Total bile acid better reflects the early biochemical biliary drainage response after endoscopic retrograde cholangiopancreatography

Abstract

BACKGROUND

Currently, early assessment of the biliary drainage response after therapeutic biliary endoscopic retrograde cholangiopancreatography (ERCP) commonly relies on bilirubin-based markers, but these may not fully reflect the earliest postoperative biochemical improvement. We hypothesized that the total bile acid (TBA) level may provide an earlier and more sensitive signal of biochemical recovery after biliary drainage.

AIM

To compare the early postoperative trajectories of TBA, bilirubin, and cholestatic enzymes after therapeutic biliary ERCP.

METHODS

This single-center retrospective cohort study included adults undergoing therapeutic biliary ERCP. Laboratory values were assessed at baseline within 24 hours before ERCP, at the first postoperative test within 24 hours, and at the second postoperative test within 5 days. Baseline comparisons, logistic regression, generalized estimating equations, paired within-patient comparisons, and unified longitudinal models were used to evaluate whether TBA served as an earlier and more sensitive biochemical indicator of drainage response than bilirubin-based markers after therapeutic ERCP.

RESULTS

The main analytic cohort comprised 896 patients, of whom 278 achieved an early biochemical drainage response while 618 did not. TBA showed the most pronounced postoperative decline, while the decreases in direct bilirubin and total bilirubin occurred more slowly. Also, in the 836 patients with complete three-time-point measurements, the relative decline in TBA was significantly greater than that of direct bilirubin and total bilirubin at both 24 hours and within 5 days postoperatively. Exploratory analyses further showed that although the levels of gamma-glutamyl transferase and alkaline phosphatase also declined significantly after drainage, their early relative declines remained significantly smaller than that of TBA.

CONCLUSION

TBA declines earlier and more markedly than bilirubin-based and conventional cholestatic markers after biliary drainage, supporting its use as an adjunctive early biochemical indicator after therapeutic biliary ERCP.

Key Words: Total bile acid; Bilirubin; Biliary drainage; Endoscopic retrograde cholangiopancreatography; Cholestasis; Postoperative monitoring

Core Tip: Bilirubin-based markers are commonly used to assess the biliary drainage response after therapeutic endoscopic retrograde cholangiopancreatography, but their early postoperative changes may be limited. In this real-world cohort, total bile acid (TBA) declined earlier and more markedly than direct and total bilirubin. Although gamma-glutamyl transferase and alkaline phosphatase also declined after drainage, their early changes were smaller than those of TBA. These findings support TBA as a responsive adjunctive marker of early biochemical recovery after biliary drainage.

INTRODUCTION

Endoscopic retrograde cholangiopancreatography (ERCP) is a well-established therapeutic procedure for a wide range of biliary disorders, including choledocholithiasis, benign biliary strictures, and malignant biliary obstruction[1-4]. In routine clinical practice, effective biliary drainage after ERCP is usually inferred from the technical success of the procedure, symptom improvement, and postoperative biochemical recovery in patients. Among the indicators considered to assess biochemical improvements after ERCP, serum bilirubin-based markers are the most commonly used laboratory surrogates for decompression and jaundice resolution[5,6]. However, the earliest postoperative biochemical response after ERCP is often difficult to interpret, particularly within the first 24 hours, when the clinical symptoms may remain nonspecific and imaging reassessment is not routinely performed.

The levels of total bilirubin (Tbil) and direct bilirubin (Dbil) have traditionally been regarded as key markers of cholestasis and drainage efficacy[7-11]. Nevertheless, the kinetics of bilirubin may not fully reflect the earliest biological consequences of a restoration of the normal bile flow. In particular, changes in the level of bilirubin after biliary intervention are influenced not only by biliary excretion itself, but also by its hepatocellular uptake, conjugation, intracellular transport, redistribution, and systemic clearance[12-14]. As a result, bilirubin values may show a delayed postoperative improvement even when the biliary drainage has been technically successful. This limitation is clinically relevant because decisions regarding postoperative monitoring, the need for closer observation, and the timing of reassessment are often made before bilirubin has had sufficient time to decline substantially.

Recent analytical work has emphasized how bilirubin can be measured using conventional laboratory assays as well as newer spectroscopic, fluorescent, and biosensing platforms[15-17]. While these advances have improved its analytical detection, they cannot overcome the inherent biological delay in bilirubin kinetics after biliary drainage. Thus, a marker that is more directly linked to the restored bile flow may still be useful for early post-ERCP assessment.

In this regard, total bile acid (TBA) may provide a more direct biochemical reflection of the recovery in biliary outflow. Bile acids are central components of bile formation and enterohepatic circulation, and their serum concentrations are closely linked to cholestatic physiology[18,19]. Experimental and clinical studies have shown that serum bile acids are sensitive indicators of hepatobiliary dysfunction and may respond dynamically to an impaired or restored bile excretion[20-22]. In addition, it has been proposed that bile acids are not merely digestive products but may also serve as signaling molecules involved in hepatic adaptation, cholestatic injury, and metabolic regulation[23,24]. These features suggest that TBA could be a plausible candidate marker for detecting early biochemical improvements after biliary decompression.

However, the comparative early postoperative behaviors of TBA and bilirubin-based markers after ERCP have not been well characterized yet. Indeed, most previous studies evaluating the post-drainage biochemical response after ERCP have focused on bilirubin regression, particularly in malignant biliary obstruction or in the context of preoperative optimization and the timing of subsequent therapy[6,25]. By contrast, data directly comparing TBA with Tbil and Dbil during the immediate postoperative period in ERCP are limited. It therefore remains unclear whether TBA can capture early drainage-related biochemical changes more sensitively than conventional bilirubin-based markers.

Accordingly, we conducted a real-world cohort study of patients undergoing therapeutic biliary ERCP to compare the early postoperative trajectories of TBA, Dbil, and Tbil in such patients. Using predefined peri-procedural laboratory time windows and complementary longitudinal analytic approaches, we aimed to determine whether TBA could serve as an earlier and more sensitive biochemical indicator of the biliary drainage response than bilirubin-based markers after ERCP.

MATERIALS AND METHODS

Study design and patients

This was a single-center, retrospective, real-world cohort study based on a consecutively maintained ERCP database of hospitalized patients treated from 2023 to 2026. Adult patients who underwent therapeutic biliary ERCP and had available peri-procedural biochemical data were considered eligible for screening. The study protocol was reviewed and approved by the Ethics Committee of the Second Affiliated Hospital of Guangxi Medical University [approval No. 2018-KYL (003)]. The requirement for informed consent was waived because of the retrospective study design and the use of anonymized clinical data. Furthermore, to reduce selection and information bias, the cohort was derived from a consecutively maintained ERCP database, and predefined inclusion and exclusion criteria were applied before outcome assessment. The time windows for the laboratory tests and clinical classifications were prespecified, and the key procedure-related variables were manually reviewed before the statistical analysis.

Inclusion and exclusion criteria

Patients were included in the study analysis if they met all of the following criteria: (1) Age ≥ 18 years; (2) Underwent therapeutic biliary ERCP for a biliary indication, including stone disease, malignant biliary stricture, benign biliary stricture, or other biliary obstruction-related conditions; (3) Had baseline liver biochemical tests available within 24 hours before ERCP; and (4) Had at least one first postoperative biochemical assessment within 24 hours after ERCP. Also, for inclusion in the main analytic cohort, patients were required to have paired baseline and first postoperative Tbil or Dbil values. For inclusion in the trajectory-analysis cohort, an additional second postoperative biochemical assessment within 5 days after ERCP was required. Because the trajectory-analysis cohort required complete three-time-point measurements, patients with complete TBA/Dbil/Tbil measurements were compared with those without complete three-time-point data to assess potential selection bias.

Patients were excluded from the study analysis if they met any of the following criteria: (1) Had pancreatic rather than biliary disease as the primary indication; (2) Had normal pancreaticobiliary ducts or a clearly non-biliary primary problem; (3) ERCP failure without actual biliary therapy; (4) Had received stent removal only without new biliary intervention; (5) Had an altered upper gastrointestinal anatomy that could substantially affect the procedure’s characteristics, such as Billroth II reconstruction; or (6) Were missing key laboratory data that would preclude an assessment of postoperative biochemical changes. Non-initial ERCP procedures were not excluded a priori and were retained as a clinical characteristic in the analysis.

Data review and variable reconstruction

The ERCP database was cleaned and standardized before analysis. Key variables were then manually reviewed to optimize the clinical classifications, including diagnosis regrouping, the separation of biliary and pancreatic stents, drainage-strategy classification, and confirmation of whether the index procedure qualified as therapeutic biliary ERCP. The reviewed results were then incorporated into the final analysis-ready dataset.

Definitions

Laboratory data were analyzed at three predefined time points: Baseline, with values obtained within 24 hours before ERCP, the first postoperative assessment obtained within 24 hours after ERCP, and the second postoperative assessment obtained within 5 days after ERCP.

The main diagnosis of the included patients was regrouped into more detailed clinically meaningful categories, including stone disease, malignant stricture, benign stricture, and other biliary conditions. Drainage strategy was classified as endoscopic nasobiliary drainage (ENBD) only, biliary plastic stent only, ENBD plus biliary plastic stent, or no continuous drainage device. Pancreatic stents were not considered biliary drainage devices in the drainage-strategy classification.

An early biochemical drainage response was defined as a decline of at least 30% in either Tbil or Dbil on the first postoperative test. This pragmatic bilirubin-based threshold was chosen to ensure it would exceed minor early variations because no universally validated immediate post-ERCP threshold exists. TBA was not used to define the outcome to avoid incorporation bias, as TBA was the candidate marker under evaluation.

Laboratory quality control and unavailable variables

All the biochemical values herein refer to clinically reported results from specimens accepted by the hospital laboratory. Generally, hemolyzed or otherwise unacceptable specimens are not released as final results; if hemolysis is detected, the laboratory requests repeat sampling within approximately 1-2 hours.

Because this was a retrospective clinical study, experimental replicates were not applicable. Also, exact sampling timestamps, detailed post-ERCP pharmacotherapy, and prothrombin time/international normalized ratio were not systematically recorded; therefore, these factors could not be fully adjusted for.

Statistical analysis

Continuous variables are reported herein as the median (interquartile range) or mean ± SD, and categorical variables as n (%). Between-group comparisons were performed using the Mann-Whitney U, Student’s t, χ², or Fisher’s exact tests, as appropriate. Logistic regression was used to evaluate the factors associated with an early biochemical drainage response. Marker trajectories were analyzed using generalized estimating equation models adjusted for the diagnosis and drainage strategy. Paired comparisons used Wilcoxon signed-rank tests with multiplicity adjustment, and cross-marker patterns were assessed using log postoperative-to-baseline ratio models. Additional analyses included complete-case comparison, receiver operating characteristic curve/Youden cutoff for 24-hour TBA decline, alternative 20%/30%/50% bilirubin-response thresholds, albumin/creatinine sensitivity analyses, and a 24-hour-only cross-marker comparison. A two-sided P < 0.05 was considered statistically significant. The statistical analyses used R software version 4.5.2 (R Foundation for Statistical Computing, Vienna, Austria). The statistical methods were reviewed by Liang ZH, MD, PhD, from the First Affiliated Hospital of Guangxi Medical University.

RESULTS

Study population

In total, 982 patients were screened for study participation (Figure 1); 86 patients were subsequently excluded, with the remaining 896 enrolled as the main cohort, comprising 278 responders and 618 non-responders. For the paired cross-marker analyses, 836 patients who had complete three-time-point TBA/Dbil/Tbil data were included. Complete-case comparisons (Supplementary Table 1) showed broadly comparable baseline characteristics among those patients, although their baseline TBA, stone clearance, drainage strategy, and early response differed; this potential complete-case bias was considered when interpreting the results from the trajectory analyses.

Figure 1 Study flowchart of cohort derivation. The flowchart summarizes patient screening, exclusions, and final analytic cohorts for the main analysis, trajectory analysis, and cross-marker comparison. ERCP: Endoscopic retrograde cholangiopancreatography; TBA: Total bile acid; Dbil: Direct bilirubin; Tbil: Total bilirubin.

Baseline characteristics according to the early biochemical drainage response

The patients’ baseline characteristics are summarized in Table 1. It was revealed that patients with an early biochemical drainage response were more likely to have undergone initial rather than repeat ERCP (85.6% vs 71.5%, P < 0.001) and they differed significantly in terms of the diagnosis composition and drainage strategy distribution. Stone disease was more common among responders than non-responders (80.9% vs 70.6%), whereas benign stricture was less frequent (4.3% vs 12.9%). The responders also had a larger common bile duct diameter and a higher rate of complete stone clearance. In addition, the responders showed significantly higher baseline levels of Tbil, Dbil, alkaline phosphatase, gamma-glutamyl transferase, TBA, and white blood cell count than the non-responders, whereas age and sex did not differ significantly between the groups.

Table 1 Baseline characteristics according to early biochemical drainage response after endoscopic retrograde cholangiopancreatography, n (%)/mean ± SD.

Variable	No response	Response	P value
n	618	278
Age, years	57.68 ± 14.32	57.50 ± 15.69	0.867
Male sex	411 (66.5)	178 (64.0)	0.518
Initial ERCP	442 (71.5)	238 (85.6)	< 0.001
Main diagnosis			0.001
Stone disease	436 (70.6)	225 (80.9)
Malignant stricture	89 (14.4)	36 (12.9)
Benign stricture	80 (12.9)	12 (4.3)
Other biliary disease	6 (1.0)	0 (0.0)
Clonorchiasis	5 (0.8)	4 (1.4)
Biliary dilation of unknown cause	2 (0.3)	1 (0.4)
Drainage strategy			< 0.001
ENBD only	347 (56.1)	203 (73.0)
ENBD + stent	87 (14.1)	26 (9.4)
No continuous drainage device	9 (1.5)	4 (1.4)
Stent only	175 (28.3)	45 (16.2)
Common bile duct diameter, mm	10.15 ± 6.01	11.53 ± 6.16	0.002
Stone clearance			< 0.001
No/partial clearance	47 (7.6)	14 (5.0)
Complete clearance	405 (65.5)	230 (82.7)
Baseline Tbil	87.75 ± 118.75	111.36 ± 98.09	0.004
Baseline Dbil	68.85 ± 99.26	85.42 ± 75.87	0.013
Baseline ALP	234.44 ± 213.50	269.77 ± 192.43	0.019
Baseline GGT	328.31 ± 349.96	539.38 ± 585.05	< 0.001
Baseline TBA	192.09 ± 270.53	315.57 ± 537.64	< 0.001
Baseline WBC	7.97 ± 4.14	9.37 ± 4.84	< 0.001

Early biochemical drainage response was defined according to the study criteria described in the Methods section. ERCP: Endoscopic retrograde cholangiopancreatography; ENBD: Endoscopic nasobiliary drainage; Tbil: Total bilirubin; Dbil: Direct bilirubin; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transferase; TBA: Total bile acid; WBC: White blood cell count.

Overall peri-procedural biochemical changes

The overall peri-procedural biochemical changes are summarized in Table 2 and Figure 2. Tbil decreased from 49.30 (15.95, 126.40) at baseline to 39.85 (19.48, 103.83) within 24 hours and to 29.35 (15.70, 83.42) within 5 days after ERCP. Dbil decreased from 37.10 (7.80, 101.48) to 26.20 (9.40, 83.10) and then to 19.80 (8.30, 65.90). TBA showed the most pronounced decline, falling from 123.70 (22.05, 326.40) at baseline to 10.40 (2.80, 83.50) within 24 hours and further to 4.80 (1.80, 18.35) within 5 days. Table 2 also summarizes the postoperative changes in alkaline phosphatase and gamma-glutamyl transferase. All the baseline-to-postoperative comparisons were significant (all P < 0.001).

Figure 2 Early postoperative changes in total bile acid, direct bilirubin, and total bilirubin after therapeutic biliary endoscopic retrograde cholangiopancreatography. A: Median trajectories of total bile acid, direct bilirubin, and total bilirubin at baseline, within 24 hours after endoscopic retrograde cholangiopancreatography (ERCP), and within 5 days after endoscopic retrograde cholangiopancreatography; B: Relative decline from baseline at 24 hours after ERCP; C: Relative decline from baseline within 5 d after ERCP; D: Cross-marker comparison using the decline index [-log (postoperative/baseline ratio)]. Higher values indicate greater postoperative decline from baseline. Box plots show medians and interquartile ranges. ^dP < 0.0001; IQR: Interquartile range; TBA: Total bile acid; Dbil: Direct bilirubin; Tbil: Total bilirubin.

Table 2 Three-time-point changes in key liver biochemical markers before and after endoscopic retrograde cholangiopancreatography, median (interquartile range).

Marker	Baseline	24 hours after ERCP	5 days after ERCP	P value (baseline vs 24 hours after ERCP)	P value (baseline vs 5 days after ERCP)
Tbil	49.30 (15.95, 126.40)	39.85 (19.48, 103.83)	29.35 (15.70, 83.42)	< 0.001	< 0.001
Dbil	37.10 (7.80, 101.48)	26.20 (9.40, 83.10)	19.80 (8.30, 65.90)	< 0.001	< 0.001
ALP	184.00 (109.50, 294.50)	173.00 (108.00, 282.00)	151.00 (102.00, 243.00)	< 0.001	< 0.001
GGT	284.00 (108.00, 552.00)	255.00 (108.00, 457.00)	198.00 (92.00, 353.50)	< 0.001	< 0.001
TBA	123.70 (22.05, 326.40)	10.40 (2.80, 83.50)	4.80 (1.80, 18.35)	< 0.001	< 0.001

P values were calculated for paired comparisons between baseline and postoperative measurements. Sample sizes differed slightly across markers because of missing laboratory measurements. ERCP: Endoscopic retrograde cholangiopancreatography; Tbil: Total bilirubin; Dbil: Direct bilirubin; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transferase; TBA: Total bile acid.

Longitudinal postoperative trajectories

Marker-specific generalized estimating equation analyses showed distinct postoperative patterns across the evaluated biochemical markers (Table 3). For Tbil, the change within 24 hours after ERCP was not significant (β = -0.025, P = 0.211), whereas the decline by day 5 was significant (β = -0.229, P < 0.001). For Dbil, both the 24-hour and 5-day declines were significant (β = -0.063, P = 0.004 and β = -0.249, P < 0.001, respectively). By contrast, TBA showed the strongest early response, with a marked decline already observed within 24 hours (β = -1.633, P < 0.001) and a further decline by day 5 (β = -2.322, P < 0.001). Exploratory analyses further showed that the levels of gamma-glutamyl transferase and alkaline phosphatase also declined significantly after ERCP. Gamma-glutamyl transferase was decreased at 24 hours (β = -0.074, P < 0.001) and further at 5 days (β = -0.269, P < 0.001), whereas alkaline phosphatase was also decreased at both time points (β = -0.044 and -0.145, both P < 0.001).

Table 3 Generalized estimating equation analyses of postoperative time effects after endoscopic retrograde cholangiopancreatography.

Marker	Variable	β	SE	Wald χ2	P value
Tbil	Time: 24 hours after ERCP vs baseline	-0.025	0.020	1.567	0.211
	Time: 5 days after ERCP vs baseline	-0.229	0.023	98.944	< 0.001
Dbil	Time: 24 hours after ERCP vs baseline	-0.063	0.022	8.362	0.004
	Time: 5 days after ERCP vs baseline	-0.249	0.025	99.213	< 0.001
TBA	Time: 24 hours after ERCP vs baseline	-1.633	0.054	929.271	< 0.001
	Time: 5 days after ERCP vs baseline	-2.322	0.065	1265.955	< 0.001
GGT	Time: 24 hours after ERCP vs baseline	-0.074	0.015	24.117	< 0.001
	Time: 5 days after ERCP vs baseline	-0.269	0.018	217.510	< 0.001
ALP	Time: 24 hours after ERCP vs baseline	-0.044	0.008	27.219	< 0.001
	Time: 5 days after ERCP vs baseline	-0.145	0.011	163.313	< 0.001

Generalized estimating equation models were fitted separately for each marker and adjusted for diagnosis category and drainage strategy. Only the primary time effects comparing 24 hours after endoscopic retrograde cholangiopancreatography and 5 days after endoscopic retrograde cholangiopancreatography with baseline are shown. Negative coefficients indicate lower postoperative values relative to baseline. ERCP: Endoscopic retrograde cholangiopancreatography; Tbil: Total bilirubin; Dbil: Direct bilirubin; TBA: Total bile acid; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase.

Cross-marker comparisons of the TBA, Dbil, and Tbil

The cross-marker comparisons are summarized in Tables 4 and 5 and Figure 2. Among the 836 patients with complete three-time-point measurements, paired within-patient comparisons showed that the median relative decline in TBA was 79.69% at 24 hours, compared with 5.70% for Dbil and 3.98% for Tbil; furthermore, both comparisons were found to be significant after adjustment (adjusted P < 0.001). By day 5, the median relative decline in TBA reached 92.00%, whereas the corresponding declines were 18.58% for Dbil and 17.91% for Tbil (both adjusted P < 0.001). Descriptive relative-decline analyses showed the same pattern, with substantially larger declines for TBA than for the bilirubin-based markers at both postoperative time points.

Table 4 Cross-marker comparisons of total bile acid, direct bilirubin, and total bilirubin after endoscopic retrograde cholangiopancreatography (descriptive relative declines).

Marker	Time point	n	Median (interquartile range)
TBA	24 hours after ERCP	836	79.69 (42.81, 95.37)
TBA	5 days after ERCP	836	92.00 (56.23, 98.52)
Dbil	24 hours after ERCP	836	5.70 (-31.98, 34.92)
Dbil	5 days after ERCP	836	18.58 (-22.85, 52.99)
Tbil	24 hours after ERCP	836	3.98 (-35.82, 32.32)
Tbil	5 days after ERCP	836	17.91 (-24.21, 52.57)

Table 4 presents descriptive relative declines from baseline. TBA: Total bile acid; Dbil: Direct bilirubin; Tbil: Total bilirubin; ERCP: Endoscopic retrograde cholangiopancreatography.

Table 5 Cross-marker comparisons of total bile acid, direct bilirubin, and total bilirubin after endoscopic retrograde cholangiopancreatography (paired comparisons).

Time point	Comparison	n	Median decline of first marker (%)	Median decline of second marker (%)	P value	Adjusted P value
24 hours after ERCP	TBA vs Dbil	836	79.69	5.70	< 0.001	< 0.001
	TBA vs Tbil	836	79.69	3.98	< 0.001	< 0.001
5 days after ERCP	TBA vs Dbil	836	92.00	18.58	< 0.001	< 0.001
	TBA vs Tbil	836	92.00	17.91	< 0.001	< 0.001

Table 5 presents paired within-patient comparisons of relative decline between markers. Higher relative decline indicates greater postoperative reduction from baseline. ERCP: Endoscopic retrograde cholangiopancreatography; TBA: Total bile acid; Dbil: Direct bilirubin; Tbil: Total bilirubin.

In the unified log-ratio generalized estimating equation model, using TBA at 24 hours as the reference, Dbil (β = 1.716, P < 0.001) and Tbil (β = 1.742, P < 0.001) showed significantly smaller early declines. TBA continued to decline from 24 hours to day 5 (β = -0.820, P < 0.001), and the marker-by-time interactions for Dbil and Tbil remained significant (both P < 0.001). These patterns were also consistent across the diagnosis and drainage strategy (Figures 3 and 4). A 24-hour-only sensitivity analysis not requiring 5-day data showed the same finding: TBA declined by 80.20% (44.64, 95.58), compared with 6.63% (-31.58, 36.16) for Dbil and 4.63% (-36.17, 33.56) for Tbil; furthermore, the paired comparisons remained significant (both adjusted P < 0.001; Supplementary Table 2).

Figure 3 Marker trajectories stratified by main diagnosis. A: Total bilirubin trajectories according to diagnosis group; B: Direct bilirubin trajectories according to diagnosis group; C: Total bile acid trajectories according to diagnosis group. Median values with interquartile ranges are shown at baseline, within 24 hours after endoscopic retrograde cholangiopancreatography, and within 5 days after endoscopic retrograde cholangiopancreatography. IQR: Interquartile range.

Figure 4 Marker trajectories stratified by drainage strategy. A: Total bilirubin trajectories according to drainage strategy; B: Direct bilirubin trajectories according to drainage strategy; C: Total bile acid trajectories according to drainage strategy. Median values with interquartile ranges are shown at baseline, within 24 hours after endoscopic retrograde cholangiopancreatography, and within 5 days after endoscopic retrograde cholangiopancreatography. IQR: Interquartile range; ENBD: Endoscopic nasobiliary drainage.

Exploratory cutoff analysis for early TBA decline

Exploratory receiver operating characteristic curve analysis (Supplementary Figure 1) showed that the 24-hour decline in TBA could also reflect the bilirubin-defined response with moderate discrimination [area under the curve = 0.675, 95% confidence interval (CI): 0.638-0.713; n = 873]. The Youden cutoff was a 75.10% decline (sensitivity 0.730, specificity 0.526; Supplementary Table 3). Alternative bilirubin-response thresholds of 20%, 30%, and 50% yielded similar area under the curves (0.674-0.696) and cutoffs (72.85%-80.01%; Supplementary Table 4), supporting the robustness of the results while leaving the cutoff still exploratory.

Exploratory comparison with conventional cholestatic enzymes

Exploratory analyses additionally incorporating gamma-glutamyl transferase and alkaline phosphatase were also performed. Complete three-time-point data were available for 862 patients for gamma-glutamyl transferase, 844 for alkaline phosphatase, and 828 for the combined comparison with TBA. In the unified model, both enzymes showed significantly smaller early declines than TBA, and their temporal patterns remained distinct through day 5 (Table 6, Figure 5). Furthermore, sensitivity analyses excluding albumin < 30 g/L, creatinine ≥ 177 μmol/L, or either abnormality were consistent with the primary analysis; TBA remained greater than Dbil/Tbil at 24 hours and 5 days (all adjusted P < 0.001; Supplementary Table 5).

Figure 5 Exploratory comparison of total bile acid with gamma-glutamyl transferase and alkaline phosphatase. A: Post-endoscopic retrograde cholangiopancreatography (ERCP) trajectory of gamma-glutamyl transferase; B: Post-ERCP trajectory of alkaline phosphatase; C: Relative decline of total bile acid, gamma-glutamyl transferase, and alkaline phosphatase at 24 hours after ERCP; D: Relative decline of total bile acid, gamma-glutamyl transferase, and alkaline phosphatase within 5 days after ERCP. Median values with interquartile ranges or box plots are shown as appropriate. IQR: Interquartile range; TBA: Total bile acid; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase.

Table 6 Unified generalized estimating equation model comparing total bile acid, gamma-glutamyl transferase, and alkaline phosphatase after endoscopic retrograde cholangiopancreatography.

Variable	Estimate	SE	Wald χ2	P value
Marker: GGT (vs TBA at 24 hours after ERCP)	1.537	0.057	725.857	< 0.001
Marker: ALP (vs TBA at 24 hours after ERCP)	1.566	0.056	793.137	< 0.001
Time: 5 days after ERCP vs 24 hours after ERCP (for TBA)	-0.719	0.086	70.145	< 0.001
Interaction: GGT × 5 days after ERCP	0.523	0.089	34.226	< 0.001
Interaction: ALP × 5 days after ERCP	0.616	0.087	49.795	< 0.001

The unified model was based on the log-transformed postoperative-to-baseline ratio and was adjusted for diagnosis category and drainage strategy. Positive coefficients for marker terms indicate smaller relative declines than total bile acid at 24 hours after endoscopic retrograde cholangiopancreatography. Significant interaction terms indicate temporal patterns that differ from that of total bile acid between 24 hours and 5 days after endoscopic retrograde cholangiopancreatography. ERCP: Endoscopic retrograde cholangiopancreatography; TBA: Total bile acid; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase.

Factors associated with an early biochemical drainage response

Factors associated with an early biochemical drainage response are summarized in Table 7 and Figure 6. In the multivariable model, initial ERCP was found to be independently associated with a higher likelihood of an early response [odds ratio (OR) = 1.772, 95%CI: 1.163-2.748, P = 0.009]. Compared with ENBD alone, both ENBD plus stent (OR = 0.496, 95%CI: 0.284-0.838, P = 0.011) and stent-only drainage (OR = 0.564, 95%CI: 0.354-0.887, P = 0.015) were associated with lower odds of an early response. Higher baseline Tbil (OR = 1.003, 95%CI: 1.001-1.005, P < 0.001), gamma-glutamyl transferase (OR = 1.002, 95%CI: 1.001-1.002, P < 0.001), and white blood cell count (OR = 1.042, 95%CI: 1.006-1.080, P = 0.021) were also independently associated with an early biochemical drainage response.

Figure 6 Multivariable predictors of early biochemical drainage response. Forest plot showing odds ratios and 95% confidence intervals from the final multivariable logistic regression model for early biochemical drainage response after ERCP. ERCP: Endoscopic retrograde cholangiopancreatography; ENBD: Endoscopic nasobiliary drainage; Tbil: Total bilirubin; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transferase; WBC: white blood cell count.

Table 7 Multivariable logistic regression analysis of factors associated with early biochemical drainage response after endoscopic retrograde cholangiopancreatography.

Variable	OR (95%CI)	P value
Initial ERCP	1.772 (1.163-2.748)	0.009
Drainage strategy: ENBD + stent	0.496 (0.284-0.838)	0.011
Drainage strategy: Stent only	0.564 (0.354-0.887)	0.015
Common bile duct diameter, mm	1.026 (0.997-1.056)	0.083
Baseline Tbil	1.003 (1.001-1.005)	< 0.001
Baseline ALP	0.999 (0.998-1.000)	0.105
Baseline GGT	1.002 (1.001-1.002)	< 0.001
Baseline WBC	1.042 (1.006-1.080)	0.021

Only variables retained in the final multivariable model are shown. Odds ratios were estimated for early biochemical drainage response after endoscopic retrograde cholangiopancreatography. Continuous variables were modeled per unit increase. ERCP: Endoscopic retrograde cholangiopancreatography; OR: Odds ratio; CI: Confidence interval; ENBD: Endoscopic nasobiliary drainage; Tbil: Total bilirubin; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transferase; WBC: White blood cell count.

DISCUSSION

ERCP is an established therapeutic procedure for a broad range of biliary disorders, and assessment of the drainage response remains clinically important after intervention[26,27]. In routine practice, postoperative improvement is usually judged by the technical success of the procedure, symptom relief, and biochemical recovery. Among laboratory markers, bilirubin remains the most familiar indicator of cholestasis resolution and is frequently used in clinical studies and outcome definitions[5,28-30]. However, most bilirubin-based criteria were developed for assessment over days to weeks rather than for the immediate postoperative period. To address this gap, the present study focused on this early window. The study’s findings showed that TBA behaved differently from bilirubin-based markers after therapeutic biliary ERCP. Although all major cholestasis-related biochemical variables were improved after ERCP, TBA showed the earliest and most pronounced decline, whereas Dbil and especially Tbil changed more slowly. Across multiple complementary analyses, the same pattern was consistently observed. Taken together, these findings support the potential use of TBA as a more responsive early biochemical indicator of the biliary drainage response than bilirubin-based markers.

The clinical relevance of this observation lies in the limitations of bilirubin-based monitoring. Bilirubin is intuitive, widely available, and clinically familiar, but its kinetics do not necessarily reflect the earliest biological consequences of the restored bile flow. Prior work has shown that bilirubin-based clinical success is usually assessed over a longer interval, often using relative reductions over 1-2 weeks or an absolute threshold, such as a Tbil level below 2 mg/dL[5,28-31]. In contrast, clinicians often need to interpret postoperative biochemical change within the first 24 hours after the intervention, when symptoms may remain nonspecific and imaging reassessment is not routinely performed. In this setting, bilirubin may respond too slowly. This was also reflected in our data, where Tbil showed the slowest early decline and Dbil, although more responsive, still changed much less than TBA during the immediate postoperative period. These findings suggest that a reliance on bilirubin alone, particularly Tbil, may underestimate early biochemical improvements after technically successful biliary decompression.

There is also a biologically plausible explanation for why TBA may decline earlier. Bile acids are directly linked to bile formation, hepatocellular transport, canalicular secretion, and enterohepatic circulation[18,19]. In cholestasis, circulating bile acids rise substantially, and once bile flow is restored, the serum bile acid pool may decrease relatively rapidly because it is closely tied to the obstructed biliary pathway itself. By contrast, bilirubin-based improvements may be slower and less immediate, and may therefore lag behind the actual restoration of the bile flow[32,33]. This interpretation is consistent with previous studies showing that serum bile acids are sensitive indicators of hepatobiliary dysfunction and cholestasis[20,21], and with contemporary concepts of bile acids as metabolic and signaling molecules involved in hepatic adaptation and cholestatic injury[23,24]. Although the present study was not designed to investigate the mechanisms directly, the observed temporal pattern fits well with the current understanding of bile acid biology.

An additional contribution of this study is that it clarifies the relative positions of Dbil and Tbil in early postoperative monitoring. Our data do not suggest that all bilirubin-based markers perform equally poorly during the immediate post-ERCP period. Rather, they suggest a hierarchy of responsiveness. TBA appears to be the earliest and most sensitive of the markers evaluated, while Dbil occupies an intermediate position, and Tbil is the slowest to reflect early changes. This pattern is biologically reasonable because Dbil is more closely related to biliary excretory dysfunction than Tbil[34]. Within-patient comparisons are especially informative because they reduce between-patient heterogeneity and can directly address which marker changes first after ERCP in the same individual. In our cohort, TBA showed a substantially greater relative decline than either Dbil or Tbil at both 24 hours and within 5 days. The results from the unified cross-marker longitudinal model further strengthened this interpretation, showing that both bilirubin-based markers had significantly smaller early declines than TBA and that their trajectories remained significantly different over time.

Our exploratory analyses additionally extended this comparison to conventional cholestatic enzymes. Previous studies have suggested that conventional cholestatic enzymes, particularly gamma-glutamyl transferase (GGT), may provide useful information in biliary obstruction[35,36]. In the present post-ERCP setting, we found that GGT also declined significantly during the early postoperative period, supporting the view that conventional cholestatic enzymes may respond earlier than bilirubin-based markers. However, the magnitude of this early change remained substantially smaller than that of TBA, while alkaline phosphatase changed even more slowly. In the unified longitudinal model, both GGT and alkaline phosphatase showed significantly smaller relative declines than TBA, and their temporal patterns remained distinct through day 5. These exploratory analyses therefore reinforce the main conclusion that TBA is the most responsive early biochemical indicator after therapeutic biliary ERCP.

From a clinical perspective, TBA may be a useful adjunct for early post-ERCP monitoring, especially when bilirubin remains relatively unchanged during the first postoperative day. However, TBA availability varies by institution and time of day. When TBA testing is unavailable, bilirubin, GGT, and alkaline phosphatase remain acceptable, but clinicians should recognize their slower early response.

This study has key strengths to highlight, including its relatively large consecutive real-world cohort, predefined laboratory windows, manual clinical review, and complementary longitudinal analyses. However, limitations remain and should be noted. First, its retrospective single-center design may have caused selection bias and residual confounding despite our use of prespecified criteria, paired analyses, generalized estimating equation adjustment, stratified plots, multivariable models, and sensitivity analyses. Second, complete three-time-point data were not missing at random; baseline TBA, stone clearance, drainage strategy, and early response data differed from incomplete cases. However, 24-hour-only analysis showed the same patterns, suggesting that the early TBA advantage was not driven solely by the complete-case selection. Third, exact sampling timestamps, medications, and prothrombin time/international normalized ratio were unavailable, so within-window timing, pharmacotherapy, and hepatic synthetic dysfunction could not be fully adjusted for. Meanwhile, while hemolyzed specimens were not released by the laboratory quality control, the albumin/creatinine sensitivity analyses were consistent with the primary results. Fourth, the single Chinese tertiary-center setting, moderate internally derived TBA cutoff, and lack of long-term clinical endpoints limit the generalizability and clinical implementation of our findings. Prospective multicenter studies should use fixed sampling times, medication recording, external validation, and clinical endpoints to address the limitations in the present study.

CONCLUSION

TBA decreased earlier and more markedly than bilirubin-based markers and conventional cholestatic enzymes after therapeutic biliary ERCP. TBA may serve as an adjunctive early biochemical indicator of drainage response, but actionable cutoffs and multicenter validation are required before its routine implementation as such.