Extracorporeal Support Vascular Access: 3D Printing Custom Cannulas

Abstract

Extracorporeal life support (ECLS) requires specialized cannulas for vascular access that are dependent on each patient's needs, size, and unique vascular anatomy. There are currently no commercially available drainage cannulas suitable for patients smaller than 1.5 kg. This study presents the development of a novel 6 Fr cannula for blood drainage to be used in these small patients. Cannulas were manufactured using stereolithography 3D-printing on Form 3 and Form 3B printers and tested in vitro using 3% glycerol. Drainage pressure at 100 mL/min was −39±2 mmHg and −42.3±9 mmHg, for Form 3 and Form 3B, respectively. Maximum flow rate and drainage pressures were 185.8±15 mL/min and−80±2 mmHg, and 165±21 mL/min and −85.7±12 mmHg for Form 3 and Form 3B, respectively. Three Form 3B cannulas were tested in vivo in an ECLS piglet model, the best of which had a maximum blood flow rate of 139 mL/min at −32 mmHg drainage pressure. Cannulas 3D-printed using the Form 3 and Form 3B printers exhibited appropriate size and flow rates (~100 mL/kg/min) for ECLS needed to support premature patients with minimal complexity in the fabrication process. These cannulas should be optimized in preclinical research using translational models.

INTRODUCTION

Extracorporeal Life Support (ECLS) is used to support patients with cardiac and respiratory failure and relies on vascular cannulas for blood drainage and reinfusion. These cannulas must be designed to achieve efficient blood flow, safe insertion, and minimal risk of vascular injury in patients of different sizes. Current ECLS guidelines exclude infants <34 weeks estimated gestational age (EGA) and <1.5 kg due in part to vascular cannula size constraints and flow dynamics.¹ In 2017, we analyzed data from the Extracorporeal Life Support Organization (ELSO) registry on 752 neonates with a gestational age of 29-34 weeks treated with ECLS. Our study found a higher survival rate at 34 weeks (58%) compared to 29-33 weeks (48%, p=0.05). Rates of intracranial hemorrhage were similar across groups, with no correlation between hemorrhage and gestational age. These findings have prompted other research teams to reconsider current exclusion criteria.^2,3 Our laboratory is developing a novel form of ECLS for Extremely Low Gestational Age Newborns (ELGANs), defined as those born at <28 weeks EGA and <1.5 kg.⁴ Miniaturizing cannulas is essential for translating this technology, as ELGANs face high mortality and morbidity rates with limited therapeutic options.^5-8 Additionally, the absence of appropriately sized cannulas impedes ECLS research in small animal models.

3D printing has been used to manufacture laparoscopic cannulas,⁹ hippocampal guide cannulas¹⁰ and trocar cannulas¹¹ where no suitable commercial cannula existed. This work presents a novel and reliable method to 3D-print custom small vascular drainage cannulas for ECLS, potentially improving outcomes for vulnerable patient populations and enhancing animal models for translational research.

METHODS

Cannula Design and Manufacturing

Requirements:

The ECLS Laboratory at the University of Michigan is developing an ECLS system for ELGANs (<1.5 kg).^2,12-15 However, the lack of suitable commercially available cannulas remains a hurdle and limiting factor for application of this technology. This project requires developing a miniaturized drainage cannula for ELGANs. The cannula inserts at the jugular vein, ends in the mid right atrium, and must sustain flows of 80-100 mL/kg/min.¹⁶ The jugular vein internal diameter for these patients is 2-3 mm.^17,18

Novel Cannula Concept and Design:

The cannula was designed in Solidworks^® (Dassault Systems, Vélizy-Villacoublay, France) based on dimensions and features as requested by ECLS experts. It begins with a 50 mm length at 6 Fr to minimize the pressure drop (Figure 1). Over the next 40 mm, the cannula gradually expands to an 8 Fr diameter. This design has theoretically less fluidic resistance compared to a similar cannula that is 6 Fr across the entire 90 mm length. The cannulas had by default 16 side drainage holes up to 2 cm from the tip end, although some cannulas tested other configurations with 8, 16, or 32 drainage holes spaced up to 1, 2, or 4 cm from the tip end. Drainage hole sizes varied from 0.6 mm to 1.0 mm.

Figure 1: The 3D Model of The Cannula, Created in Solidworks^®.

(Dassault Systems, Vélizy Villacoublay, France).

Form 3 and Form 3B Manufacturing:

Cannulas were printed on Form 3 and Form 3B printers (FormLabs, Somerville, MA). These Soft Lithography printers stack 2D layers on top of each other to form a 3D shape.¹⁹ The cannulas printed on the Form 3 printer used the FormLabs’ Flexible 80A resin, and the cannulas printed on the Form 3B printer used the FormLabs’ BioMed Flex 80A resin. Resins were selected because they resulted in cannulas with suitable flexibility based on ECLS clinician feedback (ShoreA 80) and could withstand repeated bending, flexing and compression. The Form 3B printer produces smoother parts than the Form 3 and uses BioMed Flex 80A resin, which is deemed "biocompatible" for up to 24 hours of contact with mucosal membranes; however, there are no data on its blood compatibility. In contrast, Flexible 80A resin does not have a "biocompatible" designation but is marketed as slightly stronger, with an ultimate tensile strength of 8.9 MPa compared to 7.2 MPa for BioMed Flex 80A. Cannulas were post-processed according to FormLabs instructions. Cannulas were considered “successfully manufactured” if they were patent and intact.

Enhancing 3D Printed Surface Material Finish:

A CarboSil^® polyurethane-silicone polymer surface coating was applied to a subset of the cannulas.²⁰ A 50 mg/mL CarboSil^® polyurethane-silicone polymer (DSM Biomedical, Exton, PA) was dissolved in anhydrous tetrahydrofuran (THF, Sigma Aldrich, St. Louis, MO). The inner lumen of the cannula was filled with the solution and then drained. The excess solution was immediately removed from the cannula by gently blowing air through it. To properly cure the surface and remove all organic solvent residue, the cannula was left to dry in a negative pressure hood overnight at 21 °C.

In Vitro Testing Method

The primary objective of this study was to evaluate the performance of cannulas produced using a novel method. We assessed cannula performance by measuring the maximum flow rate (with 3% glycerol or blood) until increases in pump RPMs no longer affected flow, as well as the negative pressure at different flow rates and cannula collapse. Cannula collapse was defined as structural failure, characterized by deformation or pinching during testing.

The secondary objective of this study was to determine the manufacturing choices that would optimize cannula performance. This included Form 3 vs Form 3B printers and their respective resins, wall thickness and drainage hole size, and the use of CarboSil^® coating.

In Vitro testing setup:

Various cannulas were manufactured and tested according to ISO 10993-4²¹ and previous studies by our group.²² The circuit used the cannula as a drainage cannula connected to a centrifugal pump (3M Sarns, Maplewood, MN) with a flow sensor (Transonic MXP-6XL, Ithaca, NY) and pressure transducers (ICU Medical Triple Transpac IT Monitoring Kit w/ 3 mL Flush Device, San Clemente, CA) attached pre-pump. The cannulas were tested with a 60/40 water-glycerol mixture to simulate blood viscosity at room (25 °C) and body (37 °C) temperature.²³ Data collected included drainage pressure, flow rate, and pump speed.

A variety of cannulas were manufactured to test the performance of various configurations. Among these cannulas were those required to conduct the following sub-experiments:

1. CarboSil^® coating effect: Six (n=6) cannulas manufactured using the Form 3 printer with drainage hole diameters of 0.6, 0.7, and 0.8 mm were used for this study. They were evaluated in in vitro parametric studies at 25 °C twice, once before CarboSil^® coating and once after.

2. CarboSil^® coating cannula performance: Five (n=5) cannulas with 0.8 mm drainage holes, manufactured using the Form 3B printer, and coated with CarboSil^® were evaluated in vitro at 37 °C to evaluate the performance of cannulas fabricated on the Form 3B.

3. Evaluation of drainage hole size and wall thickness combinations: Twenty (n=20) cannulas manufactured on the Form 3, with drainage hole sizes of 0.8, 0.9, and 1.0 mm, wall thicknesses of 0.20, 0.25, and 0.30 mm, and coated with CarboSil^® were evaluated as described above.

In Vivo Testing Method

To evaluate both fluidic and mechanical performance of the cannulas, the cannulas were tested in a newborn ECLS piglet model. All experiments and procedures followed the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals. This study was approved by the UM Institutional Animal Care and Use Committee (IACUC) under protocol #11170. In preparation for this, a total of 14 cannulas were manufactured using the BioMed Flex 80A material. The three most promising cannulas were inserted into the piglet and tested in varying positions and flows.

Newborn Piglet ECLS Model:

Under general anesthesia, a healthy newborn piglet was instrumented for venovenous ECLS vascular access via modified Seldinger technique under direct visualization (Figure 2). The new 6 Fr 3D printed cannula was used for drainage via the right jugular vein while an 8 Fr Medtronic cannula was used for reinfusion via the left femoral vein. The ECLS circuit consisted of ⅛ inch tubing, a passive filling roller pump (Daris, LLC, Dexter MI), flow probes (Transonic MXP-6XL, Ithaca, NY), pressure transducers (ICU Medical Triple Transpac IT Monitoring Kit w/ 3 mL Flush Device, San Clemente, CA), and an oxygenator (Terumo Cardiovascular FX05, Ann Arbor, MI). The sweep gas used was 95% O₂ and 5% CO₂. The right femoral artery was used for intravascular hemodynamic monitoring (heart rate and systemic blood pressure) in conjunction with a pulse oximeter. The femoral vein was used to monitor central venous pressure and administration of IV fluids and medications. Arterial blood gasses were collected for gas exchange, hemoglobin, and electrolyte monitoring. The animal was anticoagulated with heparin (150 U/kg) prior to cannulation and during ECLS support following activated clotting times (ACT 220-300 s). To characterize the cannulas, pressures (pre-pump, pre-oxygenator, post-oxygenator), pump RPMs, ECLS blood flow, oxygenator resistance, and position (via Transthoracic Echocardiography, TTE) were all measured for each cannula. To assess if blood flow was affected after transitory cannula movement from the ideal venous drainage location (center of the RA), the cannulas were tested in three anatomical positions within the right atrium (RA): (i) center of the RA; (ii) 1 cm past the RA-inferior vena cava junction (RA-IVC+1 cm); and (iii) 1 cm proximal to the RA and superior vena cava junction (RA-SVC-1 cm).

Figure 2: In vivo Piglet Cannula Testing.

(A) In vivo cannula test circuit; (B) Cannulation into the jugular vein of the pig; (C) Closeup of the cannulation site.

Data Analysis

For both in vitro and in vivo studies, continuous data are presented as Average ± Standard Deviation (SD). In vitro data for all cannulas were analyzed using multivariable linear regression in R (R Foundation, Vienna, Austria), considering wall thickness, Carbosil^® coating, printer and resin types, hole size, hole count and positioning, and glycerin temperature. This analysis identified the impact of each variable on performance while controlling for others. In vivo studies additionally assessed animal hemodynamics and ECLS characteristics, measuring drainage pressure, ECLS blood flow, and cannula position. This consistency allows for effective comparisons across cannulas. Due to the limited sample size (n≤3), no statistical analysis was conducted.

RESULTS

Cannula Design and Manufacturing

Form 3 and Form 3B Manufacturing:

A total of 76 cannulas (Form 3 = 50 and Form 3B = 26) were manufactured and were fully patent with no obvious deformities. The quantity of failed cannulas was not recorded. Reasons for failure included: 1) human error; 2) divergent shrinkage rates between cannula and supports during post-processing in alcohol, which ripped apart the cannula; and 3) resin remaining stuck in the cannula and being cured during the final UV process. Images are presented in Figure 3.

Figure 3: Microscope Images of The Cannula Tip of a Cleared Cannula Printed on The Form 3.

(A) drainage holes; (B) tip body (B); (C) Comparison between a cannula produced on the Form 3 (bottom) and Form 3B (top).

The outer diameter (OD) of printed devices at the 6 Fr section was 1.72±0.15 mm (4.3 - 6.1 Fr), or about 0.28 mm smaller than the 2.0 mm (6 Fr) designed CAD model. The wall thickness was typically 0.02±0.04 mm larger than the designed CAD model, and the internal diameter was typically 0.01±0.18 mm smaller than the designed CAD model. More detailed data are presented in Supplemental Table S1.

In Vitro Testing of Novel 3D Printed Cannulas:

Collapse was not observed with any of the fabricated cannulas during in vitro and in vivo studies. A summary is presented in Figure 4. No statistically significant difference in performance between the Form 3 and Form 3B cannulas was observed, despite the Flexible 80A resin’s higher advertised strength.

Figure 4: Summary of Cannula Performance for All 3D Printed Cannulas.

Average in vitro pressure vs flow data for all cannulas is used to compare Form 3 vs Form 3B cannulas along with a Medtronic 6Fr infusion cannula and a Medtronic 8 Fr drainage cannula. Data is presented as Average ± Standard Deviation at each flow rate.

A comparison with the theoretical performance of the 6 Fr cannula is presented in Supplemental Table S2. Empirical pressure drop measurements were 20% higher than theoretical values.

Multivariable Linear Regression Results:

The linear regression results found that an increased drainage hole size corresponds to a greater maximum flow rate (0.26 mL/min per μm) (p=0.015). Further, a larger wall thickness (and therefore smaller internal lumen diameter) corresponds to a reduced maximum flow rate (−0.30 mL/min per μm), albeit the relationship is not significant (p=0.059). The effect of other variables on maximum flow rate or pressure drop at 100 mL/min was not statistically significant. Full results can be found under Supplemental Table S3.

Drainage Hole Size and CarboSil^® Coating on Form 3:

The performance of uncoated cannulas of varying drainage hole sizes is shown in Figure 5. Cannulas with larger drainage hole sizes of 0.7 mm and 0.8 mm had higher glycerol flow rates than cannulas with 0.6 mm drainage holes. Uncoated cannulas maintained a pressure of −40±4 mmHg at a flow rate of 100 mL/min, achieving a maximum flow of 200±13 mL/min at −94±2 mmHg drainage pressure. In comparison, coated cannulas sustained −41±9 mmHg at 100 mL/min with a maximum flow of 183±44 mL/min at −90±6 mmHg drainage pressure. No significant differences in pressure drop were noted between the two types at either flow rate. The two-factor linear regression analysis was not significant for predicting pressure drop as a function of drainage hole size (p=0.96) or coating (p=0.75), nor for maximum flow as a function of drainage hole size (p=0.31) or coating (p=0.39).

Figure 5: Impact of Drainage Hole Size and Coating on Cannula Performance for Form 3 Printed Cannulas.

(A) Pressure vs flow rate data for uncoated cannulas of varying drainage hole sizes printed on the Form 3 printer. The first number on the legend refers to the drainage hole size diameter in mm, while the second number identifies the individual cannula; (B) Flow rates and pressure data of cannulas of various drainage hole sizes after being coated with CarboSil^®.

CarboSil^® Coating on Form 3B:

The performance of the Form 3B cannulas both before and after being coated with CarboSil^® is shown in Figure 6. These cannulas sustained comparable flows to the cannulas printed with the non-biocompatible Flexible 80A material. Uncoated cannulas at a glycerol flow rate of 100 mL/min had a drainage pressure of −40±6 mmHg and a maximum flow of 212±44 mL/min at −98±2 mmHg. In contrast, coated cannulas showed a drainage pressure of −35±5 mmHg and a maximum flow of 164±58 mL/min at −98±1 mmHg. Post-coating, there was no significant difference in pressure drop for individual cannulas at either the 100 mL/min flow rate or maximum flow compared to pre-coating performance.

Figure 6: Impact of Coating on The Performance of Form 3B Cannulas.

Flow rates and pressure data of uncoated (A) and coated (B) cannulas with drainage hole size of 0.8 mm compared to a Medtronic 6Fr infusion cannula and a Medtronic 8 Fr drainage cannula. The first number on the legend refers to the drainage hole size diameter in mm, while the second number identifies the individual cannula.

Varying Wall Thickness vs Drainage Hole Size on Form 3:

For the study evaluating varying wall thickness and drainage hole size, the average pressures at different flow rates were calculated for coated cannulas. This is shown in Figure 7. Cannulas with wall thicknesses of 0.2 mm, 0.25 mm, and 0.3 mm sustained pressures at 100 mL/min of −27±6 mmHg, −59±16 mmHg, and −46±18 mmHg, respectively, and maximum flow rates of 241±34 mL/min, 148±32 mL/min, and 175±52 mL/min. Cannulas with drainage hole diameters of 0.8 mm, 0.9 mm, and 1.0 mm sustained pressures at 100 mL/min of −50±21 mmHg, −40±15 mmHg, and −25±4 mmHg, respectively, and maximum flow rates of 173±51 mL/min, 191±48 mL/min, and 263±30 mL/min.

Figure 7: Effect of Wall Thickness and Drainage Hole Sizes on Performance.

(A) Average flow and pressure data of cannulas of varying wall thicknesses (0.20, 0.25, 0.30 mm); (B) Average flow and pressure data of cannulas of varying drainage hole sizes (0.8, 0.9, 1.0 mm). For comparison, data for commercial 6 Fr Medtronic infusion cannulas and 8 Fr Medtronic drainage cannula are provided in each plot.

The results of the linear regression analysis are presented as Supplemental Equation S1 and S2. Larger drainage hole size results in a lower pressure drop at 100 mL/min (p=0.02) and a higher maximum flow rate (p=0.006). Thinner wall thickness also results in a lower pressure drop at 100 mL/min (p=0.02) and a higher maximum flow rate (p=0.006).

In Vivo Testing

Newborn Piglet ECLS Model:

Fourteen cannulas were prepared for in vivo testing. Of these, 3 cannulas made of the Form 3B BioMed Flex 80A material were tested using a 1.0 kg piglet with average ACT= 298±50 seconds (Supplemental Figure S1).

The first cannula (⅛ inch connector, 0.25 mm wall thickness, 16 over 2 cm hole distribution, and CarboSil^®-coated) clotted immediately. The second cannula (⅛ inch connector, 0.25 mm wall thickness, 16 over 1 cm hole distribution, uncoated) was successfully inserted via the jugular vein. As expected, the cannula sustained the highest flow rate in the center of the RA. In this position, the cannula sustained flow of 139 mL/min with a pressure drop of −32 mmHg (Figure 8A). For comparison, this cannula sustained a pressure of −102 mmHg at a maximum flow rate of 200 mL/min in vitro. The third cannula (¼ inch connector, 0.2 mm wall thickness, 16 over 2 cm hole distribution, and CarboSil^®-coated) was successfully inserted via the jugular vein. The cannula sustained the highest flow rate in the RA-IVC+1 position. Here, the cannula sustained a maximum flow of 115 mL/min at a pressure drop of −35 mmHg (Figure 8B). For comparison, this cannula sustained a pressure of −102 mmHg and a flow rate of 190 mL/min in vitro. Although this cannula performed better in the RA-IVC position, this is a dangerous anatomic position for a venous drainage cannula, especially in pediatric patients.

Figure 8: Performance of the cannulas #63 and #66 in vivo and in vitro.

Cannula #63 (A) is the second cannula tested in vivo, and Cannula #66 (B) is the third. Both cannulas were tested in 3 different anatomical positions relative to the right atrium (RA) and inferior vena cava (IVC) junction. In vitro data for each cannula is provided for reference.

Cannulas sustained a smaller pressure drop at 100 mL/min in vivo than in vitro but lower maximum flow rates.

DISCUSSION

Manufacturing Method

Form 3 and Form 3B Manufacturing:

The manufacturing method for Form 3 and Form 3B printers consistently produces patent and functional cannulas. These cannulas were measured to have an OD approximately 14% smaller than the 2.0 mm OD (6 Fr) CAD design (Supplemental Table S1). This decrease in diameter from the drawn dimensions is likely the main cause of the difference in pressure and flow between the 3D printed cannulas and the 6 Fr Medtronic control cannula. In the future, the CAD model could be made roughly 33% larger at a 2.33 mm OD, with the expectation that it would shrink down to 2.0 mm after post-processing. This change would result in a realized cannula with actual 6 Fr dimensions and improved flow and pressure performance.

Ideal 3D-Printed Cannula Characteristics:

The Form 3B cannulas appeared to perform slightly better than the Form 3 cannulas (Fig. 4), although not by a statistically significant margin. Combined with the “biocompatibility” rating, albeit not for blood, the Form 3B cannulas are more promising for future animal studies or clinical use. Even with the “biocompatibility rating”, a solvent soak to remove any components that may leach into the blood may also be required. Surface roughness measurements are planned for a future study.

The effect of drainage hole size and CarboSil^® coating on performance was inconclusive for the Form 3 study with n=6 (Fig. 5), so n was increased to 20 for the Form 3B study (Fig. 6). This Form 3B study demonstrated that both larger drainage holes and thinner wall thickness significantly improved performance by obtaining lower negative pressures at 100 mL/min and maximum flow rate.

Although not performed in this study to simplify logistics, the CarboSil^® coating can be impregnated with a nitric oxide releasing polymer which has antithrombogenic properties; this would mitigate the need for systemic anticoagulation, overcoming a critical hurdle to clinical application in premature infants.²⁴ The CarboSil^® coating did not exhibit statistically significant lower flow rates. However, it greatly increased the variance in performance, so further work is needed to improve consistency in the coating process.

Comparison with commercial cannulas:

Commercial 6 Fr cannulas have a slightly lower pressure drop for each given flow rate. This is primarily because they have a larger internal diameter (due to shrinkage of the printed cannulas relative to the drawn dimensions), do not have the additional 40 mm length which tapers up to 8 Fr, and likely have a smoother internal texture, although surface roughness was not quantified in this study. Further, with their steel wire reinforcement, the Medtronic cannulas are extremely difficult to kink or collapse. However, the 6 Fr infusion cannulas are too short, while the 8 Fr drainage cannulas have too great an OD for this patient population. Therefore, despite their excellent performance characteristics, these cannulas cannot be used in ELGANs. In summary, the 3D printed cannulas performed almost as well as equivalent commercial cannulas, but can meet study-specific requirements where commercial cannulas are unavailable.

This 3D printing method addresses a critical milestone in the development of ECLS technology for premature patients which requires specialized miniature cannulas.²⁵ Once the milestones for clinical translation are reached, we anticipate there to be sufficient market demand and economic incentive for commercial cannulas that supersede the 3D-printed cannulas presented here. However, until that time, the 3D printing method can act as a bridge-to-clinic.

Modifiability for Unique Use Cases:

The manufacturing of a cannula which gradually increases in diameter to accommodate a growing infant was not significantly more difficult than the manufacturing of a simple cannula. This demonstrates the ease with which this 3D-printing method can be optimized for unique use cases.

LIMITATIONS AND NEXT STEPS

The in vivo study was conducted with a single newborn piglet (1.0 kg), comparable in size to premature human patients. While only a few cannulas were tested, this initial evaluation provided proof-of-concept for the catheter before more complex and costly preclinical studies with a premature animal model. The study showed feasibility but did not assess long-term biocompatibility. Notably, one of the three cannulas clotted immediately, raising concerns about biocompatibility. Due to the experiment's duration and the drying of the jugular vein wall, no additional cannulas were tested. Further research is needed with a larger sample size to evaluate long-term biocompatibility and mechanical durability over several days to weeks. While fabricated cannulas can reach flow rates of 100 mL/min, their reliability remains insufficient. Increasing the sample size would yield more data to enhance the performance of 3D-printed cannulas. To enhance fluidic performance, cannulas with a larger OD can be manufactured. In this study, we experienced a shrinkage of 0.28±0.15 mm, resulting in an actual outer diameter of 1.72±0.15 mm (86% of the target). By adjusting the design to a 2.33 mm OD, it is anticipated that the final diameter after post-processing will be approximately 2.00±0.18 mm (6 Fr). This modification would facilitate a larger inner diameter, leading to an increased flow rate, estimated to be about 1.85 times greater than the current design, based on the ratio (2.33/2.0) raised to the fourth power. Upon confirming the reliability of these enhancements, the cannulas should be evaluated in preclinical models of ELGANs.

CONCLUSION

We have demonstrated the feasibility of manufacturing custom miniature cannulas using a 3D printing technique with flexible resins in the Form 3 and Form 3B 3D printers. In vitro drainage cannulas produced by this method demonstrate an average drainage pressure of −39 mmHg at a flow rate of 100 mL/min, and maximum flow of 186 mL/min with an average drainage pressure of −80 mmHg. In the ECLS piglet model, the maximum flow rates were 139 mL/min and 115 mL/min, making it suitable for human ELGAN patients. This method can also be adapted for experimental ECLS procedures and research in small-animal models.