NEW YORK (360Dx) – Researchers at Cleveland HeartLab have developed a pair of proteomic tests for measuring cholesterol efflux capacity (CEC) and assessing patient risk of coronary artery disease (CAD).
The lab, which is owned by Quest Diagnostics, and its collaborators published a study last month in Clinical Chemistry detailing the development and analytical validation of the assays and is now evaluating their performance in additional patient cohorts, said Cory Bystrom, vice president of research and development at Cleveland HeartLab and senior author on the paper.
The tests emerged from an effort by Bystrom and his colleagues to develop a streamlined assay for measuring CEC, a measure of HDL function that previous research has linked to cardiovascular disease.
"People have been interested in HDL function as a potential marker of cardiovascular disease for some time now," Bystrom said. "Some very well-regarded studies have supported the observation of an inverse relationship between HDL-mediated cholesterol efflux and cardiovascular disease."
This suggests that CEC measurements could be used to assess patient risk of heart disease. These measurements, however, are typically made using complex and low-throughput cell-based assays that are not amenable to clinical implementation.
"The amount of labor to implement and clinically offer a cell-based assay is profound," Bystrom said.
Given that, the HeartLab researchers set out to build a mass spectrometry-based proteomic test using measurements of lipoprotein-associated proteins as a proxy for CEC.
They selected as candidates 21 proteins that previous studies had associated with either cardiovascular disease or some other aspect of HDL function or lipid biology. They developed multiple-reaction monitoring assays to these 21 proteins, which they then quantified in a set of 70 training samples. Using that data, they developed a five-protein panel for predicting CEC, which they then tested in a validation set consisting of an additional 35 serum samples. In the validation set, the mass spec test corresponded to the cell-based CEC measurements with a Spearman rank correlation coefficient of .86, indicating good correlation between the two.
Bystrom noted, however, that one challenge in assessing the correspondence of the two assays is the relative noisiness of the cell-based CEC measurement compared to the precision of the mass spec test.
"There is good agreement between predicted cholesterol efflux and the results from the cell-based assay" he said. "But the biological assay can be quite noisy, lending an advantage to a precise mass spectrometry-based measurement."
Having developed the mass spec CEC assay, the researchers then set out to determine whether their measurements would demonstrate the same relationship to cardiovascular disease as the cell-based test.
"There are two hurdles," Bystrom said. "First is building a model that allows you to predict cholesterol efflux. Second, you need to demonstrate that the predicted cholesterol efflux measured by mass spectrometry demonstrates the same clinical validity as foundational work using the cell-based assay."
The researchers evaluated the CEC test in a set of 154 CAD patients and 74 age- and sex-matched healthy controls, finding that it distinguished between the two with an area under the curve of .62. They then used logistic regression to reweight the five proteins in the panel to achieve a better separation of the two cohorts. This reweighted panel distinguished between the two groups with an AUC of .73 and a sensitivity of 71 percent and specificity of 76 percent.
Evaluating the reweighted panel in a separate set of 92 CAD cases and controls, the researchers found it performed with an AUC of .71.
They then evaluated the panels' analytical performance, establishing that coefficients of variation for the five proteins ranged from 5.2 percent to 14 percent. They tested for biological variation by running the assays in samples collected from 29 subjects once a week for eight weeks, finding, they wrote, that "the lipoprotein proteome, and therefore [the CEC and CAD score], for an individual is generally stable over eight weeks," indicating that tests were likely to pick up clinically relevant changes, as opposed to normal biological fluctuations.
After locking down the standard operating procedure for the two tests, the researchers evaluated them in a separate cohort obtained from the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) biobank, the rights to which are owned by UK-based diagnostics firm Abcodia. That cohort consisted of 69 postmenopausal women who suffered a myocardial infarction one to two years after sample collection and healthy controls. The researchers found that the CEC model cases had significantly lower CEC measurements than controls. They also found that the CAD test returned scores indicating a probability of CAD that was significantly higher for cases (.51) than controls (.37).
Bystrom said that in their initial validation work the researchers found evidence that the test might perform particularly well in women, which influenced their decision to test the assay in the UKCTOCS samples.
He noted that diagnosis of cardiovascular disease is often more difficult in women, making it an area of unmet clinical need. He declined to comment on whether Quest had any sort of commercial relationship around the test with Abcodia, which has partnered with a number of outside firms on biomarker discovery using the UKCTOCS samples.
"The performance of the assay at this point gives us the inspiration to continue working on it to understand and refine the clinical populations where it may be useful," he said, adding that a study with collaborators achieved positive results in another patient cohort."
Bystrom also suggested it could help further research into the relationship between CEC and cardiovascular disease, noting that "the high-throughput mass spec assay can readily support large clinical studies with less analytical friction than the cell-based assay."
In addition to its potential clinical utility, the assay is also an interesting example of mass spec-based proteomic test development and validation, said Irene van den Broek, a researcher at Cedars-Sinai Precision Biomarkers Laboratories whose research has focused in part on issues around implementation of mass spec-based protein assays.
Van den Broek, who was not involved in the HeartLab research, authored a commentary in Clinical Chemistry accompanying the study in which she noted that the assay provided "an example for the implementation of high-throughput multiprotein MS assays in the clinical laboratory."
While clinical mass spec has made strides in recent years, it is still not commonly used for protein tests, and multiplexed mass spec proteomic tests like that developed by Bystrom and his colleagues are particularly rare with only handful having made it to market.
In an interview, van den Broek said that while additional studies would be needed to validate the assays' performance and additional development would be needed to ensure that, for instance, the test was transferable across different laboratories, the study offers a useful and rigorous example of a clinical mass spec development process.
She added that she was particularly interested in the fact that the assay was multiplexed, measuring five proteins, and that it did not treat those measurements as separate but rather integrated them into single scores of CEC function or CAD risk.
"Most of the [proteomics] literature is focused doing discovery and finding markers and developing them into a targeted assay," she said. "But this seems to be something that has been done in a very structured way and already applied to a large clinical study."