Ultrafiltration and backfiltration during hemodialysis

Ultrafiltration is the pressure-driven process by which hemodialysis removes excess fluid from renal failure patients. Despite substantial improvements in hemodialysis technology, three significant problems related to ultrafiltration remain: ultrafiltration volume control, ultrafiltration rate control, and backfiltration. Ultrafiltration volume control is complicated by the effects of plasma protein adsorption, hematocrit, and coagulation parameters on membrane performance. Furthermore, previously developed equations relating the ultrafiltration rate and the transmembrane pressure are not applicable to high-flux dialyzers, high blood flow rates, and erythropoietin therapy. Regulation of the ultrafiltration rate to avoid hypotension, cramps and other intradialytic complications is complicated by inaccurate estimates of dry weight and patient-to-patient differences in vascular refilling rates. Continuous monitoring of circulating blood volume during hemodialysis may enable a better understanding of the role of blood volume in triggering intradialytic symptoms and allow determination of optimal ultrafiltration rate profiles for hemodialysis. Backfiltration can occur as a direct result of ultrafiltration control and results in transport of bacterial products from dialysate to blood. By examining these problems from an engineering perspective, the authors hope to clarify what can and cannot be prevented by understanding and manipulating the fluid dynamics of ultrafiltration.     

Transmembrane passage of cytokine-inducing bacterial products across new and reprocessed polysulfone dialyzers

The widespread use of bicarbonate dialysate and reprocessed high-efficiency and "high-flux" dialyzers has raised concerns about the increased risk of reverse-transfer of dialysate contaminants into the blood compartment. To evaluate this concern, the reverse-transfer of bacterial products from contaminated bicarbonate dialysate into the blood compartment was compared during in vitro dialysis with new or reprocessed high-flux polysulfone dialyzers. In vitro dialysis was carried out at 37 degrees C by use of a counter-current recirculating loop dialysis circuit with either new high-flux polysulfone dialyzers or dialyzers reprocessed once or 20 times with formaldehyde (0.75%) and bleach (< 1%) with an automated system. Heparinized whole blood from healthy volunteers was circulated through the blood compartment, and bicarbonate dialysate was circulated in the dialysate compartment. The dialysate was challenged sequentially by 1:1000 and 1:100 dilutions of a sterile Pseudomonas aeruginosa culture supernatant (bacterial challenge). Samples were drawn from the blood and dialysate compartments 1 h after each challenge. Peripheral blood mononuclear cells (PBMC) were harvested by Ficoll-Hypaque separation from whole blood in the blood compartment and a 5 x 10(6) PBMC/mL cell suspension was prepared. Likewise, dialysate samples (0.5 mL) were added to 0.5 mL suspension of 5 x 10(6) PBMC/mL drawn at baseline. All PBMC suspensions were incubated upright in a humidified atmosphere at 37 degrees C with 5% CO2 for 24 h, and total interleukin-1 alpha (IL-1 alpha) and tumor necrosis factor-alpha (TNF alpha) cytokine production (cell-associated and secreted) was measured by radioimmunoassay. Eight experiments were performed for each arm of the study with the same donor for each arm. One hour after contaminating the dialysate with a 1:1000 dilution of the bacterial challenge, IL-1 alpha production by PBMC harvested from the blood compartment was 160 +/- 0, 171 +/- 11, and 270 +/- 35 pg, respectively, for new dialyzers, dialyzers reprocessed once, and dialyzers reprocessed 20 times (P = 0.004). One hour after challenging the dialysate with 1:100 dilution, IL-1 alpha production by PBMC harvested from the blood compartment was 188 +/- 20, 228 +/- 35, and 427 +/- 67 pg, respectively, for new dialyzers, dialyzers reprocessed once, and dialyzers reprocessed 20 times (P = 0.006). IL-1 alpha production by PBMC from dialyzers reprocessed 20 times was significantly greater than both new and dialyzers reprocessed once. However, there were no significant differences between new dialyzers and dialyzers reprocessed once. Similarly, after the 1:1000 challenge, TNF alpha production by PBMC harvested from the blood compartment was 160 +/- 0, 160 +/- 0, and 213 +/- 22 pg, respectively, for new dialyzers, dialyzers reprocessed once, and dialyzers reprocessed 20 times (P = 0.008). After the 1:100 challenge, TNF alpha production was 168 +/- 8, 188 +/- 20, and 225 +/- 32 pg, respectively, for new dialyzers, dialyzers reprocessed once, and dialyzers reprocessed 20 times (P = 0.20). These results demonstrate that reprocessing of high-flux polysulfone dialyzers with bleach increases the risk of reverse-transfer of bacterial products from contaminated dialysate, and this risk appears to increase with the number of reuses. Consequently, units that reprocess membranes with bleach and have suboptimal water quality might subject their patients to a higher risk of cytokine-related morbidity.     

Absence of bacteremia and endotoxemia despite contaminated dialyzate

Fevers associated with hemodialysis have been attributed to the transfer of relatively large endotoxin molecules and/or bacteria from contaminated dialyzate across the dialyzing membrane. We evaluated 27 patients during hollow-fiber dialysis when, due to a malfunction, dialysis fluids contained bacteria and endotoxin at levels previously reported to be associated with pyrogenic reactions. Neither endotoxin nor bacteria was detected in 54 venous and arterial blood specimens collected at the termination of hemodialysis. Temperature elevations did not occur during or within 1/2 hr after dialysis. In an extended study, 20 dialyzers were collected after single patient use and the dialyzate compartment was filled with highly contaminated dialyzate, while the blood compartment was filled with sterile pyrogen-free saline. Following 5 to 7 days incubation, bacteria were present in the blood compartments of 4 of 20 dialyzers, probably due to contamination during dialyzer handling. However, the much smaller endotoxin molecule could not be detected in the absence of bacterial contamination. These results indicate that the intact cellophane membrane is an effective barrier to endotoxin and bacteria under clinical conditions.     

Lactogenic hormone signal transduction

Dialyzers are reused in approximately three quarters of the dialysis units in the United States, but the effect of reprocessing on dialyzer performance has not been extensively evaluated. In a crossover study of six chronic hemodialysis patients, we determined urea, creatinine, phosphate, and beta2-microglobulin clearances and dialysate protein loss for two types of low-flux and two types of high-flux dialyzers during use numbers 1, 2, 5, and 15. Dialyzers were reprocessed by an automated machine using Renalin (Renal Systems, Plymouth, MN) as the germicide. Dialyzer arterial and venous blood and dialysate outflow samples were obtained at 5 and 180 minutes of each dialysis session to evaluate solute clearances. Urea, creatinine, and phosphate clearances were calculated using dialysate concentrations, whereas beta2-microglobulin clearance was calculated using plasma concentrations to include its removal by adsorption to the dialysis membrane. There was a trend for urea, creatinine, and phosphate clearances to decrease with reuse for both low-flux and high-flux dialyzers, but these differences were not statistically significant. The clearance of beta2-microglobulin and dialysate total protein concentration was small for low-flux dialyzers; these values were not dependent on reuse. There was a trend for beta2-microglobulin clearance and dialysate total protein concentration to decrease during a dialysis treatment using high-flux dialyzers. More significantly, beta2-microglobulin clearance and dialysate total protein concentration decreased substantially with the reuse of high-flux dialyzers. These observations show that the maintenance of small solute clearances during reuse of high-flux dialyzers does not ensure the maintenance of large solute clearances.     

Effect of truncation and mutation of the carboxyl-terminal region of the beta subunit on membrane assembly and activity of the pyridine nucleotide transhydrogenase of Escherichia coli

Among the several disadvantages of reprocessed dialyzers is the concern that reuse could decrease the clearance of uremic toxins, leading to a decrease in the delivered dose of dialysis. To examine this possibility in the clinical setting, the clearances of small molecular weight solutes (urea and creatinine) and middle molecular weight substances (beta 2 microglobulin) were compared during dialysis with "high-efficiency" cellulose (T220L) and "high-flux" polysulfone (F80B) dialyzers reprocessed with formaldehyde and bleach. In a crossover study, six chronic hemodialysis patients were alternately assigned to undergo 21 dialysis treatments with a single T220L dialyzer or F80B dialyzer. Each patient was studied during first use (0 reuse), 2nd reuse (3rd use), and 5th, 10th, 15th, and 20th reuse of each dialyzer. Urea, creatinine, and beta 2 microglobulin clearances were measured at blood flow rates of 300 ml/min (Qb 300) and 400 ml/min (Qb 400). Total albumin loss into the dialysate was measured during each treatment. Urea or creatinine clearance of new T220L dialyzers was not significantly different from that of new F80B dialyzers at either Qb. Urea clearance of F80B dialyzers at Qb 300 decreased from 241 +/- 2 ml/min for new dialyzers to 221 +/- 5 ml/min after 20 reuses (P < 0.001), and Qb 400 from 280 +/- 4 ml/min for new dialyzers to 253 +/- 7 ml/min after 20 reuses (P = 0.001). Similarly, with reuse, creatinine clearance of F80B dialyzers also decreased at Qb 300 (P = 0.07) and Qb 400 (P = 0.03). In contrast, urea or creatinine clearance of T220L dialyzers did not decrease with reuse at either Qb. Urea clearance of T220L dialyzers was significantly higher than that of F80B at Qb 300 at the 5th, 10th, 15th, and 20th reuse (P < 0.001, = 0.005, = 0.004, and = 0.006, respectively), and Qb 400 at the 2nd, 5th, 10th, 15th, and 20th reuse (P = 0.04, 0.008, 0.03, 0.02, and 0.008, respectively). Beta 2 microglobulin clearance of T220L dialyzers was < 5.0 ml/min across the reuses studied. Beta 2 microglobulin clearance of F80B was < 5.0 ml/min for new dialyzers, but increased to 21.2 +/- 5.3 ml/min (Qb 300) and 23.6 +/- 3.3 ml/min (Qb 400) after 20 reuses (P < 0.001). Throughout the study, albumin was undetectable in the dialysate with T220L dialyzers. With F80B dialyzers, albumin was detected in the dialysate in four instances (total loss during dialysis, 483 mg to 1.467 g). In summary, the results of this study emphasize the greater need for information on dialyzer clearances during clinical dialysis, especially with reprocessed dialyzers. A more accurate knowledge of dialyzer performance in vivo would help to ensure that the dose of dialysis prescribed is indeed delivered to the patients.     

Ab initio electronic structure calculation of a new gene system using the negative factor counting method

It has been reported that cumulative carnitine losses through dialysis membranes may worsen hyperlipidemia during long-term hemodialysis. However, carnitine supplementation has not shown a consistent beneficial response. We undertook the present study to determine if there is any hypolipidemic effect of L-carnitine on Greek dialysis patients in concert with the dialysate buffer composition (acetate or bicarbonate). A total of 28 patients (16 male, 12 female), mean age 43 years (range 21-61), with end-stage renal disease on maintenance hemodialysis for a mean period of 25 months (range 7-84) were studied. The dialysis schedule was 4 h, 3 times/week using cuprophane hollow-fiber dialyzers and acetate (n = 14) or bicarbonate (n = 14) dialysate. In all patients L-carnitine (5 mg/kg body weight) was infused intravenously 3 times/week at the end of each hemodialysis session. Blood samples for carnitine and lipid determinations were obtained before treatment, and 3 and 6 months following treatment. Even though L-carnitine did not modify most of the serum lipid levels, a significant decrease in serum triglycerides was evident in the whole group of patients (from 225 +/- 76 to 201 +/- 75 mg/dl, p = 0.03). Furthermore, L-carnitine could decrease serum triglycerides only in hypertriglyceridemic patients (from 260 +/- 64 to 226 +/- 82 mg/dl, p < 0.05). L-Carnitine resulted in a reduction of serum triglycerides in both patients on bicarbonate and on acetate dialysis, while there were no significant differences in the changes of lipid parameters after L-carnitine between the two groups of hemodialysis patients. We conclude that relatively low doses of L-carnitine supplementation could contribute to the management of some hypertriglyceridemic hemodialysis patients.     

In vivo verification of an automatic noninvasive system for real time Kt evaluation

It is generally accepted that morbidity and mortality of hemodialysis patients is related to dialysis quantitation. Currently available methods for the quantitation of dialysis require blood sampling or a continuous measurement of changes in urea concentration during treatment. These maneuvers are time consuming and expensive, and are generally performed, at most, once per month. The authors introduce an on-line, automated method for measurement of dialyzer electrolyte clearance comparable to urea clearance by using dialysate conductivity sensors placed pre and post dialyzer, and measuring conductivity at three different pre dialyzer levels. Conditions that reduce clearance, such as recirculation or fiber clotting, are automatically taken into account so that the method measures effective clearance rather than dialyzer clearance. In vitro and in vivo studies validate the method. Results are immediately available and can be used to address problems such as improper needle placement and access recirculation. In addition, repetitive electrolyte clearance data can serve to enhance quality assurance programs with respect to verifying the function of reused or new dialyzers. Appropriate algorithms can be used to calculate delivered Kt/V.     

Osteoid osteoma and osteoblastoma with clonal chromosome changes

The i-STAT hand-held analyzer assays ten tests including electrolytes, gases, urea, glucose, ionized calcium, and hematocrit. Eight different cartridges assay one to eight tests. We have previously confirmed or demonstrated that accuracy and precision for blood assays are comparable to accepted laboratory methods. We now report similar results for hemodialysis dialysate and peritoneal dialysis effluent. The i-STAT analyzer is simple to use, and dialysis nurses produced accurate results with 20 min training. The results are viewed digitally on the analyzer and automatically on a small attachable printer. i-STAT blood analysis is most valuable when results are desired immediately, anywhere, including before, during and after dialysis in hemodialysis units. Hemodialysate analysis using i-STAT can be most valuable for rapidly checking dialysis machine function such as dialysate mixing and conductivity and ramping results and dialysate concentrations prepared in the unit. Peritoneal effluent analysis is useful for rapid evaluation of membrane function.     

The synaptic process in Locusta migratoria spermatocytes by synaptonemal complex analysis

BACKGROUND: Bacterial contamination of treated water and dialysate comprises an important problem for patients undergoing haemodialysis. Both the progressive reduction of the thickness of cellulose membranes and the expanding use of high-flux membranes probably enhance the risk of pyrogenic reactions, therefore increasing the need for atoxic water and non-pyrogenic dialysis fluid. METHODS: Samples of tap water, treated water, and effluent dialysate in all 85 haemodialysis centres in Greece were examined for total heterotrophic bacteria counts employing the pour plate method, total and faecal coliforms, faecal streptococci and pseudomonas spp. using the membrane filter technique, and sulphite-reducing clostridia applying the most probable number method. Overall 255 paired samples were tested from January to March 1997. RESULTS: For total heterotrophic bacteria, the overall compliance of treated water and dialysate to the American Association of Medical Instrumentation standards (<200 c.f.u./ml for water and <2000 c.f.u./ml for dialysate) was 92.6 and 63.7% respectively, whereas the compliance of tap water samples to our national standards (total heterotrophic bacteria < 10 c.f.u./ml and absence of the other indicator bacteria) was 80.7%. The most commonly isolated bacteria were pseudomonas spp., found in 22.2% of treated water and 59.5% of dialysate samples, whereas the respective frequencies were 12.3 and 36.2% for total coliforms, 8.6 and 30.0% for faecal coliforms, 14.8 and 28.7% for faecal streptococci, and sulphite-reducing clostridia were isolated in 5.8% of dialysate samples only. Haemodialysis centres equipped with storage tanks for treated water experienced lower levels of total heterotrophic bacteria, but higher counts of total and faecal coliforms, faecal streptococci, and pseudomonas spp., although the difference was statistically significant only for faecal streptococci counts, (P<0.05). Sixty-seven haemodialysis centres were equipped with bacterial filters, but mean values of all the examined microorganisms were not statistically different from those of the other centres. Faecal streptococci counts in treated water samples were positively correlated with ageing of both haemodialysis centres (P<0.005) and purification system (P<0.05), whereas pseudomonas counts were significantly correlated with ageing of the purification system (P<0.05).     

Augmenting solute clearance in peritoneal dialysis

BACKGROUND: The removal of low molecular weight solutes by peritoneal dialysis is less than by hemodialysis. The targets for Kt/Vurea and creatinine clearance formulated in the Dialysis Outcome Quality Initiative are unlikely to be achieved in a substantial portion of peritoneal dialysis patients. Possibilities to increase small solute clearances have therefore been subject to many investigations. METHODS: A review of the literature and of recent new data on determinants of solute removal, such as residual renal function, the role of drained dialysate volume and manipulation of the diffusive capacity of the peritoneum are presented. RESULTS: The contribution of residual GFR is more important for the clearance of creatinine than for Kt/Vurea. It is even more important for the removal of organic acids that are removed from the body by tubular secretion. High dosages of furosemide increase the urinary volume and the fractional Na+ excretion, but have no effect on the magnitude of residual GFR, renal creatinine clearance, renal urea clearance, and peritoneal transport characteristics. The drained dialysate volume per day is the main determinant of the peritoneal removal of urea. Its effect decreases the higher the molecular weight of a solute. It can be augmented by using large instillation volumes, by the application of more exchanges, and by increasing peritoneal ultrafiltration. A large exchange volume is especially effective in patients with an average transport state, but in those with high solute transport rates, Kt/Vurea is especially influenced by the number of exchanges. Possibilities to increase ultrafiltration are discussed. The diffusive capacity of the peritoneum can be augmented by using low dosages of intraperitoneally administered nitroprusside. This increases solute transport most markedly when it is applied in combination with icodextrin as osmotic agent. CONCLUSIONS: Small solutes clearances cannot be increased by furosemide. Increasing the instilled volume of dialysis fluid and the number of exchanges both affect solute clearance. Studies are necessary on long-term effects of manipulation of the peritoneal membrane with nitroprusside.     

Hemodialysis versus cross hemodialysis in experimental hepatic coma

In this study, the effects of conventional hemodialysis on experimental hepatic coma have been compared with those of hemodialysis against blood from a normal donor which allows exchanges without mixing of blood. cuprophan and polyacrylonitrile membranes were compared. Cross hemodialysis with a donor resulted in rpompt but transient recovery of consciousness, whichever membrane was used. Cuprophan hemodialysis without donor had no effect. Polacrylonitrile hemodialysis without donor allowed progressive and prolonged improvement in the consciusness level and the electroencephalograms. Therefore, clearance of middle molecular weight substances was more effective than exchange with a donor. Preliminary results in man showed total recovery of consciousness in six of ten patients with acute liver failure and coma and partial recovery from complete grade IV coma to grade II encephalopathy in two patients. These two patients reacted when called by their name by opening their eyes and obeying simple orders.     

Management of tumor lysis syndrome with standard continuous arteriovenous hemodialysis: case report and a review of the literature

Tumor lysis syndrome (TLS) is a critical illness with few treatment options. This report describes the clinical course of a patient with non-Hodgkin's lymphoma, who developed TLS and required renal replacement therapy. Institution of the standard therapeutic approach, intermittent hemodialysis, was not possible because of persistent hypotension. Instead, the patient was treated with a short course of continuous arteriovenous hemofiltration (CAVH) and conventional continuous arteriovenous hemodialysis (CAVHD) with dialysate flow rate of 1 L/h), which resulted in effective control of serum uric acid, potassium, urea nitrogen, phosphorus, and extracellular fluid volume. This case is in distinction to a previous report of TLS treatment with CAVHD using 4 L/h dialysate flow rate. We conclude that continuous renal replacement therapies with standard dialysate flow rates and replacement volumes should be considered for the treatment of TLS, particulary if the syndrome is accompanied by hypotension.     

Development and maintenance of bile canaliculi in vitro and in vivo

Four types of high-flux hemodialyzers, Primus 2000 (high-flux polysulfone 2.0 m2), Altra-Flux 170 G (cellulose diacetate 1.7 m2), FLX-15 GW (polyester-polymer alloy 1.5 m2) and PAN-85 DX (polyacrylonitrile 1.7 m2) were evaluated in vivo. A total of 12 stable chronic hemodialysis patients participated in the study and each type of dialyzer was tested once in 9 of them. Blood samples for the measurement of BUN, creatinine, phosphate, uric acid, albumin and beta2-microglobulin (beta2M) were drawn before and 5 min after the end of the study dialysis. During dialysis, which was performed in all patients with a blood flow rate of 250 ml/min for 240 min, the dialysate (550-600 ml/min) was collected every hour and samples were drawn for the measurements of all the above substances. The mean total amount of low-molecular substances removed per session by each dialyzer was very close to 19.5 g for urea, 2.0 g for creatinine, 0.9 g for phosphate and 1 g for uric acid. The one-third (30-33%) of the above amounts were removed during the first hour of dialysis. Dialyzers' clearances for creatinine and uric acid were significantly higher in Primus dialyzer comparing to FLX-15 GW (p < 0.05) while the clearance for urea showed a borderline significance (p = 0.055). No difference was found either among Altra-Flux 170 G, FLX-15 GW and PAN-85 DX or between Primus and PAN-85 DX dialyzers. Phosphate clearance did not show any difference among the four dialyzers. The lowest amount of albumin removed per session was 0.75 g by PAN-85 DX and the highest 1.8 g by FLX-15 GW, while the equivalents for beta2M were 80 mg by Altra-Flux 170 G and 142 mg by PAN-85 DX. A significant adsorption of beta2M on these dialysis membranes was indicated by the combination of a satisfactory serum beta2M reduction ratio (post-/predialysis values = 0.52, 0.77, 0.60, 0.55) with a reduced beta2M clearance (23.9, 13.6, 20.2, 25.1 ml/min). During the first hour of dialysis, in comparison to the following time, the highest amounts of albumin and beta2M (expressed as percentage of total) were removed by the Primus 2000 dialyzer. Our results indicate that under conventional conditions small differences in the surface area of the high-flux dialyzers are unimportant regarding the removal of low molecules. However, the composition of the membrane seems to play an important role in the removal of high-molecular substances.     

Automated peritoneal dialysis with 'on-line'-prepared bicarbonate-buffered dialysate: technique and first clinical experiences

BACKGROUND: Automated peritoneal dialysis (APD) has the possibility of increasing the dialysis efficacy by using higher fill volumes, frequent dialysate exchanges, and tidal techniques. It is then possible to treat patients adequately without residual renal function. The drawbacks of the required high amounts of dialysis solution of up to 30 litres per session are the high costs of lactate-based dialysate bags and difficulties for the patients in handling these bags. So far, bicarbonate-based peritoneal dialysate, which may be more biocompatible, is only available for CAPD in double-chamber bags. In APD this could be overcome by 'on-line' preparation of bicarbonate-buffered dialysate using advanced technologies originally designed for on-line preparation of substitution fluid for haemofiltration. METHODS: Four patients without residual renal function were treated with APD five times weekly in a crossover study design. Patients received standard lactate-based (35 mmol/l) treatment (25 litres per session each) in weeks 1 and 3. In week 2 on-line-produced bicarbonate-buffered (37 mmol/l) dialysate was used. This dialysate was prepared by an AK 100 Ultra haemodialysis machine. The machine was modified for adding glucose from a 50% concentrate to the desired concentration of 1.7%. Electrolytes, pH, pCO2, and dialysis efficacy parameters were measured. Microbiological testing was carefully performed. RESULTS: Creatinine clearances, Kt/V, and pCO2 did not vary between the different treatment phases, whereas the pH showed a distinct increase during the bicarbonate phase. Repeated determinations of endotoxins and culturing showed no contamination of the dialysate. The composition of the produced dialysate was reproducible with respect to pH, pCO2, sodium, calcium and bicarbonate, whereas the glucose concentration varied by +/- 20%. CONCLUSIONS: On-line preparation of PD fluid with the AK 100 Ultra is easy and safe to handle. APD with dialysate containing 37 mmol/l bicarbonate provides improved acid base balance and possibly improved biocompatibility, and may lead to a significant cost reduction. Further development in order to provide smaller machines and more precise ways of achieving a desired dialysate glucose concentration is necessary.     

A preliminary report on a method for studying the permeability of expanded polytetrafluoroethylene membrane to bacteria in vitro: a scanning electron microscopic and histological study

The technique of guided tissue regeneration using expanded polytetrafluoroethylene (ePTFE) membranes has been shown to be effective in implant dentistry (bony defects, extremely thin alveolar ridges, and implants placed in fresh extraction sockets). One of the drawbacks associated with the use of membranes is their premature exposure with consequent bacterial contamination. The aim of this study was to examine the possibility that oral bacteria migrate through the occlusive portion of ePTFE membranes and to determine the time needed for microorganisms to pass from the outer surface to the inner surface of the membranes. A removable acrylic device was adapted to the molar-premolar region of one quadrant of the jaws in each of three volunteers. Five cylindrical teflon chambers were glued to the buccal aspect of each device. The chambers were divided into two rooms separated by the inner portion of a ePTFE membrane. The outer room was open to the oral cavity allowing plaque accumulation; the inner room was isolated from the oral cavity by the ePTFE membrane. One of the 5 chambers was completely closed and used as control. The test period lasted for 4 weeks. Every week, one chamber was removed from each device and processed for scanning electron microscopic and histologic examinations. Our study showed the possibility that oral bacteria may contaminate ePTFE membranes exposed to the oral cavity. One specimen showed partial bacterial penetration after 2 and 3 weeks, but after 4 weeks, all membrane specimens demonstrated bacterial contamination.(ABSTRACT TRUNCATED AT 250 WORDS)