June 2010
Feature Story

Karen Lusky

As vitamin D has been tossed around in recent years as a possible public health panacea, no one could mistake the “D” for dull. More like “D” for debate, with laboratorians and clinicians talking up a storm over the past couple of years about vitamin D testing accuracy, optimal reference ranges, and the clinical utility of widespread screening.

Dialogue on these and other topics continues, with new data and developments in play. The bets are also on about the future growth of vitamin D testing, which has been off the charts for the past three years. Some experts predict the testing boom may ultimately shape the laboratory testing landscape in ways that extend beyond vitamin D.

As for testing accuracy, a number of data sources, including the CAP, suggest that labs using various assays are getting much better at achieving similar results for 25-hydroxy vitamin D (25OHD), the analyte used to identify vitamin D deficiency.

If true, it could help resolve a major cause of consternation among clinicians who sometimes complain that monitoring a patient’s 25OHD concentrations from different labs is like watching a bouncing ball.

The CAP Bone and Growth Survey for 25OHD in 2009 and earlier “showed a fairly large difference” among the peer group mean numbers for different test manufacturers, says Nataliya Polyakov, MT(ASCP), senior technical specialist for CAP Surveys.

But a proficiency testing challenge in April 2009 proved to be a different story. That CAP Survey, which, unlike previous ones, used fresh frozen serum, showed that “the [25OHD] methods are reasonably well harmonized,” says William Roberts, MD, PhD, vice chair of the CAP Chemistry Resource Committee and medical director of ARUP’s automated core laboratory, Salt Lake City.

The serum used in the April 2009 Survey is a commutable material, Dr. Roberts explains, “meaning it should behave like authentic patient samples and not be subject to matrix effects.” The College has a fresh frozen serum product for 25OHD in development for next year.

Participants in the Survey were using the following test methods: Dia­Sorin’s chemiluminescent immunoassay Liaison as well as its radioimmunoassay; Immunodiagnostic Systems’ (IDS) automated enzyme im­mu­noassay; and liquid chromatography tandem mass spectrometry (LC-MS-MS), a laboratory-developed test. The proficiency testing challenge did not include data from users of the HPLC-UV (high-performance liquid chromatography-ultraviolet) method.

The mean values for the IDS method, LC-MS-MS, and the DiaSor­in Liaison and RIA were 23.5 ng/mL, 30.1 ng/mL, 25.9 ng/mL, and 26.4 ng/mL, respectively, Dr. Roberts reports.

“It may be a little surprising,” he says, “that the immunoassay methods agree with mass spectrometry as well as they do.” Surprising, he says, because the “mass spec quantifies D2 and D3 separately, while the immunoassay methods measure the total based on the cross-reactivity of each.”

One could question whether the test methods have always been more in line than the CAP’s previous Surveys samples have shown. But that wouldn’t account for the clamor over wide variations in 25OHD test results seen in clinical practice.

The 2009–2010 data from DEQAS (Vitamin D External Quality Assessment Scheme), an international PT provider based in the United Kingdom, also shows that “with one or two exceptions, methods are generally tending to come closer together,” says DEQAS organizer Graham Carter. LC-MS-MS is showing less positive bias (divergence from the all-laboratory trimmed mean), and the DiaSorin Liaison is showing less negative bias than before, he adds.

Moreover, a small study performed in 2009 at the University of Wisconsin, Madison, and reported in abstract form at last fall’s American Society for Bone and Mineral Research annual conference, showed that labs using LC-MS-MS and the DiaSorin Liaison produced test results that correlated well, says Neil Binkley, MD, the study’s lead researcher, an internist and geriatrician with the university’s School of Medicine and Public Health. The labs using LC-MS-MS “had a little bit of a positive bias that would not be clinically significant.”

To perform the study, Dr. Binkley and colleagues sent 25 aliquots of sera used for quality assurance purposes in the University of Wisconsin lab to eight laboratories across the nation; five labs were using Dia­Sorin’s Liaison, and three were using LC-MS-MS. The laboratory staff at all eight labs believed they were testing patient samples.

A local physician group had been concerned that there were problems with 25OHD2 testing accuracy, so the researchers made sure “over half of the sera contained endogenous” 25OHD2, Dr. Binkley says. The University of Wisconsin HPLC-UV assay served as the standard for comparing the other labs’ test results.

The study’s findings contrast sharply with those of a study performed at the University of Wisconsin in 2003. At that time, Dr. Binkley and colleagues sent blood samples, including his own, to a number of labs in the Midwest. And the variations seen in test results weren’t reassuring.

“For example,” says Dr. Binkley, “the 25OHD result on aliquots of my sample ranged from 15 to 41 ng/mL, depending on which lab” did the test.

What might account for the seemingly increased harmonization among labs using immunoassays and LC-MS-MS?

“I don’t know ... to be honest,” says DEQAS’ Carter.

Kenneth Gibbs Jr., managing director of IDS, however, believes the recent data “confirm that providers of various assay methods have worked diligently to improve performance specifications for their vitamin D products.”

Ravinder Singh, PhD, co-director of the endocrine laboratory and clinical biochemistry and immunology at Mayo Clinic in Rochester, Minn., is not convinced “things have settled down” in terms of there being significant differences in test results among assays. “It comes in waves,” he says. And “there is [still] no standardization at all.”

The National Institute of Standards and Technology, or NIST, continues to lay the groundwork for standardizing the testing, an effort not without its challenges and critics.

In February of this year, NIST published its LC-MS-MS candidate reference measurement procedure for 25OHD (Tai S, et al. Anal Chem. 2010;82(5):1942–1948) (http://pubs.acs.org/doi/abs/10.1021/ac902686). NIST’s Karen Phinney, PhD, co-author of the article, says the reference method, which involves a lengthy sample analysis, isn’t meant to be used as a routine method. Rather, NIST uses the reference method and others to assign values to its reference materials, which labs can then use in assessing the accuracy of their own 25OHD testing methods.

In July 2009 (after the April 2009 PT challenge using fresh frozen serum), NIST released Standard Reference Material (SRM) 972 for 25OHD, which includes serum-based material with four different levels of the metabolites that labs can use for quality control. In December 2009, NIST released SRM 2972, which can be used for calibration, Dr. Phinney says.

Concerns have been raised over whether SRM 972 materials will work for immunoassays. Carroll Streetman Jr., president of DiaSorin, which provides the lion’s share of assays for 25OHD testing in the U.S. and abroad, has been advising clinical practitioners that only standard 1, of the four SRM 972 serum reference materials for 25OHD measurement, is “commutable and therefore suitable for all commercial 25OHD immu­no­assays, since it is an unaltered human serum sample containing predominantly 25OHD3.” He says, “SRM ‘standards’ 2, 3, and 4 are not commutable for reasons previously recognized by DEQAS” (Carter GD, et al. J Steroid Biochem Mol Biol. 2007;103:480–482).

NIST plans to look into DiaSorin’s claims that the SRM 972 materials are not commutable with immunoassays by doing a study that will include all test methods. If the answer turns out to be that they’re not commutable, NIST will provide that information to potential purchasers of the reference material for 25OHD, Dr. Phinney says.

The NIST reference method for LC-MS-MS has overcome a testing accuracy issue that can artificially inflate LC-MS-MS 25OHD3 results: a C3-epimer potentially present in adult serum. “Initially,” says DEQAS’ Carter, “it was claimed that the ... epimer was present only in neonatal serum and thus was not a problem” when testing adults. But NIST’s candidate reference measurement procedure found the epimer in adult serum.

Using LC-MS-MS testing, University of Washington pathologist Andy Hoofnagle, MD, PhD, and colleagues have detected the C3-epimer for 25OHD3 in many adults. In some of them, the epimer accounts for up to 12 percent of their total 25OHD3, he says.

Overall, however, the C3-epimer drives up 25OHD3 by five to 10 percent, “so it’s not huge,” Dr. Hoofnagle says. “Reviewers who have looked at our data [pointed out] that the coefficient of variation of [our assay] is five to 10 percent. They didn’t buy that this was clinically important.”

For DEQAS’ Carter, the presence of the C3-epimer has raised doubts about using LC-MS-MS as a reference method. It’s possible, he cautions, that other isomers with the same mass and similar fragmentation patterns as 25OHD could also skew test results. Thus, DEQAS plans to cross check 25OHD values produced by LC-MS-MS against gas chromatography mass spectrometry values.

Meanwhile, the Centers for Disease Control and Prevention has revised its LC-MS-MS method so that it separates out the C3-epimer. The method also pro­vides “better sensitivity for 25OHD2,” says CDC laboratorian and researcher Rosemary Schleicher, PhD. If the new method pans out in validation studies, the CDC plans to use it to re-analyze some of the National Health and Nutrition Examination Survey (NHANES) samples tested for 25OHD levels using the radioimmunoassay method. Numerous researchers have used those 25OHD results in linking 25OHD concentrations to various health conditions.

“We won’t re-analyze all the NHANES samples from 1988 to 2006—about 60,000, but we will use the new method to make comparisons with the historical DiaSorin RIA data,” Dr. Schleicher says. “We plan to provide regression equations such that going forward, the LC-MS-MS data can be compared to the RIA data.”

Debate continues over the appropriate cutoff for saying patients have inadequate vitamin D concentrations. Laboratorians at ARUP performed a study on vitamin D concentrations in healthy subjects, using parathormone, or PTH, as “a surrogate for vitamin D adequacy,” Dr. Roberts says. They found that “as subjects have higher vitamin D levels, their PTH values come down.” As a result, “we think a value of somewhere around 30 ng/mL and above is probably adequate,” Dr. Roberts says. “That value might, however, be as high as 40, but probably 30, which people have identified all along as being a fairly reasonable value for normal when confirmed with the LC-MS-MS and the DiaSorin Liaison.”

Dr. Binkley at the University of Wisconsin aims to get his patients’ blood concentrations to 40 ng/mL. He thinks some of the debate over the “right number” for the cutoff reflects “whether you demand evidence or practice physiology-based medicine.” In his view, one can look to “highly sun-exposed individuals like humans used to be” to get a heads up on “normal” vitamin D status. Those people usually run from 30 ng/mL to 80 ng/mL.

Pointing to the lack of long-term outcome data showing it’s safe and efficacious to prescribe a supplement for those whose serum 25OHD values exceed 20–30 ng/mL, Dr. Hoofnagle at the University of Washington takes a more conservative approach. “We call less than 20 ng/mL abnormal and from 20 to 50 normal.” He agrees that a value of under 20 ng/mL increases the risk for developing osteomalacia as an adult and rickets as a child. And “while there have been short studies showing all sorts of good things that happen with values above 30, there are no safety data over the long term.”

There are data warning that too much vitamin D could be dangerous to one’s health, apart from acute toxicity, which experts say is rare. (Dr. Hoofnagle points out that 88.4 ng/mL is the lowest documented level of acute vitamin D toxicity reported in the literature.)

Harvard professor of medicine JoAnn Manson, MD, MPH, PhD, points to a “few concerning reports of increased risk of pancreatic and other cancers associated with very high blood levels of 25OHD,” as well as a higher risk of cardiovascular disease, vascular calcification—even all-cause mortality. “We need to better understand these findings,” says Dr. Manson, chief of preventive medicine at Brigham and Women’s Hospital. “There may be a threshold above which some risk may occur and the risks may outweigh the benefits.”

The Institute of Medicine will soon issue recommendations for vitamin D intake and sufficient 25OHD blood levels, says Dr. Manson, a member of the IOM committee formulating the recommendations, which will also address calcium intake. The final report will probably be released in late summer or early fall.

Another question is beginning to generate discussion: Is an isolated 25OHD level itself sufficient for clinicians to assess and treat vitamin D deficiency?

Endocrinologist and bone and mineral specialist D. Sudhaker Rao, MB, BS, of Detroit’s Henry Ford Health System, thinks not.

It is his “biased opinion,” he says, and one about which he’s becoming increasingly clear, that clinicians should order PTH in conjunction with 25OHD. Otherwise, you “don’t know how to interpret a low value in the context of a seasonal variation or the duration and/or magnitude of the vitamin D depletion.” Dr. Rao is director of the Bone and Mineral Research Laboratory at Henry Ford Hospital.

In a recent study at Henry Ford Hospital, Dr. Rao and colleagues looked at 486 patients over the age of 50 with 25OHD and PTH testing performed at the same time. They found that about 18 percent of the patients had low 25OHD but a normal PTH. That indicates that “the individual has a seasonal variation or that the vitamin D depletion is of recent onset,” Dr. Rao says. “Therefore, it doesn’t warrant aggressive pharmaceutical therapy, such as 50,000 international units.”

“People are going ballistic over vitamin D” dosing, he says, which he describes as everyone taking 50,000 units of vitamin D when they are found to have a low 25OHD value. He questions why pharmaceutical vitamin therapy is given when PTH values are normal. The PTH results in such cases imply that the vitamin D depletion is recent and therefore less clinically urgent. “That’s my interpretation,” Dr. Rao says.

Forty-four percent of the patients in the study at Henry Ford had low 25OHD and high PTH, which Dr. Rao calls “predictable and understandable.”

“This combination implies long-term vitamin D deficiency,” he says, which does support “a certain amount of urgency” for replenishing vitamin D stores.

About five percent of the patients in the study with normal 25OHD had high PTH and normal serum calcium. “We think these patients most likely potentially have normocalcemic primary hyperparathyroidism caused by a little tumor in the parathyroid gland.”

Thirty-three percent of the patients had normal 25OHD and normal PTH.

“Dividing individuals into these groups gives the clinician an idea of when to offer pharmacological therapy as opposed to just telling the person to take, say, 1,000 units or so of vitamin [D3] from the drugstore,” Dr. Rao says.

He and colleagues presented the study data at the 2008 American Society for Bone and Mineral Research annual conference. “I don’t think anyone paid much attention to it,” he says, which is why he submitted an abstract for the 2010 conference with additional data spanning five years.

Dr. Binkley doesn’t order PTH routinely when testing patients’ vitamin D status, noting, as does Dr. Rao, the lack of consensus about this in the bone and mineral world. There’s a relationship between vitamin D status and PTH, he agrees. But, in his view, “It’s far from a tight relationship. In one simplistic scenario, for example, a person could have vitamin D inadequacy but be taking very high dietary calcium to the point that vitamin D isn’t needed in the gut to mediate transport of the calcium in the blood. Thus, the PTH would be normal.”

Unfortunately, Dr. Binkley adds, how best to approach vitamin D repletion is not guided by strong data, though guidelines in the works from endocrinology professional groups and the American Geriatrics Society should be available in the near future.

Many laboratories are reporting ever-rising volumes of 25OHD testing. DiaSorin’s Streetman says 25OHD testing volumes have grown 80 to 90 percent “year over year for the past three years.”

At Barnes Jewish Hospital in St. Louis, vitamin D testing “has continued to climb from about 3,000 in April 2009 to 3,700 this April 2010,” says Ann M. Gronowski, PhD, associate professor of pathology at Washington University School of Medicine.

The demand for 25OHD testing at Henry Ford’s Bone and Mineral Research Laboratory has increased more than twofold since 2008, says laboratory supervisor Nayana Parikh. Henry Ford and Barnes Jewish Hospital use the DiaSorin Liaison.

In March, Watson Clinic LLP, a large multispecialty physician practice in Lakeland, Fla., which is also a Dia­Sorin Liaison user, hit 2,170 25OHD tests—its “highest volume yet,” says Joan Fitchett, BSMT, MHSA, director of laboratory services.

The costs to do the test fall into the $15 to $18 range per test and are probably no more than $20, with Medicare paying $46 to $48 and managed care mostly matching Medicare rates, Fitchett says. “We have over 200 providers, including 40 specialists, so we get a lot of OB/GYN physicians and endocrinologists ordering it for people with osteoporosis who may be on hormone replacement therapy. Many of our patients come to their physician and want the test because they read an article about vitamin D.” People get the testing, she finds, for all sorts of reasons.

Medicare payment is holding steady since last year, says Charles Root, PhD, founder and president of CodeMap, Barrington, Ill. That’s when Highmark Medicare Services, the part A administrative contractor for Delaware, New Jersey, and Pennsylvania, put out a more stringent local coverage decision, which, Dr. Root says, wasn’t “that dramatic.” However, he predicts continued local coverage policy developments over the next couple of years to specify diagnoses for reimbursement purposes. Medicare doesn’t cover the test for nutritional assessment.

Of course, between reimbursement uncertainty and lack of causal data to support widespread vitamin D supplementation, many labs continue to wait for the balloon to burst.

IDS’ Gibbs doesn’t think test volume will continue to grow at the current pace well into the future. He predicts the long-term prospects for vitamin D testing are likely to be pinned to “defining studies that provide evidence of clinical benefit” for vitamin D beyond “skeletal health.”

Dr. Binkley believes “there’s quite good evidence that vitamin D is important for bone health, falls, and fractures in older adults, and pretty good, but not perfect, evidence that it’s important in cancer.” The research group at the Medical University of South Carolina presented evidence recently on vitamin D and pregnancy-related outcomes.

The data, presented in abstract form by pediatric researcher Carol Wagner, MD, at the Pediatric Academic Societies annual meeting in Vancouver, cast vitamin D in a glowing light. “The spectacular part of the study was it showed women replete in vitamin D had lower rates of preterm labor and preterm birth, and lower rates of infection,” Dr. Wagner said in a statement released by the American Academy of Pediatrics. “The greatest effects were seen among women taking 4,000 IU of vitamin D per day” (www.aap.org/advocacy/releases/PAS2010/pregnant.htm), the statement said.

Dr. Binkley believes randomized controlled trials are needed before vitamin D deficiency can be accepted as causally related to all the other conditions and diseases, such as diabetes, hypertension, and cardiovascular disease. “In the interim, I think being cautious and taking a wait-and-see approach is prudent,” he says.

A large National Institutes of Health-funded randomized clinical trial at Harvard University could provide evidence-based answers supporting or debunking the potential of vitamin D as a pill for most every ill, though not anytime soon.

Enrollment in the study will continue through 2011. Study participants will receive five years of treatment and monitoring to examine the impact of taking 2,000 IUs a day of vitamin D3 with or without omega-3 fatty acids, says Harvard’s Dr. Manson, principal investigator of the Vitamin D and Omega-3 Trial, known as VITAL. The study will look at the impact of the nutrients, individually and in combination, on a long list of ailments, including cancer and cardiovascular disease, diabetes, hypertension, autoimmune disease, cognitive decline, depression and mood, lung function, physical function, and numerous other outcomes, she says.

Even if vitamin D doesn’t turn out to be one of the major health finds of the century, it may have another legacy. James Nichols, PhD, medical director of clinical chemistry at Baystate Health, Springfield, Mass., believes the high volume of vitamin D testing has opened the door to labs being able to invest in mass spec, which even for midsize hospitals has been prohibitively expensive.

And Dr. Hoofnagle says testing for 25OHD, a small molecule, has raised awareness of issues involving small-molecule immunoassays overall. “Testosterone and thyroid hormone are great examples of tests that we have been doing by immunoassay; we know they are a problem, especially at the very low end of testosterone, for instance.”

In general, says Washington University’s Dr. Gronowski, if you can develop a test for LC-MS-MS, “it is usually better for small molecules than immunoassays.” Her laboratory is getting set up for LC-MS-MS. “We will put immunosuppressants on it first, which is what I think a lot of people start out with when they go to mass spec.” (The laboratory has no plans at this time to do vitamin D on the platform.)

But Dr. Gronowski’s prediction is “that vitamin D will eventually be done mostly on mass spec unless a better method comes out.”

Right now, Dr. Hoofnagle concedes, “the initial capital expenditure for LC-MS-MS is high.” And for many, “it’s still an intimidating technology,” he says, agreeing it is complicated. But the cost pressures could ease, he says, once a reagent rental model becomes available.

Waters Corp., of Milford, Mass., an LC-MS vendor, is working on such a model to bring down the costs for its customers, says Donald Mason, Waters’ clinical business manager. He’d be surprised if other LC-MS vendors aren’t doing the same.

“The next step” for vitamin D testing technology, Mason predicts, “is the integration of sample preparation with liquid chromatography and mass spectrometry to produce analyzers closer to what chemistry labs are accustomed to with their chemistry and immunoanalyzers.” And that “is a plug-and-play or black-box-type analyzer.”

Several companies are and will be working on the plug-and-play devices and trying to get them through the FDA, Dr. Hoofnagle says. No one can say when that will happen.

As for vitamin D testing right now, DiaSorin’s Streetman foresees the market “embracing higher-throughput systems which are only moderately complex but ... afford far greater ease of instrumentation management” than LC-MS-MS now does, from operational and maintenance perspectives. He predicts the Liaison will “continue to be a strong player, but with several large companies working on vitamin D, it’s only a matter of time before there are more automated immunoassay players.”

When asked about LC-MS-MS for vitamin D testing, Henry Ford’s Dr. Rao raises a question he considers more pressing: Should everyone be tested? And so the discussion takes off in another direction where there’s no shortage of different opinions.

On one point, though, most everyone could perhaps agree: Vitamin D has been a powerful catalyst for getting test-makers, laboratorians, clinicians, and researchers to take a hard look at issues that will pave the way for improving patient outcomes and safety.