What causes Long Covid? An update in light of new data.

The spike protein of Covid-19 binding to a cell. Getty Images, Juan Gaertner.

If we knew what causes Long Covid, we could treat it better. There are several realistic possibilities under consideration. I think about the possible mechanisms behind Long Covid in three categories, each with its own analogy:

1. Scorched Earth. The acute infection causes damage and inflammation, with effects lingering well after the virus has been cleared.

2. Bee Sting. While the virus itself has been cleared, left behind are some remnants which cause symptoms either directly or due to the immune system’s reaction. Like a stinger, these remnants can continue to cause symptoms until removed or broken down.

3. Sleeper Cell. Tiny amounts of live virus hide in parts of the body, occasionally making copies of itself—enough to cause symptoms, either directly or by eliciting them from the immune system—The viral levels are too low (and sequestered) to be picked up by conventional assays like PCR tests of the nose we use for diagnosis.

Among these, I don’t have a single preferred theory. It’s possible that these explanations are not mutually exclusive. Further, Long Covid is likely more than one distinct entity, and that as-yet defined subtypes may have distinct causes. That said, I think the third theory is the one with the steepest climb, in terms of evidence. That doesn’t mean it’s wrong, though; scientific understandings often take time to become clear. For example, it took scientists decades to discover the connection between Epstein-Barr Virus (EBV) infection and Multiple Sclerosis (albeit there, the link isn’t related to the chronic infection aspect of EBV, but rather the initial exposure).

---

Quick aside: There are some who strongly believe in the Sleeper Cell/chronic live virus theory. While my initial perspective is that this is less likely (primarily because coronaviruses have not been known to cause chronic infections), I believe that all manners of research being done on Covid-19 will unveil much about immunobiology that we fail to grasp currently. So, I remain open-minded on this, despite my “priors.”

---

New data shows that the Covid-19 remnants can hang around in small amounts.

A new study in Lancet Infectious Diseases linking Long Covid to evidence of persistent viral remnants has made quite a stir. It’s fascinating. Some have interpreted the paper to support the Bee Sting and Sleeper Cell theories of Long Covid. Having spent a lot of time with this paper (and discussing it with some experts), I actually see it lending support to both the Scorched Earth and Bee Sting theories. (It does not really address the Sleeper Cell/chronic live virus theory.)

What they did:

Researchers in China studied a group of mostly vaccinated patients who had recently experienced a mild Covid-19 illness (late 2022) and who subsequently had a biopsy (looking for cancer and the like) or an endoscopy approximately 1, 2, or 4 months later. Whatever part of the body the procedure was done in, the clinicians took samples and tested them for evidence of persistent SARS-CoV-2 RNA (the genetic material of the virus that causes Covid-19) and persistent viral proteins. Patients also filled out surveys about their symptoms.

What they found:      

1. At 1 month post-infection, 30% of the biopsies/solid tissue samples had small amounts of viral RNA. That fell to 27% at 2 months, and 11% at 4 months.

2. At four months post-infection, 34% of the patients reported one or more Long Covid symptoms.

3. Patients who had ≥1 symptom of Long Covid were more likely to have had viral RNA detected from their biopsies/samples.

4. Patients with higher quantities of viral RNA were more likely to have ≥1 Long Covid symptom.

Some caveats:

1. These patients had high rates of cancer and other serious medical problems. That makes the observations here most applicable to similar patient populations.

2. Long Covid symptoms are not synonymous with Long Covid itself. For example, fatigue is a very common symptom among cancer patients. It’s not possible to say which patients had symptoms due to Long Covid and which were reporting symptoms of their cancers or other conditions. That said, the correlations between Long Covid symptoms and persistent genetic material are why we’re here.

3. Each patient was sampled once. That is, patients were not sampled serially at 1, 2, and 4 months but rather at 1, 2, or 4 months.

Insights and analyses (FAQs).

Let’s go through some questions that I had as I read through the manuscript.

1. Does finding viral RNA 1, 2, or 4 months after infection mean that persistent live replicating virus is present? No. Prior studies have indicated that this is unlikely. So, any detected RNA molecules are likely remnants of the virus that have been integrated into parts of our own cells, or that are being expressly kept around by our immune systems (more on that below).

2. Why would our body keep viral RNA around? It’s possible that our immune systems intentionally keep viral RNA around as an evolutionary adaptation to enhance and prolong memory response by extending how long the immune system responds to the pathogen. For example, “depots” of viral components have been shown to help keep the immune system on alert for weeks and months after other viral infections. This seems to be a part of building immune memory.

3. Does the persistent viral RNA do anything? A key question is whether the viral RNA detected in the tissue samples actually gets converted into viral proteins (like the spike protein or other components of intact and infectious viral particles). To get at that, the study stained for the presence of the proteins encoded by the viral RNA and, indeed, found evidence of them in some tissues. This brings up another question: Were the detected viral proteins remnants from the acute phase of the illness, or were they newly synthesized from the persistent RNA? Best I can tell, it’s more likely the latter, though it may be some of both. In the appendix of the study, the researchers showed whether RNA was detected in the individual samples in which the proteins were found. The agreement was quite good. When scientists detected viral proteins, they also found viral RNA; When they did not find viral proteins, RNA was also not found.

4. Did patients found to have persistent viral RNA always have Long Covid symptoms, and vice versa? No. Not every patient found to have persistent RNA (or evidence of viral proteins 1, 2, or 4 months after the acute infection) had Long Covid symptoms. Moreover, the association between RNA detection and Long Covid symptoms went down over time. This is important, although, again, the tests were not done serially, which is a limitation. The presence of viral RNA at 1 month massively increased the odds of Long Covid symptoms at 4 months; but that relationship was weaker at 2 months, and weaker still (and no longer statistically significant) by 4 months. It follows that some patients with Long Covid symptoms did not have evidence of persistent RNA and some patients with persistent RNA did not have symptoms of Long Covid.

5. How did immunocompromised patients do? Oddly, immunocompromised patients were less likely to report Long Covid symptoms, albeit they were the only ones to have persistent viral RNA found in their blood cells (rather than solid tissues). This suggests that immunocompromised patients had trouble clearing the active infection (not news), and their immune systems may not be adequately responding to the infection. Remember: symptoms are often a reflection of the immune system responding to invaders. So, it may be that compromised immune systems have trouble clearing the virus and have trouble making good memory of it. This would explain why immunocompromised individuals have both longer acute infections and are at high risk of severe outcomes in subsequent infections, despite lower odds of Long Covid symptoms (in this study population, anyway).

Tying some of this together. A grand unified theory for Long Covid?

If either the Scorched Earth or Bee Sting theories explains Long Covid, it follows that the worse the acute infection is (i.e., the higher the viral loads are) the more likely Long Covid is downstream. That correlation has been established (albeit mild acute infections can become Long Covid as well and there is some debate about this). The findings of this new study lend credence to the notion that a high viral load early on predicts Long Covid risk, because the presence of persistent viral RNA 1 month out from infection (but less so at 1 month, and not at 4 months) was found to correlate with Long Covid symptoms, as were higher viral loads.

This left me wondering: Could it be that Long Covid is a pathologic extreme of something beneficial that is inherent to our amazingly intricate immune systems? Such a nuance is common in medicine and immunology. The right dose of aspirin decreases heart attack mortality, but too much can kill you. Among bone marrow recipients, some degree of “graft versus host disease” is actually good for knocking out any remaining blood cancer cells, while too much of this can kill the patient. Indeed, the very concept of an autoimmune disease is “too much of a good thing.” This seemingly counterintuitive explanation for Long Covid—that stabilized viral RNA hiding in some tissues gets converted into viral proteins so that our immune systems will stay primed and thus be better prepared for future infections—points towards a complicated interplay between viral persistence, symptoms, and immune memory.

Indeed, what if our bodies intentionally keep some small amount of Covid-19’s genetic material around for the express purpose of converting it into proteins that our immune systems will therefore continue to respond to, making sure our immune memory is fresh and that we don’t “forget” the recently encountered pathogen too soon. Then, after some time—perhaps just enough time for that bug to have worked its way through the population of our ancestors—the immune system can afford to lose interest and stop keeping those viral remnants around. Indeed, without help from the host (i.e., us), the half-life of viral RNA is quite short. But with help from the host, viral persistence is possible, and may help keep the immune system on guard. This has been suggested in other diseases, like measles, where RNA persistence has been shown both to cause harmful very sequelae and contribute to immune memory that can last a lifetime.

An expert opinion and why getting it right matters.

I was recently discussing some of this with Dr. Ziyad Al-Aly, a Long Covid researcher. Here’s what he said (this was prior to the new paper discussed here, but it still stands):

“The viral persistence hypothesis is one of many plausible mechanistic hypotheses of Long Covid. While potentially plausible, it remains a hypothesis and we need more empirical evidence to assess it. To date, no one has isolated a live replicating virus in the post-acute phase of SARS-CoV-2 infection. There is evidence of persistent viral antigens and indirect evidence of persistently activated T cells months or years after the infection. Both findings hint at the possibility of viral persistence well beyond the acute phase of the infection. There are a few randomized controlled trials that are evaluating effectiveness of antivirals (e.g., Paxlovid) in ameliorating Long Covid symptoms. The results may help evaluate the plausibility of viral persistence as a driving mechanism behind the symptomatology of Long Covid.

Overall, I really think that Long Covid is too complex and its manifestations are too varied to be simplistically explained by one mechanistic pathway. It is likely that there are many mechanistic pathways at play. Identifying those will be important to help us understand the genesis of Long Covid and identify therapeutic interventions to treat it.” —Dr. Ziyad Al-Aly.

In other emails, we discussed why it’s important that we not rule anything in or out too soon. It would be bad if patients took treatments that don’t work; for example, so far one randomized trial looking at the antiviral Paxlovid for Long Covid apparently failed. Worse would be if something we do “on theory” backfires in practice, conferring more harm than good.

If this work convinces me of anything, it is that we will not solve Long Covid by ruling anything in (or out) prematurely, relying on inadequate data, or by guessing. This is all exceedingly complicated. However, what we learn may have a large impact not just towards helping patients with Long Covid, but also for those with a variety of post-infection syndromes.

Thanks to Dr. Ziyad Al-Aly, Dr. Benjamin Chen (my graduate advisor back when I studied HIV), and Dr. Michael Mina for fruitful discussions.

Afterword: There’s a lot here and I admit I remain heavily in the learning phase here. I’m open to your feedback and your questions!