Research Article

LANGUAGE OF DYSPNEA IN PANIC DISORDER

Giampaolo Perna, M.D., Ph.D.,n Daniela Caldirola, M.D., Caterina Namia, Psy.D., Michele Cucchi, M.D.,Giovanna Vanni, M.D., and Laura Bellodi, M.D.

Dyspnea is a key symptom in panic attacks. This study investigated differenttypes of dyspnea induced by the 35% CO2 challenge test given to patients withpanic disorder (PD). The types of dyspnea provide room for possible conjectureson neurophysiological pathways involved in the experience of breathingdiscomfort in PD and in the panic-respiration connection. Factor analysisapplied to the Dyspnea Questionnaire identified three main factors: breathingeffort, sense of suffocation, and rapid breath. Factor scores for breathing effortand sense of suffocation significantly discriminated between patients who didand those who did not report CO2-induced panic attacks. Factor scores forbreathing effort significantly discriminated between patients whose reactionresembled their unexpected panic attacks and those whose reaction did not. Adissociation between an increased central respiratory command and a decreasedmechanical efficiency of the respiratory response in patients with PD mayunderlie the breathing effort factor during the CO2 challenge. The sense ofsuffocation factor was found to be linked to chemosensitivity. Although involvedin CO2 reactivity, it may not be a central factor in unexpected panic attacks.Depression and Anxiety 20:32–38, 2004. & 2004 Wiley-Liss, Inc.

Key words: CO2; chemosensitivity; breathing; ef fort; suf focation

INTRODUCTIONDyspnea is a common sensation. It can present as akey symptom in certain medical conditions [Meek andSchwartzstein, 1999] and can also be experienced byhealthy subjects [Manning and Mahler, 2001; Simon etal., 1989]. Dyspnea is the subjective experience ofbreathing discomfort that subsumes qualitatively dis-tinct sensations with variable intensity, probably arisingfrom different pathophysiologic mechanisms [Meekand Schwartzstein, 1999]. Because dyspnea and painshare several characteristics, experimental approachesdeveloped in the study of pain have subsequently beenapplied to investigating dyspnea [Banzett and Moosavi,2001]. Advances in the use of standardized measure-ment instruments (e.g., pain questionnaires and verbaldescriptor scales) have led to the recognition that painis multidimensional whereas physiology and brainimaging studies have widely investigated the under-lying neural processing of pain sensations [Banzett andMoosavi, 2001]. Similarly, dyspnea has been studied inboth medical conditions and experimental settingswhere dyspnea is evoked in healthy subjects to focus

on the verbal descriptors of dyspnea sensations and thepossible mechanisms underlying them. Findings sug-gest that dyspnea may encompass multiple sensationsill-described by a single term and not explained bya single neurobiological mechanism [Tobin, 1990].Specific groups of breathing discomfort descriptorshave been identified and linked to possible differentunderlying neurobiological pathways such as theactivation of chemoreceptors, pulmonary or respiratorymuscle receptors and the outgoing respiratory motorcommand. Although numerous factors may influence

DEPRESSION AND ANXIETY 20:32–38 (2004)

Anxiety Disorders Clinical and Research Unit, Vita-Salute

University, Istituto Scientifico H. San Raffaele, Milan, Italy

nCorrespondence to: Dr. Giampaolo Perna, Istituto Scientifico

H. San Raffaele, Vita-Salute University, 20 via Stamira d’Ancona,

20127, Milan, Italy. E-mail: Perna.giampaolo@hsr.it

Received for publication 23 July 2003; Revised 1 March 2004;

Accepted 16 March 2004

DOI: 10.1002/da.20016

Published online 18 August 2004 in Wiley InterScience (www.

interscience.wiley.com).

&& 2004 WILEY-LISS, INC.

the subjective perception of dyspnea, verbal descriptorsand associated neural pathways are thought to beshared in dif ferent conditions [Mahler et al., 1996;Moy et al., 2000; O’Donnell et al., 1997]. Dyspnea is akey symptom in panic attacks. Experimental evidencesuggests a role for the respiratory system in thepathophysiology of panic disorder (PD) [Bellodi andPerna, 1998; Klein, 1993, 2002]. Klein [2002] empha-sized that dyspnea is a salient characteristic ofspontaneous panic but not of fear reactions. Panico-genic agents, such as sodium lactate and carbondioxide, provoke panic attacks with prominent dyspneabut without activation of the hypothalamic-pituitary-adrenal axis (HPA), whereas other agents, such asfenfluramine and yohimbine, provoke anxious reactionswithout marked dyspnea but with activation of theHPA [Hollander et al., 1989; Liebowitz et al., 1985;Sinha et al., 1999]. These findings support the idea thatpanic attacks, particularly those characterized bymarked dyspnea, are not simply equivalent to fear/stress reactions; they also provide evidence for the‘‘respiratory theory’’ of the ‘‘false suffocation alarm’’[Klein, 1993]. Despite the existence of a panic-respiration connection [Bellodi and Perna, 1998], itsunderlying neurophysiological pathways remain un-clear. This study investigated types of dyspnea inducedby the 35% CO2 challenge test in patients with PD.The types of dyspnea provides room for possibleconjectures on the neurophysiological pathwaysinvolved in the experience of breathing discomfort inPD and in the panic-respiration connection.

METHODS

PATIENTS

Over a period of 9 months, 78 outpatients with PDwith or without agoraphobia (46 women, 32 men; meanage 32.478.8 years; mean illness duration 5.676.8years) were recruited among those consecutivelyreferred to the Anxiety Disorders Clinical and ResearchUnit of San Raffaele Hospital, Milan. Psychiatricdiagnosis was obtained by senior psychiatrists usingthe MINI International Neuropsychiatric Interview forDSM IV-Plus [Sheehan et al., 1994].

Exclusion criteria comprised concurrent psychiatricdisorders, except specific phobias, significant concur-rent cardiocirculatory and respiratory diseases, perso-nal or family history of cerebral aneurysm, significanthypertension (systolic blood pressure 4180 mm Hg,diastolic blood pressure 4100 mm Hg), pregnancyor epilepsy, according to physical examination andmedical history.

The severity of clinical symptomatology in patientswith PD was measured using the Panic AssociatedSymptoms Scale (PASS), which assesses panic attacks(PASS-PA subscale), anticipatory anxiety (PASS-AAsubscale) and agoraphobia (PASS-AGO subscale)[Argyle et al., 1991] and the Fear Questionnaire (FQ)

that assesses agoraphobia, blood-injury phobia andsocial phobia [Marks and Mathews, 1979].

PROCEDURE

Patients were administered the 35% CO2 inhalationchallenge test and then asked to fill in the DyspneaQuestionnaire.

Pre-test conditions required that the subjects werenot to have taken any medication for at least 2 weeks,and had refrained from alcohol for at least 36 hr,xanthines for at least 8 hr and food or smoking for atleast 2 hr before the test. Participants gave their writteninformed consent after having received a detailedexplanation of the test procedure.

35% CO2 INHALATION CHALLENGE TEST

Subjects were tested in double-blind random fashion,described in Perna et al. [1994]. Two different gasmixtures were employed: compressed air (placebo) anda mixture of 35% CO2 and 65% O2. Both gases wereinhaled through the same self-administration mask.Vital capacity was measured by a respirometer (WrightRespirometer Mark 20; Ferraris Medical, Louisville,CO) connected to the mask. The respirometer mea-sured the gas volume delivered in each inhalation.Subjects were informed that they would be inhaling twoharmless gas mixtures containing different percentagesof CO2 and O2, and that they might experience somediscomfort, ranging from a few neurovegetative symp-toms to a definite sensation of anxiety/discomfort withseveral somatic or cognitive sensations. The words‘‘panic attack’’ were never mentioned. Each subjectinhaled one-vital capacity of 35% CO2 and 65% O2 orcompressed air in a randomized order. After eachinhalation, subjects were asked to hold their breath for4 s. An interval of 25–30 min elapsed between the twoinhalations. The test was considered valid only if thesubject had inhaled at least 80% of the previouslymeasured vital capacity.

The responses to the challenges were evaluated usinga Visual Analogue Scale for Anxiety (VAS-A) and thePanic Symptom List (PSL) [Pols et al., 1991]. TheVAS-A describes the degree of global subjective anxietyon a continuum from 0 (no anxiety present) to 100(worst anxiety imaginable), and the PSL is a self-administered questionnaire assessing each of the 13symptoms of PD (DSM III-R/IV) on a 5-point scaleranging from 0 (absent) to 4 (very intense) (range¼0–52).

Global anxiety reactivity was evaluated as D%VAS-A(percentage of maximum increment or decrementpossible on the VAS-A scale) and calculated asfollows:

a) if VAS-A (post-CO2 VAS-A values�pre-CO2 VAS-Avalues) was positive, then D%VAS-A¼VAS-A� 100/(100� VAS-A before CO2); and

Research Article: Dyspnea in Panic Disorder 33

b) if VAS-A was negative, then D%VAS-A¼VAS-A� 100/VAS-A before CO2.

Before the 35% CO2 challenge, each patientcompleted the State Trait Anxiety Inventory, stateversion (STAI) [Spielberger et al., 1970].

ASSESSMENT OF CO2 REACTIVITY

CO2-induced panic attack. According to the scalesdescribed above and the DSM IV criteria for panicattack, the reaction to the 35% CO2 challenge wasconsidered to be an induced-panic attack when asensation of fear or panic with at least D%VAS-A Z26was evoked and there was an increase of the scores of atleast four symptoms on the PSL. The VAS-A value isan ideal threshold that a receiver operating character-istic (ROC) analysis of the 35% CO2 challenge haspreviously been shown able to separate panic patientsand healthy controls with a positive predictive power of91% and a negative predictive power of 75% [Battagliaand Perna, 1995].

Similarity to unexpected panic attack. To identifythose patients whose reaction to CO2 resembled theirunexpected panic attacks, all subjects were asked tocompare their CO2-induced subjective experience withtheir typical unexpected panic attacks and to rate it as‘‘dif ferent from’’ or ‘‘similar to’’ an unexpected panicattack.

DYSPNEA QUESTIONNAIRE

We administered our Italian version of the 19-itemDyspnea Questionnaire developed by Simon et al.[1989] (Table 1). Each item (descriptor) consists of asentence describing a breathing discomfort sensation.To ensure that the original English version items wereaccurately translated in our Italian version, fourspecialists (an Italian–English native language psychia-trist, a psychologist, a pneumologist, and a specialist insport medicine) worked independently on translatingthe texts. The Italian text was accepted when three offour translators gave identical versions; discrepancies intranslation were worked out by consensus among thefour translators. Patients were instructed to choose theitems (one or more) they thought most closelydescribed the breathing sensations they experiencedduring the CO2 challenge test.

ASSESSMENT OF DYSPNEA

Because specific groups of breathing discomfortdescriptors are thought to define dyspnea better thana single descriptor [Mahler et al., 1996; Moy et al.,2000; O’Donnell et al., 1997], factor analysis (seebelow) was applied to the Dyspnea Questionnaire toidentify the groups of descriptors (factors) that bestdefined the type of dyspnea the patients experiencedduring the CO2 challenge. Dyspnea experiences were

evaluated by sub-grouping patients according to theassessment of CO2 reactivity.

STATISTICAL ANALYSIS

Factor analysis: application in our study. Weapplied factor analysis to the 19 descriptors (variables)of the Dyspnea Questionnaire. Our sample had morethan twice the number of variables subjected to factoranalysis, which is the proportion recommended toobtain a good level of reliability. The factor analysiswas carried out using principal component extractionin which the axes are not rotated. This method allowedus to extract a few hierarchically ordered factors,wherein the first factor extracted is the most importantbecause it has the highest factor weight. Each factorconsists of a group of variables with different factorweights. The higher the factor weight of each variable,the more that variable contributes to the factor. Factorrotation was carried out on the data obtained from theprevious not rotated factor analysis to obtain variableswith very high factor weights in some factors and verylow factor weights in others. We chose a varimaxrotation that implies orthogonal factors. The maximumnumber of factors extracted was selected by the CattelScree Test. To insert a variable into a factor, a factorweight of at least 0.5 was selected as the thresholdvalue. The factor scores were then computed, whichrepresent a measure of each subject’s contribution toeach factor.

Data analyses. Parametric statistical analyses wereemployed because the data were normally distributed.The continuous data were analyzed by Student’s t-test.Wilks lambda discriminant analysis was employed todetermine the factor scores that distinguished the sub-grouped patients according to the assessment of CO2

TABLE 1. Dyspnea questionnaire

Number Question

S1 My breath does not go in all the wayS2 My breathing requires effortS3 I feel that I am smotheringS4 I feel a hunger for more airS5 My breathing is heavyS6 I cannot take a deep breathS7 I feel out of breathS8 My chest feels tightS9 My breathing requires more workS10 I feel that I am suffocatingS11 I feel that my breath stopsS12 I am gasping for breathS13 My chest is constrictedS14 I feel that my breathing is rapidS15 My breathing is shallowS16 I feel that I am breathing moreS17 I cannot get enough airS18 My breath does not go out all the wayS19 My breathing requires more concentration

Perna et al.34

reactivity. The correlation between variables was testedusing Pearson’s correlation coefficient. All analyseswere made using the Statistica 5.0 software package.

RESULTSOut of 78 subjects, 67 (85%) patients were agor-

aphobic; 45 (58%) had a CO2-induced panic attack,and 33 (42%) did not; 52 (67%) patients rated theCO2-induced reaction ‘‘similar’’ to their unexpectedpanic attacks, whereas 26 (33%) did not; 7 (15.5%) of45 patients with an induced panic attack rated theCO2-induced reaction ‘‘dif ferent’’ from their unex-pected panic attacks and 14 (42.4%) of 33 without an

induced panic attack rated the CO2-induced reaction‘‘similar’’ to their unexpected panic attacks (Table 2).

The patient subgroups did not dif fer significantly inillness duration or in severity of clinical symptomatol-ogy as measured by PASS and FQ global and subscalescores, VAS-A pre-CO2 scores and STAI scores (dataavailable on request), except for the higher STAI scoresin patients with reactions similar to their panic attackscompared to those whose reactions were different(44.5710.2 and 38.2711.4, respectively; t¼ 2.71,df¼ 76, Po.05). The mean number of chosen descrip-tors was 5.3673.97. The frequencies of choice of eachdescriptor after CO2 inhalation are listed in Figure 1.Descriptor S6, ‘‘I cannot take a deep breath,’’ and S1,‘‘My breath does not go in all the way,’’ were chosen by450% of the sample, whereas S16, ‘‘I feel that I ambreathing more,’’ was chosen by only five subjects(6.4%; see Fig. 1). The frequencies of choice did notsignificantly dif fer between the genders. The choice ofthe descriptors was not influenced by age or educa-tional level. Three factors were extracted and rotatedby varimax rotation. The descriptors within each factorwere ranked starting from those with the highest factorweight to those with the lowest. The names definingeach factor were obtained by both interpreting theoverall meaning of the descriptors within that factorand by considering that the higher the factor weight,the more each descriptor contributed to the interpreta-tion of that factor. The factors breathing effort (F1),sense of suffocation (F2) and rapid breath (F3)accounted for over 80% of the variance explained(Table 3).

Fig. 1. Frequency of choice of each descriptor on the Dyspnea Questionnaire.

TABLE 2. Assessment of CO2 reactivity

CO2-induced PA(N¼ 45)n

No CO2-inducedPA (N¼ 33)nn

Similar to unexpectedPA (n¼ 52)a

38 14

No similarity to unexpectedPA (n¼ 26)b

7 19

nNumber of patients who reported a CO2-induced panic attack.nnNumber of patients who did not report a CO2-induced panicattack.aNumber of patients whose CO2-induced reaction was similar totheir unexpected panic attacks.bNumber of patients whose CO2-induced reaction was different fromtheir unexpected panic attacks.

Research Article: Dyspnea in Panic Disorder 35

Wilk’s l discriminant analysis showed that the factorscores for the factors breathing effort and sense ofsuffocation significantly discriminated between pa-tients reporting CO2-induced panic attacks and thosewho did not (F¼ 3.63, Wilk’s l¼ 0.87, P o .01) andthat the factor scores for breathing effort significantlydiscriminated between patients whose reaction re-sembled their unexpected panic attacks and thosewhose reaction did not (F¼ 3.63, Wilk’s l¼ 0.87,P o .01) (Table 4).

Linear Pearson’s correlation showed no significantcorrelation between STAI scores and factor scores ineither the whole sample or the subgroups. The wholesample was also subgrouped using as cut-off values themedian scores of the STAI and the VAS-A pre-CO2

scales (42 and 20, respectively). Wilk’s l discriminantanalysis showed that the factor scores for F1, F2, andF3 did not significantly discriminate either betweenpatients with a STAI score above and below the median(n¼ 42 and n¼ 35, respectively) or between patientswith a VAS-A pre-CO2 score above and below themedian (n¼ 39 and n¼ 39, respectively) (data availableupon request).

DISCUSSION

Our findings show that the three factors of breathingeffort (F1), sense of suffocation (F2) and rapid breath(F3) best described the sensations of breathingdiscomfort induced by the 35% CO2 challenge in theoverall sample of patients with PD. The assessment ofCO2 reactivity showed that CO2-induced panic attacksand CO2 reactions similar to unexpected panic attacksdid not overlap, indicating that the use of differentcriteria produces dif ferent rates of panic [Rassovskyand Kushner, 2003]. The discriminating factor scoresin patients subgrouped according to the assessment didnot completely overlap. The factor scores for breathingeffort and sense of suffocation significantly discrimi-nated between the patients reporting an induced panicattack and those who did not, whereas only the factorscores for breathing effort significantly discriminatedbetween the patients whose CO2 reaction resembledtheir unexpected panic attacks and those whoseresponse did not. Duration of illness, severity ofclinical symptomatology and baseline/anticipatory an-xiety did not explain the results. Baseline anxiety in

TABLE 3. Factors obtained by factor analysis of the 19 dyspnea questionnaire descriptors

Factor Descriptorn Explained variance (%)

F1: Breathing Effort S2: My breathing requires effort 34S17: I cannot get enough airS13: My chest is constrictedS11: I feel that my breath stopsS1: My breath does not go in all the wayS8: My chest feels tightS19: My breathing requires more concentration

F2: Sense of Suffocation S10: I feel that I am suffocating 27S12: I am gasping for breathS3: I feel that I am smotheringS4: I feel a hunger for more air

F3: Rapid breath S15: My breathing is shallow 20S14: I feel that my breathing is rapidS7: I feel out of breath

nBased on their factor weight, the descriptors (saturation in the factor 4.5) are ordered within each factor from the highest to the lowest value.

TABLE 4. Discriminant analysis of factor scores in subgroups of patients with PD

Score

Subgroup F1: Breathing effort F2: Sense of suffocation F3: Rapid breath

Patients with CO2-induced panic attack (n¼ 45) 0.20 (1.08)a 0.21 (1.05)b 0.08 (1.15)Patients without CO2-induced panic attack (n¼ 33) �0.26 (0.81)c �0.30 (0.85)d �0.12 (0.73)Patients whose reaction was similar to theirunexpected panic attacks (n¼ 52)

0.20 (1.06)e 0.15 (1.01) 0.01 (1.12)

Patients whose reaction was different from theirunexpected panic attacks (n¼ 26)

�0.40 (0.72)f �0.30 (0.90) �0.02 (0.71)

Values are expressed as mean (7sd).aMean is significantly different from c at Po.05.bMean is significantly different from d at Po.05.eMean is significantly different from f at Po.05.

Perna et al.36

patients whose CO2-induced response was similar totheir unexpected panic attacks was higher than in thosewhose response was different; however, it was notcorrelated with factor scores. Moreover, the factorscores did not significantly discriminate betweenpatients with high or low baseline anxiety and anxietybefore the 35% CO2 challenge test.

The dyspnea factors obtained in our sample provideroom for possible conjectures on the neurophysiologi-cal pathways involved in the experience of breathingdiscomfort in PD and in the panic–respiration con-nection. The factor ‘‘breathing effort’’ describes theconscious awareness of muscular effort during activa-tion of respiratory skeletal muscles that is thought toarise from a dissociation between a central respiratorymotor command and a mechanical response of therespiratory muscles [Manning and Schwartzstein, 1995;Meek and Schwartzstein, 1999]. The associationbetween a panic response to CO2 challenge and thefactor breathing effort may be explained by theabnormalities found in the respiratory function ofpatients with PD. A thoracic breathing pattern [Beckand Scott, 1988] and a higher irregularity andinstability in baseline breathing patterns are character-istic of patients with PD [Abelson et al., 2001;Caldirola et al., 2004; Wilhelm et al., 2001; Yeraganiet al., 2002]. These characteristics could affect theability of patients with PD to maintain an adequaterespiratory homeostasis when external or internalchanges occur. A dissociation between the increasedcentral respiratory command stimulated by CO2

and a decreased mechanical efficiency of the respira-tory response might, therefore, lead to a heightenedsense of breathing effort during the CO2 challengetest.

The factor ‘‘sense of suffocation’’ arises mainlyfrom the stimulation of chemoreceptors [Meek andSchwartzstein, 1999] and involves the activation ofthe limbic/paralimbic and cerebellar regions [Banzettet al., 2000; Corfield et al., 1995; Evans et al., 2002;Liotti et al., 2001; Parson et al., 2001]. In our study,the CO2-induced panic attacks were linked to highfactor scores for sense of suffocation, but this linkdisappeared after patients were subgrouped accordingto the similarity of CO2 responses to their un-expected panic attacks. Studies on the chemosensi-tivity of patients with PD have produced conflictingresults; the most recent study showed no abnormalchemoreflex threshold and sensitivity in PD [Katz-man et al., 2002]. Our findings suggest that althoughchemosensitivity is involved in CO2 reactivity, it maynot be central to unexpected panic attacks. This idearemains speculative, because the comparison withunexpected panic attacks was retrospective andindirect. Confirmation of this idea needs furtherinvestigation into breathing discomfort after unex-pected panic attacks. The ‘‘rapid breath’’ factor,which describes a shallow, rapid breathing patternin response to an excessive mechanical load [Mahler,

1998; Manning and Mahler, 2001] was not associatedwith CO2 reactivity.

Besides the role the respiratory control centers playin the regulation of breathing, the experience ofbreathing discomfort is probably also modulated bycontextual and cognitive influences [Meek andSchwartzstein, 1999]. In fact, brain imaging studieshave shown that CO2-induced breathlessness in healthysubjects involves the limbic/paralimbic areas [Banzettet al., 2000; Corfield et al., 1995; Liotti et al., 2001];hence, it cannot be excluded that emotional processingcould have influenced dyspnea perception in oursample. The lack of effect of cognitive behavioraltherapy on respiratory hyperactivity to CO2 or ofcognitive manipulation on doxapram-induced irregularbreathing patterns in patients with PD [Abelson et al.,2001; Gorman et al., 1997] does not support the ideathat fear/anxiety fully explains the respiratory discom-fort in PD. Recent brain imaging studies have in factshown that the consciousness of breathlessness acti-vates brain areas like the cerebellum that do not seemto be involved in conditioned fear/anxiety models;instead, such areas probably underlie basic emotionslinked to an organism’s survival functions [Evans et al.,2002; Klein, 2002; Liotti et al., 2001; Parson et al.,2001]. These anatomical circuits may be involved inKlein’s ‘‘false suffocation alarm’’ in PD [Klein, 2002].CO2-induced dyspnea may be influenced by anabnormal function in brain circuits processing physio-logical perceptions/sensations linked to basic homeo-static functions. Along this line of thought, anabnormal processing of information about the organ-ism’s internal milieu in the brain circuits subservingbasic emotions [Damasio, 1999] may lead to anabnormal modulation of the respiratory basic emotionsin patients with PD.

CONCLUSIONSIn CO2-provoked panic attacks, the sense of breath-

ing effort is the most peculiar dyspnea sensation; it maybe linked to a decreased efficiency of mechanicalresponses of the respiratory system in patients with PD,whereas the sense of suffocation may be less specificfor panic attacks. Further research is needed to explorethis idea. For example, imaging studies of specific brainstructures related to respiratory sensations may be ableto identify the brain circuits possibly involved in‘‘respiratory’’ panic. Building on such evidence, re-spiratory training of inspiratory and expiratory musclefunction could enhance the overall efficiency of therespiratory system in patients with PD.

REFERENCESAbelson JL, Weg JG, Nesse RM, Curtis GC. 2001. Persistent

respiratory irregularity in patients with panic disorder. BiolPsychiatry 49:588–595.

Research Article: Dyspnea in Panic Disorder 37

Argyle N, Deltito J, Allerup P, Maier W, Albus M, Nutzinger D,Rasmussen S, Ayuso JL, Bech P. 1991. The Panic-AssociatedSymptom Scale: Measuring the severity of panic disorder. ActaPsychiatr Scand 83:20–26.

Battaglia M, Perna G. 1995. The 35% CO2 challenge in panicdisorder: Optimization by receiver operating characteristic (ROC)analysis. J Psychiatr Res 29:111–119.

Banzett RB, Mulnier HE, Murphy K, Rosen SD, Wise RJ, Adams L.2000. Breathlessness in humans activates insular cortex. Neurore-port 11:2117–2120.

Banzett RB, Moosavi SH. 2001. Dyspnea and pain: Similarities andcontrasts between two very unpleasant sensations. American PainSociety Bulletin 11(2).

Beck JG, Scott SK. 1988. Physiological and symptom responses tohyperventilation: A comparison of frequent and infrequentpanickers. J Psychopathol Behav Assoc 10:117–127.

Bellodi L, Perna G. 1998. The panic respiration connection. Milan,Italy: MDM Medical Media.

Caldirola D, Bellodi L, Caumo A, Migliarese G, Perna G. 2004.Approximate entropy of respiratory patterns in panic disorder. AmJ Psychiatry 161:79–87.

Corfield DR, Fink GR, Ramsay SC, Murphy K, Harty HR, WatsonJD, Adams L, Frackowiak RS, Guz A. 1995. Evidence for limbicsystem activation during CO2-stimulated breathing in man.J Physiol 488:77–84.

Damasio AR. 1999. The feeling of what happens. Body and emotionin the making of consciousness. New York: Harcourt-Brace.

Evans KC, Banzett RB, Adams L, McKay L, Frackowiak R, CorfieldDR. 2002. BOLD fMRI identifies limbic, paralimbic, and cerebellaractivation during air hunger. J Neurophysiol 88:1500–1511.

Gorman JM, Browne ST, Papp LA, Martinez J, Welkowitz L, CoplanJD, Goetz RR, Kent J, Klein DF. 1997. Effect of antipanictreatment on response to carbon dioxide. Biol Psychiatry 42:982–991.

Hollander E, Liebowitz M, Gorman J, Cohen B, Fyer A, Klein DF.1989. Cortisol and sodium lactate induced panic. Arch GenPsychiatry 46:135–140.

Katzman MA, Struzik L, Vijay N, Coonerty-Femiano A, Mahamed S,Duffin J. 2002. Central and peripheral chemoreflex in panicdisorder. Psychiatry Res 113:181–192.

Klein DF. 1993. False suffocation alarms, spontaneous panics, andrelated conditions. An integrative hypothesis. Arch Gen Psychiatry50:306–317.

Klein DF. 2002. Response differences of spontaneous panic and fear.Arch Gen Psychiatry 59:567–569.

Liebowitz MR, Gorman JM, Fyer AJ, Levitt M, Dillon D, Levy G,Appleby IL, Anderson S, Palij M, Davies SO, et al. 1985. Lactateprovocation of panic attacks. II. Biochemical and physiologicalfindings. Arch Gen Psychiatry 42:709–719.

Liotti M, Brannan S, Egan G, Shade R, Made L. 2001. Brainresponse associated with consciousness of breathlessness. ProcNatl Acad Sci USA 98:2035–2040.

Mahler DA. 1998. Dyspnea in chronic obstructive pulmonary disease.Monaldi Arch Chest Dis 53:669–671.

Mahler DA, Harver A, Lentine T, Scott JA, Beck K, SchwartzsteinRM. 1996. Descriptors of breathlessness in cardiorespiratorydiseases. Am J Respir Crit Care Med 154:1357–1363.

Manning H, Schwartzstein RM. 1999. Dyspnea and the control ofbreathing. In: Altose MD, Kawakami Y, editors. Control ofbreathing in health and disease. New York-Basel: Marcel Dekker.p 105–129.

Manning HL, Mahler DA. 2001. Pathophysiology of dyspnea.Monaldi Arch Chest Dis 56:325–330.

Manning HL, Schwartzstein RM. 1995. Pathophysiology of dyspnea.N Engl J Med 333:1547–1553.

Marks IM, Mathews AM. 1979. Brief standard self-rating for phobicpatients. Behav Res Ther 17:263–267.

Meek PM, Schwartzstein RM. 1999. Dyspnea. Mechanisms, assess-ment, and management: A consensus statement. Am J Respir CritCare Med 159:321–340.

Moy ML, Woodrow Weiss J, Sparrow D, Israel E, SchwartzsteinRM. 2000. Quality of dyspnea in bronchoconstriction differsfrom external resistive loads. Am J Respir Crit Care Med 162:451–455.

O’Donnell DE, Bertley JC, Chau LK, Webb KA. 1997. Qualitativeaspects of exertional breathlessness in chronic airflow limitation:pathophysiologic mechanisms. Am J Respir Crit Care Med155:109–115.

Parson LM, Egan G, Liotti M, Brannan S, Denton D. 2001.Neuroimaging evidence implicating cerebellum in the experienceof hypercapnia and hunger for air. Proc Natl Acad Sci USA98:2041–2046.

Perna G, Battaglia M, Garberi A, Arancio C, Bertani A, Bellodi L.1994. Carbon dioxide/oxygen challenge test in panic disorder.Psychiatry Res 52:159–171.

Pols H, Zanderbergen J, de Loof C, Griez E. 1991. Attenuation ofcarbon dioxide-induced panic after clonazepam treatment. ActaPsychiatr Scand 84:585–586.

Rassovsky Y, Kushner MG. 2003. Carbon dioxide in the study ofpanic disorder: Issue of definition, methodology, and outcome. JAnxiety Disord 17:1–32.

Sheehan DV, Lecrubier Y, Janavs J, Knapp E, Weiler EBP, SheedanM, Amorim P, Baker R, Sheehan KH, Lepine JP. 1994. MiniInternational Neuropsychiatric Interview. University of SouthFlorida Institute for Research in Psychiatry, Tampa, FL andINSERM Hopital de la Salpetriere, Paris, France.

Sinha SS, Coplan JD, Pine DS, Martinez JA, Klein DF, Gorman JM.1999. Panic induced by carbon dioxide inhalation and lack ofhypothalamic-pituitary-adrenal axis activation. Psychiatry Res86:93–98.

Simon PM, Schwartzstein RM, Weiss JW, Lahive K, Fencl V,Teghtsoonian M, Weinberger SE. 1989. Distinguishable sensationsof breathlessness induced in normal volunteers. Am Rev Respir Dis140:1021–1027.

Spielberger C, Gorsuch R, Lushere R. 1969. STAI Manual for theState-Trait Inventory. Palo Alto: Consulting Psychologists Press.

Tobin MJ. 1990. Dyspnea. Pathophysiologic basis, clinical presenta-tion, and management. Arch Intern Med 150:1604–1613.

Wilhelm FH, Trabert W, Roth WT. 2001. Characteristics of sighingin panic disorder. Biol Psychiatry 49:606–614.

Yeragani VK, Radhakrishna RK, Tancer M, Uhde T. 2002. Nonlinearmeasures of respiration: Respiratory irregularity and increasedchaos of respiration in patients with panic disorder. Neuropsycho-biology 46:111–120.

Perna et al.38