Liver transplantation for acute liver failure in Europe: Outcomes over 20 years from the ELTR database

Research Article

Liver transplantation for acute liver failure in Europe: Outcomes over 20 years from the ELTR database Giacomo Germani1,2, , Eleni Theocharidou1, , Renè Adam3, , Vincent Karam3, , Julia Wendon4, , John O’Grady4, , Patrizia Burra2, , Marco Senzolo2, , Darius Mirza5, , Denis Castaing3, , Jurgen Klempnauer6, , Stephen Pollard7, , Andreas Paul8, , Jacques Belghiti9, , Emmanuel Tsochatzis1, , Andrew K. Burroughs1,⇑,  1

The Royal Free Sheila Sherlock Liver Centre, University Department of Surgery, Royal Free Hospital and UCL, London UK; 2Gastroenterology, Department of Surgical and Gastroenterological Sciences, Padua University Hospital, Padua, Italy; 3ELTR, Assistance Publique-Hôpitaux de Paris, Hôpital Paul Brousse, Centre Hepato-Biliaire, Universite´Paris-Sud, Villejuif, France; 4Liver Intensive Unit, Institute of Liver Studies, King’s College Hospital, London, UK; 5The Queen Elizabeth Hospital, Birmingham, UK; 6Medizinische Hochschule Hannover, Hannover, Germany; 7St. Jame’s & Seacroft University Hospital, Leeds, UK; 8C. U. K. GHs Essen, Essen, Germany; 9Hôpital Beaujon, Clichy, France

See Focus, pages 233–234

Background & Aims: Liver transplantation for acute liver failure (ALF) still has a high early mortality. We evaluated changes during 20 years, and identified risk factors for poor outcome. Methods: Donor, graft, and recipient variables from the European Liver Transplant Registry database (January 1988–June 2009), were analysed. Aetiologies and time periods were compared. Three and 12-month survival models were generated from separate training data sets, which were validated. A sub-analysis was performed for recipient older than 50 years. Results: Four thousand nine hundred and three patients were evaluated. One, 5- and 10-year patient, and graft survival rates were 74%, 68%, 63%, and 63%, 57%, 50%, respectively. Survival was better in 2004–2009 compared to previous quinquennia (p <0.001), despite donors >60 years increased from 1.8% to 21%. A higher incidence of suicide or non-adherence occurred in paracetamol-related ALF (p <0.001). Death or graft loss were independently associated with male recipients (adjusted OR 1.25), recipient >50 years (1.26), incompatible ABO matching (1.93), donors >60 years (1.21), and reduced size graft (1.54). For both 3- and 12-month models, incompatible ABO matching, non-viral aetiology, reduced size graft, and non-UW preservation fluid were associated with increased mortality/graft loss, whereas male recipients and age >50 years were associated only at 12 months. Both models had reasonable discriminative ability with good calibration at 3 months. Recipients >50 years, comKeywords: Liver transplantation; Acute liver failure; Europe. Received 14 November 2011; received in revised form 1 February 2012; accepted 8 March 2012; available online 18 April 2012 q DOI of original article: http://dx.doi.org/10.1016/j.jhep.2012.05.005. ⇑ Corresponding author. Address: The Royal Free Sheila Sherlock Liver Centre, Royal Free Hospital, Pond Street, NW3 2QG London, UK. Tel.: +44 02074726229; fax: +44 02074726226. E-mail address: [email protected] (A.K. Burroughs).   For the European Liver and Intestine Transplant Association (ELITA) and on behalf of the European Liver Transplant Registry (ELTR) Liver Transplant Centres listed at www.eltr.org. Abbreviation: ALF, acute liver failure.

bined with donors >60 years resulted in 57% mortality/graft loss within the first year. Conclusions: Survival after liver transplantation has improved despite increases in donor/recipient age. Recipients >50 years paired with donors >60 years had a very high mortality/graft loss within the first year. Ó 2012 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

Introduction Acute liver failure (ALF) accounted for 8% of indications for liver transplantation in Europe during 1988–2009 [1] and 7% in the USA during 1999–2008 [2]. The survival rates at 1 year have improved from 60% to 92% in large centres over the past 20 years [3–11], but are less than for cirrhosis [12]. The European Liver Transplant Registry (ELTR) holds data on 87,963 liver transplantations performed in 79,063 patients in 23 European countries over 43 years [13], and has been used to evaluate outcomes and evolution of liver transplantation in Europe [14]. The primary objective of this study was to use the ELTR database to evaluate outcomes after liver transplantation for ALF, comparing different aetiologies of ALF, and to evaluate these outcomes with respect to changes in recipient and donor characteristics, and in surgical techniques over the past 20 years. The secondary objective was to identify potential risk factors for death or graft loss after liver transplantation for ALF, so as to explore if these could be used to establish a particularly poor prognosis.

Materials and methods The methods and details used to obtain the data have been described previously [15] (Supplementary material). The quality of ELTR data is controlled regularly by

Journal of Hepatology 2012 vol. 57 j 288–296

JOURNAL OF HEPATOLOGY audit visits to the contributing centres [16]. The ELTR database does not include pre-transplant variables (i.e. renal support, ventilation, and prognostic scores) and alcoholic aetiology is not a grouping for ALF. We present analyses based on data from first transplants performed in adults (aged 16 years or over) for ALF between January 1988 and June 2009. Procedures and statistical analysis We used all ELTR variables whether continuous or categorical variables, related to the donor, the recipient, the graft, and the liver transplant technique. The donor variables were age, gender, type of donor (cadaveric, domino, living); recipient variables were: age, gender, UNOS status, and donor/recipient blood group matching (identical, compatible, incompatible); graft variables were: type of graft (full size, split liver, reduced liver, partial liver from living donor); variables related to liver transplant technique were: total ischemic time (minutes), type of transplant (orthotopic, heterotopic, and non auxiliary, auxiliary), type of preservation fluid, type of by-pass (extracorporeal, lateral clamping, none, other). Patients were grouped according to the aetiology of ALF as follows: virusrelated (HAV, HBV, HDV), paracetamol-related, other drug-related, other known causes (post-traumatic, post-operative, heat shock, toxic not drug-related, other), and unknown causes. Patients were also classified according to different time periods (1988–1993, 1994–1998, 1999–2003, and 2004–2009) to evaluate the evolution of recipient, donor, and surgical characteristics. The causes of death or graft failure after first liver transplant for ALF were evaluated: primary-non-function (PNF) or delayed function (PDF), intra-operative failure, technical problems, infection, acute or chronic rejection (combined in the analysis), renal disease, cardiovascular disease, pulmonary disease,

cerebrovascular disease, gastrointestinal disease, multi-organ failure, recurrence of primary liver disease, ‘‘de novo’’ tumours, lymphoproliferative disorders, ‘‘social’’ problems (non-adherence to immunosuppressive medications, suicide or trauma). The causes of death or graft failure were compared across different aetiologies and time periods. Patient survival according to aetiology of ALF and to time period was evaluated using the life-table method and comparisons made by the log-rank test. Discrete variables are shown as percentages and continuous variables as mean values ± SD, as all had a normal distribution (Kolmogorov–Smirnov test). Chi-square test was used to compare discrete variables, Student’s t-test for continuous variables, and ANOVA analyses for comparisons of two groups or more. Statistically significant differences were defined by a p-value 60.05. The effect of recipient, donor, and surgical variables on patient and graft survival was evaluated by Cox analysis. Variables with statistically significant risk-ratios at the univariate analysis (p value cut-off <0.05) were used for the subsequent multivariate analysis. Multivariate logistic regression was used to generate 3 and 12-month mortality scores in a randomly chosen training set (70% of transplants). We assessed model discrimination and calibration on an independent validation dataset (30% of transplants) using methods already applied to the ELTR database and previously published [14]. All the analyses were done with PASW statistics version 18.0.

Results There were 6374 patients transplanted for ALF between January 1988 and June 2009: 942 paediatric patients (<16 years) and

Table 1. Recipient, donor, surgical characteristics, and causes of death or graft loss according to the aetiology of acute liver failure (ALF) in the ELTR database.

Aetiology of ALF

Viral

Paracetamol

Other drugs

Other known

Unknown

n = 1029

n = 536

n = 496

n = 744

n = 2098

Recipient gender, % Male Female Recipient age (yr), mean ± SD

51.5 48.5 40.6 ± 13.1

33.2 66.8 34.2 ± 11.7

37.5 62.5 42.8 ± 14.8

41.7 58.3 42 ± 14.6

34.8 65.2 39.8 ± 13.5

<0.001

Donor gender (%) Male Female Donor age (yr), mean ± SD

55.3 44.7 40.1 ± 16.1

52.7 47.3 43.6 ± 16.1

55.1 44.9 42.6 ± 16.3

57.8 42.2 42.7 ± 16

56.4 43.6 39.9 ± 15.9

n.s.

Blood group matching (%) Identical Compatible Incompatible Total ischemic time (min), mean ± SD

73.4 19.2 7.3 516.3 ± 191.1

66.7 31.6 1.7 552.9 ± 183.8

72.6 22.8 4.5 523.5 ± 198

77.9 18.5 3.6 481.3 ± 152.3

69.5 23.6 6.9 524.1 ± 195.8

<0.001

32.5 25.4 41.2 0.9

18.8 61.9 19.3 0

19.4 34.8 45.5 0.3

22.4 30.8 46.1 0.7

32.8 23 43.9 0.2

<0.001

98.5 0.2 1.3

100 0 0

98.8 0.2 1

98.5 0.1 1.3

99.2 0 0.8

n.s.

92.5 3.6 2.5 1.3

92.4 3.1 4.5 0

91.4 3 4.4 1.2

94.5 2.7 1.4 1.4

93 3 3.1 0.8

0.02

98 2

96.1 3.9

98.2 1.8

98.6 1.4

97.7 2.3

0.05

99.4 0.6

98.9 1.1

99.2 0.8

99.7 0.3

99.4 0.6

n.s.

Type of by-pass (%) Extracorporeal Lateral clamping None Others Type of donor (%) Cadaveric Domino Living Graft type (%) Full size Split Reduced Partial for living Type of transplant (%) Non-auxiliary Auxiliary Type of transplant (%) Orthotopic Heterotopic

p value

<0.001

<0.001

<0.001

n.s., not significant.

Journal of Hepatology 2012 vol. 57 j 288–296

289

Research Article 529 patients with a second underlying liver disease were excluded, leaving 4903 patients for the analysis. Donor variables Male donors were 51.8% with a mean ± SD age of 41 ± 16 years (range: 2–88). Nearly 2/3 of liver transplants for ALF were performed using an identical donor/recipient ABO group match, whereas an incompatible ABO matching was used between 1.7% and 7.3% of cases across different aetiologies, being significantly higher in patients transplanted for virus-related ALF (7.3%) compared to other groups (paracetamol: 1.7%, p <0.001; other drugs: 4.5%, p = 0.015; other known: 3.6%, p <0.001; unknown: 6.9%, p <0.001) (Table 1). Recipient variables The aetiology of ALF was unknown in 43%, viral in 21%, paracetamol overdose in 11%, other drugs-related in 10%, and due to other known causes in 15%.

Female recipients were 61.2% with a mean ± SD age of 40 ± 14 years (range: 16–75). The proportion of female recipients was significantly higher in the paracetamol group (67%) compared to virus-related ALF (48.5%; p <0.001) and to ALF due to other known causes (58%; p = 0.002). Recipient age was significantly lower in the paracetamol group (34.2 ± 11.7 years) compared to other aetiologies (p <0.001 vs. all other groups) (Table 1). Surgical characteristics Whole cadaveric livers from brain stem dead donors were used in 89.6%. The split liver technique was used between 2.7% and 3.6%, living donor liver transplantation between 0% and 1.3%, and auxiliary transplant between 1.4% and 3.9% across different aetiologies. A reduced size graft was used more for paracetamol-related ALF (4.5%) compared to virus-related ALF (2.5%, p = 0.003) and to ALF due to other known causes (1.4%, p <0.001). No non-heart-beating donors were recorded in this cohort. Mean ± SD total ischaemic time was 518.9 ± 188 min (Table 1).

Table 2. Change over time of liver transplantation for acute liver failure in the ELTR database.

Transplantation era Recipient gender, % Male Female Recipient age (yr), mean ± SD Recipient age, % ≤30 31-40 41-50 51-60 >60 Donor gender, % Male Female Donor age (yr), mean ± SD Donor age, % ≤40 41-60 >60 Blood group matching, % Identical Compatible Incompatible Total ischemic time (min), mean ± SD Type of donor, % Cadaveric Domino Living Graft type, % Full size Split Reduced Partial for living Type of transplant, % Non-auxiliary Auxiliary

1988-1993

1994-1998

1999-2003

2004-2009*

n = 990

n = 1190

n = 1335

n = 1387

36.3 63.7 37.6 ± 13

38.4 61.6 38.9 ± 13.7

40.3 59.7 40.4 ± 13.8

39.6 60.4 41.7 ± 13.9

n.s.

37.7 22.7 22.2 14 3.3

33.4 22.2 21.1 17.5 5.8

29.2 22.9 21 19.2 7.7

25.2 24 22.5 19 9.3

<0.001

62.4 37.6 32.7 ± 13.4

57.3 42.7 39.3 ± 15.2

54.8 45.2 42.4 ± 15.8

51.4 48.6 46.1 ± 16.4

<0.001

70.2 27.9 1.8

48.9 43 8.1

42.3 46.5 11.3

35.3 43.7 21

<0.001

67 21.2 11.7 518.4 ± 219.7

70.7 24.4 4.9 533.4 ± 193

74.2 22.8 3 522 ± 187.7

73.1 22.3 4.5 503.7 ± 162.7

<0.001

99.9 0 0.1

99.7 0.1 0.2

98.7 0.1 1.2

98.1 0.1 1.8

<0.001

94.6 2 3.2 0.1

90.9 3.8 5 0.2

93.6 3.3 1.9 1.2

93 3.1 2.1 1.8

<0.001

99.2 0.8

96 4

98 2

98.1 1.9

<0.001

n.s., not significant. ⁄ Data available until 31/06/2009.

290

Journal of Hepatology 2012 vol. 57 j 288–296

p value

<0.001

<0.001

0.001

JOURNAL OF HEPATOLOGY Recipient age increased from 38 ± 13 to 42 ± 14 years (p <0.001), this effect being due to a significant increase in the number of recipients >60 years (3.3% to 9.3%; p <0.001), and decrease in those of 630 years (38% to 25%; p <0.001) (Table 2). Transplantation for paracetamol-induced ALF increased seven fold from 2% to 14.1%, other drug-related ALF from 9.3% to 11.8% and for other known causes from 6.9% to 22.4% (all p <0.001). However, this may represent improved diagnosis, as transplants for ALF due to unknown causes decreased from 59.5% to 33% (p <0.001) (Fig. 1). Liver transplantation for ALF due to HAV and HBV decreased significantly in the last 5 years (from 1% to 0.5% and from 17.9% to 13.2%, respectively). The number of living donor and auxiliary liver transplants significantly increased (p <0.001 and p = 0.02, respectively). Transplants performed with an incompatible donor/recipient ABO group match decreased from 11.7% to 4.5% (p <0.001). Total mean ischemic time also decreased from 518.4 ± 219.7 to 503.7 ± 162.7 min (p <0.001) (Table 2).

Evolution of liver transplantation for ALF over time We examined changes by comparing the latest period of transplantation (2004–2009) to the earliest (1988–1993) (Table 2). Donor age significantly increased over this time from a mean of 33 ± 13 to 46 ± 16 (p <0.001). Over the study period, this represents an increase in donor age of 2.5 years/year. Donors 640 years of age have halved from 70% to 35% (p <0.001), whereas donors >60 years have increased more than ten times from 1.8% to 21.1% (p <0.001) (Table 2). 1988-1993

1994-1998

22.3%

19.0%

2.0%

59.5%

52.5%

12.7%

9.3% 8.2%

6.9%

Virus

8.2%

Survival

Paracetamol 1999-2003

2004-2009

Other drugs

Overall patient survival at 1, 3, 5, and 10 years was 74%, 70%, 68%, 63%, respectively; overall graft survival at 1, 3, 5, and 10 years was 63%, 59%, 57%, 50%, respectively. The re-transplantation rate was 13%, decreasing from 20% in 1988–1993 to 6% in 2004–2009 (p <0.001). No statistically significant differences were found for overall patient and graft survival according to the aetiology of ALF. Patient survival at 1, 3, and 5 years between 2004–2009 was 79%, 75%, and 72%, respectively, significantly better than for patients transplanted between 1999–2003 (75%, 72%, and 70%, p = 0.01), or between 1994 and 1998 (69%, 67%, and 65%,

Other known Unknown

18.7%

24.2% 32.3%

33.0% 14.1%

12.7% 20.1% 10.8%

22.4% 11.8%

Fig. 1. Evolution of aetiology of acute liver failure as indication for liver transplantation, according to different time periods.

A Cumulative survival

1.0

4898 patients (log-rank p = 0.0001)

Number of exposed patients Years

0.8

1

2

3

4

5

6

7

8

9

10

11

12

574

546

520

509

496

485

471

450

427

402

372

356

0.6

1988- 991 1993

0.4

1994- 1190 1998

760

725

695

665

635

589

550

522

481

419

319

221

1999- 1334 2003

872

805

760

713

608

486

340

190

74

19

-

-

2004- 1383 2009

665

462

285

130

21

-

-

-

-

-

-

-

Total

2871

2538

2260

2017

1706

1560

1361

1162

982

840

691

577

0.2 0.0

0

Years of liver transplantation 2004-2009 1994-1998 1999-2003 1988-1993 0 2 4 6 8 10 12 Years after liver transplantation

B

4898

Number of exposed patients

Cumulative survival

1.0

Years

4898 patients (log-rank p = 0.0001)

6

7

8

9

10

11

12

0.8

1988- 991 1993

519

492

458

443

431

421

409

389

368

344

314

300

0.6

1994- 1190 1998

693

652

619

591

560

518

482

454

418

354

262

188

1999- 1334 2003

811

744

698

652

551

437

304

170

68

19

-

-

2004- 1383 2009

633

431

259

118

19

-

-

-

-

-

-

-

Total

2656

2319

2034

1804

1561

1376

1195

1013

854

717

576

488

0.4 0.2 0.0

Years of liver transplantation 2004-2009 1994-1998 1999-2003 1988-1993 0 2 4 6 8 10 12 Years after liver transplantation

0

4898

1

2

3

4

5

Fig. 2. Patient (A) and graft (B) survival after liver transplantation for acute liver failure according to different years of transplantation.

Journal of Hepatology 2012 vol. 57 j 288–296

291

Research Article Table 3. Causes of death or graft loss according to the aetiology of acute liver failure (ALF) in the ELTR database+.

Aetiology of ALF

Viral

Paracetamol

Other drugs

Other known

Unknown

p value

n = 387 (%)

n = 211 (%)

n = 203 (%)

n = 268 (%)

n = 905 (%)

Infection

73 (18.9)

26 (12.3)

38 (18.7)

55 (20.5)

168 (18.6)

n.s.

PNF or PDF

51 (13.2)

15 (7.1)

16 (7.9)

44 (16.4)

82 (9.1)

0.001

Intraoperative

13 (3.4)

10 (4.7)

9 (4.4)

18 (6.7)

31 (3.4)

n.s.

Technical

27 (7)

14 (6.6)

14 (6.9)

19 (7.1)

89 (9.8)

n.s.

Rejection

39 (10.1)

20 (9.5)

19 (9.4)

11 (4.1)

113 (12.5)

0.003

Early

Late Tumour “de novo”

12 (3.1)

2 (0.9)

3 (1.5)

2 (0.7)

14 (1.5)

n.s.

Lymphoproliferative

3 (0.8)

1 (0.5)

0 (0)

3 (1.1)

10 (1.1)

n.s.

Disease recurrence

23 (5.9)

0 (0)

2 (1)

5 (1.9)

15 (1.7)

<0.001 0.051

Other Cardiovascular

18 (4.7)

9 (4.3)

13 (6.4)

24 (9)

41 (4.5)

Pulmonary

14 (3.6)

9 (4.3)

11 (5.4)

9 (3.4)

35 (3.9)

n.s.

Cerebrovascular

41 (10.6)

20 (9.5)

10 (4.9)

22 (8.2)

68 (7.5)

n.s.

Gastrointestinal

6 (1.6)

10 (4.7)

8 (3.9)

4 (1.5)

11 (1.2)

0.003

Renal

6 (1.6)

4 (1.9)

0 (0)

3 (1.1)

12 (1.3)

n.s.

Multiple organ failure

20 (5.2)

20 (9.5)

11 (5.4)

18 (6.7)

72 (8)

n.s.

Social§

3 (0.8)

16 (7.6)

4 (2)

0 (0)

2 (0.2)

<0.001

Others

30 (8.1)

21 (10)

23 (11.8)

22 (9.7)

96 (11)

n.s.

Not available

7 (1.8)

14 (6.6)

21 (10.3)

5 (1.9)

42 (4.6)

<0.001

n.s., not significant. + Data available in 1974 patients. § Non-adherence to immunosuppressive medications, suicide, trauma.

p <0.001), or between 1988 and 1993 (66%, 62%, and 61%, p <0.001) (Fig. 2A). Graft survival at 1, 3, and 5 years was also significantly better between 2004 and 2009 (73%, 68%, and 63%, respectively) compared to 1999–2003 (67%, 63%, and 61%, p = 0.004), to 1994–1998 (59%, 55%, and 53%, p <0.001), and to 1988–1993 (51%, 47%, and 45%, p <0.001) (Fig. 2B). Overall, infections were the major cause of death/graft failure (18.2%), with no differences between aetiologies of ALF. PNF and PDF occurred more often in virus-related ALF (13.2%) and ALF due to other known causes (16.2%) compared to other groups (p <0.001). A significantly higher incidence of disease recurrence causing death or graft loss was seen in virus-related ALF (5.9%) compared to other groups (p <0.001), but decreased over time from 3.5% between 1988 and 1993 to 0.7% between 2004 and 2009 (p <0.001) in the whole cohort, and from 7.4% to 5% in the group of virus-related ALF. Importantly, the rate of ‘‘social’’ problems as causes of death or graft failure was nearly ten times higher in the paracetamol ALF (7.6%) compared to other aetiologies (p <0.001) (Table 3). Amongst this group of deaths, 81.3% were due to suicide, 12.5% due non-adherence to immunosuppressive medications and 6.3% due to trauma. Changes in causes of death or graft failure with respect to different eras of liver transplantation are shown in Table 4. The incidence both of acute and chronic rejection (p <0.001), cerebrovascular causes (p = 0.009) and disease recurrence (p = 0.055) has fallen progressively, but other causes were similar apart from an increase in PNF and PDF (p = 0.029), and gastrointestinal problems (p = 0.006) (Table 4). 292

Multivariate Cox regression Variables independently associated with death or graft loss were: recipient age >50 years (RR 1.26, 95% CI 1.10–1.44, p <0.001), incompatible donor-recipient group matching (RR 2.04, 95% CI 1.85–2.70, p <0.001), paracetamol-related ALF (RR 1.24, 95% CI 1.03–1.51, p = 0.027), ALF due to other known causes (RR 1.20, 95% CI 1.05–1.38, p = 0.007) and reduced graft size (split, reduced or partial for living) (RR 1.43, 95% CI 1.15–1.78, p <0.001) (Supplementary Table 1). Prognostic model for liver transplantation for acute liver failure Most deaths occurred within the first year (77% of overall deaths), with the majority occurring within the first three months (86% of first year deaths). Therefore, we generated a 3 and 12month predictive score, using the estimates from the multivariate logistic regression (Table 5A and B), to identify any potential risk factors of early death or graft loss, adjusted for liver transplantation era 2004–2009. For the 3-month model, we included 3300 patients randomly chosen (70.6%) in a training set, of whom 1462 (44.3%) had died by 3 months post-transplantation (Table 5A). The mean ± SD 3month predictive score was 0.81 ± 0.49 (0.65 ± 0.51 for those who died vs. 0.88 ± 0.46 for those alive; p <0.001). The C statistic was 0.63 and the Hosmer–Lemeshow statistic was 7.66 (p = 0.47), suggesting no evidence of lack of fit. When the same predictive score was calculated for the 1259 patients in the validation set (mean 0.80 ± 0.48), we noted similar results (C statistic = 0.63,

Journal of Hepatology 2012 vol. 57 j 288–296

JOURNAL OF HEPATOLOGY Table 4. Change over time of death and graft loss after liver transplantation for acute liver failure in the ELTR database+.

Transplantation era

1988-1993

1994-1998

1999-2003

2004-2009*

p value

n = 607 (%)

n = 600 (%)

n = 465 (%)

n = 302 (%)

Infection

111 (18.3)

112 (18.7)

73 (15.7)

64 (21.2)

PNF or PDF

46 (7.6)

67 (11.2)

60 (12.9)

35 (11.6)

0.029

Intraoperative

18 (3)

25 (4.2)

22 (4.7)

16 (5.3)

n.s.

Technical

55 (9.1)

48 (8)

37 (8)

23 (7.6)

n.s.

Rejection

86 (14.2)

70 (11.7)

31 (6.7)

15 (5)

<0.001

Early n.s.

Late Tumour “de novo”

12 (2)

9 (1.5)

12 (2.6)

0 (0)

n.s.

Lymphoproliferative

1 (0.2)

9 (1.5)

7 (1.5)

0 (0)

0.011

Disease recurrence

21 (3.5)

13 (2.2)

9 (1.9)

2 (0.7)

0.055

Other Cardiovascular

23 (3.8)

28 (4.7)

29 (6.2)

25 (8.3)

0.025

Pulmonary

20 (3.3)

24 (4)

23 (4.9)

11 (3.6)

n.s.

Cerebrovascular

66 (10.9)

33 (5.5)

37 (8)

25 (8.3)

0.009

Gastrointestinal

2 (0.3)

15 (2.5)

14 (3)

8 (2.6)

0.006

Renal

10 (1.6)

8 (1.3)

3 (0.6)

4 (1.3)

n.s.

Multiple organ failure

30 (4.9)

48 (8)

39 (8.4)

24 (7.9)

n.s.

Social§

4 (0.7)

9 (1.5)

7 (1.5)

5 (1.7)

n.s.

Others

56 (9.7)

60 (10.3)

46 (10.1)

30 (11.2)

n.s.

Not available

43 (7.1)

20 (3.3)

15 (3.2)

11 (3.6)

0.004

n.s., not significant. + Available in 1974 patients. ⁄ Data available until 31/06/2009. § Non-adherence to immunosuppressive medications, suicide, trauma.

Hosmer–Lemeshow test = 1.91 (p = 0.96)), indicating a similar discrimination, but better calibration than for the training sample. For the 12-month model, we included 3056 randomly chosen patients (70.2%) in the training set, of whom 1692 (55.3%) had died by 12 months (Table 5B). The mean ± SD predicted score in the training set was 0.51 ± 0.47 (0.37 ± 0.51 for those who died and 0.59 ± 0.44 for those alive at 12 months; p <0.001). The C statistic was 0.63 and the Hosmer–Lemeshow test statistic was 6.07 (p = 0.64). The score for the 1292 patients in the validation set gave a significant Hosmer–Lemeshow test (p = 0.014), suggesting that the 12-month score lacked calibration. Examination of the observed and expected number of deaths suggested that the 12-month predictive score generally underestimated the risk of mortality, although its discriminative ability remained good. The same analysis was performed in 1466 recipients older than 50 years, as they were at high risk of death or graft loss in the Cox regression analysis (Supplementary Table 1.). For the 3-month model, 1233 patients were included, of whom 535 (43.4%) had died by 3 months, whereas for the 12-month model, 1136 patients were included, of whom 535 (47.1%) had died by 12 months. The multivariate model at 3 months showed that male recipients (RR 1.54, 95% CI 1.19–1.99), patients transplanted with donors older than 60 years (RR 1.53, 95% CI 1.03–2.25) and an incompatible donor/recipient ABO group match (RR 2.52, 95% CI 1.43–4.44) were independent risk factors for death or graft loss. Similarly, the multivariable model at 12 months showed that male recipients (RR 1.69, 95% CI 1.31–2.19), donors older than 60 years (RR 1.46, 95% CI 1.01–2.20) and an incompatible donor/recipient ABO group match (RR 2.95, 95% CI 1.59–5.46) were independently

associated with death or graft loss. The C statistic from this model was 0.66 at three months and 0.63 at 12 months, and the Hosmer– Lemeshow statistic was 1.88 (p = 0.96) and 2.74 (p = 0.9), respectively, suggesting good calibration.

Discussion This study is the largest published cohort of patients transplanted for ALF, and gives a comprehensive overview of the evolution and the outcomes of liver transplantation for ALF in Europe. It spans over 20 years of liver transplantation and thus captures important changes in epidemiology [17,18], survival and other outcomes. It provided a tool to generate prognostic models based on clinical and demographic variables, which had good calibration, responding to the clinical need to identify patients in whom, due to the association of adverse risk factors, liver transplantation for ALF could be considered a procedure with unacceptable outcomes. ALF mortality rates before the era of liver transplantation ranged between 80% and 85% [19]. Current results of liver transplantation are very good considering the natural history of the disease, the multi-organ involvement, the emergency context, and the lack of other effective therapies [20]. However, long-term outcomes have not been reported extensively. To date 5-year patient survival rates are documented in only few studies [3,4,6,8,21,22] ranging between 61% and 76%, and only one study reported 10-year patient (69%) and graft (59%) survival rates [21]. In this study, the 1, 5, and 10-year patient survival rates were 74%, 68%, 63%, and the 1, 5 and 10-year graft survival rates were

Journal of Hepatology 2012 vol. 57 j 288–296

293

Research Article Table 5. Independent statistically significant variables from the multivariate logistic regression model associated with mortality or graft loss at 3 months (A) and at 12 months (B) after liver transplantation for acute liver failure in the two randomly selected training data sets.

A

3-month mortality (n = 3280) n, (%)

Intercept

Estimate

Odds ratio (95% CI)

-1.353

Recipient age (yr) ≤40 41-50 >50 Blood group matching Identical Compatible Incompatible Year of transplantation 2004-2009 1999-2003 1994-1998 1988-1993 ALF aetiology Viral Non-viral Preservation fluid Wisconsin Others Type of graft Full size Partial, split, reduced

0.78 (0.63-0.97)+

1716 (42.3) 721 (22) 843 (25.7)

-0.248 Reference 0.214

1.24 (0.98-1.57)+

2323 (72) 724 (22.4) 180 (5.6)

Reference 0.178 0.729

1.19 (0.98-1.46) 2.07 (1.49-2.89)*

827 (25.2) 895 (27.3) 855 (26.1) 703 (21.4)

Reference 0.431 0.720 1.056

1.54 (1.19-1.98)° 2.05 (1.59-2.65)* 2.88 (2.22-3.73)*

578 (17.6) 2702 (82.4)

Reference 0.361

1.44 (1.15-1.79)°

2680 (86.6) 414 (13.4)

Reference 0.385

1.47 (1.15-1.88)*

2945 (93.1) 218 (6.9)

Reference 0.527

1.69 (1.24-2.31)*

*p <0.001; °<0.01; + <0.05

B

12-month mortality (n = 3056) n, (%)

Estimate

Odds ratio (95% CI)

Intercept

-0.674

Recipient gender (male)

0.277

1.32 (1.11-1.57)°

1605 (52.5) 666 (21.8) 785 (25.7)

-0.228 Reference 0.258

0.80 (0.64-0.99)+

2171 (72) 669 (22.2) 175 (5.8)

Reference 0.081 0.838

1.08 (0.88-1.34) 2.31 (1.61-3.32)*

648 (21.2) 879 (28.8) 819 (26.8) 710 (23.2)

Reference 0.229 0.632 0.825

1.26 (0.97-1.63) 1.88 (1.44-2.45)* 2.28 (1.71-3.04)*

1395 (49.3) 1150 (40.7) 284 (10)

-0.276 Reference 0.099

549 (18) 2507 (82)

Reference 0.264

1.3 (1.05-1.62)°

2542 (88) 347 (12)

Reference 0.437

1.55 (1.19-2.01)*

2745 (93.1) 205 (6.9)

Reference 0.583

1.79 (1.28-2.50)*

Recipient age (yr) ≤40 41-50 >50 Blood group matching Identical Compatible Incompatible Year of transplantation 2004-2009 1999-2003 1994-1998 1988-1993 Donor age ≤40 41-60 >60 ALF aetiology Viral Non-viral Preservation fluid Wisconsin Others Type of graft Full size Partial, split, reduced +

*p<0.001; °<0.01; <0.05

294

Journal of Hepatology 2012 vol. 57 j 288–296

1.29 (1.02-1.65)+

0.76 (0.63-0.92)° 1.1 (0.82-1.5)

JOURNAL OF HEPATOLOGY 63%, 57%, 50%. In the United States, between 1997 and 2004, the 1-year and 5-year patient and graft survival rates were 82%, 70% and 76%, 61%, respectively [2]. Based on the same time-interval in the ELTR database, the five-year patient and graft survival rates were similar at 69% and 60%, whereas the 1-year survival rates were lower at 74% and 66%, respectively. All patients had a minimum of 1 year follow-up and thus time periods are comparable, without bias of varying follow-up times. The differences in survival occur in this first year, and then survival curves remain parallel. This suggests that improvements are related to pre-, periand immediately post-operative care. The ELTR data showed a constant and progressive improvement in the survival rate over time, with only small changes in aetiology: HAV- and HBV-related ALF decreased over time and ALF due to paracetamol increased. These improved outcomes are not only important ‘‘per se’’, but is crucial to understand that this occurred despite a dramatic increase in the donor age, with donors older than 60 years being nearly ten times more numerous between 2004 and 2009 than between 1988 and 1993. These findings reflect the worldwide increase of age of donors [23], which is an established adverse factor for both elective and emergency liver transplantation [14]. Thus, the increased prevalence of this adverse factor has been compensated by a substantial improvement in the pre-, peri- and post-operative patient management of ALF with liver transplantation. Thus, it is unlikely there is a single or most important factor responsible for the improved survival after liver transplantation for ALF in the most recent 5 years. ALF patients have always been considered super-urgent and are not under the MELD system, thus the allocation system has not changed. Selection also probably has not changed, as it is unlikely that sicker patients were not selected, but this cannot be verified from the ELTR database. The introduction of new immunosuppressants may have contributed to improved survival. This is suggested by a significant reduction in rejection incidence as a cause of death or graft loss (from 14.2% in 1988–1993 to 5% in 2004–2009). However, in addition, improved anaesthesia and intensive care will have contributed, as it has been shown following liver transplantation for cirrhosis, where progressive improvements in survival have been documented in Europe [14]. Despite the significant improvement in patient and graft survival over time, a high early mortality rate still persists after transplantation for ALF, especially within the first year. Therefore, we decided to identify potential risk factors for poor prognosis in the early post-transplant period and to generate 3- and 12-month prognostic models with the aim to stratify patients according to the risk of death or graft loss. The 3-month model was well calibrated, but the 12-month model showed a tendency to underestimate mortality rates. The most likely explanation for the lack of calibration in the 12-month model is that other factors such as recurrent disease and renal function post-operatively become important and peri-operative variables lose their importance. Moreover, the lack of information regarding the clinical status of the recipient before liver transplantation such as ventilation status, renal function, presence/absence of cerebral oedema, which are not recorded by the ELTR, could weaken the predictive value of the model and is a limitation of the models. Recipients above 50 years were at risk for increased mortality, confirming data from 236 patients in a single centre [3], and from 1457 patients (UNOS database, 1988–2003) [22], and it can be probably explained by an age-related reduction in physiological reserve [3]. Therefore, a sub-analysis was performed in recipients

>50 years, based on the 3 and 12-month prognostic models. Considering mortality or graft loss rate of more than 50% at 12 months as an unacceptable outcome, there are several combinations of three risk factors, when liver transplantation for ALF could be considered futile, especially when all risk factors are present (35% and 20% of death or graft loss at 3 and 12 months, respectively) (Fig. 3). Thus, an ABO incompatible graft remains an important adverse factor for patient and graft survival confirming previous studies [24], but in contrast to two others [25,26]. The database does not include data on therapies used to prevent consequences of ABO incompatible mismatch. Recently, a more selective approach, based on antigen-specific immunoadsorption with immunoadsorbent columns with or without rituximab, has been proposed in order to reduce the high risk of sepsis occurring with previous regimens [27]. In addition, in the whole cohort, a reduced size graft (split liver, reduced, or partial for living) compared to a full size graft, increased

A

Donor age >60 years

Donor age >60 years

32% Male recipient

33%

Male recipient

43% Incompatible ABO match

44%

Incompatible ABO match

55%

Male recipient

Incompatible ABO match

55%

Male recipient

Incompatible ABO match

65%

Donor age >60 years

Donor age >60 years

Hosmer and Lemeshow test = 1.88 (0.96) Area under the ROC = 0.66

B

Donor age >60 years

Donor age >60 years

44% Male recipient

47%

Male recipient

57% Incompatible ABO match

61%

Incompatible ABO match

70%

Male recipient

Incompatible ABO match

73%

Male recipient

Incompatible ABO match

80%

Donor age >60 years

Donor age >60 years

Hosmer and Lemeshow test = 2.74 (0.90) Area under the ROC = 0.63 Fig. 3. Risk of 3-month (A) and 12-month (B) mortality or graft loss in patients older than 50 years transplanted for ALF according to the presence of independently associated risk factors.

Journal of Hepatology 2012 vol. 57 j 288–296

295

Research Article mortality or graft loss by 1.5 times. In order to reduce the mortality risk in this specific sub-group of recipients >50 years, donors >60 years should be avoided, as well as use of reduced size grafts. ABO incompatible matching could be also circumvented [28]. The results of our models may help clinicians to decide when an outcome might be unacceptable as another recipient may benefit far more from use of a particular donor liver. However, the decisions must always be focused on the individual patient, and the model predictions should not be used as guidelines. Although in an emergency situation, it may be difficult not to choose to use particular donors, the centre, and the national transplant community need to review the use of older donors for certain recipients with ALF, as the outcomes may not justify the choices taken. Lastly, we showed that in Europe there has been a significant and constant increase in liver transplantation for paracetamolrelated ALF. Among these patients, nearly 8% died or lost the graft for ‘‘social’’ complications, with suicide and non-adherence to immunosuppressive medications accounting for nearly 94% of these deaths, 57% of which occur within the first year after transplant (and 40% within the first three months). Therefore, in the early post-transplant phase, more psychological/psychiatric, and social support should be offered to patients transplanted for paracetamol overdose, to lessen graft loss and death. In conclusion, despite a progressive and constant improvement in the survival rate after liver transplantation for ALF, a high mortality and graft loss rate still persist, especially within the first year post-transplant. Data shown in our analysis strengthens the importance of a re-evaluation of the selection of older patients with ALF as potential recipients for liver transplantation. This is particularly so, as outcomes with older donors >60 years (currently 21% of all donors) result in poor survival. Therefore, a strategy to improve outcomes could be firstly to identify patients, estimated to have an unacceptable high risk of death or graft loss within one year of liver transplantation based on utilitarian principles, and secondly to avoid in these patients, and particularly in those older than 50 years (especially if males), use of grafts from donors >60 years, ABO incompatible matching (without novel technological solutions) and a reduced size graft. Conflict of interest The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jhep. 2012.03.017. References [1] Available from . [2] 2009 Annual Report of the US Organ Procurement Transplant Network and Scientific Registry of Transplant Recipients: transplant data 1999–2008, . [3] Bernal W, Cross TJ, Auzinger G, Sizer E, Heneghan MA, Bowles M, et al. Outcome after wait-listing for emergency liver transplantation in acute liver failure: a single centre experience. J Hepatol 2009;50:306–313.

296

[4] Bismuth H, Samuel D, Castaing D, Adam R, Saliba F, Johann M, et al. Orthotopic liver transplantation in fulminant and subfulminant hepatitis. The Paul Brousse experience. Ann Surg 1995;222:109–119. [5] O’Grady JG, Alexander GJ, Thick M, Potter D, Calne RY, Williams R. Outcome of orthotopic liver transplantation in the aetiological and clinical variants of acute liver failure. Q J Med 1988;68:817–824. [6] Wigg AJ, Gunson BK, Mutimer DJ. Outcomes following liver transplantation for seronegative acute liver failure: experience during a 12-year period with more than 100 patients. Liver Transpl 2005;11:27–34. [7] Ascher NL, Lake JR, Emond JC, Roberts JP. Liver transplantation for fulminant hepatic failure. Arch Surg 1993;128:677–682. [8] Farmer DG, Anselmo DM, Ghobrial RM, Yersiz H, McDiarmid SV, Cao C, et al. Liver transplantation for fulminant hepatic failure: experience with more than 200 patients over a 17-year period. Ann Surg 2003;237:666–675, [discussion 675–666]. [9] Bjoro K, Ericzon BG, Kirkegaard P, Hockerstedt K, Soderdahl G, Olausson M, et al. Highly urgent liver transplantation: possible impact of donor–recipient ABO matching on the outcome after transplantation. Transplantation 2003;75:347–353. [10] Emond JC, Aran PP, Whitington PF, Broelsch CE, Baker AL. Liver transplantation in the management of fulminant hepatic failure. Gastroenterology 1989;96:1583–1588. [11] Lee WM. Acute liver failure in the United States. Semin Liver Dis 2003;23:217–226. [12] Bernal W, Auzinger G, Dhawan A, Wendon J. Acute liver failure. Lancet 2010;376:190–201. [13] Adam R, McMaster P, O’Grady JG, Castaing D, Klempnauer JL, Jamieson N, et al. Evolution of liver transplantation in Europe: report of the European Liver Transplant Registry. Liver Transpl 2003;9:1231–1243. [14] Burroughs AK, Sabin CA, Rolles K, Delvart V, Karam V, Buckels J, et al. 3month and 12-month mortality after first liver transplant in adults in Europe: predictive models for outcome. Lancet 2006;367:225–232. [15] Adam R, Cailliez V, Majno P, Karam V, McMaster P, Caine RY, et al. Normalised intrinsic mortality risk in liver transplantation: European Liver Transplant Registry study. Lancet 2000;356:621–627. [16] Karam V, Gunson B, Roggen F, Grande L, Wannoff W, Janssen M, et al. Quality control of the European Liver Transplant Registry: results of audit visits to the contributing centers. Transplantation 2003;75:2167–2173. [17] Escorsell A, Mas A, de la Mata M. Acute liver failure in Spain: analysis of 267 cases. Liver Transpl 2007;13:1389–1395. [18] Ichai P, Samuel D. Epidemiology of liver failure. Clin Res Hepatol Gastroenterol 2011;35:610–617. [19] Bernuau J, Rueff B, Benhamou JP. Fulminant and subfulminant liver failure: definitions and causes. Semin Liver Dis 1986;6:97–106. [20] O’Grady JG. Postoperative issues and outcome for acute liver failure. Liver Transpl 2008;14 (Suppl 2):S97–101. [21] Chan G, Taqi A, Marotta P, Levstik M, McAlister V, Wall W, et al. Long-term outcomes of emergency liver transplantation for acute liver failure. Liver Transpl 2009;15:1696–1702. [22] Barshes NR, Lee TC, Balkrishnan R, Karpen SJ, Carter BA, Goss JA. Risk stratification of adult patients undergoing orthotopic liver transplantation for fulminant hepatic failure. Transplantation 2006;81:195–201. [23] Merion RM. Current status and future of liver transplantation. Semin Liver Dis 2010;30:411–421. [24] Bismuth H, Samuel D, Castaing D, Williams R, Pereira SP. Liver transplantation in Europe for patients with acute liver failure. Semin Liver Dis 1996;16:415–425. [25] Farges O, Kalil AN, Samuel D, Saliba F, Arulnaden JL, Debat P, et al. The use of ABO-incompatible grafts in liver transplantation: a life-saving procedure in highly selected patients. Transplantation 1995;59:1124–1133. [26] Toso C, Al-Qahtani M, Alsaif FA, Bigam DL, Meeberg GA, James Shapiro AM, et al. ABO-incompatible liver transplantation for critically ill adult patients. Transpl Int 2007;20:675–681. [27] Saliba F, Ichai P, Azoulay D, Habbouchi H, Antonini T, Sebagh M, et al. Successful long-term outcome of ABO-incompatible liver transplantation using antigen-specific immunoadsorption columns. Ther Apher Dial 2010;14:116–123. [28] Tanabe M, Kawachi S, Obara H, Shinoda M, Hibi T, Kitagawa Y, et al. Current progress in ABO-incompatible liver transplantation. Eur J Clin Invest 2010;40:943–949.

Journal of Hepatology 2012 vol. 57 j 288–296