Incidence, Outcomes, and Predictors of Subphenotypes of Acute Kidney Injury among Acute Respiratory Distress Syndrome Patients: A Prospective Observational Study

INTRODUCTION

Extrapulmonary organ dysfunction is present in >80% patients of with acute respiratory distress syndrome (ARDS) even at the time of onset.¹ The severity of extrapulmonary organ dysfunction correlates with the ARDS severity.¹ Acute kidney injury (AKI) is the most common extrapulmonary organ dysfunction in ARDS patients in up to 50% of subjects.² The fact that AKI is an independent risk factor of mortality in ARDS is known.^3,4 However, the mere diagnosis of AKI with ARDS may not portend a worse outcome. AKI is just a syndromic diagnosis or broader phenotype, with the specific subphenotypes within it having different outcomes.⁵ Subphenotyping aids in prognostic enrichment (different outcomes based on a particular subphenotype) and predictive enrichment (different responses to treatment subphenotype).⁵ There are different ways of subphenotyping in AKI, like clinical subphenotypes based on serum creatinine (SCr) trajectory or biomarker-based subphenotypes.^5–8 Till date, the outcomes of patients with ARDS along with AKI based on the different subphenotypes of AKI have not been studied. Secondly, there is also a paucity of data regarding the early identification of AKI subphenotypes in a particular ARDS patient within 48 hours of IMV, rather than over 5 days, with the help of SCr trajectories.^9,10 In a recent study, the DRONE score [Driving pressure (>15 cm H₂O), Oxygenation (ratio of partial pressure of oxygen in arterial blood to the fraction of inspired oxygen (PaO₂/FiO₂) ratio <208 mmHg, and Nutritional Evaluation (modified Nutritional Risk in Critically ill score ≥4)] has been proven to be an independent predictor of mortality in ARDS patients.¹¹

Since the DRONE score included factors such as respiratory system compliance (driving pressure), oxygenation (PaO₂/FiO₂ ratio), and multiorgan dysfunction variables (the Acute Physiology and Chronic Health Evaluation [APACHE II] score and the Sequential Organ Failure Assessment [SOFA] score in the calculation of modified NUTRIC (mNUTRIC) score), it may serve as an early indicator of subphenotype of AKI in ARDS patients on invasive mechanical ventilation (IMV).

MATERIALS AND METHODS

It was a single-center prospective observational study. A total of 266 patients of ARDS were enrolled in the study. After obtaining the Institutional Ethical Committee clearance (IEC 765/2019), the study was registered in Clinical Trial Registry of India (CTRI)- CTRI/2020/04/024940. The study was conducted from September 2020 to February 2023. Written informed consent was obtained from legally authorized representatives prior to recruitment for the study.

RESULTS

The demographic data and variables of interest in the study is depicted in Table 1. Among the ARDS patients, 141 (64.1%) had AKI, and subphenotyping was possible in 123 patients. The overall incidence of AKI subphenotype 1 among the ARDS patients was 45/222 (20.27%), whereas that of AKI subphenotype 2 was 78/222 (35.13%) (Table 1). Among those with subphenotype classification, 63.4% belonged to the non-resolving (subphenotype 2) of AKI. (Table 1). Mean ± SD or median (IQR) of the physiological variables among the ARDS patients is depicted (Table 2).

A comparison of the survivors and non-survivors of ARDS revealed that the APACHE II score, SOFA score, DRONE score, and SCr levels on day 3 and day 5 of IMV were significantly higher (Table 3). Out of the 123 patients with AKI-subphenotyping done, there were 65 non-survivors (52.84%). The Table 4 shows that among these 65 non-survivors, 87.7% belonged to AKI subphenotype 2, whereas only 12.3% belonged to AKI subphenotype 1 (p < 0.001 Chi-square test, Phi and Cramer’s V strength of association 0.534, depicting moderately strong association between mortality and AKI subphenotype 2).

The AKI subphenotype 1 had a significantly higher median SCr level on day 1 of IMV as compared with the subphenotype 2 (2.71 mg/dL vs 2.15 mg/dL), whereas it was reversed in day 3 (2.04 vs 3.40 mg/dL) and 5 (1.97 vs 3.92mg/dL) of IMV (Table 5). The median SCr in AKI subphenotype 1 was higher on day 1 as compared with the SCr in AKI subphenotype 2, with a decrease on day 3 and day 5 of IMV (Figs 1 and 2). A persistent decrease in SCr in AKI subphenotype 1 is seen as compared with the persistent rise in AKI subphenotype 2 (Fig. 2).

Univariate and multivariable logistic regression was done to predict non-survivors of ARDS, and the AKI subphenotype 2 was an independent predictor (adjusted OR 8.978, 95% CI [2.790–28.897], p < 0.0010) as shown in Table 6. However, when the same variables were selected for the regression analysis and AKI subphenotype 2 was replaced with the mere presence of AKI only on either day 1,3, or 5 of IMV, then, the mere presence of AKI on days 1,3 or 5 was not an independent predictor of mortality among the ARDS patients (Table 7). Bootstrap multivariable logistic regression with 1000 samples for internal validation showed that the AKI subphenotype 2 was an independent predictor of mortality among ARDS patients (p = 0.002. 95% CI [1.147–5.364]).

Regarding the association of a high DRONE score (≥4) and AKI subphenotype, 71.1% of the patients with AKI subphenotype 1 had a low DRONE score < 4, whereas 68% of the patients with AKI subphenotype 2 had a high DRONE score (≥ 4) (p < 0.001, Chi-square test) (Table 8). The Phi and Cramer V value was 0.377 (comparatively low strength of association). The ROC plotted for the DRONE score to predict AKI subphenotype 2 depicted that the DRONE score was a reliable predictor of AKI subphenotype 2 among ARDS patients (AUC 0.706, p-value <0.001, 95% CI [0.613–0.800], cut-off score ≥ 4, 69% sensitivity, 68% specificity) (Fig. 3). The relationship between the DRONE score and AKI subphenotypes 1 and 2 are depicted which shows that there is a strong association between the AKI subphenotype 1 and the lower DRONE scores (score 1–3), and the AKI subphenotype 2 is strongly associated with higher DRONE scores (score 5–9) (Fig. 4).

Survival analysis with the Kaplan–Meier survival plot displayed a significant difference in survival between the ARDS patients with subphenotype 1 of AKI and AKI subphenotype 2 [Log-Rank (Mantel–Cox) p-value < 0.001] (Fig. 5). The median time of survival was 8 days in ARDS patients having AKI with subphenotype 2 vs 45 days in ARDS patients having AKI with subphenotype 1. There was a significant difference between the requirement for RRT among patients of AKI subphenotype 1 (35.6%) and subphenotype 2 (57.7%) (p = 0.018, Pearson Chi-square test).

DISCUSSION

We found that among the ARDS patients with AKI, a majority (63.4%) had AKI subphenotype 2. In ARDS patients, the influence of positive pressure ventilation and reduced renal perfusion, reduced glomerular filtration rate, and alterations in the neurohormonal axis in the kidney are present.^14,15 About 50% of the patients with ARDS develop AKI, and out of those, 90% of them develop AKI within 48 hours of IMV.^2,16 Respiratory compliance, PEEP, and driving pressure are all factors associated with severe AKI.^17,18 However, AKI may also be due to factors such as hypotension, hypovolemia, persistent shock, and hospital-acquired causes.⁵

The AKI due to these various causes has different outcomes and responses to treatment.⁵ Thus, subphenotypes of AKI are essential to understand their role as mortality predictors.⁵

Just like AKI may be due to various factors, and can be classified into subphenotypes, the AKI in patients with ARDS can lead to varying complications which may or may not be responsive to renal replacement therapy (RRT).¹⁴ The “traditional complications” of AKI are electrolyte derangements, uremia, and fluid overload, with poor oxygenation.¹⁴ These complications are amenable to RRT.¹⁴ However, the “non-traditional” complications of AKI are often not correctable by RRT and lead to a mortality of up to 60%.^15,19,20 These “non-traditional” complications include inflammatory lung injury, cardiac dysfunction, and immune paralysis.¹⁴

We found that even though the incidence of AKI was significantly higher in the non-survivors of ARDS, the mere presence of AKI was not an independent predictor of mortality. Rather, the non-resolving (subphenotype 2) of AKI is an independent predictor of mortality after regression analysis, which was validated even after the bootstrapping method. We found that even though patients with resolving AKI (subphenotype 1) had significantly higher SCr levels on day 1 (median 2.71 mg/dL) as compared with AKI subphenotype 2 (median 2.15 mg/dL), eventually by day 3 and day 5 of IMV, the SCr levels were significantly lower in the AKI subphenotype 1. This supports the fact that single-point classification of patients into AKI or non-AKI categories in ARDS patients based on the severity may not predict outcomes, rather subphenotyping based on serum creatinine trajectories is the reliable predictor. Our results are similar to that of Bhatraju et al. where the authors found that the creatinine trajectory classifies AKI subphenotypes with different risks for mortality, even within AKI cases of similar severity.⁶ However, in contrast to the aforementioned study where the creatinine trajectory was followed for 72 hours, we followed the trajectory for up to 5 days.⁶ We also found that the need for RRT was significantly different among the AKI patients in the two subphenotypes. (RRT in 36% of patients of AKI subphenotype 1 vs 58% of patients in AKI subphenotype 2 (Pearson Chi-square p-value = 0.018). However, the mortality was higher in AKI subphenotype 2 though more underwent RRT. This shows that AKI patients with subphenotype 2 showed poorer response to therapy or they have a difference in predictive enrichment as compared with patients with subphenotype 1. However, we did not investigate whether this differential response to RRT could possibly be due to the aforementioned “non-traditional” complications of AKI on lung and heart in patients with subphenotype 2.¹⁴

We found that the novel DRONE score (comprising of DP, PaO₂/FiO₂ ratio, and the mNUTRIC score) ≥ 4 predicted the AKI subphenotype 2 in ARDS. Our results may be due to three facts, also corroborated in recent literature. Firstly, DP has been shown to be the best ventilator parameter for predicting mortality in ARDS, and higher DP has also been associated with AKI.^21,22 Secondly, in a recent study, patients with non-recovery of AKI were associated with severe ARDS, which explains that the DRONE score encompassing poorer oxygenation can predict AKI subphenotype 2.²³

Thirdly, poor nutritional status worsens outcomes of AKI in critically ill, which is again incorporated in the DRONE score.²⁴

Apart from these, literature shows that shock and comorbid illnesses like hypertension and malignancy are associated with non-recovery of AKI in ARDS patients.²² It is notable, that co-morbid illnesses are part of the calculation of DRONE score as well since it incorporates APACHE II and mNUTRIC score. Likewise, the presence of shock is incorporated in the SOFA score used for calculation of the DRONE score.¹¹

There were certain strengths of the study. It proved that ARDS patients with AKI having initial higher creatinine levels need not necessarily have a poorer outcome. Outcomes are based on the recovery versus non-recovery subphenotypes from the initial AKI. We used a creatinine trajectory (day 1–day 5)-based AKI subphenotype, which can be used in resource-limited settings as well. It is the first study of its kind that could devise an objective score that can be used within 48 hours of IMV in ARDS patients with AKI to predict possible subphenotype outcomes on day 5 of IMV. However, there were certain limitations, being a single-center study, with small sample size. Also, apart from procalcitonin, no other renal or inflammatory biomarkers were evaluated as to its possible role in AKI subphenotype 2. We did not determine the cause of AKI based on pre-renal (hypovolemic shock), intrinsic renal (nephrotoxic medications) or post-renal causes.

CONCLUSION

The incidence of subphenotype 2 (non-resolving) AKI among ARDS patients on IMV was about 35% as compared with about 20% of subphenotype 1 (resolving) AKI. The subphenotype 2 AKI was an independent predictor of mortality. The DRONE score ≥ 4 within 48 hours of IMV in ARDS patients can predict the subphenotype 2 (non-resolving) AKI. The findings need to be validated in future.

DATA AVAILABILITY

The data will be provided by the first or corresponding author upon email request. This is due to the reason of patient data confidentiality.

ORCID

Pratibha Todur https://orcid.org/0000-0003-0967-2252

Anitha Nileshwar https://orcid.org/0000-0002-4556-1851

Souvik Chaudhuri https://orcid.org/0000-0001-8392-2366

Thejesh Srinivas https://orcid.org/0000-0002-7354-1030

REFERENCES

1. Kallet RH, Lipnick MS, Zhuo H, Pangilinan LP, Gomez A. Characteristics of nonpulmonary organ dysfunction at onset of ARDS based on the Berlin definition. Respir Care 2019;64(5):493–501. DOI: 10.4187/respcare.06165.

2. Panitchote A, Mehkri O, Hastings A, Hanane T, Demirjian S, Torbic H, et al. Factors associated with acute kidney injury in acute respiratory distress syndrome. Ann Intensive Care 2019;9(1):74. DOI: 10.1186/s13613-019-0552-5. Erratum in: Ann Intensive Care 2019;9(1):84.

3. Liu KD, Glidden DV, Eisner MD, Parsons PE, Ware LB, Wheeler A, et al. Predictive and pathogenetic value of plasma biomarkers for acute kidney injury in patients with acute lung injury. Crit Care Med 2007;35(12):2755–2761. PMID: 18074478.

4. Clemens MS, Stewart IJ, Sosnov JA, Howard JT, Belenkiy SM, Sine CR, et al. Reciprocal risk of acute kidney injury and acute respiratory distress syndrome in critically ill burn patients. Crit Care Med 2016;44(10):e915–e922. DOI: 10.1097/CCM.0000000000001812.

5. Vaara ST, Bhatraju PK, Stanski NL, McMahon BA, Liu K, Joannidis M, et al. Sub-phenotypes in acute kidney injury: a narrative review. Crit Care 2022;26(1):251. DOI: 10.1186/s13054-022-04121-x.

6. Bhatraju PK, Mukherjee P, Robinson-Cohen C, O’Keefe GE, Frank AJ, Christie JD, et al. Acute kidney injury sub-phenotypes based on creatinine trajectory identifies patients at increased risk of death. Crit Care 2016;20(1):372. DOI: 10.1186/s13054-016-1546-4.

7. Ostermann M, Zarbock A, Goldstein S, Kashani K, Macedo E, Murugan R, et al. Recommendations on acute kidney injury biomarkers from the acute disease quality initiative consensus conference: A consensus statement. JAMA Netw Open 2020;3(10):e2019209. DOI: 10.1001/jamanetworkopen.2020.19209. Erratum in: JAMA Netw Open 2020;3(11):e2029182.

8. Smith TD, Soriano VO, Neyra JA, Chen J. Identifying KDIGO trajectory phenotypes associated with increased inpatient mortality. IEEE Int Conf Healthc Inform 2019;2019:10.1109/ichi.2019.8904739. DOI: 10.1109/ichi.2019.8904739.

9. Saxena A, Meshram S V. Predictors of mortality in acute kidney injury patients admitted to medicine intensive care unit in a rural tertiary care hospital. Indian J Crit Care Med 2018;22(4):231–237. DOI: 10.4103/ijccm.IJCCM_462_17.

10. Kellum JA, Sileanu FE, Bihorac A, Hoste EA, Chawla LS. Recovery after acute kidney injury. Am J Respir Crit Care Med 2017;195(6):784–791. DOI: 10.1164/rccm.201604-0799OC.

11. Todur P, Nileshwar A, Chaudhuri S, Shwethapriya R, Shanbhag V, Tatineni S. Development and internal validation of a novel prognostic score to predict mortality in acute respiratory distress syndrome – driving pressure, oxygenation and nutritional evaluation – “DRONE Score”. J Emerg Trauma and Shock 2023.DOI: 10.4103/jets.jets_12_23.

12. Ata Ur-Rehman HM, Ishtiaq W, Yousaf M, Bano S, Mujahid AM, Akhtar A. Modified nutrition risk in critically ill (mNUTRIC) score to assess nutritional risk in mechanically ventilated patients: A prospective observational study from the Pakistani Population. Cureus 2018;10(12):e3786. DOI: 10.7759/cureus.3786.

13. Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract 2012;120(4):c179–c184. DOI: 10.1159/000339789.

14. Park BD, Faubel S. Acute kidney injury and acute respiratory distress syndrome. Crit Care Clin 2021;37(4):835–849. DOI: 10.1016/j.ccc.2021.05.007.

15. Andrivet P, Adnot S, Sanker S, Chabrier PE, Macquin-Mavier I, Braquet P, et al. Hormonal interactions and renal function during mechanical ventilation and ANF infusion in humans. J Appl Physiol (1985) 1991;70(1):287–292. DOI: 10.1152/jappl.1991.70.1.287.

16. McNicholas BA, Rezoagli E, Simpkin AJ, Khanna S, Suen JY, Yeung P, et al.CCCC Consortium. Epidemiology and outcomes of early-onset AKI in COVID-19-related ARDS in comparison with non-COVID-19-related ARDS: Insights from two prospective global cohort studies. Crit Care 2023;27(1):3. DOI: 10.1186/s13054-022-04294-5. Erratum in: Crit Care 2023;27(1):202.

17. Vashisht R, Duggal A. Acute kidney injury in acute respiratory distress syndrome: Why ventilator settings matter. Ann Transl Med 2020;8(9):573. DOI: 10.21037/atm-20-2163.

18. Leite TT, Gomes CAM, Valdivia JMC, Libório AB. Respiratory parameters and acute kidney injury in acute respiratory distress syndrome: A causal inference study. Ann Transl Med 2019;7(23):742. DOI: 10.21037/atm.2019.11.92.

19. Metnitz PG, Krenn CG, Steltzer H, Lang T, Ploder J, Lenz K, et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30(9):2051–2058. DOI: 10.1097/00003246-200209000-00016.

20. Barbar SD, Clere-Jehl R, Bourredjem A, Hernu R, Montini F, Bruyere R, et al. Timing of renal-replacement therapy in patients with acute kidney injury and sepsis. N Engl J Med 2018;379(15):1431–1442. DOI: 10.1056/NEJMoa1803213.

21. Amato MB, Meade MO, Slutsky AS, Brochard L, Costa ELV, Schoenfeld DA, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med 2015; 372:747–755. DOI: 10.1056/NEJMsa1410639.

22. Vemuri SV, Rolfsen ML, Sykes AV, Takiar PG, Leonard AJ, Malhotra A, et al. Association between acute kidney injury during invasive mechanical ventilation and ICU outcomes and respiratory system mechanics. Crit Care Explor 2022;4(7):e0720. DOI: 10.1097/CCE.0000000000000720.

23. Panitchote A, Mehkri O, Hastings A, Hanane T, Demirjian S, Torbic H, et al. Clinical predictors of renal non-recovery in acute respiratory distress syndrome. BMC Nephrol 2019;20(1):255. DOI: 10.1186/s12882-019-1439-2. Erratum in: BMC Nephrol 2019;20(1):286.

24. Ramakrishnan N, Shankar B. Nutrition support in critically ill patients with AKI. Indian J Crit Care Med 2020;24(Suppl 3):S135-S139. DOI: 10.5005/jp-journals-10071-23397.