SARS-CoV-2 Detection by Digital Polymerase Chain Reaction and Immunohistochemistry in Skin Biopsies from 52 Patients with Different COVID-19-Associated Cutaneous Phenotypes Free

Dermatologists of Lombardy, the first Italian region struck by the COronaVIrus Disease 2019 (COVID-19) due to the severe acute respiratory virus 2 (SARS-CoV-2), were among the first ones to have the opportunity to examine skin lesions of infected patients [1, 2]. Six main clinical phenotypes of COVID-19-associated cutaneous manifestations – i.e., urticarial rash, confluent erythematous-maculopapular-morbilliform rash, papulovesicular exanthem, chilblain-like acral pattern, livedo reticularis/livedo racemosa-like pattern, and purpuric “vasculitic” pattern – have been initially described [3]. In an Italian multicenter study, the two most common presentations were confluent erythematous-maculopapular-morbilliform rash and chilblain-like acral pattern, which accounted for 25.7% and 24.6% of the 187 patients included in the statistical analysis, respectively [4].

The pathophysiology underlying these skin manifestations and, in particular, the role of SARS-CoV-2 in triggering the different clinical phenotypes remain elusive. Moreover, data concerning the presence of the virus in skin samples are controversial [5‒18]. Indeed, a number of studies, mainly carried out on patients with chilblain-like lesions, failed to confirm the presence of SARS-CoV-2 in the skin [5‒11]. In contrast, isolated reports have emerged regarding patients with various COVID-19-associated cutaneous manifestations (maculopapular eruptions [13], leukocytoclastic vasculitis [14], urticarial vasculitis [15], lupus tumidus-like lesions [16] and chilblain-like lesions [17, 18]) in whom SARS-CoV-2 was detected in the skin by different methods – i.e., polymerase chain reaction (PCR) [13, 14, 19], immunohistochemistry [15, 17‒20], or FISH [16]. In order to clarify this still undefined topic and demonstrate the possible presence of SARS-CoV-2 in the skin of patients with COVID-19-associated cutaneous manifestations, we collected samples, clinical, and laboratory data from a large cohort of patients from five Dermatology Units of the Lombardy region (Italy).

Patients with COVID-19-associated cutaneous manifestations who were examined between March 1, 2020, and May 30, 2020, were included in the study. Each participating center was asked to provide data on the basis of the following patient inclusion criteria: (1) an age of 18 years or older, (2) probable or laboratory-confirmed COVID-19, and (3) the presence of COVID-19-related skin manifestations confirmed by an expert dermatologist. A COVID-19 diagnosis was considered to be laboratory-confirmed in the case of a nasopharyngeal swab with a positive result for SARS-CoV-2 RNA or positive serology result for anti-SARS-CoV-2 IgG/IgM antibodies. COVID-19 was considered probable in any patient meeting the clinical criteria (dry cough, fever, dyspnea, the sudden onset of hyposmia or hypogeusia) who had been in close contact with someone with confirmed COVID-19 in the 14 days before symptom onset. A history of new medications in the 15 days before the onset of the skin manifestations was considered an exclusion criterion. The clinical data included sex, age at the time of onset of COVID-19, the presence/absence of comorbidities, cutaneous patterns, the presence/absence of mucous lesions, the duration of skin manifestations, skin-related symptoms, systemic symptoms, the duration of systemic symptoms, the latency between the cutaneous manifestations and systemic symptoms, death, and the severity of COVID-19. Patient data also included results from skin biopsies which were used to identify the presence of SARS-CoV-2, as discussed in detail in online supplementary Material S1 (for all online suppl. material, see https://doi.org/10.1159/000530746).

Systemic symptoms were taken from the charts of hospitalized patients or reported by outpatients and assessed by a physician (a pulmonologist or a specialist in internal/emergency medicine or infectious diseases). The duration of cutaneous manifestations was directly evaluated by a dermatologist in the case of hospitalized patients or reported by outpatients. Each patient was examined at least twice (during the period of skin manifestations and after their resolution). The severity of COVID-19 was classified as asymptomatic, mild (in the presence of fever, cough, and/or gastrointestinal symptoms with no imaging sign of pneumonia), moderate (in the presence of dyspnea and/or radiologic findings of pneumonia), or severe (a need for invasive assisted ventilation, the occurrence of thromboembolic events, or death) [21] and was assessed by considering the worst systemic symptoms over the entire course of the disease, as shown in hospital records or self-reported by outpatients.

The association of SARS-CoV-2 positivity in the skin, as assessed by any of the methods employed, with each clinical phenotype and with each COVID-19 severity category was tested by means of Fisher’s exact test. Differences between patients with and without SARS-CoV-2 positivity in the skin, as assessed by any of the methods employed, in terms of duration of skin manifestations and their latency relative to systemic symptoms were tested by means of Mann-Whitney U test. A p value <0.05, two-sided, was considered statistically significant (IBM SPSS Statistics for Windows, version 26.0, IBM Corp., Armonk, NY, USA). For laboratory details, see online supplementary Material S1.

A total of 52 patients with COVID-19-associated cutaneous manifestations were considered in this study. SARS-CoV-2 positivity in the skin (by any of the investigated methods) was found in 38% of patients. The demographic and clinical features of these patients, as well as the proportion of patients with positive skin staining measured with the various methodologies are summarized in Table 1. The patients were predominantly males (n = 30; 58%), and their median age at the time of the diagnosis of COVID-19 was 57 years (interquartile range, IQR: 44.8–69.8). Of the 52 patients, 7 (11%) developed urticarial rash; 19 (36%) confluent erythematous/maculo-papular/morbilliform rash; 12 (23%) papulovesicular exanthem; 6 (11.5%) a chilblain-like acral pattern; 1 (2%) a livedo reticularis/racemosa-like pattern; and 7 (13%) a purpuric vasculitic pattern. SARS-CoV-2 positivity in the skin was not significantly associated with urticarial rash (p = 0.23), confluent erythematous/maculo-papular/morbilliform rash (p = 0.38), papulovesicular exanthem (p = 0.33), chilblain-like acral pattern (p = 0.19), livedo reticularis/racemosa-like pattern (p = 0.99), or purpuric vasculitic (p = 0.99). No association was found between SARS-CoV-2 cutaneous positivity and asymptomatic (p = 0.36), mild (p = 0.77), moderate (p = 0.99), or severe (p = 0.50) COVID-19. COVID-19 was laboratory confirmed in 38 (73%) patients and was regarded as probable in the remaining 14 (27%). No difference (p = 0.33) was shown in terms of SARS-CoV-2 cutaneous positivity between patients with laboratory-confirmed and clinically suspected COVID-19. The median duration of cutaneous manifestations was 13 days (IQR: 8–23), with no statistically significant difference between cases with (median: 19, IQR: 8.25–40) or without (median: 11, IQR: 7.25–18.25) SARS-CoV-2 positivity in the skin (p = 0.06). Overall, the median latency time between COVID-19 onset and skin manifestations was 21.5 (IQR: 7.5–29) days, ranging between 15 days (IQR: 5.5–23.5) in cases positive for SARS-CoV-2 components (by means of at least one among droplet digital PCR [ddPCR], immunohistochemistry, RNA ISH) and 23.5 (IQR: 5–30) days in negative ones; however, this difference was not statistically significant (p = 0.73). Five patients (10%) were asymptomatic, 20 (38%) had mild disease, 16 (31%) had moderate disease, and 11 (21%) had severe disease.

In 5 patients, the ddPCR approach revealed the presence of SARS-CoV-2 RNA in paraffin-embedded formalin-fixed skin specimens, albeit with minimal viral loads. However, since ddPCR quantifies target nucleic acid sequences by directly enumerating many positive partitioned reactions, without the need for a standard curve, and thus allowing a specific, ultrasensitive and absolute quantitation of nucleic acids – detection limit of 0.1 copies/µL reaction (extracted RNA) – the results obtained clearly indicated the presence of viral RNA in skin samples. Positive patients (n = 5) presented with three different phenotypes: urticarial rash (n = 1); confluent erythematous-maculopapular-morbilliform rash (n = 2); and chilblain-like acral lesions (n = 2). In this group, median latency time between COVID-19 onset and skin manifestations was 28 (IQR: 23–37.5) days and median duration of skin lesions was 18 days (IQR: 12–55). Clinical data of patients who tested positive in ddPCR are summarized in online supplementary Table 1 (online suppl. Material S2).

The immunohistochemical analysis for detection of SARS-CoV-2 nucleocapsid protein turned out positive in 12 cases. Positive staining for the endothelium of small dermal vessel was demonstrated in 3 patients, 2 of whom presented with purpuric “vasculitic” pattern and one with chilblain-like acral lesions. Positive staining for the dermal eccrine sweat glands was detected in 8 patients, 5 of whom presented with confluent erythematous/maculopapular/morbilliform rash and 3 with papulovesicular exanthem. A patient with purpuric “vasculitic” pattern showed positive nucleocapsid staining for the dermal sebaceous glands. In this group, median latency time between COVID-19 onset and skin manifestations was 8.5 days (IQR: 5–15) and median duration of skin lesions was 17.5 days (IQR: 8.5–38.8).

Immunohistochemical SARS-CoV-2 spike protein staining turned out positive in 10 cases. Positivity for the endothelium of small dermal vessel was shown in 3 patients and for the endoluminal portion of the dermal eccrine glomeruli in 7 patients. Positive patients presented with four different phenotypes: confluent erythematous-maculopapular-morbilliform rash (n = 5), chilblain-like acral lesions (n = 3), urticarial rash (n = 1), and papulovesicular exanthem (n = 1). In this group, median latency time between COVID-19 onset and skin manifestations was 23 days (IQR: 12.5–29.3) and median duration of skin lesions was 29 days (IQR: 11.8–52.5).

Immunohistochemistry for detection of ACE-2 receptor protein revealed positive staining for the endothelium of small dermal vessel in 2 patients, 1 of whom presented with confluent erythematous-maculopapular-morbilliform rash and one with chilblain-like acral lesions. In this group, median latency time between COVID-19 onset and skin manifestations was 34 days and median duration of skin lesions was 39 days. Clinical data of patients who tested positive in immunohistochemistry are summarized in online supplementary Table 1. Representative immunohistochemistry analyses are shown in Figure 1.

The RNA ISH assay confirmed SARS-CoV-2 presence in the skin lesions only in one case presenting with chilblain-like acral lesions (online suppl. Table 1). This revealed an epidermal positivity of Grade 1+ in the upper layer of parakeratosis (shown in Fig. 1). The latency time between COVID-19 onset and skin manifestations was 46 days and the duration of skin lesions was 60 days.

COVID-19 patients may present with a heterogeneous spectrum of cutaneous manifestations related to SARS-CoV-2 infection but the influence of SARS-CoV-2 in inducing the different cutaneous phenotypes needs to be clarified. Direct cytopathogenic viral effect (“viral eruption”) and indirect, virus-induced activation of the immune system (“paraviral eruption”) may contribute to different cutaneous presentations. It has been hypothesized that the COVID-19-associated varicella-like papulovesicular eruption represents a classic viral exanthem, following active viremia, while chilblain-like acral lesions are paraviral in their origin, depending on a type I interferon-polarized activation of the immune system and subsequent endothelial dysfunction [22, 23].

In our study, SARS-CoV-2 positivity in the skin, documented through at least one among the methodologies described above, was found in 38% of patients. In particular, the presence of SARS-CoV-2 RNA in lesional skin was detected by means of ddPCR in 5 patients that presented with three distinct phenotypes, i.e., chilblain-like acral lesions (2 cases), erythematous-maculopapular rash (2 cases), and urticarial rash (1 case). Immunohistochemical analysis confirmed the presence of SARS-CoV-2, revealing positive staining for SARS-CoV-2 spike protein in the endothelium of small dermal vessel and in the lining of eccrine glands of these patients. In line with these findings, all the ddPCR-negative cases were negative also on immunohistochemistry for spike protein, except for 5 patients who showed positive staining on the endothelium of dermal vessel. Thus, it may be postulated that in a small portion of patients with COVID-19-associated cutaneous manifestations, COVID-19 can be diagnosed only through skin molecular analysis due to false-negative nasopharyngeal swabs or lack of humoral immunity development leading to negative serology [13]. As proof of this, another patient in our cohort resulted SARS-CoV-2 positive on skin ddPCR and immunohistochemistry but negative on nasopharyngeal swab PCR and serology tests.

No statistically significant association was demonstrated between SARS-CoV-2 presence in the skin and COVID-19 cutaneous phenotypes or severity. Similarly, no statistically significant differences were documented between cases with or without SARS-CoV-2 skin positivity, either in terms of latency time between COVID-19 onset and appearance of COVID-19-related cutaneous manifestation or durations of the latter.

Cases of COVID-19-related cutaneous manifestations positive for SARS-CoV-2 in lesional skin were reported only anecdotally, making challenging a precise estimation of skin positivity for SARS-CoV-2 among these patients. In our cohort of patients with COVID-19-associated cutaneous manifestations, which is up to now the largest one investigated by means of SARS-CoV-2 skin ddPCR and immunohistochemistry, around 40% of patients (5 patients with immunohistochemistry and ddPCR and 15 patients with immunohistochemistry only) tested positive for SARS-CoV-2, suggesting that the virus spreads to the skin only in a minority of COVID-19 patients and that in most cases the pathophysiology of the cutaneous manifestations is “paraviral.” Based on the endothelial positivity for SARS-CoV-2 on immunohistochemistry in eight of our cases and in those published in the literature [15, 17‒20], it must be assumed that the viral spread to the skin occurs through the circulatory stream. However, viral detection in the skin was not associated with a distinct cutaneous phenotype in our series, making it conceivable that it mainly depends on the viral load and the effectiveness of the immune response – either humoral or cell-mediated – against the virus. The role of cytokine-driven inflammation, which plays a crucial part at the systemic level leading to the so-called cytokine storm [24], remains to be explored in the pathogenesis of skin lesions. Only 2 of our 7 patients with vasculitic lesions had presence of SARS-CoV-2 in the endothelium. Considering that the virus has been observed at the endothelial level in the skin, one would expect a higher frequency of SARS-CoV-2 positive vasculitis manifestations. Instead, reports of vasculitis with virus presence documented at the cutaneous level are only anecdotal, including two cases of urticarial vasculitis and a case of leukocytoclastic vasculitis, in which SARS-CoV-2 was detected by means of immunohistochemistry and PCR, respectively [14, 15].

Immunohistochemistry seemed to be the most sensitive method for viral detection, particularly anti-nucleocapsid immunohistochemistry, this being in line with the superior sensitivity of plasma/serum anti-nucleocapsid antibodies relative to anti-spike protein antibodies for the laboratory diagnosis of early infection [25]. The integration of immunohistochemical staining for nucleocapsid, spike, and ACE-2 antigens allowed us to demonstrate viral presence in the skin in 20 out of 52 patients. Interestingly, immunohistochemistry appeared to be more sensitive than ddPCR in our cohort of patients. Discordant immunohistochemical and RNA detection findings have been attributed to the presence of cleaved viral proteins [20]. On the other hand, the RNAscope ISH positivity gave us the possibility of a direct visualization of the virus while retaining tissue morphology, a feature that is lost in methods such as ddPCR. We speculate that ddPCR could have more sensitively detected viral RNA in the skin if performed at specific time points. The presence of viral particles in the epithelium on ISH confirms the fact that SARS-CoV-2 can be found either in the dermis or in the epidermis of affected individuals and that the virus may disseminate to the skin via blood vessels [26]. Positive immunohistochemical staining in sweat glands confirms the findings by Recalcati et al. [27] and Liu et al. [28], who detected SARS-CoV-2 in sweat ducts and hypothesized sweat as a possible route of transmission of SARS-CoV-2.

Key strengths of the present study are the high number of cases investigated as compared to the literature and the concurrent use of different methods, including ddPCR, RNAscope, and immunohistochemistry, to demonstrate the presence of SARS-CoV-2 in lesional skin. Indeed, only few studies used more than one method to detect the virus in lesions of patients with COVID-19-associated cutaneous manifestations [19, 20].

Another strength is the use of both positive (placenta and lungs, see online suppl. Fig. 1 in online suppl. Material S2) and negative (basal cell carcinoma and psoriasis) controls for viral detection by immunohistochemistry. One limitation of the study may consist in the possible contributory role of drugs, especially for rashes with a considerably delayed onset respective to that of COVID-19, i.e., exceeding the time period with no new drugs established as eligibility criterion.

In conclusion, the detection of SARS-CoV-2 only in 38% of our skin samples, regardless of cutaneous phenotype, is noteworthy and suggests that in most cases the pathogenesis of cutaneous lesions may be the result of an immune-mediated inflammatory response rather than a direct cytopathic effect of the virus. Moreover, lack of an efficient immune response against the virus, or higher viral loads perse, might represent critical factors for the cutaneous spread of SARS-CoV-2.

SARS-CoV-2 cutaneous persistence may depend on timing of skin lesions, viral load, and immune response.

This study protocol was reviewed and approved by IRB (Ethics Committee Milan Area 2) of the principal investigator center (Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan). Protocol number: 464_2020. The study was conducted in accordance with the Declaration of Helsinki and all of the participants enrolled in the study gave their written informed consent.

The authors have no conflicts of interest to declare.

This research was partially supported by Italian Ministry of Health (Ricerca Corrente 2023), Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan (Italy).

Conceptualization: Prof. Angelo Valerio Marzano, Dr. Chiara Moltrasio, Dr. Giovanni Genovese, Prof. Marco De Andrea, Dr. Valentina Caneparo, Dr. Pamela Vezzoli, Dr. Denise Morotti, Prof. Paolo Sena, Dr. Marina Venturini, Dr. Simonetta Battocchio, Dr. Valentina Caputo, Dr. Nathalie Rizzo, Dr. Carlo Alberto Maronese, Dr. Luigia Venegoni, Dr. Francesca Boggio, Prof. Franco Rongioletti, Prof. PierGiacomo Calzavara-Pinton, and Prof. Emilio Berti; formal analysis: Dr. Chiara Moltrasio, Prof. Marco De Andrea, Dr. Valeria Caneparo, Dr. Denise Morotti, and Dr. Luigia Venegoni; investigation: Prof. Angelo Valerio Marzano, Dr. Chiara Moltrasio, Dr. Giovanni Genovese, Prof. Marco De Andrea, Dr. Valentina Caneparo, Dr. Pamela Vezzoli, Dr. Denise Morotti, Prof. Paolo Sena, Dr. Marina Venturini, Dr. Simonetta Battocchio, Dr. Valentina Caputo, Dr. Nathalie Rizzo, Dr. Carlo Alberto Maronese, Dr. Luigia Venegoni, Dr. Francesca Boggio, Prof. Franco Rongioletti, Prof. PierGiacomo Calzavara-Pinton, and Prof. Emilio Berti; resources: Prof. Angelo Valerio Marzano, Dr. Chiara Moltrasio, Dr. Giovanni Genovese, Prof. Marco De Andrea, Dr. Valeria Caneparo, Dr. Denise Morotti, and Dr. Luigia Venegoni; data curation: Dr. Chiara Moltrasio, Dr. Giovanni Genovese, and Prof. Angelo Valerio Marzano; writing – original draft preparation: Prof. Angelo Valerio Marzano, Dr. Chiara Moltrasio, and Dr. Giovanni Genovese; writing – review and editing: Prof. Angelo Valerio Marzano, Dr. Chiara Moltrasio, Dr. Giovanni Genovese, and Dr. Carlo Alberto Maronese; supervision: Prof. Angelo Valerio Marzano; project administration: Prof. Angelo Valerio Marzano. All authors have read and agreed to the published version of the manuscript.

The data that support the findings of this study are not publicly available due to their containing information could compromise the privacy of research participants but are available from the corresponding author Angelo Valerio Marzano and the co-author Chiara Moltrasio.