Integration of multi-omics data shows downregulation of mismatch repair, purin, and tubulin pathways in triple-negative chemotherapy-resistant breast tumors

doi:10.21203/rs.3.rs-1954567/v1

Download PDF

Research Article

Integration of multi-omics data shows downregulation of mismatch repair, purin, and tubulin pathways in triple-negative chemotherapy-resistant breast tumors

https://doi.org/10.21203/rs.3.rs-1954567/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Triple-negative breast cancer (TNBC) is the most aggressive breast cancer subtype. Patients with TNBC are primarily treated with neoadjuvant chemotherapy (NAC). Response to NAC is prognostic, with reductions in low overall survival and disease-free survival rates in those patients who do not achieve pathological complete response (pCR). Based on this premise, we focused on patients with residual disease and hypothesized that paired analysis of primary and residual TNBC tumors following NAC could identify unique biomarkers associated with post-NAC recurrence.

Methods and Results: We investigated 24 TNBC tumors from 12 patients with pre- and post-NAC data, including four patients who recurred shortly after surgery(<24 months) and eight who remained recurrence-free (> 48 months). These tumors were collected from a prospective NAC breast cancer study (BEAUTY) conducted at the Mayo Clinic. Differential expression analysis of pre-NAC biopsies showed minimal gene expression differences between early recurrent and non-recurrent TNBC tumors; however, post-NAC samples demonstrated a significant alteration in expression patterns in response to intervention. Topological-level differences associated with early recurrence were implicated in 251 gene sets, and an independent assessment of microarray gene expression data from the paired NAC I-SPY1 trial confirmed 56 genesets. Underlying these 56 gene sets, 113 genes were observed to be differentially expressed in the I-SPY1 and BEAUTY post-NAC studies. An independent (n=392) breast cancer data set with relapse-free survival (RFS) data was used to refine our gene list to a 17 gene signature. A three-fold cross-validation analysis of the gene signature with the combined BEAUTY and I-SPY1 data yielded an average AUC of 0.88 for six machine learning models. Due to the limited studies with pre-and post-NAC TNBC tumor data, further validation of the signature is needed.

Conclusion: Analysis of multi-omics data from post-NAC TNBC chemoresistant tumors showed downregulation of mismatch repair, purin, and tubulin pathways. Additionally, we identified a 17 gene signature in TNBC associated with post-NAC recurrence enriched with down-regulation of immune genes.

Global cancer statistics 2020 (GLOBOCAN 2020) reported breast cancer as the most commonly diagnosed cancer in women, with an estimated 2.3 million new cases per year [1]. The molecular classification of breast cancer is currently based on the human epidermal growth factor receptor-2 (HER2) gene amplification and expression of sex hormone, estrogen, and progesterone receptors. Triple-negative breast cancer (TNBC), approximately 15% of invasive breast cancers, is the most aggressive breast cancer subtype; it lacks estrogen receptor (ER) expression, progesterone receptor (PR) expression, and HER2 gene amplification. Patients with TNBC have the highest mortality rates compared to the other breast cancer subtypes and are standardly treated with chemotherapy prior to surgery, known as neoadjuvant chemotherapy (NAC). NAC reduces the size of the tumor burden and decreases the pre-surgery likelihood of nodal involvement. Most importantly, especially in TNBC, response to NAC is associated with long-term patient outcomes [2]. However, nearly 50% of the patients with TNBC have residual disease after NAC[3]. Of those, 20–30% of TNBC patients develop an early disease progression within three years and exhibit high metastatic recurrence rates and poor long-term outcomes[2, 4].

With the advances in high-throughput technologies, large-scale breast cancer genomics and transcriptomics data have been collected and analyzed to identify biomarkers associated with prognosis as well as treatment resistance [2, 5–12]. Most TNBC studies focused on subtyping or developing signatures associated with prognosis or disease recurrence, mainly utilizing pre-treatment gene expression data from tumor samples. Few studies have evaluated transcriptomics data from TNBC tumors pre-NAC and post-NAC [2, 9–12]. These studies have examined changes in gene expression post-NAC to understand the molecular underpinnings driving treatment resistance in TNBCs. Therefore, identifying genes involved with chemotherapy-resistant disease (residual disease post-NAC) could lead to better stratification of patients and better therapeutic strategies for TNBC. Balko et al. measured gene expression for 450 transcripts using the NanoString platform from the pre and post-NAC breast tumor tissues and identified the DUSP4 deficiency as an important mechanism of TNBC drug resistance[11]. The Magbanua, et al. study (I-SPY1 breast cancer clinical trial) has shown the association between recurrence and response to cell cycle and immune gene sets[9]. Hancock et al. generated pre-and-post transcriptomics data using the Ion torrent platform. They identified increased expression of SMAD2, loss of TP53, and MYC-driven amplification association with chemorefractory TNBC tumors [2]. Importantly, the I-SPY1 and Hancock et al.'s studies are complementary in that they describe the depletion of the immune microenvironment and upregulation of markers related to stemness in chemoresistant tumors[2, 9]. In our manuscript, we investigated the topological differences in chemoresistant TNBC tumors. Specifically, we report the results of our interrogation of transcriptome sequencing data from treatment-resistant TNBC tumors. An in-silica deconvolution approach confirmed that post-NAC tissue samples adequately captured epithelial tissues (tumor mass) rather than tumor beds. Since TNBC represents a heterogeneous group of TNBC subtypes, we focused on androgen receptor (AR) negative TNBC (i.e., ER-, PR-, HER2- and AR-). We developed a signature associated with early recurrence using the post-NAC gene expression data from chemoresistant TNBC tumors. That signature was then compared with the patient's pre-NAC transcriptomics, whole-exome sequencing (WES), and Reverse Phase Protein Array (RPPA) data. Moreover, we validated the signature using I-SPY1 pre-and-post-NAC TNBC data. Finally, we confirmed that the 17-gene signature could predict post-NAC patients with a high risk of recurrence.

Materials or datasets

A chemorefractory cohort from the Mayo Clinic breast cancer study

The Breast Cancer Genome Guided Therapy Study (BEAUTY) is a prospective institutional review board–approved NAC clinical trial (NCT02022202) from March 2012 to May 2014. As previously described[3], the BEAUTY study enrolled 140 patients with invasive breast cancer across Mayo Clinic Rochester, Mayo Clinic Florida, and Mayo Clinic Arizona. All patients provided written informed consent. The analysis in this manuscript includes the subcohort of 42 patients with TNBC who received 12 weeks of weekly paclitaxel, followed by four cycles of an anthracycline-based regimen. The methods were performed in accordance with relevant guidelines and regulations and approved by the Mayo Clinic Institutional Review Board (IRB).

Eighteen of the 42 patients with TNBC had the residual disease at surgery. We obtained pre-NAC and post-NAC transcriptomics data for these 18 cases (36 tumors) and identified luminal androgen positive (LAR) cases using our in-house algorithm (LAR-Sig)[13]. Due to significant clinical, molecular, and pathological differences we previously observed between the LAR and non-LAR tumors, we focused our analysis on the non-LAR (AR negative) TNBC patients[14-18]. We also removed three samples as the post-NAC biopsies contained minimal epithelial tissue (see below). The final cohort for paired visit analysis included 24 tumors with four early recurrent (ERC, <24 months post-NAC) and eight non-recurrent samples (NRC, >= 48 months post-NAC). In addition to gene expression data from the BEAUTY study, we also investigated copy number alteration and somatic mutation data from whole-exome sequencing (WES, pre-NAC due to availability). Here we summarize our findings of interrogating paired gene expression data, paired protein expression data (described below), and pre-NAC WES data from the BEAUTY cohort.

Reverse Phase Protein Array data from the BEAUTY breast cancer study

Lysates from paired tissue preparations of the paired breast tumor tissue samples at pre-NAC and post-NAC time points from the BEAUTY study were provided for evaluation by Reverse Phase Protein Array (RPPA). Relative protein levels were determined by interpolation using a super curve provided by MD-Anderson Cancer Center for functional proteomics[19]. Of the twelve patients in the ERC and NRC groups, eight had RPPA data at pre-NAC, and six had RPPA data at post-NAC visits. We performed two-sample t-tests with pooled variances on the log2 transformed protein levels for 295 antibodies between the pre-NAC and post-NAC data.

Tumor bed assessment using digital deconvolution methods

A digital deconvolution was performed on the post-NAC samples using gene expression data to ensure sufficient epithelial cells were obtained rather than tumor beds. Twenty cell types were previously observed in a single-cell RNA sequencing experiment, originating from five patients with primary TNBC[20]. We obtained the single-cell gene expression counts and t-SNE clustering scheme and constructed a balanced dataset with prevalent and low abundance cell types. Our balanced dataset consisted of eighty nearest neighbor cells to each of the twenty centroids to adequately represent the least prevalent cell type (i.e., 106 immature perivascular-like fibroblasts, imPVL). The mean-dropout relationship (high zero counts while maintaining a high mean expression level) was evaluated, and the features were reduced to 3,205 genes (curve fitting parameters of a=1.5 and b=1.1)[21]. A deconvolution model was then constructed using the balanced TNBC single-cell data and CIBERSORTx [22, 23] method. The deconvolution was implemented with the twenty cell types merged into nine lineages. Each BEAUTY study sample with bulk RNA-Seq and phenotype was individually evaluated with pseudoreplication to achieve a sampling set larger than the nine cell types. Three paired RNA-Seq samples (two non-recurrent and one early recurrent) were removed as tumor bed biopsies.

I-SPY1 trial gene expression data

The clinical and molecular data from the I-SPY1 breast cancer trial have been described in detail previously [24, 25]. The normalized expression data were downloaded from the gene expression omnibus (GEO) database with a GEO ID: GSE32603[9]. From the I-SPY1 GEO data set (N=141, with baseline data), we selected patients with TNBC (N=39, 27.7%) that were resistant to NAC (N=23, 59.0%). We also considered only paired TNBC samples (N=9, 39.1%) having gene expression data at both pre-NAC treatment (T1) and post-NAC or surgery (TS) time points. Further, like the BEAUTY breast cancer study, we also considered the paired TNBC samples in the I-SPY cohort, with N=5 patients who had an early recurrence (in less than two years after NAC) and N=4 patients who remained non-recurrent for more than four years for gene expression data analysis.

Gene set analysis

Sample-wise pathway analysis of the RNA-Seq data was carried out using the gene set variation analysis (GSVA) method [26] for the curated (C2, n=2,232) and hallmark (H, n=50) gene sets from MSigDB version 7.1[27]. GSVA scores were calculated for the pre-NAC and post-NAC gene expression datasets. The scores were then evaluated using linear regression with empirical Bayes statistics with a limma package[28]. Gene sets were determined to be significant with a p-value < 0.05. Gene sets and individual genes were filtered based on the scores, fold-changes, and p-values observed in the I-SPY1 and BEAUTY trials.

Statistics and Bioinformatics analysis

Differential expression (DE) analysis was performed using the empirical Bayes quasi-likelihood F-test (QLF) to identify genes associated with recurrence for both the pre-NAC and post-NAC data with edgeR[29]. Over-representation analysis (ORA) was performed to identify cytobands significantly enriched in the DE genes using WebGestalt[30]. Copy number alterations associated with recurrence were evaluated with Fisher's exact test. Circos plots were generated with Rcircos[31]. Survival association analysis for the selected biomarkers were independently investigated using the online tool Kaplan-Meier (KM) plotter[32]. In addition, 1) The Cancer Genome Atlas (TCGA) paired tumor and normal-adjacent TNBC samples[33], 2) non-LAR vs. LAR TNBC gene expression cohorts 3) non-LAR gene expression cohorts from Thompson et al. [13] with respect to pathologic complete response phenotype were also surveyed using publically available gene expression data[3, 34-38].

Assessment of biomarkers using machine learning methods

The I-SPY1 and BEAUTY studies were pooled together and scaled with the combat algorithm from the sva package (v 3.14.0)[39] to remove the batch effect. Given the lack of sampling power, we implemented a three-fold cross-validation strategy with the super learner package[40]. We examined six classification algorithms, including a generalized linear model (glmnet), k nearest neighbors (k-NN), neural networks, random forests, support vector machine (SVM), and kernel SVM. The models were implemented using down selection methods (Pearson's correlation, Spearman's rank correlation, and random forest). Similarly, we evaluated clinical features (age, menopause, TNM stages, and pre-NAC and post-NAC Ki67 scores) for the BEAUTY cohort, using two-fold cross-validation with the super learner, as well as combined with the biomarkers. We monitored the area under the receiver operating curve (AUC) to indicate the model's classification performance. Further, we evaluated the Cox proportional hazards of the univariates in both the I-SPY1 and BEAUTY cohorts. In addition to the classification performance of random forests, kNN, and neural networks, we evaluated non-negative least squares (nnls), bayesian general linear model (glm), and rpart algorithms.

From the prospective BEAUTY neoadjuvant study, we analyzed 36 (18 pairs) triple-negative breast tumors with RNA-Seq data from pre-NAC and post-NAC after surgery (patients who failed to respond to paclitaxel and anthracycline/cyclophosphamide combination therapy). Based on our investigation and existing studies, it is known that the luminal androgen receptor (LAR) TNBCs are biologically and clinically distinct from non-LAR tumors. Hence, we removed the three TNBC LAR tumor pairs identified by LAR-Sig from further analysis. In addition, tumor bed analysis was performed using an in silica deconvolution method, indicating that three post-NAC biopsies contained low epithelial composition, both luminal and basal. Hence, we removed those three also from the analysis. We finally investigated 24 tumors from 12 patients with non-LAR TNBC with matched pre and post-NAC gene expression data (Table 1). Of these 12 patients, four had an early recurrence (less than two years after completion of NAC), and eight remained recurrence-free for more than four years. Furthermore, for these 12 non-LAR TNBC patients, all patients had pre-NAC WES data, and 9/12 had paired protein and phosphoprotein (RPPA) data.

Post-NAC transcriptomics analysis showed clusters of genes differentially expressed on chr1q and chr17q

We investigated post-NAC RNA-Seq data for transcriptional differences associated with the response after standard chemotherapy treatment. We evaluated 20,543 genes assayed from both pre-and post-NAC biopsies. A co-inertia plot of paired samples is presented in Figure 1A, using the first two principal components. In the plot, we noted a slight trend of separation between patients with ERC and NRC for the first principal component. This component explains 20.3% of the variance observed in the pooled dataset.

We first compared the pre-NAC transcriptome gene expression between ERC and NRC groups. We identified only 11 genes to be differentially expressed after correcting for multiple testing with false discovery rate (FDR)<0.05. Conversely, we evaluated ERC and NRC tumors in the post-NAC gene expression data. We identified 660 genes differentially expressed with at least 2-fold change and FDR < 0.05 (Figure 1B, Supplementary Table 1). Among these genes, 478 (72.4%) were upregulated in ERC tumors (Figure 1B). The list of the top 23 significant genes based upon an FDR < 1e-4 is shown in Figure 1C.

Further over representation analysis (ORA) identified 10 cytobands that were significantly enriched in the 660 DE genes (Figure 1D), including 17q (17q25.3, 33 genes, p-value less than 10^-15 and 17q25.1, 15 genes, p-value 1.01×10^-1), 6p (6p22.2, 14 genes, p-value 1.40×10^-10), 10q (10q22.3, 9 genes, p-value 6.51×10^-5), and 1q (1q, 8 genes, p-value 9.39×10^-5; 1q24.3, 3 genes, p-value 0.007; and 1q23.2, 4 genes, p-value 0.01).

Genomic analysis of pre-NAC data confirms amplifications in chromosomes 1q and 17q in the early-recurrence group (ERC) compared to the non-recurrent (NRC) group

Since WES data of post-NAC tumors were unavailable, we examined copy number alterations in the pre-NAC patient tumors between ERC and NRC groups. We identified a common gain/amplification in a large region of the chr1q (q21.3-q24.2,395 altered genes) and chr17q (q25.3, 97 altered genes) in at least 3 out of 4 recurrent patients; however, they were not detected in any of the eight non-recurrent patients (p-value <=0.012 by Fisher's exact test). The 1q and 17q gain events involved 492 genes (Supplementary Table 2). Similarly, we also identified a deletion of 4 genes (in chr4p) among the ERC group. A circos plot depicting the altered copy number regions is presented in Figure 2. We also present histogram views of the copy gains in the early-recurrent samples for the 17q and 1q regions, along with the elevated expression levels observed with these copy gains using pre-NAC and post-NAC RNA-Seq data (lower panels of Figure 2, respectively). Below the circos plot, we present the chromosome 4p15.2-15.32 region, where we detected an increased copy loss in the early recurrent patients. Again, a decrease in gene expression was observed across the region (Figure 2).

Somatic Mutation showed frequent mutations in TP53 and MUC4.

We investigated the somatic mutations in the pre-NAC sample biopsies using WES tumor and blood DNA data. We identified 392 high/moderate impact somatic mutations corresponding to 366 genes in the four early-recurrent patient tumors and 545 high/moderate impact mutations in 519 genes in the eight non-recurrent patient tumors, respectively (Supplementary Table 3). Due to the small sample size, low mutation frequency across genes and low overlap among the genes were observed between the ERC and NRC groups. It was noted that the most frequently mutated genes across our cohort were TP53 (all four ERC samples and five of eight NRC samples) and MUC4 (two of the four ERC samples and three of eight NRC samples), consistent with the previous publication[41].

RPPA analysis indicates pre-NAC proteomic differences

We evaluated protein expression (232 proteins) and phosphorylation (63 phosphorylation sites) data obtained from the pre-NAC and post-NAC tumors using RPPA. The pre-NAC RPPA data were compared between the ERC and NRC groups. We observed that 30 proteins were significantly different (p-value < 0.05), including ten phosphorylation sites (10/63, 15.9%) and 20 total proteins (20/232,8.6%) among the pre-NAC samples. In contrast, there were only four protein level variations and no phosphorylation site differences in the post-NAC RPPA data. See Supplementary Tables 4-5 for full results.

Down-regulation of Tubulin and upregulation of growth receptors in the non-recurrent post-NAC tumors

A parallel single-sample gene set analysis was performed rather than single gene expression analyses to gain more insight into the group of genes (gene sets) from the gene expression data and resolve the challenges of a small sample size study. We performed Gene Set Variation Analysis (GSVA) and obtained individual gene set scores for all non-LAR TNBC tumors with the pre-NAC and post-NAC RNA-Seq data. We evaluated three comparisons (post-NAC difference between ERC vs. NRC, pre and post-NAC (paired ERC and NRC). Among the total 2282 gene sets, we identified 251 gene sets (Supplementary Table 6) significantly altered comparing the ERC and NRC group in the post-NAC tumors (p-value < 0.05).

In addition, we performed a hierarchical clustering analysis of log fold changes for 251 gene sets, Figure 3. The log fold changes represented the change in post-NAC data and the difference between the tumors for the ERC and NRC groups. The average silhouette width, a metric of within-class similarity and between-class dissimilarity, was 0.90 for two gene set clusters and was markedly reduced for three gene set clusters (0.61). We observed that the first cluster of 61 gene sets were downregulated in the post-NAC non-recurrent patients, compared to the post-NAC early recurrent group as well as the own pre-NAC biopsies (Figure 3). Tubulin (TUBB4A) was the most frequent gene in these gene sets, present in at least eleven (18.0%) gene sets, whereas the pre-NAC expression levels for tubulin were not significantly different. Conversely, the second cluster contains 190 gene sets that were upregulated in the post-NAC non-recurrent tumors. Furthermore, this cluster was down-regulated in the post-NAC early-recurrence cohort compared to their pre-NAC ERC sample biopsies. The heatmap image is presented with the gene clusters labeled, and the most commonly represented genes in the gene sets constitute the word clouds in Figure 3. In summary, cluster 1 gene sets that were downregulated in the post-NAC NRC patients were predominately associated with tubulin. In cluster2, the upregulated gene sets include growth receptors such as EGFR, TGFBR2, and GHR, as well as tumor suppressor genes found in apoptotic gene sets.

I-SPY1 confirmation of the core gene set signature from the BEAUTY TNBC cohort

We investigated publically available data from the I-SPY1 study to validate our pathway analysis findings. Due to the platform differences between the I-SPY1 (microarray) and BEAUTY (RNA-Seq) datasets, we calculated the gene set scores for 188/251 (74.9%) gene sets using the GSVA method. Of the 188 gene sets, 56 gene sets (29.8%) demonstrated high similarity between the I-SPY1 and BEAUTY studies (Supplementary Table 7 of 56 genesets). Among these 56 gene sets, 17 gene sets (cluster 1, Figure 3) were down-regulated, and 39 (cluster 2, Figure 3) were upregulated in the post-NAC I-SPY1 early-recurrence tumors. Cluster 1 could not recapitulate the same set of genes as TUBB4A was not assayed on the microarray platform used for the ISPY trial. We observed five consistent genes in cluster 2, including ERLIN2, FCER1A, TGFBR2, PER2, and EGFR. These genes are common to gene sets, including MYC targets, TGFβ, FOXO, and PI3K.

Identification of 17-gene signature to predict recurrence of cancer in chemoresistant tumors

We identified 113 differentially expressed genes up/down-regulated in the same direction when comparing the ERC and NRC patients in the I-SPY1 and BEAUTY studies related to 56 gene sets (listed in Supplementary Table 7). We assessed the prognostic value of these 113 genes in a publically available TNBC cohort (n=392 independent dataset pooled from several TNBC studies) using the Kaplan-Meier (KM) plotter online breast cancer database. Our analysis shows that 17 genes were significantly associated (log-rank test p-value < 0.05) with relapse-free survival (RFS) in the TNBC cohort (Figure 4, Supplementary Figure 1). Further investigation into this same subcohort with positive lymph node disease (N=189) showed that 10 of the 17 genes remained significant (Table 2). Additionally, we evaluated the proportional hazards for each of the genes for the BEAUTY AND I-SPY1 trials. We observed a significant correlation (Pearson correlation coefficient ρ=0.78) of the hazard ratios between the two studies (Supplementary Table 8).

Prediction of recurrence using chemoresistant gene expression data, machine learning models, and cross-validation methods with the 17 gene signature

Since there are no independent neoadjuvant breast cancer studies with publically available data to validate our signature, transcriptomic data from the BEAUTY and I-SPY1 studies were pooled together to evaluate the 17-gene signature using machine learning models. We examined six classification algorithms, including a generalized linear model (glmnet), k nearest neighbors (k-NN), neural networks, random forests, support vector machine (SVM), and kernel SVM. The models were implemented as wrappers in the R package superlearner using gene down selection methods (Pearson's correlation, Spearman's rank correlation, and random forest). Given the lack of sampling power, we implemented a three-fold cross-validation strategy, which we believe provided the most reasonable cross-validation strategy without subsampling the data into too trivial of a sample representation. We monitored the area under the receiver operating curve as indicative of the model's classification performance, provided in Figure 5. The rank-based correlation (dark brown diamond) consistently underperformed other feature selection approaches (Supplementary Figure 2). Most importantly, the ensemble of seventeen (labeled as 'All' in Figure 5, tan diamond) genes was consistently the best performing approach. We present the distribution of AUCs observed for each classification wrapper, Figure 5. The models are ordered by increasing median AUC values, and kernel support vector machines, k nearest-neighbor (kNN), and random forest were the best performing modeling approaches, with the ensemble of all 17 genes achieving AUCs of 0.84, 0.88, and 0.88, respectively. A similar assessment of a model based on the available clinical features (as shown in Table 1) was also performed to compare with the transcriptome data-based 17-gene signature model. Due to the limited clinical data availability for I-SPY1 cohort this comparison was limited to the BEAUTY TNBC cohort. The best performance of the clinical feature-based models was AUC=0.69, and clinical feature selection failed to improve the performance of the clinical feature-based models. Overall, the 17-gene signature provided classification models that outperformed all of the clinical feature models.

Differential expression of 17 genes in post-NAC TNBC ERC and NRC cohorts compared to paired pre-NAC tumors, primary tumor (baseline) samples from TCGA, and TCGA normal-adjacent tumors and association with pCR

We investigated several datasets to confirm that the dysregulation of our signature's 17 genes were 1) specific to Non-LAR tumors and 2) specific to post-NAC Non-LAR tumor biopsies. We evaluated the expression among eleven paired TNBC primary tumors (in LARs and non-LARs) and normal adjacent samples from TCGA and observed dysregulation among the tumor samples, with 9 out of the 17 (52.9%) surveyed being differentially expressed, with 7 of 9 genes confirmed to be DE when the analysis was refined to the nine paired Non-LAR TCGA tumors (Supplementary Figures 3-4). Moreover, among these seven genes, we confirmed the Non-LAR dysregulation in a core set of five genes (COL4A6, AGGF1, TANK, CITED2, PDE2A, and MCM3) in an independent cohort of N=486 TNBC samples (98 LAR and 388 Non-LAR TNBC) (Supplementary Figures 5). These publically available microarray datasets collated from Thompson et al. 2022 study were primarily assayed by the Affymetrix U33 platform [13].

Further, we evaluated whether dysregulated expression of these 17 genes existed prior to NAC treatment. As shown in Figure 6, we verified that the differential expression of these 17 genes was specific to post-NAC tumors, while none of the genes were observed to be differentially expressed among the pre-NAC tumor biopsies. We subsequently investigated the 288 (of 388) non-LAR primary tumor TNBC samples from the independent cohort (124 pCR, 164 non pCR, 100 missing). We observed that only 2 (12.5%) of the 16 (RSPO3 was not assayed on the microarray platform) genes were differentially expressed (Supplementary Figures 6). In summary, the dysregulation of the 17 genes identified in this study were only associated with post-NAC recurrence in non-LAR TNBCs with residual disease after NAC (Supplementary Figures 7).

Triple-negative breast cancers are phenotypically heterogeneous disease that presents in a younger population of women. The TNBCs currently lack therapeutic targets, and neoadjuvant chemotherapy is the standard of care from stage I to stage III diagnosis. The 3-year survival rate of TNBC patients who achieve pathological complete response is 94%. However, the 3-year survival rate plunges to 68% for those who fail to achieve pCR, and approximately 50% will develop recurrent disease within the first 3–4 years[2]. Hence, we investigated the chemoresistant BEAUTY TNBC tumors for potential biomarkers delineating early recurrence patient's landscape from non-recurrence profiles. Investigation of transcriptional differences suggests that therapeutic selection pressure drives post-NAC differences that are non-existent prior to pre-NAC treatment. Among the 660 differentially expressed genes (FDR < 0.05 and |logFC|>1), we observed enrichment of genes in post-NAC data in a few cytoband regions (Fig. 1). Two of these regions (17q25 and 1q23-24) were confirmed to demonstrate copy gains in the pre-NAC biopsy, which coincided with the elevated expression observed in the post-NAC early recurrent samples. Three genes found in the 17q25 region are involved in sumoylation: NUP85, CBX2, and CBX4. Similarly, the cytoband 4p15 presented with a copy loss in the early recurrent tumors (Pre-NAC), and the post-NAC also demonstrated decreased expression (Fig. 2).

A parallel track of analysis using topological gene set measurements was adopted to navigate the challenges we faced, given the relatively low sampling power. To reduce the number of genomic features (20,543 genes), we conducted a GSVA analysis (with 2,282 gene sets) on the post-NAC BEAUTY gene expression data and leveraged our observations with the independent I-SPY1 studies post-NAC microarray data. We observed that 251 gene sets were altered between the ERC and NRC BEAUTY TNBC tumors(Fig. 3). Cluster analysis of the 251 gene sets identified two clusters, and the first cluster includes TUBB4A, the therapeutic target of paclitaxel. The gene sets were upregulated in the ERC and included metastasis-promoting gene sets, DNA mismatch repair, and TP53 gene sets. In contrast, the second cluster of gene sets were down-regulated in the ERC samples involving tumor suppressor gene sets, including FOXO signaling, TGF-beta signaling, and apoptosis (Fig. 3).

Due to the technology differences between the microarray (I-SPY1) and RNA-Seq (BEAUTY) data, only 188/251 gene sets from the BEAUTY study were also investigated in the I-SPY1 study. We confirmed 56 gene sets in the I-SPY1 data to be significant and concordantly altered. Further, the 56 gene sets were represented by 113 genes that were concordant in direction and differentially expressed in the two datasets. We refined the 113 genes using the online tool KM-plotter and identified seventeen genes associated with recurrence-free survival in 392 TNBC samples (Fig. 4). Cross-validation analysis of these genes demonstrated a robust ability to predict recurrence, particularly with random forest, kNN, and kernel SVM (Fig. 5). Further, dysregulation of these genes was confirmed to be specific to Non-LAR TNBC and resultant of therapeutic intervention (Fig. 6).

In summary, we confirmed that the expression of these 17 signature genes is derived from non-LAR TNBC tumors and that expression differences were specific to post-NAC treatment. Four genes (FHL2, TANK, PDE2A, and RSPO3) were well documented to be associated with breast cancer and were observed to be down-regulated in the post-NAC early recurrent samples. PDE2A is a phosphodiesterase that regulates mitochondrial respiration and mitogenic clearance[42, 43]. FHL2 is a zinc finger transcription factor that has been associated with several cancers, including ovarian and cervical cancers[44–46]. We note that the I-SPY1 trial observed an increase in interferon signaling associated with shorter recurrence-free survival among non-responding patients[9]. We observed that two molecules, RSPO3, and TANK, are related to NF-κB signaling and survival response through the induction of inflammatory molecules. RSPO3 is a matricellular protein that activates WNT signaling via the canonical and non-canonical gene sets. Its involvement in cell development and growth has implicated its association with colorectal and lung cancers. RSPO3 downregulation has been involved in prostate cancer invasiveness, and it interacts with the inflammatory mediator IL-1β[47, 48]. While two SNPs (rs17705608 and rs7309) in the TANK gene have been associated with breast cancer risk[49], it is involved in TNF-mediated signal transduction. Most importantly, the TANK, the TRAF family member, binds the NEMO (IKKγ) to induce inflammation through the IKK complex and NF-κB signal transduction[50–52]. These findings suggest that initiating inflammation signaling via NF-κB signal transduction is integral to recurrent free disease. Moreover, we observed significant downregulation (-2.91, p-value 0.044) of the key apoptotic protein, BCL2, suggesting that the inflammation must be accompanied by immune infiltration and subsequent cell death.

We systematically analyzed the in-house and publicly available TNBC gene expression NAC datasets using computational biology and machine learning methods. We concluded that the post-NAC non-LAR TNBCs change significantly during the treatment compared to the baseline pre-NAC tumors. We have shown that the 17 genes identified in this study are novel biomarkers that predict recurrence in post-NAC residual tumors. Given the nature of our study, where we needed both clinical and omics data from paired pre- and post-NAC tumors from the same TNBC patient cohort, our ability to robustly evaluate the data has been limited. However, we have made a conscious effort to limit the false discoveries by applying p-value cutoffs along with fold change thresholds. Although we observed an average AUC of 0.88 using cross-validation analysis of the publicly available dataset and our Mayo Clinic BEAUTY study, further validation of the 17-gene signature in large TNBC cohorts is necessary.

We observed notable differences in chemoresistant post-NAC TNBC tumor biopsies, significantly influencing recurrence-free survival. Among these transcriptional changes, we confirmed the dysregulation of a 17 genes signature specific to Non-LAR TNBC tumors. These genes have been implicated in initiating inflammation signaling via NF-κB signal transduction. While the gene signature needs further validation, we have been able to confirm the dentrimential impact of the depletion of the immune microenvironment.

ETHICS approval and consent to participate

All patients in the study provided written informed consent. The methods were performed in accordance with relevant guidelines and regulations and approved by the Mayo Clinic Institutional Review Board (IRB).

Availability of data and materials

All data used in this analysis are publically available. Pre-NAC BEAUTY study data are available through dbGap accession ID phs001050.v1.p1. I-SPY1 data are available through the GEO database (GSE32603).

COMPETING INTERESTS STATEMENT

The authors declare no Competing Financial or Non-Financial Interests

FUNDINGS AND ACKNOWLEDGMENTS

This work was supported by Mayo Clinic's Center for Individualized Medicine (CIM) and Mayo Clinic Breast Specialized Program of Research Excellence (SPORE) (P50CA116201) (MPG, JMC, VJS, LW, and KRK). Its contents are solely the authors' responsibility and do not necessarily represent the official views of the NIH.

AUTHOR CONTRIBUTIONS
XT, KJT, JPS, PTV, VJS, MAS, NWD, MAA, GJR, JMC, LW, RW, JCB, MPG, and KRK conceived the idea and reviewed the manuscript. XT, KJT, JPS, PTV, VJS, and KRK conducted analysis and prepared the manuscript. All authors contributed to the discussion, and overall data interpretation and approved the final manuscript.

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F: Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021, 71(3):209-249.
Hancock BA, Chen YH, Solzak JP, Ahmad MN, Wedge DC, Brinza D, Scafe C, Veitch J, Gottimukkala R, Short W et al: Profiling molecular regulators of recurrence in chemorefractory triple-negative breast cancers. Breast Cancer Res 2019, 21(1):87.
Goetz MP, Kalari KR, Suman VJ, Moyer AM, Yu J, Visscher DW, Dockter TJ, Vedell PT, Sinnwell JP, Tang X et al: Tumor Sequencing and Patient-Derived Xenografts in the Neoadjuvant Treatment of Breast Cancer. J Natl Cancer Inst 2017, 109(7).
Liedtke C, Mazouni C, Hess KR, Andre F, Tordai A, Mejia JA, Symmans WF, Gonzalez-Angulo AM, Hennessy B, Green M et al: Response to neoadjuvant therapy and long-term survival in patients with triple-negative breast cancer. J Clin Oncol 2008, 26(8):1275-1281.
Craig DW, O'Shaughnessy JA, Kiefer JA, Aldrich J, Sinari S, Moses TM, Wong S, Dinh J, Christoforides A, Blum JL et al: Genome and transcriptome sequencing in prospective metastatic triple-negative breast cancer uncovers therapeutic vulnerabilities. Mol Cancer Ther 2013, 12(1):104-116.
Jiang YZ, Ma D, Suo C, Shi J, Xue M, Hu X, Xiao Y, Yu KD, Liu YR, Yu Y et al: Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell 2019, 35(3):428-440 e425.
Cancer Genome Atlas N: Comprehensive molecular portraits of human breast tumours. Nature 2012, 490(7418):61-70.
Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM, Dunning MJ, Speed D, Lynch AG, Samarajiwa S, Yuan Y et al: The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 2012, 486(7403):346-352.
Magbanua MJ, Wolf DM, Yau C, Davis SE, Crothers J, Au A, Haqq CM, Livasy C, Rugo HS, Investigators IST et al: Serial expression analysis of breast tumors during neoadjuvant chemotherapy reveals changes in cell cycle and immune pathways associated with recurrence and response. Breast Cancer Res 2015, 17:73.
Shenoy A, Belugali Nataraj N, Perry G, Loayza Puch F, Nagel R, Marin I, Balint N, Bossel N, Pavlovsky A, Barshack I et al: Proteomic patterns associated with response to breast cancer neoadjuvant treatment. Mol Syst Biol 2020, 16(9):e9443.
Balko JM, Giltnane JM, Wang K, Schwarz LJ, Young CD, Cook RS, Owens P, Sanders ME, Kuba MG, Sanchez V et al: Molecular profiling of the residual disease of triple-negative breast cancers after neoadjuvant chemotherapy identifies actionable therapeutic targets. Cancer Discov 2014, 4(2):232-245.
Balko JM, Cook RS, Vaught DB, Kuba MG, Miller TW, Bhola NE, Sanders ME, Granja-Ingram NM, Smith JJ, Meszoely IM et al: Profiling of residual breast cancers after neoadjuvant chemotherapy identifies DUSP4 deficiency as a mechanism of drug resistance. Nat Med 2012, 18(7):1052-1059.
Thompson KJ, Leon-Ferre RA, Sinnwell JP, Zahrieh DM, Suman VJ, Metzger FO, Asad S, Stover DG, Carey L, Sikov WM et al: Luminal androgen receptor breast cancer subtype and investigation of the microenvironment and neoadjuvant chemotherapy response. NAR Cancer 2022, 4(2):zcac018.
Burstein MD, Tsimelzon A, Poage GM, Covington KR, Contreras A, Fuqua SA, Savage MI, Osborne CK, Hilsenbeck SG, Chang JC et al: Comprehensive genomic analysis identifies novel subtypes and targets of triple-negative breast cancer. Clin Cancer Res 2015, 21(7):1688-1698.
Jezequel P, Kerdraon O, Hondermarck H, Guerin-Charbonnel C, Lasla H, Gouraud W, Canon JL, Gombos A, Dalenc F, Delaloge S et al: Identification of three subtypes of triple-negative breast cancer with potential therapeutic implications. Breast Cancer Res 2019, 21(1):65.
Jezequel P, Loussouarn D, Guerin-Charbonnel C, Campion L, Vanier A, Gouraud W, Lasla H, Guette C, Valo I, Verriele V et al: Gene-expression molecular subtyping of triple-negative breast cancer tumours: importance of immune response. Breast Cancer Res 2015, 17:43.
Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y, Pietenpol JA: Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest 2011, 121(7):2750-2767.
Lehmann BD, Jovanovic B, Chen X, Estrada MV, Johnson KN, Shyr Y, Moses HL, Sanders ME, Pietenpol JA: Refinement of Triple-Negative Breast Cancer Molecular Subtypes: Implications for Neoadjuvant Chemotherapy Selection. PLoS One 2016, 11(6):e0157368.
Hu J, He X, Baggerly KA, Coombes KR, Hennessy BT, Mills GB: Non-parametric quantification of protein lysate arrays. Bioinformatics 2007, 23(15):1986-1994.
Wu SZ, Roden DL, Wang C, Holliday H, Harvey K, Cazet AS, Murphy KJ, Pereira B, Al-Eryani G, Bartonicek N et al: Stromal cell diversity associated with immune evasion in human triple-negative breast cancer. EMBO J 2020, 39(19):e104063.
Kobak D, Berens P: The art of using t-SNE for single-cell transcriptomics. Nat Commun 2019, 10(1):5416.
Steen CB, Liu CL, Alizadeh AA, Newman AM: Profiling Cell Type Abundance and Expression in Bulk Tissues with CIBERSORTx. Methods Mol Biol 2020, 2117:135-157.
Chen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA: Profiling Tumor Infiltrating Immune Cells with CIBERSORT. Methods Mol Biol 2018, 1711:243-259.
Esserman LJ, Berry DA, DeMichele A, Carey L, Davis SE, Buxton M, Hudis C, Gray JW, Perou C, Yau C et al: Pathologic complete response predicts recurrence-free survival more effectively by cancer subset: results from the I-SPY 1 TRIAL--CALGB 150007/150012, ACRIN 6657. J Clin Oncol 2012, 30(26):3242-3249.
Hylton NM, Blume JD, Bernreuter WK, Pisano ED, Rosen MA, Morris EA, Weatherall PT, Lehman CD, Newstead GM, Polin S et al: Locally advanced breast cancer: MR imaging for prediction of response to neoadjuvant chemotherapy--results from ACRIN 6657/I-SPY TRIAL. Radiology 2012, 263(3):663-672.
Hanzelmann S, Castelo R, Guinney J: GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics 2013, 14:7.
Liberzon A, Birger C, Thorvaldsdottir H, Ghandi M, Mesirov JP, Tamayo P: The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 2015, 1(6):417-425.
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK: limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015, 43(7):e47.
Law CW, Alhamdoosh M, Su S, Dong X, Tian L, Smyth GK, Ritchie ME: RNA-seq analysis is easy as 1-2-3 with limma, Glimma and edgeR. F1000Res 2016, 5.
Liao Y, Wang J, Jaehnig EJ, Shi Z, Zhang B: WebGestalt 2019: gene set analysis toolkit with revamped UIs and APIs. Nucleic Acids Res 2019, 47(W1):W199-W205.
Zhang H, Meltzer P, Davis S: RCircos: an R package for Circos 2D track plots. BMC Bioinformatics 2013, 14:244.
Gyorffy B: Survival analysis across the entire transcriptome identifies biomarkers with the highest prognostic power in breast cancer. Comput Struct Biotechnol J 2021, 19:4101-4109.
TCGA: Comprehensive molecular portraits of human breast tumours. Nature 2012, 490(7418):61-70.
Shepherd JH, Ballman K, Polley MC, Campbell JD, Fan C, Selitsky S, Fernandez-Martinez A, Parker JS, Hoadley KA, Hu Z et al: CALGB 40603 (Alliance): Long-Term Outcomes and Genomic Correlates of Response and Survival After Neoadjuvant Chemotherapy With or Without Carboplatin and Bevacizumab in Triple-Negative Breast Cancer. J Clin Oncol 2022:JCO2101506.
Santonja A, Sanchez-Munoz A, Lluch A, Chica-Parrado MR, Albanell J, Chacon JI, Antolin S, Jerez JM, de la Haba J, de Luque V et al: Triple negative breast cancer subtypes and pathologic complete response rate to neoadjuvant chemotherapy. Oncotarget 2018, 9(41):26406-26416.
Loibl S, O'Shaughnessy J, Untch M, Sikov WM, Rugo HS, McKee MD, Huober J, Golshan M, von Minckwitz G, Maag D et al: Addition of the PARP inhibitor veliparib plus carboplatin or carboplatin alone to standard neoadjuvant chemotherapy in triple-negative breast cancer (BrighTNess): a randomised, phase 3 trial. Lancet Oncol 2018, 19(4):497-509.
Miyake T, Nakayama T, Naoi Y, Yamamoto N, Otani Y, Kim SJ, Shimazu K, Shimomura A, Maruyama N, Tamaki Y et al: GSTP1 expression predicts poor pathological complete response to neoadjuvant chemotherapy in ER-negative breast cancer. Cancer Sci 2012, 103(5):913-920.
Hatzis C, Pusztai L, Valero V, Booser DJ, Esserman L, Lluch A, Vidaurre T, Holmes F, Souchon E, Wang H et al: A genomic predictor of response and survival following taxane-anthracycline chemotherapy for invasive breast cancer. JAMA 2011, 305(18):1873-1881.
Leek JT, Johnson, W. Evan, Parker, Hilary S., Fertig, Elana J., Jaffe, Andrew E., Storey, John D.: sva: Surrogate Variable Analysis. In., 3.14.0 edn. Vienna, Austria: R Foundation for Statistical Computing; 2015.
van der Laan MJ, Polley EC, Hubbard AE: Super learner. Stat Appl Genet Mol Biol 2007, 6:Article25.
Chang CS, Kitamura E, Johnson J, Bollag R, Hawthorn L: Genomic analysis of racial differences in triple negative breast cancer. Genomics 2019, 111(6):1529-1542.
Lobo MJ, Reverte-Salisa L, Chao YC, Koschinski A, Gesellchen F, Subramaniam G, Jiang H, Pace S, Larcom N, Paolocci E et al: Phosphodiesterase 2A2 regulates mitochondria clearance through Parkin-dependent mitophagy. Commun Biol 2020, 3(1):596.
Acin-Perez R, Russwurm M, Gunnewig K, Gertz M, Zoidl G, Ramos L, Buck J, Levin LR, Rassow J, Manfredi G et al: A phosphodiesterase 2A isoform localized to mitochondria regulates respiration. J Biol Chem 2011, 286(35):30423-30432.
Wang C, Lv X, He C, Davis JS, Wang C, Hua G: Four and a Half LIM Domains 2 (FHL2) Contribute to the Epithelial Ovarian Cancer Carcinogenesis. Int J Mol Sci 2020, 21(20).
Jin X, Jiao X, Jiao J, Zhang T, Cui B: Increased expression of FHL2 promotes tumorigenesis in cervical cancer and is correlated with poor prognosis. Gene 2018, 669:99-106.
Cao CY, Mok SW, Cheng VW, Tsui SK: The FHL2 regulation in the transcriptional circuitry of human cancers. Gene 2015, 572(1):1-7.
Mesci A, Lucien F, Huang X, Wang EH, Shin D, Meringer M, Hoey C, Ray J, Boutros PC, Leong HS et al: RSPO3 is a prognostic biomarker and mediator of invasiveness in prostate cancer. J Transl Med 2019, 17(1):125.
Skaria T, Bachli E, Schoedon G: RSPO3 impairs barrier function of human vascular endothelial monolayers and synergizes with pro-inflammatory IL-1. Mol Med 2018, 24(1):45.
Resler AJ, Malone KE, Johnson LG, Malkki M, Petersdorf EW, McKnight B, Madeleine MM: Genetic variation in TLR or NFkappaB pathways and the risk of breast cancer: a case-control study. BMC Cancer 2013, 13:219.
Lafont E, Draber P, Rieser E, Reichert M, Kupka S, de Miguel D, Draberova H, von Massenhausen A, Bhamra A, Henderson S et al: TBK1 and IKKepsilon prevent TNF-induced cell death by RIPK1 phosphorylation. Nat Cell Biol 2018, 20(12):1389-1399.
Van Quickelberghe E, De Sutter D, van Loo G, Eyckerman S, Gevaert K: A protein-protein interaction map of the TNF-induced NF-kappaB signal transduction pathway. Sci Data 2018, 5:180289.
Zhang W, Wang J, Zhang Y, Yuan Y, Guan W, Jin C, Chen H, Wang X, Yang X, He F: The scaffold protein TANK/I-TRAF inhibits NF-kappaB activation by recruiting polo-like kinase 1. Mol Biol Cell 2010, 21(14):2500-2513.

Table 1. Demographic and clinical characteristics of TNBC patients included in this analysis.

	NRC (N=8)	ERC (N=4)	Total (N=12)
Age
Mean (SD)	51.4 (11.0)	55.8 (13.1)	52.8 (11.3)
Range	36 - 70	40 - 67	36 – 70
Menopausal.status
peri	1 (12.5%)	0 (0.0%)	1 (8.3%)
post	4 (50.0%)	3 (75.0%)	7 (58.3%)
pre	3 (37.5%)	1 (25.0%)	4 (33.3%)
T.stage
T2	3 (37.5%)	2 (50.0%)	5 (41.7%)
T3	5 (62.5%)	2 (50.0%)	7 (58.3%)
N.stage
N0	6 (75.0%)	3 (75.0%)	9 (75.0%)
N1	2 (25.0%)	1 (25.0%)	3 (25.0%)
M.stage
M0	8 (100.0%)	4 (100.0%)	12 (100.0%)
Race
NA	0 (0.0%)	1 (25.0%)	1 (8.3%)
American Indian or Alaska Native	1 (12.5%)	0 (0.0%)	1 (8.3%)
Caucasian	7 (87.5%)	3 (75.0%)	10 (83.3%)

Table 2. List of the 17 gene signature.

Gene	BEUATY logFC	BEUATY P-value	ISPY logFC	ISPY P-value	RFS p-value	RFS (LN+) p-value	Overall	Location	Type(s)	Cytoband
CD3G	-1.4	0.0211	-0.506	0.133	0.0079	0.13	6.014	Plasma Membrane	Transmembrane receptor	11q23.3
FHL2	-1.06	0.0014	-0.345	0.518	0.0008	0.0004	5.98	Nucleus	Transcription regulator	2q12.2
SNRPF	0.609	0.0492	0.743	0.072	0.0493	0.13	4.27	Nucleus	other	12q23.1
MCCC2	-0.942	0.00356	-0.118	0.73	0.0083	0.039	3.699	Cytoplasm	enzyme	5q13.2
TBXA2R	1.24	0.0119	0.088	0.812	0.0162	0.017	3.219	Plasma Membrane	G-protein coupled receptor	19p13.3
MCM3	0.907	0.0111	0.478	0.343	0.0139	0.0017	3.091	Nucleus	enzyme	6p12.2
TANK	-1.04	0.00396	-0.275	0.622	0.028	0.17	2.995	Cytoplasm	other	2q24.2
CITED2	-1.11	0.032	-1.403	0.046	0.0032	0.13	2.918	Nucleus	Transcription regulator	6q24.1
AGGF1	-0.606	0.039	-0.626	0.099	0.0278	0.48	2.853	Cytoplasm	other	5q13.3
MED1	-0.676	0.0181	-0.14	0.569	0.0333	0.077	2.775	Nucleus	Transcription regulator	17q12
RSPO3	-2.32	0.00114	-0.692	0.152	0.0422	0.05	2.092	Extracellular Space	kinase	6q22.33
PDE2A	-1.38	0.0313	-0.301	0.435	0.0498	0.053	1.641	Cytoplasm	enzyme	11q13.4
RPS6KA4	0.744	0.0391	0.633	0.135	0.0147	0.017	1.458	Cytoplasm	kinase	11q13.1
COL4A6	-3.27	0.00197	-0.2	0.722	0.0009	0.019	1.019	Extracellular Space	other	Xq22.3
NR1D2	-0.818	0.0358	-0.591	0.236	0.022	0.014	-0.476	Nucleus	ligand-dependent nuclear receptor	3p24.2
LCP2	1.82	0.00225	0.091	0.742	0.0407	0.22	-0.817	Cytoplasm	other	5q35.1
THRA	-1.32	0.00635	-0.11	0.819	0.0266	0.19	-0.577	Nucleus	nuclear hormone receptor	17q21.1

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Integration of multi-omics data shows downregulation of mismatch repair, purin, and tubulin pathways in triple-negative chemotherapy-resistant breast tumors

Status:

Version 1

Abstract

Figures

Background

Methods

Results

Discussion

Conclusions

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1