Type 1 autoimmune pancreatitis: clinical features and independent predictors of histopathological confirmation via EUS-guided fine-needle aspiration/fine-needle biopsy

Abstract

Background

In diagnosing type 1 autoimmune pancreatitis (AIP), serum IgG4 (sIgG4) can be false-negative. EUS-guided fine-needle aspiration/biopsy (EUS-FNA/FNB) pathology is key for diagnosis, but clinical features’ impact on pathologic confirmation is unclear. This study analyzed their link and factors improve diagnostic accuracy.

Methods

We analyzed data from a single-center retrospective study at Changhai Hospital (Jan 2009-Jan 2024). Type 1 AIP was diagnosed per International Consensus Diagnostic Criteria (ICDC). Patients with surgical diagnosis, no EUS, or incomplete biopsy data were excluded; eligible cases were grouped into “Confirmed”/“Unconfirmed” per ICDC. Baseline data, laboratory indicators, imaging, and EUS-FNA/FNB data were collected. Statistical analyses (ROC, χ² tests, multivariate logistic regression) were done with R 4.4.0.

Results

A total of 182 suspected type 1 AIP patients were enrolled; 84.07% were male, 88.46% middle-aged/elderly. Common symptoms: abdominal discomfort (65.93%), obstructive jaundice (43.41%). sIgG4 > 2×ULN (twice upper normal limit) occurred in 64.84%. Multivariate analysis: pathological confirmation rate 65.12% (EUS-FNB) vs. 18.75% (EUS-FNA) (P < .001, former higher). For IgG4-positive cells: 82.56% confirmation rate (> 10 cells/high-power field[HPF]) vs. 16.67% (< 10 cells/HPF) (P < .001). EUS-FNB (OR = 3.56, 95% CI: 1.55–8.18, P = .003) and IgG4-positive cell count (> 10 cells/HPF) (OR = 15.71, 95% CI: 6.96–35.46, P < .001) were independent confirmation predictors. Gender, age, sIgG4 had limited value; renal involvement, retroperitoneal fibrosis were auxiliary indicators.

Conclusions

Following systematic multi-dimensional factor screening, pathological confirmation of type 1 AIP relies on two key factors: EUS-FNB and histopathological detection of IgG4-positive cells (> 10 cells/HPF). Integrating these core diagnostic modalities with additional indicators—such as auxiliary markers of extrapancreatic involvement (e.g., renal involvement)—further enhances diagnostic precision, which facilitates the refinement of clinical diagnostic workflows for type 1 AIP.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12876-026-04694-9.

Introduction

In 1995, Yoshida et al. [1] first proposed the concept of autoimmune pancreatitis (AIP). AIP has two main subtypes—type 1 (lymphoplasmacytic sclerosing pancreatitis, LPSP, linked to IgG4-related disease [IgG4-RD]) and type 2 (idiopathic duct-centric pancreatitis, IDCP)—with some scholars recently suggesting a type 3 (unclassified AIP) [2, 3]. Type 1 AIP, linked to IgG4 - related disease (IgG4 - RD), features periductal lymphoplasmacytic infiltration, storiform fibrosis, and obliterative venulitis [2, 4, 5]. Type 2 AIP is marked by granulocyte epithelial lesions [2, 6, 7]. Type 1 AIP is more common in Asia, while type 2 prevails in Western countries, and its prevalence has been on the rise [8–14].

The 2011 International Consensus Diagnostic Criteria (ICDC) guides AIP diagnosis, integrating imaging, serum IgG4 (sIgG4), extrapancreatic involvement, histopathology, and steroid response for type 1 AIP [15]. Histopathology (e.g., duct-centric inflammation, storiform fibrosis) is pivotal, though final diagnosis requires multi-data integration [16–19].

Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) and biopsy (EUS-FNB) are mainstream for pancreatic histology [20–24]. Previous studies show EUS-FNB and 19G needles improve tissue acquisition and diagnostic accuracy [20, 21, 23, 24], but no reports have addressed how clinical features differ between histopathologically confirmed and unconfirmed type 1 AIP groups (per ICDC) after EUS-FNA/FNB—a gap that limits clinicians’ ability to optimize sampling strategies for better diagnostic outcomes.

This study focuses on patients suspected of type 1 AIP who underwent EUS-guided sampling. It aims to systematically analyze a range of relevant factors in these patients, including demographics, clinical symptoms, laboratory indicators, imaging findings, and parameters related to EUS-FNA/FNB procedures. Patients are stratified based on their histopathological confirmation status (per ICDC) [15], with subsequent comparison of factor variations between confirmed and unconfirmed groups, and identification of independent predictors associated with histopathological confirmation. Ultimately, this study seeks to provide a comprehensive reference for clinical decision-making—such as selecting individualized sampling methods—and support improvements in the histopathological diagnostic rate for type 1 AIP.

Methods

Patients and study design

This retrospective study analyzed 297 patients diagnosed with type 1 AIP at Changhai Hospital (Shanghai, China) from January 2009 to January 2024. The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Changhai Hospital (Ethics No.: CHEC2025-403). As a retrospective study, the requirement for informed consent was waived, and patient anonymity was strictly maintained throughout the study with no data posing a risk of patient identification.

All included cases met the ICDC [15], which integrates pancreatic imaging findings, serological indicators, extrapancreatic involvement, histopathological features, and response to steroid therapy. Patients were excluded if they had interfering diseases (e.g., cholangiocarcinoma, pancreatic cancer), insufficient pancreatic tissue samples for histopathological analysis, a final diagnosis based on surgical pathology (rather than EUS-guided sampling), no EUS examination, or EUS examination without subsequent EUS-FNA or EUS-FNB. All enrolled patients underwent routine hemostatic function screening prior to the procedure. No standardized pre-procedural interventions (e.g., antiplatelet/anticoagulant discontinuation, coagulopathy correction) were uniformly applied to all patients. For those with mild coagulation dysfunction (not meeting the exclusion criteria), EUS-FNA was selected after clinical risk assessment due to its relatively lower invasiveness, without targeted coagulation correction. For patients without coagulation abnormalities, the choice of puncture method was based on clinical practice, with no differential pre-procedural management between the two groups.

Enrolled patients were divided into two groups based on EUS-FNA/FNB histopathological results: “Confirmed diagnosis” (histopathologically confirmed type 1 AIP) and “Unconfirmed diagnosis” (suspected type 1 AIP or insufficient evidence for diagnosis). We compared baseline characteristics, laboratory results, pancreatic/extrapancreatic imaging findings (computed tomography [CT]/magnetic resonance imaging [MRI]/EUS), and histopathological diagnosis-related factors between the two groups, and analyzed the diagnostic efficacy of clinical features and factors associated with histopathological confirmation.

Laboratory tests

Baseline characteristics: gender, age, admission symptoms, smoking/drinking/family histories, diabetes mellitus, hypertension. Age stratified into young (18–44 years), middle-aged (45–59 years), elderly (≥ 60 years). Smoking history (tobacco abuse): ≥5 cigarettes/day [25]; drinking history: ≥50 g alcohol/day [26]. Family history: immune/pancreatic diseases (e.g., type 1 AIP, IgG4-related disease [IgG4-RD], Sjögren’s syndrome) in first-degree (e.g., parents, children) or second-degree (e.g., grandparents) relatives.

Laboratory tests included liver function and tumor markers; key indicators: sIgG4 (< 2.0 g/L), total bilirubin (TB, ≤ 23 mmol/L), alanine aminotransferase (ALT, ≤ 50 U/L), aspartate aminotransferase (AST, ≤ 40 U/L), CA19-9 (< 37 U/mL). sIgG4 stratified by AIP criteria [15]: <1×upper limit of normal (ULN), 1–2×ULN, > 2×ULN. All tests followed Changhai Hospital’s standard clinical protocols.

Imaging examinations

Imaging data included CT (plain/enhanced), MRI (plain/enhanced), and EUS. For CT/MRI, the more detailed/accurate results were used (available data if lacking either); both scan types were evaluated. EUS used Olympus EU-ME1/EU-ME2, Aloka Alpha-5, or Olympus UMG20-29R (20 MHz). In accordance with the ICDC [15], pancreatic parenchyma/duct (categorized by lesions) and extrapancreatic involvement were assessed, with rare/unclear findings grouped as “Unknown.” All evaluations were jointly conducted by Changhai Hospital’s radiologists (≥ 10 years of experience). Interrater consistency was quantified using the kappa statistic, yielding a kappa value of 0.82 (excellent agreement).

Histopathological diagnosis of Type 1 AIP

Histopathological samples were obtained via EUS-FNA/FNB by endoscopists (8–30 years of experience, ≥ 1000 annual EUS procedures) using needles from Wilson-Cook, Boston Scientific, Medi-Globe, Olympus, and Cook Medical. Tissues were formalin-fixed, paraffin-embedded, sectioned, and subjected to HE/immunohistochemical staining. Diagnostic criteria [15]: (1) periductal lymphoplasmacytic infiltration with fibrosis (no granulocytes); (2) storiform fibrosis; (3) obliterative venulitis; (4) > 10 IgG4-positive cells/high-power field[HPF]. For the counting of IgG4-positive cells: Under high-power field (×400), the “hot spot” (area with the highest density of IgG4-positive cells) was first identified, then the number of positive cells was counted in 3 consecutive non-overlapping HPFs within this hot spot, and the average value was used to determine whether the threshold (> 10 cells/HPF) was met. Cases meeting ≥ 3 criteria were classified as Level 1, and those meeting ≥ 2 as Level 2; both Level 1 and 2 were considered histopathologically confirmed, while cases judged as “suspected” or “insufficient evidence” were unconfirmed. Interpreted by pathologists (> 10 years of experience); discrepancies resolved by senior pathologists (> 20 years). Interrater consistency was quantified using the kappa statistic, yielding a kappa value of 0.82 (excellent agreement).

Statistical analysis

Statistics, charts, and figures were generated using R software (Version 4.4.0; The R Foundation, Vienna, Austria). Categorical variables: n (%) (frequency/percentage). ROC curves and AUC assessed clinical features’ diagnostic efficacy for distinguishing confirmed/unconfirmed cases. Categorical inter-group differences: χ² test or Fisher’s exact test (per data distribution). Multivariate logistic regression identified confirmation-linked features; only variables with P < .100 in univariate analysis were included. P-values adjusted by Bonferroni correction (two-tailed test); P < .050 was statistically significant.

Results

Baseline characteristics of patients with Type 1 AIP

A total of 297 patients initially clinically diagnosed with type 1 AIP were screened for eligibility. 20 patients were excluded due to final diagnosis via surgical pathology (instead of EUS-guided sampling), 57 patients for not undergoing EUS, and 38 patients as they had EUS but no subsequent EUS-FNA/FNB or incomplete histopathological data. Finally, 182 patients were enrolled, with baseline characteristics summarized in Table 1.

Table 1.

Clinical features of the study patients (n = 182)

Clinical Features	Overall
Gender, n (%)
Male	153 (84.07%)
Female	29 (15.93%)
Age(years), n (%)
18–44	21 (11.54%)
45–59	72 (39.56%)
≥60	89 (48.90%)
Symptoma, n (%)
Abdominal discomfort	120 (65.93%)
Obstructive jaundice	79 (43.41%)
Abdominal discomfort and Obstructive jaundice	35 (19.23%)
Weight loss	13 (7.14%)
Diarrhea	11 (6.04%)
Asymptomatic	5 (2.75%)
Diabetes mellitus, n (%)	48 (26.37%)
Hypertension, n (%)	30 (16.48%)
Family history of pancreatic disease, n (%)	1 (0.55%)
Smoking history, n (%)	68 (37.36%)
Drinking alcohol history, n (%)	46 (25.27%)
sIgG4, n (%)
<1×ULN	33 (18.13%)
1 ~ 2×ULN	31 (17.03%)
>2×ULN	118 (64.84%)
TB(µmol/L), n (%)
≤23	97 (53.30%)
>23	85 (46.70%)
ALT(U/L), n (%)
≤50	96 (52.75%)
>50	86 (47.25%)
AST(U/L), n (%)
≤40	95 (52.20%)
>40	87 (47.80%)
CA19-9(U/mL), n (%)
<37	139 (76.37%)
≥37	43 (23.63%)
CT/MRI-PP, n (%)
Diffuse enlargement	124 (68.13%)
Focal mass	54 (29.67%)
Basically normal	4 (2.20%)
CT/MRI-PD, n (%)
Long or multiple narrow sections	4 (2.20%)
Focal main pancreatic duct dilation	40 (21.98%)
Basically normal	80 (43.96%)
Unknown	58 (31.87%)
EUS-PP, n (%)
Diffuse enlargement	125 (68.68%)
Focal mass	52 (28.57%)
Basically normal	5 (2.75%)
EUS-PD, n (%)
Long or multiple narrow sections	9 (4.95%)
Focal main pancreatic duct dilation	37 (20.33%)
Basically normal	119 (65.38%)
Unknown	17 (9.34%)
Diameter size of puncture needles, n (%)
19G	13 (7.14%)
22G	146 (80.22%)
25G	23 (12.64%)
Puncture type, n (%)
EUS-FNA	108 (59.34%)
EUS-FNB	74 (40.66%)
Pathology diagnosis, n (%)
Unconfirmed diagnosis	96 (52.75%)
Confirmed diagnosis	86 (47.25%)
Pathology, n (%)
Suspected type 1 AIP	42 (23.08%)
Level 1 histopathological evidence for type 1 AIP	37 (20.33%)
Level 2 histopathological evidence for type 1 AIP	49 (26.92%)
Insufficient histopathological evidence for type 1 AIP	54 (29.67%)
IgG4 cells/HPF, n (%)
<10 cells/HPF	95 (52.20%)
>10 cells/HPF	87 (47.80%)
Other organ involvementb, n (%)
Biliary stricture	83 (45.60%)
Retroperitoneal fibrosis	7 (3.85%)
Kidney	15 (8.24%)
Other manifestations or no involvement	85 (46.70%)

Abbreviations: ULN upper limit of normal, AIP autoimmune pancreatitis, PP pancreatic parenchyma, PD pancreatic duct, EUS-FNA EUS-guided fine-needle aspiration, EUS-FNB EUS-guided fine-needle biopsy, HPF high-power field

^aSymptoms are non-mutually exclusive (patients may have multiple concurrent symptoms). “Abdominal discomfort and Obstructive jaundice” denotes coexistence of these two; other categories may also overlap. Thus, counts may exceed total sample size, which is reasonable

^bOrgan involvement categories are non-mutually exclusive (patients may have concurrent multiple organ involvement), so the sum of counts may exceed the total sample size, which is reasonable

Enrolled patients included 153 males (84.07%) and 29 females (15.93%), with 88.46% being middle-aged or elderly; the most common clinical symptoms were abdominal discomfort (65.93%) and obstructive jaundice (43.41%), and 19.23% had both symptoms, while the proportions of patients with diabetes, hypertension, smoking, and alcohol consumption histories were 26.37%, 16.48%, 37.36%, and 25.27%, respectively (only 1 patient had a relevant family history).

In terms of laboratory indicators, 64.84% of patients had sIgG4 levels exceeding twice the upper limit of normal (2×ULN). The abnormal rates of TB, ALT, AST, and CA19-9 were 46.70%, 47.25%, 47.80%, and 23.63%, respectively.

For imaging findings, pancreatic parenchyma with “diffuse enlargement” was common in CT/MRI (68.13%) vs. EUS (68.68%). For pancreatic duct changes, “long-segment/multiple-segment stenosis” was observed in CT/MRI (2.20%) vs. EUS (4.95%), while “focal main pancreatic duct dilatation” was noted in CT/MRI (21.98%) vs. EUS (20.33%).

In terms of tissue sampling methods, EUS-FNA was performed in 59.34% of patients vs. EUS-FNB in 40.66%, with 22G needles most commonly used (80.22%).Histopathologically, 47.25% were confirmed with type 1 AIP (Level 1: 20.33% vs. Level 2: 26.92%), and IgG4-positive cells were < 10 cells/HPF in 52.20% vs. >10 cells/HPF in 47.80%.

For extrapancreatic organ involvement, the bile duct was the most frequently affected (45.60%), followed by the kidney (8.24%) and retroperitoneal fibrosis (3.85%).

Correlation between different clinical features and histopathological diagnosis

Bar charts were used to visualize the relationship between the clinical features of type 1 AIP and histopathological diagnosis status (“Unconfirmed diagnosis” vs. “Confirmed diagnosis”), with results shown in Fig. 1 and Fig. S1.

Fig. 1.

Correlation between clinical features and histopathological diagnosis status in patients with type 1 AIP. Stacked bar charts show the proportion of histopathologically “Confirmed diagnosis” (teal) and “Unconfirmed diagnosis” (red) cases across different features. A Gender distribution; B Age stratification (years); C Presence of obstructive jaundice; D Presence of abdominal discomfort combined with obstructive jaundice; E Stratified sIgG4 levels relative to the ULN; F ALT levels; G AST levels; H TB levels; I CA19-9 levels; J EUS-PP features; K EUS-PD features; L Puncture type (EUS-FNA vs. EUS-FNB); M Diameter size of puncture needles (19G, 22G, 25G); N Abundance of IgG4-positive cells per HPF; O Presence of retroperitoneal fibrosis; P Presence of AIP-related renal involvement. (Abbreviations: AIP: autoimmune pancreatitis; sIgG4: serum IgG4; ULN, upper limit of normal; PP: pancreatic parenchyma; PD: pancreatic duct; EUS-FNA: EUS-guided fine-needle aspiration; EUS-FNB: EUS-guided fine-needle biopsy; HPF: high-power field.)

Gender (Fig. 1A), age (Fig. 1B), obstructive jaundice (Fig. 1C), and the combination of abdominal discomfort with obstructive jaundice (Fig. 1D) had low discriminative power for distinguishing histopathologically confirmed from unconfirmed cases; similarly, isolated abdominal discomfort (Fig. S1A), weight loss (Fig. S1B), diarrhea (Fig. S1C), and asymptomatic status (Fig. S1D) also failed to differentiate the two groups. Medical histories including diabetes mellitus (Fig. S1E), hypertension (Fig. S1F), smoking (Fig. S1G), and alcohol consumption (Fig. S1H) showed no obvious indicative effect on histopathological confirmation, and only 1 patient had a family history of pancreatic disease (Fig. S1I), making it impossible to analyze this factor due to uneven case distribution.

For serum biomarkers, stratified sIgG4 levels (Fig. 1E) revealed that the proportion of histopathologically confirmed cases was relatively high in the group with sIgG4 > 2×ULN, whereas other serum indicators including ALT (Fig. 1F), AST (Fig. 1G), TB (Fig. 1H), and CA19-9 (Fig. 1I) had no obvious indicative effect on histopathological confirmation.

Regarding pancreatic imaging features, pancreatic parenchyma evaluated by EUS (Fig. 1J) showed no significant difference in the proportion of confirmed vs. unconfirmed cases, consistent with findings from CT/MRI (Fig. S1J). In contrast, pancreatic duct features assessed by EUS (Fig. 1K) showed a weak indicative effect on confirmation (especially for long-segment/multiple-segment stenosis), which was also observed in CT/MRI evaluations (Fig. S1K).

In terms of puncture-related factors, samples obtained via EUS-FNB (Fig. 1L) had a significantly higher proportion of clear type 1 AIP diagnoses than those via EUS-FNA, supporting EUS-FNB as the preferred method for improving diagnostic efficiency. Puncture needle diameter size (Fig. 1M) had no strong discriminative effect, but the diagnostic accuracy appeared to decrease from 19G to 22G to 25G needles.

Histopathologically, abundant IgG4-positive cells (> 10 cells/HPF) (Fig. 1N) — a key indicator of type 1 AIP — was significantly more common in confirmed cases, and all confirmed cases were classified as Level 1 or Level 2 (Fig. S1L). Cases with retroperitoneal fibrosis (Fig. 1O) or AIP-related renal involvement (Fig. 1P) had a relatively high proportion of clear diagnoses, suggesting these complications can serve as auxiliary diagnostic evidence. In contrast, cases with local/multiple high-grade biliary strictures or combined distal biliary strictures (Fig. S1M) showed no significant difference in the proportion of confirmed vs. unconfirmed diagnosis.

Differences in clinical features between histopathologically confirmed and unconfirmed groups

Differences in clinical features (including demographics, clinical symptoms, laboratory indicators, imaging findings, and puncture/pathological data) between the “Unconfirmed diagnosis” and “Confirmed diagnosis” groups were analyzed, with detailed results shown in Table 2 and Table S1.

Table 2.

Comparison of clinical features of “Unconfirmed diagnosis ” and “ Confirmed diagnosis”

Variables	Unconfirmed diagnosis (n = 96)	Confirmed diagnosis (n = 86)	p
Gender, n (%)			0.371
Male	78 (81.25%)	75 (87.21%)	0.371
Female	18 (18.75%)	11 (12.79%)	0.371
Age(years), n (%)			0.291
18–44	9 (9.38%)	12 (13.95%)	0.464
45–59	35 (36.46%)	37 (43.02%)	0.450
≥60	52 (54.17%)	37 (43.02%)	0.176
Symptoma, n (%)
Abdominal discomfort and Obstructive jaundice	24 (25.00%)	11 (12.79%)	0.058
sIgG4, n (%)			0.401
<1×ULN	19 (19.79%)	14 (16.28%)	0.673
1 ~ 2×ULN	19 (19.79%)	12 (13.95%)	0.396
>2×ULN	58 (60.42%)	60 (69.77%)	0.245
TB (µmol/L), n (%)			0.487
≤23	54 (56.25%)	43 (50.00%)	0.487
>23	42 (43.75%)	43 (50.00%)	0.487
ALT(U/L), n (%)			0.580
≤50	53 (55.21%)	43 (50.00%)	0.580
>50	43 (44.79%)	43 (50.00%)	0.580
AST(U/L), n (%)			0.192
≤40	55 (57.29%)	40 (46.51%)	0.192
>40	41 (42.71%)	46 (53.49%)	0.192
CA19-9(U/mL), n (%)			0.525
<37	71 (73.96%)	68 (79.07%)	0.525
≥37	25 (26.04%)	18 (20.93%)	0.525
EUS-PP, n (%)			0.798
Diffuse enlargement	64 (66.67%)	61 (70.93%)	0.646
Focal mass	29 (30.21%)	23 (26.74%)	0.725
Basically normal	3 (3.12%)	2 (2.33%)	1.000
EUS-PD, n (%)			0.415
Long or multiple narrow sections	3 (3.12%)	6 (6.98%)	0.311
Focal main pancreatic duct dilation	19 (19.79%)	18 (20.93%)	0.995
Basically normal	67 (69.79%)	52 (60.47%)	0.244
Unknown	7 (7.29%)	10 (11.63%)	0.454
Puncture type, n (%)			< 0.001
EUS-FNA	78 (81.25%)	30 (34.88%)	< 0.001
EUS-FNB	18 (18.75%)	56 (65.12%)	< 0.001
Diameter size of puncture needles, n (%)			0.283
19G	5 (5.21%)	8 (9.30%)	0.434
22G	76 (79.17%)	70 (81.40%)	0.849
25G	15 (15.62%)	8 (9.30%)	0.290
Pathology, n (%)			< 0.001
Suspected type 1 AIP	42 (43.75%)	0 (0)	< 0.001
Level 1 histopathological evidence for type 1 AIP	0 (0)	37 (43.02%)	< 0.001
Level 2 histopathological evidence for type 1 AIP	0 (0)	49 (56.98%)	< 0.001
Insufficient histopathological evidence for type 1 AIP	54 (56.25%)	0 (0)	< 0.001
IgG4 cells/HPF, n (%)			< 0.001
<10 cells/HPF	80 (83.33%)	15 (17.44%)	< 0.001
>10 cells/HPF	16 (16.67%)	71 (82.56%)	< 0.001
Other organ involvementb, n (%)
Biliary stricture	44 (45.83%)	39 (45.35%)	1.000
Retroperitoneal fibrosis	3 (3.12%)	4 (4.65%)	0.709
Kidney	5 (5.21%)	10 (11.63%)	0.193
Other manifestations or no involvement	47 (48.96%)	38 (44.19%)	0.620

Abbreviations: ULN upper limit of normal, AIP autoimmune pancreatitis, PP pancreatic parenchyma, PD pancreatic duct, EUS-FNA EUS-guided fine-needle aspiration, EUS-FNB EUS-guided fine-needle biopsy, HPF high-power field

^a Symptoms are non-mutually exclusive (patients may have multiple concurrent symptoms). “Abdominal discomfort and Obstructive jaundice” denotes coexistence of these two; other categories may also overlap. Thus, counts may exceed total sample size, which is reasonable

^b Organ involvement categories are non-mutually exclusive (patients may have concurrent multiple organ involvement), so the sum of counts may exceed the total sample size, which is reasonable

No statistically significant differences were observed between the two groups in terms of gender (male: 81.25% vs. 87.21%; female: 18.75% vs. 12.79%; P = .371), individual clinical symptoms (abdominal discomfort: 68.75% vs. 62.79%, P = .490; obstructive jaundice: 42.71% vs. 44.19%, P = .959; weight loss: 9.38% vs. 4.65%, P = .344; asymptomatic: 3.12% vs. 2.33%, P = 1.000), age distribution (middle-aged: 36.46% vs. 43.02%, P = .45; elderly: 54.17% vs. 43.02%, P = .176), and medical histories (diabetes, hypertension, family history, alcohol consumption, smoking; all P > .05) (Table 2, Table S1).

Similarly, there were no statistical differences between the two groups in sIgG4 levels (sIgG4 > 2×ULN: 60.42% vs. 69.77%, P = .245), pancreatic parenchyma/pancreatic duct features (assessed by CT/MRI and EUS; all P > .05), and other serum indicators (ALT, AST, TB, CA19-9; all P > .05) (Table 2, Table S1).

Puncture method had a significant impact on the integrity of tissue samples for diagnosis: the proportion of obtaining typical type 1 AIP samples was significantly higher in the EUS-FNB group than in the EUS-FNA group (18.75% vs. 65.12%, P < .001), indicating that EUS-FNB is clinically preferred for improving diagnostic accuracy (Table 2). Puncture needle type had no strong discriminative effect on histopathological confirmation (P = .283) (Table 2). Abundant IgG4-positive cells (> 10 cells/HPF) is a typical manifestation of type 1 AIP, and its proportion was significantly higher in the “Confirmed diagnosis” group than in the “Unconfirmed diagnosis” group (16.67% vs. 82.56%, P < .001), serving as a key indicator for histopathological confirmation (Table 2). All cases in the “Confirmed diagnosis” group were classified as Level 1 (43.02%) or Level 2 (56.98%) (Table 2).

Regarding extrapancreatic involvement, retroperitoneal fibrosis (3.12% vs. 4.65%, P = .709), AIP-related renal involvement (5.21% vs. 11.63%, P = .192) and biliary strictures (45.83% vs. 45.35%, P = 1) exhibited no statistically significant differences between the two groups (Table 2).

Clinical factors associated with histopathological confirmation of Type 1 AIP

To further identify clinical features associated with histopathological confirmation of type 1 AIP, multivariate logistic regression analysis was performed. To intuitively visualize the correlation between clinical features and pathological confirmation, receiver operating characteristic (ROC) curves were plotted for all included variables, and bar charts were generated based on the corresponding area under the curve (AUC) values.

As shown in Fig. 2, variables including high counts of IgG4-positive cells in histopathological sections, puncture type, liver enzyme levels, and AIP-related renal imaging findings exhibited relatively strong correlations with the histopathological confirmation rate.

Fig. 2.

The relationship between a single clinical feature and pathological diagnosis. A. The ROC curves for the predictive ability of different clinical features on pathological diagnosis. B. The AUCs for the predictive ability of different clinical features on pathological diagnosis. (Abbreviations: sIgG4: serum IgG4; PP: pancreatic parenchyma; PD: pancreatic duct; HPF: high-power field; AUC: area under the curve; ROC: receiver operating characteristic)

Subsequently, univariate logistic regression analysis was conducted first, and variables with P < .100 were included in the multivariate logistic regression analysis. Variables with P < .050 in the multivariate analysis were then incorporated into the final logistic regression model. As presented in Table 3, the variables “abdominal discomfort combined with obstructive jaundice”, “diarrhea”, “EUS-FNB”, and “IgG4-positive cells > 10 cells/HPF” were initially included in the multivariate logistic regression analysis. The final results indicated that only two variables were independently associated with histopathological confirmation of type 1 AIP: EUS-FNB (OR = 3.56, 95% CI: 1.55–8.18, P = .003)and IgG4-positive cells > 10 cells/HPF (OR = 15.71, 95% CI: 6.96–35.46, P < .001).

Table 3.

Univariable and multivariable logistic regression analysis of factors associated with the confirmed diagnosis of type 1 AIP

Dependent: Pathology diagnosis	Unconfirmed diagnosis (n = 96)	Confirmed diagnosis (n = 86)	OR (univariable)	OR (multivariable)	OR (final)
Gender, n (%)
Male	78 (81.25%)	75 (87.21%)
Female	18 (18.75%)	11 (12.79%)	0.64 (0.28–1.43, p=.275)
Age, n (%)
18–44	9 (9.38%)	12 (13.95%)
45–59	35 (36.46%)	37 (43.02%)	0.79 (0.30–2.11, p=.642)
≥ 60	52 (54.17%)	37 (43.02%)	0.53 (0.20–1.40, p=.201)
Symptoma, n (%)
Abdominal discomfort	66 (68.75%)	54 (62.79%)	0.77 (0.41–1.42, p=.398)
Obstructive jaundice	41 (42.71%)	38 (44.19%)	1.06 (0.59–1.91, p=.841)
Abdominal discomfort and Obstructive jaundice	24 (25.00%)	11 (12.79%)	0.44 (0.20–0.96, p=.040)	0.48 (0.17–1.38, p=.172)
Weight loss	9 (9.38%)	4 (4.65%)	0.47 (0.14–1.59, p=.226)
Diarrhea	9 (9.38%)	2 (2.33%)	0.23 (0.05–1.10, p=.065)	0.24 (0.03–1.85, p=.171)
Asymptomatic	3 (3.12%)	2 (2.33%)	0.74 (0.12–4.52, p=.743)
sIgG4, n (%)
< 1×ULN	19 (19.79%)	14 (16.28%)
1 ~ 2×ULN	19 (19.79%)	12 (13.95%)	0.86 (0.32–2.33, p=.762)
> 2×ULN	58 (60.42%)	60 (69.77%)	1.40 (0.64–3.06, p=.393)
CA19-9(U/mL), n (%)
< 37	71 (74%)	68 (79.1%)
≥ 37	25 (26%)	18 (20.9%)	0.75 (0.38–1.50, p=.418)
TB(µmol/L), n (%)
≤ 23	54 (56.25%)	43 (50.00%)
> 23	42 (43.75%)	43 (50.00%)	1.29 (0.72–2.31, p=.399)
ALT(U/L), n (%)
≤ 50	53 (55.21%)	43 (50.00%)
> 50	43 (44.79%)	43 (50.00%)	1.23 (0.69–2.21, p=.482)
AST(U/L), n (%)
≤ 40	55 (57.29%)	40 (46.51%)
> 40	41 (42.71%)	46 (53.49%)	1.54 (0.86–2.77, p=.147)
EUS-PP, n (%)
Diffuse enlargement	64 (66.67%)	61 (70.93%)
Focal mass	29 (30.21%)	23 (26.74%)	0.83 (0.43–1.59, p=.579)
Basically normal	3 (3.12%)	2 (2.33%)	0.70 (0.11–4.33, p=.701)
EUS-PD, n (%)
Long or multiple narrow sections	3 (3.12%)	6 (6.98%)
Focal main pancreatic duct dilation	19 (19.79%)	18 (20.93%)	0.47 (0.10–2.18, p=.338)
Basically normal	67 (69.79%)	52 (60.47%)	0.39 (0.09–1.63, p=.195)
Unknown	7 (7.30%)	10 (11.60%)	0.71 (0.13–3.87, p=.696)
Puncture type, n (%)
EUS-FNA	78 (81.25%)	30 (34.88%)
EUS-FNB	18 (18.75%)	56 (65.12%)	8.09 (4.11–15.93, p<.001)	3.23 (1.38–7.53, p=.007)	3.56 (1.55–8.18, p=.003)
IgG4 cells/HPF, n (%)
< 10 cells/HPF	80 (83.33%)	15 (17.44%)
> 10 cells/HPF	16 (16.67%)	71 (82.56%)	23.67 (10.92–51.29, p<.001)	16.43 (7.17–37.64, p<.001)	15.71 (6.96–35.46, p<.001)
Diameter size of puncture needles, n (%)
19G	5 (5.21%)	8 (9.30%)
22G	76 (79.17%)	70 (81.40%)	0.58 (0.18–1.84, p=.352)
25G	15 (15.62%)	8 (9.30%)	0.33 (0.08–1.36, p=.126)
Other organ involvementb, n (%)
Biliary stricture	44 (45.83%)	39 (45.35%)	0.98 (0.55–1.76, p=.948)
Retroperitoneal fibrosis	3 (3.12%)	4 (4.65%)	1.51 (0.33–6.96, p=.595)
Kidney	5 (5.21%)	10 (11.63%)	2.39 (0.78–7.31, p=.125)

Abbreviations: ULN upper limit of normal, AIP autoimmune pancreatitis, PP pancreatic parenchyma, PD pancreatic duct, EUS-FNA EUS-guided fine-needle aspiration, EUS-FNB EUS-guided fine-needle biopsy, HPF high-power field

^aSymptoms are non-mutually exclusive (patients may have multiple concurrent symptoms). “Abdominal discomfort and Obstructive jaundice” denotes coexistence of these two; other categories may also overlap. Thus, counts may exceed total sample size, which is reasonable

^bOrgan involvement categories are non-mutually exclusive (patients may have concurrent multiple organ involvement), so the sum of counts may exceed the total sample size, which is reasonable

The results of both univariate and multivariate logistic regression analyses were visualized using forest plots to clearly display the effect sizes and statistical significance of each variable (Fig. 3, Fig. S2).

Fig. 3.

Forest plots of multivariable logistic regression models for Unconfirmed diagnosis vs. Confirmed diagnosis in patients with type 1 AIP. (Abbreviations: EUS-FNB: EUS-guided fine-needle biopsy; HPF: high-power field.)

Discussion

The core aim of this study was not to validate the diagnostic efficacy of EUS-FNB [20], but to systematically identify multi-dimensional factors for histopathological confirmation of type 1 AIP. Previous studies mostly focused on single-indicator validation—Zhou et al. [20] compared EUS-FNA/FNB sampling efficacy, while Zhang et al. [5] verified IgG4-positive cells’ histological value. Addressing the gap of lacking an integrated diagnostic model for type 1 AIP [3, 21], we comprehensively evaluated 29 factors (clinical symptoms, laboratory results, imaging findings, procedural details) via univariate and multivariate logistic regression, identifying two independent predictors: EUS-FNB (OR = 3.56, 95% CI: 1.55–8.18, P = .003) and IgG4-positive cells > 10/HPF (OR = 15.71, 95% CI: 6.96–35.46, P < .001). This “procedural method + histopathology” concept, distinct from previous fragmented validation, offers practical clinical reference.

Notably, this study provides key negative evidence lacking in some studies [20, 21, 27]: sIgG4 levels, pancreatic duct features, and diffuse pancreatic enlargement have no independent predictive value for histopathological confirmation. These findings have important clinical implications—while these indicators have some clinical reference value, they have limitations as independent grounds for histopathological confirmation, especially sIgG4: approximately 50% of patients with type 1 AIP have normal sIgG4 levels [27, 28]. Consistent with Masamune et al. [11], type 1 AIP primarily affects middle-aged and elderly males (male-to-female ratio ~ 3:1), with abdominal discomfort and obstructive jaundice as main symptoms. Although 19.23% of patients present with both abdominal discomfort and jaundice (attributed to pancreatic swelling compressing adjacent nerves and the distal common bile duct), this symptom lacks specificity for histopathological confirmation (P = .058). Weight loss (7.14%) and diarrhea (6.04%) are rare and show no intergroup differences, consistent with previous reports [29–32].

Comorbidities (diabetes: 26.37%; hypertension: 16.48%) and lifestyle factors (smoking: 37.36%; alcohol consumption: 25.27%) were evenly distributed between the histopathologically confirmed and unconfirmed groups (all P > .05), indicating they do not affect histopathological confirmation. Serological results showed 64.8% of patients had sIgG4 levels exceeding twice the upper limit of normal (2×ULN), but no significant difference was observed between the two groups (P = .245)—emphasizing diagnosis requires integrating imaging and histopathology over sole serology [15, 33]. As a core serological marker for type 1 AIP, approximately 50% of patients have normal sIgG4 levels, making them prone to misdiagnosis as chronic pancreatitis or pancreatic cancer [27, 28, 34–38].The abnormal rates of liver function parameters (ALT, AST, TB) and CA19-9 were 47.25%, 47.80%, 46.70%, and 23.63%, respectively, with no significant intergroup differences—nevertheless, the histopathological confirmation rate trended higher in patients with abnormal sIgG4 (69.77% vs. 60.42%), ALT (50.00% vs. 44.79%), AST (53.49% vs. 42.71%) or TB (50.00% vs. 43.75%). This aligns with Wakabayashi et al.’s [39] report of 33% focal AIP patients with CA19-9 > 100U/ml, supporting that elevated CA19-9 is not specific to pancreatic cancer [39]. Given China’s rising pancreatic cancer mortality and its nonspecific early symptoms [40], our diagnostic framework aids differentiation from type 1 AIP to improve prognosis. Moreover, Abnormal liver function may link to biliary tract involvement in type 1 AIP [41, 42], but none predict histopathological confirmation.

In terms of imaging findings, “diffuse pancreatic enlargement” is the most common imaging feature of type 1 AIP [27], detected in approximately 68% of cases via CT/MRI and EUS. However, there was no significant intergroup difference (P > .05), indicating it cannot distinguish histopathologically confirmed from unconfirmed cases. The detection rate of “long-segment or multiple-segment pancreatic duct stenosis” was low (CT/MRI: 2.20%; EUS: 4.95%); its prevalence was 6.98% in the confirmed group and 3.12% in the unconfirmed group. Though not statistically significant, it has weak indicative value. Additionally, “focal main pancreatic duct dilatation” (approximately 20%) has no predictive value, consistent with the notion that type 1 AIP cannot be diagnosed by imaging alone [15].

Multiple studies confirm EUS-FNB yields better tissue quality than EUS-FNA [20, 21, 24]. Zhou et al. [20] showed EUS-FNB had a significantly higher IgG4-positive cell detection rate (74.2% vs. 27.9%). However, most focus solely on sampling efficacy without clarifying if its advantages are independent. We confirmed EUS-FNB as an independent predictor of type 1 AIP histopathological confirmation (P < .001) via multivariate logistic regression (adjusting 28 confounders). This fills the gap of previous efficacy-focused studies [20, 43], clarifying its advantage is independent of symptoms, serology, and imaging. In our cohort, 65.12% EUS-FNB vs. 34.88% EUS-FNA in the confirmed group, and 81.25% EUS-FNA in the unconfirmed group. Consistent with Yoon et al.’s meta-analysis [24], this supports EUS-FNB’s preferential clinical use. Regarding needle size, 22G was most used (80.22%), with small samples for 19G (7.14%) and 25G (12.64%), showing no significant difference in confirmation rates (P > .05). However, previous multicenter randomized controlled trials confirmed 22G Acquire needles yield more tissue (mean 11.4 mm vs. 5.4 mm) and higher accuracy (87% vs. 67%) than 20G Procore for pancreatic masses (including AIP). This suggests clinical value in selecting 22G needles, and the needle size-related results in this study may be limited by sample size [43].

Histopathological findings further confirm the above diagnostic approach. Our study shows 82.56% of the confirmed group had IgG4-positive cells > 10/HPF, while 83.33% of the unconfirmed group had < 10/HPF. This is consistent with a recent mechanistic study demonstrating that IgG4-positive plasma cells are key pathogenic cells in type 1 AIP, and their differentiation from naïve B cells drives pancreatic immune-mediated injury [44]. Multivariate logistic regression further confirmed IgG4-positive cells > 10/HPF as an independent predictor of histopathological confirmation (P < .001). It also had the highest AUC in ROC analysis, reflecting superior overall diagnostic performance for type 1 AIP compared to other clinical features. Unlike Zhang et al. [5], who only emphasized the histological presence of IgG4-positive cells, this study quantified the indicator’s diagnostic efficacy. We also validated it as an independent predictor, providing tangible clinical evidence for histopathological findings. Additionally, extrapancreatic lesions (e.g., renal involvement, retroperitoneal fibrosis) were more common in the confirmed group (no significant intergroup difference), consistent with type 1 AIP’s nature as a systemic IgG4-related disease [42, 45]. Despite no significant intergroup difference in our cohort, such lesions can still serve as auxiliary clues for clinical suspected diagnosis.

This study has several limitations: First, as a single-center retrospective study, it has limited sample representativeness and extrapolation, requiring multicenter validation. Second, small sample sizes for 19G needles and rare extrapancreatic lesions reduce statistical power of relevant analyses. Third, confounding factors (e.g., pre-puncture glucocorticoid therapy, IgG4-related disease antibodies) were not included, potentially missing key diagnostic factors. Fourth, lack of long-term follow-up data prevents exploring associations between clinical features and prognosis (e.g., recurrence, steroid sensitivity), nor was data on clinical progression, treatment duration, or immunosuppressive therapy needs collected. It is important to note that patients in the “Unconfirmed” group were managed clinically without repeat biopsy; some were subsequently diagnosed through serology and imaging follow-up, while others remained under surveillance.Future multicenter prospective studies should validate our diagnostic model, systematically collect prognosis-related data (e.g., treatment regimens, cycles, recurrence, immunosuppressive therapy use), clarify the link between diagnosis and outcomes, and provide comprehensive evidence for precise management of type 1 AIP.

Conclusions

In conclusion, based on systematic multi-dimensional factor screening, this study analyzed 182 clinically suspected type 1 AIP patients, confirming the disease mainly affects middle-aged/elderly males. Most clinical features (gender, age, comorbidities, sIgG4) have limited value in distinguishing histopathologically confirmed vs. unconfirmed cases. EUS-FNB has a significantly higher histopathological confirmation rate than EUS-FNA, making it the preferred tissue sampling method for the disease. Histopathologically, > 10 IgG4-positive cells/HPF is a key diagnostic indicator. Extrapancreatic lesions (e.g., renal involvement, retroperitoneal fibrosis) are auxiliary diagnostic cues. These findings guide optimization of type 1 AIP’s diagnostic pathway, emphasizing prioritizing EUS-FNB for sampling and integrating IgG4-positive cell histopathology with clinical/imaging data to improve accuracy. In clinical practice, this emphasizes that in suspected cases, clinicians should prioritize EUS-FNB sampling and specifically request histopathological assessment of IgG4-positive cell density to maximize diagnostic yield.

Abbreviations

AIP

autoimmune pancreatitis

EUS

Endoscopic Ultrasound

EUS-FNA/FNB

EUS-guided fine-needle aspiration/biopsy

HPF

high-power field

ICDC

International Consensus Diagnostic Criteria

IgG4-RD

IgG4-related disease

PP

pancreatic parenchyma

PD

pancreatic duct

sIgG4

serum IgG4

ULN

upper limit of normal

Authors’ contributions

XRT, JYL, DLW and YYZ: Responsible for the data curation and writingoriginal draft; HXC, JHX, ZHY, MRJ, WSL and CL: Responsible for the investigation and formal analysis; DYZ and LQS: Responsible for the resources and validation; and ZDJ and HJH: Responsible for the conceptualization and project administration. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (grants 82370658, 82170657 and 82370655).

Data availability

The datasets used in the current study can be obtained from the corresponding authors upon reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted according to the guidelines laid down in the Declaration of Helsinki. This study was approved by the Ethics Committee of Changhai Hospital (Ethics No.: CHEC2025-403) and was granted an exemption from the requirement for informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.