Pathway-based signatures predict patient outcome, chemotherapy benefit and synthetic lethal dependencies in invasive lobular breast cancer.

BACKGROUND: Invasive Lobular Carcinoma (ILC) is a morphologically distinct breast cancer subtype that represents up to 15% of all breast cancers. Compared to Invasive Breast Carcinoma of No Special Type (IBC-NST), ILCs exhibit poorer long-term outcome and a unique pattern of metastasis. Despite these differences, the systematic discovery of robust prognostic biomarkers and therapeutically actionable molecular pathways in ILC remains limited. METHODS: Pathway-centric multivariable models using statistical machine learning were developed and tested in seven retrospective clinico-genomic cohorts (n = 996). Further external validation was performed using a new RNA-Seq clinical cohort of aggressive ILCs (n = 48). RESULTS AND CONCLUSIONS: mRNA dysregulation scores of 25 pathways were strongly prognostic in ILC (FDR-adjusted P < 0.05). Of these, three pathways including Cell-cell communication, Innate immune system and Smooth muscle contraction were also independent predictors of chemotherapy response. To aggregate these findings, a multivariable machine learning predictor called PSILC was developed and successfully validated for predicting overall and metastasis-free survival in ILC. Integration of PSILC with CRISPR-Cas9 screening data from breast cancer cell lines revealed 16 candidate therapeutic targets that were synthetic lethal with high-risk ILCs. This study provides interpretable prognostic and predictive biomarkers of ILC which could serve as the starting points for targeted drug discovery for this disease.

Epigenetic activation of SOX11 is associated with recurrence and progression of ductal carcinoma in situ to invasive breast cancer.

BACKGROUND: Risk of recurrence and progression of ductal carcinoma in situ (DCIS) to invasive cancer remains uncertain, emphasizing the need for developing predictive biomarkers of aggressive DCIS. METHODS: Human cell lines and mouse models of disease progression were analyzed for candidate risk predictive biomarkers identified and validated in two independent DCIS cohorts. RESULTS: RNA profiling of normal mammary and DCIS tissues (n = 48) revealed that elevated SOX11 expression correlates with MKI67, EZH2, and DCIS recurrence score. The 21T human cell line model of DCIS progression to invasive cancer and two mouse models developing mammary intraepithelial neoplasia confirmed the findings. AKT activation correlated with chromatin accessibility and EZH2 enrichment upregulating SOX11 expression. AKT and HER2 inhibitors decreased SOX11 expression along with diminished mammosphere formation. SOX11 was upregulated in HER2+ and basal-like subtypes (P < 0.001). Longitudinal DCIS cohort (n = 194) revealed shorter recurrence-free survival in SOX11+ than SOX11- patients (P = 0.0056 in all DCIS; P < 0.0001 in HER2+ subtype) associated with increased risk of ipsilateral breast event/IBE (HR = 1.9, 95%CI = 1.2-2.9; P = 0.003). DISCUSSION: Epigenetic activation of SOX11 drives recurrence of DCIS and progression to invasive cancer, suggesting SOX11 as a predictive biomarker of IBE.

Development of a deep-learning model tailored for HER2 detection in breast cancer to aid pathologists in interpreting HER2-low cases.

AIMS: Over 50% of breast cancer cases are "Human epidermal growth factor receptor 2 (HER2) low breast cancer (BC)", characterized by HER2 immunohistochemistry (IHC) scores of 1+ or 2+ alongside no amplification on fluorescence in situ hybridization (FISH) testing. The development of new anti-HER2 antibody-drug conjugates (ADCs) for treating HER2-low breast cancers illustrates the importance of accurately assessing HER2 status, particularly HER2-low breast cancer. In this study we evaluated the performance of a deep-learning (DL) model for the assessment of HER2, including an assessment of the causes of discordances of HER2-Null between a pathologist and the DL model. We specifically focussed on aligning the DL model rules with the ASCO/CAP guidelines, including stained cells' staining intensity and completeness of membrane staining. METHODS AND RESULTS: We trained a DL model on a multicentric cohort of breast cancer cases with HER2-IHC scores (n = 299). The model was validated on two independent multicentric validation cohorts (n = 369 and n = 92), with all cases reviewed by three senior breast pathologists. All cases underwent a thorough review by three senior breast pathologists, with the ground truth determined by a majority consensus on the final HER2 score among the pathologists. In total, 760 breast cancer cases were utilized throughout the training and validation phases of the study. The model's concordance with the ground truth (ICC = 0.77 [0.68-0.83]; Fisher P = 1.32e-10) is higher than the average agreement among the three senior pathologists (ICC = 0.45 [0.17-0.65]; Fisher P = 2e-3). In the two validation cohorts, the DL model identifies 95% [93% - 98%] and 97% [91% - 100%] of HER2-low and HER2-positive tumours, respectively. Discordant results were characterized by morphological features such as extended fibrosis, a high number of tumour-infiltrating lymphocytes, and necrosis, whilst some artefacts such as nonspecific background cytoplasmic stain in the cytoplasm of tumour cells also cause discrepancy. CONCLUSION: Deep learning can support pathologists' interpretation of difficult HER2-low cases. Morphological variables and some specific artefacts can cause discrepant HER2-scores between the pathologist and the DL model.

Mutation detection in saliva from oral cancer patients.

OBJECTIVES: The incidence of head and neck squamous cell carcinoma (HNSCC) continues to increase and although advances have been made in treatment, it still has a poor overall survival with local relapse being common. Conventional imaging methods are not efficient at detecting recurrence at an early stage when still potentially curable. The aim of this study was to test the feasibility of using saliva to detect the presence of oral squamous cell carcinoma (OSCC) and to provide additional evidence for the potential of this approach. MATERIALS AND METHODS: Fresh tumor, whole blood and saliva were collected from patients with OSCC before treatment. Whole exome sequencing (WES) or gene panel sequencing of tumor DNA was performed to identify somatic mutations in tumors and to select genes for performing gene panel sequencing on saliva samples. RESULTS: The most commonly mutated genes identified in primary tumors by DNA sequencing were TP53 and FAT1. Gene panel sequencing of paired saliva samples detected tumor derived mutations in 9 of 11 (82%) patients. The mean variant allele frequency for the mutations detected in saliva was 0.025 (range 0.004 - 0.061). CONCLUSION: Somatic tumor mutations can be detected in saliva with high frequency in OSCC irrespective of site or stage of disease using a limited panel of genes. This work provides additional evidence for the suitability of using saliva as liquid biopsy in OSCC and has the potential to improve early detection of recurrence in OSCC. Trials are currently underway comparing this approach to standard imaging techniques.

Reporting on invasive lobular breast cancer in clinical trials: a systematic review.

Invasive lobular breast cancer (ILC) differs from invasive breast cancer of no special type in many ways. Evidence on treatment efficacy for ILC is, however, lacking. We studied the degree of documentation and representation of ILC in phase III/IV clinical trials for novel breast cancer treatments. Trials were identified on Pubmed and clinicaltrials.gov. Inclusion/exclusion criteria were reviewed for requirements on histological subtype and tumor measurability. Documentation of ILC was assessed and ILC inclusion rate, central pathology and subgroup analyses were evaluated. Inclusion restrictions concerning tumor measurability were found in 39/93 manuscripts. Inclusion rates for ILC were documented in 13/93 manuscripts and varied between 2.0 and 26.0%. No central pathology for ILC was reported and 3/13 manuscripts had ILC sub-analyses. ILC is largely disregarded in most trials with poor representation and documentation. The current inclusion criteria using RECIST v1.1, fall short in recognizing the unique non-measurable metastatic infiltration of ILC.