Background Friedreich ataxia (FRDA) is a neurodegenerative disease caused by mutations in the frataxin (FXN) gene, resulting in reduced expression of the mitochondrial protein frataxin. Improved understanding of the pathophysiology of the disease has led to a growing need for informative biomarkers to assess disease progression and response to therapeutic intervention.
Objective To evaluate the performance of frataxin measurements as a diagnostic tool using two different immunoassays.
Methods Clinical and demographic information was provided through an ongoing longitudinal natural history study on FRDA. Frataxin protein levels from multiple cell types in controls, carriers and FRDA patients were measured and compared using a lateral flow immunoassay and a Luminex xMAP-based immunoassay. Receiver operating characteristic curve analyses were then performed to evaluate the sensitivity, specificity, and positive and negative predictive values for each immunoassay.
Results For whole blood and buccal cells, analysing FRDA patients and carriers together in a cohort resulted in higher sensitivities and positive predictive values compared with analyzing controls and carriers together, with similar results between each tissue type. We then compared the usefulness of a lateral flow immunoassay with a multianalyte Luminex xMAP-based immunoassay, and showed that both assays demonstrate high positive predictive values with low rates of false negatives and false positives.
Conclusions Frataxin measurements from peripheral tissues can be used to identify FRDA patients and carriers. While multiple cell types and assays may be useful for diagnostic purposes, each assay and cell type used has its advantages and disadvantages depending on study design and scope.
- MITOCHONDRIAL DISORDERS
- MOVEMENT DISORDERS
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Friedreich ataxia (FRDA) is an autosomal recessive neurodegenerative disease caused by mutations in the FXN gene encoding the mitochondrial protein frataxin.1 Most mutations result in decreased frataxin levels, which lead to the clinical manifestations of disease. The majority (96–98%) of patients are homozygous for GAA trinucleotide repeat expansions of ∼40–1700 repeats within the first intron of the FXN gene, with the repeat length on the shorter allele correlating with age of onset and disease severity.1–4 FRDA most commonly presents before the age of 18 years, and is characterised by progressive limb and gait ataxia, proprioceptive loss, absent tendon reflexes, dysarthria, and loss of ambulation 10–15 years after onset of disease.5–7 Additionally, many patients develop progressive scoliosis, diabetes mellitus, pes cavus, pyramidal weakness and optic atrophy.5 ,6 Approximately 50–75% of FRDA patients develop cardiomyopathy or cardiac complications that contribute to premature death.6 ,8 There is no approved treatment for FRDA, but several potential therapies aimed at mitochondrial dysfunction or increasing expression of frataxin are in development.9–11 As these compounds progress to clinical studies, there is a need to quantify the impact of treatments. At the molecular level, studies have focused on markers of potential consequences of reduced frataxin as a measure of FRDA-related pathology or pharmacological response. These include activities of ISC-containing enzymes, urinary and intracellular markers of oxidative stress, mitochondrial iron levels, mitochondrial DNA levels, and 31P-MRS as a measure of energy production.12–17 Since FRDA is caused by frataxin deficiency, measurement of frataxin itself should be useful to screen, diagnose and assess response of the disease process.
Traditionally, FRDA is diagnosed by measuring GAA-repeat length or detecting point mutations; however, this approach cannot be used to monitor treatment or efficiently used for high-throughput population screening. Assessing frataxin protein levels may overcome these shortfalls. Frataxin protein has typically been measured using western blot or ELISA, neither of which is feasible for rapid measurements from large numbers of samples in clinical research settings or for analysing specific tissues with low frataxin concentrations (eg, dried blood spots (DBS) or buccal cells). A few studies with small sample sizes have focused on quantification of frataxin from patient samples; however, these procedures typically are untested in many peripheral cell types used in clinical studies, and are limited by narrow reference ranges.18–20
A commercial lateral flow immunoassay was recently developed for the rapid quantification of mitochondrial proteins from a variety of cell types without the need to isolate mitochondria.21 Using this ‘dipstick’ assay, a significant loss of frataxin protein was identified from multiple cell types in FRDA patients, including whole blood and buccal cells, demonstrating potential usefulness in large clinical settings.22 ,23 A second immunoassay with multiplex capabilities using xMAP technology was also recently developed and validated for the measurement of frataxin levels from DBS. This assay also may be useful for population screening and monitoring purposes.24 While these immunoassays can measure significant differences between controls, carriers and FRDA patients, and they can be used for diagnostic purposes, their performance characteristics have not yet been directly compared.
Here, we use the lateral flow immunoassay for the measurement of differences in frataxin levels in multiple peripheral cell types collected by non-invasive and minimally invasive means from controls, carriers and FRDA patients to determine how well such measurements perform. We also measured frataxin protein levels in patients with point mutations and late-onset FRDA (LOFA) to ascertain the effect of atypical genotypes on design of an immunoassay. We then evaluated the frataxin protein measurements from DBS samples by a second xMAP-based Luminex immunoassay for its performance as compared to the lateral flow immunoassay.
Clinical and demographic information, including molecular confirmation of FXN gene mutations, was provided through an ongoing longitudinal natural history study on FRDA.25 Samples were collected from these patients at the University of Pennsylvania/Children's Hospital of Philadelphia, the University of California at Los Angeles Medical Center, Emory University, University of Florida, and other patients referred to physicians at these and other institutions. Patients with other disorders also participated as a ‘diseased’ control group to account for referral bias. These subjects had other ataxia or movement disorders in which FRDA was excluded. Carriers and normal healthy volunteers also provided samples for analysis (table 1).
Buccal cell and whole blood sample preparation
Whole blood and buccal cells were collected from controls, carriers and patients, and protein was extracted for frataxin analysis as described.22 Whole blood was collected in K2 EDTA BD Vacutainer tubes (REF 367844) and frozen immediately at –80°C until used. Buccal cells were harvested using MasterAmp Buccal Swab Brushes (Epicentre) and protein extracts were stored at –80°C until use. Protein concentration of buccal cell extracts was determined using the bicinchoninic acid (BCA) protein assay (Pierce) before frataxin protein quantity analysis.
Lateral flow immunoassay protocol
Frataxin protein quantity dipstick assay kits (Abcam/MitoSciences, MSF31) were used to measure frataxin levels in whole blood and buccal cells as described.22
Luminex xMAP-based immunoassay protocol
For each subject, 50 µL of whole blood was spotted of Protein Saver 903 Cards (Whatman), and allowed to dry. Protein was eluted from DBS, mixed with frataxin antibody-coupled microspheres, and analysed for frataxin content using the Luminex LX200 instrument.24 Calibration of frataxin content was performed using purified human frataxin (courtesy of Dr Grazia Isaya, Mayo Clinic).
Data analysis was performed using STATA SE V.11 and MS Office Excel 2007. All frataxin levels as measured by dipstick assays were expressed as a percentage of average controls, and frataxin levels measured by Luminex assay were normalised to purified recombinant frataxin and expressed as ng/mL whole blood. One-way analysis of variance (ANOVA) followed by Scheffe's posthoc analysis was used to compare mean frataxin levels in whole blood and buccal cells. Cutoff values for the lateral flow immunoassay frataxin tests were selected based on mean frataxin values and proximity to the upper left corner of the receiver operating characteristic (ROC) curves. Simple and multiple regression models analysed relationships between frataxin levels and multiple independent variables.
Clinical and demographic information for lateral flow immunoassay samples
To expand previous findings and investigate usefulness of frataxin protein measurements as a diagnostic tool, frataxin protein levels were measured from whole blood and buccal cells in a larger cohort of subjects than in previous studies (table 1). Approximately equal numbers of male and female volunteers were enrolled for each group, and the mean age and age of onset (where applicable) were similar for both cell types analysed. Across both cell types, GAA1 ranged from 41 to 1100, GAA2 from 160 to 1585, and age of onset from 2 to 63 years, showing the wide spectrum of disease severity in this study.
Evaluation of lateral flow immunoassay on whole blood samples
Frataxin levels were initially measured in whole blood drawn from controls (n=67), carriers (n=143), and FRDA patients (n=292), who were divided into LOFA (n=35), point mutation FRDA patients (pFRDA, n=12), and classic FRDA patients (cFRDA, n=246). In accordance with previous findings, mean frataxin levels were significantly reduced in carriers (67.8% of control; 95% CI 64.9 to 70.7%) and cFRDA patients (26.1% of control; 95% CI 24.1 to 28.2%), with significant decreases also seen in LOFA and pFRDA patients (figure 1A).22 The higher frataxin levels in LOFA patients (mean, 53.9% of control; 95% CI 45.6 to 62.1%) relative to other FRDA patients suggest that these patients may be difficult to identify through detection of low frataxin levels.
ROC analyses were then performed using measured frataxin levels in whole blood to assess the ability of the assay to identify samples as having ‘disease’ or ‘no disease.’ The ROC curves were constructed in two different ways to assess either the ability of the test to distinguish FRDA patients from controls and carriers, or to identify carriers and FRDA patients together separated from controls. In both analyses, the dipstick assay had an area under the curve greater than 0.950 for each potential method (figure 1B). Based on these ROC curves and mean frataxin levels measured in the samples, classification criteria for the dipstick assay were established to label an unknown sample as ‘control,’ ‘carrier/potential FRDA,’ or ‘FRDA’; frataxin levels of greater than 85% of control were used as the ‘control’ classifier, 45–85% of control as the ‘carrier/potential FRDA’ classifier, and less than 45% of control as the ‘FRDA’ classifier. These criteria were used to determine which method was most accurate, and demonstrated highest clinical usefulness (figure 1B). Designing the methodology to identify carriers and FRDA patients yielded the highest sensitivity (0.938) and positive predictive value (PPV, 0.976) (table 2). This indicates that identifying carriers and FRDA patients will result in fewer false negatives due to potential inclusion of LOFA patients with frataxin levels similar to carriers. The negative predictive value (NPV) was low (0.679) due to the relatively small number of control samples assayed compared with carriers and FRDA patients. Taken together, these results suggest that using the dipstick assay to identify patients and carriers based on frataxin levels in whole blood is a valid tool for complimentary diagnosis in large numbers of individuals.
Evaluation of lateral flow immunoassay in buccal cells
While collection of whole blood is convenient, protein extraction from buccal cells can be used in essentially any setting. Here, the potential usefulness of measuring frataxin protein from buccal cell extracts using the lateral flow immunoassay was examined. Similar to whole blood, buccal cell extracts were collected from a large number of controls (n=93), carriers (n=271) and FRDA patients, which were separated into LOFA (n=28), pFRDA (n=16) and cFRDA (n=288) cohorts (figure 2A). In agreement with previous studies, significant decreases in mean frataxin levels were measured in buccal cells from carriers (56.9% of control; 95% CI 53.9 to 59.9%) and cFRDA patients (19.7% of control; 95% CI 18.1 to 21.2%).22 LOFA patients had mean frataxin levels in buccal cells (39.8% of control; 95% CI 30.4 to 49.2%) that were significantly higher than cFRDA or pFRDA patient levels (p<0.001), similar to whole blood.
ROC curve analyses to plot sensitivities and specificities of measured frataxin levels in buccal cells showed similar results to those in whole blood. The area under the curve was greater than 0.925 for each potential method (figure 2B). Using the ROC curves and frataxin levels measured in buccal cells, another set of classification criteria was created for the dipstick assay to label an unknown buccal cell sample as ’control,’ ‘carrier/potential FRDA,’ or ‘FRDA’; here, frataxin levels of greater than 80% of control was the ‘control’ classifier, 40–80% of control was the ‘carrier/potential FRDA’ classifier, and less than 40% of control was the ‘FRDA’ classifier. These criteria differ slightly from those used for whole blood; this may be due to potential saturation of frataxin signal by control blood samples leading to artificially elevated frataxin levels in carriers and patients when expressed as a percentage of those controls. These classifiers were used to construct 2×2 tables to calculate which method was most accurate and clinically useful based on the above criteria (figure 2B). As with whole blood, the PPV (0.969) and sensitivity (0.922) were highest when grouping buccal cells from FRDA patients and carriers into the same cohort, with a relatively low NPV (0.615) based on a small control population (table 2). These values mirror those from whole blood; thus, measuring frataxin protein from either buccal cells or whole blood to identify carriers and FRDA patients may be a viable diagnostic tool.
Comparison of frataxin measurements between two different immunoassays
Next, frataxin was measured from DBS eluents using a clinically validated, multianalyte, Luminex xMAP-based immunoassay to determine its comparative effectiveness for detecting controls (n=23), carriers (n=76), and FRDA patients (n=141), including LOFA (n=20) and pFRDA patients (n=4) (table 3, figure 3A). Age of onset correlated strongly with GAA1 (r=0.748, p<0.001) and frataxin levels (r=0.643, p<0.001) in DBS.
Of those individuals assessed by Luminex assay, frataxin levels were measured in parallel via dipstick assay in controls (n=17), carriers (n=54), and FRDA patients (n=90), including LOFA (n=15) and pFRDA patients (n=4) (figure 3B). Frataxin levels measured by Luminex assay correlated strongly with frataxin levels as measured by dipstick assay (r=0.854, p<0.001), demonstrating reproducibility of frataxin measurements with parallel dipstick and Luminex assays. In agreement with previous studies, mean frataxin levels were reduced in carriers (70.6% of control; 95% CI 66.3 to 74.9%) and cFRDA patients (29.7% of control; 95% CI 26.1 to 33.3%) using the dipstick assay (figure 3A). Mean frataxin levels were also reduced in carriers (15.2 ng/mL; 95% CI 14.0 to 16.4 ng/mL) and cFRDA patients (5.6 ng/mL; 95% CI 4.9 to 6.2 ng/mL) relative to controls (28.5 ng/mL; 95% CI 26.4 to 30.6 ng/mL) with the Luminex assay (figure 3B); differences in frataxin levels between controls, carriers and FRDA patients in both assays were statistically significant (p<0.001).
To evaluate how well the Luminex assay performs as a comparable method, two separate ROC analyses on the measured frataxin values were conducted to determine which individuals to identify from the test, and the results were compared with values from dipstick assay measurements (figure 3C,D). In each circumstance for both assays, the area under the ROC curve was greater than 0.915, indicating sufficient sensitivity and specificity for both assays. The area under the curve was higher for the dipstick assay (figure 3C) than the Luminex assay (figure 3D), although fewer samples were analysed by dipstick analysis. Based on the shape of the ROC curves and mean values from the frataxin measurements, specific frataxin levels were chosen as cutoff points for the dipstick assay to classify each sample as a ‘control,’ ‘carrier/potential FRDA patient,’ or ‘FRDA patient.’ For the Luminex assay, the absolute frataxin concentration and a ratio between the method's multiple analytes (frataxin and ceruloplasmin) were used to generate a postanalytical algorithm that calculates the likelihood of FRDA. Through this analysis, frataxin levels above 23 ng/mL were identified as ‘control,’ 13–22 ng/mL were ‘carrier/potential FRDA,’ and samples with less than 12 ng/mL frataxin were labelled as ‘FRDA.’ Using these criteria, 2×2 tables were created to examine the ROCs and predictive value of both assays for each group of subjects that were identified from the test (figure 3C,D). Analysing FRDA patients and carriers together in a single cohort demonstrated the highest or close to the highest sensitivities and PPVs in the dipstick and Luminex assays (table 4). For identification of carriers/FRDA patients, the Luminex assay had higher sensitivity (0.963 vs 0.910), specificity (0.957 vs 0.765), PPV (0.995 vs 0.970), and NPV (0.733 vs 0.500) values compared with the dipstick assay. NPV values were relatively low for both assays, which likely reflect the low numbers of control samples analysed. These results suggest that frataxin measurements from eluted DBS by the Luminex assay to identify carriers and FRDA patients are more accurate than the dipstick assay, with lower false positive and false negative rates. Still, both assays exceeded expectations, and the method of testing used will depend on the study context, the clinical setting, and the composition of the study cohort.
Here, we demonstrate that frataxin measurements from peripheral tissues using the lateral flow immunoassay and a Luminex xMAP-based immunoassay are useful as diagnostic tools for FRDA. Both have high PPV and a low rate of false negatives. These assays can rapidly assess peripheral tissue samples and potentially identify individuals as carriers or patients during the presymptomatic stage of the disorder. Additionally, they are immediately applicable to natural history studies that systematically follow frataxin levels over time or in response to pharmacological intervention. While neurological exams and performance-based measures are typically used to assess disease progression, they reflect the clinical rather than the biochemical status of disease.25–27 By contrast, frataxin levels represent a standardised biochemical marker with presumed stability over time in untreated individuals, and is a target of disease-modifying pharmacological treatment.28
In the present study based on ROC curve analyses, we generated criteria for classifying measured frataxin from whole blood and buccal cells as representative of control, carrier, or FRDA patient levels.22 There were different contexts to consider in using the dipstick assay. For example, the goal might be to detect FRDA patients among a large cohort, or to identify all carrier and FRDA samples. Here, the dipstick test was more useful for identifying FRDA subjects and carriers together rather than FRDA subjects alone. Sensitivity and PPV values were highest in this scenario for whole blood and buccal cells and were comparable between the two tissues. This approach avoids exclusion of FRDA patients with higher frataxin levels, particularly LOFA patients and some pFRDA patients, but maintains a low false negative rate. In a diagnostic setting, a positive result for this test would indicate an individual is either a carrier or an FRDA patient, and could then be assessed by DNA analysis for GAA-repeat expansions. All false positives in this study were explained by a few control subjects with frataxin levels in the upper carrier range; as the controls were not genetically confirmed, one or more may actually be carriers based on a carrier frequency of about 1 in 100 for FRDA. While attempts were made to minimise the rate of false positives, a positive test would only result in assessment by DNA analysis, a relatively minor procedure. As a result, false positives are less problematic than false negatives when reflexing to molecular analysis for FRDA. In this study, the false negatives were all LOFA patients or carriers, and not cFRDA patients. This approach focuses on high PPV and high sensitivity to reduce the risk of ‘missing’ any FRDA patients. A low NPV was associated with buccal cells and whole blood; however, this value is heavily likely to reflect the small number of controls in the study underestimating the NPV of the test. In future clinical studies, the number of controls sampled may be much larger relative to the number of false positives, producing a higher NPV value.
Additionally, frataxin protein was also quantified with an xMAP-based Luminex assay from eluted DBS (figure 3A) in controls, carriers and patients to determine its usefulness as a diagnostic tool compared to the lateral flow immunoassay. Frataxin levels from DBS correlated strongly with levels from whole blood measured by the dipstick assay, and were significantly different between controls, carriers and FRDA patients (figure 3B), demonstrating that multiple assays detect similar relative amounts of frataxin. Both assays had areas under the ROC curve greater than 0.915 (figure 3C,D); however, the Luminex assay demonstrated a slightly higher sensitivity, specificity, NPV and PPV when identifying carriers and FRDA patients from a cohort (table 4), indicating that Luminex assay measuring frataxin from DBS leads to a lower rate of false negatives and false positives.
While both assays are easy to perform, there are advantages and disadvantages of each assay. The nature of the Luminex assay allows for multiplexing of analytes to investigate multiple markers of disease or markers for multiple diseases from a single sample. Only a small amount of blood is required for the Luminex assay, which is ideal for newborn screening. By contrast, the dipstick assay is clinically useful in settings where a Luminex instrument may not be available, and can be administered and quantified by anyone with minimal training. As with the Luminex assay, the dipstick assay is applicable to multiple cell types, including the rapidly and non-invasively collected buccal cell extracts from cheek swabs. Measurements from multiple buccal cell samples are ideal in clinical trial settings where blood may not be easily collected or stored. Additionally, the patient population may include younger children who are adverse to blood draws; family members of FRDA patients and normal healthy volunteers are also more likely to provide a buccal cell sample. This makes dipstick assays readily useful in studies requiring repeated measures as a part of the protocol.
Both immunoassays are useful for complementary diagnostic purposes and have their own advantages associated with them; the choice of assay used will, therefore, depend on the study question of interest, the composition of the study population, and the study design. In some ways they also offer advantages over molecular techniques focusing on GAA repeat size. Long GAA repeats are difficult to correctly size, and testing based on GAA repeat size inherently cannot detect the 3–5% of subjects with point mutations. There is some evidence that GAA repeat length has substantial somatic variability.29–31 Only one patient was reported to have no GAA repeats and still have the complete phenotype of FRDA (Schmucker et al, 4th Friedreich Ataxia Research Alliance International Scientific Conference, Strasbourg, France, 2011). While GAA repeat size does correlate with age of onset, it is not always clearly predictive of progression rate.25 ,32 All these represent limitations of assays based on GAA triplet repeat length. While direct testing of frataxin protein may not overcome all these, it seems likely that frataxin testing will be complementary to GAA repeat testing in clinical usefulness.
The authors wish to thank all FRDA patients, their relatives and all other individuals who volunteered blood and buccal cell samples for this study. The authors would also like to thank Drs Susan Perlman (UCLA), Sub Subramony (University of Florida), and George Wilmot (Emory) who provided samples from their FRDA patients.
Contributors All authors have seen and approved the final manuscript. Eric C Deutsch was responsible for performing frataxin assay, experimental design, interpreting data, statistical analysis, drafting the manuscript, and critical revision. Devin Oglesbee was responsible for performing frataxin assays, experimental design, interpreting data, and critical revision. Nathaniel R Greeley was responsible for performing frataxin assays, experimental design, and interpreting data. David R Lynch was responsible for experimental design, interpreting data, and critical revision.
Funding This work was funded by grants from the Friedreich Ataxia Research Alliance to DRL and to DO.
Competing interests None.
Patient Consent Obtained.
Ethics approval These studies were approved by the Institutional Review Board at The Children's Hospital of Philadelphia (IRB Study #2609).
The Institutional Review Board at The Children's Hospital of Philadelphia (IRB Study #2609).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no unpublished data available.