Independent risk factors predicting acute graft rejection in cardiac transplant recipients treated by triple drug immunosuppression

J

THoRAc CARDIOVASC SURG

1989;98:1113-21

Independent risk factors predicting acute graft rejection in cardiac transplant recipients treated by triple drug immunosuppression To assess independent risk factors predicting the occurrence of clinically significant acute rejection episodes in the first 6 months after cardiac transplantation, we performed a multivariate stepwise logistic regression analysis. Forty-three recipients, undergoing transplantation between September 1986 and May 1988, were eligible for analysis and received standardized, low-dose triple drug maintenance immunosuppression with cyclosporine, azathioprine, and prednisolone. Immusoprophylaxls was supplemented perioperatively with either a polyclonal (antithymocyte globulin, N = 26) or a monoclonal (OKT3, N = 17) anti-T-cell antibody. Investigated, conceivable risk factors comprised recipient and donor age, ischemic time, perioperative anti-T-cell antibody prophylaxis, recipient preoperative status, underlying disease, previous cardiac operation, and histocompatibility parameters (mismatches for HLA-A, HLA-B, HLA-DR, HLA-B+DR, HLA-A+B+DR, and RhJD] antigen, HLA-DRw6 positive recipient, and identity for ABO system). Univariate analysis suggested significant influence of the type of antibody used perioperatively (p = 0.0024) and the number of mismatches for HLA-A+B+DR (p = 0.0037) and for HLA-B+DR (p = 0.0043). Stepwise logistic regression yielded the number of mismatches for HLA-B+DR (p = 0.0029) and the type of antibody used perioperatively (p = 0.0031) as being highly significant predictors of acute cardiac rejection. Six-month freedom from rejection was 100%,41 %, and 27% for recipients with two, three, and four mismatches for HLA-B+DR and 59% versus 22 % for recipients with polyclonal versus monoclonal antibody prophylaxis. Similar to results with kidney transplantation, these results indicate that a poor donor/recipient match for combined HLA-B+DR loci constitutes an independent risk factor for acute graft rejection in low-dose triple drug immunosuppressed cardiac recipients, which stimulates the potential concept of prospective HLA matching. In our experience OlIT3 prophylaxis provides significantly less effective prevention of acute rejection than a comparable course of antithymocyte globulin.

Giinther Laufer, MD, Johannes Miholic, MD, Axel Laczkovics, MD, Gregor Wollenek, MD, Christoph Holzinger, MD, Agathe Hajek-Rosenmeier, MD,a Gabriele Wuzl, MD, Wolfgang Schreiner, PhD, Peter Buxbaum, MD, and Ernst Wolner, MD, Vienna, Austria

h e results of cardiac transplantation have improved in the past few years, as documented by the annual official report from the International Society for Heart Transplantation.' This improvement is unequivocally related to the introduction of cyclosporine into maintenance Department of Surgery II and Institute for Blood Group Serology: University of Vienna. Vienna. Austria. Received for publication Dec. 19. J 988. Accepted for publication March 2 I. 1989. Address for reprints: Giinther Laufer. MD. Department of Surgery II. The University of Vienna. Spitalgasse 23. A-I 090 Vienna. Austria.

12/1/13095

immunotherapy. Regarding cyclosporine-containing immunosuppressive schedules, triple drug maintenance immunotherapy involving the combined use of cyclosporine, azathioprine, and prednisolone has been demonstrated to provide superior actuarial l-year survival rates in the worldwide cardiac transplant registry.? Apparently, better results have been achieved by a reduction of acute rejection- and infection-related deaths. Low-dose triple drug therapy allows safe employment of cyclosporine in low doses.' which minimizes drug-associated adverse effects such as nephrotoxicity, hepatotoxicity, and arterial hypertension. Despite improved maintenance immunosuppression, 1 1 13

The

1 1 14

Surgery

Table I. Characteristics ofpatient

population-Continuous variables Variable

Recipient age Mean ± SD Range Donor age Mean ± SD Range Ischemictime Mean ± SD Range

Journal of

Thoracic and Cardiovascular

Laufer et al.

No rejection

Rejection

Total

IN= 21)

IN=22)

IN= 43)

43.4 ± 11.5 17 ± 56

45.5 ± 7.1 35 ± 59

44.4 ± 9.5 17 ± 59

29.5 ± 8.0 16 ± 46

26.0 ± 7.2 15 ± 41

27.7 ± 7.7 15 ± 46

127.1 ± 44.1 75±215

132.0 ± 39.9 60 ± 210

129.6 ± 41.6 60 ± 215

acute rejection remains a definite problem after cardiac transplantation and impacts on morbidity and mortality. Death from acute rejection still accounts for 24.4% of all causes of death in the worldwide registry.? and acute rejection has been identified as the most powerful risk factor for death in another study." Additionally, frequent acute rejection episodes may compromise long-term graft function and promote graft atherosclerosis.' This underlines the importance of a further decline of acute rejection episodes, which may be achieved by influencing important risk factors for the rejection process. To identify independent, significant risk factors for the occurrence of an acute rejection episode within 6 months of transplantation, we conducted a multivariate stepwise logistic regression analysis. The evaluation of incremental risk factors predicting clinically relevant allograft rejection under conditions of optimized (triple dug) immunosuppression was the primary goal of this study. The knowledge of such risk factors might contribute to an improvement in the results of cardiac transplantation.

Patients and methods Study population. Between March 1984 and November 1988, 101 orthotopic heart transplants were performed in 99 recipients at the Second Department of Surgery, University of Vienna. Starting in May 1986, all patients received standard, low-dose triple drug immunosuppression with cyclosporine, azathioprine, and prednisolone, supplemented prophylactically with either antithymocyte globulin (ATG, rabbit or horse) or monoclonal murine antibody OKT3 in the perioperative period. Since June 1987 antibodies were administered according to a prospective randomization (polyclonal versus monoclonal). Consecutive patients undergoing transplantation from September 1986 to May 1988 are the basis of this investigation. Fortytwo patients received low-dose triple drug immunosuppression and had a follow-up period of at least 6 months. One other patient died of unrecognized acute rejection (sudden death) after 5 weeks. These 43 recipients fulfilled the criteria for participation in the study. Four additional patients who underwent transplantation in this period died within 6 months of causes not related to acute rejection: bacterial sepsis, acute pancreatitis,

postoperative multiorgan failure, and an unknown cause (rejection ruled out) and were excluded from the analysis. Additionally, a 6-year-old child underwent orthotopic transplantation in the same period but was treated with a different immunosuppression protocol (no perioperative antibody prophylaxis) and thus did not enter the study. All recipients were examined on a monthly basis in the outpatient clinic of our institution. No patient was lost to follow-up. Prophylactic immunosuppression. Immunosuppression was conducted in three stages as preoperative, perioperative, and maintenance immunotherapy. Four different groups of immunosuppressants were used: cyclosporine, azathioprine, steroids. anti-T-cell antibodies. Cyclosporine. Cyclosporine 2 to 4 mg /kg was given orally in two divided doses before transplantation. In the perioperative course (first postoperative week) oral cyclosporine was gradually increased according to renal function parameters (serum creatinine and blood urea nitrogen) and urinary output. to obtain a trough cyclosporine target level between 200 and 400 ng/ml (whole blood high-performance liquid chromatography method) after 7 days. Cyclosporine was continued to maintain this level for I month, the dosage subsequently being reduced to achieve a target level of ISO to 250 ng/rnl from the second to the sixth month. After 6 months the dosage was further decreased to maintain a target level of 100 to ISO ng/rnl. The drug was administered three times daily in equally divided doses. Cyclosporine dosing was guided exclusively by the results from whole blood high-performance liquid chromatography drug monitoring. Compared with other reports of immunosuppressive protocols in the literature, 6 the cyclosporine regimen used is likewise a low-dose cyclosporine regimen. Azathioprine. Preoperatively, azathioprine 2 mg/kg was administered intravenously as a single dose. Postoperatively, this dosage was continued orally and further modified to keep the white blood cell count around 4000/mm 3 , which indicated an adequate amount of immunosuppression. Steroids. Methylprednisolone 500 mg was given intraoperatively after release of the aortic crossclamp, followed by 3 doses of 125 mg/S hr. Afterward, steroids were continued as oral prednisolone 1.5 rug/kg that was rapidly tapered down to 0.2 mg/kg after I week. This was given as maintenance therapy in an alternate-day fashion, 80% of the calculated dose being administered on one day and 20% on the other day. Anti-T-cell antibodies. Because of the delayed, low-dose cyclosporine regimen, prophylactic immunotherapy was supplemented with anti-T-cell antibodies in the immediate postoperative period. Patients received either polyclonal ATG ( 10 mg/ kg) (N = 26) from the rabbit (ATG, Fresenius) or the horse (ATGAM, Upjohn Co., Kalamazoo, Mich.) or 5 mg monoclonal murine antibody (OKT3, Ortho-mune monoclonal antibodies, Ortho Diagnostic Systems, Inc., Raritan. N.J.; N = 17) for 7 to 10 days after transplantation. From June 1987 patients were prospectively randomized to receive either the polyclonal or monoclonal immunoprophylaxis. Antirejection treatment. In cases of clinically significant acute rejection (definition in next paragraph), patients were treated with a daily bolus of 1gm methylprednisolone for 3 days. If steroids failed to reverse rejection, ATG or OKT3 was given as a rescue treatment. Definition of rejection. Acute graft rejection was diagnosed on the basis of results from endomyocardial biopsies that were performed on a routine schedule or when clinically indicated. Biopsy results were graded in categories of absent, mild, mod0S

Volume 98 Number 6 December 1989

Acute rejection after cardiac transplantation

I I I

5

Table II. Characteristics ofpatient population-Categorical variables No rejection (N = 2!) Variable

Recipient sex' Male Female Prophylactic antibody Polyclonal Monoclonal Underlying disease CAD Congestive CMP Valvular disease Preoperative condition Stable at home Stable in hospital Intravenous inotropic support Previous cardiac surgery Yes :'010 Mismatch HLA-A 0 I

2 Mismatch HLA-B 0 I

2 Mismatch HLA-DR 0 I

2 Mismatch HLA-B+DR 2 3 4 Mismatch HLA-A+B+DR 2 3 4

5 6 HLA-DRw6 Positive Negative ABO Identical Not identical Rhesus Compatible Not compatible

Rejection (N= 22)

Tow! (N= 43)

No.

%

No.

%

No.

%

21 0

100 0

20 2

91 9

41 2

95 5

17 4

81 19

9 13

38 62

26 17

61 39

7 13 2

32 59 9

7 II

33 52 15

14 24 5

33 56 12

6 5 10

29 24 47

9 5 8

41 36

15 10 18

35 23 42

4 17

19 81

6 16

27 73

10 33

77

3 8 10

14 38 48

I

8 13

5 36 59

4 16 23

9 37 54

0 9 12

0 43 57

0 4 18

0 18 82

0 13 30

0 30 70

I II

0 7 15

0 32 68

I

9

5 52 43

18 24

2 42 56

5 12 4

24 57 19

0 II II

0 50 50

5 23 15

12 53 35

I 4 4

I

12 0

5 19 19 57 0

2 13 6

27

5 6 25 6

2 12 14 58 14

3 18

14 86

2 91

9 20

5 38

12 88

15 6

71 29

16 6

73

31 12

72

27

16 5

76 24

21

95 5

37 6

86 14

3

0 I

I

23

0 5 9 59

23

28

*Sex of the recipient was not evaluated as a variable in the logistic regression. CAD. Coronary artery disease; CMP. cardiomyopathy.

erate, severe, or resolved acute rejection according to the classic criteria of Billingham." Moderate or severe acute rejection on biopsy was considered as clinically significant rejection and antirejection treatment was instituted. In the present analysis patients were divided into those who had one or more clinically significant acute rejections (N = 22) within 6 months of follow-

up and those who did not. (N = 21). This served as the binary, dependent variable for logistic regression. Investigated risk factors. Three continuous and 12 categorical variables were used as independent variables and were investigated as potential risk factors for acute allograft rejection. These variables contained the following:

1 1 16

The Journal of Thoracic and Cardiovascular Surgery

Laufer et al.

Table III. Risk factors for acute rejection within 6 months: Univariate analysis Variable

p Value

Recipient age Donorage Ischemic time Preoperative condition Underlying disease Cardiac operation Prophylactic antibody Mismatch HLA-A Mismatch HLA-B Mismatch HLA-DR Mismatch HLA-B+DR Mismatch HLA-A+B+DR HLA-DRw6 ABO identity Rhesus mismatch

0.4899 0.1407 0.7069 0.3879 0.8518 0.5347 0.0024 0.3018 0.0816 0.0724 0.0043 0.0037 0.6056 0.9266 0.0712

Recipient age Donor age Ischemic time Perioperative anti- T-cell antibody prophylaxis: Comparison of polyclonal versus monoclonal prophylaxis Recipient's preoperative status: Extent of hemodynamic derangement and function at the time of transplantation Underlying disease: Categories of heart disease---eoronary artery disease, congestive cardiomyopathy, valvular heart disease Previous cardiac operation Donor-recipient mismatchesfor HLA-A. HLA-B, HLA-DR. HLA-B+DR. and HLA-A+B+DR. Tissue typing was done according to standardized methods. The National Institutes of Health technique was performed for definition of HLA-A and HLA-B antigens. HLA-DR antigens were determined by the two-color fluorescence technique. H LA-DRw6 positivity of recipient Donor-recipient mismatch for Rho(D) antigen Donor-recipient identity for ABO-system Because only two patients were female, the recipient sex was not examined as a variable in the logistic regression. Data on history, number, and type of pretransplant blood transfusions were unreliable and incomplete so that this potentially important variable was not considered for inclusion in the analysis. Statistical methods. Patient data were prospectively sampled and stored in a computerized data base. They were summarized by means of descriptive statistics, characterizing continuous variables by the mean ± standard deviation and categorical variables by proportions. Risk factors of acute allograft rejection were evaluated by stepwise logistic regression using BMDp lO standard statistical software. This analysis selects predictor variables in a stepwise manner and estimates the coefficients for a logistic regression. As a first step, a univariate analysis of the association between single putative risk factors and acute allograft rejection within 6 months was performed. The p values obtained thereby reflect the significance of a variable in univariate logistic regression. All continuous and categorical variables that were significant in univariate analysis were subsequently entered into the multivariate stepwise logistic regression (see appendix). Freedom from acute

allograft rejection was calculated by the univariate produetlimited Kaplan- Meier method stratified according to significant variables identified in the logistic regression analysis. Results of Kaplan-Meier analysis were compared with the log-rank and Wilcoxon tests. Statistical significance was assumed at a p value <0.05.

Results Rejection and distribution of potential risk factors. Twenty-two heart transplant recipients exhibited at least one clinically significant acute rejection episode and 51% of the study population was thus affected. During the 6month period of follow-up, fatal rejection occurred only once in a female patient 5 weeks after the transplantation. Twenty-one patients (49% of the study population) were free from clinically significant acute rejection over 6 months of postoperative observation. Continuous and categorical variables investigated as potential risk factors for acute rejection are characterized by descriptive statistics and depicted in Tables I and II.

Predictors of rejection Univariate analysis. The univariate analysis suggested a strong influence of the prophylactic antibody (p = 0.(024) and number of mismatches for the HLA-A+B+DR loci (p = 0.(037) and for the HLA-B+DR loci (p = 0.(043). A borderline significant influence was obtained from mismatch for the Rho(D) antigen (p = 0.0712) and from the number of mismatches for the HLA-DR loci (p = 0.0724). The choice of prophylactic antibody (polyclonal versus monoclonal) indicated the strongest influence and was therefore the first variable that entered the stepwise logistic regression. The p values of univariate analysis for all variables are summarized in Table III. Multivariate analysis. After correction by stepwise logistic regression, the number of mismatches for the HLA-B+DR loci and the choice of prophylactic antibody (polyclonal versus monoclonal) revealed the strongest influence with p values of 0.0029 and 0.0031, respectively. The donor age demonstrated borderline significance for occurrence of acute rejection (p = 0.0657). Patients at high risk for acute graft rejection were cardiac recipients with four mismatches for the HLA-B+DR loci and monoclonal antibody prophylaxis. The risk of graft rejection decreased with a lower number of mismatches for the HLA-B+DR loci and polyclonal antibody prophylaxis. None of the five patients with two mismatches for the HLA-B+DR loci had an acute rejection episode in the period observed. Nine of 26 recipients with polyclonal ATG prophylaxis had an acute rejection episode, whereas 13 of 17 recipients with monoclonal OKT3 prophylaxis had an episode of rejection. Patients without graft rejection showed a weak tendency toward higher age of the

Volume 98 Number 6

Acute rejection after cardiac transplantation

December 1989

-.. c

-•.........................................................................................._

1.00

,

.S U


......E 0

.

...

0.60


....

c:

,

t:

.

0

0

B+DR=2 (N=5)

_.-

B+DR=3 (N=23) B+DR=4 (N=15)

_.

13

..

_._._._._ ....

8

..-. 5

0040

.S

c.

5

.........•.......

..... -------

0.80

1 117

•. _._._._.- ._._.- ._._.- ._._.- ._. _.- ._._.. 3

0.20

c.

0.00 0.0

2~

4~

6.0

months posttransplant

Fig. 1. Univariate Kaplan-Meier calculation for freedom from acute rejection in cardiac recipients stratified according to donor/recipient mismatches for combined HLA-B+DR loci. Numbers in the graph indicate patients at risk 2, 4, and 6 months after transplantation, respectively.

donor. None of the other variables yielded any additional information. The proposed importance of mismatch for Rha(D) antigen was not maintained after multivariate analysis (p = 0.2587). Inasmuch as the mismatch for HLA-DR is part of that for HLA-B+DR, this variable lost its importance after logistic regression (p = 0.4069). Significant variables, corresponding regression coefficients, and p values are displayed in Table IV. Predictive logistic model. Entering significant variables, corresponding regression coefficients, and the constant into the logistic model yielded the predicted probability of acute rejection for each recipient. With a probability of 0.5 used as a cut point, the model identified 86% of recipients with rejection and 86% of those without rejection correctly. Freedom from acute rejection. These results were obtained by stratification according to the number of mismatches for the HLA-B+DR loci,2-4 antibody prophylaxis (polyclonal, monoclonal), and median donor age (28 years, > 28 years). Univariate Kaplan-Meier calculations revealed freedom from clinically significant graft rejection after 6 months in 100%of recipients with two mismatches for the HLA-B+DR loci compared with 41% with three and 27%with four mismatches for the HLA-B+DR loci (Fig. 1). These differences revealed a p value of 0.0123 (log rank) and 0.0166 (Wilcoxon) comparing two versus four mismatches for HLA-B+ DR and a p value ofO.0271 (log rank) and 0.0358 (Wilcoxon) comparing two versus three mismatches for HLA-B+DR. Although recipients with three mismatches for HLA-B+DR had better freedom from rejection than recipients with four mismatches for

Table IV. Risk factors for acute rejection within 6 months: Multivariate, stepwise logistic regression Variable

Coefficient

Standard error

Mismatch HLA-B+DR Prophylactic antibody Donor age Constant

2.98 1.77 -0.11 -5.96

0.97 0.58 0.62 2.75

p Value

0.0029 0.0031 0.0657

HLA-B+DR, the differences between both strata in the univariate Kaplan-Meier method lack statistical significance (log rank p = 0.3886, Wilcoxon p = 0.4672). Recipients with polyclonal antibody prophylaxis had 59% freedom from acute rejection after 6 months compared with 22% of recipients with the monoclonal antibody (Fig. 2), a highly significant difference (log rank p = 0.0012, Wilcoxon p = 0.0009). This confirmed the results from the logistic regression analysis. However, this univariate method (entering the donor age as categorical variable) failed to reproduce the borderline significant correlation between donor age (as a continuous variable) and acute rejection in the multivariate model. Six months after transplantation, a donor age of 28 years or less was associated with 45% freedom from acute rejection and a donor age over 28 years with 42% (Fig. 3, log rank p = 0.7861, Wilcoxon p = 0.5225). Discussion Acute allograft rejection after cardiac transplantation has remained an important complication that impacts

The Journal of

1 1 18

Thoracic and Cardiovascular

Laufer et al.

Surgery

c

1.00

.~

ti

.~

...

~

0.80

0.60

c

0.40

.~

t:

e c.

...c.o

18

. ...

•... _._ .... I

ATG (N=26)

...........

OKT3 (N=I7)

16 12

.... I

_____

..

-.

•. ~ 5

_. _. _. _._.- ._._. _.- ._. _.-._. - ._. _. _. _. _. -._. _.

• 4

0.20

0.00 +----.-"""'T""-.....--,-..........-....-----.-.....-......----.---.--r 2~ 4~ 6.0 0.0

months posttransplant Fig. 2. Univariate Kaplan-Meier calculation for freedom from acute rejection in cardiac recipients stratified according to the type of prophylactic antibody used perioperatively. Six-month freedom from acute rejection is 59'k versus 22% for ATG versus OKT3 prophylaxis. Numbers in the graph indicate patients at risk 2, 4, and 6 months after transplantation, respectively.

significantly on morbidity and mortality." The acute rejection response is a complex process, which is induced by major (MHC) and minor histocompatibility alloantigens expressed on the grafted tissue. Therefore the extent of disparity of class I and/or II MHC antigens is an important assumption for immunologic intolerance, whereas the final occurrence of graft rejection is a more complex interaction of two additional factors. First, according to experimental transplant studies in the genetically well studied mice model, immune-response-gene-regulated immunoresponsiveness of the transplant recipient is considered an important determinant for rejection. 11-13 This suggests that the strength of the individual immune response to foreign alloantigens is genetically controlled and may be important in the setting of clinical organ transplantation. Additionally, immunoresponsiveness is modified by factors other than those that are genetically determined. Modulation of immunoresponsiveness by suppression (i.e., different type of immunosuppressive drug) or sensitization (i.e., previous transplant, blood transfusion, pregnancy) of the immune system has a powerful impact on the successful prevention of rejection and induction of graft acceptance. Second, the strength of rejection is likewise determined by the expression of alloantigens on the graft during both the induction and effector phase of the immune response. 14, 15 Thus the interaction of these mechanisms in the setting of clinical cardiac transplantation decides whether the graft is rejected or not. Each of the risk factors considered potentially important in the present analysis is or may be related

to one or more of the aforementioned basic determinants of the immune response. Results of this analysis indicate that the number of alloantigen disparity for the HLA-B+ DR loci (p = 0,0029) and the type of prophylactic antibody used perioperatively (p = 0.0031) have a highly significant influence on whether or not immunologic intolerance willoccur within 6 months, Results in renal transplantation from a large multicenter trial analyzing more than 8000 first cadaver kidney transplantsl'' clearly identified the disparity for histocompatibility matching of combined HLA-B and HLA-DR loci as an important impact on graft survival. HLA-B+DR matching always produced an effect regardless of whether cyclosporine was used or not, and zero mismatches had an approximately 20% better graft survival than four mismatches. Isolated mismatch analysis for class II MHC antigen was somewhat less important, whereas mismatch analysis for class I MHC antigens exhibited a weak correlation, Although there is a stronger expression of target alloantigens in the kidney than in the cardiac graft, 17, 18 the kidney data suggest that cardiac grafts poorly matched for HLA-B+DR loci might be rejected more frequently. Yacoub and associates 19 reported a significant effect on 2-year graft survival after cardiac transplantation by matching for HLA-DR and HLA-B+DR. These findings are confirmed by results from the present multivariate analysis and suggest that the extent of disparity for the HLA-B+DR alloantigens has a significant impact on acute rejection in the setting of clinical

Volume 98 Number 6

Acute rejection after cardiac transplantation

December 1989

e

1.00 donor age

0

ti QJ

..... ...

QJ

I I I9

--.._ ..... _.

0.80

<= 28a (N=22) > 28a (N=21)

E

...... ......c 0

,

0.60

.=-... 0

Co

...0

12

• .•. _.t:.7_.:::_:-::._-:-:._~._::-.:::;; ..: - - - - - - - - - - _

QJ QJ

II

e·-·_·_·_·_·_···_·_·-·-·_·_·_···_·- _. 9

0.40

9

7

0.20

Co

0.00 0.0

2.0

4.0

6.0

months posttransplant Fig. 3. Univariate Kaplan-Meier calculation for freedom from acute rejection in cardiac recipients stratified according to median donor age (28 years). Six-month freedom from acute rejection is 45% for a donor age of 28 years or less and 42% for a donor age over 28 years. Numbers in the graph indicate patients at risk 2, 4, and 6 months after transplantation, respectively.

cardiac transplantation. If not otherwise indicated by priority of a high urgency status, the practical consequences would be the institution of prospective matching for HLA-B+DR. Thus patients with well-matched grafts may require less intensive immunosuppression to prevent acute rejection, a situation that would decrease the incidence of opportunistic infection and reduce adverse drug (cyclosporine) effects, such as chronic nephrotoxicity'? and arterial hypertension in the long-term follow-up. Although kidney transplant recipients who have the HLA antigen DRw6 have been demonstrated to reject their graft faster than HLA-DRw6 negative recipients," we were not able to detect a similar correlation in cardiac recipients receiving three immunosuppressive drugs. Recent studies indicate that differences in kidney graft survival in HLA-DRw6 positive and HLA-DRw6 negative recipients are related to pretransplant blood transfusion policy.l? and this may explain lack of significance in cardiac recipients. In contrast to a report from Nakatani and colleagues.P this study also failed to show a higher risk for acute rejection in recipients with ABO-nonidentical hearts than with ABO-identical hearts. Similarily, matching for the Rho(D) antigen disclosed no significant risk factor of acute rejection in the multivariate model. Sensitization of the recipient, which implies lyrnphocytotoxic antibody production, may also increase the incidence and severity of cardiac rejection. In this regard, a positive donor jrecipient lymphocytotoxic cross-match has been shown to impact adversely, but not significantly, on cardiac graft survival. I 9 In our patient series, however,

lymphocytotoxic cross-match studies were not available for the whole population. Inasmuch as the data about the number and type (fresh or stored whole blood, packed red cells) of pretransplant blood transfusions were incomplete and unreliable, previous cardiac surgery served as an approximative indicator as to whether the individual had been previously exposed to foreign alloantigens or not. According to the well-studied kidney transplant model, blood transfusion does not necessarily mean sensitization but may also induce a status of donor-specific, reduced immunoresponsiveness. The theoretical importance of this variable on acute cardiac rejection was not confirmed in the present investigation. Other important indicators for sensitization may be pregnancy and female gender. Although included in the descriptive statistics, neither variable was examined in the logistic regression, because there were only two women. Interestingly, both of them had had multiple pregnancies and rejected the transplant. All patients received the same standard low-dose triple drug maintenance immunosuppression but differed in the immediate postoperative period. Patients treated prophylactically with monoclonal OKT3 antibodies were at significantly higher risk for development of an acute rejection episode and compared unfavorably with patients receiving polyclonal ATG antibodies. Although OKT3 is undoubtedly effective for rescue treatment of rejection in cadaver kidney-" and heart transplantation." the value of prophylactic perioperative administration is a matter of controversy. Inferior prevention of rejection by OKT3 prophylaxis compared with ATG prophylaxis has been

The Journal of

1 120

Laufer et al.

reported by Kormos and co-workers.i" Another study found superior prevention of rejection by OKT3 prophylaxis.I' but it may be of particular importance that OKT3 was administered for as long as 14 days and a steroid taper used after OKT3 discontinuation. The enthusiastic report of this group was not confirmed in our series. The immunosuppressive protocol in our study was exactly the same for the ATG and the OKT3 groups, particularly with respect to the duration of antibody prophylaxis. Therefore, in our experience, polyimmunotherapy involving OKT3 as a prophylactic agent provides less effective prevention of acute rejection than a comparable course of ATG. Comparison, however, is not obtained for OKT3 versus ATG but rather for one protocol versus another. OKT3 (anti-T3) antibodies have been demonstrated to elicit, in vitro, a variety of immune responses (T-cell activation) that are potential promoters of an allograft rejection.P: 29 These immune response-stimulating properties are considered to be caused by the intimate association of the T3 antigen with the specific antigen receptor of T cells. Successful reversals of acute kidney allograft rejection with OKT3 have been associated with a high incidence of repeated rejection episodes.i" This observation and the unfavorable results of prophylactic OKT3 in our study may conceivably represent the result of the ability of OKT3 to activate T cells. Donor age was disclosed as a variable with borderline significant impact on acute rejection. This suggests that hearts from younger donors are more easily rejected than those from older donors. However, because of the limited number of patients and borderline significance, a statement such as this does not appear to be justified. Nevertheless, it may be hypothesized that fewer alloantigens are expressed on the grafted target tissue of older donors, and this may explain the weak tendency toward less acute rejection in the multivariate model. Recipient age, ischemic time, type of underlying disease, and immediate pretransplant condition provided no further information for prediction of acute rejection. In summary, a poor donor/recipient match for combined HLA-B+DR loci and monoclonal OKT3 antibody prophylaxis have been identified as significant, independent risk factors predicting acute cardiac allograft rejection in cardiac recipients receiving low-dose triple drug maintenance immunosuppression. Similar to kidney transplantation, the combined HLA-B+DR disparity had a higher impact than another combination (HLA-A+B+DR) or matching for single HLA loci alone. In an attempt to improve the results of heart transplantation by reduction of rejection-associated complications and achievement of reduced adverse drug effects by a less intensive immunosuppression, prospective match-

Thoracic and Cardiovascular Surgery

ing for the combined HLA-B+DR loci should be stimulated. Although caution is required for comparison of one antibody with another, it is assumed that in our experience OKT3 monoclonal antibody was a significantly less effective prophylactic immunosuppressant than polyclonal ATG after heart transplantation. We gratefully acknowledge the financial support obtained from Sandoz-Austria, Vienna, Austria. REFERENCES I. Kaye MP. The registry of the International Society for Heart Transplantation: fourth official report-I 987. J Heart Transplant 1987;6:63-7. 2. Fragomeni LS, Kaye MP. The registry of the International Society for Heart Transplantation: fifth official report1988. J Heart Transplant 1988;7:249-53. 3. Laufer G, Laczkovics A, Wollenek G, Schreiner W, Kober I, Wolner E. Incidence and severity of acute cardiac allograft rejection with two different low-dose cyclosporine maintenance protocols. Ann Thorac Surg 1988;46:382-8. 4. Kirklin JK, Naftel DC, McGiffin DC, McVay RF, Blackstone EH, Karp RB. Analysis of morbid events and risk factors for death after cardiac transplantation. J Am Coli Cardiol 1988;I 1:917-24. 5. Uretsky BF, Murali S, Reddy S, et al. Development of coronary artery disease in cardiac transplant patients receiving immunosuppressive therapy with cyclosporine and prednisone. Circulation 1987;76:827-34. 6. Spratt P, Esmore D, Baron D, Shanahan MX, Farnsworth AE, Chang VP. Effectiveness of minimal dosage cyclosporine in limiting toxicity and rejection. J Heart Transplant 1986;5:8-12. 7. Deeb GM, Kolff J, McClurken JB, et al. Antithymocyte gamma globulin, low dosage cyclosporine, and tapering steroids as an immunosuppressive regimen to avoid early kidney failure in heart transplantation. J Heart Transplant 1987;6:79-83. 8. Kahan BD. Immunosuppressive therapy with cyclosporine for cardiac transplantation. Circulation 1987;75:40-56. 9. Billingham ME. Diagnosis of cardiac rejection by endomyocardial biopsy. J Heart Transplant 1982;1:25-30. 10. Engelman L. Stepwise logistic regression. In: Dixon WJ, ed. BMDP statistical software. Los Angeles: University of California Press 1985:330-44. II. Butcher GW, Howard Jc. Genetic control of transplant rejection. Transplantation 1982;34:161-6. 12. Isakov N, Bach FH. Genetic control of variability in response to class-I-antigen-disparate thyroid grafts. Transplantation 1985;39:303-9. 13. Schwartz RH. Immune response (Ir) genes of the murine histocompatibility complex. Adv Immunol 1986;38:31202. 14. Milton AD, Fabre JW. Massive induction of donor type class I and class II major histocompatibility complex anti-

Volume 98 Number 6 December 1989

gens in rejecting cardiac grafts in the rat. J Exp Med 1985;161:98-112. 15. Steiniger B, Klempnauer J, Wonigeit K. Expression of class I and class II major histocompatibility complex antigens during heart allograft rejection in the rat. Transplant Proc 1985;17:1907-10. 16. Opelz G. Correlation of HLA matching with kidney graft survival in patients with or without cyclosporine treatment. Transplantation 1985;40:240-3. 17. Daar AS, Fuggle SV, Fabre JW, Ting A, Morris PJ. The detailed distribution of MHC class II antigens in normal human organs. Transplantation 1984;38:293-8. 18. Daar AS, FuggleSV, Fabre JW, Ting A, Morris PJ. The detailed distribution of HLA-A, -B, -C antigens in normal human organs. Transplantation 1984;38:287-92. 19. Yacoub M, Festenstein P, Doyle P, et ai. The influence of HLA matching in cardiac allograft recipients receiving cyclosporine and azathioprine. Transplant Proc 1987;19: 2487-9. 20. Myers BD, Ross J, Newton L, Luetscher J, Perlroth M. Cyclosporine-associated chronic nephrotoxicity. N Engl J Med 1984;311:699-705. 21. Hendricks GFJ, Claas FHJ, Persijn GG, Witvliet MD, Baldwin W, van Rood 11. HLA-DRw6 positive recipients are high responders in renal transplantation. Transplant Proc 1983;15:1136-41. 22. Lagaaij EL, Persijn GG, Termijtelen A, Hendricks G FJ, de Lange P, van Rood 11. Evidence that the difference in kidney graft survival in DRw6+ and DRw6- recipients may be explained by a blood transfusion policy that is disadvantageousfor DRw6recipients. Transplantation 1987;44:78891. 23. Nakatani T, Aida H, Macris MP, Frazier OH. Effect of ABO blood type on survival ofCSA-treated cardiac transplant patients [Abstract]. J Heart Transplant 1988;7:81. 24. Ortho Multicenter Transplant Study Group. A randomized clinical trial ofOKT3 monoclonal antibody for acute rejection of cadaveric renal transplants. N Engl J Med 1985; 313:337-42. 25. Gilbert EM, DeWitt CW, Eiswirth CC, et ai. Treatment of refractory cardiac allograft rejection with OKT3 monoclonal antibody. Am J Med 1987;82:202-6.

Acute rejection after cardiac transplantation

1 12 1

26. Kormos R, HerIan DB, Curran M, Hardesty RL, Armitage J, Griffith BP. Monoclonal vs. polyclonal antibody therapy for prophylaxis against rejection following cardiac transplantation [Abstract]. J Heart Transplant 1988;7:82. 27. Bristow MR, Gilbert EM, Renlund DG, DeWitt CW, Burton NA, O'Connell JB. Use of OKT3 monoclonal antibody in heart transplantation: review of the initial experience. J Heart Transplant 1988;7:1-11. 28. Van Wauwe JP, De May JR, Grossens JG. OKT3: a monoclonal human T-lymphocyte antibody with potent mitogenic properties. J Immunol 1980;124:2708-13. 29. Suthanthiran M, Wiebe ME, Stenzel KH. Effect of immunosuppressants on OKT3 associated T cell activation: clinical implications. Kidney Int 1987;32:362-7.

Appendix Stepwise logistic regression selects variables in a stepwise manner according to the level of significance determined by univariate analysis. The coefficients for a logistic regression model are estimated to predict the probability of an event (acute cardiac allograft rejection within 6 months after operation). The dependent variable y represents the probability of acute cardiac allograft rejection, XI to Xk represent the risk factors analyzed, and bo, b, ... bk the unknown parameters to be estimated. With this notation, the basic formula of logistic regression can be stated as follows: E(y) = exp(bo + b l XI + +~ Xk) I + exp(bo + b, XI + + ~ Xk) where E denotes expectation. Whereas the first parameter bo represents the constant, the other parameters correspond to one of the risk factors. Practical computation was conducted with the BMDPLR program. During the stepwise procedure, variables are entered or removed. As a limit of significance for the removal or entry of a variable, a p value of 0.15 or 0.10, respectively, is chosen. The selection of variables to be entered or removed from the model is based on a maximum likelihood ratio test. This test evaluates the additional information provided by any particular variable for the prediction of acute allograft rejection. The goodness of fit of the final model is evaluated by means of a classification analysis, assessing the proportion of correct predictions for acute cardiac rejection.