May 3rd, 2010
(Mis)Information on Fenofibrate in the Clinic
Allan Brett, MD
Several days ago, I — a general internist — saw a 70 year old diabetic woman with coronary disease who was stented 3 years ago and has been asymptomatic since then. During our office visit, I noted that a nurse practitioner who does lipid management in her cardiologist’s practice had added fenofibrate to her statin just a week ago. The relevant note stated that: “on the basis of her Berkeley lipid panel, will add Tricor.” Her most recent fasting lipids, on 40 mg of atorvastatin daily, were as follows: Total cholesterol 119, LDL 37 (yes, 37!!), HDL 38, and triglycerides 220.
For this statin-treated patient, the study most relevant to adding fenofibrate (Tricor) is the recently published ACCORD Lipid trial (see here for the article and here for a CaridoExchange interview with the lead author). I printed out the article for the patient, and we carefully reviewed the results together — no overall benefit with fenofibrate; significantly worse outcomes with fenofibrate than with placebo among women; and a non-significant trend toward better outcomes with fenofibrate in the subgroup of patients with TG>204 and HDL<34. Although I gave the patient the option to discuss the data further with her cardiologist, she opted to stop the drug at our visit. Understandably, she expressed confusion as to why the drug was prescribed when randomized trial data do not support its use in patients like her.
That very same day, I received an unsolicited e-mail from an organization called “Residual Risk Reduction Initiative” (R3i). Clicking on a link to R3i’s website, I found commentary on the ACCORD Lipid trial (including a video interview with a study investigator that no longer appears to be available) that highlighted the non-significant subgroup analysis and downplayed the overall results of the trial. I was unable to find information on who funds R3i.
Although neither ACCORD Lipid nor the previously published FIELD study showed overall benefit for fenofibrate therapy in diabetic patients, fenofibrate enthusiasts (some of whom have financial conflicts of interest) point to subgroup analyses, limitations in the trials, and favorable changes in surrogate endpoints in order to rationalize expanding use of this drug. But the simple fact is that no decisive evidence exists to support use of fenofibrate — which costs about $1800 per year — in patients like the one presented above. To justify the addition of fenofibrate to statin therapy in diabetic patients, we need a trial with clinical endpoints that focuses squarely on patients with substantially high triglycerides and low HDL levels.
I welcome commentary on two other issues raised by this case:
What hard evidence exists for using costly Berkeley Heart panels to guide treatment?
How should general internists like me communicate with cardiologists who institute non-evidence-based treatments for our shared patients?
8 Responses to “(Mis)Information on Fenofibrate in the Clinic”
NEJM Journal Watch — Recent Cardiology Articles- Strategies and Criteria for the Diagnosis and Management of Myocarditis
- Sex-Based Disparity in CABG Outcomes: Are We Moving Toward Parity?
- Can Genetic and Social Risk Scores Improve Cardiac Risk Prediction?
- Extensive Ablation vs. Pulmonary Vein Isolation in Persistent Atrial Fibrillation
- Can a Novel Stent Design Improve Long-Term Outcomes of Percutaneous Coronary Intervention?
Utility of Berkeley, evidenced based treatment
The first question to ask is whether we are happy with the current state of evidence based medicine sufficiently to limit our practice to treatments that have be subject to double blind randomized prospective trials. Keep in mind, although a laudable goal to only practice evidence based medicine, a very small minority of medicine currently practiced meets this definition.
A 70 year old diabetic treated with a statin alone is still at moderately high risk for a coronary event as the statin only reduces her risk by at best 40%. Assuming a baseline risk of 30% over the next 10 years, add a statin and her risk is still 18% over the next 10 years, a bit higher than I feel comfortable leaving un-addressed.
By current standards, evidence based medicine begins and ends with a statin. Can we do better?
Berkeley adds Apo-B, and LP(a), both of which are significant measures of risk not found on a conventional lipid assay. It also will look at other atherogenic particles as well as evaluating likely response to high dose statin with KIF-6 and plaque stability with LP-PLA-2. All of these add value and can help direct therapy. Evidence based? Not by current standards Of value? I think so.
I would need to see the Berkeley results before passing judgement on the use of a fenofibrate. I would likely use niacin and/or high dose fish oil derived omega-3 fatty acids before considering a fenofibrate in a subject such as this. There is reason to believe that fenofibrates reduce microvascular complications of diabetes and might be a reason to consider that class of medicine in this patient.
Communication with our cardiologist colleagues is essential. My bigger beef with how cardiologists treat my patients is when they put in a stent to open up a minimally obstructive vessel in the non- STEMI setting, and have the patient think they just saved their lives. In truth, the stent did nothing to reduce the person’s MI risk or risk of death, exposed them to unnecessary risk, and mandate that they use dual antiplatelet treatment for an extended period of time. I think we should pick our battles where they count and the question of inappropriate stenting is a much bigger concern than the use of a medication to improve lipids in the absence of “evidence based guidelines”.
Competing Interests: I use serial calcium imaging to determine adequacy of medical therapy in primary and secondary prevention of heart attacks. This is not “evidence based practice” but it results in a near elimination of heart attacks and coronary vascular death in my practice.
Therapy guided by plaque progression
To my mind the question is moot. The patient in question has CAD – if they are a plaque progressor (which is something I am going to find out in 6 months from baseline), I am going to do everything possible to reduce their risk, going from most evidence-based approaches down to less evidence-based approaches (recall that in FIELD, fenofibrate did prevent MI; and in VA-HIT, gemfibrozil reduced cardiovascular death, stroke, and MI). Just treating purely to lipid numbers is bogus. What we need in this case scenario is some idea of the biology of the disease in this scenario – do they have a lot of plaque? is that plaque vulnerable? are they jumping plaque grades over time? Then talk to me about +/- fenofibrate.
Show Me The Money
Allen, your experience is not unique. The Berkeley test often results in patients being prescribed therapy (supplements and/or medications) that are unproven, expensive and potentially — or actually — harmful.
Doing the test is easy….knowing what to do (or avoid) with the information is not.
The Berkeley Heart panel’s been around since 1996…..and so far no evidence that the “personalized disease management” has improved cardiovascular outcome. This doesn’t pass the mustard in the era of “comparitive effectiveness.”
So here is the rub…
The philosophy of more information and more medication is better has been discredited many times – and yet persists – perhaps because we are uncomfortable with uncertainty and believe that we each have a way, yet to be proven, that can help our patients. The problem is that these strategies are unproven and expensive and can lead to more unproven tests and procedures – and patients are often unaware of the lack of evidence to support their use. We need not be paralyzed by lack of evidence – but patients who are treated with strategies that have yet to prove their value need to have that information disclosed to them. For Berkeley to be worthwhile we need to show that the information provided is worthwhile — and meaningful to future decisions — and will ultimately lead to better outcomes by patients. As for fenofibrate – it has failed in two trials — should we continue to assume it is beneficial until we have 10 such trials? 20?
uncertainty about the value of tests
This is a great discussion. Of all the tests we routinely use in medicine, I would say that less than 10% have been proven through the gold standard of an outcome-driven randomized trial. Many of them just seem to make “sense” – chest x-ray for patients presenting with dyspnea, EKG for others with chest pain, and so forth. It seems that tests that were established before the advent of the evidence-based medicine movement are now so firmly entrenched that they won’t ever be evaluated by this gold standard. While I agree there should not be a lower standard for diagnostic/prognostic tests than new therapies, we have to keep in mind common sense and pathophysiology of the disease process in front of us as well as the evidence base. In terms of Framingham risk factors and the Framingham score, the AU-ROC is 0.78, meaning over- or under-estimation of risk in nearly 1 in every 4 patients. That is our gold standard for comparing new tests in cardiovascular prevention. And there is no RCT comparing Framingham driven care to anything else – usual care, guideline-based care (which is now based on Framingham!), or newer tests. This is lamentable and perplexing. I can therefore understand the enthusiasm of those who use Berkley and CAC to stratify risk in their patients.
well…
Good points Dan – but there is a difference in using a test for diagnosis [you have a patient with chest pain and you use an ECG – hardly need a trial of the utility of the test] and for screening. We are still confronting questions of who should get a routine CXR (should anyone?) or a routine ECG (should anyone?) — particularly in low risk populations. That is the situation with the new tests too – they are expensive (certainly in aggregate) and can lead to more downstream tests and procedures of uncertain benefit. What makes senses does not always pan out as a superior strategy — we have lots of examples of that. So where should the burden of proof be – should we assume benefit for something that seems sensible — or do we need some evidence that the strategy actually gets better results?
I agree that clinical trial evidence should trump wisdom
I agree that clinical trial evidence should trump wisdom. PS, back to your original question, “for fenofibrate – it has failed in two trials — should we continue to assume it is beneficial until we have 10 such trials? 20?” The Lancet published today a favourable article on fibrate use, abstract below: Effects of fibrates on cardiovascular outcomes: a systematic review and meta-analysis. Background: Several clinical trials have reported inconsistent findings for the effect of fibrates on cardiovascular risk. We undertook a systematic review and meta-analysis to investigate the effects of fibrates on major clinical outcomes. Methods: We systematically searched Medline, Embase, and the Cochrane Library for trials published between 1950 and March, 2010. We included prospective randomised controlled trials assessing the effects of fibrates on cardiovascular outcomes compared with placebo. Summary estimates of relative risk (RR) reductions were calculated with a random effects model. Outcomes analysed were major cardiovascular events, coronary events, stroke, heart failure, coronary revascularisation, all-cause mortality, cardiovascular death, non-vascular death, sudden death, new onset albuminuria, and drug-related adverse events. Findings: We identified 18 trials providing data for 45 058 participants, including 2870 major cardiovascular events, 4552 coronary events, and 3880 deaths. Fibrate therapy produced a 10% RR reduction (95% CI 0–18) for major cardiovascular events (p=0·048) and a 13% RR reduction (7–19) for coronary events (p<0·0001), but had no benefit on stroke (−3%, −16 to 9; p=0·69). We noted no effect of fibrate therapy on the risk of all-cause mortality (0%, −8 to 7; p=0·92), cardiovascular mortality (3%, −7 to 12; p=0·59), sudden death (11%, −6 to 26; p=0·19), or non-vascular mortality (−10%, −21 to 0·5; p=0·063). Fibrates reduced the risk of albuminuria progression by 14% (2–25; p=0·028). Serious drug-related adverse events were not significantly increased by fibrates (17 413 participants, 225 events; RR 1·21, 0·91–1·61; p=0·19), although increases in serum creatinine concentrations were common (1·99, 1·46–2·70; p<0·0001). Interpretation: Fibrates can reduce the risk of major cardiovascular events predominantly by prevention of coronary events, and might have a role in individuals at high risk of cardiovascular events and in those with combined dyslipidaemia.
Response from the author of “(Mis)Information on Fenofibrate in the Clinic”
I appreciate the thoughtful responses to my presentation. One theme in the responses is “what should we do when RCT-level evidence doesn’t exist?” For my patient, in fact we do have a trial focused squarely on patients like her – diabetic patients already on statins – and the outcome of the trial was negative. Dan mentions the recent Lancet meta-analysis, but as the authors note in the fourth paragraph of their discussion section, “The ACCORD trial was the only study in which background statin therapy was routinely given.”
I would like to inject cost containment more explicitly into this debate. We have generally been taught “always put your patient’s best interests first,” and “don’t perform ad hoc rationing at the bedside.” I agree with these principles, but they imply that the proposed intervention confers some reasonable chance of benefit. It is professionally irresponsible for us to complain, on the one hand, that government or insurers are intruding on clinical practice and cutting reimbursement while, on the other hand, we practice as if a bottomless pit of money was available for unproven therapies or tests. The debate on cost containment is not some abstraction, divorced from daily clinical decisions; rather, it is centrally ABOUT our daily clinical decisions. Ordering 2 Berkeley panels, prescribing fenofibrate, and seeing the patient several extra times to monitor her therapy, will cost more than $3000 annually. That is nearly half the annual per capita Medicare expenditure in my state (see http://content.nejm.org/cgi/content/full/360/9/849). Does it make sense to spend nearly half of her “allotment” on interventions unlikely to benefit her?
Let me propose 3 “rules of thumb” that would create an ethical imperative NOT to order a costly intervention:
(a) There are no relevant clinical trials, and no professional consensus supporting the intervention, or
(b) There are no relevant clinical trials, the intervention is based on surrogate endpoints, and clinical trials of related interventions (in this case, to lower cardiac risk) have shown that we cannot trust surrogate endpoints (e.g, intensive glucose-lowering therapy, rosiglitazone, hormone therapy, torcetrapib), or
(c) There is at least one well-done clinical trial showing no benefit
Until U.S. physicians begin to practice according to principles something like these, there is little chance of creating a health program that expands access and contains costs.