Contribution of low population immunity to the severe Omicron BA.2 outbreak in Hong Kong

Monitoring population protective immunity against SARS-CoV-2 variants is critical for risk assessment. We hypothesize that Hong Kong’s explosive Omicron BA.2 outbreak in early 2022 could be explained by low herd immunity. Our seroprevalence study using sera collected from January to December 2021 shows a very low prevalence of neutralizing antibodies (NAb) against ancestral virus among older adults. The age group-specific prevalence of NAb generally correlates with the vaccination uptake rate, but older adults have a much lower NAb seropositive rate than vaccination uptake rate. For all age groups, the seroprevalence of NAb against Omicron variant is much lower than that against the ancestral virus. Our study suggests that this BA.2 outbreak and the exceptionally high case-fatality rate in the ≥80 year-old age group (9.2%) could be attributed to the lack of protective immunity in the population, especially among the vulnerable older adults, and that ongoing sero-surveillance is essential.

S evere acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of COVID-19, is characterized by efficient person-to-person transmission, propensity to cause severe disease, and frequent emergence of variants that escape population immunity [1][2][3] . The rapid surge of severe cases which require hospitalization, especially apparent during the emergence of novel variants, overwhelms the healthcare systems.
Forecasting the severity of upcoming COVID-19 waves is critical in formulating public health preparedness plans. Knowledge of the population level immunity against SARS-CoV-2 can help to predict the impact of COVID-19 on the healthcare system. COVID-19 vaccination or SARS-CoV-2 infection induce antibody response against SARS-CoV-2 4,5 . A population with high vaccination uptake rate or incidence of infection theoretically have a high level of population immunity. However, several factors can affect the humoral immunity elicited by vaccination or infection 6 . First, some COVID-19 patients or vaccine recipients do not develop sufficient protective antibody responses. Lower neutralizing antibody (NAb) titers are found in patients with asymptomatic or mild COVID-19, immunocompromised patients and older adults [7][8][9][10][11][12][13] . Second, the NAb titers elicited by inactivated whole virion or adenovirus-vectored vaccines are lower than those elicited by mRNA vaccines [14][15][16] . Third, NAb titers decline within a few months after infection or vaccination [17][18][19][20][21][22] . Fourth, NAb elicited from prior infection or vaccination may not be effective against new variants with mutations in the spike protein, as in the case of the Omicron variant 16,22 .
In view of the shortcomings of using history of vaccination or infection to predict population immunity, we need a more reliable method to determine population level immunity. NAb titers showed a good correlation with vaccine effectiveness against symptomatic infection 14,23 . In a clinical trial of the mRNA-1273 vaccine, the estimated vaccine efficacy decreased from 96% for a NAb titer of 1000 IU/ml to 78% for a NAb titer of 10 IU/ml 24 . Furthermore, individuals positive for anti-spike antibody, which correlates with NAb status, are less likely to have COVID-19 reinfection to cognate variants 25 .
Under zero-case strategy, Hong Kong had a very low incidence of COVID-19, with a cumulative count of just 12,655 cases (0.17% of the population) as of December 2021. There have been four COVID-19 waves in Hong Kong between 2020 and 2021 [26][27][28] . The first wave occurred from January to early March 2020, with the majority being imported cases from mainland China. The second wave occurred from mid-March to May 2020, and consisted of imported cases returning from places outside Asia and the associated local cases. The third and fourth waves occurred from July to August 2020, and from November 2020 to April 2021, respectively. Both the third and fourth waves consisted mainly of locally acquired cases. Local transmission of SARS-CoV-2 was almost absent from May to December 2021.
The COVID-19 vaccination program was launched in Hong Kong in February 2021. Our previous serosurveillance studies showed that <1% of sera collected from the Hong Kong population before March 2021 tested positive for SARS-CoV-2 IgG against the nucleoprotein, but the seropositive rate started to increase from April 2021 29,30 . In the current study, we first screened 1800 serum specimens collected between January and December 2021 with a surrogate virus neutralization test (sVNT), which correlate well with conventional live virus neutralization test (cVNT) against the ancestral virus 22,31 . For sera tested positive by sVNT, we assessed the NAb against the ancestral virus and Omicron sublineages BA.1 and BA.2 using a cVNT. We interpreted our serosurveillance findings with the epidemiological and genomic information of the fifth wave of COVID-19 in Hong Kong, which exceeded 1 million cases by March 18, 2022 32,33 .

Results
Sampled population. We retrieved a total of 1800 archived serum specimens collected between January and December 2021 from Queen Mary Hospital, with 300 serum specimens collected every two months. We retrieved approximately the same number of serum specimens from each age group from 0-9 years old to ≥80 years old depending on availability (Supplementary Table S1).  (Fig. 1a). The geometric mean sVNT levels also peaked in September/October, and declined in all age groups in November/December except for the 10-19 year-old age group ( Supplementary Fig. S2). Subgroup analysis of different age groups revealed stark differences in the positive rate in the sVNT among different age groups (Fig. 1b-j). For sera collected from the 0-9 year-old individuals, who were not eligible for COVID-19 vaccines in 2021, all specimens tested negative in the sVNT (Fig. 1b). For the 10-19 year-old age group, the seropositive rate rose sharply after July 2021, which coincided with the commencement of COVID-19 vaccination program for adolescents aged 12-17 years old on 14th June 2021 (Fig. 1c) 35 . In November/December, the 10-19 year-old age group had the highest sVNT seropositive rate, and there was a statistically significant trend towards lower seropositive rate for older age groups (age trend by Cochran-Armitage trend test; P < 0.0001 for age groups between 30-39 and ≥80 years) (Fig. 1k). Notably, none of the sera from the ≥80 year-old age group tested positive in the sVNT in November/December. The 10-19 year-old age group also had the highest sVNT activity (Fig. 3), and there was a statistically significant trend towards lower sVNT activity for older adults (P = 0.0014); excluding the 0-9 year-old age group (Fig. 3f).

COVID-19 vaccination in
Conventional live virus neutralization tests. The commercial sVNT assay in this study was designed based on the ancestral virus. Our previous study showed that this assay has a high concordance with the result of cVNT NAb against the ancestral virus, but poor correlation with cVNT NAb against the Omicron sublineage BA.1 due to immune escape 22 . Hence, for specimens tested positive in the sVNT, we used a cVNT to determine the NAb activity against the Omicron sublineages BA.1 and BA.2 and the ancestral virus. Overall, 86.0% (227/264) of sVNT positive specimens had detectable NAb against the ancestral virus (Fig. 1a). In contrast, only 7.2% (19/264) and 30.3% (80/264) had detectable NAb against BA.1 and BA.2, respectively. Live virus NAb against BA.1 was only found in sera from individuals between 10 and 69 years old, whereas live virus NAb activity against BA.2 could be found in sera from all age groups except for the 0-9 year-old age group (Fig. 1b-j). Assuming that all sVNT negative sera did not have detectable NAb against BA.1 or BA.2 (since our previous study showed that sVNT has 100% sensitivity for NAb against the Omicron variant 22 ), the proportion of individuals with detectable NAb against BA.1 and BA.2 in November/December 2021 were 2.3% (7/300) and 7.3% (22/300), respectively.
The absence of seropositive sera in the 0-9 year-old age group, the temporal association of seropositive rate and two-dose vaccination rate, and the low incidence of locally acquired COVID-19 cases after April 2021, suggested that NAb were elicited by COVID-19 vaccination in most of the seropositive individuals in our population. While the two-dose vaccination rate and sVNT positive rates were very similar in the younger age groups, there was a greater discrepancy between the vaccination and sVNT seropositive rates in the older age groups (Fig. 1). In November/December, the sVNT seropositive rate was <50% for the 40-49 year-old or older age groups. The greater discrepancy between vaccination and sVNT seropositive rates may be related to the lower proportion of individuals in the older age group receiving BNT162b2 (Fig. 2), as the seropositive rate and antibody titer has been shown to be higher for BNT162b2 than CoronaVac 16,36-38 .  (Fig. 4a). There was also a cluster of pet-storerelated Delta variant AY.127 41 (Fig. 4b). For the age groups from 10-59, the hospitalization rate was between 0.08-0.12%. The overall case-fatality rate was 0.60%, but the case-fatality rates were 15.3-fold and 2.4-fold higher for the ≥80 (9.19%) and 70-79 (1.46%) year-old age groups, respectively 44 (Fig. 4c).    45 .

Discussion
Serosurveillance is an effective tool to reveal the hidden burden of infection, especially at the beginning of the COVID-19 pandemic 30 . In addition, serosurveillance using NAb assays is useful in assessing the level of protective immunity in a population. respectively. In particular, older adults had the lowest levels of protective immunity. In November/December 2021, none of the ≥80 year-old age group and only 12.5% of the 70-79 year-old age group had a positive sVNT, while none of the individuals ≥70 years old had detectable NAb against BA.1 or BA.2. The lack of protective immunity against the Omicron BA.2 sublineage, especially among older adults, likely contributed to the severe fifth wave in Hong Kong which exponentially increased in February 2022 and overwhelmed the healthcare system in Hong Kong 46 .
Vaccination is associated with a much lower risk of death 33 . Unfortunately, the vaccination rate was particularly low among older adults aged 80 years or above. In a study conducted in mainland China, the two main reasons for vaccine hesitancy among adults aged 60 years or above were concern for vaccine safety and low infection risk 47 . In another study, Lau et al. showed that the presence of health conditions was one of the major factors associated with vaccine hesitancy or resistance among adults in Hong Kong 48 . Many older adults with comorbidities believe that vaccination may exacerbate their underlying diseases. In Hong Kong, the low incidence of COVID-19 in Hong Kong before the fifth wave have led to a low perceived infection risk.
The sVNT seropositive rate was very similar to the two-dose cumulative vaccination rate among the 10-19 and 20-29 year-old age groups. However, we observed a greater discrepancy between two-dose vaccination rate and sVNT seropositive rate for older adults, which may be related to the weaker immune response in this age group. Previous studies showed that older adults have poorer antibody response after vaccination compared to younger adults. Saure et al. showed that the IgG seropositive rate after vaccination was lower among adults aged ≥60 years than those aged 18-39 years 17 . Wei et al. showed that seropositivity after vaccination reduced faster for vaccine recipients aged over 75 years 12 . Another possible reason for the low seropositive rate among older adults is the vaccine preference among these age groups. Over 50% of adults aged ≥60 years have chosen Cor-onaVac. Head-to-head comparison showed that CoronaVac recipients had lower seropositive rates than BNT162b2 recipients after 2 doses 13,16 . Furthermore, seropositive rate declined more rapidly for the CoronaVac than BNT162b2 recipients 17 . Our results support the World Health Organization's recommendation that older adults should receive three doses of CoronaVac 49 .
The incidence of COVID-19 was very low in Hong Kong as a result of elimination strategy. Hence, the NAb detected by sVNT or cVNT for most individuals is likely elicited by vaccination rather than by infection. For the 0-9 year-old age group, none of the serum specimens tested positive in the sVNT. This is expected as the vaccination program for the 0-9 year-old age group was not yet started in 2021 and the prevalence of infection is likely to be very low as the number of laboratory-confirmed cases was only 0.1% (547/538500) of the population.
There was a decrease in seropositive rates in November/ December. This coincided with a low vaccination update rate since September (Supplementary Fig. S1), suggesting that there is waning of humoral immunity. Our previous study showed that for natural infection, waning of NAb level occurs most rapidly in the first 4 months 22 . For vaccination, there is also evidence of waning immunity within the first few months 50 .
The spike protein of Omicron sublineage BA.2, especially the RBD, differs from that of BA.1. There were conflicting data regarding the difference in NAb titer against BA.1 and BA.2 for post-vaccine sera. In our previous study using serum from vaccine recipients or COVID-19 patients 51  sVNT is a convenient assay which can be conducted in clinical or research laboratories without biosafety-level 3 facilities. sVNT is a competitive assay which measures antibodies that prevent the binding between SARS-CoV-2 RBD and the human receptor ACE2. Our previous study showed that sVNT has a sensitivity and specificity of 97.9% and 94.9%, using cVNT as the gold standard 55 . Our current study also showed a good correlation between sVNT and live virus NAb test against the ancestral virus for sVNT seropositive individuals. The sVNT can be easily modified to measure antibodies against the RBD of Omicron sublineages, and this would allow a more convenient assay for serosurveillance in the future.
There are several limitations in this study. First, as we are using anonymized serum specimens, we do not have clinical information or vaccination history from these individuals. Therefore, we are not able to perform further analysis to study factors that affect NAb response. In a population study conducted in UK, Wei et al showed that ethnicity, sex, social deprivation, and long term health conditions were associated with seropositivity after vaccination 12 . Second, this study included hospital patients instead of the "general population". Therefore, our data may not be representative of people without comorbidities. However, those with underlying diseases would have a much greater risk of having severe disease, and therefore our estimate would be better in predicting the burden to the healthcare systems. Third, there were fewer specimens in the pediatric age groups as the volume of archived serum specimens were not sufficient for testing. Fourth, we did not test T cell immunity. Future studies are required to understand the relationship between population T cell immunity and susceptibility to infection.
Our study highlights the value of NAb surveillance in assessing population immunity. We demonstrated that the level of protective immunity, especially against antigenically distinct variants, may not be reliably predicted by vaccination history. The low level of protective immunity among older adults was associated with an exceptionally high case-fatality rate. Vaccine studies should put more focus on the older adults. Improving the immunogenicity among older adults should be a priority in vaccine research.

Methods
Patients. Anonymised archived serum samples from the clinical biochemistry laboratory of Queen Mary Hospital in Hong Kong were used for serological surveillance as we described previously 29,30 . The archived specimens encompassed all age groups from 0-9 years old to ≥80 years old. A total of 300 serum specimens were randomly selected and tested in each two-month period from January/February to November/December 2021, with an approximately equal number of archived specimens in each 10-year age group depending on availability (Supplementary Table S1). Specimens were excluded if there was an insufficient volume of serum. Results of the surrogate NAb and live virus NAb against the ancestral virus for serum specimens collected between January and April 2021 was reported in our previous publication 29 , but the result of NAb against the Omicron sublineages BA.1 or BA.2 has not been reported before. For viral genome sequencing, 382 sequences were reported in our previous publications [39][40][41] . In the current study, we have performed genome sequencing for 163 randomly selected archived respiratory specimens collected from patients admitted to Queen Mary Hospital. This study was approved by the Institutional Review Board of the University of Hong Kong/ Hospital Authority Hong Kong West Cluster (UW 21-313 and UW 18-141). Written informed consent was waived since archived anonymized specimens were used.
Surrogate neutralization test and live virus neutralization test. sVNT test was performed using iFlash-2019-nCoV neutralization antibody assay according to manufacturer's instructions (Shenzhen YHLO Biotech Co. Ltd., Shenzhen, China) 56 . The manufacturer's cutoff for seropositivity is 15 AU/ml, while the maximum measurable result is 800 AU/ml. cVNT was performed on VeroE6/TMPRSS2 cells (JCRB cell bank of Okayama University; Cat#JCRB1819). The cVNT NAb titer was determined by an end-point dilution and cytopathic effect as we described previously 16 . Briefly, serum specimens were heat inactivated at 56°C for 30 min and were serially diluted in 2-folds with MEM containing 1% fetal calf serum (FBS) (Gibco). Duplicates of each diluted serum were mixed with 100 TCID 50 of an ancestral virus (GISAID accession number: EPI_ISL_434571), an Omicron sublineage BA.1 virus (GISAID accession number: EPI_ISL_7138045) 51 , or an Omicron sublineage BA.2 virus (GISAID accession number: EPI_ISL_9845731) 51 at 37°C for 1 h. After incubation, 100 μL of the serum-virus mixture was then added to VeroE6/TMPRSS2 cells that were seeded in 96-well plates 24 h before infection. The cells were incubated with the mixture at 37°C. After incubation for 3 days, cytopathic effect was examined. The cVNT was determined as the highest dilution with 50% inhibition of cytopathic effect. A cVNT of ≥10 was considered positive. For statistical analysis, a value of 5 was assigned if the cVNT is <10.
Whole genome sequencing and genome data analysis. Whole genome sequencing was performed using the Oxford Nanopore MinION device (Oxford Nanopore Technologies) as we described previously 16 . For the determination of viral lineage, nanopore sequencing was performed following the Nanopore protocol -PCR tiling of COVID-19 (Version: PTC_9096_v109_revH_06Feb2020) according to the manufacturer's instructions with minor modifications (Oxford Nanopore Technologies) as we described previously 26,27 . Briefly, extracted RNA was first reverse transcribed to cDNA using SuperScript TM IV reverse transcriptase (ThermoFisher Scientific, Waltham, MA, USA). PCR amplification was then performed using the hCoV-2019/nCoV-2019 Version 3 Amplicon Set (Integrated DNA Technologies, Coralville, IA, USA) with the Q5® Hot Start High-Fidelity 2X Master Mix kit (New England Biolabs, Ipswich, Massachusetts, United States) according to the Nanopore protocol. PCR products were purified using 1x AMPure XP beads (Beckman Coulter, Brea, CA, USA) and quantified using Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, Massachusetts, United States). The purified DNA was then normalized for end-prep and native barcode ligation reactions according to the PCR tiling of COVID-19 virus protocol with Native Barcoding Expansion 96 (EXP-NBD196, Oxford Nanopore Technologies). Barcoded libraries were then pooled, purified with 0.4x AMPure XP beads and then quantified using Qubit dsDNA HS Assay Kit. Purified pooled libraries were ligated to sequencing adapters and sequenced with the Oxford Nanopore MinION device using R9.4.1 flow cells for 24-48 h.
For bioinformatics analysis, the recommended ARTIC bioinformatics workflow (version 1.2.1) was used with minor modifications applied as described previously 26,27 . The modifications include reducing the minimum length at the guppyplex step to 350 to allow potential small deletions to be detected and increasing the "-normalise" value to 999999 to incorporate all the sequenced reads and the high accurate mode was used for basecalling with an increased QC passing score from 7 to 10. The sequence NC_045512.2 obtained from NCBI was used as the reference and the alignment files produced by Medaka were inspected using Integrative Genomics Viewer (IGV) (2.8.0) to verify the mutations called by the ARTIC pipeline. SARS-CoV-2 lineage was assigned using the PANGOLIN software suite (v4.0.6; accessed on May 6, 2022) 57 . All sequences were deposited onto the NCBI and GISAID database (Supplementary Data 1).
Time-resolved phylogenetic tree. Time-resolved phylogenetic tree was constructed using TreeTime program (version 0.9.0-b.2) as we described previously 27 . In addition to the described homoplastic position masking, we also removed the highly diverged sequences suggested by the TreeTime program. Sequences collected from December 1, 2021 to March 11, 2022 were aligned using MAFFT aligner (v7.310). The maximum-likelihood whole genome phylogenetic tree construction and phylodynamic analysis were performed using IQ-TREE (multicore version 2.2.0 COVID-edition) and TreeTime. For the construction of the phylogenetic tree, 100 standard nonparametric bootstrap replicates were used, and the option -czb was used to mask the unrelated substructure of the tree with branch length representing mutation count of less than 1. The resulting tree, sample collection date and the aligned sequences were used as an input for the TreeTime pipeline to create the phylodynamic tree. The outlier sequences (flagged by the TreeTime program) were removed when visualizing the tree.
Data for two-dose vaccination rate and COVID-19 outbreak in HK. The data for the two-dose vaccination uptake rate in Hong Kong was obtained from the Hong Kong government webpage 34 . The data of the Hong Kong population was obtained from Census and Statistics Department 58 .
Statistical analysis. Statistical analysis was performed using SPSS 26.0 (IBM SPSS Statistics) and GraphPad PRISM 9.1.1 (GraphPad Software, San Diego CA, USA). The Cochran-Armitage trend test was used to assess the age trend in the seropositive rate of neutralization test. The simple linear regression was used to assess the age trend in the sVNT titer. The geometric mean antibody titer for each age group was compared to the ≥80 years age group using the one way ANOVA with Dunn's multiple comparisons test. For the purpose of statistical analysis, an MN titer of <10 was considered as 5. A P value of <0.05 was considered statistically significant.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The source data on sVNT and cVNT titers in this study have been deposited into GitHub (https://github.com/SMUAbdullah/paper-Omicron-BA.2-outbreak-Hong-Kong). The genome sequences have been deposited into the NCBI GenBank and GISAID database (Supplementary Data 1). Data of the vaccination update rate in Hong Kong was obtained from the Food and Health Bureau of the HKSAR government website https://data.gov. hk/en-data/dataset/hk-fhb-fhbcovid19-vaccination-rates-over-time-by-age. The data of the Hong Kong population was obtained from the Census and Statistics Department of the HKSAR government website https://www.censtatd.gov.hk/en/web_table.html?id=1A.
Received: 1 April 2022; Accepted: 15 June 2022; Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/.