Salivary gland tumors are rare and morphologically diverse, posing both diagnostic and scientific challenges. This study presents the first phase of the SALV-Dataset Registry; a nationwide, expertly curated, and fully digitized clinicopathological resource, designed to support research and develop artificial intelligence (AI) tools assisting in salivary gland tumor pathology diagnostics. Salivary gland tumor resections diagnosed at the Leiden University Medical Center (1999–2024) were collected through the Dutch national network and registry for histo- and cytopathology (PALGA). In total, 685 cases were included. Hematoxylin- and eosin-stained slides were digitized and independently reviewed by three teams of head and neck pathologists, in line with the 2023 WHO Classification of Head and Neck Tumours. Discordant and ambiguous cases were resolved in consensus meetings, with access to immunohistochemistry, molecular analysis, and clinical data. Interobserver agreement among the three teams was quantified (Fleiss’ kappa), and agreement between the original and consensus diagnosis was determined (Cohen’s kappa). Of the 685 tumors, 75

INTRODUCTION:The prognostic value of FIGO grading in endometrial cancer (EC) varies across molecular subgroups: it is no longer considered relevant in POLEmut and p53abn tumours, but remains prognostic in NSMP tumours. Its relevance is unclear in mismatch repair deficient (MMRd) EC. Using a large collection of molecularly classified EC, this study evaluated the prognostic role of FIGO grading in MMRd EC. METHODS:Data from three randomised trials (PORTEC-1 (n = 714); PORTEC-2 (n = 427); PORTEC-3 (n = 660)) and five clinical cohorts (n = 1357) were pooled, yielding patients with stage I-III endometrioid and non-endometrioid EC. Tumours were molecularly classified according to the 2020 WHO diagnostic algorithm, with central pathology review by an expert gynaecopathologist. Cause-specific cumulative incidence was estimated using the Aalen-Johansen method and compared using Gray's test. The independent prognostic value of FIGO grading was assessed using multivariable cause-specific Cox regression models. RESULTS:In total, 2621 (83.0%) of the 3158 EC cases were molecularly classified, including 730 (27.9%) MMRd tumours: 405 (55.5%) were low-grade and 325 (44.5%) were high-grade. Median follow-up was 7.2 years. Five-year cumulative incidence of overall recurrence (18.0% vs. 18.8%, p = 0.67) and cancer-specific death (12.3% vs. 14.2%; p = 0.44) did not differ between low- and high-grade MMRd EC. In multivariable analyses, corrected for age, stage, histotype, LVSI and adjuvant treatment, FIGO grading was not independently associated with overall recurrence (HR 0.83 [95%CI 0.56-1.24]; p = 0.37) or cancer-specific death (HR 0.97 [0.62-1.51]; p = 0.89). CONCLUSION:In conclusion, FIGO grading has no prognostic role in MMRd EC and may be omitted in risk stratification and adjuvant treatment decisions.

Computational pathology leverages deep learning to extract clinically relevant information from digitized tumor slides, predicting histopathological subtypes, molecular alterations, and patient outcomes. Recent pipelines increasingly rely on foundation models trained on large pan-cancer datasets to generate generalizable features. In endometrial cancer (EC), their comparative performance for clinical diagnostic tasks remains unexplored. This study evaluates the performance of seven state-of-the-art foundation models across morphological, molecular, and prognostic tasks using a large EC dataset of 3,293 patients from randomized trials and clinical cohorts. In addition, their performance was compared to two versions of an EC-specific feature extractor (EsVIT) exclusively trained on EC. The foundation models H-OPTIMUS-0, CONCH, and VIRCHOW2 achieved the highest mean performance, but the best-performing foundation model varied by task. The top-performing foundation model outperformed EsVIT across all tasks, which highlights the superiority of foundation models over the domain-specific feature extractor EsVIT in EC. Selecting the optimal foundation model for novel tasks remains challenging due to performance plateaus and limited information on the training datasets, requiring rigorous benchmarking and domain insight to reach maximum potential.

OBJECTIVES:Identifying the primary tumor in cancer of unknown primary (CUP) remains a major clinical challenge. While WGS-based tissue-of-origin (TOO) prediction has improved diagnostic precision, its use is constrained by cost and availability. We investigated whether a low-complexity surrogate, combining SMARCA4 mutations and smoking history, can identify a lung cancer origin in CUP (CUP-Lung). METHODS:We retrospectively identified 305 provisional CUP cases from two Dutch CUP referral centers (2022-2025). Integration of whole-genome sequencing (WGS)-based TOO prediction with clinicopathological data identified 58 patients (18.9%) with CUP-Lung. Associations between SMARCA4 mutations, smoking history, and CUP-Lung were assessed using comparative and regression analyses. RESULTS:SMARCA4 mutations were present in 36.2% of CUP-Lung cases, significantly higher than in the overall CUP cohort (10.7%) and TCGA lung cancer datasets (6-8%). CUP-Lung cases had higher tumor mutational burden (median 21.5 vs. 12.6 mut/Mb, p < 0.001) and enriched smoking-related mutational signatures (p < 0.001). Smoking history was reported in 87.9% of cases. Smoking history (OR 4.9, 95% CI 2.2-10.9, p < 0.001) and SMARCA4 mutation (OR 10.3, 95% CI 4.5-23.8, p < 0.001) independently predicted CUP-Lung, together conferring a 76.0% probability of lung origin. No statistical differences were found between cases with and without detectable pulmonary involvement. CONCLUSION:Combined smoking history and SMARCA4 mutations support lung cancer classification even in the absence of detectable pulmonary involvement enabling organ-directed treatment when WGS is unavailable.

Accurate assessment of the tumor immune microenvironment is increasingly important for risk stratification in endometrial carcinoma (EC). Existing immune profiling methods often rely on immunohistochemistry and digital quantification to distinguish inflamed and excluded phenotypes, whereas the Desert phenotype-marked by minimal immune infiltration-may be identifiable on routine hematoxylin and eosin (H&E) slides. Therefore, we developed a qualitative histological scoring system to define the Desert phenotype on H&E slides and used it to examine clinicopathological and molecular associations, evaluate prognostic relevance, and assess interobserver reproducibility. Five expert gynecological pathologists defined consensus criteria for the Desert phenotype in a molecularly classified development cohort (n = 20) and applied them to a testing cohort from the randomized PORTEC-3 trial (n = 380), which included patients with high-risk EC randomized to radiotherapy or chemoradiotherapy. In the PORTEC-3 cohort, 28% of tumors displayed the Desert phenotype. Desert EC were enriched for the no specific molecular profile (NSMP) and p53-abnormal (p53abn) molecular classes and exhibited a distinct mutational profile, including a higher prevalence of CTNNB1 and AKT1 mutations, an effect that was driven by the NSMP subgroup. Desert EC had worse recurrence-free survival compared with Non-Desert EC. Across molecular classes, worse overall survival was only observed in Desert p53abn EC compared to Non-Desert p53abn EC. Interobserver agreement was moderate (κ = 0.57). To conclude, the Desert immune phenotype, as identified on routine H&E slides with moderate agreement, is a biologically distinct and prognostically significant subset of EC. Future work to improve reproducibility is essential to validate its potential for clinical use both for risk stratification and for guiding immunotherapy.

This study represents a omprehensive characterization of high-grade endometrial carcinoma (EC) of no specific molecular profile (NSMP) to improve our understanding of their poor clinical outcome. A previously molecularly classified cohort of 412 high-grade EC from the Danish Cancer Registry was extensively reviewed by 2 expert pathologists blinded for associated clinical and molecular data. Immunohistochemistry (IHC) was performed to determine ER, PR, and L1CAM status and a 10% cut-off value was applied for positivity. Shallow whole-genomic sequencing (sWGS) and next-generation sequencing (NGS) was performed to describe the molecular landscape. Survival analysis was performed using the Kaplan-Meier method, and survival difference was tested using the log-rank test. Of the 57 high-grade NSMP tumors, ER negativity was found in 30 (53%). All clear cell NSMP EC (n=12, 21%) were ER negative. L1CAM overexpression was found in 29 high-grade NSMP EC (53%) and showed overlap (n=20, 69%) with ER negativity. A high frequency of copy number (CN) events and fraction genome altered (FGA) was observed, with the median number of CN events clustering by ER status (28 vs. 43, P <0.05). Overall, the cohort showed a 52% (CI: 31.6%, 72.4%) 5-yr overall survival (OS) and 61% (CI: 42.6%, 79.4%) 5-yr disease-specific survival (DSS). No significant additional prognostic refinement was found when stratifying for ER status (5-yr OS: 46% vs. 65%, P =0.068). High-grade NSMP ECs are a heterogenous group of tumors with high prevalence of loss of ER, L1CAM overexpression, and substantial copy number alterations. Within this group, no prognostic effect of ER was identified, providing support for grouping these tumors into one risk group. This work adds to the growing body of evidence that both high-grade and/or loss of ER expression can be used to identify NSMP EC patients with a poor clinical outcome.

The relationship between bacterial activity and tumorigenesis has gained attention in recent years, complementing the well-established association between viruses and cancer. A recent study employed immunodetection of lipopolysaccharide (LPS) to demonstrate the presence of intracellular bacteria within cancer cells across various cancer types, including breast cancer. The authors proposed that these bacteria might play a role in tumor development. We sought to replicate these findings using the same experimental methods on an independent cohort of breast cancer cases. Our analysis of 129 samples revealed no evidence of LPS expression within cancer cells. Instead, LPS immunoreactivity was observed in ducts or immune cells, specifically macrophages, as expected. These discrepancies in LPS immunodetection warrant caution in interpreting the original findings, and further research is needed to clarify the potential role of intracellular bacteria in cancer development.

OBJECTIVE:Patterns of recurrence may impact the possibilities for salvage treatment and prognosis of patients with endometrial carcinoma (EC). We evaluated the recurrence rate and distribution pattern of the molecular EC subgroups in patients with stage I high-grade disease without adjuvant treatment and those staged by lymphadenectomy. METHOD:412 high-grade EC from the Danish Gynecological Cancer Database were molecularly profiled and classified into POLE mutant (POLEmut), mismatch repair deficient (MMRd), p53-abnormal (p53abn) or no specific molecular profile (NSMP) EC. Patients with stage II-IV (FIGO 2009) or residual disease after surgery were excluded. Crude and actuarial recurrence rates were calculated. RESULTS:Stage I high-grade POLEmut and MMRd EC rarely recurred (5-year overall recurrence rate 7 % (95 % CI 3-16) and 6 % (95 % CI 2-22), respectively), also when not receiving adjuvant treatment. Stage I high-grade NSMP and p53abn EC had high recurrence rates (5-year overall recurrence rate 29 % (95 % CI 16-48) and 35 % (95 % CI 27-45), respectively), mostly presenting with abdominal (NSMP EC n = 1 (3.0 %); p53abn EC n = 28 (22.4 %)) or distant recurrences (NSMP EC n = 8 (24.2 %); p53abn EC n = 21 (16.8 %)). CONCLUSION:Stage I high-grade EC present more frequently with abdominal and distant recurrences rather than isolated loco-regional recurrences, independently of molecular subgroup. Stage I high-grade POLEmut EC and MMRd EC have a favorable prognosis with few recurrences, even with no adjuvant treatment. Stage I high-grade NSMP and p53abn EC have a high recurrence rate, frequently with abdominal or distant recurrences, underscoring the need to investigate more effective adjuvant systemic treatments for these patients.

Background: Five years of adjuvant endocrine therapy (ET) is the standard of care for treating early-stage hormone receptor–positive (HR+) breast cancer (BC) to reduce the risk of disease recurrence. However, many patients discontinue ET early due to poor tolerability and toxicity. Validated biomarkers are critically important to optimize the selection of patients for adjuvant and extended (post–5 year) ET. The Breast Cancer Index (BCI) is a gene expression–based signature recommended in clinical practice guidelines to aid patient selection for ET. The BCI prognostic score reports an individualized risk of overall (0–10 years) and late (post–5 years) distant recurrence (DR). BCI was previously shown to identify patients with favorable BC-specific survival in the Stockholm trial. In the current study, untreated patients from this trial were used to identify an adjusted BCI cut-point for defining a group with minimal risk of DR, which was subsequently validated in a large cohort of patients from the Netherlands Cancer Registry (NCR), who did not receive any adjuvant endocrine therapy. Methods: 283 patients from the untreated arm of the Stockholm trial were used for BCI cut-point selection to define a minimal risk group with a 10-year risk of DR of ≤5%. Performance of the adjusted BCI model was initially assessed in the tamoxifen-treated arm of the Stockholm trial (n=317) and then evaluated in a cohort from NCR (n=1247). Women with HR+ N0 breast cancer who were ≥70 years old and did not receive any adjuvant ET were identified from the NCR database and used for BCI validation. FFPE blocks of primary tumor specimens were collected from participating hospitals across the Netherlands and sent to the Leiden University Medical Center for central processing. Sections were shipped to Biotheranostics CLIA-certified and CAP-accredited laboratory for testing blinded to the clinical data. Kaplan-Meier analysis and Cox proportional hazards regression were used to analyze the Stockholm cohorts with DR as the study endpoint. Cumulative incidence analysis and Fine-Gray model with death as a competing risk event, were used to analyze the NCR patients. Results: The adjusted BCI model stratified patients into four different risk groups: minimal, low, intermediate and high risk. In the Stockholm cohort, the minimal risk group (20%) showed a 10-year risk of DR of 2.3% and 4.3% in the untreated (n=283) and treated arm (n=317), respectively, whereas the 10-year risk of DR was 15.5% and 5% for the low risk (40%), 19.8% and 11.7% for the intermediate risk (23%), and 35.9% and 21.1% for the high risk (17%) groups, respectively. The minimal risk and low risk groups in the tamoxifen-treated arm demonstrated very similar risk profiles. When assessed in the NCR cohort (n = 1247, 55% T1, 42% T2, 22% grade 1, 61% grade2, 94% HER2-, 99.8% no chemotherapy) who did not receive any adjuvant endocrine therapy, the adjusted BCI model was significantly prognostic (p = 0.003) with subdistribution hazard ratios (sHR) for low, intermediate, and high-risk vs. minimal risk of 1.67 (95% CI: 0.81-3.45), 2.39 (95% CI: 1.14-5.01) and 3.23 (95% CI: 1.55-6.74), respectively. The minimal (16%), low (41%), intermediate (24%) and high-risk (19%) groups showed a 10-year risk of DR of 4.5%, 7.5%, 10.3%, 13.5%, respectively, with death as a competing risk event. Conclusions: The current study demonstrates that an adjusted BCI minimal risk cut-point may identify postmenopausal women with HR+ N0 breast cancer, who are at sufficiently low risk of DR and thus unlikely to derive clinically meaningful benefit from adjuvant ET. These results support BCI as a biomarker to help guide the selection of HR+ breast cancer patients who could be spared from or consider shorter duration of adjuvant ET to avoid potentially serious side effects from these therapies. Citation Format: Marie-France Jilderda, Yi Zhang, Valerie Rebattu, Ranelle Salunga, Vincent Smit, Jenna Wong, Linda de Munck, Amanda Anderson, Esther Bastiaannet, Kai Treuner, Gerrit Jan Liefers. An Adjusted Breast Cancer Index Model to Identify Women with Hormone Receptor–Positive (HR+) Breast Cancer at Minimal Risk of 10-year Distant Recurrence (DR) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P1-09-13.

BackgroundNumerous studies have shown that older women with endometrial cancer have a higher risk of recurrence and cancer-related death. However, it remains unclear whether older age is a causal prognostic factor, or whether other risk factors become increasingly common with age. We aimed to address this question with a unique multimethod study design using state-of-the-art statistical and causal inference techniques on datasets of three large, randomised trials.MethodsIn this multimethod analysis, data from 1801 women participating in the randomised PORTEC-1, PORTEC-2, and PORTEC-3 trials were used for statistical analyses and causal inference. The cohort included 714 patients with intermediate-risk endometrial cancer, 427 patients with high-intermediate risk endometrial cancer, and 660 patients with high-risk endometrial cancer. Associations of age with clinicopathological and molecular features were analysed using non-parametric tests. Multivariable competing risk analyses were performed to determine the independent prognostic value of age. To analyse age as a causal prognostic variable, a deep learning causal inference model called AutoCI was used.FindingsMedian follow-up as estimated using the reversed Kaplan-Meier method was 12·3 years (95% CI 11·9–12·6) for PORTEC-1, 10·5 years (10·2–10·7) for PORTEC-2, and 6·1 years (5·9–6·3) for PORTEC-3. Both overall recurrence and endometrial cancer-specific death significantly increased with age. Moreover, older women had a higher frequency of deep myometrial invasion, serous tumour histology, and p53-abnormal tumours. Age was an independent risk factor for both overall recurrence (hazard ratio [HR] 1·02 per year, 95% CI 1·01–1·04; p=0·0012) and endometrial cancer-specific death (HR 1·03 per year, 1·01–1·05; p=0·0012) and was identified as a significant causal variable.InterpretationThis study showed that advanced age was associated with more aggressive tumour features in women with endometrial cancer, and was independently and causally related to worse oncological outcomes. Therefore, our findings suggest that older women with endometrial cancer should not be excluded from diagnostic assessments, molecular testing, and adjuvant therapy based on their age alone.FundingNone.

Universal tumor screening in endometrial carcinoma (EC) is increasingly adopted to identify individuals at risk of Lynch syndrome (LS). These cases involve mismatch repair-deficient (MMRd) EC without MLH1 promoter hypermethylation (PHM). LS is confirmed through the identification of germline MMR pathogenic variants (PV). In cases where these are not detected, emerging evidence highlights the significance of double-somatic MMR gene alterations as a sporadic cause of MMRd, alongside POLE/POLD1 exonuclease domain (EDM) PV leading to secondary MMR PV. Our understanding of the incidence of different MMRd EC origins not related to MLH1-PHM, their associations with clinicopathologic characteristics, and the prognostic implications remains limited. In a combined analysis of the PORTEC-1, -2, and -3 trials (n = 1254), 84 MMRd EC not related to MLH1-PHM were identified that successfully underwent paired tumor-normal tissue next-generation sequencing of the MMR and POLE/POLD1 genes. Among these, 37% were LS associated (LS-MMRd EC), 38% were due to double-somatic hits (DS-MMRd EC), and 25% remained unexplained. LS-MMRd EC exhibited higher rates of MSH6 (52% vs 19%) or PMS2 loss (29% vs 3%) than DS-MMRd EC, and exclusively showed MMR-deficient gland foci. DS-MMRd EC had higher rates of combined MSH2/MSH6 loss (47% vs 16%), loss of >2 MMR proteins (16% vs 3%), and somatic POLE-EDM PV (25% vs 3%) than LS-MMRd EC. Clinicopathologic characteristics, including age at tumor onset and prognosis, did not differ among the various groups. Our study validates the use of paired tumor-normal next-generation sequencing to identify definitive sporadic causes in MMRd EC unrelated to MLH1-PHM. MMR immunohistochemistry and POLE-EDM mutation status can aid in the differentiation between LS-MMRd EC and DS-MMRd EC. These findings emphasize the need for integrating tumor sequencing into LS diagnostics, along with clear interpretation guidelines, to improve clinical management. Although not impacting prognosis, confirmation of DS-MMRd EC may release patients and relatives from burdensome LS surveillance.

Predicting distant recurrence of endometrial cancer (EC) is crucial for personalized adjuvant treatment. The current gold standard of combined pathological and molecular profiling is costly, hampering implementation. Here we developed HECTOR (histopathology-based endometrial cancer tailored outcome risk), a multimodal deep learning prognostic model using hematoxylin and eosin-stained, whole-slide images and tumor stage as input, on 2,072 patients from eight EC cohorts including the PORTEC-1/-2/-3 randomized trials. HECTOR demonstrated C-indices in internal (n = 353) and two external (n = 160 and n = 151) test sets of 0.789, 0.828 and 0.815, respectively, outperforming the current gold standard, and identified patients with markedly different outcomes (10-year distant recurrence-free probabilities of 97.0%, 77.7% and 58.1% for HECTOR low-, intermediate- and high-risk groups, respectively, by Kaplan-Meier analysis). HECTOR also predicted adjuvant chemotherapy benefit better than current methods. Morphological and genomic feature extraction identified correlates of HECTOR risk groups, some with therapeutic potential. HECTOR improves on the current gold standard and may help delivery of personalized treatment in EC.

PURPOSERecent success of human epidermal growth factor receptor 2 (HER2)-targeted antibody-drug-conjugate trastuzumab-deruxtecan in HER2-low and HER2-positive tumors has sparked interest in examining the HER2 status of tumors not traditionally associated with HER2 amplification. Despite the increasing number of systemic treatment options, patients with advanced endometrial cancer (EC) still face a poor prognosis. This study evaluates HER2-low status in over 800 EC, correlating HER2 with both molecular and clinical features.METHODSHER2 status was determined by immunohistochemistry (IHC) and dual in situ hybridization (DISH) on four studies of previously classified high-risk EC (PORTEC-3 and Medical Spectrum Twente cohort), recurrent or metastatic EC (DOMEC), and a primary stage IV cohort. EC was classified as HER2-negative (IHC 0), HER2-low (IHC 1+/2+ without amplification), or HER2-positive (IHC 3+ or DISH-confirmed amplification). Survival analysis was performed using the Kaplan-Meier method. Cox proportional hazards models assessed the independence of any prognostic impact of HER2 status.RESULTSHER2 status was determined in 806 EC: 74.8% were HER2-negative, 17.2% HER2-low, and 7.9% HER2-positive. HER2-low was found across all molecular classes and histotypes. The highest rates of HER2-low and HER2-positive tumors were in recurrent or metastatic EC (35.6% and 15.6%), followed by primary stage IV EC (29.9% and 12.4%) and high-risk EC (14.2% and 6.8%). HER2 status had no independent prognostic value.CONCLUSIONA quarter of high-risk, metastatic, or recurrent EC exhibited HER2 overexpression. The presence of HER2 overexpression in all clinical and molecular categories highlights the need for broad testing and offers treatment options for a wide range of patients.

Supplementary Material and Methods

Figure S3: Concordance p53 IHC and sequencing