The HDAC9-associated risk locus promotes coronary artery disease by governing TWIST1.

Genome wide association studies (GWAS) have identified thousands of single nucleotide polymorphisms (SNPs) associated with the risk of common disorders. However, since the large majority of these risk SNPs reside outside gene-coding regions, GWAS generally provide no information about causal mechanisms regarding the specific gene(s) that are affected or the tissue(s) in which these candidate gene(s) exert their effect. The 'gold standard' method for understanding causal genes and their mechanisms of action are laborious basic science studies often involving sophisticated knockin or knockout mouse lines, however, these types of studies are impractical as a high-throughput means to understand the many risk variants that cause complex diseases like coronary artery disease (CAD). As a solution, we developed a streamlined, data-driven informatics pipeline to gain mechanistic insights on complex genetic loci. The pipeline begins by understanding the SNPs in a given locus in terms of their relative location and linkage disequilibrium relationships, and then identifies nearby expression quantitative trait loci (eQTLs) to determine their relative independence and the likely tissues that mediate their disease-causal effects. The pipeline then seeks to understand associations with other disease-relevant genes, disease sub-phenotypes, potential causality (Mendelian randomization), and the regulatory and functional involvement of these genes in gene regulatory co-expression networks (GRNs). Here, we applied this pipeline to understand a cluster of SNPs associated with CAD within and immediately adjacent to the gene encoding HDAC9. Our pipeline demonstrated, and validated, that this locus is causal for CAD by modulation of TWIST1 expression levels in the arterial wall, and by also governing a GRN related to metabolic function in skeletal muscle. Our results reconciled numerous prior studies, and also provided clear evidence that this locus does not govern HDAC9 expression, structure or function. This pipeline should be considered as a powerful and efficient way to understand GWAS risk loci in a manner that better reflects the highly complex nature of genetic risk associated with common disorders.

Biopsy and blood-based molecular biomarker of inflammation in IBD.

OBJECTIVE: IBD therapies and treatments are evolving to deeper levels of remission. Molecular measures of disease may augment current endpoints including the potential for less invasive assessments. DESIGN: Transcriptome analysis on 712 endoscopically defined inflamed (Inf) and 1778 non-inflamed (Non-Inf) intestinal biopsies (n=498 Crohn's disease, n=421 UC and 243 controls) in the Mount Sinai Crohn's and Colitis Registry were used to identify genes differentially expressed between Inf and Non-Inf biopsies and to generate a molecular inflammation score (bMIS) via gene set variance analysis. A circulating MIS (cirMIS) score, reflecting intestinal molecular inflammation, was generated using blood transcriptome data. bMIS/cirMIS was validated as indicators of intestinal inflammation in four independent IBD cohorts. RESULTS: bMIS/cirMIS was strongly associated with clinical, endoscopic and histological disease activity indices. Patients with the same histologic score of inflammation had variable bMIS scores, indicating that bMIS describes a deeper range of inflammation. In available clinical trial data sets, both scores were responsive to IBD treatment. Despite similar baseline endoscopic and histologic activity, UC patients with lower baseline bMIS levels were more likely treatment responders compared with those with higher levels. Finally, among patients with UC in endoscopic and histologic remission, those with lower bMIS levels were less likely to have a disease flare over time. CONCLUSION: Transcriptionally based scores provide an alternative objective and deeper quantification of intestinal inflammation, which could augment current clinical assessments used for disease monitoring and have potential for predicting therapeutic response and patients at higher risk of disease flares.

Polygenic risk score for alcohol drinking behavior improves prediction of inflammatory bowel disease risk.

Epidemiological studies have long recognized risky behaviors as potentially modifiable factors for the onset and flares of inflammatory bowel disease (IBD); yet, the underlying mechanisms are largely unknown. Recently, the genetic susceptibilities to cigarette smoking, alcohol and cannabis use [i.e. substance use (SU)] have been characterized by well-powered genome-wide association studies (GWASs). We aimed to assess the impact of genetic determinants of SU on IBD risk. Using Mount Sinai Crohn's and Colitis Registry (MSCCR) cohort of 1058 IBD cases and 188 healthy controls, we computed the polygenic risk score (PRS) for SU and correlated them with the observed IBD diagnoses, while adjusting for genetic ancestry, PRS for IBD and SU behavior at enrollment. The results were validated in a pediatric cohort with no SU exposure. PRS of alcohol consumption (DrnkWk), smoking cessation and age of smoking initiation, were associated with IBD risk in MSCCR even after adjustment for PRSIBD and actual smoking status. One interquartile range decrease in PRSDrnkWk was significantly associated to higher IBD risk (i.e. inverse association) (with odds ratio = 1.65 and 95% confidence interval: 1.32, 2.06). The association was replicated in a pediatric Crohn's disease cohort. Colocalization analysis identified a locus on chromosome 16 with polymorphisms in IL27, SULT1A2 and SH2B1, which reached genome-wide statistical significance in GWAS (P < 7.7e-9) for both alcohol consumption and IBD risk. This study demonstrated that the genetic predisposition to SU was associated with IBD risk, independent of PRSIBD and in the absence of SU behaviors. Our study may help further stratify individuals at risk of IBD.

Integrative Analysis of the Inflammatory Bowel Disease Serum Metabolome Improves Our Understanding of Genetic Etiology and Points to Novel Putative Therapeutic Targets.

BACKGROUND & AIMS: Polygenic and environmental factors are underlying causes of inflammatory bowel disease (IBD). We hypothesized that integration of the genetic loci controlling a metabolite's abundance, with known IBD genetic susceptibility loci, may help resolve metabolic drivers of IBD. METHODS: We measured the levels of 1300 metabolites in the serum of 484 patients with ulcerative colitis (UC) and 464 patients with Crohn's disease (CD) and 365 controls. Differential metabolite abundance was determined for disease status, subtype, clinical and endoscopic disease activity, as well as IBD phenotype including disease behavior, location, and extent. To inform on the genetic basis underlying metabolic diversity, we integrated metabolite and genomic data. Genetic colocalization and Mendelian randomization analyses were performed using known IBD risk loci to explore whether any metabolite was causally associated with IBD. RESULTS: We found 173 genetically controlled metabolites (metabolite quantitative trait loci, 9 novel) within 63 non-overlapping loci (7 novel). Furthermore, several metabolites significantly associated with IBD disease status and activity as defined using clinical and endoscopic indexes. This constitutes a resource for biomarker discovery and IBD biology insights. Using this resource, we show that a novel metabolite quantitative trait locus for serum butyrate levels containing ACADS was not supported as causal for IBD; replicate the association of serum omega-6 containing lipids with the fatty acid desaturase 1/2 locus and identify these metabolites as causal for CD through Mendelian randomization; and validate a novel association of serum plasmalogen and TMEM229B, which was predicted as causal for CD. CONCLUSIONS: An exploratory analysis combining genetics and unbiased serum metabolome surveys can reveal novel biomarkers of disease activity and potential mediators of pathology in IBD.

EnsembleCNV: an ensemble machine learning algorithm to identify and genotype copy number variation using SNP array data.

The associations between diseases/traits and copy number variants (CNVs) have not been systematically investigated in genome-wide association studies (GWASs), primarily due to a lack of robust and accurate tools for CNV genotyping. Herein, we propose a novel ensemble learning framework, ensembleCNV, to detect and genotype CNVs using single nucleotide polymorphism (SNP) array data. EnsembleCNV (a) identifies and eliminates batch effects at raw data level; (b) assembles individual CNV calls into CNV regions (CNVRs) from multiple existing callers with complementary strengths by a heuristic algorithm; (c) re-genotypes each CNVR with local likelihood model adjusted by global information across multiple CNVRs; (d) refines CNVR boundaries by local correlation structure in copy number intensities; (e) provides direct CNV genotyping accompanied with confidence score, directly accessible for downstream quality control and association analysis. Benchmarked on two large datasets, ensembleCNV outperformed competing methods and achieved a high call rate (93.3%) and reproducibility (98.6%), while concurrently achieving high sensitivity by capturing 85% of common CNVs documented in the 1000 Genomes Project. Given this CNV call rate and accuracy, which are comparable to SNP genotyping, we suggest ensembleCNV holds significant promise for performing genome-wide CNV association studies and investigating how CNVs predispose to human diseases.

Genetic regulation of the placental transcriptome underlies birth weight and risk of childhood obesity.

GWAS identified variants associated with birth weight (BW), childhood obesity (CO) and childhood BMI (CBMI), and placenta is a critical organ for fetal development and postnatal health. We examined the role of placental transcriptome and eQTLs in mediating the genetic causes for BW, CO and CBMI, and applied integrative analysis (Colocalization and MetaXcan). GWAS loci associated with BW, CO, and CBMI were substantially enriched for placenta eQTLs (6.76, 4.83 and 2.26 folds, respectively). Importantly, compared to eQTLs of adult tissues, only placental eQTLs contribute significantly to both anthropometry outcomes at birth (BW) and childhood phenotypes (CO/CBMI). Eight, six and one transcripts colocalized with BW, CO and CBMI risk loci, respectively. Our study reveals that placental transcription in utero likely plays a key role in determining postnatal body size, and as such may hold new possibilities for therapeutic interventions to prevent childhood obesity.

High-Throughput Characterization of Blood Serum Proteomics of IBD Patients with Respect to Aging and Genetic Factors.

To date, no large scale, systematic description of the blood serum proteome has been performed in inflammatory bowel disease (IBD) patients. By using microarray technology, a more complete description of the blood proteome of IBD patients is feasible. It may help to achieve a better understanding of the disease. We analyzed blood serum profiles of 1128 proteins in IBD patients of European descent (84 Crohn's Disease (CD) subjects and 88 Ulcerative Colitis (UC) subjects) as well as 15 healthy control subjects, and linked protein variability to patient age (all cohorts) and genetic components (genotype data generated from CD patients). We discovered new, previously unreported aging-associated proteomic traits (such as serum Albumin level), confirmed previously reported results from different tissues (i.e., upregulation of APOE with aging), and found loss of regulation of MMP7 in CD patients. In carrying out a genome wide genotype-protein association study (proteomic Quantitative Trait Loci, pQTL) within the CD patients, we identified 41 distinct proteomic traits influenced by cis pQTLs (underlying SNPs are referred to as pSNPs). Significant overlaps between pQTLs and cis eQTLs corresponding to the same gene were observed and in some cases the QTL were related to inflammatory disease susceptibility. Importantly, we discovered that serum protein levels of MST1 (Macrophage Stimulating 1) were regulated by SNP rs3197999 (p = 5.96E-10, FDR<5%), an accepted GWAS locus for IBD. Filling the knowledge gap of molecular mechanisms between GWAS hits and disease susceptibility requires systematically dissecting the impact of the locus at the cell, mRNA expression, and protein levels. The technology and analysis tools that are now available for large-scale molecular studies can elucidate how alterations in the proteome driven by genetic polymorphisms cause or provide protection against disease. Herein, we demonstrated this directly by integrating proteomic and pQTLs with existing GWAS, mRNA expression, and eQTL datasets to provide insights into the biological processes underlying IBD and pinpoint causal genetic variants along with their downstream molecular consequences.

Expression quantitative trait loci (eQTLs) in human placentas suggest developmental origins of complex diseases.

Epidemiologic studies support that at least part of the risk of chronic diseases in childhood and even adulthood may have an in utero origin, and the placenta is a key organ that plays a pivotal role in fetal growth and development. The transcriptomes of 159 human placenta tissues were profiled by genome-wide RNA sequencing (Illumina High-Seq 2500), and linked to fetal genotypes assessed by a high density single nucleotide polymorphism (SNP) genotyping array (Illumina MegaEx). Expression quantitative trait loci (eQTLs) across all annotated transcripts were mapped and examined for enrichment for disease susceptibility loci annotated in the genome-wide association studies (GWAS) catalog. We discovered 3218 cis- and 35 trans-eQTLs at ≤10% false discovery rate in human placentas. Among the 16 439 known disease loci of genome-wide significance, 835 were placental eSNPs (enrichment fold = 1.68, P = 7.41e-42). Stronger effect sizes were observed between GWAS SNPs and gene expression in placentas than what has been reported in other tissues, such as the correlation between asthma risk allele, rs7216389-T and Gasdermin-B (GSDMB) in placenta (r2=27%) versus lung (r2=6%). Finally, our results suggest the placental eQTLs may mediate the function of GWAS loci on postnatal disease susceptibility. Results suggest that transcripts in placenta are under tight genetic control, and that placental gene networks may influence postnatal risk of multiple human diseases lending support for the Developmental Origins of Health and Disease.

Genome-wide association study identifies four novel loci associated with Alzheimer's endophenotypes and disease modifiers.

More than 20 genetic loci have been associated with risk for Alzheimer's disease (AD), but reported genome-wide significant loci do not account for all the estimated heritability and provide little information about underlying biological mechanisms. Genetic studies using intermediate quantitative traits such as biomarkers, or endophenotypes, benefit from increased statistical power to identify variants that may not pass the stringent multiple test correction in case-control studies. Endophenotypes also contain additional information helpful for identifying variants and genes associated with other aspects of disease, such as rate of progression or onset, and provide context to interpret the results from genome-wide association studies (GWAS). We conducted GWAS of amyloid beta (Aß42), tau, and phosphorylated tau (ptau181) levels in cerebrospinal fluid (CSF) from 3146 participants across nine studies to identify novel variants associated with AD. Five genome-wide significant loci (two novel) were associated with ptau181, including loci that have also been associated with AD risk or brain-related phenotypes. Two novel loci associated with Aß42 near GLIS1 on 1p32.3 (ß = -0.059, P = 2.08 × 10-8) and within SERPINB1 on 6p25 (ß = -0.025, P = 1.72 × 10-8) were also associated with AD risk (GLIS1: OR = 1.105, P = 3.43 × 10-2), disease progression (GLIS1: ß = 0.277, P = 1.92 × 10-2), and age at onset (SERPINB1: ß = 0.043, P = 4.62 × 10-3). Bioinformatics indicate that the intronic SERPINB1 variant (rs316341) affects expression of SERPINB1 in various tissues, including the hippocampus, suggesting that SERPINB1 influences AD through an Aß-associated mechanism. Analyses of known AD risk loci suggest CLU and FERMT2 may influence CSF Aß42 (P = 0.001 and P = 0.009, respectively) and the INPP5D locus may affect ptau181 levels (P = 0.009); larger studies are necessary to verify these results. Together the findings from this study can be used to inform future AD studies.

Rare loss-of-function mutations of PTGIR are enriched in fibromuscular dysplasia.

AIMS: Fibromuscular dysplasia (FMD) and spontaneous coronary artery dissection (SCAD) are related, non-atherosclerotic arterial diseases mainly affecting middle-aged women. Little is known about their physiopathological mechanisms. We aimed to identify rare genetic causes to elucidate molecular mechanisms implicated in FMD and SCAD. METHODS AND RESULTS: We analysed 29 exomes that included familial and sporadic FMD. We identified one rare loss-of-function variant (LoF) (frequencygnomAD = 0.000075) shared by two FMD sisters in the prostaglandin I2 receptor gene (PTGIR), a key player in vascular remodelling. Follow-up was conducted by targeted or Sanger sequencing (1071 FMD and 363 SCAD patients) or lookups in exome (264 FMD) or genome sequences (480 SCAD), all independent and unrelated. It revealed four additional LoF allele carriers, in addition to several rare missense variants, among FMD patients, and two LoF allele carriers among SCAD patients, including one carrying a rare splicing mutation (c.768 + 1C>G). We used burden test to test for enrichment in patients compared to gnomAD controls, which detected a putative enrichment in FMD (PTRAPD = 8 × 10-4), but not a significant enrichment (PTRAPD = 0.12) in SCAD. The biological effects of variants on human prostaclycin receptor (hIP) signalling and protein expression were characterized using transient overexpression in human cells. We confirmed the LoFs (Q163X and P17RfsX6) and one missense (L67P), identified in one FMD and one SCAD patient, to severely impair hIP function in vitro. CONCLUSIONS: Our study shows that rare genetic mutations in PTGIR are enriched among FMD patients and found in SCAD patients, suggesting a role for prostacyclin signalling in non-atherosclerotic stenosis and dissection.

A plasma proteogenomic signature for fibromuscular dysplasia.

AIMS: Fibromuscular dysplasia (FMD) is a poorly understood disease that predominantly affects women during middle-life, with features that include stenosis, aneurysm, and dissection of medium-large arteries. Recently, plasma proteomics has emerged as an important means to understand cardiovascular diseases. Our objectives were: (i) to characterize plasma proteins and determine if any exhibit differential abundance in FMD subjects vs. matched healthy controls and (ii) to leverage these protein data to conduct systems analyses to provide biologic insights on FMD, and explore if this could be developed into a blood-based FMD test. METHODS AND RESULTS: Females with 'multifocal' FMD and matched healthy controls underwent clinical phenotyping, dermal biopsy, and blood draw. Using dual-capture proximity extension assay and nuclear magnetic resonance-spectroscopy, we evaluated plasma levels of 981 proteins and 31 lipid sub-classes, respectively. In a discovery cohort (Ncases = 90, Ncontrols = 100), we identified 105 proteins and 16 lipid sub-classes (predominantly triglycerides and fatty acids) with differential plasma abundance in FMD cases vs. controls. In an independent cohort (Ncases = 23, Ncontrols = 28), we successfully validated 37 plasma proteins and 10 lipid sub-classes with differential abundance. Among these, 5/37 proteins exhibited genetic control and Bayesian analyses identified 3 of these as potential upstream drivers of FMD. In a 3rd cohort (Ncases = 506, Ncontrols = 876) the genetic locus of one of these upstream disease drivers, CD2-associated protein (CD2AP), was independently validated as being associated with risk of having FMD (odds ratios = 1.36; P = 0.0003). Immune-fluorescence staining identified that CD2AP is expressed by the endothelium of medium-large arteries. Finally, machine learning trained on the discovery cohort was used to develop a test for FMD. When independently applied to the validation cohort, the test showed a c-statistic of 0.73 and sensitivity of 78.3%. CONCLUSION: FMD exhibits a plasma proteogenomic and lipid signature that includes potential causative disease drivers, and which holds promise for developing a blood-based test for this disease.

High-Throughput Identification of the Plasma Proteomic Signature of Inflammatory Bowel Disease.

BACKGROUND: The molecular aetiology of inflammatory bowel disease [IBD] and its two subtypes, ulcerative colitis [UC] and Crohn's disease [CD], have been carefully investigated at genome and transcriptome levels. Recent advances in high-throughput proteome quantification has enabled comprehensive large-scale plasma proteomics studies of IBD. METHODS: The study used two cohorts: [1] The CERTIFI-cohort: 42 samples from the CERTIFI trial of anti-TNFα-refractory CD patients; [2] the PROgECT-UNITI-HCs cohort: 46 UC samples of the PROgECT study, 84 CD samples of the UNITI I and UNITI II studies, and 72 healthy controls recruited in Mount Sinai Hospital, New York, USA. The plasma proteome for these two cohorts was quantified using high-throughput platforms. RESULTS: For the PROgECT-UNITI-HCs cohort, we measured a total of 1310 proteins. Of these, 493 proteins showed different plasma levels in IBD patients to the plasma levels in controls at 10% false discovery rate [FDR], among which 11 proteins had a fold change greater than 2. The proteins upregulated in IBD were associated with immunity functionality, whereas the proteins downregulated in IBD were associated with nutrition and metabolism. The proteomic profiles were very similar between UC and CD. In the CERTIFI cohort, 1014 proteins were measured, and it was found that the plasma protein level had little correlation with the blood or intestine transcriptomes. CONCLUSIONS: We report the largest proteomics study to date on IBD and controls. A large proportion of plasma proteins are altered in IBD, which provides insights into the disease aetiology and indicates a potential for biomarker discovery.

Pericytes impair capillary blood flow and motor function after chronic spinal cord injury.

Blood vessels in the central nervous system (CNS) are controlled by neuronal activity. For example, widespread vessel constriction (vessel tone) is induced by brainstem neurons that release the monoamines serotonin and noradrenaline, and local vessel dilation is induced by glutamatergic neuron activity. Here we examined how vessel tone adapts to the loss of neuron-derived monoamines after spinal cord injury (SCI) in rats. We find that, months after the imposition of SCI, the spinal cord below the site of injury is in a chronic state of hypoxia owing to paradoxical excess activity of monoamine receptors (5-HT1) on pericytes, despite the absence of monoamines. This monoamine-receptor activity causes pericytes to locally constrict capillaries, which reduces blood flow to ischemic levels. Receptor activation in the absence of monoamines results from the production of trace amines (such as tryptamine) by pericytes that ectopically express the enzyme aromatic L-amino acid decarboxylase (AADC), which synthesizes trace amines directly from dietary amino acids (such as tryptophan). Inhibition of monoamine receptors or of AADC, or even an increase in inhaled oxygen, produces substantial relief from hypoxia and improves motoneuron and locomotor function after SCI.

A common haplotype lowers PU.1 expression in myeloid cells and delays onset of Alzheimer's disease.

A genome-wide survival analysis of 14,406 Alzheimer's disease (AD) cases and 25,849 controls identified eight previously reported AD risk loci and 14 novel loci associated with age at onset. Linkage disequilibrium score regression of 220 cell types implicated the regulation of myeloid gene expression in AD risk. The minor allele of rs1057233 (G), within the previously reported CELF1 AD risk locus, showed association with delayed AD onset and lower expression of SPI1 in monocytes and macrophages. SPI1 encodes PU.1, a transcription factor critical for myeloid cell development and function. AD heritability was enriched within the PU.1 cistrome, implicating a myeloid PU.1 target gene network in AD. Finally, experimentally altered PU.1 levels affected the expression of mouse orthologs of many AD risk genes and the phagocytic activity of mouse microglial cells. Our results suggest that lower SPI1 expression reduces AD risk by regulating myeloid gene expression and cell function.

Blood and Intestine eQTLs from an Anti-TNF-Resistant Crohn's Disease Cohort Inform IBD Genetic Association Loci.

OBJECTIVES: Genome-wide association studies (GWAS) have identified loci reproducibly associated with inflammatory bowel disease (IBD) and other immune-mediated diseases; however, the molecular mechanisms underlying most of genetic susceptibility remain undefined. Expressional quantitative trait loci (eQTL) of disease-relevant tissue can be employed in order to elucidate the genes and pathways affected by disease-specific genetic variance. METHODS: In this study, we derived eQTLs for human whole blood and intestine tissues of anti-tumor necrosis factor-resistant Crohn's disease (CD) patients. We interpreted these eQTLs in the context of published IBD GWAS hits to inform on the disease process. RESULTS: At 10% false discovery rate, we discovered that 5,174 genes in blood and 2,063 genes in the intestine were controlled by a nearby single-nucleotide polymorphism (SNP) (i.e., cis-eQTL), among which 1,360 were shared between the two tissues. A large fraction of the identified eQTLs were supported by the regulomeDB database, showing that the eQTLs reside in regulatory elements (odds ratio; OR=3.44 and 3.24 for blood and intestine eQTLs, respectively) as opposed to protein-coding regions. Published IBD GWAS hits as a whole were enriched for blood and intestine eQTLs (OR=2.88 and 2.05; and P value=2.51E-9 and 0.013, respectively), thereby linking genetic susceptibility to control of gene expression in these tissues. Through a systematic search, we used eQTL data to inform 109 out of 372 IBD GWAS SNPs documented in National Human Genome Research Institute catalog, and we categorized the genes influenced by eQTLs according to their functions. Many of these genes have experimentally validated roles in specific cell types contributing to intestinal inflammation. CONCLUSIONS: The blood and intestine eQTLs described in this study represent a powerful tool to link GWAS loci to a regulatory function and thus elucidate the mechanisms underlying the genetic loci associated with IBD and related conditions. Overall, our eQTL discovery approach empirically identifies the disease-associated variants including their impact on the direction and extent of expression changes in the context of disease-relevant cellular pathways in order to infer the functional outcome of this aspect of genetic susceptibility.

Test of four colon cancer risk-scores in formalin fixed paraffin embedded microarray gene expression data.

BACKGROUND: Prognosis prediction for resected primary colon cancer is based on the T-stage Node Metastasis (TNM) staging system. We investigated if four well-documented gene expression risk scores can improve patient stratification. METHODS: Microarray-based versions of risk-scores were applied to a large independent cohort of 688 stage II/III tumors from the PETACC-3 trial. Prognostic value for relapse-free survival (RFS), survival after relapse (SAR), and overall survival (OS) was assessed by regression analysis. To assess improvement over a reference, prognostic model was assessed with the area under curve (AUC) of receiver operating characteristic (ROC) curves. All statistical tests were two-sided, except the AUC increase. RESULTS: All four risk scores (RSs) showed a statistically significant association (single-test, P < .0167) with OS or RFS in univariate models, but with HRs below 1.38 per interquartile range. Three scores were predictors of shorter RFS, one of shorter SAR. Each RS could only marginally improve an RFS or OS model with the known factors T-stage, N-stage, and microsatellite instability (MSI) status (AUC gains < 0.025 units). The pairwise interscore discordance was never high (maximal Spearman correlation = 0.563) A combined score showed a trend to higher prognostic value and higher AUC increase for OS (HR = 1.74, 95% confidence interval [CI] = 1.44 to 2.10, P < .001, AUC from 0.6918 to 0.7321) and RFS (HR = 1.56, 95% CI = 1.33 to 1.84, P < .001, AUC from 0.6723 to 0.6945) than any single score. CONCLUSIONS: The four tested gene expression-based risk scores provide prognostic information but contribute only marginally to improving models based on established risk factors. A combination of the risk scores might provide more robust information. Predictors of RFS and SAR might need to be different.