College of American Pathologists
Printable Version

  Cholesterol—rethinking set practices


CAP Today




March 2010
Feature Story

Karen Lusky

LDL and HDL cholesterol still hold sway in predicting the risk of vascular disease, but the lipids picture is changing in ways that could lead to better testing and treatment.

For one, the evidence-based debates are heating up about the best way to measure the risk associated with elevated low-density lipoprotein levels and other atherogenic particles. Questions continue about the interplay of inflammation with LDL-C in risk prediction and treatment outcomes. And high-density lipoprotein cholesterol’s do-gooder status has undergone a bit of a shift as research shows that more HDL-C isn’t always helpful and might even be harmful, which means drugmakers have to be cautious about how their HDL-raising drugs work.

A meta-analysis published in the Nov. 11, 2009 issue of the Journal of the American Medical Association found that LDL cholesterol calculated from fasting samples used to measure triglycerides, total cholesterol, and HDL cholesterol is no more effective than using the non-HDL cholesterol level to predict the risk of vascular disease (Di Angel­antonio E, et al. JAMA. 2009;302:1993–2000). That’s good news because testing for non-HDL cholesterol doesn’t require a fasting sample. All that’s needed is simple arithmetic: Subtract HDL cholesterol from total cholesterol without regard to triglyceride. (Non-HDL-C is a measure that includes low-, intermediate-, and very-low-density lipoprotein cholesterol.)

The meta-analysis, which encompassed more than 300,000 patients and 10,000 major adverse events, also showed that triglycerides do not provide any independent information about coronary heart disease risk, says Russell Tracy, PhD, director of the laboratory for clinical biochemistry research at the University of Vermont. His laboratory participated in one of the multicenter studies pulled together in the JAMA report, issued by the Emerging Risk Factors Collaboration.

The data showed, too, that apolipoprotein B (apo B), a proxy for the number of LDL particles, is as effective as either LDL cholesterol or non-HDL cholesterol in assessing vascular disease risk.

The authors conclude that discussions about whether to measure cholesterol levels or apolipoproteins “should hinge more on practical considerations (eg, cost, availability, and standardization of assays) than on major differences in strength of epidemiological associations.” And, as Dr. Tracy pointed out in a CAP TODAY interview, “Most labs are set up to do cholesterol and HDL with automated instruments.”

The findings of the study, says Gary L. Horowitz, MD, are in line with what he presented last fall in a session at CAP ’09: a clinical case showing the value of using non-HDL cholesterol in lieu of calculating LDL cholesterol. Doing so avoids the danger of a laboratory unknowingly using a nonfasting sample, which will produce an LDL-C that is falsely low, giving a “false sense of reassurance,” warns Dr. Horowitz. He is associate professor of pathology at Harvard Medical School, director of clinical chemistry at Beth Israel Deaconess Medical Center, Boston, and chair of the CAP Chemistry Resource Committee.

Daniel Hoefner, PhD, section head of clinical chemistry at Marshfield (Wis.) Clinic, says most phlebotomists and laboratorians suspect that a significant portion of those who tell their doctors they are fasting are in fact not fasting. Even worse, “We have heard that clinicians will ask some patients to tell the phlebotomist they have been fasting even when they haven’t, because they know the lab may refuse to do the lipid panel on a nonfasting sample,” Dr. Hoefner says. Whether that’s because the clinician is looking at the non-HDL-C value Dr. Hoefner doesn’t know.

At CAP ’09, Dr. Horowitz shared the real-world clinical case to illustrate the edge that non-HDL cholesterol has over a calculated LDL cholesterol result. The case involved Mr. Smith, a nonsmoker with a family history of early myocardial infarction. The results of his physical examination, including his blood pressure, were normal.

Mr. Smith’s primary care physician ordered an initial nonfasting sample for lipidsprobably to prevent the patient from having to return to the office, Dr. Horowitz surmises. Mr. Smith’s calculated LDL cholesterol result based on the nonfasting sample was 33 mg/dL lower than the one on a fasting sample a week later (142 mg/dL versus 175 mg/dL), putting him in different risk categories. Yet the patient’s non-HDL cholesterol results were consistent in the fasting and nonfasting samples (192 mg/dL and 200 mg/dL, putting him unequivocally in the high-risk category).

To explain why a nonfasting sample skews LDL cholesterol results, Dr. Horowitz discussed the Friedewald equation, used to calculate LDL-C. The equation in its simplified form is as follows:

LDL-C = total cholesterol − HDL cholesterol − TG/5.

Total cholesterol is composed of HDL-C, LDL-C, and VLDL-C (very-low-density lipoprotein cholesterol). The triglycerides in the equation are a stand-in for VLDL-C, so VLDL-C is subtracted from total cholesterol. The total cholesterol of some patients will also include intermediate-density lipoprotein (IDL) and lipoprotein (a), though most people have negligible amounts, Dr. Horowitz explained.

The Friedewald equation “works remarkably well,” he noted, except for nonfasting samples, because after a meal, a lot of triglycerides get packaged in chylomicrons circulating in the blood. And the chylomicron triglycerides “will be ascribed to VLDL-C, thus causing overestimation of VLDL-C and an underestimation of LDL,” making the LDL-C value appear to be “falsely healthy,” he said. And even when triglycerides are less than 400 mg/dL (Mr. Smith’s nonfasting TG was 250), there’s “no guarantee the sample is fasting.”

“There’s another little secret” about using the Friedewald equation, Dr. Horowitz warns. “If you have IDL particles—and you don’t know if a patient does—the equation isn’t accurate.”

Calculating non-HDL-C by simply subtracting HDL-C from total cholesterol sidesteps all of the aforementioned problems. And non-HDL-C “takes into account IDL-C and Lp(a)c, both of which are probably atherogenic,” Dr. Horowitz said. Thus, in his view, labs should always report non-HDL cholesterol. “The whole concept of good and bad cholesterol is correct. We just want to modify the bad to be non-HDL.”

The therapeutic goal for non-HDL is the cutpoint for LDL cholesterol plus 30 mg/dL, according to the National Cholesterol Education Panel’s Adult Treatment Panel III (ATP III) guidelines.

It can be argued that the requirement to have a fasting sample can be circumvented with a direct LDL cholesterol assay, which Dr. Hoefner notes is “minimally influenced by elevated triglycerides.” The problem, he cautions, is that the assay is not highly standardized. And while the assay tends to provide “quite accurate results” when testing “fairly normal” people, he says, it becomes increasingly problematic when people with dyslipidemias are tested. Thus, “In the group of people where you especially need to have correct results, you are getting less accurate values and, like calculated LDL-C, even a dyslipidemic patient’s ‘true’ LDL-C [by the gold standard β-quantification method] is generally lower in postprandial samples,” Dr. Hoefner points out. So from a biochemical perspective, “it’s bad science to use nonfasting blood for LDL-C, regardless of the method.”

Non-HDL-C doesn’t have that problem. Dr. Horowitz, who calls non-HDL-C a “winner” and “probably the marker of choice,” says, “It’s like estimated GFR for renal function in the sense that it requires only a calculation from data we already have, but the calculation is much simpler.” Dr. Hoefner, too, gives it a thumbs-up: “It’s cheap, it’s easy, it’s robust, and it appears to be better than LDL-C for assessing risk. I think we should be using it.”

What do the ATP III guidelines have to say about it? Gary Myers, PhD, chief of the clinical chemistry branch at the Centers for Disease Control and Prevention, says the guidelines recommend using non-HDL-C as a surrogate for apo B because non-HDL-C can be calculated, using total cholesterol and HDL-C, with what’s already in a lipid panel, avoiding the expense of another test.

But substituting non-HDL-C for apo B will not provide the same information or an equally robust risk prediction, some say.

“The problem with non-HDL cholesterol,” says John Contois, PhD, manager of research and development at Maine Standards Co. LLC, Windham, Me., is that, like LDL cholesterol, “it doesn’t reflect the number of atherogenic particles.” Apo B, a component of all atherogenic particles, provides a direct measure of the number. As a primary screening measure, non-HDL cholesterol clearly trumps LDL cholesterol, he says. That’s the finding of the AACC’s Lipoproteins and Vascular Diseases Division Working Group on Best Practices, whose position statement was published in Clinical Chemistry (Contois J, et al. 2009;55:407–419). The evidence shows, according to the statement, that apo B is first in risk prediction ability, followed by non-HDL cholesterol, with LDL cholesterol coming in third. The working group wrote, “We agree that a greater emphasis on non-HDL-C rather than LDL-C will improve patient care.” Because apo B has been validated more extensively in epidemiological studies and clinical trials, and because it reflects the number of atherogenic particles (unlike non-HDL-C), it would be the best measure to eventually replace LDL-C, the group says.

Ultimately, “the decision whether to use non-HDL and HDL cholesterols and/or apolipoproteins is up to the physician,” Dr. Contois says. And depending on the clinical situation, these tests may be complementary. “Rather than embrace a reductionist approach” limiting the number of tests, laboratory professionals “should provide doctors with the best tools—not the fewest,” he says.

Dr. Hoefner, a co-author of the position statement, is in favor of using non-HDL cholesterol for screening. He says, “Once you identify people at risk, there is certainly much more value to following them with apo B or LDL particle number.”

“From a clinical perspective,” says cardiologist and lipid expert Allan Sniderman, MD, Edwards professor of cardiology at McGill University Health Centre, Montreal, “it is important to appreciate that statins lower LDL-C and non-HDL-C more than they lower apo B. A patient can be at his or her cholesterol target, including the non-HDL target, but not at the apo B target.” Dr. Sniderman points out that apo B has been measured in all of the clinical trials. “The treatment goal in Canada [for apo B] is less than 80 mg/dL. In my view, the data support a target of 65 mg/dL in high-risk patients.”

No current U.S. guidelines set a target goal for apo B, Dr. Contois says. But the AACC position statement recommends a cutoff of less than 100 mg/dL, and less than 80 for high-risk patients.

While making an overall switch in the U.S. from LDL cholesterol to non-HDL-C or apo B wouldn’t rise to the level of a sea change, it would be likely to create a few waves, experts say.

The CDC’s Dr. Myers notes that since 1986, “we have had a cholesterol-focused health care [system]” where the message has been to “know your number and lower your cholesterol.” And “non-HDL ... is hard to understand. What is a non-anything? It’s harder for the public to pick up that idea. But we do change these things as new information becomes available.”

The National Cholesterol Education Panel’s ATP IV guidelines may be out later this year, Dr. Myers says. And if they recommend apo B or any other markers over LDL cholesterol or total cholesterol, an educational effort will be required.

There could also be a payment barrier to using non-HDL cholesterol or apo B. The Medicare Physician Quality Reporting Initiative rewards physicians for ordering LDL cholesterol on patients with particular diagnoses, Dr. Hoefner says.

“A clinician who switches to apo B in lieu of LDL-C—or even if he or she were to use non-HDL-C without an LDL-C—will lose that incentive payment,” he says. Dr. Hoefner discovered that firsthand recently when he started adding a comment on laboratory reports suggesting the clinician order apo B or non-HDL-C when the lab could not calculate LDL-C owing to elevated triglycerides.

All of the focus on LDL-C can overshadow HDL-C, which also figures in the risk prediction equation, and according to the meta-analysis published in JAMA, in a fairly powerful way.

The JAMA report, says Dr. Tracy, notes that the risk of coronary artery disease is lowest when HDL cholesterol is above 70 mg/dL—and highest when it’s below 34. And the risk is “graded all the way” between those two points, he notes. “The highest adjusted risk for non-HDL is around 2.2, and the highest adjusted hazard ratio for very low HDL is around 1.5.”

The JAMA report says “current findings suggest that therapy directed at HDL-C as well as non-HDL-C may generate substantial additional benefit.” Coronary heart disease risk, the authors note, is about two-thirds lower in those with 15 mg/dL higher HDL-C and 80 mg/dL lower non-HDL-C, “which are alterations that are attainable with, say, extended-release niacin plus a potent statin.” The authors add that long-term randomized trials are needed to see if these lipidmodifying regimens can ante up the “epidemiologically expected risk reduction.”

The authors caution that the failure of the HDL-C-raising drug torcetrapib “has raised questions about the value of raising HDL-C cholesterol and highlighted the need to characterize more reliably the relationship between HDL-C and vascular risk, particularly at high HDL-C levels.”

“The prevailing wisdom or dogma before the torcetrapib trial,” Dr. Tracy says, “was that LDL particles deposit in the artery walls and HDL particles help pull it out.” Yet even before that clinical trial, “some folks doing research on HDL had some degree of suspicion that not all HDL is the same and that not necessarily all would be entirely good for you.”

Prescribing niacin to raise HDL-C seemed to help patients, Dr. Tracy says. But raising HDL-C with torcetrapib not only did not show any benefit, there were more cardiovascular events in the treated group—and the trial was halted, he says.

Ivan Pacold, MD, MS, director of preventive cardiology at Loyola University Medical Center, Maywood, Ill., notes that in the clinical trial testing torcetrapib, the drug appears to have raised blood pressure, which could have offset its HDL-C-raising benefit. But he points out that when you give a drug that boosts HDL-C, you really don’t know if the higher HDL level is composed of the active or beneficial kind of HDL. That is, is it doing the job it’s supposed to do—which is, as Dr. Pacold puts it, to “scavenge cholesterol from the arteries?”

Dr. Pacold thinks the benefit of niacin comes from a multitude of effects—not solely raising HDL-C. It also lowers triglycerides and the LDL-C level, he says. He predicts that if a national clinical trial underway now, the AIM-HIGH study, which is testing the impact of a sustained-release niacin and a statin, turns out well, niacin may be the next important addition for people with cardiovascular disease or at risk for it.

The clinical trial is fully enrolled and is expected to be completed in 2012 or 2013, says Michael Miller, MD, director of preventive cardiology at the University of Maryland Medical Center, the principal investigator for the study at the center. (For more details, visit

Inflammatory markers also factor in the risk prediction. Dr. Contois points to data showing that C-reactive protein is elevated with vascular inflammation, though the marker isn’t specific for that. Even so, “It does appear to add information to the traditional risk factors, including LDL cholesterol.” And the JUPITER (Justification for the Use of Statins in Primary Prevention) trial is appearing to find that various stat­ins can benefit people with normal LDL cholesterol levels but elevated CRP, he adds. The study showed a reduction in cardiovascular disease endpoints among those treated with rosuva­statin. That group also had a reduction in CRP and a further decline in LDL cholesterol.

There’s an anomaly in the study that affects interpretation of the data, however. “About 40 percent of the subjects had metabolic syndrome, which is associated with an increased number of LDL particles,” Dr. Contois says. Thus, even though the study subjects all had normal LDL cholesterol values, that doesn’t necessarily mean they didn’t have elevated numbers of LDL particles. “Thus, the question still remains whether the [treatment benefits] seen were due to reduction in LDL particles, along with a further reduction in LDL cholesterol, or due to lowering CRP—or both,” he says.

“Lipoprotein-associated phos-pholipase A2 [Lp-PLA2] has also been implicated in vascular inflammation and appears to be more specific than CRP,” Dr. Contois continues. And statin therapy also reduces Lp-PLA2. “There hasn’t been as much outcome data for Lp-PLA2 as for CRP,” he says, and adds, “Until data suggest otherwise, I would not dissuade clinicians from using either test.”

Harvard cardiologist J. Michael Gaziano, MD, MPH, doesn’t think inflammatory markers are helpful in further risk-stratifying a low-risk or a high-risk patient. They can be of help, however, in figuring out whether an intermediate-risk person “really fits more in a high- or low-risk category,” he says. “If the person has had an MI or is young with a very low risk score, the inflammatory markers [CRP or Lp-PLA2, or both] aren’t likely to change what you do for the patient.”

Dr. Sniderman’s take on cardiovascular risk and inflammation is more one that focuses on targeting apo B rather than LDL cholesterol. “Atherosclerosis is the outcome of the trapping of apo B particles in the arterial wall. There’s no question that inflammation is involved in the process, but we don’t have specific, effective anti-inflammatory therapies right now,” he points out. And if you can avoid the problem, which is LDL “lodging in the artery wall, you don’t have to fix it. That’s what prevention should be.”

The preventive effort should start far earlier than it typically does, in his view. Lowering LDL-C, non-HDL-C, and apo B will reduce the number of cardiovascular disease events by 30 percent to 40 percent in people who are in their 50s and 60s, Dr. Sniderman says. “But I think if you started before arteries are horrifically damaged, you could reduce events by 90 percent.” He thus advocates starting to do a complete lipoprotein assessment, including apo B, of patients at age 25. (In his opinion, screening in developed countries should also include triglycerides. But he doesn’t worry too much about whether the person is fasting as long as anything abnormal is followed up.)

While the evidence and recommendations for testing are sorted out, the search for new cardiovascular risk factors presses on.

Dr. Myers predicts new factors will be found that will “push the envelope to assess risk” and “touch people missed by the traditional lipids, just like CRP supposedly does.”

“Are any of those going to make the transition from research to clinical use?” he asks, adding, “That’s the real question.”

Dr. Hoefner advises against expecting a “magic bullet” of a risk marker, given that cardiovascular disease is markedly multifactorial, but he says that doesn’t mean people shouldn’t continue to look for other ways to predict risk.

And Dr. Tracy predicts that in five to 10 years, genetic personalized information will be part of risk assessment. Couple that with a sharper understanding of lipid metabolism and better assays for the different lipid types, he says, and risk assessment “will be in a much stronger place.”

Karen Lusky is a writer in Brentwood, Tenn.