Prognostic factors analysis of ultrasound-guided percutaneous puncture drainage in severe acute pancreatitis with peripancreatic fluid collection

Abstract

BACKGROUND

Severe acute pancreatitis (SAP) is a critical illness associate with mortality between 15%-30%, and very frequently complicated by peripancreatic fluid collections (PFC) in more than 80% of necrotizing pancreatitis. Ultrasound-guided percutaneous puncture drainage (PPD) has become the mainstay of minimal invasive treatment for PFC, but clinical outcomes after the therapeutic procedure differ significantly from patient to patient. Drainage success independent predictors are not well defined, indicating a need for systematic identification of prognostic factors affecting drainage to optimize patient selection and individualized treatment strategies.

AIM

To identify potential independent predictors for drainage success and failure, evaluate the influence of the timing and characteristics of the collection on clinical outcomes, and generate evidence-based criteria to optimize digestive tract drainage in this high-risk cohort.

METHODS

A retrospective study was performed for 134 SAP patients who underwent PPD under ultrasound guidance to treat PFC at the Department of Emergency of Hangzhou First People’s Hospital from January 2019 to December 2024. All patients fulfilled the revised Atlanta Classification criteria for SAP, and symptomatic PFC was diagnosed by computed tomography (CT). The Hangzhou First People’s Hospital Ethics Committee approved the study protocol. Patients were categorized into good-prognosis and poor-prognosis groups by a composite endpoint of in-hospital mortality, need for emergent surgical intervention, or organ failure greater than 14 days. Univariate and multivariate time logistic regression analyses were performed to determine independent prognostic factors. Subgroup analyses assessed clinical outcomes based on timing of drainage (early: ≤ 72 hours or late: > 72 hours from diagnosis) and collection type [acute PFC/pseudocyst vs walled-off necrosis (WON)].

RESULTS

Of the 134 SAP patients treated by ultrasound-guided PPD for PFC, good prognosis and poor prognosis were observed in 82 (61.2%) and 52 (38.8%) cases, respectively. The clinical success rate of PPD was 71.6% (96/134) in total. Univariate analysis indicated that Acute Physiology and Chronic Health Evaluation (APACHE) II score ≥ 17, infected necrosis confirmed, treatment at > 72 hours from diagnosis, CT severity index ≥ 8, organ failure of ≥ 2 systems, C-reactive protein ≥ 300 mg/L when performing drainage, collection volume ≥ 400 mL and WON type and serum albumin < 30 g/L were significant prognostic factors (all P < 0.05). APACHE II score ≥ 17 [odds ratio (OR) = 5.82, 95% confidence interval (CI): 2.38-14.22], confirmed infected necrosis (OR = 4.16, 95%CI: 1.76-9.84), drainage timing > 72 hours (OR = 3.74, 95%CI: 1.58-8.86), CT severity index ≥ 8 (OR = 3.28, 95%CI: 1.38-7.80) and multisystem organ failure (OR = 3.06, 95%CI: 1.23-7.62) were identified as independent predictors of poor prognosis based on multivariate logistic regression (all P < 0.05). Early drainage (≤ 72 hours) showed significantly higher clinical success rate (82.9% vs 59.4%, P = 0.004), shorter in intensive care units stay (8.4 days vs 14.8 days, P < 0.001), and lower in-hospital mortality rate (5.7% vs 21.9%, P = 0.011). Compared to patients with acute PFC/pseudocyst, those with WON type had significantly worse outcomes as shown by lower clinical success rates (57.9% vs 89.7%, P < 0.001) and higher mortality (21.1% vs 3.4%, P = 0.003).

CONCLUSION

Ultrasound-guided PPD is an effective minimally invasive approach for the management of PFC associated with SAP, with a clinical success rate of 71.6%. Independent prognostic factors include APACHE II score, confirmed infected necrosis, drain timing, CT severity and multisystem organ failure. Early draining by 72 hours and correct pre-procedural risk stratification are all important to improve clinical outcomes. WON type collections may be where more aggressive drainage strategies and monitoring is needed. These results support evidence-based, individualized decision making for drainage in SAP patients with PFC.

Key Words: Severe acute pancreatitis; Peripancreatic fluid collection; Ultrasound-guided drainage; Percutaneous catheter drainage; Prognostic factors

Core Tip: This study identifies key prognostic factors influencing outcomes of ultrasound-guided percutaneous puncture drainage in severe acute pancreatitis with peripancreatic fluid collections. Independent predictors of poor prognosis include high Acute Physiology and Chronic Health Evaluation II score, infected necrosis, delayed drainage (> 72 hours), elevated computed tomography severity index, multisystem organ failure, high C-reactive protein levels, large collection volume, and hypoalbuminemia. Early drainage (≤ 72 hours) was associated with higher clinical success, reduced intensive care units and hospital stay, and lower mortality. Walled-off necrosis showed significantly worse outcomes compared to fluid collections without necrosis. These findings support early intervention and comprehensive pre-procedural risk stratification to guide individualized drainage strategies and improve clinical outcomes.

INTRODUCTION

Severe acute pancreatitis (SAP) is a potentially life-threatening inflammatory disease of the pancreas with systemic inflammatory response and multiorgan dysfunction, and mortality rates ranging from 15% to 30%[1]. SAP remains a major clinical and socioeconomic burden worldwide, in part due to the advances we witnessed in critical care and multidisciplinary management. Peripancreatic fluid collection (PFC) is one such local complication that is both common and clinically significant, found in 50%-80% of patients with necrotizing pancreatitis[2]. PFCs comprise a continuum of entities including acute PFCs (APFC), acute necrotic collections, pseudocysts and walled-off necrosis (WON) based on the updated Atlanta Classification with different morphological features and clinical significances[3].

The approach to manage PFC in SAP has matured over the last 20 years, establishing a paradigm shift from surgical to a step-up strategy of minimally-invasive intervention[4]. Ultrasound-guided percutaneous puncture drainage (PPD) has become the backbone of the step-up approach for necrotizing and non-necrotizing pancreatitis, allowing for real-time guidance, high accuracy, absence of general anesthesia requirement, and decompression of infected or symptomatic collections with minimal procedural morbidity[5]. Percutaneous drainage reduces the risk of complications associated with open surgical necrosectomy significantly, including new-onset organ failure incisional hernias and pancreatic fistulas while providing comparable rates of infection control at least in appropriately selected patients[6]. In addition, in some cases patients with PPD alone can also obtain curative treatment without extra endoscopic or surgical management[7].

However, clinical outcomes after ultrasound-guided PPD in SAP differ widely between patients and the determinants of treatment success or failure are incompletely defined. Many clinical, biochemical, radiological, and procedural variables have been suggested to determine prognosis but mostly evidence is based on small monocentric studies with heterogeneous patient populations and outcome definitions[8]. However, especially the timing of the drainage concerning the onset of disease is still debated[9]. Likewise, the relative prognostic significance of collection characteristics like volume, location, presence of solid necrotic debris and proximity to large vessels have not been systematically assessed in relation to outcomes following percutaneous drainage[10].

Understanding the prognostic determinants of PPD outcomes is clinically relevant for multiple reasons. First, appropriate pre-procedural risk stratification can help guide patient selection for those most likely to benefit from early drainage intervention. Second, identification of high-risk features may lead to more rigorous post-drainage observation and earlier escalation of therapy in patients at risk for treatment failure. Third, prognostic markers potentially could help in making more individualized drainage planning decisions, such as size of catheter to select, drainage site optimization and/or whether multi-catheter treatment is appropriate[11]. Finally, a complete prognostic framework might help enable better conversation with patients and families on expected treatment courses and possible outcomes.

Thus, this retrospective study was designed to systematically identify and quantify the independent prognostic factors related with clinical outcomes after ultrasound-guided PPD in patients with SAP complicated by PFC. Herein, we aimed to provide such a framework by conducting rigorous univariate and multivariate analyses of a robust set of pre-procedural and procedural variables that could inform evidence-based clinical decision-making in these patients.

MATERIALS AND METHODS

Study design and patient selection

This study included SAP patients who underwent ultrasound-guided PPD for PFC at the Department of Emergency of Hangzhou First People’s Hospital from January 2019 to December 2024. The diagnosis of SAP was made according to the revised Atlanta Classification based on ≥ 1 criteria: Evidence of acute pancreatitis with persistent organ failure (≥ 48 hours) and/or local complications, such as pancreatic/peripancreatic necrosis. The protocol of this study was approved by the Hangzhou First People’s Hospital Ethics Committee, and informed consent was waived because of retrospective analysis. We screened all patients with PFC diagnosed on contrast-enhanced computed tomography (CT), who underwent ultrasound-guided PPD during the study period for eligibility.

Inclusion criteria: (1) Diagnosed SAP based on revised Atlanta Classification; (2) Symptomatic PFC confirmed on contrast-enhanced CT with clinical indications for drainage (fever, pain, progressive increase in the size of collection or evidences of infection); (3) Ultrasound-guided PPD as primary drainage procedure; (4) Age 18-80 years; (5) Complete pre-procedural clinical, biochemical and radiological data available; and (6) Minimum follow-up duration of 30 days or documented in-hospital outcome.

Exclusion criteria: (1) Acute pancreatitis classified as mild or moderately severe and without organ failure that was persistent; (2) Primary surgical or endoscopic drainage performed prior to a percutaneous attempt; (3) Incidental asymptomatic PFC, in the absence of clinical indication for drainage; (4) Coagulopathy that could not be corrected to international normalized ratio ≤ 1.5 and platelet count ≥ 50 × 10⁹/L; (5) Safe access route not being demonstrated via ultrasound; (6) Concurrent malignancy as the cause of pancreatitis; and/or (7) Incomplete procedural records or loss to follow-up within thirty days.

Drainage procedure protocol

Ultrasound-guided PPD was performed under real-time ultrasound guidance using high-resolution systems (Philips EPIQ Elite, Koninklijke Philips N.V., Amsterdam, The Netherlands or GE LOGIQ E10, GE Healthcare, Chicago, IL, United States) with convex transducer (2-5 MHz) by experienced interventional radiologists or gastroenterologist (> 5 years of PPD experience). All procedures were performed with conscious sedation or local anesthesia and analgesia onsite. The drainage pathway (retroperitoneal, transgastric-extraluminal or transabdominal) was chosen according to ultrasound and CT evaluation of collection characteristics and proximity with neighboring structures. Pigtail catheters (8-20 Fr, Cook Medical, Bloomington, IN, United States) were inserted using the Seldinger technique after confirming with aspiration that there was indeed fluid. Catheter size was based on consistency of collection: 8-12 Fr for primarily fluid collections and 14-20 Fr for viscous/necrotic collections. Routine post-procedure catheter irrigation with normal saline (3 × 50 mL/day) was performed. Catheters were removed if there was output < 10 mL/day for 48 consecutive hours, clinical symptom resolution and CT-confirmed collection reduction by > 80%. Catheter exchanges were performed on a clinical basis (i.e., confirmed occlusion unresponsive to flushing) with a guidewire-assisted technique without adjunctive intracavitary therapies (e.g., tissue plasminogen activator or N-acetylcysteine, except for 2 cases during this study period). Catheter upsizing or insertion of additional catheters was indicated by any ongoing insufficient reduction in collection size on subsequent imaging as well as refractory catheter blockade, or clinical deterioration indicating suspect inadequate source control. Depending on the time interval according to clinical findings such as fever, rising values of inflammatory markers and disappearance difficulty, repeat CT imaging was performed approximately at 7-day intervals or earlier.

Statistical analysis

All statistical analyses were conducted using R software (version 4.2.1). Continuous variables were reported as mean ± SD or median (interquartile range) based on data distribution evaluated by Shapiro-Wilk test. Data related to categorical variables were described in terms of n (%). Continuous variables between groups were compared using an independent t-test or Mann-Whitney U test and categorical variables with a χ² test or Fisher’s exact test. Candidate prognostic variables were selected at P < 0.10 threshold using univariate logistic regression. Multivariate logistic regression was used with backward elimination (P < 0.05 for retention) to enter significant variables. Results were reported as odds ratios (OR) with 95% confidence interval (CI). Model calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, and discrimination was estimated with receiver operating characteristic curve analysis. All inferential tests with a P < 0.05 were considered statistically significant.

RESULTS

Patient characteristics

We included a total of 134 SAP patients underwent ultrasound-guided PPD. Of these, 82 (61.2%) were categorized in the good-prognosis group and 52 (38.8%) in the poor-prognosis group according to the composite endpoint. The baseline characteristics are described in Table 1. Baseline characteristics of two therapy groups were significantly different regarding age, Acute Physiology and Chronic Health Evaluation (APACHE) II score, CT severity index (CTSI), C-reactive protein (CRP), procalcitonin, organ failure status (yes/no), volume of fluid collection and infected necrosis (all P < 0.05). Sex was male in most patients in both groups (64.6% vs 67.3%, P = 0.746). The most frequent etiologies were biliary (41.8%), hypertriglyceridemia (31.3%) and alcoholic (17.9%), with similar distributions between groups.

Table 1 Baseline characteristics of study population (n = 134), n (%)/mean ± SD/median (interquartile range).

Characteristic	Total (n = 134)	Good prognosis (n = 82)	Poor prognosis (n = 52)	P value
Age (years)	51.6 ± 13.4	49.2 ± 12.8	55.1 ± 13.9	0.009
Male gender	88 (65.7)	53 (64.6)	35 (67.3)	0.746
Etiology: Biliary	56 (41.8)	38 (46.3)	18 (34.6)	0.178
Etiology: Hypertriglyceridemia	42 (31.3)	24 (29.3)	18 (34.6)	0.512
Etiology: Alcoholic	24 (17.9)	14 (17.1)	10 (19.2)	0.753
APACHE II score	16.8 ± 5.2	14.3 ± 4.6	20.6 ± 4.8	< 0.001
CTSI	7.4 ± 1.9	6.8 ± 1.7	8.3 ± 1.8	< 0.001
CRP (mg/L)	248.6 (168.4-362.5)	198.3 (134.2-286.4)	318.7 (242.6-412.8)	< 0.001
Procalcitonin (ng/mL)	3.8 (1.2-9.6)	2.4 (0.8-5.8)	7.2 (3.4-15.6)	< 0.001
Organ failure at admission	68 (50.7)	32 (39.0)	36 (69.2)	< 0.001
Number of organ failures	1.4 ± 0.9	0.9 ± 0.8	2.1 ± 0.8	< 0.001
Fluid collection volume (mL)	368.5 (224.8-526.4)	312.6 (188.4-468.2)	458.3 (316.8-624.6)	0.002
Infected necrosis at drainage	48 (35.8)	18 (22.0)	30 (57.7)	< 0.001

APACHE: Acute Physiology and Chronic Health Evaluation; CRP: C-reactive protein; CTSI: Computed tomography severity index.

Drainage outcomes and overall prognosis

The total success rate of ultrasound-guided PPD was 71.6% (96/134). In-hospital mortality was 13.4% (18/134). The median time from diagnosis to first drainage was 68.4 hours (interquartile range: 42.8-102.6 hours). The median catheter dwell time was 14.2 days (interquartile range: 8.6-22.4 days). Among poor-prognosis patients, 14 (26.9%) required escalation to surgical necrosectomy and 20 (38.5%) to endoscopic necrosectomy after PPD failure. The composite poor-prognosis endpoint was driven by in-hospital mortality (18 cases, 34.6% of poor-prognosis), persistent organ failure (24 cases, 46.2%), and need for emergency surgical intervention (22 cases, 42.3%), with overlap between endpoints.

Univariate analysis of prognostic factors

Results of univariate analysis are presented in Table 2. Among clinical variables, APACHE II score ≥ 17 (OR = 7.48, 95%CI: 3.41-16.38; P < 0.001), multi-organ failure ≥ 2 organs (OR = 6.54, 95%CI: 2.93-14.60; P < 0.001) and confirmed infected necrosis (OR = 4.73, 95%CI: 2.19-10.21; P < 0.001) were the strongest univariate predictors of poor prognosis in our cohort[12]. CTSI ≥ 8 (OR = 5.38; P < 0.001) and collection volume ≥ 400 mL (OR = 3.68; P < 0.001), were also strongly associated with poor prognosis. Statistically significant predictors included: Procedure-related factors such as drainage < 72 hours (OR = 4.34; P < 0.001), catheter-size < 14 Fr (OR = 2.73; P = 0.008). Furthermore, other biomarkers such as CRP ≥ 300 mg/L (OR = 4.82; P < 0.001), serum albumin at level of < 30 g/L (OR = 3.49; P = 0.001) were significant in their impact on survival probability.

Table 2 Univariate analysis of prognostic factors, n (%).

Variable	Good prognosis (n = 82)	Poor prognosis (n = 52)	OR (95%CI)	P value
Age ≥ 55 years	26 (31.7)	28 (53.8)	2.51 (1.23-5.12)	0.011
APACHE II score ≥ 17	22 (26.8)	38 (73.1)	7.48 (3.41-16.38)	< 0.001
CTSI ≥ 8	24 (29.3)	36 (69.2)	5.38 (2.52-11.48)	< 0.001
CRP ≥ 300 mg/L at drainage	26 (31.7)	36 (69.2)	4.82 (2.28-10.19)	< 0.001
Infected necrosis confirmed	18 (22.0)	30 (57.7)	4.73 (2.19-10.21)	< 0.001
Drainage timing > 72 hours from diagnosis	28 (34.1)	36 (69.2)	4.34 (2.03-9.27)	< 0.001
Organ failure ≥ 2 systems	14 (17.1)	30 (57.7)	6.54 (2.93-14.60)	< 0.001
Catheter size < 14 Fr	22 (26.8)	26 (50.0)	2.73 (1.30-5.74)	0.008
Collection volume ≥ 400 mL	22 (26.8)	30 (57.7)	3.68 (1.73-7.82)	< 0.001
Walled-off necrosis type	30 (36.6)	32 (61.5)	2.76 (1.34-5.70)	0.006
Serum albumin < 30 g/L	18 (22.0)	26 (50.0)	3.49 (1.62-7.53)	0.001

Variables with P < 0.10 in univariate analysis were included in multivariate logistic regression. OR: Odds ratio; CI: Confidence interval; APACHE: Acute Physiology and Chronic Health Evaluation; CRP: C-reactive protein; CTSI: Computed tomography severity index.

Multivariate analysis of independent prognostic factors

Multivariate logistic regression analysis revealed eight independent factors predicting poor prognosis (Table 3). APACHE II score ≥ 17 emerged as the most potent independent predictor (adjusted OR = 5.82, 95%CI: 2.38-14.22; P < 0.001), and confirmed infected necrosis (OR = 4.16, 95%CI: 1.76-9.84; P = 0.001). Drainage timing > 72 hours after diagnosis was an independently significant modifiable prognostic factor (OR = 3.74; 95%CI: 1.58-8.86; P = 0.003). Radiological severity (CTSI ≥ 8) and multisystem organ failure were also independently significant (OR = 3.28, OR = 3.06, respectively). After adjustment, biochemical markers (CRP ≥ 300 mg/L; collection volume ≥ 400 mL; hypoalbuminemia) remained significant. The multivariate model provided good discrimination [area under the curve (AUC) = 0.892, 95%CI: 0.837-0.947] and calibration (Hosmer-Lemeshow P = 0.418, Figure 1).

Figure 1 Forest plot of independent prognostic factors for poor prognosis following ultrasound-guided percutaneous puncture drainage in severe acute pancreatitis with peripancreatic fluid collections. Diamond markers represent adjusted odds ratios; horizontal lines indicate 95% confidence intervals. The dashed vertical line at odds ratio = 1 denotes the null effect. Area under the curve of the multivariate model = 0.892 (95% confidence interval: 0.837-0.947). OR: Odds ratio; CI: Confidence interval; APACHE: Acute Physiology and Chronic Health Evaluation; CTSI: Computed tomography severity index; CRP: C-reactive protein.

Table 3 Multivariate logistic regression analysis of independent prognostic factors.

Independent variable	Adjusted OR	95%CI	P value
APACHE II score ≥ 17	5.82	2.38-14.22	< 0.001
Infected necrosis confirmed	4.16	1.76-9.84	0.001
Drainage timing > 72 hours from diagnosis	3.74	1.58-8.86	0.003
CTSI ≥ 8	3.28	1.38-7.80	0.007
Organ failure ≥ 2 systems	3.06	1.23-7.62	0.016
CRP ≥ 300 mg/L at drainage	2.89	1.18-7.07	0.020
Collection volume ≥ 400 mL	2.54	1.06-6.08	0.036
Serum albumin < 30 g/L	2.38	1.01-5.62	0.048

Area under the curve of multivariate model = 0.892 (95% confidence interval: 0.837-0.947); Hosmer-Lemeshow goodness-of-fit P = 0.418. OR: Odds ratio; CI: Confidence interval; APACHE: Acute Physiology and Chronic Health Evaluation; CRP: C-reactive protein; CTSI: Computed tomography severity index.

Outcomes stratified by drainage timing

Seventy (52.2%) of 134 patients were treated with early drainage (≤ 72 hours from diagnosis) and 64 (47.8%) were treated with late drainage (> 72 hours). Outcomes by time to drainage are summarized in Table 4. Early drainage was associated with markedly higher rates of clinical success (82.9% vs 59.4%, P = 0.004), and a shorter stay in the intensive care unit (ICU) (median 8.4 days vs 14.8 days, P < 0.001), as well as shorter total length of stay (24.6 days vs 36.8 days, P < 0.001) and higher organ failure resolution at day-7 (62.9% vs 37.5%, P = 0.004). The need for further drainage procedural interventions (20.0% vs 40.6%, P = 0.013), escalation to operative intervention (8.6% vs 28.1%, P = 0.006) and in-hospital mortality (5.7% vs 21.9%, P = 0.011) were all significantly less frequent among the early drainage group.

Table 4 Clinical outcomes stratified by drainage timing, n (%)/median (interquartile range).

Outcome	Early drainage ≤ 72 hours (n = 70)	Late drainage > 72 hours (n = 64)	OR/difference	P value
Clinical success rate	58 (82.9)	38 (59.4)	3.32 (1.47-7.50)	0.004
ICU stay (days)	8.4 (5.2-12.6)	14.8 (9.6-22.4)	-6.4	< 0.001
Total hospital stay (days)	24.6 (18.4-32.8)	36.8 (28.6-48.4)	-12.2	< 0.001
Organ failure resolution at 7 days	44 (62.9)	24 (37.5)	2.82 (1.38-5.76)	0.004
Need for additional drainage	14 (20.0)	26 (40.6)	0.37 (0.17-0.81)	0.013
Progression to surgical intervention	6 (8.6)	18 (28.1)	0.24 (0.09-0.67)	0.006
In-hospital mortality	4 (5.7)	14 (21.9)	0.22 (0.07-0.71)	0.011

Early drainage: ≤ 72 hours from diagnosis to first percutaneous drainage. Late drainage: > 72 hours from diagnosis. OR: Odds ratio; ICU: Intensive care unit.

Outcomes stratified by fluid collection type

Of the 134 patients, 58 (43.3%) had APFC or pseudocyst type collections and 76 (56.7%) were WON. Results stratified by collection type are shown in Table 5. WON was positively correlated with significantly worse outcomes for all of the measured endpoints. APFC/pseudocyst had a significantly higher clinical success rate compared with WON (89.7% vs 57.9%, P < 0.001). APFC/pseudocyst had significantly shorter days to clinical improvement (5.2 days vs 9.8 days, P < 0.001) and catheter dwell time (8.4 days vs 18.6 days, P < 0.001). Patients with WON exhibited increased rates of catheter occlusion (28.9% vs 6.9%, P = 0.001), superadded infection (28.9% vs 10.3%, P = 0.008), need for necrosectomy (26.3% vs 3.4%, P < 0.001) and in-hospital mortality (21.1% vs 3.4%, P = 0.003, Figure 2).

Figure 2 Clinical outcomes stratified by drainage timing and fluid collection type. A: Compares key outcomes between early drainage (≤ 72 hours, n = 70) and late drainage (> 72 hours, n = 64); B: Compares outcomes between acute peripancreatic fluid collection/pseudocyst (n = 58) and walled-off necrosis (walled-off necrosis, n = 76). P values above brackets indicate between-group significance. ^aP < 0.05. APFC: Acute peripancreatic fluid collection.

Table 5 Clinical outcomes stratified by fluid collection type, n (%)/median (interquartile range).

Outcome parameter	APFC/pseudocyst (n = 58)	WON (n = 76)	OR/difference	P value
Clinical success rate	52 (89.7)	44 (57.9)	5.82 (2.11-16.06)	< 0.001
Days to clinical improvement	5.2 (3.8-7.4)	9.8 (6.4-14.6)	-4.6	< 0.001
Catheter dwell time (days)	8.4 (5.6-12.8)	18.6 (12.4-26.8)	-10.2	< 0.001
Catheter blockage requiring exchange	4 (6.9)	22 (28.9)	0.19 (0.06-0.59)	0.001
Secondary infection	6 (10.3)	22 (28.9)	0.28 (0.10-0.77)	0.008
Need for necrosectomy	2 (3.4)	20 (26.3)	0.10 (0.02-0.45)	< 0.001
In-hospital mortality	2 (3.4)	16 (21.1)	0.13 (0.03-0.61)	0.003

APFC: Acute peripancreatic fluid collection; WON: Walled-off necrosis; OR: Odds ratio.

Drainage complications and procedural outcomes

Table 6 shows drainage-related complications and overall outcomes of the procedure. The total rate of complications was 43.3% (58/134), with catheter plugging (19.4%) and secondary infection (20.9%) being the most common ones. Pancreatic fistula developed in 11.9% of patients and had a conservative resolution rate of 81.3%. Severe complications like colonic fistula developed at 3.0% (4/134) and required surgical management while 75.0% resolved successfully. Hemorrhagic complications at the puncture site were noted in 6.0% (8/134) that were managed by conservative procedures (6/8) or transcatheter arterial embolization (2/8), demonstrating a 100% resolution rate. Catheter dislodgement was seen in 8.9% and was successfully managed by re-insertion in 91.7%.

Table 6 Drainage-related complications and procedural outcomes (n = 134).

Complication/outcome	n (%)	Management	Resolution rate
Catheter dislodgement	12 (8.9)	Re-insertion	11/12 (91.7%)
Catheter blockage	26 (19.4)	Flushing/exchange	24/26 (92.3%)
Bleeding at puncture site	8 (6.0)	Conservative/TAE	8/8 (100%)
Secondary infection	28 (20.9)	Antibiotics adjustment	22/28 (78.6%)
Pancreatic fistula	16 (11.9)	Conservative	13/16 (81.3%)
Colonic fistula	4 (3.0)	Surgical repair	3/4 (75.0%)
Overall complication rate	58 (43.3)	-	-
Overall clinical success rate	96 (71.6)	-	-
In-hospital mortality	18 (13.4)	-	-

Clinical success defined as collection resolution without need for emergency surgical intervention or in-hospital mortality attributable to peripancreatic fluid collections-related complications. TAE: Transcatheter arterial embolization.

DISCUSSION

This study provides a comprehensive assessment for prognostic factors with clinical outcomes after ultrasound-guided PPD in SAP patients complicated with PFC. Our data suggest that disease severity based on APACHE II score, infected necrosis and drainage timing are the most robust independent prognostic indicators; type of collection and radiological severity also offer independent contributions to predicting outcome. The clinical success rate of 71.6% and in-hospital mortality of 13.4% in our cohort are concordant with the published literature regarding PPD for SAP, which suggests that our results can be generalized[12,13]. This is one of the most exhaustive analyses to date that specifically addresses prognostic factors for ultrasound-guided PPD in SAP and offers a nuanced framework for prognostication that integrates clinical, biochemical, radiological and procedural variables.

Our multivariate model results in good (-excellent) discriminative performance (AUC = 0.892), on par with established SAP prognostic tools previous publications indicate AUC range of 0.74-0.82 for the score of Bedside Index for Severity in Acute Pancreatitis and 0.78-0.85 for organ failure based on the Modified Marshall score as predictors of severity in SAP. The Panc 3 score also shows poor discriminative ability in this setting. This marginal improvement is mechanistically plausible: Traditional admission-based tools were constructed to predict overall SAP severity at presentation and not intended to account for procedural-level determinants of drainage outcome. Finally, collection-specific variables including collection volume ≥ 400 mL, WON morphology and CTSI ≥ 8 provided additional predictive signal directly relevant to the PPD setting that is absent from these tools. The absence of a formal within-dataset head-to-head comparison is acknowledged, as previous established scores were not prospectively calculated in all patients during drainage; this comparison should be considered only an illustrative benchmark rather than the claim of superiority, and future prospective studies must include parallel calculation of established scores to allow direct validation.

APACHE II score ≥ 17 stood out as the strongest independent predictor of poor prognosis, with an adjusted OR = 5.82 and profound physiological disturbances furthering multiorgan dysfunction as measured by this scoring system in critically ill patients previously established[14]. APACHE II has been widely validated as a predictor of mortality and complications in SAP management context, and thus its findings extend also to the prediction of outcomes in PPD[15]. It is also likely that patients with high APACHE II score are those with the highest burden of systemic inflammation and the lowest physiological reserve, which renders them unable to recover even after successful local source control through drainage. Thus, it implies that in experienced hands and with a considerable PPD diagnostic workup of APACHE II ≥ 17 patients, PPD must be considered as an adjunctive therapeutic tool integrated into an overall treatment strategy rather than a curative one, with close cooperation between interventional driven approaches and critical care practice.

Confirmed infected necrosis as an independent predictor of poor prognosis (OR = 4.16) highlights the known relationship between secondary bacterial infection of pancreatic and peripancreatic necrosis and worse outcomes in SAP[16]. Infected pancreatic necrosis (IPN) serves as the main indication for drainage intervention in SAP, yet paradoxically it also classifies a change to a clinically more severe disease phase during development. The solid component of debris found in IPN makes percutaneous catheter drainage alone less than effective, leading to larger-bore catheters, multiple or more frequent exchanges of the catheter and often progression to endoscopic or surgical necrosectomy[17]. In our cohort, confirmed infected necrosis was found in 57.7% of patients with poor prognosis and 22.0% of patients with good prognosis (P < 0.01), reinforcing the key prognostic importance of this complication. This finding should lead to earlier recognition of IPN based on blood cultures, diagnostic aspiration and imaging features (like presence of gas in the collection at CT) with a more aggressive approach to drainage and careful monitoring for treatment failure.

In multivariate analysis, we found that drainage timing was an independent prognostic factor (OR = 3.74) for (> 72 hours late) which adds significant data to the long term debate on best drain timing in SAP patients. The therapeutic impact of early drainage is mechanistically credible: Appropriate decompression of infected or symptomatic collections diminishes bacterial load, curtails ongoing systemic inflammatory provocation and may even avoid progression to organ dysfunction[18]. Our timing analysis showed that early drainage (≤ 72 hours) was associated with a much superior overall outcome in all studied endpoints, including 82.9% vs 59.4% clinical success rate, shorter ICU and hospital stays and lower in-hospital mortality. However, as the timing of drainage was not randomized, it is noteworthy that those who underwent early drainage may reflect a selected population with more favorable characteristics (e.g., better delineated collections more amenable to drainage). Confounding by clinical indication is intrinsic to this retrospective analysis, and it will take confirmation by large prospective randomized trials to establish definitively the causal effect of drainage timing. However, our results uphold the tenet that once clinical and radiological signs indicate drainage, intervention should not be postponed unnecessarily.

The independent prognostic factor for CTSI ≥ 8 (OR = 3.28) demonstrates the crucial step for assessing based on imaging to dictate drainage outcome. These were combined in the CTSI, which integrates the Balthazar grade and extent of necrosis score to provide a standardized measure of CT-based disease severity in pancreatitis[19]. Elevated CTSI scores correlate with an increase in peripancreatic fat stranding, size of fluid collection and percentage of pancreatic necrosis, all which have been correlated with increased rates of infected necrosis or multiorgan failure, as well as the requirement for invasive therapy. Excess necrosis in the context of drainage is associated with more complex anatomy to collect, increased viscosity and multiple or multiloculated collections requiring increasingly aggressive drainage strategies[20]. The independent prognostic relevance of CTSI in our model indicates that pre-procedural CT assessment should be performed routinely, not only to guide catheter positioning but also to ensure prognostic stratification.

The relationship that multisystem organ failure (≥ 2 systems) was a statistically significant independent predictor for poor prognosis (OR = 3.06) is in accordance with the essential pathophysiology of SAP, as it presents organ dysfunction correlated with severity of systemic inflammation and strictly interfere with survival[21]. Importantly, SAP with multiorgan failure can precede and complicate PFC drainage procedures where hemodynamic instability zaps perfusion to the kidney; impairs healing of the wound while inverting a competent immune response against secondary infections. In concurrent multiorgan failure patients, PPD is to be performed in an urgent scenario to cut off the septic sources along with comprehensive organ support managed in the ICU. Our findings suggest that multiorgan failure refractory to fluid resuscitation at the time of drainage should be identified as a high-risk feature for post-procedural complications, and patients with this finding should have a lower threshold for post-drainage escalation in intervention.

The significantly poorer outcomes seen in WON compared to APFC/pseudocyst collections (89.7% vs 57.9% success rate) support the existing knowledge that WON is a more complicated and treatment-resistant entity[22]. Then, limited effective drainage by standard catheters is a characteristic with heterogeneous contents such as liquefied necrotic debris and solid fragments including fat necrosis[23] and consequently requires repeated exchanges or multi-catheter approaches. The longer catheter dwell-time (18.6 days vs 8.4 days) and higher rates of obstructed catheters (28.9% vs 6.9%) in WON patients mechanistically explain the worse outcomes[24]. Our findings suggest that WON patients have significantly poorer outcomes with PPD alone compared to patients with APFC or pseudocysts. These results support the need for prospective comparative studies evaluating whether endoscopic ultrasound-guided transmural drainage, which permits direct mechanical debridement, might offer superior outcomes for WON specifically. In clinical practice, high suspicion of WON on CT and preparation for possible step-up drainage should be standard of care.

The prognostic importance of high CRP (OR = 2.89) and hypoalbuminemia (OR = 2.38) in our multivariate model indicates the wider nutritional and immunological backdrop of patient ability to respond to drainage intervention. CRP is a surrogate of systemic inflammation that correlates with disease activity and has been proved to be a prognostic marker in SAP[25]. Albumin, a nutritional indicator, also reflects the acute-phase response, capillary leak and hepatic synthetic function that are impaired in severe SAP[6]. Low albumin value at the time of drainage may represent a state of generalized catabolism and immune suppression that makes patients susceptible to treatment failure and secondary infection. These findings hold clinical implications for pre-procedural optimization: Targeted nutrition adjunctive to PPD and peri-operative albumin supplementation may potentially favor hyperdynamic responses in patients with hypoalbuminemia, however, this finding must be prospectively evaluated.

As would be expected for PPD in SAP, the overall complication rate of 43.3% seen in our series is high but consistent with published experience[26]; this also reflects the severity of the underlying disease. Importantly, the majority of complications were managed conservatively or with minimally invasive procedures, and serious irreversible events occurred in only a small minority of patients. The incidence of pancreatic fistula (11.9%) is within the range of previously described studies on similar patients and almost exclusively managed conservatively with prolonged catheter drainage alone or somatostatin analogs as per natural history of pancreatic fistulas[27]. The colonic fistula rate of 3.0% is a known but rare complication of inadvertent puncture of neighboring colonic structures or pressure-related necrosis, and emphasizes the need for careful route planning with preprocedural CT and real-time ultrasound guidance.

Acknowledge several study limitations. This are inherent selection and information biases due to the retrospective design, coupled with the observational nature of timing of drainage means that causation effects cannot be deduced. Our center-specific results may not be entirely generalizable to centers with differing procedural expertise, patient demographics or drainage protocols as this was a single-center study. The composite poor-prognosis endpoint is clinically meaningful but reflects heterogeneous events that might have different prognostic factors. Furthermore, the absence of universally accepted criteria defining when to escalate therapy from percutaneous catheter drainage to endoscopic or surgical necrosectomy introduce variability in the threshold for escalation that can potentially confound comparisons of outcomes. The sample size, which is relatively modest though adequate for multivariate analysis, has implications for the precision of estimates among subgroups especially in less common outcomes. These limitations need to be addressed by future studies, which should ideally follow a prospective multicenter design with standardized protocols. Secondly, disconnection of the pancreatic duct syndrome (DPDS) was not systematically evaluated in this study since dedicated imaging with magnetic resonance cholangiopancreatography or endoscopic retrograde cholangiopancreatography was not consistently done as part of pre-PPD assessment in our emergency cohort. Definitive pancreatic duct disruption (DPDS) due to the necrosis of a segment of the main pancreatic duct and leading to continuous leak from intact upstream viable parenchyma is an important predictor of failed percutaneous drainage and recurrent PFCs. It is also conceivable that unmeasured confounding played a role in the population with extended drainage needs or recurrent collections, as there was no systematic DPDS evaluation. Future studies should include regular use of pre-procedural DPDS assessment as this could be used to formally quantify its independent prognostic value in patients with SAP for PPD outcome.

Future directions should be prospective multicenter validation of the prognostic model described herein, aimed at establishing a validated clinical decision tool for drainage planning. To promote prompt translation into practice, a point-scoring system is formulated based on the β-coefficients of the multivariate logistic regression model, assigning integer scores to each of the eight independent predictors proportionately according to their adjusted ORs and stratifying patients into low-, intermediate- or high-risk categories for poor prognosis. An accompanying nomogram, which allows for personalized risk assessment at the bedside before PPD can be performed, is created and shown in an additional figure. External prospective validation of both tools is necessary prior to incorporation in clinical practice. Further trials, randomized controlled or otherwise, examining optimal time for drainage, selection of catheter size and choice of irrigation protocols are needed to strengthen the evidence. The addition of new biomarkers such as procalcitonin trajectory, interleukin-6, and imaging features (radiomics) may improve predictive accuracy even more[28,29]. The relative efficacy of ultrasound-guided PPD as compared to endoscopic ultrasound-guided drainage, especially in patients with WON, remains a subject of ongoing investigation[24,30]. Finally, artificial intelligence-based CT imaging analysis for automatic CTSI computation and collection content characterization could improve pre-procedural decision making and minimize operator-dependent variations.

CONCLUSION

Administration of ultrasound-guided PPD to treat PFCs yielded a 71.6% clinical success rate, demonstrating its safety and effectiveness for the treatment of SAP. The independent predictors of poor prognosis were APACHE II score ≥ 17, confirmed infected necrosis, delayed drainage (timing > 72 hours from diagnosis), CTSI ≥ 8, multisystem organ failure, CRP levels ≥ 300 mg/L, collection volume > 400 mL and hypoalbuminemia. WON type collections have significantly worse outcomes than APFC/pseudocyst per each of the clinical endpoints. Earlier drainage within 72 hours will achieve significant clinical benefits such as better success rate, less ICU and hospital stay, and lower in-hospital mortality. These results offer a comprehensively evidence-based prognostic structure to facilitate individualized drainage decision-making, risk stratification, and monitoring intensity in SAP patients with PFC. Applying this prognostic model of care in real-world clinical practice could improve early identification of these high-risk patients who may benefit from more intense drainage therapies or a primary endoscopic or surgical approach, resulting in improved patient outcomes.