Comparison of CT findings of coronavirus disease 2019 (COVID-19) pneumonia caused by different major variants

Patient backgrounds were standardized by applying the following inclusion criteria: (1) admitted to a single institution in Tokyo, Japan after being confirmed with COVID-19 with either real-time reverse transcription-polymerase chain reaction (RT-PCR) or rapid antigen test, (2) underwent at least one chest CT during hospital stay, (3) with clinical severity of mild or moderate based on the guidance statement issued by the Ministry of Health, Labour and Welfare [10]. Those categorized as severe case requiring ICU admission or mechanical ventilation were excluded. According to the weekly surveillance report on SARS-CoV-2 variant issued by a local municipal bureau in Tokyo (the Bureau of Social Welfare and Public Health), the majority of community-acquired cases were Alpha variant from April 19 to July 18 in 2021 and Delta variant from July19 to September 26 in 2021 [2]. Therefore, those who were being confirmed during the above-described periods were deemed to be infected with the variants dominating in the respective periods (hereafter referred to as Alpha variant and Delta variant, respectively). For wild type subgroup (hereafter referred to as wild type), conveniently sampled patients who were diagnosed with COVID-19 from September 1 to December 24 in 2020 before the first case of variant type of COVID-19 was confirmed in Japan were recruited. The study population comprised 118 patients (70 men; mean age, 56.8 years ± 20.7) with 128 CT scans for wild type, 137 patients (93 men; mean age, 49.4 years ± 17.0) with 146 CT scans for Alpha variant, and 137 patients (94 men; mean age, 45.4 years ± 12.4) with 140 CT scans for Delta variant.

Medical records were reviewed for the clinical and imaging findings of the subject patients. The following data were extracted from the medical records: age, gender, duration from onset to CT, symptoms and signs, underlying comorbidities, and smoking history. Disease severities were defined according to the guidance statement issued by the Ministry of Health, Labour and Welfare [10]. Disease severity was defined as mild if one had SpO2 ≥ 96% and without respiratory symptoms other than cough and not requiring oxygen administration; and moderate if one had SpO2 < 96% and dyspnea on breath (DOB) and pneumonia and requiring oxygen administration.

Enhanced or non-enhanced CT was performed using a 256-row multi-detector CT unit (Revolution CT; GE Healthcare, Milwaukee, WI, USA) with the following parameters: tube voltage, 100 kVp; collimation, 0.625 mm, helical pitch, 1.375, field of view, 36 cm; matrix size, 512 × 512 with optimized effective current under automatic exposure control (GSI Assist; GE Healthcare) based on the x-ray attenuation on anterior–posterior and lateral scout images. CT images were acquired during a single inspiratory breath-hold. A 2.0-mm gapless section was reconstructed for chest CT images before being reviewed on the picture archiving and communication system (PACS) monitor.

Image analysis was performed on a PACS monitor independently by two board-certified radiologists (A.F. and S.I., with 33 and 7 years of experience, respectively), who were blinded to the clinical data including patient management and COVID-19 subtype. Interobserver disagreements were resolved by consensus during a joint reading to determine the results. Parameters evaluated included: primary change, i.e., either one of pure ground-glass opacity (GGO), GGO with reticular opacity (crazy-paving pattern), or GGO with consolidation, or absence of the above lesions; repairing change, i.e., fibrotic strips, bronchial distortion, or subpleural line. Axial zonal distributions of the lesions were classified as peripheral predominantly (i.e., involving mainly the peripheral one-half of the lung), central predominantly (i.e., involving mainly the central one-half of the lung), or both. The numbers of affected lobes and laterality of the lesions were also evaluated. The extension of the lung opacities was evaluated using a semiquantitative chest CT severity score, with which lung lesions were estimated as the percentage per lobe [11]. For each lung, the extent of anatomical involvement was visually scored on a scale of 0–5 as follows: score 1, 1–5% involvement; score2, 6–25% involvement; score 3, 26–50% involvement; score 4, 51–75% involvement; and score 5, 76–100% involvement. Total lung scores were calculated as the sum of the individual lobe scores. The presence or absence of reversed halo sign, centrilobular nodule, tree-in-bud, pleural effusion, thoracic lymphadenopathy (as defined by lymph node size of ≥ 10 mm in short-axis dimension), and fatty liver was also assessed. CT findings were compared after being stratified by chronological disease stages (days 1–4, days 5–8, days 9–12, and days 13 or more).

Quantitative variables were expressed as mean ± standard deviation or median (interquartile range) based on the normality of data as defined by the Shapiro–Wilk test. Categorical variables were presented as the percentage of the total. The comparisons of quantitative variables were performed using a one-way repeated measures analysis of variance (ANOVA), Kruskal–Wallis test, Student's t-test or Mann–Whitney U test based on the normality of data and categorical data using the Pearson χ² test or Fisher’s exact test. Post hoc family-wise error correction for multiple comparisons was adjusted using the Bonferroni method. Statistical analysis was done using the Python environment (version 3.7.3) by using SciPy (version 1.2.1) [12]. All P values correspond to two-sided tests and the statistical significance level was set at Bonferroni corrected P < 0.05.

Demographics and clinical characteristics of the study population were summarized in Table 1. Patient age was significantly different between subgroups with more aged patients in wild type (all, p < 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 0.01; wild type vs. Delta variant, Bonferroni corrected p < 0.001; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.3). Clinical severity was significantly different between subgroups with more moderate patients in Delta variant than the other two (all, p = 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 1.0; wild type vs. Delta variant, Bonferroni corrected p = 0.003; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.03). Cardiovascular diseases was the only comorbidities that differed significantly being more frequent in wild type than the other two (all, p < 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 0.03; wild type vs. Delta variant, Bonferroni corrected p < 0.001; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.3). There was no statistically significant difference in gender, onset to CT, symptoms, comorbidities other than cardiovascular diseases, or smoking status.

The percentage of chest CT primary change was summarized in Fig. 1. Details of the CT findings were summarized in Tables 2, 3, 4, 5. After being stratified by chronological disease stages, patient age was still significantly different between the subgroups on days 1–4 (all, p = 0.03; wild type vs. Alpha variant, Bonferroni corrected p = 0.06; wild type vs. Delta variant, Bonferroni corrected p = 0.1, Alpha variant vs. Delta variant, Bonferroni corrected p = 1.0) on days 5–8 (all, p = 0.005; wild type vs. Alpha variant, Bonferroni corrected p = 0.4; wild type vs. Delta variant, Bonferroni corrected p = 0.003, Alpha variant vs. Delta variant, Bonferroni corrected p = 0.2). There was no statistically significant difference in clinical severity between the subgroups in any of the disease stages.

Regarding primary changes, GGO with consolidation was more frequent in the order of Delta variant, Alpha variant, and wild type in all disease stages. The proportion of each primary change in each disease stage were as follows. On days 1–4, “no involvement” was observed in 39%, 41%, and 11%, respectively, in wild type, Alpha variant, and Delta variant; pure GGO in 41%, 41%, and 22%; GGO with reticulation in 18%, 9%, and 28%; and GGO with consolidation in 3%, 9%, and 39% (all, p = 0.002; wild type vs. Alpha variant, Bonferroni corrected p = 1.0; wild type vs. Delta variant, Bonferroni corrected p = 0.01; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.003) (Table 2). On days 5–8, “no involvement” was observed in 14%, 5%, and 8%, respectively, in wild type, Alpha variant, and Delta variant; pure GGO in 46%, 34%, and 8%; GGO with reticulation in 31%, 42%, and 39%; and GGO with consolidation in 10%, 19%, and 45% (all, p < 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 0.2; wild type vs. Delta variant, Bonferroni corrected p < 0.001; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.003) (Table 3). On days 9–12, “no involvement” was observed in 13%, 3%, and 2%, respectively, in wild type, Alpha variant, and Delta variant; pure GGO in 19%, 3%, and 9%; GGO with reticulation in 44%, 60%, and 39%; and GGO with consolidation in 25%, 35%, and 50% (all, p = 0.06) (Table 4). On days 13 or more, “no involvement” was observed in 0%, 0%, and 0%, respectively, in wild type, Alpha variant, and Delta variant; pure GGO in 36%, 13%, and 0%; GGO with reticulation 21%, 38%, and 29%; and GGO with consolidation in 43%, 50%, and 71% (p = 0.08) (Table 5). Repairing changes were more frequent in the order of Delta variant, Alpha variant, and wild type in all disease stages. The proportion of repairing changes of the wild type, Alpha variant, and Delta variant were as follows: 28%, 15%, and 61% on days 1–4 (all, p = 0.03; wild type vs. Alpha variant, Bonferroni corrected p = 0.8; wild type vs. Delta variant, Bonferroni corrected p = 0.3; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.03) (Table 2); 41%, 48%, and 80% on days 5–8 (all, p < 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 1.0; wild type vs. Delta variant, Bonferroni corrected p < 0.001; Alpha variant vs. Delta variant, Bonferroni corrected p < 0.001) (Table 3); 63%, 73%, and 96% on days 9–12 (all, p = 0.001; wild type vs. Alpha variant, Bonferroni corrected p = 1.0; wild type vs. Delta variant, Bonferroni corrected p = 0.003; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.003) (Table 4); and 86%, 86%, and 94% on days 13 or more (p = 0.815) (Table 5). There was no statistically significant difference in the presence of subsidiary findings including pleural effusion, lymphadenopathy, fatty liver, axial distribution of the opacities, and laterality of the involved lobes.

Chest CT severity scores of each subtype were summarized in Table 6. Median (with [interquartile range]) total chest CT scores of wild type, Alpha variant, and Delta variant were as follows: 2 [0–4], 1 [0–3.75], 6.5 [3–8] on days 1–4 (all, p = 0.008.; wild type vs. Alpha variant, Bonferroni corrected p = 1.0; wild type vs. Delta variant, Bonferroni corrected p = 0.01; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.002); 5 [2.5–8], 7 [4–10], and 9 [5–11] on days 5–8 (all, p = 0.02; wild type vs. Alpha variant, Bonferroni corrected p = 0.2; wild type vs. Delta variant, Bonferroni corrected p = 0.005; Alpha variant vs. Delta variant, Bonferroni corrected p = 0.5); 7 [5.25–9.25], 8 [7–12], and 11 [10–14.25] on days 9–12 (p = 0.04); and 10 [8.7–14.5], 12 [10.6–14.75], and 13 [12.2–14] on days 13 or more (p = 0.7). Representative CT images of each subtype were shown in Fig. 2.