Influence of different intraoperative fluid management on postoperative outcome after abdominal tumours resection

Before the procedure, thoracic epidural catheter was inserted in the left lateral position (Th 7–8 or Th 9–10 for rectal surgery) and the tested with 3 ml of 2% lidocaine was performed. After monitoring and placement of the intravenous line, the infusion of dexmedetomidine was started (0.5 mcg/kg/hour). The continuous infusion ended after skin suture at the end of the procedure.

Then a standard induction to general anaesthesia (propofol, sufentanyl, rocuronium) was performed. Anaesthesia was maintained by iv infusion of propofol. The depth of anaesthesia in both groups was adjusted to keep the bispectral index (BIS) 40–55.²⁵ Analgesia was provided by 15 ml of 0.25% epidurally levobupivacaine, with a 15 mcg sufentanyl supplement. 1–2 hours after epidural bolus of local anaesthetic, patient-controlled epidural analgesia (PCEA) was started with constant infusion rate and additional patient-controlled boluses for postoperative analgesia (PCEA (0.125% levobupivacaine 200 ml, morphine 4 mg, clonidine 0.075 mg; infusion rate 5 ml/h, bolus 5 ml, lock time 30 minutes). Relaxation was provided with rocuronium and monitored with the train-of-four monitor (TOF). Sugammadex was provided to reverse neuromuscular block at the end of operation.

The haemoglobin level was measured every two hours or at the events with blood loss over 500ml. It was kept above 80 g/L. A fall in haemoglobin was coped with blood transfusion. Oesophageal measured body temperature was kept in the range between 36 and 37 °C.

Postoperatively, the patients were transferred to postoperative recovery and then to Abdominal Surgery high dependency unit (HDU).

Monitors that calculate stroke volume (SV) and cardiac output (CO) from a standard radial arterial line (LiDCO Rapid, LiDCO Cardiac Sensor Systems, Cambridge, UK) were applied. The near-infrared spectroscopy (NIRS) monitor (INVOS, Medtronic, USA) was used in the protocol group. As a non-invasive technology that continuously monitors regional tissue oxygenation, it was used unilaterally to monitor cerebral oxygenation in the left hemisphere. A baseline prior to induction was recorded. Baseline values of the nominal stroke index (SI), cardiac index (CI), BIS, mean arterial pressure (MAP), and regional oxygen saturation (rSO2) were recorded.

The patients have received 2 ml/kg/h of balanced fluids + replacement for fluid loss (with a ratio of 3 units of balanced crystalloids per every unit of blood loss, until the Hb 80. Then the blood transfusion was started. In the event of immediate blood loss of more than 500 ml, colloids were administered with the ratio to blood loss 1:1. The exact number of fluids given, dependent on monitored hemodynamic variables.

Phenylephrine was administered when the SVV was below 13% of variation, CI was in normal range and there was hypotension. CI, MAP, and SI were maintained within 80% of baseline values. In the event of a 20% fall in regional cerebral oxygenation (rSO2) or values rSO2% below 60 in the absence of a fall in CI or blood loss, we adjusted ventilation so that PaCO2 was kept in the high normal range (5–5,5kPa).

The same anesthetic regime was used in PG; there was no hemodynamic monitor. Measurement of cerebral oxygenation was also absent. The patients received 2 ml/kg/h of balanced fluids and additional fluid for replacement for fluid loss as in protocol group. If there were no clinical signs of hypovolemia, phenylephrine was used in treatment of hypotension.

Postoperatively we collected the following data: length of stay, length of stay in HDU, re-admission to HDU or intensive care unit (ICU), quality of wound healing, reoperations, 30-day mortality. We have observed complications (sepsis, pneumonia, acute respiratory infection, pleural effusion, myocardial infarction, pulmonary embolism, stroke, infection).

To reduce unintended variations, strict discharge criteria were implemented. A hemodynamically stable patient without active infection, proper wound healing, and who has completed the first phase of rehabilitation to assisted mobility (or mobility comparable to preoperative) was discharged. If due to administrative reasons formal discharge was not possible, we considered him discharged if all the criteria were met.

Preoperative mini mental state test (MMSE) examination was conducted.²⁶ The aim was to measure possible postoperative cognitive decline. Postoperatively, the same testing was conducted after patient was admitted to the ward from high dependency unit.

The results were analysed using R: A Language and Environment for Statistical Computing. The results of intraoperative management, the results of postoperative creatinine, the demographics of the patients, and the length of stay are presented as the median and interquartile range. Groups were compared using the Mann-Whitney U test, the level of significance of 0.05 was considered statistically significant.

Intraoperative observations, postoperative complications, and ASA classification are presented as the absolute number of patients with a certain intervention/observation. Groups are compared using the Chi-square test or Fisher's exact test, where appropriate. A level of significance of 0.05 is considered statistically significant.

When comparing postoperative complications, several comparisons are made on the same sample. The level of significance was adjusted accordingly to the Bonferroni correction, and a p-value of 0.001 is considered statistically significant.

The use of multimodal monitoring resulted in differences in intraoperative management. The amount of fluid infused was higher in PG and vasoactive drugs were used more often. That suggests a trend towards more dynamic microcirculation. The most noticeable change in the postoperative period (related to differences in operative management) is a significant difference in the level of procalcitonin. Detailed discussion of those conclusions is provided below.

The results of the number of fluids given during the surgical procedure present an unexpectedly high fluid load in our PG. This group has received almost twice the number of fluids given in the CG. Intraoperative blood loss is comparable, and PG has a large positive intraoperative fluid balance. Thacker et al., reports the relation between higher fluid load and longer stay.²⁷ However, the length of stay was similar in both groups in our study. The choice of fluid (colloides or crystalloides) does not seem to have impact on overall morbidity.²⁸

One study compared goal-directed therapy (GDT) with standard fluid therapy in cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC). The study found that the use of a fluid therapy protocol combined with GDT was associated with a significant reduction in morbidity, length of hospital stay, and mortality compared to standard fluid therapy.²⁹ Similarly, Yu et al. conducted a controlled before-and-after study to evaluate the benefits of intraoperative goal-directed fluid therapy in major gynaecologic oncology surgery. The study found that the implementation of goal-directed fluid management was associated with a reduced risk of postoperative morbidities, particularly surgical site infections.³⁰ Another study analysed the impact of intraoperative fluid balance during pancreatoduodenectomy on the development of postoperative pancreatic fistula (POPF). The study found that fluid balance was significantly associated with the development of POPF, highlighting the importance of appropriate fluid management in pancreatic surgery.³¹

In addition, questions are raised about what the optimal goals of hemodynamic parameters (healthy population-derived normal values, preoperative values, maximal values) should be. Studies have shown that optimising cardiac output and oxygen delivery to higher values intraoperatively (supranormal) did not affect postoperative complications rate, intensive care unit stay, or hospital stay length.^{32, 33, 34} The question of fluid concentration has also been raised. The liberal approach can lead to oedema of the intestines and other tissues that may be responsible for poor tissue healing and other complications.³⁵ In abdominal surgery protocol-based fluid restriction reduced the incidence of perioperative complications such as cardiopulmonary events and altered intestinal motility while improving wound and anastomotic healing and reducing hospital stay compared to liberal fluid management.^9,10 One of the trials has shown a 52% lower rate of major postoperative complications in the restrictive group than in the conventional group.¹¹

Our study presents opposite results where the optimised group received a larger number of fluids. Our protocol has clearly defined steps when to add inotropes or fluid. One reason for the fluid load would be vasodilation due to epidural analgesia (all patients in our study have epidural analgesia), although vasoconstrictor (phenylephrine) in boluses was predicted to counteract the effect.³⁶ When comparing studies, 60 – 80% of the included patients have epidural analgesia.^21,37 Lopes et al., reports a significant decrease in ICU and hospital stay in the intervention group with an even greater difference in the number of infused crystalloids and colloids. They report a total volume of fluids infused 7 ml/kg /h in CG (roughly the same as in our study) and 21 ml/kg/h in the intervention group (12,5 in our study).³⁸ The choice of fluids and their timing also differs greatly.²⁰ The optimised group has received a larger volume of fluids, especially, although not exclusively, colloids.³⁹ Rare studies report other factors that greatly affect patient fluid status at the beginning, for example how long prior to the procedure are fasted, are fast track protocols implemented, etc. When trying to explain why sometimes one fluid regime (for example, restrictive) improves the outcome for the most, but not for all, we must realise that instead of restrictive or liberal there is only patient-directed fluid regime. Every patient should receive as much fluid as needed and at an appropriate time.⁴⁰

Significantly more patients with PG require vasopressor support with phenylephrine. Some articles suggest that anaesthesia after induction also causes venodilatation and not only arteriolar vasodilation (and consequently the decrease in MAP due to a change in volume out of the arterial tree into the dilated venous compartment).^{41, 42, 43} Phenylephrine infusion before induction minimises this effect, but this is hardly the complete explanation of the difference. To keep hemodynamic parameters as close to starting values as possible, the anaesthetist in PG probably reacted earlier than in CG. In CG, if MAP was kept to some extent (due to reflex mechanisms) there was no information about hemodynamic changes that would require intervention (fall in cardiac output and stroke index). The number of MAP falls below 70 mmHg and 50 mmHg is similar in both groups, but that does not mean that the duration of hypotension is the same.

BIS was used in both groups. This is in accordance with hospital policy, as total intravenous infusion was used to prevent intraoperative awareness.⁴⁴ In the context of multimodal monitoring, we omit an important variable that, without doubt, influences the outcome. Probably not only cognitive decline, but mortality and morbidity in general are related to too deep anaesthesia, a common occurrence without monitoring, especially in elderly people.^{25,45, 46, 47}

Our study does not confirm the benefit of using a cerebral oximetry monitor during major abdominal surgery, at least it does not influence the results as presented in this study. Cerebral oxygenation monitoring cannot be considered as a monitor of overall tissue oxygenation.⁴⁸ The incidence of renal impairment can be considered one of the measures of adequate oxygenation.

Neither the length of stay in the HDU nor the hospital stay decrease in PG. There are some postoperative complications such as the need for revision surgery, indication for antibiotic treatment third day after the procedure, or readmission to HDU that occur largely in the CG. Only comparison of individual complication does not show a statistically significant difference, but if we sum up all three, there is an obvious and statistically significant difference. Some other studies report more convincing but similar results.⁴⁹

C-reactive protein after surgery was elevated in both groups. There were no significant differences between the groups. A significant difference in the highest postoperative levels (in first 3 days) of procalcitonin was noted. Despite this, no clinically or microbiologically evident bacterial infection was confirmed. The level of procalcitonin increases in response to a pro-inflammatory stimulus, especially of bacterial origin.⁵⁰ The median value in CG is above the reference range. Since the surgical management in both groups was similar, different fluid status might explain stronger inflammatory response to tissue trauma in one group. It is important to note that procalcitonin is not a specific marker for bacterial infections and can be influenced by various factors, including noninfectious inflammatory reactions and tissue trauma.^51,52 Therefore, the role of procalcitonin in noninfectious tissue trauma is still not well-defined, and further research is needed to determine its diagnostic utility in this context.⁵³

Troponin leak was observed in 3 patients with PG. The increase is only marginally above the laboratory threshold value for positive. Acute myocardial infarction was ruled out in those patients with a high degree of confidence. Anyway, this can be related to a higher fluid load in the PG.

No major short-term differences in cognitive function are seen. Cognitive changes, related to the anaesthesia are much more subtile.^{54, 55, 56, 57, 58}

The patients involved in the study are very homogeneous in terms of surgical procedures, perioperative surgical management, and comorbidities. Demographic characteristics of both groups were similar. The type of anaesthesia was the same (total intravenous anesthesia [TIVA] with propofol and epidural analgesia) in all the patients observed. Compared to other prospective studies, the number of patients included is comparable.⁵⁹

Multimodal monitoring would probably provide more benefit, if used throughout the entire HDU stay not only during the surgical procedure.