Statement of hypothesis
Though in several observational data, a significant proportion of COVID-19 patients have presented with silent hypoxia with preserved respiratory compliance, thus giving it a terminology of atypical ARDS [1, 9], only a few have attempted to relate this phenomenon with increased intrapulmonary shunt [1, 2, 3]. Despite this, based on these observations, many researchers proceed to evoke a hypothesis of bradykinin as a mediator for pulmonary vasoplegia and inflammation in COVID-19. However, no study till date have demonstrated elevated bradykinin levels in either serum or bronchoalveolar lavage fluid in COVID-19 patients. Although, we agree with the Gattinoni et al. [1] Reynolds et al. [2] and Brito-Azevedo et al. [3] findings of increased intrapulmonary shunt in COVID-19 pneumonia but, we are doubtful of their conclusion of pulmonary vasodilation as the predominant phenomenon resulting in increased intrapulmonary shunt and severe hypoxia. Instead, to explain the phenomenon of silent hypoxemia in COVID-19 respiratory failure, we developed an alternate pathophysiological model – Epithelial-Endothelial crosstalk hypothesis – involving upregulation of ACE-Ang II- AT1R pathways producing a mosaic pulmonary perfusion pattern of non-homogenous pulmonary vasoconstriction with resultant intrapulmonary shunting and dead-space [10, 11]. We base this on the following arguments:
First , blood flow through intrapulmonary arteriovenous anastomoses has been demonstrated in approximately 30% of healthy adult humans at rest, and in 100% of healthy adult humans during exercise and inhalation of reduced oxygen gas mixtures [12]. Similarly, pathological conditions such as high-altitude pulmonary edema, bronchopulmonary dysplasia and congenital diaphragmatic hernia are associated with intrapulmonary shunts as the predominant mechanism for hypoxia [12]. In all these conditions raised pulmonary vascular resistance (PVR) is a common denominator and non-homogenous pulmonary vasoconstriction has been linked as the mechanism for the recruitment of intrapulmonary arteriovenous anastomosis and increased shunt [13], but not pulmonary vasodilation. Catecholamine infusions at rest, by increasing cardiac output, can increase the shunt fraction by increasing blood flow through these pathways and negatively impact pulmonary gas exchange [12]. Notably, 66% patients in the Brito-Azevedo et al. [3] study were on vasopressors while Reynolds et al. [2] did not provide information about vasopressors in their study. Furthermore, platypnea and orthodeoxia, the classical clinical features of significant pulmonary vasodilation-induced shunts, as seen in hepatopulmonary syndrome [14] (an entity closely compared to the intrapulmonary shunt of COVID-19 pneumonia [2]), are not a part of COVID-19 symptomatology.
Second , although TCD and TTSCE are well-known modalities to determine intrapulmonary shunt, neither TCD nor TTSCE can determine the pathophysiological process responsible of the development of shunt: anatomical, vasodilation or non-homogenous pulmonary vasoconstriction.
Third , since COVID-19 lung compliance at the stage of silent hypoxia is largely preserved and the areas of consolidation are minimal [1, 3], the fraction of gasless tissue would be expected to be minimal [1]. Thus, impaired hypoxic pulmonary vasoconstriction (HPV) as a cause for severe hypoxia in early COVID-19 patients should not be the sole mechanism. Studies based on in-silico [15] and mathematical [16] modelling also reinforce that even complete loss of HPV could not recreate severe hypoxia observed in COVID-19 pneumonia.
Fourth , the hypothesis of Ang II-induced downregulation of ACE activity is not supported by translational research on renin-angiotensinogen-angiotensin pathways. While increases in bradykinin levels are possible secondary to angiotensin-type-2 receptor (AT2R)-mediated Ang II action, this is often a counterregulatory mechanism [17] and hence, should not predominate. Previous studies on ARDS found a correlation between pulmonary capillary endothelium bound ACE activity and severity of ARDS [18]. High ACE levels in bronchoalveolar lavage (BAL) fluids have been observed in ARDS despite a reduction in serum ACE levels [19]. Though, no study on COVID-19 has actually measured the ACE levels in BAL fluid and ACE protein expression in lungs, a recent study measuring the circulating ACE levels in COVID-19 patients demonstrated decreased circulatory levels of ACE in severe COVID-19 patients that normalizes during the recovery phase [20]. This reduced serum ACE levels may reflect loss of enzyme release from a damaged pulmonary vascular endothelium, but not be the true representative of ACE activity in the lung compartment [20] that may actually be increased [21]. High tissue ACE activity increases degradation of bradykinin to inactive metabolites, inhibiting vasodilator pathways in lungs. Similar mechanism can explain the low incidence of rhinorrhea and nasal congestion in SARS-CoV-2, unlike other viral upper respiratory tract infections wherein bradykinin is considered to be the most important mediator. The presence of ACE in the superficial lamina propria and the endothelium of superficial blood vessels of human nasal mucosa also suggests the possible role of ACE in limiting the inflammatory effects of bradykinin in nasal mucosa [22]. A report of increased cerebrospinal fluid levels of ACE in two patients with transient COVID-19 encephalitis (˂ 3 days) further highlights the pathological role of local ACE upregulation in COVID-19 [23].
Fifth, and most important is to understand the normal physiological and pathological roles of Ang II in humans. In health, AT2R and masR, but not angiotensin-type-1 receptor (AT1R), are the only angiotensin receptor types expressed on pulmonary endothelium and AT1R are expressed on the pulmonary vascular smooth cells and mesenchymal cells located beneath the pulmonary endothelium [24]. Also, the locally produced Ang II in the vascular lumen could not crosses the intact endothelium. Thus, in healthy alveolar capillaries, AngII-AT2R and Ang(1-7)-masR induced nitric-oxide (NO)-mediated vasodilatory effect is overactive and maintains low PVR, explaining why Ang II infusions in physiological dose ranges failed to increase PVR [25, 26]. However, with high-dose Ang II infusions AT1R activity predominates and can produce intense non-homogenous pulmonary vasoconstriction and ventilation-perfusion mismatching akin to that produced by hypoxia [27, 28]. Ang II infusion has also been demonstrated to produce non-cardiogenic acute pulmonary edema [21]. Also, on one hand, the subpressor levels of Ang II activate AMP-activated protein kinase (AMPK) and increases ACE2, the high levels of Ang II (pathological state) inhibit AMPK, upregulates ACE and downregulates ACE2 via the AT1R-ERK/p38 MAPK pathway [29]. Thus, Ang II has beneficial as well as harmful vascular effects depending upon the Ang II levels as well as effector receptors expressions. Importantly, COVID-19 patients have high systemic Ang II levels that are linked with disease severity [30] and decrease in Ang II levels following human recombinant soluble ACE2 therapy correlated with the clinical improvement [31].