The Effect of Recipient's Age on Lung Transplant Outcome

Authors


* Corresponding author: Michael Hutcheon, michael.hutcheon@uhn.on.ca

Abstract

Selection criteria for organ transplantation have evolved over time. Age has been revisited periodically. We studied the outcome of lung transplant adjusted by age in a single center transplant population. We matched the 42 lung graft recipients older than 60 years transplanted by July 1999 to younger controls by lung disease, transplant era within 2 years, type of transplant and gender. The female to male ratios were 17/25 among the older cohort (median age 61.6 years), and 15/27 (median age 51.9 years) among the matched younger. Survival analysis demonstrated a significant difference: at 1 year, 60% versus 86%, and at 5 years, 37% versus 57%, for older and younger, respectively, p = 0.005. Excess annual mortality, calculated with the declining exponential approximation to life expectancy (DEALE), showed an older/younger ratio of 1.9. Eleven deaths occurred within 6 months among the older patients, 10 due to infection. After 6 months, there were 20 more deaths, 6 due to malignancy, 5 to Bronchiolitis Obliterans Syndrome (BOS), 3 to infection and 6 to other causes. Among the younger there were 6 deaths within 6 months and 12 more thereafter; among the latter, 8 were due to BOS.

Despite stringent selection, lung transplant recipients older than 60 years show increased mortality even after adjusting for their expected higher age-related mortality.

Introduction

Guidelines for lung transplantation traditionally have included age as a criterion to be considered in the selection of potential candidates (1,2). The perception of increased complications and decreased survival rates among the elderly as compared to the possibility of benefiting younger persons with better outcomes lies at the heart of the dilemma. This becomes even more relevant when the scarcity of available donor organs is factored into the decision of transplanting either a younger or an older candidate in comparable conditions. Such a decision should be based on allocation of a scarce resource justified by ethically relevant considerations (3). Objective data should help making such decisions less controversial.

Studies in renal, liver and heart transplantation have shown that older patients can be successfully transplanted (4–6). On the other hand, ISHLT Registry data show that increasing age is a risk factor for increased mortality following lung transplantation (7). In order to gain insight into the outcome of older lung transplant recipients we studied the experience of the Toronto Lung Transplant Program in recipients aged greater than 60 years. We carried out a retrospective matched double cohort study and analyzed outcome adjusting by age.

Methods

As of July 1999, 42 patients older than 60 years have received a lung transplant, no retransplants. We wrote a computer program to find a control among first lung transplant recipients younger than 60 years matching by lung disease, type of transplant (single or bilateral), transplant era within (plus or minus) 2 years and gender. The process was run blindly (outcomes were not in the selection algorithm). For morbidity within 3 months after transplant, we looked at length of stay in hospital, number of infections and acute rejection episodes. Acute rejection was considered probable when there was at least a 10% fall in FEV1, the absence of an infectious explanation for symptoms or a decline in FEV1, no histopathological confirmation of acute rejection and the patient responded to bolus solumedrol treatment with resolution of symptoms and objective findings. Acute rejection was considered definite when there was at least a 10% fall in FEV1, no other cause was identified to explain the drop in FEV1 and there was histopathological confirmation of Grade II or greater rejection as defined by the International Society for Heart and Lung Transplantation or, in the absence of clinical findings, when Grade II or greater rejection was found on surveillance biopsy.

In addition we looked at other complications. As a measure of level of activity, the 6-min walk test before and 3 months after transplant was evaluated. Causes of death were arbitrarily classified by period of time as ‘early’ within 6 months of transplant and ‘late’ after that. Standard lung procurement, preservation, surgical procedure, immunosuppression, prophylactic antibiotics and treatment of infections and rejections were utilized in all patients (8) (see Appendix 1). Charts were reviewed in accordance with UHN Ethics Research Board guidelines.

Statistical analysis

Survival probability estimates were obtained using the Kaplan–Meier method and distributions were compared via the log-rank test. Length of stay in hospital was compared via number of days out of hospital within 30 days after transplant by the Wilcoxon rank sum test. Proportions were compared via chi-squared test. The 6-min walk test was analyzed via ANCOVA. Statistical significance was defined as p < 0.05. Tests were two-tailed. Stratified analysis using conditional logistic regression was modeled separately for death at 1 and 5 years to adjust for possible confounding factors in mortality (9). Analyses were performed using SAS software (SAS for Windows version 8.2, SAS Institute, Cary, NC).

With the declining exponential approximation of life expectancy (DEALE) method (10,11), we calculated age-specific mortality rates based on expected survival for the median ages of our groups using Canada Census data (12). From our survival curves we obtained the 5-year survival probability estimates and calculated the compound average yearly mortality and disease specific excess mortality rates as the difference between compound and population annual mortality rates (11). Relative risk and power analyses were carried out with EpiInfo software (version 6.04 for DOS, CDC, Atlanta, GA).

Results

As of July 1999, 42 patients older than 60 years had been transplanted. All were included in our study. We were able to find a match for all of them.

Demographics and matching

Median age at the time of the transplant was 61.6 years (interquartile range 60.3–64.4) for the older group and 51.9 years (44.8–54.3) for the younger. There were 17 women and 25 men in the older cohort, and 15 women and 27 men in the matched younger cohort. Underlying diseases were pulmonary fibrosis in 15 pairs (36%), emphysema in 12 pairs (29%) α1-antitrypsin deficiency (ATD) in 7 pairs (17%), bronchiectasis in 3 pairs (7%), sarcoid, scleroderma, eosinophilic granuloma, pulmonary hypertension and pulmonary venooclusive disease 1 pair each (2.4%). Twenty-six pairs received bilateral lung transplants, eight pairs single lung grafts and eight more were not matched by type of transplant; the elderly received single whereas the younger double lung transplants. We were able to match 36 pairs by transplant era, as proposed, within ±2 years of the older patient's transplant date. Median time difference between elderly and matched younger controls was 162 days with 10th percentile of −620 days and 90th percentile of 551 days. Eighteen older patients were transplanted after his/her younger matched control. Eight controls could not be matched by era, there were four done before their respective cases and four after theirs. The time differences for these were up to 10 years. All pairs were matched by diagnosis; 26 (62%) of pairs were matched by all four criteria, 10 (24%) more by three criteria, five (12%) by two criteria and one (2%) pair by diagnosis only (Table 1).

Table 1.  Characteristics of matched pairs
Pair no.Underlying diseaseTime difference (days)Type of transplantGender
  1. Bolded: feature not matched.

1Pulmonary fibrosis  141DF
2Pulmonary fibrosis1084DF
3Pulmonary fibrosis  206DM
4Pulmonary fibrosis −224DM
5Pulmonary fibrosis −310DM
6Pulmonary fibrosis −419SM
7Pulmonary fibrosis  512DM
8Pulmonary fibrosis  280SF
9Pulmonary fibrosis−2768SF
10Pulmonary fibrosis  205SM
11Pulmonary fibrosis −236SM
12Pulmonary fibrosis −249SM
13Pulmonary fibrosis  408SM
14Pulmonary fibrosis−3564SM
15Pulmonary fibrosis4457SM
16Emphysema  −81SF
17Emphysema −157DF
18Emphysema −508DF
19COPD  537DF
20Emphysema  826DF
21Emphysema   60DM
22Emphysema  128DM
23Emphysema −309DM
24Emphysema  481DM
25Emphysema −618SF
26Emphysema −620SF
27Emphysema1563SF
28Pulmonary hypertension −320SF
29Emphysema (A1ATD) −318DF
30Emphysema (A1ATD) −167DM
31Emphysema (A1ATD) −296DM
32Emphysema (A1ATD) −342DM
33Emphysema (A1ATD) −388DM
34Emphysema (A1ATD) −394DF
35Emphysema (A1ATD)  469DM
36Bronchiechtasis  120DF
37Bronchiechtasis−2771DF
38Bronchiechtasis  551DM
39Sarcoidosis−1115SF
40Eosinophilci granuloma  −35SM
41Scleroderma  428DM
42Pulmonary veonooclusive Disease −557DM

Baseline co-morbidities

Four patients had been diagnosed with diabetes mellitus, two in each cohort and five had been diagnosed and were being treated for arterial hypertension, four of them in the older cohort of whom two were diabetic too.

Survival

Survival estimates 3 months after transplant were (mean ± standard error) 81 ± 8% for the older and 86 ± 5% for the younger group. At 1 year, survival had dropped to 60 ± 8% in the older cohort and it was 86 ± 5% in the younger group. At 2, 3, 4 and 5 years, estimated survival was 50 ± 8%, 50 ± 8%, 45 ± 8% and 37 ± 8% for the older, and 83 ± 6%, 71 ± 7%, 67 ± 7% and 57 ± 8% for the younger cohort, respectively, with a log-rank test p-value of 0.005 (Figure 1). Separate stratified analysis for mortality at 1 and 5 years by type of transplant and for era of transplant (matched vs. nonmatched), showed similar estimated odds ratios (OR)s among strata (data not shown). Results of the multivariable 1- and 5-year conditional logistic regression models are shown in Table 2.

Figure 1.

Survival curves for older and younger cohorts of lung transplant recipients (survival estimate ± standard error).

Table 2.  Conditional logistic regression OR estimates for mortality at 1 and 5 years
Multivariable model Factor95 % CIPR> ChiSq
EstimateLLUL
  1. CI = confidence interval; BMI = body mass index; LL = lower limit; UL = upper limit; CMV = cytomegalovirus; D+/R−= positive donor/negative recipient, FR = female recipient of female donor.

1-year survival 
 Age cohort older vs. younger5.681.70218.9480.0047
 CMV D+/R- vs. any other combination1.1280.2574.9410.8734
 SEX FR/FD vs. any other combination0.3020.0691.3260.1126
 Recipient BMI0.9840.941.0310.5024
 Donor weight0.9810.9491.0130.239
 Donor age1.0240.9871.0630.2123
 Univariable model 
 Age cohort elderly vs. young5.7271.85917.6450.0024
5-year survival 
 Age cohort older vs. younger3.3941.2099.5280.02
 SEX FR/FD vs. any other combination2.4690.6978.740.1612
 Cause of death (anxial/hypoxia)0.490.0356.8430.596
 Recipient BMI0.9920.9691.0160.5246
 Donor weight1.0170.9861.050.2864
 Donor age1.0020.9691.0350.9205
 Univariable model 
 Age cohort elderly vs. young2.9251.2057.0990.0177

Causes of death

Of the 11 early deaths in the older cohort, 10 were due to or associated with infection. Sepsis was found in 9 (82%) including: one pulmonary artery occlusion that led to lung necrosis and sepsis; one tracheal dehiscence with sepsis; one airway necrosis with sepsis; and one patient with a lymphoma who died of sepsis. Another patient died of CMV pneumonia. The only death not infection related was due to air embolism. Seven of those eleven patients had received bilateral lung grafts.

The six early deaths among the younger patients were due to infection in two (33%) cases, (one sepsis, one viral pneumonia); CAD in two (33%) and primary graft failure in two (33%). Five had received bilateral lung grafts. Late deaths in the older cohort were due to malignancy in six (30%), of which three were PTLD. Bronchiolitis Obliterans Syndrome (BOS) was the cause in five (25%). In the other nine (45%) it was due to infection in three, noncompliance in two (10%), and one each (5% each) of CAD, CVA, liver failure and multi-system organ failure. Twelve had received bilateral lung grafts. In the younger group, death was due to BOS in six (60%), sepsis in two (20%) and one each of CA and MI (10% each). Eight had received bilateral lung grafts (Table 3).

Table 3.  Mortality by period after transplant
 OlderYoungerRR95% CIp-Value1
  1. 1Chi-square or Fisher's exact test.

Early (≤6 months) 
Cause 
 Infection1022.73(0.87–8.59)0.028
 Other14 
 Total116 
Late (>6 month) 
 BO560.42(0.17–1.04)0.108
 CA613.00(0.42–21.65)0.371
 Other931.50(0.30–13.07)0.694
 Total2010 

Annual mortality rates

Statistics Canada publishes survival tables for males and females separately (13). We calculated a weighted average for the proportion of females to males, 0.51–0.49, respectively, for the 50–70 years older population. Based on population life expectancies of 28.7 and 19.4 years for those 52 and 62 years of age (median ages of our cohorts) the calculated annual age-specific mortality rates were 0.035 and 0.052, respectively. For the lung graft recipients, the compound annual mortality rates based on 5-year Kaplan–Meier survival probability estimates of 0.57 for the younger and 0.37 for the elderly were 0.112 and 0.199, respectively. The excess mortality rate for the younger was 0.112 − 0.035 = 0.077 and that for the older 0.199 − 0.052 = 0.147 such that the older/younger excess mortality rate ratio was 1.9 (see Appendix 2 for details).

Length of stay in hospital

The median number of days out of hospital within 30 days after transplant was not different between the groups. They were (median and interquartile range) 2.5 (0–10) for the older and 1.5 (0–11) for the younger with a Wilcoxon rank sum test p-value of 0.941.

Acute rejection, infectious episodes and other complications

Within 90 days, six and eight patients developed at least one episode of acute rejection among the older and younger cohorts, respectively. The relative risk (RR and 95% confidence limits) was 0.7 (0.28–1.98) with p-value of 0.771. There were 15 patients with at least one infectious episode within 90 days among the older cohort and nine among the younger. The RR was 1.7 (0.82–3.38) with p-value of 0.147. The power to declare this RR significant with an alpha of 0.05 was 24%. Assessment and treatment of complications was done using the same protocol for all patients (8). There were 13 older recipients with at least one ‘other complication’ within 90 days and 11 among the younger; two patients in each cohort had two complications and none had three or more during that period. Those included osteoporosis, arrhythmias, myocardial infarction (MI), coronary artery disease (CAD), convulsions, avascular hip necrosis, ruptured brain aneurysm, cancer, kidney failure and heart failure. The RR of developing at least one complication was 1.18 (0.6–2.3), with chi-squared p-value of 0.809. Among the younger there were patients with a cardiac complication (six with supra ventricular arrhythmias, two other acute myocardial infarction, both fatal). Among the older, there were six patients with a cardiac complication (all supra ventricular arrhythmia).

Six-minute walk test

Mean ± standard deviation pretransplant walking distances were 256 ± 118 and 293 ± 160 m for 40 older and 41 younger lung graft candidates, respectively. After transplantation (3 months) results were available for 31 of the 34 survivors in the older cohort and for 35 of the 36 in the younger cohort; (adjusted) mean walking distances after transplant were 512 ± 94 and 591 ± 93 m, respectively, and were significantly different from baseline (p < 0.001). However, there was no significant difference in improvement between cohorts; interaction p-value was 0.1536.

Discussion

We have shown that, although selected lung transplant candidates aged greater than 60 may undergo transplantation successfully, excess mortality rate doubles that of a younger matched cohort after taking into account their expected age related increased mortality rate. The increased mortality among the elderly transplant recipients appears to be the result of infectious causes in the early posttransplant period and malignancy is a frequent cause of death later. We found no significant differences in 90-day acute rejection rate, infection incidence rate, or length of stay in hospital. Six-minute walking distance after 3 months was not significantly longer for the younger. The increased RR for infections was not significant, but we did not reach enough power to test it.

At first glance it would appear that our study confirms previous reports of increased mortality by age in lung graft recipients. However, those reports have modeled outcome (death) as a function of age whereas this is the first study on the impact of lung transplant comparing outcome after adjusting for (removing) the expected age related mortality. ISHLT registry data show that as lung recipients' age increases so does the odds ratio of death at 1 and 5 years (7). It is well known that the mortality rate in the general population is a function of age (10). Hence, it should not surprise us to find higher mortality in older individuals. It is the competing risks (14) such as atherosclerosis (CAD, cerebrovascular accident, peripheral artery disease), osteoporosis, cognitive dysfunction, COPD, malignancy (lung, breast, prostate, colon, bladder) and complications of diverticular disease that account for most deaths among the elderly. Direct comparison of survival estimates between older and younger patients is therefore potentially misleading. Excess mortality, calculated by the DEALE method, permitted us to assess the effect of lung transplantation after taking into account the increased age-related mortality. The excess mortality ratio older/younger of 1.9 among lung graft recipients that we found clearly indicates that, despite the thorough assessment and selection process the older patients underwent in order to be considered candidates for transplantation, they still died at a higher rate than their younger counterparts. Though the DEALE is most accurate if survival follows an exponential curve, it is not a concern when dealing with relatively short survival times. Besides, it is much easier to understand and calculate as compared to other methods such as GAME or Markov models (15).

The increased mortality among the older cohort was observed mainly during the first 2 years after transplant. Within the first 6 months, infectious causes accounted for most deaths, whereas malignancy and BOS accounted for half of them thereafter. Our logistic regression models for mortality at 1 and 5 years (without adjusting for expected age related mortality) showed that the OR for age-group remained large and significant when adjusting for known factors for mortality. The proportions of patients with pretransplant co-morbidities in both groups was too small (<16%) to be analyzed in association with outcome. Older patients may be more susceptible to infections because of the immunosuppressive medication prescribed on top of their already decreased immunity (15). Although our results failed to reach power to declare the risk ratios statistically significant, we found that RR for incidence of infectious episodes was 1.7 (0.82–3.38) whereas that for acute rejection episodes within 90 days was 0.7 (0.28–1.98). These results point in the same direction as those reported previously for lung (16), kidney (17) and heart transplant (18,19).

Limitations

Generalizability of results may be limited in a single center study. It could be argued that the 60-year cut-off value is arbitrary as aging is a process that occurs over a lifetime and that there is no specific age at which one gets old. The 20-year annual age-specific mortality rate for men in Canada doubles at 50, triples at 61 and quadruples at 68; for women in Canada it doubles at 55, triples at 66 and quadruples at 72 (13). So, the slope of the survival curve starts increasing markedly at around 50 and increases quite obviously around 60 years. Levy et al. (20) studied liver transplant outcomes in older recipients. They did analyses shifting the age cut-off in an attempt to establish where to set the threshold for older recipients. They used the 60-year cut-off for their initial analysis, then they lowered it to 50 years and their results did change but when they shifted it up to 65, results did not change, thus supporting the 60-year cut-off as a good discriminator. Given the known association between infection with Epstein–Barr virus and malignancy (lymphoma) it would be important to know their status. We could not find Epstein–Barr virus serology results for most recipients and donors studied. The fact that we could not get a perfect match for all factors, especially type of transplant and era could have determined the results. However, analysis of mortality by those factors stratifying by matched versus non-matched showed that those factors were non-significant and with ORs close to unity. On the other hand, there are advantages to analyzing data from a single center: a 10-year lapse does not appear too long to see an effect by ‘era’. Similar or consistent protocols have been in place even though the program may have evolved in relation to technologies and treatments such as CMV therapy, diagnostic methods and prophylaxis protocols; protocols for donor selection organ procurement are more likely to be consistent. Similarly, in our program except for increasing age, the selection criteria for potential recipients have not undergone major changes. Our matching was not perfect but quite good. Thirty-six pairs (86%) were matched by at least three features. Having matched by underlying disease, type of transplant, era and gender should have made our groups comparable for survival and provided us with smaller CI for the calculated RR, and the 6 min walk. While this was a retrospective study, survival is an easily defined endpoint and data were collected prospectively and all patients have been followed up for at least 5 years.

While our results are not as optimistic as those in most publications on kidney, heart and liver, they should not discourage centers from transplanting selected older patients. However, they all should be aware of the increased risk. Chronological age by itself should not be a disqualifying criterion (5,6,21–23) as ‘physiological age’ (22) appears more important along with several other factors. However, objective definition of physiological age has remained elusive. Studies of lung transplant in older patients with expected life corrected for quality of life are required to assist in the selection of older candidates. Currently, older patients (seventh decade) should be considered for lung transplantation on a case-by-case basis only. There is no information, yet, on lung transplants in patients in the eighth decade of life. From the ISHLT registry and our results it is clear that patients aged greater than 60 have an increased risk of death and if a decision is made to assess such a patient for lung transplant all parties involved (transplant team, patient and family) should be well aware that the risk of death is increased and that information on outcome based on quality of life for the older patient is lacking. A thorough assessment, which includes evaluations to assess and/or rule out the aforementioned competing risks (including but not limited to cardiac echo and coronary angiography, carotid and peripheral vascular system imaging, endoscopy: colon and bladder) on top of the ‘routine’ assessment that younger patients undergo, is probably very important. The final decision must balance the magnitude of the increased risk in this population against potential benefits of transplant to both younger and older populations given organ availability. On a different note, this selection process, on the other hand, may invalidate statistical comparisons with younger patients, as it will create non-comparable populations.

The goal of organ transplantation is not simply to prolong life. Indeed, it may not prolong life at all. The goal is to offer patients with end organ damage additional years of quality life. Hence, quantifying survival time is not enough to form a comprehensive idea of the impact of lung transplant in the different age groups. Information on quality of life is lacking and necessary to assess the real impact of lung transplant. The trend toward increased infection and the clear difference in early mortality due to infections may suggest that tailoring immunosuppression and different drug regimens may be necessary for the older patients. Our study was not powered to detect such a difference. Hence making recommendations based on our results would not be appropriate. Further studies on age, immunologic system, and immunosuppression are needed. Until those studies are done we have decided to use a lower immunosuppressive drug regimen for individuals over 55 years of age in our program.

In conclusion, in our view lung transplant should be offered only to selected candidates older than 60 years. Candidates should be stringently scrutinized and be aware that with the current selection and postransplant management approach they still face higher mortality than the younger ones. Future studies including quality of life and objective assessment of physiological age should assist in the selection of those elderly candidates who may benefit from the procedure. Finely tuned immunosuppression regimens may improve the outcomes of those highly selected recipients.

Appendices

Appendix 1

Posttransplantation Surveillance Protocol (All Patients)

Chest radiographs were taken weekly for the first 8–12 weeks and then monthly for the first year after transplantation. FEV1 monitored daily with home electronic device and laboratory-based spirometry was done weekly/biweekly in the first month and then monthly. Bronchoscopy, with bronchoalveolar lavage specimens and transbronchial biopsies were performed at 2 and 6 weeks, then at 3, 6, 9, 12, 18 and 24 months, and yearly thereafter.

Immunosuppression Protocol

Five hundred mg IV of methylprednisolone were administered intraoperatively. Cyclosporine, azathioprine, and prednisone were started immediately postoperatively. Whole blood trough target cyclosporine levels were between 250 and 350 ng/mL over the initial 3 months and gradually reduced to between 150 and 250 ng/mL after 1 year. Azathioprine was started at 1 mg/kg/day but was reduced if liver enzyme abnormalities or leukopenia occurred. Prednisone was started at 0.5 mg/kg/day and reduced over 1 year to a dose of 15 mg on alternate days. Acute rejection episodes were managed with 1 g IV of methylprednisolone daily for 3 days followed by a regimen of prednisone tapering to baseline.

Antiviral Prophylaxis

CMV seropositive and/or received an organ from a seropositive donor were given 5 mg/kg BID IV ganciclovir for 14 days and then 5 mg/kg o.d. q M/W/F for 3 months. Hyperimmune globulin was also used in D+/R− patients before 1998. CMV-negative recipients of CMV-negative donors (D−/R−) were given oral acyclovir 400 mg TID for 3 months as herpes prophylaxis.

Appendix 2

DEALE (Declining Exponential Approximation to Life Expectancy)

Annual mortality rate(age)= 1/Life Expectancy(age)

Age specific life expectancies are obtained from published life tables

Excess mortality rate = Annualized disease specific mortality rate − Age specific annual mortality rate

Annualized disease specific mortality rates are calculated as the negative inverse of t times the natural log of survival at time t, where t is the time at which the survival is estimated using survival analysis (i.e. –1/t × ln t year survival estimate).

For example:

Life expectancy for a 52-year-old: 28.7 years

Annual mortality rate(52): 1/28.7 = 0.035

Five-year survival probability for patients with median age of 52.5 at the time of transplant: 0.57 (estimated by survival analysis of our data).

Annualized disease specific mortality rate: −1/5 ln(0.57) = 0.112

Excess mortality rate = 0.112 – 0.035 = 0.077.

Ancillary