In middle March 2020, a 57-year-old Italian man, with a 9-month history of EGPA reported fever, dry cough, anosmia, and headache soon after a household contact with a confirmed COVID-19 case. He did not seek confirmatory test, as commonly occurred in the very first period of the pandemic in Italy, being classified as a probable COVID-19 case. He recovered spontaneously in a few days and resumed his life without limitations. At that time, immunosuppressant therapy for EGPA included low-dose steroids (methylprednisolone 0.15 mg/kg/day) and mycophenolate mofetil (1000 mg per day), and he had received the third 1000 mg rituximab (RTX) infusion in March 2020, 6 days before the probable COVID-19 (the first two infusions were administered 15 days apart, 3 months earlier). The patient had a history of adult-onset asthma and nasal polyps with olfactory impairment. EGPA — diagnosed 9 months earlier due to fever, acute dyspnea, acute kidney injury, and peripheral neuropathy with hypereosinophilia, high serum eosinophilic cationic protein levels, perinuclear-ANCA pattern on indirect immunofluorescence assay, high anti-MPO antibody levels, and increased IgE levels — was in remission and under almost complete symptoms control with only mild paresthesia of the right foot digits.
Almost 6 months later (day 0; 173 days after symptoms onset of probable SARS-CoV-2 infection), before acceding to the hospital outpatient service for EGPA visit in total well-being, the patient underwent a screening test for SARS-CoV-2 on nasopharyngeal swab (NS), by Allplex™ 2019-nCoV (Seegene Technologies Inc; Seoul, South Korea) reverse transcriptase–polymerase chain reaction real-time (RT-PCR), which resulted positive. The patient was confined at home, showed no signs or symptoms of COVID-19, and serial follow-up testing was started. Between day 0 and day 231 (> 7 months), SARS-CoV-2 RNA was invariably detected by RT-PCR on a total of 18 consecutive NSs (Fig. 1). Furthermore, on days 40, 108, and 143, SARS-CoV-2 was detected on NS specimens by using in vitro Vero cell cultures, while culture yielded negative on NS specimens collected on days 185 and 214. Due to the persistent SARS-CoV-2 positivity, mycophenolate mofetil was discontinued on day 101, and RTX infusions were suspended after a fourth administration in June 2020. Based on anecdotal reports, two infusions of high-titer (1:640) COVID-19 convalescent plasma (days 101 and 172) and a 5-day course of remdesivir (days 167–172) were provided [13, 14]. In both cases, the SARS-CoV-2 RT-PCR quantification cycle (Cq) value showed a temporary increase soon after the administration, suggesting a partial reduction of the viral load in response to the treatment, although SARS-CoV-2 continued to be detected on NSs and Cq values decreased again within the subsequent weeks (Fig. 1).
Viral clearance was eventually obtained on day 238, 7 days after receiving the SARS-CoV-2-specific mAbs cocktail bamlanivimab 700 mg and etesevimab 1400 mg (Eli-Lilly). The patient remained completely asymptomatic for the whole period, and blood tests performed on days 101, 167, and 231 did not show meaningful abnormalities, including inflammatory biomarkers within normal values. Notably, no secondary cases of COVID-19 were detected among his close contacts, including relatives, work colleagues, and friends, both in the pre- and post-confinement period (i.e., before and after day 0).
Following the genetic investigation performed on nasopharyngeal samples collected on days 35, 143, and 207, Pangolin v3.1.16 and Nextclade v.1.7.2 assigned all the sequenced samples in the B.1 lineage and 20A clade, respectively, the origin of which is considered linked with the Northern Italian outbreak in February 2020. This lineage is characterized by a synonymous substitution (nt. 3037) in ORF1ab, and by the non-synonymous substitutions P304L (nt. 14408) in ORF1b and D614G (nt. 23403) in the Spike protein (S) . In addition to D614G, known to be associated with a slight increase of transmissibility, several other substitutions were identified in the S locus in all sequenced samples, including (i) T95I, which was reported in several lineages including the variants of concern/interest (VOC/VOI) Delta (B.1.617.2) and Mu (B.1.621), included in vaccine breakthrough infections ; (ii) D138E and H146N in the N-terminal domain (NTD), which were reported in Alpha, Beta and Delta variants with an extremely low frequency (≤ 0.001%) according to GISAID (last accessed on April 29, 2022); (iii) G446V, which was located in the receptor-binding domain (RBD) and was related with in vitro decreased sensitivity to convalescent sera  and monoclonal antibody REGN10987 ; and (iv) N679K, which was described both in the South African C.1.2 and the Brazilian P.1.3 and P.1.4 variants and is located just upstream of the S1/S2 junction, which may enhance Furin cleavage potentially increasing infectivity (Table 1) . Interestingly, while some substitutions in the S protein were constantly detected in all samples within the 7-month time span of infection (e.g., T95I, D138E, H146N, G446V, N679K), several others showed marked heterogeneity in terms of samples where these were identified and of relative frequency (rf) within each sample (Table 1; Fig. 2). As an example, while in the samples of days 143 and 207 a 141–144 LGVY deletion was identified in the S gene with a 98% rf, in the remaining one, it showed an 87% rf (i.e., about 9% of reads with a wild-type sequence were present). This phenomenon could be related to the simultaneous presence of highly similar, but not identical, SARS-CoV-2 subpopulations undergoing micro-evolution during the long-lasting viral carriage.
The Q675H substitution was identified in the last sequenced samples only, showing an rf of about 35% (Table 1). This mutation is located near the polybasic cleavage site at the S1/S2 junction and was reported in several SARS-CoV-2 VOI circulating worldwide . Other micro-evolution events were identified for amino acidic substitutions occurring within the ORF1a and ORF1b (Table 1). Among them, the M85-V86 deletion in ORF1a was not identified in the first available sample; its rf constantly increased from 12.5 to 97% in samples collected on days 316 and 380, respectively. To date, the role of 2 or 3 amino-acid deletions has not been clarified, even if the deletion events seem to induce a structural change in a flexible region of the non-structural protein 1 (NSP1 protein) . The R2613N (originating from a 3 nucleotides mutation in ORF1b) showed an rf of 77% in the first available sample and of 85% in the sample collected on day 143, where a second variant was concomitantly detected at the same position, namely R2613K, although with a lower rf (i.e., 14%). R2613K became the only one detectable on day 207. Likewise, the Q1140K and P2612L alterations, which emerged as subpopulations since day 208 (82%), were detected with a frequency > 90% in ORF1a and in ORF1b on day 316, respectively, providing evidence for an additional intra-host micro-evolution event. Overall, the Q1140K substitution showed a low frequency with only 22 genomes deposited in GISAID (GISAID database accessed with virSEAK JSI’s SARS-CoV-2 tool (https://virseak.bio/virus/), last accessed on April 29, 2022), providing further evidence of an intra-host evolution.
As far as immunological response is concerned, serological status evaluated on days 40 and 101 demonstrated the absence of neutralizing antibodies. Further serological assays to detect IgG antibodies against the SARS-CoV-2 Spike protein and nucleoprotein were performed on days 167 and 231 (LIAISON®SARS-CoV-2 S1/S2 IgG, DiaSorin Inc., USA; ARCHITECT®SARS-CoV-2 N IgG Immunoassay, Abbot, USA) with negative results. These findings are in agreement with previous RTX treatment (a fourth 1000-mg infusion was administered in June 2020, and it led to complete B cell depletion, which was still demonstrated on days 167 and 231).
Cellular immunity to SARS-CoV-2, assessed on day 101 while the patient was under MMF 1000 mg/day and low-dose steroids (MPDN 0.025–0.05 mg/kg/day), was also significantly impaired. Indeed, the patient displayed absence of circulating CD4 + T cells specific for a pool of peptides spanning the spike, membrane, and nucleoprotein proteins, as compared to a non-infected individual and a subject with history of paucisymptomatic COVID-19 (Fig. 3). It should be noted that the patient’s CD4 + T cells also displayed a significantly impaired cytokine production when stimulated with a superantigen like Staphylococcus enterotoxin B (SEB), thus confirming a profound drug-induced immune suppression. Considering optimal EGPA disease control, MMF was suspended. Drug discontinuation led to a partial recovery of T cell functionality. Indeed, when evaluated on days 167 and 231, respectively 2 months and 4 months after MMF discontinuation, circulating CD4 + T cells showed increased ability to produce cytokines following stimulation with SEB. Moreover, we observed an increase in the frequency of SARS-CoV-2 reactive CD4 + T cells (Fig. 3).
As a proxy for innate immune response, potentially triggered by persistent viral replication, we retrospectively collected measurements of inflammatory parameters on days 101, 167, and 231. According to the asymptomatic status of COVID-19 in the patient, values of interleukin (IL)-6, IL-8, ferritin, and C-reactive protein were in the normal range.