Read on Twitter
Reposting to make some minor corrections. Building on @kristjanmoore and using info available up to today, I assign a 90% probability to at least one of the new variants being >30% more transmissible (see https://pandemic.metaculus.com/questions/6031/more-transmissible-variant-to-infect-10m/). See my reasoning+other points below
As always, let's start with the base rate. I'll only consider SARS-2 since it's already been around ~1 year. Have there previously been mutations that have made SARS-2 more transmissible? If so, how often do they seem to occur — and how much more transmissible do they make it?
The mutation rate of SARS-2 is ~9.8 × 10^−4 substitutions per site/year, resulting in 1-2 mutations/ month for a given lineage, which is comparable to that of other coronaviruses. Note the vast majority of mutations either harm the virus or have no effect https://www.nature.com/articles/s41467-020-19818-2
The spike protein contains the part of the SARS-2 virus, the receptor binding domain, that binds to the human ACE2 receptor and uses it to enter our cells. Any mutations in this spike might cause changes in vaccine efficacy and/or viral transmissibility https://www.nature.com/articles/s41467-020-18319-6
OK, so to date (excluding the UK/SA variants in question) have there been any mutations in the spike protein that we're fairly confident has affected vaccine efficacy and/or viral transmissibility? To my knowledge, there are none for the former and just one for the latter — D614G
Generally there are two broad approaches, epidemiological and virological, to assessing whether a new strain/variant has increased transmissibility. Very strong cases via both approaches have been made for the argument that D614G is more transmissible
Epidemiological: the frequency of the D614G mutation went from ~0 to ~1 between Feb and June. But is this bc of selection pressure or random chance? This is the issue with the epi approach — while it is much quicker, it is less able to establish causality https://www.nature.com/articles/d41586-020-02544-6
sometimes things like this happen through chance via a founder effect — a decrease in genetic diversity resulting from a new population being established by a small # of individuals. We'd expect this to be esp likely w/ SARS-2 since ~80% of cases are caused by only ~20% of ppl
Virological: this slower approach is much better positioned to establish causality. It usually uses animal models/pseudotyped viruses to directly study how efficiently a new variant can infect cells by binding to ACE2.
the virological approach has, in my view, clearly established the higher transmissibility of the D614G mutation. See https://www.nature.com/articles/s41586-020-2895-3 and https://science.sciencemag.org/content/370/6523/1464
OK, so now that the virological approach has apparently confirmed the epi, let's go back to what the epi approach suggests: the D614G variant is ~20% more transmissible (variant R0 is~4.0 vs ~3.1 for nonvariant), which is the result of selective pressure https://www.medrxiv.org/content/10.1101/2020.07.31.20166082v2.full.pdf
Now, let's use the above D614G explanation both as part of our base rate (we now know this epi-proposed, virology-confirmed process has happened at least once in the past year; ~1/yr?) and as a story to explain where we're at with understanding these new variants.
With the new variants, we are at the initial stage in which epidemiological evidence suggests greater transmissibility, but have not yet had confirmation of this hypothesis from laboratory experiments that directly study relative infectiousness.
On 8 Dec, infectious disease experts in the UK noticed that the frequency of a new strain, B.1.1.7, was increasing in southeastern England both in frequency as a proportion of overall cases and possibly contributing to an absolute increase in cases https://www.sciencemag.org/news/2020/12/mutant-coronavirus-united-kingdom-sets-alarms-its-importance-remains-unclear
NERVTAG, an expert advisory group, informed the UK Gov on 18 Dec that based on "preliminary modelling data" (epi data from CoG-UK) they think the new strain spreads "significantly more quickly." As such, PM Johnson imposed a new regional lockdown on 19 Dec https://www.gov.uk/government/news/prime-minister-announces-tier-4-stay-at-home-alert-level-in-response-to-new-covid-variant
CoG-UK (COVID-19 Genomics Consortium UK) is the group that released a report on 18 Dec in which they suggest this new B.1.1.7 strain arose from selection pressure in a chronically-infected individual and that it has multiple mutations in the spike sequence https://virological.org/t/preliminary-genomic-characterisation-of-an-emergent-sars-cov-2-lineage-in-the-uk-defined-by-a-novel-set-of-spike-mutations/563
CoG-UK does not give an estimate of how much more transmissible B.1.1.7 is, but does note "associated with an unexpectedly large number of genetic changes including in the receptor-binding domain and associated with the furin cleavage site" https://virological.org/t/preliminary-genomic-characterisation-of-an-emergent-sars-cov-2-lineage-in-the-uk-defined-by-a-novel-set-of-spike-mutations/563
At this point, I'm already very concerned —this same CoG-UK team, including many of the same experts, also correctly previously argued that the D614G mutation results in greater transmissibility (~20% more), which has recently been confirmed in lab studies https://www.medrxiv.org/content/10.1101/2020.07.31.20166082v2.full.pdf
On 23 Dec, the CMMID COVID-19 working group released a report using epi data to estimate that B.1.1.7 strain, which they refer to as VOC 202012/01, is "56% more transmissible (95% credible interval across three regions 50-74%) than preexisting variants" https://cmmid.github.io/topics/covid19/uk-novel-variant.html
A preprint released today finds lower Ct (greater SARS-2 RNA content) values for B.1.1.7 samples, from which it's "possible to infer a relatively higher viral load in these specimens" — meaning the ppl these samples come from are perhaps more contagious https://www.medrxiv.org/content/10.1101/2020.12.24.20248834v1
OK, so we're starting to get a lot of evidence that *strongly* suggests this new strain is more readily transmissible. But, as Dr. Racaniello notes, there is no virologic data that confirms this and so we can't rule out the founder effect
It will take a few weeks for animal model data to confirm the existing epi data, but I think the evidence is strong enough to state the following using info available to me up until today:
-80% chance this specific new strain is >30% more transmissible
-90% chance *any* new strain before mid-2021 is >30% more transmissible
-65% chance this specific new strain is >50% more transmissible
-71% chance *any* new strain before mid-2021 is >50% more transmissible
-90% chance *any* new strain before mid-2021 is >30% more transmissible
-65% chance this specific new strain is >50% more transmissible
-71% chance *any* new strain before mid-2021 is >50% more transmissible
Why should you trust me? I've been keeping close track of COVID as part of my work for @metaculus
(and recently started forecasting on it too)and I've placed 1st among 1000s of forecasters in both the GJP2.0 Covid-19 Forecasting Tournament and GJO's Coronavirus Outbreak challenge
(and recently started forecasting on it too)and I've placed 1st among 1000s of forecasters in both the GJP2.0 Covid-19 Forecasting Tournament and GJO's Coronavirus Outbreak challenge
I'll revisit this subject in a few weeks. But basically as of now I think this warrants sufficient concern so as to prompt an immediate re-imposition of tough mitigation measures to bridge us over until at least the point where nearly all of the >55 yrs population is vaccinated
