Abstract
Background:
Prior studies identified thresholds for mean arterial pressure (MAP<65 mmHg) and central venous pressure (CVP>12 mmHg) beyond which risk for cardiac surgery-associated acute kidney injury (AKI) increases. Optimal hemodynamic targets- i.e., where active protection from AKI is observed- are unclear; however, current guidelines suggest maintaining MAP>65 and CVP 8–12. The aim of this study was to identify hemodynamic ranges associated with both increased and decreased risk of AKI by evaluating narrow ranges of MAP, CVP, and joint exposure to MAP and CVP concurrently.
Methods:
In a retrospective cohort study of adults undergoing coronary artery bypass surgery, we fine-mapped the association between AKI and the total number of minutes spent in each of the following narrow hemodynamic ranges: 14 MAP ranges in increments of 5 mmHg (45–115), 10 CVP ranges in increments of 2 mmHg (0–20), and 70 joint MAP/CVP ranges. Separate multivariable regression models estimated adjusted odds ratios (aOR) for each range including adjustments for correlations and multiple comparisons across ranges. Joint MAP/CVP ranges were grouped into five hemodynamic zones based on contiguity of the ranges and similarity of ORs observed across ranges in a color-coded heatmap. The five MAP/CVP zones were included in a single regression model to assess risk for AKI associated with time spent in each hemodynamic zone, independent of time spent in other zones.
Results:
In 1199 participants, incidence of AKI was 28%. For every 5-min spent in each hemodynamic range, risk of AKI was significantly increased in MAP range 45–50 (aOR 1.18; p=.002), 50–55 (aOR 1.13;p=.001), and 55–60 mmHg (aOR 1.06;p=.001); and significantly decreased in MAP range 90–95 mmHg (aOR 0.85;p<.001). Risk of AKI was significantly increased in CVP range 16–18 mmHg (aOR 1.07;p=.002) and significantly decreased in CVP range 4–6 mmHg (aOR 0.97;p=.025). In joint analyses, both MAP and CVP contributed to AKI risk estimates; risk decreased as CVP decreased within every MAP range and was significantly lower for joint ranges of CVP<8 and MAP>75. In analyses containing all five MAP/CVP hemodynamic zones, risk estimates suggested protection from AKI in zone 1 (high MAP/low CVP) and increased risk of AKI in zones 3–5 (low MAP/high CVP).
Conclusions:
Fine-mapping identified narrow ranges of MAP, CVP, and joint MAP/CVP associated with both AKI risk and protection. This report is among the first to characterize the association between joint MAP/CVP and AKI. Contrary to current guidelines, there was no evidence for protection associated with MAP 65–75 or CVP 8–12 mmHg.
Keywords: Cardiac surgery associated acute kidney injury, mean arterial pressure, central venous pressure, arterial hypotension, venous congestion, hemodynamic management, cardiac surgery
Introduction
Cardiac surgery-associated acute kidney injury (AKI) is common and related to subsequent chronic kidney disease and mortality.1,2 Intraoperative hemodynamic perturbations, most notably those related to arterial hypotension and cardiopulmonary bypass (CPB), are thought to be important causes of AKI.1–6 Venous congestion is also a potential contributor to AKI.7–13 Hemodynamic management strategies have garnered substantial interest for preventing AKI because, unlike many AKI risk factors, hemodynamic parameters can be modified.14 Current recommendations (Class IIa-moderate evidence) to reduce AKI15,16 call for implementation of Kidney Disease Improving Global Practice (KDIGO) guidelines,17 which include careful hemodynamic monitoring and protocol-based management of fluids and vasoactive agents; however, optimal hemodynamic goals and interventions have not been defined. Based on current evidence, standard clinical practice is to avoid hypotension, defined as MAP<65 mmHg, and to maintain CVP in the 8–12 mmHg range.17,18
Prior studies report an increase in AKI risk with time spent below a MAP threshold of 65 mmHg5,6 and/or above a CVP threshold of 12 mmHg.9,10 These studies provide valuable insights into the association between intraoperative hemodynamics and postoperative AKI; however, optimal hemodynamics- i.e. hemodynamic ranges at which protection from AKI is observed- remain unknown. Avoidance of harmful hemodynamic ranges may not equate to active protection. In addition, the nature of the relation- i.e. linear or non-linear- between AKI and MAP (or CVP) through the full set of hemodynamic ranges encountered is unclear. Some prior studies suggest risk of AKI is non-linear beyond a threshold, e.g., risk associated with MAP<55 is greater than MAP 55–65,6 and greater for CVP>16 than CVP>12.10 It remains unclear if AKI risk is constant above MAP of 65 (e.g., is risk associated with MAP 65–75 the same as 85–95 and below CVP 12?). Although the difference between MAP and venous pressure is the physiologic determinant of organ blood flow, the relation of a joint exposure of MAP and CVP with AKI has not been examined. Finally, some studies suggest differential effects on organ injury from hemodynamic ranges during vs. outside of CPB5,6,10; however, the associations of MAP, CVP, and joint MAP/CVP with AKI during each phase of cardiac surgery are not well defined.
To address these knowledge gaps, we conducted a retrospective cohort analysis of patients undergoing isolated coronary artery bypass grafting (CABG) surgery by fine-mapping the association of AKI to narrow ranges of MAP and CVP, separately, and as they occurred concurrently using a novel joint MAP/CVP exposure. Using state-of-the-art statistical methods,19 we quantified associations of AKI with each narrow hemodynamic range. We hypothesized AKI risk could be increased or decreased in non-linear fashion across the full set of these hemodynamic ranges. In sensitivity analyses, we examined the associations of hemodynamic ranges for MAP and CVP to AKI during each phase of cardiac surgery (pre-CPB, intra-CPB, and post-CPB) and in clinical sub-groups; and we explored the association of AKI to several hemodynamic zones suggested by fine-mapping of the novel joint MAP/CVP exposure.
Methods
The protocol for collection, analysis, and reporting of patient data was reviewed and approved by the Johns Hopkins Medicine Institutional Review Board (IRB- IRB00233545, Approved February 6, 2020). The requirement for written informed patient consent was waived by the IRB.
Data Sources and Covariates
Patients, identified from the Johns Hopkins Society of Thoracic Surgeons (STS) registry, were eligible for inclusion if they underwent isolated CABG requiring CPB at Johns Hopkins Hospital between July 1, 2016, and October 31, 2019. Patients were excluded sequentially for: 1) end stage renal disease or baseline serum creatinine >4mg/dL; 2) surgical procedure requiring more than one aortic cross clamp period; 3) intraoperative use of extracorporeal membrane oxygenation; 4) multiple surgical procedures during the same hospitalization (such as reoperation for bleeding or graft failure); and 5) insufficient hemodynamic (data in the electronic health record (EHR).
Data elements from the STS were merged with data from the EHR. The following covariates were included to adjust for baseline and intraoperative differences among participants:4–8 age, male/female, body mass index, comorbidities (history of hypertension, diabetes, myocardial infarction, congestive heart failure, stroke, peripheral vascular disease, chronic lung disease), medications (beta blocker, angiotensin converting enzyme inhibitor/receptor blocker, statin), preoperative left ventricular ejection fraction (LVEF), baseline hematocrit, baseline creatinine, surgical posting details (redo sternotomy, emergent case, use of intra-aortic balloon pump), and intraoperative characteristics (total procedure length, length of CPB, vasopressor inotrope dose, transfusion of packed red blood cells, and total crystalloid administered). Time-series MAP and CVP measurements were extracted and aligned at 1 minute resolution. Time-stamped intraoperative events, including anesthesia start and stop times and CPB start and stop times were aligned to hemodynamic data. All variables were verified and all missing or outlying observations were manually extracted from the original EHR record yielding a dataset unique for its size and accuracy. Preprocessing of time series data, including strategies to manage artifacts, outliers, and missing data, was conducted as previously published with modification20,21 (Supplement/Appendix1/p2–4).
Primary Exposures: Pressure ranges for MAP, CVP, and joint MAP and CVP
MAP was extracted from radial/femoral arterial line data. MAP and CVP catheters were placed in all patients in accordance with standard clinical protocol. Placement of CVP was confirmed by ultrasonography. MAP and CVP transducers were leveled at the tragus and zeroed per protocol. Anesthetic management was directed by anesthesiologists with subspecialty training in cardiac anesthesiology and echocardiography.
MAP values were divided into 14 different ranges in increments of 5 mmHg from 45–115 mmHg (e.g., 45–50, 50–55, etc.), and CVP values were divided into increments of 2 mmHg from 0–20 mmHg (0–2, 2–4, etc.). For each MAP or CVP range, we determined the cumulative number of minutes (“time spent”) in each range. For the joint concurrent MAP/CVP exposure, MAP was divided into increments of 10 mmHg and CVP into increments of 2 mmHg. This resulted in 70 joint MAP/CVP ranges (e.g., MAP 45–55/CVP 8–10 mmHg, MAP 55–65 mmHg/CVP 8–10 mmHg, etc.). For each joint range, we determined the cumulative number of minutes MAP and CVP values occurred concurrently within each of the 70 joint ranges.
Acute Kidney Injury Outcome (AKI)
The primary outcome was AKI at 48 hours after surgery as defined by KDIGO22 criteria using change in serum creatinine; i.e., an increase in serum creatinine ≥ 0.3 mg/dl from baseline. As a secondary outcome, we defined AKI by KDIGO criteria using the 48-hour criteria or increase in serum creatinine ≥1.5X baseline within 7 days after surgery. Consistent with recent studies, KDIGO urine output criteria were not included due to potential confounding by diuretic administration.10
Statistical Analysis
Baseline and intraoperative covariates were examined and compared by the primary outcome and tertile of exposure to MAP<65 mmHg and CVP>12 mmHg. Continuous variables are reported as mean ±SD, and categorical variables as proportion (%). The cumulative time of exposure in each hemodynamic range was expressed as the total number of minutes within each MAP, CVP, or joint MAP/CVP range and reported as mean ±SD and as proportion of subjects with at least 5 minutes in each range.
Primary Analysis
Logistic regression was used to determine associations between AKI and each narrow hemodynamic range. Separate regression models were generated for each of the following ranges: 14 MAP ranges, 10 CVP ranges, and 70 concurrent MAP/CVP ranges. By regressing AKI on total minutes in each separate range, odds ratios (OR) and 95% confidence intervals (CI) were determined for each range. This approach was termed “fine-mapping” (Supplement/Appendix2/p6–7). ORs/95% CIs are reported unadjusted and adjusted for covariates and expressed per five minutes spent in each range to enhance interpretability of point estimates. Because intraoperative covariates- i.e., vasoactive/inotropes, transfusions, and crystalloids- may be mediators, we report separate results from regression models excluding and including intraoperative covariates. In addition, for regression models adjusted for all covariates, we also reported the 95% CIs with correlation and multiplicity adjustment19, which corrects for correlation between hemodynamic ranges and for multiple testing. Correlation and multiplicity adjustment calculates the correlation between tests and adjusts the 95% CI by the number of correlated tests to maintain the Family Wise Error Rate (see Supplement/Appendix2/p5). ORs for each of the 70 joint MAP/CVP range were plotted in a heatmap color-coded by OR estimate. After visual inspection of this heatmap, we grouped the joint MAP/CVP ranges into five hemodynamic zones based on contiguity of the ranges and similarity of ORs for use in exploratory analyses. The five MAP/CVP zones were included in a single regression model to assess independence of risk estimates for AKI across zones.
Sensitivity Analyses
To account for correlation among neighboring ranges in the fine-mapping approach, we examined the associations of AKI with a set of wider hemodynamic ranges for MAP and CVP (wider ranges are less correlated) and compared results to those obtained from fine-mapping. We also compared risk estimates for each hemodynamic range for the entire surgery to those obtained during each of the following surgical phases: pre-CPB, intra-CPB, and post-CPB. Additionally, we examined results stratified by clinically relevant subgroups: LVEF above vs below 40%, estimated glomerular filtration rate (eGFR) above vs below 60 mL/min/1.73m2, and patients presenting in shock (defined by emergent surgery and/or preoperative balloon pump) vs those without.
Results were considered significant at p<0.05. The sample size was determined by including every available patient who met inclusion/exclusion criteria within the study period. No a priori power calculation was performed. Analyses were performed using R version 4.3.2.23 All code is publicly available with a simulated data set (https://github.com/jhuwit/fine_mapping_sim).
Results
In total, 1372 individuals underwent isolated CABG; after exclusions, 1199 were analyzed (Fig. 1). Characteristics of the study sample stratified by AKI are shown in Table 1. Of the 1199 subjects, 338 (28%) developed AKI within 48 hours after surgery. Patients who developed AKI were older, had more comorbidities, lower hematocrit, and higher creatinine at baseline. Those who developed AKI also had higher STS predicted mortality, longer surgical and CPB times, and were more likely to receive blood transfusion. Time below MAP of 65 mmHg was divided into tertiles (Supplement/Appendix3a/p6). Higher tertiles of time below MAP of 65 mmHg were associated with higher vasoactive inotrope dosing, longer durations of CPB and surgery, and more blood transfusions. Similarly, time above CVP of 12 mmHg was divided into tertiles with higher tertiles associated with higher BMI, congestive heart failure and lung disease, and longer durations of CPB and surgery (Supplement/Appendix3b/p7).
Figure 1:
Patient Flow Diagram
Table 1:
Characteristics of study sample.
| Total N= 1,199 |
AKI N= 338 |
No AKI N= 861 |
p-value | |
|---|---|---|---|---|
| Baseline characteristics | ||||
| Female, n (%) | 270 (23) | 85 (25) | 185 (21) | 0.197 |
| BMI, kg/m2, mean (SD) | 29.7 (8.1) | 31.1 (6.8) | 29.2 (8.5) | <0.001 |
| Hypertension, n (%) | 1,016 (85) | 304 (90) | 712 (83) | 0.002 |
| Diabetes, n (%) | 181 (15) | 46 (14) | 135 (16) | 0.417 |
| Myocardial infarction, n (%) | 411 (34) | 121 (36) | 290 (34) | 0.530 |
| Congestive heart failure, n (%) | 320 (27) | 125 (37) | 195 (23) | <0.001 |
| Intra-aortic balloon pump, n (%) | 88 (7.3) | 26 (7.7) | 62 (7.2) | 0.865 |
| Stroke, n (%) | 77 (6.4) | 20 (5.9) | 57 (6.6) | 0.752 |
| Peripheral vascular disease, n (%) | 111 (9.3) | 40 (12) | 71 (8.2) | 0.069 |
| Chronic lung disease, n (%) | 188 (16) | 70 (21) | 118 (14) | 0.004 |
| Beta blocker, n (%) | 707 (59) | 210 (62) | 497 (58) | 0.183 |
| ACEI/ARB, n (%) | 196 (16) | 52 (15) | 144 (17) | 0.633 |
| Statin, n (%) | 622 (52) | 174 (51) | 448 (52) | 0.914 |
| Hematocrit, %, mean (SD) | 39.2 (5.4) | 37.1 (6.0) | 40.0 (5.0) | <0.001 |
| Redo-sternotomy, n (%) | 27 (2.3) | 5 (1.5) | 22 (2.6) | 0.361 |
| Emergency case, n (%) | 34 (2.8) | 12 (3.6) | 22 (2.6) | 0.459 |
| STS mortality risk, %, mean (SD) | 1.8 (2.8) | 2.6 (3.9) | 1.5 (2.1) | <0.001 |
| Intra-operative characteristics | ||||
| Length of Procedure, minutes, mean (SD) | 395 (85) | 413 (109) | 388 (72) | <0.001 |
| Length of cardiopulmonary bypass, minutes, mean (SD) | 98 (36) | 102 (40) | 96 (34) | 0.009 |
| Vasopressor inotrope dose*, weighted dosing with no units, mean (SD) | 0.40 (0.36) | 0.39 (0.34) | 0.40 (0.37) | 0.603 |
| Total Crystalloid, ml, mean (SD) | 2,280 (2,229) | 2,171 (2,162) | 2,323 (2,255) | 0.289 |
| 2 or more units | 212 (18) | 96 (28) | 116 (13) |
Abbreviations: BMI, body mass index, ACE/ARB, angiotensin converting enzyme inhibitor/angiotensin II receptor blocker, eGFR, estimated glomerular filtration rate, STS, Society of Thoracic Surgeons
Values are mean (SD) or otherwise n (%) as noted
Vasopressor inotrope dose calculated as previously published by de La Hoz et. al.5
Association of AKI with Mean Arterial Pressure
Cumulative time spent in each MAP range is shown in Supplement/Appendix4/p8.
Associations between time spent in each MAP range and AKI are shown in Fig. 2. Visual inspection of point estimates through all MAP ranges suggests a nonlinear association between AKI and MAP. Specifically, a significant increase in odds of AKI was associated with the amount of time spent in each of the following MAP ranges (ORs shown are per 5 minutes in each range): 45–50 (adjusted OR 1.18;1.06–1.30; p=.002), 50–55 (aOR 1.13;1.05–1.21; p=.001), and 55–60 mmHg (aOR 1.06;1.03–1.09; p=.001). The odds of AKI decreased as MAP ranges increased, with a significant decrease in odds associated with the amount of time spent in MAP range 90–95 (aOR 0.85 per 5-minutes in range;0.78–0.92; p<.001).
Figure 2: Association of acute kidney injury (AKI) with cumulative time spent in narrow ranges of mean arterial pressure (MAP).
Separate logistic regression models estimated odds ratios (OR) and 95% confidence intervals for AKI within each MAP range. ORs are expressed per five minutes in range. Four sets of models are shown for each range- unadjusted, adjusted without intraoperative covariates, adjusted for all covariates, and adjusted plus correction for correlation of data and multiple testing. Brackets denote numbers included in range while parentheses denote numbers up to the integer.
Sensitivity analyses examined the association of AKI with a series of wider MAP ranges (10, 15, and 20 mmHg increments). These analyses showed significantly increased risk at MAP<65 and significantly decreased risk at MAP 90–105 (Supplement/Appendix 5a/p9), consistent with the fine-mapping results. Associations between AKI and MAP ranges during the pre-CPB, intra-CPB, and post-CPB phases are shown in Supplement/Appendix5b/p10. ORs for each MAP range were generally similar among the phases and comparable to the entire surgical period (with some exception at the highest MAPs), demonstrating increased risk at MAP <60 and decreased risk at MAP 90–95. ORs followed a similar pattern when stratified by LVEF, eGFR, and shock (Supplement/Appendix5c/p11), with point estimates suggesting increased risk at MAP<60 and decreased risk at MAP ~90–100 for all sub-groups.
Association of AKI with Central Venous Pressure
Cumulative time spent in each CVP range is shown in Supplement/Appendix6/p12.
Associations between time spent in each CVP range and AKI are shown in Fig. 3. Visual inspection of point estimates through all CVP ranges suggests a nonlinear association with AKI, with risk rising from a low at 4–6 to a high at 16–18 mmHg. In adjusted models, CVP ranges >8 mmHg were associated with higher point estimates for AKI for each 5-minutes in range, with significantly increased odds of AKI observed for CVP 16–18 mmHg after correlation and multiplicity adjustment (aOR 1.07; 1.03–1.12; p=.002). CVP ranges of 12–14 and 14–16 mmHg were also associated with increased risk in adjusted models, but CIs included one after correlation and multiplicity adjustment. The CVP range of 4–6 mmHg was associated with significantly decreased odds of AKI in adjusted models (aOR 0.97; 0.95–1.00; p=.025), but the association was no longer significant after correlation and multiplicity adjustment (aOR 0.97; 0.94–1.01; p=0.108).
Figure 3: Association of acute kidney injury (AKI) with cumulative time spent in narrow ranges of central venous pressure (CVP).
Separate logistic regression models estimated odds ratios (OR) and 95% confidence intervals for AKI within each CVP range. ORs are expressed per five minutes in range. Four sets of models are shown for each range- unadjusted, adjusted without intraoperative covariates, adjusted for all covariates, and adjusted plus correction for correlation of data and multiple testing. Brackets denote numbers included in range while parentheses denote numbers up to the integer.
Sensitivity analyses examined the association of AKI using a series of wider CVP ranges (4, 6, and 8 mmHg increments). In each analysis, risk estimates for AKI were lower at CVP <8 and higher at CVP >8 mmHg, and odds of AKI were significantly decreased in CVP range 4–8 mmHg (aOR 0.98; 0.96–1.00; p=0.045), consistent with the fine-mapping results (Supplement/Appendix7a/p13). Associations between AKI and CVP ranges during the pre-CPB, intra-CPB, and post-CPB phases are shown in Supplement/Appendix7b/p14. ORs for each CVP range were generally similar among the phases and comparable to the entire surgical period, with point estimates suggesting increased risk for CVP 16–18 and decreased risk for CVP 4–8 mmHg. Unsurprisingly, associations of CVP ranges with AKI during CPB were more variable. ORs followed a similar pattern stratified by LVEF, eGFR, and shock (Supplement/Appendix7c/15), with point estimates suggesting highest risk at CVP 16–18 and lowest risk at CVP 4–6.
Association of AKI with Joint Mean Arterial Pressure/Central Venous Pressure Ranges
Cumulative time spent in each joint MAP/CVP range is shown in Supplement/Appendix8/p16.
Visual inspection of the joint MAP/CVP plots suggested greatest unadjusted risk at ranges with low MAP and high CVP (upper left corner, Fig. 4a) and lowest risk with high MAP and low CVP (lower right corner, Fig. 4a). For example, each 5-minutes at MAP of 55–65 mmHg/CVP 16–18 mmHg was associated with significantly increased odds of AKI (OR 1.57, 1.31–1.91, p<.001), while each 5-minutes at MAP of 95–105 mmHg and CVP 6–8 mmHg was associated with significantly decreased odds of AKI (OR 0.71, 0.56–0.87, p=.003). Additionally, for any given CVP (row in Fig 4a), point estimates for AKI risk tended to decrease as MAP increased; and, for any given MAP (column in Fig 4a), point estimates tended to decrease as CVP decreased. After adjustment for covariates (Supplement/Appendix9a/p17) and correlation and multiplicity adjustment (Supplement/Appendix9b/p17), fewer individual ranges were statistically significant, but results were qualitatively similar.
Figure 4: Association of acute kidney injury (AKI) with cumulative time spent in narrow ranges of joint concurrent MAP/CVP.
MAP and CVP were examined concurrently at 1-minute intervals. We determined the cumulative time when both MAP and CVP were within each of 70 joint MAP/CVP ranges (a) or 5 joint MAP/CVP zones (b). Separate logistic regression models estimated odds ratios (OR) and 95% confidence intervals for AKI within each joint MAP/CVP range or zone. ORs are expressed per five minutes in range.
ORs ≥ 1.03 are shaded red, ORs ≤0.97 are shaded blue, with intensity of color denoting increasing effect size. An asterisk denotes p <.05. (a) Unadjusted ORs/95% CIs in each of the 70 joint MAP/CVP ranges. (b) After visual inspection of the plot in 4a, the 70 joint MAP/CVP ranges were grouped into five hemodynamic zones based on contiguity of the ranges and similarity of odds ratios with AKI outcome. Unadjusted and adjusted (for all covariates) ORs/95% CIs are displayed for each of the 5 MAP/CVP zones.
We categorized the 70 joint MAP/CVP ranges into five hemodynamic zones based on similarity and contiguity of risk estimates and determined the association of AKI with time spent in each zone, unadjusted and adjusted for all covariates (Fig. 4b). Odds of AKI were significantly increased for zone 5 (upper left corner) (aOR 1.04;1.02–1.06; p<.001) and decreased for zone 1 (lower right corner) (aOR 0.88; 0.80–0.94; p=.001). In order to explore the independent risk of AKI associated with each of the 5 MAP/CVP zones, we assessed all 5 zones in a single regression model adjusted for all covariates. In this model, point estimates suggested protection from AKI in zone 1 and increased risk in zones 3–5 with aOR/95% CI as follows in zones 1–5, respectively: 0.93/0.84–1.03;(p=.181), 1.02/0.99–1.05;(p=.165), 1.03/1.01–1.05; (p=.002), 1.03/1.01–1.05; (p=0.01), and 1.06/1.03–1.09 (p<.001).
In sensitivity analyses for the joint MAP/CVP exposure by zone, risk estimates for each phase of surgery were generally similar to those of the entire surgical period (Supplement/Appendix10a/p18), with point estimates suggesting greatest risk in zones 4 and 5 and greatest protection in zone 1. Point estimates for each hemodynamic zone were generally similar when stratified by LVEF, eGFR, and shock (Supplement/Appendix10b/p19–20) as well.
We also evaluated AKI at 48 hours and 7 days as a secondary outcome. An additional 42 patients developed AKI within 7 days for a total of 380/1199 (32%). In total, 312 (82%) developed stage 1 injury, 61 (16%) stage 2, and 7 (2%) stage 3. Results were similar to the primary AKI outcome (Supplement/Appendix 11/p21).
Discussion
We sought to identify hemodynamic ranges for MAP, CVP, and joint MAP/CVP associated with both increased and decreased risk for AKI. To accomplish this goal, we deployed a novel fine-mapping approach that evaluated associations within narrow hemodynamic ranges and adjusted for correlated data and multiple comparisons.19 We identified significantly increased risk for AKI associated with MAP ranges 45–60, and we identified significantly decreased risk for MAP 90–95, not previously reported. We identified significantly increased risk for AKI associated with CVP 16–18 and decreased risk associated with CVP 4–6. Hemodynamic ranges for a novel joint MAP/CVP exposure were also associated with both increased and decreased risk of AKI. Both MAP and CVP contributed to AKI risk estimates, with odds of AKI decreasing as MAP increased and CVP decreased through the set of observed ranges. Contrary to current practice, we found no evidence that MAP 65–75 or CVP 8–12 was associated with protection from AKI.
This is among the first studies to evaluate a series of narrow MAP ranges for associations with both increased and decreased risk of AKI. Most prior studies examined pre-determined thresholds for high risk (e.g. MAP<65) or time spent in the “worst” risk range.5,6,9,10,24 These methods are not well-suited to identify hemodynamic ranges associated with protection from AKI. Without a priori knowledge of low-risk ranges or a “best” range, we examined associations of AKI with all hemodynamic ranges. We observed decreased odds of AKI (i.e. protection) associated with MAP ranges 90–105 not previously reported. In addition to ranges associated with decreased risk, we identified several MAP ranges (45–60) associated with increased risk. Our findings are consistent with prior reports that suggested a non-linear increase in risk of AKI below a threshold of MAP<65.5,6 In our study, point estimates of AKI risk were not consistently <1 (indicating protection) until MAP>80. Another cardiac surgery study reported neither increased risk nor protection associated with MAP 65–75.6 Together, these observations suggest that the currently recommended threshold of MAP>65 may not be optimal to confer protection from AKI.
This is also the first study to assess a series of narrow CVP ranges, rather than a prespecified “high risk” threshold, to identify CVP ranges associated with both increased and decreased risk of AKI. We observed decreased odds of AKI (i.e. protection) associated with CVP 4–6 and similar results at CVP 6–8. Point estimates suggested increased risk of AKI for CVP ranges above 8, with odds of AKI significantly increased at CVP 16–18. A reduction in risk of AKI at modest CVPs has not been reported. Increasing risk associated with sequential CVP ranges >12 is consistent with findings from Lopez at al.10 Our results do not suggest that CVP 8–12 confers protection from AKI.
Furthermore, very few studies24 have evaluated the joint contributions of MAP and CVP to AKI, despite evidence that low MAP5,6 and high CVP9,10 contribute separately to AKI. Using a novel joint concurrent MAP/CVP exposure, our results suggest that both MAP and CVP contribute to risk estimates for AKI, with non-linear changes in risk visualized when one variable was constant and the other varied (see fig 4a). These findings are consistent with a prior study that reported that both CVP and perfusion pressure (i.e., MAP minus CVP)- a parameter approximating our concurrent joint MAP/CVP exposure, were associated with AKI.24 In exploratory analyses, we estimated the risk of AKI associated with joint MAP/CVP after categorizing exposure into five risk zones. Results from these analyses were consistent with all others, demonstrating greatest risk at MAP<75/CVP>12 and lowest risk (protection) at MAP>75/ CVP<8.
In sensitivity analyses, associations between AKI and hemodynamic ranges for MAP, CVP, and joint MAP/CVP were remarkably consistent by phase of cardiac surgery and clinical subgroups. A few prior studies examined associations of hemodynamics to renal injury by phase of cardiac surgery.5,6,10 Our results suggest that MAP and CVP have potential to contribute to both harm and protection from renal injury during each phase of cardiac surgery. Additional studies are needed to clarify the contributions of MAP and CVP to AKI in each phase of surgery.
Current consensus recommendations15,16 are for implementing KDIGO guidelines in cardiac surgical patients at high risk for AKI, despite limited evidence.25 While hemodynamic monitoring and protocol-based treatment for hemodynamics are advised,17 there is no consensus on monitoring strategy, hemodynamic goals, or therapeutic interventions. In standard clinical practice, continuous arterial pressure monitoring and hypotension avoidance are common, as is CVP monitoring and transesophageal echocardiography. Cardiac output monitoring is not routine in CABG at our center and most US hospitals. In the PrevAKI randomized clinical trial, implementation of a KDIGO bundle, which included optimization of hemodynamics (MAP, stroke volume variation, and cardiac index) guided by cardiac output monitoring, reduced absolute risk of AKI by 17%.18 Others have proposed a conceptual approach called protective hemodynamics based on a meta-analysis of clinical trials of blood pressure targets in perioperative and critical care patients.26 It is interesting to note these trials, which target specific blood pressure goals (not CVP or cardiac output), suggest lower mortality in patients randomized to a lower vs. higher MAP target; however, the vast majority of these trials were from non-cardiac patients and those in the intensive care unit (not operating room). Only two intraoperative strategies in cardiac surgery are currently known to provide protection from AKI (moderate quality evidence): goal directed oxygen delivery during CPB and remote ischemic preconditioning.25,27
In aggregate, our results from both separate and joint MAP/CVP analysis suggest that current guidance to maintain MAP >65 and CVP 8–12 (or <12) may not be optimal to confer protection from AKI;17,18 however, a causal association cannot be made between hemodynamic ranges and protection or risk of AKI from this observational study. It is possible that hemodynamics are only a surrogate for severity of illness or other physiologic processes that directly cause AKI. Randomized clinical trials are needed to determine the causes and treatments for AKI and to identify optimal hemodynamic goals for individual patients.
Limitations
This single center CABG cohort study may not generalize to all centers or all cardiac surgical procedures. Our non-randomized study is subject to confounding, which we addressed with adjustment for numerous confounders. These adjustments had minimal impact on point estimates. Although we included more than 1000 patients, the logistic regression analyses for fine-mapping were constrained by small sample sizes and few outcomes in some individual ranges, particularly for analyses of joint MAP/CVP ranges. As a result, ORs in adjusted models for some MAP/CVP ranges were unstable. Modest sample sizes in some ranges may also have increased the risk of type 2 errors, which could have been compounded by rigorous adjustments to correct for multiple testing. Larger studies are needed to validate the associations between joint MAP/CVP ranges and AKI.
Finally, this study, like most other published studies relating MAP and CVP to AKI, does not account for all parameters that can contribute to AKI. We had no information on cardiac output, which is a critical determinant of renal blood flow3 or other determinants of renal oxygen delivery hemoglobin concentration and hemoglobin oxygen saturation. Examining the independent contributions to AKI of MAP, CVP, cardiac output (a surrogate for renal blood flow), and renal oxygen delivery are important areas for future study.
Conclusion
Using a novel fine-mapping approach, we report novel findings on hemodynamic ranges for MAP (MAP 90–95) and CVP (4–6) that are associated with decreased risk of (i.e. protection from) AKI, and we confirm prior reports of increased risk for MAP ranges<65 and CVP ranges>12. We report that both MAP and CVP contribute to risk of AKI using a novel concurrent joint MAP/CVP exposure and identify hemodynamic ranges of the joint MAP/CVP exposure associated with both increased and decreased risk of AKI. Contrary to prior reports, we found no evidence to suggest protection from AKI associated with MAP 65–75 or CVP 8–12 mmHg.
Supplementary Material
Key Points.
Question:
In a cardiac surgery cohort, will analysis of the association between acute kidney injury (AKI) and time spent in narrow ranges of mean arterial pressure (MAP), central venous pressure (CVP), and joint MAP/CVP identify ranges of AKI protection as well as AKI risk?
Findings:
Adjusted logistic regression of time spent within narrow hemodynamic ranges (“fine-mapping”) identified narrow ranges of MAP, CVP, and joint MAP/CVP associated with both AKI risk and protection.
Meaning:
Fine mapping identifies narrow MAP, CVP, and joint MAP/CVP ranges associated with AKI in patients undergoing coronary bypass surgery. This report is among the first to characterize the association between joint MAP/CVP and AKI. Contrary to current guidelines, there was no evidence for protection associated with MAP 65–75 or CVP 8–12 mmHg.
Funding
This work was supported by a Society of Cardiovascular Anesthesiologists Award (LG), NIH/NIA RF1 AG072387(CB), NIH/NIA P30AG021334 (KBR), NIH/NCTAS KL2 (AK)
Conflicts of interests
Dr. Goeddel reports funding for a separate investigator initiated retrospective study from Retia Medical, Dr. Brown reports a data share and research grant from Medtronic, Dr. Khanna consults for Medtronic, Edwards Life Sciences, Philips Research North America, Baxter, GE Healthcare, Potrero Medical, Retia Medical and Caretaker Medical.
References
- 1.Thiele RH, Isbell JM, Rosner MH. AKI associated with cardiac surgery. Clin J Am Soc Nephrol. Mar 6 2015;10(3):500–14. doi: 10.2215/CJN.07830814 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Cheruku SR, Raphael J, Neyra JA, Fox AA. Acute Kidney Injury after Cardiac Surgery: Prediction, Prevention, and Management. Anesthesiology. Oct 9 2023;doi: 10.1097/aln.0000000000004734 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Nadim MK, Forni LG, Bihorac A, et al. Cardiac and Vascular Surgery-Associated Acute Kidney Injury: The 20th International Consensus Conference of the ADQI Group. J Am Heart Assoc. Jun 1 2018;7(11)doi: 10.1161/JAHA.118.008834 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Barbu M, Hjärpe A, Martinsson A, et al. Cardiopulmonary bypass management and acute kidney injury in cardiac surgery patients. Acta anaesthesiologica Scandinavica. Mar 2024;68(3):328–336. doi: 10.1111/aas.14357 [DOI] [PubMed] [Google Scholar]
- 5.de la Hoz MA, Rangasamy V, Bastos AB, et al. Intraop Hypotension and AKI, Stroke, and Mortality during and outside Cardiopulmonary Bypass: Retrospective Cohort Study. Anesthesiology. Jun 1 2022;136(6):927–939. doi: 10.1097/ALN.0000000000004175 [DOI] [PubMed] [Google Scholar]
- 6.Ngu JMC, Jabagi H, Chung AM, et al. Defining an Intraoperative Hypotension Threshold in Association with De Novo Renal Replacement Therapy after Cardiac Surgery. Anesthesiology. Jun 2020;132(6):1447–1457. doi: 10.1097/ALN.0000000000003254 [DOI] [PubMed] [Google Scholar]
- 7.Damman K, van Deursen VM, Navis G, Voors AA, van Veldhuisen DJ, Hillege HL. Increased central venous pressure is associated with impaired renal function and mortality in a broad spectrum of patients with cardiovascular disease. Journal of the American College of Cardiology. Feb 17 2009;53(7):582–588. doi: 10.1016/j.jacc.2008.08.080 [DOI] [PubMed] [Google Scholar]
- 8.Chen KP, Cavender S, Lee J, et al. Peripheral Edema, Central Venous Pressure, and Risk of AKI in Critical Illness. Clin J Am Soc Nephrol. Apr 7 2016;11(4):602–8. doi: 10.2215/CJN.08080715 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chen L, Hong L, Ma A, et al. Intraoperative venous congestion rather than hypotension is associated with acute adverse kidney events after cardiac surgery: retrospective cohort study. BJA. May 2022;128(5):785–795. doi: 10.1016/j.bja.2022.01.032 [DOI] [PubMed] [Google Scholar]
- 10.Lopez MG, Shotwell MS, Morse J, et al. Intraoperative venous congestion and acute kidney injury in cardiac surgery: an observational cohort study. British journal of anaesthesia. Mar 2021;126(3):599–607. doi: 10.1016/j.bja.2020.12.028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wang L, Zhang LJ, Liu J, Hu JF, Xiao LQ, Chen X. Insight Into Kidney Protection by Vacuum-Assisted Venous Drainage in Adult Cardiac Operation- A Multicenter Study. Circulation journal: official journal of the Japanese Circulation Society. Mar 24 2023;87(4):551–559. doi: 10.1253/circj.CJ-22-0308 [DOI] [PubMed] [Google Scholar]
- 12.Wang L, Hu L, Yan Dai Q, Qi H, Wang Z, Chen X. Intraoperative central venous pressure during cardiopulmonary bypass is an alternative indicator for early prediction of acute kidney injury in adult cardiac surgery. J Cardiothorac Surg. Apr 23 2024;19(1):262. doi: 10.1186/s13019-024-02734-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yang Y, Ma J, Zhao L. High central venous pressure is associated with acute kidney injury and mortality in patients underwent cardiopulmonary bypass surgery. J Crit Care. Dec 2018;48:211–215. doi: 10.1016/j.jcrc.2018.08.034 [DOI] [PubMed] [Google Scholar]
- 14.Rasmussen SB, Boyko Y, Ranucci M, de Somer F, Ravn HB. Cardiac surgery-Associated acute kidney injury - A narrative review. Perfusion. Oct 31 2023:2676591231211503. doi: 10.1177/02676591231211503 [DOI] [PubMed] [Google Scholar]
- 15.Brown JR, Baker RA, Shore-Lesserson L, et al. The Society of Thoracic Surgeons/Society of Cardiovascular Anesthesiologists/American Society of Extracorporeal Technology Clinical Practice Guidelines for the Prevention of Adult Cardiac Surgery-Associated Acute Kidney Injury. The Annals of thoracic surgery. Jan 2023;115(1):34–42. doi: 10.1016/j.athoracsur.2022.06.054 [DOI] [PubMed] [Google Scholar]
- 16.Peng K, McIlroy DR, Bollen BA, et al. Society of Cardiovascular Anesthesiologists Clinical Practice Update for Management of Acute Kidney Injury Associated With Cardiac Surgery. Anesthesia and analgesia. Oct 1 2022;135(4):744–756. doi: 10.1213/ane.0000000000006068 [DOI] [PubMed] [Google Scholar]
- 17.Group KDIGOKAKIW. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl. 2012;2:1–138. [Google Scholar]
- 18.Meersch M, Schmidt C, Hoffmeier A, et al. Prevention of cardiac surgery-associated AKI by implementing the KDIGO guidelines in high risk patients identified by biomarkers: the PrevAKI randomized controlled trial. Intensive Care Med. Nov 2017;43(11):1551–1561. doi: 10.1007/s00134-016-4670-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Crainiceanu CM, Goldsmith J, Leroux A, Cui E Functional Data Analysis with R. 2024:pages 47–53. [Google Scholar]
- 20.Mathis MR, Sathishkumar S, Kheterpal S, et al. Complications, Risk Factors, and Staffing Patterns for Noncardiac Surgery in Patients with Left Ventricular Assist Devices. Anesthesiology. Mar 2017;126(3):450–460. doi: 10.1097/ALN.0000000000001488 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Goeddel LA, Erlinger S, Murphy Z, et al. Association Between Left Ventricular Relative Wall Thickness and Acute Kidney Injury After Noncardiac Surgery. Anesthesia and analgesia. Apr 25 2022;doi: 10.1213/ANE.0000000000006055 [DOI] [PubMed] [Google Scholar]
- 22.Khwaja A KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract. 2012;120(4):c179–84. doi: 10.1159/000339789 [DOI] [PubMed] [Google Scholar]
- 23.(2023) RCT. R_: Language and Environment for Statistical Computing. R Foundation for Statistical Computing.
- 24.Jacquet-Lagrèze M, Berthod C, Schweizer R, et al. Venous congestion is more accurate than hypotension in predicting acute kidney injury after cardiac surgery. Comment on Br J Anaesth 2021; 126: 599–607. British journal of anaesthesia. Sep 2021;127(3):e81–e82. doi: 10.1016/j.bja.2021.05.028 [DOI] [PubMed] [Google Scholar]
- 25.Hariri G, Collet L, Duarte L, et al. Prevention of cardiac surgery-associated acute kidney injury: a systematic review and meta-analysis of non-pharmacological interventions. Critical care. Sep 12 2023;27(1):354. doi: 10.1186/s13054-023-04640-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.D’Amico F, Landoni G. Protective hemodynamics: a novel strategy to manage blood pressure. Curr Opin Crit Care. Dec 1 2024;30(6):629–636. doi: 10.1097/mcc.0000000000001205 [DOI] [PubMed] [Google Scholar]
- 27.D’Amico F, Pruna A, Putowski Z, et al. Low Versus High Blood Pressure Targets in Critically Ill and Surgical Patients: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Critical care medicine. Sep 1 2024;52(9):1427–1438. doi: 10.1097/ccm.0000000000006314 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.