Risk factors of embolism for the cardiac myxoma patients: a systematic review and metanalysis

Background The risk factors contributing to embolism in cardiac myxoma (CM) are yet controversial. This systematic review and meta-analysis aimed to clarify the risk factors of embolism for the CM patients. Methods PubMed, Embase, Cochrane library, Web of Science, China National Knowledge Infrastructure, Wan Fang, and Wei Pu databases were searched from inception to June 2019. Statistical analysis was conducted using Stata version 14.0. The pooled odds ratio or mean difference with 95% confidence interval was estimated for each risk factor. Results Herein, 12 studies, encompassing 1814 patients, were included. The pooled results suggested that New York Heart Association (NYHA) class I/II (P < 0.01), hypertension (P = 0.03), irregular tumor surface (P < 0.01), tumor in atypical location (P = 0.01), narrow base of tumor (P < 0.01), and increased fibrinogen (FIB) (P < 0.01) are significant risk factors of embolism in CM patients. However, sex, age, body mass index, smoking, left ventricular ejection fraction, diabetes, hyperlipidemia, atrial fibrillation, valvular heart disease, coronary heart disease, tumor size, platelet count, white blood cells, and hemoglobin were not associated with embolism (all P > 0.05). Conclusions NYHA class (I/II), hypertension, irregular tumor surface, atypical tumor location, the narrow base of tumor, and increased FIB were significant risk factors of embolism in CM patients. For CM patients with these factors, early surgery might be beneficial to prevent embolism.


Background
Cardiac myxomas (CMs) are the most prevalent type of primary cardiac tumors in adults [1] that account for approximately half of all benign lesions. The manifestations of CM patients include obstruction, preoperative embolization, and constitutional symptoms. Embolization is a major and fatal complication that occurs in 20-45% of CM patients [2]; it includes cerebral embolism and peripheral embolism. Cerebral embolism accounts for about 50% of the embolic events, mainly acute stroke symptoms. Peripheral embolism may involve extremity, visceral, and coronary arteries [3]. The right-sided CMs are associated with pulmonary embolism. These embolic events are significant causes of mortality in CM patients [4]; however, the exact factors related to the occurrence of embolism are not yet clearly identified.
Although, several risk factors of embolism are recently reported [5][6][7][8][9][10][11][12][13][14][15][16], some are controversial. For instance, a previous study [17] showed that the small size of the tumor was an independent risk factor for embolism, whereas other studies neither found any association [14] nor presented a contrary conclusion [13]. Some studies reported that the male gender was associated with an increased risk of embolism [18], while others failed to find this association [6,7]. Therefore, we conducted this systematic review with meta-analysis to clarify the risk factors of embolism in CM patients.

Data sources and search strategy
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement was followed [19]. The PubMed, Embase, Cochrane library, Web of Science, China National Knowledge Infrastructure, Wan Fang, and Wei Pu databases were used for further studies. The search strings are reported in Additional file 1. Studies published up to June 30, 2019, were eligible. No language restriction was applied. Reference lists of selected studies were checked to ensure complete coverage.

Eligibility criteria
The studies that fulfilled the following criteria were eligible for inclusion in this meta-analysis: 1) All CM patients who underwent surgical resection and were confirmed by pathological examination; 2) Patients in each study were classified into two groups (embolic and non-embolic); 3) Embolic events were diagnosed by clinical symptoms or imaging examination. If the same population was presented in more than one publication, the study with the largest sample was included.
We excluded the studies that met at least one of the following criteria: 1) Sample size < 50; 2) Abstracts from conferences, letters to the editor, and reviews; 3) Incomplete data.

Data extraction and quality assessment
Two authors (L.P and J.W) reviewed all retrieved articles and extracted data independently. The titles and abstracts were first screened to identify potentially eligible articles, and then, full texts were read to confirm their eligibility for inclusion in this meta-analysis. The extracted data included the following information: first author, year of publication, country, study type, sample size, age, incidence of embolism, and risk factors that include patient characteristics (sex, age, New York Heart Association (NYHA) class and atrial fibrillation (AF)), tumor characteristics (tumor surface and tumor location), and hematological parameters (white blood cell (WBC) and platelet count (PLT)). The NYHA class was divided into two groups (I/II vs. III/IV). The tumor surface was classified as irregular and regular [7]. The tumor location was classified into "typical location" (tumor arise from the interatrial septum at the border of the fossa ovalis in the left atrium) and "atypical location" (tumor arise from other sites of the left atrium or in the other cardiac chambers) [20]. The extracted data were cross-checked, and any disagreements were resolved by discussion or consultation with the third author (L.G).
The quality of the included studies was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist [21] for Case Series. It contains ten items, encompassing clear criteria for inclusion, the information of participants, and the statistical methods used. Each item was determined by yes, no, unclear, and not applicable.

Statistical analysis
Stata 14.0 was used for the statistical analysis of all data. The odds ratio/mean difference (OR/MD) and 95% confidence interval (CI) were calculated using the fixedeffect model or the random effect model. The heterogeneity among studies was quantified using the Cochran's Q test and chi-square (I 2 ) test. I 2 = 25% was considered low, 50% was moderate, and 75% was high. When I 2 > 50% or P-value < 0.05 was identified for substantial heterogeneity, we used the random effect model; otherwise, a fixed-effect model was adopted. A sensitivity analysis was conducted when substantial heterogeneity was presented. Publication bias was assessed by the Egger's test. All tests were two-sided, and the results were considered statistically significant at P < 0.05.

Search results and characteristics of included studies
A total of 12 studies [5][6][7][8][9][10][11][12][13][14][15][16] encompassing 1814 patients who underwent surgery with CM were selected for the current analysis. The detailed study selection progress is   Fig. 2 Forest plot for NYHA class between embolism: 4 studies were included, I 2 = 0.0%, fixed-effect model was adopted; the result showed NYHA I/II is a risk factor shown in Fig. 1, and the main characteristics of included studies are summarized in Table 1.

Meta-analysis of risk factors
Based on the data available from the included studies, the risk factors were classified as patient characteristics, tumor characteristics, and hematological parameters in the current review. A total of 20 risk factors were individually analyzed using a fixed-effect or a random effect model to estimate the association with embolism in CM patients. The main characteristics of each risk factor are summarized in Table 2.

Sensitivity analysis
High heterogeneity across studies was detected regarding age (I 2 = 76.8%), BMI (I 2 = 83.7%), and coronary heart disease (I 2 = 84.2%). Sensitivity analysis showed a similar trend among various studies except for the study by Yin et al. [9]. After the exclusion of Yin's study, the amongstudy heterogeneity was not detected (all I 2 = 0.00%, all P > 0.05); also, the overall effect did not differ significantly (all P > 0.05).
In addition, tumor size had high heterogeneity (I 2 = 83.3%), and sensitivity analysis showed a similar trend among studies except for the study by Kalçık et al. [13]. After this study was excluded, the among-study heterogeneity was found to be moderate (I 2 = 58.6%, P = 0.03), while the overall effect did not differ significantly (P = 0.19).

Publication bias
General considerations suggest that use of the Egger's test with substantially fewer 10 studies would be unwise [22]; hence, the publication bias was assessed for sex and tumor surface. The Egger's test not detect any publication bias on sex (P = 0.096) and tumor surface (P = 0.051).

Discussion
In the current meta-analysis, a total of 20 potential factors were studied. The results suggested that the NYHA class, hypertension, tumor surface, tumor location, tumor base size, and FIB were independent risk factors associated with embolism.
Among patient characteristics, we found that 2/12 factors (NYHA class and hypertension) were significantly associated with embolism. The NYHA class is the most commonly used indicator of clinical response to cardiac function. The current study showed that better cardiac   function (class I/II) was at a significantly higher risk of embolism than the worse function (class III/IV). This finding might be attributed to the fact that most embolic group patients were diagnosed after an acute embolic event; also, the tumor blockage symptoms were relatively less, and the overall cardiac function was better than that of non-embolic group. Conversely, the non-embolic group patients have a prolonged course of the disease and are likely to show hemodynamic changes; this phenomenon was associated with obstructive heart failure. However, the LVEF was not associated with embolism. We think this difference may attribute to the fact that some patients exist heart failure with preserved ejection fraction, and they also could reach NYHA class III/IV [23]. Hence, additional quantitative research about cardiac function is imperative. Some studies reported that hypertension might be a major risk factor of deep venous thromboembolism [24,25]. Interestingly, the present study found that hypertension also was a risk factor of embolism in CM patients, which could be because high blood pressure increases the activity of PLT, making the blood hypercoagulable [26].
Regarding tumor characteristics, we found that three factors are related to embolism: tumor surface, location, and base. Macroscopically, the surface of the tumor is classified into two types [27,28] as follows: Type 1 is characterized by an irregular surface and soft consistency, while type 2 presents a regular surface and compact consistency. Consistent with the previous studies [29][30][31], the current study found that embolic events are often associated with type 1 myxoma. This correlation might occur because the type 1 myxoma is prone to be friable, which leads to the shedding of tumor fragments into the bloodstream [13]. In addition, we found that atypical location is a risk factor of embolism. This conclusion is consistent with the result of the study by Deng et al. [6]. Thus, we hypothesized that the atypical location plays a significant role in hemodynamics than typical location. Lastly, the narrow tumor base was also a risk factor in this study owing to its great mobility, which tends to generate fragments [11].
In terms of hematological parameters, the present study designated increased FIB as a risk factor of embolism. It is a substrate for thrombin and is directly involved in the clotting process. In addition, FIB promotes the aggregation of PLT, increases blood viscosity, and aggravates red blood cell adhesion, all of which promote thrombosis [32]. Thus, we inferred that increased FIB promotes thrombosis on the surface of CM, which contributes to embolism.
To the best of our knowledge, this is the first metaanalysis to investigate the risk factors of embolism in patients with CM, with some practical clinical implications. Nevertheless, this meta-analysis also has some limitations. First, since all enrolled studies were retrospective case series, residual confounders and unidentified factors Fig. 7 Forest plot for FIB between embolism: 2 studies were included, I 2 = 0.0%, fixed-effect model was adopted; the result showed increased FIB is a risk factor were inevitable in observational studies. Second, all the included studies originated from Asia, which might lead to selection bias. Third, the number of overlapped risk factors in each of the studies was small for a comprehensive statistical analysis. Finally, with respect to the tumor size, the data of our meta-analysis showed high heterogeneity. Although one major source of heterogeneity was detected by sensitivity analysis, other differences between the studies should be considered.

Conclusions
This systematic review and meta-analysis identified the following significant risk factors of embolism for CM patients: NYHA class I/II, hypertension, irregular tumor surface, atypical tumor location, the narrow base of tumor, and increased FIB. For CM patients with these factors, early surgery may be beneficial for preventing embolism.