Outcomes of Patients After Liver Transplantation
Outcomes of Patients After Liver Transplantation
In this study, we present the largest analysis of liver transplant waitlist candidates with HPS to date. Our findings confirm that waitlisted liver transplant patients with exceptions for HPS can receive transplants with outcomes similar to other liver transplant candidates. However, our observation that transplant recipients with HPS and marked hypoxemia (room-air PaO2 ≤44.0 mm Hg) have worse post-transplantation survival refutes recent reports and demonstrates an association between pre-transplantation room-air oxygenation and post-transplantation mortality. Although these data need to be validated in future cohorts, they suggest that increasing hypoxemia is or can be a marker of worsened post-transplantation outcomes in HPS transplant recipients meeting automatic MELD exception criteria. Candidates with HPS exception points had decreased pre-transplantation mortality compared with non-HPS patients, due to a greater chance of receiving a transplant and better overall survival. This suggests that current exception policy might overprioritize waitlisted HPS patients. Although the increased priority was instituted, in part, to avoid compromising post-transplantation outcomes, striking a balance between pre-transplantation waitlist mortality and post-transplantation outcomes exposes a limitation in current MELD-based prioritization. As the UNOS Liver and Intestine Committee is currently re-examining exception point policies, these data can be used to potentially modify exception policies for waitlist candidates with HPS to better reflect the relationship between room-air oxygenation and pre- and post-transplantation outcomes.
In 2002, liver allocation transitioned from a system that accounted for severity of illness and waiting time, to one based solely on medical urgency using the MELD score. Although the MELD score predicts 90-day pre-transplantation mortality, it is an imperfect measure of overall survival and poorly predicts post-transplantation outcomes or survival benefit. Since the inception of MELD-based allocation, UNOS policy has dictated that liver transplantation candidates with certain conditions are eligible to receive exception points, either through standardized protocols (ie, hepatocellular carcinoma) or a case-by-case peer-review process. The most common indication for MELD exception points is hepatocellular carcinoma, developed based on data demonstrating that tumor burden was strongly associated with post-transplantation outcomes. Therefore, increased waitlist priority is given to ensure expedited transplant to maximize post-transplantation outcomes (Milan criteria). The HPS exception policy was predicated on this philosophy as well. However, although there are robust data with which to derive and validate hepatocellular carcinoma criteria to predict post-transplantation survival, the largest prior study evaluating the association between room-air oxygenation and post-transplantation survival in HPS derived from 75 patients. In addition, the previously defined oxygenation cut off to define higher-risk HPS transplant recipients was developed in a cohort of 24 transplant recipients. Our study provides the most robust data to date on the relationship between pre-transplantation room-air oxygenation and post-transplantation survival in HPS patients.
One earlier report of outcomes of HPS exception patients used OPTN/UNOS data from February 27, 2002 through March 1, 2007 showed similar results. However, our analysis expands the size, depth, and granularity of this work. First, we reviewed each exception narrative in detail, thereby confirming HPS status, allowing us to exclude patients who were misclassified by UNOS coding as having HPS. Second, we extracted detailed data related to HPS in order to evaluate the association between oxygenation and post-transplantation outcomes in the largest cohort of HPS patients. Lastly, our sample size was nearly 4 times larger.
Although the data presented here demonstrate that current exception policy provides access to transplantation for HPS patients and results in excellent post-transplantation outcomes, it raises several issues. First, in contrast to recent reports, we find an association between pre-transplantation room-air oxygenation and post-transplantation survival, with a decline in post-transplantation survival for those with an initial room-air PaO2 ≤44.0 mm Hg. This finding extends results from a small prospective study early in the MELD era and can define a lower limit of PaO2 for awarding HPS MELD exception points. However, these data must be taken in context, as the 5-year post-transplantation patient survival in HPS patients with the lowest values of PaO2 is still at or above a threshold many would consider acceptable for a transplant recipient. Therefore, the transplant community must decide what degree of hypoxemia makes a patient too high risk. The low risk of waitlist dropout, combined with excellent post-transplantation outcomes in those with less severe hypoxemia, suggest that it might be possible to optimize post-transplantation outcomes for patients with HPS without disadvantaging the broader transplant population by decreasing the initial number of exception points for HPS patients, while offering additional priority to those whose PaO2 values decline toward higher-risk values. Such a policy modification would increase pre-transplantation waiting times and delay transplantation, which can result in HPS patients developing progressive hypoxemia, which could adversely impact both pre- and post-transplantation survival. We did find that HPS MELD exception patients waitlisted in "high-MELD" regions (UNOS regions 1, 5, and 9) with longer waiting times had increased pre-transplantation waitlist removal for death or clinical deterioration (14.5% vs 6.8%; P < .001) compared with all other regions. However, these data do not allow us to determine if increased waiting time resulted in progressive hypoxemia, and whether HPS was the cause of pretransplantation death or waitlist removal, and requires additional investigation.
We also find that despite estimates that 5%–15% of liver waitlist candidates have HPS meriting an automatic MELD exception (PaO2 < 60 mm Hg) and up to 30% have HPS overall based on the alveolar-arterial gradient, <2% of waitlist candidates applied for an HPS exception. This observation suggests that HPS remains an under-recognized complication of liver disease, despite its impact on patient survival. Third, although nearly all of the HPS waitlist candidates we analyzed had room-air oxygenation data documented in the OPTN dataset, a substantial number had incomplete data to stringently phenotype HPS (severity of gas exchange abnormalities using the alveolar-arterial gradient, degree of intrapulmonary shunting) or fully characterize pulmonary function. Since 2010, the HPS MELD exception process has been streamlined and applications now require only a PaO2 measure with simple acknowledgement that other criteria for HPS are met (portal hypertension, intrapulmonary shunt, and no evidence of underlying pulmonary disease). These observations underscore important challenges in optimizing the current HPS MELD exception policy.
Our study has limitations. First, we were unable to employ the strict criteria defining HPS used in prospective multi-center studies. However, we are confident that most, if not all, of the patients had HPS based on the data documenting hypoxemia and shunting in nearly 90% of patients. Second, our HPS cohort only included patients who applied and received MELD exception points. This might have led to the inclusion of some HPS patients in the non-HPS cohort. However, we do not believe this explains our results in comparison with those of the Pulmonary Vascular Complications in Liver Disease Study Group, as they analyzed overall survival of patients evaluated for transplantation, only 55% of whom were waitlisted, as opposed to our cohort who, by definition, was waitlisted. Third, room-air oxygenation data were unavailable in approximately 10% of the cohort. The results were unchanged both with exclusion of this cohort and in sensitivity analyses where they were recoded into different oxygenation categories (data not shown). Fourth, we could not definitively determine whether progressive respiratory failure contributed to the increased mortality in the cohort with the lowest PaO2 values. However, HPS transplant recipients with a PaO2 ≤44.0 mm Hg were significantly more likely to die (29.2% [19 of 65] vs 18.2% [102 of 560]; P = .03) and have respiratory failure as the cause of death based on OPTN/UNOS coding when compared with the other HPS transplant recipients who died post-transplantation (36.9% [7 of 19] vs 10.8% [11 of 102]; P = .003). In addition, 4 of 7 deaths attributable to respiratory failure in the lowest PaO2 group occurred within the first 70 days of transplantation. Lastly, we were unable to explain with the available OPTN/UNOS data why the HPS patients with the highest values of PaO2 had increased post-transplantation mortality. These patients might, in fact, have been misclassified as HPS, and instead had other cardiopulmonary disease (eg, chronic obstructive pulmonary disease) causing hypoxia and increased mortality, or they might have had other comorbidities contributing to an increased pre-transplantation mortality that prompted the transplantation center to apply for exception points. However, these possibilities could not be ascertained in OPTN/UNOS data and require additional study.
In conclusion, we find that HPS patients granted MELD exceptions have superior overall survival relative to non-HPS patients due to less waitlist drop out in the pretransplantation period. Post-transplantation survival in HPS vs non-HPS patients is not different, but significantly declines in HPS patients with more severe hypoxemia at initial evaluation based on available OPTN/UNOS data. Re-evaluation of UNOS policy for HPS MELD exceptions might be appropriate to optimize outcomes for patients with HPS without disadvantaging the broader transplantation population. An increase rather than decrease in data collected regarding these patients is needed to guide policy.
Discussion
In this study, we present the largest analysis of liver transplant waitlist candidates with HPS to date. Our findings confirm that waitlisted liver transplant patients with exceptions for HPS can receive transplants with outcomes similar to other liver transplant candidates. However, our observation that transplant recipients with HPS and marked hypoxemia (room-air PaO2 ≤44.0 mm Hg) have worse post-transplantation survival refutes recent reports and demonstrates an association between pre-transplantation room-air oxygenation and post-transplantation mortality. Although these data need to be validated in future cohorts, they suggest that increasing hypoxemia is or can be a marker of worsened post-transplantation outcomes in HPS transplant recipients meeting automatic MELD exception criteria. Candidates with HPS exception points had decreased pre-transplantation mortality compared with non-HPS patients, due to a greater chance of receiving a transplant and better overall survival. This suggests that current exception policy might overprioritize waitlisted HPS patients. Although the increased priority was instituted, in part, to avoid compromising post-transplantation outcomes, striking a balance between pre-transplantation waitlist mortality and post-transplantation outcomes exposes a limitation in current MELD-based prioritization. As the UNOS Liver and Intestine Committee is currently re-examining exception point policies, these data can be used to potentially modify exception policies for waitlist candidates with HPS to better reflect the relationship between room-air oxygenation and pre- and post-transplantation outcomes.
In 2002, liver allocation transitioned from a system that accounted for severity of illness and waiting time, to one based solely on medical urgency using the MELD score. Although the MELD score predicts 90-day pre-transplantation mortality, it is an imperfect measure of overall survival and poorly predicts post-transplantation outcomes or survival benefit. Since the inception of MELD-based allocation, UNOS policy has dictated that liver transplantation candidates with certain conditions are eligible to receive exception points, either through standardized protocols (ie, hepatocellular carcinoma) or a case-by-case peer-review process. The most common indication for MELD exception points is hepatocellular carcinoma, developed based on data demonstrating that tumor burden was strongly associated with post-transplantation outcomes. Therefore, increased waitlist priority is given to ensure expedited transplant to maximize post-transplantation outcomes (Milan criteria). The HPS exception policy was predicated on this philosophy as well. However, although there are robust data with which to derive and validate hepatocellular carcinoma criteria to predict post-transplantation survival, the largest prior study evaluating the association between room-air oxygenation and post-transplantation survival in HPS derived from 75 patients. In addition, the previously defined oxygenation cut off to define higher-risk HPS transplant recipients was developed in a cohort of 24 transplant recipients. Our study provides the most robust data to date on the relationship between pre-transplantation room-air oxygenation and post-transplantation survival in HPS patients.
One earlier report of outcomes of HPS exception patients used OPTN/UNOS data from February 27, 2002 through March 1, 2007 showed similar results. However, our analysis expands the size, depth, and granularity of this work. First, we reviewed each exception narrative in detail, thereby confirming HPS status, allowing us to exclude patients who were misclassified by UNOS coding as having HPS. Second, we extracted detailed data related to HPS in order to evaluate the association between oxygenation and post-transplantation outcomes in the largest cohort of HPS patients. Lastly, our sample size was nearly 4 times larger.
Although the data presented here demonstrate that current exception policy provides access to transplantation for HPS patients and results in excellent post-transplantation outcomes, it raises several issues. First, in contrast to recent reports, we find an association between pre-transplantation room-air oxygenation and post-transplantation survival, with a decline in post-transplantation survival for those with an initial room-air PaO2 ≤44.0 mm Hg. This finding extends results from a small prospective study early in the MELD era and can define a lower limit of PaO2 for awarding HPS MELD exception points. However, these data must be taken in context, as the 5-year post-transplantation patient survival in HPS patients with the lowest values of PaO2 is still at or above a threshold many would consider acceptable for a transplant recipient. Therefore, the transplant community must decide what degree of hypoxemia makes a patient too high risk. The low risk of waitlist dropout, combined with excellent post-transplantation outcomes in those with less severe hypoxemia, suggest that it might be possible to optimize post-transplantation outcomes for patients with HPS without disadvantaging the broader transplant population by decreasing the initial number of exception points for HPS patients, while offering additional priority to those whose PaO2 values decline toward higher-risk values. Such a policy modification would increase pre-transplantation waiting times and delay transplantation, which can result in HPS patients developing progressive hypoxemia, which could adversely impact both pre- and post-transplantation survival. We did find that HPS MELD exception patients waitlisted in "high-MELD" regions (UNOS regions 1, 5, and 9) with longer waiting times had increased pre-transplantation waitlist removal for death or clinical deterioration (14.5% vs 6.8%; P < .001) compared with all other regions. However, these data do not allow us to determine if increased waiting time resulted in progressive hypoxemia, and whether HPS was the cause of pretransplantation death or waitlist removal, and requires additional investigation.
We also find that despite estimates that 5%–15% of liver waitlist candidates have HPS meriting an automatic MELD exception (PaO2 < 60 mm Hg) and up to 30% have HPS overall based on the alveolar-arterial gradient, <2% of waitlist candidates applied for an HPS exception. This observation suggests that HPS remains an under-recognized complication of liver disease, despite its impact on patient survival. Third, although nearly all of the HPS waitlist candidates we analyzed had room-air oxygenation data documented in the OPTN dataset, a substantial number had incomplete data to stringently phenotype HPS (severity of gas exchange abnormalities using the alveolar-arterial gradient, degree of intrapulmonary shunting) or fully characterize pulmonary function. Since 2010, the HPS MELD exception process has been streamlined and applications now require only a PaO2 measure with simple acknowledgement that other criteria for HPS are met (portal hypertension, intrapulmonary shunt, and no evidence of underlying pulmonary disease). These observations underscore important challenges in optimizing the current HPS MELD exception policy.
Our study has limitations. First, we were unable to employ the strict criteria defining HPS used in prospective multi-center studies. However, we are confident that most, if not all, of the patients had HPS based on the data documenting hypoxemia and shunting in nearly 90% of patients. Second, our HPS cohort only included patients who applied and received MELD exception points. This might have led to the inclusion of some HPS patients in the non-HPS cohort. However, we do not believe this explains our results in comparison with those of the Pulmonary Vascular Complications in Liver Disease Study Group, as they analyzed overall survival of patients evaluated for transplantation, only 55% of whom were waitlisted, as opposed to our cohort who, by definition, was waitlisted. Third, room-air oxygenation data were unavailable in approximately 10% of the cohort. The results were unchanged both with exclusion of this cohort and in sensitivity analyses where they were recoded into different oxygenation categories (data not shown). Fourth, we could not definitively determine whether progressive respiratory failure contributed to the increased mortality in the cohort with the lowest PaO2 values. However, HPS transplant recipients with a PaO2 ≤44.0 mm Hg were significantly more likely to die (29.2% [19 of 65] vs 18.2% [102 of 560]; P = .03) and have respiratory failure as the cause of death based on OPTN/UNOS coding when compared with the other HPS transplant recipients who died post-transplantation (36.9% [7 of 19] vs 10.8% [11 of 102]; P = .003). In addition, 4 of 7 deaths attributable to respiratory failure in the lowest PaO2 group occurred within the first 70 days of transplantation. Lastly, we were unable to explain with the available OPTN/UNOS data why the HPS patients with the highest values of PaO2 had increased post-transplantation mortality. These patients might, in fact, have been misclassified as HPS, and instead had other cardiopulmonary disease (eg, chronic obstructive pulmonary disease) causing hypoxia and increased mortality, or they might have had other comorbidities contributing to an increased pre-transplantation mortality that prompted the transplantation center to apply for exception points. However, these possibilities could not be ascertained in OPTN/UNOS data and require additional study.
In conclusion, we find that HPS patients granted MELD exceptions have superior overall survival relative to non-HPS patients due to less waitlist drop out in the pretransplantation period. Post-transplantation survival in HPS vs non-HPS patients is not different, but significantly declines in HPS patients with more severe hypoxemia at initial evaluation based on available OPTN/UNOS data. Re-evaluation of UNOS policy for HPS MELD exceptions might be appropriate to optimize outcomes for patients with HPS without disadvantaging the broader transplantation population. An increase rather than decrease in data collected regarding these patients is needed to guide policy.