Integrated biomarkers and renal Doppler for early acute kidney injury diagnosis in hepatitis C virus cirrhosis

Table 1 Baseline characteristics and liver function by acute kidney injury status, n (%)/mean ± SD

Variable	Non-AKI (n = 42)	Pre-renal AKI (n = 43)	Hepatorenal syndrome (n = 20)	Acute tubular necrosis (n = 15)	P value
Age (years)	52.4 ± 11.0	54.1 ± 10.8	56.6 ± 9.7	55.1 ± 10.4	0.12
Male sex	24 (57.1)	31 (72.1)	14 (70.0)	11 (73.3)	0.34
Model for End-Stage Liver Disease includes serum sodium score	16.8 ± 4.2	20.3 ± 5.1b	24.2 ± 5.4	23.7 ± 5.3	< 0.001
Serum creatinine (mg/dL)	0.9 ± 0.3	1.8 ± 0.4a	2.0 ± 0.6	2.2 ± 0.7	< 0.001
Fractional excretion of sodium (%)	0.08 ± 0.1	0.05 ± 0.1a	0.11 ± 0.1c	0.26 ± 0.2b,c	< 0.001
Total bilirubin (mg/dL)	2.3 ± 1.2	3.8 ± 1.9a	5.7 ± 2.0	5.3 ± 2.2	< 0.001
Albumin (g/dL)	3.2 ± 0.5	2.9 ± 0.4a	2.6 ± 0.3	2.4 ± 0.4	< 0.001
International normalised ratio	1.3 ± 0.2	1.5 ± 0.3a	1.9 ± 0.4	1.8 ± 0.3	< 0.001
Alanine aminotransferase (U/L)	32 ± 14	36 ± 16	40 ± 17	39 ± 15	0.09
Aspartate aminotransferase (U/L)	45 ± 18	49 ± 20	61 ± 22a	58 ± 21a	0.03
AKI onset	-				0.55
Prevalent (at admission)	-	30 (69.8)	12 (60.0)	10 (66.7)
Incident (post-admission)	-	13 (30.2)	8 (40.0)	5 (33.3)

^aP < 0.05 vs non-acute kidney injury.

^bP < 0.01 vs pre-renal acute kidney injury.

^cP < 0.01 vs hepatorenal syndrome using analysis of variance and Tukey’s correction.

AKI: Acute kidney injury.

Table 2 Biomarker accuracy for acute kidney injury detection

Marker	Area under the curve (95%CI)	Cutoff	Sensitivity (%) (95%CI)	Specificity (%) (95%CI)
Neutrophil gelatinase-associated lipocalin	0.91 (0.87-0.95)	> 150 ng/mL	90 (85-94)	85 (78-90)
Cystatin C	0.86 (0.81-0.91)	> 1.5 mg/L	84 (77-89)	80 (72-86)
Renal resistive index	0.89 (0.83-0.94)	> 0.74	82 (74-88)	88 (81-93)
Creatinine	0.73 (0.66-0.80)	> 1.2 mg/dL	68 (60-76)	62 (54-70)

Sensitivity and specificity values are provided with their 95%CI.

Table 3 Specialized tests by acute kidney injury subtype, mean ± SD

Test	Pre-renal acute kidney injury (n = 43)	Hepatorenal syndrome (n = 20)	Acute tubular necrosis (n = 15)	P value
Granular casts ≥ 5/high-power field	0% (0/43)	0% (0/20)	100% (15/15)	< 0.001
Renin (ng/mL)	5.8 ± 2.1	12.1 ± 4.0	6.3 ± 2.4	0.01
Aldosterone (pg/mL)	98 ± 32	240 ± 68	112 ± 44	0.03

P < 0.05 vs pre-renal acute kidney injury using Kruskal-Wallis and Dunn’s correction.

Table 4 Factors independently associated with 30-day mortality in cirrhotic patients with acute kidney injury (Cox proportional hazards model)

Variable	Adjusted hazard ratio (95%CI)	P value
Neutrophil gelatinase-associated lipocalin > 150 ng/mL	3.1 (1.7-5.5)	< 0.001
Renal resistive index > 0.74	2.2 (1.3-3.7)	0.003
Model for End-Stage Liver Disease includes serum sodium > 25	3.9 (2.1-7.2)	0.01

P < 0.05 is considered statistically significant in the Cox regression model.

Table 5 Adjusted hazard ratio for 30-day mortality by acute kidney injury stage (Cox model)

AKI stage	n	30-day mortality count	30-day mortality percentage	Cox P value	Comment	Hazard ratio (95%CI)
No AKI	42	1	2.4%	-	Baseline comparator	1.0 (reference)
Stage 1	28	2	7.1%	0.14	Mild injury, lower risk	2.8 (0.7-10.5)
Stage 2	25	5	20.0%	0.008	Intermediate severity	4.6 (1.5-13.8)
Stage 3	25	14	56.0%	< 0.001	High-risk group, dialysis/hepatorenal syndrome	9.3 (3.7-23.4)

Adjusted hazard ratio (95%CI) for 30-day mortality by acute kidney injury stage (Cox proportional hazards model adjusted for Model for End-Stage Liver Disease includes serum sodium, infection status, and baseline neutrophil gelatinase-associated lipocalin). AKI: Acute kidney injury.

Table 6 Internal validation performance of machine learning models for acute kidney injury subtyping

Model features	Area under the curve from repeated nested cross-validation (95%CI)	Accuracy	F1-score	Brier score	Calibration slope	Sensitivity (95%CI)	Specificity (95%CI)
Neutrophil gelatinase-associated lipocalin + RRI (XGBoost)	0.92 (0.89-0.95)	86%	0.85	0.12	0.96	89% (83-93)	84% (77%-89%)
Creatinine + cystatin C	0.81 (0.75-0.86)	74%	0.73	0.18	0.91	76% (69-82)	71% (63%-78%)
RRI alone	0.78 (0.72-0.84)	72%	0.70	0.20	0.89	73% (66-80)	69% (61%-76%)

Area under the curve from repeated nested cross-validation: (1) 5-fold inner; (2) 10-fold outer; and (3) 10 repeats. Brier score: Lower is better (perfect = 0). Calibration slope: 1 = perfect calibration. Confidence intervals for area under the curve derived from bootstrapping (1000 replicates). RRI: Renal resistive index.

Table 7 Volume assessment and acute kidney injury reclassification via point-of-care ultrasound

Classification	Clinical only (n = 24)	After point-of-care ultrasound (n = 24)
Pre-renal	8	14
Hepatorenal syndrome	6	10
Acute tubular necrosis	10	0

Point-of-care ultrasound was associated with reclassification of 10/24 ambiguous cases (41.7%), with all 10 clinically diagnosed acute tubular necrosis cases reassigned to either pre-renal (6 cases) or hepatorenal syndrome (4 cases) based on inferior vena cava collapsibility and lung B-lines.

Table 8 Unadjusted 30-day mortality and survival by acute kidney injury stage (Kaplan-Meier)

Acute kidney injury stage	n	30-day mortality (%)	Survival (%)	Deaths (n)
Stage 1	28	7.1%	92.9%	2
Stage 2	25	20.0%	80.0%	5
Stage 3	25	56.0%	44.0%	14

Unadjusted 30-day mortality counts, percentages, and survival by acute kidney injury stage (Kaplan-Meier).

Table 9 Net reclassification improvement of the neutrophil gelatinase-associated lipocalin-renal resistive index model compared to creatinine-based diagnosis

Outcome	Net reclassification improvement (95%CI)	P value
Acute kidney injury subtyping	0.41 (0.26-0.57)	< 0.001
30-day mortality risk stratification	0.36 (0.14-0.52)	0.002

A positive net reclassification improvement indicates improved reclassification over the baseline creatinine model.

Table 10 Net clinical benefit of neutrophil gelatinase-associated lipocalin - renal resistive index vs creatinine at selected thresholds

Threshold probability	Neutrophil gelatinase-associated lipocalin-renal resistive index net benefit	Creatinine net benefit
10%	0.12	0.04
25%	0.18	0.06
40%	0.20	0.08

Net benefit is calculated as (true positives - weighted false positives)/total patients. Higher values indicate better clinical decision-making utility.

Table 11 Calibration metrics of the neutrophil gelatinase-associated lipocalin - renal resistive index model for acute kidney injury subtyping

Risk decile	Predicted risk	Observed acute kidney injury rate
1 (lowest)	0.08	0.05
5 (median)	0.42	0.39
10 (highest)	0.83	0.80
Calibration slope	0.96	-
Brier score	0.12

Good calibration indicates that predicted probabilities correspond well to actual observed outcomes across risk levels.

Table 12 Intensive care unit resource utilization with neutrophil gelatinase-associated lipocalin-renal resistive index vs standard care

Outcome	Neutrophil gelatinase-associated lipocalin-renal resistive index Triage	Standard care	Absolute reduction
Total ICU admissions	36	50	-14
Correct ICU allocation (hepatorenal syndrome)	93.3%	86.7%	+6.6%
Unnecessary ICU transfers (pre-renal)	5	19	-14

Intensive care unit triage based on neutrophil gelatinase-associated lipocalin and renal resistive index resulted in improved allocation and reduced overuse of intensive care beds. ICU: Intensive care unit.

Table 13 Factors associated with acute kidney injury development in cirrhotic patients (multivariable logistic regression)

Variable	Odds ratio (95%CI)	P value
Model for End-Stage Liver Disease includes serum sodium (MELD-Na) > 22	2.9 (1.5-5.6)	0.002
Localized infection (non-septic)	2.4 (1.2-4.8)	0.015
Bilirubin > 5 mg/dL	2.1 (1.0-4.3)	0.049
Albumin < 2.8 g/dL	2.0 (1.0-4.0)	0.048
Nephrotoxic agents	2.5 (1.2-5.1)	0.013

P < 0.05 indicates statistical significance for all listed variables.