Enhancing mass spectrometry imaging accessibility using convolutional autoencoders for deriving hypoxia-associated peptides from tumors.

Mass spectrometry imaging (MSI) allows to study cancer's intratumoral heterogeneity through spatially-resolved peptides, metabolites and lipids. Yet, in biomedical research MSI is rarely used for biomarker discovery. Besides its high dimensionality and multicollinearity, mass spectrometry (MS) technologies typically output mass-to-charge ratio values but not the biochemical compounds of interest. Our framework makes particularly low-abundant signals in MSI more accessible. We utilized convolutional autoencoders to aggregate features associated with tumor hypoxia, a parameter with significant spatial heterogeneity, in cancer xenograft models. We highlight that MSI captures these low-abundant signals and that autoencoders can preserve them in their latent space. The relevance of individual hyperparameters is demonstrated through ablation experiments, and the contribution from original features to latent features is unraveled. Complementing MSI with tandem MS from the same tumor model, multiple hypoxia-associated peptide candidates were derived. Compared to random forests alone, our autoencoder approach yielded more biologically relevant insights for biomarker discovery.

Clinical characterisation of women with persistent genital arousal disorder: the iPGAD-study.

Persistent Genital Arousal Disorder (PGAD) is a rare condition-mostly in women-where patients perceive prolonged genital arousal without any sexual desire or stimulation. Etiopathological considerations reach from peripheral to central issues over local disturbance of the pudendal nerve to neuropathy, psychosocial, and pharmacological theories. Since well controlled clinical studies about PGAD in conjunction with a mental and somatic health status are missing, this study is a detailed clinical investigation of PGAD patients compared to healthy controls. 26 women who fulfilled diagnostic criteria for PGAD were compared to 26 age matched healthy controls. Investigations included comparison of vegetative, gynaecological and sexual history, psychiatric features as well as a (neuro-)radiological, neurophysiological and gynaecological examination. Moreover, a detailed clinical characterisation of PGAD symptoms was performed. PGAD symptoms were mostly characterised as tingling or prickling and were permanently present. In over 80%, PGAD symptoms were located in the clitoris. Almost 70% reported radiations to other regions of the body. Most frequent trigger factors were tight clothes, mental stress, driving a car/bus/bicycle and sexual intercourse. Relieving factors were mainly distraction, relaxation, physical exercise, masturbation and swimming. In group comparisons, PGAD presented with significant higher rates of sexual dysfunctions, spontaneous orgasms, swelling of the genitals, extraordinary lubrication as well as higher rates in depression, agoraphobia, generalized anxiety disorder and lifetime panic disorder. Significantly more PGAD patients were diagnosed with restless legs symptoms. In contrast childhood traumatization, somatization disorder, suicidality, gynaecological as well as neurophysiological examination of the pudendal nerve were not different between the groups. MRI of the brain, pelvis and spinal cord was unsuspicious and incidental findings - including Tarlov cysts or pelvic venous congestion - were equally distributed among the groups. In summary, our study provides a careful characterization of women with PGAD highlighting a serious mental burden, most probably as a consequence of PGAD. With the current set of clinical investigations there was no evidence of a clear causal relationship to a specific clinical finding as it has been previously discussed. Future studies and additional techniques will have to further explore where and how in the peripheral or central nervous systems PGAD develops.

Orthotopic Glioblastoma Models for Evaluation of the Clinical Target Volume Concept.

In times of high-precision radiotherapy, the accurate and precise definition of the primary tumor localization and its microscopic spread is of enormous importance. In glioblastoma, the microscopic tumor extension is uncertain and, therefore, population-based margins for Clinical Target Volume (CTV) definition are clinically used, which could either be too small-leading to increased risk of loco-regional recurrences-or too large, thus, enhancing the probability of normal tissue toxicity. Therefore, the aim of this project is to investigate an individualized definition of the CTV in preclinical glioblastoma models based on specific biological tumor characteristics. The microscopic tumor extensions of two different orthotopic brain tumor models (U87MG_mCherry; G7_mCherry) were evaluated before and during fractionated radiotherapy and correlated with corresponding histological data. Representative tumor slices were analyzed using Matrix-Assisted Laser Desorption/Ionization (MALDI) and stained for putative stem-like cell markers as well as invasion markers. The edges of the tumor are clearly shown by the MALDI segmentation via unsupervised clustering of mass spectra and are consistent with the histologically defined border in H&E staining in both models. MALDI component analysis identified specific peaks as potential markers for normal brain tissue (e.g., 1339 m/z), whereas other peaks demarcated the tumors very well (e.g., 1562 m/z for U87MG_mCherry) irrespective of treatment. MMP14 staining revealed only a few positive cells, mainly in the tumor border, which could reflect the invasive front in both models. The results of this study indicate that MALDI information correlates with microscopic tumor spread in glioblastoma models. Therefore, an individualized CTV definition based on biological tumor characteristics seems possible, whereby the visualization of tumor volume and protein heterogeneity can be potentially used to define radiotherapy-sensitive and resistant areas.

Influence of Systemic Therapy on the Expression and Activity of Selected STAT Proteins in Prostate Cancer Tissue.

Signal Transducer and Activator of Transcription (STAT) proteins have been identified as drivers of prostate cancer (PCa) progression and development of aggressive castration-resistant phenotypes. In particular, STAT3, 5, and 6 have been linked to resistance to androgen receptor inhibition and metastasis in in vitro and in vivo models. This descriptive study aimed to validate these preclinical data in tissue obtained from patients with PCa before and while under androgen-deprivation therapy. Therefore, STAT3, 5, and 6 expressions and activity were assessed by immunohistochemistry. The data revealed that STAT3 and 5 changed in PCa. However, there was no relationship between expression and survival. Moreover, due to the heterogeneous nature of PCa, the preclinical results could not be transferred congruently to the patient's material. A pilot study with a longitudinal patient cohort could also show this heterogeneous influence of systemic therapy on STAT3, 5, and 6 expressions and activity. Even if the main mechanisms were validated, these data demonstrate the urge for better patient-near preclinical models. Therefore, these data reflect the need for investigations of STAT proteins in a longitudinal patient cohort to identify factors responsible for the diverse influence of system therapy on STAT expression.

Automated detection of the HER2 gene amplification status in Fluorescence in situ hybridization images for the diagnostics of cancer tissues.

The human epidermal growth factor receptor 2 (HER2) gene amplification status is a crucial marker for evaluating clinical therapies of breast or gastric cancer. We propose a deep learning-based pipeline for the detection, localization and classification of interphase nuclei depending on their HER2 gene amplification state in Fluorescence in situ hybridization (FISH) images. Our pipeline combines two RetinaNet-based object localization networks which are trained (1) to detect and classify interphase nuclei into distinct classes normal, low-grade and high-grade and (2) to detect and classify FISH signals into distinct classes HER2 or centromere of chromosome 17 (CEN17). By independently classifying each nucleus twice, the two-step pipeline provides both robustness and interpretability for the automated detection of the HER2 amplification status. The accuracy of our deep learning-based pipeline is on par with that of three pathologists and a set of 57 validation images containing several hundreds of nuclei are accurately classified. The automatic pipeline is a first step towards assisting pathologists in evaluating the HER2 status of tumors using FISH images, for analyzing FISH images in retrospective studies, and for optimizing the documentation of each tumor sample by automatically annotating and reporting of the HER2 gene amplification specificities.

Comparative preclinical evaluation of a polymer-free sirolimus-eluting stent in porcine coronary arteries.

BACKGROUND:: Polymer-free drug-eluting stents (DES) without permanent-polymer coating may be associated with rapid vessel healing, providing a rationale to reduce dual-antiplatelet therapy (DAPT). The aim of the current study was to compare vessel healing of a polymer-free sirolimus-eluting stent (PF-SES), its bare metal stent (BMS) analogue to a permanent polymer-based sirolimus-eluting stent (SES) with proven effectiveness in porcine coronary arteries. MATERIAL AND METHODS:: An ultrathin-strut cobalt-chromium PF-SES, its BMS analogue and an SES with a permanent polymer were used to study vessel healing and their antistenotic potential. Stents were implanted in porcine coronary arteries for histopathologic analysis at 7, 28 and 180 days. In an additional in vitro study, the thrombogenicity of PF-SES was compared with a fluoropolymer-coated everolimus-eluting stent (EES) which demonstrated low stent thrombosis rates in numerous studies. RESULTS:: In the animal study, neointimal growth and injury scores were minimal and inflammation scores were low in the neointima and adventitia in all study groups. After 28 days, neointimal area was lowest in PF-SES when compared with SES and BMS (1.48 ± 0.55 mm² versus 2.43 ± 0.69 mm² versus 1.90 ± 0.85 mm², respectively, p < 0.05) and endothelialization of luminal surfaces was nearly complete in all groups, though SES show the least coverage with occasional adherent luminal inflammatory cells ( p > 0.05). At 180 days, neointimal area and thickness were most pronounced in SES ( p < 0.05) and comparable with BMS implantations, which were characterized by nearly completed vessel healing. PF-SES and BMS had complete endothelialization, absence of fibrin and sustained low inflammatory reaction when compared with the permanent polymer-based SES (inflammation score: PF-SES 0.41 ± 0.74 versus SES 2.52 ± 1.72 versus BMS 0.30 ± 0.65, respectively, p < 0.05 BMS versus SES). Granuloma formation and fibrin accumulation were most pronounced in SES but did not reach statistical significance, p > 0.05). In the in vitro thrombogenicity study, the PF-SES confirmed comparable antithrombogenic properties with regard to the parameters fibrin and platelet binding, and platelet aggregation when compared with the EES. CONCLUSIONS:: As compared with BMS, the ultrathin-strut cobalt-chromium PF-SES showed similar endothelialization at 28 days and comparable healing characteristics at 180 days efficacious inhibition of neointimal proliferation in porcine coronary arteries with low inflammation responses and a BMS-like endothelialization at 180 days. In addition, in an in vitro model, the PF-SES also confirmed low thrombogenicity as compared with the EES.

High-throughput sequencing of the entire genomic regions of CCM1/KRIT1, CCM2 and CCM3/PDCD10 to search for pathogenic deep-intronic splice mutations in cerebral cavernous malformations.

Cerebral cavernous malformations (CCM) are vascular lesions of the central nervous system that can cause headaches, seizures and hemorrhagic stroke. Disease-associated mutations have been identified in three genes: CCM1/KRIT1, CCM2 and CCM3/PDCD10. The precise proportion of deep-intronic variants in these genes and their clinical relevance is yet unknown. Here, a long-range PCR (LR-PCR) approach for target enrichment of the entire genomic regions of the three genes was combined with next generation sequencing (NGS) to screen for coding and non-coding variants. NGS detected all six CCM1/KRIT1, two CCM2 and four CCM3/PDCD10 mutations that had previously been identified by Sanger sequencing. Two of the pathogenic variants presented here are novel. Additionally, 20 stringently selected CCM index cases that had remained mutation-negative after conventional sequencing and exclusion of copy number variations were screened for deep-intronic mutations. The combination of bioinformatics filtering and transcript analyses did not reveal any deep-intronic splice mutations in these cases. Our results demonstrate that target enrichment by LR-PCR combined with NGS can be used for a comprehensive analysis of the entire genomic regions of the CCM genes in a research context. However, its clinical utility is limited as deep-intronic splice mutations in CCM1/KRIT1, CCM2 and CCM3/PDCD10 seem to be rather rare.