This is the paper that I was so excited to read yesterday; there's a link to the full PDF at that site. It's an amazing paper in terms of the obstacles the researchers had to overcome to make a reasonable analysis.
tl;dr - Multiple lines of evidence all agree that D614G appears to be more infectious but not more deadly than the original strain; it's now the main strain pretty much everywhere in Europe and the US from April onward. While it spreads faster, the mutation may make it more vulnerable to vaccines (i.e. vaccines work better) because it is more exposed to antibody deactivation. It also showed up more quickly and widely in younger people for reasons that aren't clear, especially weird since transmission across age groups is very common. It appears to cause a higher viral load but not more severe symptoms or outcomes, and this higher viral load may cause a small bias in the statistics by making it more likely to show a positive test result. (They compensate for this and many other biases in the analysis.) The location of the mutation is NOT a location being targeted by any of the vaccine candidates, so there's no reason to expect any problems in that respect.
Second, the quick highlights. A mutation in the spike protein, changing the amino acid at position 614 from D to G, appeared early on in the epidemic, originating from China along with the original strain. This strain then became the dominant form in many locations (including Europe and the US)...but did it take over because it's more contagious, or was it random luck as the epidemic took off?
Previous studies found that the change to the structure made the spike more effective at attaching and getting the RNA into the host cell - so that's consistent with higher infection rates. Tests in cell cultures showed it replicated faster, but that's in vitro and may not apply. Tests in animals showed it spread faster, but that doesn't prove it in humans. The way it became the dominant strain in many locations at the same time was fairly conclusive, but not airtight evidence for higher spread.
These researchers wanted to take advantage of the immense database of sequenced SARS-CoV-2 in the UK. (There was a huge effort there that was carried off very well, with proper data on the sources, making it an incredibly valuable resource.) They performed three statistical tests to look at how the 614D and 614G variants spread in clusters, to answer the question "can we be >95% certain that the spread was from differences in contagion and not other factors?"
Two of the three tests gave a conclusive "yes", one gave a "can't tell" result likely due to the high statistical uncertainty in the method. ("Can't tell" doesn't mean it didn't spread that way, just that this particular test does not provide clear evidence.)
In the first two tests, they were able to quantify that the 614G variant seems to spread 21% to 27% faster than the original, with a 95% confidence interval that ranges from a few percent faster to 50% faster. 25% would be huge, taking an R0 of 3.0 and raising it to 3.75, or an Rt of 1.0 to 1.25, which is the difference between no growth and the growth rates in the most-affected US states today. 😕
Finally, some notes on the things the researchers had to deal with to get valid results. Their main technique was to compare the course of outbreaks of each variant as they popped up in the UK. Of course, the D variant outbreaks were mostly in the early, pre-lockdown stage of the epidemic, and the G were most common in the later stage, where widespread mitigation measures changed the nature of cluster. They had to compare outbreaks at similar dates to each other.
They also had to be sure each outbreak cluster really was a chain from one source, so they compared the genomes for the number of mutations to filter out G variants that may have been imported from elsewhere and just happened to appear at the right time.
There were many other similar pitfalls, and the researchers were remarkably diligent at working through them all. They also examined the reported symptom severity and outcomes between the variants, correcting for age and other influences, and found no significant differences.
This is also a great example of cross-discipline science giving us pieces of the answer. The fact that 614G became the dominant form in country after country gave us a broad epidemiological hint that it was likely to be more infectious. Molecular modeling identified the physical change to the spike protein, making it stiffer and more likely to open up, facilitating transfer but also making it potentially more vulnerable to antibody neutralization. In vitro studies showed MUCH faster replication, animal studies showed faster spread, and now a rigorously controlled study of outbreak clusters gives us the 21-27% likely increase in transmission.
I need to look up what the variants were across Asia in Jan-Mar, and what they are currently, to see if this may be part of the puzzle of why eastern countries have had so much more success.
Similarly, there's a new strain first isolated in Spain over the summer, that has become the dominant form there, in the UK, and a few other places in the months that followed. (It contains the 614G variation and a few more changes.) Because it's so much more recent and not as globally dispersed, it's no clear yet if it has different characteristics but it's looking like it might.