Cholesterol treatment guidelines
Evidence vs. emminence
As long ago as 2005 an audit of just 18 patients, admitted to an emergency department, reported that there were 3,679 pages of clinical guidelines associated with their conditions. Imagine what it is like now.
What started out as support for evidence-based medicine, and as a means for clinicians to keep abreast of the latest scientific knowledge, has devolved into something increasingly conflicted, suspect and unwieldy. There are concerns with conflicts of interest among panellists (financial and professional), an often unsatisfactory evidence-base that can be influenced by industry, and real-world applicability. It seems any organisation can release guidelines and attempts by authorities, such as the respected US Institute of Medicine (IOM), to establish standards for the drawing up and monitoring of guidelines have largely failed. Furthermore, the implementation of guidelines can be mandated (by hospitals for example), form the basis for funding treatment, and be central to medico-legal issues. There is even suggestion that guideline compliance be an assessment of clinician performance in the US.
However, a patient has no obligation to be treated according to any guideline, and a deeper understanding of the development of certain guidelines may offer a perspective for decision making. This applies particularly to preventative medicine guidelines.
Herein, as part of the statin series of posts, I will delve into the US cardiovascular practice guidelines as they pertain to cholesterol and statins. These have been pivotal in maintaining the central role of statins in the prevention or treatment of cardiovascular disease (CVD).
In the beginning: the NHLBI
The 2016 American College of Cardiologists and American Heart Association (ACC/AHA) report was commissioned by the National Heart, Lung and Brain Institute (NHLBI), which is a US government organisation within the National Institutes of Health (NIH). Before this, the NHLBI issued the guidelines itself, through a National Cholesterol Education Program (NCEP). There were three sets of guidelines issued by NCEP before responsibility was handed over to ACC/AHA. It is revealing to rummage through some of this history.
The first (1988) NCEP report
The first guidelines were published in 1988: “Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults”.
My italics. The report has just the one focus — blood cholesterol.
The prevailing hypothesis leading up to these guidelines was that consuming animal fat increased blood cholesterol that in turn increased the risk of CVD. This is the diet-heart hypothesis that continues to undermine public health today (and that is still an hypothesis — it has never been established to be true). Consequently, there was no doubt at that time that blood cholesterol needed to be kept low, and the NCEP report was about that.
Which means that the provenance of contemporary cardiovascular guidelines is firmly set in the management of blood cholesterol.
A digression on cholesterol units:
Because US guidelines are being reviewed here, the units used will be mg/dl (milligrams per deciliter). The rest of the world uses mmol/L (millimoles per litre). I will follow the convention that the US spelling for liter is abbreviated by lower case ‘l’, whereas the litre uses uppercase ‘L’. The unit of mg/dl measures concentration by weight per volume (density), whereas mmol/L gives a measure of the number of cholesterol molecules per volume (a mole is a very large number). In the case of cholesterol, to convert mg/dl to mmol/L, divide by 40 (strictly 39). This number differs for different molecules according to their molecular weight (for example, it is ~88 for triglycerides, because triglycerides are heavier molecules than cholesterol). I will provide a conversion for the first appearance of any value.
The 1988 NCEP report, continued…
The guideline recommended measuring total cholesterol (TC) in everyone. If it was less than 200 mg/dl (5 mmol/L), give general dietary advice and retest in 5 years. If TC was > 240 mg/dl (6 mmol/L), measure low-density lipoprotein cholesterol (LDL-C) and consider drug therapy on that basis.
TC was used as a screening tool because, at the time, it was an easier and less costly test than LDL-C. However, clinical recommendations were made on the basis of LDL-C concentration.
The TC threshold (240) seems to have been derived from the 75th percentile of the estimated normal distribution of TC in the US population (using data from the 3rd National Health and Nutrition Examination Survey — NHANES III). That is, according to NHANES III, 75% of people surveyed had a TC < 240 mg/dl, and this seems to have been rather arbitrarily chosen as a cut-off value. It could also have been the 80th percentile (it is not clear), but either way, the value was not chosen on clinical grounds. The cut-off meant that 20–25% of the adult population could be considered for cholesterol-lowering drug therapy. I have seen no justification for the 200 mg/dl lower cut-off.
People with TC between these limits (200–240 mg/ dl), were divided into low-risk (without CVD and no more than one other risk factor — e.g. male, smoking, hypertension etc), and high-risk (with CVD and/or multiple risk factors). They were given explicit dietary advice and referred for measurement of LDL-C.
It is worth reminding ourselves that even today, LDL-C is not actually measured by standard blood tests — it is estimated from TC, high-density lipoprotein cholesterol (HDL-C) and triglycerides (TGs), according to the Friedewald equation. Thus, the centrepiece of cholesterol and cardiovascular guidelines is, and always has been, an indirectly determined metric. The Friedewald equation can overestimate LDL-C if TG is low, as is common in ketosis (perhaps ~1 mmol/L).
For the low-risk group, the guidelines recommended that drug treatment commence if LDL-C was greater than 190 mg/dl (4.8 mmol/L), aiming to achieve a target of less than 160 mg/dl (4 mmol/L). For the high-risk group, the threshold was set to 160 mg/dl, with a target of 130 mg/dl (3.2 mmol/L).
The document is unclear, and even contradictory, as to how these thresholds and targets were determined — 190 mg/dl may have been the LDL-C level in people from NHANES III with a TC of 240 mg/dl, but I cannot be certain. For this crucial clinical directive, the authors of the guidelines remained coy and evasive.
Notice that 190, 160 and 130 are going down in steps of 30. In later guidelines, we will see two further steps in this series — 100 and 70 mg/dl (2.5 and 1.8 mmol/L). The thinking underpinning the value of 30 remains arcane.
The second and third NCEP guidelines
These came out in 1993 and 2001. The three sets of NCEP guidelines have become known as the Adult Treatment Panels (ATP-I, II and III). For ATP-II there was a greater emphasis on HDL-C, and three levels of risk were grouped: no CVD and low-risk; no CVD but high risk; with CVD. LDL-C treatment threshold for the ‘with CVD’ group was set to 130 mg/dl (target 100 mg/dl). ATP-III recommended assessing risk with a global risk calculator (Framingham Risk Equation) for the first time, and a complete lipoprotein profile as the preferred initial test (rather than TC). The LDL-C thresholds and targets were kept the same.
In 2004, an update to ATP-III was issued that lowered treatment threshold for people with CVD to 100 mg/dl with a target of below 100 mg/dl and optionally below 70 mg/dl. Lower treatment thresholds were optionally introduced for other risk groups. Overall, the trend was towards increased medicalisation.
Throughout the ATP series of guidelines, the core concepts were preserved, probably because the panellists were too. The ATP-III panel membership was identical to that of ATP-II, and almost half of the panellists on ATP-II had come from ATP-I. This was not an environment for change or critique. With increasing eminence over time, the panellists themselves were locked into the narrative they had created. By the time ACC/AHA guidelines came along in 2013, LDL-C had been at the centre of cardiovascular practice guidelines for 25 years and the future had been set.
An abortive ATP-IV
The NHLBI, through NCEP, now embarked on ATP-IV, however, these guidelines never materialised. The story gets a bit murky around here. Drug industry conflicts of interest among panellists were beginning to raise alarms, the league of panellists may have been causing concern, and to make matters worse, another NHLBI guideline (on blood pressure) recommended reducing medication in some people (the treatment threshold for systolic blood pressure was increased from 140 to 150 mmHg in people over 60 years of age). This deprived industry of sales, and clinicians the option to treat. Powerful forces were aroused.
For these, and probably other reasons, the NHLBI had a rethink and decided to get out of the clinical guidelines business. They handed responsibility for hypertension guidelines and cholesterol guidelines to the ACC/AHA to deal with.
Both of these organisations receive funding from, and have partnerships with, the pharmaceutical industry. Perhaps unsurprisingly, therefore, the ACC/AHA guidelines on hypertension dismissed the NHLBI’s 150 mmHg recommendation, and introduced a new and lower one: 130–140 mmHg was designated as a pre-hypertension, thus increasing the opportunity for medication.
The ACC/AHA Cholesterol Guidelines
These were first published in 2013: “ACC/AHA Guidelines on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults”
The guidelines are about treatment, consistent with the clinical background of the ACC/AHA. Statins are the treatment of choice. Lifestyle change, such as diet, is given secondary importance (although there is a supplement on lifestyle). The guidelines recommend estimating CVD risk using a new calculator — the Pooled Cohort Equation (PCE).
LDL-C concentration, combined with the PCE, determine the initiation of statin treatment. There are no guidelines for ceasing statin treatment once commenced. Notably, for the first time, there are no LDL-C treatment targets either, and aiming for the lowest achievable LDL-C has been abandoned as well. The guidelines only suggest measuring LDL-C after therapy to test for patient compliance. For therapy, the guidelines offer either a high-dose statin (that they expect to reduce LDL-C on average by 50%) or a moderate-dose statin (reducing LDL-C by ~30%). The guidelines explicitly give these doses for each of the mainstream statins. There is no recommendation for the use of other cholesterol-lowering medications, either on their own or in conjunction with statins.
The ACC, on its own, issued an opinion statement in 2016 that restored LDL-C treatment targets and made recommendations for non-statin options.
The lifetime treatment of healthy people without a history of CVD is a boon for statin sales. Furthermore, the decision to initiate treatment in healthy individuals demands the strongest possible evidence-base. So, I will dig a bit deeper here.
The guidelines identify two primary-prevention candidate groups. The first consists of individuals with LDL-C greater than 190 mg/dl. Nothing else matters and therapy is advised: “The guideline recognizes that adults ≥21 years of age with primary, severe elevations of LDL–C (≥190 mg/dL) have a high lifetime risk for ASCVD [atherosclerotic CVD] events. Thus, at age 21, these individuals should receive statin therapy.” The recommendation is a high-dose statin.
Bearing in mind that there are no guidelines for stopping therapy, this means that some individuals would be sentenced to decades on statins. The evidence-base for the potential adverse effects of this recommendation is nil. Remember: “First, do no harm.”
The second primary-care candidates are 40–75 years of age and have LDL-C in the range 70–190 mg/dl. For these people, the PCE for CVD risk should be applied. If the PCE reports a risk greater than 7.5%, then treatment should be commenced (high-moderate dose statin).
Treatment can still be considered in this second group if: PCE risk is greater than 5%; LDL-C >160 mg/dl; the lifetime risk, as opposed to 10-year risk, is high or; there are other risk factors not incorporated in the PCE such as family history, cardiac imaging data or inflammation markers. The panel is not giving up on statin treatment easily.
The 190 mg/dl upper-threshold has propagated along the guidelines since ATP-I. The ACC/AHA guidelines do not revisit this number, or offer any explanatory comments.
The 70 mg/dl lower-threshold appeared in the ATP-III 2004 update for people with CVD, however, now it is part of primary prevention. There is a circularity underlying this: drug companies fund clinical trials; trials recruit people over as wide a range as practical, e.g. 70–190 mg/dl, to maximise the market; the trials show an overall benefit (however small); guideline panellists use the trials to set their guidelines, and 70–190 mg/dl becomes the recommended range for treatment. This is one way drug companies can influence guidelines.
Conflicts of Interest (COIs)
The guidelines are statin-treatment instructions — a prescription for prescribing. No matter that the ACC and AHA have relationships with, and receive funding from, statin manufacturers.
The ACC doesn’t seem bothered with their (or an individual’s) COIs with industry. They call them relationships with industry (RWI) rather than conflicts, and have this to say: “Although the ACC recognizes the need to disclose relationships for the purpose of transparency, it does not believe that these “relationships” are necessarily conflicts. When all relationships are disclosed with the appropriate detail regarding category and amount, and managed appropriately, the ACC believes that potential bias (conflict) can be avoided …”. Notably, the ACC website (and even their Annual Report) does not specify the amounts they receive in donation from industry.
It also depends on whether potential conflicts are disclosed in full by individuals, it can be a value judgement. I’ve not seen any penalty, except professional reputation, for non-disclosure.
The ACC/AHA panellists were asked to provide their RWIs, but only for the period over which the guidelines were developed (2008–13). About half of the panellists declared RWIs. There was one panellist who was also a member of ATP-III, and who provided this financial disclosure at that time: “… grants and/or research support from Pfizer, Merck, Parke-Davis, and AstraZeneca … consultant for Kos Pharmaceuticals, Abbott, and Merck … honoraria from Abbott, Bristol-Myers Squibb, Novartis, Merck, Kos Pharmaceuticals, Parke-Davis, Pfizer, and DuPont”. His disclosure for the ACC/AHA report was: “None”.
Are previous COIs still conflicts, and under what circumstances does their influence expire? What about future COIs? Does membership of a panel whos recommendations support drug company profitability also increase the likelihood of future funding from drug companies?
Then, there are professional conflicts. A panel member who has spent his professional life pursuing the cholesterol hypothesis would be conflicted by any competing hypothesis. And, for that matter, would be more likely to be invited to join a cholesterol guideline panel.
By and large, none of this is ‘manageable’ as claimed by the ACC. Instead, it should be understood that drawing up clinical guidelines is a human endeavour, and subject to human nature and failings. It should not be taken to be a purely objective and scientific process. Hence, the integrity of guidelines, not only their rigour, requires evaluation.
The strength of recommendations
As is customary, the ACC/AHA guidelines provided a two-tiered grade for each of their recommendations.
- The first tier is the grade of scientific and clinical evidence: A, B and C corresponding to high, marginal and low.
- The second tier is the strength of the panel’s recommendation: I, II and III corresponding to treat, perhaps treat and don’t treat.
The ACC/AHA grading for both primary-prevention groups was B-I.
In greater detail: ‘B’ means “Limited populations evaluated, a single randomised controlled trial (RCT) or non-randomised studies”; and ‘I’ means “Treatment should be administered — useful and effective.”
These are therefore strong recommendations based on scant data. In fact, there are no RCTs to support the LDL-C > 190 mg/dl threshold. The ACC/AHA explained it like this, in a footnote to Table 4.: “No RCTs included only individuals with LDL–C ≥190 mg/dL. However, many trials did include individuals with LDL–C ≥190 mg/dL and all of these trials consistently demonstrated a reduction in ASCVD events. Therefore, individuals with primary LDL–C >190 mg/dL should be treated with statin therapy.”
This is a specious and desperate argument. It could be applied to arbitrary LDL-C levels, e.g 160 mg/dl, and with more confidence (because there will be more people > 160 mg/dl). Worse, it is a clear admission that there have been no RCTs comparing a group >190 mg/dl with a group <190 mg/dl, on which they could base this firm treatment recommendation.
Later, the guidelines contradict themselves when they admit: ”… although in most clinical trials, individuals with LDL–C ≥190 mg/dL were not included due to their need for treatment …”. It comes down to a difference between ‘many trials’ and ‘most trials’.
Recommendations to treat LDL–C > 190 mg/dl have been in place since 1988. Therefore, it is now unethical to recruit people over this limit into a RCT in which some may receive a placebo. The guidelines have stymied themselves.
None of the risk calculators I have examined includes LDL-C as a risk factor for CVD. This applies to the US (three different calculators, including the PCE and Framingham), United Kingdom (two calculators), European Union, Canada, New Zealand or Australia. The guidelines started out all about lowering cholesterol to reduce CVD risk. It must have been awkward when risk equations uniformly discarded LDL-C as a risk. I have never seen this discord addressed.
Guidelines are not as objective as they may seem, and more often amount to the expert opinion of panellists, often chosen according to their contributions to the prevailing hypothesis. They become expert, because they are on an Expert Panel. A 2009 analysis of all ACC/AHA guidelines found only 19% of class ‘I’ (firm) recommendations had grade ‘A’ (strong) evidence to support them, and this is increasingly the trend.
For 30 years now, the LDL-C hypothesis has commanded thinking about CVD risk and cause. This has had the effect of marginalising competing hypotheses, such as insulin resistance, chronic inflammation or even viral causes.
Which brings me to the big question not covered here — just what is the evidence for the cholesterol hypothesis? How do we know that cholesterol causes heart disease, and therefore that we should be concerned about it and lower it? It is common to attribute causation to association, but vitally important not to get this wrong (otherwise we might think we can reduce the risk of rain by reducing the number of umbrellas). The evidence surrounding the cholesterol hypothesis will need to be a separate post.
Disclaimer: I am not a medical doctor. Nothing herein is, nor should be taken to be, medical advice.
Links to Cholesterol Guidelines:
ATP-I: Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
ATP-II: Summary of the Second Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel II)
ATP-III: Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III)
ACC/AHA: 2013 ACC/AHA Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults
ACC 2016 statement: ACC Expert Consensus Decision Pathway on the Role of Non-Statin Therapies for LDL-Cholesterol Lowering in the Management of Atherosclerotic Cardiovascular Disease Risk: A Report of the American College of Cardiology Task Force on Clinical Expert Consensus Documents