Study in Childhood MASENDYCZ,l JENNIFER S. LUND,'

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, June 1990, p /90/ $02.00/0 Copyright C 1990, Amerian Soiety for Mirobiology Vol. 28, No. 6 Comparison of Rotavirus Immunoglobulin A Coproonversion with Other Indies of Rotavirus Infetion in a Longitudinal Study in Childhood MASENDYCZ,l JENNIFER S. LUND,' BARBARA S. COULSON,l* KEITH GRIMWOOD,lt PAUL J. NORMAN MERMELSTEIN,2 RUTH F. BISHOP,' AND GRAEME L. BARNES' Departments of Gastroenterology' and Immunology,2 Royal Children's Hospital, Parkville 3052, Vitoria, Australia Reeived 7 November 1989/Aepted 14 February 1990 In order to determine the sensitivity and reliability of antirotaviral feal immunoglobulin A (IgA) as an indiator of rotavirus reinfetion, the antibody responses to rotavirus of 44 infants with severe rotavirus gastroenteritis reruited on admission to a hospital were studied. Fees were olleted daily during hospitalization and weekly thereafter, and sera were obtained every 4 months, for 6 to 32 months (median, 17 months). Antirotaviral IgG, IgA, and IgM were measured by enzyme immunoassay in all samples. Rotavirus antigen, rotavirus-neutralizing antibody, and total IgA were measured in fees. The results showed that use of an IgA index (ratio of speifi IgA to total IgA) was unneessary to identify opro-iga onversion to rotavirus. The other markers of rotavirus infetion tested showed a high level of preditive auray of oproonversion in rotavirus-neutralizing antibody. Copro-IgM, serum IgM, and virus in fees were insensitive measures of neutralizing antibody oproonversion. Seroonversion in IgG or IgA was deteted in 46% of neutralizing oproonversions. The most sensitive marker, present in 92% of neutralizing oproonversions, was antirotaviral feal IgA onversion. This orrelation of feal IgA with feal neutralizing antibody suggests that oproonversions in IgA represent true elevations in antirotaviral IgA with neutralizing apaity. A oproonversion in IgA appears to indiate genuine rotavirus infetion. Copro-IgA onversions in fees olleted weekly are likely to be more sensitive markers of rotavirus reinfetion than are seroonversion and virus detetion ombined in epidemiologial studies of aute diarrhea in hildren and in rotavirus vaine trials. Rotaviruses are a major ause of severe aute diarrhea in humans and animals. These viruses infet the mature epithelial ells on the villi of the small intestine (14), and animal studies have identified a loal intestinal antibody as the most important fator in passive host protetion against rotaviral disease (24, 30). Vaine programs have evaluated orally administered, live rotaviruses as vaine andidates. These animal rotaviruses appear to be naturally attenuated in humans but share antigeni harateristis with human rotaviruses (31, 34). Clinial trials of these vaine andidates have shown that the protetion onferred against disease is serotype speifi (9, 19), suggesting that suessful vaination may require ontrolled exposure of young hildren to rotavirus of more than one serotype. During vaine trials, preditions of effiay on the basis of monitoring of seroresponses by omplement fixation, enzyme immunoassay (EIA) for immunoglobulin G (IgG) and IgA, and neutralizing antibody have been less than the vaine effiay measured by linial protetion against diarrhea (32). Thus, it appears that serum immune responses may not be the most sensitive measure of vaine take, at least when animal (heterologous) vaines are given. It is likely that antibodies produed by ells in the gut muosa in response to infetion and sereted into the gut lumen are the best markers of an immune response. However, the antibody response of the intestine in vainated hildren has not been studied extensively (20, 33). In natural, presumably primary, severe aute rotavirus infetion, we * Corresponding author. t Present address: Department of Pediatris, University of Melbourne, Royal Children's Hospital, Parkville 3052, Vitoria, Australia have shown that seroonversions in antirotaviral IgM and IgG are sensitive indiators of an immune response and reflet the response in duodenal fluid. Both feal IgA and serum IgA showed a high preditive auray of duodenal IgA levels at 1 and 4 months postinfetion, but the stronger orrelation of duodenal IgA was with feal IgA (16). Thus, the measurement of opro-iga most aurately reflets the duodenal antibody response to primary rotavirus infetion. The aim of this study was to analyze in detail the ongoing intestinal and serum antibody responses to rotavirus in infants and young hildren monitored longitudinally for at least 1 year following severe aute rotavirus infetion. In partiular, the study was designed to determine the relevane and reliability of hanges in rotavirus-speifi opro- IgA, rotavirus-neutralizing oproantibody, and serum IgG, IgM, and IgA as sensitive markers of seondary rotavirus infetion. To this end, rotavirus antibody levels were measured in stools olleted weekly and in serum olleted at least every 4 months. Measurement of rotavirus-speifi antibodies in fees by using EIA deteted many signifiant inreases in opro-iga and neutralizing oproantibody not aompanied by virus exretion or by seroonversions. It is postulated that IgA oproonversion may be a reliable and sensitive marker of rotavirus reinfetion in young hildren. MATERIALS AND METHODS Patients and samples. Patients were 44 hildren, 3 weeks to 39 months old, admitted with aute rotavirus diarrhea to the infetious diseases ward of the Royal Children's Hospital, Melbourne, Australia, between April 1984 and September The linial, demographi, and laboratory findings of these hildren while hospitalized have been desribed previously (16). Approval for the study was given by the Human

2 1368 COULSON ET AL. Ethis Committees of the Royal Children's Hospital and the World Health Organization. Informed signed onsent was obtained from the parents of the hildren prior to enrollment in the study. Blood and feal speimens were olleted from eah hild at 2 to 7 days (median, 5.5 days), 24 to 66 days (median, 33.5 days), and 100 to 282 days (median, 122 days) after the onset of symptoms and at approximately 4-month intervals thereafter. Although olletion intervals varied from hild to hild, for eah individual hild, blood and fees were olleted on the same day. Additional feal speimens were olleted daily while eah hild was in the hospital and at 7- to 10-day intervals from 39 (89%) hildren for more than 1 year and 13 (30%) hildren for more than 2 years. Overall, the hildren were monitored for 6 to 32 months (median, 17 months) after the onset of symptoms. All families were in telephone ontat every 3 weeks with a researh nurse (J.S.L.), who visited their homes eah month, at whih time a linial history of the hild was obtained. All of the mothers were requested to keep a diary of the health of the hild, and approximately one-third did so for the duration of the study. Parents were asked to ontat J.S.L. or a pediatriian (K.G.) immediately when episodes of aute diarrhea ourred. Extra feal speimens were olleted during diarrheal episodes. Additional blood samples were obtained 2 to 6 weeks following the diarrhea in 39 (35%) of the episodes. Fees were refrigerated on olletion and stored at -70 C within 4 h, if obtained in the hospital. Fees olleted at home every 7 to 10 days by mothers were immediately frozen at -4 C in domesti freezers, olleted eah month, transferred to the hospital in a ar freezer, and then stored at -70 C. A 10% (wt/vol) suspension of fees (0.1 g in 1.0 ml) in phosphate-buffered saline (ph 7.2) was prepared on the day of assay for antiviral antibody. After entrifugation of the suspension at 2,000 x g for 10 min, the resulting supernatant was kept on ie until tested and then was stored immediately at -70 C. This supernatant was assayed for rotavirus antigen by monolonal EIA with EDTA treatment (13). Following Lyphogel onentration, the supernatant was examined for rotavirus by eletron mirosopy (EM) as previously desribed (13). Viruses and ells. Human rotavirus strains used in all assays inluded RV-4 (serotype 1), RV-5 (serotype 2), and RV-3 (neonatal; serotype 3) previously adapted to ulture in our laboratory (1); Wa (serotype 1) and ST-3 (neonatal; serotype 4) were provided by R. G. Wyatt (Bethesda, Md.); KU (serotype 1) was provided by T. Kutsuzawa (Japan). Simian rotavirus SA11 was supplied by H. Malherbe (San Antonio, Tex.). All viruses were propagated in MA104 ells in the presene of trypsin as previously desribed (11). EIAs for rotavirus antibodies. Sera and fees were examined for the presene of antirotaviral IgA, IgM, and IgG by EIA, with SA11 virus as the antigen (22). Serum-speifi IgG was measured at a single dilution by diret EIA and was expressed in units derived from a standard urve (5). In serum, speifi IgM titers were estimated by apture EIA with affinity-purified goat anti-human IgM (TAGO Immunoglobulins) as the oating reagent and monolonal antibody to rotavirus VP6 (13), onjugated to horseradish peroxidase to detet virus bound via speifi IgM. This eliminated the problem of false-positive results that were observed with some sera by using the diret IgM EIA (12). Serum IgA and feal antibodies (IgM and IgG) were titrated by a modifiation of a diret EIA (22) in whih sample dilutions were inubated in virus-oated plates overnight at 4 C, skim milk J. CLIN. MICROBIOL. powder replaed bovine serum albumin to blok bakground reations, and bound enzyme substrate was deteted with 3,3'-5,5'-tetramethylbenzidine. Feal extrats did not show false-positive IgM titers by diret EIA. Feal antirotaviral IgA was estimated at a single dilution (1:100) of fees by diret EIA and was expressed in units derived by omputer from log-logit analysis of data obtained from titration of a referene feal extrat. These EIA systems for IgA, IgM, and feal IgG to rotavirus were desribed previously (12). Definitions of immune onversions to rotavirus. IgG seroonversion was defined as an inrease in EIA units per milliliter of more than 30% between onseutive samples from the same hild (5). Seroonversion in IgA to rotavirus was defined as a fourfold or greater rise in titer between sequential samples. A positive IgM response in serum or fees was defined as a titer equal to or greater than 1:200. An IgA oproonversion was defined as a threefold inrease in EIA units per milliliter, to at least 100 units. These definitions of onversions in IgA and IgM have been validated (12). Total IgA determination in fees. This assay was idential to that desribed previously for measurement of total IgA in neonatal breast milk (25). In brief, affinity-isolated, goat anti-human IgA (TAGO) was employed as a oating reagent and bound IgA was deteted with the oating antibody, whih had been peroxidase onjugated (TAGO). The Australian Serum Protein Standard, ASPS 78-1, a referene preparation of human serum immunoglobulin, was tested in serial fivefold dilutions on eah plate. This data was used to express the total IgA level in fees in units of mirograms per milliliter, derived by omputer from log-logit analysis, as desribed previously for rotavirus-speifi IgA estimation in fees (12). Calulation of speifi IgA index. The ratio of the antirotaviral IgA level in units per gram of stool to the total IgA level in mirograms per gram of stool was designated as the speifi IgA index. Definition and validation of signifiant inreases in the IgA index. Both speifi and total IgA levels were estimated by EIA in 795 feal speimens olleted from a subgroup of 19 hildren reruited into the study during These speimens overed the first 101 to 495 (median, 420) days of follow-up only. Twelve of these hildren showed a signifiant rise (threefold inrease to.100 units) in speifi IgA levels during the aute phase of their rotavirus gastroenteritis. The IgA index in these 12 hildren inreased from 4- to 39-fold (median, 16-fold) to a minimum of 8.1. In the seven hildren in whom a rotavirus-speifi IgA response was not deteted, the IgA index dereased up to 10-fold or inreased up to 3.8-fold (median, 5-fold derease). Thus, an inrease in the IgA index from.4-fold to :>9-fold was defined as signifiant in demonstrating later rotavirus infetions. Five additional episodes of diarrhea during whih rotavirus reinfetion was deteted ourred in the group of 44 hildren during surveillane. The feal samples in whih total IgA was estimated overed three of these. In addition, five episodes of diarrhea during whih Giardia lamblia (2), adenovirus (2), and an unidentified "small virus" (1) were present were also identified. In eah of the three episodes of rotavirus diarrhea, speifi IgA units, total IgA units, and the speifi IgA index inreased signifiantly by 1 week following the time of the diarrhea and virus detetion. The speifi IgA units showed a greater inrease (12- to 262-fold) than did the total IgA units (1- to 10-fold), so that the IgA index was elevated. In ontrast, no signifiant hanges were deteted in the

3 VOL. 28, 1990 speifi IgA units or in the speifi IgA index during the five nonrotaviral episodes of gastroenteritis, although the hildren infeted with G. lamblia and adenovirus all showed inreases in their total IgA levels. Fluoresent fous redution neutralization assay in fees. The fluoresent fous redution neutralization method was adapted from that desribed previously for estimation of rotavirus-neutralizing antibody levels in hybridoma supernatant fluids (11), and its development will be desribed elsewhere (B. S. Coulson, and P. J. Masendyz, J. Clin. Mirobiol., in press). Rotavirus stoks (RV-4, Wa, Ku, RV-5, RV-3, and ST-3) were ativated with trypsin prior to being inubated with feal extrats. All extrats were reated with eah of the rotaviruses listed above. Stoks were diluted to ontain 2.5 x 104 fluoresing ell-forming units per ml of rotavirus and 5,ug of porine trypsin (Sigma Chemial Co.) per ml in Dulbeo modified Eagle medium. After inubation at 37 C for 30 min, the ativated virus was further diluted to 2.0 x 103 fluoresing ell-forming units per ml in Dulbeo modified Eagle medium ontaining 1,ug of trypsin per ml and 1% (vol/vol) fetal alf serum found to be free of rotavirus antibodies by EIA and fluoresent fous redution neutralization assay (Dulbeo modified Eagle mediumtrypsin-fetal alf serum). The diluted virus was mixed with an equal volume of feal extrat diluted 1:20 and 1:200 in Dulbeo modified Eagle medium-trypsin-fetal alf serum in dupliate, giving a final dilution of stool of 1:200 and 1:2,000, respetively. The mixture was inubated at 37 C for 1 h, and 50,ul per well was inoulated in dupliate onto washed onfluent monolayers of MA104 ells in mirodilution plates. After entrifugation at 1,200 x g for 30 min, the plates were inubated at 37 C in an atmosphere of 5% C02 and 95% relative humidity overnight. The ell supernatants were removed, and the monolayers were fixed in 70% (vol/vol) aetone for 5 min and then were air dried. Rabbit hyperimmune antiserum to SA1l (50,lI) at optimal dilution (1:500) in phosphate-buffered saline was added to eah well, and the plates were inubated at 37 C for 30 min. The plates were washed in phosphate-buffered saline before addition of 25,ul of fluoresein isothioyanate-labeled sheep anti-rabbit IgG F(ab')2 per well (Silenus, Australia) diluted 1:100 in phosphate-buffered saline. After 30 min at 37 C, the plates were washed and air dried and wells were examined for speifi fluoresene, as desribed previously (11). The neutralization titer of eah feal sample was expressed as the reiproal of the feal dilution giving 50% redution in the number of fluoresing ells. Feal extrats (n = 40) ontaining no antirotaviral IgA, IgM, or IgG by EIA all gave reiproal titers <200 by the fluoresent fous redution neutralization assay. Samples giving titers below 200 were therefore onsidered to be negative for neutralizing antibody to the rotavirus being tested. A signifiant elevation in neutralizing antibody, indiating rotavirus infetion, was onsidered to be a fourfold inrease in titer to at least 1:400. Statistial methodology. The paired Student t test was used to test the signifiane of elevations in IgG and IgA levels over time. Correlation between total IgA, speifi IgA, and the IgA index was tested by using Pearson's test on logarithmially transformed data. RESULTS Of the 44 hildren reruited for follow-up, 4 withdrew from the study after 6 months, so their data are not inluded in this report. The remaining 40 hildren were monitored for 12 to 32 months (mean, 21 months). Samples were olleted ROTAVIRUS IgA COPROCONVERSION 1369 from 26 (65%) hildren until the end of the first rotavirus season after their initial infetion (12 to 24 months; mean, 17 months) and from the remaining 14 hildren until the end of their seond rotavirus season following reruitment (24 to 32 months; mean, 29 months). For the purposes of this study, the rotavirus season is onsidered to over the six older months, i.e., from May to Otober. Longitudinal analysis of rotavirus-speifi antibody isotypes in serum. Antirotaviral antibody flutuations deteted in serum during the first 4 months of follow-up have been previously reported (16). Sine these flutuations are onsidered to relate to the rotavirus infetion whih brought the hild to hospital, they are not onsidered in this analysis, whih desribes antirotaviral antibody flutuations observed during the later 5 to 32 months of follow-up. Of the 40 hildren monitored until the end of the first rotavirus season after that in whih they were reruited, 33 (83%) showed one (n = 23), two (n = 9), or three (n = 1) seroonversions after the first 4 months of follow-up. These seroonversions were deteted in antirotaviral IgG only (43%), IgA only (30%), or both IgG and IgA simultaneously (27%). The overall seroonversion rate was 1.0 per hild per year. Of the 14 hildren monitored until the end of their seond rotavirus season, 9 (64%) showed a single seroonversion. The antibody isotypes involved were both IgG and IgA (56%) and IgG only (44%). The overall seroonversion rate for the 14 hildren -n the 12 to 18 months inluding their seond season was 0.7 per hild per year. The overall seroonversion rate for the 40 hildren during the 2 to 3 years inluding two seasons was 0.8 per hild per year. Antirotaviral IgM was deteted in only one hild on one oasion and was aompanied by a seroonversion in IgA. Overall, 35 of the total 54 seroonversions were assoiated with an episode of diarrhea (65%), inluding only 3 (5%) in whih rotavirus was deteted in stools by EIA or EM. Rotavirus was found in feal speimens from two other hildren without seroonversion (see below). The geometri mean titers (IgA) and units (IgG) of antirotaviral antibodies measured in the serum samples of all 40 hildren studied are shown at 4-month intervals from reruitment (Fig. 1). Large elevations in both IgG and IgA ourred following the severe, presumably primary, episode of rotavirus gastroenteritis. Further signifiant elevations in IgG ourred between 12 and 16 months and between 24 and 28 months after the severe gastroenteritis (paired t test; P < and P < 0.05, respetively). As the hildren were reruited during the period from April to September in two suessive years, the inreases in geometri mean titer of speifi IgG from 12 to 16 months and from 24 to 28 months orresponded to the ends of the two suessive winters (rotavirus seasons) following the severe episode of rotavirus gastroenteritis. A signifiant rise in the geometri mean titer of speifi IgA was reorded only between 4 and 8 months after the severe gastroenteritis (paired t test; P < 0.05). The rise in geometri mean titer of speifi IgA from 12 to 16 months was not statistially signifiant (paired t test; P > 0.05). Comparison of antirotaviral IgA units and the speifi IgA index for longitudinal analysis of rotavirus oproantibody. There were 795 feal speimens from 19 hildren in whih speifi and total IgA levels were measured and the IgA index was alulated. There was a weak but signifiant orrelation between speifi and total IgA levels for eah feal speimen (r, = 0.348; P < 0.01). However, the stronger orrelation was between speifi IgA and the IgA index (r, = 0.743; P < 0.005). These results show that the measured

4 1370 COULSON ET AL. J. CLIN. MICROBIOL. a lm -I : o F- z > L 0 z 1600 « b- o CE LA. o CE 400 w NUMBER OF DAYS AFTER ONSET OF SEVERE ROTAVIRUS DIARRHEA FIG. 1. Geometri mean titers (IgA) and geometri mean units (IgG) of antirotaviral antibodies in sera of hildren at eah 4-month interval following hospitalization with rotavirus gastroenteritis. Symbols: *, IgG; O, IgA. Numbers refer to number of serum samples tested at eah time point. variations in speifi IgA in fees were not the result of oinidental hanges in total IgA onentration. The speifi IgA level and the IgA index alulated as desribed previously are shown in Fig. 2 and 3 for two representative hildren. The hild represented in Fig. 2 showed very high levels of antirotaviral opro-iga within 1 week of the onset of aute severe rotavirus gastroenteritis. A peak was also deteted in the total IgA levels, but at a lower level, so the IgA index for this period was also signifiantly elevated. Similarly, the elevation in antirotaviral opro-iga at day 80 was refleted in both the total IgA levels and the IgA index. Flutuations in the levels of total IgA ourred from day 90 to day 460, unaompanied by signifiant variations in the other parameters. An extreme elevation in opro-iga and C,J 1000 o. o o O o b- AAA A oe I < 400- O200- IgA index at 480 days was due to rotavirus-assoiated diarrhea but was not aompanied by any alteration in total IgA levels. Figure 3 represents a hild, aged 21 months at hospitalization, in whom no opro-iga onversion or speifi opro- IgA was deteted during hospitalization with aute rotavirus gastroenteritis. Two seondary boosts in opro-iga ourred. The first, at day 80, was aompanied by a rise in the IgA index, although total IgA levels were stable. The seond opro-iga boost from days 250 to 260 was aompanied by diarrhea and a signifiant elevation in both total IgA levels and the IgA index. Speifi IgA oproonversions as defined here (see Materials and Methods) were deteted in 24 instanes from the 795 speimens olleted from 19 hildren. These events were 0 x o<5 50 C C NUMBER OF DAYS AFTER ONSET OF SEVERE ROTAVIRUS DIARRHEA FIG. 2. Comparison of the profiles of antirotaviral IgA units, IgA indies, and levels of total IgA in stools of a 3-month-old hild. C, IgA oproonversion; O, diarrhea reorded. Symbols: *, diarrhea with rotavirus present in the stools; A, IgM deteted in the stools , IgG seroonversion;., IgA seroonversion.

5 VOL. 28, 1990 ROTAVIRUS IgA COPROCONVERSION 1371 J o <0O gr :Z e à A <p x 50 w z 25 1~~~~~~~~~~~~~~~ 9 CD NUMBER OF DAYS AFTER ONSET OF SEVERE ROTAVIRUS DIARRHEA FIG. 3. Comparison of the profiles of antirotaviral IgA units, IgA indies, and levels of total IgA in the stools of a 21-month-old hild. Symbols are desribed in the legend to Fig. 2. analyzed to determined whether or not speifi IgA oproonversion was as sensitive a measure of rotavirus infetion as was the IgA index. Of 24 speifi IgA oproonversions, 20 (83%) were assoiated with a signifiant rise in the IgA index. Of these 20, 18 (90%) were also assoiated with other measures of rotavirus infetion (at least one of rotavirus in stools, opro-igm, seroonversion, or neutralizing antibody oproonversion). Three of the four IgA oproonversions not showing an elevated IgA index were also assoiated with other markers of rotavirus infetion. These oproonversions did not show an elevated IgA index, sine the total IgA level was also inreased. The false-negative and false-positive rates for speifi opro-iga onversion as a measure of rotavirus infetion were 5% (1 of 21) and 4% (1 of 24), respetively. The IgA index showed a false-negative rate of 17% (4 of 24) and a false-positive rate of 5% (1 of 21). We onluded that speifi opro-iga measurement alone was more sensitive an estimate of rotavirus infetion than was the IgA index. Overall analysis of the longitudinal data was thus performed with speifi opro-iga unit values only. This avoided the need for total IgA estimation in all stools olleted. Comparison of IgA oproonversion with onventional measures of rotavirus infetion during rotavirus reinfetion. Rotavirus-speifi opro-iga units were estimated in all stool speimens (3,150) olleted from the 40 hildren (mean, 79 speimens per hild) in whom antibody isotypes in serum were studied. These speimens were also analyzed for the presene of rotavirus by EM and EIA and for elevated levels of opro-igm to rotavirus. A subset of 422 feal speimens olleted from 10 hildren (mean, 22 per hild) was also titrated for neutralizing antibody to rotaviruses RV-4, Wa, Ku, RV-5, RV-3, and ST-3. The 115 oproonversions reorded in the 3,150 stools tested were aompanied by seroonversion in 61 ases (53%), omprising IgG seroonversion alone (46%), IgA seroonversion alone (28%), IgG and IgA seroonversion (25%), and onversion in IgG, IgA, and IgM (1%). Seroonversion in the absene of IgA oproonversion or any other marker ourred on three oasions. The inidene of IgA oproonversion with aompanying seroonversion was not age related. On all of the five oasions when opro-iga onversions were assoiated with the presene of rotavirus in stools, mild to moderate diarrhea was reorded. Four ases were aompanied by seroonversion or opro-igm. All 11 antirotaviral IgM-ontaining stools also showed opro-iga onversion. Six (55%) were assoiated with seroonversion, and two (18%) were assoiated with the presene of rotavirus in stools. Copro-IgG to rotavirus was not deteted in any stool sample tested. The relation of signifiant inreases in neutralizing oproantibody against at least one rotavirus serotype to the other markers of rotavirus infetion tested is shown in Table 1. Although seroonversion, rotavirus detetion, and oproonversion in IgM all showed a high preditive auray of oproonversion in neutralizing antibody (92 to 100%), they were insensitive markers, being present in only 4 to 46% of neutralizing antibody oproonversions. In ontrast, rotavirus oproonversion in IgA was both a highly preditive (92%) and sensitive (92%) marker of oproonversion in rotavirus-neutralizing antibody. Two neutralizing antibody onversions and one IgA oproonversion were not aompanied by any other marker of rotavirus infetion. Longitudinal analysis of rotavirus-speifi IgA in stools. Levels of speifi opro-iga units were estimated in stool speimens olleted from 40 hildren, the same population in whih antibody isotypes in serum were studied. Of the 40 hildren monitored for 12 to 18 months, inluding one rotavirus season, 36 (90%) showed at least one oproonversion. Up to 6 oproonversions per hild were reorded, TABLE 1. Comparison of rotavirus-neutralizing oproantibody onversions with other markers of rotavirus reinfetion Soure Marker % (No. positive/no. tested) Sensitivity Preditive auray Serum Speifi IgG onversion 29 (7/24) 88 (7/8) Speifi IgA onversion 33 (8/24) 100 (8/8) Speifi IgM onversion 4 (1/24) 100 (1/1) Speifi IgA, IgG, or IgM 46 (11/24) 92 (12/13) onversion Fees Rotavirus deteted (EM or EIA) 4 (1/24) 100 (1/1) Speifi IgM onversion 13 (3/24) 100 (3/3) Speifi IgA onversion 92 (22/24) 92 (22/24) a Sensitivity, Perentage of hildren with a neutralizing oproantibody onversion who also had a partiular marker at the given site. b Preditive auray, Perentage of hildren with a partiular marker at the given site who also had a neutralizing oproantibody onversion.

6 1372 COULSON ET AL. with a rate of 1.6 per hild per year. Of the 14 hildren monitored for 2 to 3 years, inluding two rotavirus seasons, 12 (86%) showed IgA oproonversions to rotavirus in the 12 to 18 months inluding the seond season, ranging from 1 to 4 per hild, with a rate of 1.7 per hild per year. The episode of severe rotaviral gastroenteritis whih provided the starting point for longitudinal follow-up in the hildren was probably a primary infetion (16) as judged by finding low or no rotavirus antibodies in aute-phase speimens and by deteting IgM serum antibody in 100% of the hildren. The peak level of speifi opro-iga was identified where possible for eah hild produing opro-iga following the primary infetion and eah later opro-iga onversion (representing reinfetion). Of the 97 pairs of primary and reinfetions, 79 (81%) showed greater peak opro-iga levels during the reinfetions than during the primary infetion. In six of seven hildren with five to eight reinfetions, the peak opro-iga level to rotavirus inreased by two- to eightfold, in a stepwise fashion from the first to the fourth or fifth reinfetion. Peak levels were lower during the last three or four opro-iga onversions, whih ourred during the seond season of follow-up. DISCUSSION Infetion of hildren with rotavirus is urrently onsidered proven if rotavirus partiles or antigen are deteted in stools or if seroonversion in IgG, IgM, IgA, or neutralizing antibody an be demonstrated. In severe aute gastroenteritis where infetion is usually primary, these riteria are onsistently met (5, 6, 10, 16, 17, 26, 29). However, rotavirus infetion is often mild or asymptomati (5, 8, 15). In this study, although all hildren exreted rotavirus during their episode of aute rotaviral gastroenteritis, virus was deteted at the time of later IgA oproonversions on only five oasions and was never found during IgA oproonversions not aompanied by diarrhea. Conentration of feal extrats by 10-fold and pretreatment with EDTA failed to inrease this number. All hildren studied had antirotaviral IgA and IgG in serum, and rotavirus was not found in stools ontaining high levels of antirotaviral opro-iga. It is likely that preexisting immunity may have kept virus growth in the intestine to levels below the limits of detetion by EIA and EM. One of the rotavirus-positive diarrheal episodes ourred during the brief hospitalization of the hild for other reasons, when all stools passed were obtained. Colletion of all stools during eah diarrheal episode may improve virus detetion. When more sensitive antigen assays, perhaps based on probes for viral RNA, are available, the feal samples of this study ould be retested. In alves, experimental seondary rotavirus infetions resulted in immune onversions in serum and fees but viral exretion was not deteted (3). Seroonversion following rotavirus reinfetion was deteted on 81 oasions, muh more frequently than was rotavirus antigen in stools. The major isotype involved was IgG (58%), followed by IgA (41%) and IgM (1%). IgG and IgA seroonversions ourred simultaneously in 75% of the ases, and these antibodies persisted for prolonged periods. In ontrast, speifi IgM was deteted only one. All of these hildren had shown an IgM immune response in serum during their primary infetion, with 91% responding in IgG and 68% responding in IgA (16). The IgM levels during primary infetion peaked at least 6 days after onset and often fell to baseline within 1 month. Sera may not have been olleted near enough to the time of rotavirus infetion to J. CLIN. MICROBIOL. detet speifi IgM in rotavirus reinfetions not temporally assoiated with diarrhea. However, serum was olleted less than 4 weeks after diarrhea in 44%, and less than 10 days after diarrhea in 15%, of reported diarrheal episodes. Children showed inreasing numbers of rotavirus seroonversions with inreased length of follow-up, but these rarely involved speifi IgM. At the end of eah of the two suessive rotavirus seasons (winters) during whih the hildren were studied, their geometri mean levels of speifi IgG rose signifiantly. This suggests that reinfetion with rotavirus also tended to be a seasonal event and reflets our finding that most of the hildren in the population studied seroonverted to rotavirus eah season. Similar IgG and IgA profiles with inreasing age were shown in the Federal Republi of Germany (7). The hildren monitored for one omplete rotavirus season after reruitment showed a higher seroonversion rate (1.0 per hild per year) than those monitored during their seond season (0.7 per hild per year). The overall seroonversion rate of 0.8 per hild per year is similar to rates of infetion (0.3 to 1.2 episodes per hild per year) reported in other longitudinal studies in both developed and developing ountries. These rates were alulated by using rotavirus detetion and seroonversion in IgG or IgA or both by EIA (6, 26, 27, 29), rotavirus detetion and seroonversion by indiret immunofluoresene (17), and rotavirus detetion alone (19, 21) Ṫhe weekly olletion of stools from 44 hildren for 6 to 32 months has enabled us to study hildhood antirotaviral oproantibody profiles in great detail. Speifi IgM was rarely deteted, and speifi IgG was never found. Marked flutuations in speifi opro-iga, whih were not the result of oinidental hanges in total IgA onentration, ourred in many hildren, although total IgA levels were sometimes elevated during proven rotavirus infetions. Thus, use of the IgA index, although giving similar results to speifi opro- IgA alone, was shown to be unneessary to identify opro- IgA onversion to rotavirus. In a small study, Hjelt et al. (18) ame to the same onlusion. The IgA index has been used for rotavirus opro-iga measurement without prior verifiation (4, 23). Analysis of the serum and muosal antibody responses to severe aute rotavirus gastroenteritis in these hildren showed that opro-iga, followed by serum IgA, were the best preditors of duodenal IgA at 1 and 4 months after infetion (16). It is thus likely that the opro-iga flutuations measured subsequently in this study, aompanied by virusneutralizing ativity, reflet similar antibody hanges in the duodenum. This is the site at whih virus-neutralizing antibodies have been shown to exert protetive apabilities in animal models (24, 30). All of the other markers of rotavirus infetion tested showed a high level of preditive auray in relation to onversion in neutralizing oproantibody (92 to 100%). However, the sensitivities of the markers in relation to neutralizing oproantibody varied widely. Detetion of opro-igm, virus in stools, or serum IgM was an insensitive measure of oproonversion in neutralizing antibody. Seroonversion in IgG or IgA or both was a moderately sensitive marker, being reorded in 46% of oproonversions. The most sensitive marker, present in 92% of neutralizing antibody oproonversions, was rotavirus-speifi opro-iga onversion. Taken together with the validation of the unadjusted rotavirus-speifi IgA level as a reliable indiator of flutuations in rotavirus opro-iga, the orrelation of opro-iga

7 VOL. 28, 1990 with neutralizing oproantibody suggests that oproonversions represent true elevations in antirotaviral IgA with neutralizing apaity. A oproonversion in IgA appears to indiate genuine rotavirus infetion. Without measurement of neutralizing oproantibodies, Bernstein et al. (4) onluded from a human rotavirus hallenge study in adults that inreases in feal antibody may be a reliable indiator of rotavirus infetion, even in the absene of detetable virus shedding or seroonversion. Taking oproonversion in speifi IgA or neutralizing antibody as the standard measure, this study shows that use of seroonversion and virus detetion only as measures of rotavirus reinfetion would lead to underestimation of virus infetion by approximately 200%. On the basis of the inidene of oproonversion in IgA, the more aurate estimate of rotavirus infetion in this study is 1.6 per hild per year for the period overing the first rotavirus season and 1.7 per hild per year for the period overing the seond season after reruitment. In ontrast to that observed for seroonversion rates, no drop in oproonversion rate in the seond season was observed, although the peak opro-iga levels reorded during the oproonversions were lower in the seond season than the first. It is likely that seroonversions in the presene of already elevated serum IgA and IgG rotavirus antibody levels are diffiult to detet. Copro-IgA levels often delined to baseline before a new oproonversion was observed. Overall, 1.7 rotavirus oproonversions per hild per year were reorded in this study. This is in exess of-the rates (0.3 to 1.2) determined in other reports (6, 17, 19, 21, 26, 27, 29), but oproonversions were not measured in these studies. Shinozaki et al. (28) also reported a rate of 1.7 rotavirus infetions per hild per year from opro-iga titer inreases to rotavirus in seven hildren tested monthly for 1 year. Two of the five hildren exreting rotavirus and showing opro-iga onversions during reinfetion did not seroonvert. This suggests that the muosal and systemi antibody responses to rotavirus an be unrelated, as was postulated by Angeretti et al. (2). The preliminary results from the longitudinal study indiated that an anamnesti opro-iga response, with higher peak opro-iga levels, ourred in 81% of rotavirus reinfetions. Anamnesti responses in opro-iga have been doumented previously in five hildren with seondary rotavirus infetion (35) and in a larger group of hildren reeiving rhesus rotavirus vaine (33). Suffiient virus for serotyping was obtainable only from one of the five hildren exreting rotavirus during reinfetion, and this showed serotype 1 by monolonal EIA (11; Fig. 2, day 480). The primary infeting virus was also serotype 1 (Fig. 2, day 3). It was not possible to determine the infeting serotype from the inreases in antibody levels measured by EIA, as antigens of different serotypes (1 to 4) gave similar EIA units and titers (12). During reinfetion, inreases in neutralizing oproantibody titers usually ourred to more than one serotype simultaneously, so this method was also unsuitable for determining the serotype of the infeting virus during reinfetion. In onlusion, the results of the study presented here suggest that speifi antirotaviral opro-iga is a reliable indiator of rotavirus reinfetion and that it is a more sensitive measure than is rotavirus detetion or seroonversion. Confirmation of this finding in hildren studied longitudinally from birth in both developed and developing ountries is required. Measurement of rotavirus-speifi opro-iga in stools, olleted weekly during epidemiologial ROTAVIRUS IgA COPROCONVERSION 1373 studies and vaine trials, is likely to provide an aurate estimation of rotavirus infetion and may assist in the measurement of vaine take and vaine effiay. ACKNOWLEDGMENTS This study was supported by grants from the National Health and Medial Researh Counil of Australia, the Diarrheal Diseases Control Program of the World Health Organization, and the Royal Children's Hospital Researh Foundation. The advie of Don Roberton is muh appreiated. We are grateful for the assistane of Leanne Uniomb, Sandra Lawrane, Lorraine Adams, Neil Franis, Lynette Vaelioja, Jane Lee, Val Williams, and the nursing staff of the infetious diseases ward, Royal Children's Hospital. LITERATURE CITED 1. Albert, M. J., and R. F. Bishop Cultivation of human rotaviruses in ell ulture. J. Med. Virol. 13: Angeretti, A., M. T. Magi, C. Merlino, B. Ferrara, and A. N. Ponzi Speifi serum IgA in rotavirus gastroenteritis. J. Med. Virol. 23: Bahmann, P. A., and R. G. Hess Loal immunity in rotaviral infetions. Ann. Reh. Vet. 14: Bernstein, D. I., J. M. Ziegler, and R. L. Ward Rotavirus feal IgA response in adults hallenged with human rotavirus. J. Med. Virol. 20: Bishop, R. F., E. Cipriani, J. S. Lund, G. L. Barnes, and C. S. Hosking Estimation of rotavirus immunoglobulin G antibodies in human serum samples by enzyme-linked immunosorbent assay: expression of results as units derived from a standard urve. J. Clin. Mirobiol. 19: Blak, R. E., H. B. Greenberg, A. Z. Kapikian, K. H. Brown, and S. Beker Aquisition of serum antibody to Norwalk virus and rotavirus and relation to diarrhea in a longitudinal study of young hildren in rural Bangladesh. J. Infet. Dis. 145: Brussow, H., H. Werhau, L. Lerner, C. Mietens, W. Liedtke, J. Sidoti, and J. Sotek Seroonversion patterns to four human rotavirus serotypes in hospitalized infants with aute rotavirus gastroenteritis. J. Infet. Dis. 158: Cameron, D. J. S., R. F. Bishop, A. A. Veenstra, and G. L. Barnes Nonultivable viruses and neonatal diarrhea: fifteen-month survey in a newborn speial are nursery. J. Clin. Mirobiol. 8: Christy, C., P. Madore, M. E. Pihihero, C. Gala, P. Pinus, D. Vosefski, Y. Hoshino, A. Kapikian, and R. Dolin Field trial of rhesus rotavirus vaine in infants. Pediatr. Infet. Dis. J. 7: Clark, H. F., K. T. Dolan, P. Horton-Slight, J. Palmer, and S. A. Plotkin Diverse serologial responses to rotavirus infetion of infants in a single epidemi. Pediatr. Infet. Dis. J. 4: Coulson, B. S., K. J. Fowler, R. F. Bishop, and R. G. H. Cotton Neutralizing monolonal antibodies to human rotavirus and indiations of antigeni drift among strains from neonates. J. Virol. 54: Coulson, B. S., K. Grimwood, R. F. Bishop, and G. L. Barnes Evaluation of end-point titration, single dilution and apture enzyme immunoassays for measurement of antirotaviral IgA and IgM in infantile seretions and serum. J. Virol. Methods 26: Coulson, B. S., L. E. Uniomb, G. A. Pitson, and R. F. Bishop Simple and speifi enzyme immunoassay using monolonal antibodies for serotyping human rotaviruses. J. Clin. Mirobiol. 25: Davidson, G. P., and G. L. Barnes Strutural and funtional abnormalities of the small intestine in infants and young hildren with rotavirus enteritis. Ata Paediatr. Sand. 68: Friedman, M. G., A. Galil, B. Sarov, M. Morgalith, G. Katzir, K. Midthun, K. Taniguhi, S. Urasawa, A. Z. Kapikian, R. Edelman, and I. Sarov Two sequential outbreaks of

8 1374 COULSON ET AL. rotavirus gastroenteritis: evidene for symptomati and asymptomati reinfetions. J. Infet. Dis. 158: Grimwood, K., J. C. S. Lund, B. S. Coulson, I. L. Hudson, R. F. Bishop, and G. L. Barnes Comparison of serum and muosal antibody responses following severe aute rotavirus gastroenteritis in young hildren. J. Clin. Mirobiol. 26: Gurwith, M., W. Wenman, D. Hinde, S. Feltham, and H. Greenberg A prospetive study of rotavirus infetion in infants and young hildren. J. Infet. Dis. 144: Hjelt, K., P. C. Grauballe, P. O. Shiotz, L. Anderson, and P. A. Krasilnikoff Intestinal and serum immune response to a naturally aquired rotavirus gastroenteritis in hildren. J. Pediatr. Gastroenterol. Nutr. 4: Lanata, C. F., R. E. Blak, R. del Aguila, A. Gil, H. Verastegui, G. Gerna, J. Flores, A. Z. Kapikian, and F. E. Andre Protetion of Peruvian hildren against rotavirus diarrhea of speifi serotypes by one, two, or three doses of the RIT 4237 attenuated bovine rotavirus vaine. J. Infet. Dis. 159: Losonsky, G. A., M. B. Rennels, Y. Lim, G. Krall, A. Z. Kapikian, and M. M. Levine Systemi and muosal immune responses to rhesus rotavirus vaine MMU Pediatr. Infet. Dis. J. 7: Mata, L., A. Simhon, J. J. Urrutia, R. A. Kronmal, R. Fernandez, and B. Garia Epidemiology of rotaviruses in a ohort of 45 Guatamalan Mayan Indian hildren observed from birth to the age of three years. J. Infet. Dis. 148: MLean, B., S. Sonza, and I. H. Holmes Measurement of immunoglobulin A, G, and M lass rotavirus antibodies in serum and muosal seretions. J. Clin. Mirobiol. 12: Nishio, O., Y. Ishihara, S. Isomura, H. Inoue, and S. Inouye Long-term follow-up of infants from birth for rotavirus antigen and antibody in the fees. Ata Paediatr. Jpn. 30: Offit, P. A., and H. F. Clarke Protetion against rotavirusindued gastroenteritis in a murine model by passively aquired gastrointestinal but not irulating antibodies. J. Virol. 54: Roberton, D. M., P. J. Forrest, E. Frangoulis, and N. Mermelstein Early indution of seretory immunity in infany: speifi antibody in neonatal breast milk. Arh. Dis. Child. J. CLIN. MICROBIOL. 61: Rodriguez, W. J., H. W. Kim, C. D. Brandt, R. H. Shwartz, M. K. Gardner, B. Jeffries, R. H. Parrott, R. A. Kaslow, J. I. Smith, and A. Z. Kapikian Longitudinal study of rotavirus infetion and gastroenteritis in families served by a pediatri medial pratie: linial and epidemiologi observations. Pediatr. Infet. Dis. J. 6: Ryder, R. W., N. Singh, W. C. Reeves, A. Z. Kapikian, H. B. Greenberg, and R. B. Sak Evidene of immunity indued by naturally aquired rotavirus and Norwalk virus infetion on two remote Panamanian islands. J. Infet. Dis. 151: Shinozaki, T., K. Araki, H. Ushijima, B. Kim, and R. Fujii Frequeny of suessive rotavirus infetions among infants in a nursery home measured by oproantibody onversion. Eur. J. Pediatr. 145: Simhon, A., L. Mata, M. Vives, L. Rivera, S. Vargas, G. Ramirez, L. Lizano, G. Catarinella, and J. Azofeifa Low endemiity and low pathogeniity of rotaviruses among rural hildren in Costa Ria. J. Infet. Dis. 152: Snodgrass, D. R., and P. W. Wells Passive immunity in rotaviral infetions. J. Am. Vet. Med. Asso. 173: Stuker, G., L. S. Oshiro, and N. J. Shmidt Antigeni omparisons of two new rotaviruses from rhesus monkeys. J. Clin. Mirobiol. 11: Vesikari, T., A. Z. Kapikian, A. Delem, and G. Zissis A omparative trial of rhesus monkey (RRV-1) and bovine (RIT 4237) oral rotavirus vaines in young hildren. J. Infet. Dis. 153: Wright, P. F., T. Tajima, J. Thompson, K. Kokubun, A. Kapikian, and D. T. Karzon Candidate rotavirus vaine (rhesus rotavirus strain) in hildren: an evaluation. Pediatris 80: Wyatt, R. G., A. Z. Kapikian, and C. A. Mebus Indution of ross-reative serum neutralizing antibody to human rotavirus in alves after in utero administration of bovine rotavirus. J. Clin. Mirobiol. 18: Yamaguhi, H., S. Inouye, M. Yamauhi, T. Morishima, S. Matsuno, S. Isomura, and S. Suzuki Anamnesti response in feal IgA antibody prodution after rotaviral infetion of infants. J. Infet. Dis. 152: