Autoantibody-Mediated Dysfunction of Sympathetic Neurons in Guillain-Barré SyndromeSympathetic Neuron Dysfunction in Guillain-Barré FREE

Autonomic dysfunction is a common and serious complication of Guillain-Barré syndrome (GBS). It occurs in up to two-thirds of all patients and can present with various symptoms, including orthostatic hypotension, abnormal sweating, gastrointestinal dysfunction, and bowel abnormalities.^{1 - 4} The most life-threatening autonomic disturbances are related to the cardiovascular system and include tachycardia, sustained hypertension, and blood pressure fluctuations.^{4 - 5} Most of these symptoms are characterized by a functional overactivity of the sympathetic arm and are difficult to treat, which increases the risk for a fatal outcome in those patients.^{6 - 7}

Despite its clinical importance, little is known about the underlying pathological mechanism of GBS-associated autonomic dysfunction. Prospective studies applying autonomic function test batteries suggest a functional imbalance of sympathetic and parasympathetic nerve fibers as a cause of cardiovascular dysautonomia in GBS.⁶ The few available pathological studies in humans generally report no or only mild changes in the autonomic nervous system. They indicate that sympathetic and parasympathetic nerve fibers are relatively spared in contrast to marked pathological changes in the motor and sensory fibers.^{4 ,8 - 9}

There is increasing and consistent evidence that autoantibodies play an important role as mediators of immune damage in the pathogenesis of GBS.^{10 - 11} Anti-ganglioside antibodies, which are associated with axonal GBS subtypes and the Miller-Fisher syndrome, have been shown to affect neural injury at the level of the nodes of Ranvier and the neuromuscular junction.¹¹ Other neural autoantigens have also been postulated to be a target of an aberrant B-cell response, based on experimental studies in which serum samples and purified IgG from patients with acute inflammatory demyelinating polyradiculoneuropathy induced demyelination and blockage of synaptic transmission.^{12 - 14}

In the current study, we sought to determine whether a misguided humoral immune response in GBS interferes with the function of autonomic nerve fibers. We evaluated the effects of purified IgG from GBS patients on catecholamine synthesis and release in sympathetic neurons and in an in vitro model of synaptic transmission between sympathetic neurons and cardiomyocytes.

Plasma filtrates from 7 patients (3 with GBS, 2 with chronic inflammatory demyelinating polyradiculoneuropathy [CIDP], and 2 with relapsing-remitting multiple sclerosis [RRMS]) were collected for this study. Their clinical features are described in the Table. Filtrates were obtained through therapeutic plasma exchange, with which all subjects were treated, consisting of 4 to 6 courses in the acute stage (GBS) or for steroid-refractory relapses (CIDP or RRMS) between 2005 and 2007. Per procedure, 1 to 1.5 plasma volumes were exchanged and replaced with human albumin and an electrolyte solution. Plasma filtrates from the first or second plasma exchange were collected and frozen for further use, with patient's informed consent being obtained.

The IgG fraction was purified from the plasma filtrate by precipitation and subsequent purification by protein G affinity chromatography according to the manufacturer's instructions (Affiland, Liège, Belgium). To elute low-molecular proteins, IgG fractions were subsequently dialyzed for 24 hours against 10 L of phosphate-buffered saline. Content and quantity of the eluate was determined by capillary electrophoresis using a Paragon CZE 2000 system (Beckman Coulter, Fullerton, California). All serum samples and purified IgG were screened and were found to be negative for IgG and IgM reactivity against the gangliosides GM1, GD1a, GD1b, and GQ1b.

Superior cervical ganglia and nodose ganglia of the vagal nerves were dissected from postnatal rats (aged 1-3 days), digested with trypsin, 0.25%, and collagenase I, 0.1%, washed, and further mechanically dissociated. The cells were then plated (100 000-200 000 for polymerase chain reaction [PCR], enzyme-linked immunosorbent assay, and Western blot; and 5000 for immunocytochemistry) on collagen-coated plastic dishes in Dulbecco modified Eagle medium (Invitrogen, Carlsbad, California) supplemented with fetal bovine serum, 10% (Hyclone, Logan, Utah), and 100-ng/mL nerve growth factor. Cell viability was determined by a fluorescence-based Live-Death Viability Kit (Molecular Probes, Eugene, Oregon) according to the manufacturer's instructions. After incubation with IgG, cell viability was determined by counting 100 to 200 cells per experimental condition from 3 randomly selected fields. In these and all subsequent experiments, cells were incubated with different human IgG in a concentration of 0.1 mg/mL, based on preliminary experiments and previous studies using comparable doses of purified human IgG in other in vitro paradigms.^{15 - 18} Each experimental condition (incubation with IgG from 1 patient with GBS, CIDP, or RRMS) was done in 3 to 4 wells and repeated 2 to 3 times. Because the underlying disease is the independent variable for our experiments, data are summarized and analyzed in a way to detect differences between each disorder.

To assess synaptic transmission of sympathetic neurons, a coculture system of sympathetic neurons and cardiomyocytes was used as described previously.^{19 - 20} Superior cervical ganglia and cardiomyocytes were isolated from 1- to 3-day-old rats. The dextroposterior half of the cardiac ventricles was dissected and dissociated for 30 minutes at 37°C with collagenase I, 0.1%, and subsequently for 10 minutes with collagenase I, 0.1%, and trypsin, 0.25%, in Dulbecco modified Eagle medium. After extensive washing, cardiomyocytes were triturated with a heat-polished Pasteur pipette and strained through a cell strainer (70 μm; BD Biosciences, Bedford, Maine), and approximately 100 000 cardiomyocytes were plated onto 35-mm collagen-coated plastic dishes together with 25 000 neurons. Sympathetic neurons and cardiomyocytes were cultured for 48 hours to allow the extension of neurites and establishment of synapses in MAH food (L15CO2 plus 10% fetal bovine serum, 6-mg/mL dextrose, 2mM glutamine, penicillin and streptomycin, 1-mg/mL 6,7, dimethyl-5,6,7,8-tetrahydropterine, 5-mg/mL glutathione, 50-mg/mL ascorbic acid, and 100-ng/mL nerve growth factor). The medium was replaced after 24 hours with MAH food containing 10μM cytosine arabinofuranoside for fibroblast elimination. The cardiomyocyte beat rate was determined after 72 hours in culture during a period of 2 minutes in 7 to 10 cell clusters per each experimental condition. The position of the cell clusters in which beat rate was established was then marked on the culture dish bottom to allow identification of the same clusters after 6 and 24 hours. In some experiments, the cultures were preincubated for 1 hour with prazosin (2μM) or propranolol (1μM) to block noradrenergic transmission as described previously.²¹

Sympathetic neurons or neurons from the nodose ganglia of the vagal nerve were plated on collagen-coated glass coverslips in 24 well plates. To stain living cells, human IgG was added (0.1 mg/mL) for 3 hours before cells were fixed in paraformaldehyde, 4%, for 30 minutes and blocked with bovine serum albumin, 5%, in phosphate-buffered saline for 1 hour. Subsequently, cells were developed with fluorescein isothiocyanate–conjugated specific anti-human IgG (1:500) for 2 hours. For double staining, the fixed cells were blocked and incubated overnight at 4°C with antibodies against tyrosine hydroxylase (1:1000; rabbit; Chemicon, Temecula, California), vesicular acetylcholine transporter (1:1000; goat, Chemicon), or β-III-tubulin (1:1000; mouse; Promega, Madison, Wisconsin) before they were developed with fluorescence-labeled specific secondary antibodies. Coverslips were mounted and images were acquired using a fluorescence microscope (Carl Zeiss, Jena, Germany). Cocultures of sympathetic neurons and cardiomyocytes were fixed in paraformaldehyde, 4%, for 30 minutes, blocked with bovine serum albumin, 5%, in Triton X, 0.1%, and incubated overnight with antibodies against tyrosine hydroxylase (1:1000), β-III-tubulin (1:1000), synaptophysin (1:500; rabbit, Dako, Glosturp, Denmark), or α-actinin (mouse, 1:1000). After washing, the cultures were developed with fluorescence-labeled specific secondary antibodies, and images were acquired using an inverted fluorescence microscope.

Tyrosine hydroxylase and choline acetyltransferase RNA expression in sympathetic neurons was investigated by quantitative real-time PCR. Total RNA was extracted from cell cultures after incubation with human IgG using TriZOL RNA isolation reagent (Invitrogen). Each human IgG was used 3 to 4 times in a different set of cell cultures. Reverse transcription was performed using TaqMan reverse transcription reagents (Applied Biosystems, Foster City, California) according to the manufacturer's instructions using an automated thermocycler (GeneAmp PCR System 9700; Applied Biosystems). Quantitative real-time PCR was performed using a real-time PCR kit that detects and quantitates nucleic acid sequences.²² Each reaction was performed in a total volume of 30 μL, including 50% TaqMan Universal PCR Master Mix. Primers and probes for tyrosine hydroxylase, choline acetyltransferase, p75NTR, and the housekeeping gene 18S were designed by and purchased from Invitrogen. Primer sequences are available on request. Real-time PCR ran for 42 cycles (15 seconds at 95°C and 60 seconds at 58°C). Samples were normalized to 18S ribosomal RNA and analyzed with the ABI PRISM 7700 Sequence Detection System using the comparative cycle of threshold method.

Supernatants (100 μL) from cocultures of sympathetic neurons and cardiomyocytes were collected before and 24 hours after incubation with human IgG (0.1 mg/mL). Samples were stored at −20°C in the presence of ethylenediaminetetraacetic acid (1mM) and sodium metabisulfite (4mM) to prevent catecholamine degradation. Noradrenaline levels were determined using a commercially available competitive enzyme-linked immunosorbent assay (Labor Diagnostika Nord, Nordhorn, Germany) in a total of 30 samples (10 GBS, 10 CIDP, and 10 RRMS) according to the manufacturer's instructions. In this assay, noradrenaline is first extracted, acylated, and then derivatized enzymatically. The noradrenaline in the samples competes with a fixed number of binding sites for antiserum. After removal of free antigen and free antigen-antiserum complexes by washing, the antibody bound to the solid phase is detectable by a peroxidase conjugate secondary antibody and measured at 450 nm. The corrected sensitivity of this assay is 2 ng/mL.

Sympathetic neurons were cultured as described above. Twenty-four hours after incubation with human IgG, the cells were harvested and lysed in buffer (Tris 50mM, 0.1% sodium dodecyl sulfate, 150mM sodium chloride, 0.5% sodium deoxycholate, and 1% NP-40) in the presence of protease inhibitors. Protein concentration was determined by bicinchoninic acid protein assay (Pierce, Rockford, Illinois), and 20 μg of protein was subjected to electrophoresis on 15% sodium dodecyl sulfate–polyacrylamide gel and subsequently transferred to a nitrocellulose membrane. After washing in 0.1% Tween-phosphate-buffered saline, blots were blocked in 5% milk powder for 1 hour. Tyrosine hydroxylase and β-actin were detected with anti–tyrosine hydroxylase antibody (1:1000) and anti-β-actin antibody (1:1000; BioVision, Mountain View, California), both incubated overnight at 4°C; horseradish-peroxidase–conjugated goat anti-rabbit and anti-mouse antibodies (1:1000; Invitrogen, Hamburg, Germany) were applied.

For statistical analysis, 1-way analysis of variance (Kruskal-Wallis test with the Dunn post hoc test) was used to compare group data. P < .05 was considered statistically significant.

In a first set of experiments, we determined if purified IgG from patients with GBS, CIDP, or RRMS bind to rodent autonomic tissue. Rodent sympathetic neurons displayed a clearly distinct staining pattern. Five of the 7 IgG fractions that were used stained sympathetic neurons, including all GBS-IgG samples (Figure 1A). Human IgG stained predominantly the sympathetic cell bodies, whereas neurites and neighboring Schwann cells remained unstained. In contrast, sensory neurons derived from the nodose ganglion of the vagal nerve were not stained by human IgG. The staining patterns of the different IgG are summarized in the Table. There was no correlation between staining intensity and patient's history of autonomic dysfunction.

Because human IgG can recognize antigens on the surface of rodent sympathetic neurons, we next asked if the survival or function of those cells might be compromised by the presence of human IgG from patients with GBS, CIDP, or RRMS. Cell survival was not altered in cell cultures exposed to IgG from patients with GBS or other human IgG (Figure 1B). In contrast, by analyzing the noradrenaline release before and after 24 hours' incubation with different IgG, noradrenaline levels decreased in cell cultures, which were incubated with CIDP-IgG or RRMS-IgG, but not in those incubated with purified GBS-IgG (Figure 1C). The synthesis of noradrenaline requires several enzymatic steps, including the oxidation of the amino acid tyrosine into dihydroxyphenylalanine. This reaction is catalyzed by tyrosine hydroxylase and represents the rate-limiting step in the noradrenaline synthesis.²³ We therefore determined the expression levels of tyrosine hydroxylase in sympathetic neurons, which were exposed to different IgG using quantitative PCR and Western blot. Tyrosine hydroxylase levels were upregulated in sympathetic neurons, which were incubated with GBS-IgG, but not in those exposed to IgG from patients with CIDP or RRMS on messenger RNA (mRNA) (Figure 2A) and protein level (Figure 2B and C). This indicates that the incubation with GBS-IgG stimulates the expression of tyrosine hydroxylase in sympathetic neurons, which is the key enzyme for the synthesis of catecholamines in sympathetic neurons.

It is known that during development, a proportion of sympathetic neurons can display a predominantly cholinergic phenotype by synthesizing acetylcholine instead of noradrenaline. Under the cell culture conditions described here, more then 95% of all β-III-tubulin–positive neurons also stained positive for tyrosine hydroxylase, whereas no positive staining could be detected with an antibody against vesicular acetylcholine transporter, a marker for cholinergic neurons (data not shown).²⁴ We also determined the mRNA levels of p75 neurotrophin receptor and choline acetyltransferase, a marker for cholinergic neurons by quantitative PCR in IgG-exposed sympathetic neurons. Choline acetyltransferase mRNA was below the detection limit in most samples, and the average mRNA levels of choline acetyltransferase and p75 did not yield any difference after exposure with different IgG (data not shown).

To further explore the functional relevance of these findings, we used a cell culture model for synaptic transmission in the autonomic nervous system. In this model, sympathetic neurons extend their axons toward clusters of cardiomyocytes and eventually form functional synapses (Figure 3A). Changes in synaptic transmission can be assessed by analyzing the cardiomyocyte beat rate. The formation of synapses was confirmed by staining with synaptophysin, a presynaptic marker that specifically stains synaptic vesicle glycoprotein (Figure 3B). Cardiomyocytes, which were cultured without sympathetic neurons, did not show any alteration of the beat rate during incubation with GBS-IgG (Figure 3D), which excludes the possibility of a purely pharmacological effect directed to cardiomyocytes. In contrast, when GBS-IgG was applied to cocultures of sympathetic neurons and cardiomyocytes, we observed a significant increase of the average beat rate (Figure 3E). Figure 3C shows a representative cluster of cardiomyocytes with innervating sympathetic neurons. A video of the GBS-IgG–induced changing of the beat rate from the same cluster (Figure 3C, right) is available http://www.archneurol.com. The increased beat rate of GBS-IgG–exposed cocultures could be inhibited by preincubation with prazosin and propranolol, 2 drugs that block α- and β-adrenergic receptors on cardiomyocytes (Figure 3F).

A potential inhibitory effect of intravenous immunoglobulin (IVIg) on the activation of tyrosine hydroxylase by GBS-IgG was explored in 2 different paradigms. To test whether IVIg neutralizes tyrosine hydroxylase–activating IgG, GBS-IgGs were preincubated for 30 minutes at 37°C with 10-mg/mL IVIg (GBS-IgG and IVIg). Because of the possibility that IVIg blocks the binding of GBS-IgG to sympathetic neurons, we incubated sympathetic neurons with 10-mg/mL IVIg for 1 hour prior to the administration of purified GBS-IgG. While preincubation of GBS-IgG and IVIg did not prevent upregulation of tyrosine hydroxylase RNA, the incubation of sympathetic neurons with IVIg blocked the GBS-IgG–induced increase of tyrosine hydroxylase RNA (Figure 2A). Similarly, cocultures of sympathetic neurons and cardiomyocytes that were incubated with IVIg prior to GBS-IgG exposure did not show an increased beat rate after 24 hours compared with cultures exposed to GBS-IgG only (Figure 3F).

Several lines of evidence suggest that an aberrant humoral immune response is involved in the pathogenesis of GBS. This is supported by studies demonstrating that IgG from patients with GBS exert complement-dependent and -independent effects on neurons and Schwann cells in different experimental paradigms.^{13 ,15 ,25 - 27} In this study, we explored the possibility that IgG from patients with GBS may also interfere with the synaptic transmission of autonomic nerve fibers. We found that IgG from patients with GBS can increase the synthesis and the release of noradrenaline from sympathetic neurons. The functional relevance of our findings is emphasized by the observation that the increased catecholamine synthesis alters the synaptic transmission between sympathetic neurons and cardiomyocytes. This implies that in GBS, circulating antibodies directed against sympathetic neurons may contribute to autonomic dysfunction via functionally relevant changes of noradrenaline synthesis.

To ensure that IgG of patients with GBS caused the change in the noradrenaline synthesis, we used highly purified IgG fractions in which the removal of the nongammaglobulin protein fractions was confirmed by capillary electrophoresis, thus excluding any unspecific effect of other plasma proteins including complement. Furthermore, IgG from patients with other autoimmune-mediated neurological diseases (CIDP and RRMS) did not show any changes in noradrenaline synthesis, which indicates that human IgG per se does not alter transmitter synthesis in sympathetic neurons. This is important, because our staining experiments demonstrate that IgG from some of the control patients can bind to sympathetic neurons as well.

Alteration of neuronal cell function through signal transduction of neuronal-specific autoantibodies has been proposed to cause neuronal dysfunction in a number of neurological disorders, including GBS,^{11 ,28 - 29} autoimmune autonomic ganglionopathy,^{18 ,30 - 31} Lambert-Eaton myasthenic syndrome,³² and Sydenham chorea.³³ The observed staining patterns suggest that the binding of GBS-IgG to the cell surface of neuronal perikarya could lead to an altered intracellular signaling that activates tyrosine hydroxylase. This enzyme appears to be an attractive target, because it is exclusively synthesized in the neuronal cell body and subsequently transported by slow axonal transport to the terminal nerve endings.³⁴ In addition, it is the rate-limiting enzyme for the catecholamine biosynthesis, its activity can be stimulated by extracellular signaling molecules, and it is controlled by several mechanisms, such as neuronal activity and end product inhibition under normal conditions.²³

Antibodies that alter the function of autonomic nerve fibers have also been reported in autoimmune autonomic ganglionopathy and Lambert-Eaton myasthenic syndrome. In autoimmune autonomic ganglionopathy, approximately 50% of patients have antibodies against the neuronal nicotinic acetylcholine receptor, and these antibodies have been found to impair the synaptic transmission in autonomic ganglia.^{18 ,30} Lambert-Eaton myasthenic syndrome is a neurological disorder caused by IgG autoantibody-mediated downregulation of P/Q-type voltage-gated calcium channels at motor nerve terminals.^{32 ,35} Rodent sympathetic and parasympathetic neurons exposed to IgG from 4 patients with Lambert-Eaton myasthenic syndrome containing voltage-gated calcium channel antibodies displayed a significantly reduced release of neurotransmitters, which was indirectly determined by changes of muscle contractions in the innervated organs.³² Although we had no direct evidence for increased stress to neurons exposed to GBS-IgG, we cannot exclude that the increase in tyrosine hydroxylase expression is part of a broader stress response of sympathetic neurons. This explanation would imply a different pathomechanism than a specific IgG autoantibody–mediated alteration of autonomic nerve fiber function like in autoimmune autonomic ganglionopathy and Lambert-Eaton myasthenic syndrome. Subsequent studies are required to explore the underlying mechanism of changes in the noradrenaline synthesis induced by GBS-IgG.

In contrast to autoimmune autonomic ganglionopathy and Lambert-Eaton myasthenic syndrome, it still remains elusive which antigens are targeted by autoantibodies in most patients with GBS. There is considerable evidence that antibodies against gangliosides are involved in the pathogenesis of some GBS subtypes, such as the Miller-Fisher syndrome and the acute motor axonal neuropathy.^{36 - 42} However, in those variants, generalized autonomic dysfunction is rarely seen,⁴³ which argues against a relevance of anti-ganglioside antibodies as a cause for autonomic dysfunction. The absence of anti-ganglioside immunoreactivity in the serum samples of our patients rather indicates that other yet undetermined antigens are involved in the GBS-IgG–mediated increase of noradrenaline synthesis.

The variety of syndromes involved in autonomic dysfunction in patients with GBS^{1 ,4 ,44 - 46} suggests that sympathetic and parasympathetic function can be either reduced or overactive in GBS. However, prospective case series have shown that most symptoms are related to a sustained overactivity of the sympathetic arm.^{6 - 7} Among cardiovascular symptoms, several studies have consistently reported that the most common abnormality is sinus tachycardia, followed by sustained or paroxysmal blood hypertension.^{4 ,6 - 7 ,44} By studying heart rate variability in patients with GBS by power spectrum analysis, Flachenecker and colleagues⁶ demonstrated that in GBS the sympathovagal balance of the cardiovascular innervation is clearly shifted to a sympathetic predominance during the early disease course. Similarly, sympathetic nerve activity of the skin has been reported to be increased during the acute phase of GBS in patients with autonomic dysfunction.⁷ Further support for an excessive sympathetic activity in GBS comes from studies that have shown increased noradrenaline levels in plasma and elevated catecholamine derivatives in the urine of patients with GBS, which correlated with the autonomic dysfunction.^{4 ,47 - 49} Our findings are consistent with those clinical observations and provide a novel link between an aberrant humoral immune response and autonomic dysfunction in GBS. A disturbed balance of the complex interaction between sympathetic and parasympathetic innervations may explain other symptoms of autonomic dysfunction not directly related to increased sympathetic outflow, which sometimes even occur in the same patient. It may also explain the lack of a strong correlation between clinical signs of dysautonomia and in vitro noradrenaline upregulation of our GBS samples. Our study does not preclude that other pathologic mechanisms are involved in the autonomic failure in GBS, such as cell-mediated structural breakdown of sympathetic and parasympathetic nerve fibers.

Although GBS is considered to be a strictly peripheral nerve disorder, a high proportion of patients with GBS can experience mental state abnormalities during a severe disease course.^{46 ,50} Interestingly, these mental abnormalities are strongly associated with clinical signs of autonomic dysfunction and are most likely the result of complications such as periodic hypertensive encephalopathy and/or posterior reversible leukoencephalopathy, which are occasionally observed in GBS.^{45 ,51 - 52} Another hypothesis that could be addressed in further studies is that circulating autoantibodies, which increase the activity of tyrosine hydroxylase in postganglionic sympathetic neurons, could also induce changes in the neurotransmitter synthesis of catecholaminergic neurons in the central nervous system.

Our observation that IVIg protects sympathetic neurons from increased noradrenaline synthesis expands previous observations that IVIg can neutralize pathogenic effects of GBS-associated autoantibodies in vitro; IVIg is an established treatment for GBS and although its efficacy with regard to the autonomic dysfunction has never been studied in clinical trials, case reports suggest that it significantly reduces symptoms of dysautonomia.⁵³ The beneficial effect of IVIg has been attributed to several mechanisms, including inactivation of the complement, neutralization by anti-idiotypic antibodies, and reduced binding of autoantibodies to the target. Our results imply that IVIg may contain antibodies, which bind to the cell surface, thus preventing GBS-IgG to access their antigens on the surface of sympathetic neurons. In contrast, a neutralizing effect by anti-idiotypic antibodies seems unlikely, because preincubation of GBS-IgG with IVIg had no protective effect in our model.

In summary, our study reveals a new mechanism through which IgG autoantibodies contribute to the pathogenesis of GBS. Autoantibodies that recognize targets on the surface of postganglionic sympathetic neurons can interfere with the noradrenaline synthesis and the synaptic transmission. This pathogenic effect may contribute to autonomic dysfunction, which can be frequently observed during the acute stage of the disease. Whether these results are applicable to other autoantibody-mediated disorders in which dysautonomia may occur warrants further investigation.

Correspondence: Helmar C. Lehmann, MD, Department of Neurology, Heinrich-Heine-University, Moorenstrasse 5, 40225 Düsseldorf, Germany (helmar.lehmann@uni-duesseldorf.de).

Accepted for Publication: August 4, 2009.

Author Contributions:Study concept and design: Lehmann, Meyer zu Hörste, and Kieseier. Acquisition of data: Lehmann, Jangouk, Kierysch, and Hartung. Analysis and interpretation of data: Lehmann, Jangouk, Meyer zu Hörste, and Kieseier. Drafting of the manuscript: Lehmann, Jangouk, Kierysch, Meyer zu Hörste, and Kieseier. Critical revision of the manuscript for important intellectual content: Lehmann, Jangouk, Meyer zu Hörste, and Hartung. Statistical analysis: Lehmann and Meyer zu Hörste. Administrative, technical, and material support: Jangouk, Meyer zu Hörste, Hartung, and Kieseier. Study supervision: Hartung and Kieseier.

Financial Disclosure: None reported.

Funding/Support: This work was supported by a grant from the Forschungskommission of the Heinrich-Heine University. Dr Lehmann is currently supported by grant LE2368/1-1 from the German Research Foundation.

Additional Contributions: Drs Stephen C. Reingold and Karlheinz Reiners provided helpful suggestions.