Central Venous Catheter Malfunction in Children

Central venous catheters (CVCs) are commonly used in children for maintenance hemodialysis (HD), although they are known to have a high complication rate with poor blood flow, thrombosis, and infections (1,2), as compared with arteriovenous fistulas. It is unknown if specific features of the CVC design contribute to the high complication rate. Computational fluid dynamics (CFD) is an engineering tool for predicting fluid flow inside a geometry, such as a vessel or a medical device, and for in silico measurements of hemodynamics parameters that are difficult or impossible to replicate in vivo or in vitro. Geometries of different dimensions and sizes can be compared to determine the factors that improve or reduce optimal blood flow.

In this multidisciplinary study, we investigate flow characteristics in pediatric CVCs by combining computational simulations and clinical data to provide a comprehensive fluid dynamics characterization of different CVC designs for pediatric applications.

Four models of CVCs that are routinely used across Europe were recreated from microcomputed tomography scans and studied within appropriate-sized anatomic models by means of numeric simulations (Ansys Fluent; Ansys Inc., Canonsburg, PA). The four CVCs, which differed in design of lumen, tip, and configuration of side holes, were the Tesio 6.5F, Hemo-Cath 8F, Pediatric Split Cath 10F, and Split Cath III 14F (MedCOMP, Harleysville, PA).

Computational analyses showed that, after CVC insertion, the velocity of blood flow increased due to a reduced vessel lumen (Figure 1Aa), therefore increasing shear stress on the vein wall (Figure 1Ab). A maximum three-fold increase of shear stress was recorded with the 10F model. With smaller CVC models, asymmetric eddies were identified (Figure 1Ac). Similar changes in hemodynamics due to catheter placement increase the risk of venous thrombosis in adult patients on HD (3). When the catheters were used in arterial configuration, blood entered the lumen mainly through the most proximal side holes, regardless of the CVC design and size. The presence of a low-velocity zone (Figure 1Ad), due to the relatively lower blood flow through the tip, was observed in all CVC designs. This might trigger platelet activation and consequent blood clotting (4). Similar to previous computational studies (4), shear stress levels were measured at the arterial tip of the catheters. Regions close to the proximal side holes recorded the highest shear stresses (Figure 1Ae), with maximum values between 105.25 and 255.43 Pa. For comparison, in physiologic conditions, blood shear stresses range from 0.1 to 1 Pa in veins, 5 Pa in arteries, and 6 Pa in arterioles (5). Large areas of stagnation were recorded in the 6.5F model. The arterial lumen of the Hemo-Cath (8F) recorded the highest percentage of shear stress, above the 10 Pa threshold defined by Mareels et al. (4) in 2007. The average shear stress values were comparable among the four models.

Figure 1.

Reconstructed models of the catheters and results from the computational and clinical studies. (A) Three-dimensional models of the central venous catheters (CVCs) included in this study, with magnification on the details of the side holes. (a) Tesio (6.5F); (b) Hemo-Cath (8F); (c) Pediatric Split Cath (10F); and (d) Split Cath III (14F). (B) Computational fluid dynamics to study changes in hemodynamics parameters in an anatomic model of the superior vena cava (SVC) and the right atrium (RA) in the presence of a Hemo-Cath 8F CVC and under clinical working conditions. Central picture: velocity pathlines for the Hemo-Cath 8F. Blood is aspirated from the arterial lumen in the SVC, whereas the venous lumen takes blood back to the RA. Lateral pictures illustrate several results from different sections of the geometry; color maps are reported with the corresponding legend ranging from the minimum (blue) to the maximum (red) values measured. (a) Velocity contours plotted at the cross-section of the SVC, before and after CVC insertion. Blood velocity inside the vein increases in the presence of the catheter; increases range between 3% and 15% in average velocity and between 21% and 34% in V_max. (b) Wall shear stress plotted on the SVC (front and back views are shown). Increased shear stress is recorded after catheter insertion; the Pediatric Split Cath registered a three-fold greater increase, whereas, in the remaining CVC models, differences ranged between 40% and 48%. (c) Asymmetric eddies (red arrows) are found in the SVC cross-section when the catheter is placed inside the vein. (d) A region of low velocity or blood stagnation is shown at the tip of the arterial lumen. (e) Close up of the shear stress distribution in the region close to the arterial side holes, where higher values of shear stress are measured (red arrows). (Right panel) A region of blood stagnation can also be found in the Split Cath 10F (top) together with high shear stress levels in the region close to the most proximal arterial side holes (bottom). (C) Table summarizing the most important results obtained from both the engineering simulations and the observations in clinical patients. Maximum blood velocity was measured at the smallest cross-section of the vein before and after catheter placement, while shear stress values were computed in a volume of fluid containing the arterial tip of the catheters. SS, shear stress.

In a pilot study of children on HD, we evaluated outcomes of 57 CVCs from time of insertion. All patients had one of the above four catheters, depending on size. Clinical observations showed that thrombosis (defined as the need for thrombolytic treatment at three consecutive HD sessions, or CVC removal due to occlusion) was the first type of CVC dysfunction in 92% of cases. At 90 days, the estimated thrombosis-free survival was 35%, 60%, and 100% in the Hemo-Cath, Split Cath, and Split Cath III groups, respectively (P=0.01). Of the CVCs removed, 89% presented at least one event of catheter dysfunction (defined as the composite end point of thrombosis or infection). Infection and thrombosis were the main causes of CVC replacement, accounting for 30% and 27% of CVC removals, respectively. Dysfunction-free survival was 21%, 49%, and 66% at 90 days in the Hemo-Cath, Split Cath, and Split Cath III groups, respectively (insufficient data were available for the Tesio). Although the smallest CVCs had the shortest dysfunction-free survival, this did not reach statistical significance (P=0.06), implying that other, potentially modifiable, features of the CVC design may also play a role in determining catheter function. At the time of catheter dysfunction, the blood flow rate within the CVC was only 77% of the highest achievable flow (P<0.001). Infection rates were comparable among groups.

Results from the computational study were in accordance with our pilot clinical data (Figure 1C). There was a significantly higher rate of thrombosis in the Hemo-Cath compared with the split-tip models (P=0.01), which could not be accounted for by the smaller lumen diameter of Hemo-Caths and was consistent with the computed shear stress: 43% of values were >10 Pa versus the 24% found in the Split Cath and 25% in the Split Cath III.

In conclusion, this proof-of-concept study identified critical fluid dynamic parameters of CVCs that correlated with adverse clinical outcomes. Findings will be verified in large, multicenter clinical studies. Future work will use computational fluid dynamics to design new CVCs optimized for children.

Disclosures

R. Shroff reports having consultancy agreements with AstraZeneca; receiving honoraria from, and serving on a speakers bureau for, Amgen and Fresenius Medical Care; receiving consultancy and advisory board honoraria from AstraZeneca, Fresenius Medical Care, and Humacyte; and receiving research grant funding from Fresenius Medical Care. All remaining authors have nothing to disclose.

Funding

This work is supported by the Kidney Research UK Paediatric Innovation grant Paed_IN_004_20190926.

Author Contributions

C. Bruno, C. Capelli, and R. Shroff conceptualized the study; C. Bruno and R. Moumneh were responsible for data curation and investigation; C. Bruno, R. Moumneh, and E. Sauvage were responsible for formal analysis; C. Capelli and R. Shroff were responsible for funding acquisition and project administration; C. Bruno, C. Capelli, and R. Moumneh were responsible for methodology; O. Arthurs, I. Simcock, L. Stronach, and K. Waters were responsible for resources; C. Bruno was responsible for visualization; C. Capelli, E. Sauvage, S. Schievano, and R. Shroff provided supervision; C. Bruno, C. Capelli, and R. Shroff wrote the original draft; and O. Arthurs, C. Capelli, R. Moumneh, E. Sauvage, S. Schievano, R. Shroff, I. Simcock, L. Stronach, and K. Waters reviewed and edited the manuscript.