Antigenic variation in current influenza A viruses: evidence for a high frequency of antigenic `drift' for the Hong Kong virus

Bull. Org. mond. Sante 11974, 51, 1-11Bull. Wid Hith Org. Antigenic variation in current influenza A viruses: evidence for a high frequency of antigenic 'drift' for the Hong Kong virus G. C. SCHILD,1 J. S. OXFORD,1 W. R. DOWDLE,2 M. COLEMAN,2 M. S. PEREIRA,3 & P. CHAKRAVERTY 3 A new antigenic variant of the Hong Kong (H3N2) subtype of influenzavirus type A is described. The variant, A/Port Chalmers/1/73 (H3N2), was first isolated in Australasia in the autumn of 1973 and subsequently became the predominant influenza A variant in most areas ofthe world, replacing the previously prevalent strain AlEngland/42/72 (H3N2). The 1973 variant shows antigenic differences from former Hong Kong variants in both haemagglutinin and neuraminidase antigens. The application of immuno-double-diffusion tests and single-radial-diffusion tests in the antigenic analysis ofnew variants of the influen- zavirus is also described. It is emphasized that since new variants of the Hong Kong virus have appeared in the successive years 1971, 1972, and 1973, the annual frequency of antigenic " drift " for the Hong Kong virus is higher than was recorded for the " Asian " influenzavirus (H2N2) in the first 5 years of the latter's prevalence from 1957 to 1962, during which period little antigenic variation occurred. The influenza A virus possesses two antigenically variable glycoproteins located at the virus surface, the haemagglutinin (HA) and the neuraminidase (NA). It is well established that antibody to these two antigens is associated with immunity to influenza. The current subtype of influenza A virus, which first appeared 6 years ago with the isolation in mid-1968 of the A/Hong Kong/1/68 (H3N2) virus (5), contained haemagglutinin of a subtype (H3) distinct from that of the previously prevalent subtype (H2) and neuraminidase of the same subtype (N2) as that of the 'Asian ' strains which had circulated in the years 1957-67. In the first 2 years of its prevalence the Hong Kong virus produced severe epidemics throughout the world: in the United Kingdom and many other countries not before the winter of 1969-70, but in the USA in the first winter of its prevalence, 1968-69 (4). The earliest evidence of antigenic variation of epidemio- logic importance in the Hong Kong virus was 1 WHO World Influenza Centre, National Institute for Medical Research, Mill Hill, London, NW7, England. 2 WHO Collaborating Centre for Reference and Research on Influenza, Centre for Disease Control, Atlanta, Georgia, USA. 3 Virus Reference Laboratory, Central Public Health Laboratory, Colindale Avenue, London, NW9, England. represented by the strain A/Hong Kong/107/71, which showed antigenic 'drift ' away from the 1968 strain in both its HA and NA antigens and which produced outbreaks in South-East Asia and the south Pacific area (13). This variant, however, failed to gain worldwide prevalence (11). Early in 1972, another variant, A/England/42/72, appeared which showed further antigenic variation in the HA antigen while the NA antigen was identical with that of the A/Hong Kong/107/71 strain (2, 11). We now describe a new antigenic variant of the Hong Kong virus, showing further antigenic drift in both its HA and NA antigens. The new variant was first detected in autumn 1973 (14) when it pro- duced widespread outbreaks in Australia and New Zealand. Subsequently, in the autumn and winter of 1973-74 it was isolated during outbreaks or from sporadic cases in the United Kingdom, then in several other countries in Europe, the Middle and Far Eastern and Pacific areas, and the North American and African continents. The outbreaks in most areas were of slight to moderate impact. During this period, isolates of A/England/ 42/72 became infrequent in most areas. In the United Kingdom the new variant appeared less than one year after the previous variant, A/England/ 3230 -1- 2 G. C. SCHILD ET AL. 42/72, had become prevalent. The present paper describes the antigenic characteristics of the new variant, preliminary studies of the distribution of antibody to the virus in the United Kingdom and the USA, and the immune response produced by currently available inactivated influenza vaccines. Circumstantial evidence is discussed which suggests that antigenic variants of the Hong Kong virus have appeared more frequently since this subtype first appeared in 1968 than variants of the 'Asian ' influenza A subtype during a comparable period after the latter's emergence. MATERIALS AND METHODS Haemagglutination-inhibition (HI) tests The HI tests were performed by standard methods (12). Sera were treated with receptor-destroying enzyme ( Vibrio cholerae filtrate) before use. Neuraminidase-inhibition (NI) tests The NI tests were carried out as described by Aymard-Henry et al. (1). The source of neuramini- dase was purified influenzavirus particles. Antisera against purified neuraminidase or purified viruses were used. Immuno-double-diffusion (IDD) tests The IDD tests in agar gels have been described previously (9). Purified, concentrated influenzavirus preparations containing 10-15 mg viral protein per ml were used as the antigen after disruption with detergent (sodium (methylamino)acetyl sulfate,a Geigy NL 97, 1 % final concentration) to liberate the diffusible antigens. Preparation of antisera to purified haemagglutinin and neuraminidase antigens Purified haemagglutinin was obtained by treatment of purified influenzavirus with a proteolytic enzyme, bromelain, to remove the haemagglutinin which was then purified by centrifugation on sucrose gradients (3). Neuraminidase antigens were separated from purified influenzavirus particles, disrupted by 1 % sodium dodecyl sulfate, at pH 8.9 by electro- phoresis on cellulose acetate strips (7). Antisera were collected from rabbits 4 weeks after a single intramuscular dose of purified antigen (approxim- ately 300 ,tg protein) together with Freund's complete adjuvant. When potent antiserum was needed, a further ' booster ' dose of antigen (300 pg) with a Sodium " sarcosyl " sulfate. adjuvant was given and the serum collected one week later. Single-radial-diffusion tests (SRDT) and SRD- absorption tests Antibodies to the influenza haemagglutinin and neuraminidase antigens were assayed in single- radial-diffusion tests in which preparations of intact, purified influenzaviruses were suspended in agarose gel at a standard (final) concentration of 0.2 mg viral protein per ml of gel (approximately 4 x 10' haemagglutinating units (HAU) per ml as described previously (10). Test sera in volumes of 10 p1l were added to 3-mm circular wells cut in the gel. In this simple two-component system antibody potency was expressed as the annulus area (mm2) of the opalescent zones produced by 10-pl volumes of antisera. Absorption tests were carried out by incubating mixtures of antiserum (20 pl) and purified virus (20 lI) for 3 h at room temperature. As a control, saline was used instead of virus. The residual unbound antibody in the mixtures was determined by adding 5-,l volumes of each mixture to the wells in single-radial-diffusion immunoplates and comparing the areas of the resulting opalescent zones with those produced by control (unabsorbed) sera. As a preliminary to the absorption studies, the minimal quantities of homologous virus required to remove all the antibody from the antisera to haemagglutinin and neuraminidase were determined by adding graded concentrations of virus (20 pl), ranging from 106 HAU/ml to 10' HAU/ml, to 20-Ml volumes of each antiserum. Throughout the absorption studies, the concentrations of homolo- gous virus used were calculated on the basis that the viral protein contained a dose 20% greater than the minimal absorbing dose of homologous virus for the antiserum under test. In comparative studies, the variant strains were used at the same protein concentration as the homologous virus. The amount of antibody absorbed by the variant strains was calculated as the percentage reduction in the areas of opalescence between control (unab- sorbed) and test sera. Virus strains Virus strains were from the collections of the WHO World Influenza Centre and the WHO Collaborating Centre for Influenza in the USA; except for the 1973 isolates, these strains have been described previously (11). Viruses were grown ANTIGENIC VARIATION IN INFLUENZA A VIRUSES 3 Table 1. Characterization of influenza A strains isolated in Australia, New Zealand, and the United Kingdom in October 1973 by haemagglutination-inhibition tests HI titre Virus strains postinfection ferret serum to to purified HA A/HK/1/68 A/H K/i 07/71 A/Eng/42/72 A/PC/1/73 b A/HK/i/68 A/PC/1/73 Reference strains: A/Hong Kong/1/68(H3N2) 7680a 960 7680 640 25600 3200 A/Hong Kong/5/72(H3N2)C 320 10240 640 1 920 1 200 1 200 A/England/42/72(H3N2) 1 920 320 15 360 3840 6400 6 400 Australian isolates: A/Perth/2/73 60 120 640 1 920 1 200 51 200 A/Perth/4/73 640 240 1 920 1 920 1 200 51 200 New Zealand isolates: A/Dunedin/4/73 320 120 1 280 1 920 1 200 A/Pt. Chalmers/1/73 480 160 1 280 1 920 1 200 51 200 United Kingdom isolates: A/England/929/73 640 320 2560 1 920 1 200 51 200 A/England/954/73 640 320 1 280 1 920 1 200 51 200 a Reciprocal Hi titre. b Postinfection ferretserato A/Dunedin/4/73 and A/England/954/73 gave HI reactions closelysimilartothoseof A/Port Chalmers/i/73 c The strain A/Hong Kong/5/72 is identical with A/Hong Kong/107/71. in the allantoic cavity and their purification from RESULTS egg allantoic fluid was by standard procedures (11). Stocks of virus for reference purposes were re- Evidence of further antigenic drift in the HA and isolated at their terminal dilution endpoints in NA antigens of current influenza A viruses embryonated eggs in order to eliminate the possi- Tables 1 and 2 give the results of HI and NI bility of obtaining mixed populations. tests, which characterize the surface antigens Table 2. Characterization of influenza A strains by neuraminidase-inhibition tests Immune rabbit serum to Source strain of purified N2 (1957)- purified N2 (1968)- AEnld427 A/otCamr1 3neuraminidase neuraminidase derived neuraminidase derived A/England/42/72 A/Port Chalmers/i/73 from A/Singapore/i /57 from A/Hong Kong/i /68 A/Singapore/1/57 1 oooa 50 10 10 A/Hong Kong/i /68 300 1250 160 20 A/England/42/72 100 450 640 50 A/Port Chalmers/1/73 10 14 100 200 A/England/929/73 10 13 100 200 A/England/954/73 10 12 140 200 a Serum dilution inhibiting viral neuraminidase activity by 50 %. G. C. SCHILD ET AL. (HA and NA) of the influenza A viruses isolated in 1973 in Australia and the United Kingdom and which compare them with the antigens of the formerly prevalent variants, A/Hong Kong/1/68, A/Hong Kong/107/71, and A/England/42/72. For these tests, postinfection ferret sera and hyperim- mune rabbit sera, which were prepared against purified haemagglutinin and neuraminidase antigens, were employed. The results provided clear evidence that the HA and NA antigens of the 1973 isolates differed from those of the previously prevalent variant, A/England/42/72. The 1973 isolates from New Zealand, western Australia, and the United Kingdom appeared to form an antigenically closely related group of viruses. The isolate A/Port Chal- mers/l/73 was taken as being typical of these isolates and is proposed as the reference strain for the 1973 variant. In order to confirm that the HA and NA antigens of the 1973 variants belonged to the same subtype, H3 and N2, as the antigens of the Hong Kong virus, immuno-double-diffusion (IDD) tests were carried out with hyperimmune sera to the purified HA (H3) and NA (N2) antigens (see Fig. 1). Anti- serum to purified HA of A/Hong Kong/1/68 and A/England/42/72 viruses gave precipitin lines (Fig. la, lb) with all of the test viruses isolated from 1968 to 1973. However, the precipitin lines between different influenzavirus variants in adjacent wells did not show complete identity and 'spurs ' were Fig. 1. Immuno-double-diffusion tests: 1 a (top left), 1 b (top right), 1 c (bottom left), and 1 d (bottom right). Centre wells contain undiluted rabbit antisera. Peripheral wells contain antigens (virus concentrates treated with approx. 1 % sodium sarcosyl sulfate to disrupt the virus particles). Wells marked A and B contain antiserum to purified haemagglutinin of A/Hong/Kong/I /68 and A/England/42/72, respectively, and C and D contain antiserum to purified neuraminidase of A/Singapore/I /57 and A/Hong Kong/l /68, respectively. Wells marked 57,68, 72, and 73 contain virus concentrates of A/Singapore/l /57 (H2N2), A/Hong Kong/l /68 (H3N2), A/England/42/72 (H3N2), and A/Port Chalmers/1/73 viruses, respectively. Spur formation between precipitin lines produced by antigens in adjacent wells indicates patterns of changing antigenic determinants (antigenic drift) in the haemagglutinin or neuraminidase antigens. However, the formation of common (shared) precipitin reactions between all test viruses is evidence that the haemagglutinin and neuraminidase antigens of all test viruses are of common subtypes, namely H3 and N2. 4 ANTIGENIC VARIATION IN INFLUENZA A VIRUSES detected between lines given by successive variants. The detection of such 'spurs ' suggested that homologous HA antigen possessed antigenic deter- minants, which were not shared by the strains tested in the adjacent wells, although the observed cross-reactions showed that other antigenic deter- minants were shared by all the test strains isolated from 1968 to 1973. Similarly (but not shown in the figures), antiserum to the purified HA of A/Port Chalmers/1/73 virus gave reactions with all the test viruses used (A/Hong Kong/1/68, A/England/ 42/72, and A/Port Chalmers/1/73), and ' spur ' formation between these strains tested in adjacent wells was demonstrated. The pattern of precipitin reactions thus indicated that the HA of A/Port Chalmers/1/73 virus was of the same subtype (H3) as the prototype A/Hong Kong/1/68 virus. The antisera used in these studies were specific for HA of subtype H3 and failed to react with virus strains containing other subtypes of HA (HO or H2) found amongst human influenza A viruses. Cross-absorption experiments were performed to confirm that the haemagglutinins of A/Hong Kong/1/68 and A/Port Chalmers/1/73 viruses each contained antigenic determinants (strain specific determinants) that were not shared. Antiserum to the HA of A/Hong Kong/1/68 virus was absorbed by the addition of purified A/Port Chalmers/1/73 virus to remove antibodies to antigenic determinants shared by both strains (cross-reactive determinants). The pattern of reactions of the absorbed serum in IDD tests is shown in Fig. 2. No reaction was detected with the 1973 variant but distinct precipitin lines were found with the 1968, 1971, and 1972 variants. Similarly (but not shown in the figure), antiserum to the HA of Port Chalmers virus, after absorption with Hong Kong/68 virus, gave IDD reactions with the 1972 and 1973 variants but not with the 1968 and 1971 variants. The results of the absorption experiments, together with the pattern of spur formation shown in Fig. la and lb, indicated that the HA antigens of the 1968 and 1973 variants contained antigenic determinants that were not shared, and thus provided evidence that during antigenic ' drift ' certain antigenic determinants of the HA antigen underwent complete antigenic change while other determinants remained unaltered and provided the basis for the recognition of the antigen as H3. IDD studies were also performed with antisera to purified NA antigens. Antiserum against purified NA (N2) of A/Singapore/1/57 (H3N2) virus and A/Hong Kong/1/68 virus gave cross-reactions with all test strains of influenza A viruses successively prevalent in man from 1957 to 1973. ' Spur ' for- mation between the precipitin reactions given by influenzavirus strains tested in adjacent wells (Fig. Ic and Id) indicated progressive antigenic drift in the NA and, as in the case of the HA antigens, the findings suggested that antigenic drift in the NA antigen (N2) involved complete change in some determinants while other determinants re- mained antigenically stable from 1957 to 1973. The anti-neuraminidase sera used in these studies were specific to the antigens of subtype N2 and failed to react with NA antigens of other subtypes. The test strains were also compared by single- radial-diffusion. This technique enables a high degree of accuracy and reproducibility for antibody assays and is thus of value for the quantification of antigenic differences between influenza variants. The comparative SRDT reactions of sera on immunoplates containing A/Hong Kong/1/68, A/ England/42/72, and A/Port Chalmers/1/73 are Fig. 2. Immuno-double-diffusion reactions. The centre well E contains a rabbit antiserum to the purified HA of A/Hong Kong/I/68 virus after absorption with purified concentrated A/PC/i/73 virus in order to obtain a strain specific population of anti-HA antibody. This population of antibodies is directed against antigenically variable determinants of HA, which are not possessed by the 1973 variant but precipitin lines are detected against the 1972, 1971, and 1968 viruses. 5 G. C. SCHILD ET AL. VIRUS STRAINS A|HK/1168 - A/ENG/42/72-0 NPC/1l73-f ANTISERA H3 N2 (1968) (1968) A/HK/68 AAEN(72 A$IC73 I I I l l Fig. 3. Studies of antigenic variation employing single-radial-diffusion tests with immune anti-HA and anti-NA sera and postinfection ferret sera. Antigens are marked A/H K/1 /68, A/England/42/72, and A/PC/1/73 and contain the respective intact, concentrated, and purified viruses at a standard concentration of 0.2 mg viral protein per ml agarose gel. Antisera H3(1968) and N2(1968) represent vertical rows of wells containing rabbit serum to purified HA and NA, respectively of A/Hong Kong/i /68 virus. A/H K/68, A/Eng./72, and A/PC/73 represent vertical rows of wells containing postinfection ferret serum to the appropriate virus strain. The annulus areas of the opalescent zones indicate the relative numbers of antibody molecules reacting on each immunoplate. Table 3. Serological cross-reactions of A/Hong Kong/1/68(H3N2) with its variants in single-radial-immuno- diffusion tests Areas of opalescence per 10 ju1 of serum (mm2) antisera to isolated psifcinfre eaat Purified virus A/Hong Kong/1/68 antigens postinfection feret sera a to incorporated in anti-pure H3 anti-pure N2 A/Port Chalmera'immunoplate haemagglutinin neuraminidase A/Hong Kong/i /68 A/England/42/72 ,1173m, (chick origin) (rabbit origin) area % homol. area % homol. area % homol. area % homol. area % homol.reaction reaction reaction reaction reaction A/Hong Kong/1/68(H3N2) 63.8 100 130 100 43.2 100 30.3 66 5.5 8.6 A/England/42/72(H3N2) 34.8 55 56.5 43 13.4 31 45.7 100 37.1 58 A/Port Chalmers/1/73(H3N2) 22.2 34 16.7 13 6.1 14 33.6 73 63.8 100 a Zones given by postinfection sera reflect essentially antibody to haemagglutinin, since such sera are found to contain high levels of antibody to haemagglutinin and low amounts of antibody to neuraminidase. 6 ANTIGENIC VARIATION IN INFLUENZA A VIRUSES Table 4. Antigenic comparison of A/Hong Kong/1/68(H3N2) and its variants. Quantitative antibody absorption tests employing single-radial-diffusion (SRD) for antibody assay a Percentage reduction in zone area after absorption Antiserum absorbing virus A/Hong Kong/1i/68 A/Hong Kong/i 07/71 A/England/42/72 A/Port Chalmers/1/73 Hyperimmune antiserum to A/Hong Kong/1i/68 antigens: anti-haemagglutinin (H3) iooa 32.5 34.3 21.8 anti-neuraminidase (N2) 100 19.0 18.1 2.7 Postinfection ferret serum to: A/Hong Kong/i /68 virus 100 40.5 51.0 19.2 a Assays based on zone area (mm2) of opalescence in immunoplates containing 0.2 mg purified A/Hong Kong/i /68 virus per ml gel. shown in Fig. 3 and measurements of the relative areas of opalescent zones are given in Table 3. Antisera to the purified haemagglutinin and the purified neuraminidase of A/Hong Kong/l/68 virus gave smaller opalescent zones in tests with A/ England/42/72 and A/Port Chalmers/l/73 viruses than with the homologous virus. The zone areas in tests with A/Port Chalmers/l/73 were 34% and 13% of the zones with homologous virus for the anti-haemagglutinin and anti-neuraminidase sera, respectively, and they provide confirmatory evidence of the considerable degree of antigenic variation in the HA and NA antigens between 1968 and 1973. The successive variants of A/Hong Kong/68 virus were also tested for their ability to absorb antibody for the homologous A/Hong Kong/1/68 virus from anti-haemagglutinin and anti-neuraminidase sera (Table 4). The ability to absorb antibody from these sera decreased successively for the 1972 and 1973 isolates. While absorption with A/Port Chalmers/l/73 virus reduced the zone area given by anti-HA antibody for A/Hong Kong/1/68 virus by 21.8 %, the same virus preparation reduced the zone size of anti-NA antibody by only 2.7%. Since increasing the dose of absorbing viruses by up to fivefold did not further reduce the amount Table 5. Frequency and level of haemagglutination-inhibiting antibody in normal serum samples collected from young adults in the United Kingdom and the USA, October-November 1973 Cumulative percentage of sera with antibody at stated values to HI antibody titre A/Hong Kong/i /68 A/England/42/72 A/Port Chalmers/1/73(H3N2) (H-3N42) U.K. USA U.K. USA U.K. USA 1: < 20 19 8 38 39 72 67 1: >20 81 92 62 61 28 33 1: > 40 54 88 50 41 17 20 1:>80 34 72 40 26 7 11 1 :>160 17 51 29 13 1 3 1 :>320 3 27 12 8 0 1 1 :>640 2 14 6 0 0 0 1:>1280 0 6 2 0 0 0 geometric mean titres 65 108 47 23 17 12 7 G. C. SCHILD ET AL. of residual antibody in the anti-HA and anti-NA sera, it seems that the residual antibody represents CIOX0 0 a population of molecules not capable of reacting with the particles of absorbing virus. The absorbed o > A\ antibody represents antibody specific to the antigenic .C determinants of the HA (or NA) which are cross- 2 0 _ O oNe o reactive between strains, while the residual antibody mI o4,, is specific to the antigenically variable determinants Zi _ E of the HA (or NA). These results of absorptionZ studies also confirm that extensive antigenic variation Ie>so o w o N in the HA and NA antigens occurred between E A\ 1968 and 1973, and furthermore suggest that the 0@ Xo X > o O ~ degree of change in the NA antigen was greater 0, A\ than that for the HA antigen in this period. E _ C 0 to < Antibody in human sera to the A/Port Chalmersl o .s A\ N O X 0 1/73 variant and response to vaccination2 ~~~r_N O N A\ The frequencies and titres of HI antibody in sera FCI O N 0 04 from young adults in the United Kingdom and the ( NA N USA in November 1973 are shown in Table 5. 00 0 v oCD Only 30% of such sera contained detectable HI z antibodies (titres of 1:20 or greater) to A/Port I a . Chalmers/1/73 virus whereas 61% of sera had HI N o' ¢= at ae '|' antibody to A/England/42/72 virus. Thus, a large N )N , *o 0 0 C proportion of the sera tested lacked antibody to the new variant suggesting that a high proportion~~~~~~~~~~~ A (d of the populations tested may be potentially suscep- -O @ cgltibleto infection.LU .> *_ co. The inactivated influenza vaccines available for o XEzco. E C routine use at the present time contain A/England/ E X tEE 42/72 virus. It was of importance to establish o whether such vaccine could induce antibody to 0 C. ai 0 0 ° the 1973 variant. Table 6 shows the serum antibody A\ > levels in persons in England before and after C N .0 .'C O O >_ 0 ° receiving vaccine containing 600 i.u. of A/England/tm0 leN 0 _ 4/7 (RC0sz> c A\ _ O X42/72 (MRC7 strain), and in the USA before and .' < nc- after receiving a commercial vaccine containingCCQ I A\ |~ cs 700 CCA units of the same strain. In England, C. >N< 89% of vaccinees developed a fourfold or greater .c b A\ un O _ oo o X Z HI antibody response to both the homologous 4 0 zo . .- virus and to the new variant A/Port Chalmers/1/73. mM A\ co am X ;°o 3 Q In the USA, the figures were 86% and 76%, respect- E.52 0 10(N 0 o ively. However, the geometric mean HI antibody su U.v 8 X titre after vaccination was significantly lower for qo< > 8 the A/Port Chalmers/1/73 variant than for A/ 00~~~~~~~~~~~20.2E E E England/42/72. The geometric mean HI titres for > X @ @ @) ,58 ' both A/England/42/72 and A/Port Chalmers/.1/73 .- M @ @ @ .: > Q > after vaccination were lower in the USA (177 and .a) ES. g a. O. 90, respectively) than in England (980 and 240, 0. ( co respectively). The lower titre in the USA may be0D@ X oQ Q a characteristic of the geriatric population under study. 8 ANTIGENIC VARIATION IN INFLUENZA A VIRUSES Table 7. Antigenic cross-reactions of recent influenza A viruses as indicated by haemagglutination-inhibition tests Postinfection ferret sera to Virus strains reference strains new variants A/HK/1/68 A/HK/107/71 A/Eng./42/72 A/PC/1/73 A/Hann./61/73 A/P. Rico/1/74 A/Eng./635/74 A/Hong Kong/1/68 3840 480 3840 320 60 320 60 A/Hong Kong/107/71 160 7680 320 960 240 80 240 A/England/42/72 640 160 7680 1 280 320 160 480 A/Port Chalmers/1/73 160 320 1 280 2560 1 280 480 640 A/Hannover/61/73 < 40 240 1 280 2 560 2560 480 3840 A/Puerto Rico/1/74 120 120 160 640 120 1 920 160 A/England/635/74 40 120 160 640 960 160 3 840 More recently, 3 additional variants typified by A/Hannover/61/73, A/Puerto Rico/1/74, and A/ England/635/74 have been detected (Table 7). The A/Puerto Rico/1/74 strain became the predominant virus in the USA in the first half of 1974. Strains resembling A/England/635/74 and A/Hannover/ 61/73 made up a small but increasing proportion of the strains isolated in Europe between December 1973 and April 1974, the remainder of the European isolates being closely similar to the A/Port Chalmers/ 1/73 virus. These recent findings provide additional evidence of the high degree of antigenic variability of the current influenza A subtype. DISCUSSION The data presented in this paper provide evidence of further antigenic drift in both HA and NA surface antigens of the Hong Kong variant of influenza A virus. The strain selected as the proto- type of the most recent variant is A/Port Chalmers/ 1/73 (H3N2) although the earliest reported outbreaks, which were associated with the new variant, occurred in western Australia in August 1973 (Mackay- Scollay, personal communication). In addition to the conventional HI and NI tests, immuno-double-diffusion and single-radial-diffusion (together with antibody absorption) tests were found to be useful and sensitive procedures for the quantitation of the degree of antigenic variation among strains and they are recommended for adoption as routine procedures. In particular, double diffusion and radial diffusion tests have considerable advantages in assessing antigenic variation since they provide simple two-component test systems, which are independent of factors that could influence the results of HI tests, such as avidity of the test virus for erythrocytes or antibody and the effects of nonspecific inhibitors in serum samples. Evidence from all the test systems in the present study demonstrated antigenic drift in the HA and NA antigens of the Port Chalmers strain. The immunodiffusion results suggest that during antigenic drift some antigenic determinants of the antigen remain unchanged while others undergo a complete change. Since the antigenically variable HA deter- minants are critical sites for virus neutralization (Virelizier and Schild, unpublished observation), any changes in such determinants have important immunological consequences. A previous survey (2) indicated that antigenic drift in the NA antigen of the Hong Kong virus occurred as early as 1969, while the haemagglutinin antigen did not change before 1971. Studies of the NA antigen, as well as the HA antigen, are thus essential to a full knowledge of antigenic drift. Preliminary surveys of the frequency and titre of HI antibody to the new Port Chalmers variant in randomly collected sera in the United Kingdom and the USA indicated that only a small proportion of individuals (17% and 20% respectively) had antibody at titres considered to be commensurate with immunity to infection (1/40 or greater). Thus it appears that the majority of the population in these countries may be susceptible to infection with the 1973 variant. Vaccines containing the 9 G. C. SCHILD ET AL. A/England/42/72 (H3N2) variant elicited HI anti- body at titres of 1/40 or greater to the Port Chalmers virus in 83% of vaccinees in England and the USA. However, since the geometric mean HI titres in these sera, when measured with the Port Chalmers variant, were two- to fivefold lower than the titres detected using the homologous A/England/42/72 virus, it might be expected that a vaccine containing the 1972 variant would be less than optimum in protecting against the new variant. In anticipation of proposals that influenza vaccines for use in the coming season should contain the new Port Chalmers variant as the influenza A component, a recombinant virus (designated MRC1 1) with the growth capacity of A/PR8/34 (HONI) virus and the surface antigens (H3N2) of the Port Chalmers variant was prepared by standard procedures (8) for use as a potential vaccine strain. Preliminary data suggest that vaccines containing A/Port Chalmers/l/73 virus induce antibody to the three newer variants, A/ Hannover/61/73, A/Puerto Rico/1/74, and A/ England/635/74 as well. In contrast to the preceding influenza A subtype (' Asian ' influenzavirus), which was prevalent from 1957 to 1968 and which underwent limited antigenic drift in its HA antigen by 1964 and again in 1967, the current Hong Kong virus which first appeared in 1968 has so far produced (in the years 1971, 1972, and 1973) at least three variants differing in the HA or NA or both. The high frequency of antigenic variation in the Hong Kong virus is probably not attributable to increased surveillance or different serologic techniques. The same basic technique for the preliminary identification of strains, that is, HI tests with postinfection ferret or chicken sera, has been applied unchanged since 1957. Also, increased surveillance brought about by the intensification of the World Health Organization Influenza Programme might be expect- ed to result only in the earlier detection of variants, which would in any case be detected when they become prevalent. Evidence from a number of sources further suggests that the Hong Kong virus differs from the ' Asian ' subtype in a number of its biological properties. Thus the A/Hong Kong/ 1/68 virus differed from previous human influenza A subtypes in its ability to be readily isolated from clinical specimens in tissue culture and in the allantoic cavity of embryonated hens' eggs without prior adaptation (6). Further, the A/Hong Kong/ 1/68 virus is unique amongst human influenza A viruses in possessing a wide natural host range and has been isolated from a wide variety of species including swine, domestic fowl, cattle, and wild sea birds (reviewed in (15) and unpublished data). Whether the surface HA and NA antigens will continue to ' drift ' at an increasing rate or whether the frequent variation seen so far indicates the approaching end of the H3N2 period is a matter for speculation, but certainly this high degree of antigenic variability emphasizes the need for intensive surveillance of influenza in order to ensure the availability of vaccines containing the appro- priate HA and NA antigens and to ensure prompt detection of the next subtype of influenza A virus that might become prevalent in man. ACKNOWLEDGEMENTS The authors are indebted to the many laboratories submitting influenza isolations for investigation; to Dr Helen Mair, Dr J. Smith, and Dr J. Dadswell of the Public Health Laboratory Service, the United Kingdom; and to Dr G. Noble and Dr R. Rosenberg of the US Public Health Service for kindly supplying paired human sera from vaccinees. Robert Newman, Victor Law, and Rose Gonsalves of the N.I.M.R., and Avis Sisson, Donna Sasso, Wilma Yarbrough, and Carol Reed of the CDC provided excellent technical assistance. R,SUMt VARIATION ANTIGtNIQUE DES VIRUS GRIPPAUX A ACTUELS: PREUVE D'UNE FRtQUENCE ELEVtE DE e DRIVE s ANTIGENIQUE DU VIRUS DE HONG KONG Durant l'automne 1973, un nouveau variant antige- nique du sous-type Hong Kong de la grippe A a 6t6 identifi6 en Australasie; il est par la suite devenu la souche pr6dominante dans la plupart des regions du globe. Ce variant, A/Port Chalmers/1/73 (H3N2), est caracterse par une h6magglutinine et une neuraminidase 10 ANTIGENIC VARIATION IN INFLUENZA A VIRUSES 11 differentes de celles du virus A/England/42/72 (H3N2), apparu une annee auparavant. Le pr6sent article expose les caract6ristiques antig6- niques du nouveau variant et les m6thodes utilisees pour l'analyse antig6nique. On y fait aussi 6tat des r6sultats de recherches sero-epidemiologiques pr6liminaires men6es au Royaume-Uni et aux Etats-Unis d'Amerique et d'6tudes sur la r6ponse immunitaire A 1'6gard du nouveau virus chez des sujets vaccines par les vaccins antigrippaux actuellement disponibles. De nouveaux variants du virus de Hong Kong ont fait leur apparition en 1971, 1972 et 1973. Cette frequence de # d6rive * antig6nique est plus 6lev&e que celle enregis- tr6e pendant les cinq premieres ann6es (1957-1962) de predominance des virus de la grippe asiatique. REFERENCES 1. AYMARD-HENRY, M. ET AL. Bull. Wld Hith Org., 48: 199-202 (1973). 2. AYMARD-HENRY, M. ET AL. In: Symposia Series in Immunobiological Standardization, Basel, S. Karger, 1973, Vol. 20, pp. 28-38. 3. BRAND, C. M. & SKEHEL, J. J. Nature New Biology, 238: 145-147 (1972). 4. COCKBURN, W. C. ET AL. Bull. Wid Hlth Org., 41: 345-348 (1969). 5. COLEMAN, M. T. ET AL. Lancet, 2: 1384-1386 (1968). 6. COLEMAN, M. T. & DOWDLE, W. R. Bull. Wld. Hlth. Org., 41: 415-418 (1969). 7. LAVER, W. G. J. molec. Biol., 9: 109 (1964). 8. MCCAHON, D. & SCHILD, G. C. J. gen. Virol., 15: 73-77 (1972). 9. SCHILD, G. C. J. gen. Virol., 15: 99-103 (1972). 10. SCHILD, G. C. ET AL. J. gen. Virol., 16: 231-236 (1972). 11. SCHILD, G. C. ET AL. Bull. Wld Hlth Org., 48: 269-278 (1973). 12. WHO Expert Committee on Influenza. Wid Hlth Org. tech. Rep. Ser., No. 64, 1953. 13. Wkly epidem. Rec., 46 (50): 518 (1971). 14. Wkly epidem. Rec., 48 (44): 421 (1973). 15. Bull. Wid Hlth Org., 47: 439-542 (1972).