1

Inactivation of hepadna virus by EAW

Original articlesJournal of Antimicrobial Chemotherapy (2000) 46, 363-368

Inactivation of a hepadnavirus by electrolyzed acid water

Masami Tagawaa, Taketo Yamaguchia, Osamu Yokosukaa, Shoichi Matsutania,Toyoyuki Maedab and Hiromitsu Saishoa

First Department of Medicine, School of Medicine, Chiba University, 1-8-1 Inohana,Chuo-ku, Chiba260-8670, Japan and Department of Medical Instruments, KaigenCo.Ltd, Osaka, Japan

Glutaraldehyde is used a disinfectant for endoscopes, but is an irritant and so shouldbe replaced by an alternative. Electrolyzed acid water (EAW) has a bactericidaleffect, and an endoscopic washing device using EAW has been developed in Japan.To investigate the effect of EAW on the infectivity of viruses, we treated duckhepatitis B virus (DHBV), which has similar properties to hepatitis B virus, with EAW,and determined the number of remaining infectious virus particles in a bioassaysystem. One-day-old Pekin ducks were inoculated with duck serum containing 105.5

ID50 DHBV; the serum had previously been incubated with 100 volumes of EAW orion-exchanged water at room temperature for 7 min. DHBV infection was indicatedby detection of viral DNA in duck serum samples 1-8 weeks afterinoculation.Treatment of serum with EAW diminished DHBV infectivity whereastreatment with ion-exchanged water did not. The virus load was estimated to havebeen reduced to 101-103 ID50 during the first 1 min and to <100.5 ID50 in the next 6 minof incubation when compared with the control. Thus, EAW directly inactivates DHBVand its clinical application is recommended.

2

Introduction

Cleaning and disinfection of medical devices and instruments contaminated withblood and related materials are important in the prevention of infection bymicroorganisms, including viruses. Glutaraldehyde (2%) was recommended forcleaning endoscopic instruments by O'Connor & Axon in 19831 and by the WorkingParty of the British Society of Gastroenterology(BSG) Endoscopy Committee in19882. It has since been the most commonly used disinfectant in the world for thispurpose. However, it has irritant and sensitizing properties3-10 and the Working Partyof BSG has suggested ways in which exposure of healthcare personnel toglutaraldehyde can be eliminated or minimized by using alternative disinfectants orautomated washer disinfectors11. According to the guidelines of the BSG, alternativesto glutaraldehyde have to be at least as microbicidal as glutaraldehyde, non-irritantand compatible with both endoscope components and decontamination equipment.

• Several different electrochemical solutions containing a mixture of radicalswith oxidizing properties have been reported to be microbicidal and arepossible alternatives to glutaraldehyde. Electrolysed acid water (EAW), asolution generated by electrolysis at the anode of 0.05% NaCl solution, has alow pH, a high oxidization-reduction potential (ORP) and bactericidal effects12-15. Iwasawa et al.12 reported that growth of 147 bacterial isolates, includ ingmethicillin-resistand Staphylococcus aureus, was completely inhibited bytreatment with 100 volumes of EAW for 5 s with the exception of two Bacillusspp. This effect was equivalent to that of a 0.1% sodium hypochlorite (NaOCl)solution and much more potent than that of a 0.5% chlorhexidine gluconatesolution. It has been reported that EAW reduces hepatitis B virus (HBV)antigenicity,13-16 but reduction in viral infectivity by EAW has not yet beenconfirmed.

• Duck hepatitis B virus (DHBV),17 an avian hepadnavirus, resembles HBV in itsmolecular structure and other properties, and experimental infection is easilyestablished in young ducks. The virucidal efficacy of glutaraldehyde,18 NaOCland sodium dichloroisocyanurate(NaDCC)19 have already been assessedusing the DHBV model. The presence of virus can be detected in suchbioassay systems with a sensitivity of 100 copies (genome-equivalent DHBV,estimated by quantification of DHBV DNA) per animal; this sensitivity isequivalent to that of the polymerase chain reaction amplification method.20

Detection of infectious virus is more appropriate than the detection of nucleicacids or virus antigens for determining the activity of disinfectants. Here wepresent the results of an investigation into the effect of EAW on viral infectivity,in which ducks ware inoculated with EAW-treated DHBV.

3

Materials and methods

Generation of EAW

EAW was generated as previously reported16 by the Cleantop WM-1 (Kaigen Co.Ltd, Osaka, Japan), a portable endoscope-cleaning device incorporating anelectrolysis apparatus. Briefly, 10 l of tap water containing 0.05% NaCl waselectrolysed at room temperature for 45 min with a current of 3 A. The anodic andcathodic sides of the solution were separated by a cationic membrane (Nation 450,Dupont, Wilmington, DE, USA); the solution on the anodic side, which contained theelectrolysed products, was called EAW and used for the study. The pH, ORP andchlorine concentration were measured using a pH meter (Horiba, Kyoto, Japan),ORP meter (Horiba) and freechlorine meter (Hach, CO, USA).

Infection study

Serum stock of DHBV. A frozen stock of duck serum containig wild-type infectiousDHBV (DHBV16)21 (22.8 ng DHBV DNA/ml, 6.6 × 109 DHBV genomes/ml, 108.3

ID50/ml) was used as the inoculum. The ID50 of this serum was calculated to be 101.5

genomes by an infection study using sequentially diluted infective serum (data notshown). Infection with 100.5ID50 in 1 day old ducklings established a DHBV carrierstate: none of the ducklings inoculated with 10-0.5 ID50 became DHBV carriers 8weeks after inoculation. Inocula containing 10-0.5-105.5ID50 were also used as controlsin the quantitative analysis.

Inoculum preparation. (i ) EAW-treated group. Serum containing 105.5ID50DHBV(1.5µl, 107 genomes) was mixed with 100 volumes of EAW and incubated at roomtemperature for 7 min. The treated serum was neutralized by adding 1/10 volume (15µl) of 10 × phosphate-buffered saline (PBS). In some experiments, inocula weremixed with 500 volumes of EAW or were incubated for 1 min. (×) Ion-exchangeddistilled water (IDW) -treated group. The same amount of serum was treated withIDW in the sameway as with EAW.

Inoculation of ducks and serum sampling. One-day-old Pekin ducks (Anasdomesticus) were purchased from a DHBV-free commercial supplier in Chiba, Japan.Oneday-old ducklings became chronic DHBV carriers after infection with infectiousserum.22 Inocula, prepared as described above, were injected intravenously into twoto eight 1-day-old ducklings per group, and serum samples were obtained 1, 2, 3, 4,6 and 8 weeks after inoculation. The animal handling protocols followed theGuidelines for Animal Experimentation, Inohana Campus, Chiba University and wereapproved by the Animal Ethics Committees of Chiba University School of Medicine.

4

Quantification of DHBV DNA in serum

DNA was extracted from the serum samples and analysed by slot-blot hybridizationwith a 32P-labelled DHBV DAN probe as reported previously.23 DHBV DNA in thesample was quantified using a BAS2000 image analyser (Fuji, Tokyo, Japan) usingknown amounts of cloned DHBV DNA as the control.

Statistical analysis

Statistical analysis by the Kruskal-Wallis method was performed for the controlgroup. The significance between control and each EAW- and IDW-treated group wasanalysed by the Mann-Whitney U-test.

Results

Properties of EAW

The pH of the EAW was 2.15, the ORP 1153 mV and the chlorine concentration 7.15ppm. The pH of EAW did not change during incubation with serum, but increased to5.59 after addition of PBS. The ORP gradually declined to 620 mV during theincubation with serum and decreased to 453 mV after the addition of PBS. Thechlorine concentration had decreased rapidly, to 0.3 ppm, after 1 min of incubationwith serum; it did not change during the incubation or when PBS was added.

Infectivity of DHBV treated with EAW and IDW

The numbers of DHBV-positive ducks in the groups receiving untreated and treatedinoculum are shown in the Table. The inoculum with 105.5 ID50 of DHBV incubatedwith 100 volumes of EAW for 7 min caused fewer infections (none of eight duckswere infected 8 weeks after inoculation) than in the control group (all seven ducksinfected). Six ducks inoculated with the serum treated with 100 or 500 volumes ofIDW for 7 min were all infected 1 - 8 weeks after inoculation.

In the group given an inoculum treated with 100 volumes of EAW for 1 min, DHBVDNA was detected in two of five ducks 1 week after inoculation, and 3/3 were stillpositive for DHBV DNA after 8 weeks. Treatment with 500 volumes of EAW reducedthe infectivity to 0/4 ducks even though the time of incubation was only 1 min.

5

DHBV titre

The DHBV titre of the serum after inoculation with the control and treated serum isshown in the Figure. Inoculation with 105.5 ID50 of DHBV resulted in infection in allducks with high titres of viraemia (10 ng/ml DHBV DNA) after 1 - 8 weeks (Figure,panel a). Inoculation with 10-0.5of ID50 resulted in detectable DHBVDNA54 weeksafter inoculation but chronic infection was not established with this amount ofinfectious virus (Figure,panel b) Treatment with 100 volumes of EAW for 7 min(Figure, panel c) and 500 volumes of EAW for 1 min (panel f) reduced the infectionrate (Table) and the viral titre in the inoculated ducks. Ducks inoculated with the IDW-treated serum showed high-titre viraemia and the time course of infection was similarto that of the untreated control (Figure, panel d). In the group given an inoculumtreated with 100 volumes of EAW for 1 min (Figure, panel e), no DHBV wasdetectable after 1 week in three ducks and the titre was 0.18 and 0.08 ng/ml in theother two ducks (DJ1654 and DJ1657, respectively), much lower then in the controlgroup (Figure, panel a).

Statistical analysis of DHBV titre

As previously reported,the viral titre in the early phase of infection correlated with theamount of infectious virus in the inoculum.24 One week after inoculation, the DHBVtitres in the control group inoculated with 10-0.5,100.5,101.5, 102.5,103.5,104.5, and 105.5

ID50 of DHBV were 0.36, 0.07, 0.12, 0.42, 13.7, 49.2 and 39.3 ng/ml serum, respectively.

The P value with Kruskal-Wallis analysis was 0.042, indicating that each group wassignificantly independent and that the DHBV titre after 1 week was related to theamount of infectious virus inoculated.

To confirm that EAW treatment reduced viral infectivity, we analysed the DHBV titrestatistically in ducks treated with 100 volumes of EAW for 7 min and in untreatedcontrols 1 week after inoculation.The P value (Mann Whitney U-test) was 0.019,indicating that the infectivity of DHBV was reduced significantly by the EAWtreatment.Treatment with 100 volumes of EAW for 1 min also reduced the DHBV titresignificantly after 1 week (P = 0.045); treatment with 500 volumes of EAW for 1 minreduced the DHBV titre after 1 week but it was not statistically significant (P =0.064).The reduction in DHBV titre in ducks treated with 100 or 500 volumes of IDWwas not significant (P > 9.999 and P = 0.248, respectively).

To assess the level of infectious virus remaining in the inoculum treated with 100volumes of EAW for 7 min, DHBV titres after 1 week were compared with those incontrol ducks inoculated with various amounts of DHBV.The P values were 0.335, 0.075, 0.045, 0.014, 0.008, 0.019 and 0.019 for the groupsinoculated with 10-0.5,100.5,101.5, 102.5,103.5,104.5, and 105.5 ID50 of DHBV,respectively.These results indicate that the level of infectious DHBV in the inoculum was reducedfrom 105.5 to 100.5 ID50 treatment with EAW.Five ducks were inoculated with serum that has been incubated with 100 volumes ofEAW for 1 min. The titres after 1 week were significantly different from those ingroups receiving 103.5,104.5 and 105.5 ID50 (P = 0.022, 0.045 and 0.045, respectively).These data indicated that the amount of infectious virus in the inoculum was <103.5

ID50 DHBV after 1 min of incubation.

6

Discussion

This is the first report on the direct of an electrochemically modified saline solutionwith a strong oxidizing property on hepadnavirus infectivity. DHBV infectivity wascompletely lost after incubation of the inoculum with 100 volumes of EAW for 7 min,and with 500 volumes for 1 min. The inactivation of the virus was irreversible afteradding PBS to neutralize pH and adjusting the salt to a physiological concentration.

We quantified infectious DHBV remaining in the treated inoculum by tow methods: (i)lack of DHBV infection 8 weeks after inoculation was taken as an indication that theinoculum contained less that the ID50 of DHBV; (ii) the DHBV titre 1 week afterinoculation of treated inoculum was compared with that in controls and thesignificance of the difference was assessed.

Infectivity was lost after incubation of serum with 100 volumes of EAW for 7 min atroom temperature and with 500 volumes of EAW for 1 min, indicating that thesetreatments reduced the number of infectious viruses from 105.5 ID50 to < 100.5

ID50,and to 5 10-0.5 ID50 in some preparations.It was also estimated that the infectivitywas reduced to about 101-3 ID50 at the end of the first minute of treatment. As dilutionof the virus with IDW did not result in any inactivation, it was concluded that EAWdirectly affected the virus particles and reduced their infecticity.

Electrolysis of a low concentration of NaCl in water produces Cl2,HClO,H+Cl-,O2 andO3 at the anode; the solution at the anode site is characterized by a low pH, a highORP and low concentrations of Cl2.the bactericidal effect of EAW is explained by thelow pH (<2.7), 25 ORP exceeding 970 mV and the presence of Cl2; these factorshave synergic effects.15,26 The minimal chlorine concentration required for thevirucidal effect of NaOCl and NaDCC, as assessed by in vitro DNA polymeraseassay of DHBV, was 3×103 ppm, which was equivalent to NaOCl and NaDCCsolutions of about 3%. The presence of Cl2 is not the only reason for the virucidaleffect of EAW.

Abe et al.13 and Morita et al.16 have reported that EAW-treated HBV cannot bedetected by radiommunoassay13 or enzyme immunoassay16 with HBsAb, an antibodythat recognizes the tertiary structure of HBsAg on the surface of the virion, indicatingthat EAW affected the antigenicity of the surface antigen. Morita et al.16 analysed theeffect on HIV-1 recombinant reverse transcriptase and HIV-1 particles in vitro anddemonstrated that naked and packaged reverse transcriptase activities were reducedby EAW in a time- and concentration-dependent manner. EAW may denature theprotein, destroying its immunoreactivity and enzymatic function. From our results andthose of Abe et al.13 and Morita et al.,16 the mechanism of actions could involve: (i)irreversible denaturation of surface proteins, including preS proteins, which arenecessary for attachment of virions to the receptor on hepatocytes ; or (ii) loss offunction of DNA polymerase or of the viral DNA genome. Hence, EAW is expected tohave virucidal effects for a wide variety of viruses, including non-enveloped viruses.

7

A DHBV bioassay has been used previously to assess the efficacy of glutaraldehyde,18 NaOCl and NaDCC as disinfectants.19 Because these disinfectants are irritant andtoxic for intravenous injection, virions purified by ultracentrifugation or small amountsof virus were injected into young ducks. Murray et al18 treated duck serum with anequal volume of 2.1% glutaraldehyde (pH 5) for 10 min at room temperature. Asolution equivalent to 102 ID50 was used to inoculate 1-day-old ducks; infectivity wasreduced from 100% to 60% and the DHBV titre was reduced by t103.27 Tsiquaye &Barnard19 treated duck serum with NaOCl and NaDCC solutions in which the finalchlorine concentration was 3.0 × 103 ppm. None of the ducks inoculated with 103.2

ID50 of DHBV became infected. In our experiment, serum was treated with 100volumes of EAW for 7 min and none of the ducks inoculated with 105.5 ID50 DHBVwere infected. Thus, EAW had a strong disinfectant activity, comparable to those ofglutaraldehyde, NaOCl and NaDCC. Even though EAW appears to be virucidal itsclinical use awaits more information about: its effect on bacterial spores andmycobacteria; its stability during handling; it is safety for medical personnel and theenvironment; the safety to patients of solution remaining on the instrument; cost-benefit analysis; and its applicability to automated washing machines. None of the 1-day-old ducks (mean body weight 55 g) inoculated intravenously with ≤700 µl ofEAW after neutralization died because of acute toxicity. No sensitization to EAW hasbeen reported during 2.5 years of clinical use of Cleantop WM-1 in endoscopic unitsin hospitals in Japan. In this report, a virucidal effect of EAW was demonstrated in theDHBV bioassay system. EAW is an alternative to the currently used glutaraldehyde.Further investigation into the mechanisms of activity against viruses and othermicroorganisms is warranted.

8

Table

Detection of DHBV DNA in the serum of ducks in control, EAW-treated and IDW-treated groups

Numbers of DHBV-positive ducks/numbers of ducks analysed are shown. All duckswere inoculated with 105.5ID50 of untreated (control) or treated DHBV.aSerum samples from other ducks were not obtained at these points. bOne duck died in an accident.

9

Figure

DHBV DNA in the serum of ducks after inoculation with untreated (a and b) or treated(c - f) duck serum. Inoculum contained 105.5ID50 (107 genomes) (a, c - f) or 10-0.5ID50(101 genomes) (b) of DHBV before treatment. (a and b) Mean (1) and S.E. (bar) ofDHBV DNA titre of each group. (c - f) Inoculum was incubated with 100 volumes ofEAW for 7 min (c; t, DJ1646; t, DJ1467; t, DJ1648; ,, DJ1649; ,, DJ1611; ,, DJ1612; ,,DJ1613; ,, DJ1614), 100 volumes of IDW for 7 min (d; ,, DJ1623; ,, DJ1624;,,DJ1625), 100 volumes of EAW for 1 min (e; ,, DJ1654; ,, DJ1655; ,, DJ1656; ,,DJ1657; ,, DJ1658) and 500 volumes of EAW for 1 min (f; ,, DJ1659; ,, DJ1660; ,,DJ1661; ,, DJ1662). DHBV DNA titre of individual ducks 1 - 8 weeks after inoculationare indicated

10

References1 O'Connor, H. J. & Axon, A. T. R (1983).

Gastrointestinal endoscopy: infection and disinfection. Gut24, 1067-77.2 Working Party of the British Society of Gastroenterology. (1988).

Cleaning and disinfection of equipment for gastrointestinal flexible endoscopy: interimrecommendations of a Working Party of the British Society of Gastroenteroloty. Gut29, 1134-51.

3 Axon, A. T. R., Banks, J., Cockel, R., Deverill, C. E. A. & Newmann,C. (1981). Disinfection in upper- digestive-tract endoscopy in Britain.Lancet i , 1093-4.

4 Hansen, K. S. (1983). Occupational dermatoses in hospital cleaning women. Contact Dermatitis 9, 343-51.

5 Corrado, O.J., Osman, J. & Davies, R. J. (1986). Asthma and rhinitis after exposure to glutaraldehyde in endscopy units. Human Toxicology 5, 325-8.

6 Norback, D. (1988). Skin and respiratory symptoms from exposure to alkaline glutaraldehydein medical services. Scandinavian Journal of Work, Environment and Health 14, 366-71.

7 Jachuck, S. J., Bound,C.L., Steel, J.& Blain , P. G. (1989). Occupational hazard in hospitalstaff exposed to 2 per cent glutaraldehyde in an endoscopy unit. Journal of the Society ofOccupational Medicine 39, 69-71.

8 Fowler, J. F. (1989). Allergic contact dermatitis from glutaraldehyde exposure. Journal ofOccupational Medicine 31, 852-3.

9 Wiggins, P., McCurdy, S.A. & Zeidenberg, W. (1989). Epistaxis due to glutaraldehydeexposure. Journal of Occupational Medicine 31, 854-6.

10 Calder, I. M., Wright, L. P. & Grimstone, D. (1992). Glutaraldehyde allergy in endoscopyunits. Lancet 339, 433.

11 Working Party of the British Society of GastroenterologyEndoscopy Committee. (1998). Cleaning and disinfection of equipment for gastrointestinalendoscopy. Gut42, 585-93.

12 Iwasawa, A., Nakamura, Y. & Mizono, T. (1993). Bactericidal activity of aqua oxidation wateragainst clinically isolated strains. Journal of Japanese Environmental Infection 8, 11-6.

13 Abe, S., Miya, Y. & Okuda, R. (1994). Inactivation of hepatitis B virus by high oxidationpotential water. Japanese Journal of Restorative Dentistry 37, 1616-23.

14 Okuda, R., Sasazaki, H., Kanehira, M., Okabe, T., Abe, S., Tagami, A. et al. (1994). Electronmicroscopical study of bactericidal actions of high oxidation potential water. JapaneseJournal of Restorative Dentistry 37, 755-65.

15 Ito, K., Nishida, T. & Murai, S. (1996). Inhibitory effects of acid water prepared by anelectrolysis apparatus on early plaque formation on specimens of dentine. Journal of ClinicalPeriodontology 23, 471-6.

16 Morita, C., Sano, K., Morimatsu, S., Kiura, H., Goto.T., Kohno, T. et al. (2000). Disinfectionpotential of electrolyzed solution containing sodium choloride at low concentrations. Journalof Virological Methods 85, 163-74.

17 Mason, W. S., Seal, G. & Summers, J. (1980). Virus of Pekin ducks with structural andbiological relatedness to human hepatitis B virus. Journal of Virology36, 829-36.

18 Murray, S. M., Freiman, J. S., Vickery, K., Lim, D., Cossart, Y.E. & Whiteley, R. K. (1991). Duck hepatitis B virus: a model to assess efficacy ofdisinfectants against hepadnavirus infectivity. Epidemiology and Infection 106, 435-43.

19 Tsiquaye, K. N. & Barnerd, J. (1933). Chemical disinfection of duck hepatitis B virus: a modelfor inactivation of infectivity of hepatitis B virus. Journal of Antimicrobial Chemotherapy32,313-23.

20 Larzul, D., Guigue, F., Sninsky, J.J., Mack, D.H., Brechot, C. & Guesdon, J. L. (1988).Dtection of hepatitis B virus sequences in serum by using in vitro enzymatic amplication.

11

Journal of Viro-logical Methods 20, 227-37.

21 Mandart, E., Kay, A. & Galibert, F. (1984). Nucleotide sequence of a cloned duck hepatitis BVirus genome: comparison with woodchuck and human hepatitis B Virus sequences. Journalof Virology 49, 782-92.

22 Omata, M., Uchiumi, K., Ito, Y., Yokosuka, O., Mori, J., Terao, K. et al. (1983). Duck hepatitisB virus and liver diseases. Gastroenterology 85, 260-7.

23 Tagawa, M., Omata, M., Yokosuka, O., Uchiumi, K., Imazeki, F. & Okuda, K. (1985). Earlyevents in duck hepatitis B virus infection: sequential appearance of viral deoxyribonucleicacid in the liver, pancreas, kidney, and spleen. Gastroenterology 89, 1224-9.

24 Jibert, A. R., Botten, J. A., Miller, D. S., Bertram, E. M., Hall, P. M., Kotlarski, J. et al. (1998).Characterization of age- and doserelated outcomes of duck hepatitis B virus infection.Virology 244, 273-82.

25 Nottle, W. A. (1982). Oral Microbiology with Basic Microbiology and lmmunology, 4th edn,pp. 28 and 73. C. V. Mosby, St Louis, MO.

26 Sakai, T., Shibata, A., Bandai, M. & Nakane, F. (1993). Application. of electrolyzed waterproduced by Oxylyzer for dental region. Part 1. General rules for usage. Journal of JapaneseProsthodontics Society 37, 920-7.

27 Kobayashi, H., Tsuzuki, M., Koshimizu, K., Toyoma, H., Yoshihana, N., Shikata, T. et al.(1984). Susceptibility of hepatitis B virus to disinfectants or heat. Journal of ClinicalMicrobiology 20 ,214-6.

Received 15 December 1999; returned 29 February 2000; revised 24 March 2000; accepted 19 April 2000

12

Disinfection potential of electrolyzed solutions containing sodium chlorideat low concentrationsJournal of Virological Method 85 (2000) 163-174

Chizuko Moritaa, Kouichi Sanoa,*, Shinichi Morimatsua, Hiromasa Kiuraa,b, ToshiyukiGotoa, Takehiro Kohnoa, Wu Honga, Hirofumi Miyoshic, Atsuo Iwasawad, YoshikoNakamurad, Masami Tagawae, Osamu Yokosukae, Hiromitsu Saishoe, ToyoyukiMaedaf, Yoji Katsuokab

aDepartment of Microbiology, Osaka Medical College, Takatsuki-shi, Osaka 569-8686 JapanbDepartment of Urology, Osaka Medical College, Takatsuki-shi, Osaka 569-8686JapancDepartment of Internal Medicine, Kobe Teishin Hospital, Kobe 651-8798 JapandDepartment of Clinical Pathology, Showa University, Fujigaoka Hospital, Kanagawa227-8501 JapaneFirst Department of Medicine, School of Medicine, Chiba University, Chiba 260-8670, JapanfDepartment of Medical Instruments, Kaigen Co. Ltd., Osaka 541-0045, Japan

Received 5 July 1999; received in revised form 1 November 1999; accepted 2November 1999