Biomarkers of the ageing immune system and their association with frailty – A systematic review

Introduction: Ageing is associated with several physiological changes, including changes in the immune system. Age-related changes in the innate and adaptive immune system are thought to contribute to frailty. Understanding the immunological determinants of frailty could help to develop and deliver more effective care to older people. This systematic review aims to study the association between biomarkers of the ageing immune system and frailty. Methods: The search strategy was performed in PubMed and Embase, using the keywords “ immunosenescence ” , “ inflammation ” , “ inflammaging ” and “ frailty ” . We included studies that investigated the association of bio- markers of the ageing immune system and frailty cross-sectionally in older adults, without an active disease that affects immune parameters. Three independent researchers selected the studies and performed data extraction. Study quality was assessed using the Newcastle-Ottawa scale adapted for cross-sectional studies. Results: A total of 44 studies, with a median number of 184 participants, was included. Study quality was good in 16 (36 %), moderate in 25 (57 %) and poor in 3 (7 %) of studies. The most frequently studied inflammaging biomarkers were IL-6, CRP and TNF- α . Associations with frailty were observed for increased levels of (i) IL-6 in 12 of 24 studies, (ii) CRP in 7 of 19 studies, and (ii) TNF- α in 4 of 13 studies. In none of the other studies were associations observed of frailty with these biomarkers. Different types of T-lymphocyte subpopulations were studied but each subset was studied only once, and the study sample sizes were low. Conclusion: Our review of 44 studies on the relation between immune biomarkers and frailty identified IL-6 and CRP as the biomarkers that were most consistently associated with frailty. T-lymphocyte subpopulations were investigated but too infrequently to draw strong conclusions yet, although initial results are promising. Addi- tional studies are required in order to further validate these immune biomarkers in larger cohorts. Furthermore, prospective studies in more uniform settings and larger cohorts are needed to further investigate the association with immune candidate biomarkers for which potential associations with ageing and frailty were previously observed, before these can be used in clinical practice to help assess frailty and improve the care treatments of older patients.


Introduction
The global population is ageing rapidly. Ageing is associated with several physiological changes, including changes in the immune system. Age related changes in the innate and adaptive immune system are thought to contribute to frailty, but exact relations between immune parameters and frailty remain to be established. Frailty is highly prevalent; and found in 20 % to 30 % of the older population over 75 years (Topinkova, 2008). A frequently used definition of frailty in geriatric medicine is a clinical state characterized by a decline in functioning across multiple physiological systems, accompanied by increased vulnerability to stressors which results in high risk of poor health outcomes, including falls, incident disability, hospitalization and mortality (Fried et al., 2001). Mechanistically, frailty appears to be a multifaceted deregulation of several biological pathways and systems. Recent studies have addressed the correlation of biomarkers and the frailty clinical phenotype to a certain extent.
Among the possible mechanisms that contribute to the occurrence of frailty are the age-related changes that occur in the immune system. As the innate immune system serves as the first line of defense against injury and infections, it gives an immediate response to external stressors and, as a results, plays a crucial part in the development and shaping of immune responses which, in turn, play a central role in inflammation and immune protection against infections (Yao et al., 2011). Emerging evidence suggests that the immune system is altered in frailty. These alterations are referred to as "immunosenescence" and "inflammaging". These phenomena are characterized by age-related imbalances in immune responses and by alterations in the underlying cellular mechanisms. "Immunosenescence" refers to the decline of (predominantly) the adaptive immune system, and is characterized by reductions in the numbers as well as the antigen-recognition repertoire of naive T and B cells. These reductions in the adaptive immune system are thought to result from age-related declines in hematopoietic stem cell numbers and thymic involution. Inflammaging refers to an agerelated over-activation of the innate immune system, resulting in a state of chronic, low-grade, sterile inflammation. Inflammaging is thought to be triggered both by the age-related declines in the adaptive immune system (as a compensatory response), as well as by the agerelated accumulation of (immune-reactive) debris (Franceschi et al., 2000). Studies have demonstrated elevated CRP, cytokine and chemokine levels, and an abnormal white blood cell distribution in older adults, which were suggested to reflect a dysregulated inflammatory state related to advancing age and which have been linked to adverse outcome of various diseases, such as cancer and Covid-19 (Choudhary et al., 2021).
It is however unclear whether these dysregulations in the innate and adaptive immune system are related to frailty. The relationship between the degenerated immune system and adverse outcomes could be based on underlying confounders such as comorbidities or malnutrition which can also lead to a state of chronic low-grade inflammation in older people. However, there is also evidence that chronic exposure to inflammatory mediators may be in part responsible for the development of chronic diseases (Furman et al., 2019). As immune changes may be one of the mechanisms underlying the development of frailty in older adults, understanding the immunological determinants for frailty may help to develop and deliver more effective care to older people. The aim of the present review is to study the association between biomarkers of the ageing immune system and frailty.

Search strategy
The present systematic review was performed with the assistance of a trained librarian. This review was conducted following the PRISMA guidelines (http://www.prisma-statement.org/) for reporting and design of systematic reviews. We systematically queried PubMed and Embase for citation until December 31st, 2021. The search strategy only included MeSH terms and studies published in English. We used the keywords "immunosenescence", "inflammation", "inflammaging", "frailty". The full literature search strategy can be found in the Appendix 1.

Eligibility criteria
Three authors (NdG, SPM, JP, FVdB and ETVH) independently selected the studies according to the following criteria. Any disagreement was resolved through discussion. Included studies were those (1) investigating frailty and immune biomarkers measurements at identical timepoint, (2) reporting blood biomarkers of the immune system or local tumor related immune biomarkers -such as tumor-infiltrating lymphocytes, as those cellular markers may reflect the inflammation status of frail patients -and (3) reporting frailty with standard tools, used in the medical research, assessing the different domains of frailty (unintentional weight loss, exhaustion, low energy expenditure, low grip strength, and/or slowed walking speed), with an available description of the scoring process. These instruments, of which the Fried Frailty Scale has been most used, have been validated against gold standard Comprehensive Geriatric Assessment and are strongly predictive of mortality and other adverse events (Lee et al., 2020). Studies were excluded if these (1) comprised a population with an active disease with potentially major effects on both frailty and the immune profile at baseline (such as HIV, hemotologic malignancies, active infection, …)we chose not to exclude studies investigating cancer patients, as cancer is often diagnosed in older populations and the impact of solid cancer on the immune system is assumed to be relatively limited (in contrast to for example hematologic malignancy or HIV), (2) only investigated dementia and cognitive impairments, as cognitive functions comprise different pathways, (3) performed the measurements shortly after an intervention such as vaccination, or (4) did not use clear diagnostic criteria for frailty or used only one component or one geriatric impairment of the frailty phenotype (e.g only fatigability) for its diagnosis.

Data extraction
Three authors (NdG, SPM, ETVH) extracted data from the selected studies in a Microsoft Excel spreadsheet. The following information was extracted: (1) characteristics of the study population (including sample size, demographics, country in which the study was performed); (2) setting in which the study was performed; (3) diagnostic criteria for frailty; (4) immunological parameters assessed with corresponding methods of measurement; (5) measured biomarkers; (6) type and number of adjustments in the multivariate analyses; (7) main associations reported.

Description of analysis and presentation of data
We used a table to describe the studies and extract all the results. The table described the following information: author, country of the study performed, study design, type of population studied, aim of the study, inclusion and exclusion criteria, diagnosis (if applicable), type of treatment (if applicable), biomarkers, number of participants, age of participants, type of analysis, interpretation and conclusion.
From the extracted data, we counted the studies that investigated each biomarker. Furthermore, we determined studies demonstrating statistically significant association with the biomarker.

Quality assessment
Study quality was assessed by three authors (NdG, SPM, ETVH) using the Newcastle-Ottawa Scale (NOS) adapted for cross-sectional studies (Herzog et al., 2013). The NOS assigns a maximum of 10 points based on three quality parameters: selection, comparability, and outcome. Quality was assessed as poor if the score was below 5 points, moderate if the score was 5 or 6, good if the score was 7 or 8 and high if the score was 9 or 10. In case of disagreement between two authors, consensus was reached after discussion.

Data search results and characteristics of included studies
The flowchart of study selection is shown in Fig. 1. The systematic searches resulted in a total of 1647 records. By the initial screening of titles and abstracts, 490 items were identified after removing duplicates and irrelevant records. After excluding 435 records for irrelevant source material that did not describe studies in accordance with the inclusion criteria, 65 records underwent full-text review. After reading the full text, 44 papers met the criteria for inclusion. Table 1 shows details of the selected studies. The studies included a total of 18,419 participants with a median of 184 participants. Older patients were over 60 years old. The included studies used different definitions of frailty; the Fried score was the most frequently used to measure frailty (in 24 studies, 55 %). Studies were population based (11 studies, 25 %) or included older adults from hospitals or clinical research centers (8 studies, 18 %, with 6 disease specific studies), longterm care facility (9 studies, 20 %), general practice (6 studies, 14 %) and 3 studies did not report the study population. The quality of the studies, assessed with a based New-Castle Ottawa Scale tool, was moderate in 25 studies (57 %), good in 16 studies (36 %) and poor in 3 studies (7 %).
In total, 143 unique biomarkers were studied. Fig. 2 presents the 10 most studied biomarkers of the innate immune system. IL-6 was evaluated in 24 studies, CRP was reported in 19 studies, TNF-α was measured in 13 studies, and white blood cell (WBC) count was assessed in 10 studies. All other biomarkers of the immune system were studied in <8 studies. Twelve out of 24 (50 %) demonstrated statistically significantly higher IL-6 serum levels in frail individuals and five studies observed non-significantly higher IL-6 serum levels in frail individuals. For instance, Adriaensen et al. (2014) performed a high-quality crosssectional study on 394 community-dwelling older adults, with a mean age of 85 years. Adriaensen et al. measured frailty with a global functioning score, which was determined with the activities of daily living score (ADL), the short physical performance battery (SPPB), minimental state examination (MMSE) and Geriatric Depression Scale (GDS)-15. In total, 21 % was frail and 26 % was mild frail. IL-6 serum levels were significantly higher in patients with frailty. Compared to individuals with lower levels of IL-6, the study reported a higher odd of functional impairment occurring in individuals with slightly elevated IL-6 (adjusted OR (%95CI), 4.16 (1.6-10.9)) or highly elevated IL-6 levels (adjusted OR (%95CI), 4.35 (1.7-114)).
Higher CRP levels were significantly associated with the presence of frailty in 7 of 19 studies (36 %) and non-significantly in 5 studies (26 %). Higher TNF-α levels were significantly associated with the presence of frailty in 4 out of 13 studies (30 %) and non-significantly in 2 studies. The high-quality study from Collerton et al. (2012) investigated CRP and TNF-α serum levels in 811 older adults from the Newcastle 85+ study. The frailty status of the patients was evaluated using an approximation of the Cardiovascular Health study and Rockwood Frailty Index (RFI) assessing the MMSE and disability score from ADL. When models were before screening          fully adjusted for covariates, IL-6 and TNF-a were similarly associated with frailty. Lower basal IL-6 level (OR (%95CI), 0.50 (0.31-0.79)) and lower basal TNF-α levels (0.62 (0.39-0.98)) were associated with a lower risk of frailty and higher CRP levels (1.78 (1.12-2.85)) remained positively and significantly associated with a higher risk of frailty. Fig. 3 presents the 10 most studied biomarkers of the adaptive immune system and their association with frailty. The total lymphocyte count was investigated in 11 studies. In 8 out of 11 studies, the total lymphocyte count was not significantly associated with frailty. However, 3 studies observed a significant negative association of total lymphocyte count with frailty. For instance, Nunez et al. (Núñez et al., 2020), demonstrated in a high-quality study that low lymphocytes percentage was associated with a higher risk of frailty (p = 0.001) and moreover low lymphocytes percentages were also associated with risk of long-term mortality. WBC count was investigated in 10 studies but only one study with a moderate quality, observed a significant positive association with frailty (Gilmore et al., 2021).
T-cell subsets were only studied in two studies. The largest crosssectional study including 1072 participants, from Johnstone et al. (2017), demonstrated that higher percentages of naïve CD4 + T-cells (p = 0.001) and effector memory CD8 + T-cells (p = 0.02) were associated with a lower Frailty Index, whereas a higher percentage of CD8 + central memory T-cells was associated with a higher Frailty Index (p = 0.02).

Discussion
This systematic review shows that only 44 studies investigated the association of biomarkers of the ageing immune system with frailty. There studies reported many and diverse biomarkers of immunosenescence and inflammaging, especially cytokines and cellular biomarkers.
Our review shows that elevated levels of IL-6 and CRP were consistently associated with frailty. Lymphocytes subpopulations were investigated too infrequently within small cohort studies only to draw any conclusions. Only one in three studies had high quality.
A large body of literature recognizes IL-6, CRP and TNF-α as relevant biomarkers for the ageing innate immune system. Predictably, as IL-6 and TNF-α are predominantly monocyte derived, Leng et al., have observed changes in the monocytes compartment with ageing (Leng et al., 2009). However, only a small number of studies reported those biomarkers as significantly associated with frailty. Our findings are consistent with the recently published systematic review, by Xu et al., showing that peripheral inflammatory biomarkers are related to frailty (Xu et al., 2022). However, the review was restricted to a limited number of pre-specified biomarkers including lymphocytes, IL-6, CRP, and TNF-α. Our review adds the open search for any biomarkers.
There are several explanations for the observed association between IL-6, CRP and TNF-α with frailty.
CRP has been widely recognized as a marker for systemic inflammation. Walker et al. demonstrated in a prospective study that increasing CRP levels during midlife increased the risk of frailty later in life (Walker et al., 2017). Studies have shown that increases in CRP plasma levels are associated with increased risk of sarcopenia, cardiovascular diseases, disability, and cognitive decline in older individuals. Elevated CRP levels were also associated with increased risk of mortality in frail older patients (Puzianowska-Kuznicka et al., 2016;Nouvenne et al., 2016;Giovannini et al., 2011). However, many diseases can cause increased CRP levels.
Measurement of CRP plasma levels is a frequently used screening test in daily clinical practice. Clinicians use it as a tool to diagnose infections or clinical conditions closely associated with underlying inflammatory mechanisms. Therefore, it seems be a rather unspecific biomarkers of the ageing immune system.
There is growing evidence that IL-6 and TNF-α play a central role in pathogenesis of chronic and age-related diseases. In acute inflammation, IL-6 promotes the expansion and activation of T cells and differentiation Previous studies demonstrated an association with increased levels of proinflammatory serum markers in older adults and in individuals with dementia or Parkinson's disease. It has been suggested that IL-6 may advance the onset of age-related diseases. Results for TNF-α are heterogenous. TNF-α has previously been associated with exhaustion and chronic fatigue syndrome which share a pathophysiological core with the frailty condition (Furtado et al., 2020;Morey et al., 2015). Additionally, cytokine networks involving up-and downregulation of other cytokines may add additional layers to the heterogeneity of the immune responses among frail adults. Therefore, it may be important to consider patterns or profiles comprising numerous circulating cytokines rather than focusing on the effect of individual cytokines.
Total white blood cell counts, and lymphocyte counts are the most studied biomarkers of adaptive immunity. Xu et al. discusses the relevance of leukocytes and lymphocytes in ageing . Emerging research shows that an elevated lymphocyte rather than leukocyte count has been associated with frailty and especially with low physical activity and grip strength (Fernandez-Garrido et al., 2018). Lymphocytes subpopulations were investigated infrequently and in rather small cohort studies. Therefore, solid conclusions cannot be drawn. However, there are some promising results.
During ageing, T lymphocytes are strongly affected due to changes in the proportion of T lymphocyte subpopulations that have undergone cellular senescence, such as a decrease in CD4 + T cells and an increase in CD8 + T cells (Shirakawa and Sano, 2021). Moreover, cells of the innate (monocytes, macrophages, basophils, etc.) and adaptive (CD4 + and CD8 + lymphocytes, etc.) immune system can exhibit alterations in functions and phenotypes. Only few studies reported an association of lymphocytes with frailty. For instance, Johnstone et al. (2017), demonstrated that lower levels of naïve CD4 + T cells and higher levels of CD8 + central memory T-cells were predictive of higher scores of the frailty index. Another study by Ng et al. investigated thoroughly T-cell subsets, demonstrating loss of CD28, an established hallmark of immunosenescence (Rodriguez et al., 2020), as predictive of frailty (Ng et al., 2015). Only these two studies investigating lymphocytes subpopulations had a high quality. Johnstone et al. and Ng et al. included a large sample of elderly nursing home residents, increasing the power of the study to detect meaningful differences between groups (Johnstone et al., 2017;Ng et al., 2015). The studies controlled for confounding factors, such as age, sex, comorbidities, and medication use, reducing the risk of bias in the results. The lack of association in other studies may be due to the vulnerability of frail individuals to acute and subacute diseases that affect inflammatory parameters. As a results, adjusting for these confounders in the analyses, particularly in low sample sizes, could decrease the significance of the results and explain the lack of association of the markers and frailty. In the line of our results, studies were generally  small and too few to draw any strong conclusion about the association of lymphocytes and frailty, based on currently available literature. This review has several strengths. The review validates previous literature and additionally, highlights the potential cellular biomarkers. To the best of our knowledge, this present study is the first to conduct an open search for any biomarkers, enlightening a large panel of cellular markers by describing the studies thoroughly and assessing their quality. We conducted an extensive literature search to identify biomarkers and performed an adequate quality assessment for cross-sectional studies. Moreover, our study was systematically conducted and results were systematically reported. Results were described for each distinct markers, uncovering the research gap and opening the path for further investigations. However, we only included studies focusing on physical frailty, as cognitive-decline-related frailty covers a large spectrum of cognitive diseases, including a very wide panel of markers of interest. Therefore, other aspects of frailty such as cognitive status (e.g., dementia), which is of relevance in frailty, were not considered. Further studies on cognitive decline should be required to understand frailty and inflammation. Furthermore, the heterogeneity of the studies did not allow to draw very strong conclusions. Non-standardized naming and measurements of the biomarkers make the comparison between studies more difficult; several studies defined subsets of CD4 and CD8 differently, but looked at the same marker, therefore the total number of biomarkers found can be ambiguous. Underlying data or following standardized naming conventions could be used to improve the comparability between studies.
In conclusion, our review of 44 studies on the relation between immune biomarkers and frailty identified IL-6 and CRP as the biomarkers that were most consistently associated with frailty. T-lymphocyte subpopulations were investigated but too infrequently to draw strong conclusions yet, although initial results were promising. Additional studies are required in order to further validate these immune biomarkers in larger cohorts. Furthermore, prospective studies in more uniform settings and larger cohorts are needed to further investigate the association with immune candidate biomarkers for which potential associations with ageing and frailty were previously observed, before these can be used in clinical practice to help assess frailty and improve the care treatments of older patients.

Funding
This work has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No 860173.