Registry of the International Society for Heart and Lung Transplantation: Twenty-second Official Adult Heart Transplant Report—2005

Abstract

In this, the 22nd official report of the Registry of the International Society for Heart and Lung Transplantation, we present information pertaining to >70,000 heart transplants performed worldwide. As in last year’s report,1Taylor D.O. Edwards L.D. Boucek M.M. et al.The registry of the International Society for Heart and Lung Transplantation 21st official adult heart transplant report—2004.J Heart Lung Transplant. 2004; 23: 797-803Abstract Full Text Full Text PDF Scopus (252) Google Scholar we focus our principal analysis for this report on recently transplanted patients rather than the entire cohort of heart recipients. With each subsequent report we find improving short-term outcomes. However, medication toxicities and complications of immunosuppression, as well as the ever-present threat of cardiac allograft vasculopathy, continue to limit the long-term survival of the typical heart transplant recipient. As a new addition to this year’s analysis, we utilize models developed in multivariate analyses to predict survival for a variety of hypothetical recipients to demonstrate the potential utility of such an approach for prospective risk stratification and candidate selection based on actual patient data rather than anecdotal assessment. All figures and tables from this report, and a more comprehensive set of Registry slides, are available at www.ishlt.org/registries/. Survival rates were calculated using the Kaplan-Meier method2Kaplan E.L. Meier P. Nonparametric estimation from incomplete observations.J Am Stat Assoc. 1957; 53: 457-481Crossref Scopus (48139) Google Scholar and compared using the log-rank test. Multivariate analyses were performed using Cox proportional hazards regression.3Cox D.R. Oakes D. Analysis of survival data. Chapman and Hall, London1984Google Scholar The results of the multivariate analyses are reported as relative risks (RRs) with either a corresponding p-value and/or 95% confidence limits. Factors with RR significantly >1 indicate that the factor is associated with an increased likelihood of the event (e.g., mortality, development of coronary allograft vasculopathy) occurring. Conversely, RR <1 indicates that the event is less likely to occur when that factor is present. Based on the RRs and baseline survivals obtained by the Cox proportional hazards models, predicted survival rates were computed for specified patient/donor/transplant profiles. The number of heart transplant procedures reported to the Registry each year continues to decrease (Figure 1). As pointed out in prior reports,1Taylor D.O. Edwards L.D. Boucek M.M. et al.The registry of the International Society for Heart and Lung Transplantation 21st official adult heart transplant report—2004.J Heart Lung Transplant. 2004; 23: 797-803Abstract Full Text Full Text PDF Scopus (252) Google Scholar, 4Taylor D.O. Edwards L.B. Mohacsi P.J. et al.The registry of the International Society for Heart and Lung Transplantation 20th official adult heart transplant report—2003.J Heart Lung Transplant. 2003; 22: 616-624Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar this appears to be due to decreased reporting from centers outside of the USA (www.ishlt.org/registries/). Analyses of a variety of worldwide transplant registries suggest that >2000 heart transplants are performed annually at non-ISHLT reporting centers, placing the annual number of heart transplants at >4,000 worldwide (P. Mohacsi, personal communication, manuscript in preparation). Figure 2 depicts the distribution of transplants stratified by average annual heart transplant center volume from January 1, 1997 through June 30, 2004. Forty-three percent of the reporting transplant centers averaged <10 procedures per year and these centers account for approximately 13% of all heart transplant procedures. Higher-volume centers (those averaging >30 procedures per year and representing approximately 13% of centers) account for 39% of the total number of procedures.Figure 2Distribution of heart transplants stratified by average annual center volume for heart transplants performed January 1, 1997 through June 30, 2004.View Large Image Figure ViewerDownload (PPT) In the overall Registry experience (1982 to 2004), the primary indication for adult heart transplantation has been equally balanced between coronary heart failure and non-coronary cardiomyopathy (45% each). Valvular (3% to 4%), adult congenital (2%), re-transplantation (2%) and miscellaneous causes make up the remainder. These indications for transplantation have not changed significantly over the last 10 years (www.ishlt.org/registries/). Figure 3 demonstrates the age distribution of heart transplant recipients by era. Note that, over the last 20 years, the percentage of recipients >65 years of age has steadily increased, while the percentage of recipients at 40 to 49 years has steadily decreased. The characteristics of the most recent cohort transplanted from January 1, 2002 to June 30, 2004 are shown in Table 1. Note that 48% of recipients are now receiving intravenous inotropic support and almost 27% are on some type of mechanical circulatory support (21% with a left ventricular assist device [LVAD]). This is a significant increase compared with the time period of January 1, 1999 to June 30, 2001, when 34% were on inotropic support and 15% were on mechanical support (11% on LVAD support).4Taylor D.O. Edwards L.B. Mohacsi P.J. et al.The registry of the International Society for Heart and Lung Transplantation 20th official adult heart transplant report—2003.J Heart Lung Transplant. 2003; 22: 616-624Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar However, substantially fewer recipients are hospitalized immediately before transplant in the most recent cohort, 48.2% vs 72.3% (1999 to 2001).4Taylor D.O. Edwards L.B. Mohacsi P.J. et al.The registry of the International Society for Heart and Lung Transplantation 20th official adult heart transplant report—2003.J Heart Lung Transplant. 2003; 22: 616-624Abstract Full Text Full Text PDF PubMed Scopus (188) Google Scholar This likely reflects the current practice of outpatient management of LVAD bridge-to-transplant and inotrope bridge-to-transplant patients where possible. The percentage of recipients with their most recent panel-reactive antibody (PRA) level being >10% has increased from 5.0% to 9.1% during this same time period.4Taylor D.O. Edwards L.B. Mohacsi P.J. et al.The registry of the International Society for Heart and Lung Transplantation 20th official adult heart transplant report—2003.J Heart Lung Transplant. 2003; 22: 616-624Abstract Full Text Full Text PDF PubMed Scopus (188) Google ScholarTable 1Characteristics of Recent Adult Heart Transplant Recipients (January 1, 2002 Through June 20, 2004)Recipient age (years)50.8 ± 12.2 (18–76)Donor age (years)33.0 ± 12.7 (10–60)Recipient/donor gender (% male)76.6%/67.8%Recipient weight (kg)76.7 ± 14.4 (40–110)Recipient body mass index25.6 ± 4.1 (13.0–44.5)Recipient/donor diabetes mellitus19.7%/1.8%Recipient amiodarone use (USA only)28.8%Recipient/donor cigarette-smoking history41.4%/30.4%Ischemic time (hours)3.1 ± 1.0 (0.4–6.0)Most recent panel-reactive antibodies ≥10%9.1%Pulmonary vascular resistance (Wood units)2.6 ± 1.4 (1.0–10.0)HLA mismatches 0–24.4% 3–439.5% 5–656.0%Diagnosis Coronary artery disease42.6% Cardiomyopathy46.0% Valvular2.7% Re-transplant2.2% Congenital2.4% Other causes4.0%Donor cause of death Head trauma62.8% Stroke26.9% Other10.4%Pre-operative support Hospitalized at time of transplant48.3% On intravenous inotropes47.9% Left ventricular assist device20.6% Intra-aortic balloon pump5.7% Ventilator2.5% Right ventricular assist device0.6% Total artificial heart0.5% Extracorporeal membrane oxygenation0.4%Data expressed as mean ± standard deviation (range) (N = 6,808). Percentages calculated based only on known responses. Open table in a new tab Data expressed as mean ± standard deviation (range) (N = 6,808). Percentages calculated based only on known responses. Figure 4 demonstrates the trends in anti-lymphocyte antibody use for induction immunosuppression over the last 3.5 years (January 2001 to June 2004). Peri-operative anti-lymphocyte antibody use appears to have leveled off at approximately 45% to 50% of patients transplanted; however, there is a very recent 5% increase in 2004. The peri-operative use of OKT3 as an induction agent continues to decrease (only 4% of heart transplant procedures) and peri-operative use of IL-2-receptor antibodies and polyclonal anti-thymocyte antibodies continues to increase (now used in 26% and 23% of heart transplant procedures, respectively). Figure 5 portrays maintenance immunosuppression protocols at 1-year post-transplant for 3 snapshots in time—1998, 2002 and the first half of 2004. Tacrolimus is currently the most common calcineurin inhibitor, used by 50% of patients transplanted during the first half of 2004, with cyclosporine use dropping to 47%. Mycophenolate mofetil remains the predominant anti-proliferative agent, used in >75% of patients. Rapamycin use doubled in the last 2 years with 14% of patients alive at 1-year post-transplant during the first half of 2004 while receiving it as part of their immunosuppressive protocol, compared with only 7% during 2002. Prednisone use has decreased over the last 6 years but is still used by almost 75% of patients at 1-year post-transplant.Figure 5Maintenance immunosuppression agents at 1 year post-transplantation reported for 1998, 2002 and January to June 2004. Different patients analyzed in each time period. MMF, mycophenolate mofetil.View Large Image Figure ViewerDownload (PPT) Heart transplant Kaplan-Meier survival curve and graft half-time calculations are shown in Figure 6. Note that the survival curve for the entire cohort (1982 to 2003) continues to demonstrate that, after the steep fall in survival during the first 6 months, survival decreases at a very linear rate (approximately 3.4% per year), even well beyond 15 years post-transplant. In addition, there does not appear to be a point beyond which the slope of the survival curve decreases to approach that seen for the general population. The slight decrease in the slope of the curve during Years 18 to 21 is most likely an artifact of data reporting. The same pattern was seen during the last 3 years with regard to the curves for each annual report; however, the tail of the curve resumes its prior slope with additional patient follow-up in the subsequent years. As noted in prior years’ analyses, survival improves with each successive 5- to 6-year era (Figure 7). In fact, the continued improvement in survival is likely more significant than is apparent from the survival curves given that the risk profile of recipients continues to increase. In the multivariate analysis (see later), transplants performed in 2002 experienced an 18% lower risk of 1-year mortality than those performed in 2000. Overall survival stratified by recipient age for patients transplanted in the most recent era (1996 to 2003) and survival curves stratified by diagnosis for recent transplants (1996 to 2003) are presented on the Registry website (www.ishlt.org/registries/). Survival for re-transplantation has improved significantly compared with the prior era. Patients undergoing re-transplantation at >12 months after initial transplantation have 1-year survival rates of approximately 82% (similar to the contemporary cohort of primary transplants at 83%) (www.ishlt.org/registries/).Figure 7Kaplan-Meier survival by era for heart transplants performed between January 1982 and June 2003.View Large Image Figure ViewerDownload (PPT) Figure 8 demonstrates the survival curves for recipients undergoing simultaneous heart and kidney transplant, compared with the cohort receiving heart transplant alone during the same time period. As noted in previously reported single-center and smaller registry series, combined (simultaneous or same donor) heart/kidney transplantation can be performed in select patients, with similar survival to the overall cohort of heart recipients. However, survival for patients receiving a kidney transplant later after a heart transplant (median 8 years after heart transplant) is significantly lower than for heart-only recipients at the same time-points after heart transplant (84% vs 93% for the first year after renal transplant report and Year 8 to Year 9 for heart-only, respectively, and 49% vs 65% at 5 years after renal transplant report and Year 8 to Year 13 for heart-only, respectively) (Figure 9). Yet it appears that the bulk of the mortality difference occurs very early after kidney transplant, suggesting that a “successful” renal transplant (crudely defined as patient alive 6 months after renal transplant) returns the recipient to the “usual” heart transplant survival curve.Figure 9Kaplan-Meier survival for heart transplant recipients undergoing subsequent (late) kidney transplantation compared with heart-alone transplantation performed between January 1982 and June 2003. Median time to kidney transplant = 8.0 years after heart transplant.View Large Image Figure ViewerDownload (PPT) To keep up with the evolution of immunosuppressive therapy and post-transplant management, the cohorts analyzed must be as current as possible to provide clinically useful data. Accordingly, we focus this year’s analysis of 1-year mortality on the most recent cohort, transplanted from January 2000 through June 2003, and compared findings with the cohort transplanted during the 4 preceding years (January 1996 through December 1999). Table 2 depicts the categoric risk factors for 1-year mortality. Having adult congenital heart disease as the indication for heart transplantation remains a powerful risk factor for 1-year mortality, with 75% of the first-year mortality occurring within the first month post-transplantation. Requiring temporary mechanical circulatory support (short-term, extracorporeal), requiring dialysis at the time of transplant, requiring mechanical ventilation at the time of transplant, having coronary artery disease as the indication for transplant, and being hospitalized immediately before transplant are risk factors for 1-year mortality in the most recent era, as in the prior era.Table 2Risk Factors for Mortality Within 1 Year for Adult Heart Transplants Performed Between January 1996 and June 20031996 to 1999 (N = 13,365)2000 to 6/2003 (N = 9,743)NRelative riskp-valueNRelative riskp-valueCategoric factors Diagnosis of congenital heart disease2031.770.00012212.51<0.0001 Temporary circulatory supportaTemporary circulatory support includes ECMO and Abiomed.462.40.00011292.07<0.0001 Ventilator2631.590.00022441.730.0001 Dialysis1601.560.00182221.690.0001 Hospitalized at transplant (including ICU)9,4941.3<0.00016,0841.58<0.0001 Donor history of cancer1501.220.27191371.580.0117 Female recipient/male donor1,2441.110.2229711.460.0001 Donor cause of death: cerebrovascular/stroke3,4190.980.69722,6581.270.0002 Female recipient/female donor1,5491.070.42981,2731.250.0292 Pulsatile long-term VAD304bVAD type collected for only part of era. Only those with known type were included in this group.0.990.9341,2341.220.0394 Repeat transplant3061.91<0.00012251.180.3747 Transplant year = 2000 vs 2002/20032,8921.180.0101 Diagnosis of coronary artery disease6,0301.130.0124,3241.170.0128 PRA >10%6061.40.00024921.130.3018 HLA-DR mismatches (per mismatch)591 (0 MM)1.160.0003433 (0 MM)1.10.0596 Intra-aortic balloon pump4791.410.00064271.020.8722 HLA-B mismatches (per mismatch)237 (0 MM)1.170.0015155 (0 MM)1.020.7367 Repeat transplant3061.91<0.00012251.180.3747 Infection requiring intravenous drug therapy within 2 weeks of transplant6911.210.03667980.990.9348 Prior transfusions1,3671.20.01291,4540.90.2329 Intravenous inotropes4,8110.79<0.00013,4010.75<0.0001Continuous factors Recipient age<0.0001<0.0001 Donor age<0.0001<0.0001 Recipient weight pre-transplant0.00060.53 Recipient body mass index pre-transplantcTwo separate models were fit to estimate the impact of recipient weight and recipient body mass index. All other factors were included in both of the models; the parameter estimates shown in the table for all factors except recipient body mass index were obtained from the model that included recipient weight.0.040.43 Weight ratio: donor weight/recipient weight0.00020.11 Center volume (inverse relationship)0.0001<0.0001 Ischemia time<0.0001<0.0001 Recipient bilirubin pre-transplant0.00040.0004 Recipient serum creatinine pre-transplant<0.0001<0.0001 Recipient diastolic pulmonary artery pressure0.00870.0073ECMO, extracorporeal membrane oxygenation; PRA, panel-reactive antibodies; VAD, ventricular assist device.a Temporary circulatory support includes ECMO and Abiomed.b VAD type collected for only part of era. Only those with known type were included in this group.c Two separate models were fit to estimate the impact of recipient weight and recipient body mass index. All other factors were included in both of the models; the parameter estimates shown in the table for all factors except recipient body mass index were obtained from the model that included recipient weight. Open table in a new tab ECMO, extracorporeal membrane oxygenation; PRA, panel-reactive antibodies; VAD, ventricular assist device. Although a diagnosis of congenital heart disease and the use of temporary mechanical support (short-term, extracorporeal) are the 2 strongest independent risk factors for 1-year mortality, it is important to note that these represent only 2.3% and 1.3% of the recipients, respectively, during this era. Pre-transplant inotrope use is associated with a similar reduction in mortality in both eras. Several factors have differing impacts during the 2 eras. Repeat transplantation, a powerful predictor of risk in the earlier era, is less predictive in the current era (RR = 1.2, p = 0.35). Likewise, intra-aortic balloon pumping at the time of transplant seems to have less effect in the current era despite similar recipient utilization. A donor history of cancer and cerebrovascular donor cause of death are both significant risk factors in the current era, yet not during the prior era. Female recipients of either male or female donor hearts are at increased risk for mortality in the most recent era, in contrast to the prior era. Patients with pulsatile (long-term) mechanical circulatory support before their transplant have a 22% increased risk in the current era, as compared with no significant risk in the prior era. There are 1,234 recipients who received these devices in the current era compared with only 304 in the prior era. The difference in survival remains when only the so-called “implantable” devices (Thoratec HeartMate and Novacor WorldHeart) are considered (www.ishlt.org/registries/). The database only recently began differentiating the various types of ventricular assist devices; thus, in future analyses with larger numbers the impact of individual devices can be more thoroughly investigated. For more information regarding mechanical circulatory support, one is referred to the ISHLT Mechanical Circulatory Support Device (MCSD) Registry website (www.ishlt.org/registries/mcsdDatabase/) and Deng et al.5Deng M.C. Edwards L.B. Hertz M.I. et al.Mechanical Circulatory Support Device Database of the International Society for Heart and Lung Transplantation second annual report—2004.J Heart Lung Transplant. 2004; 23: 1027-1034Abstract Full Text Full Text PDF PubMed Scopus (63) Google Scholar, 6Deng MC, Edwards LB, Taylor DO, et al. Mechanical Circulatory Support Device Database of the International Society for Heart and Lung Transplantation: third annual report—2005. J Heart Lung Transplant (in press).Google Scholar As mentioned earlier, year of transplant is a significant risk factor, with an 18% decrease in risk in 2002 compared with 2000. Significant continuous factors include recipient and donor age (Figure 10, Figure 11), donor heart ischemic time, center volume (inverse relationship), recipient pulmonary artery (PA) diastolic pressure, recipient bilirubin and recipient creatinine (www.ishlt.org/registries/). Recipient weight, recipient body mass index (BMI) and donor/recipient weight ratio are not significant predictors of 1-year mortality in the latest era, in contrast to the prior era when all 3 were significant (www.ishlt.org/registries/). Several previously identified categoric risk factors appear to have less impact on mortality in the current era than in the prior era (Table 2). Several additional factors that appear not (or no longer) to be risk factors include symptomatic cerebrovascular disease, history of pregnancy, recipient history of malignancy, donor history of hypertension, donor history of diabetes, donor height and donor/recipient height ratio (see online Registry slide set for complete list).Figure 11Effect of donor age on 1-year mortality for heart transplants performed between January 1996 and June 2003 by era of transplantation (for 1996 to 1999 analysis: N = 13,365; for 2000 to 6/2003 analysis: N = 9,743).View Large Image Figure ViewerDownload (PPT) Because mortality during the first year is 1.4 times that of the next 4 years combined, risk factors for 1-year mortality remain powerful predictors for 5-year outcomes. Thus, these factors can conceal the effects of important factors that may impart their effect over longer periods of time. In an attempt to adjust for this effect, we again present the risk factor analysis for 5-year survival only in those patients who have survived the first year post-transplant (i.e., 5-year “conditional” survival). For those interested in risk factors for non-conditional 5-year survival, the data are available in the complete Registry slide set (www.ishlt.org/registries/). Table 3 highlights the categoric risk factors for 5-year mortality conditional on survival to 1 year. In the cohort of >9,000 patients transplanted from January 1996 through June 1999, risk factors for 5-year “conditional” mortality include previous transplant, stroke before transplant discharge, cerebrovascular event before transplant, allograft vasculopathy identified during Year 1, diabetes before transplant, treatment for rejection during Year 1, treatment for infection before transplant discharge and HLA-DR mismatches. As in the 1-year risk factor analysis, year of transplant is a significant independent risk factor, with a 20% reduction in risk between the beginning of the era (1996) and the end (1998/1999). Continuous variables that remain predictors of 5-year conditional mortality in the most recent era include recipient age, donor age and recipient weight (www.ishlt.org/registries/). As noted in last year’s report,1Taylor D.O. Edwards L.D. Boucek M.M. et al.The registry of the International Society for Heart and Lung Transplantation 21st official adult heart transplant report—2004.J Heart Lung Transplant. 2004; 23: 797-803Abstract Full Text Full Text PDF Scopus (252) Google Scholar the risk curve for recipient age is U-shaped, with the younger and older groups having greater risk of 5-year conditional mortality than the age group between 50 and 55 years.Table 3Risk Factors for Mortality Within 5 Years, Conditional on Survival to 1 Year, for Adult Heart Transplants Performed Between January 1996 and June 1999 (N = 9,120)NRelative riskp-value95% confidence interval for Relative riskCategoric factors Repeat transplant1811.70.00721.16–2.52 Stroke prior to transplant discharge1001.660.01381.11–2.48 Cerebrovascular event prior to transplant2991.40.00941.09–1.81 Coronary artery vasculopathy during Year 14321.40.00251.13–1.74 Diabetes9991.350.00031.15–1.58 Treated for rejection during Year 12,5741.3<0.00011.15–1.47 Treated for infection prior to transplant discharge1,1661.260.00231.09–1.47 Transplant year = 1996 vs 1998/19992,8251.240.00121.09–1.42 HLA-DR mismatches (per mismatch)425 (0 MM)1.160.00471.05–1.28Continuous factors Recipient age0.002 Donor age0.0003 Recipient weight pre-transplant0.0014 Recipient body mass index pre-transplantaTwo separate models were fit to estimate the impact of recipient weight and recipient body mass index. All other factors were included in both of the models; the parameter estimates shown in the table for all factors except recipient body mass index were obtained from the model that included recipient weight.0.18a Two separate models were fit to estimate the impact of recipient weight and recipient body mass index. All other factors were included in both of the models; the parameter estimates shown in the table for all factors except recipient body mass index were obtained from the model that included recipient weight. Open table in a new tab Selected factors that are not (or no longer) significant risk factors for 5-year conditional mortality include: gender; ventricular assist device at the time of transplantation; ventilator requirement; intravenous inotropes or hospitalization at time of transplant; center volume; recipient height, weight or body mass index; panel-reactive antibody test positivity; and dialysis at time of transplant. A full list of non-significant factors can be found in the Registry slide set (www.ishlt.org/registries/). As with last year’s analysis,1Taylor D.O. Edwards L.D. Boucek M.M. et al.The registry of the International Society for Heart and Lung Transplantation 21st official adult heart transplant report—2004.J Heart Lung Transplant. 2004; 23: 797-803Abstract Full Text Full Text PDF Scopus (252) Google Scholar we extended the survival analysis out to 10 years, recognizing the limitations of analyzing a cohort initially transplanted from 1990 through June 1993. This year we include an additional 5,000 patients in the analysis (n = 10,982). As in the 5-year analysis just presented, we performed a “conditional survival analysis,” this time analyzing only patients who survived the first 3 years to focus on factors impacting late survival. Table 4 highlights the categoric risk factors for 10-year mortality conditional on 3-year survival. Previous transplant and pre-transplant diagnosis of coronary artery disease are predictors of 10-year conditional mortality. In addition, any pre-transplant diagnosis other than coronary artery disease (CAD), cardiomyopathy, valvular heart disease or congenital heart disease increases the 10-year risk by 24%. As in the 5-year analysis, HLA-DR mismatches increase the risk, but only by 9% per mismatch. Just as in the 1- and 5-year analyses, year of transplant is a significant risk factor.Table 4Risk Factors for Mortality Within 10 years, Conditional on Survival to 3 Years, for Adult Heart Transplants Performed Between January 1990 and June 1993 (N = 10,982)NRelative riskp-value95% confidence interval for Relative RiskCategoric factors Repeat transplant1641.370.03711.02–1.83 Diagnosis other than cardiomyopathy, coronary artery disease, congenital heart disease or valvular heart disease5381.240.03781.01–1.52 Diagnosis of coronary artery disease4,9461.19<0.00011.1–1.29 Transplant year: 1990 vs 1992/19932,3641.110.02991.01–1.21 HLA-DR mismatches (per mismatch)534 (0 MM)1.090.01081.02–1.16 Female/no previous pregnancy1,2120.830.0060.73–0.95Continuous factors Recipient age<0.0001 Donor age<0.0001 Recipient BMI pre-transplant<0.0001 Donor weight0.0004 Open table in a new tab Surprisingly, only a 2-year difference in the time of transplant (1990 vs 1992/1993) is significant, with an 11% decrease in risk for those transplanted in 1992/1993. Female gender (if no previous pregnancy) appears to be protective. Recipient and donor age remain powerful continuous variables, just as in the 1- and 5-year analyses. However, the risk curve for recipient age is no longer U-shaped as in the 5-year analysis, but is relatively flat up to age 40 to 45 years when the risk begins to increase progressively with increasing age (www.ishlt.org/registries/). Recipient obesity (as reflected by BMI at the time of transplant) is a significant continuous variable, whereas donor weight demonstrates an inverse correlation (lower donor weight associated with greater risk) (www.ishlt.org/registries/). With larger numbers in this year’s analysis, center volume is no longer a significant predictor for 10-year conditional mortality. Adjudication of primary cause of death is particularly problematic in multicenter registries. This is due not only to actual difficulties in assigning priority to the multiple competing events, but also to the difficulties inherent in a registry reporting format utilizing non-uniform definitions. These limitations aside, review of the stated causes of death after heart transplantation can provide useful information. Included in this year’s Registry slide set (www.ishlt.org/registries/) is a table with the breakdown of causes of death from heart transplant recipients transplanted from January 1992 through June 2004, categorized by time post-transplant. Within the first 30 post-transplant days, graft failure (primary and non-specific) accounts for 40% of deaths, followed by multiorgan failure (14%) and non-cytomegalovirus (non-CMV) infection (13%). From 31 to 365 days, non-CMV infection accounts for almost 33% of the deaths, followed by graft failure (primary and non-specific) (18%) and acute rejection (12%). After 5 years, allograft vasculopathy (CAV) and late graft failure (likely due to allograft vasculopathy) together account for 30% of deaths, followed by malignancies (including lymphoma) (23%), and non-CMV infections (10%). In addition, this year we stratified the causes of death by era to investigate potential trends. Comparing the 1998 to 2004 deaths reported with the 1992 to 1997 deaths we see a significant decrease in primary and non-specific graft failure within the first 30 days during the most recent era (57% vs 43%) (www.ishlt.org/registries/). This year we again present data on the incidence of treated rejection during the first post-transplant year and its relationship to the various immunosuppressive protocols. Patients receiving OKT3 as part of an induction protocol have higher rejection rates during the first year after transplant than those receiving polyclonal or interleukin-2 antibody induction, and those receiving no antibody induction (Figures 12). Likewise, patients who at transplant discharge are receiving tacrolimus-based immunosuppression, particularly when in combination with mycophenolate mofetil, appear to have lower rejection rates (Figure 13). For this analysis, rejection is defined as the percent of patients treated during Year 1 and as the average number of rejection episodes. However, our analysis is limited by the fact that the Registry does not collect data on untreated mild-to-moderate rejection episodes (ISHLT Grades 1A, 1B and 2) or details regarding antibody-mediated rejection. More importantly, this is not a prospective, randomized “study,” thus “cause-and-effect” conclusions cannot be drawn. Further study, possibly utilizing propensity-matching analysis, will be needed to better evaluate these relationships.Figure 13Percentage of adult heart transplant recipients treated for rejection during the first year post-transplantation stratified by immunosuppressive medications at transplant discharge, age and gender.View Large Image Figure ViewerDownload (PPT) For evaluation of the most common post-transplant morbidities, this year’s analysis includes those transplants performed from April 1994 through June 2003. This year we compare the 1-year morbidities during the first part of the era (1994 to 1999) to the later part (2000 to 2003). It appears that the incidence rates of renal insufficiency, hyperlipidemia and diabetes increased over time (Table 5). Five- and 8-year incidences are included in the full slide set (www.ishlt.org/registries/). Similar to last year’s analysis, we examine CAV by determining risk factors for early CAV (occurring within 3 years post-transplant) and for late CAV (occurring within 7 years, conditional on having no CAV at 1 year), now with >6,000 recipients in the 3-year analysis and >1,400 recipients in the 7-year analysis. As in prior, smaller-cohort analyses of early CAV, donor hypertension remains a categoric risk factor for early disease, whereas female donor and recipient are associated with lower risk. However, in this larger analysis, pre-transplant coronary artery disease is not a significant predictor for early CAV. In addition, prior transfusions appear to be protective of early CAV and an episode of infection requiring intravenous antibiotics within 2 weeks of transplant increases the risk of early CAV.Table 5Cumulative Prevalence of Post-transplant Morbidity in Survivors Within 1 Year for Transplants Performed Between April 1994 Through June 2003OutcomeTransplants: April 1994 through December 1999Transplants: January 2000 through June 2003Within 1 yearTotal Number With Known ResponseWithin 1 yearTotal Number with known ResponseHypertension72.60%(N = 9,659)76.80%(N = 5,668)Renal dysfunction All25.70%(N = 9,534)31.70%(N = 5,727) Abnormal creatinine <2.5 mg/dl15.20%22.10% Creatinine >2.5 mg/dl9.10%7.80% Long term Dialysis1.20%1.50% Renal transplant0.20%0.30%Hyperlipidemia49.80%(N = 10,225)68.70%(N = 6,035)Diabetes23.70%(N = 9,634)30.20%(N = 5,674)Coronary artery vasculopathy8.70%(N = 8,577)7.00%(N = 5,125) Open table in a new tab Within this larger cohort of recipients, year of transplant is an important predictor of early CAV. When compared with the most recent cohort (2001/2002), recipients transplanted in 1996, 1997 or 1998 had a greater risk of early CAV. Recipient age (inverse relationship), donor age, center volume and recipient BMI are independent continuous risk factors (www.ishlt.org/registries/). Categoric risk factors for the development of angiographic CAV of any severity within 7 years after transplantation, conditional on having no significant CAV at 1 year, include donor history of diabetes and cerebrovascular event as donor cause of death, whereas female donor and HLA-B mismatches appear protective. With this larger cohort, pre-transplant coronary artery disease, HLA-DR mismatch, donor hypertension and hospitalization for rejection during Year 1 are no longer predictive. Significant continuous risk factors include recipient age (inverse relationship), donor age, recipient BMI and center volume (www.ishlt.org/registries/). Figure 14 demonstrates the Kaplan-Meier freedom from severe renal dysfunction and freedom from CAV. Note the steady slope of the curves such that, at 9.5 years, only 47% of patients are free of angiographic CAV and only 60% are free of severe renal dysfunction (defined as creatinine >2.5, dialysis or renal transplant). The demographics of post-transplant malignancies have not changed significantly since last year’s analysis. A separate analysis has been performed and a manuscript on this topic is currently in preparation. We have updated the Registry slide set to include 8-year cumulative data, which now reveal an incidence of malignancy of 26% by 8 years, the large majority being skin cancers (occurring in 18% of 8-year survivors) (www.ishlt.org/registries/). This year we utilized the Cox proportional hazards analysis to estimate survival rates for several hypothetical patients. This type of analysis may be very helpful to the clinician in several situations. Accurate prediction of short- and/or long-term survival for a particular patient based on up-to-date risk factor analysis might allow better decision-making when confronted with candidate selection. Most centers make candidate selection decisions on the basis of individual (or group) “intuition,” based on prior experience or published data. Often our experiential recall is tainted by the last “great” or “horrible” case, and the published data are often outdated. Another valuable application of such an analysis could be the prediction of post-transplant morbidity to allow tailoring of the medical therapy and follow-up. For example, if a patient’s risk of early rejection and infection could be accurately predicted pre-transplant, the immunosuppressive regimen could be adjusted accordingly a priori rather than post hoc as we currently do. In addition, the ideal situation would allow future post-transplant events to enter the analysis and new curves could be generated for future events. Clearly, that is what we currently do individually in daily practice, but without the power of sophisticated analyses of large databases. Figure 15 is an example of an estimated survival curve (generated from the 5-year survival analysis of patients transplanted from 1995 to 1999) for what many would consider a “high-risk” candidate—a 30-year-old multi-gravida female with pre-transplant PRA of 50%, hospitalized on inotropes, transplanted in 1999, and receiving a donor heart from a 30-year-old woman. The curves demonstrate quite reasonable 1- and 5-year survival when compared with the average profile for the cohort (86% vs 88% and 76% vs 80%, respectively). Figure 16 demonstrates the 10-year predicted survival of the same patient (generated from the 10-year conditional analysis of patients transplanted from 1990 to 1993). Note that, if this patient survives to Year 3 post-transplant, her predicted survival to 10 years is actually better than the cohort average (77% vs 73%). As a second example case, Figure 17 demonstrates an estimated survival curve for what is an increasingly common candidate—a 63-year-old man with CAD, PRA of <10%, creatinine of 1.8, a pulsatile long-term VAD, weight of 95 kg, height of 70 inches, a 3-hour cold ischemia time and average (cohort mean) donor and other risk factors. Note that the 1- and 5-year survival is significantly lower than the cohort average (79% vs 88% and 65% vs 80%, respectively) and, even if he survives 3 years, his predicted subsequent survival to 10 years is markedly inferior to the cohort average (54% vs 73%) (Figure 18). The juxtaposition of these 2 cases clearly demonstrates the importance of separating the risk factors for short- and long-term survival, many of which are “competing.”Figure 16Ten-year conditional predicted survival curve for a hypothetical 30-year-old woman (conditional on survival to 3 years post-transplant).View Large Image Figure ViewerDownload (PPT)Figure 17Five-year predicted survival curves for hypothetical 63-year-old man.View Large Image Figure ViewerDownload (PPT)Figure 18Ten-year conditional predicted survival curve for hypothetical 63-year-old man (conditional on survival to 3 years post-transplant).View Large Image Figure ViewerDownload (PPT) For a registry or database to be clinically useful and directly impact patient care it must provide up-to-date and accurate information. At first glance, the discrepancies between successive year’s reports suggest problems with data validity or analysis. However, it is more likely that the changing risk factors and outcomes are more a reflection of changing medical management and patient selection—proof that we actually do “learn from our mistakes.” Likewise, as recipient and donor characteristics continue to change over time, today’s “risk factors” will no longer be tomorrow’s, and newly identified factors and factors that we currently do not view as risk factors based on prior analyses will become important.