Russell Timothy W, Townsley Hermaleigh, Hellewell Joel, Gahir Joshua, Shawe-Taylor Marianne, Greenwood David, Hodgson David, Hobbs Agnieszka, Dowgier Giulia, Penn Rebecca, Sanderson Theo, Stevenson-Leggett Phoebe, Bazire James, Harvey Ruth, Fowler Ashley S, Miah Murad, Smith Callie, Miranda Mauro, Bawumia Philip, Mears Harriet V, Adams Lorin, Hatipoglu Emine, O'Reilly Nicola, Warchal Scott, Ambrose Karen, Strange Amy, Kelly Gavin, Kjar Svend, Papineni Padmasayee, Corrah Tumena, Gilson Richard, Libri Vincenzo, Kassiotis George, Gamblin Steve, Lewis Nicola S, Williams Bryan, Swanton Charles, Gandhi Sonia, Beale Rupert, Wu Mary Y, Bauer David L V, Carr Edward J, Wall Emma C, Kucharski Adam J
Centre for Mathematical Modelling of Infectious Diseases, London School of Hygiene & Tropical Medicine, London, UK.
Francis Crick Institute, London, UK; National Institute for Health Research Biomedical Research Centre and Clinical Research Facility, University College London Hospitals NHS Foundation Trust, London, UK.
Lancet Infect Dis. 2025 Jan;25(1):80-93. doi: 10.1016/S1473-3099(24)00484-5. Epub 2024 Sep 11.
The emergence of SARS-CoV-2 variants and COVID-19 vaccination have resulted in complex exposure histories. Rapid assessment of the effects of these exposures on neutralising antibodies against SARS-CoV-2 infection is crucial for informing vaccine strategy and epidemic management. We aimed to investigate heterogeneity in individual-level and population-level antibody kinetics to emerging variants by previous SARS-CoV-2 exposure history, to examine implications for real-time estimation, and to examine the effects of vaccine-campaign timing.
Our Bayesian hierarchical model of antibody kinetics estimated neutralising-antibody trajectories against a panel of SARS-CoV-2 variants quantified with a live virus microneutralisation assay and informed by individual-level COVID-19 vaccination and SARS-CoV-2 infection histories. Antibody titre trajectories were modelled with a piecewise linear function that depended on the key biological quantities of an initial titre value, time the peak titre is reached, set-point time, and corresponding rates of increase and decrease for gradients between two timing parameters. All process parameters were estimated at both the individual level and the population level. We analysed data from participants in the University College London Hospitals-Francis Crick Institute Legacy study cohort (NCT04750356) who underwent surveillance for SARS-CoV-2 either through asymptomatic mandatory occupational health screening once per week between April 1, 2020, and May 31, 2022, or symptom-based testing between April 1, 2020, and Feb 1, 2023. People included in the Legacy study were either Crick employees or health-care workers at three London hospitals, older than 18 years, and gave written informed consent. Legacy excluded people who were unable or unwilling to give informed consent and those not employed by a qualifying institution. We segmented data to include vaccination events occurring up to 150 days before the emergence of three variants of concern: delta, BA.2, and XBB 1.5. We split the data for each wave into two categories: real-time and retrospective. The real-time dataset contained neutralising-antibody titres collected up to the date of emergence in each wave; the retrospective dataset contained all samples until the next SARS-CoV-2 exposure of each individual, whether vaccination or infection.
We included data from 335 participants in the delta wave analysis, 223 (67%) of whom were female and 112 (33%) of whom were male (median age 40 years, IQR 22-58); data from 385 participants in the BA.2 wave analysis, 271 (70%) of whom were female and 114 (30%) of whom were male (41 years, 22-60); and data from 248 participants in the XBB 1.5 wave analysis, 191 (77%) of whom were female, 56 (23%) of whom were male, and one (<1%) of whom preferred not to say (40 years, 21-59). Overall, we included 968 exposures (vaccinations) across 1895 serum samples in the model. For the delta wave, we estimated peak titre values as 490·0 IC (95% credible interval 224·3-1515·9) for people with no previous infection and as 702·4 IC (300·8-2322·7) for people with a previous infection before omicron; the delta wave did not include people with a previous omicron infection. For the BA.2 wave, we estimated peak titre values as 858·1 IC (689·8-1363·2) for people with no previous infection, 1020·7 IC (725·9-1722·6) for people with a previous infection before omicron, and 1422·0 IC (679·2-3027·3) for people with a previous omicron infection. For the XBB 1.5 wave, we estimated peak titre values as 703·2 IC (415·0-3197·8) for people with no previous infection, 1215·9 IC (511·6-7338·7) for people with a previous infection before omicron, and 1556·3 IC (757·2-7907·9) for people with a previous omicron infection.
Our study shows the feasibility of real-time estimation of antibody kinetics before SARS-CoV-2 variant emergence. This estimation is valuable for understanding how specific combinations of SARS-CoV-2 exposures influence antibody kinetics and for examining how COVID-19 vaccination-campaign timing could affect population-level immunity to emerging variants.
Wellcome Trust, National Institute for Health Research University College London Hospitals Biomedical Research Centre, UK Research and Innovation, UK Medical Research Council, Francis Crick Institute, and Genotype-to-Phenotype National Virology Consortium.
严重急性呼吸综合征冠状病毒2（SARS-CoV-2）变体的出现和2019冠状病毒病（COVID-19）疫苗接种导致了复杂的暴露史。快速评估这些暴露对针对SARS-CoV-2感染的中和抗体的影响，对于为疫苗策略和疫情管理提供信息至关重要。我们旨在按既往SARS-CoV-2暴露史，研究个体水平和人群水平抗体动力学对新出现变体的异质性，探讨其对实时估计的影响，并研究疫苗接种活动时间的影响。
我们的抗体动力学贝叶斯分层模型，估计了针对一组SARS-CoV-2变体的中和抗体轨迹，这些变体通过活病毒微量中和试验进行定量，并根据个体水平的COVID-19疫苗接种和SARS-CoV-2感染史进行分析。抗体滴度轨迹用分段线性函数建模，该函数取决于初始滴度值、达到峰值滴度的时间、设定点时间以及两个时间参数之间梯度的相应增减速率等关键生物学量。所有过程参数均在个体水平和人群水平进行估计。我们分析了伦敦大学学院医院-弗朗西斯·克里克研究所遗产研究队列（NCT04750356）参与者的数据，这些参与者在2020年4月1日至2022年5月31日期间通过每周一次的无症状强制性职业健康筛查，或在2020年4月1日至2023年2月1日期间通过症状性检测对SARS-CoV-2进行监测。遗产研究纳入的人群为克里克研究所员工或伦敦三家医院的医护人员，年龄超过18岁，并签署了书面知情同意书。遗产研究排除了无法或不愿签署知情同意书的人以及非合格机构雇佣的人。我们对数据进行分段，以纳入在三种受关注变体（德尔塔、BA.2和XBB 1.5）出现前150天内发生的疫苗接种事件。我们将每一波的数据分为两类：实时数据和回顾性数据。实时数据集包含在每一波出现日期之前收集的中和抗体滴度；回顾性数据集包含每个个体在下一次SARS-CoV-2暴露（无论是疫苗接种还是感染）之前采集所有样本。
我们在德尔塔波分析中纳入了335名参与者的数据，其中223名（67%）为女性和112名（33%）为男性（中位年龄40岁，四分位间距22 - 58岁）；在BA.2波分析中纳入了385名参与者的数据，其中271名（70%）为女性和114名（30%）为男性（41岁，22 - 60岁）；在XBB 1.5波分析中纳入了248名参与者的数据，其中191名（77%）为女性，56名（23%）为男性，1名（<1%）不愿透露性别（40岁，21 - 59岁）。总体而言，我们在模型中纳入了1895份血清样本中的968次暴露（疫苗接种）。对于德尔塔波，我们估计无既往感染的人的峰值滴度值为490.0国际单位（95%可信区间224.3 - 1515.9），在奥密克戎变异株出现之前有既往感染的人为702.4国际单位（300.8 - 2322.7）；德尔塔波分析未纳入有过奥密克戎感染的人。对于BA.2波，我们估计无既往感染的人的峰值滴度值为858.1国际单位（689.8 - 1363.2），在奥密克戎变异株出现之前有既往感染的人为1020.7国际单位（725.9 - 1722.6），有过奥密克戎感染的人为1422.0国际单位（679.2 - 3027.3）。对于XBB 1.5波，我们估计无既往感染的人的峰值滴度值为703.2国际单位（415.0 - 3197.8），在奥密克戎变异株出现之前有既往感染的人为1215.9国际单位（511.6 - 7338.7），有过奥密克戎感染的人为1556.3国际单位（757.2 - 7907.9）。
我们的研究表明，在SARS-CoV-2变体出现之前实时估计抗体动力学是可行的。这种估计对于理解SARS-CoV-2暴露的特定组合如何影响抗体动力学，以及研究COVID-19疫苗接种活动时间如何影响人群对新出现变体的免疫力具有重要价值。
惠康信托基金会（Wellcome Trust）、英国国家卫生研究院大学学院医院生物医学研究中心（National Institute for Health Research University College London Hospitals Biomedical Research Centre）、英国研究与创新署（UK Research and Innovation）英国医学研究理事会（UK Medical Research Council）、弗朗西斯·克里克研究所（Francis Crick Institute）以及基因型到表型国家病毒学联盟（Genotype-to-Phenotype National Virology Consortium）。
