The impact of nitroglycerine and volume on gastric tube microperfusion assessed by indocyanine green fluorescence imaging

The influence of hypervolemia and intraoperative administration of nitroglycerine on gastric tube microperfusion remains unclear The present study aimed to investigate the impact of different hemodynamic settings on gastric tube microperfusion quantified by fluorescence imaging with Indocyanine green (ICG-FI) as a promising tool for perfusion evaluation. Three groups with seven pigs each were formed using noradrenaline, nitroglycerin, and hypervolemia for hemodynamic management, respectively. ICG-FI, hemodynamic parameters, and transit-time flow measurement (TTFM) in the right gastroepiploic artery were continuously assessed. Fluorescent microspheres (FM) were administered, and the partial pressure of tissue oxygen was quantified. The administration of nitroglycerine and hypervolemia were both associated with significantly impaired microperfusion compared to the noradrenaline group quantified by ICG-FI. Even the most minor differences in microperfusion could be sufficiently predicted which, however, could not be represented by the mean arterial pressure measurement. Histopathological findings supported these results with a higher degree of epithelial damage in areas with impaired perfusion. The values measured by ICG-FI significantly correlated with the FM measurement. Using tissue oxygenation and TTFM for perfusion measurement, changes in microperfusion could not be comprehended. Our results support current clinical practice with restrictive volume and catecholamine administration in major surgery. Hypervolemia and continuous administration of nitroglycerine should be avoided.

Esophagectomy with reconstruction via gastric tube pull-up remains the most successful treatment in nonmetastatic esophageal carcinoma 1 . However, the formation of gastroesophageal anastomosis can be complicated, resulting in several life-threatening events such as leakage with mediastinitis, sepsis, and lethal outcomes [2][3][4] . Leakage occurrence has been associated with reduced microvascular blood flow and consecutive hypoxemia 5,6 . Due to the surgical technique of gastric tube formation with ligation of all gastric vessels except the right gastroepiploic artery and its arcade, a decrease in microperfusion can be observed mainly in the upper part of the gastric tube 7,8 .
Up to now, several strategies to improve the gastric tube microperfusion [9][10][11][12] or to reduce venous congestion have been reported [13][14][15] . However, there is increasing evidence that excessive intravenous fluid administration has a relevant impact on developing anastomotic complications 16 .
Indocyanine green fluorescence imaging (ICG-FI) is a promising tool for assessing and quantifying tissue microperfusion. Several retrospective clinical studies indicate a benefit of the technology in terms of anastomotic leakage rate [17][18][19] . Recently, we validated the ICG-FI by assessing gastric tube perfusion in a porcine model. Perfusion could be predicted by using the slope of fluorescence intensity (SFI), background subtracted the Institute for Surgical Research at the University Center Hamburg-Eppendorf (Hamburg, Germany).
A 6 F arterial catheter was introduced to the right carotid artery for permanent assessment of heart frequency (HF), and median arterial blood pressure (MAP), as well as a central venous catheter into the right jugular vein for delivery of infusions, heparin, and ICG as well as the measurement of central venous pressure (CVP). A 4F arterial catheter with an embedded thermistor was inserted into the right femoral artery for continuous hemodynamic monitoring of stroke volume, cardiac output, systemic vascular resistance (SVR), and global end-diastolic volume (GEDV) by the determination of thermodilution. These parameters were documented by the PiCCO device (Pulsion Medical Systems, Munich, Germany) based on arterial pulse contour analysis. For the FM application, a catheter was placed in the left atrium. A pigtail catheter was inserted into the abdominal aorta via the right femoral artery to withdraw reference blood samples with a known constant withdrawal rate to calculate the flow rate per tissue sample 7 .
First, a midline laparotomy was performed. A catheter was placed into the bladder for urinary drainage. The gastric tube formation was performed under ligation of all arteries despite the gastroepiploic artery and its arcade. The dissection of the minor curvature was performed using Endo-GIA (Covidien, black cartridge), resulting in a gastric tube with a three-centimeter diameter. In this way, the perfusion depended on a single vessel leading to a standardized "one-vessel-model". Three regions of interest (ROI) were defined: fundus (D1), corpus (D2), and prepyloric (D3). Figure 1 shows a macroscopic and an associated fluorescence angiographic image of the gastric tube with marked ROIs in the defined areas. After that, a flexible polarographic measuring probe (Licox,  Baseline measurements were done after completing the preparation. The hemodynamic, metabolic, and tissue oxygenation parameters were determined after blood pressure modulation. Mean arterial blood pressure (MAP) was modified by using norepinephrine and propofol, resulting in 3 measurements: hypotension (MAP < 60 mmHg, T1), normotension (MAP 60-90 mmHg, T2), and hypertension (MAP > 90 mmHg, T3) 7 . Animals were randomly assigned to the different treatment groups: Control (control; n = 7), Nitroglycerine (Nitro; n = 7), and hypervolemia (Vol; n = 7). As the administration of catecholamines represents the standard of hemodynamic management in clinical praxis, calculated values of the control group represent the reference values when comparing the different groups. Nitroglycerin was applicated by continuous intravenous administration of nitroglycerin (3.5 µg/kg/min). Hypervolemia was defined by a global end-diastolic volume (GEDV) > 700 ml and at least 5 L of crystalloids and colloids (ratio 2:1). Every time the target MAP was reached, a 30-min stabilization period was respected. ICG-FI and FM were performed at each measurement. Hemodynamic parameters, TTFM, and tPO 2, were captured for each measurement 7 . After finishing the measures, the animals were euthanized by a veterinarian using T 61 under deep anesthesia. The experimental setting is shown in simplified form in Fig. 2. Fluorescent microspheres. FMs of 15 μm in diameter (Molecular Probes, Eugene, Ore) were used to measure the median blood flow (ml/min/g) at each measurement. Microspheres labeled with different fluorescent colors were randomly selected for the application. Excitation and emission wavelengths for each of the fluorescence microspheres were as follows: blue, 356/424 nm; blue-green, 427/468 nm; yellow-green, 495/505 nm; orange, 534/554 nm; red, 570/598 nm; crimson, 612/638 nm; and scarlet, 651/680 nm. They did not interfere with the excitation and emission wavelength of ICG. Each vial of FMs was placed in an ultrasonic water bath for 5-10 min to disperse the microspheres and vortexed twice for 3 min to ensure proper mixing before injection. Approximately 3.33 × 10 6 FMs were suspended in physiological saline solution to a volume of 10 mL and constantly injected over 60 s into the left atrium at each measurement. Reference blood samples were withdrawn in anticoagulated (5 mL of 3.13% sodium citrate) syringes with a constant-rate withdrawal pump at 3.18 mL/ min over 3 min. Injection of FMs was started when the withdrawn blood reached the suction syringe. At the end of the experiments, the gastric tube was excised and fixed in 10% formaldehyde for 6-8 days. Afterwards it was dissected into 28 tissue pieces with a mean weight of 3.5 ± 0.2 g 7 . According to the standard method published by Glenny et al., the tissue pieces were processed for blood flow determination using spectrofluorometry 21 . Fluorescence intensity (FI). Perfusion was assessed with the FI-system (LLS GmbH, Ulm, Germany).
ICG was administered intravenously through a central venous line to ensure adequate mixing. The dose was www.nature.com/scientificreports/ adapted to the body weight (0.02 mg/kg body weight). The bowel was illuminated with near-infrared light at a wavelength of 785 nm provided by infrared laser diodes with a total output of 80 mW in a field of view of 10 cm in diameter. The fluorescence emission of the excited dye was detected by an infrared-sensitive charged-couple device camera system equipped with a band-pass filter for the selective transmission of light at the central wavelength of 830 nm. The dynamic range of the camera was 54 dB. The camera`s signals were digitized with a frame grabber card that provides a resolution of 8 bits. Images were acquired at a rate of 25 frames per second and were recorded in real-time. The charge-coupled device imaging system was positioned on the exposed surface of the gastric tube at a distance of ≈ 25 cm. The camera distance was measured and calibrated after each measurement. The FI was displayed in real-time on a computer monitor and analyzed using a digital image processing system with a temporal resolution of 20 ms and spatial resolution of ≈ 0.2 mm at a penetration depth of 4 mm 7 .
Image and data analysis. The analysis of fluorescence data has already been described in detail before 22 .
Briefly, to obtain a quantitative measure for the fluorescence intensity, we calculated both the mean value and the SD of the measured pixel intensities in the region of interest on the tissue wall for each image in a sequence of 60 s after the injection of the ICG. This region of interest was 30 × 30 pixels, corresponding to an area of ≈ 8 × 8 mm on the gastric surface, depending on the distance from the camera. To measure gastric perfusion, we calculated three different parameters derived from the time-dependent fluorescence signal 7 .
• Background-Subtracted Peak Fluorescence Intensity (BSFI) To calculate BSFI from the time-dependent fluorescence intensity, the initial intensity value before the injection of ICG was subtracted from the peak fluorescence intensity during the first passage of the dye through the gastric tube 22 . • The slope of Fluorescence Intensity (SFI) This parameter represents by the maximal slope during the increase of the time-dependent fluorescence intensity induced by the first wave of the dye, which passes the capillaries of the gastric tissue 22 . • Time to slope (TTS) TTS is defined as the time between injection of the ICG and the first fluorescent signal at the ROI. TTS of areas D2 and D1 were placed in relation to area D3 as a TTS ratio 7 .
Each ICG-FI sequence was recorded online for 120 s with real-time digitizing. BSFI, SFI, and TTS data were analyzed offline directly after the images were recorded with a customized software package "Meteroarchive VCL LLS Fluoreszenzangiographie V 1.0" (LLS GmbH, Ulm, Germany) 7 . A scheme of the calculation of the parameters is shown in Fig. 3.

Histopathology.
Representative specimens of the gastric tube ROIs were assessed and stored in 3.5% buffered formalin. Afterwards, all samples were routinely processed and embedded in paraffin, and 5-µm slices were stained with hematoxylin and eosin 7 . The slices were examined by an experienced pathologist (blinded to the treatment) using an established scoring system 23 . Mucosa taken from the anterior gastric corpus was divided into the following grades under a light microscope: grade 0: normal gastric mucosa; grade 1: surface mucosa cells were damaged; grade 2: in addition to extensive luminal damage, cells lining the gastric pits were also disrupted and exfoliated; grade 3: cell destruction extended into the gastric gland 23 . Statistics. Data from all animals were expressed as mean ± SD. Comparisons of hemodynamics, TTFM, tPO 2, FM, and ICG-FI data among the hemodynamic states were performed by ANOVA. Data for SFIcor, BSFI-

Results
ICG-FI images of the gastric tube perfusion were obtained in all 21 animals. Systemic hemodynamic parameters were recorded continuously and are summarized in Table 1. The predefined MAP level could be achieved in all groups. Gastric Fig. 6. The control group showed an increasing flow with rising MAP. Hypervolemia also resulted in increasing flow at normotension while hypertension showed decreasing flow. Substitution of nitroglycerin in group 2 resulted in no artery flow changes. There were no significant differences between the groups and the time points of measurement.

TTFM-flow. Flow in the gastro-epiploic artery measured by TTFM is shown in
Tissue oxygenation. The local tissue oxygenation in the control group showed significantly decreasing values from the prepyloric region to the tip. In group 2 area D1 had a significantly decreased oxygenation compared to D2 and D3. Application of volume led to the highest tissue oxygenation in area D2. There were no significant changes within an area at the different measurement times. The tissue oxygenation of all groups is shown as box plots in Fig. 7.

Discussion
Anastomotic leakage remains a life-threatening complication after esophagectomy 24 . Due to the singular vascular supply of the gastric tube, impaired perfusion should be emphasized as an independent risk factor 25 . Up to now, various techniques for intraoperative measurement of local tissue perfusion have been described 26,27 . Among these, ICG-FI is a promising tool for a real-time, non-invasive assessment of gastric tube perfusion 28 . Several retrospective studies using ICG-FI during esophagectomy have reported a reduction of the anastomotic leakage rate of up to 69% compared to standard-of-care procedures [29][30][31][32][33] . Previously, we validated the quantification of ICG-FI by using fluorescent microspheres as the gold standard for experimental perfusion assessment. We were the first to show that ICG-FI can sufficiently predict the local tissue perfusion of the gastric tube by using the described quantification parameters 7 . Moreover, in an additional study on intraoperative evaluation of gastric tube perfusion in patients undergoing esophagectomy we quantified the perfusion using SFI, BSFI and TTS. In fact, perfusion reduction of up to 32% was not associated with anastomotic leakage 20 . The present study aimed was to investigate the influence of hypervolemia and continuous administration of nitroglycerine on gastric tube perfusion quantified by ICG-FI and FM. In our recent study, we objectified two key aspects of gastric tube microperfusion. By establishing three different settings of hemodynamic management according to different intraoperative settings, we have observed the influence of hypervolemia and continuous administration of nitroglycerine on the gastric tube perfusion. FM, as the gold standard of experimental tissue perfusion measurement, demonstrated a significantly reduced microperfusion of the whole gastric tube in the setting of hypervolemia or after administration of nitroglycerine compared to the control group treated with catecholamines. In addition, there was a loss of hemodynamic coherence resulting in increased systemic MAP without improving the gastric tube microperfusion. Furthermore, hypervolemia was associated with a reduction of perfusion level compared to normotension, whereas administration of nitroglycerine resulted in no significant changes in microperfusion levels with increasing MAP. TTFM showed a comparable pattern of flow changes, which, however, were not statistically significant. The local tpO 2 was consistent with the FM results of group 2 without significant changes of microperfusion between the measurements. However, the perfusion changes in group 3 could not be detected by tPO 2 measurement. Therefore, TTFM and tPO 2 are inappropriate for the evaluation of microperfusion. Moreover, histopathological findings supported these results with a higher degree of epithelial damage after administration of nitroglycerine or volume, especially in the corpus and at the tip of the gastric tube.
In addition, ICG-FI has been validated as a suitable technology for evaluating gastric tube microperfusion by confirming the FM results. While MAP as the standard parameter for intraoperative and postoperative hemodynamic management was comparable in all groups at the different measurements, ICG-FI, especially SFI, was able to confirm the changes of microperfusion not only between different MAP levels, but also between the groups with different treatment at identical MAP. Furthermore, SFI and BSFI correlated significantly with FM results underlining the consistency of the two methods for evaluating microperfusion. Even though FM represents the gold standard in experimental microperfusion measurement, it remains a technically non-trivial, invasive, and, in terms of instrumentation and interpretation, complex and time-consuming method. The perfusion evaluation with FM is only suitable for experimental animal settings. In comparison, ICG-FI is a non-invasive technology, which can be evaluated, and interpreted immediately. In addition, ICG-FI can be used clinically on humans to verify animal experimental results.
In contrast to BSFI and SFI, TTS-ratio could only reliably show the perfusion changes within the respective groups. This is mainly due to the calculation of the ratio to the prepyloric region as the baseline, since the delay in the first fluorescence signal between the groups was not considered.
Several strategies for perioperative hemodynamic management have been published. The use of catecholamines as a standard method for perioperative blood pressure management and their influence on the perfusion of the gastric tube are controversially discussed. Theodorou et al. observed an increasing blood lactate level after hemorrhagic shock and consecutive administration of norepinephrine in an experimental porcine model. They concluded that vasoconstriction reduced the gastric tube microperfusion 34 . In contrast, Al-Rawi et al. demonstrated a sufficient increase of microcirculation of the gastric tube after administration of catecholamine in 12 patients 35 . Moreover, in a recently published retrospective analysis of 494 patients undergoing esophagectomy, intraoperative vasopressor use was not associated with anastomotic leakage 36 . Our results support the use of catecholamine administration for hemodynamic management in patients undergoing esophagectomy and reconstruction with gastric tube pull-up. The control group showed significantly better perfusion levels, and increasing MAP was associated with perfusion improvement.
As a hypothesis, nitroglycerine as a vasodilator agent might reduce venous congestion in the gastric tube. In a porcine model, van Bommel et al. investigated the effect of continuous application of nitroglycerine during gastric tube formation. Nitroglycerine could improve the microperfusion in the corpus and fundus region compared to a control group 37 . However, in a prospective, randomized, controlled trial, nitroglycerine was administrated intravenously during gastric tube reconstruction in patients undergoing an esophagectomy. There was no significant difference regarding anastomotic leakage rate, microvascular blood flow, and microvascular hemoglobin oxygen saturation compared to the control group 38  www.nature.com/scientificreports/ their results, we could demonstrate that a continuous administration of nitroglycerine was even associated with impaired perfusion at all measurement times compared to the control group. Perioperative volume management during major surgery affects the surgical outcome 39 . In several clinical studies, hypervolemia was associated with increased rates of anastomotic leakage [40][41][42][43] . Moreover, in a prospective clinical trial comparing liberal fluid management with goal-directed fluid therapy in patients undergoing pancreaticoduodenectomy visceral tissue edema measured by CT-scan could be identified as an independent risk factor for severe surgical complications 44 . Furthermore, hypervolemia could be identified as an independent risk factor for higher postoperative morbidity in patients undergoing esophagectomy 41 . Supporting those already published data, we could demonstrate that hypervolemia is associated with significantly reduced microperfusion of the gastric tube using ICG-FI for quantification. Increased morbidity, especially anastomotic leakage, can be explained by impaired microperfusion being an independent risk factor for anastomotic leakage.
Due to the large experimental animal setting, results should be interpreted with caution regarding translation into humans. Another limitation is the small cohort size due to the smallest necessary number of test animals.
Based on our results, we conclude that hypervolemia and continuous application of nitroglycerine are associated with impaired microperfusion of the gastric tube as measured by ICG-FI and FM compared to hemodynamic management using norepinephrine. Our results support current clinical practice with restrictive volume and catecholamine administration in major surgery. Hypervolemia should be avoided in esophagectomy with gastric tube pull-up. Furthermore, continuous administration of nitroglycerine is not suitable for improving microperfusion of the gastric tube. In addition, an evaluation and valid quantification of gastric tube microperfusion using ICG-FI is possible by determining SFI and BSFI. The smallest differences in microperfusion could be sufficiently predicted, whichcould not be represented by the MAP measurement. However, further validation of ICG-FI with prospective clinical trials and survival studies is needed.

Data availability
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.