Renal transplantation is the treatment of choice in end-stage renal disease (ESRD).1 However, despite a significant improvement in short- and long-term graft survival, related in part to the currently available immunosuppressive agents,2 over 30% of renal transplant (RTx) recipients return to dialysis within the first 5 years. Moreover, patients with graft failure have become a growing proportion of incident and prevalent dialysis patients worldwide.3 The rate of patients returning to hemodialysis (HD) after renal graft failure is approximately 12% per year in our dialysis units.
There is increasing evidence that late nephrological referral of ESRD patients is associated with increased morbidity and mortality when compared to an early referral and current data suggest that between 20% and 57% of patients in Europe and the United States are not referred to the nephrologist on time.4-11 A possible reason for these results is that early referral patients are started on dialysis with a functioning vascular access (VA) fashioned weeks before, after careful planning.12 At the same time in incident HD patients, the use of catheters, whether tunneled or not, is associated with a worse prognosis when compared with an arteriovenous fistula.12
The clear majority of RTx patients worldwide is managed by nephrologists and hence is part of an early referral group. Hence, the relative role of VA type in the morbidity and mortality of this population can be dissected out from other prognostic factors. Therefore, our objective was to determine whether there is a relationship between the type of VA and mortality in patients returning to HD after allograft failure.
MATERIALS AND METHODS
We carried out a retrospective, observational, cohort study of individuals returning to HD with a failing RTx per the type of VA at the time of initiation of HD. The study had approval by our institutional Institutional Review Board.
We collected data in 138 patients who started HD with a failing RTx at our center between January 1995 and December 2014. All patients were followed up by nephrologists at our center where we perform RTxs as well as we run a chronic dialysis clinic, therefore they belong to an early referral group. We excluded patients who did not complete 1 month of dialysis treatment and those entered in our Continuous Ambulatory Peritoneal Dialysis program. All patients were placed on a thrice weekly regimen, using polysulfone dialysers, a bicarbonate dialysis bath, controlled ultrafiltration, and with a target equilibrated KT/V above 1.2.
Sixty-one (44.2%) patients received a kidney from a living donor and 55.8% from a deceased donor. 34.3% received induction with polyclonal antibodies (thymoglobulin), and 86.2% were immunosuppressed with a calcineurin inhibitor-based (cyclosporine or tacrolimus) triple therapy regimen. 67.4% presented at least 1 episode of acute rejection and 95 (68. 8%) of 138 patients lost their kidney to biopsy demonstrated chronic allograft nephropathy.
The study population was divided into 2 groups per the type of VA at the initiation of HD. The programmed VA (PVA) group included all patients with either an arteriovenous fistula or a polytetrafluorethylene graft (PTFE graft). The unprogrammed VA (UPVA) included those patients started on dialysis using a dual lumen temporary catheter either tunneled (silicone dual lumen PermCath from Quinton or Tesio from Medcomp) or nontunneled (polyurethane catheters from Quinton, Arrow, or Medcomp) The decision on the type of the VA at entry was made by the attending nephrologist.
We collected demographic data (age, sex, VA at initiation of HD, original cause of CKD in the native kidneys, time on HD before the transplant, type of donor, graft survival, time on HD after the RTx) and relevant laboratory data just before starting dialysis (creatinine, estimated glomerular filtration rate [eGFR] by Modification of Diet in Renal Disease formula, hemoglobin (Hb), serum albumin, intact parathyroid hormone, calcium-phosphate product, hepatitis B and C status). At the start of HD, the risk of death was stratified into 3 categories using Khan comorbidity index which has been previously validated for HD populations and is based on age, presence or absence of diabetes mellitus (DM) and organ-specific diseases.13 The low-risk group includes patients younger than 70 years, without DM or comorbidities. The moderate-risk group includes patients between 70 and 80 years of age or patients with DM or those with significant cardiovascular conditions. The high-risk group includes patients older than 80 years, or older than 70 years but with DM, or younger than 70 years with DM and significant cardiovascular disease or involvement of 2 or more organs or cancer (excluding nonmelanoma skin cancer).
The initiation of HD therapy was considered the beginning of treatment. Patients were recruited until December 31, 2014. The study was closed at the end of the observation period which lasted until December 31, 2016. Patients were censored if and when they received a new transplant, were switched to Continuous Ambulatory Peritoneal Dialysis, or were lost to follow-up. Death from any cause was considered the main endpoint.
Continuous variables were compared using a t test and categorical variables using a χ2 test. Nonparametric variables were compared using the Mann-Whitney U test. The Kaplan-Meier method was used to compare the survival curves of the 2 groups and the log rank test to determine their differences.
In the univariate analysis, we included age, sex, time on HD before the RTx, type of donor, immunosuppression treatment, graft survival, time on HD after transplant failure, type of VA at the start of HD, laboratory data, hepatitis B and C status, and Khan risk index. All variables found to be statistically significant (P <0.05) were included in 2 Cox multivariate models. In model 1, all potential predictors were included: age, DM, time on HD before Tx, induction with thymoglobulin and anti-IL2, Hb, donor type, and VA type at dialysis initiation; model 2 included time on HD before Tx, Hb, induction, donor type, VA type at dialysis initiation and Khan comorbidity index but did not include age and diabetes, given that they are part of Khan comorbidity index itself. In none of the 2 models, we included variables that did not predict mortality in the univariate model, such as, serum creatinine, serum albumin, intact parathyroid hormone, and calcium-phosphate product. Associations with a P value less than 0.05 were considered statistically significant. All results were expressed as mean, median, interquartile range and SD as appropriate. Statistical analysis was performed using SPSS 10.0 for Windows (SPSS Inc, Chicago, IL).
Clinical and Demographic Data
Data are shown in Table 1. Of the 138 patients, 85 (61.6%) subjects began HD with a PVA, of which 37 (43%) was preexisting access from before the transplant, the remaining 53 (38.4%) began HD with a UPVA.
At the time of dialysis initiation, all patients were on steroids as part of their maintenance immunosuppression associated with tacrolimus and micophenolic acid in 54 (39.1%), cyclosporine and azathioprine in 38 (27.5%), cyclosporine and micophenolic acid in 11 (8%), azathioprine alone 15 (10.9%), micophenolic acid alone in 9 (6.5%), and sirolimus and micophenolic acid in 11 (8%). There were no significant differences between groups on this regard. Calcineurin inhibitors were stopped at dialysis initiation, the rest of immunosuppression was tapered over a 3-month period.
Fifty-six (40.6%) patients died during follow-up and 32 (23.2%) were retransplanted. The survival rates for the total cohort at 12, 24, and 48 months were 85.4%, 78.5%, and 71%, respectively.
We observed a statistically significant increase in mortality in the UPVA group, and this difference was more pronounced during the first and second year of follow-up. In these 2 years, 7 patients died in the PVA group versus 22 in the UPVA group (Figure 1). In the PVA group, there were 4.5 deaths per 100 patient's years at risk versus 18 deaths per 100 patient's years at risk in UPVA group. In the PVA group, there were no differences in mortality whether the VA was preexisting before transplantation or performed before dialysis initiation. Overall survival was significantly better in the PVA group (log rank test, 31.9; P < 0.0001. The same association was observed adjusting the data for the Khan comorbidity index (Figure 2). Likewise, Table 2 depicts the mortality observed in the different Khan comorbidity index staging categories according to the type of VA. Mortality was higher in patients with UPVA with a high Khan comorbidity index at 12 and 24 months.
Causes of Death
Cardiovascular causes were the most frequent, including sudden death in 36.4%, infectious causes (27.3%), neoplasms (12.7%), and undetermined (12.7%). Of the 15 (27.3%) patients who died from infectious diseases, 12 belonged to the UPVA and 3 to the PVA group. This difference was not statistically significant, perhaps due to small number of patients. No other differences in the causes of death were evident when comparing the PVA versus UPVA. We also found no differences when we analyzed causes of death at 12 and 24 months (Table 3).
In the univariate analysis (Table 4), age, (hazards ratio [HR], 1.069; 95% confidence interval [CI], 1.04-1.09), DM (HR, 2.290; 95% CI, 1.14-6.33), time on HD before Tx (HR, 1.007; 95% CI, 1.00-1.01), donor type (living donor was considered the reference group) (HR, 5.377; 95% CI, 2.66-10.83), induction with thymoglobulin and anti-IL2 (HR, 2.134; 95% CI, 1.19-3.81), Hb (HR, 1.174; 95% CI, 1.03-1.33), VA at start of HD (PVA was the reference group) (HR, 4.587; 95% CI, 2.58-8.15), Khan comorbidity index moderate versus low risk (HR, 5.116; 95% CI, 2.48-10.54) and high versus low risk (HR, 12.270; 95% CI, 5.99-25.13) showed a strong association with death.
Cox Multivariate Analysis
All the variables that showed significance in the univariate analysis were included in multivariate analysis. Results of the 2 multivariate models developed are described in Table 5. In summary, in model 1, we observed that age (HR, 1.050; 95% CI, 1.02-1.08; P < 0.001) and VA type (HR, 4.261; 95% CI, 2.24-8.00; P < 0.0001) were strongly associated with mortality. Using this second model, VA type (HR, 5.904; 95% CI, 2.83-12.31; P < 0.0001) continued to be strongly associated with death despite having been adjusted for Khan comorbidity index.
Our data indicate that subjects with a failing RTx, followed up by nephrologists (early referral) and reentering HD with a nonprogrammed VA show significantly greater mortality than those with a VA despite adjusting for Khan risk index and that most of the increased mortality is observed in the first 2 years of follow up.
At present, there is a worldwide increase in the number of patients with failing renal grafts entering renal replacement programs, and it is thus appropriate to consider the clinical status of these patients when they reenter HD programs.14 In our study, we observed that some of the analyzed variables, such as eGFR, 10 mL/min; Hb, 9.2 g/dL; and serum albumin, 3.6 g/L, fell below the recommended Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines, although they were like the values reported by Arias et al14 in their study, patients with chronic allograft nephropathy were restarted on dialysis with a creatinine clearance of 9 mL/min and a Hb of 8.9 g/dL. Also, these variables were not significantly different between groups, and renal function was similar at the beginning of replacement therapy which excludes early versus late dialysis initiation as an explanation of our results.
Our overall mortality rate was 14.6% at 1 year, lower than that reported for the general incident dialysis population by the Argentine Registry of Chronic Dialysis patients for the year 2007 (17.55%),15 even though previous studies suggest that patients with failing grafts have an increased mortality compared to the general HD population. For example, Gill et al16 determined the number of deaths between 1995 and 2003 in a single group of patients that were initially on dialysis and placed on the transplant waiting list, underwent renal transplantation, and then reentered HD due to a failing graft. They observed that the lowest mortality rate was observed during the period with a functioning graft and the highest after reentering dialysis (17.9/100 patients/year; 95% CI, 15.7-20.3). They also observed that diabetic and older patients were at increased risk at all times. Likewise, Rao et al17 analyzed the death rate of 174 436 patients included in the Scientific Registry of Transplant Recipients database and compared those reentering dialysis with a failing graft with the group on the transplant waiting list for the first transplant during the period between 1995 and 2004 and the risk of death from any cause was 78% higher in the group with the failing graft.
In our cohort and in agreement with other studies, the main causes of death were cardiovascular (36.4%) and infectious (27.3%). Patients with UPVA had more infections than those with PVA. Although this difference did not reach statistical significance (probably due to a low number of patients), we hypothesized that this could be an explanation for the survival difference observed between the groups. Gill et al's18 data based on the analysis of 4741 patients with a failing graft showed that cardiovascular causes were responsible for 36% of deaths followed by infections (17%). In a Cox regression analysis of their data, they found that older age, female sex, white race, diabetes as a cause of ESRD, congestive heart failure, peripheral vascular disease, illegal drug use, tobacco smoking, and a high GFR at the start of HD were predictors of death from any cause. Kaplan and Meier-Kriesche19 looked at the risk factors for death after a failed RTx in 78 564 patients taken from the United States Renal Data System database. Their data showed that the death rate was 3 times higher than that observed in patients with a functioning graft (94.2 vs 28.1 patients/year); that the adjusted annual cardiovascular death rate was 7 times higher after graft loss when compared with cardiovascular deaths with a functioning transplant (43.1 vs 6.9 per 100 patients/year); and that the death rate due to infectious causes was 4 times higher after restarting HD when compared with individuals with a functioning graft (16.3 vs 3.7 per 100 patients/year).
We further show that patients using catheters have a greater mortality risk than those with a programmed VA. (Log Rank p value < 0.0001). Previous studies in incident HD patients showed that catheter use is associated with an increased rate of death from any cause, but none of these looked specifically at the population restarting dialysis after a failed renal graft.
Lorenzo et al12 reported that in 538 patients a nonprogrammed initiation of dialysis (late referral) and the use of temporary catheters were independently associated with greater mortality, especially during the 1st year. Similarly, Pastan et al20 showed that in a population of 7497 patients, the use of a catheter whether tunneled or not was associated with greater all-cause mortality when compared with an AV fistula (odds ratio [OR], 1.4, 95% CI, 1.1-1.9) or a PTFE graft (OR, 1.3; 95% CI, 1.1-1.6). This increased mortality was seen at the initiation of dialysis as well as in the long-term follow-up.21 In our study, 53 (38.4%) patients entered dialysis with a catheter, of which 18 (34%) were switched to an AV fistula over time. This could explain the increased adjusted early mortality of this group that is not seen in the long-term follow-up.
These studies also showed that incident patients starting HD in a programmed way had higher hematocrit and serum albumin values, whereas we could not establish any difference in Hb, serum albumin, calcium-phosphate product, and creatinine values between the 2 groups.
Several investigators have observed that early referral to a nephrologist is advantageous for patient survival.4-12 Patients with a failing graft, by definition an early referral group, should show low mortality rates even with the use of a catheter but this was not the case in our study. In agreement with other study,18 we did not find that use of polyclonal antibodies, acute rejection, or transplant survival had an impact on all-cause mortality. On the other hand, in univariate and multivariate analyses (model 2), we found that a deceased donor source was independently associated with greater mortality (P < 0.0001; OR, 5.90; 95% CI, 2.8-12.3). In a literature search of English language journals, we could not find a similar result. A plausible hypothesis for this finding is that patients receiving a living donor kidney spend a shorter time on dialysis before the transplant, receive less immunosuppression, and are retransplanted sooner after their graft fails.
Our study has some limitations including the retrospective design, relatively small sample size and the fact that we were not able to study whether the number of admissions to our hospital were different between the 2 groups. Lastly, we found that many patients that were restarted on HD using a catheter had a high (38.5%) or an intermediate (17.3%) value for Khan comorbidity index, which might indicate that patients in this group were somehow sicker and for this reason were selected for catheter placement based on the a priori assumption that they would have a short survival. Nevertheless, after adjusting for Khan comorbidity index in model 2, VA type continued to be strongly associated with death.
Likewise, in a simpler analysis, UPVA predicted increasing mortality for each category staging of the Khan comorbidity index.
Another limitation is that the evaluation of Khan comorbidity index has been validated for incident HD patients and not for their subgroup of patients returning to HD after renal graft failure.
Finally, the lack of a control group of failing native kidney patients appears as another limitation for our study. Nevertheless, studies in this population have shown similar results with patients entering dialysis with a catheter having the worst prognosis.12,20-22 In our own nonpublished data, over 322 incidental HD patients, those who started early (MDRD > 7 mL/min) with a catheter had worse survival than those entering later (MDRD < 7 mL/min) with a fistula or AV graft.23
In this retrospective study, patients entering our 2 institutional HD programs after a failed RTx using a catheter as opposed to a previously fashioned permanent access showed a greater all-cause mortality.
1. Marcen R, Pascual J, Tato AM, et al. Renal transplant recipient outcome after losing the first graft. Transplant Proc
2. Hariharan S, Johnson CP, Bresnahan BA, et al. Improved graft survival after renal transplantation in the United States, 1988 to 1996. N Engl J Med
3. U.S. Renal Data System. 1997 Annual data report. Published 1997.
4. Jungers P, Zingraff J, Albouze G, et al. Late referral to maintenance dialysis: detrimental consequences. Nephrol Dial Transplant
5. Schmidt RJ, Domico JR, Sorkin MI, et al. Early referral and its impact on emergent first dialyses, health care costs, and outcome. Am J Kidney Dis
6. Roubicek C, Brunet P, Huiart L, et al. Timing of nephrology referral: influence on mortality and morbidity. Am J Kidney Dis
7. Metcalfe W, Khan IH, Prescott GJ, et al. Can we improve early mortality in patients receiving renal replacement therapy? Kidney Int
8. Jungers P, Massy ZA, Nguyen-Khoa T, et al. Longer duration of predialysis nephrological care is associated with improved long-term survival of dialysis patients. Nephrol Dial Transplant
9. Kinchen KS, Sadler J, Fink N. The timing of specialist evaluation in chronic kidney disease and mortality. Ann Inter Med
10. Winkelmayer WC, Owen WF Jr, Levin R, et al. A propensity analysis of late versus early nephrologist referral and mortality on dialysis. J Am Soc Nephrol
11. Stack AG. Impact of timing of nephrology referral and pre-ESRD care on mortality risk among new ESRD patients in the United States. Am J Kidney Dis
12. Lorenzo V, Martn M, Rufino M, et al. Predialysis nephrologic care and a functioning arteriovenous fistula at entry are associated with better survival in incident hemodialysis patients: an observational cohort study. Am J Kidney Dis
13. Khan IH, Catto GR, Edward N, et al. Influence of coexisting disease on survival on renal-replacement therapy. Lancet
14. Arias M, Escallada R, de Francisco AL, et al. Return to dialysis after renal transplantation. Which would be the best way? Kidney Int Suppl
15. Marinovich S, Lavorato C, Celia E, et al. Registro argentino de diálisis crónica SAN-INCUCAI 2007. Nefrología Argentina
16. Gill JS, Rose C, Pereira BJ, et al. The importance of transitions between dialysis and transplantation in the care of end-stage renal disease patients. Kidney Int
17. Rao PS, Schaubel DE, Jia X, et al. Survival on dialysis post-kidney transplant failure: results from the Scientific Registry of Transplant Recipients. Am J Kidney Dis
18. Gill JS, Abichandani R, Kausz AT, et al. Mortality after kidney transplant failure: the impact of non-immunologic factors. Kidney Int
19. Kaplan B, Meier-Kriesche HU. Death after graft loss: an important late study endpoint in kidney transplantation. Am J Transplant
20. Kinchen KS, Sadler J, Fink N, et al. The timing of specialist evaluation in chronic kidney disease and mortality. Ann Intern Med
21. Gruss E, Portolés J, Tato A, et al. Clinical and economic repercussions of the use of tunneled haemodialysis catheters in a health area. Nefrologia
22. Ozeki T, Shimizu H, Fujita Y, et al. The type of vascular access and the incidence of mortality in Japanese dialysis patients. Intern Med
23. Wasiuchnik VGP, Laham G, Abascal JM, et al. Early starting hemodialysis: co-morbidities account for the lack of survival benefit independently of the vascular access type. Hemodial Int