A central role for amyloid fibrin microclots in long COVID/PASC: origins and therapeutic implications

In many cases, individuals infected with the SARS-CoV-2 virus and suffering from COVID-19 continue (or in some case begin) to display symptoms long after the acute phase. Depending on the demographics of the hosts and variant of SARS-CoV-2, these post-acute sequelae of COVID-19 (PASC [1]) may affect ∼30% of all infected individuals [2,3], are starting to be observed even in children [4,5], and are commonly known as long COVID [6–10]. Many regulatory health bodies still do not recognize this syndrome as a separate disease entity, and refers to it under the broad terminology of 'COVID'. The symptoms of long COVID are multifarious [11,12] and include breathlessness, fatigue, chest pain, myalgia, cognitive dysfunction, innate immune responses coupled to inflammatory cytokine production, and a pro-coagulant state. Our focus is on the latter.

At present there is no established treatment for long COVID (e.g. [13,14], so from a systems point of view it is important to understand which symptoms are 'primary', and which are simply secondary effects of the primary symptoms themselves. This would allow treatment strategies to focus on the primary symptoms and their causes. We here make the case (with evidence) that much of the aetiology of long COVID can be attributed to the formation of aberrant amyloid fibrin microclots, triggered in particular by the SARS-Cov-2 spike protein, and that by inhibiting the transport of erythrocytes to capillaries, and hence O₂ transfer, it is these amyloid microclots that are mainly responsible for the various long COVID symptoms observed. The microclots may also present novel antigens that lead to the production of autoantibodies, that can exacerbate symptoms further. This understanding of the role of such microclots may be expected to lead to an effective strategy for treating long COVID (and probably for other, related conditions such as myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)).

Classically, it was assumed that proteins folded into their conformational states of lowest free energy, because unfolded proteins typically refolded into their original forms [15]. However, the discovery of prion proteins in particular showed that this was not always the case; the more common and native form of the prion protein PrP^C could be converted — with no change in primary sequence — into a thermodynamically more stable, 'rogue' version referred to as PrP^Sc (Figure 1), and this transformation was itself catalyszd by PrP^Sc (e.g. [16,17]). These 'rogue' versions commonly contain a large number of ordered β-sheets in a cross-β architecture [17–23], and are referred to generally as amyloid forms. A great many proteins are now known to be able to adopt such amyloid forms, and over 50 have been implicated in a variety of diseases ('amyloidoses') [24–26]. These kinds of amyloid structures may be stained with fluorogenic dyes [27,28] such as congo red [29], thioflavin T [30–35], which is much more visible than congo red, or a variety of conjugated oligothiophenes [36] commercialized as Amytracker^TM dyes (e.g. [37,38]).

The normal blood clotting cascade is well established (e.g. [39–41] (Figure 2), with the terminal stages of both 'intrinsic' and 'extrinsic' pathways involving the self-organised polymerization of soluble fibrinogen to insoluble fibrin, catalysed by thrombin [42]. Fibrinogen, a cigar-shaped molecule of some 5 × 45 nm, is one of the most abundant proteins in plasma (commonly present at 1.5–3.5 g L⁻¹ [41]).

The action of thrombin on fibrinogen leads to the removal of two small fibrinopeptides (Figure 3), and this initiates the thermodynamically favorable conversion to fibrin, first as small protofibrils and then as long fibrils, typically with a diameter of 50–100 nm. Factor XIII is a transglutaminase which inserts cross-links into the developing clot, and also serves to cross-link α2-antiplasmin (α2-AP) to fibrin. α2-AP and other similar molecules are major inhibitors of plasmin, the enzyme which is mainly responsible for clot proteolysis (fibrinolysis).

The structure of fibrin clots is typically characterized by the diameter of the spaghetti-like fibres and by the size of the residual pores [43]. However, in some cases these spaghetti-like structures with pores are absent. We initially characterized this phenomenon, observed in the electron microscope, as 'dense matted deposits' (e.g. [44–46]). We later discovered that such structures could be induced by highly substoichiometric amounts of bacterial lipopolysaccharide (LPS) (1 molecule LPS per 100,000,000 fibrinogen molecules), and it emerged that they too were, in fact, amyloid in character [26,47–50]. This highly substoichiometric induction (of a thermodynamically favourable process) is very important, as it effectively provides a means for a massive amplification of what may be a tiny amount of trigger material.

The list of molecules that could induce this anomalous clotting, additional to the bacterial LPS that was then our main focus, included iron ions [51–56], oestrogens [46,57], lipoteichoic acid [49,52], and serum amyloid A [58].

While the list of the many other molecules that can plausibly effect this anomalous amyloid-type clotting is unclear, such clots may also be observed in the blood of individuals with inflammatory diseases such as Alzheimer's [37,50,59–61], Parkinson's [37,48], type 2 diabetes [37,38,62–64], and rheumatoid arthritis [65–68]. Similar phenomena have also been observed in the pregnancy disorder pre-eclampsia [69], where there is also strong evidence for a microbial component [70,71].

Although this was a novel finding for 'normal' blood, it had previously been shown that β-sheet structures could be induced in fibrin artificially by mechanical means [72]. There have also been rare reports (see [26,73,74]) of amyloidogenic mutations (alleles) in the fibrinogen Aα chain. Additionally, it has long been known that aged fibrin deposits can bind amyloid dyes in tissue sections [75] and that fibrin-derived peptides can form cross-beta structures [76]. The amyloid form of the prion protein is highly resistant to proteolysis (resistance to proteolysis by proteinase K is used in an assay for PrP^Sc [77–79]), and so this provided a ready explanation for the different nature and persistence of the amyloid fibrin clots [52].

Coagulopathies [84–102], and especially the formation of extensive microclots in vivo, are a hallmark of both COVID [85,103–115] and long COVID [116,117], and we have demonstrated that these microclots too are amyloid in character [108,109,116]. Importantly, the addition of purified, recombinant SARS-CoV-2 S1 spike protein to coagulation-competent normal plasma is sufficient to induce the formation of anomalous clots [118] that adopt amyloid states that are also resistant to fibrinolysis [108]. Note that the observations of the microclots in (platelet-poor) plasma are performed without the addition of exogenous thrombin; they are naturally there in the circulation of patients with both acute and long COVID. The size of these amyloid microclots, that can be observed microscopically and stained e.g. with thioflavin T [108,109,116] (Figure 4; control vs LC plasma), are typically anywhere from 1–200 µm; this means that they can effectively block up, and inhibit blood flow through, all kinds of microcapillaries, thereby strongly lowering the availability of oxygen in tissues. As expected, they consist mainly of fibrin, but also contain many other proteins, including α2-antiplasmin [108] (and even the virus itself [119]). They also have heightened pro-inflammatory activity and elicit fibrin autoantibodies [118] (and maybe others). Elements of at least some spike protein variants of SARS-CoV-2 can also stimulate (in silico) the amyloidogenic aggregation of serum amyloid A [120]. More generally, there is considerable evidence for the induction of amyloid production by amyloidogenic proteins by viruses such as those of the Herpesviridae family, e.g. in Alzheimer's disease [121–125].

A major marker for fibrinolytic activity (Figure 2) is a polypeptide referred to as d-dimer Figure 5. It is a strong prognostic indicator for disease outcome (survival) in acute COVID [85,146–154]. It is, in effect, a composite measure of how much fibrinogen there was, how much was converted to fibrin clots (whether normal or fibrinaloids), and how much these then went on to be lysed. Each of these general steps may of course be regulated independently (i.e. proceed at a 'fast' or slow' rate, as encoded simplistically in Figure 6), so the analysis of d-dimer measurements must take all of these issues into account. Thus high levels of d-dimer must of necessity represent the production of many more clots (and sufficient fibrinolysis to be occurring) but cannot of themselves reflect how many remained after their lysis nor which of them were 'normal' and which amyloid (see Figure 7 for a simplified pathway explanation).This explains why d-dimer levels are significantly raised following infection with many of the earlier variants of SARS-CoV-2, which produce multiple fibrinaloid microclots and Long COVID. However, d-dimer is massively increased in the case of the omicron variant where much clotting takes place but seemingly not to a fibrinaloid form, and thus effective clot fibrinolysis takes place. That there may be strong dependencies on the precise variant is not really surprising (omicron differs from both alpha and delta variants by more than 20 mutations [155], more than enough to vary a protein's activity 1000-fold in typical directed evolution experiments [156]). High d-dimer thus may, but does not have to [157], reflect disease severity given a particular SARS-CoV-2 variant; it depends on the context, and in particular the type of fibrin that can serve as the substrate for its production.

While we do not yet know the details of the mechanical properties of the fibrin amyloid microclots, it is known that amyloid fibrils can typically exhibit unusually high mechanical strength and resistance to deformation (e.g. [174–177]. The stiffness also increases with the thickness of the fibres [178]. This implies strongly that the fibrinaloids are likely to be more prone to 'getting stuck' in capillaries.

The Bradford Hill criteria for causation of a disease Y by an environmental factor X [70,179] represent a useful framework for assessing the role of fibrin amyloid microclots in Long COVID, and are as follows: (1) strength of association between X and Y, (2) consistency of association between X and Y, (3) specificity of association between X and Y, (4) experiments verify the relationship between X and Y, (5) modification of X alters the occurrence of Y, and (6) biologically plausible cause and effect relationship. We think that the evidence is very strongly consistent with these criteria when X is represented by microclots and Y by Long COVID and potentially a variety of other conditions.

While long COVID is a multi-system disorder with multiple symptoms of varying severity, it remains possible that there is in fact a particular major underlying cause (or that a very small number are the main contributors). Our view is as follows: given that amyloid microclots can clog up capillaries and inhibit the transport of O₂ to tissues, this alone can in fact more or less self-evidently serve to explain a great many observations, and in particular the symptoms of both acute and long COVID. These obviously include breathlessness due to low O₂ directly, and thrombotic events such as acute myocardial infarction [180,181], stroke [181–183], etc due to the microclots. The lack of O₂ transport to tissues explains straightforwardly how a nominally respiratory disease also leads to the dysfunction of organs such as the kidney [184,185], PoTS (postural tachycardia syndrome [186]), myalgia in skeletal muscle [187,188], neurological disorders [189], and lactic acidosis [190,191] (the mass of lactate is too low to have been detected in our own COVID untargeted metabolomics experiments [192]), and potentially the benefits of hyperbaric O₂ therapy [193]. It is also worth stressing that amyloid structures themselves tend to be cytotoxic, often via membrane disruption [194–198].

As well as the post-infection diseases referred to above, the emergence of long COVID has brought to the fore its similarities to other even more widely established syndromes such as myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) [1,228], and full details are available in these recent reviews [1,228]. Berg and colleagues have also highlighted the role of coagulopathies in ME/CFS [229]. We note, too, that other 'anomalous' diseases bearing symptoms that overlap with those of ME/CFS include Gulf war syndrome [230], where again we would hazard that the analysis of fibrinaloid microclot formation would have a positive outcome. Thus, while we consider it likely that the phenomena we describe will also broadly be true for ME/CFS, we focus here on PASC/Long COVID.

An important component of severe acute COVID-19 disease accompanying pathological clotting, is virus-induced endotheliopathy resulting in systemic endotheliitis [111,231–235]. Endotheliitis is central to initiating a state of failing normal clotting physiology and it is also known to be significantly linked to coagulopathies, as it activates microthrombotic pathways and initiates endotheliopathy-associated intravascular microthrombi [236].

A consequence of cell death, such as occurs in endotheliopathy, is the release in COVID patients of the normally intracellular iron storage protein ferritin [237,238], which can then release free iron [239]. Iron dysregulation is an accompaniment to a huge number of chronic, inflammatory diseases [200,240,241], and iron dysregulation is a significant accompaniment in COVID [238,242–245]. Indeed, there is evidence that SARS-CoV-2 can release iron from haemoglobin directly [246]. The involvement of iron dysregulation would be consistent with the potentially protective effects of chelating it, e.g. with lactoferrin [247–250] or other chelators [251–254].

The important role that platelets play in acute COVID-19 (Figure 9), has been discussed in the context of disease severity, the development of endotheliopathy and as general drivers of coagulation pathology [232,233,255–257]. Platelets interact with circulating inflammatory molecules, the (damaged) endothelium itself, and also with immune cells, resulting in platelet/cellular and platelet/molecule complexes [258,259]. Platelet complexes are mediated by membrane-membrane interactions via receptor binding. Two of the molecules central in this discussion are fibrinogen and VWF, and platelets form significant complexes with them both [260]. Ultimately, platelet–cell and platelet–coagulation molecule protein complexes form part of platelet activation mechanisms and vascular remodeling and these complexes drive granule secretion, surface glycoprotein expression, and molecular activation platelet hyperactivation pathways [258,259].

Given that there is a definite background removal rate of fibrinaloid microclots, albeit slower than that of normal non-amyloid clots, it is then mostly necessary to ensure that they do not form further. Consequently, we do not see a role for 'clot-busters' as have occasionally been used when acute COVID-19 is accompanied by Acute Respiratory Distress Syndrome [261]. However, addressing only one of the pathologies (either clot pathologies or platelet hyperactivation) will always tend to fail, as was also noted in recent clinical trials where just one of the two therapies was trialed in Acute COVID-19. Lawler and co-workers [262] investigated the use of therapeutic anticoagulation with heparin in noncritically ill patients with COVID-19. (We note too that heparin, like lactoferrin [250], may have antiviral properties [263–266].) Lopes and coworkers [267] also investigated the use of therapeutic versus prophylactic anticoagulation for patients admitted to hospital with COVID-19. Earlier, Viecca and colleagues reported [268] on a single-center, investigator-initiated, proof-of-concept, case control, phase IIb study. The study [268] explored the effects of anti-platelet therapy on arterial oxygenation and on clinical outcomes in patients with severe COVID-19 with hypercoagulability. Outcomes were not significantly positive in any of the trials mentioned.

However, it must be noted that the trials were done on acute COVID-19 patients. This has led us to suggest a multipronged approach and a regime of triple anticoagulation treatment [231], where Long COVID patients might be treated by one month of dual antiplatelet therapy (DAPT) (Clopidogrel 75 mg/Aspirin 75 mg) once a day, as well as a direct oral anticoagulant (DOAC) (Apixiban) 5 mg twice a day, together with a proton pump inhibitor (PPI) (e.g. pantoprazole 40 mg/day for gastric protection). Such a treatment regime showed promise under condition of clinical practice (there were no treatment-free 'controls') [231]. However, especially because of the potential for hypocoagulation (bleeding), it must only be followed under strict and qualified medical guidance in which we recommend the regular assessment of coagulation status before and during treatment. It (as do any other methods [269]) now needs to be studied in a proper randomized controlled trial.

We do note that the use of triple therapy is not new in clinical practice. Its uses predate COVID-19, where it is successfully prescribed in patients where thromboses (which may or may not involve fibrinaloid clots) are particularly worrisome and in various coronary diseases accompanied by atrial fibrillation. Here strong anticoagulant treatments tailored to the needs of the individual patient are recommended [270], commonly involving dual or triple treatments with various anticoagulant agents. The present authors have focused on the 'triple treatment' for anticoagulation that involves an oral anticoagulant plus two drugs designed to decrease platelet activation (usually the P2Y₁₂ inhibitor clopidogrel, plus low-dose aspirin). (In addition, a gastric proton pump inhibitor is also given to reduce the likelihood of gastric bleeding.)

Many pieces of research-level evidence (especially [85,101,106,108,109,116,231,271]) suggest strongly that fibrin amyloid microclots, driven by the presence of the SARS-CoV-2 spike protein, are an inevitable accompaniment to (and a likely cause of) Long COVID. A biologically coherent explanation [272,273] can link such observations with the other observable symptoms of Long COVID, and thus serve to satisfy the logic normally required [179,274,275] to provide a causative explanation.

However, we are still potentially far from translating this kind of understanding into both diagnostics and therapeutics. Some of the elements that are needed both to make the evidence even more robust and to pass regulatory approval more widely include the following: Our fibrin amyloid diagnostic methods, presently semi-quantitative, must be subject to further optimization, made robust and quantitative, and reproduced more widelyWork is required to provide a simple, cheap and widely available point-of-care instrument to effect the necessary measurements of the number, size and nature of fibrinaloid micrclotsLongitudinal studies, and those relating fibrinaloid presence to the severity of Long COVID, will help recognise the role of different fibrinaloids in causing different symptoms; these may vary predictively between SARS-CoV-2 strainsThere would be value in assessing other blood parameters simultaneously (e.g. fibrinogen, d-dimer, von Willebrand Factor) plus suitable inflammatory cytokinesArmed with the above, we should reasonably expect clinicians to be able to obtain ethics for a suitable designed Randomized Controlled Trial of anticoagulant and platelet inhibitor regimes, coupled to the assessment, using viscoelastic assays such as Thromboelastography (TEG or ROTEM or Sonoclot), before and during such treatments of coagulation potential so as to guard against any hypocoagulation and the dangers of bleeding.Such individuals should be followed for a period after the end of treatment to assess the degree of permanenceOf course it is easy to design detailed and expensive trials, accompanied by a great many other measurements of covariates, but the above sets out what we would consider as the minimum necessary to strengthen the claims that the analysis and removal of fibrin amyloid microclots should have major utility in providing substantial benefits to those with Long COVID, and likely in related conditions as well.

Here we have argued, and focus on the fact, that Long COVID is characterized by the presence of persistent fibrin amyloid microclots that might block capillaries and inhibit the transport of O₂ to tissues, entrapping numerous inflammatory molecules, including those that prevent clot breakdown (as we have indeed recently shown) (Figure 10).

In addition to microclot formation, significant platelet dysfunction and a systemic endotheliitis drive systemic cellular hypoxia. These pathologies can explain most, if not all, of the lingering symptoms to which individuals with long COVID refer. We have noted that amyloid microclots, platelet hyperactivation and endothelial dysfunction, might form a suitable set of foci for the clinical treatment of the symptoms of long COVID [231]. Therefore, if fibrinaloid microclots are largely responsible for the symptoms of Long COVID, their removal is to be seen as paramount for relieving these symptoms and allowing the body to repair itself.