Distributions of the clinical characteristics in breast cancer patients
The study included 234 breast cancer patients with 157 (67.1%) chemosensitive and 77 (32.9%) chemoresistant. Figure 1A is the imagological examination data of chemosensitivity and chemoresistance. Table 1 showed the clinical characteristics and statistical information of 234 patients. All participants are female, with the median age of 48 years. According to American Joint Committee on Cancer (AJCC) Clinical Stage, most patients were diagnosed with stage II or stage III breast cancer. Pathological examination revealed 222 cases of invasive ductal carcinoma (94.9%). There were statistical differences in Histological type, PR, and Molecular Subtype between the sensitive group and the resistant group (P<0.05), and no significant difference was found in the other variables between the two groups (P>0.05).
Table 1
Characteristics of study subjects.
Characteristics | Total | Chemosensitive | Chemoresistant | P |
| 234 | 157 (0.671) | 77 (0.329) | |
Age (years) | | | | 0.244 |
<48 | 113 (0.483) | 80 (0.510) | 33 (0.429) | |
≥48 | 121 (0.517) | 77 (0.490) | 44 (0.571) | |
Menopausal state | | | | 0.421 |
Menopause | 68 (0.291) | 43 (0.274) | 25 (0.325) | |
Premenopausal | 166 (0.709) | 114 (0.726) | 52 (0.675) | |
AJCC Clinical Stage | | | | 0.395 |
II | 104 (0.444) | 64 (0.408) | 40 (0.519) | |
IIIA | 51 (0.218) | 38 (0.242) | 13 (0.169) | |
IIIB | 65 (0.278) | 45 (0.287) | 20 (0.260) | |
IIIC+ IV | 14 (0.060) | 10 (0.064) | 4 (0.052) | |
Histological type | | | | 0.011* |
Invasive ductal carcinoma | 222 (0.949) | 153 (0.975) | 69 (0.896) | |
Other types | 12 (0.051) | 4 (0.025) | 8 (0.104) | |
Tumor size | | | | 0.455 |
T1(size≤2) | 64 (0.274) | 39 (0.248) | 25 (0.325) | |
T2 (2༜size≤5) | 94 (0.402) | 66 (0.420) | 28 (0.364) | |
T3(size༞5) | 76 (0.325) | 52 (0.331) | 24 (0.312) | |
Lymph node staging | | | | 0.981 |
N0 | 79 (0.338) | 53 (0.338) | 26 (0.338) | |
N1 | 81 (0.346) | 55 (0.350) | 26 (0.338) | |
N2 | 58 (0.248) | 39 (0.248) | 19 (0.247) | |
N3 | 16 (0.068) | 10 (0.064) | 6 (0.078) | |
ER | | | | 0.181 |
Negative | 124 (0.530) | 88 (0.561) | 36 (0.468) | |
Positive | 110 (0.470) | 69 (0.439) | 41 (0.532) | |
PR | | | | 0.015* |
Negative | 70 (0.299) | 55 (0.350) | 15 (0.195) | |
Positive | 164 (0.701) | 102 (0.650) | 62 (0.805) | |
C-erbB-2a | | | | 0.830 |
Negative | 32 (0.137) | 22 (0.140) | 10 (0.130) | |
Positive | 202 (0.863) | 135 (0.860) | 67 (0.870) | |
Molecular Subtype | | | | 0.041* |
Luminal A | 50 (0.214) | 29 (0.185) | 21 (0.273) | |
Luminal B | 131 (0.560) | 98 (0.624) | 33 (0.429) | |
HER2 overexpression | 19 (0.081) | 10 (0.064) | 9 (0.117) | |
Triple-negative | 34 (0.145) | 20 (0.127) | 14 (0.182) | |
a C-erbB-2-positive: HER2 (+++) or Fish (+); ER, estrogen receptor; PR, progesterone receptor; C-erbB-2, human epidermal growth factor receptor 2. |
*: P value<0.05 was considered as significant. |
Statistical Analysis Of Four Snps Genotyping Results
MassArray genotyping was performed on the rs786204926, rs701848, rs12402181, rs35770269 polymorphism of 234 patients. We analysed the genotype frequency of SNP, rs786204926, rs701848, rs35770269 accords with the HW equilibrium state (Table 2). We further statistically analysed the genotype results and compared the distribution of genotypes under different genetic models. The results were listed in Table 3. In the heterozygote model, the AG genotype was significantly different from the AA genotype (OR=0.497, 95% CI=0.265-0.933, P=0.002). Under the Dominant model, the AG+GG genotype was significantly different from the AA genotype (OR=0.472, 95% CI=0.263-0.848, P=0.011). In the allele analysis, there was a statistical difference between alleles G and A (OR=0.522, 95% CI=0.320-0.850, P=0.008), Carrying the G allele or AG genotype will increase the risk of chemosensitivity in breast cancer. We further subgroup analysis showed that rs786204926 was associated with age (≥48), premenopausal, AJCC Clinical Stage II, molecular subtype (HER2 overexpression, triple-negative) in breast cancer chemoresistance (Supplementary Table 3). There is no correlation between rs701848, rs12402181, rs35770269 and breast cancer chemosensitivity. We randomly selected some samples to verify the genotyping results using Sanger sequencing method. Figure 1B showed the image of PCR products of some samples after agarose gel electrophoresis. The wild homozygous, heterozygous, and mutant homozygous sequence of the rs786204926 polymorphism were shown in Figure 1C.
Table 2
The test of Hardy-Weinberg Equilibrium of four SNPs.
SNP | Group | Genotype | P |
11 | 12 | 22 |
rs786204926 | Chemosensitive | 85 (54.1%) | 56 (35.7%) | 16 (10.2%) | 0.147 |
| Chemoresistant | 55 (71.4%) | 18 (23.4%) | 4 (5.2%) | 0.143 |
rs701848 | Chemosensitive | 57 (36.9%) | 67 (43.3%) | 33 (19.7%) | 0.114 |
| Chemoresistant | 33 (40.3%) | 30 (44.2%) | 14 (15.6%) | 0.135 |
rs12402181 | Chemosensitive | 60 (38.2%) | 89 (56.7%) | 8 (5.1%) | 0.001 |
| Chemoresistant | 33 (42.9%) | 35 (45.5%) | 9 (11.7%) | 0.952 |
rs35770269 | Chemosensitive | 54 (34.4%) | 73 (46.5%) | 30 (19.1%) | 0.549 |
| Chemoresistant | 34 (44.2%) | 34 (44.2%) | 9 (11.7%) | 0.910 |
Table 3
The association between PTEN gene polymorphisms and the risk of breast cancer chemotherapy resistance.
SNP | Model | Allele/Genotype | Chemosensitive | Chemoresistant | OR (95%CI) | P |
rs786204926 | Heterozygote | AA | 85 (0.541) | 55 (0.714) | | |
| | AG | 56 (0.357) | 18 (0.234) | 0.497 (0.265-0.933) | 0.028 |
| Homozygote | AA | 85 (0.541) | 55 (0.714) | | |
| | GG | 16 (0.102) | 4 (0.052) | 0.386 (0.123-1.216) | 0.136* |
| Dominant | AA | 85 (0.541) | 55 (0.714) | | |
| | AG+GG | 72 (0.459) | 22 (0.286) | 0.472 (0.263-0.848) | 0.011 |
| Recessive | GG | 16 (0.102) | 4 (0.052) | | |
| | AG+AA | 141 (0.898) | 73 (0.948) | 2.071 (0.668-6.421) | 0.225* |
| Additive | AG | 56 (0.357) | 18 (0.234) | | |
| | AA+GG | 101 (0.643) | 59 (0.766) | 1.817 (0.977-3.380) | 0.057 |
| Allele | A | 226 (0.720) | 128 (0.831) | | |
| | G | 88 (0.280) | 26 (0.169) | 0.522 (0.320-0.850) | 0.008 |
rs701848 | Heterozygote | TT | 57 (0.460) | 33 (0.524) | | |
| | TC | 67 (0.540) | 30 (0.476) | 0.773 (0.421-1.420) | 0.407 |
| Homozygote | TT | 57 (0.633) | 33 (0.702) | | |
| | CC | 33 (0.367) | 14 (0.298) | 0.733 (0.343-1.564) | 0.421 |
| Dominant | TT | 57 (0.363) | 33 (0.429) | | |
| | TC+CC | 100 (0.637) | 44 (0.571) | 0.760 (0.436-1.326) | 0.333 |
| Recessive | CC | 33 (0.210) | 14 (0.182) | | |
| | TC+TT | 124 (0.790) | 63 (0.818) | 1.198 (0.598-2.399) | 0.611 |
| Additive | TC | 67 (0.427) | 30 (0.390) | | |
| | TT+CC | 90 (0.573) | 47 (0.610) | 1.166 (0.668-2.035) | 0.588 |
| Allele | T | 181 (0.576) | 96 (0.623) | | |
| | C | 133(0.424) | 58 (0.377) | 0.822 (0.554-1.221) | 0.332 |
rs12402181 | Heterozygote | GG | 60 (0.403) | 33 (0.485) | | |
| | GA | 89 (0.597) | 35 (0.515) | 0.715 (0.401-1.274) | 0.254 |
| Homozygote | GG | 60 (0.882) | 33 (0.786) | | |
| | AA | 8 (0.118) | 9 (0.214) | 2.045 (0.721-5.803) | 0.173 |
| Dominant | GG | 60 (0.382) | 33 (0.429) | | |
| | GA+AA | 97 (0.618) | 44 (0.571) | 0.825 (0.474-1.436) | 0.495 |
| Recessive | AA | 8 (0.051) | 9 (0.117) | | |
| | GA+GG | 149 (0.949) | 68 (0.883) | 0.406 (0.150-1.097) | 0.068 |
| Additive | GA | 89 (0.567) | 35 (0.455) | | |
| | GG+AA | 68 (0.433) | 42 (0.545) | 1.571 (0.907-2.718) | 0.106 |
| Allele | G | 209 (0.666) | 101 (0.656) | | |
| | A | 105 (0.334) | 53 (0.344) | 1.045 (0.695-1.569) | 0.834 |
rs35770269 | Heterozygote | AA | 54 (0.425) | 34 (0.500) | | |
| | AT | 73 (0.575) | 34 (0.500) | 0.740 (0.409-1.336) | 0.317 |
| Homozygote | AA | 54 (0.643) | 34 (0.791) | | |
| | TT | 30 (0.357) | 9 (0.209) | 0.476 (0.202-1.126) | 0.087 |
| Dominant | AA | 54 (0.344) | 34 (0.442) | | |
| | AT+TT | 103 (0.656) | 43 (0.558) | 0.663 (0.380-1.158) | 0.148 |
| Recessive | TT | 30 (0.191) | 9 (0.117) | | |
| | AT+AA | 127 (0.809) | 68 (0.883) | 1.785 (0.801-3.976) | 0.152 |
| Additive | AT | 73 (0.465) | 34 (0.442) | | |
| | AA+TT | 84 (0.535) | 43 (0.558) | 1.099 (0.635-1.902) | 0.736 |
| Allele | A | 181 (0.576) | 102 (0.662) | | |
| | T | 133 (0.424) | 52 (0.338) | 0.694 (0.464-1.037) | 0.074 |
OR, Odds ratio; CI, Confidence interval. |
*: Pass Fisher's exact test calculation. |
P value <0.05 was considered as significant. |
Logistic Regression Analysis
The chemoresistance was used as the dependent variable, and related factors (sex, age, lymph node metastasis, rs786204926 genotype, etc.) were used as independent variables. Logistic regression analysis showed that age (P=0.000) was risk factors for breast cancer chemoresistance. lymph node staging (P=0.023) and rs786204926 AA/GG/AG genotype (P=0.000) were protective factors for breast cancer chemoresistance (Figure 1D).
Linkage Disequilibrium (Ld) And Haplotype Analysis
SHEsis software was used to analyse the LD of the two polymorphisms of PTEN. Complete LD was detected in the polymorphisms of rs701848 and rs786204926 (r2=0.222, D'=1.000) (Figure 2A, 2B and 2C). At the same time, haplotype analysis was performed, the chemosensitive and chemoresistant groups with haplotype estimated frequency less than 3% were excluded from further analysis. We found that TA, TG haplotypes (rs701848, rs786204926) were related to chemosensitivity. Carrying TA might increase the risk of chemoresistance. Carrying TG might decrease the risk of chemoresistance. Other haplotypes were not significantly different between the chemosensitive and chemoresistant groups (Figure 2D).
rs786204926 affects a new PTEN isoform expression association with drug sensitivity
According to The cBioPortal for Cancer Genomics (http://cbioportal.org) database analysis of samples with mutation data in invasive breast cancer (TCGA, Firehose Legacy) [33, 34], PTEN had mutations in 9% of patients (Figure 3A), and mutations in PTEN are associated with overall survival (Figure 3C). In order to study the correlation between the PTEN mutant genome and epirubicin drug sensitivity in breast cancer, we used GDSC online software for analysis. The results showed that PTEN mutant genome would increase the sensitivity of epirubicin drug in breast cancer (Figure 3C). The database had reported that there are two splice sites (X70_splice, X212_splice) on the PTEN gene. We predicted that the rs786204926 is also a splice site (Figure 3D). Mutations at the splicing site might affect AS of genes, and these effects had been detected by Alamut Visual v.2.15 and HSF. HSF predicted whether the rs786204926A>G mutation affected AS. The prediction results showed that rs786204926 is a splice acceptor site, when carrying G allele, it will destroy the wild-type acceptor site and affect splicing (Supplementary Table 4). Further using Alamut Visual v.2.14 showed that the rs786204926A>G mutation destroyed the acceptor site and created a new acceptor site. It also showed the changes between the exon splicing enhancers (ESE) after mutation, the results showed that the ESE binding ability is weakened after mutation (Figure 3E).
Higher levels of new isoforms of PTEN in chemosensitive samples of breast cancer
We randomly selected 6 breast cancer tissues after chemotherapy (2 GG types, 2 AG types, and 2 AA types). Perform RT-PCR amplification on the sample, and agarose gel electrophoresis on the products and sequence verification, the results showed that PTEN wild type had no difference among alleles (Figure 4). In PTEN mutant isoforms, AG and AA genotypes are statistically significant compared with GG alleles (P<0.05). The results show that rs786204926 affects the production of PTEN isoforms, and the levels of new PTEN subtypes are higher in breast cancer chemosensitive samples.
We randomly selected 9 breast cancer tissues after chemotherapy (3 GG types, 3 AG types, and 3 AA types). WB results showed that in PTEN wild type and mutant type, AG and AA genotypes were statistically significant compared with GG alleles (P<0.05). The AG and AA genotype in SRP40 and ASF/SF2 were statistically significant compared with the GG allele (P<0.05) (Figure 5). The results showed that SNP and splicing factor related to breast cancer chemosensitivity, and the levels of PTEN-W and PTEN-M are higher in breast cancer chemosensitive samples.
The difference response of wild and mutant PTEN isoforms to PI3K drug-resistance signalling pathway
According to the predicted amino acid sequence, we used Pymol2.3.0 for homology modeling and predicted the changes in the protein structure after mutation (Figure 6A). The results showed the spatial structure of the protein was changed after gene mutation. In wild type, the amino acids D324, Y177, Y176, Q149, K183, and R189 of INS form hydrogen bonds with R46, E45, R48, E7, and Q91 of PTEN respectively; the amino acids I92, F44, I95, Y42, K41, D43, M89, I94, I9, L29, and R49 have hydrophobic interactions with the amino acids L318, F278, I280, P281, F279, P283, Y180, D187, E284, L186, L181, Y188, N184, K147 of PTEN respectively, and constitute a hydrophobic surface. The above residues may It constitutes the active surface for the interaction between the two. In the mutant, amino acids N24, K26, K53, R28, R48, E45, and Q91 of INS form hydrogen bonds with mutant K295, V296, E294, E291, Y182, Y183, and D193 respectively; amino acids L11, E76, F98, E96, P75 of INS, K27, F44, Y42, D43, G47, I9, R49, S21, I92, P22, L29, L23 and mutant C310, I312, F285, I286, G288, L324, Q313, Y186, P289, P287, E290, N190, F284, S293, K153, T292, L192, R195, D307, G299, Q297, Y182 have a hydrophobic effect and constitute a hydrophobic surface. The above residues may constitute an active surface for the interaction of the two. The results of protein docking showed that hydrogen bonding and hydrophobic interaction increased after PTEN mutation, mediating the interaction between PTEN and PIP3, making PIP3 more easily dephosphorylated into PIP2. Compared with PTEN-W, PTEN-M is more prone to phosphorylation, leading to chemosensitivity (Figure 6B). WB results showed that AG and AA genotypes were statistically significant compared with GG alleles in P-PI3K (P<0.05) (Figure 6C). It further confirmed that PTEN-W/M leads to chemosensitivity through the PI3K-AKT pathway signaling pathway.