- Email: [email protected]
Association between changes in suspicious prostate lesions on serial MRI examinations and follow-up biopsy results
Association between changes in suspicious prostate lesions on serial MRI examinations and follow-up biopsy results Andrew B Rosenkrantz, Samuel L Rice, Natasha E Wehrli, Fang-Ming Deng, Samir S Taneja PII: DOI: Reference:
S0899-7071(14)00230-7 doi: 10.1016/j.clinimag.2014.08.008 JCT 7681
To appear in:
Journal of Clinical Imaging
Received date: Accepted date:
21 July 2014 11 August 2014
Please cite this article as: Rosenkrantz Andrew B, Rice Samuel L, Wehrli Natasha E, Deng Fang-Ming, Taneja Samir S, Association between changes in suspicious prostate lesions on serial MRI examinations and follow-up biopsy results, Journal of Clinical Imaging (2014), doi: 10.1016/j.clinimag.2014.08.008
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 proof before it is published in its final 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.
ACCEPTED MANUSCRIPT Association Between Changes in Suspicious Prostate Lesions on
T
Serial MRI Examinations and Follow-up Biopsy Results
RI P
Original Research
SC
Authors
NU
Andrew B Rosenkrantz MD.1 Samuel L Rice MD. 1 Natasha E Wehrli MD. 1* Fang-Ming Deng MD PhD.2 Samir S Taneja MD.3
Correspondence:
CE
PT
*Current affiliation: Weill Cornell Medical College New York-Presbyterian Hospital New York, NY 10065
ED
All authors: Department of Radiology NYU Langone Medical Center 550 First Avenue New York, NY 10016
MA
1: Department of Radiology 2: Department of Pathology 3: Department of Urology, Division of Urologic Oncology
AC
Andrew B. Rosenkrantz. Department of Radiology Center for Biomedical Imaging NYU Langone Medical Center 660 First Avenue, 3rd Floor New York, NY 10016
Phone: 212-263-0232 Fax: 212-263-6634 Email: [email protected] Support: The Joseph and Diane Steinberg Charitable Trust Disclosures: Author Dr. Samir S Taneja is a consultant for Hitachi and previously has been a consultant for Eigen, Healthtronics, Bayer, and Gtx, has previously received grants from the DOD and NIH for which money was paid to the institution, receives royalties from Elsevier, and previously was on the speaker’s bureau for Janssen. All other authors have no conflicts to disclose. 1
ACCEPTED MANUSCRIPT
Abstract
T
We assessed changes in prostate lesions on serial MRI examinations in predicting biopsy results.
RI P
55 men undergoing two prostate MRI examinations ≥6 months apart, followed by targeted biopsy, were included. Two radiologists assessed dominant lesions for an increase in size or
SC
suspicion score (SS). Progression on MRI had lower sensitivity (23.5%-35.3%) and higher
NU
specificity (76.2%-90.5%) than PSA velocity (sensitivity 70.6%, specificity 52.4%) for predicting positive biopsy. Highest accuracy was achieved by PSA velocity (63.6%) for positive
MA
biopsy, and by MRI (65.5%-72.7%) for Gleason >6 tumor. Findings support lesion progression
PT
ED
on MRI serving as a basis for performing subsequent targeted biopsy.
AC
CE
Key Words: prostate cancer; MRI; biopsy; PSA
2
ACCEPTED MANUSCRIPT 1. Introduction
T
A key challenge in the contemporary management of prostate cancer is the lack of an
RI P
effective means of providing non-invasive monitoring over time of a patient’s level of risk for harboring significant cancer. This ability would be important in a number of contexts. First, in
SC
men with a previous biopsy demonstrating low-risk disease, active surveillance with selective
NU
deferred radical therapy is becoming an increasingly used approach to balance preservation of quality-of-life with adequate oncologic control1. However, current active surveillance regimens
MA
call for serial trans-rectal biopsies to identify subsequent evidence of high-grade disease1, 2. The PSA velocity (PSAv), referring to the rate of change in serum PSA levels over time, is an easy
ED
and practical way to non-invasively monitor a patient as well3, and indeed is widely incorporated
PT
into surveillance algorithms4 and has been noted to be a frequent trigger of delayed radical intervention in this population5. Nonetheless, numerous studies cast strong doubt on the ability
CE
of PSAv to reliably monitor for potential progression of disease in men on surveillance,
6-8
AC
concluding that serial biopsy remains necessary and that better non-invasive markers are needed3, . Furthermore, there has recently been a surge of public skepticism regarding the large number
of biopsies performed in men that demonstrate either low-risk cancer or an absence of tumor altogether9. This concern has contributed to an increasing desire to tailor decisions to perform an initial or repeat biopsy to a patient’s individualized level of risk10, for instance by incorporating a period of non-invasive monitoring in some men to guide the decision of whether to perform a potential biopsy. However, this decision faces the similar dilemma as in the active surveillance cohort, given that numerous studies have also shown poor results for the ability of changes in PSA over time to predict the diagnosis of prostate cancer on an initial or repeat biopsy11-14. Thus, there is a compelling need to derive better tools to provide non-invasive evaluation in men 3
ACCEPTED MANUSCRIPT with suspected or known prostate cancer, which in turn could be useful for optimizing selection of men within these groups to undergo biopsy.
T
Given recent advances in multi-parametric imaging, MRI has improved substantially in
RI P
its ability to localize and depict the extent of tumor within the prostate15 and currently is the only widely available non-invasive imaging test that has an established high performance in
SC
visualizing focal tumors. In addition, findings on a single MRI examination have consistently
NU
been shown to be a significant predictor of the results of subsequent biopsy performed in men with known or suspected prostate cancer16, 17. It has thus been suggested that MRI be routinely
MA
incorporated into such clinical pathways18.
The ability of MRI to provide a distinct illustration of the extent and location of tumor in
ED
the prostate raises the possibility of serial MRI examinations being used to provide non-invasive
PT
monitoring in men contemplating undergoing prostate biopsy19. In such an approach, findings on serial MRI examinations suggesting progression of a visualized lesion could serve as a trigger
CE
for identifying those patients warranting a subsequent biopsy. If this scheme were validated,
AC
MRI could then be used as a tool to provide non-invasive monitoring and risk assessment, and thereby potentially reduce the number of biopsies performed. To our knowledge, the utility of multi-parametric MRI has not been studied in this context. Therefore, the aim of our study is to assess the role of changes in prostate lesions on serial MRI examinations in predicting the pathological results of follow-up prostate biopsies.
2. Methods 2.1 Patients
4
ACCEPTED MANUSCRIPT This retrospective HIPAA-compliant study was approved by our institutional review board with a waiver of the requirement for written informed consent.
We searched an
T
institutional database to identify men who underwent multi-parametric MRI examinations of the
RI P
prostate on separate dates separated by at least six months, with no intervening prostate therapy. The initial examination was required to have occurred after January 2011 given a major change
SC
in our institution’s routine prostate MRI protocol at that time. In men with more than two serial
NU
prostate MRI examinations, the first and last examinations were selected. This search identified an initial set of 162 men. Of these, we then identified those men who underwent a prostate
MA
biopsy after the follow-up MRI, providing a cohort of 64 men. An additional 9 cases were excluded given non-routine protocol on either the initial or follow-up MRI. These exclusions
ED
left a final included cohort of 55 men (mean age 65±8 years; range 48 to 80 years), with a mean
PT
PSA at the time of the initial MRI of 4.7±2.1 ng/mL (range 0.6 to 10.9 ng/mL). The distribution of clinical histories at the time of the initial MRI in these men was as follows: no prior prostate
CE
biopsy (n=14); previous negative prostate biopsy (n=19); prior positive biopsy, on active
AC
surveillance (n=22). The mean time interval between the initial and follow-up MRI was 14±6 months (range 6-32 months). 30 of the men underwent an additional biopsy after the initial MRI examination; however, there was a delay of at least four months between this biopsy and the follow-up MRI examination in all cases.
2.2 MRI Technique MRI was performed using a 3T system (MAGNETOM Trio, Siemens Healthcare, Erlangen, Germany) and a torso phased-array coil.
Examinations included the following
sequences of the prostate and seminal vesicles: turbo spin-echo (TSE) T2-weighted imaging in
5
ACCEPTED MANUSCRIPT the axial, coronal, and sagittal planes (TR/TE, 4000-4960/105; slice thickness, 3 mm; no interslice gap for axial and coronal planes, 0.6 mm interslice gap for sagittal plane; FOV, 180 ×
T
180 mm; matrix, 256 × 256 for axial and coronal planes, 256 x 230 for sagittal plane; parallel
RI P
imaging factor, 2; 3 signal averages for axial and coronal planes, 2 signal averages for sagittal plane), axial TSE T1-weighted imaging (TR/TE, 683/10; slice thickness, 3 mm, no interslice
SC
gap; FOV, 180 × 180 mm; matrix, 192 × 192; parallel imaging factor, 2; 2 signal averages) and
NU
axial single-shot echo-planar diffusion-weighted imaging (DWI) (TR/TE, 4100/86; slice thickness, 3 mm; no interslice gap; FOV, 200 × 200 mm; matrix, 100 × 100; parallel imaging
MA
factor, 2; 10 signal averages; b values, 50 and 1000 s/mm2) with reconstruction of the apparent diffusion coefficient (ADC) map using a standard mono-exponential fit. Dynamic contrast-
ED
enhanced MRI (DCE-MRI) of the prostate was performed using a 3D T1-weighted spoiled
PT
gradient-echo sequence (TR/TE, 2.84/0.94; flip angle, 16°; partition thickness, 3 mm; FOV, 240 × 240 mm; matrix, 128 × 128; no parallel imaging; 1 signal average) before and at multiple time-
CE
points after the intravenous administration of 0.1 mmol/kg of gadopentetate dimeglumine
AC
(Magnevist, Bayer HealthCare). The contrast material was administered as a bolus via power injector (Spectris, Medrad) and followed by a 20-mL saline flush, both administered at a rate of 3 mL/s. The temporal resolution for DCE was 5.5 seconds. 55 post-contrast measures were obtained, for a total acquisition time after contrast injection of 5 minutes 5 seconds. A biexponential model based on the raw DCE images was applied to generate parametric maps representing the maximum slope of enhancement during the contrast-enhanced acquisition and the washout of contrast material after the enhancement peak, hereafter referred to as the “maximum slope” and “washout maps,” respectively.
Commercial software was used for
generating the parametric maps (DynaCAD version 2.1.6, Invivo).
6
ACCEPTED MANUSCRIPT
2.3 Image Analysis
T
Cases were reviewed in consensus by two fellowship-trained radiologists with 2 and 6
RI P
years of experience in prostate MRI interpretation who were blinded to clinical and pathological details. The radiologists initially observed the baseline examination performed in each patient,
SC
reviewing all sequences in joint fashion. The radiologists identified the most suspicious lesion in
NU
each case, as guided by the PI-RADS criteria provided within the guidelines set forth by the European Society of Uroradiology expert panel20, which was considered to represent the
MA
dominant lesion. For the dominant lesion, the observers separately recorded the lesion’s level of suspicion on each sequence (T2WI, DWI, and DCE), again using the explicit PI-RADS criteria,
ED
recording a score of 1 for sequences on which the lesion was not visible. A score of 1 was
PT
recorded for all sequences in cases with no visible dominant lesion. The radiologists also measured the dominant lesion’s single largest diameter in the axial plane on each sequence. For
CE
DWI, lesion size was recorded using the ADC maps given a past study showing better
AC
performance of the ADC map than high b-value images in predicting tumor sizes on radical prostatectomy21; for DCE, lesion size was recorded using an individual post-contrast time-point depicting the lesion given greater clarity of lesion boundaries on these images than on postprocessed parametric color maps. No size was recorded for sequences in which the dominant lesion was not visualized.
After completing evaluation of the baseline examination, the
radiologists then observed the follow-up examination in similar fashion, again recording the suspicion level and size of the dominant lesion on each sequence. During this assessment, the radiologists were aware of which lesion was selected as the dominant lesion on the initial study. Following evaluation of all examinations, the radiologists’ readings were categorized in binary
7
ACCEPTED MANUSCRIPT fashion in terms of whether or not an increase was identified between examinations in suspicion level or size. For suspicion level, an increase of at least one category on the 1-5 PI-RADS scale
T
was considered to represent an increase; for size, given imprecision of individual size
RI P
measurements, an increase in maximal diameter of at least 3 mm was considered to represent an increase. These categorizations were based solely on the presence of an increase, irrespective of
SC
the actual values themselves on the two examinations.
NU
In addition, the electronic medical record was reviewed to record the PSA at the time of the baseline and follow-up MRI examinations for each patient, defined as the PSA obtained
MA
immediately preceding each MRI. Based on these values, the PSAv was recorded for each
ED
patient as (PSAfollow-up – PSAbaseline) / (time interval between serial PSA values)19.
PT
2.4 Reference standard
The reference standard in all cases was trans-rectal ultrasound-guided prostate biopsy
CE
performed after the follow-up MRI, which at our institution routinely includes cores targeting
AC
MRI-detected abnormalities, in addition to standard systematic cores from all sextants. Biopsy results were categorized as positive or negative for tumor, as well as positive or negative for tumor with a Gleason score greater than 6.
2.5 Statistics Receiver-operating-characteristic analysis was performed to identify optimal thresholds in terms of PSAv for predicting presence of any cancer as well as cancer with Gleason score > 6, defined as that threshold maximizing the average of sensitivity and specificity; PSAv values were then converted to a binary measure as being greater or lesser than these thresholds. This
8
ACCEPTED MANUSCRIPT step was performed to allow comparison within a single analysis of the utility of PSAv with that of changes in lesion suspicion score or size, which were also recorded in binary fashion. The
T
Cochrane Q-test was performed to compare the sensitivity, specificity, and accuracy of a panel of
RI P
measures (PSAv, change in suspicion score or size for T2WI, DWI, and DCE, as well as change in any MRI parameter) for predicting presence of tumor as well as of tumor with Gleason score
SC
greater than 6. The p-values provided from this analysis are adjusted for multiple comparisons
NU
and were considered statistically significant at p<0.05. The statistical analysis was performed
MA
using software (MedCalc for Windows, version 12.7, MedCalc Software, Ostend, Belgium).
3. Results
ED
3.1 Clinical data
PT
The follow-up biopsy was positive for tumor in 61.8% (34/55), and negative in 38.2% (21/55), of cases. In addition, the follow-up biopsy was positive for tumor with Gleason score >
CE
6 in 23.6% (13/55) of cases. In the 13 cases with Gleason score > 6, 10 had a Gleason score of
AC
3+4 and 3 had a Gleason score of 4+3. In the 55 included cases, mean PSAv was 0.66±1.50 ng/mL/year (range, -1.97 to +5.15 ng/mL/year). Representative cases are shown in Figures 1-2.
3.2 Prediction of biopsy results Of the 55 cases, 23.6% (13/55), 25.5% (14/55), and 23.6% (13/55), showed an increase in suspicion level on T2WI, DWI, and DCE, respectively; 10.9% (6/55), 18.2% (10/55), and 12.7% (7/55), showed an increase in size on T2WI, DWI, and DCE, respectively; and 41.8% (23/55) showed an increase using any MRI parameter. ROC demonstrated optimal PSAv thresholds of +0.09 ng/ml/year and +0.48 ng/ml/year for predicting presence of any tumor and tumor with
9
ACCEPTED MANUSCRIPT Gleason score > 6 on follow-up biopsy. 61.8% (34/55) and 63.6% (35/55) of cases exhibited a PSAv meeting these respective thresholds.
T
Table 1 shows the sensitivity, specificity, and accuracy of each of the measures in
RI P
predicting the presence of tumor on follow-up biopsy. Overall, change in suspicion level or size on MRI had low sensitivity (range 23.5%-35.3%) but good specificity (range 76.2%-90.5%) in
SC
predicting presence of tumor. PSAv had higher sensitivity (70.6%) but lower specificity (52.4%)
NU
in comparison with the MRI-based measures. Highest overall accuracy for predicting presence of tumor was obtained by PSAv (63.6%). The sensitivities of all of the individual MRI features,
MA
the specificities of change in size on T2WI, change in suspicion score on DWI, or change in size or suspicion score on DCE, as well as the accuracy of change in size on T2WI, were all
ED
significantly difference compared with the performance of PSAv (p<0.05). Otherwise, there was
PT
no significant difference in sensitivity, specificity, or accuracy between any combination of the individual MRI-based measures and PSAv, although there were non-significant greater
CE
sensitivities for predicating a positive biopsy based on a change in suspicion score than a change
AC
in size for all three MRI sequences. Table 2 shows the sensitivity, specificity, and accuracy of each of the measures in predicting the presence of tumor with Gleason score > 6 on follow-up biopsy. Sensitivities using the MRI parameters tended to be higher in this analysis in comparison with the sensitivity of MRI in predicting presence of tumor.
Nonetheless, as in the prior assessment, change in
suspicion level or size on MRI had lower sensitivity (range 15.4%-38.5%) and higher specificity (range 78.6%-90.5%) in comparison with PSAv (sensitivity 61.5%, specificity 54.8%).
In
comparison with the analysis for prediction of tumor, all of the MRI assessments based on change in size or suspicion score exhibited higher accuracy for predicting tumor with Gleason
10
ACCEPTED MANUSCRIPT score > 6 (range 65.5%-72.7%) than did PSAv (56.4%). The specificity of all of the MRI parameters were significantly different than that of PSAv (p<0.05); no here was significant
T
difference in sensitivity, specificity, or accuracy between any combination of the individual
RI P
MRI-based measures and PSAv, although there were non-significant greater sensitivities for predicating tumor with a Gleason score > 6 based on a change in suspicion score than a change
NU
SC
in size for all three MRI sequences.
MA
4. Discussion
In this study, we evaluated the potential role of detection of changes in focal prostate
ED
lesions on serial MRI examinations in predicting pathologic findings on follow-up biopsy. The
PT
presence of progression of a lesion on imaging was observed to have low sensitivity but high specificity in prediction of subsequent biopsy results. The low sensitivity is likely attributable to
CE
the known slow rate of biologic progression and prolonged natural disease course of prostate
AC
cancer in most cases1. Indeed, prostate cancer often has a pre-clinical phase of over a decade22. To this end, identification of either cancer or high-grade cancer that was not detected on an earlier biopsy often represents undersampling on the prior biopsy rather than true interval disease progression23. Thus, given the stability of many cases of prostate cancer over the time interval under evaluation in this study, an overt change in appearance may not be expected. On the other hand, although interval progression on imaging was observed in a small fraction of cases, this most often represented high-grade tumor. This observation is consistent with the known indolent course and often clinical insignificance of Gleason 6 prostate cancer24, in comparison with the more aggressive nature of higher Gleason score tumors that warrant therapy.
11
ACCEPTED MANUSCRIPT In comparison with serial changes on MRI, PSAv exhibited higher sensitivity and lower specificity in prediction of follow-up biopsy results. A fraction of prostate cancers are inherently
T
invisible or exhibit poor conspicuity on MRI, for reasons relating to their histologic architecture
assessing such lesions.
RI P
or other mechanisms not yet fully understood25, and PSAv may provide an alternate means of Past studies have also demonstrated a role of PSA-based data in
SC
complementing MRI for detection of some prostate cancer cases26,
27
. However, PSA lacks
NU
specificity, also changing in response to prostatic inflammation, hormonal status, hyperplasia, and other factors, contributing to its lower specificity22,
28
. When considering sensitivity and
MA
specificity together, the overall performance of PSAv was moderate in our study, consistent with results of numerous prior studies3, 6-8. For instance, Vickers et al. reported an AUC of 0.702 for
ED
PSAv in predicting biopsy results in the initial detection of prostate cancer11, and Ng et al.
PT
reported an AUC of 0.70 in predicting presence of adverse histology on follow-up biopsy in men on active surveillance29. Thus, our findings agree with past studies in indicating that PSAv is not
CE
sufficient as a non-invasive biomarker in providing guidance for biopsy decisions.
AC
One interesting observation was a somewhat higher performance for changes in lesions’ suspicion score than changes in size in predicting biopsy results, although these differences were not significant, possibly relating to the small numbers of cases with serial imaging changes as well as with positive biopsy results. Obtaining a precise measure of prostate lesion size can be difficulty when performed manually given that prostate tumors often do not form a discrete, well-marginated mass, as encountered by tumors in some other organs. In many cases, the tumor may have irregular margins, insinuating with normal tissue, as well as exhibit heterogeneous signal intensity with areas of relatively subtle signal alterations. It is also possible that other examination-related factors, such as slight differences in slice positioning and orientation, as well
12
ACCEPTED MANUSCRIPT as varying degrees of patient motion, between the examinations, may also influence apparent lesion size. These factors can introduce error into the size measurements and confound the
T
assessment for potentially subtle changes in size between studies. On the other hand, the
RI P
suspicion level is determined based on five discrete categories, each having its own distinct criteria20. It is possible that this determination, based upon the judgment of the interpreting
SC
radiologist, is less susceptible to the imprecisions that confound the size measurements.
NU
Our findings are clinically relevant in men with both suspected as well as with known prostate cancer and who are on surveillance. Specifically, there is currently a lack of a reliable
MA
means of appropriately stratifying such cohorts and optimally selecting those men who warrant biopsy. This limitation has led to a very high rate of sequential biopsies in such cohorts30. Our
ED
findings support the routine performance of biopsy following an apparent change in size or
PT
suspicion score of a focal lesion on MRI, given the very high specificity of such an occurrence, including for high-grade disease. Thus, the presence of progression on imaging identifies a
CE
subset of men who seem to be at particularly high risk and who more confidently warrant biopsy.
AC
However, given the low sensitivity of changes in imaging findings, serial imaging is not currently sufficient to avoid performing a repeat biopsy. Investigation of additional means of identifying the subset of men harboring significant disease that may have been missed on both an initial biopsy as well as on serial MRI examinations is therefore warranted. A number of limitations of our study warrant mention. First, the sample size was small, including only a small number of cases that were positive for tumor. In addition, the patient population was heterogeneous, including both patients with and without known prostate cancer prior to the initial MRI. Nonetheless, the potential value of using MRI to provide non-invasive monitoring and selection of patients for biopsy is important in both groups, and serial MRI has
13
ACCEPTED MANUSCRIPT been applied in a comparable fashion in both cohorts in our practice. Also, the time interval between the two MRI examinations was variable within the cohort. While the PSAv calculation
T
adjusts for the variability in time interval between measurements, it was not possible to adjust the
RI P
changes in MRI findings in this fashion. This variability in timing of the examinations reflects the retrospective nature of our study and the lack of a current accepted standard or guideline as to
SC
the appropriate frequency at which serial MRI examinations should be performed in this context.
NU
Finally, as this study was retrospective, as previously noted, the actual performance of serial MRI for guiding prostate cancer management in a prospective setting remains unknown, and
MA
future studies incorporating serial MRI findings into clinical pathways and decision-making remain warranted.
ED
In conclusion, we observed low sensitivity and high specificity of changes in focal In
PT
prostate lesions on serial MRI in predicting the results of subsequent prostate biopsy.
comparison, PSAv had moderate performance in terms of both sensitivity and specificity. While
CE
the highest overall for predicting a positive biopsy was achieved by PSAv, the highest accuracy
on MRI.
AC
for predicting a Gleason score greater than 6 was achieved by a change in size or suspicion score Our findings indicate that in men contemplating prostate biopsy, the biopsy is
warranted if there is apparent progression on imaging, although it is not possible at this time to avoid deferring biopsy on the basis of stability of a lesion on serial imaging. Further studies are necessary to evaluate the role of these findings in guiding clinical decisions regarding the performance of prostate biopsy in a larger prospective patient cohort.
14
ACCEPTED MANUSCRIPT References
7.
8.
9. 10. 11.
12.
13.
14.
15.
16.
17.
T
RI P
SC
NU
6.
MA
5.
ED
4.
PT
3.
CE
2.
KLOTZ L. Active surveillance for prostate cancer: overview and update. Current treatment options in oncology 2013;14:97-108. LAWRENTSCHUK N, KLOTZ L. Active surveillance for favorable-risk prostate cancer: a short review. Korean J Urol 2010;51:665-70. VAN DEN BERGH RC, ROEMELING S, ROOBOL MJ, et al. Prostate-specific antigen kinetics in clinical decision-making during active surveillance for early prostate cancer--a review. European urology 2008;54:505-16. KLOTZ L. Active surveillance versus radical treatment for favorable-risk localized prostate cancer. Current treatment options in oncology 2006;7:355-62. COEN JJ, FELDMAN AS, SMITH MR, et al. Watchful waiting for localized prostate cancer in the PSA era: what have been the triggers for intervention? BJU international 2011;107:1582-6. IREMASHVILI V, MANOHARAN M, LOKESHWAR SD, et al. Comprehensive analysis of post-diagnostic prostate-specific antigen kinetics as predictor of a prostate cancer progression in active surveillance patients. BJU international 2013;111:396-403. AYRES BE, MONTGOMERY BS, BARBER NJ, et al. The role of transperineal template prostate biopsies in restaging men with prostate cancer managed by active surveillance. BJU international 2012;109:1170-6. ROSS AE, LOEB S, LANDIS P, et al. Prostate-specific antigen kinetics during follow-up are an unreliable trigger for intervention in a prostate cancer surveillance program. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2010;28:2810-6. MCNAUGHTON-COLLINS MF, BARRY MJ. One man at a time--resolving the PSA controversy. The New England journal of medicine 2011;365:1951-3. ROOBOL MJ, CARLSSON SV. Risk stratification in prostate cancer screening. Nature reviews. Urology 2013;10:38-48. VICKERS AJ, TILL C, TANGEN CM, et al. An empirical evaluation of guidelines on prostate-specific antigen velocity in prostate cancer detection. Journal of the National Cancer Institute 2011;103:462-9. BANGMA CH, VAN SCHAIK RH, BLIJENBERG BG, et al. On the use of prostate-specific antigen for screening of prostate cancer in European Randomised Study for Screening of Prostate Cancer. European journal of cancer 2010;46:3109-19. VICKERS AJ, SAVAGE C, O'BRIEN MF, et al. Systematic review of pretreatment prostate-specific antigen velocity and doubling time as predictors for prostate cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2009;27:398-403. VICKERS AJ, WOLTERS T, SAVAGE CJ, et al. Prostate-specific antigen velocity for early detection of prostate cancer: result from a large, representative, population-based cohort. European urology 2009;56:753-60. TURKBEY B, MANI H, SHAH V, et al. Multiparametric 3T prostate magnetic resonance imaging to detect cancer: histopathological correlation using prostatectomy specimens processed in customized magnetic resonance imaging based molds. J Urol 2011;186:1818-24. PINTO PA, CHUNG PH, RASTINEHAD AR, et al. Magnetic resonance imaging/ultrasound fusion guided prostate biopsy improves cancer detection following transrectal ultrasound biopsy and correlates with multiparametric magnetic resonance imaging. J Urol 2011;186:1281-5. VOURGANTI S, RASTINEHAD A, YERRAM NK, et al. Multiparametric magnetic resonance imaging and ultrasound fusion biopsy detect prostate cancer in patients with prior negative transrectal ultrasound biopsies. The Journal of urology 2012;188:2152-7.
AC
1.
15
ACCEPTED MANUSCRIPT
26.
27.
28. 29. 30.
T
RI P
SC
NU
25.
MA
24.
ED
21. 22. 23.
PT
20.
CE
19.
AHMED HU, KIRKHAM A, ARYA M, et al. Is it time to consider a role for MRI before prostate biopsy? Nature reviews. Clinical oncology 2009;6:197-206. COAKLEY FV, CHEN I, QAYYUM A, et al. Validity of prostate-specific antigen as a tumour marker in men with prostate cancer managed by watchful-waiting: correlation with findings at serial endorectal magnetic resonance imaging and spectroscopic imaging. BJU Int 2007;99:41-5. BARENTSZ JO, RICHENBERG J, CLEMENTS R, et al. ESUR prostate MR guidelines 2012. Eur Radiol 2012;22:746-57. BLINDED SIKARIS K. Prostate cancer screening. Pathology 2012;44:99-109. DUFFIELD AS, LEE TK, MIYAMOTO H, et al. Radical prostatectomy findings in patients in whom active surveillance of prostate cancer fails. The Journal of urology 2009;182:2274-8. DONIN NM, LAZE J, ZHOU M, et al. Gleason 6 prostate tumors diagnosed in the PSA era do not demonstrate the capacity for metastatic spread at the time of radical prostatectomy. Urology 2013;82:148-52. LANGER DL, VAN DER KWAST TH, EVANS AJ, et al. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2--sparse versus dense cancers. Radiology 2008;249:900-8. VILANOVA JC, COMET J, BARCELO-VIDAL C, et al. Peripheral zone prostate cancer in patients with elevated PSA levels and low free-to-total PSA ratio: detection with MR imaging and MR spectroscopy. Radiology 2009;253:135-43. VILANOVA JC, BARCELO-VIDAL C, COMET J, et al. Usefulness of prebiopsy multifunctional and morphologic MRI combined with free-to-total prostate-specific antigen ratio in the detection of prostate cancer. AJR. American journal of roentgenology 2011;196:W715-22. GREENE KL, ALBERTSEN PC, BABAIAN RJ, et al. Prostate specific antigen best practice statement: 2009 update. The Journal of urology 2013;189:S2-S11. NG MK, VAN AS N, THOMAS K, et al. Prostate-specific antigen (PSA) kinetics in untreated, localized prostate cancer: PSA velocity vs PSA doubling time. BJU international 2009;103:872-6. PLOUSSARD G, NICOLAIEW N, MARCHAND C, et al. Risk of repeat biopsy and prostate cancer detection after an initial extended negative biopsy: longitudinal follow-up from a prospective trial. BJU international 2013;111:988-96.
AC
18.
16
ACCEPTED MANUSCRIPT
Tables
T
Table 1-Performance of PSA velocity and of changes in suspicion level or size of focal lesions
43.6% (24/55) 56.4% (31/55) 50.9% (28/55)
DWI DCE
38.2% (21/55) 41.8% (23/55) 43.6% (24/55) 54.6% (30/55)
ED
Size T2WI
DCE
AC
CE
Any MRI parameter
PT
DWI
Specificity 52.4% (10/21)
23.5% (8/34) 35.3% (12/34) 29.4% (10/34)
76.2% (16/21) 90.5% (19/21) 85.7% (18/21)
8.8% (3/34) 17.7% (6/34) 14.7% (5/34) 47.1% (16/34)
85.7% (18/21) 81.0% (17/21) 90.5% (19/21) 66.7% (14/21)
NU
Suspicion score T2WI
Sensitivity 70.6% (24/34)
SC
Accuracy 63.6% (35/55)
MA
Measure PSA velocity
RI P
on serial multi-parametric prostate MRI in predicting presence of tumor on follow-up biopsy
17
ACCEPTED MANUSCRIPT
Table 2-Performance of PSA velocity and of changes in suspicion level or size of focal lesions
T
on serial multi-parametric prostate MRI in predicting presence of tumor with Gleason score > 6
70.9% (39/55) 69.1% (38/55) 70.9% (39/55)
DWI DCE
72.7% (40/55) 65.5% (36/55) 70.9% (39/55) 63.6% (35/55)
ED
Size T2WI
DCE
AC
CE
Any MRI parameter
PT
DWI
Specificity 54.8% (23/42)
38.5% (5/13) 38.5% (5/13) 38.5% (13/13)
81.0% (34/42) 78.6% (33/42) 81.0% (34/42)
15.4% (2/13) 15.4% (2/13) 15.4% (2/13) 61.5% (8/13)
90.5% (38/42) 81.0% (34/42) 88.1% (37/42) 64.3% (27/42)
NU
Suspicion score T2WI
Sensitivity 61.5% (8/13)
SC
Accuracy 56.4% (31/55)
MA
Measure PSA velocity
RI P
on follow-up biopsy
18
ACCEPTED MANUSCRIPT
Figures and Figure Legends
T
Figure 1-80 year-old male with on active surveillance for Gleason 3+4 tumor on previous
RI P
prostate biopsy. Axial T2WI (A), ADC map (B), and parametric perfusion map from DCE (C) from initial prostate MRI examination show a hypervascular focus in the left posterolateral
SC
peripheral zone (arrow, C), without focal corresponding abnormality on T2WI or ADC map. On
NU
axial T2WI (D), ADC map (E), and parametric perfusion map from DCE (F) from follow-up MRI examination ten months later, the lesion has progressed, demonstrating a corresponding
MA
discrete focus of decreased T2 signal (arrow, A) and reduced ADC (arrow, B), while the hypervascular focus on DCE is stable (arrow, C).
PSA velocity during this time was 1.95
PT
ED
ng/ml/year. Subsequent biopsy demonstrated Gleason 4+3 tumor.
CE
Figure 2-68 year-old male on active surveillance for Gleason 3+3 tumor on previous biopsy.
AC
Axial T2WI (A), ADC map (B), and parametric perfusion map from DCE (C) from initial prostate MRI examination show an elliptical focus of decreased T2 signal (arrow, A), reduced ADC (arrow, B), and hypervascularity (arrow, C) within the anterior fibromuscular stroma. On axial T2WI (D), ADC map (E), and parametric perfusion map from DCE (F) from follow-up MRI examination twelve months later, the lesion is stable in size and appearance on all sequences (arrow, D-F). PSA decreased during this time, with a PSA velocity of -0.30 ng/ml/year. Subsequent biopsy demonstrated Gleason 3+3 tumor.
19
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1A
20
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1B
21
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1C
22
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1D
23
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1E
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1F
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2A
26
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2B
27
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2C
28
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2D
29
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2E
30
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2F
31
ACCEPTED MANUSCRIPT Highlights *Change in focal lesions during serial prostate MRI was compared with biopsy results.
T
*Change on serial MRI had lower sensitivity for tumor than PSA velocity.
RI P
*Change on serial MRI had higher specificity for tumor than PSA velocity.
*PSA velocity had higher accuracy than changes on serial MRI in predicting tumor.
AC
CE
PT
ED
MA
NU
SC
*Changes on serial MRI had higher accuracy than PSA velocity for high grade tumor.
32
S0899-7071(14)00230-7 doi: 10.1016/j.clinimag.2014.08.008 JCT 7681
To appear in:
Journal of Clinical Imaging
Received date: Accepted date:
21 July 2014 11 August 2014
Please cite this article as: Rosenkrantz Andrew B, Rice Samuel L, Wehrli Natasha E, Deng Fang-Ming, Taneja Samir S, Association between changes in suspicious prostate lesions on serial MRI examinations and follow-up biopsy results, Journal of Clinical Imaging (2014), doi: 10.1016/j.clinimag.2014.08.008
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 proof before it is published in its final 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.
ACCEPTED MANUSCRIPT Association Between Changes in Suspicious Prostate Lesions on
T
Serial MRI Examinations and Follow-up Biopsy Results
RI P
Original Research
SC
Authors
NU
Andrew B Rosenkrantz MD.1 Samuel L Rice MD. 1 Natasha E Wehrli MD. 1* Fang-Ming Deng MD PhD.2 Samir S Taneja MD.3
Correspondence:
CE
PT
*Current affiliation: Weill Cornell Medical College New York-Presbyterian Hospital New York, NY 10065
ED
All authors: Department of Radiology NYU Langone Medical Center 550 First Avenue New York, NY 10016
MA
1: Department of Radiology 2: Department of Pathology 3: Department of Urology, Division of Urologic Oncology
AC
Andrew B. Rosenkrantz. Department of Radiology Center for Biomedical Imaging NYU Langone Medical Center 660 First Avenue, 3rd Floor New York, NY 10016
Phone: 212-263-0232 Fax: 212-263-6634 Email: [email protected] Support: The Joseph and Diane Steinberg Charitable Trust Disclosures: Author Dr. Samir S Taneja is a consultant for Hitachi and previously has been a consultant for Eigen, Healthtronics, Bayer, and Gtx, has previously received grants from the DOD and NIH for which money was paid to the institution, receives royalties from Elsevier, and previously was on the speaker’s bureau for Janssen. All other authors have no conflicts to disclose. 1
ACCEPTED MANUSCRIPT
Abstract
T
We assessed changes in prostate lesions on serial MRI examinations in predicting biopsy results.
RI P
55 men undergoing two prostate MRI examinations ≥6 months apart, followed by targeted biopsy, were included. Two radiologists assessed dominant lesions for an increase in size or
SC
suspicion score (SS). Progression on MRI had lower sensitivity (23.5%-35.3%) and higher
NU
specificity (76.2%-90.5%) than PSA velocity (sensitivity 70.6%, specificity 52.4%) for predicting positive biopsy. Highest accuracy was achieved by PSA velocity (63.6%) for positive
MA
biopsy, and by MRI (65.5%-72.7%) for Gleason >6 tumor. Findings support lesion progression
PT
ED
on MRI serving as a basis for performing subsequent targeted biopsy.
AC
CE
Key Words: prostate cancer; MRI; biopsy; PSA
2
ACCEPTED MANUSCRIPT 1. Introduction
T
A key challenge in the contemporary management of prostate cancer is the lack of an
RI P
effective means of providing non-invasive monitoring over time of a patient’s level of risk for harboring significant cancer. This ability would be important in a number of contexts. First, in
SC
men with a previous biopsy demonstrating low-risk disease, active surveillance with selective
NU
deferred radical therapy is becoming an increasingly used approach to balance preservation of quality-of-life with adequate oncologic control1. However, current active surveillance regimens
MA
call for serial trans-rectal biopsies to identify subsequent evidence of high-grade disease1, 2. The PSA velocity (PSAv), referring to the rate of change in serum PSA levels over time, is an easy
ED
and practical way to non-invasively monitor a patient as well3, and indeed is widely incorporated
PT
into surveillance algorithms4 and has been noted to be a frequent trigger of delayed radical intervention in this population5. Nonetheless, numerous studies cast strong doubt on the ability
CE
of PSAv to reliably monitor for potential progression of disease in men on surveillance,
6-8
AC
concluding that serial biopsy remains necessary and that better non-invasive markers are needed3, . Furthermore, there has recently been a surge of public skepticism regarding the large number
of biopsies performed in men that demonstrate either low-risk cancer or an absence of tumor altogether9. This concern has contributed to an increasing desire to tailor decisions to perform an initial or repeat biopsy to a patient’s individualized level of risk10, for instance by incorporating a period of non-invasive monitoring in some men to guide the decision of whether to perform a potential biopsy. However, this decision faces the similar dilemma as in the active surveillance cohort, given that numerous studies have also shown poor results for the ability of changes in PSA over time to predict the diagnosis of prostate cancer on an initial or repeat biopsy11-14. Thus, there is a compelling need to derive better tools to provide non-invasive evaluation in men 3
ACCEPTED MANUSCRIPT with suspected or known prostate cancer, which in turn could be useful for optimizing selection of men within these groups to undergo biopsy.
T
Given recent advances in multi-parametric imaging, MRI has improved substantially in
RI P
its ability to localize and depict the extent of tumor within the prostate15 and currently is the only widely available non-invasive imaging test that has an established high performance in
SC
visualizing focal tumors. In addition, findings on a single MRI examination have consistently
NU
been shown to be a significant predictor of the results of subsequent biopsy performed in men with known or suspected prostate cancer16, 17. It has thus been suggested that MRI be routinely
MA
incorporated into such clinical pathways18.
The ability of MRI to provide a distinct illustration of the extent and location of tumor in
ED
the prostate raises the possibility of serial MRI examinations being used to provide non-invasive
PT
monitoring in men contemplating undergoing prostate biopsy19. In such an approach, findings on serial MRI examinations suggesting progression of a visualized lesion could serve as a trigger
CE
for identifying those patients warranting a subsequent biopsy. If this scheme were validated,
AC
MRI could then be used as a tool to provide non-invasive monitoring and risk assessment, and thereby potentially reduce the number of biopsies performed. To our knowledge, the utility of multi-parametric MRI has not been studied in this context. Therefore, the aim of our study is to assess the role of changes in prostate lesions on serial MRI examinations in predicting the pathological results of follow-up prostate biopsies.
2. Methods 2.1 Patients
4
ACCEPTED MANUSCRIPT This retrospective HIPAA-compliant study was approved by our institutional review board with a waiver of the requirement for written informed consent.
We searched an
T
institutional database to identify men who underwent multi-parametric MRI examinations of the
RI P
prostate on separate dates separated by at least six months, with no intervening prostate therapy. The initial examination was required to have occurred after January 2011 given a major change
SC
in our institution’s routine prostate MRI protocol at that time. In men with more than two serial
NU
prostate MRI examinations, the first and last examinations were selected. This search identified an initial set of 162 men. Of these, we then identified those men who underwent a prostate
MA
biopsy after the follow-up MRI, providing a cohort of 64 men. An additional 9 cases were excluded given non-routine protocol on either the initial or follow-up MRI. These exclusions
ED
left a final included cohort of 55 men (mean age 65±8 years; range 48 to 80 years), with a mean
PT
PSA at the time of the initial MRI of 4.7±2.1 ng/mL (range 0.6 to 10.9 ng/mL). The distribution of clinical histories at the time of the initial MRI in these men was as follows: no prior prostate
CE
biopsy (n=14); previous negative prostate biopsy (n=19); prior positive biopsy, on active
AC
surveillance (n=22). The mean time interval between the initial and follow-up MRI was 14±6 months (range 6-32 months). 30 of the men underwent an additional biopsy after the initial MRI examination; however, there was a delay of at least four months between this biopsy and the follow-up MRI examination in all cases.
2.2 MRI Technique MRI was performed using a 3T system (MAGNETOM Trio, Siemens Healthcare, Erlangen, Germany) and a torso phased-array coil.
Examinations included the following
sequences of the prostate and seminal vesicles: turbo spin-echo (TSE) T2-weighted imaging in
5
ACCEPTED MANUSCRIPT the axial, coronal, and sagittal planes (TR/TE, 4000-4960/105; slice thickness, 3 mm; no interslice gap for axial and coronal planes, 0.6 mm interslice gap for sagittal plane; FOV, 180 ×
T
180 mm; matrix, 256 × 256 for axial and coronal planes, 256 x 230 for sagittal plane; parallel
RI P
imaging factor, 2; 3 signal averages for axial and coronal planes, 2 signal averages for sagittal plane), axial TSE T1-weighted imaging (TR/TE, 683/10; slice thickness, 3 mm, no interslice
SC
gap; FOV, 180 × 180 mm; matrix, 192 × 192; parallel imaging factor, 2; 2 signal averages) and
NU
axial single-shot echo-planar diffusion-weighted imaging (DWI) (TR/TE, 4100/86; slice thickness, 3 mm; no interslice gap; FOV, 200 × 200 mm; matrix, 100 × 100; parallel imaging
MA
factor, 2; 10 signal averages; b values, 50 and 1000 s/mm2) with reconstruction of the apparent diffusion coefficient (ADC) map using a standard mono-exponential fit. Dynamic contrast-
ED
enhanced MRI (DCE-MRI) of the prostate was performed using a 3D T1-weighted spoiled
PT
gradient-echo sequence (TR/TE, 2.84/0.94; flip angle, 16°; partition thickness, 3 mm; FOV, 240 × 240 mm; matrix, 128 × 128; no parallel imaging; 1 signal average) before and at multiple time-
CE
points after the intravenous administration of 0.1 mmol/kg of gadopentetate dimeglumine
AC
(Magnevist, Bayer HealthCare). The contrast material was administered as a bolus via power injector (Spectris, Medrad) and followed by a 20-mL saline flush, both administered at a rate of 3 mL/s. The temporal resolution for DCE was 5.5 seconds. 55 post-contrast measures were obtained, for a total acquisition time after contrast injection of 5 minutes 5 seconds. A biexponential model based on the raw DCE images was applied to generate parametric maps representing the maximum slope of enhancement during the contrast-enhanced acquisition and the washout of contrast material after the enhancement peak, hereafter referred to as the “maximum slope” and “washout maps,” respectively.
Commercial software was used for
generating the parametric maps (DynaCAD version 2.1.6, Invivo).
6
ACCEPTED MANUSCRIPT
2.3 Image Analysis
T
Cases were reviewed in consensus by two fellowship-trained radiologists with 2 and 6
RI P
years of experience in prostate MRI interpretation who were blinded to clinical and pathological details. The radiologists initially observed the baseline examination performed in each patient,
SC
reviewing all sequences in joint fashion. The radiologists identified the most suspicious lesion in
NU
each case, as guided by the PI-RADS criteria provided within the guidelines set forth by the European Society of Uroradiology expert panel20, which was considered to represent the
MA
dominant lesion. For the dominant lesion, the observers separately recorded the lesion’s level of suspicion on each sequence (T2WI, DWI, and DCE), again using the explicit PI-RADS criteria,
ED
recording a score of 1 for sequences on which the lesion was not visible. A score of 1 was
PT
recorded for all sequences in cases with no visible dominant lesion. The radiologists also measured the dominant lesion’s single largest diameter in the axial plane on each sequence. For
CE
DWI, lesion size was recorded using the ADC maps given a past study showing better
AC
performance of the ADC map than high b-value images in predicting tumor sizes on radical prostatectomy21; for DCE, lesion size was recorded using an individual post-contrast time-point depicting the lesion given greater clarity of lesion boundaries on these images than on postprocessed parametric color maps. No size was recorded for sequences in which the dominant lesion was not visualized.
After completing evaluation of the baseline examination, the
radiologists then observed the follow-up examination in similar fashion, again recording the suspicion level and size of the dominant lesion on each sequence. During this assessment, the radiologists were aware of which lesion was selected as the dominant lesion on the initial study. Following evaluation of all examinations, the radiologists’ readings were categorized in binary
7
ACCEPTED MANUSCRIPT fashion in terms of whether or not an increase was identified between examinations in suspicion level or size. For suspicion level, an increase of at least one category on the 1-5 PI-RADS scale
T
was considered to represent an increase; for size, given imprecision of individual size
RI P
measurements, an increase in maximal diameter of at least 3 mm was considered to represent an increase. These categorizations were based solely on the presence of an increase, irrespective of
SC
the actual values themselves on the two examinations.
NU
In addition, the electronic medical record was reviewed to record the PSA at the time of the baseline and follow-up MRI examinations for each patient, defined as the PSA obtained
MA
immediately preceding each MRI. Based on these values, the PSAv was recorded for each
ED
patient as (PSAfollow-up – PSAbaseline) / (time interval between serial PSA values)19.
PT
2.4 Reference standard
The reference standard in all cases was trans-rectal ultrasound-guided prostate biopsy
CE
performed after the follow-up MRI, which at our institution routinely includes cores targeting
AC
MRI-detected abnormalities, in addition to standard systematic cores from all sextants. Biopsy results were categorized as positive or negative for tumor, as well as positive or negative for tumor with a Gleason score greater than 6.
2.5 Statistics Receiver-operating-characteristic analysis was performed to identify optimal thresholds in terms of PSAv for predicting presence of any cancer as well as cancer with Gleason score > 6, defined as that threshold maximizing the average of sensitivity and specificity; PSAv values were then converted to a binary measure as being greater or lesser than these thresholds. This
8
ACCEPTED MANUSCRIPT step was performed to allow comparison within a single analysis of the utility of PSAv with that of changes in lesion suspicion score or size, which were also recorded in binary fashion. The
T
Cochrane Q-test was performed to compare the sensitivity, specificity, and accuracy of a panel of
RI P
measures (PSAv, change in suspicion score or size for T2WI, DWI, and DCE, as well as change in any MRI parameter) for predicting presence of tumor as well as of tumor with Gleason score
SC
greater than 6. The p-values provided from this analysis are adjusted for multiple comparisons
NU
and were considered statistically significant at p<0.05. The statistical analysis was performed
MA
using software (MedCalc for Windows, version 12.7, MedCalc Software, Ostend, Belgium).
3. Results
ED
3.1 Clinical data
PT
The follow-up biopsy was positive for tumor in 61.8% (34/55), and negative in 38.2% (21/55), of cases. In addition, the follow-up biopsy was positive for tumor with Gleason score >
CE
6 in 23.6% (13/55) of cases. In the 13 cases with Gleason score > 6, 10 had a Gleason score of
AC
3+4 and 3 had a Gleason score of 4+3. In the 55 included cases, mean PSAv was 0.66±1.50 ng/mL/year (range, -1.97 to +5.15 ng/mL/year). Representative cases are shown in Figures 1-2.
3.2 Prediction of biopsy results Of the 55 cases, 23.6% (13/55), 25.5% (14/55), and 23.6% (13/55), showed an increase in suspicion level on T2WI, DWI, and DCE, respectively; 10.9% (6/55), 18.2% (10/55), and 12.7% (7/55), showed an increase in size on T2WI, DWI, and DCE, respectively; and 41.8% (23/55) showed an increase using any MRI parameter. ROC demonstrated optimal PSAv thresholds of +0.09 ng/ml/year and +0.48 ng/ml/year for predicting presence of any tumor and tumor with
9
ACCEPTED MANUSCRIPT Gleason score > 6 on follow-up biopsy. 61.8% (34/55) and 63.6% (35/55) of cases exhibited a PSAv meeting these respective thresholds.
T
Table 1 shows the sensitivity, specificity, and accuracy of each of the measures in
RI P
predicting the presence of tumor on follow-up biopsy. Overall, change in suspicion level or size on MRI had low sensitivity (range 23.5%-35.3%) but good specificity (range 76.2%-90.5%) in
SC
predicting presence of tumor. PSAv had higher sensitivity (70.6%) but lower specificity (52.4%)
NU
in comparison with the MRI-based measures. Highest overall accuracy for predicting presence of tumor was obtained by PSAv (63.6%). The sensitivities of all of the individual MRI features,
MA
the specificities of change in size on T2WI, change in suspicion score on DWI, or change in size or suspicion score on DCE, as well as the accuracy of change in size on T2WI, were all
ED
significantly difference compared with the performance of PSAv (p<0.05). Otherwise, there was
PT
no significant difference in sensitivity, specificity, or accuracy between any combination of the individual MRI-based measures and PSAv, although there were non-significant greater
CE
sensitivities for predicating a positive biopsy based on a change in suspicion score than a change
AC
in size for all three MRI sequences. Table 2 shows the sensitivity, specificity, and accuracy of each of the measures in predicting the presence of tumor with Gleason score > 6 on follow-up biopsy. Sensitivities using the MRI parameters tended to be higher in this analysis in comparison with the sensitivity of MRI in predicting presence of tumor.
Nonetheless, as in the prior assessment, change in
suspicion level or size on MRI had lower sensitivity (range 15.4%-38.5%) and higher specificity (range 78.6%-90.5%) in comparison with PSAv (sensitivity 61.5%, specificity 54.8%).
In
comparison with the analysis for prediction of tumor, all of the MRI assessments based on change in size or suspicion score exhibited higher accuracy for predicting tumor with Gleason
10
ACCEPTED MANUSCRIPT score > 6 (range 65.5%-72.7%) than did PSAv (56.4%). The specificity of all of the MRI parameters were significantly different than that of PSAv (p<0.05); no here was significant
T
difference in sensitivity, specificity, or accuracy between any combination of the individual
RI P
MRI-based measures and PSAv, although there were non-significant greater sensitivities for predicating tumor with a Gleason score > 6 based on a change in suspicion score than a change
NU
SC
in size for all three MRI sequences.
MA
4. Discussion
In this study, we evaluated the potential role of detection of changes in focal prostate
ED
lesions on serial MRI examinations in predicting pathologic findings on follow-up biopsy. The
PT
presence of progression of a lesion on imaging was observed to have low sensitivity but high specificity in prediction of subsequent biopsy results. The low sensitivity is likely attributable to
CE
the known slow rate of biologic progression and prolonged natural disease course of prostate
AC
cancer in most cases1. Indeed, prostate cancer often has a pre-clinical phase of over a decade22. To this end, identification of either cancer or high-grade cancer that was not detected on an earlier biopsy often represents undersampling on the prior biopsy rather than true interval disease progression23. Thus, given the stability of many cases of prostate cancer over the time interval under evaluation in this study, an overt change in appearance may not be expected. On the other hand, although interval progression on imaging was observed in a small fraction of cases, this most often represented high-grade tumor. This observation is consistent with the known indolent course and often clinical insignificance of Gleason 6 prostate cancer24, in comparison with the more aggressive nature of higher Gleason score tumors that warrant therapy.
11
ACCEPTED MANUSCRIPT In comparison with serial changes on MRI, PSAv exhibited higher sensitivity and lower specificity in prediction of follow-up biopsy results. A fraction of prostate cancers are inherently
T
invisible or exhibit poor conspicuity on MRI, for reasons relating to their histologic architecture
assessing such lesions.
RI P
or other mechanisms not yet fully understood25, and PSAv may provide an alternate means of Past studies have also demonstrated a role of PSA-based data in
SC
complementing MRI for detection of some prostate cancer cases26,
27
. However, PSA lacks
NU
specificity, also changing in response to prostatic inflammation, hormonal status, hyperplasia, and other factors, contributing to its lower specificity22,
28
. When considering sensitivity and
MA
specificity together, the overall performance of PSAv was moderate in our study, consistent with results of numerous prior studies3, 6-8. For instance, Vickers et al. reported an AUC of 0.702 for
ED
PSAv in predicting biopsy results in the initial detection of prostate cancer11, and Ng et al.
PT
reported an AUC of 0.70 in predicting presence of adverse histology on follow-up biopsy in men on active surveillance29. Thus, our findings agree with past studies in indicating that PSAv is not
CE
sufficient as a non-invasive biomarker in providing guidance for biopsy decisions.
AC
One interesting observation was a somewhat higher performance for changes in lesions’ suspicion score than changes in size in predicting biopsy results, although these differences were not significant, possibly relating to the small numbers of cases with serial imaging changes as well as with positive biopsy results. Obtaining a precise measure of prostate lesion size can be difficulty when performed manually given that prostate tumors often do not form a discrete, well-marginated mass, as encountered by tumors in some other organs. In many cases, the tumor may have irregular margins, insinuating with normal tissue, as well as exhibit heterogeneous signal intensity with areas of relatively subtle signal alterations. It is also possible that other examination-related factors, such as slight differences in slice positioning and orientation, as well
12
ACCEPTED MANUSCRIPT as varying degrees of patient motion, between the examinations, may also influence apparent lesion size. These factors can introduce error into the size measurements and confound the
T
assessment for potentially subtle changes in size between studies. On the other hand, the
RI P
suspicion level is determined based on five discrete categories, each having its own distinct criteria20. It is possible that this determination, based upon the judgment of the interpreting
SC
radiologist, is less susceptible to the imprecisions that confound the size measurements.
NU
Our findings are clinically relevant in men with both suspected as well as with known prostate cancer and who are on surveillance. Specifically, there is currently a lack of a reliable
MA
means of appropriately stratifying such cohorts and optimally selecting those men who warrant biopsy. This limitation has led to a very high rate of sequential biopsies in such cohorts30. Our
ED
findings support the routine performance of biopsy following an apparent change in size or
PT
suspicion score of a focal lesion on MRI, given the very high specificity of such an occurrence, including for high-grade disease. Thus, the presence of progression on imaging identifies a
CE
subset of men who seem to be at particularly high risk and who more confidently warrant biopsy.
AC
However, given the low sensitivity of changes in imaging findings, serial imaging is not currently sufficient to avoid performing a repeat biopsy. Investigation of additional means of identifying the subset of men harboring significant disease that may have been missed on both an initial biopsy as well as on serial MRI examinations is therefore warranted. A number of limitations of our study warrant mention. First, the sample size was small, including only a small number of cases that were positive for tumor. In addition, the patient population was heterogeneous, including both patients with and without known prostate cancer prior to the initial MRI. Nonetheless, the potential value of using MRI to provide non-invasive monitoring and selection of patients for biopsy is important in both groups, and serial MRI has
13
ACCEPTED MANUSCRIPT been applied in a comparable fashion in both cohorts in our practice. Also, the time interval between the two MRI examinations was variable within the cohort. While the PSAv calculation
T
adjusts for the variability in time interval between measurements, it was not possible to adjust the
RI P
changes in MRI findings in this fashion. This variability in timing of the examinations reflects the retrospective nature of our study and the lack of a current accepted standard or guideline as to
SC
the appropriate frequency at which serial MRI examinations should be performed in this context.
NU
Finally, as this study was retrospective, as previously noted, the actual performance of serial MRI for guiding prostate cancer management in a prospective setting remains unknown, and
MA
future studies incorporating serial MRI findings into clinical pathways and decision-making remain warranted.
ED
In conclusion, we observed low sensitivity and high specificity of changes in focal In
PT
prostate lesions on serial MRI in predicting the results of subsequent prostate biopsy.
comparison, PSAv had moderate performance in terms of both sensitivity and specificity. While
CE
the highest overall for predicting a positive biopsy was achieved by PSAv, the highest accuracy
on MRI.
AC
for predicting a Gleason score greater than 6 was achieved by a change in size or suspicion score Our findings indicate that in men contemplating prostate biopsy, the biopsy is
warranted if there is apparent progression on imaging, although it is not possible at this time to avoid deferring biopsy on the basis of stability of a lesion on serial imaging. Further studies are necessary to evaluate the role of these findings in guiding clinical decisions regarding the performance of prostate biopsy in a larger prospective patient cohort.
14
ACCEPTED MANUSCRIPT References
7.
8.
9. 10. 11.
12.
13.
14.
15.
16.
17.
T
RI P
SC
NU
6.
MA
5.
ED
4.
PT
3.
CE
2.
KLOTZ L. Active surveillance for prostate cancer: overview and update. Current treatment options in oncology 2013;14:97-108. LAWRENTSCHUK N, KLOTZ L. Active surveillance for favorable-risk prostate cancer: a short review. Korean J Urol 2010;51:665-70. VAN DEN BERGH RC, ROEMELING S, ROOBOL MJ, et al. Prostate-specific antigen kinetics in clinical decision-making during active surveillance for early prostate cancer--a review. European urology 2008;54:505-16. KLOTZ L. Active surveillance versus radical treatment for favorable-risk localized prostate cancer. Current treatment options in oncology 2006;7:355-62. COEN JJ, FELDMAN AS, SMITH MR, et al. Watchful waiting for localized prostate cancer in the PSA era: what have been the triggers for intervention? BJU international 2011;107:1582-6. IREMASHVILI V, MANOHARAN M, LOKESHWAR SD, et al. Comprehensive analysis of post-diagnostic prostate-specific antigen kinetics as predictor of a prostate cancer progression in active surveillance patients. BJU international 2013;111:396-403. AYRES BE, MONTGOMERY BS, BARBER NJ, et al. The role of transperineal template prostate biopsies in restaging men with prostate cancer managed by active surveillance. BJU international 2012;109:1170-6. ROSS AE, LOEB S, LANDIS P, et al. Prostate-specific antigen kinetics during follow-up are an unreliable trigger for intervention in a prostate cancer surveillance program. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2010;28:2810-6. MCNAUGHTON-COLLINS MF, BARRY MJ. One man at a time--resolving the PSA controversy. The New England journal of medicine 2011;365:1951-3. ROOBOL MJ, CARLSSON SV. Risk stratification in prostate cancer screening. Nature reviews. Urology 2013;10:38-48. VICKERS AJ, TILL C, TANGEN CM, et al. An empirical evaluation of guidelines on prostate-specific antigen velocity in prostate cancer detection. Journal of the National Cancer Institute 2011;103:462-9. BANGMA CH, VAN SCHAIK RH, BLIJENBERG BG, et al. On the use of prostate-specific antigen for screening of prostate cancer in European Randomised Study for Screening of Prostate Cancer. European journal of cancer 2010;46:3109-19. VICKERS AJ, SAVAGE C, O'BRIEN MF, et al. Systematic review of pretreatment prostate-specific antigen velocity and doubling time as predictors for prostate cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2009;27:398-403. VICKERS AJ, WOLTERS T, SAVAGE CJ, et al. Prostate-specific antigen velocity for early detection of prostate cancer: result from a large, representative, population-based cohort. European urology 2009;56:753-60. TURKBEY B, MANI H, SHAH V, et al. Multiparametric 3T prostate magnetic resonance imaging to detect cancer: histopathological correlation using prostatectomy specimens processed in customized magnetic resonance imaging based molds. J Urol 2011;186:1818-24. PINTO PA, CHUNG PH, RASTINEHAD AR, et al. Magnetic resonance imaging/ultrasound fusion guided prostate biopsy improves cancer detection following transrectal ultrasound biopsy and correlates with multiparametric magnetic resonance imaging. J Urol 2011;186:1281-5. VOURGANTI S, RASTINEHAD A, YERRAM NK, et al. Multiparametric magnetic resonance imaging and ultrasound fusion biopsy detect prostate cancer in patients with prior negative transrectal ultrasound biopsies. The Journal of urology 2012;188:2152-7.
AC
1.
15
ACCEPTED MANUSCRIPT
26.
27.
28. 29. 30.
T
RI P
SC
NU
25.
MA
24.
ED
21. 22. 23.
PT
20.
CE
19.
AHMED HU, KIRKHAM A, ARYA M, et al. Is it time to consider a role for MRI before prostate biopsy? Nature reviews. Clinical oncology 2009;6:197-206. COAKLEY FV, CHEN I, QAYYUM A, et al. Validity of prostate-specific antigen as a tumour marker in men with prostate cancer managed by watchful-waiting: correlation with findings at serial endorectal magnetic resonance imaging and spectroscopic imaging. BJU Int 2007;99:41-5. BARENTSZ JO, RICHENBERG J, CLEMENTS R, et al. ESUR prostate MR guidelines 2012. Eur Radiol 2012;22:746-57. BLINDED SIKARIS K. Prostate cancer screening. Pathology 2012;44:99-109. DUFFIELD AS, LEE TK, MIYAMOTO H, et al. Radical prostatectomy findings in patients in whom active surveillance of prostate cancer fails. The Journal of urology 2009;182:2274-8. DONIN NM, LAZE J, ZHOU M, et al. Gleason 6 prostate tumors diagnosed in the PSA era do not demonstrate the capacity for metastatic spread at the time of radical prostatectomy. Urology 2013;82:148-52. LANGER DL, VAN DER KWAST TH, EVANS AJ, et al. Intermixed normal tissue within prostate cancer: effect on MR imaging measurements of apparent diffusion coefficient and T2--sparse versus dense cancers. Radiology 2008;249:900-8. VILANOVA JC, COMET J, BARCELO-VIDAL C, et al. Peripheral zone prostate cancer in patients with elevated PSA levels and low free-to-total PSA ratio: detection with MR imaging and MR spectroscopy. Radiology 2009;253:135-43. VILANOVA JC, BARCELO-VIDAL C, COMET J, et al. Usefulness of prebiopsy multifunctional and morphologic MRI combined with free-to-total prostate-specific antigen ratio in the detection of prostate cancer. AJR. American journal of roentgenology 2011;196:W715-22. GREENE KL, ALBERTSEN PC, BABAIAN RJ, et al. Prostate specific antigen best practice statement: 2009 update. The Journal of urology 2013;189:S2-S11. NG MK, VAN AS N, THOMAS K, et al. Prostate-specific antigen (PSA) kinetics in untreated, localized prostate cancer: PSA velocity vs PSA doubling time. BJU international 2009;103:872-6. PLOUSSARD G, NICOLAIEW N, MARCHAND C, et al. Risk of repeat biopsy and prostate cancer detection after an initial extended negative biopsy: longitudinal follow-up from a prospective trial. BJU international 2013;111:988-96.
AC
18.
16
ACCEPTED MANUSCRIPT
Tables
T
Table 1-Performance of PSA velocity and of changes in suspicion level or size of focal lesions
43.6% (24/55) 56.4% (31/55) 50.9% (28/55)
DWI DCE
38.2% (21/55) 41.8% (23/55) 43.6% (24/55) 54.6% (30/55)
ED
Size T2WI
DCE
AC
CE
Any MRI parameter
PT
DWI
Specificity 52.4% (10/21)
23.5% (8/34) 35.3% (12/34) 29.4% (10/34)
76.2% (16/21) 90.5% (19/21) 85.7% (18/21)
8.8% (3/34) 17.7% (6/34) 14.7% (5/34) 47.1% (16/34)
85.7% (18/21) 81.0% (17/21) 90.5% (19/21) 66.7% (14/21)
NU
Suspicion score T2WI
Sensitivity 70.6% (24/34)
SC
Accuracy 63.6% (35/55)
MA
Measure PSA velocity
RI P
on serial multi-parametric prostate MRI in predicting presence of tumor on follow-up biopsy
17
ACCEPTED MANUSCRIPT
Table 2-Performance of PSA velocity and of changes in suspicion level or size of focal lesions
T
on serial multi-parametric prostate MRI in predicting presence of tumor with Gleason score > 6
70.9% (39/55) 69.1% (38/55) 70.9% (39/55)
DWI DCE
72.7% (40/55) 65.5% (36/55) 70.9% (39/55) 63.6% (35/55)
ED
Size T2WI
DCE
AC
CE
Any MRI parameter
PT
DWI
Specificity 54.8% (23/42)
38.5% (5/13) 38.5% (5/13) 38.5% (13/13)
81.0% (34/42) 78.6% (33/42) 81.0% (34/42)
15.4% (2/13) 15.4% (2/13) 15.4% (2/13) 61.5% (8/13)
90.5% (38/42) 81.0% (34/42) 88.1% (37/42) 64.3% (27/42)
NU
Suspicion score T2WI
Sensitivity 61.5% (8/13)
SC
Accuracy 56.4% (31/55)
MA
Measure PSA velocity
RI P
on follow-up biopsy
18
ACCEPTED MANUSCRIPT
Figures and Figure Legends
T
Figure 1-80 year-old male with on active surveillance for Gleason 3+4 tumor on previous
RI P
prostate biopsy. Axial T2WI (A), ADC map (B), and parametric perfusion map from DCE (C) from initial prostate MRI examination show a hypervascular focus in the left posterolateral
SC
peripheral zone (arrow, C), without focal corresponding abnormality on T2WI or ADC map. On
NU
axial T2WI (D), ADC map (E), and parametric perfusion map from DCE (F) from follow-up MRI examination ten months later, the lesion has progressed, demonstrating a corresponding
MA
discrete focus of decreased T2 signal (arrow, A) and reduced ADC (arrow, B), while the hypervascular focus on DCE is stable (arrow, C).
PSA velocity during this time was 1.95
PT
ED
ng/ml/year. Subsequent biopsy demonstrated Gleason 4+3 tumor.
CE
Figure 2-68 year-old male on active surveillance for Gleason 3+3 tumor on previous biopsy.
AC
Axial T2WI (A), ADC map (B), and parametric perfusion map from DCE (C) from initial prostate MRI examination show an elliptical focus of decreased T2 signal (arrow, A), reduced ADC (arrow, B), and hypervascularity (arrow, C) within the anterior fibromuscular stroma. On axial T2WI (D), ADC map (E), and parametric perfusion map from DCE (F) from follow-up MRI examination twelve months later, the lesion is stable in size and appearance on all sequences (arrow, D-F). PSA decreased during this time, with a PSA velocity of -0.30 ng/ml/year. Subsequent biopsy demonstrated Gleason 3+3 tumor.
19
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1A
20
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1B
21
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1C
22
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1D
23
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1E
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 1F
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2A
26
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2B
27
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2C
28
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2D
29
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2E
30
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
SC
RI P
T
Fig. 2F
31
ACCEPTED MANUSCRIPT Highlights *Change in focal lesions during serial prostate MRI was compared with biopsy results.
T
*Change on serial MRI had lower sensitivity for tumor than PSA velocity.
RI P
*Change on serial MRI had higher specificity for tumor than PSA velocity.
*PSA velocity had higher accuracy than changes on serial MRI in predicting tumor.
AC
CE
PT
ED
MA
NU
SC
*Changes on serial MRI had higher accuracy than PSA velocity for high grade tumor.
32