Clinical Biochemistry 48 (2015) 443–447
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Sweating the small stuff: Adequacy and accuracy in sweat chloride determination☆ Mari L. DeMarco a,⁎, Dennis J. Dietzen b,c, Sarah M. Brown b,c a b c
Department of Pathology and Laboratory Medicine, St. Paul's Hospital, University of British Columbia, Vancouver, Canada Department of Pathology and Immunology, Washington University School of Medicine, St Louis, MO, USA Department of Pediatrics, Washington University School of Medicine, St Louis, MO, USA
a r t i c l e
i n f o
Article history: Received 5 November 2014 Received in revised form 9 December 2014 Accepted 10 December 2014 Available online 18 December 2014 Keywords: Sweat chloride Sweat rate Cystic fibrosis Biological variability Limit of quantitation
a b s t r a c t Objectives: Sweat chloride testing is the gold standard for diagnosis of cystic fibrosis (CF). Our objectives were to: 1) describe variables that determine sweat rate; 2) determine the analytic and diagnostic capacity of sweat chloride analysis across the range of observed sweat rates; and 3) determine the biologic variability of sweat chloride concentration. Methods: A retrospective analysis was performed using data from all sweat chloride tests performed at St. Louis Children's Hospital over a 21-month period. Results: A total of 1397 sweat chloride tests (1155 sufficient [≥75 mg], 242 insufficient [b75 mg]), were performed on 904 individuals. The sweat weight collected from forearms was statistically greater than that collected from legs. There was a negligible correlation between sweat weight and chloride concentration (r = −0.06). The mean individual biologic CV calculated from individuals with two or more sweat collections ≥ 75 mg was 13.1% (95% CI: 11.3–14.9%; range 0–88%) yielding a reference change value of 36%. Using 60 mmol/L as the diagnostic chloride cutoff, 100% of CF cases were detected whether a minimum sweat weight of 75, 40, or 20 mg was required. Conclusions: 1) Collection of sweat from forearms is preferable to upper legs, particularly in very young infants; 2) sweat chloride concentrations are not highly dependent upon sweat rate; 3) a change in sweat chloride concentration exceeding 36% may be considered a clinically significant response to cystic fibrosis transmembrane receptor targeted therapy, and 4) sweat collections of less than 75 mg provide clinically accurate information. © 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Sweat chloride analysis is the gold standard for diagnosis of cystic fibrosis (CF) [1]. Newborn screening for CF using immunoreactive trypsinogen is performed in all 50 of the United States and many other countries [2]. Positive newborn screens are confirmed with sweat chloride analysis [3]. With the advent of cystic fibrosis transmembrane receptor (CFTR)-modifying therapeutics such as ivacaftor [4], sweat chloride analysis may also be used to monitor response to therapy [5]. Sweat chloride analysis is a multi-step process that requires: 1) stimulation of sweat production with a cholinergic agonist (e.g., pilocarpine); 2) collection of sweat using filter paper, gauze, or plastic capillary tubing; and 3) chloride analysis, most commonly achieved by coulometric titration. Despite over 50 years of clinical use, some questions that remain are: 1) how much sweat is necessary for diagnosis, 2) which sites are Abbreviations: CF, cystic fibrosis; CFTR, cystic fibrosis transmembrane receptor; QNS, quantity not sufficient; CLSI, Clinical and Laboratory Standards Institute. ☆ This work was presented in part at the American Association for Clinical Chemistry Annual Meeting, Houston, TX, July 29, 2013. ⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, St Paul's Hospital, 1081 Burrard St, Vancouver BC V6Z 1Y6, Canada. E-mail address: [email protected] (M.L. DeMarco).
most responsive to sweat stimulation, 3) what is the biologic and analytical variation associated with sweat chloride analysis, and 4) what is the capacity for sweat chloride testing to reflect changes in severity of disease? There are minimum sweat volume requirements for collection by gauze and plastic capillary methods; 75 mg of sweat for the former and 15 μl of sweat by the latter. Collections less than the minimum (QNS) are considered inappropriate for diagnosis. QNS rates are especially high in newborns [6,7]. Bilateral stimulation is encouraged as a means of decreasing QNS rates [8] but collections from multiple sites must be analyzed independently as opposed to pooling specimens. The site for sweat induction may be an important determinant of sweat production. Common sites for stimulation include the leg, arm, or a combination of a leg and an arm. While it is suggested that one area may produce more sweat than another (e.g., forearm versus thigh) due to variations in sweat gland density, this hypothesis has not been rigorously examined. According to the Clinical and Laboratory Standards Institute (CLSI) guidelines, the current 75 mg sweat requirement for gauze/paper methods is based on a sweat rate of 1 g/m2/min. In the original paper by Gibson and Cooke, the average sweat weight collected on 2.5 cm diameter filter paper disks was 76 mg, range 18–135 mg [9]. Regardless of
http://dx.doi.org/10.1016/j.clinbiochem.2014.12.011 0009-9120/© 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
444
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
amount of sweat collected, no patient with CF had a sweat chloride concentration less than 80 mmol/L, and none of the controls had concentrations greater than 60 mmol/L. Goldberg et al. showed no correlation between sweat weight and sweat chloride concentration (n = 463, R2 = 0.002, p N 0.05) [10]. Our extensive experience as a CF testing center suggests that a lower sweat weight requirement that does not compromise diagnostic accuracy would facilitate rapid results and decrease parental stress associated with a delayed diagnosis. Finally, the variability in sweat chloride concentration between two different collections can be considerable. This imprecision almost certainly is derived from the sweat collection process, as chloride titration is precise (CV b 2%) at diagnostically relevant concentrations. Assessing the variation in this testing system is important for understanding differences between repeat values. This intra-individual variance, which has not been previously determined, will become increasingly important as sweat chloride is adopted as a biomarker of therapeutic effect. This study, therefore, examined the relationship between sweat rate and sweat chloride concentration, the contribution of intraindividual variation and variability in the sweat collection system to the imprecision of the sweat chloride measurement, and the relative capacity of arms and legs to produce sufficient sweat for analytical testing. Analytical considerations in this study are restricted to the Gibson and Cooke (gauze) method of sweat collection, although findings from analysis of sweat site capacity may have broader implications. Materials & Methods Sweat chloride analysis Quantitative pilocarpine iontophoresis sweat chloride testing was performed in the St. Louis Children's Hospital Core Laboratory following the Gibson and Cooke Technique [9]. Laboratory protocols follow the standards of the Clinical Laboratory Standards Institute CLSI-34 A3 guidelines [11]. Briefly, sweat was collected on pre-weighed 2×2-inch sterile gauze pads (Kendall Curity #3381, Marshfield, MA) over 30 minutes following pilocarpine nitrate iontophoresis (Gibson-Cooke Sweat Test Apparatus, Farrall Instruments, Grand Island, NE). Chloride ion concentration was determined by coulometric titration using a Biodynamics LyteTek-Cl Chloride Analyzer. Whenever feasible, sweat collections from two sites (upper leg and forearm) were performed at each patient encounter. According to CF Foundation and CLSI guidelines, the minimum sweat weight currently required for a sufficient chloride analysis is 75 mg but for purposes of this study, insufficient collections were retained and analyzed but not communicated to ordering physicians. This study was conducted under approval from the Washington University Human Research Protection Office. Limit of chloride quantitation The limit of chloride quantitation of the chloridometer was evaluated by adding known amounts of NaCl in solution (0.1, 0.4, 0.7, 1, 2, 3, and 4 μmoles of chloride) to pre-weighed gauze pads. Samples were tested in duplicate for 5 consecutive days. For purposes of this study the limit of quantitation (LOQ) was defined as the lowest amount of chloride detectable with an imprecision of less that 20% (CV) over at least 5 days of testing. Study design Analysis was performed using data from all sweat chloride tests performed at St. Louis Children's Hospital, a designated CF Center, over a 21-month period (January 2011 to September 2012). A total of 1397 sweat chloride measurements (1155 specimens ≥ 75 mg and 242 specimens b 75 mg), were performed on 904 individuals (Table 1). Variables included in the data analysis were: (1) age, (2) sweat site (forearm v. upper leg), (3) patient location (outpatient
Table 1 Descriptive statistics for sweat chloride testing performed at St. Louis Children's Hospital from January 2011 to September 2012. Number
Category
1397 1155 242 904 967 907 60
Sweat chloride tests Sufficient sweat collections QNS sweat collections Patients Patient encounters Patient encounters (age N 3 months) Patient encounters (age ≤ 3 months)
v. inpatient), (4) sweat weight, (5) sweat chloride concentration, and (6) confirmatory diagnosis of CF (genetic testing and/or two positive sweat chloride tests). CF Foundation interpretive guidelines for sweat chloride results were utilized (Table S1) [12]. Statistical analysis Descriptive statistics were calculated using median and ranges for continuous variables and percentages for categorical variables. The Mann–Whitney U test was used to examine the difference in sweat weights from different collection sites (forearms v. upper legs). Data analyses were performed with IBM SPSS version 20.0 (Armonk, NY), and GraphPad Prism version 6.0 (La Jolla, CA). Deming regression was performed with R version 3.1.1 (The R Foundation for Statistical Computing) and the cp-R interface [13]. The reference change value was calculated with 95% confidence as follows: 2 2 RCV ¼ √2 1:96 √ CV A þ CV I
ð1Þ
where CVA is the analytical coefficient of variation and CVI is the intraindividual biological coefficient of variation and as measured in the study, may contain contributions from variability in the sweat collection process. Results Imprecision of chloride measurement and limit of chloride quantitation In the case of sweat chloride measurement, imprecision and LOQ is dictated by both the analytic instrument and the amount of chloride presented. Our daily quality control regimen consists of separate 200 μL solutions containing 45 and 75 mmol/L NaCl with 9 and 15 μmoles of chloride ion, respectively. Imprecision of chloride measurement at these concentrations was 2.5% and 2.0%, respectively, over the length of the study. For purposes of this study we defined the LOQ as the molar quantity of chloride ion measurable with imprecision of less than 20% (CV) over 5 days. As shown in Fig. 1a, the LOQ for chloride ion is 0.4 μmoles. Using this limit, the minimum required sweat weight for precise analysis is a function of the desired LOQ of sweat chloride measurements (Fig. 1b) and was calculated as follows: Min: sweat weight ðmgÞ ¼
LOQ titratable chloride ion ðμmolÞ 1 mmol 106 mg water LOQ sweat chloride ðmmol=LÞ 1000 μmol 1 L water
ð2Þ In Eq. (2), the LOQ for chloride ion is determined empirically (0.4 μmol in our case), the LOQ of sweat chloride measurement is the desired lower limit of quantitating chloride concentration for diagnostic purposes, and the density of collected sweat is approximated to the density of water at room temperature. In our system, therefore, a sweat weight of 40 mg and 20 mg yield a lower LOQ equal to 10 mmol/L and 20 mmol/L, respectively. Both concentrations are well below the indeterminate
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
b)
100
200
Minimum required sweat weight (mg)
a)
80 CV (%)
445
60 40 20
160 120 80 40 0
0 0
1 2 3 4 Chloride ion on gauze (µmol)
0 20 40 60 80 100 Desired sweat chloride LOQ (mmol/L)
Fig. 1. Analytic sensitivity of chloride titration and derivation of minimum sweat weight. (a) The analytic sensitivity of the chloridometer used for sweat chloride testing was evaluated by repeated measurements of various known quantities of chloride ion. Imprecision of chloride measurement is expressed as the coefficient of variation (CV) in percent. (b) The required sweat weight as a function of the desired minimum limit of sweat chloride concentration.
concentration range of 30–60 mmol/L for infants less than six months of age and 40–60 mmol/L for patients greater than 6 months of age. Patient demographics and QNS rates The age of patients tested during the 21-month period ranged from 4 days to 65 years (median = 1.5 years). Of the 1397 sweat collections, 242 (17.3%) were b 75 mg. Following CLSI guidelines for calculating QNS rates, 6.7% of all patient encounters resulted in no valid sweat collection. For patients ≤ 3 months (n = 60) and N 3 months (n = 907) the QNS rates were 20% and 4.1%, respectively. Of the 23 patients with sweat chloride concentrations indicative of CF, 21 (2.3% of our patient population) were given a confirmatory diagnosis of CF after either genetic testing or multiple sweat chloride tests greater than 60 mmol/L. Of the remaining two patients, one was lost to follow up and the other was closely followed by clinicians at our hospital but was never diagnosed with CF. Biological and analytical variation The reference change value reflects the critical difference between two consecutive measurements that cannot be explained by the combination of biologic and analytic variability [14]. We assessed biologic variability in 284 patients with two or more sufficient sweat collections. The median within-subject biologic CV calculated from these individuals was 8.3% (mean 13.1%, 95% CI: 11.3–14.9%; range 0–88%) (Fig. 2). These data correspond to a reference change value of 36% (95% probability).
Fig. 2. Intra-individual biologic variation calculated from the sweat chloride concentrations determined from sufficient sweat collections. Variability of chloride concentration between 2 or more sufficient sweat collections (≥75 mg) from a single individual was calculated and expressed as coefficient of variation (CV) in percent. Distribution of this variability in 284 patients is displayed.
Sweat weight from arms and legs. The sweat weight collected from forearms (n = 781) was statistically greater (P b 0.0001) than that collected from upper legs (n = 458) (Fig. 3). This relationship was statistically significant among all age groups when both sufficient and insufficient collections were included (Fig. 3).
Sweat rate and chloride concentration The induction of a “maximal” sweat rate is often cited as the rationale for a minimum sweat weight of 75 mg in the Gibson-Cooke approach. We, therefore, assessed the chloride concentration across the wide range of sweat rates we encountered (Fig. 4). When all collections were considered (n = 1348), there was no clinically significant correlation between sweat weight and chloride concentration (r = −0.06). As specimens with chloride concentrations greater than 60 mmol/L may have had an undue influence on the correlation, these were removed in a separate analysis, which yielded similar results (r = −0.08; n = 1300). In order to avoid the influence of more imprecise measurements at very low sweat rates, we assessed the correlation of chloride concentration on sweat weight for specimens with greater than 20 mg sweat
Fig. 3. Comparison of sweat weights obtained from arms versus legs. Amount of sweat generated by location of pilocarpine administration and age is displayed. Bars represent the median and interquartile range of values and whiskers correspond to the 2.5th and 97.5th percentiles of the distribution. From left, the first two bars represent all sweat collections irrespective of adequacy requirements from the arm (n = 781) and leg (n = 458). Bars 3 and 4 display sweat weight distributions only from those sweat specimens of at least 75 mg from arms (n = 705) or legs (n = 333). Bars 5–8 display amount of generated sweat from arms (n = 707) and legs (n = 134) in patients N 6 months of age as well as arms (n = 76) and legs (n = 324) from patients ≤ 6 months of age. The final two bars display the amount of sweat generated from arms (n = 76), and legs (n = 125) from patients of all ages that did not meet the 75 mg weight requirement.
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
Chloride Ion (mmol/L)
446
140 120 100 80 60 40 20 0
0
200
400
600
800
1000
Sweat Weight (mg) Fig. 4. Negligible dependence of sweat chloride concentration on sweat rate. Sweat chloride concentration is plotted as a function of the amount of sweat generated in all 1358 specimens (r = −0.06).
(r = − 0.11; n = 1250) and those with greater than 40 mg sweat (r = − 0.12; n = 1195). These analyses indicate only a negligible dependence of electrolyte concentration on sweat rate. Sweat weight and diagnostic accuracy In the absence of a significant positive correlation between sweat weight and chloride concentration, we directly compared the analytic and diagnostic accuracy of QNS collections (b 75 mg) with sufficient collections (≥ 75 mg). There were 74 patients with at least one QNS collection and at least one subsequent sufficient collection. Again, in order to avoid the influence of high imprecision we restricted our analyses to 101 pairs of insufficient (from 20 and 74 mg) and sufficient collections (≥75 mg) from 60 patients. Fig. S1 displays the quantitative agreement between sweat chloride concentrations in QNS and sufficient specimens from the same patient. Chloride concentrations from insufficient specimens were highly correlated with sufficient collections: ClQNS = 0.97 × Clsufficient – 2.06, r = 0.86, P b 0.0001. Diagnostic agreement was also excellent (Table 2). There were no false negative or false positive diagnoses for CF if specimens weighing 20–74 mg were used to make the diagnosis (60 mmol/L cutoff). There were five instances of indeterminate classification of negative specimens and four instances of negative classification of indeterminate specimens. In each of these nine cases, chloride concentrations for sufficient specimens were near the cutoff between negative/indeterminate classifications and disagreement likely the result of measurement imprecision. Similarly, when comparing sweat chloride results between sufficient collections (≥ 75 mg of sweat) from the same patients (n = 284), there were 14 individuals for whom the interpretation of sweat chloride concentrations resulted in both indeterminate and negative classifications. Again, chloride concentrations from these specimens with adequate sweat weight were near the cutoff between negative/ indeterminate classifications. Discussion In one of the seminal papers on CF testing from 1963, Gibson and di Sant-Agnese observed that sweat electrolyte concentration is related to Table 2 Diagnostic concordance for patients (n = 60) with at least one sufficient and one QNS result ≥ 20 mg. Sufficient collection (≥75 mg) QNS (20–74 mg)
CF unlikely
Indeterminate
Indicative of CF
CF unlikely Indeterminate Indicative of CF
41 5 0
4 3 0
0 0 7
sweat rate [15]. These authors noted in their cohort of 29 individuals (10 of which had CF), that for the majority of these individuals, sweat electrolyte concentrations were greatest at higher sweat rates. While this experiment did not specifically examine chloride concentrations as a function of sweat rate, data from this experiment has been cited as the evidence for a positive correlation between sweat rate and chloride ion concentration and the reason for setting the minimum allowable sweat rate at 75 mg over 30 min (using 2 × 2-inch gauze or filter paper) so to as achieve maximally stimulated sweat glands [11]. In the current study specifically examining chloride concentration, we observed a negligible negative correlation between sweat rate and chloride ion concentration and thus no support for the concept of “maximally” stimulating sweat glands for testing. With no evidence of correlation between sweat rate and chloride ion concentration, we evaluated the analytic and diagnostic accuracy of sweat collections less than 75 mg. There was excellent analytic and diagnostic concordance between specimens considered adequate by current criteria and those currently considered inadequate (20 mg to 75 mg). The ability to use collections of less than 75 mg for clinical purposes would obviate the need to discard viable specimens, prevent unnecessary delays in diagnosis, and decrease parental stress associated with repeat testing. Replication of this study at other centers, in a large number of specimens, is certainly indicated. Our center currently maintains the 75 mg requirement (CF Foundation standards); but we hope that these data prompt a re-evaluation of minimum sweat requirements in order to salvage clinical data associated with collections heretofore thought to be “insufficient.” Concerning specimen adequacy, we also observed that the chosen sweat site can dramatically alter QNS rates. The CF Foundation guidelines recommend duplicate testing in a single visit to increase the likelihood of a sufficient collection during the current patient encounter. As there are no data to show which combination of testing sites (upper leg and forearm) result in the lowest QNS rates, we wanted to explore this in our patient population. We found that arms generate more sweat than legs, independent of patient age. Based on these results we altered our testing center's protocol from duplicate testing on an arm and leg, to both arms, whenever possible. During the 21-month study period we did not meet CF Foundation target QNS rates of b10% for patients ≤3 months of age, and b5% for patients N3 months of age. For the younger group, a large percentage of these tests were done on acutely ill inpatients likely leading to the inflated QNS rate. Based on the results of this study and general observations, we made a few key modifications to the sweat testing policy at our hospital: (1) sweat testing is limited to outpatients (exception for inpatients that can safely ambulate to the outpatient testing center), (2) patients must remain in the outpatient testing center post-iontophoresis until the gauze is collected, (3) testing is to be performed in duplicate at each visit on both arms, whenever possible. Since implementing these protocols, CLSI-defined QNS rates dropped from 20% to 3.3% for patients ≤ 3
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
months of age and from 4.1% to 0.36% for patients N 3 months of age (as measured from the 8 months following the change in protocol). Lastly, we were interested in evaluating the capacity of sweat chloride testing to reflect changes in severity of the disease. Historically sweat chloride testing has been largely used in a diagnostic capacity, but with the advent of disease modifying therapy (e.g. ivacaftor), its role may expand to monitoring response to therapy [5]. As an objective assessment of the significance of differences in serial sweat chloride results from individuals, we calculated the reference change value. A patient's sweat chloride concentration can be considered to have undergone a statistically significant change if the concentration in serial measurements changes more than ± 36%. This is not to say that changes b 36% cannot reflect meaningful alterations in the patient's status, but that the interpretation of such a change may be obscured by variations from analytical and non-disease-related biological factors. As sweat chloride concentration is the best currently available marker of CFTR function, an objective interpretation of patient results (such as with a reference change value) is needed when evaluating the utility of sweat chloride testing to monitor response to therapy. It should also be noted that while sweat chloride concentration reflects CFTR function, it does not necessarily reflect lung dysfunction [16] and other disease manifestations. Moreover, given that a large change in sweat chloride concentration is required to meet statistical significance, there is a need for other markers of disease pathology that could assist in monitoring incremental and/or early changes as a result of therapy. Conflict of interest The authors have no potential, perceived or real conflicts of interest. The authors did not receive any financial assistance to conduct this study or write this manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.clinbiochem.2014.12.011.
447
References [1] Farrell PM, Rosenstein BJ, White TB, Accurso FJ, Castellani C, Cutting GR, et al. Guidelines for diagnosis of cystic fibrosis in newborns through older adults: cystic fibrosis foundation consensus report. J Pediatr 2008;153:S4–S14. [2] Wagener JS, Zemanick ET, Sontag MK. Newborn screening for cystic fibrosis. Curr Opin Pediatr 2012;24:329–35. [3] Kirk JM. A continuing role for sweat testing in an era of newborn screening for cystic fibrosis. Clin Biochem 2011;44:487–8. [4] Eckford PDW, Li C, Ramjeesingh M, Bear CE. Cystic fibrosis transmembrane conductance regulator (CFTR) potentiator VX-770 (ivacaftor) opens the defective channel gate of mutant cftr in a phosphorylation-dependent but atp-independent manner. J Biol Chem 2012;287:36639–49. [5] Ramsey BW, Davies J, McElvaney NG, Tullis E, Bell SC, Dřevínek P, et al. A cftr potentiator in patients with cystic fibrosis and the G551D mutation. N Engl J Med 2011; 365:1663–72. [6] Legrys VA, McColley SA, Li Z, Farrell PM. The need for quality improvement in sweat testing infants after newborn screening for cystic fibrosis. J Pediatr 2010;157: 1035–7. [7] Eng W, LeGrys VA, Schechter MS, Laughon MM, Barker PM. Sweat-testing in preterm and full-term infants less than 6 weeks of age. Pediatr Pulmonol 2005;40:64–7. [8] Sands D, Oltarzewski M, Nowakowska A, Zybert K. Bilateral sweat tests with two different methods as a part of cystic fibrosis newborn screening (CF NBS) protocol and additional quality control. Folia Histochem Cytobiol/Pol Acad Sci, Pol Histochem Cytochem Soc 2010;48:358–65. [9] 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. [10] Goldberg S, Schwartz S, Francis M, Stankiewicz H, Izbicki G, Picard E. Does sweat volume influence the sweat test result? Arch Dis Child 2010;95:377–81. [11] CLSI. Sweat testing: sample collection and quantitative chloride analysis; approved guideline. CLSI document C34-A3. 3rd ed. Clinical and Laboratory Standards Institute; 2009. [12] LeGrys VA, Yankaskas JR, Quittell LM, Marshall BC, Mogayzel Jr PJ. Diagnostic sweat testing: the cystic fibrosis foundation guidelines. J Pediatr 2007;151:85–9. [13] Holmes DT. cp-R, an interface the R programming language for clinical laboratory method comparisons. Clinical Biochemistry. [14] Harris EK, Yasaka T. On the calculation of a “reference change” for comparing two consecutive measurements. Clin Chem 1983;29:25–30. [15] Gibson LE, Disant Agnese PA. Studies of salt excretion in sweat. Relationships between rate, conductivity, and electrolyte composition of sweat from patients with cystic fibrosis and from control subjects. J Pediatr 1963;62:855–67. [16] Barry PJ, Jones AM, Webb AK, Horsley AR. Sweat chloride is not a useful marker of clinical response to ivacaftor. Thorax 2014;69:586–7.
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Sweating the small stuff: Adequacy and accuracy in sweat chloride determination☆ Mari L. DeMarco a,⁎, Dennis J. Dietzen b,c, Sarah M. Brown b,c a b c
Department of Pathology and Laboratory Medicine, St. Paul's Hospital, University of British Columbia, Vancouver, Canada Department of Pathology and Immunology, Washington University School of Medicine, St Louis, MO, USA Department of Pediatrics, Washington University School of Medicine, St Louis, MO, USA
a r t i c l e
i n f o
Article history: Received 5 November 2014 Received in revised form 9 December 2014 Accepted 10 December 2014 Available online 18 December 2014 Keywords: Sweat chloride Sweat rate Cystic fibrosis Biological variability Limit of quantitation
a b s t r a c t Objectives: Sweat chloride testing is the gold standard for diagnosis of cystic fibrosis (CF). Our objectives were to: 1) describe variables that determine sweat rate; 2) determine the analytic and diagnostic capacity of sweat chloride analysis across the range of observed sweat rates; and 3) determine the biologic variability of sweat chloride concentration. Methods: A retrospective analysis was performed using data from all sweat chloride tests performed at St. Louis Children's Hospital over a 21-month period. Results: A total of 1397 sweat chloride tests (1155 sufficient [≥75 mg], 242 insufficient [b75 mg]), were performed on 904 individuals. The sweat weight collected from forearms was statistically greater than that collected from legs. There was a negligible correlation between sweat weight and chloride concentration (r = −0.06). The mean individual biologic CV calculated from individuals with two or more sweat collections ≥ 75 mg was 13.1% (95% CI: 11.3–14.9%; range 0–88%) yielding a reference change value of 36%. Using 60 mmol/L as the diagnostic chloride cutoff, 100% of CF cases were detected whether a minimum sweat weight of 75, 40, or 20 mg was required. Conclusions: 1) Collection of sweat from forearms is preferable to upper legs, particularly in very young infants; 2) sweat chloride concentrations are not highly dependent upon sweat rate; 3) a change in sweat chloride concentration exceeding 36% may be considered a clinically significant response to cystic fibrosis transmembrane receptor targeted therapy, and 4) sweat collections of less than 75 mg provide clinically accurate information. © 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Sweat chloride analysis is the gold standard for diagnosis of cystic fibrosis (CF) [1]. Newborn screening for CF using immunoreactive trypsinogen is performed in all 50 of the United States and many other countries [2]. Positive newborn screens are confirmed with sweat chloride analysis [3]. With the advent of cystic fibrosis transmembrane receptor (CFTR)-modifying therapeutics such as ivacaftor [4], sweat chloride analysis may also be used to monitor response to therapy [5]. Sweat chloride analysis is a multi-step process that requires: 1) stimulation of sweat production with a cholinergic agonist (e.g., pilocarpine); 2) collection of sweat using filter paper, gauze, or plastic capillary tubing; and 3) chloride analysis, most commonly achieved by coulometric titration. Despite over 50 years of clinical use, some questions that remain are: 1) how much sweat is necessary for diagnosis, 2) which sites are Abbreviations: CF, cystic fibrosis; CFTR, cystic fibrosis transmembrane receptor; QNS, quantity not sufficient; CLSI, Clinical and Laboratory Standards Institute. ☆ This work was presented in part at the American Association for Clinical Chemistry Annual Meeting, Houston, TX, July 29, 2013. ⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, St Paul's Hospital, 1081 Burrard St, Vancouver BC V6Z 1Y6, Canada. E-mail address: [email protected] (M.L. DeMarco).
most responsive to sweat stimulation, 3) what is the biologic and analytical variation associated with sweat chloride analysis, and 4) what is the capacity for sweat chloride testing to reflect changes in severity of disease? There are minimum sweat volume requirements for collection by gauze and plastic capillary methods; 75 mg of sweat for the former and 15 μl of sweat by the latter. Collections less than the minimum (QNS) are considered inappropriate for diagnosis. QNS rates are especially high in newborns [6,7]. Bilateral stimulation is encouraged as a means of decreasing QNS rates [8] but collections from multiple sites must be analyzed independently as opposed to pooling specimens. The site for sweat induction may be an important determinant of sweat production. Common sites for stimulation include the leg, arm, or a combination of a leg and an arm. While it is suggested that one area may produce more sweat than another (e.g., forearm versus thigh) due to variations in sweat gland density, this hypothesis has not been rigorously examined. According to the Clinical and Laboratory Standards Institute (CLSI) guidelines, the current 75 mg sweat requirement for gauze/paper methods is based on a sweat rate of 1 g/m2/min. In the original paper by Gibson and Cooke, the average sweat weight collected on 2.5 cm diameter filter paper disks was 76 mg, range 18–135 mg [9]. Regardless of
http://dx.doi.org/10.1016/j.clinbiochem.2014.12.011 0009-9120/© 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
444
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
amount of sweat collected, no patient with CF had a sweat chloride concentration less than 80 mmol/L, and none of the controls had concentrations greater than 60 mmol/L. Goldberg et al. showed no correlation between sweat weight and sweat chloride concentration (n = 463, R2 = 0.002, p N 0.05) [10]. Our extensive experience as a CF testing center suggests that a lower sweat weight requirement that does not compromise diagnostic accuracy would facilitate rapid results and decrease parental stress associated with a delayed diagnosis. Finally, the variability in sweat chloride concentration between two different collections can be considerable. This imprecision almost certainly is derived from the sweat collection process, as chloride titration is precise (CV b 2%) at diagnostically relevant concentrations. Assessing the variation in this testing system is important for understanding differences between repeat values. This intra-individual variance, which has not been previously determined, will become increasingly important as sweat chloride is adopted as a biomarker of therapeutic effect. This study, therefore, examined the relationship between sweat rate and sweat chloride concentration, the contribution of intraindividual variation and variability in the sweat collection system to the imprecision of the sweat chloride measurement, and the relative capacity of arms and legs to produce sufficient sweat for analytical testing. Analytical considerations in this study are restricted to the Gibson and Cooke (gauze) method of sweat collection, although findings from analysis of sweat site capacity may have broader implications. Materials & Methods Sweat chloride analysis Quantitative pilocarpine iontophoresis sweat chloride testing was performed in the St. Louis Children's Hospital Core Laboratory following the Gibson and Cooke Technique [9]. Laboratory protocols follow the standards of the Clinical Laboratory Standards Institute CLSI-34 A3 guidelines [11]. Briefly, sweat was collected on pre-weighed 2×2-inch sterile gauze pads (Kendall Curity #3381, Marshfield, MA) over 30 minutes following pilocarpine nitrate iontophoresis (Gibson-Cooke Sweat Test Apparatus, Farrall Instruments, Grand Island, NE). Chloride ion concentration was determined by coulometric titration using a Biodynamics LyteTek-Cl Chloride Analyzer. Whenever feasible, sweat collections from two sites (upper leg and forearm) were performed at each patient encounter. According to CF Foundation and CLSI guidelines, the minimum sweat weight currently required for a sufficient chloride analysis is 75 mg but for purposes of this study, insufficient collections were retained and analyzed but not communicated to ordering physicians. This study was conducted under approval from the Washington University Human Research Protection Office. Limit of chloride quantitation The limit of chloride quantitation of the chloridometer was evaluated by adding known amounts of NaCl in solution (0.1, 0.4, 0.7, 1, 2, 3, and 4 μmoles of chloride) to pre-weighed gauze pads. Samples were tested in duplicate for 5 consecutive days. For purposes of this study the limit of quantitation (LOQ) was defined as the lowest amount of chloride detectable with an imprecision of less that 20% (CV) over at least 5 days of testing. Study design Analysis was performed using data from all sweat chloride tests performed at St. Louis Children's Hospital, a designated CF Center, over a 21-month period (January 2011 to September 2012). A total of 1397 sweat chloride measurements (1155 specimens ≥ 75 mg and 242 specimens b 75 mg), were performed on 904 individuals (Table 1). Variables included in the data analysis were: (1) age, (2) sweat site (forearm v. upper leg), (3) patient location (outpatient
Table 1 Descriptive statistics for sweat chloride testing performed at St. Louis Children's Hospital from January 2011 to September 2012. Number
Category
1397 1155 242 904 967 907 60
Sweat chloride tests Sufficient sweat collections QNS sweat collections Patients Patient encounters Patient encounters (age N 3 months) Patient encounters (age ≤ 3 months)
v. inpatient), (4) sweat weight, (5) sweat chloride concentration, and (6) confirmatory diagnosis of CF (genetic testing and/or two positive sweat chloride tests). CF Foundation interpretive guidelines for sweat chloride results were utilized (Table S1) [12]. Statistical analysis Descriptive statistics were calculated using median and ranges for continuous variables and percentages for categorical variables. The Mann–Whitney U test was used to examine the difference in sweat weights from different collection sites (forearms v. upper legs). Data analyses were performed with IBM SPSS version 20.0 (Armonk, NY), and GraphPad Prism version 6.0 (La Jolla, CA). Deming regression was performed with R version 3.1.1 (The R Foundation for Statistical Computing) and the cp-R interface [13]. The reference change value was calculated with 95% confidence as follows: 2 2 RCV ¼ √2 1:96 √ CV A þ CV I
ð1Þ
where CVA is the analytical coefficient of variation and CVI is the intraindividual biological coefficient of variation and as measured in the study, may contain contributions from variability in the sweat collection process. Results Imprecision of chloride measurement and limit of chloride quantitation In the case of sweat chloride measurement, imprecision and LOQ is dictated by both the analytic instrument and the amount of chloride presented. Our daily quality control regimen consists of separate 200 μL solutions containing 45 and 75 mmol/L NaCl with 9 and 15 μmoles of chloride ion, respectively. Imprecision of chloride measurement at these concentrations was 2.5% and 2.0%, respectively, over the length of the study. For purposes of this study we defined the LOQ as the molar quantity of chloride ion measurable with imprecision of less than 20% (CV) over 5 days. As shown in Fig. 1a, the LOQ for chloride ion is 0.4 μmoles. Using this limit, the minimum required sweat weight for precise analysis is a function of the desired LOQ of sweat chloride measurements (Fig. 1b) and was calculated as follows: Min: sweat weight ðmgÞ ¼
LOQ titratable chloride ion ðμmolÞ 1 mmol 106 mg water LOQ sweat chloride ðmmol=LÞ 1000 μmol 1 L water
ð2Þ In Eq. (2), the LOQ for chloride ion is determined empirically (0.4 μmol in our case), the LOQ of sweat chloride measurement is the desired lower limit of quantitating chloride concentration for diagnostic purposes, and the density of collected sweat is approximated to the density of water at room temperature. In our system, therefore, a sweat weight of 40 mg and 20 mg yield a lower LOQ equal to 10 mmol/L and 20 mmol/L, respectively. Both concentrations are well below the indeterminate
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
b)
100
200
Minimum required sweat weight (mg)
a)
80 CV (%)
445
60 40 20
160 120 80 40 0
0 0
1 2 3 4 Chloride ion on gauze (µmol)
0 20 40 60 80 100 Desired sweat chloride LOQ (mmol/L)
Fig. 1. Analytic sensitivity of chloride titration and derivation of minimum sweat weight. (a) The analytic sensitivity of the chloridometer used for sweat chloride testing was evaluated by repeated measurements of various known quantities of chloride ion. Imprecision of chloride measurement is expressed as the coefficient of variation (CV) in percent. (b) The required sweat weight as a function of the desired minimum limit of sweat chloride concentration.
concentration range of 30–60 mmol/L for infants less than six months of age and 40–60 mmol/L for patients greater than 6 months of age. Patient demographics and QNS rates The age of patients tested during the 21-month period ranged from 4 days to 65 years (median = 1.5 years). Of the 1397 sweat collections, 242 (17.3%) were b 75 mg. Following CLSI guidelines for calculating QNS rates, 6.7% of all patient encounters resulted in no valid sweat collection. For patients ≤ 3 months (n = 60) and N 3 months (n = 907) the QNS rates were 20% and 4.1%, respectively. Of the 23 patients with sweat chloride concentrations indicative of CF, 21 (2.3% of our patient population) were given a confirmatory diagnosis of CF after either genetic testing or multiple sweat chloride tests greater than 60 mmol/L. Of the remaining two patients, one was lost to follow up and the other was closely followed by clinicians at our hospital but was never diagnosed with CF. Biological and analytical variation The reference change value reflects the critical difference between two consecutive measurements that cannot be explained by the combination of biologic and analytic variability [14]. We assessed biologic variability in 284 patients with two or more sufficient sweat collections. The median within-subject biologic CV calculated from these individuals was 8.3% (mean 13.1%, 95% CI: 11.3–14.9%; range 0–88%) (Fig. 2). These data correspond to a reference change value of 36% (95% probability).
Fig. 2. Intra-individual biologic variation calculated from the sweat chloride concentrations determined from sufficient sweat collections. Variability of chloride concentration between 2 or more sufficient sweat collections (≥75 mg) from a single individual was calculated and expressed as coefficient of variation (CV) in percent. Distribution of this variability in 284 patients is displayed.
Sweat weight from arms and legs. The sweat weight collected from forearms (n = 781) was statistically greater (P b 0.0001) than that collected from upper legs (n = 458) (Fig. 3). This relationship was statistically significant among all age groups when both sufficient and insufficient collections were included (Fig. 3).
Sweat rate and chloride concentration The induction of a “maximal” sweat rate is often cited as the rationale for a minimum sweat weight of 75 mg in the Gibson-Cooke approach. We, therefore, assessed the chloride concentration across the wide range of sweat rates we encountered (Fig. 4). When all collections were considered (n = 1348), there was no clinically significant correlation between sweat weight and chloride concentration (r = −0.06). As specimens with chloride concentrations greater than 60 mmol/L may have had an undue influence on the correlation, these were removed in a separate analysis, which yielded similar results (r = −0.08; n = 1300). In order to avoid the influence of more imprecise measurements at very low sweat rates, we assessed the correlation of chloride concentration on sweat weight for specimens with greater than 20 mg sweat
Fig. 3. Comparison of sweat weights obtained from arms versus legs. Amount of sweat generated by location of pilocarpine administration and age is displayed. Bars represent the median and interquartile range of values and whiskers correspond to the 2.5th and 97.5th percentiles of the distribution. From left, the first two bars represent all sweat collections irrespective of adequacy requirements from the arm (n = 781) and leg (n = 458). Bars 3 and 4 display sweat weight distributions only from those sweat specimens of at least 75 mg from arms (n = 705) or legs (n = 333). Bars 5–8 display amount of generated sweat from arms (n = 707) and legs (n = 134) in patients N 6 months of age as well as arms (n = 76) and legs (n = 324) from patients ≤ 6 months of age. The final two bars display the amount of sweat generated from arms (n = 76), and legs (n = 125) from patients of all ages that did not meet the 75 mg weight requirement.
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
Chloride Ion (mmol/L)
446
140 120 100 80 60 40 20 0
0
200
400
600
800
1000
Sweat Weight (mg) Fig. 4. Negligible dependence of sweat chloride concentration on sweat rate. Sweat chloride concentration is plotted as a function of the amount of sweat generated in all 1358 specimens (r = −0.06).
(r = − 0.11; n = 1250) and those with greater than 40 mg sweat (r = − 0.12; n = 1195). These analyses indicate only a negligible dependence of electrolyte concentration on sweat rate. Sweat weight and diagnostic accuracy In the absence of a significant positive correlation between sweat weight and chloride concentration, we directly compared the analytic and diagnostic accuracy of QNS collections (b 75 mg) with sufficient collections (≥ 75 mg). There were 74 patients with at least one QNS collection and at least one subsequent sufficient collection. Again, in order to avoid the influence of high imprecision we restricted our analyses to 101 pairs of insufficient (from 20 and 74 mg) and sufficient collections (≥75 mg) from 60 patients. Fig. S1 displays the quantitative agreement between sweat chloride concentrations in QNS and sufficient specimens from the same patient. Chloride concentrations from insufficient specimens were highly correlated with sufficient collections: ClQNS = 0.97 × Clsufficient – 2.06, r = 0.86, P b 0.0001. Diagnostic agreement was also excellent (Table 2). There were no false negative or false positive diagnoses for CF if specimens weighing 20–74 mg were used to make the diagnosis (60 mmol/L cutoff). There were five instances of indeterminate classification of negative specimens and four instances of negative classification of indeterminate specimens. In each of these nine cases, chloride concentrations for sufficient specimens were near the cutoff between negative/indeterminate classifications and disagreement likely the result of measurement imprecision. Similarly, when comparing sweat chloride results between sufficient collections (≥ 75 mg of sweat) from the same patients (n = 284), there were 14 individuals for whom the interpretation of sweat chloride concentrations resulted in both indeterminate and negative classifications. Again, chloride concentrations from these specimens with adequate sweat weight were near the cutoff between negative/ indeterminate classifications. Discussion In one of the seminal papers on CF testing from 1963, Gibson and di Sant-Agnese observed that sweat electrolyte concentration is related to Table 2 Diagnostic concordance for patients (n = 60) with at least one sufficient and one QNS result ≥ 20 mg. Sufficient collection (≥75 mg) QNS (20–74 mg)
CF unlikely
Indeterminate
Indicative of CF
CF unlikely Indeterminate Indicative of CF
41 5 0
4 3 0
0 0 7
sweat rate [15]. These authors noted in their cohort of 29 individuals (10 of which had CF), that for the majority of these individuals, sweat electrolyte concentrations were greatest at higher sweat rates. While this experiment did not specifically examine chloride concentrations as a function of sweat rate, data from this experiment has been cited as the evidence for a positive correlation between sweat rate and chloride ion concentration and the reason for setting the minimum allowable sweat rate at 75 mg over 30 min (using 2 × 2-inch gauze or filter paper) so to as achieve maximally stimulated sweat glands [11]. In the current study specifically examining chloride concentration, we observed a negligible negative correlation between sweat rate and chloride ion concentration and thus no support for the concept of “maximally” stimulating sweat glands for testing. With no evidence of correlation between sweat rate and chloride ion concentration, we evaluated the analytic and diagnostic accuracy of sweat collections less than 75 mg. There was excellent analytic and diagnostic concordance between specimens considered adequate by current criteria and those currently considered inadequate (20 mg to 75 mg). The ability to use collections of less than 75 mg for clinical purposes would obviate the need to discard viable specimens, prevent unnecessary delays in diagnosis, and decrease parental stress associated with repeat testing. Replication of this study at other centers, in a large number of specimens, is certainly indicated. Our center currently maintains the 75 mg requirement (CF Foundation standards); but we hope that these data prompt a re-evaluation of minimum sweat requirements in order to salvage clinical data associated with collections heretofore thought to be “insufficient.” Concerning specimen adequacy, we also observed that the chosen sweat site can dramatically alter QNS rates. The CF Foundation guidelines recommend duplicate testing in a single visit to increase the likelihood of a sufficient collection during the current patient encounter. As there are no data to show which combination of testing sites (upper leg and forearm) result in the lowest QNS rates, we wanted to explore this in our patient population. We found that arms generate more sweat than legs, independent of patient age. Based on these results we altered our testing center's protocol from duplicate testing on an arm and leg, to both arms, whenever possible. During the 21-month study period we did not meet CF Foundation target QNS rates of b10% for patients ≤3 months of age, and b5% for patients N3 months of age. For the younger group, a large percentage of these tests were done on acutely ill inpatients likely leading to the inflated QNS rate. Based on the results of this study and general observations, we made a few key modifications to the sweat testing policy at our hospital: (1) sweat testing is limited to outpatients (exception for inpatients that can safely ambulate to the outpatient testing center), (2) patients must remain in the outpatient testing center post-iontophoresis until the gauze is collected, (3) testing is to be performed in duplicate at each visit on both arms, whenever possible. Since implementing these protocols, CLSI-defined QNS rates dropped from 20% to 3.3% for patients ≤ 3
M.L. DeMarco et al. / Clinical Biochemistry 48 (2015) 443–447
months of age and from 4.1% to 0.36% for patients N 3 months of age (as measured from the 8 months following the change in protocol). Lastly, we were interested in evaluating the capacity of sweat chloride testing to reflect changes in severity of the disease. Historically sweat chloride testing has been largely used in a diagnostic capacity, but with the advent of disease modifying therapy (e.g. ivacaftor), its role may expand to monitoring response to therapy [5]. As an objective assessment of the significance of differences in serial sweat chloride results from individuals, we calculated the reference change value. A patient's sweat chloride concentration can be considered to have undergone a statistically significant change if the concentration in serial measurements changes more than ± 36%. This is not to say that changes b 36% cannot reflect meaningful alterations in the patient's status, but that the interpretation of such a change may be obscured by variations from analytical and non-disease-related biological factors. As sweat chloride concentration is the best currently available marker of CFTR function, an objective interpretation of patient results (such as with a reference change value) is needed when evaluating the utility of sweat chloride testing to monitor response to therapy. It should also be noted that while sweat chloride concentration reflects CFTR function, it does not necessarily reflect lung dysfunction [16] and other disease manifestations. Moreover, given that a large change in sweat chloride concentration is required to meet statistical significance, there is a need for other markers of disease pathology that could assist in monitoring incremental and/or early changes as a result of therapy. Conflict of interest The authors have no potential, perceived or real conflicts of interest. The authors did not receive any financial assistance to conduct this study or write this manuscript. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.clinbiochem.2014.12.011.
447
References [1] Farrell PM, Rosenstein BJ, White TB, Accurso FJ, Castellani C, Cutting GR, et al. Guidelines for diagnosis of cystic fibrosis in newborns through older adults: cystic fibrosis foundation consensus report. J Pediatr 2008;153:S4–S14. [2] Wagener JS, Zemanick ET, Sontag MK. Newborn screening for cystic fibrosis. Curr Opin Pediatr 2012;24:329–35. [3] Kirk JM. A continuing role for sweat testing in an era of newborn screening for cystic fibrosis. Clin Biochem 2011;44:487–8. [4] Eckford PDW, Li C, Ramjeesingh M, Bear CE. Cystic fibrosis transmembrane conductance regulator (CFTR) potentiator VX-770 (ivacaftor) opens the defective channel gate of mutant cftr in a phosphorylation-dependent but atp-independent manner. J Biol Chem 2012;287:36639–49. [5] Ramsey BW, Davies J, McElvaney NG, Tullis E, Bell SC, Dřevínek P, et al. A cftr potentiator in patients with cystic fibrosis and the G551D mutation. N Engl J Med 2011; 365:1663–72. [6] Legrys VA, McColley SA, Li Z, Farrell PM. The need for quality improvement in sweat testing infants after newborn screening for cystic fibrosis. J Pediatr 2010;157: 1035–7. [7] Eng W, LeGrys VA, Schechter MS, Laughon MM, Barker PM. Sweat-testing in preterm and full-term infants less than 6 weeks of age. Pediatr Pulmonol 2005;40:64–7. [8] Sands D, Oltarzewski M, Nowakowska A, Zybert K. Bilateral sweat tests with two different methods as a part of cystic fibrosis newborn screening (CF NBS) protocol and additional quality control. Folia Histochem Cytobiol/Pol Acad Sci, Pol Histochem Cytochem Soc 2010;48:358–65. [9] 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. [10] Goldberg S, Schwartz S, Francis M, Stankiewicz H, Izbicki G, Picard E. Does sweat volume influence the sweat test result? Arch Dis Child 2010;95:377–81. [11] CLSI. Sweat testing: sample collection and quantitative chloride analysis; approved guideline. CLSI document C34-A3. 3rd ed. Clinical and Laboratory Standards Institute; 2009. [12] LeGrys VA, Yankaskas JR, Quittell LM, Marshall BC, Mogayzel Jr PJ. Diagnostic sweat testing: the cystic fibrosis foundation guidelines. J Pediatr 2007;151:85–9. [13] Holmes DT. cp-R, an interface the R programming language for clinical laboratory method comparisons. Clinical Biochemistry. [14] Harris EK, Yasaka T. On the calculation of a “reference change” for comparing two consecutive measurements. Clin Chem 1983;29:25–30. [15] Gibson LE, Disant Agnese PA. Studies of salt excretion in sweat. Relationships between rate, conductivity, and electrolyte composition of sweat from patients with cystic fibrosis and from control subjects. J Pediatr 1963;62:855–67. [16] Barry PJ, Jones AM, Webb AK, Horsley AR. Sweat chloride is not a useful marker of clinical response to ivacaftor. Thorax 2014;69:586–7.
