HPS reportedly occurs in 4–32% of patients with cirrhosis2,9,10,13,17; however, our previous study reported a relatively higher prevalence of 41.5%.16 This discordance may be due to the non-standardized diagnosis method and heterogeneous cohort compositions. HPS has received less attention compared to other cirrhosis-related complications. In general, HPS does not have any symptoms when in a stable state; therefore, clinicians usually do not perform special screening tests or confirmative echocardiography tests for cirrhotic patients who do not show respiratory symptoms. Therefore, in terms of clinical impact, longitudinal data on HPS are insufficient. The present study analyzed long-term follow-up data to determine the prognostic impact of HPS on cirrhosis and ACLF. In the overall cohort, HPS was a significant prognostic factor for short-term survival independent of underlying liver dysfunction. In particular, HPS was more commonly observed in decompensated cirrhotic patients16,20, suggesting that active efforts to identify HPS in decompensated patients can help predict the prognosis in this population.
The exact mechanism by which HPS influences prognosis is not well established. The cause of mortality is multifactorial, while the main causes of death are related to liver-related complications.1,13,21 A previous animal study identified intestinal endotoxemia as an important mechanism in the development of HPS.22,23 Moreover, our previous study reported significantly increased levels of factors in HPS group, such as lipopolysaccharides, nitric oxide, and endothelin, that are related to bacterial translocation.16 In the present study, events related to bacterial translocation (bacteremia, spontaneous bacterial peritonitis [SBP]) were more likely to occur in the HPS group than in the non-HPS group. Seven cases of SBP and four cases of bacteremia were observed in the HPS group, while eight cases of SBP and three cases of bacteremia occurred in the non-HPS group. This difference may explain why ACLF more commonly developed in patients with HPS in our study, as the development of ACLF is closely related to the precipitation of infection.16,24,25 This study showed that HPS was related to a higher risk of ACLF independent of underlying cirrhosis severity. In addition, lung involvement of ACLF was more common in patients with HPS than in those without HPS. Thus, HPS can result in a poor prognosis in patients with cirrhosis via a high incidence of ACLF development with respiratory failure.
In patients with HPS, underlying liver dysfunction estimated using the CTP and MELD scores is related to mortality.1,2,8,10,26,27 This finding was also observed in the development of ACLF; moreover, the underlying severity of cirrhosis was a risk factor for the development of ACLF.25,28 However, the severity of HPS was not a risk factor for poor prognosis in patients with HPS. Contrast-enhanced transthoracic echocardiography with saline is accepted as the most practical method of HPS diagnosis.1,7,19,29 We assessed the HPS grade according to these echocardiographic findings. Our previous study reported a difference in the levels of lipopolysaccharides, nitric oxide, and endothelin according to the echocardiographic severity grade of HPS16. Therefore, we expected to observe a difference in prognosis according to the HPS severity. However, there was no relationship between the HPS severity and mortality and ACLF development in the HPS group. The exact reason for this finding is not clear but suggests the existence of a pulmonary shunt itself, rather than the severity of shunts, as a risk factor for the development of respiratory complications in acute decompensation. Further study on this topic, including pathogenesis, is needed in a well-designed study with a larger population.
The strengths of this study were its prospective observational design and relatively long follow-up period of over 5 years.
The severity of HPS was classified according to echocardiographic findings, and it was evaluated as a prognostic factor. This is the first study to report the relationship between ACLF and HPS.
However, this study has a few limitations. For instance, we were not able to perform repeated serologic evaluations for bacterial translocation; therefore, we could not present data on the direct relationship between HPS, bacterial translocation, and the development of ACLF. In addition, a relatively large number of participants were lost to follow-up.
In conclusion, HPS was a poor prognostic factor for cirrhosis and a risk factor for ACLF. CTP class B/C and MELD score ≥ 18 were also risk factors for mortality and occurrence of ACLF in the long-term follow-up. The HPS group showed more frequent development of infection-related complications and lung involvement of ACLF than that seen in the non-HPS group. Further evaluation of the underlying pathogenesis and related factors is needed through a well-designed prospective study with a larger population.