CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Kidney International, Vol. 39 (1991), PP. 909—919

Beta2-microglobulin kinetics in end-stage renal failure

Ross A. ODELL, PETER SLOWIACZEK, JOHN B. MORAN, and KLAUS SCHINDHELM

Centre for Biomedical Engineering, University of New South Wales, Kensington; and St. Vincent's Hospital, Fitzroy, Australia

Beta2.microglobulin kinetics in end-stage renal failure. The kinetics of [15—17]. The basis for these speculations is the observed rise in /32-microglobulin (f32m) were studied in five anephric or anuric hemodi- serum 2m levels, to an extent greater than can be explained by alysis patients. Human l32m was isolated from peritoneal dialysate using ion-exchange and gel chromatography and radiolabeled with 1251.hemoconcentration, during dialysis with cellulosic membranes. Patients were injected with 10 sCi labeled /32m. In one study (N =4), Others have argued that a proper correction of post-dialysis plasma activity was measured over 72 hours. In a second study (N =4), levels should be based instead on the contraction of the patients received low-flux dialysis 24 hours after injection and high- extracellular volume, in which case no anomalous increase in clearance dialysis (BeIlco BL655) at 48 hours. Plasma activities were fitted to a three-compartment, variable volume model. Endogenous f32m f32m levels is seen [18, 19]. The /32m generation rate in dialysis levels (radioimmunoassay) were 56 6 mg/liter. The /32m distribution patients has not been measured. volume was 12.7 2.0 liter (0.20 0.03 liter/kg) and the non-renal One possible means of preventing is to lower clearance was 3.0 0.4 mI/mm. The generation rate, 9.9 1.7 mg/hr serum f32m levels through increased extracorporeal removal. A (0.16 0.04 mglkg/hr), was similar to that measured in subjects with measure of performance might be the fraction of weekly pro- normal renal function. The three compartment model derived from the turnover data gave an adequate fit of the arterial concentrations ofduction that is removed. High-flux dialysis, hemofiltration and endogenous and exogenous f32m during low-flux (nil 132m clearance) and hemodiafiltration all remove measurable amounts of /32m and high-clearance (132m clearance of 19 mI/mm) dialysis. Simulations based can lower serum levels both acutely and in the longer term [15, on this model indicate that extracorporeal treatment can at best remove 20—22], but at present there is no comprehensive model of /32m about 50% of weekly production. These results suggest that $2mkinetics in dialysis patients with which to evaluate these ther- production is not increased in dialysis patients, that there is substantial non-renal f32m clearance, and that the amount of f32m that can beapies. The standard one-compartment model does not ade- removed by extracorporeal therapy is therefore limited. quately describe 2m kinetics during and between dialyses [23, 241, suggesting appreciable intravascular/extravascular resis- tance to mass transfer. Two or more compartments are clearly required. Another important unknown quantity is the non-renal Dialysis-related amyloidosis is a common complication ofclearance of f32m in these patients. The larger this quantity, the long-term dialysis. Consequences of deposition includesmaller the fraction of weekly production which can be re- carpal tunnel syndrome, joint pathology, and bone destructionmoved by a given extracorporeal therapy. through cyst formation [1—5]. A major component of this form This study was undertaken to measure, in dialysis patients of amyloid is f32-microglobulin (/32m) [6, 7], and though otherwithout renal function, the 132m generation rate and extrarenal species may be present as well [8, 9], attention has focused onclearance, and to develop a compartmental model for predicting this 12,000 Dalton . 132m kinetics during dialysis. f32m is normally cleared by the kidneys at a rate comparable to GFR [10, 11], then reabsorbed and catabolized in the tubules, Methods and serum levels are inversely related to GFR [121. Clearance by conventional dialyzers is negligible as these membranes are Preparation of /32-microglobulin impermeable to /32m. Production of f32m in normals is 9 mg/ hr/70 kg [10]. Production may be increased in proliferative Human /32m was isolated from peritoneal dialysate (HIV and disorders [13] and [14] as indicated by highhepatitis B negative). Twenty liters of dialysate were concen- serum levels in the presence of normal renal function. trated 100-fold by ultrafiltration through a cuprophan dialyzer The etiology of dialysis amyloidosis and a strategy for pre-(Asahi AM 710). The concentrate had a /32m concentration of venting it, or at least retarding its progression, remain elusive.approximately 400 mg/mI. It was applied to a Sephacel DEAE High serum /32m levels are an obvious precondition. There has(Pharmacia Fine Chemicals, Uppsala, Sweden) column equili- also been speculation that /32m generation is increased bybrated with 20 mrvt Tris HCI at pH 8.5 and eluted with a hemodialysis through contact with particular membranes, me-discontinuous gradient of NaCI in the same buffer. Fractions chanical damage, or the fall in plasma osmolarity during dialysiscontaining f32m were recovered at 250 to 300 mM NaC1. These were reconcentrated by ultrafiltration (YM5 membrane, Ami- con, Danvers, Massachusetts, USA), then further purified by Received for publication February 26, 1990 two passes on a Sephacryl S200 (Pharmacia) column (4.4 cm X and in revised form December 27, 1990 75 cm) eluted with phosphate buffered saline, pH 7.3 (PBS). Gel Accepted for publication December 28, 1990 filtration of the product showed a single peak at a position © 1991 by the International Society of Nephrology equivalent to 12 kD and showed immunologic reactivity as

909 910 Ode/I et a!: /32mkineticsin renal failure

5000

2500 100 600 (a)

E a 0 > >300 50

Co

Fig. 1. A. Chromogratogram of the injectate (1.0 X 0 0 30cm column). B. Chromatograms of plasma samples taken one hour (solid line, left scale, mean 0 10 20 30 of 4) and 24 hours (dashed line, right scale, mean of 5) after injection (1.0 x 25 cm column). Eluent flow Fraction rate was 0.25 mI/mm, fractions were 1.0 ml. assessed by radloimmunoassay. The purity of this product was Table 1. Patient data confirmed by isoelectric focusing and SDS-PAGE. CAPD HD For each of the two studies performed, 2m was radio- Patient Sex iodinated using a chloramine-T procedure. Approximately 6 mg Age years Diagnosis /32m in 2.6 ml PBS, pH 7.3 was reacted on ice with 800 1 57 M — 12.5 Chronic GN Ci — Na1251 (Amersham, England, UK) and 60 1d chloramine-T 2 65 F 9 Staghorn calculi 3 53 F 4.6 1 Goodpasture's syndrome solution (2.5 mg/mi). After two minutes 60 tl of a solution of 4 64 M — 14.5 Chronic GN sodium metabisuifite (2.5 mg/mi) and potassium iodide (5 mgI a 34 M 3.5 1.5 Wegener's granulomatosis ml) was added to stop the reaction. The entire solution was Abbreviation chronic GN is chronic glomerulonephritis loaded onto a Sephacryl S-200HR column (1.6 cm x 80 cm) aanuric equilibrated with 1% pooled (Common- wealth Serum Laboratories, Parkville, Victoria, Australia) in sterile Duibecco's phosphate buffered saline, pH 7.3. Fractions corresponding to the f32m protein peak were collected, pooled,the second day and four hours of high-clearance dialysis on the and sterilized by filtration (Millex GV 0.2 jim, Millipore,third (in one case, fourth) day. Bedford, Massachusetts, USA) into sterile septum-sealed vials. All patients studied had been maintained on thrice weekly A portion of each vial was sampled for pyrogenicity and sterilityhemodialysis for at least one year prior to the study, four with testing. The specific activity of the product was 20 CiImgcuprophan dialyzers and one with a cellulose acetate dialyzer. protein. A chromatogram of the injectate (Fig. lA) shows aThe patients had been on dialysis, including CAPD, for 9 3 single protein peak and a small free iodide peak. SDS-PAGE ofyears. No patient displayed symptoms or signs of amyloidosis. the injectate (labeled /32m plus 1% human serum albumin) andPatient details are given in Table 1. unlabeled 2m demonstrated low molecular weight bands that uptake of 125Jwasblocked by oral administration of co-migrated and corresponded with the main band in a com-sodium iodide. mercial human 2m preparation (Sigma Chemicals, St. Louis, Missouri, USA). Dialysisstudies Fourhour dialyses were performed using Gambro AK1O Turnover studies machines and sterile dialysate produced by ultrafiltration Two turnover studies were conducted. In the first, fourthrough a Belico BL655 high-flux dialyzer. Low-flux dialysis anephric hemodialysis patients were injected with 10 Ci la-was performed with Gambro cuprophan dialyzers (N =3)and beled f32m on the morning after dialysis. Plasma was sampled atone CDAK cellulose acetate dialyzer. High-clearance dialysis 5, 10, 15, and 30 minutes, 1, 2, 4, 6, 24, 48, and 72 hours. Thewas performed with Belico BL655 cellulose acetate dialyzers. patients were not dialyzed during this period. In the secondBlood flow was 200 mllmin and dialysate flow was 300 mllmin. study, four hemodialysis patients, including three of the firstThe mean weight loss during dialysis was 1.5 kg. group and an additional (anuric) patient, were injected with 10 Arterial plasma was sampled from the cannula immediately Ci labeled 2m on the third day after their previous dialysis.after cannulation, from the blood line at 5 minutes, 1, 2, 3 and Plasma was sampled at 5, 10, 15, and 30 minutes, 1, 2, 4, and 244 hours after the start of dialysis, and by venipuncture from the hours. The patients received four hours of low-flux dialysis onfistula 30 minutes post-dialysis. Venous plasma was sampled at Odell et al: f32mkineticsin renal failure 911

G V1dC1Idt= —(K01+K12+K13+dV1/dt)C1+K12C2+K13C3+G — V2dC2/dt =K12C1 (K12+ dV2/dt)C2

— V3dC3/dt =K13C1 (K13+ dV3/dt)C3

dV1/dt =41IdVd/dt

dV2/dt =412dVd/dt dV2/dt = K01 = Knr + Kr + Kd 413dVd/dt Fig. 2. Thethree compartment modeloff32mdistribution.Kis the K01 =Kr+ Knr + Kd total clearance, K12, K13 are intercompartmental clearance parameters or permeability area products. =Vi/IVi = V1/V The distribution volume, d'isthe sum of the compartmental volumes. The following were assumed: the generation rate, all one hour (low-flux) or at I and 3 hours (high-clearance) forclearances, and weight gain were constant; weight changes clearance measurements. Dialysate was sampled at 5 minutes,were equivalent to changes in distribution (extracellular) vol- and 1, 2, 3 and 4 hours. ume; and the ratios of compartmental volumes to the distribu- tion volume (41k) were constant. Analytical methods The differential equations were solved numerically using a Plasma and dialysate samples were counted (Packard Auto-fourth-order Runge-Kutta routine [25]. The parameters V1, V2, gamma 5650)intactand as supernatants of a 10% TCA (wt/vol,V3, K12, K13, and K01 (= Knr, since Kr = Kd = 0) were final concentration) precipitation. Gel chromatography (Sepha-determined from the plasma disappearance curves by chi- cry! S200, 1.0 x 25 cm column) of plasma samples taken onesquare (weighted least square) fitting using the Marquardt- hour and 24 hours after injection showed no evidence of bindingLevenberg method [25]. The variance of each data point was of labeled f32m to plasma constituents (Fig. 1). Endogenous 2mcalculated from the counting error and an assumed 1% coeffi- concentrations were measured by RIA (Pharmacia). Plasmacient of variation due to other sources of error; each data point samples were diluted 1:20 before analysis. Dialysate sampleswas then weighted by the inverse of the variance. The estimated were assayed undiluted. volumes refer to the time of injection. Superiority of three- over two-compartment fits was assessed by analysis of variance [26]. Modeling The generation rate was calculated from the estimated clear- The plasma disappearance curves of TCA-precipitable activ-ance and the mean serum /32m concentration over the study ity were fitted to the three compartment model in Figure 2. Theperiod: same model was used to simulate dialysis. The model assumes G = KnrCj(endogenous) endogenous input to, and elimination from, compartment one only. Tracer input (bolus) was to compartment one (plasma)Dialyzer clearance was calculated from blood-side measure- and C1(t) was measured. Transfer between compartments andments using: to the environment are expressed in terms of clearances and concentrations. In this formulation, the fluxes between com- Kd = (QIC —Q0C0)IC1 partments 1 and 2 are Q1 Qb(l — H) flux(l—2)= K12C1 Q0 = Qpi— Qf flux (2—* 1) =K12C2 and from dialysate-side measurements using: and total elimination is = QdOCdO/CPI elimination =K01C1 = (Kr+ Knr + Kd)CI where Qb andQd are blood, plasma, and dialysate flow where Kr, Knr, and Kd are renal, non-renal and dialyzerrates, Qisthe ultrafiltration rate, subscripts i and o denote inlet clearance, respectively. The material balance on compartmentand outlet, and H is the hematocrit. 1, with allowance for changing volume, is then Results are expressed as mean SD.

d(V1C1)Idt= V1dC1/dt+ C1dV1/dt Results = + K12 + K13)C1 + K12C2 + K13C3 + G TCA-precipitable and non-precipitable 125Jactivityare —(K01 shown for one patient (first study) in Figure 3. The rise in where G is the endogenous production rate. Material balancesnon-precipitable activity, after the initial dilution, suggests on compartments 2 and 3 are derived similarly and, afterpossible catabolism of the injected protein. The disappearance rearranging, the differential equations describing the compart-curves of precipitable activity are plotted in Figures 4 and 5 for mental concentrations become: the first and second studies, respectively. 912 Ode!!etal:f32mkinetics in renal failure

5000 twice as large, comparable to the iodine space [28], was assumed. The dialysis results are summarized in Table 3. Dialyzer clearances of labeled and endogenous /32m were similar (Wilcoxon matched pairs test, P =0.1)and clearances of both species were nil with low flux dialysis. With high-clearance ii. 1000 dialysis, clearances measured on the blood side were higher > than those measured on the dialysate side. Two possible U explanations are uptake by the membrane and loss of f32m from E the dialysate samples to the tube walls. During low flux dialysis C 0 endogenous fl2m levels rose 14 8% while labeled 2m levels rose 1 4%. Levels of both solutes fell an average of 23% during high-clearance dialysis. Mean arterial concentrations of endogenous and labeled f32m 100 during dialysis are shown in Figure 7, along with the concen- 0 24 48 72 trations predicted from the three-compartment model, using the individual patient parameters in Table 2 with appropriate ad- Time, hours justments to the volumes. A J32m clearance of 19 mllmin Fig. 3. Concentrations of TCA-precipitable (0) and non-precipitable (0)activity,patient 2, first study. (high-clearance) or zero (low-flux) was assumed. The agree- ment between model and data indicates that volume changes alone are sufficient to explain the rise in f32m concentration during low flux dialysis and that the model has predictive validity for low dialyzer clearances, at least. There is no The three-compartment model gave significantly better fits of the 72-hour turnover data (first study) than did the two-com-evidence that endogenous and exogenous /32m behaved differ- partment model (P < 0.001). Clearance was systematicallyently. overestimated (7 2%) and distribution volume underesti- The likely impact of therapies such as hemofiltration and mated (9 3%) by the two-compartment model. Since accuratehemodialfiltration which can remove substantial amounts of estimation of clearance and distribution volume require that the132m was assessed by simulation, under the assumption that tail of the disappearance curve be well-defined, in the secondgeneration and non-renal clearance are independent of concen- study it was assumed that no f32m was removed during thetration. Theoretical concentration profiles and removal rates low-flux dialysis and the three-compartment model was fitted tounder different regimens were calculated using the mean param- -the-data over--48-(or 72)hollrs. Pre-dialysis, but not intra- oreter values in Table 2 and the following assumptions: three post-dialysis, data points were included in the fit and allowancetre-atments- -of -the- -same- -duration-spaed at 2, 2 and 3 day was made for the change in volume during the first dialysis. intervals; constant rate of weight gain; and the same post- The parameter estimates from the two studies are given indialysis weight for all dialyses. The initial 32m concentration Table 2. The clearance was 3.00.4 mi/mm, considerably lesswas set at 50 mg/liter and the model was run for three simulated than the 7.7 mi/mm measured in anephric sheep [27]. Theweeks. The results summarized in Figures 8 to 11 refer to the estimated plasma volume was 3.4 0.5 liter (53 9 ml/kg) andthird week, though concentrations stabilized by the second the distribution volume was 12.7 2.0 liter (200 30 mllkg),week. comparable to the expected plasma and extracellular volumes. Concentration profiles during the midweek dialysis are shown The f32m generation rate was 9.9 1.7 mg/hr (11.1 2.9in Figure 8 for zero and 2 liter weight loss. For comparison, mg/hr/70 kg), similar to the generation rate measured in normalsprofiles based on a one-pool model with the same total distri- by Karisson et al [10]. bution volume are also shown. Nonlinearity, seen even when Under the assumption that free (non-precipitable) 125Jactiv-volume is held constant, and the post-dialysis rebound reflect ity might be indicative of catabolism of the /32m molecule, thethe substantial intravascular/extravascular barrier. Aside from fractional conversion of protein-bound to free 125jwasesti-dialyzer clearance the important factors are the intrabody mated by performing a mass balance. The increase in free 125Jclearances (compare the one and three-pool models) and weight activitywas compared to the decrease in bound 125!activityloss. Renal clearance up to 5 mi/mm and variations in genera- over an interval when disequilibrium between compartmentstion rate had little effect on post/pre-ratios. Nor was there a was expected to be minimal. In the first study the chosensignificant difference in post/pre-ratios between the first and interval was 24 to 72 hours; in the second, it was the intervalsecond weeks of the simulation. from 30 minutes post-dialysis (24 hours) to the start of the Weekly profiles (with Kr = 0) are shown in Figure 9 for following dialysis. The changes in concentration are plotted inclearances up to 80 mI/mm. With Kd = 0, concentrations rise Figure 6. The total mass of labeled f32m at any time wasduring dialysis and fall during the interval. The other profiles obtained from the three-compartment model, and the mass ofhave a more conventional shape, with concentration highest free label was calculated as the product of the plasma concen-after the long interval. In Figure 10, weekly removal as a tration and an assumed distribution volume. The fractionalfraction of total f32m production is plotted against Kdtd/V for conversion was 37 13% if the distribution volume of free labelseveral levels of residual renal function. While a six hour was assumed equal to that of f32m, and 65 26% if a volumeprocedure gives slightly greater removal, for the same Kdtd/V, Ode!! et a!: f32mkineticsin rena! failure 913

5000

> 1000 C.,

CD E CD CD

200 0 24 48 72 Fig.4. Concentrationsof TCA-precipitab!e Time,hours activity, first study.

5000

E > > 1000 0 CD CD E DC 0CD

200 Fig.5.Concentrations of TCA-precipitab!e 0 24 48 72 activity, second study. Low-fluxdialysisat 24 hours,high-clearance dialysis at 48 (or 72) Time, hours hours.

thana four hour procedure, the difference is not large. Thisbecause the therapy is intermittent and competing with the 3 suggests that Kdtd/V may be useful for quantifying the amountmi/mm non-renal clearance, not because of intrabody mass of dialysis with respect to J32m. It is also apparent that inter-transfer resistance. Finally, midweek predialysis levels for mittent therapy can, at best, remove about 50 to 60% ofdifferent levels of Kr are plotted against Kd in Figure 11. Again, production, less if renal function is present. Comparison withthe influence of residual renal function is clear. The decrease in the one-pool model indicates that removal is limited primarilypre-dialysis 2m levels to be expected with an increased Kd, 914 Ode!!et a!: f32m kinetics in renal failure

Table 2. Parameter estimates for three compartment model Serum G Weight V1 Knr V2 K12 V3 K13 Vd f32m mgi Patient kg liter mI/mm liter mI/mm liter mI/mm liter mg/liter hr First study 1 76 3.1 2.6 3.2 75 6.1 26 12.4 58 9.0 2 53 2.8 2.9 1.7 57 6.3 31 10.8 63 10.8 3 58 3.9 3.5 3.8 93 6.0 22 13.6 52 11.0 4 66 3.9 2.7 2.7 77 6.6 28 13.3 56 9.0 Mean 63 3.4 2.9 2.8 75 6.3 27 12.5 57 10.0 SD 10 0.6 0.4 0.9 15 0.3 4 1.2 5 1.1 Second study 1 77 3.4 3.3 2.3 79 7.2 39 12.4 52 10.3 2 56 2.7 3.4 1.9 66 7.4 32 12.0 63 12.7 4 66 3.8 2.7 2.1 105 10.7 35 16.6 55 8.9 5 58 3.3 2.6 1.9 81 4.8 28 10.1 47 7.3 Mean 64 3.3 3.0 2.1 83 7.5 34 12.9 54 9.8 SD 10 0.4 0.4 0.2 16 2.4 5 2.7 7 2.3 Per kg, both studies Mean 0.053 0.047 0.039 1.25 0.109 0.48 0.201 0.159 SD 0,009 0.010 0.011 0.25 0.027 0.09 0.032 0.041

150

100

> a0

C a a 50 I) 0 C

Fig. 6. The increase in concentration of non- precipitable 125jplottedagainst the decrease in 0 concentration of precipitable 125J,Theincrease and 0 100 400 500 decrease are relative to the concentrations at t 24 200 300 hr (30 minutes after low-flux dialysis in the second Decreasein bound activity, cpm/mI study).

both in absolute andrelativeterms, is strongly dependent onrequired to fit turnover data in dialysis patients. The parameter Kr. estimates, except for the clearance, were similar to those previously determined in anephric sheep [27]. The estimated Discussion distribution volume, 20% of body weight, was similar to the Whereas a two-compartment model provides an adequate fit17% measured in normals [10] and the 18% estimated from fJ2m to f32m kinetics when renal function is normal [10, 271, thekinetics during hemofiltration [19]. turnover data in this study were better fitted by three compart- There was no evidence that /32m generation was increased in ments: a small, rapidly equilibrating compartment, presumablythese patients. Generation rates were similar to those deter- representing the viscera with their more open capillary archi-mined in normals [10], perhaps slightly higher when adjusted for tecture, and a larger, more slowly equilibrating compartment,body weight. The rise in f32m levels during low-flux dialysis was both communicating with the plasma compartment. Chanard etadequately explained by contraction of the extracellular vol- al [291 have also recently reported that three exponentials areume, as first suggested by Bergstrom and Wehie [18], and as Ode!! et a!: f32m kinetics in renal failure 915

Table 3.Clearances (mllmin) and post-dialysis/pre-dialysis profiles of Floege et a! [23] when replotted on a logarithmic concentration ratios for labeled and endogenous /32m during scale, and can also be inferred from the post-procedure rebound low-flux and high-clearance dialysis reported by Canaud et al [24], which was interpreted by them as Labeled Endogenous extra generation of 2m. Volume changes, whether due to fluid f32m /32m removal, as modeled here, or to fluid shifts between the Low-flux extracellular and intracellular spaces [19], which was not con- Clearance: blood side (N =4) 2 5 —1 8 sidered, also affect the concentration profiles. Residual renal dialysate size (N =20) 1 1 0 1 Cpost/Cpre (N =4) 1.01 0.04 1.14 0.08 function and small changes in generation rate had negligible High-clearance effect. Clearance: blood side (N =9) 17 8 22 4 Over the longer term, the simulations showed that concen- dialysate side (N =18) 2 5 —1 8 trations stabilized one week alter beginning a new therapeutic 0.77 0.05 0.77 0.06 Cpost/Cpre (N =4) regimen, indicating that two to three weeks is sufficient time to evaluate a procedure. The serial measurements of Blumberg and Burgi [32] illustrate this well. The simulations also showed that the reduction in predialysis /32m levels resulting from a demonstrated by others using myoglobin [301 and inulin [19] asgiven clearance is smaller than might be expected, particularly extracellular volume markers. It is perhaps worth noting thatwhen renal function is present. Studies of high-removal tech- simulation showed that an assumed 100% increase in /32mniques have achieved reductions of 20 to 37% in predialysis f32m generation rate during dialysis would increase post-dialysislevels [20—22, 32], but comparison with the model is not levels an additional 10% only. It is clear then that the hypoth-possible in most cases because sufficient information, especially esis that amyloidosis results from enhanced f32m production isclearances, was rarely given. Shinzato, Kobayakawa and not tenable. Maeda [21] reported a decrease in predialysis levels of 27% with A second hypothesis is that amyloidosis is simply a conse-HDF using the Hospal AN69 HF with a clearance estimated to quence of prolonged exposure to elevated j32m levels and thatbe 35 ml!min under the conditions employed (Fig. 1 in their development of amyloidosis can be slowed only by reducingpaper). Adjusting for the different mean patient weight (50 kg) those levels. The non-renal clearance of 3 mI/mm and the modeland longer procedure (4.5 hr), the expected decrease according of 132m kinetics offer some guidance here. The non-renal clear-to the model is 31% for mid-week pre-dialysis levels and 23% ance is such that a small additional clearance, acting continu-for the first dialysis of the week, indicating rough agreement. ously, can have a significant impact on f32m levels. ThusFinally, the inherent limitations of intermittent therapy are Vincent et a! [31] showed that f32m levels in hemodialysis clearly illustrated in Figure 10, which suggests that extracorpo- patients increased markedly with time and that this increasereal removal can account for, at best, about half of weekly 132m was largely explained by greater diuresis in short-term patients.production. Assuming amyloid deposition is proportional to Similarly, Blumberg and Burgi [32] found a significant correla-132m concentration, this would double the symptom-free inter- tion between the reciprocal of serum f32m and creatinine clear-val. The idea that hemofiltration or any other method can ance. In terms of our results, this relation can be written remove total weekly /32m production [39] and thereby prevent 1/C =(Knr+ Kr)/G =(18+ 6Kr) X103(liter/mg) amyloidosis is fallacious. Over a ten year period the average individual produces, and which fits their data reasonably well. Others have stressed thedisposes of, close to a kilogram of f32m. In the anephric patient importance, in comparative studies, of matching groups forundergoing conventional dialysis, disposal routes are limited to similar residual function [33]. Periodic measurement of renaltotal or partial catabolism and deposition of the intact molecule function in longitudinal studies would seem equally important.or fragment thereof as amyloid. The clearance measured in the CAPD also provides a small f32m clearance on the order of 0.7turnover studies represents the sum of these processes. Depo- to 1.0 mI/mm [32, 34, 35] and assuming Knr =3mllmin and nosition of '311-labeled 32m in joints a few days after injection has renal function, one would expect /32m levels in CAPD patientsbeen demonstrated [40] and 2m fragments have also been to be 75 to 80 percent of those in hemodialysis patients. Thoughidentified in amyloid [8]. These several pathways cannot be measured ratios range widely, from 60 to 100% [32, 36—38],distinguished with certainty from the present data, but assum- when concentrations were corrected for renal function the ratioing that J32m fragments constitute a minor fraction of amyloid, was approximately 75% [36, 37]. that excretion and thyroid uptake of free 125J were negligible, The simulations of intermittent therapy were based on aand that non-precipitable 125J was a catabolic product, the compartmental model that was derived from turnover studiesestimated conversion of bound to free 125Jactivitysuggests that conducted under conditions that approximated steady state andone-half or more of the injected protein may have been catab- was tested successfully against minor perturbations of 132molized. Thus a third hypothesis regarding the pathogenesis of levels. While extrapolation beyond these conditions entailsamyloidosis is that some other factor in the presence of elevated some uncertainty, the simulations provide at least a roughf32m levels favors deposition as amyloid over catabolism. Re- perspective on the problem of manipulating f32m levels. Pre-duced proteolysis [41] and chronic subclinical [42] dicted concentration profiles during dialysis (or other proce-have been suggested, as has aluminum exposure [43], although dure) did not follow a single exponential decay, indicating thethe increased burden of chelatable aluminum in patients with inappropriateness of a one-pool model. This was seen in ouramyloidosis may be effect rather than cause. It could also be experimental results (Fig. 7), is evident in the concentrationthat the previously thought to cause amyloidosis by 916 Odell et al: f32m kinetics in renal failure

1.5

1.0 0 0

C-)

0.5 Fig. 7. Arterial concentrations of endogenous (square) and labeled (circle) f32m during low-flux (filled symbol) and high-clearance (open symbol) Time, hours dialysis. Lines represent model predictions.

1.5

1.0

0 0.5 0

Fig. 8. Concentration of 2m during and after dialysis 0.2 for extracorporeal clearances Kd =0to 80 mI/mm. 2 0 1 2 3 4 5 6 Weight =64kg. Three-pool model, weight loss =kg (solid lines) or zero (dashed lines). One-pool model, Time, hours weight loss =2kg (dotted lines). increasing /32m production [151 do so by creating conditionsthis is so, reduction of f32m levels might have a greater thai favoring amyloid formation [44]. proportional effect of the rate of amyloid formation. Finally, it is conceivable that f32m itself is a pathogenetic factor. There is accumulating evidence that /32m is not simply a Appendix 1. Nomenclature structural adjunct to the MHC class I molecule but has signif- icant biological activity in its own right. It has been shown tovi volume of compartment i stimulate collegenase synthesis in synovial cells [45], to influ-Vd distribution volume ence the activity of bone cells [46], and to have chemotacticci concentration in compartment i activity [47]. It is possible, then, that elevated /32m levels aloneG generation rate (mass/time) can induce a pathology which results in amyloid deposition. If renal clearance (volltime) Odell et alt 2m kinetics in renal failure 917

60

L g)

E o40 Co C G) C) C 0 C) Co E U) a-. 20

Fig. 9. Weekly 2m concentration profile according to —2 —1 0 1 2 3 4 5 6 7 8the three-pool model with K, =0, =4hr, weight = Time,days 64kg, weight gain =1kg/day, Kd =0to 80 mI/mm.

Kd, mi/rn/n Kd, mi/rn/n

20 40 60 80 20 40 60 80 80 60

C- 50 60 0) E 0C 40 Co Co C > a) 040 C) E 0C a) C) 30 0)

(0 20 0) 20 a-

10 0 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 Kt/vol KU . Fig.11. Midweek predialysis f32m levels versus Kdtd/V andKdfor K, = Fig.10. Extracorporeal removal (% of production) of /32m versus0 to 5 mI/mm. Three-pool model, weight =64kg, td =4hr. KdtIV for different levels of renal function. Weight =64kg. Three pool model, td =4hr (solid lines); three pool model, td =6hr (dashed line); one-pool model, td =4hr (dotted line). Upper scale gives Kd for td= 4 hr. Qd clialysate flow rate, mllmin Q, ultrafiltrationrate, mllmin C plasma concentration Knr non-renal clearance ho (subscript)inlet/outlet or arterial/venous Kd extracorporeal clearance H hematocrit K1 clearance between compartments i and j, (vol/time) fraction of total distribution volume change Acknowledgment assigned to compartment The assistance of ENKA AG (Wuppertal, F.R.G.) isgratefully Q plasmaflow rate, mllmin acknowledged. 918 Ode!!et a!: p2m kinetics in renal failure

Reprint requests to Dr. Ross A. Ode!!, Centre for Biomedical21. SHINZATO T, KOBAYAKAWA H, MAEDA K: Comparison of various Engineering, University of New South Wales, P0 Box 1, Kensington, treatment modes in terms of p2-microglobulin removal: Hemodial- New South Wales, 2033 Australia. ysis, hemofiltration, and push/pull HDF. Artif Organs 13:66—70, 1989 22. SUZUKI M, TAKAHASHI 5, HIRASAWA Y: Therapeutic approach to References dialysis-related osteoarthropathy. Nephrol Dial Transplant 4 (suppl 3):32—37,1989 1. BARDIN T, ZINORAFF J,KUNTZD, DRUEKE T: Dialysis related 23. FLOEGE3, WILKS M, SHALDON 5, KOCH KM, SMEBY L: p2- amyloidosis. Nephrol Dial Transplant 1:151—1 54, 1986 microglobulin kinetics during haemofiltration. Nephro! Dial Trans- 2. WALTS AE, GOODMAN MD, MATORINPA:Amyloid, carpal tunnel plant 3:784—789,1988 syndrome, and chronic hemodialysis. Am J Nephrol 5:225—226, 24.CANAUDB, A5SOUNGA A, IMBERT E, KAAKI M, AZNAR R, 1985 ROBINET-LEvY M, MI0N C: Simplified /32m kinetic modelling 3. BARDIN T, KUNTZ D, ZINORAFF J, Voismt MC, ZELMAR A, analysis applied to a daily post dilutional hemofiltration program. LAN5AMAN 3: Synovial amyloidosis in patients undergoing long- (abstract) Blood Purif 5:275, 1987 term hemodialysis. Arthr Rheum 28:1052—1058, 1985 25.PtassWH, FLANNERY BP, TEUKOLSKY SA, VETTERLING WT: 4. MUNOZ-GOMEZ J, BEROADA-BARADO E, GOMEZ-PEREZ R, LL0- Numerical Recipes: The Art of Scientific Computing. Cambridge, PART-BUISAN E, SusIAs-SoBaavIA E, ROTES-QUEROL 3, SOLE- Cambridge University Press, 1986 ARQUES M: Amyloid arthropathy in patients undergoing periodical 26. LANDAW EM, DISTEFANO 33 III: Multiexponential, multicompart- haemodialysis for chronic renal failure: A new complication. Ann mental, and noncompartmental modeling. II. Data analysis and Rheum Dis 44:729—733, 1985 statistical consideration. Am J Physiol 246:R665—R677, 1984 5. HUAUX JP, NOELH,MALGHEMJ,MALDEGUE B, NAGANT DE 27. ODELL RA, SLOWIAçZEK P, SCHINDHELM K: p2-Microglobulin DEuxdilAisNEs C: Erosive azotemic osteoarthropathy: Possible kinetics in an ovine model. Trans Am Soc Artif Organs 35:625—629, role of amyloidosis. Arthr Rheum 28:1075—1076, 1985 1989 6. GEJYO F, YAMADA T, ODANI 5, NAKAGAWA Y, AIwAwA M, 28. BERMAN M: Iodine kinetics, in The Thyroid and Biogenic Amines, KuNIT0M0 T, KATAOKA H, SuzuIcI M, HIRA5AWA Y, SHIRAHAMA edited by RALL JE, KOPIN Ji, Amsterdam, North Holland Publish- T, COHEN A, SCHMID K: A new form of amyloi4 protein associated with hemodialysis was identified as p2-microglobulin. Biochem ingCo, 1972, p 182 29.CHANARD J, VINCENT C, CAUDWELL V, BERNIEH B, LAVAUD 5, Biophys Res Comm 129:701—706, 1985 TOUPANCE 0, REvILLARD JP: Plasma '251-p2-rnicroglobulin turn- 7.GotavicPD, CASEY TT, STONE Wi, DIRAIMONDO CR, PRELLI FC, FRANGIONE B: Beta-2-microglobulin is an amyloidogenic pro- over in chronic haemodialysis patients. (abstract) Nephrol Dial tein in man. J Clin Invest 76:2425—2429, 1985 Transplant 4:492,1989 8. LINKE RP, BOMMER J, RITZ E, WALDHERR R, EULITZ M: Amyloid 30.FLOEGE 3, EISENBACH GM, MERSCHER 5, COLTON CK, SHALDON Stones Qf uremic patients consist of beta2-microglobulin 5: Plasma p2-microglobulin and myoglobin kinetics during cu- fragments. Biochem Biophys Res Comm 136:665—671, 1986 prophan hemodialysis. (abstract)BloodPurif 6:353—354, 1988 9. SAITO A, QGAWA H, CHUNG TO, OHKuBO I: Accumulation of 31.VINCENTC, REVILLARD JP, GALLAND M, TRAEGER J: Serum serum amyloid P and its deposition in the carpal tunnel region of p2-microglothnin hemodialyzed patients.Nephron 21:260—268, long-term hemodialysis patients. Trans Am Soc Art if Intern Organs 1978 32. BLUMBERG BURGI 33:521—523,1987 A, W: Behavior of p2-microglobulin in patients 10.Kt.assor.t FA, GROTH G, SEGE K, WIBELL L, PETERSON PA: with chronic renal failure undergoing hemodialysis, hemodiafiltra- Turnover in humans of /32-microglobulin: The constant chain of tion and continuous ambulatory peritoneal dialysis (CAPD). Clin HLA-ãntigens.EarJ Clin Invest 10:293—300,1980 Nephro! 27:245—249,1987 11.GAUTIER C, NGUYEN-SIMONNET H, VINCENT C, REVILLARD JP, 33. TIELEMANSC, DRATWA M,BERGMANNP.GOLDMANM,FLAMION PELLET MV: Renal tubular absorption of p2 microglobuliñ. Kidney B, COLLART F, WENS R: Continuous ambulatory peritoneal dialy- 1nt26:l70—l75, 1984 sis, protective against dialysis-associated amyloid? Nephron 53: 12. KAmssON FA, WIBELL L, EVRIN PE: /32-microglobulin in clinical 174—175, 1989 medicine. Scand J Clin Lab Invest 40 (suppl l54):27—37, 1980 34. SETHID, MURPHY CMB, BROWN EA, MUELLER BR, GOWER PE: 13. SCHARDUN GHC, STATIUS VAN En LW: /32-microglobulin: Its Clearance of beta-2-niicroglobulin using continuous ambulatory significance in the evaluation of renal function. Kidney mt 32:635— peritoneal dialysis. Nephron 52:352—355, 1989 641, 1987 35.KREDIETRT, KOOMEN 0CM, KOOPMAN MG, HOEK FJ, STRUIJK 14. MANICOURT D, BRAUMAN H, ORLOFF 5: Plasma and urinary levels DO, BOESCHOTEN W, Amzs L: The peritoneal transport of serum of /2 microglobulinin rheumatoid arthritis. AnnRheum Dis 37:328— proteinsand neutral dextran inCAPD patients. Kidney Int 35:1064- 332, 1978 1972,1989 15. FLOEGE3, ORANOLLERAS C, BINGEL M, DESCHODT 0, BRANGER 36. TIELEMANSC, DRATWA M,BERGMANNP,GOLDMANM,FLAMION B, OULES R, KOCH KM, SHALDON 5: /32-microglobulin kinetics B,COLLARTF,WENSR:Continuous ambulatory peritoneal dialysis during haemodialysis and haemofiltration. Nephrol Dial Transplant vs. haemodialysis: A lesser risk of amyloidosis? NephrolDial 1:223—228, 1987 Transplant 3:291—294,1988 16. SHALDON 5, KOCH KM, DINARELLO CA, COLTON CK, KNUDSEN 37. SETHID,GOWERPE:Dialysis arthropathy, j32-microglobulin and P3, FLOEGE 3, GRANOLLERAS C: (32-microglobülin and haemodial- the effect of dialyZer membrane. NephrolDial Transplant 3:768— ysis. Lancet i:925—926, 1987 772, 1988 17. KNUDSEN P3, LEON 3, NG A-K, SHALDON 5, FLOEGE 3, KOCH 38. BALLARDIEFW,KLRRDNS,TENNENT0, PEPYS MB:Haemodi- KM: Hemodialysis-related induction of beta-2-microglobulin and alysiS versus CAPD: Equal predisposition to amyloidosis? Lancet -l synthesis and release by mononuëlear phagocytcs. i:795—796,1986 Nephron 53:188—193,1989 39. VIENKEN3, HARDING M:Dialysis membranes and /32-microglob- 18. BERGSTROMJ, WEHLE B:No change in corrected beta2-microglob- ulin:Interactions, limitations and perspectives. Nephrol Dial ulin concentration after cuprophane haemodialysis. Lancet i:628— Transpl 4 (suppl 3):27—3l,1989 629,1987 40.FLOEGE J, NONNA5T-DANIEL B, GIELOW P, BRANDIS A, SPINDLER 19. H0NIG R, MAR5EN T, SCHAD 5, BARTH C, POLLOK M, BALDAMUS F, HUNDE5HAGEN H, KOCH KM, SHALDON 5: Specific imaging of CA: Correction of beta2-microglobulin changes to changes of dialysis-related amyloid deposits using °11-beta-2-microglobulin. distribution volume. Int J Artif Org 11:459—464, 1988 Nephron 51:441—447,1989 20. BOMMER 3, SEELIG P, SEELIG R, GEERLING5 W, BOMMER 0, RITZ 41.ARGILE5 A,MouRAD 0, ATKINSRC,MIONC: Haemodialysis- E: Determinants of plasma p2microglobulinconcentration: Possi- associated amyloidosis and anti-proteases. (abstract) Nephrol Dial ble relation to membrane biocompatibility. Nephrol Dial Trans- Transplant 4:492,1989 plant 2:22—25,1987 42. CURATOLA0, MALARA E,SISCA5,PO5TORINOM, CUTRUPI 5, Odell et al: j32m kinetics in renal failure 919

MAGGIORE Q:Biologicalmarkers of inflammation and carpal tunnel BECKERMAN B, BENSON MD: Autocrine induction of collegenase syndrome in dialysis patients. (abstract) Nephrol Dial Transplant by serum amyloid A-like and f3-microglobulin-like . Sci- 4:492—493, 1989 ence 243:655—657, 1989 43. YVER L, BLANCHIER D, BUIQUANG D, CABANNE JF, CHARMES JP, 46. CANALIS E, MCCARTHY T, CENTRELLA M: Growth factors and the MEFTAHI H: Does aluminum induce dialysis amyloidosis? (ab- regulation of bone remodelling. J Gun Invest 81:277—281, 1988 stract) Nephrol Dial Transplant 2:450—451, 1987 47. DARGEMONT C, DUNON D, DEUGNIER MA, DENOYELLE M, 01- 44. STONE WJ, HAKIM RM: Beta-2-microglobulin amyloidosis in long- RAULT JM, LEDERER F, LE KHD, GODEAU F, THIERY JP, IMH0F term dialysis patients. Am J Nephrol 9:177—183, 1989 BA: Thymotaxin, a chemotactic protein, is identical to /32-micro- 45. BRINCKERHOFF CE, MITCHELL TI, KARMILOWICZ MJ, KLUVE- globulin. Science 246:803—805, 1989