Researchers have identified a pattern of deletions in the spike (S) glycoprotein of SARS-CoV-2 that can prevent antibody binding. Virus lineages featuring this mechanism are currently being transmitted between individuals globally, they say. Their results - reported after analyzing nearly 150,000 S gene sequences collected from many parts of the world - exhibit a form of virus "escape" that resulted from a common, strong selective pressure; for example, the authors identified at least nine instances where deletion variants arose in patients whose COVID-19 infections were persistent. So far, the strongest indicator of protection against SARS-CoV-2 appears to be humoral immunity, such as by antibodies, to the S protein. Recently approved vaccines deliver the S protein, training the body to recognize and attack it. Whether SARS-CoV-2's S protein will evolve the ability to evade the immune system through mechanisms like deletions, which virus proofreading cannot correct, remains unknown. Using samples from a cancer patient infected with COVID-19 who died 74 days after COVID-19 diagnosis, Kevin McCarthy and colleagues cloned S genes and identified variants with deletions. This prompted them to investigate patient metadata sequences deposited in a database, called GISAID, containing whole-genome sequences of SARS-CoV-2. They identified more patients with deletions in the S protein. Searching nearly 150,000 sequences in the full GISAID database, they found 1,108 additional instances of viruses that exhibited deletions in S. By examining the atomic structure of the S protein, McCarthy and colleagues mapped these deletions to defined antigenic sites. This led them to develop the hypothesis that these deletions would produce viruses that no longer bind a published strongly neutralizing antibody. Using a panel of deletion mutations, they confirmed this hypothesis. The deletions they uncovered have repeatedly arisen in genetically distinct viruses, at various times, in multiple countries, the authors say. During evaluation of this manuscript," they note, "multiple lineages with altered antigenicity and perhaps increased transmissibility have emerged and spread." They conclude: "These real-world outcomes demonstrate the predictive potential of this and like approaches and show the need to monitor viral evolution carefully and continually."