Strategies for Implementing Machine Learning Algorithms in the Clinical Practice of Radiology.

Despite recent advancements in machine learning (ML) applications in health care, there have been few benefits and improvements to clinical medicine in the hospital setting. To facilitate clinical adaptation of methods in ML, this review proposes a standardized framework for the step-by-step implementation of artificial intelligence into the clinical practice of radiology that focuses on three key components: problem identification, stakeholder alignment, and pipeline integration. A review of the recent literature and empirical evidence in radiologic imaging applications justifies this approach and offers a discussion on structuring implementation efforts to help other hospital practices leverage ML to improve patient care. Clinical trial registration no. 04242667 © RSNA, 2024 Supplemental material is available for this article.

Clinical correlates of CT imaging-derived phenotypes among lean and overweight patients with hepatic steatosis.

The objective of this study is to define CT imaging derived phenotypes for patients with hepatic steatosis, a common metabolic liver condition, and determine its association with patient data from a medical biobank. There is a need to further characterize hepatic steatosis in lean patients, as its epidemiology may differ from that in overweight patients. A deep learning method determined the spleen-hepatic attenuation difference (SHAD) in Hounsfield Units (HU) on abdominal CT scans as a quantitative measure of hepatic steatosis. The patient cohort was stratified by BMI with a threshold of 25 kg/m2 and hepatic steatosis with threshold SHAD ≥ - 1 HU or liver mean attenuation ≤ 40 HU. Patient characteristics, diagnoses, and laboratory results representing metabolism and liver function were investigated. A phenome-wide association study (PheWAS) was performed for the statistical interaction between SHAD and the binary characteristic LEAN. The cohort contained 8914 patients-lean patients with (N = 278, 3.1%) and without (N = 1867, 20.9%) steatosis, and overweight patients with (N = 1863, 20.9%) and without (N = 4906, 55.0%) steatosis. Among all lean patients, those with steatosis had increased rates of cardiovascular disease (41.7 vs 27.8%), hypertension (86.7 vs 49.8%), and type 2 diabetes mellitus (29.1 vs 15.7%) (all p < 0.0001). Ten phenotypes were significant in the PheWAS, including chronic kidney disease, renal failure, and cardiovascular disease. Hepatic steatosis was found to be associated with cardiovascular, kidney, and metabolic conditions, separate from overweight BMI.

Automated Lung Ultrasound Pulmonary Disease Quantification Using an Unsupervised Machine Learning Technique for COVID-19.

COVID-19 is an ongoing global health pandemic. Although COVID-19 can be diagnosed with various tests such as PCR, these tests do not establish pulmonary disease burden. Whereas point-of-care lung ultrasound (POCUS) can directly assess the severity of characteristic pulmonary findings of COVID-19, the advantage of using US is that it is inexpensive, portable, and widely available for use in many clinical settings. For automated assessment of pulmonary findings, we have developed an unsupervised learning technique termed the calculated lung ultrasound (CLU) index. The CLU can quantify various types of lung findings, such as A or B lines, consolidations, and pleural effusions, and it uses these findings to calculate a CLU index score, which is a quantitative measure of pulmonary disease burden. This is accomplished using an unsupervised, patient-specific approach that does not require training on a large dataset. The CLU was tested on 52 lung ultrasound examinations from several institutions. CLU demonstrated excellent concordance with radiologist findings in different pulmonary disease states. Given the global nature of COVID-19, the CLU would be useful for sonographers and physicians in resource-strapped areas with limited ultrasound training and diagnostic capacities for more accurate assessment of pulmonary status.

Ultrasound Radiomics for the Detection of Early-Stage Liver Fibrosis.

Objective: The study evaluates quantitative ultrasound (QUS) texture features with machine learning (ML) to enhance the sensitivity of B-mode ultrasound (US) for the detection of fibrosis at an early stage and distinguish it from advanced fibrosis. Different ML methods were evaluated to determine the best diagnostic model. Methods: 233 B-mode images of liver lobes with early and advanced-stage fibrosis induced in a rat model were analyzed. Sixteen features describing liver texture were measured from regions of interest (ROIs) drawn on B-mode images. The texture features included a first-order statistics run length (RL) and gray-level co-occurrence matrix (GLCM). The features discriminating between early and advanced fibrosis were used to build diagnostic models with logistic regression (LR), naïve Bayes (nB), and multi-class perceptron (MLP). The diagnostic performances of the models were compared by ROC analysis using different train-test sampling approaches, including leave-one-out, 10-fold cross-validation, and varying percentage splits. METAVIR scoring was used for histological fibrosis staging of the liver. Results: 15 features showed a significant difference between the advanced and early liver fibrosis groups, p < 0.05. Among the individual features, first-order statics features led to the best classification with a sensitivity of 82.1−90.5% and a specificity of 87.1−89.8%. For the features combined, the diagnostic performances of nB and MLP were high, with the area under the ROC curve (AUC) approaching 0.95−0.96. LR also yielded high diagnostic performance (AUC = 0.91−0.92) but was lower than nB and MLP. The diagnostic variability between test-train trials, measured by the coefficient-of-variation (CV), was higher for LR (3−5%) than nB and MLP (1−2%). Conclusion: Quantitative ultrasound with machine learning differentiated early and advanced fibrosis. Ultrasound B-mode images contain a high level of information to enable accurate diagnosis with relatively straightforward machine learning methods like naïve Bayes and logistic regression. Implementing simple ML approaches with QUS features in clinical settings could reduce the user-dependent limitation of ultrasound in detecting early-stage liver fibrosis.

Toward Reduction in False-Positive Thyroid Nodule Biopsies with a Deep Learning-based Risk Stratification System Using US Cine-Clip Images.

Purpose: To develop a deep learning-based risk stratification system for thyroid nodules using US cine images. Materials and Methods: In this retrospective study, 192 biopsy-confirmed thyroid nodules (175 benign, 17 malignant) in 167 unique patients (mean age, 56 years ± 16 [SD], 137 women) undergoing cine US between April 2017 and May 2018 with American College of Radiology (ACR) Thyroid Imaging Reporting and Data System (TI-RADS)-structured radiology reports were evaluated. A deep learning-based system that exploits the cine images obtained during three-dimensional volumetric thyroid scans and outputs malignancy risk was developed and compared, using fivefold cross-validation, against a two-dimensional (2D) deep learning-based model (Static-2DCNN), a radiomics-based model using cine images (Cine-Radiomics), and the ACR TI-RADS level, with histopathologic diagnosis as ground truth. The system was used to revise the ACR TI-RADS recommendation, and its diagnostic performance was compared against the original ACR TI-RADS. Results: The system achieved higher average area under the receiver operating characteristic curve (AUC, 0.88) than Static-2DCNN (0.72, P = .03) and tended toward higher average AUC than Cine-Radiomics (0.78, P = .16) and ACR TI-RADS level (0.80, P = .21). The system downgraded recommendations for 92 benign and two malignant nodules and upgraded none. The revised recommendation achieved higher specificity (139 of 175, 79.4%) than the original ACR TI-RADS (47 of 175, 26.9%; P < .001), with no difference in sensitivity (12 of 17, 71% and 14 of 17, 82%, respectively; P = .63). Conclusion: The risk stratification system using US cine images had higher diagnostic performance than prior models and improved specificity of ACR TI-RADS when used to revise ACR TI-RADS recommendation.Keywords: Neural Networks, US, Abdomen/GI, Head/Neck, Thyroid, Computer Applications-3D, Oncology, Diagnosis, Supervised Learning, Transfer Learning, Convolutional Neural Network (CNN) Supplemental material is available for this article. © RSNA, 2022.

Machine learning improves early detection of liver fibrosis by quantitative ultrasound radiomics.

Progression of liver fibrosis to cirrhosis, a severe non-reversible process, is one of the most critical risk factors in developing hepatocellular carcinoma and liver failure. Detection of liver fibrosis at an early stage is therefore essential for better patient management. Ultrasound (US) imaging can provide a noninvasive alternative to biopsies. This study evaluates quantitative US texture features to improve early-stage versus advanced liver fibrosis detection. 157 B-mode US images of different liver lobes acquired from early and advanced fibrosis rat cases were used for analysis. 5-6 regions of interest were placed on each image. Twelve quantitative features that describe liver texture changes were extracted from the images, including first-order histogram, run length (RL), and gray level co-occurrence matrix (GLCM). The diagnostic performance of individual features was high with AUC ranging from 0.80 to 0.94. Logistic regression with leave-one-out cross-validation was used to evaluate the performance of the combined features. All features combined showed a slight improvement in performance with AUC = 0.95, sensitivity = 96.8%, and specificity = 93.7%. Quantitative US texture features characterize liver fibrosis changes with high accuracy and can differentiate early from advanced disease. Quantitative ultrasound, if validated in future clinical studies, can have a potential role in identifying fibrosis changes that are not easily detected by visual US image assessments.

Deep Learning in Kidney Ultrasound: Overview, Frontiers, and Challenges.

Ultrasonography is a practical imaging technique used in numerous health care settings. It is relatively inexpensive, portable, and safe, and it has dynamic capabilities that make it an invaluable tool for a wide variety of diagnostic and interventional studies. Recently, there has been a revolution in medical imaging using artificial intelligence (AI). A particularly potent form of AI is deep learning, in which the computer learns to recognize pixel or written data on its own without the selection of predetermined features, usually through a specific neural network architecture. Neural networks vary in architecture depending on their task, and key design considerations include the number of layers and complexity, data available, technical requirements, and domain knowledge. Deep learning models offer the potential for promising innovations to workflow, image quality, and vision tasks in sonography. However, there are key limitations and challenges in creating reliable and safe AI models for patients and clinicians.

Quantification of abdominal fat from computed tomography using deep learning and its association with electronic health records in an academic biobank.

OBJECTIVE: The objective was to develop a fully automated algorithm for abdominal fat segmentation and to deploy this method at scale in an academic biobank. MATERIALS AND METHODS: We built a fully automated image curation and labeling technique using deep learning and distributive computing to identify subcutaneous and visceral abdominal fat compartments from 52,844 computed tomography scans in 13,502 patients in the Penn Medicine Biobank (PMBB). A classification network identified the inferior and superior borders of the abdomen, and a segmentation network differentiated visceral and subcutaneous fat. Following technical evaluation of our method, we conducted studies to validate known relationships with visceral and subcutaneous fat. RESULTS: When compared with 100 manually annotated cases, the classification network was on average within one 5-mm slice for both the superior (0.4 ± 1.1 slice) and inferior (0.4 ± 0.6 slice) borders. The segmentation network also demonstrated excellent performance with intraclass correlation coefficients of 1.00 (P < 2 × 10-16) for subcutaneous and 1.00 (P < 2 × 10-16) for visceral fat on 100 testing cases. We performed integrative analyses of abdominal fat with the phenome extracted from the electronic health record and found highly significant associations with diabetes mellitus, hypertension, and renal failure, among other phenotypes. CONCLUSIONS: This work presents a fully automated and highly accurate method for the quantification of abdominal fat that can be applied to routine clinical imaging studies to fuel translational scientific discovery.

Cross-Modal Data Programming Enables Rapid Medical Machine Learning.

A major bottleneck in developing clinically impactful machine learning models is a lack of labeled training data for model supervision. Thus, medical researchers increasingly turn to weaker, noisier sources of supervision, such as leveraging extractions from unstructured text reports to supervise image classification. A key challenge in weak supervision is combining sources of information that may differ in quality and have correlated errors. Recently, a statistical theory of weak supervision called data programming has shown promise in addressing this challenge. Data programming now underpins many deployed machine-learning systems in the technology industry, even for critical applications. We propose a new technique for applying data programming to the problem of cross-modal weak supervision in medicine, wherein weak labels derived from an auxiliary modality (e.g., text) are used to train models over a different target modality (e.g., images). We evaluate our approach on diverse clinical tasks via direct comparison to institution-scale, hand-labeled datasets. We find that our supervision technique increases model performance by up to 6 points area under the receiver operating characteristic curve (ROC-AUC) over baseline methods by improving both coverage and quality of the weak labels. Our approach yields models that on average perform within 1.75 points ROC-AUC of those supervised with physician-years of hand labeling and outperform those supervised with physician-months of hand labeling by 10.25 points ROC-AUC, while using only person-days of developer time and clinician work-a time saving of 96%. Our results suggest that modern weak supervision techniques such as data programming may enable more rapid development and deployment of clinically useful machine-learning models.

Spatial Characterization of Tumor Perfusion Properties from 3D DCE-US Perfusion Maps are Early Predictors of Cancer Treatment Response.

There is a need for noninvasive repeatable biomarkers to detect early cancer treatment response and spare non-responders unnecessary morbidities and costs. Here, we introduce three-dimensional (3D) dynamic contrast enhanced ultrasound (DCE-US) perfusion map characterization as inexpensive, bedside and longitudinal indicator of tumor perfusion for prediction of vascular changes and therapy response. More specifically, we developed computational tools to generate perfusion maps in 3D of tumor blood flow, and identified repeatable quantitative features to use in machine-learning models to capture subtle multi-parametric perfusion properties, including heterogeneity. Models were developed and trained in mice data and tested in a separate mouse cohort, as well as early validation clinical data consisting of patients receiving therapy for liver metastases. Models had excellent (ROC-AUC > 0.9) prediction of response in pre-clinical data, as well as proof-of-concept clinical data. Significant correlations with histological assessments of tumor vasculature were noted (Spearman R > 0.70) in pre-clinical data. Our approach can identify responders based on early perfusion changes, using perfusion properties correlated to gold-standard vascular properties.

Liver Imaging Reporting and Data System (LI-RADS) v2018: diagnostic value of ancillary features favoring malignancy in hypervascular observations ≥ 10 mm at intermediate (LR-3) and high probability (LR-4) for hepatocellular carcinoma.

OBJECTIVE: This study was conducted in order to assess the diagnostic accuracy of LI-RADS v2018 ancillary features (AFs) favoring malignancy applied to LR-3 and LR-4 observations on gadoxetate-enhanced MRI. METHODS: In this retrospective dual-institution study, we included consecutive patients at high risk for hepatocellular carcinoma (HCC) imaged with gadoxetate disodium-enhanced MRI between 2009 and 2014 fulfilling the following criteria: (i) at least one LR-3 or LR-4 observation ≥ 10 mm; (ii) nonrim arterial phase hyperenhancement; and (iii) confirmation of benignity or malignancy by pathology or imaging follow-up. We compared the distribution of AFs between HCCs and benign observations and the diagnostic performance for the diagnosis of HCC using univariate and multivariate analyses. Significance was set at p value < 0.05. RESULTS: Two hundred five observations were selected in 155 patients (108 M, 47 F) including 167 (81.5%) LR-3 and 38 (18.5%) LR-4. There were 126 (61.5%) HCCs and 79 (28.5%) benign lesions. A significantly larger number of AFs favoring malignancy were found in LR-3 and LR-4 lesions that progressed to HCC compared to benign lesions (p < 0.001 and p = 0.003, respectively). The most common AFs favoring malignancy in HCCs were hepatobiliary phase (HBP) hypointensity (p < 0.001), transitional phase hypointensity (p < 0.001), and mild-moderate T2 hyperintensity (p < 0.001). Sensitivity and specificity of AFs for the diagnosis of HCC ranged 0.8-76.2% and 86.1-100%, respectively. HBP hypointensity yielded the highest sensitivity but also the lowest specificity and was the only AF remaining independently associated with the diagnosis of HCC at multivariate logistic regression analysis (OR 14.83, 95% CI 5.81-42.76, p < 0.001). CONCLUSIONS: Among all AFs, HBP hypointensity yields the highest sensitivity for the diagnosis of HCC. KEY POINTS: • LR-3 and LR-4 observations diagnosed as HCC have a significantly higher number of ancillary features favoring malignancy compared to observations proven to be benign. • The presence of three or more ancillary features favoring malignancy has a high specificity (96.2%) for the diagnosis of HCC. • Among all ancillary features favoring malignancy, hepatobiliary phase hypointensity yields the highest sensitivity, but also the lowest specificity for the diagnosis of HCC.

A New Multimodel Machine Learning Framework to Improve Hepatic Fibrosis Grading Using Ultrasound Elastography Systems from Different Vendors.

The purpose of the work described here was to determine if the diagnostic performance of point and 2-D shear wave elastography (pSWE; 2-DSWE) using shear wave velocity (SWV) with a new machine learning (ML) technique applied to systems from different vendors is comparable to that of magnetic resonance elastography (MRE) in distinguishing non-significant (

Point Shear Wave Elastography Using Machine Learning to Differentiate Renal Cell Carcinoma and Angiomyolipoma.

The question of whether ultrasound point shear wave elastography can differentiate renal cell carcinoma (RCC) from angiomyolipoma (AML) is controversial. This study prospectively enrolled 51 patients with 52 renal tumors (42 RCCs, 10 AMLs). We obtained 10 measurements of shear wave velocity (SWV) in the renal tumor, cortex and medulla. Median SWV was first used to classify RCC versus AML. Next, the prediction accuracy of 4 machine learning algorithms-logistic regression, naïve Bayes, quadratic discriminant analysis and support vector machines (SVMs)-was evaluated, using statistical inputs from the tumor, cortex and combined statistical inputs from tumor, cortex and medulla. After leave-one-out cross validation, models were evaluated using the area under the receiver operating characteristic (ROC) curve (AUC). Tumor median SWV performed poorly (AUC = 0.62; p = 0.23). Except logistic regression, all machine learning algorithms reached statistical significance using combined statistical inputs (AUC = 0.78-0.98; p < 7.1 × 10-3). SVMs demonstrated 94% accuracy (AUC = 0.98; p = 3.13 × 10-6) and clearly outperformed median SWV in differentiating RCC from AML (p = 2.8 × 10-4).

Automatic inference of BI-RADS final assessment categories from narrative mammography report findings.

We propose an efficient natural language processing approach for inferring the BI-RADS final assessment categories by analyzing only the mammogram findings reported by the mammographer in narrative form. The proposed hybrid method integrates semantic term embedding with distributional semantics, producing a context-aware vector representation of unstructured mammography reports. A large corpus of unannotated mammography reports (300,000) was used to learn the context of the key-terms using a distributional semantics approach, and the trained model was applied to generate context-aware vector representations of the reports annotated with BI-RADS category (22,091). The vectorized reports were utilized to train a supervised classifier to derive the BI-RADS assessment class. Even though the majority of the proposed embedding pipeline is unsupervised, the classifier was able to recognize substantial semantic information for deriving the BI-RADS categorization not only on a holdout internal testset and also on an external validation set (1900 reports). Our proposed method outperforms a recently published domain-specific rule-based system and could be relevant for evaluating concordance between radiologists. With minimal requirement for task specific customization, the proposed method can be easily transferable to a different domain to support large scale text mining or derivation of patient phenotype.

A multi-model framework to estimate perfusion parameters using contrast-enhanced ultrasound imaging.

PURPOSE: Contrast-enhanced ultrasound imaging has expanded the diagnostic potential of ultrasound by enabling real-time imaging and quantification of tissue perfusion. Several perfusion models and curve fitting methods have been developed to quantify the temporal behavior of tracer signal and standardize perfusion quantification. While the least-squares approach has traditionally been applied for curve fitting, it can be inadequate for noisy and complex data. Moreover, previous research suggests that certain perfusion models may be more relevant depending on the organ or tissue imaged. We propose a multi-model framework to select the most appropriate perfusion model and curve fitting method for each diagnostic application. METHODS: Our multi-model approach uses a system identification method, which estimates perfusion parameters from the model with the best fit to a given time-intensity curve. We compared current perfusion quantification methods that use a single perfusion model and curve fitting method and our proposed multi-model framework on bolus 3D dynamic contrast-enhanced ultrasound (DCE-US) in vivo images obtained in mice implanted with a colon cancer, as well as on simulation data. The quality of fit in estimating perfusion parameters was evaluated using the Spearman correlation coefficient, the coefficient of determination (R2 ), and the normalized root-mean-square error (NRMSE) to ensure that the multi-model framework finds the best perfusion model and curve fitting algorithm. RESULTS: Our multi-model framework outperforms conventional single perfusion model approaches with least-squares optimization, providing more robust perfusion parameter estimation. R2 and NRMSE are 0.98 and 0.18, respectively, for our proposed method. By comparison, the performance of the traditional approach is much more dependent upon the selection of the appropriate model. The R2 and NRMSE are 0.91 and 0.31, respectively. CONCLUSIONS: The proposed multi-model framework for perfusion modeling outperforms the current approach of single perfusion modeling using least-squares optimization and more robustly estimates perfusion parameters when using empiric data labeled by an expert as the gold standard. Our technique is minimally sensitive to issues affecting the accuracy of perfusion parameter estimation, including rise time, noise, region of interest size, and frame rate. This framework could be of key utility in modeling different perfusion systems in different tissues and organs.

A radiogenomic analysis of hepatocellular carcinoma: association between fractional allelic imbalance rate index and the liver imaging reporting and data system (LI-RADS) categories and features.

OBJECTIVE: To evaluate the association between the liver imaging reporting and data system (LI-RADS) categories and features and the fractional allelic imbalance (FAI) rate index of hepatocellular carcinoma (HCC). METHODS: The institutional review board approved this retrospective study. Medical records collected between January 2008 and December 2013 were reviewed to find patients with histologically confirmed HCC, FAI analysis, and CT or MR imaging of the liver. The final population included 71 patients (54 males, 17 females). Three radiologists reviewed the images using the LI-RADS v. 2014. The association between FAI and LI-RADS categories and features was tested using the Spearman's rank correlation coefficient (rho) and the Wilcoxon rank-sum test [low FAI (<40%) vs high FAI (≥40%)]. A p value < 0.007 was used as the threshold for statistical significance after application of the Bonferroni correction for multiple comparisons. RESULTS: HCCs were classified as LR-3 (n = 4), LR-4 (n = 22), and LR-5 (n = 45). There was a positive correlation (rho = 0.264) between FAI rate index and LI-RADS category, although not statistically significant after Bonferroni correction (p = 0.024). 14 of the 20 (70%) HCCs with high FAI (≥40%) were categorized as LR-5, 6/20 (30%) as LR-4 and none as LR-3 (p = 0.377). Among the evaluated LI-RADS imaging features, only lesion size showed a statistically significant different distribution in tumors with high FAI compared to those with low FAI. HCCs with FAI ≥40% were larger (56 ± 42 mm) compared to those with FAI <40% (36 ± 30 mm; p = 0.005). CONCLUSION: There was a positive correlation, although not statistically significant, between the LI-RADS diagnostic categories and the FAI rate of HCC. Tumors with high FAI were larger compared to those with low FAI. Advances in knowledge: HCCs with high (≥40%) FAI are larger compared to those with low (<40%) FAI.

Differences in Antipsychotic-Related Adverse Events in Adult, Pediatric, and Geriatric Populations.

In recent years, antipsychotic medications have increasingly been used in pediatric and geriatric populations, despite the fact that many of these drugs were approved based on clinical trials in adult patients only. Preliminary studies have shown that the "off-label" use of these drugs in pediatric and geriatric populations may result in adverse events not found in adults. In this study, we utilized the large-scale U.S. Food and Drug Administration (FDA) Adverse Events Reporting System (AERS) database to look at differences in adverse events from antipsychotics among adult, pediatric, and geriatric populations. We performed a systematic analysis of the FDA AERS database using MySQL by standardizing the database using structured terminologies and ontologies. We compared adverse event profiles of atypical versus typical antipsychotic medications among adult (18-65), pediatric (age < 18), and geriatric (> 65) populations. We found statistically significant differences between the number of adverse events in the pediatric versus adult populations with aripiprazole, clozapine, fluphenazine, haloperidol, olanzapine, quetiapine, risperidone, and thiothixene, and between the geriatric versus adult populations with aripiprazole, chlorpromazine, clozapine, fluphenazine, haloperidol, paliperidone, promazine, risperidone, thiothixene, and ziprasidone (p < 0.05, with adjustment for multiple comparisons). Furthermore, the particular types of adverse events reported also varied significantly between each population for aripiprazole, clozapine, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone (Chi-square, p < 10-6). Diabetes was the most commonly reported side effect in the adult population, compared to behavioral problems in the pediatric population and neurologic symptoms in the geriatric population. We also found discrepancies between the frequencies of reports in AERS and in the literature. Our analysis of the FDA AERS database shows that there are significant differences in both the numbers and types of adverse events among these age groups and between atypical and typical antipsychotics. It is important for clinicians to be mindful of these differences when prescribing antipsychotics, especially when prescribing medications off-label.

Pathway analysis of genome-wide data improves warfarin dose prediction.

BACKGROUND: Many genome-wide association studies focus on associating single loci with target phenotypes. However, in the setting of rare variation, accumulating sufficient samples to assess these associations can be difficult. Moreover, multiple variations in a gene or a set of genes within a pathway may all contribute to the phenotype, suggesting that the aggregation of variations found over the gene or pathway may be useful for improving the power to detect associations. RESULTS: Here, we present a method for aggregating single nucleotide polymorphisms (SNPs) along biologically relevant pathways in order to seek genetic associations with phenotypes. Our method uses all available genetic variants and does not remove those in linkage disequilibrium (LD). Instead, it uses a novel SNP weighting scheme to down-weight the contributions of correlated SNPs. We apply our method to three cohorts of patients taking warfarin: two European descent cohorts and an African American cohort. Although the clinical covariates and key pharmacogenetic loci for warfarin have been characterized, our association metric identifies a significant association with mutations distributed throughout the pathway of warfarin metabolism. We improve dose prediction after using all known clinical covariates and pharmacogenetic variants in VKORC1 and CYP2C9. In particular, we find that at least 1% of the missing heritability in warfarin dose may be due to the aggregated effects of variations in the warfarin metabolic pathway, even though the SNPs do not individually show a significant association. CONCLUSIONS: Our method allows researchers to study aggregative SNP effects in an unbiased manner by not preselecting SNPs. It retains all the available information by accounting for LD-structure through weighting, which eliminates the need for LD pruning.

Genetic variants associated with warfarin dose in African-American individuals: a genome-wide association study.

BACKGROUND: VKORC1 and CYP2C9 are important contributors to warfarin dose variability, but explain less variability for individuals of African descent than for those of European or Asian descent. We aimed to identify additional variants contributing to warfarin dose requirements in African Americans. METHODS: We did a genome-wide association study of discovery and replication cohorts. Samples from African-American adults (aged ≥18 years) who were taking a stable maintenance dose of warfarin were obtained at International Warfarin Pharmacogenetics Consortium (IWPC) sites and the University of Alabama at Birmingham (Birmingham, AL, USA). Patients enrolled at IWPC sites but who were not used for discovery made up the independent replication cohort. All participants were genotyped. We did a stepwise conditional analysis, conditioning first for VKORC1 -1639G→A, followed by the composite genotype of CYP2C9*2 and CYP2C9*3. We prespecified a genome-wide significance threshold of p<5×10(-8) in the discovery cohort and p<0·0038 in the replication cohort. FINDINGS: The discovery cohort contained 533 participants and the replication cohort 432 participants. After the prespecified conditioning in the discovery cohort, we identified an association between a novel single nucleotide polymorphism in the CYP2C cluster on chromosome 10 (rs12777823) and warfarin dose requirement that reached genome-wide significance (p=1·51×10(-8)). This association was confirmed in the replication cohort (p=5·04×10(-5)); analysis of the two cohorts together produced a p value of 4·5×10(-12). Individuals heterozygous for the rs12777823 A allele need a dose reduction of 6·92 mg/week and those homozygous 9·34 mg/week. Regression analysis showed that the inclusion of rs12777823 significantly improves warfarin dose variability explained by the IWPC dosing algorithm (21% relative improvement). INTERPRETATION: A novel CYP2C single nucleotide polymorphism exerts a clinically relevant effect on warfarin dose in African Americans, independent of CYP2C9*2 and CYP2C9*3. Incorporation of this variant into pharmacogenetic dosing algorithms could improve warfarin dose prediction in this population. FUNDING: National Institutes of Health, American Heart Association, Howard Hughes Medical Institute, Wisconsin Network for Health Research, and the Wellcome Trust.