Current management guidelines for COVID-19 reflect the assumption that critically ill patients infected with SARS-CoV-2 develop acute respiratory distress syndrome (ARDS). However, emerging data and clinical reports increasingly suggest an alternative view that severe COVID-19 reflects a confluence of vascular dysfunction, thrombosis, and dysregulated inflammation.

Severe COVID-19 is distinct from ARDS and cytokine-release syndromes

Published data [1, 2], anecdotal observations, and discussions with colleagues worldwide indicate that COVID-induced respiratory phenotypes are distinct from typical ARDS in several ways [2]. COVID-19 patients develop profound hypoxemia early in their disease course. However, overt respiratory dysfunction at these early stages is unusual. Pulmonary compliance in intubated COVID-19 patients appears to be only modestly decreased, and patients are therefore relatively easy to ventilate [1]. Gattinoni et al. [1] described hyperperfusion of poorly ventilated lung, perhaps due to vasodilation and the loss of hypoxic vasoconstriction. While benefit from low-to-moderate levels of PEEP and prone positioning have been noted, these authors suggest that they result more from hemodynamics effects than lung recruitment [1]. Later in the course of COVID-19, some patients develop a phenotype more consistent with ARDS [1, 2].

The reported inflammatory response in COVID-19 is also not consistent with either typical ARDS or cytokine-release syndromes (CRS) or “cytokine storm.” Qin et al. [3] report mean interleukin-6 levels were 25 (SD: 10–55) pg/mL (normal range: 7 pg/mL). Other smaller COVID-19 reports have ranged from 7 to 125 pg/mL. These findings contrast with interleukin-6 elevations seen in typical ARDS and in CRS. Sinha et al. report mean interleukin-6 levels of 282 (111–600) pg/mL in “hypoinflammatory” ARDS [9] and 1618 (517–3205) pg/mL in hyperinflammatory ARDS [4], 10- to 60-fold higher than reported in the Wuhan data. Among CRS patients, mean interleukin-6 levels are frequently as high as 10,000 pg/mL [5]. Other inflammatory cytokines (e.g., interleukin-8, interleukin-1β) demonstrate similar patterns. In summary, COVID-19 is associated with only mild inflammatory cytokine elevation and demonstrates physiology and immunology that are difficult to reconcile with ARDS or CRS. An alternative mechanism of disease therefore seems likely.

Vasculopathy and dysregulated inflammation in COVID-19

The combination of observed physiology and emerging pathologic evidence points toward a vascular disease process as contributing factor in COVID-19 pathogenesis. Pulmonary shunting is consistent with intense vasodilation and endothelial dysfunction. The observation that 89% of hospitalized patients in Rome showed subsegmental vascular enlargement on their admission computed tomography scan supports this view [6]. Reports of increased respiratory dead space suggest lung-vascular thrombosis from thrombotic microangiopathy or pulmonary embolism. The latter was recently reported in up to 40% of hospitalized COVID-19 patients [7]. Autopsies performed on patients who died early on were indeed notable for lung-vascular congestion [8]. Vascular disease may also explain massive D-dimer elevations, while antiphospholipid antibodies were recently reported in COVID-19 [9]. A constellation of multi-system organ involvement, low-grade inflammation, lymphopenia, hypercoagulability, and heterogenous microvascular dysfunction is a classic description of many systemic vasculopathies, such as vasculitides (Supplement Table).

Reported findings indicate that immunosuppression, endothelial activation, and direct viral-mediated tissue damage, rather than hyperinflammatory injury, mediate COVID-induced organ dysfunction. For example, a recent autopsy study found no renovascular or interstitial inflammation, but noted endothelial activation, occasional frank necrosis, and copious virions in renal tissue [10]. If direct infection drives injury, vascular tissue is expected to be quite susceptible as it highly expresses angiotensin-converting enzyme-2 (ACE-2), which is essential for coronavirus uptake.

SARS-CoV-2 initiates cellular infection by binding ACE-2 on the surface of human cells, including endothelial cells (Fig. 1) [11]. Attachment promotes disordered cytokine paracrine signaling, including both pro- and anti-inflammatory molecules, and pro-apoptotic mediators [12]. Chemokine-mediated lymphocyte recruitment and subsequent infection of lymphocytes, which also express ACE-2, likely contribute to lymphocyte apoptosis, natural killer and B cell suppression, and T cell exhaustion, as noted by Qin et al. [3]. These findings are consistent with clinical lymphopenia, which correlates with mortality [3]. Viral injury, disordered cytokine release, and damage-associated molecular patterns (DAMPs) induce localized microvascular inflammation, which triggers endothelial activation, leading to vasodilation and pro-thrombotic conditions.

Fig. 1
figure 1

(1) The SARS-CoV-2 virus infects an endothelial cell by binding to ACE-2. Cellular infection initiates localized inflammation, endothelial activation, tissue damage, and disordered cytokine release. Membrane fusion also interrupts AngII metabolism, leading to an increase in AngII and a decrease in Ang (1–7), augmenting inflammation, endothelial activation, and leukocyte and platelet recruitment. (2) Pulmonary endothelial activation leads to the ACE-1 shedding phenomenon, where ACE-1 is rapidly liberated from the cell membrane. This produces an initial rapid rise in AngII, which can induce a positive feedback loop enhancing local inflammation, coagulation, and capillary leak. (3) The transiently increased ACE-1 dissipates, leading to subphysiologic AngII concentrations. Low AngII in this phase leads to vasodilation, worsened capillary leak, and impaired endothelial conductance and autoregulation. Low systemic AngII also upregulates ACE-2, possibly increasing susceptibility to SARS-CoV-2 in remote tissue. ACE, angiotensin-converting enzyme; AngI, angiotensin-I; AngII, angiotensin-II; Ang (1–7), angiotensin (1–7); DAMPs, damage-associated molecular pattern molecules

Angiotensin-II (AngII) is primarily metabolized by endothelial ACE-2 to the vasodilatory and anti-inflammatory peptide angiotensin (1–7). In the early phases of infection, ACE-2 consumption by viral entry would be predicted to increase local AngII concentration. Among the known effects of AngII are vasoconstriction, endothelial activation, and pro-inflammatory cytokine release. Platelet activation by AngII may further enhance a pro-thrombotic milieu. AngII also has potent chemotactic effects that may accelerate lymphocyte recruitment and suppression. Importantly, admission AngII levels in COVID-19 patients are reportedly twice the normal levels [13]. In addition, pulmonary vascular inflammation specifically leads to a phenomenon known as ACE-1 “shedding,” where endothelial surface-bound ACE-1 is released into the interstitium [14]. This phenomenon initially produces a sharp increase in local AngII that gradually decreases until concentrations are well below physiologic levels [14]. Because AngII downregulates ACE-2 expression, transition to a state of AngII deficiency is consistent with the increased ACE-2 expression noted at autopsy [10]. Of note, AngII reportedly increases microvascular permeability under basal conditions but decreases permeability during inflammation, an effect attributed to the inflammation-induced shift from type-1 to type-2 receptor expression [15].

Implications for therapeutics and research

Postulating that COVID-19 is a vascular and hypoinflammatory disease has important implications for ongoing research. However, more evidence is needed to explore vascular injury through biomarkers, imaging, and histology. If COVID-19 is indeed primarily a vascular disorder, early invasive mechanical ventilation should be initiated cautiously. Investigations should examine the impact of liberal versus restrictive invasive mechanical ventilation strategies.

We reiterate that there is no evidence that COVID-19 patients develop “cytokine storm.” This finding suggests that the use of cytokine-blockade agents should meet with skepticism in the absence of randomized evidence. However, anticoagulation should be a key priority for investigation. Similarly, given the putative role of AngII deficiency, a randomized trial of angiotensin-II treatment in COVID-19 patients who have progressed to shock is warranted. Illness progression may also be an effect modifier, with potential benefits of anti-inflammation and angiotensin blockade earlier in critical disease and harms at later time points.

Conclusion

COVID-induced respiratory failure involves physiologic, clinical, and immunologic phenotypes that are not consistent with either ARDS or cytokine-release syndromes. COVID-19 instead reflects immunosuppression and features compatible with vascular disease.