EURURO Vol. 71 No. 4

samples were then thawed and centrifuged at

[5_TD$DIFF]

1900

for 15 min to remove cellular debris. Exosome isolation

from plasma was performed using the exoRNeasy kit

(Qiagen, Valencia, CA, USA) on exoEasy spin columns, and

RNA was extracted from the vesicles bound to the silica

membrane using the QIAzol phenol/guanidine-based lysis

solution. Chloroform was then added to QIAzol-samples

and centrifuged at 12 000

for 15 min at 4

C. The

aqueous phase containing RNA was recovered and applied

to the RNeasy MinElute spin column, in which the

RNA binds to the membrane and contaminants are

discarded. The RNA was finally eluted in 20

l of the

elution buffer.

2.4.

Analysis of

[6_TD$DIFF]

AR-V7 on plasma-derived exosomal RNA

The investigational part of this study included the assess-

ment of AR-V7 in CRPC patients. Other mutations of the

androgen-related pathway were not examined because of

the limited amount of RNA available and lack of strong

scientific evidence concerning their role in resistance. The

analysis of AR-V7 in RNA was performed by ddPCR using the

One-Step RT-ddPCR kit, as described earlier. VCaP RNA was

used to spike clinical samples and blank specimens as the

control. Results were reported as copies of mutant allele per

milliliter of plasma.

2.5.

Data analysis

The ddPCR QuantaSoft software determined the absolute

target concentration as copies/ml in the samples. Statisti-

cal analyses were performed separately in the AR-V7-

positive (AR-V7

[5_TD$DIFF]

) and AR-V7-negative (AR-V7 ) cohorts.

PSA response rates (RRs) were compared with the Fisher

exact test. Time-to-event outcomes (ie, clinical or radio-

graphic progression-free survival [PFS] and overall surviv-

al [OS]) were evaluated by the Kaplan-Meier method, and

survival-time differences were compared by the log-rank

test. Additional statistics were performed by paired

test.

All tests were two sided, and

values 0.05 were

considered statistically significant. The performance of

the test was assessed by the performance of a binary

classifier system because its discrimination threshold is

varied. The receiver operating characteristic curve (ROC

curve) was created by plotting the true-positive rate

against the false-positive rate at various threshold

settings. SPSS software v2.0 (IBM Corp., Armonk, NY,

USA) was used for calculations.

Results

Using ddPCR we were able to detect the AR-V7 transcript,

even at low concentrations (ie, 2 copies/ml) (Supplementa-

ry Fig. 1). No AR-V7 signal was detected in the unspiked

control plasma or in blank wells, confirming the specificity

of the assay. Supplementary

Table 1

reports the median

number of copies spiked per unit volume of plasma and

measured by ddPCR, 95% confidence intervals (CIs), and

coefficients of variation.

3.1.

Patient characteristics and AR-V7 expression

Overall, 36 patients were enrolled, of whom 26 received

abiraterone and 10 enzalutamide; median follow-up time

was 9 mo (range: 2.0–31.0). Baseline characteristics for the

study population are reported in

Table 1 .

A total of 14 of

36 patients (38.8%) were AR-V7

(100–2400 copies)

( Fig. 1

)

before the start of HT. Seven of 14 AR-V7

patients had a

plasma sample at baseline (median: 500 copies/ml) and at

progression (median: 887 copies/ml); although there was an

increase, the difference was not statistically significant

(paired

test,

= 0.25)

( Table 2 )

. The AR-FL was

[7_TD$DIFF]

used as an

internal control to validate the extraction process; AR-V7

samples were true negative because the AR-FL signal was

detectable. The starting volume of plasma samples (1 vs 2 ml)

[(Fig._2)TD$FIG]

Fig. 2 – (a) Progression-free survival of androgen receptor splice variant

7–positive (AR-V7

) versus AR-V7-negative (AR-V7

) patients; (b) overall

survival of AR-V7

versus AR-V7

patients.

AR-V7 = androgen receptor splice variant 7.

E U R O P E A N U R O L O G Y 7 1 ( 2 0 1 7 ) 6 8 0 – 6 8 7

684