In this report, we have shown a 100% peritransplant and 6-month survival, and a 93% 1-year survival for LT in HPS. Although comparisons are limited by the fact that this is a single, small series, these data are similar to recent overall survival statistics for both DDLT and LDLT in the United States (15).
Three particular findings are of potential importance and will be explored. First, overall survival and survival in patients with severe HPS were higher than those reported previously. Second, this is the largest report of adult-to-adult LDLT in HPS, and the only report of LDLT in adults with severe HPS; this appeared to be effective in our patients. Third, biliary and vascular complications were more frequent than those reported in routine LT, implying the need for increased vigilance and strategies to minimize these complications.
Our results differ from those reported by other major centers in the four largest comparable HPS transplant series (reporting at least 15 cases) (7–10) and one literature review (11) published to date (Table 4). Mortality differences were particularly pronounced in subjects with severe hypoxemia. One-year mortality in severe HPS was 14% in our series, compared to a mortality of 67% within the transplant hospitalization (9), 30% at 3 months (11) and ≥40% at 6 months posttransplant (7). Existing data may have (1) discouraged physicians from referring patients with severe HPS for LT, (2) discouraged patients with severe HPS from pursuing LT and (3) formed the basis for transplant refusal in certain severe cases. In this context, our results may have important implications for current practice.
At our centers, there was no apparent selection bias favoring patients with less severe HPS for transplantation, as nontransplanted patients had a higher mean PaO2 than the transplanted cohort. Differences in the severity of HPS also could not account for the improved survival compared to other reports, as our series contained the largest proportion of patients with severe HPS and had the lowest mean PaO2 (51 mmHg). There were no evident differences in demographic or baseline characteristics related to survival, compared to other series (16). Our mean cohort age (51 years) was the highest, and distribution of recipient liver disease was similar to that in the other series, with the exception of a higher and lower prevalence of alcoholic cirrhosis in Taille et al. (52%) and Swanson et al. (8%), respectively, and a higher prevalence of biliary atresia in Krowka et al. (given a large number of pediatric cases), none of which would be expected to alter transplant outcomes significantly (16). The severity of liver disease in our study was similar to that in three of the four other series with which it could be compared (no data on severity of liver disease was available from Krowka et al.). Our patients had a similar mean MELD score (14) to that in Swanson et al.'s study (MELD 13) (7). Although none of the other series reported MELD scores, the proportion of patients in Child's-Pugh (CP) class A was comparable to ours (33%) in Taille et al.'s study (30%) (10) and in Krowka et al.'s 2004 study (24%) (8). Only Arguedas et al. (9) had a much lower percentage of CP class A patients (12%). However, this study noted identical mean CP scores in survivors (9 ± 2) and nonsurvivors (9 ± 1) (9), suggesting no significant relationship between CP class and mortality in this population—a finding confirmed by Swanson et al. (7). Although the MELD scoring system for organ allocation was widely introduced in 2002, this system was not adopted at either of our centers and therefore did influence our results.
Improved survival in this series was seen despite comparable rates of postoperative complications to those in other reporting series. Several factors may account for this. First, our patients may have benefited from medical and technological advances in care. Prior series included transplants performed as remotely as 1968 (11), 1985 (7), 1991 (10) and 1996 (8,9), with most performed in the 1990s and none later than 2002. In contrast, 20/21 transplants in our series were performed in 2004 or later and 15/21 in 2006 or later. Modern advances include changes in LT practices including experience, techniques, surgical materials, perioperative anesthetic care (17,18) and ICU care (19). Accordingly, overall LT survival has improved significantly between 1997 and 2005 (15). Furthermore, improvements in the management of certain particularly common complications in HPS patients may have impacted survival. For example, the widespread use and availability of endosopic retrograde cholangiopancreatography and magnetic resonance cholangiopancreatography may have improved the early diagnosis and management of biliary complications, which were the cause of death in four of five patients in Krowka et al.'s series (8). Also, manipulation of conventional ventilator settings is generally ineffective in the management of severe postoperative hypoxemia in HPS (10), but high-frequency oscillator ventilation has also only recently been used in adult patients (20). This technology was lifesaving in one case of respiratory failure in our series, and respiratory failure accounted for all perioperative mortalities in Taille et al.'s report (10), and for the majority of overall deaths in Arguedas et al.'s report (9) and in Krowka et al.'s review (11).
Second, patients may have benefited from pre- and perioperative care by a specialized HPS clinic and team. Preoperatively, important features may have included optimization of concurrent lung disease and frequent follow-up for oxygen titration. Our data demonstrated a rapid preoperative rate of decline in PaO2 and a high variability in this rate between patients. Given that there are no established clinical predictors of this rate, frequent oxygen titration was used to match increasing oxygen demands with appropriate oxygen therapy to preserve functional status and muscle mass, and to prevent deconditioning. Given that functional status is a significant predictor of surgical survival (16), this may have contributed to favorable outcomes. Other possible consequences of suboptimal pre- and perioperative oxygen delivery that may have been minimized include delayed wound healing, decreased resistance to bacterial infection, biliary and vascular anastamotic ischemia and depressed immune function (10,21–23). Postoperatively, availability of HPS specialists working in conjunction with ICU and transplant specialists may also have been beneficial, as it led to the early use of HPS-specific strategies such as inhaled NO and Trendelenberg positioning for management of refractory postoperative hypoxemia (10,24–27). The key elements of this specialized care model which might be transferable to other centers were: (1) frequent follow-up with oxygen titration and (2) knowledge of HPS-specific strategies to address refractory hypoxemia in the postoperative period, which could be protocolized.
Living donor liver transplantation
Previous English language reports of LDLT in HPS include a series of two adults with moderate HPS (PaO2 69 and 67 mmHg) (28), a report of one child with severe HPS (PaO2 40 mmHg) (29) and a report of one child with moderate HPS (PaO2 50–60 mmHg) (27); all patients demonstrated rapid improvement in oxygenation post-LT. Egawa et al. reported outcomes in 19 children and 2 adults with intrapulmonary shunting who received LDLT, however it is unclear how many of these patients had HPS (gas exchange data not reported). Although all survivors had an improvement in MAA shunt fraction, 1-year mortality was 38% in that series (21). In our series, all five LDLT patients survived the transplant hospitalization (four of five had severe HPS), and LDLT was not associated with an increase in complications. All three LDLT recipients who were followed for >3 months had a resolution in hypoxemia within 3–6 months, which was comparable to DDLT recipients. This suggests that the initially reduced liver mass and hepatic synthetic function in LDLT do not impede reversal of intrapulmonary shunting, and is congruent with previous observations that hepatic synthetic dysfunction is not correlated with baseline HPS severity or incidence (30,31).
Although a longer follow-up period and more patient data will be required, our results may suggest that LDLT is effective in adults with severe HPS. The main advantage of LDLT is that it usually reduces an individual's transplant waiting time. Given (1) the rapidly progressive hypoxemia in HPS (in our report and others (7)); (2) the increased pretransplant mortality in these patients (6), which correlates with the severity of hypoxemia (4,7) and (3) the current median liver transplant waiting time of nearly 1 year (15), LDLT may be a particularly important strategy to consider in HPS patients.
Two of five LDLT (40%) and 6/16 DDLT (38%) recipients had biliary complications (excluding cholestasis alone). These rates were higher than previously reported rates in non-HPS patients in both LDLT and DDLT (24–34% and 10–20%, respectively) (32,33). In patients with HPS, Taille et al. (10) did not report on all biliary complications, but did note a high rate of bile leaks (26%). Increased biliary complications such as anastomotic leaks and stenoses may result from exaggerated tissue hypoxia at the level of the anastamosis in patients with HPS, as has been suggested by other authors (8,21). Although preoperative PaO2 was not a significant predictor of biliary complications, this may have been due to small numbers. Vascular complications were seen in 29% of patients and included caval and hepatic artery anastamotic strictures, which may also have resulted from anastamotic hypoxia; a similarly elevated overall vascular complication rate of 22% was reported by Taille et al. (10).
Overall, these results suggest a need for increased vigilance and a lower threshold for investigation for biliary and vascular complications in the postoperative period. We speculate that strategies to maximize perioperative tissue oxygen delivery, including titrating inspired oxygen to achieve maximum hemoglobin saturation (100% where possible) and maintaining hemoglobin above 100 mg/L (34) might reduce the risk of these complications.
Limitations in this study include its small size, limiting our ability to draw statistically significant conclusions. Referrals to our centers may have been biased in favor of patients who were more likely to be candidates for liver transplant, with less comorbidities. It is unclear whether more severe HPS patients were more likely to be referred to our center given the existence of a specialized clinic and transplant expertise, or less likely to be referred given existing poor transplant outcomes reported in the literature. Within our centers, severe cases were more likely to be recognized and diagnosed with HPS by clinical suspicion. Although these factors prevent any generalization about the distribution of severity of HPS, they do not alter our conclusions about the potential benefits of LT in severe HPS. However, given expertise in LDLT and existence of a dedicated HPS clinic at our centers, generalization to all LT centers should be made with caution.