Nobuhiro Yuki, M.D., Ph.D., and Hans-Peter Hartung, M.D.
N Engl J Med 2012; 366:2294-2304
June 14, 2012
The Guillain–Barré syndrome, which is characterized by acute areflexic paralysis with albuminocytologic dissociation (i.e., high levels of protein in the cerebrospinal fluid and normal cell counts), was described in 1916.1 Since poliomyelitis has nearly been eliminated, the Guillain–Barré syndrome is currently the most frequent cause of acute flaccid paralysis worldwide and constitutes one of the serious emergencies in neurology. A common misconception is that the Guillain–Barré syndrome has a good prognosis — but up to 20% of patients remain severely disabled and approximately 5% die, despite immunotherapy.2 The Miller Fisher syndrome, which is characterized by ophthalmoplegia, ataxia, and areflexia, was reported in 1956 as a likely variant of the Guillain–Barré syndrome, because the cerebrospinal fluid of affected patients showed albuminocytologic dissociation.3 Furthermore, frank Guillain–Barré syndrome has developed in some patients with the Miller Fisher syndrome.4
Various studies of the immunopathogenesis of the Guillain–Barré syndrome suggest that the disease actually encompasses a group of peripheral-nerve disorders, each distinguished by the distribution of weakness in the limbs or cranial-nerve–innervated muscles and underlying pathophysiology (Figure 1Figure 1
Spectrum of Disorders in the Guillain–Barré Syndrome and Associated Antiganglioside Antibodies.).5-7 There is substantial evidence to support an autoimmune cause of this syndrome, and the autoantibody profile has been helpful in confirming the clinical and electrophysiological relationship of the typical Guillain–Barré syndrome to certain other peripheral-nerve conditions. This review considers the current understanding, diagnosis, and management of the Guillain–Barré syndrome.
The reported incidence of the Guillain–Barré syndrome in Western countries ranges from 0.89 to 1.89 cases (median, 1.11) per 100,000 person-years, although an increase of 20% is seen with every 10-year rise in age after the first decade of life.8 The ratio of men to women with the syndrome is 1.78 (95% confidence interval, 1.36 to 2.33).
Two thirds of cases are preceded by symptoms of upper respiratory tract infection or diarrhea. The most frequently identified infectious agent associated with subsequent development of the Guillain–Barré syndrome is Campylobacter jejuni, and 30% of infections were attributed to C. jejuni in one meta-analysis,9 whereas cytomegalovirus has been identified in up to 10%.10,11 The incidence of the Guillain–Barré syndrome is estimated to be 0.25 to 0.65 per 1000 cases of C. jejuni infection, and 0.6 to 2.2 per 1000 cases of primary cytomegalovirus infection.12 Other infectious agents with a well-defined relationship to the Guillain–Barré syndrome are Epstein–Barr virus, varicella–zoster virus, and Mycoplasma pneumoniae .10,11,13
During a 1976 mass immunization against A/New Jersey/1976/H1N1 “swine flu” in the United States, people who received the vaccine were at increased risk for the development of the Guillain–Barré syndrome.14 Other seasonal influenza vaccines have not been associated with the same increase in risk. With the pandemic influenza A (H1N1) outbreak in 2009, there was great concern that vaccination against H1N1 might also trigger the Guillain–Barré syndrome, but that did not occur.15
The first symptoms of the Guillain–Barré syndrome are numbness, paresthesia, weakness, pain in the limbs, or some combination of these symptoms. The main feature is progressive bilateral and relatively symmetric weakness of the limbs, and the weakness progresses over a period of 12 hours to 28 days before a plateau is reached.16 Patients typically have generalized hyporeflexia or areflexia. A history of upper respiratory infectious symptoms or diarrhea 3 days to 6 weeks before the onset is not uncommon.
The differential diagnosis is wide, and detailed neurologic assessment localizes the disease to the peripheral nerves rather than to the brain stem, spinal cord, cauda equina, neuromuscular junction, or muscles. The presence of distal paresthesia increases the likelihood that the correct diagnosis is the Guillain–Barré syndrome. If sensory involvement is absent, disorders such as poliomyelitis, myasthenia gravis, electrolyte disturbance, botulism, or acute myopathy should be considered. Hypokalemia shares some features with the Guillain–Barré syndrome but is commonly overlooked in the differential diagnosis. In patients with acute myopathy, tendon jerks are preserved and serum creatine kinase levels are increased. If paralysis develops abruptly and urinary retention is prominent, magnetic resonance imaging of the spine should be considered, to rule out a compressive lesion.
Nerve-conduction studies help to confirm the presence, pattern, and severity of neuropathy. These studies are essential for research, given specific criteria for categorizing the diagnosis,17 but nerve-conduction studies are not obligatory for the recently proposed Brighton criteria for diagnosis, which were developed for use in resource-poor environments.16 Once the diagnosis of an acute peripheral neuropathy is clear, the Guillain–Barré syndrome is the likely diagnosis in the majority of patients. However, clinicians should consider alternative causes, such as vasculitis, beriberi, porphyria, toxic neuropathy, Lyme disease, and diphtheria.
A lumbar puncture is usually performed in patients with suspected Guillain–Barré syndrome, primarily to rule out infectious diseases, such as Lyme disease, or malignant conditions, such as lymphoma. A common misconception holds that there should always be albuminocytologic dissociation. However, albuminocytologic dissociation is present in no more than 50% of patients with the Guillain–Barré syndrome during the first week of illness, although this percentage increases to 75% in the third week.18 Some patients with human immunodeficiency virus infection and the Guillain–Barré syndrome have pleocytosis.
The Guillain–Barré syndrome generally follows a monophasic course and typically does not recur, but two or more episodes have been reported in 7% of patients.19 The mean interval between recurrences in these patients was 7 years. Although hyporeflexia or areflexia is a hallmark of the Guillain–Barré syndrome, 10% of patients have normal or brisk reflexes during the course of the illness. Thus, the possibility of the Guillain–Barré syndrome should not be excluded in a patient with normal or brisk reflexes if all other features are supportive of the diagnosis.20 Clinical deterioration after initial improvement or stabilization with immunotherapy suggests that the treatment had a transient effect or that chronic inflammatory demyelinating polyneuropathy is present.21
Natural History and Prognostic Models
In the majority of patients, the Guillain–Barré syndrome continues to progress for up to 1 to 3 weeks after the onset of symptoms. 22 Two thirds of patients are unable to walk independently when maximum weakness is reached.2 Respiratory insufficiency occurs in 25% of patients, and major complications, including pneumonia, sepsis, pulmonary embolism, and gastrointestinal bleeding, develop in 60% of intubated patients.23 Among severely affected patients, 20% remain unable to walk 6 months after the onset of symptoms. The variations in the rate and extent of recovery in the Guillain–Barré syndrome make prognostication difficult.
One clinical scoring system that has been developed uses the patient's age, the presence or absence of antecedent diarrhea, and disease severity to predict whether a patient will be able to walk independently at 1, 3, or 6 months (see the table in the Supplementary Appendix, available with the full text of this article at NEJM.org).24,25 Another prognostic scale uses the number of days between the onset of weakness and hospital admission, the presence or absence of facial or bulbar weakness, and the severity of the limb weakness to predict the likelihood that respiratory insufficiency will develop.26 Both scales, validated in their respective patient populations, can be useful in the care of patients with the Guillain–Barré syndrome.
Demyelinating and Axonal Subtypes
The histologic features of the Guillain–Barré syndrome support a classification that includes demyelinating and axonal subtypes — acute inflammatory demyelinating polyneuropathy and acute motor axonal neuropathy.27,28 The classification is based on nerve-conduction studies (Fig. 1 in the Supplementary Appendix),29,30 and there is a notable difference in the geographic distribution of subtypes of the syndrome. In Europe and North America, the demyelinating Guillain–Barré syndrome accounts for up to 90% of cases,29 whereas in China, Japan, Bangladesh, and Mexico, the frequency of the axonal Guillain–Barré syndrome ranges from 30% to 65% and the frequency of the demyelinating Guillain–Barré syndrome ranges from 22% to 46%.22,30-32 In an Italian cohort, nerve-conduction studies overestimated the incidence of the demyelinating Guillain–Barré syndrome when the studies were performed early in the course of the disease, and subsequent nerve-conduction studies in the same patients indicated a need for reclassification: the proportion of cases that were classified as the demyelinating subtype decreased from 67% to 58%, and the proportion classified as the axonal subtype increased from 18% to 38%.33 It is generally thought that test results obtained early in the course of illness may lead to misclassification of the subtype and that serial nerve-conduction studies are therefore important for accurate subtype classification (see the Supplementary Appendix).
In acute motor–sensory axonal neuropathy, there is clear involvement of the sensory fibers, but detailed studies have suggested that mild changes occur in the sensory nerves of some patients with acute motor axonal neuropathy.34 Acute motor-conduction-block neuropathy is a mild form of acute motor axonal neuropathy but does not progress to axonal degeneration.35
There are localized forms of the Guillain–Barré syndrome that are distinguished by involvement of certain muscle groups or nerves. Facial diplegia with paresthesia is a localized form of the demyelinating Guillain–Barré syndrome,36 whereas pharyngeal–cervical–brachial weakness, which is characterized by acute weakness of the oropharyngeal, neck, and shoulder muscles, represents a localized form of the axonal Guillain–Barré syndrome.37,38
The Miller Fisher Syndrome
The Miller Fisher syndrome appears to be more common among patients with the Guillain–Barré syndrome who live in eastern Asia than among those who live in other parts of the world, occurring in up to 20% of patients in Taiwan and 25% of patients in Japan.4,39 Most patients with the Miller Fisher syndrome have evidence of infection 1 to 3 weeks before the development of ophthalmoplegia or ataxia; in one study, 20% of patients had C. jejuni infection and 8% had Haemophilus influenzae infection.40
The presence of distal paresthesia is associated with the Miller Fisher syndrome. Careful clinical assessment and focused investigations such as brain imaging and electrophysiological examinations can rule out other conditions, such as brain-stem stroke, Wernicke's encephalopathy, myasthenia gravis, and botulism. The disease peaks at a median of 1 week, and improvement often starts at a median of 2 weeks.4 Recovery from ataxia and recovery from ophthalmoplegia take a median of 1 and 3 months, respectively. By 6 months after the onset of neurologic symptoms, most patients have recovered from ataxia and ophthalmoplegia.
Postmortem Studies and Clinicopathological Correlation
The classic pathological findings in acute inflammatory demyelinating polyneuropathy are inflammatory infiltrates (consisting mainly of T cells and macrophages) and areas of segmental demyelination, often associated with signs of secondary axonal degeneration, which can be detected in the spinal roots, as well as in the large and small motor and sensory nerves.27 There is evidence of early complement activation, which is based on antibody binding to the outer surface of the Schwann cell and deposition of activated complement components; such complement activation appears to initiate the vesiculation of myelin (Figure 2Figure 2
Possible Immunopathogenesis of the Guillain–Barré Syndrome.).41 Macrophage invasion is observed within 1 week after complement-mediated myelin damage occurs.
In acute motor axonal neuropathy, IgG and activated complement bind to the axolemma of motor fibers at the nodes of Ranvier, followed by formation of the membrane-attack complex.42 The resultant nodal lengthening is followed by axonal degeneration of motor fibers with neither lymphocytic inflammation nor demyelination.28,43 There are autopsy reports indicating that the neurologic signs of the Miller Fisher syndrome overlap with those of the Guillain–Barré syndrome (ophthalmoplegia and ataxia in the former and substantial limb weakness in the latter),44 which suggests that the available immunohistochemical and electron-microscopical studies do not accurately differentiate the demyelinating subtype from the axonal subtype of the Guillain–Barré syndrome.41,42 The pathological features of the “pure” Miller Fisher syndrome remain uncertain because almost all patients eventually have a complete recovery and fatal cases are very rare.
Gangliosides, which are composed of a ceramide attached to one or more sugars (hexoses) and contain sialic acid (N-acetylneuraminic acid) linked to an oligosaccharide core, are important components of the peripheral nerves. Four gangliosides — GM1, GD1a, GT1a, and GQ1b — differ with regard to the number and position of their sialic acids, where M, D, T, and Q represent mono-, di-, tri-, and quadri-sialosyl groups (Figure 1). IgG autoantibodies to GM1 and GD1a are associated with acute motor axonal neuropathy and its more extensive and less extensive subtypes, acute motor–sensory axonal neuropathy and acute motor-conduction-block neuropathy, respectively, but not with acute inflammatory demyelinating polyneuropathy.5,34,35 Motor and sensory nerves express similar quantities of GM1 and GD1a, but their expression within various tissues may differ.45 This could explain the preferential motor-axon injury seen in acute motor axonal neuropathy.
IgG autoantibodies to GQ1b, which cross-react with GT1a, are strongly associated with the Miller Fisher syndrome, its incomplete forms (acute ophthalmoparesis and acute ataxic neuropathy), and its central nervous system variant, Bickerstaff's brain-stem encephalitis, which includes acute ophthalmoplegia, ataxia, and impaired consciousness after an infectious episode.6,7,46 Patients with pharyngeal–cervical–brachial weakness are more likely to have IgG anti-GT1a antibodies, which may cross-react with GQ1b; they are also less likely to have IgG anti-GD1a antibodies, which suggests a link to the axonal Guillain–Barré syndrome.37
The localization of these target ganglioside antigens has been associated with distinct clinical patterns of ophthalmoplegia, ataxia, and bulbar palsy. GQ1b is strongly expressed in the oculomotor, trochlear, and abducens nerves, as well as muscle spindles in the limbs.46,47 The glossopharyngeal and vagus nerves strongly express GT1a and GQ1b, possibly accounting for dysphagia.48
In some situations, antibodies against complexes consisting of two different gangliosides, rather than antibodies against a single ganglioside, develop in patients with the Guillain–Barré syndrome, suggesting specific recognition of a new conformational epitope formed by these glycolipids (see the Supplementary Appendix).49
Some evidence supports the presence of molecular mimicry between gangliosides and antecedent infectious agents in patients with the Guillain–Barré syndrome and those with the Miller Fisher syndrome. Lipooligosaccharide is a major component of the outer membrane of C. jejuni. Studies have shown that bacterial isolates from patients with the Guillain–Barré syndrome bear GM1-like or GD1a-like lipooligosaccharide, and those from patients with the Miller Fisher syndrome have lipooligosaccharides mimicking GQ1b (Fig. 2 in the Supplementary Appendix).40,50 In another study, an H. influenzae isolate from a patient with the Miller Fisher syndrome also carried a GQ1b-mimicking lipooligosaccharide.51
Experimental autoimmune neuritis, which can be induced by immunization with peripheral-nerve proteins or transferred to animals by T cells sensitized to these proteins, resembles the demyelinating Guillain–Barré syndrome clinically and pathologically.52 However, evidence in support of the concept that such autoreactive T-cell or autoantibody responses occur in a sizable proportion of patients is lacking, suggesting that experimental autoimmune neuritis is not a valid model of the Guillain–Barré syndrome.
A rabbit model of the axonal Guillain–Barré syndrome produced by sensitization with GM1 or GM1-like lipooligosaccharides of C. jejuni from patients with the disorder supports the role of molecular mimicry in the pathogenesis of this disease.53,54 In the rabbit model, anti-GM1 antibodies bound to nodes and activated complement, resulting in the formation of a membrane-attack complex at the anterior roots of the spinal nerve, which was followed by the disappearance of the sodium-channel cluster (Fig. 3 in the Supplementary Appendix). Such a constellation of abnormalities might induce nerve-conduction failure and muscle weakness. Axonal degeneration occurred at a later stage in this model.
In a murine model, the passive transfer of anti-GM1 or anti-GD1a antibodies produced a replica of the axonal Guillain–Barré syndrome in the presence of human complement, providing supportive evidence of the pathogenic roles of antiganglioside antibodies and complement in the development of the axonal Guillain–Barré syndrome.55,56 Eculizumab, a human monoclonal antibody that binds to and blocks cleavage of the complement component C5, prevented dysfunction and structural nerve damage in the murine model.57
The presence of anti-GM1 or anti-GD1a antibodies impeded axonal regeneration after peripheral-nerve injury in a murine model,45 and erythropoietin, which is used to treat anemia in chronic kidney disease, enhanced nerve regeneration.58 The activation of RhoA and Rho kinase appears to prevent neurite outgrowth induced by anti-GM1 or anti-GD1a antibodies.59 An inhibitor of Rho kinase, fasudil, which has been used in the prevention of cerebral vasospasm in patients with subarachnoid hemorrhage, has shown therapeutic effects in experimental autoimmune neuritis.60
Associated Infectious Diseases
Infection with cytomegalovirus or Epstein–Barr virus is associated with the demyelinating Guillain–Barré syndrome, whereas C. jejuni infection is associated with the axonal Guillain–Barré syndrome and with the Miller Fisher syndrome.10,40,61,62 The pathogenesis of the demyelinating Guillain–Barré syndrome has yet to be clarified, despite the documentation of characteristic histologic changes.27,41 In contrast, our understanding of the underlying pathogenesis of the axonal Guillain–Barré syndrome and the Miller Fisher syndrome has been, in part, elucidated.
Infection by C. jejuni carrying GM1-like or GD1a-like lipooligosaccharide induces anti-GM1 or anti-GD1a antibodies in some patients. These autoantibodies bind to GM1 or GD1a expressed on the motor nerves of the limbs, resulting in the axonal Guillain–Barré syndrome (Fig. 2 in the Supplementary Appendix). In contrast, infection by C. jejuni bearing GQ1b-mimicking lipooligosaccharide elicits the generation of anti-GQ1b antibodies. The autoantibodies, on binding to GQ1b expressed in the oculomotor nerves and muscle spindles, may cause the Miller Fisher syndrome.
Even in developed countries, 5% of patients with the Guillain–Barré syndrome die from medical complications such as sepsis, pulmonary emboli, or unexplained cardiac arrest, perhaps related to dysautonomia.23 Thus, management requires measures for the early detection of such complications (Table 1Table 1
Management of the Guillain–Barré Syndrome.).63 Ideally, all patients should remain under hospital observation until it has been established that there is no evidence of clinical progression.65 Whenever feasible, patients should be treated in a critical care unit, where adequate resources are available to allow continuous cardiac and respiratory monitoring. Patients with very mild weakness and the ability to walk independently are unlikely to require any treatment beyond supportive care.
Even in the absence of clinical respiratory distress, mechanical ventilation may be required in patients with at least one major criterion or two minor criteria. The major criteria are hypercarbia (partial pressure of arterial carbon dioxide, >6.4 kPa [48 mm Hg]), hypoxemia (partial pressure of arterial oxygen while the patient is breathing ambient air, <7.5 kPa [56 mm Hg]), and a vital capacity less than 15 ml per kilogram of body weight, and the minor criteria are inefficient cough, impaired swallowing, and atelectasis.66 An early assessment of swallowing will identify patients at risk for aspiration, necessitating the placement of a nasogastric tube.23 Selective decontamination of the digestive tract decreases the time that patients remain on a ventilator.67
Serious and potentially fatal autonomic dysfunction, such as arrhythmia and extreme hypertension or hypotension, occurs in 20% of patients with the Guillain–Barré syndrome.23 Severe bradycardia may be preceded by wide swings (exceeding 85 mm Hg) in systolic blood pressure from day to day. Bradycardia may be so marked that it causes asystole, warranting the use of a temporary cardiac pacemaker.
When patients with the Guillain–Barré syndrome are not ambulatory, prophylaxis against deep-vein thrombosis, consisting of subcutaneous heparin and the use of compression stockings, is important.23 Other possible complications include urinary retention and constipation, which may be addressed by bladder catheterization and the use of laxatives, respectively. The implementation of early and active individualized rehabilitation programs will maximize the chances of a favorable outcome.68
Pain, in the form of dysesthesia or muscular, radicular, arthralgic, and meningitic pain, has been reported to precede weakness in one third of patients with the Guillain–Barré syndrome: two thirds of all patients have pain during the acute phase and one third a year later.69 Early recognition and treatment are important, and opioids, gabapentin, and carbamazepine may be effective, whereas glucocorticoids are not.70 Severe fatigue has been reported in 60% of patients71 and, when persistent, may respond to a program of strengthening, aerobic, and functional exercise.72
The Guillain–Barré syndrome can affect the central nervous system. In one study, vivid dreams, hallucinations, or psychosis affected one third of patients.73 These changes occurred during the progression or plateau phase of the syndrome and disappeared as the patients recovered.
Plasma exchange was the first treatment that was found to be effective in hastening recovery in patients with the Guillain–Barré syndrome,74 and it appeared to be most effective when it was started within the first 2 weeks after disease onset in patients who were unable to walk. An electrophysiological examination is not always required for the initiation of immunotherapy. Plasma exchange nonspecifically removes antibodies and complement and appears to be associated with reduced nerve damage and faster clinical improvement, as compared with supportive therapy alone.74 The usual empirical regimen is five exchanges over a period of 2 weeks, with a total exchange of 5 plasma volumes. One trial showed that patients who could walk with or without aid but could not run benefited from two exchanges of 1.5 plasma volumes, but more severely affected patients required at least four exchanges.75
Treatment with intravenous immune globulin, initiated within 2 weeks after disease onset, is reported to be about as effective as plasma exchange in patients with the Guillain–Barré syndrome who cannot walk independently.64,76 It is thought that immune globulin may act by neutralizing pathogenic antibodies and inhibiting autoantibody-mediated complement activation, resulting in reduced nerve injury and faster clinical improvement, as compared with no treatment,77,78 although no comparative studies have been performed. In general, intravenous immune globulin has replaced plasma exchange as the treatment of choice in many medical centers because of its greater convenience and availability. According to the standard treatment regimen, immune globulin is given at a total dose of 2 g per kilogram of body weight over a period of 5 days. 64,76 The pharmacokinetics of immune globulin varies among patients, and some patients have a smaller rise in serum IgG after the administration of immune globulin.79 These patients are likely to have a poorer outcome, with fewer able to walk unaided at 6 months. A second course of immune globulin in severely unresponsive patients was reported to be beneficial in one study.80 Whether this observation holds will be assessed in an international study by the Inflammatory Neuropathy Consortium.
The combination of plasma exchange followed by a course of intravenous immune globulin is not significantly better than plasma exchange or immune globulin alone.64 Neither prednisolone nor methylprednisolone can significantly accelerate recovery or affect the long-term outcome in patients with the Guillain–Barré syndrome.81,82 One study showed that combined administration of immune globulin and methylprednisolone was not more effective than immune globulin alone, although an analysis corrected for known prognostic factors suggested a short-term effect.83
There have been no randomized trials of treatment for patients with the Miller Fisher syndrome. In retrospective analyses, intravenous immune globulin, but not plasmapheresis, resulted in a slight hastening of recovery from ophthalmoplegia or ataxia in patients with the Miller Fisher syndrome, although the time to complete recovery remained unchanged.4,84
The Guillain–Barré syndrome, an acute immune-mediated neuropathy, still carries a grave prognosis. The syndrome is manifested as a spectrum of peripheral-nerve disorders with several clinical variants that are characterized by the distribution of weakness of the limbs or cranial-nerve–innervated muscles, underlying pathological abnormalities, and associated autoantibodies.5-7,34,35,37,85
The most frequent antecedent infection is C. jejuni infection, which is associated with 30% of cases of the Guillain–Barré syndrome and 20% of cases of the Miller Fisher syndrome.9,40 Molecular mimicry between the bacterial and peripheral-nerve components appears to elicit autoantibodies and induce the development of the axonal subtype of the Guillain–Barré syndrome or the Miller Fisher syndrome after enteritis with C. jejuni.40,50,53,54 Eculizumab, erythropoietin, and fasudil, which have been used in the treatment of other, unrelated medical conditions, have shown promise in animal models of the Guillain–Barré syndrome,57,58,60 but clinical studies are lacking.
Dr. Hartung reports receiving consulting fees from Biogen Idec, Novartis Pharma, Merck Serono, Bayer HealthCare, and Genzyme and consulting and lecture fees from Baxter, CSL Behring, and Talecris. No other potential conflict of interest relevant to this article was reported.
We thank Dr. N. Shahrizaila (Department of Medicine, University of Malaya) for her editorial assistance with earlier drafts of the manuscript; Dr. N. Kokubun (Department of Neurology, Dokkyo Medical University) for preparing Fig. 1 in the Supplementary Appendix, and for helpful suggestions; Dr. K. Susuki (Department of Neuroscience, Baylor College of Medicine) for preparing Fig. 3 in the Supplementary Appendix; and Prof. A. Uncini (Department of Neuroscience and Imaging, G. d'Annunzio University) and Prof. S. Kuwabara (Department of Neurology, Graduate School of Medicine, Chiba University) for their critical review of an earlier version of the manuscript.
From the Department of Medicine, National University of Singapore, Singapore (N.Y.); and the Department of Neurology, Heinrich Heine University, Düsseldorf, Germany (H.-P.H.).
Address reprint requests to Dr. Yuki at the Department of Medicine, National University of Singapore, Unit 09-01, Centre for Translational Medicine, 14 Medical Dr., Singapore 117599, Singapore, or at email@example.com.