The functional effects of suppression of hypothermia-induced cell swelling in liver preservation by cold storage

CRYOBIOLOGY

28, 150-158 (1991)

The Functional Effects of Suppression of Hypothermia-Induced Swelling in Liver Preservation by Cold Storage

Cell

RALF SUNDBERG, AAMER AR’RAJAB, BO AHREN, AND STIG BENGMARK Department of Surgery, Lund University, S-221 85 Lund, Sweden It is known that cellular edema and functional impairment develop during anaerobic cold storage of organs. The extent of both is related to the storage time and the composition of the preservation solution used. We studied hypothermia-induced cell swelling and its effect on liver function after cold storage preservation with either Eurocollins (EC), a number of modified EC solutions in which glucose was replaced by various concentrations of raRinose, or UW solution. After 24 h storage, tissue swelling as determined by total tissue water (TfW) in rat liver tissue slices was most pronounced in slices incubated in Eurocollins, whereas the TTW was only moderately increased in slices stored in modified Eurocollins containing 90 to 120 mM raftinose. In contrast, slices incubated in UW solution had a TTW equal to normal rat liver tissue. Furthermore, intact rabbit livers preserved with Eurocollins had an increase in the whole organ weight, while there was no weight change after preservation with the modified solution containing 120mM rafIinose (M120). In contrast, a pronounced weight loss was observed after preservation with UW solution. After cold storage, the livers were reperfused for 2 h at 38°C in an isolated perfusion circuit (IPL) with an acellular perfusate. Bile flow was significantly greater in livers preserved in Ml20 than in those preserved with the conventional Eurocollins. However, the bile flow in the livers stored in Ml20 was inferior to that in the livers preserved with UW solution, which in turn was equal to that in control livers. The release of alanineaspartate-aminotransferase into the perfusate was higher in livers preserved with Eurocollins, with or without modification, than in the livers preserved with UW solution. It is concluded that although the replacement of the glucose in Eurocollins with ratEnose did indeed prevent hypothermia-induced cell swelling as evidenced by the absence of weight changes in the intact rabbit liver, these livers were still not as well preserved as those stored in UW. Therefore, the mechanism of action of UW is not solely based upon suppression of cell swelling nor upon the absence of glucose in this solution, but upon other factors which are discussed in this paper. o test Academic PESS, II-K.

For a long time, the liver has been regarded as a difficult organ to preserve by simple cold storage. While Collins solution, in the original study, allowed 30 h cold storage of dog kidneys (5), it has permitted only up to 18 h cold storage of the dog liver (2). This solution and its more recent derivative Eurocollins (8) have for more than a decade been the standard solutions for clinical liver preservation, and storage periods longer than 6-8 h have not been regarded as safe (1, 2). However, the recent demonstration of dog liver preservation for up to 48 h with UW solution (14), followed by successful clinical liver preservation for at least 24 h (17), has made it apparent that the Colhns-

Received October 19, 1989; accepted May 2, 1990

type solutions are not so well suited for preservation of the liver as for other organs. Based on theoretical considerations, a number of advantages of the UW solution have been put forward (1). Among these, two have been stressed as perhaps the most important. Both relate to glucose, which is the main impermeant that counteracts hypothermia-induced cell swelling in Collinstype solutions. First, glucose being relatively impermeable to kidney cells is permeable to liver cells during hypothermia (3) and is thus ineffective for the control of liver cell volume. Second, the glucose in Collins solution by entering the cell may cause intracellular acidosis through anaerobic glycolysis, and due to organ specific metabolic differences, this could be more

150 OOll-2240/91 $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

LIVER

FUNCTION

AFTER

pronounced in the liver than in the kidney (21, 27). UW solution circumvents these problems since it does not contain glucose, but instead the lactobionate anion and the trisaccharide raftinose both of which are impermeable to the liver cell (1, 28). In the present study, we have attempted to improve Eurocollins solution by replacing the glucose with raffinose. We compared this modified solution with Eurocollins and UW solution by testing the ability of these solutions to counteract hypothermia-induced cell swelling, using a tissue slice method, and by studying weight changes of whole organs. In addition, the function of livers preserved in these solutions was evaluated during isolated perfused liver (IPL) reperfusion. MATERIALS

AND

METHODS

Animals Male Wistar rats, weighing 300-350 g, were purchased from Anticimex (Stockholm, Sweden). Swedish country breed rabbits of both sexes, weighing 2-3 kg, were purchased from local vendors. The animals had free access to water and food pellets and were not fasted prior to the experiments. Preservation Solutions The composition of the solutions used is shown in Table 1. UW solution was gener-

COLD

151

STORAGE

ously donated by DuPont Critical Care (Waukegan IL). Perfadex was purchased from KabiBaxter (Stockholm, Sweden). Eurocollins solution and the modifications thereof, containing 120 mM glucose, or in which glucose was gradually replaced with 30,60,90, or 120 mM rafftnose (M30, M60, M90, and M120, respectively), were made with chemicals purchased from Sigma and Aldridge. All solutions were stored at 4°C until used. Rat Liver Tissue Slice Experiments Three rats were anesthetized with ether (Casco Nobel Industrifarg, Stockholm, Sweden), one for each of the duplicate experiments. The abdomen was opened through a midline incision, and the liver was rapidly excised and placed in a chilled bowl. One-millimeter-thick slices were cut by hand using a razor blade. Six slices from each rat were put in open glass beakers containing 20 ml of each of the solutions described in Table 1 and also in normal saline. They were then incubated at 0°C for 24 h, without stirring or shaking. All slices were then gently blotted on a filter paper in a standardized fashion, weighed “wet,” oven dried at 7O”C, and then weighed again. Total tissue water (TTW) was calculated by the formula (wet weight - dry weight)/dry weight, as described by Little (19). TTW was also determined in fresh rat liver tissue in the same way.

TABLE 1 Composition of Solutions, mmoliliter

Na K Cl pi

HCO, Glucose Raffinose Lactobionate

Eurocollins

MO

M30

M60

M90

Ml20

UW”

10 115 15 100 10 195 0 0

10 115 15 100 10 120 0 0

10 115 15 100 10 90 30 0

10 115 15 loo 10 60 60 0

10 115 15 100 10 30 90 0

10 115 15 100 10 0 120 0

30 120 0 25 0 0 30 100

LIUW solution also contains 3 m&4glutathione, 5 rnA4adenosine, 1 mA4allopurinol, 5 mM MgSO, 5% hydroxyethyl starch, 100 IU/liter insulin, and 0.5 mliliter Bactrim.

152

SUNDBERG

Rabbit Liver Experiments Organ harvesting. Twenty-four rabbits were anesthetized with pentobarbital (25 mg/kg). Following iv administration of 500 IU of heparin, the livers were rapidly extirpated according to a standard technique (25), with an in situ flush through the portal vein with 100-150 ml of chilled (4°C) Perfadex, Eurocollins, M120, or UW solution (experimental groups 1-4, respectively). There were 6 livers in each group. The bile duct was cannulated with a plastic catheter for collection of bile and a cholecystectomy and ligation of the cystic duct were performed. The livers were then stored on ice in 100-150 ml of the respective solutions for 24 h except the controls (Group 1) which were reperfused immediately. The livers were weighed before and immediately after the storage period, and the percentage of weight change was calculated. Isolated perfusion. The isolated perfusion circuit was similar to the one previously described (24). In short, the livers were flushed with 100 ml Perfadex at room temperature to remove the cold storage solution and to render the preserved livers approximately the same temperature as controls. The livers were then floated in a plastic chamber and reperfused with a Watson Marlow Model 502s rollerpump through the portal vein at 38”C, at a flow rate of 3 ml/g/min. The perfusate was a KrebsHenseleit bicarbonate solution (18), with 30 g/liter of dextran 60 for oncotic support, 10 mM glucose, and 50 mg/liter of penicillin G and streptomycin. Dextran 60 was a generous gift from Pharmacia (Uppsala Sweden). The reperfusion medium was gassed with a mixture of 0, and CO, through a membrane oxygenator to give a pH of 7.4 f 0.2. Samples of the perfusate were taken at 30, 60, and 120 min for analysis of alanineaspartate-aminotransferase (ASAT), which was performed using a routine clinical method (26). Bile was collected from the cannulated common duct.

ET AL.

Statistics Results were calculated as mean values ? standard deviation. Statistical comparisons were performed with the KruskalWallis nonparametric analysis of variance. If a difference was found among the groups, this was further analyzed with the Wilcoxon nonparametric test for unpaired data. A P value less than 0.05 was considered significant. RESULTS

Rat Liver Tissue Slice Experiment As seen in Fig. 1, the TTW increased markedly in the slices incubated in saline to values of 4.54 + 0.20 g water/g dry weight (mean + SD), as compared to 2.19 + 0.08 in fresh control tissue (P < 0.01). The TTW was moderately increased in the slices incubated in Eurocollins (3.28 + 0.22), the increase being significantly lower than for those preserved in saline (P < 0.01). In the modified Eurocollins solutions where 120 mm glucose was gradually replaced by raffmose (MO, M30, M60, M90, and M120), TTW decreased linearly with increasing concentrations of raffinose (3.85 2 0.28, 3.51 ? 0.20, 2.99 + 0.20, 2.79 ? 0.21, 2.65 2 0.29, respectively, P < 0.01 between each step in concentration, except for the last one; NS). The TTW of slices incubated in MO was significantly higher than that in EC (P < O.Ol), whereas those stored in M30 were not significantly different from those preserved in EC. In contrast, the TTW of the M60, M90, and Ml20 groups was significantly lower than that in EC (P < 0.01). The total tissue water of slices incubated in UW solution was lower than in the other groups of preserved slices (2.24 2 0.14, P < 0.01) and did not differ significantly from the water content of fresh control tissue. Isolated Perfused Rabbit Liver Experiments The weight changes in whole rabbit livers cold stored for 24 h in Eurocollins, M120,

LIVER

FUNCTION

AFTER

COLD

STORAGE

153

5” E .P $ F u UI $ 5 L 0

432l0

Mean, S.D. 1. Changes in TTW in 24-h cold-stored rat liver tissue slices. Eighteen slices from three rat livers were used in each group. The bars indicate means f SD. FIG.

and UW solution are shown in Fig. 2. Livers stored in Eurocollins gained 10.1 +3.4% (mean ? SD) of the pre-preservation weight. The weight was essentially unchanged in the livers preserved in M120, - 1.1 + lS%, whereas there was a marked percentage weight loss of 15.4 + 4.7% for the livers stored in UW solution. All differences between the groups were significant (P < 0.01).

FIG,

All the livers produced bile after normothermic reperfusion. Bile production commenced within 15 min and was constant throughout the experiments. The amount of bile produced during 2 h per 100 g liver tissue is presented in Fig. 3. The bile production of the livers preserved for 24 h with UW solution did not differ from controls but was significantly better than for the other two groups preserved in Ml20 or Eu-

t4eanfS.D. 2. Weight changes during 24 h cold storage of rabbit livers. The bars indicate means k SD.

154

SUNDBERG

ET AL.

Controls Eurocollins

t-

20 M

3 w. fi

.

10

z

0 Mean* SD. FIG. 3. Bile production during 2 h IPL reperfusion of 24-h cold-stored rabbit livers. The bars

indicate means k SD.

rocollins (Z’ < 0.01). The livers stored in Eurocollins had a mean bile production which was 33% of the mean value for controls, whereas the mean for the livers stored in Ml20 was 52% of that for controls. This difference was significant at the P < 0.01 level. The amount of ASAT released into the perfusate at 30, 60, and 120 min after normothermic reperfusion is shown in Fig. 4. It

g

20 :’ 18 1416 1

H

is seen that there was an increasing enzyme release during the reperfusion period in all groups of preserved livers. There was no difference between the livers stored in Eurocollins and M120. However, the ASAT levels at 120 min were significantly higher in the livers stored in Eurocollins and Ml20 than in those stored in UW solution and controls (P < 0.01). Also, the livers preserved in UW solution released more en-

Controls Eurocollins Ml 20

q

q w

“, 12: 10: v1 z 8: ‘;;i 62. 4204 30

60

120

Reperfusion time, minutes FIG. 4. ASAT release during 2 h IPL reperfusion of 24-h cold-stored rabbit livers. The bars indicate means k SD.

LIVER

FUNCTION

AFTER

COLD

STORAGE

155

in this respect. However, they were not as effective as UW solution in preventing tissue swelling. In fact, the TTW of slices inDISCUSSION cubated in UW solution was not different The use of hypothermia remains the main from the control tissue water content. principle in organ preservation. Hypother- However, this is an expected finding, since mia reduces the rate of cellular metabolism, the basic composition of UW solution was and thus delays the occurrence of cell originally based upon the results of tissue death. However, one well-documented un- slice experiments (28). UW solution condesired side effect of hypothermia is cell tains 130 mM of substances, namely lactoswelling. This occurs due to depression of bionate (100 mM) and rafIinose (30 mM), the activity of membrane bound Na/ which are included because of their cell K-dependent ATPase, loss of membrane membrane impermeability. The difference demonstrated in tissue hydration following potential, and the presence of intracellular impermeable anions (20, 22). Since such storage in UW solution or Ml20 could posswelling may lead to cell lysis, and interfere sibly be explained by the difference in imwith organ reperfusion, it is an aim in organ permeant content (10 mM) or by an addipreservation to minimize its occurrence. tional osmotic effect due to the colloid This can be achieved by including cell present in UW solution. Whereas UW solution seems to have an membrane impermeable molecules in the cold storage solutions (1, 7, 12, 22, 28). It appropriate composition to maintain cell has been postulated that a concentration of volume during cold storage in the tissue 110-140 mJ4 of such substances would bal- slice model, we subsequently found, that in ance the osmotic force of the cell mem- the intact organ, it causes a dehydration of brane impermeable intracellular anions, the tissue. This is shown by the remarkable and thereby prevent the development of hy- decrease in whole organ weight after 24 h of cold storage in this solution. This observapothermia-induced cell swelling (1). In this study, we have attempted to cor- tion is further supported by a concomitant relate cell swelling during hypothermia to electron microscopy study, in which we oborgan function after reperfusion in an IPL served shrinking of hepatocytes and dilamodel. First, we replaced the glucose in tion of the sinusoidal space in rat livers cold Eurocollins solution with the trisaccharide stored for 24 h in UW solution (23). In the raffinose at various concentrations in order intact rabbit liver, the solution containing to find the optimal amount of impermeants. 120 mM raftinose (M120), which was choFor this purpose, a tissue slice method was sen for further study in the IPL model, was chosen. Since it was our experience from not associated with any significant weight change and therefore appeared to adepreliminary experiments that it was difficult to dissolve more than 120 mmol raftinose quately maintain cell volume during cold per liter, and since this impermeant concen- storage. In this respect, the Ml20 solution tration comes close to that of UW solution, was superior to the UW solution. The difwe chose to study modified EC solutions ference observed between the tissue slice containing 120 mM glucose or raflinose at model and the weight changes in whole rabvarious concentrations. Of the raffinose bit livers demonstrates the weakness of the concentrations studied, we found that 90 former technique. The slices are incubated and 120 mM were the most effective in pre- in a volume of fluid per unit tissue mass venting the swelling of rat liver tissue relatively much greater than the volume in slices. These raffinose-supplemented solu- which whole organs are preserved. Since tions were superior to Eurocollins solution the slices are thin, equilibration with the

zyme than the controls at all time intervals (P c 0.05).

156

SUNDBERG

preservation medium will take place easily, and therefore the volume changes will be greater than in the intact organ. Also, in the whole organ, the effect of colloid may become more apparent, since the oncotic pressure will block the entrance of water into the tissue from the preservation medium bathing the organ across the organ capsule or from the infused preservation medium across the endothelial lining. Taken together, the measurement of weight changes in whole rabbit livers during cold storage is a more reliable technique, demonstrating that hypothermia-induced cell swelling can be effectively prevented by replacing the glucose in Eurocollins solution with raffinose. In the IPL model, two viability parameters were studied, namely bile flow and ASAT release into the pet&sate. Of these two, previous studies have shown that bile flow is the most sensitive and discriminative for evaluating preservation-induced injury, since it correlates with the preservation time in cold storage or perfusion preservation (15, 25), it discriminates between different preservation media (13, 15), and it has also been valuable for studying the effect of pharmacological donor pretreatment (24,25). In the present study, improved bile production was seen when the glucose in Eurocollins was replaced with 120 mM raflinose. A possible explanation could be that omitting the glucose would result in less tissue acidosis during cold storage of the liver, as suggested by Belzer and Southard (1). However, this hypothesis is contradicted by the recent findings of Fuller and Buzra, who by measuring intracellular pH with 31PNMR spectroscopy during preservation of rat livers, could not demonstrate any difference between livers cold stored in solutions with or without glucose (10). It is therefore more likely that the increase in bile flow after replacing glucose with raffinose in Eurocollins is due to improved cell volume control only. The improvement in liver function demonstrated with this mod-

ET AL.

ification of Eurocollins does not, however, reach the level of function found in livers preserved with UW solution. The present study thus clearly demonstrates that livers preserved in UW solution become dehydrated. In contrast, livers preserved in Ml20 maintained their weight. Although there is a possibility that dehydration could improve reperfusion of the organ, it seems unlikely that a reduction in cell volume below normal during preservation would be advantageous in itself. Since dehydration increases the concentration of the intracellular enzymes and their substrates, it will, at least from a theoretical standpoint, increase the rate of metabolism. If so, this could aggravate the injury due to ischemia. Therefore, it seems more reasonable that the maintenance of normal tissue hydration, as with M120, would be more beneficial. However, despite this, UW solution preserved liver function well. Therefore, it seems more likely that the superiority of UW solution is not solely due to its efficiency in preventing hypothermiainduced swelling, nor due to the absence of glucose. It has previously been shown in an IPL model, that bile production as good as control values can be obtained after preservation with a simplified version of UW solution, containing only lactobionate, raffinose, phosphate, and magnesium sulfate, and that the release of liver enzymes is decreased if glutathione is also included (13). Of these components, Ml20 lacks lactobionate, magnesium sulfate, and glutathione. Glutathione is obviously an important ingredient. For example, it has been shown that the contractile force of hearts preserved in UW solution decreased if glutathione was omitted (29). The role of magnesium in UW solution has not been studied and is as yet unclear. Collins demonstrated that a functional impairment occurred in kidneys stored in Collins solution if magnesium sulfate was omitted (6). However, in another study the elimination of magnesium from Collins solution had no

LIVER

FUNCTION

AFTER

effect (16), and Eurocollins, which contains no magnesium, has been shown to be an effective preservation medium for kidneys (2). Another interesting ingredient in UW solution is the lactobionate molecule itself. It has been shown previously, that lactobionate in UW solution cannot be replaced by another anionic organic impermeant , gluconate, without a reduction in viability (13). Lactobionate has been observed to be a strong chelator of calcium and iron (J. H. Southard, personal communication). This feature may be of importance in minimizing cell injury due to calcium inflow (4, 9) and free radical formation (11). In view of the inferior functional performance, despite adequate cell volume control in livers preserved with M120, which lacks these ingredients, compared to that of the livers preserved with UW solution, further studies concerning the role of these agents in organ preservation fluids are necessary. REFERENCES

1. Belzer, F. O., and Southard, J. H. Principles of organ preservation by cold storage. Transplanration 45, 673-676 (1988). 2. Benichou, J., Halgrimson, C. G., Weil, R., III, Koep, L. J., and Starzl, T. E. Canine and human liver preservation for 6 to 18 hr by cold infusion. Transplantation 24, 407411 (1977). 3. Cahill, G. F., Jr., Ashmore, J., Scott Earle, A., and Zotta, S. Glucose penetration into liver. Amer. J. Physiol. 192, 491496 (1958). 4. Cheung, J. Y., Bonventre, J. V., Malis, C. D., and Leaf, A. Calcium and ischemic injury. N. Engl. J. Med. 314, 1670-1676 (1986). M. B., 5. Collins, G. H., Bravo-Shugarman, Terasaki, P. Kidney preservation for transportation: Initial perfusion and 30 hours ice storage. Lancer 6, 1219-1222(1%9). 6. Collins, G. M., Hartley, L. C., and Clunie, G. J. A. Kidney preservation for transportation. Experimental analysis of optimal perfusate composition. &it. J. Surg. 59, 187-189 (1972). 7. Downes, G. L., Hoffman, R. M., Huang, J., and Belzer, F. 0. Mechanism of action of washout solutions for kidney preservation. Transplunfarion 16, 46-53 (1973). 8. Dreikom, K., Horsch, R., and Rohl, L. 48- to 96 hour preservation of canine kidneys by initial

COLD

157

STORAGE

perfusion and hypothermic storage using the Euro-Collins solution. Eur. Ural. 6, 221-224 (1980). 9. Farber, J. L. Calcium and the mechanisms of liver necrosis. In “Progress in Liver Diseases,” (H. Popper and F. Schaffner, Eds.) Vol. 7, Chap. 20. Grune & Stratton, NY, 1982. 10. Fuller, B. J., Buzra, A. L., Proctor, E., Myles, M., Gadian, D. G., and Hobbs, K. E. F. Control of pH during hypothermic liver storage: Role of storage solution. Transplantation 45, 239-241 (1988). 11. Green, C. J., Healing, G., Simpkin, S., Lunec, J., and Fuller, B. J. Desferrioxamine reduces susceptibility to lipid peroxidation in rabbit kidneys subjected to warm ischemia and reperfusion. Camp. Biochem. Physiol. 85, 113-117 (1986). 12. Jamart, J., and Lambotte, L. Differential effect of swelling and anoxia on kidney function and its consequences on the mechanism of action of intracellular organ preservation solutions. Transplantation 34, 176-182 (1982). 13. Jamieson, N. V., Lindell, S., Sundberg, R., Southard, J. H., and Belzer, F. 0. An analysis of the components in UW solution using the isolated perfused rabbit liver. Transplantation 46, 512-516 (1988). 14. Jamieson, N. V., Sundberg, R., Lindell, S. et al. Preservation of the canine liver for 24-48 hours using simple cold storage with UW solution, Transplantation 46, 517-522 (1988). 15. Jamieson, N. V., Sundberg, R., Lindell, S., Southard, J. H., and Belzer, F. 0. A compatison of cold storage solutions for hepatic preservation using the isolated perfused rabbit liver. Cryobiology 25, 30&3 10 (1988). 16. Jensen, E. H., and Kemp, E. Kidney preservation III. The importance of the composition of perfusion fluids in the transplantation of rabbit kidneys. &and. J. Ural. Nephrol. 6, 284-(1972). 17. Kalayglou, M., Sollinger, H. W., Stratta, R. J., et al. Extended preservation of the liver for clinical transplantation. Lance? 19, 617-619 (1988). 18. Krebs, H. A., and Henseleit, K. Untersuchungen under die Harnstoffbildung in Tier Korper. Hoppe-Seylers

Z. Physiol.

Chem. 21, 33-66

(1932). 19. Little, J. R. Determination of water and electrolyte in tissue slices. Ann. Biochem. 7, 87-95 (1964). 20. Macknight, A. D. C., and Leaf, A. Regulation of cellular volume. Physiol. Rev. 57, 510-573 (1977). 21. Newsholme, E. A., and Start, C. “Regulation in metabolism,” pp. 247-291. Wiley, London, 1974.

158

SUNDBERG

22. Pegg, D. E. Organ preservation. Surg. Clin. N. Amer. 66, 616-632 (1986). 23. Sundberg, R., Alumets, J., Ar’Rajab, A., Ahren, B., and Bengmark, S. Ultrastructure of liver preservation with Eurocollins and UW solution. Transplantation Proc., in press. 24. Sundberg, R., Ar’Rajab, A., Ahren, B., and Bengmark, S. Chlorpromazine donor pretreatment improves liver preservation quality with UW solution in an experimental model. Transplantation, 48, 742-744 (1989). 25. Sundberg, R., Lindell, S., Jamieson, N. V., Southard, J. H., and Belzer, F. 0. Effects of chlorpromazine and methylprednisolone on perfusion preservation of rabbit livers. Cryobiology 25, 417-424 (1988).

ET AL.

26. The Commute on Enzymes of the Scandinavian Society or Clinical Chemistry and Clinical Physiology. Recommended methods for determination of enzymes in blood. Stand. J. Clin. Lab. Znv. 33, 292-306 (1974). 27. Vesell, E. S. Significance of the heterogenity of lactic dehydrogenase activity in human tissues. Ann. N.Y. Acad. Sci. 94, 912-931 (1963). 28. Wahlberg, J. A., Southard, J. H., and Belzer, F. 0. Development of a cold storage solution for pancreas preservation. Cryobiology 23,477482 (1986). 29. Wicomb, W. N., and Collins, G. M. 24-hour rabbit heart storage with UW solution: Effects of low-flow perfusion, colloid and shelf storage. Transplantation

48, 6-9 (1989).