Chest CT-based differential diagnosis of 28 patients with suspected corona virus disease 2019 (COVID-19)

Objectives: The chest CT findings that can distinguish patients with corona virus disease 2019 (COVID-19) from those with clinically suspected COVID-19 but subsequently found to be COVID-19 negative have not previously been described in detail. The purpose of this study was to determine the distinctions among patients with COVID-19 by comparing the imaging findings of patients with suspected confirmed COVID-19 and those of patients initially suspected to have COVID-19 who were ultimately negative for the disease. Methods: 28 isolated suspected in-patients with COVID-19 were enrolled in this retrospective study from January 22, 2020 to February 6, 2020. 12 patients were confirmed to have positive severe acute respiratory syndrome corona virus 2 (SARS-CoV-2) RNA results, and 16 patients had negative results. The thin-section CT imaging findings and clinical and laboratory data of all the patients were evaluated. Results: There were no significant differences between the 12 confirmed COVID-19 (SARS-Cov-2-positive) patients and 16 SARS-CoV-2-negative patients in epidemiology and most of the clinical features or laboratory data. The CT images showed that the incidence of pure/mixed ground-glass opacities (GGOs) was not different between COVID-19 and SARS-CoV-2-negative patients [9/12 (75.0%) vs 10/16 (62.5%), p = 0.687], but pure/mixed GGOs in the peripheral were more common in patients with COVID-19 [11/12 (91.7%) vs 6/16 (37.5%), p = 0.006]. There were no significant differences in the number of lesions, bilateral lung involvement, large irregular/patchy opacities, rounded opacities, linear opacities, crazy-paving patterns, halo signs, interlobular septal thickening or air bronchograms. Conclusions: Although peripheral pure/mixed GGOs on CT may help distinguish patients with COVID-19 from clinically suspected but negative patients, CT cannot replace RT-PCR testing. Advances in knowledge: Peripheral pure/mixed GGOs on-chest CT findings can be helpful in distinguishing patients with COVID-19 from those with clinically suspected COVID-19 but subsequently found to be COVID-19 negative.

epidemiology and most of the clinical features or laboratory data. The CT images showed that the incidence of pure/mixed ground-glass opacities (GGOs) was not different between COVID-19 and SARS-CoV-2-negative patients [9/12 (75.0%) vs 10/16 (62.5%), p = 0.687], but pure/mixed GGOs in the peripheral were more common in patients with COVID-19 [11/12 (91.7%) vs 6/16 (37.5%), p = 0.006]. There were no significant differences in the number of lesions, bilateral lung involvement, large irregular/patchy opacities, rounded opacities, linear opacities, crazy-paving patterns, halo signs, interlobular septal thickening or air bronchograms. Conclusions: Although peripheral pure/mixed GGOs on CT may help distinguish patients with COVID-19 from clinically suspected but negative patients, CT cannot replace RT-PCR testing. advances in knowledge: Peripheral pure/mixed GGOs on-chest CT findings can be helpful in distinguishing patients with COVID-19 from those with clinically suspected COVID-19 but subsequently found to be COVID-19 negative.
Radiological examination, especially thin-section chest CT, is a non-invasive, quick method and has the advantages of discovering lung pathological changes early and reflecting the severity of the disease. Some previous radiographical studies of chest CT findings in patients with COVID-19 indicate that bilateral ground-glass opacities (GGOs) or consolidation on-chest CT should remind radiologists to make a possible diagnosis of COVID-19. 5 However, all these data came from confirmed patients, and the characteristics to be aware of in suspected patients are not clear.
In this study, we aimed to discover the key points for diagnosis by comparing the chest CT characteristics of 12 patients with confirmed COVID-19 and 16 clinically suspected but SARS-CoV-2-negative patients.

meThODS anD maTerIalS Patients
This retrospective study was approved by the Ethic Committees of the Third Affiliated Hospital of Sun Yat-Sen University. As a retrospective study, informed consent was waived. We included all patients admitted to the Third Affiliated Hospital of Sun Yat-Sen University in Guangzhou from January 22, 2020 to February 6, 2020 ( Figure 1). Our inclusion criteria (suspected patients) were patients with any epidemiological risk history and any two of the clinical characteristics as follows: (1) epidemiological history within 14 days prior to the onset of the disease: history of travel to or residence in Wuhan and its surrounding areas; close contact with COVID-19 patients or other patients from Wuhan with fever; clustered cases; and (2) clinical characteristics: fever and/or respiratory symptoms, imaging findings of pneumonia, normal or decreased total white blood cell count, or decreased lymphocyte count in the early stages of disease. Patients were confirmed by SARS-COV-2 RNA detection with respiratory or blood samples by the Guangdong Centers for Disease Control and Prevention (CDC). If the SARS-COV-2 RNA test was negative twice more than 24 h apart, the suspected patient was defined as negative. The data on demographic, epidemiological, clinical and laboratory characteristics were abstracted from medical records. 6 A total of 1119 patients were screened in the fever clinic at the Third Affiliated Hospital of Sun Yat-Sen University, and 28 suspected patients were enrolled, including 12 confirmed and 16 negative patients.

Chest CT
All these patients performed thin-section CT scans. The median duration from the onset of initial symptoms to CT scan was 4 days, ranging from 1 day to 2 weeks. All CT examinations were performed using an Aquilion ONE scanner (Toshiba Medical Systems; Tokyo, Japan) or an IQon Spectral scanner (Philips Healthcare, Best, the Netherlands). All CT images were obtained with the patient in the supine position during end-inspiration and without contrast medium. The CT protocol is as follows: 120 kV; automatic tube current (180 mA-440 mA); iterative reconstruction technique; detector, 32/160 mm; rotation time, 0.5 sec; section thickness, 1 mm; collimation, 0.5 mm; pitch, 1.5; and matrix, 512*512. An image reconstruction kernel was used to smooth the lung with a thickness of 1 mm and an interval of 1 mm. The CT images were viewed with lung (window width, 1,600 HU; window level, −600 HU) and mediastinal (window width, 300 HU; window level 40 HU) window settings.
All thin-section CT images were reviewed by two experienced cardiothoracic radiologists (SidongXie and Xiuzhen Chen) using a viewing console. The images were reviewed independently and reached a decision in consensus. In case of disagreements between the two primary radiologists' interpretations, a third cardiothoracic radiologist with 20 years of experience (Jie Qin) adjudicated a final decision.
The major CT findings were described using international standard nomenclature defined by the Fleischner Society glossary and peer-reviewed literature on viral pneumonia. [7][8][9][10] In all 28 patients, the initial CT findings were evaluated for the following features: (1) type of opacity: pure/mixed GGO or consolidation; (2) lesion location: peripheral pulmonary/ subpleural or central distribution/scattered; (3) morphology of lesions: large irregular/patchy opacities, rounded opacities or linear opacities; (4) the number of lung lobes affected by GGOs or consolidation and whether there was bilateral lung involvement; (5) interstitial involvement of the lung: peribronchovascular interstitial thickening, interlobular septal thickening, tree-in-bud pattern, honeycombing, subpleural curvilinear line, and pleural thickening; (6) crazy-paving pattern; (7) halo sign; (8) air bronchogram; (9) lymphadenopathy (lymph node size of ≥10 mm in short-axis diameter); and (10) underlying lung diseases such as emphysema or bronchiectasis; and (11) other abnormalities (e.g., solid pulmonary nodule, cavitation and pleural effusion). Statistical analysis All data were statistically analysed using SPSS v.22.0 (IBM Corp, USA). Continuous variables are expressed as the mean (SD) or median (IQR) and were compared with the t-test or Mann-Whitney U test; categorical variables are expressed as frequency (%) and were compared by χ² test or Fisher's exact test between the confirmed and suspected but negative patients. P values < 0.05 were considered to be a statistically significant difference.

reSulTS
Baseline characteristics, clinical features and laboratory data All 28 clinically suspected patients admitted to The Third Affiliated Hospital of Sun Yat-Sen University were enrolled, including 12 confirmed patients with COVID-19 and 16 SARS-CoV-2-negative patients. The confirmed patients included six males and six females; the median age was 40   years; in contrast, four men and twelve females with a median age of 35 years composed the suspected but negative patients. Patients in both groups mainly manifested fever and cough. The confirmed COVID-19 patients had more myalgia (91.6% vs 31.2%, p = 0.002), and the median number of neutrophils in SARS-CoV-2-negative patients was higher than that in confirmed patients (3.2 × 10 9 /L vs 4.6 × 10 9 /L,p < 0.05). The baseline characteristics, epidemiological factors, clinical features and laboratory data of the 12 confirmed COVID-19 patients and 16 SARS-CoV-2-negative patients are shown in Table 1.

CT imaging findings
The initial thin-section chest CT findings of 28 clinically suspected patients with COVID-19 are presented in Table 2. The main chest CT findings of clinically suspected patients included bilateral multi-focal lesions in the lungs, halo signs and pure/ mixed GGOs, which were more frequently located in the peripheral pulmonary/subpleural areas in the confirmed COVID-19 patients than in the suspected but negative patients (91.7% vs 37.5%, p = 0.006). The number of lesions, bilateral lung involvement, large irregular/ patchy opacities, rounded opacities and linear opacities, crazypaving patterns, interlobular septal thickening, air bronchograms, peribronchovascular interstitial thickening, tree-in-bud patterns, subpleural curvilinear lines and pleural thickening were apparent in confirmed and suspected but negative patients, and there were no significant differences. (Figures 2-4)

DISCuSSIOn
To screen for patients with COVID-19 and control the spread of the epidemic, clinically suspected patients need to be isolated. Chest imaging findings are important criteria for suspected cases. In our study, clinically suspected patients had bilateral multi-focal lung lesions, halo signs and pure/mixed GGO opacities, which were more peripheral/subpleural in the confirmed COVID-19 patients. Because COVID-19 mainly involves the respiratory system, chest CT is more sensitive than chest X-ray (CXR) in differential diagnosis initial evaluation, response evaluation and follow-up of COVID-19. CXR is usually no abnormal findings in the early stage of COVID-19, whereas CT findings may be present even before symptom onset. 11,12 Furthermore, CT imaging have been proven to have diagnostic value in several studies with initially false-negative RT-PCR screening results. 4,13,14 Therefore, to identify COVID-19 patients early and control the source of the infection, CT is often a first-line investigation for COVID-19 in mainland China. Many studies have concluded that the CT findings of COVID-19 include bilateral pulmonary parenchymal pure/ mixed GGOs in the lung periphery, accompanying crazypaving patterns, consolidation, intralobular interstitial thickening and interlobular septal thickening 3-5, 15, 16 . The common CXR findings are similar to those described for CT: bilateral, peripheral, consolidation and/or GGOs. 12,15 when the disease developed from the intermediate to rapid progression stages, CXR may show bilateral lung diffuse consolidative opacitiesthe features of acute respiratory distress syndrome (ARDS). 12 However, unlike in confirmed COVID-19 patients, whether these imaging findings exist in the suspected but negative patients has not been reported. It is clinically important to identify negative patients from suspected COVID-19 patients.
In our study, pure/mixed GGOs were also common in negative patients. Many pulmonary infectious diseases, such as influenza virus pneumonia, and non-infectious conditions,  including interstitial pulmonary oedema, pulmonary haemorrhage, respiratory bronchiolitis, hypersensitivity pneumonitis, organising pneumonia and alveolar proteinosis, can show similar imaging manifestations. 8,9 In addition, the type of exudative lesions is related to the different stages of COVID-19 pneumonia. Pan et al 16 reported that typical mild COVID-19 pneumonia initially presented as small subpleural, unilateral or bilateral GGOs in the lower lobes, which then changed into crazy-paving patterns and subsequent consolidation in a few days.
However, pure/mixed GGOs in the peripheries was a special imaging feature for differential diagnosis to distinguish patients with COVID-19 from clinically suspected patients in our study. The lesions of COVID-19 patients were predominately distributed in the peripheral/subpleural pulmonary regions. 3,16,17 We speculate that this is related to the pathophysiological mechanism of the disease. COVID-19 is caused by SARS-CoV-2, which is approximately 50-200 nm in diameter and is prone to staying in the terminal bronchioles and causing lung damage. In addition, SARS-CoV-2 mainly invades cells containing angiotensin-converting enzyme 2 (ACE2) receptors, which are found in regions rich in bronchial epithelial cells and type 2 alveolar epithelial cells, especially the latter. [18][19][20] As a result, the peripheral lung is attacked first and suffers the most damage. Nevertheless, the GGOs of both H1N1 pneumonia and SARS are also distributed more peripherally. 21,22 Therefore, the single factor of peripheral/subpleural lung distribution in COVID-19 is not very unique.
In addition, other CT features, including interlobular septal thickening, crazy-paving patterns, halo signs and consolidation, were not specific for a diagnosis of COVID-19 and could be seen in the negative patients. The crazy-paving pattern has been shown to occur in many other diseases, such as usual interstitial pneumonia, infection, pulmonary oedema, haemorrhage, ARDS, alveolar proteinosis, bronchiolitis obliterans organising pneumonia (BOOP), andradiation pneumonitis. 10,23 In our study, more confirmed COVID-19 patients had myalgia, but there were no differences between the two groups patients in other clinical symptoms. The neutrophil count in SARS-CoV-2-negative patients was higher than that in confirmed patients. This suggests the possibility of bacterial infection in suspected but negative patients, which may be an identifying point.
Because the clinical features and imaging findings are not unique to confirmed patients with COVID-19, aetiologic evidence, including SARS-CoV-2 RNA and serum IgM antibodies, remains the diagnostic standard for COVID-19.
There are several limitations to our study. First, the sample size of confirmed and suspected but negative patients was small. Second, the pathogens of most patients who were SARS-CoV-2 negative were not clear. Third, there were no pathological results of lung tissue available to investigate the correlation between radiological and histopathological findings.
In conclusion, although peripheral pure/mixed GGOs on CT may help distinguish patients with COVID-19 from clinically suspected but negative patients, CT cannot replace RT-PCR testing. aCknOWleDgmenT