Abstract
Background:
Prior studies of post-lung transplant acute kidney injury (AKI) have not accounted for confounding effects of primary graft dysfunction (PGD). We sought to test the impact of PGD on AKI risk factors and on the association of AKI with mortality.
Methods:
We included patients transplanted at the University of Pennsylvania from 1 hours. We used multivariable logistic regression to test the impact of PGD on AKI risk factors and Cox models to test association of AKI with one-year mortality adjusting for PGD and other confounders.
Results:
Of 299 patients, 188 (62.9%) developed AKI with 142 (75%) cases occurring by post-operative day 4. In multivariable models, PGD was strongly associated with AKI (OR 3.76, 95% CI 1.72–8.19, p=0.001) but minimally changed associations of other risk factors with AKI. Both AKI (HR 3.64, 95% CI 1.68–7.88, p=0.001) and PGD (HR 2.55, 95% CI 1.40–4.64, p=0.002) were independently associated with one-year mortality.
Conclusions:
Post-lung transplant AKI risk factors and association of AKI with mortality were independent of PGD. AKI may therefore be a target for improving lung transplant mortality rather than simply an epiphenomenon of PGD.
Keywords: acute kidney injury, lung transplantation, primary graft dysfunction, mortality, risk factors
Introduction
Acute kidney injury (AKI), the syndrome of rapidly dropping glomerular filtration rate (GFR) in response to an acute stressor, is common in the postoperative period after lung transplantation. Recent reports indicate that AKI incidence is as high as 70%, with 17–37% of recipients developing stage 2–3 AKI and 6–8% requiring renal replacement therapy (RRT).1–3 These studies have suggested that stage 2–3 AKI is associated with a three-fold increase in long-term mortality.2 As such, prevention and treatment of AKI could represent attractive targets to improve outcomes.
Studies to date, however, have not adequately accounted for primary graft dysfunction (PGD) when determining AKI risk factors and outcomes. PGD, an acute lung injury syndrome that develops in the allograft within the first 72 hours post-reperfusion, is a major driver of transplant outcomes during the first year.4 The studies of AKI clinical risk factors and outcomes that reported findings regarding PGD and AKI were underpowered due to small sample size or substantial missingness in PGD phenotyping, and contained minimal description of the impact of PGD on AKI risk factors and outcomes.5,6 Additionally, the timing of incident AKI relative to PGD has not been well-described.
Better understanding the relationship of PGD and AKI has the potential to inform management strategies that balance risks to the allograft and kidneys, particularly in regard to fluid and calcineurin inhibitor therapy. In addition, determining AKI’s association with long-term outcomes independent of PGD would lend greater justification to targeting post-transplant AKI for prevention and treatment. Using a prospective cohort of lung transplant recipients at our institution, we therefore sought to elucidate 1) the impact of PGD on AKI risk factors, 2) whether the association of AKI with mortality is independent of PGD, and 3) the post-transplant temporal relationship of AKI and PGD.
Patients and Methods
Study population
We included patients who underwent lung transplantation at the University of Pennsylvania (Penn) from June 2005 - October 2012 and were enrolled in the previously described prospective Lung Transplant Outcomes Group (LTOG) cohort.4 The time period was used to coincide with the Lung Allocation Score (LAS) waitlist system era and provide adequate time for outcome assessments.7 At Penn, all patients of age ≥18 years who were waitlisted for lung-only transplantation were approached for informed consent. No patients with a glomerular filtration rate (GFR) under 50 ml/min/1.73 m2 were listed for transplant. This study was approved by the Penn Institutional Review Board. A STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement is included in a supplemental file.
Clinical data collection and protocols
As part of the LTOG protocol, baseline and perioperative recipient clinical data were collected through the third postoperative day as previously described.4 We collected additional variables pertinent to AKI, including serum creatinine (SCr) measurements, history of diabetes mellitus, and pre-transplant mechanical ventilation. We also assessed vital status through one year post-transplant. We included a limited number of donor variables available in the LTOG database.
During the study period, all lung transplant recipients received induction therapy with intravenous methylprednisolone 500mg per transplanted lung as well as either basiliximab or daclizumab. Tacrolimus was initiated on the first day after transplantation with a goal trough concentration of 8–12 ng/ml, and either azathioprine or mycophenolate was used for cell cycle inhibition. Intraoperative cardiopulmonary bypass was used at the discretion of the operating surgeon. The most common indication was bilateral transplantation, though elevated intraoperative pulmonary arterial pressure and inability to tolerate single-lung ventilation prompted occasional use in patients undergoing unilateral transplantation. Of note, Penn did not transplant patients bridged with pre-transplant extracorporeal life support during the study period.
Defining AKI
We defined and staged AKI according to the consensus Kidney Disease Improving Global Outcomes (KDIGO) SCr and RRT criteria.8 Baseline SCr was the value measured immediately prior to the transplant operation. AKI was defined by 1) a 0.3 mg/dL SCr increase over any 48-hour period to a level ≥0.3 mg/dL above baseline, 2) a 50% increase in SCr over any 7-day period to a level ≥50% above baseline, or 3) need for acute RRT. We applied these criteria to the entire transplant hospitalization in order to determine the postoperative day of the first AKI episode. Each case of AKI was staged over the following 7 days (stage 2: SCr ≥200% of baseline; stage 3: SCr ≥300% of baseline, SCr ≥4mg/dL, or RRT use).8
Defining PGD
Based on prior studies, we defined PGD as grade 3 at 48 or 72 hours according to International Society for Heart and Lung Transplant (ISHLT) guidelines.4,9,10 We also performed sensitivity analyses defining PGD as grade 3 at any time during the first 72 hours. For both PGD definitions, onset was defined as the first day meeting grade 3 criteria.10 Two Penn physicians trained to interpret PGD reviewed chest radiographs through postoperative day 3 for the presence of diffuse parenchymal allograft infiltrates (classification Κ=0.95, reported previously).4 Conflicts were adjudicated by a third reviewer.
Statistical analysis
We tested the associations of each recipient and donor covariate with AKI using the unpaired t-test or Wilcoxon rank-sum test for continuous variables, and the χ2 or Fisher’s exact test for categorical variables. We tested the association of these covariates with AKI stage using Cuzick’s non-parametric test of trend.11 We constructed multivariable logistic regression models of risk factors for AKI by considering all baseline and operative variables that were associated with AKI at p<0.20 as well as any previously reported risk factors for AKI after lung transplantation.1,2 We fit logistic regression models with and without PGD to assess its impact on other AKI risk factors. Because PGD and AKI onset occurred within one day of each other in 70% of coincident cases (Figure 1A, B), and because clinical manifestations of PGD and AKI may not precisely reflect the timing of their underlying pathophysiologic processes, we did not infer a causal relationship between PGD and AKI from these models.
Figure 1.
Relative timing of primary graft dysfunction (PGD) and acute kidney injury (AKI). PGD was defined by meeting grade 3 criteria at 48 or 72 hours, and incident postoperative day was the first day such patients met grade 3 criteria. A. Overlay of the frequency of incident PGD and AKI by postoperative day. B. Distribution of the number of days by which PGD preceded AKI among the 50 patients who developed both PGD and AKI. Negative values represent cases in which AKI preceded PGD.
We tested the associations of AKI and AKI stage with one-year mortality using the log-rank test and Cox proportional hazard models to adjust for confounders.12 We pre-specified four explanatory variables to avoid overfitting,13 given the limited number of deaths: AKI, PGD, age, and transplant procedure (bilateral v. unilateral). For these analyses, we used AKI occurring by day 14 post-transplant, since AKI occurring later was more likely to be a non-specific marker of prolonged hospitalization due to complicated post-transplant course. One-year mortality was the primary outcome because it is clinically relevant while limiting the impact of intercurrent confounders such as chronic lung allograft dysfunction, and because of the established association of PGD with one-year mortality.4,14,15 In secondary analyses, we determined the association of AKI with 90-day mortality using logistic regression to adjust for PGD.
Because of patterns we noted in the distribution of AKI timing (Figure 1A), we performed exploratory analyses to assess whether AKI risk factors and outcomes varied by day of AKI onset. We divided AKI into cases occurring during the initial peak through postoperative day 5 (early) and those occurring during the second peak from day 6–14 (delayed). We utilized multinomial logistic regression to determine the associations of patient characteristics with early and delayed AKI, and Cox models to determine the associations of early and delayed AKI with one-year mortality.
Stata/IC 13 (StataCorp LP, College Station, TX) was used for all analyses, and a two-sided p<0.05 was considered statistically significant. Recipient missing data were rare and were addressed with complete case analysis. Recipients were not excluded in the event of missing donor data; donor data missingness is noted in tables as applicable.
Results
During the study period, 303 Penn LTOG enrollees underwent lung-only transplantation. Of these, 299 (98.7%) had complete recipient clinical and outcomes data and were included in the analysis. AKI developed in 188 (62.9%) patients during the transplant hospitalization: 108 (36.1%) stage 1, 47 (15.7%) stage 2, and 33 (11.0%) stage 3. Nineteen patients (6.4%) required RRT. AKI occurred a median of 2 (IQR 1–3) days post-operatively, with 95% of cases occurring by day 14. AKI that progressed to stage 2 or 3 was evident particularly early, with 78/80 (97.5%) of cases meeting AKI criteria by day 5 and a median of 2 (IQR 2–4) days from allograft reperfusion to peak AKI stage. There were two peaks in AKI incidence: an early peak through day 5 that overlapped substantially with PGD incidence, and a smaller delayed peak occurring from days 8–14 (Figure 1A).
Baseline characteristics by AKI status are shown in Table 1. Key baseline and operative variables associated with AKI included pre-transplant diagnosis of cystic fibrosis, bilateral transplant, cardiopulmonary bypass use, and units of blood products transfused during the operation. In the multivariable model, independent risk factors for AKI included increasing body mass index (BMI) and diagnoses of interstitial lung disease (ILD) and cystic fibrosis (CF), while preoperative angiotensin converting enzyme (ACE)-inhibitor use was protective (Table 2, Model 1). Bilateral transplantation did not remain significantly associated with AKI in the multivariable model; adjustments for recipient pre-transplant diagnosis, age, and mean pulmonary arterial pressure were largely responsible for this attenuation of effect. There was no trend in AKI rate by year of transplant (p=0.997), and adjusting Model 1 for transplant year minimally impacted the odds ratios of other covariates. Of note, there was little variability in induction immunosuppression or potentially nephrotoxic antimicrobial medications over the first 3 post-transplant days. Specifically, basiliximab was used in 94% of patients and vancomycin in 97%, while piperacillin-tazobactam, amphotericin, and colistin were used in 2%, <1%, and 0%, respectively, precluding meaningful analysis of associations with AKI.
Table 1.
Characteristics by AKI status and stage.
| AKI stage | |||||||
|---|---|---|---|---|---|---|---|
| No AKI (n=111) |
AKI (n=188) | p | 1 (n=108) | 2 (n=47) | 3 (n=33) | p | |
| PGD | |||||||
| Grade 3 at 48h or 72h | 10 (9.0) | 50 (26.6) | <0.001 | 19 (17.6) | 11 (23.4) | 20 (60.6) | <0.001 |
| Grade 3 any time 0–72h | 22 (21.8) | 79 (42.0) | <0.001 | 35 (32.4) | 21 (44.7) | 23 (69.7) | <0.001 |
| Recipient baseline | |||||||
| Age (years) | 59.9 (55.5–64.1) | 58.6 (50.6–62.4) | 0.009 | 59.5 (51.2–62.4) | 57.4 (50.1–62.7) | 55.2 (51.3–59.6) | 0.002 |
| Male sex | 61 (55.0) | 118 (62.8) | 0.183 | 74 (68.5) | 26 (55.3) | 18 (54.6) | 0.981 |
| Race | 0.110‡‡ | 0.298‡‡ | |||||
| Caucasian | 93 (83.8) | 164 (87.2) | 93 (83.8) | 44 (93.6) | 28 (84.9) | ||
| African American | 10 (9.0) | 20 (10.6) | 12 (11.1) | 3 (6.4) | 5 (15.2) | ||
| Other | 8 (7.2) | 4 (2.1) | 4 (3.7) | 0 (0) | 0 (0) | ||
| Diabetes mellitus | 12 (10.8) | 26 (13.8) | 0.449 | 16 (14.8) | 5 (10.6) | 5 (15.2) | 0.629 |
| BMI (kg/m2) | 26.3 (22.3–30.1) | 26.8 (22.2–30.7) | 0.734 | 27.7 (24.1–30.9) | 24.6 (20.4–30.0) | 24.5 (21.1–29.9) | 0.378 |
| Baseline renal function | |||||||
| Serum Cr (mg/dl) | 0.84 (0.70–0.97) | 0.80 (0.70–1.00) | 0.871 | 0.82 (0.69–1.00) | 0.80 (0.62–1.00) | 0.80 (0.77–1.10) | 0.521 |
| eGFR (ml/min/1.73m2) | 92.7 (76.4–98.3) | 93.5 (74.5–104.4) | 0.233 | 94.3 (75.3–104.6) | 94.1 (76.1–101.9) | 88.7 (64.5–108.8) | 0.707 |
| eGFR<60 ml/min/1.73m2 | 6 (5.4) | 14 (7.5) | 0.495 | 7 (6.5) | 1 (2.1) | 6 (18.2) | 0.107 |
| ABO blood type | 0.986‡‡ | 0.934‡‡ | |||||
| A | 45 (40.5) | 73 (38.8 | 42 (38.9) | 17 (36.2) | 14 (42.4) | ||
| AB | 2 (1.8) | 4 (2.1) | 1 (0.9) | 2 (4.3) | 1 (3.0) | ||
| B | 15 (13.5) | 25 (13.3) | 13 (12.0) | 7 (14.9) | 5 (15.2) | ||
| O | 49 (44.1) | 86 (45.7) | 52 (48.2) | 21 (44.7) | 13 (39.4) | ||
| ABO match level 1 | 109 (98.2) | 180 (95.7) | 0.332‡‡ | 102 (94.4) | 45 (95.7) | 33 (100) | 0.983‡‡ |
| Transplant diagnosis | 0.005‡‡ | 0.013 | |||||
| COPD | 45 (40.5) | 74 (39.4) | 46 (42.6) | 19 (40.4) | 9 (27.3) | ||
| ILD | 56 (50.5) | 67 (35.6 | 42 (38.9) | 15 (31.9) | 10 (30.3) | ||
| CF | 2 (1.8) | 24 (12.8) | 9 (8.3) | 9 (19.2) | 6 (18.2) | ||
| PAH | 3 (2.7) | 10 (5.3) | 4 (3.7) | 2 (4.3) | 4 (12.1) | ||
| Sarcoidosis | 3 (2.7) | 7 (3.7) | 4 (3.7) | 1 (2.1) | 2 (6.1) | ||
| Other | 2 (1.8) | 6 (3.2) | 3 (2.8) | 1 (2.1) | 2 (6.1) | ||
| mPAP (mm Hg) | 24 (20–28) | 26 (21–32) | 0.039 | 25 (20–32) | 24 (22–29) | 29 (23–35) | 0.011 |
| Preoperative meds | |||||||
| Corticosteroids | 48 (43.2) | 94 (50.0) | 0.258 | 52 (48.2) | 24 (51.1) | 18 (54.6) | 0.195 |
| ACE inhibitor | 18 (16.2) | 12 (6.4) | 0.006 | 5 (4.6) | 4 (8.5) | 3 (9.1) | 0.110 |
| Supplemental O2 | 97 (87.4) | 159 (84.6) | 0.503 | 95 (88.0) | 39 (83.0) | 25 (75.8) | 0.105 |
| Mechanical ventilation | 1 (0.9) | 5 (2.7) | 0.418 | 1 (0.9) | 0 (0) | 4 (12.1) | 0.004 |
| Donor and allograft | |||||||
| Age, years† | 37.6 (25.8–48.2) | 34.2 (22.5–48.0) | 0.433 | 36.0 (23.5–49.2) | 30.0 (20.8–45.5) | 34.2 (22.4–47.9) | 0.237 |
| Male sex | 63 (56.8) | 121 (64.4) | 0.192 | 74 (68.5) | 28 (59.6) | 19 (57.6) | 0.495 |
| Race‡ | 0.594 | 0.626‡‡ | |||||
| Caucasian | 72 (65.5) | 124 (66.7) | 70 (65.4) | 35 (76.1) | 19 (57.6) | ||
| African American | 27 (24.6) | 38 (20.4) | 23 (21.5) | 7 (15.2) | 8 (24.2) | ||
| Other | 11 (10.0) | 24 (12.9) | 14 (13.1) | 4 (8.7) | 6 (18.2) | ||
| Smoking history§ | 15 (15.6) | 28 (18.0) | 0.561 | 12 (13.6) | 9 (22.0) | 7 (26.9) | 0.146 |
| Ischemic time¶ | |||||||
| First lung | 233 (186–283) | 230 (198–275) | 0.619 | 224 (198–266) | 232 (194–275) | 245 (211–290) | 0.170 |
| Second lung | 315 (239–390) | 323 (280–367) | 0.581 | 327 (280–354) | 312 (265–361) | 341 (295–427) | 0.218 |
| Operative | |||||||
| Transplant type | 0.004 | <0.001 | |||||
| Unilateral | 56 (50.5) | 63 (33.5) | 44 (40.7) | 13 (27.7) | 6 (18.2) | ||
| Bilateral | 55 (49.6) | 125 (66.5) | 64 (59.3) | 34 (72.3) | 27 (81.8) | ||
| Cardiopulmonary bypass | |||||||
| Use | 54 (48.7) | 118 (62.8) | 0.017 | 61 (56.5) | 32 (68.1) | 25 (75.8) | 0.001 |
| Minutes†† | 245 (200–271) | 263 (231,281) | 0.010 | 263 (238–277) | 253 (198–271) | 277 (253–309) | 0.003 |
| RBC units transfused | 0 (0–2) | 1 (0–3) | 0.002 | 0 (0–2) | 1 (0–3) | 2 (0–3) | <0.001 |
| Lowest MAP (mm Hg) | 57 (50–64) | 58 (50–63) | 0.910 | 59 (50–63) | 60 (50–65) | 55 (49–60) | 0.235 |
Data are shown as n (%) for categorical variables, and median (interquartile range) for continuous variables. P values for covariate association with AKI from Wilcoxon rank-sum or χ2 tests, as appropriate, unless otherwise specified. P values for association with AKI stage (ordered none, 1, 2, 3) from Cuzick’s non-parametric test of trend unless otherwise specified. Definition of abbreviations: PGD=primary graft dysfunction, grade 3 at 48 or 72 hours post-reperfusion; BMI= body-mass index; Cr=creatinine; eGFR=estimated glomerular filtration rate according to CKD-EPI equation47; COPD=chronic obstructive pulmonary disease; ILD: interstitial lung disease; CF=cystic fibrosis; ACE=angiotensin-converting enzyme; mPAP=mean pulmonary arterial pressure measured at the time of transplant listing (n=276) or at beginning of transplant operation (n=23); RBC= red blood cells (intraoperative); MAP=mean arterial pressure (intraoperative); AKI=acute kidney injury.
Donor age missing in 14 subjects with AKI, 31 subjects without AKI.
Donor race missing in 3 cases.
Donor smoking history missing for 15 subjects with AKI, 33 subjects without AKI.
Ischemic times are missing for one subject who had AKI stage 1; ischemic times for second lung are for those who underwent bilateral lung transplantation.
Data listed only for those who were placed on cardiopulmonary bypass.
Fisher’s exact test used instead of χ2 due to low cell counts.
Table 2.
Multivariable logistic regression models of risk factors for acute kidney injury.
| Covariate | Model 1 (without PGD) | Model 2 (with PGD) | ||
|---|---|---|---|---|
| OR (95% CI) | p | OR (95% CI) | p | |
| Primary graft dysfunction | - | - | 3.76 (1.72–8.19) | 0.001 |
| Age (per 10 years) | 1.23 (0.82–1.84) | 0.316 | 1.26 (0.83–1.91) | 0.274 |
| Male sex | 1.53 (0.90–2.60) | 0.116 | 1.65 (0.96–2.85) | 0.070 |
| Body mass index (per 5 kg/m2) | 1.37 (1.03–1.82) | 0.031 | 1.29 (0.96–1.72) | 0.091 |
| Baseline eGFR<60 ml/min/1.73m2 | 1.95 (0.62–6.10) | 0.253 | 1.94 (0.59–6.41) | 0.278 |
| Transplant diagnosis | ||||
| COPD | Ref | Ref | ||
| ILD | 0.54 (0.30–0.97) | 0.041 | 0.51 (0.28–0.94) | 0.032 |
| CF | 12.82 (2.10–78.23) | 0.006 | 13.70 (2.20–85.49) | 0.005 |
| Other† | 1.27 (0.46–3.53) | 0.649 | 0.95 (0.33–2.78) | 0.927 |
| ACE inhibitor use | 0.27 (0.12–0.64) | 0.003 | 0.25 (0.10–0.62) | 0.003 |
| mPAP (per 10 mm Hg) | 1.20 (0.89–1.62) | 0.241 | 1.17 (0.85–1.61) | 0.332 |
| Bilateral transplant | 1.51 (0.83–2.75) | 0.172 | 1.44 (0.78–2.67) | 0.240 |
Other=pulmonary arterial hypertension, sarcoidosis, congenital heart disease, or bronchiolitis obliterans syndrome.
Definition of abbreviations: PGD=primary graft dysfunction, grade 3 at 48 or 72 hours post-reperfusion; eGFR=estimated glomerular filtration rate according to CKD-EPI equation47; COPD=chronic obstructive pulmonary disease; ILD: interstitial lung disease; CF=cystic fibrosis; ACE=angiotensin-converting enzyme; mPAP=mean pulmonary arterial pressure measured at the time of transplant listing (n=276) or at beginning of transplant operation (n=23); Ref=reference against which the other pulmonary diagnoses are compared.
PGD-AKI association and impact on AKI risk factors
Grade 3 PGD was present at 48 or 72 hours in 50/188 (26.6%) with AKI versus 10/111 (9.0%) without AKI (p<0.001, Table 1). The likelihood of PGD increased substantially with higher AKI stage (p<0.001 for trend, Table 1). In cases of coincident AKI and PGD (n=50), 35 (70%) met criteria for both within one day of each other (Figure 1B). Only 2 (4.0%) met AKI criteria prior to the first day of grade 3 PGD.
Despite the strong association of PGD and AKI, including PGD in the multivariable model of AKI risk factors (Table 2, Model 2) minimally changed the association of baseline characteristics with AKI. The BMI-AKI association was no longer statistically significant after adjustment for PGD, though the change in odds ratio point estimate was modest.
One hundred and one patients (33.8%) met grade 3 PGD criteria at any time during the first 72 hours. Similar to the primary PGD definition, grade 3 PGD at any time was strongly associated with AKI and AKI stage (p<0.001, Table 1). While a greater number of these PGD cases was incident on days 0 and 1 (Supplemental Figure 1), PGD and AKI were again coincident within 1 day of each other in a high percentage (63.3%) of cases in which both developed.
Long-term outcomes from AKI
Of 299 patients, 48 (16.1%) died within one year after transplant. AKI by postoperative day 14 and corresponding AKI stage were associated with significantly increased risk of one-year mortality (log-rank test p<0.001 for AKI, log-rank test for trend p<0.001 for AKI stage; Supplemental Figure 2), with rates of 6.7%, 15.0%, 14.9%, and 54.6% for no AKI, and AKI stages 1, 2, and 3, respectively. Increased mortality was evident within days to weeks after transplant among those with stage 3 AKI, whereas the separation of survival curves compared with no AKI was delayed for stage 2 (starting ~1 month) and stage 1 (starting ~3 months).
Adjusting for PGD and other confounders had a modest effect on the association of AKI with one-year mortality (Table 3, Figure 2A), and in a similar model using AKI stage (Figure 2B). The association of AKI stage with 90-day mortality was also robust to adjustment for PGD (OR 2.42 per AKI stage, 95% CI 1.56–3.76, p<0.001). One-year mortality hazard was highest among those with both AKI and PGD, though AKI without PGD was also significantly associated with mortality (Figure 3). The interaction of AKI and PGD for association with mortality was not statistically significant (likelihood-ratio test, p=0.514).
Table 3.
Multivariable Cox proportional hazards models showing the impact of acute kidney injury and primary graft dysfunction on the other’s association with one-year mortality.
| Covariate | Model 1 (without PGD) | Model 2 (without AKI) | Model 3 (PGD & AKI) | |||
|---|---|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | |
| Primary graft dysfunction | - | - | 3.11 (1.71–5.63) | <0.001 | 2.55 (1.40–4.64) | 0.002 |
| Acute kidney injury | 4.15 (1.93–8.94) | <0.001 | - | - | 3.64 (1.68–7.88) | 0.001 |
| Age (per 10 years) | 1.37 (0.93–2.02) | 0.116 | 1.34 (0.89–2.00) | 0.159 | 1.40 (0.93–2.11) | 0.107 |
| Bilateral transplant | 0.81 (0.42–1.56) | 0.535 | 0.83 (0.43–1.58) | 0.569 | 0.76 (0.40–1.45) | 0.403 |
Definition of abbreviations: PGD=primary graft dysfunction, grade 3 at 48 or 72 hours post-reperfusion; AKI=acute kidney injury through post-transplant day 14, which constituted 180/188 (95.7%) of all AKI cases (see methods for rationale).
Figure 2.
Adjusted survival probabilities over one year post-transplant are shown by acute kidney injury (AKI) developing by day 14 (A) and associated AKI stage (st) (B). Cox models were used to adjust hazard ratios (HR, shown with 95% confidence intervals) for primary graft dysfunction, recipient age, and transplant type (unilateral v. bilateral).
Figure 3.
Adjusted survival probabilities over one year post-transplant are shown by AKI-PGD strata. Using AKI developing by day 14 (see methods for rationale), strata included Neither (n=105), PGD only (n=14), AKI only (n=134), and AKI & PGD (n=46). Cox models were used to adjust hazard ratios (HR, shown with 95% confidence intervals) for recipient age and transplant type (unilateral v. bilateral).
AKI stage remained strongly associated with one-year mortality in the subgroup (n=60) with PGD (HR 2.37 per stage, 95% CI 1.39–4.04, p=0.002; Supplemental Figure 3). In survival analyses of the entire cohort, there was a significant interaction of PGD and time—PGD had a stronger association with mortality early after transplant with attenuation over time. Inclusion of this interaction in Cox models, however, minimally changed the associations of AKI, age, and transplant procedure with mortality. Cox models using a PGD definition of grade 3 at any time during the first 72 hours yielded similar findings to the primary one-year mortality analyses for both AKI (HR 3.66, 95% CI 1.69–7.93, p=0.001) and AKI stage (stage 1: HR 2.50, 95% CI 1.06–5.92, p=0.037; stage 2: 2.85, 1.00–8.09, p=0.049; stage 3: 14.64, 5.78–37.09, p<0.001).
Early versus delayed AKI
AKI through day 5 (early AKI, first peak of Figure 1A) accounted for 151 (83.9%) cases within the first 14 days, while delayed AKI (days 6–14, second peak of Figure 1A) occurred in only 29 (16.1%). All characteristics significantly associated with AKI in the primary analysis (Table 1) remained significantly associated with early AKI (Supplemental Table 1). While odds ratio point estimates suggested an attenuation and possible reversal of the association of PGD, bilateral transplant, and cardiopulmonary bypass with delayed AKI, analysis was limited by small numbers and these findings were not statistically significant. In multivariable Cox models (Supplemental Table 2), both early and delayed AKI had significant associations with one-year mortality. Adjustment for PGD modestly decreased the hazard ratio for early AKI but had no effect on the association of delayed AKI with mortality.
Discussion
We demonstrated a strong association of PGD with AKI and AKI stage. Despite this, adjusting for PGD did not have a substantial impact on the association of AKI with one-year mortality or on risk factors for AKI. The fact that AKI was associated with mortality even in the absence of PGD underscores the importance of AKI as an independent contributor to lung transplant outcomes, and raises the possibility that patients at high risk for AKI may derive clinical benefit from renal-protective strategies.
While prior single-center studies have detailed incidence, risk factors, and outcomes of AKI after lung transplantation,1–3,5,16 none has adequately accounted for potential confounding effects of PGD. Bennett et al did report higher grade 3 PGD rates in 135 lung transplant recipients with AKI (22% versus 13% in those without AKI), but this finding was not statistically significant.6 Our better-powered study showed a convincing association of PGD with AKI, though despite this link PGD did not have a significant impact on AKI risk factors. There are several plausible mechanisms by which PGD and AKI may be related. PGD is characterized by systemic immune dysregulation17–20 and its management often includes both prolonged mechanical ventilation and attempts to maintain a negative fluid balance, all of which may have harmful renal effects.21,22 Conversely, increasing serum creatinine or oliguria may prompt fluid challenges, which could precipitate allograft pulmonary edema and hypoxemia. Finally, early post-reperfusion systemic derangements in inflammation and endothelial activation may be a common mechanistic antecedent to both lung allograft and renal injury.23–28 Though AKI more commonly followed than preceded PGD, we do not specifically infer a causal effect of PGD on AKI from our study since 1) 70% of coincident cases occurred within one day of each other, and 2) SCr rise, which defines AKI, is a delayed marker of renal injury.29 It therefore remains unclear whether the systemic effects of PGD contribute to AKI or if both are downstream effects of other post-reperfusion systemic factors. The substantial number of AKI cases without PGD suggests that additional mechanisms may also contribute to renal injury.
We found that AKI was strongly associated with 90-day and one-year mortality independent of PGD. This finding expands on prior studies, which did not account for PGD, and indicates that AKI is not a mere PGD epiphenomenon. Early renal-protective strategies therefore may have the potential to substantially impact transplant outcomes. There is a natural tension, however, between kidney and allograft protection regarding post-transplant fluid management and calcineurin inhibitor initiation. Thus, developing robust AKI prediction tools to identify a sub-group at high risk, and that go beyond current models limited to prediction of RRT-requiring AKI only,30 may help to test the impact of renal-protective strategies on outcomes. It is also notable that post-lung transplant rates of stage 2–3 AKI and RRT in both our study and multiple others were 4–5 times those seen after non-transplant cardiac surgery despite younger age and lower prevalence of pre-operative chronic kidney disease.31–33 Whether these unexpectedly high rates reflect novel molecular mechanisms of renal injury is unclear, as few studies have explored the molecular epidemiology of AKI after lung transplantation.34 Further studies may identify AKI mechanisms that could be targeted without risk to the allograft and therefore applied more broadly.
In testing AKI risk factors for this study, we found several novel or rarely reported characteristics associated with post-lung transplant AKI. BMI has not previously been reported as a risk factor for AKI after lung transplantation, though we and others have noted this association in critical illness populations.35–38 Obesity is hypothesized to increase AKI risk via an underlying proinflammatory state, intraabdominal hypertension, and underestimated fluid requirements, though precise mechanisms remain unclear.39 ACE inhibitors, which have well-described long-term renal-protective effects,40 were associated with reduced AKI risk. We did not have detailed data on dosage, duration, and whether such medications were held prior to transplantation, limiting the ability to draw conclusions about possible mechanisms. CF has been noted as a post-lung transplant AKI risk factor in only one other study to date.41 Patients with CF frequently use inhaled or intravenous aminoglycoside and other nephrotoxic medications that may prime their kidneys for further injury.42–44 Further study of these relatively novel risk factors in larger cohorts may aid in developing post-transplant AKI risk stratification tools, and exploring mechanisms underlying their associations with AKI is warranted. In addition, such larger studies may aid in better defining risk factors for and outcomes from early versus delayed AKI. Differences in risk factors, combined with testing for molecular markers of renal endothelial or tubular injury, may point toward subgroup-specific nephrotoxic mechanisms and inform tailored renal-protective strategies.
Our study was limited by single-center design, limiting generalizability, and relatively small sample size, which precluded adjustment of mortality analyses beyond pre-specified confounders, prevented robust analyses of subgroups such as those with delayed AKI, and limited our ability to test potential interactions between AKI risk factors such as pre-transplant diagnosis, transplant procedure, and hemodynamic factors. To our knowledge, however, this is the first study to describe the PGD-AKI relationship in depth and fully account for the effects of PGD on AKI risk factors and outcomes. In addition, we utilized only those transplanted in the LAS era to reflect current practice, unlike the largest prior studies which included patients transplanted as remotely as 1990.1–3 Our study, like prior ones in lung transplant populations, also lacked urine output data. Prospective studies including urine output data may be useful for early AKI detection and describing AKI endotypes.45,46 We also did not have detailed data on post-transplant fluid balance and other time-varying factors that may contribute to AKI risk, particularly for the smaller subset with delayed AKI. Larger, prospective multicenter studies may help to address some of these limitations and account for the effects of evolving practices such as bridging to transplant using mechanical ventilation and extracorporeal life support.15
In conclusion, we demonstrated a strong association of PGD with AKI after lung transplantation. AKI risk factors and mortality, however, were largely independent of PGD, suggesting that reducing AKI rates may have the potential to improve clinically relevant transplant outcomes. Additional studies are warranted to understand mechanisms underlying the high rates of post-lung transplant AKI, more fully define the impact of mild and delayed AKI, and identify high-risk individuals in whom to test renal-protective strategies.
Supplementary Material
Acknowledgments:
MGSS supported by K23DK097307, R01DK111638, and the University of Pennsylvania University Research Foundation. EC supported by K23HL116656. JMD supported by K23HL121406. JDC supported by R01HL087115 and K24HL115354.
Footnotes
Conflict of interest statement:
The authors have no potential sources of conflict of interest that are directly relevant or related to the work described in the manuscript.
Data accessibility statement:
The datasets generated in the conduct of this study may be made available on reasonable request. Requests may be made to the corresponding author, but are subject to the review and approval of the Lung Transplant Outcomes Group Ancillary Studies Committee.
References
- 1.Fidalgo P, Ahmed M, Meyer SR, et al. Incidence and outcomes of acute kidney injury following orthotopic lung transplantation: a population-based cohort study. Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association. September 2014;29(9):1702–1709. [DOI] [PubMed] [Google Scholar]
- 2.Wehbe E, Brock R, Budev M, et al. Short-term and long-term outcomes of acute kidney injury after lung transplantation. The Journal of Heart and Lung Transplantation. 3// 2012;31(3):244–251. [DOI] [PubMed] [Google Scholar]
- 3.Rocha PN, Rocha AT, Palmer SM, Davis RD, Smith SR. Acute renal failure after lung transplantation: incidence, predictors and impact on perioperative morbidity and mortality. Am J Transplant. June 2005;5(6):1469–1476. [DOI] [PubMed] [Google Scholar]
- 4.Diamond JM, Lee JC, Kawut SM, et al. Clinical risk factors for primary graft dysfunction after lung transplantation. Am J Respir Crit Care Med. March 1 2013;187(5):527–534. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wehbe E, Duncan AE, Dar G, Budev M, Stephany B. Recovery from AKI and short- and long-term outcomes after lung transplantation. Clin J Am Soc Nephrol. January 2013;8(1):19–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bennett D, Fossi A, Marchetti L, et al. Postoperative acute kidney injury in lung transplant recipients. Interactive CardioVascular and Thoracic Surgery. 2019;28(6):929–935. [DOI] [PubMed] [Google Scholar]
- 7.Egan TM, Murray S, Bustami RT, et al. Development of the new lung allocation system in the United States. Am J Transplant. 2006;6(5 Pt 2):1212–1227. [DOI] [PubMed] [Google Scholar]
- 8.Kellum JA, Lameire N, Aspelin P, et al. Kidney disease: Improving global outcomes (KDIGO) acute kidney injury work group. KDIGO clinical practice guideline for acute kidney injury. Kidney International Supplements. 2012;2(1):1–138. [Google Scholar]
- 9.Christie JD, Carby M, Bag R, Corris P, Hertz M, Weill D. Report of the ISHLT Working Group on Primary Lung Graft Dysfunction part II: definition. A consensus statement of the International Society for Heart and Lung Transplantation. The Journal of heart and lung transplantation : the official publication of the International Society for Heart Transplantation. October 2005;24(10):1454–1459. [DOI] [PubMed] [Google Scholar]
- 10.Christie JD, Bellamy S, Ware LB, et al. Construct validity of the definition of primary graft dysfunction after lung transplantation. The Journal of heart and lung transplantation : the official publication of the International Society for Heart Transplantation. November 2010;29(11):1231–1239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cuzick J A wilcoxon-type test for trend. Statistics in Medicine. 1985;4(4):543–547. [DOI] [PubMed] [Google Scholar]
- 12.Cox DR. Regression Models and Life-Tables. Journal of the Royal Statistical Society. Series B (Methodological). 1972;34(2):187–220. [Google Scholar]
- 13.Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol. December 1996;49(12):1373–1379. [DOI] [PubMed] [Google Scholar]
- 14.Christie JD, Sager JS, Kimmel SE, et al. IMpact of primary graft failure on outcomes following lung transplantation*. Chest. 2005;127(1):161–165. [DOI] [PubMed] [Google Scholar]
- 15.Yusen RD, Edwards LB, Kucheryavaya AY, et al. The Registry of the International Society for Heart and Lung Transplantation: Thirty-second Official Adult Lung and Heart-Lung Transplantation Report—2015; Focus Theme: Early Graft Failure. The Journal of Heart and Lung Transplantation.34(10):1264–1277. [DOI] [PubMed] [Google Scholar]
- 16.Fidalgo P, Ahmed M, Meyer SR, et al. Association between transient acute kidney injury and morbidity and mortality after lung transplantation: a retrospective cohort study. J Crit Care. Dec 2014;29(6):1028–1034. [DOI] [PubMed] [Google Scholar]
- 17.Diamond JM, Lederer DJ, Kawut SM, et al. Elevated plasma long pentraxin-3 levels and primary graft dysfunction after lung transplantation for idiopathic pulmonary fibrosis. Am J Transplant. Nov 2011;11(11):2517–2522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Diamond JM, Meyer NJ, Feng R, et al. Variation in PTX3 is associated with primary graft dysfunction after lung transplantation. Am J Respir Crit Care Med. September 15 2012;186(6):546–552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Hoffman SA, Wang L, Shah CV, et al. Plasma cytokines and chemokines in primary graft dysfunction post-lung transplantation. Am J Transplant. Feb 2009;9(2):389–396. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shah RJ, Emtiazjoo AM, Diamond JM, et al. Plasma complement levels are associated with primary graft dysfunction and mortality after lung transplantation. Am J Respir Crit Care Med. June 15 2014;189(12):1564–1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Koyner JL, Murray PT. Mechanical Ventilation and the Kidney. Blood Purification. 11/19 2010;29(1):52–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.van den Akker JPC, Egal M, Groeneveld JAB. Invasive mechanical ventilation as a risk factor for acute kidney injury in the critically ill: a systematic review and meta-analysis. Critical Care. 05/27 2013;17(3):R98–R98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Christie JD, Robinson N, Ware LB, et al. Association of protein C and type 1 plasminogen activator inhibitor with primary graft dysfunction. Am J Respir Crit Care Med. January 1 2007;175(1):69–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bouchard J, Malhotra R, Shah S, et al. Levels of Protein C and Soluble Thrombomodulin in Critically Ill Patients with Acute Kidney Injury: A Multicenter Prospective Observational Study. PLoS ONE. 2015;10(3):e0120770. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Covarrubias M, Ware LB, Kawut SM, et al. Plasma intercellular adhesion molecule-1 and von Willebrand factor in primary graft dysfunction after lung transplantation. Am J Transplant. November 2007;7(11):2573–2578. [DOI] [PubMed] [Google Scholar]
- 26.Kelly KJ, Williams WW Jr., Colvin RB, Bonventre JV. Antibody to intercellular adhesion molecule 1 protects the kidney against ischemic injury. Proc Natl Acad Sci U S A. January 18 1994;91(2):812–816. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Shah RJ, Diamond JM, Lederer DJ, et al. Plasma monocyte chemotactic protein-1 levels at 24 hours are a biomarker of primary graft dysfunction after lung transplantation. Translational research : the journal of laboratory and clinical medicine. December 2012;160(6):435–442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Munshi R, Johnson A, Siew ED, et al. MCP-1 gene activation marks acute kidney injury. Journal of the American Society of Nephrology : JASN. January 2011;22(1):165–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cruz DN, Mehta RL. Acute kidney injury in 2013: Breaking barriers for biomarkers in AKI[mdash]progress at last. Nature reviews. Nephrology. 02//print 2014;10(2):74–76. [DOI] [PubMed] [Google Scholar]
- 30.Grimm JC, Lui C, Kilic A, et al. A risk score to predict acute renal failure in adult patients after lung transplantation. The Annals of thoracic surgery. January 2015;99(1):251–257. [DOI] [PubMed] [Google Scholar]
- 31.Thakar CV, Christianson A, Freyberg R, Almenoff P, Render ML. Incidence and outcomes of acute kidney injury in intensive care units: A Veterans Administration study. Crit Care Med. 2009;37. [DOI] [PubMed] [Google Scholar]
- 32.Thakar CV, Arrigain S, Worley S, Yared J-P, Paganini EP. A Clinical Score to Predict Acute Renal Failure after Cardiac Surgery. Journal of the American Society of Nephrology. January 1, 2005. 2005;16(1):162–168. [DOI] [PubMed] [Google Scholar]
- 33.Parikh CR, Coca SG, Thiessen-Philbrook H, et al. Postoperative biomarkers predict acute kidney injury and poor outcomes after adult cardiac surgery. Journal of the American Society of Nephrology : JASN. September 2011;22(9):1748–1757. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Forker CM, Miano TA, Reilly JP, et al. Post-reperfusion plasma endothelial activation markers are associated with acute kidney injury after lung transplantation. Am J Transplant. April 24 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Shashaty MG, Meyer NJ, Localio AR, et al. African American race, obesity, and blood product transfusion are risk factors for acute kidney injury in critically ill trauma patients. J Crit Care. October 2012;27(5):496–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Shashaty MG, Kalkan E, Bellamy SL, et al. Computed tomography-defined abdominal adiposity is associated with acute kidney injury in critically ill trauma patients*. Crit Care Med. July 2014;42(7):1619–1628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Soto GJ, Frank AJ, Christiani DC, Gong MN. Body mass index and acute kidney injury in the acute respiratory distress syndrome. Critical Care Medicine. 2012;40(9):2601–2608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Druml W, Metnitz B, Schaden E, Bauer P, Metnitz PGH. Impact of body mass on incidence and prognosis of acute kidney injury requiring renal replacement therapy. Intensive Care Medicine. 2010;36(7):1221–1228. [DOI] [PubMed] [Google Scholar]
- 39.Winfield RD, Delano MJ, Lottenberg L, et al. Traditional resuscitative practices fail to resolve metabolic acidosis in morbidly obese patients after severe blunt trauma. Journal of Trauma - Injury, Infection and Critical Care. 2010;68(2):317–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Taal MW, Brenner BM. Renoprotective benefits of RAS inhibition: from ACEI to angiotensin II antagonists. Kidney international. 2000;57(5):1803–1817. [DOI] [PubMed] [Google Scholar]
- 41.Sikma MA, Hunault CC, van de Graaf EA, et al. High tacrolimus blood concentrations early after lung transplantation and the risk of kidney injury. European Journal of Clinical Pharmacology. 2017;73(5):573–580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ramsey BW, Pepe MS, Quan JM, et al. Intermittent administration of inhaled tobramycin in patients with cystic fibrosis. Cystic Fibrosis Inhaled Tobramycin Study Group. The New England journal of medicine. January 7 1999;340(1):23–30. [DOI] [PubMed] [Google Scholar]
- 43.Flume PA, Mogayzel PJ Jr., Robinson KA, et al. Cystic fibrosis pulmonary guidelines: treatment of pulmonary exacerbations. Am J Respir Crit Care Med. November 1 2009;180(9):802–808. [DOI] [PubMed] [Google Scholar]
- 44.Flume PA, O’Sullivan BP, Robinson KA, et al. Cystic fibrosis pulmonary guidelines: chronic medications for maintenance of lung health. Am J Respir Crit Care Med. November 15 2007;176(10):957–969. [DOI] [PubMed] [Google Scholar]
- 45.Macedo E, Malhotra R, Bouchard J, Wynn SK, Mehta RL. Oliguria is an early predictor of higher mortality in critically ill patients. Kidney Int. 2011;80. [DOI] [PubMed] [Google Scholar]
- 46.Kellum JA, Sileanu FE, Murugan R, Lucko N, Shaw AD, Clermont G. Classifying AKI by Urine Output versus Serum Creatinine Level. Journal of the American Society of Nephrology. January 7, 2015. 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Annals of Internal Medicine. 2009;150(9):604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.