Our institutional review board approved the study. Written informed consent was obtained from patients who underwent enhanced CT examinations.

To explore the development of RE, the records of patients with pulmonary or mediastinal malignancies and CVO from October 2014 to December 2021 were reviewed. The detailed inclusion criteria were as follows: patients with pulmonary or mediastinal malignancies with CVO, patients who had chest CT scans with unenhanced, arterial, and venous phase, and patients who had RE on enhanced CT. There were 3133 oncology patients with CVO during this period. Those with other collateral veins but not ICVP (n = 991) and those without collateral veins (n = 1721) were excluded. Among the patients with ICVP filling with contrast (n = 421), 342 patients who did not have RE were excluded. Finally, 79 patients who demonstrated RE (RE group: 57.5 ± 12.3 years, 70 men) on chest CTs were included (Figure S1).

Moreover, to analyze the differing CT features between RE and sclerotic metastasis, 53 patients with malignancies with CVO and sclerotic metastasis (SM) (SM group: 54.8 ± 11.1 years, 35 men) in the same period were selected randomly for comparison.

Clinical data (e.g., sex, age, primary tumor types) were obtained from medical records.

Chest CT was performed using a 64-detector spiral CT system (Discovery CT750 HD, GE System) (RE group, n = 27; SM group, n = 17), a 96-detector spiral CT system (SOMATOM Force CT, Siemens Healthcare) (RE group, n = 29; SM group, n = 16), or an 80-detector spiral CT system (UIH uCT 780, United Imaging Intelligence, Shanghai, China) (RE group, n = 23; SM group, n = 20). The acquisition parameters were as follows: 120 kVp, 150–300 mA of automatic adjustment, slice thickness of 5 mm, and pitch of 0.984:1.

For the baseline staging or efficacy evaluation of the malignant tumors, unenhanced and contrast-enhanced CT scans (30–35 s, arterial phase; 55–60 s, venous phase) were routinely obtained for all patients in our center. Enhanced images were obtained after a bolus intravenous injection of 1.5 ml/kg of nonionic contrast agent (Omnipaque 300, GE Healthcare, or Ultravist 370, Bayer Healthcare) through the antecubital vein at a rate of 3 ml/s, which was reduced to 2.5 ml/s or 2.0 ml/s for patients who had received previous radiotherapy or chemotherapy. The axial and multiplanar reformation images were routinely reconstructed at a slice thickness of 2 mm and an interval of 1 mm. The CT images were reconstructed using Standard (soft) kernel and Bone kernel for the GE system, Br44d kernel and Qr50d kernel for the Siemens system, and B_Soft kernel and B_VSHARP kernel for the United Imaging Intelligence system.

The contrast material was injected via the left (RE group, n = 37; SM group, n = 23) or right (RE group, n = 42; SM group, n = 30) antecubital veins.

All enrolled patients had CVO. Considering that the ICVP drains into the superior vena cava via the azygos venous system as the main pathway [10,11,12,13], the extent of the CVO was classified as follows (Figure S2): type 1, patients with obstruction of the superior vena cava above the azygos arch level, with or without obstruction of the brachiocephalic vein (unilateral or bilateral); and type 2, patients with superior vena cava obstruction straddling the azygos arch, with or without azygos arch obstruction, with or without brachiocephalic vein obstruction (unilateral or bilateral).

The communication between the ICVP and other collateral veins (the paravertebral collateral veins, or the anterior or lateral thoracic venous plexus) were recorded.

According to the diagnosis of intravertebral enhancement in previous studies [1,2,3,4], we diagnosed RE based on the following findings: no abnormality in unenhanced scans, obvious enhancement in the arterial phase, changes in density and morphology in the venous phase, and no abnormality in other imaging examinations (emission CT [ECT], PET-CT, or MRI) during the same period.

Sclerotic metastases typically present as radiodense bone lesions that are round/nodular with relatively well-defined margins on CT in patients with malignant tumors [4, 14, 15]. To compare the CT features of RE to sclerotic metastasis, osteolytic bone metastases with soft tissue masses were excluded. Benign dense bone lesions (bone island, etc.) were also excluded. All cases were diagnosed according to typical findings of CT, ECT, PET-CT or MRI during the same period and follow-up imaging, in the absence of a pathological diagnosis.

Two experienced authors (one with more than 20 years of experience with subspecialty training in musculoskeletal radiology and one with 8 years of experience) reviewed the CT images and decisions were reached by consensus. The images were reviewed with the mediastinal soft tissue window (window width, 250–350 HU; window level, 30–50 HU), and bone window (window width, 950–1500 HU; window level, 250–400 HU).

The following features were reviewed: (1) the enhancement patterns of the ribs (nodular/patchy, linear, and mixed) in arterial phase; the changes of the morphology and boundary of RE and sclerotic metastasis in venous phase; (2) for each patient, the number and level of involving ribs, left/right/bilateral ribs, anterior/posterior ribs, the shape (swelling, normal) and the cortex (interrupted, normal) of the involving ribs; (3) for patients with multiple involving ribs, unilateral/bilateral, adjacent/discrete.

In the RE group, the CT values of the patchy and mixed RE regions in the unenhanced, arterial phase and venous phase scans; and the surrounding normal bone in unenhanced scan were measured. The largest RE region was measured if there were several RE regions. Linear REs were not measured. The measurement of the CT value is as follows (Figure S3): in the arterial phase, the largest slice of patchy RE is selected, the conformal region of interest is drawn along the edge, and each RE region is measured three times. The average of the three measurements is the CT value. The measurement in unenhanced scans and the venous phase will be consistent with the measurement in the arterial phase in terms of the position and range of the region of interest. The CT value of the surrounding normal ribs is measured within 1 cm around the RE region on the unenhanced image, avoiding the outline of cortical bone. Likewise, all CT values of the sclerotic metastases in the SM group were measured (Figure S3).

Further review was performed to determine whether intravertebral enhancement was present, RE was misdiagnosed on the original imaging report or whether the RE or sclerotic metastasis persisted on follow-up imaging.

The age of patients was compared between the RE and SM groups via the unpaired t-test, and categorical variables were analyzed using Pearson's chi-squared test. The one-way ANOVA test was applied to compare the CT values in three phases in both the RE and SM groups. Tukey's multiple comparison test was employed to compare the differences in the CT values in the RE group. Using the unpaired t-test, the following differences in CT values were compared between and within RE and SM groups (Table S1): between the RE region/ sclerotic metastasis and the surrounding normal rib, the surrounding normal rib, the RE region/ sclerotic metastasis in unenhanced CT, and during the arterial and venous phases.

SPSS version 22 (SPSS Inc.) and GraphPad Prism 9.0 (GraphPad Software Inc.) were applied to analyze the data and create graphs. Statistical significance was set at p < 0.05.