AURA2, NCT02094261). The positive percentage agreement and
negative percentage agreement between tissue and plasma testings
were 61.4% and 78.6%, respectively [60]. It is not dissimilar to the
results obtained for detection of exon 19 deletions and L585R
mutations in ctDNA using the same test. Notably, liquid biopsy
analysis of EGFR mutations (as well as KRAS mutation) at diagno-
sis or relapse was prospectively validated in 180 patients with
advanced-stage NSCLC in a study conducted at the Dana-Farber
Cancer Institute (Boston, Massachusetts, USA). Mutations were
detected using a digital PCR platform with high specificity (>79%)
and with a median turnaround time of only 3 days, compared with
12 days and 27 days for tissue genotyping of newly diagnosed and
relapsed tumors, respectively. In addition to NSCLC, the eligibility
of targeted therapy has extended to many cancer types, including
melanoma and colorectal and breast cancers. Currently, most deci-
sions are made based on biopsy results from tumor tissue speci-
mens. But, such assessments could potentially be replaced with
liquid biopsy approaches in the future. This replacement might be
particularly important when surgery is not indicated and tissue
biopsy samples are difficult to obtain or when genetic heterogeneity
could complicate decision-making based on limited tissue speci-
mens. However, extensive researches are necessary to prove the
utility and reliability of ctDNA analysis in such diverse settings.
Of note, the preliminary results from a large validation trial of
Guardant360 were presented at the 2016 ASCO Annual Meeting
[61]. The trial was a study of a digital DNA sequencing technology
encompassing a broad panel of 70 genes, including all current
clinically actionable genes with approved targeted drugs. The
study included 15,191 patients with advanced-stage lung, breast,
colorectal, or other cancers [62]. Accuracy was assessed indirectly
by comparing the frequencies of specific DNA changes in ctDNA
with those of tissue-based DNA changes from patients included in
The Cancer Genome Atlas (TCGA) and directly by matched plasma
and tissue samples from a subset of almost 400 patients with
NSCLC or colorectal cancer. Correlations between the TCGA
and ctDNA data ranged from 92 to 99% across multiple cancer
genes and different classes of gene alterations. However, the EGFR
T790M resistance mutation was only detected in plasma DNA from
patients treated with EGFR TKIs, which was not detectable in
pretreatment samples. The matched plasma-tissue comparison
showed a PPV of 87%. Importantly, the PPV increased to 98%
when the plasma and tissue samples were collected<6 months
apart. The Guardant360 test is quite sensitive, since mutated
ctDNA accounting for 0.4% of the total cfDNA in the blood
could be detected. The Guardant360 test enabled the detection
of resistance mutations in EGFR, ALK, and KRAS that were not
detectable in the matched tumor biopsy samples in almostThe Introduction and Clinical Application of Cell-Free Tumor DNA 57