According to a report from Japan's National Clinical Database, approximately 950,000 gastroenterological surgeries were performed throughout the country between 2011 and 2013 [1]. In these surgeries, hydroxyethyl starch (HES) was the most frequently used volume expander to maintain circulating plasma volume in patients with hemorrhage, while also reducing the need for albumin products.

Previous studies examined the use of HES in patients with severe sepsis and those admitted to intensive care units [2,3,4]. In both groups, a higher mortality risk and a greater need for renal replacement therapy were identified in patients administered HES rather than another fluid resuscitation preparation, such as Ringer's solution or saline. Given the frequent administration of HES in surgical settings, it is important to determine whether it is associated with kidney injury.

Prior to the approval of 6% HES 130/0.4, one of the most prevalent types of HES in the surgical setting in Japan was 6% HES 70/0.5 (molecular mass: 70 kDa; molar substitution ratio: 0.5). Although a previous study conducted in Japan found no evidence of an association between HES use and kidney injury, it was limited to patients with massive hemorrhage [5]. Therefore, in this study, we conducted retrospective full-cohort and propensity-score-based analyses to investigate the potential association between 6% HES 70/0.5 use and the incidence of acute kidney injury (AKI) in gastroenterological surgery patients treated at a Japanese university hospital.

The patient selection process for the full-cohort and propensity-score-based analyses is presented in Fig. 1. Among the 2,394 gastroenterological surgery patients initially included in this study, 2,218 were eligible for the analysis. Of these, 66 (3.0%) AKI patients (AKI group) were included in the full-cohort analysis: 40 (60.6%) patients categorized as "risk", 15 (22.7%) as "injury", and 11 (16.7%) as "failure" based on the RIFLE criteria. In the full-cohort analysis, the incidence of AKI in the HES groups 0 ml, 1–999 ml, and ≥ 1,000 ml was 3.2%, 2.7%, and 2.9%, respectively (P = 0.83).

According to the propensity scores of the three HES groups, 35 AKI patients and 1,269 non-AKI patients were matched between the observed and expected groups. Thus, there were 926 matched patients in the 0 ml HES group, 80 matched patients in the 1–999 ml HES group, and 298 matched patients in the ≥ 1,000 ml HES group. Among the matched patients, the incidence of AKI in the three HES groups was 2.5%, 0%, and 4.0%, respectively (P = 0.11). Thus, the RIFLE criteria results were similar to those of the full-cohort analysis.

Patient characteristics and outcomes are shown in Table 1. The differences between the AKI and non-AKI patients in the full-cohort analysis were significant for the following variables: sex, CKD stage, use of preoperative diuretics, ASA-PS classification, emergency surgery, surgical site, estimated blood loss volume, total crystalloids infusion, total RBC transfusion, and total albumin infusion. In the propensity-score-based analysis, the differences between the AKI and non-AKI patients were significant for ASA-PS classification, scheduled or emergency surgery, estimated blood loss, use of NSAIDs during surgery, and total albumin infusion. The difference in creatinine levels between the two groups were significant in both the full-cohort and propensity score-based analyses (P < 0.01 and P < 0.04, respectively). The mean propensity scores (95% CI) in the groups with 0 ml, 1–999 ml, and ≥ 1,000 ml of total infused HES were 0.60 (0.59–0.60), 0.39 (0.38–0.39), and 0.72 (0.69–0.74), respectively. In our series of patients, the maximum amount of administered HES was 4,500 ml; among the four patients administered ≥ 3,000 ml, one (25%) patient developed AKI.

Our study found no evidence of a significant effect of HES administration on AKI incidence after gastroenterological surgery, as the incidence in patients administered ≥ 1,000 ml was similar (about 0.3% lower) to that in patients not administered HES. Because the molecular mass and degree of substitution of 6% HES 70/0.5 are relatively low, the effect time of this plasma augmenter is relatively short, such that it is unlikely to influence postoperative kidney status. Our results are consistent with those of some, but not all, previous studies evaluating other surgical settings [5,8]. For example, Mårtensson and Bellomo [9] reported that intraoperative HES is harmful to postoperative renal function. Conversely, in a retrospective analysis of 846 surgical patients with massive hemorrhage (≥ 1,000 ml intraoperative blood loss), Endo et al. [5] found no association between 6% HES 70/0.5 administration and postoperative renal function. Our study also included patients with blood loss volumes of < 1,000 ml and the analyses were based on the categorization of patients according to HES infusion volume and estimated blood loss. In a retrospective study of 31 patients who had survived ≥ 1 month after an intraoperative blood loss of ≥ 500 ml, Suzuki et al. [10] showed that there was no significant difference between the amount of HES administered to patients who developed AKI and those who did not (53 vs. 55 ml/kg); they also failed to identify a correlation between HES dose and changes in serum creatinine. Ahn et al. [11] evaluated the risk of AKI in similar numbers of AKI and non-AKI patients; they concluded that HES should be used with caution in patients with decreased renal function who are undergoing thoracic surgery. Our study results may have been influenced by selection bias regarding renal function.

We also found no evidence of a relationship between perioperative medication and comorbidity. However, this does not necessarily rule out an association between these variables and postoperative AKI because our study focused on the association between intraoperative HES and postoperative AKI, such that a degree of selection bias for these variables may have been introduced.

The postoperative AKI incidence in our study (2.97%) is within the range reported for non-cardiac and cardiac surgical cases (0.8–14% and 5–20%, respectively) [12,13,14,15,16]. In a systematic review conducted by Gillies et al. [17], the incidence of AKI in patients administered HES intraoperatively was 5.4%, whereas in patients who had not been administered HES it was 3.6%. Although those authors addressed the impact of HES on AKI incidence, they did not include any studies in which patients received HES 70/0.5 (Hespander), the form administered to our patients. Despite the differences in the molecular mass and molar substitution ratio of HES 70/0.5 vs. other preparations, the incidence of AKI in this study was similar to the incidences reported in that review.

RBC transfusion is a risk factor for AKI after cardiac surgery [18,19], but an analogous association was not detected in our gastroenterological surgery patients. The multiple logistic regression model of our full-cohort analysis did not show any significant association between AKI and RBC transfusion, nor did the univariate analysis in the propensity-score-based analysis. Our omission of the RBC transfusion as a variable in the latter analysis was in accordance with the purpose of this study, which was to evaluate the association between 6% HES 70/0.5 and AKI in gastroenterological surgery patients; other factors that may contribute to the development of postoperative AKI were therefore not considered. Thus, whether RBC transfusion is also a risk factor for AKI in other clinical settings remains to be determined.

This study had several limitations that should be taken into account in the interpretation of its findings. First, the ASA-PS classification is a comprehensive evaluation of overall physical status; it does not provide insights into the individual comorbidities of each patient, except for hypertension and peripheral vascular disease. Other studies have shown that the incidence of postoperative AKI is affected by risk factors including male sex, older age, heart disease, hypertension, peripheral vascular disease, diabetes, chronic obstructive pulmonary disease, and hypoalbuminemia [13,14,15,16]. Except for hypertension and peripheral vascular disease, these individual comorbidity risk factors were not included as covariates in this study and their potential association with AKI in gastrointestinal surgery patients was therefore not addressed. However, our study did show a significant association between ASA-PS classification and postoperative AKI. By contrast, CKD stage was not a significant factor related to AKI incidence. The possibility that different CKD stages may necessitate changes in anesthetic management should be considered, especially in the selection of intravenous infusions and the use of NSAIDs. Although we included intraoperative use of NSAIDs, crystalloid infusion, RBC transfusion, and albumin use, we were unable to account for preoperative medications (except NSAIDs, angiotensin converting enzyme inhibitors, and angiotensin II receptor blockers) due to data limitations. Third, data regarding perioperative patient volume state, such as stroke volume variation (SVV) and pulse pressure variation (PPV), were not available for analysis, nor were data on postoperative urine volume. Although perioperative goal-directed therapy protocols are targeted at achieving hemodynamic optimization, neither SVV nor PPV is routinely monitored at our center, especially in patients undergoing minimally to moderately invasive surgery. We were therefore unable to determine the effect of perioperative hypovolemic state on postoperative AKI. Evaluation using the Clavien-Dindo classification may have provided information on postoperative leakage, infection, or ileus partly caused by hypervolemia and edema, and especially that associated with crystalloid use, whereas colloid administration would have reduced intestinal tract edema. In this study, AKI incidence was only slightly higher in patients who did not receive HES than in those who did. However, our definition of AKI relied only on the serum creatinine component of the RIFLE criteria. The inclusion of the above-cited information may alter the results. Nonetheless, a previous report demonstrated that AKI can be evaluated with the method used in this study [5]. Moreover, although we assumed that increases in blood loss volumes would require the infusion of larger amounts of HES, in patients with hemorrhage a portion of the infused HES may have leaked such that not all of the HES would have been metabolized by the kidneys. Our study differs from a previous study from Japan [5] in that it included blood loss volumes and HES infusion amounts in assessing AKI. Finally, this single-center study included only 66 patients with postoperative AKI vs. 2,152 non-AKI patients. The inclusion of a larger number of these patients in the logistic regression analysis would have increased the statistical power of our study. Our approach, however, was to stratify the independent variables into three different groups, with a minimum of 10 AKI cases for each variable, to achieve an adequate sample size in the final model. The statistical power was 0.94 at an alpha level of 0.05. Also, while the results indicated a relatively high frequency of AKI among patients who had been administered large quantities of HES (> 3,000 ml), there were only three non-AKI patients and one AKI patient. Our results must therefore be confirmed in a study with a larger sample size. At a volume of 1,000 ml, there was no significant increase in the incidence of AKI in our patients.

In conclusion, both the full-cohort analysis and the propensity-score-based analysis demonstrated a lack of association between clinical 6% HES 70/0.5 use and postoperative AKI incidence in gastroenterological surgery patients treated in a single center. Randomized clinical trials will be needed to definitively determine whether HES plays a role in the development of postoperative AKI.