Effect of Convalescent Plasma Therapy on Time to Clinical Improvement in Patients With Severe and Life-threatening COVID-19 A Randomized Clinical Trial
Of the many therapies that have been discussed for the treatment of COVID-19, convalescent plasma has gained significant momentum in the medical community. Its appeal has gained traction in society because the science behind it makes sense, and it provides an opportunity for patients that have recovered from COVID to donate their antibodies to those that are ill.
Plasma is the yellowish viscous material that your red blood cells float around in. It carries important proteins and nutrients. One of these groups are antibodies. Antibodies are proteins made by white blood cells that help your body detect and fight disease. Your body has made unique antibodies for each virus you have been infected with throughout your life. You can think of them like plungers floating around, with each plunger having a unique suction cup that will stick to the surface of a specific virus. When it gets stuck your white blood cell grabs a hold of the handle and destroys the virus. These antibodies generally stick around in your blood for 6 months and then your white blood cells stop making them since the threat is gone. They can remake them when necessary, should you encounter the infection again.
The nice thing about antibodies is that they are not specific to individual humans. One human can use other humans’ antibodies. Heck, a human can use other animals’ antibodies. Did you ever wonder what is used as the antidote to snake venom? So, theoretically the antibodies from one human should work for another human. This is the theory behind convalescent plasma. Convalescence simply means time after recovery from an illness. Convalescent plasma has been used historically during small pox outbreaks, influenza, and measles. We also use antibodies to treat certain auto-immune disease, immunodeficiencies, and Guillain-Barre syndrome.
This study was a randomized open-label study performed in Wuhan China during the peak of the COVID-19 outbreak. Open label indicates that everyone involved knew which patients received the treatment. One group was randomized to convalescent plasma plus supportive care and the other just to supportive care. They enrolled patients with severe/ life-threatening disease admitted to hospitals in Wuhan.
Methods.
Randomization – Patients were randomized 1:1 to convalescent plasma or supportive care only. Randomization was stratified based on severe COVID vs. life-threatening COVID on disease severity scale. Stratification is a separate randomized sequenced to ensure one group does not fill with more severe illness due to random chance.
Blinding – The trial was open label indicating that all participants were aware of group allocation. Unblinding can influence providers and adjudicators of outcome to attempt to manipulate the results in favor or against an intervention. Also, supportive care was not protocolized potentially influencing the results.
Trial completion – The trial was stopped early due to inability to enroll patients. Funny, in the middle of March all of the COVID cases in China disappeared. When this happened the Chinese Academy of Medical Sciences stopped funding the study. The authors were left with no other choice to publish the results or bury the study. Fishy? Aren’t you glad you live in America?
Follow-up – There was no reported loss to follow-up. However, the protocol only followed patients for 28 days after enrollment. In the discussion the authors comment that some of these patients were still in “limbo” and it is unknown if they would have recovered if observed longer.
Intention to Treat – They performed a per-protocol analysis. This indicates that patients were analyzed in whatever group they ended up in. If a patient was randomized to the control group, but was transferred to the plasma group, they would be analyzed in the plasma group. This is frowned upon from an evidence-based medicine perspective because the randomization process is not preserved. However, only one patient crossed over, so I’ll give them a pass.
Statistics – Again, the researchers used a change in severity scale of at least two points as their primary endpoint. They calculated their sample size to be 100 patients in each group. This has been the bane of a many COVID studies. They overestimate how many patients are going to make the 2-point jump and therefore underestimate the number needed to enroll. The study was likely underpowered to begin with.
Results
They ended up with 51 patients in the control group and 52 patients in the experimental group. The groups were equally divided up between severe COVID and life-threatening COVID. The randomization process appeared successful. They do not report any significant differences between the two groups. The median age of a patient enrolled was 70. The time from onset of symptoms to enrollment was on average 30 days! That’s quite a long time before patients were eligible to receive the treatment.
They found no significant difference in improvement between the plasma group and the control for the all those enrolled. There was a difference in patients with severe COVID (95% CI 1.07-4.32). Improvement occurred in 21/23 severe COVID patients given plasma. There was no difference in life threatening COVID.
There was no mortality difference.
In severe COVID patients, discharge trended towards significance in the plasma group. This may have been due to faster recovery or issues related to early termination of the protocol for some patients.
There were no significant findings, or even trends towards significant results in the life-threatening COVID group. These were the most ill patients requiring either aggressive oxygenation or mechanical ventilation
Patients in the convalescent plasma group were more likely to test negative for COVID 24, 48, and 72 hours after receiving plasma.
Meaning
I had high hopes for convalescent plasma. The science behind it makes sense. There are anecdotes of success in previous pandemics. You can harvest it and store it for a significant amount of time in hospital blood banks for future use. Unfortunately, this trial was a bit of a mess and not much can be gleaned from it. Even though there were some areas that showed promise, it certainly could not be considered a robust therapy based on this evidence.
I often talk about maintaining prognosis throughout trials. In the beginning, randomization is used to equalize prognosis between the groups. Blinding is used during the trial. Intention to treat and completion of follow-up is used at the end. In this case they lose the equalizing power of randomization by the absence of blinding and the issues they had with the trial being stopped early. The protocol of a study should isolate the treatment, so no other factors could influence the outcome. In this study there were many other interventions used on different patients that could have played an important role in their care. Also, given that they could not follow several patients beyond the termination of the trial left several without outcomes limiting the interpretation.
Unfortunately, the study was underpowered. There were very few events overall. Particularly of hard endpoints like mechanical ventilator days, deaths, and discharge from the hospital. Several of the endpoints trended towards significance. Improvement in disease status appeared to be trending as well as reduced days to discharge. However, most of this was influenced by patients with severe COVID. Patients with life threatening COVID did not appear to benefit at all from the intervention.
Life threatening COVID continues to be a tough problem to solve. Death rates are very high, most interventions impact the end results very little, and studies investigating it remain underpowered. This is similar to studies in ARDS from other viral and bacterial illness. With death rates in the 60% range it is very difficult to power a study to see a significant difference in an intervention.
Finally, there was a significant delay from symptom onset to actually administration of the treatment. Most did not receive it until 30 days after symptom onset. This could have seriously dampened any treatment effect. It would’ve been very nice to see some result from patients a few days after symptom onset. Citizens, blood curators, and hospitals are going to great lengths to obtain and store this liquid gold. It would’ve been nice to see if all this effort has any impact in treating this disease.
This study certainly highlights the infrastructure weakness in medical research that exists. In the information age, it’s really sad for the data from these patients to be wasted. The entire world is connected. A system should exist that allows collaboration across multiple health systems in different countries. The protocol was not complicated. All that needed to be accomplished was the randomization and then the administration of the plasma. After that, there was not much to do other than collect the outcomes. We could’ve debated the validity of the results afterwards. At least, if they had the patients, there would be some idea if it worked.
Hospitals in the US have been using plasma since April to treat certain COVID patients. If each hospital randomized even a handful of patients and uploaded the outcomes onto a database this trial could be completed in a quarter of the time with 5x the patients. But collaboration remains fragmented, and data is not transferred. It’s a shame that this trial was started, stopped early, and we are left with no new information. The medical research community could use some innovation.