Evolution of 18F-FDG PET/CT Findings in Patients After COVID-19: An Initial Investigation

Visual Abstract

Thr oughout the United Kingdom, during February and March 2020 there was a rapid spread of coronavirus disease 2019 (COVID-19), which may result in viral pneumonitis and acute respiratory distress syndrome (1). The median time from symptom onset to intensive care admission was 10 d, although only 5% of patients were admitted (1). This is when antiviral responses are at a peak, suggesting that pneumonitis is a consequence of adaptive immunity (2).
Persistent respiratory symptoms affect at least one third of hospitalized COVID-19 patients, some of whom will have post-COVID-19 lung disease (PCLD) (3). Steroids are critical in reducing mortality from COVID-19, but their role in PCLD is less clear, and identifying those who might benefit may be difficult.
Currently, 18 F-FDG PET/CT has no role in the management of patients with COVID-19 (4), and there has been little investigation into the quantification and evolution of 18 F-FDG uptake in COVID-19 (Supplemental Table 1; supplemental materials are available at http://jnm.snmjournals.org). Given the growing role of 18 F-FDG PET/CT in interstitial lung diseases, the primary aim of this preliminary study was to assess the temporal evolution of 18 F-FDG uptake in COVID-19 and to correlate this evolution with clinical progression and recovery. A secondary aim was to investigate whether steroids could alter this evolution.

MATERIALS AND METHODS
The Institutional Review Board approved this retrospective study and waived the requirement to obtain informed consent. The challenges of the pandemic constrained the methodologic design, necessitating a retrospective approach.

Patient Selection
All studies performed in the department over the first U.K. peak of the coronavirus pandemic (March-April 2020) and from September 2020 to February 2021 (second peak) were assessed for acute COVID-19 by following the British Society of Thoracic Imaging guidelines or a confirmed history of COVID-19 in the electronic health record system (5). These studies included some of patients without positive polymerase chain reaction (PCR) test results, because of the poor availability of PCR tests in the early period. Also included were studies performed for persistent (.4 wk) respiratory symptoms, in keeping with PCLD, and studies of patients who had recovered from COVID-19 after the initial period. Ongoing treatment with steroids and other immunosuppressive drugs was recorded. Formal lung function tests were not performed because of infection risks. Acute studies between May and September 2020 were not examined because of the low prevalence and incidence of COVID-19 in London during that time (Supplemental Fig. 1; Supplemental Table 2).

F-FDG PET/CT Imaging Protocol
Patients fasted for at least 6 h, and blood glucose levels were recorded before injection of 400 MBq of 18 F-FDG adjusted for weight in keeping with the guidelines of the Administration of Radioactive Substances Advisory Committee (6). After an uptake time of 63. 10.9 min, whole-body PET scans of supine patients with their arms above their head were acquired at a rate of 2 min per bed position using a GE Healthcare Discovery 710 PET/CT scanner. A nonenhanced low-dose CT scan was acquired for anatomic coregistration and attenuation correction. Images were reconstructed using a resolution recovery iterative algorithm. All images were reviewed by at least one dually accredited radiologist-nuclear medicine physician. Quantification was performed by investigators with at least 10 y of experience in quantifying PET/ CT images of diffuse lung disease. PET analysis was performed with masking of clinical history and CT analysis.

Determination of Temporal Stage
After review of the clinical, CT, and electronic health records, the number of days since disease onset was estimated, and the acute COVID-19 cases were assigned to 1 of 2 temporal groups: early or late COVID-19 (7). Early COVID-19 (approximately #1 wk after disease onset) was defined predominantly as CT findings of ground-glass opacities with or without associated interlobular thickening. Late COVID-19 (.1 wk to #4 wk after disease onset) was defined as CT findings of increasing consolidation and signs of resolution marked by subpleural sparing, development of a fibrous stripe, and crescentic consolidation or a reversed halo or atoll sign. Patients who were asymptomatic after 28 d were classed as recovered patients. In addition, patients who were imaged because of persistent symptoms after 28 d were described as having PCLD. The CT component was correlated with other cross-sectional images to reduce the likelihood of incorrect classification due to breathing artifacts.

Quantitative 18 F-FDG PET Analysis
All images were processed using a standard protocol on a dedicated imaging workstation (ADW Volume Viewer, version 4.6; GE Healthcare), which calculated the lung target-to-background ratio (TBR lung 5 SUV max /SUV mins ) following methods described previously (8-10).

Statistics
The difference in 18 F-FDG PET uptake measures within the lung against temporal staging and pretreatment with steroids were assessed using the nonparametric Mann-Whitney test. Results were depicted using box-and-whisker plots. All statistical analyses were performed using SPSS, version 25.0 (IBM).
In 18 of the 50 patients, imaging was performed because of persistent shortness of breath and respiratory symptoms in keeping with PCLD. All 18 had been admitted to the hospital and had required oxygen. Fifteen of these patients previously had PCR tests positive for COVID-19, and COVID-19 was clinically diagnosed in the others. Nine had ongoing treatment with steroids for PCLD; the other 9 were not receiving treatment for their PCLD. All PCLD patients underwent repeated PCR testing confirming that they were PCR-negative before PET imaging (Supplemental Table 5).

Temporal Stage
After review of the CT component of the PET/CT (lung windows) and available clinical history, 8 (35%) of the 23 acute COVID-19 patients were determined to represent early COVID-19 and 15 (65%) late ( Fig. 1; Supplemental Table 5).
Association of Pulmonary 18 F-FDG Uptake with Temporal Staging in Early-and Late-Stage Disease 18 F-FDG uptake analysis of lung lesions in acute-disease patients demonstrated an increasing TBR lung over time, with progression from low-avidity ground-glass changes in the early stage to avid consolidation during the late stage (median values in the early stage: SUV max , 1.6, and TBR lung , 6.4; median values in the late stage: SUV max , 4.0, and TBR lung , 13.7). In acute-disease patients, TBR lung differed significantly between the early and late stages, with late-stage patients having a higher TBR lung than earlystage patients (P 5 0.001; Fig. 2). Among these acute-disease patients, a significant positive correlation was observed between TBR lung and estimated time since onset (r s 5 0.60, P 5 0.003; Fig.  3). This correlation was stronger when limited to acute-disease patients estimated to be in the first 3 wk of infection (n 5 18, r s 5 0.81, P , 0.001).

Pulmonary 18 F-FDG Uptake in PCLD
There was a lower TBR lung in patients who had received treatment with high-dose steroids (P 5 0.003) (Fig. 2) (median values in steroid-treated patients: SUV max , 2.4, and TBR lung , 6.62; median values in untreated patients: SUV max , 5.8, and TBR lung , 18.1). TBR lung was lower in asymptomatically recovered patients (median SUV max , 1.2; median TBR lung , 4.6) than in either untreated PCLD patients or those treated with steroids (P , 0.001 and P 5 0.020, respectively; P , 0.001 on Kruskal-Wallis testing for all 3 groups).

DISCUSSION
To our knowledge, this study was the first attempt to characterize the evolution of pulmonary 18 F-FDG uptake in patients with COVID-19 assigned a temporal stage (early stage to late stage to PCLD) based on clinical context and CT findings.
The increase in lung avidity with time suggests increasing lung inflammation (11,12) in acute COVID-19. In most cases, 18 F-FDG uptake would then be expected to decrease with viral clearance and establishment of immunity. There is, however, a subset of COVID-19 patients with delayed recovery who continue to show significant 18 F-FDG uptake, reminiscent of our findings in interstitial lung disease (8,9,13,14), and raising the possibility that COVID-19 pneumonitis is associated with an activated host immune response rather than direct viral pathology (12,15,16). It would be useful to understand the ability of lung avidity to predict the clinical course or the likelihood that post-COVID-19 interstitial lung disease will develop in this patient cohort.
The RECOVERY study (Randomized Evaluation of COVID-19 Therapy), which this study predates, demonstrated a survival benefit from steroid use in hypoxic patients with COVID-19 (15). In our study, several patients went on to develop an inflammatory organizing pneumonia characterized by persistent and increasing 18 F-FDG uptake. Steroid therapy is a recognized treatment for organizing pneumonia and other inflammatory interstitial lung diseases (15), and 18 F-FDG uptake was consistently lower in those cases treated with postdischarge steroids. Our findings raise the question of whether steroid administration has a role not just in acute hypoxia but in the later stages of COVID-19 and in PCLD. This question has been debated (15), with calls for a randomized, controlled trial to define the role of steroid therapy more widely. Although imaging may be useful, it is hard to determine from CT whether parenchymal changes indicate reversible inflammation or irreversible fibrosis. It is possible that 18 F-FDG PET/CT may offer a sensitive and specific biomarker to guide and rationalize steroid treatment.
Given the challenges of nuclear medicine imaging in the pandemic, this study has methodologic limitations. They are directly related to the infectious and emergent epidemic, the workload and severe capacity restraints of PET/CT departments, the need to protect staff and sterilize equipment, and the medical instability of seriously ill COVID-19 patients. These challenges limit patient numbers, preventing the use of a control group and longitudinal 18 F-FDG PET/CT imaging. Diagnostic CT will likely remain the most practical way to investigate acute COVID-19, although PET imaging may give potential mechanistic insights. However, PCLD patients are not currently believed to be an infection risk, and performing longitudinal 18 F-FDG PET/CT studies in this population may thus be realistic and feasible. This study was not prospectively designed to examine the use of steroids in PCLD; however, statistically significant lower 18 F-FDG uptake was observed in PCLD patients who received steroids than in those who did not. Finally, the lack of PCR testing in the first wave, as well as the high incidence of asymptomatic cases throughout the pandemic, creates uncertainties about prevalence, and retrospective analyses may therefore suffer from selection bias. Despite the design limitations, the findings of this study offer some insight into the development of pulmonary disease in COVID-19 patients and can help provide the evidence to justify performing formal prospective studies on this topic in the future.
CONCLUSION 18 F-FDG uptake in COVID-19 patients increases with time after infection and correlates with severity. Persistent 18 F-FDG uptake is seen in patients with PCLD disease. These findings suggest that future studies may be directed at the use of 18 F-FDG PET/CT to clarify the disease trajectory and may aid management of those patients with persistent respiratory symptoms RGB FIGURE 3. 18 F-FDG uptake (TBR lung ) against estimated time after onset of disease (on logarithmic scale), with superimposed regression using 23 acute (early and late) patients (F 1,23 5 14.94, P , 0.001; Spearman r s 5 0.595, P 5 0.003). Steroid treated 5 at least 10 d of high-dose steroid treatment. RGB FIGURE 2. 18 F-FDG uptake (TBR lung ) by temporal stage.