Risk Factors for Death After Heart Transplantation:Does a Single-Center Experience Correlate WithMulticenter Registries?James F. McCarthy, FRCSI, Patrick M. McCarthy, MD, Malek G. Massad, MD,Daniel J. Cook, PhD, Nicholas G. Smedira, MD, Vigneshwar Kasirajan, MD,Marlene Goormastic, MPH, Kathy Hoercher, MS, and James B. Young, MDCardiac Transplantation and Mechanical Circulatory Assist Program, Department of Thoracic and Cardiovascular Surgery, TheTransplant Center Histocompatibility Laboratory, and Heart Failure and Cardiac Transplant Medicine, The Cleveland ClinicFoundation, Cleveland, Ohio

Background. Risk factors for death after heart trans-plantation (Tx) are frequently documented from multi-center registries. Although this information is helpful, itreflects a whole range of experiences and results, andmay not translate to a particular center. This study wasperformed to (1) evaluate pre-Tx factors affecting mortal-ity in a single-center experience, and (2) compare thesefactors with risk factors obtained from multicenter reg-istry reports.

Methods. Review of our transplant database betweenJanuary 1984 and December 1995 identified 405 adultswho received a primary heart Tx. Multiple factors wereanalyzed, including demographics, Tx era, cytomegalovi-rus status, United Network for Organ Sharing status ofrecipient, presence of pulmonary hypertension, previouscardiac operations, mechanical ventilation or circulatorysupport, ischemia time, number of rejection episodes,and preoperative flow cytometry crossmatching.

Results. One- and 5-year survival rates were 87.8% and73.4%, respectively (Kaplan-Meier). Contrary to multi-center registry reports, our data indicate that reoperativeprocedures, left ventricular assist device support, increas-

ing donor and recipient age, and ischemia time up to 4.2hours are not risk factors for death after Tx. Likewise,mode of donor death is not a risk factor affecting out-come. Significant risk factors for mortality identified bymultivariate analysis included early transplant era (1984to 1989; p 5 0.002), female donor (p 5 0.042), cytomega-lovirus-seropositive donor (p 5 0.048), high pulmonaryvascular resistance (p 5 0.018), and intraaortic balloonpump support (p 5 0.03). It also identified a positiveB-cell flow cytometry crossmatch (p 5 0.015) to be a riskfactor with univariate analysis.

Conclusions. Our data identify a group of recipients,reportedly at high risk in multicenter registries, who arenot at increased risk of death after Tx. This informationsupports the growing experience with older donors andrecipients and with bridged transplants, and has allowedus to expand our donor pool. These prognostic factors atevaluation allow more liberal selection of patients anddonors for transplantation.

(Ann Thorac Surg 1998;65:1574–9)© 1998 by The Society of Thoracic Surgeons

Heart transplantation currently provides the mosteffective treatment for end-stage heart disease.

However, in view of donor shortage, the number of hearttransplants performed annually in the United States andworldwide has plateaued [1] and will continue to do soshort of attempts to expand the donor pool. Analysis ofrisk factors is, therefore, important to make informedjudgments in any particular potential donor–recipientsituation. To date, data analysis from numerous multi-center registries has formed the basis of our risk stratifi-cation knowledge. Studies from the Cardiac TransplantResearch Database and the Registry of the InternationalSociety for Heart and Lung Transplantation continue to

list criteria such as increasing donor and recipient age,use of mechanical ventricular assistance, and cold isch-emia time as risk factors that affect survival aftertransplantation [1, 2]. Although data from these mul-ticenter registries are extremely helpful, the informa-tion obtained is extrapolated from a number of differ-ent centers with different management strategies andranges of experience, and as a result, may not alwaysrepresent or translate to a particular center’s experi-ence. It is therefore probable that favorable results of alarge center’s experience may be masked in a multi-center study by the inclusion of smaller volume pro-grams with higher mortality [3]. In this study, wesought to analyze the risk factors for mortality afterheart transplantation at our center, and to comparethese risk factors with data obtained from multicenterregistry reports.

Accepted for publication Dec 2, 1997.

Address reprint requests to Dr McCarthy, Department of Thoracic andCardiovascular Surgery, Desk F25, The Cleveland Clinic Foundation, 9500Euclid Ave, Cleveland, OH 44195 (e-mail: mccartp@cesmtp.ccf.org).

© 1998 by The Society of Thoracic Surgeons 0003-4975/98/$19.00Published by Elsevier Science Inc PII S0003-4975(98)00138-6

Material and Methods

Patient CharacteristicsReview of our transplant database between January 1984and December 1995 identified 405 adult patients (18 yearsof age or older) who underwent orthotopic heart trans-plantation at the Cleveland Clinic Foundation. The clin-ical records of these patients were reviewed retrospec-tively. The patients were analyzed in two transplant eras:those who received transplants between 1984 and 1989,and those who received transplants between 1990 and1995. Description of the donor and recipient characteris-tics are listed in Table 1. There were 315 (78%) male and90 (22%) female recipients with a mean (6 standard errorof the mean) age of 48.7 6 0.5 years (range, 18 to 66 years).In comparison, 69% of donors were men. The mean 6SEM donor age was 30.0 6 0.6 years (range, 10 to 67years). All patients had end-stage heart disease that wascaused by an ischemic cardiomyopathy in 192 patients(47%) and a nonischemic dilated cardiomyopathy in theremainder (53%). The factors analyzed are listed in Table2, and include donor and recipient age, gender, race, andcytomegalovirus status; donor mode of death; cause ofheart failure; United Network for Organ Sharing status;presence of pulmonary hypertension and previous car-diac operations in the recipient; mechanical ventilatorysupport or circulatory support to transplantation with theintraaortic balloon pump (IABP) or left ventricular assistdevice (LVAD; HeartMate, Thermo Cardiosystems Inc,Woburn, MA); organ ischemia time up to 4.2 hours (mean2.3 hours); flow cytometry crossmatch; number of epi-sodes of cellular rejection; and survival. Myocardial pres-ervation was attained with modified Buckberg solution.

The decision to implant an LVAD was prompted by aprogressive deterioration in the patient’s hemodynamicsand organ function. The hemodynamic indications forLVAD use included transplantation candidates who hada pulmonary capillary wedge pressure of 20 mm Hg orgreater coupled with a cardiac index of 2.0 L z min21 z m22

or less, or a systolic blood pressure of 80 mm Hg or less,despite maximal inotropic agents and IABP support [4].Details of the device description and implantation tech-nique have been reported [5].

Cellular rejection was graded based on the gradingsystem adopted from the Billingham criteria and thecriteria of the International Society for Heart and LungTransplantation [6, 7], whereby grades IA, IB, and II wereconsidered mild rejection, grades IIIA and IIIB wereconsidered moderate, and grade IV severe rejection.Early rejection referred to all rejection episodes occurringwithin 30 days of the transplant. Only patients who hadmoderate and severe rejection were considered in theforthcoming analysis of cellular rejection after transplan-tation. The technique of flow cytometry crossmatchinghas been described in previous publications [8–10].Maintenance immunosuppression consisted of a triple-drug combination of cyclosporine, azathioprine, and ste-roids since the initiation of the program. Patients withcompromised renal function were selectively inducedwith OKT3 monoclonal antibody after transplantation.This was followed by conversion to cyclosporine-basedimmunosuppression when their renal function im-proved. Episodes of acute rejection were initially treatedwith intravenous methylprednisolone for 3 days. Recur-rent or refractory rejection was treated with steroids andOKT3. Other modalities such as plasmapheresis, immu-noglobulin, and cytolytic therapy were used in cases ofacute rejection when a significant humoral componentwas suspected.

Statistical AnalysisWhen indicated, data are presented as mean 6 standarderror of the mean. In all, 22 covariates were analyzed aspotential risk factors for death after transplantation (Ta-ble 2). This analysis included all deaths, both early andlate. Actuarial survival curves were generated by the

Table 2. Covariates Analyzed for Risk of Mortality AfterHeart Transplantation

DonorVariables

RecipientVariables Other

Age Age Transplant eraSex Sex Cold ischemia timeRace Race Flow cytometry crossmatchCMV status CMV statusMode of

deathDiagnosis

Hypertension UNOS statusCardiac arrest Pulmonary

vascularresistance

Previous cardiacoperation

Mechanicalventilation

HeartMateLVAD bridged

Intraaorticballoon pump

Early rejectionepisodes

CMV 5 cytomegalovirus; LVAD 5 left ventricular assist device;UNOS 5 United Network for Organ Sharing.

Table 1. Description of Recipient and Donor Characteristics

VariableNo. of

Patients % of Total

Transplantation era 1984–1989 92 23Transplantation era 1990–1995 313 77Previous cardiac operations 176 44Male donor 278 69Male recipient 315 78White recipient 337 83CMV-positive recipient 280 72HeartMate LVAD bridged 47 12Donor hypertensive 27 7Donor cardiac arrest 47 12CMV-positive donor 195 49

CMV 5 cytomegalovirus; LVAD 5 left ventricular assist device.

1575Ann Thorac Surg McCARTHY ET AL1998;65:1574–9 RISK AFTER HEART TRANSPLANTATION

Kaplan-Meier method, and differences were analyzed bythe log-rank test. Significant risk factors for mortalitywere identified using the Cox proportional-hazardsmodel. Three Cox models were run for the patients withcomplete data for each set of variables. A p value lessthan 0.05 indicated significance.

Results

Patient demographics are presented in Table 1. Thirty-two patients (7.9%) were on mechanical ventilatory sup-port at the time of transplantation, 29 (7.2%) were onIABP support, and 21 (5.2%) were on both, whereas 47(12%) were on an LVAD. The patients were followed upfor a mean 6 SEM duration of 41.2 6 1.6 months (longest,143 months). The mean 6 SEM number of episodes ofcellular rejection within the initial 30 days after trans-plantation was 0.78 6 0.04 episodes. There were 99 deathsamong the 405 patients (24%). The causes of early (,3months after transplantation) and late deaths (.3months) are listed in Table 3. Figure 1 demonstrates theKaplan-Meier survival curves for those patients trans-planted in the early (1984 to 1989) and late (1990 to 1995)eras. The overall 1-, 3-, and 5-year actuarial survival rateswere 87.8%, 80.1%, and 73.4%, respectively.

The significant risk factors for death after transplanta-

tion identified by multivariate analysis are outlined inTable 4. These included five significant variables: earlytransplant era (1984 to 1989), female donor, cytomegalo-virus seropositive donor, IABP support, and increasedpulmonary vascular resistance (.3 Wood units) in therecipient. Univariate analysis showed positive B-cell flowcytometry crossmatching (data available in 101 patients)and pulmonary hypertension (systolic .50 mm Hg) inthe recipient to also significantly affect survival. Theactuarial Kaplan-Meier survival estimate for patientswith pulmonary vascular resistance less than 3 Woodunits was 74% at 5 years compared with only 66% forthose with pulmonary vascular resistance of 3 Woodunits or more (p 5 0.04, log-rank test). All other riskfactors identified from multicenter registry reports [1, 11]were not significantly associated with death, includingreoperation, LVAD support as a bridge to transplanta-tion, increasing donor and recipient age, ischemia timesof up to 4.2 hours, and mode of donor death. Tables 5 and6 depict survival correlated with increasing donor andrecipient age. Interestingly, mechanical ventilatory sup-port to transplantation did not impact survival by uni-variate analysis. The actuarial 5-year survival was 73% inthe group with ventilatory support compared with 71%without it (not significant, log-rank test).

Comment

Heart transplantation continues to be a viable therapeu-tic option for certain patients with end-stage heart dis-ease, with a reported 1-year survival of 83% worldwide[1]. Previous data clearly demonstrate that the number oftransplants performed at a single center affect survival

Table 5. Impact of Donor Age on Survival

Age (y)No. of

Patients1-Year

Survival5-Year

Survival

,20 83 92.8% 85.9%20–29 136 87.5% 70.3%30–39 94 86.0% 71.6%40–49 67 81.9% 70.4%.50 25 96.0% 60.3%

Total 405 87.8% 73.4%

Table 3. Early (,3 months) and Late (.3 months) KnownCauses of Death After Heart Transplantation Identified Froma Large Single-Center Experience (Cleveland ClinicFoundation 1984–1995)

Cause of Death ,3 Months After Tx .3 Months After Tx

Rejection 11 (14%) 25 (32%)a

Infection 8 (11%) 5 (7%)Graft failure 4 (5%) 1 (1%)Cerebrovascular 3 (4%) 1 (1%)Hemorrhage 1 (2%) 0Malignancy 0 7 (9%)Other 4 (5%) 7 (9%)

Total 31 (40.3%) 46 (59.7%)

a Includes chronic rejection and cardiac allograft vasculopathy.

Tx 5 transplantation.

Fig 1. Actuarial patient survival for cardiac transplants in two eras.

Table 4. Risk Factors for Mortality After HeartTransplantation Identified From a Large Single-CenterExperience (Cleveland Clinic Foundation 1984–1995)

Variable Risk Ratio p Value

Transplantation era1984–89

2.1 0.002

Female donor 1.5 0.042CMV-positive donor 1.6 0.048PVR .3 Wood units 1.8 0.018IABP 2.2 0.03

CMV 5 cytomegalovirus; IABP 5 intraaortic balloon pump; PVR 5pulmonary vascular resistance.

1576 McCARTHY ET AL Ann Thorac SurgRISK AFTER HEART TRANSPLANTATION 1998;65:1574–9

[11, 12]. The risks of death at early and intermediate timepoints are substantially higher in low-volume transplantcenters, which constitute a large proportion of the centersperforming transplantation in the United States [12]. Of7,893 cardiac transplants reviewed in one registry report,as much as 25% of the differences in observed patientmortality were related to center-to-center variability [12].

Multicenter registries have the advantage over single-center studies of evaluating a large number of patients ina relatively short period of time. Therefore there is anincreased ability to detect risk factors and differencesamong patient groups. However, because of the largevariability in each center’s institutional protocols andresults, risk stratification may not translate to a particularcenter’s experience. Previous reports from our institutionhave clearly demonstrated that presence of an implant-able LVAD in bridged patients was not a risk factor fordeath after transplantation [13] in contrast to multicenterregistry data [1, 14]. Not infrequently, other organ failuredevelops in patients with heart failure who are waitingfor a donor. These patients have demonstrated improve-ment in the state of their multiorgan dysfunction and intheir physical condition after bridging [13, 15]. It seemslogical then to assume that if early results after LVADimplantation are optimal these patients should be in areasonably good condition at the time of transplantation[13]. Interestingly although the presence of an LVAD hasnot been demonstrated to be a risk factor for death aftertransplantation at our center, presence of an IABP beforetransplantation does constitute a significant risk, proba-bly because of the hemodynamic instability of the patientrequiring acute support. Pulmonary vascular resistancewas also identified as a risk factor as demonstrated inTable 7 and as previously shown by other groups [16, 17].This remains so despite our practice of implanting over-sized high-quality hearts in these recipients. However wehave just begun to use nitric oxide in some of thesepatients and this may affect early survival.

Not surprisingly, early transplant era is a risk factor formortality. This has been the case in other reports [18, 19],and probably reflects a combination of a learning curvefor the transplantation team, and also a different erawhen the diagnostic and therapeutic options were not atthe present level. The majority of the transplantations(77%) were in the later era and the early learning expe-rience was consolidated here. The improved survival inrecent years has the added benefit of allowing expansion

of the donor pool [3]. This has been our practice since1990 and increasing donor and recipient age have not hadan adverse impact on survival, as has been reported bysome authors [20]. Indeed judicious use of hearts fromdonors aged more than 50 years, in the presence of anormal coronary angiogram, has allowed us to performtransplantation in older recipients, who are the mainbeneficiary of this group, with an increasing frequency.Other risk factors identified by multicenter registry re-ports were not significantly associated with death in ourseries, including reoperations, ischemia time, and modeof donor death. Previous cardiac operation has beendescribed as a risk factor in single- and multicenterreports [21, 22].

Risk factors for death after transplantation identifiedby multivariate analysis of data from the registry of theInternational Society for Heart and Lung Transplantation[1] identified recipient risk factors that were not found toaffect survival at our center. These included need for leftventricular assistance or ventilatory support preopera-tively, progressive age, and race. Likewise, the donor ageand race, and organ ischemia time, which were includedas risk factors in the registry report, were not found toaffect survival in our study.

It should be emphasized that risk stratification is adynamic process that may change among transplant eras.Certain risk factors identified at our center have beenneutralized over time, whereas some persist and yet othersemerge. Severe rejection was a risk factor in our earlyexperience [23]. Now it apparently is not, perhaps reflectinga more aggressive approach to this problem with earlierdiagnosis, improved immunosuppressive therapy, and in-creasing use of plasma exchange. Likewise the duration oforgan ischemia, although still less than 4.2 hours, waspreviously a risk factor but now appears not to be [23]. Thisfinding is supported by some authors [24] and refuted byothers [1, 25]. Intraaortic balloon pump counterpulsationhas recently emerged as a risk factor that may reflect thedebilitated condition of this subgroup of patients. Thereforein the era of worsening donor shortages and longer waits,transfer of IABP-dependent patients to LVAD support maybe advantageous.

This study has a number of limitations. The number of

Table 7. Correlation of Early Mortality and PulmonaryVascular Resistancea

PVR(Wood units)

EarlyMortality(,3 mo)

,2 6/71 (8.5%)2–2.99 6/61 (9.8%)3–3.99 4/43 (9.3%)4–4.99 4/22 (18.2%)$5 4/14 (28.6%)

Total 24/211 (11.4%)

a 211 patients with complete and verifiable data.

PVR 5 pulmonary vascular resistance.

Table 6. Impact of Recipient Age on Survival

Age (y)No. of

Patients1-Year

Survival5-Year

Survival

18–30 25 79.6% 62.9%30–39 55 90.8% 73.0%40–49 97 92.7% 68.5%50–59 178 85.3% 77.1%.60 50 87.4% 78.7%

Total 405 87.8% 73.4%

1577Ann Thorac Surg McCARTHY ET AL1998;65:1574–9 RISK AFTER HEART TRANSPLANTATION

patients analyzed is relatively small when compared withlarge registries [1, 25]. It concentrates only on mortality asan end point, and is retrospective. However, it shows thatmulticenter registry results do not always translate to asingle center, and thus it is of significance. This isparticularly important when considering the relativelyfew large-volume centers and the relatively larger num-ber of low-volume centers.

In conclusion, data from our center identified a groupof recipients, reportedly at high risk in multicenter reg-istries, who were not found to be at increased risk ofdeath after transplantation. This information supportsthe growing experience with older donors and recipientsand with bridged transplants, and has allowed us toexpand our donor pool to include older donors andlong-distance procurement. Our data also suggest thatthe experience level of the center, the presence of highpulmonary vascular resistance in the recipient, and pos-itive B-cell reactions on prospective flow cytometry cross-matching may be significant factors affecting outcome.These prognostic factors may allow more liberal selectionof patients and donors for transplantation.

References

1. Hosenpud JD, Novick RJ, Breen TJ, Keck B, Daily P. TheRegistry of the International Society for Heart and LungTransplantation: twelfth official report—1995. J Heart LungTransplant 1995;14:805–15.

2. Bourge RC, Stevenson LAW, Naftel DC, et al. Risk factors fordeath in critically ill patients awaiting cardiac transplantation:a multi-institutional study. Circulation 1995;92(Suppl 1):I-702.

3. Hoercher K, Young JB, Stewart RW, McCarthy PM, SmediraNG, Kendall K. The use of older heart donors may notimpact mortality or morbidity: a single center experience[Abstract]. J Heart Lung Transplant 1997;16:83.

4. Frazier OH, Rose EA, Macmanus Q, et al. Multicenterclinical evaluation of the HeartMate 1000 IP left ventricularassist device. Ann Thorac Surg 1992;53:1080–90.

5. McCarthy PM, Wang N, Vargo RL. Preperitoneal insertion ofthe HeartMate 1000 IP implantable left ventricular assistdevice. Ann Thorac Surg 1994;57:634–8.

6. Billingham ME, Cary NRB, Hammond ME, et al. A workingformulation for the standardization of nomenclature in thediagnosis of heart and lung rejection: heart rejection studygroup. J Heart Transplant 1990;9:587–93.

7. Nakhleh RE, Jones J, Goswitz JJ, Anderson EA, Titus J.Correlation of endomyocardial biopsy findings with autopsyfindings in human cardiac allografts. J Heart Lung Trans-plant 1992;11(3 pt 1):479–85.

8. Hurley JP, Cook DJ, McCarthy PM, et al. Flow cytometrycrossmatching: a method for monitoring antidonor antibod-ies in heart transplant recipients. Transplant Proc 1995;27:1301–3.

9. Cook DJ, Klingman LL, Koo AP, Goldfarb D, Dennis VW,

Hodge EE. Quantitative flow cytometry cross-matching forprecise measurement of donor-specific alloreactivity. Trans-plant Proc 1994;26:2866–7.

10. Iwaki Y, Cook DJ, Teresaki PI, et al. Flow cytometry cross-matching in human cadaver kidney transplantation. Trans-plant Proc 1987;19:764–6.

11. Laffel GL, Barnett AI, Finkelstein S, Kaye MP. The relationbetween experience and outcome in heart transplantation.N Engl J Med 1992;327:1220–5.

12. Hosenpud JD, Breen TJ, Edwards EB, Daily OP, HunsickerLG. The effect of transplant center volume on cardiac trans-plant outcome: a report of the United Nework for OrganSharing Scientific Registry. JAMA 1994;271:1844–9.

13. Massad M, McCarthy PM, Cook DJ, et al. Does successfulbridging with the implantable left ventricular assist deviceaffect cardiac transplantation outcome? J Thorac CardiovascSurg 1996;112:1275–83.

14. Hosenpud JD, Edwards EB, Lin HM, Daily OP. Influence ofHLA matching on thoracic transplant outcomes. Circulation1996;94:170–4.

15. Frazier OH, Rose EA, McCarthy PM, et al. Improved mor-tality and rehabilitation of transplant candidates treatedwith a long-term implantable left ventricular assist system.Ann Surg 1995;222:327–36.

16. Griepp RB, Stinson EB, Dong E Jr, Clark DA, Shumway NE.Determinants of operative risk in human heart transplanta-tion. Am J Surg 1971;122:192–7.

17. Kirklin JK, Naftel DC, McGriffin DC, McVay RF, BlackstoneEH, Karp RB. Analysis of morbid events and risk factors fordeath after cardiac transplantation. J Am Coll Cardiol 1988;11:917–24.

18. Hauptman PJ, Kartashov AI, Couper GS, et al. Changingpatterns in donor and recipient risk: a ten year evolution inone transplant center. J Heart Lung Transplant 1995;14:654–8.

19. McGriffin DC, Kirklin JK, Naftel DC, Bourge RC. Competingoutcomes after heart transplantation: a comparison of erasand outcomes. J Heart Lung Transplant 1997;16:190–8.

20. Drinkwater DC, Laks H, Blitz A, et al. Outcomes of patientsundergoing transplantation with older donor hearts. J HeartLung Transplant 1996;15:684–91.

21. De Maria R, Minoli L, Parolini M, et al. Prognostic determi-nants of six month morbidity and mortality in heart trans-plant recipients. J Heart Lung Transplant 1996;15:124–35.

22. Uthoff K, Wahlers T, Cremer J, Borst HG. Previous openheart operations: a contribution to impaired outcome aftercardiac transplantation? Ann Thorac Surg 1997;63:117–23.

23. Stewart RW. Cardiac transplantation at the Cleveland ClinicFoundation—the first ten years. In: Terasaki PI, Cecka JM,eds. Clinical transplants. Los Angeles: UCLA Tissue TypingLaboratory, 1994:127–31.

24. Briganti EM, Bergin PJ, Rosenfeldt FL, Esmore DS, RabinovM. Successful long-term outcome with prolonged ischemictime cardiac allografts. J Heart Lung Transplant 1995;14:840–5.

25. Young JB, Naftel DC, Bourge RC, et al. Matching the heartdonor and heart transplant recipient. Clues for successfulexpansion of the donor pool: a multivariable, multi-institutional report. J Heart Lung Transplant 1994;13:353– 65.

INVITED COMMENTARY

“Lies, damned lies and statistics”—Benjamin Disraeli, British Prime Minister 1874–1880

McCarthy and colleagues present the Cleveland Clinicexperience in cardiac transplantation and subject their

series of patients to univariate and multivariate analysislooking for preoperative risk factors that predict out-come. In their series of 405 patients, they demonstrateexcellent 1-year survival and fail to confirm several preop-erative risk factors that have been demonstrated to be

1578 McCARTHY ET AL Ann Thorac SurgRISK AFTER HEART TRANSPLANTATION 1998;65:1574–9

© 1998 by The Society of Thoracic Surgeons 0003-4975/98/$19.00Published by Elsevier Science Inc PII S0003-4975(98)00339-7