We conducted a study of severe adenovirus infection in children in National Taiwan University Hospital (NTUH) and Chang Gung Memorial Hospital, two medical centers located in northern Taiwan. From November 2010 to June 2011, children with severe adenovirus infections were included in this study. Severe adenovirus infection was defined as children with laboratory-confirmed adenovirus infection by viral culture or polymerase chain reaction (PCR) plus the requirement of intensive care. Viral culture was performed routinely in most children hospitalized for suspicious viral infections in these two medical centers. Real time PCR was performed optionally according to the primary care physician’s clinical suspicion. Hospital-acquired adenovirus infection that was defined as the onset of adenovirus infections occurred 6 days after hospitalization according to the median duration (about 6 days) of incubation period [3].

This study was conducted in Taiwan only, not outside of this country of residence, and the Institutional Review Board approval was obtained form Chang Gung Memorial Hospital (No. 100-2518B). This was a retrospective study without intervention or obtaining clinical specimens and all the data were analyzed anonymously, so informed consent was waived. The waiving of informed consent was also approved by the Institutional Review Board of Chang Gung Memorial Hospital. Every patient had written informed consent for that their data could be used for research under the regulation of the Taiwan law although we did not obtain informed consent for this specific study.

Patients’ demographic data, clinical manifestations, intensive care requirement, laboratory results, images, diagnosis, complications and sequelae were collected and analyzed. Respiratory failure was defined as the need of using positive pressure ventilation, which included both invasive modes (e.g., pressure-control ventilation, high frequency oscillatory ventilation) and noninvasive modes (e.g., bilevel positive airway pressure, or continuous positive airway pressure). Acute respiratory distress syndrome (ARDS) was diagnosed if the presentations fulfilled four criteria: 1. acute onset, 2. PaO2/FiO2 less than 200 mmHg, 3. bilateral infiltrates seen on chest radiographs and 4. pulmonary wedge pressure less than 18 mmHg or no clinical evidence of left atrial hypertension, which were defined by American-European Consensus Conference Criteria for Acute Respiratory Distress Syndrome [10]. Diagnosis of disseminated intravascular coagulation (DIC) was based on laboratory data after evaluation with the scoring system proposed by International Society on Thrombosis and Haemostasis [11].

Hexon gene sequencing was used for serotyping. DNA extraction was performed by using a MagNA Pure LC Total Nucleic Acid Isolation kit (Roche, Germany). For molecular typing, primary PCR primer sets were 5′-TACAACATYGGCTACCAGGG-3′ and 5′-GAGAASGGBGTRCGSAGGTA-3′. Nested PCR was performed with outer PCR products, using primers of 5′-AACTTCCAGCCYATGAG-3′ and 5′-GGRTCCACCTCRAARGTC-3′. The HAdV positive PCR products were purified by gel extraction and were then sequenced using a BigDye Terminator Ready Reaction Cycle Sequencing Kit and an automated sequencer ABI 3730 (Applied Biosystems, Foster City, CA, USA). Obtained sequences were compared with reference sequences of adenoviruses in GenBank using BLAST and the serotypes were determined if the sequences yielded an identity score≧90%.

Because serotype 7 accounted for most cases and deaths in this study, we constructed the phylogenetic tree of adenovirus serotype 7 from 11 cases of severe adenovirus infection and from 3 patients of non-severe adenovirus. The nucleotide sequence homology was inferred from the identity scores obtained using the BLAST program (National Center for Biotechnology Information, Bethesda, MD). A phylogenetic dendrogram was constructed using the neighbor-joining method with the MEGA program 4.0 (Sudnir Kumar, Arizona State University) and the reliability of the tree was estimated with 500 bootstrap pseudo-replicates.

Totally, 45 patients were included. The genotype distribution (Figure 1A) was serotype 7 for 22 cases (49%), serotype 3 for 19 cases (42%) and serotype 2 for 4 cases (9%). This epidemic was initially associated with type 3 and later with type 7 (Figure 1B). Median age (range) was 2.75 years (0.08–15.43 years), and 87% (39/45) of the patients were below 5 years of age. Male to female ratio was 1.65 (28 to 17). More than half of the patients had major underling diseases (Table 1) and both neurological diseases and respiratory diseases played important roles. Of these 45 patients, 19 (42%) were bed-ridden, 21 (47%) had contact and/or cluster history and 11 (24%) had hospital-acquired adenovirus infection.

All patients were febrile. The median (range) peak temperature was 39.9°C (38.0–42.0). 95% of patients experienced fever higher than 39°C and 50% experienced fever higher than 40°C. After 10 fatal cases were excluded, the median (range) duration of fever was 10 days (2–30), 69% (24/35) had fever for equal to or over 7 days and 29% (10/35) had fever for equal to or over 14 days. In addition to fever, every patient presented with dyspnea and productive cough. Non-exudative pharyngotonsillitis (85%) and coryza (68%) were also common. 21 patients (47%) had gastrointestinal symptoms including vomiting, diarrhea or abdominal pain. Seizure attack was noticed in 10 patients (22%). Conjunctivitis (7%) and skin rash (9%) were less common.

The first chest X-ray taken in most patients revealed increased lung infiltration and/or consolidation bilaterally (Figure 2A). With disease progression, consolidation was more often observed and more extensive (Figure 2B). Bilateral whiteout was seen in 8 (15%) patients (Figure 2C). No cavitating or necrotizing lesion was noted.

The major diagnosis was pneumonia in 40 (89%) patients and bronchopneumonia in 5 (11%) patients. Encephalitis was diagnosed in 7 (16%) patients by pediatric neurologists according to clinical and electroencephalography findings.

The median hospital stay was 16 days and the median ICU stay was 7 days (Table 2). Respiratory failure with the need of mechanical ventilation and hypotension were the most important indications of ICU care. The median duration of mechanical ventilator use was 12 days while the median duration of inotropic agent use was 4 days. Eleven (24%) patients developed ARDS, high frequency oscillatory ventilation (HFOV) was used in 12 (27%) patients and extracorporeal membrane oxygenation (ECMO) was applied in 6 (13%) patients. Definite or possible DIC was diagnosed in 29 patients because both fibrin degradation product (FDP) and D-dimer noticeably elevated in these patients.

Chest ultrasonography was used to verify the presence of pleural effusion and was done in 26 patients. Pleural effusion was detected in 85% (22/26) of patients. Pleurocentesis was performed in 16 patients. Chest tube was inserted in 4 patients and median chest tube placement time was 8 days (8–23). Two patients underwent video-assisted thoracoscopic surgery (VATS) but no empyema was noted.

Samples collected between fever onset and 8 days after defervescence were included for analysis. This time interval was chosen because we found that viral load at 8 days after defervescence was still high in most patients. If PCR was taken as standard, adenovirus was more easily isolated from throat swabs (91%), sputum (88%) and rectal swabs (86%) but with relative difficultly from pleural effusion. The positive isolation rate of pleural effusion was only 39%, although the geometric mean (range) of viral load of pleural effusion was 1.5×10⁹ copies/ml (1.7×10⁵–2.8×10¹⁰). Lumbar puncture was performed in 9 children if central nervous system involvement was suspected. Cerebrospinal fluid (CSF) culture for both bacteria and viruses was negative in all 9 samples and there was no pleocytosis, elevated protein or decreased glucose in all CSF samples.

Co-infection played a minor role in severe adenovirus infection. There was no definite bacterial co-infection. All patients had negative blood and negative urine culture results for bacteria. Urine pneumococcal antigen test was all negative when it was performed in 29 patients. Possible Mycoplasma pneumoniae co-infection was diagnosed in 4 patients according to positive PCR or positive IgM for Mycoplasma pneumoniae.

Respiratory virus co-infection was also rare although the study period compassed the respiratory syncytial virus, influenza virus and parainfluenza virus circulating season in Taiwan. Viral antigen tests and PCR were performed frequently. There was only one case with positive influenza A PCR test. One throat swab yielded adenovirus and parainfluenza virus simultaneously. Only one patient had positive sputum respiratory syncytial virus antigen and the other one had positive Chlamydia antigen, respectively. 21 patients (47%) had gastrointestinal symptoms. However, only 2 patients had positive stool norovirus antigen and there was no positive stool rotavirus antigen.

Clinical laboratory data were collected between fever onset and 8 days after defervescence for analysis; Table 3 shows the results. Pleural effusion was grossly bloody in 44% (7/16) of patients and had extremely high lactate dehydrogenase (LDH), elevated protein level but normal glucose (Table 3) and pleocytosis (the median white cell count 1522/µL and the median lymphocyte percentage was 61%). Leukopenia and/or thrombocytopenia were seen in more than 60% of patients. Nevertheless, during serial follow up, both leukocytosis and leukopenia could be detected in the same patient. Left shift (>5% proportion of immature neutrophil cells) [12] was noted in 42% of patients. Leukocytosis with left shift was usually noted in the early course while leucopenia and thrombocytopenia were most often found when disease severity progressed. Like pleural effusion, extremely high serum LDH was noted. Liver enzyme elevation was more prominent for aspartate aminotransferase (AST). Elevated creatinine kinase, hypoalbuminemia and hyponatremia were also characteristic.

10 (22%) patients died and they all had major underlying diseases. 7 (70%) fatal cases were infected with serotype 7 (Figure 1A). The median duration from disease onset to death was 25 days (range from 4 to 71 days). Most deaths were associated with acute respiratory distress syndrome.

Complication occurred in 24%. Except 1 with pneumothorax and 3 with asthma attacks, 6 patients had nosocomial infection (including ventilator associated pneumonia, infected pressure sore, and central venous catheter associated infection), 4 patients had ECMO-related complications, including thrombus and aneurysm formation, and encephalopathy after prolonged sedation was diagnosed in 2 patients.

If survived, 77% (27/35) had full recovery. 4 patients had neurological sequelae, including epilepsy, muscle weakness and developmental regression and another 4 patients had respiratory sequelae: 2 of them were bronchiolitis obliterans and the other 2 needed home BiPAP use.

We compared the hexon sequences of 11 severe adenovirus serotype 7 infection samples and 6 non-severe adenovirus serotype 7 infection samples. They are all very similar. Only two samples had one and three different nucleotides, respectively. After translated to amino acid, we only observed one sample to have one amino acid change (Histidine to Asparagine). Phylogenetic tree was presented in Figure 3. Our serotype 7 sequences shared 100% homology with the strains from America, Japan, Korea and China, and one of them, human adenovirus 7 strain 0901/HZ/ShX/CHN/2009 (GenBank GU230989), caused a severe lower respiratory tract disease outbreak in infants in Shaanxi Province, China recently [7].