EURURO Vol. 72 No. 5

667 860

2.3. Genomic instability and mRNA abundance profiling
Within the Canadian cohort, tumor tissues of 242 and 156 cases were
successfully processed for DNA and total RNA, respectively. The tissues
were obtained either as fresh-frozen prostatectomy specimens from the
UHN Pathology BioBank and the Genito-Urinary BioBank of CHUdeQ-UL
or as fresh-frozen single ultrasound-guided needle biopsy, and
processed as previously described . Full details on tumor
sampling and processing are outlined in the Supplementary material.
Copy number aberrations (CNAs) were profiled by single-nucleotide
polymorphism microarrays performed using 50–200 ng of DNA by
Affymetrix OncoScan FFPE Express 3.0 (Thermo Fisher Scientific[15_TD$DIFF], Santa
Clara, CA, USA). For microarray analysis, 500 ng of total RNA was assayed
using the Affymetrix HuGene 2.0 array at the Centre for Applied
Genomics, The Hospital for Sick Children, Toronto, Canada. Analyses for
gene-level CNA and mRNA abundance were as previously described ,
and are elaborated in the Supplementary material. PGA was calculated
by dividing the number of base pairs that are involved in a copy number
change over the total length of the genome. Chromothripsis scores,
generated by ShatterProof , were obtained from a previously
published whole genome sequencing study . Tumors with a
maximum ShatterProof score of >0.517 were considered to be
chromothriptic.
2.4. Measurement of focal tumor hypoxia in UHN radiotherapy
cohort
Intraprostatic measurements of pO2 were obtained before radiotherapy
in 102 patients, using an ultrasound-guided transrectal needle piezo-
electrode, as previously described For each patient, between 40 and
80 pO2 measurements were obtained along two to four measurement
tracks in regions of the prostate gland expected to contain tumor, based
on the diagnostic biopsies. HP20 (the percentage of pO2 measurements
<20 mmHg) was used for the hypoxia analysis Tumors were
determined to be hypoxic if their HP20 values were higher than the
median HP20 of the cohort. No patient received neoadjuvant hormonal
therapy prior to pO2 measurements.
2.5. TMA construction and SChLAP1 RNA-ISH
TMAwas constructed using prostatectomy specimens from the EMC cohort.
Fromeach specimen, three separate tissue cores of 0.6 mmwere included in
the TMA. Details on TMA construction were as previously described
. TMA cores were selected from the most representative sections of the
prostatectomy specimens based on tumor size and highest Gleason score.
SChLAP1 RNA-ISH was then performed on 1-wk-old 5mm sections using
RNAscope according to the manufacturer’s protocol (Advanced Cell
Diagnostics, Inc., Hayward, CA, USA). Full staining protocol[1_TD$DIFF], including the
probe sequence, is outlined in the Supplementary material[2_TD$DIFF], and has been
previously published Scoring of TMA was performed by estimation of
mean RNA-ISH signal spots in the tumor areas (A.M.H.). SChLAP1 positivity
was classified by the presence of 5 signal spots. Identification of IDC/CA in
the TMA was performed by two independent observers who were not
involved in the scoring of RNA-ISH signal intensity.
2.6. Statistical analysis
All statistical analyses were performed using R v.3.2.2 package. Fisher
exact test, Mann–Whitney U test, and Kruskal–Wallis test were used to
examine associations of IDC/CA subpathology with clinical indices at
diagnosis (age, cT category, diagnostic Gleason score, or International
Society of Urological Pathology grading system for prostate cancer based
on Gleason score [ISUP grade] [3_TD$DIFF] , and prostate-specific antigen [PSA]),
and molecular factors of PGA, hypoxia, and SChLAP1 expression. Risk of
biochemical relapse for IDC/CA was assessed in a Canadian intermedi-
ate-risk cohort, and separately in an MSKCC low- to high-risk cohort for
validation. For risk of metastasis, both cohorts, including low- and high-
risk patients from the Canadian cohort, were pooled to enrich for events.
Biochemical relapse was defined using the Phoenix criteria for
radiotherapy patients; two consecutive PSA values of >0.2 ng/ml or
salvage treatment with radiotherapy and/or androgen deprivation for
prostatectomy patients . Visceral, skeletal, or non-regional nodal
relapses that were clinically detected or biopsy proven were recorded as
distant metastases. Biochemical relapse-free rate (bRFR) and metastasis-
free rate (mFR) were measured from the time of starting treatment to
event. The bRFR and mFR were estimated using the Kaplan–Meier
method and compared using the log-rank test stratified by the cohorts.
Associations between clinical covariates (age, cT category, Gleason score,
ISUP grade, and PSA), IDC/CA, and PGA with biochemical relapse or
metastasis were assessed using Cox proportional hazards ratio (HR),
stratified for the cohort as defined by site and treatment given.
Multivariable Cox models, including only covariates with univariable
p < 0.05, were used to estimate adjusted HRs of IDC/CA in the presence of
clinical covariates and PGA. Harrell’s C-index was generated for
discrimination of the different multivariable models (clinical, clinica-
l + IDC/CA, and clinical + IDC/CA + PGA). Age, PSA, and PGA were
considered as continuous variables throughout all analyses. All p values
are two-sided, and significance was defined at a = 0.05.
3. Results
IDC/CA+ prostate cancers were associated with increased
biochemical and metastatic relapses in the Canadian
(167 biochemical and 30 metastatic relapses in
674 NCCN-defined low- to high-risk prostatectomy and
radiotherapy cases; see Supplementary Tables 1 and 2 for
clinical characteristics) and MSKCC cohorts[18_TD$DIFF] (71 biochemical
and 22 metastatic relapses in 258 NCCN-defined low- to
high-risk prostatectomy cases; Supplementary Table 1)
. IDC/CA+ prostate cancers were associated with more
advanced clinicopathological indices, including higher cT
category, ISUP grade, and overall NCCN risk category
(p < 0.001 for all; Supplementary Table 3). IDC/CA was
predictive of an increased risk of biochemical relapse
(multivariable analyses of clinical + IDC/CA: HRCanadian
2.17 [95% confidence interval {CI} = 1.53–3.07], p < 0.001;
HRMSKCC 2.32 [95% CI = 1.32–4.08], p = 0.0035; A and B,
, and Supplementary Table 4[19_TD$DIFF] [prostatectomy
cohort]). IDC/CA was also independently predictive of
metastatic relapse (multivariable analysis of clinica-
l + IDC/CA: HRpooled 3.31 [95% CI = 1.76–6.21], p < 0.001),
including ISUP grade , , and Supplementary
Fig. 2). Therefore, in two separate cohorts, the presence of
IDC/CA subpathology predicted for aggressive risk features
that manifested in treatment relapse and metastases.
Next, we investigated if IDC/CA+ prostate cancers were
associated with genomic instability and tumor hypoxia
). We observed that IDC/CA+ prostate cancers had
increased hypoxic tumor subpopulations compared with
IDC/CA– prostate cancers (64.0% vs 45.5%, p = 0.17; Sup-
plementary Fig. 3). In a subset of 476 (242 Canadian and
234 MSKCC) cases, IDC/CA+ tumors were significantly
associated with higher PGA (median of 7.2 vs 3.0,
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 6 6 5 – 6 7 4 667