Sublingual microcirculation in pancreatico-biliary surgery: An observational study

1.1. Background

Pancreatico-biliary resection remains one of the biggest challenges to anesthesiologists and surgeons in abdominal surgery. Although there is a decline in mortality rates after pancreatectomy, attributed to concentration of these procedures in high-volume hospitals [1], the morbidity rates are still high [2]. The clinical morbidity rates are ranging from 35% to 62% after pancreaticoduodenectomy, and from 33% to 50% after distal pancreatectomy in large cohort studies [2]. The fluid management seems to be an important factor in the reduction of the complication rate in gastrointestinal surgery. [3] However, these findings are still questionable in pancreatectomies. Some studies showed that restrictive fluid management reduces postoperative morbidity after pancreatectomy [4, 5], while other studies contradict these findings. [6, 7] Recently, a large randomized controlled trial addressed this issue by randomizing patients that underwent pancreatic resection to receive either liberal or restrictive perioperative fluid regimen [8]. The study showed that there was no significant difference between these regimens in terms of perioperative complications. Due to the lack of consensus on the optimal implementation of fluid management, individualized goal-directed fluid therapy have been introduced in the perioperative setting [9]. This goal-directed fluid therapy has demonstrated to improve tissue oxygen tension and microcirculatory perfusion in major abdominal surgery compared to restrictive fluid regimen. [10] There are several methods that are used as guide for this goal-directed fluid therapy such as central venous pressure, Doppler-guided fluid boluses or stroke volume variation. However, these methods are not convenient for the conscious surgical ward patient. The current parameters for fluid management in the surgical ward are maintaining cardiac output, blood pressure and urine output [11]. However, recent studies have shown that the measurement of microcirculation is considered to be an important adjunct measurement to conventional hemodynamic monitoring [12]. Microcirculation is the delivery site of oxygen to tissue cells and is essential for the maintenance of cellular life. Thus the function of organs is directly dependent on the function of their respective microcirculation [13]. The aim of fluid therapy is to promote tissue perfusion to sustain oxygen delivery to the organs [14]. This organ oxygenation can be divided in three parts. The first part is carried out by the systematic circulation, i.e. cardiac output, blood pressure and oxygenated red blood cells. The second part is the locoregional site of the organ, the microcirculation, with the main objective to distribute flow to the organ. The last part is at a cellular level, where mitochondria utilize the oxygen. So physiologically there are two main factors that determine oxygenation of organ tissue: the convective transport of oxygen by red blood cells to the microcirculation of the organs and the passive diffusion of oxygen from this microcirculation to the cell to initiate a chain reaction that results in ATP. [12] A big shortcoming of the conventional fluid therapy monitoring (cardiac output, urine output and blood pressure) is that they only account for the convective factor of oxygenation, not taking into account the diffusive factor. Simplified, one can imagine that despite optimal flow in the capillaries, the farther away a target cell is from the oxygenated red blood cell or the lower the oxygen solubility of the tissue cell is, the more difficult oxygenation of the targeted cell becomes. This diffusive capacity of the microcirculation has been quantified by the capillary density filled with the oxygenated red blood cells. Indeed, Pranskunas et al. illustrated this in a prospective observational study that examined patients with clinical signs of impaired organ perfusion and their microcirculation. Patients with abnormal microcirculation responded to fluid challenge by an improvement of their microcirculation and more importantly by a reduction of the number of clinical signs of impaired organ perfusion. By contrast, in patients with signs of impaired organ perfusion that had normal microcirculation, fluid challenge did not improve their microcirculation or their number of signs of impaired organ perfusion. These findings were found to be independent from stroke volume responsiveness after fluid challenges [15]. This study was conducted in intensive care unit patients and not strictly in surgical patients.

1.2. Objectives

We conducted a prospective observational study that tried to identify perioperative differences in sublingual microcirculatory parameters that can be used for goal-directed fluid therapy. Furthermore, a correlation between perioperative clinical outcomes and microcirculation were examined.

2.1. Study design

A prospective cohort study of all patients undergoing pancreatic resection were enrolled to identify their perioperative microcirculatory changes.

2.2. Setting and participants

Patients that were scheduled for an open elective surgery for pancreatico-biliary pathology were prospectively enrolled between May 2014 and December 2014. Informed consent was obtained from all patients, in adherence to the institutional review board (Erasmus Medical Center, ’s Gravendijkwal 230, 3000 CA Rotterdam, 24 January 2014, under number NL45915.078.13 by prof. dr. Tilanus).

2.3. Variables and measurement

Sublingual microcirculation was measured using the Cytocam (Braedius Medical, Naarden the Netherlands) that is based on video microscopy and IDF. [16] The measurement was performed in three different time points: one day before surgery (T0), within the first 24 hours after surgery (T1) and on the fourth day after surgery (T2)(Video 1,2 and 3). OPEN IN VIEWER

Videostill image 1.2,3 The sublingual microcirculation of a patient one day before undergoing distal pancreatectomy with splenectomy (T0), one day after (T1) and four days after (T2) the surgery. Images were obtained using the IDF device At each time point, microcirculation was recorded in three different spots of the sublingual mucosa for a minimum of 2 seconds per sequence by an independent researcher not participating in the care of the patient. Video sequences were repeatedly taken until getting three different sequences of good quality were obtained. Afterwards, video sequences were offline blindly analyzed by a video analyst who had no involvement in the image data acquisition. The video quality was assessed with a standardized microcirculation image score, the Massey score [17]. Video sequence analysis was performed using Automated Vascular Analysis (AVA 3.0 software; MicroVision Medical, Amsterdam, the Netherlands) [18]. The images were randomly presented to the video analyst. Microcirculatory parameters were described as: Microvascular flow index (MFI (AU)), Total vessels density (TVD (mm/mm2/)), Perfused vessels density (PVD (mm/mm2)), and proportion of perfused vessels (PPV (%)).These parameters were obtained in each time point and for small vessels alone (diameter <20mm) and for all vessels.

2.4. Data sources

The following data were recorded at baseline: age, sex, height, weight, comorbidity and ASA physical status classification. Data collection of the following data items were obtained during perioperative stay: epidural placement, duration of the operation, type of operation, systemic hemodynamic variables, routine laboratory results and fluid balance. The fluid balance was monitored between the beginning of the surgery until free drinking advise that was given by the attending physician, as part of the local hospital protocol. After discharge, the following data were obtained: length of stay, Clavien-Dindo classification of surgical complications, 30-days and 90-days mortality and histopathological diagnosis. Clavien-Dindo classification is a general surgical classification for postoperative complications, with grade I and II complications can be corrected by noninvasive intervention [19]. While grade III or higher are more severe complications that only could be resolved by invasive intervention. In our study a complication was scored as clinically significant if it was as Clavien-Dindo III or higher.

2.5. Bias

In order to exclude patient selection bias, all patients scheduled for pancreatico-biliary resection were asked to participate in this study. There was no deviation from the protocolled treatment for this patient in the ward, besides the measurement of the microcirculation.

2.6. Study size

As this is a study conducted to evaluate if there are microcirculatory changes perioperative, no formal sample size calculations have been done.

2.7. Quantitative variables and statistical methods

Baseline characteristics were calculated using descriptive statistics. The Shapiro-Wilkinson test and distribution plots are used to verify the normality of distribution of continuous variables. Continuous normally distributed data have been expressed as mean with their standard deviation. Continuous not normally distributed variables have been shown as median with their interquartile range. The data from different time points were analyzed using a linear model for repeated measurements. Furthermore, a Student’s t test or Mann-Whitney U test was conducted to test differences between groups for clinical outcomes. Statistical analysis was conducted using SPSS (version 21.0). A two-sided p-value <0.05 was considered as statistically significant.

3.1. Participants and descriptive data

A total of 19 patients gave their informed consent, two of those patients had a deviated time point of the video sequencing and another patient did not meet the image quality guidelines. Therefore, 16 patients where eligible for analysis. The median age was 58 years and 11 (69%) patients were male. The indication for elective abdominal surgery was for a benign disease in seven patients and (pre)malignant in nine patients. Eleven patients underwent an open pancreaticoduodenectomy, two patients an open distal pancreatectomy with splenectomy, one patient a laparoscopic distal pancreatectomy with splenectomy, one patient an extra-hepatic bile resection and one patient only adhesiolysis. Epidural analgesia was given in 15 (94%) patients. All patients received 2 liters of intravenous crystalloids in the first 24 hours postoperative. Additional fluid therapy with intravenous crystalloids was given in 6 (38%) patients. Furthermore, 10 (62%) of the patients received packed cells in the first 24 hours after surgery. All patients characteristics are shown in Table 1.

A linear regression for repeated measurements showed a significant difference in systolic blood between 24 hours postoperative (T1) and 4 days postoperative (T2) (138 (95% CI 128-148) vs 117 (95% CI 105-129) mm Hg, p=0.01). Furthermore a significant difference after 24 hours postoperative (T1) compared to baseline (T0) was found for diastolic blood pressure (67 (95% CI 59-74) vs. 81 (95% CI 73-89) mm Hg, p=0.02) and for heart rate (87 (95% CI 79-94) vs. 76 (95% CI 68-83) strokes/minute, p=0.001). Additionally mean arterial pressure showed a significant difference between the time points 24 hours postoperative (T1) compared to baseline (T0) (83 (95% CI 75-92) vs. 99 (95% CI 90-107) mm Hg, p=0.02) and between 24 hours postoperative (T1) and 4 days postoperative (T2) (83 (95% CI 75-92) vs. 97 (95% CI 89-105) mm Hg, p=0.03). In a multivariate analysis fluid balance was associated with the significant differences of systemic blood pressure (p=0.03) and mean arterial pressure (p=0.03) between 24 hours postoperative (T1) and 4 days postoperative (T2).

Video sequences were performed in three different time points, and per time point three movies were obtained, except in one patient who died before the third time point and another who was discharged before the third time point. Therefore, in total 138 images were analyzed by an independent video analyst. A linear regression for repeated measurements showed a significant difference in the TVD of small vessels and all vessels after 24 hours postoperative (T1) compared to baseline (T0) (TVDs mean 22.0 (95% CI 19.2 -24.9) vs 24.8 (95% CI 22.1 -27.4) mm/mm2, p=0.046; TVDa mean 26.6 (95% CI 23.8 – 29.4) vs 23.8 (95% CI 21.0 – 26.6) mm/mm2, p=0.02) (Fig. 1). Furthermore, a significant difference was found in MFI small vessels between baseline (T0) and 4 days postoperative (T2) (p=0.02), and between 24 hours postoperative (T1) and 4 days postoperative (T2) (p=0.047). Same significant difference was seen in MFI all vessels between baseline (T0) and 4 days postoperative (T2) (p=0.04), but not between the other time points. There was no significant difference in PVD or PPV for small or all vessels in between the three time points. All the microcirculatory differences over the different time points where independent from fluid balance. All data are shown in Table 2.

OPEN IN VIEWER

Fig.1

Boxplots representing medians with interquartile range, whiskers extent to the most and least extreme scores of TVD small vessels (left) and TVD all vessels (right).

OPEN IN VIEWER

Fig.2

Boxplots of the difference in PVD and PPV in small and all vessels in the first 24 hours perioperative. Boxplots representing median with interquartile range, whiskers to the most and least extreme score.

A total of 8 (50%) patients had a perioperative complication; six of them had a Clavien-Dindo III or higher complication. These Clavien-Dindo III or higher complications were intra-abdominal abscess in three patients, pancreatic fistula in one patient, bile leakage in one patient and liver necrosis in one patient. There was no difference in the TVD and PVD of small and all vessels between the patients with these clinical significant complications and without these complications on baseline or 24 hours postoperative. In addition no significant differences were found in systolic blood pressure, diastolic blood pressure, heart rate or mean arterial pressure for these significant complications between baseline (T0) or 24 hours postoperative (T1). However at baseline patients with significant complications had a higher PPV in small vessels (97.4% vs. 84.2 %, p=0.02), a higher PPV in all vessels (97.5 vs. 85.0, p=0.002) and a higher MFI in all vessels (3.0 vs. 2.8, p=0.005). Furthermore, the difference between 24 hours postoperative and baseline in patients without clinical significant complication and with gave a significant difference for PVD small vessels (delta PVDs mean -0.1 vs -8.2, p=0.01), PVD all vessels (delta PVDa mean -0.1 vs -7.4, p=0.01), PPV small vessels (delta PPVs median 7.0 vs -15.5, p=-0.01), PPV all vessels (delta PPVa median 6.4 vs -15.0, p=0.01) and MFI all vessels (delta MFIa mean 0.11 vs -0.18, p=0.01) (Fig. 2). For the systemic hemodynamic parameters only heart rate showed a significant difference between 24 hours postoperative and baseline in patients without significant complications and with (delta HR mean 5 vs 21, p=0.002). All data are presented in Table 3.

4.1. Key results

There is a growing evidence that goal-directed fluid therapy can play a major factor in the postoperative course after pancreatico-biliary resections.(5, 7, 10, 20) This study tried to identify microcirculatory parameters that change during the perioperative course after pancreatico-biliary surgery and therefore could be utilized as goal-directed fluid therapy targets. First of all there were significant differences in the TVD of small and all vessels 24 hours postoperative compared to baseline measurement. This could be explained by a “third space” fluid loss. These third space deficits are usually seen directly postoperative with fluids accumulating in the interstitial space [21]. However, these differences were minimal in terms of clinical meaning (TVDs mean 22.0 vs 24.8 and TVDa mean 26.6 vs 23.8). Furthermore, a significant difference was found in MFI small vessels and all vessels between 4 days postoperative and baseline and in MFI small vessels between 4 days postoperative and 24 hours postoperative. In addition, there were several differences seen in the systemic hemodynamic parameters between the different time points. Nevertheless, all these differences were clinically not meaningful as the differences were too small to detect by eye. In order to identify patients that would benefit from goal-directed fluid therapy we found a clinical significant difference for PVD and PPV small and all vessels. The change between preoperative measurement and 24 hours after surgery in these two parameters showed a great indicator for the development of clinical significant complications. Patients with no clinical significant complication (Clavien-Dindo <III) had almost no change in PVD and a positive change in PPV over the course of 24 hours after surgery. In the contrary, patients with clinical significant complication (Clavien-Dindo III) showed a decrease in PVD and PPV over the first 24 hours after surgery. The relation between PVD and PPV with more complications can be explained by the loss of coherence which can occur between micro-and macrocirculation [22]. This can occur when there is an increased need for oxygenated red blood cells (PPV) to be delivered to the microcirculation while this does not occur due to generalized postoperative stress causing an impairment in the function of the microcirculation to deliver this extra demand. These are important indicators as it can help to identify and treat patients that will develop a significant clinical complication. The assessment of the sublingual microcirculation can be used to identify these complications earlier. Furthermore, the microcirculation can be used as a target in the goal-directed fluid therapy

Several intensive care unit studies have shown the benefit of microcirculatory guided fluid therapy [15, 23–26]. Although, these clinical implications of this microcirculatory monitoring and microcirculatory targeted fluid therapy need to be addressed in prospective trials as this is yet to be addressed in surgical ward patients. The design of these studies should first focus on the PVD and PPV goal-directed fluid therapy, as with our study we could not take any conclusion about the responsiveness of these parameters to fluid therapy. In advanced and probably much larger studies, this microcirculatory guided interventions, be it fluid therapy or others, should be examined to the extent of reducing postoperative complications in pancreatic surgery.

4.2. Limitation

This study has limitations as the sample size was small. Furthermore, there is in some extent of heterogeneity in surgical procedures. Not all patients eventually underwent a pancreatic resection and not all procedures have the same approach (open vs laparoscopic). Although, all patients were scheduled for a pancreatic resection this indication was diverted during operation as one patient had an oncological radical margin without pancreatic resection and the other had a difficult procedure where the attending surgeon made the call to stop the surgery. However, these different procedures and approaches did not seem to alter the microcirculatory changes. Moreover, the different time points of the microcirculatory monitoring are limited and ideally a more detailed monitoring with daily measurements, especially during operation can give more insights in the timing of goal-directed fluid therapy. Likewise, a more detailed and longer fluid balance registration could have provided more information on the influence of fluid balance on microcirculation. This prolonged registration and other additional registrations were avoided as this was a prospective observational study that studied the standard patient care with the minimal addition of bedside sublingual microcirculation measurements.

4.3. Interpretation and generalizability

In conclusion, there are perioperative microcirculatory changes after open pancreatico-biliary surgery. These changes could be utilized in the future to identify perioperative complications and intervene at an earlier stage. The most significant microcirculatory indicator for a perioperative complicated course is the change of PVD and PPV. If these two indicators decrease over the course of the first 24 hours postoperative, there is very high chance for a complicated course after pancreatico-biliary surgery.