We retrospectively selected patients with invasive breast cancer who underwent surgery after NAC at the general surgery unit of Samsung Medical Center between December 2004 and June 2016. There were 287 patients with HER2-positive breast cancer among 967 selected patients. All patients underwent at least three cycles of NAC before surgery. The NAC regimens without trastuzumab included anthracycline-based (doxorubicin with cyclophosphamide) or anthracycline and taxane-based (doxorubicin and docetaxel with or without cyclophosphamide) regimens. The NAC regimens including trastuzumab were anthracycline-based, taxane-based (docetaxel), or anthracycline and taxane-based (doxorubicin with docetaxel and cyclophosphamide) regimen. We collected the sequential specimen blocks including pre-NAC, post-NAC, and recurrent specimens of each patient. Patients were excluded if they did not have either pre-NAC or post-NAC specimen blocks. Finally, a total of 100 patients were included in this study. All clinical parameters, including patient age, menopause status, clinical tumor-node-metastasis (TNM) stage, NAC regimen, surgery type, follow-up period after diagnosis, survival, and tumor recurrence were reviewed based on electronic medical records from initial diagnosis to last recorded follow-up until July 2016. All hematoxylin and eosin (H&E)-stained slides and immunohistochemistry slides were reviewed by two pathologists (Y.S. and E.Y.C.). Tumor recurrence was defined as involving both local recurrence and distant metastasis, while pCR was defined as no invasive tumor in the breast or lymph nodes, and no lymphovascular invasion. This study was approved by the Institutional Review Board (IRB) of Samsung Medical Center (SMC 2016-08-005). The requirement for formal written informed consent was waived by the IRB.

Immunohistochemical tests for estrogen receptor (ER), progesterone receptor (PR), and HER2 were performed on 10% formalin-fixed and paraffin-embedded (FFPE) blocks of pre-NAC and post-NAC biopsy specimens. After rehydration and antigen retrieval in citrate buffer, the tissue sections were stained for ER (clone 6F11; Novocastra, Newcastle upon Tyne, UK), and PR (clone 16; Novocastra). Immunohistochemical positivity for ER and PR was determined according to the American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP) guidelines using a cutoff of >1% stained tumor nuclei and evaluated by Allred score [22]. HER2 positivity was defined as HER2 3+ expression by immunohistochemistry (clone 4B5; Ventana, Tucson, USA) or 2+ expression by immunohistochemistry with amplification by silver in situ hybridization (SISH) [23]. HER2 SISH analysis was performed using INFORM DDISH HER-2 DNA SISH probe kits (BenchMark XT; Ventana) on 10% FFPE blocks. HER2 positivity was determined according to ASCO/CAP guidelines using a cutoff HER2 amplification ratio of ≥2.0 and/or an average HER2 copy number of ≥6.0 signals per cell.

PIK3CA mutation analysis was performed on corresponding FFPE blocks. All H&E slides were reviewed, and the target area containing invasive carcinoma and ductal carcinoma in situ (DCIS) was marked on the slides. Genomic DNA was extracted from the 5-µm-thick unstained slides of the FFPE block using manual microdissection. Mutational analysis of PIK3CA exon 9 and 20, which were reported as hotspots, was performed by directional sequencing of polymerase chain reaction (PCR) fragments amplified from genomic DNA using sequencing probes, similar to that reported in a previous study [24]. The forward primer for exon 9 was 5′-TGTGAATCCAGAGGGGAAAA-3′, and the reverse primer was 5′-TGCTGAGATCAGCCAAATTCA-3′; the forward primer for exon 20 was 5′-TTGCATACATTCGAAAGACC-3′, and the reverse primer was 5′-CCTATGCAATCGGTCTTGC-3′. The PCR products were subjected to Sanger sequencing to detect the mutation.

The associations between PIK3CA mutations and clinicopathological parameters were statistically evaluated using several tests, including chi-square test, Fisher exact test, Student t-test, Mann-Whitney U-test, and Kruskal-Wallis test. For the survival analyses, OS was defined as the time between the date of diagnosis and the date of death or last follow-up. DFS was defined as the time between the date of diagnosis and the date of tumor recurrence or last follow-up. Survival analyses were assessed using a Kaplan-Meier curve and the log-rank test. Multivariate analysis was used to evaluate the prognostic value of OS and DFS using a Cox proportional hazards model. Binary logistic analysis was performed to evaluate the predictive significance of pCR. A two-sided p-value less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS version 20.0 (IBM Corp., Armonk, USA).

Average patient age was 47.5±10.0 years. PIK3CA mutations were observed in the sequential specimens of 17 patients. Among these, five patients had mutations in exon 9, and the rest had mutations in exon 20. The frequency of PIK3CA mutation in the pre-NAC, post-NAC, and recurrent specimens was 12.0% (12/100), 13.6% (11/81), and 20.0% (3/15), respectively. PIK3CA mutation analysis could not be performed on the post-NAC specimens of 19 patients due to insufficient amount of residual tumor. Among these 19 uninvestigated patients, 14 showed pCR but low DCIS volume, and five had invasive carcinoma smaller than 3 mm in post-NAC specimens.

In 23 cases, post-NAC specimens did not provide enough residual invasive carcinoma in the resected breast for analysis. In these cases, post-NAC PIK3CA mutation analysis was conducted in residual DCIS. Not all of the analyzed residual DCIS specimens revealed a PIK3CA mutation.

Patients in the mutant PIK3CA group had a higher pathol-ogic N stage (ypN stage, p=0.008) and underwent total mastectomy more frequently than those in the wild-type group (p=0.034). Patients with PIK3CA mutations tended to have larger tumor size after NAC (p=0.075). There was no significant difference between the two groups when comparing other parameters. The baseline characteristics of the study population are summarized in Table 1.

In total, 31 patients achieved pCR after NAC. Table 2 summarizes the pCR rate under varying conditions. There was no difference between the pCR rate of patients who had at least one PIK3CA mutation in sequential specimens and that of patients with no PIK3CA mutation in all sequential specimens (17.6% vs. 33.7%, p=0.191). PIK3CA mutation in post-NAC specimens correlated with a lower pCR rate than that observed in the post-NAC wild-type specimens (0.0% vs. 24.3%, p<0.001). However, there was no statistical difference in pCR rate between the pre-NAC mutant group and the pre-NAC wild-type group (25.0% vs. 31.8%, p=0.750).

The significance of pCR rates associated with the site of PIK3CA mutation was assessed in the group with at least one mutation in sequential specimens and in the group with mutant PIK3CA in the pre-NAC specimens. No significant difference was observed in the pCR rates of these two groups (at least one mutation in sequential specimen: wild-type vs. exon 9 vs. exon 20, 33.7% vs. 20.0% vs. 16.7%, p=0.199; pre-NAC specimen: wild-type vs. exon 9 vs. exon 20, 31.8% vs. 25.0% vs. 25.0%, p>0.999).

The survival curves for each condition are depicted in Figure 1. Patients with at least one PIK3CA mutation had a significantly shorter median DFS than those in the wild-type group (58.3 months vs.119.3 months, p=0.020) (Figure 1A). Survival analysis of the post-NAC specimens revealed a trend for shorter DFS in the mutant PIK3CA group (60.2 months vs. 113.5 months, p=0.148) (Figure 1B). There was no significant difference between the DFS of the mutant group and that of the wild-type group in pre-NAC specimens (51.8 months vs. 103.9 months, p=0.477) (Figure 1C).

Sequential mutation analysis indicated no significant difference in median DFS between the group exhibiting exon 9 mutation and the group exhibiting exon 20 mutation (119.3 months vs. 47.2 months vs. 59.8 months, p=0.063) (Figure 1D). Survival analysis of post-NAC specimens showed significant median DFS differences among the three mutation site groups (exon 9 vs. exon 20 vs. post-NAC wild-type: 41.8 months vs. 74.0 months vs. 113.5 months, p=0.039) (Figure 1E). Among the mutation sites, a PIK3CA mutation in exon 9 was found to yield a significantly shorter median DFS than other groups (exon 9 vs. exon 20, p=0.048; exon 9 vs. post-NAC wild-type, p=0.017). There was no significant difference in median OS according to PIK3CA mutation status (84.5 months vs. 118.0 months, p=0.984) (Figure 1F).

Univariate analysis revealed that PIK3CA mutation and post-NAC tumor size were significantly associated with a worse DFS in sequential specimens (PIK3CA mutation: hazard ratio [HR], 3.143, 95% confidence interval [CI], 1.137–8.689, p=0.027; post-NAC tumor size: HR, 1.022, 95% CI, 1.009–1.036, p=0.001). These results are presented in Supplementary Table 1 (available online). Multivariate analysis also determined that PIK3CA mutation was a significant contributing factor for a shorter DFS (HR, 3.550; 95% CI, 1.001–12.589; p=0.050). Post-NAC tumor size and the presence of pre-NAC clinical node metastasis were also determined to be significant contributors by multivariate analysis (post-NAC tumor size: HR, 1.029, 95% CI, 1.007–1.052, p=0.010; pre-NAC clinical node metastasis: HR, 0.003, 95% CI, 0.000–0.047, p<0.001). These results are presented in Table 3. All the patients in this study underwent NAC, and a total of 97 patients had lymph node metastasis at the time of diagnosis due to characteristics of NAC. One out of three patients (33.3%) who were not clinically diagnosed with lymph node metastasis (cN0) had tumor recurrence at a higher rate than that observed at other stages. This could explain why a shorter DFS was observed in the cN0 stages.

Univariate analysis for OS in the sequential specimens revealed the significance of post-NAC tumor size (HR, 1.031; 95% CI, 1.012–1.051; p=0.001) and the presence of lymphovascular invasion (HR, 4.140; 95% CI, 1.035–16.562; p=0.045). Supplementary Table 2 (available online) summarizes these results. However, these factors were not found to be significant in multivariate analysis. As shown in Table 4, older age and the presence of pre-NAC clinical lymph node metastasis were additional significant factors for shorter OS (age: HR, 1.109, 95% CI, 1.004–1.225, p=0.042; pre-NAC clinical lymph node metastasis: HR, 0.027, 95% CI, 0.001–0.633, p=0.025).

Due to the small sample size, it was difficult to identify all variables for pCR analysis. Supplementary Table 3 (available online) summarizes the univariate analysis for pCR. According to univariate evaluation, trastuzumab treatment as part of NAC was a significant factor for pCR (HR, 4.610; 95% CI, 1.453–14.628; p=0.009). Similar significance was obtained using multivariate analysis for pCR (HR, 4.766; 95% CI, 1.464–15.519; p=0.010). In both univariate and multivariate analyses, PIK3CA mutation did not correlate with pCR (univariate: HR, 0.714, 95% CI, 0.179–2.843, p=0.633; multivariate: HR, 0.988, 95% CI, 0.228–4.279, p=0.987). These results can be viewed in Table 5.

For evaluation of the predictive impact of PIK3CA mutation, the difference in the rate of pCR after NAC was statistically analyzed. In the trastuzumab-treated group, patients who had at least one PIK3CA gene mutation in sequential specimens experienced a 20.0% pCR rate, whereas patients who had purely wild-type PIK3CA experienced a 43.1% pCR rate; this difference, however, was not significant (p=0.294). Pre-NAC subgroup analysis yielded corresponding results with no significant predictive effect associated with the PIK3CA mutation (pre-NAC mutant group: 33.3% vs. pre-NAC wild-type: 40.3%, p>0.999). Supplementary Table 4 (available online) summarizes these results.

In nine out of 100 cases, PIK3CA mutation status changed in sequential specimens. In four cases, the pre-NAC PIK3CA mutation was lost after NAC (MT-WT group). New PIK3CA mutations developed in five cases (WT-MT group), and of these, tumor recurrence was observed in four. In 83 cases, PIK3CA remained purely wild-type in sequential specimens (WT-WT group), while purely PIK3CA mutants were found in the sequential specimens (MT-MT group) of eight cases. The clinicopathological details of patients who experienced changes in their mutation status are described in Table 6.

The DFS curves of the mutation change groups are plotted in Figure 2. Because there was no tumor recurrence in the MT-WT group, the Kaplan-Meier method could not be employed for survival analysis. The log-rank test was used, revealing a significant difference in the survival curves (p=0.016). More specifically, there was a statistically significant difference between the purely wild-type group and developed PIK3CA mutation group (WT-WT vs. WT-MT, p=0.005). While the survival curve for the MT-MT group showed a worse trend than that for the MT-WT group (MT-MT vs. MT-WT, p=0.079), further analysis between the subgroups showed no significant differences.