We investigated the loss of somatic mutations in TP53 and PIK3CA in breast cancer tissue after neoadjuvant chemotherapy (NCT) and the clinical relevance of the observed mutation profiles. Samples were derived from three cohorts: Cohort 1 consisting of 206 patients undergoing NCT with matched pre- and postchemotherapy tumor tissues; Cohort 2 consisting of 158 additional patients undergoing NCT; and Cohort 3, consisting of 81 patients undergoing chemotherapy with prechemotherapy tumor tissues. In the first cohort, somatic mutations in TP53 or PIK3CA were identified in 24.8% of the pre-NCT tumor samples but in only 12.1% of the post-NCT tumor samples (P < 0.001). Patients with initial TP53 and PIK3CA mutations who became negative for the mutations after NCT had a higher Miller–Payne score (P = 0.008), improved disease-free survival, and improved overall survival than those with no change or the opposite change. The association of loss of mutations in TP53 and PIK3CA and improved survival was successfully validated in the second cohort. In addition, 28.4% of the tumors showed intratumoral heterogeneity of somatic mutations in TP53 or PIK3CA, whereas 71.6% were homogeneous, either with or without the mutations. Our data reveal the novel concept that chemotherapy may reduce mutation frequency in patients with breast cancer. Furthermore, the loss of somatic mutations in TP53 and PIK3CA may be translated to biomarkers for prognosis via further verification, which may optimize the choice of sequential therapy and improve patient survival. Cancer Res; 74(13); 3399–407. ©2014 AACR.

Neoadjuvant chemotherapy (NCT) provides an opportunity to record breast cancer phenotype in a setting where both pre- and postchemotherapy samples can be obtained and the response to chemotherapy can be documented (1). It is of considerable importance to elucidate genetic changes and to identify prognostic biomarkers associated with NCT, which can provide a more accurate assessment of individual treatment. The altered gene expression profiles of breast cancer under NCT have been described (2–4). However, to the best of our knowledge, a comparison of the exomes and genomes of pre- and post-NCT samples does not yet exist. We hypothesized that mutation shift might occur in breast tumors during NCT, and a comparison of the exomes of pre- and post-NCT samples might reveal genetic alterations associated with prognosis.

Recent advances in next-generation sequencing (e.g., whole genome sequencing and exome sequencing) provide further opportunities to more completely characterize the molecular architecture of breast cancer (5, 6). Integrated molecular analyses of breast cancer have revealed that TP53 and PIK3CA are the most frequently mutated genes, each with a mutation frequency >35% (7). Using exome sequencing, we revealed that breast cancers might undergo loss of TP53 and PIK3CA mutations after NCT. This study investigates the loss of TP53 and PIK3CA mutations after NCT and its clinical relevance. This phenomenon, in turn, may serve as prognostic biomarkers and optimize the choice of sequential therapy.

Study cohorts

Patients selected for the current study fulfilled the following inclusion criteria: (i) female patients diagnosed with unilateral histologically confirmed invasive ductal carcinoma (IDC); breast carcinoma in situ (with or without microinvasion) were excluded; (ii) pathologic examination of tumor specimens was carried out by the Department of Pathology in Fudan University Shanghai Cancer Center (FUSCC, Shanghai, PR China); (iii) patients without any evidence of metastasis at diagnosis. Patients in the first cohort and second cohort had all received three to six cycles of NCT (paclitaxel and carboplatin based) and had residual invasive tumors in the operated breast. In the first cohort, patients were treated at FUSCC between January 1, 2003 and December 31, 2006. In the second cohort for validation of the findings from the first cohort, patients were treated between January 1, 2007 and December 31, 2009. Of the 242 patients who were originally enrolled in the first cohort, 36 patients were considered as pathologic complete remission (pCR) and excluded from the study. Similarly, 30 of the 188 patients initially enrolled in the second cohort were considered pCR and further excluded. As a result, the first cohort consisted of 206 female non-pCR patients with histologically confirmed IDC. The second cohort (the validation cohort) consisted of 158 female non-pCR patients with IDC. Matched samples from biopsies and surgical resection were collected before and after neoadjuvant treatment. Pathologic response was assessed postoperatively using the Miller–Payne scoring system (8, 9), with grade 5 considered to be pCR and grade 1 to 4 to be non-pCR. The third cohort consisted of 81 randomly selected patients with histologically confirmed IDC who had undergone prior surgical resection and received adjuvant chemotherapy. All patients in the third cohort were treated at FUSCC between January 1, 2003 and December 31, 2009.

The molecular subtypes of breast cancer according to immunohistochemical (IHC) profiles were categorized as follows: Luminal A = ER/PR+, HER2, and Ki67 < 14%; Luminal B = ER/PR+ and HER2+ or Ki67 ≥ 14%; HER2+ = ER, PR, and HER2+; and basal-like = ER, PR, HER2, and CK5/6+ or EGFR+ (10, 11). Tumors with IHC phenotypes of ER, PR, HER2, CK5/6, and EGFR were excluded. The staining and the interpretation of ER, PR, HER2, Ki-67, CK5/6, and EGFR have been previously described (10, 12). The clinicopathologic characteristics of the patients are listed in Table 1.

Table 1.

Clinicopathologic characteristics of patients with breast cancer who received neoadjuvant and adjuvant chemotherapy

CharacteristicsCohort 1: NCT (n = 206); patient no. (%)Cohort 2: NCT (n = 158); patient no. (%)P1aCohort 3: CT (n = 81); patient no. (%)P2b
Age at diagnosis, y   0.573  0.526 
 Mean 51 52  53  
 SD 9.2 10.1  8.9  
 ≤50 102 (49.5) 76 (48.1)  36 (44.4)  
 >50 104 (50.5) 82 (51.9)  45 (55.6)  
      
Menopausal status   0.554  0.215 
 Premenopausal 107 (51.9) 87 (55.1)  37 (45.7)  
 Postmenopausal 99 (48.1) 71 (44.9)  44 (54.3)  
      
(Pre-NCT) Tumor size, cm   0.339   
 Mean 4.6 4.9  3.1 <0.001 
 SD 1.6 1.7  1.2  
 <2 41 (19.9) 32 (20.2)  31 (38.3)  
 2–5 76 (36.9) 48 (30.4)  37 (45.7)  
 >5 89 (43.2) 78 (49.4)  13 (16.0)  
      
(Pre-NCT) Node status   0.521  0.068 
 Negative 55 (26.7) 47 (29.7)  31 (38.3)  
 Positive 151 (73.3) 111 (70.3)  50 (61.7)  
      
(Pre-NCT) Tumor grade   0.503  0.221 
 1–2 60 (29.1) 41 (25.9)  28 (34.6)  
 3 146 (70.9) 117 (74.1)  53 (65.4)  
      
Molecular subtype   0.126  0.601 
 Luminal A 87 (42.2) 58 (36.7)  38 (46.9)  
 Luminal B 49 (23.8) 31 (19.6)  18 (22.2)  
 HER2+ 24 (11.7) 16 (10.1)  8 (9.9)  
 Basal-like 46 (22.3) 53 (33.6)  17 (21.0)  
      
Pathologic response (Miller–Payne score)   0.041  NA 
 SD/PD or 1 51 (24.8) 23 (14.6)  NA  
 2 49 (23.8) 46 (29.1)    
 3 54 (26.2) 55 (34.8)    
 4 52 (25.2) 34 (21.5)    
      
Estimated 3-year DFS   0.732  0.052 
 Percentage survival 0.768 0.752  0.811  
 SE 0.032 0.038  0.048  
      
Estimated 3-year OS   0.800  0.078 
 Percentage survival 0.860 0.852  0.901  
 SE 0.026 0.031  0.036  
CharacteristicsCohort 1: NCT (n = 206); patient no. (%)Cohort 2: NCT (n = 158); patient no. (%)P1aCohort 3: CT (n = 81); patient no. (%)P2b
Age at diagnosis, y   0.573  0.526 
 Mean 51 52  53  
 SD 9.2 10.1  8.9  
 ≤50 102 (49.5) 76 (48.1)  36 (44.4)  
 >50 104 (50.5) 82 (51.9)  45 (55.6)  
      
Menopausal status   0.554  0.215 
 Premenopausal 107 (51.9) 87 (55.1)  37 (45.7)  
 Postmenopausal 99 (48.1) 71 (44.9)  44 (54.3)  
      
(Pre-NCT) Tumor size, cm   0.339   
 Mean 4.6 4.9  3.1 <0.001 
 SD 1.6 1.7  1.2  
 <2 41 (19.9) 32 (20.2)  31 (38.3)  
 2–5 76 (36.9) 48 (30.4)  37 (45.7)  
 >5 89 (43.2) 78 (49.4)  13 (16.0)  
      
(Pre-NCT) Node status   0.521  0.068 
 Negative 55 (26.7) 47 (29.7)  31 (38.3)  
 Positive 151 (73.3) 111 (70.3)  50 (61.7)  
      
(Pre-NCT) Tumor grade   0.503  0.221 
 1–2 60 (29.1) 41 (25.9)  28 (34.6)  
 3 146 (70.9) 117 (74.1)  53 (65.4)  
      
Molecular subtype   0.126  0.601 
 Luminal A 87 (42.2) 58 (36.7)  38 (46.9)  
 Luminal B 49 (23.8) 31 (19.6)  18 (22.2)  
 HER2+ 24 (11.7) 16 (10.1)  8 (9.9)  
 Basal-like 46 (22.3) 53 (33.6)  17 (21.0)  
      
Pathologic response (Miller–Payne score)   0.041  NA 
 SD/PD or 1 51 (24.8) 23 (14.6)  NA  
 2 49 (23.8) 46 (29.1)    
 3 54 (26.2) 55 (34.8)    
 4 52 (25.2) 34 (21.5)    
      
Estimated 3-year DFS   0.732  0.052 
 Percentage survival 0.768 0.752  0.811  
 SE 0.032 0.038  0.048  
      
Estimated 3-year OS   0.800  0.078 
 Percentage survival 0.860 0.852  0.901  
 SE 0.026 0.031  0.036  

NOTE: All comparisons by χ2 test except age, tumor size (Student t test), and estimated 3-year survival (log-rank test).

Abbreviations: CT, chemotherapy; NA, not applicable.

aP value between the first and second cohorts.

bP value between NCT and CT.

Follow-up for patients in the three cohorts was completed on December 31, 2013. The median length of follow-up was 63 months (range, 5–97 months) in the first cohort, 42 months (range, 4–82 months) in the second cohort, and 64 months (range, 5–106 months) in the third cohort. Our definition of disease-free survival (DFS) events included: the first recurrence of disease at a local, regional, or distant site, the diagnosis of contralateral breast cancer, and death from any causes (13). The overall survival (OS) was calculated from the date of diagnosis to the date of death or last follow-up. Patients without events or death were censored at the last follow-up. This study was approved by the Ethics Committee of FUSCC, and each participant signed an informed consent document.

Sample collection and processing

Tumor samples and paired blood samples were obtained from patients undergoing surgical treatment at FUSCC in accordance with the appropriate institutional review boards. For the first and second cohorts, tumor tissues were macrodissected to avoid the influence of stromal tissues (<10%). For each case in the third cohort, one section 15 μm in thickness was stained with hematoxylin and eosin and microdissected using the AS Laser Microdissection system (Leica Microsystems). In total, 30 to 40 tumor foci, each containing approximately 100 cancer cells, were obtained. QIAamp DNA Mini Kits (Qiagen) were used to extract DNA from the tissues. The quality and concentration of the extracted DNA were determined using a NanoDrop 2000 (Thermo Fisher Scientific). The extracted DNA was then used for mutation analysis.

Exome capture and sequencing

DNA extracted from tumor samples and paired blood samples were sent to Beijing Genomics Institute-Shenzhen for exome sequencing as described previously (14–17). Exome capture was done using the SureSelect Human All Exon Kit (Agilent), guided by the manufacturer's protocols. In brief, the qualified genomic DNA samples were randomly fragmented to 200 to 500 bp, followed by end-repair, A-tailing, and paired-end index adapter ligation. Final libraries were validated by Bioanalyzer analysis (Agilent) and quantitative PCR. Each captured library was loaded on the Hiseq 2000 platform and paired-end sequencing was performed with read lengths of 90 bp, providing at least 100-fold average depth for each sample. SOAPaligner 2.20 (18) was used to align the reads to human reference hg19 with parameters set to “-a -b -D -o -v 2 -r 1 -t -n 4.” Reads that aligned to the designed target region were collected for single nucleotide variation (SNV) identification and subsequent analysis. For SNV calling, the Variant Quality Score Recalibration method was applied to estimate the probability that each variant is a true genetic variant or a machine artifact. For indel calling, Burrows–Wheeler Aligner (BWA, V.0.5.8) was used to gap align the sequence reads to the human reference hg19. GATK IndelRealigner was used to perform local realignment of the BWA-aligned reads and GATK IndelGentotyperV2 was used to call indels. The variants were functionally annotated and were subsequently categorized into missense, nonsense, read-through, or splice-site mutations and coding indels, which are likely to be deleterious compared with synonymous and noncoding mutations.

Mutation analysis

The detection of mutations in frozen tumor tissues and paired blood DNA was performed by Sanger sequencing of all exons of TP53 and PIK3CA. PCR amplification was performed on an ABI 9700 Thermal Cycler using standard conditions. After amplification, the products were purified using a QIAquick PCR Purification Kit (Qiagen) and directly sequenced using an ABI PRISM 3730 Automated Sequencer (Invitrogen). PCR products were directly sequenced in forward and reverse directions. Detected mutations were confirmed by resequencing of tumor and matched normal blood DNA from new PCR product.

Pyrosequencing, a more sensitive method, was applied to quantitatively re-evaluate the cases with loss of TP53 or PIK3CA mutation after NCT as described previously (19, 20). Allele frequencies were outputted as relative percentages using the Biotage PSQ software.

Statistical analysis

Frequency tabulation and summary statistics were provided to characterize the data distribution. Student t test was used to compare continuous variables. To determine the combination of TP53 and PIK3CA mutation status, we defined the variable “TP53 and PIK3CA status.” The “TP53 and PIK3CA status” was considered wild-type only when neither TP53 nor PIK3CA mutation was detected; otherwise, it was classified as mutant. The McNemar test was applied to compare the change in mutation status following the treatment. The Cochran–Armitage trend test was used to test whether a change in mutation status was associated with clinical outcome in terms of the Miller–Payne Score. Survival curves were constructed using the Kaplan–Meier method, and the univariate survival difference was determined with the log-rank test. Adjusted HRs with 95% confidence intervals (CI) were calculated using Cox proportional hazards models. All statistical analyses were performed using Stata statistical software, version 10.0 (StataCorp). A two-sided P < 0.05 was considered statistically significant.

Exome sequencing in two cases for somatic mutation identification

We performed exome sequencing to examine the exomes (100-fold coverage) of two paired samples (pretreatment biopsies and posttreatment tumors), which were separate from the following cohorts. One pair (P1) was from a 62-year-old Chinese woman, and the other (P2) was from a 58-year-old Chinese woman, both of whom were diagnosed with histologically confirmed IDC and received four cycles of NCT (paclitaxel and carboplatin) before undergoing mastectomy (21, 22). Each had an approximately 50% reduction in bidimensionally measurable tumor diameter after NCT. Matched samples from biopsies and surgical resection were collected before and after NCT. Tumors were confirmed as basal-like with IHC phenotypes of ER, PR, HER2, CK5/6+, and EGFR+ by two independent pathologists in the Department of Pathology at FUSCC.

By comparing pre- and posttreatment tumor tissues, we identified 23 somatic SNVs in P1 and P2 (Supplementary Table S1). Among these variations, we identified and validated TP53 (c.C298T) and PIK3CA (c.A3140G) as significantly mutated genes only in the pretreatment tumors, not in the paired posttreatment samples, suggesting that TP53 and PIK3CA mutations might be lost in breast cancer after NCT. TP53 and PIK3CA have been identified as the most frequently mutated genes in breast cancer, each with a mutation frequency >35% (7). Thus, we supposed that loss of TP53 and PIK3CA mutations might be a common phenomenon after NCT.

TP53 and PIK3CA mutation shift in tumor tissues before and after NCT

We assessed the change in TP53 and PIK3CA mutation status in the first and second cohorts. In pretreatment tumor samples, the frequency of TP53 mutations was higher in both HER2+ and basal-like breast cancer, and PIK3CA mutations were basically associated with Luminal A, Luminal B, and HER2+ status (Supplementary Fig. S1). This was in concordance with the data from the Cancer Genome Atlas (TCGA; ref. 7) and gave confidence to the surrogate molecular subtyping using IHC markers. Both in Cohort 1 and 2, four patients had both PIK3CA and TP53 mutations in the pre-NCT samples, and two patients had both PIK3CA and TP53 mutations in the post-NCT samples. PIK3CA and TP53 mutations are not mutually exclusive in either cohort (Supplementary Table S2, Fisher exact test; in the first cohort, P = 0.769 for pre-NCT and P = 0.217 for post-NCT; in the second cohort, P = 0.749 for pre-NCT and P = 0.169 for post-NCT). All TP53 and PIK3CA somatic mutations identified before and after NCT are listed in Supplementary Table S3.

In the first cohort, 51 patients (24.8%) had TP53 or PIK3CA mutations before NCT; this decreased to 25 (12.1%) after treatment (Fig. 1A). Supplementary Figure S2 indicated high viable tumor cellularity in the post-NCT tumors where mutation loss occurred. We also reanalyzed the DNA samples from which mutation loss occurred using pyrosequencing, a more sensitive method, to quantitatively examine the loss of mutations. Compared with pretreatment samples, the posttreatment tumors with loss of PIK3CA or TP53 mutations had significantly decreased frequency (<10%) of mutant alleles (Supplementary Fig. S3). A total of 160 patients (77.7%) retained their initial status [145 wild-type (WT), 15 mutant (MT)]; 10 patients (4.8%) had a WT-to-MT change, and 36 patients (17.5%) had an MT-to-WT change. In the second cohort for validation, 42 patients (26.6%) had TP53 or PIK3CA mutations before NCT; this decreased to 20 (12.7%) after treatment (Fig. 1B). Eight patients (5.1%) had a WT-to-MT change, and 30 patients (19.0%) had an MT-to-WT change; the remaining 120 patients (75.9%) retained their initial status (108 WT, 12 MT). The decrease in mutation rate after chemotherapy was statistically significant in both cohorts (P < 0.001; McNemar test for both cohorts). Taken together, these results suggest that loss of TP53 and PIK3CA mutations is common in breast cancer tissues after NCT.

Figure 1.

TP53 and PIK3CA mutation shift in tumor tissues before and after NCT in the first (A, n = 206) and second cohorts (B, n = 158).

Figure 1.

TP53 and PIK3CA mutation shift in tumor tissues before and after NCT in the first (A, n = 206) and second cohorts (B, n = 158).

Close modal

We then evaluated the change in TP53 and PIK3CA mutation status separately (Supplementary Table S4 for the first cohort and Supplementary Table S5 for the second cohort). For TP53 mutations, the McNemar test revealed that the decrease in the mutation rate after chemotherapy was statistically significant in both cohorts (P = 0.013 for the first cohort and P = 0.035 for the second cohort). For PIK3CA mutations, a similar decrease in the mutation rate was also observed (P = 0.016 for the first cohort and P = 0.019 for the second cohort). Moreover, molecular subtype was not significantly associated with changes in TP53 and PIK3CA mutations (Supplementary Table S6, P = 0.902 for the first cohort and P = 0.915 for the second cohort). However, when separately analyzed, HER2+ and basal-like tumors were associated with a higher rate of TP53 mutation loss (Supplementary Table S7, P = 0.018) in the first cohort, but this was not successfully validated in the second cohort (Supplementary Table S7, P = 0.178). Molecular subtype was not significantly associated with PIK3CA mutation loss (Supplementary Table S8, P = 0.257 for the first cohort and P = 0.404 for the second cohort).

Influence of mutation shift on tumor response to NCT

We investigated the effect of mutation loss on tumor response to NCT. Patients with Miller–Payne grade 5 (recognized as pCR) were initially excluded from this study because they had no residual cancer tissue after NCT for analysis. Non-pCR patients were divided into four groups: MT-to-WT, WT-to-WT, MT-to-MT, and WT-to-MT. In the first cohort, 106 (51.4%) had significant pathologic partial responses (PR) defined as Miller–Payne grade 3 or 4 (Table 2). In the second cohort, 89 (56.3%) had significant PRs after NCT. Loss of mutation was significantly associated with a better pathologic response (P < 0.001 for both the first cohort and second cohort; Cochran–Armitage trend test). In addition, we examined the influence of TP53 (Supplementary Table S9) or PIK3CA (Supplementary Table S10) mutation loss separately on tumor response to NCT. Loss of TP53 mutation was significantly associated with better clinical response (P < 0.001 for the first cohort and P = 0.001 for the second cohort; Cochran–Armitage trend test). Loss of PIK3CA mutation was also significantly associated with a better pathologic response (P = 0.001 for the first cohort and P < 0.001 for the second cohort). These data reveal that loss of the TP53 and PIK3CA mutations is associated with improved clinical response to NCT.

Table 2.

Association of TP53 and PIK3CA mutation shift with tumor response to NCT in the first cohort (n = 206) and second cohort (n = 158)

Miller–Payne Score
SD/PD or 1234Total
CohortMutation changeN (%)N (%)N (%)N (%)N (%)
First MT-to-WT 3 (5.9) 5 (10.2) 9 (16.7) 19 (36.5) 36 (17.5) 
 WT-to-WT 32 (62.7) 40 (81.6) 42 (77.8) 32 (61.5) 145 (70.4) 
 MT-to-MT 10 (19.6) 2 (4.1) 2 (3.7) 1 (1.9) 15 (7.3) 
 WT-to-MT 6 (11.8) 2 (4.1) 1 (1.9) 1 (1.9) 10 (4.9) 
 Total 51 (100) 49 (100) 54 (100) 52 (100) 206 (100) 
       
Second MT-to-WT 1 (4.3) 5 (10.9) 9 (16.4) 15 (44.1) 30 (19.0) 
 WT-to-WT 10 (43.5) 37 (80.4) 43 (78.2) 18 (52.9) 108 (68.4) 
 MT-to-MT 7 (30.4) 2 (4.3) 2 (3.6) 1 (2.9) 12 (7.6) 
 WT-to-MT 5 (21.7) 2 (4.3) 1 (1.8) 0 (0) 8 (5.1) 
 Total 23 (100) 46 (100) 55 (100) 34 (100) 158 (100) 
Miller–Payne Score
SD/PD or 1234Total
CohortMutation changeN (%)N (%)N (%)N (%)N (%)
First MT-to-WT 3 (5.9) 5 (10.2) 9 (16.7) 19 (36.5) 36 (17.5) 
 WT-to-WT 32 (62.7) 40 (81.6) 42 (77.8) 32 (61.5) 145 (70.4) 
 MT-to-MT 10 (19.6) 2 (4.1) 2 (3.7) 1 (1.9) 15 (7.3) 
 WT-to-MT 6 (11.8) 2 (4.1) 1 (1.9) 1 (1.9) 10 (4.9) 
 Total 51 (100) 49 (100) 54 (100) 52 (100) 206 (100) 
       
Second MT-to-WT 1 (4.3) 5 (10.9) 9 (16.4) 15 (44.1) 30 (19.0) 
 WT-to-WT 10 (43.5) 37 (80.4) 43 (78.2) 18 (52.9) 108 (68.4) 
 MT-to-MT 7 (30.4) 2 (4.3) 2 (3.6) 1 (2.9) 12 (7.6) 
 WT-to-MT 5 (21.7) 2 (4.3) 1 (1.8) 0 (0) 8 (5.1) 
 Total 23 (100) 46 (100) 55 (100) 34 (100) 158 (100) 

NOTE: P < 0.001 for both the first and second cohorts (Cochran–Armitage trend test).

Influence of mutation shift on patient survival

We further questioned whether the loss of these mutations influenced DFS and OS. A striking association between loss of mutations in TP53 and PIK3CA and patient survival was observed in both cohorts (Fig. 2). Kaplan–Meier plots showed that mutation change status was associated with DFS (P = 0.033) and OS (P = 0.045) in the first cohort, and this was likely due to the better survival for MT-to-WT group. This observation was supported by similar results from the second cohort (P = 0.038 for DFS and P = 0.058 for OS). However, neither TP53 (Supplementary Fig. S4) nor PIK3CA (Supplementary Fig. S5) mutation loss alone was significantly associated with DFS, which might be attributable to the limited number of cases with TP53 or PIK3CA mutation loss separately.

Figure 2.

Kaplan–Meier estimates of DFS and OS according to TP53 and PIK3CA mutation shift in the first (A, n = 206) and second cohorts (B, n = 158).

Figure 2.

Kaplan–Meier estimates of DFS and OS according to TP53 and PIK3CA mutation shift in the first (A, n = 206) and second cohorts (B, n = 158).

Close modal

Furthermore, we used a multivariate Cox proportional hazards regression model accounting for age, menopausal status, tumor size, lymph node status, grade, molecular subtype, pathologic response, and TP53 and PIK3CA status to identify independent predictors of DFS in these two cohorts (Table 3). In the first cohort, we observed that node status, pathologic response, PIK3CA status, and TP53 and PIK3CA mutation status were independent predictors of DFS after multivariate adjustment (Supplementary Tables S11 and S12). Compared with the MT-to-WT group, the rest (WT-to-WT, WT-to-MT, and MT-to-MT) were significantly associated with worse DFS (HR, 1.96; 95% CI, 1.12–4.88; P = 0.031; Table 3). In the second cohort, the adjusted multivariate regression revealed that tumor size, node status, pathologic response, PIK3CA status, and TP53 and PIK3CA mutation status were independent predictors of DFS (Supplementary Tables S11 and S12). The rest (WT-to-WT, WT-to-MT, and MT-to-MT) was also significantly associated with poorer DFS (HR, 1.68; 95% CI, 1.06–3.18; P = 0.042; Table 3) relative to an MT-to-WT change. Compared with TP53 or PIK3CA status alone, the combination of TP53 and PIK3CA mutation status was a more robust independent prognostic factor of DFS in both cohorts. Thus, loss of mutations in TP53 and PIK3CA is most likely associated with improved survival.

Table 3.

Multivariate analysis of disease-free survival for patients who underwent NCT in the first cohort (n = 206) and second cohort (n = 158)

Cohort 1 (n = 206)Cohort 2 (n = 158)
VariableaHR (95% CI)PHR (95% CI)P
TP53 Status 
 MT-to-WT 1.00 (Ref.) 0.485 1.00 (Ref.) 0.644 
 Restb 1.45 (0.91–3.16)  1.17 (0.86–2.73)  
 
PIK3CA Status 
 MT-to-WT 1.00 (Ref.) 0.322 1.00 (Ref.) 0.409 
 Restb 1.68 (0.59–5.20)  1.57 (0.89–2.59)  
 
TP53 and PIK3CA Statusc 
 MT-to-WT 1.00 (Ref.) 0.031 1.00 (Ref.) 0.042 
 Restb 1.96 (1.12–4.88)  1.68 (1.06–3.18)  
Cohort 1 (n = 206)Cohort 2 (n = 158)
VariableaHR (95% CI)PHR (95% CI)P
TP53 Status 
 MT-to-WT 1.00 (Ref.) 0.485 1.00 (Ref.) 0.644 
 Restb 1.45 (0.91–3.16)  1.17 (0.86–2.73)  
 
PIK3CA Status 
 MT-to-WT 1.00 (Ref.) 0.322 1.00 (Ref.) 0.409 
 Restb 1.68 (0.59–5.20)  1.57 (0.89–2.59)  
 
TP53 and PIK3CA Statusc 
 MT-to-WT 1.00 (Ref.) 0.031 1.00 (Ref.) 0.042 
 Restb 1.96 (1.12–4.88)  1.68 (1.06–3.18)  

NOTE: Bold values indicate statistical significance.

Abbreviation: Ref., reference.

aAdjusted by Cox proportional hazards models including age, menopausal status, tumor size, node status, tumor grade, molecular subtype, and pathologic response.

bCases with WT-to-WT, WT-to-MT, or MT-to-MT changes.

cTo determine the combination of TP53 and PIK3CA mutation status, we defined the variable “TP53 and PIK3CA status.” The “TP53 and PIK3CA status” was considered WT only when neither TP53 nor PIK3CA mutation was detected; otherwise, it was classified as MT.

Intratumoral heterogeneity of mutations in tissue samples

The phenotype of mutation shift prompted us to investigate its biologic nature. We further performed genetic heterogeneity analysis on samples from 81 patients with histologically confirmed IDC who had undergone prior surgical resection and received chemotherapy. Twenty-two cases were TP53/PIK3CA mutation positive. Of these 22 cases, a total of 748 tumor foci were fractionated. Samples from four of the 22 cases consisted of only MT cells, whereas the other 18 samples consisted of both WT and MT cells. Among the 59 cases identified in routine detection as TP53 and PIK3CA WT by Sanger sequencing, 1,880 tumor foci were fractionated for analysis of TP53 and PIK3CA mutations. Five of the 59 cases consisted of both WT and MT cells. Taken together, 28.4% of the samples (23 of 81) had intratumoral heterogeneity in TP53 or PIK3CA mutation. In Fig. 3, we present typical images of tissues with TP53 mutation (p.R175H) and PIK3CA mutation (p.H1047R) heterogeneity. The remaining 71.6% of the samples (58/81) contained only TP53/PIK3CA mutation–positive cells or only WT cells. These results reveal high intratumoral heterogeneity of TP53 and PIK3CA in prechemotherapy tumors and partially explain the genetic basis of mutation shift after chemotherapy.

Figure 3.

Multiple foci microdissection and analysis of intratumoral heterogeneity of TP53 and PIK3CA somatic mutations in the third cohort (n = 81). The arrows indicate somatic mutations. A, representative images of tissue with TP53 mutation heterogeneity (p.R175H). B, representative images of tissue with PIK3CA mutation heterogeneity (p.H1047R). Scale bar, 50 μm.

Figure 3.

Multiple foci microdissection and analysis of intratumoral heterogeneity of TP53 and PIK3CA somatic mutations in the third cohort (n = 81). The arrows indicate somatic mutations. A, representative images of tissue with TP53 mutation heterogeneity (p.R175H). B, representative images of tissue with PIK3CA mutation heterogeneity (p.H1047R). Scale bar, 50 μm.

Close modal

To our knowledge, our study is the first to suggest that NCT influences mutation status in tissue samples and to demonstrate that significant loss of mutations in TP53 and PIK3CA can occur during NCT. We highlighted the positive prognostic impact of loss of mutations in TP53 and PIK3CA under NCT in breast cancer. The decreases in the TP53 and PIK3CA mutation rates were significantly associated with better pathologic response (higher Miller–Payne score) and improved survival.

Compared with data released by TCGA (7), our data showed lower mutation rates and different mutation spectra of TP53 and PIK3CA in the pretreatment tumors in Chinese population, indicating that there might be some difference in the mutational evolution of breast cancer in different populations. Since few Asian (or Chinese) patients were included in the TCGA cohort, it does raise the question of ethnic diversity for future research. As previously reported, the mutation rates and spectra of TP53 and PIK3CA are quite different among different populations. For PIK3CA mutations in Chinese population, the mutation rates were 12.3% (7/57) reported by Wang and colleagues (23) and 7.5% (9/120) reported by Tong and colleagues (24). For TP53 mutations, the frequency of mutation reported by Hao and colleagues (25) was 22.7% (10/44), which was consistent with the results from Fukushima and colleagues (26), who reported 20.0% (10/50) of the tumors harboring TP53 mutations in the Japanese population. Furthermore, the strength of our study is the reproducibility of the finding in an independent validation sample set. Taken together, our findings, which were in accordance with the previous reports by other Chinese or Asian investigators, revealed unique mutation rates and spectra of TP53 and PIK3CA in Chinese population.

The shift in tumors from mutation status to wild-type status observed in this study suggests that both mutant and nonmutant cancer cells coexist in the same tumor. To identify intratumoral heterogeneity, we microdissected and analyzed mutation status in more than 2,500 tumor foci of 81 tumors from patients with IDC who underwent prior surgery before chemotherapy, and 28.4% of tumors contained both mutant and wild-type foci. These findings are in accordance with the previous reports by Hernandez and colleagues (27) and Kalinsky and colleagues (28).

TP53 and PIK3CA are the most commonly mutated genes, and they play pivotal roles in the biology of breast cancer (7). The p53 tumor suppressor protein (encoded by TP53) is involved in the cell cycle, checkpoint control, repair of DNA damage, and apoptosis (29–31). PIK3CA, encoding the catalytic subunit of phosphoinositide 3-kinase, regulates cell survival, apoptosis, proliferation, motility, growth, and cytoskeletal rearrangement (32–34). Notably, the majority of mutation changes after chemotherapy were from mutant status to wild-type, suggesting that cancer cells harboring certain TP53 and PIK3CA mutations might be more sensitive to chemotherapy than those without mutation. We further analyzed the relationships between the shift of mutation status and tumor response to NCT. We found that patients with a MT-to-WT shift in TP53 and PIK3CA status were more likely to achieve a high Miller–Payne score after chemotherapy; they also had higher DFS rate relative to patients with no change or the opposite change. When considered with previous studies (35, 36), our investigation suggests that mutation shift may be related to the heterogeneity of intratumoral mutations and to different chemosensitivity levels of mutant and wild-type cells. Both mutant and wild-type cancer cells coexist in untreated tumors with TP53 or PIK3CA mutations. For patients who had loss of mutations after NCT, we hypothesize that their cancer cells with certain TP53 or PIK3CA driver mutations might be sensitive to chemotherapeutic agents. Therefore, cytotoxic therapy would primarily remove the mutant cells and consequently shift the evolutionary landscape in favor of the nonmutant subclones. Thus, when analyzing the residual tumors using Sanger sequencing, the percentage of mutant alleles would be too low to detect, and the genotypes of the tumors would be considered to have changed from MT to WT. These patients generally had smaller residual tumors, benefited more from adjuvant chemotherapy and had relatively favorable prognosis. In contrast, for patients who underwent MT-to-MT change or WT-to-MT change after NCT, their cancer cells with TP53 or PIK3CA driver mutations might be resistant to chemotherapeutic agents. In the residual tumors, the mutant alleles would remain or become enriched. Consequently, these patients benefited less from chemotherapy and had relatively poor prognosis. Interestingly, in the third cohort, five patients (6.2%) had a low frequency and abundance of TP53/PIK3CA mutations, which were detected only through microsamples, thus explaining in part why the mutation status of some patients shifted from negative to positive after chemotherapy.

One of the limitations of the current study is the lack of sufficient pre-NCT tumor tissue for microdissection and mutation analysis. We therefore used the third cohort for the analysis of intratumoral heterogeneity of TP53 and PIK3CA mutations, and the results from the three interconnecting cohorts provide a solid foundation to support our conclusions. In addition, more research is required to elucidate the mechanism by which loss of mutations in TP53 and PIK3CA regulates chemosensitivity. The regimen used in these patients was a combination of taxanes and platinum. In fact, both taxanes (mitotic inhibitors and antimicrotubule agents) and platinum (DNA-damaging agents) might be relevant for the mutation loss. However, the precise mechanism of mutation loss is beyond the current study. Further studies are warranted to explore molecular mechanisms resulting in the chemotherapy-related loss of mutations in TP53 and PIK3CA.

In conclusion, chemotherapy may affect somatic mutation status in patients with breast cancer, lowering the number of TP53 and PIK3CA mutations. This may be attributable to the heterogeneity of intratumoral mutations and the different sensitivities of mutant and wild-type tumor cells to chemotherapy. These findings might help to optimize the choice of sequential therapy and improve patient survival. To enable the selection of the most effective therapeutic option, additional preclinical and clinical studies based on our findings are needed to further examine drug resistance mechanisms and to validate our findings in other independent cohorts.

No potential conflicts of interest were disclosed.

Conception and design: Y.-Z. Jiang, K.-D. Yu, J. Bao, Z.-M. Shao

Development of methodology: Y.-Z. Jiang, K.-D. Yu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-Z. Jiang, J. Bao, W.-T. Peng

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-Z. Jiang, K.-D. Yu, J. Bao, W.-T. Peng

Writing, review, and/or revision of the manuscript: Y.-Z. Jiang, K.-D. Yu, J. Bao, W.-T. Peng, Z.-M. Shao

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.-D. Yu, Z.-M. Shao

Study supervision: K.-D. Yu, Z.-M. Shao

The authors thank Jiong Wu, Jin-Song Lu, Guang-Yu Liu, Gen-Hong Di, and Zhen-Zhou Shen for excellent data handling.

This work was supported by National Natural Science Foundation of China (81370075, 81001169), Training Plan of Excellent Talents in Shanghai Municipality Health System (K.-D. Yu), Research and Innovation Project of Shanghai Municipal Education Commission (2014; K.-D. Yu), Shanghai International Science and Technique Cooperation Foundation (12410707700), International S&T Cooperation Program of China (ISTCP No. 09), and the Shanghai Key Laboratory of Breast Cancer (12DZ2260100).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Ellis
MJ
,
Ding
L
,
Shen
D
,
Luo
J
,
Suman
VJ
,
Wallis
JW
, et al
Whole-genome analysis informs breast cancer response to aromatase inhibition
.
Nature
2012
;
486
:
353
60
.
2.
Modlich
O
,
Prisack
HB
,
Munnes
M
,
Audretsch
W
,
Bojar
H
. 
Immediate gene expression changes after the first course of neoadjuvant chemotherapy in patients with primary breast cancer disease
.
Clin Cancer Res
2004
;
10
:
6418
31
.
3.
Buchholz
TA
,
Stivers
DN
,
Stec
J
,
Ayers
M
,
Clark
E
,
Bolt
A
, et al
Global gene expression changes during neoadjuvant chemotherapy for human breast cancer
.
Cancer J
2002
;
8
:
461
8
.
4.
Lee
SC
,
Xu
X
,
Lim
YW
,
Iau
P
,
Sukri
N
,
Lim
SE
, et al
Chemotherapy-induced tumor gene expression changes in human breast cancers
.
Pharmacogenet Genomics
2009
;
19
:
181
92
.
5.
Shah
SP
,
Roth
A
,
Goya
R
,
Oloumi
A
,
Ha
G
,
Zhao
Y
, et al
The clonal and mutational evolution spectrum of primary triple-negative breast cancers
.
Nature
2012
;
486
:
395
9
.
6.
Stephens
PJ
,
Tarpey
PS
,
Davies
H
,
Van Loo
P
,
Greenman
C
,
Wedge
DC
, et al
The landscape of cancer genes and mutational processes in breast cancer
.
Nature
2012
;
486
:
400
4
.
7.
Cancer Genome Atlas Network
. 
Comprehensive molecular portraits of human breast tumours
.
Nature
2012
;
490
:
61
70
.
8.
Ogston
KN
,
Miller
ID
,
Payne
S
,
Hutcheon
AW
,
Sarkar
TK
,
Smith
I
, et al
A new histological grading system to assess response of breast cancers to primary chemotherapy: prognostic significance and survival
.
Breast
2003
;
12
:
320
7
.
9.
Silver
DP
,
Richardson
AL
,
Eklund
AC
,
Wang
ZC
,
Szallasi
Z
,
Li
Q
, et al
Efficacy of neoadjuvant Cisplatin in triple-negative breast cancer
.
J Clin Oncol
2010
;
28
:
1145
53
.
10.
Kennecke
H
,
Yerushalmi
R
,
Woods
R
,
Cheang
MC
,
Voduc
D
,
Speers
CH
, et al
Metastatic behavior of breast cancer subtypes
.
J Clin Oncol
2010
;
28
:
3271
7
.
11.
Voduc
KD
,
Cheang
MC
,
Tyldesley
S
,
Gelmon
K
,
Nielsen
TO
,
Kennecke
H
. 
Breast cancer subtypes and the risk of local and regional relapse
.
J Clin Oncol
2010
;
28
:
1684
91
.
12.
Goldhirsch
A
,
Wood
WC
,
Coates
AS
,
Gelber
RD
,
Thurlimann
B
,
Senn
HJ
. 
Strategies for subtypes–dealing with the diversity of breast cancer: highlights of the St. Gallen international expert consensus on the primary therapy of early breast cancer 2011
.
Ann Oncol
2011
;
22
:
1736
47
.
13.
Yu
KD
,
Huang
AJ
,
Fan
L
,
Li
WF
,
Shao
ZM
. 
Genetic variants in oxidative stress-related genes predict chemoresistance in primary breast cancer: a prospective observational study and validation
.
Cancer Res
2012
;
72
:
408
19
.
14.
Yan
XJ
,
Xu
J
,
Gu
ZH
,
Pan
CM
,
Lu
G
,
Shen
Y
, et al
Exome sequencing identifies somatic mutations of DNA methyltransferase gene DNMT3A in acute monocytic leukemia
.
Nat Genet
2011
;
43
:
309
15
.
15.
Li
Y
,
Vinckenbosch
N
,
Tian
G
,
Huerta-Sanchez
E
,
Jiang
T
,
Jiang
H
, et al
Resequencing of 200 human exomes identifies an excess of low-frequency non-synonymous coding variants
.
Nat Genet
2010
;
42
:
969
72
.
16.
Zang
ZJ
,
Cutcutache
I
,
Poon
SL
,
Zhang
SL
,
McPherson
JR
,
Tao
J
, et al
Exome sequencing of gastric adenocarcinoma identifies recurrent somatic mutations in cell adhesion and chromatin remodeling genes
.
Nat Genet
2012
;
44
:
570
4
.
17.
Jiang
YZ
Yu
KD
Peng
WT
Di
GH
Wu
J
Liu
GY
et al 
Enriched variations in TEKT4 and breast cancer resistance to paclitaxel
.
Nat Commun
2014
;
5
:
3802
.
18.
Li
R
,
Yu
C
,
Li
Y
,
Lam
TW
,
Yiu
SM
,
Kristiansen
K
, et al
SOAP2: an improved ultrafast tool for short read alignment
.
Bioinformatics
2009
;
25
:
1966
7
.
19.
Lai
LA
,
Kostadinov
R
,
Barrett
MT
,
Peiffer
DA
,
Pokholok
D
,
Odze
R
, et al
Deletion at fragile sites is a common and early event in Barrett's esophagus
.
Mol Cancer Res
2010
;
8
:
1084
94
.
20.
Amos-Landgraf
JM
,
Kwong
LN
,
Kendziorski
CM
,
Reichelderfer
M
,
Torrealba
J
,
Weichert
J
, et al
A target-selected Apc-mutant rat kindred enhances the modeling of familial human colon cancer
.
Proc Natl Acad Sci U S A
2007
;
104
:
4036
41
.
21.
Chen
XS
,
Nie
XQ
,
Chen
CM
,
Wu
JY
,
Wu
J
,
Lu
JS
, et al
Weekly paclitaxel plus carboplatin is an effective nonanthracycline-containing regimen as neoadjuvant chemotherapy for breast cancer
.
Ann Oncol
2010
;
21
:
961
7
.
22.
Yu
KD
,
Liu
GY
,
Zhou
XY
,
Zhou
Y
,
Wu
J
,
Chen
CM
, et al
Association of HER-2 copy number and HER-2/CEP-17 ratio with neoadjuvant taxane-containing chemotherapy sensitivity in locally advanced breast cancer
.
Oncologist
2012
;
17
:
792
800
.
23.
Wang
L
,
Zhang
Q
,
Zhang
J
,
Sun
S
,
Guo
H
,
Jia
Z
, et al
PI3K pathway activation results in low efficacy of both trastuzumab and lapatinib
.
BMC Cancer
2011
;
11
:
248
.
24.
Tong
L
,
Yang
XX
,
Liu
MF
,
Yao
GY
,
Dong
JY
,
Ye
CS
, et al
Mutational analysis of key EGFR pathway genes in Chinese breast cancer patients
.
Asian Pac J Cancer Prev
2012
;
13
:
5599
603
.
25.
Hao
XD
,
Yang
Y
,
Song
X
,
Zhao
XK
,
Wang
LD
,
He
JD
, et al
Correlation of telomere length shortening with TP53 somatic mutations, polymorphisms and allelic loss in breast tumors and esophageal cancer
.
Oncol Rep
2013
;
29
:
226
36
.
26.
Fukushima
T
,
Onda
M
,
Abe
R
,
Otake
T
,
Kimijima
I
,
Tsuchiya
A
. 
p53 mutations and overexpressions in Japanese breast cancer
.
Eur J Surg Oncol
1995
;
21
:
595
600
.
27.
Hernandez
L
,
Wilkerson
PM
,
Lambros
MB
,
Campion-Flora
A
,
Rodrigues
DN
,
Gauthier
A
, et al
Genomic and mutational profiling of ductal carcinomas in situ and matched adjacent invasive breast cancers reveals intra-tumour genetic heterogeneity and clonal selection
.
J Pathol
2012
;
227
:
42
52
.
28.
Kalinsky
K
,
Heguy
A
,
Bhanot
UK
,
Patil
S
,
Moynahan
ME
. 
PIK3CA mutations rarely demonstrate genotypic intratumoral heterogeneity and are selected for in breast cancer progression
.
Breast Cancer Res Treat
2011
;
129
:
635
43
.
29.
Kastan
MB
,
Bartek
J
. 
Cell-cycle checkpoints and cancer
.
Nature
2004
;
432
:
316
23
.
30.
Brosh
R
,
Rotter
V
. 
When mutants gain new powers: news from the mutant p53 field
.
Nat Rev Cancer
2009
;
9
:
701
13
.
31.
Lehmann
BD
,
Pietenpol
JA
. 
Targeting mutant p53 in human tumors
.
J Clin Oncol
2012
;
30
:
3648
50
.
32.
Barbareschi
M
,
Buttitta
F
,
Felicioni
L
,
Cotrupi
S
,
Barassi
F
,
Del
GM
, et al
Different prognostic roles of mutations in the helical and kinase domains of the PIK3CA gene in breast carcinomas
.
Clin Cancer Res
2007
;
13
:
6064
9
.
33.
Shaw
RJ
,
Cantley
LC
. 
Ras, PI(3)K and mTOR signalling controls tumour cell growth
.
Nature
2006
;
441
:
424
30
.
34.
Bader
AG
,
Kang
S
,
Zhao
L
,
Vogt
PK
. 
Oncogenic PI3K deregulates transcription and translation
.
Nat Rev Cancer
2005
;
5
:
921
9
.
35.
Bai
H
,
Wang
Z
,
Chen
K
,
Zhao
J
,
Lee
JJ
,
Wang
S
, et al
Influence of chemotherapy on EGFR mutation status among patients with non-small-cell lung cancer
.
J Clin Oncol
2012
;
30
:
3077
83
.
36.
Mok
TS
,
Wu
YL
,
Thongprasert
S
,
Yang
CH
,
Chu
DT
,
Saijo
N
, et al
Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma
.
N Engl J Med
2009
;
361
:
947
57
.

Supplementary data