EURURO Vol. 72 No. 4

562 656

associated with increased responsiveness to checkpoint
inhibitors in melanoma and nonsmall cell lung cancer
. In RCC, nivolumab is approved for use in post-
VEGF-targeted therapy, and our finding of increased GA
frequency in this setting (a surrogate for mutational
burden) may explain the observed activity. Front-line
studies assessing nivolumab monotherapy in mRCC have
shown diminished response rates .
Several limitations of the current study must be noted.
First, treatment-related data was not collected across all
patients whose samples were submitted to a clinical
laboratory. In the subset of patients in whom this data was
collected, receipt of first-line or later therapy was inferred
based on the nature of agents received. Although most
patients receive agents other that sunitinib, pazopanib, and
temsirolimus in the first-line setting, there is a small cohort
of patients (< 10% based on current market data) that may
receive other treatments. We acknowledge as well that
patients who are characterized as receiving postfirst-line,
or secondary therapies, may have received these treat-
ments in a variety of settings (eg, agents such as nivolumab
or cabozantinib may be rendered as third-, fourth-, or fifth-
line treatment). We further acknowledge that ctDNA
collection could have occurred at varying time points
during the course of treatment (eg, at the time of initiation
of therapy, while on treatment, or closer to the time of
progression). To be clear, the contrast between first-line
and postfirst-line therapies does not reflect a comparison
of individual patients who have made the transition across
this therapeutic continuum. Nonetheless, we feel that
these data offer hypothesis-generating evidence of how
tumor genomics may be altered from early to late
metastatic disease. Another limitation of the current study
is the lack of histologic characterization in many patients.
One might propose that the low cumulative frequency of
RCC-specific GAs (eg, VHL) might be secondary to abundant
nonclear cell histologies or potential misclassification, or
perhaps challenges in exon 1 coverage of the VHL gene.
Many patients in the series had no specified histology, and
therefore, one cannot exclude the possibility that these
were non-RCC histologies (eg, lymphoma, neuroendocrine
tumors, etc.). However, other series including genomic
assessment of metastatic RCC specimens have reflected a
relatively lower frequency of VHL alteration, as shown in
Supplementary Finally, another major
limitation of the study is the lack of clinical outcome data
associated with the individual patients included herein.
This annotation may ultimately help ascertain the predic-
tive role of alterations detected in ctDNA. Of course, one
must be weary of response and progression-free survival
data collected in a retrospective fashion, as this is always
subject to investigator bias.
In the current series, tissue-based genomic profiling
data was not available. There are limited published
reports that compare the current ctDNA platform
(Guardant360, Redwood City, CA, USA) to tissue-based
platforms. In one series including nine patients (none
with genitourinary tumors), only 10 of 45 alterations
(22%) seen across the entire cohort were found in both
platforms These small hypothesis-generating stud-
ies need to be confirmed in larger series. One may also ask
whether the results we provide are reproducible on an
intrapatient level. In total, 13 patients had more than one
assessment of their ctDNA with second samples obtained
21–360 (median: 157) d after the first sample. We
assessed the positive predictive agreement between time
one and time two samples stratified by time intervals
(Supplementary . In the patient with samples taken
within 1 mo, eight of the nine (89%) time 1 sample
alterations were seen in the time 2 sample, with 80%
overall positive agreement (1 new alteration detected at
time 2). For patients with the second ctDNA samples
drawn 2–3 mo after the first and greater than 3 mo after
the first draw, the overall positive agreement was 41%
and 28%, respectively. Although one might conclude the
lack of complete concordance reflects the precision or
accuracy of the test, inaccurate test performance would
result in consistent lack of interpatient concordance
independent of intervening time. The decreasing concor-
dance with increasing time is indicative of clonal
evolution under treatment pressures . Finally, it
is important to note that the test appears to have high
specificity for cancer cases. In a series of 222 healthy
volunteers assessed with Guardant360, only one patient
was found to have an alteration (TP53 mutation)
5. Conclusions
We provide the largest experience to date assessing ctDNA
in patients withmRCC. GAs were detected in themajority of
patients, with frequent alterations in TP53, VHL, and EGFR
and NF1. At present, ctDNA provides a potential mechanism
to identify unique targets. The landscape of alterations
presented herein delineates the presence of unique
alterations (eg, HER2, EGFR, and BRCA1 mutations) that
could prompt consideration of nonconventional therapies
for mRCC, and possibly facilitate enrollment in prospective
basket trials which enlist patients based on mutational
status. The ongoing American Society of Clinical Oncology
Targeted Agent and Profiling Utilization Registry trial is one
such study that permits enrollment of patients bearing
alterations in ctDNA, and aligns patients with a wide panel
of associated therapies (NCT02693535). The cumulative
assessment of multiple patients using ctDNA, as done
herein, offers a unique opportunity to characterize
differences in GAs that may occur across varying disease
states (eg, line of treatment, histology, etc.). Variations seen
across first-line and postfirst-line therapy suggest poten-
tial mechanisms of therapeutic resistance (eg, TP53
mutation) and also provide biological rationale for the
activity seen with currently approved agents for refractory
disease (eg, everolimus in the context of emerging PIK3CA
and NF1 alterations). The noninvasive nature of ctDNA
testing makes it an attractive method of obtaining real-
time genomic data as compared to serial biopsies of
metastatic sites. Through embedding these assays in
prospective clinical trials in mRCC, ctDNA may yield
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 5 5 7 – 5 6 4 562