| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Medicine, 2 Department of Physiology and Biophysics, 3 Department of Pediatrics, and 4 Cystic Fibrosis Research Center, University of Alabama at Birmingham, Birmingham, Alabama; and 5 Department of Pediatrics, University of Colorado, Denver, Colorado
Correspondence and requests for reprints should be addressed to Steven M. Rowe, M.D., M.S.P.H., University of Alabama at Birmingham, THT 215, 1900 University Boulevard, Birmingham, AL 35294-0006. E-mail: smrowe{at}uab.edu
ABSTRACT
Advances in our understanding of cystic fibrosis pathogenesis have led to strategies directed toward treatment of underlying causes of the disease rather than treatments of disease-related symptoms. To expedite evaluation of these emerging therapies, early-phase clinical trials require extension of in vivo cystic fibrosis transmembrane conductance regulator (CFTR)–detecting assays to multicenter trial formats, including nasal potential difference and sweat chloride measurements. Both of these techniques can be used to fulfill diagnostic criteria for the disease, and can discriminate various levels of CFTR function. Full realization of these assays in multicenter clinical trials requires identification of sources of nonbiological intra- and intersite variability, and careful attention to study design and statistical analysis of study-generated data. In this review, we discuss several issues important to the performance of these assays, including efforts to identify and address aspects that can contribute to inconsistent and/or potentially erroneous results. Adjunctive means of detecting CFTR including mRNA expression, immunocytochemical localization, and other methods are also discussed. Recommendations are presented to advance our understanding of these biomarkers and to improve their capacity to predict cystic fibrosis outcomes.
Key Words: ion transport • cystic fibrosis transmembrane conductance regulator biomarkers • cystic fibrosis therapy • nasal potential difference • sweat chloride
Advancements in our understanding of cystic fibrosis (CF) pathogenesis have identified therapeutic targets that correct underlying genetic or ion transport defects. The evaluation and prioritization of new treatment strategies (such as gene transfer and cystic fibrosis transmembrane conductance regulator [CFTR]–activating agents) will require accurate assays to measure CFTR activity in vivo to demonstrate biological activity and facilitate drug development (1). This article reviews detection and quantification of CFTR-dependent and -independent ion transport in proof-of-mechanism, early-phase CF clinical trials, focusing on the transepithelial nasal potential difference (NPD) assay and measurements of sweat electrolytes as biomarkers of CFTR activity.
NASAL POTENTIAL DIFFERENCE
Clinical and Biological Relevance
On the basis of current working models of CF lung disease pathogenesis, the NPD has clinical and biological relevance. The PD is specific to ion transport abnormalities within the causal pathway of lung disease, and clear differences between findings in normal subjects and patients with CF provide a guide for anticipated NPD parameters that would reflect biological activity. The transepithelial NPD measurement, first described by Knowles and coworkers, estimates the net ion conductance across the nasal airway epithelium and reveals bioelectric hallmarks associated with CF (2–4). Measurement of the PD while sequentially perfusing compounds that isolate sodium transport (Ringer's PD and the change [inhibition] in PD after amiloride perfusion) and chloride transport (perfusion with zero chloride solution and CFTR-activating agents such as isoproterenol, terbutaline, genistein, or adenosine; and/or ATP to stimulate CFTR-independent Cl– transport) allows estimates of Na+ and Cl– conductance, respectively (5–7). Patients with CF exhibit enhanced Na+ absorption, reduced or absent CFTR-mediated Cl– secretion, and enhanced ATP-stimulated Cl– transport, all of which have been validated in studies of human airway tissues ex vivo. These characteristic NPD findings can fulfill diagnostic criteria for CF, particularly when other assessments are inconclusive, although normal values are currently institutional and have not been established outside of defined research protocols (8, 9).
The NPD can distinguish phenotypic severities of CF and associated CFTR dysfunction in single-center study designs. Wilschanski and colleagues compared normal individuals with clinically normal heterozygotes, subjects with congenital bilateral absence of the vas deferens due to CFTR mutations, and patients with CF subdivided by pancreatic function. NPD parameters (including both Na+ and Cl– transport) discriminated between the CF phenotypes (and levels of CFTR function) in aggregate, although there was significant overlap, particularly in subjects with mild phenotypes (10). These studies and those of Knowles, Alton, and others provide predictions of clinically meaningful changes in NPD parameters in response to CFTR-active agents, Na+ transport inhibitors, and/or activators of Cl– conductance (2, 7, 9, 11). The effects of experimental agents (i.e., CFTR correctors and potentiators) on ion transport are expected to be rapid in general, based on the known kinetics of CFTR transcription, translation, membrane trafficking, epithelial surface half-life, and other in vitro studies (12, 13). Other cross-sectional studies have shown correlation between the NPD and disease severity (including pancreatic function and pulmonary status), although Thomas and coworkers reported that the latter correlation was limited to men (14, 15). Fajac and coworkers showed that Cl– conductance correlated with pancreatic status (but not lung function), while measures of Na+ transport correlated with FEV1 (16). Overall, inconsistencies in the relationship between NPD measures and genotype and/or phenotype seen across studies (14, 17, 18) likely reflect the limited correlation between phenotype and ion transport due to CFTR-independent effects, and other sources of phenotypic variability (such as environment or modifier genes). Underpowered studies and technical limitations of the NPD also contribute to the disparity.
Taken together, NPD measurements can distinguish predicted CFTR activity in aggregate if conducted in a rigorous and controlled fashion with an adequate sample size, although significant overlap between phenotypes and population differences may impair detection of low-level CFTR activity.
Feasibility and More Recent Experience
NPD measurements are safe and well tolerated, but require a cooperative patient who can tolerate manipulation of the nasal catheter or, in the case of infants, conscious sedation. Most patients with CF tolerate the procedure as long as severe nasal diseases (e.g., polyposis or nasal inflammation) do not interfere with the assay. The procedure itself requires approximately 30 minutes for the study subject; setup and maintenance of dedicated equipment by experienced and trained personnel is an additional requisite, and the study must be performed in a quiet atmosphere with minimal electrical interference. Costs are substantial, related to use of disposable equipment, ongoing support of NPD operators and laboratory support personnel, and institutional commitments of dedicated space. Because of these constraints, the NPD is generally limited to performance at dedicated sites in early-phase clinical trials and/or a select subset of research subjects enrolled in larger studies. Measurement of PD of the lower airways through bronchoscopic technique is feasible and could elucidate differences in electrolyte transport of the upper and lower airway (a case most likely to be important for measures of sodium transport) (2, 11, 19), but requires further validation in rigorously controlled trials for use in multicenter studies.
Nasal administration trials.
Nasal administration represents an attractive approach to evaluate agents intended to restore ion channel activity, with a number of practical and theoretical advantages. The nasal epithelium is a faithful representation of the histologic and ion transport features of the pulmonary epithelium, supporting its use as a biomarker for the lower airway. Nasal administration studies of this sort can be performed in a relatively rapid fashion, permitting prioritization of potential agents. The local route can also be used when concerns regarding systemic toxicity of an agent may be limiting. Because the nasal epithelium undergoes relatively rapid cell turnover, evaluation of the nasal epithelium after a washout period provides an additional control to validate results.
Researchers performing nasal administration of CFTR gene transfer agents were among the first to report rescue of Cl– transport on the basis of the NPD, but consistent efficacy across studies has been limited in part by the lack of standardization of NPD protocols (including use of chloride-free solution), variable efficiency of vectors, and heterogeneous transgene expression. Viral vectors, including adenovirus (20, 21) and adeno-associated adenovirus (22), have demonstrated transgene delivery by NPD, but positive results have not been universal (23). Improvements in Cl– transport, using lipid-based gene transfer vectors, were initially limited to small (about 2 mV) aggregate changes (24–26) or improvements in subgroups of subjects (27–29). More recent preparations have demonstrated somewhat improved transfer (30–32). Konstan and colleagues reported successful gene transfer to the nasal mucosa using compacted DNA nanoparticles, with relatively large polarizing PD improvement after isoproterenol perfusion in seven of eight subjects (33). Gene transfer studies have generally failed to demonstrate improvements in Na+ transport (even when increased Cl– conductance was seen). These results are consistent with in vitro studies indicating that the threshold to restore Cl– transport requires correction of a much lower percentage of cells (3–6%) compared with that needed to correct Na+ hyperabsorption (more than 90%) (34–36). Correction of NPD findings without detection of transgene expression seen in some studies may also indicate that a relatively small number of corrected cells can normalize NPD findings through a parallel electrical arrangement of nasal cells in relation to the measurement of electromotive force.
Topical dosing in the nose has been successfully used in studies testing the capacity of gentamicin to restore mutant CFTR function by suppressing premature stop mutations (37, 38). As opposed to gene transfer studies, improvements in both Na+ and Cl– transport were seen, potentially reflecting more homogeneous CFTR protein expression/correction across the epithelium. Chloride secretagogues, Na+ channel inhibitors, and other agents have also been administered in nasal dosing regimens, demonstrating biological effects in early-phase studies (2, 39–43). Experience suggests that inclusion of a placebo arm (rather than relying on crossover design alone) allows for more robust comparisons of NPD parameters, and may help avoid conflicting conclusions based on the variance of the assay (such as when a topical agent may disrupt mucosal integrity [23, 44, 45]).
Multicenter trials of CFTR modulators.
With the large number of new CFTR-active agents approaching early-phase evaluation, and the relative rarity of eligible subjects with CF (particularly when limited to specific CFTR mutations), a multicenter approach is required to expeditiously evaluate new CF treatment strategies that target mutant CFTR. Despite this need, multicenter trial experience with the NPD is fairly limited (45–48). Whereas numerous single-center trials (evaluating both nasal and systemic therapies) have shown improved ion transport on the basis of the NPD (6, 24, 26, 37–40, 43, 49–52), no multicenter trial evaluating a direct CFTR modulator has demonstrated improvement in NPD parameters in patients with CF. Despite negative results, experience has improved our understanding of the use of this assay in the context of clinical trials, as described subsequently.
Sensitivity, Specificity, and Analysis Methods
The bioelectric properties of the nasal epithelium have been examined in numerous studies to bridge encouraging in vitro observations to the human condition (2, 4), including studies of gene transfer (11, 23–29, 33, 34, 49, 53, 54), CFTR-activating agents (5, 6, 55, 56), restoration of mutant CFTR function and localization to the epithelial cell membrane (37, 38, 40, 51, 52), and regulators of non-CFTR ion transport (41, 43, 57–59). In many of these studies, the findings in the nasal epithelium have reflected observations made with in vitro or ex vivo model systems. The positive findings from various strategies suggest that the nasal respiratory epithelium has adequate sensitivity to detect and quantify the bioactivity of experimental agents, but the lack of a "gold standard" for CFTR modulation (or other components of defective ion transport) in vivo limits precise estimates of NPD sensitivity in this regard. Because of these limitations, researchers have resorted to examining relationships between ion transport and various levels of CFTR dysfunction provided by genotype–phenotype correlations (10), providing potential benchmarks for treatment effect. Whether a drug effect that achieves these targets will translate into clinically detectable improvements remains an area of intense interest. Because NPD results describe bioelectric findings of the nasal airway and only indirectly measure CFTR activity, evaluations of agents that have caused bioelectric changes through non–CFTR-dependent pathways must be interpreted carefully, and may not necessarily reflect CFTR function. Together, the accumulated evidence supports the measurement of nasal ion transport in proof-of-concept studies and in early phases of drug development.
A number of statistical methods have been used in previous studies, representing a potential challenge to NPD analysis. Multiple and repeated measures from each subject further increase complexity. For example, Na+ transport can be quantified by (1) the mean basal PD (including the average of multiple measurements at various distances within the inferior meatus), (2) the most polarizing PD obtained within the inferior meatus, (3) the stable PD obtained after perfusion of Ringer's solution during an NPD tracing, or (4) the change in PD after perfusion with amiloride (Figure 1). Similarly, CFTR-dependent chloride transport can be quantified as (1) the change in PD after zero Cl– perfusion, (2) the change in PD after perfusion with a CFTR agonist (in general, 0.01 mM isoproterenol), or (3) the sum of these changes (or "total CFTR-dependent chloride secretion," which is typically less variable than its individual components). Analysis of the mean of left and right nostril NPD measures from each subject (as opposed to considering each nostril individually) reduces variability, and improves assay sensitivity and specificity (46). Limiting analysis to the most polarizing nostril is an alternative approach, but can lead to erroneous conclusions (and lack of specificity) if a treatment increases the distribution of an NPD parameter without producing a change in the mean value. This is of particular relevance to trials that do not use a placebo arm, or in studies with nasally administered compounds that potentially disrupt mucosal integrity. Analysis of NPD results incorporating both Na+ and Cl– measures into a single value (total change in PD or 
NPD) was shown to reduce overlap between CF phenotypes and decreased CFTR activity in smokers compared with control subjects, and in subjects with CF with various amounts of predicted CFTR function, suggesting increased power to discriminate CFTR activity (10, 60). A measure such as NPD would have most use in studies in which changes in both Na+ and Cl– are anticipated. Ongoing analysis and comparison with traditional NPD measurements should clarify the use of this measurement in assessment of CFTR-restorative therapies. More detailed analyses of various statistical methods quantifying multicenter NPD results are currently in progress. These and other studies are intended to develop consensus in the field regarding best statistical methods for NPD analysis. Until then, the primary method for NPD analysis should be established for each study a priori to abrogate statistical concerns due to multiple comparisons.
|
F508 CFTR-corrective agent (8-cyclopentyl-1,3-dipropylxanthine [CPX]), investigators failed to demonstrate biological effects by NPD outcome measures (47). Post hoc analysis showed that NPD variability was significantly different between the four study sites (and much greater than conventional laboratory measures), suggesting this may have contributed to the negative results (61). In particular, within-subject variance for total chloride secretion was significantly larger than between-subject variance (indicating relatively low reproducibility), and was particularly large at a single site [87 vs. 21 (mV)2 at the other three centers]. These and other concerns led to the development of standardized operating procedures (SOP) among Cystic Fibrosis Therapeutics Development Network (CF-TDN) centers (62). In a more recently completed multicenter trial evaluating nasal gentamicin and tobramycin to suppress premature stop mutations (five of six sites used the newly developed SOP), improved variance of stimulated chloride conductance was seen compared with that reported by McCarty and colleagues (47, 61) (Table 1). Restricting analysis to the five CF-TDN sites that used all aspects of the NPD SOP demonstrated further reductions in variability and improved reproducibility (46). Commensurate with these improvements, a reduced number of subjects is anticipated to be necessary to detect significant changes in Cl– conductance (Figure 2). Improved power is dependent on a consistent effect magnitude, thus reducing the coefficient of variation.
|
|
|
|
To facilitate consistency and to maintain quality assurance among centers, a central nasal PD coordinating center (termed the Center for CFTR Detection) has been established at the University of Alabama at Birmingham to continue earlier efforts initiated by the previous NPD Interpretive and Coordinating Center at the University of North Carolina (Chapel Hill). Several NPD workshops sponsored by the Cystic Fibrosis Foundation have been completed at the University of Alabama at Birmingham, providing hands-on training to NPD operators. To ensure that appropriate procedures are maintained at local sites, individual NPD centers undergo annual site visits by TDN coordinating center personnel. In addition, individual NPD operators are required to meet agreed-on qualification standards to perform NPDs in CF-TDN studies. All sites use common essential equipment including voltmeters, data capture recorders, and measuring devices. Centralized processing and evaluation of tracings by the Center for CFTR Detection allows ongoing quality assurance and scoring by a single blinded investigator in randomized trials. The Cystic Fibrosis Foundation has supported these procedures and additional efforts to pursue incremental improvements in the performance of the assay. For example, studies at our center comparing NPD measurements made using two commonly employed devices (calomel/agar bridge vs. silver chloride pellet/saline bridge) on the same subjects have not demonstrated significant differences in basal PD reproducibility in normal individuals; whether these findings apply to measures of isolated Cl– transport or extend to individuals with CF is under further evaluation, and may lead to recommendations regarding equipment use if clear differences are established. Further studies of this sort are included among the recommendations to further advance the use of NPD as a biomarker of CFTR activity (Table 3A).
SWEAT CHLORIDE
Clinical and Biological Relevance
First described in the early 1950s, the elevation of sweat electrolyte levels in CF has been extraordinarily important both clinically and biologically (69). Determination of sweat electrolytes is still the key diagnostic test for CF, having proved remarkably robust. In addition, elevation of sweat electrolytes explains the hypoelectrolytemia that can be a significant complication of CF (70). Hypoelectrolytemia often occurs in infants or toddlers at presentation and may be accompanied by severe morbidity and even mortality (71). In addition to diagnostic and clinical importance, electrolyte determinations in isolated human sweat ducts provided the first evidence that chloride transport was the primary abnormality in CF (72). Sweat electrolyte levels have also been useful in the characterization of patients with CF in genotype– phenotype studies (73).
Although mechanisms that control sweat electrolyte levels are complex and incompletely understood (74, 75), several features of sweat gland physiology suggest that determination of sweat electrolytes may be a suitable biomarker in early-phase trials of agents with CFTR activity. For example, CFTR is highly expressed in the sweat duct (76). Furthermore, in vitro studies demonstrate that CFTR inhibition affects sweat duct electrolytes (77). Compounds that stimulate CFTR also affect sweat duct electrolytes in vitro (78). The most compelling evidence that sweat electrolyte determination holds potential as an outcome measure, however, has its basis in cross-sectional studies of patients with various CFTR mutations (10, 79–82). As shown in Figure 3, genotype–phenotype correlations for disease severity are also associated with alterations in sweat chloride.
|
Other features of sweat gland physiology make the determination of sweat electrolytes attractive as an outcome measure for CFTR activity. The elevation in sweat chloride in CF is reliably present just days after birth and is essentially fixed over decades (83, 84), allowing the test to be applied to subjects of any age. Moreover, as opposed to almost every other organ involved in CF (e.g., lung, pancreas, intestine, and sinus), the sweat gland is not damaged by disease pathogenesis, whereas other organs show evidence of secondary damage with time that may impair measurement of CFTR activity (85). In the lung, for example, structural airway damage may not be reversible even if full CFTR activity is restored. Preservation of sweat gland function in CF allows testing of CFTR activity unimpeded by inflammation, fibrosis, or other organ damage. Whether the protective effects of preserved sweat duct anatomy outweigh the potential concern of being removed from the primary pathophysiologic process responsible for morbidity warrants further evaluation. This concern was highlighted by McCarty and coworkers (47), who reported no change in sweat electrolytes associated with F508 CFTR modulator CPX, although results may have been due to lack of effect rather than limitations of the sweat biomarker.
Several features of sweat physiology constitute additional caveats in evaluating its usefulness as an outcome measure in CF trials. The sweat duct is lined with purely absorptive epithelia (86) whereas the airway, pancreas, and gastrointestinal tract are lined with secretory epithelia (87), although fluid absorption is normally dominant in the larger airways. It is not clear that an agent that is efficacious in absorptive epithelia will demonstrate activity in secretory epithelia. Coupling of CFTR activity with the epithelial sodium channel (ENaC) also differs between tissues (88, 89). In the sweat duct, failure of CFTR activity results in diminished ENaC absorptive activity whereas in the airway the opposite occurs. Given the importance of ENaC activity in CF pathophysiology (90), this difference raises questions as to how well changes in sweat electrolytes reflect alterations in the airway microenvironment.
Sweat electrolytes also appear to be more sensitive to salt intake and aldosterone metabolism than the NPD (91). In normal individuals, sweat sodium can change by 30 mEq/L or more by varying salt intake or by administering aldosterone or aldosterone blockers (92, 93). Sweat chloride can be similarly affected. In patients with CF, aldosterone and salt leak can influence sodium excretion in sweat, and chloride to a lesser degree (93–96). Consideration of sodium homeostasis therefore becomes important in the use of sweat chloride as an outcome measure; further studies are required to determine how best to control for these variables in trials incorporating sweat electrolytes.
Both sodium and chloride can be easily measured in a single sweat collection. Most studies indicate that sweat chloride shows less variability than sweat sodium in CF (83). In addition, it is believed that the primary defect in CFTR in the sweat gland is failure of chloride transport, and thus chloride levels are more closely related to the primary defect. Whether sweat sodium or sodium:chloride ratios can provide useful information in trials of agents with CFTR activity, particularly with respect to coupling with ENaC or salt status, deserves further study. Measurement of sweat potassium may also be of some use (74), but clearly needs further investigation before widespread use in clinical trials. CFTR is also expressed in sweat coil epithelia, leading to a small but detectable difference in sweat rate between subjects with CF and control individuals in response to ß-adrenergic stimulation after cholinergic blockade (97). This test holds some potential as an outcome measure (particularly when combined with sweat electrolytes) but involves measuring small volumes that are difficult to collect in a consistent manner.
In summary, elevation of sweat chloride levels in CF has both clinical and biological relevance. Abnormalities in sweat chloride occur early in the disease and remain consistent because of the lack of secondary damage to the sweat gland. The clear correlation between CFTR activity and sweat electrolyte concentration among subjects with various CF mutations suggests sweat chloride will be affected by therapies that alter CFTR activity. If an agent restores CFTR function of a subject with minimal or absent CFTR activity (as seen in those with severe mutations) to 50% of normal, we would anticipate that sweat chloride levels will decrease to near normal levels, as seen in asymptomatic carriers of CF. If a more modest effect is seen (e.g., about 10% of CFTR activity is restored), then a decrease of 10–15 mmol/L would be expected, resulting in values similar to those with one or more mild CFTR mutations. On the basis of in vitro studies of CFTR antagonists and agonists, the change in sweat chloride with an efficacious compound could be seen after relatively short duration (days to weeks).
Feasibility
Fortunately, sweat electrolyte determinations do not place a great deal of burden on patients, are only infrequently associated with complications, and have been performed in clinical laboratories for 40 years. The procedure essentially involves stimulation of sweat glands with pilocarpine via iontophoresis to increase the rate of sweat production, a procedure required to discriminate subjects with CF from control individuals (98). Sweat is collected for 30 minutes on each arm, and captured by gauze, filter paper, or a coil and prepared for analysis. The entire procedure requires 45 minutes and the patient may ambulate during the study. Approximately 1% of the time there can be an area of erythema at the site of iontophoresis consistent with a first-degree burn, but more serious complications are rare. The sweat test is performed thousands of times per year in patients of all ages, including infants as young as 1 week, for diagnostic purposes. Prior clinical experience and widespread use limit ethical concerns.
To maximize its use in multicenter trials, sweat electrolyte testing requires a standardized approach to collection and measurement. The CF-TDN in conjunction with Vertex Pharmaceuticals, Inc. (Cambridge, MA), has addressed this question in anticipation of clinical testing of an agent with CFTR activity on surface-localized mutant alleles (i.e., class III, IV, and V mutations). Despite guidelines for sweat testing from the Clinical and Laboratory Standards Institute (99), a survey of TDN centers showed a number of differences in sweat collection and analysis among centers. To better standardize methodology among centers, all sites participating in this study have transitioned to sweat collection using the Macroduct sweat collection system (Wescor, Logan, UT) (Figure 4). This method has been used by a number of large centers and appears to minimize operator differences compared with the standard Gibson-Cooke gauze or filter paper collection system (100, 101). In addition, a standard operating procedure was developed for sweat collection, shipping and analysis at a central laboratory, as well as a qualification process for research personnel in performing the procedure (available on request from the CF-TDN Coordinating Center). Sweat contains few cells and little protein or other macromolecules and is apparently not affected by freeze–thaw effects.
|
Sensitivity and Specificity to Treatment Effects
We do not know the sensitivity and specificity of sweat electrolyte changes in response to treatment because there is no therapy shown to modify chloride absorption in the sweat gland through modulation of CFTR activity. A study of oral lithium supplementation in CF showed a trend toward decreased sweat chloride (94 to 87 mmol/L), but it was not statistically significant (p < 0.07) (102). The increase in sweat chloride after discontinuation of lithium was statistically significant, however. This is the only study to show effects on sweat chloride (albeit of uncertain significance); furthermore, it is not known how lithium affects the pathophysiologic cascade in CF, limiting conclusions regarding sensitivity and specificity. A study of fatty acid supplementation in CF demonstrated a decrease in sweat sodium but no change in sweat chloride (103). This study raises the question of specificity of a treatment effect: agents that influence ENaC activity are likely to alter sweat sodium and possibly even sweat chloride. Overall, the best information we have regarding sensitivity of sweat electrolytes is based on consideration of sweat electrolytes by genotype in affected individuals.
One possible way to approach sensitivity and specificity in vivo would be to study sweat electrolytes or conductivity after local application of agents to the skin followed by iontophoresis to ensure contact with the sweat duct (analogous to topical nasal treatments for NPD). Limitations in iontophoresis would restrict its use to polar molecules. A study of this sort may be particularly useful when safety, regulatory, or pharmacokinetic considerations limit systemic administration of a potential therapeutic compound.
Overall, a great deal needs to be learned about sensitivity and specificity of sweat electrolytes as an outcome measure in CF and it is only through a series of multicenter trials that this can be approached.
Reproducibility
A better understanding of the reproducibility of sweat electrolyte measurements is imperative for future use of this outcome measure. There are reports that present repeat measurements over a few weeks (51, 61, 104), supplying an estimate of the inherent biological variability in sweat chloride, but these studies are not comprehensive. In addition, factors that likely contribute to inherent biological variation in sweat chloride measurement are still incompletely understood. There is only one multicenter trial that has examined intersite variability in sweat chloride measurements (61). Similar to problems encountered during NPD measures in this trial, each of the four sites in this study used a different approach to sweat collection and analysis. There was considerable variability among sites in the variance seen within and between patients. Some sites reported markedly less variability than others. This study underscores the need to harmonize sweat collection and analysis for multicenter trials. A series of studies is needed to provide more precise information on reproducibility of sweat electrolytes.
On the basis of the study by Ahrens and coworkers, the coefficient of variation of sweat chloride in patients with CF is between 10 and 20%, including both within-patient and between-patient variance (61). This suggests that agents that are expected to produce a change of 15–20 mEq/L in sweat chloride from a baseline of 100 mEq/L (typical for CF) could be effectively studied in trials enrolling dozens of patients rather than hundreds. The sample size estimates are likely to decrease with standardization of collection and central analysis.
Summary of the Potential of Sweat Chloride as an Outcome Measure
Sweat electrolytes hold great potential in the detection of CFTR activity for systemically administered agents. The clinical and biological relevance of sweat chloride is well established, and testing is feasible in the context of multicenter trials. Standardization of the procedure is relatively straightforward, and is currently in progress for use in ongoing trials. An improved understanding of the sensitivity, specificity, and reproducibility is mandatory. Nevertheless, genotype–phenotype correlations support the potential for sweat electrolytes in this setting. Concerns regarding whether the absorptive epithelium of the sweat gland truly reflects the secretory epithelia of the airway and other organs and whether salt and aldosterone status will confound findings deserve further attention. A list of needed studies to further develop sweat chloride as an outcome measure in CF is given in Table 3B.
OTHER MEASURES OF CFTR IN STUDY SUBJECTS
In addition to measures of function, localization of protein and measures of mRNA expression have been proposed as approaches to CFTR detection in early-phase trials. Quantification of CFTR cDNA mRNA transcript levels has been used to demonstrate transgene delivery and expression (using vector-specific primers [105, 106]), and changes in total CFTR transcript levels may potentially be detected in treatments predicted to improve mRNA stability or CFTR protein expression (107). Technological advancements may allow relatively straightforward quantification of RNA specific transcript levels, using automated real-time reverse transcriptase-polymerase chain reaction techniques (46, 108, 109). High baseline CFTR mRNA levels might be used to predict subjects most likely to respond to novel treatments that are dependent on adequate CFTR mRNA substrate. For example, high CFTR mRNA levels have been associated with treatment responses to gentamicin in Israeli subjects harboring stop mutations (110). When measured in gene therapy trials, failure to detect CFTR mRNA transgene expression has not always been associated with the absence of improved Cl– conductance (24); this disparity indicates detection of mature mRNA from nasal samples may have limited sensitivity compared with measures of CFTR activity, even when using the best available methods (e.g., because of low-level contamination by human RNase). Heterogeneous, high expression of transgene in individual cells also has the potential to overestimate overall expression because of sampling error.
Immunocytochemical (e.g., immunofluorescence [IF]-based) detection of CFTR location in brushed human nasal airway cells can be used to characterize effects of drugs modulating CFTR translation or membrane trafficking. Immunohistochemical detection does not allow evaluation of CFTR activity (the same limitation applies to mRNA transcript levels); therefore, cellular localization has minimal use in therapies addressing class III and IV alleles (surface-localized mutants) where CFTR processing is essentially normal. Relatively simple methods permit procurement of nasal cells by curettage at local sites, with processing and blinded scoring at specialized centers. Different techniques for processing these samples have been summarized by Harris and coworkers (111). The procedure has been used in clinical trials examining suppressors of premature stop mutations with variable results (37, 46). Most difficulties noted in airway cell IF relate to nonspecificity of antibody staining and effects of sample processing on cellular staining characteristics. Approaches to improve identification at the cell surface (where CFTR would be anticipated to be active) versus subsurface (where it would not) could be bolstered by refinement of more sensitive and specific anti-CFTR antibodies and colocalization with other cell surface markers; however, processing and scoring increase in complexity with these techniques. Future use of IF in clinical trials also requires prospective validation of staining patterns seen in normal subjects and those expressing various classes of CFTR mutations, as only limited information is available regarding staining patterns in normal and CF tissues (112, 113). Moreover, the role of IF will likely be limited to an adjunctive measure within clinical trials, as it remains primarily a qualitative or semiquantitative measure. Application of biochemical methods commonly used in the laboratory to examine cells grown in culture (e.g., Western blot, immunoprecipitation, and cell surface biotinylation), adapted to human airway, skin, or gastrointestinal samples, would also help confirm the efficiency of agents addressing aberrant cellular processing but are still in the exploratory phase and may be limited by sensitivity, available antibodies, and inadequate recovery of excised tissue.
Quantifying CFTR activity by fluorescence-based halide efflux on brushed nasal cells ex vivo has been used with limited success (28, 114, 115). Technical limitations include the need to perform the assay in specialized centers (with expensive imaging equipment), the short time frame required between nasal cell procurement and assay, challenges of studying nonadherent living cells (with beating cilia), and the reportedly poor correlation between percentage of cells with halide efflux and respiratory symptoms in individuals heterozygous for CFTR mutations (114). These issues currently limit the widespread adoption of this technique within multicenter clinical trials.
As clinical therapies intended to restore function to mutant protein are in the early stages of development, and our understanding of CFTR and other biomarkers of ion transport has important limitations, more than one measure of efficacy is required in early-phase trials, and should be based on the pharmacologic activity of the agent under evaluation. For example, systemic agents directed at correcting F508 CFTR misprocessing might include measures of CFTR activity (including NPD and sweat chloride), as well as one or more nonfunctional indicators of protein rescue (e.g., IF or other assays of epithelial cells ex vivo); reverse transcriptase-polymerase chain reaction has minimal use in this setting. Including multiple methodologies will help ascertain the spectrum of biological activity, address issues of sensitivity and specificity, allow correlation of these CFTR-specific biomarkers internally and with established clinical biomarkers (e.g., lung function, growth), and facilitate the identification of barriers to protein rescue, as failure to demonstrate functional rescue may not be sufficient to rule out a therapy with modest bioactivity. In total, these adjunctive methods offer the potential to confirm mechanism of action, and require further validation and refinement to clarify their full use (Table 3B).
CONCLUSIONS
In vivo measures of CFTR activity are in use to demonstrate biological activity of new CF treatment strategies and currently serve as biomarkers in CF clinical studies. Although NPD and sweat electrolyte assays have been shown to correlate with predicted CFTR activity in subjects with certain CF genotypes and phenotypes, variability of assay performance has thus far been an important limitation to demonstrating treatment effects in multicenter trials. Inclusion of NPD measurements in phase 3 studies of CFTR or other ion transport active agents may allow this biomarker to serve as a surrogate end point, provided that clear correlation with established surrogates (e.g., lung function) or clinical efficacy measures (e.g., exacerbation frequency) is established. Current research aimed at reducing nonbiological variability is underway, and should lead to technical modifications that enhance the capacity of these assays to demonstrate meaningful biological effects in early-phase studies. These and other measures of CFTR activity/biogenesis will assist in the evaluation and prioritization of new therapeutic agents in emerging CF clinical trials.
ACKNOWLEDGMENTS
The authors thank Drs. M. Knowles, R. Oster, M. Ashlock, N. Hamblett, M. Boyle, E. Sorscher, and Z. Bebok for contributions and helpful suggestions. The authors are also grateful for contributions from members of the CF-TDN Outcomes Working Group, including L. Boitano, K. F. Bruce, M. Cohen, P. Cornelisse, Z. Davies, D. Escobar, A. Gentossio, P. Hiatt, J. Hocevar-Trnka, D. Hornick, G. Hovick, S. Hurban, J. Johnson, L. Larson, K. McCoy, K. Packer, M. Pian, B. Plant, L. Rose, R. Rowbotham, B. Trapnell, D. Wetmore, T. White, and H. Young; and the NPD operators and research coordinators within the CF-TDN for making these studies possible.
FOOTNOTES
Supported by the National Institutes of Health (1 K23 DK075788-01 and 1 U01 HL081335), the Cystic Fibrosis Foundation (CLANCY06A0, CLANCY96LO, CLANCY05Y2, R464-CR02, and ACCURS05A0), and Cystic Fibrosis Foundation Therapeutics, Inc. (CLANCY02Y0 and ACCURS98Y).
The University of Alabama and the University of Colorado are members of the Cystic Fibrosis Therapeutics Development Network.
Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
(Received in original form March 15, 2007; accepted in final form May 14, 2007)
REFERENCES
F508 cystic fibrosis and residual chloride secretion. Lancet 1999;353:984–985.[CrossRef][Medline]F508 cystic fibrosis transmembrane conductance regulator [abstract]. Pediatr Pulmonol Suppl 2006;41:247.F508-homozygous cystic fibrosis patients: partial restoration of nasal epithelial CFTR function. Am J Respir Crit Care Med 1998;157:484–490.[Medline]F508). N Engl J Med 1990;323:1517–1522.[Abstract]F508 CFTR in cystic fibrosis sweat gland. Nat Genet 1992;1:321–327.[CrossRef][Medline]
was 0.05 and power was 0.80.
