ONCOLOGY LETTERS 21: 420, 2021

The feasibility of circulating tumor DNA analysis as a

marker of recurrence in triple-negative breast cancer
SATOSHI OKAZAKI1,2, TAKAAKI SASAKI1, SHUNSUKE YASUDA1,2, MASAHIRO ABE1,2, NANA YOSHIDA1,2,
RYOHEI YOSHIDA1, KEI ISHIBASHI1,2, YOSHINORI MINAMI1, SHUNSUKE OKUMURA1, SHINICHI CHIBA3,
HIDEHIRO TAKEI4, RYUSUKE HAYASHI5,6, TOSHIHIRO NAGATO5,6, HIROYA KOBAYASHI6,
AYUMU SUGITANI7, YUSUKE ONO7, YUSUKE MIZUKAMI7,8,
MASAHIRO KITADA1,2 and YOSHINOBU OHSAKI1

1
Respiratory Center, 2Breast Center, 3Center for Advanced Research and Education, Asahikawa Medical University;
4
Department of Surgical Pathology, Asahikawa Medical University Hospital;
5
Department of Otolaryngology-Head and Neck Surgery; 6Department of Pathology, Asahikawa Medical University,
Asahikawa, Hokkaido 078-8510; 7Institute of Biomedical Research, Sapporo Higashi Tokushukai Hospital,
Sapporo, Hokkaido 065-0033; 8Division of Gastroenterology and Hematology/Oncology,
Department of Medicine, Asahikawa Medical University, Asahikawa, Hokkaido 078-8510, Japan

Received March 26, 2020; Accepted September 16, 2020

DOI: 10.3892/ol.2021.12681

Abstract. Triple‑negative breast cancer (TNBC) has a poorer revealed to be activated in patients harboring PIK3CA H1047R
prognosis than other breast cancer subtypes; therefore, iden‑ and AKT1 E17K mutations; therefore, the PI3K/AKT pathway
tifying markers of early recurrence is important. The present may be a promising candidate for targeted therapy in these
study aimed to establish a liquid biopsy protocol for droplet patients.
digital PCR‑based detection of frequently mutated genes
in patients with TNBC. Tumor DNA from 36 patients with Introduction
TNBC who relapsed within 2 years after surgical resection
was retrospectively analyzed. Somatic mutational profiles were Triple‑negative breast cancer (TNBC) accounts for ~15% of
evaluated using targeted sequencing to identify frequently all breast cancer cases and is defined by a lack of estrogen and
mutated genes and genes associated with molecularly targeted progesterone receptors' expression and the absence of human
therapies. The association between genetic alterations and epidermal growth factor receptor‑2 (1). Molecular studies have
associated protein phosphorylation was investigated using demonstrated that TNBCs are a heterogeneous group of tumors
immunohistochemical analysis. Recurrent hot spot mutations with different clinical features, prognoses, genetic‑molecular
in the plasma were monitored over time. Mutation‑specific alterations, and responses to treatment (2). TNBCs are more
probes were used to successfully detect mutations in the blood likely to recur earlier after surgery than other subtypes, and
samples of patients who were positive for PIK3CA H1047R and many patients have poor prognosis (3). To increase success
AKT1 E17K mutations. Somatic mutations in AKT1 (14.9%) rate during TNBC treatment, personalized medicine designed
and PIK3CA (25.5%) were frequently identified in the data. for each patient's tumor profile individually taking genetic
Robust phosphorylation of AKT and S6RP was more common alterations in each patient into account is necessary.
in tumors with PIK3CA H1047R and AKT1 E17K mutational Currently, for recurrence detection, serum tumor marker
background than in tumors with wild‑type PIK3CA and AKT1. evaluation or computed tomography (CT) are used; however,
In conclusion, the present study evaluated a high‑sensitivity more sensitive strategies should be developed to improve detec‑
detection system for frequently mutated genes that was also tion reliability. To identify relevant mutations, more accurate
applicable for cell‑free DNA. The PI3K/AKT pathway was and cost‑effective tools, such as strategies that use cell‑free
DNA (cfDNA), are required in clinical settings. However, the
analysis of cfDNA using next‑generation sequencing (NGS) is
expensive and time‑consuming. Among new technologies for
cfDNA quantification, droplet digital PCR (ddPCR) provides
Correspondence to: Dr Takaaki Sasaki, Respiratory Center,
Asahikawa Medical University, 2-1 Midorigaoka Higashi, Asahikawa,
the highest sensitivity for detecting and tracking actionable
Hokkaido 078-8510, Japan mutations at frequencies as low as 0.01‑0.05% (4). In hormone
E-mail: takaaki6@asahikawa-med.ac.jp receptor‑positive breast cancer, the utility of liquid biopsy
to analyze hormone therapy‑resistant mutations has been
Key words: triple-negative breast cancer, early recurrence, liquid studied (5,6). In case of TNBC, liquid biopsy for detecting
biopsy, genetic testing, digital PCR minimal residual disease during neoadjuvant chemotherapy
was also studied and reported by some groups. These studies
2 OKAZAKI et al: CIRCULATING TUMOR DNA ANALYSIS AS A MARKER OF RECURRENCE IN TNBC

have demonstrated that circulating tumor DNA (ctDNA) level should be screening for cfDNA targeted amplicon sequencing
during neoadjuvant chemotherapy was associated with patient was employed.
survival (7,8). However, there is still no valid protocol that can Between 20 and 100 nanograms of cfDNA were used for
be used for diagnosis of disease recurrence, which is focusing library construction with the Ion AmpliSeq Cancer Panel v2,
on specific mutations. which targets thousands of mutational hotspot regions in
Here, we explore the practical utility of liquid biopsy 50 cancer‑associated genes using the Ion AmpliSeq Library
using ddPCR‑based detection of frequently mutated genes Kit (Thermo Fisher Scientific, Inc.) (10). The sequencing was
in samples of TNBC patients. We also describe the clinical performed on an Ion PGM System according to the manu‑
course of metastatic TNBC patients whose genome harbors facturer's protocol, and sequencing leads were multiplexed,
AKT1 E17K and PIK3CA H1047R mutations, which have quality‑filtered, and aligned to the human reference genome
been reported to be oncogenic driver mutations in breast (GRCh37) using the Torrent Suite software (ver. 5.0.4; Thermo
cancer that lead to PI3K/AKT pathway activation and tumor Fisher Scientific, Inc.). Variants were identified with the
progression. Variant Caller software (ver. 5.0.4.0; Thermo Fisher Scientific,
Inc.) and filtered using Ion Reporter Software (Thermo Fisher
Materials and methods Scientific, Inc.). The quality of all variants calls was manu‑
ally confirmed by IGV software (ver. 2.3.59; Broad Institute,
Patients and breast cancer samples. We examined 57 Cambridge, MA, USA).
consecutive TNBC patients who underwent mastectomy or
breast‑conserving surgery at the Asahikawa Medical University Mutation detection by ddPCR. For the detection of the AKT1
Hospital between April 2000 and March 2017. Thirty‑six c.49G>A mutation, a specific primer/probe set was utilized
patients who relapsed within 2 years postoperatively were (dHsaCP2000031; Bio‑Rad Laboratories). For PIK3CA and
enrolled in the study. Tissue resected metastatic specimens of TP53 mutations, specific assays were designed in‑house and
these patients were also evaluated. All cases were reviewed synthesized by Integrated DNA Technologies (IDT); nucleo‑
by two experienced pathologists and categorized according to tide sequences of primers and probes are shown in Table SI).
the WHO classification (9). After surgical resection, patients ddPCR assays were performed as previously described (11).
underwent follow‑up examination every 3 months for 2 years Reaction mixtures were made of ddPCR Supermix for probes
and every 6 months thereafter. The diagnosis of recurrence was (no dUTP; Bio‑Rad), primer, wild‑type‑ and mutant‑specific
made based on abnormal findings in the conserved mammary probes, and template DNA in a total volume of 22 µl. Droplets
glands or by imaging diagnosis during the follow‑up period. were generated by mixing the reaction mixture with Droplet
When patients had tumor recurrence, they were treated with Generation Oil (Bio‑Rad Laboratories) in a QX200 droplet
chemotherapy (e.g., eribulin, tegafur‑uracil, bevacizumab, generator (Bio‑Rad Laboratories). PCR was performed on a
paclitaxel, or vinorelbine). Veriti Thermal Cycler (Thermo Fisher Scientific, Inc.) using the
This study protocol, including tumor sample collection following cycling conditions: 10 min at 95˚C, 40 cycles of 94˚C
and genetic analysis, was approved by the Asahikawa Medical for 30 sec followed by 60˚C for TP53 or 57˚C for PIK3CA for
University Research Ethics Committee (approval nos. 17043 one minute, followed by 98˚C for 10 min. Samples were trans‑
and 18118). Written informed consent was obtained from all ferred to the QX200 Droplet Reader (Bio‑Rad Laboratories)
the patients before their enrollment for genetic analysis. to measure the fluorescence of 6‑fluorescein amidite and
hexachloro‑fluorescein probes. Droplets were scored as posi‑
Sample collection and processing. Genomic DNA was tive or negative based on their fluorescence intensity, which
extracted from formalin‑fixed paraffin‑embedded (FFPE) was determined by gating thresholds defined with positive and
primary and recurrent tumor samples. Two cases were negative controls. Finally, absolute copy number input in the
excluded because the specimens contained too few tumor reaction and the ratio of mutated fragments were calculated by
cells. The tumor cell content ranged widely (from 20 to 90%), QuantaSoft ver 1.7 (Bio‑Rad Laboratories) based on a Poisson
with 44 (94%) samples containing >60% tumor cells. Viable distribution. Samples were scored as positive for the mutation
tumor areas were dissected with a razor blade, and DNA was when at least three mutation‑containing droplets were detected
extracted using the GeneRead DNA FFPE Kit (Qiagen). by the mutant‑specific probe as reported previously (12).
Plasma samples were collected from AKT1 E17K
(c.49G>A)‑ and PIK3CA H1047R (c.3140A>G)‑positive Immunohistochemical analysis. Immunohistochemistry (IHC)
patients using PAXgene Blood ccfDNA Tube (Qiagen). Plasma was performed using the Envision™ HRP System (#K5361;
was then isolated by centrifuging the samples at 1,400 x g Dako) using formalin‑fixed sections, as described previ‑
for 10 min at room temperature. The cell‑free plasma was ously (13). Two samples with a low tumor cellularity were
transferred into 2‑ml collection tubes and stored at ‑80˚C until excluded. Heat‑induced antigen retrieval was performed for
purification. For cfDNA purification, QIAamp Circulating phospho‑AKT (pAKT; #4060) and phospho‑S6RP (pS6RP;
Nucleic Acid kit (Qiagen) was utilized. The genomic DNA #4858; Cell Signaling Technology, Inc.) antibodies in EDTA
concentration was measured using the Qubit dsDNA HS buffer at pH 9.0, and endogenous peroxidase activity was
Assay kit (Thermo Fisher Scientific, Inc.) and the Qubit 3.0 inhibited according to the manufacturer's instructions. The
fluorometer (Thermo Fisher Scientific, Inc.). slides were then incubated at 4˚C overnight with anti‑pAKT
at 1/25 and anti‑pS6RP at 1/100 [diluted in Dako Real
Targeted amplicon sequencing. To identify frequently mutated Antibody Diluent (Dako)], followed by incubation with an
in our TNBCs and to determine which genetic alteration HRP‑conjugated secondary antibody and substrate.
 ONCOLOGY LETTERS 21: 420, 2021 3

Table I. Demographic and clinicopathological features of

patients.

Characteristics N %

Total 36
Median age at surgery, years (range) 56.6 (30-81)
Menopausal status
Premenopausal 13 36.1
Postmenopausal 23 63.9
T stage
T1 (≤2.0 cm) 11 30.6
T2 (2.1-5.0 cm) 18 50
T3 (>5.0 cm) 5 13.9
T4 (direct extension to the chest 2 5.5 Figure 1. OS in TNBC patients who relapsed within 2 years after surgery. OS,
wall or skin) overall survival; TNBC, triple‑negative breast cancer.
No. of positive lymph nodes
0 16 44.4
Statistical analyses. Patients who were still alive at the most
1-3 10 27.8
recent follow‑up were censored for survival on the date of that
4-9 7 19.5
follow‑up. The Kaplan‑Meier method was used to estimate
≥10 2 5.5 overall survival (OS) rates. Fisher's exact test was used to
Unknown 1 2.8 compare our data on the frequency of gene mutations with that
Stage in the METABRIC database, which is the largest global study
I 7 19.5 of breast cancer tissue samples ever performed (14). Pearson's
II 15 41.6 correlation coefficient was calculated to examine the relevance
III 10 27.8 of allele frequency obtained by targeted amplicon sequencing
IV 4 11.1 and ddPCR. The Mann‑Whitney U test was used to assess
IHC score differences between PIK3CA H1047R‑ and AKT1
Histological grade
E17K‑mutant samples and wild‑type samples. A two‑sided
1 2 5.5 P‑value of <0.05 was considered statistically significant. All
2 7 19.5 statistical analyses were two‑sided and performed using R soft‑
3 23 63.9 ware (version 3.4.1; The R Foundation, Vienna, Austria) (15).
Unknown 4 11.1
Histological type Results
Invasive ductal 32 89
Invasive lobular 2 5.5 Characteristics of early recurrent TNBCs. The clinico‑
Others 2 5.5 pathological characteristics of the 36 enrolled patients are
summarized in Table I. The 5‑year OS rate was 19% (Fig. 1).
Ki67 LI The median age at surgery was 56.6 years. Over half (63.9%)
≤14 2 5.5 of the patients were postmenopausal, and 89.0% had invasive
>14 22 61.2 ductal carcinoma histology. Four patients (11.1%) were initially
Unknown 12 33.3 diagnosed with stage IV cancer. Neoadjuvant and adjuvant
Chemotherapy received treatments were administered to 35 of 36 patients (97.2%).
Yes 35 97.2 Neoadjuvant chemotherapy was performed in 11 patients.
No 1 2.8 Oral chemotherapy (tegafur‑uracil) was administered to four
patients. One patient did not receive chemotherapy because
Median PFS, months (range) 12.6 (0-24)
of liver cancer treatment. The median time to progression
Median OS, months (range) 31.9 (1-175) and median OS from surgery were 12.6 months (range:
0‑24 months) and 31.9 months (range: 1‑175 months), respec‑
LI, labeling index; PFS, progression-free survival; OS, overall survival.
tively.

Genetic alterations in early recurrent TNBCs. Of the 36 cases,

the tumor content was low in two cases, so they were excluded
Staining for pAKT and pS6RP was evaluated by consensus from DNA analysis. There were 13 patients who had undergone
by two observers blinded to the clinical data. For scoring, the resection of the recurrence, each with eight lymph nodes, two
highest staining intensity in malignant cells was classified lungs, two brains, and one pleura. Of 1,835 variants obtained
as 0, no staining; 1, weak staining; 2, moderate staining; or from 47 samples, 72 variants were selected by coverage, type
3, strong staining (Fig. S1). of function, sequencing quality, filtering of single‑nucleotide
4 OKAZAKI et al: CIRCULATING TUMOR DNA ANALYSIS AS A MARKER OF RECURRENCE IN TNBC

Figure 2. Mutation frequencies and IHC score in 47 TNBC samples. DNA was extracted from 34 primary and 13 recurrent tumors and IHC was performed
on 47 similar samples. Mutation analysis of cancer‑related genes that were mutated in at least 15% of 47 TNBC samples show phospho‑AKT (upper) and
phospho‑S6RP (lower) IHC scores of the samples. IHC, immunohistochemistry; TNBC, triple‑negative breast cancer; RP, ribosomal protein.

polymorphisms (SNPs) using Variant Caller software, and most common mutation in our study (7/47, 14.9%) (Fig. 3A).
manual selection of false‑positives (Fig. S2) (16). Forty‑eight For PIK3CA gene (Fig. 3B), PIK3CA H1047R was the most
were single nucleotide variants (66.7%) and 24 were small frequent mutation (5/47, 10.6%). In contrast, for mutations in
insertion‑deletions (33.3%). Of these variants, 54 and seven TP53 gene, there were various mutation sites, but no charac‑
were unique to the primary and metastatic lesions, respectively, teristic mutations were found (Fig. 3C).
and 11 were found in both. Thirty‑three of 34 tumors (97.1%)
obtained from primary tumors had the following muta‑ Patterns of expression of phospho‑AKT and phospho‑S6RP.
tions (average 5.2, range 1‑12 mutations): TP53 (27, 79.4%); AKT1 and PIK3CA were reported as driver oncogenes whose
NOTCH1 (16, 47.1%); PTEN (13, 38.2%); PIK3CA (11, 32.3%); activity was sufficient to transform a normal cell into a
CDKN2A (11, 32.3%); APC (10, 29.4%); VHL (8, 23.5%); and cancerous one. PI3K/AKT signaling pathway activity can be
AKT1 (7, 20.6%) (Fig. 2). All 13 recurrent samples harbored detected by the phosphorylation of AKT and S6RP proteins.
mutations (average 2.5, range: 1‑5 mutations), and mutation Therefore, we investigated the correlation between genetic
in KDR was the most frequent (8, 61.5%). The comparison of alterations and protein phosphorylation in tumor samples by
genetic mutations in 13 metastatic tumors and corresponding IHC (Fig. 2). Representative images of tumor cells with strong,
primary tumors were shown in Fig. S3. The left side column moderate, weak, and no staining for pAKT1 and pS6RP are
shows the data of the primary tumor and the right column shown in Fig. S1. Tumors with PIK3CA H1047R or AKT1 E17K
shows the data of the metastatic tumor. AKT1 mutations were mutations had higher expression level of phospho‑AKT (pAKT)
not found in these samples. For PIK3CA mutation, samples and phospho‑S6RP (pS6RP) than PIK3CA and AKT1 wild‑type
from 4 patients were detected including from 3 primary tumors (P=0.044 and P=0.105, respectively; Table SII),
tumors. The sites and frequencies of mutations in TP53, suggesting that genetic alteration of AKT1 or PIK3CA leads to
AKT1, and PIK3CA are shown in Fig. 3. AKT1 E17K was the induction of abnormal activation of survival pathways in cancer.
 ONCOLOGY LETTERS 21: 420, 2021 5

Figure 3. Sites distribution of (A) AKT, (B) PIK3CA and (C) TP53 mutation. The table at the bottom shows a comparison between mutations detected in our
data (upper line) and in the METABRIC cohort (lower line). P‑values were calculated using Fisher’s exact test.

Figure 4. Specific mutation frequencies of TNBC as detected by NGS and ddPCR. We designed or purchased allele‑specific probes for ddPCR. We then used
these probes to validate the mutation (AKT1 E17K, PIK3CA H1047R, and TP53 R175H and R248Q) allele frequencies in NGS and ddPCR data of the DNA
derived from 47 specimens of early recurrent TNBC. P‑values and R were calculated using Pearson’s correlation coefficient. TNBC, triple‑negative breast
cancer; NGS, next‑generation sequencing; ddPCR, droplet digital polymerase chain reaction.
6 OKAZAKI et al: CIRCULATING TUMOR DNA ANALYSIS AS A MARKER OF RECURRENCE IN TNBC

Figure 5. Clinical course of case reports. (A) A 58‑year‑old woman with a diagnosis of TNBC. The graph shows the change in the percentage of PIK3CA
H1047R mutations detected in cfDNA using ddPCR and the CA15‑3 levels during the clinical course after surgery. CT shows multiple liver metastases
13 months after surgery and also the progression of liver metastases 25 months after surgery. CT shows further progression of the liver metastases 41 months
after surgery. Arrowhead indicates liver metastasis. (B) A 77‑year‑old woman with Stage IIIB TNBC with direct extension to the skin and lymph node metas‑
tasis. The graph shows the change in the percent E17K mutation detected in cfDNA using ddPCR and the CEA levels during the clinical course after surgery.
CT shows lymph node progression 11 months after surgery and 48 months after surgery. CT shows new left cervical lymph node metastasis. Arrowhead
indicates lymph node metastasis. cfDNA, cell-free DNA; ddPCR, droplet digital polymerase chain reaction; CT, computed tomography; CA15-3, carbohydrate
antigen 15-3; CEA, carcinoembryonic antigen.

Mutation allele frequencies of FFPE samples. We sought survived. Among them, consent was obtained in three cases,
to develop a highly sensitive blood‑based method for detec‑ and plasma was collected and analyzed. Patients nos. 1422 and
tion of genetic alterations that can be used as a clinical 2443 were harboring PIK3CA H1047R mutation and patient
application for serial monitoring. We designed or purchased no. 2152 was harboring AKT1 E17K mutation in primary
allele‑specific probes for ddPCR that yielded 0.01% detection tumor, however in patient no. 1422, PIK3CA H1047R mutation
sensitivity between mutant and wild‑type alleles (Table SI; was not detected in lymph node metastasis site nor cfDNA due
Fig. S4; Appendix S1). Genomic DNA extracted from to the adjuvant chemotherapy after surgery (Fig. S5A).
HCT116, SKBR‑3 or HCC70 cells was used as a ddPCR In other TNBC cohort including 3 cases with early
mutation control for PIK3CA H1047R, TP53 R175H or TP53 recurrence within 2 years after surgery, 2 cases of late recur‑
R248Q, respectively, and genomic DNA extracted from rence, 4 cases without metastasis were analyzed for cfDNA.
HCC38 cells was used as a wild‑type control for PIK3CA or Within this cohort, 2 patients (patient nos. 2667 and 3019)
TP53 (Table SIII). were harboring PIK3CA H1047R (Fig. S5B and C) or a
Then the frequencies of mutations identified by ddPCR patient (patient no. 996) was harboring AKT1 E17K muta‑
were compared to the ones previously determined by NGS tion (Fig. S5D) and all corresponding cfDNA from plasma was
using DNA derived from the FFPE tissue samples containing detected in the same mutations as primary lesion.
47 specimens of early TNBC recurrence. As shown in Fig. 4, Collectively, 6 patients with metastatic TNBC patients
AKT1 E17K, PIK3CA H1047R, and TP53 R175H and R248Q harboring PIK3CA H1047R or AKT1 E17K mutation,
mutations detected by NGS significantly correlated with 5 patients (83.3%) were detected corresponding mutation by
frequencies in ddPCR (r=0.93, P<0.001). digital PCR methods.

Results of digital PCR detection of clinical samples. In our Clinical course of a PIK3CA H1047R‑positive TNBC patient.
early recurrent TNBCs patients, 4 patients out of 36 patients still A 58‑year‑old woman (patient no. 2443) with a diagnosis of
 ONCOLOGY LETTERS 21: 420, 2021 7

TNBC (T3N0M0, Stage IIb) received neoadjuvant chemo‑ Therefore, circulating‑tumor DNA (ctDNA) in cell‑free
therapy with 5‑fluorouracil/epirubicin/cyclophosphamide plasma represents a very low fraction of the total amount of
followed by docetaxel. After neoadjuvant chemotherapy, circulating DNA, but there are several reports in which ctDNA
she underwent mastectomy and sentinel lymph node biopsy. in TNBC patients was analyzed. Chen et al (7) reported that
Histological examination of the resected specimens revealed ctDNA was detected in four out of the 33 early‑stage TNBC
pathological complete response. Needle biopsy revealed inva‑ patients after neoadjuvant chemotherapy by NGS analysis.
sive lobular carcinoma with high proliferative rate (MIB‑1 All four cases relapsed early after surgery, and disease‑free
index 28%). Thirteen months after the surgery, multiple survival (DFS) was significantly inferior compared to cases
liver metastases were found. Furthermore, 39 months after in which ctDNA could not be detected. On the other hand,
surgery, a liquid biopsy was performed (and the cfDNA was ctDNA could not be detected in nine of the recurrent cases,
extracted, concentration: 20.6 ng/ml of plasma). Genetic and the sensitivity remained at 31%. Riva et al (8) reported
mutation analysis of cfDNA indicated the presence of that in 27 out of 36 early TNBC patients before treatment, it
PIK3CA c.3140A>G mutation (H1047R), as did genetic anal‑ was possible to detect ctDNA, and in patients, for whom it was
ysis of the primary site, and the allele frequency was 0.4% as possible to detect ctDNA even after 1 cycle of neoadjuvant
defined by the liquid biopsy. One month later, progression chemotherapy, DFS and OS were significantly lower. These
of liver metastases was detected by CT (Fig. 5A). Liquid findings suggest that the detection and quantification of ctDNA
biopsies were performed every three months thereafter, and is a very promising tool for assessing a response to neoadjuvant
the increase of mutant allele frequency was correlated to chemotherapy. Our detection system using ddPCR can be used
increase of CA‑15‑3. As shown in Fig. S5B and C, mutations for liquid biopsy because of its high sensitivity. Importantly,
were detectable by liquid biopsy in two other patients, and using this protocol, we were able to understand the disease
one of them had brain metastasis while a tumor marker was progression during chemotherapy in patients with metastatic
negative. TNBC. In patient no. 2152 in particular, an increase in the
mutation copy ratio was observed reflecting on the disease
Clinical course of an AKT1 E17K‑positive TNBC patient. A state progression more sensitively than the tumor marker did.
77‑year‑old woman (patient no. 2152) underwent left breast In solid tumors other than breast cancer, biomarkers have been
resection and axillary lymph node dissection. She was diag‑ used as a tool to detect recurrence. These biomarkers include
nosed with Stage IIIB TNBC with direct extension to the skin prostate‑specific antigen for prostate cancer, carcinoembry‑
and lymph node metastasis. Examination of the biopsy sample onic antigen (CEA) for colon cancer, and cancer antigen 125
revealed invasive ductal carcinoma with high proliferative for ovarian cancer (19). On the other hand, in breast cancer or
rate (MIB‑1 index 38%). Eleven months after surgery, CT non‑small cell lung cancer, Clinical Practice Guidelines does
confirmed lymph node progression. Fifty-two months after not recommend the surveillance tumor marker testing for the
surgery, a liquid biopsy was performed and the concentra‑ patients treated with curative intent. It because biomarkers
tion of extracted cfDNA was 16.6 ng/ml of plasma. Seven measured postoperatively were not sensitive nor specific for
months later, a second liquid biopsy was performed and the cancer relapse (20,21).
concentration of extracted cfDNA was 18.8 ng/ml of plasma. In our previous study, the usefulness of cfDNA measure‑
Ten months later, a third liquid biopsy was performed and the ments rather than tumor markers in NSCLC patients was
concentration of cfDNA was 13.5 ng/ml of plasma. Genetic indicated (11). Ideally, if we could design digital PCR
mutation analysis of cfDNA indicated the presence of AKT1 probes/primers for all the detected genetic mutations by
c.49G>A (E17K) as in the primary site, and the mutation allele NGS, all patients could be eligible for detection from cfDNA.
frequency was 0.2% in the first biopsy and 0.4% in the second However, NGS was not performed for the purpose of screening
biopsy (Fig. 5B). In the second liquid biopsy, although tumor of functioning genetic mutations in TNBCs, but rather for
markers were undetectable, the �������������������������
AKT1���������������������
c.49G>A (E17K) muta‑ genetic mutations that should be searched in cfDNA. In the
tion could be detected. In the third liquid biopsy, the allele present study, we identified frequently mutated and associated
frequency rose to 2.4%, and one month later, metastasis of with molecularly targeted therapies using NGS based genetic
left the cervical lymph node was detected by CT. As shown alteration screening and we focused on AKT1 and PIK3CA
in Fig. S5D, mutation was detectable by liquid biopsy in mutations and showed that analysis of cfDNA over time could
another patient. be useful for determining recurrence.
The prognosis is poor for metastatic TNBC and TNBC
Discussion that recurs within 2‑3 years (1). Although clinical trials
have examined various treatments (22‑24), no cure has
We studied a high‑sensitivity detection system based on been established. In this study, we found that PI3K/AKT
ddPCR that enabled the detection of low‑level copy muta‑ signaling, and in particular PIK3CA H1047R or AKT1 E17K
tions and the identification of mutations in plasma samples mutations, were significantly associated with high incidence
of patients with metastatic TNBC. ddPCR is a relatively of metastasis as compared to the METABRIC cohort.
inexpensive method and has rapid turnover. The ddPCR assay The METABRIC database contains the details of genetic
is especially useful when there is insufficient genomic DNA mutations for 269 TNBC cases, which we compared to our
for NGS analysis. Circulating DNA in the cell‑free plasma results (Fig. 3). The frequency of AKT1 mutation was higher
fraction originates from many different cells, including in our study than in METABRIC. AKT1 E17K was the most
lymphocytes and neoplastic cells (17). Their DNA is released common mutation in both groups (Fig. 3A). For PIK3CA
into the circulation in the process of cellular destruction (18). gene, the mutation sites were generally consistent between
8 OKAZAKI et al: CIRCULATING TUMOR DNA ANALYSIS AS A MARKER OF RECURRENCE IN TNBC

our data and those in METABRIC; however, the mutational amplicon sequencing. The PI3K/AKT pathway was active in
frequency was not statistically significant, although there patients harboring AKT1 E17K and PIK3CA H1047R muta‑
was a tendency for higher mutational frequency in our study tions, suggesting an inhibitor of this pathway needed to be
compared with that shown in METABRIC (Fig. 3B). In both explored as a possible therapeutic agent. Finally, we success‑
groups, PIK3CA H1047R was the most frequent mutation fully detected predefined genetic mutations from circulating
(our data: 5/47, 10.6% and METABRIC: 15/269, 5.6%). free DNA using ddPCR method; however, for early recurrence
While Takeshita et al (5) reported that the presence of detection or treatment monitoring, we suggest using multiple
PIK3CA H1047R or E542K in cfDNA was associated with target detection strategy with high detection sensitivity.
favorable prognosis, in our cohort, patients with recurrence
within two years after surgery were corrected in this study, Acknowledgements
therefore we could not verify whether PIK3CA H1047R was
prognosis marker or not. In contrast, the mutations in TP53 The authors would like to thank Dr Munehiko Ogata and
were generally consistent, but no statistically significant Ms. Mayumi Suzuki (Sapporo Higashi Tokushukai Hospital)
difference in its frequency was observed (Fig. 3C). The for technical support during genetic analyses.
frequencies of TP53 R175H and R248Q mutations were
higher in METABRIC (13/269, 4.8% and 7/269, 2.6%, Funding
respectively), but these mutations were also common in our
study (3/47, 6.4%). A previous report suggested that Asian This work was supported by JSPS KAKENHI [grant
TNBC patients were more likely to harbor PI3K����������
/���������
AKT muta‑ no. 18K15262 (to RY)]
tions than patients of other ethnicities (25). AKT1 E17K
is considered a driver oncogene (26‑28), and AKT inhibi‑ Availability of data and materials
tors have been assessed in preclinical and clinical trials
in multiple cancer types known to harbor this mutation, The datasets generated and/or analyzed during the current
suggesting that AKT‑targeting therapy may be effective for study are not publicly available due to their containing genetic
patients with activated AKT signaling (29). Based on the information that could compromise the privacy of research
above, Asian TNBC patients have a higher frequency of this participants but are available from the corresponding author
mutation, and AKT inhibitors may be an effective treatment on reasonable request.
option for this population.
There are limitations in this study. First, the number of Authors' contributions
patient samples was small; however, we analyzed 36 patients
with TNBC who relapsed within 2 years after surgical resec‑ Conception and design: SaO and TS. Provision of study
tion, the frequency of AKT1 E17K was higher in this cohort, materials, and acquisition, analysis and interpretation of
and there may be clinical benefits in monitoring this mutation, data: SaO, TS, SY, MA, NY, RY, KI, YoM, ShO, SC, HT,
as it is a marker of recurrence. A long‑term surveillance study RH, TN, HK, AS, YOn, YuM, MK and YOh. All authors
in a large cohort of patients is warranted to confirm these wrote, read and approved the final manuscript.
important findings. Second, the liquid biopsy was performed
on patients whose tumor has already metastasized. For early Ethics approval and consent to participate
detection, it is necessary to confirm similar events in patients
whose tumor has not yet metastasized. This study protocol, including tumor sample collection and
Nevertheless, we believe that molecular target therapy for genetic analysis, was approved by the Asahikawa Medical
druggable driver oncogene will be applicable for the treatment University Research Ethics Committee (approval nos. 17043
of TNBC in the near future. Intervention with kinase inhibi‑ and 18118). For the prospective study (approval nos. 17043),
tors could improve prognosis as seen in other types of cancers. written informed consent was obtained from all patients before
So, it is important to know what kinds of driver oncogene are their enrollment for genetic analysis. For the retrospective
present in TNBC patients with poor prognosis that can be study for survival analysis of 36 patients with TNBC (approval
detected in cfDNA mutation analysis by cost‑effective and no. 18118), informed consent was waived by the Asahikawa
highly sensitive way. Medical University Research Ethics Committee and an oppor‑
In this study, considering the possibility for AKT inhibitor tunity for opt‑out regarding the study was provided through
treatment as a therapy for TNBC, liquid biopsy was performed the institutional website.
on patients with AKT1 and PIK3CA mutations. Based on
the results obtained, detecting single mutation detection like Patient consent for publication
PIK3CA and AKT for surveillance of TNBC which comprised
~32% of TNBC was not sufficient for the early detection of Written informed consent for publication of clinical details
TNBC recurrence; however, we believe that using predefined and /or clinical images was obtained from the patients.
genetic alteration, multiple targets in ddPCR, e.g., TP53
R175H/R248Q or high‑sensitive NGS‑based liquid biopsy Competing interests
could be employed to yield higher detection rates.
In conclusion, in our TNBC early recurrent patient cohort, YMiz and YOn receive funding from Hitachi High
AKT1 E17K (14.9%) and PIK3CA H1047R (10.6%) were Technologies, Inc. for the experimental equipment. No poten‑
common in surgically resected samples detected by target tial conflicts of interest were disclosed by the other authors.
 ONCOLOGY LETTERS 21: 420, 2021 9

References 15. Kanda Y: Investigation of the freely available easy‑to‑use software

‘EZR’ for medical statistics. Bone Marrow Transplant 48:
452‑458, 2013.
1. Massihnia D, Galvano A, Fanale D, Perez A, Castiglia M, 16. McCall CM, Mosier S, Thiess M, Debeljak M, Pallavajjala A,
Incorvaia L, Listì A, Rizzo S, Cicero G, Bazan V, et al: Beierl K, Deak KL, Datto MB, Gocke CD, Lin MT, et al: False
Triple negative breast cancer: Shedding light onto the role of positives in multiplex PCR‑based next‑generation sequencing
pi3k/akt/mtor pathway. Oncotarget 7: 60712‑60722, 2016. have unique signatures. J Mol Diagn 16: 541‑549, 2014.
2. Cossu‑Rocca P, Orrù S, Muroni MR, Sanges F, Sotgiu G, Ena S, 17. Schwarzenbach H, Hoon DS and Pantel K: Cell‑free nucleic acids
Pira G, Murgia L, Manca A, Uras MG, et al: Analysis of PIK3CA as biomarkers in cancer patients. Nat Rev Cancer 11: 426‑437,
mutations and activation pathways in triple negative breast 2011.
cancer. PLoS One 10: e0141763, 2015. 18. Bettegowda C, Sausen M, Leary RJ, Kinde I, Wang Y, Agrawal N,
3. Dent R, Trudeau M, Pritchard KI, Hanna WM, Kahn HK, Bartlett BR, Wang H, Luber B, Alani RM, et al: Detection of
Sawka CA, Lickley LA, Rawlinson E, Sun P and Narod SA: circulating tumor DNA in early‑ and late‑stage human malig‑
Triple‑negative breast cancer: Clinical features and patterns of nancies. Sci Transl Med 6: 224ra24, 2014.
recurrence. Clin Cancer Res 13: 4429‑4434, 2007. 19. Barak V, Holdenrieder S, Nisman B and Stieber P: Relevance of
4. Oxnard GR, Paweletz CP, Kuang Y, Mach SL, O'Connell A, circulating biomarkers for the therapy monitoring and follow‑up
Messineo MM, Luke JJ, Butaney M, Kirschmeier P, investigations in patients with non‑small cell lung cancer. Cancer
Jackman DM, et al: Noninvasive detection of response and Biomark 6: 191‑196, 2010.
resistance in EGFR‑mutant lung cancer using quantitative 20. Colt HG, Murgu SD, Korst RJ, Slatore CG, Unger M and
next‑generation genotyping of cell‑free plasma DNA. Clin Quadrelli S: Follow‑up and surveillance of the patient with lung
Cancer Res 20: 1698‑1705, 2014. cancer after curative‑intent therapy: Diagnosis and Management
5. Takeshita T, Yamamoto Y, Yamamoto‑Ibusuki M, Tomiguchi M, of Lung Cancer, 3rd ed: American College of Chest Physicians
Sueta A, Murakami K and Iwase H: Clinical significance of Evidence‑Based Clinical Practice Guidelines. Chest 143 (Suppl 5):
plasma cell‑free DNA mutations in PIK3CA, AKT1, and ESR1 e437S‑e454S, 2013.
gene according to treatment lines in ER‑positive breast cancer. 21. Emens LA and Davidson NE: The follow‑up of breast cancer.
Mol Cancer 17: 67, 2018. Semin Oncol 30: 338‑348, 2003.
6. Takeshita T, Yamamoto Y, Yamamoto‑Ibusuki M, Tomiguchi M, 22. Albain KS, Nag SM, Calderillo‑Ruiz G, Jordaan JP, Llombart AC,
Sueta A, Murakami K, Omoto Y and Iwase H: Analysis of Pluzanska A, Rolski J, Melemed AS, Reyes‑Vidal JM,
ESR1 and PIK3CA mutations in plasma cell‑free DNA from Sekhon JS, et al: Gemcitabine plus Paclitaxel versus Paclitaxel
ER‑positive breast cancer patients. Oncotarget 8: 52142‑52155, monotherapy in patients with metastatic breast cancer and prior
2017. anthracycline treatment. J Clin Oncol 26: 3950‑3957, 2008.
7. Chen YH, Hancock BA, Solzak JP, Brinza D, Scafe C, Miller KD 23. Perez EA, Hillman DW, Mailliard JA, Ingle JN, Ryan JM,
and Radovich M: Next‑generation sequencing of circulating Fitch TR, Rowland KM, Kardinal CG, Krook JE, Kugler JW, et al:
tumor DNA to predict recurrence in triple‑negative breast cancer Randomized phase II study of two irinotecan schedules for
patients with residual disease after neoadjuvant chemotherapy. patients with metastatic breast cancer refractory to an anthra‑
NPJ Breast Cancer 3: 24, 2017. cycline, a taxane, or both. J Clin Oncol 22: 2849‑2855, 2004.
8. Riva F, Bidard FC, Houy A, Saliou A, Madic J, Rampanou A, 24. Talbot DC, Moiseyenko V, Van Belle S, O'Reilly SM,
Hego C, Milder M, Cottu P, Sablin MP, et al: Patient‑Specific Alba Conejo E, Ackland S, Eisenberg P, Melnychuk D,
Circulating Tumor DNA Detection during Neoadjuvant Pienkowski T, Burger HU, et al: Randomised, phase II trial
Chemotherapy in Triple‑Negative Breast Cancer. Clin Chem 63: comparing oral capecitabine (Xeloda) with paclitaxel in patients
691‑699, 2017. with metastatic/advanced breast cancer pretreated with anthracy‑
9. Lakhani SR, Ellis IO and Schnitt SJ, Tan pH and van de clines. Br J Cancer 86: 1367‑1372, 2002.
Vijver MJ (eds): WHO Classification of Tumours of the Breast. 25. Jiang YZ, Ma D, Suo C, Shi J, Xue M, Hu X, Xiao Y, Yu KD,
Vol 4. 4th edition. IARC WHO Classification of Tumours, 2012 Liu YR, Yu Y, et al: Genomic and transcriptomic landscape of
10. Yoshida R, Sasaki T and Ohsaki Y: EGFR and KRAS mutations triple‑negative breast cancers: Subtypes and treatment strategies.
in triple‑mutated lung cancer. J Thorac Oncol 12: e92‑e93, 2017. Cancer Cell 35: 428‑440.e5, 2019.
11. Yoshida R, Sasaki T, Umekage Y, Tanno S, Ono Y, Ogata M, 26. de Bruin EC, Whiteley JL, Corcoran C, Kirk PM, Fox JC,
Chiba S, Mizukami Y and Ohsaki Y: Highly sensitive detection Armisen J, Lindemann JPO, Schiavon G, Ambrose HJ and
of ALK resistance mutations in plasma using droplet digital Kohlmann A: Accurate detection of low prevalence AKT1 E17K
PCR. BMC Cancer 18: 1136, 2018. mutation in tissue or plasma from advanced cancer patients.
12. Ono Y, Sugitani A, Karasaki H, Ogata M, Nozaki R, Sasajima J, PLoS One 12: e0175779, 2017.
Yokochi T, Asahara S, Koizumi K, Ando K, et al: An improved 27. Kim MS, Jeong EG, Yoo NJ and Lee SH: Mutational analysis
digital polymerase chain reaction protocol to capture low‑copy of oncogenic AKT E17K mutation in common solid cancers and
KRAS mutations in plasma cell‑free DNA by resolving acute leukaemias. Br J Cancer 98: 1533‑1535, 2008.
'subsampling' issues. Mol Oncol 11: 1448‑1458, 2017. 28. Rudolph M, Anzeneder T, Schulz A, Beckmann G, Byrne AT,
13. Ishibashi K, Kumai T, Ohkuri T, Kosaka A, Nagato T, Hirata Y, Jeffers M, Pena C, Politz O, Köchert K, Vonk R, et al: AKT1
Ohara K, Oikawa K, Aoki N, Akiyama N, et al: Epigenetic (E17K) mutation profiling in breast cancer: Prevalence,
modification augments the immunogenicity of human leukocyte concurrent oncogenic alterations, and blood‑based detection.
antigen G serving as a tumor antigen for T cell‑based immuno‑ BMC Cancer 16: 622, 2016.
therapy. OncoImmunology 5: e1169356, 2016. 29. Hyman DM, Smyth LM, Donoghue MTA, Westin SN, Bedard PL,
14. Pereira B, Chin SF, Rueda OM, Vollan HK, Provenzano E, Dean EJ, Bando H, El‑Khoueiry AB, Pérez‑Fidalgo JA,
Bardwell HA, Pugh M, Jones L, Russell R, Sammut SJ, et al: The Mita A, et al: AKT inhibition in solid tumors with AKT1
somatic mutation profiles of 2,433 breast cancers refines their mutations. J Clin Oncol 35: 2251‑2259, 2017.
genomic and transcriptomic landscapes. Nat Commun 7: 11479,
2016.