The role of intraoperative parameters on predicting laparoscopic abdominal surgery associated acute kidney injury

This was a single-center prospective cohort study performed at King Chulalongkorn Memorial Hospital, Bangkok, Thailand between June 2012 and December 2013.

Eligible participants were adults aged 18 years or older, with American Society of Anesthesiologist (ASA) status between I and III, who underwent laparoscopic abdominal surgery for at least 2 h, long enough to impact kidney function [9]. Exclusion criteria included patients with pre-existing chronic kidney disease (estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m²) and patients on non-steroidal anti-inflammatory drugs (NSAIDs) 1 week before surgery, because peri-operation kidney function might be affected. All patients received the standard treatment for postoperative care including fluid management and hemodynamic optimization). Every patient received standard dose of inhaled anesthetic agent with the maximum dose at 1.3 minimal alveolar concentration. We used age, dose of nitrous oxide, hemodynamic status, body temperature, and intraoperative opioid dose to adjust the dose of inhaled anesthetic agents. Intraoperative CO₂ inflation was limited at 15 mmHg.

The study was approved by The Institutional Review Board (IRB No. 393/55), Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand. All participants accepted the protocol, and provided written informed consent.

Blood samples for measurement of serum creatinine and urine samples for testing urine neutrophil gelatinase associated lipocalin levels (uNGAL) were serially obtained at five time points from the start of surgery: 0, 6, 24, 48 and 72 h. We measured uNGAL using ELISA technique (R&D, Minneapolis, USA). Furthermore, we collected and measured urine output (UO) every 30 min during the operation period, and on days 1, 2 and 3 after operation.

We collected patient demographic data, pre-operative clinical characteristics, co-morbid conditions (hypertension, diabetes and dyslipidemia) and ASA status. We also collected intraoperative data including type of operation (urology, gynecology, colorectal surgery, upper general abdominal surgery or lower general abdominal surgery), type of inhaled anesthetic agents, operation time, inflation time, intra-abdominal pressure, exposure index, duration of intraoperative hypotension, amount of blood loss, amount of intravenous fluid administered and UO. Operation time was defined as the duration of the procedure from the initiation of skin incision to the completion of skin closure according to anesthesia records. Inflation time was the time after the creation of pneumoperitoneum by CO₂ inflation. Intra-abdominal pressure was the average pressure after the creation of pneumoperitoneum by CO₂ inflation. Exposure index, the new parameter, was the product of inflation time and intra-abdominal pressure. Intraoperative hypotension was defined by the mean arterial pressure (MAP) < 65 mmHg during surgery.

AKI was diagnosed using standard Kidney Disease Improving Global Outcomes (KDIGO) criteria [10]. However, due to UO being available only in 24 h increments we had to make the following modifications to the standard UO criteria: ≥ 0.5 ml/kg/h as no AKI; 0.3–0.5 ml/kg/h as stage 2; < 0.3 ml/kg/h as stage 3. Baseline serum creatinine was defined as the most recent pre-hospital creatinine value up to 1 year prior to hospital admission [11, 12].

Body weight on admission was used for calculating the rate of urine flow per kilogram per hour. For AKI staging patients were assigned to their worst KDIGO category according to either serum creatinine or UO.

Statistical analyses were performed using STATA version 15.0, with statistical significance set at p < 0.05. Comparisons between groups were performed with Fisher’s Exact test for categorical variables and using Student’s t-Test or Kruskal-Wallis one-way analysis of variance by ranks for continuous variables. Categorical data is summarized as counts and percent. Continuous data is summarized as mean (standard deviation) or median (25th, 75th percentile). For comparison of serum creatinine and urine NGAL between AKI and Non-AKI groups over the follow-up times (baseline, 6, 24, 48, 72 h), we used multivariable mixed model adjusted for age, diabetes status, urological surgery status and also the clustering effect from repeated measurements within patients.

We used unadjusted and adjusted logistic regression to individually test the association between the three main exposure variables (operation time, inflation time and exposure index) and AKI development. Risk factors considered for adjustment were priori risk factors including age, diabetes, baseline eGFR and type of operation (urological surgery). For analyses of developing AKI prediction, Area under the receiver operating characteristic (AUC-ROC) curve and reclassification analyses were used to measure the predictive performances of each models. The optimal cutoff points were determined by the largest sum of sensitivity and specificity.

The peak serum creatinine was seen 24 h after surgery with a significant difference between AKI and non-AKI patients: mean (SD) 1.01 (0.41) vs 0.78 (0.30) mg/dL, p < 0.013 (Fig. 2). Serum creatinine returned to baseline on postoperative day 3 in both groups. The median uNGAL level of AKI patients increased from 5.1 ng/mL at time 0 to 20.4 ng/mL 72 h after surgery. There median uNGAL level was significantly higher in the AKI group than in the non-AKI group at 72 h: median (Q1,Q3), 20.4 (6.4, 65.5) vs. 7.2 (3.0, 13.2) ng/mL, p = 0.037. By performing a paired time-series analysis, uNGAL level in the AKI group was higher than the non-AKI group, p = 0.016 (Fig. 3).

As shown in Table 2, the mean inflation time was significantly higher in the AKI group than in the non- AKI group: mean (SD) 192 (86.4) vs 151.1 (70.7) min, p = 0.045. Similarly, mean exposure index was significantly higher in AKI as compared to non-AKI patients: 2325.9 (1083.5) vs. 1806.1 (827.7) mmHg-min, p = 0.035. However, intra-abdominal pressure was not different between groups. In addition, duration of intraoperative hypotension, amount of blood loss, intravenous fluid and intraoperative urine output were not different between groups (Fig. 4).

Table 3 listed the derived sensitivities, specificities, and predictive values for operation time, inflation time, and exposure index at the cutoff that provided the maximum summation of sensitivity and specificity. The sensitivity and specificity of operation time, inflation time, and exposure index were optimal at the 255 min cutoff with AUC 0.61 (95% CI) 0.46–0.75; 150 min cutoff with AUC 0.63 (95% CI 0.49–0.78); and 1950 mmHg-mim with AUC 0.64 (95% CI 0.49–0.80), respectively.

In the multivariate model, adjusted for age, diabetic status, baseline eGFR, and type of operation (urological surgery), exposure index was the risk factors of AKI development with the OR of 1.47 per 500 exposure index increment, 95% CI 1.05–2.04, p = 0.024. Whereas inflation time had marginally significant association with the development of AKI, with OR of 1.24 per 30 min increment, 95% CI 1.01–1.58, p = 0.048 (Table 4). The AUC-ROC curve of operation time, inflation time, and exposure index for predicting postoperative AKI were 0.61, 0.63, and 0.64, respectively (Fig. 5a). While the model combining of intraoperative parameters with adjusted factor showed the AUC-ROC were 0.66 (operation time with adjusted factors), 0.69 (inflation time with adjusted factor), 0.71 (exposure time with adjusted factors), and 0.71 (operation time + inflation time + exposure time + adjusted factors) (Fig. 5b).

We also assessed the capability of intraoperative parameters (to ‘reclassify’ the degree of risk of AKI. Based on the clinical prediction model (operation time, inflation time, and exposure index), subjects were categorized into pre-specified ‘low’, ‘intermediate’, and ‘high’ risk of AKI (Table 5) using cutoffs of 0–30%, 30 to 60%, and > 60%, respectively. We then compared the proportions of reclassified subjects across each of these three risk groups when intraoperative parameters was added to the clinical prediction model. Among 23 patients who had AKI, 26.1% were reclassified as increased risk of when intraoperative parameters was added to the clinical prediction model and 22.0% were reclassified as lower risk. Among 41 non-AKI patients, 12.2% were reclassified as increased risk for AKI whereas 22% were reclassified as lower risk for AKI. There was no overall significant improvement in reclassification among AKI and non-AKI when intraoperative parameters were combined with clinical model (net reclassification index (NRI) 14.1%, p = 0.41). However, using the relative IDI, the reclassification of risk of AKI improved by 1.8% when intraoperative parameters were combined with clinical model (p = 0.025).