The study involved five regional pediatric units (Prince of Wales Hospital, United Christian Hospital, Princess Margaret Hospital, Queen Elizabeth Hospital, and Pamela Youde Nethersole Eastern Hospital) that provided continuous follow-up of 47 pediatrics patients (21 girls, 26 boys; age range, 1.5-17 years; median age, 13.6 years) with serologically confirmed SARS. These patients had undergone a surveillance thin-section CT examination 6 months after diagnosis. All these patients, although asymptomatic at rest, were found to have reduced oxygen consumption at peak exercise (defined as a respiratory exchange ratio > 1.0, heart rate > 85% of the age-predicted value, or when the child became exhausted) compared with healthy control subjects based on peak oxygen consumption (Vo₂ measured in liters per minute) running test.

Sixteen of the 47 patients had abnormal thin-section CT findings. All patients with or without abnormal CT findings were clinically followed up once every 2 months, and the CT findings did not affect the patients' management. These 16 adolescents (10 girls and six boys; age range, 10-17 years; median age, 14.4 years) were recruited for an extended 12-month follow-up thin-section CT examination due to parental concern. All subjects were asymptomatic at rest at the time of the CT examination and were free from any acute upper respiratory tract infection or illness for at least 2 weeks before the CT study. The study was approved by the institutional ethics review board of the Chinese University of Hong Kong. Written informed consent was obtained from parents.

Thin-section CT was performed from the lung apices to the diaphragm using a low-radiation-dose technique (50-80 mAs) [7, 8]. Scanning was performed on an MDCT unit (LightSpeed 16, GE Healthcare) using a high-spatial-frequency reconstruction algorithm with a 1.25-mm section thickness and 7-mm gap. The subjects were scanned in a supine position during breath-holding at full inspiration and at maximal expiration, except nine patients who were too young to follow the instructions; in those patients, only inspiratory images were obtained. All of these patients did not have an abnormality detected on thin-section CT 6 months after SARS diagnosis. The images were photographed at conventional lung window settings (window level, -700 H; window width, 1,000-1,500 H).

All CT images were reviewed independently by two pediatric radiologists for the presence and distribution of residual CT abnormalities. The reviewers were aware of the SARS status of the subjects but were not aware of the previous CT findings of the patients at 6 months and were blinded to the clinical information and pulmonary status of the subjects. Each segment of the lung was identified and assessed specifically for the presence or absence of ground-glass opacification, reticular pattern, nodules, parenchymal scar, bronchial dilatation, bronchial wall thickening, and air trapping. Ground-glass opacification was defined as increased lung parenchyma attenuation that did not obscure the underlying vascular architecture [9]. Reticular pattern was diagnosed when either coarse linear or curvilinear opacities or fine subpleural reticulation without substantial ground-glass opacities was present. The presence of a parenchymal band, irregular interface, or architectural distortion was considered as a positive sign of parenchymal scar [10]. Bronchiectasis was diagnosed when the bronchi appeared larger than the accompanying pulmonary arteries [11]. Bronchial wall thickening was described when the bronchial walls were seen as discrete structures in the distal third of the lung parenchyma. The diagnosis of focal air trapping was based on the presence of a focal area of abnormally low attenuation within the lung parenchyma that was present on the inspiratory scans or that became more exaggerated on expiratory scans [12].

The area of the abnormalities on thin-section CT was also evaluated. The percentage of abnormalities over the lung area was traced on each thin-section image using a workstation (Advantage 4.2 [GE Healthcare] for Windows [Microsoft]). The percentage of abnormal area and volume over the total area and volume of lung covered in the examination (note a 7-mm gap between consecutive slices) was calculated. A radiologist would grade the severity of the disease as follows: 0% involvement was considered normal; less than 30% involvement, mild; 30-70% involvement, moderate; or more than 70% involvement, severe.

Two weeks after review of the initial film package, the two radiologists reviewed the 6-month thin-section CT studies in a similar manner with the names of the patients hidden. Finally, the two sets of films were compared and a final consensus was reached whether there was progression or regression of the CT changes.

Clinical information about the patients during the course of the illness was retrieved from their records for correlation. The data included demographic characteristics, clinical symptoms, duration of fever and hospitalization, details of treatment, ICU admission, and blood test results. The serial chest radiographs of each child during their acute illness were also reviewed 3 weeks after interpreting the CT images. All of the radiographic examinations were retrospectively reviewed independently by three radiologists on separate settings and without knowledge of the thin-section CT findings.

For each patient, the radiograph that showed the most extensive disease involvement was chosen for assessment of the distribution and percentage involvement of parenchymal abnormalities. The number of regions with consolidative changes was counted, and the anatomic distribution was recorded. Each lung was divided into three zones: upper, middle, and lower. Each zone was considered to span one third of the craniocaudal distance of the lung on the frontal radiograph. The extent of parenchymal abnormalities on each radiograph was assessed visually to estimate the percentage of area occupied in each zone on each side. The percentage involvement of the six lung zones was then averaged to obtain the overall mean percentage of involvement. This method has been used in an earlier study of adults patients at the same institution [13]. All the serial chest radiographs of the patients were assessed, and the number of days from onset of fever to complete resolution of radiographic opacification was recorded for each patient.

Pulmonary function tests were performed in 38 of the 47 patients at 6 and 12 months after diagnosis. The remaining nine patients were too young to cooperate for the lung function tests. For those 16 patients with CT performed at the 12-month follow-up, the tests were performed within the same week of the CT examination. The tests were performed by a team of supervised technicians according to the recommended standard [14]. Forced expiratory volume in 1 sec (FEV₁) and forced vital capacity (FVC) were measured with spirometry (Pulmonary Function System with BreezeSuite Software, Medical Graphics). The total lung capacity (TLC) was measured by body plethysmography (MedGraphics Elite Series Plethysmography, Medical Graphics). The pulmonary function results were expressed as percentages of the predicted normal values. All subjects were assessed for the presence of any obstructive abnormality (defined by FEV₁/FVC ratio < 80% predicted) or restrictive abnormality (defined as TLC < 80% predicted).

The statistical analysis was based on all 47 patients who underwent thin-section CT at the 6-month follow-up, including those 16 patients with repeated CT examination at the 12-month follow-up. They were divided into two groups: patients with normal findings on either the first CT scan at 6 months or the subsequent CT scan at 12 months were classified as group 1 and patients with persistent abnormal findings on both of the examinations were referred to as group 2.

The clinical data, extent of radiographic opacification, and pulmonary function test results of group 1 and group 2 were analyzed. The results were expressed as medians with an interquartile range. Potential risk factors for persistent CT abnormalities at 12 months were first evaluated individually using the chi-square test or Mann-Whitney test. Risk factors with a p value of less than 0.25 were then analyzed by multivariate logistic regression analysis using a forward stepwise selection strategy.

The lung function test results of 38 children at 6- and 12-month follow-up were compared using the paired Student's t test. The results were analyzed with the Statistical Package for Social Science (SPSS version 11.0) for Windows (Microsoft). Two-tailed probability values of less than 0.05 were considered significant.

Sixteen of the 47 patients had abnormal findings on thin-section CT at 6-month follow-up. Seven children (44%) had residual ground-glass opacification, and 11 children (69%) had air trapping. Three patients had both features of residual ground-glass opacification and air trapping. All three of these patients required oxygen supplement (6-13 days), and two required ventilatory support (4-6 days) during the acute course of the disease due to acute respiratory distress. Six patients (38%) had a small parenchymal scar, of which three had associated ground-glass opacification and two had associated air trapping. Only one patient in this group had a scar as the sole abnormal feature on thin-section CT. The CT features at 6 months and the number of children involved are summarized as Table 1.

View Larger Version

TABLE 1: Distribution and Classification of Thin-Section CT Abnormalities 6 and 12 Months After the Diagnosis of Severe Acute Respiratory Syndrome in a Group of 47 Pediatric Patients

Ground-glass opacification and reticular pattern—All seven patients with ground-glass opacification at 6 months after SARS diagnosis showed either partial or complete resolution at 12 months. Two patients had complete resolution of the ground-glass opacification (overall extent at 6 months was 5.8% and 6.7%, respectively). The ground-glass opacification resolved in three patients, but there were new reticular opacities in the previous area of abnormality, which constituted < 10% of the overall lung volume covered in the examination (Figs. 1A, 1B). Only one of these three patients required oxygen supplement for 6 days during the acute illness, and none of them received mechanical ventilation.

Two patients had persistent residual ground-glass opacities (Fig. 2) that were non-confluent and scattered, although the extent was reduced. These two patients required ICU care and received mechanical ventilation for 4-6 days during the active disease stage and had the largest extent of ground-glass opacification at 6 months among all patients (30% and 20% of overall lung volume covered). The extent of ground-glass opacification of these two patients decreased to 16.7% and 9.2%, respectively, at 12 months.

Air trapping—Air trapping remained similar in extent in nine (82%) of 11 patients, which ranged from 2.5% to 62%. The remaining two (18%) of 11 patients showed a slight increase in the extent of air trapping in the same involved segments (extent increased from 2.5% and 35% at 6-month follow-up to 5.8% and 45%, respectively, at 12-month follow-up). Of those 11 patients with air trapping, three had all lobes involved. The other eight patients had predominantly bilateral lower lobe involvement. All of the areas of focal air trapping were subsegmental and predominantly subpleural (Fig. 3). Four patients had only small focal areas of air trapping that constituted less than 5% of the total lung area.

Parenchymal scar—The parenchymal scars were unchanged in all six patients. All these scars were small (1-4 cm in maximal extent) (Fig. 4) and were not associated with traction bronchiectasis or volume loss. Bronchiectasis, bronchial wall thickening, and nodules were not shown in this cohort.

Most patients had more than one abnormal CT feature. The CT features at 12 months and the number of children involved are summarized as Table 1. A summary of the serial changes between the two CT examinations is given in Table 2.

View Larger Version

TABLE 2: Thin-Section CT Findings of 16 Pediatric Patients 6 and 12 Months After the Diagnosis of Severe Acute Respiratory Syndrome

There was no significant discrepancy between the independent grading of abnormalities on the two CT examinations by the two radiologists about air trapping, residual ground-glass opacification, and parenchymal scar. Reticular opacities were missed in one patient by one radiologist initially and were diagnosed to be present during panel review. For all abnormalities, variation of the measured area was less than 10%.

There were significant intergroup differences (p < 0.05) observed in age distribution, requirement of oxygen supplementation, lowest lymphocyte count and duration of abnormal lymphocyte count during the acute course of the disease, duration of use of ribavirin (Virazole, ICN Pharmaceuticals), requirement for steroid therapy (hydrocortisone or methylprednisolone), overall percentage of opacification on acute chest radiograph, and number of days taken for complete radiographic resolution. A summary of these data is given in Table 3.

View Larger Version

TABLE 3: Significant Clinical Parameters of Subjects' Characteristics Stratified According to Thin-Section CT Results at 6 and 12 Months After Diagnosis of Severe Acute Respiratory Syndrome

In general, group 2 patients, those who had persistent CT abnormalities at 12 months, were older, were more likely to require oxygen supplementation, received ribavirin treatment for a longer duration, and underwent additional hydrocortisone or methylprednisolone therapy. Group 2 patients also had a lower lymphocyte count, higher overall percentage of opacification on the acute chest radiograph, and longer time for complete resolution of radiographic abnormalities.

View larger version (162K)

Fig. 1A —14-year-old girl with severe acute respiratory syndrome. Transverse thin-section CT scan obtained 6 months after diagnosis shows subtle area of ground-glass opacification (arrowheads) is present at junction of posterior segment of right upper lobe and apical segment of right lower lobe.

View larger version (161K)

Fig. 1B —14-year-old girl with severe acute respiratory syndrome. Transverse thin-section CT scan obtained 12 months after diagnosis shows ground-glass opacification is resolved but smaller area of reticular pattern (arrows) is present at posterior segment of right upper lobe.

View larger version (183K)

Fig. 2 —15-year-old girl who required ventilatory support during acute illness of severe acute respiratory syndrome. Transverse thin-section CT scan obtained 6 months after diagnosis shows patchy areas of residual ground-glass opacification (arrowheads) in left lower lobe.

View larger version (228K)

Fig. 3 —11-year-old boy who had bilateral lower lobe consolidation during acute illness of severe acute respiratory syndrome. Transverse expiratory thin-section CT scan shows small subpleural area of air trapping (arrow) is present in left lung base.

There was no difference between the two groups with respect to the sex of the patients, duration of hospital stay, duration of fever, clinical presentation, requirement of ventilatory support, and serum creatine phosphokinase and lactate dehydrogenase levels.

View larger version (182K)

Fig. 4 —Transverse thin-section CT image in 8-year-old girl who had unifocal right upper lobe consolidation during acute illness of severe acute respiratory syndrome. Small parenchymal scar (arrow) is present at previous area of consolidation and remains unchanged between 6 and 12 months after diagnosis.

Forward stepwise logistic regression revealed that the most abnormal lymphocyte count (p = 0.03), the use of ribavirin (p = 0.03), the overall percentage of opacification on the acute chest radiograph (p = 0.042), and the number of days taken for complete radiographic resolution (p = 0.025) were significant factors associated with abnormal thin-section CT.

All nine patients who underwent no pulmonary function test at 6-month follow-up were in group 1. Of the 15 patients who had a thin-section CT abnormality detected at 12-month follow-up, only one child had mild restrictive deficit (TLC > 70% predicted) and his lung function showed no significant change at the 6- and 12-month follow-ups. For the group 1 patients, two had a mild obstructive deficit (> 70% predicted) and one had a restrictive deficit (> 70% predicted) at the 6-month follow-up, and their pulmonary function test results were normal at 12 months. For most children in this study who had relatively normal lung function test results at 6 months, their mean FEV₁ and the ratio FEV₁/FVC showed further improvement (paired Student's t test, p < 0.05) at 12 months (Table 4).

View Larger Version

TABLE 4: Results of Conventional Lung Function Tests in 38 Pediatric Patients 6 and 12 Months After Diagnosis of Severe Acute Respiratory Syndrome (SARS)

Lower respiratory tract disease is common in children. The clinical course of most viral infections is typically mild and is self-limited with uneventful recovery. However, severe complications might occur and long-standing pulmonary sequelae as a result of small airway damage have been described in patients after adenovirus pneumonia and Mycoplasma pneumonia [17-20]. Chest radiography is poor at depicting even advanced obliterative bronchiolitis [21], whereas thin-section CT can show indirect signs of small airways disease, such as mosaic attenuation pattern (defined as scattered irregular areas of high and low attenuation), pulmonary vascular abnormalities, bronchial abnormalities, and air trapping on expiratory CT [22]. Currently, thin-section CT is recommended as the diagnostic tool of choice in the investigation of small airways disease in view of its high spatial and contrast resolution and relative noninvasiveness compared with lung biopsy.

For adults patients, the temporal pattern of lung abnormalities on thin-section CT has been reported, which showed evidence of reticulation after the second week that persisted in half of all patients evaluated after 4 weeks [23]. In our group of children, all had radiographic resolution of consolidative changes on discharge from the hospital (range, 8-46 days; mean, 16 days). At 6- and 12-month follow-ups, mild CT changes that resembled those reported in adult patients who underwent a short-term follow-up examination (within 50 days after the acute disease) [23] were noted in some of the patients in our cohort. These changes included residual ground-glass opacities, reticular opacities, and parenchymal bands, although the degree of abnormalities was much milder in our group than in the adult group. Severe lung changes reported in adults, such as honeycombing, traction bronchiectasis, and fibrosis with irregular interfaces [15], were not observed in this pediatric cohort.

The difference observed between adult and pediatric groups might be accounted for by the fact that less severe disease was seen in children and the follow-up from disease onset (12 months) was longer than in previous adult series [15, 16, 23, 24]. The result suggested that patients with a more severe course of the disease were more likely to develop persistent subclinical CT abnormalities. The two major clinical indicators for chronic pulmonary changes identified include the degree of lymphopenia and the extent of radiographic opacification at presentation.

The duration of ribavirin treatment is most likely a marker of more severe disease rather than a causative factor in lung damage. Ribavirin was used in all cases of possible or probable SARS in our locality because the response of the causative agent to ribavirin was unknown in the initial phase. The drug was given to all subjects in this cohort as part of the treatment protocol (Appendix 1). The duration of ribavirin treatment was determined by the clinical condition of the patient. To our knowledge, pulmonary toxicity secondary to ribavirin has not been reported in the literature [25]. The main adverse reaction with this drug was hemolysis, which was not found in our patients. The thin-section CT abnormalities in our group of patients were therefore unlikely related to the drug effect and were most likely secondary to the disease itself. None of the blood and sputum cultures showed coexisting infection. The longest duration of ventilation support in our subjects was only 6 days and there would be little contribution to development of chronic lung changes.

Air trapping was observed in a proportion of our pediatric subjects after the acute infection of SARS; this finding has not been reported, to our knowledge, in previous adult series because thin-section CT was performed at the end of inspiration only. Therefore, this feature might have been underdetected in adults. Furthermore, the presence of air trapping in the pediatric group could be explained by the greater contribution of small airways to airway resistance in young children. In our group, the degree of air trapping was mild to moderate (< 70%) and was identified only on expiratory scans. The mosaic pattern was not identified in any of our patients.

Air trapping is secondary to collateral air drift into the alveoli beyond the narrowed or obstructed bronchus or bronchiole [12]. It causes a decrease in pulmonary blood flow and subsequently leads to gradual atrophy of the involved portion of lung tissue. There is also evidence to suggest air trapping might precede bronchiectasis [26]. Air trapping on thin-section CT was found in 15% and 37% of children with a history of adenovirus pneumonia and Mycoplasma pneumonia, respectively [27, 28]. Residual lung function abnormalities have also been reported in association with air trapping shown on thin-section CT. Mok et al. [29] found long-term impairment of small airway function even in asymptomatic children after Mycoplasma pneumonia. In our cohort, the majority of the patients (82%) did not show an increase in the extent of air trapping on the second follow-up examination, which was 6 months after the first follow-up examination; however, a mild increase in extent (up to 10%) was observed in two patients. Close monitoring of lung function and of thin-section CT follow-up of these children might be warranted to detect further progression.

In this cohort, although resolution was observed in children with persistent ground-glass opacification, new changes of reticular pattern were observed in a number of patients. This pattern has been reported in adult survivors of acute respiratory distress syndrome [30] and was found to be strongly related to the duration of mechanical ventilation. We did not find such association in our cohort, which might be accounted for by the small sample size. The clinical significance was not certain at this stage. Further follow-up might be warranted in this group of patients.

Parenchymal scars did not show any change in all patients with that finding, and the presence of parenchymal scar alone probably does not warrant follow-up. Bronchiectasis, which occurs most commonly secondary to adenovirus or bacterial infection, is not a feature in pediatric SARS.

Although 32% (15 of 47) of the patients had persistent CT abnormalities at 12 months after diagnosis, only one patient showed a persistent abnormality on pulmonary function tests at 12 months. This discrepancy may be accounted for by the fact that only mild abnormalities were found on thin-section CT. Chang et al. [31] found that a proportion of children with postinfectious small airway damage had normal lung function. In another study, researchers reported that the correlation between the extent of thin-section CT abnormalities and lung function test results was poor [32]. In our study, although the lung function tests of the majority of patients who had SARS infection were already within the normal range 6 months after diagnosis, repeated lung function tests performed at the 12-month follow-up showed further improvement. This improvement, although not clinically significant, suggested that the previous insult of SARS infection caused subclinical impairment of lung function in this group of patients who gradually improved with time.

The main limitation of the study was the small sample size of subjects. This limitation was, however, an intrinsic problem of patient selection because only a small number of pediatric patients were infected by SARS in our locality and most had only mild infection; therefore, only a limited number of patients had residual CT abnormalities. The small sample size made it difficult to draw definitive conclusions about the results at the present stage. The CT abnormalities might represent sequelae of lung disease before SARS. One cannot rule out this possibility because of the lack of premorbid CT, although this possibility is considered unlikely because, according to the medical records, all subjects were healthy before the SARS infection.

The findings of our study have broadened our understanding of the pulmonary sequelae of this new disease. There was, however, no significant correlation between CT abnormalities and clinical symptoms. The absence of a suitable control group with which to compare the CT findings in our cohort of SARS patients is also a significant limitation of our study. Most children with other viral chest infections do not undergo thin-section CT or follow-up CT. The absence of a control group might handicap the interpretation of our findings. Reports from other centers that also treated patients with SARS would allow better assessment of the significance of CT abnormalities and the cost-effectiveness to follow up these patients.

In conclusion, up to 32% of children affected with SARS showed thin-section CT abnormalities up to 12 months after diagnosis despite clinical resolution and unremarkable pulmonary function test results. Persistent CT abnormalities, although mild, were more likely to develop in patients with more severe disease. The two major clinical indicators were the degree of lymphopenia and the extent of radiographic opacification at presentation. Interval changes were observed in patients with ground-glass opacification and air trapping but not in those with parenchymal scars.