- Email: [email protected]
Enzymatic determination of sodium and chloride in sweat
ClinicalBiochemisu'y,Vol. 29, No. 1, pp. 33-37, 1996 Copyright© 1996The CanadianSocietyof ClinicalChemists Printed in the USA. All rightsreserved 0009-9120/96 $15.00 + .00 ELSEVIER
0009-9120(95)02003-5
Enzymatic', Determination of Sodium and Chloride in Sweat RICHARD P. TAYLOR and TIMOTHY J. JAMES Department of Clinical Biochemistry, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK Objective: To develop methods based on enzyme activation for the analysis of sweat sodium and chloride using 13-galactosidase and a-amylase, respectively. Methods: Both were monitored kinetically on the Cobas Fara centrifugal analyzer. The sweat, collected with the Macroduct TM system, was diluted no more than five-fold for the volumes obtained of 16 to 80 p,L, median 32.5 p,L. The sodium assay utilized a sodium-binding cryptand to maximize linearity. Results: Between-run coefficients of variation (%) at 10, 20, and 50 mmol/L were 3.6, 4.5, and 1.3 for sodium and 7.1,6.1, and 6.0 for chloride, respectively. The sodium method showed excellent agreement with flame photometry (y = 0.997x + 0.742; r = 0.998), and chloride with a mercuric thiocyanate method (y = 0.995x + 0.485; r = 0.996), giving equivalent discrimination between patients with and without cystic fibrosis. Conclusions: The methods enable the rapid analysis on the same analyzer of both sodium and chloride in a single dilution of sweat collections of low volume.
K E Y WORDS: sweat; sodium; chloride; cystic fibrosis; enzyme activation; spectrophotometry; centrifugal analyzer; a m y l a s e ; galactosidase; cryptand.
Introduction
he measurement of sweat sodium and chloride T continues to be useful in the diagnosis of cystic fibrosis (CF) (1). Sweat sodium has traditionally been determined by flame photometry and chloride by coulometric, titrimetric, or colorimetric techniques (2). However, there continue to be some disadvantages associated with these methods (3). For example, if sodium is measured by flame photometry two analytical instruments are required and multiple dilutions of t:he sweat specimen must be made. Most titrimetric and colorimetric chloride methods have reagents containing mercury, which is an environmental toxin. They are also subject to matrix effects and m a y have limited working ranges. Ion-selective electrode (ISE) methods for
Correspondence: Dr. R.P. Taylor, D e p a r t m e n t of Clinical Biochemistry, Level 4, J o h n Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK. M a n u s c r i p t received: M a y 11, 1995; accepted: J u n e 1, 1995. CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
sweat sodium and chloride are also in use (2). ISEs for chloride and sodium may be available on the same analyzer for the analysis of serum, but generally require validation for sweat electrolyte analysis (2). In addition, the chloride electrode may have limited use for serum measurement and may be a relatively expensive approach to sweat analysis. We have developed methods for serum sodium and chloride based on the ion-dependent activation of enzyme activity for the analysis of sweat. They can be run on the same analyzer, are suitable for small volumes of sweat, and minimize manipulation of the specimen prior to analysis. Ion-dependent enzyme activity has been used for the automated analysis of the serum constituents sodium, potassium, chloride, and phosphate (4-7). The sodium assay utilizes the sodium-dependent activity of [3-galactosidase (EC 3.2.1.23), with cleavage of the substrate o-nitrophenyl-~-D-galactopyranoside (ONPG) to release o-nitrophenol: ONPG--(~-galactosidase/Na+)-. o-nitrophenol + galactose To achieve linearity over a range appropriate for diluted sweat (0 to 50 mmol/L sodium), a binding agent Kryptofix 221 was employed as used previously for serum (4) but re-optimization of the assay was necessary. For the analysis of chloride the ion-dependence of a-amylase is exploited. The principle of the assay is the reactivation by chloride of a-amylase that has been deactivated by the addition of EDTA to chelate calcium ion (6). The regeneration of activity is proportional to the chloride concentration. The reactivated s-amylase cleaves the substrate 2-chloro-4nitrophenyl-~,D-maltoheptaoside and, in the presence of c~-glucosidase and B-glucosidase, the product 2-chloronitrophenol is released and is measured at 405 nm. Materials and methods
Sweat was collected using the Wescor 3700-SYS Macroduct system (8) and sweat conductivity was TM
33
TAYLOR AND JAMES
measured with a Wescor Sweat-Chek conductivity analyzer (Chemlab Scientific Products Ltd, Hornchurch, Essex, UK). The enzymatic methods for sodium and chloride were developed on a Cobas Fara analyzer (Roche Diagnostics, Welwyn Garden City, Herts, UK). For chloride, results were compared with a mercuric thiocyanate method (Nycomed, Birmingham, UK) also determined on the Cobas Fara (9,10). Enzymatic sodium results were compared with flame photometry on a Corning 450 instrument (Coming Diagnostics, Halstead, Essex, UK). The binding agent Kryptofix 221 (4,7,13,16,21pentaoxa-l,10-diazabicyclo[8.8.5] tricosane) was purchased from Fluorochem (Old Glossop, Derbyshire, UK). Dithiothreitol, ~-galactosidase [Grade VI from E. coli, and prepared as recommended, (4)], ONPG and [ethylene bis (oxyethylenenitrilo)] tetra acetic acid (EGTA) were obtained from Sigma (Poole, Dorset, UK). The enzymatic chloride reagents were obtained as the serum chloride method from Boehringer Mannheim (Lewes, East Sussex, UK). Other reagents were of Analar grade from Merck (Poole, Dorset, UK). TM
REAGENT PREPARATION
Working standards of 5 mmol/L and 50 mmol/L sodium chloride were prepared from a 200 mmol/L stock solution. For sodium, the enzyme reagent (Reagent 1) contained: 450 mmol/L Tris buffer at various pH values, 6 mmol/L dithiothreitol, 11.25 mmol/L magnesium sulphate, 24 mmol/L lithium chloride, 0.5 mmol/L lithium EGTA, 1125 U/L ~-galactosidase, and various concentrations of Kryptofix 221. The substrate reagent (Start reagent) contained 45 mmol/L ONPG. This gave final reaction concentrations of: 300 mmol/L Tris, 4 mmol/L dithiothreitol, 7.5 mmol/L magnesium sulphate, 16.0 mmol/L lithium chloride, 0.33 mmol/L lithium EGTA, 750 U/L ~-galactosidase, and 1.5 mmol/L ONPG. For chloride, the enzyme reagent (Reagent 1) contained: 100 mmol/L 2-morpholino-ethanesulphonic acid (MES) buffer, pH 7, 30 mmol/L ethylenediaminetetraacetic acid (EDTA), 22.4 kU/L a-amylase, 89.6 kU/L a-glucosidase, and 89.6 kU/L ~-glucosidase. The substrate reagent (Start reagent) contained: 10 mmol/L MES buffer (pH 4), 30 mmol/L EDTA, 3 mmol/L calcium EDTA, and 1.5 mmol/L 2-chloro-4-nitrophenyl-~,D-malto-heptaoside. INSTRUMENT SETTINGS
For sodium, the sample (10 }~L) was added to 200 ~L of Reagent 1 and 40 }~L of water. After a 90-s incubation, 10 }xL of Start reagent and a further 40 }~L of water were added and the reaction rate monitored at 420 nm for 150 s. For assay development, 31 readings were taken at 5 s intervals. For chloride, the sample (2 ~L) and Reagent 1 (80 ~L) were mixed and incubated for 5 min. Start re-
34
agent (80 }xL)was added and the reaction was monitored at 405 nm for 490 s, taking 50 readings at 10-s intervals during assay development. Both methods were run at 37 °C and used the kinetic reaction mode and linear regression calibration mode functions of the analyzer. SPECIMEN COLLECTION AND PREPARATION
Sweat specimens were collected by the CF nurse using the Macroduct system (8). Immediately after collection, the sweat conductivity was measured with the Sweat-Chek conductivity analyzer and specimens were then either frozen until analysis or assayed without delay. The sweat was expelled from the plastic capillary tube into an autoanalyzer cup and immediately aspirated with a variable volume positive displacement pipette. The volume aspirated with the pipette was recorded. The sweat was dispensed into a Fara analyzer cup and diluted with water using the same pipette tip, typically dispensing twice the volume of sweat aspirated to obtain a three-fold dilution. The median volume collected was 32.5 ~L (range 16.0-80.0 ~L) and for collections of low volume, dilutions up to five-fold were made to obtain a minimum of 75 ~L of diluted sweat. Results on patient specimens were multiplied by the dilution factor to obtain the sweat concentration. For the method comparison, both assays for each analyte were performed on the same day, usually within 1 h. The two chloride assays were sampled from the same analytical cup. For sodium analysis by flame photometry, an aliquot (10 ~L) was taken with internal standard (10 ~L lithium nitrate, 150 mmol/L), diluted with water (2 mL) and aspirated directly into the flame photometer. TM
TM
Results ASSAY DEVELOPMENT
In the enzymatic sodium assay, ~-galactosidase activity is dependent on sodium concentration, pH, and the cryptand concentration (4). The dependence of assay sensitivity on Kryptofix 221 concentration and pH are displayed in Figures 1 and 2, respectively. The Kryptofix 221 concentration was varied at a fixed pH of 8.7, at which pH Kryptofix binds sodium effectively (Figure 1) (4). The absorbance decreases with increasing cryptand concentration. At 2.0 mmol/L, the response was linear for sodium concentrations up to 50 mmol/L and the absorbance change was sufficient to give adequate assay sensitivity. At this Kryptofix concentration, increasing the pH above 8.7 increased the absorbance change but there was some deviation from linearity (Figure 2). At lower pH, the loss of linearity was accompanied by a decreased absorbance change. Thus, at a pH of 8.7 and a Kryptofix 221 concentration of 2.0 mmol/L, the linear range extended from 0 to 50 mmol/L sodium, appropriate for actual sodium con-
CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
ENZYMATIC DETERMINATION OF SODIUM A N D CHLORIDE IN S W E A T A
4) C
E ,< <3
0.6
~
0.6
0.5
~
0.5
~
0,4
~
0.2
~
0.1
~
0,0
0.4
B
4) (.) t..(2 0 U) .(2 ,,<
0
10
20'
30'
40'
5'0
Sodium (retool/L) Figure 1 -- Enzymatic rate as a function of sodium concentration at pH 8.7 and different Kryptofix 221 concentrations in working reagent: [] zero; • 0.9 retool/L; • 2.0 mmol/L; © 2.5 retool/L; • 5.0 mmol/L. centrations up to 150 mmol/L in sweat diluted threefold. Absorbance measurements made at 5-s intervals demonstrated t h a t the sodium reaction showed no lag phase. The chloride reaction was monitored at 10-s intervals and was shown to have a linear course after an initial lag of 120 s (6). In routine use, the reaction rate was measured between 120 and 490 s. The absorbance of the 50 mmol/L calibrant with a sample volume of 2 ~L was 0.25 over this interval. The assay was observed to be linear between 0 and 100 mmol/L (observed C1- = 0.988 [expected C1-] + 0.874; r = 0.999). This covered a range well in excess of t h a t observed in diluted sweat specimens. Imprecision was assessed from 20 replicates on the same reaction rotor (within-run) and 20 replicates on different days ,:between run) of 10, 20, and 50 mmol/L sodium chloride, which would correspond to 30, 60, and 150 mmol/L when analyzing sweat diluted three-fold (Table 1). For sodium, within-run CVs were less t h a n 2% and between-run were 4.5% or less. The within-run CVs for chloride were 3.6% to 5.8% and between-run were 6.0% to 7.1%. Assay sensitivity was determined by measuring the absorbance in the assays of 20 replicates of dis-
0
10
20
30
Sodium
40
50
(mmol/L)
Figure 2 -- Enzymatic rate as a function of sodium concentration at 2.0 mmol/L Kryptofix 221 and different pH values: [] 8.0; • 8.3; • 8.7; O 9.0; • 9.3. tilled water. The sensitivities, defined as 3 standard deviations above the m e a n absorbance of water, were 0.97 mmol/L for sodium and 0.70 mmol/L for chloride. The potential for interference in the assays by other constituents of sweat was i n v e s t i g a t e d by their addition to water and to 10 and 50 mmol/L solutions of sodium chloride at concentrations of at least one third their concentration in sweat. In the sodium assay 50 mmol/L lactate and 5 mmol/L calcium chloride, m a g n e s i u m sulphate, a m m o n i u m chloride, and potassium bicarbonate gave less t h a n 1 mmol/L of apparent sodium in undiluted sweat. In the chloride assay, a 50 mmol/L l a c t a t e and 5 mmol/L magnesium sulphate and potassium bicarbonate gave less t h a n 1 mmol/L of apparent chloride in undiluted sweat. The dye erioglaucine sodium salt, used in the Macroduct sweat collector to monitor sweat volume, at 100 ~mol/L was also measured as less t h a n 1 mmol/L in both assays. Working sodium reagents were useable for at least 7 weeks stored in the dark at 4 °C. Over this period, the absorbance yield declined by 17%. Kryptofix 221 is light sensitive but showed no apparent deterioration when stored at room temperature and shielded from light for at least 4 months. The work-
TABLE 1 Imprecisionof Enzymatic Methods
Analyte Concentration Sodium (n = 20) Within-run Between-run Chloride (n = 20) Within-run Between-run
10 mmol/L
20 mmol/L
50 mmol/L
Mean
C V (%)
Mean
C V (%)
Mean
C V (%)
10.0 10.5
1.8 3.6
20.1 19.5
1.6 4.5
49.6 50.2
0.9 1.3
10.7 10.4
5.8 7.1
20.3 20.2
4.6 6.1
51.8 50.9
3.6 6.0
CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
35
TAYLOR AND JAMES
ing chloride reagents were useable for at least 4 weeks, over which period the absorbance yield declined by 30%. CLINICAL SPECIMENS
Fifty patients' sweat samples were analyzed for sodium with the enzymatic and flame photometer methods, with excellent agreement observed between the methods (Figure 3). The relationship was a n a l y z e d by D e m i n g r e g r e s s i o n (11). Twelve samples were from patients with diagnosed CF retested for this study; median flame sodium result = 92.2 mmol/L, range 68.1-138; median enzymatic result = 97.4 mmol/L, range 64.5-136. Thirty-eight samples were from non-CF patients; median flame sodium result = 16.4 mmol/L, range 4.0-37.5; median enzymatic result = 16.0 mmol/L, range 8.0-42.8. In the same patient specimens, the enzymatic chloride assay showed excellent agreement with the mercuric thiocyanate method (Deming regression; Figure 4). For the known CF patients the results were: median thiocyanate result 105 mmol/L, range 62.8-130; median enzymatic result 104 mmol/L, range 60.4-131 and, for the non-CF children, median thiocyanate result 12.3 mmol/L, range 6.0-25.0; median enzymatic result 12.0 mmol/L, range 6.3-27.0. Thus, in these patients the enzymatic sodium and chloride methods gave similar discriminatory information to the nonenzymatic methods between the patients with and without CF (Figure 5). The conductivity results were: CF patients: median 117, range 85-141; non-CF patients: median 36, range 24-48. The conductivity (y) correlated well with enzymatic sodium (linear regression equation ofy on x : y = 0.948x + 20.4, r = 0.971), enzymatic chloride (y = 0.846x + 27.2, r = 0.988), and enzymatic (sodium + chloride)(y = 0.452x + 23.7, r = 0.987).
150 0 !
125"
~,
100 -
O)
75
~
==
so
i
-
i
25
Chloride:
•
50
mercuric
i
,
75
i
100
•
i
,
125
thiocyanate
=
150
(mmol/L)
F i g u r e 4 - - Correlation between e n z y m a t i c a n d mercuric t h i o c y a n a t e chloride a s s a y s in s w e a t from 50 p a t i e n t s . Regression equation: e n z y m a t i c C l - = 0.995 colorimetric C1- + 0.485; 95% CI of slope 0.957 to 1.035; 95% CI of intercept - 2 . 1 5 2 to 3.122; r = 0.996.
Discussion
Enzyme activation assays have been described for the analysis of sodium and chloride in serum (4,6). We have demonstrated their use for the analysis of sodium and chloride in sweat. The enzymatic assays provide methods that are simple and convenient and applicable to a wide range of autoanalyzers (4,6). We have adapted the chloride assay to a Bayer RA1000 analyzer in addition to a Cobas Fara. The sodium and chloride methods can be performed from the same specimen cup and analyzer, thereby reducing manipulation errors. Sweat volumes in the order of 30 ~LL are sufficient, with a requirement for only a three-fold dilution. The small sample volumes of 10 ~L or less required in the assays enable duplicate or repeated analyses to be made if required. Both assays agreed well with existing methods and were linear in the required range for sweat diluted threeto five-fold. An additional dilution of specimen for flame photometry is not required. Imprecision of the 150 -
O
12+.
125 • A
~1
®~
75
0
100"
I
--
:
"
s.
•
5O
~ 0
i 0
25
Sodium:
i
i 50
flame
•
i 75
•
i
-
100
photometer
+
•
12S
i 1S0
(mmollL)
Figure 3 - - Correlation between enzymatic and flame photometric sodium assays in sweat from 50 patients. Regression equation: e n z y m a t i c N a + = 0.997 flame N a + + 0.742; 95% confidence i n t e r v a l (CI) of slope 0.971 to 1.024; 95% CI of i n t e r c e p t - 0 . 9 4 6 to 2.431; r = 0.998. 36
2s
C ~ 0
i i
8
n
o i
i i
I i
ES FS EC MC Figure 5 - - Discrimination between patients diagnosed
with CF (n = 12) and those without CF (n = 38). ES enzymatic sodium; FS flame-photometric sodium; EC enzymatic chloride; MC mercuric thiocyanate chloride. • CF patients; C) non-CF patients. CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
ENZYMATIC DETERMINATION OF SODIUM A N D CHLORIDE IN S W E A T
sodium assay was similar to t h a t reported for a sweat ISE method (12). The chloride assay imprecision was slightly greater than for an ISE method but, as only 2 ~L of specimen was assayed, it is likely t h a t imprecision could be improved by increasing the volume of diluted sweat in the assay. The Macroduct system allows the simple collection of sweat without ew~poration and with minimal manipulation after collection (8). However, the volume collected is less than for the traditional Gibson and Cooke method, which has a minimum acceptable collection in the order of 75 to 100 ~L (1,2,13). The Macroduct has a minimum acceptable volume in the order of 15 ~L (2). In one large series, a mean volume of 60 ~L was collected with the Macroduct system (14). Some autoanalyzers with ion-selective electrodes can analyze sodium and chloride simultaneously, b u t m a y use modifications of urine methods requiring large sample volumes (100 ~L) and up to 20-fold dilutions, giving a required working range below 10 mmol/L (12). In our series, the range of volumes collected was 16 to 80 ~L with a median of 32.5 ~L but this was sufficient for the enzymatic methods, which had effi~ctive working ranges up to at least 50 mmol/L in the assays. The enzymatic methods could simplify equipment requirements. The chloride method can also be used for serum chloride and therefore could obviate the need for a chloride ISE. Because flame photometry is often used only for sweat testing, the enzymatic sodium method could eliminate the need for a flame photometer. Chloride is the analyte of choice for sweat testing if only one is measured, an approach that appears to be commonly adopted (2). However, there are advantages in measuring both analytes. They can be valuable for providing an additional quality assurance check on the assays (2). Measuring both can also improve the diagnostic power of sweat testing, particularly when equivocal sodium and chloride concentrations are obtained, where inspection of the sum of the sodium and chloride and of their relative concentrations can imp:rove discrimination (13,15). The successful introduction of conductivity measurement as a first test on sweat has obviated the requirement to carry out further tests on a proportion of specimens deemed unequivocally normal (14). The conductivity m e a s u r e m e n t can be made very rapidly and has generated an expectation of rapid results while the patient is still in the clinic. In those cases w h e r e the conductivity result is equivocal, there is a need to minimize the delay in obtaining further results (13,15). The measurement of both sodium and chloride would provide additional information to give the most definitive results. The enzymatic sodium and chloride methods could fulfill these requirements easily and rapidly on instruments available in most laboratories. Results for both analytes can be obtained in 20 min on
CLINICAL BIOCHEMISTRY, V O L U M E
29, F E B R U A R Y 1996
the same analyzer. The methods could be suited to analyzers situated in outpatient clinics.
Acknowledgements W e thank Dr. A. Thomson and the cystic fibrosis specialist nurses in the Department of Paediatrics for their cooperation throughout this study. W e are grateful to Boehringer Mannheim U K for the provision of a chloride reagent kit.
References I. Gibson LE, Cooke RE. A test for concentration of electrolytes in sweat in cystic fibrosisof the pancreas utilizing pilocarpine by iontophoresis. Pediatrics 1959; 23: 545-9. 2. LeGrys VA, Burnett RW. Current status of sweat testing in North America. Arch Pathol Lab M e d 1994; 118: 865-7. 3. Tietz N W , Pruden EL, Siggaard-Andersen O. Electrolytes, blood gases and acid-base balance. In: Textbook of ClinicalChemistry.Pp. 1184-7. Philadelphia: W.B. Saunders Company, 1986. 4. Berry M N , Mazzachi RD, Pejakovic M, Peake MJ. Enzymatic determination of sodium in serum. Clin C h e m 1988; 34: 2295-8. 5. Kimura S, Asari S, Hayashi S, et al. N e w enzymatic method with tryptophanase for determining potassium in serum. Clin C h e m 1992; 38: 44-7. 6. Ono T, Taniguchi J, Mitsumaki H, et al. A new enzymatic assay of chloride in serum. Clin C h e m 1988; 34: 552-3. 7. Tedokon M, Suzuki K, Kayamori Y, Fujita S, Katayama Y. Enzymatic assay of inorganic phosphatewith use of sucrose phosphorylase and phosphoglucomutase. Clin C h e m 1992; 38: 512-5. 8. Cole DEC, Boucher MJ. Use of a new sample-collection device (Macroduct TM) in anion analysis of hum a n sweat. Clin C h e m 1986; 32: 1375-8. 9. Frey MJ. A quantitative colorimetric method for the determination of serum chloride using the Technicon R A 1000 system. Clin C h e m 1983; 29: 1255. 10. Taylor RP, Polliack AA, Bader DL. The analysis of metabolites in h u m a n sweat: analytical methods and the potential application to investigation of pressure ischaemia of soft tissues.A n n Clin Biochem 1994; 31:
18-24. 11. Strike PW. Medical laboratory statistics. Bristol: J Wright & Sons Ltd., 1981. 12. Barbour HM. Development and evaluation of the simultaneous determination of sweat sodium and chloride by ion-selective electrodes. A n n Clin Biochem 1991; 28: 150-4. 13. Green A, Dodds P, Pennock C. A study of sweat sodium and chloride; criteria for the diagnosis of cystic fibrosis. A n n Clin Biochem 1985; 22: 171-6. 14. Hammond KB, Turcios NL, Gibson LE. Clinical evaluation of the macroduct sweat collection system and conductivity analyzer in the diagnosis of cystic fibrosis.J Pediatr 1994; 124: 255-60. 15. Henderson MJ, Littlewood JM, Miller M. Interpretation of sweat sodium and chloride concentrations. A n n Clin Biochem 1986; 23:109 (letter).
37
0009-9120(95)02003-5
Enzymatic', Determination of Sodium and Chloride in Sweat RICHARD P. TAYLOR and TIMOTHY J. JAMES Department of Clinical Biochemistry, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK Objective: To develop methods based on enzyme activation for the analysis of sweat sodium and chloride using 13-galactosidase and a-amylase, respectively. Methods: Both were monitored kinetically on the Cobas Fara centrifugal analyzer. The sweat, collected with the Macroduct TM system, was diluted no more than five-fold for the volumes obtained of 16 to 80 p,L, median 32.5 p,L. The sodium assay utilized a sodium-binding cryptand to maximize linearity. Results: Between-run coefficients of variation (%) at 10, 20, and 50 mmol/L were 3.6, 4.5, and 1.3 for sodium and 7.1,6.1, and 6.0 for chloride, respectively. The sodium method showed excellent agreement with flame photometry (y = 0.997x + 0.742; r = 0.998), and chloride with a mercuric thiocyanate method (y = 0.995x + 0.485; r = 0.996), giving equivalent discrimination between patients with and without cystic fibrosis. Conclusions: The methods enable the rapid analysis on the same analyzer of both sodium and chloride in a single dilution of sweat collections of low volume.
K E Y WORDS: sweat; sodium; chloride; cystic fibrosis; enzyme activation; spectrophotometry; centrifugal analyzer; a m y l a s e ; galactosidase; cryptand.
Introduction
he measurement of sweat sodium and chloride T continues to be useful in the diagnosis of cystic fibrosis (CF) (1). Sweat sodium has traditionally been determined by flame photometry and chloride by coulometric, titrimetric, or colorimetric techniques (2). However, there continue to be some disadvantages associated with these methods (3). For example, if sodium is measured by flame photometry two analytical instruments are required and multiple dilutions of t:he sweat specimen must be made. Most titrimetric and colorimetric chloride methods have reagents containing mercury, which is an environmental toxin. They are also subject to matrix effects and m a y have limited working ranges. Ion-selective electrode (ISE) methods for
Correspondence: Dr. R.P. Taylor, D e p a r t m e n t of Clinical Biochemistry, Level 4, J o h n Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK. M a n u s c r i p t received: M a y 11, 1995; accepted: J u n e 1, 1995. CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
sweat sodium and chloride are also in use (2). ISEs for chloride and sodium may be available on the same analyzer for the analysis of serum, but generally require validation for sweat electrolyte analysis (2). In addition, the chloride electrode may have limited use for serum measurement and may be a relatively expensive approach to sweat analysis. We have developed methods for serum sodium and chloride based on the ion-dependent activation of enzyme activity for the analysis of sweat. They can be run on the same analyzer, are suitable for small volumes of sweat, and minimize manipulation of the specimen prior to analysis. Ion-dependent enzyme activity has been used for the automated analysis of the serum constituents sodium, potassium, chloride, and phosphate (4-7). The sodium assay utilizes the sodium-dependent activity of [3-galactosidase (EC 3.2.1.23), with cleavage of the substrate o-nitrophenyl-~-D-galactopyranoside (ONPG) to release o-nitrophenol: ONPG--(~-galactosidase/Na+)-. o-nitrophenol + galactose To achieve linearity over a range appropriate for diluted sweat (0 to 50 mmol/L sodium), a binding agent Kryptofix 221 was employed as used previously for serum (4) but re-optimization of the assay was necessary. For the analysis of chloride the ion-dependence of a-amylase is exploited. The principle of the assay is the reactivation by chloride of a-amylase that has been deactivated by the addition of EDTA to chelate calcium ion (6). The regeneration of activity is proportional to the chloride concentration. The reactivated s-amylase cleaves the substrate 2-chloro-4nitrophenyl-~,D-maltoheptaoside and, in the presence of c~-glucosidase and B-glucosidase, the product 2-chloronitrophenol is released and is measured at 405 nm. Materials and methods
Sweat was collected using the Wescor 3700-SYS Macroduct system (8) and sweat conductivity was TM
33
TAYLOR AND JAMES
measured with a Wescor Sweat-Chek conductivity analyzer (Chemlab Scientific Products Ltd, Hornchurch, Essex, UK). The enzymatic methods for sodium and chloride were developed on a Cobas Fara analyzer (Roche Diagnostics, Welwyn Garden City, Herts, UK). For chloride, results were compared with a mercuric thiocyanate method (Nycomed, Birmingham, UK) also determined on the Cobas Fara (9,10). Enzymatic sodium results were compared with flame photometry on a Corning 450 instrument (Coming Diagnostics, Halstead, Essex, UK). The binding agent Kryptofix 221 (4,7,13,16,21pentaoxa-l,10-diazabicyclo[8.8.5] tricosane) was purchased from Fluorochem (Old Glossop, Derbyshire, UK). Dithiothreitol, ~-galactosidase [Grade VI from E. coli, and prepared as recommended, (4)], ONPG and [ethylene bis (oxyethylenenitrilo)] tetra acetic acid (EGTA) were obtained from Sigma (Poole, Dorset, UK). The enzymatic chloride reagents were obtained as the serum chloride method from Boehringer Mannheim (Lewes, East Sussex, UK). Other reagents were of Analar grade from Merck (Poole, Dorset, UK). TM
REAGENT PREPARATION
Working standards of 5 mmol/L and 50 mmol/L sodium chloride were prepared from a 200 mmol/L stock solution. For sodium, the enzyme reagent (Reagent 1) contained: 450 mmol/L Tris buffer at various pH values, 6 mmol/L dithiothreitol, 11.25 mmol/L magnesium sulphate, 24 mmol/L lithium chloride, 0.5 mmol/L lithium EGTA, 1125 U/L ~-galactosidase, and various concentrations of Kryptofix 221. The substrate reagent (Start reagent) contained 45 mmol/L ONPG. This gave final reaction concentrations of: 300 mmol/L Tris, 4 mmol/L dithiothreitol, 7.5 mmol/L magnesium sulphate, 16.0 mmol/L lithium chloride, 0.33 mmol/L lithium EGTA, 750 U/L ~-galactosidase, and 1.5 mmol/L ONPG. For chloride, the enzyme reagent (Reagent 1) contained: 100 mmol/L 2-morpholino-ethanesulphonic acid (MES) buffer, pH 7, 30 mmol/L ethylenediaminetetraacetic acid (EDTA), 22.4 kU/L a-amylase, 89.6 kU/L a-glucosidase, and 89.6 kU/L ~-glucosidase. The substrate reagent (Start reagent) contained: 10 mmol/L MES buffer (pH 4), 30 mmol/L EDTA, 3 mmol/L calcium EDTA, and 1.5 mmol/L 2-chloro-4-nitrophenyl-~,D-malto-heptaoside. INSTRUMENT SETTINGS
For sodium, the sample (10 }~L) was added to 200 ~L of Reagent 1 and 40 }~L of water. After a 90-s incubation, 10 }xL of Start reagent and a further 40 }~L of water were added and the reaction rate monitored at 420 nm for 150 s. For assay development, 31 readings were taken at 5 s intervals. For chloride, the sample (2 ~L) and Reagent 1 (80 ~L) were mixed and incubated for 5 min. Start re-
34
agent (80 }xL)was added and the reaction was monitored at 405 nm for 490 s, taking 50 readings at 10-s intervals during assay development. Both methods were run at 37 °C and used the kinetic reaction mode and linear regression calibration mode functions of the analyzer. SPECIMEN COLLECTION AND PREPARATION
Sweat specimens were collected by the CF nurse using the Macroduct system (8). Immediately after collection, the sweat conductivity was measured with the Sweat-Chek conductivity analyzer and specimens were then either frozen until analysis or assayed without delay. The sweat was expelled from the plastic capillary tube into an autoanalyzer cup and immediately aspirated with a variable volume positive displacement pipette. The volume aspirated with the pipette was recorded. The sweat was dispensed into a Fara analyzer cup and diluted with water using the same pipette tip, typically dispensing twice the volume of sweat aspirated to obtain a three-fold dilution. The median volume collected was 32.5 ~L (range 16.0-80.0 ~L) and for collections of low volume, dilutions up to five-fold were made to obtain a minimum of 75 ~L of diluted sweat. Results on patient specimens were multiplied by the dilution factor to obtain the sweat concentration. For the method comparison, both assays for each analyte were performed on the same day, usually within 1 h. The two chloride assays were sampled from the same analytical cup. For sodium analysis by flame photometry, an aliquot (10 ~L) was taken with internal standard (10 ~L lithium nitrate, 150 mmol/L), diluted with water (2 mL) and aspirated directly into the flame photometer. TM
TM
Results ASSAY DEVELOPMENT
In the enzymatic sodium assay, ~-galactosidase activity is dependent on sodium concentration, pH, and the cryptand concentration (4). The dependence of assay sensitivity on Kryptofix 221 concentration and pH are displayed in Figures 1 and 2, respectively. The Kryptofix 221 concentration was varied at a fixed pH of 8.7, at which pH Kryptofix binds sodium effectively (Figure 1) (4). The absorbance decreases with increasing cryptand concentration. At 2.0 mmol/L, the response was linear for sodium concentrations up to 50 mmol/L and the absorbance change was sufficient to give adequate assay sensitivity. At this Kryptofix concentration, increasing the pH above 8.7 increased the absorbance change but there was some deviation from linearity (Figure 2). At lower pH, the loss of linearity was accompanied by a decreased absorbance change. Thus, at a pH of 8.7 and a Kryptofix 221 concentration of 2.0 mmol/L, the linear range extended from 0 to 50 mmol/L sodium, appropriate for actual sodium con-
CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
ENZYMATIC DETERMINATION OF SODIUM A N D CHLORIDE IN S W E A T A
4) C
E ,< <3
0.6
~
0.6
0.5
~
0.5
~
0,4
~
0.2
~
0.1
~
0,0
0.4
B
4) (.) t..(2 0 U) .(2 ,,<
0
10
20'
30'
40'
5'0
Sodium (retool/L) Figure 1 -- Enzymatic rate as a function of sodium concentration at pH 8.7 and different Kryptofix 221 concentrations in working reagent: [] zero; • 0.9 retool/L; • 2.0 mmol/L; © 2.5 retool/L; • 5.0 mmol/L. centrations up to 150 mmol/L in sweat diluted threefold. Absorbance measurements made at 5-s intervals demonstrated t h a t the sodium reaction showed no lag phase. The chloride reaction was monitored at 10-s intervals and was shown to have a linear course after an initial lag of 120 s (6). In routine use, the reaction rate was measured between 120 and 490 s. The absorbance of the 50 mmol/L calibrant with a sample volume of 2 ~L was 0.25 over this interval. The assay was observed to be linear between 0 and 100 mmol/L (observed C1- = 0.988 [expected C1-] + 0.874; r = 0.999). This covered a range well in excess of t h a t observed in diluted sweat specimens. Imprecision was assessed from 20 replicates on the same reaction rotor (within-run) and 20 replicates on different days ,:between run) of 10, 20, and 50 mmol/L sodium chloride, which would correspond to 30, 60, and 150 mmol/L when analyzing sweat diluted three-fold (Table 1). For sodium, within-run CVs were less t h a n 2% and between-run were 4.5% or less. The within-run CVs for chloride were 3.6% to 5.8% and between-run were 6.0% to 7.1%. Assay sensitivity was determined by measuring the absorbance in the assays of 20 replicates of dis-
0
10
20
30
Sodium
40
50
(mmol/L)
Figure 2 -- Enzymatic rate as a function of sodium concentration at 2.0 mmol/L Kryptofix 221 and different pH values: [] 8.0; • 8.3; • 8.7; O 9.0; • 9.3. tilled water. The sensitivities, defined as 3 standard deviations above the m e a n absorbance of water, were 0.97 mmol/L for sodium and 0.70 mmol/L for chloride. The potential for interference in the assays by other constituents of sweat was i n v e s t i g a t e d by their addition to water and to 10 and 50 mmol/L solutions of sodium chloride at concentrations of at least one third their concentration in sweat. In the sodium assay 50 mmol/L lactate and 5 mmol/L calcium chloride, m a g n e s i u m sulphate, a m m o n i u m chloride, and potassium bicarbonate gave less t h a n 1 mmol/L of apparent sodium in undiluted sweat. In the chloride assay, a 50 mmol/L l a c t a t e and 5 mmol/L magnesium sulphate and potassium bicarbonate gave less t h a n 1 mmol/L of apparent chloride in undiluted sweat. The dye erioglaucine sodium salt, used in the Macroduct sweat collector to monitor sweat volume, at 100 ~mol/L was also measured as less t h a n 1 mmol/L in both assays. Working sodium reagents were useable for at least 7 weeks stored in the dark at 4 °C. Over this period, the absorbance yield declined by 17%. Kryptofix 221 is light sensitive but showed no apparent deterioration when stored at room temperature and shielded from light for at least 4 months. The work-
TABLE 1 Imprecisionof Enzymatic Methods
Analyte Concentration Sodium (n = 20) Within-run Between-run Chloride (n = 20) Within-run Between-run
10 mmol/L
20 mmol/L
50 mmol/L
Mean
C V (%)
Mean
C V (%)
Mean
C V (%)
10.0 10.5
1.8 3.6
20.1 19.5
1.6 4.5
49.6 50.2
0.9 1.3
10.7 10.4
5.8 7.1
20.3 20.2
4.6 6.1
51.8 50.9
3.6 6.0
CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
35
TAYLOR AND JAMES
ing chloride reagents were useable for at least 4 weeks, over which period the absorbance yield declined by 30%. CLINICAL SPECIMENS
Fifty patients' sweat samples were analyzed for sodium with the enzymatic and flame photometer methods, with excellent agreement observed between the methods (Figure 3). The relationship was a n a l y z e d by D e m i n g r e g r e s s i o n (11). Twelve samples were from patients with diagnosed CF retested for this study; median flame sodium result = 92.2 mmol/L, range 68.1-138; median enzymatic result = 97.4 mmol/L, range 64.5-136. Thirty-eight samples were from non-CF patients; median flame sodium result = 16.4 mmol/L, range 4.0-37.5; median enzymatic result = 16.0 mmol/L, range 8.0-42.8. In the same patient specimens, the enzymatic chloride assay showed excellent agreement with the mercuric thiocyanate method (Deming regression; Figure 4). For the known CF patients the results were: median thiocyanate result 105 mmol/L, range 62.8-130; median enzymatic result 104 mmol/L, range 60.4-131 and, for the non-CF children, median thiocyanate result 12.3 mmol/L, range 6.0-25.0; median enzymatic result 12.0 mmol/L, range 6.3-27.0. Thus, in these patients the enzymatic sodium and chloride methods gave similar discriminatory information to the nonenzymatic methods between the patients with and without CF (Figure 5). The conductivity results were: CF patients: median 117, range 85-141; non-CF patients: median 36, range 24-48. The conductivity (y) correlated well with enzymatic sodium (linear regression equation ofy on x : y = 0.948x + 20.4, r = 0.971), enzymatic chloride (y = 0.846x + 27.2, r = 0.988), and enzymatic (sodium + chloride)(y = 0.452x + 23.7, r = 0.987).
150 0 !
125"
~,
100 -
O)
75
~
==
so
i
-
i
25
Chloride:
•
50
mercuric
i
,
75
i
100
•
i
,
125
thiocyanate
=
150
(mmol/L)
F i g u r e 4 - - Correlation between e n z y m a t i c a n d mercuric t h i o c y a n a t e chloride a s s a y s in s w e a t from 50 p a t i e n t s . Regression equation: e n z y m a t i c C l - = 0.995 colorimetric C1- + 0.485; 95% CI of slope 0.957 to 1.035; 95% CI of intercept - 2 . 1 5 2 to 3.122; r = 0.996.
Discussion
Enzyme activation assays have been described for the analysis of sodium and chloride in serum (4,6). We have demonstrated their use for the analysis of sodium and chloride in sweat. The enzymatic assays provide methods that are simple and convenient and applicable to a wide range of autoanalyzers (4,6). We have adapted the chloride assay to a Bayer RA1000 analyzer in addition to a Cobas Fara. The sodium and chloride methods can be performed from the same specimen cup and analyzer, thereby reducing manipulation errors. Sweat volumes in the order of 30 ~LL are sufficient, with a requirement for only a three-fold dilution. The small sample volumes of 10 ~L or less required in the assays enable duplicate or repeated analyses to be made if required. Both assays agreed well with existing methods and were linear in the required range for sweat diluted threeto five-fold. An additional dilution of specimen for flame photometry is not required. Imprecision of the 150 -
O
12+.
125 • A
~1
®~
75
0
100"
I
--
:
"
s.
•
5O
~ 0
i 0
25
Sodium:
i
i 50
flame
•
i 75
•
i
-
100
photometer
+
•
12S
i 1S0
(mmollL)
Figure 3 - - Correlation between enzymatic and flame photometric sodium assays in sweat from 50 patients. Regression equation: e n z y m a t i c N a + = 0.997 flame N a + + 0.742; 95% confidence i n t e r v a l (CI) of slope 0.971 to 1.024; 95% CI of i n t e r c e p t - 0 . 9 4 6 to 2.431; r = 0.998. 36
2s
C ~ 0
i i
8
n
o i
i i
I i
ES FS EC MC Figure 5 - - Discrimination between patients diagnosed
with CF (n = 12) and those without CF (n = 38). ES enzymatic sodium; FS flame-photometric sodium; EC enzymatic chloride; MC mercuric thiocyanate chloride. • CF patients; C) non-CF patients. CLINICAL BIOCHEMISTRY, VOLUME 29, FEBRUARY 1996
ENZYMATIC DETERMINATION OF SODIUM A N D CHLORIDE IN S W E A T
sodium assay was similar to t h a t reported for a sweat ISE method (12). The chloride assay imprecision was slightly greater than for an ISE method but, as only 2 ~L of specimen was assayed, it is likely t h a t imprecision could be improved by increasing the volume of diluted sweat in the assay. The Macroduct system allows the simple collection of sweat without ew~poration and with minimal manipulation after collection (8). However, the volume collected is less than for the traditional Gibson and Cooke method, which has a minimum acceptable collection in the order of 75 to 100 ~L (1,2,13). The Macroduct has a minimum acceptable volume in the order of 15 ~L (2). In one large series, a mean volume of 60 ~L was collected with the Macroduct system (14). Some autoanalyzers with ion-selective electrodes can analyze sodium and chloride simultaneously, b u t m a y use modifications of urine methods requiring large sample volumes (100 ~L) and up to 20-fold dilutions, giving a required working range below 10 mmol/L (12). In our series, the range of volumes collected was 16 to 80 ~L with a median of 32.5 ~L but this was sufficient for the enzymatic methods, which had effi~ctive working ranges up to at least 50 mmol/L in the assays. The enzymatic methods could simplify equipment requirements. The chloride method can also be used for serum chloride and therefore could obviate the need for a chloride ISE. Because flame photometry is often used only for sweat testing, the enzymatic sodium method could eliminate the need for a flame photometer. Chloride is the analyte of choice for sweat testing if only one is measured, an approach that appears to be commonly adopted (2). However, there are advantages in measuring both analytes. They can be valuable for providing an additional quality assurance check on the assays (2). Measuring both can also improve the diagnostic power of sweat testing, particularly when equivocal sodium and chloride concentrations are obtained, where inspection of the sum of the sodium and chloride and of their relative concentrations can imp:rove discrimination (13,15). The successful introduction of conductivity measurement as a first test on sweat has obviated the requirement to carry out further tests on a proportion of specimens deemed unequivocally normal (14). The conductivity m e a s u r e m e n t can be made very rapidly and has generated an expectation of rapid results while the patient is still in the clinic. In those cases w h e r e the conductivity result is equivocal, there is a need to minimize the delay in obtaining further results (13,15). The measurement of both sodium and chloride would provide additional information to give the most definitive results. The enzymatic sodium and chloride methods could fulfill these requirements easily and rapidly on instruments available in most laboratories. Results for both analytes can be obtained in 20 min on
CLINICAL BIOCHEMISTRY, V O L U M E
29, F E B R U A R Y 1996
the same analyzer. The methods could be suited to analyzers situated in outpatient clinics.
Acknowledgements W e thank Dr. A. Thomson and the cystic fibrosis specialist nurses in the Department of Paediatrics for their cooperation throughout this study. W e are grateful to Boehringer Mannheim U K for the provision of a chloride reagent kit.
References I. Gibson LE, Cooke RE. A test for concentration of electrolytes in sweat in cystic fibrosisof the pancreas utilizing pilocarpine by iontophoresis. Pediatrics 1959; 23: 545-9. 2. LeGrys VA, Burnett RW. Current status of sweat testing in North America. Arch Pathol Lab M e d 1994; 118: 865-7. 3. Tietz N W , Pruden EL, Siggaard-Andersen O. Electrolytes, blood gases and acid-base balance. In: Textbook of ClinicalChemistry.Pp. 1184-7. Philadelphia: W.B. Saunders Company, 1986. 4. Berry M N , Mazzachi RD, Pejakovic M, Peake MJ. Enzymatic determination of sodium in serum. Clin C h e m 1988; 34: 2295-8. 5. Kimura S, Asari S, Hayashi S, et al. N e w enzymatic method with tryptophanase for determining potassium in serum. Clin C h e m 1992; 38: 44-7. 6. Ono T, Taniguchi J, Mitsumaki H, et al. A new enzymatic assay of chloride in serum. Clin C h e m 1988; 34: 552-3. 7. Tedokon M, Suzuki K, Kayamori Y, Fujita S, Katayama Y. Enzymatic assay of inorganic phosphatewith use of sucrose phosphorylase and phosphoglucomutase. Clin C h e m 1992; 38: 512-5. 8. Cole DEC, Boucher MJ. Use of a new sample-collection device (Macroduct TM) in anion analysis of hum a n sweat. Clin C h e m 1986; 32: 1375-8. 9. Frey MJ. A quantitative colorimetric method for the determination of serum chloride using the Technicon R A 1000 system. Clin C h e m 1983; 29: 1255. 10. Taylor RP, Polliack AA, Bader DL. The analysis of metabolites in h u m a n sweat: analytical methods and the potential application to investigation of pressure ischaemia of soft tissues.A n n Clin Biochem 1994; 31:
18-24. 11. Strike PW. Medical laboratory statistics. Bristol: J Wright & Sons Ltd., 1981. 12. Barbour HM. Development and evaluation of the simultaneous determination of sweat sodium and chloride by ion-selective electrodes. A n n Clin Biochem 1991; 28: 150-4. 13. Green A, Dodds P, Pennock C. A study of sweat sodium and chloride; criteria for the diagnosis of cystic fibrosis. A n n Clin Biochem 1985; 22: 171-6. 14. Hammond KB, Turcios NL, Gibson LE. Clinical evaluation of the macroduct sweat collection system and conductivity analyzer in the diagnosis of cystic fibrosis.J Pediatr 1994; 124: 255-60. 15. Henderson MJ, Littlewood JM, Miller M. Interpretation of sweat sodium and chloride concentrations. A n n Clin Biochem 1986; 23:109 (letter).
37