Relationships Between Plasma Lipids Species, Gender, Risk Factors, and Alzheimer’s Disease

Abstract

Background:

Lipid metabolism is altered in Alzheimer’s disease (AD); however, the relationship between AD risk factors (age, APOE ɛ4, and gender) and lipid metabolism is not well defined.

 

Objective:

We investigated whether altered lipid metabolism associated with increased age, gender, and APOE status may contribute to the development of AD by examining these risk factors in healthy controls and also clinically diagnosed AD individuals.

 

Methods:

We performed plasma lipidomic profiling (582 lipid species) of the Australian Imaging, Biomarkers and Lifestyle flagship study of aging cohort (AIBL) using liquid chromatography-mass spectrometry. Linear regression and interaction analysis were used to explore the relationship between risk factors and plasma lipid species.

 

Results:

We observed strong associations between plasma lipid species with gender and increasing age in cognitively normal individuals. However, APOE ɛ4 was relatively weakly associated with plasma lipid species. Interaction analysis identified differential associations of sphingolipids and polyunsaturated fatty acid esterified lipid species with AD based on age and gender, respectively. These data indicate that the risk associated with age, gender, and APOE ɛ4 may, in part, be mediated by changes in lipid metabolism.

 

Conclusion:

This study extends our existing knowledge of the relationship between the lipidome and AD and highlights the complexity of the relationships between lipid metabolism and AD at different ages and between men and women. This has important implications for how we assess AD risk and also for potential therapeutic strategies involving modulation of lipid metabolic pathways.

 

 

INTRODUCTION

Alzheimer’s disease (AD) is a progressive neurodegenerative disease associated with cognitive impairment and is the most common form of dementia. Increasing age is a major risk factor for AD [1] and is driving the current increases in AD. Aging is also associated with changes in lipid metabolism in brain regions playing crucial roles in cognitive function [2, 3] which is reflected in the periphery [4].

The ɛ4 allele of the apolipoprotein E gene (APOE) is the strongest genetic risk factor for the sporadic form of AD [5]. Individuals with two copies of the ɛ4 allele have been reported to have a 14-fold increased risk of developing both late onset and early onset sporadic AD. Given the role of apolipoprotein E protein (ApoE) in lipid transport and metabolism, investigating the influences of different APOE alleles on lipid metabolism is warranted.

Recent studies have indicated that gender may be a potential risk factor for sporadic AD, and that females are more susceptible to developing sporadic AD, even after accounting for differences in longevity [6, 7]. In addition, gender may influence other risk factors, such that men will be affected differently to women [7]. We have previously reported on strong associations between many lipid species and gender [4, 8].

Lipids play an essential role in all mammalian systems. Recent lipidomic studies have identified multiple lipid classes and species associated with AD, at both preclinical and clinical stages [9, 10]. These studies have suggested that phospholipid, ceramide, and sulfatide metabolism may be involved in AD pathogenesis. Significant differences in ether lipids and plasmalogens have also been reported in AD studies which investigated both brain and serum lipid species [11–13]. While the exact mechanism(s) of AD pathogenesis are yet to be fully elucidated, considerable evidence indicates the involvement of aberrant lipid metabolism [11, 14]. In this study we sought to better define the associations of AD and its major risk factors (age, gender, and APOE) with the plasma lipidome, and in particular, define how age, gender, and APOE interact with the associations between AD and the plasma lipidome.

 

 

MATERIALS AND METHODS

Participants

The current study utilized plasma samples and data from the Australian Imaging, Biomarkers and Lifestyle (AIBL) flagship study of aging, which recruited 1,112 individuals over the age of 60 years into a longitudinal study (follow-up at 18-month intervals) for AD. In this paper, cross-sectional data to the latest available time point were used to maximize the number of participants available because at the later time points there were more dropouts. Time points were used as a covariate to remove any association based on time point.

The AIBL study was approved by the institutional ethics committees of Austin Health, St Vincent’s Health, Hollywood Private Hospital and Edith Cowan University. Written informed consent was given by all volunteers before participating in the study. Individuals with a history of non-AD dementia, schizophrenia, bipolar disorder, significant current depression, Parkinson’s disease, cancer (other than basal cell skin carcinoma) within the last two years, symptomatic stroke, uncontrolled diabetes, or current regular alcohol use exceeding two standard drinks per day for women or four per day for men, were excluded from the study [15]. Details of the selection criteria can be found in Ellis et al. [15].

 

Blood collection and storage, APOE genotyping

All study volunteers fasted for a minimum of 10 h overnight prior to blood collection. Blood draw was carried out using standard serologic techniques and fractionated as described previously by Ellis et al. [15] followed by storage at –80°C for research. Within the current study, lipid metabolites were measured in the plasma.

Apolipoprotein E (APOE) genotype was determined following extraction and purification of genomic DNA from 0.5 ml whole blood. Each sample was genotyped for the presence of the three APOE variants (ɛ2, ɛ3, and ɛ4) based on TaqMan SNP genotyping assays for rs7412 (C 904973) and rs429358 (C 3084793) as per the manufacturer’s instructions (AB Applied Biosystems by Life Technologies, Victoria, Australia).

 

Lipid extraction and lipidomic analysis

Lipid species were extracted from 10μL plasma, with the addition of an internal standard mix (Supplementary Table 3), using the single phase butanol/methanol extraction method as described previously [16]. 10μL aliquots of plasma were mixed with 100μL of 1-butanol:methanol (1:1 v/v) with 5 mM ammonium formate containing an internal standard mix. The mixture was vortexed for 10 s and sonicated for 60 min in a sonic water bath at 25°C and then centrifuged (16,000×g, 10 min, 20°C). The supernatant was transferred into a 0.2 mL glass insert with Teflon insert cap for analysis by ultra-performance liquid chromatography electrospray ionization tandem mass spectrometry.

The solvent system consisted of solvent A) 50% H2O/30% acetonitrile/20% isopropanol (v/v/v) containing 10 mM ammonium formate) and solvent B) 1% H2O/9% acetonitrile/90% isopropanol (v/v/v) containing 10 mM ammonium formate. The following mass spectrometer conditions were used; gas temperature, 150°C, gas flow rate 17 L/min, nebulizer 20 psi, Sheath gas temperature 200°C, capillary voltage 3500 V and sheath gas flow 10 L/min.

Plasma quality control (PQC) samples consisting of a pooled set of 6 healthy individuals were incorporated into the analysis at 1 PQC per 18 plasma samples. Technical quality control samples (TQC) consisted of PQC extracts which were pooled and split into individual vials to provide a measure of technical variation from the mass spectrometer only. These were included at a ratio of 1 TQC per 18 plasma samples. TQCs were monitored for changes in peak area, width and retention time to determine the performance of the LC-MS/MS analysis and the PQCs were subsequently used to align for differential responses across the analytical batches.

 

Characterization of phospholipid isomers

The fatty acid composition of phospholipids was characterized by either collision induced dissociation (CID) in negative ionization mode, for neutral or negatively charged species, or by CID of lithium adducts in positive ionization mode for positively charged species and subsequently used to infer structural composition on our chromatography. Fragmentation in negative mode enabled the identification of fragment ions corresponding to the fatty acyl species. Fragmentation of the lithium adducts in positive ion mode enabled identification of fragment ions corresponding to the loss of the fatty acyl constituents. Characterization of the phospholipid fatty acids was performed on whole lipid extracts of PQC samples. For the analysis of lithium adducts of phosphatidylcholine, alkylphosphatidylcholine, alkenylphosphatidylcholine, and sphingomyelin, the 10 mM ammonium formate in the extraction buffer and running solvents were substituted with 200μM lithium acetate, which provided the ability to monitor for lithium adducts for additional structural information.

 

Liquid chromatography mass spectrometry

Analysis of plasma extracts was performed on an Agilent 6490 QQQ mass spectrometer with an Agilent 1290 series HPLC system and a ZORBAX eclipse plus C18 column (2.1×100 mm, 1.8μm, Agilent) with the thermostat set at 60°C [8]. Mass spectrometry analysis was performed in positive ion mode, with dynamic scheduled multiple reaction monitoring (MRM). Mass spectrometry settings and MRM transitions for each lipid class are shown in Supplementary Table 3. Detailed description of the method is presented in Huynh et al. [8].

 

Data integration, batch alignment, and statistical analysis

Data integration was carried out using Agilent Mass Hunter version 8.0. Each batch was aligned using its PQC samples and log transformed prior to statistical analysis. A total of 582 lipid species were used for analysis. Statistical analysis was carried out using R version 3.4.1 for linear and logistical modelling with covariates. For linear regression, data was not mean centered and standard deviation scaled to allow the coefficients to be back transformed into percentage change for interpretation. For logistic regression, data was mean centered and standard deviation scaled. Sporadic missing values (∼5%) were observed for (BMI, total cholesterol, HDL-C, and triglycerides) in the AIBL clinical database. To avoid exclusion of entire samples due to such missing values, we utilized multiple imputations by chained equations [17] to estimate these parameters utilizing other related clinical variables (age, sex, BMI, HDL-C, total cholesterol, and triglycerides). Every time we imputed, we used a subset of these variables (age, sex, BMI, HDL-C, total cholesterol, and triglycerides).

Associations between risk factors (age, gender, and APOE ɛ4) and lipid species (as outcomes) were determined using linear regression of the latest available samples from individuals who did not develop MCI or AD. These analyses were adjusted for the other risk factors (age, gender, and APOE ɛ4) as appropriate, in addition to BMI, total cholesterol, HDL-C, triglycerides, site of collection and time point, statin treatment, and omega 3 supplementation as indicated. β-coefficients and 95% confidence intervals were converted to percentage change (percentage change =  (10∧β-coefficient–1)×100). Covariates involving statin treatment or DHA intake (fish oil, krill oil) were treated as binary variables. Correction for multiple comparison was by the method of Benjamini and Hochberg. For interaction analysis with age, gender, or APOE ɛ4, binary variables were used, where male = 1, age >75 = 1 and one or more copies of ɛ4 = 1.

Interaction analyses (2-df) tested each lipid metabolite individually on a binary risk factor (+additional covariates as stated) with AD as the outcome. For each binary risk factor (Age >75, Sex, APOE ɛ4) two models were used with the assignment of the risk factor switched between 0 and 1 for each model. This provided associations (coefficient, CI, and significance) between each lipid and AD for both risk groups (i.e., male and female) to aid in interpretation of the interaction.

 

Naming convention of lipid species

The lipid naming convention used here follows the guidelines established by the Lipid Maps Consortium and the shorthand notation established by Liebisch et al. [18, 19]. Glycerophospholipids typically contain two fatty acid chains and in the absence of detailed characterization are expressed as the sum composition of carbon atoms and double bonds (i.e., the phosphatidylcholine species, PC(38:6). However, where an acyl chain composition has been determined the naming convention indicates this (i.e., PC(38:6) is renamed to PC(16:0_22:6), where the sn1 and sn2 positions are unknown or to PC(16:0/22:6) where the sn positions are known). This is also extended into other lipid classes and subclasses.

Species separated chromatographically but incompletely characterized were labelled with an (a) or (b), for example the phosphatidylcholine plasmalogen species, PC(P-17:0/20:4) (a) and (b) where (a) and (b) represent the elution order. Triglyceride species are measured using a single neutral loss experiment and are represented as their sum composition with their neutral loss experiment (i.e., TG(56:2) [NL-18:2]).

 

 

RESULTS

Associations of plasma lipid species with risk factors for Alzheimer’s disease

We examined the associations between lipid species and risk factors for AD (age, gender, and APOE ɛ4 status) utilizing the most recent time-point sample of non-MCI/AD individuals (controls). Participant characteristics (at the time of sample) are shown in Supplementary Table 1.

Age was associated with 195 lipid species (corrected for multiple comparisons) after adjusting for gender, BMI, site of collection, and collection time-point. When further adjustment for clinical lipids (total cholesterol, HDL-C, and triglycerides) was performed, age was associated with 153 lipid species. A summary of these associations is presented in Fig. 1, and further details can be found in Supplementary Table 2. Acylcarnitine species showed the strongest association with increasing age (Fig. 1Supplementary Table 2). Significant negative associations were also observed with ethanolamine plasmalogens (Fig. 1B, Supplementary Table 2).

Fig.1

Associations between Alzheimer’s disease risk factors and plasma lipid species. Linear regression analyses of risk factors against plasma lipid species were performed adjusting for covariates as indicated. The most recent available sample of each healthy control individual were used (n = 696). A, B) Associations between plasma lipids and aging. A) Lipids with a p-value less than 3.01×10–6 (top 20) are highlighted in blue. B) Ethanolamine plasmalogens, PE(P), with p < 0.05 are highlighted in light blue. C, D) Associations between lipids and gender. C) Lipids with a p-value less than 3.25×10–13 (top 20) are highlighted in red. D) Glycerophospholipids esterified with a 22:6 fatty acid with p < 0.05 (DHA, docosahexaenoic acid) are highlighted in orange. E, F) Associations between lipids and APOE ɛ4 status. E) Lipids with a p-value less than 1.13×10–2 (top 20) are highlighted in green. F) Ethanolamine plasmalogens, PE(P), with p < 0.05 are highlighted in light green.

Associations between Alzheimer’s disease risk factors and plasma lipid species. Linear regression analyses of risk factors against plasma lipid species were performed adjusting for covariates as indicated. The most recent available sample of each healthy control individual were used (n = 696). A, B) Associations between plasma lipids and aging. A) Lipids with a p-value less than 3.01×10–6 (top 20) are highlighted in blue. B) Ethanolamine plasmalogens, PE(P), with p < 0.05 are highlighted in light blue. C, D) Associations between lipids and gender. C) Lipids with a p-value less than 3.25×10–13 (top 20) are highlighted in red. D) Glycerophospholipids esterified with a 22:6 fatty acid with p < 0.05 (DHA, docosahexaenoic acid) are highlighted in orange. E, F) Associations between lipids and APOE ɛ4 status. E) Lipids with a p-value less than 1.13×10–2 (top 20) are highlighted in green. F) Ethanolamine plasmalogens, PE(P), with p < 0.05 are highlighted in light green.

There were 385 lipid species associated with gender (299 species when covariates included clinical lipids). Sphingolipids with a d18:2 base exhibited a stronger association than other lipid species with a particularly strong association noted for the sphingomyelin, SM(d18:2/14:0) (Fig. 1C). Nearly all glycerophospholipids esterified with docosahexaenoic acid were positively associated with being female (Fig. 1D, Supplementary Table 2).

By comparison the associations between plasma lipid species and the APOE ɛ4 allele were relatively weak. There were 31 lipid species associated with APOE ɛ4 prior to correction for multiple comparisons (45 species when the analysis was also adjusted for clinical lipids). In general, the APOE ɛ4 allele was associated positively with lysophosphatidylethanolamine and acylcarnitine species, but negatively associated with phosphatidylethanolamine ether lipids including plasmalogens, PE(O) and PE(P), along with dihydroceramides, Cer(d18:0) species, and alkyldiacylglycerols, TG(O), (Fig. 1E, F, Supplementary Table 2). However, these were not significant after correction for multiple comparisons.

As APOE ɛ2 is considered to be protective against the risk of developing AD relative to APOE ɛ4, we examined the lipid associations with ɛ4 relative to ɛ2, excluding ɛ3 homozygotes. In total 127 species have an uncorrected p-value less than 0.05, 39 lipid species were associated with APOE ɛ4 after correction for multiple comparisons (Supplementary Table 2). In addition to the species described above, which typically showed stronger associations, several dihydroceramides and ceramides were negatively associated with ɛ4 (relative to ɛ2), while some cholesteryl ester species showed positive associations, other cholesteryl species showed negative associations. The phosphatidylethanolamine and lysophosphatidylethanolamine species were positively associated with ɛ4 relative to ɛ2.

 

Common lipid associations with Alzheimer’s disease and its risk factors

We examined the data for associations between lipid species and clinically diagnosed AD using logistic regression adjusting for the same covariates as for the risk factor analyses (Supplementary Table 2).

Comparing associations between lipid species and age in the control group, with lipid species and clinically diagnosed AD, 108 lipid species associated in the same direction with an uncorrected p-value less than 0.05 (i.e., increased with aging and increased in AD). From these, 47 lipid species that were negatively associated with AD were found to decrease with increasing age; whereas the other 61 lipid species positively associated with AD were found to increase with increasing age. There are also four lipid species, which associated opposingly between lipid species and age in the control group with lipid species and clinically diagnosed AD (Fig. 2 and Supplementary Table 4). In general, increasing age and AD were both found to be associated with decreases in ether lipids containing polyunsaturated fatty acids, higher levels of certain sphingolipid species (dependent on their acyl composition), increases in odd and branched fatty acids, decreases in ubiquinone and increases in species of phosphatidylethanolamine and triacylglycerol.

Fig.2

Common associations between plasma lipid species with Alzheimer’s disease, aging, gender, and APOE ɛ4 status. Lipids with uncorrected p-values < 0.05 were used for this analysis. Associations were adjusted for covariates as outlined in Supplementary Table 4. Associations of risk factors with plasma lipid species were performed on n = 696 controls. Associations of AD with plasma lipid species were performed on n = 696 controls and n = 268 AD. A) Common associations of plasma lipid species with age, APOE ɛ4, and AD. B) Common associations of plasma lipid species with age, gender, and AD.

Common associations between plasma lipid species with Alzheimer’s disease, aging, gender, and APOE ɛ4 status. Lipids with uncorrected p-values < 0.05 were used for this analysis. Associations were adjusted for covariates as outlined in Supplementary Table 4. Associations of risk factors with plasma lipid species were performed on n = 696 controls. Associations of AD with plasma lipid species were performed on n = 696 controls and n = 268 AD. A) Common associations of plasma lipid species with age, APOE ɛ4, and AD. B) Common associations of plasma lipid species with age, gender, and AD.

When examining the associations between lipid species and APOE ɛ4 allele status, and between lipid species and AD, there were 18 species that associated with both AD and APOE status. As the associations with APOE status were performed on control samples only and the associations with AD were performed adjusting for APOE status, these common lipid associations are independently associated with both outcomes. All of these species were negatively associated (Fig. 2 and Supplementary Table 4). Out of the 18 species associated with both APOE status and AD, 14 were ether lipid species.

There were 13 lipid species associated with aging, APOE ɛ4, and AD in a similar manner (i.e., all negatively associated). These species were predominantly ether lipids (Fig. 2 and Supplementary Table 4). We also performed these analyses adjusting for DHA supplementation and statin use, but this resulted in little difference to the associations (Supplementary Table 5).

Comparing associations between lipid species and gender in the control group, with lipid species and clinically diagnosed AD, 75 lipid species associated in the same direction with an uncorrected p-value less than 0.05 while 44 lipid species were associated in the opposite direction. Out of the 44 opposing species, 14 were sphingolipid species (Fig. 2 and Supplementary Table 4).

When examining the associations between lipid species and gender and between lipid species and age, 114 lipid species were associated with gender and age. 56 of the 114 lipid species were positively associated, 26 of the 114 lipid species were negatively associated, and the remaining 32 were associated in opposing direction. 15 of the positively associated lipid species were sphingolipid species and 16 of the positively associated lipid species were phosphatidylethanolamine species. 9 of the 32 opposing lipids were acylcarnitine species (Fig. 2 and Supplementary Table 4).

There were 53 lipid species associated with gender, aging, and AD. Of these 53 lipid species, 25 were positively associated, with 15 of the 53 lipid species negatively associated and the remaining 13 species were opposingly associated. 10 of the positively associated lipid species were phosphatidylethanolamine species (Fig. 2 and Supplementary Table 4).

 

Interaction of risk factors with the associations of lipid species with Alzheimer’s disease

We next tested whether lipid species were associated differently with AD; depending on age, gender, and APOE ɛ4 status. Interaction with age was tested by defining a binary cut-off at age 75, and interactions with APOE ɛ4 were tested using a binary cut for one or more ɛ4 alleles. Characteristics of this split are shown in Supplementary Table 1. There were 451 individuals (n = 390/61 controls/AD) 75 or younger and 513 individuals (n = 306/207 controls/AD) older than 75 at sample collection. Males and females had similar proportions of control and AD subjects (288/109 versus 408/159 control/AD for males and females respectively) whereas within the ɛ4+ve group, there were a much higher proportion of AD subjects (non ɛ4, 523/99 control/AD and ɛ4, 173/169 control/AD). We identified 107 lipid species, which demonstrated significant interactions with age (Fig. 3A, Supplementary Table 6). Multiple sphingolipid species showed a significant interaction with age; where species that were negatively associated with AD showed a stronger association in the younger age group (Fig. 3). Our results indicate that individuals that develop AD earlier present with a larger change in their sphingolipidome, particularly long acyl chain species thought to be synthesized by ceramide synthase 2. There was also an overall negative association with dihydroceramide seen in the younger group.

Fig.3

Interactions of age and gender with the associations between plasma lipid species and Alzheimer’s disease. Regression analysis with an interaction term identified lipid species that associated with AD differently in relation to age and gender. Analysis were adjusted for age, gender, BMI, total cholesterol, HDL cholesterol, triglycerides, site, and time-point of collection (n = 696 controls, 268 AD). A) Logistic regression examining lipid species associations with AD with an age interaction (stratified above and below 75 years of age). B) Logistic regression examining lipid species associations with AD with a gender interaction.

Interactions of age and gender with the associations between plasma lipid species and Alzheimer’s disease. Regression analysis with an interaction term identified lipid species that associated with AD differently in relation to age and gender. Analysis were adjusted for age, gender, BMI, total cholesterol, HDL cholesterol, triglycerides, site, and time-point of collection (n = 696 controls, 268 AD). A) Logistic regression examining lipid species associations with AD with an age interaction (stratified above and below 75 years of age). B) Logistic regression examining lipid species associations with AD with a gender interaction.

In both age groups, we observed negative associations between AD and DHA-esterified ether lipids. However, additional species were negatively associated with AD in the younger age group, including linoleic acid (18:2) and arachidonic acid (20:4) containing species. In contrast, lipid species containing particular omega-6 fatty acids such as adrenic acid, 22:4 n-6, and docosapentaenoic acid 22:5 n-6 showed stronger positive associations with AD in the older age group (Fig. 3 and Supplementary Table 5).

We next examined whether gender significantly interacted with the association between lipid species and AD. We found 71 lipid species with a significant gender interaction. The strongest interactions with gender were observed in species esterified with polyunsaturated fatty acids from both the omega-3 and omega-6 pathways. The association with AD for the ethanolamine plasmalogen, PE(P-16:0/20:5) had an odds ratio of 0.63 (95% CI 0.50–0.78, corrected p-value = 7.39×10–4Supplementary Table 2). When examined with an interaction term (p = 1.18×10–2) males were not significantly associated, with an odds ratio = 0.83 (95% CI 0.61–1.14), whereas females showed a stronger association, with an odds ratio = 0.49 (95% CI 0.36–0.66, corrected p-value 1.47×10–4) (Fig. 3 and Supplementary Table 6). This same effect was seen for all species with a significant interaction, as it was found that the association between lipid species and AD was markedly weaker in males than in females (Fig. 3 and Supplementary Table 6).

We next explored why polyunsaturated fatty acid-esterified phospholipids associated differently with AD, dependent on gender. Initial analysis highlighted differences in DHA containing lipid species between healthy control males and females (Fig. 1D). It was interesting to note that DHA lipid species are different between cognitively normal males and females (1D) even after additional adjustments for omega-3 supplementation and statin use (Fig. 4A), however these gender-specific differences was not observed in individuals with AD (Fig. 4C). Plotting unadjusted concentrations of ether lipids (22:6)-esterified highlighted the differences in concentrations of DHA ether lipids between cognitively normal men and women (Fig. 4B). The associations of gender with DHA-esterified ether lipids (Fig. 1D) were lost in the AD sub-cohort and adjustment for omega-3 supplementation had no further effect (Fig. 4C, D).

Fig.4

Associations of DHA containing lipid species with gender in control and Alzheimer’s disease groups. A) Linear regression of lipid species against gender, adjusting for clinical covariates, omega-3 supplementation and statin use using most recent samples of cognitively normal controls (n = 696). B) Concentrations of total PE(O) and PE(P) species esterified with a 22:6 fatty acid in cognitively normal males (n = 288) and females (n = 408), and in AD males (n = 109) and females (n = 159). p-values were obtained from a Dunn’s test after Kruskal-Wallis analysis. Black lines represent the median with 95% confidence intervals. C, D) Linear regression adjusting for clinical covariates using most recent samples of AD individuals (n = 268). C) No adjustment for omega-3 supplementation. D) Adjusted for omega-3 supplementation.

Associations of DHA containing lipid species with gender in control and Alzheimer’s disease groups. A) Linear regression of lipid species against gender, adjusting for clinical covariates, omega-3 supplementation and statin use using most recent samples of cognitively normal controls (n = 696). B) Concentrations of total PE(O) and PE(P) species esterified with a 22:6 fatty acid in cognitively normal males (n = 288) and females (n = 408), and in AD males (n = 109) and females (n = 159). p-values were obtained from a Dunn’s test after Kruskal-Wallis analysis. Black lines represent the median with 95% confidence intervals. C, D) Linear regression adjusting for clinical covariates using most recent samples of AD individuals (n = 268). C) No adjustment for omega-3 supplementation. D) Adjusted for omega-3 supplementation.

We observed very little interaction between APOE alleles and the lipid associations with AD. Only 10 lipid species had a significant interaction value. In general lipid associations were stronger for those without an ɛ4 allele (Supplementary Table 6).

 

 

DISCUSSION

In this study we have exploited the comprehensive lipidomic analysis of the AIBL cohort to examine in greater detail the relationship between plasma lipid species, AD risk factors (age, gender, and the APOE ɛ4 allele) and clinically diagnosed AD. We further explored the interaction of these risk factors with the associations between the lipidome and AD.

 

Ether lipids are associated with APOE ɛ4, aging, and Alzheimer’s disease

On examination of two major risk factors for AD, namely age and possession of APOE ɛ4 alleles, we observed common associations within the ether lipid classes, as 10 of the 13 lipid species that were independently negatively associated with aging, the ɛ4 allele, and AD, were ether lipids. Our observations are consistent with previous studies reporting lipid abnormalities in AD pathogenesis [20]. In particular a decrease in total phosphatidylethanolamine plasmalogen and a decrease in some phosphatidylcholine-plasmalogen species have been reported [9] in AD brains. The potential mechanisms of ether lipids in mediating AD risk are of great interest. It has been shown that the AβPP secretases are tightly associated with cell membranes and specifically with lipid rafts [21, 22]. In particular, the formation of Aβ is dependent on the β- and γ-secretases which are both expressed and regulated by lipid rafts; whereas in contrast, α-secretase, the enzyme that is essential for non-amyloidogenic proteolysis of AβPP, is expressed and exerts its activity via non-lipid raft membrane domains [23]. Ether lipids have been found to associate with lipid rafts [24], and Rothhaar et al. [25] identified a strong relationship between this subgroup of ether lipids—the plasmalogens (the major brain phospholipids)—and γ-secretase activity. It was found that the addition of phosphatidylcholine plasmalogens to AD brain tissue caused a reduction in γ-secretase activity [25]. Work by Grimm et al. identified dysregulation of ether lipid biosynthetic enzymes in the presence of Aβ in patients with AD [26]. In particular, alkyl-dihydroxyacetone phosphate-synthase (AGPS), a rate-limiting enzyme in plasmalogen synthesis, has been shown to be regulated by AβPP processing. Increased Aβ levels, which are observed in AD, led to peroxisomal dysfunction and reduced AGPS protein stability. This causes a reduction in AGPS protein levels and eventually a reduction in plasmalogen synthesis [26].

Another link between plasmalogens and AD comes from the finding that Aβ peptides increase phospholipase A2 activity, which is responsible for the degradation of plasmalogens [27, 28]. This may lead to a vicious cycle whereby Aβ peptides reduce plasmalogen levels and the reduced plasmalogen levels directly increase γ-secretase activity, leading to a higher production of Aβ peptides. Furthermore, oxidative stress is known to be an early event of AD pathogenesis which results in membrane damage, including the loss of plasmalogens, thereby exacerbating oxidative stress as they act as antioxidants [29], although not all studies agree on the antioxidant activity of plasmalogens [20].

Since ether lipids have been proposed to have functional roles in several aspects of metabolism that relate to the onset and progression of AD, this raises the possibility that ether lipids may be protective for AD and that the decreased levels observed with increasing age and the APOE ɛ4 allele may mediate the increased risk associated with these factors. However, further studies are required to validate these hypotheses.

 

Interactions between age and gender and plasma lipid levels, and their association with Alzheimer’s disease

Individuals that develop AD at a younger age are likely to exhibit more severe pathology and perform worse in cognitive tests [30, 31]. Several ceramide species, thought to be regulated by ceramide synthase 2, showed an association with AD; these ceramide species also showed a significant interaction with age, with the younger group (under 75 years) showing a stronger association between ceramide species and AD than the older group. Thus, the dysregulation of the sphingolipidome appears to be associated with earlier onset and a more severe form of the disease. Whether such changes in sphingolipid metabolism facilitate the pathogenesis of AD remains to be determined, as does the specific role of ceramide synthase 2 in the dysregulation of these lipid species.

Almost half of the lipid species measured in this study differed between males and females even after adjustment for multiple anthropometric measures and clinical lipids. We observed a dramatic shift in the effect size of the associations between lipid species and AD when using gender as an interaction term. In general, odds ratios for females were stronger (lower for negative associations, higher for positive associations) compared to males. This was particularly true for lipid species containing polyunsaturated fatty acid species, which were typically higher in females.

Several studies have indicated differences in MCI and AD prevalence between genders [32, 33], where females appear to have a higher risk [34]. Gender differences in mortality and survival rates in relation to AD have also been reported [35]. Potential biological mechanisms for these differences in risk may relate to differences in lipid metabolism. One early study examining dietary intake of omega-3 fatty acids and incident AD identified a significant interaction with gender [36], where the protective effect of dietary omega-3 fatty acids was predominately seen in women. It has previously been reported that conversion of DHA precursors to DHA is higher in women compared to men, resulting in higher circulating DHA levels [37–39].

Our results indicate higher DHA levels in cognitively normal females, but this gender difference is not observed in the AD subset, as both males and females show lower levels of lipid species esterified with omega-3 fatty acids relative to the control group. This contributes to the significant interaction between gender and the association of these lipid species with AD. Thus, in females, the higher levels of omega-3 esterified lipid species appear not to be protective, but rather show a greater decrease in AD compared to the decrease observed in males (i.e., resulting in no difference in absolute concentrations between males and females with AD). This would support the concept that omega-3 fatty acids are not in themselves protective against AD, and may explain the failure of intervention trials with omega-3 supplementation. Rather, we observe lower levels of omega-3 esterified lipid species in AD, which may represent reverse causation, in other words lower levels of omega-3 esterified lipid species may simply be part of the pathology, not a cause but the fact that the levels decrease more markedly in females may argue that the pathology is more aggressive in the latter. Further studies are needed to elucidate the mechanism behind these changes and differences in lipid levels.

The method employed in this current study enabled us to perform an in-depth examination of 582 lipid species and investigations into the relationship of age, gender, risk factors, and AD. Limitations acknowledged within the current study include the cross-sectional design of the study, and therefore further studies are required to validate the current findings longitudinally in an independent cohort. While this study is relatively large in comparison to many lipidomic and AD studies in the literature, statistical analyses to estimate interaction effects typically require larger studies to identify associations of similar effect size. This resulted in limited powered to identify all potential interaction effects, in particular disease associations derived from different APOE alleles. A second limitation, inherent to the design of the AIBL study, is the cohort enrichment for memory complainers and those diagnosed with a memory problem. Further validation should include a population cohort. Further, due to the size of the cohort examined in this study, we employed a high throughput method, which did not allow the complete separation of all isomeric species, which may have limited our ability to detect some associations.

 

Conclusion

We find in this study that ether lipids and plasmalogens associate with all risk factors of AD and with AD itself. While APOE ɛ4 was only weakly associated with the plasma lipidome, the inclusion of APOE ɛ2 showed stronger associations suggesting the protective effect of APOE ɛ2 may be more tightly coupled to lipid metabolism than the risk associated with the APOE ɛ4 allele. Both age and sex showed significant interactions with the associations of the lipidome with AD, although these were largely different. The interaction of age was predominantly with sphingolipids and ether lipids, while sex showed a strong interaction with the associations between omega-3 esterified lipid species and AD. This study extends our existing knowledge of the relationship between the lipidome and AD and highlights the complexity of the relationships between lipid metabolism and AD at different ages and between men and women. This has important implications for how we assess AD risk and also for potential therapeutic strategies involving modulation of lipid metabolic pathways. With further validation, lipid species have a potential to contribute as prognostic indicators of AD.

 

ACKNOWLEDGMENTS

The authors are grateful to the AIBL study participants and their families for their valuable contribution to this study.

Funding for the AIBL study was provided in part by the study partners [Commonwealth Scientific Industrial and research Organization (CSIRO), Edith Cowan University (ECU), Mental Health Research institute (MHRI), National Ageing Research Institute (NARI), Austin Health, CogState Ltd.]. The AIBL study has also received support from the National Health and Medical Research Council (NHMRC) and the Dementia Collaborative Research Centres program (DCRC2), as well as funding from the Science and Industry Endowment Fund (SIEF) and the Cooperative Research Centre (CRC) for Mental Health – funded through the CRC Program (Grant ID:20100104), an Australian Government Initiative. K.H was supported by a Dementia Australia Research Foundation Scholarship. PJM is supported by a Senior Research Fellowship from the National Health and Medical Research Council of Australia. This work was also supported in part by the Victorian Government’s Operational Infrastructure Support Program.

A provisional patent has been filed. Application number: App Number 2018901220; DEMENTIA RISK ANALYSIS; Baker Heart and Diabetes Institute, Edith Cowan University.

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/19-1304r1).

 

SUPPLEMENTARY MATERIAL

REFERENCES

[1]

Hebert LE , Weuve J , Scherr PA , Evans DA (2013) Alzheimer disease in the United States (2010-2050) estimated using the 2010 census. Neurology 80, 1778–1783.

[2]

Smidak R , Kofeler HC , Hoeger H , Lubec G (2017) Comprehensive identification of age-related lipidome changes in rat amygdala during normal aging. PLoS One 12, e0180675.

[3]

Hancock SE , Friedrich MG , Mitchell TW , Truscott RJ , Else PL (2015) Decreases in phospholipids containing adrenic and arachidonic acids occur in the human hippocampus over the adult lifespan. Lipids 50, 861–872.

[4]

Weir JM , Wong G , Barlow CK , Greeve MA , Kowalczyk A , Almasy L , Comuzzie AG , Mahaney MC , Jowett JB , Shaw J , Curran JE , Blangero J , Meikle PJ (2013) Plasma lipid profiling in a large population-based cohort. J Lipid Res 54, 2898–2908.

[5]

Corder EH , Saunders AM , Strittmatter WJ , Schmechel DE , Gaskell PC , Small GW , Roses AD , Haines JL , Pericak-Vance MA (1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261, 921–923.

[6]

Sinforiani E , Citterio A , Zucchella C , Bono G , Corbetta S , Merlo P , Mauri M (2010) Impact of gender differences on the outcome of Alzheimer’s disease. Dement Geriatr Cogn Disord 30, 147–154.

[7]

Kim S , Kim MJ , Kim S , Kang HS , Lim SW , Myung W , Lee Y , Hong CH , Choi SH , Na DL , Seo SW , Ku BD , Kim SY , Kim SY , Jeong JH , Park SA , Carroll BJ , Kim DK (2015) Gender differences in risk factors for transition from mild cognitive impairment to Alzheimer’s disease: A CREDOS study. Compr Psychiatry 62, 114–122.

[8]

Huynh K , Barlow CK , Jayawardana KS , Weir JM , Mellett NA , Cinel M , Magliano DJ , Shaw JE , Drew BG , Meikle PJ (2019) High-throughput plasma lipidomics: Detailed mapping of the associations with cardiometabolic risk factors. Cell Chem Biol 26, 71–84.e74.

[9]

Grimm MO , Grosgen S , Riemenschneider M , Tanila H , Grimm HS , Hartmann T (2011) From brain to food: Analysis of phosphatidylcholins, lyso-phosphatidylcholins and phosphatidylcholin-plasmalogens derivates in Alzheimer’s disease human post mortem brains and mice model via mass spectrometry. J Chromatogr A 1218, 7713–7722.

[10]

Cheng H , Xu J , McKeel DW Jr. , Han X (2003) Specificity and potential mechanism of sulfatide deficiency in Alzheimer’s disease: An electrospray ionization mass spectrometric study. Cell Mol Biol (Noisy-le-grand) 49, 809–818.

[11]

Goodenowe DB , Cook LL , Liu J , Lu Y , Jayasinghe DA , Ahiahonu PW , Heath D , Yamazaki Y , Flax J , Krenitsky KF , Sparks DL , Lerner A , Friedland RP , Kudo T , Kamino K , Morihara T , Takeda M , Wood PL (2007) Peripheral ethanolamine plasmalogen deficiency: A logical causative factor in Alzheimer’s disease and dementia. J Lipid Res 48, 2485–2498.

[12]

Han X , Holtzman DM , McKeel DW Jr. (2001) Plasmalogen deficiency in early Alzheimer’s disease subjects and in animal models: Molecular characterization using electrospray ionization mass spectrometry. J Neurochem 77, 1168–1180.

[13]

Guan Z , Wang Y , Cairns NJ , Lantos PL , Dallner G , Sindelar PJ (1999) Decrease and structural modifications of phosphatidylethanolamine plasmalogen in the brain with Alzheimer disease. J Neuropathol Exp Neurol 58, 740–747.

[14]

Chatterjee P , Lim WL , Shui G , Gupta VB , James I , Fagan AM , Xiong C , Sohrabi HR , Taddei K , Brown BM , Benzinger T , Masters C , Snowden SG , Wenk MR , Bateman RJ , Morris JC , Martins RN (2016) Plasma phospholipid and sphingolipid alterations in presenilin1 mutation carriers: A pilot study. J Alzheimers Dis 50, 887–894.

[15]

Ellis KA , Bush AI , Darby D , De Fazio D , Foster J , Hudson P , Lautenschlager NT , Lenzo N , Martins RN , Maruff P , Masters C , Milner A , Pike K , Rowe C , Savage G , Szoeke C , Taddei K , Villemagne V , Woodward M , Ames D (2009) The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: Methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. Int Psychogeriatr 21, 672–687.

[16]

Alshehry ZH , Barlow CK , Weir JM , Zhou Y , McConville MJ , Meikle PJ (2015) An efficient single phase method for the extraction of plasma lipids. Metabolites 5, 389–403.

[17]

White IR , Royston P , Wood AM (2011) Multiple imputation using chained equations: Issues and guidance for practice. Stat Med 30, 377–399.

[18]

Fahy E , Subramaniam S , Brown HA , Glass CK , Merrill AHJr. , Murphy RC , Raetz CR , Russell DW , Seyama Y , Shaw W , Shimizu T , Spener F , van Meer G , VanNieuwenhze MS , White SH , Witztum JL , Dennis EA (2005) A comprehensive classification system for lipids. J Lipid Res 46, 839–861.

[19]

Liebisch G , Vizcaino JA , Kofeler H , Trotzmuller M , Griffiths WJ , Schmitz G , Spener F , Wakelam MJ (2013) Shorthand notation for lipid structures derived from mass spectrometry. J Lipid Res 54, 1523–1530.

[20]

Dorninger F , Forss-Petter S , Berger J (2017) From peroxisomal disorders to common neurodegenerative diseases – the role of ether phospholipids in the nervous system. FEBS Lett 591, 2761–2788.

[21]

Ehehalt R , Keller P , Haass C , Thiele C , Simons K (2003) Amyloidogenic processing of the Alzheimer beta-amyloid precursor protein depends on lipid rafts. J Cell Biol 160, 113–123.

[22]

Cordy JM , Hussain I , Dingwall C , Hooper NM , Turner AJ (2003) Exclusively targeting beta-secretase to lipid rafts by GPI-anchor addition up-regulates beta-site processing of the amyloid precursor protein. Proc Natl Acad Sci U S A 100, 11735–11740.

[23]

Vetrivel KS , Thinakaran G (2010) Membrane rafts in Alzheimer’s disease beta-amyloid production. Biochim Biophys Acta 1801, 860–867.

[24]

Pike LJ , Han XL , Chung KN , Gross RW (2001) Lipid rafts are enriched in plasmalogens and arachidonate-containing phospholipids and the expression of caveolin does not alter the lipid composition of these domains. FASEB J 15, A20-A20.

[25]

Rothhaar TL , Grösgen S , Haupenthal VJ , Burg VK , Hundsdörfer B , Mett J , Riemenschneider M , Grimm HS , Hartmann T , Grimm MOW (2012) Plasmalogens inhibit APP processing by directly affecting γ-secretase activity in Alzheimer’s disease. ScientificWorldJournal 2012, 15.

[26]

Grimm MOW , Kuchenbecker J , Rothhaar TL , Groesgen S , Hundsdoerfer B , Burg VK , Friess P , Mueller U , Grimm HS , Riemenschneider M , Hartmann T (2011) Plasmalogen synthesis is regulated via alkyl-dihydroxyacetonephosphate-synthase by amyloid precursor protein processing and is affected in Alzheimer’s disease. J Neurochem 116, 916–925.

[27]

Sanchez-Mejia RO , Newman JW , Toh S , Yu GQ , Zhou Y , Halabisky B , Cisse M , Scearce-Levie K , Cheng IH , Gan L , Palop JJ , Bonventre JV , Mucke L (2008) Phospholipase A2 reduction ameliorates cognitive deficits in a mouse model of Alzheimer’s disease. Nat Neurosci 11, 1311–1318.

[28]

Sanchez-Mejia RO , Mucke L (2010) Phospholipase A2 and arachidonic acid in Alzheimer’s disease. Biochim Biophys Acta 1801, 784–790.

[29]

Zoeller RA , Nagan N , Gaposchkin DP , Legner MA , Lieberthal W (1999) Plasmalogens as endogenous antioxidants: Somatic cell mutants reveal the importance of the vinyl ether. Biochem J 338, 769–776.

[30]

Reid W , Broe G , Creasey H , Grayson D , McCusker E , Bennett H , Longley W , Sulway MR (1996) Age at onset and pattern of neuropsychological impairment in mild early-stage alzheimer disease: A study of a community-based population. Arch Neurol 53, 1056–1061.

[31]

Jacobs D , Sano M , Marder K , Bell K , Bylsma F , Lafleche G , Albert M , Brandt J , Stern Y (1994) Age at onset of Alzheimer’s disease: Relation to pattern of cognitive dysfunction and rate of decline. Neurology 44, 1215–1215.

[32]

Andersen K , Launer LJ , Dewey ME , Letenneur L , Ott A , Copeland JR , Dartigues JF , Kragh-Sorensen P , Baldereschi M , Brayne C , Lobo A , Martinez-Lage JM , Stijnen T , Hofman A (1999) Gender differences in the incidence of AD and vascular dementia: The EURODEM Studies. EURODEM Incidence Research Group. Neurology 53, 1992–1997.

[33]

Petersen RC , Roberts RO , Knopman DS , Geda YE , Cha RH , Pankratz VS , Boeve BF , Tangalos EG , Ivnik RJ , Rocca WA (2010) Prevalence of mild cognitive impairment is higher in men: The Mayo Clinic Study of Aging(CME). Neurology 75, 889–897.

[34]

Podcasy JL , Epperson CN (2016) Considering sex and gender in Alzheimer disease and other dementias. Dialogues Clin Neurosci 18, 437–446.

[35]

Lapane KL , Gambassi G , Landi F , Sgadari A , Mor V , Bernabei R (2001) Gender differences in predictors of mortality in nursing home residents with AD. Neurology 56, 650–654.

[36]

Morris MC , Evans DA , Bienias JL , Tangney CC , Bennett DA , Wilson RS , Aggarwal N , Schneider J (2003) Consumption of fish and n-3 fatty acids and risk of incident Alzheimer disease. Arch Neurol 60, 940–946.

[37]

Burdge GC , Calder PC (2005) Conversion of alpha-linolenic acid to longer-chain polyunsaturated fatty acids in human adults. Reprod Nutr Dev 45, 581–597.

[38]

Burdge GC , Wootton SA (2002) Conversion of alpha-linolenic acid to eicosapentaenoic, docosapentaenoic and docosahexaenoic acids in young women. Br J Nutr 88, 411–420.

[39]

Burdge GC , Jones AE , Wootton SA (2002) Eicosapentaenoic and docosapentaenoic acids are the principal products of alpha-linolenic acid metabolism in young men*. Br J Nutr 88, 355–363.

 

Total Cholesterol and All-cause Mortality by Sex and Age: A Prospective Cohort Study Among 12.8 Million Adults

Sang-Wook Yi
Jee-Jeon Yi  
Heechoul Ohrr 

Abstract

It is unclear whether associations between total cholesterol (TC) levels and all-cause mortality and the optimal TC ranges for lowest mortality vary by sex and age. 12,815,006 Korean adults underwent routine health examinations during 2001–2004, and were followed until 2013. During follow-up, 694,423 individuals died. U-curve associations were found. In the TC ranges of 50–199 and 200–449 mg/dL, each 39 mg/dL (1 mmol/L) increase in TC was associated with 23% lower (95% CI:23%,24%) and 7% higher (6%,7%) mortality, respectively. In the age groups of 18–34, 35–44, 45–54, 55–64, 65–74, and 75–99 years, each 1 mmol/L higher TC increased mortality by 14%, 13%, 8%, 7%, 6%, and 3%, respectively (P < 0.001 for each age group), for TC ≥ 200 mg/dL, while the corresponding TC changes decreased mortality by 13%, 27%, 34%, 31%, 20%, and 13%, respectively, in the range < 200 mg/dL (P < 0.001 for each age group). TC had U-curve associations with mortality in each age-sex group. TC levels associated with lowest mortality were 210–249 mg/dL, except for men aged 18–34 years (180–219 mg/dL) and women aged 18–34 years (160–199 mg/dL) and 35–44 years (180–219 mg/dL). The inverse associations for TC < 200 mg/dL were stronger than the positive associations in the upper range.

Introduction

Reduction of total cholesterol (TC) has been an integral part of public health campaigns, such as Healthy People 2020 in the US and Under 5 in Norway1,2,3, as well as cardiovascular disease (CVD) risk prediction models. This goal has primarily been supported by the success of statin trials showing that statin therapy reduced mortality from ischemic heart disease (IHD)4,5. “The lower, the better” cholesterol hypothesis has been accepted by many health professionals. However, the statin trials were mainly performed in persons at a high risk of heart disease, especially in men with manifest CVD, in whom heart disease mortality constituted approximately 50% of all deaths6.

Although disease-specific morbidity and mortality, such as IHD mortality, have their analytical merits, all-cause mortality is arguably the most important endpoint for patients or the general population when assessing risk factors and the effectiveness of a treatment or a public health intervention for life-threatening diseases7. The target TC levels for public health interventions in the general population should be determined after careful consideration of the levels associated with the lowest mortality in the general population. The associations of TC and all-cause mortality, however, have been relatively infrequently examined, and the associations have been inconsistent: positive linear8, inverse9, U-curve10,11,12, and reverse-L-curve13,14 associations have all been found. Moreover, cholesterol levels differ by sex and age15,16,17,18. It is unclear whether and to what extent the associations of cholesterol with mortality differ by sex and age3,17.

Through a large prospective cohort study among over 12 million participants, we examined whether the association between TC levels and all-cause mortality varied by sex and age, and estimated the sex- and age-specific levels of TC associated with the lowest mortality. Additionally, detailed estimates of the mean (and median) concentrations of TC according to sex and age are presented.

Methods

Study population and follow-up

Ninety-seven percent of the Korean population receives compulsory health insurance through the National Health Insurance Service (NHIS). The Korean Metabolic Risk Factor (KOMERIT) study included 12,845,017 NHIS beneficiaries 18–99 years of age who underwent routine health examinations from 2001 to 200419. Persons (n = 26,136) with missing information on serum total cholesterol, fasting glucose, blood pressure, and body mass index (BMI) were excluded, as were 3,665 individuals with extreme anthropometric measures and another 210 with a missing date of the health examination. The final study population included 12,815,006 participants, who were followed until December 31, 2013 through the Resident Register of Korea. The authors were granted access to the anonymized data by the NHIS, without specific informed consent from the participants according to Korean law. This study was approved by the Institutional Review Board of Catholic Kwandong University with a waiver of informed consent. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cohort studies was used to guide the reporting of our study.

Data collection

Serum TC and fasting glucose were assayed using enzymatic methods. Blood pressure was measured once in a seated position using a standard mercury sphygmomanometer, and the systolic blood pressure was measured as the first Korotkoff sound. Weight and height were measured to the nearest kilogram and centimeter, respectively19. BMI was calculated by weight in kilograms divided by the square of height in meters (kg/m2). Information on smoking history, alcohol use, and known heart disease or cancers was collected via a self-administered questionnaire. A standard protocol officially registered by the Korean government was applied for health examinations and data collection. External quality assessments of clinical chemistry were regularly performed20.

Statistical analysis

Baseline TC concentrations were mainly categorized into 18 groups (mg/dL; <120, 120–129 to 270–279 in increments of 10, ≥280). The cholesterol category with the lowest mortality (220–229 mg/dL) in all participants was used as the reference. Three groups (<200 [reference, desirable], 200–239 [borderline high], and ≥240 [high]), defined according to the cut-points proposed by the National Cholesterol Education Program (NCEP) of US, were used in an additional analysis21. In the spline analysis, a restricted cubic spline transformation of TC with 5 knots (138, 170, 191, 213, and 260 mg/dL; 5th, 27.5th, 50th, 72.5th, and 95th percentiles in all participants) was used to evaluate non-linear associations.

The hazard ratios (HRs) for death were calculated using Cox proportional hazards models stratified by age (years) at baseline (18–24, 25–34, 35–44, 45–54, 55–64, 65–74, or 75–99). In the multivariable model, the following variables were adjusted for: age at baseline (continuous variable; within each age group), sex, smoking status (current smoker, former smoker, never smoker, or missing information), alcohol use frequency (none, 2 days/month-2 days/week, 3–7 days/week, or missing information), physical activity (at least once a week; yes or no), systolic blood pressure (<120, 120–139, or ≥140 mm Hg), fasting glucose (<100, 100–125, or ≥126 mg/dL), and BMI (<18.5, 18.5–24.9, 25–29.9, or ≥30 kg/m2).

The apparent optimal ranges of TC for survival were determined by a general inspection of the curvilinear association. Generally, the interval of 40 mg/dL (roughly 1 mmol/L) with the lowest risk (the lowest unweighted geometric mean of HRs in 4 consecutive TC categories in the categorical analysis, and in 5 consecutive 10-mg/dL TC levels in the spline analysis [for example, the points of 200, 210, 220, 230, 240 mg/dL]), were considered the optimal ranges.

Subgroup analyses by sex and age, as well as various categorical, spline, and linear analyses, served as sensitivity analyses. All p-values were 2-sided. All analyses used SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Ethics approval

This study was approved by the Institutional Review Board of Kwandong University (Gangneung, Republic of Korea).

Results

During a mean 10.5 years of follow-up, 454,546 men and 239,877 women died. At baseline, the participants’ mean ± SD age was 44.4 ± 14.2 years, their mean TC level was 194.2 ± 49.0 mg/dL (Table 1), and 11.2% of participants had high TC levels (≥240 mg/dL). Individuals with higher TC levels were older and had higher levels of fasting glucose, systolic blood pressure, and BMI (Table 1). People with TC ≥240 mg/dL tended to be non-drinkers and were more likely to have comorbid heart disease, stroke, or cancer. The number of individuals was highest in the TC range of 180–189 mg/dL (Table 1, Supplementary Fig. 1).

Table 1 Participants’ characteristics according to total cholesterol categories.

TC concentrations according to sex and age

Men had on average higher TC levels than women between 24–25 to 48–49 years old, while women had higher levels than men in the age ranges of 18–23 years and ≥50 years (Fig. 1, Supplementary Table S1). Among men, the mean TC levels increased from 159.0 mg/dL at 18–19 years to a maximum of 201.4 mg/dL at 50–51 years, and among women, the mean levels increased from 170.5 mg/dL at 20–21 years to a maximum of 212.4 mg/dL at 56–57 years. The decrease in TC levels after the peak values were reached was greater in men than in women. The gradient of increase in TC levels with age was steepest from 18–19 to 28–29 years in men, while it was steepest from 44–45 to 52–53 years in women (Fig. 1).

Figure 1
figure1

Mean and median concentrations of total cholesterol. To convert cholesterol from mg/dL to mmol/L, multiply by 0.02586.

Associations between total cholesterol and mortality

U-curve associations between TC levels and mortality were found in both men and women (Fig. 2). The TC range associated with the lowest mortality was 210–249 mg/dL (Supplementary Table S2). When age was further considered, U-curve associations were observed regardless of sex or age (Fig. 3), and the optimal TC range for survival was 210–249 mg/dL for each age-sex group, except for men at 18–34 years (180–219 mg/dL) and for women at 18–34 years (160–199 mg/dL) and at 35–44 years (180–219 mg/dL) (Supplementary Table S3).

Figure 2
figure2

Age-adjusted hazard ratios associated with 18 total cholesterol (TC) categories, according to sex. TC categories (mg/dL: <120, 120–129 to 270–279 by 10, ≥280, 220–229 as reference). The midpoint was used as a representative value for each TC category, except for both ends (112 and 296), for which the median of all participants was used. *Hazard ratios and 95% confidence intervals were calculated using Cox hazards models stratified by baseline age (years: 18–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, 85–99). Age at baseline was adjusted as a continuous variable within each age group. To convert TC from mg/dL to mmol/L, multiply by 0.02586.

Figure 3
figure3

Hazard ratios* associated with 18 total cholesterol (TC) categories for mortality by sex and age. TC categories (mg/dL: <120, 120–129 to 270–279 by 10, ≥280, 220–229 as reference). The midpoint was used as a representative value for each TC category, except for both ends (112 and 296), for which the median of all participants was used. *Hazard ratios and 95% confidence intervals were calculated using Cox hazards models stratified by baseline age (years: 18–24, 25–34, 35–44, 45–54, 55–64, 65–74, 75–84, 85–99), after adjustment for age at baseline (continuous variable), smoking status, alcohol use, physical activity, known history of heart disease, stroke, or cancer, body-mass index, systolic blood pressure, and fasting glucose levels. To convert TC from mg/dL to mmol/L, multiply by 0.02586.

 

In the spline analysis (Fig. 4, Supplementary Fig. 2), the TC ranges associated with the lowest mortality were approximately 200–240 mg/dL, except for men at 18–34 years (approximately 180–220 mg/dL) and for women at 18–34 years (approximately 160–200 mg/dL) and at 35–44 years (approximately 180–220 mg/dL).

Figure 4
figure4

Hazard ratios* using spline transformed total cholesterol (TC) levels for mortality by sex and age. 5 knots (138, 170, 191, 213, and 260 mg/dL) were used with 220 mg/dL as reference. *Hazard ratios and 95% confidence intervals were calculated using Cox hazards models with the same method as in Fig. 3. To convert TC from mg/dL to mmol/L, multiply by 0.02586.

 

When assuming linear associations for TC levels of 50–449, 50–199, and 200–449 mg/dL, each 39 mg/dL (1 mmol/L) increase in TC was associated with 8% lower (HR = 0.92, 95% CI = 0.917–0.922), 23% lower (HR = 0.77, 95% CI = 0.76–0.77), and 7% higher (HR = 1.07, 95% CI = 1.06–1.07) mortality, respectively (Fig. 5).

Figure 5
figure5

Hazard ratios* per each 39 mg/dL (1 mmol/L) increase in total cholesterol (TC), according to TC range and age. *Hazard ratios and 95% confidence intervals were calculated using Cox hazards models with the same method as in Fig. 3. To convert TC from mg/dL to mmol/L, multiply by 0.02586.

 

At cholesterol levels <200 mg/dL (Table 2), inverse associations were the strongest in men aged 45–54 years and women aged 55–64 years, the age group with the highest mean TC level in both sexes (Pinteraction [age] <0.001). At cholesterol levels ≥200 mg/dL, the HRs per 39 mg/dL (1 mmol/L) higher levels were highest in the youngest groups (aged 18–44 years), and lowest in the oldest group (aged 75–99) years in both sexes (Pinteraction [age] <0.001).

Table 2 HRsa per 39 mg/dL (1 mmol/L) TC increase according to sex, age, and TC range.

 

The associations were modestly stronger in men than in women at TC levels of 50–449, 50–199, and 200–449 mg/dL (Pinteraction [sex] <0.001 in each TC range), when all ages were combined. At cholesterol levels <200 mg/dL, men had stronger inverse associations than women in age groups <65 years.

Associations across standard classifications of TC

Compared to the desirable levels of <200 mg/dL (Supplementary Table S3), borderline high levels of 200–239 mg/dL were associated with a lower risk of mortality in each age-sex group except for women aged 18–34 years, while high levels of ≥240 mg/dL were associated with a decreased risk in both sexes and each age group except for women aged 18–44 years and men aged 18–34 years, in whom high levels were associated with increased mortality.

Discussion

A U-shaped relationship between TC and mortality was observed in each age-sex group. TC levels associated with the lowest mortality were 210–249 mg/dL in both sexes in all age groups, except for the youngest groups of men, aged 18–34 years (180–219 mg/dL), and women aged 18–34 years (160–199 mg/dL) and 35–44 years (180–219 mg/dL). At TC levels of 50–199 and 200–449 mg/dL, each 39 mg/dL (1 mmol/L) increase in TC was associated with 23% lower (95% CI = 23–24%) and 7% higher (6–7%) mortality, respectively. Inverse associations in the lower TC range were strongest at the ages for which the mean TC levels were highest (men aged 45–54 years and women aged 55–64 years), while positive associations in the upper TC range were strongest in the youngest ages (<45 years) in both sexes. Both the inverse associations in the lower TC range and the positive associations in the upper TC range weakened with advancing age beyond the ages with the strongest associations.

Previous cohort studies have reported inconsistent results on the shape of associations between TC and all-cause mortality, including positive linear, inverse, U-curve, and reverse-L-curve (or reverse-J-curve) associations3,8,9,10,11,13,14,17,22. Some previous studies suggested different shapes of associations by sex and age3,17. The associationbetween TC and mortality was substantially modified by age and, to a lesser degree, by sex, in our study. Our study clearly demonstrated that the shape of association is a U-curve in each sex and each age group, including those aged 75–99 years (mean age: 79.0 years), which constituted 154,321, 80,776, and 18,080 elderly people aged 75–79, 80–84, and ≥85 years, respectively. Considering the weaker effect size associated with high TC with advancing age in the elderly, it is no surprise that previous studies conducted mainly in elderly populations found generally inverse or reverse-L-curve associations13,14. Additionally, the previously reported positive associations in younger adults8, may be explained by the stronger positive associations and lower optimal range in younger ages observed in our study, combined with the higher TC concentrations and larger proportions of morbidity and mortality from heart diseases in Western populations.

The NCEP experts classified TC levels into 3 categories: <200, 200–239, and ≥240 mg/dL, as desirable, borderline high, and high levels, respectively, mainly based on the association between TC and IHD21. In the current study, however, TC levels of 210–249 mg/dL and approximately 200–240 mg/dL were associated with the lowest mortality in the categorical and spline analyses, respectively. Our study suggested that the optimal ranges for overall survival are higher than that those for IHD. Similarly, a higher optimal range for overall survival than for IHD mortality has also been reported for BMI23. In contrast, the optimal ranges for all-cause mortality and IHD mortality were similar for fasting glucose and blood pressure24,25,26,27. Cholesterol levels might be a marker of general health, rather than a marker specific for CVD28. Even within CVD subtypes, TC ranges associated with lowest risk have not been consistent. For example, for stroke, TC levels <200 mg/dL were not associated with the lowest mortality in prospective cohort studies29,30, and randomized trials have not provided clear evidence of whether lipid-lowering therapies, including statins, reduce stroke mortality6,31. Hemorrhagic stroke, respiratory diseases (especially chronic obstructive pulmonary disease), digestive diseases (especially liver disease), and several cancers have been suggested to be associated with lower TC levels10,30,32,33,34; thus, the ranges associated with lowest risk might be even higher for these diseases than those for all-cause mortality. However, we could not examine whether the associations differed by cause of death, due to data unavailability.

Reverse causality has been suggested as an explanation of higher mortality associated with low cholesterol levels. However, a long term follow-up study in a Japanese-American population showed that individuals with low cholesterol levels maintained over a 20-year period had the worst all-cause mortality, and concluded that reverse causality was unlikely to account for the higher mortality associated with low cholesterol entirely14.

Lower optimal ranges for survival at younger ages than at older ages have also been observed for BMI19, whereas consistent ranges have been found regardless of sex and age for blood pressure and fasting glucose26,27,29. Whether different proportions of cause-specific mortality by age lead to the lower optimal range at younger ages needs to be investigated.

The sex- and age- specific levels of TC in the current study of Koreans were lower than those reported in other high-income countries, including Japan, England, and the US15,16,17,35,36. The distribution of TC levels by sex and age, however, were generally similar to those of other regional and ethnic populations, although detailed information is not always available. TC levels peaked at 50–51 years in men and at 56–57 years in women, and after the peak age, the levels decreased more rapidly in men than in women. The crossover point of the mean TC levels between sexes occurred at the age of 50–51 years, exactly at the median age of menopause37. The steep decline in estrogen corresponds well to the sharp increase in TC in women that was observed around the time of menopause in the current study.

Randomized trials have provided evidence that statin therapy may lower the overall mortality risk in persons with increased cardiovascular risk, mostly due to the reduction of mortality from heart disease5,6. The evidence, however, may not be definitive enough to claim that “the lower the cholesterol, the better” for all-cause mortality reduction in the general population with relatively low heart disease risk38.

The current cholesterol guidelines are heavily based on heart disease risk and recommend a TC range of <200 mg/dL as desirable. TC range <200 mg/dL, however, may not be necessarily a sign of good health when other diseases are considered. The diseases associated with lower TC levels and potential mechanisms have not been conclusively identified. Since the inverse associations in lower TC range were stronger than the positive associations in upper TC range, identification of diseases associated with lower TC levels and further understanding of the mechanisms of such associations may help improve health outcomes in the general population. Pending more research for clarification, careful evaluation and management might increase the chance of preventing and diagnosing potentially life-threatening diseases at an earlier stage in adults with low TC levels.

A very large number of participants, the prospective nature of the study, and complete follow-up for death are clear strengths of this study. Another major strength is that the study participants were ethnically homogeneous and lived in a similar environment covered by the same health care system. Another strength is that this study estimated mortality risk associated with TC levels down to below 120 mg/dL. However, there are limitations. First, the use of lipid-lowering medication was unaccounted for. The risk associated with high cholesterol might have been underestimated. However, in Korea, IHD mortality accounted for only approximately 5% of all-cause mortality, and only 10% of people with hypercholesterolemia received lipid-lowering therapy39. Therefore, the impact of not considering medication use is likely to be modest, and the TC levels in this study generally reflect levels without lipid-lowering medications. Additionally, this study could not determine whether statin-induced low cholesterol increases mortality. Second, other lipid measures, such as low-density lipoprotein and high-density lipoprotein cholesterol levels, were unavailable. Recent dyslipidemia management guidelines are more closely focused on these sub-fractions of cholesterol, so the direct application of our findings to individual patient care might be somewhat limited. Further study is needed to determine the sex- and age-specific associations of cholesterol fractions. Third, information on cause-specific mortality was not available. Fourth, the generalizability of our findings may be affected by the fact that the study participants were homogeneously Korean. The U-curve associations may be generalized to other ethnic populations, since the shape of the associations was generally the same for each sex and each age group, despite their varying cardiometabolic risk profiles. However, some results, such as the magnitude of relative risk associated with TC and the TC range associated with the lowest mortality, may vary by ethnic groups with different distributions of cause-specific mortality and dyslipidemia-related healthcare utilization.

In conclusion, U-curve relationships between TC and mortality were found, regardless of sex and age. TC ranges associated with the lowest mortality were 210–249 mg/dL in each sex-age subgroup, except for the youngest groups of men, aged 18–34 years (180–219 mg/dL), and women aged 18–34 years (160–199 mg/dL) and 35–44 years (180–219 mg/dL). Inverse associations in the range <200 mg/dL were more than 3-fold stronger than positive associations for cholesterol levels ≥200 mg/dL, except for the youngest adults. Positive associations in the upper TC range were strongest for youngest adults and weakened with advancing age. TC levels <200 mg/dL may not necessarily be a sign of good health. Identification and proper management of diseases associated with lower TC levels might improve survival.

Data Availability

The data supporting the findings of this study are available from the NHIS, but restrictions apply to the availability of these data, which were used under license for the current study; therefore, the data are not publicly accessible.

References

  1. 1.

    Benjamin, E. J. et al. Heart Disease and Stroke Statistics-2017 Update: A Report From the American Heart Association. Circulation. 135, e146–e603 (2017).

    Article Google Scholar 

  2. 2.

    US Department of Health and Human Services. Healthy People 2020. 2020 Topics & Objective: Heart Disease and Stroke. https://www.healthypeople.gov/2020/topics-objectives/topic/heart-disease-and-stroke/objectives AccessedNovember 21, 2018.

  3. 3.

    Petursson, H., Sigurdsson, J. A., Bengtsson, C., Nilsen, T. I. & Getz, L. Is the use of cholesterol in mortality risk algorithms in clinical guidelines valid? Ten years prospective data from the Norwegian HUNT 2 study. J Eval Clin Pract. 18, 159–68 (2012).

    Article Google Scholar 

  4. 4.

    Cholesterol Treatment Trialists Collaborators, Mihaylova, B. et al. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 380, 581–90 (2012).

    Article Google Scholar 

  5. 5.

    Chou, R., Dana, T., Blazina, I., Daeges, M. & Jeanne, T. L. Statins for Prevention of Cardiovascular Disease in Adults: Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA. 316, 2008–2024 (2016).

    Article Google Scholar 

  6. 6.

    Cholesterol Treatment Trialists Collaboration, Baigent, C. et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet 376, 1670–81 (2010).

    Article Google Scholar 

  7. 7.

    PDQ Adult Treatment Editorial Board. Levels of Evidence for Adult and Pediatric Cancer Treatment Studies (PDQ(R)): Health Professional Version. 2017 May 19. Available from: https://www.ncbi.nlm.nih.gov/books/NBK65748/PDQ Cancer Information Summaries [Internet]. Bethesda (MD): National Cancer Institute (US), 2002.

  8. 8.

    Stamler, J. et al. Relationship of baseline serum cholesterol levels in 3 large cohorts of younger men to long-term coronary, cardiovascular, and all-cause mortality and to longevity. JAMA. 284, 311–8 (2000).

    CAS Article Google Scholar 

  9. 9.

    Casiglia, E. et al. Predictors of mortality in very old subjects aged 80 years or over. Eur J Epidemiol. 9, 577–86 (1993).

    CAS Article Google Scholar 

  10. 10.

    Jacobs, D. et al. Report of the Conference on Low Blood Cholesterol: Mortality Associations. Circulation. 86, 1046–60 (1992).

    CAS Article Google Scholar 

  11. 11.

    Kirihara, Y., Hamazaki, T., Ogushi, Y., Tsuji, H. & Shirasaki, S. The Relationship between Total Blood Cholesterol Levels and All-cause Mortality in Fukui City, and Meta-analysis of This Relationship in Japan. J Lipid Nutr 17, 67–78 (2008).

    CAS Article Google Scholar 

  12. 12.

    Lin, Y. C. et al. Different effect of hypercholesterolemia on mortality in hemodialysis patients based on coronary artery disease or myocardial infarction. Lipids Health Dis. 15, 211 (2016).

    Article Google Scholar 

  13. 13.

    Petersen, L. K., Christensen, K. & Kragstrup, J. Lipid-lowering treatment to the end? A review of observational studies and RCTs on cholesterol and mortality in 80+ -year olds. Age Ageing. 39, 674–80 (2010).

    Article Google Scholar 

  14. 14.

    Schatz, I. J. et al. Cholesterol and all-cause mortality in elderly people from the Honolulu Heart Program: a cohort study. Lancet. 358, 351–5 (2001).

    CAS Article Google Scholar 

  15. 15.

    Lawes, C. M. M., Hoorn, S. V., Law, M. R. & Rodgers, A. Chapter 7. High cholesterol. In: Ezzati, M., Lopez, A. D., Rodgers, A., Murray, C. J. eds. Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors. Vol. 1. (Geneva: World Health Organization, 2004).

  16. 16.

    Roth, G. A. et al. High total serum cholesterol, medication coverage and therapeutic control: an analysis of national health examination survey data from eight countries. Bull World Health Organ. 89, 92–101 (2011).

    Article Google Scholar 

  17. 17.

    Ulmer, H., Kelleher, C., Diem, G. & Concin, H. Why Eve is not Adam: prospective follow-up in 149650 women and men of cholesterol and other risk factors related to cardiovascular and all-cause mortality. J Womens Health (Larchmt). 13, 41–53 (2004).

    Article Google Scholar 

  18. 18.

    Park, J. H. et al. Effects of age, sex, and menopausal status on blood cholesterol profile in the korean population. Korean Circ J. 45, 141–8 (2015).

    CAS Article Google Scholar 

  19. 19.

    Yi, S. W., Ohrr, H., Shin, S. A. & Yi, J. J. Sex-age-specific association of body mass index with all-cause mortality among 12.8 million Korean adults: a prospective cohort study. Int J Epidemiol. 44, 1696–705 (2015).

    Article Google Scholar 

  20. 20.

    Min, W. K. et al. Annual Report on External Quality Assessment in Clinical Chemistry in Korea. J Lab Med Qual Assur. 2003(25), 1–14 (2002).

    Google Scholar 

  21. 21.

    National Cholesterol Education Program Expert Panel on Detection E. Treatment of High Blood Cholesterol in A. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 106, 3143–421 (2002).

    Article Google Scholar 

  22. 22.

    Song, Y. M., Sung, J. & Kim, J. S. Which cholesterol level is related to the lowest mortality in a population with low mean cholesterol level: a 6.4-year follow-up study of 482,472 Korean men. Am J Epidemiol. 151, 739–47 (2000).

    CAS Article Google Scholar 

  23. 23.

    Prospective Studies Collaboration, Whitlock, G. et al. Body-mass index and cause-specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet. 373, 1083–96 (2009).

    Article Google Scholar 

  24. 24.

    Emerging Risk Factors Collaboration, Sarwar, N. et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet 375, 2215–22 (2010).

    Article Google Scholar 

  25. 25.

    Lewington, S., Clarke, R., Qizilbash, N., Peto, R. & Collins, R. Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet. 360, 1903–13 (2002).

    Article Google Scholar 

  26. 26.

    Yi, S. W. et al. Low Systolic Blood Pressure and Vascular Mortality Among More Than 1 Million Korean Adults. Circulation. 133, 2381–90 (2016).

    Article Google Scholar 

  27. 27.

    Yi, S. W. et al. Association between fasting glucose and all-cause mortality according to sex and age: a prospective cohort study. Sci Rep. 7, 8194 (2017).

    ADS Article Google Scholar 

  28. 28.

    Ko, D. T. et al. High-Density Lipoprotein Cholesterol and Cause-Specific Mortality in Individuals Without Previous Cardiovascular Conditions: The CANHEART Study. J Am Coll Cardiol. 68, 2073–2083 (2016).

    CAS Article Google Scholar 

  29. 29.

    Prospective Studies Collaboration, Lewington, S. et al. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55,000 vascular deaths. Lancet 370, 1829–39 (2007).

    Article Google Scholar 

  30. 30.

    Yi, S. W., Shin, D. H., Kim, H., Yi, J. J. & Ohrr, H. Total cholesterol and stroke mortality in middle-aged and elderly adults: A prospective cohort study. Atherosclerosis. 280, 211–7 (2018).

    Article Google Scholar 

  31. 31.

    De Caterina, R. et al. Cholesterol-lowering interventions and stroke: insights from a meta-analysis of randomized controlled trials. J Am Coll Cardiol 55, 198–211 (2010).

    Article Google Scholar 

  32. 32.

    Kitahara, C. M. et al. Total cholesterol and cancer risk in a large prospective study in Korea. J Clin Oncol. 29, 1592–8 (2011).

    CAS Article Google Scholar 

  33. 33.

    Neaton, J. D. et al. Serum cholesterol level and mortality findings for men screened in the Multiple Risk Factor Intervention Trial. Multiple Risk Factor Intervention Trial Research Group. Arch Intern Med. 152, 1490–500 (1992).

    CAS Article Google Scholar 

  34. 34.

    Strohmaier, S. et al. Total serum cholesterol and cancer incidence in the Metabolic syndrome and Cancer Project (Me-Can). PLoS One. 8, e54242 (2013).

    ADS CAS Article Google Scholar 

  35. 35.

    Arai, H. et al. Serum lipid survey and its recent trend in the general Japanese population in 2000. J Atheroscler Thromb. 12, 98–106 (2005).

    Article Google Scholar 

  36. 36.

    Carroll, M. D., Kit, B. K., Lacher, D. A., Shero, S. T. & Mussolino, M. E. Trends in lipids and lipoproteins in US adults, 1988-2010. JAMA. 308, 1545–54 (2012).

    CAS Article Google Scholar 

  37. 37.

    Gold, E. B. The timing of the age at which natural menopause occurs. Obstet Gynecol Clin North Am. 38, 425–40 (2011).

    Article Google Scholar 

  38. 38.

    Leibowitz, M. et al. Association Between Achieved Low-Density Lipoprotein Levels and Major Adverse Cardiac Events in Patients With Stable Ischemic Heart Disease Taking Statin Treatment. JAMA Intern Med. 176, 1105–13 (2016).

    Article Google Scholar 

  39. 39.

    Lee, Y. H. et al. Serum cholesterol concentration and prevalence, awareness, treatment, and control of high low-density lipoprotein cholesterol in the Korea National Health and Nutrition Examination Surveys 2008–2010: Beyond the Tip of the Iceberg. J Am Heart Assoc. 3, e000650 (2014).

    PubMed PubMed Central Google Scholar 

Download references

Acknowledgements

The authors thank the staff at the Big Data Steering Department at the NHIS of Korea for providing the data and support.

Author information

Affiliations

Contributions

S.W.Y. conceived the study concept and design, acquired the data, statistically analyzed the data and wrote the first draft. S.W.Y., J.J.Y. and H.O. analyzed and interpreted the data and contributed to critical revision of the manuscript. All authors have read and approved the final submitted version of the manuscript. S.W.Y. is the study guarantor.

Corresponding author

Correspondence to Sang-Wook Yi.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

Hyperbaric Oxygen Could Improve Cognitive Ability

The hyperbaric oxygen market is growing rapidly as scientists discover more uses in preventing cognitive decline.

An innovative form of hyperbaric oxygen therapy (HBOT) may be able to improve cognitive ability in aging adults. Researchers at the Shamir Medical Center, and Tel Aviv University found for the first time in a peer-reviewed study, that HBOT improvements in cerebral blood flow could boost the cognitive performance of healthy adults.

Longevity.Technology: As the population ages, cognitive decline within otherwise healthy adults is becoming a growing problem. As solutions such as HBOT prove their effectiveness in increasing oxygen saturation, demand is likely to grow.

The main areas of improvement came in the speed with which information was processed, executive function and cognitive function, all of which decline with age. There was also a correlation between positive cognitive gains and cerebral blood flow in specific regions.

The study was designed by Professor Shai Efrati and Dr Amir Hadanny on the HBOT protocol developed over the past decade at Sagol Center. It involved a randomised controlled clinical trial of 63 healthy adults over the age of 64 who were either in a control group or received HBOT treatment.


 

“In our study, for the first time in humans, we have found an effective and safe medical intervention that can address this unwanted consequence of our age-related deterioration.”

 


 

There is a growing interest in the value of HBOT treatments. During the process, a patient breathes normal air which increases the oxygen solubility and stimulates the release of stem cells and other factors which can promote healing. It is being used in a growing number of environments around the world to treat issues such as non-healing wounds.

However, as this study shows, there is also increasing interest in the regenerative potential of HBOT. Delivering high levels of oxygenation at high pressure increases oxygen levels within tissue while targeting genes which are sensitive to both oxygen and pressure. The result is an improved metabolism in which the target genes proliferate stem cells, reduce inflammation, generate new blood vessels and repair mechanisms.

“Age-related cognitive and functional decline has become a significant concern in the Western world. Major research efforts around the world are focused on improving the cognitive performance of the so-called ‘normal’ aging population,” said Professor Efrati. “In our study, for the first time in humans, we have found an effective and safe medical intervention that can address this unwanted consequence of our age-related deterioration.” [1]

The market for HBOT is growing rapidly. According to Transparent Market Research, the market is expected to grow at 7.4% year on year and reach a valuation of $284.8 million. More affordable prices, technological advancements and improved performance, as well as the search for solutions which can arrest cognitive decline, are just some of the reasons behind the rise [2].

As technology improves, it is possible to purchase at home devices for $5,000 or less if you have room, making it increasingly attractive for older people looking for ways to reduce age related cognitive decline and other issues.

[1] https://www.sciencedaily.com/releases/2020/07/200715123143.htm
[2] https://3wnews.org/uncategorised/2976820/hyperbaric-oxygen-therapy-devices-market-poised-to-garner-maximum-revenues-by-2025/

Phil Newman
Editor-in-Chief

Phil has over 25 years of C-level management, marketing and business development expertise in Europe and North America. His creative background has helped him shape unconventional strategies for commercial growth – garnering both awards and investor ROI.

Phil has wide experience of technology transfer and the commercialisation of innovations from both private and institutional sources and this led to his interest in Longevity and the founding of Longevity.Technology.

Artificial gut aims to expose the elusive microbiome

The microbiome is a collection of trillions of bacteria that reside in and on our bodies. Each person’s microbiome is unique—just like a fingerprint—and researchers are finding more and more ways in which it impacts our health and daily lives. One example involves an apparent link between the brain and the bacteria in the gut. This brain-gut “axis” is believed to influence conditions such as Parkinson’s disease, depression, and irritable bowel syndrome. However, many studies into the brain-gut axis have stalled because of one central problem: the lack of an adequate testable model of the gut.

Current testing platforms cannot emulate the human gut accurately and cheaply enough for large-scale studies. The needs something new, which is what a team at MIT Lincoln Laboratory is tackling in a project funded through the Technology Office. Researchers there aim to create the perfect artificial gut.

“The question from the mechanical side is, how do you emulate the colon?” says Todd Thorsen, the project’s principal investigator from the Biological and Chemical Technologies Group. “Bacteria in the colon occupy lots of ecological niches.”

Thorsen is referring to the complexity of the human gut, which includes a community of 100 trillion microbes that all have specific, and sometimes clashing, needs. For example, certain types of bacteria in the gut will die in the presence of oxygen, while others need it to survive. The gut also contains both hard and soft mucus that allows different types of bacteria to grow. All of these conditions need to be mimicked in a single platform in order to properly maintain and test microbiome samples—and that’s not an easy task.

“Until now, no one has been able to culture a microbiome sample and maintain it,” says David Walsh from the Biological and Chemical Technologies Group, who led the device’s development and fabrication. “If we can maintain a culture, we can do things like add toxins and therapeutics to see how they change the culture over time.”

To address the problem, the laboratory team developed a multimaterial platform made of permeable silicon rubber and other plastics, such as polystyrene, all of which are cheap and can be rapidly prototyped. The two components of the platform emulate the essential oxygen and mucosal gradients.

The above photo (left) shows the component that controls the oxygen gradient. Air diffuses through the plastic while the blue ports allow researchers to change the local oxygen concentrations at different positions within the adjacent microculture chambers. The right photo shows the component that controls mucus, which is welled up into the device from below. Both components implement careful geometry to yield the precise conditions found in the gut.

“The final system will allow us to tackle real-world problems,” Walsh says. Those problems, in addition to unraveling the brain-gut axis, include developing resilience to current and emerging pathogens, combating biological warfare, and more.

This year, the research team is partnering with the University of Alabama at Birmingham, Northeastern University, and the University of California at San Francisco to implement their first tests of microbiome samples to study links to Parkinson’s disease. The laboratory’s role is to use the artificial gut to culture microbiome samples taken from people with and without Parkinson’s disease and test what happens when different suspected adverse influencers are added. The goal is to correlate how changes in the caused by exposure to certain toxins may induce Parkinson’s-like nerve damage.

The laboratory will also continue advancing other aspects of the project. Some examples include building a tubular core-shell origami-like gut that rolls up during assembly to emulate the colon and the surrounding vascularized tissue, and developing modeling software to predict how microbial communities might change over time.

by Anne McGovern, Massachusetts Institute of Technology

Is this proof depression can be a physical illness? Matthew Leeming says he cured his depression with anti-inflammatory drugs

Having no money is ‘sick-making’, as Evelyn Waugh’s characters used to say.

At the age of 47 I found myself living in Dubai, broke after a business venture went horribly wrong, and pole-axed with treatment-resistant depression.

I couldn’t work I felt so awful. Getting out of bed and into a taxi to a friend’s office where I pretended to occupy myself each day was an ordeal.

It was 2012 and for 20 years I’d been on a type of antidepressant called a tricyclic which, until then, had kept the depression which had only once before rendered me unable to work, at bay.

But it stopped being effective, I think through the shock of the business problems.

Trying to explain what depression feels like is difficult because it is in part an absence: an absence of feeling, an absence of self-respect, a lack of power to do anything and be optimistic.

I know of someone who had suffered depression and cancer and who said depression was worse. ‘When I had cancer, I wanted to live. But when I had depression I wanted to die,’ they said.

The brain itself has no pain receptors but I think everyone with depression would agree that the disease is the mental equivalent of pain — concentrated anguish, unrelieved negative thoughts, hatred of oneself.

But you also feel physically ill, as if you had been hit by a train. It’s not that you can’t be bothered to get out of bed: it is as physically difficult to do so, as when you have flu.

Living with depression is like carrying a large weight around. That weight is heavier in the morning perhaps because in many depressed people levels of cortisol (a stress hormone) are higher in the morning.

The cortisol levels and feelings of illness suggested to me that depression might be a physical disease. Although I didn’t know about the cortisol link at the time, when I first saw a doctor about the depression I said: ‘This is physical.’

But it meant nothing to him. He probably took this thought to be another symptom. However, 20 years later, I read about the work of Carmine Pariante, a professor of biological psychiatry — a brand new discipline — at King’s College, London, which suggested depression can indeed be an inflammatory physical illness.

‘In some cases, depression is due to normal bodily processes going wrong, the product of a malfunctioning immune system,’ Professor Pariante told me.

At the heart of his case are two well-attested observations. The first is the effects of interferon, a drug that stimulates the immune system to use inflammation to destroy the hepatitis virus lodged inside the liver.

One study Professor Pariante was involved with, which was published in the journal Neuropsychopharmacology in 2016, found that a third of patients given interferon for hepatitis treatment developed depression, and these were the patients who had the strongest inflammation.

‘This suggests that the depression in these patients might be caused by the inflammation,’ says Professor Pariante.

Second, people with rheumatoid arthritis often become depressed. Rheumatoid arthritis is an auto immune condition that occurs when the immune system perceives the body’s chemical messengers as invading bacteria and secretes an inflammatory chemical, tumour necrosis factor (TNF), in a bid to destroy them resulting in join pain and swelling.

Around 20 per cent of patients with this disease are also depressed according to Edward Bullmore, a professor of psychiatry at the University of Cambridge. Most doctors regard the depression as a response to the misery of the disease. But Professor Pariante believes the TNF causes the depression. ‘We now know that chemicals secreted by the body to signal the increased inflammation, such as TNF, can also directly affect brain cells and brain function, inducing depressive symptoms,’ he explains.

An anti-inflammatory drug called Remicade, developed to treat rheumatoid arthritis, that blocks the action of TNF has been a remarkably effective treatment for the disease he says.

In some patients it also caused an apparently miraculous lifting of the patient’s depression. The effect can be so dramatic that nurses call it the ‘Remicade high’.

Remicade — also known as infliximab — was recently trialled as an antidepressant but was only shown to be effective for patients with both depression and high inflammatory markers.

However TNF is just one of many inflammatory proteins in our bodies and according to Professor Pariante the resulting inflammation they bring causes ‘sickness behaviour’, something our body does to rid itself of damaging viruses and toxins while the lethargy and lack of motivation gives it a chance to recover.

‘Increased inflammation can affect our emotions and behaviour and induce symptoms that resemble depression, such as fatigue, malaise, aches and pains, bad mood and reduced interest in socialising,’ Professor Pariante says. ‘If you remember how you felt last time you had a really bad infection, you will recognise these symptoms. In fact, inflammatory chemicals change the function of brain areas that are important for anxiety and depression.’

If Professor Pariante is right, the inflammation theory suggests that in many cases depression is your mind telling your body you are ill when you are not — you are actually suffering inflammation.

Not all depression is caused like this but Professor Pariante estimates that 40 per cent of cases may involve inflammation.

Professor Pariante’s work turned on a mental lightbulb for me. A professor agreed that my disease was physical!

And because I was living in Dubai I was able to do something about it. That’s because you can buy Celebrex, a non-steroidal anti-inflammatory used for treating arthritis, over-the-counter there (but only on prescription in the UK). Internet research following reading about Pariante’s work produced papers showing its clinical effectiveness when used with antidepressants.

So I bought a packet. I took 200mg a day (I guessed at the dose) for a week with no effect but then one evening I took 400mg, the dose prescribed daily for arthritis. Within half an hour the ghastly feelings were leaving me, the anxiety in my stomach contracting as if after a strong drink following a bad day at work. Within two weeks I stopped taking the antidepressants — for the first time in 20 years. I felt normal.

Chekhov (a doctor as well as a writer) once said ‘If many remedies are prescribed for an illness, you may be certain that the illness has no cure’ and there are certainly a huge number of antidepressants on the market, none of which are universally effective. But this could be because depression has a number of causes, one of which it seems is inflammation.

Professor Pariante says that ‘depression could be like a fever — a symptom of a variety of underlying pathologies’.

So what does this mean for treatment? Professor Edward Bullmore in his book The Inflamed Mind predicts that future research will involve investigating drugs like Remicade and Celebrex rather than looking for new antidepressants in the mould of Prozac.

But despite my positive experience with anti-inflammatories Professor Bullmore warns: ‘Doctors and psychiatrists will want to see positive clinical trial data before recommending anti-inflammatory treatment for depression.

‘It is worth remembering that all anti-inflammatory drugs, including Celebrex, have side-effects and it is not advisable to start taking them until the evidence for therapeutic benefit is clearer. Hopefully, there will be further progress in the next few years to get us to that point but the current situation is still a scientific work in progress.’

I don’t think patients should suffer while there are effective drugs much less dangerous than lithium — which is what is generally prescribed for treatment-resistant depression — available.

If Professor Pariante is right, people with depression may gain more than a new treatment. We may receive sympathy. The inflammation theory provides depressed people with an acknowledgement that they are physically ill and can’t just pick up their mat and walk.

By MATTHEW LEEMING FOR THE DAILY MAIL – PUBLISHED: 17:42 EDT, 23 September 2019 | UPDATED: 03:27 EDT, 24 September 2019

In Tiny Doses, An Addiction Medication Moonlights As A Treatment For Chronic Pain

Lori Pinkley, a 50-year-old from Kansas City, Mo., has struggled with puzzling chronic pain since she was 15.

She’s had endless disappointing visits with doctors. Some said they couldn’t help her. Others diagnosed her with everything from fibromyalgia to lipedema to the rare Ehlers-Danlos syndrome.

Pinkley has taken opioids a few times after surgeries but says they never helped her underlying pain.

“I hate opioids with a passion,” Pinkley says. “An absolute passion.”

Recently, she joined a growing group of patients using an outside-the-box remedy: naltrexone. It is usually used to treat addiction, in a pill form for alcohol and as a pill or a monthly shot for opioids.

As the medical establishment tries to do a huge U-turn after two disastrous decades of pushing long-term opioid use for chronic pain, scientists have been struggling to develop safe, effective alternatives.

When naltrexone is used to treat addiction in pill form, it’s prescribed at 50 mg, but chronic-pain patients say it helps their pain at doses of less than a tenth of that.

Low-dose naltrexone has lurked for years on the fringes of medicine, but its zealous advocates worry that it may be stuck there. Naltrexone, which can be produced generically, is not even manufactured at the low doses that seem to be best for pain patients.

Instead, patients go to compounding pharmacies or resort to DIY methods — YouTube videos and online support groups show people how to turn 50 mg pills into a low liquid dose.

Some doctors prescribe it off-label even though it’s not FDA-approved for pain.

University of Kansas pain specialist, Dr. Andrea Nicol has recently started prescribing it to her patients, including Pinkley. Nicol explains that for addiction patients, it works by blocking opioid receptors — some of the brain’s most important feel-good regions. So it prevents patients from feeling high and can help patients resist cravings.

At low doses of about 4.5 mgs, however, naltrexone seems to work completely differently.

“What it’s felt to do is not shut down the system, but restore some balance to the opioid system,” Nicol says.

Some of the hype over low-dose naltrexone has included some pretty extreme claims with limited research to back them, like using it to treat multiple sclerosis and neuropathic pain or even using it as a weight-loss drug.

In the past two years, however, there’s been a big increase in new studies published on low-dose naltrexone, many strengthening its claims as a treatment for chronic pain, though most of these were still small pilot studies.

Dr. Bruce Vrooman, an associate professor at Dartmouth’s Geisel School of Medicine, was an author of a recent review of low-dose naltrexone research. Vrooman says that when it comes to treating some patients with complex chronic pain, low-dose naltrexone appears to be more effective and well-tolerated than the big-name opioids that dominated pain management for decades.

“Those patients may report that this is indeed a game changer,” Vrooman says. “It may truly help them with their activities, help them feel better.”

So how does it work? Scientists think that for many chronic pain patients, the central nervous system gets overworked and agitated. Pain signals fire in an out-of-control feedback loop that drowns out the body’s natural pain-relieving systems.

They suspect that low doses of naltrexone dampen that inflammation and kick-start the body’s production of pain-killing endorphins — all with relatively minor side effects.

Despite the promise of low-dose naltrexone, its advocates say few doctors know about it.

The low-dose version is generally not covered by insurance, so patients typically have to pay out of pocket to have it specially made at compounding pharmacies.

Advocates worry that the treatment is doomed to be stuck on the periphery of medicine because, as a 50-year-old drug, naltrexone can be made generically.

Patricia Danzon, a professor of health care management at the Wharton School at the University of Pennsylvania, explains that drug companies don’t have much interest in producing a new drug unless they can be the only maker of it.

“Bringing a new drug to market requires getting FDA approval and that requires doing clinical trials,” Danzon says. “That’s a significant investment, and companies — unsurprisingly — are not willing to do that unless they can get a patent and be the sole supplier of that drug for at least some period of time.”

And without a drug company’s backing, a treatment like low-dose naltrexone is unlikely to get the big promotional push out to doctors and TV advertisements that have turned drugs like Humira or Chantix into household names.

“It’s absolutely true that once a product becomes generic, you don’t see promotion happening, because it never pays a generic company to promote something if there are multiple versions of it available and they can’t be sure that they’ll capture the reward on that promotion,” Danzon says.

The drugmaker Alkermes has had huge success with its exclusive rights to the extended-release version of naltrexone, called Vivitrol. In a statement for this story, the company says it hasn’t seen enough evidence to support the use of low-dose naltrexone to treat chronic pain and therefore is remaining focused on opioid addiction treatment.

Pinkley says she is frustrated that there are so many missing pieces in the puzzle of understanding and treating chronic pain, but she, too, has become a believer in naltrexone.

She has been taking it for about a year now, at first paying $50 a month out of pocket to have the prescription filled at a compounding pharmacy. In July, her insurance started covering it.

“I can go from having days that I really don’t want to get out of bed because I hurt so bad,” she says, “to within a half-hour of taking it, I’m up and running, moving around, on the computer, able to do stuff.”

This story is part of NPR’s reporting project with KCUR and Kaiser Health News.

This story is part of NPR’s reporting project with KCUR and Kaiser Health News.

Drug Find Turns Aging into Thing of the Past

A cocktail of drugs has been found to reverse a critical element of the ageing process for the first time.

Scientists said a clinical trial, carried out at Stanford University in California, suggested that growing old could one day become a treatable condition.

The study involved nine men aged 51 to 65 who took three existing drugs — a growth hormone and two diabetes medicines — for one year. The drugs appeared to alter chemical compounds attached to their DNA, reversing changes that accumulate over time.

The effect was equivalent to taking an average of two and a half years off their biological age, the researchers said. The subjects’ defences against infection and cancer also appeared to be boosted.

Steve Horvath, of the University of California, Los Angeles, said: “I was very surprised. I did not think it was possible to find age reversal. Our study is only a first step that demonstrates feasibility, but it suggests that a cocktail of relatively safe substances can already achieve what appeared to be a distant dream from science fiction novels.”

The work focuses on “epigenetic” changes inside cells, which involve chemicals that latch on to portions of DNA. These act like switches, controlling the activity of individual genes.

Epigenetic changes that accumulate with age appear to make people more vulnerable to diseases such as cancer.

Professor Horvath has developed techniques that can assess a person’s epigenetic status to predict with a high degree of accuracy how old they are and how long they have to live. People who smoke, for instance, are likely to be substantially older in epigenetic terms than their chronological age.

The main aim of the trial, which was carried out at the Stanford medical centre in California, was to rejuvenate the thymus gland, which plays an important role in fighting infection.

It was already thought that growth hormone could stimulate thymus regeneration but it can also promote diabetes, so the cocktail of drugs used in the trial included two widely used anti-diabetic medicines, dehydroepiandrosterone (DHEA) and metformin. The reversal of the ageing process was discovered as a side-effect.

Gregory Fahy, lead author of a study published in the journal Aging Cell, said: “The implication is that ageing may be a treatable condition.”

Professor Horvath said: “Our study results strongly suggest that these subjects are in a better shape to fight off infections because their thymus and immune system were in much better shape than before. The treatment actually lowered the risk of cancer.”

The researchers stress that the study was small and with no placebo control arm. A trial with 100 subjects is planned.