Xihui Lin

PD-L1 and Tumor Infiltrating Lymphocytes as Prognostic Markers in Resected NSCLC.

Introduction Immune checkpoint inhibition has shifted treatment paradigms in non-small cell lung cancer (NSCLC). Conflicting results have been reported regarding the immune infiltrate and programmed death-ligand 1 (PD-L1) as a prognostic marker. We correlated the immune infiltrate and PD-L1 expression with clinicopathologic characteristics in a cohort of resected NSCLC. Methods A tissue microarray was constructed using triplicate cores from consecutive resected NSCLC. Immunohistochemistry was performed for CD8, FOXP3 and PD-L1. Strong PD-L1 expression was predefined as greater than 50% tumor cell positivity. Matched nodal samples were assessed for concordance of PD-L1 expression. Results Of 522 patients, 346 were node-negative (N0), 72 N1 and 109 N2; 265 were adenocarcinomas (AC), 182 squamous cell cancers (SCC) and 75 other. Strong PD-L1 expression was found in 24% cases. In the overall cohort, PD-L1 expression was not associated with survival. In patients with N2 disease, strong PD-L1 expression was associated with significantly improved disease-free (DFS) and overall survival (OS) in multivariate analysis (HR 0.49, 95%CI 0.36–0.94, p = 0.031; HR 0.46, 95%CI 0.26–0.80, p = 0.006). In this resected cohort only 5% harboured EGFR mutations, whereas 19% harboured KRAS and 23% other. KRAS mutated tumors were more likely to highly express PD-L1 compared to EGFR (22% vs 3%). A stromal CD8 infiltrate was associated with significantly improved DFS in SCC (HR 0.70, 95%CI 0.50–0.97, p = 0.034), but not AC, whereas FOXP3 was not prognostic. Matched nodal specimens (N = 53) were highly concordant for PD-L1 expression (89%). Conclusion PD-L1 expression was not prognostic in the overall cohort. PD-L1 expression in primary tumor and matched nodal specimens were highly concordant. The observed survival benefit in N2 disease requires confirmation.

The Immune Microenvironment, Genome-wide Copy Number Aberrations, and Survival in Mesothelioma

Introduction: Results of recent clinical studies of immune checkpoint inhibitors in malignant pleural mesothelioma (MPM) have dampened initial enthusiasm. However, the immune environment and targets of these treatments such as programmed cell death protein 1 and its ligand programmed death ligand 1 (PD‐L1) have not been well characterized in MPM. Using a large cohort of patients, we investigated PD‐L1 expression, immune infiltrates, and genome‐wide copy number status and correlated them to clinicopathological features. Methods: Tissue microarrays were constructed and stained with PD‐L1(clone E1L3N [Cell Signaling Technology, Danvers, MA]), cluster of differentiation 4, cluster of differentiation 8, and forkhead box P3 antibodies. PD‐L1 positivity was defined as at least 5% membranous staining regardless of intensity, and high PD‐L1 positivity was defined as at least 50%. Genomic DNA from a representative subset of 113 patients was used for genome‐wide copy number analysis. The percent genome alteration was computed as a proxy for genomic instability, and statistical analyses were used to relate copy number aberrations to other variables. Results: Among 329 patients evaluated, PD‐L1 positivity was detected in 130 of 311 (41.7%), but high PD‐L1 positivity was seen in only 30 of 311 (9.6%). PD‐L1 positivity correlated with nonepithelioid histological subtype and increased infiltration with cluster of differentiation 4–positive, cluster of differentiation 8–positive, and forkhead box P3–positive lymphocytes. High PD‐L1–positive expression correlated with worse prognosis (hazard ratio = 2.37, 95% confidence interval: 1.57–3.56, p < 0.001) in univariate analysis but not in multivariate analysis. Higher percent genome alteration was associated with epithelioid histological subtype and poorer survival (hazard ratio = 1.59, 95% confidence interval: 1.01–2.5, p = 0.04) but not PD‐L1 expression. Conclusions: PD‐L1 expression was associated with nonepithelioid MPM, poor clinical outcome, and increased immunological infiltrates. Increased genomic instability did not correlate with PD‐L1 expression but was associated with poorer survival.

A Crowdsourcing Approach to Developing and Assessing Prediction Algorithms for AML Prognosis

Acute Myeloid Leukemia (AML) is a fatal hematological cancer. The genetic abnormalities underlying AML are extremely heterogeneous among patients, making prognosis and treatment selection very difficult. While clinical proteomics data has the potential to improve prognosis accuracy, thus far, the quantitative means to do so have yet to be developed. Here we report the results and insights gained from the DREAM 9 Acute Myeloid Prediction Outcome Prediction Challenge (AML-OPC), a crowdsourcing effort designed to promote the development of quantitative methods for AML prognosis prediction. We identify the most accurate and robust models in predicting patient response to therapy, remission duration, and overall survival. We further investigate patient response to therapy, a clinically actionable prediction, and find that patients that are classified as resistant to therapy are harder to predict than responsive patients across the 31 models submitted to the challenge. The top two performing models, which held a high sensitivity to these patients, substantially utilized the proteomics data to make predictions. Using these models, we also identify which signaling proteins were useful in predicting patient therapeutic response.

BPG: Seamless, Automated and Interactive Visualization of Scientific Data

We introduce BPG, an easy-to-use framework for generating publication-quality, highly-customizable plots in the R statistical environment. This open-source package includes novel methods of displaying high-dimensional datasets and facilitates generation of complex multi-panel figures, making it ideal for complex datasets. A web-based interactive tool allows online figure customization, from which R code can be downloaded for seamless integration with computational pipelines. BPG is available at http://labs.oicr.on.ca/boutros-lab/software/bpg

Fast Nonnegative Matrix Factorization and Applications to Pattern Extraction, Deconvolution and Imputation

Nonnegative matrix factorization (NMF) is a technique widely used in various fields, including artificial intelligence (AI), signal processing and bioinformatics. However existing algorithms and R packages cannot be applied to large matrices due to their slow convergence, and cannot handle missing values. In addition, most NMF research focuses only on blind decompositions: decomposition without utilizing prior knowledge. We adapt the idea of sequential coordinate-wise descent to NMF to increase the convergence rate. Our NMF algorithm thus handles missing values naturally and integrates prior knowledge to guide NMF towards a more meaningful decomposition. To support its use, we describe a novel imputation-based method to determine the rank of decomposition. All our algorithms are implemented in the R package NNLM, which is freely available on CRAN.

Network-Based Biomarkers Enable Cross-Disease Biomarker Discovery

Biomarkers lie at the heart of precision medicine, biodiversity monitoring, agricultural pathogen detection, amongst others. Surprisingly, while rapid genomic profiling is becoming ubiquitous, the development of biomarkers almost always involves the application of bespoke techniques that cannot be directly applied to other datasets. There is an urgent need for a systematic methodology to create biologically-interpretable molecular models that robustly predict key phenotypes. We therefore created SIMMS: an algorithm that fragments pathways into functional modules and uses these to predict phenotypes. We applied SIMMS to multiple data-types across four diseases, and in each it reproducibly identified subtypes, made superior predictions to the best bespoke approaches, and identified known and novel signaling nodes. To demonstrate its ability on a new dataset, we measured 33 genes/nodes of the PIK3CA pathway in 1,734 FFPE breast tumours and created a four-subnetwork prediction model. This model significantly out-performed existing clinically-used molecular tests in an independent 1,742-patient validation cohort. SIMMS is generic and can work with any molecular data or biological network, and is freely available at: https://cran.r-project.org/web/packages/SIMMS.

Abstract 3124: Prognostic biomarkers for pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) has the lowest 5-year-survival rate of common cancers ( Among the 67 published risk loci we tested using a multivariate Cox proportional hazard model, we found a strong positive association of the single nucleotide polymorphism rs4785367 (RefSNP alleles: C/T on forward strand; MAF = 0.474) with overall survival of PDAC donors: HR = 0.426; CI = 0.268 - 0.686; p-value = 0.00029. A more detailed analysis at the genotype level revealed that the presence of the homozygous minor allele has a stronger effect than either the heterozygous or homozygous major allele. The SNP falls within the intergenic region between the ZNF423 and TMEM188 genes, within the exon 2 of lncRNA RP11-305A4.3 and overlaps a CTCF regulatory domain. Preliminary gene expression analysis from RNA sequencing data on a subset of PDAC donors (n = 28) shows that patients carrying the minor allele have significant higher TMEM188 expression than of the major type (p-value = 0.012), suggesting that this allele may influence the course of PDAC via TMEM188 activity. A recent study linked the gene product to activation of NK cells, which in turns increases the defense mechanism against the pathogens, infections and transformed tumors. These findings suggest a possible molecular mechanism influencing the course of PDAC. We are also exploring the effect of the presence of the minor allele on the regulatory CTCF region, by applying an integrative pipeline for risk SNP analysis to pancreatic cancer. This will possibly detect the effect on the NANOG motif binding and/or on CTCF looping. In summary, the present study detected the rs4785367 as a prognostic biomarker for pancreatic cancer, with the novelty of increased TMEM188 gene expression being linked to the presence of the alternate allele in PDAC patients. Further investigations on this and on assessing the effect of the polymorphism on the regulatory CTCF feature are in progress. Citation Format: Cristina Baciu, Robert Grant, Hansen He, Musaddeque Ahmed, Robert E. Denroche, Lee Timms, Gun Ho Jang, Ayelet Borgida, Xihui Lin, Paul C. Boutros, Dianne Chadwich, Sheng-Ben Liang, Sagedeh Shahabi, Michael H.A. Roehrl, Sean Cleary, Julie M. Wilson, John D. McPherson, Lincoln Stein, Steven Gallinger. Prognostic biomarkers for pancreatic cancer. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3124.