Back

Modeling and Predicting Antibody Durability for mRNA-1273 Vaccine for SARS-CoV-2 Variants

Luo, G.; Hu, Z.; Letterio, J.

2021-05-04 infectious diseases
10.1101/2021.05.04.21256537 medRxiv
Show abstract

Recently, the antibody titer levels have been followed in 33 adults who received the mRNA-1273 vaccine for 6 months. With single dose estimated effectiveness of 92.1%, we combine this knowledge with corresponding antibody levels to model and estimate the long-term durability of mRNA-1273 vaccine. Additionally, we integrate studies about differences in antibody neutralization to SARS-CoV-2 variants to understand how variants can affect the durability of the vaccine. The estimated days after first injection for binding antibodies to fall below levels of those from day 15 is 411 days. The estimated days after first injection to fall below the lower limit of detection of 20 GMTs is 327 days for pseudovirus neutralization and 461 days for live virus neutralization. Our model has pseudovirus neutralization against variant B.1.351 falling below 20 GMT on day 100; variant P.1 on day 202, variant B.1.429 on day 258; and variant B.1.1.7 on day 309. The data used contains many limitations including the small sample size with older age bias, sensitivity of the neutralization assays, and limited data on variants. Still, we believe mRNA-1273 two dose vaccine can provide over a year of protection against COVID-19 from the initial D614G variant. It is likely by the second year, protection against COVID-19 will fall below single dose efficacy. Therefore, there should be consideration for a booster shot a year after the first set of vaccines. If there is an observed increase in variants with higher resistance such as B.1.351 and P.1, a booster vaccine against the newer variants should be considered to increase protection against resistant variants.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.