Assessment of nodal staging and risk factors for nodal involvement in gallbladder cancer: retrospective study Anita Balakrishnan PhD FRCSa, Petros Barmpounakis PhDb, Nikolaos Demiris PhDb, Bodil Andersson MD PhDc, Alejandro Brañes MDd, Xavier de Aretxabala MDe, Malin Sternby Eilard MD PhDf, Paul Gibbs MD FRCSa, Simon J F Harper MD FRCSa, Emmanuel L Huguet PhD FRCSa, Asif Jah FRCSa, Vasilis Kosmoliaptsis PhD FRCSa, Javier Lendoire MD PhDg, Siong S Liau MD FRCSa, Shishir Maithel MDh, Jack L Martin PhD FRCSa, Colin Noel MD, MMedi, Raaj K Praseedom FRCSa, Alejandro Serrablo MD PhDj, Volkan Adsay MDk for the OMEGA study investigators aDepartment of HPB Surgery, Cambridge University Hospitals NHS Foundation Trust, and Department of Surgery, University of Cambridge, Hills Road, Cambridge, CB2 0QQ, United Kingdom bCambridge Clinical Trials Unit – Cancer Theme, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge, CB2 0QQ, United Kingdom, and Department of Statistics, Athens University of Economics and Business, Athens, Greece cDepartment of Surgery, Lund University, Skane University Hospital, Lund, Sweden dDepartment of HPB Surgery, Hospital Sotero del Rio, Av. Concha y Toro 3459, Puente Alto, Región Metropolitana, Chile eDepartment of Digestive Surgery, Hepato-Pancreato-Biliary Surgery Unit, Surgery Service, Gallbladder Consortium Chile, Sotero del Rio Hospital and Clinica Alemana, Santiago, Chile f Transplantation Center, Sahlgrenska University Hospital and Department of Surgery, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden gDepartment of Surgery, University of Buenos Aires, Hospital Dr Cosme Argerich, Buenos Aires, Argentina hDepartment of Surgery, Emory University School of Medicine, Atlanta, Georgia, 30322 United States of America iGastrointestinal Surgery and HPB Surgery, Department of Surgery, University of the Free State, Bloemfontein, South Africa jDepartment of HPB Surgery, Miguel Servet University Hospital, Zaragoza, Spain kDepartment of Pathology, Koç University Hospital, Istanbul 34010, Turkey; Koç University Research Center for Translational Medicine (KUTTAM), Istanbul 34010, Turkey Corresponding author: Anita Balakrishnan, PhD FRCS Consultant Hepatopancreatobiliary Surgeon, Department of HPB Surgery, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge, CB2 0QQ, United Kingdom Email: ab2031@cam.ac.uk Tel: +44 1223 256 151 This manuscript is being submitted for publication as an original article and has not been previously submitted, published or presented. No funding was required for this study. The authors can confirm there are no conflicts of interest to declare. The collaborating authors within the OMEGA Study Investigators are listed in full at the end of the manuscript. Abstract Background: Nodal assessment in gallbladder cancer (GBC) remains challenging, particularly in incidental GBC. This understages node positive patients resulting in prognostic inaccuracy and insufficient adjuvant treatment. This study aimed to identify risk factors for positive nodes in GBC and compare prognostic discrimination of available nodal staging parameters. Methods: This international cohort study assessed GBC resections performed between 1st January 2010 and 31st December 2020. Logistic regression was used to identify risk factors for node-positive status and develop a risk prediction score for positive nodes. Nodal staging models including nodal site, positive node number and positive node ratio were compared for greatest prognostic discrimination in GBC. Results: A total of 3676 patients underwent GBC resection across 133 centres in 41 countries. T-stage (T2 ,p=0.012, T3, p=0.002, and T4, p<0.001), lymphovascular and perineural infiltration (LVPI, , p<0.001) and tumour differentiation (p<0.001) carried the greatest risk of positive nodes. These three parameters comprised the OMEGA Node Positivity Prediction Score (OMEGA-NOPPS) with C- statistics of 0.81 (95% CI 0.78- 0.84, train dataset) and 0.79 (0.73-0.85, test dataset) for identification of node-positive status, highlighting a 20% or greater increased risk of positive nodes in poorly differentiated tumours with LVPI despite T1 disease. Conclusions: Data from this large multicentre study confirmed that the number of positive nodes is the most discriminative prognostic model for nodal staging in GBC. The OMEGA-NOPPS stratification provided three simple parameters to risk stratify nodal involvement. Incidental GBC with LVPI and poorly-differentiated tumours, including early T-stages, should be considered for further treatment. Introduction: Gallbladder cancer (GBC) is rare but aggressive, and the commonest tumour of the biliary tract1,2. There have been no large prospective studies on surgical management of GBC, hence robust evidence for treatment strategies is lacking. Smaller studies lack applicable findings, while cancer registry datasets have insufficient granularity to impact practice. One key aspect of staging GBC is identifying nodal involvement, which suggests systemic dissemination and indicates adjuvant treatment. The Operative Management of Gallbladder Cancer (OMEGA) international study confirmed nodal stage as a key prognostic marker associated with recurrence-free survival (RFS)3–5. Yet lymphadenectomy rates remain low despite international guidelines and patients with incidental GBC (histologically identified following cholecystectomy for suspected benign disease) often had lymphadenectomy omitted at initial surgery 6–10. The resulting absence of nodal assessment may understage the 3.4% to 20% of T1 tumours reported to have positive nodes4,11–14. Better stratification of the risk of nodal involvement could identify patients with early but aggressive tumours and correspondingly higher risk of node positive disease who would benefit from adjuvant treatment. Furthermore there is ongoing debate regarding the optimal nodal parameters to guide prognosis, disputing the widely used 8th AJCC classification in favour of other parameters such as positive lymph node ratios (PLNR) 15–21. In this publication the large OMEGA study dataset was interrogated to identify the optimal prognostic nodal parameter and to develop a risk score for nodal involvement in GBC. Methods Data collection Collaborating centres were recruited to the OMEGA study by invitation disseminated via emails to all members of the three international HPB associations (the European-African HPB Association, the Americas HPB Association, and the Asia-Pacific HPB Association). Inclusion and exclusion criteria Patients were included in this study if they had undergone surgery between 1st January 2010 and 31st December 2020 for either pre-operatively diagnosed or incidental (identified following cholecystectomy for benign disease) GBC. Exclusion criteria were high-grade dysplasia only, metastatic disease at surgery, or macroscopic tumour remaining at the end of resection (R2). Demographic and pathologic data collection Clinical parameters, operative details, pathological findings as well as follow-up and survival data were obtained from institutional databases. Comorbidities were incorporated in the Charlson Comorbidity Index (CCI). Operative details Extent of surgery was defined as cholecystectomy only, wedge resection (taking the liver margin at the gallbladder bed), resection of liver segments IVb and V, or major hepatectomy (right, extended left or extended right hemihepatectomy). Data was also obtained on specific nodal groups resected, PLNN, total lymph nodes excised (TLNE), extrahepatic bile duct resection (EBDR), resection of additional organs and surgical approach. Complications within 30 post-operative days were classified using the Clavien- Dindo scale22. Histology was recorded using the 8th American Joint Committee on Cancer (AJCC) classification (2018) for GBC, incorporating parameters such as differentiation status and lymphovascular or perineural invasion (LVPI)23. Follow-up and survival RFS was defined as the interval between date of surgery and date of first recurrence on imaging. Ethical approval Ethical approval was obtained from the Research and Development Office at Cambridge University Hospitals NHS Foundation Trust (the lead site) and the United Kingdom Research Ethics Committee (IRAS ID 285918). Other participating centres obtained further approvals as needed. This study was conducted and reported in compliance with the STROBE guidelines for cohort studies.18 Statistical analysis Factors associated with TLNE were analysed using Poisson and Negative Binomial regression models and data expressed as risk ratios (RR). Binary data were analysed using logistic regression, recursive partitioning classification tree and random forest models. Survival data were analysed using the Cox model and Kaplan Meier plots. Separate multivariable analyses, excluding patients without lymphadenectomy, were employed to assess the association of PLNR and AJCC nodal classification with RFS due to collinearity between those variables. Recursive partitioning classification was also used to select the most suitable cut-off value for PLNN and PLNR. The predictive ability and goodness-of-fit of the prognostic models were compared using the Akaike Information Criterion (AIC) and C-statistic. To develop the risk prediction model for node involvement, the dataset was split into 80% and 20% train and test sets and performed 10-fold cross-validation repeated 10 times on the train dataset to select hyperparameters for the tree and random forest. The predictive discrimination of the risk score was assessed on the test dataset as a separate validation cohort. The final risk score was derived by multiplying the coefficients of the logistic regression model by 3, rounding to the nearest integer, then adding the transformed coefficient. No data imputation was used and patients were only included in this analysis if data on all the parameters were available. Full statistical methodology is available in Supplementary Data. Results The OMEGA study identified 4138 patients who had undergone surgery for GBC in 41 countries across 133 participating centres (Supplementary Figure S1). After excluding 78 patients with only high grade dysplasia and 384 patients with metastatic disease/ R2 resection, 3676 patients were finally included for analysis. Median OS and RFS for the whole cohort were 51.2 (49.3-52.8) months and 35.2 (34.3-36.9) months, with median follow up of 45.3 (24.1-80.5) months3. TLNE Some 3227 patients underwent lymphadenectomy, and 1820 patients had > 6 nodes excised. The median TLNE was 5 (interquartile range 2-9). No lymphadenectomy was performed in 48.1% (n=90) and 18.6% (n=75) of patients with T1a and T1b disease respectively. Greater TLNE was associated with T2 tumours and above on multivariable analysis compared to T1a (Table 1, Figure 1a), and poorer TLNE with R1 (margin positive) disease (RR 0.80, 95% CI 0.74-0.88 versus R0). Extent of liver resection showed no association with TLNE, while EBDR was associated with more TLNE compared to no EBDR (RR 1.20 [1.12-1.28]. A laparoscopic approach was associated with less TLNE than an open approach while a robotic approach was associated with more TLNE versus open (RR 0.86 [0.78-0.96] and RR 1.39 [1.11-1.77] respectively). Extending nodal dissection beyond the cystic node to regional (perihilar), coeliac and para-aortic nodes was associated with a 3, 4 and 6-fold increased TLNE respectively, with no associated increased morbidity or mortality on a separate multivariable analysis (Supplementary Figure S2a). TLNE was not associated with age, gender, comorbidities, or the need for a second operation for disease clearance (Figure 1a). T-stage Positive lymph nodes (N+) were identified in 1353 patients across the entire cohort (Table 1). Some 1081 had N1 disease (up to 3 positive nodes) while 272 patients had N2 disease (>3 positive nodes). N+ were predominantly associated with higher T stages on multivariable analysis, but were also present in 2.6% of T1a cases (n=5) and 8.4% T1b cases (n=34, Table 1, Figure 1b). Poorly differentiated tumours and LVPI were also associated with N+ (OR 1.74 [1.31-2.32] p<0.001, and OR 3.25 [2.61-4.06] p<0.001 respectively). Suspected nodal involvement on pre-operative CT, MRI or PET-CT scans was also significantly associated with N+ (OR 2.39 [1.85-3.09], p<0.001). TLNE had a small but significant association with N+ (OR 1.04 [1.03-1.06], p<0.001). Non-regional node dissection was not associated with increased N+ (Supplementary Figure S2b). Prognostic association with RFS Higher T-stages, margin positive disease, and N+ had the greatest association with RFS on multivariable analysis (Figure 2). Non-regional node dissection and TLNE had no significant effect on RFS (Supplementary Figure S3a and b). Survival recursive partitioning tree models identified the PLNN of 3 nodes as the most discriminatory cut- off for RFS , and associated with worse RFS on multivariable analysis (p<0.001,Figure 2 and Supplementary Figure S4a). N+ in any nodal group was associated with worse RFS compared to N0 disease; particularly para-aortic nodes (p<0.001, Supplementary Figure S5a). Using the same recursive partitioning methodology as above, a threshold of >0.1 was identified as the best discriminator of RFS for PLNR, and associated with worse RFS on multivariable analysis (p<0.001, Supplementary Figures S4c, S5c). PLNR was associated with a 3-fold hazard to RFS when assessed as a continuous variable (p<0.001, Supplementary Figure S5d). Scoring system for node positive status in incidental GBC The three parameters chosen for the OMEGA-NOPPS, namely T-stage, tumour differentiation and LVPI, had all exhibited significant association with N+ and are routinely available in patients with incidental GBC (Table 3). This analysis utilised 1141 individuals from the main cohort for whom the relevant data were available, split into 80:20 train:test datasets. The train dataset (n=912) exhibited a C-statistic of 0.81 (95% CI 0.78-0.84), sensitivity of 70.6% and specificity of 76.9%. The separate test or validation cohort showed good generalizability of the model (n=229), with a C-statistic of 0.79 (0.73-0.85), sensitivity of 72.3% and specificity of 74.8%. Calibration curves showed an agreement between predicted and observed outcomes, without signs of overfitting or underfitting (Supplementary Figure S7). Notably, OMEGA-NOPPS quantifies the additional risk conferred by tumour differentiation status and LVPI on top of T-stage. Discussion In this multicentre study T-stage, differentiation and LVPI were identified as key risk factors for node positive disease and positive node number was confirmed as the most important prognostic nodal factor for RFS. This study showed that coeliac and para-aortic nodal dissection could be performed safely although good nodal yields can be obtained with regional lymphadenectomy alone. EBDR was associated with more lymph nodes retrieved, in keeping with some studies, however increasing data indicates that good TLNEs can still be obtained without EBDR, which is no longer routinely recommended in GBC resection due to increased morbidity7,24–27. The 8th AJCC classification was introduced in 2018, stratifying GBC nodal stage by PLNN rather than nodal site, however subsequent studies have queried the value of the 8th AJCC classification compared to other staging parameters such as PLNR16,18,23,28–32. The 8th AJCC classification had a small but clear advantage over the other parameters in this study, which unlike previous studies combines a large cohort with the necessary granularity and a high lymphadenectomy rate17,18,28,31,32. In contrast to other nodal staging prognostic models, the 8th AJCC classification requires no complex calculations and is widely used in histopathological reporting internationally. This makes it feasible, practical, and easily reproducible for prognostication of node positive disease and decisions regarding adjuvant treatment. This study showed that poorly differentiated tumours with LVPI would carry 20% or higher risk of node positive disease. Cholecystectomy alone is increasingly considered sufficient treatment for T1 tumours, however up to 10% of such patients could harbour positive nodes, overlooked without a lymphadenectomy, and thus not receive adjuvant treatment 6–10,12,14,36. OMEGA-NOPPS therefore informs consideration of further surgery in early stage tumours, and, for higher scores, may allow consideration of adjuvant treatment without the cost and morbidity of confirmatory lymphadenectomy. Given the encouraging promise offered by ongoing adjuvant and neoadjuvant chemo- and immunotherapy trials in GBC, this is a key opportunity to maximise benefit in a group of patients for whom long-term survival may be possible37–40. The 3 parameters in OMEGA-NOPPS are routine histopathology parameters mandated by international histopathological reporting guidelines, readily available in incidental GBC, and easily incorporated in tumour board discussions 41,42 . Additionally, the use of parameters beyond T-stage may counterbalance any inadvertent understaging of tumour extent in cholecystectomy specimens. OMEGA-NOPPS complements existing scores for GBC recurrence, survival and adjuvant treatment, facilitating more personalised prognostication and management35,43,44. Importantly, the wide global population within this study should render this score internationally applicable in both high and low incidence areas and allow further validation in a prospective multicentre setting. The main strength of this study is the large number of patients and centres recruited which allows a truly global picture of the practice of GBC surgery, with a high lymphadenectomy rate and incorporating a level of detail not available in single-centre cancer registries. Nevertheless, this study has a number of limitations. Given the retrospective nature of the study, specific data on the number of nodes retrieved and exact nodal sites was not always available, and the existence of anatomically continuous nodal chains rendered further detailed breakdown of adjacent nodal subgroups difficult. Lymphadenectomy rates for T1a and T1b disease in this study were much higher than other studies, particularly compared to registry-based data where nodal staging was absent in almost half the patients with T1b disease10,45. This may be due to increased lymphadenectomy rates in recent years, but may also represent differences in treatment strategies as well as overall and recurrence-free survival rates between tertiary HPB centres in this study and the more heterogenous data pool in national registries46. The exact reason for initial cholecystectomy was not always known, hence it was not possible to separately analyse patients with incidental early stage GBC. OMEGA-NOPPS relies on pathological variables only available post- operatively in cases of incidental gallbladder cancer. Improved imaging techniques and novel molecular biomarkers are necessary for better prediction of nodal metastases for pre-operatively detected GBC and thus a key area for future research. The data from this study showed that identification of four or more positive nodes has better prognostic discrimination than nodal ratio or more distant positive nodes, highlighting the importance of a thorough regional lymphadenectomy to accurately stage gallbladder cancer. The new OMEGA-NOPPS model provides a simple three parameter score to risk stratify nodal involvement and highlight patients with LVPI and poorly-differentiated incidental GBC who may otherwise be overlooked for further treatment, ensuring a more nuanced discussion and personalised treatment of this rare and aggressive tumour. References: 1. Roa JC, García P, Kapoor VK, Maithel SK, Javle M, Koshiol J. Gallbladder cancer. Nat Rev Dis Primers. 2022;8(1):69. doi:10.1038/s41572-022-00398-y 2. Ferlay J, Ervik M, Lam F, et al. Global Cancer Observatory: Cancer Today. Lyon, France: International Agency for Research on Cancer. Published online 2020. Accessed October 12, 2022. https://gco.iarc.fr/today 3. Balakrishnan A, Barmpounakis P, Demiris N, et al. Surgical outcomes of gallbladder cancer: the OMEGA retrospective, multicentre, international cohort study. EClinicalMedicine. 2023;59:101951. doi:10.1016/j.eclinm.2023.101951 4. Goel M, Pandrowala S, Patel P, Patkar S. Node positivity in T1b gallbladder cancer: A high volume centre experience. Eur J Surg Oncol. 2022;48(7):1585-1589. doi:10.1016/j.ejso.2022.03.013 5. Bridgewater J, Fletcher P, Palmer DH, et al. Long-Term Outcomes and Exploratory Analyses of the Randomized Phase III BILCAP Study. J Clin Oncol. 2022;40(18):2048-2057. doi:10.1200/JCO.21.02568 6. Benson AB, D’Angelica MI, Abrams TA, et al. Hepatobiliary cancers, version 2.2014. J Natl Compr Canc Netw. 2014;12(8):1152-1182. doi:10.6004/jnccn.2014.0112 7. Aloia TA, Járufe N, Javle M, et al. Gallbladder cancer: expert consensus statement. HPB (Oxford). 2015;17(8):681-690. doi:10.1111/hpb.12444 8. Vogel A, Bridgewater J, Edeline J, et al. Biliary tract cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol. Published online November 10, 2022:S0923-7534(22)04699-3. doi:10.1016/j.annonc.2022.10.506 9. Kemp Bohan PM, O’Shea AE, Ellis OV, et al. Rates, Predictors, and Outcomes of Portal Lymphadenectomy for Resectable Gallbladder Cancer. Ann Surg Oncol. 2021;28(6):2960- 2972. doi:10.1245/s10434-021-09667-8 10. Vo E, Curley SA, Chai CY, Massarweh NN, Tran Cao HS. National Failure of Surgical Staging for T1b Gallbladder Cancer. Ann Surg Oncol. 2019;26(2):604-610. doi:10.1245/s10434-018-7064-7 11. Shao J, Lu HC, Wu LQ, Lei J, Yuan RF, Shao JH. Simple cholecystectomy is an adequate treatment for grade I T1bN0M0 gallbladder carcinoma: Evidence from 528 patients. World J Gastroenterol. 2022;28(31):4431-4441. doi:10.3748/wjg.v28.i31.4431 12. Lee H, Kwon W, Han Y, Kim JR, Kim SW, Jang JY. Optimal extent of surgery for early gallbladder cancer with regard to long-term survival: a meta-analysis. J Hepatobiliary Pancreat Sci. 2018;25(2):131-141. doi:10.1002/jhbp.521 13. Rhodin KE, Goins S, Kramer R, et al. Simple versus radical cholecystectomy and survival for pathologic stage T1B gallbladder cancer. HPB (Oxford). 2024;26(4):594-602. doi:10.1016/j.hpb.2024.01.012 14. Pehlivanoglu B, Akkas G, Memis B, et al. Reappraisal of T1b gallbladder cancer (GBC): clinicopathologic analysis of 473 in situ and invasive GBCs and critical review of the literature highlights its rarity, and that it has a very good prognosis. Virchows Arch. Published online December 29, 2022. doi:10.1007/s00428-022-03482-6 15. Birnbaum DJ, Viganò L, Russolillo N, Langella S, Ferrero A, Capussotti L. Lymph node metastases in patients undergoing surgery for a gallbladder cancer. Extension of the lymph node dissection and prognostic value of the lymph node ratio. Ann Surg Oncol. 2015;22(3):811-818. doi:10.1245/s10434-014-4044-4 16. Amini N, Spolverato G, Kim Y, et al. Lymph node status after resection for gallbladder adenocarcinoma: prognostic implications of different nodal staging/scoring systems. J Surg Oncol. 2015;111(3):299-305. doi:10.1002/jso.23813 17. Amini N, Kim Y, Wilson A, et al. Prognostic Implications of Lymph Node Status for Patients With Gallbladder Cancer: A Multi-Institutional Study. Ann Surg Oncol. 2016;23(9):3016-3023. doi:10.1245/s10434-016-5243-y 18. Wang J, Liu F, Ma W, Hu H, Li F. Metastatic lymph node ratio as an important prognostic factor in advanced gallbladder carcinoma with at least 6 lymph nodes retrieved. Langenbecks Arch Surg. 2023;408(1):382. doi:10.1007/s00423-023-03119-5 19. Xiao Z, Shi Z, Hu L, et al. A new nomogram from the SEER database for predicting the prognosis of gallbladder cancer patients after surgery. Ann Transl Med. 2019;7(23):738. doi:10.21037/atm.2019.11.112 20. Moazzam Z, Alaimo L, Endo Y, Lima HA, Pawlik TM. Predictors, Patterns, and Impact of Adequate Lymphadenectomy in Intrahepatic Cholangiocarcinoma. Ann Surg Oncol. 2023;30(4):1966-1977. doi:10.1245/s10434-022-13044-4 21. Zheng W, Jiang W, Wu Q, et al. Comparisons of different lymph node staging systems for predicting overall survival of node-positive patients with renal cell carcinoma: a retrospective cohort study using the Surveillance, Epidemiology and End Results database. BMJ Open. 2023;13(4):e068044. doi:10.1136/bmjopen-2022-068044 22. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg. 2004;240(2):205-213. doi:10.1097/01.sla.0000133083.54934.ae 23. Amin, M. 8th AJCC Staging Manual. Springer; 2017. 24. Lv TR, Liu F, Hu HJ, et al. The role of extra-hepatic bile duct resection in the surgical management of gallbladder carcinoma. A first meta-analysis. Eur J Surg Oncol. 2022;48(3):482-491. doi:10.1016/j.ejso.2021.11.131 25. Fuks D, Regimbeau JM, Le Treut YP, et al. Incidental gallbladder cancer by the AFC- GBC-2009 Study Group. World J Surg. 2011;35(8):1887-1897. doi:10.1007/s00268-011- 1134-3 26. Kosuge T, Sano K, Shimada K, Yamamoto J, Yamasaki S, Makuuchi M. Should the bile duct be preserved or removed in radical surgery for gallbladder cancer? Hepatogastroenterology. 1999;46(28):2133-2137. 27. Gani F, Buettner S, Margonis GA, et al. Assessing the impact of common bile duct resection in the surgical management of gallbladder cancer. Journal of Surgical Oncology. 2016;114(2):176-180. doi:10.1002/jso.24283 28. Chen C, Rui Z, Yuhan W, et al. Optimal Lymph Node Staging System in Evaluating Prognosis of Gallbladder Carcinoma: A Multi-institutional Study. Ann Surg Oncol. 2021;28(13):8142-8151. doi:10.1245/s10434-021-10749-w 29. Lee W, Jeong CY, Kim YH, et al. Validation of the prognostic performance in various nodal staging systems for gallbladder cancer: results of a multicenter study. Langenbecks Arch Surg. 2019;404(5):581-588. doi:10.1007/s00423-019-01807-9 30. Maegawa FB, Ashouri Y, Hamidi M, Hsu CH, Riall TS. Gallbladder Cancer Surgery in the United States: Lymphadenectomy Trends and Impact on Survival. Journal of Surgical Research. 2021;258:54-63. doi:10.1016/j.jss.2020.08.041 31. Shirai Y, Sakata J, Wakai T, Ohashi T, Ajioka Y, Hatakeyama K. Assessment of lymph node status in gallbladder cancer: location, number, or ratio of positive nodes. World J Surg Onc. 2012;10(1):87. doi:10.1186/1477-7819-10-87 32. Negi SS, Singh A, Chaudhary A. Lymph Nodal Involvement as Prognostic Factor in Gallbladder Cancer: Location, Count or Ratio? Journal of Gastrointestinal Surgery. 2011;15(6):1017-1025. doi:10.1007/s11605-011-1528-4 33. Dominguez DA, Aversa JG, Hagerty BL, et al. Prognostic impact of lymphovascular invasion in pT1-T3 gallbladder adenocarcinoma. J Surg Oncol. 2020;122(7):1401-1408. doi:10.1002/jso.26160 34. Ruff SM, Cloyd JM, Pawlik TM. Annals of Surgical Oncology Practice Guidelines Series: Management of Primary Liver and Biliary Tract Cancers. Ann Surg Oncol. 2023;30(13):7935-7949. doi:10.1245/s10434-023-14255-z 35. Ethun CG, Postlewait LM, Le N, et al. A Novel Pathology-Based Preoperative Risk Score to Predict Locoregional Residual and Distant Disease and Survival for Incidental Gallbladder Cancer: A 10-Institution Study from the U.S. Extrahepatic Biliary Malignancy Consortium. Ann Surg Oncol. 2017;24(5):1343-1350. doi:10.1245/s10434-016-5637-x 36. Balakrishnan A, Jah A, Lesurtel M, et al. Heterogeneity of management practices surrounding operable gallbladder cancer – results of the OMEGA-S international HPB surgical survey. HPB. Published online July 2022:S1365182X22015143. doi:10.1016/j.hpb.2022.06.014 37. The OPT-IN study, Clinical Trials.gov Identifier:NCT04559139. https://clinicaltrials.gov/ct2/show/NCT04559139 38. Goetze TO, Bechstein WO, Bankstahl US, et al. Neoadjuvant chemotherapy with gemcitabine plus cisplatin followed by radical liver resection versus immediate radical liver resection alone with or without adjuvant chemotherapy in incidentally detected gallbladder carcinoma after simple cholecystectomy or in front of radical resection of BTC (ICC/ECC) - a phase III study of the German registry of incidental gallbladder carcinoma platform (GR)- the AIO/ CALGP/ ACO- GAIN-trial. BMC Cancer. 2020;20(1):122. doi:10.1186/s12885-020- 6610-4 39. Ozer M, Goksu SY, Sanford NN, et al. A Propensity Score Analysis of Chemotherapy Use in Patients With Resectable Gallbladder Cancer. JAMA Netw Open. 2022;5(2):e2146912. doi:10.1001/jamanetworkopen.2021.46912 40. Stein A, Arnold D, Bridgewater J, et al. Adjuvant chemotherapy with gemcitabine and cisplatin compared to observation after curative intent resection of cholangiocarcinoma and muscle invasive gallbladder carcinoma (ACTICCA-1 trial) - a randomized, multidisciplinary, multinational phase III trial. BMC Cancer. 2015;15:564. doi:10.1186/s12885-015-1498-0 41. Albores-Saavedra J, Kloppel G, Adsay N, et al. Tumours of the Gallbladder and Extrahepatic Bile Ducts. Carcinoma of the Gallbladder and Extrahepatic Bile Ducts. Fred T. Bosman, Fatima Carneiro, Ralph H. Hruban, Neil D. Theise, Eds. In: WHO Classification of Tumours of the Digestive System.; 2010. 42. Roa J, Adsay N, Arola J, Tsui W, Zen Y. Tumours of the Gallbladder and Extrahepatic Bile Ducts. Carcinoma of Gallbladder Edited by International Agency for Research in Cancer and the WHO Classification of Tumours Editorial Board. In: WHO Classification of Tumours of the Digestive System. 5th ed.; 2019. 43. Sahara K, Tsilimigras DI, Kikuchi Y, et al. Defining and Predicting Early Recurrence after Resection for Gallbladder Cancer. Ann Surg Oncol. 2021;28(1):417-425. doi:10.1245/s10434-020-09108-y 44. Xiang J, Zhang X, Weber SM, et al. Identification of patients who may benefit the most from adjuvant chemotherapy following resection of incidental gallbladder carcinoma. Journal of Surgical Oncology. 2021;123(4):978-985. doi:10.1002/jso.26389 45. Kwon W, Kim H, Han Y, et al. Role of tumour location and surgical extent on prognosis in T2 gallbladder cancer: an international multicentre study. Br J Surg. 2020;107(10):1334- 1343. doi:10.1002/bjs.11618 46. Hamad A, Cloyd JM, Dillhoff M, et al. Comparison of lymph node evaluation and yield among patients undergoing open and minimally invasive surgery for gallbladder adenocarcinoma. Surg Endosc. 2021;35(5):2223-2228. doi:10.1007/s00464-020-07635-z Figure legends: Figure 1: A) Multivariable analysis of factors associated with total lymph nodes excised and B) multivariable analysis of factors associated with positive node number. A) Forest plot and multivariable Negative Binomial regression analysis of factors associated with total lymph nodes excised (TLNE) and B) Forest plot and logistic regression analysis of factors associated with positive node number (N+). EBDR – extrahepatic bile duct resection, HR – hazard ratio, CI – confidence intervals, LVPI – lymphovascular or perineural invasion Figure 2: Multivariable analysis of factors associated with recurrence-free survival. Forest plot and multivariable Cox proportional hazard regression analysis of factors associated with recurrence-free survival (RFS). EBDR – extrahepatic bile duct resection, HR – hazard ratio, CI – confidence intervals Table 1: Demographic and clinical data according to nodal stage Parameter N0 N1 N2 Nx Median TLNE (IQR) All patients (n=3676) 1874 (51.0%) 1081 (29.4%) 272 (7.4%) 449 (12.2%) 5 (2,9) Age <60 (n=1044) 550 (52.7%) 316 (30.3%) 86 (8.2%) 92 (8.8%) 5 (4,6) Age >60 (n=2632) 1324 (50.3%) 765 (29.0%) 186 (7.1%) 357 (13.6%) 6 (4,8) Male (n=1252) 635 (50.7%) 361 (28.8%) 91 (7.3%) 165 (13.2%) 5 (2,9) Female (n=2424) 1239 (51.1%) 720 (29.7%) 181 (7.5%) 284 (11.7%) 5 (2,10) CCI <5 (n=2133) 1115 (52.2%) 646 (30.3%) 166 (7.9%) 206 (9.7%) 6 (3,10) CCI >5 (n=1321) 655 (49.6%) 376 (28.4%) 99 (7.5%) 191 (14.5%) 5 (1,9) Tumour differentiation Well (n=837) 522 (62.3%) 155 (18.5%) 27 (3.2%) 133 (15.9%) 5 (1,9) Moderate (n=1489) 731 (49.1%) 503 (33.8%) 109 (7.3%) 146 (9.8%) 5 (2,9) Poor (n=795) 305 (38.4%) 290 (36.5%) 114 (14.3%) 86 (10.8%) 6 (2,7) LVPI present (n=1644) 591 (35.9%) 710 (43.1%) 208 (12.7%) 135 (8.2%) 6 (3,11) LVPI absent (n=1176) 788 (67.0%) 170 (14.4%) 23 (2.0%) 195 (16.6%) 4 (1,8) Histology T1a (n=187) 92 (49.2%) 4 (2.1%) 1 (0.5%) 90 (48.1%) 1 (0,4) T1b (n=403) 294 (73.0%) 31 (7.7%) 3 (0.7%) 75 (18.6%) 4 (1,8) T2 (n=1700) 991 (58.3%) 461 (27.1%) 68 (4.0%) 180 (10.6%) 5 (2,9) T3 (n=1186) 453 (38.2%) 487 (41.1%) 151 (12.7%) 95 (8.0%) 6 (3,10) T4 (n=200) 44 (22.0%) 98 (49.0%) 49 (24.5%) 9 (4.5%) 9 (5,15) R0 (n=3187) 1748 (54.8%) 869 (27.3%) 213 (6.7%) 357 (11.2%) 5 (2,9) R1 (n=472) 120 (25.4%) 210 (44.5%) 58 (12.3%) 84 (17.8%) 4 (1,8) Access Open (n=3162) 1625 (51.4%) 996 (31.5%) 261 (8.3%) 280 (8.9%) 6 (2,10) Laparoscopic (n=465) 218 (46.9%) 74 (15.9%) 9 (1.9%) 164 (35.2%) 1 (0,6) Robotic (n=36) 24 (66.7%) 9 (25.0%) 1 (2.8%) 2 (3.6%) 6 (3,11) EBDR performed (n=976) 380 (38.9%) 412 (42.2%) 141 (14.4%) 43 (4.4%) 7 (3,12) EBDR not performed (n=2700) 1494 (55.3%) 669 (24.8%) 131 (4.9%) 406 (15.0%) 4 (1,8) Extent of resection Cholecystectomy alone (n=557) 197 (35.4%) 77 (13.8%) 10 (1.8%) 273 (49.0%) 0 (0,2) Wedge resection (n=1407) 817 (58.1%) 426 (30.3%) 79 (5.6%) 85 (6.0%) 6 (3,10) SIVb/V resection (n=1397) 755 (54.0%) 436 (31.2%) 131 (9.4%) 75 (5.4%) 6 (3,9) Major hepatectomy (n=315) 105 (33.3%) 142 (45.1%) 52 (16.5%) 16 (5.1%) 7 (3,13) Extent of nodal dissection Cystic (n=142) 100 (70.4%) 23 (16.2%) 0 (0%) 19 (13.4%) 1 (1,6) Regional (n=2332) 1286 (55.1%) 779 (33.4%) 184 (7.9%) 33 (3.6%) 6 (3,9) Coeliac (n=267) 140 (52.4%) 92 (34.4%) 33 (12.3%) 2 (0.7%) 7 (4,13) Para-aortic (n=417) 236 (56.6%) 126 (30.2%) 51 (12.2%) 4 (1.0%) 11 (7,17) Single stage surgery (n=2139) 882 (41.2%) 692 (32.3%) 196 (9.2%) 369 (17.3%) 5 (1,9) Two stage surgery (n=1537) 992 (64.5%) 389 (25.3%) 76 (4.9%) 80 (5.2%) 5 (3,9) High prevalence (n=1209) 638 (52.8%) 371 (30.7%) 96 (7.9%) 104 (8.6%) 8 (4,12) Low prevalence (n=2467) 1236 (50.1%) 710 (28.8%) 176 (7.1%) 345 (14.0%) 4 (1,8) Abbreviations: TLNE – total lymph nodes excised, CCI – Charlson Comorbidity Index, LVPI - lymphovascular or perineural invasion, EBDR – extrahepatic bile duct resection Table 2: Discriminative ability of prognostic models for nodal staging Prognostic model AIC C-index (SE) Model thresholds HR (95% CI)* p-value PLNR continuous 14720.27 0.7581 (0.0069) N/A 3.00 (2.44- 3.69) p<0.001 PLNR categorical (>0.1) 14692.51 0.7615 (0.0068) PLNR <0.1 1.00 (Ref) N/A PLNR >0.1 2.22 (1.93- 2.55) p<0.001 AJCC 7th edition 14694.96 0.7603 (0.0069) N0 1.00 (Ref) N/A N1 2.19 (1.9- 2.52) p<0.001 N2 2.64 (1.9- 3.67) p<0.001 AJCC 8th edition 14660.80 0.7633 (0.0068) N0 1.00 (Ref) N/A N1 2.04 (1.76- 2.36) p<0.001 N2 3.66 (2.95-4.53) p<0.001 *Hazard ratios (HR) imported from individual multivariable analyses using each of the prognostic models, as per Supplementary Figures 2 and 3. Abbreviations: AIC – Akaike Information Criterion, CI – confidence intervals, PLNR – positive lymph node ratio. Table 3: OMEGA Node Positivity Prediction Score (OMEGA-NOPPS) stratification for nodal involvement in gallbladder cancer. As an example, a patient with a T1b tumour would be expected to have a 11.9% risk of N+ if they had a well-differentiated tumour without LVPI, but a 50% risk with a poorly-differentiated tumour and LVPI. Individual parameters and scores T stage T1a 0 T1b 3 T2 5 T3 7 T4 10 Tumour differentiation Well 0 Moderate 1 Poor 2 Lymphovascular or perineural invasion No 0 Yes 4 Final scores and associated risk Final risk score Probability of positive nodes 0-2 Up to 6.5% 2-4 6.5 – 11.9% 4-6 11.9 – 20.9% 6-8 20.9 – 33.9% 8-10 33.9 – 50.0% 10-12 50.0 – 66.1% 12-14 66.1 – 79.1% 14-16 79.1 – 88.1% Collaborators: Tomoyuki Abe Moh'd Abu Hilal Maria del Mar Achalandabaso Boira Mustapha Adham Mohamed Adam Maryam Ahmad Bilal Al-Sarireh Maite Albiol Nassir Alhaboob Adnan Alseidi Houssem Ammar Akshay Anand Bodil Andersson Pantelis Antonakis Veronica Araya Stanley W Ashley Georgi Atanasov Fabio Ausania Ricardo Balestri Abhirup Banerjee Sudeep Banerjee Simon Banting Giedrius Barauskas Fabian Bartsch Andrea Belli Simona Beretta Frederik Berrevoet Ramesh Singh Bhandari Gerardo Blanco Fernandez Louisa Bolm Mathieu Bonal Emre Bozkurt Andries E Braat Luke Bradshaw Konstantinos Bramis Alejandro Branes Lyle Burdine Matthew Byrne Maria Caceres Maria Jesus Castro Santiago Benjamin Chan Lynn Chong Ahmet Çoker Maria Conde Rodriguez Daniel Croagh Alyn Crutchley Carmen Cutolo Mathieu D'Hondt Daniel D'Souza Freek Daams Raffaele Dalla Valle José Davide Mario de Bellis Marieke de Boer Celine de Meyere Philip de Reuver Matthew Dixon Panagiotis Dorovinis Gabriela Echeverría Bauer Maria Eduarda Hasan Eker Joris Erdmann Mert Erkan Evangelos Felekouras Emanuele Felli Eduardo Fernandes Eduardo Figueroa Rivera Andras Fulop Daniel Galun Michael Gerhards Poya Ghorbani Fabio Giannone Luis Gil Emmanouil Giorgakis Mario Giuffrida Felice Giuliante Ioannis Gkekas Miguel Gomez Bravo Bas Groot Koerkamp Oscar Guevara Alfredo Guglielmi Aiste Gulla Rahul Gupta Amit Gupta Marta Gutiérrez Abu Bakar Hafeez Bhatti Jeroen Hagendoorn Zain Hajee Abdul Rahman Hakeem Hytham Hamid Sayed Hassen Stefan Heinrich Roberto Hernandez-Alejandro Ryota Higuchi Daniel Hoffman David Holroyd Daniel Hughes Arpad Ivanecz Satheesh Iype Isabel Jaen Torrejimeno Shantanu Joglekar Robert Jones Klaus Kaczirek Harsh Kanhere Ambareen Kausar Zhanyi Kee Jessica Keilson Jorg Kleef Johannes Klose Brett Knowles Jun Kit Koong Nagappan Kumar Supreeth Kunnuru Paleswan Joshi Lakhey Andrea Laurenzi Yeong Sing Lee Felipe Leon Voon Meng Leow Jean-Baptiste Lequeu Mickael Lesurtel Elisabeth Lo Stefan Löb Elizabeth Lockie Peter Lodge Dolores López Garnica Victor Lopez Lopez Linda Lundgren Nikolaos Machairas Dhiresh Maharjan Deep Malde Marc Mankarious Guillaume Martel Julie Martin Michele Mazzola Arianeb Mehrabi Ricardo Memeo Flavio Milana George Molina Leah Monette Haluk Morgul Dimitrios Moris Antonios Morsi-Yeroyannis Nicholas Mowbray Francesk Mulita Edoardo Maria Muttillo Malith Nandasena Pueya Rashid Nashidengo Arash Nickkholgh Colin Byron Noel Masayuki Ohtsuka Arturs Ozolins Sanjay Pandanaboyana Nikolaos Pararas Alessandro Parente June Peng Arkaitz Perfecto Valero Julie Perinel Konstatinos Perivoliotis Teresa Perra Patrick Pessaux Natalie Petruch Gaetano Piccolo Laszlo Piros Alberto Porcu Viswakumar Prabakaran Raj Prasad Mikel Prieto Calvo Florian Primavesi Eva Maria Pueyo Periz Alberto Quaglia Jose M Ramia Angel Ashwin Rammohan Francesco Razionale Ricardo Robles Campos Manas Roy Sophie Rozwadowski Luis Ruffolo Natalia Ruiz Andrea Ruzzenante Lily Saadat Mohamed Amine Said Edoardo Saladino Gabriel Saliba Per Sandstrom Carlo Alberto Schena Anthony Scholer Christoph Schwarz Lorenzo Serafini Leyre Serrablo Pablo E Serrano Deepak Sharma Aali Sheen Vishwanath Siddagangaiah Michael Silva Saurabh Singh Ajith Siriwardena Michal Skalski Mante Smig Faris Soliman Abhinav Arun Sonkar Donzília Sousa Silva Ernesto Sparrelid Harry V M Spiers Parthi Srinivasan Malin Sternby Eilard Oliver Strobel Urban Stupan Miguel Angel Suarez-Munoz Manisekar Subramaniam Teiichi Sugiura Robert Sutcliffe Hilko Swank Shibojit Talukder Lillian Taylor Prabin Bikram Thapa Catherine The Asara Thepbunchonchai Caman Thieu Navneet Tiwari Guido Torzilli Chutwichai Tovikkai Blaz Trotovsek Savvas Tsaramanidis Georgios Tsoulfas Katsuhiko Uesaka Garzali Umar Lucio Urbani Michail Vailas Ronald van Dam Peter van de Boezem Stijn van Laarhoven Tomas Vanagas Mike Van Dooren Manon Viennet Luca Vigano Aarathi Vijayashanker Celia Villodre Toshifumi Wakai Aklile Workneh Li Xu Masakazu Yamamoto Zhiying Yang Robert Young Marko Zivanovic