Bojing Liu

Background: Patients with ER-positive/HER2-negative early breast cancer of low or intermediate grade are generally considered to have a favourable prognosis, yet a number will recur. This risk is poorly captured by standard clinicopathological factors. AI-based medical devices for whole slide image (WSI) analyses used for risk-stratification often predict on slide-level thereby overlooking spatial localisation of the biomarkers. We hypothesise that the spatial distribution of a previously reported AI-based marker provides independent prognostic value beyond its global score. Patients and Methods: This Swedish multicentre cohort study includes 6106 patients with primary invasive breast cancer: 3078 from the Stockholm region (training) and 3028 from Skåne (independent test). Tile-level high-grade morphology scores were generated with ViT_DG, a vision-transformer adaptation of DeepGrade, and the presence of clusters of high-risk tiles localised at the invasive tumour front was defined as a binary biomarker. Primary analyses were pre-specified in the ER-positive/HER2-negative/lymph-node-negative and Grade 1-2 subgroup. Association with breast-cancer recurrence-free interval (BCRFI) and distant metastasis were assessed using multivariable Cox proportional hazards models (HR). Results: In the primary subgroup (n=1536 training; n=1403 test), tumour front clusters were present in 37% and 39% of patients. Cluster-positive status was associated with a 2.30-fold increased risk in the external test. Cluster-negative patients achieved a 10-year BCRFI of 95.3%, equivalent to Grade 1 patients and better than Grade 2 (92.4%), whereas cluster-positive patients (90.2%) did worse than the Grade 3 reference. The association was retained in patients classified as low-risk by ViT_DG (test set multivariable HR 2.86), identifying high-risk patients not captured by the global score. Conclusions: The absence of high-risk morphology clusters at the tumour front is an independent prognostic indicator, identifying a low-risk group independently of the global AI biomarker classification. More broadly, preserving the spatial distribution of AI-derived predictions enables prognostic stratification beyond slide-level aggregation. ClinicalTrials.gov identifier: NCT02306096

Self-supervised learning (SSL) automates the extraction and interpretation of histopathology features on unannotated hematoxylin-and-eosin-stained whole-slide images (WSIs). We trained an SSL Barlow Twins-encoder on 435 TCGA colon adenocarcinoma WSIs to extract features from small image patches. Leiden community detection then grouped tiles into histomorphological phenotype clusters (HPCs). HPC reproducibility and predictive ability for overall survival was confirmed in an independent clinical trial cohort (N=1213 WSIs). This unbiased atlas resulted in 47 HPCs displaying unique and sharing clinically significant histomorphological traits, highlighting tissue type, quantity, and architecture, especially in the context of tumor stroma. Through in-depth analysis of these HPCs, including immune landscape and gene set enrichment analysis, and association to clinical outcomes, we shed light on the factors influencing survival and responses to treatments like standard adjuvant chemotherapy and experimental therapies. Further exploration of HPCs may unveil new insights and aid decision-making and personalized treatments for colon cancer patients.

Objective: This study was undertaken to examine the association between montelukast use, β2‐adrenoreceptor (β2AR) agonist use, and later Parkinson disease (PD).Methods: We ascertained use of β2AR agonists (430,885 individuals) and montelukast (23,315 individuals) from July 1, 2005 to June 30, 2007, and followed 5,186,886 PD‐free individuals from July 1, 2007 to December 31, 2013 for incident PD diagnosis. We estimated hazard ratios and 95% confidence intervals using Cox regressions.Results: We observed 16,383 PD cases during on average 6.1 years of follow‐up. Overall, use of β2AR agonists and montelukast were not related to PD incidence. A 38% lower PD incidence was noted among high‐dose montelukast users when restricted to PD registered as the primary diagnosis.Interpretation: Overall, our data do not support inverse associations between β2AR agonists, montelukast, and PD. The prospect of lower PD incidence with high‐dose montelukast exposure warrants further investigation, especially with adjustment for high‐quality data on smoking. ANN NEUROL 2023;93:1023–1028

Histological Grade is a well-known prognostic factor that is routinely assessed in breast tumours. However, manual assessment of Nottingham Histological Grade (NHG) has high inter-assessor and inter-lab variability, causing uncertainty in grade assignments. To address this challenge, we developed and validated a three-level NHG-like deep learning-based histological grade model. The primary performance evaluation focuses on prognostic performance. This observational study is based on two patient cohorts (SöS-BC-4, N=2421 (training and internal test); SCAN-B-Lund, N=1262 (test)) that include routine histological whole slide images together with patient outcomes. A Deep Convolutional Neural Network (CNN) model with an attention mechanism was optimised for the classification of the three-level histological grading (NHG) from hematoxylin and eosin-stained WSIs. The prognostic performance was evaluated by time-to-event analysis of Recurrence-free survival (RFS) and compared to clinical NHG grade assignments in the internal test set as well as in the fully independent external test cohort. We observed effect sizes (Hazard Ratio) for grade 3 vs 1, for the conventional NHG method (HR=2.60 (1.18-5.70 95%CI, p-value = 0.017)) and the deep learning model (HR = 2.27, 95%CI: 1.07-4.82, p-value = 0.033) on the internal test set after adjusting for established clinicopathological risk factors. In the external test set, the unadjusted HR for NHG 1 vs 2 was estimated to be 2.59 (p-value = 0.004) and NHG 1 vs 3 was estimated to be 3.58 (p-value < 0.001). For predGrade, the unadjusted HR for grade 1 vs 2 HR=2.52 (p-value = 0.030), and 4.07 (p-value = 0.001) for grade 1 vs 3. In multivariable analysis, HR estimates for neither NHG nor predGrade were found to be significant (p-value >0.05). We tested for differences in HR estimates between NHG and predGrade in the independent test set, and found no significant difference between the two classification models (p-value > 0.05), confirming similar prognostic performance between conventional NHG and predGrade. Routine histopathology assessment of NHG has a high degree of inter-assessor variability, motivating the development of model-based decision support to improve reproducibility in histological grading. We found that the proposed model provides similar prognostic performance as NHG. The results indicate that deep CNN-based models can be applied for breast cancer histological grading.

To examine whether irritable bowel syndrome (IBS) was related to the future risk of Parkinson’s disease (PD), we conducted a nested case-control study in the Swedish total population including 56,564 PD cases identified from the Swedish Patient Register and 30 controls per case individually matched by sex and year of birth. Odds ratios (ORs) with 95% confidence intervals (CIs) for having a prior diagnosis of IBS were estimated using conditional logistic regression. We furthermore conducted a cohort study using the Swedish Twin Registry following 3046 IBS patients identified by self-reported abdominal symptoms and 41,179 non-IBS individuals. Through Cox proportional hazard models, we estimated hazard ratios (HRs) and 95% CIs for PD risk. In the nested case-control study, 253 (0.4%) PD cases and 5204 (0.3%) controls had a previous IBS diagnosis. IBS diagnosis was associated with a 44% higher risk of PD (OR = 1.44, 95% CI 1.27–1.63). Temporal relationship analyses showed 53% and 38% increased risk of PD more than 5 and 10 years after IBS diagnosis, respectively. In the cohort analysis based on the Swedish Twin Registry, there was no statistically significantly increased risk of PD related to IBS (HR = 1.25, 95% CI = 0.87–1.81). Our results suggest a higher risk of PD diagnosis after IBS. These results provide additional evidence supporting the importance of the gut–brain axis in PD.

Summary: Background: The vagus nerve provides essential parasympathetic innervation to the gastrointestinal system and is known to have anti‐inflammatory properties.Aims: To explore the relationship between vagotomy and the risk of inflammatory bowel disease (IBD) and its major categories: Crohn's disease (CD) and ulcerative colitis (UC).Methods: A matched cohort comprising 15 637 patients undergoing vagotomy was identified through the Swedish Patient Register from 1964 to 2010. Each vagotomised patient was matched for birth year and gender with 40 nonvagotomised individuals on the date of vagotomy. We estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for IBD using flexible parametric models adjusted for matching variables, year of vagotomy, birth country, chronic obstructive pulmonary disease and comorbidity index.Results: We observed 119 (0.8%) patients with vagotomy developed IBD compared to 3377 (0.5%) IBD cases in nonvagotomised individuals. The crude incidence of IBD (per 1000 person‐years) was 0.38 for vagotomised patients and 0.25 for nonvagotomised individuals. We observed a time‐dependent elevated risk of IBD associated with vagotomy, for instance, the HR (95% CI) was 1.80 (1.40‐2.31) at year 5 and 1.49 (1.14‐1.96) at year 10 post‐vagotomy. The association appeared to be stronger for truncal than selective vagotomy and limited to CD (HR was 3.63 [1.94‐6.80] for truncal and 2.06 [1.49‐2.84] for selective vagotomy) but not UC (1.36 [0.71‐2.62] for truncal and 1.25 [0.95‐1.63] for selective vagotomy).Conclusions: We found a positive association between vagotomy and later IBD, particularly for CD. The finding indirectly underlines the beneficial role of the vagal tone in IBD.

Objective: To examine whether vagotomy decreases the risk of Parkinson disease (PD). Methods: Using data from nationwide Swedish registers, we conducted a matched-cohort study of 9,430 vagotomized patients (3,445 truncal and 5,978 selective) identified between 1970 and 2010 and 377,200 reference individuals from the general population individually matched to vagotomized patients by sex and year of birth with a 40:1 ratio. Participants were followed up from the date of vagotomy until PD diagnosis, death, emigration out of Sweden, or December 31, 2010, whichever occurred first. Vagotomy and PD were identified from the Swedish Patient Register. We estimated hazard ratios (HRs) with 95% confidence intervals (CIs) using Cox models stratified by matching variables, adjusting for country of birth, chronic obstructive pulmonary disease, diabetes mellitus, vascular diseases, rheumatologic disease, osteoarthritis, and comorbidity index. Results: A total of 4,930 cases of incident PD were identified during 7.3 million person-years of follow-up. PD incidence (per 100,000 person-years) was 61.8 among vagotomized patients (80.4 for truncal and 55.1 for selective) and 67.5 among reference individuals. Overall, vagotomy was not associated with PD risk (HR 0.96, 95% CI 0.78–1.17). However, there was a suggestion of lower risk among patients with truncal vagotomy (HR 0.78, 95% CI 0.55–1.09), which may be driven by truncal vagotomy at least 5 years before PD diagnosis (HR 0.59, 95% CI 0.37–0.93). Selective vagotomy was not related to PD risk in any analyses. Conclusions: Although overall vagotomy was not associated the risk of PD, we found suggestive evidence for a potential protective effect of truncal, but not selective, vagotomy against PD development.

Parkinson’s disease (PD) may take decades to develop and early life exposures such as infection may be important. However, few epidemiological studies have evaluated early life risk factors in relation to PD risk. We therefore examined such associations in a prospective analysis of 3 545 612 individuals born in Sweden between 1932 and 1970 without PD on January 1, 2002. Incident PD cases were identified using the Swedish Patient Register during 2002–2010. Information on sibship size, number of older and younger siblings, multiple births, parental age, birth month and season was obtained from the Swedish Multi-Generation Register. Monthly data on national burden of influenza-like illness during 1932–1970 were obtained from the Swedish Public Health Agency. Hazard ratios with 95% confidence intervals (CI) were estimated using Cox proportional hazards regression. During the follow-up, 8779 incident PD cases were identified. As expected, older age, male sex, parental occupation as farmers, and family history of PD were associated with higher PD risk. Overall, early life factors, including flu burden in the year of birth, were not associated with PD risk, although we did find a lower PD risk among participants with older siblings than those without (HR = 0.93, 95%CI: 0.89, 0.98). Our study therefore provided little support for important etiological contributions of early life factors to the PD risk late in life. The finding of a lower PD risk among individuals with older siblings will need confirmation and further investigation.