Iron deficiency and internalizing disorders in youth

Internalizing disorders in adolescents are common and impairing, with recent data suggesting increasing incidence (Twenge, Cooper, Joiner, Duffy, & Binau, Reference Twenge, Cooper, Joiner, Duffy and Binau2019). Many factors are likely implicated, given the heterogeneous nature of depressive and anxiety disorders. That iron deficiency would contribute to the recent change in the prevalence of internalizing disorders in adolescents is plausible for several reasons: (1) iron plays a critical role in the brain, potentially impacting the structure and function of mood-relevant areas, (2) iron deficiency is common in adolescence (CDC, 2002, 2014; Gupta et al., Reference Gupta, Hamner, Suchdev, Flores-Ayala and Mei2017; Looker et al., Reference Looker, Cogswell and Gunter2002), and (3) available evidence suggests that replenishing iron stores may improve internalizing symptoms (Mikami et al., Reference Mikami, Okazawa, Kimoto, Akama, Onishi, Takahashi and Matsumoto2019).

To our knowledge, however, the prevalence of iron deficiency in adolescents with internalizing disorders has not been examined. We found that 32% of our participants had sF < 15 ng/mL. This high prevalence of iron deficiency, which may be partially accounted for by the over-representation of Hispanic females in our study (Gupta et al., Reference Gupta, Hamner, Suchdev, Flores-Ayala and Mei2017; Looker et al., Reference Looker, Cogswell and Gunter2002; WHO, 2001), is particularly troubling given that our participants had undergone a thorough screening to rule out many general medical conditions that could cause iron deficiency. Moreover, the pre/perinatal history and the medical record review exclude the possibility that our findings represent chronic sequelae of early-life or concurrent anemia (online Supplementary Data). As such, to the best of our knowledge, our participants were healthy, had received good prenatal care, had not had perinatal anemia, but did have low iron intake despite coming from households with a median income nearly double the national average (Table 1). This low iron intake is consistent with recent data showing a trend for reduction in iron intake in the US population, with an associated increase in the prevalence of iron deficiency anemia. The large effect sizes we found for the association between iron deficiency and internalizing symptoms highlight the potential for replenishing iron stores to reverse the recently documented increase in the prevalence of depressive and anxiety disorders in adolescents (Mojtabai & Olfson, Reference Mojtabai and Olfson2020; Mojtabai, Olfson, & Han, Reference Mojtabai, Olfson and Han2016)

Brain iron homeostasis during iron deficiency

Understanding the interplay between iron deficiency and brain structure and function across the lifespan requires one to consider three inter-related factors: (1) iron transport into the brain, (2) hierarchy of iron distribution to bodily systems, and (3) how the presence of iron deficiency has been defined. Once the BBB matures during infancy, iron transport into the brain becomes tightly regulated, protecting it from daily fluctuations in systemic iron levels (Wade et al., Reference Wade, Chiou and Connor2019). Preclinical and clinical studies have established that iron is prioritized for erythropoiesis, for evident survival advantage. As the body is faced with insufficient iron intake to meet its needs, iron reserves are tapped in a relatively hierarchical order (e.g. liver, skeletal muscle, heart, etc.) and, with increasing iron deficiency severity, iron is eventually diverted even from the brain to the bone marrow (Ennis, Dahl, Rao, & Georgieff, Reference Ennis, Dahl, Rao and Georgieff2018; Guiang, Georgieff, Lambert, Schmidt, & Widness, Reference Guiang, Georgieff, Lambert, Schmidt and Widness1997; Zamora, Guiang, Widness, & Georgieff, Reference Zamora, Guiang, Widness and Georgieff2016). Finally, historically, iron deficiency has been the focus of public health interventions due to its association with anemia. Current guidelines to diagnose ID have been based on anemia-relevant markers, such as bone marrow iron content and sF below which anemia develops (WHO, 2011). They are not based on the assessments of brain iron content or the emergence of neuropsychiatric manifestations. Considering these three factors together, it thus follows that, compared to infants who may not benefit from the protective effect of the BBB yet, brain iron content in adolescents will only be impacted once a threshold (in terms of severity of body iron stores depletion) has been crossed. However, what severity of iron deficiency is needed to overcome the homeostatic mechanisms of the BBB, resulting in brain iron depletion, is unknown. What appears certain is that brain iron is drawn upon before anemia develops. This is supported by studies in adolescents and young adults with iron deficiency but without anemia finding improvement in neurocognitive and emotional functioning following iron supplementation (Ballin et al., Reference Ballin, Berar, Rubinstein, Kleter, Hershkovitz and Meytes1992; Beard et al., Reference Beard, Hendricks, Perez, Murray-Kolb, Berg, Vernon-Feagans and Tomlinson2005; Corwin et al., Reference Corwin, Murray-Kolb and Beard2003; Fordy & Benton, Reference Fordy and Benton1994; Karl et al., Reference Karl, Lieberman, Cable, Williams, Young and McClung2010; Low et al., Reference Low, Speedy, Styles, De-Regil and Pasricha2016; Rangan et al., Reference Rangan, Blight and Binns1998; Vahdat Shariatpanaahi et al., Reference Vahdat Shariatpanaahi, Vahdat Shariatpanaahi, Moshtaaghi, Shahbaazi and Abadi2007; Verdon et al., Reference Verdon, Burnand, Stubi, Bonard, Graff, Michaud and Favrat2003). This appears consistent with our findings both concerning internalizing symptom severity and subcortical structures volumes.

Notably, we found a larger association between internalizing symptom severity and iron deficiency (a categorical variable) compared to sF (a continuous variable). This is consistent with the BBB's role in protecting the brain from fluctuations in body iron until the homeostatic mechanisms are overwhelmed (Wade et al., Reference Wade, Chiou and Connor2019). When iron deficiency is defined more liberally, as sF < 20 ng/mL, the association with symptom severity remains significant, albeit of a smaller magnitude, suggesting that the cutoff for sF of <15 ng/mL to define iron deficiency may be overly conservative, having been established based on hematological outcomes without regard to the function of other organs, like the brain. In fact, for those participants with a hemoglobin level obtained within one year before study entry, anemia was quite uncommon with only one person affected (online Supplementary Data). In patients with restless leg syndrome, iron supplementation aims to raise sF > 50 ng/mL. However, the association between peripheral and brain iron content has not been well examined in general and not at all in adolescents. Identifying the threshold below which iron deficiency starts drawing on brain iron is an area that requires urgent attention, with potentially widespread clinical implications.

Iron deficiency and brain structure and function

Brain iron is also critical for myelin formation (Bastian et al., Reference Bastian, von Hohenberg, Georgieff and Lanier2019; Beard & Connor, Reference Beard and Connor2003; Lange & Que, Reference Lange and Que1998). Oligodendrocytes are enriched with iron-requiring enzymes involved in lipid metabolism, needed for initial myelin deposition as well as for maintaining its integrity (Bourre et al., Reference Bourre, Pascal, Durand, Masson, Dumont and Piciotti1984; Cammer, Reference Cammer1984a, Reference Cammer and Norton1984b; Connor & Menzies, Reference Connor and Menzies1996; Tansey & Cammer, Reference Tansey and Cammer1988). Two brain imaging studies have explored the association between brain iron content and brain function in children. Peterson et al. used data collected in healthy 12- to 21-year-olds, enrolled in the National Consortium on Alcohol and Neurodevelopment in Adolescence study (Peterson et al., Reference Peterson, Kwon, Luna, Larsen, Prouty, De Bellis and Pfefferbaum2019). They repurposed scans obtained for functional MRI and diffusion tensor imaging (DTI) to track brain iron distribution, replicating the fact that several subcortical nuclei are iron-rich, and that iron accumulates with increasing age but plateaus by early adulthood. Additionally, working memory speed was inversely associated with iron signal in the left dentate nucleus and substantia nigra (Peterson et al., Reference Peterson, Kwon, Luna, Larsen, Prouty, De Bellis and Pfefferbaum2019). Similarly, Carpenter et al. used brain imaging to estimate brain iron content in 39 healthy children (56% female, mean age 9.5 ± 1.3 years), finding again a positive association between age and iron content in the basal ganglia (Carpenter et al., Reference Carpenter, Li, Wei, Wu, Xiao, Liu and Egger2016). Moreover, the right caudate iron content was positively associated with spatial IQ (Carpenter et al., Reference Carpenter, Li, Wei, Wu, Xiao, Liu and Egger2016).

Using structural brain imaging, we found an inverse association between body iron status and the volumes of the putamen and the left caudate. The effect sizes were smaller for the right hemisphere structures, perhaps reflecting the lateralization in brain iron content observed in some studies (Langkammer et al., Reference Langkammer, Schweser, Krebs, Deistung, Goessler, Scheurer and Reichenbach2012; Xu, Wang, & Zhang, Reference Xu, Wang and Zhang2008). This inverse association may reflect the fact that brain iron moderates the decrease in subcortical structures volume observed during late childhood and adolescence (Raznahan et al., Reference Raznahan, Shaw, Lerch, Clasen, Greenstein, Berman and Giedd2014; Wierenga et al., Reference Wierenga, Langen, Ambrosino, van Dijk, Oranje and Durston2014). It may also reflect iron deficiency-induced impairment in dopaminergic signaling (Beard et al., Reference Beard, Chen, Connor and Jones1994; Burhans et al., Reference Burhans, Dailey, Beard, Wiesinger, Murray-Kolb, Jones and Beard2005; Erikson et al., Reference Erikson, Jones and Beard2000, Reference Erikson, Jones, Hess, Zhang and Beard2001; Jellen et al., Reference Jellen, Lu, Wang, Unger, Earley, Allen and Jones2013; Nelson et al., Reference Nelson, Erikson, Pinero and Beard1997; Pino et al., Reference Pino, da Luz, Antunes, Giampa, Martins and Lee2017). Preclinical studies have linked stimulant-induced reduction in dopamine D₂ receptor density in the ventral striatum with an increase in putamen volume (Chang et al., Reference Chang, Cloak, Patterson, Grob, Miller and Ernst2005; Churchwell, Carey, Ferrett, Stein, & Yurgelun-Todd, Reference Churchwell, Carey, Ferrett, Stein and Yurgelun-Todd2012; Groman, Morales, Lee, London, & Jentsch, Reference Groman, Morales, Lee, London and Jentsch2013; Jan, Lin, Miles, Kydd, & Russell, Reference Jan, Lin, Miles, Kydd and Russell2012; Jernigan et al., Reference Jernigan, Gamst, Archibald, Fennema-Notestine, Mindt, Marcotte and Grant2005). Similarly, clinical studies have found enlarged putamen in patients with stimulant use disorders (Ersche et al., Reference Ersche, Acosta-Cabronero, Jones, Ziauddeen, van Swelm, Laarakkers and Williams2017, Reference Ersche, Barnes, Jones, Morein-Zamir, Robbins and Bullmore2011, Reference Ersche, Jones, Williams, Turton, Robbins and Bullmore2012; Jacobsen, Giedd, Gottschalk, Kosten, & Krystal, Reference Jacobsen, Giedd, Gottschalk, Kosten and Krystal2001). Finally, our finding is also consistent with increased basal ganglia volumes following extended treatment with ‘typical’ antipsychotics, characterized by potent dopamine D₂ antagonist activity (Corson, Nopoulos, Miller, Arndt, & Andreasen, Reference Corson, Nopoulos, Miller, Arndt and Andreasen1999; Navari & Dazzan, Reference Navari and Dazzan2009). The basal ganglia are a key node in the cortico-striato-thalamo-cortical loops, subserving a multitude of neuropsychological processes implicated in psychopathologies, such as inhibitory control and reward processing (Drysdale et al., Reference Drysdale, Grosenick, Downar, Dunlop, Mansouri, Meng and Liston2017; Janiri et al., Reference Janiri, Moser, Doucet, Luber, Rasgon, Lee and Frangou2019; Pizzagalli et al., Reference Pizzagalli, Holmes, Dillon, Goetz, Birk, Bogdan and Fava2009; Wei & Wang, Reference Wei and Wang2016; Williams, Reference Williams2016). Whether iron supplementation would reverse these structural changes requires future studies.

Timing of iron deficiency and persistence of neuropsychiatric sequelae

Given the key role iron plays in a broad set of metabolic processes, it is not surprising that iron deficiency would be associated with cognitive and neuropsychiatric deficits (Vulser et al., Reference Vulser, Wiernik, Hoertel, Thomas, Pannier, Czernichow and Lemogne2016). However, the nature, severity, and chronicity of these effects are closely tied to the time during development when iron deficiency ensues (Barks, Hall, Tran, & Georgieff, Reference Barks, Hall, Tran and Georgieff2019; Georgieff, Reference Georgieff2017). The earlier the exposure, the broader the impact on neurocognitive functioning given the brain's substantial iron-dependent metabolic needs early in life, when brain structures underlying basic neurocognitive processes are rapidly developing (Barks et al., Reference Barks, Hall, Tran and Georgieff2019; Georgieff, Reference Georgieff2017; Georgieff, Brunette, & Tran, Reference Georgieff, Brunette and Tran2015; Lozoff & Georgieff, Reference Lozoff and Georgieff2006). This is thought to involve irreversible impairment in gene expression, affecting neuronal growth and plasticity (Georgieff et al., Reference Georgieff, Brunette and Tran2015). These ‘sensitive periods’ have been shown both in animal and clinical studies (Barks et al., Reference Barks, Hall, Tran and Georgieff2019; Georgieff, Reference Georgieff2017; Mudd et al., Reference Mudd, Fil, Knight, Lam, Liang and Dilger2018). For instance, children exposed to iron deficiency in-utero or in infancy, show motor, cognitive, emotional, and social deficits long after iron stores had been replenished (Barks et al., Reference Barks, Hall, Tran and Georgieff2019; Doom et al., Reference Doom, Richards, Caballero, Delva, Gahagan and Lozoff2018; Lozoff, Jimenez, Hagen, Mollen, & Wolf, Reference Lozoff, Jimenez, Hagen, Mollen and Wolf2000; Lozoff et al., Reference Lozoff, Smith, Kaciroti, Clark, Guevara and Jimenez2013). Additionally, in children with ADHD, Turner et al. found that a small mean corpuscular volume (a marker of iron deficiency) in the toddler years predicted poor response to psychostimulant treatment in elementary school (Turner, Xie, Zimmerman, & Calarge, Reference Turner, Xie, Zimmerman and Calarge2010). However, in contrast to what appears as persistent sequelae when iron deficiency occurs early in life, cognitive deficits in children and adults with iron deficiency, with or without anemia, can improve with iron repletion (Chmielewska et al., Reference Chmielewska, Dziechciarz, Gieruszczak-Bialek, Horvath, Piescik-Lech, Ruszczynski and Szajewska2019; Grantham-McGregor & Ani, Reference Grantham-McGregor and Ani2001; Low et al., Reference Low, Speedy, Styles, De-Regil and Pasricha2016; McCann & Ames, Reference McCann and Ames2007). For example, iron supplementation in 5- to 8-year-old children with ADHD and women with postpartum depression reduces symptom severity, particularly in those with iron deficiency (Konofal et al., Reference Konofal, Lecendreux, Deron, Marchand, Cortese, Zaim and Arnulf2008; Sever, Ashkenazi, Tyano, & Weizman, Reference Sever, Ashkenazi, Tyano and Weizman1997; Wassef, Nguyen, & St-Andre, Reference Wassef, Nguyen and St-Andre2019). In other words, which sequelae arise closely depend on which brain areas are most metabolically actively and/or rapidly developing at the time of iron deficiency, making them particularly vulnerable to its impact (Lozoff & Georgieff, Reference Lozoff and Georgieff2006; Vulser et al., Reference Vulser, Wiernik, Hoertel, Thomas, Pannier, Czernichow and Lemogne2016).

Some limitations of our pilot study must be acknowledged. First, our findings are based on a relatively small sample size, requiring replication in a larger study. Second, given that internalizing disorders disproportionately affect adolescent females, we did not recruit males. As such, whether our findings extend to males remains to be seen. Third, because this was a cross-sectional evaluation, the direction of the causal association between iron deficiency and our outcomes of interest cannot be established. While low sF is the most specific non-invasive marker of iron deficiency, with excellent reproducibility (Belza, Ersboll, Henriksen, Thilsted, & Tetens, Reference Belza, Ersboll, Henriksen, Thilsted and Tetens2005), neither the duration of iron deficiency was available, nor the presence of anemia ruled out as we did not measure hemoglobin. Because some symptoms associated with anemia may overlap with internalizing symptoms (Murray-Kolb, Reference Murray-Kolb2011), future studies should exclude the confounding effect of anemia to best examine the independent effect of iron deficiency on the brain. Given the exploratory nature of this study, no correction for multiple comparisons was made. As shown in Table 1, 62% of the participants were estimated to be in the luteal phase of their menstrual cycle, a time when ferritin tends to be higher (Kim, Yetley, & Calvo, Reference Kim, Yetley and Calvo1993). As such, the rate of iron deficiency may have been even higher had we enrolled the participants upon the onset of their menstrual phase. Finally, measuring C-reactive protein would have ruled out cases of inflammation, where ferritin would have been elevated. However, the stringent inclusion/exclusion criteria and the high prevalence of iron deficiency suggest that our participants did not have acute inflammation.

In summary, given iron's role in multiple metabolic processes affecting brain structure and function, iron deficiency can have a wide-ranging impact on brain development. In youth, this may compound their proclivity to develop internalizing disorders and compromise treatment response. Future studies, ideally longitudinal, should examine how changes in iron status during pubertal maturation may moderate brain development and the emergence of psychopathology. This risk may impact males and females differently, given that iron deficiency disproportionately affects menstruating females, particularly of minority background. Moreover, future interventions should seek to examine the clinical benefits of replenishing iron stores.