADC map demonstrates focal restriction in the peri-urethral region (arrows). (C) DCE image demonstrates bright signal corresponding to the area with focal restriction, consistent with malignancy (arrow).
A
B
C
D
Figure 9 69 year-old male with targeted biopsy-proven Gleason score 7 (3+4) adenocarcinoma. (A) Axial T2W image. Arrow denotes the expected location of the abnormality, although it is not well visualized. (B) DCE image at the same level demonstrates focal early enhancing signal (arrow). (C) ADC map at the same level demonstrates focal restriction consistent with malignancy (arrow). (D) DWI with b=1400 s/mm2 , demonstrating bright signal consistent with malignancy (arrow).
A
B
C
Figure 10 65 year-old male with Gleason score 6 (3+3) diffuse infiltrative prostate cancer on histology. (A) Axial T2W image demonstrates no focal hypointense lesions. (B) ADC map demonstrates vague ADC restriction foci bilaterally. (C) DCE demonstrates no focal early enhancements, with vague enhancing regions seen bilaterally, a non-specific finding.
A
B
* *
*
*
C
*
*
Figure 11 Normal central zone. (A), (B) and (C), axial T2W, coronal T2W, and ADC map demonstrating the appearance of a normal central zone on a multiparametric MR exam (*), respectively.
A
B
C
D
Figure 12 Appearance of neurovascular bundle (NVB). (A) Axial T2W image at the level of low mid gland. Note the location of NVB at 5 o’clock and 7 o’clock locations in close proximity to the prostatic capsule (arrows). (B), (C) NVB at mid gland and superior mid gland levels (arrows), respectively. (D) ADC map, on which NVB appears as focal restriction (arrow). It is useful to visualize NVB across multiple slices on T2W images so that it is not mistaken for a subcapsular lesion.
A
B
C
D
Figure 13 51-year-old male who had a staging prostate MR for recently diagnosed bilateral prostate adenocarcinoma of Gleason score 8 (3+5). PSA = 6.3 ng/mL. (A) Axial T1W image shows diffuse bilateral bright signal indicative of post biopsy hemorrhage 3 weeks post biopsy (arrows). (B) Axial T2W image at the same level demonstrates diffuse low signal in the peripheral zone bilaterally with no dominant lesion. (C) ADC map shows no focal restriction. (D) ADC map from a repeat exam 8 weeks later demonstrates a large region of restriction in the right peripheral zone, as well as a smaller area of restriction anteriorly, consistent with malignancy (arrows).
A
B
C
Figure 14 75 year-old male with Gleason score 6 (3+3) adenocarcinoma at targeted biopsy of left transition zone. (A) Axial T2W image at mid-gland shows numerous areas of BPH without focal abnormality. (B) and (C) ADC map and extrapolated DWI (b = 2000 s/mm2 ) identified the lesion (arrows).
A
B
C
D
Figure 15 71 year-old male with history of superficial bladder cancer previously treated with intravesicle instillation of BCG, now presents with a rising PSA and abnormal DRE. Biopsy revealed granulomatous prostatitis. (A) Coronal and (B) axial T2W images demonstrate diffuse low signal throughout the prostate gland. (C) ADC map demonstrates diffuse diffusion restriction throughout the prostate gland. (D) DCE image demonstrates diffuse early enhancement.
Common Technical and Anatomical Pitfalls in the Evaluation of Multiparametric Prostate MRI Xiaozhou Liu, Sadhna Verma MD
www.elsevier.com/locate/enganabound
PII: DOI: Reference:
S0037-198X(15)00042-5 http://dx.doi.org/10.1053/j.ro.2015.08.004 YSROE50519
To appear in:
Seminar in Roentgenology
Cite this article as: Xiaozhou Liu, Sadhna Verma MD, Common Technical and Anatomical Pitfalls in the Evaluation of Multiparametric Prostate MRI, Seminar in Roentgenology, http://dx.doi.org/10.1053/j.ro.2015.08.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Common technical and anatomical pitfalls in the evaluation of multiparametric prostate MRI
!"#$%& ' () ) *( + ( ,( # -( !"#$ %& * &' (. ((/(&& 0 1(2
Common technical and anatomical pitfalls in the evaluation of multiparametric prostate MRI Introduction Multiparametric MRI is the single most versatile imaging assessment of the prostate, useful in delineating local soft-tissue anatomy and in tumor visualization, while providing superior soft-tissue resolution. Multiparametric MRI readily detects extracapsular and seminal vesicle invasion, both of which are important staging and prognosticating factors. [1-3] MRI is also helpful in assisting surgical planning, especially if a neurovascular-bundle-sparing technique is considered. [4] As no sequence alone provides sufficient characterization, a multiparametric approach is the current standard. Despite recent technological advancements, radiologists must be aware of the potential pitfalls and diagnostic challenges in interpreting prostate MR. Below we discuss a few such pitfalls that are commonly seen in clinical practice.
Acquisition related artifacts Diffusion-weighted imaging (DWI) DWI is based on the principle of random motion of water molecules in tissue. In cancer, tissues with tightly packed cells, water diffusion tends to be more restricted. T2 “shine through” artifacts due to the high water content in the peripheral zone can be reduced by calculating apparent diffusion coefficient (ADC). This allows malignancies to be detected as focal restriction on ADC maps (figure 1). DWI sequences with ADC maps are prone to artifacts, especially metal implants such as hip replacement, owing to the relatively low signal-to-noise ratio (figure 2). Although correlation between Gleason score and ADC values have been reported, [5-8] there is significant overlap between benign and malignant
lesions, reducing ADC specificity for malignancy. [9] At our institution, we use multiple b values and an additional high-b-value sequence up to 2000 s/mm2 for prostate DWI acquisition. In our experience, and consistent with recent studies, [10-13] high b values are associated with improved lesion detection (figure 3). However, the high b value may also lead to contour distortions of the gland on DWI. This is caused by random eddy currents in the human tissue produced by the rapidly changing magnetic fields of DWI echo-planar sequences. These eddy currents create irregular magnetic fields that interfere with the external field and degrade the images. [14] As a result, prostatic capsule, focal lesions, and internal boundaries may appear distorted or obscured on high-b-value images (figure 3D). For ADC map to be informative, an appropriate windowing level is critical. Figure 4 demonstrates an example of how optimal window accentuates a focal lesion that would be difficult to visualize on sub-optimal windows. Determining an optimal window can be a technically complicated process as it is affected by the various scanning parameters. A prior study [14] has advocated specific values for the window and level settings (window width of 1.650 x 10-6 mm2/s and level of 1.675 x 10-6 mm2/s). In our experience, standardized window and level settings are useful to ensure consistent imaging quality, after manual adjustments to account for technical differences between scanners. Dynamic contrast enhancement (DCE) In DCE imaging, a bolus of IV contrast is injected, followed by a series of rapid T1-weighted (T1W) image acquisitions. Prostate cancer demonstrates increased early enhancement compared to normal tissue due to increased vascularity and vessel permeability and due to tissue factors secreted by the carcinoma. [15, 16]
Although experience with DCE-MRI is limited, focal early enhancement is suggestive of
malignancy. DCE-MRI combined with T2W images have also shown high sensitivity and specificity (90% and 88%, respectively) for lesions greater than 0.5 cm. [17, 18] In our experience, DCE-MRI is an important sequence as part of a multiparametric exam. However, it has limited value on its own, as some nonmalignant lesions such as prostatitis could also demonstrate early focal enhancement (figure 5).
Interpretive errors Errors in detection when a reader fails to perceive an abnormality
Occasionally, prostate tumors may have unusual or uncommon imaging presentations, leading to diagnostic dilemma or false-negative interpretations. Awareness of these less common presentations is important, since these are also prone to being missed on standard biopsy. Subcapsular crescentic tumors Small subcapsular tumors can be diagnostically challenging, as they tend to be crescentic in shape and may be difficult to differentiate from the crescentic-shaped capsule on T2W images. It has been previously hypothesized [19] that peripheral zone tumors may initially exhibit subcapsular spread during early growth. Subcapsular tumors are important to identify for optimal surgical planning, given the propensity for extraprostatic extension. As demonstrate in figure 6, such a tumor may be indistinct on T2W images from the adjacent capsule. The flattening of the capsule by an endorectal coil, if used, may also hinder identification. DWI and DCE facilitate detection by more clearly showing small subcapsular signal abnormalities. the lesion was revealed on the ADC map. Distal apical tumors A very distal apical tumor can be difficult to detect on a multiparametric MR exam, due to the relatively small size and lack of clear capsule in distal apical region (figure 7). In addition, a tumor in this location is often difficult to biopsy due to anatomical factors, such as need for sampling tissue at an oblique angle from the needle tip. A previous study has reported relatively low negative predictive value of a standard 12-core TRUS biopsy for distal apical tumors, in which among patients who underwent a 12-core biopsy followed by radical prostatectomy, 28% of patients were positive at the apex on biopsy, and 65% were positive for distal apical malignancy on final pathology specimen. [20] As such, it is crucial for a radiologist to inspect this region closely for suspicious lesions, since the distal apical region has been
reported as the most common site of positive margins after surgery. [21] Such a tumor typically demonstrates similar features as other peripheral zone tumors (T2 hypo-intense with focal restriction on ADC map and early enhancement on DCE), therefore correlation with functional sequences such as ADC is helpful (figure 7, note that the abnormality is readily visible on ADC map). Peri-urethral tumors The peri-urethral region is generally not well visualized, partly related to non-specific decreased T2 signals in this area, as well as effacement by surrounding benign prostatic hyperplasia (BPH) nodules. Such a tumor is often not well-visualized on an anatomical sequence, but is sometimes detectable on DWI or DCE images (figure 8). Detection of peri-urethral tumors is important due to staging and surgical planning considerations. Anterior horn tumors While transition zone occupies the bulk of anterior prostate in most men, a portion of the peripheral zone extends along the lateral margin of the transition zone to the anterior prostate margin, described as “anterior” or “anterolateral” horn of peripheral zone. Tumors in this region can be diagnostically challenging – they are difficult to detect on a standard biopsy given the anterior location, as well as on anatomical MR images, due to the variable size and positioning of transition zone at the apex. Also, depending on the extent of BPH, stromal nodules may bulge into the apex and distort anterior anatomy in unpredictable ways. Such a tumor has MR features similar to elsewhere in the peripheral zone, and correlation with functional sequences is helpful (figure 9). Infiltrative tumors A subtype of prostate carcinoma manifests as diffusely infiltrative tumor. Histologically, the tumor is comprised of scattered malignant glands intermixed with normal prostate tissue, rather than forming a mass lesion. These tumors are challenging to diagnose, as the gland may appear completely unremarkable on multiple sequences in a multiparametric exam (figure 10). Radiologists, urologist and patients should
be aware of this known limitation of multiparametric prostate MRI with current technology, and take this into account when considering clinical management following a negative exam. Finally, malignancies in areas with stromal BPH have traditionally posed great diagnostic challenges. This will be discussed in greater details later in this review. Errors of interpretation when an abnormality is identified but attributed to a wrong cause Normal anatomic structures mistaken for malignancy The anatomical central zone was initially described by McNeal [22] in 1981. The central zone surrounds the ejaculatory ducts. Tumors arising within the central zone account for approximately only 2.5% of prostate carcinoma but have a propensity to invade the seminal vesicles and are associated with poor prognoses overall. [23] The traditional school of thought [24] is that as the transition zone expands with aging, the central zone is gradually effaced and can no longer be distinguished from the transition zone on a standard MRI. The transition zone and the central zone are collectively referred to as the central gland. Recent developments in imaging now allows visualization of the central zone as a distinct structure to some extent, with one report suggesting that it may be identified in up to 80% of patients with prostate cancer. [24] On a multiparametric prostate MRI, the central zone is an area near the base of the prostate that is T2 hypo-intense with decreased signal on ADC map, mimicking a peripheral zone lesion. The central zone can be identified as an area of butterfly shape on coronal images (figure 11). The neurovascular bundle (NVB) contains cavernous neural plexus in close proximity to the local vascular structures (including the periprostatic venous plexus) and courses along the posterolateral margin of the gland bilaterally at 5 o’clock and 7 o’clock positions. [25-27] The cavernous neural plexus supplies the corpora cavernosa, and is critical to the preservation of sexual function in a nerve-sparing prostatectomy. [26, 27] Occasionally, the NVB may appear as focal suspicious lesions along the capsule adjacent to the peripheral zone. This pitfall can be avoided by visualizing the structure on multiple images,
as NVBs are typically curvilinear, tubular structures as opposed to a round, mass-like lesion typical for prostate cancer (figure 12). Non-cancerous abnormalities mistaken for malignancy Post-biopsy hemorrhage is hypo-intense on T2W images, similar to prostate carcinoma. In our experience, areas of hemorrhage may present significant diagnostic difficulties by either mimicking or masking tumors (figure 13). This is exacerbated by the fact that the prostate gland actively secretes citrate, which has anti-coagulant properties and contributes to the prolonged presence of hemorrhage in the peripheral zone. [28, 29] As a result, a delay of several weeks prior to multiparametric MR is recommended after biopsy to allow adequate blood product resorption. There is debate about the optimal wait time, but a minimum of 4 to 8 weeks is generally recommended. [30, 31] Several strategies exist to mitigate the effect of residual hemorrhage. T1W images are often helpful in highlighting blood products and delineating its distribution in suspicious areas seen on T2W images (figure 13A). Previously, Barrett et al. have [32] have reported the hemorrhage exclusion sign where post-biopsy hemorrhage spares and outlines the tumor periphery, thus aiding in its visualization. A few prior studies have also shown that DWI remains sensitive to prostate cancer when combined with T2W images in the setting of post-biopsy hemorrhage. [33, 34]
In our experience, the manifestations of post-biopsy hemorrhage are more varied. Areas of
hemorrhage have decreased ADC values in benign tissue, and increased image distortion due to susceptibility artifacts. Incompletely resorbed hemorrhage may mask a tumor from detection (figure 13C and D), rather than outlining it. As such, DWI can have major limitations in the presence of post-biopsy hemorrhage. Stromal BPH nodules in the transition zone significantly increases the difficulty of cancer detection, [35] where 30% of prostate cancer arise. Appearance on T2W images may delineate the etiology of a hypointense transition zone lesion. As a general rule, tumors tend to be lenticular with irregular hypointensities circumscribed by poorly-defined margins, whereas stromal BPH nodules tend to be wellcircumscribed with sharp boundaries. However, imaging characteristics on anatomical sequences may not
always be reliable. For example, an occult tumor in area of BPH can be nearly imperceptible on T2W images (figure 14A). The same tumor can be visualized on ADC map and DWI with high b value (figure 14B and C). Oto et al. [36] have demonstrated that ADC values may help differentiate transition zone prostate carcinoma from stromal BPH nodules. In our experience, we have only had limited success with ADC maps for transition zone lesions. This is primarily because evolving stromal BPH nodules can have ADC values that overlap significantly with tumor. [9] A recent study [37] has shown that prostate MR spectroscopy may be useful in transition zone lesions. In the study, a significant correlation was observed between aggressiveness of a lesion and metabolite ratios (ȡ=0.58 for choline + creatine/citrate, and ȡ=0.60 for choline/creatine ratios). In summary, stromal BPH nodules pose significant challenge in prostate MR interpretation. In challenging cases, imaging alone may not be sufficient to differentiate tumors from BPH nodules. Acute and chronic prostatitis due to an infectious or inflammatory etiology can cause PSA elevations. [38] The imaging presentations can also be similar to malignancy and contribute to false-positive findings. [39, 40]
Both active processes and sequelae of prostate infection or inflammation demonstrate T2 signal
abnormalities. Highlighting the dilemma, a prior study [41] showed that a significant proportion of biopsies yielded prostatitis or inflammatory changes after a positive multiparametric prostate MR. Past observations [42] are consistent with our experience in that inflammatory lesions tend to be curvilinear with ill-defined margins as opposed to tumors which tend to be rounded mass-like lesions. However, exceptions are plenty and imaging characteristics alone are often insufficient to reach definitive diagnoses. Granulomatous prostatitis is a distinct entity that deserves attention. The etiology is diverse, with bacteria, fungi, parasites and viruses all implicated, [43] although up to 69% may have non-specific causes. [44] With an overall low incidence, granulomatous prostatitis is frequently seen in the setting of bladder cancer as a complication of bacillus Calmette-Guerin (BCG) immunotherapy. Clinically, granulomatous prostatitis may be detectable as a firm nodule on digital rectal exam, and may also cause PSA elevation. Granulomatous prostatitis can appear as a large consolidative mass with T2 and DWI signal abnormalities,
and may appear highly suspicious for prostate cancer (figure 15). Biopsy is the only way to distinguish between granulomatous prostatitis and malignancy due to a lack of differentiating characteristics on imaging. [45] In challenging cases, clinical correlation is often helpful. A patient from an area with endemic tuberculosis (parts of Asia and India) should prompt consideration for genitourinary tuberculosis and tuberculosis prostatitis. A history of prior superficial bladder cancer treated with BCG immunotherapy should prompt consideration for granulomatous prostatitis.
# "
,
* &
+ !
# "
!"3%#$ !%& '
() !" % $ *"+%," - . ' / '( ' ( ! %3#. $ .%!01.%" 2
!!" 44 $ ,,%# 3 '
! " # '( ' ( !! %#"$ 3#% 24
! $% ! # " 5 !$% $ ",#% 425
& $ ! ' 5 !$5 $ "&&%+, - .
'( ' ( ! %3 $ #*"%& 62 78
$% (7 !! $ #!%" 9 9
$
: !!*# $ &" %* 9 / 9
(
" 7 7 . !!&6 $ +#%+ ; 9
) * 5 !!+$4$ ",%, 4 4
"
(7 !!&6#$ ,, %
< '
5+
= ! $% $ &&%
, * & + !
# " ,
* & + #! # # ## #" #, #
>79
) ++"6#$ *&%&* 3 =
' (6 ++# # $ "!%+ ( ?(
,-.( / !
5++*4#$ ",%, 4
..
) & = ' ! $4$ !,%" 7=
(@ A
) ) 012 346 !!#%"&%,* : -
) 3* ) 5 ! 6%$ #*%" . ('(
( ) !!* 65$ #&,%+B #&+%+! 7=
(@ " 6 +&$ #,%"+ (
+ " () !!& 6%,$ * %*B ** - '
4
" " ! 5 ! 3#$ &+"%+! ;.@
.
) 5 !!*3%#$ ,!%" =;3
! '( ' ( ++, 3#"$ + #%* 4 5 7
! 5++ 5 $ !*% . 7;
+
! 5++# 5% $ ##+%, 3.
)
! 5++, %$ $ #&,%+! 9 C
' (' !" 3 $ &!%" 0 55 '
5 ! ! '( ' ( !!" 5"$ !*+%&# 4
6 3 ! 5 ! 3#$ *,%* 2 1'
)
* (7 !! $ #&*%+" 4 4
)
! 5 !! $ ,#%+ ;
+ ! "'( ' ( !! 53#$ * +%" A '
) ! 5 !!$6#$ *,% #
#* #& #+ "! "
" "# "" ",
9 4
) ! $ 5 ! 3$ $ ",*%* =
7
() ++, $# 6$ "!*%# 2 7
) ! 2" 2 5 !3 $ "% 1(A
7.8! ! *%3*@ ! "$ *"%,* 24
$ $ / : !!&#!$ &%+" .2 %8 :'
+ ! 3( ! 5 !!" #$ **% " . 5( 9 & 7 6 !!" 6#$ # &%"& D7 =C D : 3 (.A !!#$ #! 7
# / () !!, ,$ "*"%&
A
B
*
Figure 1 45-year-old man with Gleason score 6 (3+3) adenocarcinoma of the prostate. (A) Axial T2W image at the level of apex, which shows non-specific low signal in the right peripheral zone apex (arrow). There is moderate buldging of the prostatic capsule on the right and effacement of rectoprostatic angle (*). (B) ADC map demonstrates focal restriction in right apex (arrow) correlating with the low signal area on T2W images.
A
B
Figure 2 58-year-old male with targeted biopsy-confirmed far-anterior prostate adenocarcinoma of Gleason score 7 (4+3). This patient also has a left hip prosthesis. (A) Axial T2W image at the level of mid-gland, demonstrating an ill-defined far anterior lesion with capsular bulge (arrow). (B) ADC map at the same level demonstrating extensive metal artifact. Except for a small area in the right peripheral zone, the ADC map is un-interpretable and the anterior lesion seen on the T2W image cannot be evaluated.
A
B
C
D
b1000
b1400
Figure 3 72 year-old male with biopsy-proven Gleason score 7 (3+4) prostate adenocarcinoma. (A) Axial T2W image shows subtle ill-defined hypointensity in the left lateral mid gland peripheral zone (arrow). (B) ADC map shows equivocal, if any, diminished signal in the suspicious region. (C) DWI acquired at b=1000 s/mm2 , the lesion with subtle T2 signal abnormality is not apparent (arrow points to the expected location of the lesion). (D) DWI acquired at b=1400 s/mm2 , the lesion is now apparent (arrow). Note that spatial resolution becomes extremely poor at b=1400 s/mm2 with obscuration of the capsule and internal structures, also note the prominent “ghosting” artifact (between the arrowheads).
A
B
C
Figure 4 65 year-old male with targeted biopsy-proven bilateral adenocarcinoma of Gleason score 6 (3+3). (A) Axial T2W image demonstrates non-specific hypointensity within the mid gland peripheral zone bilaterally (arrows). (B) ADC at sub-optimal window (W = 3357, L = 1149), the lesions are not visible (arrows point to the expected location of bilateral lesions in this patient). (C) ADC map reveals focal restrictions after window optimization (arrows, W=1500, L = 1600).
A
B
D
E
C
Figure 5 60 year-old male with targeted biopsy-proven Gleason score 8 (4+4) in the right base to mid gland and prostatitis with calcifications in the left mid-gland. (A) Axial T2W image at the mid-gland level demonstrates T2 hypointense mass, as well as heterogeneous hypo-intense lesions in the left mid gland peripheral zone (arrowhead), which represent calcifications. (B) Coronal T2W image demonstrates T2 hypointense mass in the right superior base (arrow), as well as heterogeneous hypo-intense calcifications (arrowhead). (C) ADC map demonstrates focal restriction (arrow). (D) Axial DCE image at the right base shows focal early enhancement in the region of biopsy-proven malignancy (arrow). (E) Axial DCE image at mid-gland level shows focal early enhancement in the region of prostatitis with calcifications (arrowhead).
A
B
Figure 6 55 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) On this axial T2W image, the subcapsular crescentic tumor is indistinct from the prostatic capsule (arrow). (B) ADC map reveals focal restriction in a crescentic shape along the capsule, consistent with malignancy.
A
B
Figure 7 49 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) Axial T2W image in the distal apical region. An area of low signal (arrow) can be equivocally identified. (B) ADC map at the same level demonstrates focal restriction consistent with malignancy (arrows).
A
B
C
Figure 8 55 year-old male with targeted biopsy-proven Gleason score 7 (4+3) adenocarcinoma. (A) Axial T2W image at the far apical region demonstrates no obvious signal abnormality. (B) ADC map demonstrates focal restriction in the peri-urethral region (arrows). (C) DCE image demonstrates bright signal corresponding to the area with focal restriction, consistent with malignancy (arrow).
A
B
C
D
Figure 9 69 year-old male with targeted biopsy-proven Gleason score 7 (3+4) adenocarcinoma. (A) Axial T2W image. Arrow denotes the expected location of the abnormality, although it is not well visualized. (B) DCE image at the same level demonstrates focal early enhancing signal (arrow). (C) ADC map at the same level demonstrates focal restriction consistent with malignancy (arrow). (D) DWI with b=1400 s/mm2 , demonstrating bright signal consistent with malignancy (arrow).
A
B
C
Figure 10 65 year-old male with Gleason score 6 (3+3) diffuse infiltrative prostate cancer on histology. (A) Axial T2W image demonstrates no focal hypointense lesions. (B) ADC map demonstrates vague ADC restriction foci bilaterally. (C) DCE demonstrates no focal early enhancements, with vague enhancing regions seen bilaterally, a non-specific finding.
A
B
* *
*
*
C
*
*
Figure 11 Normal central zone. (A), (B) and (C), axial T2W, coronal T2W, and ADC map demonstrating the appearance of a normal central zone on a multiparametric MR exam (*), respectively.
A
B
C
D
Figure 12 Appearance of neurovascular bundle (NVB). (A) Axial T2W image at the level of low mid gland. Note the location of NVB at 5 o’clock and 7 o’clock locations in close proximity to the prostatic capsule (arrows). (B), (C) NVB at mid gland and superior mid gland levels (arrows), respectively. (D) ADC map, on which NVB appears as focal restriction (arrow). It is useful to visualize NVB across multiple slices on T2W images so that it is not mistaken for a subcapsular lesion.
A
B
C
D
Figure 13 51-year-old male who had a staging prostate MR for recently diagnosed bilateral prostate adenocarcinoma of Gleason score 8 (3+5). PSA = 6.3 ng/mL. (A) Axial T1W image shows diffuse bilateral bright signal indicative of post biopsy hemorrhage 3 weeks post biopsy (arrows). (B) Axial T2W image at the same level demonstrates diffuse low signal in the peripheral zone bilaterally with no dominant lesion. (C) ADC map shows no focal restriction. (D) ADC map from a repeat exam 8 weeks later demonstrates a large region of restriction in the right peripheral zone, as well as a smaller area of restriction anteriorly, consistent with malignancy (arrows).
A
B
C
Figure 14 75 year-old male with Gleason score 6 (3+3) adenocarcinoma at targeted biopsy of left transition zone. (A) Axial T2W image at mid-gland shows numerous areas of BPH without focal abnormality. (B) and (C) ADC map and extrapolated DWI (b = 2000 s/mm2 ) identified the lesion (arrows).
A
B
C
D
Figure 15 71 year-old male with history of superficial bladder cancer previously treated with intravesicle instillation of BCG, now presents with a rising PSA and abnormal DRE. Biopsy revealed granulomatous prostatitis. (A) Coronal and (B) axial T2W images demonstrate diffuse low signal throughout the prostate gland. (C) ADC map demonstrates diffuse diffusion restriction throughout the prostate gland. (D) DCE image demonstrates diffuse early enhancement.