In this study, we integrated an observational study based on the MIMIC-IV database with an MR analysis based on GWAS summary data to comprehensively assess the relationship between serum albumin levels and PE risk. The results of our study revealed a negative linear correlation between serum albumin levels and the risk of PE. A significantly higher risk of PE was observed in individuals with lower albumin levels. Furthermore, the utilization of two-sample MR analysis provided additional support for the causality between serum albumin and the risk of PE, suggesting that low serum albumin could potentially serve as a predictive indicator for PE risk.
PE and DVT are collectively referred to as VTE, affecting approximately 10 million people worldwide [20]. Epidemiological evidence has suggested that low serum albumin levels are associated with the occurrence of many cardiovascular diseases, ranging from ischemic heart disease, heart failure, atrial fibrillation, stroke, and venous thromboembolism itself [21]. Multiple observational studies have revealed that serum albumin levels are correlated negatively with VTE [22–24]. For instance, one retrospective study demonstrated that even a modest decrease in serum albumin levels corresponded to an increased risk of VTE, with the risk proportionally escalating as albumin levels declined [25]. In that study, taking a serum albumin level of ≥ 4 g/dL as a control, patients with an albumin level of 3-3.99 g/dL experienced a 1. 5 times greater risk of VTE, those with a serum albumin of 2.5–2.99 g/dL received a 2 times greater risk of VTE, and those with a serum albumin of < 2.5 g/dL increased the risk of VTE by almost 3 times. In addition, low serum albumin levels may increase the risk of VTE, whether in the general population or in high-risk populations [26, 27].
The aforementioned studies provide compelling evidence for the correlation between low serum albumin levels and an increased risk of VTE. However, it is worth noting that PE distinguishes itself from VTE, and the current understanding of the association between serum albumin levels and VTE risk falls short of adequately explaining the impact of serum albumin on PE risk. Moreover, previous research has predominantly focused on the prognosis of patients with PE in relation to serum albumin levels, leaving a gap in our knowledge regarding the specific connection between serum albumin levels and PE risk. Consequently, uncertainty remains regarding the exact nature and magnitude of the association between serum albumin and the risk of PE.
In order to assess the association more accurately between serum albumin and PE risk, we extracted data from 13,326 individuals (749 PE patients, 12,577 non-PE controls) in the MIMIC-IV database for a relatively large observational study. Firstly, by comparing the clinical data of PE patients and non-PE controls, we discovered serum albumin levels were significantly lower in the case group than in the control group. Subsequently, the PSM strategy of 1:1 was performed to control the effect of potential confounding factors, and the results demonstrated that serum albumin levels in the case group were still significantly lower than those in the control group. Finally, the association between serum albumin and PE risk was measured by restricted cubic spline model, in which we observed that serum albumin levels were linearly and negatively correlated with PE risk, and the risk of PE gradually increased with decreasing serum albumin levels. These results suggested that the association between serum albumin and the risk of PE may share similarities with the association between serum albumin and VTE risk.
Given the presence of potential confounders and the issue of reverse causation, making causal inferences in observational studies can be challenging. MR provides a more dependable method for establishing causality, which can overcome certain limitations of observational studies [28]. Our MR analysis revealed a potential negative causality between serum albumin and PE risk. The primary IVW method, along with additional MR methods employed as supplementary measures, yielded consistent results. A range of sensitivity analyses further confirmed the robustness and reliability of the negative correlation. In conclusion, MR analysis provided evidence supporting the negatively correlated causality between serum albumin and PE risk at the genetic level, which aligns with the findings of our observational study.
While this study confirmed the causality between serum albumin and PE risk, the mechanisms underlying this association remain poorly understood. Inflammation may be a potential mechanism linking low serum albumin to increased PE risk. There has been increasing evidence that inflammation plays an essential role in the pathogenesis of venous thromboembolism [29]. Several previous studies have also confirmed that some inflammatory markers (e.g., C-reactive protein, interleukin, and tumor necrosis factor α) are associated with an increased risk of VTE [30–32]. Folsom et al. argued that low serum albumin levels reflect inflammation or hypercoagulability [26], which may account for the correlation between low serum albumin and the occurrence and progression of VTE. Serum albumin, a negative acute-phase protein, is primarily synthesized by the liver [33]. During episodes of inflammation, the liver redirects its synthesis towards other acute-phase proteins, resulting in a decrease in serum albumin levels [34]. This points to the fact that low serum albumin forms part of the inflammatory response, indicating low serum albumin may be reliant on inflammation to influence the occurrence and development of VTE. Interestingly, in a prospective study by Kunutsor et al. [35], serum albumin was still significantly associated with VTE risk despite adjusting for several inflammation markers, such as gamma-glutamine transpeptidase, fibrinogen, and high-sensitivity C-reactive protein. This correlation persisted even when individuals with elevated levels of high-sensitivity C-reactive protein, indicating an active inflammatory process, were excluded from the analysis. Consequently, they considered that the correlation between serum albumin and VTE risk may not be reliant on inflammation and unaffected by inflammation. Another perspective suggests that low serum albumin could merely serve as an indicator of potentially poor general health conditions, such as liver disease, nephrotic syndrome, and cancer, which could contribute to the risk of VTE [35]. In summary, the underlying mechanisms linking serum albumin and VTE risk are still incomplete and subject to controversy. Further studies are necessary to resolve these uncertainties in order to gain a deeper understanding of the mechanisms involved.
The major strength of this study lies in the integration of an observational study with a Mendelian randomization analysis, allowing for a comprehensive assessment of the relationship between serum albumin and the risk of PE. The observational study enables exploration of the association between serum albumin levels and PE risk at the epidemiologic level. MR analysis effectively excludes the interference of confounding, bias, and reverse causality in traditional epidemiologic investigations. Notably, both approaches yielded almost identical results, adding to the reliability and robustness of our findings.
Nevertheless, there are several limitations in our study. Firstly, the observational study excluded many individuals with missing critical data (e.g., albumin), potentially introducing selection bias. Secondly, although adjustments were made for certain relevant confounders, there may still exist residual or unmeasured confounders, such as some inflammatory markers (e.g., C-reactive protein). Lastly, our study primarily concentrated on individuals of European-American ancestry, which may limit the generalizability of the findings to other racial and ethnic groups. To further validate the results, future studies including more diverse races and ethnicities are warranted.