An Open-Source Normothermic Perfusion System Designed for Research Scientists

摘要

Here, we present a protocol for an open-source, modifiable normothermic ex vivo perfusion system designed to overcome the high cost and limited flexibility of existing platforms, enabling more accessible, cost-efficient, and iterative use in pre-clinical biomedical research.

摘要

The integration of ex vivo perfusion machines for human organs prior to transplantation has improved outcomes for recipients and increased organ availability while revealing novel avenues of investigation for translational medicine. However, these machines have limited availability for experimental modifications while presenting a significant upfront cost investment and prohibitive per-use cost to the research scientist. Furthermore, there is a significant need for improved pre-clinical models in biomedical research that allow for authentic interrogation of cellular processes in the multi-cellular organ setting.

To provide access to similar technology for the greater research community, we constructed a cost-effective modular normothermic perfusion machine entirely from readily sourced parts run with user-modifiable Python code. Since the system is devoid of customized hardware, investigators can construct and augment the system as required for individualized research applications.

Herein, we describe the machine set-up and perfusion process in detail, convey our experience with porcine and human organs, and provide early proof-of-concept experimental results. These studies demonstrate the potential to interrogate pathophysiology and complex cellular processes using intact human tissue.

引言

Although short-duration (<12 h) normothermic machine perfusion has been previously employed for kidneys and livers prior to transplant^1,2,3,4,5, prolonged normothermic ex vivo liver perfusion became a reality in 2020 with the Liver4Life machine⁶. The intent of this technology was geared toward the expansion of viable donor organs, but the possible translational applications were readily apparent to the research community^{7,8,9,10,11,12}. For bench scientists and conventional research laboratories, these clinical tools are almost assuredly out of reach due to significant upfront and recurring costs^12,13. Clinical normothermic perfusion machines are upwards of $250,000 USD, with single-use disposables ranging from $9,500-$60,000 USD per run^{13,14,15,16,17,18,19,20,21,22,23}. Furthermore, proprietary rights and the inability to tailor the existing technology make scientific applications challenging.

The outstanding need for improved pre-clinical models^24,25,26 in biomedical research, coupled with the aforementioned limitations, inspired the creation of an open-source, modifiable perfusion system (Figure 1). Circuit design and development were guided by the principles of cost-utilization, laboratory space requirements, and mobility for translational applications. The system can be constructed for $48,000-$66,000 USD, and each perfusion cost ranges from $1,000-$2,300 USD, which was achieved through the integration of readily sourced individual components run by freely downloadable code. The system was optimized through an iterative process making use of donor porcine blood and expired human blood, porcine livers, and partial human liver segments from the operating room (Supplementary Table 1, Supplementary Table 2, and Supplementary Table 3). Organs are housed in an enclosed basin that optimally positions the specimen to minimize pressure necrosis, allow for ideal cannulation trajectory, drain runoff, and ensure normothermia. The reservoir houses approximately 2-3 L of perfusate prior to being directed toward inflow circuits. Hemodialysis, which is critical for electrolyte homeostasis and hemoconcentration in hepatic perfusions^6,27,28,29, was integrated through a side circuit with an independent roller pump. Pharmacologic infusions support physiologic functions (e.g., total parenteral nutrition (TPN), bile salts, etc.³⁰), augment vascular tone (epoprostenol, phenylephrine), prevent coagulation (enoxaparin), and infection (piperacillin-tazobactam, amphotericin). Physiologic perfusion parameters are ensured using real-time in-line sensors to obtain blood gas, pressure, and flow values. The system allows for maximal user adaptability such that the integration of hardware components can occur at any point in the circuit. Furthermore, each of the individual components within the circuit can be interchanged or modified to enable tailoring to specific research questions and experimental needs. For example, the dual inflow circuit can be altered for perfusion of organs with a single inflow (i.e., kidneys or pancreas) by simply removing the Y-connector prior to the pump heads, and claves for infusion/sampling can be installed at any point within the circuit. Additionally, the system is run with Python code that allows semi-automated perfusion based on a specified set of perfusion parameters dictated by the end user (Supplementary Figure 1). For example, sensors acquire continuous arterial blood gasses (ABG), pressure, and flow readings that act as input to implement output commands to automatically adjust oxygenator gas flow, pump speed, etc. The graphic user interface (GUI) allows for manual override of this automation if desired. The customizability of the circuit is its hallmark, and it broadens the overall applicability of this system for a wide range of experimental investigations.

Access to the system in its entirety is freely available (https://github.com/CCR-SOP/OpenSourcePerfusionSystem.git) to encourage collaboration and innovation in an effort to forge new depths of knowledge across multiple fields of translational science. 

Figure 1: Open-source perfusion system. (A) Open-source perfusion system schematic for dual inflow set-up for hepatic perfusion. (B) Three-dimensional (3D) rendering of the system during organ perfusion with software integration as demonstrated on screens. (C) Photo of perfusion machine during porcine perfusion experiment. The exposed configuration in the figure is for demonstrative purposes only. During perfusion, the sterile warming drape covers the organ to maintain a sterile enclosure.P/F Sensor: Pressure and flow sensors. HA: Hepatic artery. PV: Portal vein. HV: Hepatic vein. BD: Bile duct. Please click here to view a larger version of this figure.

方案

This protocol was conducted in accordance with the ethical standards of our institution's Human Research Ethics Committee and adheres to all applicable guidelines and regulations for research involving human subjects and tissues.

1. Machine set-up

1. Table preparation
   1. Sterilize the tabletop and mobile cart with 70% alcohol spray and wipe thoroughly. Place the mobile cart perpendicular to the left side of the table. Prior to the connection of all tubing and hardware throughout the protocol, spray with 70% alcohol or autoclave, if applicable, to sterilize.
      NOTE: Prior to any organ handling or intervention, use sterile gloves and alcohol spray to facilitate sterility prior to and throughout perfusion.
   2. Cover the workstation with sterile blue drapes and then absorbent underpads.
   3. Attach the reservoir holder to the cart leg closest to the bench with the ring projecting outward, approximately 12" below the bottom of the workstation top.
   4. Place the drum funnel on the right aspect of the workstation with a spout overhanging the front edge of the table. Cut out circular memory foam to fill the base of the drum, allowing a 4" gap for outflow near the spout. Place the diaphragm underneath foam connected to a respirator.
   5. Obtain 2 water heating blankets and place them on top of the foam insert with connector tubing oriented to the top left of the blanket. Cut off white clamps from the tubing and insert a 3/16" male Luer barb into each tube. Secure 1 inflow and 1 outflow line from the Y'ed water heating tank lines to each blanket.
   6. Obtain a sterile warmer drape and identify the center blue square. Fold in half twice and cut 2 mm off the corner of the fold to create an outflow hole. Place sterile drape on top of drum so that cut hole is in line with outflow spout.
   7. Obtain the 3D-printed outlet connector and place the magnet in the groove. Place under drape through outflow spout and align cut hole to hole in connector and magnet. Place a second magnet aligned with the outflow to secure the drape.
   8. Raise excess drape to cover the entire drum funnel and secure with rubber band/elastic ties. Fold the excess drape over the top to create a sterile barrier. Replace the second water jacket on top of the folded drape.
   9. Cut a ~30" piece of 1/4" tubing and attach the proximal aspect to the outlet piece from the bottom of the drum funnel spout and the distal aspect to a reservoir inflow barb.
2. Pump and oxygenator preparation
   1. Place pumps on mobile cart closest to workstation edge with stand and two clamps immediately behind pumps so that clamps are 6" above pumps.
   2. Obtain 2 pump heads and secure in pumps.
   3. Obtain 2 oxygenators and place within clamps attached to stands in a horizontal configuration. Orient oxygenators with barb facing outward and lateral barb pointing to the right.
      NOTE: Oxygenators with Luer connectors require an alternative configuration with the label facing outward and inner prongs facing upward.
3. Gas mixer preparation
   1. Cut two (2) 4' pieces of 1/8" tubing and insert a male 1/8" Luer into the ends of each tube. Connect one tube to each arterial and venous gas inflows.
   2. Cut two (2) 2" pieces of 1/4" tubing and attach a 1/4" female Luer onto one end. Fit each of these tubing pieces on the top right barbs of each oxygenator.
   3. Connect the arterial and venous 1/8" gas inflow tubing to the top right Luer connector of their respective oxygenators.
4. Hemodialysis circuit preparation
   1. Remove all tubing from the hemodialysis kit except that leading into the filter directly (blue, red, green, and yellow lines). Secure the filter onto the pegboard at the back of the workstation, ~12" to the right of the pumps, or in line with the reservoir.
   2. Place 1 large roller pump to the immediate right of the filter for blood inflow. Place 2 smaller peristaltic pumps to the right of the blood roller pump.
      1. Obtain two (2) 1/8^th peristaltic tubings and place within the white pump clamp. Attach two (2) female 1/8^th Luer barbs at each end of each tubing.
   3. Hang a 5 L bag of dialysate on the right lateral aspect of the workstation.
      1. Connect the dialysate bag to the hemodialysis filter inflow by attaching 1/8^th tubing to the bag's Luer connector. Attach the male 1/8th Luer to the end of this tubing and attach it to the inflow peristaltic pump tubing.
   4. Dialysate Inflow/Outflow: Obtain 3 1/8^th tubing with each end equipped with a male 1/8^th Luer.
      1. Inflow: Connect proximal end to distal aspect of pump tubing and distal end to yellow inflow tubing (with 1/8th female Luer) to hemodialysis filter.
      2. Outflow: Connect proximal end to green outflow tubing (with 1/8th female Luer) and distal end to outflow peristaltic pump tubing. Connect the 1/8th tubing to the distal aspect of the peristaltic pump and drain it into the waste bag via Luer connectors.
5. Proximal circuit creation
   1. Place reservoir within reservoir holder with exterior numbers and yellow-capped connector oriented outward.
   2. Cut ~36" of 5/16" tubing and connect the proximal end to the outflow spicket at the bottom of the reservoir. Connect the 5/16" Y connector to the distal end of this tubing.
   3. Cut two (2) 4" lengths of the 5/16" tubing and connect these to the end of the Y connector. Attach each of the ends of these 4" tubes onto the horizontal prong of each of the pump heads.
   4. Cut two (2) 4" lengths of 1/4" tubing and connect 1 to each of the vertical barbs of the pump heads. Connect the distal aspect to the left lower prong of each oxygenator.
   5. Cut two (2) 2" lengths of 1/4" tubing and connect a female Luer to one end. Add a male Luer cap and attach it to the right lower prong of each oxygenator.
6. Venous circuit creation
   1. Cut one (1) 30" length of 5/16" tubing and attach the proximal end to the outflow (right) barb of the venous oxygenator. Cut in half and place a 5/16" Y connector overthe distal cut end. Reattach the other half to the top barb of the Y connector.
   2. Side sampling circuit: Cut one (1) 2" piece of 1/4" tubing and stretch the proximal end using a needle driver. Connect this to the other barb of the 5/16" Y connector.
      1. Obtain a cuvette sensor. Connect the barbed side to the 2" tubing connected to the Y connector. Cut off 1-2" of distal cuvette tubing and connect to the male 1/4" Luer. Attach to the proximal end of the shunt sensor once calibrated, and on the distal aspect, connect the male 1/8" Luer to the 1/8" tubing.
      2. Connect the horizontal aspect of the female T connector to 1/8" tubing and male Luer. To the perpendicular aspect, connect a sampling clave. Connect a 1/8" male Luer to the remaining horizontal aspect of the T connector.
      3. Cut 12" of 1/8" tubing, attach the proximal end to the sampling clave, and attach a 1/8" male Luer to the distal end. Connect this to another female T connector. To one side of the female T connector, attach a 1/4" male Luer and connect to an inflow channel using 2" of 1/4" tubing.
   3. Return to the distal end of the 5/16" tubing and attach another 5/16" Y connector. Obtain ~55" of ¼" tubing and attach to the top end of the 5/16" Y connector.
   4. Hemodialysis blood circuit creation: Attach a 1/4" male Luer to the distal end of the 1/4" tubing. Place the tubing through the large roller pump. Connect to the blood inflow at the top of the hemodialysis filter (blue-lined tubing).
      1. Cut one (1) 24" piece of 1/4" tubing and connect a male 1/4" Luer to the proximal aspect. Connect this tubing to the blood outflow hemodialysis filter (red-lined tubing). Connect the distal aspect to an inflow barb of the reservoir.
   5. Return to complete the venous inflow circuit leading to the specimen. Cut one (1) 36" piece of 1/4" tubing and stretch it to fit onto the barb of 5/16" Y connector.
   6. Drug infusion site: Obtain 3 four-way stopcocks and connect linearly. Connect this piece to the perpendicular arm of a female T connector. To the in-line arms of the female T connector, attach two male 1/4" Luers. Insert piece 12" from the left aspect of the tubing.
   7. Attach the external flow sensors onto the 1/4" tubing after the drug infusion site. Obtain 1 9" mount and secure at the level of the drum funnel. Make a cut 2" distal to the end of flow sensors and insert pressure sensors with barbs into 1/4" tubing. Place the distal aspect of the tubing into the center of the drum funnel. Reserve one (1) 1/4"male Luer for venous cannulation into the specimen.
7. Arterial circuit creation
   1. Cut another ~30" piece of 5/16" tubing and attach the proximal end of the right outflow barb of the arterial oxygenator. Cut this tubing approximately halfway through and attach one 1 5/16" Y connector.
   2. Side sampling circuit: Create a duplicate side sampling circuit as described in step 1.6.2 and attach it to the inferior aspect of the Y connector, omitting the cuvette. When completed, attach the outflow 1/8" Luer to the T connector.
   3. Cut one (1) 36" piece of 1/4" tubing and fit it onto the barb of 5/16" Y connector.
   4. Repeat step 1.6.7. Reserve one (1) ¼" male Luer for arterial cannulation into the specimen.
8. Venous outflow creation
   1. Cut ~30" of 1/4" tubing and attach 1 male ¼" Luer to the proximal end to attach to the inferior vena cava (IVC) outflow of the specimen. Attach the distal end to one of the inflow barbs of the reservoir.
   2. Cut this tubing ~6" from the reservoir and attach 2 1/4" male Luers to each end. Insert a female T with a clave attached to the perpendicular aspect for sampling.
9. Introduce perfusate to the circuit as described in step 4. Rewarm and connect the specimen as described in step 6.1.4. or step 6.2.4. and commence perfusion.

2. CDI sensor calibration

1. Ensure the calibrator gas bottles are full and installed per the user operating manual and that the monitor is set to calibrate mode. Calibrate per manufacturer's instructions.
2. Disconnect the sensor-cable-head apparatus from the monitor once calibration is complete and immediately integrate it into the respective venous and arterial side circuits. Luer connections are created in steps 1.6.2 and 1.7.2.

3. Code Initiation

1. Download, install, and open the Python code found in the GitHub repository. Ensure proper Python and PyCharm software installation.
2. Create a unique branch for each perfusion run for accurate data recording. Once initiated, a GUI will display with two windows. Orient one window per display for easy control of hardware and display of perfusion parameters.
3. Adjust perfusion parameters manually or automate perfusion through respective buttons displayed on the GUI (Supplementary Figure 1).

4. Perfusate creation and system priming

1. Prime the circuit approximately 2 h prior to organ perfusion to allow for correction of metabolic/electrolyte derangements, warming to normothermia, and administering anticoagulation, antibiotics, and other pharmaceuticals.
2. Porcine perfusate
   1. Obtain 2 L of type OO whole porcine blood and introduce it to the reservoir by draining through an open 50-100 mL syringe. Take approximately 1 mL of whole blood and measure baseline pH, electrolytes, and hematocrit on a point-of-care machine.
   2. Administer aliquots of 8.4% sodium bicarbonate as necessary to counteract acidity due to cold blood preservatives. Initiate hemodialysis to normalize hyperkalemia and lactic acidemia as needed and titrate to desired hematocrit concentration. Adjust gas flow to titrate to physiologic pO2 and pCO2 levels.
   3. Administer pharmaceuticals as described below (section 5).
   4. Preserve 1-2 L type OO whole blood for intermittent boluses as needed to maintain hematocrit at ~30-40% throughout the perfusion experiment.
3. Human perfusate
   1. Obtain type-matched blood products from a blood bank (near-expired or donated for research purposes).
      NOTE: Approximately 4 units each of packed red blood cells (RBCs) and plasma are required for each perfusion experiment.
   2. Wash pRBCs to remove preservatives and normalize electrolyte derangements from the RBC pellet prior to priming. Wash RBCs twice.
      1. RBC washing: Place pRBC and 0.9% NaCl in a 1:1 ratio into a centrifuge tube. Centrifuge at 1300 × g for 15 min. Remove the supernatant with a pipette and readminister fresh 0.9% NaCl in a 1:1 ratio to the RBC pellet. Centrifuge again as above and remove 2nd supernatant for final washed RBC pellet.
   3. Introduce washed RBC pellets into the reservoir using an open 50-100 mL syringe. Place 4 units of plasma into the reservoir and begin circulation.
   4. Initiate dialysis, warming blankets, and gas inflow once perfusate is circulating.
   5. Sample perfusate and apply pharmacological interventions as needed to ensure near-physiologic perfusate levels as possible prior to organ perfusion.
   6. Wash and introduce additional washed pRBC pellets and titrate volume status through hemodialysis as necessary to maintain hematocrit levels of 30-40%.

5. Pharmacologic intervention and infusion preparation

1. Antibiotic and antifungal administration
   1. Combine two 2.5 g vials of piperacillin/tazobactam in a sterile specimen cup with 40 mL of sterile saline. Add 10 mL to the reservoir as an "antibiotic prime".
      1. Dilute the remaining mixture with 60 mL of sterile saline and divide between two 50 mL syringes. Connect one syringe to a syringe pump and program to continuously deliver 41 µL/min to the portal vein drug infusion site (step 1.6.6).
      2. Preserve the remaining antibiotic in a 4 °C refrigerator.
   2. Reconstitute one 50 mg vial of amphotericin with 10 mL of sterile water.
      1. Draw up 1 mL of concentrated amphotericin and combine it with 9 mL of sterile saline. Administer the 10 mL mixture to the reservoir to serve as an "antifungal prime" at the beginning of perfusion.
      2. Administer 1 mL of concentrated amphotericin every 12 h to the reservoir.
2. Anticoagulation
   1. Combine 1 mL (100 mg) of enoxaparin with 49 mL of sterile saline. Administer a 5 mL mixture per liter of perfusate at the beginning of perfusion for an "anticoagulation prime."
      1. Preserve the remaining enoxaparin solution in a 4 °C refrigerator.
      2. Administer the enoxaparin solution (5 mL/1 L perfusate) every 12 h to the reservoir.
3. Steroids
   1. Reconstitute 1 g of methylprednisolone with 20 mL of sterile water.
   2. Prime the system according to dosing specifications prior to perfusion initiation: Porcine livers receive 10 mL (500 mg) of methylprednisolone. Partial human livers receive 2 mL (100 mg) of solumedrol.
   3. Dilute the remaining mixture with 40 mL of saline in a 50 mL syringe and place it in a syringe pump to continuously deliver 33 µL/min to the portal vein drug infusion site (step 1.6.6).
4. Parenteral nutrition and bile salts
   1. Reconstitute 1.7 g of taurocholic acid with 50 mL of 10% amino acid + electrolyte concentrated parenteral nutrition component of TPN (unmixed with dextrose) in a sterile specimen cup. Draw up into a 50 mL syringe and place in a syringe pump to continuously deliver 83 µL/min to the portal vein drug infusion site (step 1.6.6.)
5. Vascular tone control
   1. Vasodilation: Reconstitute 0.5 mg of epoprostenol with 50 mL of sterile saline in a sterile cup and place in a 50 mL syringe. Place it in a syringe pump and connect it to the clave proximal to the hepatic artery flow sensor.
      1. Bolus or continuously infuse (10 µL/min) as needed to achieve desired mean arterial pressure and/or to counteract vasospasm.
   2. Vasoconstriction: Dilute 5 mL (50 mg) of phenylephrine with 45 mL of sterile saline in a sterile cup and place in a 50 mL syringe. Store mixture in 4 °C refrigerator and bolus as needed to maintain desired mean arterial pressure.
      NOTE: This is rarely required, and reconstitution may be omitted and implemented only as needed.
6. Pressure and flow sensor calibration
   1. Pressure calibration: Calibrate pressure sensors without perfusate flowing. Re-zero sensors if non-zero pressure reading displayed per manufacturer's instructions while the end of the tubing is open to air for 0 mmHg pressure. Ensure proper pressure measurements by gently obstructing both the proximal and distal ends of the tubing to produce elevated pressures.
   2. Flow calibration: Calibrate once the circuit is circulating perfusate prior to organ perfusion to ensure accurate measurement. Clamp tubing distal to flow sensors to cut off the flow. Re-zero flow sensors if non-zero flow rate displayed per manufacturer's instructions. Unclamp tubing to allow flow and accurate flow measurement.

6. Hepatic procurement

1. Porcine procurement
   NOTE: Porcine livers are procured from a local butcher shop in compliance with the USDA humane handling of livestock protocols. Dissection is guided by the research team to include infra- and suprahepatic vena cava and bile duct.
   1. Sterilize the liver with diluted iodine in sterile saline and place it in a sterile basin on ice. Identify the portal vein, hepatic artery, superior vena cava, and common bile duct, cannulate and secure with 4-0 silk ties.
      NOTE: The bile duct is often cannulated with a fenestrated angio-catheter due to its small diameter and to prevent biliary obstruction of variant porcine bile systems.
   2. Flush the cannulated portal vein and hepatic artery with 3 L of cold sterile saline (2 L PV:1 L HA) followed by 1 L of cold organ preservation solution (750 mL PV:250 mL HA). Submerge the liver in the cold organ preservation solution with ice for transport.
   3. Prime the system as described above (step 4.1) until physiologic parameters are met.
   4. Submerge the cannulated liver in warmed sterile saline for rapid rewarming and removal of any air emboli. Connect the appropriate inflow tubing to the portal vein and hepatic artery cannulas. Titrate pump speed for physiologic flow and pressure values.
   5. Connect the vena cava cannula to the outflow tubing to facilitate outflow to the reservoir to complete the circuit.
   6. Drain the gallbladder of residual bile prior to connecting syringe tubing to the bile duct cannula and a conical vial for collection.
2. Human procurement
   1. Prime the system as described above (step 4.1).
   2. Obtain human specimens after indicated standard of care oncologic resections. Measure in vivo flow rates by intraoperative ultrasound for calibration on the ex vivo system. Identify the segmental portal vein, hepatic artery, and bile duct during surgical dissection. Transect the triad per standard surgical technique and hand the specimen of the surgical field to the sterile back table.
   3. Identify and cannulate the segmental portal vein, hepatic artery, hepatic vein, and bile duct, secure with 4-0 silk ties or prolene suture.
      NOTE: Omit outflow cannulation if the segment does not have a dominant outflow vein to avoid congestive hepatopathy. The basin outflow then functions as the hepatic vein/IVC outflow tubing.
   4. Return to the ex vivo lab and weigh the specimen. Submerge the specimen in warmed sterile saline and connect respective inflow/outflow tubing. Connect the bile duct cannula to the syringe tubing for drainage to the conical vial.
   5. Commence perfusion and titrate pump flow for desired flow and pressure perfusion parameters.

7. Perfusion parameter titration

1. Sample perfusate from pre-hepatic (HA and PV side circuits) and post-hepatic (IVC or basin outflow) claves at numerous time points for titration of perfusion parameters based on point of care blood gas, electrolyte, and hematocrit and hemoglobin values.
2. Calibrate the CDI sensors throughout perfusion using the manual specimen analyses.
3. Adjust hemodialysis inflow and outflow accordingly to maintain volume status and hematocrit concentration of 30-40% and correct any metabolic and electrolyte derangements.
4. Manually dose aliquots of 5% dextrose, free water, 8.4% sodium bicarbonate, and 0.9% NaCl as needed to correct electrolyte abnormalities.

8. Perfusion circuit variants

1. Transportation adaptation
   1. Plug the pumps into a mobile power source for the duration of the run to facilitate continuous perfusion en route to the imaging machine. Plug the mobile power source into the AC outlet to charge the external battery during perfusion.
   2. Unplug the mobile power source from the wall AC outlet once the perfusion experiment is concluded.
   3. Disconnect gas inflow to oxygenators. Clamp off the hemodialysis side circuit and disconnect blood inflow to filter. Stop dialysate pumps.
   4. Remove flow sensors from inflow tubing. Disconnect drug infusions and turn off stopcocks. Unplug pressure sensors from the input/output plug.
   5. Carefully move the drum funnel and tubing to the mobile cart, which now is an independent unit and can be transported to the imaging suite.

结果

Representative porcine runs
Whole porcine livers were obtained from a local butcher for cost efficiency and perfused for various optimization and testing phases of machine development (Supplementary Table 1). The first 24 h of three representative liver runs are evaluated here (Figure 2). Mean hepatic arterial pressure was 80 mmHg ± 11 mmHg, and mean portal venous pressure was 7.1 mmHg ± 3.3 mmHg with flows of 107 mL/min ± 102 mL/min and 501 mL/min ± 329 mL/min, respectively. The large standard deviations for pressure and flows correlate with the large difference in liver size, with pigs ranging from 50 lbs. to 350 lbs. The dual inflow circuits allow for physiologic oxygen differentials with a mean arterial saturation (SO₂) of 95 ± 7.4% compared to mean portal venous SO₂ of 85 ± 13% (p < 0.001). The SO₂ outflow from the hepatic veins was 74 ± 18%, indicative of oxygen consumption (p < 0.01), and peripheral pH measurements were within physiologic ranges at 7.2 ± 0.09. Additional physiologic parameters such as total hemoglobin (9 ± 2.5 g/dL), hematocrit levels (28.3 ± 6.4%), glucose level (477 ± 15 mg/dL), and changes in lactate (3.0 ± 2.5 mmol) and potassium (-0.77 ± 2.3 mmol) were monitored throughout perfusion by sampling at regular intervals. For porcine perfusions, bile production was indicative of organ viability. The gallbladder was manually drained once the organ became normothermic, and the common bile duct was attached to a small tube for collection. Total bile production throughout perfusion was 41 ± 6.9 mL total, with a rate of 6.9 ± 4.0 mL/h.

Figure 2: Perfusate metrics during porcine liver perfusion (n = 3). (A) Average blood inflow rates (mL/min) of the hepatic artery and portal vein over time. (B) Average blood inflow pressures (mmHg) of the hepatic artery and portal vein over time. (C) Comparison of percent RBC O₂ saturation between hepatic artery, portal vein, and inferior vena cava demonstrating oxygen differential between dual inflows and oxygen consumption in the outflow. Statistically significant differences were determined between the portal vein and arterial saturations (p < 0.0001), the portal vein and IVC saturations (p < 0.01), and the arterial and IVC saturations (p < 0.0001). (D) Evaluation of blood pH between hepatic artery, portal vein, and inferior vena cava. (E) Hematocrit (%) and hemoglobin (g/dL) levels (top) throughout perfusion, in addition to change in PV potassium (mmol) (bottom) as a marker of hemolysis. (F) Reservoir glucose levels reflecting mean circuit glucose (mg/dL) throughout perfusion. (G) Change in lactate level (mg/dL) throughout perfusion as compared to time 0 lactate values in hepatic artery inflow, IVC outflow, and system perfusate. (H) Bile production (mL) over time throughout perfusion. Please click here to view a larger version of this figure.

Representative human runs
Three tumor-bearing human hepatic segments were perfused, with perfusion duration time ranging from 48-72 h, depending on the requirements of the experiment (Figure 3). To compare data between representative application runs, we evaluated the first 48 h of biochemical values. Perfusion approximated in vivo parameters with continuous low-pressure, high-volume portal venous inflow and pulsatile high-pressure, low-volume arterial inflow, which are displayed in real-time to allow easy titration. If the liver did not reach physiologic biochemical values within two hours of perfusion, the experiment was terminated. For successful liver perfusions, mean hepatic arterial pressure was 84 mmHg ± 11 mmHg, and mean portal venous pressure was 9 mmHg ± 4 mmHg with flows of 113 mL/min ± 66 mL/min and 317 mL/min ± 302 mL/min, respectively. The dual inflow circuits allow for physiologic oxygen differentials with a mean arterial saturation (SO₂) of 98% ± 3% compared to mean portal venous SO₂ of 92% ± 7% (p < 0.0001). The SO₂ of the outflow from the hepatic veins was 86% ± 9%, indicative of oxygen consumption (p < 0.0001), and pH measurements across the system were within physiologic ranges at 7.3 ± 0.09. Additional physiologic parameters such as total hemoglobin (10.2 g/dL ± 1.6 g/dL), hematocrit levels (30.2% ± 5.5%), glucose level (170 mg/dL ± 68 mg/dL), and changes in lactate (3.6 mmol/L ± 2.2 mmol/L) were monitored throughout perfusion by sampling at regular intervals. In a representative run, an average of 1.4 mL/h of bile was produced. Prior to and at the termination of perfusion experiments, biopsies were obtained for tissue analysis. Representative tissue sections demonstrate hepatic parenchyma, metastatic neuroendocrine tumor, and metastatic colorectal adenocarcinoma after perfusion.

Figure 3: Perfusate and tissue metrics during human liver segment(s) perfusion (n = 3). (A) Mean perfusate inflow rate (mL/min) for the hepatic artery and portal vein. (B) Mean perfusate inflow pressure (mmHg) for the hepatic artery and portal vein over time. (C) Red blood cell oxygen saturation (%) for the inflow hepatic artery, inflow portal vein, and outflow hepatic vein. Statistically significant differences were determined between the portal vein and arterial saturations (p < 0.0001), the portal vein and IVC saturations (p < 0.0001), and the arterial and IVC saturations (p < 0.0001). (D) Perfusate pH for the inflow hepatic artery, inflow portal vein, and outflow hepatic vein. (E) Red blood cell metrics throughout perfusion, including mean hematocrit (%) and hemoglobin (g/dL) perfusate levels (top) with change in perfusate potassium (mmol) from baseline over time (bottom). (F) Mean glucose concentration (mg/dL) for the inflow hepatic artery, inflow portal vein, and outflow hepatic vein over time. (G) Change in perfusate lactate concentration (mg/dL) from baseline over time for the inflow hepatic artery, outflow hepatic vein, and systemic perfusate. (H) Bile production (mL) over time. (I) Human liver segments (Couinaud II-IV) during (left) and post perfusion (right) demonstrating lack of pressure necrosis after system optimization. Superficial area of cauterization present at the top of the left image due to intraoperative hemostasis. (J) Post-perfusion tissue section demonstrating hepatic parenchyma (hematoxylin and eosin stain, 10x, inset 40x; scale bar 50 µM). (K) Post-perfusion tissue section demonstrating metastatic neuroendocrine tumor (hematoxylin and eosin stain, 10x, inset 40x; scale bar 250 µM). (L) Post-perfusion tissue section demonstrating metastatic colorectal adenocarcinoma (hematoxylin and eosin stain, 10x, inset 40x; scale bar 50 µM). Please click here to view a larger version of this figure.

Representative application: CT imaging of perfused specimen
The ability to integrate clinical data sources such as computed tomography (CT) or magnetic resonance (MR) imaging with pathology stands to advance personalized diagnostics and medicine. However, pathology reports are summary findings, and the heterogeneity inherent to solid tumors far exceeds human capabilities for delineation. A potential solution is to perform imaging ex vivo followed by immediate preparation of macroscale tissue sections to allow for radiomic integration of spatially preserving cellular data (e.g., spatial transcriptomics, highly multiplex immunofluorescence). To accommodate conventional imaging on the system, a mobile set-up was constructed, including transitioning pumps to an external battery source to provide undisrupted power during transport and imaging (Figure 4A). With the liver on the CT table, iodinated contrast was administered, and images were obtained, which demonstrate consistent radiographic features with those acquired in vivo (Figure 4B).

Representative application: Kidney and pancreas perfusion
Innate to its design, the system is adaptable to the perfusion of additional organs (Figure 4C-F), which obviates the need for dual inflow and thereby decreases the per-use costs. To demonstrate, we removed the portal venous circuit for a single arterial inflow (Supplementary Figure 2) and perfused a human kidney (Figure 4C) and a human partial pancreas (Figure 4E). Representative tissue sections following perfusion demonstrate glomeruli (Figure 4D) and Islets of Langerhans (Figure 4F).

Figure 4: Representative application of perfusion machines: CT imaging and organ modularity. (A) Mobile adaptation of the system for transport. (B) Representative CT images of the liver with tumor after administration of iodinated contrast in vivo (top) and ex vivo after perfusion (bottom). (C) Human kidney perfusion on the reconfigured machine for single inflow. (D) Post-perfusion tissue section demonstrates renal glomeruli and collecting ducts (hematoxylin and eosin stain, 8x; scale bar 250 µm). (E) Human distal pancreas perfusion on the reconfigured machine for single inflow. (F) Post-perfusion tissue section demonstrates pancreatic islets (hematoxylin and eosin stain, 8x; scale bar 250 µm). Please click here to view a larger version of this figure.

Supplementary Figure 1: Python code settings to facilitate semi-automated perfusion. Schematic depicting input and output parameters utilized by the code enabling semi-automation with the addition of commonly employed manual adjustments (Table). Please click here to download this File.

Supplementary Figure 2: Single inflow schematic used for kidney and pancreas perfusion. P/F Sensor: Pressure and flow sensors. Please click here to download this File.

Supplementary Figure 3: Red blood cell optimization. (A) Blood-only perfusion experiment demonstrated significantly less hemolysis in low RPM conditions compared to high RPM conditions (p < 0.01) using plasma-free hemoglobin as a surrogate for cell lysis. (B) Qualitative comparison of plasma hemolysis over time visually demonstrates less hemolysis in low RPM states when compared to high RPM states. (C-F) Evaluating hemolysis in systems with differing levels of resistance and basin integration with plasma surrogates: (C) A significant difference in lactate levels was determined between the parallel circuit set-up and both the closed and basin set-up (p < 0.0001 and p < 0.0001, respectively). (D) A significant difference in potassium levels was determined between the parallel circuit set-up and both the series closed and series with basin set-up (p < 0.0001 and p < 0.0001, respectively). (E) A significant difference in hematocrit levels was determined between the parallel circuit set-up and both the series closed and series with basin set-up (p < 0.0001 and p < 0.0001, respectively). (F) A significant difference in hemoglobin levels was determined between the parallel circuit set-up and series closed and series with basin set-up (p < 0.0001 and p < 0.0001, respectively). Please click here to download this File.

Supplementary Table 1: Porcine organ optimization perfusions. Thirty-five porcine perfusion experiments performed during optimization phases, including 25 partial livers, 2 kidneys, and 8 whole livers. Included are the duration of perfusion and the reason for termination. Please click here to download this File.

Supplementary Table 2: Human liver validation and application perfusions. Fourteen human partial liver perfusions, including formal left and right hepatectomies spanning to single-segment perfusion experiments. Included are pathology, experimental aim, perfusion duration, and rationale for termination. Please click here to download this File.

Supplementary Table 3: Perfusate-only optimization. Twenty-two perfusate-only optimization runs, including species type, experimental aim, duration of experiment, and reason for termination. Please click here to download this File.

讨论

Normothermic perfusion machines commonly employed prior to solid organ transplantation have enabled significant strides in ex vivo perfusion. While their use with partial organs is technically feasible in some cases^7,8,9, they remain inaccessible for the research scientist due to high costs and limited availability. In this work, we sought to provide a comparable, open-source alternative to the greater research community through the use of readily sourced components and freely accessible code to run the system for a broad range of research endpoints.

Several steps within the protocol are essential for achieving stable organ perfusion. One of the most critical steps is effective pressure offloading to prevent parenchymal collapse and sinusoidal dilation. This was achieved by implementing memory foam supported by an oscillating diaphragm to evenly distribute the organ weight and maintain near-atmospheric portal pressure for adequate drainage. The initial use of a metal grate to support the liver resulted in parenchymal collapse and sinusoidal congestion, as observed in histological analysis and ultrasound imaging during perfusion. This was partially resolved by transitioning to a perforated silicone hammock. While gross pressure necrosis was resolved, there was still notable evidence of parenchymal collapse, as evidenced by high portal vein pressures and significantly congested specimens after 48 h of perfusion. Submerging the liver in saline or blood was also attempted and was successful in relieving congestion and portal hypertension; however, it was not feasible due to the large quantity of perfusate required, difficulty in maintaining sterility, and lack of efficient draining into the reservoir. When compared to the floating technique, we observed the same near-atmospheric portal pressure when the liver was placed on a memory foam cushion with an oscillating diaphragm to disperse weight more equally (Figure 1B and Figure 3I). This design results in softer, better-perfused segments while still allowing easy, efficient drainage of blood outflow from the specimen.

A second critical step is minimizing circuit resistance that imparts shear stress on red blood cells to prevent hemolysis. This was accomplished by introducing side circuits with restrictive components to facilitate physiologic flow without subjecting the perfusate to unnecessary pressure. Hemolysis is an inevitable and continuous process during machine perfusion. In early iterations of the system, extensive hemolysis was a significant contributor to early termination. The shear stress that the system imparts on red blood cells predominantly stems from the pumps due to perfusate stasis and recirculation in addition to narrow diameters of tubes and membranes. The pumps currently employed cause some degree of hemolysis, which increases over time. Pump efficiency is an important factor in hemolysis, as higher flows could be beneficial due to lower rates of blood recycling and secondary flow, mitigated by higher volume of perfusate and/or lower pump speeds. Since large volumes of pRBCs are often difficult to acquire, we focused efforts on redesigning the system to provide physiologic flow rates at lower pump speeds, imparting lower resistance in the circuit overall³¹. To further lessen the intrinsic resistance of the circuit, we integrated side circuits in parallel, which resulted in lower lactate and potassium levels and higher hematocrit and hemoglobin levels over time (Supplementary Figure 3). Lastly, the current hemofilter is integrated into the portal vein side circuit with blood propelled by a peristaltic pump that confers a significant amount of hemolysis due to notable resistance and shear stress imparted by both the filter and its pump. One of our current optimization efforts is focused on removing this side circuit by investigating the potential implementation of hemodialysis tubing into the reservoir. Removal of the clinical-grade hemofilters would allow for subsequent removal of the peristaltic pump, further minimize hemolysis, and reduce the overall cost of the system.

Despite the utility and adaptability of the system described here, limitations should be acknowledged. Notably, the duration of viable perfusion reproducible 3-4 days. Hemolysis remains a central factor limiting perfusion longevity, as discussed above. However, we found that significantly increasing total perfusate volume can substantially extend the duration of perfusion if longer perfusion times are required, but this increases resource utilization beyond what we describe here. Optimal perfusion duration will ultimately depend on the specific experimental question being asked of the system. For example, studies investigating drug delivery kinetics or pathway inhibition may be effectively conducted over 2-12 h, whereas questions involving cell recruitment into the tumor microenvironment may require 48 h³². Extending viability beyond the current window may be important for applications such as extracellular matrix remodeling (i.e., fibrosis), but we believe most cancer-related applications will not require the extra resource expenditure. The primary focus was on oncology-oriented translational applications, aiming to maintain human tumors in a controlled ex vivo environment in which therapy can be delivered through intact parenchyma and tumor vasculature, conditions that more closely replicate the complexities of in vivo scenarios²⁵. In this scenario, the system addresses longstanding issues in current pre-clinical oncology models related to drug delivery, tumor heterogeneity, and drug metabolism, which may explain the frequent failure to translate pre-clinical success into clinical efficacy^33,34.

Compared to existing commercial machines, the system offers significant cost advantages, modular scalability, and experimental flexibility. Commercial devices are often proprietary, costly, and optimized for full-organ perfusion in transplant settings. The system is scalable in a way that commercial machines are not, such that we can envision its applicability broadly. The system presented herein is not a final product but rather a blueprint, which can be modified, adapted, and repurposed to fit a wide variety of research interests and experimental questions, filling a void for scientists across many disciplines.

Future directions for this platform include scaling down to subsegmental liver segments and perfusate volume to enable the use of autologous blood products to support immuno-oncology applications. We are also working to integrate multi-omics platforms and real-time imaging to provide dynamic monitoring of treatment response, paving the way for detailed pharmacological studies. We hope that the system will encourage collaboration and innovation in an effort to forge new depths of knowledge across multiple fields of translational science.

材料

本文使用的材料清单

姓名	公司	目录编号	评论
1/4“ 内径 x 3/8” 外径 x 1/16“ 墙 Versilon C-219-A 软质 PVC 管材	美国塑料公司	57682
10 cc 注射器，带 BD Luer-Lock 吸头	屋宇 署	302995
1cc 注射器，带 BD 鲁尔锁吸头	屋宇 署	309628
25 Fr 门户插管	XVIVO的	11.01.520
3D 打印流量传感器支架	不适用	不适用	请参阅 GitHub for SDL files
3D 打印磁铁支架	不适用	不适用	请参阅 GitHub for SDL files
48 英寸。高 x 24 英寸W 白色钉板	家得宝	109099
4 通旋塞阀，黄色，2 个内螺纹鲁尔锁，带旋转锁的外螺纹鲁尔	世界精密仪器	14057-10
50 cc 注射器，带 BD Luer-Lock 吸头	屋宇 署	309653
ACF00023 泰贡 S3 by 圣戈班 |5/16“ 内径 x 1/2” 外径 x 3/32“ 墙 |E-3603 非 DEHP |不含邻苯二甲酸盐的实验室管 |50' 包装长度	软管仓库	ACF00023
两性霉素	兽医药房	不适用	https://mms.mckesson.com/product/505635/Xgen-Pharmaceuticals-39822105505
倒钩接头，Y 型连接器，3/8“ 管内径，PP，5/PK	大众	MFLX06368-24	材料类型不重要
CDI 540 校准器	Terumo	540厘升
CDI 电缆头支架	Terumo	CDI519CL
CDI监控器	Terumo	500AVHCTCL型
CDI 极夹	Terumo	CDI517CL
Clave IV 无针管接头 100/Ca	亨利·沙因牙科	1097422
含电解质的 Clinimix	巴克斯特	2B7721	只用氨基酸面，不要与葡萄糖面混用
CO2USP 等级 50 磅	罗伯特的氧气	R25
用于 GUI  的计算机显示器;	不适用	不适用	用于查看 GUI 的任何类型的屏幕
数据采集设备	测量计算	USB-202 接口	需要 2 或 3 个，具体取决于机器的连接方式
泵台式电源	李维特尼克斯	100-40015	需要2个，一个用于HA，一个用于光伏
Eagle 1662 滚筒漏斗，带黄铜筛网，直径 18 英寸 x 高 7 英寸，黄色	全球工业	WB440139
依前列醇	兽医药房	不适用	https://mms.mckesson.com/product/1239813/Sun-Pharmaceuticals-62756005940
雌鲁尔¼“倒钩在天然尼龙洁净室中制造;10包	大众	MFLX40314-26	材料类型不重要
接头，聚丙烯，直型，内螺纹鲁尔到软管倒钩适配器，1/8“ 内径;25/PK	大众	MFLX30800-08	材料类型不重要
接头，聚丙烯，直，外螺纹鲁尔锁到软管倒钩适配器，1/8“ 内径;25/PK	大众	MFLX30800-18	材料类型不重要
接头，PVDF，直型，软管倒钩异径管，5/16“ 内径 x ¼“ ID;10/PK	大众	MFLX40614-62	材料类型不重要
流动支架	不适用	不适用	请参阅 GitHub for SDL files
流量传感器夹具	基恩士	FD-XC20R1
流量传感器控制器	基恩士	FD-XA1型
流量传感器头	基恩士	FD-XS20型
流量传感器电源	基恩士	MS2-H50型
气体混合器	多选题	GB100+2	需要2个，一个用于HA，一个用于光伏
气瓶 A，CDI 500	Terumo	中极型506
气瓶 B，CDI 500	Terumo	光碟机507
胰高血糖素试剂盒，胰高血糖素 1 mg，无菌水 1 mL，小瓶	费森尤斯	0593-03
H/S 比色皿 1/4x1/4 W/6“ EXT 1CA=10EA 70-2006-4298-4	Terumo	6934
硬壳储液罐，150 &M 过滤器	血液学公司	00205-00
加热冷垫，Mul-T-Pad，TP12E 垫尺寸 13“ x 18”，适用于 Gaymar TP700 泵	多字垫	8002-062-012	https://www.amazon.com/Heat-Cold-Mul-T-Pad-Gaymar-TP700/dp/B00HFF7876。的 请注意，Stryker 还提供与“Mul-T-Pad C2Dx 8002-062-012”类似的加热水套
胰岛素	兽医药房		https://mms.mckesson.com/product/294668/Novo-Nordisk-Pharmaceutical-00169183311
Ismatec 型 Ecoline 泵 VC-280 II	伊斯马泰克	ISM1078B-115
LTV 1200 儿科呼吸机	护理融合		https://coastbiomed.com/product/carefusion-ltv-1200-ventilator/?srsltid=AfmBOopSCKwTLmKhZXcI CgGQHfiKwj__78jNWxVaznJCS aq3RjXw2E-H. 该型号的儿科呼吸机是从医院剩余的中重新利用的，没有购买。任何商用间歇泵或曝气机都是合适的替代品。
主泵电缆	李维特尼克斯	190-10331	需要2个，一个用于HA，一个用于光伏
MaPerSol 器官保存液，1 升袋	保鲜解决方案	PS005型
Masterflex 适配器接头，鲁尔对鲁尔，三通，Avantor	大众	MFLX45508-56	材料类型不重要
Masterflex 内螺纹鲁尔 3/16“ 内径倒钩，天然 PVDF 洁净室制造	大众	MFLX40314-22	材料类型不重要
Masterflex Ismatec Reglo 数字泵，带 MasterflexLive，4 通道，8 辊;115/230伏交流电	万世高振	MFLX78018-22	https://www.avantorsciences.com/us/en/product/NA5139147/masterflex-ismatec-reglo-digital-multichannel-pumps-avantor
Masterflex 外螺纹鲁尔 1/4“ 内径倒钩，天然 PVDF 洁净室制造，10 个/包	大众	MFLX40318-27	材料类型不重要
Masterflex 外螺纹鲁尔 3/16“ 内径倒钩，天然 PVDF 洁净室制造	大众	MFLX40318-22	材料类型不重要
Medline 胎盘盆标准托盘	医疗线	DYNJSPLACENTA
美敦力 DLP 2 英寸血管插管，3 毫米斜头	缝合线	MDT30007
记忆海绵	最优惠的价格床垫		https://www.amazon.com/Best-Price-Mattress-Twin-Topper/dp/B07999TQWB/ref=sr_1_2_sspa?crid=17LQXX80WX5FK&dib=eyJ2Ijo iMSJ9.oRyEE5KwmKy1pb-VvY9MOJg _MF3wr0PmXqKcOrSEbeJWS2X 5BkFs_WSfuE1OU8GS0u_T1qM IJbsoy3omRv1wfBCy2sw2fOUIX JMqAO6BSb7t_D3j745MoXjedLbo BYypyi6_Qgoc32N8bpVcmoT_k9b LECWdi_KpEmxQx2PEok1fjlK3ywO yBXwqPnXefMt5JeW5H1R7RO3fD g4tidmRoEJ-YxfU3d_iprpFpm7A3zi 7CG_BMUAaGpBcRitwqb9Q4TB5bM 1i-GcMhSFKbzaGNd76CoVVVkZRjib c9Ix-CvY.t7xCkzB81vNhO9gYLxoNen WHEwC6JJpw9T4G6iIa5N4&dib_tag =se&关键字=内存%2Bfoam&qid =1748632609&前缀=内存%2Bfoam %2Caps%2C133&SR=8-2-SPONS&sp_csd =d2lkZ2V0TmFtZT1zcF9hdGY&th=1
带直无螺纹孔的钕磁铁	麦克马斯特·卡尔	336K834
氮气，125 CU FTGrade 4.5 高纯度	罗伯特的氧气	R32
氧气，255 CU FTGrade 4.5 高纯度	罗伯特的氧气	R5
PDMSXA-2500：PermSelect 2500 厘米平方薄膜模块 -- 带倒钩配件，不带 MCC	烫选	PDMSXA-2500型	可以进行鲁尔或倒刺接头，但连接会发生变化
盐酸去氧肾上腺素注射液，USP 10mg/mL 1mL 小瓶	健康第一	1000520
用于 ORS 流体加温系统的聚氨酯加温窗帘，无菌，44 英寸 x 66 英寸	医疗线	ORS-300H
猪猪血	兰皮尔	7204911
压力传感器监视器	PendoTECH公司	PMAT2HR
PRISMAFLEX M60 套装	百特医疗公司	106696
PrismaSol BGK0/2.5 用于连续肾脏替代治疗的替代溶液，非 PVC	巴克斯特	110240S
泵安装底板	李维特尼克斯	190-10313	需要2个，一个用于HA，一个用于光伏
泵 USB 适配器	李维特尼克斯	100-30391	需要2个，一个用于HA，一个用于光伏
PuraLev 驱动系统	李维特尼克斯	100-91025	需要2个，一个用于HA，一个用于光伏
PuraLev IP 适配器电缆	李维特尼克斯	190-10475	需要2个，一个用于HA，一个用于光伏
分流传感器 SYS500 - CA = 20EA 98-0702-1541-7	Terumo	CDI510H	可以从缝合线中购买过期的 - 工作原理相同，有时有一点 pH 漂移
一次性压力传感器，非无菌，聚砜 1/4 英寸软管倒钩	PendoTECH公司	预备-N-025
一次性压力传感器	PendoTECH公司	预备-N-025	可重复使用灭菌，但每次使用前都要测试传感器
碳酸氢钠 8.4% 溶液实验室级	实验室巷	144-55-8
牛磺胆酸钠水合物	米利波尔西格玛	86339-25克
索美曲	兽医药房	不适用	https://mms.mckesson.com/product/1212228/Dr-Reddys-Laboratories-43598013074
史赛克加热垫	亚马逊河	B01M74Y8DI
SU 泵头，PP，倒钩 3/8“，伽马 >25kGy，DCP-30.2-G25，PuraLev	李维特尼克斯	100-91071	灭菌后可重复使用
注射泵（11 Elite泵）	Harvard Apparatus	70-4504	每种输注药物需要一个
注射器管	GSS系统	ME2020
热水器和循环器	格兰特科学	TC120型
佐辛	兽医药房	不适用	https://mms.mckesson.com/product/1009456/Auromedics-Pharmaceuticals-55150011930