|Home | About | Journals | Submit | Contact Us | Français|
The effect of occlusive portal vein thrombosis (PVT) on the mortality of pediatric liver transplant candidates and recipients is poorly defined. Using standard multivariable techniques, we studied the relationship between PVT and waiting list and post-transplant survival using data from the Scientific Registry of Transplant Recipients between September 2001 and December 2007. A total of 5,087 candidates and 3,630 liver transplant recipients were evaluated during the period. PVT was found in 1.4% (n = 70) of liver transplant candidates and 3.7% (n = 136) of recipients. PVT was not associated with increased waiting list mortality (HR = 0.9; 95% CI, 0.8–1.1, p = 0.35). Conversely, PVT patients had significantly lower unadjusted survival in the post-transplant period (p = 0.01). PVT was independently associated with increased post-transplant mortality in multivariate models (HR = 2.9; 95%CI 1.6–5.3, p = 0.001 at 30 days and HR = 1.7; 95% CI, 1.1–2.4; p = 0.01 overall survival). The presence of PVT in pediatric liver candidates was not associated with increased waiting list mortality but was clearly associated with post-transplant mortality, especially in the immediate post-operative period.
Portal vein thrombosis (PVT) is a common sequelae of chronic liver disease in both adult and pediatric patients (1–4). Architectural changes in liver parenchyma secondary to chronic liver disease increase the resistance to portal vein blood flow resulting in venous stasis. Susceptibility to thrombogenesis is increased by acquired or inherited hypercoagulopathies (2, 5, 6). While patients with PVT may present with acute hepatic decompensation, it is more often diagnosed as an incidental finding during pre-transplant radiographic evaluation or during the liver transplant operation (1, 7). Occlusive PVT leads to severe portal hypertension and significant morbidity related to ascites, hepato-hydrothorax, hypersplenism, or variceal bleeding. Improvements in prophylactic management of esophageal varices have reduced the risk of bleeding, but these cirrhosis-related complications are particularly difficult to manage in children (3, 5, 8–11).
In spite of these complications, the natural history of PVT in children is poorly described and is not well understood. The current literature consists primarily of single-center reports with a wide variation in PVT incidence, which is as high as 10% in candidates and 16% in recipients (12, 13). Our previous work suggests that PVT is associated with increased mortality in adult liver transplant recipients, but the outcomes remain less clear among pediatric liver transplant candidates. (14–16). It is critical to understand the complex relationship between PVT and survival in both the waiting list and post-liver pediatric populations. This knowledge would optimize the timing of liver transplantation in children at risk for devastating complications. In short, large multi-center studies or clinical registry data are needed to best understand and care for these complex patients. Within this context, we have studied PVT in both pediatric liver transplant candidates and recipients using a national sample, derived from data from the Scientific Registry of Transplant Recipients (SRTR). We hypothesized that pediatric patients with PVT would have an increased risk of mortality both on the waitlist and following liver transplantation. In this case, transplant pediatricians may want to consider policies that facilitate early transplant for these complex patients, while still considering the implications of PVT on post-transplant survival.
Data were obtained from the Scientific Registry of Transplant Recipients (SRTR) and based on patient-level data submitted by transplant centers in the United States to the Organ Procurement and Transplantation Network (OPTN). All liver transplant candidates initially wait-listed at age < 18 years between September 2001 and December 2007 were included in the study cohort.
In the SRTR database, PVT status is reported at two different clinical events. It is reported for liver transplant candidates (recorded as of the time of listing) and for transplant recipients (recorded as of the time of transplant). As in our previous works, for the purposes of evaluating wait list mortality, the PVT field from the candidate file was used.(13, 14, 17) For analysis involving transplant recipients, the PVT field from the recipient file was used. As we have previously described, the PVT fields in the candidate and recipient files frequently did not correlate. For example, PVT was noted in 70 liver transplant candidates and 136 liver transplant recipients. This is likely for two reasons. First, portal vein thrombosis is frequently diagnosed at the time of liver transplantation. Secondly, even if portal vein thrombosis was diagnosed pre— transplant, data that is recorded in the candidate history files are rarely updated by transplant centers, with the exception of the variables that change the MELD/PELD score. It would not be clinically valid to use data from the recipient file for wait list outcome measurements if the goal of the paper is to make clinical policy recommendations, thus we did not specifically make adjustments when the two PVT covariates were not in agreement.
Each candidate was observed until death, censoring only for the earliest of living donor liver transplant, loss to follow-up, or the end of the observation period (December 31, 2007). If candidates were removed from the waitlist they were assumed to have died.
For the purposes of descriptive analysis, the study cohort of candidates was divided into two groups: liver transplant candidates with PVT (PVT candidates) and without PVT (non-PVT candidates). Univariate comparisons were made between the PVT and non-PVT groups for all candidates. A Kaplan-Meier survival curve was created to compare survival among liver transplant candidates with and without PVT, and statistical comparisons used the log-rank test. For the longer term follow up in the Kaplan-Meier curve in candidates with PVT, we do not have adequate sample size to support a valid model, thus time points after this threshold are not displayed. Cox regression was used to estimate the covariate-adjusted effect of PVT on wait list mortality. Each candidate was observed until death, censoring only for the earliest of living or deceased donor liver transplant, loss to follow-up, or the end of the observation period (December 31, 2007). Each model included an indicator for PVT status (1 = yes; 0 = no) and the following adjustment covariates: race, sex, age, dialysis status, bilirubin, international normalized ratio (INR), albumin, sodium, and history of malignancy. No covariates were time-dependent, and inactive time was included in the analysis.
For the time-to-transplant Cox model, patients began follow-up at the date of initial placement on the waiting list and were followed until the earliest of death, transplant (deceased or living), loss to follow-up, or end of study. Model adjustment covariates included race, sex, age, dialysis status, bilirubin, international normalized ratio (INR), albumin, sodium, and history of hepatoblastoma. No covariates were time-dependent, and inactive time was included in the analysis.
Univariate comparisons were made between the PVT and non-PVT groups for all recipients. A Kaplan-Meier survival curve was created to compare survival among liver transplant recipients with and without PVT, and statistical comparisons used the log-rank test. For the longer term follow up in the Kaplan-Meier curve in recipients with PVT, we do not have adequate sample size to support a valid model, thus time points after this threshold are not displayed.
To model post-transplant mortality, recipients began follow-up at the time of liver transplantation (deceased or living). For the model of 30 day survival, patients were followed until death or end of observation time (30 days post-transplant). For the model of overall survival, patients were followed until death or loss to follow-up (occurred in less than 1% of recipients). No covariates were time-dependent. Creatinine, bilirubin, albumin, INR, sodium, and dialysis were coded as of their final pretransplant values. In addition to the covariates listed above, the post-transplant mortality models also included the following recipient characteristics at the time of transplant: sex, age, etiology of liver disease, living donor status, and components of the DRI, including: donor age, donor race, cause of death, donation following cardiac death status, and split liver status.
All statistical analyses were performed using SAS (v 9.3.1; SAS Institute; Cary, NC).
The characteristics of PVT and non-PVT pediatric liver transplant wait-listed candidates are displayed in Table 1. The prevalence of PVT (reported in this national database) among 5,087 candidates was 1.4% (n = 70). Candidates with PVT were more likely to be female (70.0% versus 51.0%; p=0.002) and have a lower PELD/MELD score at the time they were placed on the transplant list (9.9 ± 13.2 versus 15.3 ± 13.7; p=0.002). Age, race, Body Mass Index (BMI), dialysis, and previous transplant did not significantly differ significantly between PVT and non-PVT candidates. Diagnosis of biliary atresia was not more common among candidates with PVT, nor was survival among biliary atresia candidates significantly different from non-biliary atresia candidates. (p = 0.72)
A Kaplan-Meier survival curve was created to compare survival among liver transplant candidates with and without PVT (Figure 1). The unadjusted survival rate was not significantly lower among liver transplant candidates with PVT (p = 0.797).
We first assessed the relationship between PVT and waitlist mortality using a uni-variate Cox regression model. PVT was not significantly associated with waitlist mortality (HR = 1.1; 95% CI, 0.5–2.4; p = 0.765). In an attempt to elucidate the characteristics independently associated with waitlist mortality among all candidates, a multivariate Cox regression analysis was performed and is presented in Figure 2. Once again, PVT was not significantly associated with waitlist mortality by multivariate analysis (HR = 1.4; 95% CI, 0.7–3.0; p = 0.382). Female gender was associated with a significantly decreased covariate-adjusted mortality risk (HR = 0.8; 95% CI, 0.6‒0.9; p=0.003), while candidates that were two years of age or younger showed increased mortality (HR = 2.6; 95% CI, 1.9–3.4; p <0.0001) compared to candidates between the ages of 3 and 11. In addition, both bilirubin and INR values were associated with waitlist mortality.
We then investigated the relationship between reported PVT status and transplant rate. A covariate-adjusted Cox regression was performed in an effort to determine candidate characteristics independently associated with liver transplant rate. PVT was not independently associated with the adjusted liver transplant rate (HR = 0.9, 95% CI, 0.8–1.1, p = 0.35). However, candidates were more likely to receive a transplant if they were female (HR = 1.1; 95% CI, 1.0–1.1; p = 0.001), age two years or younger (HR = 1.1, 95% CI, 1.0–1.1, p = 0.0002), or between the ages 12–18 (HR = 1.4, 95% CI, 1.3–1.5; p <0.0001) [age reference group candidates between 3 and 12]. Similarly, candidates with a blood sodium between 132–137 mEq/L had an increased rate of transplantation (HR = 1.2, 95% CI 1.1–1.2, p <0.0001), while a mean sodium below 132 mEq/L was not significantly correlated. As expected, PELD/MELD components were significantly associated with transplant rate: INR (1.13, 95% CI, 1.06–1.20, p<0.001), bilirubin (HR = 1.25, 95% CI 1.23–1.28, p<0.0001), albumin (HR = 0.63, 95% CI 0.59–0.67, p<0.0001), and dialysis (HR = 1.4, 95% CI 1.2–1.6, p<0.0001). Race and previous malignancy diagnosis was not associated with transplantation rate.
The characteristics of PVT and non-PVT pediatric liver transplant recipients and their donors are displayed in Table 2. A total of 3,630 patients received a liver transplant and of those, 3.7% (n = 136) were reported to have PVT. Graft donor characteristics, including age, weight, and donor risk index (DRI), were not significantly different between the groups. In addition, utilization of deceased donor split liver grafts and living donor grafts was similar between the two groups. PVT recipients were more likely to be female (64.7% versus 51.1%; p = 0.002). Recipient age, race, PELD score at transplant, albumin, sodium, and ascites did not differ between groups. PVT was not associated with diagnosis of biliary atresia, though this assessment was limited by small sample size. More specifically, there were 1496 recipients with biliary atresia listed at the primary diagnosis. Among these recipients, there were 40 with PVT and 4 mortalities.
A Kaplan-Meier survival curve was created to compare survival among liver transplant recipients with and without PVT (Figure 3). The unadjusted survival rate is significantly lower in the PVT group across the post-transplant period (p = 0.01). It appears that the survival rate is particularly lower in the first few months following transplantation and then continues to trend downward in concordance with non-PVT recipients.
We then assessed the implications of PVT on short term post-transplant survival (30 day post-transplant survival). PVT was significantly associated with 30 day mortality (HR = 2.9; 95%CI 1.6–5.3, p = 0.001). Other covariates significantly associated with short term mortality included: creatinine (HR 4.42; 95%CI 2.8–7.1, p = 0.001), INR (HR = 1.6; 95%CI = 1.1–2.4, p=0.02), recipient age 0 – 2 years (HR = 1.7; 95%CI 1.1–2.6, p = 0.014) (compared to 3 to 12 years), and donor age (HR = 1.02; 95%CI 1.01–1.04, p = 0.01). No statistical interactions between PVT and the other covariates (including age at transplant) were noted.
A univariate regression was performed to determine the effects of PVT on overall post-transplant mortality. Recipients with PVT showed increased post-transplant mortality (HR = 1.7; 95% CI, 1.2–2.5; p = 0.007). These results were corroborated by a multivariate Cox regression analysis that revealed PVT was significantly associated with adjusted post-transplant mortality (HR = 1.6; 95% CI, 1.0–2.4; p = 0.03), as displayed in Figure 4. The model was adjusted for gender, age, race, creatinine, bilirubin, INR, albumin, sodium, dialysis, previous malignancy diagnosis, graft type, and retransplantation.
With this work, we have investigated the relationship between PVT and survival among both pediatric liver transplant candidates and recipients. Our hypothesis stated that liver transplant candidates and recipients with PVT would have a significantly higher risk of mortality. In fact, we only noted a significantly higher rate of post-transplant mortality in the 3.6% of liver transplant recipients with PVT; we did not note a statistically significant relationship between PVT and wait-list mortality. Though this study has several important limitations, it does represent the largest known series studying portal vein thrombosis and pediatric liver transplantation. In this context, these analyses do not specifically support policies facilitating early transplant for pediatric patients with PVT. This conclusion should be considered within the context of the limitations of this registry based analysis.
The etiology of PVT is likely related to a combination of decreased portal venous blood flow related to either cirrhosis and/or mesenteric venous hypoplasia and hypercoagability (for example: acquired deficiencies of proteins C and S). PVT increases the post-operative risk of death following adult liver transplantation (14, 18). This study shows similar post-transplant survival outcomes, which is in agreement with previous work in pediatric liver transplant patients (13). The implications of PVT on the survival of children on the waitlist remain less clear. Though our work noted a trend towards inferior survival, it was not statistically significant. Previous work in adult liver transplant populations has noted similar findings (14, 17).
When considering the results of our study, it is important to understand the limitations of retrospective observational data. The most important limitation to consider in this study is that our ability to truly define the incidence of PVT in pediatric liver disease population is restricted. Liver transplantation was performed on 3.7% of children with PVT in our study. Smaller, single center studies (by our group) have demonstrated much higher rates of PVT in transplant recipients (10%) which may demonstrate under-detection within the national registry data (13). A prospective study of screening within a large number of centers would be needed for an accurate determination of incidence. Another important limitation involves the characterization of PVT (complete or partial). In previous work, we assigned a specific definition to PVT to include only cases of occlusive thrombosis of the main portal vein (13, 17). Such rigorous definition of PVT is not possible using OPTN data, and as a result, we are not able to confirm the reliability of the PVT data reported by transplant centers. Clearly, complete portal vein thrombosis is significantly more complex to manage at the time of transplantation. Another limitation involves the lack of data on patients with ESLD who were never listed for transplantation. Future studies may seek to include these patients, for they likely represent a particularly sick cohort of patients. Finally, there are likely associations between biliary atresia and portal vein thrombosis that are not able to be fully understood due to limitations of this dataset. More specifically, pediatric patients with biliary atresia may present with complex mesenteric venous anatomy. Clearly, the complexity of their clinical scenario is not adequately represented using national registry data. Further, this analysis in limited by both small sample size and event rates among biliary atresia patients with PVT (only 40 recipients and 4 mortalities). Additional work is needed using more robust data sources to better understand the relationship between biliary atresia, portal vein thrombosis and anatomic variations, and outcomes.
Our data suggest that much of the increased risk of post-operative mortality is directly related to the transplant operation. Certainly, there are several options for the surgical management of this complex problem and it is critical that the mesenteric vascular anatomy is fully understood prior to transplantation. Considering that mesenteric venous hypoplasia is common in the setting of biliary atresia, a complete pre-operative radiographic work-up is particularly important in these patients. Recent studies comparing preoperative assessment of hepatic and portal vasculature have shown that Doppler ultrasound can be a useful initial screening test followed by CT venography or MRV for definitive operative planning (19, 20). These studies can also help to recognize important vascular anomalies such as a pre-duodenal portal vein (PV) or absence of a PV.
In summary, PVT is associated with an increase in postoperative mortality, though the relationship between PVT and waitlist mortality remains less clear. Additional work is needed to understand the incidence of this disease, particularly focusing on children who may have PVT but who are never listed for transplantation. Although further investigations are needed to delineate the best options for management of PVT in liver transplant recipients, careful preoperative diagnosis and planning offer the best opportunity for improvement in post-operative survival outcomes. Finally, although some limitations exist, this work does not seem to support broad-based policy efforts that would facilitate early transplant for pediatric patients with PVT.
Funding: The Scientific Registry of Transplant Recipients is funded by contract number 234-2005-37009C from HRSA, US Department of Health and Human Services. The views expressed herein are those of the authors and not necessarily those of the US government.
Support: Michael J. Englesbe was supported by the National Institutes of Health - National Institute of Diabetes and Digestive and Kidney Diseases (K08 DK0827508).