COVID-19 discussion pages:
COVID-19 Lung Disease

A Pulmonary Vasculopathy

COVID-19 lung disease is not a respiratory pneumonia. Radiologically the disease is more accurately considered to be a disease of the lung vessels, a pulmonary vasculopathy.

For an in-depth explanation of the concept of COVID-19 lung disease as a pulmonary vasculopathy, please read the article below.

Watch the presentation by Dr Graham Lloyd-Jones, Director of Radiology Masterclass, given on November 11th 2020 at the annual (virtual) conference of the British Society for Haematology, titled - What can the Radiology tell us about the vasculopathy of COVID-19 lung disease?

Acute COVID-19 lung disease: A pulmonary vasculopathy - not a respiratory pneumonia

This page expresses the opinions of Dr Graham Lloyd-Jones BA MBBS MRCP FRCR, Director of Radiology Masterclass.

First published: 3rd November 2020

Most recent major update: 10th November 2020

Introduction

Having examined many images in preparation of the Radiology Masterclass COVID-19 image gallery, I find myself left with questions relating to the imaging appearances of COVID-19 pneumonia which currently are unsatisfactorily answered by the radiological literature.

It is now clear from autopsy studies that COVID-19 pneumonia is unlike other pneumonias. It is a disease primarily driven by the consequences of damage to endothelial cells of small pulmonary blood vessels [1,2,3]. In this post I explore why we should not describe COVID-19 lung disease as a ‘pneumonia’ or ‘lower respiratory tract infection’, but rather as a pulmonary vasculopathy (vascular disease of the lungs).

Background

On viewing the first CT examples of COVID-19 lung disease my initial impression was that the peripheral consolidation had a pattern resembling multiple wedge-shaped pulmonary infarcts, similar to those we see in patients with pulmonary emboli on CT pulmonary angiography (CTPA) (Figure 1). This is likely an oversimplification but since those early days I have been looking for an explanation for this appearance.

Radiologists soon learned that COVID-19 causes a pneumonia in a specific pattern. We became comfortable making a confident diagnosis; in the context of the typical clinical symptoms, anyone with patchy bilateral lung shadowing in the periphery of the lung bases could be diagnosed with COVID-19 pneumonia, regardless of their swab result (Figure 2).

The CT appearances are as characteristic as the radiographic features, bilateral ground-glass opacities being the dominant finding (Figure 3). Other observations include the predominant right lower lobe involvement, the reversed halo sign, features ascribed to ARDS in severe cases, and, in the intermediate to chronic phase, the development of organising pneumonia and fibrosis [1,4,5].

But we still need to ask the question: Why does the disease cause the distinct pattern of changes in the lungs?

Is the pattern of COVID-19 lung disease typical of pneumonia?

We know that the virus attacks cells, gaining entry via the angiotensin-converting enzyme 2 (ACE2) receptor [1] which is expressed in certain organs in the body, including the lungs, heart, kidney, and gut [2]. Inhalation and direct alveolar cell invasion seemed a rational explanation for viral entry to the body. Although this may still be the case, the radiological discussion notes the lack of features that suggest primary involvement of the airways. In the acute phase of the lung disease there is no central bronchial wall thickening, mucus plugging, lobar consolidation, abscess formation, centrilobular nodules, or tree-in-bud opacification of the small airways. Lymph node enlargement is not a key feature and pleural effusions are described as uncommon [1,4,6].

Radiologically, COVID-19 lung disease is vastly different from a pneumonia or lower respiratory tract infection, at least in the conventional sense of it being a disease driven by airways inflammation.

If COVID-19 is not a pneumonia then what is it?

The clinical and radiological discussion noted the increase in incidence of thromboembolic disease in patients with COVID-19 (Figure 4) – even in patients who have at least standard doses of thromboprophylaxis – a phenomenon which has been widely considered to be a complication of the pneumonia and systemic disease [7,8,9,10].

When we see central filling defects on a CT pulmonary angiogram (CTPA) in the context of COVID-19 we are confident in diagnosing pulmonary thromboembolic disease and comfortable calling the multiple wedge-shaped areas of peripheral opacities pulmonary infarcts. When the filling defects are not present, radiologists have described the peripheral wedge-shaped opacities as being typical of COVID-19 and concluded that there is no evidence of pulmonary thromboembolic disease.

But, there is evidence that CTPAs positive for pulmonary emboli (PE) in COVID-19 patients display a pattern of disease different from those in non-COVID-19 patients; a lower volume and more peripheral distribution of thrombotic burden is described in the context of COVID-19 [11]. We already know that peripheral vascular obstruction more often results in pulmonary infarction than even massive central clot burden [12]. Could it be that in COVID-19 we are seeing pulmonary infarcts without visible arterial filling defects centrally, as was my initial impression? Could these pulmonary infarcts be purely venous rather than due to arterial obstruction? (Pulmonary venous infarction is a known cause of ground-glass and wedge-shaped peripheral opacities [13].) One COVID-19 histological study reports that ‘combined capillary and venous thrombi were found without arterial thrombi’ [2].

Radiologically there are other features which point towards involvement of the blood vessels – the dilated peripheral vessels (Figure 5), as we see in conditions such as hepatopulmonary syndrome (a disorder mediated by systemic vasodilators causing abnormal vasodilation of peripheral lung vessels) [14-16], and the sparing of the immediate subpleural lung (Figure 6a and b), a finding which we would usually ascribe to vascular phenomena such as pulmonary haemorrhage or vasculitis [17]. Neither of these features seen in COVID-19 resemble a conventional pneumonia.

Autopsy reports mention a variety of features that we might expect to find in a pneumonia, such as diffuse alveolar damage, and alveolar exudative inflammation [18-22]. They also describe vascular congestion as a prominent finding, often with microthrombi in situ despite anticoagulation, and intra-alveolar fibrin exudation [1, 20-22]. Thickening of the alveolar capillaries is reported [18,22]. Pulmonary haemorrhage, vasculitis, and thrombotic microangiopathy are other features described [2,18,22].

These histological reports make it clear that the endothelial cells of small blood vessels of the lung are being damaged by the virus. Specifically, the terms ‘endotheliitis’ and ‘capillaritis’ are used [12,22]. We are also told that the ACE2 receptor is not only present in alveolar epithelial cells but also in the endothelial cells of the alveolar capillaries. Both are damaged [1,2,23].

Papers describe a systemic coagulopathy, the severity of which correlates with disease severity [23]. They specifically refer to the process of vascular microthombosis, or immunothrombosis, with pulmonary haemorrhage [24]. Studies also report the benefit of steroids [25], anticoagulation [3], and inhaled vasoregulatory drugs [26].

Could COVID-19 lung disease be a pulmonary vasculopathy?

In view of these papers, we need to reconsider the radiological appearances of COVID-19 lung disease. Rather than thinking of COVID-19 lung disease as a pneumonia or lower respiratory tract infection, we need to look at the disease as if it could be primarily affecting the vessels of the lungs – a vasculopathy (vascular disease) and ask the question: In what ways can we see that COVID-19 lung disease affects the blood vessels?

If we revisit the main radiological features of acute COVID-19 lung disease as seen on CT, we can see that, conceivably, they all relate to the findings described on autopsy studies. The areas of ground-glass opacification are non-specific and could relate to oedema or pulmonary haemorrhage of the lung interstitium secondary to vascular damage, rather than necessarily being secondary to alveolar damage. It is possible to speculate that the areas of peripheral consolidation could be due to pulmonary infarcts secondary to microthrombi of small vessels, with or without visible thromboembolic disease centrally. The dilated vessels we see in the lung periphery could equate to dysfunction of the local vasoregulatory system [27]. The sparing of the sub-pleural lung could perhaps be due to congestion at the level of the confluence of the interlobular pulmonary veins (Figure 6a and b), the smallest veins of the lung periphery, or similar to the processes we see in pulmonary vasculitis [17,28]. The reversed halo sign (or atoll sign) could relate to pulmonary infarction [29], rather than being caused by the organising pneumonia which is seen later in the disease process. The right lung predominance could also point to a thromboembolic process rather than preferential inhalation of virus into this lung; we know that pulmonary infarcts occur more frequently in the right lung [12]. It is perhaps even possible to speculate that the striking symmetry of lung involvement points to viral infection of the lungs via the haematogenous route with preferential infection of the periphery of the lower lobes, the most highly perfused areas of the lungs [30].

In all, there are no radiological features seen in the acute phase of COVID-19 lung disease which cannot be ascribed to some degree to a primary abnormality of the blood vessels. For this reason, it is possible, if not probable, that these vascular phenomena – such as microthrombosis – should not be considered secondary to a pneumonia or as a complication of the systemic disease, but should rather be considered as an integral part of the disease itself.

Furthermore, the disease process elsewhere in the body is also reported histologically as being driven by vessel damage [1]. If the process driving extra-pulmonary disease is also occurring in the lungs, then the phenomena of pulmonary and extrapulmonary thromboembolic disease should be considered as the same process, that of a systemic vasculopathy.

Why is the distinction between respiratory pneumonia and pulmonary vasculopathy important?

Radiologists have never before said we understand a lung disease without reference to the histology. The initial radiological descriptions from China indicated that the relationship between radiological and histological findings had not yet been investigated and so the cause of ground-glass opacities could not be established [31]. As yet, there have been no studies directly correlating imaging and histological findings, so our understanding is still based on assumption. It seems we have dismissed much of the imaging manifestations of COVID-19 lung disease as being related to airways inflammation or acute respiratory distress syndrome (ARDS). ARDS does occur later in the disease process but is described clinically as atypical [26,32,33].

Gaining understanding of a disease process is required before treatments can be optimized. There are ongoing trials of medications which seem to be based on the premise that the disease in the lungs is a conventional pneumonia and that the thromboembolic element of the disease is a complication, the treatment of which is considered when a patient has already become severely unwell [7,8]. Rather, there is good radiological rationale – as well as histological rationale – to suggest that future trials should be built on the understanding that COVID-19 lung disease is a complex vasculopathy comprising elements of vasculitis, microangiopathy with in situ thrombosis, vasodilation, and pulmonary haemorrhage. (The pulmonary haemorrhage reported histologically could either be within areas of ground-glass opacification [4,31] or in areas of peripheral consolidation as is seen in pulmonary infarction – driven by the bronchial circulation [34]).

If considering COVID-19 lung disease to be a vasculitis (more specifically an endotheliitis or capillaritis), as represented by ground-glass opacification with sub-pleural sparing, then it would perhaps not be a surprise that dexamethasone is effective [25]. If considering some or all of the wedge-shaped peripheral opacities seen radiologically in acute COVID-19 lung disease to relate to vascular congestion, or a microangiopathy with in situ thrombosis (with or without filling defects visible centrally on CTPA), then it would not be a surprise that anticoagulation is effective [3]. If considering COVID-19 lung disease as a disorder of vasoregulation in the peripheral lung vessels, as evidenced by the CT appearance of prominent peripheral pulmonary vessels, then there is perhaps radiological rationale for the benefit of inhaled vasodilators [26].

Some treatment trials are working on the basis that patients with severe COVID-19 lung disease are at high risk of thromboembolic complications and so discussion is focused on increasing anticoagulation above that given for thromboprophylaxis [24]. But, what if the lung disease visible on imaging at the time of clinical presentation already represents a large burden of thrombotic disease without visible thromboembolic disease evidenced by visible central filling defects on CTPA?

Does this hypothesis correlate with the clinical features?

If we consider the radiological features of COVID-19 lung disease to be primarily related to vascular disease, then it also helps us understand the clinical features. If the disease were driven by airways inflammation, then we would expect a productive cough, not a dry cough. The profound hypoxia could be explained by the combination of microthrombotic disease (ventilation/perfusion mismatch) [32], and by the visible vasodilation of peripheral lung vessels [15]. (This vasodilation is perhaps similar to the mechanism seen in the rare hepatopulmonary syndrome in which systemic vasodilators cause an arteriovenous shunt in the peripheral vessels of the lungs, thus leading to a profound hypoxia [14]. Interestingly, hepatopulmonary syndrome results in pulmonary vessel dilatation in a similar distribution to that seen in COVID-19 lung disease - peripheral and basal.)

The risk factors for poor outcome in COVID-19 also equate to the risk factors for thromboembolic and cardiovascular disease: age, sex, ethnicity, obesity, non-O blood group, hypertension, and diabetes [24, 35-40].

Is there a role for CT or Duel-Energy CT in COVID-19 lung disease?

Unless rt-PCR swab tests have been unavailable, CT has not been used to diagnose the disease. Until now its use has generally been limited to looking for visible thromboembolic disease or the complications of treatment [4].

Although there may not be a direct role for CT in making the diagnosis of COVID-19 lung disease, it is possible it has a role in characterising the type of changes we see in the lungs relating to vasculopathy. If studies were carried out with direct correlation of histological and radiological features it may be that we could characterise areas of abnormality in the lung visible on CT with pathological changes sufficiently to facilitate tailoring of treatments to patients in certain categories: those with vasculitic, macrothrombotic, microthrombotic, or vasodilatory type patterns.

It seems probable that the areas of peripheral subsegmental opacification seen in COVID-19 are similar in nature to multiple pulmonary infarcts. Evidence of the disease involvement of pulmonary vessels is mounting with recent studies using dual energy CT [15,16, 41]. These studies confirm that vascular abnormalities in COVID-19 lung disease are common and, importantly, show that areas of peripheral opacification correlate with perfusion defects in 96-100% of cases. This phenomenon occurred in patients with or without evidence of macrothrombosis (evidenced by visible filling defects on CTPA). Proximally dilated vessels and areas of increased blood supply are reported adjacent to these perfusion defects; features which would be atypical for conventional pulmonary infarcts.

So, the initial impression of peripheral wedge-shaped consolidation representing pulmonary infarcts is likely to be an oversimplification, but a very similar process is occurring. Overall, it would seem that a more complex mechanism is at play comprising damage to endothelial cells and inflammation of small vessels, microthrombosis in situ with perfusion defects, visible macrothrombosis, abnormal vasoregulation and arteriovenous shunting, even before the development of ARDS.

Conclusion

As yet, radiologists do not fully understand the pathogenesis of COVID-19 lung disease. We have tended to satisfy ourselves with being able to make the diagnosis without pursuing a deeper understanding. We need to be aware of the ongoing histological and haematological discussion regarding the pathogenesis of COVID-19. In the absence of direct radiological and histological correlation, radiologists need to gain understanding of the pathogenesis of COVID-19 with reference only to the indirect correlation of the radiological and histological features.

While further studies and trial results are awaited, those designing treatment trials need to know that there is radiological corroboration of the histological and haematological rationale to consider COVID-19 as a systemic vasculopathy in the acute phase, and that the key radiological manifestations of the disease visible in the lungs at the time of acute clinical presentation are primarily of a pulmonary vasculopathy, not a conventional pneumonia.

It is possible that CT and CTPA could be used to delineate specific phenotypes of vascular disease in the lungs with the purpose of tailoring treatment to specific patients.

The evidence for considering COVID-19 lung disease to be primarily a vasculopathy is now so strong that it seems likely that when we see lung changes in the acute phase of the disease on a CT, CTPA, or even a chest X-ray, these changes can be considered a marker of the extent of the vasculopathy itself, rather than as a sign of a respiratory pneumonia which may or may not later be complicated by vasculopathy.

Even more simply, this means that a chest X-ray with the typical pattern of lung shadowing, in the context of known or suspected COVID-19, could be considered an indicator of the extent of vasculopathy in the lungs, and so could be used as a clinical marker for trialling commencement of treatment for vascular entities – vasculitis, thrombosis, vasodilatation.