The characteristic symptoms of restless legs syndrome (RLS) have been known for hundreds of years and were first reported in medicine in the 1600s. Clinicians must consider potential mimics, comorbid, and associated conditions when evaluating children with RLS symptoms. The traditional differentiation of RLS from periodic limb movement disorder (PLMD) is noted in children as well as adults. Because current pediatric RLS research is sparse, this article provides the most up-to-date evidence-based as well as consensus opinion-based information on the subject of childhood RLS and PLMD. Prevalence, pathophysiology, diagnosis, treatment, and clinical associations are discussed.
The characteristic symptoms of restless legs syndrome (RLS) have been known for hundreds of years and were first reported in medicine in the 1600s. The Swedish neurologist Karl Ekbom formally described the clinical, epidemiologic, and pathophysiologic correlates of the condition in 1945. The syndrome, initially referred to as Ekbom’s disease, has 4 well-known clinical criteria: (1) an uncomfortable sensation or unexplainable urge to move the legs or other affected body part, (2) increasing symptoms with rest or inactivity, (3) a reduction of symptoms with movement, and (4) a circadian enhancement of symptoms in the evening and/or at night. Although Ekbom clearly reported all aspects of RLS occurring in children, it was not until the mid 1990s that the first case reports of children with RLS were published. Since the 1990s, much has been discovered with regards to the genetics, potential pathophysiology, and epidemiology of RLS. Most of these advances have not included specific information about children. Correlations between RLS and select pediatric populations such as attention-deficit/hyperactivity disorder (ADHD) and iron deficiency have helped generate new perspectives with regards to potential pathophysiology. The lack of pediatric-specific information about RLS in the literature has led to the use of age-adjusted adult criteria for the diagnosis of this condition in children. Caveats added to the adult diagnostic criteria are meant to increase the selectivity of these criteria for children; however, the inherent difficulties of relying on verbal descriptions in a linguistically developing population to diagnose a largely subjective disorder increase the clinical complexity of this task. An accurate diagnosis of RLS is clearly the single most important aspect of treating children with this condition. Clinicians must consider potential mimics, comorbid, and associated conditions when evaluating children with RLS symptoms. In addition, the use of more objective findings such as periodic leg movements in sleep (PLMS) and a family history of RLS increase diagnostic certainty.
The traditional differentiation of RLS from periodic limb movement disorder (PLMD) is noted in children as well as adults. The diagnostic criteria for PLMD include increased PLMS for age (>5 per hour) and a clinical sleep disturbance that is not accounted for by another sleep disorder, including RLS. Although evidence is limited to clinical case studies, data suggest that children may manifest PLMD at a younger age and develop RLS later in childhood. Observations such as these help researchers to postulate the biologic relationships between the sensory and motor components of these seemingly distinct but related disorders. In addition, clinicians can use this sort of information to assist them with cases in which diagnostic certainty is in question. Because current pediatric RLS research is sparse, this article provides the most up-to-date evidence-based as well as consensus opinion-based information on childhood RLS and PLMD. Prevalence, pathophysiology, diagnosis, treatment, and clinical associations are discussed.
Description and prevalence of RLS and PLMD in children
The adult literature suggests that RLS and PLMD is common in northern European populations and may be one of the most common inherited conditions known. Surveys indicate that between 4% and 15% of adults in the United States and Western Europe have symptoms consistent with RLS. In addition, there is evidence that up to 40% of adult RLS sufferers report symptoms starting in childhood or adolescence. Variation in prevalence figures may represent genetic heterogeneity as well as differences in survey tools used to detect RLS in large populations. In 1995, the International Restless Legs Syndrome Study Group developed standardized criteria for the diagnosis of RLS in adults. The 4 essential features (noted earlier) and 5 additional clinical features of RLS were noted in this seminal paper, including (1) sleep disturbance or daytime result of sleep disturbance, (2) involuntary movements during sleep (PLMS) and when awake, (3) neurologic examination findings consistent with secondary RLS, (4) clinical course and exacerbating factors, and (5) family history. Idiopathic or primary RLS was also distinguished from reactive or secondary RLS and its many causes including uremia, neuropathy, medications, anemia, and other causes for motor restlessness.
In 2003, a National Institutes of Health (NIH) workshop produced expert consensus criteria for the diagnosis of RLS in children and special populations. Categories of diagnostic certainty for RLS in children aged 2 to 12 years were established based on varying levels of clinical evidence ( Fig. 1 ). The use of adult criteria for adolescents (13-year-olds to 18-year-olds) was retained for the “Definite” diagnosis with the addition of “Possible” and “Probable” criteria as noted.
The adoption of diagnostic criteria in adults and children has led to less methodological variation between more recent studies. In addition, validated RLS inventories such as the International Restless Legs Syndrome Study Group rating scale, which uses a 10-question, 40-point format to detect and rate the severity of RLS symptoms, have allowed investigators to use common tools in adult studies. Aside from tools used to assess populations, prevalence rates vary based on the age and gender of the individuals included. The risk of RLS is notably increased in women compared with men, with an overall 3:2 female/male ratio. More than the age of 65 years the prevalence of RLS is reported to increase to 10% to 20%.
Recent studies that use common standard RLS criteria show a prevalence of approximately 8% to 20% in adults of varying age when all severity levels are included. Investigations that separate prevalence rates for RLS based on clinical severity report that moderate to severe RLS (episodes twice or more per week with moderate to severe distress) occurs in 2.7% to 3.9% of adults. Clinical relevance of RLS symptoms is important because it helps to establish our understanding of how many individuals may require medical attention. Using the NIH consensus criteria for the diagnosis of definite RLS in children, Picchietti and colleagues recently investigated the prevalence of RLS in a large population of US and UK children. These investigators showed that 1.9% of 8-year-olds to 11-year-olds and 2% of 12-year-olds to 17-year-olds fulfilled these criteria. The prevalence of moderately severe RLS was noted in 0.5% and 1% of 8-year-olds to 11-year-olds and 12-year-olds to 17-year-olds, respectively. Compared with adult prevalence studies, no gender preference has been noted in the pediatric population. In addition, these investigators reported a strong potential genetic component in this study, with 71% of children age 8 to 11 years and 80% of children age 12 to 17 years with RLS symptoms noted to have at least 1 parent with RLS. The development of an RLS symptom severity scale for children and adolescents was recently developed using a semistructured interview for the detection of symptoms and severity. No large-scale studies have yet used this tool or validated it in larger more varied populations.
From the mid 1990s to the present, a higher prevalence of RLS and PLMD has been noted in children with ADHD and symptoms consistent with ADHD. Although most of the available research in this area is limited by sample sizes ( n between 19 and 98 individuals), estimates for RLS in the ADHD pediatric population range between 10.5% and 44%. The estimated prevalence of ADHD in the RLS population is similarly increased between 18% and 26%. The recent US and UK population-based prevalence study by Picchietti and colleagues also shows a clear relationship between ADHD and RLS. In the US sample, 23.9% of 8-year-olds to 11-year-olds and 28.6% of 12-year-olds to 17-year-olds with definite RLS also had attention-deficit disorder (ADD)/ADHD diagnoses by self-report. The relationship between these 2 disorders is complex because sleep deprivation in children may mimic hyperactivity and/or inattention noted with ADD and ADHD. In addition, physiologic and genetic investigations suggest that these 2 conditions may share similar features such as iron deficiency and dopamine (DA) dysfunction. The potential comorbidity of these 2 conditions is clinically relevant but also provides a unique opportunity for RLS researchers to determine the common mechanisms that may help us to understand the complex pathophysiology of both RLS and ADHD.
PLMD and Periodic Leg Movements
Although PLMD is delineated as its own sleep-related movement disorder in the International Classification of Sleep Disorders (2nd edition) , Diagnostic and Coding Manual (American Academy of Sleep Medicine, 2005), many experts in the field consider PLMD as existing on a continuum with RLS. Clinically, both disorders are associated with low ferritin levels, respond to dopaminergic medications and share similar genetics. Both RLS and PLMS are noted predominately in White children versus other racial groups (odds ratio = 9.5). With the recent discovery of the dose-dependent association between the BTBD9 gene on chromosome 6p and the presence of PLMS and low serum ferritin in RLS, there remains little doubt that the motor and sensory features of RLS are related. Clinically, most RLS sufferers are noted to have PLMS on polysomnography (PSG) testing (reports vary from 80% to 100%), which suggests a common neural mechanism. However, it is clear that PLMS occurs in other disorders such as narcolepsy, Parkinson disease and Tourette syndrome as well as in situations such as pregnancy and as a result of medications (eg, elective serotonin reuptake inhibitors [SSRIs] and tricyclic antidepressants [TCAs]). Thus, although PLMS are clearly linked to RLS both clinically and genetically, they also represent a common neural pattern that can be activated in multiple ways.
Given the common and nonspecific nature of PLMS, some authorities question whether or not PLMS, and thus PLMD, should be considered abnormal. Speculation that PLMS are not necessarily pathologic is further supported by the fact that they are noted in association with other sleep disorders such as obstructive sleep apnea (OSA), as a result of OSA treatment with continuous positive airway pressure, and are noted to occur with sleep deprivation. PLMS are also well known to increase with age and may be noted in up to 30% of adults more than 50 years old.
Counter to the argument that PLMS represent normal motor activity are data that show the effect of PLMS and RLS on both health and psychological well-being. In adults, sympathetic overactivation is hypothesized to be at the heart of the association between PLMS and chronic cardiovascular conditions such as congestive heart failure and hypertension. This relationship also is believed to increase the risk for stroke and may even increase the risk for insulin resistance and type II diabetes. In addition, the effect of age on sympathetic hyperactivity suggests that younger individuals may show a more dramatic swing in autonomic response associated with PLMS, thus potentially predisposing them to even higher risks for cardiovascular and neurovascular complications. With regards to the psychological and potential developmental effect of RLS and PLMD, several adult and pediatric studies show a strong relationship with ADHD, behavioral disorders, and cognitive deficits as well as depression and anxiety. Overall, between 25% and 30% of children with RLS have symptoms consistent with ADHD and a similar number of children with ADHD fulfill the criteria for RLS. Adults with RLS are noted to carry a 4-fold to 5-fold and 13-fold increased risk for depression and panic disorder, respectively. In addition, multiple patient-reported outcome measures have shown the significant effect of RLS in adults, which is often noted as worse than the effects of hypertension, diabetes, and arthritis. Although studies showing causation are lacking in the pediatric population and there are notably few in adults, the strikingly significant overlap with medical and psychological conditions as well as reduced quality-of-life measures should not be ignored by clinicians.
A critical component to the diagnosis of PLMD aside from increased PLMS (more than 15/h in adults and 5/h in children) is a noted sleep disturbance (eg, insomnia or fragmented sleep) and/or daytime dysfunction (eg, excessive daytime sleepiness or behavioral problem). The diagnosis of PLMD in children and adolescents is made by (1) PLMS documented by PSG and exceeding a periodic limb movement index (PLMi) of 5 per hour, (2) clinical sleep disturbance, and (3) the absence of another primary sleep disorder or reason for the PLMS (including RLS). Standardized criteria commonly used to score PLMS measured via bilateral anterior tibialis electromyography were recently modified such that 4 movements must be scored in a row to qualify as periodic leg movements and each movement must be greater than 8 μV in amplitude, occur at a frequency of at least 5 seconds and no more than 90 seconds between movements, and individual movements must last at least 0.5 seconds and no longer than 10 seconds ( Fig. 2 ).
Despite the small number of investigations concerning PLMD and/or PLMS in children, there is ample evidence to support that PLMS in children, especially with RLS, are common. Studies suggest that between 8.4% and 11.9% of children may have PLMD. In addition, substantial normative data in children from studies that record PLMS via PSG and/or accelerometry show that few children or adolescents have PLMi greater than 5/h. Children with RLS are also noted to have similar rates of increased PLMS as noted in adults. One recent study of parent-child RLS pairs reported that 74% of children with definite RLS had PLMS in excess of 5/h. Thus, when PLMS are noted in children clinicians are obligated to investigate the possibility of RLS.
When PLMS is measured, the PLMi represents only the number of specific leg movements that fulfill scoring criteria per hour of sleep (eg, 5/h). Another line of investigation with regards to PLMS is the variability of intermovement intervals rather than just the overall number of PLMS. Studies of adults and children with RLS and other conditions in which PLMS are noted such as narcolepsy and ADHD show that subjects with RLS-related PLMS (and likely PLMD) have less intermovement interval variability that tends to cluster on average between 24 and 28 seconds between movements. Speculation as to the cause of this particular frequency range includes potential neural pattern generators in the spinal cord and/or diencephalon. Differentiation of RLS-related PLMS from other PLMS that fulfill the classic periodic leg movement (PLM) criteria requires further research and may not only aid in the identification of children with RLS but also help us understand the physiologic relationship between RLS and PLMS.
Another important clinical point to consider is that PLMS show marked night-to-night variability in children, as well as adults, and may preclude diagnosis on a single-night PSG.
Multiple-night testing may be necessary to quantify PLMS. Accelerometry has been found to correlate with PLMS on PSG in adults, and may be useful in the clinical evaluation of a child suspected of PLMD or RLS. Accelerometry (also referred to as actigraphy and actometry) uses a small three-dimensional electronic gravitational force detector that is strapped to the ankle during sleep. It is well tolerated by children and can be used to sample movements in a continuous fashion for a predetermined number of hours or days. When used together with a sleep log and sleep questionnaire, accelerometry data can reveal night-to-night variability, the effectiveness of therapy over time, and may also provide a more naturalistic measure of PLMS frequency and effect ( Fig. 3 ).
An intriguing area of research that further shows the PLMD-RLS continuum comes from clinical evidence that some children manifest PLMD and/or PLMS years before the symptoms of RLS develop. This retrospective longitudinal study of 18 children (mean age = 10.3 years) described the initial diagnosis of PLMD or probable/possible RLS and subsequent diagnosis of definite RLS at an average of 11.6 years later. In addition, the cohort had many of the comorbidities commonly associated with RLS such as ADHD, parasomnias, and a low serum ferritin level. Thus, despite the differentiation of RLS from PLMD on clinical grounds, it is clear that these disorders share much in terms of potential pathophysiology, comorbid conditions, and treatment. With additional research we may find that PLMD is not only on a continuum with RLS, but may represent a forme fruste of RLS.
Pathophysiology and genetics of RLS/PLMD
The mechanisms leading to the motor and sensory symptoms of RLS/PLMD are unclear. Years of clinical observation show that almost all idiopathic cases of RLS respond to dopaminergic medications such as levodopa ( l -DOPA)/carbidopa, ropinirole, and pramipexole. This finding suggests a common monoaminergic mechanism within the central nervous system. Neuroanatomic and physiologic models of diencephalic and spinal cord dopaminergic systems support an intriguing hypothesis that the sole source of DA to the spinal cord, the A11 hypothalamic cell group, may be a major contributor to the development of RLS. In addition, animal models of DA receptor knockout mice (D 2 -like receptors) suggest that loss of spinal cord gating of sensory input and motor output may also confer the symptoms of RLS/PLMD ( Fig. 4 ).
The association of iron deficiency with RLS was first noted by Nordlander in 1954. Impairment of brain iron availability is now hypothesized to play a role in the pathogenesis of RLS and PLMD based on several studies in animals and humans. Serum iron indices such as total iron, hemoglobin, and hematocrit values are usually within the normal ranges for age and gender in patients with RLS. Despite normal serum iron tests, brain iron deficiency has been implicated in human investigations using cerebrospinal fluid analysis of iron and ferritin, magnetic resonance imaging and ultrasound of the substantia nigra, and autopsy examination of brain tissue from patients with RLS. The potential mechanistic link between iron deficiency and dysfunction of central dopaminergic systems is based on evidence that iron is a cofactor for the rate-limiting enzyme, tyrosine hydroxylase, in DA synthesis. It also plays a major role in the proper functioning of postsynaptic D 2 receptors, and iron deficiency has been noted to cause downregulation of striatum and nucleus accumbens DA receptors as well as dysregulation of DA vesicular release. Correlations between peripheral serum ferritin levels and cerebrospinal ferritin levels in patients with RLS show that there is an association between the two, and that serum ferritin levels less than 50 ng/mL correlate with relative body iron storage deficiency. Recent evidence suggests that lower ferritin status in RLS may not only correlate with alterations in DA metabolism and neural transmission but may also be associated with an inability to retain intracellular ferritin. In a study of 24 women with early-onset RLS and a control group of 25 women without RLS, Earley and colleagues reported that despite equivalent serum iron indices between the 2 groups (serum ferritin, hemoglobin, total iron-binding capacity, and percent saturation) marked differences were noted in 2 proteins associated with iron trafficking into cells (soluble transferrin receptor [TfR] and divalent metal transporter 1 protein [DMT-1]) and movement out of cells (ferroportin protein). As expected, patients with RLS had higher TfR and DMT-1 levels consistent with an increased cellular need for iron. Paradoxically, they also had higher ferroportin protein levels, which would normally signify high intracellular iron levels because this protein regulates iron efflux. This finding adds another dimension to the RLS-iron story in that patients with RLS seem to have an intracellular need for iron yet the proteins responsible for regulating cellular iron content create a situation tantamount to a leaky bucket.
Genetic investigations of RLS using twin concordance, family association, familial linkage, and genomic association methods over the past 10 years have created a more complex picture. Twin studies suggest a heritability of approximately 54% for RLS. Additional genomic and linkage studies suggest that different sensorimotor phenotypes may be linked to different genetic loci.
Evidence from familial segregation analysis in Germany first showed that early-onset RLS (<30 years old) supported a single major gene model with an autosomal-dominant mode of inheritance. Late-onset RLS did not have this effect and on further analysis 2 distributions of RLS were identified based on age of onset, with 26.3 years of age as the pivotal point. Conclusions from these studies suggest that RLS is primarily genetic in younger-onset groups but also includes an environmental component for expression, which may be more relevant to the later-onset groups. In addition, multiple studies report genetic anticipation, whereby subsequent generations show earlier and earlier age of symptom onset similar to trinucleotide repeat disorders such as spinocerebellar ataxia and Huntington disease. No evidence supports a trinucleotide repeat mechanism in RLS.
Familial linkage studies have identified 5 different genetic loci thus far. The power of these types of studies is that they are able to detect rare disease alleles but also they are limited by the specific nature of the phenotype used, small sample size, genetic heterogeneity, phenocopy, penetrance, and marker allele frequencies. The first genetic locus (RLS1) was discovered in French Canadian families. It is on chromosome 12p and shows a pseudodominant pattern of inheritance, with a high allele frequency and recessive inheritance pattern. The second locus (RLS2) was found on chromosome 14q in a 3-generation Italian pedigree. Variable phenotypes or RLS with and without PLMS were included, and a common allele haplotype was identified for all positive family members. Two separate familial linkage studies did not find this association in other multifamily studies, suggesting a rare allele in this case. The third RLS linkage locus (RLS3) was described in 2 extended US family pedigrees on chromosome 9p using the phenotype of early-onset RLS. Subsequent transmission disequilibrium tests showed low significance in association with this loci and RLS in 2 other European populations. RLS4 was discovered in a South Tyrolean population on chromosome 2q. This isolated population reduced the genetic and environmental heterogeneity inherent in other multicultural studies. An autosomal-dominant mode of inheritance with a founder effect was described in 3 of 18 families. RLS5 was noted in French Canadian families on chromosome 20p in association with early age of RLS onset (<26.6 years). The autosomal-dominant pattern of inheritance in this study also pointed to a single common gene. Thus, the nature of the numerous genetic loci (12p, 14q, 9p, 2q, and 20p) identified in these studies with varying RLS phenotypes suggests that RLS/PLMD is a complex genetic trait that interacts with environmental factors.
Given the number of different and possibly rare alleles discovered in familial linkage studies or RLS, it is essential to understand if there exist more common disease-associated alleles with particular RLS phenotypes. Genomic association studies using single-nucleotide polymorphisms are able to compare multiple phenotypes in a large genetic sample. The results of 2 major genomic RLS association studies were reported in 2007. An Icelandic investigation of 306 cases and 15,664 controls reported that 3 genes (BTBD9, GLO1, and DNAH8) on chromosome 6p were associated with patients with RLS who had PLMS as a major component of their phenotype. This study showed a dose-dependent genetic association with PLMS and serum ferritin levels. Individuals with heterogenous alleles showed twice the risk for RLS with PLMS, whereas those homozygous for this variant had 4 times the risk. Serum ferritin levels were also lower in those with the addition of each BTBD9 allele. The second genomic study was performed in a German cohort of 401 familial RLS sufferers and 1644 controls. The study was replicated in 2 separate French Canadian and German populations. Four genes (MEIS1-ch. 2p, BTBD9-ch 6p, MAP2K5-ch15q, and LBXCOR1-ch 15q) were associated with RLS. In both studies, BTBD9, which is widely distributed in the brain and body, was found to be associated with RLS.
Replication of these findings using all-adult RLS-related genomic genes and RLS1–5 loci in children have not shown any clear associations. One study specifically looking at gene variants in 23 children found an 87% positive family history of RLS and a trend toward association with MEIS1 and MAP2K/LBX-COR1 variants, but not BTBD9. Another study looking at 386 children with ADHD and RLS did not find a genetic prevalence. Additional genetic research focused on children and parents with varying RLS/PLMD phenotypes may help to elucidate pathophysiologic mechanisms associated with pediatric and early-onset RLS. In particular the use of enriched samples, such as ADHD, may add much to our understanding of the underlying biology.
Pathophysiology and genetics of RLS/PLMD
The mechanisms leading to the motor and sensory symptoms of RLS/PLMD are unclear. Years of clinical observation show that almost all idiopathic cases of RLS respond to dopaminergic medications such as levodopa ( l -DOPA)/carbidopa, ropinirole, and pramipexole. This finding suggests a common monoaminergic mechanism within the central nervous system. Neuroanatomic and physiologic models of diencephalic and spinal cord dopaminergic systems support an intriguing hypothesis that the sole source of DA to the spinal cord, the A11 hypothalamic cell group, may be a major contributor to the development of RLS. In addition, animal models of DA receptor knockout mice (D 2 -like receptors) suggest that loss of spinal cord gating of sensory input and motor output may also confer the symptoms of RLS/PLMD ( Fig. 4 ).
The association of iron deficiency with RLS was first noted by Nordlander in 1954. Impairment of brain iron availability is now hypothesized to play a role in the pathogenesis of RLS and PLMD based on several studies in animals and humans. Serum iron indices such as total iron, hemoglobin, and hematocrit values are usually within the normal ranges for age and gender in patients with RLS. Despite normal serum iron tests, brain iron deficiency has been implicated in human investigations using cerebrospinal fluid analysis of iron and ferritin, magnetic resonance imaging and ultrasound of the substantia nigra, and autopsy examination of brain tissue from patients with RLS. The potential mechanistic link between iron deficiency and dysfunction of central dopaminergic systems is based on evidence that iron is a cofactor for the rate-limiting enzyme, tyrosine hydroxylase, in DA synthesis. It also plays a major role in the proper functioning of postsynaptic D 2 receptors, and iron deficiency has been noted to cause downregulation of striatum and nucleus accumbens DA receptors as well as dysregulation of DA vesicular release. Correlations between peripheral serum ferritin levels and cerebrospinal ferritin levels in patients with RLS show that there is an association between the two, and that serum ferritin levels less than 50 ng/mL correlate with relative body iron storage deficiency. Recent evidence suggests that lower ferritin status in RLS may not only correlate with alterations in DA metabolism and neural transmission but may also be associated with an inability to retain intracellular ferritin. In a study of 24 women with early-onset RLS and a control group of 25 women without RLS, Earley and colleagues reported that despite equivalent serum iron indices between the 2 groups (serum ferritin, hemoglobin, total iron-binding capacity, and percent saturation) marked differences were noted in 2 proteins associated with iron trafficking into cells (soluble transferrin receptor [TfR] and divalent metal transporter 1 protein [DMT-1]) and movement out of cells (ferroportin protein). As expected, patients with RLS had higher TfR and DMT-1 levels consistent with an increased cellular need for iron. Paradoxically, they also had higher ferroportin protein levels, which would normally signify high intracellular iron levels because this protein regulates iron efflux. This finding adds another dimension to the RLS-iron story in that patients with RLS seem to have an intracellular need for iron yet the proteins responsible for regulating cellular iron content create a situation tantamount to a leaky bucket.
Genetic investigations of RLS using twin concordance, family association, familial linkage, and genomic association methods over the past 10 years have created a more complex picture. Twin studies suggest a heritability of approximately 54% for RLS. Additional genomic and linkage studies suggest that different sensorimotor phenotypes may be linked to different genetic loci.
Evidence from familial segregation analysis in Germany first showed that early-onset RLS (<30 years old) supported a single major gene model with an autosomal-dominant mode of inheritance. Late-onset RLS did not have this effect and on further analysis 2 distributions of RLS were identified based on age of onset, with 26.3 years of age as the pivotal point. Conclusions from these studies suggest that RLS is primarily genetic in younger-onset groups but also includes an environmental component for expression, which may be more relevant to the later-onset groups. In addition, multiple studies report genetic anticipation, whereby subsequent generations show earlier and earlier age of symptom onset similar to trinucleotide repeat disorders such as spinocerebellar ataxia and Huntington disease. No evidence supports a trinucleotide repeat mechanism in RLS.
Familial linkage studies have identified 5 different genetic loci thus far. The power of these types of studies is that they are able to detect rare disease alleles but also they are limited by the specific nature of the phenotype used, small sample size, genetic heterogeneity, phenocopy, penetrance, and marker allele frequencies. The first genetic locus (RLS1) was discovered in French Canadian families. It is on chromosome 12p and shows a pseudodominant pattern of inheritance, with a high allele frequency and recessive inheritance pattern. The second locus (RLS2) was found on chromosome 14q in a 3-generation Italian pedigree. Variable phenotypes or RLS with and without PLMS were included, and a common allele haplotype was identified for all positive family members. Two separate familial linkage studies did not find this association in other multifamily studies, suggesting a rare allele in this case. The third RLS linkage locus (RLS3) was described in 2 extended US family pedigrees on chromosome 9p using the phenotype of early-onset RLS. Subsequent transmission disequilibrium tests showed low significance in association with this loci and RLS in 2 other European populations. RLS4 was discovered in a South Tyrolean population on chromosome 2q. This isolated population reduced the genetic and environmental heterogeneity inherent in other multicultural studies. An autosomal-dominant mode of inheritance with a founder effect was described in 3 of 18 families. RLS5 was noted in French Canadian families on chromosome 20p in association with early age of RLS onset (<26.6 years). The autosomal-dominant pattern of inheritance in this study also pointed to a single common gene. Thus, the nature of the numerous genetic loci (12p, 14q, 9p, 2q, and 20p) identified in these studies with varying RLS phenotypes suggests that RLS/PLMD is a complex genetic trait that interacts with environmental factors.
Given the number of different and possibly rare alleles discovered in familial linkage studies or RLS, it is essential to understand if there exist more common disease-associated alleles with particular RLS phenotypes. Genomic association studies using single-nucleotide polymorphisms are able to compare multiple phenotypes in a large genetic sample. The results of 2 major genomic RLS association studies were reported in 2007. An Icelandic investigation of 306 cases and 15,664 controls reported that 3 genes (BTBD9, GLO1, and DNAH8) on chromosome 6p were associated with patients with RLS who had PLMS as a major component of their phenotype. This study showed a dose-dependent genetic association with PLMS and serum ferritin levels. Individuals with heterogenous alleles showed twice the risk for RLS with PLMS, whereas those homozygous for this variant had 4 times the risk. Serum ferritin levels were also lower in those with the addition of each BTBD9 allele. The second genomic study was performed in a German cohort of 401 familial RLS sufferers and 1644 controls. The study was replicated in 2 separate French Canadian and German populations. Four genes (MEIS1-ch. 2p, BTBD9-ch 6p, MAP2K5-ch15q, and LBXCOR1-ch 15q) were associated with RLS. In both studies, BTBD9, which is widely distributed in the brain and body, was found to be associated with RLS.
Replication of these findings using all-adult RLS-related genomic genes and RLS1–5 loci in children have not shown any clear associations. One study specifically looking at gene variants in 23 children found an 87% positive family history of RLS and a trend toward association with MEIS1 and MAP2K/LBX-COR1 variants, but not BTBD9. Another study looking at 386 children with ADHD and RLS did not find a genetic prevalence. Additional genetic research focused on children and parents with varying RLS/PLMD phenotypes may help to elucidate pathophysiologic mechanisms associated with pediatric and early-onset RLS. In particular the use of enriched samples, such as ADHD, may add much to our understanding of the underlying biology.
Diagnosing RLS/PLMD in children
The consensus criteria for the diagnosis of RLS in children established by the NIH expert panel in 2003 include that a child should be able to state in their own words their experience of the symptoms. By raising the threshold for the RLS diagnosis in children, these criteria not only reduce potential misdiagnosis, but they also make the job of the clinician more challenging given the hurdles presented by the developmental process of verbal fluency. As previously discussed, up to 40% of adults first experience RLS symptoms in childhood and/or adolescence. In addition, PLMS and PLMD may precede the diagnosis of RLS in children by on average 11 to 12 years. Thus, clinicians must be prepared to take a careful nonleading history to elicit the salient features of RLS in children.
Recent evidence from the development of a multidimensional, self-administrated, patient-reported outcome questionnaire to assess pediatric RLS symptom severity and effect has set the stage for a more systematic approach to diagnosing RLS in childhood. The outcomes of this study show that children experience RLS symptoms within multiple domains including day and night RLS sensations, associated countermeasures (ie, rubbing or moving), and pain. In addition, the study provides a measure of effect on sleep and wake activities, as well as on emotions and tiredness. Although additional validation of this tool is needed, it is believed to provide a valid clinical measure of RLS symptoms in children 9 years of age or older, but should be used with some caution in children 6 to 8 years old. As part of this project children also communicated many nonverbal descriptions of their symptoms using a visual analogue scale and free-hand drawings of their experiences.
Although drawings can be interpreted in many ways, this format may be of particular use in younger or less fluent children to provide a starting point for elaboration and description with a clinician or parent ( Fig. 5 ).
When clinicians engage children in a discussion of RLS sensations it is important to provide a nonleading introduction to allow the child to express their experience. Given the genetic nature of RLS it is common that 1 or more adult family members may have noted similar symptoms or even have an RLS diagnosis themselves. For this reason clinicians should direct their inquiry first toward the child to help them recreate the last time they experienced something that “made it difficult to fall asleep” or “made it difficulty to lie in bed.” Often the presenting complaint from the child or parent may not seem to be related to restlessness or kicking, which may prompt further RLS/PLMD questions. Complaints are generally related to difficulty falling asleep, not wanting to go to sleep, and occasionally difficulty remaining asleep. In some cases young children may not recall any RLS symptoms at all because they may not be experiencing them at the moment of the interview. When this occurs, clinicians can direct the conversation to the routine that precedes getting into bed, what happens on a typical night when the child gets into bed, and what happens on a typical night after they fall asleep. It is helpful to have the child describe their surroundings to contextualize the sleep-onset experience because this is one of the most common times for symptoms to emerge. If a complaint is noted at other times during the day, such as sitting in class at a desk, or while doing homework after school, the clinician can use the same techniques to recreate the context in order help the child to remember how they felt and to provide their own description. Children use many imaginative ways to describe RLS symptoms such as “soda bubbles in my legs” and “ants biting my leg.” However, it is more common that children are unable to describe a sensation, and like adults they may use descriptions such as “just want to move” or “got to kick.” The terms used by children to describe the sensory experience of RLS can be further elaborated on by many forms of nonverbal communication such as drawings (see earlier discussion). It is not common for children to use the term “urge to move” in relationship to their description of RLS symptoms, but it is helpful to allow them to physically demonstrate compensatory maneuvers such as wiggling, rubbing, kicking, hitting, and even “constantly moving to find the cool spot.” After exhausting direct inquiry with a child it is sometimes helpful to incorporate the parent into the discussion to remind the child of bedtime rituals and activities that may jar the child’s recall. It is a complex medical art to direct the discussion toward understanding if the cause for a sleep disturbance is related to a sensory discomfort; however, it is more common for children and parents to complain of the secondary effects of RLS than the primary sensory discomfort required for the diagnosis.
The remaining diagnostic criteria including timing of the symptoms, exacerbating conditions, and compensatory maneuvers can be elicited by focusing on the bad-feeling symptom that the child describes. With regards to timing, it is common to note some degree of classroom difficulty as noted by the child, the child’s teacher, or parent. Difficulty sitting quietly at a desk and paying attention, as well as irritability and hyperactivity, may be notable symptoms of children with RLS. The overlap with ADHD-like symptoms is often brought up in this context and a helpful clinical pearl is to remember that the hyperactivity noted with RLS is related to an internal sensory discomfort resulting in excessive motor activity. The child with ADHD may also experience RLS, but the hyperactivity noted with ADHD is related to an external reaction to the environment resulting in variable attention. There is much overlap between these 2 clinical populations and it is important to consider the basis for excessive motor activity given that treatment strategies may be dramatically different.
Whether using the NIH consensus criteria or a semistructured interview clinicians must be aware of conditions that can mimic RLS symptoms in children. Identifiable causes of RLS symptoms are well described in the literature (often referred to as secondary RLS) and it is important to eliminate these as potential causes because treatment may vary widely. Some of the more commonly associated conditions in adults such as diabetes and pregnancy are not so common in children. Therefore, it is important to recognize other causes including joint pain associated with Osgood-Schlatter disease and arthritis, dysasthesias related to peripheral neuropathy or radiculopathy, akathesia related to antidopaminergic medications, and cutaneous pain related to dermatitis or rashes. As mentioned earlier, there is a large overlap between children with ADHD and RLS, and identification of one should prompt clinicians to ask about symptoms of the other. Another commonly associated condition with RLS is well known as growing pains. Studies of children diagnosed with growing pains unrelated to any other identifiable disorder commonly report a strong family history of RLS, and in most cases fulfill the diagnostic criteria for RLS. The prevalence of growing pains in children aged 4 to 6 years measured with validated instruments is as high as 37%. Thus, a description of typical growing pains in children with either a family history of growing pains or RLS should provide a cue for clinicians ( Box 1 ).
