Research has shown that most errors in protein biomarker measurement occur in the preanalytical phase. This issue of Strategies explores a recent study that compared the effects of sample type and sample handling on 20 biomarkers.

Even as researchers have taken advantage of multiplex platforms to develop more sensitive and complex tests, they have found these tests to be particularly vulnerable to preanalytical errors. Especially in clinical trails, sample type and sample collection are significant concerns, as samples are often collected from more than one study center. Recently, researchers from Crescendo Bioscience, Inc. and Stanford University investigated these issues. They compared plasma versus serum samples as well as two sample handling procedures to study how these variables would affect both Crescendo Bioscience’s multi-biomarker disease activity (MBDA) test and a panel of eight autoantibody assays developed on the Meso Scale Discovery platform (J Immunol Methods 2012;378:72-80). The MBDA test, marketed as the VECTRA DA test, uses an algorithm based on 12 protein biomarkers to provide a measure of rheumatoid arthritis (RA) disease activity.

Understanding how preanalytical variables affect such assays is critical for interpreting clinical results as well as for developing new tests, according to study coauthor, P. Scott Eastman, PhD. “Unlike the pharmaceutical world, in diagnostics, during the research and development phases, we don’t usually run our own prospective trials,” he said. “This means diagnostic companies very often have to depend on various groups with sample cohorts that meet the requirements of what it is we would like to study, and we can’t control the kind of samples collected or how they are handled.” Eastman is director of assay discovery at Crescendo Bioscience in South San Francisco.

To compare serum versus plasma results, the researchers collected paired serum and plasma samples from 32 subjects with RA and performed both the MBDA and autoantibody assays. They also evaluated two serum handling methods—traditional shipping at ambient temperature without centrifugation, and a protocol involving centrifugation followed by cold shipping. For this assessment, matched serum samples were collected from 10 individuals with RA.

The results of the comparisons showed that the MBDA assay was much more sensitive to preanalytical effects than the autoantibody markers. For example, in the plasma versus serum experiment, the autoantibody markers showed less than 10% median difference between the two sample types. In contrast, only eight of the 12 MBDA protein biomarkers were highly correlated. The protein concentrations had a systematic shift toward higher serum concentrations. In addition, five out of 12 protein biomarkers had shifts >15% in the median percent difference in concentration across 32 patient samples.

Comparing traditional versus protocol serum sample handling methods, the autoantibody levels between the two were again highly correlated, with a median correlation coefficient of 0.99. However, the researchers found wide differences in the MBDA protein biomarkers. Only seven of the 12 biomarkers had correlation coefficients ≥0.95, with some as low as 0.05. Of the markers that showed significant differences by traditional sample handling methods, the values of seven increased and one decreased compared to the protocol method. For example, epidermal growth factor (EGF) and interleuken-6 (IL-6) serum concentrations increased up to 40-fold, while vascular endothelial growth factor A and resistin concentrations increased 2 to 4-fold.

The authors suggested that the marked differences in some of the protein immunoassays were due to prolonged exposure with blood cells at room temperature in the traditional sample handling group. They noted that this sample handling method had the greatest effect on proteins secreted by blood cells, such as EGF and IL-6. “Once the cells start to lyse, they release some of these cytokines,” Eastman said. “That is why not all of the biomarkers were affected. Some of them are not cellular, but are free in the serum already.”

The authors’ findings reflect a concern not only for test developers, but also for clinical labs, where sophisticated assays like those the investigators used often translate into more challenging preanalytical problems, according to Susan Maynard, PhD, who was not associated with the study. “Thirty years ago analytical precision was our biggest issue,” she said. “As instruments have become much more precise and consistent, I think our bigger problems now are often preanalytical issues, especially with immunoassays.” Maynard is the director of chemistry, toxicology, and blood gases at Carolinas Medical Center in Charlotte, N.C.

Eastman stressed that preanalytical problems associated with clinical trials have far-reaching implications. “Those of us developing assays really need to be able to access as many of these clinical trial cohorts as possible, but we need to make sure the sample collection is done right,” Eastman said. He explained that he was motivated to conduct the study based on experience. In the past, Crescendo Bioscience had not been able to use samples from several clinical trails because the samples had not been handled properly or they were not the appropriate kind of sample. In one case, Eastman found that all of the biomarkers he was studying were dramatically lower in the first group of samples he had received. “When we asked the investigator, we were told that all of these samples had been in the same freezer,” Eastman said. “So we concluded that at some point that freezer went down and the event was never recorded.” In other cases, trials have only been able to offer plasma samples rather than serum, he added.

Again, clinical labs face similar challenges, Maynard emphasized. “With samples from doctors’ offices and other locations, often they are not centrifuged immediately and kept cold, though they may arrive at the lab cold,” she said. “I think we all have in our policies the kind of sample handling we expect, but we know that reality does not always reflect that.”

The authors also pointed out that preanalytical considerations are not only important for future studies of assay development, but for correctly interpreting results from trials. “The results of this study illustrate the importance of characterizing preanalytical variability to ensure test accuracy for development, validation, and clinical testing with biomarker assays,” they wrote. “This is especially critical when these assays are integrated into large clinical trials, where using standardized serum processing and handling procedures would be an essential part of the study design, directly affecting result interpretation and next phase of trials.”