Magneric Resonance Imaging, Vol. 9, pp. 429-434. 1991 0730-725X/91 $3.00 + .Ml

Printed in the USA. All rights reserved. Copyright Q 1991 Pergamon Press plc

0 Original Contribution

IN VIVO PROTON MAGNETIC RESONANCE IMAGING AND LOCALIZED SPECTROSCOPIC ANALYSIS OF POLYCYSTIC

KIDNEY DISEASE IN DBA/2FG-pcy MICE

RHEAL A. TOWNER,* TAMIO YmmmxI,t DIANA J. PHILBRICK,? BRUCE J. HOLUB,~

EDWARD G. JANZEN,*$ AND HISAHIDE TAKAHASHI~

*MRI Facility and Departments of iNutritional Sciences and $Clinical Studies, University of Guelph, Guelph, Ontario, Canada NlG 2W1, SLaboratory Animal Centre, Fujita-Gakuen Health University,

Toyoake, AICHI, Japan

Proton magnetic resonance imaging (MRI) and localized spectroscopy techniques were used to study polycystic kidney disease (PKD) in DBA/2FGpcy (pcy) mice, which are an animal model for the adult form of human PKD. A volume selective spectroscopy (VOSY) method was used to obtain localized proton spectra as well as appar- ent T1 and Tz relaxation times of the kidneys of either pcy or DBA/ZFG normal (DBA) mice. Localization of the proton spectra was based on spatial coordinates determined from the proton images. Increases in apparent kid- ney T1 and T2 relaxation times were observed for pcy mice in comparison to those measured in DBA mice. Lo- calized kidney Tl values from pcy mice were found to range from 799 to 1395 msec whereas DBA mice kidney T, values were 617 to 774 msec. Localized T2 values measured from pcy mice kidneys ranged from 90 to 172 msec in comparison to kidney T2 values of 50 to 72 msec for DBA mice. The onset of cyst formation in the kidneys of pcy mice as a function of age was also studied. As early as 4 to 5 weeks of age it was possible to detect in vivo changes in pcy mice kidney T, and T2 values.

Keywords: (VOSY).

‘H-MRI; Polycystic kidney disease (PKD); DBA/ZFG-pcy mice: Volume-selective spectroscopy

INTRODUCTION

Polycystic kidney disease (PKD) in humans is the most common structural abnormality affecting the kidney. Two basic forms of PKD are known. The rare child- hood PKD (CPKD) is an autosomal recessive disorder which usually causes renal failure within the first de- cade of life, whereas the more common adult form of PKD (ADPKD) is an autosomal dominant disorder that frequently remains asymptomatic until the fourth or fifth decades of life. ‘2’ Diagnosis of ADPCK is usually confirmed by computer tomography although there are recent reports that MRI can be used as an ef- fective supplement to the CT findings.3 Curiously, values from routine urinalyses tests often remain nor- mal until the onset of renal failure in these patients.’ Recessive forms of polycystic disease are also found in

adulthood. The DBA/2 FG-pcy (pcy) mouse has been developed as an animal model to study human PKD. The disease in pcy mice appears to resemble human ADPKD in clinical course and morphology, although inheritance is an autosomal recessive trait.4 Kidney weights are normal in the affected animals at birth, but by 25 weeks of age these organs consist of numer- ous, fluid-filled cysts and occupy almost the entire ab- dominal cavity. Death from uremia usually occurs by 8 months of age. In this mouse model of PKD, the cysts are thought to originate in the tubules, as in the case in the human form of the disease.4 Light micros- copy studies by Takahashi et al. have shown that the onset of the cystic changes may occur before birth,4 but there are no reports about the progression of cyst development in vivo. In this study, proton images and apparent kidney water proton T, and T2 determina-

RECEIVED 1 l/6/90; ACCEPTED 2/12/91. Acknowledgments-Financial assistance was provided by

the University of Guelph MRI Facility, the Natural Sciences and Engineering Research Council, and Fujita-Gakuen Health University. We thank Dr. Uwe Oehler, MRI Facility, for use of the in-house computer program.

This is contribution #007 supported by funds from the University of Guelph MRI Facility.

Address all correspondence to Dr. Rhea1 A. Towner, MRI Facility, Room 1601, Ontario Veterinary College, Uni- versity of Guelph, Guelph, Ontario, Canada NlG 2Wl.

429

430 Magnetic Resonance Imaging 0 Volume 9, Number 3, 1991

tions, obtained from a localized proton spectroscopy (VOSY) technique, were used to assess the effects of this cystic development and to monitor the severity of the cystic disease in pcy mice ranging from 10 to 192 days of age.

Kidneys have not been studied by MRI as exten- sively as other tissues due to the major problem of res- piratory motion artifacts. In spite of the considerable motion in the abdomen there are some previous in vivo studies that have been carried out on the kidneys. Leung et al. used MRI to differentiate a wide spec- trum of renal diseases in human patients, such as glomerulonephritis, acute tubular necrosis, hydrone- phrosis, perirenal hematoma, renal tumours and poly- cystic kidneys.’ Studies by Iaina et al. incorporated MRI to study experimentally induced (e.g., gentami- tin) acute renal failure in rats.6p7 Levine et al. used MRI and computer assisted tomography (CAT) com- paratively to study hemorrhagic renal cysts of human patients with autosomal dominant polycystic kidney disease.3 Magnetic resonance imaging and spectros- copy (3’P and ‘H) were used by Terrier et al. to in- vestigate acute renal ischemia in rats.8 Abdominal respiratory gating or triggering was implemented in this study to enhance the image quality of the kidneys.

EXPERIMENTAL

The abdominal region of an anaesthetized mouse (Halothane; l%-1.5%) was placed on a respiratory gating apparatus consisting of a small balloon filled with air connected to a length of Tygon tubing. The inspiration of the mouse caused an increase in pres- sure within the balloon which was transferred via an increase in air pressure to a Honeywell MicroSwitch (142 PCOlD) pressure transducer. The change in pres- sure during inspiration was used to trigger acquisition of the phase-encoding steps in the imaging sequence. Further details on the respiratory gating device were previously discussed in Brauer et a1.9

NMR measurements were made with the use of a SIS 2.0 Tesla/31 cm bore imaging system (Spectros- copy Imaging Systems Corp., Fremont, California, USA). Multiple proton image slices were taken in the transverse and coronal planes using an echo time (TE) of 30 msec and a delay time (TR) of 1.0 sec. The im- age slices were 4.7 mm thick, field of view was 6 x 8 cm’, with 128 phase-encoding steps, 4 acquisitions per step, and 256 frequency encoding points.

Localized proton spectra were taken from 4.7 x

4.7 x 4.7 mm3 voxels (0.25 G/cm gradient strength) within a kidney based on spatial coordinates deter- mined from the images. The VOSY (volume selective spectroscopy) method, lo based on stimulated echo methods,“*r2 was used for spatial localization of the

proton spectra. The sequence uses three selective 90” sine pulses to excite a stimulated echo from the region intersecting the slice planes. A 20-msec delay was used after the first and third 90” pulses, and a 65msec de- lay was used after the second 90” pulse. The number of transients used for each spectrum was eight. A tim- ing diagram for the VOSY sequence is illustrated in Fig. 1. Increasing the stimulated echo delay (d3) in- ‘creases the amount of T, relaxation experienced by the final signal. Increasing the spin-echo delay (d2) in- creases the amount of T2 relaxation experienced, by the final signal. An in-house computer program was used to determine localized volumes of the kidneys in the images.

Localized T, and T2 determinations were obtained for proton resonances within the voxels by a modifi- cation of the VOSY method.9s’0 For Tl experiments, the delay after the second 90” pulse was incremented from 30 to 230 msec in 40-msec increments, while the delay after the first and third 90” pulses was kept con- stant at 20 msec. For T2 experiments, the delay after the first and third 90” pulses was incremented from 30 to 55 msec in 5-msec increments, while the middle delay was kept constant at 65 msec. T, and T, deter- minations were calculated via nonlinear regression analysis.

RESULTS AND DISCUSSION

A series of coronal and transverse images of DBA/ 2FG (DBA) and pcy mice were obtained (Figs. 2, 3). In comparison to normal DBA mice, the kidneys of the pcy mice were dramatically enlarged and had in- creased proton signal intensities. To establish whether the increase in proton signal intensity in the images of pcy mouse kidneys was due to an increase in the water resonance or fat (methylene protons of fatty acyl moieties) an in vivo localized proton spectroscopy method (VOSY) was employed. Localized proton spec-

Echo I

I

P gP FSY gz gxr gYr

Gradient I

Tiiing 4--d2--H- - ---w d3 d2 I Fig. 1. Timing diagram for the VOSY sequence.

In vivo ‘H-MRI of polycystic kidney disease in pcy mice 0 R.A. TOWNER ET AL. 431

Fig. 3. Transverse ‘H-NMR images of DBA/2FG and DBA/2FGpcy mice. The (1) kidneys are clearly visible in the middle (left and right) of each image. Other notable anatom- ical features include the (2) spinal cord (intense region) at the top of each image, the (3) back muscles (dark grey region) surrounding the spinal cord, and the (4) intestines below the kidneys. The abdominal region of the mouse was placed on a respiratory gating device consisting of a balloon filled with air. The squares in the kidneys illustrate the regions where localized proton spectra in Figure 4 were obtained. (a) DBA/ 2FG mouse (male; 6 mo old). (b) DBA/2FGpcy mouse (fe- male; 6 mo old).

Fig. 2. Coronal I H-NMR images of DBA/ZFG and DBA/ ZFG-pcy mice. The (1) kidneys are clearly visible on the left- hand side of each image. The (2) upper chest cavity (far right) and the (3) stomach (middle) are the least intense regions. The (4) livers are the grey homogeneous areas be- tween the kidneys and the upper chest cavities. For the DBA/ 2FG mouse, (5) renal adipose tissue to the left of the kidneys is also observed. No renal adipose tissue is observed for the DBA/2FG-pcy mice. (a) DBA/2FG mouse (female; 6 mo old). (b) DBA/2FG-pcy mouse (female; 3 mo old). (c) DBA/ ZFG-pcy mouse (female; 6 mo old).

tra were taken from 4.7 x 4.7 x 4.7 mm3 voxels of both left and right kidneys based on spatial coordi- nates determined from the transverse images (Fig. 3). Localization was primarily obtained within the central

medullary region in order to eliminate any possible proton signal contamination from renal adipose tissue (especially for the smaller DBA kidneys). The renal medullary volume in the proton image slice in Fig- ure 3a was determined to be 1.24 x 10-l cm’, using a computer program designed in this laboratory, whereas the VOSY sampling volume was 1.04 x 10-l cm3. Therefore, only a minor component of the proton sig- nal (less than 10%) was possibly obtained from the renal cortex. As shown in Fig. 4, only an increase in kidney water proton resonance (6 = 4.5 ppm) was observed for the pcy mice, with no detectable peak for the lipid fatty acid methylene protons (6 = 1.0 ppm). Both water and lipid fatty acid methylene proton

432 Magnetic Resonance Imaging 0 Volume 9, Number 3, 1991

b

a

i

I I I , , 1 I, 1 I, I I I I

18 16 14 12 10 8 6 4 2 0 -2 -4 -6 -6 -10

6 (wm)

Fig. 4. (a) In vivo ‘H-NMR spectrum of a localized region (4.7 x 4.7 x 4.7 voxel) from a DBA/2FG mouse kidney taken from spatial coordinates of the transverse image in Fig. 3a. Localization was achieved by the VOSY method. The predominant resonance (6 = 4.5 ppm) is from water, with a smaller peak intensity for the methylene protons of fatty acid chains (6 = 1 .O ppm). (b) In vivo ‘H-NMR spec- trum of a localized region from a DBA/2FG-pcy mouse kid- ney (female mouse; 6 mo old). Spatial coordinates were taken from the transverse image in Fig. 3c.

peaks were detected from the kidneys of normal DBA mice.

With the use of a modification of the VOSY tech- nique, apparent T, (spin-lattice or longitudinal relax- ation time constant) and T, (spin-spin or transverse relaxation time constant) values for the proton reso- nance of water within the kidneys were determined. However, relative changes in T, and T2 values be- tween voxels of normal and damaged kidney tissue do reflect real changes in the relaxation rates due to tis- sue injury. The major advantage of the VOSY method is that localized spectra can be obtained from the re- gion of interest (ROI) with little contamination of sig-

nals from regions outside the desired ROI.‘O Little or no complications should arise with the use of the VOSY technique under in vivo conditions since the RF power irradiated during each scan is negligible. lo A disadvantage of the technique is that the final spec- trum is subject to both Tl and T2 relaxation effects accumulated during the selection procedure and there- fore only apparent relaxation values can be obtained. The proton T, and T2 values determined by the VOSY method were also uncorrected for the effects of mo- lecular diffusion.

It was observed that there was a considerable change occurring in the kidney Tl relaxation times of mice with the polycystic disease in comparison to those of normal mice (Table 1). Kidney T, values ob- tained from the pcy mice ranged from 799 to 1395 msec, which is a dramatic increase from the DBA con- trol mice kidney Tl values of 617 to 774 msec (Ta- ble 1). Typical literature values for T, of kidney in vitro and in vivo are 450 to 500 at 30 MHz for rats13 and 830 msec at 85 MHz for mice.14 Tl values are de- pendent on the external magnetic field, whereas T2 values are not significantly changed. The observed Tl values are dependent on the amount of “bound” or immobilized water present in the tissue; fast exchange between “bound” and free water gives an average re- laxation time. If more free water is present, as in cysts for example, the observed Tl value will be increased. Any disease process giving rise to an increase in water content or in the ratio of free water to bound water will cause an observed increase in T, . An increase in T, values is also characteristic for other focal diseases such as metastatic tumour nodules in tissues, such as the liver, and multiple sclerosis plaques. l5

An increase in the T2 values measured from the kidneys of mice with polycystic kidney disease was also observed, ranging from 90 to 172 msec, which is significantly increased from the DBA normal mice

Table 1. Apparent localized T1 and T, measurements in kidneys of normal male and female (DBA/2FG)

and polycystic (DBA/2FGpcy) mice

T, measurements T, measurements (msec; *SEM) (msec; & SEM)

Kidney Kidney

Strain Right Left Right Left

DBA (4) 731 * 33 697 + 32 56 + 10 64 + 12 pcy (10) 1182 f 164* 1098 + 69* 112 + 18* 98 + 18*

*Significantly different from the corresponding kidney in the DBA/ 2FG group, p < 0.05. Numbers in parentheses indicate numbers of animals.

In viva ‘H-MRI of polycystic kidney disease in pcy mice 0 R.A. TOWNER ET AL. 433

kidney T2 values of 50 to 72 msec (Table 1). Typical in vitro literature values for mouse kidney T2 range from 42 to 55 msec,14 regardless of the magnetic field. T2 is generally shorter for tissues that are more solid. The T2 values of water resonances are strongly dependent upon the amount of ‘free’ water present in the tissue.“j*” The overall observed water T2 value is a fast-exchange averaged value determined from the T2 of ‘free’ water, the very short T2 of water bound to proteins and membranes, and the fraction of total wa- ter in each pool. The increase in water T2 is likely due to an edematous response as a result of the formation of cysts. An edematous response increases the molar fraction of ‘free’ water resulting in an increase in ob- served water T,.

The increase in observed proton signal in the im- ages of mice with polycystic kidneys is therefore due not only to an increased amount of water present, as determined from localized spectroscopy, but also due to changes in T, and T,. The detected increases in wa- ter proton T, and T2 values are consistent with find- ings from those of Leung et al. and Marotti et al. for PKD in humans in vivo using two-point computations from T, and T,-weighted images.4*1s

Tissue analyses support the findings of increased water and reduced lipid contents, observed in the ‘H- NMR spectra of kidneys from the pcy mice (Fig. 4). The average water content of the pcy kidneys (87.4%) was higher than the DBA mouse kidneys (70.2%), whereas the total phospholipid, cholesterol, and tria- cylglycerol levels were reduced by 91.6%, 88.0%, and 90.7%, respectively in the pcy, compared to the DBA kidneys (H. Aukema, unpublished observations).

In a study by Iaina et al, the cortical and medullary T, and T2 water proton relaxation times were deter- mined from excised rat kidneys.6 In vitro water T,

and T2 values for normal renal cortex were found to be 473 + 1.3 msec and 47.5 + 0.2 msec, respectively; and in the medulla T, and T2 values were 691 + 35 msec and 71 f 0.4 msec, respectively.6 Any contribu- tion in proton signal from the cortex would decrease the measured T, and T2 values from the medulla. Therefore, the substantial measured increase in T,

and T2 values obtained from the kidneys of pcy mice indicates that even if there were some proton contri- bution from the cortex it would be negligible in com- parison to the overall difference in kidney T, and T,

values observed between DBA and pcy mice. In addition, histological studies indicate that the

cysts are formed in both cortical and medullary re- gions of the kidneys of pcy mice. The cut surfaces of kidneys from animals surviving until the eighth month of life typically have a rugged appearance due to cyst formation and few, if any, intact cortical and jux- tamedullary nephrons.3 Therefore the Tl and T2 values

from either the cortical or medullary regions of pcy mouse kidneys would be similar. This change is simi- lar to that reported in kidneys from ADPKD patients in which the renal cysts appear to be homogenously distributed throughout the cortical and medullary tis- sue, obscuring the renal papilla and pyramids in severe cases. I9

Studies were also done to investigate the onset of cyst formation in the kidneys of pcy mice as a func- tion of age. The disease progresses as the mouse ma- tures and seems to be optimal at 3-6 mo of age. Early morphological changes in these mice include marked distensions of the Bowman’s capsule, the proximal and distal tubules, and the collecting ducts. As the af- fected animal matures, the cysts become sufficiently large to completely obscure the corticomedullary junc- tion (Fig. 5). With the use of MRI and a localized spectroscopy technique (VOSY) it was possible to de- tect changes in pcy kidneys in vivo as early as 4-5 wk of age (Table 2) at a time when urinary protein and glucose excretions are normal and prior to the devel- opment of hematuria.’

The pcy mouse has a relatively long life span com- pared to the other murine models of PKD and thus is a useful model for studies on the development of this disease. Interestingly, urine production and several in- dices of renal function are unimpaired in these ani- mals until the onset of uremia.3 Further studies with MRI and localized spectroscopy techniques, such as

Fig. 5. Morphological differences between (a) a pcy mouse kidney, and (b) a kidney from an age-matched (105 days) DBA normal mouse. Absolute organ weights of the pcy kid- neys were 3.7 f 0.6 g compared to organ weights of 0.35 + 0.00 g for the DBA control mice (average r?r SD for 4 ani- mals/group). The kidney from the pcy mouse is uniformly studded with cysts filled with a watery fluid; these cysts are distributed to some extent in the cortex and medulla regions resulting in a loss of the cortico-medullary junction.

434 Magnetic Resonance Imaging 0 Volume 9, Number 3, 1991

Table 2. Apparent localized T, and Tz measurements in kidneys of pcy mice of various ages

T, measurements T, measurements (msec; *SEM) (msec; +SEM)

Kidney Kidney Age (day) Right Left Right Left

9 (2) 772 - 62 -

30 (4) 784 -t 63 797 f 157 122 * 34 (3) 122 * 12 (3) 60 (4) 1048 f 71* 1279 f 408 86 + 30 94* 18

*Significantly different from the corresponding kidney at 30 days of age. Numbers in parentheses indicate numbers of animals.

VOSY, will provide a better understanding of the de- velopment of PKD in this strain, and will give an ac- curate assessment of their clinical condition, and the effectiveness of various nutritional/pharmacological therapies (including dietary supplementation with the O-3 fatty acids) which have been proposed for the treatment of PKD.20~2’ These studies demonstrate

that MRI in conjunction with a localized spectroscopy technique, such as VOSY, can be used to assess the early onset of PKD in an experimental animal model for the human disease in vivo over extended periods of time. Future studies will concentrate on the tracking of cyst development in the newborn pcy mouse and the use of MRI as a means of distinguishing simple cysts from perirenal hematomas, a serious complica- tion of the disease in ADPCK.3 The incidence and se- verity of this complication has not been documented

in the pcy mouse.

1.

2.

3.

4.

5.

6.

REFERENCES

Gardner, K.D. Cystic disease of the kidney. In: SC. Massry and R.J. Glassock (Eds.), Textbook of Nephrol- ogy. Baltimore/London: The Williams & Wilkins Co., Vol. 1; Ch. 6; 1983:~. 160. Delaney, W.B.; Alder, S.; Bruns, F. J.; Lincinia, M.; Se- gal, D.B. and Fraley, D.S. Autosomal dominant poly- cystic kidney disease: Presentations, complications and prognosis. Am. J. Kidney Dis. 5:104-l 11; 1985. Levine, E.; Grantham, J.J. Perinephric hemorrhage in autosomal dominant polycystic kidney: CT and MR findings. J. Comput. Assist. Tomogr. 11: 108-l 11; 1987. Takahashi, H.; Ueyoma, Y.; Hibino, T.; Kuwahara, Y.; Suzuki, S.; Hioki, K.; Tamaoki, N. A new mouse model of genetically transmitted polycystic kidney disease. J. Urology 135:1280-1283; 1986. Leung, A.W.-L.; Bydder, G.M.; Steiner, R.E.; Bryant, D.J.; Young, I.R. Magnetic resonance imaging of the kidneys. Am. J. Radiology 143:1215-1227; 1984. Iaina, A.; Abrashkin, S.; Weininger, J. Proton MR

7.

8.

9.

10.

Il.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

study of different types of experimental acute renal fail- ure in rats. Magn. Reson. Imaging 4:241-244; 1986. Iaina, A.; Weininger, J.; Abrashkin, S.; Azoury, R. Re- nal proton magnetic resonance in experimental acute re- nal failure in rats. Renal Physiol. Biochem. 12:202-211; 1989. Terrier, F.; Lazeyras, F.; Posse, S.; Aue, W.P.; Zimmer- mann, A.; Frey, B.M.; Frey, F.J. Study of acute renal ischemia in the rat using magnetic resonance imaging and spectroscopy. Magn. Reson. Med. 12:114-136; 1989. Brauer, M.; Towner, R.A.; Renaud, I.; Janzen, E.G.; Foxall, D.L. In vivo proton nuclear magnetic resonance imaging and spectroscopy studies of halocarbon induced liver damage. Magn. Reson. Med. 9:229-239; 1989. Kimmich, R.; Hoepfel, D. Volume-selective multipulse spin-echo spectroscopy. J. Magn. Reson. 72~379-384; 1987. Frahm, J.; Merboldt, K.D.; Hanicke, W.; Haase, A. Stimulated echo imaging. J. Magn. Reson. 64:81-93; 1985. Granot, J. Selected volume excitation using stimulated echoes (VEST): Applications to spatially localized spec- troscopy and imaging. J. Magn. Reson. 70:488-492; 1986. Morris, P. Nuclear Magnetic Resonance Imaging in Medicine and Biology. Oxford: Clarendon Press; 1986: p. 264. Beall, P.T.; Amtey, S.R.; Kasturi, S.R. NMR Data Handbook for Biomedical Applications. New York: Pergamon Press; 1984:~. 62. Foster, M.A.; Hutchison, J.M.S. Practical NMR Imag- ing. Oxford, IRL Press; 1987:~. 45. Taylor, D.G.; Bore, C.F. A review of the magnetic res- onance response of biological tissue and its applicabil- ity to the diagnosis of cancer by NMR radiology. J. Comput. Assist. Tomogr. 5:122-126; 1981. Lynch, L.J. Water relaxation in heterogenous and bio- logical systems. In: J. Cohen (Ed.), Magnetic Resonance in Biology. New York: Wiley; 1983:~~. 258-267. Marotti, M.; Hricak, H.; Fritzsche, P.; Crooks, L.E.; Hedgcock, M.W.; Tanagho, E.A. Complex and simple renal cysts: Comparative evaluation with MR imaging. Radiology 162:679-684; 1987. Evan, A.P.; Gardner, K.D. Morphology of polycystic kidney disease in man and experimental models. In: H.B. Cummings and S. Klahr (Eds.), Chronic Renal Disease: Causes, Complications and Treatment. New York: Plenum Press; 1985:~~. 55-68. Takahashi, H.; Yoshida, K.; Valli, V.E.O.; Philbrick, D.J.; Yamaguchi, T.; Holub, B.J. The effects of an omega-3 fatty acid-enriched diet on organ weights and renal pathophysiology in an animal model for polycystic kidney disease. FASEB J. 3:A933 (Abs. 4110); 1989. Yamaguchi, T.; Valli, V.E.O.; Philbrick, D.J.; Holub, B. J. Effects of a dietary supplementation with omega- 3 fatty acids on the pathophysical and phospholipid compositions of kidneys from mice with polycystic kid- ney disease. Clin. Invest. Med. 12:Bl03 (Abs. C-668); 1989.