Test tube interference shows "chink in armor"

January 2005
Feature Story

Sue Parham

Between August and November 2004, clinical laboratories that use Vacutainer SST, Vacutainer SSTII, and Microtainer test tubes from BD Diagnostics received bulletins from the company, alerting them to an immunoassay interference related to the tubes that had been documented when a particular set of assays was performed on certain platforms from four manufacturers.

"Through the application of Six Sigma methods, BD scientists and engineers have been able to identify an interference between a constituent of our product and a limited number of assays," says John Boselli, vice president of quality for BD Diagnostics Preanalytical Systems. BD’s studies have confirmed, he says, that other assays on the affected instrument systems do not exhibit these interferences. "BD has also learned that there are other instrument platforms that appear not to exhibit any immunoassay interferences associated with our products," Boselli adds. (Related article: "Instruments, platforms, and assays affected by the tube-related interference").

Alan Remaley MD, PhD, and his colleagues at the National Institutes of Health were among the first labs to identify the interference in February 2004, when an NIH endocrinologist noticed that total T₃ results were discordant in comparison with other thyroid test results the laboratory generated. "We study a lot of patients with thyroid disorders, and in addition to other thyroid function tests, we perform total T₃ tests on all of our patients, even though generally total T₃ is not a widely ordered assay. I think that’s probably why this came to our attention first," explains Dr. Remaley, director of the immunoassay laboratory in the Department of Laboratory Medicine at the NIH.

Kenneth Blick, PhD, professor in the Department of Pathology and director of chemistry at the University of Oklahoma Health Sciences Center, Oklahoma City, says his lab identified a similar type of immunoassay interference with its folic acid and cortisol assays at about the same time. "One of our technologists had picked up discrepancies with cortisol, just from evaluations we were performing in the lab," Dr. Blick says. Cortisol results generated between March and May 2004 at his institution appeared to be 20 percent to 30 percent higher than those performed earlier in 2004.

Dr. Remaley and his colleagues witnessed a similar bias with the NIH’s total T₃ assay, discovering upon initial review that 25 percent to 30 percent of all of the total T₃ results generated by his lab between March and May were above the upper limit of the reference range. "Because we are a research institution, this wouldn’t necessarily raise a red flag, but upon further review, we discovered that most of those patients were euthyroid," he says.

Further, the lab noticed discrepancies between the results of total T₃ assays performed on two different instrument platforms, says Glen Hortin, MD, PhD, director of clinical chemistry, Department of Laboratory Medicine, NIH. "We have two instruments in our laboratory on which we can perform T₃ testing," Dr. Hortin says. "One was relatively severely affected, but we saw no effect on the other analyzer." This led the NIH researchers to suspect that the interference was out of the ordinary.

As more laboratories began to notice interferences with particular immunoassay tests—chiefly endocrine assays—they turned to their manufacturers for solutions. While these manufacturers worked with BD, of Franklin Lakes, NJ, to solve the problem, Drs. Remaley and Hortin joined forces with Raffick Bowen, PhD, clinical chemistry fellow in the NIH Department of Laboratory Medicine, and other NIH colleagues to study its cause and scope. Their findings appear in an article published online before print in Clinical Chemistry on Dec. 2 (www.clinchem.org/cgi/content/abstract/clinchem.2004.043349v1).

Using three types of serum test tubes—plastic Vacuette tubes from Greiner Bio-One (Longwood, Fla.), BD Vacutainer glass tubes, and BD Vacutainer plastic SST tubes—the researchers analyzed serum samples from apparently healthy volunteers, 15 males and 34 females, who ranged in age from 30-60 years. They performed 17 immunoassays (n = 49), 30 clinical chemistry tests (n = 20), and 33 immunology assays (n = 15) to determine which assays and tubes showed clinically significant interferences, and examined tube effects by adding pooled serum to different tube types.

The researchers discovered that when measured by the Immulite 2000 but not by the AxSym analyzer, total T₃ results were significantly higher for samples collected in the SST tubes (mean value: 2.81 nmol/L) than either the glass (mean value: 2.15 nmol/L) or Vacuette (mean value: 2.24 nmol/L) tubes. The effect was large enough to substantially shift the distribution of patient values, increasing the percentage of values above the reference interval from 11.3 percent to 35.8 percent. Results from several other tests statistically differed among tube types, but the differences were not considered to be clinically significant.

Importantly, they found interference not only from SST and SSTII tubes, but also from plastic plain red-top and tiger-top tubes. "Other tube types are giving falsely elevated results with the T₃ and possibly other immunoassays," Dr. Bowen says. That the plain red-top and tiger-top tubes are implicated also is no surprise, he adds, because they have internal tube coatings (i.e. silicone).

The NIH investigators also confirmed BD’s observations about the cause of the interference. "We don’t understand the mechanism of the interference fully, but we do now know that it is a tube additive," Dr. Remaley says.

The immunoassay interference observed with the total T₃ assay was the tip of the iceberg, Dr. Remaley says. "There were many tests that were affected, and some even far worse than total T₃." For example, vitamin B₁₂ showed a positive bias approaching 40 percent.

At the University of Oklahoma Health Sciences Center, the 30 percent increase seen in total T₃ was a great concern. "If somebody had had a level that was toward the upper end of the reference range," Dr. Blick says, "this could have pushed them over the cutoff and could have caused a diagnosis of hyperthyroidism."

However, says Laurence Demers, PhD, distinguished professor of pathology and laboratory medicine at Penn State University, M.S. Hershey (Pa.) Medical Center, clinicians expect variability when ordering total T₃ results, mainly because the assay has always had a great deal of variability associated with it. Thus, a 25 percent to 30 percent positive bias may not be clinically significant, Dr. Demers says: "The recovery on the analytical systems that measure total T₃ is not that good, so it doesn’t take much to upset the equilibrium that exists between plucking out an antigen with an antibody and having it interfered with by some factor that’s in the sample." In addition, for total T₃ , he says, "the variability from sample to sample in patients can run as high as 25 percent, which is about the same as the percentage of bias that’s been reported in conjunction with the tube-related interference."

Since total T₃ is a confirmatory test that’s usually ordered for patients in whom hyperthyroidism is suspected and the T₄ result is borderline elevated, total T₃ generally is not a major test in the thyroid workup, Dr. Demers notes. He says, "Although an interference such as that described in the Clinical Chemistry paper could cause a decision zone to be affected, it’s a minor change that’s generally not quite enough to cause a clinically significant change in patient management."

But other affected assays could be more problematic. The one that most concerns D. Robert Dufour, MD, professor of pathology at George Washington University, Washington, DC, in terms of the assay giving false-positive results, is the hepatitis B surface antigen. Under certain circumstances, the affected hepatitis B assay is subject to false-positives, and though the test manufacturer recommends that all positive hepatitis B surface antigens be confirmed by a neutralization assay, most clinical laboratories don’t perform this confirmatory test. "In our laboratory," Dr. Dufour says, "we do the neutralization assay with weakly positive samples, but not in all of our positives. So there is the potential that some patients were misclassified as being hepatitis B positive based on that test, particularly since it’s probably the most widely used test for hepatitis B surface antigen."

Another affected assay is a tumor marker for breast cancer, the BR test from Bayer Diagnostics of Tarrytown, NY, which has been demonstrated to have a positive bias of about 20 percent when using the affected test tubes. "With a 20 percent positive bias, a value of 35, which is the cutoff, would become a value of 42," says Martin Fleisher, PhD, chairman of the Department of Clinical Laboratories at Memorial Sloan-Kettering Cancer Center, New York. "It doesn’t sound like a big deal," he says, "but protocols are written so that patients will be treated in a certain way if a value is above or below a certain cutpoint. If that cutpoint is at 35, but then spuriously becomes a 42 because of an immunoassay interference, obviously the patient is being treated in a fashion that would be outside of the limits of the protocol."

Dr. Blick points out that not all immunoassays are "up for grabs." It’s the small-molecule competitive assays that appear to be most affected. "The sandwich, or noncompetitive, assays are okay," he says.

Since no one knows exactly when the interference began to appear, reference laboratories with affected systems are now offering to go back as far as October 2003 to re-test potentially affected samples. Clinical laboratories across the nation are having to make difficult decisions about whether to re-test samples.

"We went back and re-collected specimens and tested everything," Dr. Fleisher says. "At the time the technical bulletins were released, BD did not specifically indicate that the tests not listed were unaffected. We could not take the chance."

Because the interference varied from lot-to-lot, with some test tubes demonstrating biases and others returning accurate results, without performing some type of analysis labs were unable to determine whether the lots of tubes they were using were affected. At the same time, a similar variability was observed in selected reagent lots used in the affected assays. Fortunately, Memorial Sloan-Kettering’s lots of tubes appeared to be okay. "If we would have had a bad lot, however, it would have affected us tremendously because we have thousands of patients and numerous protocols that are being monitored all of the time, and small changes in test results are examined frequently," Dr. Fleisher says.

The University of Oklahoma Health Sciences Center had recently changed to a new type of specimen collection protocol, so only cortisol and folic acid were affected. The affected tests were sent to a reference laboratory that ran a platform not affected by the interference. "We don’t have a phlebotomy team, so all of the nurses and the house staff collect samples all over the center, throughout three hospitals, and we have stocks of tubes everywhere," Dr. Blick says. Tubes come to the laboratory via a pneumatic tube system. "For us, it was just an impossible task to go around and replace all of the tubes or instruct everybody to collect cortisol and total T₃ in glass tubes. Nevertheless, this was the solution BD was promoting at the time," he says.

BD began distributing a new test tube in October 2004, which it is referring to as the "adjusted product," and the company expects that once it reaches all of the affected laboratories, the new tubes will go a long way in remedying the situation. "In the spirit of trying to restore platform performance, to make sure we can drive the best possible clinical outcomes, we decided to take the positive action to adjust our products so that we could improve the performance of these systems. And we’ve achieved that," Boselli says.

Ana Stankovic, MD, PhD, MSPH, worldwide medical director of BD Diagnostics Preanalytical Systems, says: "Going forward, if customers are experiencing elevated results, the best strategy is to purchase the adjusted product. But if labs are unable to get the new tubes, they can continue to use BD’s plain serum tubes until the new tubes become available."

Regardless of what company’s test tube a laboratory uses to collect specimens, labs need to begin considering the test tube as a factor that can affect laboratory analyses, Dr. Hortin says. "Using mass spectrometry, Dr. Steve Drake in our laboratory looked at some of the components that are shed by test tubes, and he saw a variety of polymers in the samples that are used as surface coatings for tubes," he says (Clin Chem. 2004;50:2398-2401). Dr. Hortin now believes collection tubes are much more complex devices than laboratorians have commonly appreciated in the past. And he says immunoassay interferences related to tubes are not a problem unique to BD. "It’s a potential problem with all tubes," he says.

"Test tubes are not totally inert materials that act as containers. They have substances in them that are serving as clot activators and surfactants that are affecting how the components interact with the surface wall. In addition, there are lubricants that go in the stopper to determine how well the vacuum is contained, and substances that can leach out of the gel in the gel tubes."

The lab community can take steps to reduce the probability that the interference problem will recur. First, Dr. Demers says, if manufacturers work on eliminating other sources of variability in immunoassays, overall bias can be reduced. "It would also be helpful if manufacturers would use state-of-the-art test tubes when establishing ranges for their assays, or perform their studies in a number of different tubes."

Proficiency testing could also help ensure that tube-related immunoassay interferences are minimized. "When CAP or other proficiency testing providers send out challenges," Dr. Fleisher says, "they could ask labs to tell them what kinds of tubes they are using. Tubes are part of the test system, and it’s really the same thing as asking what type of reagents or instruments are being used to perform a test."

Labs might also consider adding new types of statistical analyses to their protocols. "One of the devices we used in our analysis of the problem was a moving average," says NIH’s Dr. Bowen. "We were actually able to track our data from the past three years, and we noticed a shift in the actual mean of our total T₃ results, so labs can probably monitor shifts in their results by using a moving average of their patient data, assuming that patient distribution doesn’t change," he says.

Adds Dr. Remaley: "It’s an insensitive technique, but it was sensitive enough to pick up this problem. Proficiency tests are incapable of detecting this because you don’t draw the sample in those tubes, so I think the only way to do this currently is to look at your own results." Diagnostic companies will have to do their part in helping eliminate these interferences, he notes.

Dr. Blick’s suggestion would be what he’s going to do in his own laboratory: "We’re going to take our control material and add it to any future new lots of BD collection tubes and mix it in such a way that the sample comes into contact with the lubricant on the cap, clot activators, etc." Then they will centrifuge to expose it to the serum separator gel, and run the assay on that processed control material to look for changes.

For the moment, labs must rely on manufacturers to provide collection tubes that will enable accurate testing. Drs. Blick and Bowen have tested BD’s adjusted product, and both say the new tubes seem to be considerably improved. To help prevent future interference issues, Boselli says, BD has formed a cross-functional instrument company liaison group. The group’s focus is to establish close working relationships with instrument company partners at all levels of product development and post-market surveillance.

Still, a larger, perhaps more troubling, quality control worry remains. "Those of us in the laboratories have always felt," Dr. Fleisher says, "that once we have validated a test on a particular instrument using particular reagents, we were home safe—we knew everything about the test and the problems that existed. But now we have to worry about not only in what tube the sample is being collected, but under what conditions it is being collected."

This situation has uncovered "a weakness in our approach and a chink in our armor," Dr. Remaley sums up, "revealing that our current practices are imperfect for detecting this type of problem."

Sue Parham is a writer in Edgewater, Md. To read more about the affected products, read the technical bulletins released by BD on the following Web site: www.bd.com/vacutainer/techbulletins. BD is encouraging laboratorians who are seeking additional information or support to contact the company at Vacutainer_TechServices@bd.com or at 800-631-0174.