Evaluating a clinical tool (FAMCAT) for identifying familial hypercholesterolaemia in primary care: a retrospective cohort study

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.

Management of Familial Hypercholesterolemia with Special Emphasis on Evinacumab

Familial hypercholesterolemia (FH) is an underdiagnosed disease that contributes to a significant number of cardiovascular incidents through high serum Low-Density Lipoprotein Cholesterol (LDL-C) values. Its treatment primarily requires healthy lifestyle and therapy based on statins, ezetimibe and Proprotein Convertase Subtilisin/Kexin Type 9 (PCSK9); however, there are also new treatment options that can be used in patients who do not respond to therapy, among which we highlight evinacumab. Elevated LDL-C values, together with clinical manifestations associated with cholesterol deposition (e.g., tendon xanthomas, xanthelasma and arcus cornealis) and family history are the main elements in the diagnosis of FH. Pathognomonic signs of FH include extensor tendon xanthomas; however, their absence does not exclude the diagnosis. Elevated LDL-C levels lead to premature Atherosclerotic Cardiovascular Disease (ASCVD), which is why early diagnosis and treatment of FH is essential. Evinacumab, a novelty in pharmacological practice, having a complex mechanism of action, causes desirable changes in lipid parameters in patients with homozygous form of familial hypercholesterolemia (HoFH). This review collects and summarizes the most important aspects of the new drug, especially being a discovery in the treatment of HoFH, giving these patients hope for a longer and more comfortable life.

Electronic health record-based facilitation of familial hypercholesterolaemia detection sensitivity of different algorithms in genetically confirmed patients

Aims

Familial hypercholesterolaemia (FH) is a disorder of LDL cholesterol clearance, resulting in increased risk of cardiovascular disease. Recently, we developed a Dutch Lipid Clinic Network (DLCN) criteria-based algorithm to facilitate FH detection in electronic health records (EHRs). In this study, we investigated the sensitivity of this and other algorithms in a genetically confirmed FH population.

Methods and results

All patients with a healthcare insurance-related coded diagnosis of 'primary dyslipidaemia' between 2018 and 2020 were assessed for genetically confirmed FH. Data were extracted at the time of genetic confirmation of FH (T1) and during the first visit in 2018-2020 (T2). We assessed the sensitivity of algorithms on T1 and T2 for DLCN ≥ 6 and compared with other algorithms [familial hypercholesterolaemia case ascertainment tool (FAMCAT), Make Early Diagnoses to Prevent Early Death (MEDPED), and Simon Broome (SB)] using EHR-coded data and using all available data (i.e. including non-coded free text). 208 patients with genetically confirmed FH were included. The sensitivity (95% CI) on T1 and T2 with EHR-coded data for DLCN ≥ 6 was 19% (14-25%) and 22% (17-28%), respectively. When using all available data, the sensitivity for DLCN ≥ 6 was 26% (20-32%) on T1 and 28% (22-34%) on T2. For FAMCAT, the sensitivity with EHR-coded data on T1 was 74% (67-79%) and 32% (26-39%) on T2, whilst sensitivity with all available data was 81% on T1 (75-86%) and 45% (39-52%) on T2. For Make Early Diagnoses to Prevent Early Death MEDPED and SB, using all available data, the sensitivity on T1 was 31% (25-37%) and 17% (13-23%), respectively.

Conclusions

The FAMCAT algorithm had significantly better sensitivity than DLCN, MEDPED, and SB. FAMCAT has the best potential for FH case-finding using EHRs.

A mixed methods study of the awareness and management of familial hypercholesterolaemia in Irish general practice

Introduction

Familial Hypercholesterolemia (FH) is one of the most common genetic disorders, with an estimated global prevalence of 1:200-500, which leads to premature cardiovascular disease. Nevertheless, public and professional awareness of FH is often lacking, with an estimated 20,000 largely undiagnosed cases in Ireland.

Purpose

The overall aim of the project was to test the feasibility of a model of care that would include electronic record screening, clinical assessment, and coding of possible FH patients across a network of general practices in Ireland. In addition, a secondary aim was to gauge the awareness and knowledge of FH across the network.

Methods

This study took part in multiple phases, employing a mixed methods design. The study included a validated questionnaire, tailored online educational resources, a retrospective chart review of patients with a history of elevated LDL cholesterol (LDLc) and an active review with a selection of those patients. Results were analyzed using SPSS V27, where descriptive statistics and relevant correlation tests were employed.

Results

Eighteen general practices agreed to take part in the study. In the initial survey, respondents rated their personal and practice familiarity with FH as slightly below average. Around one-third of respondents were not aware of FH guidelines. Of over 55,000 adult patient records searched, only 0.2% had a recorded FH diagnosis and 3.9% had ever had an LDLc above 4.9 mmol/l. Eight practices completed 198 chart reviews. Among these, 29.8% of patients had a family history recorded, and 22.2% had a family history of CVD recorded. Female patients had higher averages for highest and recent LDLc. Seventy patients underwent a clinical review-with 27% of these patients identified as "probable" or "definite FH." There was a statistically significant (p = 0.002) relationship between FH status and whether the patient had other CVD risk factors.

Conclusion

General practitioners in Ireland had similar levels of awareness of FH compared to findings from elsewhere. The activities discussed encouraged clinicians to consider FH when talking to their patients, especially those with elevated LDLc at an early age. Broader awareness of the condition could increase conversations about FH and benefit patient outcomes.

Applying implementation science to improve care for familial hypercholesterolemia

PURPOSE OF REVIEW

Improving care of individuals with familial hypercholesteremia (FH) is reliant on the synthesis of evidence-based guidelines and their subsequent implementation into clinical care. This review describes implementation strategies, defined as methods to improve translation of evidence into FH care, that have been mapped to strategies from the Expert Recommendations for Implementing Change (ERIC) compilation.

RECENT FINDINGS

A search using the term 'familial hypercholesterolemia' returned 1350 articles from November 2018 to July 2021. Among these, there were 153 articles related to improving FH care; 1156 were excluded and the remaining 37 were mapped to the ERIC compilation of strategies: assess for readiness and identify barriers and facilitators [9], develop and organize quality monitoring systems [14], create new clinical teams [2], facilitate relay of clinical data to providers [4], and involve patients and family members [8]. There were only 8 of 37 studies that utilized an implementation science theory, model, or framework and two that explicitly addressed health disparities or equity.

SUMMARY

The mapping of the studies to implementation strategies from the ERIC compilation provides a framework for organizing current strategies to improve FH care. This study identifies potential areas for the development of implementation strategies to target unaddressed aspects of FH care.

Comparing the performance of the novel FAMCAT algorithms and established case-finding criteria for familial hypercholesterolaemia in primary care

OBJECTIVE

Familial hypercholesterolaemia (FH) is a common inherited disorder causing premature coronary heart disease (CHD) and death. We have developed the novel Familial Hypercholesterolaemia Case Ascertainment Tool (FAMCAT 1) case-finding algorithm for application in primary care, to improve detection of FH. The performance of this algorithm was further improved by including personal history of premature CHD (FAMCAT 2 algorithm). This study has evaluated their performance, at 95% specificity, to detect genetically confirmed FH in the general population. We also compared these algorithms to established clinical case-finding criteria.

METHODS

Prospective validation study, in 14 general practices, recruiting participants from the general adult population with cholesterol documented. For 260 participants with available health records, we determined possible FH cases based on FAMCAT thresholds, Dutch Lipid Clinic Network (DLCN) score, Simon-Broome criteria and recommended cholesterol thresholds (total cholesterol >9.0 mmol/L if ≥30 years or >7.5 mmol/L if <30 years), using clinical data from electronic and manual extraction of patient records and family history questionnaires. The reference standard was genetic testing. We examined detection rate (DR), sensitivity and specificity for each case-finding criteria.

RESULTS

At 95% specificity, FAMCAT 1 had a DR of 27.8% (95% CI 12.5% to 50.9%) with sensitivity of 31.2% (95% CI 11.0% to 58.7%); while FAMCAT 2 had a DR of 45.8% (95% CI 27.9% to 64.9%) with sensitivity of 68.8% (95% CI 41.3% to 89.0%). DLCN score ≥6 points yielded a DR of 35.3% (95% CI 17.3% to 58.7%) and sensitivity of 37.5% (95% CI 15.2% to 64.6%). Using recommended cholesterol thresholds resulted in DR of 28.0% (95% CI 14.3% to 47.6%) with sensitivity of 43.8% (95% CI 19.8% to 70.1%). Simon-Broome criteria had lower DR 11.3% (95% CI 6.0% to 20.0%) and specificity 70.9% (95% CI 64.8% to 76.5%) but higher sensitivity of 56.3% (95% CI 29.9% to 80.2%).

CONCLUSIONS

In primary care, in patients with cholesterol documented, FAMCAT 2 performs better than other case-finding criteria for detecting genetically confirmed FH, with no prior clinical review required for case finding.

TRIAL REGISTRATION NUMBER

NCT03934320.

Performance and clinical utility of supervised machine-learning approaches in detecting familial hypercholesterolaemia in primary care

Familial hypercholesterolaemia (FH) is a common inherited disorder, causing lifelong elevated low-density lipoprotein cholesterol (LDL-C). Most individuals with FH remain undiagnosed, precluding opportunities to prevent premature heart disease and death. Some machine-learning approaches improve detection of FH in electronic health records, though clinical impact is under-explored. We assessed performance of an array of machine-learning approaches for enhancing detection of FH, and their clinical utility, within a large primary care population. A retrospective cohort study was done using routine primary care clinical records of 4,027,775 individuals from the United Kingdom with total cholesterol measured from 1 January 1999 to 25 June 2019. Predictive accuracy of five common machine-learning algorithms (logistic regression, random forest, gradient boosting machines, neural networks and ensemble learning) were assessed for detecting FH. Predictive accuracy was assessed by area under the receiver operating curves (AUC) and expected vs observed calibration slope; with clinical utility assessed by expected case-review workload and likelihood ratios. There were 7928 incident diagnoses of FH. In addition to known clinical features of FH (raised total cholesterol or LDL-C and family history of premature coronary heart disease), machine-learning (ML) algorithms identified features such as raised triglycerides which reduced the likelihood of FH. Apart from logistic regression (AUC, 0.81), all four other ML approaches had similarly high predictive accuracy (AUC > 0.89). Calibration slope ranged from 0.997 for gradient boosting machines to 1.857 for logistic regression. Among those screened, high probability cases requiring clinical review varied from 0.73% using ensemble learning to 10.16% using deep learning, but with positive predictive values of 15.5% and 2.8% respectively. Ensemble learning exhibited a dominant positive likelihood ratio (45.5) compared to all other ML models (7.0-14.4). Machine-learning models show similar high accuracy in detecting FH, offering opportunities to increase diagnosis. However, the clinical case-finding workload required for yield of cases will differ substantially between models.