Study protocol and human population
This multi-center study was collectively conducted on 511 acutely symptomatic patients proved with COVID-19 during the period from June till October 2020. It was approved by the Institutional Ethics Committee. Patient consent was waived by the Research Ethics Board, assuring the respect of the confidentiality of patient’s data and medical records. The manuscript has no overlap with any previously published work.
Inclusion criteria were acutely symptomatic COVID-19 patients (during the 10 days from the onset of first clinical complaint) with positive PCR results and complete medical records.
Exclusion criteria were (1) asymptomatic patients, (2) incomplete medical records, (3) degraded quality of CT images with respiratory motion artifacts, (4) patients with known explanation for the clinical-radiological mismatch (48 patients were already excluded before the onset of the study) such as (A) pre-existing cardio-pulmonary or extra-pulmonary comorbidities such as emphysema, interstitial lung diseases, lung cancer, morbid obesity, co-existing neurological, cardiac, or abdominal diseases. All can impact the patient original O2 saturation and yield more deterioration of his condition not proportionate to the degree of lung involvement. (B) Vascular complications proved by CT pulmonary angiography among borderline-severe or severe/critical patients, such as acute pulmonary embolism.
Clinical evaluation
The clinical evaluation was performed by two consultant pulmonologists (having long time experience in the field of chest diseases; 18–20 years).
Patients were clinically classified according to the clinical symptoms, oxygen saturation, and respiratory therapeutic requirements into three groups [9]:
-
Group [A] (mild): patients with 95–100% O2 saturation/room-air (RA), absent or type I dyspnea, respiratory rate (RR) < 30/min, and no need for O2 support.
-
Group [B] (borderline severe): patients with 93–94% O2 saturation/RA plus type II or III dyspnea and/or tachypnea (RR ≥ 30/min) or both.
-
Group [C] (severe/critical): patients with < 93% O2 saturation/RA up to ARDS.
The patients in group [B] and [C] could be indicated for initial high flow nasal oxygen therapy up to mechanical ventilation.
CT machines and scanning parameters:
Multiple MDCT machines were used: (1) SOMATOM Sensation 64, Siemens Medical Systems, Germany, (2) Canon Medical Systems; Toshiba Aquilion 64, USA, and (3) Canon Medical Systems; Toshiba Aquilion CXL/CX 128, USA.
CT scanning parameters were slice thickness: 1–1.25 mm, tube rotation: 0.6–0.9 s, detector collimation 1 mm, 120–130 kVp, and 200 mA, FOV = 350 mm × 350 mm. Intravenous contrast administration was not used.
Study design and steps
The study included two major steps:
-
(1)
First step: standardized combined morphological/volumetric CT severity analyses and proposal of a new CTSI.
-
(2)
Second step: standardized blinded/independent validation analysis for the proposed new CTSI.
Step (1): standardized combined morphological/volumetric CT severity analyses and proposal of a new CTSI
It was conducted retrospectively during the period from June till September 2020. It included 379 acutely symptomatic COVID-19 patients. They were 242 males and 137 females (63.9%:36.1%). Their age ranged from 10 to 80 years (mean age 45.42 ± 20.1 SD). They were clinically classified into three groups: Group [A] (mild 298/79% patients). Group [B] (borderline severity 57/15% patients). Group [C] (severe or critical 24/6% patients).
CT images were analyzed in consensus with the availability of medical records by two expert consultant radiologists (having long time experience in thoracic imaging; 15 and 25 years). They performed combined CT volumetric and morphologic assessment followed by severity analysis and a new combined CTSI proposal.
[I] Volumetric/quantitative assessment [size of lesions (S)]
OsiriX MD 11.0 software (Pixmeo SARL, Geneva, Switzerland) was utilized for imaging review and volumetric/quantitative assessment of all patients, so the variability of the used MSCT machines would not impact the quantitative assessment. It was utilized for automated calculation of the total and pathological lung volumes based on threshold interval adjustment during the region of interest (ROI) 2D/3D color-coded reconstruction. (0:− 1024 Hu) the interval was set for total lung volume calculation and (0: − 700 Hu) interval was almost set for pathological lung volume calculation. Lesions were classified into four grades:
-
[S1] Patchy lesions involving (< 15% of lung volume),
-
[S2] Patchy lesions involving (15–25% of lung volume),
-
[S3] Patchy lesions involving (25–50% of lung volume),
-
[S4] Diffuse lung involvement (> 50% of lung volume).
[II] Morphologic assessment (M)
Eight morphologic HRCT patterns were traced:
-
[M1] Pure ground-glass opacities (GGOs) or solid nodules with peri-focal GG (halo sign).
-
[M2] GGOs with a peripheral organization “Atoll or reversed halo sign.”
-
[M3] GGOs mixed with consolidative changes.
-
[M4] Homogeneous or “curvilinear” consolidations; the latter is parallel to the pleural lining.
-
[M5] GGOs with “air bubble sign”; are small air-filled spaces within the ground-glass or consolidative changes, representing a cut section of sub-segmental bronchiolectasis sequel to fibrosis with focal air trapping.
-
[M6] GGOs with “early secondary fibrotic changes” and architectural distortion; are irregular linear streaks and atelectatic plates, representing early scarring of cellular components, which could be associated with bronchial wall thickening and bronchiectatic changes.
-
[M7] Patchy GGOs with smooth septal thickening “crazy-paving pattern,”
-
[M8] Diffuse alveolar damage (DAD) pattern, showing diffuse lung involvement with either of the above-mentioned CT patterns or mixed (notably crazy-paving pattern and air bubble sign).
Extra-parenchymal HRCT findings were also assessed, including significant nodal enlargement (short axis > 1 cm) and pleuro-pericardial effusions.
The overall time for this combined quantitative and morphologic CT assessment was estimated and ranged from 6 to 12 min.
[III] Statistical severity analyses and CTSI proposal
-
❖ Prevalence of HRCT characteristics among the total number of patients.
-
❖ Statistical analysis of significant relation between each HRCT volumetric or morphologic characteristic and clinical severity. Chi-square tests and P value measurements were performed using an online calculator (https://www.socscistatistics.com). P value (< 0.05) was considered statistically significant.
-
❖ Calculation of the severity ratio for each HRCT abnormality among each group from the following equation [Severity ratio = Prevalence rate of HRCT finding among the total number of each patients’ group/Prevalence rate of each patients' group among total sample size]. The total severity index is then summed and the CTSI model is created.
Step (2): standardized blinded/independent validation analysis for the proposed new CTSI
It was conducted prospectively during October 2020 on other 132 patients. They included 72 males and 60 females (54.5%:45.5%). Their age ranged from 22 to 67 years (mean age 41.13 ± 11.69 SD). CT images were analyzed independently by two expert consultant radiologists (having long time experience in thoracic imaging; 10 and 30 years). They were blinded from the clinical records.
The following statistical methods were utilized:
-
❖ Inter-observer agreement (IOA) using Cohen’s Kappa test.
-
❖ Prevalence, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), using an online diagnostic test calculator (http://araw.mede.uic.edu/cgi-bin/testcalc.pl).
-
❖ Receiver operating characteristics (ROC) curve, area under the curve (AUC), confidence interval, and accuracy, using QI Macros 2020 Excel software.