Intermittent Hypoxia Induces Early Functional
Cardiovascular Remodeling in Mice
Maurice Dematteis1,2,3*, Cécile Julien1,2,4,5*, Christiane Guillermet6, Nathalie Sturm6, Sylvie Lantuejoul6,7,
Michel Mallaret4, Patrick Lévy1,2,3**, and Evelyne Gozal5,8**
1
INSERM ERI17, Grenoble, France; 2Faculté de Médecine, IFR1, Université Joseph Fourier, Grenoble, France; 3Laboratoires du Sommeil et EFCR, and
4
Laboratoire de Pharmacologie–CEIP, CHU, Hôpital A. Michallon, Grenoble, France; 5Department of Pediatrics, KCHRI, University of Louisville,
Louisville, Kentucky; 6Département de Pathologie, CHU, Hôpital A. Michallon, Grenoble, France; 7INSERM U823/UJF, Institut Albert Bonniot,
Grenoble, France; and 8Department of Pharmacology and Toxicology, University of Louisville, Louisville, Kentucky
Rationale: Intermittent hypoxia, a hallmark of sleep apnea, is a major
factor for hypertension and impaired vasoreactivity.
Objectives: To examine the temporal occurrence of these two out-
comes in order to provide insight into mechanisms and early cardio-
vascular disease identification.
Methods: Functional and structural cardiovascular alterations were
assessed in C57BL6 mice exposed to intermittent hypoxia (21–4% FIO2,
30-s cycle, 8 h/d) or air for up to 35 days. Blood pressure, heart rate, and
urinary catecholamines were measured at Days 1 and 14. Hindquarter
vasoreactivity was assessed at Days 14 and 35, including vasoconstric-
tion to norepinephrine, endothelium-, and non–endothelium-dependent
vasodilation. Aorta, heart, and hindquarter skeletal muscles were immu-
nostained for vascular markers PECAM-1 and collagen IV.
Measurements and Main Results: Hemodynamic alterations occurred from
Day 1, characterized by blood pressure surges with bradytachyarrhyth-
mia driven by cyclic hypoxia. At Day 14, blood pressure at normoxia was
elevated, with predominant diastolic increase. With hypoxia, vasopres-
sive catecholamines were elevated, blood pressure surged with a lower
hypoxic threshold, whereas heart rate fluctuations decreased. Histo-
logic alterations started from Day 14, with decreased endothelial
PECAM-1 expression in descending aorta and left heart. Impaired
peripheral vasoreactivity occurred at Day 35, including hypervaso-
constriction to norepinephrine secondary to sympathetic hyperac-
tivity, without changes in pre- and postsynaptic a-adrenoceptors or
in endothelium- and non–endothelium-dependent vasodilation.
Conclusions: Intermittent hypoxia induces sequential cardiovascular
events suggesting increased chemoreflex and depressed baroreflex,
resulting in sympathoadrenal hyperactivity, early hemodynamic al-
terations with proximal histologic remodeling, and delayed changes
in peripheral vasoreactivity. Such early alterations before overt car-
diovascular disease strengthen the need for identifying at-risk in-
dividuals for systematic treatment.
Keywords: sleep apnea; blood pressure; vasoconstriction; histopathology
Obstructive sleep apnea (OSA) is characterized by repetitive
upper airway occlusions, leading to recurring oxygen desatura-
tions and sleep fragmentation (1). Patients with OSA exhibit
increased cardiovascular morbidity and mortality, including
(Received in original form February 12, 2007; accepted in final form October 24, 2007)
Supported by a grant from Association Grenobloise pour les Insuffisants
Respiratoires à Domicile.
*These authors contributed equally to this article.
**These authors were equally contributing senior authors.
Correspondence and requests for reprints should be addressed to Maurice Dematteis,
M.D., Ph.D., Laboratoire HP2, Institut Jean Roget, Faculté de Médecine de Grenoble,
38042 Grenoble Cedex 09, France. E-mail: maurice.dematteis@ujf-grenoble.fr
This article has an online supplement, which is accessible from this issue’s table of
contents at www.atsjournals.org
Am J Respir Crit Care Med Vol 177. pp 227–235, 2008
Originally Published in Press as DOI: 10.1164/rccm.200702-238OC on October 25, 2007
Internet address: www.atsjournals.org
AT A GLANCE COMMENTARY
Current Scientific Knowledge on the Subject
Intermittent hypoxia in sleep apnea is a major factor for
hypertension and impaired vasoreactivity. Understanding
the development of these outcomes may condition the
management of apneic patients at risk for cardiovascular
complications.
What This Study Adds to the Field
Intermittent hypoxia induced early functional and histo-
logic cardiovascular remodeling in mice. These results
strengthen the need for early identification and treatment
of apneic individuals at risk for cardiovascular complica-
tions.
systemic hypertension and ischemic heart and brain diseases
(2, 3). OSA is an independent risk factor for hypertension, and
several components of OSA may contribute to hypertension,
including intermittent hypoxia (IH), hypercapnia, acute nega-
tive intrathoracic pressures, and microarousals (3). The role of
blood gas alterations has been well characterized (3). Activation
of chemoreceptors by either IH or hypercapnia increases sym-
pathetic nervous system (SNS) activity, resulting in blood pres-
sure (BP) and heart rate (HR) surges after each apneic episode
(4). Chronic repetition of these acute and short nocturnal
episodes leads to increased chemoreceptor sensitivity and SNS
outflow and, with the loss of baroreflex compensation (5), in-
creases daytime SNS activity and blood pressure (3, 4). Vascular
alterations observed in OSA include impaired endothelium-
dependent peripheral vasodilation and increased vasoconstric-
tor responsiveness (6, 7). Similar to hemodynamic changes, the
fact that impaired vasoreactivity can improve with OSA treat-
ment (8, 9) strongly supports a direct relationship between OSA
and the vascular abnormalities. However, the temporal relation-
ships between hemodynamic changes and associated vascular
alterations associated with OSA are unknown. Although a bidirec-
tional relationship with reciprocal aggravation is conceivable,
their respective chronology and contribution to atherogenesis
and cardiovascular complications are still unclear. Thus, a better
understanding of this pathophysiology may condition the man-
agement of apneic patients at risk for cardiovascular complica-
tions. Although complete cardiovascular phenotyping of patients
with OSA is required to clarify this pathophysiology, clinical
investigations may be limited by unknown or confounding factors
(disease duration, chronology of events, contribution of the differ-
ent OSA components, and genetic and environmental factors).
Experimental models allow the control of some of these limiting
factors and have produced evidence for a causal relationship
228 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 177
between IH and chronic BP increase (10) or vasoreactivity al-
terations (11, 12). We previously reported an exaggerated vaso-
constriction to norepinephrine (NE), whereas the response to
acetylcholine was unaffected in mice exposed to 2 weeks of IH
(12). The mechanisms underlying these differential alterations,
as well as their relationship to hemodynamic changes and car-
diovascular tissue remodeling, remain to be elucidated. To
address these issues, we studied the chronology and relation-
ships of functional and structural cardiovascular changes in mice
exposed to long-term IH up to 35 days. We assessed the acute
and chronic changes of BP and HR; we also assessed hindquar-
ter vasoreactivity by investigating the pre- and postsynaptic
catecholaminergic transmission for vasoconstriction, and the
endothelial and nonendothelial component for vasodilation.
Finally, we analyzed histologic cardiovascular alterations at
proximal and distal levels. To our knowledge, this is the first
comprehensive study evaluating the longitudinal effects of IH on
cardiovascular functions and structural remodeling. The findings
reported in this study strengthen the need for early identifica-
tion and treatment of individuals at risk for cardiovascular
complications. Some of the results of this study have been
previously reported in the form of an abstract (13).
METHODS
Animals
C57BL/6J adult male mice (n 5 158, 8 wk old) were randomized to 1,
14 and 35 days of IH (hypoxic group [HX]: 21–4% FIO2, 30-s cycle, 8 h/d;
Days 1, 14, and 35), with respective normoxic control animals (NX
group) exposed to air in identical chambers (n 5 8–16 per group).
Procedure-related effects were assessed using a second control group
of unhandled mice (UN group). Mice were weighed before and on the
last day of each week of exposure. Additional details are provided in
the online supplement.
Hemodynamic Study
BP and HR were measured at Days 1 and 14 with a femoral artery
catheter, digitized, then analyzed using Acqknowledge software (Biopac
Systems, Inc., Santa Barbara, CA). Three periods of recording were
selected (Figure 1): (1) a baseline period corresponding to 1 hour air/air
exposure, (2) a transition period consisting of the first 15 IH cycles with
the hypoxic plateau progressively reaching 4% FIO2, and (3) a 7-hour
period of stable IH with all hypoxic plateaus at 4% FIO2.
Once the measurements were completed, blood, urine, and tissue
were collected for hematocrit, plasma-activated transforming growth
factor (TGF)-b1, urine free catecholamines, and immunohistochemistry
analysis (see the online supplement).
2008
Vascular Reactivity Assessment
Vascular reactivity was studied in isoflurane-anesthetized mice at Days
14 and 35 by measuring the perfusion pressure (Pp) from the con-
stantly perfused hindquarter. Vasoconstriction to NE (2, 10, 50 mg) was
assessed before and after inhibition of the NE transporter (uptake-1) by
desipramine infusion (0.156 mg/min/kg), and after sympathetic inacti-
vation by exsanguination. Vasodilation was assessed in exsanguinated
mice with methoxamine-precontracted vasculature (3 mg/kg/min).
Acetylcholine (5, 50, 500 ng) was injected before and after blockade of
the NO-dependent endothelium vasodilation (L-NAME, 0.2 mg/100 ml
for 5 min). Non–endothelium-dependent vasodilation was assessed using
nitroprusside (200 mg/100 ml) (see the online supplement).
Immunohistochemistry
Tissues were frozen in liquid nitrogen. Cross-sections (10-mm thick-
ness) of heart, midthoracic aorta, and hindquarter muscle were stained
for hematoxylin–eosin and sirius red to assess tissue morphology and
collagen content. Immunostaining of two vascular markers, platelet
endothelial cell adhesion molecule (PECAM)-1 and collagen IV, were
performed. Aortic intima-media and adventitia thickness were measured
with a Nikon microscope (Eclipse 80i; Nikon France SAS, Champigny
sur Marne, France) and LUCIA-G5 software (Laboratory Imaging Ltd,
Prague, Czech Republic) (see online supplement).
Data Analysis
BP was used to determine HR, arithmetic mean BP (MBP), systolic BP
(SBP), and diastolic BP (DBP). Pulse pressure was calculated by
subtracting DBP from SBP. A double analysis was performed as follows:
(1) a global evaluation by averaging the last 15-minute recordings of
baseline period and of every hour of IH exposure at Days 1 and 14 and
(2) an intracycle analysis of both transition and stable IH periods by
comparing the hemodynamic fluctuations during normoxic and hypoxic
plateaus (NxP and HxP, respectively). For the transition period, because
the HxP progressively decreased to 4% FIO2, hemodynamic response
was assessed for each of the 15 cycles. For the stable IH period, because
the HxP were all at 4% FIO2, values from NxP or HxP of the last 10 cycles
of each hour were averaged.
Results were expressed as mean 6 SEM and analyzed using Wilcoxon
and Mann-Whitney tests for intragroup and intergroup comparisons,
respectively; a P value less than 0.05 was considered significant.
RESULTS
Body Weight Evolution
Initial body weights were similar in both mouse groups (Figure
2). NX mice progressively gained weight. In contrast, weights
of HX mice initially decreased then increased parallel to NX
mice but remaining significantly lower.
Figure 1. Schematic representa-
tion of the three periods of hemo-
dynamic measurements. BP 5
blood pressure; HR 5 heart rate;
IH 5 intermittent hypoxia.
Dematteis, Julien, Guillermet, et al.: Hypoxia and Cardiovascular Changes
Figure 2. Body weight evolution during intermittent hypoxia (IH).
Body weight alteration in mice exposed to 35 days of IH (HX) or air/air
(NX), n 5 16 per group; P , 0.05 for intergroup* and intragroup†
comparisons.
BP and HR
Hemodynamic changes were assessed during three different
periods reflecting daytime period, sleep onset, and nighttime
sleep in patients with OSA, as shown in Figure 1.
Hemodynamic measurements at baseline. NX and HX ani-
mals had similar MBP and HR at Day 1, whereas at Day 14, HX
mice had increased MBP with a trend for decreased HR (P 5
0.06) compared with NX and UN mice (Figure 3). DBP
increased more (115.1 mm Hg; HX vs. NX, 122.3 6 4.12 vs.
107.1 6 5.82 mm Hg) than SBP (113.9 mm Hg; HX vs. NX,
Figure 3. Hemodynamic parameters at baseline. Mean arterial blood
pressure and heart rate at baseline period, before any exposure to
intermittent hypoxia (Day 1) and at Day 14, in hypoxic (HX; n 5 11 at
Day 1, n 5 10 at Day 14), normoxic sham (NX; n 5 9 at Days 1 and
14), and unhandled (UN; n 5 8) mice. *P , 0.05 versus NX.
229
140.1 6 4.46 vs. 126.2 6 5.45 mm Hg). No significant difference
in BP and HR emerged between NX and UN mice at Day 14.
Hemodynamic measurements during the transition period.
During this period, BP and HR were measured at NxP and
HxP of each IH cycle (C1 to C15) to assess cardiovascular
responses to progressive FIO2 decline, during acute (Day 1) and
midterm (Day 14) IH exposure (Figure 4). This allowed to
assess chemoresponses for different levels of oxygen desatura-
tion. At Day 1, BP increased (from 111.8 6 4.95 to 119.3 6
7.27 mm Hg) when FIO2 reached 8.5% and remained elevated
for three cycles, even during NxP. BP then progressively de-
creased, with oscillations driven by cyclic hypoxia from 6.7%
FIO2 corresponding to BP surges at the end of the hypoxic plateaus
(Figure 4B).
At Day 14, baseline BP was elevated and further increased
only during hypoxic phases. These hypoxic BP surges started
from 12.2% versus 8.5% FIO2 at Day 1, with maximum oscil-
lations and elevations at 8.5%. BP subsequently decreased, but
more slowly than at Day 1, thus remaining elevated at the end
of the transition period.
Figure 4. Hemodynamic parameters during the transition period.
(A) FIO2 recording from an individual chamber. (B) Mean arterial blood
pressure at Day 1 (D1; circles) and Day 14 (D14; squares) of intermittent
hypoxia (IH). (C) Heart rate (HR) at Days 1 (circles) and 14 (squares) of
IH. Dotted lines represent minimum and maximum HR at Day 1 and
Day 14. Note the larger HR fluctuations at Day 14 compared with Day
1 (long vs. short double-headed arrows between the dotted lines). For
both B and C, normoxic and hypoxic plateaus were compared; *P ,
0.05. Ten and 11 hypoxic mice were used at Days 1 and 14,
respectively. HxP 5 hypoxic plateau; NxP 5 normoxic plateau.
230 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 177
HR fluctuations were minimal at Days 1 and 14 until reaching
8.5% FIO2, then increased with greater amplitude at Day 14
versus Day 1, independently from the normoxic or hypoxic
phases (Figure 4C).
Hemodynamic measurements during 7 consecutive hours of
IH. At Day 1, BP progressively decreased throughout the 7
hours in both HX and NX mice (Figure 5A). Motor activity also
decreased over the first 30 minutes of IH, after which the
animals remained mainly asleep. This behavioral pattern per-
sisted at Day 14.
In contrast to Day 1, Day 14 HX mice had higher baseline
BP with a steeper decrease throughout the 7 hours of assess-
ment (Figures 5A and 5B). BP elevation in HX mice was
predominantly diastolic, and decreased more rapidly in HX
2008
mice, as substantiated by slope calculations (HX vs. NX: SBP,
21.71 6 0.78 vs. 20.88 6 0.54 mm Hg/h; DBP, 22.07 6 0.92 vs.
21.45 6 0.47 mm Hg/h) (Figures 5C and 5D).
Pulse pressure was lower in HX mice at Day 14 (HX vs. NX:
15.2 6 2.90 vs. 22.6 6 2.34 mm Hg; P 5 0.072), and progressively
increased throughout the 7 hours in both groups to reach 19.5 6
4.60 and 25.9 6 2.31 mm Hg for HX and NX mice, respectively.
In contrast, no difference in HR emerged between HX and NX.
To assess baroresponses, we measured hemodynamic fluctu-
ations during NxP and HxP at each hour of the 7-hour recording
period. We observed similar amplitude of hypoxic BP surges at
Days 1 and 14 (Figures 5E and 5F). However, BP oscillations at
Day 14 occurred with reduced HR variability compared with
Day 1.
Figure 5. Hemodynamic parameters
during stable intermittent hypoxia
(IH). Mean arterial blood pressure
(BP) at Days 1 (A) and 14 (B) in
hypoxic (HX) and normoxic (NX)
mice. Systolic (C) and diastolic
(D) BP at Day 14 in HX and NX mice.
HX versus NX mice, *P , 0.05. BP
and heart rate (HR) fluctuations dur-
ing hypoxic versus normoxic plateaus
at Days 1 (E) and 14 (F). Data are
expressed in percentage of normoxic
plateau value for each hour of the 7
hours of measurement, and statistical
analysis was performed on raw data
from hypoxic versus normoxic pla-
teaus; *P , 0.05. Ten HX and eight
NX mice were used at Day 1; 11 HX
and 6 NX mice were used at Day 14.
B 5 baseline period; T 5 transition
period.
Dematteis, Julien, Guillermet, et al.: Hypoxia and Cardiovascular Changes
Biochemistry
Hematocrit, TGF-b1, and catecholamines were determined at
the end of the 8-hour exposure at Days 1 and 14 (Table 1). At
Day 14, HX animals had higher hematocrit than NX mice. No
significant difference in TGF-b1 appeared between HX and NX
animals at Days 1 and 14.
Free dopamine and NE were higher in some HX mice at Day
1, whereas epinephrine was elevated in some HX mice at Day
14. This was confirmed by the catecholamine ratios, reflecting
a delayed increase in epinephrine versus NE, and prominence
of the two vasopressive catecholamines over the dopaminergic
system at Day 14.
Vascular Reactivity
Vasoconstriction. NE induced similar dose-dependent responses
in Day 14 HX and Day 14 NX mice. In contrast, Day 35 HX
mice had higher baseline Pp (P 5 0.07) and enhanced NE
response compared with NX mice (Figure 6A and Figure E1 of
the online supplement). This increased responsiveness was
significant from 10 mg NE and reached a plateau for the two
highest doses. Contribution of NE clearance to this exaggerated
response was assessed with the NE reuptake inhibitor desipra-
mine (Figure 6B and Figure E1). Desipramine and exsanguina-
tion similarly lowered baseline Pp and the vasoconstriction
response to NE at all doses, without a significant difference
between NX and HX groups at Days 14 and 35 (Figures 6B and
6C, and Figure E1). Vehicle injection led to similar Pp than
baseline values, and no difference between NX and UN animals
emerged.
Vasodilation. During methoxamine precontraction, baseline
Pp and response to acetylcholine was similar in HX mice and in
their respective control group (Figures 6D and 6E, and Figure
E2). NO synthesis inhibition by L-NAME produced similar
elevation in baseline Pp and response to acetylcholine in HX
and NX mice, suggesting no alterations in endothelium-depen-
dent vasodilation (Figure 6E and Figure E2). Assessment of
endothelium-independent vasodilation with nitroprusside did
not reveal any Pp difference between HX and NX animals
(Figure 6E and Figure E2). Vehicle injection led to similar Pp
than baseline values regardless of the group and time point.
Histologic Cardiovascular Remodeling
HX mice exposed to 14 days of IH exhibited decreased
PECAM-1 immunostaining at the aortic and cardiac levels,
whereas no difference emerged for hindquarter skeletal muscles
TABLE 1. BIOCHEMICAL ALTERATIONS INDUCED BY INTERMITTENT HYPOXIA
Day 1
Hypoxic Mice Normoxic Mice
Blood
Hematocrit, % 41.3 6 1.90 (n 5 8) 41.0 6 2.38 (n 5 6)
TGF-bl, ng/ml 6.88 6 0.91 (n 5 4) 5.43 6 0.67 (n 5 5)
Urine
Epinephrine, nmol/L 40.3 6 15.2 (n 5 7) 34.2
Norepinephrine, nmol/L 154 6 46.2 (n 5 7) 104
Dopamine, nmol/L 1,063 6 351 (n 5 7) 341
E/NE ratio 0.68 6 0.22 (n 5 7) 0.42
(NE 1 E)/DA ratio 0.33 6 0.14 (n 5 7) 0.66
Definition of abbreviations: DA 5 dopamine; E 5 epinephrine; NE 5 norepinephrine; TGF-b1 5 transforming growth factor-b1.
Hematocrit, TGF-b1, and urinary catecholamines measured after 8 hours of intermittent hypoxia (hypoxic mice) or air/air
(normoxic mice) exposure at Days 1 and 14.
* Comparison between hypoxic and normoxic mice, P , 0.05.
†
Comparison between hypoxic mice at Days 1 and 14, P , 0.05.
231
at 14 and 35 days of IH. At the aortic level, the decreased
endothelial PECAM-1 expression predominated at the dorsal
wall of intima, without evidence of endothelial denudation
(Figure 7A). Similar reduction in staining occurred in the left
ventricular myocardium, with a gradient from the pericardium
to the endocardium, and almost no staining in the deepest layers
of the ventricle of some animals, especially in the papillary
muscles (Figure 7B). Comparatively, the right ventricle was less
affected or not affected. Collagen IV staining showed no dif-
ference in heart collagen density between HX and NX animals
at Day 14, suggesting no difference in capillary density. Re-
garding intima-media thickness, although HX mice tended to
have higher values at Day 14, implicating the media, no signif-
icant difference emerged, as for adventitia thickness (Figure 7C).
DISCUSSION
To clarify the pathophysiology of OSA-related cardiovascular
complications, we studied the chronology of functional and
structural cardiovascular changes and their relationship in mice
exposed to long-term IH. Hemodynamic alterations occurred
first at Day 1, as BP and rhythmic oscillations driven by IH
cycles. After 2 weeks of IH, BP was elevated at normoxia with
predominant diastolic increase. With hypoxia, sympathoadrenal
activity was enhanced; BP surged with lower oxygen desatura-
tion, whereas HR fluctuations decreased. Proximal histologic
cardiovascular remodeling appeared at Day 14, with decreased
endothelial PECAM-1 expression in the descending aorta and
left heart. Distal functional vascular remodeling became obvi-
ous at Day 35, characterized by increased vasoconstriction to
NE, with evidence of sympathetic hyperactivity without changes
in pre- and postsynaptic a-adrenoceptor vasoresponse, or in endo-
thelium- and non–endothelium-dependent vasodilation.
Methodologic Considerations
Considering their high breathing frequency and metabolic rate,
we exposed mice to 30-second IH cycles (i.e., 120/h of apnea
index), which compares with severe human OSA. Moreover, this
short cycle duration limited the hypoxia-induced hyperventila-
tion that may induce hypocapnia and cardiovascular changes. We
selected a 4% FIO2 nadir based on our experience and previous
studies showing that FIO2 of less than 5% produced maximal
changes in BP and HR (14). We did not record sleep to show
sleep disruption, which may have affected our results. We
exposed mice to IH during daytime, which is the preferential
sleep period in rodents. On the basis of behavioral assessment,
Day 14
Hypoxic Mice Normoxic Mice
46.3 6 1.20 (n 5 10)*† 38.7 6 1.29 (n 5 10)
4.98 6 0.34 (n 5 10) 5.27 6 0.76 (n 5 12)
6 13.2 (n 5 6) 549 6 253 (n 5 7) 164 6 93.9 (n 5 9)
6 29.7 (n 5 6) 269 6 105 (n 5 7) 145 6 53.5 (n 5 9)
6 207 (n 5 6) 684 6 246 (n 5 7) 898 6 297 (n 5 9)
6 0.12 (n 5 6) 3.20 6 1.06 (n 5 7)*† 0.77 6 0.29 (n 5 8)
6 0.15 (n 5 6) 1.35 6 0.15 (n 5 7)*† 0.51 6 0.10 (n 5 8)
232 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 177
our mice spent most of the time sleeping during IH exposure,
although microarousals could not be ruled out. However, rodents
have polyphasic sleep; they do not sleep for 8 consecutive hours.
They also sleep during nighttime, and are able to compensate for
daytime sleep disruption by shifting their preferential sleep
period to nighttime. Overall, this suggests that our results were
mainly due to hypoxia.
Hemodynamic Changes
Baseline period. The baseline period is representative of the
daytime period in patients with OSA. After 2 weeks of IH, mice
had elevated BP during normoxia. This finding provides new
evidence for a causal relationship between IH and the rapid
development of systemic hypertension independently of con-
founding factors. The magnitude of BP increase, with its pre-
dominant diastolic component, is in agreement with human OSA
pathology (2, 3, 15). The pathophysiology may involve recurrent
sympathetic stimulations through carotid and aortic chemore-
ceptor activation at each oxygen desaturation, leading to chronic
SNS hyperactivity and increased circulating catecholamines (3,
4). Urinary catecholamines, reflecting integrative sympathoa-
drenergic activity during IH exposure, uncovered a delayed in-
crease in epinephrine. Although NE preferentially reflects SNS
activation, epinephrine mainly characterizes adrenal involvement
(16), suggesting differential regulation of these two components
in response to acute and chronic hypoxia (17). The vasoreactivity
study suggested SNS hyperactivity (see below). Together, these
changes should result in increased BP, whereas HR should
2008
Figure 6. Vasoreactivity after intermittent hypoxia (IH).
Vasoconstriction (A–D). Vascular responses to norepineph-
rine (NE) in mice exposed to 14 and 35 days of IH (HX14,
n 5 8; HX35, n 5 7) or air (NX14, n 5 7; NX35, n 5 8).
(A) Before NE-reuptake blockade and hemorrhage-induced
sympathetic inhibition; (B) with desipramine inhibition of
NE reuptake; (C) with desipramine NE-reuptake blockade
and sympathetic inactivation by exsanguination; (D) vas-
cular response to methoxamine. *P , 0.05. See also Figure
E1 of the online supplement. Vasodilation (E) in mice ex-
posed to 14 and 35 days of IH (HX14, n 5 8; HX35, n 5 7)
or air (NX14, n 5 7; NX35, n 5 7). Mice were exsangui-
nated to inhibit sympathetic activity. Vascular response to
acetylcholine (Ach; 500 ng) with active nitric oxide (NO)
synthesis (left), and with inactive NO synthesis (L-NAME
L-nitro-arginine methyl ester) (middle). Note that three
doses of Ach were tested (5, 50, 500 ng), but only the
highest dose is shown in the figure. Non–endothelium-
dependent vasodilation with the NO donor sodium nitro-
prusside (200 mg) (right). See also Figure E2. HX14 5 Day
14 hypoxia; HX35 5 Day 35 hypoxia; NX14 5 Day 14
normoxia; NX35 5 Day 35 normoxia.
decrease due to vagally mediated inhibition of baroreceptors.
However, we found evidence of reduced baroreflex (see below).
Alternatively, the decreased HR observed in this study may
result from changes in blood volume and redistribution. Venous
blood return can be increased in OSA, due to peripheral vasocon-
striction and augmented renal sodium reabsorption secondary to
renin–angiotensin system activation and negative intrathoracic
pressure (3). These changes could lead to intermittent change
in ventricular pre- and afterload, diastolic dysfunction, and
enhanced DBP (3).
Transition period. The transition period represented the
apneic episodes at sleep onset and allowed assessment of
cardiovascular responses during progressive oxygen desatura-
tion. After 2 weeks of IH, BP surges occurred at a less severe
hypoxia, with slower return to baseline, and greater BP and HR
instability than at Day 1. Such alterations may underlie arrhyth-
mic complications associated with OSA (15) and may involve
chemoreceptor modifications. Carotid bodies are the major
peripheral sensors for detecting oxygen decline in arterial blood,
and chronic IH may lead to increased carotid body sensitivity and
response to hypoxia through long-term facilitation (18).
Stable IH period. Representative of the human nighttime
period, the stable IH period was characterized by a progressive
decline in BP over the 7 hours in all groups; however, this was
more pronounced in the chronic IH animals. Such daytime fall in
BP is typical in rodents and may relate to their reduced motor
activity during sleep (19). In addition, sleep regulation itself may
contribute to this decreased BP through changes in the auto-
nomic balance (increased parasympathetic vs. decreased sympa-
Dematteis, Julien, Guillermet, et al.: Hypoxia and Cardiovascular Changes 233

Figure 7. Histologic cardiovascular alterations induced by 14 days of intermittent hypoxia. (A) Platelet endothelial cell adhesion molecule (PECAM)-1
immunostaining of aorta cross-sections (10 3 10 magnification; insets, 10 3 40 magnification) decreased in hypoxic (HX) versus normoxic (NX) mice,
without endothelial denudation as suggested by the presence of endothelial cell nuclei (10 3 40 magnification, arrows). (B) PECAM-1 immunostaining
of heart cross-sections (10 3 10 magnification) decreased in HX versus NX mice, with a gradient from the periphery to the center of the section (dotted
triangle). Collagen IV and PECAM-1 immunostainings of heart cross-sections are shown in 10 3 40 magnification insets; only PECAM-1 expression decreased
in HX mice (C) Sirius red staining of aorta cross-sections. Left inset shows measurements of the wall thickness (double-headed arrows). Right inset shows
that intima media thickness in HX mice was not significantly different from NX mice. Four to five mice per group were used for immunostainings.

thetic activity) (3). However, despite their greater BP decrease,
BP was still elevated in Day 14 HX mice compared with Day 14
NX mice. Thus, similarly to apneic patients, hypoxic mice
suffered from enhanced BP throughout the nycthemeral cycle.
In addition, the sleep-related BP fall can disappear in patients
with OSA (15, 20). This nondipping pattern seems to be highly
predictive of cardiovascular complications even without daytime
hypertension (3). In contrast, Day 14 HX mice exhibited a greater
BP decrease. Beyond a species-related difference, one explana-
tion may rely on dopamine. Urinary free dopamine stems from
a local renal dopaminergic–natriuretic system, derived from
circulating dopa and acting as an autocrine/paracrine substance
(16) that may participate in the hypoxic diuretic response (21).
HX mice at Day 1 had the highest dopamine levels and the lowest
(norepinephrine 1 epinephrine)/dopamine ratio, illustrating a
predominant secretion of dopamine. In contrast, Day 14 HX
mice were not different from NX mice for dopamine, and their
catecholamine ratio was inverted compared with Day 1, reflect-
ing the predominant secretion of the two vasopressive catechol-
amines. Alternatively, hypoxic vasodilation, involving the aden-
osine–prostaglandin–NO cascade (22), may account for the
greater DBP decline. Indeed, adenosine accumulates during
hypoxia resulting from ATP degradation. However, such com-
pensatory vasodilation is transient, gradually decreasing over
days, and is replaced with neoangiogenesis to alleviate tissue
hypoxia. In that case, the nondipping pattern could occur at
a later stage of IH exposure.
Assessment of the hemodynamic variability throughout 7
hours showed similar BP fluctuations after acute and chronic
IH, but with reduced HR variability in chronic IH mice.
Although the hemodynamic alterations concomitant with the
FIO2 decline of the transition period may relate to increased
chemosensitivity, BP oscillations with reduced HR fluctuations
suggest depressed baroreflex, due to either decreased sensitivity
or increased set point (3, 5).
Functional Vascular Remodeling
The enhanced vasoconstrictor response to NE with normal
endothelium- and non–endothelium-dependent vasodilation agrees
with previous studies (7, 12), although others have demonstrated
impaired vasodilation (8, 9, 11). The discrepancies may origi-
nate from our experimental procedure. We assessed in situ vas-
cular response of the entire vessel network with its surrounding
regulation. This setting, as well as the effect of isoflurane anes-
thesia, may have masked the endothelial dysfunction previously
detected in isolated artery (11). Furthermore, the characteristics
of the hypoxic stimulus may also affect the result. In our previ-
ous study, the same IH paradigm in a noisier apparatus led to
a similar but earlier increased vasoconstriction to NE (12), sug-
gesting that IH with an additional stress factor may be more
detrimental. In contrast, different hypoxic stimuli may induce
an opposite response, as chronic sustained hypoxia in rats has
led to decreased vasoconstrictor responses (23). The delayed
changes in vasoreactivity suggest a phenomenon secondary to
234 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 177
both hypoxia and hemodynamic alterations. The exaggerated re-
sponse to NE may be due to the following: (1) an increased NE
release and or (2) a decreased NE clearance and/or (3) post-
synaptic alterations.
Increased NE release, a corollary of SNS hyperactivity/
hyperreactivity, is well established in OSA (3, 4) and agrees with
our results suggesting increased chemoreflex. Higher catechol-
amine levels may result from increased release and/or decreased
clearance.
NE reuptake via the NE transporter uptake-1 is the main
mechanism for NE clearance (16). Such an energy-dependent
mechanism can be affected by hypoxia and oxidative stress. We
used desipramine to inhibit the NE reuptake leading to NE accu-
mulation in the synaptic cleft. This resulted in presynaptic a2-
receptor–mediated inhibition of SNS activity with decreased NE
release and subsequent decreased baseline Pp. Desipramine in-
duced similar decreased baseline Pp in HX and NX mice and
lowered pressor responses to NE at all concentrations without
differences between groups. This suggests that NE-reuptake
inhibition was similar between groups and that presynaptic a2-
receptors were likely not involved in the differential NE
response of Day 35 HX mice. Exsanguination did not further en-
hance sympathetic inhibition, suggesting that the desipramine–
sympathoinhibitory effect was maximum.
Postsynaptic alterations may result from altered number
and/or affinity of postsynaptic a-adrenoceptors. However, HX
and NX mice exhibited similar responses to NE after SNS
suppression, or after administration of the a1-agonist methox-
amine. Overall, the hyperresponsiveness to NE appeared to
result from SNS hyperactivity.
Histologic Cardiovascular Remodeling
Proximal cardiovascular remodeling occurred from Day 14. We
used two vascular markers, PECAM-1 for endothelial cells and
collagen IV for perivascular basement membranes, but only
PECAM-1 was modified. Affected areas included the descending
aorta with no evidence of endothelium denudation, and the left
ventricle, predominating in the deepest layers without decreased
capillary density, whereas the right ventricle was less affected.
This topographic distribution pattern of decreased PECAM-1
expression may result from hemodynamic strains induced by IH.
PECAM-1 is a cell–cell adhesion molecule most abundantly
expressed in endothelial cells and previously believed to mediate
cell adhesion, neogenesis, and leukocyte transmigration through
the endothelium (24). Recent evidence implicates PECAM-1 as
a mechanoreceptive molecule (25). Indeed, shear stress triggers
a rapid tyrosine phosphorylation of PECAM-1, initiating a sig-
naling cascade leading to eNOS activation (26). Chronic expo-
sure to biomechanical forces may alter PECAM-1 expression and
distribution, which in turn may represent an important mecha-
nism modulating endothelial cell sensitivity to mechanical stimuli
(25). Shear stress is indeed a critical determinant of cardiovascular
homeostasis, regulating remodeling and atherogenesis. Although
the vasoreactivity experiment showed no evidence of peripheral
endothelial dysfunction, PECAM-1 alterations implicate the
aortic and cardiac endothelium in IH pathophysiology. Whether
down-regulated endothelial PECAM-1 expression in our HX
mice represents a marker of endothelial dysfunction or an early
event of atherogenesis (27, 28) remains to be determined.
Large-artery stiffness increases pulse pressure, which is asso-
ciated with several cardiovascular risk factors (29). However,
we observed a decreased pulse pressure resulting from the prom-
inent DBP increase. Increase in pulse pressure may occur later,
following appearance of structural remodeling leading to arte-
rial stiffness.
2008
We assessed TGF-b1 as an early marker of tissue remodel-
ing. TGF-b1 is a potent antiinflammatory cytokine involved in
angiotensin-mediated tissue remodeling, promoting collagen
expression and fibrosis (30). In agreement with clinical studies
(31), no difference in TGF-b1 emerged between HX and NX
mice. However, normal plasma TGF-b1 cannot exclude local-
ized tissue TGF-b1 alterations.
Cardiovascular Alterations Induced by
IH: Pathophysiologic Hypothesis
Collectively, our results suggest the following pathophysiology
that may occur in OSA. IH activation of chemoreceptors leads
to increased SNS tone resulting in BP surges with subsequent
baroreflex-mediated bradyarrhythmia. Proximal cardiovascular
remodeling as evidenced by decreased PECAM-1 staining
occurs early, whereas peripheral functional vascular remodeling
as shown by the increased vasoconstriction to NE is delayed.
While the proximal PECAM-1 changes suggest a prominent
effect of hemodynamic strains, the respective contribution of IH
and hemodynamic alterations to the delayed hypervasoconstric-
tion remains to be determined. These alterations, in turn, may
contribute to persistent hemodynamic changes. In addition,
increased BP becomes permanent because of enhanced chemo-
receptor sensitivity and sympathoadrenal activity, whereas baro-
receptor response is decreased. Peripheral vascular remodeling
may further become structural, eventually leading to athero-
genesis, and resulting in chronic tissue hypoxia; together with
impaired coagulation and augmented hematocrit levels, this
may increase risks for acute ischemic and arrhythmic compli-
cations (3). These findings support the hypothesis that IH plays
an independent role in cardiovascular consequences associated
with OSA. Occurrence of these early alterations before overt
cardiovascular disease (loss of sleep-dipping pattern, develop-
ment of daytime hypertension or increased wall vessel thick-
ness) corroborates with clinical findings (32), and strengthens
the need for early identification of at-risk individuals for sys-
tematic treatment.
Conflict of Interest Statement: None of the authors has a financial relationship
with a commercial entity that has an interest in the subject of this manuscript.
Acknowledgment: The authors thank Dr. S. Bottari for catecholamine determination.
References
1. Malhotra A, White DP. Obstructive sleep apnoea. Lancet 2002;360:
237–245.
2. Peppard PE, Young T, Palta M, Skatrud J. Prospective study of the
association between sleep-disordered breathing and hypertension.
N Engl J Med 2000;342:1378–1384.
3. Leung RS, Bradley TD. Sleep apnea and cardiovascular disease. Am J
Respir Crit Care Med 2001;164:2147–2165.
4. Somers VK, Dyken ME, Clary MP, Abboud FM. Sympathetic neural
mechanisms in obstructive sleep apnea. J Clin Invest 1995;96:1897–
1904.
5. Carlson JT, Hedner JA, Sellgren J, Elam M, Wallin BG. Depressed
baroreflex sensitivity in patients with obstructive sleep apnea. Am J
Respir Crit Care Med 1996;154:1490–1496.
6. Kato M, Roberts-Thomson P, Phillips BG, Haynes WG, Winnicki M,
Accurso V, Somers VK. Impairment of endothelium-dependent vaso-
dilation of resistance vessels in patients with obstructive sleep apnea.
Circulation 2000;102:2607–2610.
7. Kraiczi H, Hedner J, Peker Y, Carlson J. Increased vasoconstrictor
sensitivity in obstructive sleep apnea. J Appl Physiol 2000;89:493–498.
8. Kraiczi H, Caidahl K, Samuelsson A, Peker Y, Hedner J. Impairment of
vascular endothelial function and left ventricular filling: association
with the severity of apnea-induced hypoxemia during sleep. Chest
2001;119:1085–1091.
9. Imadojemu VA, Gleeson K, Quraishi SA, Kunselman AR, Sinoway LI,
Leuenberger UA. Impaired vasodilator responses in obstructive sleep