Changes in Severity of Allergy and Anxiety Symptoms Are Positively Correlated in Patients with Recurrent Mood Disorders Who Are Exposed to Seasonal Peaks of Aeroallergens

Abstract

Considering clinical and animal evidence suggesting a relationship between allergy and anxiety, we hypothesized that, from low to high aeroallergen exposure, changes in anxiety symptom scores in patients with primary mood disorders will correlate with changes in allergy symptom scores. We also anticipated that sensitization to tree pollen, as determined by allergen specific IgE antibodies, will predict a greater worsening of anxiety during exposure to tree pollen. 51 patients with unipolar or bipolar disorder (age: 19-63 years, 65% female) were recruited. Tree- pollen IgE positive subjects (12) were included as the experimental group and patients negative to a multi-allergen serological test (39) were included in the control group. Self reports of anxiety and allergy symptoms were obtained once during the peak airborne pollen counts and once during the period of low airborne pollen counts, as reported by two local pollen counting stations. Using linear regression models, we confirmed a significant positive association between allergy scores and anxiety scores (p<0.04); however, the IgE specific tree pollen positivity was not significantly associated with changes in anxiety scores. Because changes in anxiety scores relate to changes in depression scores, the relationship between allergy and anxiety involves states rather than only traits, and as such, our results lead to future efforts to uncover potential anxiety triggering, exacerbating or perpetuating role of allergens in vulnerable individuals.

Introduction

Allergic disease is a prevalent condition, with approximately 50% of the American population sensitized to at least one allergen (NHANES II and III) (1). Worldwide, a peak in suicide in the spring is consistently observed in epidemiologic research studies (2-8). In the fall, a second, smaller peak in suicides has also been observed, although it has been less well replicated (7,9-12). Such seasonal peaks in suicide correspond temporally with the peaks in aeroallergens that are measured in the environment. The spring peak is contemporaneous with high airborne tree pollen counts, and the fall peak with high weed counts, particularly ragweed in the United States. We have preliminarily reported an association between peaks in airborne tree pollen and springtime peaks in nonviolent suicide in women.

While depression is the disorder most frequently associated with suicide (14), anxiety has also been found to independently contribute to increased suicide risk (15,16). Comorbid anxiety further increases the risk of suicide (17), especially when comorbid with bipolar disorder (18) or depression (19-21). In fact, comorbid depression and anxiety also peaks in the spring and again in the fall, with the seasonality of comorbidity greater than that of depression alone (21). Comorbid anxiety and depression symptoms are also commonly reported in allergic patients (22).

We have recently found a preliminary association between symptoms of upper airway inflammation and depression (23) and between the seasonality of mood and self-reported mood sensitivity to high pollen counts (24). Additionally, in an animal model intended to examine the effects of allergic sensitization to tree pollen on depressive symptoms, we found significant anxiety-like behavior across trials in the sensitized animals following exposure to tree pollen (25). In individuals with allergic sensitization, when mast cell bound IgE antibody is crosslinked by specific allergens, an activating signal is transduced which results in mast cell degranulation and the release of inflammatory mediators and cytokines. Clinical investigation suggests that Th2-type lymphocytes are predominantly activated in allergic diseases. Th2 cells are characterized by their production of IL-4, IL-5, and IL-13.

Cytokines, administered in amounts below the threshold necessary to induce “sickness behaviors,” have been shown to induce anxiety, depression, and cognitive disturbances in healthy subjects (26). An increase in cytokine levels in the blood has been hypothesized as one potential catalyst for the decompensation of depression (27-29). Cytokine-treated patients may also experience an increase in depressive symptoms, including suicidal ideation and attempted suicide (30-33). Even a low dose of cytokine-promoting endotoxins such as lipopolysaccharides (LPS), can trigger depressive symptoms along with anxiety without bringing about other sickness behaviors (26). Certain cytokines released during LPS-induced inflammation are also released during the allergic response. For instance, mast cell degranulation releases TNF-α (34), the administration of which has been shown in animal models to be anxiogenic (33). We have seen that sensitization and subsequent exposure to tree pollen (25) and intranasal LPS administration (35) induce anxiety-like behaviors in sensitized rodents. Additionally, we have reported increased gene expression of cytokines involved with allergic inflammation (36) in the orbital cortex of suicide victims, where histopathological changes in suicide victims have been previously reported (14).

We hypothesized that a) that during the intervals of high tree pollen, the anxiety level should be greater in individuals who are sensitized, that is IgE antibody positive, to tree pollen allergens., and b) the changes in anxiety scores between the high and low pollen periods will be positively correlated with changes in allergy symptoms.

Methods

This is an auxiliary hypothesis driven analysis of data collected for a primary study on changes in depression and allergy symptoms. The protocol has been approved by the Institutional Review Board at the University of Maryland School of Medicine. We recruited 51 participants, 18 to 65 years of age with a past clinical diagnosis of Major Depressive Disorder or Bipolar Disorder, by means of radio and newspaper advertisements. Recruitment took place at two sites, Baltimore and Washington, DC, between September 2006 and June 2007.

A pre-screening interview was performed by telephone for all subjects in order to exclude those with any major medical illness, previous psychotic disorder, winter-type seasonal affective disorder, drug or alcohol dependence, or current or intended pregnancy during the course of the study.

We obtained informed consent following a thorough explanation of the study. A semi-structured questionnaire was administered that was designed to measure the comprehension of the consent, before continuing any study procedures. At this time, participants were further screened with a questionnaire and they provided a blood specimen to determine the presence of aeroallergen specific IgE antibodies as an indicator of allergic sensitization. The subject's serum was initially analyzed with a widely used multi-allergen screen, Phadiatop, ImmunoCAP 250, Phadia, Uppsala, Sweden, that detects IgE antibodies specific for any of 15 common weed, grass, tree, mite and pet epidermal aeroallergens, including the principal seasonal allergen specificities relevant to the hypothesis of our study [37]. To verify each subject's specific sensitization to spring pollen, tree, and/or fall, ragweed, aeroallergens, serum from each Phadiatop positive subject was then reanalyzed for IgE antibody to individual allergens using the ImmunoCAP 250. Quantitative levels of IgE antibody specific for the trees ash, beech, birch, elm, maple, poplar, sycamore, oak and the weed ragweed pollen allergens. IgE antibody results > 0.35 kUa/L were considered a positive result, indicative of the presence of IgE antibody to that allergen specificity in the blood (38).

Patients were additionally screened with a Structural Clinical Interview for DSM-IV (SCID) (39) using the mood disorders, substance abuse, and psychotic disorders modules. For inclusion in the study, the individual had to have unipolar or bipolar depressive disorder. Patients were excluded if they had an active substance abuse or dependence or a psychotic disorder as assessed by the SCID. Exclusion of subjects with substance abuse was considered important because, in addition to directly altering mood states, the drugs can directly alter mediator release from effector cells of allergic inflammation and potentially reduce peripheral blood cytokine levels. In addition, the use of intranasal corticosteroids or montelukast led to exclusion because of the locus of their action and its potential for reducing cytokine levels. Anti-histamines and decongestants, however, are not known for their central effects on the allergic cascade, instead acting to relieve symptoms of allergic reaction, and therefore their use was not a criterion for exclusion.

Based on the results of the blood tests, patients were subdivided into IgE -positive or sensitized subjects in the test group and Phadiatop-negative subjects who were considered the control group. The participants and the raters were masked to the specific IgE results or group assignment throughout the course of the study.

Data on daily airborne pollen counts were obtained from two local, official pollen-counting stations in Washington, DC and Baltimore that are recognized by the Aerobiology-Aeroallergen Monitoring Network of the /American Academy of Allergy Asthma and Immunology (AAAAI). Pollen data from these counting stations were e-mailed to the study center daily and, starting in February 2007, the data from these stations were recorded daily on the AAAAI website (www.aaaai.org). A research assistant entered and verified the aeroallergen data and matched the genus of tree pollen counts to the specificity of the IgE in the blood of each study subject. The periods of high or low levels of atmospheric tree pollen concentrations were identified as after January 15 and before July 1st of 2007. The pre-pollen interval began when the subject's first tree pollen allergen count exceeded 10 grains/m3. The cutoff pollen count, which identified a peak tree pollen period, was = 90 grains/m3. The actual timing of the first visit was designed to be one of convenience. For some participants, their low tree pollen period visit preceded the peak tree pollen season, and for others it followed the peak tree pollen period.

Anxiety and Allergy Rating Examinations

Anxiety symptoms were measured with the Burns Anxiety Inventory (BAI) (40), a self-reported questionnaire often used in a clinical setting. The severity of five symptoms of allergic disease, namely sneezing; rhinorrhea; itchy nose, throat or palate; itchy, watery, or red eyes; and nasal congestion, were measured using the Allergy Symptom Severity Assessment (ASSA)(41). The ASSA is also a self-reported instrument, using a scale from 0 to 4, where 0 is no symptoms; 1 is mild symptoms but no discomfort; 2 is moderate symptoms with discomfort but no interference in functioning; 3 is severe symptoms with discomfort such that it interferes with functioning; and 4 is severe symptoms such that medical attention is required. For the primary study on depression, patients were also evaluated using the Structured Interview Guide for the Hamilton Depression Rating Scale – Seasonal Affective Disorder Version (SIGH-SAD)(42).

Data Management and Analysis

Preliminary analysis included descriptive methods in which the study data were characterized with a mean and standard deviation for continuous variables, and percentage of positivity for categorical variables. The IgE anti-tree pollen positive group data were compared to the negative group data using Student's t-test and chi-square analysis. Differences in the measures of allergic disease or anxiety symptoms during the low and high pollen periods were computed using the BAI scores and the ASSA data. Linear regression analysis was performed to identify significant associations between a change in ASSA and the BAI scores. These measures were adjusted for factors including gender, and the order of the low pollen and high pollen interviews. A second model included IgE tree pollen specific IgE antibody positivity. In a third, more parsimonious model, the ASSA score was replaced by nasal congestion alone. Posthoc, change in the total SIGH-SAD depression score was added to the model to assess whether depression mediates the association between allergy and anxiety, and medication use was considered for inclusion in the models. Statistical analysis was performed using SAS version 9.1 software.

Results

To date, 51 individuals completed the study. Of these, 12 individuals (24%) were positive for tree pollen specific IgE (IgE+) and 39 (76%) were IgE antibody negative for tree pollen (IgE-). From this sample, 78.4% of the participants had their low pollen interview during their second visit, i.e., preceding their high pollen interview on the third visit. For IgE+ individuals, 41.7% had their low pollen interview before their high pollen interview. For IgE-individuals, 89.7% had their low pollen interview before their high pollen interview. Table 1 presents these characteristics of the participants as well as their demographic information.

Table 1. Characteristics of study participants, all combined, and by IgE Tree Pollen Positivity.

	Total	IgE Tree Pollen Positive
Variable	N=51	Yes, N=12 (24%)	No, N=39 (76%)	p value*
Age, mean (SD)	45.4 (11.1)	41.8 (13.1)	46.5 (10.4)	0.20
Gender male (%)	35.3	41.7	33.3	0.60
Race (%)
Caucasian	76.5	66.7	79.5	0.47
African-American	21.6	33.3	18.0
Hispanic	2.0	0.0	2.6
Allergy symptoms (%)
Low pollen	80.0	81.8	79.5	0.86
High pollen	95.7	90.9	97.2	0.36
Burns Anxiety Score, mean (SD)
Low pollen	20.3 (15.6)	22.8 (17.6)	19.6 (15.2)	0.55
High pollen	20.7 (17.2)	24.1 (23.9)	19.7 (14.8)	0.57
Low pollen visit before high pollen visit (%)	78.4	41.7	89.7	0.0004

^*p values from comparisons of tree positives and negatives, t-test comparing means, and chi-square comparing proportions.

During the low pollen period, 80.0% of all participants reported experiencing allergic symptoms, with 81.8% of IgE+ and 79.5% of IgE- groups reporting symptoms. During the high pollen season, 95.7% of all participants reported allergic symptoms, with 90.9% of IgE+ and 97.2% of IgE- reporting symptoms. During the low pollen season, the average of all participant BAI scores was 20.3±15.6 (mean±SD). The average BAI score for IgE+ individuals during that period was 22.8±17.6 (mean?SD), and for IgE- individuals it was 19.6±15.2 (mean±SD). During the high pollen season, the average of all participant scores was 20.7±17.2 (mean±SD). The average score for IgE+ individuals in that period was 24.1?23.9 (mean±SD); however, there was no significant difference with the BAI score of IgE-individuals, which then averaged 19.7 ±14.8 (mean±SD). Table 1 reports average participant scores on the BAI and ASSA by IgE status as well as combined. Table 2 lists the medications used for all participants and by tree pollen specific IgE antibody status.

Table 2. Medication usage for tree positive vs. tree negative participants.

	Anti-depressants		Mood Stabilizers		Anxiolytics		Anti-psychotics		Thyroid Hormones		Sleep Medications
	%	n	%	n	%	n	%	n	%	n	%	n
Tree pollen negative (n = 39)	90	35	28	11	26	10	18	7	21	8	13	5
Tree Positive (n = 12)	75	9	25	3	8	1	25	3	8	1	8	1
Total (n = 51)	86	44	27	14	22	11	20	10	18	9	12	6

The linear regression analyses revealed a significant association between the ASSA score and an increase in the BAI score (p=0.04; table 3). One interpretation of the regression coefficient is that, after adjustment for gender and visit order, an increase by one unit was observed between a high versus low pollen ASSA score was associated with an increase of 1.14 in the BAI score.

Table 3. Results from linear regression models for associations of change in Allergy Symptom Severity Score (Model 1) and IgE Tree positivity (Model 2) with change in Burns Anxiety Score, adjusted for gender and order of visit. In Model 3, nasal congestion replaced the allergy symptom score.

	Burns Anxiety Score (High - Low Pollen Score)
	Model 1			Model 2			Model 3
Independent Variable	β	SE	p-value*	β	SE	p-value*	β	SE	p-value*
Allergy Symptom Score (high - low pollen)	1.14	0.54	0.04	1.15	0.54	0.04	-
Gender male	2.04	3.64	0.58	1.40	3.66	0.70	1.76	3.67	0.63
IgE Tree Positive (Yes vs No)				5.93	4.88	0.23	-
Low pollen visit before high	10.99	4.27	0.01	13.87	4.87	0.01	9.52	4.26	0.03
Nasal congestion	-			-			7.67	3.78	0.05

^*p-value for coefficient from linear regression model.

In a second model, the IgE specific tree pollen positivity was not significantly associated with the BAI score and, in a multivariable model, did not affect the significance of the ASSA relationship with the BAI score.

Finally, in a third model, the presence of nasal congestion, considered independent of the ASSA score, was associated with a significant 7.67 unit increase in the BAI score, adjusted for gender and visit order. Posthoc, when we included the SIGH-SAD score of depression in the model, it was highly significantly associated with the BAI score and the ASSA score lost significance. When use of either anxiolytic or antidepressant medication was added to models, neither was significantly associated with the BAI score.

A significant association (p=0.0004) was observed for the order of the visits with change in the BAI score, such that those individuals whose low pollen visit preceded the high pollen visit showed a greater increase in the BAI score. Such order effects were taken into account in the linear regression analysis.

Discussion

In the present study, we have confirmed one of our two hypotheses. Change in the severity of allergic symptoms was positively correlated with changes in the severity of anxiety symptoms. This is consistent with previous research suggesting a connection between allergic inflammation and anxiety (22,25,43,44)]. On the other hand, we did not find, as we had hypothesized, a link between specific IgE antibody status and anxiety scores, possibly as a result of the low number of participants who were positive for tree pollen specific IgE antibody.

The results of the present study should, however, be interpreted with caution, considering several limitations. First, as this was an addition to our main depression study, we did not use the anxiety disorder module in the SCID to diagnose anxiety disorders. Second, we used the presence of pollen specific IgE antibody as an indicator of sensitization, but we did not quantify the magnitude of the sensitization. We preferred, instead, to use the specific IgE test rather than puncture skin tests because psychotropic medications may have anti-histaminic effects that could have interfered with the interpretation of the skin test.

The order of high pollen and low pollen interviews was not randomized. The post-pollen season may not represent a low pollen period in the same way as the pre-pollen season given the presence of grass pollen at the later time. Further limitations may lie in the instruments used to assess allergic and anxiety symptoms. Both the Allergy Symptoms Severity Assessment and the Burns Anxiety Inventory are self-reports. This may lead to concern regarding whether and to what extent individual features, such as neuroticism, may have altered both self-ratings and thus contributed a spurious relationship. However, data from a recent study specifically examining the role of neuroticism in affecting the association between allergy symptoms and depression have indicated that its contribution is absent in women and small in men (45).

We can speculate as to several possible mechanisms mediating a relationship between the severity of anxiety symptoms and allergic symptoms. First, the psychological effects of being ill, as well as impairments of sleep secondary to nasal obstruction or inflammation may affect mood. Second, cytokines may lead to a mood disturbance via the expression of the enzyme indoleamine-2,3-dioxygenase, which shifts the synthesis of tryptophan from serotonin to kynurenin (13,43,46). The resultant acute tryptophan depletion results in decreased brain serotonin, which may contribute to both depression and anxiety. Third, cytokines released during allergic inflammation may affect the brain directly, via nerves, surrounding tissue, or via regions that do not have a blood brain barrier, such as the circumventricular organs. Other explanations for the behavioral change may exist.

The amygdala, in mediating the fear response, is associated with anxiety through its connection to the hippocampus, which controls contextual conditioning, and the prefrontal cortex (PFC), responsible for extinction. Anxiety results when the context of fear-inducing circumstances is misconstrued or when the fear response does not become extinct even in the presence of innocuous circumstances (47). The PFC ordinarily inhibits and suppresses activity in the amygdala, often through the regulation of the hippocampus (48). However, the PFC also receives inputs from the systems, involved in the regulation of behavioral, autonomic, and endocrine responses, to which the amygdala projects. In this way, emotional stimuli serve to stimulate the PFC more than do other stimuli, resulting in a reciprocal process (49). Thus, the PFC, which usually allows for responses to stimuli to be changed as necessary, when damaged or inhibited prevents a response other than fear from occurring even when circumstances may not be noxious (48). It is possible that the orbital cortex and its regulation of the amygdala may be affected by cytokine expression, leading to a decrease in amygdala inhibition and an increase in anxiety. We have observed increased gene expression in the cytokines involved in allergic reactions in that region of the brain in victims of suicide (36).

Cytokines may also be involved in changes in the hypothalamic-pituitary-adrenal (HPA) axis, as well as in the production of corticotropin-releasing hormone (CRF), adrenocorticotropic hormone (ACTH), and cortisol or corticosterone (26,44,50,51). A sustained increase in activity in the HPA axis is associated with severe anxiety (14). A recent study using an animal model has (44) indicated that the degranulation of mast cells, a central cellular element of the anti-parasitic and allergic response, was necessary to see an increase in activity in the PVN and central amygdala. The former is responsible for hypothalamic secretion of ACTH and activation of the sympathetic autonomic nervous system; the latter for the release of CRF and its consequent stress response. In addition, Breese et al. found that cytokines might lead to a physiological change in GABAA receptor functioning, possibly serving to limit the activity of its agonists, often anxiolytic (51). Allergen-induced peripheral inflammation could also be linked with inflammatory processes in the brain underlying observable anxiety-like behaviors (44,52).

The percentage of individuals reporting upper respiratory symptoms (more than 75% during low pollen symptoms and more than 90% during high pollen season), exceeded the expected prevalence estimate based on the proportion of subjects who were IgE antibody positive. It is likely that non-allergic upper respiratory symptoms contributed to these results. Non-IgE mediated allergic rhinitis is triggered by environmental conditions and substances other than allergens which include changes in the weather, irritants, or pollutants (53). Alternatively, it has been recently shown that some individuals may have a ‘local form’ of allergy disease involving IgE produced only in the nasal mucosa. They develop nasal symptoms in the absence of detectable serum IgE or a positive skin test to prevalent environmental allergens (54). Finally, exogenous proteases from pollens and other aeroallergens may cause inflammation by mechanisms other than IgE activation of effector cells, possibly acting on molecules such as protease-activated receptors (PARs) (55-57).

We are lacking convincing evidence that such a mechanism is, indeed, operative in rhinitis. In any case, our results, which are consistent with our animal data (35,52), indicate that anxiety symptoms are correlated with upper respiratory inflammatory symptoms.

The relationship between anxiety and allergy scores became non-significant when we adjusted for depression scores for covariates. This suggests that the depression and anxiety scores are highly intercorrelated, with anxiety symptoms as a component of the depressive syndrome, or indicative of comorbidity between anxiety and depression. On the other hand, the relationship between depression scores and allergy symptom scores remained significant after adjustment for anxiety symptoms. These data suggest that the relationship between allergic disease and depression is a major phenomenon, and seemingly more robust than the one between allergic disease and measures of anxiety. However, our animal model data point towards a more consistent association between anxiety-like behaviors rather than depressive-like behaviors and with sensitization and exposure to tree pollen allergen (35,52). To further analyze this triangular relationship between allergy, depression, and anxiety, we will continue our research with primary anxiety disorders and primary non-comorbid mood disorders.

Despite its preliminary nature, the current study is conceptually important. One could suppose that the relationship between allergic disease and anxiety, or depression, represents mainly common vulnerabilities, possibly genetic or developmental. However, identifying a relationship between changes in allergy symptoms and changes in depression symptoms strongly suggests a relationship that affects states rather than traits. This concept, which leads to an investigation of the causality, triggering, precipitation, and exacerbation of anxiety disorders is important for designing potential preventive and curative interventions.

Additional research is necessary to confirm our currently reported relationship between changes in self-reported allergy symptoms and anxiety symptoms, with better instruments for measuring anxiety and more precise measurements of allergic inflammation. Studies in humans and animals, involving immunological and psychotropic interventions, will be essential for mechanistic and therapeutic understanding and may contribute to the generation of novel targets and interventions for anxiety symptoms and disorders.