Development of a risk stratification model for predicting acute renal failure in orthotopic liver transplantation recipients*

Abstract

SummaryThe purpose of this prospective observation cohort study was to develop and validate a risk stratification model for prediction of acute renal failure after liver transplantation. Data from 71 orthotopic liver transplantation recipients were used to develop a risk stratification model by binary logistic regression analysis containing the following variables: pretransplant hepatitis B and ⁄ or C infection; arterial hypertension; intra-operative mean arterial blood pressure before induction of anaesthesia; units of packed red blood cells required; hypotension (mean arterial blood pressure £50 mmHg); and maximum lactate concentration. Validation of the model on 167 consecutive patients revealed a sensitivity of 0.78, a specifity of 0.92, a negative predictive value of 0.96 and a positive predictive value of 0.62. This risk stratification model enables identification of orthotopic liver transplantation recipients with no or low risk for acute renal failure directly at the end of the transplantation procedure. Acute renal failure is a common and severe complication after major surgery and organ transplantation. In a population-based cohort analysis including 69 321 patients after non-renal organ transplantation, 7.6% of the patients developed postoperative acute renal failure (defined by 50% reduction in glomerular filtration rate or a need for urgent dialysis) [1]. Previous estimations of the incidence of acute renal failure after orthotopic liver transplantation in adults vary between 31% and 94%, depending on the definition used [2][3][4][5][6]. Its multifactorial origin complicates the identification of risk factors to develop prevention strategies. Acute renal failure can occur early in the postoperative course and seems to be related to pre-operative conditions and to the operative procedure. Impaired renal function and especially the need for post-transplant renal replacement therapy is associated with prolonged hospitalisation, increased mortality [2,[7][8][9], and significant financial cost [10].These observations prompted this prospective observational study, with the aim of developing a prediction model to identify patients at risk for acute renal failure requiring renal replacement therapy directly at the end of the transplantation procedure. Therefore we investigated only pre-and intra-operative risk factors for the development of acute renal failure after orthotopic liver transplantation, postoperative risk factors such as nephrotoxic drugs or postoperative complications having been extensively studied in the past.

Immunosuppression in liver transplant recipients with renal impairment

“…Severe forms of acute renal ''failure'' (RIFLE) requiring renal replacement therapy (RRT) have been reported in 6.5-18.3% [9][10][11][12]4,13] of cases, usually early, within the first week post-LT. In this setting, acute renal failure results from major post-operative stresses frequently associated with acute tubular necrosis, including prolonged hypotension, sepsis or septic shock, sustained pre-renal failure, primary graft non-function, or delayed function of the liver.…”

Acute Kidney Injury in the Surgical Patient

“…These scores predicted a probability of severe AKI between <1% and >20%, with the ROC-AUC varying between 0.77 and 0.84. For patients undergoing non-cardiac surgery a few predictive models, developed in small cohorts utilizing limited intraoperative data or in larger cohorts but using only severe AKI as an end-point, are further limited by the lack of validation studies and provide only modest predictive accuracy [22, 112-114]. Recently UK AKI in Cardiac Surgery Collaborators group have developed and validated a new risk prediction score for any stage AKI after cardiac surgery with ROC-AUC of 0.74, providing better discrimination compared to previously published scores [115].…”

“…The volume of physiologic data routinely acquired during intraoperative hemodynamic monitoring is rarely used in published risk scores and when used, it is usually summarized by some reductionist approach (mean, lowest value, etc) rather than applied in their continuity and complexity and almost never in automatized fashion [13, 22, 106, 112, 113, 116-121]. Lack of sophistication in both data collection and analysis of real time physiologic data has limited this approach.…”

Perioperative Risk Assessment, Prevention, and Treatment of Acute Kidney Injury

“…Kheterpal et al 14 determined that the predictive value of their previously mentioned renal risk prediction index improved from ROC-AUC of 0.77 to 0.79 if the intraoperative risk factors of use of a vasopressor infusion, mean number of vasopressor bolus doses administered, and the administration of furosemide or mannitol were added to the analysis. In liver transplantation, Rueggeberg et al 41 developed a risk stratification model for predicting AKI immediately at the end of the transplantation procedure. In it they included the intraoperative risk factors of units of packed red blood cells required, hypotension defined as mean arterial pressure less than 50 mm Hg, and maximum lactate concentration.…”

“…Their model had a sensitivity of 0.78, a specificity of 0.92, and a negative predictive value of 0.96 for AKI at the end of transplantation. 41 Although the value of each of these individual risk factors is unknown, the absence of them is reassuring.…”