Comparison of serum, EDTA plasma and P100 plasma for luminex-based biomarker multiplex assays in patients with chronic obstructive pulmonary disease in the SPIROMICS study

Abstract

Background: As a part of the longitudinal Chronic Obstructive Pulmonary Disease (COPD) study, Subpopulations and Intermediate Outcome Measures in COPD study (SPIROMICS), blood samples are being collected from 3200 subjects with the goal of identifying blood biomarkers for sub-phenotyping patients and predicting disease progression. To determine the most reliable sample type for measuring specific blood analytes in the cohort, a pilot study was performed from a subset of 24 subjects comparing serum, Ethylenediaminetetraacetic acid (EDTA) plasma, and EDTA plasma with proteinase inhibitors (P100).

Methods: 105 analytes, chosen for potential relevance to COPD, arranged in 12 multiplex and one simplex platform (Myriad-RBM) were evaluated in duplicate from the three sample types from 24 subjects. The reliability coefficient and the coefficient of variation (CV) were calculated. The performance of each analyte and mean analyte levels were evaluated across sample types.

Results: 20% of analytes were not consistently detectable in any sample type. Higher reliability and/or smaller CV were determined for 12 analytes in EDTA plasma compared to serum, and for 11 analytes in serum compared to EDTA plasma. While reliability measures were similar for EDTA plasma and P100 plasma for a majority of analytes, CV was modestly increased in P100 plasma for eight analytes. Each analyte within a multiplex produced independent measurement characteristics, complicating selection of sample type for individual multiplexes.

Conclusions: There were notable detectability and measurability differences between serum and plasma. Multiplexing may not be ideal if large reliability differences exist across analytes measured within the multiplex, especially if values differ based on sample type. For some analytes, the large CV should be considered during experimental design, and the use of duplicate and/or triplicate samples may be necessary. These results should prove useful for studies evaluating selection of samples for evaluation of potential blood biomarkers.

Figures

Figure 1

Scatterplots of coefficient of variation (CV) and reliability score for all consistently detectible analytes for 24 subjects. Coefficient of variation (CV; left panel) and reliability (right panel) are plotted as shown for either serum and EDTA plasma (top row) or P100 and EDTA plasma (bottom row). Outliers in these figures with either CV > 20% or reliability < 0.60 are discussed further in Table 1.

Figure 2

Plots indicating analytes with notable differences in CV or reliability between serum and EDTA plasma for 24 subjects. A) The ratio of CV (serum:EDTA plasma) is plotted in rank order from largest to smallest by analyte. Analytes with notably better CV in EDTA plasma (ratio >1.5; lower CV in plasma) and notably better CV in serum (CV ratio <0.667; lower CV in serum) are indicated B) The difference in reliability score between serum and plasma (serum minus plasma) is is plotted in rank order from largest to smallest by analyte. Analytes with notably better reliability in serum versus plasma (difference > +0.15) and better reliability in EDTA plasma versus serum (difference < −0.015) are indicated. Horizontal lines indicate descriptive cut-points used to define notable performance.

Figure 3

Plot indicating analytes with notable differences in mean levels between serum and EDTA plasma for 24 subjects. The ratio of measured mean analyte levels (serum:EDTA plasma) is plotted in rank order from largest to smallest by analyte. Analytes with notably higher levels measured in serum (ratio >1.5) and EDTA plasma (< 0.667) are indicated. Horizontal lines indicate descriptive cut-points used to define notable differences.