Histopathologic Features of Chilblainlike Lesions Developing in the Setting of the Coronavirus Disease 2019 (COVID-19) Pandemic

All cases originated from the Department of Pathology, Hôpital Cochin, Assistance Publique-Hôpitaux de Paris, France, and were recorded between March 2020 and April 2020. Hôpital Cochin has been a referral center for COVID-19 patients. All patients with skin changes occurring in the setting of suspected or confirmed SARS-CoV-2 infection were seen by a dermatologist. All patients presenting with cutaneous acral lesions resembling chilblains and who had a skin biopsy were included. The study was approved by the Cochin Hospital Institutional Review Board. Written informed consent was obtained from all patients. We did not include in this series cases of COVID-19 patients with acro-ischemia occurring in the intensive care unit.¹⁵  These result from low peripheral blood flow, use of vasopressors, and disseminated intravascular coagulation. Our institution has no pediatrics department; the recruitment is therefore primarily adult patients. Hematoxylin-erythrosine-saffron–stained sections were prepared and slides were examined by 2 pathologists.

Immunohistochemistry expression was evaluated on deparaffinized formalin-fixed, paraffin-embedded sections, subjecting them to antigen retrieval using the Leica Bond protocol (LEICA Microsystèmes SAS, Nanterre, France) with proprietary Retrieval ER2 (ethylenediaminetetraacetic acid solution, pH 9.0) for 20 minutes. Mouse monoclonal antibodies against CD3 (dilution 1:300; polyclonal rabbit anti-human CD3 [concentrate], Dako), CD20 (dilution 1:400; monoclonal mouse anti-human CD20cy, clone L26 [concentrate] Dako), CD4 (dilution 1:200; clone 4B12, Novocastra/Leica BioSystems), CD8 (dilution 1:100; monoclonal mouse anti-human CD8, clone C8/144B, Dako), and CD123 (dilution 1:50; mouse monoclonal anti-human CD123, clone BSB-59, BioSB/Diagomics) were used as primary antibodies and detected by the Polymer Refine Kit (Leica Biosystems) on a Leica Bond autostainer. All immunostained slides were reviewed by a pathologist. T lymphocytes were counted using CD3 immunohistochemistry on a high-magnification field, counting approximately 50 lymphocytes typically around a vessel in the dermis. CD4 and CD8 immunohistochemistry was then used to count each subpopulation and establish the CD4:CD8 ratio. Presence of CD123⁺ plasmacytoid dendritic cells (pDCs) was evaluated on a review of 1 section per case. Distribution of CD123⁺ pDCs (perivascular, periadnexal, and/or dermoepidermal junction) and the presence of clusters were also recorded. Clusters were defined as nodular aggregates containing at least 10 pDCs. We were unable to obtain validated antibodies for SARS-CoV-2 antigen immunohistochemistry. Electron microscopy was not performed.

Mean and SD were used for normally distributed data. Median and interquartile range were used for data that were not normally distributed. Categorical variables were expressed as counts and percentages. All analyses were performed using R software, version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

A total of 13 patients with CBLL were included in the analysis. They all presented with erythematous acral lesions similar to chilblains (Figures 1, A, and 3, A). Their demographic and clinical characteristics are shown in the Table. Their median age was 32 years (interquartile range, 22–36 years), and 7 (53.8%) were men. The most common associated symptoms suggestive of COVID-19 were cough (5; 38.5%), diarrhea (3; 23.1%), rhinorrhea (2; 15.4%), shortness of breath (1; 7.7%), arthralgia (1; 7.7%), myalgia (1; 7.7%), and headache (1; 7.7%). Nine patients (69.2%) had at least one of these symptoms. Patients had only mild symptoms and none developed severe COVID-19 pneumonia. The following associated disorders were present: psoriasis (2; 15.4%), Crohn disease (1; 7.7%), ulcerative colitis (1; 7.7%), and Raynaud phenomenon (4; 30.8%). Two patients (15.4%) had a prior history of deep venous thrombosis and 1 had a prior history of pulmonary embolism.

Nasopharyngeal swabs for SARS-CoV-2 reverse transcription polymerase chain reaction were negative. Patients had no systemic inflammatory response, with normal white blood cell counts and C-reactive protein levels. Lactate dehydrogenase, alanine aminotransferase, and aspartate aminotransferase levels were normal. Blood urea nitrogen, creatinine, and creatinine kinase levels were within normal range. Anti-nuclear antibodies and D-dimer were normal and lupus anticoagulant testing results were negative. Serology tests for SARS-CoV-2 were not available.

The Table and Figures 1, A through D, 2, A through D, and 3, A through D, show the histopathologic findings in skin biopsies of CBLL occurring in the context of the COVID-19 pandemic.

Histopathology showed a superficial and deep infiltrate of lymphocytes in 10 of 13 patients (83%). The intensity of the inflammatory infiltrate was scored as mild (5 of 13; 38%), moderate (7 of 13; 54%), or dense (1 of 13; 8%). This infiltrate was perivascular in 12 cases (92%). Well-defined rounded lymphocytic cuffs around vessels were particularly prominent in 1 case (Figure 1, B). The distribution of this infiltrate was peri-eccrine in 4 cases (30%) (Figure 2, A). A variable amount of karyorrhectic nuclear debris was present in most cases, without any neutrophilic infiltrate. Vacuolar alteration of the basal layer (Figure 1, C) and rare scattered apoptotic keratinocytes in the upper layers of the epidermis were each seen in 7 patients (54%). Lichenoid interface dermatitis was found in only 1 case (Figure 2, B). Epidermal keratinocytes demonstrated changes consistent with epidermal injury and increased regeneration such as ballooned and swollen superficial keratinocytes and cellular and nuclear enlargement with conspicuous nucleoli. Ischemic epidermal necrosis was seen in 2 cases (15%) (Figure 3, D). Papillary dermal edema was present in 6 cases (46%). Dilated empty or congestive blood vessels were seen in the reticular dermis in 5 patients (39%). These capillaries or postcapillary venules were often lined by plump endothelial cells, sometimes assuming a hyperchromatic hobnaillike appearance with shrinkage of the endothelial cells, or less often showing endothelial sloughing (Figure 2, C). Extravasated erythrocytes were seen in 9 patients (69%). Intraluminal fibrin thrombi were identified in 5 cases (39%). Although perivascular infiltrates were present in most cases, we did not observe true vasculitis. Necrosis of the vascular walls, fibrin extravasation around blood vessels, and fibrin deposits within blood vessel walls were not observed. Small muscular arteries were invariably spared by the inflammatory infiltrate. Small muscular veins showed a sparse lymphocytic infiltrate within the media (Supplemental Figure 1; see supplemental digital content at https://meridian.allenpress.com/aplm in the February 2021 table of contents) in 6 of 7 cases (86%; biopsy did not sample veins that were adequate for evaluation in 6 other cases). Variable alterations of eccrine glands were present in 11 of 13 cases (85%), notably modifications of the epithelial border of the eccrine duct including apoptotic or necrotic scattered cells, atrophic or regenerative aspects of the epithelium, and a dystrophic irregular epithelial lining of the secretory portion of the gland. Mild interstitial deposits of mucin in the reticular dermis were seen in 1 case with Alcian blue, pH 2.5 stain (Figure 2, D). Periodic acid–Schiff staining did not show thickening of the epidermal basal membrane in any case.

By immunohistochemistry, dermal lymphocytes were overwhelmingly CD3⁺ T cells, consisting of an admixture of CD4⁺ and CD8⁺ T subtypes with a median CD4⁺:CD8⁺ ratio of 1.09:1 (interquartile range, 0.90–2.13; evaluation performed on 11 cases). CD8⁺ T cells demonstrated a cytotoxic phenotype, with strong immunohistochemical expression of granzyme B cytotoxic molecule. There were no or rare sparse CD20⁺ B cells. Plasmacytoid dendritic cells, highlighted by immunohistochemistry for CD123, were present in 8 of 13 cases (62%). Among those cases, distribution of pDCs was perivascular in all cases, along the dermal-epidermal junction in 1 case, and around acrosyringia in another case. Numerous clusters of pDCs were present and prominent in 1 case (Figure 1, D).

Two histopathologic patterns seemed to stand out among the cases. The first histopathologic pattern resembled features seen in chilblains or chilblain lupus. The features seen in the second pattern recapitulated a thrombotic vasculopathy (Figure 3, B).

The main observed features of chilblainlike histopathologic pattern were a superficial and deep, moderate or dense, perivascular lymphocytic infiltrate with variable vacuolar interface changes and papillary dermal edema. Ten cases were classified into this pattern.

The main features seen in the thrombotic vasculopathy pattern were vascular occlusion by a fibrinoid plug accompanied by no or mild lymphocytic infiltrate and often by epidermal necrosis (Figure 3, C and D). Three cases were classified into this pattern.

As superficial thrombi can be observed in chilblains, cases showing a classic histopathologic aspect of chilblains associated with focal superficial thrombi were classified in the chilblainlike histopathologic pattern.

Two patients had a history of deep vein thrombosis and 1 had a history of pulmonary embolism. Results of the coagulation tests of both patients were normal. Histopathologically, 1 of these 2 patients had a histopathologic pattern of chilblains and the other had a pattern of thrombotic vasculopathy. The comparative characteristics by pattern are available in the Table.

We present detailed histopathologic features of a large series of CBLL developed during the COVID-19 pandemic. Several clinical series have been published reporting CBLL; however, few include a histopathologic evaluation.^17–21  Chilblainlike lesions appear related to COVID-19; indeed, their incidence has surged and then decreased in France as well as in other countries in parallel with the evolution of the pandemic. Thirty-one percent of patients were asymptomatic besides the CBLL, and other patients had one or a few mild general symptoms. However, nasopharyngeal swabs did not result in detection of SARS-CoV-2 RNA, in accordance with other series in which the polymerase chain reaction positivity rate was similarly zero or very low.^19,21–24  Because by convergent estimates, a significant proportion of SARS-CoV-2 infections are asymptomatic,^25–27  CBLL may be a marker of COVID-19 for otherwise asymptomatic or mildly symptomatic patients. The hypothesis that CBLL represent late cutaneous manifestations of COVID-19 or are associated with an early effective immune response might explain the negativity of reverse transcription polymerase chain reaction in all our cases. It is noteworthy that SARS-CoV-2 polymerase chain reactions performed on skin samples are consistently negative in the literature.^14,20  Other hypotheses have been suggested in the literature. It has been argued that CBLL are a manifestation of chilblains due to lockdown.²⁴  This is not tenable in our cases for 2 reasons. Chilblainlike lesions appeared during March 2020 and were collected during a 3-week period concomitantly with the epidemic peak in our region. Secondly, in France, the months of March and April were among the warmest recorded ever, which argues against the role of cold and humid weather, as is the case in classical chilblains. Another hypothesis evoked, without evidence, is that of a concomitant parvovirus B19 epidemic.²⁸  It is also possible that the association between CBLL and SARS-CoV-2 is the product of another confounding factor, not yet identified. Only one study has demonstrated the presence of SARS-CoV-2 by immunohistochemistry and electron microscopy in endothelial cells in CBLL skin biopsies.²¹ 

In our study, we report the existence of 2 histopathologic aspects of CBLL, a chilblainlike histopathologic pattern and a thrombotic vasculopathy pattern. It is currently unknown if these 2 patterns represent 2 different entities, different stages of the same process, or variable expressivity of the same process. The presence of superficial thrombi in chilblains is a known histopathologic feature reported in the literature.²⁹  What was surprising and led us to classify certain cases in the thrombotic vasculopathy pattern was that they showed a much less abundant or even absent inflammatory infiltrate with a less marked perivascular distribution.

Interestingly, 4 patients (31%) presenting with CBLL also had a history of chilblains or Raynaud phenomenon. Preexisting history of chilblains or Raynaud phenomenon could be a contributing factor to CBLL. A history of autoimmune diseases (Crohn disease, ulcerative colitis, and psoriasis) was also common (3 cases; 23%) in the cohort and could represent a predisposing factor.

The first histopathologic pattern of CBLL, which we named the chilblainlike histopathologic pattern, may be linked to the antiviral immune response against SARS-CoV-2. Early control of viral replication by the innate immune system plays a central role in limiting viral spread, shaping the downstream adaptive immune response and thus determining the outcome of infection.³⁰  Specifically, there is evidence that type I interferon (IFN) response (mainly IFN-α and IFN-β) has a strong antiviral effect against other coronaviruses.^31,32  Interestingly, several genetic diseases that present with clinical acral skin lesions resembling chilblains result from type I interferonopathies such as familial chilblain lupus, STING-associated vasculopathy, and Aicardi-Goutières syndrome. Most affected genes represent type I IFN target genes and/or negative regulators of the antiviral IFN-stimulatory DNA response.^33–35  This supports the hypothesis of a link between a strong innate type I IFN antiviral response against SARS-CoV-2 and CBLL. Genetic predisposition factors such as single-nucleotide polymorphisms of type I IFN pathway genes could conceivably also play a role. On the basis of similarity of the histopathologic features with lupus and chilblain lupus seen in several cases, we performed CD123 immunohistochemistry to identify pDCs. In CBLL, pDCs were present in the majority of cases with a variable density, showing a mainly perivascular distribution. Dense perivascular clusters were seen in 1 case. The presence of clusters of pDCs is a well-known diagnostic clue for lupus and can also be seen in chilblain lupus and chilblains.^36,37  Among cell types, pDCs represent a specialized dendritic cell population that express CD123 and, when activated, are capable of producing large quantities of type I IFNs.³⁸  Interestingly, it has been reported that, after infection with coronaviruses SARS-CoV and Middle East respiratory syndrome–related coronavirus, most cell types do not produce a type I IFN antiviral response, except pDCs, which express high levels of IFN-α/β.^31,39–42  Presence of pDCs in tissue could thus represent a link between a strong innate type I IFN antiviral response against SARS-CoV-2 and CBLL.

The second histopathologic pattern of CBLL, which we named the thrombotic vasculopathy pattern, may result from endothelial cell damage. Endothelial dysfunction has been proposed as the cause responsible for excess mortality of COVID-19 in patients with preexisting cardiovascular disease, hypertension, and diabetes.^4,5,43,44  The exact mechanisms of this endothelial dysfunction are unknown. It could result either from direct endothelial cell infection by SARS-CoV-2 or from immune-mediated dysfunction. Severe acute respiratory syndrome coronavirus 2 infects the host cells using the angiotensin-converting enzyme 2 (ACE2) receptor, which, among others, is expressed by endothelial cells.⁴⁵  Direct endothelial cell infection by SARS-CoV-2 has been documented on engineered human blood vessel organoids,⁴⁶  and electron microscopy has shown viral particles in endothelial cells in postmortem analysis of kidney tissue from COVID-19 patients.⁴⁷  Importantly, perturbation of normal ACE2 functions,⁴⁴  increased secretion of cytokines such as IL-1β and IL-6 resulting in diffuse microangiopathy with thrombosis,^44,48  complement-mediated microvascular injury,⁴⁹  and induction of apoptosis/pyroptosis in endothelial cells⁴⁷  could be responsible for COVID-19–induced vasculopathy.

That we did not observe extracutaneous organ involvement may suggest that pathways implicated preferentially target acral skin.

In summary, the same clinical presentation can correspond to 2 histopathologic patterns. In the current state of knowledge, it is difficult to determine whether these are 2 distinct histopathologic aspects or variants of the same process. We have individualized these 2 patterns to highlight the unusual aspect of thrombotic vasculopathy present in some CBLL, in order to alert pathologists to its existence. In light of the mechanisms mentioned above, the chilblainlike histopathologic pattern could result from type I IFN immune reaction to the virus and the pattern of thrombotic vasculopathy could reflect a localized endothelial dysfunction. Reports of livedo reticularis in mild forms of COVID-19 support the latter hypothesis.¹³  Response to viral infection might trigger diverse mechanisms leading to the 2 histopathologic patterns described.

Until the resolution of the current COVID-19 pandemic, pathologists need to be aware of the potential diagnostic pitfalls in distinguishing CBLL, chilblain lupus, and chilblains. Understanding the pathophysiology of CBLL and how it might be related to COVID-19 should advance knowledge of this illness and may lead to specific therapeutics.