Early Response to Therapeutic Plasma Exchange as a Prognostic Indicator in Pediatric Liver Transplant Recipients with Early Allograft Dysfunction

ABSTRACT

Background

Early allograft dysfunction (EAD) is a major determinant of morbidity and mortality after pediatric liver transplantation. Therapeutic plasma exchange (TPE) is increasingly used as supportive therapy, yet pediatric‐specific data remain limited. We aimed to determine whether early response to TPE predicts survival in pediatric liver transplant recipients with EAD.

Methods

In this retrospective study, 15 pediatric patients (0–18 years) who developed EAD after living donor liver transplantation and received TPE (2015–2023) were analyzed. EAD was defined by Olthoff criteria. Response was assessed within 72 h using predefined criteria. Patients were classified as responders (n = 9) or non‐responders (n = 6).

Results

Among 140 pediatric liver transplant recipients, 15 (10.7%) developed EAD. The median age was 13 months (IQR: 9.1–22.6), and 80% weighed ≤ 10 kg. Responders showed significant reductions in total bilirubin (17.0 to 9.0 mg/dL; p = 0.004), INR (3.0 to 1.5; p = 0.004), AST (2239 to 150 IU/L; p = 0.004), and ALT (2467 to 200 IU/L; p = 0.004); non‐responders showed no meaningful improvement. In‐hospital mortality occurred exclusively among non‐responders (83.3% vs. 0%; p = 0.001). Kaplan–Meier analysis revealed significant survival divergence (100% vs. 16.7% at day 45; log‐rank p = 0.0005). Pre‐TPE albumin was significantly lower in non‐responders (2.90 vs. 3.20 g/dL; p = 0.013), highlighting albumin as a potential early bedside screening parameter for identifying patients at higher risk of non‐response. One‐year overall survival was 66.7%.

Conclusions

Early response to TPE within 72 h strongly discriminates survival in pediatric EAD and may serve as a practical prognostic tool. Prospective multicenter studies are warranted to validate these findings.

Early biochemical and clinical response to therapeutic plasma exchange within 72 h was strongly associated with survival in pediatric liver transplant recipients with early allograft dysfunction. These findings suggest that early TPE response may help identify grafts with recovery potential and guide timely clinical decision‐making.

1. Introduction

Liver transplantation (LT) is the definitive treatment for end‐stage liver disease in pediatric patients. It has led to substantial improvements in long‐term survival. However, early postoperative complications compromise outcomes in some recipients. Among these, early allograft dysfunction (EAD) remains a frequent and clinically relevant problem. It is associated with increased morbidity, longer intensive care unit stays, and inferior graft survival [1].

EAD is most often defined using Olthoff et al.'s criteria, which are validated in large transplant cohorts. By this definition, EAD is diagnosed within the first seven postoperative days if at least one of the following occurs: peak AST or ALT > 2000 IU/L, total bilirubin ≥ 10 mg/dL on day 7, or INR ≥ 1.6 on day 7 [2]. This approach reliably identifies patients with impaired early graft function and predicts adverse short‐ and long‐term outcomes [3, 4].

EAD pathophysiology is complex and mostly non‐immunological. Ischemia–reperfusion injury (IRI) causes hepatocellular injury, endothelial dysfunction, and microcirculatory impairment after graft reperfusion. Prolonged cold and warm ischemia, marginal donor characteristics, and donor–recipient mismatch contribute to EAD [5, 6, 7]. EAD differs from acute rejection. It represents a transient, sometimes severe, loss of graft function. This can recover with support or progress to primary non‐function [8].

Current management of EAD is largely supportive, focusing on hemodynamic optimization, correction of coagulopathy, and prevention of secondary organ failure. However, in severe cases, conventional supportive strategies may be insufficient to stabilize the patient or allow adequate time for graft recovery. This therapeutic gap has prompted growing interest in extracorporeal liver support therapies, including therapeutic plasma exchange (TPE), as adjunctive treatment options [9].

TPE has been more frequently reported in adults with EAD and other acute liver failures after liver transplant. However, pediatric data remain limited. Pediatric recipients differ from adults in organ size, metabolism, and ability to tolerate interventions outside the body. These differences limit direct comparison to adult data [10, 11]. With no standard treatment pathway and limited pediatric research, further evaluation of TPE in pediatric EAD is needed.

We aimed to determine if early biochemical and clinical responses to TPE predict outcomes in pediatric liver transplant recipients with EAD. We specifically assessed whether a response to TPE within 72 h predicts survival and which pre‐TPE parameters are associated with non‐response.

2. Materials and Methods

2.1. Study Design and Patient Population

This single‐center, retrospective, observational study included pediatric patients aged 0–18 years. All patients had living donor liver transplants at Acıbadem University Hospital between 2015 and 2023. They were followed in the pediatric intensive care unit and received TPE due to EAD. Of 140 pediatric liver transplant recipients during the study period, 15 met diagnostic criteria for EAD and received TPE. All eligible patients were included in the final analysis (Figure 1).

FIGURE 1.

Patient flow diagram. Of 140 pediatric liver transplant recipients, 15 (10.7%) developed early allograft dysfunction (EAD) and received therapeutic plasma exchange (TPE). No patients were excluded following EAD diagnosis. Nine patients (60%) were classified as responders with 100% survival; six (40%) were non‐responders with 83.3% in‐hospital mortality. One non‐responder underwent retransplantation and survived.

All patients diagnosed with EAD underwent a comprehensive and standardized evaluation for potential underlying etiologies. These included sepsis, acute rejection, vascular or biliary complications, and drug‐related toxicity. Assessments used radiological imaging, transcystic cholangiography, microbiological investigations, analysis of donor‐specific antibodies, and systematic review of drug interactions. Routine Doppler ultrasonography was performed during the first seven postoperative days to exclude vascular complications.

At the time of EAD diagnosis and TPE initiation, no evidence of intra‐abdominal sepsis, acute rejection, vascular pathology, or biliary obstruction was found. Complications such as vascular thrombosis developed later in selected cases. All patients who met the diagnostic criteria for early allograft dysfunction during the study period received therapeutic plasma exchange according to our institutional protocol. Thus, no untreated EAD control group was available for comparison.

2.2. Definition of Early Allograft Dysfunction

EAD was diagnosed according to Olthoff et al., requiring at least one of the following within 7 days post‐transplant [2]: total bilirubin ≥ 10 mg/dL on postoperative day (POD) 7 [3]; INR ≥ 1.6 on POD 7 [4]; AST or ALT > 2000 IU/L within the first 7 days. We acknowledge that in the living donor liver transplantation (LDLT) setting—particularly with split or partial grafts—baseline aminotransferase levels may be transiently elevated due to ischemia–reperfusion stress on the reduced graft mass, which may limit the specificity of an isolated transaminase criterion for diagnosing true EAD. To minimize misclassification in our cohort, although the Olthoff framework formally requires only one of the three criteria, all 15 patients meeting the transaminase criterion in the present study also met at least one additional Olthoff criterion (elevated total bilirubin and/or elevated INR), as reflected in the pre‐TPE laboratory values (median total bilirubin 16.0 mg/dL; median INR 2.0). Patients with isolated transaminase elevation in the absence of any concomitant bilirubin or INR abnormality, or without clinical deterioration, were not classified as having EAD requiring TPE. This conservative approach permitted early identification of EAD before POD 7 and timely initiation of TPE, while reducing the risk of overdiagnosis attributable to LDLT‐specific reperfusion patterns.

2.3. Exclusion Criteria

Patients meeting any of the following criteria were excluded from the study:

Development of primary non‐function (PNF), defined as irreversible graft dysfunction within the first seven postoperative days requiring urgent retransplantation, in the absence of identifiable immunologic, vascular, or surgical causes [12].

Presence of vascular complications identified by Doppler ultrasonography or angiography.

Diagnosis of early biliary complications.

Biopsy‐confirmed acute rejection during the EAD period.

Presence of severe sepsis or septic shock at the time of initiation of TPE.

Incomplete or missing clinical and/or laboratory data.

2.4. Immunosuppressive Protocol

All patients undergoing liver transplantation received a standardized immunosuppressive protocol with tacrolimus and corticosteroids. Corticosteroid therapy began intraoperatively at 10 mg/kg (maximum 1 g). The dose was reduced by 50% every 5 days. After tapering, patients received 5 mg/day methylprednisolone for 3 months. Calcineurin inhibitor therapy started on postoperative day 2. Tacrolimus trough levels were targeted between 8 and 10 ng/mL. Immunosuppressive therapy was adjusted based on liver function, histopathology, and clinical evaluation. Mycophenolate mofetil was added selectively based on rejection risk and renal function. Basiliximab was administered to all patients as induction therapy in two intravenous doses on postoperative days 0 and 4, with weight‐based dosing per published pediatric protocols: 10 mg per dose for patients weighing < 35 kg, and 20 mg per dose for those weighing ≥ 35 kg.

All 15 patients diagnosed with EAD received empirical pulse corticosteroid therapy at 10 mg/kg/day (maximum 1 g) initiated at the time of EAD recognition. The duration of pulse therapy was individualized rather than fixed: in patients in whom liver biopsy was feasible (platelet count > 80 000/mm³, absence of active bleeding, and stable coagulation profile), histological diagnosis was typically obtained within 24–48 h and pulse steroids were tapered immediately upon biopsy‐confirmed exclusion of acute rejection. In patients with thrombocytopenia, persistent coagulopathy, or unstable clinical status precluding early biopsy, empirical pulse therapy was continued for up to a maximum of 5 days while ancillary investigations (donor‐specific antibodies, repeat Doppler imaging, microbiological work‐up) were completed and rejection was excluded by clinical–laboratory consensus. Once rejection was excluded by either approach, corticosteroids were tapered back to the standard maintenance dose. Among the 15 EAD patients in this cohort, none demonstrated histological or clinical–laboratory evidence of acute rejection.

2.5. Therapeutic Plasma Exchange (TPE) Procedure

TPE sessions were performed daily or on alternate days, depending on the patient's condition. Double‐lumen central venous catheters (internal jugular or femoral veins) and the Fresenius COM‐TEC system were used. Each session lasted approximately four hours. Fresh frozen plasma was the primary replacement fluid. If allergic reactions occurred, 5% albumin was used instead. The amount of plasma exchanged was calculated as follows: volume = [0.07 × weight (kg)] × [1 − red blood cell percentage]. Anticoagulation was with acid‐citrate‐dextrose (ACD) and heparin, depending on blood clotting and catheter function.

2.6. Determination of the Number of TPE Sessions

A fixed number of TPE sessions was not used in this study. The number of TPE sessions was tailored to each individual. Early clinical and biochemical responses were reassessed daily. Continuation of TPE was considered appropriate in the presence of at least one of the following findings: initiation of a decreasing trend in total bilirubin levels or loss of a previously increasing trend; evidence of improvement in coagulation parameters, defined as a ≥ 0.2 absolute decrease in INR; or achievement of hemodynamic stabilization or a reduction in vasopressor requirements.

2.7. Definition of Response to TPE (Responder/Non‐Responder)

Response classification (responder vs. non‐responder) was determined retrospectively using predefined objective criteria based on laboratory values and clinical parameters extracted from electronic medical records. Response status was not used to guide real‐time clinical management decisions. Response to TPE was assessed within 72 h after the start of therapy or after at least two consecutive sessions. This timeframe was chosen as it allows completion of at least two TPE sessions and is consistent with response assessment intervals used in adult and pediatric TPE liver studies [9, 13], while remaining within the clinical window before secondary complications confound graft function interpretation.

2.8. Definition of Responder

Patients were classified as TPE responders if at least two of the following criteria were met:

≥ 20% reduction in total bilirubin levels from baseline, or total bilirubin to < 10 mg/dL.

≥ 0.2 absolute decrease in INR, or INR ≤ 1.6.

A reduction in vasopressor dose requirements, or complete discontinuation of vasopressor support.

2.9. Definition of Non‐Responder

Patients were classified as TPE non‐responders if there was failure to meet at least two responder criteria despite completion of a minimum of two consecutive TPE sessions, and absence of meaningful clinical or biochemical improvement over 72 consecutive hours (plateau) or evidence of disease progression. Importantly, retransplantation and death were not used to define response status. These events were recorded separately as clinical outcomes. All fatal events and retransplantation occurred after completion of the 72‐h response assessment period.

2.10. Re‐Evaluation and Retransplantation Considerations

All non‐responders were continuously evaluated for retransplantation throughout the TPE course in collaboration with the transplant surgical team and the multidisciplinary liver transplant board. Decisions regarding re‐listing were individualized and based on clinical trajectory, exclusion of irreversible extrahepatic organ failure (severe neurological injury, refractory multiorgan failure), donor availability, and family consent. In our predominantly LDLT setting, the practical availability of an emergency replacement graft was constrained by donor evaluation logistics (workup of a new living donor candidate typically requiring several days), and a formal deceased‐donor PNF re‐listing pathway within 7 days—as routinely available in some North American and Western European systems—was not consistently accessible. These contextual factors are relevant when interpreting the timing of retransplantation decisions in our cohort. Retransplantation ultimately occurred in one patient (portal vein thrombosis) who received a second graft from a related living donor and survived.

2.11. Data Collection

Clinical, demographic, laboratory, and procedural data were retrospectively collected from the institutional electronic medical records and liver transplantation database. Collected variables included patient demographics (age, sex, body weight), primary liver disease, perioperative data, and pediatric intensive care unit (PICU) course. Pretransplant disease severity was assessed using the Pediatric End‐Stage Liver Disease (PELD) score, calculated from standard laboratory parameters and clinical criteria at the time of transplantation. Laboratory parameters recorded at baseline and during follow‐up included total bilirubin, INR, AST, ALT, serum lactate, platelet count, and other routine biochemical markers. Laboratory values were collected at standardized daily time points to allow serial trend analysis.

TPE‐related data included the timing of TPE initiation after transplantation, the number of TPE sessions, the exchanged plasma volume, the type of replacement fluid, and TPE‐related complications. Hemodynamic parameters, vasopressor requirements, and clinical status were documented before and after TPE sessions. Outcome measures included response status, retransplantation, in‐hospital mortality, and 1‐year survival.

2.12. Ethical Approval

This study was approved by the institutional ethics committee, and written informed consent was obtained from all participants' guardians (Approval No: 2023‐21/701).

2.13. Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were assessed for normality using visual (histograms) and analytical methods. Given the small sample size and non‐normal distribution of most variables, continuous data were expressed as median and interquartile range (IQR) or median (minimum–maximum), as appropriate.

Comparisons between responder and non‐responder groups were performed using the Mann–Whitney U test for continuous variables and the Fisher's exact test for categorical variables. Paired comparisons of laboratory parameters before and after TPE were analyzed using the Wilcoxon signed‐rank test. Associations between categorical variables were evaluated using Fisher's exact test, while correlations between continuous variables were assessed using Spearman's rank correlation coefficient. All statistical tests were two‐sided, and p‐values < 0.05 were considered statistically significant. Receiver operating characteristic (ROC) curve analysis was performed for pre‐TPE total bilirubin and INR as candidate predictors of non‐response, and the area under the curve (AUC) was calculated with 95% confidence intervals. Due to the limited number of outcome events, multivariable regression analysis was not feasible.

3. Results

3.1. Patient Demographics and Transplant Characteristics

Among 140 pediatric liver transplant recipients, 15 patients (10.7%) developed early allograft dysfunction and received therapeutic plasma exchange. Patient demographics and transplant characteristics are summarized in Table 1. The cohort consisted of eight females and seven males, with a median age of 13 months (IQR: 9.1–22.6). The median body weight was 8 kg (range: 5–79 kg), and 12 patients (80%) weighed less than 10 kg at the time of transplantation.

TABLE 1.

Baseline characteristics of pediatric recipients with early allograft dysfunction treated with therapeutic plasma exchange.

Characteristic	Value
Patient demographics
Total patients	15
Sex	8 females/7 males (53.3%/46.7%)
Median age (months)	13.0 (IQR: 9.1–22.6)
Median body weight (kg)	8.0 (range: 5–79)
Body weight ≤ 10 kg	12 patients (80.0%); median 6.9 kg (range: 5–10)
Median height (cm)	73 (range: 58–171)
Transplant characteristics
Donor type	100% living donors
Donor relationship	8 mothers (53.3%), 7 fathers (46.7%)
Indications for liver transplantation
Primary diagnosis	Biliary atresia with prior Kasai—60.0% (n = 9); PFIC3—20.0% (n = 3); Propionic acidemia—13.3% (n = 2); PFIC NOS—6.7% (n = 1)
PICU course
PICU length of stay (days)	16 (range: 4–45)
Therapeutic plasma exchange
Mean TPE initiation (POD)	3.7 (range: 1–7)
TPE sessions per patient (range)	2–15 (mean: 4.4)
Clinical outcomes
Responders	9/15 (60.0%)
Non‐responders	6/15 (40.0%)
In‐hospital mortality	5/15 (33.3%)—all non‐responders
Retransplantation	1/15 (6.7%)—portal vein thrombosis; survived
1‐year overall survival	10/15 (66.7%)
1‐year graft survival	9/15 (60.0%)
Death‐censored graft survival	14/15 (93.3%)

Abbreviations: IQR, interquartile range; NOS, not otherwise specified; PFIC, progressive familial intrahepatic cholestasis; PICU, pediatric intensive care unit; POD, postoperative day; TPE, therapeutic plasma exchange.

All grafts were obtained from living donors. The most common indication for liver transplantation was biliary atresia with prior Kasai portoenterostomy (60.0%, n = 9), followed by progressive familial intrahepatic cholestasis type 3 (PFIC3) (20.0%, n = 3), PFIC not otherwise specified (6.7%, n = 1), and propionic acidemia (13.3%, n = 2). The median length of stay in the PICU was 16 days (range: 4–45 days). Based on predefined response criteria, nine patients (60%) were classified as responders, and six patients (40%) as non‐responders. The distribution of primary diagnoses was broadly comparable between groups: among responders (n = 9), biliary atresia was the most frequent indication (n = 5), followed by PFIC3 (n = 2), PFIC NOS (n = 1), and propionic acidemia (n = 1); among non‐responders (n = 6), biliary atresia again predominated (n = 4), followed by PFIC3 (n = 1) and propionic acidemia (n = 1). No specific underlying liver disease was disproportionately represented among non‐responders.

Baseline demographic characteristics and pre‐TPE biochemical parameters were compared between responder and non‐responder groups (Table 2). There were no statistically significant differences between groups in terms of age, body weight, timing of TPE initiation, number of TPE sessions, or pre‐TPE laboratory parameters, including total bilirubin, INR, AST, ALT, creatinine, CRP, and procalcitonin (all p > 0.05). Pre‐TPE albumin levels were significantly lower in non‐responders compared to responders (median 2.90 (IQR: 2.75–2.98) vs. 3.20 (IQR: 2.90–3.60) g/dL; p = 0.013), with a large effect size (rank‐biserial r = 0.796). Calculated PELD scores were also comparable between groups, with a median of 30.2 (IQR: 15.4–37.8) in responders and 27.5 (IQR: 20.2–31.4) in non‐responders (p = 0.768).

TABLE 2.

Baseline characteristics of responders and non‐responders.

Variable	Responder (n = 9)	Non‐responder (n = 6)	p	Effect size (r)
Age (months)	13.0 (11.7–15.8)	15.5 (9.1–45.1)	0.689	0.15
Weight (kg)	8.00 (5.00–8.00)	7.50 (7.00–15.50)	0.510	0.04
Height (cm)	69.00 (61.00–74.00)	74.00 (67.00–95.25)	0.376	0.41
PELD Score	30.2 (15.4–37.8)	27.5 (20.2–31.4)	0.768	0.15
TPE started day (POD)	2.00 (2.00–4.00)	5.00 (2.50–6.00)	0.708	0.43
Number of TPE sessions	3.00 (3.00–5.00)	4.00 (3.25–4.00)	0.544	0.09
Pre‐TPE Total bilirubin (mg/dL)	17.00 (13.00–20.00)	15.50 (15.00–18.25)	1.000	0.52
Pre‐TPE INR	3.00 (2.00–3.00)	2.00 (2.00–2.00)	0.763	0.26
Pre‐TPE AST (U/L)	2239.00 (1757.00–2762.00)	2225.00 (1892.25–2547.25)	1.000	0.07
Pre‐TPE ALT (U/L)	2467.00 (2050.00–3120.00)	1756.50 (1026.25–2453.00)	0.224	0.31
Pre‐TPE Albumin (g/dL)	3.20 (2.90–3.60)	2.90 (2.75–2.98)	0.013*	0.80*
Pre‐TPE CRP (mg/L)	5.00 (4.00–6.00)	7.00 (6.50–29.00)	0.554	0.19
Pre‐TPE Procalcitonin (ng/mL)	1.5 (0.7–5.0)	2.2 (1.25–5.25)	0.906	0.02

Note: Data are presented as median (interquartile range). Continuous variables were compared using the Mann–Whitney U test. *Statistically significant (p < 0.05).

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRP, C‐reactive protein; INR, international normalized ratio; PELD, pediatric end‐stage liver disease; POD, postoperative day; TPE, therapeutic plasma exchange.

To explore the predictive value of pre‐TPE laboratory parameters, ROC curve analysis was performed for total bilirubin and INR as potential predictors of non‐response. Pre‐TPE total bilirubin demonstrated an AUC of 0.759 (95% CI: 0.455–1.000), with an optimal threshold of 18.2 mg/dL yielding a sensitivity of 66.7% and specificity of 77.8%. Pre‐TPE INR yielded an AUC of 0.630 (95% CI: 0.304–0.920). The 95% confidence intervals for both AUC estimates are wide—particularly for bilirubin, where the lower bound (0.455) crosses the line of no discrimination—reflecting the limited number of outcome events available for analysis. These results should therefore be interpreted as exploratory and hypothesis‐generating only, requiring prospective validation in larger pediatric cohorts (Figure 2). Given the significant intergroup difference in pre‐TPE albumin and its large effect size (rank‐biserial r = 0.796; p = 0.013), exploratory examination of pre‐TPE albumin as a predictor of non‐response was also undertaken. All non‐responders had pre‐TPE albumin values ≤ 3.0 g/dL (upper IQR bound 2.98 g/dL), whereas the majority of responders had values above this threshold (responder lower IQR bound 2.90 g/dL), suggesting that a pre‐TPE albumin cut‐off in the range of 2.95–3.0 g/dL may have potential discriminative value as a simple bedside parameter. Formal ROC validation with adequately powered cohorts is required before this threshold can be applied clinically.

FIGURE 2.

Receiver operating characteristic (ROC) curves for pre‐TPE total bilirubin and INR as predictors of non‐response to therapeutic plasma exchange. The optimal bilirubin threshold of 18.2 mg/dL (filled circle) yielded a sensitivity of 66.7% and specificity of 77.8% (area under the curve [AUC] = 0.759; 95% confidence interval [CI]: 0.455–1.000). The dashed diagonal represents chance‐level discrimination. Given the small sample size, findings should be interpreted as exploratory.

3.2. Therapeutic Plasma Exchange and Biochemical Changes

TPE was initiated promptly upon diagnosis of EAD, with a mean initiation time of postoperative day 3.7 (range: 1–7). The number of TPE sessions ranged from 2 to 15, with a mean of 4.4 sessions per patient.

When all patients were analyzed together (n = 15), TPE was associated with a significant reduction in total bilirubin levels, which decreased from a median of 16.0 mg/dL (IQR: 13.0–19.5) before TPE to 9.0 mg/dL (IQR: 8.0–14.5) after TPE (p = 0.028). Serum aminotransferase levels declined markedly following TPE. Median AST levels decreased from 2239 IU/L (IQR: 1790–2688) to 187 IU/L (IQR: 79–411) (p = 0.002), and median ALT levels decreased from 2300 IU/L (IQR: 1612–2751) to 297 IU/L (IQR: 98–342) (p < 0.001). Pre‐TPE INR was 2.0 (IQR: 1.7–3.0) and showed a decrease to 1.5 (IQR: 1.2–2.1) following TPE; however, this change did not reach statistical significance (p = 0.184). Overall pre‐post comparisons (Table 3) should be interpreted in conjunction with the response‐stratified analysis presented in Table 4.

TABLE 3.

Comparative analysis of biochemical parameters before and after therapeutic plasma exchange (n = 15).

Parameter	Pre‐TPE median (IQR)	Post‐TPE median (IQR)	p
AST (U/L)	2239 (1790–2688)	187 (79–411)	0.002*
ALT (U/L)	2300 (1612–2751)	297 (98–342)	< 0.001*
Total Bilirubin (mg/dL)	16.0 (13.0–19.5)	9.0 (8.0–14.5)	0.028*
Fibrinogen (mg/dL)	112 (102.5–149.5)	111 (96–128)	0.454
BUN (mg/dL)	21 (12.5–26.0)	14 (12.0–23.5)	0.389
Creatinine (mg/dL)	0.3 (0.2–0.4)	0.2 (0.2–0.5)	0.505
Albumin (g/dL)	2.9 (2.8–3.5)	3.0 (2.6–3.2)	0.762
CRP (mg/L)	2.7 (0.7–4.9)	1.1 (0.5–2.5)	0.095
Procalcitonin (ng/mL)	1.5 (0.8–6.0)	2.0 (0.5–4.0)	0.121
INR	2.0 (1.7–3.0)	1.5 (1.2–2.1)	0.184
PT (sec)	24.7 (18.6–38.0)	22.0 (18.4–28.0)	0.454
APTT (sec)	37.8 (26.0–52.0)	34.0 (29.0–58.5)	1.000

Note: *Statistically significant (p < 0.05); Wilcoxon signed‐rank test used for all paired comparisons.

Abbreviations: ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; BUN, blood urea nitrogen; CRP, C‐reactive protein; INR, international normalized ratio; IQR, interquartile range; PT, prothrombin time; TPE, therapeutic plasma exchange.

TABLE 4.

Response‐stratified pre‐ and post‐TPE biochemical changes.

Parameter	Responders (n = 9)			Non‐Responders (n = 6)
	Pre‐TPE	Post‐TPE	p	Pre‐TPE	Post‐TPE	p
Total Bilirubin (mg/dL)	17.0 (13.0–20.0)	9.0 (6.0–11.0)	0.004*	15.5 (15.0–18.3)	15.5 (12.0–19.0)	0.375
INR	3.0 (2.0–3.0)	1.5 (1.3–1.8)	0.004*	2.0 (2.0–2.0)	2.1 (2.0–2.3)	0.062
AST (IU/L)	2239 (1757–2762)	150 (79–400)	0.004*	2225 (1892–2547)	2100 (800–2800)	0.062
ALT (IU/L)	2467 (2050–3120)	200 (80–342)	0.004*	1757 (1026–2453)	1950 (900–2900)	0.031*
Albumin (g/dL)	3.2 (2.9–3.6)	3.1 (2.8–3.4)	0.481	2.9 (2.8–3.0)	2.8 (2.6–3.0)	0.844
Fibrinogen (mg/dL)	128 (103–150)	112 (96–130)	0.285	108 (102–148)	105 (90–120)	0.844
CRP (mg/L)	5.0 (4.0–6.0)	1.0 (0.5–2.5)	0.016*	7.0 (6.5–29.0)	5.0 (1.5–9.0)	0.438
Procalcitonin (ng/mL)	1.5 (0.7–5.0)	2.0 (1.0–3.0)	0.820	2.2 (1.3–5.3)	1.1 (0.2–4.2)	0.062
Creatinine (mg/dL)	0.3 (0.2–0.4)	0.2 (0.2–0.4)	0.285	0.3 (0.2–0.5)	0.3 (0.2–0.6)	0.844

Note: Data are presented as median (IQR). *Statistically significant (p < 0.05); Wilcoxon signed‐rank test used for all paired comparisons.

Abbreviation: CRP, C‐reactive protein.

3.3. Response‐Based Analysis of Pre–Post Changes

In the responder group (n = 9), TPE resulted in statistically significant improvements in all key biochemical markers of graft function: total bilirubin decreased from a median of 17.0 to 9.0 mg/dL (p = 0.004), INR from 3.0 to 1.5 (p = 0.004), AST from 2239 to 150 IU/L (p = 0.004), and ALT from 2467 to 200 IU/L (p = 0.004) (Table 4). Hemodynamic response was also evident, with a reduction in vasopressor dose requirements or complete discontinuation of vasopressor support observed within the first 72 h of TPE initiation.

In contrast, in the non‐responder group (n = 6), no clinically meaningful biochemical improvement was observed following TPE. Total bilirubin remained unchanged (median 15.5 to 15.5 mg/dL, p = 0.375), INR showed no significant change (2.0 to 2.1, p = 0.062), and AST remained markedly elevated (p = 0.062). Although a modest reduction in ALT was observed (p = 0.031), this isolated finding was not accompanied by improvement in other parameters and did not meet the predefined response criteria (Table 4). Importantly, the significant bilirubin reduction observed in the overall cohort analysis (Table 3) was driven entirely by the responder subgroup. Non‐responders showed no consistent biochemical improvement, indicating that the apparent overall pre–post effect should not be interpreted as a uniform treatment response across all patients.

3.4. Clinical Outcomes

In‐hospital mortality occurred in five patients (33%) and was observed exclusively among patients classified as non‐responders (5/6, 83.3%), whereas no deaths occurred in the responder group (0/9). The difference between groups was statistically significant (p = 0.001998, Fisher's exact test). Early non‐response to TPE demonstrated 100% sensitivity, 90% specificity, 83.3% positive predictive value, and 100% negative predictive value for predicting in‐hospital mortality (Table 1). However, given the small sample size and limited number of outcome events, these predictive estimates should be considered exploratory and hypothesis‐generating rather than definitive.

Kaplan–Meier survival analysis demonstrated a marked divergence in overall survival between responder and non‐responder groups. Responders maintained 100% survival throughout the follow‐up period, whereas the survival probability in non‐responders decreased progressively to 16.7% by day 45 (Figure 3). The difference in survival distributions between groups was statistically significant by the log‐rank test (χ ² = 12.04, p = 0.0005).

FIGURE 3.

Kaplan–Meier survival curves for therapeutic plasma exchange (TPE) responders (n = 9) and non‐responders (n = 6). Time was defined as days from transplantation to death or censoring at 365 days. × marks indicate death events; tick marks indicate censored observations. Log‐rank test: χ ² = 12.04, p = 0.0005.

3.5. Adverse Events During Therapeutic Plasma Exchange

A total of two TPE‐related adverse events were recorded across 66 TPE sessions, yielding an event rate of 3.0% per session. At the patient level, 2 of 15 patients (13.3%) experienced at least one complication. The reported complications were mild hypotension in one patient and hypotension accompanied by chills in another. All complications were transient and resolved with supportive management. No TPE session was terminated due to complications. No life‐threatening complications directly attributable to TPE were observed.

3.6. Retransplantation and Mortality

Retransplantation was required in one patient (6.7%) due to portal vein thrombosis. This patient was 18 months old, had undergone liver transplantation for progressive familial intrahepatic cholestasis (PFIC), received four sessions of therapeutic plasma exchange, and was classified as a non‐responder according to the prespecified 72‐h response criteria prior to the development of the vascular complication. Portal vein thrombosis was diagnosed by Doppler ultrasonography on postoperative day 12, approximately 5 days after the fourth and final TPE session, in the context of progressive graft dysfunction and clinical deterioration. The temporal separation between the last TPE session and the onset of thrombosis, together with the central anatomical location of the thrombus and the documented progressive graft injury, collectively favor a multifactorial etiology related to the underlying severe graft dysfunction and surgical–anatomic factors rather than a direct procedural complication of TPE. Although catheter‐related thrombotic events are a recognized rare complication of extracorporeal therapies, no clinical or imaging evidence linked the central venous access used for TPE to portal venous thrombosis in this case. The PVT was therefore not classified as a TPE‐attributable adverse event. The patient survived following retransplantation and remained alive at the last follow‐up.

During follow‐up, five patients (33%) died. All deaths occurred among patients classified as non‐responders based on the prespecified 72‐h response criteria. Deaths were attributed to severe graft dysfunction and complex postoperative complications rather than to TPE itself. Documented causes of death included vascular thrombosis, intracranial hemorrhage, intestinal necrosis with ileal perforation, renal failure with encephalopathy, hepatic encephalopathy, and intra‐abdominal hemorrhage. The minimum follow‐up duration for surviving patients was 12 months. Overall, one‐year patient survival was 66.7% (10/15). Overall graft survival was 60% (9/15). Death‐censored graft survival was 93.3% (14/15) (Table 1).

4. Discussion

Although significant improvements have been achieved in liver transplantation outcomes over recent decades, graft dysfunction remains one of the leading causes of morbidity and mortality, particularly in pediatric recipients [11, 14]. While retransplantation is theoretically the definitive treatment option for primary graft failure, its applicability is often limited in countries with a restricted deceased donor pool and where living donor liver transplantation predominates [15]. Therefore, supportive and bridging treatment strategies have gained increasing clinical importance [1]. Early allograft dysfunction (EAD) represents a severe clinical condition that may progress rapidly and result in death before a suitable graft becomes available, making early intervention crucial in this patient population [2, 3]. In line with our institutional protocol, TPE was initiated early in all patients diagnosed with EAD.

Although an increasing body of evidence from adult EAD series suggests that TPE may improve graft function, data in the pediatric population remain limited [1, 9, 16]. The potential benefits of TPE stem from its ability to target the multifactorial pathophysiology of EAD, characterized by systemic inflammation, oxidative stress, microcirculatory disturbances, and the accumulation of albumin‐bound hepatotoxic molecules [5, 6, 7]. By removing circulating proinflammatory cytokines, endotoxins, bilirubin, and ammonia, TPE may reduce hepatocellular stress and modulate dysregulated immune responses [6, 7]. In addition, the use of fresh frozen plasma allows replacement of coagulation factors and albumin, thereby providing temporary support for the liver's synthetic functions [17]. These mechanisms may be particularly relevant in pediatric patients, in whom smaller circulating blood volumes and more rapid bilirubin accumulation can lead to faster clinical deterioration [11]. In adult series, TPE has been associated with significant reductions in bilirubin and INR levels accompanied by improved survival, providing a plausible biological rationale for the biochemical and clinical improvements observed in our cohort [16, 17, 18].

In the present study, AST and ALT levels were markedly elevated, with median values of 2239 and 2300 IU/L, respectively, indicating severe hepatocellular injury [18]. These levels exceed those reported in previously published pediatric and adult series and may reflect the heightened vulnerability of pediatric recipients to ischemia–reperfusion injury and systemic inflammatory stress. While statistically significant reductions in AST (p = 0.002) and ALT (p < 0.001) were observed following TPE, we agree with the view that aminotransferase changes during TPE should not be considered meaningful indicators of true biological graft response. Aminotransferases are released from injured hepatocytes and are susceptible to both dilutional and adsorptive removal during extracorporeal plasma exchange; their decline during TPE thus reflects, at least in part, mechanical removal rather than restored hepatocellular synthetic function. Furthermore, in the LDLT setting, the reduced graft mass may itself contribute to lower absolute aminotransferase output during the recovery phase, independent of TPE. For these reasons, post‐TPE AST/ALT trajectories were not used in our response classification, and total bilirubin together with predefined coagulation and clinical criteria were considered the more reliable indicators of treatment response. The aminotransferase data are presented for descriptive completeness only and should not be over‐interpreted as evidence of TPE efficacy.

Across the full cohort, a significant reduction in total bilirubin levels was observed (median 16.0 to 9.0 mg/dL, p = 0.028). However, as demonstrated by the response‐stratified analysis (Table 4), this apparent improvement is entirely driven by the responder subgroup, in whom bilirubin fell from 17.0 to 9.0 mg/dL (p = 0.004). In non‐responders, bilirubin remained essentially unchanged (15.5 to 15.5 mg/dL, p = 0.375). Given the absence of a control group, it also cannot be determined whether the observed bilirubin reduction in responders is attributable to TPE or to spontaneous graft recovery [18, 19]. Although INR values decreased following TPE, this change did not reach statistical significance (Table 3). This may be explained by the transient correction of coagulation parameters following fresh‐frozen plasma infusion, the limited sample size, or differences in baseline disease severity [16, 17].

Clinical response to TPE was observed in 60% of patients, a rate comparable to those reported in previous pediatric series. Notably, mortality occurred exclusively among non‐responders, suggesting a strong association between early biochemical response and survival. An important interpretive caveat must be acknowledged: the responder classification itself was partly based on biochemical criteria which are the same parameters used to demonstrate post‐TPE improvement in Table 4. This overlap introduces a degree of circularity; however, survival outcomes were not part of the classification criteria and were assessed independently, which limits—but does not eliminate—the concern about circular reasoning [16, 19, 20].

In our cohort, only two mild adverse events were observed across 66 TPE sessions (3.0% per session). All complications were transient and successfully managed without interruption of therapy. This complication rate lies at the lower end of the 2%–6% range reported in the literature and supports the safety of TPE in pediatric patients [16, 20].

In the present cohort, retransplantation was required in only one patient, in whom portal vein thrombosis developed following the initial transplant. Although this patient had received multiple sessions of therapeutic plasma exchange (TPE), the indication for retransplantation was clearly related to a vascular complication, a condition that cannot be corrected by extracorporeal blood purification techniques. This finding underscores that TPE does not replace surgical intervention in the presence of structural complications, but rather serves as a supportive or bridging therapy to stabilize graft function [15, 21].

Overall mortality occurred exclusively among non‐responders, whereas no deaths were observed in the responder group (Figure 3). The comparability of baseline characteristics between groups is critical for interpreting this finding: responders and non‐responders did not differ significantly in age, body weight, PELD score (median 30.2 vs. 27.5; p = 0.768), pre‐TPE bilirubin, INR, AST, ALT, or inflammatory markers (Table 2). Notably, pre‐TPE INR values were numerically higher in the responder group, further reducing the likelihood that baseline disease severity alone explains the survival difference. Effect size analysis revealed that pre‐TPE albumin showed a large, statistically significant difference between groups (rank‐biserial r = 0.796, p = 0.013), with non‐responders showing notably lower levels (median 2.90 vs. 3.20 g/dL). Lower pre‐TPE albumin in non‐responders may indicate a state of more advanced or irreversible hepatocellular injury at the time of TPE initiation. Several pathophysiological mechanisms may underlie the prognostic value of pre‐TPE albumin in this population. First, low albumin in the early post‐transplant window likely reflects severe and persistent failure of hepatic synthetic function and may identify a “point of no return” beyond which functional recovery of the graft is unlikely despite extracorporeal support. Second, hypoalbuminemia in this acute setting may reflect a systemic capillary leak phenomenon, driven by ischemia–reperfusion injury and a heightened post‐transplant inflammatory response, with redistribution of albumin to the extravascular space; capillary leak syndrome is itself an independent marker of poor outcome in critically ill children. Third, albumin is the principal carrier protein for unconjugated bilirubin and several hepatotoxic metabolites; lower circulating albumin reduces the buffering capacity of plasma and may amplify the toxic burden on the recovering graft and on extrahepatic organs. These mechanisms are not mutually exclusive and may converge to identify patients in whom TPE alone is insufficient to achieve graft stabilization. Pre‐TPE albumin may therefore serve as a simple bedside parameter to identify patients at higher risk of non‐response. Exploratory ROC analysis suggested that a pre‐TPE bilirubin threshold of 18.2 mg/dL may also have moderate predictive value for non‐response (AUC = 0.759; 95% CI: 0.455–1.000; sensitivity 66.7%, specificity 77.8%); however, given the wide confidence interval and small sample, prospective validation in larger cohorts is required before clinical application [16, 18].

The number and timing of TPE sessions were comparable between groups (Table 2), suggesting that survival differences were unlikely to be explained by treatment exposure or delayed initiation alone. Importantly, all fatal events occurred after completion of the predefined 72‐h response assessment window, confirming that response classification preceded mortality. Deaths were attributed to severe graft dysfunction and complex postoperative complications rather than to TPE‐related adverse events.

Similar associations between early biochemical non‐response and poor survival have been reported in adult and limited pediatric series (Table 5) [16, 19], suggesting that failure to respond to TPE may represent a surrogate marker of irreversible graft injury rather than an indicator of treatment inefficacy. The absence of mortality among responders supports the hypothesis that early TPE initiation and subsequent biochemical improvement may contribute to graft stabilization and improved short‐term survival [16, 17]. However, given the absence of an untreated control group, it cannot be determined whether observed benefits are attributable to TPE itself or to the natural recovery trajectory of patients with inherently less severe or more reversible graft injury. Accordingly, early response to TPE should be interpreted primarily as a prognostic marker rather than as proof of therapeutic efficacy [9, 18, 22].

TABLE 5.

Comparison with published literature on therapeutic plasma exchange for early allograft dysfunction.

Study	Population	Total LT (n)	EAD n (%)	TPE n	Bilirubin change (mg/dL)	Mortality (TPE group)	Control group
Present study (2015–2023)	Pediatric LDLT (0–18 years)	140	15 (10.7%)	15	16.0 to 9.0; p = 0.028	33.3% overall; 83.3% non‐responders; 0% responders	None (all EAD received TPE)
Choe et al. 2017 [16]	Adult LDLT + DDLT	2804 (13‐years period)	143 (5.1%); 107 TPE; 36 non‐TPE	107	15.2 to 13.1; p < 0.001	1‐month: 17.8%; 1‐year: 46.2%	Concurrent unmatched EAD non‐TPE (n = 36)
Hakeem et al. 2021 [9]	Adult LDLT + pediatric	659 LDLT (465 adult)	67 (14.4% of adult LDLT); TPE subset: 29	29	13.6 to 12.3; p = 0.356	30‐day: 38%; 90‐day: 45%; 1‐year: 45%	Unmatched concurrent non‐TPE EAD (n = 24)
Camci et al. 2004 [19]	Adult + pediatric (age 1–63 years)	NR	25 (rate NR)	25	15.4 to 11.9; p = 0.001	24% (6/25 died; 2 retransplanted; 19/25 survived)	None

Abbreviations: DDLT, deceased donor liver transplantation; EAD, early allograft dysfunction; LDLT, living donor liver transplantation; NR, not reported.

4.1. Limitations

This study has several limitations. First, the retrospective single‐center design and relatively small sample size limit the generalizability of the findings. Due to the limited number of patients and outcome events, multivariable regression analysis was not feasible, and potential confounding factors could not be fully adjusted for. Second, the absence of an untreated EAD control group precludes definitive conclusions regarding the causal impact of TPE on survival. According to institutional protocol and ethical considerations, withholding TPE in patients fulfilling established EAD criteria was not considered appropriate. Importantly, this design limitation also means that responders—identified by 72‐h biochemical improvement—might in some instances have recovered spontaneously or with conventional supportive care alone without TPE; without an untreated control arm, the contribution of TPE itself to graft recovery cannot be isolated, and the response classification should be interpreted primarily as a prognostic stratifier rather than as evidence of treatment efficacy. Third, although response classification was based on a predefined 72‐h landmark assessment to minimize reverse causation bias, residual confounding cannot be entirely excluded. Several perioperative variables—including cold ischemia time, warm ischemia time, graft‐to‐recipient weight ratio, donor age, donor BMI, hepatic steatosis grade, and intraoperative blood loss—were not systematically captured in our database and could not be incorporated as adjustment variables, representing an additional source of unmeasured confounding. Fourth, formal liver biopsies were not systematically performed or available for centralized review across all 15 patients; consequently, histological characterization of severe ischemia–reperfusion injury, hyperacute or antibody‐mediated rejection, and quantitative microscopic assessment of the graft could not be incorporated into the analysis. Fifth, our institutional and regulatory context (predominantly LDLT, with limited deceased‐donor PNF re‐listing pathways within 7 days) differs from many North American and Western European systems, which may affect external generalizability—particularly regarding the timing and feasibility of urgent retransplantation in TPE non‐responders. Finally, while Kaplan–Meier survival analysis demonstrated a strong association between early TPE response and survival, these results should not be interpreted as proof of therapeutic efficacy. Future studies should specifically address these limitations: a multicenter collaborative design that includes centers not routinely employing TPE for pediatric EAD would allow propensity‐matched comparisons (matching for sex, age, primary diagnosis, body weight, PELD score, and graft type) to better isolate the contribution of TPE to outcome, and prospective biomarker validation—including pre‐TPE albumin—in larger cohorts is warranted.

5. Conclusions

Therapeutic plasma exchange appears to be a safe and feasible supportive strategy in pediatric liver transplant recipients who develop early allograft dysfunction. In this cohort, TPE was associated with significant biochemical improvement in responding patients, particularly in total bilirubin levels. Most importantly, early clinical and biochemical response within the first 72 h strongly discriminated short‐ and one‐year survival outcomes. Although the observational design precludes definitive conclusions regarding causality, the absence of mortality among responders and the marked survival difference between groups suggest that early response to TPE may serve as a valuable prognostic marker in pediatric EAD. In addition, pre‐TPE serum albumin emerged as a simple and clinically accessible parameter that may help identify, at the bedside, patients at higher risk of non‐response and warrants prospective validation as an early screening tool. TPE should be considered a supportive or bridging therapy rather than a substitute for retransplantation in cases of irreversible structural graft complications. Larger multicenter studies—ideally incorporating propensity‐matched comparisons with EAD cohorts not exposed to TPE, with matching for age, sex, primary diagnosis, body weight, and PELD score—are warranted to confirm whether an early TPE response reflects intrinsic graft recovery potential or directly contributes to improved survival outcomes.

Author Contributions

I.B., T.U.Y., and A.C. conceived and designed the study. I.B. and O.U. collected and analyzed the data. I.B. and T.U.Y. drafted the manuscript. G.S. and E.E. contributed to data interpretation and critical revision of the manuscript. R.E. contributed to clinical data and transplantation data collection. A.C. supervised the study and reviewed the final manuscript. All authors read and approved the final version and agree to be accountable for all aspects of the work.

Funding

The authors have nothing to report.

Ethics Statement

This study was approved by the institutional ethics committee of Acıbadem Mehmet Ali Aydınlar University (Approval No: 2023–21/701). Written informed consent was obtained from all participants' guardians.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The raw data supporting the conclusions of this article will not be made publicly available due to ethical restrictions and patient confidentiality requirements. The data are not publicly available as they contain information that could compromise the privacy of the patients included in the study. De‐identified data may be made available by the corresponding author upon reasonable request, subject to institutional approval.