(T1230-06-39) Prediction of Hematocrit through Potassium Measurement in Whole Blood Using Volumetric Absorptive Microsampling

Purpose: Hematocrit is an essential value to be used for evaluating various clinical conditions in patient care and drug research. The conventional method of measuring hematocrit through venipuncture is invasive and requires a phlebotomist. In situations that venous blood sampling is not favorable or possible, an alternative method is needed to determine hematocrit. The measurement of potassium concentrations in blood using a volumetric absorptive microsampling (VAMS) method may be useful in estimating hematocrit without venipuncture. This study was performed to predict hematocrit values accurately and precisely through potassium measurement in VAMS samples from pancreatic cancer patients.

Methods: VAMS samples were collected from the fingertips of pancreatic cancer patients who visited the outpatient cancer clinic. Potassium was extracted from the whole blood in VAMS tips using an aqueous solution of potassium chloride 1.6 mEq/L with vortexing at 1,200 rpm for 30 minutes. The potassium concentrations in the extracts were measured using a chemistry analyzer equipped with an Ion-Selective Electrode module. The values of hematocrit (Hct_pred) were predicted from the potassium concentrations using a previously reported equation of linear regression: Hematocrit = 18.7 × Potassium concentration - 2.8 [1]. Hct_pred values were compared with the measured values of hematocrit (Hct_meas) on the same day of VAMS collection by the clinical chemistry laboratory. The normality of Hct_pred and Hct_meas values was assessed using Q-Q plots and Shapiro-Wilk tests. The proportional and constant biases between Hct_pred and Hct_meas were evaluated from the slope and intercept of a Deming regression line. The mean bias and the corresponding 95% limit of agreement (LoA) between Hct_pred and Hct_meas were determined using a Bland-Altman plot of the differences against their mean values. A mountain plot was used to identify the median and central 95% of the differences between Hct_pred and Hct_meas.

Results: A total of 31 evaluable VAMS samples obtained from 21 participants were used in the analysis. The potassium concentrations of the VAMS samples ranged from 3.07 to 3.89 mmol/L. The Hct_pred values ranged from 24.3% to 43.3% with a mean ± standard deviation (SD) of 34.3% ± 4.6%, whereas the Hct_meas values ranged from 24.8% to 45.8% with a mean ± SD of 35.3% ± 4.8%. The Q-Q plots and Shapiro-Wilk tests demonstrated a normal distribution of the Hct_pred and Hct_meas values without significant deviation. The equation of Deming regression line for Hct_pred and Hct_meas was: Hct_meas = 1.04 × Hct_pred - 0.26. The proportional or constant bias between Hct_pred and Hct_meas was very small, as the slope and intercept included 1 and 0 with their 95% confidence intervals of 0.77 to 1.31 and -9.35 to 8.83, respectively. The Bland-Altman plot displayed a random spread of data points with no recognizable pattern. The mean bias ± SD between Hct_pred and Hct_meas was -0.99% ± 2.73%, indicating the high accuracy of the hematocrit prediction method as the mean bias was very close to 0. All differences between Hct_pred and Hct_meas were within the 95% LoA from -6.34% to 4.36%, demonstrating reasonable precision. The mountain plot depicted an approximately symmetrical distribution of the differences between Hct_pred and Hct_meas centered at -0.99%. Additionally, the central 95% of the differences lay between -4.88% to 2.99%, demonstrating the high precision of this prediction method.

Conclusion: Potassium concentrations measured in whole blood using VAMS provided an accurate and precise prediction of hematocrit values measured in a clinical chemistry laboratory. Such method appears to be acceptable in predicting hematocrit values with minimally invasive blood sampling and may be a useful alternative to venipuncture for measuring hematocrit in a situation that venipuncture is not favorable or possible.

References: [1] Capiau S, Stove C. Hematocrit prediction in volumetric absorptive microsamples. J Pharm Biomed Anal. 2020;190:113491. doi:10.1016/j.jpba.2020.113491

Acknowledgements: This study was approved by the Institutional Review Board (IRB) of Seoul National University Hospital (IRB number: H-2103-093-1205). This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2021R1F1A104996113). This work was also supported by Research Institute of Pharmaceutical Sciences and the four phase of the Brain Korea 21 Program in 2023.