Detection and Clinical Implications of Monovalent Rotavirus Vaccine-Derived Virus Strains in Children with Gastroenteritis in Alberta, Canada

ABSTRACT

While rotavirus vaccine programs effectively protect against severe rotavirus gastroenteritis, rotavirus vaccine strains have been identified in the stool of vaccinated children and their close contacts suffering from acute gastroenteritis. The prevalence of vaccine strains, the emergence of vaccine-derived strains, and their role in acute gastroenteritis are not well studied. We developed a locked nucleic acid reverse transcription real-time PCR assay (LNA-RTqPCR) to detect the monovalent rotavirus vaccine (RV1) Rotarix nonstructural protein 2 (NSP2) in children with acute gastroenteritis and healthy controls, and validated it using sequence-confirmed RV1 strains. The association between RV1-derived strains and gastroenteritis was determined using logistic regression. The new assay exhibited 100% (95% CI 91.7%, 100%) diagnostic sensitivity and 99.4% (95% CI 96.2%, 100%) diagnostic specificity, with a detection limit of 9.86 copies/reaction and qPCR efficiency of 99.7%. Using this assay, we identified the presence of RV1-derived NSP2 sequences in 7.7% of rotavirus gastroenteritis cases and 98.6% of rotavirus-positive healthy children (94.4% had previously received the RV1). Among gastroenteritis cases, those whose stool contained RV1-derived strains had milder gastroenteritis symptoms compared to that of natural rotavirus infections. We observed no significant association between RV1-derived strains and gastroenteritis (odds ratio [OR] 0.98; 95% CI 0.60, 1.72). Our study demonstrated that the new assay is suitable for monitoring RV1-derived rotavirus strain circulation and that the RV1-derived strains are not associated with development of gastroenteritis symptoms.

INTRODUCTION

Prior to the implementation of rotavirus vaccine programs, group A rotavirus was the leading cause of severe acute gastroenteritis (AGE) in young children worldwide (1). The two rotavirus vaccines recommended for worldwide use by the World Health Organization (WHO) (2) are the monovalent (Rotarix; GlaxoSmithKline Biologics, Rixensart, Belgium) and a pentavalent mixture of strains (RotaTeq; Merck & Co., Whitehouse station, NJ, USA). The monovalent rotavirus vaccine (RV1) contains a live, attenuated human G1P[8] strain RIX4414 derived from the parental strain 89-12 (3), and the pentavalent preparation (RV5) consists of five bovine-human mono-reassortant rotaviruses carrying the G1-G4 and the P[8] proteins derived from human strains (4).

Rotavirus vaccination offers cross-protection against a broad range of rotavirus serotypes and has substantially reduced the burden of rotavirus AGE (5). Further, person-to-person spread of vaccine strains can induce a serological response that offers protection for contacts of vaccinees (6). Administered orally, rotavirus vaccine strains mimic natural rotavirus infections with intestinal replication and shedding in the stool (7, 8). Rotavirus vaccine strains have been found to cause symptomatic illness, can lead to prolonged shedding in individuals with immunodeficiencies (9, 10), and symptomatic AGE from horizontal transmission to unvaccinated children has been reported (11–19).

A concern with any live-attenuated vaccines is the emergence of disease-causing vaccine revertant or reassortant strains (20). Whole-genome sequence analysis has revealed nonsynonymous point mutations in RV1 strains shed by vaccinated infants (21). Furthermore, genetic reassortments between vaccine strains and circulating wild-type strains have been identified for both RV1 and the pentavalent rotavirus vaccine (RV5) RotaTeq (11–13, 15–19, 22). While studies focusing on partial genomic sequences found RV1 and RV5 strains circulating in children with AGE (14, 23), the pathogenicity of vaccine or reassortant strains in immunocompetent children is unclear. Therefore, monitoring the prevalence of symptomatic infections associated with rotavirus vaccine strains and analysis of their clinical impact is crucial to the evaluation of vaccine program safety.

An effective reverse transcription real-time PCR method with oligonucleotides incorporating special nucleotides, restructured C-5 propynyl-dC bases, and G clamps was developed in 2014 to identify the RV1-specific NSP2 gene segment (24). However, the aforementioned assay was one step and the special oligonucleotides are proprietary. Thus, we sought to develop and validate a two-step reverse transcription quantitative real-time PCR (RT-qPCR) assay with common reagents that differentiates RV1 vaccine or vaccine-derived strains from wild-type rotavirus. Our secondary objective was to describe the epidemiology of AGE and virus shedding associated with RV1-derived strains following the implementation of Alberta’s universal RV1 vaccination program in June 2015.

MATERIALS AND METHODS

Study participants and specimen acquisition.

Ethics approval was provided by the University of Alberta and the University of Calgary ethics review boards. Caregiver consent and participant assent (as appropriate) were obtained. Eligible participants, namely, children under 18 years old with AGE and unmatched controls with no AGE symptoms, were recruited consecutively by the Alberta Provincial Pediatric EnTeric Infection TEam (APPETITE) according to study protocol (25). Children with AGE, defined as ≥3 episodes of vomiting or diarrhea in the preceding 24 h and symptom duration of <7 days, were recruited through two pediatric emergency departments (EDs) and a telephone nursing advice line in Alberta. Children in these EDs with noninfectious illnesses or children receiving routine vaccinations at a Calgary public health clinic were enrolled as asymptomatic controls.

Rectal swabs and stool specimens were collected from symptomatic children between December 2014 and August 2018. Rectal swabs were collected by inserting FLOQSwabs (Copan Italia, Brescia, Italy) into the rectum and rotated 360°. Controls only provided stool specimens. Stool specimens were collected in Fecal Collectors (V302-F, Starplex Scientific, ON, Canada). Two oral suspensions of the RV1 (GlaxoSmithKline Inc, Canada) were provided by Alberta Health Services. Specimens were transported at 4°C to the testing laboratory within 24 h of collection and frozen immediately at −70°C until analyzed within a week.

Molecular detection of rotavirus.

Details of nucleic acid extractions, reverse transcription, and rotavirus molecular detection are described in the supplemental methods.

Group A rotaviruses in stool, swab, and RV1 specimens were identified by an in-house reverse transcription real-time PCR (RT-qPCR) assay (Fig. 1), the gastroenteritis viral panel (GVP), and a multitarget bead-based molecular assay, the Luminex gastrointestinal pathogen panel (GPP) (Luminex Molecular Diagnostics, ON, Canada), as described previously (26). Other targets of GVP include adenovirus (all serotypes), astrovirus, norovirus (genogroups I and II), and sapovirus. Besides rotavirus, GPP also detects two other viral pathogens (adenovirus 40/41, norovirus GI and GII), nine bacterial pathogens (Campylobacter, Clostridioides difficile, Escherichia coli O157, enterotoxigenic E. coli (ETEC), Salmonella, Shigella, Shiga toxin-producing E. coli (STEC), Vibrio cholerae, and Yersinia enterocolitica), and three parasites (Cryptosporidium, Entamoeba histolytica, and Giardia).

FIG 1.

Rotavirus and Rotarix-derived strains detected in gastroenteritis cases (A) and healthy controls (B) in this study and their VP7/VP4 genotypes. GVP, gastrointestinal virus panel RT-qPCR; LNA, locked nucleic acid; NSP2, nonstructural protein 2; VP4, viral protein 4; VP7, viral protein 7; X, untypeable result.

VP7 and VP4 genotyping approach.

The stool or swab specimen with the lowest RT-qPCR threshold cycle (C_T) values, corresponding to higher viral load, for each AGE case and stool specimens from healthy controls that tested positive for rotavirus, and the RV1-derived strains were subjected to VP7/VP4 genotyping. VP7 (G) and VP4 (P) genotypes were determined by a conventional RT-PCR and electrophoretyping assay in which genotype was determined by banding pattern as described in the 2009 WHO manual of rotavirus detection and characterization methods (27), but employing updated G12 primers (28) (see the supplemental methods). A subset of samples that was not genotyped using the WHO method due to polymorphisms at the primer binding sites or low viral load were subjected to a method of G/P genotyping designed for samples with a low viral load (29) (supplemental methods). The sequences obtained were analyzed using MEGAv7 software and genotypes were assigned using the Basic Local Alignment Search Tool (BLAST) to query the nucleotide sequences available in GenBank (30).

Development and validation of the RV1 LNA-RTqPCR assay.

We developed an RT-qPCR assay modified from Gautam et al. (24) to differentiate the RV1-derived strain from wild-type rotavirus. We used their previously described primers but designed a new probe that incorporated a nucleic acid analog called locked nucleic acid (LNA) to raise the melting temperature (T_m) (Table 1). The RV1 LNA probe was 5 bp shorter than the original Gautam et al. probe with a Tm of 67.3°C (Table 1).

TABLE 1.

Primers and probe of LNA-RTqPCR assay used to differentiate between Rotarix and wild-type G1P[8] rotavirus

Name	Sequence	Length(bp)	Tm °C	Positiona	Amplicon size (bp)	Reference
Rotarix NSP2-F	GAA CTT CCT TGA ATA TAA GAT CAC ACT GA	29(+)	54.8	546–574	281	24
Rotarix NSP2-R	TTG AAG ACG TAA ATG CAT ACC AAT TC	26(−)	54.1	826–801		24
Rotarix LNA probe	5′ 6-FAM CAA +TA+G+AT T+GAAGTCAGT AAC+GTTT 3′ IBFQ	25(−)	67.3	780–756		this study

^aNucleotide position in Rotarix strain JX943605.

The RV1 NSP2 LNA-RTqPCR was tested on various dilutions of RV1 complementary DNA (cDNA) using TaqMan Universal PCR Master Mix, no AmpErase UNG (Life Technologies, USA). Each reaction contained 5 μl TaqMan Universal PCR Master Mix, 0.5 μl of the NSP2 primer-probe mix (18 μM primer mix and 5 μM probe), 2 μl of PCR-grade nuclease-free water (Invitrogen, USA), and 2.5 μl cDNA. The qPCR conditions consisted of a 3-min initial hold at 95°C then 45 amplification cycles consisting of 15 s at 95°C and 2 min at 60°C on the ABI 7500 real-time PCR analyzer (Applied Biosystems, USA).

Accuracy of the new LNA-RTqPCR was determined using Sanger-sequencing-confirmed RV1 cDNA (i.e., the results of LNA-RTqPCR were compared to conventional NSP2 PCR-Sanger sequencing results, both performed on the commercial RV1 vaccine strains) as the reference standard. Diagnostic sensitivity, diagnostic specificity, and inter-test agreement were calculated (see the supplemental methods). The intraassay precision (repeatability) and interassay precision (reproducibility) were determined by analyzing a series of 10-fold dilutions of RV1 vaccine cDNA in triplicate on two ABI 7500 real-time PCR analyzers carried out independently by two researchers over a period of 5 days (Table S3 in the supplemental material). The linearity and limit of detection (LOD) were established by testing two RV1 vaccine cDNA with 5-point dilution steps in triplicate over 2 days. The cross-reactivity (one aspect of analytic specificity) was determined by testing samples containing wild-type G1P[8], G9P[8], G12P[8], G2P[4], G3P[8], G9P[4], G4P[8], and G2P[8] rotaviruses. The inference (the other aspect of analytic specificity) was determined by testing samples (n = 153) that were negative for rotavirus and positive for genogroup I or II noroviruses, sapovirus, various serotypes of adenovirus, or astrovirus. We followed recommended data analysis and acceptance criteria for the validation of experiments of semiquantitative diagnostic tests (31).

Identification of RV1 vaccine-derived strains in healthy children and children with AGE.

The cDNA of all AGE cases and healthy controls that tested positive for rotavirus in the primary study was retested using our validated RV1 NSP2 LNA-qPCR approach to identify RV1vaccine-derived strains.

Patient clinical characteristics, gastroenteritis severity measure, and RV1 vaccine status.

We compared the clinical characteristics of participants whose stool contained an RV1-derived strain to that of participants infected by the wild-type G1P[8] rotavirus. AGE illness severity was quantified employing the modified Vesikari scale (MVS) score and clinical dehydration scale score reported by parents at the index ED visit (32–34). We defined MVS scores of 0 to 8 as mild, 9 to 10 as moderate, and ≥11 as severe gastroenteritis (32, 33). Doses and dates of RV1 vaccinations were obtained from the Alberta Health Immunization and Adverse Reaction to Immunization Reporting (Imm/ARI) system. We calculated the number of days between vaccination administration and stool sample submission.

Statistical analysis.

The association between the RV1-derived strain and gastroenteritis was evaluated using the attribution methods developed by Platts-Mills et al. (35). The analysis was restricted to infants (defined as <1 year old) who had received at least one dose of rotavirus vaccine. We used logistic regression to determine the association between presence of AGE (case versus control) and the C_T values for the RV1 vaccine-derived strains and other viruses, enteric bacteria, and age in months. The regression model coefficient for RV1 was multiplied by the difference between 38 and the C_T value of each RV1-derived strain-positive AGE case and exponentiated to calculate a case-specific odds ratio (OR). From the case-specific ORs, the attributable fraction (AF) was calculated as (OR−1)/OR. The case-specific ORs were averaged to determine the overall OR for the association between AGE and the RV1-derived strain. The case-specific AFs were averaged to determine the overall AF for the portion of AGE cases testing positive for the RV1-derived strain whose gastroenteritis could be attributed to the RV1-derived strain. The entire process was bootstrapped and bias-corrected with accelerated 95% confidence intervals (CIs) calculated using the boot package in R (36, 37).

The association between RV1-derived strains and AGE severity was evaluated among all cases in terms of presence of fever, duration of diarrhea and vomiting, and MVS and dehydration scores. For each clinical outcome, AGE cases with the RV1-derived strains were compared to cases with wild-type rotavirus and to all AGE cases negative for RV1-derived strains, respectively. Logistic regression was used for fever and linear regression for the other outcomes. Analyses were adjusted for codetected pathogens and age in months.

RESULTS

Development and validation of LNA-RTqPCR for RV1 vaccine NSP2 detection.

The RV1 LNA-RTqPCR detected the RV1 NSP2 gene in all dilutions of RV1 cDNA that were previously confirmed by Sanger sequencing (Fig. 2A). The diagnostic sensitivity and specificity of the RV1 LNA-RTqPCR were 100% (95% CI: 93.4% to 100%) and 99.4% (95% CI: 96.73% to 99.98%), respectively, with excellent agreement (k = 0.988, 95% CI: 0.96 to 1.0). In terms of analytical specificity, only one out of 168 samples described below was positive by LNA-RTqPCR. No cross-reactivity was observed in the rotavirus-positive samples collected prior to the use of RV1 vaccine in Alberta (n = 32), samples with rotavirus genotyped as non-G1P[8] (7 different genotypes, n = 49), samples tested negative for rotavirus but positive for norovirus (n = 27), sapovirus (n = 15), adenovirus (n = 15), or astrovirus (n = 15), and 14 out of 15 samples tested negative for all enteropathogens in healthy controls. The one negative sample from a control that tested positive for the RV1-derived strain by LNA-RTqPCR (C_T = 36.18) was confirmed by sequence as G1P[8] rotavirus (Fig. 2D).

FIG 2.

Signal amplification curves of RV1, wild-type G1P[8], rotaviruses of other genotypes, other enteric viruse-positive and enteropathogen-negative stool samples by Rotarix LNA-RTqPCR. (A) Serial dilutions of RV1 cDNA. (B) Wild-type G1P[8] samples collected prior to RV1 introduction in Alberta. (C) Other rotavirus genotypes. (D) Other enteric virus-positive and enteropathogen-negative samples collected from healthy children. Magenta curves in C and D represent RV1-positive controls. RV1, monovalent rotavirus vaccine Rotarix.

The percent coefficient of variation (% CV) ranged from 0.3 to 1.0 for intraassay precision and 0.8 to 3.9 for interassay precision for the LNA-RTqPCR (Table 2). The lowest copy number detected was 9.86 copies/qPCR. The LNA-RTqPCR efficiency was 99.7% with a slope of −3.33 and an R² value of 0.99 (Fig. S1A).

TABLE 2.

Intraassay and interassay precision of the RV1 LNA NSP2 assay^a

Type	Concn	No. of replicates	Mean CT	SD	% CV
Intraassay	Neat	3	14.6	0.1	1.0
	10−1	3	16.8	0.1	0.5
	10−2	3	20.3	0.2	0.8
	10−3	3	23.9	0.1	0.4
	10−4	3	27.5	0.1	0.3
	10−5	3	31.0	0.2	0.7
Interassay	Neat	3	14.3	0.6	3.9
	10−1	3	16.8	0.1	0.8
	10−2	3	20.4	0.2	1.2
	10−3	7	24.0	0.2	0.8
	10−4	7	27.6	0.4	1.5
	10−5	6	31.4	0.9	2.9

^aC_T, threshold cycle; SD, standard deviation; % CV, percent coefficient of variation, calculated as (SD/mean C_T) × 100%.

Prevalence of RV1-derived strains.

Of the 3,741 AGE cases and 2,187 controls, GVP results were available for 3,329 and 1,369, respectively. Of the cases tested, 574 (17.2%) tested positive for rotavirus, including 44 (1.3%) positive for RV1-derived strains (Fig. 1). Rotavirus was detected in 144 controls (10.5%); all except two tested positive for RV1-derived strains. All cases and controls with RV1-derived strains detected were enrolled between December 2015 and June 2018, i.e., after RV1 was included in the provincial immunization program. The majority of the rotavirus AGE cases in our study were infants from before the provincial implementation of rotavirus vaccine and older children who were ineligible for RV1, while a large number of controls (87%) were vaccinated at the public health clinic before submitting their specimens. The majority with RV1-derived strains, 62.5% (25/40) of gastroenteritis cases (Fig. S2) and 95.1% (117/123) of controls, were enrolled ≤4 weeks after RV1 vaccination and all were <12 months old (Table 3).

TABLE 3.

Rotarix vaccination status of participants testing positive for Rotarix (RV1) strains

Participant type	RV1-positive (n)	Rotarix vaccination (n [%])			Days between vaccination and specimen Collection (n [%])
		1 dose	2 doses	No vaccination documented	≤28 days after 1st dose	≤15 days after 2nd dose
AGE casesa	44	36 (82)	4 (9)	4 (9)	23/36 (64)	2/4 (50)
Controlsb	142	65 (46)	58 (41)	19 (13)	63/65 (97)	54/58 (93)

^aAcute gastroenteritis (AGE) cases between 68 and 131 days old.

^bControl between 69 and 106 days old.

Association of RV1-derived strains and gastroenteritis.

To assess the association between RV1-derived strains and gastroenteritis and to account for confounding by rotavirus vaccination, the analysis was limited to infants with known RV1 vaccine (n = 978). There were 591 AGE cases, including 40 (6.8%) with the RV1-derived strain, and 387 controls, including 123 (31.8%) with the RV1-derived strain (Table 4). After adjustment for age and co-detected pathogens, there was no association between the RV1-derived strain and gastroenteritis (OR 0.90; 95% CI 0.56, 1.55). Similarly, no association between RV1-derived strains and AGE was found in subgroup analyses of children vaccinated <4 weeks prior to study enrollment (OR 1.18; 95% CI 0.56, 2.87) and children <6 months old (OR 0.64; 95% CI 0.40, 1.10).

TABLE 4.

Disease, pathogen, and vaccination status of participants

Test status	Acute gastroenteritis cases (n = 3,741)			Healthy controls (n = 2,187)
	Total	Infants		Total	Infants
		Total	Vaccinateda		Total	Vaccinateda
RV1	44	44	40	142	142	123
Other rotavirus strains	530	75	11	2	2	1
Rotavirus negative	2,755	901	540	1,225	483	263
Untested	412	85	2	818	214	0

^aNumbers in boldface type indicate those included in calculations of the association between RV1 strains and gastroenteritis. The C_T-based odds ratio for the association between Rotarix strains and gastroenteritis was 0.90 (95% CI 0.56, 1.55) after adjustment for age and co-detected pathogens.

Association of RV1-derived strains and gastroenteritis severity.

Cases with RV1-derived strains detected had a mean MVS score of 10.2 (standard deviation [SD] 3.3), compared to 11.4 (SD 3.3) among all AGE cases that tested negative for RV1-derived strains and 13.0 (SD 2.0) among those with wild-type rotavirus infection. After adjustment for age and co-detected pathogens, there was no difference in MVS score between cases with RV1-derived strains and AGE cases that tested negative for RV1. However, AGE cases with RV1-derived strains had an MVS score of −7.2 (95% CI −12.5, −1.8) points lower than those with other rotavirus strains, indicating lower severity of disease (Table 5).

TABLE 5.

Association between Rotarix-derived strains and gastroenteritis severity among gastroenteritis cases, adjusted for age and co-detected pathogens

	Cases with RV1a-derived strains (n = 40)	Cases with other rotavirus strains (n = 11)b		Cases with other rotavirus strains or rotavirus-negative (n = 551)b
Symptom	n (%)	n (%)	OR (95% CI)	n (%)	OR (95% CI)
Fever	7 (18)	8 (73)	0.18 (0, 10.57)	231 (42)	0.36 (0.14, 0.87)
	Mean (SD)	Mean (SD)	Difference (95% CI)	Mean (SD)	Difference (95% CI)
MVSc	10.2 (3.3)	13.0 (2.0)	−7.2 (−12.5, −1.8)	11.0 (3.3)	−1 (−2.1, 0.2)
	Median (IQR)	Median (IQR)	Difference (95% CI)	Median (IQR)	Difference (95% CI)
Diarrhea duration (hours)	97.6 (49.6, 240.0)	96.0 (47.4, 238.9)	−68.2 (−299.5, 163.1)	99.4 (48.0, 191.6)	8.9 (−31.9, 49.7)
Vomiting duration (hours)	35.0 (0, 100.4)	94.4 (44.9, 135.7)	−190.9 (−356.1, −25.6)	53.6 (22.0, 122.1)	−11.2 (−41.4, 19.1)
Dehydration score	1.0 (0, 2.0)	3.0 (1.0, 4.0)	−2.2 (−5.3, 0.9)	2.0 (0, 3.0)	−0.5 (−1.2, 0.1)

^aRV1, monovalent rotavirus vaccine: Rotarix.

^bOR, odds ratio; CI, confidence interval; IQR, interquartile range; SD, standard deviation.

^cMVS, modified Vesikari scale.

DISCUSSION

We developed and validated the LNA-RTqPCR with a novel probe to detect RV1-derived strains in rotavirus-positive samples in a high-throughput and efficient manner. The new RV1 LNA-RTqPCR was highly sensitive (100%) and specific (99.4%) for the RV1 NSP2 gene. We identified RV1-derived strains in 7.7% of all acute rotavirus gastroenteritis cases and 98.6% of healthy controls. Infants with AGE who had RV1 had milder AGE symptoms compared to those with wild-type rotavirus, and we identified no association between RV1-derived strains and acute symptomatic gastroenteritis (0.90; 95% CI 0.56, 1.55). The RV1-derived strains identified in our study may be the true vaccine strain or a vaccine-wild type reassorted strain. Further investigation of the genome composition of these RV1-derived strains by whole-genome sequencing is required to clarify this issue.

LNA is a bicyclic nucleoside analog that has an altered ribose ring containing a methylene linker which connects the 4′-C position to the 2′-O position. The “locked” structure of LNA confers significantly increased affinity to its cDNA target, rendering it very well suited for SNP discrimination (38). Due to its short length, the LNA probe has the advantages of higher sequence specificity due to less spurious binding and better quenching due to reduced distance between the fluorophore and quencher, leading to an improved signal-to-noise ratio (39).

The diagnostic characteristics of the LNA-RTqPCR and the Gautam assay were very similar based on published data (24). The difference in the limit of detection (9.86 versus 2 copies per PCR) could be related to different qPCR standards used. Kaneko et al. reported a false positive sample by the Gautam assay in a sample with wild-type G9P[8] rotavirus confirmed by sequencing the VP7, VP4, VP6, and NSP2 genes (18). Thus far, we found RV1-derived strains identified by our assay to have the G1P[8] genotype in all 44 AGE cases and all 133 controls that we successfully genotyped. However, a sample with a low viral load that was missed by our GVP detection assay tested positive by LNA-RTqPCR, suggesting the LNA-RTqPCR assay has a higher sensitivity than GVP. This was also supported by the observation that the LNA RTqPCR produced lower C_T values for the same rotavirus-positive samples than did GVP RT-qPCR (using the TaqMan probe), which targets the NSP3 gene (unpublished observation). Our findings were consistent with a comparative sensitivity study of four probe chemistries, which reported superior performance of the LNA probe compared to a TaqMan probe and a probe containing minor groove binders (40).

Post vaccination shedding of RV1 has been detected in 35 to 80% of vaccine recipients, mostly during the first 28 days after the initial dose and 15 days after the second dose (8). In the present study, 62.5% of AGE cases and 95.1% of controls with RV1-derived strains in their stool samples had similar time frames of shedding of vaccine strains by the dates of vaccination.

Our findings were supported by a UK study conducted between 2006 and 2016 which found 8.2% of rotavirus-positive samples collected mostly from children with gastroenteritis contained RV1-derived strains. This positivity rate likely represented vaccine shedding, as RV1 was mostly detected in vaccinated children right after receiving vaccination in both studies (41). The prevalence of RV1 was low in AGE cases and high in healthy controls. Our study differs from a Japanese study and two studies from the United States in positivity rates of RV1 in children with gastroenteritis (14, 18, 29). The low rates observed in the Japanese study (1.6%, 6/372) and two U.S. studies (0.6%, 1.2%) are likely due to the small sample size in the Japanese study (n = 372) and the fact that RotaTeq is the more widely used vaccine in the United States (29). A study of Hungarian children under 5 years of age with gastroenteritis found a high percentage (17.3%) of RV1 strains in a subset of rotavirus G1P[8] positive stool specimens; however, these 55 strains were identified by VP7 and VP4 sequencing, not NSP2 RT-qPCR, and were closely related to, but not identical to, the RV1 parental strain (23).

Our analysis showed no association between RV1-derived strains and the presence of AGE after adjusting for age and co-detected pathogens, providing no evidence for RV1 as a cause of gastroenteritis in a vaccinated cohort of infants. However, among those with gastroenteritis, cases with RV1-derived strains had less severe disease than children with other rotavirus strains detected, indicated by a lower MVS and shorter vomiting duration. The unadjusted mean of the MVS was 10 (moderate) for children with RV1-derived strains and 13 (severe) (33) for children with wild-type rotavirus strains. After adjustment, the difference between the two groups increased to 7, further highlighting the contrast between these illnesses. These results are consistent with those of Gower et al. (42), who found an unadjusted difference in severe MVS prevalence between gastroenteritis cases with vaccine-derived and wild type of 9.8% versus 37.5%. With a larger sample size, they also found statistically significant differences in some severity indicators we did not (i.e., fever and dehydration), though they did not adjust for age or co-detected pathogens (42).

Asymptomatic wild-type rotavirus infection occurred in two vaccinated controls in this study. Gunawan et al. similarly identified two cases of asymptomatic wild-type rotavirus infections of G1P[8] and equine-like G3P[8] in vaccinated children (43). Rotavirus asymptomatic infections in vaccinated children are not well characterized. Rotavirus vaccination mimics the first natural infection to protect children from gastroenteritis with severe consequences but not subsequent infections by wild-type rotaviruses (44, 45).

The emergence of disease-causing vaccine revertants or wild type-vaccine reassortant viruses is a concern with live-attenuated vaccines (20). We showed that the clinical severity of AGE in RV1 NSP2-positive patients was lower than that of the wild-type G1P[8] rotavirus and other rotavirus strains (Table S3). It should also be noted that even among those without a co-detected pathogen, their gastroenteritis symptoms may have been caused by another enteropathogen for which we did not test (e.g., Aichi virus) (46) or reflect a vaccination side effect (47); thus, we cannot conclusively state that their symptoms were due to rotavirus infection.

A limitation of this study was that identification of genetic changes in the antigenic regions of RV1 and their impact on recipients was not evaluated. We did not compare VP7 and VP4 sequences of the RV1-derived strains phylogenetically to the RV1 sequences or its parental strain 89-12. The detection of significant mutations in the shed strains may provide clues to the link between the associated clinical symptoms and RV1. Moreover, because we used LNA-RTqPCR and G/P typing, we could not distinguish a sample positive for RV1-derived strains from a sample with both RV1 and wild-type G1P[8]. Additionally, whole-genome sequencing of these RV1-derived strains was not performed. Whether these rotaviruses represent revertants of the attenuated RV1 G1P[8] strain or reassortants of the vaccine strain with another wild-type strain is unclear. Whole-genome sequence analysis and phylogenetic studies (48, 49) should be performed to clarify the evolution of RV1 vaccine strains and their clinical impact. Nevertheless, we have shown that regardless of their genome composition, these RV1-derived strains are clinically benign. Bias may exist due to the enrollment of healthy children attending the public health clinic for routine vaccinations, including RV1. However, in the analysis of only children who were recently vaccinated, we demonstrated that even when both cases and controls had equal likelihood of exposure, the RV1-derived strain was still not associated with gastroenteritis.

In conclusion, our study validated and supported the use of a novel assay to identify vaccine related strains. In contrast to the clinical manifestations of wild-type rotavirus, RV1-derived strains were not associated with the development of severe AGE. However, monitoring vaccine-related safety concerns in areas where rotavirus vaccine are universally administered as a part of continued molecular surveillance of rotavirus infections remains necessary in the post-vaccine-introduction era.