A multi-step classifier addressing cohort heterogeneity improves performance of prognostic biomarkers in three cancer types

Construction and optimization of multi-platform precision pathways for precision medicine

In the enduring challenge against disease, advancements in medical technology have empowered clinicians with novel diagnostic platforms. Whilst in some cases, a single test may provide a confident diagnosis, often additional tests are required. However, to strike a balance between diagnostic accuracy and cost-effectiveness, one must rigorously construct the clinical pathways. Here, we developed a framework to build multi-platform precision pathways in an automated, unbiased way, recommending the key steps a clinician would take to reach a diagnosis. We achieve this by developing a confidence score, used to simulate a clinical scenario, where at each stage, either a confident diagnosis is made, or another test is performed. Our framework provides a range of tools to interpret, visualize and compare the pathways, improving communication and enabling their evaluation on accuracy and cost, specific to different contexts. This framework will guide the development of novel diagnostic pathways for different diseases, accelerating the implementation of precision medicine into clinical practice.

Mass cytometry analysis reveals altered immune profiles in patients with coronary artery disease

Objective

The importance of inflammation in atherosclerosis is well accepted, but the role of the adaptive immune system is not yet fully understood. To further explore this, we assessed the circulating immune cell profile of patients with coronary artery disease (CAD) to identify discriminatory features by mass cytometry.

Methods

Mass cytometry was performed on patient samples from the BioHEART-CT study, gated to detect 82 distinct cell subsets. CT coronary angiograms were analysed to categorise patients as having CAD (CAD+) or having normal coronary arteries (CAD-).

Results

The discovery cohort included 117 patients (mean age 61 ± 12 years, 49% female); 79 patients (68%) were CAD+. Mass cytometry identified changes in 15 T-cell subsets, with higher numbers of proliferating, highly differentiated and cytotoxic cells and decreases in naïve T cells. Five T-regulatory subsets were related to an age and gender-independent increase in the odds of CAD incidence when expressing CCR2 (OR 1.12), CCR4 (OR 1.08), CD38 and CD45RO (OR 1.13), HLA-DR (OR 1.06) and Ki67 (OR 1.22). Markers of proliferation and differentiation were also increased within B cells, while plasmacytoid dendritic cells were decreased. This combination of changes was assessed using SVM models in discovery and validation cohorts (area under the curve = 0.74 for both), confirming the robust nature of the immune signature detected.

Conclusion

We identified differences within immune subpopulations of CAD+ patients which are indicative of a systemic immune response to coronary atherosclerosis. This immune signature needs further study via incorporation into risk scoring tools for the precision diagnosis of CAD.

Lipidomics Profiling and Risk of Coronary Artery Disease in the BioHEART-CT Discovery Cohort

The current coronary artery disease (CAD) risk scores for predicting future cardiovascular events rely on well-recognized traditional cardiovascular risk factors derived from a population level but often fail individuals, with up to 25% of first-time heart attack patients having no risk factors. Non-invasive imaging technology can directly measure coronary artery plaque burden. With an advanced lipidomic measurement methodology, for the first time, we aim to identify lipidomic biomarkers to enable intervention before cardiovascular events. With 994 participants from BioHEART-CT Discovery Cohort, we collected clinical data and performed high-performance liquid chromatography with mass spectrometry to determine concentrations of 683 plasma lipid species. Statin-naive participants were selected based on subclinical CAD (sCAD) categories as the analytical cohort (n = 580), with sCAD+ (n = 243) compared to sCAD- (n = 337). Through a machine learning approach, we built a lipid risk score (LRS) and compared the performance of the existing Framingham Risk Score (FRS) in predicting sCAD+. We obtained individual classifiability scores and determined Body Mass Index (BMI) as the modifying variable. FRS and LRS models achieved similar areas under the receiver operating characteristic curve (AUC) in predicting the validation cohort. LRS enhanced the prediction of sCAD+ in the healthy-weight group (BMI < 25 kg/m²), where FRS performed poorly and identified individuals at risk that FRS missed. Lipid features have strong potential as biomarkers to predict CAD plaque burden and can identify residual risk not captured by traditional risk factors/scores. LRS compliments FRS in prediction and has the most significant benefit in healthy-weight individuals.

Overcoming cohort heterogeneity for the prediction of subclinical cardiovascular disease risk

Cardiovascular disease remains a leading cause of mortality with an estimated half a billion people affected in 2019. However, detecting signals between specific pathophysiology and coronary plaque phenotypes using complex multi-omic discovery datasets remains challenging due to the diversity of individuals and their risk factors. Given the complex cohort heterogeneity present in those with coronary artery disease (CAD), we illustrate several different methods, both knowledge-guided and data-driven approaches, for identifying subcohorts of individuals with subclinical CAD and distinct metabolomic signatures. We then demonstrate that utilizing these subcohorts can improve the prediction of subclinical CAD and can facilitate the discovery of novel biomarkers of subclinical disease. Analyses acknowledging cohort heterogeneity through identifying and utilizing these subcohorts may be able to advance our understanding of CVD and provide more effective preventative treatments to reduce the burden of this disease in individuals and in society as a whole.

NEMoE: a nutrition aware regularized mixture of experts model to identify heterogeneous diet-microbiome-host health interactions

BACKGROUND

Unrevealing the interplay between diet, the microbiome, and the health state could enable the design of personalized intervention strategies and improve the health and well-being of individuals. A common approach to this is to divide the study population into smaller cohorts based on dietary preferences in the hope of identifying specific microbial signatures. However, classification of patients based solely on diet is unlikely to reflect the microbiome-host health relationship or the taxonomic microbiome makeup.

RESULTS

We present a novel approach, the Nutrition-Ecotype Mixture of Experts (NEMoE) model, for establishing associations between gut microbiota and health state that accounts for diet-specific cohort variability using a regularized mixture of experts model framework with an integrated parameter sharing strategy to ensure data-driven diet-cohort identification consistency across taxonomic levels. The success of our approach was demonstrated through a series of simulation studies, in which NEMoE showed robustness with regard to parameter selection and varying degrees of data heterogeneity. Further application to real-world microbiome data from a Parkinson's disease cohort revealed that NEMoE is capable of not only improving predictive performance for Parkinson's Disease but also for identifying diet-specific microbial signatures of disease.

CONCLUSION

In summary, NEMoE can be used to uncover diet-specific relationships between nutritional-ecotype and patient health and to contextualize precision nutrition for different diseases. Video Abstract.

Microenvironment‑driven resistance to B‑Raf inhibition in a melanoma patient is accompanied by broad changes of gene methylation and expression in distal fibroblasts

The incidence of malignant melanoma is rapidly increasing and current medicine is offering only limited options for treatment of the advanced disease. For B‑Raf mutated melanomas, treatment with mutation‑specific drug inhibitors may be used. Unfortunately, tumors frequently acquire resistance to the treatment. Tumor microenvironment, namely cancer‑associated fibroblasts, largely influence this acquired resistance. In the present study, fibroblasts were isolated from a patient suffering from acrolentiginous melanoma (Breslow, 4.0 mm; Clark, IV; B‑Raf V600E mutated). The present study focused on the expression of structural and functional markers of fibroblast activation in melanoma‑associated fibroblasts (MAFs; isolated prior to therapy initiation) as well as in autologous control fibroblasts (ACFs) of the same patient isolated during B‑Raf inhibitor therapy, yet before clinical progression of the disease. Analysis of gene transcription was also performed, as well as DNA methylation status analysis at the genomic scale of both isolates. MAFs were positive for smooth muscle actin (SMA), which is a marker of myofibroblasts and the hallmark of cancer stoma. Surprisingly, ACF isolated from the distant uninvolved skin of the same patient also exhibited strong SMA expression. A similar phenotype was also observed in control dermal fibroblasts (CDFs; from different donors) exclusively following stimulation by transforming growth factor (TGF)‑β1. Immunohistochemistry confirmed that melanoma cells potently produce TGF‑β1. Significant differences were also identified in gene transcription and in DNA methylation status at the genomic scale. Upregulation of SMA was observed in ACF cells at the protein and transcriptional levels. The present results support recent experimental findings that tumor microenvironment is driving resistance to B‑Raf inhibition in patients with melanoma. Such an activated microenvironment may be viable for the growth of circulating melanoma cells.

Presence of immune cells, low tumor proliferation and wild type BRAF mutation status is associated with a favourable clinical outcome in stage III cutaneous melanoma

BACKGROUND

The variable prognosis in stage III cutaneous melanoma is partially due to unknown prognostic factors. Improved prognostic tools are required to define patients with an increased risk of developing metastatic disease who might benefit from adjuvant therapies. The aim was to examine if cellular immune markers in association with tumor proliferation rate and BRAF mutation status have an impact on prognosis in stage III melanoma.

METHODS

We have used two sets of case series with stage III disease: 23 patients with short survival (≤ 13 months) and 19 patients with long survival (≥ 60 months). Lymph node metastases were analyzed for Ki67, CD8 and FOXP3 protein expression using immunohistochemistry. BRAF mutation status was analyzed in a previous study on the same samples.

RESULTS

Low tumor proliferation rate was significantly associated with a better prognosis (p = 0.013). Presence of FOXP3+ T cells was not correlated to adverse clinical outcome. A highly significant trend for a longer survival was found in the presence of an increasing number of markers; CD8+ and FOXP3+ T cells, low tumor proliferation and BRAF wildtype status (p = 0.003). Presence of at least three of these four markers was found to be an independent favorable prognostic factor (OR 19.4, 95% CI 1.9-197, p = 0.012), when adjusting for ulceration and number of lymph node metastases. Proliferation alone remained significant in multivariate analyses (OR 26.1, 95% CI 2.0-344, p = 0.013) but with a wider confidence interval. This panel still remained independent when also adjusting for a previously identified prognostic glycolytic-pigment panel.

CONCLUSIONS

We have demonstrated that presence of immune cells in association with tumor proliferation and BRAF mutation status may further contribute to identify stage III melanoma patients with high risk of relapse.