Ultrafiltration failure in CAPD.A Shetty, DG Oreopoulos
Toronto Hospital (Western Division), Ontario, Canada., Canada
Correspondence Address: Source of Support: None, Conflict of Interest: None PMID: 0008699390
Source of Support: None, Conflict of Interest: None
Keywords: Biological Transport, Hemodiafiltration, adverse effects,Human, Peritoneal Dialysis, adverse effects,methods,Peritoneum, physiopathology,Treatment Failure,
The threat of peritoneal membrane failure has cast a shadow over the future of CAPD for many years. Various authors have defined ultrafiltration(UF) failure in different, ways: essentially it represents "expectations unfulfilled` by the peritoneum i.e. the presence of features of fluid overload in the absence of hypoalbuminemia despite the use of appropriate hypertonic dialysate and fluid restriction. Slingeneyer et al gave a clinical definition using the following criteria: 1) inability to reach the 'dry weight' and to maintain a normal blood pressure despite increasing use of a dialysate with high dextrose concentrations; 2) pitting edema in the presence of normal serum proteins and albumin (more than 60.0 and 30.0 g/L respectively); 3) persisting symptoms despite severe reduction of fluid and sodium intake. 4) need for hospital admissions for extracellular volume removal with WD (in CAPD or IPD patients) and/or with hemofiltration. Henderson et al defined loss of ultrafiltration in absolute terms as a significant diminution in effluent volume from a 4 - hour hypertonic cycle in successive measurements over several months. Mactier et al defined ultrafiltration failure in CAPD as clinical evidence of fluid overload that persists despite restriction of fluid intake and the use of three or more hypertonic (4.25% dextrose) exchanges per day. Heimburger et al defined permanent clinical loss of UF as edema and inability to reach “dry weight” despite three or more hypertonic (4.25% dextrose) exchanges per day and restricted fluid intake: symptoms of overhydration exceeding one month's duration, and finally, no established cause of UF loss such as peritonitis, dialysate leakage or catheter malfunction.
True UF failure should be differentiated from apparent UF failure. Apparent UF failure may be due to the following. 1) increased fluid and salt intake. 2) failure of the patient to perform exchanges regularly. 3) mechanical catheter problems leading to increased residual volume, which, rapidly dilutes the incoming dialysate, 4) loss of residual renal function without concurrent reduction in the fluid Intake and change in the glucose concentration of the dialysate bags, 5) hyperglycomia or hyperalbuminemia resulting in reduction of the osmotic gradient, 6) loculation of fluid within the peritoneal cavity due to adhesions that reduce the solution - membrane contact. 7) reduced capillary hydraulic pressure In shock states and 8) internal dialysate leaks such as peritoneopleural leak. Acute peritonitis can induce acute UF failure through increased capillary - venular permeability, which promotes increased glucose absorption. This improves when the peritonitis clears.
The wide variability in the reported Incidence of UF failure ranges from 1.65 to 100% ,,; In most of these studies UF failure was related to use of acetate buffered dialysate. Two early reports from France showed a high Incidence of UF failure. In 1983 Slingeneyer et al reported a 25.8% Incidence of UF failure in 58 patients on CAPD followed for over 30 months. In this study the actuarial risk of developing UF failure on CAPD was 10% at one year and 65% at 3 years. In 1984, Faller of al reported 100% incidence of UF failure In 31 patients using acetate buffered dialysate. On the contrary, UF reduced only in two (16%) of 16 patients using lactate dialysate.
Experience in other countries has been different. In Toronto 16 of 508 patients were withdrawn from CAPD because of "membrane" failure. In this study older patients developed UF failure earlier than younger patients and those who developed UF failure late had a higher incidence of peritonitis. In the French study, however, patients with UF failure were younger than those who did not have UF failure. This difference was attributed to longer survival of younger patients on CAPD. Peritonitis did not appear to bear any relationship to loss of ultrafiltration.
Verger et al also from France reported a high incidence of UF failure. Nolph et al from the USA CAPD registry reported that only 1.6% of 6838 patients followed for 3119 patient years required transfer to haemodialysis because of loss of UF and/or inadequate control of uremia.
In 1990, Heimburger et al from Sweden reported that 6.2% of their 227 patients developed permanent loss of UF. The cumulative risk of such-loss increased from 2.6% after one year to 30.9% after six years of treatment. Other recent studies of large patient populations indicate that 1.3% to 6.0% of CAPD patients will have to be withdrawn from the treatment because of UF loss,,,.
Ultrafiltration depends upon transcapillary ultrafiltration and lymphatic absorption. Transcapillary UF is governed by Starling's forces. As a corollary to Starling's law, transcapillary UF rate (UFR) is directly proportional to effective peritoneal blood flow (EPBF), hydraulic permeability of the peritoneum (HP), the effective surface area of the peritoneum (ESAP) and the transcapillary pressure gradient (TCPG).
i.e. UFR ? EPBF x HP x ESAP x TCPG (Equation 1)
[Figure - 1] depicts the various pressures acting across the transcapillary membrane. From this we can derive the following equation:
TCPG = HSP-C + COP-D + CrOP-D - HSP-D- COPP-CrOP-P (Equation 2)
where TCPG = transcapillary pressure gradient
HSP-C = capillary hydrostatic pressure
COP-D = dialysate colloid osmotic pressure.,
CrOP-D = dialysate crystalloid osmotic pressure
HSP-D = dialysate hydrostatic pressure
COP-P = plasma colloid osmotic pressure
CrOP-P = plasma crystalloid osmotic pressure
The effective peritoneal blood flow is an important determinant of UF rate. Theoretically, the blood flow in the peritoneal capillary network is between 50-70 ml/min14. However, not all of this blood is available for water and solute exchange. The effective blood flow available for dialysis is 22.4 ml/min14. If this increases, there may be an increase in early solute exchange and hence an early loss of osmotic gradient during an exchange. If it decreases as in shock states, there will be less blood available for dialysis, which again would reduce ultrafiltration.
Various disease conditions may alter the surface area of the peritoneum and both an increase or decrease in the peritoneal surface area can result in UF loss. Even though increased surface area should lead to increased UF rate, it actually decreases net LIF in CAPD because of the rapid exchange of solutes along with that of water. By this same mechanism, both an increase as well as decrease in the permeability can reduce UF, even though (according to equation (1) an increase in the permeability should increase the UF rate. In patients on CAPD, the hydraulic permeability of the peritoneum is 0.40 ml/hr/mm Hg/sq m15. The critical component of the transcapillary pressure gradient is the osmotic pressure difference, which is generated by the glucose concentration in the dialysate. The plasma osmolarity of a uremic, individual averages about 310 mosm/L and the osmolality of the dialysate is 350 for 1.5% solution and 490 for 4.25% solution. According to vant Hoff's law, 1.0 mosmol/kg H2O exerts a pressure of 19.3 mm Hg when the reflection coefficient is 1. Reflection coefficient of 1 means that the membrane is semi-permeable and the no solute is being transported. The reflection coefficient of glucose in CAPD patients has been indirectly calculated to be 0.05. The hydrostatic pressure of blood in the peritoneal capillaries is 17 mm Hg and that of the dialysate in the peritoneal cavity ranges from 3 to 20 mm Hg, depending on whether the patient is recumbent or upright. The colloid osmotic pressure of blood is 26 mm Hg and that of the dialysate is 0.1 mm Hg. The actual transcapillary UF will be slightly less than expected from these figures because the gradient decreases towards the venous end of the capillary due to gradual increase in the colloid osmotic pressure and decrease in capillary hydrostatic pressure as we go from the arterial to the venous end of the capillary. In contrast to this, we expect the solute clearance to be greater towards the venous end of the capillary because of the increase in the endothelial cell separation (up to 40 A compared to 10 A at the proximal end of the capillary) according to the hyperporosity model proposed by Nolph.
Lymphatic absorption can be- calculated from the disappearance rate of intraperitoneaily (IP) administered human albumin, autologous haemoglobin, RISA or neutral dextran and the appearance rate of 1hese substances in blood. Probably IP pressure is the main determinant of lymphatic absorption during CAPD. An increase in W pressure from 10-20 mm Hg by the application of external pressure caused a 70% increase in the lymphatic absorption rate. Other factors that affect lymphatic absorption are respiratory movements and the position of the patient - these operate through changes in intra-lymphatic pressure and intra-peritoneal hydrostatic pressure. Lymphatic absorption is higher in the supine position. In contrast to the transcapillary UF, which occurs mainly during the first hour of each exchange, lymphatic absorption is constant throughout the exchange and hence would have major impact on net UF in long dwell exchanges such as the night exchanges. The cumulative lymphatic re-absorption may vary between 200 and 450 ml over 6 hours during a 2L exchange.
Blood viscosity, certain hormones and drugs, interstitial water pressure, serum protein concentration, protein concentration in the interstitial fluid and in peritoneal fluid alli affect the UF rate by affecting one or more of the variables in equation 1.
Wu et al described two mechanisms, of UF failure. Later Verger et al named them Type I and Type II UF failure.
1) Type I - UF failure exists where there is UF failure with normal or even high solute transport. This type of failure is caused by
i) long-term CAPD - seen after several years on this modality.
ii) use of an acetate dialysate
iii) during or after acute peritonitis
2) Type II UF failure exists where there is UF failure along with impaired exchange of solutes such as glucose, creatinine, etc. These patients may have features of underdialysis. This type of UF failure is caused by
i) sequela of severe peritonitis especially that due to Staph aureus
ii) sequela of recurrent peritonitis
iii) smoldering or inadequately treated peritonitis
iv) following an abdominal operation/operations
v) peritoneal sclerosis
Type I UF loss following long-term CAPD due solely to increased lymphatic absorption sometimes has beer called. Type III UF loss.
From Equations 1 and 2 and [Figure:I], one can derive the factors associated with UF failure. Interestingly both increase as well as decrease in factors like peritoneal hydraulic permeability, surface area and blood flow cause UF failure.
Peritoneal Blood Flow: The theoretical blood flow in the peritoneal capillary network has been calculated to range between 50 and 70 ml/min14; however not all this blood is available for exchange of water and solutes. It has been calculated that effective peritonea, blood flow available for dialysis is only 22.4 ml/min14. According to the equation (1) any decrease in peritoneal blood flow would decrease UF - a fact that may explain some of the UF loss in the presence of peritoneal vascular damage, diabetes or sclerosing peritonitis.
Hydraulic - permeability, effective peritoneal surface area and lymphatic absorption rate: Most UF failure is associated with a large effective peritoneal surface area or rapid lymphatic absorption (Type I UF failure). The transport of low) molecular weight solutes depend mainly on the effective peritoneal surface area. Krediat et al found that when compared to 9 patients with good UF, 12 CAPD patients with poor net UF had higher mass transfer area coefficients (MTAC) of creatinine and glucose in 5 CAPD patients. Wideroe et al reported a decreased net UF, which was associated with increased transperitoneal transport of low molecular weight solutes. Heimburger et al reported high MTAC in seven and a high effective lymphatic absorption rate in 2 among 9 patients with permanent, net UF failure. Also Mactier et al, reported a high lymphatic absorption causing UF failure.
After observing a high incidence of UF failure in France and low incidence in North America six nephrologists from Canada, France and the USA conducted a study of this difference. This study, which involved 210 patients from 20 participating canters found that net UF and dialysate glucose concentration was much lower in the group using acetate solution While North Americans used lactate solutions almost exclusively, solutions containing acetate were in common use in France. There was no significant correlation of UF failure with age. Only the group using both acetate and lactate but not the groups using only lactate or only acetate showed a significant positive correlation between the duration of CAPD and UF loss. Peritonitis episodes did not correlate with UF loss. In the USA, acute studies during dialysis showed no significant differences in transport features when they compared acetate and lactate- containing solutions Thus it seemed likely that peritoneal transport defects are due to exposure for weeks or months or to factors unique to the French solutions. Even among the patients from France, those using lactate only had significantly better UF than those who used acetate only or acetate and lactate.
During peritonitis the net UF is lower because a high, effective lymphatic absorption rate and a large effective surface area cause a rapid dissipation of the osmotic gradient, In peritonitis, even though the surface area of the peritoneal membrane may not change, the peritoneal area participating in the exchange (i.e. the effective peritoneal surface area) may increase due to capillary vasodilatation. Krediet et al, found that, during dialysis, the maximal intra-peritoneal volume was reached at an earlier stage during a dwell due to increased transcapillary UF during the first hour; however, the volume after 4 hours was lower due to a higher effective lymphatic absorption rate. In the presence of pleuroperitoneal communication, an increase in the volume of distribution of the dialysate and a large residual volume leads to instantaneous dilution of the administered glucose. Both mechanisms produce a rapid decrease in the osmotic gradient.
Long term CAPD has been associated with type I UF failure. Heimburger et al found 7 of 9 patients with permanent UF failure had increased solute transport, two patients had high lymphatic absorption rate with normal diffusive mass transport, coefficients. Krediet et al found that when only 3.86% dialysate was used, the mean maximal UF capacity decreased from 2792 ml/24 hours during the first 6 months of CAPD to 1861 ml/24 hours alter more than 2 years. This decrease was associated with an increased absorption of glucose. In another study 27 20 patients on CAPD for more than 4 years had a lower UF (mean 487 ml after 4 hours dwell of 3.86% dialysate) than did a group just starting CAPD (mean 826 ml). Those on long-term CAPD had higher mass transfer area coefficients (MTAC) for glucose and creatinine, Also MTAC is higher and net UF is lower during 0- 3 months after the start of CAPD than after 6- 7 months: this implies that true baseline values are probably not reached before 6 months on CAPD. After 6 months on CAPD, MTAC gradually increases and the net UF declines. This evidence points to a time dependent increase in effective peritoneal surface area.
A possible cause of UF failure in patients on long term CAPD is the PD solution. The bioincompatible elements in the conventional PD solutions are low pH, high osmotality, the osmotic agent (glucose) and the buffer (lactate), Jorres et al have pointed out convincingly that dialysis fluid is cytotoxic to mesothelial cells because of its low pH, which impairs leucocyte and peritoneal macrophage viability and reduces the release of cytoprotective leucotrienes (such as TNF and IL5) from both peripheral and peritoneal polymorphonuclear leucocytes. Viassara described mechanism of glucose-induced peritoneal damage, during which tissue and cell-surface proteins undergo non-enzymatic glycosylation, which releases IL-1 and TNF. This may be the process underlying fibrosis. Carozzi et al demonstrated that an increase in peritoneal lymphocyte and macrophage calcium induce the release of large amounts of gamma interferon and IL-1, which stimulate peritoneal fibroblast proliferation. When Veech studied the role of lactate buffer, he found that increased lactate levels produce a strong increase in intracellular Ca++ content, which may lead to increased fibroblast proliferation mediated by increased release of gamma interferon and IL-1.
In sclerosing peritonitis, Verger et al found that one patient had a normal glucose absorption preceded by a phase of increased glucose absorption associated with loss of net UF. Krediet et al found an increase in trans-peritoneal solute transport in 3 patients, and a decrease in the solute transport in one patient with sclerosing peritonitis. This data suggest that sclerosing peritonitis can produce either a decrease or increase in the effective peritoneal surface. An increase in the effective peritoneal surface area in spite of sclerosis may be due to the formation of new capillaries in poorly vascularized parts of the peritoneum.
Interleukin-1 hypothesis: Shaldon et al offered this hypothesis to explain the final, common pathway mediating the loss of ultrafiltration. IL-1 is released from macrophages secondary to a variety of stimuli. Effluent from CAPD patients contained significant quantities of IL-1. IL-1 has several local peritoneal actions including stimulation of fibroblasts, which in tum increases collagen production and stimulates the endothelial cells to produce prostacyclin. Prostacyclin, a potent vasodilator, may increase glucose absorption and diminish ultrafiltration, which can account for both UF loss and sclerosing peritonitis. Interestingly acetate is a potent stimulator of IL-1 release. Mast cell mediated changes: Dobbie et al first proposed that mast cell played a role in UF loss. Most peritoneal irritants reduce the number of submesothelial mast cells for 3 - 6 days after exposure, after which these calls gather in large numbers at the site of injury and release more than 75% of their granules. These granules contain various polypeptides and nucleotides that induce hyperemia and increase particularly venular permeability and trigger collagen production. Local tissue damage releases thromboplastin, converting fibrinogen to fibrin. Fibrin organization can initiate sclerosis. Free oxygen radicals also have been implicated in the mesothelial coil damage.
Since the 1920's it has been known that the intraperitoneal use of hyperlonic glucose solutions may be associated with marked mesothelial changes. Animal experiments showed such changes as mesothelial swelling, separation, denudation, fibrosis, loss of mesothelium and loss of endothelial intercellular spaces in response to such noxious stimuli as silica, carbon, drying and indomethacin. Verger et al and Dobbie et al demonstrated mesothelial intracellular edema, disruption of organelles, interstitial edema and submesothelial deposition of collagen fibres in response to introduction of dialysate into the peritoneal cavity. On light and electron microscopy, Verger et al found large gaps between the mesothelial calls, loss of mesothelial microvilli, mesothelial desquamation and moderate hypervascularization in those with Type I UF failure and numerous adhesions and sclerosing peritonitis in those with Type II UF loss.
UF failure should be suspected when we find weight gain, edema and reduced drained volume. Before diagnosing true UF loss, we should exclude apparent UF loss by checking 1) the fluid intake (patients usually underestimate their fluid intake), 2) the residual peritoneal volume, 3) onset of UF loss: acute onset usually indicates a mechanical problem with apparent UF loss, 4) The chest cavity for any evidence of unilateral pleural effusion especially on the right side which is typical of pleuroperitoneal shunt 5) The scrotum for bilateral hydrocoele: this may indicate a communication between the peritoneum and the tunica vaginalis 6) the abdominal wail for evidence of interstitial leak of dialysate, 7) the drainage time -an increase suggests catheter dysfunction or malposition or mechanical problems (hernias or fluid sequestration), 8) gradual loss of any residual renal function and finally 9) the ultrafiltration volume with daily measurement for at least one week.
Once a real loss of UF has been demonstrated, we should do a peritoneal equilibration test (PET) for accurate diagnosis. The discriminating parameter is the ratio between the glucose concentration in the peritoneal dialysis effluent at the end of 4 hours and that in the dialysate at 0 hour (Df/Di).
Df/Di 0.3 - 0.5 indicates normal peritoneal membrane function; catheter malfunction, catheter malposition, mechanical problems such as hernias or fluid sequestration. If one can exclude all causes of apparent UF loss, suspect poor patient compliance with dialysis. Df/Di < 0.3 indicates a hyperpermeable peritoneal membrane as seen in Type I UF loss. Df / Di > 0.5 indicates hypopermeable peritoneal membrane as seen in Type II UF loss.
Those with high Df/Di for glucose also have low dialysate / plasma (D/P) creatinine values and those with low Df/Di glucose levels would have high D/P creatinine values.
The ultrafiltrate produced by patients with normal ultrafiltration is hyponatric and hypo-osmotic. Twardowski  has found a significant inverse correlation between the drainage volume and DP (dialysate/ plasma) ratio for sodium. A decrease in the dialysate sodium concentration during the dwell also differentiates true from apparent UF loss; in apparent UF loss (where water removal across the peritoneal membrane is normal) there will be the normal decrease in sodium concentration in the dialysate. In true UF loss of both types, water removal is impaired and hence there is no decrease sodium concentration in the dialysate.
CT scan of the peritoneal cavity with and without radiocontrast in the dialysate helps to recognize such conditions as sclerosing peritonitis, internal dialysate leak into the pleural space, abdominal wall or into the tunica vaginalis of the testes.
The principles of treating patients with UF loss can be understood by studying [Figure - 3], which depicts the ultrafiltration and lymphatic absorption in patients with normal UF, Type I UF loss and Type II UF loss.
Type I UF loss: Shortening the dwell time and terminating the exchange before the lymphatic absorption exceeds the transcapillary UF will attain some net UF. Maximum net UF will be attained if the dialysate is drained (at point A in [Figure - 3]) when the transcapillary UF rate will be maximum. This can be achieved by providing nightly intermittent peritoneal dialysis (NIPD), daily ambulated peritoneal dialysis (DAPD) automated peritoneal dialysis.
Resting the peritoneum for few weeks may help in regaining its UF capacity, hence another option is transferring the patient to haemodialysis for 4 - 6 weeks. Other approach in treating Type I UF failure, short of transferring the patient to haemodialysis, is to use high molecular- weight osmotic agents instead of glucose, e.g. glucose polymers, albumin etc. which are not absorbed as fast as glucose. However, the benefits promised by glucose polymer (icodextrin) during long exchanges may be limited by the accumulation of its poorly metabolized disaccharides in the blood. Albumin would be a good osmotic agent for long dwell exchanges because it is absorbed relatively slowly; it generates greater osmostic pressure than would be predicted from its osmolality because of negative charges at physiologic pH; and it may decrease or avoid negative nitrogen balance. At present, its use is prohibited by cost and limited availability but recombinant DNA technology may make its manufacture possible in the future.
The last option is to abandon CAPD and keep the patient on haemodialysis permanently.
Type I UF loss due only to increased lymphatic absorption rate cannot be treated with short-dwell exchanges; hence patients with this complication will have to be transferred to permanent haemodialysis.
Type II UF loss: In these cases shortening the exchanges would do little good compared to the benefits in Type I patients. They may improve with high dose PD in short exchanges.
Patients with severe Type II UF loss, should be transferred permanently to haemodialysis. Contrary to Type I failure, NIPD does not help much because even solute transport is defective and even at the beginning of the exchange, these patients have poor UF.
Pharmacological manipulation has been tried but no drugs have been successful. Such manipulation has focused mainly on drugs that can increase solute transport such as phosphatidylcholine, chlorpromazine, verapamil and amphotericin B. Phosphatidylcholine may act either by repelling stagnating water due to its surface-active properties, which increase the surfactant - like activity of the peritoneal membrane or it also may reduce lymphatic re-absorption. In CAPD patients with or without UF failure, IP administration of phosphatidylcholine has been found to increase net UF,,,. Reports of the effects of oral phosphatidylcholine are conflicting and it has been suggested that intraperitoneally administered phosphatidylcholine may induce peritoneal adhesions. Chlorpromazine also acts as a surface-active agent. Calcium-channel blockers appear to inhibit IL-1 and gamma- IFN release by blocking the passive endocellular Ca++ influx; hence verapamil may decrease peritoneal fibrosis and thus partially restore UF in patients with Type 1 UF failure. Nifedipine, surprisingly, has been found to decrease net ultrafiltration. Amphotericin B increases the UF rate by increasing the sodium transport, thus reducing the negative effects of a hyponatric ultrafiltrate. However, this effect is counterbalanced by chemical peritonitis caused by its solvent desoxycholate. Furosemide administered in traperitoneally also increases the concentration of sodium in the ultrafiltrate. Dopamine increases the peritoneal UF by increasing the hydrostatic pressure in the peritoneal capillary. Theoretically, vasodilators should increase UF by increasing the peritoneal blood flow. However, the administration of nitroprusside induced vasodilatation without increasing ultrafiltration. These results were observed in long exchanges and hence the impact of these agents on rapid exchanges remains to be clarified. Substances such as neostigmine, which reduce peritoneal lymphatic flow, may increase the volume of effluent.
Possibly we could prevent UF loss by preventing peritonitis and by using a more physiologic peritoneal dialysis solution. More physiologic buffers such as bicarbonate are under investigation. Another step that might reduce the rate of UF loss is the development of a better osmotic agent such as albumin. The introduction of collapsible plastic bags and the Y-set have been major breakthroughs in reducing the peritonitis rate. Educating the patient about measures to avoid contamination should not be neglected.
We thank Cristy and Lori Espino for their help in the preparation of this manuscript and Dr. JO Godden for his editorial review of the final manuscript.
[Figure - 1], [Figure - 2], [Figure - 3]