Transfer of omega-3 fatty acids across the blood–brain barrier after dietary supplementation with a docosahexaenoic acid-rich omega-3 fatty acid preparation in patients with Alzheimer's disease: the OmegAD study
Y. Freund Levi,
Department of Neurobiology, Caring Sciences and Society, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden
Little is known about the transfer of essential fatty acids (FAs) across the human blood–brain barrier (BBB) in adulthood. In this study, we investigated whether oral supplementation with omega-3 (n-3) FAs would change the FA profile of the cerebrospinal fluid (CSF).
A total of 33 patients (18 receiving the n-3 FA supplement and 15 receiving placebo) were included in the study. These patients were participants in the double-blind, placebo-controlled randomized OmegAD study in which 204 patients with mild Alzheimer's disease (AD) received 2.3 g n-3 FA [high in docosahexaenoic acid (DHA)] or placebo daily for 6 months. CSF FA levels were related to changes in plasma FA and to CSF biomarkers of AD and inflammation.
At 6 months, the n-3 FA supplement group displayed significant increases in CSF (and plasma) eicosapentaenoic acid (EPA), DHA and total n-3 FA levels (P <0.01), whereas no changes were observed in the placebo group. Changes in CSF and plasma levels of EPA and n-3 docosapentaenoic acid were strongly correlated, in contrast to those of DHA. Changes in DHA levels in CSF were inversely correlated with CSF levels of total and phosphorylated tau, and directly correlated with soluble interleukin-1 receptor type II. Thus, the more DHA increased in CSF, the greater the change in CSF AD/inflammatory biomarkers.
Oral supplementation with n-3 FAs conferred changes in the n-3 FA profile in CSF, suggesting transfer of these FAs across the BBB in adults.
The omega-3 (n-3) and omega-6 (n-6) highly unsaturated fatty acids (FAs), including docosahexaenoic acid (DHA, 22 : 6 n-3) and arachidonic acid (AA, 20 : 4 n-6), accumulate in the central nervous system (CNS) during foetal growth. By contrast, only trace amounts of eicosapentaenoic acid (EPA, 20 : 5 n-3) are found in nervous tissues. It is generally assumed that FAs in the cerebrospinal fluid (CSF) are continuously exchanged during life, but little is known about the rate of exchange and how dietary changes affect the transfer of FAs across the human blood–brain barrier (BBB). In young animals, dietary supplementation with DHA results in an increase in DHA concentration in the CNS [1-4].
Various diseases may affect the FA composition of the CNS. Thus, lower levels of DHA and other n-3 polyunsaturated FAs as well as AA have been noted in brain phospholipids and lipid rafts [5, 6] and blood lipids  in patients with Alzheimer's disease (AD), compared with control subjects. Likewise, schizophrenia and major depression have been shown to be associated with low brain DHA levels .
Accumulating evidence suggests that a high intake of fish is linked to reductions in the risk of developing AD [9, 10] and may delay cognitive decline in patients with mild cognitive impairment (MCI) and very mild AD [11, 12] and also reduces symptoms of mood disorders . Higher levels of DHA or other n-3 FAs in the blood have been correlated with a decreased risk of cognitive loss in normal ageing  and the development of dementias [10, 15]. However, supplementation with n-3 FAs in patients with moderate AD did not confer improvement or reduced progression of declining cognitive functions [11, 16]. Although these observations might establish a link between dietary n-3 FAs and CNS pathologies, few studies have investigated the correlation between dietary n-3 FAs and FA composition in the CNS .
Here, we present results from the randomized, placebo-controlled OmegAD study in which a marine oil rich in DHA was given as a dietary supplement to patients with mild to moderate AD for 6 months. We followed changes in long-chain polyunsaturated FA levels in CSF and plasma . In the present report from this study, we also investigated the relation between CSF FA levels and several inflammation markers and AD-related CSF proteins, reported in .
Materials and methods
A total of 40 patients were initially included; however, only 33 were followed up (18 receiving n-3 FA supplementation and 15 receiving placebo; Fig. 1). These patients in the present study were participants in the double-blind, placebo-controlled OmegAD study in which 204 patients with mild AD received 2.3 g n-3 FAs (high in DHA) daily or placebo for 6 months . The ethical committee at the Karolinska University Hospital approved the study protocol. All patients and caregivers provided written informed consent. The inclusion criteria were as follows: (i) a diagnosis of mild to moderate AD according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria, (ii) Mini Mental State Examination scores between 15 and 30 (inclusive) and (iii) treatment with a stable dose of acetylcholine esterase inhibitor for at least 3 months before study start. Further study details have been reported previously . Patients were assigned randomly to receive four 1 g capsules daily, each containing either 430 mg DHA and 150 mg EPA (i.e. EPAX1050TG, the Omega-3 group; Pronova Biocare A/S, Lysaker, Norway) or an isocaloric placebo oil (containing 1 g corn oil, including 0.6 g linoleic acid; the placebo group) for 6 months.
Approval was given by the local ethics committee for a subgroup of 40 patients to undergo lumbar puncture at baseline and after 6 months (Fig. 1). Per protocol analyses were performed. Thus, this report is based on CSF and plasma data from 33 patients at baseline and after 6 months of treatment with n-3 FAs or placebo.
Sample collection and analyses of CSF and plasma
Plasma was prepared from blood obtained by venipuncture. Lumbar puncture was performed according to standard methods [11, 17].
Levels of FAs in plasma were analysed by gas chromatography . Results are presented as area percentages.
Details of the CSF FA analysis methods are given in the Supporting information. In brief, FAs were extracted from CSF according to the methods of Bligh and Dyer , converted to methyl esters , and analysed using a gas chromatographic method designed to measure very low FA concentrations. Results are given primarily as concentration (ng mL−1) and, for comparisons with plasma FA values, as weight percentages.
CSF levels of IL-6, soluble interleukin-1 receptor type II (sIL-1RII) and biomarkers of AD [phosphorylated tau (P-tau), total-tau (T-tau) and β-amyloid1–42 (Aβ1–42)] were analysed by enzyme-linked immunosorbent assay .
Data are presented as mean and 95% confidence interval (CI) unless otherwise stated. Single within-group differences were analysed using a dependent Student's t-test. Changes over time, between groups and interactions, were then evaluated by repeated measures anova. For simple correlations, the Pearson's product moments correlation test was applied. A P-value < 0.05 was considered significant for all analyses, and all methods were checked for sphericity, normality and homogeneity of variance if appropriate. The analyses were performed using sas version 9.1.3 (SAS Institute, Cary, NC, USA) or statistica version 12.0 (StatSoft, Tulsa, OK, USA).
At baseline, n-3 and n-6 FAs comprised 2.2% and 11.2%, respectively, of all FAs in the CSF. The main FAs detected in CSF were, in decreasing order of magnitude, myristic (14 : 0), palmitic (16 : 0), oleic (18 : 1 n-9), stearic (18 : 0), linoleic (18 : 2 n-6), AA (20 : 4 n-6), DHA, tetracosenoic (24 : 1) and docosatetraenoic acids (22 : 4 n-6; Fig. 2 and Table 1). The approximate concentrations at baseline of AA, EPA, docosapentaenoic acid (22 : 5 n-3; DPA n-3) and DHA were 474, 32, 20 and 243 ng mL−1, respectively. Hence, there was nearly an eightfold higher concentration of DHA than EPA and DPA n-3 in the CSF, whereas the AA concentration was approximately twofold higher than that of DHA.
Table 1. Mean CSF FA concentrations (95% CI) at baseline and after 6 months of supplementation with a DHA-rich n-3 FA preparation
CSF FAs, ng mL−1 (95% CI)
Omega-3 FA group (n =18)
Placebo group (n =15)
CSF, cerebrospinal fluid; FA, fatty acid; LA, linoleic acid; AA, arachidonic acid; DTA, docosatetraenoic acid; EPA, eicosapentaenoic acid; DPA, docosapentaenoic acid n-3; DHA, docosahexaenoic acid; CI, confidence interval. aThe n-6/n-3 FA ratios are based on the original data given as ng mL−1. *P <0.05; **P <0.01 and ***P <0.001: dependent Student's t-tests for changes between baseline and at 6 months.
At baseline, there were no significant differences in the concentrations of EPA, DPA n-3 or DHA, or all n-3 FAs (i.e. the sum of α-linolenic acid, EPA, DPA n-3 and DHA), between the Omega-3 FA and placebo groups (Table 1). Absolute concentration changes (in ng mL−1) during the trial are shown in Fig. 2. Six months later, EPA, DPA n-3, DHA and all n-3 FA levels had increased significantly (t-tests) in the Omega-3 group, whereas levels of AA and docosatetraenoic acid and the ratio of n-6/n-3 FAs were lower than in the placebo group (Fig. 2 and Table 1). No significant changes were noted for the other individual FAs in the Omega-3 or the placebo groups. Of note, there was no increase in linoleic acid (but rather a decline) in the placebo group given corn oil (with a small amount of linoleic acid) as a constituent of the placebo capsules.
anova was performed, to evaluate the interaction of time and treatment effects, and changes in the levels of EPA and total n-3 FAs and the ratio of n-6/n-3 FAs remained statistically significant (Fig. 2). The change in DHA concentration was not significant (P =0.2). There were no significant changes in the total FA content of CSF (Table 1).
At trial entry, the two treatment groups did not differ significantly with regard to plasma levels of EPA, DPA n-3 and DHA (as percentages of all FAs; see Figure S1). In the Omega-3 FA group, plasma levels of DHA and EPA as well as of all n-3 FAs were significantly higher at 6 months compared with pretrial values, whereas no such changes were observed in the placebo group.
In the Omega-3 group, there were significant reductions in the percentages of plasma linoleic acid and AA (see Figure S1). In agreement with previous findings , plasma linoleic acid percentages were higher in the placebo group and lower in the Omega-3 group at 6 months compared with the respective baseline values (anova, P =0.001).
Comparison of CSF and plasma FAs
Based on the percentage values of the individual FAs in plasma and CSF at baseline for all patients (Figure S1), polyunsaturated FA values were generally much higher in plasma than in CSF; the exception to this was for DPA n-3. Thus, the approximate proportion of linoleic acid in plasma for all patients was 22%, compared with only 6% in CSF, giving a 3.6-fold higher concentration in plasma. The corresponding ratios for AA, EPA, DHA and for all n-3 FAs were 1.8, 10.0, 2.3 and 3.4, respectively, indicating that the difference between plasma and CSF was greatest for EPA, followed by linoleic acid, all n-3 FAs and DHA, and least for AA levels. Hence, a gradient between plasma and CSF was observed, which was highest for EPA (in relative terms, as absolute EPA levels were very low in CSF) and lowest for AA.
Cytokines and growth factors
Cytokine and growth factor concentrations in CSF and plasma have been presented previously . Briefly, there was no significant treatment effect of oral n-3 FA supplementation for 6 months on inflammatory (IL-6 and sIL-1RII) and AD biomarkers (P-tau, T-tau and Aβ1–42) in CSF and also no effect on inflammatory markers (IL-6, sIL-1RII and high-sensitivity C-reactive protein) in plasma.
When changes in individual CSF FA values were related to those of plasma FA (using weight percentage values of all FAs), we found that EPA and DPA n-3 levels correlated significantly, but no correlation was observed for DHA or AA levels (Fig. 3). Thus, as EPA and DPA n-3 levels increased in the plasma, they were also elevated in CSF.
Next, to determine whether changes in CSF FA levels were related to AD biomarkers in CSF, we analysed the concentrations of CSF FAs (in ng mL−1). First, we noted that changes in CSF levels of biomarkers of AD (i.e. P-tau and T-tau) and sIL-1RII correlated significantly with changes in CSF DHA concentrations. As shown in Fig. 4(a–c), the greater the increase in DHA, the greater the increase in sIL-1RII levels but the greater the decrease in P-tau and T-tau. Similarly, there were significant correlations between the changes in DPA n-3 and sIL-1RII (r =0.370, P =0.035), and between the changes in total n-3 FAs and both sIL-1RII and P-tau (r =0.414, P =0.02 and r =−0.39, P =0.026, respectively).
There were several important findings of this study. First, we have shown that it is feasible to measure the levels of individual FAs in CSF. Our results are in line with what Pilitsis et al.  reported and extend what Quinn et al.  reported in briefly. Secondly, significant increases were observed for the most important n-3 FAs in CSF during the 6 months of oral administration of this DHA-rich supplement. Thirdly, there was a good correlation between the increases in the levels of EPA, DPA n-3 and total n-3 FAs in plasma and CSF during the supplementation period. Although the changes in DHA levels did not correlate in a statistically significant manner, increases over time in both compartments (particularly in plasma) were observed. Fourthly, there were significant correlations between changes over time in CSF concentrations of DHA and in CSF concentrations of P-tau, T-tau and sIL-1RII. Together, these findings support the hypothesis that increased oral intake of n-3 FAs leads to their accumulation in central nervous tissues and may affect nervous system physiology as well as the pathogenesis and progression of AD, particularly in very mild AD.
In previous studies in transgenic mouse AD models, it was also demonstrated that increased brain DHA content, following n-3 FA supplementation, was associated with reductions in various aspects of AD pathology, for example, amounts of plaque, amyloid-β levels and improved behaviour [21-28]. Of note, we observed in the present study that changes over 6 months of the CSF AD biomarkers tau and the anti-inflammatory cytokine receptor s-IL-1RII levels showed significant covariation with CSF DHA changes, suggesting interactions. This is intriguing because there is a marked increase in CSF T-tau in AD and also in prodromal stages of the disease. This increase in T-tau reflects the neuronal degeneration in AD but is also observed in other disorders with damaged or degenerating neurons. High CSF P-tau levels have only been observed in AD and reflect the pathological phosphorylation of tau seen in this disorder . Our findings of negative correlations between CSF concentration changes of DHA and the tau proteins might reflect an effect of n-3 FA supplementation on tau phosphorylation and neurodegeneration. However, it should be noted that mean CSF levels of the tau proteins were not significantly reduced during the n-3 supplementation . Nonetheless, this lack of change does not preclude the positive effect on cognition seen in patients with very mild disease in our clinical study  and in those with MCI and AD in previous studies during n-3 supplementation [12, 30]. Further studies should explore the roles of DHA-(and EPA-) based neuroprotectins, resolvins and growth factors .
The significant (t-test) rise in DHA in the CSF of the omega-3 group was as large in absolute terms as the rise in EPA in the CSF (approximately 30 and 50 ng mL−1, respectively), despite nearly an eightfold greater level of DHA relative to EPA in the CSF. This might suggest a difference in the handling of the two FAs during the passage from blood across the BBB to the CNS. It is unlikely that minor differences in molecular weight and other physical properties would create selectivity for transfer of these the long-chain n-3 FAs across the BBB, but we cannot exclude the possibility that the DHA-rich dietary supplement specifically attenuated DHA transfer [32, 33]. A similar entry of EPA and DHA into the rat brain has been demonstrated recently by Ouellet et al. . We suggest that further investigations should be carried out to elucidate whether DHA was specifically retained in the CNS because of the AD-related deficiency of DHA in brain tissue, as observed by others [5, 6]. By contrast, EPA, a very minor constituent of the brain, seemed to enter the brain and appeared in the CSF. This finding is somewhat contrary to the usually assumed notion that the CSF represents a drainage system for the brain [34, 35]. The possibility of dietary DHA-induced FA β-oxidation and FA release to CSF should also be considered [36-38]. Finally, we were not able to determine how much DHA was retroconverted to EPA in the plasma or in the CNS, but because so much more DHA is found in the CSF (compared with EPA), it is likely that a modest increase in the retroconversion during oral supplementation would be sufficient to raise EPA levels in the CSF.
Our results might lead to further exploration of the positive effect of oral FA supplementation on preservation of memory functions in patients with MCI  and, possibly, very mild AD . However, clinical trials of n-3 FAs in patients with AD with moderate to severe disease have so far been unsuccessful in halting the progression of AD . This suggests that there might be a temporal window in the pathogenesis of AD in which n-3 FA supplementation can benefit patients with MCI but attempts to restore CNS DHA loss cannot affect the clinical course of AD progression beyond a certain time-point.
In conclusion, we have demonstrated here that specific CSF n-3 FA changes occur during supplementation with a DHA-rich n-3 FA preparation that correlate significantly with changes in various biomarkers of AD. The underlying mechanisms and consequences of these findings remain to be elucidated.
YFL, IV, TC and JP: study design, collection and analysis of data, preparing and writing the manuscript. HB, GFI, ME and L-OW: study design and preparing the manuscript. BV, MS, EH: study design, laboratory analyses and preparing the manuscript. NS: laboratory analyses, collection of data, preparing and writing the manuscript.
Conflict of interest
FA analyses were performed, and data were analysed whilst NS was employed at the National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism Intramural Research Program; he is now an employee of DSM Nutritional Products LLC., and the company produces and sells both n-3 and n-6 polyunsaturated FAs. JP has received a speaker's honorarium from Solutex Inc., Madrid, Spain. None of the other authors has any conflict of interest to declare.
We thank A.-C. Tysén-Bäckström and Andreas Svensson, RN for patient management, Siv Tengblad, BMA, for analysis of plasma n-3 FA levels and Sharon Majchrzak for CSF FA analyses.
Financial support was provided through The Regional Agreement on Medical Training and Clinical Research (ALF) between Stockholm County Council and the Karolinska Institutet (20110263, 20110604), and by Funds of Capio, Demensförbundet, Gamla Tjänarinnor, Swedish Alzheimer Foundation, Odd Fellow Sweden, Swedish Nutrition Foundation, Gun och Bertil Stohnes Foundation, Swedish Society of Physicians and Lion's Sweden. The OmegAD study was initially partly funded by Pronova Biocare A/S, Lysaker, Norway; the company was represented in the trial steering committee with regard to study design and provided the EPAX1050TG and placebo preparations, but was not involved in the data and patient collection and analyses or interpretation of scientific data.