Definition

Hepatopulmonary syndrome (HPS) is defined by abnormal oxygen exchange in association with intrapulmonary vascular dilatation (IPVD) in patients with liver disease.¹ The presence of other cardiopulmonary disease that alters gas exchange does not exclude this diagnosis.^2–5 HPS is most commonly associated with cirrhosis¹ and portal hypertension, but neither of these are required.⁶ The correlation between the degree of liver dysfunction and the presence^7–8 and severity^{3–4,7,9–10} of this syndrome is debated.

Clinical Features

HPS usually presents as dyspnea^6,11 in patients who are already known to have liver disease. HPS-induced shortness of breath often is relieved when the patient is lying down,^11–12 and therefore is referred to as platypnea. There are no consistently noted physical examination findings.^6,13 Hypoxia is often worse in the standing position (orthodeoxia),¹² and it generally can be corrected with sufficient supplemental oxygen.^1,3,4,10,14

Pathophysiology

Dilated precapillary vessels and pleural-based arteriovenous connections are noted at autopsy in cases of HPS.¹⁵ Current thinking suggests that these abnormal vessels develop due to a functional excess of pulmonary vasodilators¹; they cause hypoxia through ventilation/perfusion (V/Q) mismatching, arteriovenous (AV) shunting, and limitation of oxygen diffusion to red blood cells (RBCs) in the center of the vessel.^15–17 The hyperdynamic circulation, which is characteristic of cirrhosis, likely exacerbates this problem by decreasing RBC transit time through the pulmonary capillaries, further limiting oxygen diffusion.^15,17 Orthodeoxia is due to a worsening of V/Q mismatch and AV shunting in the standing position.¹⁸

Nitric oxide (NO) has been implicated as a key vasodilator in HPS. Exhaled NO levels are increased in patients with cirrhosis compared to healthy controls and in HPS patients compared to cirrhotic patients without HPS; NO levels correlate with the severity of cirrhosis and gas exchange abnormalities.¹⁹ In rat models of HPS induced by ligation of the common bile duct (CBDL), increased levels of endothelial²⁰ and inducible NO synthase (eNOS and iNOS, respectively) have been observed, and administration of a nitric oxide synthase inhibitor prevents the development of pulmonary vasodilation and HPS.²¹

Excess eNOS is located in the pulmonary arteries and capillaries and is associated with impaired vasoconstriction; levels of this enzyme correlate with the degree of gas exchange abnormalities.²⁰ CBDL rats demonstrate increased hepatic production of endothelin-1 (ET-1)²² and increased vascular expression of the endothelin-B receptor (ET-B)²³ in proportion to the severity of gas exchange abnormalities^22,24; interaction between ET-1 and the ET-B receptor, therefore, is believed to be the trigger for increased eNOS expression. This theory is further supported by data that show a reduction in eNOS expression and an improvement in HPS when CBDL animals are treated with endothelin-B receptor antagonists.²⁴

iNOS is expressed in macrophages found in the lungs of CBDL rats,²¹ while treatment of these rats with norfloxacin is associated with a reduction in the rate of gram-negative bacterial translocation, accumulation of pulmonary macrophages, production of iNOS, and severity of HPS.²⁵ Pulmonary macrophages in CBDL rats also have been noted to express elevated levels of heme-oxygenase-1 (HO-1), an enzyme that catalyzes formation of the vasodilating gas, carbon monoxide (CO).²⁶ Increased levels of carboxyhemoglobin (COHb) have been observed in rat²⁶ as well as human²⁷ subjects with HPS, and treatment with an HO-1 inhibitor normalizes COHb levels and partially alleviates HPS in CBDL rats.²⁶ These data suggest that CO also contributes to pulmonary vasodilation in this syndrome. Finally, tumor necrosis factor alpha (TNF-α) rises in CBDL animals in association with ET-1 and endotoxin levels, and it has been proposed to influence accumulation of the iNOS- and HO-1-producing pulmonary macrophages.²⁸ Administration of pentoxifylline, a phosphodiesterase inhibitor that suppresses production of TNF-α, is associated with a reduction in TNF-α levels, pulmonary macrophage accumulation, ET-B receptor and eNOS expression, and severity of HPS.²⁹

Diagnosis

HPS should be considered in any patient with liver disease and dyspnea or hypoxia. Evaluation begins with an arterial blood gas (ABG), with the patient resting in the seated position and breathing room air.^6,17 No specific gas exchange criteria for HPS have been universally accepted,³⁰ but a 2004 European Respiratory Society (ERS) task force advised further evaluation when the PaO₂ is less than 80 mm Hg or the alveolar-arterial oxygen gradient (A-a gradient) is 15 mm Hg or greater (≥20 mm Hg for patients over age 64).¹⁷

Accurate HPS diagnosis requires the presence of ABG changes that cannot be fully explained by comorbid cardiopulmonary disease. Conditions that frequently coexist with cirrhosis that may influence gas exchange include chronic obstructive pulmonary disease (COPD), congestive heart failure, restrictive lung disease due to ascites or hepatic hydrothorax, α₁-antitrypsin deficiency, and portopulmonary hypertension (distinguished from HPS by its increased pulmonary artery pressure and vascular resistance; in HPS, pulmonary artery pressure and vascular resistance are low).³¹ Patients should have a chest x-ray (CXR) and pulmonary function tests¹² to assess for pulmonary disease; of note, increased markings at the lung bases on CXR^1,6,8 and/or a reduced diffusion capacity for carbon monoxide (DLCO)^{3–4,7,13,18} are common findings in HPS and, in isolation, do not exclude the diagnosis. Cardiac function is evaluated by echocardiogram, often concurrently with IPVD assessment (see later discussion).

When a gas exchange abnormality is present and not fully explained by another cardiopulmonary disease, the patient should be evaluated for the presence of IPVDs. Contrast-enhanced echocardiography (CEE) is commonly used for this purpose; advantages include that it is widely available, it permits concurrent evaluation for cardiac causes of abnormal gas exchange, and it can distinguish intracardiac from intrapulmonary shunt based on the number of cardiac cycles required for agitated saline to pass from the right to left atrium.¹² CEE is highly sensitive for the presence IPVDs³⁰ and may document them in up to 82% of patients tested.³² Compared with patients without IPVDs, those with a positive CEE have a greater incidence of dyspnea³³ and abnormal CXRs,^9,33 as well as more severe cirrhosis^9,32–33 and gas exchange abnormalities.^9,33 However, many patients with IPVDs demonstrated by CEE do not have gas exchange abnormalities,^{5,8,13,32–33} and so this test is not very specific for HPS.³⁰

Technetium-99m-labeled macroaggregated albumin (^99mTc MAA) lung perfusion scanning is an alternative test for IPVDs. It is expensive, requires radiation exposure,¹³ and cannot document the site of shunting, but it is able to provide a quantitative shunt fraction^13,17,30 that correlates directly with the A-a gradient^3,10,14 and inversely with the room air PaO₂^3,10,14,34 and oxygen saturation.³⁴ Perfusion scanning is less sensitive than CEE for the detection of IPVDs,⁵ but positive results are rare in patients without HPS.^5,10,34 Because of these test characteristics, CEE has been advocated as the first-line modality for evaluating patients with liver disease and abnormal gas exchange.^5,10,17 If CEE is positive but the relative contributions of other cardiopulmonary disease and possible HPS are not clear, lung perfusion scanning can determine if HPS is present.^5,10,12,17

Prevalence

The prevalence of HPS varies greatly depending on how it is diagnosed. When abnormal gas exchange is defined by widening of the A-a gradient, more patients meet HPS criteria than when a reduction of PaO₂ is used,⁹ because hyperventilation can maintain a normal PaO₂ while the A-a gradient is still elevated due to decreased PaCO₂.¹⁵ HPS prevalence is also affected by the sensitivity of the IPVD evaluation method used; IPVDs are found more frequently with CEE compared to lung perfusion scanning⁵ and with TEE compared to TTE,³⁵ thereby increasing HPS diagnosis rates.^5,35 For example, HPS was diagnosed in 3 of 40 (7.5%) cirrhotic patients when a PaO₂ of less than 70 mm Hg and a positive lung perfusion scan were required⁵; when criteria consisted of an A-a gradient more than 15 mm Hg and a positive CEE, the prevalence of HPS among patients with cirrhosis was reported as 32% (31 of 98 patients).⁹

Prognosis

In the absence of liver transplantation, patients with HPS have a poorer functional status, reduced self-reported quality of life,⁸ and a worse survival^7–836 than non-HPS controls matched for severity of liver disease. HPS patients who die during follow-up have been noted to have greater room air PaO₂ reductions, A-a gradient elevations, and shunt fractions than those who survive,^3,7,36 but this is not a universal finding.⁸ Without a transplant, HPS patients demonstrate progressive hypoxemia.³⁶ Despite this, death is usually due to complications of liver disease,^7,17,36 and mortality from primary respiratory failure is rare.³⁶

Therapy

Multiple medical therapies have been tried for HPS without clear efficacy, including inhibitors of nitric oxide^37–39 and TNF-α production,⁴⁰ as well as antibiotics to reduce macrophage accumulation.⁴¹ Case reports have documented improvement in HPS after transjugular intrahepatic portosystemic shunt (TIPS) placement, but this therapy is still considered experimental.^12,42 Pulmonary angiography with embolization of dilated capillaries⁴³ or arteriovenous communications⁴⁴ also has been effective in case reports. Some authors advise pursuing pulmonary angiography in HPS patients with a poor response to administration of 100% oxygen,^1,3,44 since these patients are more likely to have large shunts that may improve with embolization. Oxygen therapy also has been recommended.^{1,6,11–12,17}

Liver transplantation is the only definitive therapy for HPS and should be considered when patients are symptomatic or have a PaO₂ less than 60 mm Hg.^6,17 Owing to the increased mortality associated with HPS and the lack of other effective therapies, the United Network for Organ Sharing (UNOS) has adjusted organ allocation algorithms to prioritize patients with HPS and a PaO₂ below 60 mm Hg.⁴⁵

Patients with HPS who receive liver transplants have been observed to have a greater postoperative mortality than their non-HPS counterparts in two series (33% of 9 HPS patients versus 9.2% of 76 non-HPS patients at 6 months⁴⁶; 29% of 24 HPS patients versus 8%–10% of historical controls at 1 year¹⁴), although data from the largest available HPS population (24 transplanted HPS and 30 transplanted non-HPS patients) showed no survival difference between HPS and non-HPS transplant recipients when followed for over 7 years.³⁶ A preoperative PaO₂ ≤ 50 mm or shunt fraction ≥ 20% was predictive of postoperative mortality in one report (mean PaO₂ 59 mm Hg versus 43 mm Hg, and shunt fraction 18% versus 41% in 17 survivors versus 7 nonsurvivors).¹⁴ A subsequent series of 24 transplanted HPS patients also showed a trend toward increased mortality after transplant among HPS patients with preoperative PaO₂ ≤ 50, although this difference did not reach statistical significance (P = .08).³⁶ Another small series noted a non-significant correlation between mortality and shunt fraction, observing that 3 out of 6 patients with HPS and preoperative shunt fractions over 30% died during the first 60 days after liver transplant, while only 1 of 6 HPS patients with a shunt fraction ≤ 30% died (at day 71).³ Among those patients who survive the perioperative period, improvement in or resolution of HPS is noted in the majority of cases,^4,14,36,46 although the amelioration of symptoms may require a year or more to occur.^4,14,36

Key Points