Single-radial-haemolysis test for diagnosing flavivirus infections, particularly Japanese encephalitis*

Bulletin of the World Health Organization, 64 (5): 735-740 (1986) © World Health Organization 1986 Single-radial-haemolysis test for diagnosing flavivirus infections, particularly Japanese encephalitis* SAMUEL GEORGE' & KHORSHED PAVRI 2 Use of the single-radial-haemolysis (SRH) technique for the diagnosis offlavivirus infections is described. A large number ofpaired and single convalescent serum samples collected from cases of encephalitis during two major outbreaks in Kolar district of Karnataka State in India during 1977 and 1979 were tested by this technique. The results were compared with those obtained in the haemagglutination inhibition (HI) test in all cases, and the complement fixation (CF) and neutralization tests in some cases. Japanese encephalitis virus was shown by the SRH test to be the major etiologic agent responsible for both epidemics. This was corroborated by the HI, CF and neutralization test results. The single-radial-haemolysis test wasfound to be simpler and more specific and sensitive than the haemagglutination inhibition test. The single-radial-haemolysis (SRH) test was originally developed for the detection and assay of antibody to influenza virus haemagglutinins (1, 2) but has been successfully adapted for detection of several other haemagglutinating viruses, e.g., rubella virus (3, 4), mumps virus (5, 6), coronavirus (7, 8), parainfluenza viruses (9), and measles virus (10). It has been applied to togaviruses also (11-14). In this paper we describe our attempts at the application of the SRH test for the diagnosis of flavivirus infections, with particular reference to Japanese encephalitis, which is one of the major public health problems in India. The results of the SRH test were compared with those obtained in the haemagglutination inhib- ition (HI) test, and in some cases with those of the neutralization and complement fixation (CF) tests. MATERIALS AND METHODS Viruses and immune sera. The Indian reference strains of Japanese encephalitis virus (JE, P20778 strain), West Nile virus (WN, G22886 strain) and dengue type 2 virus (DEN-2, P23085 strain) and their respective mouse immune sera were used in this study. Human sera. The following human sera were employed in the study. (1) Paired sera from 4 cases * From the National Institute of Virology, 20-A, Dr Ambedkar Road, Pune 41 1001, India. ' Senior Research Officer. Requests for reprints should be sent to this author at the National Institute of Virology Unit, Victoria Hospital, Bangalore 560 002, India. 2 Director. and single convalescent sera from 22 cases collected during the 1977 epidemic of encephalitis in Kolar district of Karnataka State, India. (2) Paired sera from 171 cases and single convalescent sera from 30 cases obtained during the 1979 epidemic of encephalitis in Kolar district. Antigens. Initially, sucrose-acetone extracted (SA) antigens were used. Later, a crude 20% suspension of infected mouse brains in borate saline, pH 9.0, was used; this antigen was termed "SRH antigen". Both the SA and SRH antigens were treated with protamine sulfate. Both were stored in the frozen state (-50 °C) and their HA titres were determined before use. Preparation of SRH immunoplates. Erythrocytes from 2-4-day-old chicks were washed three times in phosphate-buffer saline (PBS), pH 7.2, and finally suspended in 0.4% bovine albumin borate saline (BABS) at the optimal pH for virus haemaggluti- nation to make a 15% suspension. To 1 ml of the erythrocyte suspension was added 1 to 2 ml of SA or SRH antigen having an HA titre of 1280 to 2560. The mixture was kept at 4 °C for 30 minutes with occasional shaking. The antigen-sensitized erythro- cytes were washed three times in 0.4% BABS and finally suspended in the same diluent to make a 15% suspension. Erythrocytes treated with an equal volume of normal mouse brain "antigen" were used for the preparation of control plates. Normal erythrocytes were also used for the preparation of additional control plates. 4717 -735- 736 S. GEORGE & K. PAVRI To prepare one SRH plate, 0.3 ml of the 15% suspension of sensitized erythrocytes was added along with 0.1 ml of undiluted guinea pig com- plement to 2.6 ml of 1.51o agarose (Indubiose A-37) in PBS, pH 7.2, held at around 40 °C in a water- bath and poured into an immunoplate° kept on a levelling table. The gels were allowed to harden for about 30 minutes after which wells of 2.0 mm diameter were punched in the agarose, and 5 1l of inactivated serum was added to each well. The plates were covered with their lids, transferred to a humid box, and incubated at 37 °C for 18-20 hours. The diameters of the zones of haemolysis were measured with the aid of a micrometer eyepiece calibrated in 0.1 mm divisions. HI, CF and neutralization tests. These were per- formed according to standard procedures employed in our Institute (15-17). Criteria for positive sera in the SRH test A haemolytic zone of > 2.5 mm diameter was accepted as a positive reaction (13). Paired sera giving the following reactions were considered to be diagnostically positive in SRH: (a) Seroconversions, where the acute serum sample gave no haemolytic zone but the convalescent serum gave a haemolytic zone of >3.0 mm diameter. (b) Increase or decrease of > 2.0 mm between the diameters of the zones produced by the acute and convalescent sera (5, 6). (c) Both the acute and convalescent serum samples gave similar haemolytic zones of >6.0 mm diameter. Single convalescent serum samples giving haemo- lytic zones of > 6.0 mm diameter were considered positive. Sera reacting with more than one antigen were considered positive for that antigen with which the haemolytic zonal diameter was at least 2.0 mm greater than that produced by other reacting anti- gens. Sera giving haemolytic zones on control plates were excluded. Criteria for positive sera in the HI test Those serum pairs which showed conversions and those which showed a fourfold or more increase or decrease in titres were considered to be diagnostically positive. Also those pairs in which both the acute and convalescent sera had similar high titres of > 1:160 were considered to be diagnostically positive. Single convalescent serum samples with titres of ) 1:160 a Hyland Division, Travenol Laboratories, Costa Mesa, CA, USA. were also considered positive. RESULTS The homologous and heterologous reactions of JE, WN and DEN-2 viruses and their immune sera in the SRH and HI tests are given in Table 1. Haemolytic zones were consistently larger in the homologous reactions (10-12.5 mm), compared with the hetero- logous reactions (0-7.5 mm); with the DEN-2 anti- gen, haemolysis was observed only with the homo- logous antiserum. In the HI test employing the same sera, all the antigens reacted not only with the homo- logous immune sera, but also with the other two at lower titres. The 1977 encephalitis epidemic Four paired serum samples were tested in the SRH, HI, CF and neutralization tests. None of the sera produced nonspecific zones of haemolysis on the control plates containing unsensitized erythrocytes or erythrocytes treated with normal mouse brain antigen. Three pairs gave a monotypic response for JE virus in the SRH test, with a decrease in zone diameter of 2-4 mm between the first and second specimens. These three pairs reacted monospecifically for JE virus in the CF test, and they were also positive in the neutralization test with JE virus. However, in the HI test, all the three pairs gave positive JE +WN + DEN-2 reactions; the decrease in the zone diameters in these three pairs was accompanied by a decrease in the HI titres also. The fourth pair showed conversion in both SRH and HI tests for WN virus only, and the neutralization test with JE virus was negative; low levels of CF antibodies (1:4) for WN virus could be detected in the second serum sample. The reactions of the 22 single convalescent sera in the SRH test were: 10, monotypic JE; 2, monotypic WN; 9, JE + WN; and 1, negative. None of the sera produced nonspecific zones of haemolysis on the con- trol plates. In all the 10 monotypic JE cases, the CF reaction was also monotypic for JE: 8 of these 10 tested in the neutralization test with JE virus were also positive. The reactions of these 10 sera in HI were: 6, monotypic JE; 3, JE + WN; and 1, JE +WN + DEN-2. The two sera which gave a monotypic WN reaction in SRH failed to react in both CF and HI. Both were negative in the neutralization test with JE virus. Eight of the 9 sera which gave a JE +WN reaction in SRH gave a monotypic JE reaction when tested in CF. All the 9 sera with a JE +WN reaction in SRH were positive in the neutralization test with JE virus. The reaction in HI of these 9 sera were: 3, mono- typic JE; 3, JE + WN; and 3, JE + WN + DEN-2. In all, 19 out of the 22 single convalescent sera reacted in SRH with JE antigen (mean zone diameter, SINGLE-RADIAL-HAEMOLYSIS TEST FOR DIAGNOSING FLAVIVIRUS INFECTIONS Table 1. Homologous and heterologous reactions of JE, WN and DEN-2 viruses with mouse immune sera in SRH and Hi tests Immune sera against Normal mouse serum Antigens JE P20778 WN G22886 DEN-2 P23085 (control) JE P20778 10.00 3.75 0 0 (64o)b (320) (40) (< 10) WN G22886 7.5 12.5 4.25 0 (320) (1280) (40) (< 10) DEN-2 P23085 0 0 10.0 0 (40) (40) (160) (< 10) Normal mouse brain (control) 0 0 0 0 a SRH zone diameters expressed in mm. b Figures in brackets indicate Hi titres expressed as reciprocals of serum dilutions. 8.78 mm) and 11 sera reacted with WN antigen (mean zone diameter, 5.45 mm). In HI, 19 sera were positive for JE virus with a geometric mean (GM) titre of 165.95; 10 were positive for WN and 4 were positive for DEN-2 with GM titres of 80.0 and 33.64, respectively. The 1979 encephalitis epidemic Paired sera. The results of the SRH and HI tests on the 171 paired sera obtained during the 1979 outbreak of encephalitis in Kolar district are as follows. The SRH test results in the case of 138 patients (80.7%) were consistent with a recent flavivirus infection. In 13 cases antibodies to one or more of the antigens were detected but were not at the diagnostic level; the remaining 20 cases were negative in the SRH test. Fifty-seven of the 72 cross-reactive cases were diagnosed as JE on account of the zone diameters to JE virus being larger by > 2.0 mm than the diameters of the zones for the other reacting antigens; the remaining 15 cross reactions were unresolvable. In the HI test, 115 pairs (67.3%) fulfilled the diagnostic criteria for one or more of the antigens employed in the test. In 20 pairs, antibodies to one or more of the antigens were present, but only at the non-diagnostic level; 36 pairs were negative. Out of the 56 cross-reactive cases in HI, 43 were compatible with a diagnosis of JE on account of the HI titre being at least fourfold higher for JE virus. The remaining 13 cross reactions were unresolvable. Single convalescent sera. Of the 30 single con- valescent sera tested in SRH, one gave a nonspecific reaction on the control plates and was excluded. The reactions of the remaining 29 sera in the SRH test were: 13, monotypic JE; 14, JE + WN; 1, JE +WN + DEN-2; and 1, negative. In all the 15 cross-reactive cases, the zone diameter for JE virus was > 2.0 mm bigger than that of the WN and DEN-2 zones. How- ever, only 19 of the 29 single convalescent sera ful- filled the diagnostic criteria for JE virus, i.e., with zones of >6.0 mm diameter. In the HI test, out of the 29 single convalescent sera, 9 gave a monotypic JE response; 15 gave a JE + WN response; 3 gave a JE + WN + DEN-2 response; and 2 were negative. On the whole, only 13 sera fulfilled the diagnostic criteria for JE, i.e., an HI titre of > 160 and fourfold higher titre for JE in cases of cross-reactive sera. Confirmation of the specificity of the SRH reaction There were a few sera which produced haemolytic zones of varying diameters, but were negative in the HI test. Attempts to determine whether such SRH reactions in the absence of HI reactions were specific or not were undertaken by means of the neutral- ization tests. Seventy-one sera giving haemolysis with JE virus, but negative in HI, were tested for the presence of JE virus neutralizing antibodies; 48 (67.6%o) of these were found to be positive, which proved that more than two-thirds of the SRH-positive but HI-negative sera contained JE virus neutralizing antibodies. The remaining one-third of the sera might have had too low a level of antibodies to become positive in the neutralization test. All these sera except one produced small zones of < 5.0 mm (mean zone diameter, 3.21 mm), indicating low levels of antibody. Also, the neutralization test done by the intracerebral route may not have been as sensitive as the intraperitoneal neutralization test in infant mice. Another 45 sera selected at random were also tested for the presence of JE virus neutralizing antibodies and 38 were positive. When the results of the neutralization tests were correlated with those of the SRH and HI tests, it was found that 25 sera were positive in all three tests and five sera were negative in 737 S. GEORGE & K. PAVRI all of them. Thus, full agreement between the results of the SRH, HI and neutralization tests was found in 30 cases (66.6%). Twelve sera were positive in the SRH and neutralization tests but negative in HI, whereas one serum was positive in SRH and HI but negative in the neutralization test. One serum was positive in SRH only and another in the neutral- ization test only. Perfect agreement between the results of any two or all the three tests was found in 43 cases (95.5%). Correlation ofSRH and HI test results For this, all the 392 serum samples collected from the 171 cases of encephalitis during the 1979 epidemic in Kolar district were investigated. When the SRH and HI test results of these sera were correlated, it was found that relatively significantly more positive results were obtained by the SRH test than the HI test for both JE virus (X2 = 62.1 1, P< 0.05) and WN virus (x2 = 17.88, P< 0.05) (Tables 2 and 3). The individual haemolytic zone diameters of all the sera positive in SRH were correlated with their individual HI titres in the case of JE and WN antigens (Fig. 1 and 2). There was good correlation between the two methods of antibody measurement, although there was some scatter of zone diameters for each HI titre. For JE and WN viruses, the mean zone diameters of the patients' sera were regressed on the log2 HI titres and a linear relationship was found between the two. The regression coefficients were: for JE virus, r=0.945 (P<0.01) and for WN virus, r=0.909 (P<0.05). The observed mean zone diameters at each HI titre along with the expected values arrived at statistically for JE and WN antigens (260 and 111 positive sera, respectively) also showed good agreement. DISCUSSION Since flaviviruses such as Japanese encephalitis, West Nile and dengue viruses are of considerable public health importance in India, the development of a simple, sensitive and type-specific test for the Table 2. Correlation of SRH and HI test results with JE virus SRH test Positive Negative Total HI test: Positive 175 (44.6)a 8 (2.0) 183 (46.7) Negative 85 (21.7) 124 (31.6) 209 (53.3) Total 260 (66.3) 132 (33.7) 392 a Figures in parentheses indicate percentages. Table 3. Correlation of SRH and HI test results with WN virus SRH test Positive Negative Total HI test: Positive 85 (21.7)a 68 (17.3) 153 (39.0) Negative 26 (6.6) 213 (54.3) 239 (60.9) Total 111 (28.3) 281 (71.7) 392 a Figures in parentheses indicate percentages. rapid diagnosis of these diseases is extremely important. Conventional serological tests like HI, CF and the neutralization test have well-known limitations. The single-radial-haemolysis test is simple, sensitive, type-specific, and rapid and there- fore its usefulness in the diagnosis of flavivirus infections was determined. In the present study, the SRH test was applied successfully for the determination of the etiology of two outbreaks of encephalitis which occurred in India. During the 1979 outbreak, 138 out of the 171 cases (80.7%?) where paired sera were available could be diagnosed by the SRH test as due to flavivirus infection, compared with 115 (67.3%) diagnosed by the HI test. Of the 138 positive cases in SRH, 120 could clearly be identified as JE virus infections and 3 asWN virus infections. By the HI test, only 100 could be clearly diagnosed as JE virus infections and 2 as WN virus infections. The number of sera reacting with DEN-2 antigen in both SRH and HI tests was small and in no case did DEN-2 appear to be etiologically involved. Generally, primary JE cases are marked by the absence of a DEN-2 response. However, cross reaction with the closely related WN virus is known to occur. The reactions in both SRH and HI tests were mostly with the JE and WN antigens. The antibody level and the number of positives to WN virus were lower than those to JE virus. For example, the geometric mean HI titre for WN virus was only half (29.4) of that for JE virus (58.3). The mean zone diameter was also smaller for WN virus (5.4 mm), compared with that for JE virus (6.05 mm). Among the encephalitis cases in 1977, JE virus could be incriminated by SRH as the causative agent in 3 out of the 4 cases where paired sera were available and in 19 out of 22 cases where only single con- valescent sera were available; in one case, WN virus was the probable etiologic agent although neutral- ization tests could not be carried out with WN virus owing to insufficient quantity of the serum. Com- pared with this, only 2 paired sera and 13 single sera fulfilled the diagnostic criteria for JE virus by HI. 738 SINGLE-RADIAL-HAEMOLYSIS TEST FOR DIAGNOSING FLAVIVIRUS INFECTIONS 14.0 11.5 10.5 9.5 I EI 2 8.5 7.5 6.5 5.5 4.5 3.5 2.5 <10 10 20 40 80 160 320 640 1280 HI t t re E .L c 0 (n I Fig. 1. Relation between HI titres and SRH haemolytic zone diameters obtained on 260 sera positive in SRH for Japanese encephalitis virus. The comparatively high mean zone diameter of 8.78 mm for JE virus as against 5.45 mm for WN virus, and also the high geometric mean HI titre of 165.95 for JE virus as against 80.0 forWN virus, con- firm that the JE virus was the major etiologic agent in the 1977 outbreak. Our SRH and HI results also show that the 1979 epidemic of encephalitis was caused by JE virus, the WN virus being responsible for only two cases. The serological diagnosis of Japanese encephalitis and other flavivirus infections is often difficult owing to cross reactions in all the three conventional tests (HI, CF and neutralization), particularly in areas where more than one member of the group is present. In such situations, the SRH test would be advan- tageous as it was found to be less cross-reactive in the present study. Several workers have reported type specificity as one of the main advantages of the SRH test with arboviruses. Odelola (13) found no cross reaction between WN and yellow fever viruses in SRH, whereas they cross-reacted in HI. Gaidamovich & Melnikova (11) found the SRH test useful in the diagnosis of laboratory infection with Venezuelan equine encephalomyelitis virus as well as for sero- 14.0jF 12.5 11.5 10.5 9.5 8.5 7.5 6.5 5.5 4.5 3.5 - A2 2.5 A <10 10 20 40 80 160 320 640 1280 HI t t r e Fig. 2. Relation between HI titres and SRH haemolytic zone diameters obtained on 111 sera positive in SRH for West Nile virus. surveys. Gaidamovich et al. (12) employed this tech- nique for the diagnosis of dengue fever in Bangla- desh. They found no cross reaction between DEN-2 and JE viruses in SRH whereas it was present in HI. Duca et al. (14) found the SRH test simple and more specific than the classical haemagglutination inhib- ition test in the quantitative assay of specific antibody to WN and Sindbis viruses. They reported that dengue immune human and rat sera, which reacted in HI with the WN virus, were consistently negative in the SRH test with WN virus. Some advantages which make the SRH test most suitable both for serodiagnosis and for serosurveil- lance of Japanese encephalitis and other flavivirus infections are: (i) the test is simple and because it is unaffected by nonspecific inhibitors, acetone extrac- tion of sera is not necessary; (ii) since crude mouse brain antigens suffice, this results in considerable saving; (iii) sera can be tested undiluted and only microlitre volumes are required; (iv) no costly equip- ment is required. The simplicity, specificity, sensi- tivity and economy of the SRH test render it ideal for the detection of arboviral antibodies, particularly in field areas in India and other developing countries where JE, WN and dengue viruses are endemic. * . _ A a l A *- A - t. m Aa a -A V a AIt -iin A a a ma :0 "~~~~~~~~ 739 740 S. GEORGE & K. PAVRI RtSUMt L'EPREUVE D'HEMOLYSE RADIALE SIMPLE DANS LE DIAGNOSTIC DES INFECTIONS A FLAVIVIRUS, EN PARTICULIER L'ENCEPHALITE JAPONAISE En Inde, les flavivirus tels que ceux de 1'encephalite japo- naise, West Nile et de la dengue jouent un role considerable en sante publique. Les epreuves classiques destinees a diagnostiquer ces maladies, par exemple l'inhibition de l'hemagglutination (IH), la fixation du complement (FC) et la neutralisation sont laborieuses, longues a executer et couteuses. En outre, il n'est pas toujours possible de faire un diagnostic specifique de type en raison des reactions croisees observees dans ces epreuves. C'est pourquoi on a essaye d'appliquer l'epreuve d'hemolyse radiale simple au diagnos- tic de ces infections flavivirales. Pour cette epreuve, on a sensibilise des erythrocytes de poulet avec des virus de l'encephalite japonaise (EJ), West Nile (WN) et de la dengue (DEN-2) et on les a incorpores dans de l'agarose (a 1,5% dans du solute salin tamponne au phosphate) en meme temps que du complement. Des quan- tites de 5 1,1 de serum inactive de malades etaient introduites dans des cupules d'un diametre de 2 mm pratiqu&es dans le gel. Les plaques etaient incubees a 37 °C jusqu'au lendemain. Des zones d'hemolyse se sont produites autour des cupules renfermant des serums qui contenaient des anticorps. Les diametres de ces zones ont et mesures au moyen d'un oculaire de micrometre etalonne en 0,1 mm. Les echantillons eprouves comprenaient 4 paires de serum et 22 serums uniques de convalescents recueillis pendant l'epidemie d'encephalite qui a sevi en 1977 dans le district de Kolar, Etat de Karnataka, en Inde, ainsi que 171 paires de serum et 30 serums uniques de convalescents recueillis pendant l'epidemie d'encephalite de 1979 dans le meme district. Sur les 171 cas pour lesquels on disposait de paires de serum en 1979, 138 (80,7%) ont pu etre attribues a une infection a flavivirus au moyen de l'epreuve d'hemolyse radiale simple, contre 115 cas (67,3%) diagnostiques par 1'epreuve d'inhibition de l'hemagglutination. Sur les 138 cas ci-dessus, 120 ont et clairement identifies comme des infections a virus EJ, 3 comme des infections a virus WN et les 15 restants comme dus a des "flavivirus indifferenci6s". Sur les 120 cas diagnostiques comme EJ par I'hemolyse radiale simple, 63 ont donne une reaction monotypique alors que 57 presentaient des reactions croisees surtout avec WN. Sur les 115 cas diagnostiques au moyen de l'inhibition de l'hemagglutination, 100 etaient des EJ, 2 etaient dus a WN et 13 etaient des infections "a flavivirus indif- ferencies". Sur les 100 cas diagnostiques comme EJ par cette derniere methode, 57 ont donne une reaction mono- typique et 43 donnaient des reactions crois&es surtout avec WN. Au total, il a et possible d'identifier nettement l'agent etiologique dans 123 cas (72%) par l'epreuve d'hemolyse radiale simple contre 102 cas (60%) par l'epreuve d'inhibition de l'hemagglutination. En ce qui concerne l'epidemie d'encephalite de 1977, le virus EJ a pu etre incrimine comme agent etiologique par l'epreuve d'hemolyse radiale simple dans 3 des 4 cas pour lesquels on disposait d'une paire de serums et dans 19 des 22 cas pour lesquels on possedait seulement un serum unique de convalescent. Dans 1'epreuve d'inhibition de l'hemag- glutination, seuls 2 paires de serums et 13 serums uniques satisfaisaient aux crit&es diagnostiques de EJ. Dans un cas, les deux epreuves ont designe le virus West Nile comme agent etiologique. On a constate que l'epreuve d'hemolyse radiale simple offrait de nombreux avantages car elle est simple, sensible et donne lieu a moins de reactions croisees que la reaction d'inhibition de l'hemagglutination. REFERENCES 1. SCHILD, G. C. ET AL. Bulletin of the World Health Organization, 52: 43-50 (1975). 2. RUSSELL, S. M. ET AL. Journal ofgeneral virology, 27: 1-10 (1975). 3. STRANNEGARD, 0. ET AL. Journal of clinical micro- biology, 1: 491-494 (1975). 4. CLARKE, M. ET AL. Journal of hygiene (Cambridge), 79: 355-364 (1977). 5. GRILLNER, I. ET AL. Journal of clinical microbiology, 4:11-15(1976). 6. VAANAANEN, P. ET AL. Archives of virology, 52: 91-99 (1976). 7. HEIRHOLZER, J. G. & TANNOCK, G. A. Journal of clinical microbiology, 5: 613-620 (1977). 8. RISKI, H. ET AL. Scandinavian journal of infectious diseases, 9: 75-77 (1977). 9. PROBERT, M. & RUSSEL, S. M. Journal of clinical microbiology, 2: 157-161 (1975). 10. WESSLEN, L. Scandinavian journal of infectious diseases, 10: 15-20 (1978). 11. GAIDAMOVICH, S. Y. & MELNIKOVA, E. E. Inter- virology, 13: 16-20 (1980). 12. GAIDAMOVICH, S. Y. ET AL. Acta virologica, 24: 153 (1980). 13. ODELOLA, H. A. Archives of virology, 60: 325-328 (1979). 14. DUCA, M. ET AL. Bulletin of the World Health Organization, 57: 937-942 (1979). 15. CLARKE, D. H. & CASALS, J. American journal of tropical medicine and hygiene, 7: 561-573 (1958). 16. PAVRI, K. M. ET AL. Indian journal ofmedical research, 50: 153-161 (1962). 17. PRASAD, S. R. ET AL. Indian journal of medical research, 75: 1-6 (1982).