This retrospective study began with a review of data from 183 breast cancer patients who underwent surgery with SLNB, and whose SNs were examined by OSNA at Osaka University Hospital between April 2017 and March 2019. After applying the exclusion criteria, data from a total of 94 patients were ultimately used in the analysis of mutations in the primary tumor (Figure 1).

Tumors were obtained at the time of surgery and immediately frozen and stored at –80 °C until analysis by next-generation sequencing. SLNB was performed using both dye (patent blue or indocyanine green) and radiocolloid (technetium-99m tin colloid). The whole SN tissue was used for OSNA and homogenized in 4 mL of Lynorhag solution (Sysmex Corporation, Kobe, Japan); 2 μL of lysate was used for the assay, and the remainder was stored at –80 °C. To ensure adequate homogenization, large lymph nodes were divided into multiple tissue fragments for processing; thus, a single node could generate more than one lysate.

The classification of CK19 mRNA copy number per assay was as follows: >5000, (++); >250 and ≤5000, (+); >0 and ≤250 (–), and 0, (ND). As recommended by the manufacturer of the OSNA assay, (++) and (+) were considered positive for SN metastasis, and >0 and ≤250 were regarded negative for SN metastasis, despite the amplification of CK19 mRNA. In cases of a positive OSNA result, the decision as to whether or not to perform ALND was based on a nomogram used to predict non-SN metastasis (10, 18). This nomogram was developed at our institution to guide decision making regarding ALND, and its use has been investigated through a clinical trial approved by the Institutional Review Board of Osaka University Hospital (approval no. 21433).

A total of 214 SNs and 231 lysates from 94 patients were analyzed. Mutation analysis of 21 genes in primary tumors was performed by next-generation sequencing in all cases. For cases of PIK3CA mutations in the primary tumor, PIK3CA hotspot mutations (PIK3CA H1047R, E545K, and E542K) in SN lysates were analyzed using ddPCR (Figure 2). The PI3K/AKT signaling pathway–related genes (AKT1 and PTEN) were analyzed using molecular barcode–based next-generation sequencing. The detailed protocol followed that described by Yoshinami et al. (19).

Tumor DNA was extracted from fresh frozen tissue using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany), according to manufacturer instructions. Peripheral blood lymphocytes were separated from whole blood by centrifugation for 10 min at 3000 rpm (1840 × g) twice and stored in the pellet state at –80 °C until further use. Peripheral blood lymphocyte DNA was extracted from the pellet using the DNeasy Blood & Tissue Kit, according to manufacturer instructions. Sentinel lymph node DNA was extracted from 100–500 μL of OSNA lysate, using the QIAamp Circulating Nucleic Acid Kit (Qiagen), and eluted using 50 μL of desalted water.

The gene panel used in this study was an institutionally designed custom target panel generated using Agilent’s SureDesign platform (https://earray.chem.agilent.com/suredesign). The panel was not a commercially supplied next-generation sequencing kit but was specifically designed for this study to cover the whole exonic regions of 21 genes, including 12 genes frequently mutated in breast cancer (cBioPortal for Cancer Genomics, http://www.cbioportal.org) as well as ESR1 (Supplementary Table 1). The median coverage of the sequencing area was 100% (range, 92%–100%) of the whole exome for each gene. Sequence libraries were prepared from 200 ng of tumor DNA, using a SureSelect XT HS2 (Agilent Technologies, Inc., Santa Clara, CA, USA) according to manufacturer instructions, and sequenced using MGI DNBSEQ-G400RS (MGI Tech, Shenzhen, China). CLC Genomics Workbench (Qiagen) was used for variant calling.

The exclusion criteria were as follows: variant allele frequency, <5%; read depth, <100; and forward–reverse balance, <0.2, in introns, reported as single nucleotide polymorphism with variant allele frequency >0.5% at the Tohoku Medical Megabank Organization (https://www.megabank.tohoku.ac.jp/english/).

For mutation analysis, DNA was extracted from SN lysates as described above. Then, 9 μL of DNA solution was made up to 20 μL by mixing with the following solutions: 1XddPCR Supermix for probes (Bio-Rad Laboratories, Inc., Hercules, CA, USA), 900 nM for each primer, 250 nM for probe, and DNA. The sequences of the primers and probes are summarized in Supplementary Table 2.

Droplet generation oil was added to the mixture, which was subsequently transferred onto a QX100 droplet generator (Bio-Rad Laboratories, Inc.). Then, 40 μL of emulsified mixture was subjected to polymerase chain reaction using a T100 thermal cycler (Bio-Rad Laboratories, Inc.) at 95 °C for 10 min, followed by 40 cycles at 94 °C for 30 s, and 56 or 60 °C for 1 min, and 98 °C for 10 min.

Data analysis was performed using the QX100 droplet reader and QuantaSoft software, version 1.7.4 (both Bio-Rad Laboratories, Inc.). The presence of two or more dots per well was considered a positive result, and the copy numbers of three positive wells were summated. For cases divided into multiple lysates, the node was defined as positive if at least one lysate was ddPCR-positive.

R version 4.3.3 was used for statistical processing. Pearson’s correlation analysis was used to evaluate the association between copy numbers of mutated DNA and CK19 mRNA in lysates. Because both variables showed right-skewed distributions, log-transformed values were used. A p value < 0.05 was considered significant.

The study was approved by the Ethical Review Board of Osaka University Hospital (approval date and number: January 29, 2019, #18396). Before surgery, all patients provided opt-out consent for the initial storage of samples and the use of samples in this study.

Data from 94 patients were evaluated (Figure 1), and the median follow-up period was 78.0 months (range, 1–105 months). Their clinicopathological characteristics are summarized in Table 1, Figure 3.

We performed panel sequencing of 21 genes in 94 patients and found mutations in 63 primary tumors (67.0% of patients). A total of 16 genes were mutated: PIK3CA in 31 cases (33.0%), PTEN in 5 cases (5.3%; one case harbored two additional PTEN mutations), and AKT1 in 3 cases (3.2%) (Supplementary Table 1). In the 31 patients with PIK3CA mutations, mutations in hotspots H1047R, E545K, and E542K were detected in 13 (41.9%), 7 (22.6%), and 5 (16.1%), respectively (Supplementary Table 3). The breast cancer subtypes in these 31 patients were as follows: HR-positive and HER2-negative (n = 25), HER2-positive (n = 6); none had triple-negative breast cancer. Two patients with PIK3CA mutation–positive primary tumors developed recurrence during follow-up; the mutation was not a hotspot in either case.

We analyzed 59 SNs and 66 lysates in 25 patients with PIK3CA mutations: 1 SN, 2 SNs, and ≥3 SNs from each of 6, 9, and 10 patients, respectively. Intraoperative OSNA diagnosis of the 59 SNs showed that 10 (from 7 patients) were metastasis positive; 6 SNs (from 4 patients) were (++) and 4 (from 3 patients) were (+). In all 10 SNs determined metastasis positive by the OSNA, PIK3CA mutations were found. However, PIK3CA mutations were also detected in 30 of the 49 SNs (equivalent to 18 of 25 patients) diagnosed as negative for metastasis by the intraoperative OSNA (Figure 4A). For the five patients with PTEN mutations, 13 SNs and corresponding 13 lysates were analyzed. Intraoperative OSNA examination identified 2 of the 13 SNs (equivalent to 2 of 5 patients) as metastasis-positive; both cases were classified as micrometastasis. PTEN mutations were detected in only one SN, which was one of the metastasis-positive nodes. For the three patients with AKT1 mutations, 10 SNs and the corresponding 10 lysates were analyzed. Intraoperative OSNA examination showed that none of the 10 SNs were metastasis-positive, and no AKT1 mutations were detected in any SN. Because PTEN and AKT1 mutations were only rarely detected in OSNA lysates using molecular barcode–based next-generation sequencing, we focused on PIK3CA for subsequent analysis.

In the 59 SNs analyzed, a positive correlation was observed between the copy numbers of CK19 mRNA and mutated PIK3CA genes (r = 0.47, 95% confidence interval, 0.25–0.64, p < 0.01) (Figure 4B). This suggests that the PIK3CA mutation copy number might have potential as a clinical marker for evaluating tumor burden in SNs, as it showed a trend consistent with CK19 mRNA levels. Of the cases of positive PIK3CA mutation in the primary tumor, it was possible to detect the PIK3CA mutation in all SNs assessed as metastatic by OSNA, with a 100% positive predictive rate for gene mutation detection (Figure 5).

The absence of false negatives for OSNA in DNA analysis by ddPCR suggests the potential for more sensitive metastasis detection, although these findings are exploratory and hypothesis-generating. It is necessary to investigate whether, through further studies, the results of whole-genome and whole-exome sequencing could be evaluated for other genes.