Cirrhosis and hepatopulmonary syndrome

Abstract

Hepatopulmonary syndrome (HPS) is characterized as a triad: liver disease, intrapulmonary vascular dilatation and arterial hypoxemia. HPS is reported to be present in 4% to 32% of adult patients with end-stage liver disease and in 9%-20% of children. The pathogenesis of HPS has not been clearly identified. Portal hypertension causes impairment in the perfusion of the bowel and increases the enteral translocation of Gram (-) bacteria and endotoxins. This stimulates the release of vasoactive mediators, such as tumor necrosis factor-alpha, heme oxygenase-derived carbon monoxide and nitric oxide. Genetic alterations have not been associated with this syndrome yet; however, cytokines and chemokines have been suggested to play a role. Recently, it was reported that cumulated monocytes lead to the activation of vascular endothelial growth factor-dependent signaling pathways and pulmonary angiogenesis, which plays an important role in HPS pathogenesis. At present, the most effective and only radical treatment is a liver transplant (LT). Cirrhotic patients who are on the waiting list for an LT have a shorter survival period if they develop HPS. Therefore, it is suggested that all cirrhotic cases should be followed closely for HPS and they should have priority in the waiting list.

Key Words: Cirrhosis, Hepatopulmonary syndrome, Pathophysiology, Liver transplantation

Core tip: Hepatopulmonary syndrome (HPS) is an important complication of cirrhosis. HPS is a significant factor in dyspnea and cyanosis in cirrhotic cases. At present, the most effective and only radical treatment is a liver transplant. Cirrhotic patients who are on the waiting list for a liver transplant have a shorter survival period if they develop HPS. Therefore, it is suggested that all cirrhotic cases should be followed closely for HPS and they should have priority in the waiting list. This review aims to reevaluate the recent progress in the diagnosis, pathophysiology and treatment of HPS.

INTRODUCTION

Pulmonary complications related to chronic liver diseases are frequently observed. The two most significant complications among them are hepatopulmonary syndrome (HPS) and portopulmonary hypertension (POPH). HPS is observed more frequently than POPH in patients with chronic liver diseases. HPS has been reported to be present in 4% to 32% of adult patients with end-stage liver disease[1,2], and in 9%-20% of children[3-5].

Kennedy et al[6] first defined HPS in 1977. HPS is characterized as a triad: liver disease, intrapulmonary vascular dilatation and arterial hypoxemia[7]. Cirrhosis is the most common condition associated with HPS. The cause of liver disease leading to portal hypertension does not seem to affect the development of HPS. HPS has been reported in patients with prehepatic portal hypertension in the absence of chronic liver disease, in Budd-Chiari syndrome and even in patients with acute or chronic inflammatory liver disease without evidence of cirrhosis or portal hypertension[8-12].

Prognosis is poor with the development of HPS in patients waiting for a liver transplant (LT). Therefore, these patients should be followed closely. This review aims to reevaluate the recent progress in the diagnosis, pathophysiology and treatment of HPS.

PATHOPHYSIOLOGY

The pathogenesis of HPS has not been clearly identified. However, there are some important clinical clues. Although it has been observed in cases without cirrhosis and portal hypertension, most patients with HPS have cirrhosis and portal hypertension[4,13].

Portal hypertension causes impairment in the perfusion of the bowel and increases the enteral translocation of Gram (-) bacteria and endotoxins. This stimulates the release of vasoactive mediators, such as tumor necrosis factor-alpha (TNF-α), heme oxygenase (HO)-derived carbon monoxide (CO) and nitric oxide (NO)[14-18] (Figure 1).

Figure 1 Possible mechanisms of hepatopulmonary syndrome. TNF-α: Tumor necrosis factor-alpha; ET-1: Endothelin-1; ETBR: Endothelin B receptors; NO: Nitric oxide; eNOS: Endothelial NO synthase; iNOS: inducible NO synthase; HO-1: Heme oxygenase-1; CO: Carbon monoxide; VEGF: Vascular endothelial growth factor.

In clinical studies, the increase of nitric oxide production in the lung plays a role in HPS pathogenesis[16-23]. When compared with cirrhotic control patients, exhalation NO levels increase in the cases with cirrhotic HPS. Plasma endothelin-1 (ET-1) levels increase in cases with cirrhosis and intrapulmonary vascular dilatation[24-26]. ET-1 causes NO-related vasodilatation with the activation of endothelin B receptors (ET_BR) on endothelial cells[27,28]. In addition, increased phagocytosis of bacterial endotoxin in the lung promotes activation of inducible NO synthase (iNOS), which also contributes toward increased NO production. Bacterial translocation, and subsequent monocyte accumulation, may also stimulate pulmonary angiogenesis in HPS, which may partly be controlled by genetic factors[13].

Genetic alterations have not been associated with HPS yet; however, cytokines and chemokines have been suggested to play a role. It is more common in patients carrying the monocyte chemoattractant protein-1 (MCP-1) 2518G gene; conversely it is less frequent in patients with the endothelial NO synthase (eNOS) 298Asp allele and those carrying eNOS 298Asp[29]. These results suggest that the eNOS 298Asp polymorphism can prevent the development of HPS in cirrhotic patients. In addition, the G allele may be associated with higher MCP-1 expression in certain inflammatory conditions. The effect of the G allele appears to be dose dependent: cells from individuals homozygous for G at -2518 produced more MCP-1 than cells from G/A heterozygotes[30]. Higher levels of MCP-1, an inflammation marker, in HPS suggest the role of inflammation in the development of pulmonary shunts. Moreover, macrophages produce HO-1, which leads an increase in the production of CO and contributes to the vasodilatation[15]. CO is produced by the catabolism of heme with heme oxygenase[15,31].

Recently, cumulated monocytes have been observed to lead to the activation of vascular endothelial growth factor-dependent signaling pathways and pulmonary angiogenesis, which plays an important role in HPS pathogenesis[32]. Gene polymorphisms involved in the regulation of angiogenesis have also been associated with the risk of developing HPS[33].

Although other mediators, such as somatostatin analogue (octreotide), glucagon, prostacyclin, angiotensin-2, vasoactive intestinal peptide, calcitonin, substance P, atrial natriuretic factor and platelet-activating factor, may play a role in the pathogenesis of vascular changes in HPS, no clear relation was found between any of these mediators and vascular dilatation[34-39].

Important mechanisms mentioned above result in ventilation-perfusion (V/Q) mismatch, diffusion limitation of oxygen and, less commonly, direct arteriovenous connections[40,41]. The capillaries dilate to 15-500 μm (n: 8-15 μm) in HPS[42].

The hypoxia occurs because of the increased cardiac output caused by pulmonary vasodilatation and the inadequate oxygenation of the blood, which runs through enlarged pulmonary microcapillaries (Figure 2). In addition, hypoxia worsens as a result of the vasoconstrictor response to hypoxia, especially in the lower zones where there is less ventilation[43]. Lowered V/Q ratios result in elevated ΔP (A-a) O₂, which is correctable only by 100% oxygen inhalation. However, hypoxemia caused by larger arteriovenous shunts (AVS) does not respond to inhalation of 100% oxygen[44].

Figure 2 Ventilation-perfusion mismatch; diffusion limitation of oxygen. The capillaries are known to dilate to 15-500 μm (n: 8-15 μm) in hepatopulmonary syndrome (HPS).

Previous studies have shown that certain changes occur in lungs of animals with HPS. Melo-Silva et al[45] showed that in rat models, the tidal volume, minute ventilation and mean inspiratory flow were significantly reduced, chest wall pressure dissipation against the resistive and viscoelastic components and elasticity were increased, and the lung resistive pressure dissipation was lower; however, the viscoelastic pressure was higher in the HPS group. The proportion of collagen volume in the vasculature increased by 29% in the HPS animals.

Furthermore, patients with hepatic cirrhosis have an elevated plasma level of lipopolysacchride (LPS)[46]. Extra LPS was given to rats with cirrhosis, which resulted in further widening of the alveoli wall, a decreased density of cells, narrowed alveoli space and destruction of the integrity of type I cell membrane, with infiltration of polymorphs and fibrinous exudates, indicating interstitial pulmonary edema and an inflammatory reaction. There was severe stasis of the blood in alveolar walls and numerous red cells extravasated the airspace, resulting in the widespread dilatation of alveolar capillaries and the augmentation of the permeability of the microvasculature[18].

CLINICAL FEATURES

Cases of HPS can be asymptomatic or they can present with growth retardation, cyanoses, dyspnea, platypnea, orthopnea, spider angioma or finger clubbing. In a study conducted in The Mayo Clinic, 82% of 22 patients with HPS had symptoms and findings related to liver disease, and the time period from the respiratory complaints to the diagnosis of HPS was an average of 4.8 ± 2.5 years. In the same study, 18% of the patients complained of labored breathing and the liver disease was diagnosed after further tests[47]. The most frequent symptom was progressive dyspnea. Platypnea is defined as dyspnea of a patient when he/she is in an upright position. Digital clubbing and cyanosis are also frequent in HPS patients[29,48,49]. Spider angioma is not specific for the diagnosis of HPS[50]. The findings of chest radiography and decreased CO diffusion capacity determined by pulmonary function tests for the patients with HPS were nonspecific[51].

There is a poor correlation between liver disease and the level of oxygenation. While it has been reported mostly in patients with severely decompensated end-stage liver disease, namely child C patients, it has been also observed in patients with child A and child B cirrhosis[13,52].

In cirrhotic cases with suspicion of HPS, other cardiopulmonary diseases, such as pulmonary atelectasis, ascites, chronic obstructive pulmonary diseases, hepatic hydrothorax and infections should be ruled out.

Two types of HPS have been defined. While type 1 is related to precapillary dilatations, AVS is associated with type 2 HPS. In addition, while type 1 HPS responds to oxygen support, type 2 does not[53]. If PO₂≥ 600 mmHg when the patient is given 100% oxygen, it is considered that it is probably not AVS. If the PO₂ fails rise to 150-200 mmHg or over, then the hypoxia may be considered to be caused by AVS[54].

DIAGNOSIS

HPS is a serious complication frequently seen in cirrhotic patients and results from hypoxemia. The clinical findings in HPS are the same as those of hypoxemia. Contrast echocardiography (CEE) is accepted as a sensitive screening test for HPS. Other techniques, such as technetium 99m-labeled macroaggregated albumin (MAA) scan, computed tomography or pulmonary angiography, have also been used to diagnose HPS[55]. Saline microbubbles are used for CCE. A transpulmonary passage was considered present if microbubbles appeared in the left atrium at least three heartbeats after the initial appearance of contrast in the right side of the heart[56-59] (Figure 3).

Figure 3 “Bubble” or contrast echocardiogram (apical four chamber view). A stream of microbubbles filling the left atrium following systemic venous injection is shown.

Transesophageal CEE increases the imaging quality of the heart; therefore, it is more sensitive than transthoracic CEE to identify intrapulmonary vasodilatation and may detect HPS, which can go unnoticed by transthoracic CEE[60]. For patients with hypoxemic HPS, there is no study that shows it is more superior. The detection of intrapulmonary shunts by this method is not sufficient to diagnose HPS and impaired oxygenation is required to be shown.

In addition, in cases with HPS, left ventricle enlargement and higher systolic velocity in the mitral valve represent satisfactory indirect markers of HPS[41], in addition to CEE[61].

Orthodeoxia, positional modification of abnormal shunting, was defined as a fall in PaO₂≥ 5% when upright, or 4 mmHg. While the PaO₂ is normal in the horizontal position, it decreases in the upright position, depending on the increase of blood flow velocity in arteriovenous anastomosis in the basal segments of the lungs, caused by the effect of the gravity. This increases the ventilation-perfusion mismatch and hypoxia becomes apparent[62]. Orthodeoxia has been reported to be present in 20% to 80% of patients with HPS[63,64]. Hypoxemia is defined as PaO₂ < 70 mmHg, and severe hypoxemia is defined as PaO₂ < 50 mmHg. At sea level, and while breathing room air, a resting PA-aO₂ of ≥ 2.0 kPa (≥ 15 mmHg) can be considered abnormal. For the people over 64, it should be evaluated as PA-aO₂≥ 20 mmHg. Accordingly, the PA-aO₂vs PaO₂ calculation is done to grade the severity of HPS[58] (Table 1).

Table 1 Grading of the severity of hepatopulmonary syndrome[58].

Stage	PA-aO2 mmHg	PaO2 mmHg
Mild	≥ 15	≥ 80
Moderate	≥ 15	< 80 - ≥ 60
Severe	≥ 15	< 60 - ≥ 50
Very severe	≥ 15	< 50 (< 300 on 100% O2)

PA-aO₂: Alveolar-arterial oxygen tension difference; PaO₂: Arterial oxygen tension.

In recent years, transcutaneous oxygen saturation measurement with pulse oximetry has emerged a simple, low cost, and widely available technique to screen for HPS. With a threshold value of < 96%, pulse oximetry has a sensitivity and specificity of 100% and 88%, respectively, for detecting patients with a PaO₂ < 60 mmHg. A pulse oximetry value of < 94% detected all patients with a PaO₂ < 60 mmHg with an increased specificity of 93%[65,66]. Contrary to these findings, CEE may be positive despite normal arterial blood gases. In a prospective study of candidates for LT, Krowka et al[67], found that 9.7% of 31 normoxemic patients had positive CEE. These findings suggested that mild or subclinical intrapulmonary vasodilatations insufficiency in cirrhotic patients may not alter gas exchange.

In the Technetium 99m-labeled MAA scan, MAA particles are given intravenously. The diameter of the marked particles is 20-50 μm and normally they cannot pass through pulmonary veins, which have a diameter of 8-15 μm. However, in the presence of an intrapulmonary shunt, these marked particles enter the circulatory system and appear in the kidneys and the brain. In the diagnosis of HPS, a value greater than 6% is significant and specific for HPS. However, as MAA provides positive results in the presence of intracardiac shunts as well; therefore, its sensitivity is low[41,68].

Unless there is an accompanying pulmonary disease, the spirometric tests in HPS are not impaired. However, abnormal diffusion capacity for carbon monoxide (DLCO) is frequently observed in patients with HPS[69]. In one study, the DLCO was decreased in 80% of the cases[70]. However, its specificity is low; therefore, it is not used in practice.

Pulmonary angiography is more invasive and less sensitive compared with high resolution chest computed tomography CEE[71] (Figures 4 and 5).

Figure 4 Right pulmonary artery angiogram (posteroanterior projection) showing a diffuse fine reticular pattern of multiple pulmonary telangiectasias consistent with type I hepatopulmonary syndrome.

Figure 5 Normal pulmonary angiography.

TREATMENT

Currently, there are no effective medical therapies for HPS. In the past, HPS was considered as a contraindication for LT because of serious operative and perioperative complications in adults[72]. Today, LT is the only effective treatment option for patients with this condition, because of the underlying liver disease.

Although several investigations have been performed, no effective medical treatment has been found. Several attempts have been made to inhibit the development of HPS by administering nitric oxide, using diets low in L-arginine using methylene blue, which is an inhibitor of guanylate cyclase[73], aspirin[36], somatostatin[49,74], almitrine[75], N-acetylcysteine[76], indomethacin[51], garlic[77,78], mycophenolate mofetil (an inhibitor of angiogenesis and nitric oxide production)[79], pentoxifylline[80], decreasing the increased portal pressure by transjugular portosystemic shunt[81-83], and using antibiotics to decrease bacterial translocation in the bowel[84]. However, a role for any of these drugs in the long-term treatment of HPS has not been demonstrated. Currently, the most effective treatment is LT.

In cirrhotic cases with HPS, survival was significantly decreased compared with cirrhotic cases without HPS. In one study, the five-year survival rate was determined as 23% for cases with HPS and 67% for patients without HPS[85]. In studies performed at The Mayo Clinic, 33%-40% of patients with HPS died within 2.5-4 years. Most of the patients having clinically stable hepatic function worsened with the development of HPS[47]. In our center, eleven of 16 patients (68.7%) with HPS died before LT. The main causes of deaths were variceal bleeding accompanying multiorgan failure, pneumonia and sepsis[4].

Therefore, in addition to their Model for End-Stage Liver Disease (MELD) scores, it is very important for patients on the waiting list to be diagnosed as having HPS or not. However, many clinical features of HPS that might influence exception to the MELD scoring system, including standardized diagnostic criteria, pre- and post-LT mortality rates, and the rate of progression of hypoxemia, are not fully characterized[86]. In the United States, UNOS allows patients with confirmed HPS to be listed for an LT. For this reason, it is suggested that the patients with PaO₂ < 60 mmHg on room air in the sitting position, should receive an increase in their MELD scores, and during their waiting period they should get a 2- to 3-point increase in their MELD scores every three months[86]. The indications are that patients who had an LT this way showed better survival. Iyer et al[87] observed a trend (without reaching statistical significance) for better 5-year survival in the MELD exception era (since 2002) as compared to earlier HPS transplants.

The severity of preoperative hypoxemia, underlying liver disease and comorbidities appear to be factors that increase postoperative mortality[88,89]. While some studies show that the mortality rate of cases with HPS are 29%-38.5%[4,90], there are other studies that show the survival after an LT is not different to that od the cirrhotic cases without HPS[84,89]. In fact Iyer et al[87] suggested that survival post-LT was not dependent on baseline PaO₂ values obtained at the time of HPS diagnosis.

85%-100% of patients have an improvement in oxygenation within 1 year of an LT[4,53]. However, in some severe cases with hypoxemia, lung function may not recover up to one year after the operation. Most of these cases need oxygen for 5 to 700 d, and spend longer in hospital[89].

After liver transplantation and after all the factors that led to HPS have disappeared, patients recover from HPS[91,92]. In a single post mortem study, the changes after the transplantation in lungs in HPS could be related to collagen tissue deposition in pulmonary capillary and venule walls[93]. However, the recovery from HPS after transplantation shows that the pathological changes in the lungs in Type I HPS are reversible.

In our center, the perioperative and postoperative outcomes were uneventful in all patients. Even if previous reports indicated that patients with HPS require ventilatory support after an LT[67,94], two patients remained on mechanical ventilation and they were extubated two and five days post-transplant, respectively. If the post-operative hypoxemia is refractory to standard treatments, NO and trendelenburg positioning (supine position with feet elevated 15°-30° higher) and oscillator ventilation therapy are reported to be effective[89]. Experiences related with outcome of HPS after an LT are limited; portal venous thrombosis, intracranial events and multiorgan failure are seen more commonly in patients with HPS, and these complications are associated with a higher mortality rate[95]. In our series, we did not observe these complications, except for one case of acute respiratory distress syndrome, which was observed on postoperative day 1. This was probably related to systemic inflammatory response syndrome, because the patient had arterial hypoxemia, fever and ground glass opacity on the chest radiography.

In conclusion, HPS is an important complication of cirrhosis, causing dyspnea and cyanosis in cirrhotic cases. There is no relation between the development of HPS and the severity of cirrhosis. At present, the most effective and only radical treatment is an LT. Cirrhotic patients who are on the waiting list for an LT have a shorter survival period if they develop HPS. Therefore, it is suggested that all cirrhotic cases should be followed closely for HPS and should have priority in the waiting list.