Explore the vast range of titles available.

Background Breast cancer is a highly heterogeneous disease where variations of biomarker expression may exist between individual foci of a cancer (intra-tumoral heterogeneity). The extent of variation of biomarker expression in the cancer cells, distribution of cell types in the local tumor microenvironment and their spatial arrangement could impact on diagnosis, treatment planning and subsequent response to treatment. Methods Using quantitative multiplex immunofluorescence (MxIF) imaging, we assessed the level of variations in biomarker expression levels among individual cells, density of cell cluster groups and spatial arrangement of immune subsets from regions sampled from 38 multi-focal breast cancers that were processed using whole-mount histopathology techniques. Molecular profiling was conducted to determine the intrinsic molecular subtype of each analysed region. Results A subset of cancers (34.2%) showed intra-tumoral regions with more than one molecular subtype classification. High levels of intra-tumoral variations in biomarker expression levels were observed in the majority of cancers studied, particularly in Luminal A cancers. HER2 expression quantified with MxIF did not correlate well with HER2 gene expression, nor with clinical HER2 scores. Unsupervised clustering revealed the presence of various cell clusters with unique IHC4 protein co-expression patterns and the composition of these clusters were mostly similar among intra-tumoral regions. MxIF with immune markers and image patch analysis classified immune niche phenotypes and the prevalence of each phenotype in breast cancer subtypes was illustrated. Conclusions Our work illustrates the extent of spatial heterogeneity in biomarker expression and immune phenotypes, and highlights the importance of a comprehensive spatial assessment of the disease for prognosis and treatment planning.

Prognostic information is essential for decision-making in breast cancer management. In recent years, trials and clinical practice have emphasized genomic prognostication tools, despite clinicopathological methods being more affordable and accessible. PREDICT v3 is one such tool with promising results across cohorts. Advances in machine learning (ML), transfer learning, and ensemble methods provide opportunities to enhance these approaches, especially where missing data and model assumptions differ across diverse populations. This study evaluates the potential to improve survival prognostication in breast cancer. More precisely, we compare de novo ML, transfer learning from the pretrained prognostication model PREDICT v3, and a stacked ensemble approach. Data from the MA.27 trial (NCT00066573) were used for model training, with external validation on data from the Tamoxifen Exemestane Adjuvant Multinational trial (NCT00279448 and NCT00032136) and a US Surveillance, Epidemiology, and End Results cohort. Transfer learning was applied by re-estimating the parameters of the pretrained prognostic tool PREDICT v3. De novo ML included random survival forests and extreme gradient boosting, and the ensemble was implemented using weighted linear stacking of model predictions. Internal and external validation was assessed in terms of the integrated calibration index and discrimination. Shapley Additive Explanations values were used to explain model predictions and decision-curve analysis to facilitate the interpretation of performance differences. Transfer learning, de novo random survival forest, and the stacked ensemble improved calibration in MA.27 over the pretrained model (integrated calibration index reduced from 0.042 in PREDICT v3 to ≤0.007) while discrimination remained comparable (AUROC increased from 0.738 in PREDICT v3 to 0.744-0.799). In decision-curve analysis, these approaches demonstrated consistently positive net benefit across clinically relevant thresholds, while PREDICT v3 lost net benefit beyond 7.5% predicted risk. Invalid PREDICT v3 predictions were observed in 23.8% to 25.8% of MA.27 individuals due to missing information. In contrast, ML models and the stacked ensemble predicted survival despite missing data. Across all models, patient age, nodal status, pathological grading, and tumor size had the highest Shapley Additive Explanations values, indicating their importance for survival prognostication. External validation in the US Surveillance, Epidemiology, and End Results cohort confirmed the benefits of transfer learning, RSF, and ensemble in terms of calibration while maintaining discrimination at comparable levels. In contrast, generalizability was limited in the Tamoxifen Exemestane Adjuvant Multinational trial, a cohort with a substantially different distribution of clinicopathological characteristics. This study demonstrates that transfer learning, de novo RSF, and a stacked ensemble can improve prognostication compared with the pretrained PREDICT v3, particularly in the presence of missing or uncertain inputs. Transportability may be limited in cohorts with different clinicopathological profiles, requiring local validation before clinical deployment. Ultimately, better survival estimation can provide more meaningful guidance in breast cancer care. ClinicalTrials.gov NCT00066573; https://clinicaltrials.gov/study/NCT00066573, NCT00279448; https://clinicaltrials.gov/study/NCT00279448, NCT00032136; https://clinicaltrials.gov/study/NCT00032136.

Background A key step in cancer genome analysis is the identification of somatic mutations in the tumor. This is typically done by comparing the genome of the tumor to the reference genome sequence derived from a normal tissue taken from the same donor. However, there are a variety of common scenarios in which matched normal tissue is not available for comparison. Results In this work, we describe an algorithm to distinguish somatic single nucleotide variants (SNVs) in next-generation sequencing data from germline polymorphisms in the absence of normal samples using a machine learning approach. Our algorithm was evaluated using a family of supervised learning classifications across six different cancer types and ~1600 samples, including cell lines, fresh frozen tissues, and formalin-fixed paraffin-embedded tissues; we tested our algorithm with both deep targeted and whole-exome sequencing data. Our algorithm correctly classified between 95 and 98% of somatic mutations with F1-measure ranges from 75.9 to 98.6% depending on the tumor type. We have released the algorithm as a software package called ISOWN (Identification of SOmatic mutations Without matching Normal tissues). Conclusions In this work, we describe the development, implementation, and validation of ISOWN, an accurate algorithm for predicting somatic mutations in cancer tissues in the absence of matching normal tissues. ISOWN is available as Open Source under Apache License 2.0 from https://github.com/ikalatskaya/ISOWN .

Approximately 80% of all breast cancer cases are estrogen receptor-positive (ER+). This subtype is known to have distant recurrences in a subset of patients after adjuvant endocrine therapy, partially due to the heterogeneity of the disease. ER+ breast cancer has been generally classified as an immune-cold disease. Thus, to better inform treatment decisions and to consider the prospect of immunotherapy, the immune microenvironment needs to be thoroughly characterized. In this study, the proteome and transcriptome of the tumour and tumour microenvironment (TME) were characterized using the GeoMx Digital Spatial Profiler (DSP) and NanoString’s BC360 gene expression panel. Spatially resolved tumour and TME across a patient's lumpectomy demonstrated substantial heterogeneity in the expression of commonly targeted immune and tumour proteins. Results from this study demonstrated heterogeneity across the tumour and TME in ER+ breast cancer, which may be reflective of a variable immune response.

Background Ovarian cancer is the leading gynecologic cancer diagnosed in North America and because related symptoms are not disease specific, this often leads to late detection, an advanced disease state, and the need for chemotherapy. Ovarian cancer is frequently sensitive to chemotherapy at diagnosis but rapid development of drug resistance leads to disease progression and ultimately death in the majority of patients. Results We have generated paclitaxel resistant ovarian cell lines from their corresponding native cell lines to determine driver mechanisms of drug resistance using gene expression arrays. These paclitaxel resistant ovarian cells demonstrate: (1) Increased IC 50 for paclitaxel and docetaxel (10 to 75-fold) and cross-resistance to anthracyclines (2) Reduced cell apoptosis in the presence of paclitaxel (3) Gene depletion involving mitotic regulators BUB1 mitotic checkpoint serine/threonine kinase, cyclin BI (CCNB1), centromere protein E (CENPE), and centromere protein F (CENPF), and (4) Functional data validating gene depletion among mitotic regulators. Conclusions We have generated model systems to explore drug resistance in ovarian cancer, which have revealed a key pathway related to the spindle assembly checkpoint underlying paclitaxel resistance in ovarian cell lines.

HER2 overexpression/amplification is linked with poor prognosis in early breast cancer. Co-expression of HER2 and HER3 is associated with endocrine and chemotherapy resistance, driven not simply by expression but by signalling via HER2:HER3 or HER2:HER2 dimers. Proximity ligation assays (PLAs) detect protein–protein complexes at a single-molecule level and allow study of signalling pathways in situ. A cohort of 100 tumours was analyzed by PLA, IHC and FISH. HER complexes were analyzed by PLA in a further 321 tumours from the BR9601 trial comparing cyclophosphamide, methotrexate and fluorouracil (CMF) with epirubicin followed by CMF (epi-CMF). The relationships between HER dimer expression and RFS and OS were investigated, and multivariate regression analysis identified factors influencing patient prognosis. PLA successfully and reproducibly detected HER2:HER2 and HER2:HER3 protein complexes in vivo. A significant association ( P  < 0.00001) was identified between HER2 homodimerization and HER2 gene amplification. Following a minimum p value approach high levels of HER2:HER2 dimers were significantly associated with reduced relapse-free (RFS; hazard ratio = 1.72, 95% confidence interval 1.15–2.56, P  = 0.008) and overall survival (OS HR = 1.69 95% CI = 1.09–2.62, P  = 0.019). Similarly, high levels of HER2:HER3 dimers were associated with reduced RFS (HR = 2.18, 95% CI = 1.46–3.26, P  = 0.00016) and OS (HR = 2.21, 95% CI = 1.41–3.47, P  = 0.001). This study demonstrates that in situ detection of HER2 and HER2:3 protein:protein complexes can be performed robustly and reproducibly in clinical specimens, provides novel prognostic information and opens a significant novel opportunity to probe the clinical impact of cellular signalling processes.

Ki67 has potential clinical importance in breast cancer but has yet to see broad acceptance due to inter-laboratory variability. Here we tested an open source and calibrated automated digital image analysis (DIA) platform to: (i) investigate the comparability of Ki67 measurement across corresponding core biopsy and resection specimen cases, and (ii) assess section to section differences in Ki67 scoring. Two sets of 60 previously stained slides containing 30 core-cut biopsy and 30 corresponding resection specimens from 30 estrogen receptor-positive breast cancer patients were sent to 17 participating labs for automated assessment of average Ki67 expression. The blocks were centrally cut and immunohistochemically (IHC) stained for Ki67 (MIB-1 antibody). The QuPath platform was used to evaluate tumoral Ki67 expression. Calibration of the DIA method was performed as in published studies. A guideline for building an automated Ki67 scoring algorithm was sent to participating labs. Very high correlation and no systematic error (p = 0.08) was found between consecutive Ki67 IHC sections. Ki67 scores were higher for core biopsy slides compared to paired whole sections from resections (p ≤ 0.001; median difference: 5.31%). The systematic discrepancy between core biopsy and corresponding whole sections was likely due to pre-analytical factors (tissue handling, fixation). Therefore, Ki67 IHC should be tested on core biopsy samples to best reflect the biological status of the tumor.

BackgroundPancreatic ductal adenocarcinoma (PDAC) is a malignant neoplasm of the pancreas characterized by late-stage detection, with few treatment options and prognostic biomarkers. Tertiary lymphoid structures (TLS) have been found to be predictive of survival in PDAC, however, their phenotypic heterogeneity, functional significance, and relationship to genomic and transcriptomic subtypes remain poorly understood.MethodsUsing TLS histopathology detection methods we identified TLS across a cohort of over 600 patients and linked their presence to matched genomics and clinical metadata. We generated GeoMx spatial transcriptomics data for 13 PDAC patients with over 200 regions of interest (ROIs) focused on TLS and tumour. We stained and digitized matched whole-slide hematoxylin and eosin (H&E) images and single-cell spatial proteomic data using Imaging Mass Cytometry (IMC) on serial sections of the same ROIs profiled with GeoMx. Using unsupervised clustering and differential expression analyses we characterized TLS into subgroups.ResultsWe show that we can automate TLS detection in H&E from primary tumour resections and liver metastases biopsies. We quantified the percentage of PDAC samples with TLS, the number of TLS per patient and linked their presence to PDAC transcriptomic subtypes, genomic aberrations and patient metadata. We identified three distinct TLS subgroups based on whole-transcriptome TLS expression profiles, and find that TLS subgroup identity is determined by tumour proximity, and specific pathway activation within the adjacent tumour. Finally, we use IMC to deconvolve the single cell content of each subgroup.ConclusionsTLS are a known prognostic factor in PDAC, however, they have never been thoroughly characterized or linked to genomic and transcriptomic PDAC subtypes. We find previously unappreciated heterogeneity in TLS phenotypes and link these to tumour phenotypes in what is the most comprehensive characterization of TLS to date.Ethics ApprovalSamples were taken from previous studies with patient informed consent and approval from Institutional Review or Research Ethics Boards (REB # 20-0170-E, Sinai Health).

Abstract Ki67 has potential clinical importance in breast cancer but has yet to see broad acceptance due to inter-laboratory variability. Here we tested an open source and calibrated automated digital image analysis (DIA) platform to: (i) investigate the comparability of Ki67 measurement across corresponding core biopsy and resection specimen cases, and (ii) assess section to section differences in Ki67 scoring. Two sets of 60 previously stained slides containing 30 core-cut biopsy and 30 corresponding resection specimens from 30 estrogen receptor-positive breast cancer patients were sent to 17 participating labs for automated assessment of average Ki67 expression. The blocks were centrally cut and immunohistochemically (IHC) stained for Ki67 (MIB-1 antibody). The QuPath platform was used to evaluate tumoral Ki67 expression. Calibration of the DIA method was performed as in published studies. A guideline for building an automated Ki67 scoring algorithm was sent to participating labs. Very high correlation and no systematic error ( p = 0.08) was found between consecutive Ki67 IHC sections. Ki67 scores were higher for core biopsy slides compared to paired whole sections from resections ( p ≤ 0.001; median difference: 5.31%). The systematic discrepancy between core biopsy and corresponding whole sections was likely due to pre-analytical factors (tissue handling, fixation). Therefore, Ki67 IHC should be tested on core biopsy samples to best reflect the biological status of the tumor.