Revie
on
Mi
290
and clinical features, and to provide guidelines for diag-
studies emphasized that the exclusion of all other causes
A fundamental question regarding pathogenesis in
in the lung are similar to those involved in the system-
ic and splanchnic alterations in the hyperdynamic cir-
HPS has also been recognized in patients with
portal hypertension in the absence of cirrhosis (portal
vein thrombosis, nodular regenerative hyperplasia,
iver. Published by Elsevier B.V. All rights reserved.
Available online 25 July 2006
* Corresponding author. Tel.: +1 205 957 9698; fax: +1 205 975
9777.
E-mail address: mfallon@uab.edu (M.B. Fallon).
Journal of Hepatology 45 (
0168-8278/$32.00 � 2006 European Association for the Study of the L
2. Definition
HPS is classically defined by a widened alveolar-arte-
rial oxygen gradient (AaPO2) on room air (>15 mmHg,
or >20 mmHg in patients >64 years of age) with or with-
out hypoxemia resulting from intrapulmonary vasodila-
culatory state of cirrhosis. HPS is found most
commonly in the setting of cirrhosis and appears to
occur across the spectrum of etiologies of liver disease
[5,11,12]. However, whether the presence or severity of
intrapulmonary vasodilatation and HPS correlate with
the severity of underlying liver disease is controversial
and studies have found HPS more commonly in both
less and more advanced cirrhosis [5,11–14]. Recently,
nosis and treatment. HPS is whether the mechanisms of vascular alterations
1. Background
Respiratory symptoms are exceedingly common in
patients who have chronic liver disease with estimates
ranging as high as 50–70% of patients complaining of
shortness of breath [1]. The differential diagnosis of
dyspnea is extensive in such patients and numerous
causes should be considered. Over the last 15 years, pul-
monary vascular abnormalities have been increasingly
recognized as important clinical entities that influence
survival and liver transplant candidacy in affected
patients. The most common such abnormality, the hep-
atopulmonary syndrome (HPS), occurs when intrapul-
monary vasodilatation impairs arterial oxygenation.
This syndrome is now recognized as many as 15–20%
of patients undergoing evaluation for orthotopic liver
transplantation (OLT) [2]. The presence of HPS increas-
es mortality in the setting of cirrhosis and may influence
the frequency and severity of complications of portal
hypertension [3]. The purpose of this review is to pro-
vide an update on HPS, including the pathophysiology
The hepatopulm
David T. Palma,
University of Alabama at Birmingham Liver Center, MCLM
doi:10.1016/j.jhep.2006.07.002
of cardiopulmonary dysfunction was required to estab-
lish the diagnosis of HPS [7]. However, it is now evident
that HPS may coexist with other cardiopulmonary
abnormalities [8,9] and contribute significantly to gas
exchange abnormalities in this setting. In addition, the
AaPO2 normally increases with age and varies signifi-
cantly even in healthy adults. Therefore, using values
above the 95% confidence interval for the age-corrected
AaPO2 is appropriate to avoid over-diagnosis of HPS
[10].
3. Pathophysiology
3.1. Mechanisms in humans
w
ary syndrome
chael B. Fallon*
, 1918, University Boulevard, Birmingham, AL 35294, USA
tation in the presence of hepatic dysfunction or portal
hypertension [4–7]. From a practical vantage point,
identifying patients with PaO2 <70 mmHg is useful for
recognizing those with clinically significant HPS. Early
www.elsevier.com/locate/jhep
2006) 617–625
congenital hepatic fibrosis and Budd–Chiari syn-
drome) [15–18] and has been reported in the setting
of acute and chronic hepatitis in the absence of portal
hypertension [19,20]. These findings support that
advanced liver disease is not required for HPS to
develop and that the pathophysiologic events occur-
ring in the pulmonary microvasculature of patients
with HPS are unique relative to the systemic and
splanchnic circulations.
The pathogenetic hallmark of HPS is microvascular
dilatation within the pulmonary arterial circulation.
These changes may result from decreased pre-capillary
arteriolar tone alone or could involve additional
mechanisms such as angiogenesis, remodelling, and
vasculogenesis, which have been recently suggested
[21]. In human HPS, the vasodilatation is assumed
to result from excessive vascular production of vasodi-
lators, particularly nitric oxide (NO). This has been
based on the observation that exhaled NO levels, a
measure of pulmonary production, are increased in
cirrhotic patients with HPS and normalize after OLT
[22–24], as HPS resolves. In addition, a case report
revealed that acute inhibition of NO production or
action with NG-nitro-L-arginine methyl ester (L-
NAME) or cyclic GMP inhibitor methylene blue,
respectively, transiently improves HPS [25–27].
However, a recent study found that administration
of inhaled L-NAME did not acutely improve intrapul-
monary vasodilatation [21], raising the possibility that
factors other than NOS-derived NO effects on vascu-
lar tone contribute to HPS.
The exact mechanisms of increased endogenous NO
production and its relationship to the presence of portal
hypertension, the hyperdynamic circulation and the
degree of liver injury remain uncertain. In addition,
whether other mediators such as heme oxygenase-
derived carbon monoxide [28] might contribute to
intrapulmonary vasodilatation and explain the lack of
improvement of HPS with NO inhibition in some
patients has not yet been established.
3.2. Mechanisms in experimental HPS
Chronic common bile duct ligation (CBDL) in the
rat is the only established model that reproduces the
physiologic features of human HPS [29,30] (Fig. 1). It
is unique among rodent models of cirrhosis and/or
portal hypertension in that other commonly used mod-
els such as thioacetamide-induced cirrhosis and partial
portal vein ligation do not result in the development of
HPS [31]. Early studies in CBDL animals focused on
the vasoconstrictor role of eicosanoids and on an
618 D.T. Palma, M.B. Fallon / Journal of Hepatology 45 (2006) 617–625
Fig. 1. Potential mechanisms and treatments in
experimental HPS (see text for details).
and result from a gravitational increase in blood flow
through dilated vessels in the lung bases [47]. While
history and physical examination. Such an evaluation
may lead the clinician to consider alternate, more com-
al of
increase in intravascular macrophage-like cells [32,33].
Subsequent work identified increased pulmonary vascu-
lar endothelial nitric oxide synthase (eNOS) as a major
source of pulmonary NO production [34–36] and dem-
onstrated that the administration of intravenous L-
NAME improved hypoxemia after CBDL [37].
Further studies have revealed that increased hepatic
production of endothelin-1 (ET-1) with release into
the circulation is an important mechanism for trigger-
ing the increase in pulmonary eNOS and the onset of
vasodilatation after CBDL [35,38]. This effect may be
driven by a shear stress mediated increase in pulmon-
ary vascular endothelial endothelin B (ETB) receptor
expression which enhances endothelial NO production
by ET-1 [39]. Accordingly, administration of a selective
ETB receptor antagonist to CBDL animals decreases
pulmonary endothelial eNOS and ETB receptor levels
and significantly improves HPS [40]. Recent data sup-
port that biliary epithelium is an important source of
hepatic ET-1 production after CBDL and may explain
the unique susceptibility of CBDL animals to HPS
[41,42].
As experimental HPS progresses, there is a steady
accumulation of intravascular macrophages. These cells
transiently produce inducible nitric oxide synthase
(iNOS) [36,37] and progressively produce heme oxygen-
ase 1 (HO-1) [36,43]. These events contribute to further
vasodilatation through production of iNOS-derived
NO and HO-1-derived carbon monoxide (CO).
Accordingly, HO inhibition improves experimental
HPS. In addition, prolonged treatment of CBDL ani-
mals beginning at the time of ligation with norfloxacin
to inhibit bacterial translocation and tumor necrosis
factor-alpha (TNF-a) production decreases macro-
phage accumulation and prevents the transient increase
in iNOS [44], supporting that TNF-a contributes to
macrophage accumulation. Further, pentoxifylline, a
non-specific phosphodiesterase inhibitor that increases
intracellular cAMP levels and also inhibits TNF-a pro-
duction in macrophages [45], given over a similar time
frame can prevent the onset or decrease the severity of
HPS [46]. Both of these agents initiated at the onset of
liver injury influence the development of the hyperdy-
namic state and may modify ETB receptor expression
and endothelin related signalling events in the pulmon-
ary microvasculature.
Findings to date in the CBDL model suggest that a
sequence of events related in part to increased vascu-
lar shear stress and to hepatic ET-1 production may
trigger the onset of experimental HPS. The observa-
tion that hepatic and plasma ET-1 levels increase
within 1 week after CBDL [42] suggests that hepatic
ET-1 production and release may occur with relatively
modest degrees of bile duct proliferation. The finding
that macrophages accumulate in the pulmonary micro-
D.T. Palma, M.B. Fallon / Journ
vasculature and may be influenced by TNF-a inhibi-
mon diagnoses such as COPD, CHF or myocardial
ischemia. However, if the common causes of dyspnea
can be excluded, and particularly if platypnea or digital
clubbing is present, further evaluation for HPS is
orthodeoxia has been observed in a variety of condi-
tions, including post-pneumonectomy, recurrent pul-
monary thromboemboli, and atrial septal defects
(such as patent foramen ovale), it is highly specific
for HPS in the setting of liver disease [48]. The sen-
sitivity of orthodeoxia for HPS is relatively low, but
increases in cases of severe HPS [49,50]. Cough is not
a common finding in HPS. Spider angiomata are
commonly reported in HPS but are frequently seen
in cirrhotic patients without HPS. One study
observed that patients with these cutaneous lesions
had more pulmonary vasodilatation and higher alve-
olar-arterial oxygen gradients than those without vas-
cular spiders (AaPO2: 20 mmHg versus 8 mmHg) [51].
Finally, clubbing and distal cyanosis, when present in
the setting of liver disease or portal hypertension,
should raise suspicion for HPS [2].
5. Diagnosis
The diagnostic features of HPS include evidence of
liver disease or portal hypertension, an elevated age-ad-
justed alveolar-arterial oxygen gradient (AaPO2), and
evidence of intrapulmonary vasodilatation. In the pres-
ence of coexisting cardiac or pulmonary disease, estab-
lishing a diagnosis of HPS can be difficult. Fig. 2
presents an algorithm for the diagnosis of HPS. A logi-
cal evaluation of dyspnea in the patient with liver dis-
ease or portal hypertension begins with a careful
tion supports that these cells may also contribute to
vasodilatation. Fig. 1 includes potential therapeutic
targets for treatment in HPS based on experimental
data.
4. Clinical manifestations
The clinical features of HPS typically involve
respiratory complaints and findings associated with
chronic liver disease. The insidious onset of dyspnea,
particularly on exertion, is the most common com-
plaint but is non-specific. Platypnea (shortness of
breath exacerbated by sitting up and improved by
lying supine) and orthodeoxia (hypoxemia exacerbat-
ed in the upright position) are classically described
Hepatology 45 (2006) 617–625 619
warranted.
iver
al Hy
Dy
w pul
y, Exa
tests
FT's,
shun
art b
rdiac
ails).
620 D.T. Palma, M.B. Fallon / Journal of Hepatology 45 (2006) 617–625
5.1. Assessment of arterial oxygenation
In patients with liver disease found to have dyspnea
L
Port
Lo
Histor
Other
(P
Elevated A-a gradient
Hypoxemia
Low suspicion of intrinsic cardiopulmonary disease
Contrast echocardiogram
Normal Early
(< 3 he
Delayed shunting
(>3 heart beats)
No HPS HPS Intraca
Fig. 2. Diagnosis of HPS (see text for det
or clubbing, or in those undergoing transplant evalua-
tion, pulse oximetry is a simple, non-invasive screening
test for hypoxemia and a decreased SpO2 should lead
to arterial blood gas (ABG) analysis. However, caution
must be exercised in interpreting a ‘‘normal’’ SpO2 as
pulse oximetry may overestimate SaO2 in nearly one-
half of patients with cirrhosis [52]. Therefore, to reliably
detect hypoxemia ABG analysis should be considered
when the SpO2 values are 97% or less. In addition, if
hypoxemia or HPS is strongly suspected based on histo-
ry and physical exam, arterial blood gas analysis should
be performed while breathing room air regardless of
pulse oximetry. In HPS, ABGs reveal an elevated
age-adjusted AaPO2 with or without hypoxemia. The
expected upper limit of normal for room-air AaPO2 at
a given age (>95% confidence interval) can be calculated
using the following equation: AaPO2 = [0.26
age � 0.43] + 10 [10].
If gas exchange abnormalities are detected, chest
radiography and pulmonary function tests are per-
formed to evaluate for the presence of other pulmonary
abnormalities. Since cardiopulmonary disorders
unrelated to liver disease or those related to ascites
are more common than HPS, treating these abnormal-
ities prior to further evaluation for HPS is reasonable
in the absence of significant hypoxemia (PaO2
<70 mmHg).
The ERS Task Force has proposed a classification
system that uses arterial oxygen tension (PaO2) to stage
the severity of HPS. According to this system, a
Disease
pertension
spnea
se oximetry
m, CXR, ABG,
as appropriate
Chest CT)
Suspect intrinsic cardiopulmonary disease
Treat as appropriate
Symptoms persist
Contrast echocardiogramting
eats)
MAA if delayed shunting
shunt
[This figure appears in colour on the web.]
PaO2 <50 mmHg indicates very severe HPS, a
PaO2P50 and <60 mmHg suggests severe HPS, and a
PaO2P60 and <80 mmHg corresponds with moderate
HPS [4]. Staging the severity of HPS is important as a
means of predicting survival [13,53], and determining
the timing and risks of orthotopic liver transplantation
[3,6,53].
5.2. Contrast echocardiography
If hepatopulmonary syndrome is suspected, trans-
thoracic microbubble contrast echocardiography is
the preferred screening test for intrapulmonary vasodi-
latation [12]. Contrast echocardiography is performed
by injecting agitated saline intravenously during nor-
mal transthoracic echocardiography, producing micro-
bubbles that are visualized by sonography. This bolus
opacifies the right ventricle within seconds and in the
absence of right-to-left shunting, bubbles are absorbed
in the lungs. If an intra-cardiac shunt is present, con-
trast agent enters the left ventricle within three heart
beats (early shunting). If intrapulmonary shunting
characteristic of hepatopulmonary syndrome is pres-
ent, the left ventricle opacifies at least three heart
beats after the right (delayed shunting). While up to
40% of patients with cirrhosis have a positive contrast
echocardiogram [12], only a subset of these patients
whom 27 (24%) had HPS [3]. The median survival
among patients with HPS was significantly shorter
al of
have sufficient vasodilatation to cause abnormal gas
exchange and fulfill criteria for hepatopulmonary syn-
drome. If a patient with liver disease or portal hyper-
tension and hypoxemia has a positive contrast
echocardiogram in the absence of significant cardio-
pulmonary disease, the diagnosis of hepatopulmonary
syndrome has been established. A semi-quantitative
scoring system for assessing intrapulmonary shunting
during contrast echocardiography has been developed,
though it remains unclear whether the degree of
shunting correlates with the degree of gas exchange
abnormalities [54].
5.3. Macroaggregated albumin scan
In hypoxemic patients with both intrapulmonary
vasodilatation and intrinsic cardiopulmonary disease,
the technetium-labeled macroaggregated albumin scan
(MAA scan) may be useful in defining the contribu-
tion of HPS to gas exchange abnormalities. In this
test, radiolabeled aggregates of albumin measuring
approximately 20 lm in diameter are infused into
the venous system. Ordinarily, particles of this size
become trapped in the pulmonary microvasculature
and scintigraphy reveals nearly complete uptake in
the lungs. In the presence of significant intrapulmo-
nary shunting, a fraction of the macroaggregated
albumin passes through the lungs and into the sys-
temic circulation. Scintigraphy then also reveals
uptake in other organs in addition to the lung, allow-
ing the calculation of the shunt fraction. In one
study, the MAA scan was positive only in patients
with HPS and a PaO2 <60 mmHg and not in COPD
patients with a similar degree of hypoxemia [8]. How-
ever, the MAA scan is less sensitive than contrast
echocardiogram and may be most useful in determin-
ing whether HPS contributes to hypoxemia in
patients with concomitant obstructive pulmonary
disease.
5.4. Pulmonary function studies
While abnormal pulmonary function studies are fre-
quently observed in HPS, these findings are of low
specificity [55]. In the absence of concomitant obstruc-
tive or restrictive lung disease, measurements of total
lung capacity and expiratory flow rates in HPS
patients are generally normal. Diffusion impairment
is commonly seen in HPS [55]. In one study, the dif-
fusing capacity for carbon monoxide (DLCO) was less
than 80% of the predicted value in 15 of 18 patients
with HPS [49]. However, the presence of decreased
DLCO with normal spirometry is not specific for
HPS, and is routinely observed in patients with early
interstitial lung disease, vasooclusive disease, and pro-
D.T. Palma, M.B. Fallon / Journ
found anemia [56].
(10.6 months) compared to patients without HPS
(40.8 months). Mortality remained higher in those
with HPS after adjusting for severity of underlying liv-
er disease and after excluding patients who underwent
liver transplantation during follow-up. The causes of
death in patients with HPS were mainly due to com-
plications of hepatocellular dysfunction and portal
hypertension and correlated with the severity of hyp-
oxemia in HPS. These data raises the possibility that
the presence of HPS may be an important factor that
influences the progression of liver disease and the risk
of complications related to portal hypertension. Final-
ly, even modest hypoxemia related to HPS may wors-
en during sleep based on the observation that
nocturnal oxygen saturation decreased in a small
cohort of non-HPS cirrhotic patients [61].
Mortality after liver transplantation also appears to
be higher in patients with HPS compared to those
without HPS. The utility of the severity of HPS as a
predictor of outcome after liver transplantation has
been prospectively evaluated in a cohort of 24 patients
5.5. Other diagnostic techniques
Pulmonary angiography is expensive and invasive
and has a low sensitivity for detecting intrapulmonary
vasodilatation. Therefore, it is not routinely utilized in
the diagnosis of HPS. High-resolution chest computer-
ized tomography (CT) and evaluation of pulmonary
blood transit time are newer diagnostic modalities
for assessing HPS. In one study, the degree of pul-
monary microvascular dilatation observed on chest
CT correlated with the severity of gas exchange
abnormalities in a small cohort of patients with
HPS, suggesting that quantitation of intrapulmonary
vasodilatation was possible [57]. In another study, pul-
monary transit time of erythrocytes, measured by
echocardiographic analysis of human serum albumin-
air microbubble complexes through the heart, also
correlated with gas exchange abnormalities in a small
group of patients with HPS [58]. Whether these tech-
niques have diagnostic utility for HPS remains to be
determined.
6. Natural history and prognosis
The natural history of hepatopulmonary syndrome
is incompletely characterized. Most patients appear
to develop progressive intrapulmonary va