EURURO Vol. 72 No. 2

315 320

Words of Wisdom
Re: Diagnostic Accuracy of Multi-parametric Magnetic
Resonance Imaging and Transrectal Ultrasound Biopsy
in Prostate Cancer (PROMIS): A Paired Validating
Confirmatory Study
Ahmed HU, El-Shater Bosaily A, Brown LC, et al
Lancet 2017;389:815–22
Experts’ summary:
In this prospective, multicenter, paired-cohort study the
authors evaluated the utility of multi-parametric magnetic
resonance imaging (MP-MRI) as a gateway-test to invasive
testing in men with elevated prostate-specific antigen (PSA).
They also compared the diagnostic accuracies of MP-MRI and
transrectal ultrasound guided biopsy (TRUS-biopsy) in detect-
ing clinically significant prostate cancer that was defined as
Gleason score 4 + 3 or a cancer core length 6 mm or longer.
The accuracy of both tests was vetted against template prostate
mapping biopsy (TPM-biopsy). A total of 576 biopsy-naı¨ve men
underwent MP-MRI, TRUS-biopsy, and TPM-biopsy. On TPM-
biopsy, 71% (n = 408) of men had cancer, with 40% (n = 230)
harboring clinically significant disease. For clinically significant
cancer, MP-MRI was more sensitive (93%) than TRUS-biopsy
(48%, p < 0.0001) and less specific (41% for MP-MRI vs 96% for
TRUS-biopsy, p < 0.0001).
Experts’ comments:
The authors are applauded for conducting this timely and
important study. The study answers several vital questions:
(1) is MP-MRI a better screening tool than TRUS-biopsy for
detecting clinically significant prostate cancer—yes indeed;
the authors demonstrate that in biopsy-naı¨ve men with ele-
vated PSA, the MP-MRI did a much better job of catching
clinically significant disease than the TRUS-biopsy (see sensi-
tivities above), (2) is a 1.5-Tesla MR without endorectal coil
capable of detecting clinically meaningful lesions with high
sensitivity—yes absolutely; utilizing several different defini-
tions of clinically significant prostate cancer, the authors
showed that the detection sensitivities for the 1.5-Tesla
MP-MRI ranged between 87% and 93%, much higher than
those for the TRUS-biopsy (48–60%, p < 0.0001), (3) can the
results of an expert center be generalized to community
practice—yes again; provided that strict quality control mea-
sures for image acquisition, and uniform and fastidious train-
ing for radiologists is undertaken across all centers, and finally
(4) can/should MP-MRI be used as a triage test to invasive
testing in men with elevated PSA—based on aforementioned
data, probably yes. At our institution, we are adopting this
paradigm, and also utilizing MP-MRI in managing patients on
active surveillance.
Among the questions that come to mind next, and are
unanswered, the more obvious are those related to the
issues of healthcare economics and dissemination of new
technology such as cost-effectiveness, availability of the
test to all, development of validated training curricula for
the radiologists, etc. The more important questions,
however, two in particular, are those regarding the poor
specificity and definition of radiological significance. It is
not lost on the writers that the MP-MRI here is being
assessed as a screening test, but the question of poor
specificity remains, as it affects the burden of unnecessary
biopsies. In this regard, would use of a 3-Tesla MR help, and/
or is there a role for image-learning iterative computational
systems, both of which have been shown to improve signal-
to-noise ratio . From a radiological significance
standpoint, is Prostate Imaging Reporting and Data System
(PIRADS)-3 category, defined as ‘‘equivocal risk,’’ being
abused as a diagnostic safe haven—PIRADS-3 (or a Likert
score 3) was the most commonly diagnosed radiographic
lesion (n = 163 of 576) in the study, which adds little value to
patient care on account of being equivocal—should thus the
PIRADS classification be altered along the lines of the
2004 update of 1973 World Health Organization noninva-
sive urothelial cancer classification to encourage
unequivocality. The answers to these questions hopefully
will come with new research, and experience. Starting the
dialogue was important, which this study has, with resound.
Conflicts of interest: Firas Abdollah is an advisor for Genome Dx. Akshay
Sood is a member of the IDEAL collaboration at University of Oxford.
References
[1]
.
[2]
ava ilable at
journal homepage:
0302-2838/