Statins in Familial Hypercholesterolemia: Translating Evidence to Action∗
Familial hypercholesterolemia (FH) is an autosomal dominant genetic disorder caused by mutations in 3 genes: LDLR, principally, APOB, less frequently, and PCSK9, rarely. Causal mutations, of which there are more than 1,500 in LDLR alone, disrupt the ability of the liver to recycle low-density lipoprotein cholesterol (LDL-C), thereby leading to lifelong, usually severe elevations in plasma LDL-C. In adults, that generally translates to an LDL-C level >190 mg/dl accompanied by a marked increased risk of coronary artery disease (CAD) (1,2). Natural history studies in heterozygous FH patients show the median age of myocardial infarction is 50 years for men and ∼60 years for women (3), with a (>50-fold) elevated relative risk of mortality at younger ages when there are few competing atherosclerotic cardiovascular disease risk factors (4).
For more than 75 years, individuals with FH have actively been contributing to our understanding of LDL-C metabolism (2,5). Indeed, studies in FH patients have provided a linchpin for the now axiomatic principle that LDL-C is a causative factor for atherosclerosis (6). Participation of individuals with FH in clinical trials has also been essential to the approval of every major LDL-C therapy currently used, including critical proof-of-principle studies of the efficacy of statins in lowering LDL-C (7) and, more recently, of proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors (8,9). Although landmark trials of statins in primary (e.g., West of Scotland Coronary Prevention Study) and secondary (e.g., Scandinavian Simvastatin Survival Study) prevention demonstrating substantial reductions in CAD death included FH patients (although exact numbers are unknown), such large trials would have been untenable in purely FH populations. Even if we could retrospectively determine with absolute certainty the FH status of every patient who has participated in a randomized trial of statins in primary prevention, we would likely have a sample size of <500 (10). One unanticipated effect of these early successes was that it became ethically forbidden to consider long-term, randomized, placebo-controlled outcome trials of LDL-lowering agents for primary or secondary prevention in FH patients. It is therefore ironic that the FH community, which has contributed so much toward our health and well-being, cannot benefit from the same level of clinical trial evidence we expect for statins (10,11).
A wealth of observational data suggests that statin therapy alters the natural history of FH both in children and in adults (12–14). However, more insight into the quantitative effects of LDL-C lowering on cardiovascular outcomes in FH patients is critical, especially given the convergence of several factors. First, recent data from U.S. and international registries describing the current landscape of FH suggest we have room to improve the care of FH patients. For instance, in a cross-sectional analysis of 1,295 adults with FH enrolled in the CASCADE FH (CAscade SCreening for Awareness and Detection of Familial Hypercholesterolemia) registry, we found prevalence of CAD 5 to 7 times higher than that of the age-matched general U.S. population and low LDL-C goal attainment (15). Similar findings were recently described in the SAFEHEART (Spanish Familial Hypercholesterolemia Cohort Study) registry (16). Second, there is a new realization that FH is much more common (and heterogeneous) than once thought. Classic estimates of FH put the prevalence at approximately 1 in 500 population. However, modern genetic studies have now coalesced around a prevalence much closer to 1 in 200 population, at least for populations of European ancestry (17–19). Although a recent report by Khera et al. (20) highlighted some phenotypic heterogeneity, the report also demonstrated that for any given LDL-C level, those individuals with LDLR mutations had worse outcomes. Third, the introduction of PCSK9 inhibitors and their indication for patients with FH has raised major questions about who should receive these powerful but expensive LDL-lowering therapies and what their cardiovascular benefits in FH may be.
It is in this context that the investigation of Besseling et al. (21), in this issue of the Journal, was launched. The primary goal was to estimate the cardiovascular beneficial effect of statins in FH patients in the absence of randomized clinical trial data. The investigators were able to undertake this journey only because of the infrastructure in the Netherlands (created in great part by the authors themselves) over the last ∼25 years to study and treat FH. This program included a broadly based screening program for FH that, crucially, included genetic testing; as a result, the authors have access to longitudinal data from 1994 to 2013 for more than 25,000 molecularly characterized FH patients (19). The authors were able to link medical record and mortality data to a subset (22%) of these patients. After certain exclusions, data for 1,041 statin users were available to be compared to data for 518 patients who had never used statins. The relatively high number of those who had never used is surprising given the Netherlands’ track record in the excellent care of FH patients (22). Using this approach, the authors found that FH patients who were statin users had a 44% relative risk reduction for the combined endpoint of CAD and all-cause mortality compared with that of those who had never used statins. This may actually be an underestimate of the potential benefit, given the relatively modest use of high-intensity statins (∼30%) in statin users. The overall estimate of the statin benefit translates to a number needed to treat (NNT) of 222 for 1 year of statin therapy to prevent a death in FH patients. This is remarkable and far outstrips the NNT for 1 year for primary prevention in non-FH patients (NNT: ∼500) (11); indeed, it is lower than the NNT for secondary prevention in non-FH patients (NNT: ∼350, extrapolating to a 1-year time frame for mortality) (10).
In a nonrandomized, real-world dataset such as this, where the baseline attributes of 2 groups are markedly different, biases and confounders in data can easily lead to false interpretations of treatment effects. One such bias highlighted by the authors was indication bias, exemplified here by the fact that the statin-users group was older and had more comorbidities than the statin-nonusers group. Without appropriate adjustment for severity of illness, a higher event rate among statin users could be falsely blamed on statins. The authors went to great lengths to adjust for measured covariates by using inverse probability-of-treatment weighting. In this method, patients are assigned weight based on their propensity for being given a statin as judged by the relevant cardiovascular risk covariates. Therefore, patients most highly influenced were those who had a high propensity for treatment based on all measured covariates but were not treated; and those who had a low propensity for treatment who were treated. Consider the following grossly simplified, hypothetical example.
Take statin-user patient A, who has a relatively benign cardiovascular profile (female, young, normal blood pressure), and assume the propensity for statin treatment is 0.1. The weight of this patient would be: [1 ÷ 0.1 = 10], which would be much greater than that for patient B, who has a propensity score of 0.8 and has received statin therapy in light of a higher risk profile (older, male, smoker; this patient’s weight would be: [1 ÷ 0.8 = 1.25]).
The actual data measured, including outcomes, were then reinterpreted (imputed) in light of these weights to reveal treatment effects that more closely account for measured confounders. With these analyses, several potential caveats should be noted. First, the authors were unable to link patients to medical record data for 78% of consenting individuals, which limited the sample size. The smaller the sample size, the greater chance that the limited number of outcomes could have disproportionate effects in the weighted analyses. For instance, in the unweighted analysis, there was 1 death per 543 person-years among the statin nonusers, whereas in the imputed analysis, there was 1 death per 174 person-years (vs. 1 death per 833 person-years among statin users). This might influence the stability of the models and potentially our confidence in the point estimates for benefit from therapy. Second, the authors assumed that there was a minimal effect of unobserved confounders, but what determines whether an FH patient is a statin user or nonuser? As a physician who cares for FH patients, I have been struck by the heterogeneity of responses when individuals are presented with a new diagnosis of FH. It is always a bit surprising when FH patients defer statin therapy even when educated about the risks and benefits. What explains the “failure” of the 518 nonusers to embrace potentially life-saving therapy? Is there something different about an FH patient who is not taking statins that is not captured by the model? For instance, how would the model account for patients having the “ostrich syndrome” (defined in Wikipedia as “when people prefer to ‘stick their heads in the sand,’ much as an ostrich does, rather than accept some uncomfortable facts”) or account for having a nihilistic physician who does not believe in statins? Either of these scenarios could carry unmeasured confounders resulting in an increased risk of death not mediated by their failure to take statins. Finally, as FH cases here were defined molecularly, it is possible that the magnitude of the effect would not be as large in FH cases defined phenotypically or with negative genetic testing.
Despite these caveats, these data are extremely important and offer unique insights. In the absence of being able to track down historic administrative data for a larger number of molecularly defined FH patients in the Netherlands, it is hard to imagine a larger study. Data were consistent with our understanding of the pathophysiology and natural history of FH and with prior observational data.
What are the implications of this study for care of patients with FH? The evidence is now overwhelming that FH is highly morbid yet underappreciated, underdiagnosed, and undertreated. The terrible irony, strongly emphasized by this study, is that, if identified early enough and treated appropriately, the morbidity and mortality of FH can be markedly if not totally ameliorated.
As a community, we must turn our attention to the critical task of assuring that all individuals with FH have the opportunity for optimal care (Figure 1). This needs to start with finding undiagnosed and untreated FH patients “by hook or by crook,” including using innovative approaches to screen for undiagnosed FH patients in health care systems, through flagging “FH-level” lipids on laboratory reports and through adherence to guideline-based screening recommendations. We must have the ability to appropriately classify FH patients by using specific International Classification of Diseases-10 codes. We need to continue to educate providers and patients about FH and combat misinformation. For instance, for all the negative press about statins, one of this study’s big conclusions is that, for FH patients, the most important potential “side effect” of statins is increased longevity, and for those who cannot take statins or have elevated residual risk, we need to use other treatment modalities and work to remove barriers to care. We must work with the knowledge that we never find an individual with FH, we only find families with FH. We know that systematic approaches to cascade screening can save lives yet, in many countries, are rarely implemented. If we can do these things, we will move from the current world where we refer to “patients with FH” to a future where we refer to “individuals with FH.”
Besseling et al. (21) bring to mind deep-sea explorers who have used a new and innovative approach to dive deeper than anyone before and have captured images of a feature long theorized but not directly observed. We should not be overly critical that the “images” obtained are slightly blurry due to the limits of the technology but rather congratulate them and redouble our efforts to get better instruments for another, more ambitious, expedition.
1. : "Familial hypercholesterolaemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society". Eur Heart J 2013; 34: 3478.
2. : "The agenda for familial hypercholesterolemia: a scientific statement from the American Heart Association". Circulation 2015; 132: 2167.
3. : "Coronary artery disease in 116 kindred with familial type II hyperlipoproteinemia". Circulation 1974; 49: 476.
4. "Risk of fatal coronary heart disease in familial hypercholesterolaemia. Scientific Steering Committee on behalf of the Simon Broome Register Group". BMJ 1991; 303: 893.
5. : "Familial hypercholesterolemia: identification of a defect in the regulation of 3-hydroxy-3-methylglutaryl coenzyme A reductase activity associated with overproduction of cholesterol". Proc Natl Acad Sci U S A 1973; 70: 2804.
6. : "A review on the diagnosis, natural history, and treatment of familial hypercholesterolaemia". Atherosclerosis 2003; 168: 1.
7. : "Influence of combined therapy with mevinolin and interruption of bile-acid reabsorption on low density lipoproteins in heterozygous familial hypercholesterolemia". Ann Intern Med 1985; 103: 339.
8. : "PCSK9 inhibition with evolocumab (AMG 145) in heterozygous familial hypercholesterolaemia (RUTHERFORD-2): a randomised, double-blind, placebo-controlled trial". Lancet 2015; 385: 331.
9. : "Efficacy and safety of alirocumab in patients with heterozygous familial hypercholesterolemia not adequately controlled with current lipid-lowering therapy: design and rationale of the ODYSSEY FH studies". Cardiovasc Drugs Ther 2014; 28: 281.
10. : "Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins". Lancet 2005; 366: 1267.
11. : "Statins for the primary prevention of cardiovascular disease". Cochrane Database Syst Rev 2013; 1: CD004816.
12. : "Efficacy of statins in familial hypercholesterolaemia: a long term cohort study". BMJ 2008; 337: a2423.
13. "Mortality in treated heterozygous familial hypercholesterolaemia: implications for clinical management. Scientific Steering Committee on behalf of the Simon Broome Register Group". Atherosclerosis 1999; 142: 105.
14. : "Familial hypercholesterolaemia in children and adolescents: gaining decades of life by optimizing detection and treatment". Eur Heart J 2015; 36: 2425.
15. : "Treatment gaps in adults with heterozygous familial hypercholesterolemia in the united states: data from the CASCADE-FH registry". Circ Cardiovasc Genet 2016 Mar 24; . [E-pub ahead of print].
16. : "Attainment of LDL-cholesterol treatment goals in patients with familial hypercholesterolemia: 5-year SAFEHEART registry follow-up". J Am Coll Cardiol 2016; 67: 1278.
17. : "Mutations causative of familial hypercholesterolaemia: screening of 98,098 individuals from the Copenhagen general population study estimated a prevalence of 1 in 217". Eur Heart J 2016; 37: 1384.
18. : "Exome sequencing identifies rare LDLR and APOA5 alleles conferring risk for myocardial infarction". Nature 2015; 518: 102.
19. : "Homozygous autosomal dominant hypercholesterolaemia in the Netherlands: prevalence, genotype-phenotype relationship, and clinical outcome". Eur Heart J 2015; 36: 560.
20. : "Diagnostic yield of sequencing familial hypercholesterolemia genes in patients with severe hypercholesterolemia". J Am Coll Cardiol 2016; 67: 2578.
21. : "Statins in familial hypercholesterolemia: consequences for coronary artery disease and all-cause mortality". J Am Coll Cardiol 2016; 68: 252.
22. : "Review of first 5 years of screening for familial hypercholesterolaemia in the Netherlands". Lancet 2001; 357: 165.
Dr. Knowles' work on FH is supported by an American Heart Association National Innovative Research grant 15IRG222930034 and by an Amgen Full Potential Initiative grant (paid to institution not to individual).