detection limit to 0.2% (Figure 3). The results showed that we
were able to detect a true signal for blood samples spiked
with 5 PANC1 cells (ie less than 1 KRAS mutant copy per mL
blood). Moreover, we obtained a linear abundance ratio with
samples spiked with a higher number of PANC1 cells. In
contrast, Sanger sequencing showed a sensitivity close to
100 cells per 6 mL (ie about 16 copies per mL blood)
(Figure 4A) while TaqMelt and HRM had a sensitivity close to
75 and 50 cells per 6 mL, respectively (ie about 12 copies or 8
copies per mL blood) (Figure 4B and C). A summary of these
findings are presented in Table 1.
3.2. ddPCR can detect KRAS mutations in CTCs from
CRC patients
Next, we wished to evaluate the detection of mutant KRAS in
CTCs obtained from CRC patients. For this, we enrolled 35
CRC patients at any stage of the disease who were admitted
to the digestive surgery department of the Pitie-Salp etri ^ ere
University Hospital. The characteristics of this cohort are indicated
in Table 2. First, we performed CTC analysis of blood
samples obtained just before curative surgery. This analysis
showed that 90% (26/29) of patients harbored at least one
CTC per 3 mL of total blood and among then 7% (2/29) of patients
had also cell clusters named circulating tumor microemboli
or multicellular CTC clusters (Table 3 and Supp. Data
2). In contrast, the analysis of blood from ten healthy donors
revealed no CTCs. We observed a trend of increasing numbers
of CTCs according to disease stage ranging from in situ to metastatic
disease. In contrast, no correlation was observed between
the number of CTCs and serum concentrations of the
tumor markers CEA (carcinoma embryonic antigen) and
CA19.9 (Supp. data 2). Next, we performed the multiplex
KRAS genotype assay by ddPCR. 86% (30/35) was performed
successfully, since 5 samples could not be amplified by whole