Introduction
Acute kidney injury (AKI) is defined as an abrupt decrease in kidney function that could be induced by both kidney and extrarenal diseases.1 AKI after acute pulmonary embolism (PE) has been recently discussed owing to its potential high incidence and worse prognosis for those patients: the Registro Informatizado de la Enfermedad TromboEmbolica (RIETE) study reported an AKI incidence of approximately 25% to 30% after PE2 and the occurrence of AKI was independently associated with higher 30-day mortality and major bleeding.3 The diagnostic criteria of AKI, based on the Kidney Disease Improving Global Outcomes (KDIGO) guidelines, requires at least two serum creatinine (SCr) levels or constant measurement of urine output that brings difficulty in AKI identification, especially for large-scale multicentre registries.1 Thus, previous studies mostly estimated AKI according to the single SCr level on admission or collected suspected or confirmed AKI events based on judgement of physician or experience.2 4 Considering the latent progression of AKI, the AKI incidence after PE has been underestimated. Risk estimation and renal function recovery among these patients would be inaccurate.
According to the existing evidence, the higher risk of AKI among patients with PE is probably related to high-risk PE that could induce AKI due to hypofiltration of the systemic circulation. However, an AKI incidence of approximately 20% among patients with normotensive PE has been previously reported (31% in intermediate-risk and 14% in low-risk PE).3 The underlying mechanism was inferred that venous congestion induced by right ventricular (RV) dysfunction after PE might cause lower perfusion of the kidney. The RIETE study reported that nearly three quarters of the patients achieved an early recovery (within the first week), and a complete recovery rate of 73% was found among those with AKI after acute PE, while no further investigation has been performed.5 Therefore, we hypothesised that earlier anticoagulation therapy might contribute to earlier relief of clot burden and decreased tension of RV, thus improving renal function.
We performed a retrospective cohort study of patients with acute PE, using dynamic SCr levels to reveal the accurate incidence of AKI during hospitalisation, especially in normotensive patients. Afterwards, we investigated the potential risk factors for AKI and the impact of anticoagulation on renal function recovery.
Methods
Participants and study design
In this multicentre, retrospective cohort, patients who were subjectively confirmed PE with/without deep vein thrombosis (DVT) were consecutively recruited from four tertiary hospitals with high quality of PE management in Beijing, China: China-Japan Friendship Hospital, Beijing Hospital, Fuwai Hospital and Anzhen Hospital. Patients who were hospitalised from April 2015 to November 2021 and met the inclusion criteria were enrolled in this study and followed up for at least 2 years, based on an existing cohort. PE was confirmed by helical CT pulmonary angiography (CTPA), ventilation-perfusion lung scintigraphy (V/Q scan) or pulmonary angiography. Patients with an unclear history of acute onset or over 30 days from onset to PE diagnosis, haemodynamically unstable (high-risk) or without available SCr data on admission were excluded. Decisions on diagnosis and initiation, maintenance or change in the treatment pattern were at the discretion of the physicians and patients of the participating centres. Demographic data, medical history related to venous thromboembolism (VTE), risk factors for VTE, symptoms and signs on presentation, physical and laboratory examination results, imaging test results, types of diagnostic methods, diagnostic results, therapeutic management and clinical outcomes of PE during hospitalisation and the follow-up period (through telephone or outpatient clinic) were collected via an electronic data capture system. All recruited patients were followed up for up to 2 years.
The study was approved by institutional review boards and ethical committees of all the centres involved (the ethical committee of China-Japan Friendship Hospital, the ethical committee of Beijing Hospital, the ethical committee of Fuwai Hospital and the ethical committee of Anzhen Hospital). Written informed consent was obtained from all participants in the study according to the requirements of the ethical committee of each medical centre.
Definitions
Both PE severity and the simplified pulmonary embolism severity index (sPESI) score were estimated according to the 2019 ESC (the European Society of Cardiology) guidelines for acute PE.5 All available measurements of SCr after hospitalisation (including the follow-up period) for each enrolled participant were recorded for an identification of AKI (from the first time measured after hospitalisation and the last time before discharge, where available). AKI was diagnosed according to the KDIGO definition: increase in SCr by 0.3 mg/dL (26.5 µmol/L) within 48 hours; or increase in SCr to ×1.5-fold of baseline, which is known or presumed to have occurred within the prior 7 days. Baseline SCr levels were back-calculated from theoretical renal function according to the MDRD equation and in accordance with international KDIGO guidelines,6 assuming that baseline estimated glomerular filtration rate (eGFR) is 75 mL/min per 1.73 m2. For patients with only one SCr record, which was usually tested on admission, AKI and its severity were estimated by comparing this record with baseline SCr. Bleeding events included major bleeding and clinically relevant non-major bleeding were defined by the International Society on Thrombosis and Haemostasis criteria.7
According to KDIGO, AKI severity was defined as: stage 1, increase in SCr levels to ×1.5-fold of baseline SCr or ≥0.3 mg/dL (≥26.5 µmol/L) increase; stage 2, increase of more than 2.0–2.9 folds of baseline SCr; stage 3, increase of more than 3.0 folds of baseline SCr or ≥4.0 mg/dL (≥353.6 µmol/L). Contrast-induced nephropathy (CIN) was defined as an elevation of SCr of more than 25% or 0.5 mg/dL (44 µmol/L) from baseline within 48 hours following CTPA.8 The mean pulmonary arterial pressure, diameter of inferior vein (IV), the RV end-diastolic anteroposterior diameter and the transverse diameter of the RV were measured by echocardiography.
BNP/NT-proBNP elevation was determined by local clinicians, according to the cut-off value of local centre, patient’s age and other health conditions. RV failure was diagnosed by CTPA or echocardiography according to the 2019 ESC/ERS guidelines for acute PE (RV/LV diameter ratio ≥1.0 on echocardiography or CTPA).9
The day of AKI diagnosis referred to the first day that the KDIGO criteria of AKI was met. The day of AKI recovery referred to the day when SCr decreased to the baseline level. Early recovery of AKI was identified if the SCr level decreased to baseline within 7 days after AKI occurrence. The primary outcome of the study was AKI and the secondary outcome was the recovery of renal function.
Patient and public involvement
No patient involved.
Statistical analysis
Baseline characteristics were expressed in terms of descriptive statistics. Categorical variables were summarised as frequency (percentage). Continuous variables were presented as mean (SD) or median (IQR). P values were calculated by students’ t test, Mann-Whitney U test, χ2 test or Fisher exact test among different groups, where appropriate. Missing data were objectively recorded as shown in the tables. Imputation of missing values was not performed. To describe the temporal changes of SCr in high risk and patients with normotensive PE, a locally weighted smoothing method was used for the spline curves. Kaplan-Meier curves were portrayed to present the cumulative AKI rates in patients with normotensive PE by BNP/NT-proBNP elevation and/or RV failure and differences among strata were compared by Log-rank tests. Thereafter, univariable and multivariable COX regression models were performed to explore the risk factors for AKI. To estimate the effect of early anticoagulation (within 5 days, 6–10 days and more than 10 days after PE onset) during follow-up in this cohort, a logistic model was fitted to estimate the ORs and 95% CIs on AKI recovery. All tests were two sided and were considered statistically significant at a p value of <0.05. All analyses were performed using SAS V.9.4 software (Cary, North Carolina).
Results
Dynamic changes of renal function after an onset of acute PE
Of 604 patients diagnosed with PE who were identified during our study period, after screening for acute onset (less than 30 days), 115 patients were excluded due to a chronic characteristic. Consequently, a total of 461 patients with acute normotensive PE were enrolled in the final analysis (figure 1). A slight but significant increase of SCr was observed, peaking at around 20 days with an approximate elevation of 10 µmol/L, and then returned to baseline after 40 days after onset (online supplemental figure 1).
Supplemental material
Flowchart of the study. ǂincluding major bleeding and clinically relevant non-major bleeding (CRNMB) defined by the International Society on Thrombosis and Haemostasis (ISTH) criteria. AKI, acute kidney injury; PE, pulmonary embolism.
AKI incidence and risk factors
Based on the dynamic changes of SCr during hospitalisation, the incidence of AKI was 18.9% (87/461) in the normotensive patients. Among those who developed AKI, 72.4% (63/87) were at stage 1, 21.8% (19/87) were at stage 2 and 5.7% (5/87) were at stage 3. Comparisons of the baseline characteristics and outcomes between patients with and without AKI are shown in tables 1 and 2. Those with AKI had older age (median 74.0 vs 66.6 years old, p<0.0001), higher SCr levels (median 81 vs 69 µmol/L, p<0.0001), BNP (median 137.9 vs 76.9 pg/mL, p=0.0454), NT-proBNP (median 1999.0 vs 239.5 ng/mL) and D-dimer levels on admission, and higher rates of BNP/NT-proBNP elevation (72.4% vs 43.0%, p<0.0001).
Characteristics of patients with normotensive PE with/without AKI during hospitalisation
RV structural parameters, treatment and outcomes of patients with PE with/without AKI
Echocardiography revealed a significant higher mPAP and larger diameter of the IV among patients with AKI and non-AKI (48.0 mm Hg vs 41.5 mm Hg, p=0.0251, 16.0 mm vs 14.0 mm, p=0.0046, respectively). Low-molecular weight heparin was used less in patients with AKI (64.6% vs 77.0%, p=0.0201) with a numerically lower dose and the use of direct oral anticoagulants was not significantly different between those with or without AKI (6.1% vs 9.5%, p=0.3291). The rates of in-hospital all-cause death and bleeding events were both higher among the patients with AKI (5.7% vs 0.8%, p=0.0061, and 8.1% vs 1.9%, p=0.0072, respectively).
BNP/NT-proBNP elevation (HR 2.27, 95% CI 1.33 to 3.86, p=0.0026) and the history of chronic kidney disease (CKD) (HR 4.81, 95% CI 2.44 to 9.48, p<0.0001) were associated with the development of AKI during hospitalisation (table 3). Patients with AKI had higher rates of BNP/NT-proBNP elevation and RV failure, thus survival analyses were performed to identify the associations between cardiac and renal function (online supplemental figure 2). BNP/NT-proBNP elevation, RV failure or both were associated with increased AKI rates during hospitalisation.
Risk factors for in-hospital AKI of patients with normotensive PE
Earlier anticoagulation and recovery of renal function
Early recovery of renal function occurred in 44 of 87 (50.6%) among the patients with normotensive PE with AKI. After adjustment for cardiovascular diseases, CKD and age, anticoagulation administered within 5 days after the onset of acute PE (probably due to earlier diagnosis) was a protective factor for the early recovery of AKI (OR 0.258, 95% CI 0.079 to 0.844, p=0.025), indicating a reduction of approximately three fourths in the odds of unrecovered renal function, compared with those with delayed treatment (more than 10 days after PE onset) (figure 2).
Forest plot of multivariable logistic regression on the effects of earlier anticoagulation on AKI recovery. *The ORs of days to anticoagulation were estimated taking later anticoagulation (≥11 days after PE onset) as reference. OR<1 refers to earlier recovery. AKI, acute kidney injury; PE, pulmonary embolism.
Discussion
Based on a multicentre, retrospective cohort, our analysis revealed an incidence of AKI during hospitalisation among patients with acute PE and investigated the potential association between cardiac and renal function in patients with normotensive acute PE. Of note, we innovatively reported a more accurate incidence of AKI according to the dynamic laboratory examinations of SCr level and revealed the potential beneficial impact of earlier anticoagulation therapy on the early recovery of renal function. These findings provide important clues and possible rationale to explore the pathophysiology and potential management strategy of AKI after acute PE.
AKI secondary to non-renal diseases is of high incidence but tends to be ignored. Thus, the incidences of AKI varied across studies. Our analysis revealed that the incidence of AKI was 18.9% after acute normotensive PE, which was much higher than that in the general inpatient population in China (2.5% and 9.1% for community-acquired and hospital-acquired AKI, respectively).10 However, a study from Taiwan reported a lower incidence of AKI among patients with PE of 4.9%, most likely based on the International Classification of Diseases, Ninth Revision (ICD-9) in discharge diagnosis.4 The RIETE multicentre registry study, although consistent with our study, reported an AKI incidence of 29.5% at baseline according to the KDIGO criteria, based on data from 21 131 patients with PE with SCr levels only available on admission.2 The reasons may be multiple for the current differences in AKI incidences. First, the progression of AKI is latent in most cases, with a gradual escalation in creatinine or slow de-escalation in urine output during hospitalisation, especially for patients with CKD, older age, multiple comorbidities and PE of high risk.2 Thus, the application of dynamic SCr for the identification of AKI was vital in our study, and successfully provided a relatively accurate AKI incidence.
There were significant differences in baseline characteristics between the patients who developed AKI or not. Those with older age, DVT and cardiovascular and lung diseases were more likely to develop AKI. Those with AKI had higher cardiac troponin I, white cell counts and D-dimer levels than those without. This clinical finding was consistent with previous studies and indicated that patients with AKI had more severe condition of PE (including multiorgan damage and clot burden).
Except for the remarkable incidence of AKI, a significant trend of SCr elevation and recovery was observed among normotensive patients in our study. The mechanism was deduced as the lower perfusion of the kidney due to venous congestion induced by RV dysfunction after PE, and this was partly verified by the significant association between BNP/NT-proBNP, RV structural change by radiology and AKI in our study. The current opinion suggests that haemodynamically stable patients are at low risk of death, but those with signs of RV dysfunction would develop worse outcomes. Previous studies proved the association between renal failure/dysfunction and adverse outcomes of PE.11 It is widely accepted that a sudden pressure overload in the pulmonary circulation due to PE leads to RV dysfunction along with tricuspid regurgitation. High filling pressures secondary to volume overload in the RV would also result in the elevation of BNP/NT-proBNP, a marker for RV dysfunction.12 13 From our results, BNP/NT-proBNP elevation alone (without structural change in the RV) was associated with the occurrence of AKI, indicated that the cardiorenal interaction after acute PE may undermine renal function by RV dysfunction, not merely RV decompensation. Consequently, central venous pressure (CVP) rises and may lead to renal congestion and injury.14 Previous studies revealed that an elevation in right atrial pressure, which was correlated with CVP, but not a decrease in cardiac output and/or cardiac index, was associated with descending renal function.15 Renal venous hypertension can induce a decline in glomerular filtration rate (GFR),16 17 and persistent renal venous hypertension can result in the progression of kidney disease.18–21 Therefore, we inferred that the indicator for ventricular functional indexes was related to the impairment of renal function, exemplified by diameters of IV and BNP levels. The diameters of IV, as measured by echocardiography, were higher in AKI than patients with non-AKI in our study, although their RV diameters were comparable in patients with normotensive acute PE.
However, for BNP or NT-proBNP levels, we used elevation judged by clinician rather than unified cut-off values. It was suggested that the cut-off values for BNP in PE prognostic should be lower than those for congestive heart failure, while the recommended BNP or NT-proBNP levels in the assessment of RV function of PE remain undetermined.9 Additionally, the extraction of BNP/NT-proBNP is easily affected by the GFR. More studies on BNP/NT-proBNP among patients with PE with renal impairment are warranted.22
Our study innovatively illustrated the association between earlier anticoagulation therapy and early recovery of AKI, indicating a potential benefit in the protection of renal function and the outcomes of patients with acute PE. In accordance with our results, the RIETE study reported that 73% patients with AKI achieved renal recovery and among those, 80% was observed in the first 7 days, half of them recovered in the first 48 hours after the onset of AKI.5 The results highlighted the importance of individualised monitoring and decision-making in patients with AKI, as many patients will have better outcome. Delayed initiation of anticoagulation, usually due to delayed diagnosis or misdiagnosis, has been discussed in patients with acute VTE and higher rates of mortality and major bleeding were observed at 90-day follow-up but failed to reach a significant association between delayed anticoagulation and poor outcomes.23
CTPA is the currently widely applied diagnosis technique for PE and CIN is a usual cause of hospital-acquired kidney injury. Among those with available SCr obtained after CTPA (139 patients) in our study, the incidence of CIN was 12.3%, and 11 cases met the AKI criteria at the same time. Previous studies have presented incidence of CIN after CTPA of 14%24 and 13.1%.8 The incidence of CIN among patients with PE was higher than that among other patient populations (13% vs 5%),25 indicating that patients with PE may be more vulnerable to contrast and thus calling for more aggressive measurements to prevent renal impairment. Unfortunately, we were unable to track every patient’s renal function after CTPA in our cohort, which leaves uncertainty in the development and possible cause of AKI. Further analysis of the interaction between CTPA and RV dysfunction is required.
We acknowledge several limitations of our study. First, although dynamic records of SCr during hospitalisation was obtained, only available results were calculated in our analysis due to the retrospective design. Those patients who lacked frequent SCr measurements might be misdiagnosed for AKI. The actual incidence of AKI after acute PE was inevitably underestimated using real-world data. Second, only half of the patients underwent echocardiography during hospitalisation, so the evaluation of structural function of the heart and its correlation to renal function and outcomes was limited. Third, despite that it was deduced that RV function may be associated with the occurrence and early recovery of AKI, the lack of ultrasonic data on RV function at different time points made it difficult to determine the chronological order of RV and kidney function decline.
Conclusions
Renal impairment and AKI were highly prevalent among patients with normotensive PE. The occurrence of AKI was associated with right heart function. Patients developed AKI would benefit from early anticoagulation therapy for an earlier recovery of renal function.
This post was originally published on https://bmjopen.bmj.com