Perfusion http://prf.sagepub.com
Tubing failure during prolonged roller pump use: a laboratory study Giles J Peek, Kim Wong, Colin Morrison, Hilliary M Killer and Richard K Firmin Perfusion 1999; 14; 443 DOI: 10.1177/026765919901400607 The online version of this article can be found at: http://prf.sagepub.com/cgi/content/abstract/14/6/443
Published by: http://www.sagepublications.com
Additional services and information for Perfusion can be found at: Email Alerts: http://prf.sagepub.com/cgi/alerts Subscriptions: http://prf.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations (this article cites 6 articles hosted on the SAGE Journals Online and HighWire Press platforms): http://prf.sagepub.com/cgi/content/refs/14/6/443
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 443
Perfusion 1999; 14: 443–452
Tubing failure during prolonged roller pump use: a laboratory study Giles J Peek Heartlink ECMO Centre, Department of Cardiothoracic Surgery, Glenfield Hospital, Leicester, Kim Wong, Colin Morrison Department of Engineering, University of Leicester, Hilliary M Killer and Richard K Firmin Heartlink ECMO Centre, Department of Cardiothoracic Surgery, Glenfield Hospital, Leicester
Little is known about the mechanical forces acting on extracorporeal circuit tubing with prolonged roller pump use during extracorporeal membrane oxygenation (ECMO). We examined the time to tubing rupture of three different materials during actual roller pump use, mean and standard deviation (SD) (SD shown in parentheses): Tygon® (control) 243.7 h (175.4); LVA 121 h (14.3); and SRT 6.6 h (2.1). Failure times for both LVA and SRT were significantly different from the control (paired t-test, p = 0.02 and p < 0.001, respectively). The minimum failure times for Tygon and LVA were 99 and 101 h, respectively. We then examined Tygon under conditions of pure compression, demonstrating that even after 3.67 million compression cycles at full occlusion crack formation did not occur. If the tubing was overoccluded, cracks appeared within 24 h. Scanning electron microscopy of Tygon, which has been used during clinical ECMO, and the failure pattern during destruction testing demonstrate that shear stress and compression coexist during clinical ECMO. Use of under-occlusive pump settings could improve tubing life.
Address for correspondence: GJ Peek, Heartlink ECMO Centre, Department of Cardiothoracic Surgery, Glenfield Hospital, Groby Road, Leicester LE3 9QP, UK. E-mail:
[email protected] © Arnold 1999
0267–6591(99)PF326OA
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 444
444
GJ Peek et al.
Background The DeBakey roller pump1 is an elegant solution to the problem of propelling blood in the extracorporeal circuit. Unfortunately, roller pumps can cause significant blood damage and haemolysis2 if improperly calibrated (over-occluded). Tubing wear also results in liberation of particles from the tube wall into the blood (spallation). Spallation is an important source of the microemboli observed during perfusion.3,4 During the prolonged use of the roller pump during extracorporeal membrane oxygenation (ECMO), raceway tubing may actually rupture leading to air embolism and haemorrhage. In this paper, we present two experiments which give some insights into the mechanical forces which cause raceway rupture during prolonged roller pump use.
Experiment 1: destruction testing Introduction The raceway is the piece of tubing within the pump housing (boot) which is swept by the rollers. The gap between the rollers and the pump boot can be adjusted (occlusion). Maximum forward flow occurs at full occlusion, i.e. when the tubing is completely compressed. Unfortunately, full occlusion may also result in significant blood damage,2,5 as may gross under-occlusion. The ideal occlusion setting for tubing used for ECMO is said to be slightly under-occlusive. This is achieved in our centre, as in most ECMO centres, by gradual reduction in occlusion until air just refluxes past the rollers. Such under-occlusion has been shown to reduce haemolysis.2 The factors that may influence tubing life are not fully understood, but are thought to include occlusion settings, speed of revolution, pressure and temperature.6 In this experiment the longevity of two potential new ECMO tubings, LVA (Portex 800-500-575, Portex Industries, Hythe, Kent, UK) and SRT 620 (Rehau UK, Langley, Slough, UK), are compared to that of the currently used ECMO tubing, Tygon (Tygon® S-65-HL; Northern Performance Plastics, Corby, Northamptonshire, UK). Aims To measure the time to tubing rupture of Tygon, LVA and SRT under conditions exceeding those seen during clinical perfusion.
Materials and methods A closed-loop circuit was primed with Hartmann’s solution: the same circuit was used for all experiments with the raceway being replaced each time (Figure 1). The pump was servo-regulated by a bladder box and controller (Seabrook, Cincinnati, OH, USA), and by a moisture sensor (manufactured in-house) which was placed in the well of the pump. In the event of raceway rupture, fluid on the sensor caused the pump to stop and also stopped an integral timer. The two potential new ECMO tubings: LVA and SRT 620 were compared to Tygon® S-65-HL, which served as a control. The end-point was the time to tubing rupture. Five runs were performed for each material. The heater (Cincinnati Sub-Zero, Cincinnati Products Inc., Cincinnati, OH, USA) was adjusted to maintain a fluid temperature of 37°C, as we have found that failure times at room temperature are often much longer than those obtained at 37°C. Full occlusion was set at the beginning of each run and then not adjusted. Full occlusion was obtained by adjusting the rollers such that no drop in a 60-cm column of fluid occurred in 2 min. The pump (Stockert GmbH, Munich, Germany) speed was set at 200 rev/min, as it is rare in clinical practice to exceed 130 rev/min. The post-pump target pressure was 400–600 mmHg. As the revolutions per minute could not be altered, this was manipulated by infusion and withdrawal of fluid from the circuit and the use of a gate clamp to vary the post-pump resistance. In the event, it proved difficult to regulate the circuit pressure. If attempts were made to alter the circuit volume beyond a few millilitres, severe vibration developed immediately after the circuit was over- or underfilled. This was obviously exacerbated by the extremely high pump speeds used. Thus, the final circuit pressure was determined by the compliance of the individual tubing materials. Observations and adjustments were made daily. It was not possible to conduct experiments in a random order due to the availability of tubing materials and concurrent in vivo experiments. An electromagnetic flow meter (St Jude Delphin, Boston, MA, USA) was used for the second half of the experiment. Statistical methods Failure times for LVA and SRT were compared to the control, Tygon, using the paired Student’s t-
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 445
Tubing failure during prolonged roller pump use 445
Figure 1 Destruction testing circuit
test. A p value of < 0.05 was taken to indicate significance. Correlation between starting pressure and failure time was assessed using Spearman’s test. Results Failure time. Times to tubing rupture, mean and standard deviation (SD) (SD shown in parentheses): Tygon (control) 243.7 h (175.4); LVA 121 h
(14.3); and SRT 6.6 h (2.1). Failure times for both LVA and SRT were significantly different from the control (paired t-test, p = 0.02 and p < 0.001, respectively). The minimum failure times for Tygon and LVA were 99 and 101 h, respectively. Type of failure. Tubing failed in three ways (Figure 2). Failure occurred through the longitudinal
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 446
446
GJ Peek et al.
crease built up by the repeated folding of the tubing in the pump, through transverse tears, presumably caused by shearing and through scallopshaped defects in the inlet side of the raceway. All five pieces of SRT failed through scallops. All five pieces of LVA failed via creases, with one piece failing simultaneously via a tear. Three pieces of Tygon failed via a crease and two pieces via a tear. Reduction in flow with time. The flow meter was only available from half-way through the experiments; therefore, only three test runs with Tygon and the five pieces of SRT were examined. Because all the SRT samples failed in the first 9 h, meaningful results for flow reduction over time were only obtained with the Tygon. It can be seen from Figure 3 that flow gradually decreased over time despite a constant number of rev/min. This was presumed to be due to a progressive loss in the tubing’s ability to resume a circular shape during ‘diastole’, thereby reducing the stroke volume. Pressure. As previously discussed, it proved impossible to control the pressure at such high flow rates. As shear stress seems to be the major factor
Figure 2
in tubing wear (see experiment 2 below), it was decided that maintaining a constant high pump speed was more important than controlling the circuit pressure. It would be possible to design a circuit with a large venous reservoir and a variable height fluid column post-pump in which the pressure could be accurately controlled. The starting circuit pressure for Tygon was 641 mmHg (mean) with an SD of 191, falling to 373 mmHg at tubing rupture for the last piece of Tygon to rupture. The starting pressure for LVA was 618 mmHg (mean) with a SD of 58.6, increasing slightly to 638 mmHg at tubing rupture for the last piece of LVA to fail. The mean starting pressure for SRT was 600.2 mmHg with a SD of 20.36. The failure times did show an inverse relationship to starting pressure, with a Spearman’s r = 0.7, but this correlation was not statistically significant (p = 0.094). Discussion The failure times for the three materials were significantly different. The control material, Tygon, was the least predictable with failure times ranging from 99 to 496 h. The manufacturer, Norton Performance Plastics (Sue Hinson, personal communication), rec-
Types of failure
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 447
Figure 3
Flow vs time for Tygon S-65-HL
Tubing failure during prolonged roller pump use 447
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:26 pm Page 448
448
GJ Peek et al.
ognizes that many factors can contribute to the variation in pump life observed, including material components. The mixing and extrusion processes are always consistent, but lot-to-lot variation can occur. Their own in-house peristaltic pump tests are documented as guidelines. The failure time for the LVA was more predictable, being close to the mean of 122 h. The minimum failure time being 101 h, similar to the 99 h minimum for Tygon. SRT is an order of magnitude less durable than either Tygon or LVA, with no run exceeding 9 h. A further series of destruction runs was started with the SRT with gradually decreasing occlusion, but still carried out at 200 rev/min. Unfortunately, the pump seized up following two runs, and further experiments were cancelled. Under-occlusion to give flow rates of −5% and −13% of fully occluded flow resulted in failure times of 7 and 5 h, respectively. These times are similar to those seen with full occlusion. There appeared to be two failure mechanisms, resulting in three different types of tubing failure. First, folding of the tube across its axis resulting in cycles of compression and tension causes the tube to fail along a longitudinal crease. This was the predominant type of failure for the LVA, but was also evident in three of the five pieces of Tygon. This mechanism is also responsible for the loss of restitution shown by the gradual diminution of flow with time, and the eventual oval shape of the tubing at the end of each run. This type of failure is the mode 1 failure discussed in experiment 2, and has been previously described as a failure mechanism in Tygon S65-HL.7 The other types of tubing failure through transverse tears or ‘scallops’ on the inlet side of the tubing are most likely due to shearing forces. Shear is set up by the rotation of the rollers over the tubing, which is held still by the shoes and clamps of the pump boot. Further shear is caused by friction between the outer wall of the tubing and the pump boot. This mode 2 failure seems to be an important mechanism in the failure of Tygon, occurring in two out of five test runs, and is also evident on pieces of Tygon tubing used clinically (see experiment 2). The circuit pressure seemed to be controlled by the compliance of the tubing materials. This can be surmised from the similarity of pressures between materials, SRT around 600 mmHg, LVA 600–700 mmHg and Tygon again showing the most variability at 400–800 mmHg. It could be argued that the Tygon durability was variable because the pressures
were variable, or that pressures and durabilities were both related to the mechanical properties of each piece of tubing, which are known to be variable. This question would be difficult to answer due to the variability of the tubing, but an experiment with servo-control of the circuit pressure is possible, and large numbers would be required. There is an inverse correlation between median circuit pressure and failure time for Tygon, but this does not reach statistical significance.
Experiment 2: mechanical testing Introduction There is little published work concerning the forces generated in raceway tubing during roller pump use. A fuller understanding of the exact mechanisms of tubing failure would seem to be essential if tubing rupture is to be prevented. Aims 1) To measure the number of load cycles required to cause tubing failure due to compression alone, in the absence of shear. 2) To examine by scanning electron microscopy (SEM) tubing used clinically in order to determine which forces have been acting on the material. Materials and methods Compression testing. A test rig was constructed to generate pure compressive forces in a piece of 1/2inch Tygon S-65-HL tubing (see Figure 4). The apparatus consisted of a circuit filled with Hartmann’s solution. Fluid was pumped through the piece of Tygon under test by a magnetically coupled pump (RS Supplies, Corby, Northamptonshire, UK) at a flow rate of 5.6 l/min, confirmed by a flow sensor. Fluid was maintained at 37°C by a servo-regulated heating coil placed around a part of the circuit which was made from aluminium tubing. Inlet pressure (to the RS pump) was maintained at 50 mmHg and outlet pressure at 200 mmHg. The tubing under test was compressed by a plunger on a moving arm at a rate of 400 times/min (equivalent to 200 rev/min with two rollers). Full occlusion was set by calculation of the required occlusion from measurements of the tubing wall thickness, and confirmation by pressure transduction and the closed-circuit television system (CCTV). Pressure, temperature and flow were
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:27 pm Page 449
Figure 4
Apparatus for compression testing
Tubing failure during prolonged roller pump use 449
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:27 pm Page 450
450
GJ Peek et al.
recorded continuously by chart recorder, evolution of cracks was recorded by time-lapse CCTV. The tubing compliance was recorded daily. SEM. A piece of 1/2-inch Tygon S-65-HL that had been used during a clinical ECMO run for approximately 48 h at 75 rev/min was examined. The tubing was taken and examined with the naked eye. The portion with the most visible wear marks was cut out as a specimen and a mark was made parallel to the fatigue cracks (i.e. the crease). The specimen was frozen in liquid nitrogen and then struck on the mark with a hammer, fracturing it along the crease. The tube was thawed to room temperature, coated with gold film and then examined by scanning electron microscope.
reached a plateau. This was after 3.67 million stress cycles, equivalent to 305.8 h at 200 rev/min. Tubing was then cut and examined macro- and microscopically. No cracks were visible; the tubing had not failed. A further experiment was performed with the tubing over-occluded, fatigue cracks developing within 24 h. SEM. A representative electron micrograph is shown in Figure 6. The bottom of the picture is the luminal surface of the tube, and the roller was moving left to right. The dominant features in this specimen are the striations extending from the luminal surface of the tube and swaying towards the right. These marks are evidence that shearing force is acting on the tubing, in addition to the compressive forces.
Results Compression testing. Compression testing was continued until the compliance curve (Figure 5)
Figure 5
Compliance testing
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:27 pm Page 451
Tubing failure during prolonged roller pump use 451
Figure 6
A scanning electron micrograph of crack-fractured Tygon S-65-HL
Discussion Compression testing. Engineers use the word ‘failure’ in the context of fatigue testing to indicate a loss of the original mechanical properties of a material. Clinicians and perfusionists, however, would not consider the tubing to have failed until rupture occurred or looked inevitable. We know from clinical tubing use and the SEMs that fatigue cracks are present in Tygon after 24–48 h. Therefore, strictly speaking, the tubing has ‘failed’. However, in experiment 1, the tubing functioned for a minimum of 99 h before rupture occurred. The fact that the tubing can be compressed and released 3.67 million times without developing fatigue cracks indicates that if shear forces can be eliminated or reduced, tubing life is likely to be greatly prolonged. Other groups have shown that shear is also the force responsible for much of the blood damage in both roller pumps and centrifugal pumps.5,8 This has led to the development of nonocclusive roller pumps, like the Michigan/Affinity pump (Avecor) and the Collin-Cardio pump (Healthcare Materials SA). These pumps reduce shear by eliminating the pump boot, the raceway being stretched over the rollers; tube life would theoretically be much longer in these devices. However, there are limita-
tions with nonocclusive peristaltic pumps as to which types of tubing they can use, as the raceway must be distensible. The Michigan pump uses a polyurethane raceway constructed from two flat sheets formed like a ‘Popsicle’ packet. The CollinCardio pump uses oval silicone tubing. A tube life of 121 h has been recorded with the Collin-Cardio pump in our laboratory for a standard silicone material that would be expected to last 6–12 h in an occlusive roller pump. The characterization of the fatigue life of polymeric materials, based on stress or strain amplitudes, is much the same as metals. The formation of long-chain molecules is the main feature of the structural constitution of polymeric materials. Under cyclic loading, polymers display deformation modes. Crazing and shear flow are the most common modes of deformation during the fatigue process. While crazing has the connotation of brittle failure, shear banding represents the more ductile process typical of the types of polymer used in tubing.9 SEM. The fact that the tubing, which was loaded by pure compressive (mode 1) forces, did not fail indicates that this mechanism does not initiate cracks. The presence of skewed cracks in the clini-
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
07pf326.qxd 19/10/99 1:27 pm Page 452
452
GJ Peek et al.
cal tubing specimen indicates that longitudinal forces (shear or mode 2) play an important part in crack initiation and propagation.
Conclusions The overall conclusion of this mechanical comparison of Tygon, LVA and SRT is that Tygon is the superior material for ECMO tubing. However, its mechanical properties are variable. LVA seems to have sufficient durability to be considered possible ECMO tubing. Its failure time seemed consistent, with a minimum failure time similar to that seen with Tygon. SRT is not viable ECMO tubing. Spallation and biocompatability profiles would also have to be examined for LVA before it can be recommended for clinical ECMO. There appear to be two types of forces acting on the raceway tubing, resulting in failure: shear stress and compression. Compression alone was not able to initiate crack formation in Tygon. Scanning electron microscopy demonstrates the coexistence of both mechanisms during clinical tubing use, evidence further supported by the failure of Tygon by both mechanisms. Improvements in pump design to reduce shear stress, and use of under-occlusion to reduce compressive forces, might result in improved tubing durability.
References 1 DeBakey ME. Simple continuous-flow blood transfusion instrument. New Orleans Med Surg J 1934; 87: 386. 2 Berstein EF, Gleason LR. Factors influencing hemolysis with roller pumps. Surgery 1967; 61: 432–42. 3 Page US, Bigelow JC, Carter CR, Swank RL. Emboli (debris) produced by bubble oxygenators. Removed by filtration. Ann Thorac Surg 1974; 18: 164–71. 4 Solis RT, Kennedy PS, Beall AC Jr, Noon GP, DeBakey ME. Cardiopulmonary bypass. Microembolization and platelet aggregation. Circulation 1975; 52: 103–108. 5 Bernstein EF, Blackshear PL Jr, Keller KH. Factors influencing erythrocyte destruction in artificial organs. Am J Surg 1967; 114: 126–38. 6 Mortensen J, Bagley B, Cosgrove W, Yates W, Daniels A. Durability of tubing subjected to roller pumps in extracorporeal blood perfusion systems. Artif Organs 1979; 442–47. 7 Snyder EJ, McElwee DL, Hassan HM et al. Investigation of fatigue failure of S-65-HL ‘Super Tygon’ roller pump tubing. J Extracorp Technol 1996; 28: 79–87. 8 Pedersen TH, Videm V, Svennevig JL et al. Extracorporeal membrane oxygenation using a centrifugal pump and a servo-regulator to prevent negative inlet pressure. Ann Thorac Surg 1997; 63: 1333–39. 9 Suesh S. Fatigue of materials. Cambridge: Cambridge University Press, 1991.
Acknowledgements The authors wish to thank Mr R Reeves ACP, Department of Perfusion, Glenfield Hospital. Financial support for this study came from the British Heart Foundation, Norton Performance Plastics, Rehau and Sir Samuel Scott of Yews Trust.
Downloaded from http://prf.sagepub.com at PENNSYLVANIA STATE UNIV on April 10, 2008 © 1999 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.