2 N E W S & R E S E A RC H Largest UK nitrogen dioxide levels: London and Midlands Air pollution linked to higher Covid-19 mortality Scientists have unearthed a possible link between the severity of Covid-19 and air quality. The preliminary study - looking at whether areas with higher levels of air pollutants in England are associated with a larg- er number of cases/deaths from Covid-19 – was conducted by a team from the University of Cambridge. Report: Mark Nicholls Aware of the effects that air pollut- ants have on human health – and that initial symptoms of Covid-19 include fever, with some patients developing serious respiratory prob- lems – for their study, the scientific team used publicly available data from multiple air monitoring sta- tions across the country where a minimum of 2,000 SARS-CoV-2 infec- tions and 200 deaths were reported in the period from February to early April 2020. They analysed the data on total Covid-19 cases and deaths, against the levels of three major air pollut- ants, collected between 2018 and 2019, when no Covid-19 case was reported. People exposed to higher levels of certain pollutants were found to be at higher risk of Covid-19 infec- tion and/or death when taking into account differences in population density from region to region. ‘We used different approaches to reach these conclusions,’ researcher Marco Travaglio explained. ‘After gathering air pollution data from the European Environmental Agency, we then compared it to lab-con- firmed numbers of cases and deaths from Covid-19 as reported by gov- ernment official sources. ‘We used multiple statistical mod- els to account for the effects of demographics and found that air pollution is strongly associated with Covid-19 mortality, whether you live in a highly populated area or not.’ The largest number of Covid-19 deaths in England occurred across London and the Midlands, reflect- ing the geographical distribution of Covid-19 related cases. Previous studies have shown that the annual average of nitrogen dioxide con- centrations are largest in these two regions. ‘Our results provide the first evi- dence that SARS-CoV-2 case fatality is associated with increased nitro- gen oxide and nitrogen dioxide levels in England.’ London, the Midlands and the North West show i o d u t S d o o G / k c o t s r e t t u h S © the largest concentration of these air pollutants, with Southern regions displaying the lowest levels in the country, and the number of Covid- 19 deaths follows a similar trend,’ explained Travaglio, a PhD student at the Medical Research Counxil (MRC) Toxicology Unit, Cambridge. While the implications could be far-reaching, the researchers empha- sise that their findings are only preliminary and only show a cor- relation, and that further research is needed to confirm that air pollution makes Covid-19 worse. ‘Data is compelling and certainly flags up several important issues,’ he pointed out, ‘but more work needs to be done before our find- ings can turn into concrete policies that can aid against the fight against Covid-19. ‘The major implication for the time being is that our study raised interesting hypotheses regard- ing the effect of air pollution on the current epidemic and adds yet another piece to the puzzle and shows that air pollution could be linked to human health in ways that no-one imagined before.’ Supported by the Medical Research Council, the research has been car- ried out by the lab of Dr Miguel Martins at the MRC Toxicology Unit in Cambridge, with Travaglio shar- ing first authorship of the manu- script with Yizhou Yu. It also includes evidence from Northern Italy and the USA that high levels of air pollution are linked to deadlier cases of Covid-19. This comes at the height of the UK coronavirus lockdown with significant falls in car and public transport use. In addition, Department for From clinician to hospital medical director ‘Lead from the front’ Denton also had several national roles before taking up her pre- sent NNUH position in 2018. ‘One thing that struck me about being a leader is that, coming through the imaging world where there is Making the transition from clinician to a senior hospital management role can prove challenging. Professor Erika Denton did it – whilst also retaining some clinical responsibilities. A radiology background, Denton believes, is a major asset in mak- ing the move into high-level man- agement. Currently the Medical Director of the Norfolk and Norwich University Hospital (NNUH) in the east of England, she remains a consultant radiologist (with a spe- cific interest in breast imaging), and Professor of Radiology at the University of East Anglia. Denton describes her transition from clinical to management – with a few pointers from her career that prepared her for the role. She had found herself in the role with very little notice after the publication of a report by UK regulator the Care Quality Commission (CQC) into the hospital’s activities. ‘Such a role had been in my mind as being a good thing to do, but I had not anticipated it at such short notice,’ she explained. With a state school background in the United Kingdom, Denton recalled her days at medical school, which she did not particu- larly enjoy, but she loved being a junior doctor, even when on medi- cal rotation at Medway Maritime Hospital in Kent. ‘It was all a good grounding and I learned to not fear talking to anybody.’ from London Training scheme and radiol- ogy academy Moving to rural Norfolk in 1999, she became direc- tor of breast screening at the Norfolk and Norwich Hospital (which would later become the NNUH) and had other management roles at the new 1,200-bed University Hospital which opened in 2001 and employs over more than 8,000 staff to provide care for a million people. There, she campaigned for more junior doctor posts, helping to establish a train- ing scheme followed by a radiology academy. u i c n a t S n i t n a t s n o C / k c o t s r e t t u h S © Marco Travaglio is a second-year PhD student at the University of Cambridge. Prior to the Covid-19 outbreak, his pro- ject mainly involved looking at selective cell death in cell models of Parkinson’s disease but, more recently, in the light of the coronavirus crisis, he and his team have re-focused efforts towards Covid- 19 research. Energy, Food and Rural Affairs (Defra) monitoring data for nitrogen dioxide (NO2) in London, Leeds, Manchester, Cardiff, Edinburgh, Liverpool, Bristol and Newcastle, found big reductions in emissions (at the end of March 2020) com- pared with the same time last year. Despite the fall in NO2 emissions, citizens in UK cities were warned by the European Public Health Alliance (EPHA) that exposure to air pol- lution could increase their risk of dying from Covid-19. In 2019, ten hospitals in the most polluted areas of London were equipped with air quality monitors to measure levels of toxic air and help protect patients and staff after a study found 60% of hospitals and NHS facilities in inner London were located in areas that exceed air qual- ity pollutant levels. constant change, radiology sets you up really well for leadership and the agile world of medical manage- ment.’ However, she warned of the need to recognise the importance of a multi-disciplinary approach. ‘Remember,’ Denton advised her audience, ‘you are only one small EUROPEAN HOSPITAL Vol 29 Issue 3/20

4 CA R D I O LO G Y Device to benefit more cardiac patients The pioneering mini-pacemaker relevant. Garweg, who also leads his institution’s leadless pacing clinical and research program, said: ‘The first generation of the device that allows VVI pacing has similar fea- tures to a “conventional” ventricular pacemaker including accelerometer driven rate responsive pacing, auto- mated pacing capture management. But it only measured the heart activ- ity in one ventricle, so patients who needed to have the heart activity in the atrium measured as well, did not qualify. Belgium - A pioneering new genera- tion of wireless mini-pacemaker is set to benefit many more patients than before. Following a successful first-in- Europe implantation procedure at the University Hospitals of Leuven, the advance has been described as ‘the beginning of a new para- digm of cardiac pacing’ with the development of the next-genera- tion mini-pacemaker regarded as a major step forward in this field. Cardiologist and electrophysiologist Dr Christophe Garweg, from the cardiovascular department at the University Hospitals of Leuven, out- lined the work involved since 2015 in designing and implanting the first The second-generation pacing device ‘With this new type of pacemaker, we can also treat patients with a complete interruption of the heart activity between the ventricle and the atrium.’ The device utilises the new MARVEL (Micra Atrial TRacking Using a Ventricular AccELerometer) custom software developed to detect atrial contraction using the 3-axis accelerometer (AAC) of the Micra, enabling AV atrio-ventricular (AV) synchronous pacing. Manufactured by Medtronic, both devices are currently the only lead- less pacing devices available for clinical use and weigh a mere 2g, with a length of 25.9 mm, a diam- eter of 6.7 mm, and 8-13-year bat- tery life. The distal end incorporates the electrode and four retractable nitinol tines to ensure fixation in the myocardium. It has also been associated with a 63% reduction of complications, compared to histori- cal groups of patients with a con- ventional pacemaker. Continuing clinical trials While now commercially released, the Leuven department continues developed to allow atrio-ventricular synchronous pacing (VDD pacing mode). 40% of patients can use it With the new evolution of the Micra pacemaker able to measure cardiac activity in the atrium and, as such, coordinate the electrical activity between atrium and ventricle, it wid- ens the use of the mini-pacemaker to 40% of patients, up from the 16% the first generation for whom it was Thorax with implanted Micra AV generation of the Micra VR device, which allows asynchronous ven- tricular pacing (VVI pacing mode). However, until now, its use has been limited to atrial fibrillation (AF) patients, or those with contra- indication for double chamber pace- maker, which is approximately 16% of current pacing indications. Now, second-generation model, the Micra AV, has been the Entering a new age of artificial intelligence AI predicts blood flow to the heart A strong and independent predictor ‘In patients with known or sus- pected coronary artery disease,’ the study concluded, ‘reduced MBF and MPR measured automatically inline using AI quantification of cardio- vascular magnetic resonance perfu- sion mapping provides a strong, independent predictor of adverse cardiovascular outcomes.’ Professor James Moon, who led the multi-centre research team, explained that measuring blood flow to the heart has been used for many years, but non-invasive blood flow assessments, including CMR (Cardiac Magnetic Resonance) imag- ing, remain difficult to analyse in a manner precise enough to deliver a prognosis or recommend treatment. During the study, the UCL/Barts/ NIH team made CMR ‘fully quantita- tive’ in measuring the actual amount of blood flow going to heart muscle, using AI to identify where the mus- cle is and making the process of working out where the blood pool is easier and quicker. From routine CMR scans from more than 1,000 patients attending St Bartholomew’s Hospital and the Royal Free Hospital the new auto- mated AI technique – devised by Dr Peter Kellman and Dr Hui Xue at NIH – was used to analyse the images, enabling them to precisely and instantly quantify the blood flow to the heart muscle. By comparing the AI-generated blood flow results with the health outcomes of each patient, the team found that those with reduced blood flow were more likely to have adverse health outcomes. Moon, Professor of Cardiology at UCL and the Clinical Director of Imaging at Barts Heart Centre (BHC), said that AI also could show findings the eye may not have previ- ously seen. In addition to measuring blood flow to the heart muscle to highlight abnormalities and predict adverse outcomes, AI also identified what may be happening in a wider con- text with the patient, such as ath- erosclerosis and age-related stiffen- ing of arteries. ‘By just measuring the large arteries, we are missing a chunk of the story about how our hearts are over our lifetimes, and that the microvasculature is clearly Report: Mark Nicholls UK – Artificial Intelligence (AI) has, for the first time, measured blood flow to the heart to help predict which patients may suffer myocar- dial infarction or stroke. A research team at University College London and Barts Health NHS Trust and the National Institutes for Health (NIH) in the USA – are optimistic that AI analysis of perfu- sion maps will be a reliable, con- venient and detailed new biomarker in cardiac patient care. With myocardial perfusion reflect- ing the macro- and microvascular coronary circulation, the researchers explored the prognostic significance of stress myocardial blood flow (MBF) and myocardial perfusion reserve (MPR - the ratio of stress to rest MBF) in patients with suspect- ed and known coronary artery di- sease referred clinically for perfu- sion assessment. Image analysis was conducted automatically using a novel AI approach and used to seek the asso- ciations of stress MBF and MPR with death and major adverse cardiovas- cular events, including myocardial infarction, stroke, heart failure, late revascularisation, and death. Myocardial perfusion maps for a patient with a right coronary artery stenosis. There is hypoperfusion in the basal to mid inferior wall. The perfusion maps have been contoured automatically using artificial intelligence (green) and the myocardial blood flow in each myocardial segment is provided to the operator almost instantly during the scan. Dr Christophe Garweg is a cardiolo- gist and electrophysiologist with the car- diovascular department at the University Hospitals of Leuven, where he also leads the institution’s leadless pacing clinical and research program. to use the device in the context of clinical trials to further improve the technology. It has also previously participated in the Marvel 1 and 2 prospective multicentre studies to assess the feasibility and the safety of clinical use of the Marvel one and two softwares. Safety and efficacy need confirmation Garweg stressed that the safety and efficacy of the new Micra AV still must be confirmed in larger populations with a longer follow- up before being widely used in the clinical setting. In addition to the first implant in Leuven (May 28), similar implants were carried out on the same day in Linz and Santiago. longer Garweg believes term patient-benefits of this device are the absence of transvenous lead and subcutaneous pocket reducing the complication rate at long-term follow-up. (MN) James Moon is a Professor of Cardiology at UCL and the Clinical Director of Imaging at Barts Heart Centre. His research focuses on developing mapping techniques with CMR and measuring diffuse myocardial fibrosis. Having published more than 300 papers, he is also looking at making scans faster and cheaper in improving quality as CMR expands. changing with age and diabetes and other parameters probably associ- ated with chest pain and outcome. We think those are tractable with a new generation of treatments for the microvasculature disease function,’ Moon added. As a new marker of disease, he acknowledges there is still some way to go to fully understand and realise the potential of this AI advance but also sees an opportunity to develop new therapies for patients with chest pain where the coronary arter- ies do not appear abnormal. However, the researchers believe their study has demonstrated the growing potential of AI-assisted imaging technology to improve the detection of heart disease and will have more diverse applications in cardiac care ‘It’s an indication that we are entering a new era of AI,’ said Moon. EUROPEAN HOSPITAL Vol 29 Issue 3/20

6 CA R D I O LO G Y Remote cardiac monitoring Distant experts observe adverse signs examinations – without patients hav- ing to make their way to the hospital.’ Particular advantages in times of Covid-19 The latter has become particularly relevant during the Covid-19 pan- demic. ‘Many patients are anxious and worried about the risk of con- tracting the virus during a hospital visit.’ The standard six-week follow- up examination after surgery has therefore been changed to a tele- examination. Patients are called and asked about their health and whether or not they are experiencing prob- lems with their surgical scars or any other issues ‘We used to deem this in-house examination to be very important, Rauholt said. ‘However, over the last few months we’ve had really good experiences with remote follow-up examinations. Patients no longer have to worry about a nosoco- mial Covid-19 infection, and the tele- examination also causes less time pressure and allows for time more intensively spent with a patient. Opportunities and chal- lenges Home monitoring of cardiac patients is an effective tool to considerably reduce strain on patients and hos- pital staff, Rauholt concludes. ‘The system is very reliable. However, the evaluation of the reports requires experience because there is a cer- For cardiology patients fitted with an implantable cardiac monitor, cardioverter-defibrillator (ICD) or pacemaker, home monitoring is a blessing. The system also has many advantages for medical staff, as Kristina Rauholt reports. The nurse and Certified Cardiac Device Specialist for Allied Professionals (CCDS) at the Sahlgrenska University Hospital, in Sweden, has worked with home monitoring systems for cardiology patients for many years. She explains how patients and doctors benefit – and how remote monitoring has changed during the Covid-19 pandemic. fications as it would not be possible to cope with the amount of work created,’ she added. Patients benefit from this system as they do not have to go to hospital for as many check-ups. A large number of routine follow-up examinations can be carried out via telemedicine – a significant cost- and labour- saving for the hospital. ‘The quality of remote follow-up examinations is equivalent to that of face-to-face Home monitoring systems have been used at the Sahlgrenska University Hospital since 2008. They ensure that the devices work and support patients reliably. ‘Every morning we evaluate the report which the sys- tem generates automatically when an irregularity is registered,’ Rauholt explained. Home monitoring detects changes in heart rhythm, for example ventricular tachycardia or atrial fibril- lation. Patients often do not even reg- ister these changes, because at first they may not cause any symptoms. ‘One of the biggest advantages for us is that the system also alerts us to these “silent problems”,’ the expert pointed out. Vice versa, the data can also be uti- lised when patients report symptoms that indicate heart problems. ‘It’s an important means of backing up a diagnosis,’ she said. This is also the basis for adjusting medication or for synchronising the rhythm more pre- cisely with the help of the implant. The remote solution saves time, work and cost The cardiology team evaluates the data sent from the home monitoring system via the mobile network and asks patients to come to the hospital for further investigations if necessary, the expert explained. ‘The evalua- tions also help us to check whether the implant is working correctly – something which patients who have only just undergone surgery tend to worry about. If a patient knows we are continuously monitoring the implant during office hours, this pro- vides a lot of reassurance.’ The home monitoring system rais- es the alarm if the implant causes problems. ‘There are cases where there is a risk that the electrode of an implanted defibrillator will shock the patient, when this is not neces- sary. This malfunction can become dangerous if it is not detected in time.’ GPs are only rarely involved in home monitoring; they may merely use the data generated to confirm a suspected diagnosis, Rauholt said. Accident and emergency doctors, on the other hand, frequently request these evaluations when they need to check whether the ICD has already applied an electric shock. At Sahlgrenska University Hospital around 800 patients are now con- nected to the home monitoring sys- tem. Cardiologists are only informed about irregularities. ‘If all is well with the patient, we don’t need any noti- The cardiovascular impact of Covid-19 Plus: The multi-organ involvement The increased Covid-19 risk to car- diac patients was discussed dur- ing an online presentation at ECR 2020. Focusing on the prevalence of pulmonary embolism in Covid- 19 patients, Dr Karl-Friedrich Kreitner, Professor of Radiology at the Department of Diagnostic and Interventional Radiology at the Johannes-Gutenberg University in Mainz, Germany, discussed hypo- theses which can explain cardiac involvement in the course of the dis- ease, and help delegates understand imaging findings of cardiovascular complications. The ‘Non-pulmonary aspects of Covid-19’ session, which outlined pathology research on Covid-19, described the many clinical mani- festations of the infection and intro- duced the concept of multi-organ involvement. Alongside Dr Kreitner’s presentation on cardiac issues, other contributors will spoke of neuro- logic manifestations, and the disease in paediatric patients. The virus affects other organs When we met, Dr Kreitner stressed that although the virus uses the ACE-2 receptor predominantly loca- ted in lung tissue to enter a human, this receptor is also present else- where in the body. ‘It’s also in up to 10% of myocardial cells and in endothelial cells of arteries and veins,’ he explained. ‘If you are aware of this then you come to the conclusion that the infection is not exclusively localised in the lungs, but can affect other organs as well, including the heart and pulmonary vasculature.’ Among in-patients who die of the disease, Kreitner said there is evidence that not all died due to lung tissue destruction, but there were also other causes, notably pulmonary. Lab data can help clinicians to assess the risk of Covid-19 patients sustaining thromboembolic disease, he said, in particular checking the D-dimer or fibrin degradation prod- ucts (FDP) as markedly-elevated D-Dimer and FDP are common in non-survivors of Covid-19. Recommendations suggest that patients with a D-dimer level below one have a lower risk of throm- boembolism than those presenting with a level over one. ‘In these newer recommendations there is some risk stratification; that with a D-dimer between one and two you should think of controlling this value and do pulmonary embolism imaging in the further course of the disease. If presenting with values above two, you should immediately test to see whether the patient suf- fers from pulmonary embolism or not. ‘Generally, if values are below one, you can treat these patients with a prophylactic dose of low molecular weight heparin and where the value is above two, you should consider a therapeutic dose of low molecular weight heparin.’ Weighing up risks The challenge, he pointed out, is to weigh up lowering the risk of thromboembolism against the increased risk of haemorrhage, and it is very important to look at these values when the patient is admitted to hospital. ‘With cardiac involvement, Covid- 19 patients suffering a cardiovascu- lar disease have a much higher risk of dying from the condition. This is perhaps because their cardiovascu- lar system is not fit enough to face the challenges of this SARS-Cov-2 infection.’ In such cases, clinicians should turn to cardiac biomarkers, espe- cially high-sensitivity troponin and the NT-pro BNP because, when they are elevated, the individual is at a greater risk. While much of the evidence of this has come from China, Italy and the USA, Dr Kreitner discussed two patients from his department who presented with different forms of Covid-19 myocarditis. ‘One pre- sented with a typical viral myocar- ditis and the second with dilated cardiomyopathy-like appearance with signs of inflammation, so the phenotypes of cardiac involvement have a great diversity.’ EUROPEAN HOSPITAL Vol 29 Issue 3/20

8 R A D I O LO G Y Radiomics strengthens breast imaging The field of AI-enhanced imaging provides radiologists with an unprecedented opportunity to shape patient care, a leading Austrian radiologist explained at ECR 2020. Choose DL over ML algorithms Report: Mélisande Rouger Workflow with radiomics starts with image acquisition and segmenta- tion. When the region of interest is defined, radiomics analysis enables extraction of a large quantity of imaging features, and then to select, reduce, classify and model the data to answer a specific question. ‘Machine learning or deep learn- ing may be used at any step of the process and help in various aspects of daily practice,’ said Katja Pinker-Domenic, a breast radiolo- gist at the Memorial Sloan Kettering Cancer Centre, USA, and the Medical University of Vienna in Austria. Current applications When it comes to mammography, the most common use case for machine learning (ML) or deep learning (DL) is screening accuracy. Studies have shown that deep convolution- al neural networks (CNN) achieve area under curves (AUC) of about 0.9, regardless of the framework. Researchers now want to take it one step further and are looking at performance of radiologists on mam- mograms with or without the help of artificial intelligence. AI support has been shown to increase sensitivity and there is a trend towards increas- ing specificity. ‘When the radiologist is provided with a decision support system, performance improves, not reading time,’ Pinker-Domenic said. With ultrasound, the main applica- tion has been classification, to dif- ferentiate between benign and malig- nant masses. Ultrasound is recom- mended when a suspicious mass has already been spotted. However, to date, there are no guidelines that rec- ommend using AI with this modal- ity in clinical practice. ‘More studies are needed to explore advanced DL methods and prove their usefulness,’ Pinker-Domenic underlined. ‘It’s an area of interest.’ A study by Barinov et al ( Journal of Digital Imaging, June 2019) has shown that an FDA approved AI-decision support sys- tem incorporated in image analysis improved accuracy in breast cancer diagnosis for all radiologists, regard- less of their level of experience. AI-enhanced MRI has by far been the most explored area, particularly in lesion detection. ‘Research has shown that fully automated detection of breast cancer using CNN is pos- sible and this is very encouraging, in particular with the trend of abbrevi- ated MR protocols in breast cancer screening,’ she added. With CNN, the next step is clas- sification. Authors of another study applied AI based breast lesion detec- tion classification imaged to mul- tiparametric breast MR protocol with ultrafast DCE-MRI, Tw2 and DWI. The researchers could show that the final AI system that incorpo- rates information from all sequences showed significantly higher perfor- mance than DEC-MRI alone. When operating at the same sensitivity level as radiologists it was compared to, AI-enhanced MRI improved sensi- tivity and detected less false positives than humans, potentially sparing biopsy. However, in another study, radiologists did perform better than Axial first post-contrast bilateral dynamic MR images of two patients with benign-appearing small breast masses (A, B; blue arrows) in which biopsy yielded invasive ductal carcinoma and of two patients (red circles) categorised as BI-RADS 4 in which biopsy results yielded (c) pseudoangiomatous stromal hyperplasia and (d) fibroadenoma ML and DL, which performed better than ML (AUC: 0.88 vs. 0.81) for lesion classification. Radiologists should thus strive to use DL rather than ML algorithms, especially for compli- cated cases such as BRCA mutation carriers, Pinker-Domenic explained. ‘We should identify cases in which we struggle, for example small enhancing masses in BRCA muta- tion carriers. These patients often undergo high-risk MRI screening and present with small tiny lesions that probably have a benign appearance but can be cancer. Our awareness level in this patient collective is already altered compared with an average risk of breast cancer,’ she said. According to a recent study, the BI-RADS descriptors are not very useful to classify lesions, with a 53% diagnostic accuracy, 73% sensitivity and 42% specificity. By contrast, radiomics diagnos- tic accuracy is almost 80%, and sensitivity and specificity are also higher (Figure 1). ‘This shows poten- tial for AI to be used as an adjunct tool, to spare unnecessary biopsies for benign appearing small breast masses. This can be very useful in these patients, who undergo repeat- ed screening and biopsies,’ she said. Prediction and prognosis at hand With AI enhanced MRI, radiologists can also try themselves at prediction and prognosis. Molecular subtyp- ing with DCE-MRI has been studied extensively. Specific molecular sub- types seem to carry radiomics signa- tures that can be used to accurately classify those lesions with respect to receptors – the molecular subtype – with good diagnostic accuracy. These signatures have the potential to pro- vide prognostic indicators that are derived from the entire tumour and that may be used to monitor tumour biology changes during treatment. This function can be extended to contrast-enhanced mammography, which has the same underlying psy- chological mechanisms as MRI and can be used as an alternative when MRI is not available. A new standalone solution beats mammography Abbreviated MRI protocol is superior Is mammography still the best method for breast cancer screening? For a number of breast cancers, the latest scientific findings suggest otherwise. For more than a decade, Professor Christiane Kuhl MD, Director of the Clinic for Diagnostic and Interventional Radiology at the University Hospital RWTH Aachen, has researched the use of MRI in breast cancer screening. During a Bayer Healthcare webinar at the RÖKO Digital Congress the radiologist explained why she believes abbreviated MRI protocols to be a game changer which could, in fact, completely replace mammography for screening. ubiquitous, benign changes, such as cysts or fibroadenomas. Report: Karoline Laarmann Despite its overall success, mammo- graphy screening also continues to attract criticism. For one, it detects breast cancers which are prognosti- cally irrelevant, resulting in overdi- agnosis. Mammography also does not detect all breast cancers that are of prognostic relevance, thus underdiagnosing cases. “20-30% of breast cancers detected in women who participate in screening are not discovered during mammography screening but during the intervals,’ Professor Kuhl points out. ‘These interval cancers are basically noth- ing other than cases of failed early detection of fast-growing, and there- fore prognostically relevant, breast cancers.’ The reason for underdiagnosis not only lies in the density of the glan- dular breast tissue but also in the tumour biology. As the detection of breast cancer in the mammogra- phy- or digital breast tomosynthesis image is based on the visualisation of architectural distortion, microcalcifi- cation or spiculated mass, it is par- ticularly suitable to find luminal, i.e. less aggressive breast cancers, whilst biologically aggressive tumour types can appear in X-ray based images as Improving prognosis with MRI ‘We need a method for early detec- tion which offers a sensitivity profile that meets the oncological demand and finds those cancers that are prog- nostically relevant,’ Kuhl demands. In 2010 the professor of radiology already proved that MRI meets these criteria with the results of her mul- ticentric EVA-Study. This compared the efficiency of different imaging procedures for breast cancer screen- ing in women with increased risk in the family. The study showed that MRI had the highest hit rate for tumours detected, Prof Dr Christiane Kuhl has directed the Clinic for Diagnostic and Interventional Radiology at the University Hospital Aachen since 2010. Kuhl studied medicine in Bonn, the city of her birth (1966) and, in 2004, accepted a C3 professorship at the Department for Oncological Diagnostics and Interventional Radiology at the University Hospital of Bonn. Today, she is considered to be among the most renowned German breast cancer researchers. Her work has received numerous awards nationally and internationally. From her experience, Kuhl advocates the use of MRI for early breast cancer detection. women with a high-risk profile, is mainly due to high cost and lack of availability. An abbreviated version of MRI, focusing on the essentials, can now provide relief. Kuhl and her team have developed a protocol which only consists of subtracted images acquired before and after the administration of contrast medium (FAST), which are fused into a pro- jection image with maximum inten- sity (MIP). The results, first published in a study in 2014, astounded even Kuhl herself. and the interval cancer rate was zero. Kuhl presents a simple explanation: ‘When we use MRI for breast cancer screening and use contrast medium, we do nothing other than visualis- ing the angiogenesis and protease activity – tissue changes that directly correlate with carcinogenesis, cell proliferation and metastatic growth. This basically means that MRI is at the other end of the sensitivity spec- trum compared to mammography. Or, put differently: the more biologi- cally aggressive the breast cancer the more reliably it is detected with MRI.’ Fast and effective: Abbreviated MRI protocols The reason that MRI is still rarely used for early screening, i.e. only for The diagnostic precision of the abbreviated MRI protocol was equiv- alent to that of the complete MRI protocol and was no less sensi- tive and specific. Furthermore, not only was the examination time short (three minutes) but so was the average time an experienced radi- ologist needed to determine whether breast cancer was present or not (2.8 seconds), and characterising an enhancement in an image also only took 28 seconds. More than just an add-on – it works Since then, many further internation- al studies on the subject of abbrevi- ated MRI protocols versus complete MRI protocol have been carried out, EUROPEAN HOSPITAL Vol 29 Issue 3/20

1 0 R A D I O LO G Y Algorithms must meet quality criteria Deep Learning in breast cancer detection low performance, especially new BI-Rads lexicon because these soft- wares were based on segmentation. Recently two DL softwares were optimised and demonstrated very nice agreement, reaching 99% for MLO and 96% for CC view to dis- tinguish between fatty and dense breast. Another publication showed very good kappa agreement with radiologists (0.85) A French expert in breast imaging looked at the latest DL applica- tions in her field, screening their strengths and weaknesses in improving breast cancer detection. It is really important to understand which types of data sets need to be checked when evaluating an AI model for image interpretation, according to Isabelle Thomassin- Naggara, Professor of Radiology at Sorbonne University in Paris. ‘The model is usually built on enriched data cohort with a high volume of BCs, ideally including all types of BC on mammography, speculated mass, run mass and clus- ter classification. Another data set must be used for internal validation,’ she said. The model also needs to be exter- nally validated in another cohort with lower prevalence of cancer and possibly with more subtler cancers to improve its accuracy. Last but not least, the model must be evaluated with an independent data set with a prevalence that is representative of the screened population. ‘Today, an algorithm needs more than one million mammographiesto truly demonstrate its efficiency. The development of dedicated platforms must be elaborated to build larger evaluation of data sets that are more representative of the population. All these steps can be performed retro- spectively, but, before clinically vali- dating a model, a randomised trial comparing the accuracy of a model with the radiologists’ is needed,’ she said. Incremental improvement in the ‘area under curve’ (AUC) is not directly translatable to improve patient outcomes in a clinical setting and it is uncertain which examined population a commercially AI sys- tem would flag as having more than 2% of malignancy requiring addi- tional diagnostic workup in current clinical practice. A costly mistake ‘We learned from our experience of CAD in mammography that a prod- uct that promises new technologies too quickly can be a costly mistake later, by leading to find more false positives without improving cancer detection,’ she said. A number of studies have compared human vs. AI performance. A recent publication in the Lancet compared 25 studies and highlighted meth- odological limitations of validation studies of AI models, including the absence of calculation of number of subjects, absence of prospective design, presence of many statistic biases, absence of validation on independent samples and absence of comparison on the same pop- ulation of AI model and human reading. (A comparison of DL per- formance against HC professionals in detecting diseases from medical imaging: a systematic review and meta-analysis, the Lancet, October 2019) The Google paper on how its DL algorithm beat radiologists at detect- ing cancer made the news early this year and frightened a lot of radiolo- gists. But this paper showed a num- ber of flaws. ‘When you look at this publicity, you can pick major errors. Look at the quality of the mam- mogram: even an expert radiologist would not be able to make a diag- nosis correcly from such a low qual- ity mammogram,’ said Thomassin- Naggara, who pointed out further significant limitations, such as the methodology, which used two data sets with very different prevalence of malignancy – 1.6 % in the train- ing cohort vs. 22.2% in the validat- ing cohort. The vast majority of images in this study were acquired on Hologic equipment, so not rep- resentative of manufacturers’ diver- sity. In addition, out of the +25,000 women included in the cohort, only 500 cases were used to compare the AI model with the human readers. However, the main limitation was probably the poor performance of human readers overall, rather than a clearly spectacular performance No compression. No superimposition. JUST GREAT IMAGING www.ab-ct.com The French radiology commu- nity reently highlighted the need to develop cloud software to automati- cally analyse daily images that can detect defaults or instabilities, to help medical physicists: decrease the analysis time during quality con- trol (QC); create automatic analysis of QC criteria to simplify daily prac- tice; offer automatic patient qual- ity control assessment for technolo- gists, for example blurring, position, compression, artefacts; and create a report for technologists’ self-assess- ment. ‘AI with DL will change practice and the place of imaging on the patient pathway. Imaging-based risk prediction of BC and biomarkers are potentially fantastic ways to personalise medicine and improve recommendations to our patients,’ Thomassin-Naggara concluded. (MR) Isabelle Thomassin-Naggara is Professor of Radiology MD at APHP Sorbonne University in Paris, France from AI. ‘Human readers’ perfor- mance was very low and this could be improved,’ she said. By con- trast, another publication used two data sets with a normal prevalence of BC in screening populations in the US and independent validation cohorts in Sweden. ‘Four AI mod- els were tested and the results are interesting: No single AI model out- performed the radiologists. Using the AI models and human reading showed better outcome than using AI or the radiologist alone,’ she said. These results are in line with previ- ous work on an enriched cohort, which showed that radiologists’ performance with AI was better than without. ‘We have an increased specificity and sensitivity when the radiologist is helped by AI, and no limitation of time duration, which used to be a restriction with CAD,’ she added. Improving image acquisition and patient pathway The challenge in these two appli- cations is to lower radiation dose. Bilateral mammography Some software can do scatter cor- rection without using anti-scatter grid, by identifying the structures that cause scatter and subtracts the calculated scatter. The dose-saving depends on breast thickness and structure. Another AI model has been developed to help the techni- cian to achieve optimal positioning, by defining the right compression force and indicating the right expo- sure parameters. After each exposure, anonymised data are sent to an external cloud database, which can analyse the final image quality of each diag- nostic image and give advice on all acquisition parameters. Optimising the patient pathway with AI-enhanced imaging delivers the ability to automatically evaluate breast density and quickly deter- mine which patients will need ultra- sound after mammography. Up to now the different solutions to automatically assess breast densi- ty displayed low reproducibility and Radiography: Correct patient positioning is cr Perfect musculoskeletal imaging Dr Francis Zarb, a senior radiography lecturer at the University of Malta, outlined challenges in correctly positioning patients needing musuloskeletal imaging. During an interview with Sonja Buske he also offered a brief insight into the use of Artificial Intelligence (AI) in radiography today. What are the challenges in posi- tioning the patient for musculo- skeletal radiography? Zarb: The purpose of Radiography is to provide radiographic images using the lowest dose of radiation (ALARA) consistent with obtaining the required diagnostic information, taking into account economic and social factors. A radiograph can be compared to a shadow, in that it is a projection image whereby a 3-D structure is represented on a 2-D image with associated issues, such as super-imposition. A radiographic image should faithfully reproduce the anatomical structures being examined, in terms of contrast and spatial resolution. Factors that need to be considered in producing a radiographic image include the technical aspects of the exposure and patient-related factors such as patient size. These may require changes in exposure factors to compensate for the differences in the thickness of the anatomy being imaged. Also important is the patient’s co-operation, which may require the use of immobilisation aids and the use of the shortest exposure time possible together with effective communication. Image quality is not easy to define! Does incorrect positioning affect radiation dose and image qual- ity? A poor quality image does not pro- vide adequate diagnostic information and may lead to a missed or incor- rect diagnosis. A poor quality image may also lead to a repeat examina- tion, which in turn provides an extra unnecessary radiation dose to the patient. Image contrast is the ability to differentiate between tissues based on their density, i.e. the difference in attenuation of the X-ray beam by the tissues being examined. Optimisation of image contrast is achieved through the appropriate selection of exposure factors, espe- cially the kilo-voltage (kVp) and tube current (mAs) to limit scatter and image noise, both degra-ding image quality. Scatter radiation can be limited in its production through compression (reducing thickness) and collimation (restriction of the size of the X-ray beam). Scatter can also be reduced from reaching the image receptor EUROPEAN HOSPITAL Vol 29 Issue 3/20

1 2 L A B O R AT O RY A new technique to understand metabolic pathways Mass spectrometry- based metabolomics Mass spectrometry-based metabo- lomics has emerged as a powerful tool to help study chemical ecology. Recent advances in the technique make it possible to study microbial interactions from complex commu- nities. Laia Castaño-Espriu outlined the role and benefits of MS in this con- text in her presentation ‘Analysis of microbial ecology by mass spec- trometry-based metabolomics tech- niques’, at the joint meeting of the German Society for Hygiene and Microbiology (DGHM) and the Association for General and Applied Microbiology (VAAM) in Leipzig, Germany this March. Castaño-Espriu explained that the bacterial order Actinobacteria is responsible for the production of 65-70% of microbially-produced specialised metabolites with diverse biological activities, but it is esti- mated that only 10% of biosynthetic genes clusters encoding the produc- tion of these specialised metabolites are transcribed under normal labo- ratory conditions. Microbial competition enhances chemical diversity ‘It’s been observed that microbi- al competition enhances chemical diversity as an ecological advan- tage,’ she said. ‘Therefore, the study of bacterial competition is key in understanding the ecological role that secondary metabolites pose to microbes and in understanding the induction of cryptic biosynthetic gene clusters.’ As a final year PhD student at the University of Strathclyde in Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) equipment Glasgow, Castaño-Espriu’s project is based on the analysis of microbial interactions by co-cultivation and MS-based metabolomics, with a par- ticular focus on Actinobacteria that has been isolated from the marine environment as a potential source of novel chemistry. The analysis of these interactions by LC-MS revealed the production of metabolites that were co-culture specific and they found that sev- eral interactions not only enhanced chemical diversity, but also inhib- ited the growth of several clinically relevant pathogens, she explained. ‘Furthermore, we could gain a better understanding of the chemical ecol- ogy during bacterial competition, where changing conditions aggra- vate the complexity of these interac- tions. Therefore, we demonstrated that MS-based metabolomics is an exciting strategy to study micro- bial ecology and to prioritise novel chemistry.’ Metabolomics approaches – refer- ring to the study of the complete set of metabolites found in a cell, tissue, organ or organism – in gen- eral are applied to compare the chemical profiles of different experi- mental groups to find out the chem- ical response to external stimuli such as environmental stresses, she explained. ‘MS-based metabolomics approaches in microbial ecology enable the understanding of micro- bial interactions in complex com- munities. Metabolomics allows the analysis of those metabolites that are produced during microbial inter- actions, therefore, providing a snap- shot of the ecological role of these metabolites in the producers. This allows the understanding of bio- logical questions such as Why are certain metabolites produced under specific conditions?, she pointed out. By studying the secondary meta- bolism of organisms in complex communities and understanding the role of interesting metabolites, Castaño-Espriu continued, it is pos- sible to gain greater insights in the correlation between phenotypic traits and microbial interactions. utilises Secondary Identification of potentially significant metabolites Time-of- Her work Flight Ion Mass Spectrometry (ToF-SIMS) and Liquid Chromatography Mass Spectrometry (LC-MS). The MS-based metabo- lomics studies follow the steps of sample collection and preparation, generation and acquisition of MS data (metabolic profiles), data pro- cessing using a deconvolution soft- ware such as MZmine, and statistical analyses. ‘This approach,’ she said, ‘enables the identification of poten- tially significant metabolites belong- ing to the groups of interest, there- fore, prioritising these for chemical identification and further experi- ments to understand the underly- ing mechanisms of the metabolic pathways.’ MS-based metabolomics tech- niques offer advantages in fields including the effect of drugs at a metabolite level; drug develop- ment; the study of diseases such as cancer, nutrition science, tox- icology analyses to natural pro- ducts; chemical ecology research. ‘Therefore,’ Castaño-Espriu added, ‘MS-metabolomics repre- sents a multi- and interdisciplinary and Detectable: paediatrics’ tiniest Laia Castaño-Espriu is a final year PhD student at the University of Strathclyde in Glasgow under the supervision of Dr Katherine R. Duncan, with a particular research interest in metabolomics, actino- bacteria and drug discovery. approach that can be used in differ- ent fields to understand metabolic pathways.’ Metabolomics, she added, aims to deliver more rapid and reliable analyses combined with more eco- nomical methods, data standardisa- tion and the sharing of more MS data through online databases. Recent applications of metabo- lomics include disease biomark- ers and drug development, with a significant increase in the use of MS-based metabolomics observed in the area of microbial ecology in the last decade. Compared to other techniques, Castaño-Espriu believes that meta- bolomics is the best approach to accelerate drug development as it is low-costs and also rapid and effective. ‘This effectiveness makes a meta-bolomics approach ideal for other life and medical sciences.’ She feels it will become neces- sary to develop new technologies that offer greater sensitivity and spatial resolution to improve data quality, and to create databases with more MS data coverage to facilitate metabolite identification. Another challenge lies in data variability based on sample prepara- tion and the sample matrix among other factors. However, she believes the development of new instruments and new methods will aim to over- come this limitation. (MN) Device makes veins visible The device VeinViewer® from Christie Medical Holdings benefits from advanced near-infrared light technol- ogy before, during and after venipuncture and venous access, Greiner Bio-One reports. ‘VeinViewer helps you to improve the success rate of the initial venipuncture and increase patient satisfaction and safety.’ ‘Harmless near-infrared (NIR) light is directed towards the patient’s skin,’ the firm explains. ‘Haemoglobin in the blood absorbs the NIR and the surrounding tissue reflects it back to the device, where the data is processed into an image, colour is added and the image sent back to the skin to provide a real time visualisation of the blood vessels and patterns up to 10 mm below it. HD technology with digital full field help provide the clinician with clear, sharp imaging to support routine healthcare work when vein condi- tions are poor.’ The hand carried VeinViewer Flex – for use in emergencies ‘packs the latest advances in superior vascular imaging into a handheld, mobile package,’ the company reports. ‘VeinViewer can help deliver faster venipuncture procedures on your difficult patients through greatly improved first stick success. Flex gives you revolutionary HD views in real-time, so you have the best view of vasculature procedures pre-, dur- ing and post-access.’ Flex runs on Fast Swap Battery and AC power so there are no concerns about a down device and durability meets the busiest use. Durability meets demand in the busiest use in, for example, vein care centres, and departments for oncology, geriatrics, dialysis, casu- alty, emergency/triage, etc. There’s a stability to aid hands-free operation, and the Flex Imaging Unit is ergo- nomically designed for simplified procedures. The standard Universal Mode provides HD and Df2 projec- tion technology, a clear and accurate view of superficial vein patterns in real time, even in remote locations, Greiner Bio-One points out. VeinViewer Vision2 is a mobile imaging unit with an arm reach- ing up to 1.35 metres, attached to a trolley for easy transportation between beds or phlebotomy chairs. The ASSESS Imaging Suite of soft- ware is available only on VeinViewer products. ‘The choice of various EUROPEAN HOSPITAL Vol 29 Issue 3/20

1 4 D I G I TA L PAT H O LO G Y High-capacity digital image analysis Spatial statistics extract subvisual features Spatial analytics offers greater clarity in the assessment of tumours beyond routine microscopic analysis, Mark Nicholls reports. High-capacity digital image analysis enables new methods of spatial sta- tistics to extract features not imme- diately distinguishable by visual inspection. These subvisual features reflect complex properties, such as intratumour heterogeneity and have the additional benefit that they can target specific compartments in the tumour micro-environment. in this Advances field were detailed in a recent webcast organ- ised by Global Engage and spon- sored by Indica Labs, which focused on research that has identified novel prognostic signatures using new spatial analysis and machine learn- ing to measure tumour-immune cell interactions within the tumour immune micro-environment. In his presentation ‘Assessment Intratumour Heterogeneity of and Tumour/Host Interaction using Spatial Analytics’, Arvydas Laurinavicius, Professor of Pathology at Vilnius University, outlined his team’s work on spatial image analyt- ics work based on hexagonal grid subsampling of digital image data. He highlighted how machine reading can generate much more data on the proliferation rate and its spatial heterogeneity in breast cancer compared to conventional microscopic assessment. This ena- bles much richer data sets for fur- ther analyses, generated from rou- tine pathology slides. He further explained that sub- sampling of image analysis data into hexagonal grid tiles provides good opportunities to appraise the tumour texture, the so-called Haralick entropy, which estimates the spatial heterogeneity or ‘disar- rangement’ of tumour proliferation rate. Two analytical models ‘We have developed two analytical models based on hexagonal tiling of image analysis data,’ Laurinavicius added. ‘They enable quantification of subvisual features of intratumor heterogeneity and tumour/host interaction; furthermore, these two methodologies can be combined for integrated prognostic models. ‘We compute spatial entropy of biomarker expression based on the co-occurrence matrix built from all six neighbours to each hexagon.’ Beside the Ki67 heterogeneity indicators obtained by hexagonal til- ing, which outperform proliferation rates, he said it is possible to visu- alise and measure ‘Pareto hotspots’, which represent the ‘hottest 20%’ of the proliferative area in the tumour. ‘We have published and unpub- lished which indicate that heteroge- neity measurement is more informa- tive than the level of biomarker expression itself, even in the hot- spots,’ he explained. In appraising tumour-host inter- action, Laurinavicius stressed the importance of looking for best methods to measure tissue immune response features. A well-established ‘Immunoscore’ system has been shown to pro- vide independent prognostic value in colorectal cancer; remarkably, assessment of the tumour immune microenvironment may significantly improve the prognostic and pre- dictive power of testing in solid tumours, compared to current clini- cal routine. This will guide individ- named ‘Immunogradient’. Several indicators were used to quantify this, including centre of mass, which is a quantitative feature that tells whether these cells are ‘willing’ to enter the tumour or stop at the edge. ‘We can optimise the method for each type of tumour,’ Laurinavicius explained. ‘For breast cancer, we need a broader zone to collect more data to compute Immunogradient indicators, as there are less lym- phocytes. ‘We tested the prognostic value of these indicators. In all prognostic models we have the Immunogradient indicators as independent predictors along with lymph node involvement data in breast cancer and T stage in colorectal cancer. ‘Looking at patient survival prob- abilities in both cancer cohorts, we see that a good prognostic stratifi- cation can actually be achieved by any of the immunogradient family indicators.’ One feature noted in early hor- mone receptor-positive breast can- cer patients was that five years after surgery all patients do quite well with a 92% survival rate but after 10 years there is a dramatic diver- The Interface Zone between the core tumour and stroma is represented by the ribbon of coloured hexagons: red – tumour aspect; yellow – tumour edge; green – stroma (host) aspect. CD8 density profiles are extracted from the corresponding interface zone ranks; they are further used to compute immunogradient indicators. The analysis is performed on immunohistochemically stained colorectal cancer tissue section. ual therapy decisions, in particular, to benefit from new cancer immu- notherapy options. In another example, Laurinavicius pointed to the value of ‘tissue phe- nomics’ to predict prostate can- cer recurrence, by the appraisal of immune cell densities and distances within automated tumour gland/ stroma outline and fixed zones around the outline. Focus then turned to a recent- ly published study in breast and colorectal cancer patient cohorts and novel Interface Zone Immunogradient indicators com- puted from tumour infiltrating lym- phocyte density profiles across the automatically sampled tumour/host front line. ‘We looked at tumours with a slightly different approach,’ Laurinavicius continued. ‘If you want to identify the invasive mar- gin, the question comes as to where should you draw the “invasion line”; you can do it manually, or automati- cally, but should you outline every gland in every case? ‘If you consider where the fight happens between the host and the tumour, it is a rather irregular area. We explored the concept of extract- ing an interface zone, rather than just a margin, or a ribbon, of a fixed width between high-level visual fea- tures.’ A natural question, he said, was whether the hexagonal tiling could be utilised to detect these interface areas. ‘It took us a while but we implemented this procedure based on single CD8 immunochemistry and digital image analysis.’ For the digital image analysis, the researchers utilised the Indica Labs Halo AI to automatically classify the tissue sample and quantify cells. The hexagonal grid overlaid in breast cancer tissue section, stained on immuno- histochemically for Ki67 (proliferation marker). Higher intensity of red shading of the hexagons represents higher local proliferation rate. This data is used to compute texture and bimodality indicators to quantify intratumour heterogeneity of proliferative activity. the tissue content within each hexa- gon, and its surrounding hexagons, to estimate the probability of each hexagon being at the interface. The tumour/stroma edge can then be extracted as those hexagons with high ‘interface-ness’. The interface zone which reaches both into the tumour and host tissue was established around the edge by ranking hexagons according to their distance to the tumour/stroma edge. Example: colorectal cancer In a practical example for colo- rectal cancer, the interface zone was detected and ranked automati- cally and immune cell densities in each rank was used to construct an interface profile of the biomarker density. gence of the two probability curves: high Immunogradient patients maintained 87% survival, while low Immunogradient patients had 55% survival after 10 years. ‘Such knowledge about immune- response properties, obtained from the surgically excised tumour, which predicts long term survival, suggests that these patients should be moni- tored more closely and could ben- efit from immunotherapy during the course of their illness,’ Laurinavicius added. Effective combination of prognostic models The prognostic models based on tumour heterogeneity, interface and gradient, can be effectively com- bined, he said. From the image analysis data they developed a set of rules based on This reveals the immune cells’ ‘gradient towards tumour’ and was Finally, Laurinavicius pointed out that frequently spatial heterogene- EUROPEAN HOSPITAL Vol 29 Issue 3/20

1 6 N E W S & R E S E A RC H The Alzheimer’s Association International Conference Exciting findings: ’flu vaccines and P-tau217 Report: Cynthia E. Keen More than 32,000 people from over 160 countries registered for The Alzheimer’s Association International Conference (AAIC 2020) in July. This largest and most influential inter- national conference on dementia science had to be held virtually this year, when important highlights were aired. P-tau217: An accurate indica- tor of Alzheimer’s in labora- tory blood tests The ability to identify individuals at high risk of developing Alzheimer’s Disease (AD), or at early onset, with simple, inexpensive blood tests would greatly aid research to devel- op therapeutic drugs to treat and potentially prevent the disease. Early detection with such tests could also facilitate immediate treatment to help slow brain damage and would help patients and their families to prepare for the future. Advances in blood tests to identify abnormal versions of the tau pro- tein identified in AD patients were presented at the event. The stud- ies’ results hopefully will accelerate clinical trials of AD biomarker blood tests. Changes in the normal amount of tau and amyloid brain proteins cre- ate clumps in the brain known as plaques and tangles, and build-up of tau tangles is believed to correlate closely with cognitive decline. Positron-emission tomogra- phy (PET) scans and cerebrospi- nal fluid (CSF) biomarker measure- ments are the current methods to identify changes in the brain when Alzheimer’s is suspected before the dementia symptoms of AD appear. But these methods are invasive, very expensive, and not readily available. A multinational team of research- ers, led by Professor Oskar Hansson MD PhD, of Lund University’s Clinical Memory Research Unit in Malmö, Sweden, identified that the blood/ plasma levels of P-tau217 could dis- tinguish AD from other neurodegen- erative disorders with a diagnostic accuracy of 89%-98% in a diversified three-study cohort of 1,402 individu- als from Sweden, Columbia, and the USA. P-tau217 is a form of tau found in tangles and also seems to correlate closely with the build-up of amy- loid. The researchers discovered that P-tau217 was more accurate than three other current experimental bio- markers and comparable to PET imaging and cerebrospinal fluid bio- markers. It also was able to identify AD in post-mortem brain samples of patients with neurodegenerative dis- ease with an 89% accuracy. P-tau217 levels were increased about seven- fold in Alzheimer’s and started to increase 20 years before onset of cognitive impairment in individuals with a gene causing Alzheimer’s, Professor Hansson reported. ‘This test,’ he said, ’once verified and con- firmed, opens the possibility of early diagnosis of Alzheimer’s before the dementia stage, which is very impor- tant for clinical trials evaluating novel Adina Zeki Al Hazzouri PhD Albert Amran Elisabeth Thijssen Professor Oskar Hansson Suzanne Schindler MD, PhD Svetlana Ukraintseva PhD therapies that might stop or slow down the disease process.’ Another factor of interest was that the test could distinguish individuals who carried a gene mutation leading to genetic early-onset Alzheimer’s starting in middle age, as young as age 25. The study included 600 members of an extended family of 6,000 living in Columbia who carry the mutation. Study findings from the Washington University School of Medicine in St. Louis, Missouri, and the Memory and Aging Center of the UCSF Weill Institute for Neurosciences in San Francisco, also supported the prom- ise of plasma P-tau217 as a biomark- er to identify AD in blood tests. Neurologist Suzanne Schindler MD PhD, and colleagues from Washington University, used mass spectrometry to map the levels of both P-tau217 and P-tau181, high levels of which occur in the brains of AD patients. They determined that P-tau217 was more closely linked to the build-up tive study of over 9,000 individuals under the age of 60 had a lower risk of developing AD if they received one flu vaccination, and that the risk continued to decline if they received subsequent annual vaccinations. People who received at least one flu vaccination were 17% less likely to develop AD over the course of their lifetime. Multiple annual vacci- nations reduced the risk by another 13%, according to presenter Albert Amran of the McGovern Medical School. He also reported that individ- uals who received their first vaccina- tion at younger ages were associated with even greater decreases in risk. The researchers developed their cohort from Cerner Health Facts, a database of electronic health records from over 600 participating hospitals and clinics hosted by UT’s School of Biomedical Informatics. They used machine learning to analyse more than 310,000 health records to study the relationship between ’flu vaccina- tion and AD. Svetlana Ukraintseva PhD reported that the research team studied asso- ciations between the pneumococcal vaccine, with and without an accom- panying ’flu shot, after adjusting for sex, race, birth cohort, educa- tion, smoking, and the number of rs2075650 G alleles in the TOMM40 gene, a known genetic risk factor for Alzheimer’s. They did not iden- tify any additional risk reductions in individuals who also had a flu shot in this study cohort, but they did determine that individuals who were not carriers of the gene had the larg- est risk reduction of up to 40% when they had an anti-pneumonia vaccine. Young adult obesity increas- es risk of late-life dementia Mid-life obesity is a well-established- risk factor for dementia, but in what may be the first study of young adults aged 20 to 49, researchers from Columbia University’s Population Research Center in New York City, have associated large body mass r e n t r e a G n a u J / k c o t s r e t t u h S of amyloid plaques in the brain as measured by PET imaging. Based on these findings, the researchers have launched an 1,100-participant clini- cal trial to develop and validate AD blood biomarkers. Elisabeth Thijssen presented study findings from the UCSF Memory and Aging Center that showed both P-tau217 and P-tau181 are elevated in blood tests of AD patients, with a diagnostic accuracy of 96% and 91% respectively compared to PET imaging. The 617-participant cohort in their study included healthy con- trols as well as patients with AD and frontotemporal lobar degenera- tion (FTLD) spectrum. Results also showed that P-tau217 was five- fold greater and P-tau181 four-fold greater in AD patients compared to healthy controls and FTLD patients respectively. Seasonal flu and pneumonia vaccinations reduce AD risk Researchers from the University of Texas Health Science Center in Houston reported that a retrospec- Alzheimer’s disease: Tau proteins aggregate to neurofibrillary tangles in a neuron axon. The transport of synaptic vesicles is disrupted. 3d rendering. ‘One of our theories of how the ’flu vaccine may work is that some of the proteins in the ’flu virus may train the body’s immune response to better protect against AD,’ Amran said. ‘Providing people with a flu vaccine may be a safe way to intro- duce those proteins that could help prepare the body to fight off the disease.’ Similarly, individuals aged 65-75 who had a pneumococcal vaccina- tion had a reduced risk of develop- ing AD by 25% to 30%, according to a research study conducted at the Duke University Social Science Research Institute’s Biodemography of Aging Research Unit in Durham, NC. The study included over 5,000 individuals participating the Cardiovascular Health Study, a pop- ulation-based longitudinal study of coronary heart disease and stroke in adults 65 years and older. in index (BMI) with higher late-life dementia risk. Led by Adina Zeki Al Hazzouri PhD, a research team studied over 5,100 older adults, estimating their BMI starting at age 20. She reported that dementia risk was 1.8 times higher among women who were overweight and 2.5 times higher among obese women compared to women with normal BMI in early adulthood. Contrary to prior pub- lished studies, they found no asso- ciation between midlife BMI and dementia risk among women. Dementia risk for men, who repre- sented 44% of the study cohort, was 2.5 times higher for those who were obese in early adulthood, 2.0 times higher for mid-life obesity, and 1.5 times higher for men who were over- weight in mid-life. Interestingly, the researchers discovered that obesity in late life for both men and women decreased dementia risk, suggest- ing that high late life BMI may be protective. EUROPEAN HOSPITAL Vol 29 Issue 3/20