Late hepatic artery thrombosis after pediatric liver transplantation: A cross-sectional study of 34 patients

Authors


  • This study was funded by the Sigrid Juselius Foundation, the Finnish Medical Foundation, the Mary and Georg C. Ehrnrooth Foundation, and the National Graduate School of Clinical Investigation.

  • See Editorial on Page 512

Abstract

Hepatic artery thrombosis (HAT) after liver transplantation (LT) increases patient morbidity and mortality. Early HAT is considered to occur within the first month after LT, whereas late HAT occurs after the first month. Few studies have addressed late HAT after LT, especially in pediatric patients. Between 1987 and 2007, 99 patients (age < 18 years) underwent deceased donor LT. Thirty-four of 66 eligible patients (52%) underwent magnetic resonance imaging (MRI) according to protocol. On the basis of MRI findings, the patients were grouped as those who experienced late HAT and those who did not. Additionally, potential risk factors for late HAT were analyzed retrospectively. P values were adjusted for multiplicity. The median age at LT was 1.7 years [interquartile range (IQR) = 1.0-9.6 years], and the median follow-up time at MRI was 9.5 years (IQR = 4.0-16.4 years). Late HAT was diagnosed in 15 of the 34 patients [44%, 95% confidence interval (CI) = 29%-61%] undergoing MRI and in 3 of these patients with angiography preceding MRI. Ultrasonography revealed late HAT in 6 of these 15 patients with a sensitivity of 40% (95% CI = 20%-64%). The donor/recipient weight ratio remained significantly higher for the patients with late HAT versus the patients without late HAT after P values were adjusted (5.4 versus 1.9, P = 0.03). No marked differences were observed in laboratory or liver histology parameters between the groups. In conclusion, late HAT is common after pediatric LT. The donor/recipient weight ratio was higher for patients with late HAT, and this was attributable to the lower weight of the recipients. No salient features of late HAT were observed with respect to laboratory or histological parameters, at least in terms of our study's cross-sectional period. Liver Transpl 20:591–600, 2014. © 2014 AASLD.

Abbreviations
ALP

alkaline phosphatase

ALT

alanine aminotransferase

ASA

acetylic salicylic acid

AST

aspartate aminotransferase

CI

confidence interval

CK7

cytokeratin 7

CSA

cyclosporine A

GGT

gamma-glutamyltransferase

HAT

hepatic artery thrombosis

IQR

interquartile range

LMWH

low-molecular-weight heparin

LT

liver transplantation

MRI

magnetic resonance imaging

TAC

tacrolimus

US

ultrasonography

Vascular complications after liver transplantation (LT) decrease both graft and patient survival.[1] Hepatic artery thrombosis (HAT) is a major cause of retransplantation in the pediatric population.[2] Early and late forms of HAT have been described, but the exact time interval for these 2 forms has not been established. However, most studies define early HAT as occurring within the first month after LT.[3]

The incidence of early HAT is higher in pediatric patients versus adult patients, and risk factors for early HAT have been proposed even in a contradictory manner.[3] For example, in the setting of pediatric LT, both whole livers and reduced grafts have been shown to increase the risk for early HAT.[3]

The incidence of late HAT in adult LT patients has been reported to range from 0.9% to 4.7%.[4-9] In a series of both pediatric and adult patients, there were 70 occurrences of late HAT after 4234 LT procedures,[1] and the incidence of late HAT was 2.7% for children under 5 years of age who underwent LT.[10]

Patients with late HAT can be asymptomatic, although biliary complications are frequent,[5-7] and they are more frequent in late HAT cases versus early HAT cases.[8] In both early and late HAT, elevated transaminases are the most common feature.[1]

In this cross-sectional study, we report the occurrence of late HAT after pediatric LT diagnosed with magnetic resonance imaging (MRI). Additionally, we investigate factors associated with late HAT.

PATIENTS AND METHODS

Patients and Study Design

All pediatric patients (age < 18 years) who received deceased donor liver grafts between 1987 and 2007 at Helsinki University Central Hospital were evaluated for study eligibility (n = 99). After the exclusion of deceased patients (n = 32) and 1 patient lost during follow-up, the eligible study population (ie, those for whom MRI was possible) comprised 66 patients (67%). Thirty-seven of the 66 patients (56%) underwent abdominal MRI, and for 34 patients (52%), the use of the contrast medium was successful. Thus, altogether, 32 of the 66 eligible patients (48%) were not studied.

A comparison of the 34 study patients (MRI with successful use of the contrast medium) and the 32 nonstudy patients [no MRI (n = 29) or MRI without successful use of the contrast medium (n = 3)] is shown in Table 1.

Table 1. Comparison of 34 Patients Who Underwent MRI (Study Patients) and 32 Patients Who Did Not (Nonstudy Patients)
 No MRI (n = 32)aMRI (n = 34)P Value
  1. NOTE: The P values are unadjusted as stated in the Patients and Methods section.

  2. a

    This group includes 3 patients who underwent MRI but for whom the use of the contrast medium was not successful.

  3. b

    The data are presented as medians and IQRs.

  4. c

    The 95% CI for the difference in the medians was 1.2 to 13.0 years.

  5. d

    The 95% CI for the difference in the medians was −4.1 to −0.5.

  6. ||Two anastomoses and 1 anastomosis were missing for the nonstudy and study groups, respectively.

Diagnosis [n (%)]  0.01
Biliary atresia7 (22)14 (41) 
Metabolic diseases1 (3)7 (21) 
Hepatitis8 (25)7 (21) 
Other16 (50)6 (18) 
Age at LT (years)bc8.9 (2.0-15.1)1.7 (1.0-9.6)0.02
Donor/recipient weight ratiobd1.4 (1.0-3.3)3.7 (1.4-5.8)0.01
Graft type [n (%)]  0.02
Whole liver17 (53)8 (24) 
Reduced liver15 (47)26 (76) 
Arterial anastomosis [n (%)]||  0.07
Hepatic artery22 (73)16 (48) 
Aorta8 (27)17 (52) 

MRI was performed as part of the routine clinical follow-up protocol without clinical indications. Because of family refusal or for logistic reasons, only 37 of the 66 patients (56%) underwent MRI, and for 34 (52%), the contrast medium was successfully used at the time of our cross-sectional study. Two patients underwent MRI for a clinical indication, and these patients were included in the analyses. MRI was performed once for each study patient (n = 34) between 2009 and 2011, and the MRI findings were analyzed with respect to the study patients' cross-sectional biochemical variables and liver histology (determined as part of the routine clinical follow-up protocol). Additionally, a retrospective analysis of potential risk factors for late HAT was incorporated into our study on the basis of these MRI findings (see the Preoperative, Perioperative, and Postoperative Factors section). Detailed follow-up of patients with late HAT was not within the scope of this study.

The patient characteristics of the 34 study patients are shown in Table 2. The cardiothoracic surgical team of Children's Hospital operated on 32 patients, and the LT team of the Department of Transplantation and Liver Surgery operated on the other 2 patients.

Table 2. Characteristics of 34 LT Patients Together and Divided Into 2 Groups Based on MRI Findings
 All Patients (n = 34)No Late HAT (n = 19)Late HAT (n = 15)P Value
  1. NOTE: The P values have been derived from comparisons of the no–late HAT group and the late HAT group and are adjusted.

  2. a

    The data are presented as medians and IQRs.

  3. b

    The minimum was 0.66 years, and the maximum was 17.8 years.

  4. c

    The mean weight was 23.8 kg for all patients, 29.0 kg for patients without late HAT, and 17.1 kg for patients with late HAT.

  5. d

    The data are presented as means and standard deviations.

  6. e

    |The difference in the medians was −9.3 years (95% CI = −16.5 to −2.2 years).

  7. f

    The total percentages are greater than 100% because of rounding.

  8. g

    Tyrosinemia (n = 4), Wilson's disease (n = 2), and familial hypercholesterolemia (n = 1).

  9. h

    Autosomal recessive polycystic kidney disease (n = 2), hepatoblastoma (n = 2), atypical hemolytic uremic syndrome (n = 1), and familial congenital liver cirrhosis (n = 1).

  10. i

    The categorization was based on the median LT year of 2000.

  11. j

    One anastomosis was not retrievable from medical records.

Sex [n (%)]   >0.99
 Male21 (62)12 (63)9 (60) 
 Female13 (38)7 (37)6 (40) 
Age at LT (years)a1.7 (1.0-9.6)b5.2 (1.4-12.8)1.1 (0.9-3.3)0.48
Weight at LT (kg)ac12.0 (9.4-27.2)17.5 (9.6-47.5)10.0 (9.3-16.5)0.66
Age at MRI (years)d15.4 ± 7.015.5 ± 6.515.2 ± 7.9>0.99
Follow-up at MRI (years)a9.5 (4.0-16.4)5.8 (4.0-15.2)||15.1 (4.2-17.1)||0.15
Diagnosis [n (%)]f   0.40
 Biliary atresia14 (41)7 (37)7 (47) 
 Metabolic diseasesg7 (21)3 (16)4 (27) 
 Hepatitis7 (21)3 (16)4 (27) 
 Otherh6 (18)6 (32)0 
LT era [n (%)]i   0.48
 Before 200017 (50)7 (37)10 (67) 
 2000 or after17 (50)12 (63)5 (33) 
Graft type [n (%)]   0.30
 Whole liver8 (24)7 (37)1 (7) 
 Reduced liver26 (76)12 (63)14 (93) 
Arterial anastomosis [n (%)]j   0.26
 Hepatic artery16 (47)12 (67)4 (27) 
 Aorta17 (50)6 (33)11 (73) 
 Not available1 (3)   
Biliary reconstruction [n (%)]   0.73
 Roux-en-Y27 (79)14 (74)13 (87) 
 Duct-to-duct7 (21)5 (26)2 (13) 

Three of the 34 patients (8.8%) underwent combined liver-kidney transplantation. In addition, 3 of the 34 patients underwent re-LT 4, 18, and 20 years before the abdominal MRI evaluation. The graft failures were due to primary nonfunction, HAT, and chronic rejection, respectively. Three patients (8.8%) were diagnosed with late HAT before the study (0.9, 1.6, and 10.1 years after LT) with computed tomography angiography and procedure-based angiography.

The patients' medical records were retrospectively evaluated, and the national LT registry was used to retrieve some additional information. This study was approved by the ethics committee for pediatrics, adolescent medicine, and psychiatry of the Hospital District of Helsinki and Uusimaa (application number 345/13/03/03/2008). Informed consent was obtained from patients and, in the case of minors, also from parents or guardians.

Immunosuppression

Triple immunosuppression, including cyclosporine A (CSA), azathioprine, and methylprednisolone, was used for the initial immunosuppression. When it was clinically indicated, CSA was switched to tacrolimus (TAC), and azathioprine was switched to mycophenolate mofetil or mycophenolic acid. Immunosuppressant changes were made on an individual basis. The methylprednisolone dose was tapered to 0.25 mg/kg/day at 2 weeks, and this was switched to every other day at 6 months; the aim was a corticosteroid-free regimen in adulthood.

At the time of the MRI evaluation, 13 of the 34 patients (38%) were on CSA-based triple immunosuppression, and 10 (29%) were on TAC-based triple immunosuppression. Eight of the 34 patients (24%) were on CSA-based double immunosuppression, and 3 (9%) were on TAC-based double immunosuppression.

Cross-Sectional Study

The study visits took place at Helsinki University Central Hospital either in the pediatric unit (Children's Hospital; 26 patients) or in the adult unit (Department of Transplantation and Liver Surgery; 8 patients).

Imaging

Abdominal MRI was performed with a 1.5-T magnetic resonance scanner (Achieva, Philips, Best, the Netherlands) as a part of our protocol. Gadoteric acid was used as the contrast medium. Small children underwent MRI under general anesthesia (age < 7 years; n = 6). The MRI protocol varied because of differences in the available equipment. In general, T2-weighted and contrast-enhanced T1-weighted gradient-echo sequences were used. With respect to T2-weighted images, the validity of half-Fourier acquisition single-shot turbo spin-echo sequences is well established. Moreover, we usually perform a heavily T2-weighted sequence routinely used in magnetic resonance cholangiopancreatography. Because of the extremely short acquisition time, the true fast imaging with steady-state precession sequence appears very useful in the study of pediatric or uncooperative patients unable to maintain the required breath-hold. A pediatric radiologist unaware of the patients' clinical status re-evaluated MRI images.

Doppler ultrasonography (US) reports were re-evaluated, and in 3 late HAT patients diagnosed by angiography, US reports before the time of angiography were re-evaluated. MRI and US were mostly performed at the same time, but in 13 patients (38%), US and MRI were performed separately [mean = 157 ± 139 days, median = 105 days, interquartile range (IQR) = 35-233 days]. In 2 patients, MRI was performed before US.

Biochemical Variables

The laboratory parameters of interest (with an emphasis on liver function tests) were determined at the time of MRI for 16 patients (47%; <14 days apart). For 18 patients (53%), some laboratory parameters of interest (usually basic liver tests) were determined at the time of MRI, but some parameters of interest were not. For these 18 patients, all laboratory parameters of interest were pooled together with mean and median time differences between MRI and pooled laboratory values of 100 ± 83 and 89 (IQR = 34-135) days, respectively.

Liver Histology

Liver biopsy was performed as part of the routine clinical follow-up protocol regardless of patient symptoms. A detailed description of the histological methods and the results of biopsy findings have been reported previously.[11] Briefly, core needle biopsy was performed under ultrasound guidance, and biopsy specimens were fixed, embedded, and stained with conventional histochemical methods. Immunostaining for cytokeratin 7 (CK7) was also performed. Biopsy specimens were evaluated and scored by 2 pathologists unaware of the patients' clinical status. Fibrosis was coded from 0 to 4, portal inflammation was coded from 0 to 3, CK7 for periportal hepatocytes (chronic cholestasis) was coded from 0 to 3, and bile duct proliferation was coded from 0 to 2.

Thirty of the 34 patients (88%) had biopsy specimens available. For 20 of these patients (67%), MRI and liver biopsy were performed at a mean of 134 ± 143 days and at a median of 73 days (IQR = 34-196 days) apart. For 16 of these 20 patients, biopsy was performed before MRI, and for 4 patients, biopsy was performed afterward. The time difference between MRI and liver biopsy was greater than the 75th percentile (196 days) for 2 patients with late HAT (544 and 338 days) and for 3 patients without late HAT (352, 271, and 203 days).

Preoperative, Perioperative, and Postoperative Factors

The preoperative laboratory values for each patient's coagulation status were collected, and anesthesiology reports were used to obtain information about the use of frozen plasma and platelets (units), the patient's weight at LT, blood loss during LT, the length of the anhepatic phase, the cold ischemia time, and the use of antifibrinolytics.

As for postoperative factors (ie, factors other than cross-sectional study parameters), the use of low-molecular-weight heparin (LMWH) and acetylic salicylic acid (ASA) was recorded. Twenty-three patients (68%) were treated with LMWH: enoxaparin (n = 16) or dalteparin (n = 7). All the patients who did not receive LMWH treatment (n = 11) underwent LT before 1996. Thirty patients (88%) were treated with ASA, and it was used once daily (either 50 or 100 mg). ASA was used 3 to 4 months after LT according to a treatment protocol in the early LT era. Currently, the treatment period varies from one patient to another (eg, because of suspicion of potential vascular problems). Four patients (12%) did not receive ASA treatment at all for unknown reasons. The duration of the ASA treatment was calculated from the day on which the treatment ended (n = 15) or from the day on which MRI/angiography was performed (n = 15). The duration of the LMWH treatment was calculated from the day on which the treatment ended because all 23 patients received LMWH for only a short time after LT. Also, patients who were not at all treated with LMWH (n = 11) were compared to LMWH-treated patients (n = 23). Sensitivity analyses of ASA treatment were performed without 2 patients with an angiography-based diagnosis of late HAT and ASA treatment at the time of angiography. Finally, we focused only on those 15 ASA-treated patients for whom the treatment end date was determined by the clinician (clinically based ASA discontinuation date) and not by the date of MRI or angiography and on 4 patients not treated with ASA at all.

Statistics

Stata 12.1 (StataCorp LP, College Station, TX) was used for statistical calculations except for the Cochran-Armitage trend test, for which SAS 9.3 (SAS Institute, Gary, NC) was used. Patient survival and graft survival (death was treated as graft failure) from first LT were calculated with the Kaplan-Meier method. On the basis of MRI, groups with late HAT (n = 15) and without late HAT (n = 19) were formed. Normality was evaluated with the Shapiro-Wilk test with an α level of 0.05. Continuous variables are presented as medians and IQRs or as means and standard deviations. Categorical variables are presented as frequencies and percentages. Laboratory values under or over the detection level were set to the lowest or highest reported level, respectively. Missing values were treated as missing for analyses.

Comparisons of medians were performed according to the Bonett-Price method[12] with the Stata bpmedian module,[13] and comparisons of means were made with an unpaired Student t test. Comparisons between categorical variables were made with Fisher's exact test. Exact logistic regression was used to investigate surgery-related factors: the type of artery anastomosis [(0) hepatic artery or (1) aorta)] and the graft type [(0) whole liver or (1) reduced liver graft)] as independent variables and late HAT [(0) no late HAT or (1) late HAT)] as a dependent variable. Odds ratios and related confidence intervals (CIs) are reported for exact logistic regression. Ordered histological factors (fibrosis, CK7 for periportal hepatocytes, and bile duct proliferation) and late HAT grouping were analyzed with the exact Cochran-Armitage trend test. Spearman rank correlation was used between the cross-sectional laboratory values and ordered histological parameters (fibrosis, CK7 for periportal hepatocytes, and bile duct proliferation). Because of the exploratory nature of our study, P values considered to be family in a broad sense (comparisons between study and nonstudy patients not being counted; see Table 1 were adjusted with the Stata qqvalue module with simes as the method option to produce adjusted P values (ie, q values; for details, see reference [14]). CIs are not comparable to adjusted P values. All P values are 2-tailed and are presented in an adjusted form except for the aforementioned comparisons between study and nonstudy patients.

RESULTS

Patient and Graft Survival for All 99 Transplant Patients From 1987 to 2007

Patient survival from first LT is shown in Fig. 1. The 1-year patient survival rate was 84% (95% CI = 75%-90%, 83 patients at risk). The 1, 5-, and 10-year graft survival rates were 79% (95% CI = 69%-86%, 78 patients at risk), 72% (95% CI = 62%-80%, 65 patients at risk), and 65% (95% CI = 55%-74%, 44 patients at risk), respectively. HAT played a role at least to some extent in 3 patient deaths (1 late HAT) and in 4 graft losses (2 late HATs).

Figure 1.

Patient survival (with 95% CIs) from first LT for all 99 patients who underwent LT between 1987 and 2007.

Cross-Sectional Study Findings

Imaging

Late HAT was diagnosed in 15 of the 34 patients (44%, 95% CI529%-61%) who underwent MRI. In 3 of these patients, late HAT was diagnosed with angiography before MRI. Doppler US revealed late HAT in 6 of these 15 patients with a Doppler US sensitivity of 40% (95% CI = 20%-64%). The specificity of Doppler US was 100% (95% CI = 83%-100%). In 12 late HAT patients (80%), some forms of collateral arterial revascularization were evident on MRI. The mean liver volumes of the 2 groups were comparable (Table 3). This was also true for patients with reduced liver grafts (1076 cm3 as in Table 3 versus 931 cm3, 95% CI for the difference = −192 to 481, P = 0.69).

Table 3. Cross-Sectional Variables for Patients With Late HAT and Patients Without Late HAT
 No Late HAT (n = 19)Late HAT (n = 15)Difference (95% CI)P Value
  1. NOTE: The P values are adjusted.

  2. a

    The data are presented as medians and IQRs.

  3. b

    The reference range was <6 μmol/L [n = 32; there was 1 left-censored value (ie, set to 2 μmol/L)].

  4. c

    The data are presented as means and standard deviations.

  5. d

    The reference range was 70% to 130% [there was 1 right-censored value (ie, set to 170%)].

  6. e

    |The reference range was <15 minutes (n = 26: 14 in the no–late HAT group and 12 in the late HAT group).

  7. The difference varies because of rounding

ALT (U/L)a20 (14-30)29 (16-35)−9 (−19 to 1)0.37
AST (U/L)a29 (25-38)34 (29-44)−5 (−14 to 4)0.63
GGT (U/L)a16 (13-33)34 (22-59)−18 (−36 to 0)0.30
ALP (U/L)a182 (79-233)149 (107-288)33 (−75 to 141)0.83
Total bilirubin (μmol/L)a13 (8-15)9 (7-19)4 (−2 to 10)0.56
Bile acids (μmol/L)ab4.4 (3.3-11.2)9.2 (4.2-16.8)−4.8 (−11.7 to 2.2)0.49
Albumin (g/L)a39.2 (36.0-39.9)36.8 (34.4-40.8)2.4 (−0.9 to 5.7)0.48
Prealbumin (mg/L)c192 ± 59212 ± 65−20 (−63 to 24)0.66
Plasma urea (mmol/L)c6.6 ± 2.56.9 ± 2.3−0.3 (−2.0 to 1.4)0.88
Thromboplastin time (%)cd101 ± 27107 ± 27−6 (−25 to 13)0.83
Galactose half-life (minutes)c||10.4 ± 1.511.4 ± 2.2−1.1 (−2.6 to 0.4)0.48
Liver volume on MRI (cm3)c1133 ± 4251002 ± 500131 (−192 to 454)0.73
Spleen size on MRI (cm)c12.3 ± 3.112.3 ± 2.50.0 (−2.0 to 2.1)>0.99

Biochemical Variables

None of the laboratory values determined at the time of MRI differed significantly between the groups (Table 3). Nonetheless, gamma-glutamyltransferase (GGT) and bile acid levels were approximately twice as high for patients with late HAT versus patients without late HAT.

Liver Histology

Details of liver histology are shown in Table 4. Patients with late HAT seemed to have more severe chronic cholestasis (according to CK7 staining for periportal hepatocytes) than patients without late HAT. CK7 staining also showed a positive correlation with cross-sectional GGT (Spearman ρ = 0.59, P = 0.03), and positive correlations were evident with cross-sectional alanine aminotransferase (ALT, ρ = 0.39) and bile acids (ρ = 0.43), but the significance was weakened after adjustments (P = 0.26 for ALT and P = 0.20 for bile acids). Other tested correlations showed no statistical significance before adjustments [fibrosis, CK7 for periportal hepatocytes, and bile duct proliferation were correlated with ALT, aspartate aminotransferase (AST), GGT, total bilirubin, and bile acids; there were 15 correlations altogether].

Table 4. Liver Histology Findings for Patients With Late HAT and Patients Without Late HAT
Liver HistologyAll Patients (n = 30)No Late HAT (n = 16)Late HAT (n = 14)P Value
  1. NOTE: The P values are adjusted.

  2. a

    Exact Cochran-Armitage trend test.

  3. b

    Fisher's exact test.

  4. the total percentage is greater than 100% due to rounding.

Fibrosis [n (%)]   0.97a
017 (57)9 (56)8 (57) 
110 (33)6 (38)4 (29) 
21 (3)01 (7) 
32 (7)1 (6)1 (7) 
Portal inflammation [n (%)]   0.86b
024 (80)12 (75)12 (86) 
16 (20)4 (25)2 (14) 
CK7 for periportal hepatocytes [n (%)]   0.33a
022 (73)14 (88)8 (57) 
16 (20)2 (13)4 (29) 
22 (7)02 (14) 
Bile duct proliferation [n (%)]   0.40a
019 (63)7 (44)12 (86) 
110 (33)9 (56)1 (7) 
21 (3)01 (7) 

Baseline Differences Between Patients With Late HAT and Patients Without Late HAT

The baseline patient characteristics and the primary diagnoses were comparable for the 2 groups (Table 2). However, patients with late HAT more frequently received an arterial anastomosis for the aorta versus the hepatic artery, and reduced livers were used more often in these patients (Table 2). These 2 factors were analyzed as explanatory variables and late HAT was analyzed as the response variable in exact logistic regression. Overall, the regression model score was 7.37 (P = 0.25); for the type of artery anastomosis, the odds ratio was 3.9 (95% CI = 0.7-26.2, P = 0.48), and for the graft type, it was 5.8 (95% CI = 0.5-308.1, P = 0.56).

Anomalies of the hepatic artery were comparable in the 2 groups (3/18 patients in the no–late HAT group versus 4/15 patients in the late HAT group, P = 0.87; 1 patient was missing for the no–late HAT group).

Preoperative, Perioperative, and Postoperative Factors for Patients With Late HAT and Patients Without Late HAT

No marked differences in preoperative laboratory values were found between the groups (Table 5). As for perioperative factors, the median donor/recipient weight ratio was higher for patients with late HAT (5.4 versus 1.9, P = 0.03, donor/recipient weight ratio for whole group = 3.7). The donor weights were similar for the 2 groups (the median and mean donor weights were 60 and 56 kg, respectively, for the patients without late HAT and 60 and 59 kg, respectively, for the patients with late HAT; for recipient weights, see Table 2). The donor/recipient age ratio was higher for the patients with late HAT (21.7 versus 3.4, P = 0.15), and the donor age was also higher for this group (the median and mean donor ages were 36 and 33 years, respectively, for the patients with late HAT and 17 and 24 years, respectively, for the patients without late HAT). The usage of perioperative antifibrinolytics was similar (14/19 versus 10/15, P = 0.88), as were the donor causes of death (P = 0.79; details not shown).

Table 5. Preoperative, Perioperative, and Postoperative Variables for 2 Groups Based on MRI Findings
 No Late HAT (n = 19)Late HAT (n = 15)Difference (95% CI)P Value
  1. NOTE: The P values are adjusted.

  2. a

    The data are presented as means and standard deviations.

  3. b

    The difference varies because of rounding.

  4. c

    The data are presented as medians and IQRs.

  5. d

    The reference range was 70% to 130% [there were 3 left censored values (ie, set to 6%): 2 in the no–late HAT group and 1 in the late HAT group].

  6. e

    |The reference range was 23 to 33 seconds [n = 30; there were 4 right censored values (ie, set to 180 seconds): 2 in the no–late HAT group and 2 in the late HAT group].

  7. f

    The reference range was 79% to 128% (n = 22: 11 in each group).

  8. g

    n = 23 (14 in the no–late HAT group and 9 in the late HAT group).

  9. h

    Determined from the day of MRI/angiography (n = 15: 9 in the no–late HAT group and 6 in the late HAT group) or the day of clinically based stoppage (n = 15: 7 in the no–late HAT group and 8 in the late HAT group) minus the day on which ASA was started.

  10. i

    The complications were leaks and strictures during the postoperative course.

Preoperative laboratory values    
Hemoglobin (g/L)a105 ± 20101 ± 163 (−10 to 16)b0.83
Hematocrit (%)a31 ± 630 ± 41 (−3 to 4)0.88
Platelet count (×109/L)c94 (68-118)137 (79-230)−43 (−117 to 31)0.61
Thromboplastin time (%)cd51 (18-74)27 (16-58)24 (−9 to 57)0.48
Activated partial thromboplastin time (seconds)c||38 (27-85)52 (34-112)−14 (−67 to 39)0.83
Factor V (%)cf51 (26-80)35 (18-48)16 (−17 to 49)0.66
Perioperative variables    
Donor/recipient weight ratioc1.9 (1.1-4.7)5.4 (3.1-6.4)−3.5 (−5.5 to −1.4)0.03
Donor/recipient age ratioc3.4 (1.2-10.9)21.7 (7.5-36.2)−18.3 (−32.1 to −4.5)0.15
Frozen plasma at LT (U)c6 (2-10)6 (4-8)0 (−5 to 5)>0.99
Platelets at LT (U)c8 (4-16)8 (4-8)0 (−7 to 7)>0.99
Blood loss at LT (mL)c2310 (1500-6600)3300 (2050-5000)−990 (−3010 to 1030)0.66
Cold ischemia time (hours)a7.3 ± 2.38.7 ± 2.4−1.5 (−3.1 to 0.2)b0.37
Anhepatic phase (minutes)a61 ± 1467 ± 31−6 (−24 to 13)0.81
Postoperative variables    
Stay in intensive care unit (days)c3 (1-4)4 (3-7)−1 (−3 to 1)0.66
LMWH duration (days)cg12.5 (6.8-19.0)12.0 (9.5-32.5)0.5 (−15.9 to 16.9)>0.99
ASA duration (days)ch1201 (153-1454)128 (46-1196)1073 (271 to 1865)0.15
Rejection episodes [n (%)]   >0.99
No6 (32)5 (33)  
Yes13 (68)10 (67)  
CMV viremia episodes [n (%)]   0.63
No15 (79)9 (60)  
Yes4 (21)6 (40)  
Biliary complications [n (%)]i   0.83
No18 (95)13 (87)  
Yes1 (5)2 (13)  

As for postoperative factors, no marked differences were found in the duration of the LMWH treatment between the groups. The proportions of late HAT were similar for patients who were not at all treated with LMWH and patients who were treated with LMWH (6/11 versus 9/23, P = 0.79). The median duration of the ASA treatment, however, was shorter for the late HAT patients (Table 5). Excluding 2 patients with a late HAT diagnosis based on angiography and ASA usage at the time of angiography did not change the results (difference = 1100 days, 95% CI = 192-2008 days, P = 0.20). The groups were similar, however, when we focused only on patients with a clinically based ASA discontinuation date (n = 15; 92 versus 68 days, 95% CI for the difference = −543 to 592 days, P > 0.99). In addition, the frequencies of no treatment with ASA were similar (3/19 for the no–late HAT group versus 1/15 for the late HAT group, P = 0.83).

DISCUSSION

Our study of 34 patients shows that late HAT is common after LT when it is diagnosed by MRI, but it is less common when it is diagnosed with US. We observed that the donor/recipient weight ratio was higher for patients with late HAT, and this was attributable to the lower weights of the recipients at LT rather than higher donor weights.

MRI was performed for only 52% of the whole population, so our estimate of late HAT might be an underestimate or an overestimate. We can speculate that in the best case scenario, 15 late HAT cases among 66 patients would yield a prevalence of 23%; in the worst case scenario (if we assume that in addition to the 15 patients with late HAT, all 32 nonstudy patients had late HAT), the prevalence would be as high as 71%. Duffy et al.[1] reported that HAT was diagnosed in 69 of 866 pediatric LT cases, but no incidence of late HAT was reported separately for pediatric patients. In 2 older pediatric studies, the rate of late HAT varied from 2.7%[10] to 6.1%,[15] and both studies used angiography among other methods. If we focus only on diagnoses based on Doppler US, the prevalence would be 18% (6/34), which is still higher than that previously reported. McDiarmid et al.[16] showed that in 10 pediatric cases of HAT, 5 were true positives and 5 were false negatives according to US, and this was later confirmed by angiography; this meant 50% sensitivity for US. The sensitivity of Doppler US for detecting HAT diminishes in adults later after LT.[17] Both studies showed collateral formation in HAT patients.[16, 17]

In contrast to our study, Sieders et al.[18] reported that in their early HAT series of pediatric patients, the donor/recipient weight ratio was lower for HAT patients versus control patients (1.5 versus 2.2), although not significantly so. This difference was attributable to both lower donor weights and higher recipient weights for the HAT patients versus the control patients.

We found some indications that chronic cholestasis was more evident in patients with late HAT. Because the hepatic artery provides blood to the biliary tree, chronic cholestasis and HAT in theory may go hand in hand. Only a few cases of late HAT were present in a pediatric protocol liver biopsy study[19]; therefore, putting our findings (although not significant) in context with other protocol liver biopsy studies remains open. A recent study[11] from our center observed a correlation between GGT and chronic cholestasis, and this study (examining a subset of patients from ref. 11) also showed a correlation between GGT and chronic cholestasis.

We observed a difference in the median ASA treatment duration between patients with late HAT and patients without late HAT, but this was weakened after statistical adjustments. No significant differences were evident in our study, however, when we focused on patients with a clinical decision–based ASA treatment period (n = 15) or on patients receiving no ASA (n = 4). One study[6] found a lower incidence of late HAT in an ASA prophylaxis group versus patients who did not receive prophylaxis (0.4% versus 2.2%), and another study[20] found similar frequencies of late HAT in patients using ASA and patients not using ASA (1.2% versus 1.0%). HAT studies of pediatric patients have found no significant differences in early HAT incidence between patients using ASA and patients not using ASA or in the frequency of ASA or alprostadil usage between patients with HAT and patients without HAT.[21, 22] The benefits and risks of ASA in reducing HAT after LT have not been studied to our knowledge in randomized controlled trials. Therefore, one might argue that its use should not be recommended, but recommendations are frequently based on observational studies, at least in distinguished general medicine journals.[23]

Although nearly half of our study population had late HAT, we did not find any clear factors that might reflect poor patient outcomes. Late HAT, however, played something of a role in 1 patient's graft loss long after LT (shortly after MRI), and this emphasizes the importance of vigilant observation. One study[24] has previously shown that late HAT is associated with a neovascularized liver. The survival rate for 11 patients (including 10 patients with late HAT) was 90.9% at 4.5 years, and the median time from a diagnosis of late HAT to a neovascularized liver was 4.1 months.[24] Most of our study patients had arterial revascularization, which might indicate adequate protection against more deleterious disease patterns of late HAT highlighted in the recent pediatric guidelines.[25] Although we do not know exactly how long our patients had been living with late HAT, MRI might provide a reasonable radiological approach for monitoring these patients in the long run. However, the benefits and risks of MRI should be carefully balanced, especially in younger children, who usually require general anesthesia and intubation to make MRI feasible. The detection of late HAT might not be necessary if adequate collaterals have been formed. However, US seems suboptimal for detecting late HAT; therefore, MRI might be justified, especially because computed tomography involves radiation exposure. It also might be a valid approach to wait for smaller children to reach a certain age at which general anesthesia is not required to perform MRI. The long-term outcomes of our patients with late HAT remain to be seen.

The strengths of our study include radiological, biochemical, and histological parameters determined in a cross-sectional manner and a unified protocol for treating our patients at and after LT. We adjusted P values to diminish type I errors, and 10 of 72 P values (14%) were <0.05 before adjustments (n = 2, P < 0.001; n = 2, P < 0.01; and n = 6, P < 0.05). Two of these remained significant at a q value level of 5% (adjusted P value = 0.03).

We studied approximately half of the eligible study population, and this may have led to a sampling bias. Some differences between study and nonstudy patients were observed; therefore, caution is warranted in generalizing our findings. Sensitivity and specificity estimates for US should be interpreted with some caution because our study was not designed to be a diagnostic accuracy study. Additionally, there were time differences for matching MRI findings with liver histology findings and for matching MRI findings with laboratory values; therefore, some caution is warranted in the interpretation of MRI with respect to histological and biochemical findings, although the exact time of the late HAT diagnosis remains open.

In conclusion, our study shows that late HAT is common after pediatric LT. In addition, the donor/recipient weight ratio was higher for the patients with late HAT. We observed no drastic features of late HAT with respect to biochemical or histological parameters, at least in terms of our study's cross-sectional period.

Ancillary

Advertisement