Case Studies in Physiology: Temporal variations of the lung parenchyma and vasculature in asymptomatic COVID-19 pneumonia: a multispectral CT assessment

Abstract

This study reports systematic longitudinal pathophysiology of lung parenchymal and vascular effects of asymptomatic COVID-19 pneumonia in a young, healthy never-smoking male. Inspiratory and expiratory noncontrast along with contrast dual-energy computed tomography (DECT) scans of the chest were performed at baseline on the day of acute COVID-19 diagnosis (day 0), and across a 90-day period. Despite normal vital signs and pulmonary function tests on the day of diagnosis, the CT scans and corresponding quantification metrics detected abnormalities in parenchymal expansion based on image registration, ground-glass (GGO) texture (inflammation) as well as DECT-derived pulmonary blood volume (PBV). Follow-up scans on day 30 showed improvement in the lung parenchymal mechanics as well as reduced GGO and improved PBV distribution. Improvements in lung PBV continued until day 90. However, the heterogeneity of parenchymal mechanics and texture-derived GGO increased on days 60 and 90. We highlight that even asymptomatic COVID-19 infection with unremarkable vital signs and pulmonary function tests can have measurable effects on lung parenchymal mechanics and vascular pathophysiology, which may follow apparently different clinical courses. For this asymptomatic subject, post COVID-19 regional mechanics demonstrated persistent increased heterogeneity concomitant with return of elevated GGOs, despite early improvements in vascular derangement.

NEW & NOTEWORTHY We characterized the temporal changes of lung parenchyma and microvascular pathophysiology from COVID-19 infection in an asymptomatic young, healthy nonsmoking male using dual-energy CT. Lung parenchymal mechanics and microvascular disease followed different clinical courses. Heterogeneous perfused blood volume became more uniform on follow-up visits up to 90 days. However, post COVID-19 mechanical heterogeneity of the lung parenchyma increased after apparent improvements in vascular abnormalities, even with normal spirometric indices.

INTRODUCTION

COVID-19 infection is known to affect lung parenchyma and cause multisystem vascular damage, including endothelialitis, micro- or macrothrombosis, end-organ infarct, and stroke (1–4). Other viral pneumonia, such as influenza, are not associated with such vascular insults, even when the clinical severity of the disease is comparable (1, 5). Severe COVID-19 disease is associated with consumption coagulopathy, including thrombocytopenia, prolongation of prothrombin time (PT), and activated partial thromboplastin time (aPTT), as well as elevated serum D-dimer and fibrinogen (6). Derangements in vascular physiology, as well as the lung parenchyma, are associated with subsequent morbidity and mortality.

Early in the COVID-19 pandemic, attention was mainly given to the qualitative imaging features [plain chest X-ray or computerized tomography (CT) scan] of lung parenchymal abnormalities [e.g., infiltrates and/or ground-glass opacities (GGO)]. As the knowledge of the disease grew, more attention was given to comprehensive assessments of lung parenchymal, airway, microvascular, and macrovascular effects of COVID-19. Pulmonary vascular disease associated with COVID-19 pneumonia is often apparent on autopsy and histopathology reports. However, the use of methods for noninvasive, in vivo assessments of the pathophysiology in mild COVID-19 disease, or before the formation of macrothrombosis in the form of pulmonary embolism is mostly unknown.

A combination of ultra-low dose noncontrast CT coupled with dual-energy contrast-enhanced lung CT (DECT) is a unique noninvasive methodology that can provide high resolution, quantitative assessments not only of the parenchyma and airways, but also perfusion via its surrogate: pulmonary perfused blood volume (PBV) (7). A recent study among patients with severe COVID-19 pneumonia showed qualitatively heterogeneous lung perfusion with decreased PBV on DECT (8). However, temporal physiologic changes in the parenchyma and vasculature among mild or asymptomatic cases, especially early in the disease, are unknown. Quantification of parenchymal and vascular pathophysiologies early in the disease course, as well as their variations over time, can advance our knowledge of COVID-19, identify patients at risk of progression, measure treatment response, and promote new therapies that may target specific structural and functional abnormalities.

In this case study, we describe a comprehensive assessment of the lung via paired inspiratory (total lung capacity, TLC) and expiratory (residual volume, RV) noncontrast CT coupled with contrast DECT to longitudinally assess parenchymal and vascular COVID-19 effects in a healthy nonsmoking male athlete. The research volunteer was asymptomatic and suspected of having COVID-19 pneumonia based on abnormalities noted on CT imaging (later RT-PCR confirmed) that was performed as a part of an unrelated research study. We report longitudinal changes in lung inflammation, parenchymal mechanics, and perfusion using previously validated quantitative CT (qCT) metrics (9, 10).

MATERIALS AND METHODS

Participant

The patient was a 32-yr-old Caucasian male marathon runner with no significant past medical history. He enrolled as a healthy, never-smoking control subject in a longitudinal CT imaging study (NIH R01 HL130883) at the University of Iowa, which was approved by the Institutional Research Ethics Board (IRB ID: 201706713). He reported no previous history of cardiovascular or lung disease, and no fever or subjective symptoms of viral infection during any visit. He reported having run 5 mi before coming in for the baseline study. He took no prescription medications throughout the duration of this longitudinal assessment, except for independently volunteering for a COVID-19 clinical trial which remains blinded.

Experimental Procedures and Measures

The study protocol, for which the subject volunteered and provided written, informed consent, included measurements at baseline (day 0), as well as follow-up visits at days 30, 60, and 90, which matched the same study pattern as followed by a group being studied in a smoking cessation program. Imaging protocols were consistent across the visits to ensure comparability, which also included review of symptoms, monitoring of vital signs, and full inspiratory and full expiratory ultra-low dose chest computed tomography scans (research dedicated Siemens SOMATOM Force) (11–13). Pre- and postbronchodilator pulmonary function tests with diffusing capacity for carbon monoxide (DLCO) was performed on days 0 and 90 visits before CT scanning, and a 6-min walk test was performed on day 90. The CT protocol included both noncontrast- and contrast-enhanced DECT scans at days 0, 30, and 90, but only a noncontrast DECT scan at day 60. All CT scans were obtained using standardized protocols on Siemens SOMATOM Force scanner (Erlangen, Germany). Noncontrast DECT scans were obtained during breath holds at total lung capacity (TLC) and residual volume (RV). Contrast-enhanced DECT scans on days 0, 30, and 90 included an additional scan obtained precisely 17 s after the initiation of prediluted 50% saline (43 mL) and 50% iodinated contrast (43 mL, Isovue 370, Bracco Diagnostic, Princeton, NJ) injection at a flow rate of 4 mL/s into the left antecubital vein. Contrast-enhanced images were obtained during a breath hold at functional residual capacity (FRC) initiated just before scanning, and were used for detecting vascular abnormalities and quantifying regional perfused blood volume (surrogate for perfusion) (7).

Data Analysis

The CT images were qualitatively reviewed for parenchymal, vascular, and incidental findings with a consistent methodology by one of the authors (PN), a cardiothoracic radiologist with 10 yr’ experience. Quantitative lung characterization was obtained from standardized density-based histograms as well as a three-dimensional adaptive multiple feature method (3-D-AMFM) for texture analysis, using an automated supervised machine learning software dubbed the Adaptive Multiple Feature Method or AMFM (14–18). The density-based methods characterize the lung into low (emphysema-like) attenuation areas (i.e., voxels less than −950 HU), and high (inflamed- and fibrotic-like) attenuation areas (i.e., voxels between −250 HU and −600 HU). By assuming the lung to be composed of air and tissue, one can report the percent air and tissue (which includes blood and extravascular fluid) on a voxel scale (19), as well as total and regional air and tissue volumes. Short-term changes in the so-called tissue volume can serve as an index of gain or loss of inflammation. Texture features classified the lung into normal parenchymal structures, bronchovascular structures, ground-glass opacities (GGO), mixed ground-glass and reticular (GG-reticular) opacities, consolidation, honeycombing, and emphysema. Further details on these techniques and their potential applications for COVID-19 pneumonia are detailed elsewhere (9). Segmentation of the lung parenchyma from the surrounding chest wall and mediastinal structures was achieved using a multiresolution convolutional neural network (20). The TLC and RV images at each time point were then matched voxel-by-voxel using a tissue mass preserving image registration technique, which warped the TLC image to the RV image. Regional parenchymal expansion was estimated by calculating the determinant of the corresponding Jacobian matrix (21). Coefficients of variation for each Jacobian determinant map (CV-Jac) were used to assess the heterogeneity of regional parenchymal mechanics. Additional regional metrics from image matching included an Anisotropic Deformation Index (22). Perfusion (measured as PBV) to the right and left lungs, heterogeneity of the lung perfusion (measured as PBV coefficient of variation; PBV-CV), and ventilation-perfusion ratio (V̇/$\dot{\text{Q}}$) was obtained using both in-house and scanner supported software to perform dual-energy material decomposition. In this method, high- and low-energy images are decomposed into an iodine attenuation component and a virtual noncontrast (VNC) component. The VNC image is used for registration and segmentation, whereas the iodine attenuation image when normalized to iodine attenuation in main pulmonary artery (MPA) shows the percentage of blood in each voxel and when multiplied by voxel volume results in blood volume in each voxel. We have shown previously that DECT-derived PBV measurement and lung perfusion quantification methodology is comparable with dynamic CT-based lung perfusion measurements (7). By image matching of the TLC image to the FRC virtual noncontrast image, we acquire regional Jacobians which represent a change in regional lung volume. Thereafter, each region of the lung was represented as a percent regional volume change relative to the whole lung volume change, we have a measure of % volume change or % ventilation (V̇). By representing the same region by a value of PBV calculated to be a % total PBV ($\dot{\text{Q}}$), we assessed an index of regional V̇/$\dot{\text{Q}}$. These regions were set to cubes of 8 × 8 × 6 voxels (0.6 mm on a side) along X, Y, and Z axes, resulting in ∼60 mm³ volumes of interest (VOI). For comparison, the volume of an average human acinus is estimated to be 187 mm³ (SD ± 79 mm³) (23).

RESULTS

Table 1 summarizes the physical characteristics, vital signs, pulmonary function test, and 6-min walk test parameters for the subject. Qualitative review of the CT images showed a unilateral ground-glass opacity (GGO) in the left lower lobe on day 0 (Fig. 1 with no apparent pulmonary vascular abnormalities). This radiographic finding was promptly communicated to the clinical lead of the research study (AC). Though the participant was asymptomatic, the high COVID-19 positivity rate in the State of Iowa prompted a reverse transcriptase-polymerase chain reaction (RT-PCR) test by nasal swab on the same day, which was reported positive. Subsequent CT scans at 30, 60, and 90 days were reported as visually normal, with complete resolution of the lung abnormality on follow-up examinations.

Table 1.

Summary of the physical characteristics, vitals, pulmonary function test, and 6-min walk test of the participant

Parameter	Day 0	30-Day	60-Day	90-Day
Physical characteristics
Height, cm	177	–	–	176.8
Weight, kg	66.1	–	–	67
Body mass index, kg/m2	21.09	–	–	21.43
Vitals
Temperature, °C	36.8	36.7	36.7	36.5
Heart rate, min−1	58	65	44	72
Respiratory rate, min−1	16	14	16	16
Oxygen saturation ($\text{Sp}{\text{O}}_2$), %	97	95	96	96
Spirometry
Vital capacity, L	4.61	–	–	4.64
RV, L	1.1	–	–	1.14
TLC, L	5.71	–	–	5.78
FVC pre-bronchodilator, L	4.54	–	–	4.74
FEV1 pre-bronchodilator, L	3.55	–	–	3.68
FEV1/FVC pre-bronchodilator, %	78	–	–	78
FVC post-bronchodilator, L	4.62	–	–	4.62
FEV1 post-bronchodilator, L	3.94	–	–	3.87
FEV1/FVC post-bronchodilator, %	85	–	–	84
DLCO predicted, %	36.99	–	–	36.99
DLCO measured, %	39.8	–	–	38.48
COHb, %	0.08	–	–	0.08
6-min walk test
HR (pre-test), min−1	–	–	–	56
$\text{Sp}{\text{O}}_2$ (pre-test), %	–	–	–	97
Dyspnea	–	–	–	0
Fatigue	–	–	–	0
Number of laps	–	–	–	6
Distance in partial lap, m	–	–	–	30
Total distance, m	–	–	–	390
HR (post-test), min−1	–	–	–	55
$\text{Sp}{\text{O}}_2$ (post-test), %	–	–	–	96
Dyspnea (post-test)	–	–	–	0
Fatigue (post-test)	–	–	–	0

BMI, body mass index; COHb, carboxyhemoglobin; DLCO, diffusing capacity for carbon monoxide; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; HR, heart rate; RV, residual volume; $\text{Sp}{\text{O}}_2$, oxygen saturation; TLC, total lung capacity.

Figure 1.

Longitudinal inspiratory axial lung window computer tomography (CT) images. The baseline examination, day 0 (A) demonstrated a ground-glass opacity in the left lower lobe. Subsequent CT image at 30-day (B) showed complete resolution of the left lung disease. Visually, no residual or recurring lung disease is seen on 60-day (C) and 90-day (D) follow-up visits.

The qualitative and quantitative parenchymal analysis data are summarized in Table 2, and representative images showing quantification methods of parenchymal characterization are highlighted in Fig. 2 with change in the CV-Jac summarized in Fig. 3. Note that there is a difference between plethysmographic/spirometric lung volumes and CT-derived lung volumes, consistent with body posture differences (24). Density-based quantification for lung parenchyma on day 0 showed a total lung volume of 5.29 L at TLC, with nonaerated tissue volume of 1.0 L. The density-based method quantified 5.8% of the lung as high-attenuation area (HAA) with voxel densities between −250 and −600 HU. The 3-D-AMFM texture quantification (9, 17) yielded a higher percentage of abnormal lung, characterized 10.3% as GGO and 0.5% lung as mixed ground-glass reticular. The heterogeneity in regional parenchymal expansion (Fig. 3) was measured by CV-Jac (value 0.26). On day 30, quantitative analysis of the CT scan showed improvement in total lung volume (6 L at TLC), with decreased but persistent high attenuation areas (4.5%), GGOs (4.7%), and ground-glass reticular areas (0.1%). The CV-Jac decreased slightly to 0.21, suggesting some improvement in regional expansion heterogeneity. After this initial 30-day improvement, the quantitative analysis of the CT images on days 60 and 90, however, showed decreased lung volume at TLC compared with day 30 (5.53 L on day 60 and 5.45 L on day 90) with increased high attenuation areas (5% on day 60 and 5.4% on day 90), increasing GGO (7% on day 60 and 8.5% on day 90), and ground-glass reticular disease areas (0.2% on day 60 and 0.3% on day 90). Consistent with an inflammatory process at day 90, the apparent tissue volume at both TLC and RV was elevated (Table 2). The CV-Jac also increased (0.30 on day 60 and 0.28 on day 90), suggesting increased heterogeneity of the lung expansion mechanics compared with days 0 and 30.

Table 2.

Summary of the qualitative and quantitative lung parenchymal characteristics

Parameter	Day 0	30-Day	60-Day	90-Day
Qualitative (visual) assessment
Overall study	Abnormal	Normal	Normal	Normal
Abnormality	GGO	–	–	–
Quantitative analysis
Segmentation and density-based analysis
Total lung volume at TLC, L	5.3	6.0	5.5	5.45
Total lung volume at RV, L	2.8	3.4	2.5	2.3
High attenuation area at TLC, %	5.8	4.5	5.0	5.4
“Tissue” volume at TLC, L	1.00	0.98	0.96	1.03
“Tissue” volume at RV, L	1.00	0.96	0.99	1.00
3D AMFM texture analysis at TLC
Ground glass opacity, GGO, %	10.3	4.7	7.0	8.5
GGO-reticular, %	0.5	0.1	0.2	0.3

3D AMFM, three-dimensional adaptive multiple feature method; GGO, ground-glass opacities; RV, residual volume; TLC, total lung capacity.

Figure 2.

Longitudinal lung quantitative computer tomography (CT) metrics in the subject. Density threshold (left column) quantifying the high-attenuation areas (HAA, orange) as 5.8% at baseline (day 0) which decreased to 4.5% at 30-day, but increased to 5.0% at 60-day and 5.4% at 90-day. Three dimensional (3-D)-AMFM texture quantification (middle and right column) which is a more sensitive method, quantified the ground-glass (GG, green) as 10.3% and ground-glass reticular (GGR, dark blue) as 4.7% on baseline day 0 examination. This decreased on 30-day follow-up with GG 4.7% and GGR 0.1%. However, the disease increased to GG 7.0% with GGR 0.2% at 60-day and GG 8.5% with GGR 0.3% on 90-day CT.

Figure 3.

Bar diagram showing change in the Jacobian coefficient of variation, Jacobian-CV at baseline (BL), 30-day (30 D), 60-day (60 D), and 90-day (90 D). The graph shows an initial improvement in the Jacobian-CV followed by worsening at the 60-day.

Qualitative PBV images and quantitative total and lobar PBV were obtained from the contrast-enhanced DECT scans on days 0, 30, and 90. The qualitative PBV images on day 0 showed considerable heterogeneity, which subsequently improved on days 30 and 90 (Fig. 4). The qualitative and quantitative PBV parameters are summarized in Table 3. Quantitative assessment of the DECT lung scans showed that the total PBV was apparently high at the baseline examination, day 0 measuring 725 mL with highest heterogeneity of perfusion, PBV-CV 68.3%. The total PBV decreased from 725 mL at baseline to 536 mL on day 30 and 402 mL on day 90, consistent with a reduced contrast enhancement (78 HU) in the main pulmonary artery (MPA) at baseline which increases to 192 HU and 265 HU for an identical contrast delivery protocol.

Figure 4.

Qualitative visualization of lung pulmonary blood volume (PBV) images at the baseline, 30-day, and 90-day computer tomography (CT). The baseline axial CT PBV (left column-middle row) and 3-D PBV images (left column-bottom row) demonstrate heterogeneous perfusion with mottled appearance. The 30-day and 90-day images (middle and right rows, respectively) demonstrate a more homogeneous PBV distribution with resolution of the mottled appearance.

Table 3.

Summary of the qualitative and quantitative lung perfusion characteristics

Parameter	Day 0	30-Day	90-Day
Qualitative (visual) assessment
Perfusion images	Diffusely abnormal	Normal	Normal
Quantitative analysis
Main PA, HU	78	192	265
PBV, mL	725	536	402
PBV-CV, %	68.3	64.9	63.0
V̇/$\dot{\text{Q}}$ mode	0.60	0.80	0.75
Poorly perfused volume, %	17	5.5	4.9

CV, coefficients of variation; PA, pulmonary artery; PBV, pulmonary blood volume.

The PBV histogram narrowed at 30 and 90 days (Fig. 5) and mean regional PBV-CV decreased to 64.9% on day 30 and 63.0% on day 90 as plotted in Fig. 5 (inset). As plotted in Fig. 6, the total lung V̇/$\dot{\text{Q}}$ modes were 0.6, 0.8, and 0.75 at baseline, 30, and 90 days, respectively. At the same time the poorly perfused lung VOIs (V̇/$\dot{\text{Q}}$ > 4.0) dropped from 17% to 5.5% and 4.9% of the whole lung, respectively. As a means of comparison in Fig. 7, we demonstrate the PBV distribution in two nonsmoking male subjects with different ages. Note the relatively more uniform distribution of PBV compared with baseline and day 30 distribution patterns seen in Fig. 4. Identical scanning, breath hold, and contrast delivery protocols were used in all subjects depicted in this report (25).

Figure 5.

Histogram demonstrating percent distribution of pulmonary perfused blood volume (PBV). Note the large amount of lung with zero or near zero PBV. At baseline (blue curve), there is a large amount of lung with zero or near zero PBV and significantly higher PBV variability (coefficient of variation) demonstrated in the inset bar graph. Follow-up CT at 30-day (orange) and 90-day (green) demonstrated decrease in PBV heterogeneity (bar graph inset) with a markedly narrowed histogram. The increased “apparent” PBV in some regions at baseline are hypothesized to be regions of iodine trapping due to inflammation-induced capillary leak.

Figure 6.

V̇/$\dot{\text{Q}}$ histogram. V̇ is derived from Jacobians mapped via image matching TLC noncontrast lung image to the virtual noncontrast functional residual capacity (FRC) image calculated from the dual-energy computed tomography (DECT) contrast scan. $\dot{\text{Q}}$ is derived from material decomposition applied to the DECT contrast scan. In the case of regional Jacobians and regional pulmonary blood volume (PBV), both are converted to a percent of their respective total lung values. % V̇/%$\dot{\text{Q}}$ provides V̇/$\dot{\text{Q}}$. This was generated at baseline (BL), as well as at days 30 and 90. Solid bars on the right show the percent of lung volume occupied by volumes of interest (VOI)s with V̇/$\dot{\text{Q}}$ over 4.0, representing effective dead space ventilation associated with low-to-no PBV. The modes of the V̇/$\dot{\text{Q}}$ relationships were 0.6, 0.8, and 0.75 at baseline, day 30, and day 90, respectively.

Figure 7.

Axial (left column) and three-dimensional (3-D) lung PBV (right column) in two healthy nonsmoking male subjects.

DISCUSSION

This longitudinal case study characterizes the temporal physiologic and mechanical changes of the lung perfusion, as well as lung parenchyma in an athletic, adult never-smoker with asymptomatic COVID-19 disease. This study demonstrates a variable and independent course in the lung parenchyma and microvasculature. The parenchymal abnormalities improved at 30 days after the initial diagnosis but worsened at the 60- and 90-day follow-up examinations. Changes in regional parenchymal mechanics showed initial improvement by day 30, with apparent radiographic resolution of infection. However, a mild increase in the heterogeneity of regional mechanics (Jacobians) became apparent on days 60 and 90 and were accompanied by increases in ground glass texture on days 60 and 90, associated with increased lung inflammation, leading in some to fibrotic lung disease after moderate to severe COVID-19 pneumonia (26, 27). It is important to note that our estimate of “tissue” volume represents total nonaerated lung volume, and thus represents an apparent gain or loss of blood and extravascular fluid volume—possibly indicative of altered perfusion associated with COVID-19 infection or inflammation. This assumes that parenchymal tissue mass was preserved.

The lungs showed significant heterogeneity of perfusion with increased total PBV at the time of acute COVID-19 pneumonia. The total PBV decreased on days 60 and 90. However, perfusion heterogeneity also demonstrated continued improvement during the follow-up scans on days 60 and 90. We may conjecture that the severe perfusion heterogeneity and increased total PBV are a manifestation of the effects of the SARS-CoV-2 virus on the microvasculature, which improved after the resolution of the infection. DECT-based study has shown apparent increase in iodine content (and hence PBV) in the areas of lungs that show parenchymal disease, but decreased perfusion apparently in healthy lung tissue (28, 29). The areas of parenchymal disease correspond to HAA on qCT. Loss of small size vessels (1.25–5 mm²) has also been demonstrated in a multicenter study (30). These previous studies, performed at the time of acute infection, highlight COVID-19 related vascular abnormalities (including peripheral hyperperfusion in the diseased segments) with loss of smaller-sized vessels, consistent with “back-up” of blood into larger vessels.

In our case, the initial apparent increase in PBV within perfused lung regions could also be explained in part by higher iodine content in the lung parenchyma due to increased capillary leak, which improved on follow-up. Extravascular capture of iodine would serve to erroneously assign a blood volume within the inflamed lung region. However, our finding that increased total PBV was coupled to a significant dilution of iodine in the MPA (Table 3) suggests that cardiac output at baseline may have been increased compared with day 30 (and may have been further reduced at day 90). One possible explanation for a higher cardiac output during the baseline measurement is that the subject completed a 5-mi run just before participating in the imaging protocol on day 0. Of course, this is only speculative.

Our assessment of V̇/$\dot{\text{Q}}$ must be recognized as an approximation to the physiology, since we scanned the subject at static lung volumes, with an inspiratory volume at TLC. This method has been shown to compare closely with dynamic xenon CT-derived ventilation methods (31). Furthermore, we utilize PBV as a surrogate for perfusion (7). Despite these limitations, the method provides an index of lung function which highlights a significant increased respiratory dead space at baseline (i.e., a large spike in values above V̇/$\dot{\text{Q}}$ above 4.0 in Fig. 6). A shift in the mode of this V̇/$\dot{\text{Q}}$ distribution from 0.6 at baseline to 0.8 at day 30 reflects a reduction in the number of poorly perfused lung regions. A shift in the modes from 0.8 at day 30 to 0.75 at day 90 was accompanied by an increased heterogeneity of the Jacobians (increased standard deviations) at days 60 and 90 following a drop in Jacobian heterogeneity at day 30. This, as mentioned earlier, was accompanied by an increased presence of ground glass texture following an initial decrease between baseline and day 30.

There is increasing recognition of multiorgan vascular effects from COVID-19. In the lungs, dilation of vessels within (or in proximity to) affected regions have been previously described (10, 28, 32). Enlargement of the subsegmental pulmonary artery branch has been shown to be a marker of severe disease in patients with COVID-19 pneumonia (33). The presence of hypo- or hyperperfused regions, and their association with normal and ground glass regions in severe disease, has been reported using subtraction CT angiography (29). Similar changes have also been demonstrated via DECT-PBV studies (32, 34). Such vascular pathophysiology is unique to COVID-19 pneumonia, given an increasing observation of severe hypoxemia even in patients with normal lung compliance (35). The similarity of regional hypoperfusion in COVID-19 to cases of high-altitude pulmonary edema (HAPE) have been noted (36), suggesting a role for pulmonary vasodilators. When compared with healthy controls and matched patients with acute respiratory distress syndrome, patients with COVID-19 showed significantly reduced blood volume in small caliber blood vessels on CT (37) and loss of smaller vessels (as detected by CT) has been shown to serve as a predictor of survival (30). Our case-study provides unique observations related to the demonstration of longitudinal quantification of parenchymal and vascular pathophysiology in an asymptomatic patient with mild disease. Our report also shows that even in mild cases, post-COVID-19 inflammation can be associated with regional derangements in parenchymal mechanics, which can dissociate from the vascular effects of infection. Further longitudinal characterization of COVID-19 pathophysiology is needed in subjects with varying disease severity to better understand the long-term effects of this disease.

Methodological Considerations

Iodine quantification depends on the amount of contrast within the lungs. Various hemodynamic factors can alter the contrast dilution, despite standardized injection protocols. To minimize the hemodynamic effects on lung PBV quantification, the PBV was calculated by using a normalization of the regional lung iodine content to the main pulmonary artery contrast enhancement. As an imaging site for various multicenter studies, we have performed extensive technologist training for standardization of the research and scan protocol. As such, the reproducibility and standardization of this methodology are dependent on the type of CT scanner and the use of standardized imaging protocols. Although we cannot confirm the statistical confidence in this case-study, a comparison of the pathophysiological findings in the same subject at four 30-day experimental visits using standardized methodology provides an internal control for confounders.

Conclusions

This case-study reports longitudinal qualitative and quantitative lung parenchymal and vascular pathophysiology of mild asymptomatic COVID-19 pneumonia in a nonsmoker adult. Vital signs, pulmonary function tests, and 6-min walk test for this participant showed no abnormalities and were all within normal limits for his demographic. Lung microvascular and parenchymal disease may follow different temporal trajectories, with abnormalities in perfusion showing continued and sustained resolution, but abnormalities in regional parenchymal mechanics showing persistent, and perhaps worsening characteristics. Routine visual qualitative review of CT images thus may not discern such quantitative abnormalities in an otherwise asymptomatic patient.

GRANTS

This study was supported in part by National Institutes of Health (NIH) R01 HL130883, R01 HL126838, and S10OD018526.

DISCLOSURES

Eric A. Hoffman and Joseph M. Reinhardt are cofounders and shareholders of VIDA Diagnostics, a company commercializing lung image analysis software developed, in part, at the University of Iowa. Junfeng Guo is a shareholder of VIDA Diagnostics. David W. Kaczka has received a research grant from ZOLL Medical Corporation, and is a cofounder and shareholder of OscillaVent, Inc. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

P.N., J.M.R., A.P.C., E.A.H., and D.W.K. conceived and designed research; P.N., J.M.R., A.P.C., E.A.H., and D.W.K. performed experiments; P.N., A.M., S.E.G., J.G., J.M.R., A.P.C., E.A.H., and D.W.K. analyzed data; P.N., A.M., S.E.G., J.G., J.M.R., A.P.C., E.A.H., and D.W.K. interpreted results of experiments; A.M., S.E.G., J.G., J.M.R., A.P.C., E.A.H., and D.W.K. prepared figures; P.N., J.M.R., A.P.C., E.A.H., and D.W.K. drafted manuscript; P.N., J.M.R., A.P.C., E.A.H., and D.W.K. edited and revised manuscript; P.N., A.M., S.E.G., J.G., J.M.R., A.P.C., E.A.H., and D.W.K. approved final version of manuscript.