Showing papers by "University of Cagliari published in 2018"

Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning

Abstract: Deep neural networks and machine-learning algorithms are pervasively used in several applications, ranging from computer vision to computer security. In most of these applications, the learning algorithm has to face intelligent and adaptive attackers who can carefully manipulate data to purposely subvert the learning process. As these algorithms have not been originally designed under such premises, they have been shown to be vulnerable to well-crafted, sophisticated attacks, including training-time poisoning and test-time evasion attacks (also known as adversarial examples). The problem of countering these threats and learning secure classifiers in adversarial settings has thus become the subject of an emerging, relevant research field known as adversarial machine learning. The purposes of this tutorial are: (a) to introduce the fundamentals of adversarial machine learning to the security community; (b) to illustrate the design cycle of a learning-based pattern recognition system for adversarial tasks; (c) to present novel techniques that have been recently proposed to assess performance of pattern classifiers and deep learning algorithms under attack, evaluate their vulnerabilities, and implement defense strategies that make learning algorithms more robust to attacks; and (d) to show some applications of adversarial machine learning to pattern recognition tasks like object recognition in images, biometric identity recognition, spam and malware detection.

Wild patterns: Ten years after the rise of adversarial machine learning half-day tutorial

Abstract: Deep neural networks and machine-learning algorithms are pervasively used in several applications, ranging from computer vision to computer security. In most of these applications, the learning algorithm has to face intelligent and adaptive attackers who can carefully manipulate data to purposely subvert the learning process. As these algorithms have not been originally designed under such premises, they have been shown to be vulnerable to well-crafted, sophisticated attacks, including training-time poisoning and test-time evasion attacks (also known as adversarial examples). The problem of countering these threats and learning secure classifiers in adversarial settings has thus become the subject of an emerging, relevant research field known as adversarial machine learning. The purposes of this tutorial are: (a) to introduce the fundamentals of adversarial machine learning to the security community; (b) to illustrate the design cycle of a learning-based pattern recognition system for adversarial tasks; (c) to present novel techniques that have been recently proposed to assess performance of pattern classifiers and deep learning algorithms under attack, evaluate their vulnerabilities, and implement defense strategies that make learning algorithms more robust to attacks; and (d) to show some applications of adversarial machine learning to pattern recognition tasks like object recognition in images, biometric identity recognition, spam and malware detection.

DarkSide-20k: A 20 tonne two-phase LAr TPC for direct dark matter detection at LNGS

Abstract: Building on the successful experience in operating the DarkSide-50 detector, the DarkSide Collaboration is going to construct DarkSide-20k, a direct WIMP search detector using a two-phase Liquid Argon Time Projection Chamber (LAr TPC) with an active (fiducial) mass of 23 t (20 t). This paper describes a preliminary design for the experiment, in which the DarkSide-20k LAr TPC is deployed within a shield/veto with a spherical Liquid Scintillator Veto (LSV) inside a cylindrical Water Cherenkov Veto (WCV). This preliminary design provides a baseline for the experiment to achieve its physics goals, while further development work will lead to the final optimization of the detector parameters and an eventual technical design. Operation of DarkSide-50 demonstrated a major reduction in the dominant 39Ar background when using argon extracted from an underground source, before applying pulse shape analysis. Data from DarkSide-50, in combination with MC simulation and analytical modeling, shows that a rejection factor for discrimination between electron and nuclear recoils of $>3 \times 10^{9}$ is achievable. This, along with the use of the veto system and utilizing silicon photomultipliers in the LAr TPC, are the keys to unlocking the path to large LAr TPC detector masses, while maintaining an experiment in which less than $< 0.1$ events (other than $ u$ -induced nuclear recoils) is expected to occur within the WIMP search region during the planned exposure. DarkSide-20k will have ultra-low backgrounds than can be measured in situ, giving sensitivity to WIMP-nucleon cross sections of $1.2 \times 10^{-47}$ cm2 ( $1.1 \times 10^{-46}$ cm2) for WIMPs of 1 TeV/c2 (10 TeV/c2) mass, to be achieved during a 5 yr run producing an exposure of 100 t yr free from any instrumental background.

An updated checklist of the vascular flora native to Italy

Abstract: An updated inventory of the vascular flora alien to Italy, providing details on the occurrence at regional level, is presented. The checklist includes 1597 species, subspecies, and hybrids, distributed in 725 genera and 152 families; 2 taxa are lycophytes, 11 ferns and fern allies, 33 gymnosperms, and 1551 angiosperms. 157 taxa are archaeophytes and 1440 neophytes. The alien taxa currently established in Italy are 791 (570 naturalized and 221 invasive), while 705 taxa are casual aliens, 4 are not assessed, 7 are of unknown regional distribution, 47 have not been confirmed in recent times, 3 are considered extinct or possibly extinct in the country, and 40 are doubtfully occurring in Italy. This checklist allows to establish an up-to-date number (9792) of taxa constituting the whole (native and alien) Italian flora.

Low-Mass Dark Matter Search with the DarkSide-50 Experiment

Abstract: We present the results of a search for dark matter weakly interacting massive particles (WIMPs) in the mass range below 20 GeV/c2 using a target of low-radioactivity argon with a 6786.0 kg d exposure. The data were obtained using the DarkSide-50 apparatus at Laboratori Nazionali del Gran Sasso. The analysis is based on the ionization signal, for which the DarkSide-50 time projection chamber is fully efficient at 0.1 keVee. The observed rate in the detector at 0.5 keVee is about 1.5 event/keVee/kg/d and is almost entirely accounted for by known background sources. We obtain a 90% C.L. exclusion limit above 1.8 GeV/c2 for the spin-independent cross section of dark matter WIMPs on nucleons, extending the exclusion region for dark matter below previous limits in the range 1.8–6 GeV/c2.

Constraints on sub-GeV dark matter-electron scattering from the DarkSide-50 experiment

Abstract: We present new constraints on sub-GeV dark-matter particles scattering off electrons based on 6780.0 kg d of data collected with the DarkSide-50 dual-phase argon time projection chamber. This analysis uses electroluminescence signals due to ionized electrons extracted from the liquid argon target. The detector has a very high trigger probability for these signals, allowing for an analysis threshold of three extracted electrons, or approximately 0.05 keVee. We calculate the expected recoil spectra for dark matter-electron scattering in argon and, under the assumption of momentum-independent scattering, improve upon existing limits from XENON10 for dark-matter particles with masses between 30 and 100 MeV/c^{2}.

Adversarial Malware Binaries: Evading Deep Learning for Malware Detection in Executables

Abstract: Machine learning has already been exploited as a useful tool for detecting malicious executable files. Data retrieved from malware samples, such as header fields, instruction sequences, or even raw bytes, is leveraged to learn models that discriminate between benign and malicious software. However, it has also been shown that machine learning and deep neural networks can be fooled by evasion attacks (also known as adversarial examples), i.e., small changes to the input data that cause misclassification at test time. In this work, we investigate the vulnerability of malware detection methods that use deep networks to learn from raw bytes. We propose a gradient-based attack that is capable of evading a recently-proposed deep network suited to this purpose by only changing few specific bytes at the end of each mal ware sample, while preserving its intrusive functionality. Promising results show that our adversarial malware binaries evade the targeted network with high probability, even though less than 1 % of their bytes are modified.

Data Mining for Detecting Bitcoin Ponzi Schemes

Abstract: Soon after its introduction in 2009, Bitcoin has been adopted by cyber-criminals, which rely on its pseudonymity to implement virtually untraceable scams. One of the typical scams that operate on Bitcoin are the so-called Ponzi schemes. These are fraudulent investments which repay users with the funds invested by new users that join the scheme, and implode when it is no longer possible to find new investments. Despite being illegal in many countries, Ponzi schemes are now proliferating on Bitcoin, and they keep alluring new victims, who are plundered of millions of dollars. We apply data mining techniques to detect Bitcoin addresses related to Ponzi schemes. Our starting point is a dataset of features of real-world Ponzi schemes, that we construct by analysing, on the Bitcoin blockchain, the transactions used to perform the scams. We use this dataset to experiment with various machine learning algorithms, and we assess their effectiveness through standard validation protocols and performance metrics. The best of the classifiers we have experimented can identify most of the Ponzi schemes in the dataset, with a low number of false positives.

Carotid Artery Wall Imaging: Perspective and Guidelines from the ASNR Vessel Wall Imaging Study Group and Expert Consensus Recommendations of the American Society of Neuroradiology

Abstract: Identification of carotid artery atherosclerosis is conventionally based on measurements of luminal stenosis and surface irregularities using in vivo imaging techniques including sonography, CT and MR angiography, and digital subtraction angiography. However, histopathologic studies demonstrate considerable differences between plaques with identical degrees of stenosis and indicate that certain plaque features are associated with increased risk for ischemic events. The ability to look beyond the lumen using highly developed vessel wall imaging methods to identify plaque vulnerable to disruption has prompted an active debate as to whether a paradigm shift is needed to move away from relying on measurements of luminal stenosis for gauging the risk of ischemic injury. Further evaluation in randomized clinical trials will help to better define the exact role of plaque imaging in clinical decision-making. However, current carotid vessel wall imaging techniques can be informative. The goal of this article is to present the perspective of the ASNR Vessel Wall Imaging Study Group as it relates to the current status of arterial wall imaging in carotid artery disease.

Is feature selection secure against training data poisoning

Abstract: Learning in adversarial settings is becoming an important task for application domains where attackers may inject malicious data into the training set to subvert normal operation of data-driven technologies. Feature selection has been widely used in machine learning for security applications to improve generalization and computational efficiency, although it is not clear whether its use may be beneficial or even counterproductive when training data are poisoned by intelligent attackers. In this work, we shed light on this issue by providing a framework to investigate the robustness of popular feature selection methods, including LASSO, ridge regression and the elastic net. Our results on malware detection show that feature selection methods can be significantly compromised under attack (we can reduce LASSO to almost random choices of feature sets by careful insertion of less than 5% poisoned training samples), highlighting the need for specific countermeasures.

2017 WSES guidelines on colon and rectal cancer emergencies: obstruction and perforation

Abstract: Obstruction and perforation due to colorectal cancer represent challenging matters in terms of diagnosis, life-saving strategies, obstruction resolution and oncologic challenge. The aims of the current paper are to update the previous WSES guidelines for the management of large bowel perforation and obstructive left colon carcinoma (OLCC) and to develop new guidelines on obstructive right colon carcinoma (ORCC). The literature was extensively queried for focused publication until December 2017. Precise analysis and grading of the literature has been performed by a working group formed by a pool of experts: the statements and literature review were presented, discussed and voted at the Consensus Conference of the 4th Congress of the World Society of Emergency Surgery (WSES) held in Campinas in May 2017. CT scan is the best imaging technique to evaluate large bowel obstruction and perforation. For OLCC, self-expandable metallic stent (SEMS), when available, offers interesting advantages as compared to emergency surgery; however, the positioning of SEMS for surgically treatable causes carries some long-term oncologic disadvantages, which are still under analysis. In the context of emergency surgery, resection and primary anastomosis (RPA) is preferable to Hartmann’s procedure, whenever the characteristics of the patient and the surgeon are permissive. Right-sided loop colostomy is preferable in rectal cancer, when preoperative therapies are predicted. With regards to the treatment of ORCC, right colectomy represents the procedure of choice; alternatives, such as internal bypass and loop ileostomy, are of limited value. Clinical scenarios in the case of perforation might be dramatic, especially in case of free faecal peritonitis. The importance of an appropriate balance between life-saving surgical procedures and respect of oncologic caveats must be stressed. In selected cases, a damage control approach may be required. Medical treatments including appropriate fluid resuscitation, early antibiotic treatment and management of co-existing medical conditions according to international guidelines must be delivered to all patients at presentation. The current guidelines offer an extensive overview of available evidence and a qualitative consensus regarding management of large bowel obstruction and perforation due to colorectal cancer.

Interplay Among Psychopathologic Variables, Personal Resources, Context-Related Factors, and Real-life Functioning in Individuals With Schizophrenia: A Network Analysis.

Abstract: Importance Enhanced understanding of factors associated with symptomatic and functional recovery is instrumental to designing personalized treatment plans for people with schizophrenia. To date, this is the first study using network analysis to investigate the associations among cognitive, psychopathologic, and psychosocial variables in a large sample of community-dwelling individuals with schizophrenia. Objective To assess the interplay among psychopathologic variables, cognitive dysfunctions, functional capacity, personal resources, perceived stigma, and real-life functioning in individuals with schizophrenia, using a data-driven approach. Design, Setting, and Participants This multicenter, cross-sectional study involved 26 university psychiatric clinics and/or mental health departments. A total of 921 community-dwelling individuals with a DSM-IV diagnosis of schizophrenia who were stabilized on antipsychotic treatment were recruited from those consecutively presenting to the outpatient units of the sites between March 1, 2012, and September 30, 2013. Statistical analysis was conducted between July 1 and September 30, 2017. Main Outcomes and Measures Measures covered psychopathologic variables, neurocognition, social cognition, functional capacity, real-life functioning, resilience, perceived stigma, incentives, and service engagement. Results Of 740 patients (221 women and 519 men; mean [SD] age, 40.0 [10.9] years) with complete data on the 27 study measures, 163 (22.0%) were remitted (with a score of mild or better on 8 core symptoms). The network analysis showed that functional capacity and everyday life skills were the most central and highly interconnected nodes in the network. Psychopathologic variables split in 2 domains, with positive symptoms being one of the most peripheral and least connected nodes. Functional capacity bridged cognition with everyday life skills; the everyday life skills node was connected to disorganization and expressive deficits. Interpersonal relationships and work skills were connected to avolition; the interpersonal relationships node was also linked to social competence, and the work skills node was linked to social incentives and engagement with mental health services. A case-dropping bootstrap procedure showed centrality indices correlations of 0.75 or greater between the original and randomly defined samples up to 481 of 740 case-dropping (65.0%). No difference in the network structure was found between men and women. Conclusions and Relevance The high centrality of functional capacity and everyday life skills in the network suggests that improving the ability to perform tasks relevant to everyday life is critical for any therapeutic intervention in schizophrenia. The pattern of network node connections supports the implementation of personalized interventions.

Human Endogenous Retroviruses Are Ancient Acquired Elements Still Shaping Innate Immune Responses

Abstract: About 8% of our genome is composed of sequences with viral origin, namely human Endogenous Retroviruses (HERVs). HERVs are relics of ancient infections that affected the primates' germ line along the last 100 million of years, and became stable elements at the interface between self and foreign DNA. Intriguingly, HERV co-evolution with the host led to the domestication of activities previously devoted to the retrovirus life cycle, providing novel cellular functions. For example, selected HERV envelope proteins have been coopted for pregnancy-related purposes, and proviral Long Terminal Repeats participate in the transcriptional regulation of various cellular genes. Given the HERV persistence in the host genome and its basal expression in most healthy tissues, it is reasonable that human defenses should prevent HERV-mediated immune activation. Despite this, HERVs and their products (including RNA, cytosolic DNA, and proteins) are still able to modulate and be influenced by the host immune system, fascinatingly suggesting a central role in the evolution and functioning of the human innate immunity. Indeed, HERV sequences had been major contributors in shaping and expanding the interferon network, dispersing inducible genes that have been occasionally domesticated in various mammalian lineages. Also the HERV integration within or near to genes encoding for critical immune factors has been shown to influence their activity, or to be responsible for their polymorphic variation in the human population, such as in the case of an HERV-K(HML10) provirus in the major histocompatibility complex region. In addition, HERV expressed products have been shown to modulate innate immunity effectors, being therefore often related on the one side to inflammatory and autoimmune disorders, while on the other side to the control of excessive immune activation through their immunosuppressive properties. Finally, HERVs have been proposed to establish a protective effect against exogenous infections. The present review summarizes the involvement of HERVs and their products in innate immune responses, describing how their intricate interplay with the first line of human defenses can actively contribute either to the host protection or to his damage, implying a subtle balance between the persistence of HERV expression and the maintenance of a basal immune alert.

Discovery of methylated circulating DNA biomarkers for comprehensive non-invasive monitoring of treatment response in metastatic colorectal cancer

Abstract: Objective Mutations in cell-free circulating DNA (cfDNA) have been studied for tracking disease relapse in colorectal cancer (CRC). This approach requires personalised assay design due to the lack of universally mutated genes. In contrast, early methylation alterations are restricted to defined genomic loci allowing comprehensive assay design for population studies. Our objective was to identify cancer-specific methylated biomarkers which could be measured longitudinally in cfDNA (liquid biopsy) to monitor therapeutic outcome in patients with metastatic CRC (mCRC). Design Genome-wide methylation microarrays of CRC cell lines (n=149) identified five cancer-specific methylated loci ( EYA4 , GRIA4 , ITGA4 , MAP3K14-AS1, MSC ). Digital PCR assays were employed to measure methylation of these genes in tumour tissue DNA (n=82) and cfDNA from patients with mCRC (n=182). Plasma longitudinal assessment was performed in a patient subset treated with chemotherapy or targeted therapy. Results Methylation in at least one marker was detected in all tumour tissue samples and in 156 mCRC patient cfDNA samples (85.7%). Plasma marker prevalence was 71.4% for EYA4 , 68.5% for GRIA4 , 69.7% for ITGA4 , 69.1% for MAP3K14-AS1% and 65.1% for MSC . Dynamics of methylation markers was not affected by treatment type and correlated with objective tumour response and progression-free survival. Conclusion This five-gene methylation panel can be used to circumvent the absence of patient-specific mutations for monitoring tumour burden dynamics in liquid biopsy under different therapeutic regimens. This method might be proposed for assessing pharmacodynamics in clinical trials or when conventional imaging has limitations.

Current Knowledge on Endocrine Disrupting Chemicals (EDCs) from Animal Biology to Humans, from Pregnancy to Adulthood: Highlights from a National Italian Meeting

Abstract: Wildlife has often presented and suggested the effects of endocrine disrupting chemicals (EDCs). Animal studies have given us an important opportunity to understand the mechanisms of action of many chemicals on the endocrine system and on neurodevelopment and behaviour, and to evaluate the effects of doses, time and duration of exposure. Although results are sometimes conflicting because of confounding factors, epidemiological studies in humans suggest effects of EDCs on prenatal growth, thyroid function, glucose metabolism and obesity, puberty, fertility, and on carcinogenesis mainly through epigenetic mechanisms. This manuscript reviews the reports of a multidisciplinary national meeting on this topic.

Smart contracts vulnerabilities: a call for blockchain software engineering?

Abstract: Smart Contracts have gained tremendous popularity in the past few years, to the point that billions of US Dollars are currently exchanged every day through such technology. However, since the release of the Frontier network of Ethereum in 2015, there have been many cases in which the execution of Smart Contracts managing Ether coins has led to problems or conflicts. Compared to traditional Software Engineering, a discipline of Smart Contract and Blockchain programming, with standardized best practices that can help solve the mentioned problems and conflicts, is not yet sufficiently developed. Furthermore, Smart Contracts rely on a non-standard software life-cycle, according to which, for instance, delivered applications can hardly be updated or bugs resolved by releasing a new version of the software. In this paper we advocate the need for a discipline of Blockchain Software Engineering, addressing the issues posed by smart contract programming and other applications running on blockchains.We analyse a case of study where a bug discovered in a Smart Contract library, and perhaps "unsafe" programming, allowed an attack on Parity, a wallet application, causing the freezing of about 500K Ethers (about 150M USD, in November 2017). In this study we analyze the source code of Parity and the library, and discuss how recognised best practices could mitigate, if adopted and adapted, such detrimental software misbehavior. We also reflect on the specificity of Smart Contract software development, which makes some of the existing approaches insufficient, and call for the definition of a specific Blockchain Software Engineering.

Why Do Adversarial Attacks Transfer? Explaining Transferability of Evasion and Poisoning Attacks

Abstract: Transferability captures the ability of an attack against a machine-learning model to be effective against a different, potentially unknown, model. Empirical evidence for transferability has been shown in previous work, but the underlying reasons why an attack transfers or not are not yet well understood. In this paper, we present a comprehensive analysis aimed to investigate the transferability of both test-time evasion and training-time poisoning attacks. We provide a unifying optimization framework for evasion and poisoning attacks, and a formal definition of transferability of such attacks. We highlight two main factors contributing to attack transferability: the intrinsic adversarial vulnerability of the target model, and the complexity of the surrogate model used to optimize the attack. Based on these insights, we define three metrics that impact an attack's transferability. Interestingly, our results derived from theoretical analysis hold for both evasion and poisoning attacks, and are confirmed experimentally using a wide range of linear and non-linear classifiers and datasets.

Cerebral Small Vessel Disease: A Review Focusing on Pathophysiology, Biomarkers, and Machine Learning Strategies

Abstract: Cerebral small vessel disease (cSVD) has a crucial role in lacunar stroke and brain hemorrhages and is a leading cause of cognitive decline and functional loss in elderly patients. Based on underlying pathophysiology, cSVD can be subdivided into amyloidal and non-amyloidal subtypes. Genetic factors of cSVD play a pivotal role in terms of unraveling molecular mechanism. An important pathophysiological mechanism of cSVD is blood-brain barrier leakage and endothelium dysfunction which gives a clue in identification of the disease through circulating biological markers. Detection of cSVD is routinely carried out by key neuroimaging markers including white matter hyperintensities, lacunes, small subcortical infarcts, perivascular spaces, cerebral microbleeds, and brain atrophy. Application of neural networking, machine learning and deep learning in image processing have increased significantly for correct severity of cSVD. A linkage between cSVD and other neurological disorder, such as Alzheimer's and Parkinson's disease and non-cerebral disease, has also been investigated recently. This review draws a broad picture of cSVD, aiming to inculcate new insights into its pathogenesis and biomarkers. It also focuses on the role of deep machine strategies and other dimensions of cSVD by linking it with several cerebral and non-cerebral diseases as well as recent advances in the field to achieve sensitive detection, effective prevention and disease management.

Are the "Best Buys" for Alcohol Control Still Valid? An Update on the Comparative Cost-Effectiveness of Alcohol Control Strategies at the Global Level.

Abstract: Objective:Evidence on the comparative cost-effectiveness of alcohol control strategies is a relevant input into public policy and resource allocation. At the global level, this evidence has been us...

HERV Envelope Proteins: Physiological Role and Pathogenic Potential in Cancer and Autoimmunity.

Abstract: Human endogenous retroviruses (HERVs) are relics of ancient infections accounting for about the 8% of our genome. Despite their persistence in human DNA led to the accumulation of mutations, HERVs are still contributing to the human transcriptome, and a growing number of findings suggests that their expression products may have a role in various diseases. Among HERV products, the envelope proteins (Env) are currently highly investigated for their pathogenic properties, which could likely be participating to several disorders with complex etiology, particularly in the contexts of autoimmunity and cancer. In fact, HERV Env proteins have been shown, on the one side, to trigger both innate and adaptive immunity, prompting inflammatory, cytotoxic and apoptotic reactions; and, on the other side, to prevent the immune response activation, presenting immunosuppressive properties and acting as immune downregulators. In addition, HERV Env proteins have been shown to induce abnormal cell-cell fusion, possibly contributing to tumor development and metastasizing processes. Remarkably, even highly defective HERV env genes and alternative env splicing variants can provide further mechanisms of pathogenesis. A well-known example is the HERV-K(HML2) env gene that, depending on the presence or the absence of a 292-bp deletion, can originate two proteins of different length (Np9 and Rec) proposed to have oncogenic properties. The understanding of their involvement in complex pathological disorders made HERV Env proteins potential targets for therapeutic interventions. Of note, a monoclonal antibody directed against a HERV-W Env is currently under clinical trial as therapeutic approach for multiple sclerosis, representing the first HERV-based treatment. The present review will focus on the current knowledge of the HERV Env expression, summarizing its role in human physiology and its possible pathogenic effects in various cancer and autoimmune disorders. It moreover analyzes HERV Env possible exploitation for the development of innovative therapeutic strategies.

Transvaginal ultrasound vs magnetic resonance imaging for diagnosing deep infiltrating endometriosis: systematic review and meta-analysis.

Abstract: Objectives To perform a systematic review of studies comparing the diagnostic accuracy of TVS and MRI in Deep Infiltrating Endometriosis (DIE) including only studies in which patients have been underwent both techniques. Methods An extensive search of papers comparing TVS and MRI in DIE was performed in Medline (Pubmed) and Web of Sciences from January 1989 to January 2016. Studies were considered eligible if they reported on the use of TVS and MRI in the same set of patients for the preoperative detection of endometriosis in pelvic locations in women with clinical suspicion of DIE using the surgical data as a reference standard. Quality was assessed using QUADAS-2 tool. A random-effects model was used to determine overall pooled sensitivity, specificity, positive likelihood ratio (LR+) and negative likelihood ratio (LR–) and the diagnostic odds ratio (DOR). Results Of the 375 citations identified, 6 studies (n=424) were considered eligible. Pooled sensitivity, specificity, LR+ and LR– of MRI in detecting DIE in the recto-sigmoid for MRI were 0.85 (95% CI, 0.78–0.90), 0.95 (95% CI, 0.83–0.99), 18.4 (95% CI, 4.7–72.4) and 0.16 (95% CI, 0.11–0.24), respectively. Pooled sensitivity, specificity, LR+ and LR– of TVS in detecting DIE in the recto-sigmoid for TVS were 0.85 (95% CI, 0.68–0.94), 0.96 (95% CI, 0.85–0.99), 20.4 (95% CI, 4.7–88.5) and 0.16 (95% CI, 0.07–0.38), respectively. DOR was 116 (95% CI, 23-585) and 127 (95% CI, 14 - 1126), respectively. Pooled sensitivity, specificity, LR+ and LR– of MRI in detecting DIE in the rectovaginal septum for MRI were 0.66 (95% CI, 0.51–0.79), 0.97 (95% CI, 0.89–0.99), 22.5 (95% CI, 6.7–76.2) and 0.38 (95% CI, 0.23–0.52), respectively. Pooled sensitivity, specificity, LR+ and LR– of TVS in detecting DIE in the rectovaginal septum for TVS were 0.59 (95% CI, 0.26–0.86), 0.97 (95% CI, 0.94–0.99), 23.5 (95% CI, 9.1–60.5) and 0.42 (95% CI, 0.18–0.97), respectively. DOR was 65 (95% CI, 21- 204) and 56 (95% CI, 11 - 275), respectively. Pooled sensitivity, specificity, LR+ and LR– of MRI in detecting DIE in the uterosacral ligaments for MRI were 0.70 (95% CI, 0.55–0.82), 0.93 (95% CI, 0.87–0.97), 10.4 (95% CI, 5.1–21.2) and 0.32 (95% CI, 0.20–0.51), respectively. Pooled sensitivity, specificity, LR+ and LR– of TVS in detecting DIE in the uterosacral ligaments for TVS were 0.67 (95% CI, 0.55–0.77), 0.86 (95% CI, 0.73–0.93), 4.8 (95% CI, 2.6–9.0) and 0.38 (95% CI, 0.29–0.50), respectively. DOR was 32 (95% CI, 12- 85) and 12 (95% CI, 7- 24), respectively. Wide confidence intervals of pooled sensitivities, specificities and DOR were present for both techniques in all the considered locations. Heterogeneity was moderate or high for sensitivity and specificity for TVS and MRI in most locations assessed. According to QUADAS2, the quality of the studies was considered good for most domains of the included studies. Conclusions Overall diagnostic performance of TVS and MRI for detecting DIE involving recto-sigmoid, uterosacral ligaments and rectovaginal septum is similar.

Ten years of sorafenib in hepatocellular carcinoma: Are there any predictive and/or prognostic markers?

Abstract: Sorafenib has been considered the standard of care for patients with advanced unresectable hepatocellular carcinoma (HCC) since 2007 and numerous studies have investigated the role of markers involved in the angiogenesis process at both the expression and genetic level and clinical aspect. What results have ten years of research produced? Several clinical and biological markers are associated with prognosis. The most interesting clinical parameters are adverse events, Barcelona Clinic Liver Cancer stage, and macroscopic vascular invasion, while several single nucleotide polymorphisms and plasma angiopoietin-2 levels represent the most promising biological biomarkers. A recent pooled analysis of two phase III randomized trials showed that the neutrophil-to-lymphocyte ratio, etiology and extra-hepatic spread are predictive factors of response to sorafenib, but did not identify any predictive biological markers. After 10 years of research into sorafenib there are still no validated prognostic or predictive factors of response to the drug in HCC. The aim of the present review was to summarize 10 years of research into sorafenib, looking in particular at the potential of associated clinical and biological markers to predict its efficacy in patients with advanced HCC.