Incidental mediastinal masses detected on chest computed tomography scans during the COVID-19 pandemic

Abstract

OBJECTIVES

The prevalence of mediastinal masses in large-scale populations in China has been rarely reported. During COVID19 pandemic, many incidentalomas were reported due to the large amount of chest computed tomography scan performed in emergency setting.

METHODS

Retrospective analysis of emergency chest computed tomography scans (February 2020–February 2021) for COVID-19 screening, including mediastinal abnormalities (excluding lymph nodes, dysplasia, pneumomediastinum and other non-mass alterations), with computed tomography features, diagnostic workup and 1 year follow-up data were reviewed.

RESULTS

Of the 40 112 patients [mean age 54.5 (17.2) years; male-to-female ratio 1.02:1] screened for COVID-19, 293 (0.73%) had mediastinal masses of which 223 (0.56%) located in the anterior mediastinum. As participants aged, the prevalence tended to increase (P < 0.001). The prevalence was not different between the sexes (P = 0.635). An oval shape, anterior mediastinal location, and thymus involvement were the most common computed tomography characteristics. Surgery confirmed 11.3% (33 of 293) of nodal lesions, with a benign to malignant ratio of 51.4: 48.5. A computed tomography scan follow-up was conducted in 32.3% (84/260) of the patients, and in 82.1% (69/84) of cases the lesion was stable. Additionally, mediastinal masses were detected in 7.7% (20/260) of elderly patients who passed away soon after their primary disease worsened.

CONCLUSIONS

In Chinese COVID-19 screening chest computed tomography, the prevalence of all mediastinal masses and anterior mediastinal masses was 0.73% and 0.56%, respectively. Findings support risk-stratified management: growing/suspicious lesions warrant intervention versus surveillance for stable masses. Standardized protocols and multidisciplinary consensus are critical.

Open in new tabDownload slide

INTRODUCTION

Mediastinal masses are relatively common findings, with approximately half of these lesions located in the anterior mediastinum [1]. The increasing use of chest computed tomography (CT) imaging in clinical practice, particularly for lung cancer screening [2], has led to a higher detection rate of asymptomatic mediastinal lesions [3, 4]. At the beginning of 2020, COVID-19 pandemic prompted a global surge in chest CT examinations and nucleic acid tests, particularly in outpatients and emergency settings. This unprecedented volume of chest CT imaging, primarily aimed at diagnosing COVID-19, inadvertently generated a large dataset comparable to that of a population-based chest CT screening programme. While the primary focus of these scans was to assess pulmonary conditions, incidental findings such as mediastinal masses were frequently detected [2]. Several reports were published on this topic: the 1st study was from the Early Lung Cancer Action Project (ELCAP) [3], the 2nd was from the Framingham Heart Study [4], the 3rd used data from 2 health check-up centres in Korea [5] and the 4th used low-dose CT examination data from the Japanese health check-up population [6]. These studies provided valuable insights into the epidemiology of mediastinal masses across different countries. However, no similar studies have been conducted in the Chinese population, leaving a gap in understanding the prevalence and characteristics of these lesions in this demographic. The incidental detection of mediastinal masses during COVID-19 screenings offers a unique opportunity to compare these findings with those from established lung cancer screening programmes. For instance, the ELCAP and Framingham studies primarily focused on lung cancer detection, yet their incidental findings of mediastinal masses provide a useful benchmark for understanding the prevalence and significance of such lesions in asymptomatic individuals. Similarly, the Korean and Japanese studies, which utilized health check-up data, highlighted the role of routine CT imaging in identifying mediastinal abnormalities. By drawing parallels between these studies and the incidental findings during COVID-19 screenings, we can better contextualize the significance of mediastinal masses detected in the Chinese population during the pandemic.

This study aims to fill this gap by analysing the prevalence and characteristics of mediastinal masses incidentally detected during COVID-19-related chest CT screenings in China. By comparing these findings with those from other large-scale CT screening studies, we hope to contribute to a more comprehensive understanding of mediastinal masses across diverse populations and clinical contexts.

PATIENTS AND METHODS

Study population

We retrospectively collected data from outpatients and emergency patients who underwent chest CT scans between February 2020 and February 2021. The study population was restricted to patients who underwent chest CT scans for COVID-19 screening purposes. Patients with CT reports indicating mediastinal abnormalities were included, while those with the following conditions were excluded: mediastinal lymph nodes, thymic dysplasia, pneumomediastinum and other non-mass alterations (Fig. 1). This study was approved by the Ethics Committee of Xuanwu Hospital ([2022] 057). The requirement for informed consent was waived due to the retrospective nature of the study.

CT acquisition

All chest CT examinations were performed using a 256-slice multidetector row CT scanner (Somatom Definition Flash, Siemens Medical Systems, Forchheim, Germany) at 120 kVp with automatic tube current modulation, with a pitch of 1.2 and a collimation of 0.6 mm. Patients were scanned craniocaudally from the area extending from the level of the superior aperture of the thorax to the diaphragm in the supine position at full inspiration with a single breath hold (approximately 4–5 s). All CT images were reconstructed with 5 mm along with additional 1 mm slices in the axial plane of both the lung window [window level, −600 to −700 Hounsfield units (HU); window width, 1200–1500 HU] and mediastinal window (window level, 20–40 HU; window width, 400 HU).

Identifying masses in the mediastinum

Each radiology report and corresponding CT scan were independently reviewed by at least 2 experienced chest radiologists from our research team. Discrepancies in interpretation were resolved through discussion and consensus. The CT scan reports were screened according to the following keywords: [mediastin*] or [thym*]. A mediastinal mass was defined as any mass or lesion >5 mm in the long-axis diameter located in the mediastinum.

Evaluation of chest CT images

The masses were classified according to the three-compartment model developed by the International Thymic Malignancy Interest Group (ITMIG): anterior, middle or posterior mediastinum [7]. The location of the mass was also used to identify the organs that would likely be involved. The shape of the lesion was categorized into 4 groups: round, oval, triangular and irregular. The bidimensional long-axis and short-axis diameters (mm) were measured using electronic callipers and assessed mean CT attenuation by drawing a round region of interest along the border of the mediastinal lesion in the largest axial plane. For each mass, CT attenuations were obtained to determine whether it was cystic or fatty. Each thymic mass was then separately classified and described according to its shape and longest diameter. In addition, the longest diameter and CT attenuation of the thymic masses were statistically analysed.

Follow-up of mediastinal masses

If a resection was not performed, a re-evaluation with CT scan was suggested. Based on clinical experience indicating that the diameter of mediastinal masses remains relatively stable within a short-term period (1 year), follow-up chest CT scans in this study were performed 1 year after the baseline CT examination. The radiological features were then recorded and compared with the baseline features. We classify mediastinal masses with well-defined borders, homogeneous density, and lack of invasion into surrounding structures as benign masses. In contrast, malignant masses may exhibit irregular margins, heterogeneous enhancement, local invasion or associated lymphadenopathy. Incorporating these imaging characteristics will contribute to a more comprehensive understanding of the diagnostic process. The changes in the lesion size, measured on the long-axis diameter on follow-up CT scans were recorded. A size change of 2 mm or more in the long-axis diameter between baseline and follow-up CT scans was defined as an increase. Nodules increase or decrease of less than 2 mm was defined as no change, and nodules became smaller than before at 2 mm, defined as decrease. The pathology report was used to determine the pathologic diagnosis if the mass was resected.

Statistical analysis

The statistical analysis was performed using SPSS 26.0 software (SPSS Inc., Chicago, IL, USA). Categorical variables are reported as numbers (percentages). Continuous variables were assessed for normality using the Shapiro–Wilk test, with a significance threshold of P < 0.05 for rejecting the normality assumption. Normally distributed data are presented as mean (standard deviation), while non-normally distributed data are reported as median (interquartile range, IQR). All subgroup analyses involving age and sex comparisons were adjusted for multiple testing using the Bonferroni method.

RESULTS

Prevalence of mediastinal masses

Following a comprehensive review of all chest CT scans in accordance with the predefined inclusion and exclusion criteria, a total of 40 112 patients were enrolled in this study for data analysis [mean age 54.5 (17.2) years; male-to-female ratio 1.02:1]. Among the 40 112 participants, mediastinal masses were detected in 293 (0.73%), of which 0.56% (223 of 40112) were anterior mediastinal masses (Table 1). The prevalence tended to increase as the age of the participants increased (P < 0.001) (Fig. 2 and Table 2). Although men exhibited a marginally higher prevalence than women (0.75% vs 0.71%), no statistically significant difference was observed between genders (P = 0.635) (Table 2).

CT features of mediastinal masses

Of the 293 mediastinal masses, the most common finding was a thymic mass (152 cases; prevalence, 0.38%), which was mainly located in the anterior mediastinum (223 cases; prevalence, 0.56%, Table 1). A thyroid mass was detected in 6 patients (0.01%), neurogenic mass in 10 patients (0.02%), and other type of mass in 125 patients (0.31%; Table 1). In addition, the most common CT characteristic of mediastinal masses was an oval shape (n = 163; Supplementary Material, Table S1). The median values of the longest and perpendicular longest diameters were 19.0 mm (IQR 13–30 mm, range 5–116 mm) and 12.0 mm (IQR 8–17.5 mm, range 2–114 mm), respectively (Supplementary Material, Table S2). Regarding the long-axis diameter, 38.6% (113 of 293) of lesions, it was between 10 and 19 mm, followed by 11.9% (35 of 293) that was smaller than 10 mm, 23.2% (68 of 293) between 20 and 29 mm, and 26.3% (77 of 293) 30 mm or larger (Supplementary Material, Table S1 and Fig. 3A). The median CT attenuation of the lesions was 25 HU (IQR 11–41, range −56 to 121). Approximately half of the lesions (45.7%) had a CT attenuation of less than 20 HU, which corresponded to the water attenuation of the cysts (Supplementary Material, Table S1 and Fig. 3B).

CT features of thymic masses

Of the 293 mediastinal masses, 51.9% (152 of 293) masses were identified as originating from the thymus. All the thymic masses were located at the anterior mediastinum (Table 1). They were mostly ovoid in shape (n = 89; Supplementary Material, Table S3). The median values of the longest and perpendicular longest diameters and CT attenuation were 15.0 mm (IQR 11.0–22.5 mm, range 5–116 mm), 10.0 mm (IQR 7.5–15.0 mm, range 4–114 mm) and 38.5 HU (IQR 22.5–49.0 HU, range −50 to 121 HU), respectively (Supplementary Material, Table S4).

Cases with a confirmed diagnosis

The diagnosis of 33 lesions (11.3%, 33 of 293) was confirmed after surgical resection, of which 11.7% (26 of 223) were diagnosed by surgical resection of anterior mediastinal masses. A prevalence of thymic epithelial tumours (46.2%, 12/26) was reported. The characteristics and pathological features of the diagnosed cases are shown in Table 3 and Supplementary Material, Table S5. The longest diameter and CT attenuation of nodal lesions were 46.2 (27.2) mm and 35.2 (19.7) HU. Of these cases, 48.5% (16 of 33) were malignant (12 thymic epithelial tumours, 2 B-cell lymphomas, 1 germinoma, and 1 follicular dendritic cell sarcoma), and 51.5% (17 of 33) were benign (6 thymic cysts, 1 bronchogenic cyst, 2 oesophageal cysts, 3 mature teratomas, 2 schwannomas and 3 retrosternal thyroid). Among the patients who underwent resection, 39.4% (13 of 33) had at least 1 follow-up CT scan. In other words, surgical resection was performed in 4.4% of patients (13 of 293) who had at least 1 follow-up CT scan at one year or later.

Unconfirmed lesions with follow-up

All of the screened mediastinal masses received follow-up, and the results showed that 67.7% (176/260) of patients never underwent a repeat CT scan, except for 33 patients whose diagnosis was confirmed by surgical resection. Similarly, 66.5% (131/197) of patients with anterior mediastinal masses had not undergone a repeated CT scan after excluding the 26 patients who underwent surgical resection. However, 32.3% (84/260) had at least 1 follow-up CT scan at 1 year or later. The follow-up results showed that a total of 82.1% (69/84) of lesions were stable in size during follow-up, while 13.1% (11/84) showed an increase in size or elevated CT attenuation and 4.8% (4/84) exhibited a decrease in size (with one lesion disappearing completely) (Table 4).

In addition, 7.7% of patients (20/260) had died of diseases other than mediastinal masses by the time of follow-up (Supplementary Material, Table S6). The mean age of the deceased patients was 78.0 (10.9) years, and they all died of diseases common to elderly individuals: heart failure, acute cerebrovascular disease, respiratory failure and renal failure.

DISCUSSION

This large-scale retrospective cohort study analysed the prevalence of asymptomatic mediastinal masses in patients who underwent chest CT scans during the COVID-19 epidemic. We must emphasize, however, that emergency patients whose condition required a chest CT scan were excluded (e.g. pulmonary disease, chest trauma, malignant tumour follow-up). The participants who were ultimately enrolled received a chest CT scan that was performed passively due to COVID-19. This retrospective study was the first to analyse the prevalence of mediastinal/anterior mediastinal masses within the Chinese population. Previous studies were all performed on patients undergoing active chest CT scans, whereas the present study was the first to report both the prevalence and follow-up results of mediastinal masses from passively performed CT scans. Our study revealed that the prevalence of asymptomatic mediastinal masses was 0.73% (293 of 40 112) and 0.56% (223 of 40 112) for anterior mediastinal masses. An oval shape, anterior mediastinal location and thymus involvement were the most frequent imaging findings for mediastinal masses. Thirty-three patients were diagnosed by surgical resection, and approximately half of the lesions were malignant. Patients who did not undergo surgical resection received follow-up, and 82.1% of lesions (69 of 84) were stable in size, with a low percentage of changes occurring. In addition, 20 elderly patients with a severe basic disease had passed away at the time of follow-up. This highlights the importance of considering comorbidities and overall health status when managing mediastinal masses, particularly in elderly populations.

The prevalence of anterior mediastinal masses in our study was lower than that in the Japan health check-up study (1.49%, 578 of 38861), Framingham Heart Study (0.89%, 23 of 2571) and Korean health check-up data study (0.73%, 413 of 56 358) but slightly higher than that in the ELCAP study (0.45%, 41 of 9263). These variations in prevalence can be attributed to several factors, including differences in study populations, CT imaging protocols and lesion definitions.

The demographic and geographic characteristics of the study populations likely play a significant role in the observed differences. Our study population was drawn from northern China, whereas the Framingham Heart Study and ELCAP study primarily included Western populations, and the Japanese and Korean studies focused on East Asian populations. Genetic predispositions, environmental exposures, and lifestyle factors may contribute to the varying prevalence of mediastinal masses across these regions. For instance, the higher prevalence in the Japanese study (1.49%) may reflect unique population-specific risk factors or healthcare-seeking behaviours.

Variations in CT imaging protocols—especially slice thickness—also critically influence the observed prevalence rates. In our study, 1 mm axial slice CT images were used, which aligns with the Framingham Heart Study and the Korean health check-up study. Thin-slice CT (1 mm or thinner) is known to enhance the detection of small lesions, potentially explaining the higher prevalence in these studies compared to the ELCAP study, where 32% of participants had baseline CT images with slice thicknesses of 2.5 mm or 10 mm. The Japanese health check-up study, which reported the highest prevalence (1.49%), used a slice thickness of 2.5 mm, suggesting that even thicker slices may not significantly compromise detection rates in certain populations.

The criteria for defining and measuring lesions also varied among studies. In our study, a cut-off of 5 mm or larger was used to detect incidental nodular lesions, whereas the ELCAP study applied a stricter threshold of 7 mm or larger for the short-axis diameter. This difference in lesion definition likely contributed to the lower prevalence reported in the ELCAP study. The use of a 5 mm threshold in our study may have increased the detection of smaller lesions, thereby raising the observed prevalence. Thus, these factors were responsible for the difference in the prevalence of anterior mediastinal masses.

The most frequent imaging findings in our study included an oval shape, anterior mediastinal location, and thymus involvement. These characteristics are consistent with those reported in the Framingham Heart Study and the Korean health check-up study, which also emphasized the importance of thin-slice CT for accurate characterization of mediastinal masses. The Japanese health check-up study highlighted the role of routine CT imaging in identifying mediastinal abnormalities, even with thicker slice protocols, while the ELCAP study provided valuable insights into the incidental detection of mediastinal masses in the context of lung cancer screening. Among the masses originating from the thymus, 66.4% (101 of 152) were less than 2 cm in the long-axis diameter; 58.6% (89 of 152) were oval, and 56.6% (86 of 152) had a CT attenuation greater than 20 HU. Based on these CT features, a substantial proportion of lesions were indistinguishable from thymic epithelial tumours. CT attenuation is a reliable method for measuring a thymic mass [8, 9]. The Japanese health check study recommended a CT value of 29.4 HU to distinguish between thymic cysts and thymic epithelial tumours [6]. Similarly, Li et al. [10] used a CT value of 31.2 HU as the threshold for differentiating thymic cysts from microscopic thymic tumours. Most thymic masses in our study showed a soft tissue density (median 38.5 HU) and were suspected of being thymomas. Thymic cysts, however, may have a higher CT attenuation than serous fluid because of their protein-rich composition or haemorrhage [11–13]. Therefore, correctly differentiating thymic epithelial tumours from prevalent benign diseases such as thymic cysts [12, 14] or nodular thymic hyperplasia [15] is challenging, which can result in unnecessary thymectomies [14].

In our study, the proportion of benign tumours (51.5%, 17 of 33) was slightly higher than that of malignant tumours (48.5%, 16 of 33) in the 33 resected cases. Slightly more than 10% of the detected anterior mediastinal lesions (11.7%, 26/223) had a confirmatory diagnosis, and 12 of 15 malignancies were thymic epithelial tumours. Interestingly, if we assume that unconfirmed cases (88.3%, 197 of 223) had the same proportion of thymic epithelial tumours (46.2%, 12 of 26) as our confirmed cases, the overall prevalence (0.23%, 90 of 40,112) would be similar to that reported in a prior postmortem investigation (0.18%, 57 of 31 000 cases) [16]. Therefore, incidental thymic epithelial tumours are fairly prevalent. However, because not all anterior mediastinal nodular lesions are thymic epithelial tumours, incidental anterior mediastinal masses must be managed cautiously and followed regularly.

When we followed the unconfirmed cases with a CT review, the sizes of 82.1% of the masses were unchanged. This suggests that many incidental findings may be benign, but growing lesions, particularly those with irregular margins or heterogeneous enhancement, raise concerns for malignancy and warrant closer monitoring or intervention. These findings align with the Korean health check-up study, which also reported a low rate of progression among incidental mediastinal masses but emphasized the need for vigilance in cases of growing lesions. Consistent with previous studies [3–5], we also postulate that incidental mediastinal masses in asymptomatic people, including masses in the thymus gland, should be approached in a ‘conservative’ manner, particularly because the evidence presented here suggests very slow growth or stability in some people.

The management of incidental mediastinal masses requires a tailored approach based on lesion characteristics, patient demographics, and clinical context. For stable lesions with benign imaging features (e.g. well-defined borders, homogeneous density and lack of invasion), we recommend annual follow-up with thin-slice CT imaging to monitor for any changes in size or morphology. This approach is supported by the Framingham Heart Study and the Korean health check-up study, which also emphasized the importance of longitudinal monitoring for asymptomatic lesions. For lesions showing growth or suspicious features (e.g. irregular margins, heterogeneous enhancement or local invasion) should prompt earlier re-evaluation or referral to a multidisciplinary team for further assessment. Surgical intervention should be considered for lesions showing significant growth, malignant features, or clinical symptoms. Given the overlap between incidental mediastinal masses and lung cancer screening populations, integrating follow-up protocols with existing lung cancer screening programmes may improve efficiency and patient outcomes. This approach has been successfully implemented in the ELCAP study and other large-scale screening initiatives.

During the follow-up period, 20 patients died of exacerbations of their primary disease, and the mediastinal mass was not re-examined or confirmed. A mediastinal mass may be detected incidentally in clinical practice, just as it can be discovered during lung cancer screenings or health check-up programmes. According to both guidelines and a consensus [17–20], people at high risk of lung cancer should undergo an annual chest CT scan, and lung cancer screenings might also detect other diseases, such as mediastinal masses [21].

Our study has several limitations. Firstly, CT images of some participants were not reviewed, and thus some mediastinal masses might have been missed. Secondly, most participants did not have a repeat chest CT follow-up review, resulting in a low follow-up rate. Finally, the follow-up period was short and needs to be extended.

Among individuals undergoing chest CT scans during COVID-19 screening in China, the prevalence of all mediastinal masses and anterior mediastinal masses was 0.73% and 0.56%, respectively. Our findings advocate risk-stratified management: lesions with growth/suspicious features require systematic evaluation, while stable low-risk masses may undergo limited surveillance. Standardized multidisciplinary protocols are urgently needed for classification and follow-up criteria. These insights should not be construed as supporting population screening, but rather emphasize evidence-based decision-making. Generalizability requires validation through prospective multicentre studies across diverse populations.-

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

FUNDING

This work was supported by the Capital Health Research and Development of Special Fund (2022-2-2013).

Conflict of interest: The authors have no conflicts of interest to declare.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding author.

Author contributions

Gaojun Lu: Conceptualization; Data curation; Formal analysis; Methodology; Writing—original draft; Writing—review & editing. Peilong Zhang: Data curation; Methodology; Writing—review & editing. Sara Ricciardi: Conceptualization; Methodology; Writing—original draft; Writing—review & editing. Ruotian Wang: Data curation; Methodology; Writing—review & editing. Chen Wang: Data curation; Methodology; Writing—review & editing. Kun Qian: Data curation; Methodology; Writing—review & editing. Giuseppe Cardillo: Conceptualization; Supervision; Writing—review & editing. Yi Zhang: Conceptualization; Supervision; Visualization; Writing—review & editing

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Moustafa Farouk Aboollo, Paolo Scanagatta and the other, anonymous reviewer(s) for their contribution to the peer review process of this article.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• CT

  Computed tomography

 
• ELCAP

  Early Lung Cancer Action Project

 
• HU

  Hounsfield units

 
• IQR

  Interquartile range

Author notes