Carbon Dioxide Hypersensitivity in Separation-
Anxious Offspring of Parents with Panic Disorder
Roxann Roberson-Nay, Donald F. Klein, Rachel G. Klein, Salvatore Mannuzza, John L. Moulton III,
Mary Guardino, and Daniel S. Pine
the participants’ homes. Anxiety symptoms, panic attacks, and respiratory physiology (respiratory frequency and tidal volume) were
monitored during baseline and 15-min maintained CO2breathing.
Results: As hypothesized, significant offspring SAD ? parent PD interactions were obtained for anxiety symptoms, respiratory frequency,
tidal volume, and a panting index during CO2inhalation. Offspring with both SAD and parental PD exhibited more anxiety symptoms at
at particularly high risk for adult PD.
respiratory frequency, separation anxiety disorder, tidal volume
proximate causal mechanism of panic attacks. During hyperven-
tilation, more carbon dioxide (CO2) is eliminated than produced,
thus inducing respiratory alkalosis, a putative panicogen. The
hyperventilation hypothesis was previously tested by having
patients with PD hyperventilate both in room air and in 5%
CO2-enriched air (1). Five percent CO2was chosen because this
level approximates the usual lung concentration of CO2. It was
predicted that breathing 5% CO2air would prevent panic by
respiratory alkalosis, whereas hyperventilating in ambient air
would induce panic. Surprisingly, the opposite was found. The
connection between PD and CO2hypersensitivity has now been
well-documented, characterized by perturbed ventilation and
increased panic attacks and anxiety symptoms (2–11). Perturbed
ventilation includes increased respiratory frequency, coupled
with lowered tidal volume (8–9,12–14). This respiratory re-
sponse pattern during CO2exposure is akin to “panting,” which
decreases gas exchange by increasing the dead space proportion
of each breath.
Klein hypothesized a specific, evolved suffocation alarm system
(15). The initial adaptive response to smothering (e.g., maternal
ncued panic attacks, the defining clinical features of
panic disorder (PD), have been the focus of active
research. An early view proposed hyperventilation as the
overlay of neonate or sudden exposure to high CO2environ-
ment) would be hyperventilation combined with escape and
protest. However, if hyperventilation and escape failed, because
the asphyxiating environment was inescapable, panting would
be the default adaptive response, in an attempt to prevent a
further rise in partial pressure of CO2and increased respiratory
acidosis. Klein hypothesized that a hypersensitive suffocation
alarm system would lead to false alarms manifested as “uncued”
panic attacks (15).
Family studies suggest that CO2hypersensitivity represents a
vulnerability marker for PD. For instance, increased minute
ventilation has been observed in relatives of PD probands
compared with relatives of low-risk subjects (16). Healthy rela-
tives of PD patients also report more anxiety compared with
healthy relatives of healthy subjects after inhalation of 35% CO2
(17). Moreover, patients with PD who are hypersensitive to CO2
are three times more likely to have a first-degree relative with PD
than those without CO2hypersensitivity (18). A recent twin study
showed higher concordance for CO2-induced panic attacks
among monozygotic than dizygotic twins (55.6% vs. 12.5%) (19).
Given that the major source of familial risk for PD seems to be
genetic (19,20), heritable aspects of CO2-hypersensitivity and PD
might overlap significantly. Findings from a recent population-
based twin study of shared genetic influence for CO2hypersen-
sitivity and uncued panic attacks support this hypothesis (21).
This study also suggests that, among adults, a prior history of
separation anxiety disorder (SAD) moderates the association
between genetic risk for PD and CO2-hypersensitivity. A more
recent adult twin study by the same team found that genetic
factors strongly influence the covariation among childhood
separation anxiety disorder, CO2hypersensitivity, and adult PD
(22). The familiality of anxiety disorders and the putative rela-
tionship between PD and SAD prompted the current study that
relates CO2hypersensitivity to familial risk for PD in children
Studies examining CO2hypersensitivity in children with anx-
iety disorders or at risk for PD have relied on maintained 5% CO2
protocols that expose subjects to room air, followed by 5%
bia University (DFK); New York University Child Study Center (RGK, SM,
JLM), New York; Nathan S. Kline Institute for Psychiatric Research (SM,
MG), Orangeburg; Freedom from Fear (MG), Staten Island, New York;
Section on Development and Affective Neuroscience (DSP), National
Institute of Mental Health Intramural Research Program, Bethesda,
Address correspondence to Roxann Roberson-Nay, Ph.D., Virginia Com-
monwealth University, Department of Psychiatry, PO Box 980126, Rich-
mond, VA 23298; E-mail: firstname.lastname@example.org.
Received Feb 17, 2009; revised Nov 26, 2009; accepted Dec 11, 2009.
BIOL PSYCHIATRY 2010;67:1171–1177
© 2010 Society of Biological Psychiatry
CO2-enriched air for various time periods (23–25). Clinical
samples of children with anxiety disorders have been found to
request termination of the CO2task at higher rates, report more
panic symptoms, and exhibit more rapid respiratory rate in
response to CO2than healthy peers (23). Children with SAD
exhibit signs of CO2hypersensitivity consisting of relatively
greater reports of dyspnea and steeper respiratory frequency
slopes (24), much like patients with PD during exposure to
maintained 5% CO2. In contrast, this response does not occur in
children with social phobia. Children with generalized anxiety
disorder do not differ from healthy subjects on respiratory-rate or
dyspnea response to CO2but do report increased anxiety. In
sum, SAD but not social phobia or generalized anxiety disorder
is more consistently associated with CO2hypersensitivity.
Only a subset of children with SAD develop PD (26,27). It
might be that it is those individuals at familial risk for later-life PD
who selectively exhibit respiratory correlates of PD at an early
age. We hypothesized that children with SAD who also have
parents with PD would experience greater symptomatic change
and respiratory perturbations (i.e., increased respiratory rate and
decreased tidal volume) during CO2inhalation, compared with
other offspring (i.e., offspring without SAD who are at high risk
for PD [parents have PD], offspring with SAD who are at low risk
for PD [no parental PD], and offspring with neither risk factor [no
SAD and no parental PD]). We also tested the hypothesis that,
during CO2breathing, conversion to panting respiration, indexed
by a decreasing ratio of tidal volume to respiratory frequency
over time, would be significantly greater in offspring at genetic
risk for PD and SAD than in the other offspring groups.
We previously reported the separate main effects of offspring
SAD and parental PD on CO2effects in a partial sample of 142
offspring (24). This sample was too small to test an SAD ?
parental PD interaction. This report is based on the completed
sample of 212 offspring from this same high-risk study.
Methods and Materials
Two hundred twelve biological offspring of 135 families were
recruited, with at least one parent with a lifetime diagnosis of PD
(n ? 57 PD? families; n ? 88 offspring) or neither parent with
a history of PD (n ? 78 PD? families; n ? 124 offspring). Of the
57 families in which at least one parent met criteria for PD (n
mothers ? 47; n fathers ? 11), there was only 1 family in which
both the father and mother had a history of PD. Age of PD onset
was 27.30 years (SD ? 9.69) for mothers and 28.73 for fathers
(SD ? 9.40). Twenty of 47 mothers (41.7%) and 5 of 11 fathers
(45.5%) with PD were currently symptomatic. Demographic data of
parents with and without PD did not differ significantly (Table 1).
Subjects also included offspring of parents with a history of
major depression (MD). In a series of separate analyses, parental
MD was not associated with any measure of offspring CO2
hypersensitivity. Therefore, parental MD was not included in the
statistical models, and this report focuses on whether measures
of CO2sensitivity are influenced by an interaction between
offspring SAD and parental PD, regardless of parental MD.
Offspring (n ? 212) were classified on the basis of the
presence or absence of a lifetime parental diagnosis of PD and
cross-classified on the basis of current SAD, irrespective of other
ongoing anxiety disorders. The four cross-tabulated groups
were: 1) offspring with both SAD and PD (SAD?/PD?, n ? 13);
2) offspring with SAD but no parental PD (SAD?/PD?, n ? 10);
3) offspring with parental PD but not SAD (SAD?/PD?, n ? 75);
and 4) offspring with neither SAD nor parental PD (n ? 114,
SAD?/PD?). As presented in Table 2, the four risk groups did
not differ significantly on gender composition or percent of
parents with a history of MD, but children with SAD? (SAD?/
PD? and SAD?/PD?) were younger than those without SAD
(SAD?/PD? and SAD?/PD?).
Fourteen subjects had missing respiratory data, due to tech-
nical difficulties. The 198 offspring with respiratory data came
from 126 families. Those included in respiratory data analyses
were: 1) SAD?/PD?, n ? 11; 2) SAD?/PD?, n ? 9; 3)
SAD?/PD?, n ? 69; and 4) SAD?/PD?, n ? 109. The 14
subjects with missing respiratory data were included in all
analyses of symptom-based responding to CO2. The 14 offspring
without respiratory data did not differ significantly from the
others with regard to demographic data or psychopathology
measures, p values ? .05.
Offspring exclusionary criteria were psychosis, mania, perva-
sive developmental disorder, use of psychotropic medication,
IQ ?70, or an acute medical condition. Parents with PD were
past or current outpatients identified via chart review from the
New York State Psychiatric Institute (New York, New York),
Long Island Jewish Medical Center (New Hyde Park, New York),
or Freedom From Fear (Staten Island, New York). Approximately
one-half of healthy parents were recruited through a pediatric
dental clinic, with the remaining one-half recruited through
acquaintances of parents. Inclusion/exclusion criteria were the
same for parent and comparison parents with the exception that
comparison parents could not qualify for a past or present mood
or anxiety disorder.
Full disclosure of study procedures was provided to all
families. Written informed consent was obtained from parents
and offspring 18 years and older. Offspring ages 9–17 provided
assent. Participants were informed that the CO2task would
include periods of breathing room-air and a 5% CO2-enriched air
mixture and that they might experience anxiety during CO2
inhalation. This study was approved by an institutional review
Table 1. Parent Demographic Data
Parent Diagnostic Group
PD? (n ? 78)PD? (n ? 57)
Mother’s Ethnicity n (%)
Father’s Ethnicity n (%)
Mother’s Age, mean yrs (SD)
Father’s Age, mean yrs (SD)
Family SES n (%)
Score on 5-factor Hollingshead scale (33); ? denotes presence of panic
6 PD? families and 5 PD? families.
SES, socioeconomic status.
1172 BIOL PSYCHIATRY 2010;67:1171–1177
R. Roberson-Nay et al.
Parental Diagnoses. Both parents were individually admin-
istered the Structured Clinical Interview for DSM-IV disorders
(SCID) (28) by trained clinicians blind to all family information.
In the event a parent was unable to complete a clinical interview,
the spouse or ex-spouse served as the informant and completed
the Family Informant Schedule and Criteria revised for DSM-IV
(29). All biological mothers (total n ? 135) completed the SCID.
Of the 139 biological fathers (four families had two offspring with
different biological fathers), 73 completed the SCID, with the
remaining 66 cases evaluated with spouse or ex-spouse’s inter-
views. Clinicians provided clinical narratives that detailed current
and lifetime history of DSM-IV diagnoses. Integrity and reliability
of parent clinical interviews was monitored through audiotapes
as well as expert review of clinical narrative summaries. Details
concerning these procedures and diagnostic reliability are pub-
lished elsewhere (23,24).
Offspring diagnosis was based on
parent and child report. Parents and offspring were interviewed
in their home approximately 1 month before the CO2procedure
was completed. Trained psychologists blind to parent diagnoses
administered the Parent As Respondent Informant Schedule (30)
to parents and a child version to offspring. Different staff
conducted parent and offspring interviews. Fidelity of clinical
interviews was monitored through audiotapes. Clinicians wrote
clinical narrative summaries documenting DSM-IV diagnoses,
which were blindly reviewed by an expert clinician for accuracy.
A review by two raters of randomly selected parent and child
interviews (50 child and parent interviews) yielded acceptable
reliability for anxiety and mood disorders (? ? .65). Diagnoses
were coded as present if either the parent or the child report
indicated the presence of a psychiatric disorder.
Participants wore a face mask that covered their mouth and
nose throughout the experimental procedure. The face mask was
connected via gas-impermeable tubing to a three-way stopcock
valve, allowing the experimenter to switch manually from room-
air to the 5% CO2mixture. Connected to this valve was a large
multi-liter balloon reservoir located behind the participant. Dur-
ing the 25-min procedure, offspring breathed room air for 10
min, and 5% CO2-enriched air for 15 min. Participants were
unaware of this schedule. The face mask included an occlusion
pressure system device that transmitted inhalation and exhala-
tion data to a laptop computer with spirometry software. This
device computed respiratory frequency and tidal volume values.
The procedure was conducted in the home in a quiet room
selected by the family. Both a physician and technician, blind to
diagnostic status of parents and offspring, supervised it. Before
the procedure was initiated, children were instructed to signal if
they wished to terminate the CO2inhalation task for any reason.
The physician and technician remained in the room during the
entire CO2procedure, and reiterated that the child could signal if
she/he wished to discontinue. Parents were in an adjacent room
unless they or the child requested that they remain.
Assessment of Symptomatic Response to Maintained 5%
CO2. A trained technician who recorded presence/absence as
well as intensity of anxiety symptoms administered the Acute
Panic Inventory (API), modified and validated for children and
adolescents (21). The API assesses 23 symptoms rated on a 0–3
scale with 0 ? absent, 1 ? mild, 2 ? moderate, and 3-severe.
Participants were asked to point to the rating that best-described
their symptom severity. The API change scores were created by
subtracting baseline API scores (i.e., before attachment of face-
mask and delivery of CO2) from APIs assessed at three time
points: 1) after attachment of facemask and assessment of
additional baseline respiratory physiology before CO2adminis-
tration (during room air breathing), 2) after 5 min of CO2
exposure, and 3) at the end of CO2breathing.
The technician rated panic attacks during CO2inhalation. A
panic attack was operationally defined as an episode of peak
anxiety during which at least four somatic/cognitive symptoms
were reported, similar to previous studies. As in prior studies,
panic attacks required an increase in self-rated anxiety and
increases of 1 point or more on at least four API symptoms
Assessment of Ventilatory Response to Maintained 5%
Respiratory frequency (fR) and tidal volume (VT) were
continuously measured during room-air and CO2exposure with
breath-by-breath spirometry. The average of the initial 10 min of
room air breathing represents baseline values. Mean scores for fR
Table 2. Offspring Risk Group Characteristics
n ? 114
n ? 75
n ? 10
n ? 13
% Female (n)
Mean Age, yrs (SD)
Age Range, yrs
% Offspring with a Parent with
Lifetime MD (n)
% Offspring with Comorbid Anxiety
or Depressive Disorder (n)
67.5 (77) 62.7 (47)90.0 (9)61.5 (8) 3.17.37
social phobia/social anxiety disorder; PTSD, post-traumatic stress disorder; OCD, obsessive compulsive disorder.
R. Roberson-Nay et al.
BIOL PSYCHIATRY 2010;67:1171–1177 1173
and VTmeasured during CO2exposure were computed across
the first 10 30-sec epochs of the 15-min exposure. A panting
index was also created by dividing VTvalues by their corre-
sponding fR values. Analyses were restricted to the first 5 min of
the 15-min inhalation task, because ventilatory measures are
highly linear for the first 5 min, with values plateauing after
approximately 5 min of exposure to CO2-enriched air. This CO2
procedure has been used previously in adults with PD and
pediatric studies of CO2hypersensitivity (6,23,24).
Age was included as a covariate in all analyses, given that
offspring with SAD (mean ? 12.68, SD ? 2.57) were younger
than those without SAD [mean ? 15.44, SD ? 3.00; t(210) ? 4.21,
p ? .001].
To examine API change scores, the SAS Mixed procedure was
used (31). This method allows for inclusion of random effects
(i.e., parent–child clusters) and fixed effects (e.g., parental PD,
offspring SAD). The unstructured covariance structure was spec-
Exact logistic regression was used to examine panic attack
response (yes/no) in offspring. This method was chosen over
asymptotic methods, which might be unreliable when sample
sizes are small or the data are sparse, skewed, or heavily tied.
The SAS Mixed procedure was also used to analyze continu-
ous, repeated measures of fR and VT. The analysis included
random effects and fixed effects. The first order autoregressive
(i.e., type ? ar) covariance structure was specified. Several
variables were entered as covariates in statistical models, includ-
ing baseline fR and VTvalues (respectively), offspring age, and
offspring body mass index.
Panting index values did not approximate a normal Gaussian
curve, demonstrating significant positive skew. Thus, the SAS
GLIMMIX procedure, which allows for random and fixed effects
as well as non-normal distributions, was used to examine panting
index values (31). A ? distribution and log link function specified
and the first-order autoregressive covariance structure was ap-
plied. A baseline VT/fR ratio was created with measures assessed
during the baseline phase and entered as a covariate. Offspring
age and body mass index also were entered as covariates.
All tests of interactions, main effects, and follow-up contrasts
are based on two-tailed tests of significance. Main effects for
offspring SAD and parental PD are not interpreted in the
presence of a significant offspring SAD ? parental PD interac-
tion. Dependent measure estimates presented in the text and
tables are adjusted for variables included in the particular
analysis. Throughout, “?” indicates the presence of a disorder,
and “?” indicates its absence. All analyses were conducted with
SAS version 9.2 (SAS, Cary, North Carolina). Statistical signifi-
cance was defined as ? ? .05. Although we rely on two-tailed
tests of significance for follow-up contrasts, our contrasts are
directional. To bring promising data patterns to notice, p values ?
.05 and ? .10 are reported as trends.
Although most offspring completed the entire 15 min of CO2
exposure, 18% terminated prematurely, creating a range of 65
sec–15 min of CO2-breathing. No differences were found among
the four risk groups in amount of exposure to the CO2-enriched
air (p ? .77).
Interaction Effects of Offspring SAD and Parental PD on
Offspring CO2Sensitivity Symptomatic Anxiety Responses
API. Analysis of the repeated API change scores supported
the hypothesized SAD ? PD ? Trial interaction [F(2468) ? 5.43,
p ? .005], indicating that the change in anxiety symptoms during
the respiratory assessment varied as a function of both offspring
SAD and parental PD across time (Figure 1). Follow-up analysis
of the three-way interaction indicated that offspring in the
SAD?/PD? group reported more symptoms on the API initially
while wearing the facemask and breathing room air compared
with SAD?/PD? and SAD?/PD? offspring [t(468) ? ?3.00,
p ? .003; t(468) ? ?2.46, p ? .01, respectively]. No difference
was detected between offspring in the SAD?/PD? and SAD?/
PD? groups [t(468) ? ?1.49, p ? .14] on anxiety symptoms
while wearing the facemask and breathing room air.
A trend was noted for the SAD?/PD? versus SAD?/PD?
offspring contrast after 5-min exposure to 5% CO2[t(468) ? 1.76,
p ? .08], with the high-risk group reporting more anxiety
symptoms. The SAD?/PD? offspring group did not differ from
the SAD?/PD? or SAD?/PD?offspring groups [t(468) ?
?1.58, p ? .12; t(468) ? ?.53 p ? .60, respectively] after 5 min
of 5% CO2breathing. At the end of CO2breathing, however, the
SAD?/PD? group had greater elevations in anxiety symptoms
compared with the SAD?/PD? and SAD?/PD? groups
[t(468) ? ?2.76, p ? .006; t(468) ? ?2.74, p ? .006], with a
statistical trend observed for the SAD?/PD? and SAD?/PD?
group contrast [t(468) ? ?1.83, p ? .07].
Panic Attack Response. Offspring with SAD at high risk for
PD (SAD?/PD?) had a rate of panic attacks threefold higher (4
of 11, 36.4%) than the other groups, (e.g., SAD?/PD?: 11% [1 of
9]; SAD?/PD?: 10% [7 of 69]; SAD?/PD?: 11% [11 of 109]). But
results of the conditionalexacttestdidnotyieldasignificantSAD ?
PD interaction, as indicated by a nonsignificant exact median
unbiased estimate of the coefficient (1.38, p ? .58).
To increase statistical power, the high-risk group (SAD?/
PD?) was contrasted against all other risk groups combined in a
second analysis. The exact median unbiased estimate of the
coefficient for risk group status was significant (1.61, p ? .05),
with the high-risk group being approximately five times more
likely to experience a panic attack response compared with
offspring with only one or no risk factor.
Figure 1. Acute Panic Inventory (API) scores generated with the SAS mixed
model procedure. Estimated API scores are adjusted for all variables (i.e.,
baseline API, offspring age, Parental panic disorder [PD], Offspring separa-
tion anxiety disorder [SAD], and Trial effect) in the model. CO2, carbon
1174 BIOL PSYCHIATRY 2010;67:1171–1177
R. Roberson-Nay et al.
Respiratory Responses to CO2
Respiratory frequency and VTmeans and SE values across the
10 30-sec epochs are graphically depicted in Figure 2. Tests of
interactions for respiratory rate, tidal volume, and the panting
index are presented in Table 3, and follow-up contrasts are
presented in Table 4.
CO2Effects on Respiratory Frequency. As hypothesized, a
significant two-way interaction between offspring SAD and pa-
rental PD was found. Cell contrasts showed that the SAD?/PD?
group had higher fR during CO2breathing, relative to offspring
with only parental PD (SAD?/PD?). No significant differences
were found between the SAD?/PD? and SAD?/PD? or SAD?/
CO2Effects on Tidal Volume. A significant Offspring SAD ?
Parental PD interaction on VTwas obtained. The SAD?/PD?
offspring group exhibited lower VTcompared with the SAD?/
PD? group. A statistical trend was detected for the SAD?/PD?
and SAD?/PD? contrast. No statistically significant difference
was detected between the SAD?/PD? and SAD?/PD? groups.
CO2Effects on Panting Index. A significant Offspring SAD ?
Parental PD interaction on the panting ratio was found (Figure 3).
Risk group contrasts indicated that the SAD?/PD? offspring
exhibited lower panting ratios, indicating progressively more
panting, compared with SAD?/PD? and SAD?/PD? offspring.
No significant differences were found between SAD?/PD?
offspring and SAD?/PD? offspring.
The current study extends previous findings on enhanced
vulnerability to maintained 5% CO2inhalation in SAD. This is the
first study to test whether parental PD moderates the association
between childhood SAD and CO2hypersensitivity. Offspring of
PD parents with SAD were postulated to represent a distinct
clinical group, possibly at particularly high risk for PD, as
offspring with and without SAD of parents with or without a history of PD
across 5 min of 5% CO2inhalation. Respiratory frequency and tidal volume
values are adjusted for all variables (i.e., baseline, offspring age, offspring
Abbreviations as in Figure 1.
Table 3. Predictors of Offspring fR, VT, and Panting Index During Maintained 5% CO2Inhalation
dfFp dfFp dfFp
PD ? SAD ? trial
PD ? SAD
PD ? trial
SAD ? trial
Parent and offspring disorders are not mutually exclusive. ? denotes a positive association, and ? denotes a negative association between the
independent and dependent variables.
fR, respiratory rate; VT, tidal volume; CO2, carbon dioxide; PD, panic disorder; SAD, separation anxiety disorder; BMI, body mass index.
Table 4. Offspring SAD ? Parental PD Interaction Contrasts
SAD?/PD? vs. SAD?/PD?
SAD?/PD? vs. SAD?/PD?
SAD?/PD? vs. SAD?/PD?
as in Table 3.
R. Roberson-Nay et al.
BIOL PSYCHIATRY 2010;67:1171–1177 1175
manifest by CO2-hypersensitivity. In this study, parental PD
alone was not associated with CO2hypersensitivity in offspring;
however, it had significant influence on CO2response in the
offspring with SAD. Also, offspring with both SAD and parental
PD exhibited greater panic symptoms compared with SAD?/
PD? and SAD?/PD? offspring during threat (i.e., wearing
facemask during ambient air breathing). This finding for the
“threat” epoch extends our previous analysis in PD offspring by
showing that an enhanced level of anxiety symptoms occurs
before CO2inhalation predominantly among offspring with both
parental PD and offspring SAD.
Offspring with both risk factors also differed from all other
risk groups at the end of CO2-breathing. In fact, only offspring
with both SAD of parental PD exhibited increased elevations in
anxiety symptoms from 5 min of CO2breathing to termination of
CO2breathing, whereas all other risk groups’ endorsements
remained stable or decreased. A high rate of panic attacks also
emerged in this high-risk group (36%), compared with the other
three groups, who all exhibited rates in the 10%–11% range.
However, for this dichotomous index, the SAD ? PD interaction
was not significant, likely reflecting low power to detect an
interactive effect. The panic attack rate in the SAD?/PD? group
is similar to the rate reported in adult PD patients (30%–40%)
(6,8,32) during 5% CO2exposure; similarly, the rate of panic
attacks in the other three offspring groups is comparable to that
typically observed in adult healthy subjects (8,32).
Results for measures of respiratory physiology were less
consistent, because of the unexpected responsiveness of the
SAD?/PD? group. An SAD ? PD interaction was found for both
fR and VT, as hypothesized. Interactions for the respiratory
measures were driven by an SAD effect manifest only in the PD?
but not PD? stratum. Specifically, offspring with both SAD and
parental PD had the highest CO2-induced respiratory rate cou-
pled with reduced tidal volume, differing from offspring who had
parents with PD but no SAD, on measures of respiratory rate and
tidal volume. The high-risk offspring group also differed from
SAD?/PD? offspring on tidal volume. Unexpectedly, the
SAD?/PD? group did not differ from the SAD?/PD? group on
any respiration related measures. Post hoc analysis of a panting
index also suggests that youth with SAD at genetic risk for PD
might be attempting to reduce CO2levels by engaging in rapid,
shallow breathing. It should be noted that, again, not all be-
tween-group contrasts were significant, with the SAD?/PD?
group and SAD?/PD? group not differing significantly. Never-
theless, results from analyses of respiratory physiology indicate
that parental PD moderates the association between SAD and
respiratory physiology and that the ventilatory patterns observed
here are consistent with those identified among adult PD pa-
To date, much of the research examining disorder-specific
markers has yielded disappointing results. Some of the markers
investigated among the anxiety disorders include autonomic
indicators such as heart rate, heart rate variability, and skin
conductance as well as stress-related endocrine indexes such as
cortisol. None have demonstrated diagnostic specificity; rather,
each tends to be associated with global changes in arousal, and
each relates to some extent to the presence of high levels of
anxiety symptoms as opposed to any specific anxiety disorder or
symptom profile. To understand better the etiologic or patho-
physiological underpinnings of specific anxiety disorders, detec-
tion of disorder-specific markers might prove essential. In the
case of childhood SAD, respiratory system dysregulation caused
by CO2holds promise as a distinct objective marker.
A limitation of this study is that we did not systematically
record whether parents remained in the room with the child
during CO2breathing. It seems unlikely that this factor affected
outcomes, because SAD main effects on CO2-induced respiratory
perturbations also have been observed in studies using labora-
tory settings in which parents are not present during testing (23).
Although our results are generally supportive of study hypothe-
ses, tests of offspring SAD and parent PD interactions relied on
relatively small cell sizes, and contrasts directly comparing
children with both risk factors with children with one or no risk
factor did not consistently support differences, despite the pres-
ence of a significant interaction. At the same time, it is all the
more striking that support for interaction effects were obtained
with small samples. Ideally, future studies should seek to collect
larger samples and implement longitudinal follow-ups to clarify
the distinction between separation anxiety with and without
parental PD, with regard to course and comorbidity and possibly
treatment and prevention.
This work was supported by the National Institute of Mental
Health Grant R01 MH-59171 and a Grant from the Nick Traina
The authors report no biomedical financial interests or po-
tential conflicts of interest.
1. Gorman, JM, Fyer MR, Goetz R, Askanazi J, Liebowitz MR, Fyer AJ, et al.
2. Kent JM, Papp LA, Martinez JM, Browne ST, Coplan JD, Klein DF, et al.
A comparison with major depression and premenstrual dysphoric dis-
3. Papp LA, Klein DF, Gorman JM (1993): Carbon dioxide hypersensitivity,
hyperventilation, and panic disorder. Am J Psychiatry 150:1149–1157.
of control on panic attacks induced via inhalation of 5.5% carbon diox-
5. Klein DF (1998): Panic and phobic anxiety: Phenotypes, endopheno-
types, and genotypes. Am J Psychiatry 155:1147–1149.
6. Gorman JM, Fyer MR, Goetz R, Askanazi J, Liebowitz MR, Fyer AJ, et al.
7. Gorman JM, Kent J, Martinez J, Browne S, Coplan J, Papp LA (2001):
Figure 3. Log-transformed panting ratio (respiratory frequency/tidal vol-
are adjusted for all variables (i.e., baseline, offspring age, offspring body
abbreviations as in Figure 1.
1176 BIOL PSYCHIATRY 2010;67:1171–1177
R. Roberson-Nay et al.
8. Papp LA, Martinez JM, Klein DF, Coplan JD, Norman RG, Cole R, et al.
(1997): Respiratory psychophysiology of panic disorder: Three respira-
tory challenges in 98 subjects. Am J Psychiatry 154:1557–1565.
9. Pain MC, Biddle N, Tiller JW (1988): Panic disorder, the ventilatory re-
sponse to carbon dioxide and respiratory variables. Psychosom Med
Carbon dioxide induced panic attack in panic disorder in Japan. Prog
11. Lousberg H, Griez E, van den Hout MA (1988): Carbon dioxide chemo-
sensitivity in panic disorder. Acta Psychiatr Scand 77:214–218.
12. Goetz RR, Klein DF, Gully R, Kahn J, Liebowitz MR, Fyer AJ, et al. (1993):
Panic attacks during placebo procedures in the laboratory. Physiology
ing tests in panic disorder. Biol Psychiatry 38:240–245.
CO2in panic disorder. Psychiatry Res 39:13–19.
15. Klein DF (1993): False suffocation alarms, spontaneous panics, and re-
sensitivity to CO2as a trait of familial panic disorder. Biol Psychiatry
first-degree relatives of patients with panic disorder. Biol Psychiatry
18. Perna G, Cocchi S, Bertani A, Arancio C (1995): Sensitivity to 35% CO2in
healthy first-degree relatives of patients with panic disorder. Am J Psy-
19. Bellodi L, Perna G, Caldirola D, Arancio C, Bertani A, Di Bella D (1998):
CO2-induced panic attacks: A twin study. Am J Psychiatry 155:1184–
the genetic epidemiology of anxiety disorders. Am J Psychiatry 158:
21. Battaglia M, Pesenti-Gritti P, Spatola CAM, Ogliari A, Tambs K (2008): A
to hypercapnia and panic disorder. Am J Med Genet 147B:586–593.
22. Battaglia M, Pesenti-Gritti P, Medland SE, Ogliari A, Tambs K, Spatola
CAM (2009): A genetically informed study of the association between
childhood separation anxiety, sensitivity to CO(2), panic disorder, and
the effect of childhood parental loss. Arch Gen Psychiat 66:64–71.
23. Pine DS, Coplan JD, Papp LA, Klein RG, Martinez JM, Kovalenko P, et al.
24. Pine DS, Klein RG, Coplan JD, Papp LA, Hoven CW, Martinez J, et al.
(2000): Differential carbon dioxide sensitivity in childhood anxiety dis-
orders and nonill comparison group. Arch Gen Psychiatry 57:960–967.
25. Pine DS, Klein RG, Roberson-Nay R, Mannuzza S, Moulton JL III, Wolde-
adolescents: Relationship to panic disorder in parents and anxiety dis-
orders in subjects. Arch Gen Psychiatry 62:73–80.
(2003): Is childhood separation anxiety disorder a predictor of adult
27. Klein RG (1995): Is panic disorder associated with childhood separation
anxiety disorder? Clin Neuropharmacol 18:7–14.
28. Spitzer RL, Williams JB, Gibbon M, First MB (1992): The Structured Clini-
cal Interview for DSM-III-R (SCID). I: History, rationale, and description.
29. Mannuzza S, Fyer AJ (1990): Family Informant Schedule and Criteria
Anxiety Disorders Clinic.
ity of maternal reports of lifetime mental disorders in their children. J
31. Littell RC, Milliken GA, Stroup WW, Wolfinger RD (2002): SAS System for
Mixed Models. Cary, North Carolina: SAS Institute.
responses to carbon dioxide and breath holding challenges in individ-
uals with panic disorder. J Psychsom Rev 60:291–298.
33. Hollingshead AB, Redlich FC (1958): Social Class and Mental Illness: A
Community Study. New York: John Wiley & Sons.
R. Roberson-Nay et al.
BIOL PSYCHIATRY 2010;67:1171–1177 1177