A, In vivo confocal microscopy (IVCM) image of subbasal corneal nerves in a control participant. B, After tracing the nerve, NeuronJ software (National Institutes of Health, http://www.imagescience.org/meijering/software/neuronj/) quantifies the length of the traced nerves and provides corneal nerve fiber length in millimeters per millimeters squared.
Moderately affected individual (patient 5) with lower-limb skin biopsy (LLSB) indirect immunofluorescence: arrowheads indicate intraepidermal nerve fibers (IENFs) (A), and in vivo confocal microscopy (IVCM) shows diminished innervation (B). Severely affected individual (patient 3) with LLSB indirect immunofluorescence showing no IENF (C) and IVCM indicating scarce corneal nerves.
CADT indicates Compound Autonomic Dysfunction Test; ESC, electrochemical skin conductance, NIS-LL, Neuropathy Impairment Score of the Lower Limbs; ONLS, Overall Neuropathy Limitations Scale; and PND, Polyneuropathy Disability Scale. Bullets indicate individual patients’ values; solid lines, statistical significance; dashed line, statistical nonsignificance.
eTable. Clinical, Genetic, and IENFD Data of Patients
Customize your JAMA Network experience by selecting one or more topics from the list below.
Rousseau A, Cauquil C, Dupas B, et al. Potential Role of In Vivo Confocal Microscopy for Imaging Corneal Nerves in Transthyretin Familial Amyloid Polyneuropathy. JAMA Ophthalmol. 2016;134(9):983–989. doi:10.1001/jamaophthalmol.2016.1889
Small fiber neuropathy (SFN) is an important feature of transthyretin familial amyloid polyneuropathy (TTR-FAP). A practical and objective method for the clinical evaluation of SFN is needed to improve the management of this disease. In vivo confocal microscopy (IVCM) of the corneal nerves, a rapid noninvasive technique, may be used as a surrogate marker of SFN.
To determine the correlation of SFN with IVCM in patients with TTR-FAP.
Design, Setting, and Participants
A prospective, single-center, cross-sectional controlled study was conducted at the French National Reference Center for TTR-FAP from June 1, 2013, to June 30, 2014. Fifteen patients with TTR-FAP underwent a complete neurologic examination, including Neuropathy Impairment Score of the Lower Limbs, hand grip strength, and evaluation of vegetative dysfunction, as well as electrophysiologic studies (nerve conduction and electrochemical skin conductance) and intraepidermal nerve fiber density quantification. Patients and 15 controls (matched for age and sex) underwent ophthalmologic assessments, including corneal esthesiometry and IVCM.
Main Outcomes and Measures
Correlation of corneal nerve fiber length (CNFL) with the severity of SFN.
Of the 15 patients enrolled in the study, 6 were women (40%); mean (SD) age was 54.4 [13.7] years. The CNFL was shorter in the patients than in controls (13.08 vs 17.57 mm/mm2; difference of 4.49 [95% CI, 0.72 to 8.27]; P = .02). The patients’ CNFL correlated with the severity of both autonomic neuropathy assessed by the Compound Autonomic Dysfunction Test (rs = 0.66 [95% CI, 0.22 to 0.87]; P = .008) or electrochemical skin conductance (rs = 0.80 [95% CI, 0.50 to 0.93]; P < .001) and sensorimotor neuropathy assessed using the Neuropathy Impairment Score of the Lower Limbs (rs = −0.58 [95% CI, −0.84 to −0.11]; P = .02). Patients with altered sensory nerve action potentials and intraepidermal nerve fiber density had a shorter CNFL (P = .04 and P = .02, respectively). The CNFL could be measured in all patients compared with sensory nerve action potentials (11 patients [73%; 95% CI, 44% to 92%]; P < .001) and intraepidermal nerve fiber density (4 patients [27%; 95% CI, 8% to 55%]; P < .001).
Conclusions and Relevance
In these 15 patients with TTR-FAP, IVCM measurement permitted rapid, noninvasive evaluation of small-fiber alterations in patients and could be used to assess SFN in this setting. The CNFL could be measured in all patients, thus avoiding the floor effect seen with other neuropathy measures. Longitudinal studies with more cases evaluated are needed to define the place of IVCM in monitoring patients with TTR-FAP.
Transthyretin familial amyloid polyneuropathy (TTR-FAP) is a rare but severe genetic disease with autosomal dominant transmission caused by a mutation in the transthyretin gene.1 It is responsible for a disabling peripheral sensorimotor and autonomic neuropathy in adulthood,2 together with cardiac1 and ocular involvement.3 Antiamyloid therapies, including liver transplant and TTR stabilizers, can slow or halt disease progression.4 The neuropathy is assessed using various clinical markers, including the Neuropathy Impairment Score, which correlates well with locomotion, handgrip, and disease stage.5 New sensitive tools are needed to detect epidermal6 and sudomotor7 denervation to start treatment as early as possible.
Intraepidermal nerve fiber density (IENFD) measurement is an objective and quantitative method for diagnosing small fiber neuropathy (SFN),8 but IENFD is invasive, time consuming, and requires expertise, making it unsuitable for repeated examinations.
In vivo laser scanning confocal microscopy (IVCM) of the cornea is a rapid, noninvasive technique used to study subbasal corneal nerve fiber length (CNFL).9,10 The CNFL is considered a surrogate marker of denervation in diabetic sensorimotor and autonomic neuropathy.11,12 The aim of the present study was to determine whether CNFL, evaluated with IVCM, was correlated with the severity of SFN as assessed with clinical and paraclinical markers currently used to evaluate sensorimotor and autonomic SFN.13-16
Question Is in vivo confocal microscopy a good tool to evaluate small fiber alterations in familial amyloid polyneuropathy (FAP)?
Findings This cross-sectional controlled pilot study revealed that corneal nerve fiber length (CNFL) was shorter in 15 patients with FAP than in controls. Patients' CNFL correlated with the severity of autonomic and sensory-motor neuropathy as well as with small fiber neuropathy assessed by skin biopsy.
Meaning In vivo confocal microscopy could be a rapid noninvasive tool to assess small-fiber loss in FAP.
A prospective, single-center, cross-sectional controlled study was conducted at the French National Reference Center for TTR-FAP. The study population consisted of patients with TTR-FAP who had undergone neurologic examinations and skin biopsies. Their IVCM findings were compared with those obtained in healthy controls selected from staff and staff relatives and matched for sex and age (within 5 years). The controls had no history of chronic corneal disease or ocular surgery.
The study was approved by the Ethics Committee of the French Society of Ophthalmology and conformed to the Declaration of Helsinki.17 Written informed consent was obtained from all participants; no financial compensation was provided.
The patients were part of a large TTR-FAP cohort monitored in the French National Reference Center for FAP. They were selected during a 12-month period from June 1, 2013, to June 30, 2014. All patients with TTR-FAP referred for ophthalmologic evaluation were invited to enroll.
Patients were eligible for the study if had genetically confirmed TTR-FAP, were older than 18 years, and had a skin biopsy performed during routine monitoring. In each patient, clinical examination, skin biopsy, and IVCM had to be performed within the same 3-month period. The TTR-FAP diagnosis was confirmed by genetic analysis showing an established amyloidogenic mutation of the transthyretin gene18 and by tissue biopsy showing amyloid deposits.13 Patients were not eligible for the study if they had a history of a disease known to cause peripheral neuropathy (eg, chronic alcoholism and diabetes), a chronic corneal disorder, or previous ocular surgery.
All of the patients underwent a comprehensive neurologic examination that included sensory and motor testing and tendon reflexes. Lower limb sensory motor involvement was assessed with the Neuropathy Impairment Score of the Lower Limbs (NIS-LL).19 The NIS is a composite clinical scoring system widely used to assess the severity of peripheral neuropathy. The NIS-LL is a subset of the NIS that assesses function of the lower limbs, which are most affected in TTR-FAP. The NIS-LL provides a score ranging from 0 (normal) to 88 (total impairment) to the clinical abnormalities noted in the physical assessment of sensation, muscle power, and tendon reflexes. The assessment of sensation (sensory NIS-LL) ranges from 0 (normal) to 16 (complete impairment). Functional impairment was estimated by testing bilateral hand grip strength with a Jamar analog hand dynamometer,20 and the value of the dominant arm was used for correlation studies. Autonomic involvement was tested with the Compound Autonomic Dysfunction Test (CADT), which is a questionnaire designed to assess the main symptoms of autonomic dysfunction,21 with scores ranging from 16 (no symptoms of autonomic dysfunction) to 0 (all symptoms of autonomic dysfunction). Functional impairment was evaluated with 2 tests: Overall Neuropathy Limitations Scale (ONLS)22 and the Polyneuropathy Disability (PND) Scale.23 The ONLS scoring ranges from 0 (no disability) to 12 (maximum disability); the PND Scale categorizes the limitations as stage 0, no impairment; stage I, sensory disturbance but preserved walking capability; stage II, impaired walking capability but ability to walk without a stick; stage IIIA, walking only with the help of 1 stick; stage IIIB, walking only with the help of 2 sticks or crutches; and stage IV, confined to a wheelchair or bedridden.
Conventional nerve conduction studies were used to assess large nerve fiber involvement; compound muscle action potentials and sensory nerve action potentials (SNAPs) were recorded in all 4 limbs. SNAPs were recorded in an antidromic manner for the lower limbs and radial nerve and in an orthodromic manner for other nerves in the upper limbs. Data were collected on compound muscle action potential amplitudes, terminal motor latency, conduction velocity and minimal F-wave latency for motor nerves, and SNAP amplitudes and distal conduction velocity for sensory nerves. We mostly studied sural SNAP amplitudes, which are included in the routine protocol for patients with TTR-FAP since they are the first to be affected by the disease.24
Two skin specimens were obtained from each patient by means of a 3-mm punch biopsy to gather information on a length-dependent process, in keeping with the 2010 European Federation of Neurological Societies guideline.25 A distal lower limb specimen was obtained 10 cm above the lateral malleolus, and a proximal lower limb specimen was obtained from the upper lateral thigh 20 cm below the anterior iliac spine.25-27 The specimens were immediately fixed in paraformaldehyde, 4%, for 3 to 4 hours at 4°C, rinsed with phosphate buffer saline, kept in sucrose solution for 12 hours, and embedded in adragante gum, then frozen and serially sectioned with a cryostat. Sections 10-µm thick were stained with hematoxylin-eosin, and Congo red was used to detect amyloid deposits. Linear IENFD (number of fibers per millimeter) was quantified by a single pathologist (C.L.) using a bright-field immunohistochemistry protocol (indirect immunofluorescence with antibodies against protein gene product 9.5) in at least three 50-μm sections,25,27 in keeping with published guidelines and counting rules.28 Electrochemical skin conductance (ESC) was measured on the hands and feet using a simple, rapid, noninvasive sudomotor test (Sudoscan; Impeto Medical).29
A single ophthalmologist (A.R.) screened the patients for ocular manifestations of FAP and all participants for confounding causes of corneal hypoesthesia. This examination included best-corrected visual acuity, slitlamp examination with fluorescein corneal staining,30 Goldmann tonometry, and dilated ophthalmoscopic examination. Corneal sensitivity was measured at the center of the cornea31 by a single masked examiner (B.D.) using an esthesiometer with a single-use thread tip (Cochet-Bonnet).
In vivo confocal microscopy of the cornea was performed in the patients and controls by the same examiner (B.D.) as previously described (Rostock Cornea Module of the Heidelberg Retina Tomograph; Heidelberg Engineering GmbH).9 A drop of topical anesthetic (oxybuprocaine, 0.4%; MSD-Chibret) and a drop of gel tear substitute (carbomer 980, 0.2%; Europhta) were instilled into each eye before IVCM. For each patient and matched control, 15 images of the central CNF were captured at the same intensity of illumination with the microscope focused beneath the basal epithelium.
To avoid selection bias, 1 eye of each participant was randomly selected with Research Randomizer software, a free net-based program that generates sequences of random digits (https://www.randomizer.org). For each selected eye, 5 CNF images were randomly selected for analysis with NeuronJ software by 2 independent observers (A.R. and A.L.) who were unaware of the participant’s identity and neurologic results. NeuronJ is a free ImageJ plug-in (National Institutes of Health, http://www.imagescience.org/meijering/software/neuronj/) designed to trace and quantify elongated image structures, such as nerves, and is used widely in IVCM studies.32 We assessed CNFL, defined as the total length of all nerves visible per frame (expressed as millimeters per millimeters squared) (Figure 1). The results are expressed as the mean (SD) value for the 5 selected images. As reviewed by Patel and McGhee,33 among the different corneal nerve variables that can be assessed with IVCM, CNFL benefits from the best intraoperator and interoperator reproducibility and repeatability. Previous studies34,35 have shown good intraoperator and interoperator reproducibility in CNFL measures, intraobserver repeatability greater than 90%, and interobserver variation of approximately 10%.
Because the study sample was small, we used nonparametric tests: Spearman correlation coefficients and the randomized version of the Hotelling test for multivariate mean comparisons. For the Hotelling test, simultaneous 95% CIs rather than individual t-test–based 95% CIs were determined. The nonparametric Kruskal-Wallis test was used to compare means.
Seventeen patients were initially recruited, but 1 individual refused IVCM and another patient was excluded because of severe ocular surface abnormalities and a history of multiple ocular surgical interventions. The final patient population comprised 15 individuals with various stages of the disease who were members of 14 families (6 women [40%] and 9 men [60%]; mean [SD] age, 54.4 [13.7] years), with 5 different transthyretin amyloidogenic mutations (V30M in 10 cases). Fifteen healthy individuals served as controls (6 women [40%] and 9 men [60%]; mean [SD] age, 54.5 [12.9] years). The patients' PND Scale scores were as follows: stage 0, 2 patients (13%); stage I, 6 patients (40%); stage II, 3 patients (20%); stage IIIA, 2 patients (13%); and stage IIIB, 2 patients (13%). Complete clinical, neurophysiologic, genetic, and IENFD data, as well as patient identification numbers, are available in the eTable in the Supplement. Fourteen patients (93%) had symptomatic disease: 12 had a predominantly neuropathic phenotype, and 2 had a predominantly cardiac phenotype. The data were recorded anonymously before retrospective statistical analysis with SAS, version 9.4 (SAS Institute) and R, version 18.104.22.168
The patients’ CNFL values ranged from 2.6 to 28.7 mm/mm2. The CNFL and esthesiometry values were lower in the patients compared with the controls (13.08 vs 17.57 mm/mm2; difference of 4.49 [95% CI, 0.72-8.27]; and 58 vs 48 mm; difference of 10 mm [95% CI, 0.44-10.56], respectively; randomized Hotelling test, P = .02). All 5 patients with the lowest CNFL values had the V30M mutation.
The IENFD on the distal leg was null in 11 patients (73% [95% CI, 44%-92%; P < .001) (including 9 with the V30M mutation), below the 5th percentile of normative values for age and sex26 in 2 patients (13%), and normal in 2 patients (13%) (1 asymptomatic carrier and 1 person with a predominantly cardiac phenotype). Because the distal leg IENFD was null in most cases, we decided to study the association between CNFL and the mean lower-limb IENFD (the mean of the proximal and distal leg IENFD values). The patients were therefore divided into 3 tertiles according to the mean lower-limb IENFD values. The mean CNFL values in these 3 subgroups (7.94, 14.41, and 16.90 mm/mm2) were significantly different (P = .02; nonparametric Kruskal-Wallis test) (Figure 2). The CNFL values correlated with the ESC values both on the feet (Spearman rs = 0.80 [95% CI, 0.50-0.93]; P < .001) and the hands (Spearman rs = 0.61 [95% CI, 0.15-0.86]; P = .02) (Figure 3A and B).
The CNFL values correlated with the NIS-LL sensory subscore (Spearman rs = −0.51 [95% CI, −0.81 to −0.01]; P = .049) (Figure 3C). Because sural SNAPs were null in 4 of the 15 patients (27% [95% CI, 8% to 55%]; P < .001), correlation studies would not have been meaningful. The patients were therefore divided into 3 subgroups according to the SNAP tertiles. Mean CNFL values in these 3 groups (10.00, 11.54, and 17.71 mm/mm2) were different in the nonparametric Kruskal-Wallis test (P = .04). The CNFL ranged from 14.3 to 28.2 mm/mm2 in the 4 patients with normal SNAP amplitudes: 2 patients with symptomatic V30M TTR polyneuropathy, 1 V122I patient with a predominantly cardiac phenotype, and 1 asymptomatic patient. The IENFD was normal in the asymptomatic patient and abnormal in the other 3 patients.
The CNFL correlated with the CADT score (Spearman rs = 0.66 [95% CI, 0.22 to 0.87]; P = .008) (Figure 3D). The CNFL also correlated with the total NIS-LL score (rs = −0.58 [95% CI, −0.84 to −0.11]; P = .02) (Figure 3E). Correlation was noted with the dominant-arm Jamar score (Spearman rs = 0.65 [95% CI, 0.21 to 0.87]; P = .008) (Figure 3F) and the PND Scale score (Spearman rs = −0.53 [95% CI, −0.82 to −0.02]; P = .04) (Figure 3G). There was no correlation between CNFL and the Overall Neuropathy Limitations Scale score (Spearman rs = −0.50 [95% CI, −0.81 to 0.02]; P = .06) (Figure 3H).
The findings of the present study indicate that CNFL is clearly lower in patients with TTR-FAP than in matched healthy controls and that it correlates negatively with the severity of the neuropathy. Corneal nerve fiber length could be measured in all 15 patients included in this study compared with SNAP (73%) and IENFD (27%), thus avoiding the floor effect seen with these measures. Our IVCM findings therefore show that the corneal nerves are damaged in patients with TTR-FAP as demonstrated in other acquired and inherited SFNs, such as diabetes,11 Fabry disease,37 and Charcot-Marie Tooth neuropathy type 1.38
We also found that CNFL correlated with the severity of both the sensorimotor and autonomic neuropathy in TTR-FAP, as evaluated with clinical and paraclinical tests. A strong correlation was demonstrated with the classic clinical neurologic impairment score and with sural SNAP amplitudes, as well as with clinical motor neuropathy and walking status. A correlation between CNFL and the severity of diabetic sensorimotor neuropathy was shown in a study39 of 231 patients with diabetes and 61 controls assessed with quantitative sensory tests, nerve conduction study, symptom scales, and disability scores.
The CNFL also correlated with the global clinical autonomic CADT score21 and with the neurophysiologic autonomic sudomotor test. The ESC measured on the feet, which correlates with both sensory and autonomic neuropathy in patients with diabetes,15 was also strongly associated with CNFL in our study. These results are in keeping with the dual autonomic and sensory innervation of the cornea.40 A reduction in CNFL was noted in 34 patients with diabetes and autonomic neuropathy.12 In this latter study, CNFL correlated with a composite autonomic neuropathy symptom score and was highly sensitive and specific for the diagnosis of diabetic autonomic neuropathy.12 Furthermore, a correlation between CNFL and clinical sudomotor test results was recently reported.41
The CNFL also correlated with IENFD in the lower limbs. We used the mean lower limb IENFD value since fiber loss in the ankle did not permit correlation of CNFL with ankle IENFD. Our results suggest that CNFL could be a useful and sensitive marker for TTR-FAP denervation, with numerous advantages over preexisting methods. Indeed, CNFL was always measurable in our cohort of 15 patients with different severities and stages of TTR-FAP, whereas skin biopsy specimens showed null IENFD values in 11 patients (73%), and sural nerve SNAP amplitudes were null in 4 (27%). Measurement of CNFL is also simple, rapid (acquisition requires approximately 15 minutes), noninvasive, and repeatable, whereas IENFD is time consuming (1 day for the technique and 1 hour for measurement) and the results are not available for at least 1 month. Nevertheless, skin biopsy appears to be a sensitive tool in TTR-FAP,6 showing a decrease in IENFD in 59% of 46 asymptomatic patients with various TTR mutations.
In vivo confocal microscopy has been shown11,37,38 to be an effective method for assessing SFNs of various causes, with diabetes being the most extensively studied. The CNFL may be less sensitive than current methods for detecting early denervation: in this study, the 2 patients with PND Scale scores of 0 (patients 2 and 12) had normal CNFL values.
The correlation we observed between CNFL and clinical severity suggests that CNFL is likely to be reduced as the disease progresses. However, a large prospective study of serial CNFL measurement in presymptomatic patients and healthy controls will be needed to determine whether CNFL changes can serve to monitor the progression of denervation.
In vivo confocal microscopy is a reproducible method of CNFL measurement and is suitable for longitudinal studies of patients with diabetes.42 This method might also prove useful for monitoring TTR-FAP progression in patients receiving new antiamyloid therapies, such as TTR-tetramer stabilizers (tafamidis and diflunisal), small interfering RNAs, and antisense oligonucleotides targeting TTR.4,43,44
One limitation of this study is its small size, which results in wide 95% CIs that could be reduced with more participants. In addition, the semiautomated quantification of CNFL may be examiner dependent. Automated CNFL quantification would facilitate the use of IVCM for care in patients with SFNs and has been shown45 to provide results comparable to those of manual and semiautomated methods. To use IVCM for diagnosis and monitoring of TTR-FAP, it will first be necessary to determine pathologic threshold values. A multinational normative data set was recently published,46 but the study used software that is not commercially available at present.
Subbasal CNFL measurement with IVCM provides a rapid, noninvasive means of evaluating small-fiber alterations in TTR-FAP. The results of the present study correlate with clinical and paraclinical measures of sensorimotor and autonomic polyneuropathy. Longitudinal studies evaluating more cases are needed to assess the capacity of IVCM to detect changes in CNFL over time and to define the place of this method in monitoring patients with presymptomatic or symptomatic TTR-FAP.
Corresponding Author: David Adams, MD, PhD, Department of Neurology, Bicêtre Hospital, APHP, Paris-Sud University, INSERM U 1195, French Reference Center for FAP (NNERF), 94275 Le Kremlin Bicêtre, France (firstname.lastname@example.org).
Submitted for Publication: January 19, 2016; final revision received April 29, 2016; accepted May 1, 2016.
Published Online: June 30, 2016. doi:10.1001/jamaophthalmol.2016.1889.
Author Contributions: Drs Rousseau and Cauquil had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Rousseau, Cauquil, Labbé, Théaudin, Labetoulle.
Acquisition, analysis, or interpretation of data: Rousseau Cauquil, Dupas, Labbé, Baudouin, Barreau, Lacroix, Guiochon-Mantel, Benmalek, Adams.
Drafting of the manuscript: Rousseau, Cauquil, Barreau, Guiochon-Mantel, Benmalek, Labetoulle, Adams.
Critical revision of the manuscript for important intellectual content: Rousseau, Cauquil, Dupas, Labbé, Baudouin, Théaudin, Lacroix, Labetoulle, Adams.
Statistical analysis: Cauquil, Benmalek.
Administrative, technical, or material support: Cauquil, Dupas, Baudouin.
Study supervision: Cauquil, Labbé, Baudouin, Théaudin, Labetoulle, Adams.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Lacroix and Cauquil received personal fees from Pfizer outside the submitted work. Dr Adams received personal fees from Pfizer Europe, Pfizer Inc, ALNYLAM, and GSK outside the submitted work. No other disclosures were reported.
Create a personal account or sign in to: