- Email: [email protected]
Carbon dioxide induces erratic respiratory responses in bipolar disorder
Journal of Affective Disorders 112 (2009) 193 – 200 www.elsevier.com/locate/jad
Research report
Carbon dioxide induces erratic respiratory responses in bipolar disorder Dean F. MacKinnon ⁎, Brandie Craighead, Laura Lorenz Department of Psychiatry and Behavioral Sciences, Johns Hopkins University, Baltimore Maryland Received 16 January 2008; received in revised form 17 March 2008; accepted 24 March 2008 Available online 20 May 2008
Abstract Background: CO2 respiration stimulates both anxiety and dyspnea (“air hunger”) and has long been used to study panic vulnerability and respiratory control. High comorbidity with panic attacks suggests individuals with bipolar disorder may also mount a heightened anxiety response to CO2. Moreover, problems in the arousal and modulation of appetites are central to the clinical syndromes of mania and depression; hence CO2 may arouse an abnormal respiratory response to “air hunger”. Methods: 72 individuals (34 bipolar I, 25 depressive and bipolar spectrum, 13 with no major affective diagnosis) breathed air and air with 5% CO2 via facemask for up to 15 min each; subjective and respiratory responses were recorded. Results: Nearly half the subjects diverged from the typical response to a fixed, mildly hypercapneic environment, which is to increase breathing acutely, and then maintain a hyperpneic plateau. The best predictors of an abnormal pattern were bipolar diagnosis and anxiety from air alone. 25 individuals had a panic response; panic responses from CO2 were more likely in subjects with bipolar I compared to other subjects, however the best predictors of a panic response overall were anxiety from air alone and prior history of panic attacks. Limitations: Heterogeneous sample, liberal definition of panic attack. Conclusion: Carbon dioxide produces abnormal respiratory and heightened anxiety responses among individuals with bipolar and depressive disorders. These may be due to deficits in emotional conditioning related to fear and appetite. Although preliminary, this work suggests a potentially useful test of a specific functional deficit in bipolar disorder. © 2008 Elsevier B.V. All rights reserved. Keywords: Bipolar disorder; Panic disorder; Hypercapnia; Fear; Appetite; Respiration; Carbon dioxide
1. Introduction Experimental inhalation of carbon dioxide has illuminated mechanisms of respiratory physiology and anxiety and also may shed theoretical light on affective disorders. At a fixed CO2 concentration of 5%, charted over 10 min or more, the natural respiratory response resembles a ⁎ Corresponding author. Meyer 3-181, 600 N Wolfe St, Baltimore, MD, 21287. Tel.: +1 443 287 6330; fax: +1 410 955 0152. E-mail address: [email protected] (D.F. MacKinnon). 0165-0327/$ - see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2008.03.020
plateau: minute ventilation and end tidal CO2 rise sharply from baseline then settle at a steady, elevated level (Haywood and Bloete, 1969; Padget 1928; Reynolds et al., 1972). Using the same method, panic attacks can be reliably and selectively induced in individuals with prior panic attacks (Cohen and White, 1951; Papp et al., 1993; Rassovsky and Kushner, 2003; Sanderson and Wetzler, 1990). Although major depressive disorder raises the risk for spontaneous panic attacks (Roy-Byrne et al., 2000), depression alone does not confer an increased risk for CO2-induced panic (Kent et al., 2001). Panic attacks are
194
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
also common with bipolar disorder (Chen and Dilsaver, 1995; Goodwin and Hoven, 2002; MacKinnon et al., 2002); it is unknown whether bipolar disorder confers elevated risk for CO2-induced panic. The anxiogenic effect of CO2 is probably due to central nervous system arousal (Bailey et al., 2003), but the response to CO2 breathing may also hinge on its function as a mediator of respiratory appetite. Respiration normally is automatic, thus does not involve an appetitive cycle of drive, satiety, and reward. Dyspnea is an almost irresistible drive (Rafferty and Gardner, 1996); its satiation registers as reward (Peiffer et al., 2008). Yet, in the fixed 5% CO2 protocol, where drive is stimulated but satiety is blocked by the inspired CO2, there is typically an accommodation to a hypercapneic state. Why does one not continue to escalate one's respiratory effort? The plateau phenomenon suggests that the adaptive response to CO2 breathing is to find an optimal compromise of dyspnea and effort (Poon, 1987). Adaptation to a heightened plateau of CO2 and minute ventilation hinges on learning (implicitly or explicitly) that it is futile to continue to escalate respiratory effort; eucapnia can never be achieved. If this adaptive aspect of respiratory behavior is conditioned centrally (Mitchell and Johnson, 2003), as other appetitive behaviors appear to be (Everitt et al., 2003), then some perturbations of respiration may be apparent in a disorder that affects the central regulation of appetitive behavior — such as bipolar disorder (Depue and Iacono, 1989; Johnson et al., 2000). Irregular patterns of respiration were noted in our initial bipolar sample, but not analyzed formally or in detail (MacKinnon et al., 2007). In this extended sample we hypothesize that individuals with bipolar disorder will be more likely than control subjects to mount a respiratory response that deviates from the normal plateau shape. This result would be consistent with a putative functional deficit in appetitive regulation that prevents optimization of the drive to achieve the minimal CO2 possible with sustainable effort. In contrast, the CO2 sensitivity in depressed individuals has been shown to be depressed (DamasMora et al., 1978; Shershow et al., 1976), consistent with a generally inhibited appetitive drive in depression. Based on our prior observations we also expect there to be some CO2-induced panic with bipolar disorder, but less so that in studies of individuals with panic disorder. 2. Methods
excluded if they had an unstable medical condition or psychiatric status at the time of study, or could not abstain from alcohol or illicit drugs for the 2 days prior to the experiment. All were able to provide informed consent. A subset of this group has been described elsewhere, but not compared to a control group (MacKinnon et al., 2007). 2.2. Protocol Subjects initially reviewed the consent form, then met with the PI in his office to discuss the procedure, affirm their informed consent, have a diagnostic interview (if necessary) and provide data on present mental and physical health. They were then escorted to the laboratory, where rating scales were administered. The research assistant was not present during the pre-procedure interview, and the PI was not in the room during the respiratory protocol (the PI observed, without subjects' knowledge, via 2-way mirror). Subjects were randomized to receive either air or 5% CO2 in air as the first arm, and were blinded to the content of the gases. Upon completing the baseline rating scales, subjects were fitted with a face mask linked to a 4-way valve, through which they breathed compressed room air and compressed room air mixed with 5% CO2 for 10–15 min each. Subjects rested in a comfortable reclining chair throughout the procedure, and were allowed to sit back or elevate feet if desired; they were instructed not to speak during the procedure. Between trials the mask was removed and subjects rested for a minimum of 10 min, until anxiety had returned to baseline. Rating scales were completed before, at 5 min intervals during, and immediately after each trial. 2.3. Diagnostic assessment Forty seven of the 72 subjects in this analysis were also participants in psychiatric genetics research at Johns Hopkins. With subjects' consent, diagnoses were extracted from the genetics database, having been obtained via psychiatrist or masters-level research associate assessment using the Diagnostic Interview for Genetic Studies (DIGS) (Nurnberger et al., 1994). For subjects not enrolled in a genetics study, the PI administered an abbreviated version of the DIGS, focusing on depression, mania, and panic disorder items. In ambiguous cases, medical records were requested, with the subject's permission, to clarify a diagnosis.
2.1. Subjects 2.4. Subjective ratings Study subjects were primarily individuals enrolled in bipolar disorder genetics research; additionally, some responded to direct advertisement. Subjects were
Forty two of the 72 subjects completed Beck Depression and Anxiety Inventories (Beck 1967; Beck
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
et al., 1988) prior to the CO2 trial. Before and at intervals during the procedure, subjects completed a rating scale of panic attack symptoms (13 items, rated from not present to severe), and an analog rating (0 to 100 mm) of anxiety and physical discomfort. Immediately post-trial, subjects were asked whether the respiratory procedure caused them to have a panic attack, and whether they had experienced panic attacks in the past. 2.5. Respiratory data Breathing was monitored via Korr RSS 100 spirometer. Respiratory data were recorded on a desktop computer in raw form as the flow (in or out) each 0.02 s; from this the RSS software generated a “parameter file” which established breath size and time interval for each breath, based on changes in the flow pattern. All raw data were rendered graphically and inspected by the PI (who was blinded at this phase to the diagnosis and panic outcome), to detect and resolve problems with breaths missed or miscalled by the breath parameter algorithm. All 72 subjects were included in the primary analysis of panic response; for the analysis of individual response patterns and logistic regression analyses (see below) we required a minimum of 10 min of useful CO2 data. In 4 cases subjects interrupted the CO2 arm before the 10 min mark, and in 6 cases the trial was limited by technical issues (gas supply or computer problems), thus n = 62 for these analyses. To facilitate comparison of respiratory data across individuals, the primary variables tidal volume, breath frequency (time from peak inspiration to peak inspiration), and minute ventilation (sum of tidal volumes for each 60 s epoch) on CO2 were divided by the means of these measures on air alone. Thus, an adjusted minute ventilation (aMV) of 2 means the subject consumed twice as much air + CO2 per minute compared to air alone. 2.6. Data analysis A panic response was determined either by self report or by ratings that indicated moderate anxiety (N50 on the analog rating scale) plus four or more typical panic symptoms. We use the term “panic response” rather than “panic attack” to differentiate this outcome from the outcome in other studies that use a more stringent definition of “panic attack”. Statistical significance was calculated using Student's t-test (2-tailed) for continuous variables, and chi-square for dichotomous variables. These, as well as regression slope calculations, were performed using the
195
Microsoft Excel data analysis package. Logistic regression analyses were performed using Stata software. In comparisons based on affective disorder diagnosis, we used three diagnostic categories: BiPolar I (BPI: subjects were given a confident diagnosis of bipolar disorder type I by an interviewer), Other Affective Disorder (OAD: subjects were diagnosed with either bipolar II with recurrent depression, major depressive disorder, recurrent, or “bipolar NOS”, denoting subthreshold or questionable manic/hypomanic syndromes), and No Affective Disorder (NAD: subjects with no major mental illness, as well as subjects with non-affective or questionable affective diagnoses). Analysis of individual response patterns applied formal criteria to determine whether a subject's respiratory response was consistent or inconsistent with the normal plateau response described in the physiology literature. Specifically, a “normal” response had: 1) an early spike (the slope of aMV vs time for minutes 0–3 was greater than the slope during minutes 4–9 and 10–15); 2) persistent elevation (mean aMV in minutes 6–15 was greater than mean aMV for minutes 1–5); 3) net increase on CO2 vs air (mean aMV for minutes 1–15 N 1); and 4) minimal plateau variance (variance of aMV after minute 5 was less than the mean aMV variance for all subjects on air, which was 0.06). 3. Results Respiratory and subjective responses to air and 5% CO2 were obtained for 72 subjects. Thirty-four had bipolar disorder type 1. Of the remainder 25 had a major affective disorder (bipolar II (n = 7), bipolar NOS (n = 8), or recurrent major depressive disorder (n = 10)), while 8 had no major psychiatric disorder diagnosis and 5 had some other psychiatric diagnosis (schizophrenia (n = 2), antisocial personality disorder (n = 2), depression NOS (n = 1)). Over half (n = 37) had experienced at least one panic attack prior to the study, and 20 were diagnosed on interview with panic disorder. Twenty-five (35%) of the subjects experienced a panic response during the CO2 trial, by either self-report or by panic attack criteria. 3.1. Diagnostic group factors Groups defined by diagnosis–bipolar I (BPI), other affective disorder (OAD), and no affective disorder (NAD)–differed in no significant way in terms of sex, age, body size, or smoking habit (Table 1). Although individuals with bipolar I had histories of panic attacks and panic disorder more often than others, these
196
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
Table 1 Subject characteristics and response By affective diagnostic group Bipolar I
n
(BPI)
Other Affective Disorder (OAD)
34
25
By panic response to CO2
No Affective Disorder
Non-panicking
Panicking
(NAD) 13
47
25
Demographic 43 34.0 25 10
±12 ±8.7 74% 29%
47 32.1 13 7
±10 ±8.5 52% 28%
41 30.5 7 4
±12 ±10.4 54% 31%
– – – –
43 31.8 28 17
±11 ±8.3 60% 36%
47 32.8 17 7
±12 ±7.3 68% 28%
– – – –
Bipolar I Bipolar spectrum Prior panic attacks Panic disorder Panic from CO2 Antidepressant Benzodiazepine
– – 20 12 16 24 4
– – 59% 35% 47% 71% 12%
– – 13 5 6 14 1
– – 52% 20% 24% 56% 4%
– – 4 3 3 2 0
– – 31% 23% 23% 15% 0%
– – – – d a,b –
18 19 21 11 – 23 3
38% 40% 45% 23% – 49% 6%
16 6 16 9 – 17 2
64% 24% 64% 36% – 68% 8%
– – – – –
Self-rating Beck Depression⁎ Beck Anxiety⁎ Baseline anxiety
20 18 27
±6 ±13 ±30
20 21 28
±10 ±17 ±28
15 16 20
±10 ±14 ±24
– – –
18 14 20
±9 +12 ±24
22 26 38
±8.0 ±16 ±32
– e e
Air Minute ventilation (l/min) Maximum anxiety
8.5 34
±2.0 ±30
8.2 33
±2.4 ±31
9.8 24
±1.9 ±24
– –
8.6 22
±2.1 +20
8.7 48
±2.3 ±26
– e
13.3 14.9 53
±2.8 ±4.9 ±30
13.2 11.9 44
±3.0 ±5.5 ±33
14.7 17.5 38
+2.6 ±5.0 ±33
– b,c –
11.7 13.7 33
±3.5 ±5.1 ±21
14.0 15.6 69
±2.5 ±6.0 ±15
– X
Age Basal metabolic index Female Current smoker Clinical
CO2 Time (min) Minute ventilation (l/min) Maximum anxiety
Statistically significant comparisons ( p b 0.05): a) BPI vs NAD; b) BPS vs NAD; c) BPI vs BPS; d) BPI vs (BPS+ NAD); e) non-panicking vs panicking. X = biased comparison. For the comparison of antidepressant and benzodiazepine use n = 67, excluding subjects with no psychiatric diagnosis for which medications are indicated; ⁎n = 42; Other Affective Disorder = BPII, recurrent MDD, BP NOS; no affective disroder inlcudes schizophrenia (2), ASPD (2), and no mental disorder.
differences did not meet statistical significance. Anxiety and mood rating scales and anxiety/discomfort ratings on air alone were not significantly different across diagnostic categories. 3.2. Patterns of individual respiratory response Inspection of minute-by-minute data revealed widespread individual variation (Fig. 1). The modal response across all subjects was a marked elevation in minute ventilation (driven mainly by increased tidal volume) during the first 3 to 5 min, followed by a persistently flat or less acute elevation for the remainder of the CO2 exposure. But many individuals deviated markedly from this pattern. Several had what appeared to be a delayed,
muted or absent response. Many individuals, having peaked early in the course of CO2 exposure, subsequently dropped towards baseline respiration; several of these went on to peak again late in the trial. And a few individuals showed a continuous steep rise throughout the trial, with no evident plateau. Inspection of these patterns, grouped by diagnosis, suggests that subjects without major affective disorder tend to follow the normal plateau pattern, while the subjects with major affective disorders often had aberrant patterns. As described in the Methods section, we defined parameters for a normal time course of response as follows. A normal pattern was postulated to begin with a rapid increase, followed by a persistent but less acute net increase in minute ventilation (relative to air) and minimal
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
197
Fig. 1. Individual respiratory responses to carbon dioxide inhalation. Tracings for 62 individuals with CO2 trials lasting 10 min or more are grouped by affective disorder diagnosis. Each tracing represents the total volume per minute for 10–15 min for one individual breathing 5% carbon dioxide mixed with room air. All values are adjusted for the subject's respiration on air alone, where 1 is no change, N1 is an increase and b1 is a decrease from air. The upper row includes tracings deemed “normal" by algorithm defining the expected steady and sustained respiratory elevation. The lower row includes tracings that failed on one or more criteria.
variability throughout the remainder of the trial. An abnormal pattern failed at least one of these criteria. Of the 62 individuals who remained on CO2 for 10 min or longer, most (55%) of had normal responses by these criteria, and they were distributed unevenly across diagnostic categories. Normal respiratory responses were observed in 92% of the NAD subjects, versus 53% of BPI, versus 35% of the OAD (NAD vs BPI, χ2 = 5.5, p b 0.02; NAD vs OAD χ2 = 9.7, p b 0.005; BPI vs OAD χ2 = 1.6, p = 0.2). On average, NAD individuals met 3.8 of the possible 4 criteria for a normal response, BPI individuals met 3.3 of the 4, and OAD met 2.9 (NAD vs BPI, t40 = 2.3, p = 0.02; NAD vs OAD, t30 = 3.2, p b 0.01; BPI vs OAD t48 = 1.3, p = 0.2). Thus, subjects with a major affective disorder, particularly subjects with bipolar spectrum and major depressive disorders, were significantly more likely to have an aberrant response to CO2. 3.3. Predictors of panic response There was an increased risk of a panic response among those diagnosed with bipolar I compared to all other diagnoses (47% of BPI vs 24% of others; χ2 = 4.3, df = 1, p b 0.04) (Table 1). This was evidently independent of anti-panic agents: 69% of subjects with bipolar I who panicked were on antidepressant/antianxiety medications, versus 72% who did not panic (χ2 = 0.05, df = 1, p = 0.8). We did not find that panic responses were predicted directly by age, sex, body size, current smoking, prior panic attacks, or even prior diagnosis
of panic disorder. The use of antidepressants and antianxiety agents was similarly common in psychiatrically diagnosed subjects who panicked and did not panic. The most robust predictor of a panic response was the anxiety and physical discomfort experienced at baseline and while breathing air via facemask. Those who panicked scored higher on the Beck Anxiety Inventory prior to starting the experiment (26.1 vs 14.5; t40 = 2.7, p b 0.01), and rated their baseline levels of anxiety (pre-facemask), on a scale of 0–100, significantly higher than those who did not panic with CO2 (37.6 vs 19.8, t70 = 2.6, p b 0.01). Their maximum anxiety rating during the air-only part of the trial was over twice as great as that experienced by subjects who did not panic (50.4 vs 21.8, t70 = 4.4, p b 0.00005). The maximum anxiety rating over the CO2 trial, for panicking individuals, was 76.5/100, compared to 31.7/100 for non-panicking individuals. When the analysis is limited to the strictest definition of panic attack–self-report plus panic criteria–and the 9 individuals thus defined are compared to the 47 who neither reported panic nor met criteria for panic, a few new findings emerge. The group that panicked under the strict definition was significantly older (51 vs 43, t54 = 2.0, p b 0.05) and somewhat though not significantly more obese (BMI = 37.3 vs 31.9, t54 = 1.4, p = 0.17). These individuals had two minutes less time on CO2 (11.7 vs 14.0 min, t54 = 2.3, p b 0.05), but otherwise there were no notable differences in the results.
198
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
Table 2 Logistic regression analyses of selected variables to explain panic response and abnormal respiratory response (n = 62) Abnormal respiratory response to CO2
Sex (M) Antidepressant or antianxiety use CO2 induced panic History of panic attack High maximum anxiety—air Affective disorder diagnosis
OR
95%CI
Z
p
0.8 0.5 0.2 3.1 4.2 22.2
0.23–2.75 0.15–1.99 0.04–0.88 0.85–11.02 1.05–16.81 1.91–258.5
0.37 0.92 2.14 1.71 2.02 2.47
0.71 0.36 0.03 0.09 0.04 0.01
OR
95%CI
Z
p
1.3 1.8 3.9 4.0 2.5 1.5
0.35–4.65 0.46–7.08 1.04–14.75 1.06–15.42 0.65–9.55 0.24–9.45
0.37 0.85 2.02 2.04 1.33 0.43
0.72 0.39 0.04 0.04 0.18 0.67
CO2-induced panic
Sex (M) Antidepressant or antianxiety use History of panic attack High maximum anxiety—air High minute ventilation—CO2 Affective disorder diagnosis
Notes: High maximum anxiety—air: maximum anxiety rating on air of 25 or more. High minute ventilation—CO2: mean adjusted minute ventilation (aMV) of more than 1.75 times baseline minute ventilation on air alone (i.e., N60th percentile). Affective disorder diagnosis: bipolar I plus other affective disorders (bipolar II, bipolar NOS, major depression, recurrent).
3.4. Multivariate analyses We performed logistic regression analyses on the relative contributions of demographic, diagnostic, mood state and treatment factors to abnormal respiratory pattern and CO2-induced panic (Table 2). Variables for each of these factors were selected based on their salience in univariate analyses; potential confounding factors sex and antidepressant/antianxiety use were added despite nil associations. The same independent variables and same subjects were used for analyses of both CO2-induced panic and abnormal respiratory response (with exceptions for non-independence). The factors that influenced the outcome of an abnormal respiratory pattern on CO2 included having a high maximum anxiety rating on air (Odds Ratio (OR) = 4.2, p b 0.05) and any affective disorder diagnosis (OR = 22.2, p b 0.05); having a CO2-induced panic response during the experiment decreased the risk of an abnormal pattern (OR = 0.2, p b 0.05). Panic response was associated most strongly with having a history of prior panic attacks (OR = 3.9; p b 0.05), and with having a high anxiety rating on air alone (OR = 4.0; p b 0.05). Other variables of potential influence, including sex, antidepressant use, higher than average respiratory response
to CO2, and any affective disorder diagnosis (BPI + OAD), were not significant contributors to CO2-panic risk. Sex and antidepressant/antianxiety use made no impact on either outcome. 4. Discussion Individuals with a history of manic, hypomanic, and/or major depressive episodes, in contrast to control subjects without a major affective disorder diagnosis, were likely to have an abnormally flat, variable, or escalating respiratory response to CO2. This finding is not readily explained as the result of anxious arousal; although abnormal respiratory patterns were associated with CO2-independent anxiety ratings, a panic response to CO2 appeared to diminish the risk of an abnormal response. Individuals with bipolar disorder type I were more vulnerable to panic symptoms when breathing CO2; however a history of panic attacks and higher anxiety independent of CO2– both common with bipolar disorder–are better predictors of panic response. Surprisingly, the concurrent use of antidepressant or antianxiety medications did not appear to affect the panic or respiratory response, however insufficient information was available to take into account countervailing factors such as illness severity. Shared panic vulnerability to CO2 may result from biological factors common to both affective and anxiety disorders (Boyer, 2000; MacKinnon and Zamoiski, 2006), however an explanation for the abnormal patterns of respiratory response is less obvious. The common thread may be found in the mechanisms of emotion. Aversive and appetitive conditioning respectively make use of closely related structures in the amygdala (Lang and Daviset, 2006). Although these structures participate in distinct circuits and service distinct functions (Gabriel et al., 2003), it is plausible that different forms of emotional conditioning rely on the same cellular mechanisms (Kruzich and See, 2001; LeDoux 2000). Thus panic attacks and aberrant appetitive behavior may be the outcomes of pathology in the centralized facilitation of emotional learning related to fear and reward, respectively. In other words, CO2 may provoke unmanageable fear when there is failure to regard the experimentally induced arousal as benign, and an erratic respiratory pattern when there is failure to learn the optimal solution to the puzzle of insatiable respiratory drive. It is also possible that the vulnerabilities to panic and respiratory irregularity arise from different root causes, but that an effect of one disorder–e.g., stressdriven hypercortisolism in the context of a manicdepressive episode or maladaptive fear conditioned in the context of bipolar hyper- or hypoarousal–is to lower
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
the threshold of expression for the other. Thus historical factors of this kind may explain both the relationship and independence of bipolar and panic disorders and, as this study suggests, the complex relationship of anxiety and respiratory abnormality with hypercapnia. The hypothesis that respiratory perturbations reflect a deficit in behavioral adaptation assumes these phenomena are not better explained by physiologic variables unmeasured in this experiment. But the behavioral hypothesis may be sufficient to explain this phenomenon: there is evidence that training and mindset can override physiologic drive (within limits) and thus significantly influence the respiratory response to CO2 (Ley 1999; Stanescu et al., 1981). Additional evidence for the role of volition and learning comes directly from subjects; several remarked spontaneously that awareness of the experimental conditions mitigated the alarm they might have felt over irremediable breathlessness. To begin to weigh the relative influence of behavioral and physiologic factors one would measure CO2 and O2 levels, enroll a physically homogeneous and unmedicated sample, and gauge the respiratory responses to provocative stimuli that do not elicit a specific appetite, such as lactate infusion or sympathomimetic agents. Systematic assessment of treatment history and response in a medicated sample will potentially reveal an association between successful mood stabilization and normal CO2-response, and would be the essential step towards the use of CO2-response as a clinical tool. In summary, this study has applied a well-established method from respiratory physiology and anxiety research to individuals diagnosed with bipolar and depressive disorders; these individuals tend to be more anxious and to mount erratic and arguably maladaptive respiratory responses to carbon dioxide. This unexpected finding offers preliminary support for a pathophysiologic link between bipolar and panic disorders within the common mechanisms of emotional conditioning of fear and appetite. The dysfunction in bipolar disorder may be in the mechanisms by which behavior is driven by new and evolving information about potential threat and by one's state of satisfaction for any given appetite. Role of funding source This research was supported by NIMH grant K23MH001988 to the principal author, and includes data gathered for the following NIMH bipolar and depression genetics grants: MH059533, MH042243, MH059552, and MH042243. the NIMH had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Conflicts of interest No conflict declared.
199
Acknowledgements The authors wish to thank the research staff for the mood disorders genetics research program at Johns Hopkins, who were instrumental in referring subjects and locating data. References Bailey, J.E., Argyropoulos, S.V., Lightman, S.L., Nutt, D.J., 2003. Does the brain noradrenaline network mediate the effects of the CO2 challenge? J. Psychopharmacol. 17, 252–259. Beck, A.T., 1967. Measurement of depression: the depression inventory. In: Beck, A.T. (Ed.), Depression: Clinical, Experimental, and Theoretical Aspects. Harper & Row, New York, NY, pp. 186–207. Beck, A.T., Epstein, N., Brown, G., Steer, R.A., 1988. An inventory for measuring clinical anxiety: psychometric properties. J.Consult Clin. Psychol. 56, 893–897. Boyer, P., 2000. Do anxiety and depression have a common pathophysiological mechanism? Acta Psychiatr. Scand. Suppl. 102 (Suppl. 406), 24–29. Chen, Y.W., Dilsaver, S.C., 1995. Comorbidity of panic disorder in bipolar illness: evidence from the Epidemiologic Catchment Area Survey. Am. J. Psychiatry 152, 280–282. Cohen, M.E., White, P.D., 1951. Life situations, emotions, and neurocirculatory asthenia (anxiety neurosis, neurasthenia, effort syndrome). Psychosom. Med. 13, 335–357. Damas-Mora, J., Jenner, F.A., Sneddon, J., Addis, W.D., 1978. Ventilatory response to carbon dioxide in depression. Lancet 1 (8064), 616. Depue, R.A., Iacono, W.G., 1989. Neurobehavioral aspects of affective disorders. Annu. Rev. Psychol. 40, 457–492. Everitt, B.J., Cardinal, R.N., Parkinson, J.A., Robbins, T.W., 2003. Appetitive behavior: impact of amygdala-dependent mechanisms of emotional learning. Ann. N.Y. Acad. Sci. 985, 233–250. Gabriel, M., Burhans, L., Kashef, A., 2003. Consideration of a unified model of amygdalar associative functions. Ann. N.Y. Acad. Sci. 985, 206–217. Goodwin, R.D., Hoven, C.W., 2002. Bipolar-panic comorbidity in the general population: prevalence and associated morbidity. J. Affect. Disord. 70, 27–33. Haywood, C., Bloete, M.E., 1969. Respiratory responses of healthy young women to carbon dioxide inhalation. J. Appl. Physiol. 27, 32–35. Johnson, S.L., Sandrow, D., Meyer, B., Winters, R., Miller, I., Solomon, D., Keitner, G., 2000. Increases in manic symptoms after life events involving goal attainment. J. Abnorm. Psychol. 109, 721–727. Kent, J.M., Papp, L.A., Martinez, J.M., Browne, S.T., Coplan, J.D., Klein, D.F., Gorman, J.M., 2001. Specificity of panic response to CO(2) inhalation in panic disorder: a comparison with major depression and premenstrual dysphoric disorder. Am. J. Psychiatry 158, 58–67. Kruzich, P.J., See, R.E., 2001. Differential contributions of the basolateral and central amygdala in the acquisition and expression of conditioned relapse to cocaine-seeking behavior. J. Neurosci. 21, RC155. Lang, P.J., Davis, M., 2006. Emotion, motivation, and the brain: reflex foundations in animal and human research. Prog. Brain Res. 156, 3–29.
200
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
LeDoux, J.E., 2000. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184. Ley, R., 1999. The modification of breathing behavior. Pavlovian and operant control in emotion and cognition. Behav. Modif. 23, 441–479. MacKinnon, D.F., Craighead, B., Hoehn-Saric, R., 2007. Carbon dioxide provocation of anxiety and respiratory response in bipolar disorder. J. Affect Disord. 99, 45–49. MacKinnon, D.F., Zamoiski, R., 2006. Panic comorbidity with bipolar disorder: what is the manic–panic connection? Bipolar. Disord. 8, 648–664. MacKinnon, D.F., Zandi, P.P., Cooper, J., Potash, J.B., Simpson, S.G., Gershon, E., Nurnberger, J., Reich, T., DePaulo, J.R., 2002. Comorbid bipolar disorder and panic disorder in families with a high prevalence of bipolar disorder. Am. J. Psychiatry 159, 30–35. Mitchell, G.S., Johnson, S.M., 2003. Neuroplasticity in respiratory motor control. J. Appl. Physiol. 94, 358–374. Nurnberger, J.I., Blehar, M.C., Kaufmann, C.A., York-Cooler, C., Simpson, S.G., Harkavy-Friedman, J., Severe, J.B., Malaspina, D., Reich, T., 1994. Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Arch. Gen. Psychiatry 51, 849–859. Padget, P., 1928. The respiratory response to carbon dioxide. AJP Legacy 83, 384–394. Papp, L.A., Klein, D.F., Martinez, J., Schneier, F., Cole, R., Liebowitz, M.R., Hollander, E., Fyer, A.J., Jordan, F., Gorman, J.M., 1993. Diagnostic and substance specificity of carbon-dioxide-induced panic. Am. J. Psychiatry 150, 250–257.
Peiffer, C., Costes, N., Herve, P., Garcia-Larrea, L., 2008. Relief of dyspnea involves a characteristic brain activation and a specific quality of sensation. Am. J. Respir. Crit. Care Med. 177, 440–449. Poon, C.S., 1987. Ventilatory control in hypercapnia and exercise: optimization hypothesis. J. Appl. Physiol. 62, 2447–2459. Rafferty, G.F., Gardner, W.N., 1996. Control of the respiratory cycle in conscious humans. J. Appl. Physiol. 81, 1744–1753. Rassovsky, Y., Kushner, M.G., 2003. Carbon dioxide in the study of panic disorder: issues of definition, methodology, and outcome. J. Anxiety Disord. 17, 1–32. Reynolds, W.J., Milhorn Jr., H.T., Holloman Jr., G.H., 1972. Transient ventilatory response to graded hypercapnia in man. J. Appl. Physiol. 33, 47–54. Roy-Byrne, P.P., Stang, P., Wittchen, H.U., Ustun, B., Walters, E.E., Kessler, R.C., 2000. Lifetime panic-depression comorbidity in the National Comorbidity Survey. Association with symptoms, impairment, course and help-seeking. Br. J. Psychiatry 176, 229–235. Sanderson, W.C., Wetzler, S., 1990. Five percent carbon dioxide challenge: valid analogue and marker of panic disorder? Biol. Psychiatry 27, 689–701. Shershow, J.C., Kanarek, D.J., Kazemi, H., 1976. Ventilatory response to carbon dioxide inhalation in depression. Psychosom. Med. 38, 282–287. Stanescu, D.C., Nemery, B., Veriter, C., Marechal, C., 1981. Pattern of breathing and ventilatory response to CO2 in subjects practicing hatha-yoga. J. Appl. Physiol. 51, 1625–1629.
Research report
Carbon dioxide induces erratic respiratory responses in bipolar disorder Dean F. MacKinnon ⁎, Brandie Craighead, Laura Lorenz Department of Psychiatry and Behavioral Sciences, Johns Hopkins University, Baltimore Maryland Received 16 January 2008; received in revised form 17 March 2008; accepted 24 March 2008 Available online 20 May 2008
Abstract Background: CO2 respiration stimulates both anxiety and dyspnea (“air hunger”) and has long been used to study panic vulnerability and respiratory control. High comorbidity with panic attacks suggests individuals with bipolar disorder may also mount a heightened anxiety response to CO2. Moreover, problems in the arousal and modulation of appetites are central to the clinical syndromes of mania and depression; hence CO2 may arouse an abnormal respiratory response to “air hunger”. Methods: 72 individuals (34 bipolar I, 25 depressive and bipolar spectrum, 13 with no major affective diagnosis) breathed air and air with 5% CO2 via facemask for up to 15 min each; subjective and respiratory responses were recorded. Results: Nearly half the subjects diverged from the typical response to a fixed, mildly hypercapneic environment, which is to increase breathing acutely, and then maintain a hyperpneic plateau. The best predictors of an abnormal pattern were bipolar diagnosis and anxiety from air alone. 25 individuals had a panic response; panic responses from CO2 were more likely in subjects with bipolar I compared to other subjects, however the best predictors of a panic response overall were anxiety from air alone and prior history of panic attacks. Limitations: Heterogeneous sample, liberal definition of panic attack. Conclusion: Carbon dioxide produces abnormal respiratory and heightened anxiety responses among individuals with bipolar and depressive disorders. These may be due to deficits in emotional conditioning related to fear and appetite. Although preliminary, this work suggests a potentially useful test of a specific functional deficit in bipolar disorder. © 2008 Elsevier B.V. All rights reserved. Keywords: Bipolar disorder; Panic disorder; Hypercapnia; Fear; Appetite; Respiration; Carbon dioxide
1. Introduction Experimental inhalation of carbon dioxide has illuminated mechanisms of respiratory physiology and anxiety and also may shed theoretical light on affective disorders. At a fixed CO2 concentration of 5%, charted over 10 min or more, the natural respiratory response resembles a ⁎ Corresponding author. Meyer 3-181, 600 N Wolfe St, Baltimore, MD, 21287. Tel.: +1 443 287 6330; fax: +1 410 955 0152. E-mail address: [email protected] (D.F. MacKinnon). 0165-0327/$ - see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jad.2008.03.020
plateau: minute ventilation and end tidal CO2 rise sharply from baseline then settle at a steady, elevated level (Haywood and Bloete, 1969; Padget 1928; Reynolds et al., 1972). Using the same method, panic attacks can be reliably and selectively induced in individuals with prior panic attacks (Cohen and White, 1951; Papp et al., 1993; Rassovsky and Kushner, 2003; Sanderson and Wetzler, 1990). Although major depressive disorder raises the risk for spontaneous panic attacks (Roy-Byrne et al., 2000), depression alone does not confer an increased risk for CO2-induced panic (Kent et al., 2001). Panic attacks are
194
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
also common with bipolar disorder (Chen and Dilsaver, 1995; Goodwin and Hoven, 2002; MacKinnon et al., 2002); it is unknown whether bipolar disorder confers elevated risk for CO2-induced panic. The anxiogenic effect of CO2 is probably due to central nervous system arousal (Bailey et al., 2003), but the response to CO2 breathing may also hinge on its function as a mediator of respiratory appetite. Respiration normally is automatic, thus does not involve an appetitive cycle of drive, satiety, and reward. Dyspnea is an almost irresistible drive (Rafferty and Gardner, 1996); its satiation registers as reward (Peiffer et al., 2008). Yet, in the fixed 5% CO2 protocol, where drive is stimulated but satiety is blocked by the inspired CO2, there is typically an accommodation to a hypercapneic state. Why does one not continue to escalate one's respiratory effort? The plateau phenomenon suggests that the adaptive response to CO2 breathing is to find an optimal compromise of dyspnea and effort (Poon, 1987). Adaptation to a heightened plateau of CO2 and minute ventilation hinges on learning (implicitly or explicitly) that it is futile to continue to escalate respiratory effort; eucapnia can never be achieved. If this adaptive aspect of respiratory behavior is conditioned centrally (Mitchell and Johnson, 2003), as other appetitive behaviors appear to be (Everitt et al., 2003), then some perturbations of respiration may be apparent in a disorder that affects the central regulation of appetitive behavior — such as bipolar disorder (Depue and Iacono, 1989; Johnson et al., 2000). Irregular patterns of respiration were noted in our initial bipolar sample, but not analyzed formally or in detail (MacKinnon et al., 2007). In this extended sample we hypothesize that individuals with bipolar disorder will be more likely than control subjects to mount a respiratory response that deviates from the normal plateau shape. This result would be consistent with a putative functional deficit in appetitive regulation that prevents optimization of the drive to achieve the minimal CO2 possible with sustainable effort. In contrast, the CO2 sensitivity in depressed individuals has been shown to be depressed (DamasMora et al., 1978; Shershow et al., 1976), consistent with a generally inhibited appetitive drive in depression. Based on our prior observations we also expect there to be some CO2-induced panic with bipolar disorder, but less so that in studies of individuals with panic disorder. 2. Methods
excluded if they had an unstable medical condition or psychiatric status at the time of study, or could not abstain from alcohol or illicit drugs for the 2 days prior to the experiment. All were able to provide informed consent. A subset of this group has been described elsewhere, but not compared to a control group (MacKinnon et al., 2007). 2.2. Protocol Subjects initially reviewed the consent form, then met with the PI in his office to discuss the procedure, affirm their informed consent, have a diagnostic interview (if necessary) and provide data on present mental and physical health. They were then escorted to the laboratory, where rating scales were administered. The research assistant was not present during the pre-procedure interview, and the PI was not in the room during the respiratory protocol (the PI observed, without subjects' knowledge, via 2-way mirror). Subjects were randomized to receive either air or 5% CO2 in air as the first arm, and were blinded to the content of the gases. Upon completing the baseline rating scales, subjects were fitted with a face mask linked to a 4-way valve, through which they breathed compressed room air and compressed room air mixed with 5% CO2 for 10–15 min each. Subjects rested in a comfortable reclining chair throughout the procedure, and were allowed to sit back or elevate feet if desired; they were instructed not to speak during the procedure. Between trials the mask was removed and subjects rested for a minimum of 10 min, until anxiety had returned to baseline. Rating scales were completed before, at 5 min intervals during, and immediately after each trial. 2.3. Diagnostic assessment Forty seven of the 72 subjects in this analysis were also participants in psychiatric genetics research at Johns Hopkins. With subjects' consent, diagnoses were extracted from the genetics database, having been obtained via psychiatrist or masters-level research associate assessment using the Diagnostic Interview for Genetic Studies (DIGS) (Nurnberger et al., 1994). For subjects not enrolled in a genetics study, the PI administered an abbreviated version of the DIGS, focusing on depression, mania, and panic disorder items. In ambiguous cases, medical records were requested, with the subject's permission, to clarify a diagnosis.
2.1. Subjects 2.4. Subjective ratings Study subjects were primarily individuals enrolled in bipolar disorder genetics research; additionally, some responded to direct advertisement. Subjects were
Forty two of the 72 subjects completed Beck Depression and Anxiety Inventories (Beck 1967; Beck
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
et al., 1988) prior to the CO2 trial. Before and at intervals during the procedure, subjects completed a rating scale of panic attack symptoms (13 items, rated from not present to severe), and an analog rating (0 to 100 mm) of anxiety and physical discomfort. Immediately post-trial, subjects were asked whether the respiratory procedure caused them to have a panic attack, and whether they had experienced panic attacks in the past. 2.5. Respiratory data Breathing was monitored via Korr RSS 100 spirometer. Respiratory data were recorded on a desktop computer in raw form as the flow (in or out) each 0.02 s; from this the RSS software generated a “parameter file” which established breath size and time interval for each breath, based on changes in the flow pattern. All raw data were rendered graphically and inspected by the PI (who was blinded at this phase to the diagnosis and panic outcome), to detect and resolve problems with breaths missed or miscalled by the breath parameter algorithm. All 72 subjects were included in the primary analysis of panic response; for the analysis of individual response patterns and logistic regression analyses (see below) we required a minimum of 10 min of useful CO2 data. In 4 cases subjects interrupted the CO2 arm before the 10 min mark, and in 6 cases the trial was limited by technical issues (gas supply or computer problems), thus n = 62 for these analyses. To facilitate comparison of respiratory data across individuals, the primary variables tidal volume, breath frequency (time from peak inspiration to peak inspiration), and minute ventilation (sum of tidal volumes for each 60 s epoch) on CO2 were divided by the means of these measures on air alone. Thus, an adjusted minute ventilation (aMV) of 2 means the subject consumed twice as much air + CO2 per minute compared to air alone. 2.6. Data analysis A panic response was determined either by self report or by ratings that indicated moderate anxiety (N50 on the analog rating scale) plus four or more typical panic symptoms. We use the term “panic response” rather than “panic attack” to differentiate this outcome from the outcome in other studies that use a more stringent definition of “panic attack”. Statistical significance was calculated using Student's t-test (2-tailed) for continuous variables, and chi-square for dichotomous variables. These, as well as regression slope calculations, were performed using the
195
Microsoft Excel data analysis package. Logistic regression analyses were performed using Stata software. In comparisons based on affective disorder diagnosis, we used three diagnostic categories: BiPolar I (BPI: subjects were given a confident diagnosis of bipolar disorder type I by an interviewer), Other Affective Disorder (OAD: subjects were diagnosed with either bipolar II with recurrent depression, major depressive disorder, recurrent, or “bipolar NOS”, denoting subthreshold or questionable manic/hypomanic syndromes), and No Affective Disorder (NAD: subjects with no major mental illness, as well as subjects with non-affective or questionable affective diagnoses). Analysis of individual response patterns applied formal criteria to determine whether a subject's respiratory response was consistent or inconsistent with the normal plateau response described in the physiology literature. Specifically, a “normal” response had: 1) an early spike (the slope of aMV vs time for minutes 0–3 was greater than the slope during minutes 4–9 and 10–15); 2) persistent elevation (mean aMV in minutes 6–15 was greater than mean aMV for minutes 1–5); 3) net increase on CO2 vs air (mean aMV for minutes 1–15 N 1); and 4) minimal plateau variance (variance of aMV after minute 5 was less than the mean aMV variance for all subjects on air, which was 0.06). 3. Results Respiratory and subjective responses to air and 5% CO2 were obtained for 72 subjects. Thirty-four had bipolar disorder type 1. Of the remainder 25 had a major affective disorder (bipolar II (n = 7), bipolar NOS (n = 8), or recurrent major depressive disorder (n = 10)), while 8 had no major psychiatric disorder diagnosis and 5 had some other psychiatric diagnosis (schizophrenia (n = 2), antisocial personality disorder (n = 2), depression NOS (n = 1)). Over half (n = 37) had experienced at least one panic attack prior to the study, and 20 were diagnosed on interview with panic disorder. Twenty-five (35%) of the subjects experienced a panic response during the CO2 trial, by either self-report or by panic attack criteria. 3.1. Diagnostic group factors Groups defined by diagnosis–bipolar I (BPI), other affective disorder (OAD), and no affective disorder (NAD)–differed in no significant way in terms of sex, age, body size, or smoking habit (Table 1). Although individuals with bipolar I had histories of panic attacks and panic disorder more often than others, these
196
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
Table 1 Subject characteristics and response By affective diagnostic group Bipolar I
n
(BPI)
Other Affective Disorder (OAD)
34
25
By panic response to CO2
No Affective Disorder
Non-panicking
Panicking
(NAD) 13
47
25
Demographic 43 34.0 25 10
±12 ±8.7 74% 29%
47 32.1 13 7
±10 ±8.5 52% 28%
41 30.5 7 4
±12 ±10.4 54% 31%
– – – –
43 31.8 28 17
±11 ±8.3 60% 36%
47 32.8 17 7
±12 ±7.3 68% 28%
– – – –
Bipolar I Bipolar spectrum Prior panic attacks Panic disorder Panic from CO2 Antidepressant Benzodiazepine
– – 20 12 16 24 4
– – 59% 35% 47% 71% 12%
– – 13 5 6 14 1
– – 52% 20% 24% 56% 4%
– – 4 3 3 2 0
– – 31% 23% 23% 15% 0%
– – – – d a,b –
18 19 21 11 – 23 3
38% 40% 45% 23% – 49% 6%
16 6 16 9 – 17 2
64% 24% 64% 36% – 68% 8%
– – – – –
Self-rating Beck Depression⁎ Beck Anxiety⁎ Baseline anxiety
20 18 27
±6 ±13 ±30
20 21 28
±10 ±17 ±28
15 16 20
±10 ±14 ±24
– – –
18 14 20
±9 +12 ±24
22 26 38
±8.0 ±16 ±32
– e e
Air Minute ventilation (l/min) Maximum anxiety
8.5 34
±2.0 ±30
8.2 33
±2.4 ±31
9.8 24
±1.9 ±24
– –
8.6 22
±2.1 +20
8.7 48
±2.3 ±26
– e
13.3 14.9 53
±2.8 ±4.9 ±30
13.2 11.9 44
±3.0 ±5.5 ±33
14.7 17.5 38
+2.6 ±5.0 ±33
– b,c –
11.7 13.7 33
±3.5 ±5.1 ±21
14.0 15.6 69
±2.5 ±6.0 ±15
– X
Age Basal metabolic index Female Current smoker Clinical
CO2 Time (min) Minute ventilation (l/min) Maximum anxiety
Statistically significant comparisons ( p b 0.05): a) BPI vs NAD; b) BPS vs NAD; c) BPI vs BPS; d) BPI vs (BPS+ NAD); e) non-panicking vs panicking. X = biased comparison. For the comparison of antidepressant and benzodiazepine use n = 67, excluding subjects with no psychiatric diagnosis for which medications are indicated; ⁎n = 42; Other Affective Disorder = BPII, recurrent MDD, BP NOS; no affective disroder inlcudes schizophrenia (2), ASPD (2), and no mental disorder.
differences did not meet statistical significance. Anxiety and mood rating scales and anxiety/discomfort ratings on air alone were not significantly different across diagnostic categories. 3.2. Patterns of individual respiratory response Inspection of minute-by-minute data revealed widespread individual variation (Fig. 1). The modal response across all subjects was a marked elevation in minute ventilation (driven mainly by increased tidal volume) during the first 3 to 5 min, followed by a persistently flat or less acute elevation for the remainder of the CO2 exposure. But many individuals deviated markedly from this pattern. Several had what appeared to be a delayed,
muted or absent response. Many individuals, having peaked early in the course of CO2 exposure, subsequently dropped towards baseline respiration; several of these went on to peak again late in the trial. And a few individuals showed a continuous steep rise throughout the trial, with no evident plateau. Inspection of these patterns, grouped by diagnosis, suggests that subjects without major affective disorder tend to follow the normal plateau pattern, while the subjects with major affective disorders often had aberrant patterns. As described in the Methods section, we defined parameters for a normal time course of response as follows. A normal pattern was postulated to begin with a rapid increase, followed by a persistent but less acute net increase in minute ventilation (relative to air) and minimal
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
197
Fig. 1. Individual respiratory responses to carbon dioxide inhalation. Tracings for 62 individuals with CO2 trials lasting 10 min or more are grouped by affective disorder diagnosis. Each tracing represents the total volume per minute for 10–15 min for one individual breathing 5% carbon dioxide mixed with room air. All values are adjusted for the subject's respiration on air alone, where 1 is no change, N1 is an increase and b1 is a decrease from air. The upper row includes tracings deemed “normal" by algorithm defining the expected steady and sustained respiratory elevation. The lower row includes tracings that failed on one or more criteria.
variability throughout the remainder of the trial. An abnormal pattern failed at least one of these criteria. Of the 62 individuals who remained on CO2 for 10 min or longer, most (55%) of had normal responses by these criteria, and they were distributed unevenly across diagnostic categories. Normal respiratory responses were observed in 92% of the NAD subjects, versus 53% of BPI, versus 35% of the OAD (NAD vs BPI, χ2 = 5.5, p b 0.02; NAD vs OAD χ2 = 9.7, p b 0.005; BPI vs OAD χ2 = 1.6, p = 0.2). On average, NAD individuals met 3.8 of the possible 4 criteria for a normal response, BPI individuals met 3.3 of the 4, and OAD met 2.9 (NAD vs BPI, t40 = 2.3, p = 0.02; NAD vs OAD, t30 = 3.2, p b 0.01; BPI vs OAD t48 = 1.3, p = 0.2). Thus, subjects with a major affective disorder, particularly subjects with bipolar spectrum and major depressive disorders, were significantly more likely to have an aberrant response to CO2. 3.3. Predictors of panic response There was an increased risk of a panic response among those diagnosed with bipolar I compared to all other diagnoses (47% of BPI vs 24% of others; χ2 = 4.3, df = 1, p b 0.04) (Table 1). This was evidently independent of anti-panic agents: 69% of subjects with bipolar I who panicked were on antidepressant/antianxiety medications, versus 72% who did not panic (χ2 = 0.05, df = 1, p = 0.8). We did not find that panic responses were predicted directly by age, sex, body size, current smoking, prior panic attacks, or even prior diagnosis
of panic disorder. The use of antidepressants and antianxiety agents was similarly common in psychiatrically diagnosed subjects who panicked and did not panic. The most robust predictor of a panic response was the anxiety and physical discomfort experienced at baseline and while breathing air via facemask. Those who panicked scored higher on the Beck Anxiety Inventory prior to starting the experiment (26.1 vs 14.5; t40 = 2.7, p b 0.01), and rated their baseline levels of anxiety (pre-facemask), on a scale of 0–100, significantly higher than those who did not panic with CO2 (37.6 vs 19.8, t70 = 2.6, p b 0.01). Their maximum anxiety rating during the air-only part of the trial was over twice as great as that experienced by subjects who did not panic (50.4 vs 21.8, t70 = 4.4, p b 0.00005). The maximum anxiety rating over the CO2 trial, for panicking individuals, was 76.5/100, compared to 31.7/100 for non-panicking individuals. When the analysis is limited to the strictest definition of panic attack–self-report plus panic criteria–and the 9 individuals thus defined are compared to the 47 who neither reported panic nor met criteria for panic, a few new findings emerge. The group that panicked under the strict definition was significantly older (51 vs 43, t54 = 2.0, p b 0.05) and somewhat though not significantly more obese (BMI = 37.3 vs 31.9, t54 = 1.4, p = 0.17). These individuals had two minutes less time on CO2 (11.7 vs 14.0 min, t54 = 2.3, p b 0.05), but otherwise there were no notable differences in the results.
198
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
Table 2 Logistic regression analyses of selected variables to explain panic response and abnormal respiratory response (n = 62) Abnormal respiratory response to CO2
Sex (M) Antidepressant or antianxiety use CO2 induced panic History of panic attack High maximum anxiety—air Affective disorder diagnosis
OR
95%CI
Z
p
0.8 0.5 0.2 3.1 4.2 22.2
0.23–2.75 0.15–1.99 0.04–0.88 0.85–11.02 1.05–16.81 1.91–258.5
0.37 0.92 2.14 1.71 2.02 2.47
0.71 0.36 0.03 0.09 0.04 0.01
OR
95%CI
Z
p
1.3 1.8 3.9 4.0 2.5 1.5
0.35–4.65 0.46–7.08 1.04–14.75 1.06–15.42 0.65–9.55 0.24–9.45
0.37 0.85 2.02 2.04 1.33 0.43
0.72 0.39 0.04 0.04 0.18 0.67
CO2-induced panic
Sex (M) Antidepressant or antianxiety use History of panic attack High maximum anxiety—air High minute ventilation—CO2 Affective disorder diagnosis
Notes: High maximum anxiety—air: maximum anxiety rating on air of 25 or more. High minute ventilation—CO2: mean adjusted minute ventilation (aMV) of more than 1.75 times baseline minute ventilation on air alone (i.e., N60th percentile). Affective disorder diagnosis: bipolar I plus other affective disorders (bipolar II, bipolar NOS, major depression, recurrent).
3.4. Multivariate analyses We performed logistic regression analyses on the relative contributions of demographic, diagnostic, mood state and treatment factors to abnormal respiratory pattern and CO2-induced panic (Table 2). Variables for each of these factors were selected based on their salience in univariate analyses; potential confounding factors sex and antidepressant/antianxiety use were added despite nil associations. The same independent variables and same subjects were used for analyses of both CO2-induced panic and abnormal respiratory response (with exceptions for non-independence). The factors that influenced the outcome of an abnormal respiratory pattern on CO2 included having a high maximum anxiety rating on air (Odds Ratio (OR) = 4.2, p b 0.05) and any affective disorder diagnosis (OR = 22.2, p b 0.05); having a CO2-induced panic response during the experiment decreased the risk of an abnormal pattern (OR = 0.2, p b 0.05). Panic response was associated most strongly with having a history of prior panic attacks (OR = 3.9; p b 0.05), and with having a high anxiety rating on air alone (OR = 4.0; p b 0.05). Other variables of potential influence, including sex, antidepressant use, higher than average respiratory response
to CO2, and any affective disorder diagnosis (BPI + OAD), were not significant contributors to CO2-panic risk. Sex and antidepressant/antianxiety use made no impact on either outcome. 4. Discussion Individuals with a history of manic, hypomanic, and/or major depressive episodes, in contrast to control subjects without a major affective disorder diagnosis, were likely to have an abnormally flat, variable, or escalating respiratory response to CO2. This finding is not readily explained as the result of anxious arousal; although abnormal respiratory patterns were associated with CO2-independent anxiety ratings, a panic response to CO2 appeared to diminish the risk of an abnormal response. Individuals with bipolar disorder type I were more vulnerable to panic symptoms when breathing CO2; however a history of panic attacks and higher anxiety independent of CO2– both common with bipolar disorder–are better predictors of panic response. Surprisingly, the concurrent use of antidepressant or antianxiety medications did not appear to affect the panic or respiratory response, however insufficient information was available to take into account countervailing factors such as illness severity. Shared panic vulnerability to CO2 may result from biological factors common to both affective and anxiety disorders (Boyer, 2000; MacKinnon and Zamoiski, 2006), however an explanation for the abnormal patterns of respiratory response is less obvious. The common thread may be found in the mechanisms of emotion. Aversive and appetitive conditioning respectively make use of closely related structures in the amygdala (Lang and Daviset, 2006). Although these structures participate in distinct circuits and service distinct functions (Gabriel et al., 2003), it is plausible that different forms of emotional conditioning rely on the same cellular mechanisms (Kruzich and See, 2001; LeDoux 2000). Thus panic attacks and aberrant appetitive behavior may be the outcomes of pathology in the centralized facilitation of emotional learning related to fear and reward, respectively. In other words, CO2 may provoke unmanageable fear when there is failure to regard the experimentally induced arousal as benign, and an erratic respiratory pattern when there is failure to learn the optimal solution to the puzzle of insatiable respiratory drive. It is also possible that the vulnerabilities to panic and respiratory irregularity arise from different root causes, but that an effect of one disorder–e.g., stressdriven hypercortisolism in the context of a manicdepressive episode or maladaptive fear conditioned in the context of bipolar hyper- or hypoarousal–is to lower
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
the threshold of expression for the other. Thus historical factors of this kind may explain both the relationship and independence of bipolar and panic disorders and, as this study suggests, the complex relationship of anxiety and respiratory abnormality with hypercapnia. The hypothesis that respiratory perturbations reflect a deficit in behavioral adaptation assumes these phenomena are not better explained by physiologic variables unmeasured in this experiment. But the behavioral hypothesis may be sufficient to explain this phenomenon: there is evidence that training and mindset can override physiologic drive (within limits) and thus significantly influence the respiratory response to CO2 (Ley 1999; Stanescu et al., 1981). Additional evidence for the role of volition and learning comes directly from subjects; several remarked spontaneously that awareness of the experimental conditions mitigated the alarm they might have felt over irremediable breathlessness. To begin to weigh the relative influence of behavioral and physiologic factors one would measure CO2 and O2 levels, enroll a physically homogeneous and unmedicated sample, and gauge the respiratory responses to provocative stimuli that do not elicit a specific appetite, such as lactate infusion or sympathomimetic agents. Systematic assessment of treatment history and response in a medicated sample will potentially reveal an association between successful mood stabilization and normal CO2-response, and would be the essential step towards the use of CO2-response as a clinical tool. In summary, this study has applied a well-established method from respiratory physiology and anxiety research to individuals diagnosed with bipolar and depressive disorders; these individuals tend to be more anxious and to mount erratic and arguably maladaptive respiratory responses to carbon dioxide. This unexpected finding offers preliminary support for a pathophysiologic link between bipolar and panic disorders within the common mechanisms of emotional conditioning of fear and appetite. The dysfunction in bipolar disorder may be in the mechanisms by which behavior is driven by new and evolving information about potential threat and by one's state of satisfaction for any given appetite. Role of funding source This research was supported by NIMH grant K23MH001988 to the principal author, and includes data gathered for the following NIMH bipolar and depression genetics grants: MH059533, MH042243, MH059552, and MH042243. the NIMH had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Conflicts of interest No conflict declared.
199
Acknowledgements The authors wish to thank the research staff for the mood disorders genetics research program at Johns Hopkins, who were instrumental in referring subjects and locating data. References Bailey, J.E., Argyropoulos, S.V., Lightman, S.L., Nutt, D.J., 2003. Does the brain noradrenaline network mediate the effects of the CO2 challenge? J. Psychopharmacol. 17, 252–259. Beck, A.T., 1967. Measurement of depression: the depression inventory. In: Beck, A.T. (Ed.), Depression: Clinical, Experimental, and Theoretical Aspects. Harper & Row, New York, NY, pp. 186–207. Beck, A.T., Epstein, N., Brown, G., Steer, R.A., 1988. An inventory for measuring clinical anxiety: psychometric properties. J.Consult Clin. Psychol. 56, 893–897. Boyer, P., 2000. Do anxiety and depression have a common pathophysiological mechanism? Acta Psychiatr. Scand. Suppl. 102 (Suppl. 406), 24–29. Chen, Y.W., Dilsaver, S.C., 1995. Comorbidity of panic disorder in bipolar illness: evidence from the Epidemiologic Catchment Area Survey. Am. J. Psychiatry 152, 280–282. Cohen, M.E., White, P.D., 1951. Life situations, emotions, and neurocirculatory asthenia (anxiety neurosis, neurasthenia, effort syndrome). Psychosom. Med. 13, 335–357. Damas-Mora, J., Jenner, F.A., Sneddon, J., Addis, W.D., 1978. Ventilatory response to carbon dioxide in depression. Lancet 1 (8064), 616. Depue, R.A., Iacono, W.G., 1989. Neurobehavioral aspects of affective disorders. Annu. Rev. Psychol. 40, 457–492. Everitt, B.J., Cardinal, R.N., Parkinson, J.A., Robbins, T.W., 2003. Appetitive behavior: impact of amygdala-dependent mechanisms of emotional learning. Ann. N.Y. Acad. Sci. 985, 233–250. Gabriel, M., Burhans, L., Kashef, A., 2003. Consideration of a unified model of amygdalar associative functions. Ann. N.Y. Acad. Sci. 985, 206–217. Goodwin, R.D., Hoven, C.W., 2002. Bipolar-panic comorbidity in the general population: prevalence and associated morbidity. J. Affect. Disord. 70, 27–33. Haywood, C., Bloete, M.E., 1969. Respiratory responses of healthy young women to carbon dioxide inhalation. J. Appl. Physiol. 27, 32–35. Johnson, S.L., Sandrow, D., Meyer, B., Winters, R., Miller, I., Solomon, D., Keitner, G., 2000. Increases in manic symptoms after life events involving goal attainment. J. Abnorm. Psychol. 109, 721–727. Kent, J.M., Papp, L.A., Martinez, J.M., Browne, S.T., Coplan, J.D., Klein, D.F., Gorman, J.M., 2001. Specificity of panic response to CO(2) inhalation in panic disorder: a comparison with major depression and premenstrual dysphoric disorder. Am. J. Psychiatry 158, 58–67. Kruzich, P.J., See, R.E., 2001. Differential contributions of the basolateral and central amygdala in the acquisition and expression of conditioned relapse to cocaine-seeking behavior. J. Neurosci. 21, RC155. Lang, P.J., Davis, M., 2006. Emotion, motivation, and the brain: reflex foundations in animal and human research. Prog. Brain Res. 156, 3–29.
200
D.F. MacKinnon et al. / Journal of Affective Disorders 112 (2009) 193–200
LeDoux, J.E., 2000. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184. Ley, R., 1999. The modification of breathing behavior. Pavlovian and operant control in emotion and cognition. Behav. Modif. 23, 441–479. MacKinnon, D.F., Craighead, B., Hoehn-Saric, R., 2007. Carbon dioxide provocation of anxiety and respiratory response in bipolar disorder. J. Affect Disord. 99, 45–49. MacKinnon, D.F., Zamoiski, R., 2006. Panic comorbidity with bipolar disorder: what is the manic–panic connection? Bipolar. Disord. 8, 648–664. MacKinnon, D.F., Zandi, P.P., Cooper, J., Potash, J.B., Simpson, S.G., Gershon, E., Nurnberger, J., Reich, T., DePaulo, J.R., 2002. Comorbid bipolar disorder and panic disorder in families with a high prevalence of bipolar disorder. Am. J. Psychiatry 159, 30–35. Mitchell, G.S., Johnson, S.M., 2003. Neuroplasticity in respiratory motor control. J. Appl. Physiol. 94, 358–374. Nurnberger, J.I., Blehar, M.C., Kaufmann, C.A., York-Cooler, C., Simpson, S.G., Harkavy-Friedman, J., Severe, J.B., Malaspina, D., Reich, T., 1994. Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Arch. Gen. Psychiatry 51, 849–859. Padget, P., 1928. The respiratory response to carbon dioxide. AJP Legacy 83, 384–394. Papp, L.A., Klein, D.F., Martinez, J., Schneier, F., Cole, R., Liebowitz, M.R., Hollander, E., Fyer, A.J., Jordan, F., Gorman, J.M., 1993. Diagnostic and substance specificity of carbon-dioxide-induced panic. Am. J. Psychiatry 150, 250–257.
Peiffer, C., Costes, N., Herve, P., Garcia-Larrea, L., 2008. Relief of dyspnea involves a characteristic brain activation and a specific quality of sensation. Am. J. Respir. Crit. Care Med. 177, 440–449. Poon, C.S., 1987. Ventilatory control in hypercapnia and exercise: optimization hypothesis. J. Appl. Physiol. 62, 2447–2459. Rafferty, G.F., Gardner, W.N., 1996. Control of the respiratory cycle in conscious humans. J. Appl. Physiol. 81, 1744–1753. Rassovsky, Y., Kushner, M.G., 2003. Carbon dioxide in the study of panic disorder: issues of definition, methodology, and outcome. J. Anxiety Disord. 17, 1–32. Reynolds, W.J., Milhorn Jr., H.T., Holloman Jr., G.H., 1972. Transient ventilatory response to graded hypercapnia in man. J. Appl. Physiol. 33, 47–54. Roy-Byrne, P.P., Stang, P., Wittchen, H.U., Ustun, B., Walters, E.E., Kessler, R.C., 2000. Lifetime panic-depression comorbidity in the National Comorbidity Survey. Association with symptoms, impairment, course and help-seeking. Br. J. Psychiatry 176, 229–235. Sanderson, W.C., Wetzler, S., 1990. Five percent carbon dioxide challenge: valid analogue and marker of panic disorder? Biol. Psychiatry 27, 689–701. Shershow, J.C., Kanarek, D.J., Kazemi, H., 1976. Ventilatory response to carbon dioxide inhalation in depression. Psychosom. Med. 38, 282–287. Stanescu, D.C., Nemery, B., Veriter, C., Marechal, C., 1981. Pattern of breathing and ventilatory response to CO2 in subjects practicing hatha-yoga. J. Appl. Physiol. 51, 1625–1629.