Abstract
With an estimated worldwide-pooled prevalence of approximately 5% in children (Polanczyk, de Lima, Horta, Biederman, & Rohde, 2007), ADHD is one of the most common childhood neuropsychiatric conditions. According to the criteria of the Diagnostic and Statistical Manual of Mental Disorders (4th ed. text rev.; DSM-IV-TR; American Psychiatric Association [APA], 2000), ADHD is defined by a pervasive and age-inappropriate pattern of inattention, hyperactivity-impulsivity, or both. Despite an extensive worldwide literature (Wolraich, 1999), the genetic (Mick & Faraone, 2008) and environmental etiological factors (Millichap, 2008), as well as the exact pathophysiology underlying ADHD, are not well understood.
Intriguing albeit preliminary observations suggest that iron deficiency may be involved in the pathophysiology of ADHD, at least in a subset of patients (Cortese et al., 2008; Konofal, Lecendreux, Arnulf, & Mouren, 2004). Iron is an essential trace metal, which plays a central role in a multitude of biological processes, including many essential brain functions (Andrews, 1999). The iron deficiency hypothesis of ADHD is grounded on several lines of evidence. First, iron is a cofactor of enzymes necessary for the synthesis and catabolism of the monoaminergic neurotransmitters (Youdim, 2000), which are implicated in the pathophysiology of ADHD. Second, iron deficiency is associated with decreased dopamine transporter expression (Beard, Connor, & Jones, 1993); variation in the dopamine transporter gene has been linked to genetic vulnerability for ADHD (Mick & Faraone, 2008). Third, iron deficiency may lead to dysfunction in the basal ganglia (Youdim, Ben-Shachar, & Yehuda, 1989), which are believed to play a significant role in the pathophysiology of ADHD (Biederman & Faraone, 2005). Fourth, iron deficiency has been reported in children with cognitive and behavioral impairments that prominently include poor attention and hyperactivity (Lozoff et al., 2006).
Peripheral (i.e., in the body) iron status may be estimated by serum ferritin levels. To date, 14 studies assessing serum ferritin levels in children with ADHD have been published (Calarge, Farmer, DiSilvestro, & Arnold, 2010; Cortese et al., 2011; Cortese, Konofal, Bernardina, Mouren, & Lecendreux, 2009; Juneja, Jain, Singh, & Mallika, 2010; Kiddie, Weiss, Kitts, Levy-Milne, & Wasdell, 2010; Konofal et al., 2004, 2007; Menegassi et al., 2010; Millichap, Yee, & Davidson, 2006; O. Oner, Alkar, & Oner, 2008; O. Oner et al., 2010; P. Oner, Dirik, Taner, Caykoylu, & Anlar, 2007; P. Oner & Oner, 2008; Sever, Ashkenazi, Tyano, & Weizman, 1997). The main characteristics of these studies are summarized in Table 1, reporting demographic information of the participants, tools used to diagnose ADHD, prevalence of ADHD subtypes and psychiatric comorbidities, inclusion/exclusion criteria, data on pharmacological treatment, mean value of serum ferritin, and key findings for each study. We excluded from this table one case report (Konofal, Cortese, Lecendreux, Arnulf, & Mouren, 2005) and one trial (Konofal et al., 2008) that assessed only participants with selected values of serum ferritin (i.e., <30 ng/mL).
Characteristics of Studies Assessing Serum Ferritin Levels in Individuals With ADHD.
Note: ATMX = atomoxetine; K-SADS-E: Schedule for Affective Disorders and Schizophrenia for School-Aged Children, Epidemiologic; DYS: dysthimic disorder; K-SADS-PL = Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version; DSM-IV = Diagnostic and Statistical Manual of Mental Disorders–4th ed.; SF = serum ferritin; ADHD subtypes: C = combined subtype; I = predominantly inattentive; HI = predominantly hyperactive-inattentive subtype; ODD = oppositional defiant disorder; LD = learning disorders; AD = anxiety disorders; PD = phonological disorder; ED = elimination disorder; SAD = separation anxiety disorder; DD = depressive disorders; EaD = eating disorders; SUD = substance use disorders; PDD = pervasive developmental disorders; DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders–4th ed.. text rev. ; NS = not specified; CPRS = Conners’ Parent Rating Scale; CTRS =Conners’ Teacher Rating Scale; GAD = generalized anxiety disorder; P = simple phobias; OCD = obsessive compulsive disorder; TS = Tourette syndrome; MDD = major depressive disorder; MR = mental retardation; SD = speech disorder; RLS = restless legs syndrome; DSM-III-R = Diagnostic and Statistical Manual of Mental Disorders–3rd ed., revised (American Psychiatric Association, 1987).
Most of the studies included in Table 1 have been correlational because they have explored the relationship between serum ferritin levels and ADHD symptoms severity within the ADHD group, without specifically assessing and including a comparison group. Findings from these correlational studies have been mixed: Whereas seven studies (Calarge et al., 2010; Cortese et al., 2009; Konofal et al., 2004, 2007; O. Oner, et al., 2008; O. Oner et al., 2010; P. Oner & Oner, 2008) found an inverse significant relationship between serum ferritin levels and ADHD symptoms severity, three other studies (Corteseet al., 2011; Juneja et al., 2010; Menegassi et al., 2010) failed to confirm a significant relationship. Also, the only three studies that included a control group have yield mixed findings: Whereas Konofal et al. (2004) and Juneja et al. (2010) reported significantly lower serum ferritin levels in ADHD versus controls, Menegassi et al. (2010) failed to replicate this finding.
All available studies on iron deficiency in ADHD have included school-age children. Most of the studies have used semistructured interviews according to formal criteria to confirm the diagnosis of ADHD. Most of the studies have included children with all the three subtypes of ADHD (i.e., combined, predominantly inattentive, and predominantly hyperactive-impulsive types) although some articles did not specify the ADHD subtype (Kiddie et al., 2010; Konofal et al., 2004, 2007; Millichap et al., 2006; P. Oner & Oner, 2008; Sever et al., 1997) or included only children with the combined subtype (O. Oner et al., 2010). Although most of the studies included all the three ADHD subtypes, none of them reported serum ferritin levels in relation to the subtype. This likely reflects the lack of statistical power for this analysis because the reviewed studies included small or relatively small sample (mean number of participants with ADHD across studies was approximately 48). This is unfortunate because there is a growing interest to better characterize the ADHD subtypes in terms of clinical phenotype, underlying neurobiology, and response to treatment, and it is not clear to which extent the current Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994) subtypes reflect different underlying neurobiological pathways (for a critical review, see Nigg, Tannock, & Rohde, 2010). Therefore, an exploratory investigation of a possible relationship of iron status to DSM-IV ADHD subtypes is warranted. With regard to inclusion and exclusion criteria, most of available studies excluded medical/neurological conditions affecting serum ferritin levels. However, studies are heterogeneous with respect to medication status of the participants. Whereas five studies (Cortese et al., 2009; Menegassi et al., 2010; O. Oner, et al., 2008, 2010; P. Oner et al., 2007) included stimulant-naïve participants, all the other studies reported data on children with current or previous history of pharmacological treatment with stimulants or other ADHD drugs. One of the well-known adverse effects of stimulant treatment (although generally manageable in the clinical practice) is appetite reduction (Graham et al., 2011). As pointed out (D’Amato, 2005), appetite reduction may lead to decreased oral intake of iron-rich foods, with consequent decrease in serum ferritin levels. Therefore, assessing iron status in stimulant-naïve individuals may avoid a possibly relevant bias. Very few studies have explored the relationship between serum ferritin levels and factors other than ADHD symptoms severity in individuals with ADHD. Cortese et al. (2009) found a significant inverse relationship between serum ferritin levels and severity of sleep-wake transition disorders. Calarge et al. (2010) reported evidence suggesting that serum ferritin levels are directly related to response to stimulant treatment. Only one study (O. Oner, et al., 2008) assessed the relationship between serum ferritin levels and cognitive/neuropsychological measures, reporting negative results.
Given the limitations and mixed findings in previous studies, further research in larger stimulant-naïve samples including a control group is warranted. Moreover, some underexplored aspects in this field of research, such as the degree of association between serum ferritin levels and ADHD severity, ADHD subtypes, or measures of cognitive functioning, deserve investigation. The main aim of this study was to overcome limitations and unresolved issues of previous research by comparing serum ferritin levels in a larger sample of 101 stimulant-naïve children with ADHD and 93 matched controls. The secondary aim was to assess the association of serum ferritin levels to measures of ADHD symptoms severity, ADHD subtypes, and IQ.
Method
The present study was conducted at the Child Neuropsychiatry Outpatient Service of “Azienda Sanitaria Locale (ASL) Roma A,” Rome (Italy). This was a joint collaborative project with the chair of pediatrics at Second Faculty of Medicine, “Sapienza” University of Rome. Children were referred to the Child Neuropsychiatry Outpatient Service following parent complaints of restlessness and/or inattention and/or “school problems.”
Participants
ADHD participants
All 6- to 14-year-old children newly consecutively diagnosed with ADHD in the Outpatient Service of “ASL Roma A” from January 2009 to September 2010 were included in the present study. The age range was selected to include as much of the age range most representative of “classic” childhood ADHD. Exclusion criteria were as follows: (a) intellectual deficiency (Full Scale IQ < 70 on the Wechsler Intelligence Scale for Children–Third Edition [WISC-III]–Italian Version [Orsini & Picone, 2006], plus impairment in adaptive function, according to Criterion 2 of the DSM-IV-TR [APA, 2000]); (b) any neurological diseases—Criteria (a) and (b) were intended to reduce neurobiological heterogeneity which may affect iron status; and (c) any chronic conditions or diseases (e.g., anemia or celiac disorder) as well as acute inflammatory conditions that could affect peripheral iron status. All children with ADHD recruited in the study were stimulant naïve since at first diagnosis. We point out that in Italy, psychostimulant treatment is allowed only after an accurate diagnostic process confirming a formal diagnosis of ADHD. Following the diagnosis, children with ADHD are registered in a “National Surveillance Register” (see Panei et al., 2004), and then the treatment can be started. Therefore, it is infrequent to find children treated with psychostimulants before a formal diagnosis of ADHD.
Controls
A control group matched for age and gender was randomly recruited among the children seen by family pediatricians in routine care in the same local area. The family pediatricians collaborating with the authors were asked to refer children suitable as control participants for the study. Therefore, the pediatricians were asked to refer only healthy children, that is, children without chronic or acute medical conditions as well as without known mental disorders according to their medical records and a detailed clinical interview.
The study was conducted in accordance with the Declaration of Helsinki (International Committee of Medical Journal Editors, 1989). Written informed consent was obtained from the parents of all participants, and written assent was obtained from all children.
Procedures
Psychiatric and psychometric evaluation
Psychiatric diagnoses were established according to DSM-IV-TR criteria (APA, 2000) and confirmed by means of the semistructured interview Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version (K-SADS-PL; Kaufman et al., 1997) conducted by an experienced child psychiatrist (R. D.). The parents and schoolteachers of all the children in the ADHD group were also asked to fill out the ADHD–Rating Scale (ADHD-RS) adapted for the Italian population (Marzocchi & Cornoldi, 2000). The WISC-III–Italian Version (Orsini & Picone, 2006), was used to estimate the IQ. For logistic reasons, it was not possible to perform the assessment with the K-SADS-PL, the ADHD-RS, and the WISC-III in the control participants. However, before being referred to the study, control participants were carefully interviewed by the local pediatricians who were specifically trained in the diagnosis of ADHD according to the DSM-IV-TR. Therefore, ADHD in the control participants was ruled out on the basis of clinical interview.
Medical assessment
Medical history, neurological and physical examinations, and electroencephalogram during sleep were performed in all ADHD participants to exclude comorbid medical and neurological conditions. Blood samples were collected by a registered nurse in the morning from each participant to obtain serum ferritin (the main parameter of this study), as well as other measures, including serum iron, a complete blood count, globular volume, and hemoglobin values. Serum ferritin levels were determined by means of the enzyme-linked immunosorbent assay (ELISA) method (http://www.bio.davidson.edu/courses/genomics/method/ELISA.html).
Statistical Analysis
Demographic and clinical data are presented as means and standard deviation and categorical data. The ADHD-RS scores are presented as raw scores (minimum total score = 0, maximum total score = 54, minimum total score for ADHD = 12, minimum score in each of the two subscales [predominantly inattentive, predominantly hyperactive-impulsive] = 0, maximum score in each of the two subscales = 27, minimum score in each subscale for ADHD = 12). Two-tailed t tests for independent samples were used to compare age, serum ferritin levels, red cell number, and globular volume value in children with ADHD versus controls. One-way ANOVA was used to compare age, serum ferritin levels, red cell number, and globular volume value among children with ADHD inattentive subtype, ADHD hyperactive-impulsive subtype, combined subtype, and control group. A chi-square test and odds ratio (OR) with relative confidence intervals (CI) were used for categorical variables. A correlation analysis using Pearson’s correlation was performed between serum ferritin values and behavioral data (i.e., ADHD symptoms severity, assessed with the ADHD–Rating Scale–Total score [ADHD-RS-TOT], ADHD–Rating Scale–Hyperactive Impulsive score [ADHD-RS-HI], and ADHD–Rating Scale–Inattentive score [ADHD-RS-I]) as well as cognitive scores (IQ: Full Scale IQ [FSIQ], Verbal IQ [VIQ], Performance IQ [PIQ]). A probability level of p < .05 was used to indicate statistical significance. Statistical analyses were performed using SPSS v15.0 (SPSS, Inc., Chicago, IL, USA).
Results
In all, 113 children were referred for ADHD symptoms evaluation, and 96 controls were assessed for inclusion criteria. A total of 7 children with ADHD-like symptoms presented with Full Scale IQ < 70 and were therefore excluded; 5 children in the ADHD group and 3 in the control group were excluded from statistical analysis due to nonavailability of ADHD-RS or refusal to give blood sample. No other children were excluded because of exclusion criteria. Therefore, analyses were conducted on a sample of 101 children with ADHD (males: n = 92) and 93 controls (males: n = 82).
Demographic and biochemical data are reported in Tables 2 and 3. Psychiatric comorbid disorders in the ADHD group were as follows: 42 ADHD children presented with oppositional defiant disorder (ODD), 16 with generalized anxiety, 4 with major depressive disorder, and 1 with dysthymic disorder. None of the participants were treated with antiepileptic or oral antipsychotic. Two-tailed t tests showed that age, serum ferritin, and globular volume did not significantly differ between children with ADHD and controls. Red cells number was significantly higher in children with ADHD than in controls, although marginally (p = .05). According to the cutoff used by Konofal et al. (2004) to indicate iron deficiency in children (30 ng/mL), patients with ADHD were more likely to present the lowest ferritin values, although not significantly (OR = 0.78, CI 95% = [0.44, 1.37]; p = .392). One-way ANOVA showed that age, serum ferritin, red cells, and globular volume did not significantly differ among children with ADHD inattentive subtype, ADHD hyperactive-impulsive subtype, combined subtype, and controls (Table 3).
Demographic and Biochemical Data of Participants With ADHD and Controls.
Chi-square or t test for independent samples.
Demographic and Biochemical Data of ADHD Groups.
Chi-square or ANOVA analysis.
Table 4 reports descriptive data regarding FSIQ, VIQ, PIQ, ADHD-RS-TOT, ADHD-RS-HI, and ADHD-RS-I. All correlations between serum ferritin levels and the aforementioned cognitive/behavioral data were not significant, either collapsing all the participants in one group or considering children with ADHD and controls separately (data not shown).
Descriptive Cognitive and Behavioral Data of ADHD Group.
Note: FSIQ = Full Scale IQ; VIQ = Verbal IQ; PIQ = Performance IQ; ADHD-RS-TOT = ADHD–Rating Scale–Total score; ADHD-RS-HI = ADHD–Rating Scale–Hyperactive-Impulsive score; ADHD-RS-I = ADHD–Rating Scale–Inattentive score.
Discussion
To our knowledge, this is the largest study that assessed serum ferritin levels in stimulant-naïve children with ADHD and controls matched for gender and age. We found a nonsignificant difference in serum ferritin levels between children with ADHD and controls. Moreover, the associations between serum ferritin levels and ADHD symptoms severity, ADHD subtypes, or IQ were not significant.
Considering the other available controlled studies on iron status in ADHD, our main finding (i.e., non significant difference in serum ferritin levels between children with ADHD and comparisons) is in line with the report by Menegassi et al. (2010), but in contrast with the results by Konofal et al. (2004) and Juneja et al. (2010). Among others, two factors may explain these discordant results. First, it has been pointed out that psychostimulants or other psychotropic medications may affect appetite and, consequently, alter serum ferritin levels (D’Amato, 2005). Juneja et al. (2010), who found significantly lower serum ferritin levels in children with ADHD compared with controls, did not specify the medication status of their participants previously to the study inclusion, and, therefore, their study is not informative in this respect. Therefore, it is possible that the significant differences reported were accounted for by stimulant treatment not mentioned in the text. Konofal et al. (2004), who also found significantly lower serum ferritin levels in children with ADHD compared with controls, used a 2-month wash-out period from previous psychostimulant treatment. However, as reported by Menegassi et al. (2010), it is not clear whether this prior period of medication suspension would be sufficient to normalize possible nutritional deficiencies due to previous treatment. Whereas, Menegassi et al. reported no significant difference in serum ferritin levels between stimulant-naïve children with ADHD and controls. Indeed, they also found no significant differences between controls and children with ADHD treated with methylphenidate. However, serum ferritin levels were lower (although not significantly) in children with ADHD treated with stimulants. Because the treatment lasted only for 3 months, we cannot exclude, on a theoretical ground, that a longer period of treatment would have led to significantly lower serum ferritin values in children with ADHD treated with stimulants versus controls, although this remains speculative because there is no empirical evidence indicating to which extent and how methylphenidate affects serum ferritin levels. Second, different techniques to measure serum ferritin levels between our study (ELISA) and the study by Konofal et al. (2004; Elecsys Enzymun-Test) were used. We are not aware of a head-to-head comparison of these two techniques, so we cannot exclude different specificity and sensitivity of these two techniques, leading to possible different results in serum ferritin values. Third, our study was conducted in a different country (Italy) compared with those by Konofal et al. (2004) and Juneja et al. (2010; France and India, respectively). However, we do not think that these differences may account for the discrepancy in the results among our study and those by Konofal et al. (2004) and Juneja et al. (2010) because it is unlikely that differences in techniques or geographic location affected selectively on ADHD or controls.
We point out that the mixed findings on serum ferritin levels reported in available controlled studies (including the present one) of children with ADHD should not suggest that iron deficiency is not involved in the pathophysiology of ADHD. Indeed, serum ferritin is a marker of peripheral (i.e., not in the brain) iron status. The extent to which serum ferritin correlates with brain iron levels remains unclear (Cortese et al., 2011). In the study by Cortese et al. (2011), there was a trend, which, however, did not reach statistical significance, for correlation of serum ferritin levels with brain iron levels in most of the brain regions assessed. Previous studies of other disorders have yielded mixed findings. Whereas one study (Argyropoulou et al., 2000) found a significant relationship in beta-thalassemia major, and another (Christoforidis et al., 2007) reported only a moderate correlation (r = .56) in the same disease, other authors (Godau, Klose, Di, Schweitzer, & Berg, 2008) failed to find a significant correlation in restless legs syndrome (RLS). Therefore, there is no solid evidence to state that ferritin is a highly reliable marker of brain iron, although it may roughly estimate it. Because brain iron is what is expected to affect neuronal functions and myelination of white matter (Lozoff, 2011; Georgieff, 2008) expected to underpin ADHD symptoms, we suggest that, besides an assessment of peripheral iron markers, an estimation of brain iron levels is crucial to establish a possible role of iron deficiency in the pathophysiology of ADHD. A reduced amount of peripheral iron may have an impact on central levels of iron. However, normal peripheral ferritin levels do not always necessarily reflect normal brain iron. For instance, a dysfunction in the blood-brain barrier would lead to decreased entry of iron in the brain, and, therefore, to reduced brain iron levels in the presence of normal peripheral ferritin values. To this regard, Cortese et al. (2011) recently published a pilot study assessing brain iron levels in ADHD. They found significant lower levels of estimated brain iron in the thalamus bilaterally in ADHD children versus controls. Although serum ferritin levels were correlated with estimated brain iron, this association failed to reach statistical significance in most of the brain regions assessed. Recently, a deficit in the entry of iron in the brain has been reported in patients with RLS (Connor et al., 2011). It has been observed that patients with RLS appear to have marginal central nervous system iron levels (Allen & Earley, 2007). These levels can become insufficient for an appropriate brain functioning even with normal peripheral iron (Allen & Earley, 2007). Interestingly, it has been reported that RLS may be comorbid with ADHD (Cortese et al., 2005), thus suggesting that these two disorders might share common underlying pathophysiological mechanisms. Accordingly, we speculate that dysfunction in the blood-brain barrier or iron transport mechanisms also in children with ADHD (with or without RLS) may account for a possible mismatch between peripheral and central iron. Unfortunately, given the unavailability of a formal Italian translation of the official RLS research criteria for RLS in children, we could not collect methodologically sound data about the prevalence of RLS in our sample.
With regard to the second aim of our study, the lack of significant association between serum ferritin levels and ADHD symptoms severity is in contrast with the results by seven studies (Calarge et al., 2010; Cortese et al., 2009; Konofal et al., 2004, 2007; O. Oner, et al., 2008, 2010; P. Oner & Oner, 2008) but in agreement with three other studies (Cortese et al., 2011; Juneja et al., 2010; Menegassi et al., 2010). As it can be the case with mixed results from small or relatively small samples, we cannot exclude that positive results were accounted for in a Type I error. Moreover, the lack of significant correlation between serum ferritin levels and measures of cognitive function confirms, in a larger sample, the results by O. Oner, et al. (2008). The same considerations on central versus peripheral iron status apply also to the results related to our second aims.
Our results should be considered in the light of some limitations. First, we did not perform a formal evaluation of the nutritional status and eating patterns of the participants. We note that, unfortunately, none of the available studies has reported a specific assessment of eating patterns. As previously stated, D’Amato (2005) pointed out that possible differences in iron status between those with ADHD and controls may be accounted for by appetite abnormalities. These might be related not only to stimulant treatment but also to other factors independent of stimulant treatment, such as, for example, putative alterations of eating patterns in ADHD (Cortese, Bernardina, & Mouren, 2007), which might, on a theoretical ground, alter iron intake. However, although the evaluation of eating patterns might have been of relevance on the case of significant differences between ADHD children and controls, to shed light on possible reasons underlying the difference, it seems less necessary given our results showing no significant differences. Second, the K-SADS-PL was not performed in the control participants. However, ADHD in the controls was ruled out by means of a clinical interview by the local pediatricians specifically trained in recognizing the symptoms of ADHD according to the DSM-IV criteria. Third, only White participants were included. Further studies should assess the impact of race on the relationship between ADHD and iron status because it has been reported that mean serum ferritin levels vary according to race in children (Brotanek, Gosz, Weitzman, & Flores, 2007) and adults (Zacharski, Ornstein, Woloshin, & Schwartz, 2000). Notwithstanding these limitations, we think that our study adds to the previous literature on iron status and ADHD addressing underexplored issues on the relationship between serum ferritin levels and ADHD. Our study suggests no relationship between serum ferritin levels and ADHD. However, we think that the key result of this article should not be a reason to dismiss research on the relationship between ADHD and iron status. Considering the preliminary results by Cortese et al. (2011) along with our study, the largest in the field based on serum ferritin and free of possible bias, we conclude that if we rely only on serum ferritin, we might miss the opportunity to find a true relationship between low iron status and ADHD. Therefore, we advocate a high level and methodologically sound modern investigation of the relationship between ADHD and iron status.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Renato Donfrancesco is consultant and coinvestigator in studies sponsored by Shire Pharmaceuticals and consultant for Novartis. He has received financial support to attend medical meetings from Shire Pharmaceuticals. Samuele Cortese has received financial support to attend medical meetings from Eli Lilly & Company (2007-2009) and Shire Pharmaceuticals (2009-2010), and has been coinvestigator in studies sponsored by GlaxoSmithKline (2006), Eli Lilly & Company (2007-2008), and Genopharm (2008). He has served as a consultant for Shire Pharmaceuticals (2009-2010).