Branislav Zagrapan

Background: Immunohistochemical assessment of estrogen receptor (ER), progesterone receptor (PR), proliferation marker Ki67, and human epidermal growth factor receptor 2 (HER2) represents a central component of the diagnostic workflow in invasive breast carcinoma, directly influencing therapeutic decision-making, prognosis, and patient management. In this study, we evaluated the use of Indica Labs’ clinical image analysis platform for AI-assisted breast cancer biomarker scoring in a prospective observer study conducted at the University of Bern. The study compared biomarker assessment by novice observers and trained pathologists with and without AI assistance, with a focus on interobserver agreement, scoring consistency, and the potential role of AI-assisted workflows in diagnostic practice and training environments. Methods: Fifty breast cancer cases comprising 250 immunohistochemically stained slides for ER, PR, HER2, and Ki67 were independently assessed by three medical students and two board-certified pathologists in two sequential evaluation rounds. In the first round, biomarkers were scored using conventional digital pathology assessment without AI support (Visual Dx). Following a four-week washout period, the same cases were re-evaluated with assistance from HALO Breast IHC AI (AI-Assisted Dx). Observer agreement and scoring concordance were analyzed to compare performance between novice observers and experienced pathologists under unassisted and AI-assisted conditions. Results: AI-assisted assessment increased interobserver agreement at clinically relevant cutoffs across all four biomarkers in both novice observers and pathologists. Similarly, Fleiss’ kappa values improved for all biomarkers following AI assistance, with an average increase of 0.42 across observer groups. When all participants were analyzed collectively, agreement at clinical decision thresholds also improved, indicating greater concordance between novice observers and experienced pathologists during AI-assisted evaluation. Conclusions: HALO Breast IHC AI supported both novice observers and experienced pathologists by improving scoring concordance and interobserver agreement across clinically relevant breast cancer biomarkers. These findings suggest that AI-assisted workflows may contribute to greater standardization and reproducibility in biomarker assessment and may also provide supportive value in educational and training settings.

Background: Chronic endometritis (CE) is a subtle, often asymptomatic endometrial inflammation marked by CD138⁺ plasma cell infiltration and linked to recurrent implantation failure (RIF), recurrent pregnancy loss (RPL), and unexplained infertility. Emerging evidence implicates endometrial microbiome dysbiosis in CE. Objective: To systematically review and conduct meta-analysis on the association between CE and endometrial microbiome alterations and their reproductive implications. Methods: We searched MEDLINE, Embase, Web of Science, Scopus, Cochrane CENTRAL, and Google Scholar for studies diagnosing CE via CD138 immunostaining, assessing microbiota with molecular techniques. Data extraction, quality assessment, and meta-analysis were performed. Results: Twenty-two studies including 4022 women were analyzed. CE was associated with reduced prevalence of Lactobacillus-dominated microbiota and increased detection of non-Lactobacillus species, particularly Streptococcus spp., Enterococcus spp., Escherichia coli, Staphylococcus spp., Ureaplasma spp., and Gardnerella vaginalis. In the meta-analysis (2947 women), Enterococcus spp. and Ureaplasma spp. were significantly more prevalent in women with CE, whereas Streptococcus spp., E. coli, Staphylococcus spp. and G. vaginalis showed non-significant trends. Only E. coli and Streptococcus spp. showed significant heterogeneity between-studies. Conclusions: CE is linked to microbial dysbiosis with reduced Lactobacillus dominance and enrichment of potentially pathogenic taxa, notably Enterococcus and Ureaplasma spp. These findings suggest that the endometrial microbiome contributes to chronic inflammation and adverse reproductive outcomes, yet heterogeneity and limited evidence call for standardized diagnostics and robust trials before clinical implementation.

Background: Prognosis and treatment planning for colorectal cancer patients depend on accurate assessment of lymph nodes. However, the histopathological diagnosis of lymph node metastasis under a microscope is time-consuming and prone to error, especially when micro-metastases are involved. We integrated MetAssist 1.0, a deep learning model developed at our institute for detecting lymph node metastases in haematoxylin and eosin-stained whole-slide images, into routine diagnostics as an AI-assisted supporting tool. We evaluated its feasibility and clinical utility prospectively within a routine quality control workflow.Methods: Embedded in routine clinical practice, MetAssist 1.0 was prospectively evaluated as a supporting tool operating alongside standard microscopy. Across 1,783 lymph node slides from 139 colorectal cancer patients (May 2023-September 2025), MetAssist 1.0 analysed digitized slides and flagged discordant classifications for targeted re-review. To assess potential impacts on inter-observer agreement and pathologists’ workload, we conducted a reader study on a randomly selected sub-cohort (256 slides; three readers) with unassisted, assisted, and first-read paradigms.Findings: After pathologists re-examined the discordant slides flagged by the model, sensitivity improved to 99.4% (157/158; 95% CI 96.5-99.9%), while specificity remained at 77.9% (1,266/1,625; 95% CI 75.8–79.9%). Free-text feedback collected during prospective use allowed us to characterize common causes of false-positive results, including tissue folds, non-lymph node tissue, and benign structures. Relative to unassisted review, pairwise agreement increased by 6.2–6.9% with assistance and by up to 13.3% in the first-read approach. Total review time decreased for all readers by approximately 60–95% (e.g., from 61 to 3 minutes).Interpretation: Taken together, these results support the feasibility of prospective clinical integration of MetAssist 1.0 as a supporting tool and highlight practical strengths and limitations for real-world use.

Abdominal aortic aneurysm (AAA) and peripheral artery disease (PAD) share pathophysiological mechanisms including the activation of the fibrinolytic and innate immune system, which explains the analysis of D-dimer and myeloperoxidase (MPO) in both conditions. This study evaluates the diagnostic marker potential of both variables separately and as a combined MPO/D-dimer score for identifying patients with AAA versus healthy individuals or patients with PAD. Plasma levels of MPO and D-dimer were increased in PAD and AAA compared to healthy controls (median for MPO: 13.63 ng/mL [AAA] vs. 11.74 ng/mL [PAD] vs. 9.16 ng/mL [healthy], D-dimer: 1.27 μg/mL [AAA] vs. 0.58 μg/mL [PAD] vs. 0.38 μg/mL [healthy]). The combined MPO/D-dimer score (median 1.26 [AAA] vs. −0.19 [PAD] vs. −0.93 [healthy]) showed an improved performance in distinguishing AAA from PAD when analysed using the receiver operating characteristic curve (area under the curve) for AAA against the pooled data of healthy controls + PAD: 0.728 [MPO], 0.749 [D-dimer], 0.801 [score]. Diagnostic sensitivity and specificity ranged at 82.9% and 70.2% (for score cut-off = 0). These findings were confirmed for a separate collective of AAA patients with 35% simultaneous PAD. Thus, evaluating MPO together with D-dimer in a simple score may be useful for diagnostic detection and the distinction of AAA from athero-occlusive diseases like PAD.

An abdominal aortic aneurysm (AAA) is a multifactorial disease with a variety of genetic and environmental risk factors, but the exact mechanism of AAA formation and progression is still not well understood. The present study investigated the frequency of cytomegalovirus (CMV), Epstein‐Barr virus (EBV), and papillomavirus types 6 and 11 (HPV6 and HPV11), their impact on clinical manifestations of cardiovascular diseases, and their possible association with inflammation in patients with AAA and healthy volunteers. Genotyping of CMV UL75, EBV LMP‐1, and HPV6, and HPV11 E6 was performed by polymerase chain reaction (PCR), while the viral DNA loads were measured by quantitative real‐time PCR. Cytokine levels were determined by enzyme‐linked immunosorbent assays. The CMV UL75 was detected more frequently in the blood of patients with AAA than in the blood of healthy volunteers (32.7% vs. 6.3%, p < .0001). Neither EBV LMP‐1 nor HPV6 E6 was found in blood and aortic wall biopsies, while the HPV11 E6 was detected in 36.4% of AAA walls. The CMV infection in patients with AAA was associated with an increased risk of hypertension and coronary artery disease (OR, 9.057; 95% CI, 1.141–71.862; p = .037; and OR, 2.575; 95% CI, 1.002–6.615; p = .049, respectively). Additionally, CMV‐infected patients with AAA had higher tumor necrosis factor‐α levels compared with noninfected subjects (p = .017). Our findings suggest that CMV infection can stimulate local inflammation in the aorta but is not a direct cause of most abdominal aortic aneurysms.

TLRs are a family of signaling sensors that play a crucial role in the host immune response and are involved in the modulation of inflammatory processes. To study their contribution to abdominal aortic aneurysm (AAA) formation and development, we determined the frequency of TLR2, TLR3, TLR4, and TLR9 single-nucleotide polymorphisms (SNPs) and investigated the association between polymorphisms and the risk of AAA incidence. A total of 104 patients with AAAs and 112 healthy, unrelated volunteers were screened for the presence of TLR2 (2029C/T and 2258G/A), TLR3 (1377C/T, 1234C/T, and −7C/A), TLR4 (896A/G, 1196C/T, and 3266G/A), and TLR9 (−1237T/C, −1486T/C, 1174G/A, and 2848C/T) SNPs by using PCR-RFLP analysis. The heterozygous genotype of the TLR2 2029C/T SNP was more common in patients with AAA than in healthy subjects (p < 0.0001) and was associated with at least an 8-fold increased risk of AAA incidence (p < 0.001). The wild-type genotype of the TLR3 −7C/A SNP was associated with a 3-fold increased risk of hypertension (p = 0.026). The heterozygous TLR3 genotype 1377C/T and −7C/A SNPs were less common in patients with AAA than in healthy subjects (p < 0.0001 and p = 0.0004, respectively) and were associated with a decreased risk of AAA occurrence (p < 0.001 and p = 0.0012, respectively). No relation to AAA risk was found for TLR4 SNPs. Heterozygous genotypes of the TLR2 2029C/T and TLR3 1377C/T and −7C/A SNPs may serve as genetic biomarkers of AAA incidence. Key Points The TLR2 2029C/T SNP was more common in AAA patients than in healthy subjects. The homozygous recessive genotype of the TLR3 −7C/A SNP exhibited low expression. The wild-type genotype of the TLR3 −7C/A SNP was associated with hypertension.

The pathogenesis of abdominal aortic aneurysm (AAA) involves a central component of chronic inflammation which is predominantly mediated by myeloid cells. We hypothesized that the local inflammatory activity may be reflected in systemic alterations of neutrophil and monocyte populations as well as in soluble factors of myeloid cell activation and recruitment. To establish their marker potential, neutrophil and monocyte sub-sets were measured by flow cytometry in peripheral blood samples of 41 AAA patients and 38 healthy controls matched for age, sex, body mass index and smoking habit. Comparably, circulating factors reflecting neutrophil and monocyte activation and recruitment were assayed in plasma. Significantly elevated levels of CD16+ monocytes, activated neutrophils and newly released neutrophils were recorded for AAA patients compared with controls. In line, the monocyte chemoattractant C-C chemokine ligand 2 and myeloperoxidase were significantly increased in patients' plasma. The diagnostic value was highest for myeloperoxidase, a mediator which is released by activated neutrophils as well as CD16+ monocytes. Multivariable regression models using myeloid activation markers and routine laboratory parameters identified myeloperoxidase and D-dimer as strong independent correlates of AAA. These two biomarkers were combined to yield a diagnostic score which was subsequently challenged for confounders and confirmed in a validation cohort matched for cardiovascular disease. Importantly, the score was also found suited to predict rapid disease progression. In conclusion, D-dimer and myeloperoxidase represent two sensitive biomarkers of AAA which reflect distinct hallmarks (thrombus formation and inflammation) of the pathomechanism and, when combined, may serve as diagnostic and prognostic AAA score warranting further evaluation.

Thrombospondin-1 (TSP-1) is primarily expressed by platelets and endothelial cells (ECs) and rapidly released upon their activation. It functions in haemostasis as a bridging molecule in platelet aggregation, by promoting platelet adhesion to collagen and by protecting von Willebrand factor strings from degradation. In blood of patients undergoing surgery and in co-cultures of neutrophils with platelets or ECs, we observed proteolysis of the 185 kDa full-length TSP-1 to a 160-kDa isoform. We hypothesized that TSP-1 processing may alter its haemostatic properties. Selective enzyme inhibitors in co-cultures revealed that neutrophil proteases elastase and cathepsin G mediate TSP-1 processing. The cut site of cathepsin G was mapped to TSP-1 amino acids R237/T238 by Edman sequencing. Formation of neutrophil extracellular traps protected TSP-1 from complete degradation and promoted controlled processing to the 160-kDa isoform. Haemostatic properties were tested by platelet aggregation, adhesion, coagulation and string formation under flow. Platelets from TSP-1 deficient mice did not differ from wild-type in platelet aggregation but showed severe impairment of platelet adhesion to collagen and string formation under flow. Reconstitution experiments revealed that the 160-kDa TSP-1 isoform was markedly more potent than the 185-kDa full-length molecule in restoring function. Thus, TSP-1 processing by neutrophil proteases yields a 160-kDa isoform which shows enhanced potency to promote platelet adhesion and string formation. This finding reveals a novel mechanism of neutrophil-mediated thrombus formation and provides first evidence for the impact of TSP-1 proteolysis on its haemostatic properties.

Heme oxygenase‐1 (HO‐1), encoded by HMOX1 gene and regulated by Nrf2 transcription factor, is a cytoprotective enzyme. Its deficiency may exacerbate abdominal aortic aneurysm (AAA) development, which is also often associated with hyperlipidemia. Beneficial effects of statins, the broadly used antilipidemic drugs, were attributed to modulation of Nrf2/HO‐1 axis. However, the effect of statins on Nrf2/HO‐1 pathway in patients with AAA has not been studied yet. We analyzed AAA tissue from patients treated with simvastatin (N = 28) or without statins (N = 14). Simvastatin treatment increased HO‐1 protein level in AAA, both in endothelial cells (ECs) and in smooth muscle cells (SMCs), but increased Nrf2 localization was restricted only to vasa vasorum. Nrf2 target genes HMOX1, NQO1, and GCLM expression remained unchanged in AAA. In vitro studies showed that simvastatin raises HO‐1 protein level slightly in ECs and to much higher extent in SMCs, which is not related to Nrf2/ARE activation, although HMOX1 expression is upregulated by simvastatin in both cell types. In conclusion, simvastatin‐induced modulation of HO‐1 level in ECs and SMCs in vitro is not related to Nrf2/ARE activity. Likewise, divergent HO‐1 and Nrf2 localization together with stable expression of Nrf2 target genes, including HMOX1, in AAA tissue denotes Nrf2 independency.