About 13 million preterm neonates are born worldwide every year (Ohuma et al., 2023). One million of these infants die from the pathology which enforces preterm birth and from complications of prematurity itself. The most immature patient group are extremely preterm infants (EPI) with a gestational age of less than 28 weeks. EPI regularly need neonatal intensive care treatment to allow survival despite their extreme immaturity (Blencowe et al., 2013; Dougherty et al., 2022). Recent progress in clinical treatment, especially non-invasive respiratory support, has significantly improved outcomes in this patient group (Göpel et al., 2015). Nonetheless, pulmonary gas exchange, including mechanical ventilation, continues to cause lifelong organ damage, particularly to the immature lung. The most important in this context is Bronchopulmonary Dysplasia (BPD), which affects more than half of the extremely preterm infants (Hurskainen et al., 2021; Regin et al., 2022).

Artificial amnion and placenta technologies (AAPT) are promising approaches in reducing the prematurity-related lung injury (van der Hout-van der Jagt et al., 2021). In lamb and piglet models, several working groups have successfully achieved the transfer of extremely immature fetuses to extrauterine, fluid-based environments (Hornick et al., 2016; Miura et al., 2016; Ch et al., 2021; Eixarch et al., 2023; Charest-Pekeski et al., 2022; Hornick et al., 2018). Artificial placentas are also developed as rescue treatment for preterm respiratory failure in conventional neonatal intensive care treatment and for use in an air based environment with occlusion of the airway (Omecinski et al., 2023; Kosaka et al., 2025; Sanchez-Martinez et al., 2023; van et al., 2025; Church et al., 2018; Schoberer et al., 2012).

Gas exchange during AAPT happens extracorporeally within membrane oxygenators (ECMO) which are connected to the fetus via the umbilical cord. Large bore cannulation of umbilical vessels is an essential technique for this approach. Cannulation of umbilical vessels for AAPT treatment will require the development of cannulas specifically designed for this purpose. These need to fit tightly into the vessel, resist traction and have a low resistance. Reliable long-term operability is crucial in this treatment because failure is immediately life-threatening and can cause lifelong neurocognitive impairment. Although the hypoxia tolerance of the newborn and especially the preterm brain is probably longer than the 3–5 min in normothermic adults (Singer, 1999), we assume that the cannulation time should not exceed 3 minutes to be regarded as safe for ethical considerations. Therefore, the EXIT umbilical vessel cannulation procedure ought to be developed, time optimized and trained in animal models or in ex vivo human test setups before being introduced into clinical practice. The human ex vivo model, in this case, is superior to the animal models due to animal ethics and biosimilarity considerations. The umbilical cords of piglets and lamb differ from human ones in anatomy regarding number of vessels, vascular diameter, and tissue consistency (Luis Manuel, 2018; Randall, 1989; Jordan, 1919).

Clinical and technical experience with ECMO-cannulation of umbilical vessels is scarce. To date, there is a lack of short and long-term data on human umbilical cord use for extracorporeal circulation. Cornish Et Al. (1991) performed cannulation of human umbilical veins as a supplemental drainage to the jugular venous access during ECMO rescue therapy on three subjects (CORNISH et al., 1991). Bjøro and Stray-Pedersen, (1986) introduced the concept of utilizing reperfused human umbilical cord cutoffs to draw conclusions about physiological in vivo conditions for medication purposes (Bjøro and Stray-Pedersen, 1986). The effect of vasoactive substances on the umbilical vascular tone was tested, confirming the possibility of reviving umbilical tissue (Bjøro and Stray-Pedersen, 1986). The cutoffs were reperfused in a test stand using a low-viscosity perfusate with a pH value similar to amniotic fluid (Bjøro and Stray-Pedersen, 1986). This test stand was, however, not intended for cannulation testing nor the simulation of extracorporeal life support (ECLS) (Bjøro and Stray-Pedersen, 1986). Only unphysiological low flow rates could be achieved using this test setup (Bjøro and Stray-Pedersen, 1986).

We present an ex vivo human umbilical cord test setup using autologous placental blood intended for the simulation and testing of human umbilical ECMO cannulation. The ex vivo model of blood-perfused human-umbilical cords features adjustable blood flow and–pressure. We were able to show the successful parallel cannulation of all three human umbilical vessels with commercially available ECMO cannulae. To our knowledge, for the first time.

A total number of 67 umbilical cords were harvested. 20 (29,8%) of the donated umbilical cords were excluded since they did not meet the length criteria of >15 cm. Two cords (3.0%) were excluded due to maternal infections, and an additional two (3.0%) due to handling mistakes in the cannulation process. 43 cords were cannulated terminally of which 28 cords were used for preliminary training purposes. This resulted in n = 15 umbilical cords being included in the validation study.

This work describes the development and establishment of an ex vivo test setup which allows to perform cannulation experiments on reperfused human umbilical cord vessels. The method is novel and is suitable to close relevant gaps in translational research of artificial placenta technology. The presented setup and cannulation method enable a safe and ethically acceptable method for training in umbilical cord cannulation, particularly for AAPT or neonatal ECMO, with the umbilical cord serving as the access point. Compared to the experiments of Bjoro et al. (Bjøro and Stray-Pedersen, 1986), we developed a comprehensive circuit that includes arterial and venous umbilical vessels and allows for the addition of further circuit components, such as oxygenators. Pressure losses are still in the supraphysiological range but much closer to physiological pressures compared to the previously reported setup (Bjøro and Stray-Pedersen, 1986). Further improvements in terminal cannulation strategy (use of larger cannulae) might allow physiological pressures and flow in future studies. All material needed for these experiments can be harvested after childbirth from repelled tissue, so the method inherently carries low risk, is more ethical as it does not involve patients. It allows for the reduction of the need for animal studies. Moreover, it is superior to the quality of animal experiments for biosimilarity reasons: the anatomy of animal umbilical cords differs significantly from the human one, concerning the number and architecture of the vessels, their elasticity, vascularization, length, and tear resistance (Luis Manuel, 2018; Randall, 1989; Jordan, 1919).

Previous studies on umbilical cord cannulation by different research groups reported the general feasibility of cannulation procedures within their respective animal models. (Hornick et al., 2016; Charest-Pekeski et al., 2022; Hornick et al., 2018; Schoberer et al., 2012; Arens et al., 2011). With these results, the feasibility in humans was assumed. Hornick et al. suggested to evaluate the possible cannula diameter with sonographic measurements (Hornick et al., 2016). However, the translation of umbilical cannulation results from the animal model to the human neonate has been controversially discussed (Bie et al., 2021). Vasospasm and differences in vessel properties are reported as important challenges (Ch et al., 2021; Bie et al., 2021). With the ex vivo test setup presented in this work, the feasibility of ECMO cannulation in an ex vivo human umbilical cord was shown. The vasospasm of the human umbilical cord and its prevention may be examined in a reproducible laboratory environment under the given limitations by the ex vivo test stand. In addition, it is now possible to translate the results from animal models to a preclinical human setup in order to examine their validity.

Beyond artificial placenta technology (AAPT), the test setup could likewise be used to train and develop treatments and surgical techniques in fetal, perinatal and neonatal care, such as for scenarios like cordocentesis. To adequately train healthcare professionals with cannulation techniques in a high-fidelity setting, the environment can be adapted to such tasks by integrating the test setup into a maternal manikin with an artificial uterus. This way, human patient care practices can be trained, embedding cannulation within broader perinatal procedures rather than isolating it as a standalone skill.

While the medical simulation field already offers artificial umbilical cords (Sawyer et al., 2018), they are intended for neonatal care training. Also, integration of natural umbilical cords within artificial simulation setups has been described (Ghazali et al., 2023). However, in both cases, none exists for training umbilical cannulation for blood circuit establishment with AAPT.

In the process of establishing this novel approach, we had to use different commercially available vascular catheter devices for the different purposes. For the terminal cannulation, a cannula size of 7 Fr (arteries) and 8 Fr (vein) shortened to a length of 6.5 cm was found to be a viable compromise regarding cannulation time, insertion length, and the goal of atraumatic cannulation. Umbilical cord cannulation had to be trained in the sense of a progressive “learning curve”. Since a continuous improvement in outcome, time, and possible cannula sizes was observed during the training on the 28 umbilical cords, the cannulation skill improved. Further improvement could be achieved by, for instance, adding ultrasound guidance during the cannulation or by developing new cannulas specifically designed for umbilical cords. The technique is representative of a clinical skill that can be learned and trained. The total number of 43 terminally cannulated cords suggests the independence of terminal cannulation from umbilical cord anatomy, thereby opening broad possibilities for ethical clinical training and testing on human umbilical cords in a wide span of gestational ages.

AAPT applications aim to be a treatment alternative for extremely preterm neonates. In this study, most umbilical cords were obtained from term-born neonates with a gestational age (GA) above 36+0 weeks. At this GA, the mean vascular diameters are approximately 4.2 mm for the arteries and 8.2 mm for the vein, compared to 3.25 mm and 6.5 mm, respectively, at 24 weeks of GA (Kiserud and Acharya, 2004; Kiserud et al., 2006; Di Naro et al., 2001) Owing to the linear viscoelastic properties of the umbilical cord (Arens et al., 2011), cords from term neonates can still be perfused under the physiological flow and pressure conditions of a younger GA, thereby functionally mimicking earlier developmental stages. This approach avoids the need to harvest cords from preterm neonates and mitigates the associated ethical concerns. The observed flow rates and resulting vessel diameters in our experiments support the validity of this assumption. Since there is no change in the mechanical behavior of the umbilical cord, the same assumption can be made for umbilical cords with abnormalities, such as a single artery. Nevertheless, due to the exclusion of such umbilical cords from the proof-of-concept study, this cannot be verified.

The target flow could be established in all 15 included samples, showing the reproducibility of the setup. Corresponding pressure loss is, however, still supraphysiological and needs to be addressed in a future improved terminal cannulation strategy. VAA and VA pressure loss over blood flow can be described using linear regression (cf. Figure 2) to define operating conditions in future experiments.

ECMO cannulation of at least one artery and one vein was shown in ten out of 15 cases (66%). The fastest cannulation time was measured to be 2.45 min for all three vessels, showing that the required maximum duration of 3 min for cannulation is feasible. However, the sealing and fixation technique of the cannulae using cable ties and compression tape was aimed at a fast and tight connection for this feasibility study and was developed only for the laboratory use of the ex vivo test setup. Clinical application of ECMO cannulae to the umbilical cord and respective strategies against dislocation of the cannulas still need to be developed. The fixation was applied before the vessels were fully dilated, which could create a bending effect and cause higher resistance. The fixation method using a cable tie might have further caused trauma to both the cord and vessel wall. Histopathological examination of the reperfused vessels might clarify the options for optimization of the employed method.

The ECMO cannulation flow of 30 mL/min was chosen based on preliminary experiments to have a pressure load of at least 30 mmHg on the cannulae. The choice is verified by the pressure loss results illustrated in Figure 2. From the linear regression follows a pressure loss of 41.75 mmHg at 30 mL/min, which is slightly higher than the targeted 30 mmHg. Bypassing the ex vivo setup with up to six ECMO cannulae can reduce the pressure loss to 9 mmHg, making physiological operation points possible in this setup. The umbilical cord is perfused continuously instead of in a pulsatile manner as it is in utero. Preliminary tests using a pulsatile pump resulted in several suction events during flow rate ramp-up. The addition of pulsatility might be possible after a sufficient dilation of the vessels and the achievement of the target flow range.

This study developed and demonstrated the successful use of an ex vivo test setup in which human umbilical cords can be effectively and reproducibly cannulated and operated at physiological flow conditions. The ex vivo test setup closes the gap between existing animal models and the human neonate and thereby offers a low-risk and ethical method to investigate cannulation devices and procedures for AAPT applications and umbilical cord-related therapies in general.