Improving the Detection of Potential Cases of Familial Hypercholesterolemia: Could Machine Learning Be Part of the Solution?

BACKGROUND: Familial hypercholesterolemia (FH), while highly prevalent, is a significantly underdiagnosed monogenic disorder. Improved detection could reduce the large number of cardiovascular events attributable to poor case finding. We aimed to assess whether machine learning algorithms outperform clinical diagnostic criteria (signs, history, and biomarkers) and the recommended screening criteria in the United Kingdom in identifying individuals with FH-causing variants, presenting a scalable screening criteria for general populations. METHODS AND RESULTS: Analysis included UK Biobank participants with whole exome sequencing, classifying them as having FH when (likely) pathogenic variants were detected in their LDLR, APOB, or PCSK9 genes. Data were stratified into 3 data sets for (1) feature importance analysis; (2) deriving state-of-the-art statistical and machine learning models; (3) evaluating models' predictive performance against clinical diagnostic and screening criteria: Dutch Lipid Clinic Network, Simon Broome, Make Early Diagnosis to Prevent Early Death, and Familial Case Ascertainment Tool. One thousand and three of 454 710 participants were classified as having FH. A Stacking Ensemble model yielded the best predictive performance (sensitivity, 74.93%; precision, 0.61%; accuracy, 72.80%, area under the receiver operating characteristic curve, 79.12%) and outperformed clinical diagnostic criteria and the recommended screening criteria in identifying FH variant carriers within the validation data set (figures for Familial Case Ascertainment Tool, the best baseline model, were 69.55%, 0.44%, 65.43%, and 71.12%, respectively). Our model decreased the number needed to screen compared with the Familial Case Ascertainment Tool (164 versus 227). CONCLUSIONS: Our machine learning-derived model provides a higher pretest probability of identifying individuals with a molecular diagnosis of FH compared with current approaches. This provides a promising, cost-effective scalable tool for implementation into electronic health records to prioritize potential FH cases for genetic confirmation.

Overview of a collaborative global effort to address the burden of familial hypercholesterolaemia.

This is an overview of the EAS Familial Hypercholesterolaemia (FH) Studies Collaboration (FHSC) global consortium and registry (established 2015), which broadly addresses the global burden of FH. Eighty-seven National Lead Investigators from 74 countries form this expanding global consortium, and this global registry currently includes pooled data on 70,000 participants from participating countries to facilitate FH surveillance. Published first results from this global registry concluded that FH is diagnosed late, and management of LDL-cholesterol falls below guideline recommendations, and therefore earlier detection of FH and wider use of combination therapy is required. Further FHSC studies will follow on updated data including new countries, participants and variables, and non-DNA genetic information, and on the remaining cohorts in the registry. FHSC cross-sectional collaborative global studies are expected to promote FH detection earlier in life to subsequently initiate early lipid lowering therapy to reduce lifelong exposure to cumulative LDL-cholesterol thus reducing cardiovascular disease risk.

Ensemble machine learning methods in screening electronic health records: A scoping review.

Background: Electronic health records provide the opportunity to identify undiagnosed individuals likely to have a given disease using machine learning techniques, and who could then benefit from more medical screening and case finding, reducing the number needed to screen with convenience and healthcare cost savings. Ensemble machine learning models combining multiple prediction estimates into one are often said to provide better predictive performances than non-ensemble models. Yet, to our knowledge, no literature review summarises the use and performances of different types of ensemble machine learning models in the context of medical pre-screening. Method: We aimed to conduct a scoping review of the literature reporting the derivation of ensemble machine learning models for screening of electronic health records. We searched EMBASE and MEDLINE databases across all years applying a formal search strategy using terms related to medical screening, electronic health records and machine learning. Data were collected, analysed, and reported in accordance with the PRISMA scoping review guideline. Results: A total of 3355 articles were retrieved, of which 145 articles met our inclusion criteria and were included in this study. Ensemble machine learning models were increasingly employed across several medical specialties and often outperformed non-ensemble approaches. Ensemble machine learning models with complex combination strategies and heterogeneous classifiers often outperformed other types of ensemble machine learning models but were also less used. Ensemble machine learning models methodologies, processing steps and data sources were often not clearly described. Conclusions: Our work highlights the importance of deriving and comparing the performances of different types of ensemble machine learning models when screening electronic health records and underscores the need for more comprehensive reporting of machine learning methodologies employed in clinical research.

Prevalence of Familial Hypercholesterolemia Among the General Population and Patients With Atherosclerotic Cardiovascular Disease: A Systematic Review and Meta-Analysis.

BACKGROUND: Contemporary studies suggest that familial hypercholesterolemia (FH) is more frequent than previously reported and increasingly recognized as affecting individuals of all ethnicities and across many regions of the world. Precise estimation of its global prevalence and prevalence across World Health Organization regions is needed to inform policies aiming at early detection and atherosclerotic cardiovascular disease (ASCVD) prevention. The present study aims to provide a comprehensive assessment and more reliable estimation of the prevalence of FH than hitherto possible in the general population (GP) and among patients with ASCVD. METHODS: We performed a systematic review and meta-analysis including studies reporting on the prevalence of heterozygous FH in the GP or among those with ASCVD. Studies reporting gene founder effects and focused on homozygous FH were excluded. The search was conducted through Medline, Embase, Cochrane, and Global Health, without time or language restrictions. A random-effects model was applied to estimate the overall pooled prevalence of FH in the general and ASCVD populations separately and by World Health Organization regions. RESULTS: From 3225 articles, 42 studies from the GP and 20 from populations with ASCVD were eligible, reporting on 7 297 363 individuals/24 636 cases of FH and 48 158 patients/2827 cases of FH, respectively. More than 60% of the studies were from Europe. Use of the Dutch Lipid Clinic Network criteria was the commonest diagnostic method. Within the GP, the overall pooled prevalence of FH was 1:311 (95% CI, 1:250-1:397; similar between children [1:364] and adults [1:303], P=0.60; across World Health Organization regions where data were available, P=0.29; and between population-based and electronic health records-based studies, P=0.82). Studies with ≤10 000 participants reported a higher prevalence (1:200-289) compared with larger cohorts (1:365-407; P<0.001). The pooled prevalence among those with ASCVD was 18-fold higher than in the GP (1:17 [95% CI, 1:12-1:24]), driven mainly by coronary artery disease (1:16; [95% CI, 1:12-1:23]). Between-study heterogeneity was large (I2>95%). Tests assessing bias were nonsignificant (P>0.3). CONCLUSIONS: With an overall prevalence of 1:311, FH is among the commonest genetic disorders in the GP, similarly present across different regions of the world, and is more frequent among those with ASCVD. The present results support the advocacy for the institution of public health policies, including screening programs, to identify FH early and to prevent its global burden.

Overview of the current status of familial hypercholesterolaemia care in over 60 countries - The EAS Familial Hypercholesterolaemia Studies Collaboration (FHSC).

BACKGROUND AND AIMS: Management of familial hypercholesterolaemia (FH) may vary across different settings due to factors related to population characteristics, practice, resources and/or policies. We conducted a survey among the worldwide network of EAS FHSC Lead Investigators to provide an overview of FH status in different countries. METHODS: Lead Investigators from countries formally involved in the EAS FHSC by mid-May 2018 were invited to provide a brief report on FH status in their countries, including available information, programmes, initiatives, and management. RESULTS: 63 countries provided reports. Data on FH prevalence are lacking in most countries. Where available, data tend to align with recent estimates, suggesting a higher frequency than that traditionally considered. Low rates of FH detection are reported across all regions. National registries and education programmes to improve FH awareness/knowledge are a recognised priority, but funding is often lacking. In most countries, diagnosis primarily relies on the Dutch Lipid Clinics Network criteria. Although available in many countries, genetic testing is not widely implemented (frequent cost issues). There are only a few national official government programmes for FH. Under-treatment is an issue. FH therapy is not universally reimbursed. PCSK9-inhibitors are available in ∼2/3 countries. Lipoprotein-apheresis is offered in ∼60% countries, although access is limited. CONCLUSIONS: FH is a recognised public health concern. Management varies widely across countries, with overall suboptimal identification and under-treatment. Efforts and initiatives to improve FH knowledge and management are underway, including development of national registries, but support, particularly from health authorities, and better funding are greatly needed.

Height and body fatness and colorectal cancer risk: an update of the WCRF-AICR systematic review of published prospective studies.

PURPOSE: There is no published dose-response meta-analysis on the association between height and colorectal cancer risk (CRC) by sex and anatomical sub-site. We conducted a meta-analysis of prospective studies on the association between height and CRC risk with subgroup analysis and updated evidence on the association between body fatness and CRC risk. METHODS: PubMed and several other databases were searched up to November 2016. A random effects model was used to calculate dose-response summary relative risks (RR's). RESULTS: 47 studies were included in the meta-analyses including 50,936 cases among 7,393,510 participants. The findings support the existing evidence regarding a positive association of height, general and abdominal body fatness and CRC risk. The summary RR were 1.04 [95% (CI)1.02-1.05, I² = 91%] per 5 cm increase in height, 1.02 [95% (CI)1.01-1.02, I² = 0%] per 5 kg increase in weight, 1.06 [95% (CI)1.04-1.07, I² = 83%] per 5 kg/m2 increase in BMI, 1.02 [95% (CI)1.02-1.03, I² = 4%] per 10 cm increase in waist circumference, 1.03 [95% (CI)1.01-1.05, I² = 16%] per 0.1 unit increase in waist to hip ratio. The significant association for height and CRC risk was similar in men and women. The significant association for BMI and CRC risk was stronger in men than in women. CONCLUSION: The positive association between height and risk of CRC suggests that life factors during childhood and early adulthood might play a role in CRC aetiology. Higher general and abdominal body fatness during adulthood are risk factors of CRC and these associations are stronger in men than in women.

Carbohydrates, glycemic index, glycemic load, and breast cancer risk: a systematic review and dose-response meta-analysis of prospective studies.

Context: The investigation of dose-response associations between carbohydrate intake, glycemic index, glycemic load, and risk of breast cancer stratified by menopausal status, hormone receptor status, and body mass index (BMI) remains inconclusive. Objective: A systematic review and dose-response meta-analyses was conducted to investigate these associations. Data Sources: As part of the World Cancer Research Fund/American Institute for Cancer Research Continuous Update Project, PubMed was searched up to May 2015 for relevant studies on these associations. Study Selection: Prospective studies reporting associations between carbohydrate intake, glycemic index, or glycemic load and breast cancer risk were included. Data Extraction: Two investigators independently extracted data from included studies. Results: Random-effects models were used to summarize relative risks (RRs) and 95%CIs. Heterogeneity between subgroups, including menopausal status, hormone receptor status, and BMI was explored using meta-regression. Nineteen publications were included. The summary RRs (95%CIs) for breast cancer were 1.04 (1.00-1.07) per 10 units/d for glycemic index, 1.01 (0.98-1.04) per 50 units/d for glycemic load, and 1.00 (0.96-1.05) per 50 g/d for carbohydrate intake. For glycemic index, the association appeared slightly stronger among postmenopausal women (summary RR per 10 units/d, 1.06; 95%CI, 1.02-1.10) than among premenopausal women, though the difference was not statistically significant (Pheterogeneity = 0.15). Glycemic load and carbohydrate intake were positively associated with breast cancer among postmenopausal women with estrogen-negative tumors (summary RR for glycemic load, 1.28; 95%CI, 1.08-1.52; and summary RR for carbohydrates, 1.13; 95%CI, 1.02-1.25). No differences in BMI were detected. Conclusions: Menopausal and hormone receptor status, but not BMI, might be potential influencing factors for the associations between carbohydrate intake, glycemic index, glycemic load, and breast cancer.

Blood concentrations of carotenoids and retinol and lung cancer risk: an update of the WCRF-AICR systematic review of published prospective studies.

Carotenoids and retinol are considered biomarkers of fruits and vegetables intake, and are of much interest because of their anti-inflammatory and antioxidant properties; however, there is inconsistent evidence regarding their protective effects against lung cancer. We conducted a meta-analysis of prospective studies of blood concentrations of carotenoids and retinol, and lung cancer risk. We identified relevant prospective studies published up to December 2014 by searching the PubMed and several other databases. We calculated summary estimates of lung cancer risk for the highest compared with lowest carotenoid and retinol concentrations and dose-response meta-analyses using random effects models. We used fractional polynomial models to assess potential nonlinear relationships. Seventeen prospective studies (18 publications) including 3603 cases and 458,434 participants were included in the meta-analysis. Blood concentrations of α-carotene, ß-carotene, total carotenoids, and retinol were significantly inversely associated with lung cancer risk or mortality. The summary relative risk were 0.66 (95% confidence interval [CI]: 0.55-0.80) per 5 µg/100 mL of α-carotene (studies [n] = 5), 0.84 (95% CI: 0.76-0.94) per 20 µg/100 mL of ß-carotene (n = 9), 0.66 (95% CI: 0.54-0.81) per 100 µg/100 mL of total carotenoids (n = 4), and 0.81 (95% CI: 0.73-0.90) per 70 µg/100 mL of retinol (n = 8). In stratified analysis by sex, the significant inverse associations for ß-carotene and retinol were observed only in men and not in women. Nonlinear associations were observed for ß-carotene, ß-cryptoxanthin, and lycopene, with stronger associations observed at lower concentrations. There were not enough data to conduct stratified analyses by smoking. In conclusion, higher blood concentrations of several carotenoids and retinol are associated with reduced lung cancer risk. Further studies in never and former smokers are needed to rule out confounding by smoking.

An update of the WCRF/AICR systematic literature review on esophageal and gastric cancers and citrus fruits intake.

PURPOSE: The 2007 World Cancer Research Fund/American Institute for Cancer Research expert report concluded that foods containing vitamin C probably protect against esophageal cancer and fruits probably protect against gastric cancer. Most of the previous evidence was from case-control studies, which may be affected by recall and selection biases. More recently, several cohort studies have examined these associations. We conducted a systematic literature review of prospective studies on citrus fruits intake and risk of esophageal and gastric cancers. METHODS: PubMed was searched for studies published until 1 March 2016. We calculated summary relative risks and 95 % confidence intervals (95 % CI) using random-effects models. RESULTS: With each 100 g/day increase of citrus fruits intake, a marginally significant decreased risk of esophageal cancer was observed (summary RR 0.86, 95 % CI 0.74-1.00, 1,057 cases, six studies). The associations were similar for squamous cell carcinoma (RR 0.87, 95 % CI 0.69-1.08, three studies) and esophageal adenocarcinoma (RR 0.93, 95 % CI 0.78-1.11, three studies). For gastric cancer, the nonsignificant inverse association was observed for gastric cardia cancer (RR 0.75, 95 % CI 0.55-1.01, three studies), but not for gastric non-cardia cancer (RR 1.02, 95 % CI 0.90-1.16, four studies). Consistent summary inverse associations were observed when comparing the highest with lowest intake, with statistically significant associations for esophageal (RR 0.77, 95 % CI 0.64-0.91, seven studies) and gastric cardia cancers (RR 0.62, 95 % CI 0.39-0.99, three studies). CONCLUSIONS: Citrus fruits may decrease the risk of esophageal and gastric cardia cancers, but further studies are needed.

Synthesis of a peroxime proliferator activated receptor (PPAR) alpha/gamma agonist via stereocontrolled Williamson ether synthesis and stereospecific SN2 reaction of S-2-chloro propionic acid with phenoxides.

The stereospecific synthesis of the PPAR alpha/gamma agonist 1 was accomplished via ethylation of the optically pure trihydroxy derivative 6, itself derived via an enzymatic resolution. The ethylation can be accomplished without epimerization only under strict control of the reaction conditions and the choice of base (sodium tert-amylate), temperature (-30 degrees C), order of addition, and solvent (DMF). The key diastereospecific SN2 reaction of the phenol 4 with S-2-chloropropionic acid is best achieved via the sodium phenoxide of 4 derived from Na0 as the reagent of choice. The structure elucidation and key purification protocols to achieve pharmaceutical purity will also be described.