Abnormalities in CSF concentrations of ferritin and transferrin in restless legs syndrome

Methods.

We enrolled 16 patients with idiopathic RLS4 who had periodic leg movements in sleep of more than 15 per hour, a 1-year history of daily RLS symptoms, and a positive clinical response to levodopa. Exclusion criteria were iron deficiency (ferritin < 18 ng/dL), renal/metabolic disorders, neuropathy, brain or spinal cord injuries, chronic inflammatory processes, and chronic pain syndromes. Eight of the 16 RLS subjects had a family history of RLS and had onset of RLS symptoms before 40 years of age. The other eight patients had no family history and had initial symptom onset after 50 years of age. All subjects discontinued dopaminergic or opiate agents 2 weeks before the procedure. During this withdrawal period, temazepam was used for as long as 3 days before the procedure. Eight age-matched control subjects were paid $400 for undergoing the procedure. Control subjects did not have a history of chronic sleep disturbance (>3 weeks of symptoms), daytime hypersomnolence, or RLS. The lumbar puncture was performed between 9:30 am and 10:30 am on an outpatient basis. Six milliliters of CSF and 5 mL of serum were stored at −80 °C until assayed for iron, transferrin, and ferritin according to previously published methods.5,6

The primary hypothesis is that patients with RLS will have a decrease in CSF ferritin and an increase in CSF transferrin, which would be expected if, as we proposed, there are low brain iron concentrations. This was tested using the Student’s two-tailed t-test. We explored the relationship between serum and CSF values for each of the three indices by linear regression analysis for each group separately. The differences between regression lines for each group were elevated by a multiple regression model. The significant effects were determined by t-test for the regression variables. Differences between control subjects and patients for age were tested using the standard two-tailed t-test, and for gender were tested using the chi-square test.

Results.

A comparison between the eight RLS subjects with a family history and the eight without a history showed no difference (p > 10) on any serum or CSF measures. Thus, additional analyses were performed using the values from all 16 RLS subjects. The mean age of the RLS patients was 64.2 years (range, 44 to 79 years) and of the control subjects was 61.2 years (range, 49 to 80 years). There were no differences between control subjects and RLS patients for serum transferrin (p = 0.16) or ferritin (p = 0.44; figure 1). The serum iron values of the RLS patients were higher than those of the control subjects (1,110 ± 80 μg/L versus 650 ± 50 μg/L; p < 0.002). However, serum iron values for both groups were within the normal range (250 to 1,750 μg/L). In support of our primary hypothesis, CSF ferritin levels were markedly lower (1.11 ± 0.25 ng/mL versus 3.5 ± 0.55 ng/mL; p = 0.0002) and the transferrin levels were higher (26.5 ± 5.1 mg/L versus 6.71 ± 1.6 mg/L; p = 0.018) in RLS subjects than in the control subjects (figure 2). CSF iron levels were not significantly different. The correlation (figure 3A) between the serum and CSF ferritin levels for the control subjects was r = 0.72 (p = 0.046). The correlation (see figure 3A) between the serum and CSF ferritin for the RLS patients was r = 0.64 (p = 0.007). The slopes for the RLS and control regression lines were 0.04 (95% CI, 0.0014 to 0.01) and 0.1 (95% CI, 0.0003 to 0.2), respectively. The multiple regression for the overall model gave r = 0.85 with p = 0.001. The y-intercepts (p = 0.001) but not the slopes (p = 0.12) were different for the two regression lines. There was no significant correlation between serum and CSF values for iron or for transferrin. A conjoint plot of CSF ferritin versus CSF transferrin concentrations showed that all RLS conjoint values fell outside of the normal range (figure 3B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Serum concentrations for iron, ferritin, and transferrin in patients with restless legs syndrome (RLS) and in control subjects (Normal). The mean for the individual values is indicated by the dark bar.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. CSF concentrations for iron, ferritin, and transferrin in patients with restless legs syndrome (RLS) and control subjects (Normal). The mean for the individual values is indicated by the dark bar.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. (A) A linear regression plot of the relationship between serum ferritin and CSF ferritin for patients with restless legs syndrome (RLS; •) and control subjects (□). (B) Conjoint plot of CSF ferritin and CSF transferrin values from RLS patients (○). The distribution of control conjoint values is contained within the area demarcated by the rectangular box.

Discussion.

A rise or fall in brain iron leads to a concomitant rise or fall in ferritin, reflecting its importance as the primary iron storage protein in the brain.3 Transferrin, which is the primary iron transport protein in the brain, reflects an increased or decreased tissue need for iron.3 For example, iron deficiency in rats leads to a reduction in ferritin and a rise in transferrin in the brain.7 Therefore, the interpretation of the current findings—low CSF ferritin and high CSF transferrin—is that total brain concentrations of iron in RLS patients are lower than in people without RLS. Although CSF iron levels were normal, they were relatively low compared with the high transferrin concentration, which is consistent with the concept of a decreased availability of iron in the brain. Furthermore, by using CSF ferritin and CSF transferrin as a conjoint value, we were able to segregate completely RLS subjects from control subjects (see figure 3B). The use of CSF ferritin–transferrin conjoint values may prove to be the first effective biologic marker of this disorder.

Serum iron levels, distinct from tissue iron levels, are highly variable and may be affected by diet, stress, sleep behavior, and individual circadian patterns.8 Therefore, one or several of these factors could account for the higher serum iron levels seen in patients with RLS. The serum results clearly indicate that systemic concentrations of iron were not low and thus could not account for the CSF findings. The blunted serum-to-CSF ferritin relationship seen in the RLS group suggests that there is an alteration in the blood–brain barrier iron transport mechanism, which may account for the current CSF finding. The cerebral capillary transferrin receptor is an important iron-controlling protein in the interface between blood and CSF.3 Therefore, the role of brain capillary transport of iron may be the key to understanding the mechanism by which low brain iron levels are maintained in RLS.

RLS has a well-defined circadian pattern, with symptoms being dominant at nighttime.4 Serum iron has a marked circadian variation, with as much as a 30 to 50% drop in serum iron concentration at night.8 Low brain iron concentration in patients with RLS may create a greater dependence within the brain on the serum level. The nighttime drop in serum iron may translate into a clinically remarkable drop in brain levels in patients with RLS and thus create symptoms. An early-morning rebound in serum iron levels might follow as compensation for lower nighttime levels and may explain the higher levels seen in our patients in the morning. The mechanism by which low brain iron concentration causes RLS is unknown. There is circumstantial evidence for a role of the dopaminergic system in the pathophysiology of RLS.9 Tyrosine hydroxylase is the rate-limiting enzyme in the production of dopamine and requires iron as a cofactor for hydroxylation. Also, iron deficiency may affect dopamine receptors indirectly.10

There are probably several mechanisms involved in the pathophysiology of RLS. However, we believe that in patients with idiopathic RLS, a reduction in the relative availability of iron in the brain is a contributing factor to the symptoms.