Journal O\ CIinical Pharmacy and Therapeutics (1994) 19, 199-202

A comparison of the endotoxin-retentive abilities of two '96-h' in-line intravenous filters C. Richards* and P. F. Grassbyt 'Sterile Product Unit, University Hospital of Wales, Card$, and tPharmaceutica1 Unit, St M a y ' s Hospital Penarth, Cardif U.K.

SUMMARY

The ability of two in-line intravenous filters, the Pall Intravenous Set Saver Filter ELD96 and the Codan LV. S SET-P, to retain endotoxin released by Pseudomonas amginosa during filtration of amino-acid-based parenteral nutrition (PN) solu- tions was investigated. When challenged with 10' cells of P. aeruginosa during simulated clinical PN infusion, the Pall ELD96 was shown to be capable of providing an effluent free of detectable endo- toxin (<0-3125 EWml). In contrast, the Codan I.V. S SET-P allowed significant amounts of endotoxin to pass through.

INTRODUCTION

The use of intravenous (i.v.) fluid filters as a routine part of i.v. therapy has been shown to reduce a variety of complications which can arise from particulate, microbial and air infusion. With conventional filters, however, it is important that the filter and all admin- istration equipment is changed every 24 h as trapped Gram-negative bacteria release significant amounts of endotoxin, which is not retained. Infusion of endo- toxin produces pyrogenic, toxic and inflammatory reactions (I).

The Pall Intravenous Set Saver Filter (ELD96, Pall Biomedical), which contains a membrane called Posidyne@, was developed to retain endotoxin for 96 h. This enabled the life of the i.v. filters and all administration equipment upstream of the filter to be safely extended to 96 h, which can lead to cost savings on administration equipment.

Several studies have confirmed that the ELD96 filter retains endotoxin, and provides 96 h of protection, after infusion with a range of i.v. solutions contami- nated with Escherichia coli, Pseudomonus aeruginosa and Klebsiella pnetrrnoniae (2, 3).

The ELD96 has also been demonstrated to retain P. aeruginosa associated endotoxin in infant parenteral nutrition (PN) solutions (4). which have been shown to support the growth of bacteria and fungi. This has resulted in the use of ELD96 filters for the administra- tion of PN fluids (which do not contain lipid) to infants in the special care baby unit of the University Hospital of Wales. The serious consequences of bacterial con- tamination of infant i.v. administration lines were therefore reduced.

Recently another filter, Codan I.V. S SET-P, which contains a different membrane (polysulphone), has become available which claims a 96-h use period. The ability of the Codan N filter to retain endotoxin for 96 h has not been independently investigated. The aim of this study was to determine the ability of this filter to retain endotoxin, compared with the Pall ELD96 and a conventional filter which does not retain endo- toxin, under simulated clinical conditions during filtra- tion of PN solutions inoculated with P. aeruginosa.

MATERIALS AND METHODS

Filters

The filters studied were the Pall Intravenous Set Saver (ELDS96LL; Pall Biomedical Ltd, Portsmouth, U.K.), the Codan I.V. S SET-P (Medizinsche Gerate GmbH, Lensahn, Germany) and the Ivex-I1 (Millipore SA, 67 Molshein, France).

Parenteral nutrition solufion

The PN formulation used is shown in Table 1.

Preparation of bacterial inoculum

A suspension of P. aeruginosa NCIB 8626 was prepared by inoculating 100d tryptone soya broth with a

199

200 C. Richards and P. F. Grassby

Table 1. Formula for infant PN solution (300 ml)

Constituent Concentration

Nitrogen Glucose Sodium Potassium Calcium Magnesium Phosphate Chloride Zinc CGpper Vitamins

0.5 g 120.0 kcal 10.2 mmol 5.9 mmol 1.2 mmol 0.3 mmol 1.4 mm01 16.4 mmol

7 7 pmol 0.5 pmol 2.0 ml

single colony from a tryptone soya agar plate and incubating this culture overnight in a shaking water bath at 30°C. Aliquots of the overnight culture were centrifuged (2,600 g for 10 min). The supernatant was discarded and the cell pellet resuspended in sufficient PN solution to give a cell suspension of approximately 1 x lo7 bacteridml.

The total bacterial count was assessed using a counting chamber (Neubauer haemocytometer-BDH Ltd, Dagenham, Essex, U.K.) and then the cell suspen- sion diluted appropriately with PN solution to give lo7 cells/ml. This represented the working challenge suspension.

Evaluation of filter endotoxin retention

A 360-d bag of PN solution was aseptically con- nected to an i.v. solution administration set flravenol Laboratories Ltd).and a filter. The administration set and filter were primed with infusion solution. The administration solution was inserted into an i.v. solu- tion pump (Ivac 531, Ivac UK, Harrow, U.K.) and the pump was set to deliver at a flow rate of 10 &. The outlet tubing of the filter was placed into a IO-ml sterile, pyrogen-free container, the pump was started and approximately 2 ml of infusion solution was collected in this container. The pump was stopped and, using a sterile syringe and needle, a 3.6-ml bolus of the 1 x lo7 P. ueruginosalml was injected into the PN bag. The bag was mixed well. Flow through the filter was initiated at 10 ml/h and the effluent from the filter

collected in a I-litre sterile, pyrogen-free container. At I, 6, 24, 48, 72 and 96 h, 2-ml filtrate samples were collected in 10-ml sterile, pyrogen-free containers. The 2-ml samples were subsequently assayed for endotoxin levels.

Immediately after initiating flow through the filter, a I-ml sample was withdrawn from the bag of PN solution. This sample was serially diluted and 0.I-ml aliquots of appropriate dilutions plated onto tryptone soya agar and incubated for 48 h at 30°C. The number of colony forming units (cfu)/ml was counted after incubation. This represented the initial viable count of bacteria in the bag of PN solution. After a 24-h flow of 250 ml PN solution through the filter, a further I-ml sample was withdrawn from the bag of PN solution. This sample was serially diluted and 0-I-ml aliquots of appropriate dilutions were plated onto tryptone soya agar and incubated (48 h at 30°C). The number of &ml was counted after incubation. This represented the viable count of bacteria in the bag of PN solution after 24 h.

The 360-ml PN solution bags (plus 3.6 x lo7 P. aeruginosu) and I-like sterile, pyrogen-free containers were replaced every 24 h for 96 h. The viable counts were also repeated for each infusion bag.

Limulus amoebocyte lysate assay for detecting endotoxins

All endotoxin assays were performed using glass- ware depyrogenated by heating at 260°C for 4 h. Filtrate samples (0.4 rnl) were mixed with 0.1 ml Limulus amoebocyte lysate (Atlas Bioscan Ltd, Bognor Regis, West Sussex, U.K.) reconstituted with pyrogen-free water (Atlas Bioscan Ltd) and incubated in a LAL5000 endotoxin assay module (Atlas Bio- scan Ltd) at 37°C. The rate of turbidity development (which increases with increasing endotoxin concen- tration) was measured for each sample. Positive controls were performed, using a range of E. coli control standard endotoxin concentrations at 0.03125, 0.0625, 0.125, 0.25, 0.5, 1, 5, 10 and 50 endotoxin units (EU)/ml. The sensitivity of the lysate was confirmed as 0.0325 EU/ml. Negative controls were performed using the PN solution. As P. aerugi- nosu was used as the challenge organism and as the endotoxin released is different from the E. coli con- trol standard endotoxin, all values are quoted as endotoxin unit-equivalents (EUE).

lnfravenous filters and endotoxin retention 201

Table 2. Endotoxin concentration (EUUml) downstream of test filters Hours ELD96 ELD96 ELD96 Codan Codan Codan Ivex H.P.

*=<0.3125 EUWml.

RESULTS

Control experiments demonstrated that the PN solu- tion analysed inhibited the LAL assay used. Diluting the PN solution 1:10 overcame this inhibition. Thus, all test samples were subsequently diluted at this level and the dilution factor taken into account when calculating the results.

The data obtained when challenging the filters with P. aeruginosa are shown in Table 2. In three separate experiments no endotoxin PO-3125) was found downstream of the ELD96 filters. In contrast, the Codan I.V. filter allowed significant amounts of endo- toxin to pass through in all three experiments. The h e x H.P. filter was used as a control to confirm that endotoxin was detectable downstream of a filter containing a non-endotoxin-retentive cellulose acetate membrane.

The total bacterial challenge for each filter tested was approximately 10'. This was calculated by multi- plying the daily bacterial challengdml (approximately 1 x 10s/ml) by the infusion volume (250 ml) by the number of days challenge (4). The viable count was found to decrease over each 24-h period, indicating that a proportion of the bacteria were dying.

P. aeruginosa was used in this study because it has been shown to be impractical to use E. coli as the challenge organism as the growth of this organism in PN is so variable that controlled experiments are impossible (4).

DISCUSSION

The data reported here for the Pall ELD96 comple- ment those obtained by other workers, who have demonstrated endotoxin retention for up to 96 h from E. coli in 5% dextrose (2), E. coli and Klebsiella pneumo-

nie in 5% dextrose, and 0.9% NaCl(3) and P. ueruginosa in paediatric total parenteral nutrition solutions (4).

The level of bacterial loading used in these experi- ments was chosen to complement other workers who consider I x 10' as a suitable challenge (2). This level is high, but it is achieved in clinical practice.

There was considerable variation in the perfor- mance of individual Codan filters, which also appeared to perform worse than the control filters. As both the Codan filters and the control filters have no specific mechanism to remove endotoxin, it is possible that some endotoxin retention occurred as a result of the physical entrapment of endotoxin aggregates in the filter media. Variability in the performance is possibly due to the size and degree of endotoxin aggregates present. These will be markedly affected by slight changes in such factors as pH and ionic strength of the solution.

In addition to the improvement in patient care afforded by a 96-h endotoxin-retentive filter, such a filter will provide considerable cost savings as only one administration set is required during the life of the filter. In several studies in this country and others, savings of f 8,637-f 35,000 have been estimated (5-7). However, this may be an underestimate of the potential savings as the cost savings achieved by the prevention of the clinical complications of i.v. therapy remain undocumented.

The Codan LV. S SET-P Infusion filter retails at considerably less than the PALL ELD96, and offers potentially greater savings. The filter is being mar- keted with a 96-h life, and the company have stated that 'it can be left in siftc for 96 hours'. The results of this study demonstrate that this filter does not afford protection from endotoxin release for longer than 24 h at the endotoxin challenge level used in this experi- ment. The Pall ELD96 filter did retain endotoxin at this

202 C. Richards and F. F. Grassby

challenge level for 96 h. As discussed, endotoxin infusion will give rise to a range of complications, which will need to be countered. This could take the form of increased doses of antibiotics, more aggressive clinical investigations and possible removal or resiting of the i.v. line; thus, irrespective of the deleterious effects on the patient the cost of treatment will be high.

Finally, this represents a new area in which pharma- cists can develop an active expertise in the selection and usage of i.v. medical devices by independent investigations.

REFERENCES

1. Wardle, EN. (1986) Endotoxins and respiratory and renal failure. Care of the Critically 111, 2, 54-56.

2. Baumgartner, TG, el al. (1986) Bacterial endotoxin retention by in-line intravenous filters. Arnericun Ioumal of Hospital Pharmacy, 43, 681484. Spielberg R. (1985) Evaluation of the endotoxid bacterial retention capabilities of IV filters during simu- lated extended infusions. Pall Technical Report IV-IDOI. Richards C, Thomas P. (1990) Use of endotoxin reten- tive intravenous filters with paediatric total parenteral nutrition solutions. ]oumal of Clinical Pharmacy and lhrapeutics, 15, 53-58.

5. Cousins D. (1988) Cost savings in IV therapy. Care of the Critically 01, 4 , 1-4.

6. Barber N, Jacklin A. (1987) CCU drug costs-the pharmacists role. International Cure World, 4, 80-82.

7. Stromberg G, Wahlgren J. (1989) Saving money with effective inline filters. Intensive Cure Nursing, 5 , 109-113.

3.

4.