RPS 1601 - Updates on pancreatic cysts and gland dysfunction

Author Block: A. Saccomanno¹, D. Squecco², G. Morana³; ¹Varese/IT, ²Modena/IT, ³Treviso/IT
Purpose: To assess the risk of progression and clinical outcomes in patients with Branch-Duct Intraductal Papillary Mucinous Neoplasms (BD-IPMN) ≤2 cm followed by magnetic resonance imaging (MRI) for at least 10 years, stratified by initial cyst size (1–10 mm vs 11–20 mm).
Methods or Background: We performed a retrospective observational study at Ca’ Foncello Hospital, Treviso. A total of 174 patients with BD-IPMN ≤2 cm diagnosed between 2007 and 2025 were included, all undergoing MRI follow-up for ≥10 years. Outcomes were time to progression, defined as appearance of worrisome features (WF) or high-risk stigmata (HRS), time to surgery, and overall survival (OS). Kaplan–Meier survival analysis and multivariate Cox regression were used to identify independent predictors.
Results or Findings: Median follow-up was 12 years. New WF/HRS occurred in fewer than 15% of patients, with cumulative progression rates of 10.8% at 5 years and 12.6% at 10 years. Stratified by size, 10-year progression was 8.5% in cysts 1–10 mm and 21.4% in cysts 11–20 mm (p=0.028). Surgery was required in 11 patients (6.3%), yielding one high-grade dysplasia and two pancreatic ductal adenocarcinomas. Ten-year OS exceeded 87% in both groups. Multivariate analysis confirmed age at diagnosis (HR 1.22, p<0.001) and active smoking (HR 4.69, p=0.042) as independent predictors.
Conclusion: BD-IPMN ≤2 cm show indolent long-term behavior, with most patients free from WF/HRS even after 10 years. Lesions 11–20 mm have a slightly higher risk of progression, but overall rates remain low.
Limitations: This is a retrospective, single-center study; not all follow-up MRIs were performed with contrast media; the number of surgical and oncological events was limited.
Funding for this study: No external funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: I. Dudás, Z. Somogyiné Nagy, A. Kubovje, A. I. Vigh, T. Marjai, P. Maurovich-Horvat, B. K. Budai; Budapest/HU
Purpose: This study aimed to compare image quality and radiation dose between single-source and dual-source photon-counting detector CT (PCD-CT) for pancreatic cystic lesion (PCL) follow-up imaging.
Methods or Background: We retrospectively identified 20 patients with pancreatic cystic lesions scanned with single-source PCD-CT between 2024 October and 2024 December who had previous dual-source PCD-CT scans at our institution with the same abdominal imaging protocol as part of their follow-up. In addition, anthropomorphic phantoms were scanned on both systems. Virtual monoenergetic images (VMI) at 40 and 70 keV were reconstructed. Circular regions of interest were placed in the aorta, pancreas parenchyma, liver parenchyma, cyst fluid, subcutaneous fat, and iliopsoas muscles. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated. Pancreatic duct depiction, PCL delineation, image noise, and overall quality were assessed using a 5-point modified Likert scale.
Results or Findings: The radiation dose was not significantly different between the scanners in either the unenhanced, pancreatic, or venous phases. At 70 keV, postcontrast CNR did not differ between scanners, whereas at 40 keV, single-source PCD-CT demonstrated significantly higher CNR in the aorta (p=0.014), pancreas (p=0.002), liver (p<0.001), and iliopsoas (p=0.017) for the venous phase. SNR was significantly higher for single-source in multiple tissues at both 40 and 70 keV, most prominently in the venous phase. Subjective assessments favored single-source PCD-CT for image quality, though phantom-based noise power spectrum analysis showed no significant differences.
Conclusion: Single-source PCD-CT provides improved SNR, CNR, and subjective image quality compared with dual-source PCD-CT for pancreatic cystic lesion imaging, without increased radiation dose. However, distinction cannot be made whether these improvements originate from hardware or software differences.
Limitations: This was a single-center study with a retrospective study design.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the institutional ethics committee.

Author Block: B. Abdolalizadeh, N. Madrid Scheller, S. Jawad, T. Skårup Kristensen, C. Palnæs Hansen, C. Ewertsen; Copenhagen/DK
Purpose: Intraductal papillary mucinous neoplasms (IPMNs) are common pancreatic cystic lesions with variable malignant potential, making management and surveillance challenging. This study aimed to evaluate imaging characteristics, demographic factors, and clinical outcomes of patients with IPMNs referred to multidisciplinary team conferences at a tertiary reference center between 2019 and 2023.
Methods or Background: We included all patients referred to the weekly IPMN MDT conference from January 1, 2019, to December 31, 2023. Exclusion criteria were cystic lesions other than IPMN, uncertain IPMN diagnosis, patients with only a single CT scan available, and cysts <5 mm. Demographic data, imaging features including baseline and newly developed worrisome features (WFs), and histological outcomes of surgically resected patients were collected from electronic health records.
Results or Findings: A total of 1082 patients were included (mean age 69.8 years, 57.1% female). Branch-duct IPMN was the predominant subtype (95.3%). Worrisome features (WFs) were present in 207 patients (19.1%) at baseline, and an additional 47 patients (4.1%) developed new WFs during follow-up. Rapid cyst growth was observed in 6.8% according to Fukuoka criteria and in 10.3% according to the updated Kyoto 2024 criteria. Surgical resection was performed in 62 patients (5.7%), with malignant transformation or high-grade dysplasia confirmed in 31 cases (2.9%).
Conclusion: Malignant transformation was infrequent. The presence of worrisome features, including rapid cyst growth did not predict malignancy. These findings support a more individualised and potentially less intensive surveillance strategy.
Limitations: The limitations of the study are its single-centre , potential selection bias, and interobserver variability in WF assessment.
Funding for this study: No funding was provided for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: legal department at Rigshospitalet (ref. 23052217) and the Danish Data Protection Agency (RH -2015-07, nr. 03616)

Author Block: C. Reca, R. Levi, C. Bonifacio, A. Laghi; Milano/IT
Purpose: To compare the diagnostic value of Abbreviated MRI protocols (APs)(with/without DWI) versus a comprehensive protocol(CP) for Branch-Duct Intraductal Papillary Mucinous Neoplasms(BD-IPMN) surveillance, evaluating inter-reader agreement, cost-effectiveness and time efficiency.
Methods or Background: This single-center retrospective study included patients with BD-IPMN undergoing baseline and follow-up MRI between January2018 and April2024. CP included axial and coronal T2WI, axial fat-saturated T1-weighted images before and after contrast injection (T1WI), 2D and 3D MR MRCP and DWI images.The two APs were obtained from CP follow-up examinations: AP with DWI consisted of axial and coronal T2WI, 2D-3D MRCP, axial fat-saturated T1WI images and DWI, AP without DWI comprised only axial and coronal T2WI and 2D-3D MRCP. A senior radiologist(>20 years of experience in pancreatic imaging) and a junior radiologist(4th-year-radiology resident) independently and blinded analyzed each protocol using Kyoto criteria and recommended management. Inter-reader and inter-protocol agreement were analyzed using Cohen’s and Fleiss’ Kappa. Acquisition times and healthcare costs were compared using national reimbursement rates(p<0.05 statistical significant).
Results or Findings: Overall, 40 patients(25 women and 15 men) with a median age of 67 years(range 48-87) were included. No statistically significant differences were observed among the 3protocols in identifying worrisome features, including high-risk stigmata, cyst size, main pancreatic duct dilation, and mural nodules.The inter-reader agreement was almost perfect(Fleiss' Kappa=0.82). Substantial inter-reader agreement(Cohen’s Kappa>0.6) was observed for all protocols, with almost perfect agreement(Cohen’s Kappa>0.8) in some instances. APs demonstrated up to a 48%reduction in acquisition time and a 27.4%cost reduction compared to CP. The addition of DWI did not significantly improve diagnostic accuracy.
Conclusion: APs, particularly without DWI, are reliable and cost-effective alternatives to CP for BD-IPMN follow-up, maintaining comparable diagnostic performance. Substantial inter-reader agreement highlights the reproducibility of APs, even among less experienced radiologists.
Limitations: Retrospective
Funding for this study: n/a
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: n/a

Author Block: S. Boccioli, S. Bogani, C. Badii, G. Danti, S. Paolucci, V. Miele; Florence (FI)/IT
Purpose: The aim of the study was to develop a radiomic MRI model able to predict the future development of worrisome features (WF) in branch duct-intraductal papillary mucinous neoplasms (BD-IPMNs). By analyzing quantitative imaging features from MRI scans, we try to identify biomarkers that could be used as early indicators of disease progression, helping personalized management and reducing unnecessary interventions.
Methods or Background: We performed a single-center retrospective study including patients with BD-IPMNs who had at least two contrast-enhanced MRI exams between January 2011 and March 2025. Patients were divided into two groups depending on whether they developed WF or HRS during follow-up. Lesions were manually segmented on T2-weighted images and 107 radiomic features were extracted using PyRadiomics. Feature selection and model building were done using LASSO logistic regression with weighting to account for class imbalance.
Results or Findings: A total of 194 patients were included. After a mean follow-up of 4 years and 5 months, 143 patients did not develop WF/HRS (Group 0) while 51 did (Group 1). Out of 107 radiomic features, 52 were initially associated with progression and 9 shape-based features remained significant after correction. The predictive model achieved an AUC of 0.70 (95% CI 0.62–0.79) and a precision-recall AUC of 0.49, showing moderate predictive ability.
Conclusion: Radiomic MRI models may help identify BD-IPMN patients at higher risk of developing WF or HRS. By providing quantitative information beyond conventional imaging, it could support earlier and more personalized clinical decisions. Further studies are needed to confirm and refine these findings.
Limitations: This study was retrospective, conducted at a single center and included a relatively small sample, which may limit generalizability. MRI scans from different vendors could introduce variability. Moreover the biological significance of radiomic features remains unclear.
Funding for this study: No
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the local Ethical Committee of Careggi University Hospital (July 13th, 2021; protocol number: 20256_oss) and conducted in accordance with the Declaration of Helsinki and its later amendments.

Author Block: A. Chernomorets, D. Ben-Ami Shor, R. Tzadok, H. Yashar, S. Lazar, Y. Katz, R. Kessner; Tel Aviv/IL
Purpose: -To evaluate the growth of Side-Branch Intraductal Pancreatic Mucinous Neoplasm (SB-IPMN) lesions during a long follow-up period.
-To compare between SB-IPMN lesions of different sizes.
Methods or Background: The final reports of all the patients that underwent MRI-MRCP at our institution between the years 2011-2021 were reviewed for the diagnosis of SB-IPMN. 483 adult patients were diagnosed with SB-IPMN, however 115 patients were excluded due to: misclassification at diagnosis, clinical or radiological follow-up of less than 12 months, non-detectable lesion on revision or pancreatic malignancy at diagnosis. We evaluated the growth of the largest cystic lesion during follow-up, according to the first and last examinations of the patients. The medical records of the patients were reviewed for demographic data, history of hepatobiliary diseases, and history of non-pancreatic malignancy. Finally, we compared between 3 size groups of SB-IPMN: < 1 cm at diagnosis (smallest), 1-1.9 cm (intermediate) and >= 2 cm (largest).
Results or Findings: Our final study population included 368 patients, 254 females and 114 males. The mean radiological follow-up was 54.3 months and the mean clinical follow-up was 67.4 months. The mean absolute growth of the largest cystic lesion during follow-up was 1.3 mm, for the entire population. Only two patients (0.5%) developed pancreatic malignancy, one from the smallest lesions group and one from the largest lesions group. The comparison between the size groups revealed no differences regarding the mean follow-up length, growth during follow-up and development of malignancy. The mean growth during follow-up was 1.4 mm in the smallest and intermediate lesions groups and 0.4 mm in the largest lesions group.
Conclusion: Our results suggest that the follow-up guidelines for SB-IPMN should be less strict than for other types of IPMN.
Limitations: The study is retrospective.
Funding for this study: None.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Tel Aviv Sourasky Medical Center Institutional Review Board approval (2021-2025)

Author Block: A. Del Gaudio, D. De Santis, B. Masci, B. Catalano, M. Zerunian, G. Tremamunno, T. Polidori, D. Caruso, M. Francone; Rome/IT
Purpose: To evaluate the feasibility of replacing true non-contrast (TNC) images with virtual non-contrast (VNC) reconstructions generated with dual-layer spectral CT (DLCT).
Methods or Background: From December 2024 to February 2025, consecutive inpatients who underwent contrast-enhanced abdominal-CT using DLCT scanner (Spectral CT 7500, Philips Healthcare) were retrospectively enrolled. Each examination included a TNC acquisition, followed by arterial and portal venous phases. VNC images were reconstructed from the arterial (VNCa) and portal venous (VNCp) datasets using vendor-provided spectral decomposition algorithms. An independent radiologist measured attenuation values (HU) on TNC, VNCa, and VNCp images for the liver, spleen, pancreas, kidneys, abdominal aorta, portal vein, and hypodense liver and kidney lesions. Signal-to-noise ratio (SNR) and contrast-to-noise ratio(CNR) were calculated for parenchymal organs and vessels across all image types. Lesion size was assessed on TNC, VNCa,and VNCp images.
Results or Findings: One hundred patients (mean age: 67.1 ± 11.4 years; 52 males)with 160 hypodense abdominal lesions were analyzed.
VNCa and VNCp reconstructions showed lower attenuation than TNC in the liver, spleen, and pancreas(all P < .001).No significant HU differences were observed in the portal vein and kidneys(all P > .540). Aortic HU was higher in VNCa but lower in VNCp compared with TNC(P < .001).Mean HU of hypodense lesions was significantly higher in TNC compared with both VNCa and VNCp(P < .001).
SNR was consistently higher in VNCa and VNCp compared to TNC(P < .001).
CNR was significantly higher in VNCa compared with TNC for the aorta and spleen, and higher in VNCa than VNCp for the aorta. No other significant differences were observed.
Conclusion: VNC images from DLCT underestimated attenuation relative to TNC but provided comparable or higher objective image quality and preserved lesion size measurement.
Limitations: Retrospective single-center study, single-vendor reconstruction
Funding for this study: N/A
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: F. Omboni¹, M. Ronot², C. R. A. Corallo³, C. Triantopoulou⁴, N. Kartalis⁵, M. Zins², G. Zamboni¹; ¹Verona/IT, ²Paris/FR, ³Leeds/UK, ⁴Athens/GR, ⁵Stockholm/SE
Purpose: To describe a series of cystic panNETs to determine specific imaging features in a large cohort of patients
Methods or Background: This is a retrospective multicenter observational study (5 centers in 4 countries) which included cystic panNETs diagnosed with FNA/pathology or by 68GaDOTATOC-PET-CT. Cross-sectional imaging was reviewed for each patient (145 MRI, 11 CT with contrast, 1 non-contrast CT).
Results or Findings: 157 patients (85 women, 72 men, mean age 57,5 yrs) with 166 cystic panNETs were included; 9 had multiple lesions. Diagnosis was made by FNA/pathology (143 lesions) or by positivity at DOTATOC-PET-CT (23 lesions). Nine lesions were functioning. ki-67 ranged from 1% to 19% (G1:103; G2: 21; G3: 0). Ki-67% was not available/indeterminate for 42 lesions.
Lesions were predominantly located in the body-tail (125) with a mean diameter of 24mm (5-90mm) and a mean wall thickness of 1.9mm. Most patients (130/157) had a solitary lesion; 36 panNETs were associated with other pancreatic lesions (18 IPMN, 14 solid/cystic panNETs, 4 nonspecific cysts). 142 lesions had round margins.
On MRI, cyst fluid was predominantly hyperintense on T2-weighted sequences (139/166 lesions) and hypointense on T1-sequences (152/166 lesions). Five lesions contained enhancing mural nodules. MPD was dilated in 13 patients, in 12/13 caused by the cystic lesion.
Contrast-enhancement was present in 162 lesions, in 96/162 constant in all phases; 4 lesions showed no enhancement. 15 lesions had wall calcifications. No lesion showed macroscopic nodal or metastatic spread at baseline.
Conclusion: In our retrospective multicenter series, we analyzed radiologic features of cystic panNETs. The lack of extra-pancreatic spread at diagnosis, even in large tumors with high Ki67, raises the possibility of an indolent course. Long-term follow-up studies are needed to confirm this hypothesis.
Limitations: Retrospective study
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: F. Omboni¹, M. Ronot², C. R. A. Corallo³, C. Triantopoulou⁴, N. Kartalis⁵, M. Zins², G. Zamboni¹; ¹Verona/IT, ²Paris/FR, ³Leeds/UK, ⁴Athens/GR, ⁵Stockholm/SE
Purpose: To evaluate a population of resected or followed-up cystic panNETs to determine natural evolution.
Methods or Background: This retrospective multicentric study included 121 patients (69 women, 52 men, mean age 60 yrs) with 126 cystic panNETs diagnosed by FNA/histology (115 lesions) or 68Ga-DOTATOC-PET-CT (11 lesions). Ki-67 was available for 90 lesions (range 1%-15% (79 G1; 11 G2). For each patient, all available imaging was assessed, at least baseline and most recent scans (119 MRI; 7 CT). Patients were divided in two subgroups: resected versus followed-up.
Results or Findings: At baseline, no lesion showed macroscopic nodal or metastatic spread.
Resected group (55 lesions): mean diameter 29,9 mm (17 pT1, 26 pT2, 12 pT3). Thirty-three were resected after ≥ 6 months follow-up (mean interval 29 months). Three tumors >40 mm (44, 52, 70 mm), showed peripancreatic nodal involvement (all T3N1M0; 2G1 + 1G2). No lesion < 40 mm have nodal spread. After a median 31,7 months follow-up, no recurrence was observed in 49/55 (6 lost to follow-up).
Follow-up group (71 lesions): fifty-nine were followed for >6 months (mean 46,9 months). Mean diameter was 17,5 mm; among the 15 lesions >20 mm, 10 remained stable. Eleven progressed radiologically after a mean of 65.2 months. Detailed data were available for 8/11 growth was limited (<50% of initial diameter). None developed nodal or metastatic spread.
Conclusion: In this series, cystic G1-G2 cystic panNETs <40 mm showed an indolent course, with no nodal or metastatic spread during follow-up. In such cases, surveillance may represent a safe alternative to upfront resection.
Limitations: Retrospective study
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: S. Nalliah¹, S. N. F. Hostrup¹, E. B. Mark¹, M. H. Liedenbaum², T. Engjom², I. H. S. Haldorsen², A. M. Drewes¹, S. S. Olesen¹, J. B. Frøkjær¹; ¹Aalborg/DK, ²Bergen/NO
Purpose: To evaluate whether a radiomics-based AI model can (1) classify patients with chronic pancreatitis (CP), (2) identify CP-related complications, and (3) provide quantitative imaging biomarkers of disease severity on routine CT.
Methods or Background: CP is a progressive inflammatory disease associated with pain, diabetes, and exocrine pancreatic insufficiency (EPI). Conventional imaging typically detects advanced disease, limiting early diagnosis and accurate staging. Radiomics can reveal subtle imaging alterations beyond visual assessment, offering an opportunity for earlier detection and monitoring. We evaluated whether CT-based radiomics could classify CP, detect complications, and provide biomarkers of disease severity. The study included 468 participants: a training cohort of 359 from Aalborg (201 CP, 148 controls) and a test cohort of 109 (68 external CP patients from Bergen, 41 Aalborg controls). CT scans underwent automated pancreatic segmentation, followed by radiomics feature extraction (PyRadiomics) and feature selection using LASSO regression. AI models were trained to classify CP and complications (EPI, diabetes, pain). Performance was evaluated on the test cohort, and model probability scores were correlated with fecal elastase levels and disease severity.
Results or Findings: Forty-six radiomics features were associated with CP and its complications, including markers related to pancreatic volume, calcifications, and ductal dilatation. The CP vs. healthy classification model achieved excellent performance (AUC = 0.97). For complications, AUCs were 0.80 for EPI, 0.63 for diabetes, and 0.59 for pain. Model probability scores correlated with fecal elastase levels (p < 0.001) and increased with disease severity (p = 0.004).
Conclusion: Radiomics-based CT analysis allows accurate CP classification and provides quantitative markers of disease severity and complications. These results support its potential as a non-invasive tool for diagnosis, staging, and longitudinal monitoring of CP
Limitations: Retrospective design, and no prospective clinical validation limit this study.
Funding for this study: The study was funded from the North Denmark Region Health Innovation Fund (journal number: 2024-0015).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective, multi-center study was approved by the Danish National Committee on Health Research Ethics (Ref: 2308560) and The Regional Ethical Committee in Western Norway (Ref: 2019/1037), with data-sharing agreements between Aalborg University Hospital and Haukeland University Hospital approved by the Scandinavian Baltic Pancreas Club (SBPC) steering committee.

Author Block: T. J. Vogl, H. Adwan, I. Al Haj Ibrahem; Frankfurt/DE
Purpose: To retrospectively analyze the accuracy and safety of CT-guided percutaneous core-needle biopsies of pancreatic lesions suspected to be malignant based on MRI or CT scans.
Methods or Background: This study evaluated CT-guided percutaneous biopsies of suspicious pancreatic lesions performed at our university hospital. Biopsy was performed using a 17G coaxial needle and an 18G core biopsy needle. Data on patient characteristics, lesions, procedures, and histologic results were recorded.
Results or Findings: In total, 90 patients (58.9% males, mean: 65±12.2 years) underwent CT-guided percutaneous biopsies. The lesions had a mean size of 36.8±12.2 mm and were predominantly located in the pancreatic head 61.1% (55/90). Technical success was achieved in all biopsies 100% (90/90). 96.7% of the procedures (87/90) were performed using direct access routes, while 3.3% (3/90) required indirect transhepatic or transgastric approaches. Among the biopsies, 65.6% (59/90) confirmed malignancy, with adenocarcinoma as the most common malignant subtype representing 55.6% (50/90) of all cases. The rate of non-malignant findings was 26.7% (24/90) including chronic pancreatitis at a rate of 5.6% (5/90) as well as pancreatic pseudocysts and pancreatic cystadenoma each at a rate of 2.2% (2/90). Seven cases were identified as false negative, but malignancy was later confirmed after re-biopsy or surgery. The diagnostic accuracy was 92.2% (83/90). The rate of major complications was 1.1% (1/90), of minor complications 2.2% (2/90).
Conclusion: This study shows that CT-guided pancreatic biopsy is a safe procedure with high diagnostic accuracy.
Limitations: Retrospective study design
Funding for this study: No funding
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Approval of the ethics committee of the Johann-Wolfgang-Goethe University Frankfurt

Author Block: S. Seyrek, A. AVCIOGLU, S. S. Taflan, B. Tozlu, Ö. B. Kuş, M. G. Kartal, B. Gultekin, A. Sadic, A. K. Uzum; Istanbul/TR
Purpose: Late-onset autoimmune diabetes mellitus (LADA) and type 2 diabetes mellitus (DM) are are often misdiagnosed (10–20%) but differ in treatment and prognosis. In this study, multiparametric MRI techniques capable of reliably distinguishing between these two entities within the clinical gray zone were compared.
Methods or Background: This prospective single-center study enrolled 104 outpatients (51 LADA, 53 type 2 DM) during July 2024–August 2025. Participants underwent non-contrast abdominal MRI. Pancreatic volume (PV, cm³), proton density fat fraction (PDFF, %), T1 relaxation time (ms), and T2* relaxation time (ms) were measured. In addition, C-peptide levels and fecal elastase were assessed to evaluate endocrine and exocrine pancreatic function.
Results or Findings: Compared with type 2 DM, patients with LADA demonstrated significantly lower PV (35.7 vs. 50.1 cm³, p = 0.0002), with consistent reductions across the head (p < 0.0001) and corpus (p = 0.0069). PDFF values were significantly lower in LADA patients both in the whole pancreas (2.3 vs. 5.6%, p = 0.0005) and pancreatic head (p = 0.012). C-peptide was reduced in LADA (0.83 vs. 1.62 ng/ml, p < 0.0001), whereas fecal elastase showed no difference (p = 0.71). Anthropometric measures, including body mass index (26.1 vs. 29.6 kg/m², p = 0.0007) was also significantly lower in LADA patients. Regarding relaxation times, no significant differences were observed for pancreatic T1 (p = 0.28) or T2* (p = 0.028). Additional analyses revealed smaller psoas muscle area (2.98 vs. 4.97 cm², p < 0.0001) and reduced vertebral bone marrow fat fraction (42.0 vs. 49.0%, p = 0.003) in LADA.
Conclusion: Multiparametric MRI, particularly pancreatic volume and PDFF, provides reliable markers to distinguish LADA from type 2 DM within the clinical gray zone.
Limitations: Patient compliance
Funding for this study: This study was supported by the Scientific Research Projects Coordination Unit of Istanbul University (Project No: TTU-2024-40922)
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Ethics Committee of Istanbul University/Istanbul Faculty of Medicine (Approval No:2023/2373, Date: 29/12/2023). All participants provided informed consent

Author Block: T. Ota¹, A. Busse-Coté², M. Hori¹, R. Sartoris², A. Vanzulli², V. Vilgrain², N. Tomiyama¹, M. Ronot²; ¹Suita/JP, ²Paris/FR
Purpose: Abdominal aortic calcification (AAC) has been linked to complications after pancreatoduodenectomy (PD), but the value of volumetric calcification score (VCS) and radiomic features (RFs) remains unclear. We aimed to evaluate whether volumetric AAC metrics and RFs predict postoperative complication severity and to develop integrated prediction models (PMs).
Methods or Background: This retrospective study included 380 patients undergoing PD. Clinical variables was summarized as propensity score (PS). AAC was segmented on non-contrast CT to calculate ACV volume (ACV), VCS, and 107 RFs. PMs combining PS, VCS, and selected RF were built using multivariate logistic regression, support vector machine (SVM), and neural network (NN), and PMs were validated by two radiologists. Performance was assessed using area under the receiver operating characteristic curve (AUC) and accuracy.
Results or Findings: Of 380 patients, 177 with measurable AAC were included for radiomic analyses, and randomly assigned to training (n=141) and validation (n=36) cohorts. PS was significantly higher in patients with major versus minor complications (p < 0.0001). Neither ACV nor VCS differed significantly between groups. Among RFs, "busyness" showed the strongest discriminatory potential (p = 0.078). Three PMs incorporating PS, VCS, and busyness were constructed. In training, NN-based PM2 (PS + busyness) achieved the best performance (AUC 0.94; accuracy 86.5%). In validation, PM2 with SVM and NN maintained robust accuracy (~80%) and AUCs of 0.86–0.88.
Conclusion: Volumetric AAC measures alone were insufficient predictors of complication severity. Combining clinical data with radiomic texture “busyness” in non-linear models improved accuracy, offering a promising preoperative tool for PD risk stratification.
Limitations: Single-centre, retrospective design.
Funding for this study: N/A
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Institutional Review Board of Hôpital Beaujon, AP-HP