Patient and technique survival on peritoneal dialysis in patients with failed renal allograft: A case control study

http://www.kidney-international.org & 2006 International Society of Nephrology Patient and technique survival on peritoneal dialysis in patients with failed renal allograft: A case control study S Mujais 1 and K Story 1 1 Renal Division, Baxter Healthcare Corporation, McGaw Park, Illinois, USA Failed renal allograft is becoming one of the most frequent causes of dialysis initiation in countries with developed transplant programs. The majority of patients initiate hemodialysis (HD) as their next renal replacement modality and concerns about the success of peritoneal dialysis (PD) in this patient population has been expressed. We evaluated patient and technique outcome in a cohort of 494 patients in the United States who initiated PD after a failed renal allograft in the years 2000 2003, and compared the outcomes to those of two case-matched groups: patients new to dialysis or patients transferred from HD who started PD during the same period. Patients starting PD after a failed allograft had patient survival and technique survival similar to case-matched controls. Transplantation was lower in patients with failed allograft than controls. The high success of PD in patients with failed allograft suggests that it is beneficial to utilize this modality more frequently in this patient group than current practice.. doi:10.1038/sj.ki.5001930 KEYWORDS: transplant failure; peritoneal dialysis Correspondence: S Mujais, Renal Division, Baxter Healthcare Corporation, 1620 Waukegan Rd, MPGR-A2E, McGaw Park, Illinois 60085-9815, USA. E-mail: salim_mujais@baxter.com Failed renal allograft is becoming a frequent cause of dialysis initiation in countries with developed transplant programs. 1 3 By the nature of the selection process for transplant eligibility, most patients with failed renal allograft are younger and have few co-morbidities compared to the overall population with end-stage renal disease (ESRD). 1 3 These same characteristics have been associated with a greater use of peritoneal dialysis (PD). 2,4 Gill et al. 2 reported that 16% of patients with failed transplant were on PD compared to 12% of the overall ESRD population. Nevertheless, the majority of patients with a failed allograft initiate hemodialysis (HD) as their next renal replacement modality. 2 The reasons very few patients initiate PD after failed allograft may be related to the perceived uncertainty about the success of PD in this patient population because of the paucity of outcome studies 5 7 and concerns about an increased risk of therapy-related infections. 6,8 The increased risk of infection is assumed to be sustained after allograft failure because of continuation of immunosuppressive therapy to preserve residual allograft function. 6,8 10 Although many studies have addressed the issue of post-allograft failure dialysis, 3,5 7,11 the vast majority have been single center studies with small sample sizes and data collection over many years, hence telescoping and obscuring changes in practice. 12 Populationbased studies have been few 1,13 15 and have typically not separated HD from PD, so it is uncertain whether the results are weighted by the overrepresentation of patients on HD. The present analysis was undertaken to provide insight into the status of patient and technique survival in the United States in contemporary patients who started PD after a failed allograft in comparison to matched concurrent PD patients who were new to renal replacement therapy (RRT) or were transferring to PD from HD. RESULTS Demographic characteristics The demographic profiles of the three case-matched groups (failed allograft, new to dialysis, and transfer from HD) are illustrated in Table 1. As expected from case-matching, patients in the three groups were identical in terms of age, gender distribution, prevalence of diabetes (25% with diabetes mellitus), and PD modality utilization (65% S133

Table 1 Demographic profile of the matched patient groups Failed transplant New to RRT Transfer from HD Number of patients 494 491 479 Male:Female 240:254 238:253 231:248 DM:Non-DM (%) 125:369 (25.3:74.7) 122:369 (24.85:75.15) 121:358 (25.3:74.7) APD:CAPD (%) 321:173 (65:35) 319:172 (65:35) 310:169 (65:35) Age (years) 39.68714.62 39.77714.60 40.47714.16 Cohort year 2000 123 121 114 2001 139 139 136 2002 116 116 114 2003 116 115 115 Center activity (number of patients) 0 4 67 67 64 5 9 95 93 88 10 19 171 171 170 20 29 98 97 95 X30 63 63 62 APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; DM, diabetes mellitus. automated PD (APD)). The average age of all groups was lower than that of the PD population in general, 4 likely owing to the selective process for transplantation. Similarly, the prevalence of diabetes was lower in these groups than in the ESRD population on PD overall. 4 The choice of PD modality (APD vs continuous ambulatory PD (CAPD)) matches the historical cohorts from which the groups were selected. 4 The distribution of the patients over the various annual incident cohorts was important to examine and was included in the matching to ascertain that secular trends in care practices did not inadvertently affect measured outcomes (Table 1). Similarly, dialysis center activity (number of patients started on PD in a calendar year) has been previously found to affect outcomes of interest, 4,16,17 hence its inclusion in the case-matching. Patient survival by reason for PD Patients with failed allograft and their case-matched controls had identical survival during the period of observation (Table 2). Survival in all groups was 90% in the first year and exceeded 80% in year 2. These survival rates are higher than those of the overall PD population, 4 reflecting the select nature of the groups in this study. In the Cox Regression analysis, survival in all three groups was adversely affected by age (hazard ratio (HR) ¼ 1.031, Po0.0001, a 3% increase in risk of death for each year older), and was better in nondiabetic patients (HR ¼ 0.417, Po0.0001, a 58% decrease in risk of death for non-diabetic compared to diabetic patients). Center size, center activity, gender, and PD submodality had no significant effect on patient survival in these groups. Patient status before initiating PD (failed transplant, new to dialysis, or transfer from HD) had no impact on patient survival. Although these groups had very low mortality, the causes of death were similar to those expected for patients with ESRD with cardiac causes accounting for 62% of the total. There was no difference in percent of patients that died owing to an infection among the three groups. Table 2 Patient survival 1-year 2-year 3-year 4-year Failed transplant 90.7171.42 85.0871.94 78.8372.69 75.8073.33 New to dialysis 91.1071.40 87.6471.76 81.9672.54 74.3773.97 Transfer from HD 89.8671.50 83.3472.06 76.8872.73 73.0373.38 HD, hemodialysis. Table 3 Technique survival (remaining on PD vs transferring to HD) 1-year 2-year 3-year 4-year Failed transplant 77.2171.97 64.2272.44 53.7272.93 47.7573.47 New to dialysis 82.7471.81 69.0372.45 57.3973.05 52.0873.57 Transfer from HD 74.7472.07 64.5572.43 54.9272.87 48.7273.37 HD, hemodialysis; PD, peritoneal dialysis. Technique success by reason for PD The rate of transfer from PD to HD was similar among patients in the three groups (Table 3). In multivariate analysis, center activity was the major predictor of technique success in all three groups (HR ¼ 0.84, Po0.0001, a 16% decrease in risk of technique failure for each step increase in center activity category as defined in Table 1). Age had a small effect on risk of transfer to HD (HR ¼ 1.001, Po0.005, a 1% increase in transfer to HD for each 10-year increase in age). Patient status before initiating PD (failed transplant, new to dialysis, or transfer from HD) had no impact on technique success. Reasons for technique failure A comparison of the major causes of transfer to HD for the groups is shown in Table 4. The highest causes of transfer to HD in the first year on PD for all patients were infection S134

Table 4 Proportional distribution of causes of transfer to HD Failed transplant (%) New to RRT (%) Transfer from HD (%) Catheter problem 18.4 18.9 19.4 Inadequate dialysis 20.4 16.5 15.7 Infection 27.2 25.2 20.5 Missing 1.9 0.5 0.5 Other 8.7 8.1 10.6 Other medical 13.6 12.8 12.4 Psychological 5.8 16.0 19.0 Fluid management 3.9 1.9 1.8 HD, hemodialysis; RRT, renal replacement therapy. Table 5 Percent of patients not receiving transplant after starting PD 1 year 2 year 3 year 4 year Failed transplant 92.3271.30 82.2572.17 74.2972.93 70.0473.65 New to dialysis 85.5171.70 75.4772.31 60.1573.21 52.5073.89 Transfer from HD 94.4371.15 85.6572.07 80.9472.61 75.5973.55 HD, hemodialysis; PD, peritoneal dialysis. (peritonitis and catheter-related) and catheter malfunction. The proportional contributions of various causes to transfer to HD showed a significant overall difference (Po0.007). This difference appeared to be driven mostly by lower psychosocial reasons for transfer in the failed transplant group. Transplant by reason for PD The rate of transplant was lower among patients with a failed allograft than among patients new to dialysis, but similar to that of patients transferred from HD (Table 5). In the Cox regression analysis, age (Po0.0001), gender (Po0.01), center size (Po0.05), and patient status before initiating PD (Po0.0001) were all significant parameters affecting the likelihood for transplant. The presence of diabetes or whether on APD or CAPD did not affect likelihood of transplant in the model. DISCUSSION The results of the present study can be summarized as follows: patients with failed allograft and case-matched controls had identical survival during the period of observation. The rate of transfer from PD to HD was similar among patients in the three groups. The rate of transplant was lower among patients with a failed allograft than casematched controls. The evaluation of outcome of patients with allograft failure returning to dialysis is critically dependent on the choice of the comparator group. Previous studies have utilized either no control group 1 or a variety of control groups defined as either all patients initiating PD 5,7,14 or any form of dialysis at the same center 3,15 or patients at the same center matched for age and diabetes, 6 or even actuarial data. 11 The profile of patients initiating dialysis after transplant failure appears in certain limited respects to be similar to that of the general incident ESRD population. Gill et al. 1,2 described similar levels of glomerular filtration rate, serum albumin, and hematocrit despite presumably being under the care of specialty physicians. Hence the use of incident patients as the comparator group appears to be reasonable as the delineated factors described by them have a demonstrated impact on outcomes 1,2 and need to be considered. We chose to use an individual matching approach to define the comparator group. 18,19 Subject level matching as used in this study allows for direct control of known measured confounders (age, diabetes, PD submodality, and center size) and for the control of unmeasured confounders that are difficult to measure or obtain (such as change in comorbidity during the course of the observation period as most co-morbidity data is typically obtained at a single time point at the initiation of dialysis). Matching for cohort year also allows for time comparability. The outcome of patients starting PD after renal allograft failure has been evaluated in a limited number of studies (Table 6). 5 7,14 Davies 5 prospectively compared the clinical outcomes of patients returning to dialysis with failed allografts to those of patients new to dialysis between 1989 and 2001. When controlled for age and co-morbidity, no significant differences in survival were seen between groups. Survival of patients with failed allograft commencing PD was not significantly different from that of patients new to dialysis. Similarly, technique failure was not different between the groups. Sasal et al. 6 examined the difference in outcome between 42 patients with failed renal transplantation and returning to PD compared with patients new to PD between 1989 and 1996. The two groups were matched for age and presence of diabetes. The failed-transplant group had a considerably worse outcome than those new to PD with higher mortality and technique failure. These findings are in contrast to those of another Canadian study involving all centers reporting to the Canadian Organ Replacement Registry, which found no difference in overall survival in patients with allograft failure returning to dialysis and patients new to dialysis. 20 Duman et al. 7 compared the outcomes of 34 patients with failed kidney allograft returning to PD with those of 82 PD patients who were new to dialysis. Patients with diabetes were excluded from the study. All failed-transplant patients were using immunosuppressive agents during the first 3 months of PD. Mortality rate and 1-, 3-, and 5-year patient and technique survival were similar in the two groups. de Jonge et al. 21 (abstract; de Jonge et al., Perit Dial Int 24(Suppl 2): S37, 2004) compared outcomes of patients returning to HD or PD after renal transplant failure with those of patients new to PD. The post-transplant PD group was similar to the group new to PD with regard to mortality, transfer to HD, technique failure, and retransplantation. More recently, Badve et al. 14 studied the effects of failed kidney transplantation on outcomes on PD in Australian and S135

Table 6 Studies of patient and technique survival on PD in patients with failed transplant Author Period Failed transplant Control Patient survival Technique success Davies 5 1989 2001 28 469 Similar Similar Sasal et al. 6 1989 1996 42 43 Lower Lower Duman et al. 7 NA 34 82 Similar Similar De Jonge et al. (abstract) NA 21 136 Similar Similar Badve et al. 14 1991 2004 309 13947 Similar Similar Present study 2000 2003 494 491/479 Similar Similar NA, not available; PD, peritoneal dialysis. New Zealand patients evaluated between 1991 and 2004. Patients commencing PD after renal allograft failure experienced outcomes comparable with failed native kidneys. The present study is the largest to date of the course of patients with failed allograft who are subsequently treated by PD. Patient survival and technique success in this group of patients were similar to those of matched patients. With a single exception, 6 our results are consistent with the preponderance of data in the literature regarding patient survival and technique success 5,7,14 (abstract; de Jonge et al., Perit Dial Int 24(Suppl 2): S37, 2004) (Table 6). Because of the practice of continuing immunosuppression after allograft failure, sometimes for very long periods, concern has been raised about a predisposition to infectious complications. The preponderance of published data, however, suggests that peritonitis rates are not different between patients on PD after a failed allograft and other PD patients 6,7,14,22,23 (abstract; de Jonge et al., Perit Dial Int 24(Suppl 2): S37, 2004), (abstract; Manga et al., Perit Dial Int 22: 143, 2002). In the present study, the rate of technique failure owing to infection was not different between the two groups. Andrews et al., 8 however, retrospectively examined the incidence and outcome of peritonitis in 39 immunosuppressed and 146 non-immunosuppressed patients treated with CAPD. Immunosuppressed patients had more episodes of peritonitis, required more frequent hospital admissions, and required more surgeries to remove infected catheters compared with non-immunosuppressed patients. The immunosuppressed group, however, consisted of patients receiving immunosuppressive drugs for a variety of ailments including systemic lupus erythematosus, vasculitis, glomerulonephritis, and previous transplant. The mixed composition of the group is a strong confounding factor and the results, therefore, cannot be accepted as characterizing postallograft failure patients. Treatment-related characteristics of patients with failed transplant initiating PD have been minimally studied. A predisposition to higher peritoneal transport profile in patients on PD as a result of allograft failures has been reported by Hebert et al. 24 in a study of 19 patients. Eight (42%) of these patients were determined to be high transporters, a value higher than generally observed in a PD population. 25 A normal peritoneal equilibration test distribution, however, has been observed in the reports by Davies, 5 de Jonge et al. (abstract; de Jonge et al., Perit Dial Int 24(Suppl 2): S37, 2004), Manga et al. (abstract; Manga et al., Perit Dial Int 22: 143, 2002), Duman et al., 7 and Badve et al., 14 suggesting that the report by Hebert et al. 24 may represent an aberrant occurrence owing to small sample size. Contradictory findings have also been reported for the rate of decline of residual renal function. Davies 5 has suggested that patients on PD after a failed transplant have a faster rate of decline of residual renal function than other patients on PD. Schiffl et al. 26 observed similar findings in eight patients with renal transplant failure who were returning to CAPD compared to 16 matched (demographic and renal characteristics) CAPD patients who had never received a transplant. Considering the complexity of the factors that may influence residual renal function decline, including the nature of the underlying renal disease in the native organs and the transplant, the withdrawal of immunosuppression, and the effects of medications, it would be premature to conclude that patients with allograft failure are characterized uniformly by this profile. de Jonge et al. (abstract; de Jonge et al., Perit Dial Int 24(Suppl 2): S37, 2004) were unable to reproduce the findings in a population that had a more usual distribution of peritoneal transport profile. In conclusion, the high success of PD in patients with failed allograft suggests that it is beneficial to utilize this modality more frequently in this patient group than current practice. MATERIALS AND METHODS The present analysis is based on data from 494 patients who started PD after a failed renal allograft. Patients belonged to cohorts of patients in the United States that started PD between the years 2000 and 2003 and were followed until June 2005. Information about these patients is tracked in the Baxter Healthcare Corporation On- Call TM system. 4 This system, which is compliant with the Health Insurance Portability and Accountability Act (HIPAA), includes demographic data; treatment history for renal disease, including reason for starting PD (new to dialysis, transfer from HD or failed allograft); causes for any changes in treatment modality (transfer from PD to HD or transplant); patient outcome; and general information about PD therapy such as submodality use (CAPD or APD) and dialysis prescription. As this information is gathered as a component of the Baxter dialysis supplies home delivery system, it has the distinct advantage of reflecting actual verified rather than reported conditions. De-identified information from this system forms the basis of this analysis. 4 S136

Patients starting PD because of a failed renal allograft were matched 1:1 to patients new to RRT or patients transferring from HD and initiating PD in the same time period. Matching was carried out by age (71 year), gender, presence of diabetes mellitus, PD submodality (CAPD vs APD), cohort year, and dialysis center characteristics (overall PD patients census and activity as reflected in the number of patients initiating PD in a calendar year). The majority of these factors have been identified as predictors of outcome on PD. 4 To evaluate technique success and patient disposition, a detailed list of possible causes of transfer from PD is maintained within the system to allow for categorization of disposition (death, transplant, or transfer to HD) and primary causes of modality transfer. All events entered in the system are dated, allowing for determination of rates of occurrences. For this analysis, causes of transfer from PD were grouped in the following broad categories: infection (peritonitis and catheter-related infection), catheter problem, inadequate dialysis, ultrafiltration failure/fluid management issues, psychosocial causes (psychological and social/learning), and other medical causes. The life-table method was used to compute estimates of actuarial patient and technique survival. A stepwise Cox Regression analysis was used to calculate estimates of HRs and determine which parameters were related to patient and technique survival, and transplantation rate. Data were censored at the following events: switch to HD, transplantation, death, loss to follow-up, and recovery of native renal function. The exceptions were death for patient survival calculation and transfer to HD for the technique survival calculation. ACKNOWLEDGMENTS We thank Ms Rona McGreevy for expert editorial help in the preparation of this paper. REFERENCES 1. Gill JS, Abichandani R, Kausz AT, Pereira BJ. Mortality after kidney transplant failure: the impact of non-immunologic factors. Kidney Int 2002; 62: 1875 1883. 2. Gill JS, Abichandani R, Khan S et al. Opportunities to improve the care of patients with kidney transplant failure. Kidney Int 2002; 61: 2193 2200. 3. de Francisco AL, Arias M. Survival markers in dialysis. Nefrologia 2001; 21: 137 149. 4. Guo A, Mujais S. Patient and technique survival on peritoneal dialysis in the United States: evaluation in large incident cohorts. Kidney Int Suppl 2003; 64(Suppl 88): S3 S12. 5. Davies SJ. Peritoneal dialysis in the patient with a failing renal allograft. Perit Dial Int 2001; 21(Suppl 3): S280 S284. 6. Sasal J, Naimark D, Klassen J et al. Late renal transplant failure: an adverse prognostic factor at initiation of peritoneal dialysis. Perit Dial Int 2001; 21: 405 410. 7. Duman S, Asci G, Toz H et al. Patients with failed renal transplant may be suitable for peritoneal dialysis. Int Urol Nephrol 2004; 36: 249 252. 8. Andrews PA, Warr KJ, Hicks JA, Cameron JS. Impaired outcome of continuous ambulatory peritoneal dialysis in immunosuppressed patients. Nephrol Dial Transplant 1996; 11: 1104 1108. 9. Smak Gregoor PJ, Zietse R, van Saase JL et al. Immunosuppression should be stopped in patients with renal allograft failure. Clin Transplant 2001; 15: 397 401. 10. Jassal SV, Lok CE, Walele A, Bargman JM. Continued transplant immunosuppression may prolong survival after return to peritoneal dialysis: results of a decision analysis. Am J Kidney Dis 2002; 40: 178 183. 11. Howard RJ, Reed AI, Van Der Werf WJ et al. What happens to renal transplant recipients who lose their grafts? Am J Kidney Dis 2001; 38: 31 35. 12. Meier-Kriesche HU, Ojo AO, Port FK et al. Survival improvement among patients with end-stage renal disease: trends over time for transplant recipients and wait-listed patients. J Am Soc Nephrol 2001; 12: 1293 1296. 13. Meier-Kriesche HU, Kaplan B. Death after graft loss: a novel endpoint for renal transplantation. Transplant Proc 2001; 33: 3405 3406. 14. Badve SV, Hawley CM, McDonald SP et al. Effect of previously failed kidney transplantation on peritoneal dialysis outcomes in the Australian and New Zealand patient populations. Nephrol Dial Transplant 2006; 21: 776 783. 15. Rao PS, Schaubel DE, Saran R. Impact of graft failure on patient survival on dialysis: a comparison of transplant-naive and post-graft failure mortality rates. Nephrol Dial Transplant 2005; 20: 387 391. 16. Huisman RM, Nieuwenhuizen MG, Th de Charro F. Patient-related and centre-related factors influencing technique survival of peritoneal dialysis in The Netherlands. Nephrol Dial Transplant 2002; 17: 1655 1660. 17. Schaubel DE, Blake PG, Fenton SS. Effect of renal center characteristics on mortality and technique failure on peritoneal dialysis. Kidney Int 2001; 60: 1517 1524. 18. Etminan M. Pharmacoepidemiology II: the nested case control study a novel approach in pharmacoepidemiologic research. Pharmacotherapy 2004; 24: 1105 1109. 19. Costanza MC. Matching. Prev Med 1995; 24: 425 433. 20. Rao PS, Schaubel DE, Saran R. Impact of graft failure on patient survival on dialysis: a comparison of transplant-naive and post-graft failure mortality rates. Nephrol Dial Transplant 2005; 20: 381 391. 21. de Jonge H, Bammens B, Lemahieu W et al. Comparison of peritoneal dialysis and haemodialysis after renal transplant failure. Nephrol Dial Transplant %R 10.1093/ndt/gfl010 2006; 21: 1669 1674. 22. Gregoor PJ, Kramer P, Weimar W, van Saase JL. Infections after renal allograft failure in patients with or without low-dose maintenance immunosuppression. Transplantation 1997; 63: 1528 1530. 23. Krishnan M, Lok CE, Jassal SV. Epidemiology and demographic aspects of treated end-stage renal disease in the elderly. Semin Dial 2002; 15: 79 83. 24. Hebert LA, Wilmer WA, Falkenhain ME et al. Renoprotection: one or many therapies? Kidney International 2001; 59: 1211 1226. 25. Mujais S, Vonesh E. Profiling of peritoneal ultrafiltration. Kidney Int Suppl 2002; S81: S17 S22. 26. Schiffl H, Mucke C, Lang SM. Rapid decline of residual renal function in patients with late renal transplant failure who are re-treated with CAPD. Perit Dial Int 2003; 23: 398 400. S137