Effects of intraoperative and early postoperative normal saline or Plasma-Lyte 148^® on hyperkalaemia in deceased donor renal transplantation: a double-blind randomized trial

Abstract

Background

Administration of saline in renal transplantation is associated with hyperchloraemic metabolic acidosis, but the effect of normal saline (NS) on the risk of hyperkalaemia or postoperative graft function is uncertain.

Methods

We compared NS with Plasma-Lyte 148^® (PL) given during surgery and for 48 h after surgery in patients undergoing deceased donor renal transplantation. The primary outcome was hyperkalaemia within 48 h after surgery. Secondary outcomes were need for hyperkalaemia treatment, change in acid–base status, and graft function.

Results

Twenty-five subjects were randomized to NS and 24 to PL. The incidence of hyperkalaemia in the first 48 h after surgery was higher in the NS group; 20 patients (80%) vs 12 patients (50%) in the PL group (risk difference: 0.3; 95% confidence interval: 0.05, 0.55; P=0.037). The mean (sd) peak serum potassium was NS 6.1 (0.8) compared with PL 5.4 (0.9) mmol litre⁻¹ (P=0.009). Sixteen participants (64%) in the NS group required treatment for hyperkalaemia compared with five (21%) in the PL group (P=0.004). Participants receiving NS were more acidaemic [pH 7.32 (0.06) vs 7.39 (0.05), P=0.001] and had higher serum chloride concentrations (107 vs 101 mmol litre⁻¹, P<0.001) at the end of surgery. No differences in the rate of delayed graft function were observed. Subjects receiving PL who did not require dialysis had a greater reduction in creatinine on day 2 (P=0.04).

Conclusions

Compared with PL, participants receiving NS had a greater incidence of hyperkalaemia and hyperchloraemia and were more acidaemic. These biochemical differences were not associated with adverse clinical outcomes.

Clinical trial registration

Australian New Zealand Clinical Trials Registry: ACTRN12612000023853.

Hyperkalaemia is a common complication of kidney transplantation, occurring in 25–40% of recipients.^1–3 Although the outcomes of severe postoperative hyperkalaemia in kidney transplantation are not well studied, adverse consequences of hyperkalaemia in this setting include significant haemodynamic and neurological effects that result in respiratory paralysis or cardiac arrest if not treated urgently.^4,5 The aims of perioperative fluid administration in deceased donor renal transplantation include maintenance of adequate kidney transplant perfusion and the avoidance of electrolyte and acid–base disturbances, including hyperkalaemia.⁶ Sodium chloride 0.9%, known as normal saline (NS), is potassium free and, partly for this reason, is the current standard of care for i.v. fluid therapy for kidney transplantation.⁷

Despite the preferential use of NS in renal transplantation, previous studies in live donor^2,8–11 and deceased donor renal transplant recipients¹² report that NS causes hyperchloraemic metabolic acidosis, which can increase the risk of hyperkalaemia. Significant geographical, institutional, and subspecialty variation persists regarding fluid therapy,¹³ highlighting the uncertainty and paucity of evidence-based data surrounding the choice of best fluid in renal transplant recipients. Accordingly, we performed a double-blinded randomized trial comparing the effects of intraoperative and early postoperative NS or acetate-buffered crystalloid solution (Plasma-Lyte 148^®; Baxter Healthcare, Toongabbie, NSW, Australia; PL) on hyperkalaemia, the need for hyperkalaemia treatment, acid–base status, and graft function in patients receiving a deceased donor renal transplant.

Methods

The Austin Health Research and Ethics Committee approved this study (HREC no. H2011/04526). The trial was prospectively registered with the Australian New Zealand Clinical Trials Registry (ACTRN12612000023853). Between September 2012 and April 2015, the trial was conducted at Austin Hospital, a university teaching hospital in Melbourne with expertise in renal transplantation. Inclusion criteria included adult patients (age ≥18 yr) undergoing deceased donor renal transplantation. Exclusion criteria included preoperative hyperkalaemia >6.0 mmol litre⁻¹ that was not corrected before transplantation, pregnancy, chronic liver disease (liver function tests >1.5 times normal value), known allergic reaction to study solutions, and patients undergoing multi-organ transplantation. All participants provided written informed consent.

A statistician generated a computerised randomization sequence of 50 allocation codes, 25 for each group. Independent research staff sealed the allocation codes into sequentially-numbered opaque envelopes. The sequence was decoded after data analysis. Study participants, anaesthetists, surgeons, nephrologists and all perioperative clinicians and renal nursing staff caring for patients were blinded to the trial fluid intervention. Baxter Healthcare (Toongabbie, NSW, Australia) provided either NS or PL as 1000 ml identical blinded carrier solution containers. The PL contained sodium (140 mmol litre⁻¹), potassium (5 mmol litre⁻¹), magnesium (3 mmol litre⁻¹), chloride (98 mmol litre⁻¹), acetate (27 mmol litre⁻¹), and gluconate (23 mmol litre⁻¹). The NS contained sodium (154 mmol litre⁻¹) and chloride (154 mmol litre⁻¹).

Before surgery, all subjects were fasted for a minimum of 2 h for clear fluids and 6 h for solids. Immediately upon arrival in the operating room, subjects were randomized. After preoxygenation, general anaesthesia was induced using propofol (1–3 mg kg⁻¹ i.v.), fentanyl (1–3 μg kg⁻¹ i.v.), and cisatracurium (0.15–0.2 mg kg⁻¹ i.v.). Anaesthesia was maintained with sevoflurane or desflurane in a 50:50 oxygen–air mixture, maintaining a bispectral index of 40–60. Additional doses of fentanyl and cisatracurium were administered as appropriate.

All participants were prescribed a minimum of 2000 ml trial fluid during surgery, in accordance with Victorian Kidney Transplant Collaborative Renal Transplant Protocol. There was no standardized hourly rate, and the timing of fluid administration was at the discretion of the anaesthetist. Additional trial fluid boluses were permitted, with the rate, timing, and number of boluses individualized and based on estimates of surgical blood loss. Electrocardiography, pulse oximetry, capnography, blood pressure, urine output, and core body temperature were monitored. Use of invasive arterial blood pressure monitoring was at the discretion of the anaesthetist. Measurement of central venous pressure (CVP) was not used for the assessment of volume responsiveness because of the very poor relationship between CVP and blood volume and the inability of CVP or change in CVP to predict haemodynamic response to a fluid challenge. Other advanced haemodynamic monitoring devices were not used. Mean arterial pressure was maintained within 20% of the baseline preoperative value and supported with vasoactive therapy where appropriate. Open-label fluids (dextrose 5%, colloids, or blood products) were used at the discretion of the anaesthetist. Blood was transfused for haemoglobin <80 g litre⁻¹, or <90 g litre⁻¹ when further bleeding was anticipated. During surgery, the patient’s temperature was kept constant at 36 °C with a forced air-warming device, and arterial partial pressure of CO₂ was maintained at 35–40 mm Hg. Frusemide (125 mg i.v.) was administered 5 min before reperfusion according to standard institutional practice.

After surgery, all subjects received a maintenance infusion of trial fluid, with a starting rate of the previous hour’s urine output plus 30 ml. The trial fluid was continued as the designated crystalloid fluid for the first 48 h after surgery. The treating medical staff could alter the rate of trial fluid administration as clinically required or cease the infusion if crystalloid therapy was no longer needed. Other open-label fluids or blood products could be given if required in the opinion of the treating doctor. Any additional fluid boluses of trial crystalloid solution could be administered to any patient if volume supplementation was required. After 48 h the study fluid ceased, and the rate and type of any further fluid was at the discretion of the treating clinician. Immunosuppression was according to the local standard of care, with a regimen of tacrolimus (0.15 mg kg⁻¹ in divided doses initially and then adjusted to achieve a therapeutic trough tacrolimus concentration of 5–10 ng ml⁻¹), mycophenolate (1000 mg twice daily), prednisolone, and basiliximab commenced before surgery and continued throughout. Perioperative plasma exchange was used for higher immunological risk patients with known donor-specific anti-human leucocyte antigen antibodies. All recipients had negative T-cell and B-cell complement-dependent cytotoxicity cross-matches with donor sera before transplantation.

The primary outcome was the incidence of hyperkalaemia within 48 h of surgery. Hyperkalaemia was defined as potassium ≥5.5 mmol litre⁻¹ [normal range (NR) potassium in our laboratory is 3.5–5.5 mmol litre⁻¹]. The highest serum potassium value was also recorded, as was the number of subjects requiring treatment for hyperkalaemia. Secondary outcomes included hyperkalaemia during hospital admission, treatment for hyperkalaemia, requirement for postoperative dialysis, delayed graft function (DGF; hours to first dialysis after surgery, acid–base derangements), postoperative complications, and hospital length of stay. Treatment for hyperkalaemia was according to hospital policies under the direction of the treating clinician, with therapies including the following: binding resins; insulin, glucose, or both; calcium gluconate i.v.; and dialysis. Delayed graft function was defined as either the requirement for dialysis because of graft dysfunction in the first 7 days post-transplant or a failure of serum creatinine to reduce by 20% in the first 72 h after renal transplant. Post-transplant graft function was also assessed using the creatinine reduction ratio (CRR) on postoperative day 2 and kinetic estimated glomerular filtration rate (KeGFR), which are validated metrics associated with graft outcome.^14–16 Renal perfusion was assessed with technetium-99m mercaptoacetyltriglycine (Tc99m MAG3) nuclear medicine scan and Doppler ultrasound performed on postoperative day 1. All other complications were defined and classified according to the European Perioperative Clinical Outcome definitions,¹⁷ and recorded and graded according to the Clavien–Dindo method.¹⁸ Potassium concentrations and all other biochemistry parameters were sampled before surgery, immediately on arrival at the postoperative anaesthesia recovery unit, and at 24 and 48 h after surgery. Renal function (both serum creatinine and estimated GFR by the Chronic Kidney Disease Epidemiology Collaboration equation)¹⁹ was also measured at 1 week, 1, 3, 6, and 12 months post-transplantation.

Statistical analysis

Sample size calculations were based on a pilot study from 44 patients receiving renal transplants in our institution.³ A difference in serum potassium of 0.5 mmol litre⁻¹ at 48 h was considered clinically important. Assuming a mean serum potassium of 5.7 mmol litre⁻¹ in the NS group and a mean serum potassium of 5.2 mmol litre⁻¹ in the PL group, with an sd of 0.7 mmol litre⁻¹, an α value of 0.05, and a desired power of 0.80, a total of 23 subjects were required in each group. We received ethics approval to recruit 50 subjects. Statistical analysis was performed using commercial statistical software STATA/IC v.13 (College Station, Texas, USA). Data were reported as the mean (sd) if normally distributed or median [interquartile range (IQR)] if not normally distributed. Student’s unpaired t-test with Welch’s correction or the Wilcoxon–Mann–Whitney rank sum test was used to detect a difference in continuous characteristics between groups depending on the satisfiability of the distributional assumptions. Fisher’s exact test was used to investigate differences between groups in categorical outcomes, including the primary outcome of hyperkalaemia (serum potassium >5.5 mmol litre⁻¹) in the first 48 h after surgery. A two-tailed P-value of <0.05 was used as a threshold for statistical significance. To analyse potassium values throughout the four prespecified time points, we used a random-effect generalized linear regression model, with individual subject treated as a random effect. Owing to the exploratory nature of this analysis with regard to secondary end points, no correction for multiple testing was undertaken. The study was reported in accordance with the CONSORT guidelines for reporting randomized trials.²⁰

Results

During the study period, 70 patients underwent deceased donor renal transplantation. The CONSORT diagram is presented in Fig. 1. Thirteen patents were excluded (refused consent, n=5; inclusion criterion not met, n=2; transplant proceeded in absence of study investigators, n=6). Fifty-seven patients provided informed consent. Seven participants did not proceed to transplant because the donor kidney was considered unsuitable. Twenty-five subjects were randomized to NS and 25 to PL. One subject in the PL group was excluded because of a protocol violation (the blinded trial fluids were not administered). There were no other study protocol breaches or violations, and sampling of blood at the proposed study times was consistent in all participants. Table 1 reports baseline characteristics, which were similar with respect to preoperative biochemistry, age, cause of renal failure, donor age, and proportion of kidney donations after circulatory and brain death.

Primary outcome: hyperkalaemia within 48 h

During the first 48 h period after surgery, there was a higher incidence of hyperkalaemia after NS (80%) than with PC [50%; risk difference: 0.3; 95% confidence interval (CI): 0.05, 0.55; P=0.04). Likewise, 64% of subjects were treated for hyperkalaemia after NS vs 21% after PL (risk difference: 0.43; 95% CI: 0.18, 0.68; P=0.004; Table 2). In the first 48 h after surgery, the mean (sd) highest potassium concentrations were 6.1 (0.8) mmol litre⁻¹ after NS compared with 5.4 (0.9) mmol litre⁻¹ after PL (mean difference: −0.65; 95% CI: −1.12, −0.17; P=0.009). Medical management for hyperkalaemia with insulin, dextrose, or both, resonium, or calcium gluconate was instituted in 10 subjects in the NS group vs seven subjects in the PL group (P=0.55). Thirteen subjects required dialyses for hyperkalaemia in the NS group vs four in the PL group (P=0.02).

Intraoperative variables

Fifteen subjects receiving NS and 19 receiving PL had arterial blood gases checked during surgery. The peak mean (sd) potassium value was 5.4 (1.1) mmol litre⁻¹ in the NS group compared with 4.9 (0.8) mmol litre⁻¹ in the PL group (mean difference: −0.42; 95% CI: −1.13, 0.28; P=0.23). Four subjects were treated for hyperkalaemia with insulin and dextrose in the NS group vs one in the PL group (P=0.27; Table 3). The median (IQR) volume of intraoperative trial fluid administered in NS group was 2500 (2000, 3000) ml compared with 3000 (2000, 3000) ml in the PL group (P=0.92). Additional intraoperative fluid, blood, and use of vasoactive medications were similar. Subjects receiving NS were more acidaemic [mean (sd) pH 7.32 (0.06) vs 7.39 (0.05) (pH normal range: 7.35–7.45); P=0.002] and hyperchloraemic [median (IQR) 107 (105, 110) vs 101 (99.8, 103.3) mmol litre⁻¹; (normal range for chloride: 98–107 mmol litre⁻¹); (P<0.01)] compared with PL. Eight subjects (32%) produced urine in the NS group vs 11 (46%) in the PL group (P=0.39). Perioperative fluid administration is summarized in Table 3. The median (IQR) duration of surgery was 2.8 (2.5, 3.2) h in the NS group and 3.0 (2.6, 3.3) h in the PL group (Wilcoxon–Mann–Whitney P=0.32). The median (IQR) cold ischaemic time in NS group time was 11.8 (6.3; 14.3) h and in the PL group 10.8 (9.8, 13.2) h (Wilcoxon–Mann–Whitney P=0.74). Likewise, there were no differences in warm ischaemic time between the NS and PL groups; 34 (27, 41) vs 33 (24, 43) min (Wilcoxon–Mann–Whitney P=0.56).

Longitudinal analysis, delayed graft function, and complications

The electrolyte and biochemical values measured at each time point are presented in Table 4. For any given time point, subjects receiving PL had lower potassium values by an average of −0.34 mmol litre⁻¹ (95% CI: −0.63, −0.05; P=0.02) compared with NS. Medical management for hyperkalaemia with insulin, dextrose, or both, resonium, or calcium gluconate was instituted in five subjects in the NS group and no subjects in the PL group. There was no requirement for dialysis to manage hyperkalaemia in either group. No significant difference in the incidence of DGF (Table 5) was identified. Although fewer subjects in the PL group required dialysis in the first 48 h, this difference was not statistically significant. Subjects in the PL group who did not require dialysis had a more rapid reduction in creatinine on day 2 (P=0.04), with a trend towards better KeGFR in the PL group at 48 h (P=0.12; Fig. 2). Clinical and radiological markers of renal function after surgery are summarized in Table 4.

The incidence of complications was similar between groups (Supplementary material File 1). Two subjects in each group were electively admitted to intensive care after surgery for cardiovascular monitoring, optimization of fluid status, and vasoactive therapy. Seven subjects in the NS group and nine subjects in the PL group required perioperative blood transfusion. The median (IQR) length of hospital stay was 6.5 (6, 13.2) days after NS and 7.1 (6, 10.4) days after PL (P=0.18). There was no 30 day mortality. Follow-up blood samples obtained at 1 week, 1, 3, and 6 months and 1 yr intervals showed a similar improvement in markers of renal function over time in both groups (Table 5).

Discussion

We performed a prospective randomized double-blind controlled study comparing the effects of NS and PL as perioperative i.v. fluid therapy in deceased donor renal transplant recipients. Compared with PL, subjects receiving NS were more acidaemic during surgery, and after surgery had higher serum potassium and a greater incidence of hyperkalaemia. Moreover, there was a higher rate of interventions to correct hyperkalaemia in the first 48 h after surgery. Finally, although subjects in the PL group who did not receive dialysis had a greater CRR on day 2, there was no significant difference in the rate of DGF.

Relationship to previous findings

Potura and colleagues recently investigated the incidence of hyperkalaemia in the context of deceased donor renal transplantation.¹² Similar to our findings, they reported no significant difference in the incidence of intraoperative hyperkalaemia in recipients receiving open-label NS or an acetate-buffered solution, and also reported significant intraoperative metabolic acidosis and hyperchloraemia. In contrast to the present study, however, they continued open-label trial fluid for only 4–5 h after surgery, whereas we continued blinded fluids for 48 h postsurgery, allowing a more detailed and robust evaluation of the impact of fluid intervention on biochemical and clinical outcomes throughout this time period.

The higher incidence of hyperkalaemia in the NS group occurred despite the fact that, in comparison to PL, less NS in total was administered during the study period. This trend reflects a trend to a lower urine output in the NS group, because the routine practice was to link the volume of fluid administration to the observed urine output. Overall, data suggest that the mechanism for the higher incidence of hyperkalaemia in the NS group relates to its chemical properties (i.e. high chloride, low pH), rather than the fluid volume.

Several studies in live donor kidney transplant recipients have also reported an increased incidence of metabolic acidosis in patients receiving NS compared with balanced crystalloid solutions.^2,8–10 Similar to our study, some of these groups also evaluated renal function as a key outcome.²⁸ These studies found no significant difference in markers of renal function between groups, although they were not powered to do so. Although iatrogenic hyperchloraemia is associated with renal impairment and reduction of renal cortical perfusion in both animal and human studies,^21,22 this has not consistently translated into major deleterious renal functional outcomes for renal transplant recipients.

Despite the limited power of the present study, we observed a modest efficacy signal towards better renal outcomes in the PL group than the NS group. Compared with previous reports,^23–26 we observed a high rate of both DGF overall (haemodialysis within 7 days post-transplant) and the need for haemodialysis within the first 48 h post-transplant. This limits any analysis of changes in serum creatinine or estimated GFR because of the use of haemodialysis, leading to exclusion of many subjects from such analysis. However, fewer subjects were dialysed in the first 48 h in the PL group. Moreover, we observed a statistically significant and clinically meaningful increase in CRR in subjects in the PL group and a corresponding trend towards better KeGFR in the PL group at 48 h. The binary (yes/no) measure of the requirement for dialysis within 7 days post-transplant fails to account for the following factors: (i) severity and duration of DGF; (ii) dialysis for hyperkalaemia and fluid overload with good graft function; and (iii) poor graft function where dialysis is not needed. Large studies have demonstrated that DGF is associated with inferior outcomes regardless of the need for dialysis.²⁷ The CRR on day 2 was used as a measure of early graft function in non-dialysed subjects because it is inversely correlated with 12 month graft function, and low CRR predicts graft loss.^{14 28} The CRR2 also measures relative rather than absolute change in creatinine concentration, and is thus less likely to be confounded by non-graft factors (e.g. recipient muscle mass or fluid status) and is more useful in predicting subsequent graft function.^29,30

The rate of donation after circulatory death (overall 30%) was similar between groups, so this did not account for the high rate of DGF observed. Furthermore, donor age seems comparable to other centres. The higher DGF rate may be attributable to donor time to death, pharmacological protection of transplant kidneys during cold perfusion, or expanded-criteria donors, or different clinician thresholds for initiation of dialysis compared with other centres.³¹ Such variables were not collected as part of this study. No differences in kidney function were observed between groups at 1, 3, 6, and 12 months, although the study was not powered to detect such differences. It is possible that the choice of fluid has less influence on the longer-term function of the transplant than other more well-established donor and recipient characteristics.

Implications

Compared with PL, NS leads to inferior biochemical outcomes in relationship to hyperchloraemic metabolic acidosis and hyperkalaemia. These findings indicate that concerns about the potassium found in PL exposing patients to a greater risk of hyperkalaemia are unjustified when compared with the effects of NS. More importantly, they suggest that, on physiological and biochemical grounds, PL may be superior to NS in patients receiving deceased donor kidney transplants. It is possible that less hyperkalaemia would occur with a physiologically balanced, low-chloride formulation that does not contain potassium; however, to our knowledge, no such formulation is presently available for clinical use.

Strengths and limitations

Our study has several methodological strengths. The fluid intervention was blinded and randomized, thus minimizing allocation and selection bias. Blinding also avoided the potential for differences between groups in terms of management of electrolyte and acid–base disturbances attributable to physician awareness of the allocated fluid therapy. The extended use of blinded fluids for 48 h after surgery was strictly adhered to in all participants, with electronic medical recording allowing a detailed and accurate tally of fluid use and fluid balances, reinforcing the internal validity of the key outcome variables reported. Previous studies comparing balanced fluids with NS in kidney transplantation were largely restricted to the intraoperative period and may not have delivered a sufficient volume of trial fluids to determine the full effect of fluid choice. This is also the first fluid intervention study in deceased donor renal transplant recipients to report renal outcomes at 1 week, 1, 3, 6, and 12 months after trial fluid intervention.

Limitations of the study include inadequate power to compare the effects of NS and PL on short- or long-term graft function and other postoperative complications; addressing this will require a larger, multicentre study.

We are uncertain whether the immunosuppressive therapy contributed to the incidence of hyperkalaemia in each group; we consider this unlikely because all participants received the same immunosuppressive protocol. The rate of DGF in the present study was high, possibly reducing the generalizability of the findings; however, this reinforces the relevance of the results to patients at high risk of poor graft function and associated hyperkalaemia. Further studies are needed to determine whether our results are applicable to live donor kidney transplant recipients or to other patients at risk of perioperative acute kidney injury. Interestingly, we observed a strong signal in the incidence of hyperkalaemia in participants receiving PL in donation after brain death recipients when compared with donation after cardiac death recipients, with statistically significant and clinically meaningful reductions in hyperkalaemia immediately after surgery and 24 and 48 h after surgery. Given our small sample size, these findings should be interpreted cautiously and may reflect a type 1 error. Finally, we cannot provide information on the mechanism of fluid-associated hyperkalaemia, and whether the greater incidence of hyperkalaemia observed with NS was related to its acidaemic or hyperchloraemic properties, and whether these impacted mean renal artery flow velocity or renal cortical tissue perfusion as reported by others.^{21 22}

Conclusions

In deceased donor kidney transplantation recipients receiving PL or NS during surgery and for 48 h after surgery, use of NS was associated with hyperchloraemic acidosis and more frequent postoperative hyperkalaemia, with greater use of medical interventions to control serum potassium. These biochemical differences were not associated with adverse clinical outcomes. Given the potential for serious consequences from postoperative hyperkalaemia, this study supports the preferential use of buffered crystalloid solutions for perioperative fluid management in patients undergoing deceased donor renal transplantation. Further studies in larger populations at average risk of DGF are required to confirm these findings and to determine the effects of different perioperative fluids on kidney transplant function.

Authors’ contributions

Study design: L.W., R.B., P.M.

Patient recruitment: L.W., P.M., F.I., D.S., G.E.

Data collection: L.W., L.H., P.M., G.E.

Data analysis: L.C., L.H., L.W., M.C.

Writing up the manuscript: L.W., P.M., R.B., M.C., F.I.

Supplementary material

Supplementary material is available at British Journal of Anaesthesia online.

Declaration of interest

L.W. and R.B. have received honoraria from Baxter Healthcare. Study conception, design, trial management, data collection, data analyses, and writing of the manuscript have been executed completely independently of Baxter Healthcare and any other external organizations.

Funding

Baxter Healthcare (compounded and supplied the blinded fluid solutions to Austin Hospital); The Department of Anaesthesia Research Fund at Austin Hospital.

References