Severe Acute Respiratory

Radiology

K. T. Wong, FRCR
Gregory E. Antonio,
FRANZCR Syndrome: Radiographic
David S. C. Hui, MD
Nelson Lee, MD
Edmund H. Y. Yuen, FRCR
Appearances and Pattern of
Alan Wu, MD
C. B. Leung, MD
Progression in 138 Patients1
T. H. Rainer, MD
Peter Cameron, MD PURPOSE: To retrospectively evaluate the radiographic appearances and pattern of
Sydney S. C. Chung, MD progression of severe acute respiratory syndrome (SARS).
Joseph J. Y. Sung, MD
Anil T. Ahuja, FRCR MATERIALS AND METHODS: Chest radiographs obtained at clinical presentation
and during treatment in 138 patients with confirmed SARS (66 men, 72 women;
mean age, 39 years; age range, 20 – 83 years) were assessed. Radiographic appear-
Index terms: ances of pulmonary parenchymal abnormality, distribution, and extent of involve-
Lung, radiography, 68.11
ment on initial chest radiographs were documented. Recognizable patterns of
Pneumonia, acute interstitial, 68.21
Severe acute respiratory syndrome radiographic progression were determined by comparing the overall mean percent-
age of lung involvement for each patient on serial radiographs.
Published online before print
10.1148/radiol.2282030593 RESULTS: Initial chest radiographs were abnormal in 108 of 138 (78.3%) patients and
Radiology 2003; 228:401– 406 showed air-space opacity. Lower lung zone (70 of 108, 64.8%) and right lung (82 of
Abbreviations:
108, 75.9%) were more commonly involved. In most patients, peripheral lung involve-
SARS ⫽ severe acute respiratory ment was more common (81 of 108, 75.0%). Unifocal involvement (59 of 108, 54.6%)
syndrome was more common than multifocal or bilateral involvement. No cavitation, lymphad-
WHO ⫽ World Health Organization enopathy, or pleural effusion was demonstrated. Four patterns of radiographic progres-
sion were recognized: type 1 (initial radiographic deterioration to peak level followed by
1
From the Departments of Diagnostic Ra- radiographic improvement) in 97 of 138 patients (70.3%), type 2 (fluctuating radio-
diology and Organ Imaging (K.T.W., graphic changes) in 24 patients (17.4%), type 3 (static radiographic appearance) in 10
G.E.A., E.H.Y.Y., A.T.A.), Medicine and
patients (7.3%), and type 4 (progressive radiographic deterioration) in seven patients
Therapeutics (D.S.C.H., N.L., A.W., C.B.L.,
J.J.Y.S.), Accident and Emergency Medi- (5.1%). Initial focal air-space opacity in 44 of 59 patients (74.6%) progressed to
cine (T.H.R., P.C.), and Surgery (S.S.C.C.), unilateral multifocal or bilateral involvement during treatment.
Prince of Wales Hospital, Chinese Univer-
sity of Hong Kong, 30-32 Ngan Shing St, CONCLUSION: Predominant peripheral location; common progression pattern
Shatin, Hong Kong SAR. Received April from unilateral focal air-space opacity to unilateral multifocal or bilateral involve-
15, 2003; revision requested April 17; re-
vision received April 22; accepted April ment during treatment; and lack of cavitation, lymphadenopathy, and pleural
29. Address correspondence to K.T.W. effusion are the more distinctive radiographic findings of SARS.
(e-mail: [email protected]). © RSNA, 2003
See also the other article by Wong
et al in this issue.
Author contributions:
Guarantor of integrity of entire study,
A.T.A.; study concepts, A.T.A., K.T.W., Editor’s Note: Although the 138 patients described in this report were also included in a report
G.E.A., S.S.C.C., J.J.Y.S.; study design,
A.T.A., K.T.W., G.E.A.; literature research, published online by the New England Journal of Medicine (www.nejm.org; April 7, 2003), the
A.T.A., K.T.W., G.E.A.; experimental stud- analysis of the radiographic findings for these patients in the Radiology report has been performed
ies, K.T.W., D.S.C.H., N.L., A.W., C.B.L., in much greater detail.
T.H.R., P.C., J.J.Y.S., S.S.C.C.; data acquisi- —Anthony V. Proto, MD, Editor
tion, K.T.W., G.E.A., D.S.C.H., N.L.,
E.H.Y.Y., A.W., C.B.L., T.H.R., P.C.; data In early March 2003, there was an outbreak of atypical pneumonia in Hong Kong. The
analysis/interpretation, K.T.W., G.E.A.,
A.T.A.; manuscript preparation and edit-
World Health Organization (WHO) defined the illness as severe acute respiratory syn-
ing, K.T.W., G.E.A., A.T.A., D.S.C.H.; drome (SARS). At the time of writing this article, there have been 1,059 reported cases in
manuscript definition of intellectual con- Hong Kong and more than 2,890 cases worldwide (1), including 32 deaths in Hong Kong
tent, K.T.W., G.E.A., A.T.A., J.J.Y.S., related to the illness.
D.S.C.H., P.C.; manuscript revision/re-
At our institution, over 200 confirmed cases of SARS have been treated (2). Imaging plays
view, K.T.W., A.T.A., G.E.A., D.S.C.H.,
P.C.; manuscript final version approval, a crucial role in diagnosis and in monitoring of disease progress during medical treatment.
K.T.W., A.T.A., D.S.C.H., J.J.Y.S. From our experience, the radiographic appearances of SARS at the time of initial presen-
© RSNA, 2003 tation are variable, ranging from normal to widespread opacification. In addition, patients
show different radiologic progression during treatment.

401
 Because the role of imaging is central to
the diagnosis and the care of the patients, TABLE 1 TABLE 2
Profile of 138 Subjects with SARS Location of Lung Opacities on Initial
radiographers and radiologists should be Radiographs in 108 Subjects
aware of the radiographic appearances of Subject Category No. of Subjects*
this disease and the infection-control Right
Health care workers 69 (50.0)† Lung Right and/or
Radiology

guidelines to prevent transmission of the Medical students 16 (11.6) Zone* Lung Left Lung Left Lung
disease. The purpose of our study was to Patients 53 (38.4)
retrospectively evaluate the radiographic Upper 13 (12.0) 9 (8.3) 18 (16.7)
* Number in parentheses is the percentage. Middle 42 (38.9) 35 (32.4) 50 (52.8)
appearances and patterns of progression in † Includes 20 physicians, 34 nurses, 15 al-
Lower 52 (48.1) 41 (38.0) 70 (64.8)
patients with SARS. lied health workers. Laterality 82 (75.9) 67 (62.0) ...

Note.—Number in parentheses is the per-

MATERIALS AND METHODS centage.
* Zone height defined as one-third of
Subjects (K.T.W., G.E.A., H.Y.Y.) who were un- craniocaudal extent of lung.
Between March 11 and 25, 2003, 138 aware of the clinical progress of the sub-
subjects (66 men, 72 women; mean age, jects. Each lung was divided into three
39 years; age range 20 – 83 years) were zones: upper, middle, and lower. Each
identified as being secondary (history of zone spanned one-third of the craniocau- TABLE 3
direct contact with index case at our in- dal distance of the lung on the frontal Appearance of Lung Opacities on
radiograph and was evaluated separately. Initial Radiographs in 108 Subjects
stitution) or tertiary (history of direct
contact with secondary cases) cases of The observers assessed the presence,
No. of
SARS at our institution. There were 66 appearances, distribution, and size of Appearance Subjects*
men and 72 women, with a mean age of lung parenchymal abnormalities on each
chest radiograph in all subjects. The ap- No. of lesions
39 years, age range of 20 – 83 years. Sixty- Unifocal 59 (54.6)
nine of the 138 were health care workers pearances were categorized as follows: Multifocal unilateral 8 (7.4)
and an additional 16 were medical stu- air-space shadow, reticular shadow, nod- Multifolcal bilateral 41 (38.0)
dents who were present for clinical teach- ular shadow, or mass. The anatomic dis- Pattern of involvement
tribution was noted to be central if the Peripheral 81 (75.0)
ing in the index ward (Table 1). The re- Central 13 (12.0)
maining 53 were inpatients in the same abnormality predominantly involved the Mixed peripheral and central 14 (13.0)
medical ward or those who had visited medial half of the zone and peripheral if No. of involved zones
their relatives in that ward. The diagnosis it predominantly involved the lateral One 59 (54.6)
half. The size of the lesion was assessed Two 28 (25.9)
of SARS was based on WHO diagnostic More than two 21 (19.4)
criteria (3). This retrospective study was by visually estimating the percentage
approved by our Institutional Review area occupied in each zone on each side * Number in parentheses is the percentage.
Board; patient informed consent was not to determine the overall mean percent-
required. age of involvement by averaging the per-
centage involvement of the six lung tered intravenously in pulsed fashion
zones. Associated findings, in particular (0.5 g for three consecutive days) to 107
Chest Radiography and Evaluation
the presence of cavitation, lymphade- subjects whose clinical condition so indi-
Frontal chest radiographs were ob- nopathy, and pleural effusion, were also cated.
tained at initial clinical presentation and assessed.
during treatment. The initial chest radio- Serial frontal chest radiographs ob- Data Analysis
graph was obtained an average of 2.5 tained during treatment were also retro-
days (range, 0 –10 days) after onset of fe- spectively reviewed by the same radiolo- The radiographic patterns at the time
ver. Only frontal chest radiographs were gists in consensus. All subjects included of clinical presentation in all 138 sub-
obtained (posteroanterior for subjects in this study underwent serial follow-up jects, as assessed on initial chest radio-
who could stand, anteroposterior for chest radiography for at least 14 days graphs, were analyzed and categorized as
those who could not). All radiographic (unless deceased). For each follow-up ra- normal or unifocal, unilateral multifocal,
examinations were performed with com- diograph, the extent of lung parenchy- or bilateral multifocal abnormalities. The
puted radiography equipment (Mobilett mal involvement was assessed by using distribution of lung parenchymal in-
Plus; Siemens, Erlangen, Germany) by us- the same method as for the radiograph volvement in terms of central or periph-
ing a standardized technique (75 kV, 4 obtained at initial clinical presentation. eral involvement and the zones involved
mAs, 180-cm film-focus distance for pos- Follow-up radiographs were obtained were noted. We determined if there were
teroanterior; 70 kV, 4 mAs, 100-cm film- daily during the hospital stay. All sub- recognizable patterns of radiographic
focus distance for anteroposterior; broad jects were given a combination of ribavi- progression by comparing the overall
tube focus for both). The images were rin (Derbin BLC, United Kingdom) (an percentage of lung involvement for each
assessed by using a picture archiving and antiviral agent, administered orally; ini- subject on serial radiographs.
communication system viewer with a tial dose of 2.4 g followed by 1.2 g three
2,048 ⫻ 2,048-pixel monitor (Magicview times daily) and corticosteroid (pred- RESULTS
version VA22E; Siemens). nisolone; Clonmel Healthcare, Ireland)
Appearances of Abnormalities at
The frontal chest radiographs obtained (0.5–1.0 mg per kilogram of body weight
Presentation
at clinical presentation and at follow-up per day; ) for treatment. The steroid
during treatment were retrospectively re- methylprednisolone (Solu-medrol; Phar- Chest radiographs obtained at presen-
viewed in consensus by three radiologists macia Upjohn, Belgium) was adminis- tation were abnormal in 108 of 138 sub-

402 䡠 Radiology 䡠 August 2003 Wong et al

Radiology

Figure 2. Schematic depicts the four patterns of radio-

graphic progression determined from serial chest radio-
graphs. See Table 4 for time to peak(s).
Figure 1. Frontal chest radiograph in a 23-year-old man with SARS
shows a focal ill-defined air-space opacity predominantly involving
the periphery of right lower zone. Note lack of cavitation, lymphad-
enopathy, and pleural effusion. changes with no discernible radiographic
peak or change in overall mean lung in-
volvement of less than 25% for more
than 10 days; and type 4, progressive ra-
jects (78.3%). The radiographic pattern (49 of 108, 45.4%). Bilateral disease was diographic deterioration.
observed in all 108 subjects was air-space present in 41 (38.0%) subjects. At presen- The type 1 pattern was the most com-
opacities with ill-defined margins. None tation, the overall mean lung involvement monly observed (97 of 138, 70.3%), with
of these lesions showed a reticular or was 4.7% (range, 0.8%– 63.3%). radiographic deterioration to a peak (mean
nodular pattern or a mass. There was no time from onset of fever to peak, 8.6
evidence of cavitation in the area of air- Radiographic Progression days ⫾ 3.1 [SD]; range, 2–17 days) followed
space consolidation, lymphadenopathy, by radiographic improvement (Fig 3).
Of the 30 subjects with an initial nor-
or pleural effusion (except in one subject Twenty-four of 138 subjects (17.4%) had
mal chest radiograph, 29 showed evi-
with concomitant congestive cardiac fail- the type 2 pattern, with two distinct radio-
dence of air-space opacities on subse-
ure who had a small pleural effusion) on
quent follow-up chest radiographs after graphic peaks at 6.3 days ⫾ 3.0 and 13.5
any chest radiograph at initial presenta-
an average of 3.1 days (range, 1–7 days). days ⫾ 3.7. Ten subjects (7.3%) had the
tion. Chest radiographs obtained at pre- static type 3 radiographic appearances for
The remaining subject had normal-ap-
sentation were normal in 30 of 138 sub- most of the time during treatment. Among
pearing follow-up radiographs, but signs
jects (21.7%). seven subjects with Type 4 pattern, radio-
and symptoms were strongly suggestive
of SARS and the diagnosis was made with graphs showed progressive deterioration
Distribution of Abnormalities at until the lungs became completely consol-
the aid of thin-section computed tomog-
Presentation idated or the subject died. Six died during
raphy (CT) of thorax. Thus, of the total of
The location and appearance of lung 138 subjects, all but one showed air-space the study period, and one remains criti-
opacities on initial radiographs are shown opacity, either initially (n ⫽ 108) or sub- cally ill at the time of this writing, requir-
in Tables 2 and 3, respectively. The right sequently (n ⫽ 29) on radiographs. ing intensive care and assisted ventilation.
lung (82 of 108 subjects, 75.9%) was in- At review of follow-up radiographs, we Of the 59 subjects with unilateral focal
volved in more subjects than the left (67 of were able to identify four patterns of radio- air-space opacity on initial radiographs,
108, 62.0%). The disease showed a predi- graphic progression (Table 4, Fig 2): type 1, 44 (74.6%) progressed to unilateral mul-
lection for the lower zone (70 of 108, initial radiographic deterioration to peak tifocal (n ⫽ 10) or bilateral multifocal
64.8%). Involvement of peripheral lung level, followed by radiographic improve- (n ⫽ 34) air-space opacities during hospi-
parenchyma (81 of 108, 75.0%) (Fig 1) was ment, with maximum difference in overall talization. In fifteen (25.4%), the opacity
more common than a mixed peripheral mean lung involvement greater than 25%; remained unilateral focal in terms of lung
and central pattern (14 of 108, 13.0%) or a type 2, fluctuating radiographic changes involvement.
central pattern (13 of 108, 12.0%) at the with at least two radiographic peaks and Confluent air-space opacities diffusely
time of presentation. Unifocal involve- an intervening trough, which differed by involving both lungs, compatible with
ment (59 of 108, 54.6%) was slightly more more than 25% for overall mean lung acute respiratory distress syndrome, were
common than multifocal involvement involvement; type 3, static radiographic observed in 11 of 138 subjects (8.0%)

Volume 228 䡠 Number 2 SARS: Radiographic Appearances and Pattern of Progression 䡠 403
Radiology

Figure 3. Serial radiographic appearances in a 23-year-old woman

with SARS, type 1 pattern. (a) Frontal chest radiograph obtained at
clinical presentation shows unilateral focal air-space opacity in the
right middle zone. (b) Follow-up frontal chest radiograph obtained 5
days later shows progression of radiographic changes, with multifocal
bilateral air-space opacities in both lungs. (c) Subsequent follow-up
chest radiograph obtained after another 7 days shows radiographic
improvement in extent of pulmonary parenchymal air-space opacities
after successful medical therapy with a combination of oral ribavirin
and corticosteroids.

during the course of the disease (Fig 4);

this occurred almost exclusively in sub-
jects with poor clinical outcome (six died
and five required prolonged assisted ven-
tilation at the end of the study period).

DISCUSSION

SARS was recognized as a global health

hazard in March 2003. At our institution
it initially affected mainly health care
professionals but soon spread to involve
inpatients, outpatients and their con-
tacts. With the convenience of air travel,
the disease has now spread to all parts of
the world, with patients appearing in
other parts of Asia, Europe, North Amer-
ica, and Australia. In view of the world-
wide increase in number of confirmed
cases, the WHO issued a global alert for
the first time in more than a decade (1). implicated as the causative agent (4). It is tant for the protection of health care
The disease is highly infectious and, at suspected that infection is transmitted by workers, as recommended by the Centers
the time of this writing, only preliminary means of droplets and, possibly, fomites; for Disease Control and Prevention (5).
data have been reported on the causative hence, both respiratory and contact in- On the basis of our results, air-space
agent of SARS. A coronavirus has been fection-control precautions are impor- opacification is the pattern seen on chest

404 䡠 Radiology 䡠 August 2003 Wong et al

 features, including high fever (temper-
TABLE 4 ature ⬎ 38°C), chills, rigor, myalgia,
Patterns of Radiographic Progression of SARS in 138 Subjects
and laboratory findings such as leuko-
Progression penia and thrombocytopenia in pa-
Pattern* No. of Subjects† Time to Peak (d)‡ tients with recent exposure, are very
Radiology

Type1 97 (70.3) 8.6 ⫾ 3.1 suggestive of SARS. The presence of an

Type2 24 (17.4) 6.3 ⫾ 3.0, 13.5 ⫾ 3.7§ air-space opacity on chest radiographs
Type3 10 (7.3) Not applicable has been as helpful in confirmation of
Type4 7 (5.1) Not applicable the diagnosis.
* Pattern types are defined in the Results section. Radiographic progression to unilateral
† Number in parentheses is the percentage.
multifocal or bilateral involvement oc-
‡ Data are mean ⫾ SD.
§ Numbers are for first peak and second peak.
curred in most subjects with unilateral
focal air-space opacity on the initial chest
radiograph obtained during treatment
with a combination of an antiviral agent
and corticosteroids. We found that only a
small percentage (7.2%) of subjects with
SARS showed a static radiographic ap-
pearance (type 3 pattern) for most of the
time during treatment. On the other
hand, progressive radiographic deteriora-
tion despite medical treatment seems to
be associated with poor prognosis, with
all deaths in our series occurring in pa-
tients with the type 4 pattern.
There are some limitations to our
early study: (a) The subjects in this
study were a heterogeneous group, and
we have not taken into account the se-
verity of clinical symptoms and clinical
outcome. (b) Visual estimation of the
percentage of lung involvement may
appear to be subjective; however, in our
opinion this appears to be the most
practical method in real life, in an epi-
demic crisis in which it is not possible
to design detailed computer-based
models at short notice. (c) A frontal ra-
diograph alone may not be accurate in
helping identify central versus periph-
Figure 4. Frontal chest radiograph in a 76-year-old man with SARS eral lesions; however, in patients who
who was undergoing medical treatment shows diffuse confluent air- underwent thin-section CT, this local-
space opacities involving both lungs and normal heart size. These ization proved to be accurate (6). (d)
findings are compatible with radiologic features of acute respiratory This report does not deal with clinical
distress syndrome. and radiologic comparison, particularly
the timing and type of treatment regi-
men and the timing of the radiographic
radiographs in patients with SARS. All diographic feature of SARS in our study. response. It also does not include eval-
but one of 138 subjects showed air-space The opacities occupied a peripheral or uation of any outcome indicators. Such
opacification of varying extent and dis- mixed peripheral and central location an evaluation requires detailed clinical,
tribution at some stage of the disease. In in 88% of subjects. Important absent immunologic, and statistical analyses,
one subject, the initial and early-progress findings included the lack of cavitation, which may be possible when more data
radiographs were normal and the diagno- lymphadenopathy, or pleural effusion. and experience become available.
sis was made with the aid of thin-section The radiographic appearance of pe- In conclusion, SARS has become a
CT of thorax. The CT scans in this subject ripheral air-space opacities in SARS is global health hazard and its high infec-
showed a small (approximately 2-cm) indistinguishable from other causes of tivity is alarming. Imaging plays an im-
area of ground-glass opacification with atypical pneumonia, such as Myco- portant role in the diagnosis and moni-
intralobular interstitial and interlobular plasma, Chlamydia, and Legionella (7,8), toring of response to therapy. The
septal thickening in the posterior costo- and overlap with other types of viral predominant peripheral location; com-
phrenic recess of the right lung. We have pneumonia in adults (9). Since imaging mon progression pattern from unilateral
recently reported the thin-section CT alone cannot help differentiate SARS focal air-space opacity to unilateral mul-
findings in patients with SARS (6). from other diseases, the clinical mani- tifocal or bilateral involvement during
The peripheral location of air-space festation is indispensable for diagnosis. treatment; and lack of cavitation, lymph-
opacification was another common ra- The presence of characteristic clinical adenopathy, and pleural effusion are the

Volume 228 䡠 Number 2 SARS: Radiographic Appearances and Pattern of Progression 䡠 405
 more distinctive radiographic findings of .who.int/csr/sars/casedefinition/en/. Accessed 7. Goodman LR, Goren RA, Teptick SK. The
this potentially lethal disease. April 12, 2003. radiographic evaluation of pulmonary in-
4. Update: outbreak of severe acute respira- fection. Med Clin North Am 1980; 64:553–
tory syndrome—worldwide, 2003. MMWR 574.
References
Morb Mortal Wkly Rep 2003; 52:241–248. 8. Macfarlane JT, Miller AC, Roderick Smith
1. World Health Organization. Cumulative 5. Centers for Disease Control and Preven- WH, Morris AH, Rose DH. Comparative
number of reported cases (SARS) from 1
Radiology

tion. Updated interim domestic infection radiographic features of community ac-

February to 27 March 2003. Available at:
control guidance in the health care and quired Legionnaires disease, pneumococ-
www.who.int/csr/sarscountry/2003_04_11.
Accessed April 12, 2003. community setting for patients with sus- cal pneumonia, mycoplasma pneumonia,
2. Lee N, Hui D, Wu A, et al. A major out- pected SARS. Available at: www.cdc.gov and psittacosis. Thorax 1984; 39:28 –33.
break of severe acute respiratory syndrome /ncidod/sars/infectioncontrol.htm. Accessed 9. Kim EA, Lee KS, Primack SL, et al. Viral
in Hong Kong. N Engl J Med [serial online]. April 4, 2003. pneumonias in adults: radiologic and
April 14, 2003. Available at: nejm.org/ear- 6. Wong KT, Antonio GE, Hui DSC, et al. pathologic findings. RadioGraphics 2002;
lyrelease/sars.asp. Thin-section CT of severe acute respiratory 22(spec issue):137S–149S.
3. World Health Organization. Case defini- syndrome: evaluation of 73 patients ex-
tions for surveillance of severe acute respi- posed to or with the disease. Radiology
ratory syndrome (SARS). Available at: www 2003; 228:395– 400.

406 䡠 Radiology 䡠 August 2003 Wong et al