Adenovirus Genotypes Associated With Severe Acute Respiratory Infections Outbreak in Children, in Antioquia, Colombia, 2022–2023

Abstract

Background:

Human Adenovirus (HAdV) is a common cause of acute respiratory infections, typically mild in healthy individuals. However, in late 2022, an outbreak of severe acute respiratory infection caused by HAdV emerged among children in Colombia and other countries.

Methods:

We described an HAdV outbreak between February 2022 and April 2023. Children with severe acute respiratory infection and HAdV infection confirmed by polymerase chain reaction were included in 4 institutions in Antioquia, Colombia. Our study investigated the clinical manifestations and circulating HAdV genotypes before, during and after this HAdV outbreak.

Results:

A total of 133 HAdV cases were analyzed, 37 (27.8%) cases were classified as the preoutbreak group, 88 (66.1%) as the outbreak and 8 (6.0%) as the postoutbreak group. Predominant symptoms were fever (87.0%), rhinorrhea (57.1%) and dyspnea (36.8%). The need for intensive care unit admission and supplemental oxygen increased during the outbreak and peaked in the postoutbreak period. Phylogenetic analysis revealed that 71.4% (10/14) of preoutbreak sequences belonged to genotype HAdV-C89, while during the outbreak, 75.6% (28/37) were HAdV-B3. Clinical symptoms did not significantly differ between HAdV-C89 and HAdV-B3 infections, but children infected with HAdV-B3 were significantly older.

Conclusions:

This study highlights the shifting dynamics of HAdV genotypes in children and their epidemiologic impact. The emergence of HAdV-B3 in the post-COVID-19 period contributed to a severe acute respiratory infection outbreak, emphasizing the need for ongoing surveillance.

Human Adenovirus (HAdV) is a common cause of mild respiratory infections in children worldwide.¹ The incidence of severe HAdV cases is higher among immunocompromised individuals and children under 4 years old, with mortality and/or respiratory failure requiring ventilation as high as 30% in these populations.² A better understanding of HAdV epidemiology and clinical variations associated with genotypes may help predict outbreaks, implement control measures and develop vaccines.

HAdV is a nonenveloped, nonsegmented, double-stranded DNA virus with a genome that encodes 3 major capsid proteins (hexon, penton base and fiber).³ Based on these, the viruses are classified into A to G species, with several serotypes associated with different conditions. HAdV-B (serotypes 1, 3, 7, 16, 21) and HAdV-C (serotypes 1, 2, 5, 6) are the most frequently associated with respiratory infections.⁴ Recombination events in HAdV are well-documented and significantly contribute to the emergence of novel genotypes.⁵

The molecular epidemiology of HAdV remains incompletely understood, primarily due to the emergence of new recombinant genotypes.⁶ This genetic variability can result in increased pathogenicity and the ability to cause outbreaks; continuous surveillance and advanced genetic analysis techniques are necessary to keep pace with the rapid evolution of HAdV. Understanding the molecular epidemiology of these viruses can support the development of effective diagnostic tools, vaccines and treatment strategies.²

Antibody neutralization patterns have historically been used to classify serotypes 1–51; however, a genomic approach based on the major capsid proteins has been used recently, and new genotypes have been described for HAdV-B and C (HAdV-B 55, 66, 68, 76, 77, 78, 79, 114, 106 and HAdV-C 57, 89, 104, 108) according to the HAdV Working Group.⁷

HAdV-C89 is a novel species C genotype defined by unique penton base substitutions (A363E, P364 deletion and L154Q), forming a distinct phylogenetic cluster with atypical capsid variability, unlike most HAdV-C strains with limited recombination. These features may impact clinical severity.⁸ In contrast, HAdV-B3, a major cause of respiratory outbreaks, belongs to species B and shows highly conserved hexon, penton base and fiber genes, maintaining strong phylogenetic stability.⁹

Surveillance data on HAdV in South America remains limited, although before the SARS-CoV-2 pandemic, HAdV-B serotypes 3, 7 and 66 were frequently identified as causative agents of respiratory infections.^10,11 The onset of the SARS-CoV-2 pandemic and the subsequent mitigation measures significantly altered the transmission dynamics of various respiratory viruses, including HAdV.¹²

In Colombia, the proportion of HAdV severe acute respiratory infection (SARI) increased from 8.1% in the first semester to 27.1% in the second semester of 2022.¹³ Interestingly, a significant proportion of those patients required admission to the intensive care unit (ICU) and ventilatory support¹⁴; the emergence of a new serotype was a possible explanation for the observed unusual increase in severe HAdV respiratory infections. This study aims to identify potential factors responsible for a SARI outbreak in children caused by HAdV in Antioquia (Colombia) by describing the clinical manifestations and the genotypes circulating during 2022–2023.

METHODS

Sample Collection

From February 2022 to April 2023, SARI cases with polymerase chain reaction (PCR)-confirmed HAdV infection from 4 healthcare institutions in Antioquia, Colombia, were enrolled: San Vicente Foundation Hospital, Medellín and Rionegro, San Juan de Dios Hospital, Yarumal and San Rafael Hospital, Yolombó. A previous study identified 37 HAdV infections in children between February and June 2022.¹⁵

Patients ≤18 years met the case definition if they had SARI within 15 days, required medical observation or hospitalization, and had PCR-confirmed HAdV infection. Patients without medical records were excluded. Parents or guardians’ informed consent was obtained, and children assented. Then, medical staff collected nasopharyngeal swabs, and clinic-epidemiologic data.

Participants were classified by enrollment date: preoutbreak (February 28–June 21, 2022), outbreak (July 8–December 22, 2022) and postoutbreak (January 14–April 6, 2023) (Fig. 1). The San Vicente Foundation Hospital ethics committee approved this study (IRB 022023).

FIGURE 1.

Flow chart of methodology. Flow chart summarizing the study design, case selection, laboratory testing, genotyping and statistical analysis performed in the investigation of HAdV-associated severe acute respiratory infections in children, Antioquia, Colombia, 2022–2023.

Laboratory Testing

Initial laboratory screening of samples was conducted by direct immunofluorescence assay (D3 Ultra direct fluorescent antibody Respiratory Virus Screening identification Kit, Diagnostic Hybrids Inc., Athens, OH) following the manufacturer´s instructions. The molecular testing was conducted at the Genomic One Health Laboratory of the Universidad Nacional of Colombia. Samples collected from February to June 2022 were analyzed using 4 Allplex respiratory panels assays (Seegene Inc., Seoul, South Korea) (https://www.seegene.com/software/seegene_viewer).¹⁵

Samples collected after July 2022 were confirmed with adenovirus-targeted PCR. Genomic DNA extraction from 400 µL of respiratory specimens was done using the automated MagMAX Viral/Pathogen Nucleic Acid Isolation kit (Thermo Fisher), following the manufacturer’s instructions. Amplification and detection of viral DNA were performed with a 7500 fast qPCR system instrument (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA) using iTaq Universal Probes One-Step (Bio-Rad, Berkeley, CA). PCR primers and probe sequences used were as follows: HAdV-panForward 5’-GCCCCAGTGGTCTTACATGCACATC-3’, HAdV-panReverse 5’-GCCACGGTGGGGTTTCTAAACTT-3’ and HAdV-panProbe 5’-FAM-TGCACCAGACCCGGGCTCAGGTACTCCGA-3’.¹⁶ Cycling conditions were as follows: 95 °C for 3 minutes and 40 cycles of 95 °C for 15 seconds, 60 °C for 30 seconds and 65 °C for 30 seconds. PCR products were sequenced using hexon gene-specific primers. Genomic DNA extraction from 400 µL of respiratory specimens was done using the automated MagMAX Viral/Pathogen Nucleic Acid Isolation kit (Thermo Fisher), following the manufacturer’s instructions. To type HAdVs, the hexon gene hypervariable regions 1–6 (HVR1−6) were amplified using nested PCR primers¹⁷ (Table, Supplemental Digital Content 1, https://links.lww.com/INF/G378). Amplified products were separated on 1% agarose gels and purified with the Norgen DNA gel extraction kit (Qiagen, Chatsworth, CA).

Sequencing Library Preparation

Sequencing targeted the HAdV hexon gene. DNA libraries were prepared using the Nextera XT Library Prep protocol (Illumina, San Diego, CA) per the manufacturer’s instructions.

Genomic DNA tagmentation was performed with 5 µL of DNA (10 ng input). Libraries were then amplified using tagmented DNA in a PCR program that added Index 1 (i7) and Index 2 (i5) adapters, with Nextera XT Index Kit v2 Set B (Illumina).

PCR amplicon (library) clean-up was performed with 0.6X Illumina Purification Beads, yielding 30 µL of library in Resuspension Buffer.

Library Normalization and Pooling

Libraries were quantified using Qubit DNA HS reagent (Thermo Fisher Scientific, Waltham, MA), and amplicon size was verified on a High Sensitivity D1000 screen tape Agilent 4150 TapeStation System (Agilent Technologies, Santa Clara, CA). The pooled library concentration was then confirmed to be ~2nM.

Library Denaturing and MiSeq Loading

Denaturation and dilution were performed following the manufacturer’s instructions. The final pool was sequenced on the Illumina MiSeq platform to obtain 150 bp paired-end reads using the MiSeq Reagent Nano Kit v2 (300 cycles, Cat No: MS-103-1003).

Genome Assembly

Raw sequencing data were analyzed for quality using FastQC v0.11.8 (Babraham Bioinformatics, Babraham Institute, Cambridge, United Kingdom).¹⁸ Then, adapter contamination and low-quality reads were trimmed using Trimmomatic v0.39 (Usadel Lab, RWTH Aachen University, Aachen, Germany).¹⁹ Viral genomes were assembled using Spades v3.15 (Center for Algorithmic Biotechnology, St. Petersburg State University, St. Petersburg, Russia)²⁰ and compared with reference sequences obtained from public databases using the Basic Local Alignment Search Tool.²¹ Study sequences were mapped using BWA v0.7.17 (Burrows-Wheeler Aligner, Heng Li, Broad Institute of MIT and Harvard, previously Wellcome Trust Sanger Institute, MA)²² and Samtools v1.15 [Wellcome Sanger Institute (part of the SAM/BAM tools suite), Hinxton, Cambridge, United Kingdom]²³ and the quality parameters were calculated with Qualimap v2.2.2 (Biomedical Genomics Group, Universitat de Barcelona, Barcelona, Spain).²⁴ Sequences with 50% gene (Hexon and Penton) or lower coverage were excluded.

Phylogenetic Analysis

Reference sequences of HAdV were obtained from the National Center for Biotechnology Information virus (NCBI virus, https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/) using the terms “Human Mastadenovirus,” “Serotype” (ie, C), and “genotype” (ie, 2) according to the Human Adenovirus Working Group (http://hadvwg.gmu.edu/) classification. Sequences with a minimum average depth of 8X and a coverage threshold of 70X were included in the phylogenetic analysis. Three datasets were generated for the hexon gene, including species (A–G), genotype B and genotype C. Additionally, one dataset was created for genotypes C2 and C89, and penton gene phylogenetic analysis was performed to differentiate these genotypes.^8,25 All four datasets underwent alignment using the MAFFT software.²⁶ Phylogenetic relationships for each alignment were obtained with the Maximum Likelihood Estimation method from IQ-tree v2.1.2 [Lam-Tung Nguyen and colleagues, Centre for Integrative Bioinformatics Vienna (CIBIV), University of Vienna & Medical University of Vienna, Vienna, Austria]²⁷ and the best-fit substitution model was estimated by Bayesian information criterion using the ModelFinder function of IQ-tree. To assess the robustness of each tree node, a resampling bootstrap of 1000 replicates was done. Bootstrap values were labeled on the tree branches. Visualization and coloring/edition of the resulting phylogenetic tree was conducted using iTOL.²⁸

Statistical Analysis

R software and its package EpiEstim were used to analyze demographic and clinical data.²⁹ The Shapiro–Wilk test assessed normality of continuous data, expressed as mean ± SD if normal or median [interquartile range (IQR)] if nonnormal. Levene’s test evaluated variance homogeneity. One-way ANOVA compared normally distributed samples with equal variance. Fisher exact, Mann–Whitney or Kruskal-Wallis tests assessed differences. All quantifications were duplicated.

RESULTS

Patient Characteristics and Clinical Features

A total of 135 children tested positive for HAdV; 2 were excluded due to unavailable data. Thus, 133 patients met the inclusion criteria and were enrolled between February 2022 and April 2023. During the initial phase (February–May 2022), HAdV-associated acute respiratory infection cases increased slightly, peaking at 15 in May before dropping to 3 in June, defining the preoutbreak period. The outbreak phase started with a sharp monthly rise in cases for 2 months, 24 cases in August and 28 in September 2022. Subsequently, monthly case counts decreased through January 2023, indicating the end of the outbreak. The sample collection was completed in March 2023, with case frequencies remaining stable and within the expected range (Fig. 2). Overall, 37 (27.8%) cases were classified as preoutbreak, 88 (66.1%) as the outbreak and 8 (6.0%) as the postoutbreak (Table 1).

FIGURE 2.

Tracking of HAdV severe acute respiratory infection in Antioquia, February 2022–April 2023. The distribution of various human adenovirus genotypes in Antioquia, Colombia, from February 2022 to March 2023. Areas represent the percentage prevalence of strains obtained in each period: HAdV-C89 (red), HAdV-C1 (green), HAdV-B78 (yellow), HAdV-C5 (purple), HAdV-B3 (blue), HAdV-B7 (orange) and all HAdV SARI cases enrolled (gray).

TABLE 1.

Clinical Characteristics of Children With HAdV Acute Respiratory Infection in Antioquia in the Preoutbreak, Outbreak, and Postoutbreak

Characteristic	Preoutbreak, N = 37*	Outbreak, N = 88*	Postoutbreak, N = 8*	Overall, N = 133*	P value†
Date onset symptoms	February 28, 2022–June 21, 2022	July 8, 2022–December 22, 2022	January 14, 2023–April 6, 2023
Sex					>0.9
Female	17 (46%)	40 (45%)	4 (50%)	61 (46%)
Male	20 (54%)	48 (55%)	4 (50%)	72 (54%)
Age (yr)					0.6
Median (IQR)	2.00 (1.00, 3.00)	2.00 (1.00, 4.00)	1.00 (1.00, 4.00)	2.00 (1.00, 4.00)
Range	0.00, 6.00	0.00, 7.00	0.00, 8.00	0.00, 8.00
Medical record
None	22 (59%)	48 (55%)	4 (50%)	74 (56%)	0.8
Asthma	3 (8.1%)	7 (8.0%)	1 (13%)	11 (8.3%)	0.8
ROBS‡	8 (22%)	7 (8.0%)	0 (0%)	15 (11%)	0.090
Neoplasia	1 (2.7%)	3 (3.4%)	2 (25%)	6 (4.5%)	0.055
Congenital malformations	0 (0%)	15 (17%)	1 (13%)	16 (12%)	0.011
Other§	3 (8.1%)	8 (9.1%)	0 (0%)	11 (8.3%)	>0.9
Diagnosis
Asthmatic crisis	10 (27%)	10 (11%)	1 (13%)	21 (16%)	0.081
Bronchiolitis	3 (8.1%)	14 (16%)	2 (25%)	19 (14%)	0.3
Pneumonia	12 (32%)	25 (28%)	2 (25%)	39 (29%)	0.9
Rhinopharyngitis	8 (22%)	34 (39%)	3 (38%)	45 (34%)	0.2
Sinusitis	2 (5.4%)	2 (2.3%)	0 (0%)	4 (3.0%)	0.7
Tracheitis	1 (2.7%)	0 (0%)	0 (0%)	1 (0.8%)	0.3
Other¶	1 (2.7%)	3 (3.4%)	0 (0%)	4 (3.0%)	>0.9
Antibiotics hospital	14 (38%)	32 (39%)	4 (50%)	50 (39%)	0.8
Intensive care unit	4 (11%)	17 (19%)	5 (63%)	26 (20%)	0.007
Deaths	0 (0%)	3 (3.4%)	0 (0%)	3 (2.3%)	0.6

^*Median (IQR) or Frequency (%).

^†Fisher’s exact test; Kruskal-Wallis rank sum test.

^‡RBOS: recurrent bronchial obstruction syndrome.

^§Other medical records: sickle cell anemia, seizures, chronic kidney disease, Down syndrome.

^¶Other diagnoses: Croup, tonsilitis, otitis.

The most reported symptoms were fever (87.0%, 116/133), rhinorrhea (57.1%; 76/133), dyspnea (36.8%; 49/133), vomiting (18.0%; 24/133), diarrhea (14.3%; 19/133) and adynamia (12.0%; 16/133). Conjunctivitis was not frequently reported (2.2%; 3/133). Most enrolled children (129/133, 96.9%) were hospitalized; the remaining participants were monitored at home, without antibiotics or oxygen therapy during follow-up.

No significant differences were found in sex, age, symptoms, chest radiograph abnormalities or acute-phase reactants (white blood cell count, lymphocytes and C-reactive protein) on the day of initial hospitalization across periods (Table 1 and Table, Supplemental Digital Content 2, https://links.lww.com/INF/G378). Although Table 1 presents overall comparisons across the 3 periods, individual analyses were performed to explore specific trends of interest. Children with congenital malformations were more likely to be affected during the outbreak than preoutbreak (P = 0.015). ICU admission was higher in the postoutbreak than preoutbreak (P = 0.013) and outbreak (P = 0.028).

No patients had evidence of kidney dysfunction on admission (median creatinine 0.29 mg/dL, IQR: 0.24–0.35; data from 26 patients). Liver enzyme data were available for only 15 patients, with alanine aminotransferase and aspartate aminotransferase elevations in 3 children (2 preoutbreak and 1 outbreak). Chest radiographs (n = 99) showed normal imaging in 39.4% (39/99), interstitial pattern (39.4%; 39/99), ground-glass opacities (11.1%; 11/99), ground-glass opacities with reticulation (4.0%; 4/99) and other abnormalities (3.0%; 3/99).

Antibiotics were prescribed in 37.6% (50/133) of the patients, with ampicillin/sulbactam (54.0%; 27/50) and ceftriaxone (28.0%; 14/50) as the most frequently prescribed. None of the patients received antiviral treatment.

Phylogenetic Diversity

Of the 133 patients, 60 had sufficient volume for sequencing (Table, Supplemental Digital Content 3, https://links.lww.com/INF/G378). Of those, 54 (54/133, 40.6%) met the minimum sequencing parameters to be phylogenetically classified: 14 in the preoutbreak (17/37, 37.8%), 37 in the outbreak (37/88, 42%) and 3 in the postoutbreak (3/8, 37.5%) period. Phylogenetic analysis of the hexon gene revealed that all sequences clustered in serotypes B (n = 36) or C (n = 23) (Fig. 3A). Further analysis of the B genotypes identified 29 samples as B3, one as B78, and one as B7. Similarly, analysis of the C genotypes identified 2 and 6 samples as C5 and C1, respectively (Fig. 3B). Additionally, penton gene phylogenetic analysis was performed to differentiate 10 samples close to C2 and C89 genotypes (Figs. 3C and 3D). The phylogenetic analysis demonstrated clear and robust differentiation among adenovirus genotypes, with branch support consistently between 99% and 100% (Fig. 3). These high values strongly validate genotype assignments, particularly providing confidence in differentiating closely related genotypes such as C2 and C89. During the preoutbreak period, the predominant genotype was HAdV-C89 (71.4%, 10/14), while during the outbreak, it was HAdV-B3 (75.6%, 28/37) (Fig. 2).

FIGURE 3.

Serotype and genotype phylogenetic relationships of HAdV. Maximum likelihood trees were constructed using the IQ-Tree V 2.3.3 program with ultrafast bootstrapping (UFBoot) and 1000 replicates. ModelFinder was calculated as the best-fit substitutional model according to Bayesian information criteria. Strains from this study were categorized by symbols corresponding to their respective batches: pentagons, stars and circles represent strains from preoutbreak, outbreak, and postoutbreak, respectively. Red stars correspond to codetection of B3/C1. Each tree was constructed as follows: A: Phylogenetic tree of partial hexon sequence from species A–G HAdV serotypes. B: B genotypes phylogenetic tree of hexon partial sequence. C: C genotypes phylogenetic tree of hexon partial sequence. D: C2 and C89 genotypes phylogenetic tree of full penton sequence. Bootstrap values greater than 80 are shown in the branches.

Clinical Presentation of HAdV-C89 Infection Compared With HAdV-B3

Fever, cough and rhinorrhea were the most frequent symptoms in patients infected by HAdV-C89 and B3. No significant differences were detected in age, sex, comorbidities, symptoms and antibiotic use between those infected by HAdV-C89 and HAdV-B3. While 28.5% (8/29) of patients reported diarrhea in the HAdV-B3 group, none reported that symptom in the HAdV-C89 group (P = 0.086) (Table 2).

TABLE 2.

Comparison of Demographic, Clinical Features, and Outcomes Between Children With HAdV-C89 and B3 Respiratory Infection

Characteristic	C89, N = 10*	B3, N = 29*	Overall, N = 39*	P value†
Sex				0.7
Female	6 (60%)	15 (52%)	21 (54%)
Male	4 (40%)	14 (48%)	18 (46%)
Age (yr)				0.032
Median (IQR)	2.0 (1.00, 2.00)	3.0 (2.00, 4.0)	3.00 (1.50, 4.0)
Range	1.0, 3.0	0.0, 7.0	0.0, 7.0
Medical record
None	7 (70%)	17 (59%)	24 (62%)	0.7
Asthma	1 (10%)	2 (6.9%)	3 (7.7%)	>0.9
ROBS‡	1 (10%)	1 (3.4%)	2 (5.1%)	0.5
Neoplasia	0 (0%)	2 (6.9%)	2 (5.1%)	>0.9
Congenital malformations	0 (0%)	4 (14%)	4 (10%)	0.6
Other§	1 (10%)	3 (10%)	4 (10%)	>0.9
Symptoms
Cough	10 (100%)	26 (90%)	36 (92%)	0.6
Fever	10 (100%)	27 (93%)	37 (95%)	>0.9
Rhinorrhea	8 (80%)	23 (79%)	31 (79%)	>0.9
Dyspnea	3 (30%)	6 (21%)	9 (23%)	0.7
Vomit	3 (30%)	4 (14%)	7 (18%)	0.3
Odynophagia	0 (0%)	2 (6.9%)	2 (5.1%)	>0.9
Conjunctivitis	0 (0%)	1 (3.4%)	1 (2.6%)	>0.9
Headache	0 (0%)	1 (3.4%)	1 (2.6%)	>0.9
Diarrhea	0 (0%)	8 (28%)	8 (21%)	0.086
Adynamia—asthenia	0 (0%)	4 (14%)	4 (11%)	0.6
Seizures	0 (0%)	1 (3.4%)	1 (2.8%)	>0.9
Current diagnosis
Asthmatic crisis	2 (20%)	3 (10%)	5 (13%)	0.6
Bronchiolitis	0 (0%)	1 (3.4%)	1 (2.6%)	>0.9
Pneumonia	5 (50%)	6 (21%)	11 (28%)	0.11
Rhinopharyngitis	3 (30%)	18 (62%)	21 (54%)	0.14
Otitis	0 (0%)	1 (3.4%)	1 (2.6%)	>0.9
Intensive care unit	2 (20%)	1 (3.4%)	3 (7.7%)	0.2
Radiologic findings				0.5
Anormal	3 (30%)	15 (52%)	18 (46%)
Normal	3 (30%)	6 (21%)	9 (23%)
No Image	4 (40%)	8 (28%)	12 (31%)
Antibiotics	5 (50%)	9 (38%)	14 (41%)	0.7
Deaths	0 (0%)	0 (0%)	0 (0%)

^*Median (IQR) or Frequency (%).

^†Fisher´s exact test; Wilcoxon rank sum test.

^‡RBOS: Recurrent Bronchial Obstructive Syndrome.

^§Other medical records: Kidney diseases, Down Syndrome, diabetes, and febrile seizures.

No significant differences were detected in white blood cell, PCR results, onset symptoms or hospitalization duration between those infected with HAdV-C89 and HAdV-B3. However, children infected with HAdV-B3 were significantly older than those infected with HAdV-C89 (P = 0.032).

Coinfections

In samples before the outbreak, coinfection was observed in 78.5% (11/14). The most frequent were Haemophilus influenzae (11/14, 78.5%), Rhinovirus (10/14, 71.4%), Enterovirus (10/14, 71.4%) and Bocavirus (7/14, 50.0%). Five children had HAdV-B3/C1 coinfection, all of them in the outbreak: the mean age was 2 years (IQR: 1–2), 2 had comorbidities (Down syndrome and congenital heart malformation), but none of them were admitted to the ICU.

DISCUSSION

The dynamics of circulating HAdV genotypes between February 2022 and March 2023 were analyzed. An increase in SARI cases was observed between July and December 2022, coinciding with an outbreak in which HAdV B3 emerged as the predominant genotype. Before July 2022, HAdV C89 was the dominant genotype, with no cases of HAdV B3 detected. The increased number of SARI cases during the outbreak could not be attributed to changes in patient selection criteria and was therefore attributed to the emergence of HAdV B3 during the later stages of the SARS-CoV-2 pandemic.

Recent studies from the United States,³⁰ India,³¹ Japan,³² Pakistan³³ and Vietnam³⁴ reported increased SARI cases by HAdV in 2022–2023, with HAdV-B3 as the predominant outbreak genotype. Several factors may explain this rise, including viral characteristics, host immune response and coinfections. First, mutations in the hypervariable region of the hexon gene may enable immune escape in older children. Second, while HAdV is widespread and HAdV-B3 infections typically occur in infancy, our study found that infected children were significantly older. Reduced HAdV-B3 circulation in 2020–2021 may have expanded the susceptible population, delaying primary infections beyond infancy. Additionally, older children may mount a stronger immune response, leading to increased lung inflammation and greater oxygen requirements. Finally, Nguyen et al.³⁴ reported frequent HAdV coinfections with other respiratory viruses, consistent with our findings. Previous research showed that a higher number of pathogens per sample increases the likelihood of severe lower respiratory disease.¹⁵

Reports on the epidemiology of HAdV in Colombia and Latin America are scarce. According to Herrera-Rodríguez et al.³⁵ and Rojas et al.¹⁰ HAdV-B3, B7, C1 and C2 were the main causes of respiratory manifestations due to HAdV between 1997 and 2012, while the circulation of HAdV-C89 has not been previously described in Colombia. Dhingra et al.⁸ identified HAdV-C89 as an HAdV-C2 with a novel penton base sequence, and Tahmasebi et al., HAdV-C89 with an additional mutation in Brazil, both using fecal samples for sequencing. Our results support that HAdV-C89 could cause SARI with similar symptoms to HAdV-B3. Further studies are needed to understand the clinical manifestations of these new strains.

A recent study identified the circulation of HAdV-C, including C89, causing acute respiratory infections between 2019 and 2022.³⁶ Also, other studies reported a predominance of HAdV-C during this period, when measures to prevent COVID-19 transmission were in place.³⁷ These findings suggest that HAdV-C1, C2, C5, and C89 may employ transmission mechanisms beyond respiratory and contact routes to sustain their spread.

Our study contrasts with another recently published study on HAdV in Colombia.³⁸ We found a lower proportion of children requiring ICU stay (19% vs. 97%), a shorter median hospital stay (9 days vs. 18 days) and a lower mortality rate (3.4% vs. 11%). These differences could be due to variations in study design, inclusion criteria and patient populations. Our study included children from four hospitals with different complexities, allowing a wide clinical disease presentation.

Our study has several key strengths. Collecting samples during the transition from strict nonpharmacologic interventions allowed us to explore the microbiologic impact of returning to normal social interactions. During periods of mandatory mask-wearing, social distancing, and school closures, new HAdV genotypes emerged as causes of SARI, while common respiratory HAdVs like HAdV-B3 spread when these measures relaxed or ceased. Additionally, our sequencing strategy using the Respiratory Pathogen ID/AMR Enrichment Panel Kit allowed us to identify HAdV-C89 as the dominant genotype in early 2022, as it provided hexon and penton base gene sequences for classification. We recommend using metagenomic or multiple enrichment panels for respiratory pathogens in situations where HAdV-C is expected, as analyzing hexon and penton base genes is crucial for identifying new recombinant genotypes.

Our study has some limitations. First, including using different approaches for samples collected before the outbreak compared with those collected during and after the outbreak. Nevertheless, the predominance of HAdV-B during and after the outbreak enabled targeted sequencing of the hexon gene, minimizing the risk of misclassification. Second, due to the small number of postoutbreak cases, the statistical power to detect differences in this group is limited. We applied exact tests to account for this constraint, but findings related to the postoutbreak period should be interpreted. Larger postoutbreak surveillance is needed to confirm findings. Finally, the study did not systematically test for viral or bacterial coinfections in the outbreak and postoutbreak groups, which may have influenced disease severity and limited conclusions regarding the isolated effect of specific HAdV genotypes.

In conclusion, SARI cases caused by HAdV increased in Colombia during late 2022. Our findings indicate that HAdV-C89 predominated before the outbreak, while HAdV-B3 became dominant during the outbreak. The novel HAdV-C89 caused SARI with symptoms similar to those of HAdV-B3 infections. The social distancing and other public health measures during the COVID-19 pandemic may have contributed to the emergence of new HAdV genotypes. When those measures were stopped, the emergence of HAdV-B3 led to an outbreak of respiratory illness in children. The postpandemic resurgence of respiratory viruses is a stark reminder of the importance of integrating the lessons learned from COVID-19 into our surveillance and preparedness strategies to build more robust and equitable health systems capable of responding effectively to future outbreaks.