Introduction

Cardiac lymphoma is usually secondary to cardiac involvement by systemic lymphoma, and less than 25% of lymphomas have cardiac involvement.1,2 In contrast, primary cardiac lymphomas (PCLs) are extremely rare, accounting for only 2% of primary cardiac tumors.3,4,5,6 As of July 2018, only 232 cases of PCLs have been reported in the literature.7 PCLs are highly invasive lymphomas, and the predominant histopathological type is diffuse large B-cell lymphoma, which accounts for 80% of published cases.8,9 The average age of onset is 63 years, and the male-to-female ratio is 3:1.4 The prognosis of this disease is poor, and the median survival time after diagnosis is 7 months.8,10 Early diagnosis in conjunction with effective chemotherapy can relieve the pathogenetic condition, control the progress of the disease effectively, and result in good outcomes with long-term survival.3,11,12 PCLs most frequently arise on the right side of the heart, particularly in the right atrium. This requires differentiation from cardiac angiosarcoma, which is the most common primary cardiac malignancy, often occurring on the right side of the heart.13 There is a lack of comparative studies on different primary cardiac malignant tumors, especially PCLs and primary cardiac angiosarcomas (PCAs). Previous literature reports a considerable overlap of imaging characteristics between cardiac lymphomas and angiosarcomas, which brings great challenges to the diagnosis of PCLs based on morphologic features.14 PET/CT or PET/MRI with contrast-enhanced imaging provides a combination of morphological tumor characterization and visualization of tumor metabolism. It has proven to be an indispensable problem-solving tool and is preferred for evaluation in lymphomas. Therefore, we retrospectively compared 18F-FDG PET/CT with CECT images of patients with PCLs and PCAs to provide diagnostic and differential diagnoses.

Methods

Study Design and Population

The local Institutional Review Board approved this retrospective study and waived the requirement for informed consent. We retrospectively collected clinical and imaging data of 139 consecutive patients with cardiac masses suspected by transthoracic echocardiography (TTE) and/or thoracic CT. All patients preoperatively underwent 18F-FDG PET/CT, followed by thoracic CECT on the same day. All patients only received symptomatic treatment before the above examination. The final diagnosis was confirmed by histopathology.

Inclusion Criteria

Patients were included if they met all the following inclusion criteria: (i) the patients had no previous history of lymphoma; (ii) the patients only received symptomatic treatment and had no history of chemotherapy, radiotherapy, or surgical resection before the PET/CT scan; (iii) cardiac masses were confined to the heart or pericardium without extra-cardiac involvement on 18F-FDG PET/CT imaging; and (iv) the diagnosis was confirmed by histopathology.

Exclusion Criteria

Patients were excluded for any of the following: (i) incomplete image data sets, (ii) lack of the final histological diagnosis, (iii) if extra-cardiac lesions were found on 18F-FDG PET/CT imaging or 18F-FDG uptake of bone marrow or subcutaneous fascia was increased diffusely, and (iv) bone marrow infiltration was confirmed by histopathology.

Patient Preparation and PET/CT Protocol

All patients with suspected cardiac tumors received special diet preparation (high-fat, low-carbohydrate, and protein-permitted diet) and fasting for more than 12 hours before 18F-FDG PET/CT examination. A detailed description of the preparation and the PET/CT protocol was provided in our previous study. 15

Imaging Analysis

All images were transferred to the workstation (Syngo MI Workplace, version VA30A, Siemens Healthcare) and reviewed in standard planes. CECT images were reviewed by a radiologist (Dr. JH) who received standard training in cardiac imaging. PET/CT images were evaluated by an experienced nuclear physician (Dr. WX). An observer (Dr. SW) with more than ten years of experience in radiology and PET/CT diagnosis combined the above two imaging modalities to make the final assessment.

FDG PET/CT

For the quantitative evaluation of the suppression of myocardial 18F-FDG uptake by the special diet preparation, we used the left ventricular ratio (LVR) of left ventricular (LV) myocardial uptake to the LV blood pool uptake. To reduce observer dependence and improve reproducibility, we used the maximum standardized uptake value (SUVmax). We chose to measure the ventricular septum as myocardial intake to avoid interference caused by lymphoma involving the LV free wall. The SUVmax of the blood pool was measured in the LV area between the right inferior pulmonary vein and the coronary sinus. Considering that the interventricular septum thickness was not greater than 1.3 cm, we chose a region of interest (ROI) with a diameter of 1.5 ± 0.1 cm to measure the SUVmax of the ventricular septum and the left ventricular blood pool. According to the CECT image, the non-contrast CT image and PET image were matched to the same position to display the interventricular septum and ventricular cavity. For PET images, the ROI obtained in the CT image of each patient was taken as a template, and it was extrapolated to the PET image and subsequently quantified (Supplementary Material 1 for measurement methods). LVR was calculated by dividing the LV wall’s SUVmax by the LV blood pool SUVmax. The calculation formula used is as follows:

$$ {\text{LVR}} = \frac{{{\text{SUVmax}}\,{\text{of}}\,{\text{LV}}\,{\text{wall}}}}{{{\text{SUVmax}}\,{\text{of}}\,{\text{LV}}\,{\text{blood}}\,{\text{pool}}}}. $$

For quantitative evaluations, the intensity- and volume-based PET parameters of cardiac lesions were analyzed. Intensity-based parameters included SUVmax, SUVmean, and tumor-to-liver ratio (TLR). TLR was calculated by dividing the cardiac lesion SUVmax by the liver SUVmean. Volume-based parameters included metabolic tumor volume (MTV) and total lesion glycolysis (TLG). A threshold of 41% of the SUVmax was used to delineate the MTV and TLG.16,17 For qualitative evaluations, the diagnostic criteria for PET/CT are that a tumor with diffuse metabolism is diagnosed as lymphoma, while a tumor with focal metabolism is diagnosed as angiosarcoma.

CECT

The enhancement pattern and tumor morphology were analyzed using CECT images. The pattern was categorized as heterogeneous or homogeneous enhancement, represented by 0 and 1, respectively (0 = heterogeneous and 1 = homogeneous enhancement). The tumor morphology was categorized as focal and diffuse lesions, represented by 0 and 1, respectively (0 = focal and 1 = diffuse lesions).18 The diagnostic criteria for CECT are as follows: if the tumor is a homogeneous enhancement, the diagnosis is lymphoma; if the tumor is a heterogeneous enhancement, the diagnosis is angiosarcoma; if the tumor is a diffuse lesion, the diagnosis is lymphoma; if the tumor is a focal lesion, the diagnosis is angiosarcoma. When there is a conflict between the diagnostic criteria of enhancement pattern and tumor morphology and considering the possibility of intrapericardial metastasis in angiosarcoma, we specify the enhancement pattern as the primary principle. We also evaluated the coronary vessel floating sign in the arterial phase, represented by 0 and 1, respectively, to indicate whether there was such a sign (0 = no and 1 = vessel floating sign). Pleural effusion and pericardial effusion were analyzed on CECT, and 0, 1, 2, and 3 were used to represent none, mild, moderate, and severe, respectively.

Statistical Analysis

Continuous variables were presented as mean ± standard deviation, and categorical variables were presented as frequencies and percentages. The One-Sample t test was used to compare the mean of LVR to one (test value). The Mann-Whitney U test or Student’s t test was used to compare two independent groups. Receiver operating characteristic (ROC) analysis was performed to determine the area under the ROC curve (AUC) for the intensity- and volume-based parameters. Statistical analyses were performed using SPSS software (v.26.0) and MedCalc software (v.19.0.7). A two-tailed probability value of < .05 was considered statistically significant.

Results

Population

Nine and eight patients with PCL and PCA were enrolled. All the patients had histopathological and immunohistochemical confirmation. The characteristics of the patients are shown in Table 1. A statistical difference in age between PCLs and PCAs (P = .025) was found. However, no statistical differences were found in gender (P = 1.000), blood glucose level (P = .099), and 18F-FDG-injected dose (P = .286).

Table 1 Characteristics of patients (N = 17)
Full size table

Of the nine cases with PCLs, five cases were confirmed by fine needle biopsies, two cases by cytological examinations of pleural and pericardial effusions, and the other two cases by partial resections. Of the eight cases with PCAs, two cases were confirmed by fine needle biopsies, one case by partial resection, and the other five cases by complete resections. Histologically, nine PCLs were classified as diffuse large B-cell lymphomas, the most common type of non-Hodgkin’s lymphoma. None of the patients with PCLs had AIDS, as proven by a negative anti-HIV test.

18F-FDG PET/CT and CECT Images

From 18F-FDG PET/CT images

A quantitative evaluation of 18F-FDG uptake agreement between the myocardium and blood pool was performed. In one case of PCLs (Case 5), the tumor filled the right ventricular cavity. It was poorly demarcated from the septum, so the SUVmax of the interventricular septum could not be measured. Therefore, in this case, the SUVmax of the LV free wall was measured. The interventricular septum was measured in the remaining cases (Supplementary Material 1). There was no significant difference in SUVmax between myocardial and blood pool uptake (Figure 1, Table 2). The mean of LVR was 0.97 (95% CI 0.87 to 1.07). There was no significant difference in the mean of LVR and the test value (Supplementary Material 2).

Figure 1
figure 1

Mean comparison of 18F-FDG uptake in the myocardium and blood pool. There were no differences in myocardial and blood pool uptake in the overall (A) or primary cardiac angiosarcoma group (B) or primary cardiac lymphoma group (C). MTV metabolic tumor volume; TLR tumor-to-liver ratio, TLG total lesion glycolysis

Full size image
Table 2 Quantitative evaluation of 18F-FDG uptake agreement between the myocardium and cardiac blood pool
Full size table

All the differences were statistically significant in SUVmax (t = 3.790, P = .002), SUVmean (t = 4.273, P = .001), MTV (U = 13.00, P = .027), TLR (U = 10.00, P = .011), and TLG (U = 4.0, P = .001) between PCLs and PCAs. The mean and median of the above subjects between PCLs and PCAs are detailed in Table 3 (Figure 2).

Table 3 18F-FDG PET/CT and CECT imaging features of primary cardiac angiosarcomas and primary cardiac lymphomas
Full size table
Figure 2
figure 2

Comparison of 18F-FDG uptake of tumors. There were significant differences in the mean of SUVmax (A) and SUVmean (A) and the median of TLR (A), MTV (B), and TLG (C) between primary cardiac angiosarcomas and primary cardiac lymphomas.

Full size image

ROC analysis revealed that the highest AUC for SUVmean was 0.958 (P < .001, 95% CI 0.738-1.000), and the optimal cutoff value of SUVmean was 5.17, which could generate 100.0% sensitivity and 88% specificity (Table 4 and Supplementary Material 3).

Table 4 ROC analysis for discriminating between primary cardiac angiosarcomas and primary cardiac lymphomas
Full size table

From CECT Images

There were significant differences in the enhancement pattern of tumors (P = .002) and the tumor morphology (P = .015). In nine cases of PCLs, seven cases showed homogeneous enhancement and the other two cases showed heterogeneous enhancement. All angiosarcomas showed heterogeneous enhancement. In nine cases of PCLs, eight cases showed diffuse lesions and the remaining one case showed focal lesions. In eight cases of angiosarcomas, six cases showed focal lesions and the remaining two cases showed diffuse lesions. The vessel floating sign was found in eight cases of PCLs (Figure 3). No floating sign was found in the remaining case because the lesion was located in the cardiac cavity. All lesions with angiosarcomas were located in the right atrium, and no sign of the vessel floating sign was found. A statistical difference in the tumor location between PCLs and PCAs (P = .002) was found. No statistical differences were found in pleural effusion (P = .527) or pericardial effusion (P = .349).

Figure 3
figure 3

Right coronary vessel floating signs. A, B, C, and D represent four different patients with vessel floating signs. A and B show the right coronary artery (arrows) in two patients, respectively. Multiplanar reconstruction (1) and maximum intensity projection (2) show that the right coronary artery is encapsulated by the tumor without arterial invasion or compression (coronary vessel floating sign). The transverse (C1) and multiplanar reconstruction (C2) images showed that the tumor encased the right coronary artery (arrows) with no narrowing of the arterial lumen. Multiplanar reconstruction (D1) and curved multiplanar reformatted image (D2) showed that the tumor invaded the right coronary artery (arrows), and the wall of the latter was surrounded without stenosis.

Full size image

Diagnostic Performance of Two Modalities

PET/CT alone showed 89% sensitivity, 75% specificity, and 82% diagnostic accuracy. Thoracic CECT alone showed 78% sensitivity, 100% specificity, and 88% diagnostic accuracy. The combination of the two modalities showed 89% sensitivity, 88% specificity, and 88% diagnostic accuracy. Combined PET/CT and cutoff values (SUVmean > 5.17) showed 100% sensitivity, 88% specificity, and 94% diagnostic accuracy. And combining both modalities and cutoff values (SUVmean > 5.17) showed 100% sensitivity, 100% specificity, and 100% diagnostic accuracy (Table 5 and Supplementary Material 4). There were no significant differences between PET/CT alone or thoracic CECT alone and combined both modalities, with or without cutoff (P > .05).

Table 5 Diagnostic performance of various parameters
Full size table

Outcome and Prognosis

Of the nine PCL cases, three patients achieved complete remission after chemotherapy with PET/CT (Figure 4 and 5 and Supplementary Material 5-6), two patients died, two patients showed partial remission after chemotherapy with CT, and the other two patients showed partial remission after chemotherapy with trans-thoracic echocardiography. Of the eight PCA cases (Figure 6, Supplementary Material 7), one patient underwent cardiac transplantation, two patients died, two patients developed distant metastases, and the other three patients were lost to follow-up.

Figure 4
figure 4

Primary cardiac lymphomas. MIP PET image before chemotherapy (A), 18F-FDG PET (B), CT (C), fusion images (D), and MIP PET image after chemotherapy (E). Transverse (1), coronal (2), and sagittal images (3). A MIP image showing foci of uptake in the heart region before chemotherapy. B, C, and D showing a huge irregular mass with increased intense 18F-FDG uptake (SUVmax: 18.3) in the atrioventricular groove (arrows). E MIP image showing an absence of clearly pathologic 18F-FDG uptake in the heart region after chemotherapy, mild uptake in the bilateral lung region, and pulmonary inflammation confirmed by CT.

Full size image
Figure 5
figure 5

Primary cardiac lymphomas. MIP PET image before chemotherapy (A), 18F-FDG PET (B), CT (C), fusion images (D), and MIP PET image after chemotherapy (E). Transverse (1), coronal (2), and sagittal images (3). A MIP image showing a huge lobulated mass with intense 18F-FDG uptake in the heart region before chemotherapy and foci of 18F-FDG uptake in the left upper lung region, which was confirmed to be a pacemaker. B, C, and D showing a huge irregular mass with increased intense 18F-FDG uptake (SUVmax: 18.2) in the whole heart region. (E) MIP image showing the mediastinal and bilateral hilar lymph nodes with mild 18F-FDG uptake and the bilateral masseter muscle with asymmetrical 18F-FDG uptake after chemotherapy.

Full size image
Figure 6
figure 6

Primary cardiac angiosarcomas in the right atrium. MIP PET image (A), 18F-FDG PET (B), CT (C), and fusion images (D). Transverse (1), coronal (2), and sagittal images (3). A MIP image showing multiple foci of uptake in the heart and mediastinal region. B, C, and D showing an irregular mass with increased uneven 18F-FDG uptake (SUVmax: 14.8) in the right atrium.

Full size image

Discussion

This study is the first to describe the 18F-FDG PET/CT and CECT imaging features of PCLs. This study provides the first comparative characterization of metabolic features between PCLs and PCAs, both of which arise in the right side of the heart. Combining of tumor enhancement pattern and morphology, as well as intensity- and volume-based PET parameters, helps distinguish PCLs from primary cardiac angiosarcomas and improves the accuracy of the diagnosis of PCLs (up to 100% in combination with cutoff values). Although the diagnostic accuracy tended to be higher in combined PET/CT and CECT than in a single modality, the differences were not statistically significant. The reason for this may be related to the small number of cases enrolled.

The current study found that the intensity-based PET parameters (SUVmax, SUVmean, and TLR) were significantly higher in PCLs than in PCAs. The mean value of SUVmax for PCLs was 17.7 (vs 8.34, P < .05), which was similar to the mean value of SUVmax (16.7) in the two cases with PCLs previously reported in our study. 15 It is also close to the mean value of SUVmax of two secondary cardiac lymphomas (15.2) reported by Rahbar et al. 19, six cases of secondary cardiac lymphoma (17.3) reported by Qin et al. 20 and seven cases of secondary cardiac lymphoma (19.4) reported by Meng et al.. 21 The mean value of SUVmax was significantly lower than that reported by Kikuchi et al. 22 in five cases of cardiac dominant lymphoma (25.9).

The current study found that the volume-based PET parameters (MTV and TLG) were significantly higher in PCLs than in PCAs (P < .05), implying that PCLs lesions are more extensively involved and diffuse. In contrast, PCAs lesions are more focal, consistent with the morphological characteristics of the tumor on CECT images.

The enhancement pattern of tumors was significantly different (P = .02) between PCLs and PCAs. In nine cases of PCLs, seven cases showed homogeneous enhancement and the other two cases showed heterogeneous enhancement. All angiosarcomas showed heterogeneous enhancement. It has been reported that the contrast enhancement of cardiac lymphoma may be homogeneous or heterogeneous with either modality. 23

The tumor morphology was significantly different (P = .015) between PCLs and PCAs. Eight of the nine PCLs showed diffuse lesions, and the remaining one showed focal lesions. Six of eight angiosarcomas showed focal lesions and the remaining two showed diffuse lesions. Out of the nine PCLs, the tumors diffusely involved the heart wall and completely encased the coronary artery in eight cases, also called “vessel floating sign.” No floating sign was found in the remaining case because the lesion was in the cardiac cavity. In contrast, all lesions with angiosarcomas were in the right atrium, and no sign of the vessel floating sign was found. Since the PCL lesion mostly originates in the right atrium or atrioventricular sulcus, the lesion mostly encircles the right coronary artery, forming a right coronary artery floating sign, but there is no stenosis, which has been confirmed by coronary angiography.24 The coronary artery floating sign is a specific characteristic of cardiac lymphoma.14,22,25 Interestingly, one case of PCL had no vessel floating sign; the lesions were all located in the right atrial cavity, showing focal nodular protrusion into the cardiac cavity, without obvious infiltration, which was easily confused with angiosarcoma. Therefore, the diagnosis was very challenging.

It is worth noting that after dietary preparation, there was no significant increase in the myocardial glucose metabolism in all cases, which is helpful for the description of tumor lesions. However, for some patients who cannot follow a diet, such as those who are fasting, the American Society of Nuclear Cardiology recommends low doses of IV heparin for suppression of myocyte glucose uptake.26

Unlike secondary cardiac lymphoma with superficial lymph node involvement, PCLs are difficult to diagnose by biopsy; therefore, it is important to use other non-invasive imaging methods. 18F-FDG-PET/CT has proven to be an irreplaceable problem-solving tool and is preferred for routine staging and follow-up.

Limitations of Our Study

The limitations of this study are the small number of cases and the inherent defects of its retrospective nature. Second, there might be a selection bias. This limitation arises from our strict adherence to the narrow definition of PCLs.4 According to this definition, cases were enrolled only if the lesion located in the heart was not accompanied by other organ or lymph node involvement, which potentially excluded cases with concurrent cardiac and other organs or lymph node involvement.

Conclusion

The average age of patients with PCL was significantly higher than that of patients with PCA. The intensity- and volume-based PET parameters of PCL were significantly higher than those of PCA. The enhancement pattern and tumor morphology were also different. The coronary artery floating sign is a specific characteristic of cardiac lymphoma. According to these characteristics, the two most common types of primary cardiac malignant tumors could be differentiated.

New Knowledge Gained

Primary cardiac lymphomas (PCLs) are very rare. The most common location of the disease is similar to that of primary cardiac angiosarcomas (PCAs). Combined with the enhancement characteristics of the lesions and the intensity- and volume-based parameters of 18F-FDG uptake, the two types of primary cardiac malignancies can be differentiated.