Utility of Pulse Oximetry Screening for Hepatopulmonary Syndrome

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:749 754 Utility of Pulse Oximetry Screening for Hepatopulmonary Syndrome MIGUEL R. ARGUEDAS, HARPREET SINGH, DOROTHY K. FAULK, and MICHAEL B. FALLON Liver Center, Division of Gastroenterology & Hepatology, Department of Medicine, University of Alabama at Birmingham, Birmingham, Alabama Background & Aims: Hepatopulmonary syndrome is characterized by oxygenation abnormalities caused by intrapulmonary vasodilatation in the setting of liver disease and/or portal hypertension. This syndrome occurs in approximately 15% 30% of cirrhotic patients and influences mortality and transplant candidacy. However, no specific screening guidelines are established. We evaluated pulse oximetry with contrast echocardiography in detecting hepatopulmonary syndrome in a cohort of patients undergoing evaluation for liver transplantation. Methods: One hundred twenty-seven consecutive patients referred for liver transplantation evaluation were prospectively enrolled and underwent pulse oximetry, contrast echocardiography, and arterial blood gas measurements on room air. Demographic, clinical, and laboratory data were recorded and analyzed. Results: Forty-one (32%) patients were found to have hepatopulmonary syndrome. There were no significant differences in demographic or clinical features compared with patients without hepatopulmonary syndrome, with the exception of pulse oximetry and oxygenation abnormalities. With a threshold value of <96%, pulse oximetry had a sensitivity and specificity of 100% and 88%, respectively, for detecting patients with a partial pressure of oxygen <60 mm Hg. Receiver operator characteristic analysis revealed that a pulse oximetry value of <94% detected all patients with a partial pressure of oxygen <60 mm Hg with an increased specificity of 93%. In addition, higher pulse oximetry thresholds reliably identified HPS patients with less severe hypoxemia, albeit with lower specificity. Conclusions: Pulse oximetry is a simple, low cost, and widely available technique that reliably predicts the presence and severity of hypoxemia in patients with hepatopulmonary syndrome. Institution of pulse oximetry screening might enhance detection and improve management of hepatopulmonary syndrome in cirrhosis. The presence of intrapulmonary vascular dilatation resulting in abnormal arterial gas exchange in the setting of liver disease and/or portal hypertension defines hepatopulmonary syndrome (HPS). 1 HPS occurs in approximately 15% 30% of patients being evaluated for orthotopic liver transplantation (OLT). 1,2 Mortality in cirrhotic patients with HPS is significantly increased relative to cirrhotic patients without HPS and is related to the degree of hypoxemia and severity of liver disease. 3,4 Currently, OLT is the only effective treatment for HPS, and survival is better in HPS patients who undergo OLT than in those not undergoing transplantation. 1,3,4 However, post-olt mortality is increased relative to non-hps patients, particularly when hypoxemia is severe or liver disease is advanced. 3 6 On the basis of these findings, patients in the U.S. with HPS and a partial pressure of oxygen (PaO 2 ) 60 mm Hg are eligible for increased priority for OLT (HPS Model for End-Stage Liver Disease [MELD] exception). 7 Currently, there are no specific guidelines for screening for HPS in cirrhotic patients. Clinical features have low sensitivity for detecting the presence of HPS, 2,8 and the diagnosis requires arterial blood gas (ABG) measurements to identify gas exchange abnormalities and microbubble transthoracic contrast echocardiography (CE) to detect intrapulmonary vascular dilatation. 1,9 CE alone is not a useful screening test because 50% 60% of cirrhotic patients have evidence of intrapulmonary vascular dilatation, but only a subset are found to have abnormal arterial gas exchange and fulfill criteria for HPS. 1 Similarly, ABGs are invasive and uncomfortable and are associated with potential risks, limiting their utility as a screening tool. Therefore, a simple and reliable means of detecting hypoxemia in patients with intrapulmonary vascular dilatation could facilitate identification and management of HPS, particularly in those who are potential OLT candidates and would benefit from HPS MELD exception. In prior work, we found that digital pulse oximetry (oxygen percent saturation [SpO 2 ]) performed with a standardized protocol in a pulmonary function laboratory at the time of arterial blood analysis was a useful tool for detecting hypoxemia in cirrhosis. Use of an SpO 2 threshold of 96% or less correctly identified all patients with a PaO 2 60 mm Hg. 10 However, these findings have not been validated in an outpatient clinic setting, and the utility of routine pulse oximetry as a screening test for HPS with hypoxemia has not been established. The present study was undertaken to prospectively evaluate pulse oximetry with CE for detecting HPS with hypoxemia in a cohort of patients with cirrhosis undergoing evaluation for OLT. Materials and Methods Subjects and Data Collection We performed a prospective cohort study of 127 consecutive patients with cirrhosis referred for initial liver transplant evaluation to the University of Alabama at Birmingham Liver Center outpatient clinic between November 2004 Sep- Abbreviations used in this paper: ABG, arterial blood gas; AUC, area under the curve; CE, contrast echocardiography; CI, confidence interval; HPS, hepatopulmonary syndrome; MELD, Model for End-Stage Liver Disease; OLT, orthotopic liver transplantation; P(A-a)O 2, alveolararterial oxygen gradient; PaO 2, partial pressure of oxygen; PCO 2, partial pressure of carbon dioxide; PFTs, pulmonary function tests; ROC, receiver operating characteristic; SD, standard deviation; SpO 2, oxygen percent saturation. 2007 by the AGA Institute 1542-3565/07/$32.00 doi:10.1016/j.cgh.2006.12.003

750 ARGUEDAS ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 6 tember 2005. Cirrhosis was defined histologically or by a combination of characteristic clinical, laboratory, and radiologic findings. Demographic, clinical, and laboratory data were recorded. SpO 2 measurements were performed by clinic staff during the initial clinic evaluation by using a Datascope Passport pulse oximeter (Datascope Corp, Paramus, NJ) applied to the index finger, with the patient in the sitting position. ABG measurements on room air (sitting position) were obtained in the Pulmonary Function laboratory, and CE was performed by using an established protocol. 9,11 ABGs and CE were performed within 72 hours of SpO 2 measurements, and personnel performing and interpreting each of the studies were blinded to the results of the others. HPS was defined by (1) a positive contrast echocardiogram consistent with intrapulmonary shunting, (2) an alveolar-arterial oxygen gradient [P(A-a)O 2 ] value greater than 20 mm Hg, 12,13 and (3) lack of significant findings on chest radiography or pulmonary function tests (PFTs) of sufficient severity to explain arterial oxygenation abnormalities. All study subjects were at least 18 years of age and were not selected on the basis of sex, race, socioeconomic status, and etiology of liver disease. Demographic, clinical, and laboratory data were recorded and incorporated into a database. The Institutional Review Board at the University of Alabama at Birmingham approved the study. Data Analysis Descriptive data are expressed as proportions, and quantitative data are summarized as mean standard deviation (SD) or median and interquartile range, as appropriate. Comparisons between groups for normally distributed quantitative data were performed with a Student t test or one-way analysis of variance with Bonferroni correction, as appropriate. For non-normally distributed data, the Mann-Whitney U test was used. Qualitative data were analyzed by 2 analysis or Fisher exact test, as appropriate. Correlations were obtained by using either the Pearson or Spearman tests as appropriate. We assessed the ability of pulse oximetry to detect hypoxemia in patients with positive CEs by calculating the area under the curve (AUC) of receiver operating characteristic (ROC) curves by using nonparametric methods. The diagnostic value of pulse oximetry was assessed by calculating sensitivity and specificity. For all comparisons, statistical significance was defined as a P value less than.05. All statistical work was performed by using specialized software (SPSS 11.5; SPSS Inc, Chicago, IL). Results Baseline Characteristics Our cohort consisted of 127 patients, one of whom had been diagnosed previously with HPS and referred to our center for this reason. The mean age of the cohort was 53 9 years, 75 (59.5%) were male, and 120 (95%) were white. The most common etiologies of liver disease included hepatitis C (46%), cryptogenic/nonalcoholic steatohepatitis (21%), and ethyl alcohol liver disease (19%). The average MELD score of the cohort was 14 5, and according to Child Pugh classification, 21% of patients were Child Pugh class A, 69% were Child Pugh class B, and 10% were Child Pugh class C. Significant findings on chest radiographs were reported in 12 patients (9%). Five of these patients had HPS and 7 did not. These findings were moderate to large pleural effusions (4 patients), hyperinflation (4 patients), and prominent interstitial markings (4 patients). PFTs revealed obstructive lung disease (forced expiratory volume in 1 second/forced vital capacity 70%) in 17 patients. Fifteen of these patients had mild obstruction, and 2 patients had moderate disease according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria. 14 Results of Hepatopulmonary Syndrome Evaluation Figure 1 outlines the results of the evaluation for HPS in our cohort by using pulse oximetry in conjunction with CE. Specifically, 85 patients (67%) had evidence of intrapulmonary shunting as evidenced by a positive CE. Of these patients, 68 (80%) had an SpO 2 96%, all of whom had a PaO 2 60 mm Hg. Twenty-five of these patients (37%) had a P(A-a)O 2 20 mm Hg, thereby fulfilling criteria for the diagnosis of HPS. Of these Figure 1. Results of testing for HPS in transplant candidates. PaO 2 and P(A-a)O 2 values are in mmhg. *Patient with porlopulmonary hypertension.

June 2007 PULSE OXIMETRY IN HPS 751 individuals, only one had an abnormal chest radiograph (prominent interstitial markings), and none had abnormalities on PFTs. The remaining 43 patients (63%) had a mean PaO 2 level of 95 12 mm Hg and a normal P(A-a)O 2 (mean SD, 8 7 mm Hg). Four patients had radiographic abnormalities (pleural effusions 3, hyperinflation 1), and 7 patients had evidence of obstruction on PFTs (mild 6, moderate 1). Seventeen patients with a positive CE (20%) had an SpO 2 96%. Nine of these patients (53%) had a PaO 2 level 60 mm Hg. All these patients fulfilled criteria for HPS. Chest radiographs revealed prominent interstitial markings in one patient and hyperinflation in another. PFTs revealed mild obstruction in this latter patient. All of the remaining 8 patients (47%) had a PaO 2 level of 60 mm Hg and a P(A-a)O 2 20 mm Hg. Seven of these patients fulfilled criteria for HPS, with 2 patients having prominent interstitial markings on chest radiography. One patient had mild obstruction on PFTs. The remaining patient was found to have portopulmonary hypertension at right heart catheterization. Forty-two patients (33% of the total cohort) had a negative contrast echocardiogram. Of these patients, 38 (90%) had an SpO 2 96%, all of whom had a normal PaO 2 and P(A-a)O 2 values on ABG analysis. Chest radiographs revealed a pleural effusion in one patient, whereas another had hyperinflation. PFTs revealed obstruction in 6 patients (mild 5, moderate 1). Four patients (10%) had an SpO 2 96%, all of whom had no or mild blood gas exchange abnormalities on ABG testing. Chest radiographs revealed hyperinflation in 1 patient, and PFTs revealed mild obstruction in 2 patients. Group Comparisons In total, 41 patients (32%) fulfilled diagnostic criteria for HPS, and their characteristics are presented in Supplementary Table 1. (Supplementary material available online only at www.cghjournal.org). Table 1 summarizes the clinical, biochemical, and blood gas analysis data of patients with HPS compared with patients without HPS. Patients with HPS were older than patients without HPS (56 vs 52 years, P.009). Otherwise, there were no significant differences in basic demographics (sex, race), severity or etiology of cirrhosis, or in the prevalence of abnormal chest radiographs or PFTs between HPS and non-hps patients. Dyspnea on exertion was present in 63% of patients with HPS compared with 59% in non-hps patients (P.85). Cyanosis was observed in 7% of HPS patients but was absent in non-hps patients (P.03). Digital clubbing was present in 22% of patients with HPS compared with 5% of patients without HPS (P.005), whereas spider angiomas were present in 39% and 38% of HPS and non-hps patients, respectively (P.85). In patients with HPS, SpO 2, PaO 2, and partial pressure of carbon dioxide (PCO 2 ) were significantly lower and the P(A-a)O 2 was significantly higher than in patients without HPS. Figure 2 demonstrates the distribution of PaO 2 values in patients with HPS. Seven of 41 (17%) patients with HPS had a PaO 2 60 mm Hg (mean SD, 48 10 mm Hg), which would make them eligible for HPS MELD exception. When this patient subgroup was compared with those patients with HPS with a PaO 2 60 mm Hg, there were no statistically significant differences in demographic or clinical parameters, with the exception of having significantly more pronounced hypoxemia (48 vs 76 mm Hg, P.001) and a wider P(A-a)O 2 (58 vs 29 mm Hg, P.001). Similarly, patients with HPS and a PaO 2 60 mm Hg were not significantly different compared with patients without HPS in demographic or clinical parameters, with the exception of SpO 2 (89% vs 98%, P.001), PaO 2 (48 vs 91 mm Hg, P.001), and a wider P(A-a)O 2 (58 vs 11 mm Hg, P.001). Predictive Value of Oxygen Percent Saturation Measurements We evaluated the ability of SpO 2 measurements to predict a PaO 2 60 mm Hg and the presence of HPS in the setting of a positive CE. There was a significant positive correlation between SpO 2 and PaO 2 values (r.66, P.0001). With an SpO 2 cutoff value of 96% based on the results of prior studies, 10 the sensitivity and specificity for detecting patients with a PaO 2 60 mm Hg were 100% (95% confidence interval [CI], 68% 100%) and 88% (95% CI, 84% 90%), respectively. With this same cutoff value to detect all patients with a PaO 2 70 mm Hg, the sensitivity and specificity were 79% (95% CI, 56% 92%) and 91% (95% CI, 88% 93%), respectively. Finally, to detect all patients with HPS, the sensitivity and specificity of this cutoff were 40% (95% CI, 31% 58%) and 95% (95% CI, 89% 98%). We also assessed different SpO 2 levels and calculated the sensitivity and specificity for predicting hypoxemia and HPS. To do so, we constructed ROC curves to evaluate the AUC to define useful SpO 2 cutoffs to detect the presence of HPS (Figure 3). According to this analysis, using a threshold pulse oximetry value of 100% detected all patients with HPS (sensitivity 100%, specificity 19%), with an AUC of 0.76 (95% CI, 0.67 0.85). To detect patients with HPS and a PaO 2 70 mm Hg, the ROC-derived SpO 2 threshold of 97% was associated with a sensitivity of 100% and a specificity of 65%, with an AUC of 0.94 (95% CI, 0.89 0.99). Finally, to detect patients with HPS and a PaO 2 60 mm Hg, the ROC-derived SpO 2 threshold of 94% was associated with a sensitivity of 100% and a specificity of 93%, with an AUC of 0.99 (95% CI, 0.97 1.00). Table 2 summarizes the test performance characteristics (sensitivity and specificity) of different SpO 2 cutoff values in the diagnosis of HPS. Discussion HPS is an important complication of cirrhosis and portal hypertension that is associated with adverse patient outcomes. 3,4 OLT is the only effective treatment for HPS, but mortality is increased after transplantation in HPS patients relative to those without HPS, particularly when hypoxemia is severe. 5 Therefore, patients with HPS and significant hypoxemia are eligible for increased priority for OLT. However, there are no established screening guidelines for HPS in cirrhotic patients. The present study was undertaken to define the utility of pulse oximetry, in the setting of a positive CE, in detecting hypoxemic patients with HPS. A cutoff SpO 2 value 96% detects all patients with HPS and hypoxemia (PaO 2 70 mm Hg) with a specificity of 88%. ROC analysis demonstrates that using an SpO 2 94% also detects all patients with HPS and a PaO 2 60 mm Hg who would be candidates for MELD exception related to HPS with a high specificity (93%). In addition, we found that different SpO 2 cutoff values reliably identified varying degrees of severity in patients with HPS. These results demonstrate that pulse oximetry is a simple, low cost, and reliable screening tool for the detection of HPS.

752 ARGUEDAS ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 6 Table 1. Comparison Between HPS patients and Non-HPS Patients Variable HPS (n 41) Non-HPS (n 86) P value Age (mean SD, y) 56 8 52 8.005 Gender (male, %) 49 65.06 Race (white, %) 95 100.2 Etiology of cirrhosis (viral, %) 45 49.57 MELD (mean SD) 14 4 14 5.91 Child-Pugh class (A/B/C, %) 15/73/12 24/66/10.43 Abnormal chest x-ray (%) 12 8.13 Abnormal PFTs (%) 5 17.06 SpO 2 (%) 95 4 98 2.001 PaO 2 (mean SD, mm Hg) 71 14 91 13.001 P(A-a)O 2 (mean SD, mm Hg) 35 15 11 9.001 PCO 2 (mean SD, mm Hg) 32 5 35 5.002 Pulse oximetry is a well-established method for noninvasive evaluation of arterial oxygenation. 15 Numerous studies have found that it is accurate and reliable for assessing arterial oxygenation in patients without liver disease, and a recent study has found similar results in cirrhotic patients. 10,15 In both situations, SpO 2 might overestimate oxygen saturation measured directly on ABG analysis by between 1.5% 3.5% and in cirrhotic patients resulted in a higher SpO 2 than might be expected to trigger evaluation with ABG to detect hypoxemia. In cirrhotic patients, the overestimation of oxygen saturation by SpO 2 was similar to that in patients without liver disease and was not influenced by severity of liver disease or bilirubin levels, supporting that SpO 2 is of similar utility in patients with and without liver disease. 10 Our results in an outpatient cirrhosis clinic setting found similar SpO 2 thresholds to trigger obtaining an ABG as previously observed in cirrhotic patients, thereby validating the utility of pulse oximetry for detecting hypoxemia in this setting. In addition to validating the utility of pulse oximetry in patients with cirrhosis, we demonstrated that by using a simple SpO 2 -based screening algorithm, we were able to detect all Figure 3. ROC curves of SpO 2 measurements for the detection of varying degrees of severity of HPS. patients with HPS of sufficient severity to be eligible for increased priority for OLT by performing ABGs when the SpO 2 was 94% in patients with positive CEs. SpO 2 measurements were obtained during routine vital sign assessment by liver clinic staff without specialized training and blinded to the study, supporting that our results and approach are generalizable to a variety of outpatient clinic settings. Using an SpO 2 of 94% to trigger further evaluation would have required ABG determinations in 12.6% of the cohort. In addition, by using higher SpO 2 thresholds, we were able to reliably identify HPS patients with less severe hypoxemia as well, albeit with lower specificity and the need for ABG determinations in a greater proportion of the cohort. One of the important implications of this work is the demonstration that SpO 2 measurement provides a widely available and objective means for screening for the presence and severity of HPS. From a practical perspective, Table 2. Test Performance Characteristics of Oxygen percent saturation (SpO 2 ) Measurements as a Screening Tool for HPS SpO 2 positive to All patients (sensitivity [%]/ specificity [%] PaO 2 70 mm Hg (sensitivity [%]/ specificity [%] PaO 2 60 mm Hg (sensitivity [%]/ specificity [%] Figure 2. PaO 2 values in patients with HPS. 90% 7/100 21/100 43/100 91% 10/100 29/100 57/100 92% 19/99 43/99 86/98 93% 21/98 50/98 86/97 94% 29/95 57/93 100/93 95% 40/95 79/91 100/88 96% 50/89 93/85 100/81 97% 64/68 100/65 100/61 98% 88/44 100/37 100/35 99% 97/19 100/15 100/14 100% 100/0 100/0 100/0

June 2007 PULSE OXIMETRY IN HPS 753 clinicians can use this information to determine which patients to target for further evaluation. In the current study, we performed CE in all patients undergoing OLT evaluation to define the utility of SpO 2 measurements as a screening tool for HPS. Practice guidelines recommend that all OLT candidates undergo standard echocardiography to screen for portopulmonary hypertension. 16 Our data support that one screening approach for HPS would be to perform SpO 2 measurements during initial evaluation and then perform either standard echocardiography or CE with ABG, depending on whether the SpO 2 value was low. The SpO 2 value chosen as a threshold for performing CE and ABG will determine the number of contrast and ABG studies done and the severity of hypoxemia that would be detected. On the basis of the adverse effects of HPS on survival in patients being considered for OLT, 3,4 our current practice is to perform CE and ABGs in patients found to have an SpO 2 97% to detect all HPS patients with a PaO 2 70 mm Hg. This approach permits us to consider oxygen therapy in those who desaturate with exertion and more closely monitor for progression of HPS and would result in performing CE in 46% of patients being evaluated for OLT. An alternate approach would be to perform CE and ABGs only in patients found to have an SpO 2 94% to target detection to HPS patients with a PaO 2 60 mm Hg who would be eligible for MELD exception for HPS. Such an approach would result in 13% of patients undergoing CE and ABGs rather than standard echocardiography. One factor that might influence decision making is the specific cost associated with contrast administration and CE interpretation, which is not standardized between centers. It is important to recognize that our results can only be generalized to centers at or near sea level because PaO 2 and SpO 2 values might be significantly influenced by altitude, and we have not validated results at higher altitudes. In addition, our cohort of patients undergoing OLT evaluation does not include a large number of patients with significant hypoxemia caused by underlying cardiopulmonary disease. As mentioned previously, prior studies support that SpO 2 is a useful tool for detecting hypoxemia in such patients without liver disease, 15 and in our cohort, SpO 2 was useful in detecting hypoxemia of varying severity in HPS patients. In addition, inclusion of a larger number of patients with intrinsic cardiopulmonary disease in our cohort would not be anticipated to alter the sensitivity of SpO 2 screening for HPS but could result in a lower specificity. Nonetheless, detecting non-hps hypoxemic patients would also be important, particularly if OLT is being considered. The prevalence of HPS in our cohort undergoing routine screening was 32%, a value within the range observed previously. 1 In prior studies, the prevalence of HPS has been found to vary on the basis of the criteria used to define arterial oxygen abnormalities. 17 We used a P(A-a)O 2 of 20 mm Hg as abnormal on the basis of prior work. 5 Among patients with HPS, 17% had a PaO 2 60 mm Hg, 34% had a PaO 2 70 mm Hg, and 68% had a PaO 2 80 mm Hg. On the basis of limited data supporting that oxygen values decline on average 5 mm Hg per year in patients with HPS, 4 17% (7/41) of our HPS patients had a PaO 2 between 60 70 mm Hg and would be at risk of deteriorating to less than 60 mm Hg during a 1- to 2-year time frame. Patients in this group, who are OLT candidates, should be considered for surveillance for worsening hypoxemia. Whether serial SpO 2 measures are sufficiently sensitive to detect worsening oxygenation is an area for future study, although we found a significant positive correlation between SpO 2 and PaO 2 values in our cohort. The clinical characteristics in our patients with HPS were similar to those without HPS and are also in line with a number of previous studies. 4,18 20 Specifically, we found that HPS occurred across a spectrum of etiologies and severities of liver disease and was not more common in advanced liver disease. The observation that HPS is not more common in advanced disease is a consistent finding in studies from the U.S., but this differs from several European centers. 3,21 The cause for this observation remains unclear but might reflect different geographic referral or screening patterns. Although chest radiograph abnormalities were more common in HPS patients, 4 of the 5 HPS patients with abnormalities had increased lower lobe interstitial markings in the absence of significant PFT abnormalities, a finding commonly observed and attributed to vascular engorgement in patients with established HPS. 22 In addition, we found that clinical features including the presence of dyspnea, cyanosis, clubbing, and spider angiomas were poorly sensitive for detecting severity of hypoxemia and HPS in our cohort. Similar results for several of these clinical features have been previously reported, 8,17 supporting the need for an accurate and reliable means for screening for hypoxemia in HPS. In summary, the current study reveals that clinically significant HPS is common in patients undergoing OLT evaluation and demonstrates that pulse oximetry is a simple, low cost, and widely available screening test for HPS in this patient group. SpO 2 measurements accurately predict the severity of hypoxemia in patients with HPS and, in doing so, appropriately target further diagnostic (CE and ABGs) or therapeutic interventions, such as HPS MELD exception for OLT candidates. Together, these findings support that the implementation of pulse oximetry screening in patients being evaluated for OLT and in cirrhotic patients in Gastroenterology/Hepatology clinics might improve the detection and management of HPS. Supplementary Data Note: To access the supplementary materials accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at www.cghjournal.org. References 1. Rodriguez-Roisin R, Krowka MJ, Herve P, et al. Pulmonary-hepatic vascular disorders (PHD). Eur Respir J 2004;24:861 880. 2. Fallon M, Abrams G. Pulmonary dysfunction in chronic liver disease. Hepatology 2000;32:859 865. 3. Schenk P, Schoniger-Hekele M, Fuhrmann V, et al. Prognostic significance of the hepatopulmonary syndrome in patients with cirrhosis. Gastroenterology 2003;125:1042 1052. 4. Swanson K, Wiesner R, Krowka M. Natural history of hepatopulmonary syndrome: impact of liver transplantation. Hepatology 2005;41:1122 1129. 5. Arguedas M, Abrams GA, Krowka MJ, et al. Prospective evaluation of outcomes and predictors of mortality in patients with hepatopulmonary syndrome undergoing liver transplantation. Hepatology 2003;37:192 197. 6. Taille C, Cadranel J, Bellocq A, et al. Liver transplantation for hepatopulmonary syndrome: a ten-year experience in Paris, France. Transplantation 2003;79:1482 1489. 7. United Network for Organ Sharing: 2006. Available at: http:// www.unos.org. Accessed October 12, 2006.

754 ARGUEDAS ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 6 8. Martinez G, Barbera J, Visa J, et al. Hepatopulmonary syndrome in candidates for liver transplantation. J Hepatol 2001;34:756 758. 9. Abrams GA, Jaffe CC, Hoffer PB, et al. Diagnostic utility of contrast echocardiography and lung perfusion scan in patients with hepatopulmonary syndrome. Gastroenterology 1995;109:1283 1288. 10. Abrams GA, Sanders MK, Fallon MB. Utility of pulse oximetry in the detection of arterial hypoxemia in liver transplant candidates. Liver Transpl 2002;8:391 396. 11. Arguedas MR, Drake BB, Kapoor A, et al. Carboxyhemoglobin levels in cirrhotic patients with and without hepatopulmonary syndrome. Gastroenterology 2005;128:328 333. 12. Harris E, Kenyon A, Nisbet H, et al. The normal alveolar-arterial oxygen-tension gradient in man. Clin Sci Mol Med 1974;46:89 104. 13. Harris E, Seelye E, Whitlock R. Revised standards for normal resting dead-space volume and venous admixture in men and women. Clin Sci Mol Med 1978;55:125 128. 14. Pauwels R, Buist A, Ma P, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: National Heart, Lung, and Blood Institute and World Health Organization Global Initiative for Chronic Obstructive Lung Disease (GOLD) executive summary. Repir Care 2001;46: 798 825. 15. Jensen L, Onyskiw J, Prasad N. Meta-analysis of arterial oxygen saturation monitoring by pulse oximetry in adults. Heart Lung 1998;27:387 408. 16. Murray K, Carithers R. AASLD practice guidelines: evaluation of the patient for liver transplantation. Hepatology 2005;41:1407 1432. 17. Schenk P, Fuhrmann V, Madl C, et al. Hepatopulmonary syndrome: prevalence and predictive value of various cut offs for arterial oxygenation and their clinical consequences. Gut 2002; 51:853 859. 18. Abrams G, Nanda N, Dubovsky E, et al. Use of macroaggregated albumin lung perfusion scan to diagnose hepatopulmonary syndrome:a new approach. Gastroenterology 1998;114:305 310. 19. Krowka M, Wiseman G, Burnett O, et al. Hepatopulmonary syndrome: a prospective study of relationships between severity of liver disease, Pao2 response to 100% oxygen, and brain uptake after 99mTc MAA lung scanning. Chest 2000;118:615 624. 20. Whyte M, Hughes J, Peters A, et al. Analysis of intrapulmonary right to left shunt in hepatopulmonary syndrome. J Hepatol 1998; 29:85 93. 21. Vachiery F, Moreau R, Hadengue A, et al. Hypoxemia in patients with cirrhosis: relationship with liver failure and hemodynamic alterations. J Hepatol 1997;27:492 495. 22. McAdams HP, Erasmus J, Crockett R, et al. The hepatopulmonary syndrome: radiologic findings in 10 patients. Am J Roentgenol 1996;166:1379 1385. Address requests for reprints to: Michael B. Fallon, MD, MCLM 280, 1918 University Blvd, Birmingham, Alabama 35294. e-mail: mfallon@ uab.edu; fax: (205) 975-9777. Supported by RO3 DK065958 (M.B.F.).

754.e1 ARGUEDAS ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 6 Supplementary Table 1. Demographic, Clinical, and Biochemical Characteristics of HPS Patients Patient no. Age (y) Gender Etiology Child Pugh class MELD SpO 2 (%) PaO 2 (mm Hg) P(A-a)O 2 (mm Hg) CXR FEV 1 /FVC (%) 1 55 F HCV C 16 92 57 44 NL NA 2 43 M HHC B 11 88 35 79 NL 67 3 39 M Alcohol B 14 91 49 64 NL 71 4 56 F PBC A 6 94 56 48 NL 71 5 55 F Cryptogenic B 19 90 52 59 NL 75 6 68 F Cryptogenic B 14 80 33 64 Interstitial 80 7 51 F HCV C 20 92 57 51 Interstitial 86 8 68 F HCV B 8 95 66 41 Hyperinflation 69 9 63 F Alcohol/AIH B 22 93 61 51 NL NA 10 53 M NASH B 10 97 68 50 Interstitial NA 11 52 M HCV B 7 96 62 54 NL 71 12 65 F HCV B 16 95 69 34 NL 74 13 66 F HCV B 8 96 63 39 NL 80 14 51 M HCV/alcohol B 13 95 65 49 Interstitial 84 15 44 M HCV/alcohol B 13 99 81 23 NL 76 16 58 M HHC B 19 99 80 33 NL 73 17 57 M HCV/alcohol B 13 98 71 35 NL NA 18 52 M Cryptogenic A 9 98 81 27 NL 58 19 47 M HCV/alcohol A 10 97 74 25 NL 69 20 54 F Alcohol B 23 97 76 31 NL 72 21 45 M HCV/alcohol B 14 98 76 28 NL 75 22 65 F PBC B 17 94 72 41 NL 80 23 46 F HCV B 10 92 71 30 NL 80 24 51 F NASH A 8 98 76 32 NL 82 25 49 F HCV B 10 97 75 27 NL 89 26 65 M Alcohol B 14 94 94 23 NL 84 27 67 M Cryptogenic B 19 95 73 23 NL 76 28 70 M HHC B 20 95 83 21 NL 72 29 47 M HCV B 13 99 79 20 NL 79 30 47 M 1 -Antitrypsin deficiency B 12 98 85 22 NL 80 31 48 F HCV B 17 97 86 21 NL 73 32 55 F HCV A 12 98 89 20 NL 79 33 56 M Cryptogenic B 14 98 80 26 NL 81 34 57 M Alcohol B 14 97 84 20 NL NA 35 60 M HCV/alcohol B 14 98 74 23 NL NA 36 63 F NASH C 19 98 77 22 NL NA 37 63 F Alcohol C 17 96 84 20 NL 76 38 64 F HCV B 8 96 84 22 NL 80 39 64 F HCV C 18 100 76 22 NL 76 40 64 F HCV B 12 98 87 21 NL 77 41 65 F HCV B 15 99 79 24 NL 85 CXR, chest radiograph; HHC, hereditary hemochromatosis; PBC, primary biliary cirrhosis; AIH, autoimmune hepatitis; NASH, nonalcoholic steatohepatitis; FEV 1, forced expiratory volume in 1 second; FVC, forced vital capacity; NL, normal; NA, not available.