Search

Background Personalized medicine requires the integration and analysis of vast amounts of patient data to realize individualized care. With Surgomics, we aim to facilitate personalized therapy recommendations in surgery by integration of intraoperative surgical data and their analysis with machine learning methods to leverage the potential of this data in analogy to Radiomics and Genomics. Methods We defined Surgomics as the entirety of surgomic features that are process characteristics of a surgical procedure automatically derived from multimodal intraoperative data to quantify processes in the operating room. In a multidisciplinary team we discussed potential data sources like endoscopic videos, vital sign monitoring, medical devices and instruments and respective surgomic features. Subsequently, an online questionnaire was sent to experts from surgery and (computer) science at multiple centers for rating the features’ clinical relevance and technical feasibility. Results In total, 52 surgomic features were identified and assigned to eight feature categories. Based on the expert survey (n = 66 participants) the feature category with the highest clinical relevance as rated by surgeons was “surgical skill and quality of performance” for morbidity and mortality (9.0 ± 1.3 on a numerical rating scale from 1 to 10) as well as for long-term (oncological) outcome (8.2 ± 1.8). The feature category with the highest feasibility to be automatically extracted as rated by (computer) scientists was “Instrument” (8.5 ± 1.7). Among the surgomic features ranked as most relevant in their respective category were “intraoperative adverse events”, “action performed with instruments”, “vital sign monitoring”, and “difficulty of surgery”. Conclusion Surgomics is a promising concept for the analysis of intraoperative data. Surgomics may be used together with preoperative features from clinical data and Radiomics to predict postoperative morbidity, mortality and long-term outcome, as well as to provide tailored feedback for surgeons. Graphical abstract

Surgery faces a paradigm shift since it has developed rapidly in recent decades, becoming a high-tech discipline. Increasingly powerful technological developments such as modern operating rooms, featuring digital and interconnected equipment and novel imaging as well as robotic procedures, provide several data sources resulting in a huge potential to improve patient therapy and surgical outcome by means of Surgical Data Science. The emerging field of Surgical Data Science aims to improve the quality of surgery through acquisition, organization, analysis, and modeling of data, in particular using machine learning (ML). An integral part of surgical data science is to analyze the available data along the surgical treatment path and provide a context-aware autonomous action by means of ML methods. Autonomous actions related to surgical decision-making include preoperative decision support, intraoperative assistance functions, as well as robot-assisted actions. The goal is to democratize surgical skills and enhance the collaboration between surgeons and cyber-physical systems by quantifying surgical experience and making it accessible to machines, thereby improving patient therapy and outcome. The article introduces basic ML concepts as enablers for autonomous actions in surgery, highlighting examples for such actions along the surgical treatment path.

StructuredAbstractBackgroundLack of anatomy recognition represents a clinically relevant risk in abdominal surgery. Machine learning (ML) methods can help identify visible patterns and risk structures, however, their practical value remains largely unclear.Materials and MethodsBased on a novel dataset of 13195 laparoscopic images with pixel-wise segmentations of eleven anatomical structures, we developed specialized segmentation models for each structure and combined models for all anatomical structures using two state-of-the-art model architectures (DeepLabv3 and SegFormer), and compared segmentation performance of algorithms to a cohort of 28 physicians, medical students, and medical laypersons using the example of pancreas segmentation.ResultsMean Intersection-over-Union for semantic segmentation of intraabdominal structures ranged from 0.28 to 0.83 and from 0.23 to 0.77 for the DeepLabv3-based structure-specific and combined models, and from 0.31 to 0.85 and from 0.26 to 0.67 for the SegFormer-based structure-specific and combined models, respectively. Both the structure-specific and the combined DeepLabv3-based models are capable of near-real-time operation, while the SegFormer-based models are not. All four models outperformed at least 26 out of 28 human participants in pancreas segmentation.ConclusionsThese results demonstrate that ML methods have the potential to provide relevant assistance in anatomy recognition in minimally-invasive surgery in near-real-time. Future research should investigate the educational value and subsequent clinical impact of respective assistance systems.HighlightsMachine learning models to reduce surgical risks that precisely identify 11 anatomical structures: abdominal wall, colon, intestinal vessels (inferior mesenteric artery and inferior mesenteric vein with their subsidiary vessels), liver, pancreas, small intestine, spleen, stomach, ureter and vesicular glandsLarge training dataset of 13195 real-world laparoscopic images with high-quality anatomy annotationsSimilar performance of individual segmentation models for each structure and combined segmentation models in identifying intraabdominal structures, and similar segmentation performance of DeepLabv3-based and SegFormer-based modelsDeepLabv3-based models are capable of near-real-time operation while SegFormer-based models are not, but SegFormer-based models outperform DeepLabv3-based models in terms of accuracy and generalizabilityAll models outperformed at least 26 out of 28 human participants in pancreas segmentation, demonstrating their potential for real-time assistance in recognizing anatomical landmarks during minimally-invasive surgery.

Cancer of unknown primary (CUP) account for 2-5% of all cancer diagnoses, wherein standard investigations fail to reveal the original tumor site. Basket trials allocate targeted therapeutics based on actionable somatic mutations, independent of tumor entity. These trials, however, mostly rely on variants identified in tissue biopsies from solid tumors. Since liquid biopsies (LB) represent the overall tumor genomic landscape, they may provide an ideal diagnostic source in patients with CUP. The most informative liquid biopsy compartment, however, remains to be identified. We aimed to compare the utility of genomic variant analysis for therapy stratification in CUP patients in two liquid biopsy compartments: circulating cell-free (cf) and extracellular vesicle (ev) DNA. CfDNA and evDNA from 23 patients with CUP were analyzed contemporaneously using a targeted gene panel covering 151 cancer-associated genes. LB revealed a total of 22 somatic mutations in evDNA and/or cfDNA in 11/23 patients. Out of the 22 identified somatic variants, 14 are classified as Tier I druggable somatic variants according to the MetaKB knowledgebase. Comparison of variants identified in evDNA and cfDNA revealed an overlap of 58% of somatic variants in both LB compartments, whereas over 40% of variants were only found in one or the other compartment. In our cohort, we observed substantial overlap between somatic variants identified in evDNA and cfDNA of CUP patients. Nonetheless, interrogation of both LB compartments can potentially increase the rate of druggable alterations, stressing the significance of liquid biopsies for possible primary-independent basket and umbrella trial inclusion. Significance Analysis of genomic variants in liquid biopsies has potential as a diagnostic tool for CUP samples. We compared cfDNA and evDNA by a targeted NGS approach and identified druggable targets that could help to stratify patients for inclusion into clinical trials.

Artificial intelligence (AI) and computer vision (CV) are beginning to impact medicine. While evidence on the clinical value of AI-based solutions for the screening and staging of colorectal cancer (CRC) is mounting, CV and AI applications to enhance the surgical treatment of CRC are still in their early stage. This manuscript introduces key AI concepts to a surgical audience, illustrates fundamental steps to develop CV for surgical applications, and provides a comprehensive overview on the state-of-the-art of AI applications for the treatment of CRC. Notably, studies show that AI can be trained to automatically recognize surgical phases and actions with high accuracy even in complex colorectal procedures such as transanal total mesorectal excision (TaTME). In addition, AI models were trained to interpret fluorescent signals and recognize correct dissection planes during total mesorectal excision (TME), suggesting CV as a potentially valuable tool for intraoperative decision-making and guidance. Finally, AI could have a role in surgical training, providing automatic surgical skills assessment in the operating room. While promising, these proofs of concept require further development, validation in multi-institutional data, and clinical studies to confirm AI as a valuable tool to enhance CRC treatment.

IntroductionComplex oncological procedures pose various surgical challenges including dissection in distinct tissue planes and preservation of vulnerable anatomical structures throughout different surgical phases. In rectal surgery, violation of dissection planes increases the risk of local recurrence and autonomous nerve damage resulting in incontinence and sexual dysfunction. This work explores the feasibility of phase recognition and target structure segmentation in robot-assisted rectal resection (RARR) using machine learning.Materials and MethodsA total of 57 RARR were recorded and annotated with respect to surgical phases and exact locations of target structures (anatomical structures, tissue types, static structures, and dissection areas). For surgical phase recognition, three machine learning models were trained: LSTM, MSTCN, and TransSVNet. Based on pixel-wise annotations of target structures in 9037 images, individual segmentation models based on DeepLabV3 were trained. Model performance was evaluated using F1 score, Intersection-over-Union (IoU), accuracy, precision, recall, and specificity.ResultsThe best results for phase recognition were achieved with the MSTCN model (F1 score: 0.82 ± 0.01, accuracy: 0.84 ± 0.03). Mean IoUs for target structure segmentation ranged from 0.14 ± 0.22 to 0.80 ± 0.14 for organs and tissue types and from 0.11 ± 0.11 to 0.44 ± 0.30 for dissection areas. Image quality, distorting factors (i.e. blood, smoke), and technical challenges (i.e. lack of depth perception) considerably impacted segmentation performance.ConclusionMachine learning-based phase recognition and segmentation of selected target structures are feasible in RARR. In the future, such functionalities could be integrated into a context-aware surgical guidance system for rectal surgery.

Clinically relevant postoperative pancreatic fistula (CR-POPF) is a common severe surgical complication after pancreatic surgery. Current risk stratification systems mostly rely on intraoperatively assessed factors like manually determined gland texture or blood loss. We developed a preoperatively available image-based risk score predicting CR-POPF as a complication of pancreatic head resection. Frequency of CR-POPF and occurrence of salvage completion pancreatectomy during the hospital stay were associated with an intraoperative surgical (sFRS) and image-based preoperative CT-based (rFRS) fistula risk score, both considering pancreatic gland texture, pancreatic duct diameter and pathology, in 195 patients undergoing pancreatic head resection. Based on its association with fistula-related outcome, radiologically estimated pancreatic remnant volume was included in a preoperative (preFRS) score for POPF risk stratification. Intraoperatively assessed pancreatic duct diameter (p p p p p p = 0.001), and estimated pancreatic remnant volume (p

High histone deacetylase (HDAC) 8 and HDAC10 expression levels have been identified as predictors of exceptionally poor outcomes in neuroblastoma, the most common extracranial solid tumor in childhood. HDAC8 inhibition synergizes with retinoic acid treatment to induce neuroblast maturation in vitro and to inhibit neuroblastoma xenograft growth in vivo. HDAC10 inhibition increases intracellular accumulation of chemotherapeutics through interference with lysosomal homeostasis, ultimately leading to cell death in cultured neuroblastoma cells. So far, no HDAC inhibitor covering HDAC8 and HDAC10 at micromolar concentrations without inhibiting HDACs 1, 2 and 3 has been described. Here, we introduce TH34 (3-(N-benzylamino)-4-methylbenzhydroxamic acid), a novel HDAC6/8/10 inhibitor for neuroblastoma therapy. TH34 is well-tolerated by non-transformed human skin fibroblasts at concentrations up to 25 µM and modestly impairs colony growth in medulloblastoma cell lines, but specifically induces caspase-dependent programmed cell death in a concentration-dependent manner in several human neuroblastoma cell lines. In addition to the induction of DNA double-strand breaks, HDAC6/8/10 inhibition also leads to mitotic aberrations and cell-cycle arrest. Neuroblastoma cells display elevated levels of neuronal differentiation markers, mirrored by formation of neurite-like outgrowths under maintained TH34 treatment. Eventually, after long-term treatment, all neuroblastoma cells undergo cell death. The combination of TH34 with plasma-achievable concentrations of retinoic acid, a drug applied in neuroblastoma therapy, synergistically inhibits colony growth (combination index (CI)

The prognosis of advanced stage neuroblastoma patients remains poor and, despite intensive therapy, the 5-year survival rate remains less than 50%. We previously identified histone deacetylase (HDAC) 8 as an indicator of poor clinical outcome and a selective drug target for differentiation therapy in vitro and in vivo. Here, we performed kinome-wide RNAi screening to identify genes that are synthetically lethal with HDAC8 inhibitors. These experiments identified the neuroblastoma predisposition gene ALK as a candidate gene. Accordingly, the combination of the ALK/MET inhibitor crizotinib and selective HDAC8 inhibitors (3-6 µM PCI-34051 or 10 µM 20a) efficiently killed neuroblastoma cell lines carrying wildtype ALK (SK-N-BE(2)-C, IMR5/75), amplified ALK (NB-1), and those carrying the activating ALK F1174L mutation (Kelly), and, in cells carrying the activating R1275Q mutation (LAN-5), combination treatment decreased viable cell count. The effective dose of crizotinib in neuroblastoma cell lines ranged from 0.05 µM (ALK-amplified) to 0.8 µM (wildtype ALK). The combinatorial inhibition of ALK and HDAC8 also decreased tumor growth in an in vivo zebrafish xenograft model. Bioinformatic analyses revealed that the mRNA expression level of HDAC8 was significantly correlated with that of ALK in two independent patient cohorts, the Academic Medical Center cohort (n = 88) and the German Neuroblastoma Trial cohort (n = 649), and co-expression of both target genes identified patients with very poor outcome. Mechanistically, HDAC8 and ALK converge at the level of receptor tyrosine kinase (RTK) signaling and their downstream survival pathways, such as ERK signaling. Combination treatment of HDAC8 inhibitor with crizotinib efficiently blocked the activation of growth receptor survival signaling and shifted the cell cycle arrest and differentiation phenotype toward effective cell death of neuroblastoma cell lines, including sensitization of resistant models, but not of normal cells. These findings reveal combined targeting of ALK and HDAC8 as a novel strategy for the treatment of neuroblastoma.

Histone deacetylase 8 (HDAC8), a unique class I zinc-dependent HDAC, is an emerging target in cancer and other diseases. Its substrate repertoire extends beyond histones to many nonhistone proteins. Besides being a deacetylase, HDAC8 also mediates signaling via scaffolding functions. Aberrant expression or deregulated interactions with transcription factors are critical in HDAC8-dependent cancers. Many potent HDAC8-selective inhibitors with cellular activity and anticancer effects have been reported. We present HDAC8 as a druggable target and discuss inhibitors of different chemical scaffolds with cellular effects. Furthermore, we review HDAC8 activators that revert activity of mutant enzymes. Isotype-selective HDAC8 targeting in patients with HDAC8-relevant cancers is challenging, however, is promising to avoid adverse side effects as observed with pan-HDAC inhibitors.

Drug resistance is a leading cause for treatment failure in many cancers, including neuroblastoma, the most common solid extracranial childhood malignancy. Previous studies from our lab indicate that histone deacetylase 10 (HDAC10) is important for the homeostasis of lysosomes, i.e. acidic vesicular organelles involved in the degradation of various biomolecules. Here, we show that depleting or inhibiting HDAC10 results in accumulation of lysosomes in chemotherapy-resistant neuroblastoma cell lines, as well as in the intracellular accumulation of the weakly basic chemotherapeutic doxorubicin within lysosomes. Interference with HDAC10 does not block doxorubicin efflux from cells via P-glycoprotein inhibition, but rather via inhibition of lysosomal exocytosis. In particular, intracellular doxorubicin does not remain trapped in lysosomes but also accumulates in the nucleus, where it promotes neuroblastoma cell death. Our data suggest that lysosomal exocytosis under doxorubicin treatment is important for cell survival and that inhibition of HDAC10 further induces DNA double-strand breaks (DSBs), providing additional mechanisms that sensitize neuroblastoma cells to doxorubicin. Taken together, we demonstrate that HDAC10 inhibition in combination with doxorubicin kills neuroblastoma, but not non-malignant cells, both by impeding drug efflux and enhancing DNA damage, providing a novel opportunity to target chemotherapy resistance.

Histone deacetylases (HDACs) are important modulators of epigenetic gene regulation and additionally control the activity of non-histone protein substrates. While for HDACs 1–3 and 6 many potent selective inhibitors have been obtained, for other subtypes much less is known on selective inhibitors and the consequences of their inhibition. The present report describes the development of substituted benzhydroxamic acids as potent and selective HDAC8 inhibitors. Docking studies using available crystal structures have been used for structure-based optimization of this series of compounds. Within this study, we have investigated the role of HDAC8 in the proliferation of cancer cells and optimized hits for potency and selectivity, both in vitro and in cell culture. The combination of structure-based design, synthesis, and in vitro screening to cellular testing resulted in potent and selective HDAC8 inhibitors that showed anti-neuroblastoma activity in cellular testing.