Unexplained Dyspnea: Hepatopulmonary Syndrome without Cirrhosis?

The Clinical Challenge

A 64-year-old man with a history of hepatic steatosis and obstructive sleep apnea was referred for evaluation of exertional dyspnea. He had previously been a runner, though over the 5 years before presentation, he had stopped running because of gradually increasing dyspnea that he also noticed while walking up inclines or stairs. He denied associated chest pain, cough, and constitutional symptoms. He was a prior 5–pack-year smoker who had quit more than 30 years previously, and his family history was significant for sarcoidosis in a sibling.

On the physical examination, blood pressure was 116/74 mm Hg, heart rate was 65 beats/min, temperature was 36.3 °C, and oxygen saturation was 93% on ambient air without orthodeoxia, dropping to 89% with brisk ambulation. Respiratory effort was normal at rest, and lungs were clear to auscultation. The patient did not have jugular venous distention or cardiac murmurs, and there was physiologic splitting of S2, with a normal P2. He was without ascites, peripheral edema, skin rash, telangiectasias, or clubbing.

A complete blood count and comprehensive metabolic panel were unremarkable (Table 1). Computed tomography pulmonary angiography imaging (CTPA) was negative for pulmonary emboli and demonstrated only mild apical scarring. An arterial blood gas measurement showed a pH of 7.48, a carbon dioxide partial pressure of 31 mm Hg, an oxygen partial pressure (Po₂) of 62 mm Hg, and an alveolar–arterial (A–a) Po₂ difference of 49 mm Hg without carboxyhemoglobinemia or methemoglobinemia. Pulmonary function testing revealed a diffusing capacity of the lung for carbon monoxide (Dl_CO) that was 57% of the predicted value with normal forced expiratory volume in 1 second (FEV₁), forced vital capacity (FVC), FEV₁:FVC, and total lung capacity (Table 2). Nuclear stress testing was negative for cardiac ischemia. Transthoracic echocardiography (TTE) demonstrated normal ventricular size and function, normal biatrial size, no significant valvular abnormalities, and normal estimated right ventricular (RV) systolic pressure. Agitated saline was injected at rest in the supine position, without appearance in the left heart. With cough, however, transit occurred after six or seven cardiac cycles, consistent with intrapulmonary shunting. Prior records revealed evidence of hepatic steatosis on abdominal ultrasound, raising concern for hepatopulmonary syndrome (HPS).

Table 1.

The initial complete blood count and comprehensive metabolic panel

Measurement	Value
White blood cell count, ×103/μl	10.5
Hemoglobin, g/dl	15.3
Hematocrit, %	45.7
Platelet count, ×103/μl	221
Sodium, mM/L	139
Potassium, mM/L	4.5
Chloride, mM/L	100
Bicarbonate, mM/L	24
Blood urea nitrogen, mg/dl	23
Creatinine, mg/dl	1.05
Glucose, mg/dl	95
Calcium, mg/dl	10.0
Total protein, g/dl	7.3
Albumin, g/dl	4.7
Total bilirubin, mg/dl	0.6
Alkaline phosphatase, IU/L	113
AST (SGOT), IU/L	22
ALT (SGPT), IU/L	24

Definition of abbreviations: ALT = alanine aminotransferase; AST = aspartate aminotransferase; SGOT = glutamic-oxaloacetic transaminase; SGPT = serum glutamic-pyruvic transaminase.

All values were within reference ranges.

Table 2.

Significant values obtained via pulmonary function testing

PFT Parameter	Value	Percentage Predicted
FVC, L	4.85	108
FEV1, L	3.54	106
FEV1:FVC ratio	0.73	N/A
TLC, L	7.05	103
DlCO, ml/min/mm Hg	13.16	57
DlCO/VA, ml/min/mm Hg	2.07	61

Definition of abbreviations: Dl_CO = diffusing capacity of the lung for carbon monoxide; FEV₁ = forced expiratory volume in 1 second; FVC = forced vital capacity; N/A = not applicable; PFT = pulmonary function testing; TLC = total lung capacity; VA = alveolar volume.

FVC, FEV₁, FEV₁:FVC ratio, and TLC were all within normal limits. However, Dl_CO and Dl_CO/VA were reduced. This isolated reduction in Dl_CO suggested the presence of pulmonary vascular disease.

Questions

• 1.What is the typical evaluation for a patient suspected of having intrapulmonary shunting?
• 2.In HPS, what findings might be expected during cardiopulmonary exercise testing (CPET)?
• 3.In addition to hypoxemia, what is another major cause of dyspnea in HPS?

Clinical Reasoning

This previously healthy patient presented with exercise-limiting dyspnea, mild hypoxemia, and an elevated resting A–a Po₂ difference. His testing revealed an isolated reduction in Dl_CO and a supine bubble study positive only with cough. These findings in the absence of anemia, parenchymal lung disease, or acute pulmonary embolism suggested the presence of pulmonary vascular disease, including pulmonary hypertension, HPS, or pulmonary arteriovenous malformation (PAVM). To exclude chronic thromboembolic disease, a ventilation–perfusion ($\dot{\text{V}}$_A/$\dot{\text{Q}}$) scan (Figure 1) was performed and revealed no perfusion defects, with basilar-predominant perfusion, markedly reduced perfusion in the upper lung fields, and normal ventilation. Recent CTPA lacked evidence of chronic thromboembolic disease or parenchymal changes in the areas of hypoperfusion. However, there was a striking increase in the caliber of the basilar pulmonary arteries compared with CTPA performed 5 years prior (Figure 2). CTPA was performed in the context of transient hypoxemia after outpatient orthopedic surgery. The combination of a basilar lung perfusion pattern, an elevated A–a Po₂ difference, and TTE showing late transit of bubbles with normal RV function and RV systolic pressure was suggestive of intrapulmonary shunting, either from PAVM or a disease process involving intrapulmonary vascular dilatation (IPVD), such as in HPS.

Figure 1.

The ventilation (left) and perfusion (right) scan that was obtained to evaluate for chronic thromboembolic disease after initial workup suggested the possibility of pulmonary vascular disease. This image revealed basilar-predominant perfusion.

Figure 2.

Comparison of prior chest computed tomography (CT) the patient underwent for a different reason 5 years before presentation (left) and chest CT acquired for current workup (right). Images focus on the peripheral vasculature in the lung bases and reveal increased pulmonary vessel caliber compared with the previous study.

The Clinical Solution

Given high suspicion for basilar-predominant intrapulmonary shunting, exercise stress TTE with agitated saline was performed in the upright position. This test demonstrated late transit of bubbles after six or seven cardiac cycles at rest (Video 1), with normal RV size and function at peak exercise (Video 2), consistent with this physiology. CPET revealed reduced peak O₂ consumption, high ventilatory equivalents for CO₂, significantly elevated CO₂-dead space at rest without the expected decrease during exercise, normal O₂ pulse, and borderline breathing reserve (Table 3). After delayed venous-phase CTPA excluded PAVM, IPVD was the only remaining diagnosis that explained the patient’s clinical picture. Small PAVMs related to hereditary hemorrhagic telangiectasia were considered, but small PAVMs without IPVD or pulmonary hypertension would not explain the elevated dead space or the reduced diffusion impairment (1).

Table 3.

Significant values obtained via cardiopulmonary exercise testing

CPET Parameter	Value
$\dot{\text{V}}$o2 peak, ml/kg/min	20.1
$\dot{\text{V}}$o2 peak, percentage predicted	72
Peak heart rate, percentage predicted beats/min	80
Peak minute ventilation, L/min	96
Indirect maximum voluntary ventilation, L/min	137
Breathing reserve, percentage predicted	30
Oxygen saturation nadir	90
Ve:$\dot{\text{V}}$co2 nadir	47
Vd:Vt at rest	0.55
Vd:Vt at peak exercise	0.56
O2 pulse peak, percent predicted	94

Definition of abbreviations: CPET = cardiopulmonary exercise testing; Vd = dead space volume; Ve:$\dot{\text{V}}$co₂ = ventilatory equivalents for carbon dioxide; $\dot{\text{V}}$o₂ = oxygen consumption; Vt = tidal volume.

Percentage predicted breathing reserve was calculated as [1 − (peak minute ventilation/indirect maximum voluntary ventilation) × 100]; indirect maximum voluntary ventilation was calculated as forced expiratory volume in 1 second × 38. $\dot{\text{V}}$o₂ peak was mildly reduced, O₂ pulse was normal, and breathing reserve was borderline. Importantly, Vd:Vt was significantly elevated. In all, these findings are consistent with hepatopulmonary syndrome and intrapulmonary vascular dilatation physiology.

Video 1.

Download video file ^{(22.6MB, mp4)}

Transthoracic echocardiography with agitated saline was performed in the upright position demonstrating transit of bubbles into the left heart after six or seven cardiac cycles at rest. The right ventricular size and function are normal.

Video 2.

Download video file ^{(9.8MB, mp4)}

Transthoracic echocardiography with agitated saline was performed in the upright position and with exercise demonstrating late transit of bubbles into the left heart and preserved right ventricular size and function at peak exercise.

Laboratory analysis showed normal liver synthetic function, and a combination of hepatic duplex ultrasound, abdominal computed tomographic angiography, and liver elastography revealed steatosis without portal hypertension or portosystemic shunt. Liver biopsy demonstrated mild fibrosis, steatosis, and nonnecrotizing granulomas, followed by positron emission tomography computed tomography (CT) revealing fluorodeoxyglucose-avid mild bilateral hilar lymphadenopathy (Figure 3). Evaluation for causes of nonnecrotizing granulomatosis including antineutrophil cytoplasmic antibodies, quantitative immunoglobulins, and anti–nuclear antibody testing was unremarkable, and the patient ultimately received a diagnosis of HPS due to hepatic sarcoidosis. Repeat CPET after a steroid trial lacked evidence of improvement. We did not believe that the patient had active hepatic sarcoidosis given the stable duration of symptoms and no response to prednisone, therefore a decision was made to clinically monitor the patient. Supplemental oxygen was deemed unnecessary given that his overnight polysomnography showed no nocturnal desaturation while on continuous positive-pressure ventilation and his oxygen fell to only 90% with ambulation and during CPET. In addition, CPET suggested that a major etiology of his dyspnea was elevated CO₂-dead space. We cannot exclude the possibility of an exaggerated hypoxic ventilatory response related to his liver disease, previously reported in patients with cirrhosis (2).

Figure 3.

Positron emission tomography computed tomography images centered on mild fluorodeoxyglucose uptake around the hilum, a finding consistent with hilar lymphadenopathy from sarcoidosis.

The Science behind the Solution

HPS is a condition characterized by impaired arterial oxygenation due to intrapulmonary vascular dilations in the setting of liver disease. The pathophysiology behind the gas-exchange abnormalities and accompanying dyspnea in HPS has been well characterized. An understanding of these mechanisms assisted in making the diagnosis of HPS and ultimately sarcoidosis in this patient, despite initially lacking history that might point to underlying cirrhosis.

$\dot{\text{V}}$_A/$\dot{\text{Q}}$ Mismatch as a Major Cause of Arterial Hypoxemia in HPS

$\dot{\text{V}}$_A/$\dot{\text{Q}}$ mismatch has been described as the major contributor to hypoxemia in patients with HPS (3). The characteristic IPVDs contributing to this pathophysiology may be visible on CT, described in previous studies as an increased caliber in the peripheral pulmonary vasculature (3). These IPVDs result in nonuniform perfusion, with affected lung units receiving increased blood flow relative to their ventilation (4). This increased perfusion relative to ventilation results in low $\dot{\text{V}}$_A/$\dot{\text{Q}}$ units and arterial hypoxemia if extensive enough. In our patient, the importance of this mechanism was most evident in his $\dot{\text{V}}$_A/$\dot{\text{Q}}$ scan, demonstrating marked basilar-predominant perfusion and a progressive increase in basilar pulmonary artery caliber on CT, similar to previously described cases (4, 5).

Another contributor to $\dot{\text{V}}$_A/$\dot{\text{Q}}$ mismatch in HPS is impaired hypoxic pulmonary vasoconstriction due to dysfunctional vessel tone and vascular dilatations. Prior studies have demonstrated that in patients with HPS, exposure to a fraction of inspired oxygen of 0.11 does not affect $\dot{\text{V}}$_A/$\dot{\text{Q}}$ matching, while hyperoxia (fraction of inspired oxygen of 1.0) results in paradoxical worsening of $\dot{\text{V}}$_A/$\dot{\text{Q}}$ mismatch (6). These results can be explained by the dysfunctional vessel tone leading to impaired hypoxic pulmonary vasoconstriction at low arterial oxygen pressure and failure to inhibit hypoxic pulmonary vasoconstriction at high arterial oxygen pressure. These responses contrast with the expected improvement in $\dot{\text{V}}$_A/$\dot{\text{Q}}$ matching seen in many disorders of the lung parenchyma in which appropriate initiation or inhibition of hypoxic pulmonary vasoconstriction is intact (2).

Intrapulmonary Vascular Dilatation and Diffusion–Perfusion Limitation

The pulmonary vascular abnormalities seen in HPS result in a diffusion–perfusion limitation that contributes to both reduced Dl_CO and hypoxemia. The angiogenesis that contributes to IPVDs increases the distance that gas diffuses within the dilated pulmonary capillary bed to reach all red cells (2). In addition, although absent in this case, cirrhosis is a hyperdynamic condition associated with increased cardiac output, meaning that red blood cells have shorter transit time for gas exchange to take place as they pass through the pulmonary capillary (7). The combination of these factors results in the diffusion–perfusion limitation of HPS, reducing the amount of gas exchange and limiting full oxygenation.

True Shunt from Arteriovenous Communications

In addition to IPVDs, another vascular abnormality described in patients with HPS is arteriovenous communications (2). When present, pulmonary arterial blood can enter the pulmonary venous circulation without participating in gas exchange. Arteriovenous communications can be visualized as regions of decreased accumulation of labeled, macroaggregated albumin on single-photon emission CT (8). This imaging finding is related to the pass-through of macroaggregated albumin that would otherwise be trapped in the normal pulmonary vessels.

In addition, the reduced perfusion extends into surrounding normal lung tissue, suggesting that a component of vascular steal takes place and contributes to the $\dot{\text{V}}$_A/$\dot{\text{Q}}$ mismatch discussed above (8). These communications are true right-to-left shunts, meaning that the resultant hypoxemia is not correctable with supplemental oxygen (2, 7).

Dead-Space Ventilation as a Cause of Dyspnea

Although many of the above mechanisms played a role in our patient, his degree of exercise-limiting dyspnea was out of proportion to his mild hypoxemia. There are two potential explanations for this. This first is that he may have had an exaggerated hypoxic ventilatory response to mild hypoxemia. Abnormal ventilatory responses to hypoxemia and hypercapnia have been reported in patients with cirrhosis and, together with changes in sex hormones (progesterone), are believed to account for hyperventilation and respiratory alkalosis (2, 9). A second explanation for the patient’s dyspnea was the increased resting and exercise CO₂-dead space demonstrated during CPET. The fact that his CO₂-dead space did not fall with exercise as his tidal volume increased suggested that dead space paradoxically increased with exercise. Presumably, this was related to a relative increase of cardiac output directed to abnormal lower resistance basilar pulmonary arteries during exercise. Prior studies have examined the findings of CPET in patients with HPS and have demonstrated associations with reduced oxygen consumption and increased dead space (10, 11). In lung units with IPVDs, blood flow is increased without relative differences in local ventilation, resulting in a low $\dot{\text{V}}$_A/$\dot{\text{Q}}$ ratio and hypoxemia. Other lung units, however, receive less blood flow as a result, resulting in a greater area that is ventilated but poorly perfused. These high $\dot{\text{V}}$_A/$\dot{\text{Q}}$ areas contribute to increased dead space_, as they did in our patient, leading to a higher total minute ventilation required to achieve the same alveolar minute ventilation. It is also important to consider that both shunt and $\dot{\text{V}}$_A/$\dot{\text{Q}}$ heterogeneity across the entire spectrum from low to high $\dot{\text{V}}$_A/$\dot{\text{Q}}$ regions affect arterial carbon dioxide pressure and therefore physiologic dead space (12). Particularly during exercise, these physiologic impairments manifest as dyspnea as the patient attempts to increase their respiratory rate and tidal volume to compensate for the increased dead space.

Proposed Mechanisms of HPS without Cirrhosis

Current proposed mechanisms for the pathogenesis of IPVDs include the production of or failure to produce vasoactive mediators such as endothelin-1 or BMP9 (bone morphogenic protein 9) within the dysfunctional liver. There may also be impaired hepatic clearance of intestinal bacteria that translocate into the portal circulation (7, 13). These factors combine to promote vasodilation and inflammation that result in the development of IPVDs. Cases of HPS in ischemic hepatitis demonstrate that the underlying liver dysfunction can be due to noncirrhotic causes and, in cases of portosystemic shunt such as Abernethy malformations, that IPVD can develop in the absence of evident liver disease altogether (14, 15). There have been several reports of HPS complicating sarcoidosis with cirrhosis (16, 17). Although our patient did not have cirrhosis, we hypothesize that granulomatous inflammation alone was sufficient to alter the balance of vasoactive mediators and result in IPVD. Given the lack of progression of his symptoms, it is possible that this process was a self-limited case of sarcoidosis and no longer active.

Evaluating Patients with Suspected HPS

In patients with histories of liver dysfunction presenting with dyspnea, hypoxemia, or an elevated A–a Po₂ difference without underlying parenchymal lung disease, HPS should be considered in the differential diagnosis. Screening for this condition should also be performed in liver transplantation candidates, as HPS has implications for candidacy (18). A lack of significant hypoxemia on pulse oximetry is not enough to exclude a diagnosis of HPS (19). Gas-exchange abnormalities can be subtle, and therefore, an arterial blood gas assessment is required, with current criteria defining an A–a Po₂ difference greater than 15 mm Hg (or 20 mm Hg in adults older than 65 yr) as consistent with HPS (20).

Evidence of intrapulmonary vascular dilatation is the other key component of diagnosing HPS. Typically, the initial test to obtain evidence of this intrapulmonary shunt is agitated saline–enhanced TTE. Normally, agitated saline bubbles are unable to pass through the capillaries of the pulmonary circulation and do not appear in the left heart. In the presence of an intrapulmonary shunt, bubbles appear in the left heart three or more cardiac cycles after appearing in the right heart (21). Rarely, transesophageal echocardiography is needed to confirm an intrapulmonary shunt by directly visualizing bubbles in the pulmonary veins (22).

If other comorbid respiratory conditions are present and complicating the diagnosis, additional testing can be pursued to better evaluate for the presence of HPS. CTPA, $\dot{\text{V}}$_A/$\dot{\text{Q}}$ scan, and single-photon emission CT can all be helpful, as discussed above. In addition, macroaggregated albumin lung perfusion scanning has been used to characterize the magnitude of shunt related to HPS (18). Other potentially helpful tests in which the physiology of HPS has been characterized include CPET and measurement of exhaled nitric oxide, which will be elevated in patients with HPS (23).

Treatment in HPS

Studies of numerous therapies for HPS, including pentoxifylline, norfloxacin, and sorafenib, have not demonstrated a clear benefit (7, 24). There are some data in support of using garlic extracts containing allicin, and in rare instances when the vascular pattern of IPVD is amenable, a patient may benefit from coil embolization (20, 25). It is postulated that garlic leads to the production of nitric oxide, which dilates nonaffected blood vessels, diverting blood flow away from the maximally dilated abnormal vessels and resulting in more uniform blood flow throughout the lungs. Other than supplemental oxygen, the only treatment for HPS with clear benefit is liver transplantation, and the condition is now considered an indication for the procedure. After liver transplantation, eventual resolution of HPS is expected (20).

Conclusions

Although HPS has been described in entities other than liver cirrhosis, to our knowledge, this is the first described case of a patient with HPS in the setting of only steatosis and mild fibrosis from hepatic sarcoidosis. The characteristic intrapulmonary vascular dilatations of this condition cause numerous pathophysiological manifestations, including $\dot{\text{V}}$_A/$\dot{\text{Q}}$ mismatch, diffusion–perfusion limitation, and increased CO₂-dead space. An understanding of these principles was critical in explaining the patient’s dyspnea out of proportion to hypoxemia and ultimately in making this uncommon diagnosis.

Answers

• 1.What is the typical evaluation for a patient suspected of intrapulmonary shunting?

  The initial test to evaluate for evidence of an intrapulmonary shunt is agitated saline–enhanced TTE, with bubbles appearing in the left heart three or more cardiac cycles after their appearance in the right heart.

• 2.In HPS, what findings might be expected during cardiopulmonary exercise testing (CPET)?

  Previous studies have demonstrated that compared with patients with normoxemia, patients with HPS exhibit lower peak oxygen consumption, a higher A–a Po₂ difference, and elevated CO₂-dead space without significant differences in breathing reserve, as seen in this case.

• 3.In addition to hypoxemia, what is a major cause of dyspnea in HPS?

  The elevated CO₂-dead space in HPS can be a major driver of dyspnea, as patients must increase their total minute ventilation to maintain their alveolar minute ventilation and CO₂ elimination, particularly during exercise.