Acute kidney injury in hospitalized patients with nonmalignant pleural effusions: a retrospective cohort study

Pleural effusion (PE) is a common clinical manifestation in hospitalized patients and is associated with significant morbidity and mortality. It affects more than 3,000 people per million population each year [1]. Nonmalignant pleural effusions (NMPE) are usually associated with organ dysfunction such as pneumonia and heart failure [2].

Acute kidney injury (AKI) is a common syndrome of complex etiology and pathophysiological mechanisms leading to abrupt loss of renal function. Hospital acquired AKI (HA-AKI) is reported to happen in 9.1% patients that not only associated with short-term adverse outcomes, but also has a long-term negative impact on survival [3‐6]. Even mild AKI is considered to be an independent predictor of various adverse outcomes [7, 8].

Large effusions can lead to a number of pathophysiological changes such as hypoxia and hemodynamic compromise putting patients at risk of AKI [9, 10]. What's more, Siniorakis et al. hypothesized a pleurorenal interaction, suggesting that neurohormonal derangements occur in PE patients, leading to renal hemodynamic alterations and AKI [11]. However, the incidence of AKI in hospitalized patients with NMPE is not well understood and there are few data on the relationship between effusion volume and AKI. Therefore, we conducted a large cohort of consecutive patients based on the electronic medical record (EMR) to determine the incidence of AKI in NMPE and its association with the size of effusion in hospitalized NMPE patients.

A total of 86,645 inpatients were screened between January 1, 2018, and December 31, 2021. Among them, 63,395 (73.2%) underwent chest radiological tests (CT/X-ray), and 6,956 of them had pleural effusions. After excluding 2019 cases with a discharge diagnosis of malignant diseases, there were 4937 cases, accounting for 7.8% of those admitted to internal medicine. After excluding 422 cases with sCr tests fewer than 2 during hospitalization, 935 cases diagnosed end-stage renal disease, regular dialysis, or kidney transplantation, 372 cases having CA-AKI and 161 cases having HA-AKI before chest image exams, 3047 cases were included in the final analysis (Fig. 1). Among the 3047 patients, 2019 (66.3%) cases of effusions were diagnosed with CT scan, 924 (30.3%) were diagnosed by orthostatic plain radiograph, and 104 (3.4%) were diagnosed by supine radiograph.

Our results showed that of the 3047 patients, 360 (11.8%) developed AKI during hospitalization. Although malignant pleural effusions usually accompanied with poor outcome, worse prognosis in patients with NMPE due to organ dysfunction was being noticed [19]. AKI should be considered as a systemic problem leading to multi-organ injury and even mild AKI is related to short or long adverse events. Thus, AKI might be crucial to be prevented and treated to improve NMPE patients’ prognosis.

We hypothesized that the volume of effusions related to AKI, which was verified in multivariate logistic regression after adjusting all the covariates we thought related to AKI. Large pleural effusions can alter respiratory mechanics and lead to abnormal gas exchange, reduction in diaphragmatic function or pleural irritation. Animal models have been used to study the effects of varying pleural effusion volumes and have shown that increased pleural effusion volume is associated with decreased PaO2, increased intrapulmonary shunts and a significant decrease in left ventricular preload [20‐22]. These have been verified in human patients. Razazi K et al. reported that drainage of large pleural effusions improved oxygenation and end-expiratory lung volume [9]. Another prospective study showed improved preload, systolic and diastolic function and hemodynamic changes after drainage of large effusions [10]. What's more, in a case report of a patient with acute kidney injury in the course of a large pleural effusion, the authors suggested an interdependence between pleural effusion and kidney injury, which was attributed to neurohormonal derangements, affecting the secretion of natriuretic peptides and/or antidiuretic hormone [11]. In another study of patients having non-cardiac pleural effusion with normal blood pressure and renal function, the renal artery resistance index (RI) and pulsatility index (PI) of patients having effusion were higher than those normal controls [23]. The above hypotheses, suggesting that pleural effusion may lead to complicated pathophysiological changes in kidney through neurohumoral regulation, are proposed by clinical manifestations and rationality needs further verification. We also need consider that pleural effusion is one of the prognostic scores of pneumonia and acute pancreatitis [24, 25], and pleural effusion volume is associated with poor outcomes in these acute conditions [26, 27]. For non-specific conditions, a prospective multicenter study of hospitalized patients with PE showed that large effusions were associated with higher mortality than small effusions at 1 year after adjustment for other factors [13]. AKI was known to be a measurable prognostic marker in several types of disease, as demonstrated by many studies [28‐30]. Therefore, size of effusions could be expected to be related to AKI for their common association with disease severity which can’t be fully adjusted.

A meta-analysis showed that older age was associated with a higher risk of AKI, but this effect was attenuated by lower eGFR or higher albumin-to-creatinine ratio (ACR), and eGFR and ACR were consistent, strong risk factors for AKI [31]. In our study, the odds ratio of age was 1.68 (95%CI 1.49–1.92; p < 0.001) in univariate analysis, but in multivariate analysis and LASSO analysis, the odds ratio of age was attenuated with no statistically significant difference, although it was selected by LASSO analysis, which might be due to other stronger risk factors. Proteinuria and eGFR remained independent risk factors in our study. 37.5% of patients had pneumonia, so infection was a very common problem in NMPE patients if abdominal and intracranial infection were added. We found vancomycin/teicoplanin was an independent contributor to AKI with a relatively high odds ratio (OR = 1.97, 95% CI 1.38–2.81). The exact mechanism by which vancomycin/teicoplanin induces nephrotoxicity is unclear, which was reported to be associated with oxidative stress [32]. Although vancomycin/teicoplanin is not the preferred anti-infective agent because of its side effects, it is one of the few antibiotics of choice for methicillin-resistant Staphylococcus aureus (MRSA) infections. Clinicians should carefully monitor urine output and creatinine when using vancomycin/teicoplanin. Our results suggest that both spironolactone and loop diuretics are risk factors for AKI in multivariate analysis with a relative high odds ratio. This may be related to concomitant heart failure, renal insufficiency and cirrhosis, or reduced renal perfusion with diuretics. Therefore, a better way to use diuretics in NMPE patients needs to be further investigated. In addition, the use of NSAIDs was also an independent risk factor for HA-AKI with a high odds ratio, which may be related to the hemodynamic changes or damage to the renal interstitium from NSAIDs [33]. In addition, uncontrolled infection causing fever or rapid volume depletion due to diaphoresis after NSAIDs use could be confounders, so NMPE patients may be at high risk when using NSAIDs and should be used cautiously or in reduced doses when it's necessary to improve symptoms. The Charlson comorbidity index (CCI) was a predictor of AKI in many diseases [34‐36]. Previous studies showed that the number of comorbidities and CCI were predictors of both short-term and long-term mortality in patients with PE [13, 37]. We hypothesized that NMPE patients hospitalized with multiple comorbidities may have had an acute change in condition during admission leading to acute kidney injury, considering a high burden of comorbidity was quite common in NMPE patients (the median CCI was 2 in our study and 5 in the previous research) [13]. The odds ratio of CCI was 1.69 in our univariate analysis (95%CI 1.52–1.87), but was not significant after adjustment for other variables. We found that effusion volume might work as a mediator between heart failure, pneumonia or eGFR and AKI, although the effect was small. This may support our hypothesis that the volume of PE was associated with AKI. Researches have shown that in the course of progression of pneumonia-related pleural effusion, imbalance of various pro- and anti-inflammatory factors, abnormal activity of the fibrinolytic system play an important role, which have been considered as mechanisms for the development of AKI in a variety of models [38‐40] as well. Thus, there may be a common pathway for the development of AKI and increase of pleural effusion. Besides, we hypothesized that the mechanism of the mediating effect of effusions on heart failure and eGFR might be hemodynamic changes. The above findings of an underlying mediating effect of volume can only be considered hypothesis-generating, showing possible associations but not necessarily causal relationships. It cannot be excluded that similar changes obtained by other approaches than the volume of effusions may contribute.

In fact, in the majority of cases, clinicians couldn't determine the course of the effusion because of its insidious onset. Meanwhile, most patients had more than one etiology and only a few of them had fluid examination results [37]. Thus, much remains to be done. Our study found AKI incidence was high among NMPE patients and analyzed the risk factors and the correlation between the size of effusion and AKI, attempting to provide some implications for clinicians to improve the prognosis of the underlying high mortality patients through the development of AKI. However, it has some limitations. First, measurement bias was likely because supine radiographs are only moderately sensitive and specific for assessing pleural effusions, but it only accounted for 3.4% in our study and we included these patients to get as close to the real scenario as possible. Second, due to the retrospective nature of the study, only a proportion of patients had urine output data, which may lead to an underestimation of the incidence of AKI grade [41]. In addition, we are unable to assess the fluid balance status of patients because of the lack of reliable data such as central venous pressure (CVP). Some important risk factors such as shock and sepsis might be missed, and instead, we included vasoactive drugs (used before AKI) as covariates in the multiple model. Third, we didn't follow up patients to get long-term outcomes. Forth, the outcome in our study was unbalanced, which might introduce sparse effects [42]. To evaluate this effect, we used Firth-type penalization and there was no significant difference between original multivariate logistic regression and Firth’s logistic regression model (see Additional file 1). Finally, this was a single-center study and there may be unmeasured confounders, so the causal relationship needs future studies supported by a larger sample from multiple centers.

The incidence of AKI is high among NMPE patients and large effusion volume is independent associated with AKI compared to small size. The effusion size acts as a mediator in heart failure, pneumonia, and eGFR.