Analysis of well-annotated next-generation sequencing data reveals increasing cases of SARS-CoV-2 reinfection with Omicron

SARS-CoV-2 Next-Generation Sequencing Data

We report our analysis of the Danish COVID-19 Genome Consortium accessible through GISAID that examined individual cases of host reinfection across VOCs, including sub-lineages3. There are currently 21,708 reinfection entries available spanning March 1st 2020 to August 28th 2022. Each entry reported the exact collection date of both the initial infection and reinfection for the same individual, along with NGS sequences for the second infection’s viral genome. The primary infection and reinfection time-points were recorded as host metadata, allowing for the period between infections to be measured. A smaller portion of dataset entries (7595) reported the viral Pango lineage of both infections, in addition to the collection date, and NGS data for the initial infection and reinfection samples from the same individual. The Pango lineage nomenclature uses NGS to phylogenetically classify a virus based on its genomic composition leading to the identification of viral strains16,17. The majority of the dataset did not have NGS data for the initial infection, and instead relied on a reverse transcriptase polymer chain reaction (rtPCR) assay to determine if the subject was positive for SARS-CoV-2 on the reported date. The rtPCR test does not specify the viral Pango lineage, and therefore those samples were not included in the analysis shown in Fig. 1. Because the samples still contained a confirmed initial infection, they were only included in Fig. 2a, as the NGS of reinfection provided confirmation of variant Pango lineage required for analysis. A small subset of cases, 70, (<1% of total cases) had identical Pango lineages for the initial infection and reinfection. These cases were removed from the dataset because they did not satisfy the strict Pango lineage filtering methodology. This threshold was chosen because even with small nucleotide or amino acid differences within the same Pango lineage, those entries could be consistent with persistent and unresolved infection rather than reinfection18.

Fig. 1: Reinfection Case Percent and Intervals by Variant.
figure 1

a Heatmap showing the frequency of total reinfections between two variants for both an initial infection and reinfection in Denmark. Raw counts shown below frequency value in parenthesis. No data was available for blank white squares. n = 7467 reinfection cases. b Heatmap showing the reinfection frequency between an initial Omicron infection and a second Omicron infection by sub-linage in Denmark. Raw counts shown below frequency value in parenthesis. No data was available for blank white squares. n = 340 Omicron-to-Omicron reinfection cases. c Scatterplot showing the time between cases (weeks) for the first and second infection of different variants in Denmark. Means of groups are shown with black bars. The red square highlights a number of early Omicron-to-Omicron cases mentioned in the text that occur before a 10-week period. n = 7467 reinfection cases. d Scatterplot for the time between cases (weeks) for Omicron-to-Omicron infections by lineage in Denmark. Means of groups are shown with black bars. The red square highlights early Omicron-to Omicron cases mentioned in the text that occur before a 10-week period for specific sub-lineage. n = 340 Omicron-to-Omicron reinfection cases.

Fig. 2: Reinfection percentages and sequencing counts.
figure 2

a Denmark sequencing percentages overtime stratified by variant represented as line plot with a bar plot overlay representing second infection percentages stratified by variant that include NGS and rtPCR data. b Worldwide sequencing counts stratified by continent over a period of one-month intervals. The data is also shown as percentage of cases sequenced over periods of one month in supplementary figure 2.

Reinfection case percent and intervals by variant

For each NGS-characterized VOC discovered and reported with associated Pango lineage data, we found an increasing reinfection frequency favoring reinfection with the Omicron VOC (p < 0.0001, Chi-squared test) (Fig. 1a). 26% of individuals infected with the original viral strain showed increasingly higher reinfection frequencies with subsequent variants. Those initially infected with the Alpha variant had no cases of reinfection due to Alpha; however, increasing frequencies of reinfection from Delta (2.3%, 169 cases) and Omicron (25.1%, 1875 cases) were observed (Fig. 1a). For those initially infected with Delta, reinfection due to the Delta variant was limited (<1%, 18 cases), but 41% (3060 cases) were reported for Omicron variant reinfections. Thus far in the pandemic, reinfection within the same variant but different sub-lineages, other than for Omicron, was found to be small (0.3%, 24 cases), yet a higher number of individuals initially infected with Omicron report a reinfection due to Omicron sub-lineages (4.6%, 340 cases). There is a high frequency of reinfection with Omicron among all those reinfected since March 2020, during which time a total of 93.2% reinfections were due to Omicron. These results suggest that a primary infection with either the Original, Alpha, or Delta variant does not provide sufficient protection against reinfection, in particular for an Omicron reinfection.

In order to investigate this further, we stratified Omicron-to-Omicron reinfection cases by their major sub-lineage designation (Fig. 1b). The BA.2, BA.4, and BA.5 lineages closely share spike amino acid sequences compared to other Omicron lineages. There are only three mutational differences in the spike protein between BA.2 and both BA.4 and BA.5: del69-70, L452R, and F486V8. Despite this similarity, a high frequency of Omicron-to-Omicron reinfections was reported with significant difference between the observed count distribution and the expected count distribution, showing that there is a relationship between the Omicron sub-lineages for reinfection and is not due to chance (p < 0.0001, Chi-squared test) (Fig. 1b). Individuals initially infected with the BA.1 sub-lineage accounted for a high frequency of total Omicron-to-Omicron reinfections (62%, 211 cases) with the second infections predominately caused by BA.2 (20%, 68 cases) or BA.5 (30%, 102 cases) (Fig. 1b). Similarly, individuals initially infected with BA.2 showed comparably high frequencies of reinfection (38%, 129 cases) with BA.5 (26.2%, 89 cases) (Fig. 1b). Infections designated as “Other” in Fig. 1b did not have a common nomenclature defining the lineage. The high frequencies of reinfection suggest that the characterized difference in the BA.1 or BA.2 and BA.5 spike protein is high enough to hinder the ability of post-infection neutralizing antibodies from BA.1 or BA.2 to bind to BA.5 spike protein12,16. The lineages of BA.1, BA.4, and other lesser annotated Omicron variants had lower levels of re-infectivity compared to BA.2 and BA.5.

The decline of neutralizing antibodies following an infection raises a point of concern for how long natural immunity will last in a given individual. Pre-Omicron models estimated more than 90% effectiveness of initial post-infection immunity, but these estimates decrease to less than 10% after 108 weeks19. Figure 1c and d show the time between infections by first and second VOC seen throughout the pandemic. A stepwise trend of increasing time to reinfection is demonstrated between the different VOC due to their timing of emergence, spread, and dominance in the pandemic (Fig. 1c). In particular, Omicron-to-Omicron reinfections events appear in as little as 3 weeks after the initial infection, with a mean of 22 weeks (Fig. 1c). Of these Omicron-to-Omicron reinfections, 50 of the total 340 (14.7%) cases occur within 10 weeks of the initial infection; marked by a red box in Fig. 1c. A t-test was conducted for the significance of the time between these Omicron-to-Omicron reinfections that occur before and after 10 weeks. These two sub-groups demonstrate significance of the mean reinfection times (p < 0.0001, 95% CI: 17.17–20.19). This led to further analysis to determine what could be causing this significant bimodal distribution within the Omicron-Omicron reinfection group.

These reinfections from Fig. 1c are stratified by sub-lineage in Fig. 1d. The reinfection occurrence before 10 weeks, marked by a red box in Fig. 1d, is predominantly associated with BA.1 first and then reinfection with BA.2. Analysis of Variance (ANOVA) was conducted (supplementary data 1) between these groups in Fig. 1d shows significance between the difference in means for the majority of pairings. The variant sub-lineage analysis in Fig. 1 was achievable due to the availability of NGS data processed using bioinformatics to reveal the Pango lineage of samples in the dataset.

Reinfection case percent

Using a combined dataset of both NGS and rtPCR samples as described prior, a higher number of reinfections has been reported in the past eight months (December 2021 to July 2022) compared to other VOCs in the last year (Fig. 2a). The proportion of Omicron sub-lineages against total infections in Denmark is detailed in supplementary fig. 1. During the global Omicron infection wave, Denmark had a 19.5% peak proportion of reinfection of the total sampled cases in July 2022. In comparison, a peak at 1.4% of total cases during the Delta wave were reported as reinfections. This difference highlights the high level of reinfections observed during the progression of the pandemic into the Omicron VOC. Furthermore, the data shows that when the dominant variant of infection was nearly evenly distributed at 50% between Delta and Omicron, there was not an even distribution of reinfections. The reinfections at that time, December 2021, were caused by Delta in 0.6% of cases, and by Omicron in 1.6% of cases. This is an important finding between Delta and Omicron as the likelihood of exposure was roughly the same, but the likelihood of reinfection was not.

Read more here: Source link