I’m not even convinced Sona will match a 96/96% performance in clinical trials. Please stop abusing those stats. Pretty sure we’ll get there on specificity but I’ll be incredibly impressed if either company meets 96% sensitivity “in the wild”. But as discussed before, you don’t need to hit that to identify and isolate the most infectious spreaders, and in do so put a massive dent in the R factor.
AfamaMan, I think he meant a £1 profit per test. That sounds good and all and will be great to help the most people possible, but could be risky to pitch it initially at a low profit margin... Companies can usually only afford to run at a low profit margin if both demand and the cost basis are stable/ predictable. Selling at a low margin is a big risk if they a/ produce a large quantity of product without sales agreed in advance, and/or b/ haven’t road-tested the full supply chain and underlying assumed cost basis in anger. A higher margin of profit is not necessarily greedy, it just builds in a cushion against these risks which enables a business to be more resilient. So absolutely agree with Ken’s point about prices most likely coming down in the medium to long term once demand/cost/competitive landscape are clear.
Up front... I wonder if massive orders demonstrating demand well in excess of possible supply might give them the confidence to set the price low? There is also the element of what the competition are doing as well. I seem to recall seeing someone quote that this Sona price of $50 (in Canadian CAD mind) included a hefty 70-80% profit margin (sorry, can’t recall source), which would give them considerable leverage to reduce their price if need be.
Thanks for the link BBN - that was a great piece of research. It was 12 days after Vir announced the independent lab verification by two labs of their antibodies blocking live virus that they announced a deal. Avacta announced independent verification of blocking a model virus by one lab on June 18th - not quite as far advanced as the aforementioned Vir announcement, but it is now 19 days past that point.
Too true, Will, and I hope you get to safely see your granddad in person again soon.
As well as bad news for the usefulness of antibody tests, this could be bad news for vaccinating our way out of this problem, in the scenario where a decline in antibodies means little or no lasting immunity is gained.
If it turns out antibody tests +/- vaccines are not the way out of this... There will be a greater and longer-lasting need for some combination of rapid/antigen testing, PCR/lab testing and neutralising therapies... Avacta has a finger in each of these pies - seems like great strategic positioning from the BoD.
I make this point with no joy though. I found it sobering today to see the UK stats put in the context that an estimated 1/1000 people in the UK (c.60k) have died of Covid-19 already this year... and with about 90% of the population still to catch it? It seems this is a nascent industry that may unfortunately continue in some form for far longer than our conservative demand forecast models.
Just to manage the expectations of our more skittish visitors... That video interview was published on Thursday 2nd July, in which Al’s exact words were: “We’re in the process of optimising... and that’ll just take a couple of weeks.” Another way of interpreting this is a couple weeks after the point he said that, which would mean optimisation to be completed next week - then maybe we immediately get an RNS to say so, or maybe we have to wait from then until there are initial performance data to publish. An RNS next week would also fit precisely with the broker note that suggested about 3 weeks of optimisation from the previous RNS.
My quiet hope this week is for an RNS on the neutralising therapy, based on analogy with the Vir timelines, as laid out elsewhere by BBN. But that’s just an educated guess with a hefty smear of uncertainty that thwarts a simple 1:1 read-across on timescales. We could also see BAMS or manufacturing partners.
It’s going to be an interesting month one way or another. By the end of it we could have the perfect storm (from an investing point of view - tragic from a human point of view) of a burgeoning second wave in the UK, and numerous positive Covid-19 test/therapy updates from Avacta.
Agreed hants, happy to read all genuine views, but there is a bit much unnecessary infighting going on. On the positive side, at least the blatant derampers and green boxes are in remission the last few days (hence the SP creeping up again), but I feel on the negative side, without the obvious common enemy, the usual suspects here tend to turn on each other. If we could all just keep posting interesting and respectfully contrasting views and research on the company, that’d be great thanks. Of course I love a bit of the banter too so long as it’s all given and taken in good spirit. It’s easy to ignore if you don’t like that. Love and respect to all, enjoy the rest of the weekend.
I would also disagree on that. Everyone has been eloquently suggesting otherwise all day.
There has been a broader question to how sensitive an antigen LFA for SARS-cov-2 can be. I remember an ‘expert’ from the WHO on radio 4 a couple months back saying that antigen tests were limited to low sensitivity - maybe 40-80% at best, and this because there is no amplification of the RNA in the sample, as there is in PCR. The first antigen test out by Quidel (albeit not sure if an LFA) has an LoD of only 6 million copies/ml (not good). There is everything to prove to the market for the format.
We know Avacta has made a high performance LFA for Zika virus before. That reads across well for Affimers on test strips. The Sona test might have different reagents, but has a lot in common with the Avacta test - same developers, same virus, and same compressed timescale. They’ve shown it is possible, and with what looks like a good LoD. They’ve shown it can be done and set a high benchmark around which the Avacta test will probably sit, one way or the other. I think we’re all sold on the qualities of Affimers as reagents of choice - else why would you be invested here? There is a good chance they will work better, as they are highly selective, high affinity and smaller than antibodies, not to mention faster/cheaper to manufacture and more stable at a range of temperatures. In any case, the market’s big enough for both tests, so both being high quality would be great for the world. Sona have also shown that if the market is waiting to rerate on successful test results, well there’s a lot still to price in. Lots to love here. I’m not sure why you’re so determined to disagree... But then here I go arguing the point, so who am I to criticise?
What happened CS? You used to be more reasonable and polite on here until you jumped on the bashwagon of late. Hope you and the wife recovered well from the covid-19 you said you had tested positive with.
Yep we don’t have a proven delivered test with results yet. But pointing out the lack of certainty is the language we always see from the most cynical of derampers, so I’m surprised to see you doing so, unless one of them has hacked your account! In which case, congratulations, you are reaping an epic wealth of replies. Of course, the nature of investing in opportunities with unproven potential and high upside (especially if overweight in them), is you want to make damn sure as much as you can you do your research to understand the probability of success, and anything that influences your perception of that probability is very reasonable to consider. This news is one such thing, from our antigen test’s closest analogue - a very much related test. The Avacta test’s cousin, if you will. It doesn’t create any certainty, but can certainly inform our understanding of the Avacta test’s likely performance and overall chance of success. If you would rather go all in and just hope for the best... well good luck, you’ll need it.
Have a lovely weekend all, cheers.
Just be careful with those 96%s! To manage your expectations... These were lab results from ‘spiked’ model samples, not clinical results from actual infected patient samples, so we can’t compare them to Al’s target %s just yet. But still great news, and - tentatively - the LoD looks good.
It shows what’s possible in this type of assay, from this developer, for this virus. If we believe there are advantages in the Affimer reagents (cost, stability, higher concentration > higher performance?) and that they will work at least as well as the competition, then hearing that Sona’s product is performing well is obviously great news.
After a load of digging and a couple educational rabbit holes, I’m personally coming around to this data, as much as we've found the units of LoD in TCID50/ml are frustrating to work with. The best comparator I’ve found (perhaps still not perfect, but decent) for TCID50 concentrations of live virus in positive samples would be figure 1 in this paper, by which Sona’s LoD would look pretty darn good! Note that LoDs given in TCID50 from tests validated against a heat-treated or weakened form of the virus, are not a good comparator. Remember Sona had to use a live viral culture after issues working with heat-treated virus.
Also as pointed out by BBN, and as stated in the press release, these are laboratory validation results, not clinical - which will come later. So nothing dodgy - apologies if I was involved in giving this impression! Although we still don’t know clinical S&S, they got a great result, and deserve their congratulations right now. Avacta’s time will come.
Duly noted, Ophidian, although it seems to me like it’s still a unit of concentration that can be scaled once you know the conversion factor? That’s what the Arnaout et al. paper does.
But point taken... so, let’s look at TCID50 for some live virus samples from elsewhere in North America without conversion to another unit...
And if that fails, google:
Predicting infectious SARS-CoV-2 from diagnostic samples Jared Bullard
Paper from researchers in Winnipeg (which holds a fond place in my heart after I played in a sports tournament there a couple years back!)
See figure 1 - seems to show that a sensitivity of 10^2 TCID50/ml for live virus is pretty darn good. Leaving quite a lot of wiggle room vs most of these measurements. Notwithstanding any possible variations in strain, sample treatment, etc. Makes me feel much more bullish about the Sona (and Avacta) tests after seeing that.
Hi BBN, thanks for the response. You are right of course, and I acknowledged this fact when I posted the follow up about this being laboratory, not clinical validation. The irony is I’ve actually read all those news items, and your thread, and posted on this at length myself! It is great news for Sona, of course, and bodes well in blazing a trail for Avacta to pass through the potential technical and regulatory hurdles.
From reading other comments here, I think a lot of people saw the sensitivity/specificity and thought these correspond to the 80/95% target and mean it’s a “high performing” test, and perhaps I was led into that trap as well initially, which is why I pointed out that interpretation doesn’t match with the detail in the press release. Apologies, it was late.
However I still think there is merit in digging into this LOD performance to see how it might compare with the UK’s Target Product Profile LoD in copies/ml, and from there get a read across to potential clinical sensitivity. To this end, the first paper I linked (Arnaout et al.) is very helpful.
But as discussed, because of the units of TCID50 it’s hard to get an exact read across to equivalent copies/ml. From the Arnaout et al. (Harvard/BIDMC, Boston) paper, “The first EUA antigen detection assay, the Quidel Sofia2 SARS Antigen FIA, has an LoD of approximately 6 million [copies/ml] in a contrived universal transport medium sample collection. Although the package insert indicates the LoD using TCID50 units, the BEI Resources control material referenced lists both TCID50 and genome copies/mL, allowing the calculation of the latter and an associated estimated clinical sensitivity of 31%, i.e., it would miss 7 in 10 infected patients.” ” This package insert gives LoD at a TCID50 of 8.5*10^2 per ml and details the exact viral culture used, allowing the conversion. Based on:
Unfortunately TCID50 is dependent on the viral preparation, so we can’t necessarily assume the same relationship between Sona’s reported TCID50 and copies/ml as for the Quidel or any other test - if we did, Sona’s LoD would be c.1.5 * 10^6 copies/ml. However, the Quidel test was validated with heat-deactivated virus, whereas Sona’s used live virus culture. Since (per wiki) TCID50 “quantifies the amount of virus required to kill 50% of infected hosts or to produce a cytopathic effect in 50% of inoculated tissue culture cells”, it seems intuitive that the same TCID50 should equate to fewer copies/ml for live viral cultures, implying a potentially much better sensitivity for Sona. I’d be happy to hear if anyone has a more informed view on this last point, or can find any equivalence data for live SARS-cov-2 cultures. I can find nothing on the MRIGlobal website.
In fairness they do say “laboratory validation”. Clinical validation no doubt will follow.Its a good result. My point is, don’t confuse these numbers with clinical validation numbers or compare against the clinical sensitivity targets.
Excellent s/s numbers if they can pass them off. I dunno though... They say “To generate the sensitivity data, the remnants of each negative sample were spiked with gamma irradiated COVID-19 virus and the tests rerun to determine the positive results, generating the above result of 96%.”
So, these are not ‘real’ positive clinical samples. How representative they are would depend on what range of viral loads they were spiked with. On which, they state “Current studies show positive COVID-19 patients presenting symptoms have viral loads in the 10^4 - 10^6 [TCID50] range.”
As stated below, it’s not clear exactly how that relates to copies/ml, but this paper here gives a neat distribution plot for patient viral loads in copies/ml, as estimated on an Abbott PCR system with an LOD of 100 copies/ml. Note the range spans 7-8 orders of magnitude (albeit covers asymptomatic as well), but the 10^4 to 10^6 TCID50 range quoted by Sona seems oddly narrow to me:
And here are a range of LOD data for molecular SARS-cov-2 tests: https://www.biocentury.com/article/304801/limits-of-detection-for-fda-authorized-covid-19-diagnostics
The most sensitive ones quoted in TCID50 are about 0.01 TCID50/ml, whereas the best in class for copies/ml are at 40 & 100 copies/ml.
I’m a bit loathe to assume any sort of direct conversion, but let’s just say Sona’s reported LOD of 210 TCID50 (per ml? -unstated) looks ~20,000 times less sensitive than the Roche cobas molecular (PCR) test. It’s a heck of an achievement nonetheless, but seems like a bit of creative reporting to get to 96% estimated clinical sensitivity from there, if you ask me. Not trying to put it down or anything, but understanding these nuances will be important when comparing tests.
Not an expert, just numerate.
It’s a great point, Tom. I missed that link across in the first post. RingTheBell links somewhere in this thread to a lancet paper with a nice plot of viral load in swabs vs saliva, showing generally good agreement.
In short RD, I would say no.
A 90% sensitive test would mean roughly a reduction by a factor of 10 in the R factor, if everyone was being tested all the time and immediately isolated if infected, and those with false negatives were free to keep infecting others without ever testing positive, and if all people were equally infectious...
But in reality it’s a complex system with a lot of variables, and R is estimated across the whole population, for whom:
1/ Not absolutely everyone is going to be tested absolutely all the time
2/ Those who turn up false negative on one test might be detected on the next if their viral load increases, or it was a test failure the first time.
3/ Regular testing will preferentially detect the individuals shedding highest viral loads, who would be among the more infectious.
Thanks RTB. I missed that step of correlating oropharyngeal saliva swab to saliva samples, good to hear it roughly correlates - do you recall if that was with a conversion factor, or ~1:1?
I think you put nicely what I was trying to get at - that anything meeting or exceeding the “acceptable” LOD of 1000 copies/ml will be a game changer, and more than sufficient to detect many (most?) asymptomatic and presymptomatic infectious cases.
There was this paper as well:
Once again, Figure 2’s a belter - relates a proxy to sensitivity directly against viral load.
PL, 13M - hi! Hope you’re right about detection “within the range”, but as discussed, I don’t think we need to achieve that to be world beating. Anyway, we’ll see! Indeed... we will see. The only thing that’s left to do... is see. And, speculate at length in the meantime.
Going back to my original post,