Liver and inflammatory biomarker relationships to depression symptoms in healthy older adults

Introduction: Early identification and management of physical and mental illness is vital to maintain quality of life as we age. Markers of peripheral inflammation and liver function show elevations with aging, and are also associated with depression symptoms, suggesting a similar pattern in both aging and clinical groups. Methods: The current study examined the relationship between such markers and measures of depression/ negative mood in 284 healthy older adults using data from the Australian Research Council Longevity Inter- vention (ARCLI). Baseline data in adults aged 60 – 75 included mood symptoms via Profile of Mood States and Beck Depression Inventory II, and peripheral inflammatory (TNF- α , IL-6, hs-CRP) and liver markers (GGT, ALT, AST, AST:ALT ratio) derived from blood samples. Results: The inflammation and liver enzyme relationship significantly predicted mood symptoms scores. Results showed that a significant relationship between C-reactive protein (CRP) and negative mood scores on Total Mood Disturbance and four of the six subscales (all p < .01) was dependent upon higher levels of gamma-glutamyl transferase (GGT). Discussion: Higher levels of normal-range liver metabolic and peripheral inflammatory markers are observed with negative mood in a healthy older sample experiencing the biological impact of aging, but in the absence of clinical depression symptoms, suggesting a possible role of oxidative stress or other biological mechanisms occurring with aging in depression etiology. Lifestyle interventions are discussed.


Introduction
Increased age is the main risk factor in many developed nations for chronic illness, including chronic liver diseases (CLDs; Kim et al., 2015;Prasad et al., 2012). Among older populations, depression is associated with greater functional disability and reduced quality of life (Haroon et al., 2018;Huang et al., 2015;Prina et al., 2015), and it is poorly detected, under-diagnosed, and inadequately treated (Fiske et al., 2009;Hegeman et al., 2012), suggesting higher prevalence than is reported. It is highly comorbid with diabetes, cardiovascular disease and metabolic disorders, with increasing evidence of common biological pathways, exhibiting accelerated cellular senescence that is typically observed with progressive systemic dysfunction in aging (Alexopoulos, 2005;D'Mello and Swain, 2014;Hayflick & Moorhead, 1961;Huang et al., 2015;Mitchell & Subramaniam, 2005;Popović et al., 2015;Russ et al., 2015). Understanding and identifying the causal factors that underpin the comorbidity of chronic disease and depression are essential in order to prevent and treat emerging conditions and maintain optimal quality of life as we age.
The liver is a vital multi-functional organ involved in digestive, detoxification and inflammatory processes (Jenne and Kubes, 2013;Kerner et al., 2005;Pepys and Hirschfield, 2003;Racanelli and Rehermann, 2006). It also exhibits progressive dysfunction as a function of poor diet, alcohol use, sedentary lifestyles, and rising obesity in Western nations. Chronic liver disease (CLD) conditions such as non-alcoholic fatty liver disease (NAFLD; Higgins, 1931;Horvath et al., 2014;Taub, 2004), is increasing proportionately in older patients (Mahady and Adams, 2018;Polyzos and Mantzoros, 2016). While prevalence rates vary across liver diseases, depression is significantly higher with CLD such as NAFLD than the general population (Huang et al., 2017;Popović et al., 2015;Weinstein et al., 2011), correlated with severity of histological impact (Lee et al., 2013;Tomeno et al., 2015;Youssef et al., 2013). Longitudinal research on depression and liver dysfunction suggests that emerging symptoms of the former signal declining liver health (Elwing et al., 2006;D'Mello and Swain, 2014;Fraser et al., 2009;Tomeno et al., 2015). Both decreased oxidative and regenerative capabilities of the liver, and cellular senescence are common biological mechanisms of CLD and depression (Schmucker, 2005;Stahl et al., 2018), but evidence of a relationship between depression symptoms in non-clinical samples exhibiting prodromal markers of liver dysfunction or in asymptomatic liver disease/NAFLD is insufficient (Labenz et al., 2020;Shao et al., 2021), due in part to lack of consensus on reference limits for NAFLD, and liver enzyme profiles (Botros and Sikaris, 2013;Hall and Cash, 2012;Sorbi et al., 1999).
Given the evidence in clinical disorders of patterns in biological inflammation and liver function markers that suggest a replication of cellular senescence, and feasibly attenuation to mood, these patterns in aging non-clinical groups should elucidate the role of selected markers have in influencing mental wellbeing. As previous research has identified these patterns in clinical cohorts, the possibility that such relationships are detectable before either liver markers or low mood reach clinical significance has remained largely unexplored. The current study investigated the relationship between markers of inflammation, liver function and mood in a healthy older sample aged 60-75, where we hypothesized that elevations in liver enzyme levels will be associated with negative mood. Furthermore, a peripheral cytokine-liver enzyme interaction is expected, specifically that a pro-inflammatory response would occur with raised liver enzymes, and would better influence mood scores than either marker alone. In doing so, a better understanding of the relationship between biological markers and mental wellbeing can serve to optimise healthy aging.

Design and participants
The current study used baseline data from the Australian Research Council Longevity (ARCLI) study, a 12-month longitudinal investigation of nutraceutical supplements on cognitive performance and cardiovascular health in older individuals, aged 60-75 years for whom demographics, markers of liver function, inflammation, and mood were provided (n = 284; males = 121). The study was registered on the Australian and New Zealand Clinical Trails Registry (ACTRN12611000587910; refer protocol; Stough et al., 2012). Research conducted as part of the ARCLI clinical trial met protocols set by The Code of Ethics of the World Medical Association (Declaration of Helsinki). Informed consent was obtained for each participant at the first inperson study session.
Newspaper and radio spots in addition to pamphlet and flyer distribution were used to recruit participants who were screened for eligibility twice by phone and during a face-to-face interview. All screening and study measures took place at the Centre for Human Psychopharmacology at the Hawthorn campus of Swinburne University of Technology, Melbourne, Australia.

Assessment of medical eligibility
The ARCLI study protocol required a healthy cohort with no neurological, cognitive impairment, psychiatric disorders, or chronic disease or illness within the last five years, apart from minor physical conditions and associated medications (arthritis, asthma). Chronicity in physical illness was defined as any condition or symptoms lasting longer than 6 weeks within the previous five years. Participants with mild conditions such as arthritis not requiring opioid or steroidal medication, or medically managed cardiovascular disorders (hypertension and/or hypercholesterolemia) were included in the study but assessed comprehensively (refer Section 2.2.4).

Medication and supplement use
Individuals were ineligible to participate if taking acetylcholine esterase inhibitors, antipsychotics, antidepressants, anxiolytics, dietary supplements for cognitive enhancement or illicit drugs. Acceptable supplements were minerals, vitamin D, and omega 3.
Stable cardiovascular conditions managed with antihypertensives, statins or antiplatelet agents were allowable. Other allowable medication included hormone replacement, non-steroidal anti-inflammatories, and prophylactic medications (such as salbutamol for asthma).

Cardiovascular conditions
Cardiovascular status was assessed case-by-case via an established medical committee comprising of a research nurse, medical practitioner, and ARCLI chief investigator. Enrolment in the study occurred if conditions were hypertension, and/or hypercholesterolemia; were medically managed; and within normal ranges following measurement of cholesterol and blood pressure. The ARCLI protocol paper details this information in greater depth (Stough et al., 2009).

Assessment of cognitive impairment and depression symptoms
Cognitive health was assessed by The Mini Mental Status Examination (MMSE; Folstein et al., 1975). Used as a measure to screen for cognitive impairment, the MMSE tested cognitive function with a maximum score of 30 with a required score of >24 for inclusion in the study.

Assessment of alcohol consumption
Alcohol use criteria was set at <14 standard drinks (equating to 10 g of alcohol) for females and < 28 standard drinks for males per week, as set by the Australian National Health and Medical Research Council Guidelines (2009).

Assessment of liver function
Inclusion to the study involved assessment of liver markers where outside ranges were referred to the Medical Committee on a case-by-case basis, often involving re-test before inclusion to the study (refer Section 2.3.2 for markers and ranges).

Inflammatory markers
Participant blood samples were taken between 8:30 and 10:30 am following overnight fast from 10 pm. Samples were collected by a registered nurse or qualified venepuncture technician from the cubital fossa via in a BD Vacutainer® serum separator tube containing spraycoated silica and a polymer gel for serum separation, with serum frozen and stored at -80 • C until analysis. Inflammatory biomarkers Creactive protein (CRP), Tumour necrosis factor alpha (TNF-α) and interleukin-6 (IL-6) were analysed using xMAP® multiplex technology. Ten minutes prior to analysis, plasma samples were thawed and centrifuged at 2500 g. Assay was performed using MILLIPLEX®MAP kit in accordance with kit protocols, with plates being read on MAGPIX® platforms. Intra-and inter-assay coefficient of variance (CV = [SD/m] *100) was approximately <15 % (between <5 % and < 20 %). The lower limits of quantification were CRP 0.40 picograms per millilitre; IL-6 0.18 pg/ml; TNF-α 0.43 pg/ml. Healthy ranges of CRP are typically reported as <10 mg/ml, with a range of 4-9 mg/l denoting low-grade inflammation and increased risk (Dinh et al., 2019), TNF-α range was <20 pg/ml, and IL-6 range was <16.4 pg/ml (Zhou et al., 2010).

Liver function markers
Patterns in enzymes, offer information as to etiology and course of hepatic abnormalities (Giannini et al., 2005;Melvin et al., 2012;Ruttman et al., 2005;Tynjälä et al., 2012). The serum liver enzymes aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyl transferase (GGT), and AST:ALT ratio were used as markers of liver health for their pivotal role in this assessment (Goldberg, 1980;Hall and Cash, 2012;Hossain et al., 2016;Kim et al., 2019;Kwo et al., 2017). GGT can be highly variable in healthy samples and rarely assessed alone. In most acute hepatocellular disorders, the ALT is greater than or equal to AST. The De Ritis ratio is a numerical relationship between AST and ALT with a typical range 0.5 to 0.7, > 1 suggestive of liver complications such as NAFLD/non-alcoholic steatohepatitis (NASH) (De Ritis et al., 1957), and a ratio of >2 is indicative of alcoholic liver disease. Cardiovascular, endocrine, and metabolic disorders have common biological features, and this ratio is able to identify patterns that point to dysfunction, in particular, this ratio represents a marker of risk for NAFLD (Botros and Sikaris, 2013;Rief et al., 2016;Sorbi et al., 1999). An AST:ALT ratio of >2 is suggestive of alcoholic liver disease, and a ratio of <1 suggestive of NAFLD/non-alcoholic steatohepatitis (NASH). The ratio is thought to be more accurate than the non-specific nature of enzymes individually (Rief et al., 2016), with stronger evidence for pathology (such as alcoholic liver disease) if ratio levels are observed together with elevated GGT (Moussavian et al., 1985).
Samples were collected by a registered nurse or venepuncture technician and sent via courier to a commercial pathology laboratory for analysis and reported for each participant via a liver function test (LFT). Normal ranges at this laboratory were defined as 5-51 U/l for GGT, and 5-41 U/l for ALT and AST (Kerner et al., 2005;Kim et al., 2019;Pratt and Kaplan, 2000).

Mood measures 2.3.3.1. Profile of Mood States. The Profile of Mood States (POMS;
McNair et al., 1992) is a 65-item, five-point Likert scale self-reporting questionnaire comprising five negative affect subscales; depressiondejection, tension-anxiety; anger-hostility; fatigue-inertia; confusionbewilderment and one positive affect subscale, vigour-activity, and an aggregate 'Total Mood Disturbance (TMD)' score. Disregarding the vigour-active subscale, higher scores are indicative of greater negative mood and psychological distress (Nyenhuis et al., 1999). Previous literature supports POMS as a reliable and valid measure of mood state in healthy aging populations (Andrade and Rodríguez, 2018;Gibson, 1997). Beck et al., 1996). Depression symptoms were quantified via the BDI-II Likert-type scale self-report inventory, widely used in both clinical and research settings (Beck et al., 1996), with validity and reliability as an assessment tool within older adults (Gallagher et al., 1982;Segal et al., 2008). Higher scores on items such as sadness statement (e.g., 'I feel sad much of the time'), indicated more severe depressive symptoms up to a maximum score of 63. For the purposes of this study in a non-clinical group, scores were classified as follows (as per Beck et al., 1996); 1-10 (normal); 11-16 (mild mood disturbance); 17-20 (borderline clinical depression); 21-30 (moderate clinical depression); 31-40 (severe depression); and over 40 (extreme depression).

Health and lifestyle factors
2.3.4.1. Body mass index. Participation required a body mass index (BMI) within the healthy range of between 18.5 and 29.9. BMI is a putative marker of general health calculated through body weight (in kilograms) divided by the square of height (in centimetres; kg/m 2 ). The resultant score denotes health status in one of four potential categories: underweight, normal weight, overweight or obese. Healthy weight is considered to be a BMI between 18.5 and 24.9 for adults aged between 18 and 65 years of age; 25 or above overweight, and BMI of 30 or above denotes obesity (Keys et al., 1972; World Health Organization (WHO), 2019).

Procedure
Participants attended two sessions at the study site (Swinburne University of Technology, Melbourne) seven days apart; one screening visit and subsequent visit for data collection. Written and informed consent was collected before comprehensive screening for eligibility.
Sociodemographic information was collected from all participants including age in years; sex(male/female); BMI; medical conditions; alcohol consumption per week; allowable medication use (cardiovascular or not); and eligible vitamins and supplements as determined by the Medical Committee. Participant blood samples were collected at the second session following overnight fasting and without alcohol or caffeine, or vigorous exercise, nor a high fat or sugar meal in the previous 24-h period.

Statistical analysis
Data were analysed using IBM SPSS Statistics for Windows, version 26. Demographic data were reported via means and standard deviations and percentages for continuous and categorical data respectively; and comparisons performed via t-tests for continuous variables and chi-square for categorical data. Using a factorial analysis of variance (ANOVA), custom main effects models were utilized with participant age, sex, BMI, weekly alcohol consumption, inflammation markers (CRP, IL-6 or TNF-α) and liver enzyme markers (ALT, AST or GGT) as fixed main effects. To assess the a priori relationship of inflammation on liver function relationship to mood levels (main and subscales individually), and the moderating effect of inflammation tested via interactions in the model (example: CRP*GGT). Where appropriate, post hoc analyses were employed better gauge the relationships between predictors in the models. Statistical findings were considered significant if probability value was p < .05, with reporting of confidence intervals to establish parameter estimates.

Results
The study comprised 284 healthy Australian adults (male = 121) aged 60-75 with a mean standard age of 65.9 (SD 3.98). Participants had a mean 15.89 (SD 4.01) years of education and average weekly alcohol consumption of 5.13 (SD 5.13) standard drinks. Participants mean BMI was 26.51 (males; SD 4.27); and 24.92 (females; SD 6.05) which were overweight and borderline healthy respectively. Sex differences were observed in some of the measures -some expected -such as BMI -t(276) 2.21, p = .034, with males 1.5 points higher, but also reported for ALT, t (280) 3.85, p ≤0.001 which was higher in males; for GGT t(280) 3.04, p = .003, higher in males; and for AST/ALT ratio, t(280) -3.78, p ≤0.001, which was higher in females. Finally, female scores on the MMSE were significantly higher, t(220)-4.10, p ≤0.001. Means and standard deviations for all demographic variables and screening measures are detailed in Table 1.
Mean BDI-II score was 2.33 (SD 3.31). The Profile of Mood States total, and subscales were analysed separately; mean Total Mood Disturbance (TMD) score was 57.81 (SD 42.65). All means and standard deviations for TMD and the six mood subscales are shown in Table 3.

Linear models for the prediction of mood from liver function and inflammation
Using a 'liver enzyme x inflammatory marker' model in the prediction of BDI-II and POMS total and subscale scores, for TMD scores the interaction effect was significant for the moderation of CRP on GGT's prediction of mood, F(1,189) = 10.67, p = .001. CRP had a moderating effect on GGT, to the extent that CRP increases resulted in a 0.23 unit reduction in GGT, 95 % C.I. [0.10, 0.41] (refer to Fig. 1 for calculated interactions of CRP on GGT slope (coefficients) for all scales; TMD and subscales).
Models using IL-6 and TNF-a as markers of inflammation, and AST and ALT as liver markers, as predictors were not significant.
Further analysis of this interaction was conducted with GGT split by quartile values as per Table 4 and plotted to gauge trends for TMD scores. Lower GGT levels (quartiles 1 and 2) exhibited less effect on mood with higher CRP, and in these cases, increased CRP was associated with better mood via lower scores. Conversely higher GGT levels were associated with increased negative mood. Quartile 4, in particular, displayed a steep increase to negative mood with higher CRP.
For individual POMS subscales from liver enzymes and inflammatory markers, scores for four of the six mood subscales were significantly predicted by interactions between CRP and GGT.
Depression-dejection scores were significantly modulated by the interaction of CRP with GGT, F(1,189) = 10.72, p = .001 with the slope for GGT decreasing 0.03 for every unit increase in CRP, 95 % C.I. [0.02, 0.08]. Similar to the TMD scale, the interaction plot indicated differing trends, with minimal change to depressed mood with quartiles 1 and 2, but increased scores with quartiles 3 and 4.
The interaction of CRP with GGT also predicted scores for tensionanxiety F(1,189) = 6.069, p = .015 with the slope for GGT decreasing by 0.02 for every unit increase in CRP, 95 % C.I. [0.01, 0.08]. As with the previous scales, quartile 3 and 4 exhibited higher negative mood.
Anger-hostility scores were also significantly predicted by the interaction of CRP with GGT F(1,189) = 13.56, p < .001, with the slope for GGT decreasing 0.09 for every unit increase in CRP, 95 % C.I. [0.03, 0.10]. Again, inspection of the interaction plot indicated increases to negative mood for quartiles 3 and 4. Confusion-bewilderment scores were also able to be significantly predicted by CRP interacting with GGT F(1,189) = 6.96, p = .009 with slope for GGT decreasing 0.07 for every unit increase in CRP, 95 % C.I. [0.01, 0.09]. The interaction plot indicated a reduction to confusionbewilderment scores with increased CRP in quartiles 1 and 2, but little impact for quartile 3, and an increase to negative mood in quartile 4.
The subscales, fatigue-inertia (p = .267) and vigour-activity (p = .283) were not significant in these models. Finally, no models were significant for scores on the BDI-II measure (all, p > .05).

Discussion
The current study investigated the association between mood and biological markers of inflammation and liver function in a sample of older healthy Australian adults between the ages of 60 and 75. Results supported the hypothesis that a peripheral inflammation and liver marker relationship is associated with reduced mood, specifically, that levels of the liver enzyme GGT appear to be moderated by inflammatory cytokine acute-phase protein CRP levels in making a contribution to negative mood, particularly where GGT is higher. More specifically, the moderation of CRP on GGT's positive relationship to negative mood scores were reported on POMS' Total Mood Disturbance and four of the six POMS subscales.
That other markers were not found to be associated with mood suggest a unique inter-relationship of acute-phase protein in inflammatory processes via CRP, and GGT's role in either liver health, or less directly via mechanisms such as oxidative stress. The lack of relationships observed with IL-6 and TNF-αdespite reports in the literaturesuggest that they may have a role in attenuation to mood in circumstances of more acute-phase inflammatory processes, and less that of liver functionality, in the light of the current findings.
Notably, BDI-II scores were not associated with patterns of inflammation and liver function markers, and a likely reason for this observation may be the 'floor effect' reported for this sample -mean BDI-II scores were 2.33 (SD 3.31) which is in the lower end of the 'no or minimal depression' score range, and it is lower than means reported in population studies (typical ranges of 4-8 points; Smarr and Keefer, 2011). Moreover the BDI-II literature has a shortfall in data from healthy older samples, and so despite being a putative measure, it may not be sufficiently validated in healthy older cohorts as a measure of nonclinical depression symptoms (Edelstein et al., 2015). In the context of the current study, this sample may be 'overtly' healthy in terms of the data's capacity to demonstrate clinically meaningful levels of depressive symptoms, and thereby the BDI-II may be limited in quantifying relationships between the measures.
With regard to the main findings, all significant models from the POMS exhibited a similar trend of increased negative mood with higher CRP and GGT. Despite GGT range of 0 to 30 international units per litre is typically listed as normal, there is suggestion from the presented data that the upper 25 % of GGT (>26 iu/l) may be associated with mild sub-clinical negative mood. Moreover, results suggest there may be a threshold GGT must be at in order for CRP to have this negative association.
One example is confusion-bewilderment mood subscale, which captures cognitive inefficiency and disorganised thinking occurring in negative mood. Here the trend of increased CRP and lower GGT levels (quartiles 1 and 2) was associated with lower negative mood. In these cases, given that CRP was significant in the model (as with four of the five subscales), the data indicates that CRP may be more directly related to mood levels, as is observed in cases of acute-phase behaviours related to sickness response. One possible clinical phenomenon underpinning observed differences in mood could be non-specific inflammatory conditions observed in older cohorts, such as osteoarthritis -and a probable role of GGT in oxidative stress processes related to this inflammatory response, where elevations are reported in the literature (Spooner et al., 1982).
More generally, CRP elevations in older samples are reported in conditions of physical trauma, infections and chronic conditions such as arthritis, and cardiovascular disease; with levels above 10 mg/l suggesting a diagnosis (Morley and Kushner, 1982;Dinh et al., 2019). Elevated CRP is predictive of increased mortality risk from cardiovascular disease, diabetes and metabolic disorders in outcomes studies (Ridker, 2007), and although associated with depression in older samples, appears less predictive when controlling for cardiovascular disorder presence and factors such as smoking (Tiemeier et al., 2003).
As for GGT, in addition to liver function, is a biomarker of oxidative stress and it is symptomatic of the role oxidative stress plays in inflammatory processes (Fraser et al., 2009;Lee and Jacobs, 2005;Whitfield, 2001). GGT's chief role is the metabolism of glutathione and glutathionylated xenobiotics, however, excesses are associated with prooxidant activity and cellular damage, with increase to reactive oxygen species and nitric oxide cascades (ROS and NO;Koenig and Seneff, 2015;Whitfield, 2001). Elevated GGT levels (typically ranging >30 ul/l) are considered indicative of liver disease (Mistry and Stockley, 2010), and/or excess alcohol consumption (Conigrave et al., 2003;Lee and Jacobs, 2005).
GGT has been identified as a marker of oxidative stress associated with cardiovascular disease and diabetes (Lee et al., 2007), feasibly due to its role as a glutathione precursor in catabolism of endogenous and exogenous compounds, and pathway production of reactive oxygen species (Lee et al., 2004b). Even GGT levels within a healthy range can be predictive of oxidative damage (Lee et al., 2004a), and future elevations of CRP levels (Lee and Jacobs, 2005). While GGT and CRP indicate a strong positive relationship, they are negatively correlated with antioxidant levels (Lee and Jacobs, 2009) thus lending further support to a role in oxidative stress processes, and suggest, that despite a strong interrelationship, there are still features of their production that are separate.
The identification of GGT's association with inflammatory mechanisms and as a marker of OS also lends support to the hypothesis that OS may be the common causal mechanism between CLDs and depression (Huang et al., 2017;Lee et al., 2004b;Mistry and Stockley, 2010). While the nature of their relationship is to date unclear, it has been suggested that the elevation of GGT levels precedes those of CRP in systemic inflammation activation, (Lee and Jacobs, 2005). This model of inflammatory response with CRP levels altered in response to elevation of GGT is consistent with results from the current study, where CRP level's impact on mood appeared dependent upon levels of GGT. It is difficult to gauge the mechanistic relationship between these markers -acute phase modulation of CRP levels could be raised in response to elevated GGT; thereby reflecting an increased inflammatory state and increased negative mood. It is equally likely that CRP and GGT may be rising independently of one another and signalling unrelated mechanisms that in turn impact mood level. Were there to be a functional relationship, findings from the current analyses suggest that CRP may be acting in response to GGT, as would be expected given metabolic processes of the liver and subsequently a CRP-moderated inflammatory response.
The role of CRP as a marker in both mood and inflammatory conditions has been identified by previous research, with CRP found to be associated with depressive symptoms only via its moderating effect on other factors (Chu et al., 2019). Given the current observations, it likely that these patterns represent surrogate markers of generalised inflammation and therefore serve to elucidate biological processes in mood even within normal ranges, but that they display a liver functioninflammation relationship observed in clinical groups, yet with absence of liver damage/alcoholism in a healthy non-clinical sample experiencing cellular aging.

Study limitations
In the current study, medical and lifestyle histories were limited to a five-year period, and a more detailed life history could better enable the identification of causal relationships among influencing factors. This is especially for older individuals with histories of life stressors, higher alcohol use, or fatty diets. There is a well-established relationship between long term alcohol consumption and liver health (Osna et al., 2017;Smart & Mann, 1992;van Beek et al., 2013). Questionnaires designed to measure the drinking habits among older adults for the preceding year (Midanik et al., 2013) and life-course (Bell and Britton, 2015) have both been found to capture more realistic measurements of lifetime alcohol intake than weekly standard drinks, as mid-life drinking habits represent a risk factor for later liver health and fatty liver development in older years (Britton et al., 2019). Additionally, factors as diverse as childhood socioeconomic status, and cumulative lifetime depressive episodes influence reported CRP levels in later life (Bell et al., 2016;Copeland et al., 2012;Liu et al., 2017). Furthermore, major life stressors impact via chronic stress responses and the HPA axis, in turn producing a cascade of endocrine, immunological and oxidative stress processes, exacerbating systemic aging (Lavretsky and Newhouse, 2012;Sterling and Eyer, 1988). The addition of a life stress scale may aid in quantifying such environmental impacts. Taken together, a more detailed life history would better enable the identification of causal relationships among such variables. An additional limitation is omission of diet and nutrient status, which can influence liver function and inflammation status. Elevated GGT is associated with higher meat and alcohol and lower consumption of fruit (Lee et al., 2004b). While the association between alcohol intake and GGT levels is well documented (Moses & Kamali, 2019), the effect of diet is less well explored and understood. Secondly, many foods have pro-and anti-inflammatory properties evidenced to moderate CRP, in particular, diet patterns of red meat, high-fat dairy products and refined grains -very frequent in Western nations such as Australia (King et al., 2003). Finally, data in the current study were derived from cross-sectional analyses of a single time point, which limits conclusions regarding inflammation and liver function markers in a predictive capacity-which could be derived from studies of longitudinal design. To better assess the impact of age, use of a second comparator younger cohort may establish whether this association relates to factors such as cellular senescence in non-clinical otherwise healthy older people.

Conclusions and future work
The current study explored the role of inflammation, markers of liver health, and reduced mood in a healthy older population, and found  negative mood was observed with inflammation, but only where GGT levels were high on the normal ranges. These results suggest there is interplay of mechanisms between liver function and inflammation contributing to reduced mood, even in healthy samples, and likely indicates the trajectory of reduction to cellular function and generalised inflammation occurring as a function of age, or inflammaging. GGT and CRP are more closely related to lowered mood than to liver dysfunction. These patterns are similar to that observed in clinical liver dysfunction and may be of utility in understanding precedents to the development of liver disease, and potential biomarkers for risk of declining health with aging. Future research will benefit from addressing the limitations, as well as giving consideration to nutrition and diet in the way it may impact biological processes such as the observed relationships. Further research should consider examining how raised GGT/CRP and low mood respond to interventions geared to lowered oxidative stress, for mood improvement.
Findings contribute to the knowledge gap of mechanisms between depressed mood and biological markers of inflammation and liver function, which is not frequently investigated outside of clinical groups, despite cellular senescence being a prominent biological feature of aging, and therefore likely to partly underpin alterations to mood states. These findings extend previous research identifying an association between liver function and mood levels. By using a healthy sample, the GGT-inflammation-mood relationship/association can be inferred to be independent of pre-existing health issues/conditions.

Acknowledgements and Funding
This baseline data was derived from the larger ARCLI clinical trial. Sincere thanks go to the trial participants for their time and efforts. ARCLI was funded by the Australian Research Council, Horphag, Blackmores, and Soho-Flordis. ARCLI also received generous donations from Swinburne Alumni including Doug Mitchell, and Roderic O'Connor.

CRediT authorship contribution statement
AP conceived the methodology included in this manuscript, performed edits, and supervised CP who contributed to the literature review and discussion. CS as chief investigator was responsible for conception and supervision of the ARCLI study and contributed to edits of the manuscript. EB assisted with statistical analyses in this methodology. MH contributed to edits of the manuscript. AK contributed to edits of the manuscript. EO was the ARCLI trial general practitioner, medical committee member, and contributed to edits of the manuscript. KS conducted trial testing and data collation, drafted the manuscript and managed edits.