RENAL REPLACEMENT THERAPY AND ORTHOTOPIC LIVER TRANSPLANTATION: THE ROLE OF CONTINUOUS VENO-VENOUS HEMODIALYSIS ¹

Abstract

Background. 

The need for renal replacement therapy (RRT) either beforeor after orthotopic liver transplant (OLTX) has been reported to be a poorprognostic indicator for survival. Use of continuous veno-venous hemodialysis(CVVHD) for RRT has been reported in three series of OLTX patients with high90-day mortality rates of 57–60%. We have examined our patientpopulation to determine the effect of necessity and type of RRT on patientsurvival afterOLTX.

Methods. 

We analyzed 1535 OLTX that were performed at ourinstitution from 1985 through 1999, 1037 from 1985 to 1995 (period I) and 498from 1996 to 1999 (period II). Combined liver-kidney transplants were excludedfrom analysis. Hospital dialysis unit records and a prospectively maintaineddatabase on all OLTX patients served as the source of data. Patients wereclassified into groups defined on whether or not they received RRT, when theyreceived RRT, and the type of RRT. Groups were compared for preoperativeintensive care unit status, time on the waiting list, laboratory variables,90-day postoperative mortality, 1-year patient survival, and absolutesurvival.

Results. 

Use of RRT increased from 8.29% in period I to 12.45% inperiod II, along with increased median waiting times. In period I, patientsreceiving preoperative RRT had a 90-day mortality (0%) and a 1-year survival(89.5%) almost identical to those patients who never required RRT (1.7% and90.6%). Patients who developed acute renal failure postoperatively requiringRRT, however, had a 90-day mortality of 28.6% and a 1-year survival of 55%. Inperiod II, patients requiring RRT had a 90-day mortality of 39.7% and a 1-yearactuarial survival of 54.5% compared with 6.9% and 88.6% in patients neverrequiring RRT. Patients treated with CVVHD had a 90-day mortality of 42%compared with 25% in patients treated with hemodialysis alone. However,patients receiving CVVHD both pre- and postoperatively had a 90-day mortalityof 27.7% vs. 50% in those patients who only received CVVHD postoperatively.Patients who developed acute renal failure postoperatively, which requiredRRT, regardless of therapy, had a 1-year survival of only 41.0% compared witha 1-year survival of 73.6% in those patients started on RRT preoperatively, P =0.03.

Conclusions. 

The need for RRT has increased along with waiting time inOLTX patients. Patients developing the need for RRT postoperatively have anincreased 90-day mortality and lower 1-year survival with the highest beingpresent in patients receiving CVVHD, which was started postoperatively. Thesefindings may reflect a trend toward a sicker population awaiting OLTX andemphasize the negative impact of renal failure on survival afterOLTX.

INTRODUCTION

Patientswho require renal replacement therapy (RRT) before or after orthotopic livertransplant (OLTX) have been reported to have a higher mortality compared withthose not requiring RRT ^(1–4).Furthermore, many centers have reported that preoperative renal failure aloneis a poor prognostic indicator in both primary and repeat OLTX ^(3–7).Our center has previously reported that patients undergoing OLTX, who requiredpreoperative RRT or had preoperative renal failure (serum creatinine>2mg/dl), did not have an increased mortality compared with those patients whodid not have preoperative renal failure or who never required RRT, butpatients who required postoperative RRT did have an increased mortality ^{(8, 9)}.These analyses were performed on patients transplanted before 1995 at a timewhen only hemodialysis (HD) was utilized in our population. Since then, theuse of continuous RRT (continuous veno-venous hemodialysis, CVVHD) has becomemore readily available in our center. There are several reasons why the use ofCVVHD might be effective in patients with liver failure who undergo OLTX,including better control of intracranial pressure, better cardiovascularstability, and the removal of cytokines ^(10–17).Despite these theoretical advantages, there are few reports of the use ofcontinuous RRT in OLTX patients, and these reports indicate an early mortalityof 57–60% in patients who required this modality of treatment after OLTX ^{(4, 18–20)}.This paper will review our experience with the use of RRT and CVVHD inpatients undergoing OLTX from 1996 to 1999 compared with our earlierexperience from 1985 to1995.

MATERIALS ANDMETHODS

From June 1985 through the end of 1999, a total of1727 OLTX were performed in 1563 patients at Baylor University Medical Centerin Dallas. Combined liver-kidney transplants were not included in thisanalysis. Data on patients at our center is prospectively collected andmaintained in a database as previously described ^{(8, 9)}.In addition to the routine collection of data, all of the hospital inpatientdialysis records for the years 1996–1999 were examined to identify thosepatients undergoing OLTX who underwent either preoperative or postoperative HDor CVVHD. After eliminating transplants with incomplete data (n=89) andcombined liver-kidney transplants (n=40), a total of 1535 OLTX wereincluded in the analysis, 1037 from 1985 to 1995 and 498 from 1996 to 1999.All statistical analyses were performed utilizing the SAS statistical systemas previously described ^{(8, 9)}.Categorical variables were compared between groups utilizing a 2 tailedFisher’s exact test for 2×2 tables and with likelihood ratiochi-square for tables>2×2. Continuous variables between groups werecompared with Wilcoxon 2 sample test. Actuarial survival was estimated byutilizing Kaplan-Meier methods with significance testing utilizing Wilcoxontesting. Hemodialysis was performed with biocompatible dialysis membranes andminimal or no heparin. Duration of dialysis was usually 4 hr, 3–4 timesa week, depending on clinical status. Standard bicarbonate baths were utilizedwith potassium adjusted according to clinical status. Blood flow rate wasusually 300 ml/min with dialysate flow rates of 500–600 ml/min. CVVHDwas usually performed by utilizing the Prisma CVVHD system (Cobe) with AN69membranes and usually no heparin. A few early patients received CVVHD byutilizing a Gambro machine with biocompatible membranes. During 1999, somepatients receiving continuous RRT were treated with continuous veno-venoushemodiafiltration (CVVHDF) with 500–1000 ml of prefilter fluidreplacement. Blood flow rates were usually 100–150 ml/hr with dialysateflow rates of 1000 ml/hr. Standard peritoneal dialysate was utilized (Dianeal,1.5% glucose) with potassium, bicarbonate, or calcium added as clinicallyindicated. For CVVHDF, replacement fluid was usually D5 1/2 NS withbicarbonate and potassium as indicated. There were no standard treatments;each was individualized. A few patients ⁽⁹⁾ received both modalities. Eight patients received the majority of theirtherapy with CVVHD and were counted with the CVVHD groups. One patientreceived only 2 days CVVHD and the remainder of treatments with HD and wascounted with the HDgroup.

RESULTS

From1985 through 1995, patients requiring preoperative RRT represented 1.83% ofthe total, and those requiring postoperative RRT represented 6.46% of thetotal. Patients requiring RRT at any time represented 8.29% of the total. Aspreviously reported, except for the presence of hepatorenal syndrome, therewas no way to distinguish patients who would require RRT postoperatively fromthose who would not require RRT ⁽⁸⁾.We have reexamined that data and compared those patients requiringpreoperative RRT (including those continuing RRT postoperatively ⁽¹⁰⁾,n=19, group 1 to those who never required RRT (group 3, n=951)and those patients requiring postoperative RRT only, n=67, group 2, togroup 3. These results for preoperative intensive care unit (ICU) status,median waiting time, mean donor age, glomerular filtration rate (GFR) at timeof initial pretransplant evaluation, 90-day mortality, and 1-year survival arepresented in Table 1. Patients receiving RRT preoperatively were more likely tobe in the ICU preoperatively, have worse renal function, and have shorterwaiting times. Despite the need for RRT, these patients had no increase in90-day mortality and no decrease in 1-year survival when compared to group 3(no RRT). This includes the patients who required the continued need forpostoperative RRT. Patients requiring RRT postoperatively only (i.e.,developed renal failure postoperatively) were also likely to be in the ICUpreoperatively, have worse renal function, and shorter waiting times comparedwith those never requiring RRT. However, patients who developed the need forRRT postoperatively had a higher 90-day mortality and a worse 1-year survivalcompared with those who never required RRT or those patients who were startedon RRT preoperatively.

Table 1: Characteristics of patients requiringpreoperative or postoperative RRT compared with those who do not require RRT,1985–1995

We nextexamined those patients transplanted between 1996 and 1999. There were nopatients who required only preoperative RRT. There were 19 patients whorequired pre- and postoperative RRT (3.9%, 1 HD and 18 CVVHD) and 43 patientswho required postoperative only RRT (8.6%, 11 HD and 32 CVVHD). Compared with1985–1995, the total need for RRT has increased from 8.29% to 12.45%. Wecompared the patients who required RRT (all modalities pre- orpostoperatively, n=62) to those patients who never required RRT(n=436). The results of this analysis are presented in Tables 2 and3. Patients requiring RRT from 1996 to 1999 had worsepreoperative laboratory values compared with those who never required RRT.Furthermore, they were more consistently in the ICU preoperatively, had worseChild’s-Turcotte Pugh scores, were more likely to be a retransplantpatient, and despite shorter waiting times, had an increased 90-day mortalityand a decreased 1-year survival. The patients who required RRT also hadincreased donor age. Of note, the GFR was measured at the time of initialevaluation, not at the time of transplant. Furthermore, the RRT group includespatients who had normal renal function immediately preop. We then determinedthe 90-day mortality rates in patients requiring RRT, according to thedifferent type of RRT. There were 43 patients who required postoperative RRTfor new onset acute renal failure. Their 90-day mortality was 44%. All thepatients treated with CVVHD (n=50) had a 90-day mortality of 42%. Asshown in Table 4,the patients who developed renal failure requiring CVVHD postoperatively hadan increased 90-day mortality compared with patients who received eitherhemodialysis or CVVHD both pre- and postoperatively. This did not, however,reach statistical significance(P =0.09, HD vs. post-onlyCVVHD, P =0.08, pre- andpost-CVVHD vs. post-CVVHD). The causes of death in patients who received CVVHD(all) were multiorgan system failure in 19 (15 sepsis, 2 primary nonfunction,2 hepatic artery thrombosis), liver abscess due to vancomycin resistant Enterococcus in 1, central pontinemyelinosis in 1, and pulmonary hypertension with cardiac arrest in 1. From1996 to 1999, 38 patients who required RRT survived past 90 days. Seven havesubsequently died at 4, 4, 6, 6, 6, 7, 11, and 35 months posttransplant.Therefore, we looked at actual 1-year survival in the RRT patients. For thisanalysis, we grouped the patients into those who started RRT preoperativelyregardless of therapy (n=19, 1 HD and 18 CVVHD) and those who startedRRT postoperatively regardless of therapy (n=43, 11 HD and 32 CVVHD).As shown in Table 5,patients starting RRT postoperatively have a significantly higher 1-yearmortality. Of the surviving HD patients, 1/6 developed end-stage renal disease(ESRD) and received a kidney transplant after a course of maintenance HD. Ofthe surviving CVVHD patients, 4/25 developed ESRD, of which 3 remain onmaintenance HD and 1 has received a kidney transplant. The mean creatinine ofthe survivors (not including ESRD patients) is 1.89±0.75 mg/dl (range0.6–3.5 mg/dl).

Table 2: Comparison of preoperative laboratory values inOLTX patients who did or did not require RRT,1996–1999

Table 3: Comparison of patients who required RRT duringOLTX compared with those who did not require RRT,1996–1999

Table 4: Comparison of 90-day mortality by type of RRTin patients undergoing OLTX,1996–1999

Table 5: Actual 1-year survival of OLTX patients basedon timing of RRT,1996–1999

DISCUSSION

Developmentof the need for RRT is a devastating complication in the patient who isawaiting liver transplantation or in the patient who has undergone livertransplantation. It has been suggested, particularly in retransplantcandidates that the development of the need for RRT should precludetransplantation ^{(6, 7)}.Our center, on the other hand, from 1985 to 1995 had not found that the needfor preoperative RRT was a poor prognostic indicator. We had noted that thedevelopment of the need for RRT postoperatively was a poor prognosticindicator ^{(7, 8)}.However, we had previously been unable to identify prospectively (except forhepatorenal syndrome) those patients in whom the need for postoperative RRTwould develop. This made it difficult to allocate scarce livers and avoidplacing them into patients who would not do well. Although these patients weremore likely to be in the ICU preoperatively and to have worse pretransplantrenal function, we still find it difficult to determine preoperativelypatients that would do well and patients that would not. Indeed, even thegroup that one would expect to have poor survival (hepatorenal patients ondialysis preoperatively), had excellent 1-year survival in our experiencebefore 1996 ⁽⁹⁾.We did observe that the development of renal failure postoperatively, whichrequired RRT, occurred most often in the patients having complicatedpostoperative courses with events such as sepsis, primary nonfunction, andhepatic artery thrombosis (data not shown), events that are difficult topredict.

The time period from 1996 to 1999 has brought amarked change to our transplant center. Patients have had to wait longer fororgans (see Table 1 and Table 3).Although not directly quantified, they have appeared to be more debilitatedbefore transplant as manifested by laboratory values. More of the patientshave required RRT both pre- and postoperatively (8.29%, 1985–1995;12.45%, 1996–1999). Although we cannot exclude a selection bias in theearlier time period, we do not feel that our selection criteria have changed.Ominously, the 90-day mortality in patients not requiring RRT has risen from1.7% to 6.9%. Discouragingly, 90-day postoperative mortality has risen inthose patients developing postoperative renal failure severe enough to requireRRT from 32.8% in 1985–1995 to 44% in 1996–1999, although thisdifference did not achieve significance. Our 90-day mortality of 44% in allpatients requiring CVVHD is improved compared with previously published 90-daymortality of 58%, 59%, and 60%^{(4, 18, 19)}.However, we may only be prolonging the inevitable, because their have beenthree additional deaths in the HD group (50% survival) and four additionaldeaths in the CVVHD group, all in the post-only CVVHD group. Therefore, thesurvival to this point in the post-CVVHD group is only 37.5% (total mortalityof 62.5%). However, none of the pre- and post-CVVHD group has subsequentlydied, and they maintain a 72.8% survival to this point. Including the 1patient who started HD preoperatively with these 18 patients yields 19patients who started RRT preoperatively and continued postoperatively. Ofthese 19, only 5 have died for a survival of 73.6%. Therefore, althoughpreoperative need for RRT decreases survival after OLTX, these patients can bemanaged effectively and achieve reasonable survival. The absolute 1-yearsurvival of all the patients who developed postoperative renal failurerequiring RRT is 5/11 HD, 11/32 CVVHD, or 18/43 total (41.8%). This is worsethan the earlier time period.

This was not a randomizedtrial and the choice of therapy was left to the consulting nephrologists.However, sicker patients with cardiovascular instability, need for pressors,or lower blood pressures generally received CVVHD. Our results are verysimilar to those reported by Fraley et al. ⁽⁴⁾.Their series had 28 patients who started on RRT before OLTX and 83 who startedon RRT postoperatively. Of the 28 patients treated with HD, the mortality was21.4%. The 16 patients started on CVVHD preoperatively had a mortality of 44%,higher than ours. The 39 patients started on CVVHD postoperatively had amortality of 67%, very similar to ours. Although the populations may not beequivalent (27% of their OLTX population required RRT vs. only 12.25% ofours), their trends are the same. They quantitated the illness of theirpatients with Apache II scores and reached a similar conclusion that patientsrequiring CVVHD were indeed sicker.

Our overallimpression that we are dealing with a sicker population of patients issupported by a few observations. First, 90-day mortality has risen in patientsnot requiring RRT. Second, 1-year survival had decreased for patients who havestarted RRT preoperatively from 89.5% to 73.6%. Third, the development ofpostoperative renal failure requiring RRT (all types) resulted in a highermortality in 1996–1999 compared with 1985–1995. The 1-yearsurvival for patients developing postoperative renal failure requiring RRTfrom 1985 to 1995 was 55% compared with a survival of only 41.8% in similarpatients from 1996 to 1999. This cannot be attributed to patients from 1985 to1995 who were too sick to receive conventional HD, and therefore, died withoutRRT (i.e., an underestimate of the mortality associated with RRT), because the90-day mortality in non-RRT patients during that time period was very low.There are several potential explanations for this observation. First, patientsin the earlier time period, who were felt to be unstable or unable to tolerateHD, may not have been accepted as transplant patients. This many explain thelow mortality in the patient requiring preoperative RRT in the earlier period,but does not explain the increased mortality in patients requiringpostoperative RRT in the second period. Second, perhaps CVVHD leads to anincreased mortality by itself. This is not supported by the finding ofequivalent 90-day mortality in patients who started RRT preoperatively,regardless of modality in the second period. Third, patients arrive attransplant in a more debilitated condition. Thus, when acute renal failureappears, the outcome is worse. The data from Fraley et al. support this ideaand we favor it. This theory is supported by several findings; waiting time isincreased, liver function tests are worse, Child’s-Pugh scores arehigher, and preoperative renal function is worse in the RRTpatients.

CVVHD represents a therapy with greatpotential. It clearly can be performed in patients who are not stable enoughto have traditional intermittent HD. Davenport et al. ^(10–13) and Detry et al. ⁽¹⁴⁾ have demonstrated that CVVHD is better tolerated than HD in the liver failurepatient who develops the need for RRT as demonstrated by better cardiovascularstability and less severe fluctuations in intracranial pressure. Furthermore,it has the theoretical advantage of removing inflammatory cytokines(particularly tumor necrosis factor, interleukin (IL)-6, IL-8, and IL-10) whencompared with HD ^(15–17).This may explain the better cardiovascular stability seen in patients treatedwith this modality. Despite these presumed advantages, several studies incritically ill patients and patients receiving liver transplants have failedto show a survival advantage in patients treated with CVVHD ^(18–20).This, however, may not be the real question. The real question is how manypatients can be salvaged with this technique that would otherwise die, becausethey are unable to tolerate other therapies such as intermittentHD?

Our own series unfortunately cannot answer thisquestion. It was not a randomized study, and therefore, no conclusions can bedrawn about the utility of CVVHD versus HD in the treatment of these patients.Choice of CVVHD versus HD was made by the attending nephrologist and was oftendictated by patient stability and the availability of staff and equipment. Itis clear, however, that those patients who developed the need for RRTpostoperatively from 1996 to 1999 and were treated with CVVHD did poorly. Thiscould reflect the severity of their underlying complications and isdemonstrated in the cause of death. Another possible conclusion is thatperhaps RRT should have been started preoperatively, but we have been unableto determine patients that will require RRT postoperatively. One cannotdetermine whether CVVHD alone allowed us to salvage patients who would haveotherwise died due to inability to tolerate intermittent HD. However, one canconclude that the use of CVVHD is feasible in the very ill patient awaiting orafter OLTX and that it is a useful tool for the physician caring for thesepatients.

Development of ESRD after OLTX is clearlyhigher in this group of patients. The incidence of ESRD in our entire livertransplant population is only 0.3% at 3 years, rising to 8.5% at 11 years. (T.A. Gonwa, unpublished data). In a series from England, the risk of developingESRD after OLTX was 2% at 10 years ⁽²¹⁾.We have previously reported an increased risk of ESRD in OLTX patients who hadHRS pretransplant (8.9%) ^{(8, 9)}.From 1996 to 1999, 5/31 (16%) of the patients who required RRT and survivedlong-term have already reached ESRD after a very short follow-up. Furthermore,the mean creatinine is already elevated in this group of patients(1.89±0.75 mg/dl) comparable to the 1-year creatinine of 1.9±0.6mg/dl in our earlier group of HRS patients ⁽⁸⁾.Thus, we expect even more of these patients to progress to ESRD. Furtherefforts at identifying risk factors for acute renal failure or earlierintervention may be helpful in improving short-term outcome and preservinglong-term renal function in liver transplantrecipients.