Outcome of Pulmonary Spherical Ground-glass Opacities on Ct in Patients With Coronavirus Disease 2019 (Covid-19): A Retrospective Analysis

Background: Pulmonary spherical ground-glass opacities (GGOs) are commonly detected on initial chest CT scan in patients with coronavirus disease 2019 (COVID-19).We aimed to investigate the evolution of spherical GGOs to better understand their clinical signicance. Materials and Methods:A retrospective study of 33 consecutive patients with conrmed COVID-19 and pulmonary spherical GGOs was performed from January 21, 2020, to March 6, 2020. The initial and follow-up CT images and clinical data were reviewed. The initial CT manifestations of spherical GGOs and their subsequent changes were mainly evaluated. Results:A total of 101 pulmonary spherical GGOs, including 38 with and 63 without consolidation, were found in 33 patients. Of the 101 spherical GGOs, 71 (70.3%) and 30 (29.7%) showed progression and direct absorption on follow-up CT images, respectively. GGOs with consolidation were more likely to progress than those without (84.2% vs. 61.9%, p = 0.017). The 71 progressed lesions mainly showed an increase in size and/or density and most (70.4%) of them extended toward the pleura and developed from spherical to patchy. Internal consolidation appeared and increased in 18 (25.4%) and 22 (31.0%) lesions, respectively. During absorption, all the previous progressed and directly absorbed lesions exhibited a simultaneous decrease in size and density. On each patient’s nal CT, more lesions with progression had a residual mixed GGO (40.8% vs. 6.7%, p = 0.002) and fewer had pure GGO (39.4% vs. 60.0%, p = 0.016) than those with direct absorption. Conclusion: In patients with COVID-19, most pulmonary spherical ground-glass opacities would progress, especially those with consolidation, and develop into patchy, subpleural lesions. with CT


Introduction
The current outbreak of coronavirus disease 2019 (COVID-19) was rst reported in Wuhan, China, on December 31, 2019, and it subsequently spread worldwide [1]. With an increasing number of con rmed cases, the World Health Organization declared COVID-19 a global health emergency. Although real-time reverse transcription polymerase chain reaction (RT-PCR) remains the reference standard for diagnosing COVID-19 [2], the emergence of false negatives [3,4] and the requirement for signi cant professional skills restricted prompt diagnosis of suspected cases. Some investigations in the primary epidemic area reported that chest CT had greater sensitivity for COVID-19 diagnosis compared with initial RT-PCR [5,6], making chest CT an important tool in the early detection, diagnosis, and evaluation of the disease course [7].
To date, there have been several studies on imaging of COVID-19 [8][9][10][11]. Pan et al. [8] and Chung et al. [9] investigated the initial and follow-up CT ndings of 63 and 21 con rmed patients, respectively, and found that the CT manifestations of lesions were various and changed rapidly. Pan et al. [10] studied the temporal CT changes of COVID-19 from initial diagnosis until recovery and identi ed four stages of evolution on chest CT. They also found that spherical ground-glass opacities (GGOs) were the initial CT manifestation, which was consistent with the result of Chung et al. [8]. However, no research has focused on the development of spherical GGOs and their changes following treatment.
Because extensive, patchy GGOs and GGOs with consolidation were common at advanced stage of COVID-19, we hypothesized that some spherical GGOs were the precursors of patchy GGOs with or without consolidation. If this hypothesis is con rmed, it can aid in the earlier diagnosis of suspected cases. This study investigated the outcome of spherical GGOs on CT to reveal their evolution and relationship with patchy lesions. The results will allow radiologists and clinicians to better understand the evolution of spherical GGOs on chest CT, hopefully diagnose the disease earlier, and identify patients with a worse prognosis.

Materials And Methods
This retrospective study used non-identi able patient data for analysis and was approved by the institutional review board. Therefore, the requirement for informed consent was waived.

Study Population
From January 21, 2020, to March 6, 2020, patients with con rmed COVID-19 were retrospectively reviewed. Inclusion criteria: (1) presence of at least two follow-up CT scans, (2) the patients' clinical data were complete.
Exclusion criteria: (1) patients without pulmonary lesions, (2) patients without lesions presenting as spherical GGOs on initial chest CT, and (3) the interval between initial and follow-up CTs was > 7 days.

Ct Examinations
All patients underwent a non-contrast CT upon admission to the hospital with follow-up CT scans at < 7 days.
Chest CT scans were performed using two MDCT scanners (Philips Brilliance iCT and Siemens SOMATOM go.Top) with the following parameters: tube voltage, 120 kV; automatic tube current modulation; beam pitch, 0.758/1.5; detector collimation, 0.625/0.6 mm; rotation time, 0.5 s; matrix size, 512 × 512; and section thickness and interval, 5.0 and 5.0 mm, respectively. All patients were in a supine position and scanned from the thoracic inlet to the lung base at the end of inspiration. CT images were reconstructed using a medium sharp reconstruction algorithm with a section thickness of 1 mm.

Analysis Of Ct Characteristics
Image analysis was evaluated on a PACS workstation (Vue PACS, Carestream Health, Inc) by two experienced radiologists (Z.G.C and F.J.L with 10 and 20 years of experience in thoracic radiology, respectively). They were blinded to clinical data and independently reviewed the images on both lung (width, 1500 HU; level, − 500 HU) and mediastinal (width, 350 HU; level, 40 HU) settings. Discrepancies were resolved by consensus.
For each spherical GGO, the initial CT scan was evaluated for the following characteristics: (a) lesion location, (b) relationship with the pleura (close to pleura, under the pleura, distant from the pleura), (c) lesion size, (d) CT value, (e) consolidation (yes or no), (f) uniformity of density (homogenous or heterogeneous), and (g) lesion border (illde ned or well-de ned). The relationship between the lesion and the pleura was considered as under the pleura if the distance between lesion and pleura was ≤ 2 cm; if > 2 cm, it was considered as distant from the pleura. All patients underwent follow-up chest CT scans, and the following results were reviewed: (a) the size and density changes of previous spherical GGOs, (b) changes of lesion border, (c) changes of consolidation, (d) the appearance of air bronchogram, and (f) the manifestations of residual lesions on the latest CT scan. Based on the changes of initial lesions on follow-up CT scans, spherical GGOs were divided into two groups: those that showed progression and those that were directly absorbed.

Clinical Features
Clinical and laboratory data of the patients were collected by one radiologist (W.J.L). Clinical data, including age, gender, clinical type, initial symptoms, length of hospitalization, numbers of scans, and the interval between the adjacent scans were recorded. Laboratory ndings such as white blood cell count, neutrophil count and percentage, lymphocyte count and percentage, C-reactive protein, erythrocyte sedimentation rate, and lactate dehydrogenase, were also recorded. These laboratory tests were performed upon admission to the hospital.

Statistical Analysis
All data were analyzed using SPSS 20.0 (SPSS, Chicago, Ill). Data were expressed as mean ± standard deviation for continuous variables and as numbers and percentages for categorical variables. The continuous variables used the analysis of Variance or Wilcoxon rank-sum test, and categorical variables were analyzed using the Pearson χ2 test or Fisher exact test. A p value of < 0.05 was considered signi cant.

Initial Ct Findings Of Spherical Ggos
The mean duration from illness onset to initial CT examination was 3 days. Initial chest CT ndings of 101 spherical GGOs are presented in Table 2

Follow-up Ct Scans
There was a total of 91 (mean, 3 ± 1; range, 2-6) follow-up CT scans with a mean interval of 5 ± 1 days (range, 1-10 days). The CT manifestations and changes of spherical GGOs during progressing to the peak or direct absorption are presented in Table 3. As progressing to the peak, the lesions usually showed a simultaneous increase in size and density (84.5%), a well-de ned border (81.7%), and newly emerging consolidation (25.4%) or increased consolidation (31.0%) ( Fig. 2A, B).  Data are expressed as n (%) Mixed GGO indicates GGO with brous stripes and/or consolidation Abbreviations: GGOs, ground-glass opacities.
Based on the initial CT manifestations of all pulmonary lesions, patients were divided into three groups: those with only spherical GGOs (group 1), those with spherical GGOs and a few patchy lesions (group 2), and those with patchy lesions and a few spherical GGOs (group 3). The absorption of spherical GGOs in different groups on the nal CT is presented in Table 4. Compared with groups 1 and 2, more spherical GGOs in group 3 progressed (p = 0.000). The absorption of lesions in the different groups differed signi cantly (p = 0.002). Compared with lesions in groups 1 and 2, those in group 3 had more residual GGOs (p = 0.000), especially the mixed GGOs (p = 0.007).  Abbreviations: GGOs, ground-glass opacities.

Discussion
The chest CT manifestations of coronavirus disease 2019 (COVID-19) pneumonia are various, including groundglass opacities (GGOs), brous stripes, and consolidation [12]. Previous studies investigating the clinical characteristics and CT changes of COVID-19 pneumonia had con rmed that spherical GGOs were one of the initial, common radiological manifestations [8,9,[13][14][15]. However, there were some cases with only patchy lesions on the initial scan; therefore, the clinical signi cance of spherical GGOs was unclear. Prior to understanding the signi cance of spherical GGOs, several questions must be answered. First, whether there is relationship between spherical and patchy GGOs. Second, how do spherical GGOs change following treatment. Third, whether there are differences in the outcomes of spherical GGOs with different initial manifestations. Last, whether the short-term follow-up CT manifestations of spherical GGOs are identical or not. In this study, all these issues were studied.
In the present study, most spherical GGOs were faint, heterogeneous, and ill-de ned, with a subpleural distribution on initial CT scan. During follow-up, most showed progression in size and/or density. During progression, their borders became more well-de ned and internal consolidation either appeared or increased. On subsequent axial CT imaging, most of the initial spherical GGOs lost their round shape and appeared as patchy lesions. The lesions extended toward the pleura with further expansion along the subpleural zone, indicating that distal alveoli were involved. Therefore, spherical GGOs could be seen as the precursors of patchy ones. This development process might be characteristic for COVID-19 pneumonia, and there should be a signi cant suspicion for COVID-19 when spherical GGOs showed such changes in an epidemic area.
Since spherical GGOs were early-stage lesions of COVID-19, it was important to know which kind of lesions would progress. We found that spherical GGOs with consolidation were more likely to progress than pure GGOs, suggesting that initial lesions with consolidation are more severe and have a greater possibility of progression. In addition, an increase in lesion size was often accompanied with the appearance or increase of internal consolidation. This indicates exacerbation of the lesion, which is consistent with a previous study [8]. Another study also noted that consolidation indicated disease progression or greater severity for COVID-19 [14]. In addition to internal consolidation, other pulmonary lesions may be related to lesion progression. Compared with patients with no or few patchy GGOs, more spherical lesions in patients with extensive patchy GGOs progressed. By understanding the progression of lung ndings over time, it should be possible to identify patients with a greater possibility for disease progression.
During absorption, progressed and directly absorbed spherical GGOs exhibited a simultaneous decrease in size and density. However, the progressed lesions had more residua on the nal CT. It is well known that lesion progression can cause more severe lung tissue injury, which results in a slower recovery. Progressed spherical GGOs usually developed into more extensive patchy lesions or had greater consolidation resulting in more severe alveolar and interstitial injury and edema. In contrast, initially pure spherical GGOs appear to be more easily controlled; therefore, most did not develop and had fewer residua.
There were two limitations to this study. First, the current results do not represent the natural development of spherical GGOs because most of the patients received treatment after the initial CT scan. However, we found some important changes in these lesions that can be used for resolving some clinical questions. Second, because some patients with COVID-19 pneumonia were asymptomatic at the early stage, they may have only typical patchy lesions on the initial CT scan. Therefore, it is impossible to con rm whether these patchy lesions originated from spherical ones due to a lack of earlier data. However, it is quite common that spherical GGOs develop into patchy ones, which may aid in the early diagnosis of this disease.

Conclusions
In conclusion, spherical GGOs are commonly detected in patients with early-stage COVID-19 pneumonia. Following treatment, most of them will progress, especially those with consolidation, and a few will absorb directly.
Progressed lesions usually develop into extensive patchy lesions or have more signi cant consolidation, and absorb more slowly than directly absorbed lesions. These characteristics may be helpful in the early diagnosis of this disease and in identifying patients with a potentially worse prognosis.