Positive Bubble Study in Severe COVID-19 Indicates the Development of Anatomical Intrapulmonary Shunts in Response to Microvascular Occlusion.

To the Editor:

We read with interest the recent article by Reynolds and colleagues (1) describing the transcranial Doppler bubble study findings in patients with coronavirus disease (COVID-19) with acute respiratory distress syndrome. The authors conclude that pulmonary vascular dilatation may be present in COVID-19, analogous to the microvascular changes that occur in hepatopulmonary syndrome (HPS), as a contributory mechanism of hypoxemia in COVID-19 acute respiratory distress syndrome. Although the findings on bubble study are indisputable, we share several concerns with the conclusions in the article.

First, in HPS (2), even though both / mismatch from overperfusion (capillary and precapillary dilatation) and anatomical shunt (abnormal arteriovenous communications) contribute to hypoxemia, positive bubble study is solely due to abnormal arteriovenous connections. As the diameter of the saline microbubbles is typically more than 24 μm and the diameter of pulmonary capillaries rarely exceeds 15 μm even after capillary distension (1, 3), the microbubbles are unlikely to pass through the capillaries. Thus, a positive bubble study in patients with severe COVID-19 does not automatically imply capillary dilatation or loss of hypoxic vasoconstriction but rather only the presence of abnormal pulmonary arteriovenous connections or an intracardiac shunt.

Second, although peripheral vessel dilatation is observed in COVID-19 on imaging studies (4), and frequently interpreted as abnormal vasoregulation, this cannot be equated to vasodilatation at the microcirculatory level. The converse may be true, as several imaging studies indicate that the subsegmental vascular dilation is a result of distal microvascular occlusion. Quantitative computed tomographic analysis by Lins and colleagues (5) revealed a marked loss of blood volumes in small vessels in patients with COVID-19 with increased blood volumes in medium and large vessels. This suggests increased pulmonary vascular resistance at the small-vessel level due to either microthrombi or arteriolar vasoconstriction. Furthermore, perfusion imaging studies using dual-energy computed tomographic imaging by Patel and colleagues (4) have shown a universal presence of perfusion defects in severe COVID-19, attributed mostly to microvascular thrombosis, involving a median extent of 46% of the entire lung.

Third, anatomical intrapulmonary shunts are present physiologically and may open up in response to increases in flow and pulmonary vascular resistance, akin to “pop-off” valves. For instance, exercise has been found to open up these shunts, contributing to increased alveolar–arterial oxygen gradient (3).

Fourth, diffuse pulmonary microvascular thrombosis and associated chemokine-mediated pulmonary vasoconstriction is sufficient to explain the atypical clinical features in COVID-19 such as silent hypoxemia and abrupt clinical deterioration (6). The mechanism of hypoxemia, similar to other pulmonary vaso-occlusive disorders, is flow redirection, resulting in overperfusion of the nonoccluded segments of the lung with reduced / ratios. Additionally, pulmonary vasoconstriction, if present, decreases red blood cell transit time (TRBC = microcirculatory volume/microcirculatory flow) in the alveolar capillaries, especially when exposed to higher flows, resulting in diffusion limitation, further exacerbating hypoxemia (Figure 1). As pulmonary infarction is not immediate after pulmonary vascular occlusion, lung compliance may be normal during the initial stages of pulmonary vascular occlusion, with preserved work of breathing. Dyspnea may, therefore, be absent despite profound hypoxemia in the initial stages of respiratory failure. However, minor changes in mixed venous saturation due to increased effort or deterioration in right ventricular function may cause a quick downward spiral resulting in rapid clinical deterioration. Progression of early COVID-19 respiratory failure thus mimics large pulmonary embolism, with similar lung mechanics and hemodynamics.

Figure 1.

Schematic representation of the mechanisms of hypoxemia in coronavirus disease (COVID-19) lung injury and clinical implications. Pulmonary in situ thrombosis results in hypoxemia owing to flow diversion and overperfusion of other areas with intact perfusion, leading to reduced / ratio in these segments. The dead spaces that occur because of loss of perfusion do not participate in gas exchange and are not relevant for oxygenation. Extensive perfusion loss is required before becoming clinically apparent owing to its moderate effect on effective / ratio, nonlinear relationship between / ratio and PaO (7, 8), and the presence of early compensatory mechanisms. However, when these mechanisms are exhausted, clinical deterioration is quick, aggravated by right heart compromise owing to acute pressure overload and reduced mixed venous saturations. Silent hypoxemia can be explained by a state of increased tissue extraction maintaining tissue oxygenation, at the expense of low arterial and venous oxygen saturations. This is facilitated by the subacute nature of the disease process and the lack of significant parenchymal injury at this stage; the work of breathing is not significantly increased, and dyspnea is minimal. CO = cardiac output; critical Do2 = critical oxygen delivery; O2 = oxygen; RV = right ventricular; SILI = self-induced lung injury; SO = mixed venous oxygen saturation; TRBC = red blood cell transit time in the alveoli.

In summary, the positive shunt study in severe COVID-19 indicates that abnormal arteriovenous communications open up in response to extensive small-vessel occlusion as the disease progresses. The findings fail to explain the initial severe hypoxemia in COVID-19 with preserved lung mechanics, as the degree of transpulmonary microbubble transit directly correlates with worsening lung compliance. The comparison with HPS is not appropriate owing to the evidence against microcirculatory dilatation in COVID-19. To conclude, anatomical pulmonary shunts do not contribute significantly to hypoxemia in early atypical COVID-19 respiratory failure and the distinct clinical features are best explained by progressive pulmonary vascular occlusion and subsequent diffuse lung injury due to various natural (infarction and oxidative damage) and iatrogenic sequelae.