IL-1β, IL-6, TNF- α and CRP in Elderly Patients with Depression or Alzheimer’s disease: Systematic Review and Meta-Analysis

We carried out systematic review and meta-analysis to evaluate whether peripheral levels of pro-inflammatory markers including Interleukin-1 beta (IL-1β), Interleukin-6 (IL-6), Tumor Necrosis Factor-α (TNF- α) and C-Reactive Protein (CRP) are significantly higher in elderly with depression and Alzheimer’s disease. We searched Pubmed, PsycINFO and Embase, and thirty-four relevant studies (2609 with Depression, 1645 with Alzheimer’s disease and 14363 Controls) were included. Compared with controls, IL-1β (pooled standardized mean difference [SMD]: 0.642; 95% confidence interval [CI]: 0.078–1.206; significant heterogeneity: I2 = 86.28%) and IL-6 (pooled SMD: 0.377; 95% CI: 0.156–0.598; significant heterogeneity: I2 = 88.75%) were significantly elevated in depression. There was no difference in TNF-α (p = 0.351) and CRP (p = 0.05) between those with depression and controls. Compared with controls, IL-1β (pooled SMD: 1.37, 95% CI: 0.06–2.68, significant heterogeneity: I2 = 96.01%) was significantly elevated in Alzheimer’s disease. There were no differences in IL-6 (p = 0.138), TNF-α (p = 0.451) and CRP (p = 0.07) between elderly with Alzheimer’s disease and controls. After Bonferroni adjustment, only IL-6 remained significantly higher in depression. Elderly with depression have higher IL-6 than controls, while those with Alzheimer’s disease did not have higher peripheral inflammatory markers.

Inflammation may play a pathological role in psychiatric disorders in elderly because a pro-inflammatory state is associated with ageing. Aging facilitates the pro-inflammatory state by disrupting peripheral immune system leading to excessive innate immune activity with release of pro-inflammatory cytokines and decrease in anti-inflammatory molecules 30 . The disruption in the periphery-central nervous system (CNS) immune communication via the blood-brain barrier (BBB) partly attributed by the peripheral injury further contributes to the pro-inflammatory shift of CNS. There is also altered discordant CNS response to inflammation with the upregulation of genes controlling inflammatory processes, as demonstrated by an increase in pro-inflammatory gene expression in late life of mice 31 and humans 32 . Microglia and astrocytes are the two main cell components of the CNS that play a pivotal role in the central neuro-inflammatory process responsible for neuro-biological hemostasis. With the pro-inflammatory shift of the aging CNS, there is proliferation of both activated microglia and astrocytes resulting in continuous production of pro-inflammatory cytokines 33 . This process is further amplified in depression and Alzheimer's disease, whereby chronic stress 34 and amyloid beta (Aβ) 35 respectively disrupts the microglia activity, leading to further release of the pro-inflammatory cytokines, interleukins and chemokines. In addition, depression and Alzheimer's disease cause alteration in the BBB, allowing permeability of the peripheral inflammatory cells into the CNS 36 , further enhancing a state of neuro-inflammation with consequent tissue damage. For depression, Martinez-Cengotitabengoa et al. 37 performed a systematic review and found that elevation of peripheral IL-8, IL-6, and TNF-α could indicate vulnerability to late-life depression. This review was based on qualitative review of 6 studies and the findings require further verification by a meta-analysis. For all-cause dementia, a meta-analysis found that elevation of peripheral IL-6 and CRP levels were associated with increased risk of developing dementia 38 . Nevertheless, another meta-analysis found that there was no significant difference in mean CRP levels between elderly with and without Alzheimer's disease 39 . The role of TNF-α as causative and or as contributor to pathogenesis of dementia remains inconclusive 40,41 .
Hitherto, no meta-analysis has specifically compared the peripheral levels of proinflammatory markers including IL-1β, IL-6, TNF-α and CRP between elderly with depression or Alzheimer's disease and controls without any psychiatric disorder. Convergent evidence of higher peripheral IL-1β, IL-6, TNF-α and CRP levels would support the psychoneuroimmunology theory as contributory to depressive syndromes and Alzheimer's disease in elderly. Therefore, we conducted a systematic review, meta-analysis and meta-regression. Our hypothesis was that elderly with depression or Alzheimer's disease had higher levels of IL-1β, IL-6, TNF-α and CRP than controls without psychiatric disorders. We aimed to identify factors that could influence the peripheral levels of these inflammatory markers.

Methods
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline for conducting and reporting meta-analysis and systematic reviews 42 .
Search Strategy. A comprehensive literature search was conducted through the major databases of PubMed, PsycINFO and Embase from inception to March 16, 2017. Searches including the following broad terms: cytokines, interleukins (interleukin, IL*, IL-1β, IL-6), tumor necrosis factor (TNF, tumor necrosis factor), c-reactive protein (CRP, c-reactive protein), elderly (late onset, late life, geriatric, elderly), mood disorders (depress*, depression, depressive, depressed, major depressive disorder, bipolar, manic-depressive) and Alzheimer's disease (Alzheimer's disease, Alzheimer, dementia, Alzheimer's dementia, AD) where * indicates truncation 43 . A total of 2478 titles and abstracts were shortlisted, and references of the retrieved trials and review articles were also searched.
Data extraction. The following information was extracted from each article: family name of the first author, country, year of publication, sample size (number of controls and number of elderly with depression or dementia), levels of pro-inflammatory cytokines or CRP, measure of depression or dementia, mean age, proportion of female gender, duration of education, severity of depression or dementia, proportion of diabetes, cerebrovascular disorder and cardiovascular disease. All information was recorded on a standardized data collection form and was checked by the first and last author. The quality of the 18 studies was appraised using adaptations of the Newcastle-Ottawa cohort scale for cross-sectional and case-control studies, which assessed the selection of study participants, comparability of outcome groups and assessment of outcome measures.
Selection criteria. The two authors (AN and RCH) independently reviewed all abstracts and articles to evaluate whether the study would meet criteria for inclusion in the meta-analysis. If there was any disagreement between the authors about the inclusion of a study, a third author (MWZ) reviewed the study and made a decision. Studies were included if they fulfilled the following criteria: (1) Compared the blood levels of IL-1β, IL-6, TNF-α and CRP between elderly patients with depression or Alzheimer's disease versus controls; (2) Patients with depression, Alzheimer's disease and controls were at least 60 years old; (3) Study had sufficient information to calculate effect size. Studies were excluded if they: (1) studied patients with depression or Alzheimer's disease who were younger than 60 years; (2) were animal studies; (3) did not measure levels of IL-1β, IL-6, TNF-α and CRP in SCIeNTIfIC REPoRtS | (2018) 8:12050 | DOI:10.1038/s41598-018-30487-6 the blood; (4) did not include a control group; (5) had a sample size of less than 30. If two studies were conducted in the same city within 5 years of publication, we would remove one of the studies that had smaller sample size to avoid overlapping of participants [44][45][46] . A total of 2444 articles were rejected because they did not meet the inclusion criteria. Thirty-four relevant articles were extracted and included for this meta-analysis.
Statistical analyses. All statistical analysis was performed using the "metafor" function in R based on DerSimonian-Laird random-effects model 47 and based on methods established by previous study comparing levels of biological markers between a disease group and controls without the disease 48 . The random-effects model was used because it assumes varying effect sizes between studies, because of differing study design and study population 49 . Differences in mean levels of IL-1β, IL-6, TNF-α and CRP between elderly who had Alzheimer's disease or depression and controls were calculated using a standardized mean difference based on random-effect model. Random-effect model attempted to generalize findings beyond the included studies by assuming that the selected studies are random samples from a larger population 50 . Test of heterogeneity were conducted with the Q statistic that is distributed as a X 2 variate under the assumption of homogeneity of effect sizes 43 . Between-study heterogeneity was assessed with the I 2 statistic. As a guide, I 2 values of 25% may be considered low, 50% as moderate, and 75% high 3,6 . Regression test was performed to examine publication bias if there are 10 or more studies for the outcome. Bonferroni adjustment for multiple testing in meta-analysis produced a rejection p-value of 0.05 divided by the total number of outcomes 51 . In this meta-analysis, there were 4 outcomes (IL-1β, IL-6, TNF-α and CRP). As a result, the rejection p value was 5/4 = 0.0125.
For results with considerable heterogeneity, meta-regression was performed to identify demographic and disease-related variables which might contribute to the heterogeneity if there were 5 or more studies with the variables. The regression coefficients and the associated z values and p values were reported in the meta-regression analysis 48 . Patient-related factors included mean age, gender, years of education, severity of depression or Alzheimer's disease, history of diabetes, hyperlipidemia, coronary heart disease 52 .

Results
From an initial 2,478 potentially relevant articles, we assessed 176 full-text articles for eligibility. Finally, we included 34 articles in our analysis (Fig. 1). These studies comprising 2609 elderly with depression, 1645 elderly with Alzheimer's disease and 14363 controls.

Meta-Analysis for Elderly with and without Depression.
Bonferroni adjustment for multiple testing produced a rejection p-value of 0.0125. This would mean that the outcome involving peripheral levels of IL-1β was no longer statistically significant. However, the outcome involving peripheral levels of IL-6 remained statistically significant.

Discussion
In this meta-analysis, we found that elderly with depression had significantly higher peripheral levels of IL-1β (p = 0.026), IL-6 (p < 0.001) but not TNF-α (p = 0.351) and CRP (p = 0.05). After Bonferroni adjustment, only peripheral levels of IL-1β remained significantly higher in elderly with depression than controls. However, elderly with Alzheimer's disease only had significantly higher peripheral levels of IL-1β (p = 0.047) but not IL-6 (p = 0.138), TNF-α (p = 0.735) or CRP (p = 0.0712). After Bonferroni adjustment, none of the above inflammatory markers remained significant. Heterogeneity was noted in the studies for depression and Alzheimer's disease. Factors such as age range of the elderly, proportion of gender in subjects, severity of the condition (depression or Alzheimer's disease), presence and extent of medical comorbidities, study design and methodical variance which includes the types of assay kits used to measure cytokines, may contribute to the observed heterogeneity. Nevertheless, it is not feasible to match for all these factors in the meta-analysis. In this meta-analysis, there was no publication bias. Our results with Bonferroni adjustment are different from a previous meta-analysis which was based on 7 studies and found that higher levels of IL-6 and CRP were associated with increased risk of all-cause dementia 38 . Furthermore, this previous meta-analysis did not study IL-1β and TNF-α 38 . Ravaglia et al. 88 reported that peripheral inflammatory markers are associated with increased risk of vascular dementia but not Alzheimer's disease. This postulation might explain why this meta-analysis found that all peripheral inflammatory markers were not increased in Alzheimer's disease. Many previous studies examining blood cytokine levels and cognitive impairment were also inconclusive in results 89 and this could possibly be attributed by lack of longitudinal data on cytokine expression, inadequate methodological standardization and standardization of patient characteristics. Further prospective studies are required to study the role of each proinflammatory cytokine in different types of dementia and correlate with clinical features, neuroimaging and presence of beta amyloid and neurofibrillary tangles. Our results suggest that peripheral pro-inflammatory marker IL-6 is more likely to be increased in depression as compared to Alzheimer's disease in elderly. The increased peripheral IL-6 levels in elderly with depression indicate vascular risk factors and atherosclerosis 59 which can contribute to depression.  Primary prevention strategies targeting metabolically mediated comorbidity (e.g., cardiovascular disease) may prevent depression in elderly 90 . Previous study revealed the potential therapeutic effects of antidepressant fluoxetine by reducing central and peripheral levels of IL-6 in the alleviation of depressive symptoms 91 .
In this meta-analysis, elderly with depression or Alzheimer's disease were found to have significantly higher peripheral IL-1β levels before Bonferroni adjustment. IL-1β is produced by macrophages, endothelial cells, and astrocytes 59 . IL-1β was found to be a significant risk factor for the development of depressive symptoms 92 . IL-1β crosses the blood brain barrier 93 and subsequently alters the HPA-axis 11 . IL-1β reduces the re-uptake of serotonin from the synaptic cleft resulting in depressive symptoms 94 . For Alzheimer's disease, Griffin et al. 95 found that IL-1β plays an important role in causing neurodegenerative changes. IL-1β drives the synthesis of β-amyloid precursor protein, resulting in the production and deposition of β-amyloid plaques in the brain of patients with Alzheimer's disease 95,96 . It is also involved in tau phosphorylation 97 , a key pathogenic process in Alzheimer's disease 7 .
IL-6 is produced by macrophages, T-lymphocytes, endothelial cells, smooth muscle cells and adipocytes 59 . IL-6 modulates the HPA-axis which stimulates CRH, ACTH and cortisol 98 . IL-1β and IL-6 could also work together and synergistically induce systemic immune response, leading to psycho-neuro-immunological changes in patients suffering from depression 98 . IL-6 also induces production of indoleamine 2,3-dioxygenase (IDO), leading to a decrease in tryptophan and the production of tryptophan catabolites (TRYCATs), and this is associated with depressive symptoms 99 . In contrast, peripheral IL-6 level was not significantly higher in elderly with Alzheimer's disease as compared to controls. This finding could be supported by previous research which found that elevations observed in peripheral IL-6 levels preceded the onset of Alzheimer's disease but not during the course of Alzheimer's disease 70 .
CRP is an acute-phase protein of hepatic origin that increases following IL-6 secretion by macrophages and T cells. In this meta-analysis, there was no significant difference in peripheral CRP levels in elderly with depression and Alzheimer's disease as compared to controls before and after Bonferroni's correction. Eriksson et al. 100 proposed that increased peripheral levels of CRP in mid-life played a more detrimental role leading to Alzheimer's disease but not in older people. Our findings revealed that peripheral CRP level was not increased but IL-1β was significantly raised in elderly with Alzheimer's disease as compared to controls before Bonferroni adjustment. Licastro et al. 86 postulated that the IL-1β increment in elderly with Alzheimer's disease might be due to central immune response.
TNF-α is produced by macrophages, adipocytes and astrocytes 70 . In this meta-analysis, peripheral levels of TNF-α were not significantly increased in elderly with depression as well as Alzheimer's disease as compared to controls before and after Bonferroni adjustment. Firstly, our findings might suggest that peripheral TNF-α is a less sensitive marker as compared to IL-1β and IL-6. Secondly, our findings support previous postulation that  the levels of peripheral TNF-α was found to be significantly lower in mild to moderate Alzheimer's disease as compared to severe Alzheimer's disease 101 . When we included participants with different severity of Alzheimer's disease, the peripheral TNF -α levels were not significantly higher than controls.
Limitations. This meta-analysis has several limitations. Firstly, one major limitation is lack of repeated measurements for inflammatory markers in elderly with depression and Alzheimer's disease in this meta-analysis. We cannot ignore the limitations of performing a meta-analysis on cross-sectional studies. The causality of high levels of peripheral inflammatory markers in elderly with depression cannot be established 102 . Based on the current study, it is not possible to delineate whether inflammation plays a causal role in depression or dementia, or if inflammatory processes in cardiovascular diseases that are common in the elderly lead to increased depressive symptoms in elderly or vice versa. Secondly, this meta-analysis focused on peripheral inflammatory markers which might not reflect inflammatory process in the central nervous system.

Conclusion
In conclusion, our meta-analysis found that elderly with depression had elevated peripheral levels of IL-1β, IL-6 but not CRP and TNF-α, and elderly with Alzheimer's disease only had increased peripheral levels of blood IL-1β. After correction for multiple testing, only peripheral levels of IL-6 remained significantly higher in elderly with depression. The involvement of peripheral IL-6 in elderly with depression reflects a high proportion of metabolic comorbidities which are modifiable. Further studies are necessary to evaluate the association between high peripheral IL-6 levels and the subsequent risk of developing depression in old age.