Introduction

Pediatric acute-onset neuropsychiatric syndrome (PANS) is a psychiatric disorder characterized by an abrupt onset of obsessive compulsive symptoms and/or severe eating restrictions along with at least two other severely debilitating cognitive, behavioral, or neurological symptoms.1,2 PANS has been reported and characterized by different centers around the world.3,4,5,6,7,8 A recent MRI study from our institution showed microstructural changes throughout the brain as indicated by increased median diffusivity in patients with PANS compared to controls.9 The deep gray matter (e.g., the thalamus, basal ganglia, and amygdala) demonstrated the most profound increases in diffusivity consistent with the cardinal clinical symptoms. These findings go along with previous imaging studies and suggest inflammation in these regions.10,11 A growing body of evidence supports the role of inflammation in PANS.10,11,12,13,14,15,16,17,18,19,20

Iron deficiency remains the most common childhood nutritional deficiency in the United States.21,22 Low dietary iron intake, poor absorption of iron, chronic diseases, and chronic blood loss are causes of iron deficiency and iron deficiency anemia.23,24 Children with anxiety, depression, tic disorders, attention-deficit/hyperactivity disorder, febrile seizures, breath holding spells, and fibromyalgia are more likely iron deficient.25,26,27,28,29,30,31,32,33,34 The medical literature supports the role of iron deficiency contributing to the development of neurologic issues, through a negative effect on neurodevelopment in children.35,36,37 Iron deficiency occurring in pregnant mother and fetus may even impact prenatal neurodevelopment and the neurological impact may last even after iron repletion.38,39,40

The total body iron stores are estimated by measuring serum ferritin levels. Ferritin varies less than serum iron and is more sensitive to iron store depletion.41 It is also a well-known acute phase reactant, and nonspecific elevations occur in infections, inflammation, hematochromatosis, and malignancy.41,42,43 During infection, increased ferritin levels play a role in host defense against bacterial growth and may also modulate inflammation.43 While evidence suggests the association between PANS, inflammation, and streptococcal infections,8,9,10,11,44,45 patients seen in our PANS clinic appeared to have hypoferritinemia (low serum ferritin levels) on laboratory workup which is usually ordered during their PANS flares (abrupt worsening of neuropsychiatric symptoms) or during the phase following flares which are characterized by fatigue and cognitive difficulties.46 This paradoxical observation in PANS, particularly during flares, remains unexplained.

To confirm our clinical observation, we performed this study with the objective to examine the prevalence of iron deficiency in patients with PANS, and compare it to a national cohort from the Centers for Disease Control and Prevention (CDC).22 We examined the likelihood of hypoferritinemia coinciding with a PANS flare by comparing the groups of patients with and without hypoferritinemia during a PANS flare, with regard to demographics, diets, and clinical characteristics. We also explored the potential association between hypoferritinemia and neuropsychiatric symptoms over time.

Methods

Study design, source of data, and participants

The Stanford Institutional Review Board approved this study (IRB#26922). Consent was obtained from parents of minors and adult patients, and assent obtained from patients aged 7–17 years. Subjects were recruited from the Stanford PANS/Immune Behavioral Health Clinic. Questionnaires were given to patients prior to each clinic visit inquiring about medical conditions, psychiatric symptoms, and treatments. In addition, patients completed a demographic and dietary questionnaire during the study period (September 3, 2012 to March 30, 2018). Laboratory work to evaluate iron stores and anemia was requested on all new patients to our clinic during the above study period. However, we were unable to obtain these studies in some patients for the following reasons: patient unwilling or too anxious to do a lab draw or need to prioritize other laboratory tests. To supplement data from the patient questionnaire, we also reviewed electronic medical records for psychiatric symptoms at clinic entry and for history of eating restriction.

For this study, we included consecutive patients who met strict PANS criteria.2,47 We excluded patients who declined research, had <3 clinic visits, and never had ferritin levels checked. To avoid potential referral bias, we restricted our study cohort to the community patients living within 90 miles from our clinic and the seven surrounding counties.

Definitions of variables

The first serum ferritin test result available after PANS onset was included for analysis. Hypoferritinemia was defined by serum ferritin levels <7 ng/ml in children and youth aged ≤ 15 years, and <18 ng/ml in adolescents aged >15 years.48 A more stringent criterion was used in our study, compared to the WHO (<15 ng/ml) and CDC reports (<9 ng/ml), because some publications report serum ferritin levels can be as low as 6–7 ng/ml in healthy people.49,50 This strategy helps avoid overestimation of iron deficiency in our sample.

We defined iron deficiency by the presence of hypoferritinemia and concurrently either of the following: (a) a low transferrin saturation, or (b) a high total iron binding capacity (TIBC). Iron deficiency anemia was diagnosed when a patient had iron deficiency and a concurrent low hemoglobin with reference to the laboratory.

Eating restriction was diagnosed if the following two criteria were met: (a) the chart mentioned restricted food intake (reduced appetite, restriction due to fear of choking, etc.), distress or resistance with food, and (b) this new eating restriction resulted in weight loss of more than two pounds documented in the chart.

A PANS flare was defined by neuropsychiatric deterioration reported by parents, teachers, and patients together with the worsening of psychometric test scores on the questionnaire and confirmed by clinical interview.

Statistical analysis

We calculated the proportion of patients that had hypoferritinemia and iron deficiency with and without anemia. Since adolescent females are at an elevated risk for iron deficiency,22,51 we checked the robustness of our result by excluding females over 12 years old at the time of ferritin tests, in order to minimize the confounding effect by periodic menstrual blood loss. For the fact that some patients did not have ferritin or iron workup, we performed a series of sensitivity analyses in order to ascertain the range of uncertainty of iron deficiency in patients with PANS. We made three separate assumptions for the risk of iron deficiency in patients who did not have the full iron workup (untested patients). We first assumed untested patients came from a random sample of our PANS population; second, we assumed untested patients had the same risk of iron deficiency as the general population matched for age and sex; last, we assumed none of our untested patients were iron deficient. Under each assumption, we recalculated the prevalence rate, and compared it to the age- and sex-matched prevalence estimates of iron deficiency in the general population.22 Odds ratio and 95% confidence interval were reported.

Prior studies have suggested hypoferritinemia can coincide with psychiatric symptoms in other disorders;25,26,27,28,29,30,31 thus, we compared characteristics of patients with and without hypoferritinemia during a PANS flare, using chi-square or Fisher’s exact tests for categorical variables, and two-sample t tests or Mann–Whitney U test, whenever appropriate. We chose to restrict the comparison to “flare state” since the disease state itself can be a confounder. Factors to be examined included demographics, socioeconomic status, dietary factors, body mass, psychiatric symptom severity, and caregiver burden. We used maternal education and annual household income as the proxy of socioeconomic status.52

To explore the association between changes in serum ferritin levels over time and psychiatric disease severity, we performed mixed models with random time. Outcome measures were patient’s global impairment (GI) and caregiver burden inventory (CGBI) scores, both of which are rated by parents before each clinic visit. Global impairment score is a parent-rated scale ranging from 0 to 100 (0 = no impairment, 100 = worst impairment), and was validated in our PANS population.53 Caregiver burden inventory score is a measure of caregiver burden.54 Its use is considered valid and reliable in the PANS population.46,55 These two scores within 10 days of ferritin tests were collected. They were used as an outcome measure in separate mixed models while hypoferritinemia (yes/no) was a time-varying variable, and other covariates in the model included time-in-clinic, sex, PANS flare (defined as an abrupt worsening of neuropsychiatric symptoms; yes/no), and chronic state of illness (defined as ongoing neuropsychiatric symptoms for at least 9 months continuously; yes/no).46

Statistical analysis was performed using Statistical Analysis Systems software program (SAS® University Edition, the United States). All statistical tests were considered statistically significant if two-sided p value was <0.05.

Results

Our study cohort included 79 consecutive community patients with PANS and available ferritin test results (Fig. 1). The median time ± interquartile range (IQR) between PANS onset and the first ferritin tests was 3.1 ± 4.8 years. The mean age ± standard deviation (SD) of PANS onset was 8.7 ± 3.8 years (Table 1). The majority of our patients were male (63%) and non-Hispanic White (94%). The socioeconomic status was high as reflected by a low percentage (3%) of families having annual household income <US$50,000 and a high percentage (78%) of mothers attaining college education or above. At clinic presentation, common symptoms reported by our study patients included obsessive compulsive symptoms, anxiety, irritability, emotional lability, motor and somatic symptoms.

Fig. 1
figure 1

Inclusion and exclusion flowchart to study ferritin levels in patients with pediatric acute-onset neuropsychiatric syndrome (PANS).

Full size image
Table 1 Clinical characteristics of 79 consecutive patients with pediatric acute-onset neuropsychiatric syndrome (PANS) who met the study entry criteria.
Full size table

Based on our definition of hypoferritinemia, 21/79 (27%) of our study cohort had hypoferritinemia (Table 2). This number dropped slightly to 21% after exclusion of girls older than 12 years old. Three quarters of hypoferritinemic cases were observed during a PANS flare. The prevalence of iron deficiency in our study cohort of patients who had iron workup was 14%, and about half of patients with iron deficiency also had anemia. After excluding girls older than 12 years, prevalences of hypoferritinemia, iron deficiency, and iron deficiency anemia dropped slightly.

Table 2 Ferritin and iron profiles in 79 consecutive patients with pediatric acute-onset neuropsychiatric syndrome (PANS) who met the study criteria.
Full size table

Our sensitivity analyses under the first two assumptions of iron deficiency prevalence in 96 patients who did not have ferritin (n = 68) and iron workup (n = 28) showed that our patient sample had a higher odds of iron deficiency by 1.4–2.0-fold than the sex- and age-matched population in the United States (Table 3).22 The last assumption, which was the most restrictive one by assuming all untested patients were not iron deficient, gave a lower odds of iron deficiency than the sex- and age-matched population (odds ratio 0.7, 95% CI 0.4–1.1, p = 0.11).

Table 3 Iron deficiency prevalence in a cohort of consecutive patients with pediatric acute-onset neuropsychiatric syndrome (PANS) and the general population in the United States.
Full size table

Table 4 shows a comparison between patients with and without hypoferritinemia observed during a PANS flare. The latter group included patients with normal serum ferritin levels during a PANS flare (n = 49) and cases at remission (n = 14). There was no evidence of demographic differences (including age of PANS onset, sex, race/ethnicity, annual household income, and maternal education) between the two groups. Vegetarian and “no red meat” diets were common and comparable in both groups. At clinic entry, the hypoferritinemic group were older and had worse GI scores (p < 0.01) than the other group (Table 4). BMI, eating restriction, and socioeconomic status (reflected by annual household income and maternal education) were similar. However, at the time of ferritin tests, the hypoferritinemic group had a higher rate of chronic PANS illness (69% vs 46%) and higher GI scores (63.8 vs 43.9, p < 0.01) when compared to the other group. At the time of ferritin tests, GI and CGBI scores were both higher in the hypoferritinemic group, which is compatible with the definition of flares, in which psychometric impairment scores would be increasing. Comorbid inflammatory diseases were also more commonly seen in the hypoferritinemic group. As expected, low hemoglobin (Hb), low mean corpuscular volume (MCV), and high red blood cell distribution width (RDW) were also more common in the hypoferritinemic group.

Table 4 Comparison of demographic and clinical characteristics in 79 study patients with and without hypoferritinemia during a pediatric acute-onset neuropsychiatric syndrome (PANS) flare.
Full size table

The mixed models showed that hypoferritinemia at different time points were not significantly associated with psychometric impairment scores such as GI (GI 4.84, SE 3.93, p = 0.22) and CGBI (CGBI 4.67, SE 5.66, p = 0.39). In contrast, the disease state (flare or not, chronic illness or not) was associated with these psychometric impairment scores with statistical significance (p < 0.01).

Discussion

In our PANS study cohort, a quarter (27%) of tested patients had hypoferritinemia. If we expand the denominator and include untested patients (i.e., no ferritin test obtained), the rate was halved (14%). In patients who had an iron workup, 14% (7/51) had iron deficiency and 6% (3/51) had iron deficiency anemia. The odds of iron deficiency in our patients was 1.4–2.0-fold higher than in the age- and sex-matched general population in the United States;22 this finding is especially remarkable since we used more stringent cut-off levels for ferritin than the CDC comparison population. Three quarters of hypoferritinemic cases were observed during a PANS flare. When comparing the two groups of patients with and without hypoferritinemia during a PANS flare, we did not observe differences in demographics or dietary factors. However, hypoferritinemia during a PANS flare was usually associated with chronic PANS state and higher global impairment and caregiver burden at the time of ferritin tests.

An increased rate of hypoferritinemia and iron deficiency in our PANS cohort does not appear to be associated with typical demographic factors linked to iron deficiency. In our community PANS cohort, only 3% had annual household income <US$50,000; 78% of patients’ mothers attained college education or above; and 94% of patients were non-Hispanic White (Table 1); these factors are known to be associated with adequate dietary intake and normal iron levels.51 Eating restriction and vegetarian diets were found in roughly one third of our study patients. However, these patients were usually fed with other iron-rich food, such as green leafy vegetables, white meat, or iron fortified products, according to the parents.

Other factors could potentially explain a higher rate of hypoferritinemia and iron deficiency seen in our PANS cohort. First, a subgroup of PANS, called pediatric autoimmune neuropsychiatric disorders associated with streptococcal infections (PANDAS), has been associated with Streptococcal infections.56,57 Extracellular bacterial infections such as streptococcal, promote the expression of hepcidin, which acts to block iron exit from enterocytes in the duodenum and iron-recycling macrophages.58 This reduces iron delivery to the circulation and subsequently results in anemia. Since the gut plays a role in iron metabolism, microbiome shift in the gut may be part of the picture. A recent study has also showed that gut microbiome changes in patients with PANS/PANDAS.59 Use of medications and varying eating patterns during PANS flares may further complicate the picture.

Accumulating evidence points to a neuroinflammatory process as a cause of neuropsychiatric symptoms in PANS/PANDAS,9,10,11,12,13,14,15,16,17,18,19,20 but the role of iron has not yet been understood. In PANS/PANDAS, the inflammatory processes are more pronounced in certain structures, e.g., basal ganglia, thalamus, amygdala.9,10,11 In murine models, proinflammatory stimuli such as lipopolysaccharide or β-amyloid (Aβ) enhance brain microglia to preferentially take up the non-transferrin bound iron, therefore increasing iron deposition in inflamed regions.60 The excessive iron in the brain then releases radical oxygen species, and other proinflammatory factors, further exacerbating neuroinflammation, apoptosis, and brain injury. This process might be similar to the iron accumulation and brain injury involved in many neurodegenerative diseases such as multiple sclerosis, pantothenate kinase-associated neurodegeneration, Alzheimer’s disease, and Parkinson’s disease where increased basal ganglion iron has been observed on brain imaging.61,62,63,64 Reduced iron delivery from the gut and macrophages as a result of increased hepcidin expression coupled with increased iron delivery to the inflamed brain are possible explanations for a higher rate of hypoferritinemia in our PANS patients, especially during a PANS flare.

Hypoferritinemia has also been observed in Behcet’s disease, which causes inflammation in blood vessels (vasculitis) and neuropsychiatric disease; this finding is remarkable especially since we used more stringent cut-off levels for ferritin than the CDC comparison population. One study of Behcet’s disease reported low serum ferritin in 15% of patients.65 Interestingly, the same HLA epitope (HLA-Bw4) associated with Behcet’s disease has also been shown to be strongly associated with PANS compared to healthy controls.66,67 These similarities in patients with PANS and Behcet’s disease may suggest similar underlying inflammatory processes in these diseases.

By the end of the study period, most of the hypoferritinemic patients (11/16, 69%) had normalization of ferritin levels after receiving either IVIG (n = 6, with average of 2.8 infusions preceding the resolution), methylprednisolone (n = 5), prednisone bursts (n = 6), NSAID (n = 8), or a combination of these therapies. Only three of these patients took concurrent iron supplements. Of the remaining 5/16 patients, three did not have ferritin levels rechecked by the end of the study, and two failed to normalize ferritin levels before the end of this study. Further studies on the effects of immunomodulation on hypoferritinemia and iron deficiency are required to guide clinicians in treating patients with hypoferritinemia during PANS flares.

Study strengths and limitations

To our best knowledge, this study is the first study to describe the prevalence of hypoferritinemia and iron deficiency in a cohort of patients with PANS. We used a stringent cut-off value for hypoferritinemia to estimate its prevalence and the prevalence of iron deficiency. The present conservative estimate, which may or may not be underestimated, was still higher than the general population.

Our study has several limitations. First, while we intended to systematically measure ferritin levels in every patient, various factors impeded this evaluation, including behavioral/anxiety reactions to blood draws, and phlebotomy blood volume limits to prevent excessive phlebotomy in small children. Post-hoc comparative analysis of the included and excluded patients showed that patients excluded for missing ferritin levels (but otherwise would meet inclusion criteria) from the study cohort were younger at their first clinic visit (9.9 years in the excluded vs 11.4 years in the included, p = 0.03). They did not differ in sex ratio, race/ethnicity, or symptoms at clinic presentation. Thus, our results are only generalizable to older children at this time. Second, for the estimation of iron deficiency prevalence, we matched patients by age and sex, but due to limitations in the CDC data available to us, we were unable to match 1:1 for race and ethnicity. However, we compared our PANS cohort (94% were non-Hispanic White) to the CDC non-Hispanic White data; the difference is likely minimal. Third, our data about socioeconomic status and diet preference were cross-sectional at the clinic entry, and we did not have follow-up data to address possible changes over time. We collected data regarding eating restriction and body mass at the time of ferritin tests through chart review and patient questionnaires, in order to delineate any significant changes in these parameters that may potentially provide information about food supply and intake. Last, our cohort size was small; thus, our study results need to be reproduced in larger cohorts and preferably with systematic acquisition of ferritin and iron tests, although this will likely always be a challenge in young children with anxiety-related disorders.

Future directions

Given accumulating research evidence on the inflammatory processes underlying PANS, it would be informative to perform serial measurements of serum iron, ferritin, hepcidin, and erythrocyte kinetics in patients with PANS, particularly during flare vs remission states, in order to elucidate a potential role of iron homeostasis in this illness. In addition, using MRI techniques to quantify and map iron deposition in the brain, especially the thalamus, basal ganglia, and amygdala, will provide more information for understanding the pathology of PANS illness with relation to iron dysregulation.68,69

Conclusion

Our study shows an increased rate of hypoferritinemia and iron deficiency among a community cohort of youth with PANS. The odds of iron deficiency in our study patients was 1.4–2.0-fold higher than in the age- and sex-matched general population in the United States. Three quarters of hypoferritinemic cases were observed during a PANS flare, and when compared to patients without hypoferritinemia during a PANS flare, these cases presented with worse global impairment at clinic entry. At the time of ferritin tests, these patients were more likely to have chronic PANS illness and a comorbid inflammatory disease. A larger study cohort with routine evaluation of ferritin and status of iron stores in patients with PANS is needed to corroborate our findings and examine the relationship with neuropsychiatric symptom severity. MRI brain techniques to quantify and map the iron deposition in brain structures will provide additional information about iron hemostasis.