Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 7 (1), 15742

Computational Investigation of Homologous Recombination DNA Repair Deficiency in Sporadic Breast Cancer

Affiliations

Computational Investigation of Homologous Recombination DNA Repair Deficiency in Sporadic Breast Cancer

Yue Wang et al. Sci Rep.

Abstract

BRCAness has important implications in the management and treatment of patients with breast and ovarian cancer. In this study, we propose a computational framework to measure the BRCAness of breast and ovarian tumor samples based on their gene expression profiles. We define a characteristic profile for BRCAness by comparing gene expression differences between BRCA1/2 mutant familial tumors and sporadic breast cancer tumors while adjusting for relevant clinical factors. With this BRCAness profile, our framework calculates sample-specific BRCA scores, which indicates homologous recombination (HR)-mediated DNA repair pathway activity of samples. We found that in sporadic breast cancer high BRCAness score is associated with aberrant copy number of HR genes rather than somatic mutation and other genomic features. Moreover, we observed significant correlations of BRCA score with genome instability and neoadjuvant chemotherapy. More importantly, BRCA score provides significant prognostic value in both breast and ovarian cancers after considering established clinical variables. In summary, the inferred BRCAness from our framework can be used as a robust biomarker for the prediction of prognosis and treatment response in breast and ovarian cancers.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Figure 1
Figure 1
Overview of our computational framework. In short, we first compared the gene expression profiles between familial and sporadic breast tumors considering related clinical factors (e.g. age, grade, tumor size, ER status and Her2 status) to generate a BRCAness profile. Second, by integrating gene expressions of given breast cancer samples, we calculated sample-specific BRCA scores. The scores inferred the BRCAness of patients with the higher score the higher likelihood to be BRCAness. Then, we showed that the BRCA score classifies familial from sporadic breast tumors, correlates with genomic instability, predicts patients’ survival and predicts chemotherapy response. Lastly, we further applied the BRCAness profile defined with breast cancer profiles to ovarian cancer. The corresponding BRCA scores also can classify familial from sporadic samples and predict prognosis.
Figure 2
Figure 2
BRCA score classifies familial from sporadic breast cancer patients. (a) Boxplot for comparisons of BRCA scores in germline mutate-BRCA1 (green box), germline mutate-BRCA2 (tawny box) and sporadic (blue box) breast cancer samples using gene expression of germline mutate-BRCA1 (left), germline mutate-BRCA2 (middle), germline mutate-BRCA1/2 (both BRCA1 and BRCA2, right) as reference to compare with those of sporadic ones, respectively. Mann-Whitney Wilcoxon test P-values were calculated to show the differences of BRCA scores between germline mutate-BRCA1 and sporadic breast cancer samples, germline mutate-BRCA2 and sporadic breast cancer samples. (b) ROC curves for the accuracy of classifying familial from sporadic breast cancer patients using BRCA scores calculated by comparing gene expression of germline mutate-BRCA1 (left), germline mutate-BRCA2 (middle), germline mutate-BRCA1/2 (right) as reference to sporadic ones, respectively. Black curve: comparison of patients with germline BRCA1 mutations to sporadic ones. Magenta curve: comparison of patients with germline BRCA2 mutations to sporadic ones. Cyan curve: comparison of patients with germline BRCA1 and BRCA2 mutations to sporadic ones. AUC scores were shown. The BRCA scores calculated based on germline BRCA1/2 mutations achieved the best AUCs. (c) Boxplot for BRCA scores calculated by integrating BRCAness profile and gene expression profile offered by GSE50567 which contains profiles for familial BRCA1 and BRCA2 (BRCA), other familial with non-BRCA mutations, sporadic and normal breast cancer samples. Whitney Wilcoxon test P-values were listed. (d) Same with (c) but for GSE27830 which contains profiles for four familial breast cancer samples including BRCA1, BRCA2, CHEK2 and other mutations (non-mutations of aforementioned genes). (e) Same with (c) but for GSE19177 which contains profiles for three groups of familial breast cancer samples including BRCA1, BRCA2 and non-BRCA1/2 mutations.
Figure 3
Figure 3
BRCA score correlates with breast cancer genomic features. (a) Correlation between BRCA scores and CNV burden. Spearman correlation coefficient and corresponding P-value were listed. (b) Correlation between BRCA scores and log-10 transferred mutation burden. Spearman correlation coefficient and corresponding P-value were listed. (c) We ranked TCGA breast cancer patients based on their BRCA scores in a decreasing order. We then compared the difference between top ranked (from 1% to 20%) patients and the remaining according to CNV, DNA methylation and gene expression of the 27 HR genes. The differences were calculated through the Mann-Whitney-Wilcoxon Test. Negative log-10 transferred P-values were shown. (d) Boxplot for BRCA scores in patients with 27 HR genes deletions and scores in the rest patients. And Boxplot for BRCA scores in patients with only BRCA1 and BRCA2 deletions and scores in the rest patients. Mann-Whitney-Wilcoxon Test P-values were listed. (e) Boxplot for BRCA scores in patients with 27 HR genes mutations and scores in the rest patients. And Boxplot for BRCA scores in patients with only BRCA1 and BRCA2 mutations and scores in the rest patients. Mann-Whitney-Wilcoxon Test P-values were listed.
Figure 4
Figure 4
BRCA score predicts prognosis for breast cancer patients. (a) Boxplot for BRCA scores of patients in different grades. Mann–Whitney U-test P-value was listed. (b) Same as (a) but for different cancer stages. (c) Distribution of BRCA scores based on breast cancer subtypes. Then, we ranked the BRCA scores and showed the correlations with TP53 mutation, ER status, PR status, Her2 status, triple negative breast cancer (TNBC) and molecular breast cancer subtypes. We compared BRCA scores in TP53 mutation vs. TP53 wild type, ER+ vs. ER−, PR+ vs. PR−, Her2+ vs. Her2−, TNBC vs. the other breast cancer samples, Basal-like vs. non-Basal-like, HER2-Enriched vs. non-HER2-Enriched, Luminal A vs. non-Luminal A, Luminal B vs. non-Luminal B and Normal vs. non-Normal patients. P-values were calculated using Mann–Whitney Wilcoxon test. (dg) Kaplan-Meier plots for BRCA scores comparison of the (d) discovery, (e) validation datasets in the METABRIC dataset, (f) Ur-Rehman dataset, (g) Vijver dataset. Patients with low BRCA scores (green curve) had better survival than those with high BRCA scores (red curve). Hazard ratio (HR) and log-rank test P-value were shown.
Figure 5
Figure 5
BRCA score predicts chemotherapy for breast cancer samples. (a) ROC curves for the accuracy of classifying pathologic complete response (pCR) from residual disease (RD) breast cancer patients. The BRCA scores were calculated using gene expression of germline mutate-BRCA1 (red), germline mutate-BRCA2 (blue), germline mutate-BRCA1/2 (green) as reference to compare with those of sporadic ones, respectively. AUC scores were listed. (b) Barplot for the mean AUC scores of 10-fold cross validation using clinical information (gray), clinical information + BRCA1 based BRCA scores (C+B1, red), clinical information + BRCA2 based BRCA scores (C+B2, blue) and clinical information + BRCA1/2 based BRCA scores (C+B1/2, green) to classify pCR from RD patients. The corresponding average AUCs were listed above each bar. Standard deviations were plotted with the error bars. (c) Boxplot for BRCA scores in patients with pCR (dark orange) and RD (chardonnay). BRCA scores were calculated using the BRCA1-based profiles. Mann-Whitney-Wilcoxon Test P-value was listed. (d) Barplot for pCR patients’ fractions in different BRCA score groups. The chardonnay bar is RD patients and the dark orange bar is pCR patients. Corresponding pCR rate of each group were listed above the bars.
Figure 6
Figure 6
BRCA scores implements in ovarian cancer. (a) Boxplot for BRCA scores in germline mutate-BRCA1, germline mutate-BRCA2 and sporadic ovarian cancer samples. Patients carrying germline BRCA1 mutations had much higher BRCA scores than sporadic ones (Mann-Whitney-Wilcoxon Test P = 7e-04). (b) ROC curves for the accuracy of classifying familial from sporadic ovarian cancer patients using BRCA scores. Black curve: comparison of patients with germline BRCA1 mutations to sporadic ones. Magenta curve: comparison of patients with germline BRCA2 mutations to sporadic ones. Cyan curve: comparison of patients with germline BRCA1 and BRCA2 mutations to sporadic ones. AUC scores were shown. (c) Kaplan-Meier plots for patients’ prognosis in the Bonome ovarian cancer dataset. BRCA-like patients (red curve) had better survival than non BRCA-like ones (green curve). Hazard ratio (HR) and log-rank test P-value were shown. (d) Same as (c) but for TCGA ovarian cancer samples.

Similar articles

See all similar articles

Cited by 3 articles

References

    1. McGuire A, et al. Effects of age on the detection and management of breast cancer. Cancers (Basel) 2015;7(2):908–29. doi: 10.3390/cancers7020815. - DOI - PMC - PubMed
    1. Collaborative Group on Hormonal Factors in Breast, C Familial breast cancer: collaborative reanalysis of individual data from 52 epidemiological studies including 58,209 women with breast cancer and 101,986 women without the disease. Lancet. 2001;358(9291):1389–99. doi: 10.1016/S0140-6736(01)06524-2. - DOI - PubMed
    1. Evans DG, et al. Familial breast cancer: summary of updated NICE guidance. BMJ. 2013;346:f3829. doi: 10.1136/bmj.f3829. - DOI - PubMed
    1. Miki Y, et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science. 1994;266(5182):66–71. doi: 10.1126/science.7545954. - DOI - PubMed
    1. Wooster R, et al. Localization of a breast cancer susceptibility gene, BRCA2, to chromosome 13q12–13. Science. 1994;265(5181):2088–90. doi: 10.1126/science.8091231. - DOI - PubMed
Feedback