Ovarian cancer is detectable from peripheral blood using machine learning over T-cell receptor repertoires.

The extraordinary diversity of T cells and B cells is critical for body maintenance. This diversity has an important role in protecting against tumor formation. In humans, the T-cell receptor (TCR) repertoire is generated through a striking stochastic process called V(D)J recombination, in which different gene segments are assembled and modified, leading to extensive variety. In ovarian cancer (OC), an unfortunate 80% of cases are detected late, leading to poor survival outcomes. However, when detected early, approximately 94% of patients live longer than 5 years after diagnosis. Thus, early detection is critical for patient survival. To determine whether the TCR repertoire obtained from peripheral blood is associated with tumor status, we collected blood samples from 85 women with or without OC and obtained TCR information. We then used machine learning to learn the characteristics of samples and to finally predict, over a set of unseen samples, whether the person is with or without OC. We successfully stratified the two groups, thereby associating the peripheral blood TCR repertoire with the formation of OC tumors. A careful study of the origin of the set of T cells most informative for the signature indicated the involvement of a specific invariant natural killer T (iNKT) clone and a specific mucosal-associated invariant T (MAIT) clone. Our findings here support the proposition that tumor-relevant signal is maintained by the immune system and is coded in the T-cell repertoire available in peripheral blood. It is also possible that the immune system detects tumors early enough for repertoire technologies to inform us near the beginning of tumor formation. Although such detection is made by the immune system, we might be able to identify it, using repertoire data from peripheral blood, to offer a pragmatic way to search for early signs of cancer with minimal patient burden, possibly with enhanced sensitivity.

The temporal behavior of the murine T cell receptor repertoire following Immunotherapy.

Immunotherapy is now an essential tool for cancer treatment, and the unique features of an individual's T cell receptor repertoire are known to play a key role in its effectiveness. The repertoire, famously vast due to a cascade of cellular mechanisms, can be quantified using repertoire sequencing. In this study, we sampled the repertoire over several time points following treatment with anti-CTLA-4, in a syngeniec mouse model for colorectal cancer, generating a longitudinal dataset of T cell clones and their abundance. The dynamics of the repertoire can be observed in response to treatment over a period of four weeks, as clonal expansion of specific clones ascends and descends. The data made available here can be used to determine treatment and predict its effect, while also providing a unique look at the behavior of the immune system over time.

GENTLE: a novel bioinformatics tool for generating features and building classifiers from T cell repertoire cancer data.

BACKGROUND: In the global effort to discover biomarkers for cancer prognosis, prediction tools have become essential resources. TCR (T cell receptor) repertoires contain important features that differentiate healthy controls from cancer patients or differentiate outcomes for patients being treated with different drugs. Considering, tools that can easily and quickly generate and identify important features out of TCR repertoire data and build accurate classifiers to predict future outcomes are essential. RESULTS: This paper introduces GENTLE (GENerator of T cell receptor repertoire features for machine LEarning): an open-source, user-friendly web-application tool that allows TCR repertoire researchers to discover important features; to create classifier models and evaluate them with metrics; and to quickly generate visualizations for data interpretations. We performed a case study with repertoires of TRegs (regulatory T cells) and TConvs (conventional T cells) from healthy controls versus patients with breast cancer. We showed that diversity features were able to distinguish between the groups. Moreover, the classifiers built with these features could correctly classify samples ('Healthy' or 'Breast Cancer')from the TRegs repertoire when trained with the TConvs repertoire, and from the TConvs repertoire when trained with the TRegs repertoire. CONCLUSION: The paper walks through installing and using GENTLE and presents a case study and results to demonstrate the application's utility. GENTLE is geared towards any researcher working with TCR repertoire data and aims to discover predictive features from these data and build accurate classifiers. GENTLE is available on https://github.com/dhiego22/gentle and https://share.streamlit.io/dhiego22/gentle/main/gentle.py .

Distribution equality as an optimal epidemic mitigation strategy.

Upon the development of a therapeutic, a successful response to a global pandemic relies on efficient worldwide distribution, a process constrained by our global shipping network. Most existing strategies seek to maximize the outflow of the therapeutics, hence optimizing for rapid dissemination. Here we find that this intuitive approach is, in fact, counterproductive. The reason is that by focusing strictly on the quantity of disseminated therapeutics, these strategies disregard the way in which this quantity distributes across destinations. Most crucially-they overlook the interplay of the therapeutic spreading patterns with those of the pathogens. This results in a discrepancy between supply and demand, that prohibits efficient mitigation even under optimal conditions of superfluous flow. To solve this, we design a dissemination strategy that naturally follows the predicted spreading patterns of the pathogens, optimizing not just for supply volume, but also for its congruency with the anticipated demand. Specifically, we show that epidemics spread relatively uniformly across all destinations, prompting us to introduce an equality constraint into our dissemination that prioritizes supply homogeneity. This strategy may, at times, slow down the supply rate in certain locations, however, thanks to its egalitarian nature, which mimics the flow of the pathogens, it provides a dramatic leap in overall mitigation efficiency, potentially saving more lives with orders of magnitude less resources.

A T cell repertoire timestamp is at the core of responsiveness to CTLA-4 blockade.

Biology of the response to anti-CTLA-4 involves the dynamics of specific T cell clones. Reasons for clinical success and failure of this treatment are still largely unknown. Here, we quantified the dynamics of the T cell receptor (TCR) repertoire, throughout 4 weeks involving treatment with anti-CTLA-4, in a syngeneic mouse model for colorectal cancer. These dynamics show an initial increase in clonality in tandem with a decrease in diversity, effects which gradually subside. Furthermore, response to treatment is tightly connected to the shared and public parts of the T cell repertoire. We were able to recognize time-dependent behaviors of specific TCR sequences and cell types and to show the response is dominated by specific motifs. We see that a single, specific time point might be useful to inform a physician of the true response to treatmentThe research further highlights the importance of temporal analyses of the immune response.

Breast cancer is marked by specific, Public T-cell receptor CDR3 regions shared by mice and humans.

The partial success of tumor immunotherapy induced by checkpoint blockade, which is not antigen-specific, suggests that the immune system of some patients contain antigen receptors able to specifically identify tumor cells. Here we focused on T-cell receptor (TCR) repertoires associated with spontaneous breast cancer. We studied the alpha and beta chain CDR3 domains of TCR repertoires of CD4 T cells using deep sequencing of cell populations in mice and applied the results to published TCR sequence data obtained from human patients. We screened peripheral blood T cells obtained monthly from individual mice spontaneously developing breast tumors by 5 months. We then looked at identical TCR sequences in published human studies; we used TCGA data from tumors and healthy tissues of 1,256 breast cancer resections and from 4 focused studies including sequences from tumors, lymph nodes, blood and healthy tissues, and from single cell dataset of 3 breast cancer subjects. We now report that mice spontaneously developing breast cancer manifest shared, Public CDR3 regions in both their alpha and beta and that a significant number of women with early breast cancer manifest identical CDR3 sequences. These findings suggest that the development of breast cancer is associated, across species, with biomarker, exclusive TCR repertoires.

CDR3 and V genes show distinct reconstitution patterns in T cell repertoire post-allogeneic bone marrow transplantation.

Restoration of T cell repertoire diversity after allogeneic bone marrow transplantation (allo-BMT) is crucial for immune recovery. T cell diversity is produced by rearrangements of germline gene segments (V (D) and J) of the T cell receptor (TCR) α and ß chains, and selection induced by binding of TCRs to MHC-peptide complexes. Multiple measures were proposed for this diversity. We here focus on the V-gene usage and the CDR3 sequences of the beta chain. We compared multiple T cell repertoires to follow T cell repertoire changes post-allo-BMT in HLA-matched related donor and recipient pairs. Our analyses of the differences between donor and recipient complementarity determining region 3 (CDR3) beta composition and V-gene profile show that the CDR3 sequence composition does not change during restoration, implying its dependence on the HLA typing. In contrast, V-gene usage followed a time-dependent pattern, initially following the donor profile and then shifting back to the recipients' profile. The final long-term repertoire was more similar to that of the recipient's original one than the donor's; some recipients converged within months, while others took multiple years. Based on the results of our analyses, we propose that donor-recipient V-gene distribution differences may serve as clinical biomarkers for monitoring immune recovery.

Age-related loss of gene-to-gene transcriptional coordination among single cells.

A long-standing model holds that stochastic aberrations of transcriptional regulation play a key role in the process of ageing. While transcriptional dysregulation is observed in many cell types in the form of increased cell-to-cell variability, its generality to all cell types remains doubted. Here, we propose a new approach for analysing transcriptional regulation in single-cell RNA sequencing data by focusing on the global coordination between the genes rather than the variability of individual genes or correlations between pairs of genes. Consistently, across very different organisms and cell types, we find a decrease in the gene-to-gene transcriptional coordination in ageing cells. In addition, we find that loss of gene-to-gene transcriptional coordination is associated with high mutational load of a specific, age-related signature and with radiation-induced DNA damage. These observations suggest a general, potentially universal, stochastic attribute of transcriptional dysregulation in ageing.

Predicting and affecting response to cancer therapy based on pathway-level biomarkers.

Identifying robust, patient-specific, and predictive biomarkers presents a major obstacle in precision oncology. To optimize patient-specific therapeutic strategies, here we couple pathway knowledge with large-scale drug sensitivity, RNAi, and CRISPR-Cas9 screening data from 460 cell lines. Pathway activity levels are found to be strong predictive biomarkers for the essentiality of 15 proteins, including the essentiality of MAD2L1 in breast cancer patients with high BRCA-pathway activity. We also find strong predictive biomarkers for the sensitivity to 31 compounds, including BCL2 and microtubule inhibitors (MTIs). Lastly, we show that Bcl-xL inhibition can modulate the activity of a predictive biomarker pathway and re-sensitize lung cancer cells and tumors to MTI therapy. Overall, our results support the use of pathways in helping to achieve the goal of precision medicine by uncovering dozens of predictive biomarkers.

A single nucleotide variant of human PARP1 determines response to PARP inhibitors.

The introduction of novel cancer drugs and innovative treatments brings great hope for cancer patients, but also an urgent need to match drugs to suitable patients, since certain drugs that benefit one patient may actually harm others. The newly developed poly-ADP ribose polymerase (PARP) inhibitors (PARPis) are a group of pharmacological enzyme inhibitors used clinically for multiple indications. Several forms of cancer tend to be PARP dependent, making PARP an attractive target for cancer therapy. Specifically, PARPis are commonly used in BRCA-associated breast cancers patients, since unrepaired single-strand breaks are converted into double-strand breaks and BRCA-associated tumors cannot repair them by homologous recombination so that PARPi leads to tumor cell death, by a mechanism called "Synthetic Lethality". Unfortunately, not all patients respond to PARPi, and it is not currently possible to predict who will or will not respond. Here, we present a specific genomic marker, which reflects a single-nucleotide polymorphism of human PARP1 and correlates in vitro with response to PARPi, throughout all indications. In addition, we report that this SNP is associated with re-shaping mRNA, and mRNA levels, and influences the final protein structure to expose new binding sites while hiding others. The status of the SNP is therefore critical to patients' care, as it relates responses to PARPi to the PARP1-SNP carried.

Immune Computation and COVID-19 Mortality: A Rationale for IVIg.

COVID-19 infection tends to be more lethal in older persons than in the young; death results from an overactive inflammatory response, leading to cytokine storm and organ failure. Here we describe immune regulation of the inflammatory response phenotype as emerging from a process that is analogous to machine-learning algorithms used in computers. We briefly describe some strategic similarities between immune learning and computer machine learning. We reason that a balanced response to COVID-19 infection might be induced by treating the elderly patient with a wellness repertoire of antibodies obtained from healthy young people. We propose that a beneficial training set of such antibodies might be administered in the form of intravenous immunoglobulin (IVIg).

A modeling algorithm for exploring the architecture and construction of bird nests.

The wide variety of nest architectural designs exhibited by passerine birds allowed them to diversify into a wide variety of ecological niches and terrestrial habitats. At present, very little is known about the mechanics of building these structures. Digitizing natural biological structures such as bird nests provides the opportunity to explore their structural properties and behavior under specific conditions by means of computational manipulations, simulations, and analyses. This study describes a generic algorithm for the digitization and exploration of complex interlocked bird nests, and validates it on nests built by the Dead-Sea Sparrow (Passer moabiticus) in branches of trees using stiff dry branches. This algorithm takes as input computerized tomographic scans of the nest, identifies and isolates each branch entity within the three-dimensional data, and finally extracts the characteristics of each branch. The result is a reliable three-dimensional digital model of the nest that contains a complete geometric dataset per each of its components, e.g. dimensions and contact points with neighboring components, as well as global properties, e.g. density distribution and network structure. Based on these, we were able to simulate various models of the nest construction process. Altogether, the described algorithm and possible derivatives thereof could be a valuable tool in studying the structure-function relationships of similarly complex biological objects, and may provide further insights into the potential selective mechanisms underlying historical evolution of this distinct nest form.

Digitizable therapeutics for decentralized mitigation of global pandemics.

When confronted with a globally spreading epidemic, we seek efficient strategies for drug dissemination, creating a competition between supply and demand at a global scale. Propagating along similar networks, e.g., air-transportation, the spreading dynamics of the supply vs. the demand are, however, fundamentally different, with the pathogens driven by contagion dynamics, and the drugs by commodity flow. We show that these different dynamics lead to intrinsically distinct spreading patterns: while viruses spread homogeneously across all destinations, creating a concurrent global demand, commodity flow unavoidably leads to a highly uneven spread, in which selected nodes are rapidly supplied, while the majority remains deprived. Consequently, even under ideal conditions of extreme production and shipping capacities, due to the inherent heterogeneity of network-based commodity flow, efficient mitigation becomes practically unattainable, as homogeneous demand is met by highly heterogeneous supply. Therefore, we propose here a decentralized mitigation strategy, based on local production and dissemination of therapeutics, that, in effect, bypasses the existing distribution networks. Such decentralization is enabled thanks to the recent development of digitizable therapeutics, based on, e.g., short DNA sequences or printable chemical compounds, that can be distributed as digital sequence files and synthesized on location via DNA/3D printing technology. We test our decentralized mitigation under extremely challenging conditions, such as suppressed local production rates or low therapeutic efficacy, and find that thanks to its homogeneous nature, it consistently outperforms the centralized alternative, saving many more lives with significantly less resources.

Astrocyte-specific transcriptome analysis using the ALDH1L1 bacTRAP mouse reveals novel biomarkers of astrogliosis in response to neurotoxicity.

Neurotoxicology is hampered by the inability to predict regional and cellular targets of toxicant-induced damage. Evaluating astrogliosis overcomes this problem because reactive astrocytes highlight the location of toxicant-induced damage. While enhanced expression of glial fibrillary acidic protein is a hallmark of astrogliosis, few other biomarkers have been identified. However, bacterial artificial chromosome - translating ribosome affinity purification (bacTRAP) technology allows for characterization of the actively translating transcriptome of a particular cell type; use of this technology in aldehyde dehydrogenase 1 family member L1 (ALDH1L1) bacTRAP mice can identify genes selectively expressed in astrocytes. The aim of this study was to characterize additional biomarkers of neurotoxicity-induced astrogliosis using ALDH1L1 bacTRAP mice. The known dopaminergic neurotoxicant 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP; 12.5 mg/kg s.c.) was used to induce astrogliosis. Striatal tissue was obtained 12, 24, and 48 h following exposure for the isolation of actively translating RNA. Subsequently, MPTP-induced changes in this RNA pool were analyzed by microarray and 184 statistically significant, differentially expressed genes were identified. The dataset was interrogated by gene ontology, pathway, and co-expression network analyses, which identified novel genes, as well as those with known immune and inflammatory functions. Using these analyses, we were directed to several genes associated with reactive astrocytes. Of these, TIMP1 and miR-147 were identified as candidate biomarkers because of their robust increased expression following both MPTP and trimethyl tin exposures. Thus, we have demonstrated that bacTRAP can be used to identify new biomarkers of astrogliosis and aid in the characterization of astrocyte phenotypes induced by toxicant exposures. OPEN SCIENCE BADGES: This article has received a badge for *Open Materials* because it provided all relevant information to reproduce the study in the manuscript. The complete Open Science Disclosure form for this article can be found at the end of the article. More information about the Open Practices badges can be found at https://cos.io/our-services/open-science-badges/. Cover Image for this issue: doi: 10.1111/jnc.14518.

Resistance to paclitaxel is associated with a variant of the gene BCL2 in multiple tumor types.

Paclitaxel, the most commonly used form of chemotherapy, is utilized in curative protocols in different types of cancer. The response to treatment differs among patients. Biological interpretation of a mechanism to explain this personalized response is still unavailable. Since paclitaxel is known to target BCL2 and TUBB1, we used pan-cancer genomic data from hundreds of patients to show that a single-nucleotide variant in the BCL2 sequence can predict a patient's response to paclitaxel. Here, we show a connection between this BCL2 genomic variant, its transcript structure, and protein abundance. We demonstrate these findings in silico, in vitro, in formalin-fixed paraffin-embedded (FFPE) tissue, and in patient lymphocytes. We show that tumors with the specific variant are more resistant to paclitaxel. We also show that tumor and normal cells with the variant express higher levels of BCL2 protein, a phenomenon that we validated in an independent cohort of patients. Our results indicate BCL2 sequence variations as determinants of chemotherapy resistance. The knowledge of individual BCL2 genomic sequences prior to the choice of chemotherapy may improve patient survival. The current work also demonstrates the benefit of community-wide, integrative omics data sources combined with in-lab experimentation and validation sets.

Design principles of biologically fabricated avian nests.

Materials and construction methods of nests vary between bird species and at present, very little is known about the relationships between architecture and function in these structures. This study combines computational and experimental techniques to study the structural biology of nests fabricated by the edible nest swiftlet Aerodramus fuciphagus on vertical rock walls using threaded saliva. Utilizing its own saliva as a construction material allows the swiftlets full control over the structural features at a very high resolution in a process similar to additive manufacturing. It was hypothesized that the mechanical properties would vary between the structural regions of the nest (i.e. anchoring to the wall, center of the cup, and rim) mainly by means of architecture to offer structural support and bear the natural loads of birds and eggs. We generated numerical models of swiftlet nests from µCT scans based on collected swiftlet nests, which we loaded with a force of birds and eggs. This was done in order to study and assess the stress distribution that characterizes the specific nest's architecture, evaluate its strength and weak points if any, as well as to understand the rationale and benefits that underlie this natural structure. We show that macro- and micro-scale structural patterns are identical in all nests, suggesting that their construction is governed by specific design principles. The nests' response to applied loads of birds and eggs in finite element simulations suggests a mechanical overdesign strategy, which ensures the stresses experienced by its components in any loading scenario are actively minimized to be significantly smaller than the tensile fracture strength of the nests' material. These findings highlight mechanical overdesign as a biological strategy for resilient, single-material constructions designed to protect eggs and hatchlings.

The Immune System Computes the State of the Body: Crowd Wisdom, Machine Learning, and Immune Cell Reference Repertoires Help Manage Inflammation.

Here, we outline an overview of the mammalian immune system that updates and extends the classical clonal selection paradigm. Rather than focusing on strict self-not-self discrimination, we propose that the system orchestrates variable inflammatory responses that maintain the body and its symbiosis with the microbiome while eliminating the threat from pathogenic infectious agents and from tumors. The paper makes four points: The immune system classifies healthy and pathologic states of the body-including both self and foreign elements-by deploying individual lymphocytes as cellular computing machines; immune cells transform input signals from the body into an output of specific immune reactions.Rather than independent clonal responses, groups of individually activated immune-system cells co-react in lymphoid organs to make collective decisions through a type of self-organizing swarm intelligence or crowd wisdom.Collective choices by swarms of immune cells, like those of schools of fish, are modified by relatively small numbers of individual regulators responding to shifting conditions-such collective inflammatory responses are dynamically responsive.Self-reactive autoantibody and T-cell receptor (TCR) repertoires shared by healthy individuals function in a biological version of experience-based supervised machine learning. Immune system decisions are primed by formative experience with training sets of self-antigens encountered during lymphocyte development; these initially trained T cell and B cell repertoires form a Wellness Profile that then guides immune responses to test sets of antigens encountered later. This experience-based machine learning strategy is analogous to that deployed by supervised machine-learning algorithms. We propose experiments to test these ideas. This overview of the immune system bears clinical implications for monitoring wellness and for treating autoimmune disease, cancer, and allograft reactions.

Comparing Transcriptome Profiles of Neurons Interfacing Adjacent Cells and Nanopatterned Substrates Reveals Fundamental Neuronal Interactions.

Developing neuronal axons are directed by chemical and physical signals toward a myriad of target cells. According to current dogma, the resulting network architecture is critically shaped by electrical interconnections, the synapses; however, key mechanisms translating neuronal interactions into neuronal growth behavior during network formation are still unresolved. To elucidate these mechanisms, we examined neurons interfacing nanopatterned substrates and compared them to natural interneuron interactions. We grew similar neuronal populations under three connectivity conditions, (1) the neurons are isolated, (2) the neurons are interconnected, and (3) the neurons are connected only to artificial substrates, then quantitatively compared both the cell morphologies and the transcriptome-expression profiles. Our analysis shows that whereas axon-guidance signaling pathways in isolated neurons are predominant, in isolated neurons interfacing nanotopography, these pathways are downregulated, similar to the interconnected neurons. Moreover, in nanotopography, interfacing neuron genes related to synaptogenesis and synaptic regulation are highly expressed, that is, again resembling the behavior of interconnected neurons. These molecular findings demonstrate that interactions with nanotopographies, although not leading to electrical coupling, play a comparable functional role in two major routes, neuronal guidance and network formation, with high relevance to the design of regenerative interfaces.

Immunological analysis of phase II glioblastoma dendritic cell vaccine (Audencel) trial: immune system characteristics influence outcome and Audencel up-regulates Th1-related immunovariables.

Audencel is a dendritic cell (DC)-based cellular cancer immunotherapy against glioblastoma multiforme (GBM). It is characterized by loading of DCs with autologous whole tumor lysate and in vitro maturation via "danger signals". The recent phase II "GBM-Vax" trial showed no clinical efficacy for Audencel as assessed with progression-free and overall survival in all patients. Here we present immunological research accompanying the trial with a focus on immune system factors related to outcome and Audencel's effect on the immune system. Methodologically, peripheral blood samples (from apheresis before Audencel or venipuncture during Audencel) were subjected to functional characterization via enzyme-linked immunospot (ELISPOT) assays connected with cytokine bead assays (CBAs) as well as phenotypical characterization via flow cytometry and mRNA quantification. GBM tissue samples (from surgery) were subjected to T cell receptor sequencing and immunohistochemistry. As results we found: Patients with favorable pre-existing anti-tumor characteristics lived longer under Audencel than Audencel patients without them. Pre-vaccination blood CD8+ T cell count and ELISPOT Granzyme B production capacity in vitro upon tumor antigen exposure were significantly correlated with overall survival. Despite Audencel's general failure to induce a significant clinical response, it nevertheless seemed to have an effect on the immune system. For instance, Audencel led to a significant up-regulation of the Th1-related immunovariables ELISPOT IFNγ, the transcription factor T-bet in the blood and ELISPOT IL-2 in a dose-dependent manner upon vaccination. Post-vaccination levels of ELISPOT IFNγ and CD8+ cells in the blood were indicative of a significantly better survival. In summary, Audencel failed to reach an improvement of survival in the recent phase II clinical trial. No clinical efficacy was registered. Our concomitant immunological work presented here indicates that outcome under Audencel was influenced by the state of the immune system. On the other hand, Audencel also seemed to have stimulated the immune system. Overall, these immunological considerations suggest that DC immunotherapy against glioblastoma should be studied further - with the goal of translating an apparent immunological response into a clinical response. Future research should concentrate on investigating augmentation of immune reactions through combination therapies or on developing meaningful biomarkers.

Tamoxifen-resistant breast cancer cells are resistant to DNA-damaging chemotherapy because of upregulated BARD1 and BRCA1.

Tamoxifen resistance is accountable for relapse in many ER-positive breast cancer patients. Most of these recurrent patients receive chemotherapy, but their chemosensitivity is unknown. Here, we report that tamoxifen-resistant breast cancer cells express significantly more BARD1 and BRCA1, leading to resistance to DNA-damaging chemotherapy including cisplatin and adriamycin, but not to paclitaxel. Silencing BARD1 or BRCA1 expression or inhibition of BRCA1 phosphorylation by Dinaciclib restores the sensitivity to cisplatin in tamoxifen-resistant cells. Furthermore, we show that activated PI3K/AKT pathway is responsible for the upregulation of BARD1 and BRCA1. PI3K inhibitors decrease the expression of BARD1 and BRCA1 in tamoxifen-resistant cells and re-sensitize them to cisplatin both in vitro and in vivo. Higher BARD1 and BRCA1 expression is associated with worse prognosis of early breast cancer patients, especially the ones that received radiotherapy, indicating the potential use of PI3K inhibitors to reverse chemoresistance and radioresistance in ER-positive breast cancer patients.