In the current study, we demonstrated for the first time that the survival rate was markedly distinguished between patients with LI and NLI in DCM, with lower survival rate in the former, which was consistent across stratified studies of different age and gender. In the PSM cohort, the survival rate was still lower in patients with LI. LI emerged powerfully negative effect on survival, regardless of adjustment for potential confounders. After PSM, the same results still hold. Using PSM method, the influence of potential covariates was eliminated, the comparability between the two groups of patients was higher, and the reliability of the results was strengthened. The majority of patients included in previous studies had only slight ALF (liver function indicators < 2–3 times the upper limit of normal) [4, 6, 8, 9, 13], and the value of LI has not been thoroughly affirmed. In clinical setting, there is a significant difference between the treatment and management of patients with slight ALF and LI. Our research focusing on LI is beneficial to clinical work and makes up for the shortcomings of previous studies. This is the first literature on the prognostic potential of LI in DCM, and the results have positive implications for the stratified management of patients with DCM.
When AST, ALT and TB were included as a continuous variable in the final model, they did not show an independent predictive value for prognosis in patients with DCM. The reason for this may be that in this study, the changes of liver function indicators within the normal range or slight increase had no potential clinical implication. Only when LI occurred, the prognostic value of liver function indicators was highlighted. Although previous studies [4, 13] do not entirely support the ideas presented here, the omission of vital measures such as NT-proBNP [4], LAD and LVEDD [13] from the final models may have weakened the applicability of their conclusions to patients with DCM.
There are no epidemiological studies on AFL or LI in outpatients with cardiovascular disease. A reported ALF occurs in 1–4% of asymptomatic patients [15]. We speculated that LI in outpatients with DCM has extremely low incidence and is not easily perceived. Moreover, whether the LI in outpatients with DCM is related to drugs or infectious diseases requires a comprehensive and complex evaluation process. In conclusion, we considered that LI in inpatients compared with LI in outpatients is easier to be found and judged, and has higher application value.
With the early diagnosis and etiological intervention, the development of comprehensive drug therapy and the improvement of device therapy, the survival rate of DCM patients has been considerably improved [22]. During a median follow-up of 2.4 years, approximately 31.4% of patients with DCM had a primary endpoint event in our study, which is apparently extremely high. The high mortality rate can be explained by the following reasons. First, according to the inclusion criteria and Chinese guidelines, patients with NYHA I or LVEF ≥ 45% were not included, who have better cardiac function and a favorable prognosis. Second, the rate of device therapy was low (4.4%). Third, because of the retrospective design, there was an 8.4% loss to follow-up in our cohort.
The liver is supplied by both the portal vein (70%) and the hepatic artery (30%), and its blood flow accounts for about one-fifth to one-quarter of the cardiac output [23]. It is involved in the uptake and transformation of indirect bilirubin as well as the excretion of direct bilirubin. Under the condition of hypoxia, the ability of the liver to uptake, transformation and excretion of bilirubin is weakened, which will inevitably lead to the rise of TB [24, 25]. In addition, the increase in TB may be due to hemolysis caused by congestive liver [26]. Diminished hepatic blood flow can be caused by a compensatory increase in vascular resistance sparked by reduced cardiac output and mesenteric vasoconstriction by an activated renin–angiotensin system [27, 28]. When rats developed intestinal ischemia–reperfusion injury, portal venous blood flow to the liver decreased by 66% during ischemia and hepatic arterial blood flow decreased by 80% during reperfusion, resulting in transient acute liver dysfunction (decreased bile secretion, increased ALT, and reduced ATP in liver tissue) [29]. In a model of systemic hypoperfusion induced by cardiac tamponade in pigs, decreased aortic blood flow triggered depletion of the hepatic artery buffering response, leading to liver dysfunction without evidence of cellular damage [30]. At present, the possible reason for liver dysfunction covers hepatic venous congestion secondary to right ventricular dysfunction, ischemia attributed to reduced hepatic blood flow, and hypoxemia attributed to by hepatic artery hypoxia [31]. It is generally accepted that transaminase abnormalities were dominated by hepatocyte necrosis caused by ischemia and hypoxia, while cholestasis caused by hepatic congestion was accompanied by an increase in TB, but it is impractical to definitely distinguish between the two mechanisms, as they commonly coexist and mutually exacerbate the detrimental effects on the liver [31, 32]. ALF was not only observed in hypotensive patients, but also applied to the anoxic condition of the person with normal BP [33]. Furthermore, Rossello X et al. found that normal blood pressure did not equate to no hypoperfusion in acute HF, and patients with higher SBP were less likely to be accompanied by hypoperfusion [34]. Thus, it was reasonable for the patient with LI to have normal but lower SBP.
DCM is characterized by enlarged LVEDD and reduced LVEF, and its underlying molecular mechanism involves activation of the neurohumoral system, which forms the pathophysiological basis of LI occurrence [35, 36]. The effect of LI on prognosis of patients with DCM should be considered from AST, ALT and TB. Many studies have verified that AST, ALT and TB were associated with increased right atrial pressure, pulmonary wedge pressure, and central venous pressure, as well as decreased cardiac index [37,38,39,40]. This implied that the presence of LI may be a sign of tissue hypoperfusion or systemic congestion. Although proxies for congestion (i.e., NT-proBNP) and perfusion (i.e., SBP and LVEF) were incorporated into the final multivariate model, LI may represent hemodynamic changes that cannot be captured by traditional measures. Given that NT-proBNP is a serological marker for diagnosing HF and reflecting ventricular wall tension [41], patients with LI had higher NT-proBNP and may have more severe circulatory congestion. Therefore, it was not surprising that there was a higher proportion of NYHA IV in patients with LI. LI exhibited irreplaceability in terms of prognosis, which indicated that LI not only reflected the temporary alterations of hemodynamics of DCM, but also reflected the potential pathological state and disease development stage of DCM, affecting the long-term prognosis of DCM patients. Compared with invasive hemodynamic monitoring, liver function testing is convenient, inexpensive, and acceptable, which provides the possibility of repeated testing and clinical follow-up.
The protective effect of higher SBP was consistent with previous studies [42, 43], and the underlying mechanism may be that patients with higher SBP had better tissue perfusion and cardiac function. Meanwhile, NT-proBNP consistently presented prognostic value that cannot be ignored [44]. The negative effect of LVEDD in this study was not surprising, since it had been demonstrated that expansion of LVEDD increased the risk of HF and sudden cardiac death [45, 46]. Interestingly, LVEF was not included in the final model before PSM, which may be due to the interference of confounding factors. Nonetheless, LVEF remained an irreplaceable prognostic factor after PSM.
Vincenzo Nuzzi et al. found that permanent atrial fibrillation (AF) had a similar effect on the prognosis of DCM patients as LI in this paper, while paroxysmal or persistent AF presented neutral results [47]. However, the adverse effects of permanent AF were not evident in our study. The reasons may be as follows: on one hand, with the retrospective data, we cannot make a precise distinction between permanent, paroxysmal, and persistent AF. Consequently, studies mixing the above three may influenced the presentation of the results. On the other hand, whether patients underwent radiofrequency ablation after discharge and whether sinus rhythm was restored, which were not available and may also have interfered with the results. Kidney function included as a categorical variable (eGFR ≥ 60 mL/min/1.73 m2) in univariate COX analysis was not significantly associated with the prognosis of patients with DCM, which may be due to the exclusion of patients with severe chronic kidney disease. Furthermore, the use of RAS inhibitor may improve kidney function, thus affecting its prognostic effect on patients with DCM. More importantly, the vast majority of patients did not have chronic kidney disease and the decrease in eGFR may be a transient phenomenon.
As far as we know, the reverse action mechanism of cardiogenic LI on the heart may include the following three aspects. First, the development of LI may promote the release of inflammatory factors, activate cell signaling pathways, and cause myocardial dysfunction, similar to those found in cirrhotic cardiomyopathy [48]. Second, the possible changes in the metabolic function of hepatocytes alter the medication metabolism process and influence its efficacy [49]. Finally, the liver is a transit point for material metabolism, and LI may affect its energy supply to the heart and other vital organs [50].
We acknowledge that this report had some limitations. The general flaws of retrospective studies were also not immune to this study. First, in addition to liver function indicators, we had no additional liver morphological, pathological, and hemodynamic information to adequately assess the severity of LI and its impact on prognosis. Second, we did not analyze the duration and recovery of LI, and these data may have different effects on the results. Third, the absence of myocardial injury-specific markers prevented us from reliably ruling out the role of myocardial injury in the outcome. Fourth, the lack of data on the assessment of right heart function prevented us from considering all factors that might have influenced the results. Finally, variations in patient medication and device therapy were not recorded during follow-up. This may constitute a latent factor that interfered with the results, even if the baseline levels of drugs were balanced.