Inferring blood volume collected on plasma-separating dried blood spots using mass and image analysis

BackgroundPlasma-separating dried blood spots (psDBS) allow measurement of cancer biomarkers such as cell-free DNA concentration, enabling distributed sample collection, as well as ambient storage and shipment. However, unlike volumetric blood tube collections, blood volume collected using psDBS can vary between individuals. We evaluated whether analysis of psDBS mass, area, and color can be used to infer blood volume, and enable downstream measurement of circulating analyte concentrations. MethodsWhole blood from healthy individuals and cancer patients was collected and used to prepare psDBS with volumes of 50 to 250 {micro}L. Total area and mean grey value (MGV) of the erythrocyte region were measured using image segmentation. We developed a linear regression model to predict whole blood volume using mass, area, and MGV. DNA from psDBS was extracted, then quantified using a quantitative PCR assay targeting L1PA2. ResultsWe analyzed 261 psDBS samples from 43 healthy individuals and 118 cancer patients. Using 110 samples from 10 healthy individuals, the linear regression model showed a strong correlation between actual and predicted blood volume (Pearson r = 0.97, RMSE = 5.1 {micro}L). Using 151 samples from 33 separate healthy individuals and 118 samples from cancer patients, the model predicted blood volume with high accuracy (RMSE = 15.1 {micro}L and 13.3 {micro}L, respectively). Plasma DNA concentration in psDBS was moderately correlated with matched blood tubes (Spearman rho = 0.62). ConclusionOur results show estimating blood volume collected on psDBS is feasible using mass and image analysis. This approach enables calculation of circulating concentration for plasma analytes measured using psDBS.