The Differing Pathophysiologies That Underlie COVID-19-Associated Perniosis and Thrombotic Retiform Purpura

A Case Series

C.M. Magro; J.J. Mulvey; J. Laurence; S. Sanders; A.N. Crowson; M. Grossman; J. Harp; G. Nuovo


The British Journal of Dermatology. 2021;184(1):141-150. 

In This Article


We have outlined two disparate cutaneous manifestations of COVID-19: (i) a highly inflammatory process with minimal vascular injury, manifesting as acral perniosis without significant systemic symptoms and designated as COVID-19-associated perniosis, and (ii) a pauci-inflammatory thrombotic complement-driven microvascular injury syndrome associated with microangiopathic ARDS.

Chilblains or perniosis represents a robust lymphocyte-rich vasocentric and perieccrine inflammatory reaction typically found on the digits. Two main variants have been described: (i) cold-associated idiopathic perniosis, typically seen in young women often with accompanying Raynaud phenomenon, and (ii) secondary perniosis of systemic disease including systemic lupus erythematosus and other autoimmune connective tissue disease (CTD), antiphospholipid antibody syndrome, genetic interferonopathies and viral infections, where cold exposure is an uncommon triggering event.[17]

COVID-19-associated chilblain-like lesions can be considered a form of secondary perniosis. A number of clinical descriptive studies during the COVID-19 pandemic have reported 'COVID toes', reflecting the frequent localization of the eruption to toes, although lesions also occur on the fingers. It is described most frequently in children but has also been reported in adults, including older adults in their eighties and nineties. The cases are typically asymptomatic or mild in character. Key differentiating features clinically from idiopathic perniosis include the absence of any temporal association with cold exposure and no evidence of Raynaud phenomenon. Many cases reported in the literature have not had any COVID-19 testing, and when COVID-19 nasopharyngeal swab or serological antibody testing was performed the results were typically negative.[9,10,18–20] Histological findings encompass a superficial and deep lymphocytic vascular reaction and lymphocytic eccrine hidradenitis similar to conventional perniosis.[9,10,18–20]

The biopsies of our cases of COVID-19-associated perniosis showed commonalities with idiopathic perniosis, including lymphocytic infiltration around vessels throughout the dermis and lymphocytic eccrine hidradenitits.[17] Key discriminating features include pronounced histiocytic infiltration, which included myeloperoxidase-activated histiocytes and CD11c, and CD123-positive monocyte-derived dendritic cells and some degree of frank vasculitic change that could be seen in reticular dermal blood vessels.

The livedoid and retiform skin lesions thus far examined in severe COVID-19 exhibit a pattern different from those seen in COVID-19-associated perniosis. The biopsies in severe cases were pauci-inflammatory and included features of occlusive fibrin thrombi and striking microvascular complement deposition. The distinct morphologies suggested two different pathophysiological patterns: the mild COVID-19 disease that prevails in children and the severe or critically ill COVID-19 disease that occurs in older adults with certain risk factors like obesity and diabetes mellitus.

Idiopathic/familial perniosis, chilblain lupus and Aicardi–Goutières syndrome are associated with mutations in TREX1 and RNASEH2A. Aicardi–Goutières syndrome is characterized by perniosis, periodic fevers and microcephaly, with progressive intellectual impairment due to cerebral lymphocytic vasculitis. The end result of these mutations is excessive interferon signalling. It has been suggested that these nucleases are involved in removing nucleic acids that accumulate during apoptosis, and when this process fails, an interferon signal is generated that activates the immune system. MXA, the surrogate marker for type I interferon, would be expected to be significantly upregulated in idiopathic/familial perniosis, chilblain lupus and perniosis in the setting of Aicardi–Goutières syndrome.[21]

In this current study, MXA was uniformly expressed at high levels in our cases of idiopathic/familial perniosis, as revealed by striking MXA expression in the epidermis, endothelium and inflammatory cells. The cases of COVID-19-associated perniosis showed extensive expression in the epidermis, endothelium and inflammatory cells for MXA, similar to idiopathic perniosis and secondary perniosis in the setting of excessive interferon signalling (i.e. chilblain lupus and Aicardi–Goutières syndrome). The nature of the infiltrate in this small study cohort is reflective of the type I interferon-enriched microenvironment, being one rich in CD14+ CD11c+ monocyte-derived dendritic cells including many with a classic CD4+ CD123+ plasmacytoid dendritic cell phenotype, activated myeloperoxidase-positive macrophages, and T cells with only rare B cells.[22–24]

The negative nasopharyngeal swabs in our two patients with COVID-19 perniosis may reflect the efficacy of interferon-driven monocyte and T-cell responses to clear the virus in these patients, and appear to be an almost ubiquitous finding in patients with COVID-19-associated perniosis.[9,10,17–19] Furthermore, serological testing was negative in one of our patients despite demonstration of SARS-CoV-2 viral proteins and RNA in the patient's skin lesions. More studies are needed to determine why patients with COVID-19-associated perniosis could have negative antibody testing, but the timing and sensitivity of the test and the individual immune responses to the virus are all likely contributors. A similar type I interferon-induced T helper 1 and monocyte response is seen in classic Kawasaki disease and accounts for the T-cell and monocyte-rich arteritis seen in these patients. By extension it is possible that the recently described multiorgan Kawasaki-like illness rarely seen in children with COVID-19 infection could develop as a result of excessively high levels of type I interferon.[11,25,26]

In contradistinction, the absent MXA expression in the skin biopsies from severe or critically ill patients with COVID-19 suggests a lack of a type I interferon response. Certain disease states can suppress type I interferon response, including obesity, which is associated with worse COVID-19 outcomes.[27] The production of the adipokine leptin is proportional to the body mass index, and its production is regulated by suppressor of cytokine signalling (SOCS)3. Terán-Cabanillas and Hernández showed that when SOCS3 is upregulated in a high-leptin state, in addition to leptin it suppresses the type I interferon response.[28] Type I interferons, unlike the more discussed type II interferons (i.e. interferon-γ), can be induced in all cell types, but are silent in a state of health. In the case of COVID-19, type I interferons are induced by single-stranded RNA of the virus binding to Toll-like receptors 7 and 8 in endosomes. The type I interferons produced are released and bind to the type I interferon homodimer receptor, which is ubiquitous in the body.[29] This leads to signalling through signal transducer and activator of transcription (STAT)1 and STAT2, which starts the production of a host of protective proteins, including interferon-induced transmembrane protein (IFITM)3, which acts as a complement regulator.[30] A polymorphism in IFITM3 has been shown to lead to poor outcomes in COVID-19 and is a topic that warrants further study.[31]

We propose that a strong type I interferon response may speed viral elimination, whereas a blunted type I interferon response would allow for massive viral replication leading to complement activation through the mannan-binding lectin pathway. It has been noted that interferon I response is often impaired in critically ill patients,[31] and that association plays out in transcriptional analysis of interferon in SARS-CoV-2-infected, angiotensin-converting enzyme 2-expressing cells in vitro.[32] Not surprisingly, SARS-CoV-2 is susceptible to type I interferon pretreatment,[32] as shown in a currently unpublished study from Vineet Menachery, lending further credence to the benefits of a strong type I interferon response. This hypothesis is shared by a group at Yale, who wrote an editorial that reviewed some of the existing literature on COVID-19-associated perniosis and suggested its potential pathophysiological commonality with idiopathic perniosis as a type of interferonopathy, although specific data supporting that hypothesis were not presented.[33]

The amount and microanatomical distribution of viral protein expression in COVID-19-associated perniosis, in contrast to the thrombotic purpuric lesions of severe or critically ill COVID-19, reflects the importance of the antiviral effects of type I interferons. In the cases of perniosis studied, there was no expression of SARS-CoV-2 protein in endothelium, and only a few cells of probable monocytic lineage contained SARS-CoV-2 protein and RNA. In the retiform thrombotic lesions, there was extensive localization of SARS-CoV-2 protein in the endothelium, although without any evidence of viral replication. There is evidence that the SARS-CoV-2 spike protein is able to engage the mannose-binding lectin (MBL) pathway, resulting in MASP-2 activation and subsequent formation of the C3 convertase and ultimately C5b-9 (i.e. the membranolytic attack complex). This is the effector mechanism of endothelial cell injury and subsequent thrombosis that we and others have previously reported.[5,33,34] Furthermore, MBL activation leads to further systemic amplification of the alternative pathway and activation of the coagulation pathway, along with well-established feedback-loop mechanisms. The striking expression of the proapoptotic marker caspase 3 likely reflects the sequelae of MBL activation, which is known to result in apoptosis via caspase 3 activation.[35]

The extensive degree of interleukin-6 upregulation in endothelium could be due to its elaboration by endothelial cells via MBL activation, as suggested in two separate studies.[36,37] Interleukin-6 may be involved in the pathogenesis of the vascular thrombosis through its effects on platelet aggregation and activation,[38,39] or upon angiotensin II regulation,[40,41] or it may simply be a bystander resulting from hypoxia.

In conclusion, patients with COVID-19-associated perniosis demonstrate a robust immune reaction characterized by a T-cell and interferon-rich microenvironment, which we believe leads to a mild course of the viral illness. The extent of interferon-driven inflammation that can occur in children is clearly effective in eliminating the virus, but could also have deleterious consequences if unleashed in a multiorgan inflammatory context. In the pauci-inflammatory thrombosed skin of very ill patients with COVID-19, interferon signalling is essentially absent, the implication being a permissive microenvironment that allows unchecked viral replication and triggers extensive complement activation of both the alternative pathway and the MBL pathway with catastrophic sequelae.