Clinical and radiological imaging as prognostic predictors in COVID-19 patients

Background Since the announcement of COVID-19 as a pandemic infection, several studies have been performed to discuss the clinical picture, laboratory finding, and imaging features of this disease. The aim of this study is to demarcate the imaging features of novel coronavirus infected pneumonia (NCIP) in different age groups and outline the relation between radiological aspect, including CT severity, and clinical aspect, including age, oxygen saturation, and fatal outcome. We implemented a prospective observational study enrolled 299 laboratory-confirmed COVID-19 patients (169 males and 130 females; age range = 2–91 years; mean age = 38.4 ± 17.2). All patients were submitted to chest CT with multi-planar reconstruction. The imaging features of NCIP in different age groups were described. The relations between CT severity and age, oxygen saturation, and fatal outcome were evaluated. Results The most predominant CT features were bilateral (75.4%), posterior (66.3%), pleural-based (93.5%), lower lobe involvement (89.8%), and ground-glass opacity (94.7%). ROC curve analysis revealed that the optimal cutoff age that was highly exposed to moderate and severe stages of NCIP was 38 years old (AUC = 0.77, p < 0.001). NCIP was noted in 42.6% below 40-year-old age group compared to 84% above 40-year-old age group. The CT severity was significantly related to age and fatal outcome (p < 0.001). Anterior, centrilobular, hilar, apical, and middle lobe involvements had a significant relation to below 90% oxygen saturation. A significant negative correlation was found between CT severity and oxygen saturation (r = − 0.49, p < 0.001). Crazy-paving pattern, anterior aspect, hilar, centrilobular involvement, and moderate and severe stages had a statistically significant relation to higher mortality. Conclusion The current study confirmed the value of CT as a prognostic predictor in NCIP through demonstration of the strong relation between CT severity and age, oxygen saturation, and the fatal outcome. In the era of COVID-19 pandemic, this study is considered to be an extension to other studies discussing chest CT features of COVID-19 in different age groups with demarcation of the relation of chest CT severity to different pattern and distribution of NCIP, age, oxygen saturation, and mortality rate.


Background
On December 2019, an unidentified cause of pneumonia affected a cluster of a population in Wuhan, China [1]. On January 7, 2020, a novel strain of coronavirus was extracted from the patients' respiratory tract secretions given the name of severe acute respiratory syndrome coronavirus 2 (SARSCoV2) [2,3]. On March 11, 2020, WHO characterized COVID-19 as a pandemic [4]. On September 1, 2020, confirmed cases of COVID-19 reported to WHO were 25, 298,875, including 847,602 deaths with the highest number of reported positive cases was in the USA, while the lowest number was in the Western Pacific [5].
Several studies reported a higher sensitivity of chest CT in comparison to real-time reverse transcriptase-polymerase chain reaction (RT-PCR) (98% vs. 71%, p < 0.001) [6]. RT-PCR test results take hours or even days to be available, which may delay COVID-19 patients' triaging. Furthermore, RT-PCR tests are resource-constrained, which may limit its accessibility to all suspected patients. So, chest CT is considered now as a useful supplementary tool for RT-PCR [7]. Fang et al. [6] advocated using chest CT as a screening tool for SARS-CoV-2 for patients with clinical and epidemiologic features matched with COVID-19 infection even with negative RT-PCR.
Several observational studies, systematic reviews, and meta-analysis have been published discussing the clinical, laboratory, and imaging features of COVID-19 [8][9][10][11][12][13]. However, the relations between radiological features and the clinical aspects of COVID-19, particularly the relations between CT severity and age, oxygen saturation, and fatal outcome, need to undergo further investigations. These relations can confirm the prognostic value of chest CT in COVID-19 patients. Consequently, in this study, we tried to enrich the database with COVID-19 characteristics in Egypt by describing the imaging features in different age groups and underlining the relations between CT severity score, age, oxygen saturation, and fatal outcome.

Ethical considerations
The Zagazig University ethics committee approved this study (approval no. 6381; approved May 10, 2020). A written informed consent was obtained from all patients in this study. The study was conducted according to the ethical principles of the declaration of Helsinki. This manuscript was reported in adherence with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE statement guidelines).

CT image acquisition
Non-contrast enhanced CT scans were obtained on the day of patients' hospital admission using Aquillion lightning (Anon, Japan). The scanning area extended from the level of the upper thoracic inlet to the inferior level of the costophrenic angle with the following parameters: detector X collimation widths = 16 × 1/0.5 mm (adult/child); tube voltage = 120/ 80 kV (adult/child). The tube current was regulated by sure exposure 3D; AIDR (adaptive iterative dose reduction) 3D. The CT scans were acquired at the end of inspiration in co-operative patients. Reconstructed images were obtained with a slice thickness/interval = 1/1 mm. The reconstructed images were transferred to the workstation and picture archiving and communication systems (PACS) for postprocessing.

Image analysis
All CT images were interpreted in consensus by two radiologists (W.M and M.M with 12 and 8 years of chest imaging experience, respectively). The radiologists were blinded to the patients' clinical data and laboratory results. The following features were evaluated on CT images: (a) distribution of the lung insult: (1) unilateral or bilateral, (2) involved axial aspect of the lung: posterior or anterior, (3) regional distribution within the lung: hilar, centrilobular, or pleural-based, (4) the most predominate distribution: centrilobular, pleural-based, or no dominate distribution, (5) lobar involvement: lower, upper, middle, or apical, and (6) presence of lower lobe predominance; (b) imaging pattern based on terms provided by Fleischner Society [14]: (1) ground glass opacity   (4) severe (≥ 60%).

Statistical analysis
Analyses were done using SPSS version 20.0 (IBM, Armonk, NY). The normality of distribution was assessed using the Kolmogorov-Smirnov test. Normally distributed data were presented as mean and standard deviation (SD), and categorical data as frequency and percentage. The difference between groups was analyzed by using Chi-square and Fisher exact tests (for categorical data). The correlation was done to detect the linear relationship between two numerical variables using the Pearson correlation coefficient. ROC curve was used to detect the cutoff value of age highly exposed to be infected with COVID-19. P value ≤ 0.05 was considered significant.

The relation between positive CT findings and age
On comparison between below and above 40-year-old age groups regarding positive CT findings, a strong significant relation was noted (p < 0.001) ( The relation between CT severity score and age A significant relation was noted between age and CT severity score (p ˂ 0.001) ( Table 4). The ROC curve analysis revealed that > 38-year-old was the optimal cutoff value of age that was highly exposed to develop moderate and severe stages of NCIP (AUC = 0.77, 95% CI = 0.72-0.82 p < 0.001).
The relation between CT severity score and CT features All lobar and regional lung distribution had nearly the same impact on CT severity score except posterior lung affection (p = 0.57) ( Table 5).     The relation between CT features, severity score, and oxygen saturation Most CT features had a significant relation with oxygen saturation except for pleural-based lower lobe distribution and GGO (Table 6). A significant negative correlation was noted between CT severity and oxygen saturation (r = − 0.49, p ˂ 0.001) (Fig. 6). A significant relation was noted on comparing moderate and severe stages of the CT severity score with oxygen saturation, with 73.3% of severe cases developed ≤ 90% oxygen saturation (p < 0.001).
The relation between CT features, severity score, and fatal outcome A significant relation was statistically confirmed between CT severity score and fatal outcome manifested by increasing the number of deaths in moderate and severe stages (p < 0.001). A significant relation was noted between the apex and hilar distribution and fatal outcome (p < 0.001) ( Table 7).

Discussion
Our results demonstrated that the incidence of NCIP was statistically higher in the above 40-year-old age group with progressively increasing the CT severity score with advanced age. The most susceptible age to develop moderate and severe stages of NCIP was > 38 years old. The severe stage of NCIP included the highest percentage of patients with ≤ 90% oxygen saturation and with fatal outcome.
In our study, the CFR was 4.3%. Previous studies [13,[15][16][17] showed CFR ranging between 4.3 and 15%. This value may be attributed to the presence of variable stages of the disease severity in addition to variable age groups in our admitted patients. In our research, the CFR was age-linked, increased from 0.8% in the young age group to 16.6% in the old age group. This finding is endorsing with Ferguson et al. [18], who reported that CFR increased with age from < 0.6 to 2.2% at 60 years old and reached over 9.3% at 80 years old.
Chest CT findings of NCIP regarding different age groups were described in two studies [26,27]. Song et al. [26] divided the studied population into above and below 50 years and found severe lung involvement with consolidations in above 50-year-old patients, while others younger than 50 years had more GGOs. Another study [27] classified 72 symptomatic patients into above and below 60 years and documented severe multilobar  . However, in our study, GGO and consolidation were observed more in the above 40-year-old group, but with no statistically significant relation (p = 0.29 and 0.06, respectively). A significant relation was proved between CT severity and older age groups. Bilateral and anterior aspect involvement of the lung had an obvious relation with the severe stage. However, no specific pattern or distribution had a predilection to the severe stage more than the others.
Resting oxygen saturation < 95% is considered abnormal [29]. Karimi et al. [30] established oxygen saturation < 93% with oxygen assistance or < 90% at room air as a sign of severe pneumonia. Other studies also depended on oxygen saturation ≤ 90% by pulse oximetry to define hypoxia in the pneumonia severity index score [31,32]. According to Andrea et al. [33], mixed GGO and consolidation was noted in critically ill patient with lower oxygen saturation while GGO only was noted in non-critically ill patient with higher oxygen saturation. In our study, GGO and consolidation had nonsignificant relation to oxygen saturation while crazingpaving was the main pattern associated with below 90 % oxygen saturation. Up to our knowledge, no study has correlated the distribution of NCIP and oxygen saturation. In our study anterior, centrilobular, hilar, apical, and middle lobe involvements had a significant relation to below 90% oxygen saturation. A significant negative correlation between CT severity and oxygen saturation was statistically proved.
In our study, crazy-paving, anterior aspect, hilar, centrilobular involvement, and moderate and severe stages had a statistically significant relation to higher mortality. The relation between crazy-paving pattern and higher mortality may be explained by the autopsy results from cases with crazing-paving that revealed diffuse alveolar damage with different stages of inflammation and fibrosis [34].
Finally, the relations mentioned above approve the prognostic value of CT in NCIP, which may predict the outcome of the COVID-19 patients and alter the management strategy in a trial to decrease the disease morbidity and mortality.
We encountered limitations in our study. First, chest CT was done for all patients on the day of hospital admission regardless of the onset of symptoms. Second, we did not acquire pulmonary CT angiography to evaluate the possibility of thromboembolic lung affection. Third, we did not take laboratory findings of our patients into consideration. Lastly, we did not consider the impact of comorbidity factors on CT severity, oxygen saturation, and fatal outcome.

Conclusion
In COVID-19 patients, CT severity is age-related, and most severe cases occur in the old age group. The prognostic value of CT in NCIP emits from the confirmed relations between CT imaging features, CT severity, oxygen saturation, and outcome of the patient. In consensus, CT could be considered the main determining factor in the management strategy of COVID-19 patients.