t2 mapping in prostate cancer - clinical mriclinical-mri.com/wp-content/uploads/2019/03/t2... · t2...

Dow

nloadedfrom

https://journals.lww.com

/investigativeradiologyby

wZm

ankU5H

cfvZW3rs+aU

YHwivBH

BM4VekjPiq0N

SYUbm

+0QKw

TJ37JnH4BgT/Iqvjj2nQ

kOaPp5qsU

0Y/UMsb1YseyXE6vH

eYCjN5dw

Fklua1zAQ6TvR

MDUz3ym

9sKpJon

02/11/2019

Downloadedfromhttps://journals.lww.com/investigativeradiologybywZmankU5HcfvZW3rs+aUYHwivBHBM4VekjPiq0NSYUbm+0QKwTJ37JnH4BgT/Iqvjj2nQkOaPp5qsU0Y/UMsb1YseyXE6vHeYCjN5dwFklua1zAQ6TvRMDUz3ym9sKpJon02/11/2019

T2 Mapping in Prostate CancerJulia Mai, Mohamed Abubrig, MD, Thomas Lehmann, PhD, Tom Hilbert, PhD, Elisabeth Weiland, PhD,

Marc O. Grimm, MD, Ulf Teichgräber, MD, and Tobias Franiel, MD

Objectives: The aim of the study was to determine the quantitative T2 values inprostate tissue and evaluate them for detection and grading of prostate cancer.Materials andMethods:After approval from the local ethics committee, morpho-logical T2-weighted (T2w) images, apparent diffusion coefficient (ADC) mapsfrom diffusion-weighted images, quantitative T2maps, and calculated T2w imagesfrom 75 men (median age, 66.3 years; median PSA, 8.2 ng/mL) were acquired at3 T magnetic resonance imaging (MRI). Data were retrospectively evaluated fortheir distinction between prostate pathologies.Eight hundred fifty-seven areas of normal gland (n = 378), prostate cancer

(54x Gleason score 6, 98x Gleason score 7, 25x Gleason score 8), benign prostatichyperplasia (BPH) nodes (n = 150), prostatitis (n = 119), and precancerous lesions(n = 33)were determined on calculated andmorphological T2w images.Histologicalcriterion standards were whole gland sections (16 patients), MRI-guided in-bore bi-opsies (32 patients), MRI/transrectal ultrasound-fusion biopsies (15 patients), andsystematic 12-core transrectal ultrasound-guided biopsies (12 patients). Significancewas assumed to be P < 0.05.Results: The quantitative T2 values vary significantly between prostate cancerand normal gland tissue (area under the curve [AUC], 0.871), cancer and BPHnodes (AUC = 0.827), and Gleason score 6 and 7 or higher (AUC, 0.742). Thequantitative T2 values decrease with increasing Gleason scores and correlate sig-nificantly with the ADC values (r = 0.806).The detection accuracy of prostate cancer on calculated (AUC = 0.682) and

morphological T2w images (AUC = 0.658) is not significantly different.Conclusions:Quantitative T2 values seem to be suitable for distinguishing betweenprostate cancer and normal gland tissue or BPH nodes. Similar to the ADC values,they offer an indication of the aggressiveness of the prostate cancer.

Key Words: magnetic resonance imaging, prostate cancer, T2 mapping,multiparametric MRI, quantitative T2, prostate cancer aggressiveness,diffusion imaging, apparent diffusion coefficient, Gleason score, 3 T

(Invest Radiol 2019;54: 146–152)

S tate-of-the-art magnetic resonance (MR) diagnosis of the prostate isbased on T1-weighted and T2-weighted (T2w) imaging, diffusion-

weighted, and dynamic contrast-enhanced sequences. T2-weighted im-aging is of particular significance because it provides a good contrastbetween anatomical characteristics of the organ. Therefore, T2w im-ages are of particular help in concluding a diagnosis. In contrast toquantitative T2 values from T2maps, which are based on multiple echotimes, the T2w signal intensities may vary locally as a result of radiofre-quency inhomogeneities of the transmit and receive coil, and are only aqualitative measure.

Quantitative T2 values are ideally independent from the hard-ware, as they are acquired by several echo times and reflect the absoluterelaxation of the protons contained in the tissue voxels regardless oftheir relative position to the coil. It therefore seems sensible to include

the absolute T2 relaxation values in the diagnosis of prostate cancerand to investigate their diagnostic accuracy.

A number of studies have already taken up this approach1–5;however, most of these studies examined the diagnostic accuracy ofquantitative T2 values for prostate cancer only in the peripheral zone(PZ). For the transitional zone (TZ), in which the distinction betweenbenign prostatic hyperplasia (BPH) and prostate cancer can be difficult,only limited data are available yet.1,5 For this reason, an absolute param-eter that can give indications for or against cancer would be highly ben-eficial in that case.

Furthermore, there have been indications that the quantitative T2values correlate with the aggressiveness of the prostate cancer, whichcan be characterized by the Gleason score.5,6 With regard to this, thequestion arises of whether T2 maps can be used to differentiate con-cretely between the individual Gleason scores.

The aim of this work is to determine the quantitative T2 values ofprostate cancer with the Gleason score 6, 7, and 8, normal prostate tis-sue, prostatitis, BPH, and precancerous lesions for PZ and TZ separated,and their evaluation with regard to detection, distinction, and grading ofprostate cancer in comparison to the corresponding apparent diffusioncoefficient (ADC) values, as well as the suitability of the T2w calculatedfrom quantitative T2 values for diagnosis compared with the acquiredmorphological T2w images.

MATERIALS AND METHODS

Study Design and PopulationFor this retrospective, single-center cohort study, approved by

the local ethics committee, 86 men with the following inclusion criteriawere consecutively included in our study between January 1, 2015,and January 10, 2016. Inclusion criteria were that first, the patient hadreceived all magnetic resonance imaging (MRI) sequences relevant tothe study as part of the multiparametric MRI (mpMRI) examination(T2w, T2 maps, and diffusion-weighted imaging), and second therewas a histopathological criterion standard in the form of a biopsy ora prostatectomy. Exclusion criteria were nonrepresentative prostate bi-opsies (n = 2), a time interval of more than 7 months between biopsyand MRI (n = 8), and an unusable diffusion-weighted imaging se-quence (n = 1). Finally, we considered 75 patients to be a study popu-lation (Table 1, Fig. 1). They had not been included in any other studypublished yet.

Magnetic Resonance ImagingAll patients received an mpMRI of the prostate at 3 T

(MAGNETOM Prisma; Siemens Healthcare Erlangen, Germany) using abody phased array receiver coil (18-channel design with 18 integrated pre-amplifiers,with3rowsof6elementseach,dimensions:385�590�65mm[length � width � height]) and into the table integrated 32 channel spinearray receiver coil.

Siemens Healthcare supported the study by providing 2 softwareprototypes such as model-based accelerated T2 mapping and ZoomITsingle-shot diffusion epi.7,8 The following sequences were acquired inthe study: axial fast spin-echo T2w, axial diffusion-weighted imaging,as well as axial quantitative T2 maps (Table 2). Synthetic T2w imageswere created based on the T2 maps using the same echo time as the

Received for publication June 29, 2018; and accepted for publication, after revision,September 3, 2018.

From the Department of Radiology, University Hospital Jena, Jena, Germany.Conflicts of interest and sources of funding: no funding.Correspondence to: Julia Mai, Department of Radiology, University Hospital Jena,

Am Klinikum 1, 07747 Jena, Germany. E-mail: [email protected] © 2018 Wolters Kluwer Health, Inc. All rights reserved.ISSN: 0020-9996/19/5403–0146DOI: 10.1097/RLI.0000000000000520

ORIGINAL ARTICLE

146 www.investigativeradiology.com Investigative Radiology • Volume 54, Number 3, March 2019

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.

mailto:[email protected]

www.investigativeradiology.com

morphological T2w. Apparent diffusion coefficient values were used asa reference quantitative parameter, because DWI is well known and rou-tinely used in prostate cancer diagnostic imaging.

Histopathologic AnalysisSixteen (21.3%) of the 75 patients received a DaVinci robot–

assisted prostatectomy. The preparations were all cancerous and wereprocessed in exactly the same way as the method published in 2011.9

Thirty-two patients (42.7%) underwent a targeted, MR-guided in-borebiopsy using a 1.5 T MRI in accordance with the previously publishedmethod.10,11 Four of these patients had prostate cancer in at least 1 ofthe 2 to 4 extracted samples. No cancer was found in 28 patients.

The remaining 27 patients (56%) received a transrectal ultra-sound-guided biopsy either as a 12-core systematic biopsy (12 of 27)or as an elastic fusion biopsy (15 of 27) with an additional 1 to 3 biop-sies from the MR suspected areas.11 Of these 27 patients, 10 (37%)were positive and 17 (63%) negative for prostate cancer.

Determination of Quantitative T2 and ADC Values forCorrelation With Histopathology

To determine the quantitative T2 and ADC values, lesion-based regions of interest (ROIs) were placed according to thehistopathologic reference.12,13

For the patients with an available whole-mount section, the in-vestigated areas were defined through histopathological findings andthen transferred to the T2 maps and DWI sequences (Fig. 2). On aver-age, we placed 23 (7 to 33) ROIs in the whole prostate.

For the other patients who had an MRI-guided in-bore biopsy,fusion biopsy, or systematic TRUS-guided biopsy, both parameterswere determined only for the biopsied areas.

For the MR-guided biopsies, the precise location of the biopsywas known and documented, and therefore these lesions could be di-rectly transferred to the T2 map and the DWI (Fig. 3). The systematic

TABLE 1. Study Population

Feature Data

Age, yMean (range) 66.3 (48–79)

Serum prostate specific antigen, ng/mLMedian (range) 8.2 (4.4–52.8)

Time between MRI and MR-guided biopsy, dMedian (range) 2 (2–142)

Time between MRI and transrectalultrasound-guided biopsy, dMedian (range) 23 (2–210)

Time between MRI and prostatectomy, dMedian (range) 54.5 (1–176)

MRI indicates magnetic resonance imaging.

FIGURE 1. Study flow chart with inclusion and exclusion criteria.

TABLE 2. MRI Acquisition Parameters

MorphologicT2w FSE

Diffusion-WeightedImaging T2 Mapping

Sequence Spin-echo EPI Multiechospin-echo

Acquisitionduration, min:s

06:19 04:36 07:09

Repetition time, ms 8430 4000 6820Echo time, ms 142 58 10, 20, 30, 40,

50, 60, 70, 80,90, 100, 110,120, 130, 140,150, 160

Slice thickness, mm 3 3 3Acquisition matrix 320 � 310 90 � 72 320 � 320Field of view, mm 160 � 160 160 � 128 260 � 260No. averages 4 01.01.36 1Parallel imagingfactor

2 2 2

Nominal voxelsize, mm3

0.5 � 0.5 �3.0

0.9 � 0.9 �3.0

0.8 � 0.8 �3.0

b-values, s/mm2 NA 50/1000 NADiffusion mode NA 3D diagonal NANo. echoes 1 1 16Undersamplingfactor

NA NA 5

MRI indicates magnetic resonance imaging; T2w, T2-weighted; FSE, fastspin-echo; NA, not applicable.

Investigative Radiology • Volume 54, Number 3, March 2019 T2 Mapping in Prostate Cancer

© 2018 Wolters Kluwer Health, Inc. All rights reserved. www.investigativeradiology.com 147



TRUS-guided biopsies were always taken according to a fixed scheme.14

A peripheral and a medial sample were taken in 3 different craniocaudaldirections (apical, medial, and basal) on each side of the prostate and ac-cordingly the ROIs were placed.

For the corresponding RIOs, the mean T2 and ADC values werecalculated in the T2 maps and DWI. These mean values were deter-mined in each patient for the following tissue types: prostate cancerof different Gleason scores as well as for normal gland tissue, prostati-tis, BPH nodes, and precancerous lesions (atypical small acinar prolif-eration [ASAP], low- and high-grade prostatic intraepithelial neoplasia[PIN]), in each case separately for the PZ and TZ.

Image Quality Assessment of Morphological andCalculated T2w

Two readers (reader 1, radiologist with more than 10 years of ex-perience with prostateMRI; reader 2, 1 year of experiencewith prostateMRI) assessed both sequences independently in a randomized orderand blinded to the indication of mpMRI examination, clinical, and histo-pathological results. Nine criteria of image quality were rated on a Likert

scale of 1, poor, to 5, excellent. These criteriawere general image quality,contrast between PZ and TZ, representation of the inner architecture ofthe PZ as well as the TZ, visibility of the lesion with the highest PIRADSscore, presentation of the capsule, presentation of the neurovascularbundle, and the representation of the rectoprostatic angle.15 Further-more, in the images we evaluated the risk of extracapsular growth from1, definitely not present, to 5, definitely present, and the severity ofmotion artifacts.

Tumor Detection on the Morphological andCalculated T2w

With the 36-sector model according to PI-RADS Version 2, ascore of 1 (most likely benign) to 5 (most likely malignant) was givenfor each sector in both image series.16 Furthermore, each sector wasassigned with the respective histopathological criterion standard (normal,BPH, prostatitis, ASAP, PIN, or prostate cancer), if available. The sectorswithout histology were not considered in the statistical analysis.

FIGURE 2. A, Whole gland section. B, Axial T2 map. C, axial ADC map.

FIGURE 3. A, Whole gland during MR-guided in-bore biopsy. B, Axial T2map. C, Axial ADC map.

Mai et al Investigative Radiology • Volume 54, Number 3, March 2019

148 www.investigativeradiology.com © 2018 Wolters Kluwer Health, Inc. All rights reserved.



Statistical AnalysisThe statistical analysis was performed with SPSS 24 (SPSS for

Windows, SPSS Inc, Chicago, IL). The mean T2 values were correlatedwith the mean ADC values using a Spearman correlation. The diagnos-tic accuracy of T2 and ADC values was determined with receiver oper-ating characteristic curves, binary logistic regression models with theT2 and ADC values as predictors, and DeLong test in SAS 9.4 (SASInstitute, Cary, NC). Significance was assumed to be P < 0.05.

RESULTSEight hundred fifty-seven ROIs with the respective mean T2 and

mean ADC values, from the 75 patients were included in the statisticalanalysis (Table 3).

Quantitative T2 Values for the Differentiation BetweenProstate Cancer and Other Prostate Pathologies

Based on the quantitative T2 values, a significant distinction canbe made between prostate cancer and other entities in both PZ and TZ(Table 4). This is best achieved for the distinction between prostate can-cer and normal gland tissue (AUC, 0.871 [0.840–0.902]; P < 0.01), butalso benign altered prostate tissue. For example, a BPH node or prosta-titis can be significantly distinguished from prostate cancer (AUC, 0.831

[0.778–0.884]; P < 0.01). Normal prostate tissue and benign tissue alter-ations do not show significant differences in T2 values (Fig. 4). With an85% sensitivity, this results in a cutoff of 134 milliseconds (specificity65%) in the PZ, and 104 milliseconds (specificity 68%) in the TZ to dif-ferentiate between prostate cancer and normal prostate tissue.

Comparison of the Quantitative T2 Values With theADC Values

The correlation of the T2 values with the ADC values resulted ina correlation coefficient of 0.772 (P < 0.01). The connection does notseem linear (Fig. 5).

The ADC values demonstrate a similar AUC as the T2 values inall questions (Table 4). In the distinction of normal gland tissues andprostate cancer, the AUC for ADC values is slightly lower than for T2values. These differences are, however, not significant in any of the cases.Therefore, the performance of the T2 values seems to be comparable tothe performance of the ADC values in terms of tumor detection andgrading. A combination of both parameters yielded a small diagnosticimprovement for (1) precancerous lesions versus prostate cancer,(2) prostate cancer versus other, and (3) normal versus prostate cancer,in each case for the whole prostate. This is, however, not significant inany case.

Differentiation of the Cancer Aggressiveness WithQuantitative T2 Values and ADC Values

The correlation of the Gleason scorewith the T2 values showed asignificant inverse correlation with rS = −0.261 (P < 0.01) (Fig. 6). Thecorrelation of the ADC values with the Gleason score also showed anegative correlation with rS = −0.160 (P = 0.03), that is, however,slightly less significant.

By use of the quantitative T2 values, it is also possible to differ-entiate between low-grade and intermediate-/high-grade tumors in thePZ (0.742 [0.633–0.850]; P < 0.01). This is not possible in the TZ. De-spite the correlation, the 95% confidence intervals of the individualGleason scores overlap (Fig. 6). Because the correlation with rS(T2) of−0.261 and rS(ADC) of −0.160 is weak or very weak, the quantitativeT2 values and ADC values only allow limited conclusions on the under-lying degree of aggressiveness.

Image Quality of the Morphological andCalculated T2w

Reader 1 rated the image quality of the morphological T2w asbetter in comparison to the calculated T2w image for all nine criteria.Notably, the rating “excellent” was given in most cases for the morpho-logical T2w images.

TABLE 3. Number of Examined Areas Divided According toHistological Entity for Peripheral Zone and Transitional Zone

Histopathological AreasPeripheral

Zone, n = 539TransitionalZone, n = 318

Normal 60.67% (327/539) 16.04% (51/318)Prostatitis 16.51% (89/539) 9.43% (30/318)BPH nodes 0 47.17% (150/318)Precancerous lesions 5.01% (27/539) 1.89% (6/318)ASAP 1.31% (6/539) 0PIN 3.70% (21/539) 1.89% (6/318)

Prostate cancer 17.81% (96/539) 25.47% (81/318)Gleason score 3 + 3 = 6 5.19% (28/539) 8.18% (26/318)Gleason score 3 + 4 = 7a 6.31% (34/539) 7.86% (25/318)Gleason score 4 + 3 = 7b 3.15% (17/539) 6.92% (22/318)Gleason score 4 + 4 = 8 3.15% (17/539) 2.52% (8/318)

BPH indicates benign prostate hyperplasia; ASAP, atypical small acinar prolif-eration; PIN, prostatic intraepithelial neoplasia.

TABLE 4. AUC of the Receiver Operating Curves for the Respective Questions for Mean T2 Values and Mean ADC

AUC Values (Confidence Interval) of ROC Analysis Quantitative T2 Values ADC Values Regression Coefficient T2 + ADC

Normal vs PCA 0.871 (0.840–0.902) 0.848 (0.814–0.882) 0.877 (0.848–0.907)PCAvs BPH, prostatitis, normal 0.829 (0.795–0.862) 0.825 (0.790–0.859) 0.844 (0.812–0.876)PZ: PCAvs normal 0.846 (0.804–0.889) 0.831 (0.784–0.877) 0.853 (0.812–0.894)TZ: PCAvs BPH, prostatits 0.827 (0.776–0.877) 0.799 (0.742–0.856) 0.839 (0.791–0.887)TZ: PCAvs BPH 0.840 (0.790–0.891) 0.820 (0.765–0.876) 0.859 (0.811–0.906)PZ: prostatitis vs PCA 0.846 (0.790–0.902) 0.838 (0.781–0.894) 0.862 (0.811–0.913)PZ: normal vs prostatitis 0.486 (0.416–0.555) 0.526 (0.461–0.590) 0.585 (0.512–0.657)PZ: GS = 6 vs GS ≥ 7 0.742 (0.633–0.850) 0.746 (0.635–0.858) 0.755 (0.649–0.862)TZ: GS = 6 vs GS ≥ 7 0.431 (0.312–0.289) 0.361 (0.221–0.502) 0.645 (0.504–0.787)Precancerous tissue and GS = 6 vs GS ≥ 7 0.696 (0.621–0.772) 0.697 (0.620–0.774) 0.705 (0.629–0.782)

AUC indicates area under the curve; ADC, apparent diffusion coefficient; PZ, peripheral zone; TZ, transitional zone; GS, Gleason score; ROC, receiver operating characteristic.





For reader 2, 6 of the 9 criteria showed equally good qualities inthe morphological and calculated T2w. The other 3 characteristics (vis-ibility of PZ and TZ, PZ and TZ distinction, visibility of lesions) wererated as better on the calculated T2w by reader 2.

Detection of Prostate Cancer With PIRADS UsingMorphological and Calculated T2w

The highest sensitivity and specificity for the differentiationbetween prostate cancer and noncancer tissue in the PZ results in aPIRADS score of 3 as a cutoff. This is why all lesions with 3 or higherwere evaluated as “cancer positive,” and all others as “cancer negative.”The 2 readers evaluated the morphological and calculated T2w in termsof their diagnostic value for the differentiation of prostate cancer versusother entities.

For reader 1, the AUC for morphological T2w was 0.658(0.597–0.719) and for calculated T2w AUC was 0.682 (0.624–0.740).For morphological T2w reader 2 obtained an AUC value of 0.601(0.538–0.665) and for calculated T2w AUC of 0.640 (0.581–0.700).

Reader 1 achieved the highest diagnostic accuracy with the cal-culated T2w. However, the difference between the detection rates inmorphological and calculated T2w was not significant for either reader(P = 0.47 and 0.28). In addition, the higher diagnostic accuracy ofreader 1 in comparison to reader 2 is not significant (P = 0.14).

Seminal Vesicle InvasionThe probability of seminal vesicular invasion was determined

on both morphological and calculated T2w images. The histologicallyassessed seminal vesicles of the prostatectomized patients served as areference. All 16 prostatectomy preparations showed tumor-free semi-nal vesicles. Reader 2 gave a score of 1 or 2 for all of these patients(seminal vesicle invasion not likely). Reader 1 evaluated the seminalvesicle invasion with a score of 3 in 1 of the 16 patients in both se-quences (seminal vesicle invasion possibly present). However, the im-age quality of these 2 sequences was only moderate in this specificpatient. The specificity for the detection of a seminal vesicle invasionbased on morphological or calculated T2w alone is 93.8% for reader

FIGURE 4. Receiver operating curves of quantitative T2 values and ADC values (1a, 2, 3) as well as the regression model from both values (1b).Differentiation between prostate cancer and normal tissue in the PZ, TZ, and whole prostate.





1 and 100% for reader 2. As none of the patients had seminal vesicleinvasion, it was not possible to determine the sensitivity.

DISCUSSIONOur study has shown that quantitative T2 values in both PZ and

TZ are suitable for the detection of prostate cancer, which can be signif-icantly differentiated from normal gland tissue as well as benign changesand precancerous lesions in the prostate. In addition, in the PZ, it is pos-sible to differentiate between Gleason scores 6, 7, and 8, and distinguishnonsignificant cancer from significant cancer. In the TZ, this is not pos-siblewith T2 values alone. T2 values andADC values showa strong cor-relation with rS = 0.772 (P < 0.01), and with regard to the ADC values,there is a significant negative correlation between T2 values and theGleason score.

Although reader 1 rated morphological T2w images better inquality, there is no significant difference in the diagnostic accuracy be-tween morphological and calculated T2w for both readers. Sensitivityand specificity show that the diagnosis on the basis of the T2 mapsalone is not sufficient.

It has already been shown for some quantitative parameters thatthey can distinguish prostate cancer from healthy gland tissue. The pa-rameter that can best distinguish alone is the ADC value.1,5,17–20 TheAUC for the differentiation between normal prostate tissue and prostatecancer in the PZ was 0.845; 0.689 and 0.82 in previous studies. Only 2of these studies also investigated the distinction of the cancer fromhealthy tissue in the whole gland (AUC, 0.74).1,5

It has been demonstrated that ADC values as well as quantitativeT2 values are linked to the cell density.21 The strong correlation be-tween the 2 parameters is also indicated by our results. In addition,we found that combining both values did not achieve a significant informa-tion gain. This correlation between the 2 parameters was further confirmedin another study with r = 0.770–0.804.22 Both parameters therefore seemto be equally suitable for the detection of a prostate cancer.

As for the ADC values, the same applied with the quantitativeT2 values. A lower T2 value is suspected to be cancer, and the T2 valuecan be used to distinguish between prostate cancer and normal tissue.Previous studies have shown that the mean T2 value of the cancer areasis significantly lower than the mean T2 value of normal glandularareas2,5,23 with an AUC of 0.74 in the PZ, and 0.613 and 0.85 in the

whole prostate.1,23 The AUC values here are comparable with our re-sults (PZ: AUC = 0.846; whole prostate: AUC = 0.871).

In addition, the prevalence of benign prostate hyperplasia tendsto increase in old age, which, in an MRI examination, might be identi-fied as a suspected cancer.24,25We demonstrated that it is possible to dis-tinguish between benign changes and prostate cancer using quantitativeT2 values.2,3 The only study known to us in which PZ and TZ were in-vestigated separately showed that, even in the TZ with an AUC of 0.79,cancer can be distinguished from noncancerous areas.5 This was alsoconfirmed in our study (AUC = 0.827). We additionally investigatedthe distinguishability between prostate cancer and other benign patholo-gies such as prostatitis and also precancerous lesions, which can be sig-nificantly differentiated from cancer by T2 values as well.

With regard to the ADC values, a significant inverse correlationwith the Gleason score has already been shown.14,17 Similarly, the quan-titative T2 values decrease with increasing Gleason scores.5 We wereable to demonstrate that the individual tumor grades are distinguishablefrom each other using the T2 values in the PZ. However, the AUCs of thereceiver operating curves were only moderatewith 0.718 (Gleason score6 vs 7) and 0.662 (Gleason score 7 vs 8), which means that T2 mappingcan provide evidence of aggressiveness, albeit without allowing adefinite conclusion.

In the clinical practice, it is relevant to differentiate low-grade tu-mors, which are not significant, from higher-grade tumors, which is sig-nificantly possible with T2 values. There is only one study known to us,which demonstrated this.23

Although in our study, we only examined data from one scanner,T2 values are ideally reproducible across scanners. Therefore, here, de-fined thresholds should be translatable to other systems. However, theirdiagnostic accuracy must be further investigated.5

The ratings of morphological and calculated T2w were discrep-ant for the 2 readers and also their detection rates for prostate cancer inthe 2 sequences were different. However, this difference is not signifi-cant, and it has already been published, which the evaluation of prostateMRI through PIRADS classification depends on the reader's experi-ence.26 Considering this, calculated T2w seems to be equally suitablefor cancer detection as morphological T2w. We are not aware of anystudy, which investigated this approach so far.

Lastly, T2 mapping can give both quantitative and qualitative in-formation about a cancer suspicious area in MRI. In our study, the

FIGURE 6. Quantitative T2 values in milliseconds according to Gleasonscore (degree sign indicates outliers; asterisk indicates extreme outliers,with values more than 3 times the height of the boxes).

FIGURE 5. Correlation of T2 values in milliseconds and ADC values in10−6 mm2/s (rS = 0.772).





detection rate for prostate cancer does not differ significantly betweencalculated T2w and morphological T2w, and the suspected lesion canadditionally be characterized by the quantitative T2 values. ThereforeT2 mapping could add a diagnostic gain to the mpMRI protocol. Fur-thermore, the quantitative T2 values can be included in new approachesin prostate cancer diagnostics such as computer-aided MRI diagnosis.Hereby, some first promising results were published.27

However, adding another sequence would extend the acqui-sition duration.

Regarding the scan times, there are already approaches to limitthe amount of acquired datawithout, however, restraining image qualityand validity. Those investigations have been made for mpMRI itself aswell as for T2 mapping.3,28 In both studies, they successfully reducedscan times with an equal gain of information, which offers the opportu-nity to perform sufficient diagnostic within an acceptable time frame.

Our study has some limitations. First, a selection bias may haveoccurred, because it is a retrospective study. Despite the randomization,the evaluation of the image quality of the morphological and calculatedT2w can lead to an observation bias as the MR images were notanonymized. As the ROIs on T2w, T2 map, and ADC map were manu-ally plotted, minor inaccuracies may have occurred. Also in histopathol-ogy from biopsy samples, there can always occur false-negative results.

In 65 patients (86.7%), there was less than 3 months in betweenmpMRI and biopsy, in 68 patients (90.7%), less than 4 months, and in70 patients (93.3%), less than 5 months. There can be small biologicalchanges in the prostate in this time. In addition, in 4 patients, small hy-perintense areas occurred in T1w, because they had a TRUS-guided orin-bore biopsy before mpMRI within less than 9 weeks; these patientswere not excluded, as the areas were to small and did not result inhypointensities in T2w.

In summary, we can conclude that quantitative T2 values in thePZ as well as in the TZ seem to be suitable to differentiate between pros-tate cancer and normal gland tissue or BPH nodes, and therefore, the pos-sibility of routinely using T2 mapping should be further investigated.

ACKNOWLEDGMENTSWe thank Felix Güttler, René Aschenbach, and Friedrich-Carl

von Rundstedt for organizing the study. We also thank Elisabeth Heymand Caroline Allen for their support.

REFERENCES

1. Metzger GJ, Kalavagunta C, Spilseth B, et al. Detection of prostate cancer: quan-titative multiparametric MR imaging models developed using registered correla-tive histopathology. Radiology. 2016;279:805–816.

2. Hoang Dinh A, Souchon R,Melodelima C, et al. Characterization of prostate can-cer using T2 mapping at 3T: a multi-scanner study. Diagn Interv Imaging. 2015;96:365–372.

3. Liu W, Turkbey B, Senegas J, et al. Accelerated T-2 mapping for characterizationof prostate cancer. Magn Reson Med. 2011;65:1400–1406.

4. Roebuck JR, Haker SJ, Mitsouras D, et al. Carr-Purcell-Meiboom-Gill imaging ofprostate cancer: quantitative T-2 values for cancer discrimination.MagnReson Im-aging. 2009;27:497–502.

5. Hoang Dinh A, Melodelima C, Souchon R, et al. Quantitative analysis of prostatemultiparametric MR images for detection of aggressive prostate cancer in the pe-ripheral zone: a multiple imager study. Radiology. 2016;280:117–127.

6. Gibbs P, Liney GP, Pickles MD, et al. Correlation of ADC and T2 measurementswith cell density in prostate cancer at 3.0 Tesla. Invest Radiol. 2009;44:572–576.

7. Hilbert T, Kober T, Sumpf TJ, et al. MARTINI andGRAPPA -when speed is taste.Proc Intl Soc Mag Reson Med. 2014;22:1504.

8. Stocker D,Manoliu A, Becker AS, et al. Image quality and geometric distortion ofmodern diffusion-weighted imaging sequences in magnetic resonance imaging ofthe prostate. Invest Radiol. 2018;53:200–206.

9. Samaratunga H, Montironi R, True L, et al. International Society of UrologicalPathology (ISUP) Consensus Conference on Handling and Staging of RadicalProstatectomy Specimens. Working group 1: specimen handling. Mod Pathol.2011;24:6–15.

10. Beyersdorff D, Taymoorian K, Knosel T, et al. MRI of prostate cancer at 1.5and 3.0 T: comparison of image quality in tumor detection and staging. Am JRoentgenol. 2005;185:1214–1220.

11. Franiel T, Stephan C, Erbersdobler A, et al. Areas suspicious for prostate cancer:MR-guided biopsy in patients with at least one transrectal US-guided biopsy witha negative finding-multiparametric MR imaging for detection and biopsy plan-ning. Radiology. 2011;259:162–172.

12. Turkbey B, Merino MJ, Gallardo EC, et al. Comparison of endorectal coil andnonendorectal coil T2W and diffusion-weighted MRI at 3 Tesla for localizingprostate cancer: correlation with whole-mount histopathology. J Magn Reson Im-aging. 2014;39:1443–1448.

13. Woodfield CA, Tung GA, Grand DJ, et al. Diffusion-weighted MRI of peripheralzone prostate cancer: comparison of tumor apparent diffusion coefficient withgleason score and percentage of tumor on core biopsy. Am J Roentgenol. 2010;194:W316–W322.

14. Siddiqui MM, Rais-Bahrami S, Turkbey B, et al. Comparison of MR/ultrasoundfusion-guided biopsy with ultrasound-guided biopsy for the diagnosis of prostatecancer. J Am Med Assoc. 2015;313:390–397.

15. Heijmink SW, Futterer JJ, Hambrock T, et al. Prostate cancer: body-array versusendorectal coil MR imaging at 3 T - comparison of image quality, localization,and staging performance. Radiology. 2007;244:184–195.

16. Weinreb JC, Barentsz JO, Choyke PL, et al. PI-RADS Prostate Imaging -Reporting and Data System: 2015, Version 2. Eur Urol. 2016;69:16–40.

17. Itou Y, Nakanishi K, Narumi Y, et al. Clinical utility of apparent diffusion coeffi-cient (ADC) values in patients with prostate cancer: can ADC values contribute toassess the aggressiveness of prostate cancer? J Magn Reson Imaging. 2011;33:167–172.

18. Pickles MD, Gibbs P, Sreenivas M, et al. Diffusion-weighted imaging of normaland malignant prostate tissue at 3.0T. J Magn Reson Imaging. 2006;23:130–134.

19. Vargas HA, Akin O, Franiel T, et al. Diffusion-weighted endorectal MR imagingat 3 T for prostate cancer: tumor detection and assessment of aggressiveness.Radiology. 2011;259:775–784.

20. Langer DL, van der Kwast TH, Evans AJ, et al. Prostate cancer detection withmulti-parametric MRI: logistic regression analysis of quantitative T2, diffusion-weighted imaging, and dynamic contrast-enhanced MRI. J Magn Reson Imaging.2009;30:327–334.

21. Langer DL, van der Kwast TH, Evans AJ, et al. Intermixed normal tissue withinprostate cancer: effect on MR imaging measurements of apparent diffusion coef-ficient and T2—sparse versus dense cancers. Radiology. 2008;249:900–908.

22. Wu LM, Chen XX, Xuan HQ, et al. Feasibility and preliminary experience ofquantitative T2*Mapping at 3.0 T for detection and assessment of aggressivenessof prostate cancer. Acad Radiol. 2014;21:1020–1026.

23. Wu LM, Zhao ZZ, Chen XX, et al. Comparison of T2(*) mapping with diffusion-weighted imaging in the characterization of low-grade vs intermediate-grade andhigh-grade prostate cancer. Brit J Radiol. 2016;89:20151076.

24. Alanazi AB, Alshalan AM, Alanazi OA, et al. Epidemiology of senile prostaticenlargement among elderly men in Arar, Kingdom of Saudi Arabia. ElectronPhysician. 2017;9:5349–5353.

25. Kirkham AP, Emberton M, Allen C. How good is MRI at detecting andcharacterising cancer within the prostate? Eur Urol. 2006;50:1163–1174.

26. Ke Z, Wang L, Min XD, et al. Diagnostic performance and interobserver consis-tency of the prostate imaging reporting and data system version 2: a study on sixprostate radiologists with different experiences from half a year to 17 years. ChinMed J (Engl). 2018;131:1666–1673.

27. Greer MD, Lay N, Shih JH, et al. Computer-aided diagnosis prior to conven-tional interpretation of prostate mpMRI: an international multi-reader study.Eur Radiol. 2018.

28. Weiss J, Martirosian P, Notohamiprodjo M, et al. Implementation of a 5-minutemagnetic resonance imaging screening protocol for prostate cancer in men withelevated prostate-specific antigen before biopsy. Invest Radiol. 2018;53:186–190.





t2 mapping in prostate cancer - clinical mriclinical-mri.com/wp-content/uploads/2019/03/t2... · t2...

Documents