Multiple-dose pharmacokinetics of ganglioside GM, after intravenous and intramuscular administration to healthy volunteers Ganglioside GM, multiple-dose pharmacokinetics were investigated in five healthy male volunteers. Doses of 100 mg were administered either intravenously or intramuscularly for 21 days, and the washout was followed-up for a further 21 days. The highly specific binding of the P-subunit of cholera toxin was used to quantify ganglioside GM, levels in plasma, urine, and feces. This dose regime increased the ganglioside GM, steady-state plasma levels two to three orders of magnitude above the endogenous levels of 0.132 mglL. (coefficient of variation, 8.9%). Large and variable amounts of ganglioside GM, were found in feces before and during treatment without relation to the dosage. No ganglioside GM, could be detected in urine at any time. Plasma kinetics were linear with a biexponential disposition. Exogenously administered ganglioside GM, was confined mainly to the blood volume as indicated by a steady3.57 L and appears to be excreted mainly in the form of metabstate volume of distribution of 6.98 olites. The total clearance was very slow at 1.61 +. 0.37 mVmin. Absorption after intramuscular administration was slow (time to reach maximum concentration >12 hours) and yielded steady-state THER concentrations somewhat lower compared with the intravenous infusion. (CLINPHARMACOL 1991;50:141-9.)

*

K. Ludwig Rost, MD, Jiirgen Brockmoller, MD, Willi Weber, MD, PhD, and Ivar Roots, MD ~ e r l i n , - ~ m z a n ~ The monosialotetrahexosyl ganglioside GM,, a constituent of mammalian neuronal cell membranes,' is thought to play a role in axonal growth, differentiation, and repair.2 It is also believed to mediate several membrane signal transduction p r o c e ~ s e s .However, ~ the molecular mechanisms of action involved in these different functions are poorly understood. The therapeutic interest in ganglioside GM, arose with the discovery that it accelerated regeneration of nervous tissue, partly in cooperation with nerve growth factor. Long-term ganglioside GM, treatment of animals with experimentally induced damage of the central nervous system not only increased sprouting

From the Institute of Clinical Pharmacology, Klinikum Steglitz, Free University of Berlin. Supported by Fidia Pharmaforschung, Munich, Germany. Received for publication Jan. 2, 1991; accepted April 18, 1991. Reprint requests: Ivar Roots, MD, Institute of Clinical Pharmacology, Klinikum Steglitz, Hindenburgdamm 30, D-1000 Berlin 45, Germany 13/1/30292

and enhanced recovery but also augmented behavioral recovery. Moreover, ganglioside GM, is thought to facilitate the short-term recovery and to therefore reduce mortality in acute cerebrovascular injury by protection of membrane f ~ n c t i o nThese .~ experiments indicate a potentially important pharmacologic role for ganglioside GMl in facilitating the recovery in nerve cell distress or impairment. Clinical investigations to evaluate the therapeutic potential of ganglioside GM, or ganglioside mixtures led to an amelioration of symptoms in several types of peripheral neuropathies5 and functional recovery in patients who have had stroke^.^" However, the capricious nature of these positive results dampened the initial optimism. An essential step toward ironing out some of these inconsistencies is the investigation of ganglioside GM, pharmacokinetics. Such information would enable the optimization and standardization of a ganglioside GM, therapy. To date the kinetic properties have been examined , ' ~ tritium-labeled ganglioin rats8-" and m i ~ e ' ~ with side GM,. Parenteral administration is necessary be-

141

142 Rost et al. cause of degradation by intestinal enzymes. Ganglioside GM, is absorbed from intramuscular and subcutaneous sites of administration. In the mouse the intramuscular application lead to 10-fold lower levels than the intravenous r o ~ t e . ~ . 'In~ ,its' ~micellar form, ganglioside GM, is bound to plasma albumin, forming stable lipoprotein complexes.'4 It is distributed quickly to the liver where it is extensively metabo~ i z e d , ~and , ' ~ to a much lesser extent it also concentrates in the lung, kidney, spleen, muscle, and brain, where it is incorporated i n t r a c e l l ~ l a r l ~Elimination .~~'~ occurs almost exclusively in the form of metabol i t e ~ . ~Tissue " ~ peak levels in the experimental animals were reached within 2 hours of intravenous administration, followed by a rapid ~learance.".'~In contrast, Lang8 reported a slow elimination half-life of 60 to 70 hours for ganglioside GM, and its metabolite~.~ In this pharmacokinetic investigation, ganglioside GM, was determined by exploiting its highly specific binding to the P-subunit of the cholera Aware that the pharmacokinetics of ganglioside GM, in humans are to a great extent unknown, we avoided argumentation within a special compartment model. Instead, we described the kinetics using linear system analysis, which is based on a minimum of assumptions.

MATERIAL AND METHODS Study design. This project was an open multipledose pharmacokinetic phase I study. Twenty-one doses of highly purified ganglioside GM, (Sygen, Fidia, Abano Terme, Italy) were administered to five volunteers on 21 consecutive days. Three received an infusion of 100 mg ganglioside GM, for Y2 hour daily. The remaining two volunteers received a daily 100 mg injection into the glutei medii at various defined sites. The washout phase was examined for 21 days after the last dose. Before treatment, endogenous ganglioside GM, levels were determined in plasma on 3 different days and in urine and feces on 2 different days. The time-concentration curves over 24 hours were investigated on the first and last day of administration at 0, 10, 20, 30, 45, and 60 minutes and at 1Y2, 2, 3, 5, 7, 9, 12, and 24 hours for intravenous administrations, and at 0, 30, and 60 minutes and at 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, and 24 hours for intramuscular administrations. On these days volunteers were served a controlled low-fat diet. Plasma peak levels were estimated weekly immediately after intravenous and 12 hours after intramuscular administration, respectively. Altogether, 15 trough levels were sampled between 8 and 9 AM immediately before dos-

CLIN PIIARMACOL THER AUGUST 1991

age. Such plasma samples were also taken during the washout phase on study days 22, 25, 28, 35, and 42. Urine and feces were sampled in two 24-hour collection intervals on 2 consecutive days before ganglioside GM, treatment, then on days 1 and 2 (first dose), on days 21 and 22 (last dose), and on days 27 and 42 during the washout phase. General and local side effects of the medication were monitored daily. This study was performed in accordance with the Declaration of Helsinki. Approval of the ethics committee of the Klinikum Steglitz was obtained. Subjects. Five healthy male volunteers (27 to 33 years of age) participated after giving their informed consent. The volunteers were nonsmokers and did not take any additional drugs during the whole period of investigation. The abuse of cocaine, amphetamines, morphine, LSD, cannabis, benzodiazepines, and barbiturates was excluded by monitoring during the first and third week of the study. All volunteers were negative for human immunodeficiency virus and hepatitis A and B. Before and during the 6-week period of investigation, a total of five examinations including ECG, blood pressure, and blood and urine laboratory parameters were performed to ensure the health of the volunteers. Blood laboratory parameters included total and differential blood counts, creatinine, alanine aminotransferase, aspartate aminotransferase, y-glutamyltransferase, serum proteins, electrolytes, glucose, cholesterol, and triglycerides. Semiquantitative urine parameters included leucocytes, nitrite, pH, protein, glucose, ketone bodies, blood, and bilirubin. No noticeable deviation from normal status was observed in any of the volunteers. Drug assay. Plasma recovered from 10 ml of EDTA-treated blood samples and aliquots of 24-hour urine samples were stored at -80" C pending ganglioside GM, extraction. The 24-hour feces samples were diluted to 1 L with double-distilled water before homogenization. Aliquots of the homogenates were freeze-dried and stored at -80" C. Extraction and determination were performed with a modification of the method of Kirschner et a1.16 Ganglioside GM, was extracted three times from 0.5 ml plasma as follows: 2 ml tetrahydrofurane (THF, Merck, Darmstadt, Germany) was mixed with the plasma. The sample was centrifuged and the supernatant collected. The pellets were resuspended in 0.5 ml phosphate buffer, pH 6.8, and the procedure was repeated two more times. Diethylether (2.25 ml) was mixed with the pooled supernatants. After centrifugation the aqueous phase was collected. Doubledistilled water (0.85 ml) was added to the remaining

VOLUME SO NUMBER 2

organic phase, and the above procedure repeated again. The organic phase was discarded and the vial rinsed with 0.2 ml double-distilled water, which was subsequently pooled with the aqueous extracts. These extracts were dried overnight in a Speed Vac concentrator (Bachofer, Reutlingen, Germany ), redissolved in the original volume of 0.5 ml with double-distilled water, and stored at -20" C until determination. Microtiter plates (immunoplate maxisorb F69, NUNC, Roskilde, Denmark) were preincubated with 100 p1 of 1 p,mol/L ganglioside GM, solution for 90 minutes (ganglioside GM, was provided by Fidia Pharmaforschung, Munich, Germany). The wells were washed with double-distilled water and filled with the incubation solution for 90 minutes before use. The incubation solution contained 10 mmol/L phosphate buffer, pH 6.8, 2% bovine serum albumin, and 0.05% of the detergent Tween 20 (Serva, Heidelberg, Germany). One hundred microliters of the extracts from plasma, native urine, or redissolved feces were preincubated for 90 minutes with 400 p1 choleratoxin (P-subunit)-horseradish peroxidase conjugate (CT-HRP, List Biological Laboratories Inc., Campbell, Calif.). Seventy-five microliters of this solution was incubated for 90 minutes on the microtiter plate and then discarded. The wells were rinsed five times with incubation solution. To detect the CT-HRP bound to the wells, 100 p1 of the color reagent was placed in each well and incubated for 15 minutes. The color reagent was prepared by adding 100 p1 of 10 mg/ml 3,3' ,5 3'-tetramethylbenzidine (Sigma, Deisenhofen, Germany) in dimethylsulfoxide to 20 ml of 0.1 mol/L sodium acetate-citric acid buffer, pH 6.0; 1.3 p1 of 30% hydrogen peroxide was added immediately before use. The color development was stopped by the addition of 50 p1 2.5N H,SO,. Finally, the plates were read by use of a microtiter plate reader at 450 nm. Each extract was measured in triplicate. Unknown samples were quantified with pure ganglioside GM, external standards prepared daily in eight concentrations ranging from 0.037 to 2.35 mg/L. The unknown concentrations were read from the standard curve plotted on a semilogarithmic scale. Ganglioside GMl plasma levels were determined in two independent measurements. The values presented herein are means of both determinations. The quantification limit of the assay was about 0.030 mg/L. The intra-assay variance with the same extract was 3%, the inter-assay variance, including the extraction procedure, was 15%. Data analysis. The linear system approach was used to analyze ganglioside GMl plasma data." This method of data analysis describes the response of a

Kinetics @fianglwsideGM,

143

system, for example, the plasma concentration- time course, to a given input, such as an intravenous infusion, by the convolution of a system disposition function with an input function. A common disposition function used to describe pharmacokinetic data is the unit impulse response function: Cs (t)

=-

1 "

vc

C cie " " I=

I

in which C6(t) is the dose-normalized response of a system to an idealized intravenous bolus, V, is the volume of the sampling compartment, n the number of exponential terms, A, the i-th rate constant, and C, the fraction of the initial plasma concentration associated with the i-th exponential term. In the case of a biexponential function, C, = 1 - C,. An intravenous input of variable length can be described by a step function, which is a rectangular impulse of length equal to the duration of the input: P,(t)

=

[R,; 0

5

t 5 TI or [O; t

> TI

(2)

in which T is the step or infusion duration and R, is the dose rate D,/T of the k-th infusion with the k-th dose D. Convolution of the input P for unit dose rate, with the unit impulse response C, gives the unit step response C,:

Stopping an infusion at time T is equivalent to adding a negative input of size R,. Therefore the plasma concentration at time t after an infusion of dose rate R, and duration T was described by the following equations: Ck(t) = R,C,(t), for t

5

T

(4)

Equation 4 represents concentrations during infusion and equation 5 represents concentrations after infusion. In the case of intramuscular administration firstorder absorption was assumed:

in which D, is the k-th dose and k, the absorption rate constant. Complete absorption was assumed. Convolution of this input function with the unit impulse response gives the response to the k-th intramuscular injection:

CLIN PHARMACOL THER AUGUST 1991

144 Rost et al. Table I. Pretreatment endogenous levels of ganglioside GM, in plasma, urine, and feces

Plasma (mg1L) CV% Urine (mg/L) Feces (mgl24 hours)

Vol. I

Intravenous group Vol. 2

Vol. 5

Intramuscular group Vol. 3 Vol. 4

0.192 10.3 ND 19.2

0.11 1 8.2 ND 440.5

0.144 10.6 ND 42.6

0.111 7.7 ND 79.7

0.099 7.9 ND 346.1

Vol., Volunteer; ND, not detectable. Mean endogenous plasma concentrations and their coefficient of variation (CV%) as derived by the fitting procedure applied on the whole concentration course and mean values of two independent samples of urine and feces

The cumulative response to the consecutive inputs was calculated by the principle of ~ u ~ e r p o s i t i o n ' ~ :

in which the first moment of the unit impulse response17 was as follows:

in which C, is the endogenous baseline level and m the number of doses given at times T ~ Nonlinear . regression analysis was performed with the AR program of BMDP.19 Parameters were estimated by use of the iteratively reweighted least-squares method2' with the reciprocal of the squared true response values as weights; that is, the constant coefficient of variation model was assumed. The criterion of minimum area between the confidence intervals of the fitted curve2' was used to select the appropriate structural model:

Data are given as mean 2 SD or coefficient of variation (CV).

in which t is the appropriate Student value with N-P degrees of freedom, P is the number of parameters, N is the number of observations, and MSE is the mean square error of the fit delivered by BMDP-AR. The clearance (CL), volume of distribution at steady state (V,,), and the mean residence time of the sampling compartment (MRT,) and at steady state (MRTss) were calculated according to the following formulas17: CL

1 AUC,

= -

vc MRT --CL AUMC, MRTss = AUC, V,, = CL MRT,,

RESULTS The endogenous plasma concentration calculated as the means of three independent measurements in all five volunteers before ganglioside GM, administration was 0.132 mg/L (CV, 8.9%) This concentration laid well over the assay quantification level of 0.03 mg/L. In feces the mean amount excreted was 186 k 194 mgI24 hr (n = 5). No ganglioside GM, could be detected in urine at any time in any of the volunteers (Table I). Fig. 1 depicts the trough levels of all five volunteers over the whole kinetic time course. Levels of all volunteers ranged between 12 and 75 mg/L. After 6 days a first steady state was reached because of the half-life (t~,~a) of about 1 day. The tl,+x was the dominant kinetic phase, which accounts for 90% of the AUC,. Thereafter, steady-state concentrations increased only very slightly because of the minor influence of the Pphase. The mean trough concentrations of each volunteer were calculated from the 10 values between days 8 and 22 and amounted to 32.4 r 3.7, 54.1 k 5.2, and 40.5 r 5.1 mg/L in volunteers 1, 2, and 5 (intravenous administration), respectively, and to 27.8 2 1.7 and 34.3 k 4.0 mg/L in volunteers 3 and 4 (intramuscular administration). Plasma levels after intravenous treatment tended to be somewhat higher compared with the intramuscular levels. The peak levels during steady state immediately after intravenous administration were 55.2, 82.7, and 58.2 mg/L in volunteers 1 , 2, and 5, respectively. A biexponential response function was optimal for the whole kinetic time course on the basis of the crite-

VOLUME SO NJMBER 2

Kinetics fdanglwside GM,

Time

145

[dl

Fig. 1. Trough plasma levels during the 21-day intravenous (open symbols) and intramuscular (closed symbols) administration period and during a 21-day washout phase. Open circles, Volunteer 1; open triangles, volunteer 2; open squares, volunteer 5; solid triangles, volunteer 3; solid diamonds, volunteer 4.

Table 11. Kinetic parameters after multiple intravenous or intramuscular administration of 100 mg ganglioside GM, per day for 21 consecutive days Intravenous administration Vol. 1 Vol. 2 Vol. 5

Intramuscular administration Vol. 3 Vol. 4

t~/,u, half-life derived from the first rate constant, hl,; [I/#, half-life derived from the second rate constant, h2,; MRTc, mean residence time of the sampling compartment; MRTss, mean residence time at steady state; V,, volume of the the sampling compartment; V,,, volume of distribution at steady state; CL, clearance; tl/,k,, half-life of absorption.

rion described above21 (Fig. 2). The individual kinetic parameters are listed in Table 11. From this table mean values + SD for the system parameters V, and V,, are calculated to be 5.11 + 1.99 L and 6.98 + 3.57 L (n = 5), respectively. The clearance of 1.61 + 0.37 ml/min ( n = 5) was very low in all volunteers. The

mean residence times MRT, of 52.4 + 13.2 hours and MRTss of 69.2 -t 2 1.1 hours (n = 5) did not differ very much. Multiple dose ganglioside GM, kinetics were shown to be linear for both dosage forms because both of the 24-hour concentration-time courses on the first and last dosing day, as well as the interme-

CLIN PHARMACOI, THER AUGUST 1991

146 Rost e t al. V o l u n t e e r N o . 5.

100

V o l u n t e e r N o . 3.

loo

i .v.

7

i.m.

3

Fig. 2. Multiple dose plasma kinetics with intravenous (A) and intramuscular (B) administration of ganglioside GM,. Curves were fitted according to equation 8 with biexponential response functions (equation 3 and 7).

diate trough concentrations could be fitted with one parameter vector, that is, the superposition principle was valid. Because of the long terminal tl,,P the plasma levels measured after the washout phase on day 42 were still approximately three times higher than the initial endogenous concentrations. When the 24-hour kinetic concentration-time courses were taken on their own a monoexponential dispositon function was adequate (Fig. 3). This is explained by the fact that the relatively long terminal tlI2P of 3 to 5 days cannot be detected over such a

short time interval. In addition, peak plasma levels did not decline after the infusion but persisted or even increased up to 2 hours after termination. After intramuscular injection, peak levels were not attained until 12 hours after administration. Under ganglioside GM, treatment the amount excreted in feces varied between 10 and 250 mgl24 hours with a mean of 94 + 62 mgl24 hours (n = 5 ) . Therefore ganglioside GM, treatment seems to have no effect on feces levels, although such a conclusion is rather uncertain in light of the large variance. No

VOLUME SO NUMBER 2

Kinetics ofBanglwside GM, V o l u n t e e r No. 5.

I

.v

147

V o l u n t e e r No. 3. 1 . m . 100

i

Day 1

D a y 21

Day 21

._. 7

Fig. 3. Plasma kinetics after the first (day 1) and the last dose (day 21) of ganglioside GM,. A, Day 1, intravenous administration; B, Day 1, intramuscular administration; C, Day 21, intravenous administration; D, Day 21, intramuscular administration.

ganglioside GM, was detected in urine at any stage of the experiment. Clinical and laboratory examinations did not reveal any significant changes throughout the study. There were no signs of any general or local incompatibility that may have influenced the kinetic parameters.

DISCUSSION This is the first pharmacokinetic study in humans using the specific binding of the @-subunit of cholera toxin to ganglioside GM,. It was demonstrated that endogenous ganglioside GMl concentrations in plasma were low compared with the large amount excreted per day in the feces. Animal experiments have demonstrated that ganglioside GMl is extensively metabolized by the liver9 and is excreted almost exclusively in the form of metabolites in the urine and to a much lesser extent in the In light of this ev-

idence and due to the fact that there seemed to be no difference between levels in the feces before and after parenteral administration, the large amount of GM, excreted may be of dietary origin, or as a result of mucosal cells sloughed into the intestine.' The comparison of the pharmacokinetic profile of ganglioside GM, from this study with those of previous authors is complicated by different methods of measurement. Bellato et al. described monoexponential kinetics for rats, whereas other authors showed biphasic plasma concentration- time curves for the tritiated compound after a single d ~ s e . ~ , ' In ~ . this '~ study on humans, multiple-dose pharmacokinetics were best described by a biexponential function. However, because of the dominance of the a phase, the P phase of 3 to 5 days is virtually undetectable within the 24-hour kinetics at study days 1 and 21. Exogenously administered ganglioside GM, re-

'

CLIN PHARMACOL THER AUGUST 1991

148 Rostetal. mains confined to the blood volume and is poorly distributed throughout the body, as indicated by the small difference between V, and Vs,. We did not observe the 10-fold higher ganglioside GM, levels after intravenous compared with intramuscular administration reported for mice and rats.8'" The tl/, values of 2Y2 to 3 hours (intravenous) and 15 to 20 hours (intramuscular) in rats1' were considerably shorter than both ~I/,CY and t,/,P in our study. The marginal difference of the MRTss compared with the MRT, and the low clearance also seem to be in opposition to the results of Tettamanti et a1,13 who found that ganglioside GM, had a "dual metabolic fate" in mice, with a fast metabolism of about 80% in the liver and an uptake of the remaining ganglioside GM, into a different kinetic space with a slow turnover. This space probably represents the distribution of ganglioside GM, in muscle, kidney, spleen, lung, and adrenal^,^^""^ as well as the incorporation in neuronal t i s s ~ e s . ~With ~ , ~regard ~ to these animal data, one would have expected lower plasma levels of ganglioside GM, in humans because of a fast metabolism in the liver and a larger distribution at steady state. The observation of prolonged peak levels of up to 2 hours after intravenous administration (Fig. 3 , A ) probably reflects a distribution phase within the plasma compartment. Ganglioside GMl is an amphiphilic glycolipid forming micelles in aqueous sohtions even at concentrations below 10@ m o l / ~ , *that ~ is, in the range of endogenous levels. Because of its size and negative charge it could be attached to the vascular epithelium. Up to 80% of ganglioside GM, in human plasma forms stable complexes with albumin.I4 Both of these effects would certainly lead to a prolongation of the distribution of ganglioside GM, in the blood stream, and it is therefore conceivable that increasing concentrations may be observed after termination of the infusion. This plasma binding property of ganglioside GM, probably accounts for its slow clearance and perhaps also for the discrepancy discussed above with data from the study by Tettamanti et a1.13 To date only one other pharmacokinetic study of ganglioside GM, on humans has been published. That investigation focused on the intramuscular and subcutaneous route of administration in patients with Alzheimer's disease.25 With use of tritium-labeled ganglioside GM,, up to threefold higher plasma levels were obtained compared with those obtained in our study. The terminal tlI2P of about 100 hours and the time to reach apparent steady-state plasma levels after intra-

muscular administration were, however, in good accordance with our study. We thank our colleague Michael Looby, BSc, for stimulating discussion.

References 1. Wiegandt H. The gangliosides. In: Agranoff BW, Aprison MH, eds. Advances in neurochemistry, vol. 4. New York: Plenum Press, 1982:149-223. 2. Schengrund CL. The role(s) of gangliosides in neural differentiation and repair. A perspective. Brain Res Bull 1990;24:131-41. 3. Fishman PH. Gangliosides as cell surface receptors and transducers of biological signals. In: Ledeen RW, Hogan EL, Tettamanti G, Yates AJ, Yu RK, eds. New trends in ganglioside research: neurochemical and neuroregenerative aspects. Fidia Research Series, vol. 14, Padova, Italy: Liviana Press, 1988:183-201. 4. Karpiak SE, Li YS, Mahadik SP. Gangliosides (GM1 and AGF2) reduce mortality due to ischemia: protection of membrane function. Stroke 1987;18:184-7. 5. Massarotti M. Ganglioside therapy of peripheral neuropathies: a review of clinical literature. In: Tettamanti G, Ledeen RW, Sandhoff K, Nagai Y, Toffano G, eds. Gangliosides and neuronal plasticity. Fidia Research Series, vol. 6. Padova, Italy: Liviana Press, 1986:46579. 6. Argentino C, Sacchetti ML, Toni D, et al. GMI ganglioside therapy in acute ischemic stroke. Stroke 1989;20:1143-9. 7. Samson JC. GMl ganglioside treatment of central nervous system injury: clinical evidence for improved recovery. Drug Dev Res 1990;19:209-24. 8. Lang W. Pharmacokinetic studies with3H-labeled exogenous gangliosides injected intramuscularly into rats. In: Rapport MM, Gorio A, eds. Gangliosides in neurological and neuromuscular function, development and repair. New York: Raven Press, 1981:241-51. 9. Ghidoni R, Trinchera M, Venerando B, Fiorilli A, Sonnino S, Tettamanti G. Incorporation and metabolism of exogenous GM1 ganglioside in rat liver. Biochem J 1986;237:147-55. 10. Ghidoni R, Trinchera M, Venerando B, Fiorilli A, Tettamanti G. Metabolism of exogenous GMl and related glycolipids in the rat. In: Tettamanti G, Ledeen RW, Sandhoff K, Nagai Y, Toffano G, eds. Gangliosides and neuronal plasticity. Fidia Research Series, vol. 6. Padova, Italy: Liviana Press, 1986:183-200. 11. Bellato P, Milan F, Toffano G. Disposition of exogenous tritium-labeled GMl monosialoganglioside in the rat. Clin Tri J 1989;26:39-48. 12. Orlando P, Cocciante G, Ippolito G, Massari P, Roberti S, Tettamanti G. The fate of tritium labeled GM1 ganglioside injected in mice. Pharm Res Comm 1979;11:759-73.

VOLUME 50 NUMBER 2

13. Tettarnanti G, Venerando B, Roberti S, et al. The fate of exogenously administered brain gangliosides. In: Rapport MM, Gorio A, eds. Gangliosides in neurological and neuromuscular function, development and repair. New York: Raven Press, 1981:225-40. 14. Tornasi M, Roda LG, Ausiello C, et al. Interaction of GM1 ganglioside with bovine serum albumin. Formation and isolation of multiple complexes. Eur J Biochem l98O;ll l:315-24. 15. Fishrnan PH. Recent advances in identifying the functions of gangliosides. Chem Phys Lipids 1986;42:13751. 16. Kirschner G, Bassan M, Facco MP, Ferrari G, Callegaro L. Competitive binding assay for quantitative determination of GMI ganglioside in plasma and cerebrospinal fluid. J Pharm Biomed Anal [In press]. 17. Veng-Pederson P. Meantime parameters in pharmacokinetics. Definition, computation and clinical implications (part 11). Clin Pharmacokinet 1989;17:424-40. 18. Weber W, Nitz M, Looby M. Nonlinear kinetics of the thiamine cation in humans: saturation of nonrenal clearance and tubular reabsorption. J Pharmacokin Biopharm 1990;18:501-23. 19. Dixon WJ, ed. BMDP Biomedical computer programs. Berkeley, California: University of California Press, 1987.

20. Sheiner LB. Analysis of pharmacokinetic data using parametric models. 11. Point estimates of an individual's parameters. J Pharrnacokinet Biopharm l985;13:5 15-40, 21. Imbirnbo BP, Imbimbo E, Daniotti S, Verotta D, Bassotti G. A new criterion for selection of pharmacokinetic multiexponential equations. J Pharm Sci 1988;77:784-9. 22. Guzman-Harty M, Warner JK, Mancini ME, Pearl DK, Yates AJ. Effect of crush lesion on radiolabelling of ganglioside in rat peripheral nerve. J Neurochem 1988;50:237-42. 23. Lang W, Stotzem CD. Investigations on the incorporation of a tritiated ganglioside, GM1, in the injured and intact sciatic nerves of the rat. J Neurochem 1989;53:929-34. 24. Corti M, Canth L, Sonnino S, Tettamanti G. Aggregation properties of gangliosides in aqueous solutions. In: Ledeen RW, Hogan EL, Tettamanti G, Yates AJ, Yu RK, eds. New trends in ganglioside research: neurochemical and neuroregenerative aspects. Fidia Research Series, vol. 14. Padova, Italy: Liviana Press, 1988:7991. 25. Svennerholm L, Gottfries CG, Blennow K, et al. Parenteral administration of GMl ganglioside to presenile Alzheimer patients. Acta Neurol Scand 1990;8 1:48-53.