INTRODUCTION

In December 2019, cases of pneumonia caused by a new coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) appeared in Wuhan, China. The resulting disease, Coronavirus disease-2019 (COVID-19) quickly spread worldwide within 3 to 4 months (1). By December 29, 2021, COVID-19 had been diagnosed in 281,808,270 patients worldwide, and 5,411,759 patients had died (2).

Computed tomography (CT) plays an important role in the diagnosis and management of COVID-19 pneumonia (3). CT findings continue to be analyzed. A better understanding of the changes in CT findings during the disease will facilitate the diagnosis of the disease, help determine the stage of the disease at the time of the CT examination and help identify complications that may develop. Serial CT scans of the thorax help clinicians better understand the development and progression of the disease. Our aim was to evaluate the temporal changes in thoracic CT findings during the disease.

METHODS

Patients

This study was approved by the Ethics Committee of University of Health Seciences Turkey, Prof. Dr. Cemil Tascioglu City Hospital (protocol number: 192) and the Republic of Turkey Ministry of Health, COVID-19 Scientific Research Committee. All patients gave informed consent. In 78 patients who were admitted to our hospital between March 10 and May 10 2020, COVID-19 was confirmed in the first or control reverse-transcription polymerase chain reaction, and at least two CT scans were performed in our hospital during the diagnosis and follow-up processes; the scans were retrospectively evaluated. Inclusion criteria were (1) known date of onset of the patient’s symptoms and (2) two or more thoracic CT evaluations performed in our hospital as of the date of admission. An age of younger than 18 was an exclusion criterion. Of the 78 patients, 23 were excluded from the study because their symptom onset dates were unknown. Data from 55 patients were included in the final analysis.

Computed Tomographic Imaging

Patients underwent at least two CT examinations in our picture archiving and communication system. All CT scans were performed without intravenous contrast material with the patients in the supine position during end inspiration. A Philips Ingenuity Core (Philips Medical Systems, Amsterdam, The Netherlands) was used for all chest CT examinations. The tube voltage was set at 120 kV with automatic tube current modulation. Thin-section CT slices were reconstructed by a lung algorithm with a slice thickness of 1.5 mm.

Image Evaluation

CT images were evaluated by two radiologists with 6 and 15 years of experience, respectively. Final decisions were reached by consensus. For the cases in which the radiologists disagreed, a third thoracic radiologist with 20 years of experience made a final decision. Ground-glass opacification (GGO), consolidation, the tree-in-bud sign, a crazy-paving pattern, and septal thickening were defined as elsewhere (4). GGO, honeycombing, consolidation, linear opacities, air bronchogram, bronchiectasis, pleural effusion, and mediastinal lymphadenopathy were among the findings recorded on thin-section CT. The extent of disease evident on thin-section CT was also evaluated. The distribution of abnormalities was also recorded. Lesions are involving mainly the peripheral one-third of the lung as peripheral, and lesions involving mainly the central two-third of the lung as central.

We quantified the CT images by using a previously published method (5-7) of scoring (0-5) for each lobe; the total possible score ranged from 0 to 25 (Table 1).

We categorized the CT scans according to the period between the onset of initial symptoms and the date when they were performed: Stages 1 (0-4 days, in 43 patients), 2 (5-9 days, in 31 patients), 3 (10-14 days, in 18 patients), 4 (15-21 days, in 8 patients), 5 (22-28 days, in 7 patients), and 6 (>28 days, in 25 patients) (6,7).

Statistical Analysis

To perform statistical analysis, we used NCSS 10 software (NCSS, LLC, East Kaysville, UT, USA). Normal distribution was checked with the Shapiro-Wilk test, histograms, Q-Q plots, and box plot graphics. The data were calculated as means and standard deviations, medians, minimums and maximums, frequencies, and percentages. We used Kruskal-Wallis One-Way analysis of variance to compare the total CT scan scores in each stage, and we used Dunn’s test for multiple comparisons. Intragroup variables were evaluated with Friedman’s repeated measures analysis of variance, and multiple comparisons were made with the Bonferroni-corrected Wilcoxon test (limit of significance was set as p<0.0167). Intragroup paired comparisons were evaluated with the Wilcoxon test. Fisher’s Exact probability test was used to compare nominal variables. A p value of less than 0.05 indicated bidirectional statistical significance.

RESULTS

Data from 55 patients (26 women and 29 men) were included in the analysis; the patients’ mean age was 44.6±11.5 years. Each patient had a median of 2.62±0.9 chest CT scans (Table 2). In total, 132 CT scans were evaluated.

The most prevalent symptoms at presentation were dry cough [in 40 patients (73%)] and fever [in 28 (51%)]. The distribution of lesions and changes in CT patterns at the six different disease stages were evaluated. At all stages, most of the lesions were peripheral and subpleural in location and in most patients, involvement was bilateral. The most common CT findings in all stages were GGO, consolidation, and crazy-paving patterns (Table 3).

Four of the 55 patients showed no pneumonia in the initial CT scan (Figure 1). Of the 43 patients in whom CT scans were performed in stage 1, 13 (30.2%) showed a unilateral involvement, and 9 (20.9%) showed a single lobe involvement (Table 3). The most common abnormalities in stage 1 were GGO, in 39 patients (91%); crazy-paving pattern, in 12 (28%); and consolidation, in 11 (26%). The incidence of crazy-paving pattern differed between stages (p<0.001) and was observed mostly in stages 2 and 3 (Figure 1). The frequency of consolidation also differed between the stages (p<0.001); it was observed least in stage 6 and most in stages 3 and 4 (Figure 2). The frequency of linear opacification differed between stages as well (p=0.04) and was observed mostly in stages 4, 5, and 6. Of the 25 patients who underwent CT examinations in stage 6, 22 (88%) exhibited GGO, 8 (32%) exhibited linear opacities, and only 2 (8%) exhibited no radiological abnormality (Figure 3).

Total CT scores differed between stages (p<0.001; Table 4). The total CT scores were highest in stages 3 (9.0±3.4) and 4 (10.5±6.2). The score of stage 1 was lower than those of stages 2, 3, and 4 (p<0.001). When we evaluated 43 patients in whom first CT scans were performed in stage 1, again total CT scores were highest in stages 3 and 4. Evaluation of the score differences between lobes revealed that the right middle lobe scores were lower than the other lobe scores. Right lower lobe scores were higher than the right upper and middle lobe scores (p=0.011 and p<0.001, respectively) and left lower lobe scores were higher than left upper lobe scores (p<0.001).

DISCUSSION

In our study, we evaluated the changes in CT patterns in pneumonia caused by COVID-19 from the onset of symptoms throughout the disease process.

In the early stage of the disease (0-4 days), GGO was the main radiological abnormality; crazy-paving patterns and consolidation were less common. This followed in prior reports (6-8). In stage 2 (5-9 days), the CT abnormalities became more diverse; crazy-paving patterns and consolidation increased significantly. In stages 3 and 4, the frequency of consolidation peaked (83.3% and 87.5%, respectively); then it decreased gradually. Again, GGO was the most prominent abnormality in the late stages (5 and 6). Findings of residual disease persisted in 23 (92%) of the 25 patients who underwent imaging in stage 6.

The highest total CT scores were recorded in stages 3 and 4. This result was different from that in Pan et al.’s (6) study; they reported that after 14 days, the infection was brought under control and that the consolidations started to be absorbed. The reason for this difference in results may be that only patients with mild COVID-19 pneumonia were included in Pan et al.’s (6) study. Our finding of the persistence of high CT scores was different finding from that of Pan et al. (6), but it was similar to those of Wang et al. (8). The increase in the incidence of GGO in the late stages was also similar to findings in Wang et al.’s (8) study. Increases in GGO in the late stages were also shown in patients with SARS (9). Moreover, in our study, the incidence of linear lines increased in the late stages (4, 5, and 6). Pan et al. (6) did not observe crazy-paving patterns after 14 days; in our study, in contrast, crazy-paving patterns were observed in stages 4 (25%) and 5 (28.6%).

CT scores were higher in both lower lobes, which followed in previous studies (6,7). Other results in our study that corroborated the results of previous studies included peripheral-subpleural lesions and predominant GGO and consolidation. This pattern of involvement was also similar to those in SARS and Middle East respiratory syndrome (MERS). CT findings in COVID-19 pneumonia showed typical lung damage from viral pneumonia, as in SARS and MERS (9,10).

Study Limitations

This study had two major limitations. First, group analysis of patients with mild and severe pneumonia was not performed. Second, scanning intervals among patients were not uniform because of the retrospective nature of this study.

CONCLUSION

In conclusion, the most common CT finding in COVID-19 pneumonia is bilateral peripheral GGO. On CT, lung involvement increases rapidly after the onset of the symptoms, peaks at approximately 10-14 days and continues at relatively high levels for some time thereafter. CT findings thus show a specific pattern during the disease process. On the basis of these patterns, clinicians can determine whether the disease is progressing or whether the patient is recovering. Knowing which pattern is dominant at which stage can help clinicians predict the course of the disease and determine the prognosis of the patient.

• The most common CT appearances in all stages were GGO, crazy-paving pattern, and consolidation.