Consistent with well-established biochemical properties of coronaviruses, sialylated glycan attachments between SARS-CoV-2 spike protein (SP) and host cells are key to the virus’s pathology. SARS-CoV-2 SP attaches to and aggregates red blood cells (RBCs), as shown in many pre-clinical and clinical studies, causing pulmonary and extrapulmonary microthrombi and hypoxia in severe COVID-19 patients. SARS-CoV-2 SP attachments to the heavily sialylated surfaces of platelets (which, like RBCs, have no ACE2) and endothelial cells (having minimal ACE2) compound this vascular damage. Notably, experimentally induced RBC aggregation in vivo causes the same key morbidities as for severe COVID-19, including microvascular occlusion, blood clots, hypoxia and myocarditis.
1. Introduction
The Molecular Composition of Glycans on SARS-CoV-2 SP and the RBC
The arrangement and chemical composition of the SARS-CoV-2 SP glycans have been determined, with those at its 22 N-glycosylation sites having a total of nine sialic acid (SA) terminal residues
[1][2][3][4][5][6][7][8][9][10][11] and its four O-glycans having a total of three SA terminal residues
[11]. This provides a basis for exploring these viral SP attachments to host cells, notably red blood cells (RBCs), platelets, leukocytes and endothelial cells
[1]. RBCs and platelets have densely distributed sialoglycoproteins but no ACE2 receptors on their surfaces
[12][13]; the same holds for leukocytes and most other blood cells
[14][15][16]. Endothelial cells likewise have a heavily sialylated surface coating (glycocalyx), with about 28,000 SA-tipped CD147 receptors but only about 175 ACE2 receptors per cell
[17][18].
Of particular interest are attachments of SARS-CoV-2 SP to the RBC, the latter coated with one million SA-tipped glycophorin A (GPA) molecules and a total of 35 million SA monosaccharides per cell
[19][20][21]. The heavily sialylated GPA strands are spaced about 14 nm apart on the RBC surface and extend out 5 nm
[19]. Band 3 protein is another molecule on the RBC surface, with 1.2 million copies per RBC, which extends > 10 nm from the RBC surface
[19][22] and is glycosylated by poly-N-acetyllactosamine, a sialylated branched-chain glycan
[23][24][25][26]. GPA and poly-N-acetyllactosamine, the two most abundant glycans on the RBC membrane
[25], have been found to mediate hemagglutination by various bacterial and viral pathogens
[26][27][28][29].
Hemagglutination as caused by these pathogen–glycan attachments is of particular interest in view of a primal defense mounted by RBCs along with platelets against pathogens having SA terminal moieties by attaching to them and delivering them to leukocytes or conveying them to macrophages in the liver and spleen for phagocytosis
[20][30][31][32][33][34][35][36].
A clear experimental demonstration of binding between SARS-CoV-2 SP and sialylated glycans on host cells was provided using NMR spectroscopy
[37]. It was found, in particular, that a site on the SP N-terminal domain (NTD) binds to α2,3 and α2,6 sialyl N-acetyllactosamine, which are components or variants thereof of the sialylated poly-N-acetyllactosamine glycans of the band 3 strands extending from the RBC surface. Intriguingly, this SP-to-glycan binding was found to be much more pronounced for α2,3 than for α2,6 SA-linked N-acetyllactosamine
[37], while α2,3 vs. α2,6-linked SA is likewise much more prevalent in sialylated poly-N-acetyllactosamine of adult (vs. fetal) RBCs
[24].
Possibilities for binding are indicated as well between SARS-CoV-2 SP and/or glycans at its glycosylation sites and GPA on the RBC surface, with GPA, as noted, having no known physiological role other than this type of immune adherence. The positive electrostatic potential of SARS-CoV-2 SP
[38] supports its binding to the negatively charged, densely distributed SA on the RBC surface, most on its million GPA strands
[39][40]. SA in its predominant human form, Neu5Ac, is the most common terminal residue of GPA
[19][22][41]. For the N- and O-glycans on SARS-CoV-2 SP, the most common terminal residues are galactose (Gal), with 27 total, and Neu5Ac (SA), with 12 total
[9][10][11][42]. Through binding configurations proposed by Varki and Schnaar (2017)
[43] and others
[37][44][45][46], multivalent bonds can form via α2–3 and α2–6 linkages from Neu5Ac on GPA to Gal on glycans populating SARS-CoV-2 SP glycosylation sites.
2. In Vitro, In Vivo and Clinical Studies Demonstrate Induction of RBC Aggregation by SARS-CoV-2 SP
Many in vitro, in vivo and clinical studies demonstrate that SARS-CoV-2 SP attaches to RBCs and induces RBC aggregation. Boschi et al. (2022) found that SARS-CoV-2 SP from each of the Wuhan, Alpha, Delta and Omicron strains induced RBC clumping (hemagglutination) when mixed with human RBCs in phosphate-buffered saline (PBS)
[38]. To explore whether bridging of adjacent RBCs by SARS-CoV-2 SP via glycan bonds might be the cause of this observed hemagglutination, an agent with indicated high-affinity binding to multiple SARS-CoV-2 SP glycan-binding sites
[47], the macrocyclic lactone ivermectin (IVM), was added to the mix of SP and RBCs both before and after hemagglutination formed. IVM blocked the formation of hemagglutination when added to the initial mix and reversed hemagglutination over the course of 30 min when added after it formed
[38].
The same SP-induced RBC clumping effect as noted above was demonstrated in zebrafish embryos, which have blood cell glycosylation patterns
[48] and capillary diameters
[49] similar to those of humans. When SARS-CoV-2 SP was microinjected into the common cardinal vein of a zebrafish embryo at a concentration similar to that obtained in critically ill COVID-19 patients, it caused the formation of small RBC clumps and an associated reduction in blood flow velocity within 3–5 min after injection, accompanied by thrombosis in capillaries, arteries and veins
[50]. When SP was coinjected with a mixture of heparan sulfate and heparin (molecular mass of each ≤ 30 kDa), however, with both of these glycosaminoglycans having strong binding affinity to SARS-CoV-2 SP
[50][51][52], the extent of thrombosis was markedly reduced
[50].
SARS-CoV-2 SP attachments to RBCs were demonstrated directly by Lam et al. (2021) through immunofluorescence analysis of RBCs from the blood of nine hospitalized COVID-19 patients
[53]. For these patients at hospital admission, the mean percentage of RBCs having SARS-CoV-2 SP traces was 41%. This finding suggests that concentrations of SARS-CoV-2 SP in blood as typically reported in other studies using plasma or serum may significantly understate actual values due to high-affinity binding to RBCs, which are removed from plasma and serum. SARS-CoV-2 SP and pseudovirus were each found to bind to nanoparticle arrays bearing SA derivatives
[7] and to SA-tipped CD147 receptors, likewise detected using nanoarrays
[54]. Nanoarray methods are required to detect SARS-CoV-2 SP glycan attachments, because these methods allow bindings to form multivalently, whereas univalent bindings are weak
[1] and not detectable by microarray methods
[55].
The presence of SARS-CoV-2 SP in the blood of COVID-19 patients and its induction of hemagglutination in vitro and in vivo would suggest that the same would occur clinically, which is indeed the case. In three publications that used scanning electron microscopy to examine blood from the cubital vein blood of patients with mild-to-moderate cases of COVID-19, all hospitalized but none requiring intensive care, a team of investigators observed blood cell clumping and other anomalies
[56][57][58]. The first study found stacked RBC aggregates (rouleaux) ranging in size from 3–12 cells.
Light microscopy examination of smears from the blood of 20 hospitalized COVID-19 patients with anemia detected large, stacked RBC clumps (rouleaux), in 85% of those patients
[59]. Another study, which examined the sublingual microcirculation of 38 COVID-19 patients in intensive care using video microscopy, found that the mean number of RBC microaggregates detected in these patients was 15 times the mean number for 33 healthy volunteers
[60]. These RBC microaggregates were found in two-thirds of the COVID-19 patients vs. two of the 33 healthy volunteers.
Paralleling these studies that document RBC aggregation in severe COVID-19 are many that report microvascular occlusion. Postmortem examinations of hundreds of patients who died from COVID-19 in many studies consistently found microthrombi in the pulmonary microvasculature in most patients
[61][62][63][64][65][66][67][68][69][70][71][72]. Microthrombi in alveolar capillaries were nine times as prevalent in postmortem COVID-19 patients compared to influenza patients
[64]. RBC clumping and microthrombi in the lungs have been regarded as likely causes of hypoxemia in severe COVID-19 patients
[56][58][73][74], which in turn is closely associated with mortal outcomes
[75].
3. Glycan Bindings from SARS-CoV-2 SP to Platelets and Endothelial Cells Cause Endothelial Damage, Inflammation and Coagulation
Attachments of SARS-CoV-2 SP to the heavily sialylated
[12][13][18] surfaces of platelets and endothelial cells cause endothelial damage, platelet activation and associated coagulation which, as with the attachments to RBCs, contribute to the severe morbidities of COVID-19. Platelets, having no ACE2 receptors, like RBCs
[14][15], act with RBCs in a role that was termed “immune adherence”
[33], attaching to and clearing pathogens
[35][36], and are found enmeshed with RBCs in blood cell clumps in COVID-19 patients
[57]. The degree of sialyation of the endothelial cell surface is exemplified by the 28,000 SA-tipped CD147 receptors vs. the 175 ACE2 receptors per endothelial cell
[17]. For glomerular endothelial cells from a conditionally immortalized human cell line, the enzyme neuraminidase, which hydrolyzes SA, removed more than 50% of the cells’ surface coating (glycocalyx)
[18]. The endothelial cell thus provides a prime target for the SARS-CoV-2 virus, and indeed, both whole virus and viral SP have been found on endothelial cells in clinical and in vivo COVID-19 infections
[64][71][76][77][78][79][80][81].
These SARS-CoV-2 viral or SP attachments to the endothelium can be perilous to the human host, with trillions of RBCs each flowing once per minute through the lungs and then the extrapulmonary vasculature
[82] and with the cross-sectional diameter of most capillaries so small that RBCs distort their shape to squeeze through
[83]. Thus, SARS-CoV-2 virus particles or SP attached to endothelial cells or RBCs could create resistance to blood flow or even potentially rip off a piece of an endothelial cell or the entire cell
[1]. Indeed, one study found that serum levels of circulating endothelial cells (CECs) in mild-to-moderate COVID-19 patients were up to 100 times the levels for matched controls. The study also found that each of these CECs from the COVID-19 patients typically had several holes in their membranes approximately the size of the SARS-CoV-2 viral capsid (the viral envelope)
[56].
4. Experimentally Induced RBC Clumping In Vivo: Parallels to Severe COVID-19
Studies dating back to the 1940s in dogs, rabbits, mice, hamsters and other animals closely examined the effects of IV injection of high-molecular-weight dextran (HMWD), generally of molecular weight (MW, loosely equivalent to molecular mass) ≥ 100 kDa or other blood cell-agglutinating agents. In several studies, blood cell aggregation was induced within minutes to hours after IV injection of HMWD
[84][85][86][87][88][89], with molecular bridging of RBCs by HMWD molecules being a hypothesized mechanism for this effect
[90][91][92][93]. After HMWD injection in vivo, small clumps of RBCs formed and then enlarged into longer stacked clumps (rouleaux) and, in some cases, into vast trees with branches of hundreds of stacked RBCs
[85][86][94].
In vitro, the addition of HMWD to blood likewise induced RBC aggregation
[95][96] and did so as well when added to RBCs in PBS
[97][98]. The same RBC disaggregating effect of LMWD was observed in vitro
[99], possibly caused by competitive binding to RBCs that limited bridging between adjacent RBCs by larger molecules.
Even in healthy humans or animals, RBC clumps can transiently form under conditions of slow blood flow, e.g., in deep veins of the lower limbs, but they typically disaggregate as they move into regions of faster blood flow
[100][101][102][103][104][105][106][107][108][109][110] and are rarely problematical in healthy subjects
[89][111][112]. Yet under pathological conditions in diseases such as diabetes, malignant hypertension and malaria
[103][112][113][114][115], these RBC aggregates can persist and grow via a positive feedback loop whereby the clumps cause decreased blood flow velocity with a concomitant reduction in shear forces that in turn causes further aggregation
[100][102][103][104][105][106][107][109][112].
The capability of LMWD to rapidly reverse RBC aggregation and associated microvascular occlusion caused by injection of HMWD, as noted above, distinguishes blood clumping, e.g., as induced by HMWD, from clotting, in which blood cell clumps harden into fibrin-enmeshed clots via the coagulation cascade. Indeed, several mammalian diseases are associated with increased levels of RBC aggregation and microvascular occlusion which do not typically cause blood clotting, although risks of this complication are increased
[112][113]. Blood cell clumping and clotting are not completely unrelated phenomena, however, given the potential of RBC aggregation to trigger deep vein thrombosis
[116][117] and the role of fibrinogen, an essential promoter of blood clotting, in blood cell clumping as well
[100][118][119][120].
4.1. Induced RBC Aggregation Causes Microvascular Occlusion, Hypoxia, Blood Clots, and Redistribution of Blood Flow from Smaller to Larger Blood Vessels
When HMWD or other agglutinating agents were injected into animals at sufficient concentrations to overwhelm the host’s ability to safely sequester the RBC aggregates formed, these clumps caused microvascular occlusion as detected in a variety of host tissues
[113], including the myocardium
[94][121], muscle
[122] and abdominal cavity
[94] of rats; the conjunctival vessels of dogs, cats and rabbits
[88][123]; the cheek pouch of hamsters
[89][111]; and the kidney, liver, ear chamber, bone marrow and heart tissue of rabbits, including the myocardium and pericardium
[85][86][87][124][125]. In the myocardium of rabbits and rats, the degree of myocardial tissue damage was correlated with the observed degree of intravascular aggregation of blood cells
[85][87][94], with hypoxia resulting from vascular occlusion proposed to be the cause of tissue damage
[85][87].
Associated with the microvascular occlusion that it triggered, experimentally induced RBC aggregation caused decreased velocity of blood flow
[84][86][87][88][89][107][113][121], increased blood viscosity
[84][113][123][126], increased incidence of blood clotting
[85][103][113] and decreased oxygen tension in arteries, veins and tissues, with accompanying hypoxic damage to body organs
[85][87][113][127][128]. Another effect caused by induced blood cell clumping as observed in the conjunctiva of cats, dogs and rabbits and bone marrow of rabbits was a reduction in blood flow in the capillaries and other small vessels having cross-sectional diameters of about 10 μm or smaller
[88][124], indicative of a shift of blood flow into the larger vessels.
4.2. Corresponding Morbidities in Severe COVID-19
As considered above, SARS-CoV-2 SP, like HMWD dextran, induces RBC aggregation, and the same morbidities caused by experimentally induced RBC aggregation have been commonly observed for cases of severe COVID-19. These morbidities of severe COVID-19 include microvascular occlusion in the lungs and other organ systems
[61][62][63][64][65][66][67][68][69][70][71][72][129][130], hypoxia
[73][131], arterial and venous thromboembolisms
[63][69][71][72][132][133][134][135], disseminated intravascular coagulation
[69][132][133][134][135][136][137] and multiorgan damage associated with these vascular aberrations and hypoxia
[61][137][138].
4.3. Redistribution of Blood Flow from Smaller to Larger Blood Microvessels in COVID-19 Patients
Another effect of experimentally induced RBC aggregation, the redistribution of blood flow from microvessels to blood vessels of larger cross-sectional diameter, as described above, is also paralleled in COVID-19 is. Osiaevi et al. (2023) compared videomicroscopic imaging of the sublingual microvasculature of 16 critically ill COVID-19 patients, 17 patients with long COVID and 15 healthy controls
[139]. The density of functional capillaries (having flowing RBC content ≥ 50%) with cross-sectional diameter 4–10 μm was sharply reduced for active COVID-19 patients vs. controls, with values for long COVID patients roughly halfway between those for active COVID-19 patients and healthy controls. The study investigators concluded from these and other measures of microvascular health that the long COVID patients had significant microvasculature impairment, lasting even 18 months after infection for some
[139].
Rovas et al. (2021) reported similar sharp reductions in densities of functional capillaries at the lower end of the 4–25 μm cross-sectional diameter range in the sublingual microvasculature of COVID-19 patients vs. healthy controls
[138]. The extent of reduction in density of functional capillaries of diameter 4–6 μm in the COVID-19 patients correlated with their oxygenation index (PaO
2/FiO
2) and with an index of multiorgan failure and associated mortality risk. Rovas et al. concluded from these correlations that the observed reduction in sublingual small capillary density was another manifestation of the pathological clogging of capillaries as also observed in pulmonary microthrombi at autopsies of COVID-19 patients.
5. Major Risk Factors of Age, Diabetes and Obesity for COVID-19 Severity Correlate with Increased Propensity to RBC Aggregation
The most significant risk factor for severe COVID-19 is age, with several studies showing a multifold increased risk of fatal outcomes with older age
[131][140][141]. One multivariate analysis of 17 million subjects in the UK reported a sixfold increased mortality for ages 70 through 79 vs. 50 through 59 years
[142]. A meta-analysis of 612,000 subjects in several countries conducted in 2020 found a mortality rate of 22.8% for ages 70–79 years vs. 0.3% for ages ≤ 29 years
[143].
This multifold increase in COVID-19 mortality with older age aligns with a much greater extent of microvascular occlusion in older vs. younger healthy subjects, linked to both a significantly greater propensity to RBC aggregation and slower blood flow with increased age. Microscopic examinations of the bulbar conjunctiva of healthy subjects found that 30% of those of ages 56–75 years had aggregation in the smaller venules and capillaries, as compared with a 3% rate of such aggregation of those of ages 16–35 years
[144]. This tenfold increased rate of microvascular occlusion in the older subjects corresponds to much greater RBC aggregation and slower blood flow with increased age. One study that measured RBC aggregability by multiple detection methods found a statistically significant increase in this value in the blood of middle-aged versus young adults
[145].
As noted, RBC aggregate formation in vivo depends not only on aggregability under static conditions but also on the degree of shear forces that promote disaggregation, as associated with velocity of blood flow
[100][104][105][106]. It is, thus, noteworthy that blood flow is slower with increased age
[146][147][148][149][150][151][152]. Mean velocity of capillary flow under fingernail and toenails for subjects of mean age 63 years was half of that for subjects of mean age 26 years
[146]. Older subjects had 23%
[147] and 40%
[148] diminished flow velocities vs. younger subjects for capillary flow in other tissues. Arterial blood flow velocities were 26–27% lower for older vs. younger subjects in two studies
[150][151].
6. SARS-CoV-2 SP Unattached to Whole Virus Induces Microvascular Occlusion In Vivo
6.1. Myocardial Damage as a Signal of Microvascular Occlusion
A clinical window into morbidities associated with RBC aggregation is provided by the myocardium—the heart muscle—which is among the tissues most susceptible to the damaging effects of experimentally induced RBC aggregation and ensuing microvascular occlusion. Several studies found that injection of HMWD (high-MW dextran) caused myocardial damage
[85][87][113][153] and/or electrocardiogram (ECG) changes
[94][113][126][153] characteristic of myocarditis. In one study, 40 min after HMWD injection, ECG abnormalities were apparent, and HMWD induced lasting myocardial damage
[153]. Both the degree of myocardial damage
[85][87] and of ECG abnormalities
[94] correlated with the extent of microvascular occlusion. Clinically, for hospitalized patients with coronary heart disease, the number of microthrombi per field of observation in the bulbar conjunctival microcirculation was found to be correlated with both the extent of ECG and symptomatic abnormalities
[94].
6.2. Myocardial Damage Experimentally Induced by SARS-CoV-2 SP in the Absence of Whole Virus
Induction of myocarditis by SARS-CoV-2 SP in the absence of whole virus was evidenced in two rodent studies by IV injection of BNT162b2, the Pfizer-BioNTech mRNA vaccine, an experimental system in which SP is generated by host cells, distinct from intramuscular (IM) injection used for clinically administered COVID-19 vaccinations. Clinical cases of SARS-CoV-2 SP found in endothelial cells after IV mRNA vaccination
[154][155][156] support the possibility that SP could be generated by nucleated endothelial cells in blood vessels post-vaccination. In mice, after a second IV vaccine dose, 67% had grossly visible white patches over the visceral pericardium and all showed changes of myopericarditis, compared with only mild degenerative changes in the myocardium in the intramuscular (IM)-injection group
[157].
6.3. Clinical Signs of Microvascular Occlusion and Myocarditis after Exposure to SARS-CoV-2 SP
Further insights into microvascular occlusion caused by SARS-CoV-2 SP in the absence of whole virus in a clinical setting were provided by optical coherence tomography angiography (OCT-A) imaging of the retinal microvasculature. Determinations of the vascular density (VD) of flowing blood vessels in various retinal layers of human subjects, an indicator of microvascular occlusion, found that the CoronaVac vaccine, made from inactivated whole virus, caused no changes after vaccination
[158][159]. The Pfizer-BioNTech mRNA vaccine caused small but statistically significant reductions in VD vs. controls at three days
[160] and at two and four weeks
[159] after vaccination. Reductions in many of these VD values at two weeks after vaccination were statistically significant at
p < 0.001; most of these resolved by four weeks after vaccination, but seven of these VD reductions persisted at statistically significant levels at that time
[159].
The significance of these findings derives not from the occasional ocular adverse effects that have been reported after mRNA COVID-19 vaccinations
[161][162] but rather from indications that ocular microvascular occlusion mirrors a pathology elsewhere in the body
[112][163]. Myocardial injury is another indicator of microvascular occlusion, as noted above, which opens another diagnostic window, PET-CT scanning, since fluorodeoxyglucose F18 (FDG) uptake in myocardial tissue has been found to track myocardial injury
[164][165].
7. Decreased Clinical Severity of COVID-19 by Agents That Inhibit RBC Aggregation
7.1. Fluvoxamine
Fluvoxamine (FLV), a selective serotonin reuptake inhibitor (SSRI), attracted interest from prominent medical researchers
[166][167][168] after early clinical trials indicated promising results for COVID-19 treatment
[169][170][171][172]. Although rapid recovery from severe illness was not generally observed, one study showed a significant reduction in residual symptoms of COVID-19 at 14 days after start of FLV treatment vs. untreated controls
[169], and another showed significant reductions in emergency room visits or hospitalizations
[171].
A plausible biochemical mechanism is the sharp reduction by FLV in serum levels of serotonin, which is a powerful inducer of RBC and platelet aggregation. In vitro, serotonin caused marked aggregation of RBCs, platelets and leukocytes
[88]. In vivo, injection of serotonin resulted in blood cell aggregates being trapped in small venules and capillaries in the ocular conjunctival vasculature
[88]. In dogs, a serotonin antagonist prevented an increase in pulmonary alveolar dead space, an indication of pulmonary vascular obstruction, after hemorrhagic shock
[173].
7.2. Hydroxychloroquine (HCQ)
The application of HCQ, an aminoquinoline, for treatment of COVID-19, as developed by an infectious disease team at Aix-Marseilles University in France
[174][175][176], has been the subject of significant controversy, a review of which is not attempted here. However, it is of note that HCQ has been found to have pronounced activity in reducing blood cell aggregation and associated microvascular occlusion. In 44 human subjects with vascular conditions including coronary artery and cerebrovascular disease, all having initial manifestations of microvascular occlusion, ocular conjunctival microvasculature was observed over a nine-month period following the start of HCQ treatment
[177].
7.3. Ivermectin (IVM)
To identify potential therapeutics for COVID-19, four in silico studies collectively screened over 1000 molecules for binding to SARS-CoV-2 SP and other SARS-CoV-2 viral targets
[52][178][179][180]. In each of these studies, the strongest or close-to-strongest binding affinity to SP was obtained for IVM, a macrocyclic lactone with multifaceted antiparasitic and antimicrobial activity, distributed in four billion doses for human diseases worldwide since 1987
[181][182][183]. Aminpour et al. (2022) found by molecular docking computations that IVM binds with high affinity (<−7.0 kcal/mol) to seven sialoside-binding sites or other glycan-binding sites on SARS-CoV-2 S1, six on the N-terminal domain (NTD) and one on the receptor-binding domain (RBD). These binding energy values of <−7.0 kcal/mol were obtained for the RBD in both the open (“up”) and closed (“down”) positions
[47]. As a measure of significance of this binding energy value, binding energies of <−7.0 kcal/mol predicted efficacy for a large set of HIV inhibitors with 98% sensitivity and 95% specificity in another study
[184]. Additional molecular modeling studies of IVM binding to SARS-CoV-2 SP
[185][186][187][188][189], including one by Lehrer and Rheinstein (2020)
[190], likewise found strong binding affinities for IVM.
Competitive binding by IVM to SP glycan-binding sites is thus a likely biochemical mechanism for the in vitro inhibition and reversal by IVM of aggregation of human RBCs by SARS-CoV-2 SP as noted above
[38]. In early 2020, two Florida physicians, Jean-Jacques and Juliana Rajter, were intrigued by an indication of a clinical parallel to this in vitro effect—observations that several COVID-19 patients with severe respiratory impairment and SpO2 deficits experienced normalized breathing function within 1–2 days after treatment with IVM
[191][192]. Months later, Rajter et al. (2020) reported results of a propensity-matched case control study of COVID-19 patients treated with a low dose of IVM plus standard of care (SOC) at four Florida hospitals, which yielded a 40% reduction in mortality vs. controls given only SOC
[193]. After that study was concluded, through mid-2021, more than 20 randomized controlled trials (RCTs) of IVM treatment of COVID-19 were conducted, with six of seven meta-analyses reporting notable reductions in deaths and with a mean 0.31 relative risk of mortality vs. controls (a 69% reduction)
[194].
By 2022, however, several other RCTs reported that IVM yielded no statistically significant benefits for COVID-19 treatment, as summarized in an August 2022 editorial which declared that it was “time to stop using ineffective COVID-19 drugs”
[195]. Curiously, however, the editorial prominently cited in support a June 2022 meta-analysis of ten RCTs for IVM treatment of COVID-19 encompassing 3472 patients
[196] which actually reported as the first finding in its results section a twofold reduction in deaths in its pooled IVM treatment vs. placebo groups (pooled log odds ratio of −0.67, 95% CI = −1.20 to −0.13, with low heterogeneity). The mortality reduction was less (log OR = −0.12) for the RCTs rated as having a low risk of bias, but included in that group, weighted to account for 63% of that pooled log OR, was a study of dubious credibility.
Coauthors of the aforementioned study, the TOGETHER trial
[197], have repeatedly refused to disclose four of their key outcome numbers, namely per protocol deaths and hospitalizations, treatment vs. placebo
[198], which are of key importance given critiques of the primary outcome used in all arms of the TOGETHER trial by the U.S. Food and Drug Administration
[199] and National Institutes of Health
[200]. Instead, a TOGETHER trial coauthor directed inquiring scientists to the ICODA data repository, listed as the data source in the study’s data sharing statement
[198][201]. After two months of futile attempts by scientists to obtain the data from ICODA, however, on 7 June 2022, an ICODA manager reported that the TOGETHER trial data were never held by that data repository and that she had instructed its authors to stop citing it as their data source
[198][202].
In another prominently cited RCT of IVM treatment of COVID-19 that reported negative conclusions
[203], IVM was substituted for placebo doses for 38 patients, a mistake caught a month later, and blinding was broken by use of sugar water as the placebo for one-third of the study’s patients (liquid IVM has a bitter taste). Adverse events that are distinctive to the high IVM dose used (transient and non-critical) occurred at almost identical rates in the placebo and IVM arms, while over-the-counter sales of IVM surged in the study region during the study period
[204].
The RCT evidence for IVM-based treatments of COVID-19 is thus mixed; however, in rare cases, efficacy of a drug has been conclusively established without RCT findings when it has achieved consistent major clinical benefits in the face of an established baseline of null effect. For example, the 96% cure rate for peptic ulcers by a triple therapy achieved in a 1990 clinical trial
[205] provided conclusive evidence of the therapy’s efficacy, given a baseline of palliative but rarely curative results for that chronic condition
[206][207]. The associated discovery of
H. pylori as the underlying pathogen for peptic ulcers was honored with the Nobel prize for medicine in 2005
[208]. For penicillin, early in vitro and mouse studies provided convincing indications of marked antibacterial activity. Alexander Fleming found in 1929 that transparent regions formed around penicillin embedded in agar plates of several species of cultured bacteria, indicating inhibition of bacterial growth
[209]. In the absence of RCT evidence, penicillin production was then ramped up to industrial scale, saving the lives of thousands of soldiers during World War II
[210].
For cases of moderate and severe COVID-19 in patients on room air, there is a consistent baseline of null effect in a 1–2 week timeframe: the magnitude of reductions in SpO2 levels correlate with the extent of pulmonary damage, and neither of these normalize in that timeframe
[211][212][213][214][215][216][217]. With that backdrop of null effect, three studies of severe COVID-19 patients on room air treated with IVM-based regimens observed sharp increases in SpO2 after 1 day of treatment
[218][219][220][221] while SpO2 decreased during the same 1-day period in a fourth group of such patients under standard care.
For one of these three studies, Stone et al. (2022)
[218], taking into account some missing values for the 34 treated patients at <48 h post-treatment, paired t-test calculations were performed for post-treatment minus pre-treatment SpO2 values for the study patients at +12 h, +24 h and +48 h after the start of IVM administration. These paired t-test values were highly significant, with
p < 10
−6 in each case. One patient in the study had an increase in SpO2 from 79% recorded at the first IVM dose to 95% three hours later, and four other patients had increases of 12 or more in SpO2% within 12 h after the first IVM dose. These sharp, rapid improvements parallel the disaggregation of RBC clumps observed in vitro over the course of 30 min by Boschi et al. (2022) and can be explained by rapid clearance of RBC aggregates in the vasculature and corresponding increases in efficiency of oxygenation in pulmonary and extrapulmonary tissues.
In 2020, Peru provided a unique setting to track clinical efficacy of IVM-based treatment for COVID-19 with close consideration of confounding factors, using excess deaths data from its national health system, which aligned with WHO monthly summary data
[222]. Treatment with IVM and adjunct agents was deployed at intensive, moderate or limited levels under semi-autonomous policies in its 25 states, enabling comparisons with reductions in excess deaths at 30 days after peak values, state by state. A Kendall tau calculation yields a two-tailed
p-value of 0.002 for reductions in excess deaths correlated with level of IVM use in Peru’s 25 states. On a national scale, during four months of IVM use in 2020, before a new president of Peru elected on November 17 restricted its use, there was a 14-fold reduction in nationwide excess deaths, and then a 13-fold increase in the two months following the restriction of IVM use
[222]. This set of real-world national health data, accompanied by extensive additional data by which potential confounding influences can be tracked, provides another significant indication of efficacy of IVM treatment of COVID-19.
8. A Comparison of Degree of Clinical Susceptibility to COVID-19 and RBC Aggregability in Various Animal Species
Susceptibility to COVID-19 and severity of this disease have been tracked for dozens of mammalian species, as reported in a summary figure by Meekins et al. (2021)
[223]. RBC aggregability values and related values of blood viscosity at low shear velocity have been tracked for many mammalian species as well, as reported by Baskurt and Meiselman in 2013
[224]. A correlation calculation between these two values, by species, provides a test of whether RBC aggregability is likely associated with COVID-19 morbidity.
The COVID status of mammalian species was reported by Meekins et al. using designators for viral shedding, clinical signs, mortality and transmission. Scholars derived a composite COVID status index from the first three of these indicators (transmission was not used) with values of 0 for none of these three, 1 for viral shedding only, 2 for clinical signs and 3 for clinical signs and mortality. For RBC aggregability, an aggregation index shown in Baskurt and Meiselman for 22 mammalian species was used. They also reported values of blood viscosity under low-shear conditions for 27 mammalian species that were closely correlated with the corresponding RBC aggregation index for species having values shown in both figures. For a species tracked in Baskurt and Meiselman that reported blood viscosity but not RBC aggregability, the latter value was interpolated from the blood viscosity value. Correspondence between RBC aggregability and blood viscosity was established using the values of each for cattle and horses; these species had the minimum and maximum values of all species tracked by Baskurt and Meiselman, respectively, for both of these indices.
9. Conclusions
The central role of sialylated glycan attachments between SARS-CoV-2 SP and RBCs and other blood cells in the severe morbidities of COVID-19 is founded on well-established biochemistry of coronaviruses and RBCs and established here through multiple channels of substantiation. Many preclinical and clinical studies show that SARS-CoV-2 SP attaches to and aggregates RBCs. Experimentally induced RBC clumping in vivo causes the same morbidities and the same redistribution of blood flow from smaller to larger blood vessels as for severe COVID-19. The key risk factors of increased age, diabetes and obesity for COVID-19 morbidity are each associated with significantly increased RBC aggregation. SARS-CoV-2 SP in the absence of whole virus as generated experimentally by IV injection of mRNA COVID vaccines in vivo, which caused SP to be generated in the absence of whole virus, induced microvascular occlusion.
Three generic agents which attracted prominent interest as COVID-19 therapeutics all yielded significant reductions in RBC aggregation. For mammalian species, the degree of clinical susceptibility to COVID-19 correlates with the aggregation propensity of RBCs with p = 0.033. These in vitro, in vivo and clinical findings, together, provide a convincing demonstration that RBC aggregation induced by SARS-CoV-2 SP through sialylated glycan attachments and resulting microvascular occlusion is key to the morbidities of severe COVID-19. These insights can support therapeutic and preventative strategies for evolving variants of this disease and for long COVID, while imaging of the retinal or sublingual microvasculature of active or long COVID patients can provide important support to these efforts.