**Retrospective Case-Control Study of 2017 G2P[4] Rotavirus Epidemic in Rural and Remote Australia**

**Bianca F. Middleton 1,2,\* , Margie Danchin 3,4,5, Helen Quinn 6,7, Anna P. Ralph 1,8 , Nevada Pingault <sup>9</sup> , Mark Jones <sup>10</sup> , Marie Estcourt <sup>10</sup> and Tom Snelling 1,11,12**


Received: 24 August 2020; Accepted: 22 September 2020; Published: 265 September 2020 -

**Abstract:** Background: A widespread G2P[4] rotavirus epidemic in rural and remote Australia provided an opportunity to evaluate the performance of Rotarix and RotaTeq rotavirus vaccines, ten years after their incorporation into Australia's National Immunisation Program. Methods: We conducted a retrospective case-control analysis. Vaccine-eligible children with laboratory-confirmed rotavirus infection were identified from jurisdictional notifiable infectious disease databases and individually matched to controls from the national immunisation register, based on date of birth, Aboriginal status and location of residence. Results: 171 cases met the inclusion criteria; most were Aboriginal and/or Torres Strait Islander (80%) and the median age was 19 months. Of these cases, 65% and 25% were fully or partially vaccinated, compared to 71% and 21% of controls. Evidence that cases were less likely than controls to have received a rotavirus vaccine dose was weak, OR 0.79 (95% CI, 0.46–1.34). On pre-specified subgroup analysis, there was some evidence of protection among children <12 months (OR 0.48 [95% CI, 0.22–1.02]), and among fully vs. partially vaccinated children (OR 0.65 [95% CI, 0.42–1.01]). Conclusion: Despite the known effectiveness of rotavirus vaccination, a protective effect of either rotavirus vaccine during a G2P[4] outbreak in these settings among predominantly Aboriginal children was weak, highlighting the ongoing need for a more effective rotavirus vaccine and public health strategies to better protect Aboriginal children.

**Keywords:** rotavirus; rotavirus vaccines; vaccine effectiveness; case control

#### **1. Introduction**

Rotavirus is a leading cause of severe dehydrating diarrhoeal illness in children and continues to be responsible for the deaths of 118,000 to 183,000 children every year [1]. Many of these deaths occur in resource-poor settings [2].

In 2006, two oral rotavirus vaccines, Rotarix and RotaTeq, were licensed for use and in 2009 the World Health Organization endorsed their use globally [3]. Subsequent epidemiological studies have confirmed a strong protective effect of vaccination on rotavirus morbidity in high- and upper middle-income countries (vaccine efficacy [VE] >84%) [2]. However, in low-income countries, despite a large reduction in the absolute number of cases of gastroenteritis, measured vaccine efficacy has been lower (45–57%) and in some settings there is evidence of decreased protection in the second year of life [2,4–6].

The incorporation of rotavirus vaccines into the Northern Territory immunisation schedule in 2006 and then into the Australian National Immunisation Program (NIP) in 2007, resulted in a substantial and sustained decrease in rotavirus hospitalisations [7]. However, among Aboriginal and Torres Strait Islander children living in the hyperendemic settings of rural and remote Australia, the decrease in rotavirus hospitalisation was less dramatic and not sustained, with Aboriginal children living in the Northern Territory (NT) remaining more than 20 times more likely to be hospitalised with rotavirus than their non-Aboriginal counterparts [7]. An early vaccine effectiveness study in this setting also suggested reduced effectiveness against heterotypic strains and poor protection in the second year of life [8].

In 2017, an epidemic of G2P[4] rotavirus arose in the Northern Territory and subsequently spread to adjoining rural and remote regions of Western Australia (WA). These two jurisdictions cover a large geographic area which is sparsely populated; they have a higher proportion of resident Aboriginal and Torres Strait Islander people, many of whom live in rural and remote communities. The rotavirus epidemic occurred at a time when the Northern Territory exclusively administered Rotarix and Western Australia exclusively administered RotaTeq as part of the jurisdictional implementation of the NIP. We evaluated the protective effectiveness of both vaccines in these high-burden settings, ten years after the incorporation of rotavirus vaccines into the NIP.

#### **2. Materials and Methods**

#### *2.1. Study Setting*

The Alice Springs and Barkly regions of the Northern Territory, and the Kimberley, Pilbara and Goldfields regions of Western Australia are large but sparsely populated administrative health regions. Ranging from the semi-arid south, to the arid center and tropical north, these five regions encompass more than 2,500,000 km<sup>2</sup> , but are home to a combined total of just 174,000 people [9]. Children aged <5 years represent between 7–9% of the population, and between 5 and 41% of the population in each of these regions identify as being Aboriginal and/or Torres Strait Islander (hereafter respectfully referred to as 'Aboriginal') [9]. Many of these children live in towns or small remote communities. Rotarix and RotaTeq rotavirus vaccines have been licensed for use in Australia since June 2006. The Northern Territory immunisation program has funded the administration of Rotarix exclusively since October 2006. The Western Australian immunisation program funded the administration of Rotarix from July 2007 to June 2009, RotaTeq from July 2009 to June 2017, and Rotarix from July 2017.

#### *2.2. Study Design*

We conducted a retrospective, population-based, case control study of children age-eligible for at least 1 dose of rotavirus vaccine (those born after the introduction of Rotarix rotavirus vaccine to the NT schedule—after 1 July 2006 and aged ≥6 weeks, and those born after the introduction of RotaTeq rotavirus vaccine to the WA schedule—after 1 May 2009 and aged ≥6 weeks) who had laboratory positive and notified rotavirus infection during the 2017 G2P[4] rotavirus epidemic in the NT and WA. Cases were individually matched to controls sampled from the national immunisation register. As a secondary analysis, we also compared cases who were age-eligible for full rotavirus vaccination (those born after 1 July 2006 and aged ≥24 weeks in the NT and those born after 1 May and aged ≥32 weeks in WA) with un-matched control children diagnosed with non-rotavirus gastrointestinal infections sampled from disease notification registers.

#### *2.3. Data Sources*

Rotavirus is a notifiable disease in the NT and WA. Data regarding rotavirus cases and disease register controls were ascertained from the two jurisdictional-based notifiable infectious disease databases—The Northern Territory Notifiable Disease System (NTNDS) managed by the NT Centre for Disease Control, and the Western Australian Notifiable Infectious Disease Database (WANIDD) managed by the WA Department of Health.

To estimate baseline vaccine coverage in the case-referent population, matched population controls were sampled from the Australian Immunisation Register (AIR), a comprehensive population-based register which contains vaccination data for all children registered with Australia's universal health insurance scheme, Medicare (~99% of the population).

#### *2.4. Participants*

#### 2.4.1. Population-Based Analysis

Rotavirus cases were vaccine-eligible children aged ≥6 weeks with laboratory positive and notified rotavirus infection between 1 March and 30 June 2017. Cases were drawn from the Alice Springs and Barkly regions of the NT, and the Kimberley, Pilbara and Goldfields regions of WA. To be vaccine-eligible, children had to be born on or after 1 July 2006 in the NT (for Rotarix) and on or after 1 May 2009 in WA (for RotaTeq).

De-identified population controls were selected from the Australian Immunisation Register and matched to each case by date of birth (±14 days), Aboriginal status and location of residence (listed residential postcode within either the Alice Springs, Barkly, Pilbara, Goldfields or Kimberley regions). Up to 10 eligible controls were randomly selected for each case.

#### 2.4.2. Disease Register Analysis

Rotavirus cases were selected as above, but because individual matching was not feasible, the analysis was restricted to children old enough to be fully vaccinated: age ≥24 weeks (for Rotarix) in the NT and ≥32 weeks (for RotaTeq) in WA.

Disease register controls were vaccine-eligible children (aged ≥24 weeks or ≥32 weeks in the NT and WA respectively), with microbiologically confirmed, non-rotavirus and non-vaccine preventable, notifiable gastrointestinal infections, notified between 1 January and 31 December 2017. Controls were selected form the Alice Springs and Barkly regions of the NT, and the Kimberley, Pilbara and Goldfields regions of WA. Non-rotavirus notifiable gastrointestinal infections included campylobacter, shigella, salmonella and cryptosporidium, and controls were excluded if they were also identified as a rotavirus case. Age, Aboriginal status, sex and location of residence were obtained from the disease register for inclusion in the regression analysis.

#### *2.5. Immunisation Status*

The immunisation status of all rotavirus cases, population controls and disease register controls were determined from the Australian Immunisation Register (AIR). Full vaccination was defined as AIR-documented receipt of at least two doses of Rotarix for children living in the NT and at least three doses of RotaTeq for children living in WA. Partial vaccination was defined as AIR-documented receipt of one dose only of Rotarix for children living in the NT and either one or two doses only of RotaTeq for children living in WA. Unvaccinated children were defined as those registered on the AIR, but without documented receipt of any rotavirus vaccines. In circumstances where a child had a vaccine dose recorded as dose two or dose three on the register, but where an earlier dose was not recorded, it was assumed the missing dose had been given [10]. A vaccine dose was considered administered on the date recorded as administered on the register (i.e., without any post-vaccination censoring). A vaccine dose was considered invalid if (1) administered too early (before six weeks of age or <28 days from prior vaccine dose), (2) it exceeded the recommended number of vaccine doses in the schedule (>2 doses of Rotarix or >3 doses of RotaTeq) or (3) the administered vaccine was different to the prior vaccine (mixed Rotarix/RotaTeq vaccination schedule). Children were excluded from selection as cases and controls if they had an invalid vaccine dose. Children were also excluded from the analysis if they were recorded as having received the non-programmatic vaccine for their resident jurisdiction (i.e., RotaTeq but living in the NT, or Rotarix but living in WA).

#### *2.6. Statistical Analysis*

Conditional logistic regression was used to determine the odds ratio (OR) of vaccination for rotavirus cases compared with matched population controls from the immunisation register. Additional models were fit to compute the OR for any dose of vaccine (full and/or partial vaccination) vs none, full vaccination vs none, partial vaccination vs none, and full vs partial vaccination. Subgroup analyses were by jurisdiction (NT versus WA), and by age (<12 months versus ≥12 months).

For the disease register analysis, ordinary logistic regression was used to determine the odds ratio of vaccination for rotavirus cases compared with disease register controls. Age (months), sex, Aboriginal status (Aboriginal vs non-Aboriginal) and jurisdiction of residence (NT vs WA) were included in the model, together with an interaction term for Aboriginal status and jurisdiction of residence.

Assuming a baseline population vaccine coverage of 80%, we estimated that 80 matched sets of cases and population controls, with 10 controls for each case, would have at least 80% power to detect a significant real-world vaccine effectiveness of 45% (OR = 0.55).

All analysis was performed using Stata, version 15.1 (Stata).

#### *2.7. Ethics Committee Approvals*

Approval was granted by the Central Australian Human Research Ethics Committee (CAHREC 18-3219), the Human Research Ethics Committee of the Northern Territory Department of Health and Menzies School of Health Research (HREC 18-3248), the Department of Health Western Australian Human Research Ethics Committee (DOH HREC 2018/30), the Western Australian Aboriginal Health Ethics Committee (HREC 891) and the Charles Darwin University Human Research Ethics Committee (H19040). Approval to access data held by the Australian Immunisation Register was granted by the Australian Government Department of Health.

#### **3. Results**

The rotavirus epidemic occurred between 1 March and 30 June 2017. A total of 194 vaccine-eligible children aged ≥6 weeks were identified as rotavirus cases from which 171 were eligible for inclusion in the study (see Figure 1).

The median age of rotavirus infection was 19 months (range from 1 to 94 months). Most rotavirus cases were among children who identified as Aboriginal and/or Torres Strait Islander (NT 86%, WA 75%). Genotype results were available for only 60% of rotavirus cases, however, of those typed, all were G2P[4] strains. A total of 99 children were documented as having been hospitalised with rotavirus infection—78% of rotavirus cases in the NT and 39% of rotavirus cases in WA. Hospitalisation status was unknown for 15% of WA rotavirus cases (see Table 1).

**Figure 1.** Selection of rotavirus cases and matched population controls from the Australian Immunisation Register. \* Vaccine-Eligible: children eligible by date of birth to have received at least one dose of Rotarix vaccine (those born after 1 July 2006 in the Northern Territory) or at least one dose of RotaTeq vaccine (those born after 1 May 2009 in Western Australia).

Among rotavirus cases, 65% were fully vaccinated, 25% partially vaccinated and 10% unvaccinated; among matched population controls from the immunisation register, 71% were fully vaccinated, 21% partially vaccinated and 8% unvaccinated. In the population-based analysis, the odds ratio of receipt of any doses of rotavirus vaccine versus none was 0.79 (95% CI, 0.46–1.34). For the NT and WA, the OR of any doses versus none was 1.10 (95% CI, 0.50–2.41) and 0.56 (95% CI, 0.27–1.16), respectively. For children aged <12 months and for children aged ≥12 months, the ORs were 0.48 (95% CI, 0.22–1.02) and 1.22 (95% CI, 0.55–2.73), respectively. The OR of full versus partial vaccination was 0.65 (95% CI, 0.42–1.01) (see Table 2 and Figure 2).

Of the 171 notified rotavirus cases above, 149 were age eligible for inclusion in the disease register analysis (aged ≥24 weeks or ≥32 weeks in the NT and WA, respectively). A total of 347 vaccine-eligible children were identified as having non-rotavirus gastrointestinal infections in the twelve-month period from 1 January and 31 December 2017. Of these children, 299 were eligible for inclusion (Supplementary Materials Figure S1). The median age of disease register controls was older than that of rotavirus cases, 29 months vs. 20 months (Supplementary Materials Table S1). Disease register controls were less

likely to be hospitalised than rotavirus cases (35% vs 58%) and, in WA, were less likely to identify as Aboriginal (41% vs 74%).

In the disease register analysis, 73%, 19% and 8% of cases were fully vaccinated, partially vaccinated and unvaccinated, respectively, compared with 83%, 12% and 5% of controls. The adjusted OR of any doses of rotavirus vaccine versus none was 0.58 (95% CI, 0.24–1.39); for WA and NT children, the adjusted ORs were 0.30 (95% CI, 0.09–0.98) and 1.40 (95% CI, 0.34–5.80), respectively, and for children aged <12 months and ≥12 months old, the adjusted ORs were 0.28 (95% CI, 0.03–2.83) and 0.81 (95% CI, 0.29–2.28), respectively. The adjusted OR of full vs. partial vaccination was 0.63 (95% CI, 0.35–1.13) (see Table 2 and Supplementary Materials Table S2).


**Table 1.** Baseline characteristics of rotavirus cases.


**Table 2.** Odds ratio of vaccination in rotavirus cases versus controls in the population-based analysis and the disease register analysis.

≥ ≥

≥

Additional analyses were performed as requested after peer review—including restricting the population-based analysis to children age-eligible for full vaccination only (aged ≥24 weeks in the NT and ≥32 weeks in WA), restricting the population-based analysis to children aged <5 years and restricting the population-based analysis to Aboriginal children only. An additional analysis was also run without the 'missing dose assumption', i.e., in circumstances where a child had a vaccine dose recorded as dose two or three on the register but where an earlier dose was not recorded, the cases and controls were reclassified as 'partially vaccinated' (Supplementary Materials Table S3). This resulted in the reclassification of 26 population controls as partially vaccinated, but no change to the classification of rotavirus cases. The results of the additional analyses were broadly in keeping with the per-protocol analysis.

#### **4. Discussion**

In the context of a G2P[4] rotavirus epidemic with 171 laboratory confirmed rotavirus notifications, we failed to find evidence that either rotavirus vaccine provided strong protection against rotavirus gastroenteritis. This contrasts with the large decrease in rotavirus morbidity and mortality observed globally in young children following the licensing of the oral two rotavirus vaccines, Rotarix and RotaTeq, in 2006 [2,7,11].

The 2017 G2P[4] rotavirus epidemic in the Northern Territory and adjoining regions of rural and remote Western Australia predominantly affected Aboriginal and Torres Strait Islander children (NT 86%, WA 75%). Two thirds of cases (65%) were fully vaccinated, and cases were only slightly less likely to have received a vaccine dose than matched population controls sampled from the immunisation register (OR of 0.79 is equivalent to a VE of 21% where VE = 1—OR). There was some evidence of protection among the subgroup of children <12 months old, although all 95% confidence intervals included one (no effect) and there was significant overlap in the confidence intervals across the subgroup analyses. We found little evidence of a protective effect for full vaccination overall (OR of full vs. no vaccination 0.83 (95% CI, 0.43, 1.58)), although there was some evidence that fully vaccinated children were better protected than unvaccinated children in Western Australia (OR of full vs no vaccination for WA 0.40 (95% CI, 0.18–0.93)). We also found some evidence that fully vaccinated children were moderately better protected than partially vaccinated children (OR of full vs. partial vaccination 0.65 (95% CI, 0.42–1.01)). These findings are consistent with recently published vaccine effectiveness studies evaluating the performance of Rotarix in New South Wales and both Rotarix/RotaTeq in Western Australia. In both studies, VE estimates were highest for fully vaccinated children aged <12 months, and there was evidence of increasing vaccine effectiveness with increasing doses of both Rotarix and RotaTeq vaccines [12,13].

Rotarix is a live, monovalent, attenuated oral rotavirus vaccine derived from the most common human rotavirus strain G1P[8], and RotaTeq is a pentavalent (G1, G2, G3, G4, P[8]) human–bovine reassortant vaccine [14]. While post-licensure studies have reported similar vaccine effectiveness levels for Rotarix and RotaTeq [2], very few studies have directly compared the effectiveness of each vaccine in the same setting or during the same outbreak [15–17]. While there is good evidence that RotaTeq is protective against G2P[4] strains [18], post-licensure studies have shown mixed results for the effectiveness of Rotarix against G2 strains [8,19] and in some jurisdictions using Rotarix, G2P[4] has emerged as the dominant circulating genotype [20–23]. An earlier study of a 2009 G2P[4] outbreak amongst NT Aboriginal infants failed to show that the rotavirus vaccine provided strong protection (OR 0.81 (95% CI, 0.32–2.05)) [8]. In our study, all rotavirus samples sent for genotypic analysis from the five administrative health regions between March and June 2017 were identified as G2P[4]. Given the epidemic was well-defined in time and geography, it is reasonable to assume that G2P[4] accounted for all epidemic cases; this study provides a unique opportunity to evaluate the performance of both Rotarix and RotaTeq during the same G2P[4] epidemic and in similar, albeit geographically distinct, populations. While the point estimate of the OR was consistently lower in the jurisdiction using Rotateq (consistent with better effectiveness), the confidence intervals were wide and overlapping. Small rotavirus case numbers in both jurisdictions and programmatic differences in how cases are ascertained limit our ability to draw conclusions about the comparative effectiveness of the vaccines in this study.

While there was evidence of a protective effect among younger children, our estimates suggest that a strong protective effect of vaccination is unlikely among older children. The median age of rotavirus infection was 19 months with a substantial proportion of cases occurring among children aged 12–23 months (NT 44%, WA 39%). Decreased vaccine protection in the second year of life and persistent burden of rotavirus disease have been reported in other high-burden low-resource settings [2,5,24]. Possible determinants of poor vaccine response include high levels of maternally-derived, vaccine-neutralising anti-rotavirus antibodies, poor infant nutrition, intestinal microbiota imbalance, environmental enteropathy, comorbid infections such as HIV and a high diversity of circulating rotavirus strains [25]. In the population included in this study, children are very unlikely to have been HIV infected, but other infective comorbidities are common. Apart from reduced vaccine-induced protection, programmatic restrictions, including upper age-limits for rotavirus vaccine administration may also diminish the program. An early rotavirus vaccine, RRV-TV, caused intussusception in a small number of vaccinated older infants [26] and despite reassuring phase 3 clinical trial safety results, the manufacturers of Rotarix and RotaTeq have conservatively recommended upper age limits on the administration of their vaccines—24 weeks for Rotarix and 32 weeks for RotaTeq. In practice, this limits opportunity to complete the full vaccination schedule and eliminates the possibility of catch-up of missed vaccinations in later childhood [25]. Delayed and/or incomplete vaccination is more common among Australian Aboriginal children [27] and in one observational study, two-dose DTPa coverage increased by a further 16% after the upper age limit of rotavirus vaccine administration (from 75% to 91% in Aboriginal infants), whereas two-dose rotavirus vaccine coverage increased by only 3% (from 75% to 78% in Aboriginal infants) [28]. This suggests that relaxing the upper age restrictions for rotavirus vaccines, as recommended by WHO for countries with high rotavirus burden [3], could be considered as a strategy for improving vaccine uptake and schedule completion.

The validity of case-control methods is largely dependent on adequate control of confounders, that is, factors which are causally related to both vaccination and baseline risk of disease [29]. In our setting, vaccination coverage is influenced by age, Aboriginal status, geographical location and calendar time; age and Aboriginal status remain the two strongest baseline risk factors for rotavirus gastroenteritis requiring hospitalisation [7], and epidemics are clustered in geographic space and time. Our study therefore sought to control for these potential confounders by directly matching cases to population controls on age (date of birth), Aboriginal status and location of residence, and by confining the analysis to the defined outbreak period. In the disease register analysis, these factors were not matched but were captured and adjusted for in the regression analysis. This study could not directly measure socio-economic status for individual cases and controls, although Indigenous status and remoteness of residence may be considered surrogate measures, with the Alice Springs, Barkly, Kimberley and Goldfields regions encompassing some of the most socially disadvantaged regions in Australia, as measured by the Index of Relative Socioeconomic Advantage and Disadvantage.

While the jurisdiction-based notifiable infectious disease databases are believed to capture all laboratory-confirmed rotavirus cases during the epidemic, we acknowledge that not all children with rotavirus gastroenteritis present for medical care, are referred for testing, or complete testing when it is recommended. Rotavirus vaccines have been found to be more effective in preventing severe disease requiring hospitalisation than asymptomatic and other less severe forms of infection [2]. While we were not able to directly ascertain disease severity, most cases in this study are likely to have had either moderate or severe gastroenteritis because all sought medical care (in order to be hospitalised), and 78% and 39% were hospitalised in the NT and WA respectively.

It is also acknowledged that the propensity to seek medical care for rotavirus gastroenteritis symptoms may be associated with the propensity to access medical care for other reasons, including vaccination, and this is a potential source of bias in the population-based analysis which may have caused us to underestimate vaccine protection. The disease register analysis is less likely to be affected by this bias because the vaccination status of rotavirus cases was compared to that of other children with (non-vaccine preventable), notifiable gastrointestinal clinical infections, i.e., children with clinical presentations which are likely to have been indistinguishable from rotavirus infection and who also underwent microbiological testing. The results of the disease register nested analysis were limited by small numbers, especially in the subgroup analyses, but were in broad agreement with the population-based analysis.

Rotavirus gastroenteritis cannot be reliably distinguished from other causes of non-bloody diarrhea on clinical grounds, and so only laboratory confirmed cases reported to the notifiable infectious disease databases were included. The sensitivity and specificity for detecting rotavirus in stool samples using commercially available EIA is high, although false positives and false negatives have been reported [30]. This is noted as a limitation of the nested disease register case-control study, where an assay error may result in misclassification of a case as a control, or vice versa, which would have caused us to underestimate vaccine protection.

While the Australian Immunisation Register provides credible individual and population-level data regarding vaccine coverage by vaccine type, date-of-birth, location of residence and Aboriginal status, controls were matched to cases based on their location of residence, as recorded on the register in October 2019, which may or may not accurately reflect their jurisdiction of residence between March and June 2017. It is unclear what, if any, bias this may have caused.

#### **5. Conclusions**

The incorporation of two rotavirus vaccines into the Australian NIP in 2007 has resulted in a substantial and sustained decrease in rotavirus morbidity across most of Australia, although Aboriginal and Torres Strait Islander children remain at increased risk of severe rotavirus disease requiring hospitalisation [7]. Our evaluation of the 2017 G2P[4] rotavirus epidemic in remote Australia suggests that rotavirus vaccination provided little protection against notifiable rotavirus disease for children living in rural and remote Australia, with the likely exception of children aged <12 months for whom moderate evidence of protection was found.

The admission of an additional 99 children with gastroenteritis to small regional and remote hospitals over fourteen weeks highlights the ongoing public health importance of rotavirus and the need for strategies to better protect Aboriginal children. Our data indicate a likely benefit from full rather than partial vaccination, underscoring the importance of completing the rotavirus schedule. Schedule completion could be enhanced by relaxing the upper age limit of rotavirus vaccination as has been recommended by the World Health Organisation for high-burden settings [3].

Our study also reports a high percentage of rotavirus cases in children aged 12–23 months and decreased vaccine protection among children older than 12 months. It is plausible that administering an additional or booster dose of rotavirus vaccine to slightly older children (beyond manufacturer upper age limit restrictions) may extend protection into the second year of life. Scheduling a third dose of Rotarix vaccine (at between 6 and 11 months old) is currently under investigation in the NT [31].

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-0817/9/10/790/s1, Figure S1: Selection of rotavirus cases and un-matched disease register controls for disease register nested case-control study. Table S1. Baseline characteristics of rotavirus cases and disease register controls for the unmatched disease-register nested case-control study. Table S2. Odds Ratio of vaccination in rotavirus cases versus controls in the matched population-based analysis and the disease register nested analysis (full results). Table S3. Odds Ratio of vaccination in rotavirus cases versus controls in additional population-based analysis (i) children age-eligible for full vaccination only (aged ≥ 24 weeks in the NT and ≥32 weeks in WA), (ii) children aged <5 years only, (iii) Aboriginal children only, and (iv) 'missing dose assumption' removed.

**Author Contributions:** Conceptualization, T.S., M.D. and B.F.M.; methodology, T.S., M.D., H.Q., A.P.R. and M.E.; formal analysis, T.S. and B.F.M.; data curation, B.F.M. and H.Q.; writing—original draft preparation, B.F.M.; writing—review and editing, T.S., M.D., H.Q., A.P.R., M.E., N.P. and M.J.; supervision, T.S. and M.D.; project administration, B.F.M.; funding acquisition, B.F.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** B.F.M. is supported by National Health and Medical Research Council (NHMRC) Post-Graduate Scholarship Grant (1134095), RACP Paediatrics and Child Health Division NHMRC Scholarship, Australian Academy of Science Douglas and Lola Douglas Scholarship. T.S. and A.P.R. are supported by a NHMRC Career Development Fellowship (1111657 and 1142011 respectively). M.D. is supported by a David Bickart Clinician Research Fellowship, University of Melbourne.

**Acknowledgments:** We acknowledge the support of the Menzies Child Health Indigenous Reference Group, the Kimberley Aboriginal Health Planning Forum and the Pilbara Aboriginal Health Planning Forum. We also acknowledge the support and assistance of Peter Markey and Heather Cook at the Northern Territory Centre for Disease Control, Paul Effler, Robyn Gibbs, Carolien Giele and Clare Huppatz from the Western Australian Communicable Disease Control Directorate, Rob Baird from Territory Pathology, and Julie Bines and Susie Roczo-Farkas from the Enteric Diseases Group, Murdoch Children's Research Institute.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Characterisation of a G2P[4] Rotavirus Outbreak in Western Australia, Predominantly Impacting Aboriginal Children**

**Celeste M. Donato 1,2,3,\* , Nevada Pingault <sup>4</sup> , Elena Demosthenous 1,3, Susie Roczo-Farkas <sup>1</sup> and Julie E. Bines 1,2,5**

	- nevada.pingault@health.wa.gov.au

**Abstract:** In May, 2017, an outbreak of rotavirus gastroenteritis was reported that predominantly impacted Aboriginal children ≤4 years of age in the Kimberley region of Western Australia. G2P[4] was identified as the dominant genotype circulating during this period and polyacrylamide gel electrophoresis revealed the majority of samples exhibited a conserved electropherotype. Full genome sequencing was performed on representative samples that exhibited the archetypal DS-1-like genome constellation: G2-P[4]-I2-R2-C2-M2-A2-N2-T2-E2-H2 and phylogenetic analysis revealed all genes of the outbreak samples were closely related to contemporary Japanese G2P[4] samples. The outbreak samples consistently fell within conserved sub-clades comprised of Hungarian and Australian G2P[4] samples from 2010. The 2017 outbreak variant was not closely related to G2P[4] variants associated with prior outbreaks in Aboriginal communities in the Northern Territory. When compared to the G2 component of the RotaTeq vaccine, the outbreak variant exhibited mutations in known antigenic regions; however, these mutations are frequently observed in contemporary G2P[4] strains. Despite the level of vaccine coverage achieved in Australia, outbreaks continue to occur in vaccinated populations, which pose challenges to regional areas and remote communities. Continued surveillance and characterisation of emerging variants are imperative to ensure the ongoing success of the rotavirus vaccination program in Australia.

**Keywords:** rotavirus; outbreak; Aboriginal; Indigenous; G2P[4]; gastroenteritis; Western Australia; whole genome sequencing; vaccine

#### **1. Introduction**

Group A rotaviruses, belonging to the Reoviridae virus family, remain one of the main aetiological agents of acute gastroenteritis in infants and young children worldwide, estimated to have caused 128,500 deaths and 258,173,300 episodes of diarrhea among children <5 years of age in 2016 [1]. The substantial decrease in the global burden of rotavirus disease over the last decade can be attributed to varied public health measures, such as improved sanitation, as well as the inclusion of rotavirus vaccines into the National Immunisation Programs (NIPs) of over 100 countries worldwide [2]. In Australia, the live-attenuated vaccines Rotarix® (monovalent, human G1P[8] strain) and RotaTeq®(pentavalent, humanbovine reassortant vaccine comprising G1P[5], G2P[5], G3P[5], G4P[5], and G6P[8] strains) were introduced into the NIP in mid-2007, with a state-based vaccine selection method in place up until mid-2017, after which a national tender process was initiated, with all states and territories now using Rotarix [3,4].

**Citation:** Donato, C.M.; Pingault, N.; Demosthenous, E.; Roczo-Farkas, S.; Bines, J.E. Characterisation of a G2P[4] Rotavirus Outbreak in Western Australia, Predominantly Impacting Aboriginal Children. *Pathogens* **2021**, *10*, 350. https:// doi.org/10.3390/pathogens10030350

Academic Editor: David Allen

Received: 9 February 2021 Accepted: 12 March 2021 Published: 16 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Group A rotavirus strains are classified into G and P genotypes based on the outer capsid proteins VP7 and VP4, respectively. To date, 36 G types and 51 P types have been characterised from humans and varied animal species [5]. The most prevalent genotypes in humans are G1, G2, G3, G4, G9, and G12, in combination with P[4], P[6], and P[8] [6,7]. A whole genome classification nomenclature has been developed to describe the genome constellation of strains; Gx-P[x]-Ix-Rx-Cx-Mx-Ax-Nx-Tx-Ex-Hx, denoting the VP7-VP4- VP6-VP1-VP2-VP3-NSP1-NSP2-NSP3-NSP4-NSP5/6 genes, with x referring to the various recognised genotypes for each gene. There are three major genotype constellations: Wa-like (G1-P[8]-I1-R1-C1-M1-A1-N1-T1-E1-H1), DS-1-like (G2-P[4]-I2-R2-C2-M2-A2-N2-T2-E2- H2), and AU-1-like (G3-P[9]-I3-R3-C3-M3-A3-N3-T3-E3-H3) [8].

Western Australia (WA) is the largest state in Australia and is sparsely populated, with 80% of the 2.5 million residents residing in the capital city of Perth. Approximately 4% of the WA population identify as Indigenous (hereafter respectfully referred to as Aboriginal to recognise that Aboriginal people are the original inhabitants of WA), the proportion is higher outside Perth [9]. The Kimberley (KIMB) is a remote region that encompasses an area of 421,451 square kilometres. In 2016, the population was 36,392, and 45% of the region's population identified as Aboriginal, living in towns and communities of varying sizes. The KIMB region has a younger population compared to other regions of WA, with a higher percentage of children aged 0–14 years (25%) [10]. For the period 2011–2015, the enteric disease notification rate (salmonellosis, cryptosporidiosis, rotavirus, campylobacteriosis, and shigellosis) for children in the KIMB region was 5.2 times higher than for all children in WA, with rotavirus accounting for 5% of notifications. For all enteric infection notifications, the rate for Aboriginal children was 2.4 times the non-Aboriginal rate [10].

Rotavirus became a notifiable disease in WA from July 2006 [11]. The Communicable Disease Control Directorate (CDCD) and Public Health Units (PHUs) in the Department of Health WA (WA Health) investigate clusters and outbreaks of rotavirus. Initially, Rotarix was used in WA, from July 2007 to February 2009, then vaccine selection was changed to RotaTeq. In July 2017, the rotavirus vaccine used in WA reverted back to Rotarix [12,13]. In 2017, the estimated vaccine coverage in eligible children <12 months of age was 83.5% in Aboriginal children nationally and 89.5% in non-Aboriginal children [14]. The vaccine coverage for WA was 81.5% in 2015, the most recent data available, compared to a national coverage of 85.4% [15].

Sporadic community-wide rotavirus outbreaks have occurred in different states and territories around Australia. Outbreaks due to G2P[4] strains occurred in Perth (1993), Melbourne (1994), and Sydney (2001) [16,17]. Widespread outbreaks impacting remote communities in the Northern Territory have occurred due to G2P[4] strains in 1993, 1999, 2004, and 2009 [18–22]. Outbreaks due to G2P[4] strains were reported in 2010 in South Australia and Western Australia [23]. An outbreak caused by G2P[4] occurred in New South Wales in 2012, predominantly impacting children aged 5–9 years [24]. In 2017, multiple G2P[4] outbreaks were reported in the Northern Territory, South Australia, and Western Australia [3].

The aim of this study was to describe the epidemiology and burden of disease during an outbreak of rotavirus in the metropolitan (METRO) region of Perth and the remote Kimberley (KIMB) region of WA in 2017. Whole genome sequencing was performed to characterise the rotavirus strain circulating during this outbreak and place it in the context of global strains.

#### **2. Results**

#### *2.1. Descriptive Epidemiology*

In 2017, there were 519 notified cases of rotavirus infection in WA (19.1 cases per 100,000 population), making rotavirus the third most commonly notified enteric infection in WA. A marked increase in rotavirus notifications was noted in the second quarter of

2017 (April/May/June, 2Q17), with 236 cases, compared to the five-year second quarter average (2012–2016) of 100.8 cases (Figure 1).

**Figure 1.** Monthly rotavirus notification rates between January 2012 and November 2017.

Within the 2Q17, the highest number of cases was seen in May (*n* = 122), compared to 52 cases in April and 62 cases in June. Of the 122 cases in May, 80 were aged ≤4 years (Figure 2), with the majority of cases aged <1 year (*n* = 26) and 1 year (*n* = 31). While cases were seen in all PHUs, Aboriginal people from the KIMB (*n* = 46) region and non-Aboriginal people from the metropolitan (METRO) region (*n* = 36) were the two most affected groups (Figure 3). Examining children ≤4 years, Indigenous status, and PHU more closely, Aboriginal children from the KIMB region were disproportionally represented (40/80 cases) (Figure 4).

**Figure 2.** Distribution of the number of rotavirus cases in Western Australia in May, 2017 by age (years) and Indigenous status.

**Figure 3.** Distribution of the number of rotavirus cases from May, 2017, all ages, by Indigenous status and Public Health Unit (PHU) boundaries, reflecting WA Health administrative regions: Central/Wheatbelt (CENT), Goldfields (GOLD), Great Southern (GSTH), Kimberley (KIMB), Metropolitan Perth (METRO), Midwest (MIDW), Pilbara (PILB), and South West (STHW).

**Figure 4.** Distribution of rotavirus cases in May, 2017, aged ≤4 years, by Aboriginal status and Public Health Unit (PHU) boundaries, reflecting WA Health administrative regions: Central/Wheatbelt (CENT), Goldfields (GOLD), Great Southern (GSTH), Kimberley (KIMB), Metropolitan Perth (METRO), Midwest (MIDW), Pilbara (PILB), and South West (STHW).

Vaccination status was known for 97% of the total May cases (118/122). Of these, 41% were fully vaccinated (48/118), 24% were partially vaccinated (29/118), and 35% were not vaccinated (41/118) (Table 1). Only five of the unvaccinated cases were eligible to have been vaccinated, with the remaining 36 cases ineligible due to age.

Hospitalisation status was known for 78% of cases (95/122). For those with known hospitalisation status, 38% (36/95) were hospitalised as a result of their infection, of which 64% were Aboriginal people (23/36) and 36% were non-Aboriginal people (13/36). Children aged ≤4 years represented 86% of hospitalisations (31/36), of which Aboriginal children accounted for 71% (22/31). Of the hospitalised cases, 39% were fully vaccinated (14/36) and 36% were partially vaccinated (13/36). A further 22% were not vaccinated

(8/36), five of which were ineligible for vaccination due to age. Vaccination status was unknown for one case.


**Table 1.** Vaccination status of rotavirus cases from May 2017.

1 Individuals ≥11 years of age were considered ineligible to have ever received a rotavirus vaccine dose based on age. <sup>2</sup> 52 samples were not sent to Murdoch Children's Research Institute for genotyping.

#### *2.2. Genotyping*

Cases were designated to the month of May based on the optimal date of onset (ODOO). Of the 122 cases from the month of May, a stool sample was available for 70 and were sent for genotype analysis at the National Rotavirus Reference Centre (NRRC), Murdoch Children's Research Institute in Melbourne, Australia. The predominant genotype identified was G2P[4] (94%, 66/70) (Table 2), with the majority of G2P[4] cases in the KIMB region (61%, 40/66). In the KIMB region, Aboriginal people were disproportionately represented, accounting for 90% of cases (36/40) (Table 2).

**Table 2.** Genotype results for 70 rotavirus positive samples (ODOO\* May, 2017).


<sup>1</sup> Public Health Unit (PHU) boundaries, reflecting WA Health administrative regions: Goldfields (GOLD), Kimberley (KIMB), Metropolitan Perth (METRO), Midwest (MIDW), Pilbara (PILB), and South West (STHW). \*OODO: Optimal date of onset.

#### *2.3. Vaccination and Hospitalisation Status of Genotyped Cases*

Rotavirus vaccination information was available for 94% (66/70) of cases notified in May with genotyping results (Table 1). Of these, 53% (35/66) were fully vaccinated, 20% (13/66) were partially vaccinated, 4% (3/66) were not vaccinated but were eligible based on age, and 23% (15/66) were not vaccinated due to age (Table 1). Three quarters of G2P[4] cases were either fully or partially vaccinated (47/62). The majority of cases that were partially or fully vaccinated had received only the RotaTeq vaccine. Two fully vaccinated and one partially vaccinated case had received the Rotarix vaccine. Two fully vaccinated cases had received a combination of Rotarix and RotaTeq vaccines. Almost all Aboriginal cases had a known vaccination status (40/41); 60% (24/40) were fully vaccinated, 30% (12/40) were partially vaccinated, and 10% (4/20) were not vaccinated. Of the non-Aboriginal cases with known vaccination status (26/29), 42% (11/26) were fully vaccinated, 4% (1/26) were partially vaccinated, and 54% (14/26) were not vaccinated.

Hospitalisation status was known for 86% of the genotyped cases in May (60/70), with 19 cases hospitalised. Of these, 47% were fully vaccinated (9/19), 26% were partially vaccinated (5/19), 16% were not vaccinated (3/19), and vaccination status was unknown for two cases (10%).

#### *2.4. Sequence Analysis of G2P[4] Samples*

A total of 38 G2P[4] samples were analysed using polyacrylamide gel electrophoresis to visualise the electropherotype pattern. The majority of samples had a highly similar electropherotype, indicating that a relatively conserved strain was circulating during the outbreak (data not shown).

Three samples were selected for whole genome sequencing, which were representative of the dominant electropherotype: RVA/Human-wt/AUS/WAPC2769/2017/G2P[4] (1-yearold, fully vaccinated, KIMB), RVA/Human-wt/AUS/WAPC2784/2017/G2P[4] (2-year-old, fully vaccinated child, METRO), and RVA/Human-wt/AUS/WAPC2824/2017/G2P[4] (3 year-old, fully vaccinated KIMB). The three samples exhibited the archetypal DS-1-like genome constellation: G2-P[4]-I2-R2-C2-M2-A2-N2-T2-E2-H2.

The 11 genes of each sample were successfully sequenced, with the exception of the VP2 gene of RVA/Human-wt/AUS/WAPC2769/2017/G2P[4] and RVA/Humanwt/AUS/WAPC2824/2017/G2P[4] for which only 69.7–70.3% of the open reading frame (ORF) could be determined. The sample volumes were exhausted, attempting to resolve the approximate 800-base pair (bp) region at the 3′ prime end of the gene without success.

The coding regions of each gene of RVA/Human-wt/AUS/WAPC2784/2017/G2P[4] and RVA/Human-wt/AUS/WAPC2824/2017/G2P[4] were highly conserved, with RVA/Humanwt/AUS/WAPC2769/2017/G2P[4] displaying some minor variability: VP1 (99.81–99.94% nucleotide (nt) and 99.82–99.91% amino acid (aa) similarity), VP2 (99.84–100% nt and 99.84–100% aa similarity), VP3 (99.84–99.92% nt and 99.88–100% aa similarity), VP4 (99.79–99.91% nt and 99.61–99.87% aa similarity), VP6 (99.92–100% nt and 100% aa similarity), VP7 (99.89–100% nt 99.69–100% aa similarity), NSP1 (99.86–99.93% nt and 100% aa similarity), NSP2 (99.90–100% nt and 100% aa similarity), NSP3 (99.47–99.79% nt and 99.39–99.68% aa similarity), NSP4 (99.62–99.81% nt and 99.43–100% aa similarity), and NSP5/6 (99.86–99.93% nt and 100% aa similarity).

#### *2.5. Phylogenetic Analysis*

Phylogenetic analysis of the 11 genome segments was conducted to investigate the genetic relationships of the three outbreak samples RVA/Human-wt/AUS/WAPC2769/2017/G2P[4], RVA/Human-wt/AUS/WAPC2784/2017/G2P[4], and RVA/Human-wt/AUS/WAPC2824/ 2017/G2P[4] to previously characterised Australian samples and global strains (Figure 5a–k). In the VP7 tree, the outbreak samples clustered with contemporary G2P[4] samples from Japan and Taiwan detected in 2016 and 2017 shared 99.74–100% nt and 99.24–100% aa similarity (Figure 5a). The outbreak samples did not cluster closely to previously characterised Australian samples. The most closely related were two samples from Victoria detected in 2010, sharing 99.47–99.61% nt and 99.62–100% aa similarity (Figure 5a). In the VP4 tree, the outbreak samples clustered with the same contemporary G2P[4] samples from Japan as in the VP7 tree and shared 99.61–99.83% nt and 99.48–100% aa similarity (Figure 5b). Again, the outbreak samples did not cluster closely to previously characterised Australian samples; most closely related to the same two samples from Victoria (RVA/Human-w/AUS/CK20040/2010/G2P[4] and RVA/Human-wt/AUS/CK20060/2010/G2P[4]) that shared 99.48–99.66% nt and 99.30–99.70% aa similarity (Figure 5b).

**Figure 5.** *Cont.*

*Pathogens* **2021**, *10*, 350

**Figure 5.** *Cont.*

*Pathogens* **2021**, *10*, 350

**Figure 5.** Maximum likelihood phylogenetic trees of (**a**) VP7, (**b**) VP4, (**c**) VP1, (**d**) VP2, (**e**) VP3, (**f**) VP6, (**g**) NSP1, (**h**) NSP2, (**i**) NSP3, (**j**) NSP4, and (**k**) NSP5/6 2017 Western Australia outbreak G2P[4] samples. The position of strains sequenced in this study are highlighted in red and with square symbols, previously characterised G2P[4] outbreak samples from the Northern Territory detected in 1999, 2004, and 2010 are denoted with triangle symbols. All Australian samples are in bold. Ultrafast bootstrap values≥95% are shown.

Across the VP1, VP2, VP3, and VP6 gene trees, the outbreak samples from this study continued to form conserved clusters with the contemporary Japanese samples RVA/Humanwt/JPN/MI1132/2016/G2P[4], RVA/Human-wt/JPN/K-21-16/2016/G2P[4], RVA/Humanwt/JPN/K-3-16/2016/G2P[4], RVA/Human-wt/JPN/Tokyo17-16/2017/G2P[4], and RVA/ Human-wt/JPN/CH1020/2016/G2P[4] as observed in the VP7 and VP4 trees (Figure 5c–f). The samples RVA/Human-w/AUS/CK20040/2010/G2P[4] and RVA/Human-wt/AUS/ CK20060/2010/G2P[4] were consistently the most closely related Australian samples to those from the 2017 outbreak. Across all trees, the 2017 outbreak samples fell within a clade that was comprised of a conserved group of G2P[4] strains from Belgium and Hungary that were detected in 2012, and the Australian samples RVA/Human-wt/AUS/CK20049/2010/G2P[4], RVA/Human-wt/AUS/CK20050/2010/G2P[4], RVA/Human-wt/AUS/CK20052/2010/ G2P[4], RVA/Human-wt/AUS/CK20056/2010/G2P[4], and RVA/Human-wt/AUS/RCH041/ 2010/G2P[4].

Across the NSP1, NSP2, NSP3, NSP4, and NSP5 gene trees, the 2017 outbreak samples exhibited the same pattern across all trees: clustering with same group of contemporary Japanese samples, and falling within conserved clades comprised of G2P[4] strains from Belgium and Hungary that were detected in 2012, and Australian 2010 samples (Figure 5g–k).

The 2017 outbreak samples were not closely related to the samples RVA/Humanwt/AUS/V233/1999/G2P[4], RVA/Human-wt/AUS/336190/2004/G2P[4] and RVA/ Human-wt/AUS/V203/2009/G2P[4], which were associated with prior outbreaks in the Northern Territory, often clustering in separate lineages or distinct clades. This suggests that the current outbreak variant was not derived from the prior G2P[4] outbreak variants that had undergone genetic drift or reassortment over the intervening years but were more closely related to a G2P[4] variant that has been detected in Japan, Hungary, and other regions of the world. It may be derived from the Australian 2010 G2P[4] variant that has undergone moderate genetic drift during global circulation.

#### *2.6. Comparison of the Outbreak Samples to the G2 VP7 Gene Component of the RotaTeq Vaccine*

The VP7 gene of the 2017 outbreak samples possessed 93.37–99.48% nt and 94.79–95.01% aa similarity with the G2 VP7 gene of RotaTeq. The amino acid differences between the outbreak samples and RotaTeq were analysed and 16 residues differed between the G2 component of RotaTeq and the two outbreak samples RVA/Human-wt/AUS/WAPC2824/2017/G2P[4] and RVA/Human-wt/AUS/WAPC2769/2017/G2P[4]. RVA/Human-wt/AUS/WAPC2784/ 2017/G2P[4] had 17 residues that differed. The altered residues that fell between amino acid 78 and 312 were mapped to the surface of the VP7 monomer to highlight mutations in proximity to the VP7 antigenic epitopes 7-1a, 7-1b, and 7-2 [25] (Figure 6). Mutations were observed in all three samples in antigenic epitope regions: positions A87T and D96N in antigenic region 7-1a, and S213D in region 7-1b. Additionally, the mutation D145G in the antigenic epitope region 7-2 was observed in RVA/Human-wt/AUS/WAPC2784/2017/G2P[4]. The three outbreak samples exhibited the residues D96N and S213D, which are amino acid changes that have been shown to escape neutralisation with monoclonal antibodies [26].

**Figure 6.** A surface representation of the VP7 monomer depicting the amino acid residues that differ between the 2017 G2P[4] WA outbreak samples and the G2 component of the RotaTeq vaccine strain (PDB ID: 3FMG). The antigenic epitopes are coloured as 7-1a in cyan, 7-1b in mid blue, and 7-2 in dark blue. The conserved residues that differ between the 2017 samples and the G2 component of the RotaTeq vaccine strain are shown in red and the residue that differed only in RVA/Humanwt/AUS/WAPC2784/2017/G2P[4] is shown in salmon.

#### **3. Discussion**

Rotavirus was gazetted as a notifiable disease in WA in 2006 in part to monitor the effectiveness of the rotavirus vaccine when it was added to the childhood immunisation schedule in Australia in mid-2007 [11]. The introduction of rotavirus vaccines has lessened the once prominent seasonality of rotavirus infection in Australia [27,28]. Following campylobacteriosis and salmonellosis, rotavirus was the third most commonly notified enteric infection in the population of WA in 2017 [29]. A large increase in rotavirus notifications was noted in the second quarter of 2017, with the highest number of cases noted in May, indicating an outbreak occurring prior to the onset of winter (Figure 1).

A total of 236 rotavirus notifications were recorded in the second quarter of 2017, compared to the five-year second quarter average of 100.8 notifications, highlighting the scale of the outbreak. The five-year second quarter average was somewhat skewed by an outbreak in the second quarter of 2015 (Figure 1) that affected all WA regions, and predominantly affected non-Aboriginal people. Multiple outbreaks related to child care and aged care facilities were noted during this time, with the predominant strain identified as G12P[8] [30].

In contrast to 2015, the increase in the second quarter of 2017 was noted to disproportionally affect young Aboriginal children in the KIMB region, which is in the north of the state. A number of towns and Aboriginal communities in the KIMB region were affected. The KIMB PHU investigated the increase in notifications, with assistance from local government environmental health officers. Several public health interventions were implemented as a result of their investigations, including the distribution of a public health alert to local hospitals and Aboriginal medical service providers, liaising with environmental health officers and community health staff to provide public health advice for affected communities, and an interview on local radio.

It is noteworthy that an increased burden of rotavirus disease was reported elsewhere in Australia for 2017. Multiple outbreaks were recorded across Australia, due to equine-like G3P[8] in New South Wales and G8P[8] in New South Wales and Victoria [3]. In addition to the WA outbreak herein described, outbreaks due to G2P[4] were also reported in the Northern Territory (NT) and South Australia [3]. It is thought that the 2017 G2P[4] outbreak began in the NT and subsequently spread to rural and remote regions of WA adjacent to the border between these states [31]. A companion study described the weak protective effect of either Rotarix or RotaTeq vaccination in the setting of this outbreak [31]. Suboptimal vaccine-effectiveness, particularly in the second year of life, has been reported in other high-burden, low-resource settings [32]. There are varied factors that could contribute to a reduced vaccine response, such as poor infant nutrition, the intestinal microbiota, co-morbid infections, as well as high levels of maternally derived anti-rotavirus antibodies [33].

The inclusion of rotavirus vaccines into the Australian NIP in 2007 has resulted in a considerable and sustained decrease in rotavirus morbidity across most of Australia, with a 71% decline in rotavirus-coded hospitalisations of children aged <5 years reported [34]. However, the observed decrease in hospitalisations has been less in Aboriginal and Torres Strait Islander children; they remain at greater risk of severe rotavirus disease requiring hospitalisation than their non-Indigenous counterparts [34]. Following rotavirus vaccine introduction in WA, significant declines in rotavirus-coded hospitalisation rates have been observed in all children aged <5 years, up to 79% among non-Aboriginal and up to 66% among Aboriginal children [35]. During the outbreak peak in May 2017 (122 cases), over a third of cases were hospitalised as a result of their infection, with Aboriginal people representing two thirds of these hospitalisations. As would be expected with rotavirus infection, the vast majority of cases hospitalised were ≤4 years of age, and Aboriginal children accounted for 71% of hospitalisations in this age group. Compared to their non-Indigenous counterparts, the paediatric Aboriginal population exhibit a greater burden of disease due to infections, and large, biannual rotavirus outbreaks have been reported in the Northern Territory [18–22]. Continued surveillance is critical to elucidate the complex factors that contribute to the occurrence of these outbreaks.

When the vaccination status of cases from the May peak was compared to hospitalisation status, vaccination did not appear to impact on whether a case was hospitalised. Fully or partially vaccinated children represented 75% of hospitalised cases (27/36) compared to unvaccinated eligible children accounting for 8% of hospitalised cases (3/36). Vaccination status was unknown for 3% of cases (1/36) and the remaining 14% of cases (5/36) were ineligible to have been vaccinated based on age. In May 2017, RotaTeq was the vaccine prescribed in the WA vaccination schedule and the vast majority of cases who were either fully or partially vaccinated were vaccinated with RotaTeq. Whilst a genotype-specific vaccine effectiveness has not been estimated for children in WA, the vaccine effectiveness of three doses of RotaTeq has been estimated at 82% (95% CI: 59–92) [36].

Full genome sequencing was performed on representative samples from the outbreak. These samples were found to be most closely related to Japanese G2P[4] strains detected in 2016 and 2017 across all genes in the genome. In one associated paper, these closely related samples were reported as minor G2P[4] variants circulating in the Mie prefecture in 2017 [37]. However, this variant was also detected in Tokyo in 2017, where G2P[4] was the dominant genotype, accounting for 40% of samples [38]. The outbreak samples also consistently clustered with Hungarian G2P[4] from 2012, where this genotype accounted for 13.5% of the samples genotyped in 2012 [39]. The WA outbreak samples clustered within a clade that also included G2P[4] strains from Australia that were circulating in 2010. These samples were collected during 2010–2011 when there was a substantial increase in G2P[4] strains in Australian states using the RotaTeq vaccine; G2P[4] strains replaced G1P[8] as the dominant genotype for the first time since vaccine introduction [23]. Overall, this suggests that the strain circulating during the 2017 WA outbreak is a global variant that was previously detected in Australia and has continued to be successfully transmitted in various regions around the world for over almost a decade. Based on the available sequencing, the majority of samples exhibit a relatively conserved genome that has not

undergone substantial reassortment, with the diversity observed indicative of genetic drift over the years. It is highly likely that this variant represents a re-introduction into Australia rather than reflecting genetic drift that has only occurred in the Australian population. The 2017 variant was not closely related to G2P[4] strains that had caused prior outbreaks in the Northern Territory in 1999, 2004, and 2009 [18].

The VP7 gene of the 2017 WA outbreak samples was compared to the G2 VP7 gene component of the RotaTeq vaccine. A total of 16 residues differed between the G2 component of RotaTeq and the two outbreak samples RVA/Human-wt/AUS/WAPC2824/2017/G2P[4] and RVA/Human-wt/AUS/WAPC2769/2017/G2P[4], and RVA/Human-wt/AUS/WAPC2784/ 2017/G2P[4] had 17 residues that differed. However, this is not unexpected as the RotaTeq G2 VP7 gene is derived from a strain that was circulating in 1992; global strains have undergone extensive genetic drift over the intervening years. Three of these altered residues in all three outbreak samples were observed in antigenic epitopes at positions A87T and D96N in antigenic region 7-1a, and S213D in region 7-1b [25]. Altered residue D145G in region 7-2 was only observed in RVA/Human-wt/AUS/WAPC2784/2017/G2P[4]. Residues D96N and S213D have been shown to escape neutralisation with monoclonal antibodies [26]. The observed altered residues A87T, D96N, and S213D have been observed in the majority of G2P[4] strains circulating globally over the last two decades [40]. In particular, mutations A87T, D96N, D145G, and S213D were observed in G2P[4] strains associated with outbreaks in children in Indonesia in 2018 and a nosocomial outbreak in adults within a German hospital [41,42]. Genetic drift in VP7 antigenic epitope regions could adversely impact the effectiveness of the RotaTeq vaccine against G2P[4] strains. However, large-scale studies combining genetic and antigenic characteristics of circulating variants are required to further elucidate this. It is possible that genetic drift between circulating variants and the vaccine strain, in combination with host-related facts that impact vaccine effectiveness in this population contribute to the occurrence of these outbreaks.

A limitation of this study was that a stool sample was available for 70/122 cases from the May peak. Not genotyping all samples could result in the proportion of the different genotypes being over- or underestimated. However, it does not alter the result that G2P[4] was the dominant genotype in the KIMB region as 41/53 samples were available and genotyped. The 70 samples available for genotyping were representative of the age distribution of rotavirus cases in WA during this period. However, more samples from Aboriginal cases were genotyped compared to non-Aboriginal cases (76% vs. 45%) and this could have overestimated the proportion of G2P[4] cases reported. Similarly, more samples were genotyped from the remote areas of the KIMB and PILB regions, which may also have overestimated the proportion of G2P[4] cases seen. Given this study largely focuses on the KIMB region, it is unlikely that this had any major impact on the overall results of the study.

#### **4. Materials and Methods**

#### *4.1. Notification Data*

Data on WA cases of rotavirus were obtained from the WA Notifiable Infectious Disease Database (WANIDD). The notifications contained in WANIDD are received from medical practitioners and pathology laboratories under the provisions of the Public Health Act 2016 and subsequent amendments, and are retained in WANIDD if national case definitions are met. Rotavirus was listed as a notifiable disease in WA in July 2006 [11]. Data was extracted from WANIDD by optimal date of onset (ODOO) for the time period 01/01/2012 to 31/12/2017 and exported to Microsoft® Excel 365 (Microsoft®, Version 1808, Redmond, WA, USA). The ODOO is a composite of the 'true' date of onset provided by the notifying doctor or obtained during case follow-up, the date of specimen collection for laboratory notified cases, and when neither of these dates is available, the date of notification by the doctor or laboratory, or the date of receipt of notification, whichever is earliest. Notification data are broken down by regions that are based on Public Health Unit (PHU) boundaries, reflecting WA Health administrative regions: Central/Wheatbelt

(CENT), Goldfields (GOLD), Great Southern (GSTH), Kimberley (KIMB), Metropolitan Perth (METRO), Midwest (MIDW), Pilbara (PILB), and South West (STHW).

#### *4.2. Vaccination Status*

Records of vaccine administration were submitted to the Australian Immunisation Register (AIR) (curated by Services Australia, Australian Government). The AIR includes vaccines administered under the national immunisation program, school programs, and privately. CDCD staff accessed AIR to determine the rotavirus vaccine status of notified cases.

#### *4.3. Rotavirus Positive Faecal Samples*

A total of 122 faecal samples collected from children and adults presenting to hospital or general practice clinics with severe gastroenteritis in Western Australia during May, 2017 were determined to be rotavirus positive by a local diagnostic laboratory. Seventy de-identified rotavirus positive specimens were sent to the National Rotavirus Reference Centre (NRRC) at the Murdoch Children's Research Institute. A further 27 samples did not have adequate remaining volume and were not sent for genotyping. There is no agreement with private pathology laboratories to forward samples for genotyping. As a result, 25/122 (20%) of samples were not genotyped. Where possible, metadata, including date of collection, date of birth, gender, and postcode, were collected. Samples were stored at −80 ◦C until analysis, allocated a unique laboratory code, and entered into a REDCap database.

#### *4.4. Genotyping*

Viral RNA was extracted from 10–20% (*w*/*v*) faecal extracts using the QIAamp Viral RNA mini extraction kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. Rotavirus G- and P-genotyping was performed using a hemi-nested multiplex RT-PCR assay [43]. First-round RT-PCR reactions were performed using the One Step RT-PCR kit (QIAGEN, Germany), using the VP7 (VP7F/VP7R), or the VP4 primer pair (VP4F/VP4R) [44,45]. The second-round genotyping PCR reactions were performed using the AmpliTaq® DNA Polymerase with Buffer II (Applied Biosystems, Foster City, CA USA), together with specific oligonucleotide primers for G types (1, 2, 3, 4, 8, and 9) or P types ([4], [6], [8], [9], [10], and [11]) as previously described [4]. Gel electrophoresis of second-round PCR products was performed to determine the G- and P- genotype of each sample.

#### *4.5. Conformation of Vaccine-Line Strains*

Sequencing of VP6 and VP7 genes was performed for suspect RotaTeq samples with mixed G types or were P non-typeable as previously described [18].

#### *4.6. Polyacrylamide Gel Electrophoresis*

The 11 segments of rotavirus dsRNA were separated on 10% *w*/*v* polyacrylamide gel with 3% *w*/*v* polyacrylamide stacking gel at 25 mA for 16 h. The genome migration patterns (electropherotypes) were visualised by silver staining according to the established protocol [46].

#### *4.7. Whole Genome Sequencing*

Each of the 11 genes were reverse transcribed and amplified by PCR using the OneStep RT-PCR Kit (QIAGEN, Hilden, Germany) using gene-specific sense and antisense primers (primer sequences available upon request). RNA was denatured and reverse transcribed for 30 min at 45 ◦C, followed by PCR activation for 15 min at 95 ◦C. Then, 40 cycles of amplification for 10 s at 94 ◦C, 1 min at 55 ◦C, and 3 min at 68 ◦C, followed by a final extension for 10 min at 68 ◦C were performed. The amplicons were gel purified using the Wizard® SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA) according to the manufacturer's instructions.

The purified products were pooled in equimolar concentrations and subjected to standard library construction for Illumina sequencing using the Nextera XT DNA Library Preparation Kit following the manufacturer's recommendations for dual-indexed barcoding (Illumina Inc., San Diego, CA, USA). Normalised samples were pooled and sequenced using 500-cycle (2 × 250-bp paired-end) MiSeq reagent kits (v2; Illumina Inc., San Diego, CA, USA).

#### *4.8. Sequence Assembly*

Raw reads were trimmed for quality and adapters using BBDuk Adapter/Quality Trimming Version 38.37, duplicate reads were removed using Dedupe Duplicate Read Remover version 38.37 and pair-end reads were merged using BBMerge Paired Read Merger version 38.37, all performed within Geneious Prime. Reads were mapped to reference rotavirus genomes using the Bowtie2 mapper within Geneious Prime [47].

#### *4.9. Assignment of Genotypes*

The genotypes of each of the 11 genome segments were determined using the online RotaC v2.0 rotavirus genotyping tool (http://rotac.regatools.be, accessed on 18 January 2021) in accordance with the recommendations of the Rotavirus Classification Working Group (RCWG) [8].

#### *4.10. Phylogenetic Analysis*

Nucleotide similarity searches were performed using the BLAST server on the Gen-Bank database at the National Center for Biotechnology Information, USA (www.ncbi. nlm.nih.gov, accessed on 18 January 2021). The nucleotide and amino acid sequences of each gene were compared with sequences available in the GenBank database that possessed the entire open reading frame. Multiple nucleotide and amino acid alignments were constructed using the Multiple Sequence Comparison by Log Expectation (MUSCLE) algorithm in Geneious Prime [48].

The best-fit nucleotide substitution model for each gene tree were tested and selected in IQTREE v1.6 using the using the Bayesian Information Criteria [49]. The selected nucleotide substitution models were GTR+F+R3 (VP1, VP3), GTR+F+G4 (VP4), TIM+F+G4 (NSP1, NSP2, NSP3), TIM+F+I+G4 (VP2) TN+F+G4 (VP6, NSP4), and HKY+F+G4 (VP7, NSP5/6). The maximum likelihood trees were inferred using IQTREE v1.6 with the robustness of branches assessed by 1000 bootstrap replicates using the ultrafast bootstrap feature [50]. The resulting trees were visualised and edited in FigTree v1.4.4 (http://tree. bio.ed.ac.uk/software/figtree/, accessed on 18 January 2021). Nucleotide and amino acid distance matrixes were calculated using the p-distance algorithm in MEGAX [51]. Structural analysis of the VP7 protein (PDB ID: 3FMG) was performed using the PyMOL Molecular Graphics System, Version 1.2r3pre (Schrödinger, Inc, New York, NY, USA).

#### *4.11. Accession Numbers*

The nucleotide sequences for genes described in this study have been deposited in GenBank under the accession numbers MW275246–MW275278.

#### **5. Conclusions**

This G2P[4] outbreak disproportionately impacted Aboriginal children ≤4 years of age in the remote Kimberley region of Western Australia. The G2P[4] variant circulating was closely related to contemporary Japanese G2P[4] samples, suggesting a global variant that exhibited the altered residues A87T, D96N, and S213D compared to the G2 component of the RotaTeq vaccine, residues that have been observed in the majority of G2P[4] strains circulating globally over the last two decades. Despite national vaccine coverage of 85.4%, outbreaks continue to occur in vaccinated populations in Australia, in particular impacting Aboriginal populations. These outbreaks pose particular challenges to regional areas and

remote communities. Continued surveillance and characterisation of emerging variants are imperative to ensure the ongoing success of the rotavirus vaccination program in Australia.

**Author Contributions:** Conceptualization, C.M.D., N.P. and S.R.-F.; Data curation, C.M.D., N.P. and S.R.-F.; Formal analysis, C.M.D., N.P. and S.R.-F.; Funding acquisition, C.M.D. and J.E.B.; Investigation, C.M.D., N.P. and S.R-F.; Methodology, C.M.D., N.P., E.D. and S.R.-F.; Project administration, J.E.B.; Resources, C.M.D., N.P., S.R.-F. and J.E.B.; Software, C.M.D. and N.P.; Validation, C.M.D., N.P. and S.R.-F.; Visualization, C.M.D. and N.P.; Writing—original draft, C.M.D. and N.P.; Writing—review and editing, E.D., S.R.-F. and J.E.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Australian National Health and Medical Research Council Project grant (1163346). The Australian Rotavirus Surveillance Program is supported by research grants from the vaccine companies Commonwealth Serum Laboratories (bioCSL)/Sequis (2010–2018) and the Australian Government Department of Health (2010–2018). Funding for this study was also provided by GlaxoSmithKline Biologicals SA (2010–2016, study ID116120 2017–2018). GlaxoSmithKline Biologicals SA was provided the opportunity to review a preliminary version of this manuscript for factual accuracy, but the authors are solely responsible for final content and interpretation. The authors received no financial support or other form of compensation related to the development of the manuscript. The Murdoch Children's Research Institute is supported by the Victorian Government's Operational Infrastructure Support program. CMD is supported through the Australian National Health and Medical Research Council with an Early Career Fellowship (1113269). The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

**Institutional Review Board Statement:** This activity has been reviewed by the Royal Children's Hospital Human Research Ethics Office and assessed as a quality assurance and evaluation activity conducted on behalf of the Australian Government Department of Health and Western Australian Department of Health.

**Informed Consent Statement:** Patient consent was waived as data on WA cases of rotavirus were obtained from the WA Notifiable Infectious Disease Database (WANIDD). The notifications contained in WANIDD are received from medical practitioners and pathology laboratories under the provisions of the Public Health Act 2016 and subsequent amendments.

**Data Availability Statement:** The nucleotide sequences for genes described in this study have been deposited in GenBank under the accession numbers MW275246-MW275278.

**Acknowledgments:** The authors thank H. Tran for providing technical support, and S. Thomas for constructive feedback.

**Conflicts of Interest:** The Australian Rotavirus Surveillance Program is supported by research grants from the vaccine companies Commonwealth Serum Laboratories (bioCSL)/Sequis (2010–2018) and GlaxoSmithKline (2010–2016, study ID116120 2017–2018). C.M.D. has served on an advisory board for GSK (2019), all payments were paid directly to an administrative fund held by Murdoch Children's Research Institute. All other authors declare no competing interests.

#### **References**


*Article*
