Search

Endometrial carcinoma (EC) is one of the most common gynecological cancers in the world. In this work we apply Cox proportional hazards (CPH) and optimal survival tree (OST) algorithms to the retrospective prognostic modeling of disease-specific survival in 842 EC patients. We demonstrate that the linear CPH models are preferred for the EC risk assessment based on clinical features alone, while the interpretable, non-linear OST models are favored when patient profiles are enriched with tumor molecular data. By studying the OST decision path structure, we show how explainable tree models recapitulate existing clinical knowledge prioritizing L1 cell-adhesion molecule and estrogen receptor status indicators as key risk factors in the p53 abnormal EC subgroup. We believe that visually interpretable tree algorithms are a promising method to explore feature interactions and generate novel research hypotheses. To aid further clinical adoption of advanced machine learning techniques, we stress the importance of quantifying model discrimination and calibration performance in the development of explainable clinical prediction models.

There are three prognostic stratification tools used for endometrial cancer: ESMO-ESGO-ESTRO 2016, ProMisE, and ESGO-ESTRO-ESP 2020. However, these methods are not sufficiently accurate to address prognosis. The aim of this study was to investigate whether the integration of molecular classification and other biomarkers could be used to improve the prognosis stratification in early-stage endometrial cancer. Relapse-free and overall survival of each classifier were analyzed, and the c-index was employed to assess accuracy. Other biomarkers were explored to improve the precision of risk classifiers. We analyzed 293 patients. A comparison between the three classifiers showed an improved accuracy in ESGO-ESTRO-ESP 2020 when RFS was evaluated (c-index = 0.78), although we did not find broad differences between intermediate prognostic groups. Prognosis of these patients was better stratified with the incorporation of CTNNB1 status to the 2020 classifier (c-index 0.81), with statistically significant and clinically relevant differences in 5-year RFS: 93.9% for low risk, 79.1% for intermediate merged group/CTNNB1 wild type, and 42.7% for high risk (includ-ing patients with CTNNB1 mutation). The incorporation of molecular classification in risk stratification resulted in better discriminatory capability, which could be improved even further with the addition of CTNNB1 mutational evaluation, This research was funded by the Instituto de Salud Carlos III (ISCIII), cofinanced by the European Development Regional Fund ‘A way to achieve Europe’ (FEDER, PI17/01723 to D.H.), and by Academy of Finland grant (No. 317680 to J.T.). B.Z. is supported by the faculty funded PhD position in the Integrative Life Science Doctoral Programme, University of Helsinki

Application of machine and deep learning methods in drug discovery and cancer research has gained a considerable amount of attention in the past years. As the field grows, it becomes crucial to systematically evaluate the performance of novel computational solutions in relation to established techniques. To this end, we compare rule-based and data-driven molecular representations in prediction of drug combination sensitivity and drug synergy scores using standardized results of 14 high-throughput screening studies, comprising 64 200 unique combinations of 4153 molecules tested in 112 cancer cell lines. We evaluate the clustering performance of molecular representations and quantify their similarity by adapting the Centered Kernel Alignment metric. Our work demonstrates that to identify an optimal molecular representation type, it is necessary to supplement quantitative benchmark results with qualitative considerations, such as model interpretability and robustness, which may vary between and throughout preclinical drug development projects., Graphical Abstract Graphical Abstract

Application of machine and deep learning methods in drug discovery and cancer research has gained a considerable amount of attention in the past years. As the field grows, it becomes crucial to systematically evaluate the performance of novel computational solutions in relation to established techniques. To this end we compare rule-based and data-driven molecular representations in prediction of drug combination sensitivity and drug synergy scores using standardized results of 14 throughput screening studies, comprising 64 200 unique combinations of 4 153 molecules tested in 112 cancer cell lines. We evaluate the clustering performance of molecular representations and quantify their similarity by adapting the Centered Kernel Alignment metric. Our work demonstrates that to identify an optimal molecular representation type it is necessary to supplement quantitative benchmark results with qualitative considerations, such as model interpretability and robustness, which may vary between and throughout preclinical drug development projects.Abstract Figure

Drug combination therapy has the potential to enhance efficacy, reduce dose-dependent toxicity and prevent the emergence of drug resistance. However, discovery of synergistic and effective drug combinations has been a laborious and often serendipitous process. In recent years, identification of combination therapies has been accelerated due to the advances in high-throughput drug screening, but informatics approaches for systems-level data management and analysis are needed. To contribute toward this goal, we created an open-access data portal (https://drugcomb.fimm.fi) where the results of drug combination screening studies are accumulated, standardized and harmonized. Through the data portal, we provided the web server to analyze and visualize users’ own drug combination screening data. The users have an option to upload their data to DrugComb, as part of a crowdsourcing data curation effort. To initiate the data repository, we collected 437,932 drug combinations tested on a variety of cancer cell lines. We showed that linear regression approaches, when considering chemical fingerprints as predictors, have the potential to achieve high accuracy of predicting the sensitivity and synergy of drug combinations. All the data and informatics tools are freely available in DrugComb to enable a more efficient utilization of data resources for future drug combination discovery.

Drug combination therapy has the potential to enhance efficacy, reduce dose-dependent toxicity and prevent the emergence of drug resistance. However, discovery of synergistic and effective drug combinations has been a laborious and often serendipitous process. In recent years, identification of combination therapies has been accelerated due to the advances in high-throughput drug screening, but informatics approaches for systems-level data management and analysis are needed. To contribute toward this goal, we created an open-access data portal called DrugComb (https://drugcomb.fimm.fi) where the results of drug combination screening studies are accumulated, standardized and harmonized. Through the data portal, we provided a web server to analyze and visualize users’ own drug combination screening data. The users can also effectively participate a crowdsourcing data curation effect by depositing their data at DrugComb. To initiate the data repository, we collected 437 932 drug combinations tested on a variety of cancer cell lines. We showed that linear regression approaches, when considering chemical fingerprints as predictors, have the potential to achieve high accuracy of predicting the sensitivity of drug combinations. All the data and informatics tools are freely available in DrugComb to enable a more efficient utilization of data resources for future drug combination discovery.

The Magnetic Imaging Defectoscope (MID) is a tool designed to record the time response to high-energy electromagnetic pulses. Metal presence around the tool is evaluated by matching measured and numerically modelled magnetisation decays. The air response decays faster than if the tool is placed inside a metal pipe. If multiple pipes are present around the tool, the near pipe responds at earlier times and the more distant ones at later times. This phenomenon is key to detecting metal presence in each individual pipe. The MID tool features two sensors, short and long, to ensure reliable detection of responses from a wide range of multi-barrier completions and differentiate between defects in each barrier independently. Despite this simple concept, the response detected by the tool has a complex time behaviour and cannot be modelled in a simple analytical way, which put time-domain technologies on hold for many years. The recent advances in computer speed and multicore parallel computing enable accurate numerical modelling of complex responses and determining metal presence in two metal barriers separately. In addition to thickness profiles for each barrier, the MID tool generates differential (DELTA) data panels, based on the difference between numerical models and actual responses, that visualise electromagnetic signatures of metal structures and can be used to locate and recognise various completion components, such as packers or SSD. The paper describes the basics of the time-domain magnetic-pulse technology, specifications of the tool itself and laboratory test data. This technology has been tested in a dual-string producer of the Raudhatain field operated by the Kuwait Oil Company. The long and short strings were suspected to be communicating due to identical water cut trends. The MID tool was run once in memory mode on slickline through the long string and detected a 1.5-inch corrosion hole between packers that created communication with the short string. This communication flow left a footprint in a shut-in temperature log right across the corrosion hole. Another corrosion hole was located in the casing below the tubing shoe, immediately above the perforations. During a subsequent workover, the tubing was retrieved and a 1.5-inch corrosion hole was located in the exact place where it was identified by the MID tool.

This paper presents downhole magnetic imaging defectoscopy (MID) in memory mode, its key principles, differences from other corrosion logging technologies, and some results of its application in oil wells. The MID technology is designed to check the integrity of magnetic and non-magnetic tubing and casing strings in oil and gas wells. It can be used to detect various defects, corrosion and mechanical wear, and to assess the quality of perforations. The MID tool contains two high-sensitivity sensors: a short generator/receiver coil and a long one with short relaxation times to analyse responses even at early times of 0.1 ms. The short coil generates a short, low-amplitude electromagnetic pulse and records a response from the first metal barrier, normally tubing, where as the long coil generates a long, high-amplitude electromagnetic pulse and records a signal travelling a much longer distance of up to 13 inches to record the total response from tubing and casing. Mathematical processing of these responses can determine the thicknesses of the first and second metal barriers. An independent power supply allows logging to be performed on slickline, which substantially reduces operating costs and enables the use of magnetic imaging defectoscopy for well monitoring to prevent leak and corrosion related incidents and thus increase environmental safety in the field. The wide use of memory magnetic imaging defectoscopy will substantially reduce the number of well incidents and expenses for remedial jobs.

BACKGROUND: Cytotoxic chemotherapy/agents can cause a range of side effects in cancer patients including anemia, neutropenia and thrombocytopenia, resulting in increased morbidity and mortality. These cytopenias are in part due to massive depletion of bone marrow progenitors. A common target of many chemotherapies is the megakaryocyte (MK), a rare progenitor representing 0.2% of the bone marrow. Loss of MK lineage cells results in thrombocytopenia, therefore, discovering ways to protect MKs from cytotoxic drugs could prevent this life-threatening condition. Here, we show that eltrombopag (EP), a small molecule, non-peptide thrombopoietin receptor (TPO-R) agonist, protects healthy MKs from different cytotoxic agents. We further explored the impact of cytotoxic drugs and EP treatment on signaling molecules active in MKs. METHODS: Peripheral blood and bone marrow samples were collected from healthy donors after informed consent and following protocols in accordance with the Declaration of Helsinki. CD34+ positive cells were isolated by immuno-magnetic bead separation from the samples and expanded for 8-10 days with a cytokine cocktail to induce MK differentiation. After MK expansion, cells were transferred to 96-well plates pretreated with 14 different drugs including cytarabine, gemcitabine, paclitaxel, carboplatin, cisplatin, doxorubicin, vincristine, etoposide, midostaurin, ruxolitinib, panobinostat, azacitidine, venetoclax and navitoclax. The drugs were plated in 5 different doses in a 10,000-fold concentration range, and cells added with or without EP. After 3 days incubation, the cells were analyzed on a high throughput flow cytometer using Annexin V and 7AAD to distinguish live from dead cells. To understand the impact of the drugs and EP on signaling molecules downstream of TPO-R, we analyzed phosphorylation of AKT, ERK, STAT3 and STAT5 in populations defined by MK markers CD41a, CD42b and CD110 (TPO-R). RESULTS: We found that EP overall supports MK survival in the presence of cytotoxic agents. The greatest net effect was observed when EP was combined with BCL2 inhibitors (venetoclax, navitoclax), which resulted in increased MK cell maturation compared to inhibitor alone, while gemcitabine plus EP showed the next best response. EP combined with etoposide, vincristine, cytarabine, paclitaxel, cisplatin and ruxolitinib resulted in an increase of immature MKs without a reduction in mature MK cell numbers. Higher numbers of MKs were observed when EP was combined with midostaurin, carboplatin, panobinostat and doxorubicin, compared to treatment without EP; although these numbers were lower than other tested drugs. Phosphoflow analysis revealed intriguing differences in EP-induced signaling compared to signaling induced by recombinant human thrombopoietin (rhTPO) (Figure 1). Also, EP-induced signaling did not correlate with TPO-R expression. EP activated AKT in all MK subsets, but rhTPO induced AKT only in immature MKs. While rhTPO induced ERK phosphorylation in different MK subsets, EP had no effect on ERK activation. Furthermore, rhTPO activated STAT3 only in mature MKs, but EP induced STAT3 in all MK populations. In contrast, both EP and rhTPO induced phosphorylation of STAT5 in all MK subsets. Overall signaling was inhibited by a selected set of agents (venetoclax, etoposide, midostaurin) compared to basal signaling, but addition of EP prevented the inhibition. Similar to the cell survival analysis, venetoclax combined with EP resulted in complete recovery of signaling activity compared to venetoclax alone. But the response was not as striking for etoposide or midostaurin combination compared to treatment alone. CONCLUSION: Using MKs expanded from CD34+ cells, we found that EP supports MK survival in the presence of several different cytotoxic agents, with the most striking rescue observed when EP was combined with BCL2 inhibitors. Importantly, EP and rhTPO induced distinct signaling patterns, with EP signaling independent of TPO-R expression, suggesting alternate EP targets. The tested drugs inhibited basal signaling in MKs, but this inhibition was prevented by EP. These results suggest addition of EP to cancer treatment regimens may prevent the inhibitory effects of some agents on MKs and support future investigations to demonstrate clinical efficacy of EP in combinations with some of these cytotoxic agents to alleviate thrombocytopenia. Figure 1. Figure 1. Disclosures Javarappa: Novartis Pharma AG: Research Funding. Tsallos:Novartis: Research Funding. Marques Ramos:Novartis: Employment. Pallaud:Novartis: Employment. Heckman:Celgene: Research Funding; Novartis: Research Funding; Orion Pharma: Research Funding.