- Email: [email protected]
Clinical evaluation of the macroduct sweat collection system and conductivity analyzer in the diagnosis of cystic fibrosis
Clinical evaluation of the macroduct sweat collection system and conductivity analyzer in the diagnosis of cystic fibrosis Keith B. H a m m o n d , MS, Nelson L. Turcios, MD, a n d Lewis E. Gibson, MD From the Departments of Pediatrics and Pathology, Divisionof Clinical Pathology and Laboratory Medicine, Universityof Colorado School of Medicine, Denver, the Department of Pediatrics, New Jersey Medical School, Newark, and the Department of Pediatrics, Loyola University Medical Center, Maywood, Illinois
The purposes of this study were to c o m p a r e sweat tests used in diagnosing cystic fibrosis (CF), as performed with the Macroduct c o l l e c t i o n system, with those utilizing the more laborious quantitative pilocarpine iontophoresis test (QPIT), and to ascertain the e f f i c a c y of the Sweat-Chek c o n d u c t i v i t y analyzer in eliminating some possibly unnecessary chloride analyses. A Macroduct sweat test was performed on one arm and a QPIT on the other on 1090 patients, 93 of whom had CF. Of these, 514 patients (43 with CF) also had a c o n d u c t i v i t y determination on the Macroduct sweat sample. All subjects were referred to the laboratory of one of us (K.B.H.) for sweat testing. Of the QPII" samples, 0.7% were inadequate, as were 6.1% of those from the Macroduct system. When sodium and chloride concentrations from the two tests were compared, the standard errors of the estimate were 3.90 and 3.85, respectively. Agreement within 8 mEq/L could then be e x p e c t e d with 95% c o n f i d e n c e limits. With use of the Sweat-Chek analyzer, no patient with CF was found to have a c o n d u c t i v i t y of less than 90 mmol/L, whereas 430 (91%) of the non-CF subjects had a c o n d u c t i v i t y of less than 50 mmol/L. None of those 430 subjects had a sweat chloride value >32 mmol/L. We c o n c l u d e that the Macroduct c o l l e c t i o n system provides results e q u a l l y as satisfactory as those provided by the QPIT and that the Sweat-Chek analyzer frequently eliminates the necessity of measuring chloride concentrations. (J PEDIATR1994;124:255-60.)
Early diagnosis of cystic fibrosis necessitates performing many negative sweat tests on children without CF to diagnose the disease in a few. The quantitative pilocarpine ionSupported by Wescor Inc., in the donation of the Sweat-Chek analyzer to our laboratory. All other equipment and consumables from Wescor were purchased in the normal way. Submitted for publication April 30, 1993; accepted Sept. 3, 1993. Reprint requests: Lewis E. Gibson, MD, Loyola University Medical Center, 2160 South First Ave., Maywood, IL 60153. Copyright © 1994 by Mosby-Year Book, Inc. 0022-3476/94 $3.00 + 0 9/20/51264
tophoresis test,* when performed correctly, 1 is very accurate, but time and skill are required to prevent evaporation during collection, to determine sweat weight with a chemical balance, and to calculate the chemical composition of the eluted sample. To simplify the test, many laboratories have adopted alternative methods that have inherent problems. 24 These problems may seem small, but their importance becomes apparent when one patient receives an inaccurate diagnosis. *Used here to mean that sweat is quantified by weighing a collection pad, as distinct from Macroduct collection system. Sweat collected in the Macroduct system may be quantified with a micropipet.
255
256
Hammond, Turcios, and Gibson
The Journal of Pediatrics February 1994
Table I. Average electrolyte concentrations in sweat collected simultaneously by two methods QPIT Na +
Non-CF patients* Average 18.2 SD 11.8 SE 0.4 Range 3-77 Patients with CFt Average 87.9 SD 16.9 SE 1.8 Range 35-131
Macroduct CI-
Na +
Cl
15.9 8.7 0.3 5-65
18.4 12.3 0.4 3-85
16.1 9.2 0.3 3-69
100.8 12.6 1.3 46-124
86.1 16.1 1.7 43-123
100.5 11.9 1.3 54-124
Values are expressed as millimolesper liter. *Sodiumconcentrationswere measured in 934 non-CFpatients and chloride concentrations in 926. ?Sodium concentrationsweremeasured in 88 patients with CF and chloride concentrations in 88.
The Macroduct collection system 57 (Wescor Inc., Logan, Utah) is simple but avoids many problems. Sweat is collected within a coil of plastic tubing from which signifCF QPIT
Cystic fibrosis Quantitative pilocarpine iontophoresis test
icant evaporation cannot occur. Weighing and elution are eliminated. Sweat can be taken from the coil and analyzed immediately for ionic composition, or the sweat can initially be put through the Sweat-Chek conductivity analyzer 8 (Wescor) before chemical analysis. To determine the accuracy of sweat tests utilizing this collection method, the composition of sweat collected in the Macroduct system was compared with the concentrations determined simultaneously by the Q P I T method. In addition, the efficacy of the Sweat-Chek conductivity analyzer was studied by comparing the conductivity reading with measurements of chloride, sodium, and potassium in the same specimens. METHODS The subjects were patients referred to the laboratory of one of us (K.B.H.) for sweat testing between November 1982 and December 1992. Only patients who simultaneously underwent Macroduct and Q P I T tests were included. The 1090 patient values reported represent 63.8% of all sweat tests performed during the 10-year period. Only initial tests were used; additional studies were eliminated. Other reasons for elimination were the presence of an intravenous line in one arm, loss of a sample, and interference with other studies. Any other eliminations were entirely random. Measurements of conductivity, as well as
comparison with the QPIT, were performed after the Sweat-Chek analyzer was acquired in March 1987. The diagnosis of CF was made on the basis of a positive sweat test result plus one of the following: clinical symptoms, a positive result on a newborn screen, or a positive result on gene mutation analysis. A positive sweat test result was defined as a chloride concentration greater than 60 mmol/L. Simultaneous sweat tests were performed on 1090 patients, 93 of whom had CF, with the use of the Q P I T method on one arm and the Macroduct system on the other. The age range of the patients with CF was 1 week to 37 years (median = 10 weeks); the control subjects' age range was 3 days to 78 years (median = 28 weeks). In the Q P I T procedure, 2.5 x 2.5 cm copper electrodes were placed over gauze pads saturated with 0.5°70 pilocarpine nitrate (positive electrode) or 0.2% sodium nitrate (negative electrode). The current was 2 m A for 5 minutes and the collection time was 30 minutes. Collection was on two 4.25 cm diameter Whatman No. 42 filter paper circles (Whatman Inc., Clifton, N.J.). Volumes <50 mg were considered inadequate. Chloride analysis was performed with mercuric nitrate titration by the method of Schales and Schales. 9 Sodium and potassium concentrations were determined with a flame photometer. The Macroduct system includes a battery-operated current source that, during a 5-minute period, raises, keeps constant, and then lowers the current. The maximum current during the procedure is 1.5 mA. The electrodes are made to hold pilocarpine-containing gel disks 2.8 cm in diameter. After iontophoresis, the area is cleaned and dried and the Macroduct collector attached with straps. This collector consists of a slightly concave plastic disk with a hole in the center. This hole is connected to a small (internal diameter 0.025 inch) plastic tube coiled over the top of the disk. Sweat, secreted under pressure, is forced through the hole into the tubing. There is virtually no dead space. A small amount (10 nmol) of water-soluble dye on the concave surface of the disk allows easy visualization of the sweat collected. Sweat was collected for a period of 30 minutes. The Sweat-Chek conductivity meter has two small stainless steel nipples. The plastic tube from the Macroduct can be attached to one nipple so that the sweat can be pushed With positive pressure through the conductivity cell. Another plastic tube attached to the second nipple allows sweat that has passed through the conductivity cell to be introduced into a small container for subsequent analysis. The 4 #1 cell where conductivity is measured is kept at a constant temperature of approximately 40 ° C. Calibration is easily verified by introducing standard solutions. Deionized water and air were passed through the cell between all determinations. Repeat standardization showed that this procedure prevented carryover contamination. When Macroduct samples were compared with those oh-
The Journal of Pediatrics Volume 124, Number 2
Hammond, Turcios, and Gibson
257
120 _]
110
-d
100
E
+ r~ Z
Y : 0.974 X g : 0.986
90
•
e w ~' ~ _9~e _,, ~ I ~ -
Sy x = 3.90 n = 1022
80
•
+ 0.645
70 60
bC] d] 0 02 C] <12 5L
50 4O 3O 20 10
~ 0 0 0 f i k ' ~; ) O~l) (1~)~ i , I v
I ~ I ~ i , 1 , I
0 10
0
20
30
A
ii0 I00 90 80
40 50 60 QPIT Na +
Y : 0.989 X r = 0989 Sy x = 3.85 n : 1014
+ 0399
£3 0 02 (D <<
~ ~ " •
eO~ll~- " ~ //~-
70 F-C]
70 8 0 9 0 100 110 120 (mmol/L)
o//
60 so
o ~bc4"
40
~ ~ o
2O 310 0 l
.i
~' , ~- i
,
i
, ~
,
I
,
r
,
i
,
i
,
I
,
f
,
I
,
i
0
0
B
i0
20
30
40 50 60 70 80 90 i00 ii0 120 OPmT CI(mmol/L)
Fig. 1. Comparison of sodium (A) and chloride (B) concentrations in sweat simultaneouslycollected from patients with CF (dark circles) and non-CF patients (clear circles) with the QPIT and Macroduct collectionsystems, r, Regression coefficient; Sy • x, standard error of the estimate.
tained by the QPIT, sweat was transferred directly from the Macroduct tubing to a small container kept tightly capped until analysis of the sample. When the conductivity readings were compared with ionic composition, conductivity was determined before chemical analysis. In both cases, sweat obtained by the Macroduct system was removed from the container with a micropipet before chemical analysis; samples smaller than 15 ~1 were considered insufficient. Although unnecessary for routine sweat testing, sweat weight was measured during this study on all Macroduct samples. Data comparisons were performed with a commercial
statistical program (In Plot; Graph Pad Software, San Diego, Calif.) using the least squares method of linear regression. The systematic errors were calculated with a commercial statistical package (LABSTAT; Prism Associates, Inc., Atlanta, Ga.) and were derived from the corresponding bias plots, and not from the linear regression plots.
RESULTS When simultaneous Macroduct and QPIT sweat tests were performed, eight patients (0.7%) tested with the QP1T had insufficient sweat. All but one of these eight patients were less than 11 weeks of age. With the Macroduet system,
258
Hammond, Turcios, and Gibson
E E + bE + + Z
140 130 120 110 100 90 80 70 6O 5O 40 3O 2O 10 0 0
The Journal of Pediatrics February 1994
0°
Y = r : Sy × n :
lO 2 0
0.974 X 0987 : 3.88 514
_ t1~ ~li;Y~ _ ,~
30 40 50 60 70 80 90 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0
Conductivity
A
- 1.49
(Equivalent
120
om°o~ooo
110
Y = F= Sx × n =
100
-2
E E t
o
B
9O 80 70! 60'
5O 40 3O 2O 10 0 0
NaCI mmol/L)
X
- 1521
e~/~ eo~-~-
514
~ ° ° ° ° J :I~iO°c~O000 o
o I
0961 0974 = 554
, i , i i q, i , i
i
102030405060708090100110120130140 Conductivity (Equivalent NaCL mmol/L)
Fig. 2. Comparison of conductivitywith the sum of sodium and potassium concentrations (A) and with the chloride concentration (B) performed on the same sweat samplestaken from patients with CF (dark circles) and non-CF patients (clear circles), r, Regression coefficient;Sy . x, standard error of the estimate. 67 patients (6.1%) had insufficient sweat; five of these had CF. Most of these failures occurred during the first 3 months of life; of 649 subjects more than 11 weeks of age, 25 (4.0%) had insufficientsweat. The average yield from the QPIT was 217 mg and from the Macroduct system, 60 tA. Sodium concentrations were measured on 1022 paired sweat specimens and chloride concentrations on 1014 paired sweat specimens. There was close agreement between the QPIT and the Macroduct system in both sodium and chloride values (Table I; Fig. 1). Paired t tests revealed a statistically signifi-
cant difference (bias) between sodium values in the patients with CF (p = 0.002), but this difference was not clinically important. No statistically significant differences were seen between sodium values in the non-CF group (p -- 0.127) or between chloride values in the CF group (p = 0.593) or the non-CF group (p = 0.126). The systematic error (average bias) for values within the clinical decision range of 40 to 60 mmol/L (n = 41) was 1.2 mmol/L and 1.5 mmol/L for sodium and chloride, respeetively. Random error, as measured by the standard error of the estimate (Sy, x), was 3.90 mmol/L for sodium (Fig. 1, A) and 3.85 mmol/L for chlo-
The Journal of Pediatrics Volume 124, Number 2
Hammond, Turcios, and Gibson
259
40 0 O
30 o°O
oo
o
.
i
•
:>.
20 _.J o D -0 c
°
%o
o
•
~,..... ee • o •
•
|-
-6 E E
10
oO
o
"
.-
•
0
•
• e
th ru
c5 -10
,
0
I
,
I
,
l
,
P
,
t
r
I
,
f
,
I
,
I
,
I
~
I
,
I
,
10 20 30 40 50 60 70 80 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0
Conductivity (Equivalent NaCI retool/L) Fig. 3. Comparison of bias and conductivity. The bias is presumably unmeasured anion.
T a b l e II. Electrolyte concentrations and conductivity measurements made on the same sweat sample taken from non-CF and CF patients Electrolyte c o n c e n t r a t i o n s ( m m o l / L ) Na +
Non-CF patients (n = 471) Average 18.5 sD 11.7 SE 0.5 Range 5-74 Patients with CF (n = 43) Average 87.0 SD 14.8 SE 2.0 Range 63-123
K+
Na + + K+
Cl-
Conductivity
12.5 3.3 0.2 5-28
31.0 11.6 0.5 13-83
16.4 8.5 0.4 5-60
33.4 11.2 0.5 13-87
21.4 7.3 hl 9-36
108.4 12.2 1.9 84-135
98.8 9.8 1.5 77-117
113.1 9.9 1.5 90-136
Values (except conductivity) are expressed as millimoles per liter.
ride (Fig. 1, B). On average, within 95% confidence limits, the Macroduct values predicted the QPIT values within approximately 8 mmol/L. This degree of random error was similar to that found with paired QPIT determinations in the same laboratory in which the Sy, x value was 3.09 mmol/L for sodium and 3.13 mmol/L for chloride. The sweat conductivityvalues correlated well (r = 0.987) with the total measured cation ( N a + + K+; Fig. 2, A). There was also a good correlation with the sweat chloride values (r = 0.974; Fig. 2, B). The non-CF population had a mean conductivity value of 33.4 mmol/L (range, 13 to 87 mmol/L), and the C F population had a mean of 113.1 mmol/L (range, 90 to 136 mmol/L). The corresponding mean values and ranges for sodium, potassium, chloride, and conductivity are shown in Table II. The linear regression plot of sweat chloride versus
conductivity is shown in Fig. 2, B. Fitting a second-order polynomial to these data confirmed the visual impression that a straight line did not adequately describe the data. For this reason, the bias (conductivity minus chloride) was plotted against conductivity (Fig. 3). The bias increased proportionally with the conductivity in both the non-CF and the C F patients, although to differing degrees. Calculation of the average biases in the non-CF group showed an increase from 15.73 mmol/L (SE = 0.24) at conductivities of 20 to 40 mmol/L (n = 356) to 22.09 mmol/L (SE = 0.52) at conductivities of 40 to 60 mrnol/L (n = 85). In the patients with CF the average bias increased from 12.60 mmol/L (SE = 0.92) at conductivities <115 rnmol/L (n = 24) to 16.45 mmol/L (SE = 1.23) at conductivities >115 mmol/L (n = 19). These biases are due to the presence of unmeasured anions. The relatively smaller biases in
260
Hammond, Turcios, and Gibson
the patients with CF are presumably due to the smaller amounts of bicarbonate ion present in the sweat.l° The average bias for all subjects was 15.21 mmol/L. In this study one patient with CF initially had the following values: for QPIT, sodium 35 retool/L, potassium 13 retool/L, and Chloride 46 mmol/L, and for Macroduct analysis, sodium 43 mmol/L, potassium 12 mmol/L, and chloride, 54 mmol/L. He had hypochloremic alkalosis; after this was corrected, his chloride concentration was repeatedly >70 mmol/L. The initial analysis was used for this study. With this exception, all repeated sweat tests were consistent with the initial analysis. DISCUSSION The QPIT sweat test as originally described t is remarkably accurate, but it requires careful application of the c01lecting pad to prevent evaporation, the use of a chemical balance, and the calculation of concentrations from eluted specimens containing widely varying quantities of sweat. For these reasons many laboratories use less satisfactory methods. 24 To avoid weighing, investigators have collected sweat under a plastic cup and scraped the sweat off the skin. In this procedure, water may evaporate from the sweat on the skin, increasing its salt concentration, while distilled water collects on the cool undersurface of the cup. There is no good method of mixing these two liquids, so either falsely low or falsely high concentrations may be measured. When conductivity is measured without temperature control to estimate salt content, the error is compounded. Surface electrodes can measure the chloride concentration of test solutions accurately, but without a method of controlling evaporation they give inaccurate results. They may also be pressure sensitive. The Macroduct collection system allows the collection of pure sweat without evaporation. Weighing is eliminated because a micropipet is used to quantify sweat for analysis. The data show that there is no clinically significant difference between the concentrations of chloride and sodium obtained with the QPIT method and with the Macroduct system. The Macroduct system does have a higher failure rate. The material costs for the Macroduct system of $10.00 per test are considerably higher than for the QPIT, in which material costs are negligible. This difference may be offset by lower labor costs for the Macroduct system. The Sweat-Chek conductivity meter shows that the sum of the sodium and potassium concentrations very closely approximates conductivity. The chloride concentration averages 15.21 mmol/L less than the conductivity equivalent.
The Journal of Pediatrics February 1994
Presumably this is caused by the presence of lactate, bicarbonate, and other unmeasured anions. Sweat chloride concentration provides a slightly better discriminationthan sodium between subjects with and those without CF. For this reason, chloride remains the preferred analyte for diagnostic confirmation. If it had been decided to eliminate the chloride determination for all subjects with conductivity readings less than 50 mmol/L, as suggested here, none of the 43 patients with CF who were tested would have been missed. Their lowest conductivity determination was 90 mmol/L. Of the control subjects, 430 (91%) would not have received a chloride determination; none of these subjects had a sweat chloride concentration exceeding 32 mEq/L. The QPIT is accurate but requires skill and time. The macroduct system, especially when used with the SweatChek analyzer, is simple and rapid; our work demonstrated that it is also reliable. Its availability should lead to improved accuracy of sweat test results in laboratories where less reliable methods have been used. REFERENCES
1. Gibson LE, Cooke RE. A test for concentration of electrolytes in sweat in cystic fibrosis of the pancreas utilizing pilocarpine by iontophoresis.Pediatrics 1959;23:545-9. 2. Gibson LE. The decline of the sweat test. Clin Pediatr 1973;12:450-3. 3. Denning CF, Huang NN, Cuasay LR, et al. Cooperativestudy comparingthree methods of performingsweat tests to diagnose cystic fibrosis. Pediatrics 1980;66:752-7. 4. Rosenstein B, Laugbaum T. Sweat testing in CF: not to be taken lightly. J Respir Dis 1982;3:71-6. 5. Barlow WK, Webster HL. A simplifiedmethod of sweat collection for diagnosisof cystic fibrosis. In: Lawson D, ed. Cystic fibrosis: horizons. Proceedings of the 9th International Cystic Fibrosis Congress. Brighton, England, June 9-15, 1984. New York: Wiley & Sons, 1984:204. 6. Carter EP, Barrett AD, Heeley AF, Kuzemko JA. Improved sweat test method for the diagnosisof cystic fibrosis.Arch Dis Child 1984;59:919-22. 7. Hammond KB, Turcios N, Gibson LE. An evaluation of the Wescor Sweat-Chek conductivityanalyzer [Abstract]. Pediatr Pulmonol 1988;162(suppl 2):139, 8. Licht TS, Stern M, Shwachman H. Measurement of the electrical conductivityof sweat [Abstract]. Clin Chem 1957;3:37. 9. Schales O, Schales SS. A simple and accurate method for the determination of chloride in biological fluids. J Biol Chem 1941;140:879-84. 10. Quinton PM. Abnormalities in electrolyte secretion in cystic fibrosis sweat glands due to decreased anion permeability. In: Quinton PM, Martinez JR, Hopfer U, eds. Fluid and electrolyte abnormalities in exocrine glands in cystic fibrosis. San Francisco: San Francisco Press, 1982.
The purposes of this study were to c o m p a r e sweat tests used in diagnosing cystic fibrosis (CF), as performed with the Macroduct c o l l e c t i o n system, with those utilizing the more laborious quantitative pilocarpine iontophoresis test (QPIT), and to ascertain the e f f i c a c y of the Sweat-Chek c o n d u c t i v i t y analyzer in eliminating some possibly unnecessary chloride analyses. A Macroduct sweat test was performed on one arm and a QPIT on the other on 1090 patients, 93 of whom had CF. Of these, 514 patients (43 with CF) also had a c o n d u c t i v i t y determination on the Macroduct sweat sample. All subjects were referred to the laboratory of one of us (K.B.H.) for sweat testing. Of the QPII" samples, 0.7% were inadequate, as were 6.1% of those from the Macroduct system. When sodium and chloride concentrations from the two tests were compared, the standard errors of the estimate were 3.90 and 3.85, respectively. Agreement within 8 mEq/L could then be e x p e c t e d with 95% c o n f i d e n c e limits. With use of the Sweat-Chek analyzer, no patient with CF was found to have a c o n d u c t i v i t y of less than 90 mmol/L, whereas 430 (91%) of the non-CF subjects had a c o n d u c t i v i t y of less than 50 mmol/L. None of those 430 subjects had a sweat chloride value >32 mmol/L. We c o n c l u d e that the Macroduct c o l l e c t i o n system provides results e q u a l l y as satisfactory as those provided by the QPIT and that the Sweat-Chek analyzer frequently eliminates the necessity of measuring chloride concentrations. (J PEDIATR1994;124:255-60.)
Early diagnosis of cystic fibrosis necessitates performing many negative sweat tests on children without CF to diagnose the disease in a few. The quantitative pilocarpine ionSupported by Wescor Inc., in the donation of the Sweat-Chek analyzer to our laboratory. All other equipment and consumables from Wescor were purchased in the normal way. Submitted for publication April 30, 1993; accepted Sept. 3, 1993. Reprint requests: Lewis E. Gibson, MD, Loyola University Medical Center, 2160 South First Ave., Maywood, IL 60153. Copyright © 1994 by Mosby-Year Book, Inc. 0022-3476/94 $3.00 + 0 9/20/51264
tophoresis test,* when performed correctly, 1 is very accurate, but time and skill are required to prevent evaporation during collection, to determine sweat weight with a chemical balance, and to calculate the chemical composition of the eluted sample. To simplify the test, many laboratories have adopted alternative methods that have inherent problems. 24 These problems may seem small, but their importance becomes apparent when one patient receives an inaccurate diagnosis. *Used here to mean that sweat is quantified by weighing a collection pad, as distinct from Macroduct collection system. Sweat collected in the Macroduct system may be quantified with a micropipet.
255
256
Hammond, Turcios, and Gibson
The Journal of Pediatrics February 1994
Table I. Average electrolyte concentrations in sweat collected simultaneously by two methods QPIT Na +
Non-CF patients* Average 18.2 SD 11.8 SE 0.4 Range 3-77 Patients with CFt Average 87.9 SD 16.9 SE 1.8 Range 35-131
Macroduct CI-
Na +
Cl
15.9 8.7 0.3 5-65
18.4 12.3 0.4 3-85
16.1 9.2 0.3 3-69
100.8 12.6 1.3 46-124
86.1 16.1 1.7 43-123
100.5 11.9 1.3 54-124
Values are expressed as millimolesper liter. *Sodiumconcentrationswere measured in 934 non-CFpatients and chloride concentrations in 926. ?Sodium concentrationsweremeasured in 88 patients with CF and chloride concentrations in 88.
The Macroduct collection system 57 (Wescor Inc., Logan, Utah) is simple but avoids many problems. Sweat is collected within a coil of plastic tubing from which signifCF QPIT
Cystic fibrosis Quantitative pilocarpine iontophoresis test
icant evaporation cannot occur. Weighing and elution are eliminated. Sweat can be taken from the coil and analyzed immediately for ionic composition, or the sweat can initially be put through the Sweat-Chek conductivity analyzer 8 (Wescor) before chemical analysis. To determine the accuracy of sweat tests utilizing this collection method, the composition of sweat collected in the Macroduct system was compared with the concentrations determined simultaneously by the Q P I T method. In addition, the efficacy of the Sweat-Chek conductivity analyzer was studied by comparing the conductivity reading with measurements of chloride, sodium, and potassium in the same specimens. METHODS The subjects were patients referred to the laboratory of one of us (K.B.H.) for sweat testing between November 1982 and December 1992. Only patients who simultaneously underwent Macroduct and Q P I T tests were included. The 1090 patient values reported represent 63.8% of all sweat tests performed during the 10-year period. Only initial tests were used; additional studies were eliminated. Other reasons for elimination were the presence of an intravenous line in one arm, loss of a sample, and interference with other studies. Any other eliminations were entirely random. Measurements of conductivity, as well as
comparison with the QPIT, were performed after the Sweat-Chek analyzer was acquired in March 1987. The diagnosis of CF was made on the basis of a positive sweat test result plus one of the following: clinical symptoms, a positive result on a newborn screen, or a positive result on gene mutation analysis. A positive sweat test result was defined as a chloride concentration greater than 60 mmol/L. Simultaneous sweat tests were performed on 1090 patients, 93 of whom had CF, with the use of the Q P I T method on one arm and the Macroduct system on the other. The age range of the patients with CF was 1 week to 37 years (median = 10 weeks); the control subjects' age range was 3 days to 78 years (median = 28 weeks). In the Q P I T procedure, 2.5 x 2.5 cm copper electrodes were placed over gauze pads saturated with 0.5°70 pilocarpine nitrate (positive electrode) or 0.2% sodium nitrate (negative electrode). The current was 2 m A for 5 minutes and the collection time was 30 minutes. Collection was on two 4.25 cm diameter Whatman No. 42 filter paper circles (Whatman Inc., Clifton, N.J.). Volumes <50 mg were considered inadequate. Chloride analysis was performed with mercuric nitrate titration by the method of Schales and Schales. 9 Sodium and potassium concentrations were determined with a flame photometer. The Macroduct system includes a battery-operated current source that, during a 5-minute period, raises, keeps constant, and then lowers the current. The maximum current during the procedure is 1.5 mA. The electrodes are made to hold pilocarpine-containing gel disks 2.8 cm in diameter. After iontophoresis, the area is cleaned and dried and the Macroduct collector attached with straps. This collector consists of a slightly concave plastic disk with a hole in the center. This hole is connected to a small (internal diameter 0.025 inch) plastic tube coiled over the top of the disk. Sweat, secreted under pressure, is forced through the hole into the tubing. There is virtually no dead space. A small amount (10 nmol) of water-soluble dye on the concave surface of the disk allows easy visualization of the sweat collected. Sweat was collected for a period of 30 minutes. The Sweat-Chek conductivity meter has two small stainless steel nipples. The plastic tube from the Macroduct can be attached to one nipple so that the sweat can be pushed With positive pressure through the conductivity cell. Another plastic tube attached to the second nipple allows sweat that has passed through the conductivity cell to be introduced into a small container for subsequent analysis. The 4 #1 cell where conductivity is measured is kept at a constant temperature of approximately 40 ° C. Calibration is easily verified by introducing standard solutions. Deionized water and air were passed through the cell between all determinations. Repeat standardization showed that this procedure prevented carryover contamination. When Macroduct samples were compared with those oh-
The Journal of Pediatrics Volume 124, Number 2
Hammond, Turcios, and Gibson
257
120 _]
110
-d
100
E
+ r~ Z
Y : 0.974 X g : 0.986
90
•
e w ~' ~ _9~e _,, ~ I ~ -
Sy x = 3.90 n = 1022
80
•
+ 0.645
70 60
bC] d] 0 02 C] <12 5L
50 4O 3O 20 10
~ 0 0 0 f i k ' ~; ) O~l) (1~)~ i , I v
I ~ I ~ i , 1 , I
0 10
0
20
30
A
ii0 I00 90 80
40 50 60 QPIT Na +
Y : 0.989 X r = 0989 Sy x = 3.85 n : 1014
+ 0399
£3 0 02 (D <<
~ ~ " •
eO~ll~- " ~ //~-
70 F-C]
70 8 0 9 0 100 110 120 (mmol/L)
o//
60 so
o ~bc4"
40
~ ~ o
2O 310 0 l
.i
~' , ~- i
,
i
, ~
,
I
,
r
,
i
,
i
,
I
,
f
,
I
,
i
0
0
B
i0
20
30
40 50 60 70 80 90 i00 ii0 120 OPmT CI(mmol/L)
Fig. 1. Comparison of sodium (A) and chloride (B) concentrations in sweat simultaneouslycollected from patients with CF (dark circles) and non-CF patients (clear circles) with the QPIT and Macroduct collectionsystems, r, Regression coefficient; Sy • x, standard error of the estimate.
tained by the QPIT, sweat was transferred directly from the Macroduct tubing to a small container kept tightly capped until analysis of the sample. When the conductivity readings were compared with ionic composition, conductivity was determined before chemical analysis. In both cases, sweat obtained by the Macroduct system was removed from the container with a micropipet before chemical analysis; samples smaller than 15 ~1 were considered insufficient. Although unnecessary for routine sweat testing, sweat weight was measured during this study on all Macroduct samples. Data comparisons were performed with a commercial
statistical program (In Plot; Graph Pad Software, San Diego, Calif.) using the least squares method of linear regression. The systematic errors were calculated with a commercial statistical package (LABSTAT; Prism Associates, Inc., Atlanta, Ga.) and were derived from the corresponding bias plots, and not from the linear regression plots.
RESULTS When simultaneous Macroduct and QPIT sweat tests were performed, eight patients (0.7%) tested with the QP1T had insufficient sweat. All but one of these eight patients were less than 11 weeks of age. With the Macroduet system,
258
Hammond, Turcios, and Gibson
E E + bE + + Z
140 130 120 110 100 90 80 70 6O 5O 40 3O 2O 10 0 0
The Journal of Pediatrics February 1994
0°
Y = r : Sy × n :
lO 2 0
0.974 X 0987 : 3.88 514
_ t1~ ~li;Y~ _ ,~
30 40 50 60 70 80 90 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0
Conductivity
A
- 1.49
(Equivalent
120
om°o~ooo
110
Y = F= Sx × n =
100
-2
E E t
o
B
9O 80 70! 60'
5O 40 3O 2O 10 0 0
NaCI mmol/L)
X
- 1521
e~/~ eo~-~-
514
~ ° ° ° ° J :I~iO°c~O000 o
o I
0961 0974 = 554
, i , i i q, i , i
i
102030405060708090100110120130140 Conductivity (Equivalent NaCL mmol/L)
Fig. 2. Comparison of conductivitywith the sum of sodium and potassium concentrations (A) and with the chloride concentration (B) performed on the same sweat samplestaken from patients with CF (dark circles) and non-CF patients (clear circles), r, Regression coefficient;Sy . x, standard error of the estimate. 67 patients (6.1%) had insufficient sweat; five of these had CF. Most of these failures occurred during the first 3 months of life; of 649 subjects more than 11 weeks of age, 25 (4.0%) had insufficientsweat. The average yield from the QPIT was 217 mg and from the Macroduct system, 60 tA. Sodium concentrations were measured on 1022 paired sweat specimens and chloride concentrations on 1014 paired sweat specimens. There was close agreement between the QPIT and the Macroduct system in both sodium and chloride values (Table I; Fig. 1). Paired t tests revealed a statistically signifi-
cant difference (bias) between sodium values in the patients with CF (p = 0.002), but this difference was not clinically important. No statistically significant differences were seen between sodium values in the non-CF group (p -- 0.127) or between chloride values in the CF group (p = 0.593) or the non-CF group (p = 0.126). The systematic error (average bias) for values within the clinical decision range of 40 to 60 mmol/L (n = 41) was 1.2 mmol/L and 1.5 mmol/L for sodium and chloride, respeetively. Random error, as measured by the standard error of the estimate (Sy, x), was 3.90 mmol/L for sodium (Fig. 1, A) and 3.85 mmol/L for chlo-
The Journal of Pediatrics Volume 124, Number 2
Hammond, Turcios, and Gibson
259
40 0 O
30 o°O
oo
o
.
i
•
:>.
20 _.J o D -0 c
°
%o
o
•
~,..... ee • o •
•
|-
-6 E E
10
oO
o
"
.-
•
0
•
• e
th ru
c5 -10
,
0
I
,
I
,
l
,
P
,
t
r
I
,
f
,
I
,
I
,
I
~
I
,
I
,
10 20 30 40 50 60 70 80 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0
Conductivity (Equivalent NaCI retool/L) Fig. 3. Comparison of bias and conductivity. The bias is presumably unmeasured anion.
T a b l e II. Electrolyte concentrations and conductivity measurements made on the same sweat sample taken from non-CF and CF patients Electrolyte c o n c e n t r a t i o n s ( m m o l / L ) Na +
Non-CF patients (n = 471) Average 18.5 sD 11.7 SE 0.5 Range 5-74 Patients with CF (n = 43) Average 87.0 SD 14.8 SE 2.0 Range 63-123
K+
Na + + K+
Cl-
Conductivity
12.5 3.3 0.2 5-28
31.0 11.6 0.5 13-83
16.4 8.5 0.4 5-60
33.4 11.2 0.5 13-87
21.4 7.3 hl 9-36
108.4 12.2 1.9 84-135
98.8 9.8 1.5 77-117
113.1 9.9 1.5 90-136
Values (except conductivity) are expressed as millimoles per liter.
ride (Fig. 1, B). On average, within 95% confidence limits, the Macroduct values predicted the QPIT values within approximately 8 mmol/L. This degree of random error was similar to that found with paired QPIT determinations in the same laboratory in which the Sy, x value was 3.09 mmol/L for sodium and 3.13 mmol/L for chloride. The sweat conductivityvalues correlated well (r = 0.987) with the total measured cation ( N a + + K+; Fig. 2, A). There was also a good correlation with the sweat chloride values (r = 0.974; Fig. 2, B). The non-CF population had a mean conductivity value of 33.4 mmol/L (range, 13 to 87 mmol/L), and the C F population had a mean of 113.1 mmol/L (range, 90 to 136 mmol/L). The corresponding mean values and ranges for sodium, potassium, chloride, and conductivity are shown in Table II. The linear regression plot of sweat chloride versus
conductivity is shown in Fig. 2, B. Fitting a second-order polynomial to these data confirmed the visual impression that a straight line did not adequately describe the data. For this reason, the bias (conductivity minus chloride) was plotted against conductivity (Fig. 3). The bias increased proportionally with the conductivity in both the non-CF and the C F patients, although to differing degrees. Calculation of the average biases in the non-CF group showed an increase from 15.73 mmol/L (SE = 0.24) at conductivities of 20 to 40 mmol/L (n = 356) to 22.09 mmol/L (SE = 0.52) at conductivities of 40 to 60 mrnol/L (n = 85). In the patients with CF the average bias increased from 12.60 mmol/L (SE = 0.92) at conductivities <115 rnmol/L (n = 24) to 16.45 mmol/L (SE = 1.23) at conductivities >115 mmol/L (n = 19). These biases are due to the presence of unmeasured anions. The relatively smaller biases in
260
Hammond, Turcios, and Gibson
the patients with CF are presumably due to the smaller amounts of bicarbonate ion present in the sweat.l° The average bias for all subjects was 15.21 mmol/L. In this study one patient with CF initially had the following values: for QPIT, sodium 35 retool/L, potassium 13 retool/L, and Chloride 46 mmol/L, and for Macroduct analysis, sodium 43 mmol/L, potassium 12 mmol/L, and chloride, 54 mmol/L. He had hypochloremic alkalosis; after this was corrected, his chloride concentration was repeatedly >70 mmol/L. The initial analysis was used for this study. With this exception, all repeated sweat tests were consistent with the initial analysis. DISCUSSION The QPIT sweat test as originally described t is remarkably accurate, but it requires careful application of the c01lecting pad to prevent evaporation, the use of a chemical balance, and the calculation of concentrations from eluted specimens containing widely varying quantities of sweat. For these reasons many laboratories use less satisfactory methods. 24 To avoid weighing, investigators have collected sweat under a plastic cup and scraped the sweat off the skin. In this procedure, water may evaporate from the sweat on the skin, increasing its salt concentration, while distilled water collects on the cool undersurface of the cup. There is no good method of mixing these two liquids, so either falsely low or falsely high concentrations may be measured. When conductivity is measured without temperature control to estimate salt content, the error is compounded. Surface electrodes can measure the chloride concentration of test solutions accurately, but without a method of controlling evaporation they give inaccurate results. They may also be pressure sensitive. The Macroduct collection system allows the collection of pure sweat without evaporation. Weighing is eliminated because a micropipet is used to quantify sweat for analysis. The data show that there is no clinically significant difference between the concentrations of chloride and sodium obtained with the QPIT method and with the Macroduct system. The Macroduct system does have a higher failure rate. The material costs for the Macroduct system of $10.00 per test are considerably higher than for the QPIT, in which material costs are negligible. This difference may be offset by lower labor costs for the Macroduct system. The Sweat-Chek conductivity meter shows that the sum of the sodium and potassium concentrations very closely approximates conductivity. The chloride concentration averages 15.21 mmol/L less than the conductivity equivalent.
The Journal of Pediatrics February 1994
Presumably this is caused by the presence of lactate, bicarbonate, and other unmeasured anions. Sweat chloride concentration provides a slightly better discriminationthan sodium between subjects with and those without CF. For this reason, chloride remains the preferred analyte for diagnostic confirmation. If it had been decided to eliminate the chloride determination for all subjects with conductivity readings less than 50 mmol/L, as suggested here, none of the 43 patients with CF who were tested would have been missed. Their lowest conductivity determination was 90 mmol/L. Of the control subjects, 430 (91%) would not have received a chloride determination; none of these subjects had a sweat chloride concentration exceeding 32 mEq/L. The QPIT is accurate but requires skill and time. The macroduct system, especially when used with the SweatChek analyzer, is simple and rapid; our work demonstrated that it is also reliable. Its availability should lead to improved accuracy of sweat test results in laboratories where less reliable methods have been used. REFERENCES
1. Gibson LE, Cooke RE. A test for concentration of electrolytes in sweat in cystic fibrosis of the pancreas utilizing pilocarpine by iontophoresis.Pediatrics 1959;23:545-9. 2. Gibson LE. The decline of the sweat test. Clin Pediatr 1973;12:450-3. 3. Denning CF, Huang NN, Cuasay LR, et al. Cooperativestudy comparingthree methods of performingsweat tests to diagnose cystic fibrosis. Pediatrics 1980;66:752-7. 4. Rosenstein B, Laugbaum T. Sweat testing in CF: not to be taken lightly. J Respir Dis 1982;3:71-6. 5. Barlow WK, Webster HL. A simplifiedmethod of sweat collection for diagnosisof cystic fibrosis. In: Lawson D, ed. Cystic fibrosis: horizons. Proceedings of the 9th International Cystic Fibrosis Congress. Brighton, England, June 9-15, 1984. New York: Wiley & Sons, 1984:204. 6. Carter EP, Barrett AD, Heeley AF, Kuzemko JA. Improved sweat test method for the diagnosisof cystic fibrosis.Arch Dis Child 1984;59:919-22. 7. Hammond KB, Turcios N, Gibson LE. An evaluation of the Wescor Sweat-Chek conductivityanalyzer [Abstract]. Pediatr Pulmonol 1988;162(suppl 2):139, 8. Licht TS, Stern M, Shwachman H. Measurement of the electrical conductivityof sweat [Abstract]. Clin Chem 1957;3:37. 9. Schales O, Schales SS. A simple and accurate method for the determination of chloride in biological fluids. J Biol Chem 1941;140:879-84. 10. Quinton PM. Abnormalities in electrolyte secretion in cystic fibrosis sweat glands due to decreased anion permeability. In: Quinton PM, Martinez JR, Hopfer U, eds. Fluid and electrolyte abnormalities in exocrine glands in cystic fibrosis. San Francisco: San Francisco Press, 1982.