Main findings of the current study are: first, persistent viral replication over months occurs in immunocompromised patients, also in the Omicron era and especially after B cell depletion. Second, persistent infection may lead to slowly progressing pulmonary changes and should be suspected in immunocompromised patients suffering from recurrent fever, cough, dyspnoea and fatigue. Third, viral replication can be limited to or predominantly occur in the lower respiratory tract and therefore diagnostic specimens only from the upper respiratory may miss it. Fourth, clinical symptoms and radiological changes are at least partially reversible under antiviral therapy with prolonged and combined antiviral treatment being more often associated with sustained clinical improvement than short monotherapy.

Whilst prolonged viral replication of SARS-CoV-2 has already been described for previous variants as well as in case reports for Omicron [2, 22, 23], we would like to highlight the potential of a prolonged COVID-19 course in patients with immunosuppression leading to subacute or chronic complications or sequelae. The relevant number of patients in this monocentric study suggests a higher prevalence and probably underdiagnosed facet of the disease, than case reports in the latest literature may show. In the Omicron era, in which mortality rates are lower than under previous variants, morbidity as leading factor has to be taken more seriously, especially in this vulnerable patient collective. Two previous Italian and Japanese case series investigated the combination of (two) prolonged antiviral agents plus a monoclonal antibody in patients with ongoing viral replication and were able to interrupt SARS-CoV-2 replication effectively, even if long-term follow-up data on these patients are still missing [24, 25]. We need further studies to derive the narrowest and still effective therapy regimen.

Unlike previous variants, the Omicron variant of SARS-CoV-2 predominantly replicates in the upper airways [26, 27]. Different from this, we found a shift to the lower respiratory tract in some patients, with loss of detectable viral RNA in the upper respiratory tract. Previous studies have highlighted the possibility of intra-host evolution with emergence of variants especially in immunocompromised hosts that might be adapted to antiviral treatment [28,29,30]. Whether a change in tropism of the virus in our cases was the reason for the observed shift in the replication site is not yet clear and has to be further investigated. Also the emergence of antiviral resistance mechanisms should be taken into account [31].

In immunocompetent hosts, pulmonary inflammation is induced in the first two to three weeks of SARS-CoV-2 infection which eventually leads to scarring and chronic changes thereafter [32]. In contrast, in immunocompromised hosts, our data suggest subacute progressive pulmonary changes by prolonged or ongoing SARS-CoV-2 replication. The mechanism behind the higher frequency of fibrotic-like lung changes in patients with B cell depletion compared to patients with organ transplantation is not yet clear. The more intense and T cell involving immunosuppression in patients with organ transplantation may suppress inflammatory processes, which can lead to evolvement of fibrotic changes. Anti-fibrotic effects have for instance been discussed for mycophenolate mofetil [33]. This would suggest that at least a relevant part of the observed lung changes is linked to immunological effects, not viral replication itself, as seen for COVID-19 in immunocompetent patients [34, 35]. Another explanation might be that patients with B cell-depleting therapies have underlying diseases that predispose to interstitial lung diseases (e.g. ANCA-associated vasculitis, scleroderma) or have therapeutic regimens with drugs/radiation therapy that have pulmonary toxicities rendering the lung more vulnerable to virus-induced changes. Further radiological characterisation is necessary to review this, especially in the light of changes by the emergence of the Omicron variant.

Our study has several strengths. First, we present a relevant number of immunocompromised patients with ongoing viral replication. A significant proportion of the patients in our cohort had follow-up CT scans of the lung making an assessment of the time course of the occurrence and recurrence of lung changes feasible. In addition, many follow-up SARS-CoV-2 PCR test results were available, especially paired samples between lower and upper respiratory tract, enabling a more detailed analysis of the special risk group of immunocompromised patients.

Furthermore, for many patients, long-term follow-up data were available with a median time of 167 days. This is especially important as we observed relapses of SARS-CoV-2 replication, which could be missed if the follow-up time is too short.

Limitations of our study (apart from those inherent to the retrospective observational design) include the absence of a control group. Therefore, the estimation of the relative risk of ongoing replication or the development of chronic lung conditions is not assessable. The same holds true for the impact of antiviral therapy on the replication pattern and the course of fibrotic lung changes. A selection bias probably results from the exclusion of patients that died before giving consent and the recruitment via our ID consultation service. Deaths of all patients who died after giving consent were considered not COVID-19-related. Whether the previous ongoing replication of SARS-CoV-2 played a role in a reduction of health in general and thereby leading to acceleration of comorbid conditions is difficult to extrapolate.

Second, we tried to describe the time course of radiological changes in our patients based on the repeated CT scans. Still, the exact time point of the occurrence of changes is hard to identify because of a varying frequency of radiological imaging.

Third, we only included patients with severe immunosuppression like solid organ transplant recipients or B cell-depleted patients which make up the majority of patients with ongoing SARS-CoV-2 replication in our tertiary care setting. Patients with other types or levels of immunosuppression could also be affected by prolonged SARS-CoV-2-replication.

Fourth, the reporting of symptoms is highly subjective (besides fever). The investigated population at the same time was highly comorbid with multiple other possible causes of fever, cough or fatigue making the identification of COVID-19-related symptoms challenging. Nevertheless, to reduce over-reporting of unrelated symptoms, a thorough individual case review was performed and symptoms were only considered SARS-CoV-2-related, if no other plausible condition was identified.

Our results have implications on daily practice. In patients with substantial immunosuppression (especially B cell depletion) reporting prolonged symptoms like fever, cough or dyspnoea, persistent SARS-CoV-2 infection should be considered.

To exclude a suspected prolonged replication, lower respiratory tract samples are essential, as SARS-CoV-2 replication may be low or absent in upper airways. A high delta of viral load between the lower and the upper respiratory tract could help us to consolidate the diagnosis of ongoing replication of SARS-CoV-2 in the lower respiratory tract.

Continuous detection of a high viral load (Ct-value < 25) in respiratory material especially from the lower respiratory tract should be taken seriously. Further radiological (re-)imaging should be initiated to detect or exclude chronic lung changes, even with an initial unremarkable lung imaging, as these pulmonary changes may develop over time.

As many patients especially after B cell depletion fail to mount an effective immunity against SARS-CoV-2, antiviral therapy is a major part of most therapeutic approaches towards persistent SARS-CoV-2 infection. Therefore, it should be considered at any point in time of infection. Our data suggest that treatment is most effective in combination and over a prolonged period of time. As rebound of infection is possible after discontinuing antiviral treatment, follow-up testing is essential. The effect of antiviral treatment on chronic pulmonary changes is not yet conclusive. However, we identified some cases with a pronounced regression of pulmonary changes in CT scans after initiation of antiviral therapy, which was accompanied by clinical improvement.