Salivary Biomarkers in Psychoneuroimmunology

Abstract

As molecular biology advances, an increasing number of proteins are becoming detectable at very low levels in different biological tissues. In this regard, saliva holds vast promise. Unlike blood, saliva can be sampled 1) non-invasively; 2) across all ages (newborn to elderly); 3) in the field; 4) by study participants; and 5) many times per day. With respect to psychoneuroimmunology (PNI), physiological measures of stress such as cortisol have been well characterized. Alpha amylase provides another physiological index of stress; it is a measure of autonomic nervous system activation and is quantifiable in saliva. Other salivary measures, such as inflammatory biomarkers and immunoglobulin A (IgA), provide valuable information pertaining to the effects of stress on inflammation, mucosal immunity, and oral health. Importantly, due to various methodological issues and a lack of strong correlation between saliva and blood measures, investigators should proceed with caution in drawing conclusions from measures of salivary inflammation that pertain to systemic immunity or generalized health.

Introduction

The increased availability of salivary biomarkers has become extremely valuable to behavioral scientists, and allowed the field of stress research to make fast progress. Before salivary biomarkers, human psychobiology was investigated mainly by measurement of stress hormones in urine and blood, and by inference through indicators of physiological function, such as assessment of skin conductance or heart rate. Due to logistic difficulties or high costs associated with most blood-based measures, many early studies limited the assessment of biomarkers to a single time point, which had limitations and hampered the ability to understand the impact of stress under different (often dynamic) contexts and conditions. A typical example is the early finding that depression was associated with increased cortisol secretion; this had to be revised when salivary measurement of cortisol permitted researchers to sample more frequently, showing that depression is associated with an aberrant diurnal pattern of cortisol secretion (i.e., a flattened diurnal cycle) characterized by increased levels at some time points and lower levels at other times [1].

This increased opportunity to use saliva for non-invasive biomarker assessment has greatly increased our understanding of how acute and chronic stress, including psychiatric conditions, manifests in humans in the form of a distinct biological signature. Acute stress coincides with broadly altered functioning in a large number of physiological systems, most notably the brain, whereas research has mostly focused on activation of the hypothalamus-pituitary-adrenal (HPA) axis and the sympathetic nervous system (SNS). When a threat to an organism is perceived, both systems are activated rapidly via endocrine (i.e., cortisol [HPA], epinephrine and norepinephrine [SNS]) and neural (SNS) pathways.

Despite being activated simultaneously, each of these systems has a different temporal response pattern, with SNS effects being apparent almost immediately, and cortisol secretion being delayed by about 20 minutes. This difference in temporal dynamics between SNS hormones (mostly epinephrine) and the HPA hormone cortisol appears to serve biologically relevant functions delineating an adaptive stress response [2]. At the level of the immune system, neuro-endocrine effects of acute stress appear to have immune-activating properties. For example, sympathetic activation causes the quick release of antibodies into saliva and the release (mobilization) of effector-type lymphocytes (e.g., NK cells, differentiated cytotoxic T cells) into the blood, whereby subsequent cortisol release may promote migration of these cells from the blood into peripheral tissues. In addition to these synergistic effects, there are indications that the sequential release of SNS and HPA hormones can have antagonistic effects. For example, there is evidence that the SNS first activates inflammatory mechanisms, which are later downregulated by cortisol. In this case, the difference in temporal patterns is thought to allow for a precautionary upregulation of a system in preparation for injury, followed by a down-regulation in the absence of an infectious stimulus. However, if systems are activated too often, or fail to activate, maladaptive regulation may be observed; this is denoted as allostatic load [3].

While blood-based biomarkers initially helped us understand these response patterns, the addition of saliva-based measures of HPA axis and SNS activity have allowed tighter sampling schedules and, hence, a better description of response patterns, thus having a major impact on this line of work.

Saliva-based measures of stress system activity

The HPA axis is characterized by pronounced circadian activity, with increases in the morning (cortisol awakening response [CAR]) and declines throughout the day. One-time measurements of cortisol in blood have led to inconsistent findings, because timing of blood samples was typically not adjusted to the axis’ circadian rhythm. Years of research using salivary cortisol have allowed researchers to understand that chronic stress affects parameters of the diurnal slope which becomes flatter, and after long stress exposure results in a state of hypocortisolism and loss of rhythmicity [4]. When assessing diurnal cortisol slopes, current recommendations are to use date from more than one, ideally three or more, sampling days, to use the wake-up cortisol level as a starting point [5,6] and to employ multilevel models for data analysis [7]. Despite adhering to these guidelines, a significant proportion of cortisol variability is explained by day-to-day fluctuations [8].

Salivary alpha amylase (sAA) has been used as a proxy measure of autonomic nervous system (ANS) activity during acute and chronic stress. sAA is an enzyme secreted by acinar cells in saliva glands, in which innervation by sympathetic nerves via noradrenaline signaling induces protein release. As sAA is the predominant protein in saliva, and can be easily and relatively cheaply assessed, it rapidly became popular as a measure of sympathetic activation [9–11]. Although stress-induced increases in sAA poorly correlate with other SNS markers [12], this is a generalized issue with the assessment of SNS markers and is not regarded as a disqualifying property. Moreover, recent pharmacological work has suggested that sAA might be a marker of central noradrenergic activity [13]. The assessment and interpretation of sAA is not as straightforward as initially thought. For example, levels can vary depending on sample timing, saliva collection method, and stressor type, and it is recommended that saliva flow rate is taken into account as a potentially confounding factor. The field would greatly benefit from further methodological studies that help standardize the use of sAA in research. These cautionary notes notwithstanding, recent work has for example shown that sAA stress responses are higher in individuals with self-reported early adversity [14]* and in people with lower reported self-compassion [15]. Hence, the addition of sAA to the methodological repertoire permits non-invasive assessment of ANS function, although sAA is a less direct proxy for the SNS than cortisol is for the HPA axis.

Autonomic activation occurs with a much lower threshold (i.e., it is observable even with mild stressors; [16]) than HPA activation. As a result, sAA tends to respond to acute stressors that are not powerful enough to activate the HPA axis, such as restriction of cell phone use in college students [17]. In a recent study, women with a history of trauma showed sAA, but not cortisol, reactivity to a trauma reminder during a clinical interview; those increases were positively associated with neural reactivity of trauma-related brain areas [18]*. The authors concluded that sAA might be a more reliable marker of trauma-related reactivity and hypervigilance than measures of cortisol.

Alterations of sAA circadian patterns have also been reported in chronic stress. In a one-year longitudinal study of caregivers to brain cancer patients, amylase diurnal rhythms, which are typically characterized by a sharp decline in the first hour after waking and increase thereafter, became flatter over time [19]. Altered diurnal sAA patterns have also been found in female sexual minorities (non-heterosexual women; [20]*), and in police officers with lower subjective social status [21]. These patterns of sAA alteration support strong associations with the ANS. As discussed below, questions still needing attention are the relative contribution of sympathetic and parasympathetic components, and methodological requirements in regard to sampling protocols [22]. However, amylase diurnal rhythms appear to be more stable than those of cortisol [23].

When analyzing the effects of acute and chronic stress on health, research on stress systems alone has not proven useful for understanding pathophysiological mechanisms. Investigating the combined effects of both stress systems on target systems, such as immunity and inflammation, has the potential to fill this important gap. Inflammation, while necessary for fighting off infections and healing tissues, has a strong potential to damage the organism [24]. A slow but steady increase in systemic inflammation is a common part of the aging process, denoted as inflammaging (see [25] for review). While such slow increases of low-grade inflammation occur with aging, they are most likely not a simple consequence of becoming older, but caused by accumulation of environmental impacts on the immune system, some of which can be psychological in nature. Both acute and chronic stress exposure promote inflammation in the absence of acute infection or tissue damage [26], thereby providing a potential link between stress and adverse health outcomes. In addition, stress can weaken immune defenses through negative effects on other components of the immune system, e.g., adaptive immunity [27]. Assessing inflammation at multiple time points and in natural settings would be highly informative. Salivary biomarkers allow for this and provide a better understanding of the links between stress and inflammation, which is critical in PNI research.

Factors that modulate inflammation in saliva

Throughout our entire lifespan, oral health modulates salivary inflammation. For instance, very young children (birth to 71 months) with early childhood caries exhibited elevated salivary inflammation (based on cytokine levels) compared to age-matched controls [28]. Moreover, oral health tends to worsen with age, which limits the usefulness of salivary inflammation within PNI. By adulthood, 80% of the population has gingivitis [29] and almost half of individuals over 30 years of age have some degree of periodontal disease [30], both of which relate to higher levels of salivary cytokines and C-reactive protein (CRP) [31–33]. Other oral conditions are also associated with elevated salivary inflammation, such as oral lesions (personal observations, [34] [35]) and oral squamous cell carcinoma [36]. The local production of cytokines that occurs under these types of conditions likely accounts, at least in part, for the weak (or lack of) correlation often reported for measures of inflammation between saliva and blood (see [37] for review).

A number of systemic and inflammatory diseases or conditions have also been associated with elevated inflammation in saliva. These include, but are not limited to, acute myocardial infarction [38], heart failure [39], cardiovascular disease [40], metabolic syndrome [41], and rheumatoid arthritis [42]. Finally, a large number of factors not specifically related to disease states have been associated with higher salivary inflammation, including BMI [43], obesity [44,45], short sleep duration [46], obstructive sleep apnea [47], smoking [48], and even cell phone use, assumedly due to heat production [49]. As a result, inflammatory cytokines do not serve as reliable indicators of systemic inflammation, as they often do not correlate well with blood values (cytokine values are often higher in saliva than in blood, assumedly due to multiple sources of local production; personal observations; see [29] for review).

In contrast to cytokines, there is no local production of CRP in the mouth as it is produced by the liver. As a result, CRP levels are much more dilute in saliva than in blood, e.g., 1600x more dilute [50]. However, similar to cytokines, CRP levels are higher in saliva under conditions of poor oral health (e.g., gingivitis), likely due to 1) higher infiltration of CRP from blood and 2) the presence of blood in the mouth from damaged tissues. As a result, some studies report significant correlations for CRP between saliva and blood, whereas others do not (see [51] for review).

Theoretically, individuals with good oral health should have salivary levels of inflammation that closely relate to levels of systemic inflammation. However, this has yet to be demonstrated in empirical studies and, as discussed above, many factors outside of oral health can seemingly affect salivary inflammation.

How might researchers determine good versus poor oral health? Ideally an oral examination should be conducted by a trained clinician, but this is seldom possible. Fortunately, even a rudimentary assessment can allow for the identification of individuals with poor oral health. Some examples of simple questions are: Do you have good oral health? y/n; Rate your oral health on a scale of 1–5; Do your gums sometimes bleed when you brush or floss? Y/n. The inclusion of such questions is recommended for research studies involving salivary inflammation.

Utility of salivary inflammation in psychoneuroimmunology

As reviewed by Slavish et al. (2015), saliva may have good utility for detecting inflammatory change from baseline, for instance in response to acute stressors [37]. In this regard, salivary levels of the classic pro-inflammatory cytokines (IL-1β, IL-6, TNF-α) were found to increase fairly consistently in response to acute stress [37]. Moreover, these changes generally appeared to mirror, possibly with some time lag, similar changes in blood. Note that these findings were based on a relatively small number of studies and more work is needed in this area.

Some recent studies have demonstrated elevations in salivary inflammation in response to acute stressors, such as university examinations [52]) and public speaking [53]). In one study, the Trier Social Stress Test (TSST) and the retrieval of angry memories were each shown to increase salivary levels of IL-1β, IL-6, and TNF-α in healthy young adults. Distracting individuals related to lower salivary IL-1β levels after memory retrieval but not after the TSST, suggesting that emotional reactivity plays a role in salivary inflammation [54]*.

A number of recent studies have assessed the efficacy of interventions aimed at reducing stress on salivary inflammation. Two separate studies found that mindfulness-based stress reduction lowered salivary inflammation after four or six weeks of training [55,56]. Intriguingly, a single 20-minute session of yoga breathing reduced salivary inflammation levels compared to controls [57].

Although cytokine measures in saliva are seemingly not reliable indicators of systemic inflammation, a few studies have found relations between salivary inflammation and longer-term measures such as chronic stress. For instance, Martinez et al. (2018) studied 111 individuals living in thirty low-income Mexican-origin families with at least one immigrant [58]**. They found that household fear of deportation and family conflict (both chronic stressors) strongly related to elevated levels of inflammation in saliva [58]**. While salivary CRP (sCRP) levels do not appear to change readily to acute stress [37], a recent study reported that negative emotionality related to higher sCRP levels, whereas effortful control related to lower sCRP levels [59]. Finally, female partners of military veterans with traumatic brain injury (TBI) who reported higher blame/anger exhibited higher salivary inflammation [60]. The implications of such findings on systemic inflammation, or other parameters of health, are so far unclear.

To sum, there appears to be utility in assessing inflammatory change (from baseline) in saliva across multiple time points. It is well established that short-term stress can cause temporary increases in inflammation which in time return to baseline. Assessment of this inflammatory response and its recovery period appears viable in saliva, which can be done more easily (e.g., in the field) and in a non-invasive manner compared to blood. One additional point: given that such measures are increasingly being utilized in PNI studies, both the stability and replicability of salivary inflammatory measures need to be delineated in future research.

Stress and salivary secretory immunoglobulin A (S-IgA)

Most infections (~95%) start at mucosal surfaces such as the lining of the respiratory and gastrointestinal tracts. These surfaces are protected by various antimicrobial proteins, which are secreted by local glands and form a first line of immune defense. Among these proteins, secretory immunoglobulin A (S-IgA) represents the main agent of adaptive immunity [61,62]. S-IgA can bind to antigens, such as microbes and toxins, in a highly selective manner which prevents such antigens from attaching to or penetrating mucosal surfaces. There are several compelling reasons to study salivary S-IgA in behavioral medicine. First, the salivary glands are strategically located at the portal of entry to the respiratory and gastrointestinal tracts, and play a key role in the maintenance of oral health [63,64]. Second, salivary S-IgA may serve as a representative model of IgA-secreting B-cells across different mucosal sites [65]. In this manner, salivary S-IgA may provide a general overview of the entire secretory immune system.

A third reason is that the release of S-IgA is under strong neuroendocrine control [66]. The autonomic nerves that innervate the salivary glands robustly impact salivary S-IgA secretion, whereby activation of the sympathetic nerves enhances S-IgA output [66,67]. This mechanism is thought to explain why acute stress generally increases salivary S-IgA [68–72]. Conversely, protracted stress is typically associated with decreased levels of S-IgA, possibly through cortisol’s inhibitory effects on S-IgA concentrations [68,69,73–75].

S-IgA is released into saliva in a two-step process [63]. First, IgA-producing B lymphocytes, which have migrated into the glandular tissues, produce and release IgA. The IgA molecule is then shuttled through the glandular cell, via binding to the transporter molecule Secretory Component (SC), and subsequently released into saliva as S-IgA (i.e., the IgA-SC complex). This shuttling by SC also takes place when no IgA is bound, in which case only SC is released into saliva [63]. Hence, S-IgA and SC serve as indicators of IgA production and glandular transport capacity, respectively. By measuring both S-IgA and SC it can be estimated if stress-related S-IgA reduction is attributable to a decrease in IgA release by B-lymphocytes or a reduction in glandular transport capacity. The reductions in S-IgA concentrations during chronic stress appear to be due to lower availability of IgA; transport capability remains stable or increases in individuals who report high protracted stress levels [74]**,[76].

Human S-IgA is secreted as two distinct isotypes; IgA1 and IgA2. Differentiating between these isotypes is important as concentrations of S-IgA1, but not S-IgA2, are associated with an increased vulnerability to upper respiratory tract infections [77]. Interestingly, both acute and protracted stress appear to have a larger impact on S-IgA1 levels than on S-IgA2 [74,78].

Very little research on S-IgA has studied antibody release in response to antigen [62,79,80]. This is an area of future research, now that oral vaccines have been well-validated and are used more routinely [81]. Research on salivary S-IgA should also be developed with more attention to methodological issues. For example, flow rate should preferably be measured in addition to concentration (ug/ml) so that S-IgA output (ug/minute) can be estimated; this allows one to determine if changes in concentrations are secondary to flow rate (e.g., dry mouth during stress) [63].

In general, for optimal results in measuring salivary biomarkers (other than cortisol), we recommend avoiding absorbent materials, collecting unstimulated saliva samples (e.g., by passive drool), and recording flow rates (either by volume or weight per minute of collection) [63,82,83]. In addition, some published studies report using commercial assay kits that have not been validated for use in saliva; this may produce erroneous values. For a broader discussion of these methodological issues, see recent reviews [16,51]; see Box 1 for a summary of these issues.

Box 1 – Tips to the researcher using salivary biomarkers.

• Use a collection method that is proven to work for the salivary parameter to be measured

• To reliably assess diurnal activity of cortisol and salivary alpha-amylase, collect at least three daily profiles, and use the wake-up concentration as anchor for daily slope

• Avoid absorbent collection materials and quantify salivary flow rate

• Ensure assay kits are validated for use with saliva

• Assessing oral health (even 1–2 rudimentary questions) may be helpful in research involving salivary inflammation

Immunosenescence and telomeres

Another way that stress can alter immune function is through promoting cell senescence. Similar to chronological aging, prolonged psychological stress has been related to telomere shortening; in this manner, stress appears to contribute directly to cellular aging. Thus, oral tissues can provide insights into biological aging. The problem with using saliva samples for the determination of telomere length is that that the results are highly variable due to mixed cell types (i.e., saliva contains both leukocytes and buccal cells) [84]. A buccal (cheek) swab, which is equally non-invasive, will often yield better (less variable) results for research. For a recent review on psychological stress and telomeres see [85]*.

Future perspectives

See Box 2 for a summary of key points from the above review. As technology advances, other immune-related measures in saliva are increasingly being reported in PNI research. Such markers include cytokines with different primary functions than inflammation (e.g., IL-7, IL-12p70, IL-23, interferons), growth factors, adiponectin, anti-microbial proteins (e.g., endogenous lysozymes), and full characterizations of the oral microbiome. In addition to basic research, salivary biomarkers hold immense promise in the field of medicine. Saliva is increasingly being used to determine responses to vaccination, and has become important in the early detection of various cancers through the measurement of circulating tumor cells and tumor DNA fragments (i.e., liquid biopsies). For a recent review of the diagnostic potential of saliva in disease (i.e., to inform clinical decisions and as a predictor of post-treatment outcomes) see [86]*.

Box 2 – Summary of key points.

In saliva:

• Alpha-amylase is a good proxy for autonomic activity and reactivity and might reflect central noradrenergic activity

• Inflammation is not a reliable indicator of systemic inflammation

• Inflammatory responses to stress appear to occur reliably; in this manner, salivary cytokines may be useful measures in PNI research

• S-IgA levels tend to drop under conditions of stress, especially S-IgA1

Highlights:

• Salivary cortisol and amylase have advanced the understanding of stress biology

• Salivary inflammation appears to reliably increase following stress

• Salivary S-IgA may provide a good general overview of the secretory immune system

• Saliva may give unique insights into the effects of stress on health in PNI