520 656
after adjusting for CAPRA-S (Table 2), it is not certain that
the performance improvement in AUC from 0.76 (CAPRA-S)
to 0.80 (combined model) as described in Supplementary
Table 17 is clinically significant or relevant. Nevertheless,
this is not an uncommon observation for many ROC
analyses. However, the assay alone had sensitivity of 48%
versus 71% with CAPRA-S alone, and the combined
(metastatic assay + CAPRA-S) specificity was calculated as
61%. Furthermore, with the decision curve analysis (Sup-
plementary Fig. 6) this result is not very impressive, even
though the authors selected or ‘‘cherry picked’’ a risk
threshold of 0.25, at which their model looks the best.
Not uncommon in model building, mention is made in
the Supplementary methods that ‘‘following migration of
the metastatic assay to a platform with an improved
chemistry
. . .
a technical bias adjustment was applied to the
assay threshold which was used to dichotomize assay
scores for resection clinical validation cohort
. . .
’’ Were
thresholds different for the discovery and validation
experiments? ‘‘Locked’’ weights and biases for genes in
the model are appropriately provided, and it is not unusual
to have different offsets, yet the procedure to arrive at ‘‘the
bias adjustment’’ is not provided, and the hope is that this
was done before validation(s).
Regardless, the authors are to be congratulated on their
assay development and results. To us, the most important
question is what clinicians should do with the result. It has
prognostic potential, but is it predictive? In addition, one
would hope that if the assay were positive, treatment would
not be denied but rather ramped up or prospectively
investigated as more evidence accumulates that primary-
directed therapy can improve survival in metastatic
prostate cancer
[9] .To paraphrase the author Ray Bradbury,
we should not be ‘‘predicting the future’’, but ‘‘trying to
prevent it.’’
Conflicts of interest:
Decipher was co-developed by GenomeDx and the
Mayo Clinic. The authors are employees of the Mayo Clinic. R. Jeffrey
Karnes has received research funding and travel reimbursement from
GenomeDx.
References
[1]
Walker SM, Knight LA, McCavigan AM, et al. Molecular subgroup of primary prostate cancer presenting with metastatic biology. Eur Urol 2017;72:509–18.
[2]
Cheville JC, Karnes RJ, Therneau TM, et al. Gene panel model predic- tive of outcome in men at high-risk of systemic progression and death from prostate cancer after radical retropubic prostatectomy. J Clin Oncol 2008;26:3930–6.
[3]
Karnes RJ, Cheville JC, Ida CM, et al. The ability of biomarkers to predict systemic progression in men with high-risk prostate cancer treated surgically is dependent on ERG status. Cancer Res 2010;70: 8994–9002.
[4]
Erho N, Crisan A, Vergara IA, et al. Discovery and validation of a prostate cancer genomic classifier that predicts early metastasis following radical prostatectomy. PLoS One 2013;8:e66855.
[5]
Karnes RJ, Bergstralh EJ, Davicioni E, et al. Validation of a genomic classifier that predicts metastasis following radical prostatectomy in an at risk patient population. J Urol 2013;190:2047–53.
[6]
Cooperberg MR, Davicioni E, Crisan A, Jenkins RB, Ghadessi M, Karnes RJ. Combined value of validated clinical and genomic risk stratifica- tion tools for predicting prostate cancer mortality in a high-risk prostatectomy cohort. Eur Urol 2015;67:326–33.
[7] Karnes RJ, Choeurng V, Ross AE, et al. Validation of a genomic risk
classifier to predict prostate cancer-specific mortality in men with
adverse pathologic features. Eur Urol. In press.
http://dx.doi.org/ 10.1016/j.eururo.2017.03.036 .[8]
Lee HJ, Yousefi K, Haddad Z, et al. Evaluation of a genomic classifier in radical prostatectomy patients with lymph node metastasis. Res Rep Urol 2016;8:77–84.[9]
Leyh-Bannurah S-R, Gazdovich S, Budaus L, et al. Local therapy improves survival in metastatic prostate cancer. Eur Urol 2017;72:118–24.
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 5 1 9 – 5 2 0
520