Abstract
Background:
Increased fatty acid-binding protein 3 (FABP-3) levels have been reported in neurodegenerative diseases, including Alzheimer’s disease (AD). Cerebrospinal fluid (CSF) FABP-3 has therefore been proposed as a putative marker for dementia. Population-based studies examining whether CSF FABP-3 predicts later development of dementia are lacking.
Objective:
The aim of this study was to examine CSF levels of FABP-3 in relation to later development of dementia in elderly women and in relation to Aβ42, T-tau, P-tau181, and CSF: serum albumin ratio.
Methods:
86 non-demented women aged 70–84 years who participated in the Prospective Population Study of Women in Gothenburg, Sweden took part in a lumbar puncture in 1992–93. CSF-FABP-3, Aβ42, T-tau, P-tau181, and the CSF: serum albumin ratio were measured at baseline. Participants were examined with a neuropsychiatric exam at baseline and at follow-up in 2000. Dementia was diagnosed in accordance with DSM-III-R criteria.
Results:
Between 1992 and 2000, 8 women developed dementia (4 AD, 3 vascular dementia, 1 mixed vascular dementia and AD). Higher levels of CSF-FABP-3 at baseline were related to development of dementia (OR 1.36 CI [1.05–1.76] p = 0.022) and the subtype AD (OR 1.38 CI [1.06–1.82), p = 0.019) during follow-up. FABP-3 correlated with CSF T-tau (r = 0.88, p < 0.001), P-tau181 (r = 0.619, p < 0.001), and CSF:serum albumin ratio (r = 0.233, p = 0.031), but not with Aβ42 (r = –0.08, p = 0.444)
Conclusion:
CSF FABP-3 may be an early marker for later development of dementia, probably related to neuronal degeneration, but independent of Aβ metabolism.
Keywords
INTRODUCTION
Three neuropathological features were noticed by Alois Alzheimer when he described the brain of a patient with what came to be known as Alzheimer’s disease (AD) [1, 2]. Intracellular neurofibrillary tangles composed by the microtubule associated protein tau and extracellular plaques containing the amyloid-β (Aβ) protein have become hallmarks of the disease. The third feature, intracellular lipid accumulation has received less attention. Dyslipidemia has been implicated as a risk factor for the development of AD in both epidemiological and experimental studies [3]. Further support for the role of lipids and cholesterol in AD can be gained through the connection between lipid rafts and regulation of Aβ metabolism [4]. Moreover, the strongest genetic risk factor for the development of sporadic AD is possession of the apolipoprotein E (APOE) epsilon 4 (ɛ4) allele, which is related to lipid transport and metabolism. Apolipoproteins bind lipids, including cholesterol, to form lipoproteins, and act as ligands to low-density lipoprotein receptors thereby regulating lipid uptake metabolism, and have also been suggested to interact with the Aβ peptide. The E4 variant of APOE seems to be related to Aβ dysmetabolism and/or aggregation [5, 6]. The intracellular transport of fatty acids, cholesterol, and retinoids is facilitated by fatty acid-binding proteins (FABP). FABP polymorphisms have been associated with disorders of lipid metabolism and the development of atherosclerosis [7, 8, 7, 8].
Long chain polyunsaturated fatty acids (PUFA), such as docosahexaenoic acid (DHA; omega-3) and arachidonic acid (ARA; omega-6), are essential constitutive fatty acids in the mammalian brain. The lipid bilayer of neuronal membranes, consisting of phospholipids with DHA and ARA as main components, are known to modulate the fluidity and permeability of cell membranes and contribute to the membrane remodeling and neurite growth [9]. PUFA are transported by FABPs. Three types of FABPs have been detected in the central nervous system. FABP-3 (heart-type FABP) has a preference for binding to omega-6 PUFA, FABP-5 (epidermal-type FABP) to saturated fatty acids, and FABP-7 (brain-type FABP) to omega-3 PUFA [10–12]. They also differ in their spatial and temporal expression in the developing and mature brain. Both FABP-5 and -7 are predominantly expressed in the prenatal and perinatal brain, mainly in developing neurons and glial cells, and appear to be involved at the early stages of cell maturation and differentiation as well as cell migration [10, 13]. FABP-3 is suggested to be expressed in the postnatal brain and to participate in neurite and synapse formation [14]. FABP-3 may thus have a role in neuronal synapse formation in the mature brain, and could therefore be hypothesized to play a role in neurodegenerative diseases. In support of this, increased cerebrospinal fluid (CSF) levels of FABP-3 have been observed in patients with neurodegenerative diseases, such as AD [15–18], Parkinson’s disease with dementia, and Lewy body disease [19] as well as in vascular dementia (VaD) [15, 18].
Although FABP-3 is proposed to be involved in the pathophysiology of dementia, its role as a valid CSF biomarker for different subtypes of dementia is uncertain. Its temporal association with the established CSF biomarkers of AD total tau (T-tau), hyperphosphorylated tau (P-tau181), and amyloid-β (Aβ42) needs to be further investigated to understand the relationship between lipid dysmetabolism and the AD process. Population-based studies may yield insights into pathological mechanisms that clinical studies cannot provide by studying these markers before dementia occurs. To our knowledge there is no population-based study examining whether CSF FABP-3 predict later development of dementia in individuals free from dementia at baseline. We therefore examined the relation between FABP-3 and development of dementia in a population of older women without dementia. A further aim was to examine CSF levels of FABP-3 in relation to established AD biomarkers in a sample of women without dementia.
MATERIALS AND METHODS
Subjects
The study sample was derived from the Prospective Population Study of Women (PPSW), a population-based survey in Gothenburg, Sweden [20]. The sample was obtained from the Swedish Population Register, based on birth date, and included both those living in private households and in residential care. The original sample has been described in detail previously [20, 21]. Briefly, the study began in 1968-1969 and included a representative sample of women living in Gothenburg, Sweden born on certain dates in 1908, 1914, 1918, and 1922.
In 1992-1993, 837 surviving women were invited to participate in a neuropsychiatric examination, which for the purpose of the current study will be referred to as the baseline examination. Among 590 who agreed to take part, 88 (aged 70–84 years) consented to undergo a lumbar puncture. Two of these women were diagnosed with dementia at baseline and excluded from the current analyses, leaving 86 women born in 1908 (n = 3), 1914 (n = 7), 1918 (n = 33), and 1922 (n = 43). The mean age of the participants at baseline in 1992-1993 was 72.5 years (SD 3.1).
A follow-up neuropsychiatric examination was conducted in 2000, when 62 out of 71 surviving women accepted participation (response rate 87%).
After complete description of the study, written informed consent was obtained from all participants at each examination wave. For women with dementia at follow-up, close informants gave proxy consent. The study was approved by the Ethics Committee for Medical Research at the University of Gothenburg.
As previously reported [21, 22], there were no differences at baseline between the women who participated in the lumbar puncture (n = 88) and those who participated in the psychiatric exam only (n = 505) with regard to age, psychiatric illnesses, symptoms of depression, smoking status, alcohol intake, physical activity, systolic and diastolic blood pressures, body mass index, blood levels of cholesterol, high density lipoprotein, and triglycerides, age of menopause, history of angina pectoris, myocardial infarction, and diabetes, and use of a variety of medications including lipid-lowering agents, antihypertensive agents, and hormone replacement therapy. Further, no differences could be shown regarding Mini-Mental State Examination (MMSE) scores [23] (28.17 in women who participated in the lumbar puncture versus 27.79 in the women who did not; p = 0.277), and ratings of the personality traits neuroticism (p = 0.666) and extroversion (p = 0.941) as measured with the Eysenck Personality Inventory.
Procedures
The clinical examination was conducted at a geriatric outpatient department or in the participant’s home and included comprehensive social, functional, physical, neuropsychiatric, and neuropsychological examinations, as well as a close informant interview [24]. Information about medication use was collected and classified according to the Anatomical Therapeutic Chemical (ATC) Classification code [25].
Psychiatric examination and diagnostics
The baseline psychiatric examination was semi-structured and performed by psychiatrists in 1992-1993. The follow-up exam was carried out by trained psychiatric nurses. The Comprehensive Psychological Rating Scale (CPRS) [26] was used to rate psychiatric symptoms at baseline and follow-up. The neuropsychiatric exam has been described in detail previously [27]. A Swedish version of the MMSE was used to assess cognition during the neuropsychiatric examination [23]. Both the serial sevens and backwards spelling of the word “Konst” (Swedish word for art) was included in the test, but only the higher of these sub-scores was counted in the total score. The maximum score was 30 points. If the individual had a physical disability and could not complete the test, they were regarded as not testable. Zero scores were given if individual items were refused. This procedure has been described previously [28].
Dementia diagnoses
Dementia was diagnosed according to the DSM-III-R criteria [29] by geriatric psychiatrists at consensus meetings using a symptom algorithm with data from the psychiatric examination and the key informant interview, as described previously [27, 30]. Each symptom had to reach a level causing significant difficulties in everyday life. A final diagnosis was made from the combined information.
AD was diagnosed according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s disease and Related Disorders Association (NINCDS-ADRDA) criteria [31]. The criteria for VaD were similar to the criteria proposed by the National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherce et l'Enseignement en Neurosciences (NINDS-AIREN) [32]. VaD was diagnosed when there was a temporal relationship (within 1 year) between a history of acute focal neurological symptoms and signs (hemiparesis or motor aphasia) and the first symptoms of dementia. A diagnosis of mixed dementia was made when there was a history of stroke without temporal relationship with dementia (more than one year), and when both AD and cerebrovascular disease were supposed to have contributed to dementia. VaD and mixed dementia were combined as stroke-related dementia.
Dementias due to other causes (e.g., alcoholic dementia, normal-pressure hydrocephalus, vitamin B12 deficiency) were diagnosed as described previously [27].
CSF analyses
Lumbar punctures were carried out in the morning to avoid diurnal fluctuations. CSF samples (12 ml) were taken through the L3/L4 interspace and gently mixed to avoid gradient effects. The samples were immediately centrifuged at 2,000× g at room temperature for 10 min to eliminate cells and other insoluble materials, aliquoted in 1 ml portions, snap frozen at –80°C, stored at that temperature and brought in an unbroken freeze chain to the laboratory for analyses. CSF levels of FABP-3 were analyzed using the electrochemiluminescent ELISA technology (MSD ® Human FABP3 kit, Meso Scale Discovery, Gaithersburg, MD, USA). CSF Aβ42, T-tau, and P-tau181 were analyzed using a colorimetric ELISAs (INNOTEST ® β-AMYLOID (1-42), INNOTEST ® hTAU Ag and INNOTEST ® PHOSPHO-TAU (181P), respectively, Fujirebio-Europe, Ghent, Belgium). All analyses were performed on one occasion using one batch of reagents by board-certified laboratory technicians who were blinded to clinical data. Intra-assay coefficients of variation were below 10% .
Statistical analysis
Multiple binary logistic regression models were performed to test associations between CSF FABP-3, CSF Aβ42, T-tau, and P-tau181 and development of dementia, AD, stroke, and myocardial infarction during 8-year follow-up and to test the association between CSF FABP-3 and APOE ɛ4. Age was included in the model as an independent covariate. Pearson correlations were carried out to explore the relationships between CSF levels of FABP-3 and CSF levels of T-tau, P-tau181 and CSF/serum albumin ratio. Correlations between CSF FABP-3, Aβ42, T-tau, and P-tau181 as well as the biomarker correlation to change in MMSE during follow-up were assessed by Pearson correlation coefficient, as CSF biomarkers were normally distributed. Statistical tests were carried out using SPSS for Windows (version17, SPSS Chicago, IL). A two-tailed level of significance, p < 0.05 was used in all tests.
RESULTS
Among 86 non-demented women at baseline, 62 took part in the follow-up examination in 2000. During follow-up, eight women developed dementia (four AD, three VaD, one mixed VaD/AD). Baseline characteristics of the sample are given in Table 1. At baseline in 1992, the two women who had dementia had higher levels of FABP-3 than the other women (T-test, p = 0.009). These two women were then excluded from further analyses. The biomarker profile at baseline in relation to future dementia is given inTable 2.
Among non-demented women at baseline, FABP-3 correlated with T-tau (r = 0.88, p < 0.001) (Fig. 1), P-tau181 (r = 0.619, p < 0.001), and CSF:serum albumin ratio (r = 0.233, p = 0.031), but not with Aβ42 (r = –0.08, p < 0.444). In contrast, neither T-tau (r = 0.106, p = 0.329), P-tau181 (r = 0.153, p = 0.160) nor Aβ42 (r = –0.05, p = 0.649) correlated with CSF:serum albumin ratio. FABP-3 levels were not related to APOE ɛ4 (OR 2.22 CI [0.27–18.29] p = 0.459).
CSF FABP-3 in relation to dementia during follow-up
Mean CSF FABP-3 was higher in those who developed dementia (10.1 SD 5.8 ng/ml; p = 0.008) and AD (10.4 SD 6.9 ng/ml; p = 0.024) than in those who did not (7.1 SD 2.4 ng/ml). A binary logistic regression model including age as a covariate showed that higher levels of CSF FABP-3 in 1992 were related to development of dementia (OR 1.36 CI [1.05–1.76] p = 0.022) and the subtype AD (OR 1.38 CI [1.06–1.82), p = 0.019) during follow-up. There was no relationship between levels of CSF FABP-3 and the development of VaD (OR 1.10 CI 0.74–1.63] p = 0.642).
CSF Aβ42, T-tau and P-tau181 in relation to dementia during follow-up
Binary logistic regression models, including age as a covariate, were performed to examine whether CSF Aβ42, T-tau, and P-tau181 were related to dementia development and AD in 2000. Lower levels of CSF Aβ42 (OR 0.99 CI [0.98–0.99], p = 0.022), and higher levels of T-tau (OR1.003 CI [1.0004–1.006], p = 0.027), and P-tau181 (OR 1.12 CI [1.03–1.21], p = 0.010) were related to dementia development in 2000. Lower levels of CSF Aβ42 (OR 0.995 CI [0.989–0.99], p = 0.049), and higher levels of CSF T-tau (OR 1.004 CI [1.001–1.007], p = 0.018) were related to AD, while P-tau (OR 1.057 CI [0.991–1.13], p = 0.084) was not.
Further logistic regression models, including age, FABP-3, Aβ42, T-tau, and P-tau181 were performed to examine their independent associations with dementia development and AD in 2000. In this analysis, no biomarker was associated with dementia or AD. Due to multicollinearity between FABP-3, total tau and P-tau181, we performed further analyses. In the second regression analyses, age, Aβ42, T-tau, and P-tau181 were independent covariates for the development of dementia and AD. In this analysis, no biomarker was associated with dementia or AD. A third regression included age, FABP-3, and Aβ42. In this analysis, only Aβ42 was associated with future dementia (OR 0.995, [0.99–0.999), p = 0.049), while no biomarker was associated with AD.
Higher baseline concentration of CSF FABP-3 (r = –0.29, p = 0.028, n = 56), T-tau (r = –0.29, p = 0.028, n = 56), and P-tau181 (r = –0.42, p = 0.001, n = 56), and lower levels of Aβ42 (r = 0.42, p = 0.001), correlated with a decrease in MMSE between 1992 and 2000. When one individual with an extreme decrease in MMSE (over 21 points) at follow-up was excluded from the analyses, the correlations between CSF FABP-3, T-tau, or P-tau181 and change in MMSE were no longer significant (r = 0.169, p = 0.22, n = 55; r = 0.162, p = 0.24, n = 55; r = –0.09, p = 0.49, n = 55, respectively). However, as described previously 33] there was still a positive correlation between change in MMSE and the concentration of Aβ42 (r = 0.39, p = 0.003, n = 55).
Between 1992 and 2000, 12 participants developed a stroke. There was no association between CSF FABP-3 levels and stroke (OR 1.01 CI [0.81–1.26], p = 0.935) in a binary logistic regression model including age. Ten women developed a myocardial infarction until 2000. There was no association between CSF FABP-3 levels and development of myocardial infarction in a binary logistic regression model including age (OR 1.01, CI [0.77–1.32], p = 0.966).
DISCUSSION
To our knowledge, this is the first prospective population-based study to show an association between increased CSF levels of FABP-3 and subsequent development of dementia and AD in older women. In addition, in women without dementia, CSF FABP-3 was related to CSF T-tau levels but not to Aβ42, suggesting an association with neuronal damage.
Although our association between increased CSF levels of FABP-3 and subsequent development of dementia and AD were weak, they are supported by cross-sectional clinical studies reporting elevated CSF levels of FABP-3 in AD [15, 34]. FABP-3 has also been shown to predict progression from mild cognitive impairment (MCI) to AD dementia [16, 34], and from MCI to VaD 5.3 years prior to dementia onset [18]. Elevated levels have been demonstrated also in Creutzfeldt-Jakob disease [35]. FABP-3 is thus probably a marker for neurodegeneration rather than for different types of dementia.
AD seems to develop many decades before the onset of disease symptoms [36–40]. Hence, changes in CSF biomarkers reflecting the underlying pathogenesis may occur many decades before symptoms occur [37, 41]. Lower levels of CSF Aβ42 seem to be the earliest marker, while changes in T-tau come later [37, 42]. It is therefore important to examine whether there is a very early relation between CSF levels of FABP-3, and the biomarkers Aβ42, T-tau, and P-tau181 [16, 18]. We found that FABP-3 levels were highly correlated with T-tau and P-tau181, but not to Aβ42 . This is in line with findings from clinical dementia samples, where FABP-3 correlated with CSF T-tau levels and P-tau181 levels [15, 18]. We now show that these associations occur already at the preclinical stages of dementia. It has been shown that FABP-3 is expressed in neurons, maybe explaining its correlation with T-tau [43]. Further, circulating FABP-3 levels are used as indicators of tissue damage [44]. In the brain, decreased levels of FABP-3 have been found in the frontal, occipital, and parietal cortices in patients with Down syndrome and AD [45], while higher levels of CSF FABP-3 levels are associated with brain volume loss in individuals with mild AD [46], possibly supporting the hypothesis of leakage of FABP-3 into the interstitial fluid and further out into the CSF.
It has previously been established that lower CSF levels of Aβ42 and higher levels of T-tau and P-tau181 are related to AD development in patients with MCI [37, 40], which supports our findings that altered baseline Aβ42 and T-tau are associated with AD development at the follow-up investigation 8 years later. The non-significant findings for P-tau181 could most likely be attributed to the small sample size.
In the multiple regression analyses including the AD CSF biomarkers Aβ42, T-tau, and P-tau181 as independent covariates, the significant association between FABP-3 with future dementia was not seen. However, due to multicollinearity, it may not be appropriate to use regression analyses including all biomarkers. The high correlation between the biomarkers thus minimized the ability to isolate the independent effect of each, although the effect size in this model was similar to that seen in the other model. When we only examined CSF Aβ42 and FABP-3, which were not associated, CSF Aβ42 was associated with future dementia, whereas FABP-3 was not. Multicollinearity probably makes all these analyses uncertain. It also has to be emphasized that our sample most likely is too small for these extended regression analyses.
The pathophysiological mechanisms by which FABP-3 might play a role in dementia development warrant further consideration. First, one factor may be its preference for transporting ARA. Higher levels of FABP-3 might lead to higher load of ARA and its metabolites in the brain, which has been reported in AD [47]. ARA has pro-inflammatory properties as it is a precursor in the production of leukotrienes and prostaglandins. ARA has therefore been implicated in the pathogenesis of several neurological diseases, such as AD and Parkinson’s disease [48]. The accumulation of Aβ and tau within the CNS in AD results in activation of microglia and astrocytes involving a pro-inflammatory pathway that results in the release of potentially neurotoxic substances, including cytokines and prostaglandins, leading to degenerative changes in neurons [49, 50]. Second, FABP-3 is expressed during the late phase of brain development. It participates in neurite formation and synapse maturation, and supports neuronal function in adult age [51–53]. A disturbed balance in CSF FABP-3 levels might impair neuronal functioning by modulating the fluidity and permeability of cell membranes, and by altering the phospholipid content due to alterations in ARA availability. Third, the membrane lipid composition plays an essential role in the processing of amyloid. For example, the generation of Aβ42 involves sequential proteolytic cleavages of the amyloid-β protein precursor (AβPP). AβPP itself is an integral membrane protein and the membrane lipid composition affects the proteolytic processing of AβPP [4, 54]. This links Aβ production to FABP-3. Fourth, FABP-3 might confer higher risk for dementia by its suggested arteriosclerotic properties. In serum, higher FABP-3 levels may be involved in the pathogenic cascade that leads to arteriosclerosis of the coronary arteries [55, 56]. Higher CSF FABP-3 levels could thus possibly be linked to arterio/arteriolosclerosis explaining the correlation, albeit weak, between FABP-3 and CSF:serum albumin ratio in our study, possibly reflecting blood-brain barrier dysfunction. The latter finding might explain the higher CSF FABP-3 levels found in VaD [15, 17], and in patients with MCI that later convert to VaD [18]. However, it is more likely that FABP-3 is a marker of dyslipidemia that affects both the development of artherosclerosis, but more importantly, reflects neuronal degeneration and regeneration [57]. In our study, the low number of individuals who developed VaD precludes any conclusion regarding its association with FABP-3.
Serum FABP-3 has been related to acute stroke and its severity [14, 58–60], possibly arguing for a relation with ischemia and cell death. In our study, stroke had most often occurred a long time before the lumbar puncture, which may be one reason for the lack of association between CSF FABP-3 and stroke. FABP-3 and T-tau levels are also increased in patients with subarachnoidal hemorrhage, both in the acute phase and after 6 months [44]. Dementia diseases often show slow progression over a long time period in contrast to the acute brain damage found in stroke or intracranial bleedings. Alterations in CSF may thus be more relevant than alterations in serum levels for detection of differences in non-acute brain neurodegenerativedisorders. Increased serum levels of FABP-3 are strongly related to myocardial infarction [61] and persistent myocardial damage [62]. We did not find a relation between history of myocardial infarction and CSF levels of FABP-3, suggesting that higher CSF FABP-3 levels might reflect central neuronal degeneration rather than being due to peripheral leakage.
Strength and weaknesses of the study
Among the strengths of this study are the population-based sample, the comprehensive examinations and the prospective design. The women were also well characterized. However, some methodological considerations need to be addressed. First, CSF samples were stored at –80°C, and the long-term stability of the FABP-3 is unknown. However, all samples were treated in the same manner. The temperature chain was unbroken for the analyzed aliquots. Despite this, changes in FABP-3 levels after long-term storage might have occurred, but this would have diminished the chances to find statistically significant differences between women with and without dementia. Second, the small number of women with dementia at follow-up constitutes a major weakness. The number of participants with CSF data is relatively large: However, the number of cases with AD was limited due to the population-based study design, which most likely reduced statistical power. Third, neuropsychiatric exams were performed at two time points only, so that the time for conversion to dementia development is unknown. Fourth, lumbar punctures were performed at baseline only. We could thus not test for changes in CSF FABP-3 over time. Fifth, while we were unable to show differences between women with and without lumbar puncture regarding a number of health-related factors, including cognitive functioning and personality traits, women who took part in this examination may have been healthier, which might limit generalizability to the underlying population. However, this type of selection bias would be expected to decrease the likelihood of significant findings. Finally, data on the APOE ɛ4 allele was available for only 66% of participants, further reducing the sample size. As APOE ɛ4 was not related to FABP-3, we chose not to control for APOE ɛ4 in further analysis.
In conclusion, we found that higher CSF FABP-3 levels were related to development of dementia in a non-clinical population. Further studies are needed to understand the role of FABP-3 in dementia development, and to clarify whether it is a marker for the disease or is involved in the pathogenesis of AD.
Footnotes
ACKNOWLEDGMENTS
The Swedish Research Council (11267, 2005-8460, 825-2007-7462, 825-2012-5041, 2013-8717, 2013-61X-14002), Hjärnfonden, the Torsten Söderbergs Stiftelse at the Royal Swedish Academy of Sciences, Swedish Research Council for Health, Working Life and Wellfare (no 2001-2646, 2001-2835, 2003-0234, 2004-0150, 2006-0020, 2008-1229, 2004-0145, 2006-0596, 2008-1111, 2010-0870, AGECAP 2013-2300, 2013-2496, Epilife 2006-1506), Swedish Brain Power, The Alzheimer’s Association Zenith Award (ZEN-01-3151), The Alzheimer’s Association Stephanie B. Overstreet Scholars (IIRG-00-2159), The Knut and Alice Wallenberg Foundation, Sahlgrenska University Hospital (ALF),The Emil and Maria Palm Foundation, The Bank of Sweden Tercentenary Foundation, EU FP7 project LipiDiDiet, Grant Agreement N° 211696, Eivind och Elsa K:son Sylvans stiftelse, Stiftelsen Söderström-Königska Sjukhemmet, Stiftelsen för Gamla Tjänarinnor, Handlanden Hjalmar Svenssons Forskningsfond, Stiftelsen Längmanska Kulturfonden, Epilife Small Grant, Stiftelsen Demensfonden.
The above named Study Funding Organizations have not been involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
