Key words
prostate - MR-imaging - contrast agents - gadolinium - neoplasms
Introduction
Cancer of the prostate is the most frequent malignant tumor in men, and, after cancer
of the lung and colon, has the third-highest mortality rate. The diagnosis of prostate
cancer is based on digital-rectal examination, determination of the prostate-specific
antigen (PSA) and transrectal ultrasound-guided (TRUS) biopsy. In the case of TRUS-guided
biopsy, a total of 10 – 12 samples are taken from both lateral lobes of the prostate
following a systematic pattern. Nevertheless, in the case of 20 – 40 % of patients
suspected of prostate cancer, a definitive diagnosis of cancer can be ascertained
only after repeated biopsy, since a TRUS biopsy is limited by sampling error [1]
[2]. This represents a diagnostic problem, since a raised or increasing PSA value raises
the suspicion of prostate cancer; however the low negative predictive value of a TRUS
biopsy cannot definitively indicate that the patient actually has no cancer. A multiparametric
MRI has been shown to be a very sensitive and specific non-invasive method of localizing
potentially cancerous lesions within the prostate [3]
[4]. Multiparametric MRI includes conventional T2- and T1-weighted imagining combined
with diffusion imaging and or 1H-magnetic resonance spectroscopy and/or dynamic contrast-enhanced MR imaging (DCE-MRI).
Using this approach, T2-weighted imaging is indispensable, especially for interpretation
of changes in the transition zone [5]. Compared to T2-weighted imaging alone DCE-MRI provides a diagnostic advantage with
respect to prostate cancer detection [6]
[7]. Essential in this regard is dynamic information pertaining to contrast medium inflow,
since generally T1-weighted gradient echo sequences are used with a brief resolution
of < 5 s / measurement [8].
The first approved gadolinium-containing contrast medium for MR diagnosis was gadopentetate
dimeglumine (Gd-DTPA; Magnevist, Bayer Healthcare, Berlin, Germany). Consequently,
Gd-DTPA was the contrast medium of choice for DCE-MRI of the prostate, as it was for
all other organ systems [3]
[7]
[9]. Since the introduction of Gd-DTPA, a number of additional contrast agents have
been approved which are distinguished by their chemical properties. Consequently,
studies have been required to test these contrast media and compare them with respect
to their relative advantages and disadvantages [10]
[11]
[12]
[13]. For safety reasons, the occurrence of nephrogenic systemic fibrosis (NSF) accompanying
the use of MR contrast media with linear complexes brought to the forefront more thermodynamically
stable contrast media with macrocyclic complexes [14]
[15]. To-date, there has been no systematic comparison of the use of macrocyclic and
linear contrast media in prostate DCE-MRI, and therefore no data indicating whether
the application of gadobutrol is as reliable as Gd-DTPA in the diagnosis of prostate
cancer. Gadobutrol differs from Gd-DTPA by its macrocyclic structure, charge neutrality,
its doubled molarity as well as a higher T1 relativity (Gd-DTPA r1 = 4.1 mmol-1s-1, Gadobutrol r1 = 5.2 mmol-1s-1) [10]
[16]. T1 relativity of a substance is a measure of the signal-increasing effect on T1-weighted
sequences, as already demonstrated in numerous studies of various organ systems [17]
[18]
[19]
[20]. The purpose of this study is to delineate the differences between both contrast
media in their enhancement behavior according to contrast medium dosage and the diagnostic
value of the prostate DCE MRI based on this comparison, as well as the evaluation
of the signal intensity time (SI-t) progression of the macrocyclic contrast medium
gadobutrol with the most commonly agent Gd-DTPA in prostate DCE-MRI. Proceeding from
results of studies of other organ systems [12]
[19]
[20], this study is based on the hypothesis that gadobutrol provides a higher relative
enhancement in prostate tissue, but that there is no qualitative difference in the
enhancement progression.
Materials and Methods
Study Population
These studies included patients with histologically verified cancer of the prostate
who were examined between 1/2009 and 12/2010 with multiparametric MRI of the prostate
using Gd-DTPA or gadobutrol with 1.5 T (Magnetom Avanto, Siemens, Erlangen, Germany),
and who subsequently underwent prostatectomy. Inclusion criteria in this retrospective
study were prostatectomy after histologic verification of prostate cancer, a multiparametric
MRI performed according to standardized institute protocol; examination was performed
within a week prior to the prostatectomy; dosage of Gd-DTPA at a flow rate of 4 ml/s
or gadobutrol at 2 ml/s for the DCE-MRI. An additional inclusion criterion was a carcinoma
map created by a consensus of the pathologist and radiologist based upon the pathological
prostate sample. After the prostatectomy the entire prostate sample was marked in
color (anterior, posterior, left, right) and sliced perpendicular to the longitudinal
axis from the apex to the base (5 – 8 slices). The individual slices were quartered
into individual slices according to standard protocols, embedded in paraffin and stained
with hematoxylin and eosin. All pathologically verified carcinomas > 5 mm as well
as areas with a normal peripheral zone were graphically transferred to standardized
drawings by the pathologist. All tumor patterns were assessed using the modified Gleason
Scoring System [21]. Then, based upon the consensus of the pathologist and radiologist, the carcinoma
foci and normal peripheral zone were separately plotted and coded on printed T2w images.
As needed, anatomical landmarks were relied upon for exact identification of the tumor
foci on the most closely corresponding MR images; these included zonal structures,
prostatic urethra, bladder, ejaculatory duct and seminal vesicles. The serum creatinine
and serum PSA values were determined in all patients prior to the MRI examination.
The local ethics commission approved this study. Altogether 28 patients were included
in the study who had been examined using Gd-DTPA (Magnevist; Bayer Healthcare, Berlin,
Germany) – hereafter designated Group A, and 25 patients who had been examined using
gadobutrol (Gadovist; Bayer Healthcare, Berlin, Germany) – hereafter designated Group
B. A normally classified T2w hyperintense area in the peripheral zone of all patients
was previously analyzed in the Pathology department. In addition, 42 cancerous areas
in Group A and 34 in Group B were assessed. The median age in Group A (Gd-DTPA) was
62.0 ± 5.9 years, and 62.3 ± 5.7 years in Group B (gadobutrol). The PSA value in both
groups was not statistically different (p > 0.05), with an average value of 8.6 ± 4.8 ng/ml
for Group A, and 9.7 ± 7.6 ng/ml for Group B. The median Gleason score for prostate
cancer was 7 in both groups.
Multiparametric MRI with Dynamic Contrast Medium Examination
Using a combined 6-channel surface coil and an endorectal coil (Medrad Prostate eCoilTM, Bayer Healthcare, Berlin, Germany), all patients received a multiparametric MRI
of the prostate according to an identical examination protocol by means of a commercially-available
MRI unit (1.5 T Magnetom Avanto, Siemens, Erlangen, Germany). After localizer sequences,
the prostate was examined using T2-weighted Turbo-Spin Echo (TSE) sequences in axial
orientation (repetition time/echo time in msec, 4850/85; echo train length, 15; number
of signals acquired, 3; field of view, 160 × 160 mm) and coronal scan orientation.
In addition, a T1-weighted sequence was measured across the prostate in the axial
direction (691/12; echo train length, 3; number of signals acquired, 2; field of view,
160 × 160 mm). All sequences were measured using a matrix of 256 × 256, section thickness
of 3.0 mm, slice increment of 0.6 mm, and 100 % phase oversampling.
Maintaining the angulation of the T2-TSE, a diffusion-weighted as well as a DCE-MRI
were additionally performed. In order to calculate longitudinal relaxation rate R10
[22]
[23], a gradient echo sequence was measured with the following parameters, each with
a flip angle of 2°, 5°, 10°, and 15° (4.57/1.63; number of signals acquired, 5; field
of view, 260 × 260 mm; image matrix, 256 × 256; section thickness, 3.6 mm). The DCE-MRI
was performed using a fast gradient echo sequence was measured with a temporal resolution
of 3.9 seconds (5.19/2.02; flip angle, 15°; number of signals acquired, 1; field of
view, 260 × 260 mm; image matrix, 256 × 256; section thickness, 3.6 mm). In total,
75 individual measurements were acquired; after the third measurement, the contrast
medium was adapted to body weight (0.1 mmol/kg body weight) and applied intravenously
as a bolus. Since gadobutrol 1 M is double-concentrated compared to Gd-DTPA, a patient
weighing 70 kg received 7 ml gadobutrol and 14 ml Gd-DTPA applied intravenously. In
order to reconcile the injection time of both groups, gadobutrol was injected using
a halved flow of 2.0 ml/s compared to a rate of 4 ml/s using Gd-DTPA. After injection
of the contrast medium, there was an injection of 20 ml isotonic saline solution while
maintaining the flow rates.
Using the open two-compartment model according to Tofts and Kermode [24], the pharmacokinetic parameter maps were calculated for the exchange components
Ktrans and kep.
Evaluation
Using evaluation software (syngo Tissue 4 D, Siemens, Erlangen, Germany) the DCE-MRI
was registered to the morphological images (T2w) in order to exactly localize the
tumor area and normal peripheral zone based on the carcinoma map. Circular fields
were used to indicate tumor patterns and normal peripheral zones (BV). The SI-t enhancement
curves in the marked fields were calculated as a relative SI-t enhancement of the
initial SI without contrast medium, and the individual values were exported over time
to an external computer. In addition, the 3-dimensional pharmacokinetic parameter
maps of the prostate were calculated using the volume element marking containing the
entire prostate. Parameters Ktrans and kep were determined in the previous established fields and likewise exported ([Fig. 1]).
Fig. 1 a Axial T2-TSE with a ventral leftsided prostata cancer (PCa) of the transition zone
(Gleason 3 + 4 = 7). Referencing the prostatectomy specimen tumors were marked in
printed T2-TSE images; b regions of interest were placed in the normal peripheral zone (green) and the PCa
(red) to measure relative enhancement in DCE-MRI and calculate pharmacokinetic parameter
maps – Ktrans-parameter map shown here. c The resulting enhancement-curve in the peripheral zone (type II, plateau) and the
PCa (type III, wash-out).
Based on the values for the relative SI enhancement of the tumors and healthy peripheral
zone per time unit, all curves per patient were calculated, blinded and printed out.
A radiologist (TD) categorized all curves as Type I, II or III, whereby a Type I curve
reflected a steady increase of the enhancement progression; Type II reflected a plateau,
and Type III corresponded to a decline in relative enhancement after the peak [25]. These results were then designated (tumor, normal tissue in the peripheral zone,
contrast medium) and statistically evaluated [CS, TD].
Statistics
Quantitative data are provided as the average of the least squares from mixed linear
models with 95 % confidence intervals. Qualitative data are represented as absolute
and relative frequencies. The curves between the contrast media were analyzed using
multinomial regression analysis in order to compensate for repeated observations of
a patient. Relative peak enhancement was defined as the maximum enhancement in the
temporal progression of dynamic measurement. Using mixed linear models, this was compared
with variance components. Tissue types (tumor, normal peripheral zone), contrast medium
and the interaction of both factors served as cofactors. Using the Wilcoxon rank-sum
test, the two contrast agents were compared with one another. The temporal curve progressions
were compared using linear models including all individual measured values over the
entire duration of the measurement. Analogously, the pharmacokinetic parameters Ktrans and kep were compared between the contrast media. Here again, tissue type and interaction
of tissue type and contrast medium served as cofactors. In a further analysis, the
tumors were examined under stratification in Gleason score ≤ 6 and ≥ 7. All tests
were performed bilaterally, p-values < 0.05 were considered significant. SAS 9.2 (SAS
Institute Inc., Cary, NC, USA) was used to perform the analyses.
Results
Qualitative Analysis: Increase (Type I), Plateau (Type II) and Wash-out (Type III)
Statistically, the frequency of curve types between the contrast media did not differ
(p = 0.63; [Table 1]). In the normal peripheral zone a Type I curve (steady increase) was shown in 46.4 %
of cases in Group A, and in 48.0 % of cases in Group B. A Type II curve (Plateau)
was ascertained among 53.6 % (Group A) and 52.0 % (Group B). No Type III curve (Wash-out)
could be observed in either Group A (Gd-DTPA) or Group B (gadobutrol). In 20.6 % of
the cases, gadobutrol resulted in a wash-out in the tumor, whereas with Gd-DTPA, a
16.3 % rate was not significantly lower. Tumors in Group B (gadobutrol) more frequently
demonstrated a Type II curve than in Group A (Gd-DTPA) and correspondingly fewer Type
I curves; this was, however, statistically insignificant (p = 0.75; [Table 1]).
Table 1
Curve type distribution.
|
|
type I
|
type II
|
type III
|
p-value
|
Gd-DTPA
|
PZ
|
13 (46.4)
|
15 (53.6)
|
0
|
0.75
|
Gadobutrol
|
12 (48.0)
|
13 (52.0)
|
0
|
Gd-DTPA
|
PCa
|
21 (48.8)
|
15 (34.8)
|
7 (16.3)
|
Gadobutrol
|
12 (35.3)
|
15 (44.1)
|
7 (20.6)
|
p-value
|
|
0.63
|
|
Distribution of enhancement curves of Type I, II and III in normal peripheral zone
(PZ) and prostate cancer (PCa) using Gd-DTPA and gadobutrol; absolute number and percent
in parentheses.
In addition, the tumors, based on Gleason scores, were divided into two groups (Gleason
score ≤ 6 and ≥ 7) in order to evaluate possible differences between the contrast
media in relation to Gleason scores. Consequently it was shown that use of both contrast
agents achieved a very similar frequency distribution with respect to curve types;
there were no significant differences between the contrast media for tumors with a
Gleason score of ≤ 6 or ≥ 7 ([Table 2], p = 0.43). Tumors with a Gleason score of ≥ 7, compared to tumors with a score
of ≤ 6, more frequently demonstrated a statistically significant Type III curve for
both contrast media ([Table 2], p = 0.02).
Table 2
Curve types according to Gleason.
|
|
type I
|
type II
|
type III
|
p-value KM
|
p-value Gleason
|
Gd-DTPA
|
GS ≤ 6
|
24 (53.3)
|
19 (42.2)
|
2 (4.4)
|
0.43
|
0.02
|
Gadobutrol
|
18 (45.0)
|
19 (47.5)
|
3 (7.5)
|
Gd-DTPA
|
GS ≤ 7
|
10 (38.5)
|
11 (42.3)
|
5 (19.2)
|
Gadobutrol
|
6 (31.6)
|
9 (47.4)
|
4 (21.5)
|
Frequency of enhancement curve Types I, II and III in prostate cancer of Gleason score
(GS) ≤ 6 and ≥ 7 using Gd-DTPA and gadobutrol, absolute number and percent in parentheses.
Quantitative Analysis: Relative Peak Enhancement and Curve Gradient
The relative SI maximum in normal tissue was 1.4 times [1.20; 1.59] the initial value
in the Gd-DTPA group, whereas this value in the gadobutrol group was significantly
higher, with 1.58 [1.37; 1.78] (p = 0.04). This behavior was similar in tumor tissue.
When Gd-DTPA was used, an average peak value of 1.56 [1.41; 1.71] was achieved, whereas
with gadobutrol a statistically significant higher value of 1.76 [1.59; 1.94] was
achieved (p = 0.04, [Table 3]). These differences were additionally analyzed using a mixed linear model with the
inclusion of all measured values in the entire temporal progression. In this case,
no distinction could be established between the contrast media in the normal peripheral
zone. However, in tumor tissue a trend toward higher relative enhancement values was
evident across the entire temporal progression when gadobutrol (p = 0.05) was used
([Fig. 2]). For both tissue types, the shapes of the curves (interaction of contrast agent
and time) for Gd-DTPA and gadobutrol varied significantly (peripheral zone p = 0.003,
tumor tissue p = 0.0003). Observation of the resulting curves in the different tissues
makes it clear that the rise of the enhancement curve when gadobutrol is used is greater
than when Gd-DTPA is utilized, although the further progression is the same (except
for a parallel shift upward in the case of the gadobutrol tumor enhancement curve),
as shown in [Fig. 2].
Table 3
Peak enhancement.
|
PZ
|
PCa
|
p-value
|
Gd-DTPA
|
1.40 [1.20;1.59]
|
1.56 [1.41;1.71]
|
0.04
|
Gadobutrol
|
1.58 [1.37;1.78]
|
1.76 [1.59;1.94]
|
Relative peak enhancement in DCE-MRI using Gd-DTPA and gadobutrol in normal peripheral
zone (PZ) and prostate cancer (PCa).
Fig. 2 Relative enhancement in prostate cancer (PCA) and normal peripheral zone tissue using
gadobutrol and Gd-DTPA. Peak enhancement was significantly higher using gadobutrol
in both PCa and the normal peripheral zone (p = 0.04). Comparing the entire observation
period, a trend towards higher enhancement using gadobutrol was observed in PCa but
not in the peripheral zone. The curve shapes were significantly different caused by
a faster enhancement increase using gadobutrol (peripheral zone p = 0.003; PCa p = 0.0003).
Pharmacokinetic Parameters Ktrans and kep
For all patients the pharmacokinetic parameters Ktrans and kep were compared between the contrast media. No statistically significant differences
for Ktrans and kep could be determined for the contrast agents (Ktrans p-value = 0.41, kep p-value = 0.65) ([Table 4]). The use of Gd-DTPA as well as gadobutrol demonstrated significantly higher pharmacokinetic
parameters Ktrans and kep in tumor tissue than in normal tissue ([Table 4], Ktrans p-value < 0.0001; kep p-value 0.002). Further, when considering the Gleason categories, parameter Ktrans for tumors with a Gleason score of ≥ 7 when Gd-DTPA is used (0.12 [0.09; 0.15]) and
gadobutrol (0.14 [0.10; 0.17]) was significantly higher than for tumors with a Gleason
score of ≤ 6 (Gd-DTPA (0.082 [0.04; 0.12] and gadobutrol (0.079 [0.04; 0.12]); p = 0.02);
however, once again there was no statistically significant difference between the
contrast media (p = 0.60).
Table 4
Pharmacokinetic parameters.
|
|
PZ
|
PCa
|
p-value KM
|
p-value tissue
|
Gd-DTPA
|
Ktrans
|
0.05 [0.03; 0.07]
|
0.11 [0.09; 0.12]
|
0.408
|
< 0.0001
|
Gadobutrol
|
0.06 [0.04; 0.09]
|
0.11 [0,09; 0,13]
|
Gd-DTPA
|
kep
|
0.20 [0.13; 0.27]
|
0.31 [0.25; 0.37]
|
0.651
|
0.002
|
Gadobutrol
|
0.22 [0.14; 0.30]
|
0.33 [0.26; 0.39]
|
Pharmacokinetic parameters (Ktrans und kep) in normal peripheral zone (PZ) and prostate cancer (PCa) using Gd-DTPA and gadobutrol.
Discussion
A comparison of various MR contrast media with respect to SI-t progression and detection
of lesions had previously been performed in neuroradiological and cardiovascular magnetic
resonance imaging as well as in MR mammography [12]
[18]
[19]
[20]. A potential difference of MR contrast agents in DCE-MRI of the prostate had not
been previously performed. Therefore, the purpose of this investigation was to analyze
the use of gadobutrol and Gd-DTPA in DCE-MRI in order to evaluate possible effects
on the resulting curve types and pharmacokinetic parameters.
Several characteristics principally differentiate gadobutrol and Gd-DTPA. While Gd-DTPA
is a linear and ionic magnetic resonance contrast medium, gadobutrol has a macrocyclic
structure and is charge-neutral [10]. Gd-DTPA has a concentration of 0.5 M, whereas gadobutrol has a 1 M concentration.
Thus, examinations using gadobutrol were performed using half the flow compared to
DCE-MRI with Gd-DTPA in order to administer the identical volume of contrast medium
during the same time period. In addition, gadobutrol has about one-fourth higher T1
relaxivity (1.5 T) compared to Gd-DTPA [16].
Compared to examinations using Gd-DTPA, the current study determined that the gadobutrol-enhanced
DCE-MRI demonstrated a higher relative peak enhancement in the healthy peripheral
zone of the prostate and tumor tissues. Taking into account all measured values over
the entire time progression, a greater increase in the relative enhancement curve
was observed in the gadobutrol group, in addition to a trend toward a higher relative
enhancement. These differences can be explained by the higher T1 relaxivity of gadobutrol.
This had already been described in magnetic resonance mammography, MRI of brain metastases
or infectious lesions, as well as in MR imaging using gadobutrol and other extracellular
contrast media with higher relaxivity [12]
[19]
[20]. Additional factors such as charge differences of each contrast medium are also
considered as potential influential factors for enhancement behavior. However, charge
differences are primarily relevant in so-called delayed imaging, e. g. in cartilage
imaging (dGEMRIC) [26]. Within this study we observed SI-t progression of the DCE-MRI for about 5 minutes
immediately after administration of the contrast medium. We primarily observed differences
in peak enhancement values and increase of the initial enhancement, so that an effect
of the charge differences between both contrast agents is probably negligible in this
context.
Differences were also found regarding the number of detected lesions during imaging
of cerebral lesions of mammary carcinomas [20]
[27]
[28]. This was not reviewed in the current study. In our opinion, with respect to multiparametric
MRI of the prostate, DCE-MRI primarily aids in increasing the specificity as well
as the detection and characterization of the most aggressive lesion [29]
[30]
[31]. In this investigation we could not establish a difference in frequency among curve
types I, II or III. Based on our data, we presume that both contrast media do not
qualitatively differ from one another to a clinically relevant extent in either a
healthy prostate or tumor tissue with respect to enhancement behavior, since the resulting
curve types in both groups exhibit similar frequency. Also in comparing Gleason scores,
no statistically significant differences could be determined between the two groups.
These results are substantiated by a comparative analysis of the pharmacokinetic parameters.
With respect to Ktrans and kep values, no statistical difference between these contrast agents could be found.
The study additionally showed that 48.8 % of tumors in Group A (Gd-DTPA) and 35.3 %
in Group B (gadobutrol) demonstrated a Type I curve. Of these 38.5 % of tumors in
Group A and 31.6 % in Group B had a Gleason score of ≥ 7. In the recently published
ESUR guidelines for structured assessment and classification of prostate carcinomas,
suspected tumorous areas are graded from 1 (significant carcinoma highly unlikely)
to 5 (clinically significant carcinoma highly likely) [32]
[33]. The purpose of this study was not to evaluate the PIRADS criteria. Nevertheless,
it should be pointed out that more than 30 % of tumors with a Gleason score of ≥ 7
demonstrate a Type I curve, which, according to the current PIRADS classification
in DCE-MRI is only assigned a point value of 1. Additional criteria such as focality
and asymmetry are only provided for a Type II or Type III curve [32]. During a pending evaluation of PIRADS criteria, therefore, these additional criteria
should also be applied to tumors showing a Type I curve, analogously to magnetic resonance
mammography [34].
This study is limited by its retrospective approach and interindividual comparison.
However, all patients undergoing prostatectomy are treated so that for each carcinoma
the best available histological standard is applied with respect to diagnosis and
Gleason grading. Due to the required double contrast medium dose and the resulting
time delay, intraindividual comparison is difficult to perform in the case of patients
with histologically verified carcinoma and scheduled prostatectomy.
On the whole it can be concluded that both gadobutrol and Gd-DTPA appear to be highly
suitable for DCE-MRI. Relative enhancement in DCE-MRI using gadobutrol tends to be
higher; in the course of this study no statistically significant difference could
be found with respect to curve type frequencies and the pharmacokinetic parameters
Ktrans and kep.
Clinical Relevance
No statistically significant differences were found between both contrast medium types
in the curve type frequencies and pharmacokinetic parameters.
Both the macrocyclic gadobutrol and the linear contrast agent Gd-DTPA appear to appear
to be well-suited for DCE-MRI of the prostate for the diagnosis of prostate cancer.