A comparison of radiographic features between non-survivors and survivors from ICU.

The clinical and imaging data of 121 ICU patients with SARS-CoV-2 infection (63 survivors and 58 non-survivors) were retrospectively reviewed. The clinical results and radiographic features were compared between survivors and non-survivors. Compared with survivors, non-survivors were more likely to develop ARDS (53 [91 %] vs. 22 [35 %], P < 0.0001), shock (6 [10 %] vs. 0, P = 0.009), cardiac injury(18 [31 %] vs. 6 [10 %], P = 0.003), acute kidney injury(21 [36 %] vs. 10 [16 %], P = 0.01), and pneumothorax(5 [9%] vs. 0, P = 0.017). There were typical radiographic features for ICU patients with SARS-CoV-2 pneumonia. Extensive air-space opacities could be seen in all patients. Middle and lower lung involvement was significantly more serious than upper lung (score 6.8 ± 1.9, 7.2 ± 2.1, and 5.7 ± 1.7, respectively, P < 0.0001). Based on X-ray involvement score, non-survivors were in a more critical condition than survivors (20.3 ± 4.6 vs. 19.1 ± 3.1, P = 0.038).

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pneumonia is clinically divided into mild, common, severe and critically ill types [[1], [2], [3]]. Critical ill patients need to be treated in intensive care unit (ICU), which has a high mortality rate. High resolution CT could identify typical ground-glass opacities (GGO), multifocal patchy consolidation, and crazy-paving sign in the peripheral area of the lungs of patients with SARS-CoV-2 pneumonia [[4], [5], [6]]. Follow-up CT can be used to determine whether pulmonary infection improves or progresses [7]. However, CT scan is difficult to be performed for patients in ICU, where mobile X-ray system serves as alternative in monitoring SARS-CoV-2 pneumonia. Many previous studies reported the clinical and imaging features [4,[7], [8], [9], [10], [11], [12]] of SARS-CoV-2 pneumonia. However, there are few studies comparing radiographic manifestations between survivors and non-survivors from ICU. We retrospectively investigated 121 ICU patients with SARS-CoV-2 infection (including 63 survivors and 58 non-survivors), and aimed to clarify the difference in X-ray manifestations between the two groups.

Materials and methods

This retrospective study was approved by our ethics committee, and written informed consent was waived.

The clinical data of 121 critically ill SARS-CoV-2 patients in ICU were collected and retrieved using the Radiologists Information System (RIS).The diagnosis of SARS-CoV-2 infection was confirmed by positive RT-PCR results, and the critically ill patients were defined as those admitted to ICU who required mechanical ventilation or those had shock [3] or those had a fraction of inspired oxygen (FiO2) of at least 60 % or more [13,14]. The recorded information included: age, gender, underlying diseases, initial symptoms, X-ray imaging signs and scores.

Acquisition of chest X-ray radiographs

Chest radiographs were all taken with a mobile X-ray scanner (uDR370i; United Imaging, Shanghai, China) in ICU. Scanning protocol was as follows: 80 kV, 3.2 mAs, 100-cm film-to-focus distance, with a broad tube focus. The radiograph images were reviewed on a picture archiving and communication system (Synapse; Fujifilm).

Image analysis

Two experienced thoracic radiologists (with 8 and 10 years of thoracic diagnostic experience, respectively) without knowing the patient’s clinical data assessed the X-rays. The radiograph signs and involvement scores were determined on consensus.

The involvement of lung on X-ray was assessed using a visual scoring method according to previous studies [15,16] as follows: each lung was divided into three equidistant zones (upper, middle, and lower) from the apex to the bottom (diaphragm), adding up to six zones together. For each zone: score 0, no involvement; score 1, 1%–25 % involved; score 2, 26 %–50 % involved; score 3, 51 %–75 % involved; score 4, 76 %–100 % involved. The total score was acquired by summing the scores of all the six zones, with the maximum value of 24. If a patient had multiple X-ray examinations, all X-rays were assessed, and the highest score was recorded.

Statistical analysis

Continuous variables were expressed as mean ± SD or median (IQR), and categorical variables as number (%). All data statistical analysis was performed with SPSS (22.0, IBM, USA). We assessed differences between survivors and non-survivors using two-sample t-test for normal distribution data or Mann-Whitney U test for non-normal continuous variables, and Chi-square test or Fisher’s exact-test for categorical variables. The Kruskal-Wallis test was used to compare involvement score among upper, middle and lower lung. P value less than 0.05 was considered significant difference.

Results

121 ICU patients with SARS-CoV-2 pneumonia were included in this study. The median age was 63 years. 78 (64 %) patients were men. 74 (61 %) patients had underlying diseases. Hypertension (26 %) and chronic cardiac disease (15 %) were most common underlying diseases, followed by malignancy (12 %) and chronic pulmonary disease (12 %). The most common initial symptoms were fever (92 %), cough (68 %), and dyspnea (49 %). 75 (62 %) patients developed acute respiratory distress syndrome (ARDS), 6 (5%) with shock, 31 (26 %) with acute kidney injury, 24 (20 %) with cardiac injury, and 5 (4%) with pneumothorax. 38 (31 %) patients were treated with invasive mechanical ventilation, 4 (3 %) with extracorporeal membrane oxygenation (ECMO), and 12 (10 %) with renal replacement therapy. All patients received antiviral therapy and antibacterial agents, and 48 (40 %) patients received corticosteroids (Table 1). The patients underwent one or more mobile X-rays. 119 (98 %) patients had bilateral infiltrates on chest x-ray. Extensive air-space opacities (Fig. 1, Fig. 2) could be seen in all patients. 24 (20 %) patients had pleural effusion, and 5 (4%) had pneumothorax (Fig. 3). Middle and lower lung involvement was significantly more serious than upper lung (score 6.8 ± 1.9, 7.2 ± 2.1, and 5.7 ± 1.7, respectively, P < 0.01).

Table 1

Main clinical information of 121 critically ill patients with SARS-CoV-2 infection. Comparisons of clinical features were performed between non-survivors and survivors using two-sample t-test for normal distribution data, and Chi-square test or Fisher’s exact-test for categorical variables.

Clinical features	all patients (n = 121)	non-survivors (n = 58)	survivors (n = 63)	P value
Age
mean age (IQR)	63 (50, 74)	68 (54,80)	58 (45,73)	<0.0001
Gender
male	78 (64 %)	39 (67 %)	39 (62 %)	0.540
female	43 (36 %)	19 (33 %)	24 (38 %)
Initial symptoms
fever	111(92 %)	56 (97 %)	55(87 %)	0.065
cough	82(68 %)	42 (72 %)	40(63 %)	0.294
dyspnea	59(49 %)	31 (53 %)	28(44 %)	0.322
malaise	33(27 %)	12 (21 %)	11(17 %)	0.651
diarrhea	8(7 %)	5 (9 %)	3(5 %)	0.393
myalgia	4(3 %)	2 (3 %)	2(3 %)	0.933
hemoptysis	3(2 %)	2 (3 %)	1(2 %)	0.511
headache	1(1 %)	1(2 %)	0	0.295
Underlying diseases
tuberculosis	8(7 %)	5 (9 %)	3(5 %)	0.393
diabetes	13(11 %)	8 (14 %)	5(8 %)	0.299
hypertension	32(26 %)	20 (34 %)	12(19 %)	0.054
cerebrovascular disease	9(7 %)	5 (9 %)	4(6 %)	0.634
chronic pulmonarydisease	15(12 %)	8 (14 %)	7(11 %)	0.655
chronic cardiac disease	18(15 %)	10 (17 %)	8(13 %)	0.483
malignancy	14(12 %)	8 (14 %)	6(10 %)	0.463
Hepatitis C	3(2%)	2 (3%)	1(2%)	0.511
goiter	2(2%)	1 (2%)	1(2%)	0.953
Complications
acute respiratory distresssyndrome	75(62 %)	53(91 %)	22(35 %)	<0.0001
shock	6(5 %)	6(10 %)	0	0.009
acute kidney injury	31(26 %)	21(36 %)	10(16 %)	0.01
renal insufficiency	6(5 %)	6(10 %)	0	0.009
cardiac injury	24(20 %)	18(31 %)	6(10 %)	0.003
gastrointestinalhaemorrhage	3(2%)	3(5%)	0	0.068
pneumothorax	5(4%)	5(9%)	0	0.017
pulmonary embolism	2(2%)	2(3%)	0	0.137
hemolytic anemia	1(1 %)	1(2 %)	0	0.295
cellulitis	1(1 %)	1(2 %)	0	0.295
Treatment
non-invasive mechanical ventilation	86(71 %)	49(84 %)	37(59 %)	0.002
invasive mechanical ventilation	38(31 %)	31(53 %)	7(11 %)	<0.0001
renal replacementtherapy	12(10 %)	9(16 %)	3(5%)	0.048
corticosteroids	48(40 %)	21(36 %)	27(43 %)	0.455

SARS-CoV-2=severe acute respiratory syndrome coronavirus 2; COPD = chronic obstructive pulmonary disease.

Fig. 1

A “white lung” at a 69-year-old male with laboratory-confirmed SARS-CoV-2 pneumonia. He died of respiratory failure two days after this scan.

Fig. 2

A 71-year-old male non-survivor with hypertension and coronary heart disease. His initial symptoms were fever and diarrhea. RT-PCR for SARS-CoV-2 was positive. Multiple air space opacities could be seen on first chest X-ray. Significant progress could be seen on second X-ray 4 days later (arrows). Involvement was more serious for the lower lung compared with upper lung.

Fig. 3

A 57-year-old female non-survivor with fever and diarrhea as the initial symptom. RT-PCR for SARS-CoV-2 was positive. Typical imaging features of SARS-CoV-2 pneumonia could be seen on first chest CT. Pneumothorax could be easily seen (arrows) on chest X-ray 17 days later, as well as extensive air space opacities and pleural effusion.

For 58 non-survivors, the median duration from onset of symptoms to death was 30 days, and the median duration from ICU admission to death was 9 days. The median time from last mobile X-ray to death was 2 days.

Compared with survivors, non-survivors were more likely to develop ARDS (53 [91 %] vs. 22 [35 %], P < 0.01), shock (6 [10 %] vs. 0, P = 0.009), acute kidney injury(21 [36 %] vs. 10 [16 %], P = 0.01), cardiac injury (18 [31 %] vs. 6 [10 %], P = 0.003), pneumothorax(5 [9%] vs. 0, P = 0.017) (Table 1).

Based on X-ray involvement score, non-survivors were in a more critical condition than survivors (20.3 ± 4.6 vs. 19.1 ± 3.1, P = 0.038). As summarized in Table 2, survivors and non-survivors did not differ in involvement score on zone level, but differed significantly in score of whole lung. Non-survivors were more likely to have pneumothorax than survivors (5 [9%] vs. 0, P = 0.017).

Table 2

Main radiographic features of 121 critically ill patients with SARS-CoV-2 infection. Comparisons of radiographic features were performed between non-survivors and survivors using Mann-Whitney U test for involvement score, and Chi-square test or Fisher’s exact test for categorical variables. The Kruskal-Wallis test was used to compare involvement score among upper, middle and lower lung.

Radiographic features	Critical ill(n = 121)	Non-survivors (n = 58)	Survivors (n = 63)	P value
bilateral pneumonia	119(98 %)	57(98 %)	61(97 %)	0.608
unilateral pneumonia	2(2 %)	1(2 %)	2(3 %)
mean involvement score (standard deviation)	19.7 ± 3.8	20.3 ± 4.6	19.1 ± 3.1	0.038
pleural effusion	24(20 %)	15 (26 %)	9(14 %)	0.111
pleural thickening	8(7 %)	4 (7 %)	4(6 %)	0.904
pneumothorax	5(4 %)	5 (9 %)	0	0.017
Involvement score
Six zones
right upper	2.4 ± 0.7	2.5 ± 0.8	2.3 ± 0.7	0.24
right middle	3.4 ± 0.7	3.5 ± 0.9	3.3 ± 0.8	0.28
right lower	3.6 ± 1.0	3.7 ± 1.1	3.5 ± 1.0	0.25
left upper	2.3 ± 0.9	2.4 ± 0.9	2.2 ± 0.8	0.19
left middle	3.4 ± 0.9	3.5 ± 1.0	3.3 ± 0.9	0.12
left lower	3.6 ± 1.2	3.7 ± 1.2	3.5 ± 1.1	0.17
Three regions
upper lung	5.7 ± 1.7	5.9 ± 1.8	5.5 ± 1.7	0.09
middle lung	6.8 ± 1.9	7.0 ± 1.9	6.6 ± 1.8	0.105
lower lung	7.2 ± 2.1	7.4 ± 2.1	7.0 ± 2.0	0.08
P value	<0.0001	<0.0001	<0.0001

SARS-CoV-2=severe acute respiratory syndrome coronavirus 2. ARDS = acute respiratory distress syndrome.

As summarized in Table 3, survivors and non-survivors differed significantly in multiple laboratory findings. Fig. 1 shows a “white lung” of a non-survivor with ARDS. Fig. 2 shows the rapid progress of SARS-CoV-2 pneumonia for a non-survivor. Fig. 3 shows pneumothorax in SARS-CoV-2 pneumonia. Fig. 4 shows the evolution of involvement score (with time) in part of patients.

Table 3

Main laboratory findings of 121 critically ill patients with SARS-CoV-2 infection. Comparisons of laboratory findings were performed between non-survivors and survivors using two-sample t-test for normal distribution data, or Mann-Whitney U test for non-normal data. Medians and IQR were provided in the table.

	All patient (n = 121)	Non-survivors (n = 58)	Survivors(n = 63)	P value
Leucocyte (109/L)	9.5 [7.1, 13.61]	11.64 [9.37, 15.61]	7.22 [6.1, 8.79]	<0.0001
Platelet (109/L)	130.4 [83, 188]	118 [63, 179]	144.5 [113.5, 227.75]	0.004
Erythrocyte (1012/L)	3.61 [3.01, 4.06]	3.48 [2.71, 3.89]	3.76 [3.59, 4.17]	0.001
Neutrophils (109/L)	8.72 [6.5, 11.12]	9.44 [7.36, 12.71]	7.8 [5.45, 9.37]	<0.0001
Lymphocyte (109/L)	0.6 [0.41, 0.81]	0.5 [0.32, 0.74]	0.70 [0.47, 0.93]	<0.0001
Hemoglobin (g/L)	117 [102.2, 128]	115.5 [91, 127]	120 [112.5, 129]	0.01
Glucose (mmol/L)	8.33 [7.24, 9.76]	9.11 [7.62, 13.66]	7.44 [6.46, 9.48]	0.019
Total protein (g/L)	60.6 [58.9, 65.8]	59.9 [56.8, 65.4]	61.9 [59.8, 67.9]	0.18
Globulin (g/L)	35.1 [31.25, 38.7]	37.9 [33.3, 40.7]	32.2 [29.25, 33.85]	<0.0001
Albumin (g/L)	30.1 [27.28, 33.5]	28 [25.08, 30.95]	32.3 [28.2, 37.2]	0.588
Creatinine (μmol/L)	84 [63, 109.5]	86 [66, 179.5]	82 [59, 103.5]	0.008
Uric acid (μmol/L)	189 [136.4, 330.9]	190.5 [114.5, 309]	188 [148, 362.4]	0.537
Total bilirubin (μmol/L)	14.7 [10.4, 18.5]	17.4 [11.5, 20.4]	11.4 [8.6, 13.4]	<0.0001
Direct bilirubin (μmol/L)	8.35 [6.15, 11.4]	9.75 [6.4, 14.4]	6.8 [5.35, 9.93]	<0.0001
Indirect bilirubin (μmol/L)	8.24 [5.72, 9.65]	9.15 [5.45, 11]	7.6 [5.8, 9.35]	0.083
Urea (mmol/L)	11.35 [7.28, 15.68]	14.55 [10.23, 20.08]	7.85 [6.98, 10.17]	<0.0001
Estimated glomerular filtration rate (ml/min/1.73m2)	75.7 [62.35, 89.2]	69.3 [41.35, 89.4]	82 [73.4,88.6]	0.008
Lactate (mmol/L)	1.98[1.09, 2.95]	2.7 [1.94, 3.48]	1.15 [0.95, 1.33]	<0.0001
Alanine aminotransferase (U/L)	40 [23, 68]	43 [24, 104]	37 [21, 57]	0.054
Aspartate aminotransferase (U/L)	38.4 [22.75, 67.15]	39 [17, 77.25]	38 [24.25, 61.75]	0.700
Myoglobin (ng/mL)	214.35 [127.45, 336.6]	280.6 [152.15, 736.8]	147 [84.45, 229.55]	0.002
High sensitive cardiac troponin I (pg/mL)	127.53 [48.68, 290.5]	202.2 [68.95, 460.18]	50.55 [44.5, 77.6]	<0.0001
MB isoenzyme of creatine kinase (ng/mL)	3.57 [1.1, 6.45]	5.85 [2.63, 11.93]	1.2 [0.7, 1.9]	0.014
Lactate dehydrogenase (U/L)	398.75 [351.4, 504.4]	490 [358.5, 591]	306.5 [281.5, 368.25]	<0.0001
Creatine kinase (U/L)	179 [82, 303]	180 [43, 503]	178 [84, 230.5]	0.241
Prothrombin time (second)	15.7 [13.79, 17.23]	16.5 [15.3, 19.4]	14.3 [13.25, 15.83]	<0.0001
Fibrinogen (g/L)	5.87 [4.69, 6.13]	5.92 [4.82, 6.3]	5.22 [4.59, 5.78]	0.069
Activated partial thromboplastin time (second)	45.3 [41.6, 48.45]	46.4 [42.5, 56]	43.5 [40.65, 46.8]	0.002
Thrombin time (second)	16.1 [14.3, 17.8]	17.5 [15.7, 20.6]	14.55 [13.3, 15.6]	0.446
D-dimer (μg/mL)	3.99 [2.16, 5.97]	5.47 [2.73, 12.52]	2.22 [1.82, 2.92]	<0.0001
Prothrombin activity	72 % [62 %, 82 %]	68 % [55 %, 75 %]	79 % [64 %, 91 %]	<0.0001
International normalized ratio	1.19 [1.03, 1.42]	1.3 [1.22, 1.56]	1.08 [0.99, 1.21]	<0.0001
Fibrinogen degradation products (μg/mL)	19.55 [9.6, 35.75]	29.65 [17.13, 62.65]	9.9 [9, 11.5]	<0.0001
Procalcitonin (ng/mL)	0.52 [0.24, 1.25]	0.97 [0.27, 2.58]	0.27 [0.13, 0.41]	<0.0001
N-terminal pro-brain natriuretic peptide (pg/mL)	2425.6 [1145.25, 3489.5]	3375.5 [1491.75, 8102.75]	1263 [522, 1483.5]	<0.0001
Ferritin (μg/L)	945.45 [803.5, 1632.56]	1064.5 [814.25, 2658.5]	826.8 [616.75, 1481.5]	0.137
Hypersensitive C-reactive protein (mg/L)	105.8 [68.94, 155.43]	142.7 [80.9, 209]	73.1 [42.3, 127.1]	<0.0001

SARS-CoV-2=severe acute respiratory syndrome coronavirus 2.

Fig. 4

Seven survivors and nine non-survivors from ICU exactly underwent three X-rays. X-ray images were given a involvement score according to the following rules: each lung was divided into upper, middle, and lower zones, adding up to six zones together; for each zone, score 0 = no involvement, 1 = 1 %–25 % involved, 2 = 26 %–50 % involved, 3 = 51 %–75 % involved, 4 = 76 %–100 % involved; the total score was acquired by summing the scores of six zones. The involvement score varied during disease course. Improvement could be seen in follow-up X-rays of survivors, In contrast, progress could be seen for non-survivors.

Discussions

We retrospectively reviewed clinical data and imaging data of 121 ICU patients with confirmed SARS-CoV-2 infection. Compared with survivors, non-survivors were more likely to develop ARDS, to have underlying diseases, and had higher X-ray involvement score, higher incidence of pneumothorax.

Refractory hypoxemia occurred one week after the onset of COVID and then deteriorated into ARDS in part of cases. ARDS is the fundamental pathophysiology of severe viral pneumonia [17,18], with a mortality at 28 days near 50 % [19]. ARDS was pathologically related to diffuse alveolar damage with cellular fibromyxoid exudates [[20], [21], [22], [23], [24]]. Previous studies reported that ACE2, the receptor for SARS-CoV-2, was expressed on myocytes and vascular endothelial cells, as well as on tubular cells, glomerular epithelial cells [25], resulting in potential direct cardiac and renal attack. Acute kidney or cardiac injury was observed in our cohort, which could have been related to direct effects of the virus, hypoxia, or shock. Non-survivors were more likely to develop acute kidney injury and cardiac injury.

The diagnostic value of chest CT for COVID-19 has been validated by many studies [[26], [27], [28]]. However, CT is not easily performed for ICU patients, especially when ICU is far from CT rooms. This study found mobile X-ray provided adequate image quality. Follow-up X-ray could also be used to determine whether pneumonia improves or progresses. Extensive lung involvement (or “white lung”) is the key radiographic feature of SARS-CoV-2 ARDS. The middle and lower lung involvement was more serious than upper lung. Lung involvement was more serious in non-survivors versus survivors. For survivors, significant improvement generally occurred in the follow-up X-ray. In contrast, progress of pneumonia occurred in most of non-survivors.

As for laboratory tests, lymphocytes were less in non-survivor compared to survivor, indicating that excessive immune response played an important role on pathogenesis of fatal SARS-CoV-2, and that the degree of lymphocytopenia was related to the severity of the disease. In our study, leukocytes and neutrophils were both high in non-survivors. It suggested that in critically ill patients, perhaps neutrophils were activated to induce the immune response, causing cytokine storms. LDH was a predictor in many diseases related to inflammatory reaction and tissue damage. CRP was a widely used biochemical indicator for inflammation, such as microbial invasion or tissue damage. We found that LDH and CRP were also significantly higher in non-survivors. We speculated that the virus may trigger a series of immune responses and induces cytokine storm in vivo. The level of inflammation indicators may correlate with the severity of the disease and prognosis.

This study has some limitations. First, this study was conducted at a single-center for severe SARS-CoV-2 patients; therefore, there may be selection bias. Second, most patients didn’t have CT images at ICU, which was a more accurate imaging method in monitoring the disease course. A larger cohort study of SARS-CoV-2 pneumonia from multiple centers would help further explore the disease.

In conclusion, there were typical clinical and radiographic features of ICU patients with SARS-CoV-2 pneumonia. Extensive air space opacities or “white lung” was the key radiographic sign for critical ill patients. Compared to survivors, non-survivors had more serious lung involvement, as well as a higher incidence of pneumothorax.

Ethical statement

This study was approved by the ethics committee of Tongji Hospital of Tongji Medical College of Huazhong University of Science and Technology (approval number: HUST-TJ-20200168). Written informed consent was waived.

Funding

This study was funded by 81801663.

CRediT authorship contribution statement

Gang Wu: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing. Shuchang Zhou: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no conflict of interest.