This prospective observational cohort defined prolonged SARS-CoV-2 infection (PSI) as a positive rRT-PCR with cycle threshold (Ct) <35 persisting beyond 30 days after microbiological diagnosis with the same viral variant. Of 500 consecutive hematologic malignancy (HM) patients with SARS-CoV-2 infection captured between March 2020 and August 2023, 348 were evaluable for PSI after excluding early deaths (within 30 days) and patients lost to follow-up. PSI occurred in 156 of 348 patients (44.8%). The median duration of rRT-PCR positivity in PSI cases was 58 days (IQR 43–88) compared with a median of 18 days (IQR 13–23) in patients who cleared virus earlier.
The study was performed at Vall d’Hebron University Hospital and affiliated research centers in Barcelona. Clinical data and respiratory samples were collected prospectively. Longitudinal rRT-PCR testing with Ct assessment was scheduled every 7–14 days during the first month and every 14–28 days thereafter until negativization or Ct ≥35 in two consecutive tests together with symptom resolution.
Multiple real-time multiplex RT–PCR platforms were used depending on availability; high-throughput automated systems were also employed. Whole genome sequencing followed the ARTIC v4.1 protocol and COVIDseq (Illumina). Reads were quality trimmed and mapped to the Wuhan-Hu-1 reference (NC_045512.2); variant calling used LoFreq and consensus generation with BCFtools. Minor variants (≥5% allele frequency) in spike epitopes relevant to therapeutic monoclonal antibodies were analyzed. Median depth coverage per sample was 2037× (IQR 791–7944).
In univariable and multivariable analyses, several host, disease and treatment features correlated with PSI. Treatment-related exposures significantly associated with PSI included active therapy with bispecific antibodies (bsAb), anti‑CD20 monoclonal antibodies, Bruton tyrosine kinase inhibitors (BTKi) and immunosuppressive agents for graft-versus-host disease (GvHD). Other significant associations included:
Certain diagnoses had lower PSI rates in univariable analysis (e.g., multiple myeloma and myeloid neoplasms), while lymphoma was the most common underlying disease among PSI patients (61.5%). The authors caution that exposure groups with small counts (for example bsAb n=21; IS for GvHD n=9) render some estimates exploratory.
Among patients with PSI, 56.4% (88/156) exhibited remitting and relapsing symptoms with fluctuating viral loads. During exacerbations, 70.5% of patients experienced aggravated severity of the acute infection. Pulmonary involvement was common during persistent infection: 34% developed pneumopathy, with organizing pneumonia specifically reported. Lower respiratory sampling (bronchoalveolar lavage and pulmonary biopsy) was performed in patients with radiological pneumonia to exclude secondary infections and to characterize pneumopathy.
Early deaths within 30 days (excluded from PSI assessment) were predominantly attributable to COVID-19 (84.5% of early deaths). Those who died early were older, more often unvaccinated, and had higher rates of severe/critical acute infection compared with the evaluable cohort.
PSI had a major impact on oncologic management: 54.5% (85/156) of patients with persistent infection interrupted hematologic treatment and 19.9% (31/156) suspended treatment indefinitely. The authors report substantial morbidity linked to PSI including hospital readmissions and secondary complications noted in the introduction and background context.
Whole genome sequencing identified intra-host viral mutations across the viral genome, predominantly concentrated in the spike gene in most PSI patients. Analyses of minor and consensus variants revealed substitutions associated with reduced susceptibility to anti‑SARS-CoV-2 monoclonal antibodies. Specific substitutions observed after exposure to therapeutic monoclonal antibodies included residues reported in the study for sotrovimab (e.g., E340, R346, K356) and for tixagevimab/cilgavimab (e.g., R346, K444, G446). The authors linked these substitutions to reduced viral susceptibility, highlighting the selective pressure exerted by passive antibody therapies during prolonged infection.
The study assessed mutation counts and diversity normalized by days of positivity; emergence of mutations and Shannon diversity index comparisons were analyzed relative to therapy received. Full original sequencing data are available from the corresponding author upon request.
Descriptive statistics reported medians and interquartile ranges for continuous variables and frequencies for categorical variables. Logistic regression and Wald tests were used to evaluate differences between PSI and non-PSI patients. Two multivariable models were constructed with careful covariate selection and collinearity assessment (maximum VIF reported). Given small sample sizes for some therapies, estimates for those exposures are exploratory. No imputation of missing data was performed. Significance threshold was p < 0.05. Ethical approval was granted by the institutional committee with study identifiers cited.
Limitations explicitly noted by the authors include variable molecular diagnostic platforms across the study period, limited numbers in some exposure subgroups, and the pragmatic nature of real-world testing that could influence sampling intervals.
In this HM cohort, prolonged SARS-CoV-2 infection was frequent and associated with notable clinical morbidity, treatment interruptions and intra-host viral evolution including emergence of substitutions linked to monoclonal antibody resistance. The authors emphasize the clinical need to optimize early antiviral treatment strategies, ensure vaccination and booster coverage where possible, monitor viral clearance longitudinally (including Ct tracking and sequencing in selected cases), and tailor infection control and hematologic care plans for high-risk patients. Specific therapeutic or guideline changes were not prescriptively recommended beyond these articulated needs, and detailed raw data are available from the corresponding author.