Thursday, November 26, 2020

Of Tests and Turkeys


Treatment updates

Bamlanivimab: Lilly’s monoclonal antibody concoction has been granted emergency use authorization by the FDA for mildly-to-moderately ill outpatients, but only a comme ci comme ça from the National Institutes of Health. Its Treatment Guidelines Panel says there is not enough evidence to recommend using bamlanivimab outside clinical trials, except maybe in patients at high risk for progressing to severe disease. 

 Casirivimab and imdevimab (REGN-COV2): The FDA has also issued an emergency use authorization for the Regeneron monoclonal antibody cocktail given to Donald Trump; I suspect the NIH will issue the same caveat as for the Lilly product. Both drugs require intravenous infusion, in a medical setting capable of handling severe reactions, within a few days of a positive PCR swab. Not easy to orchestrate!

Convalescent plasma: Previous negative studies were flawed by transfusing plasma with low titers of neutralizing antibodies. Now Argentinian researchers who avoided that pitfall have, alas, similarly found no benefit in hospitalized patients. Maybe if you give it earlier in disease? Monoclonal antibodies will likely displace it.

Tocilizumab (Actrema): After flopping in several previous trials, this intravenous arthritis drug –aimed at tamping down the excessive immune response underlying some of the worst manifestations of COVID-19 – may have hit a sweet spot. All we have is a vague statement that critically ill patients were more likely to reach a “key efficacy endpoint” if they received the drug. If the researchers provide such scanty information, their results are likely less than earth-shaking.


MMR vaccine: Some think various vaccinations, such as BCG (against tuberculosis), might help prevent COVID-19 through nonspecific immune system stimulation. Now a tiny study reports that among 50 COVID-19 patients who had had MMR vaccine (measles, mumps, rubella) those with high levels of mumps antibodies were less likely to have severe disease. Among the peculiar findings was the lack of association with measles or rubella antibodies, or antibodies following actual mumps. Should I call in all my adult patients and propose vaccinating them with MMR? Not on the basis of this study alone.

Baricitinib (Olumiant): This oral immune modulator, already used for rheumatoid arthritis, has crept into the scanty panoply of approved COVID-19 drugs without my ever having noticed it before. The FDA has authorized emergency use, not on its own but together with remdesivir, for hospitalized patients with breathing problems. As compared with remdesivir alone, hospitalized patients who received the two-drug combo recovered marginally faster (7 days vs. 8) and were somewhat less likely to go badly downhill (23% vs. 28%). The drug makes you more vulnerable to various infections, and its modest benefits were statistically significant only because the study was very large; as usual, no magic bullet.

Nasal spray redux: The University of Birmingham has shouted it from the rooftops: “Anti-COVID-19 nasal spray 'ready for use in humans'.” Ummm, really? “Ready for use” turns out to mean only that its components are freely available. And “anti-COVID-19” turns out to mean only in a petri dish. There’s many a slip ‘twixt petri dish and nose.

Chinese herbs: Here, shortened, is a comment left on one of my Facebook posts: “I know a traditional TCM herbalist and his disciples in China cured COVID cases through online tongue diagnoses. Absolutely Amazing!!! Overall they treated more than 120 suspected cases and 10 cases tested positive. I asked them how confident they are in curing COVID, 100% or 99%, they replied 200%! Only people losing hope or failed current treatments will go to them, and all got cured. And another amazing fact is the low cost of complete cure: only $100-300 to buy herbs. They all volunteer to provide online tongue diagnosis and a basic health survey.” Lots of luck with this one!



Moderna: Moderna’s mRNA-1273 vaccine, which announced positive Phase 3 results a few days after Pfizer did, may be second across the finish line, but otherwise comes out ahead. First, it doesn’t require Pfizer’s absurdly low storage temperatures. Shipping can be in ordinary freezers, and once thawed the vaccine can stay 30 days in a doctor’s fridge versus Pfizer’s 5. Second, Moderna’s preliminary results may still be just a press release, but Moderna volunteers more facts to sink your teeth into, e.g. of 95 cases of COVID-19 among their volunteers, only 5 were in vaccine recipients. Third, the definition of disease was better in the Moderna trial – you had to have a respiratory symptom such as cough plus some total-body symptom such as fever (for Pfizer, a mere sore throat would do). Fourth, they’ve provided an additional piece of data that strongly hints the vaccine can save lives: all 11 cases of severe COVID-19 were in the placebo group. Fifth, many cases occurred in elderly (15) and minority (20) placebo recipients, suggesting efficacy in these crucial subgroups. Finally, they discussed the flu-like side effects in some detail. The trial has not yet been completed but Moderna, like Pfizer, will surely seek emergency use authorization ASAP. 

Sputnik V: The Russians now report 91.4% vaccine efficacy at their second interim analysis, after 39 volunteers have developed COVID-19, and claim 1.2 billion doses have already been ordered worldwide. Not for my arm. 

AstraZeneca: We already knew the Oxford vaccine stimulated antibody and T cell responses in under-55’s, but a new study says that holds for older folk too, with fewer side effects. Immediately afterward, AstraZeneca became the third company to announce preliminary Phase 3 trial results30 cases of COVID-19 in vaccine recipients versus 101 in the placebo group, 70% efficacy. Good as vaccines go, and better than  the 50-60% most experts had predicted, but not in the same league as Moderna and Pfizer’s 95%. Oddly, they found stronger results in a small subgroup that accidentally had received a smaller first dose. That “accidentally” does not inspire confidence, nor do the use of different protocols in different countries, the lack of good results among the elderly, the two volunteers who developed neurological syndromes, or the company’s general lack of transparency. AstraZeneca’s vaccine is cheap and can be distributed in refrigerators rather than freezers, so despite everything it may dominate in the developing world, but I’d guess people in industrialized countries are more likely to choose otherwise.

Pfizer bis: After Moderna’s bombshell press release came out Pfizer hastened to spruce up its own meager one, now saying their Phase 3 subjects had developed 170 cases of COVID-19, 162 in the placebo group and 8 in the vaccine group, for an efficacy of 95%. They also now claim it’s 94% efficacious in the elderly. And that they observed 10 cases of severe COVID-19, of whom 9 had received placebo. What’s your theory as to how they went from 94 cases and 90% efficacy last week to 170 and 95% this week?

And the Pfizer CEO selling off 60% of his stocks the same day his company announced stellar Phase 3 results? What are we supposed to make of that?

I doubt Pfizer will ever be the go-to vaccine. If for nothing else, because none of the other major Western contenders – Moderna, Johnson & Johnson, Novavax, AstraZeneca – require the -70º C storage that makes distribution of Pfizer’s vaccine a logistical nightmare. Even hospitals don’t have freezers that powerful, much less doctors’ offices, even in the rich countries that have made down payments on enough doses for 650 million people.

Like AstraZeneca, Pfizer is unforthcoming with information. As far as I can find their animal results never got beyond the press release stage, and until goaded by Moderna they provided only the vaguest hint of their Phase 3 data. 

As I’ve pointed out elsewhere (Surprises from Africa to OctoberRED ALERTSGood News, Bad News), Pfizer’s research protocol is particularly weak, especially compared to Johnson & Johnson, whose vaccine has the additional benefit of being one-off rather than two-dose. Starting this week, though, they’re hedging their bets with a second Phase 3 trial that tries adding a booster dose. 

Novavax should report its first-peek data by early December, J & J in January, when millions of high-risk people will already have been vaccinated. By February or March there should be complete, published Phase 3 data from all 5 companies, letting us exclude serious side effects, have some idea how long protection lasts, and know whether vaccines prevent infection as well as disease. So hopefully the rest  of us should have a broad array of effective, safe vaccines to choose from. The biggest problem will be how to churn out doses fast enough – Pfizer and Moderna think they can only make enough for 20 million Americans by the end of 2020, barely covering health care workers and nursing home residents. It won’t be easy to convince people to get vaccinated, but I think knowing  the protection is fully 95% will make that task easier – as long as everybody’s transparent about the flu-like side effects.


From Serum to Spit 


There is so much confusion around COVID-19 testing that I’ll make a stab at laying out what the various tests are trying to measure and how good they are at it. Elon Musk, who’s supposed to be a smart guy, nicely illustrated the information gap the other day by mixing up PCR tests with antigen tests and not bothering to look up stuff any 14-year-old can find on the web. 

First a few concepts. False-negative tests are negative results in people who really have COVID-19 – they determine a test’s sensitivity, the percentage of cases it picks up. False-positives are tests that come out positive in uninfected people – determining specificity, the ability to correctly identify the healthy. But what in real life counts more than theoretical specificity is positive predictive value: the probability that a person with a positive test really has the disease. This depends partly on specificity, but heavily on how much COVID-19 is around. A specificity rate of 98% may sound great, but if only 1% of the people being tested really have the coronavirus, then, mathematically, 70% of positive tests will be false positives. If 10% are carrying the virus then 20% of positive tests will be false positives. For this reason we demand sky-high specificity levels, 99.5% or more, especially when random screening is being done in communities where – as is true virtually everywhere – the virus is found in fewer than 1% of noses. When 15-20% of people being tested are actually infected, as is the case all over Italy and in many American states right now, virtually all positive tests will be true positives.

Tests used for diagnosing active infection

1)    Molecular or PCR tests 

These tests directly detect the genomic material (RNA) of SARS-CoV-2, the coronavirus that causes COVID-19, by using the polymerase chain reaction technique commonly used nowadays for infectious diseases. Molecular tests are the best we have for diagnosing active COVID-19, but they have distinct disadvantages. They need those notorious long swabs that feel like they go through your nose right into your brain – that’s because the part of the throat sitting behind the nose is where the highest concentration of coronavirus is found. The person taking the swab has to be specially trained and needs elaborate personal protective equipment. Processing requires highly sophisticated equipment found only in laboratories. Tests take several hours to perform, and they’re expensive. 

And: even the PCR tests miss lots of cases. If you include patients eventually proven to have COVID-19 either on samples from deep in the lungs or on later antibody test, then as many as 35% or even 40%-60% of cases have false negative PCR swabs. The false positive rate is close to zero, however, so a positive test can be trusted.

But PCR tests risk one special type of false positive result, in people who have already recovered from COVID-19 and are no longer infectious. They can continue to have positive PCR tests because the tests can detect even small fragments of RNA. So a positive test in someone who's had COVID-19 can mean they have leftover bits of coronavirus in their system, not that they're still infectious. A COVID-19 patient stops being able to transmit the disease to others about eight days after getting sick, but can stay PCR swab-positive for weeks or even months. Until scientists figured out what was going on this caused lots of trouble for patients, who found themselves forced into prolonged, unnecessary isolation.

About asymptomatic COVID-19: the day a positive PCR swab is taken about 75% of people feel fine. But most of them are presymptomatic, destined to fall ill within the next few days, and are highly contagious. Only about 20% of swab-positive adults are truly asymptomatic in the sense of never becoming sick at all. 

2)    Rapid tests – the rationale

Molecular PCR tests require lab facilities, fancy equipment, and hours for processing, which in practice can stretch into days. We’d all prefer tests that can be performed by personnel with little or no training and deliver results in minutes. A whole variety of technologies and formats already exist, and can be particularly useful for regular screening in high-risk environments, such as care homes or colleges, to nip new outbreaks in the bud.

3)    Antigen rapid tests

These tests, on samples obtained just inside the nose, detect the coronavirus using antibodies against molecules (antigens) on its surface. They sound great: they’re cheap, the sample is quick and easy to get, equipment is small-scale and simple, and results are ready within a half hour. I’ve used similar tests in my office for years to diagnose strep throat. Manufacturers claim their tests have sensitivities between 96.5 and 99.3% as compared to PCR. 

The catch is that in real life the results are much less brilliant. PCR tests already miss many cases, but antigen tests miss even more, catching at best only half of active COVID-19 cases. This severely limits their validity, but though a single antigen test can’t rule out active coronavirus infection, testing high-risk groups such as college students or nursing home residents 2 or 3 times a week is likely to pick up infections in time to isolate patients and put their contacts into quarantine. Given the lags in getting PCR test results, it has been argued that all population screening should be based on rapid tests.

False positive rapid tests are rare, though, with specificity rates around 99.5%. So almost all positive results will be true positives, even in communities with low rates of disease, and there’s no need to verify them with a PCR swab. But the US Food and Drug Administration is warning that those high specificity rates are attained only if the test is performed correctly, and that sloppiness can lead to false positives – not changing gloves between patients, reading results before or after the advised number of minutes, leaving an open test card around…

What would really be useful is cheap and easy home antigen tests for COVID-19. Several are ready to go and waiting for approval. But they’ll likely be harder to perform than the home pregnancy tests they’re often compared to, so easier to screw up.

4)    Molecular rapid tests 

These tests for active COVID-19, also using shallow nasal swabs, are halfway between PCR and antigen tests in complexity, cost, and accuracy. One version using CRISPR technology is supposedly fast and easy enough to be performed in a doctor’s office, which could simplify screening and contact tracing. Rapid molecular tests are better than antigen tests at detecting cases, though with somewhat more false negatives than PCR; they have the same low false positive rate as the antigen tests. They’re widely used in nursing homes in the United States, but don’t seem to have arrived in Italy.

Do-it-yourself versions are coming soon in the United States. The FDA has given emergency use approval to one that has you you stick a swab up your own nose and read off the results in half an hour. The Lucira COVID-19 All-In-One Test Kit will cost around $50 and will be prescription-only – this makes sense, because for many months the supply of tests is expected to be very limited. The manufacturer-claimed sensitivity of 94% and specificity of 98% are unlikely to be achieved in the real world. 

5)    Saliva tests

The holy grail for diagnosing COVID-19 would be an ultra-simple home spit test as accurate as a molecular swab. 

In early September an Italian company claimed to have achieved just that – using a cheek swab and yielding results in just three minutes. They assigned it the unfortunate name, in English, of “Daily Tampon.” Italian physicians immediately warned it was unproven, and within two weeks the whole thing turned out to be a bluff by a failing lamp factory.

So far the spit tests suitable for doctors’ offices are not very accurate, and never caught on. Current “home” ones let you do your spitting at home, but you have to mail the collection tubes to a laboratory for processing.

6)    Blood tests for acute infection

A while back a group from Mount Sinai in New York claimed to have developed a blood test that could detect immune markers of SARS-CoV-2 infection even before a swab became positive. But what this group is publishing now now makes me guess this test – which sounded unlikely in the first place – never went anyplace.

7)    Temperature checks

Ever wonder how many cases of coronavirus infection are picked up by those thermometers people are always pointing at your forehead? The answer: almost none. Since the beginning of the pandemic 766,044 arriving travelers have been screened at American airports for symptoms and known exposures and had their temperature measured by an infrared thermometer. All this trouble has yielded exactly 9 cases of COVID-19, and we don’t know whether any of them had fever alone, without other symptoms. Fever can be a symptom of COVID-19, but it’s a lousy screening test – only half of the patients sick enough to be hospitalized in China and the US have a fever. 

8)    Trying to hold your breath for more than 10 seconds. Right.

Tests used for diagnosing past infection

9)    Serological or antibody blood tests

The best way to know whether you’ve been exposed to SARS-CoV-2 is a blood test for antibodies against it: the IgM kind that are the first to be made and the first to disappear, and the IgG kind that stick around longer. These “serology” tests can’t diagnose acute infection, because the coronavirus takes a week to goose the immune system into making even the early IgM antibodies; by the time a blood test is positive most people are no longer infectious. Antibody testing is important for epidemiologists to understand how the pandemic evolves. A positive result can also be comforting to someone who thinks some weird illness he or she had months ago might have been COVID-19. And when a person has clinical COVID-19 but swabs are negative, checking for antibodies after 3 weeks or so may clinch the diagnosis.

Antibody tests are falsely negative in as many as as many as 20% of people who’ve had COVID-19, though, partly because antibodies can fade away rapidly. And they are somewhat vulnerable to cross-reactivity with other coronaviruses, raising the possibility of false positives.

Asymptomatic patients develop lower levels of antibodies than sick people, especially the valuable neutralizing antibodies, often falling below the threshold for detection. And their antibody titers fall off particularly fast. Fortunately losing your antibodies is not disastrous, because even patients without any more detectable antibodies usually retain a hearty enough response of the cellular branch of the immune system to protect them against reinfection – T cell lymphocytes that remember the coronavirus, and will stimulate memory B cell lymphocytes into producing antibodies again as soon as the person is exposed. 

At least one company has come out with a home antibody test, where you prick your finger yourself and then plant the blood on simple test strips. Why??? Antibody tests just say whether you’ve been infected in the past, and that’s never an urgent question.

Fishing expeditions

A wastewater treatment plant

10) Waste water testing

Back in February, a bunch of Dutch scientists took it into their heads to test the human wastewater of Amsterdam’s Schiphol Airport for the RNA of SARS-CoV-2. Amazingly, they found it, as noninfectious fragments that can’t cause disease. That intuition led to a whole new way of tracking COVID-19 outbreaks in populations and, eventually, to practical screening protocols. Many American colleges, for example, now screen dormitory sewage daily for coronavirus RNA, then swab test all the residents if they find it.

11) Pooling individual samples for testing

Another testing shortcut involves running a single test on 5-8 (even as many as 32) pooled nasal swabs or samples of saliva. If SARS-CoV-2 is detected, the individuals get tested. Widespread use of this method has already saved the British National Health Service considerable money and resources.



Purveyors of bullshit

Sapan Desai: Remember Surgisphere, the mysterious company that supposedly provided the raw data for two major COVID-19 studies that were later retractedMore has come out about its founder. Unsurprisingly, he turns out to be a shady character – a doctor so unreliable that his colleagues don’t believe what he tells them about patients, a surgeon so incompetent that he attracted three malpractice suits in 2019 alone, and a repeat offender for scientific fraud. It seems increasingly likely that the elusive Surgisphere database exists only in his imagination, and that the whole sad tale says more about the dark underbelly of the science game than we researchers would care to admit.

Scott Atlas: Trump’s favorite COVID-19 advisor hit the news twice in two days last week. On November 14 governor Gretchen Whitmer announced she’d close Michigan restaurants and in-person high schools to slow the pandemic. On November 15 Dr. Atlas responded, “The only way this stops is if people rise up.” On November 16 Stanford University distanced itself, saying, “Dr. Atlas has expressed views that are inconsistent with the university’s approach in response to the pandemic.”

Didier Raoult: remember him? The shyster who heads up the infectious diseases department of a Marseilles hospital? Raoult and hydroxychloroquine went together like a horse and carriage – it was his fake research that got Donald Trump all fired up. Now the notorious Docteur Raoult, who had been caught falsifying results in the past, may be getting his comeuppance: hauled before an ethics board, potentially stripped of his medical license.

Puya Dehgani-Mobaraki: This Perugia-based ENT is pushing Protocol “D,” which attracted a wide readership in Italy after a mention in the daily La Nazione (see it here in English). As a purveyor of bullshit he doesn’t hold a candle to the conpetition, but his COVID-19 protocol is an odd mix of science and nonsense. On the one hand it advises face masks and gives good advice for home treatment, such as pulse oximetry monitoring. On the other it throws in long lists of untested supplements, advises unproven medications such as ivermectin and hydroxychloroquine, and gives the extremely dangerous advice that heparin and corticosteroids should be used at home “if saturation under 93” – in reality, anyone with oxygen levels that low should be in the hospital.

Committee Chair Sen. Ron Johnson

The US Senate: On November 19th the Senate Homeland Security and Governmental Affairs Committee held a hearing about early treatment for COVID-19. Can you believe the main subject was hydroxychloroquine? The drug is so discredited that when push came to shove even Donald Trump didn’t take it.

Rudy Giuliani having a hair-dye malfunction

Sidney Powell at Giuliani’s wrap-up press conference: “What we are really dealing with here and uncovering more by the day is the massive influence of communist money through Venezuela, Cuba, and likely China in the interference with our elections here in the United States. The Dominion voting systems . . . created in Venezuela at the direction of Hugo Chavez . . . the mail-in ballots, many of which they had actually fabricated . . . ties of the Dominion leadership to the Clinton Foundation . . . President Trump won by a landslide. We are going to prove it.” Not relevant to COVID-19, but as long as I was on purveyors of bullshit I couldn’t resist.