July 2009: Volume 35, Number 7
Vitamin D Testing—What’s the Right Answer?
Labs Grapple with Confusing Analytics, Evidence
By Genna Rollins
A decade ago, the main reason physicians ordered a vitamin D test was if there was a concern about the patient’s bone density. However, a flurry of research now has implicated the prohormone in a host of conditions, including among others, certain cancers, type 1 diabetes, multiple sclerosis, tuberculosis, Alzheimer’s disease, psoriasis, and all-cause mortality, while also indicating that vitamin D insufficiency is prevalent. Even though vitamin D’s mechanism of action in the prevention or progression of various diseases has not been established definitively, these studies have been picked up by the popular press, causing demand for vitamin D testing to skyrocket. At the same time, there have been notable technological advancements in vitamin D testing, controversies surrounding certain tests, and considerable confusion about both analytical considerations and clinical interpretation of vitamin D results, leaving laboratorians and clinicians unsure about how to implement or interpret the test properly.
“We are ahead of the evidence. The good reasons and good justification was not there to do so many tests as we are doing now,” contended Ravinder Singh, PhD, DABCC, co-director of the endocrine lab at the Mayo Clinic. “Every year millions of tests are being ordered and now we are running into a situation where doctors are confused, labs are confused, and the only reason we’re confused is because we’re not ordering this test on the right patient population.” Volume at the Mayo Clinic, which uses a liquid chromatography tandem mass spectrometry (LCMS) method, has increased to > 61,000 tests per month in December 2008, up from >19,000 per month in July 2006.
Surge Evident in Utilization Survey
CLN’s recent email survey bears out the increased volume of testing, as well as ambiguities surrounding use of the vitamin D test (see Box, below). More than 50% of respondents indicated that the volume of 25-hydroxyvitamin D (25-OH-D) testing in their labs had increased by at least 50% in the past year, with 27% indicating that volume had gone up by more than 100%. While much more modest, increased volumes also were reported for fractionated 25-OH-D testing and 1,25-dihydroxyvitamin D (1,25 (OH)2 D) testing. In the case of the relatively new fractionated 25-OH-D test, which breaks out total 25-OH-D and individual levels of 25-hydroxyvitamin D2 and 25-hydroxyvitamin D3, 32% of participants who said their labs perform the test indicated that the volume has increased by at least 50% in the last 12 months. At the same time, while 40% of participants who said they perform the 1,25 (OH)2 D test assay indicated that there had been no change in volume during the past year, nearly one-quarter reported an increase of up to 25%, and 13% reported a 50% or higher increase in volume, which appears to be out of sync with the test’s limited diagnostic indications for assessing parathyroid disease and calcium metabolism disorders.
The out-of-kilter 1,25 (OH)2 D volume is no surprise to lab directors who are finding that not all physicians know which test —25-OH-D or 1,25 (OH)2 D—is correct for determining overall vitamin D status. “Fifty percent of physicians are still confused about what is the appropriate test. Some clients order both and one comes back normal and the other’s not. Then they call me and ask, ‘what should I believe?’ and I have to explain the difference,” observed L.V. Rao, PhD, director of core laboratories and associate professor of pathology at
UMass Memorial Medical Center in Worcester. The UMass lab, which is a referral lab for Massachusetts and surrounding states, performed 400 to 500 vitamin D tests per month 3 to 4 years ago, but now is running between 12,000 and 13,000 per month, according to Rao.
“There’s significant confusion out there,” agreed Hershel Raff, PhD, director of the endocrinology research laboratory for ACL Laboratories and professor of medicine at the Medical College of Wisconsin in Milwaukee. “Vitamin D testing’s becoming part of the routine physical in Wisconsin, but some doctors don’t really understand what they’re ordering, so they check off anything that looks close.” Raff believes the vitamin D naming conventions only muddy the waters: 25-hydroxyvitamin D2 also is known as ergocalciferol, 25-hydroxyvitamin D3, as cholecalciferol, and 1,25 (OH)2 D, as calcitriol. Some doctors mistakenly think 25-hydroxyvitamin D2 or 25-hydroxyvitamin D3 is the same as 25-OH-D or 1,25 (OH)2 D. Testing volume for 25-OH-D at ACL Laboratories has increased more than 50% in the past year. The spike over the last several years was so significant that Raff retired a manual radioimmunoassay (RIA) and brought in two automated immunoassays. “When we started we were only doing about 90 tests a week, then it got to where I would have had to hire five people to keep doing it manually,” he explained.
Overview of Vitamin D Testing
CLN Survey Reflects Volume Increases, Analytical Challenges
CLN recently conducted an e-mail survey of readers to learn about their experiences with vitamin D testing. Respondents came from a variety of settings: 37% reference labs; 21.8% academic medical centers; 20.9% private hospitals; 15.6% state/county/local hospitals; 4.7% veterans/military hospitals (n=212).
What vitamin D test does your lab perform?
How has the volume of vitamin D testing at your lab changed over the past 12 months?
Selected Comments from Survey Respondents
Survey respondents had a lot to say about vitamin D testing.
“These analytes need standardization and a consensus on appropriate reference ranges. I question the utility of most of these measurements in patient management.”
“We think we have it right at this time. Waiting to see if further research might change the sufficiency cutoff to a higher number.”
“Providers in my institution order vitamin D levels on EVERYONE due to the trendy nature of the test, in my opinion. Also, after clinical interventions are made/adjusted, our providers are rechecking Vitamin D levels monthly, and that is both clinically insignificant and really hampering our ability to ensure good turnaround time.”
“There is so much misinformation on vitamin D testing.”
“Our vitamin D volumes have skyrocketed in the past 2 years from 1-2 /month to our current volume of >200/month. We are watching the market closely for a method or methodology that will fit our lab. Space, cost per reportable, and staffing are issues we must consider to bring a new procedure in-house.”
What’s the Right Reference Range?
In concert with the confusion about which vitamin D test to order, there is no consensus about the appropriate reference range and cut-offs for deficiency or insufficiency, as well as optimal and possible toxicity status. Studies have identified various ranges for disease risk reduction, and labs use a variety of ranges, which was reflected in CLN’s survey. Reported deficiency levels for 25-OH-D started at ≤8 ng/mL and went all the way up to ≤30 ng/mL. The same discordance was reflected in reported insufficiency ranges, which varied from 8 to 20 ng/mL to <32 ng/mL. Sufficiency and/or optimal ranges also had quite a number of reported cutpoints, from 25 to 80 ng/mL and 31 to 150 ng/mL to 32 to 100 ng/mL and 50 to 80 ng/mL. Likewise, reported possible toxicity levels included >80 ng/mL, >100 ng/mL, >150 ng/mL, and >200 ng/mL. Although a consensus among vitamin D researchers is coalescing around <10 ng/mL as an indication of deficiency, 10 to 30 ng/mL for insufficiency, 30 to 100 ng/mL for sufficiency, and >100 ng/mL for possible toxicity, clearly there is room for further evidence and agreement.
This variation is due in large part to the reference ranges being based on recommended levels from studies rather than actual analysis of serum. “If you take the classic approach of determining the normal level of calcium in the blood, for example, you’d get 100 samples from your reference population and measure calcium and define the reference range as plus or minus two standard deviations,” explained Reinhold Vieth, PhD, FCACB, clinical chemist and director of the Bone and Mineral Lab at Mt. Sinai Hospital in Toronto. “The problem with vitamin D is that the reference sample will be different depending on what city you’re in, what your subject profile is, and whether it’s winter or summer. Vitamin D is somewhat like cholesterol levels, in that you’re not looking at what is ‘normal’; you’re looking at what is optimal or desirable. It’s a literature range, not a population range.”
Perhaps because of the rising volume and lack of consistent uses for vitamin D testing, National Government Services (NGS), a major Medicare contractor with claims processing in at least 16 states, proposed a new vitamin D payment policy in February. NGS planned to end payment to labs and physicians for routine testing for vitamin D deficiency, except for patients with chronic kidney disease, osteomalacia, hypercalcemia, and rickets. Following comments from professional organizations such as The Endocrine Society, NGS modified the policy to include osteoporosis, hypocalcemia and vitamin D therapy follow-up as indications for testing. The new policy was to become effective July 1, 2009 but reportedly has been extended to September 1 to allow for further review.
AACC to Hold Audioconference on Vitamin D Testing
This fall, AACC will host a 90-minute audioconference that examines the analytical challenges and clinical implications of testing patients’ vitamin D status. Moderated by André Valcour, PhD, of LabCorp, the program will address the various types of technologies used to perform vitamin D testing today, as well as examine issues such as standardization and reference ranges. For more information, visit the AACC website and look for the program under “Upcoming Events.” Supported in part by an education grant from DiaSorin.
Yet another level of confusion exists as a result of several lingering analytical issues with the various assays. The RIA manufactured by Diasorin is the predicate FDA device and the one used on most studies to establish vitamin D reference ranges. Since the RIA's introduction in the early 1990s, newer generation versions of the assay have been released, and more recently new vitamin D testing methods have emerged, including high performance liquid chromatography (HPLC) and LCMS, which are laboratory-developed tests.
According to Rao, it is not uncommon for physicians to receive vitamin D test results for the same patient but from different labs that have used different assays and produced different results, putting laboratorians in a spot to explain the discrepancies. For instance, UMass Memorial Medical Center uses both an immunoassay and LCMS to measure 25-OH-D. Rao analyzed results from the two methods and found that with both using a cut-off of <30 ng/mL for insufficiency, 68% of results from the immunoassay indicated insufficiency, whereas only 30% of LCMS results were below that level. Based upon that analysis, Rao conducted a separate study using both methods on samples from 600 patients. He found there was up to 40% positive bias towards LCMS.
Rao’s experience is not isolated. Richard Garcia-Kennedy, MD, laboratory medical director at California Pacific Medical Center (CPMC) in San Francisco conducted a similar analysis. CPMC does not perform 25-OH-D testing, and uses two different reference labs, one that operates an immunoassay method and one that runs LCMS. In the same 1-month period, results from the lab that employs the immunoassay indicated that 79.8% of patients had vitamin D levels <32 ng/mL, which was the stated lower range of normal, while the other lab reported that 46% of patients had vitamin D levels <32 ng/mL. “The forty six percent below 32 ng/ml is believable because our population is dark-skinned, we have foggy summers, and a large Asian population which is very lactose intolerant. But the eighty percent makes no sense,” said Garcia-Kennedy. Busy primary care physicians were even more befuddled by the results because the lab that uses LCMS reported <20ng/mL as the deficient or abnormal level, with additional verbiage on test reports indicating that >32ng/mL is an optimal level. “Remember, the primary physician or his or her assistant in the trenches is scanning reports for abnormal flags and will not necessarily be conversant with the issues of 20 versus 32 ng/mL for normal,” explained Garcia-Kennedy. Using the <20ng/mL metric, the LCMS-immunoassay differences were even more pronounced, and a source of vexation for the affected practices.
After reviewing the discrepancies with CPMC’s endocrinology department, Garcia-Kennedy has essentially thrown in the towel vis-à-vis vitamin D testing. “When people ask me about vitamin D, I say ‘don’t even bother measuring it. Just take 500 IU of vitamin D daily, because it makes absolutely no sense from the practitioner’s perspective.' There’s a lack of foundation here still, that makes these numbers not useful,” he contended.
The U.K.-based Vitamin D External Quality Assessment Scheme (DEQAS), the largest vitamin D proficiency testing program with more than 600 registered participants, including 225 labs from the U.S., likewise has observed extensive interlaboratory variation. A 2008 review found that the annual mean interlaboratory imperfection has generally declined from near or about 30% in the mid 1990s to just under 20% in 2007, with sporadic increases as new methods were introduced and problems with certain assays were identified. Significantly, the DEQAS analysis revealed that between April 2007 and January 2008, the LCMS method produced a mean positive bias >10% from the all-laboratory trimmed mean (ALTM). "The LCMS method was a huge advance with vitamin D, but it's giving higher values for reasons we don't understand," noted DEQAS organizer Graham Carter, MSc. CAP also performs vitamin D proficiency testing.
The Gold Standard?
While LCMS is becoming the preferred analytical method of many reference labs and has been called the gold standard for measuring vitamin D, not all lab directors see it that way. "Mass spec is the gold standard to give you an idea of the molecular weight of a compound passing through the machine, but not the amount of that compound," said Vieth. Detractors of using LCMS for vitamin D testing also point out that results are subject to each lab's individual reference procedures and calibrations, which adds to the imprecision.
"It's a wild west show. There are no standardized procedures, no FDA oversight, and both HPLC and LCMS are very user dependent and user interpretative," argued Bruce Hollis, PhD, professor of pediatric biochemistry and molecular biology and director of pediatric nutrition services at the Medical University of South Carolina in Charleston. Hollis was instrumental in developing the predicate RIA and is a paid consultant to Diasorin, a leading manufacturer of vitamin D immunoassays. Singh countered that the Mayo Clinic’s LCMS method has proven to be analytically stable with no biases or trends observed in a 2 ½ year observation period. Performance data from the lab’s calibrators indicated a CV of <3%.
Hollis has argued that the LCMS methods should be calibrated to the RIA, but Carter thinks that would be a mistake. "Intuitively you'd expect LCMS to be more specific and that it will be more likely to give 'the right answer' than an immunoassay, which is subject to all kinds of interferences, matrix effects, and the like," he explained. "So the idea of calibrating a mass spec method to an immunoassay, however good the immunoassay, is simply a bad idea."
To sort out the problems observed with the proficiency testing surveys, DEQAS has commissioned the development of a new gas chromatography-mass spectrometry (GCMS) method for measuring vitamin D and expects to use the method to check the validity of the ALTM and provide values for 25-OH-D3 and 25-OH-D2 on at least some samples, according to Carter. The GCMS method might also help further establish LCMS as the gold standard by providing a method to check it against, he added.
Meanwhile, the vitamin D immunoassays have had their own share of challenges. Diasorin recalibrated its RIA with new antibodies in the late 1990s and recently modified its automated immunoassay. IDS also recalibrated its immunoassay in 2006. These changes have caused a methodological drift, with variable results from the immunoassays over time. For instance, in 2004 the Centers for Disease Control and Prevention reanalyzed a subset of 150 banked serum samples using the updated Diasorin RIA kit which originally had been included in NHANES III (1988–1994) and had been tested with an earlier version of the Diasorin RIA kit. The results indicated that data obtained using the new RIA were on average 11.7% lower than data obtained from the old RIA.
Adding to the controversy about testing methodology, Quest Diagnostics and its subsidiary Nichols Institute Diagnostics experienced well-publicized problems with Nichol’s immunoassay for parathyroid hormone testing, and in April 2009, Quest and Nichols made a $302 million settlement with the U.S. Department of Justice to resolve civil and criminal claims concerning that test and a 25-OH-D assay, among others. Separately, Quest acknowledged in 2008 that for a period of about 18 months from early 2007 to mid 2008, the company's new LCMS method for vitamin D testing possibly was reporting erroneous results. Quest offered free retesting for affected samples, said to be <10% of the company’s total vitamin D testing volume. In a New York Times article, the problems were attributed to some faulty materials being used to calibrate test results and not all Quest labs consistently following proper procedures.
Hope on the Horizon?
Even as there has been a trail of confusion in vitamin D analytics, there is, in the eyes of many, hope on the horizon. NIST has been developing serum and solvent-based standard reference materials for 25-OH-D (SRM 972 and SRM 2972, respectively). As CLN went to press, SRM 972 was about to be released, while SRM 2972 was expected to be available by the end of July 2009, according to Karen Phinney, PhD, research chemist at NIST. In theory, the SRMs will provide both immunoassay users (in the case of SRM 972) and LCMS users (in the case of SRM 2972) a means to calibrate their assays, reducing the interlaboratory CVs for each method. One study found that harmonization to a common calibrator reduced average between laboratory variation to 9% from 30% (DOI:10.1373/clinchem.2008.107847). Mayo’s Singh has argued that the industry should aim for a CV <1% (DOI:10.1373/clinchem.2007.096156).
However welcome the SRMs may be, not all researchers see them as a panacea. For instance, Hollis believes SRM 972, which includes four serum pools, "will cause more confusion than it will solve." The first pool, which represents ‘normal’ serum, will work across the board with the various assays, according to Hollis. But he expressed concerns about the second and third pools. In those cases, he contends that the standard will work well for methods that physically destroy the proteins, but in the case of newer tests that rely on displacement of analyte from the binding protein without destroying the protein, then those standards will not work as well. He attributed this to artifacts emanating from the standards themselves, for example, equine serum used in the second pool and exogenous 25-hydroxyvitamin D2 spiked to the third pool.
NIST’s Phinney, however, defends the SRMs. "SRM 972 was developed in consultation with a panel of experts in the vitamin D field,” she explained. “Each of the four levels of SRM 972 is intended to pose a different measurement challenge, and individual users will need to assess which levels best suit their particular needs."
NIST also has been developing a reference measurement procedure (RMP) for vitamin D testing, as defined by the Joint Committee for Traceability in Laboratory Medicine (JCTLM). According to Phinney, NIST expects to publish the method within the next few months and subsequently submit it as a candidate RMP to the JCTLM.
As labs await release of the reference materials and methods and aim to improve vitamin D analytics, some, like Rao, have proposed using a conversion factor to adjust LCMS results downwards in order to use the reference ranges derived from RIA-based testing. Carter has suggested that this would be “akin to ‘cooking the books’ (albeit with the best of intentions)” (DOI:10.1373/clinchem.2009.125906). Rao found that the conversion factor only confused physicians more, as they were uncertain about which level to consider in clinical decision-making. For now many labs, UMass Memorial Medical Center among them, are reporting RIA reference ranges with LCMS results. However, laboratorians doing so may share Rao’s misgivings. “I’m uncomfortable using the same reference ranges for LCMS and our immunoassay because I know we have a difference,” he said.
Gritting our Teeth
As the vitamin D field continues to evolve, there are steps laboratorians can take to improve vitamin D testing proficiency and inform the medical community about the test’s appropriate use. Raff spends a good bit of time educating physicians one-on-one, in group presentations, and in writing, including in an internal newsletter that detailed the biochemistry of vitamin D and the differences between 25-OH-D and 1,25 (OH)2 D testing and their respective indications.
In organizations with computerized physician order entry (CPOE) systems, pop-ups that steer the physician to the correct vitamin D test can be helpful, according to Greg Sephel, PhD, FACB, clinical laboratory director for the VA Tennessee Valley Healthcare System and associate professor of pathology at Vanderbilt University in Nashville. “In teaching institutions with high provider turnover the key to communication is realtime information, and the best way to accomplish that is by setting up the computer to help with ordering patterns and interpretations,” he said. When CPOE is not available, Raff recommends removing 1,25 (OH)2 D from check-off lab requisition lists so that physicians will have to write in the test to order it.
The upswing in vitamin D testing volume is attractive for labs that want to increase their billable tests, but labs currently not performing the test should think carefully before bringing it in house, according to Sephel. “It’s not simply a make-or-buy cost decision. If you’re bringing in a new machine, you have to consider whether it’s completely, semi, or unautomated and how much labor and training it will entail, and you have to account for running controls, maintenance, proficiency testing, calibrations and quality of the test,” he explains. Those factors, along with staff shortages due to retirements and lack of space due to renovations, lead Sephel to decide for now that vitamin D testing will continue to be a send-out from his lab.
Sephel urged all laboratorians to stay on top of vitamin D testing, even if their labs do not perform the test. “For many labs this remains a send-out test, and there is a tendency not to have as much knowledge about methods and interpretations of those types of tests, but I think it’s important that we’re involved. We need to do what we can to make sure the test is being ordered and interpreted properly,” he said.
While the industry awaits analytical improvements and further clinical evidence about vitamin D, laboratorians will be challenged to maintain high-quality testing and keep physicians well-informed about testing indications. As Carter put it, “We’ll just have to grit our teeth.”