SelectivIty Factors, Detection Limit, Slope of the Electrode Function and Membrane Resistance of K-seIective Solvent Polymeric Membranes

Various ion-selective electrodes based on valinomycin (Table 1) have been widely used for determination of the K + concentration in whole blood, blood plasma, and undiluted and diluted blood serum, as well as in diluted urine [for a review see (1)]. Plasticized poly(vinylchloride) membranes (columns 2 and 3 in Table 1) have been of special interest. Although these electrodes, especially the one plasticized with dinonyladipate (I), exhibita very high stabilityofthe electromotiveforce of the cell assembly and lead to reliable determination of [KI in these media, they cannot be used for undiluted urine. This shortcoming is due to anion interference (2), which can be eliminated almost completely (2) by incorporating valinomycin into Silopren (silicone rubber, column 4 in Table 1). Such membranes are now in routine use for the bedside monitoring of undiluted urine (3). Here we report on the use of these membranes to measure [K ] in undiluted blood serum. To test their universal applicability to body fluids, including serum, we prepared silicone rubber membranes and miniaturized electrodes as described earlier (2). Using the same cell assembly and the same calibration and evaluation procedure as described recently for [Na”] determination in serum (4), we correlated measurements made with use of the silicone rubber membrane electrode and a flame photometer (Figure 1). The three solutions for the electrode calibration all contained, per liter, 140 mmol of NaCl, 1.1 mmol of CaCl2, and 0.6 mmol of MgCl2. The respective concentrations of KC1 were 1, 4, and 9 mmol/L. Results obtained by direct potentiometry agreed well with the indirect measurements made by flame photometry (upper curve in Figure 1). The correlation yields a residual standard deviation of ±0.10 mmollL and a positive bias of the potentiometrically determined potassium concentrations with a mean of 2.2%. Correcting the values for the protein and lipid volume according to Waugh (4, 5) shifts the bias to -4.3% (lower curve in Figure 1, residual standard deviation: ±0.10 mmol/L). No bias (6) or only a positive one varying from 0.7% (7) to 4.0% (8) has been found in different potentiometric studies of FK”’I in serum, if no corrections are made for protein and lipid volume. Correcting the results for typical (6.7%) protein and lipid volumes (4), we obtained negative biases of Fig. 1. Correlations of [K’] measurements in serum samples obtained by direct potentiometry (ISE), and by flame photometry [uncorrected (0); corrected (#{149}) for protein and lipid volumes, see 4, 5]. A linear regression with the uncorrectedresultsgives theuppersolidline,i.e.,y= -0.12(SDO.12)+ 1.05 (SD O.02)x,where y represents the potentiometric measurements and x flame photometric. The residual standard deviation of yonxis ±O.lOmmoftover the givenrange.Linearregression withthe corrected results gives the iowersolid line. i.e.,y = -0.10 (SD 0.11) + 0.98 (SD 0.02)x. The residual standard deviation of y on xis ±0.10 mmol/L.