Predictive factors of early postoperative graft function in human liver transplantation

Predictive Factors of Early Postoperative Graft Function in Human Liver Transplantation

F. &VIER GONZALEZ,~ ANTON1 RIMOLA,~ LUIS GRANDE,' W I A fblTOLIN,3 JUAN C. GARCIA-VALDECASAS,~ JOSE FUSTER,~ ANTONIO M. LACY,' ESTEBAN CUGAT,~ JOSE VISA' AND JOAN RODES~

'Department of Surgery, 2Liver Unit and 3Laboratory of Biochemistry, Hospital Clinic i Provincial of Barcelona, University of Barcelona, 08036 Barcelona, Spain

To identify factors predictive of early postoperative graft function, we analyzed 54 variables-including easily available clinical and laboratory data prospec- tively obtained from organ donors, transplant re- cipients and surgical procedures in 168 consecutive liver transplantations. Early postoperative graft func- tion was classified into three groups according to a scoring system ranging from 3 to 9 based on peak serum ALT values, mean bile output and lowest prothrombin activity measured during the 72 hr after transplant: group 1 (score 3 to 4, good graft function; n = 73), group 2 (score 5 to 6, moderate dysfunction; n = 50) and group 3 (score, 7 to 9, severe dysfunction; n = 45). In univariate analyses, 8 of the 54 variables analyzed were statistically significant (p < 0.05) predictors of severe graft dysfunction: high serum sodium concen- tration and brain death caused by cranial trauma in organ donors, advanced age and low prothrombin activity in transplant recipients, prolonged total ischemia time and large transfusions of red blood cells, fresh frozen plasma and platelets during surgery. After introduction of these eight variables in a multivariate analysis, only four were found to independently predict early postoperative graft function: donor serum sodium concentration, total ischemia time, platelet transfusion during surgery and recipient prothrombin activity. In 52 liver transplantations, in which the predictive value of liver tissue adenine nucleotide con- centration and several biochemical sensitive markers of donor nutritional status was also analyzed, only the ATP level in liver tissue obtained at the time of organ reperfusion was identified as an independent predictor of initial graft function. We conclude that the degree of early postoperative graft dysfunction in liver trans- plantation can be predicted on the basis of data from organ donors, transplant recipients and surgical events. (HEPATOLOGY 1994;20:565-573.)

Received September 9, 1993; accepted March 7, 1994. Other abbreviations used in the text: UW, University of Wisconsin. F.X. Gonzalez is a recipient of a grant from the Hospital Clinic i Provincial

of Barcelona. Address reprint requests to: F. Xavier Gonzaez, M.D., Department of

Surgery, Hospital Clinic i Provincial, Villarroel 170, 08036 Barcelona, Spain. Copyright 8 1994 by the American Association for the Study of Liver

Diseases. 0270-9139/94 $3.00 + 0 31/1/56795

Liver transplantation has become the treatment of choice for patients with advanced chronic liver disease and for most patients with acute liver failure (1). Survival of liver transplant recipients has been reported to be greater than 70% at 12 mo and 60% at 36 mo after transplant in different series published in recent years (1-4). However, in spite of these globally satisfactory results, several important problems still arise in patients undergoing liver transplantation. One of these problems is the great variability in early postoperative graft function observed from patient to patient with emer- gency liver transplantation required in most patients in whom severe initial graft dysfunction develops (5-7).

Several retrospective studies showed no agreement in identifying predictive factors of early postoperative graft function in human liver transplantation. Whereas some authors reported that none of numerous variables from donor and recipient subjects was useful in predicting initial graft function (8, 91, other authors found that some of these variables correlated with graft function (10-15). In addition, variables identified as predictors in these latter investigations were not the same in every study. The different results obtained in these studies may result from their being based on retrospective reviews of patient records and the data from organ donors probably being obtained at very different times in relation to organ harvest. In this context, it is important to note that many laboratory parameters assessing liver and kidney function can change rapidly in the hours before organ procurement, depending on the possible occurrence of events, particularly hemodynamic and metabolic derangements, adversely affecting liver and kidney function during this period. Therefore in this study we prospectively investigated the possible rela- tionship between early postoperative graft function and a large number of easily available clinical and laboratory parameters from organ donors, transplant recipients and surgical procedure in a series of consecutive liver transplants, in which the data from donor and recipient patients were obtained immediately before organ har- vest and immediately before transplant, respectively.

Recent experimental and clinical investigations have demonstrated that initial liver graft function correlated with energy metabolism alterations secondary to the

565

GONZALEZ ET AL

50 - 40 - 30 - 20 - 10 - 0

HEPATOLOGY September 1994

Group 111

p=0.027

1 . 1 . 1 . I . I . I .

566

> t =! m a m

L

A a z a > 3 to

% '1

l o o 90 h Group I

Group II 60

MONTHS

FIG. 1. Probability of graft survival in liver transplantation with good (group l), regular (group 2) and bad (group 3) early postoperative graft function.

graft ischemia occurring during organ harvesting and recipient operation (16-20). Similarly, a direct rela- tionship between donor nutritional status and (both) liver tissue adenine nucleotide maintenance and liver graft function has also been found in experimental studies (21-25). Therefore in this s tudy we also investi- gated the value of liver tissue adenine nucleotide levels and clinical and laboratory data assessing the nutri- tional state of organ donors in predicting early postop- erative function of liver grafts.

PATIENTS AND METHODS Between November 1989 and October 1992,178 consecutive

liver transplants were performed. Criteria used for donor liver acceptance were similar to those habitually recommended by most authors (14, 15, 26). Although donor selection criteria were not rigid, in general liver donors were accepted when the following criteria were fulfilled: (a) age younger than 60 yr; (b) no history of liver disease, alcoholism or drug abuse; and (c) serum bilirubin less than 2.5 mg/dl, serum aminotransferases less than 160 U/L (4 times upper normal limit in our laboratory) and serum alkaline phosphatase less than 300 U/L for adult donors and less than 450 U/L for child donors (upper normal limit in our laboratory, 290 U/L and 400 UIL, respectively). Higher serum aminotransferase levels did not systematically preclude donor acceptance if the level clearly showed a tendency to decline. No donor with evident bacterial infection, or with HBsAg, hepatitis C virus antibody or HIV antibody positivity was accepted. ABO blood group compat- ibility and relative size-matching between donor and recipient were required, except in emergency liver transplantation. Donor livers with macroscopic fatty infiltration or other significant alterations or marked atherosclerosis of hepatic artery, celiac axis or both observed at the time of graft harvest were systematically rejected, regardless of liver function test values. All donor livers were procured with well-described

TABLE 1. Parameters used for the calculation of scoring system assessing early postoperative graft function in liver

transplant patients Parameter Assigned value

Serum ALT (U/LY < 1,000 1 1,000-2,500 2 > 2,500 3

Bile output (ml/day)* > 100 1 40-100 2 < 40 3

> 60 (spontaneous) 1 > 60 (with fresh-frozen plasma) < 60 (despite fresh-frozen plasma)

Prothrombin activity (%Y

2 3

"Highest value in the 72 hr after transplant. Normal value in our

*Mean value during the first 72 hr after transplant. 'Lowest value in the 72 hr after transplant.

laboratory: < 40 U/L.

techniques for multiple organ harvest (27, 28). Ninety-one livers were preserved in UW solution and 87 in combined UW and Euro-Collins solutions (UW solution was administered through the portal vein (2 L), and Euro-Co\\ins solution through the hepatic artery). Both preservation solutions have recently been found to have similar effectiveness in human liver transplantation when graft ischemia time is relatively short (29). Orthotopic liver transplant procedure in the recipient was performed with standard techniques (28). In 56 patients (33%) venovenous bypass was used because patients had acute liver failure (bypass may reduce the risk of increased intracranial pressure due to excessive fluid administration) or presented hemodynamic instability during portal vein and inferior caval vein clamping. All patients had similar periop- erative intensive care and immunosuppression therapy. The perioperative immunosuppression regimen consisted of intra- venous methylprednisolone (1,000 mg intraoperatively, fol- lowed by a tapering schedule from 200 mg to 20 mg in the first 6 postoperative days), intravenous azathioprine (1.5 mg/kg/day) and cyclosporine (intravenous perfusion started on the second or third postoperative day at an initial dose of 4 m g w d a y , with rapid adjustments according to blood levels of the drug).

Early postoperative graft function was graded with a scoring system previously described (29, 30). This score was obtained from the peak serum ALT value, mean bile output and lowest prothrombin activity measured in the first 72 postoperative hr (Table 1). In each patient, the score was calculated from the sum of the assigned values for each parameter, with a possible range from 3 to 9. On the basis of this score, patients were arbitrarily classified into three groups: group 1 (good early graft function) when the score was 3 or 4, group 2 (moderate graft dysfunction) when the score was 5 or 6 and group 3 (severe graft dysfunction) when the score ranged from 7 to 9. Ten patients were excluded from the analysis of the results for the following reasons. In six patients, hypovolemic shock developed as a result of massive postoperative hemorrhage (rupture of hepatic subcapsular hematoma in one case, and bleeding from caval vein anastomosis and from unidentified source regardless of exploratory laparotomy in 1 and 4 cases, respectively) were excluded because this complication could have affected early postoperative graft function. In all these six

HEPATOLOGY Vol. 20, No. 3, 1994 GONZALEZ ET AL. 567

TABLE 2. Quantitative variables from organ donors analyzed as predictive factors of early postoperative graft function No. of patients

~ ~ ~

Variable Median (range) Cutoff value 5 cutoff >cutoff

Age (F) Dopamine infusion (mg/kg/min) ICU stay (days) Serum AST (U/L) Serum ALT (UL) Serum alkaline phosphatase (U/L) Serum y-glutamyl transpeptidase (UL) Serum bilirubin (mg/dl) Serum total protein (gm/L) Serum albumin (gm/L) Prothrombin activity (%Id Serum glucose (mg/dl) Serum creatinine (mg/dl) BUN (mgidl) Serum sodium (mEq/L) Serum potassium (mEq/L) Hematocrit value (%) Hemoglobin concentration (gm/L) White blood cell count (103/mm3)

25 (5-65) 7 (0-35) 2 (0.5-20)

44 (4-543) 34 (6-772)

119 (19-752) 18 (3-401)

0.8 (0.1-6.0) 54 (24-85) 32 (19-54) 72 (16-100)

126 (42-480) 0.9 (0.4-4.0) 20 (4-70)

147 (116-187) 3.8 (1.2-7.3) 33 (11-57) 11 (3-24)

11.4 (0.9-32.7)

45" 15O 3"

40" 40"

290 40 1.2

62 35 65"

107 1.2

25 145

4.5 36 11.5 11.0

145 (86%) 131 (78%) 124 (74%)

74 (44%) 108 (64%) 158 (94%) 141 (84%) 131 (78%) 52 (31%)

106 (63%) 70 (43%) 52 (31%)

133 (79%) 118 (70%) 71 (42%)

130 (77%) 106 (63%) 89 (53%) 81 (48%)

23 (14%) 37 (22%) 44 (26%) 94 (56%)b 60 (36%)' 10 (6%) 27 (16%) 37 (22%)

116 (69%) 62 (37%) 92 (5710)

116 (69%) 35 (21%) 50 (30%) 97 (58%) 38 (23%) 62 (37%) 79 (47%) 87 (52%)

"These cutoff values were selected according to results of previous studies (5, 12, 15,60). The remaining cutoff values represent normal limits in our laboratory.

12 cases, serum AST level was greater than 160 U/L. 'In seven cases, serum ALT level was greater than 160 U/L. dProthrombin activity could not be determined in six cases.

patients, graft function (assessed on the basis of prothrombin activity and bile flow) after organ reperfusion and during the first postoperative hours was little or moderately altered, but it rapidly worsened concomitantly with the development of hypovolemic shock. Two more patients were excluded because of death on the second postoperative day. (The cause of death was brain hemorrhage in both cases.) Finally, two other patients (one with primary sclerosing cholangitis and one undergoing liver retransplantation because of severe biliary tract complications in the first graft) were excluded because bile flow could not be measured since biliary reconstruction consisting of hepaticojejunostomy without external biliary drainage was performed. Therefore the study included the remaining 168 liver transplants, involving 146 patients. Fifty-four clinical and laboratory parameters from organ donors, transplant recipients and surgical procedures were investigated as predictive factors of early graft function. With the exception of the graft preservation solution and the necessity of venovenous bypass (both mentioned above), all these parameters are shown in Tables 2 to 5. Blood samples for determining donor laboratory data were obtained within a few minutes before initiation of donor laparotomy and maintained at 4" C until they were assayed shortly after the arrival of the donor graft extraction team at our hospital. Clinical and laboratory data of liver transplant recipients were obtained a few hours before surgery.

In a subgroup of 52 transplants, liver tissue adenine nucleotides were determined. For this purpose, liver tissue (approximately 20 mg), obtained by means of Tru-cut needle (Baxter Healthcare Corp., Pharmaseal Division, Valencia, CA) biopsies a t the time of both donor laparotomy and within 2 min before organ reperfusion in the recipient, was rapidly placed and stored in liquid nitrogen ( - 176" C) until it could be analyzed. Liver concentrations of ATP, ADP and AMP and purine catabolites (xanthine and hypoxanthine) were mea-

sured by means of HPLC (31, 32). Adenylate energy charge was calculated with the equation (ATP + %ADP)/(ATP + ADP + AMP). The sum of liver tissue nucleotides was calcu- lated by means of the simple addition of ATP, ADP and AMP. In the same 52 liver transplants, donor serum prealbumin, transferrin and retinol-binding protein levels were determined as sensitive biochemical markers of the nutritional status.

Univariate analyses (Kruskal-Wallis test for quantitative variables and x2 test for qualitative variables) were used for the comparison of parameters investigated as predictors of early liver graft function in the three groups of patients classified according to the initial graft function score. Variables found to be statistically significant (p < 0.05) in univariate analyses were subsequently introduced in a multivariate analysis (logistic-regression analysis) to identify independent predictive factors of early graft function. For performing logistic re- gression analysis, the whole series of patients was divided into two groups: one including patients with good or regular initial graft function (score, 3 to 6) and another group including patients with severe initial graft dysfunction (score, 7 to 9). Probability curves of graft survival were calculated according to the Kaplan-Meier method, and the curves were statistically compared by means of the Mantel-Cox test. All these analyses were performed with the BMDP statistical package (programs 3S, 4F, LR and 1L) (33). Results are expressed as median (range).

RESULTS Characteristics of Liver Donors and Transplant Re-

cipients. Tables 2 to 4 depict clinical and laboratory data from organ donors and transplant recipients. As shown in Tables 2 and 3, most liver donors were young, had short hospital stays before liver procurement, required low doses of vasoactive drugs and had normal or

568 GONZALEZ ET AL. HEPATOLOGY September 1994

TABLE 3. Qualitative variables from organ donors and transplant recipients analyzed as predictive factors of early

postoperative graft function

Variable No. of

patients (%I

Donor Sex

Male Female

Cause of brain death Traumatic Vascularianoxic

Nutritional status (clinically assessed) Prolonged arterial hypotension

Good Regular Poor

Recipient Sex

Male Female

Transplant indication Chronic cholestasis Other chronic liver disease Acute liver failure Retransplantation

Type of transplantation Emergency Elective

History of Ascites Gastrointestinal hemorrhage Hepatic encephalopathy Upper abdominal surgery

Good Regular Poor

A B C

Identical Compatible nonidentical Incompatible

Nutritional status

Child-Pugh classification‘

Donor-recipient A130 compatibility

120 (71) 48 (29)

114 (68) 54 (32) 32 (19)

135 (80) 30 (18)

3 (2)

92 (55) 76 (45)

26“ (15)

21 (13) 22 (13)

28 (17) 140 (83)

94 (56) 48 (29) 55 (33) 48 (29)

61 (36) 70 (42) 37 (22)

14 (11) 56 (45) 55 (44)

146 (87) 16 (10) 6 (3)

996 (59)

~ ~ ~ ~ ~~ ~ ~ ~ ~ ~ ~

“One patient also had a cholangiocarcinoma. bTwenty patients also had hepatocellular carcinoma. ‘Not calculated in patients with acute liver failure or those who

underwent retransplantation.

moderately altered liver function test results. However, in a relatively important proportion of liver donors there was at least one feature that might have made these donors be considered suboptimal, such as age greater than 45 yr, a long interval between hospital admission and liver procurement ( > 3 days), prolonged arterial hypotensive episodes, high vasoactive drug doses and marked alterations in liver function test results, par- ticularly high aminotransferase levels ( > 160 UL).

Indications for the 168 liver transplantations included in the study were as follows. Twenty-six (15%) patients

had chronic cholestatic diseases (22, PBC; 3, primary sclerosing cholangitis; and 1, Caroli’s disease), 98 (58%) had cirrhosis of hepatocellular nature (31, alcoholic cirrhosis; 10, HBsAg-associated cirrhosis; 34, cirrhosis associated with HCV antibody positivity; 2, autoimmune cirrhosis; 1, cirrhosis associated with porphyria cutanea tarda; 1, cirrhosis resulting from a,-antitrypsin defi- ciency; 19, cryptogenic cirrhosis), 1 (1%) had idiopathic portal hypertension and 21 (13%) had acute liver failure. One patient with primary sclerosing cholangitis also had cholangiocarcinoma, and 20 patients with cirrhosis also had hepatocellular carcinoma. The remaining 22 liver transplants (13%) were retransplantations. Indications for retransplantations were primary graft nonfunction in seven cases, graft rejection in nine, and miscellaneous disorders in six.

Cold, warm and total ischemia times and operative blood product transfusion requirements are shown in Table 5.

Factors Predictive of Early Graft Function. Ac- cording to the criteria described for scoring and grouping of postoperative liver graft function, there were 73 grafts (43%) in group 1, 50 (30%) in group 2 and 45 (27%) in group 3. Standard liver function tests and bile output in these three groups are shown in Table 6. Most patients from group 3 experienced gradual improvement in liver function after the third postoperative day. However, in seven patients of this group (16%), graft function continued to be impaired, with progressive increases in serum aminotransferase levels, absence of bile pro- duction and impossibility of maintaining prothrombin activity greater than 40% to 50% in spite of large amounts of fresh-frozen plasma. These seven patients were considered to have primary graft nonfunction, and they underwent emergency retransplantation. All ex- planted liver grafts showed massive or submassive necrosis.

During follow-up (median, 15 mo; range, 0.1 to 39 mo), 18 grafts (25%) from group 1, 13 (26%) from group 2 and 19 (42%) from group 3 were lost. Reasons for graft loss were retransplantation in 18 cases (primary non- function in 7, rejection in 6 and miscellaneous causes in 51, patient death due to graft failure in 2 cases (rejection in 1 and HBV infection recurrence in 1) and patient death due to causes unrelated to liver graft dysfunction in the remaining 30 cases. The probability of graft survival in the three groups is shown in Figure 1.

Among the 54 variables investigated in the univariate analyses as factors predictive of initial graft function in the whole series of 168 liver transplantations, the following 8 variables were found to significantly cor- relate with early postoperative function of the graft: cause of brain death and serum sodium concentration in liver donors, age and prothrombin activity in transplant recipients, total ischemia time and intraoperative trans- fusion of RBCs, fresh-frozen plasma and platelets (Table 7). In multivariate analysis, only four of these eight variables were identified as independent predictors of initial graft function: donor serum sodium concen- tration (p = O.OOO), total ischemia time (p = 0.0031,

HEPATOLOGY Vol. 20, No. 3, 1994 GONZALEZ ET AL. 569

TABLE 4. Quantitative variables from transplant recipients analyzed as factors predictive of early postoperative graft function

Variable Median Range

Age (Yr) Serum bilirubin (mg/dl) Serum AST (U/L) Serum ALT (U/L) Serum alkaline phosphatase (U/L) Serum y-glutamyl transpeptidase (U/L) Serum cholesterol (mg/dl) Serum total protein (gm/L) Serum albumin (gm/L) Prothrombin activity (%) Platelet count ( x 1,000/mm3) Serum creatinine (mg/dl) BUN (mg/dl)

48

106 88

337 66

145 69 30 54 90

18

4.1

0.9

17-62 0.2-57.2 14-8,121 12-10,994

121-5,833 152,628 33-1,066 45-94 18-56 8-100

21-375 0.2-3.6

6-80

TABLE 5. Surgical variables analyzed as factors predictive of early postoperative graft function

Variable

No. of patients

Median (range) Cutoff value" 5 cutoff z cutoff ~ ~~ ~~

Graft ischemia times (min) Cold 313 (105-1,155) 420 126 (75%) 42 (25%) Warm 48 (25-90) 60 152 (90%) 16 (10%) TOTAL 360 (150-1,190) 480 128 (76%) 40 (24%)

RBCs (units) 13 (0-72) 13 89 (53%) 79 (47%) Fresh-frozen plasma (L) 5 (0-21) 5 80 (48%) 88 (52%) Platelets (units) 10 (0-77) 10 90 (54%) 78 (46%)

Transfusion of blood products

"Cutoff values for ischemia times were selected according to Avolio et al. (12). Cutoff values for transfusion requirements correspond to the median values in the whole series.

platelet transfusion during the operation (p = 0.004) and recipient prothrombin activity (p = 0.031).

On consideration of the subgroup of 52 liver trans- plantations in whom liver tissue adenine nucleotide levels and biochemical markers of donor nutritional status were analyzed, only ATP and ADP concentrations and the energy charge in liver tissue obtained immedi- ately before organ reperfusion showed a significant correlation with early postoperative graft function in univariate analysis. In patients from groups 1 (n = 211, 2 (n = 20) and 3 (n = 111, the median (range) prere- perfusion values of ATP were 2.8 (0.8 to 5.31, 2.3 (0.6 to 6.7) and 0.8 (0.3 to 2.6) nmol/mg protein, respectively (p = 0.000). ADP values were 5.1 (3 to 7),4.1(2.3 to 9.4) and 2.8 (1.7 to 4.9) nmoVmg protein, respectively (p = 0.001). Energy charge values were 0.3 (0.2 to 0.51, 0.3 (0.1 to 0.5) and 0.2 (0.1 to 0.4) nmoVmg protein, respectively (p = 0.004). After introduction of these three variables in the multivariate analysis, liver tissue ATP level before reperfusion was the only variable that independently correlated with initial graft function (p = 0.000).

Table 8 shows the values of the five variables identified as independent predictors of early graft function (total ischemia time, donor serum sodium concentration, surgical platelet transfusion, recipient prothrombin activity and prereperfusion liver tissue ATP) in the

seven cases with primary nonfunction. In all these patients the value of at least one variable was worse than the median group 3 values for the same variables.

DISCUSSION An intriguing phenomenon in liver transplantation is

the great variability in initial graft function, which ranges from mild dysfunction to severe hepatic insuffi- ciency requiring emergency retransplantation. This great variability theoretically depends on the state of the graft function before its removal from the donor, the magnitude of the hepatic ischemia-reperfusion injury occurring between organ harvest and reperfusion in the transplant recipient or both (34). One additional problem is the method used to assess the degree of early postoperative graft dysfunction. In human liver trans- plantation, this assessment is currently made on the basis of the serum levels of aminotransferases and bile production and coagulation status, alone or in combi- nation. In one recent study, Greig et al. (5) classified initial graft function in transplanted livers into four grades according to the aforementioned parameters: grade I was defined by an AST level less than 1,000 U/L, bile production greater than 40 ml/day and improving coagulation; in grade 11, initial AST was greater than 1,000 U/L but decreased over the next 2 days, with improving bile production and coagulopathy; in grade

570 GONZALEZ ET AL. HEPATOLOGY September 1994

TABLE 6. Postoperative standard liver function tests and bile output in the three liver transplant groups classified according to initial graft function

Parameters Group 1 (n = 73) Group 2 (n = 50) Group 3 (n = 46)

Serum bilirubin (mgidl)". * 6.5 (1.5-27.7)' 7.9 (2.2-28.0) 12.4 (3.2-32.7) Serum AST (U/L)", * 357 (60-1,670) 496 (56-2,099) Serum ALT (U/LP 338 (52-1,218) 476 (55-3,078) 2,071 (88-9,140) Serum alkaline phosphatase (U/L)" 185 (69-766) 191 (89-930) 216 (69-1,430) Serum y-glutamyl transpeptidase (U/L)" 72 (24-418) 83 (26-414) Prothrombin activity ' 63 (42-100) 53 (21-85) 40 (20-66) Bile output (mu24 h r P e 167 (20-483) 89 (0-601) 40 (0-220)

1,912 (144-9,580)

100 (28-481)

"Highest value registered in the 3 days after surgery bp < 0.05 (calculated from Kruskal-Wallis test). 'Lowest value registered in the 3 days after surgery. dData expressed as median (range). 'Mean value in the 3 days after surgery.

TABLE 7. Variables with significant predictive value for early postoperative graft function in 168 liver transplantations

Group 1 (n = 73) Group 2 (n = 60) Group 3 (n = 45) p Value= Variables ~~ ~ ~~

Cause of donor brain death Cranial trauma Other

Donor serum sodium (mEq/L)* Recipient age (yr)' Recipient prothrombin activity (%)* Total ischemia time (min)* Surgical RBC transfusion (units)' Surgical plasma transfusion (L)* Surgical platelet transfusion (units)'

~ ~ ~~~

58 (79%) 15 (21%)

144 (116-176) 52 (18-62) 62 (8-100)

345 (150-1,190) 13 (2-60)

5.4 (0.9-17.5) 10 (0-40)

~~ ~

3 1 (62%) 19 (38%)

147 (128-169) 45 (19-59) 70 (46-94)

354 (205-933) 12 (0-56)

4.6 (0-21.0) 10 (0-50)

~~~ ~

25 (56%) 20 (44%)

47 (17-60) 46 (8-100)

155 (133-187)

390 (220-1,080) 15 (4-72)

6.2 (1.9-17.5) 20 (0-77)

0.0149 0.0001 0.0074 0.0004 0.0334 0.0268 0.0435 0.0478

"Calculated from x2 test for qualitative variables and Kruskal-Wallis test for quantitative variables. 'Data expressed as median (range).

111, AST greater than 2,500 U/L, bile flow less than 40 mVday and severe coagulopathy were present; and in grade N a rapid increase in AST level, no bile flow and severe coagulopathy were present. Greig's classification, although it is the most accurate system to date, is not completely adequate for assessing early postoperative graft function for two reasons: the lack of a clear definition of coagulopathy and the possibility of a given patient fulfilling characteristics of several grades of liver dysfunction. Therefore, in this study, as well as in previously published investigations from our group (29, 301, we have modified Greig's classification with the inclusion of well-defined cutoff values for each pa- rameter and the assignment of one to three points for each category obtained to calculate a score for every individual transplant recipient (Table 1). In addition, transfusion of fresh-frozen plasma, a very common practice during the first postoperative days that can substantially modify coagulation factor levels in this period of time, has been included in our classification. The scoring system proposed in our investigation, which represents an attempt to standardize the assessment of early postoperative graft function, allowed us to classify transplanted livers into three groups with different initial dysfunction (mild, moderate and severe) and graft outcome, as shown in Table 6 and Figure 1.

As expected from previous clinical and experimental data (5, 16, 19, 20, 35-38), in this study the total ischemia time and the degree of alteration in the liver energy metabolism were identified as independent pre- dictors of early postoperative function of liver grafts. Despite the fact that graft ischemia times were not excessively prolonged in this series, a significant inverse correlation was observed between total ischemia time and initial graft function, suggesting that short periods of ischemia are advisable in human liver transplantation even when preservation solutions allowing very pro- longed ischemia times are used. On the other hand, ATP concentration in liver tissue obtained immediately before graft reperfusion was found to correlate directly with early graft function. This correlation can be explained on the basis of the fact that liver ATP levels before graft reperfusion probably reflect the overall impact of every adverse event on liver graft energy metabolism occurring between donor hospitalization and organ implantation, including the deleterious effect of the ischemia period (18, 35, 39).

The surgical requirement of platelet transfusion was another parameter found to independently correlate with initial graft function. Two features of liver trans- plantation may explain this relationship. First, several investigations have shown that endothelial cells lining

HEPATOLOGY Vol. 20, No. 3, 1994 GONZALEZ ET AL. 571

TABLE 8. Values of the five variables identified 88 independent predictors of early postoperative graft function in the seven cases with primary nonfunction

Case no. (min) (mEqAJ (units) activity (%) (nmol/mg protein)" Total ischemia time Donor serum sodium Platelet transfusion Recipient prothrombin Liver tissue ATP level

1 455 152 20 50 0.56 2 1,030 153 2 93 0.85 3 815 157 0 21 ND 4 300 154 30 25 ND 5 455 180 20 47 ND 6 360 150 30 25 ND 7 300 164 0 92 ND

ND, not determined. "Liver tissue obtained immediately before organ reperfusion

hepatic sinusoids are the cells most rapidly and severely injured after cold ischemic storage and subsequent reperfusion of the graft and that the intensity of microcirculation damage is related to the degree of organ dysfunction and survival (37, 38, 40-42). Second, a marked decrease in platelet count during the postreper- fusion period of liver transplantation has consistently been reported by several authors (36,43,44). It has been speculated that thrombocytopenia in liver transplant patients is due to platelet entrapment in the liver graft because of the alterations of the sinusoid endothelial cells (36, 44,45). Therefore, because in liver transplan- tation the severity of endothelial cell damage is probably related to the degree of platelet consumption, our finding that platelet transfusion predicted initial graft function is not surprising. In agreement with this result, McCaughan et al. recently reported that allograft dys- function was the most important predictive factor of the platelet count decrease after liver transplantation (46).

In our series, increased serum sodium levels in liver donors before organ harvest were strongly associated with marked initial graft dysfunction in transplant recipients. This finding agrees with the results previ- ously obtained by Avolio et al., who reported a direct correlation between donor serum sodium concentration and the peak levels of postoperative serum aminotrans- ferases in human liver transplantation (12). The most probable reasons for donor hypernatremia are inade- quate general management of organ donors; the ag- gressive treatment of cerebral edema, yielding an im- portant negative water balance; and the drastic decrease of antidiuretic hormone secretion as a consequence of brain death (47-50). Although there is no clear expla- nation for the relationship between donor serum sodium concentration and early postoperative graft function, it is possible that markedly increased serum sodium levels reflect careless general management of organ donors, which would be the underlying reason for poor graft quality and, consequently, poor initial graft function. However, an alternative explanation may also be hy- pothesized, as follows. During hypernatremia, all mam- malian cells investigated to date (neurons, renal tubular cells and muscle cells) increase their intracellular osmo- lality to minimize the possible damage associated with the extracellular hypertonic state, mainly through two

adaptive changes: increased influx of sodium and cyto- plasmic accumulation of several idiogenic osmoles, particularly amino acids, methylamines and polyols (51-55). These changes are reversed after the adequate correction of hypernatremia. However, whereas intra- cellular sodium concentration rapidly normalizes, cyto- plasmic idiogenic osmole content is normalized in several days, in such a way that a too-rapid hypertonic state correction is followed by intracellular water accu- mulation because of the persistence of increased intra- cellular osmolality, with consequent cell damage (56). Although there is no study investigating liver cells in different serum osmolality states, it is possible that these cells also participate in the aforementioned adaptive mechanisms against hypernatremia. In this case, a liver graft obtained from a hypernatremic donor may be damaged after being suddenly placed in the habitually normotonic milieu of the transplant recipient because of an increase in its intracellular water content due to the persistence of the cytoplasmic accumulation of idiogenic osmoles and, possibly, to compromised sodium efflux as a consequence of the energy me- tabolism derangement occurring during ischemia (34, 57). This possibility is supported by the results obtained in an experimental study on liver transplantation in pigs demonstrating that nonviable grafts had greater intra- cellular sodium and water content after organ reper- fusion than viable grafts (58). Further investigations are needed to ascertain this hypothesis. Nevertheless, whatever the reasons by which donor serum sodium strongly correlates with early postoperative graft func- tion, the recommendation of avoiding marked hyper- natremia in organ donors and correcting hypertonic state after brain death declaration may be raised.

Preoperative prothrombin activity in liver transplant recipients was also found to be statistically significant as an independent predictor of initial graft function. However, there is no satisfactory explanation for this finding.

Interestingly, no relationship was found between early postoperative graft function and donor liver function test values. The same result has been reported by other authors (8-10,13-15); it may be due to the fact that liver donors are usually accepted in the absence of marked alterations in liver function test values or

572 GONZALEZ ET AL. HEPATOLOGY September 1994

serious liver damage. Other donor-related factors such as old age, prolonged time in the intensive care unit or requirement of high doses of vasoactive drugs (which have been found to negatively influence initial graft function in several studies [5 , 10, 11, 13, 15, 591, although not in others [8,9, 12,601) were not identified as bad predictors of early postoperative graft function in this investigation. Because age, intensive care unit stay time and dopamine dose in our donors were not substantially different than those reported in other investigations, a possible explanation for the lack of correlation between these factors and the initial graft function is the habitually short graft ischemia time used in our study, which could have blunted the potentially deleterious effect of these factors. Nevertheless, the possibility that this lack of correlation could be due to the relatively small number of donors showing these factors can not be excluded. Finally, in contrast with recently reported experimental data (21-25), no clinical and laboratory parameter assessing donor nutritional status had significant value in predicting early postop- erative liver graft function in our series.

In summary, the degree of early postoperative graft function in liver transplantation can be predicted on the basis of data from organ donors, transplant recipients and surgical events. In this investigation, prolonged graft ischemia time, high platelet transfusion require- ments during surgery, increased donor serum sodium concentration, reduced prothrombin activity in trans- plant recipients and decreased liver tissue ATP level before graft reperfusion have been found to be predictors of marked initial graft dysfunction in human liver transplantation. However, the predictive value of these variables should be reassessed in further prospective investigations from other liver transplant groups.

1.

2.

3.

4.

5 .

6.

7.

8.

9.

REFERENCES

Cosimi AB. Update on liver transplantation. Transplant Proc

Iwatsuki S, Starzl TE. Todo S, Gordon RD, Esquivel CO, Tzakis AG, Makowka L, et al. Experience in 1000 liver transplants under cyclosporine-steroid therapy: a survival report. Transplant Proc

Kilpe VE, Krakauer H, Wren RE. An analysis of liver transplant experience from 37 transplant centers as reported to Medicare. Transplantation 1993;56:554-561. Goldstein RM, Olson LM, Klintmalm GBG, Husberg BS, Nery JR, Gonwa TA, Roden JS, et al. Decreased mortality associated with orthotopic liver transplantation. Transplant Proc 1988;20:

Greig PD, Woolf GM, Sinclair SB, Abecassis M, Strasberg SM, Taylor BR, Blends LM, et al. Treatment of primary liver graft nonfunction with prostaglandin E,. Transplantation 1989;48: 447-453. Kennedy EM, Wood RP, Shaw BW Jr . Primary nonfunction. Is there a contribution from the back table bath? Transplantation

Shaw SW, Gordon RD, Iwatsuki S, Starzl TE. Hepatic retrans- plantation. Transplant Proc 1985;17:264-271. Makowka L, Gordon RD, Todo S, Ohkohchi N, Marsh JW, Tzakis AG, Yokoi H, et al. Analysis of donor criteria for the prediction of outcome in clinical liver transplantation. Transplant Proc 1987;

Grazi GL, Jovine E, Frena A, Grandi U, Bellusci R, Cavallari A,

1991 ;23:2083-2090.

1988;1:498-504.

505-507.

1990;49:739-743.

1 :2378-2382.

10.

11.

12.

13

14

15

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

Mazziotti A , et al. Influence of donor data on the outcome of liver transplantation. Transplant Proc 1991;23:2483-2484. Pruim J, Van Woerden WF, Knol E, Klompmaker IJ, de Bruijn KM, Persijn GG, Slooff MJH. Donor data in liver grafts with primary non function: a preliminary analysis by the European Liver Registry. Transplant Proc 1989;21:2383-2384. Mimeault R, Grant D, Ghent C, Duff J, Wall W. Analysis of donor and recipient variables and early graft function after orthotopic liver transplantation. Transplant Proc 1989;21:3355. Avolio AW, Agnes S, Magalini SC, Foco M, Castagneto M. Importance of donor blood chemistry data (AST, serum sodium) in predicting liver transplant outcome. Transplant Proc 199 1;23:

Ploeg RJ, Alessandro AM, Knechtle SJ, Stegall MD, Pirsch JD, Hoffmann RM, Sasaki T, et al. Malfunction of the liver after transplantation: an analysis of potential risk factors. Transplant Proc 1993;25:1659-1661. Yanaga K, Tzakis AG, Starzl TE. Personal experience with the procurement of 132 liver allografts. Transplant Int 1989;2:

Mor E, Klintmalm GB, Gonwa TA, Solomon H, Holman MJ, Gibbs JF, Watemberg I. The use of marginal donors for liver transplan- tation. Transplantation 1992;53:383-386. Kamiike W, Nakahara M, Nakao K, Koseki M, Nishida T, Kawashima Y, Watanabe F, et al. Correlation between cellular ATP level and bile excretion in the rat liver. Transplantation

Harvey PRC, Iu S, McKeown CMB, Petrunka CN, Ilson RG, Strasberg SM. Adenine nucleotide tissue concentrations and liver allograft viability after cold presewation and warm ischemia. Transplantation 1988;45: 1016- 1020. Kamiike W, Burdelski M, Steinhoff G, Ringe B, Lauchart W, Pichlmayr R. Adenine nucleotide metabolism and its relation to organ viability in human liver transplantation. Transplantation

Lanir A, Jenkins RL, Caldwell C, Lee RG, Khettry U, Clouse ME. Hepatic transplantation survival: correlation with adenine nucleo- tide level in donor liver. HEPATOLOGY 1988;8:471-475. Sumimoto K, Inagaki K, Yamada K, Kawasaki T, Dohi K. Reliable indices for the determination of viability of grafted liver immedi- ately after orthotopic transplantation. Transplantation 1988;46:

Palombo JD, Hirschberg Y, Pomposelli JJ, Blackburn GL, Zeisel SH, Bistrian BR. Decreased loss of liver adenosine triphosphate during hypothermic preservation in rats pretreated with glucose: implications for organ donor management. Gastroenterology

Palombo JD, Hirschberg Y, Pomposelli JJ , Blackburn GL, Zeisel SH, Bistrian BR. Decreased loss of liver nucleotides and energy charge during hypothermic preservation by donor pretreatment with glucose. Transplant Proc 1989;21:1299-1300. Adam R, Reynes M, Bao YIvl, Astarcioglu I, Azoulay D, Chiche L, Bismuth H. Impact of glycogen content of the donor liver in clinical liver transplantation. Transplant Proc 1993;25: 1536- 1537. Cywes R, Greig PD, Morgan GR, Sanabria JR, Clavien PA, Harvey RC, Strasberg SM. Rapid donor liver nutritional enhancement in a large animal model. HEPATOLOGY 1992;16:1271-1279. Boudjema K, Lindell SL, Southard JH, Belzer FO. The effects of fasting on the quality of liver preservation by simple cold storage. Transplantation 1990;50:943-948. Pruim J, Klompmaker IJ, Haagsma, Bijleveld MA, Slooff JH. Selection criteria for liver donation: a review. Transplant Int

Starzl TE, Hakala TR, Shaw BW, Hardesty RL, Rosenthal TJ, Griffith BP, Iwatsuki S, et al. A flexible procedure for multiple cadaveric organ procurement. Surg Gynecol Obstet 1984; 158:

Starzl TE, Iwatsuki S, Esquivel CO, Todo S, Kam I, Lynch S, Gordon RD, et al. Refinements in the surgical technique of liver transplantation. Semin Liver Dis 1985;5:349-356. Garcia-Valdecasas JC, Gonzdez FX, Grande L, Rimola A, Navasa M, Fuster J , Lacy AM, et al. The use of the University of Wisconsin

2451-2452.

137-142.

1985;39:50-55.

1988;45:138-143.

506-509.

1988;95: 1043-1049.

1993;6:226-235.

223-230.

GONZALEZ ET AL. 573 HEPATOLOGY Vol. 20, No. 3, 1994

and Euro-Collins solutions either alone or in a combined method. Transplant Int 1992;5:77-80.

30. Grande L, Rimola A, Garcia-Valdecasas JC, Mas A, Fuster J , Navasa M, Lacy A, et al. Initial poor function in liver graft can recover fully. Transplantation 1992;53:228-230.

31. Wynants J , Van Belle H. Single-run high-performance liquid chromatography of nucleotides, nucleosides, and major purine bases and its application to different tissue extracts. Anal Biochem

32. Sauter AJ. High performance liquid chromatographic determi- nation of adenine nucleotides in biological materials: improve-

1985;144:258-266.

ments and adaptations to routine analysis. J Chromatog-1985; 325:314-316.

33. Dixon WJ. BMDP statistical software manual. Berkeley, CA University of California Press, 1985.

34. Clavien PA, Harvey PR, Strasberg SM. Preservation and reper- fusion injuries in liver allografts. Transplantation 1992;53:957- 978.

35. Marubayashi S, Takenaka M, Dohi K, Ezaki H, Kawasaki T. Adenine nucleotide metabolism during hepatic ischemia and subsequent blood reflow periods and its relation to organ viability. Transplantation 1980;30:294-296.

36. Iu S, Harvey PRC, Makowka L, Petrunka CN, Ikon RG, Strasberg SM. Markers of allograft viability in the rat. Transplantation

37. Momii S, Koga A. Time-related morphological changes in cold- 1987;44:562-569.

stored rat livers. Transolantation 1990:50: 745-750. 38.

39.

40.

41.

42.

43.

Caldwell-Kenkel JC, Currin RT, Tana’ka Y, Thurman RG, Le- masters JJ. Kupffer cell activation and endothelial cell damage after storage of rat livers: effects of reperfusion. HEPATOLOGY

Karwinski W, Astrid-Mette M, Farstad M, Soreide 0. Sixty minutes of normothermic ischemia in the rat liver: correlation between adenine nucleotides and bile excretion. J Surg Res

Thurman RG, Bunzendahl H, Lemasters JJ . Role of sinusoidal lining cells in hepatic reperfusion injury following cold storage and transplantation. Semin Liver Dis 1993;13:93-100. Manner M, Shult W, Senninger N, Machens G, Otto G. Evaluation of preservation damage after porcine liver transplantation by assessment of hepatic microcirculation. Transplantation 1990;6:

Ikeda T, Yanaga K, Kishikawa K, Kakizoe S, Shimada M, Sugi- machi K. Ischemic injury in liver transplantation: difference in injury sites between warm and cold ischemia in rats. HEPATOLOGY

1991;13:83-95.

1989;46:99-103.

940-943.

1992;16:454-461. Owen CA. Rettke SR. BowieJW. Cole TL. Jensen CC. Wiesner RH.

44. Porte RJ, Knot EA, Bontempo FA. Hemostasis in liver transplan- tation. Gastroenterology 1989;97:488-501.

45. Hutchinson DE, Genton E, Porter KA. Platelet changes following clinical and experimental hepatic homotransplantation. Arch Surg

46. McCaughan GW, Herkes R, Powers B, Rickard K, Gallagher ND, Thompson JF, Sheil AGR. Thrombocytopenia post liver trans- plantation. J Hepatol 1992;16:16-22.

47. Wijnen RMH, Van Der Linden CJ. Donor treatment after pronouncement of brain death: a neglected intensive care problem. Transplant Int 1991;4:186-190.

48. Jennett B, Teasdale G. Management of head injuries. In: Con- temporary neurology series. Vol 20. Philadelphia: F.A. Davis,

49. Fiser DH, Jimenez JF, Wrape V, Woody R. Diabetes insipidus in children with brain death. Crit Care Med 1987;15:551-553.

50. Soifer BE, Gelb AW. The multiple organ donor: identification and management. Ann Intern Med 1989;110:814-823.

51. Heilig CW, Stromski ME, Blumenfeld JD, Lee JP, Gullans SR. Characterization of the major brain osmolytes that accumulate in salt-loaded rats. Am J Physiol 1989;257F:1108-1116.

52. Cserr HF, DePasquale M, Nicholson CH, Patlak CS, Pettigrew KD, Rice ME. Extracellular volume decreases while cell volume is maintained by ion uptake in rat brain during acute hyper- natremia. J Physiol 1991;442:277-295.

53. Cserr HF, DePascuale M, Patlak CS. Regulation of brain water and electrolytes during acute hyperosmolality in rats. Am J Physiol 1987;253:F522-F529.

54. Thurston JH, Hauhart RE, Naccarato EF. Taurine: possible role in osmotic regulation of mammalian heart. Science 1981;214:

55. Lohr JW, McReynolds J , Grimaldi T, Acara M. Effect of acute and chronic hypernatremia on myoinositol and sorbitol concentration in rat brain and kidney. Life Sci 1988;43:271-276.

56. Lien YHH, Shapiro JI, Chan L. Effects of hypernatremia on organic brain osmoles. J Clin Invest 1990;85:1427-1435.

57. Blankensteijn JD, Terpstra OT. Liver preservation: the past and the future. HEPATOLOGY 1991;6:1235-1250.

58. Hoshino T, Maley WR, Labs JD, Clemens MG, Burdick JF, Williams GM. Useful parameters for predicting posttransplant liver viability. Transplant Proc 1988;20:987-991.

59. Alexander JW, Vaughn WK. The use of “marginal” donors for organ transplantation. Transplantation 1991;51: 135-141.

60. Wall WJ, Mimeault R, Grand DR, Bloch M. The use of older donor livers for hepatic transplantation. Transplantation 1990;49:377- 381.

1968;97:27-33.

1986:259-266.

1373-1374.

Krom RA. Hemostatic evaluation of patients undergoing liver transplantation. Mayo Clin Proc 1987;62:761-772.