Explore the vast range of titles available.

Deep learning algorithms have been successfully used in medical image classification. In the next stage, the technology of acquiring explainable knowledge from medical images is highly desired. Here we show that deep learning algorithm enables automated acquisition of explainable features from diagnostic annotation-free histopathology images. We compare the prediction accuracy of prostate cancer recurrence using our algorithm-generated features with that of diagnosis by expert pathologists using established criteria on 13,188 whole-mount pathology images consisting of over 86 billion image patches. Our method not only reveals findings established by humans but also features that have not been recognized, showing higher accuracy than human in prognostic prediction. Combining both our algorithm-generated features and human-established criteria predicts the recurrence more accurately than using either method alone. We confirm robustness of our method using external validation datasets including 2276 pathology images. This study opens up fields of machine learning analysis for discovering uncharted knowledge. Technologies for acquiring explainable features from medical images need further development. Here, the authors report a deep learning based automated acquisition of explainable features from pathology images, and show a higher accuracy of their method as compared to pathologist based diagnosis of prostate cancer recurrence.

This study aims to determine the epigenetic mechanism regulating Kiss1 gene expression in the anteroventral periventricular nucleus (AVPV) to understand the mechanism underlying estrogen-positive feedback action on gonadotropin-releasing hormone/gonadotropin surge. We investigated estrogen regulation of the epigenetic status of the mouse AVPV Kiss1 gene locus in comparison with the arcuate nucleus (ARC), in which Kiss1 expression is down-regulated by estrogen. Histone of AVPV Kiss1 promoter region was highly acetylated, and estrogen receptor α was highly recruited at the region by estrogen. In contrast, the histone of ARC Kiss1 promoter region was deacetylated by estrogen. Inhibition of histone deacetylation up-regulated in vitro Kiss1 expression in a hypothalamic non–Kiss1-expressing cell line. Gene conformation analysis indicated that estrogen induced formation of a chromatin loop between Kiss1 promoter and the 3' intergenic region, suggesting that the intergenic region serves to enhance estrogen-dependent Kiss1 expression in the AVPV. This notion was proved, because transgenic reporter mice with a complete Kiss1 locus sequence showed kisspeptin neuron-specific GFP expression in both the AVPV and ARC, but the deletion of the 3' region resulted in greatly reduced GFP expression only in the AVPV. Taken together, these results demonstrate that estrogen induces recruitment of estrogen receptor α and histone acetylation in the Kiss1 promoter region of the AVPV and consequently enhances chromatin loop formation of Kiss1 promoter and Kiss1 gene enhancer, resulting in an increase in AVPV-specific Kiss1 gene expression. These results indicate that epigenetic regulation of the Kiss1 gene is involved in estrogen-positive feedback to generate the gonadotropin-releasing hormone/gonadotropin surge.

The agonistic/antagonistic biocharacter of selective estrogen receptor modulators (SERMs) can have therapeutic advantages, particularly in the case of premenopausal breast cancers. Although the contradictory effects of these modulators have been studied in terms of crosstalk between the estrogen receptor α (ER) and coactivator dynamics and growth factor signaling, the molecular basis of these mechanisms is still obscure. We identify a series of regulatory mechanisms controlling cofactor dynamics on ER and SERM function, whose activities require F-box protein 22 (Fbxo22). Skp1, Cullin1, F-box-containing complex (SCFFbxo22) ubiquitylated lysine demethylase 4B (KDM4B) complexed with tamoxifen-bound (TAM-bound) ER, whose degradation released steroid receptor coactivator (SRC) from ER. Depletion of Fbxo22 resulted in ER-dependent transcriptional activation via transactivation function 1 (AF1) function, even in the presence of SERMs. In living cells, TAM released SRC and KDM4B from ER in a Fbxo22-dependent manner. SRC release by TAM required Fbxo22 on almost all ER-SRC-bound enhancers and promoters. TAM failed to prevent the growth of Fbxo22-depleted, ER-positive breast cancers both in vitro and in vivo. Clinically, a low level of Fbxo22 in tumor tissues predicted a poorer outcome in ER-positive/human epidermal growth factor receptor type 2-negative (HER2-negative) breast cancers with high hazard ratios, independently of other markers such as Ki-67 and node status. We propose that the level of Fbxo22 in tumor tissues defines a new subclass of ER-positive breast cancers for which SCFFbxo22-mediated KDM4B degradation in patients can be a therapeutic target for the next generation of SERMs.

BRCA1 and 53BP1 antagonistically regulate homology‐directed repair (HDR) and non‐homologous end‐joining (NHEJ) of DNA double‐strand breaks (DSB). The histone deacetylase (HDAC) inhibitor trichostatin A directly inhibits the retention of 53BP1 at DSB sites by acetylating histone H4 (H4ac), which interferes with 53BP1 binding to dimethylated histone H4 Lys20 (H4K20me2). Conversely, we recently found that the retention of the BRCA1/BARD1 complex is also affected by another methylated histone residue, H3K9me2, which can be suppressed by the histone lysine methyltransferase (HKMT) inhibitor UNC0638. Here, we investigate the effects of the class I HDAC inhibitors MS‐275 and FK228 compared to UNC0638 on histone modifications and the DNA damage response. In addition to H4ac, the HDAC inhibitors induce H3K9ac and inhibit H3K9me2 at doses that do not affect the expression levels of DNA repair genes. By contrast, UNC0638 selectively inhibits H3K9me2 without affecting the levels of H3K9ac, H3K56ac or H4ac. Reflecting their effects on histone modifications, the HDAC inhibitors inhibit ionizing radiation‐induced foci (IRIF) formation of BRCA1 and BARD1 as well as 53BP1 and RIF1, whereas UNC0638 suppresses IRIF formation of BRCA1 and BARD1 but not 53BP1 and RIF1. Although HDAC inhibitors suppressed HDR, they did not cooperate with the poly(ADP‐ribose) polymerase inhibitor olaparib to block cancer cell growth, possibly due to simultaneous suppression of NHEJ pathway components. Collectively, these results suggest the mechanism by that HDAC inhibitors inhibit both the HDR and NHEJ pathways, whereas HKMT inhibitor inhibits only the HDR pathway; this finding may affect the chemosensitizing effects of the inhibitors. Findings show how HDAC inhibitors inhibit both homologous recombination and non‐homologous end‐joining pathways through conversion of histone modifications.

We previously reported on two women with breast lesions in whom radiological examination could not exclude malignancy. In both cases, mastectomy was performed, and histological analyses revealed papillary lesions lined by fibrovascular stroma and nuclear inverse polarity. Hematoxylin–eosin, p63, and calponin staining indicated an absence of myoepithelial cells. However, it was concluded that the lesions had been non-malignant. These women have now been under long-term surveillance (74 months for one case and 62 months for the other) and have had no disease recurrence. Mucin (MUC)1, MUC2, MUC4, MUC5AC, MUC5B, and MUC6 immunostaining has also been performed in these women to investigate further whether their tumors were malignant or benign. In both cases, the tumors were only positive for MUC1 in apical luminal apical cells, as in normal breast tissue. MUC5B immunostaining, even when weak, can detect early breast cancer but was completely negative in our two cases. Therefore, both tumors were considered benign. Our findings in these cases suggest that nuclear inverse polarity papillary lesions lacking myoepithelial cells are benign. This knowledge should decrease the number of unnecessary operations performed for this tumor and their negative impact on patients’ quality of life.

In a substantial number of patients, ductal carcinoma in situ (DCIS) of the breast will never progress to invasive ductal carcinoma, and these patients are often overtreated under the current clinical criteria. Although various candidate markers are available, relevant markers for delineating risk categories have not yet been established. In this study, we analyzed the clinical characteristics of 431 patients with DCIS and performed whole-exome sequencing analysis in a 21-patient discovery cohort and targeted deep sequencing analysis in a 72-patient validation cohort. We determined that age <45 years, HER2 amplification, and GATA3 mutation are possible indicators of relapse. PIK3CA mutation negativity and PgR negativity were also suggested to be risk factors. Spatial transcriptome analysis further revealed that GATA3 dysfunction upregulates epithelial-to-mesenchymal transition and angiogenesis, followed by PgR downregulation. These results reveal the existence of heterogeneous cell populations in DCIS and provide predictive markers for classifying DCIS and optimizing treatment.Satoi Nagasawa and Yuta Kuze et al. report a multi-omic analysis of ductal carcinoma in situ (DCIS) of the breast, including whole-exome, single-cell, and spatial transcriptome sequencing. They find that for patients under 45 years of age, HER2 amplification and GATA3 mutation are associated with higher risk of relapse, suggesting they could be used as predictive markers when deciding on a treatment course.

Peritoneal washing cytology (CY) in patients with pancreatic cancer is mainly used for staging; however, it may also be used to evaluate the intraperitoneal status to predict a more accurate prognosis. Here, we investigated the potential of deep learning of CY specimen images for predicting the 1-year prognosis of pancreatic cancer in CY-positive patients. CY specimens from 88 patients with prognostic information were retrospectively analyzed. CY specimens scanned by the whole slide imaging device were segmented and subjected to deep learning with a Vision Transformer (ViT) and a Convolutional Neural Network (CNN). The results indicated that ViT and CNN predicted the 1-year prognosis from scanned images with accuracies of 0.8056 and 0.8009 in the area under the curve of the receiver operating characteristic curves, respectively. Patients predicted to survive 1 year or more by ViT showed significantly longer survivals by Kaplan–Meier analyses. The cell nuclei found to have a negative prognostic impact by ViT appeared to be neutrophils. Our results indicate that AI-mediated analysis of CY specimens can successfully predict the 1-year prognosis of patients with pancreatic cancer positive for CY. Intraperitoneal neutrophils may be a novel prognostic marker and therapeutic target for CY-positive patients with pancreatic cancer.

The overexpression of Polo-like kinase 1 (PLK1) is associated with poor clinical outcomes in various malignancies, making it an attractive target for anticancer therapies. Although recent studies suggest PLK1’s involvement in homologous recombination (HR), the impact of its overexpression on HR remains unclear. In this study, we investigated the effect of PLK1 overexpression on HR using bioinformatics and experimental approaches. Analyzing The Cancer Genome Atlas (TCGA) and Cancer Cell Line Encyclopedia (CCLE) datasets with the Homologous Recombination Deficiency (HRD) score, we found a positive correlation between PLK1 expression and HRD score, indicating that increased PLK1 expression suppresses HR. To validate these findings, we performed cell line-based experiments, demonstrating that PLK1 overexpression attenuates RAD51 focus formation and HR, as measured by ASHRA in T47D cells. Since HR-deficient cells are hypersensitive to PARP inhibitors, we further confirmed that PLK1 overexpression increases sensitivity to PARP inhibitors, both in CCLE dataset analysis and experiments using T47D cells. Additionally, we found that the effects of PLK1 overexpression on HR suppression and increased PARP inhibitor sensitivity were mitigated by either a PLK1 kinase inhibitor or the kinase-dead mutant [T210A]. This suggests that PLK1’s impact on HR and PARP inhibitor sensitivity is mediated through its kinase activity. Moreover, analysis of clinical ovarian cancer samples revealed that higher PLK1 expression correlates with increased sensitivity to PARP inhibitors. Our results suggest that PLK1 overexpression suppresses homologous recombination, leading to enhanced sensitivity to PARP inhibition, presenting a potential therapeutic strategy for targeting cancers with overexpression of PLK1.

A subset of prostate cancer displays a poor clinical outcome. Therefore, identifying this poor prognostic subset within clinically aggressive groups (defined as a Gleason score (GS) ≧8) and developing effective treatments are essential if we are to improve prostate cancer survival. Here, we performed a bioinformatics analysis of a TCGA dataset (GS ≧8) to identify pathways upregulated in a prostate cancer cohort with short survival. When conducting bioinformatics analyses, the definition of factors such as “overexpression” and “shorter survival” is vital, as poor definition may lead to mis-estimations. To eliminate this possibility, we defined an expression cutoff value using an algorithm calculated by a Cox regression model, and the hazard ratio for each gene was set so as to identify genes whose expression levels were associated with shorter survival. Next, genes associated with shorter survival were entered into pathway analysis to identify pathways that were altered in a shorter survival cohort. We identified pathways involving upregulation of GRB2. Overexpression of GRB2 was linked to shorter survival in the TCGA dataset, a finding validated by histological examination of biopsy samples taken from the patients for diagnostic purposes. Thus, GRB2 is a novel biomarker that predicts shorter survival of patients with aggressive prostate cancer (GS ≧8).

The present study investigated whether kisspeptin–G protein-coupled receptor 54 (GPR54) signaling plays a role in mediating mating-induced ovulation in the musk shrew (Suncus murinus), a reflex ovulator. For this purpose, we cloned suncus Kiss1 and Gpr54 cDNA from the hypothalamus and found that suncus kisspeptin (sKp) consists of 29 amino acid residues (sKp-29). Injection of exogenous sKp-29 mimicked the mating stimulus to induce follicular maturation and ovulation. Administration of several kisspeptins and GPR54 agonists also induced presumed ovulation in a dose-dependent manner, and Gpr54 mRNA was distributed in the hypothalamus, showing that kisspeptins induce ovulation through binding to GPR54. The sKp-29–induced ovulation was blocked completely by pretreatment with a gonadotropin-releasing hormone (GnRH) antagonist, suggesting that kisspeptin activates GnRH neurons to induce ovulation in the musk shrew. In addition, in situ hybridization revealed that Kiss1-expressing cells are located in the medial preoptic area (POA) and arcuate nucleus in the musk shrew hypothalamus. The number of Kiss1-expressing cells in the POA or arcuate nucleus was up-regulated or down-regulated by estradiol, suggesting that kisspeptin neurons in these regions were the targets of the estrogen feedback action. Finally, mating stimulus largely induced c-Fos expression in Kiss1-positive cells in the POA, indicating that the mating stimulus activates POA kisspeptin neurons to induce ovulation. Taken together, these results indicate that kisspeptin–GPR54 signaling plays a role in the induction of ovulation in the musk shrew, a reflex ovulator, as it does in spontaneous ovulators.