Citations 2
Contemporary Adjuvant Chemotherapy for Intraductal Papillary Mucinous Neoplasms
Haugk, Beate , White, Steven, Marchegiani, Giovanni, Holmberg, Marcus, Zerbi, Alessandro, Lu, Zipeng , Inoue, Y., Choi, Munseok , Kang, Chang Moo, Visser, B., Corpelijn, Lieke, Andersson, Bodil, Bereza-Carlson, Paulina, Björnsson, Bergthor, Wennerblom, Johanna, Goh, Brian K. P., Hartman, V., Ahmad, Syed, Patel, Sameer, Berrevoet, Frederik, Mortensen, Kim Erlend, Roberts, K., Lynch, Calum, Gupta, Amit, Kleeff, Joerg, Addeo, Pietro, Friess, Helmut, Koea, Jonathan, Gomez, Dhanny , Gordon-Weeks, Alex, Doria, Cataldo, Tzimas, Georgios, Franklin, O., Månsson, C., Cabús, Santiago Sánchez, Bhogal, Ricky H. , Pathak, Samir, Honselmann, Kim Christin, Balakrishnan, Anita, Mann, Kulbir , Gill, Anthony J., Mittal, A, Samra, Jas, Athwal, Tejinderjit, Loveday, Benjamin, Bhardwaj, Neil, Menon, Krishna, Thapa, Prabin, Matsumoto, Ippei, Jamieson, Nigel, Rebours, Vinciane, Descourvières, Clémence, Davidson, BR, Soreide, Kjetil, Pandanaboyana, Sanjay, Collaborative, Adeno-IPMN2, Kumar, Ajit, González, Carol Natalia Clavijo, Lam, Shi, L’Official-German, Ivan, Bergner, Tina, Rosso, Edoardo , Giulianotti, Pier Cristoforo , Morreau, Mathew , Dili, Alexandra, Sotiropoulou, Maria , Kuzman, Matta , Büyükkasap, Ahmet Çağrı, Uludağ, Gökhan, McGuigan, Mari-Claire , Chouari, Tarak , Dogul, Ahmet, Doğu, Doğukan , Rebelo, Artur, Holewik, Franziska Bernadette, Plesa-Furda, Roxana, Revoredo, Fernando, Galanis, Ilias , Maharjan, Dhiresh, Ghorbani, Poya , Lind, Max, Ami, Katsuya , Vijayashanker, Aarathi , Alsaoudi, Tareq , Lundgren, Linda , Ellisa, Katrina, Hong, Paul , Gupta, Archit, MartinXu, Congde, Konstantoudakis, Georgios, Christophides, Thalis, Spurring, Eleanor Margaret, Iwao, Yasuhito, Pirklc, Martin, Vaitha, Ruhi , Tiwari, Navneet, Sheel, Andrea, Kotecha, Krishna , Stavrou, Gregor A., Patel, Rikhilroy, Misca, Mihaela , Hammond, John, Roalso, Marcus Thomas Thor, Soomro, Faiza H, Halimi, Asif , Kendall, Timothy, Patel, Visad, Stankic, Dunja, Dries, Pieter , Hein, Charlotte, Betram, Julia, Jacob, Rosetta, Wilson, Gregory, Redas, Souheil, Poškus, Augustas, Mosert, Paula Lucia, Šilinskaitė, Ugnė , Hirose, Yuki, Liu, Xue, Standaert, Lars, Kay, Tegan J, Nishida, Kenji, Peddakota, Vivek , Gupta, Amit , Kang, Chang Moo , Ven Fong, Zhi, Kang, Chang Moo , Lucocq, James E. , Marchegiani, Giovanni , Zerbi, Alessandro , Pagnanelli, Michele , Visser, B. , Andersson, Bodil , Bereza‐Carlson, Paulina , Björnsson, Bergþór , Wennerblom, Johanna Hansson , Labori, Knut Jørgen , Goh, Brian K P , Hartman, Vera , Berrevoet, Frederik Christiaan , Ausania, Fabio , Bertrand, Claude , Kleeff, Joerg, Addeo, Pietro , Bellotti, Ruben , Fong, Zhi Ven , Friess, Helmut , Koea, Jonathan , Gordon-Weeks, Alex Nicolas , Doria, Cataldo , Franklin, Oskar , Pathak, Samir , Honselmann, Kim C. , Balakrishnan, Anita , Harrison, Ewen M. , Gill, A J, Mittal, Anubhav, Samra, Jas , Athwal, Tejinderjit , Loveday, Benjamin P. T. , Teh, Catherine , Bhardwaj, Neil , Menon, Krishna, Matsumoto, Ippei , Jamieson, Nigel , Rebours, Vinciane , Søreide, Kjetil , Martlı, Hüseyin Fahri , Roeyen, Geert , Antonakis, Pantelis T. , Bramis, Konstantinos , Hall, Lewis A , Valle, Valentina , Connor, Saxon J. , Kapiris, Stylianos A. , Kerem, Mustafa , Rebelo, Artur , Plesa-Furda, R., Filho, Omero Pereira da Costa , Galanis, Ilias, Konstantoudakis, Georgios, Christophides, Thalis, McGuinness, Matthew James , Yaqub, Sheraz , Pellegrini, Riccardo , Perri, Giampaolo , Sheel, Andrea , Sheen, Amy , Evans, Daisy, Ghamarnejad, Omid , Karavias, Dimitrios D. , Roalsø, Marcus T. T. , Stankic, Dunja , Grubor, Nikica , Gulla, Aistė , Furukawa, Takaaki , Hirose, Yuki , Yin, Lingdi
JAMA Network Open | March 27, 2026 | Volume: 9 | Issue: 3 | Pages: e263688
10.1001/jamanetworkopen.2026.3688
Importance Adjuvant chemotherapy regimens may be administered after pancreatic resection for adenocarcinoma arising from intraductal papillary mucinous neoplasms (A-IPMNs), although the evidence supporting their use is limited. Objective To evaluate the survival benefit associated with contemporary adjuvant chemotherapy regimens after resection in A-IPMNs between 2017 and 2023. Design, Setting, and Participants This retrospective cohort study was an international, multicenter study with 69 participating centers across Europe, North America, South America, and the Asia-Pacific region. Patients undergoing resection for A-IPMNs were included. Data were analyzed from May to August 2025. Intervention Contemporary adjuvant chemotherapy regimens, such as gemcitabine-capecitabine (GemCap); 5-fluorouracil, leucovorin, oxaliplatin, and irinotecan (FOLFIRINOX), modified FOLFIRINOX (mFOLFIRINOX), and S-1. Main Outcome and Measure Overall survival (months). Results Among 1321 patients (median [IQR] age 70 [63 to 76] years; 713 males [54.0%] and 608 females [46.0%]), 781 patients (59.1%) received adjuvant chemotherapy, while in 181 patients (13.7%) it was omitted due to poor patient fitness. Of patients who received adjuvant chemotherapy, 568 patients (72.6%) received contemporary regimens, including GemCap (232 patients [29.7%]), FOLFIRINOX (176 patients [22.5%]), mFOLFIRINOX (71 patients [9.1%]), and S-1 (70 patients [9.0%]). The median (IQR) follow-up for the cohort was 64.2 (40.4 to 85.0) months, and the median overall survival was 73.8 months (95% CI, 66.4 to 81.9 months). After 90-day landmark analysis and exclusion of patients ineligible for chemotherapy, adjuvant chemotherapy vs no adjuvant chemotherapy (propensity score–matched populations, 243:243 patients) was not associated with improved overall survival (median, 82.3 months; 95% CI, 78.2 months to not applicable [NA] vs not reached; 95% CI, 75.3 months to NA; P = .58). Contemporary regimens vs no adjuvant chemotherapy (propensity score–matched populations, 309:309 patients) was not associated with longer survival, and a mean survival benefit greater than 4.2 months over 5 years was excluded (difference in restricted mean survival, 1.26 months; 95%, −1.72 to 4.24 months). Treatment outcomes did not vary by chemotherapy regimen or disease characteristic (eg, N stage or carbohydrate antigen 19-9 level). Conclusions and Relevance In this study, contemporary adjuvant chemotherapy was not associated with improved overall survival in A-IPMNs, and a randomized clinical trial is indicated.
Class imbalance correction in artificial intelligence models leads to miscalibrated clinical predictions: a real-world evaluation
Roesler, Mathias W. , Wells, Cameron, Schamberg, Gabriel, Gao, Junyi , Harrison, Ewen, O'Grady, Gregory, Varghese, Chris , Harrison, Ewen M. , Schamberg, Gabriel , Harrison, Ewen
No container title available | March 5, 2026 | Volume: No volume available | Issue: No issue available | Pages: No page available
Background Predictive models employing machine learning algorithms are increasingly being used in clinical decision making, and improperly calibrated models can result in systematic harm. We sought to investigate the impact of class imbalance correction, a commonly applied preprocessing step in machine learning model development, on calibration and modelled clinical decision making in a large real-world context. Methods A histogram boosted gradient classifier was trained on a highly imbalanced national dataset of >1.8 million patients undergoing surgery, to predict the risk of 90-day mortality and complications after surgery. Class imbalance correction strategies including random oversampling, synthetic minority oversampling technique, random under-sampling, and cost-sensitive learning were compared to the natural distribution ('natural'). Models were tested and compared with classification metrics, calibration plots, decision curve analysis, and simulated clinical impact analysis. Results The natural model demonstrated high performance (AUROC 0.94, 95% CI 0.94--0.95 for mortality; 0.84, 95% CI 0.84--0.85 for complications) and calibration (log loss 0.05, 95% CI 0.04--0.05 for mortality; 0.23, 95% CI 0.23--0.24 for complications). Class imbalance mitigation (CSL, ROS, RUS, and SMOTE) did not improve AUROC or AUPRC but increased recall and F1 scores at the expense of precision and accuracy. However, these methods severely compromised model calibration, leading to significant over-prediction of risks (up to a 62.8 % increase) as further evidenced by increased log loss across all mitigation techniques. Decision curve analysis and clinical scenario testing confirmed that the natural model provided the highest net benefit. Conclusion Class imbalance correction methods result in significant miscalibration, leading to possible harm whenused for clinical decision making.