Convalescent Blood Therapy

A simple and medically feasible strategy is available now for treating COVID-19 patients, transfuse blood plasma from recovered patients. The idea is that the antibodies from the recovered patients will help the infected patients. The idea is an old one and has been used before with some success. Here is Robert Kruse from Johns Hopkins (who also makes other suggestions):

A simple but potentially very effective tool that can be used in infectious outbreaks is to use the serum of patients who have recovered from the virus to treat patients who contract the virus in the future. Patients with resolved viral infection will develop a polyclonal antibody immune response to different viral antigens of 2019-nCoV. Some of these polyclonal anti-bodies will likely neutralize the virus and prevent new rounds of infection, and the patients with resolved infection should produce 2019-nCoV antibodies in high titer.Patients with resolved cases of 2019-nCoV can simply donate plasma, and then this plasma can be transfused into infected patients. Given that plasma donation is well established, and the transfusion of plasma is also routine medical care, this proposal does not need any new science or medical approvals in order to be put into place. Indeed, the same rationale was used in the treatment of several Ebola patients with convalescent serum during the outbreak in 2014–2015, including two American healthcare workers who became infected.

As the outbreak continues, more patients who survived infection will become available to serve as donors to make antisera for 2019-nCoV, and a sizeable stock of antisera could be developed to serve as a treatment for the sickest patients.

Kruse worries that the exponential growth of the pandemic will be too fast but I think he makes a mistake. The number of recovered patients will far exceed the number of hospitalized patients so the supply of plasma will rise more quickly than the demand.

Convalescent blood therapy was used to treat people during the 1918 flu pandemic and appeared to be useful (see here for references to papers from that time.) A recent meta-analysis of patients treated with blood therapy during the 1918 flu found good results (noting, of course, that data from a hundred years ago wasn’t ideal) :

Patients with Spanish influenza pneumonia who received influenza-convalescent human blood products may have experienced a clinically important reduction in the risk for death. Convalescent human H5N1 plasma could be an effective, timely, and widely available treatment that should be studied in clinical trials.

Blood therapy has also been used periodically since that time to treat Ebola patients, MERS patients, Junin patients and others but under non-ideal conditions where lots of things were being tried at the same time and controls were not ideal. Results have been mostly positive or non-negative, e.g. this study on 84 Ebola patients found few benefits but also small costs. Blood therapy has also been used for animals.

To implement we need a database of recovered patients. The recovered patients then needed to be tested to find those with the most antibodies. It is probably best to use recovered patients from the same location to maximize overlap although the Chinese brought plasma from China to Italy. Most of the dangers from blood transfusion such as passing on another disease are well understood and should be manageable with testing and knowledge of donors. In rare cases such as Dengue it can bad to stimulate the immune system (see discussion here).

Plasma therapy is not difficult and there are firms with expertise in the field including Takeda and Regenernon the latter of whom developed a blood based treatment for Ebola. Thus, CBP seems worthy of consideration.

Hat tip: Monique van Hoek.


Well, sounds like either placebo or mildly helpful. So, are we going to ask for a libertarian solution, like pay people to give blood, or in this socialist age pass a law requiring you to give blood on pain of some penalty?

You can either pay (like in the US) or ask (like in most of Europe). You can’t do the two things because there are studies that show that the donors tend to disappear if you start to pay. A bit like charity, that tends to be crowded out where the State enters.

By the way, this looks a great area for philanthropy. I assume that there is nothing to patent here (I wonder why Takeda and Regeneron are in this), so there is likely little money to do studies and all the abstruse protocols that the FDA requires (mostly useless, read “Bad Pharma” of Ben Goldrake if you want to know more on the subject). With a few million dollars this might become a much better understood and used therapy.

What you say is largely true Massimo. In the USA you can patent this medical method, unlike in Europe, Canada and the UK, where such medical method patents are not legal. However, in the USA you cannot sue doctors or hospitals who infringe a medical method patent (35 USC § 287(c)); though in theory you can sue somebody making a tool that is used in a medical method for contributory infringement (it's usually not easy to do this however). So unless the medical method patent involves a patented tool (in which case it's not really a pure medical method patent anyway), the medical method patent, like here using blood plasma to fight disease, is not going to make the US patent owner any money. What I have seen however is fame: some US medical method patentees like to have recognition from their peers (fame) rather than money (which as I explained above is hard to get in the USA and impossible outside the USA for medical procedure inventions). So in the USA these doctors will patent a procedure they invented so they can market it to their patients as a 'patented' medical cure, not to be confused with 'patent medicine' (sham drugs). So fame = money, in a way.

The FDA indeed is a roadblock to progress; their modus operandi is to delay, delay, delay.

Bye. Posting from the Greek countryside where the living is Covid-19, for now.

Hey Ray, look at this overly broad bulls*t patent lawsuit by defunct Theranos to stop COVID testing. This is the legal equivalent of hoarding and price gouging hand sanitizer.

what about autologous whole blood injection for the patients in an attempt to overcome the severity of the disease

The number of recovered patients is irrelevant. The limiting factor is the available equipment and time to perform the procedure.

Depends on how effective the treatment is verses the costs of doing so.

The military has used FWB without major incident for quite some time, I do not know exactly what the current kit entails, but the whole procedure can be done in the field and under combat conditions. In theory we could dispense with all the filters, bags, etc. and just dump from donor to recipient with just some needles and tubing.

I would recommend against that for a holy host of reasons, but if this sort of thing could free up vents in a triage situation, I would do it in a heartbeat (I have read zilch of the data on this sort of thing's effectiveness).

Even if we need to run it through as FFP, we have a lot of capacity for that these days. We use somewhere in the tens of millions of units of blood products every year; doubling our capacity would be more than enough to give all the MICU cases a bag and forget about it.

Getting the IgG or whatever from donor to recipient is easy, and just dumping it over is relatively low risk (compared to 10%+ mortality rate if other options are not to be had).

That said, I suspect that this will not be enough to actually make a significant dent.

"That said, I suspect that this will not be enough to actually make a significant dent."

Why not ?

Basic math. How are you going to cover 330 million American's from recovered patients. Currently there are 9. Italy has 3K recovered.

We need only plasma for sick patients, not for all the population. It is not a vaccine. And not all sick patients, only those that have worrisome symptoms, perhaps 10% of the patients.

But you may be right nevertheless, there might not be enough recovered patients. The math have been well-explained by Tim Z below, in an exponential model, to have more recovered patients than ill patients, you need the time its take to recover (say 15 days) to be shorter for the time it takes to the number of new case to double (say 5 days). We are far from this. To be more precise, the model says that the ratio of recovered patient over curently ill patients is e^{5/15}-1 = e^{1/3}-1 = 40%.
Interestingly, this is not what we see: in Italy, we have 2941 recovered case over 26,062 active cases, like 11%. But I suspect that's because there are many cases with no or only midl symptoms that are now recovered but has never been tested and are not counted.

Anyway, a question to doctors here. If we want to cure or help
significantly an infected patient in danger, the plasma of how many recovered patients do we need?

You need to think in terms of volume flows.

In order for humoral immunity to work, you need certain titers of antibody to circulate in the body. However, when I give you FFP, the titer of your serum will only be a fraction of that in donor. Your total serum volume is around 2.5 L, if I transfuse 1 unit of FFP that is going be maybe 10% of the volume. You will immediately dilute the concentration down by an order of magnitude. Worse, you are going co consume the antibodies. Some of them will simply be hydrolyzed. Others of them will bind the virus (what we want), but then leave circulation. Even if I doubled your plasma volume (which will kill a lot of these patients faster than Covid), you will have only half the effective concentration of the recovered donor.

We can get rid of some of the issues with volume overload by spinning the stuff down (assuming the antibodies are robust to centrifugation or other volume reduction techiniques), but we still run into the problem that we have no B-cells pumping out fresh antibody in response to antigens.

Which brings up the other problem. Say I flood your body with a bunch of antibodies recovered from recovered sera. Great, we lock down the virus. Except suppose we miss some, well the patient's own clonal expansion, class switching, etc. will all be delayed. After all it is exactly the epitopes we want the patient's immune system to "see" that will be hidden by our foreign antibodies (this is exactly how Rhogam prevents sensitization of Rh- mothers during delivery).

Now we have to not just flood the body with sera, but keep delivering it until the residual viral load is low enough to be taken out by the patient's delayed immune response.

This sort of treatment may increase the length of hospital stay. And that is assuming that we don't get fun cytokine responses to foreign antibodies in the presence of the coronavirus which might possibly make the pulmonary picture worse.

End of the day, we would need a lot of serum. Unless we can drive antibody maturation to very strong affinities we may need truly huge amounts of serum. And we will need to concentrate it.

And then we need to remember that absent continuous stimuli, most of the plasma cells that make antibodies will die off and the donor's serum will become increasingly weaker.

Estimating the number of needed donors is very hard, making IVIG for patients who are humorally deficient needs something like 10K:1 ratios. Antibody binding affinities literally span orders of magnitude. And, lest we forget, the humoral response does not work alone. What is sufficient to defeat the virus in a patient with robust killer T function may be insufficient for an elderly patient.

At the end of the day my back of the envelope calculations are four orders of magnitude wide.

But it could work. If we have enough people and if the most common antibody responses have high enough affinities and if antibody production lasts long enough and if it does not cause other problems ... yeah it could work. If knocking back the virus gets me a free vent with no muss or fuss, then heck yeah. Line em up the donors. Heck if it only takes a small amount serum to maintain life-sustaining respiration? I would dust off my FWB transfusion technique and get cranking if nothing else is an option.

But I place low odds on all the numbers falling on the right side of the scale.

Note that the number of recovered patients is greater than the number of confirmed patients who have recovered because many people have had mild symptoms and recovered. We need to find these people which is why testing needs to expand to not only people have it but to find people who have had it. There is also the possibility, as mentioned in the post, of importing plasma from China and other countries as they recover.

With the caveats that am neither an immunologist nor an ID doc and that I have not searched the literature, why would we assume that the asymptomatic patients have high affinity antibodies?

Antibodies increase in affinity the longer the immune response goes on and the less of the pathogen that is cleared by innate immune system. If a patient's MHCs are better at driving a quicker immune response, we might find that less symptomatic patients did not have time for the adaptive immune response to have a long period of affinity maturation. Likewise, if the initial viral exposure was smaller, there may have been relatively little in the way of epitopes to drive the B-cell response.

For example, my Hep B titers refused to seroconvert twice, inspite of receiving sufficient innoculum to drive a humoral response. Presumably my anigen presentation was deficient for some of the epitopes in the vaccine so it took a long time for me to build up detectable antibodies.

At the end of the day, antibodies are only part of what allows the immune system to clear the virus. As this is not my area of direct expertise, I have only the foggiest of clue about how the end numbers work out. If we end up needing ratios like with IVIG, the concept is going to be of extremely limited utility. Saving just 5,000 lives, under that sort of math, would require around 50,000,000 recovered individuals. I assume the math would be better, but that is within the several orders of magnitude spanned by my back of the envelope spitballs.

Good point. I believe this is the reason bats have a lot of this stuff without getting sick—very strong innate immunity.

Generally different species react differently to these things. For example, domestic dogs and cats don't appear to get COVID-19 despite plenty of exposure.

A virus can have mild or undetectable effects in its host species, but when it jumps to another species be devastating. For example, some fruit bats apparently carry ebola without symptoms.

Alex, it's not that bats have "strong" innate immunity. They have a different immune systems that allows them to be persistently or latently infected with a variety of viruses that other animals' immune systems would react more aggressively towards. They're living viral cultures and reservoirs, supporting the propagation and mutation of of many viral lines. They also shed virus particles readily, which given their life cycle and habits is unfortunate, since it increases the chances of a viral mutation crossing to another species.

To put it simply, if they actually had great immune systems, they'd just kill these viruses off. They don't. Their immune reaction allows them to tolerate being infected.

what are we living in the dark ages?
talking about convalescence plasma? you know we can synthesize any protein in mass quantity in vitro. hello, monoclonal Ab ... give it a few months, it's not like a cure for the common cold or anything.
they also talking about inject mRNA as a vaccine

This means North Korea can have all the cholera toxin they want. Yay.

Very instructive answer. Thanks.

"The number of recovered patients is irrelevant. The limiting factor is the available equipment and time to perform the procedure."

The US only has 9 recovered patients currently and China, by far the largest, only has 68K recovered patients. You aren't going to be able to inoculate the other 1 billion out of that group.

So, no, recovered patients are the bottleneck.

This has not happened yet - The number of recovered patients will far exceed the number of hospitalized patients

Maybe in a few weeks, and certainly at some point, but as of right now, the number of hospitalized people in countries in places like Italy, Spain, and France considerably exceeds the number of confirmed recovered patients.

And the recurrence of covid-19 in patients who have officially 'recovered' is at least a bit troubling.

I believe they used this technique during the Ebola outbreak a few years back, but it has limitations. (My reference "Level 4: Virus Hunters of the CDC".) Since treatment only started after an infection was underway, plasma helped, but it was not a miracle cure. It's not clear it will be a miracle cure for COVID-19, though I imagine it would be useful in many severe cases.

More seriously, plasma in people's bloodstreams is not the same as plasma in a suitable medically approved container. Plasma donation takes about an hour and requires trained staff, careful sterilization and specialized equipment. There are limits on how much more plasma can be made available. In theory, plasma could be extracted from donated blood, but this requires equipment and so on.

Maybe autologous blood can be used preemptively, premorbidly for future outbreaks. Usually a centrifuge separates all the blood products: serum vs plasma, but to use the serum for premorbid polyclonal antibody testing. And then apheresis to reintroduce “convalescent” blood back into the system.


"Kruse worries that the exponential growth of the pandemic will be too fast but I think he makes a mistake. The number of recovered patients will far exceed the number of hospitalized patients"

If the exponential growth model is right, this isn't true. The ratio of sick patients to recovered patients will be greater than 1 if the time it takes to recover (probably a week or 2?) is longer than the time it takes for the virus to double (I've heard 6 days). This is independent of how long the disease has been spreading (exponential models are convenient/horrifying like that). It also assumes all patients recover.

That said, not everyone infected needs treatment, so maybe this works anyway.

And ignore the number of dead patients.

Aside from the issues others have raised (this is not a simple treatment, and right now there aren't nearly enough potential plasma donors), this is mitigating treatment. If it's used on severely ill patient, it won't necessarily prevent the damage the infection causes. A huge part of the tissue damage is not from the viral infection, but from the immune response to the viral infection. We're talking about destruction of lung tissue, fibrosis, kidney damage - i.e., the kind of stuff that doesn't just heal once the infection is cleared.

This kind of treatment would be reserved for critically ill patients, who have been sick for a couple of weeks on average. Don't expect dramatic results on this population.

In the sort of circumstances we are seeing in Lombardy, keeping a patient off a vent is a major win. If even 15% of treated patients can manage without ventilation, we can save a lot of lives.

Sure, but how much plasma is needed? Who do you give it to? Where do you get it? We have very little idea of how robust the antibody response is, and how much plasma it would require to be useful. And frankly, the studies cited aren't that impressive (well, they were at least mostly non-negative!).

We're not anywhere near the point where we are testing to find those people who are recovered, who are free of any other infections, and are well enough to donate plasma. All that's aside from the longstanding issue of getting plasma donations. And as you note, it's probably of minimal benefit to use this technique on the really sick patients, so you'd need to treat patients who are pretty sick, but not yet ventilator sick. I've never seen an experimental treatment that wasn't initially only used on the sickest patients, and in this case I think that might be a waste of time. The alternative is to give the treatment to people who odds are won't actually need it, on the assumption that it will lower the percentage of those who would worsen without it. I'm not saying it's a dead end, but it doesn't excite me as a key to this pandemic.

I do not know the numbers, my very rough calculations give me error bars several orders of magnitude wide.

It is possible that everyone develops strong antibodies and relatively small numbers of donors are needed and that relatively small antibody amounts can improve pulmonary function. I would bet against it, but I suspect we should be making a LOT of odds against bets until one works. Dumbell investing seems prudent here.

Sounds good to me. The meta-study cited shows efficacy. There's anecdotal evidence of success using this technique coming out of China. It is a safe and well tested approach. As for the problem with getting people to donate, I doubt that will be a problem at all. You have to be a real ahole to tell someone you don't want to help others survive the pandemic right after you received medical assistance to beat it. Even if you didn't receive medical assistance to beat it, I can't think of a single person I know who wouldn't donate their blood if they had been found to have survived Covid-19. And, as this article,, points out, even if its use is focused on health care providers it could help prevent the most feared calamity of this pandemic: the collapse of the medical system.

There's also the option of having the patient bitten by a radioactive spider, or flashed with gamma rays.

You may recall that I, the old clinical diagnostics guy, went on and went in about false positive and false negative rates. Well,

"False positives likely account for over 50% of all asymptomatic positive tests"

If that is true it changes a lot of things. It explains why we might have been slow to use that test.

It might explain why we're slow to use tests even now.

It might mean that while propagation is lower then we thought, fatality is much higher.

I really think I said "went on and on". Stupid phone.

Re: Ebola. Hemorrhagic fevers in part do not respond well to blood-serum treatment because their secondary effects (wrecking of internal organ function) don't respond to the serum treatment.

Get to the infected person early enough, before secondary symptoms set in, and it should be effective (such as with Lassa Fever, another hemorrhagic fever).

COVID-19 presents the same challenge. The secondary pneumonia is caused indirectly by the virus itself. Again, get to the infected person early enough, and serum treatment can really work.

But at this point that might mean treating at false positive. That compounds the trade-offs.

A single person can donate plasma twice per 7 days. The volume donated is based on the person's weight but maxes out at 880mL (that's a lot). I don't know how much would need to be transfused to be effective, but a few thousand people donating twice a week at the right price could be useful.

If I recall correctly, that was the story about the sled dog Balto (who has a statur in Central Park) who led the sleds to take serum to rescue kids in Alaska in the 1920s. The weather was too bad for the small planes to fly in, so they had to use sleds.

Diptheria is a toxin based problem though, a virus typically takes more antibodies. Also, neutralizing the toxin does not slow the body's own immune response to the causative pathogen.

But yes, Balto and company were bringing serum to Nome to combat diptheria.

Hell is paved with good intentions

Ideally we would produce required antibodies in the lab. But while several methods of been used to make them in vitro, it's definitely not something we can rely on for COVID-19 treatment anytime soon:

Obviously, if it were easy, it would already be in use for a wide range of diseases.

Will gay men be allowed to donate plasma? Because I don't think we can have medical treatments that are unmoored from social justice.

Anyone who does not match the FDA guidelines, male or female, should not be donating blood for 12 months following various actions. IV drug use, for example.

However, people who have had transplants are to be blocked for five years. But as a troll, that fact should be easy to work up into some form of faux outrage designed to create a response, even among the bored drinking their coffee,

So happy to see my PI, Monique, get a hat tip on Tyler's blog. Dr. Baranova, a virology professor at Mason, made the below video originally in Russian and now in English. Watch for a thorough understanding of how the virus works!

Comments for this post are closed