Real-world Clinical Studies Research Articles

Psoriasis: a disease beyond the skin.

Psoriasis is a chronic, immune-mediated inflammatory dermatosis with a multifactorial etiology, affecting approximately 3% of the Italian population. This condition not only imposes a substantial burden on physical health but also significantly diminishes the personal and social quality of life of affected individuals. This study aims to assess the critical role of integrated management approaches in psoriasis, focusing specifically on the role of primary care physicians in coordinating multidisciplinary care and addressing comorbidities. A comprehensive review of current therapeutic guidelines, classification systems, and available treatment strategies was conducted. This was supplemented with real-world clinical case studies. The study evaluated intervention approaches at both primary and secondary care levels, involving collaborations between general practitioners and specialists. Special emphasis was placed on the use of clinical assessment tools such as the Psoriasis Area and Severity Index (PASI) to categorize disease severity and guide therapeutic decisions. Our findings underscore the effectiveness of a multidisciplinary and personalized approach in managing psoriasis, particularly in moderate-to-severe cases. The pivotal role of primary care physicians in the early diagnosis, ongoing monitoring, and coordination of care pathways was highlighted, leading to significant improvements in comorbidity management. The clinical cases presented illustrate how tailored therapeutic regimens, supported by close collaboration between general practitioners and dermatologists, can result in the effective resolution of psoriatic lesions and markedly enhance patients’ quality of life. This study reinforces the necessity of a comprehensive, integrated, and multidisciplinary approach in the management of psoriasis, with primary care physicians serving as crucial intermediaries between community-based and specialist care. Depending on disease severity, treatment response, and recurrence patterns, primary care physicians play a central role in directing patients to appropriate secondary care services. The integration of specialist dermatological care with ongoing community-based monitoring optimizes clinical outcomes and mitigates the overall disease burden. We advocate for the establishment of clinical networks that facilitate seamless communication between different levels of care, ensuring holistic and targeted management of both psoriasis and its associated comorbidities. Keywords: Psoriasis, primary care physician, multidisciplinary care, integrated therapy, quality of life, comorbidities.

Clinical rebound after treatment with nirmatrelvir/ritonavir in COVID-19

BackgroundNirmatrelvir/ritonavir (NM/r) is a safe and effective oral antiviral therapeutic used for treatment of mild-to-moderate COVID-19. Case reports described a clinical rebound syndrome whereby individuals experience a relapse of symptoms shortly after completing successful treatment. There is a lack of information on frequency of COVID-19 rebound after NM/r in routine clinical care, contributing factors, and clinical outcomes.MethodsWe reviewed electronic medical records to verify COVID-19 diagnosis, symptoms, and treatment with NM/r from January-June 2022. We defined COVID-19 clinical rebound as clear improvement in symptoms followed by recurrence or worsening of symptoms within 30 days of a five-day course of NM/r.ResultsWe studied 268 adults with median age 57 (IQR 47, 68), 80% White race, 85% non-Hispanic ethnicity, 55% female, 80% vaccinated and boosted against SARS-CoV-2, and 68% with any co-morbidity. Sixteen (6.0%) of studied patients were determined to have COVID-19 clinical rebound. The median time from starting NM/r to rebound was 11 days (IQR 9, 13). Notable demographic and clinical factors with higher proportion (not statistically significant) among COVID-19 rebound patients were female sex (75% rebound vs. 54.5% no rebound), Black race (12.5% rebound vs. 4.9% no rebound), presence of at least one co-morbidity (81.3% rebound vs. 67.5% no rebound), and lack of prior SARS-CoV-2 infection (100% rebound vs. 92.9% no rebound). Only one patient (6.25%) was hospitalized after COVID-19 rebound.ConclusionsCOVID-19 clinical rebound after treatment with NM/r is mild with favorable outcomes and more common than previously reported from real-world clinical care studies.

COVID-19 clinical rebound after treatment with nirmatrelvir/ritonavir.

Nirmatrelvir/ritonavir (NM/r) is a safe and effective oral antiviral therapeutic used for treatment of mild-to-moderate COVID-19. Case reports described a clinical rebound syndrome whereby individuals experience a relapse of symptoms shortly after completing successful treatment. There is a lack of information on frequency of COVID-19 rebound after NM/r in routine clinical care, contributing factors, and clinical outcomes. We reviewed electronic medical records to verify COVID-19 diagnosis, symptoms, and treatment with NM/r from January-June 2022. We defined COVID-19 clinical rebound as clear improvement in symptoms followed by recurrence or worsening of symptoms within 30 days of a five-day course of NM/r. We studied 268 adults with median age 57 (IQR 47, 68), 80% White race, 85% non-Hispanic ethnicity, 55% female, 80% vaccinated and boosted against SARS-CoV-2, and 68% with any co-morbidity. Sixteen (6.0%) of studied patients were determined to have COVID-19 clinical rebound. The median time from starting NM/r to rebound was 11 days (IQR 9, 13). Notable demographic and clinical factors with higher proportion (not statistically significant) among COVID-19 rebound patients were female sex (75% rebound vs 54.5% no rebound), Black race (12.5% rebound vs 4.9% no rebound), presence of at least one co-morbidity (81.3% rebound vs 67.5% no rebound), and lack of prior SARS-CoV-2 infection (100% rebound vs 92.9% no rebound). Only one patient (6.25%) was hospitalized after COVID-19 rebound. COVID-19 clinical rebound after treatment with NM/r is mild with favorable outcomes and more common than previously reported from real-world clinical care studies.

Exploring the association between microscopic colitis and celiac disease: A comprehensive analysis using the national in-patient data (2016-2019).

Several investigations suggested correlation between microscopic colitis (MC) and celiac disease (CD). This study aimed to examine this relationship using large-sized, population-based data with adequate control for confounding factors. This study employed the National Inpatient Sample (NIS) database over 4 years (2016-2019). Patients with/without MC in the presence/absence of CD were identified through ICD-10 codes. Univariate and multi-variate analyses involving odds ratios (OR) and 95% confidence intervals (CI) were performed. Overall, 26,836,118 patients were analyzed. Of whom, 6,836 patients had MC (n = 179 with CD and n = 6,657 without CD). The mean hospital stay was not significantly different between both groups (5.42 ± 5.44 days vs. 4.95 ± 4.66 days, P = 0.202). The univariate analysis revealed a significant association between MC and CD (OR = 22.69, 95% [19.55, 26.33], P < 0.0001). In the multi-variate analysis, which adjusted for potential confounders including age, race, hospital region, hospital teaching status, ZIP income, smoking status, alcohol overuse, hypertension, diabetes mellitus, lipidemia-related disorders, non-steroidal anti-inflammatory drug use, and selected auto-immune diseases, the association remained significant (OR = 15.71, 95% CI [13.52, 18.25], P < 0.0001). Moreover, in patients with MC, the presence of CD emerged as a significant, independent variable of in-hospital mortality in univariate (OR = 2.87, 95% [1.14, 7.21], P = 0.025) and multi-variate (OR = 3.37, 95% CI [1.32, 8.60], P = 0.011) analyses. This study establishes a probable link between MC and CD, backed by both univariate and multi-variate analyses, while also identifying CD as an independent risk factor for increased mortality among MC patients. These findings need to be validated in real-world clinical studies.

The inter-link of ageing, cancer and immunity: findings from real-world retrospective study

BackgroundAlthough the concept of declined immune function associated with cancer has been accepted extensively, real-world clinical studies focusing on analysis of the peripheral blood immune changes underlying ageing, immunity and cancer are scarce.MethodsIn this case-control study, we retrospectively analysed 1375 cancer patients and enrolled 275 age and gender matched healthy individuals. Flow cytometry was conducted to assess the immune changes. Further analysis was examined by SPSS 17.0 and GraphPad Prism 9 software.ResultsCancer patients showed obviously decreased CD3+ T, CD3+CD4+ Th, CD3+CD8+ CTL, CD19+ B, CD16+CD56+ NK cell counts and lower percentage of PD-1 (programmed cell death protein-1, PD-1) positive cells than healthy control (P < 0.0001). For cancer patients, the reference range of circulating percentage of PD-1+CD45+ cells, PD-1+CD3+ T cells, PD-1+CD3+CD4+ Th cells and PD-1+CD3+CD8+ CTL (Cytotoxic T Lymphocyte, CTL) were 11.2% (95% CI 10.8%-11.6%), 15.5% (95% CI 14.7%-16.0%), 15.4% (95% CI 14.9%-16.0%) and 14.5% (95% CI 14.0%-15.5%), respectively. Moreover, the reduction of CD3+ T, CD3+CD4+ Th, CD3+CD8+ CTL, CD19+ B cell counts accompanied with age and stage advancing (P < 0.05). CD16+CD56+ NK cells decreased with stage, but elevated in aged and male cancer patients (P < 0.05). Additionally, the percentage of PD-1 positive cells varied across cancer types, raised with age and stage. Head and neck, pancreatic, gynaecological and lung demonstrated a higher level of the percentage of PD-1 positive cells than melanoma, prostate, and breast cancer (P < 0.05).ConclusionsThis study provides the reference range of the percentage of PD-1 positive cells on peripheral blood, confirms the decreased immune cells and a series of immune changes accompanying with cancer, expands our real world evidence to better understand the interactions of ageing, cancer and immunity. Moreover, the circulating percentage of PD-1 positive cells shows similar tumor type distribution with tumor mutational burden (TMB), supports that it maybe a potential predictive biomarker for immune checkpoint inhibitor therapy.

Deep reinforcement learning for multi-class imbalanced training: applications in healthcare

With the rapid growth of memory and computing power, datasets are becoming increasingly complex and imbalanced. This is especially severe in the context of clinical data, where there may be one rare event for many cases in the majority class. We introduce an imbalanced classification framework, based on reinforcement learning, for training extremely imbalanced data sets, and extend it for use in multi-class settings. We combine dueling and double deep Q-learning architectures, and formulate a custom reward function and episode-training procedure, specifically with the capability of handling multi-class imbalanced training. Using real-world clinical case studies, we demonstrate that our proposed framework outperforms current state-of-the-art imbalanced learning methods, achieving more fair and balanced classification, while also significantly improving the prediction of minority classes.

Individualizing Medicinal Therapy Post Heart Stent Implantation: Tailoring for Patient Factors.

The field of cardiovascular medicine is undergoing a transformative shift towards personalized medicinal therapy, particularly in the context of post stent implantation. This narrative review explores the significance, challenges, and future directions of individualized treatment strategies for patients with coronary stents. The review highlights the pivotal role of personalized approaches in optimizing treatment efficacy and minimizing adverse events. Real-world clinical studies and trials underscore the importance of tailoring antiplatelet therapy based on platelet function testing, genetic testing, and risk scoring. These studies reveal that personalized medicinal treatment improves clinical outcomes by balancing preventing thrombotic events and mitigating bleeding risks. Challenges, including cost, test availability, patient adherence, and ethical considerations, are discussed in depth, shedding light on the complexities of implementing personalized approaches. Technological advancements, including omics data integration, artificial intelligence, and big data analytics, shape the future of personalized medicinal therapy. These tools enable precise pharmacogenomic selection of medications and the development of integrated risk-scoring systems. Patient engagement and education are also central, with empowered patients and remote monitoring contributing to collaborative decision-making. In conclusion, the narrative review underscores that personalized medicinal therapy post stent implantation holds immense promise for revolutionizing cardiovascular care. By embracing a comprehensive approach that considers genetics, clinical factors, and patient preferences, healthcare providers can optimize treatment outcomes and improve patient quality of life. The evolving landscape of personalized medicine offers a glimpse into a future where tailored treatment strategies become the cornerstone of precision cardiovascular care.

895-P: Effect of Long-Term Empagliflozin Therapy on Plasma Beta-Hydroxy Butyrate (ß-OHB)—Lack of Effect of Concomitant Insulin, GLP-1 Agonist, or Pioglitazone Therapy

Sodium-glucose transporter 2 (SGLT2) inhibitors are associated with increased risk of euglycemic diabetic ketoacidosis in T2DM. Possible mechanisms include increased lipolysis, elevated glucagon, and decreased plasma insulin levels. Therefore, concomitant high dose insulin as well as thiazolidinediones could potentially attenuate the rise in β-OHB because of anti-lipolytic effects. However, this has not been examined in the real-world clinical studies. We examined the effect of long-term Empagliflozin therapy on plasma β-OHB in veterans with T2DM. A retrospective chart review was performed in 43 subjects who had fasting β-OHB, C-peptide and detailed anthropometric measurements and were followed for 23 ± 2.5 months. Subjects were categorized into two groups based on fasting β-OHB concentrations (&amp;lt; or ≥2.8mg/dl). Eleven patients (24.4%) had high (6.03 ±1.22mg/dl) while 32 (74.6%) had normal β-OHB (1.42±0.09 mg/dL). There was no difference in age (59 ± 4 vs 62 ±2 yrs), BMI (35±1 vs 32 ±1 kg/m2), or the duration of diabetes (15 ± 3 vs 18±2 yrs) between the two groups. Subjects with higher β-OHB had higher HbA1c (8.9 ±0.7 vs 7.9 ± 0.3%, p&amp;lt;0.05). There was no difference in fasting C-peptide (3.2 ± 0.9 vs 3.5 ± 0.5ng/ml), total daily insulin dose (184± 43 vs 214±19 units/day), number of subjects on Pioglitazone (54 vs 35%) or GLP-1 agonist therapy (36 vs 52%) between the two groups. In conclusion, long-term therapy with Empagliflozin leads to elevated β-OHB in 25% of T2DM and it was associated with poor glycemic control. Concomitant insulin or pioglitazone therapy did not alter plasma β-OHB concentration. Disclosure L.C.Ortiz: None. J.Wortham: None. S.Pinkson: None. J.Trejo: None. A.S.Mendoza: None. X.Chen: None. D.Tripathy: None.

The Incidence and Prognosis Value of Perineural Invasion in Rectal Carcinoma: From Meta-Analyses and Real-World Clinical Pathological Features

AimsPerineural invasion (PNI) is a special type of metastasis of several cancers and has been reported as being a factor for poor prognosis in colorectal carcinoma. However, investigations of PNI in only rectal cancer and a comprehensive analysis combining meta-analyses with real-world case studies remain lacking. Materials and methodsFirst, articles from 2000 to 2020 concerning the relationship between PNI and rectal cancer prognoses and clinical features were meta-analysed. Subsequently, we carried out a retrospective analysis of 312 rectal cancer cases that underwent radical surgery in the real world. The incidence of PNI and the relationship between PNI and prognosis, as well as clinicopathological factors, were investigated. ResultsThe incidence of PNI was 23.09% and 33.01% in the meta-analysis and clinical cases, respectively. PNI occurred as early as stage I (2.94%). Moreover, neoadjuvant therapy significantly reduced the PNI-positive rate (20.34% versus 26.54%). Both meta-analysis and real-world clinical case studies suggested that PNI-positive patients had poorer prognoses than PNI-negative patients. We established an effective risk model consisting of T stage, differentiation and lymphovascular invasion to predict PNI in rectal cancer. ConclusionPNI is a poor prognostic factor for rectal cancer and could occur even in stage I. Additionally, neoadjuvant therapy could sufficiently reduce the PNI-positive rate. T stage, lymphovascular invasion and differentiation grade were independent risk factors for PNI and the risk model that included these factors could predict the probability of PNI.

Switching between Originators and Biosimilars in Dermatology: A Systematic Review of Real-World Clinical Studies

Background. Although biosimilars have been increasingly used over recent years, some concerns about a potential loss of efficacy and altered safety profile when switching from an originator to a biosimilar still exist. Interchangeability can be a challenge for dermatologists too. An extensive systematic review of published switching studies among originators and biosimilars was carried out in order to provide evidence regarding the effects derived from the switch in terms of efficacy and safety outcomes in real-life contexts. Results. Thirty-seven articles were included in this systematic review (14 studies related to adalimumab, 10 to etanercept, 12 to infliximab, and 1 each to adalimumab, etanercept, and infliximab). Studies were mainly carried out among European countries. Most of them were observational studies or register-based studies. The majority of studies enrolled patients diagnosed with psoriasis or psoriatic arthritis who underwent a single switch from the originator to the biosimilar. Overall, the studies’ results demonstrated that switching between adalimumab, etanercept, and infliximab originators and biosimilars is safe and effective in a real-life setting of patients with dermatological conditions. Only a few studies highlighted an increase in the risk of loss of efficacy as well as an increased rate of AEs, both of which were identified as the main causes of biosimilar discontinuation, probably associated with the well-known phenomenon of the nocebo effect. Conclusion. Switching from a biologic originator to its biosimilar is safe and effective. Only a few studies have evaluated the switch among biosimilars; thus, no firm conclusion can be drawn for this type of switch in terms of the efficacy and safety outcomes. Based on our results, we believe that biosimilars can be considered interchangeable with their reference products and that no additional switch studies are necessary to support switching among originators and biosimilars in clinical practice. However, the continuous monitoring of all biologics (both originators and biosimilars) in routine clinical practice is strongly needed given their peculiar safety profile.

Artificial Intelligence in Paediatric Tuberculosis.

Tuberculosis (TB) continues to be a leading cause of death in children despite global efforts focused on early diagnosis and interventions to limit the spread of the disease. This challenge has been made more complex in the context of the coronavirus pandemic, which has disrupted the "End TB Strategy" and framework set out by the World Health Organization (WHO). Since the inception of artificial intelligence (AI) more than 60years ago, the interest in AI has risen and more recently we have seen the emergence of multiple real-world applications, many of which relate to medical imaging. Nonetheless, real-world AI applications and clinical studies are limited in the niche area of paediatric imaging. This review article will focus on how AI, or more specifically deep learning, can be applied to TB diagnosis and management in children. We describe how deep learning can be utilised in chest imaging to provide computer-assisted diagnosis to augment workflow and screening efforts. We also review examples of recent AI applications for TB screening in resource constrained environments and we explore some of the challenges and the future directions of AI in paediatric TB.

ClusterMLD: An Efficient Hierarchical Clustering Method for Multivariate Longitudinal Data

Longitudinal data clustering is challenging because the grouping has to account for the similarity of individual trajectories in the presence of sparse and irregular times of observation. This paper puts forward a hierarchical agglomerative clustering method based on a dissimilarity metric that quantifies the cost of merging two distinct groups of curves, which are depicted by B-splines for the repeatedly measured data. Extensive simulations show that the proposed method has superior performance in determining the number of clusters, classifying individuals into the correct clusters, and in computational efficiency. Importantly, the method is not only suitable for clustering multivariate longitudinal data with sparse and irregular measurements but also for intensely measured functional data. Towards this end, we provide an R package for the implementation of such analyses. To illustrate the use of the proposed clustering method, two large clinical data sets from real-world clinical studies are analyzed.

Clinical biostatistics in the 2020s

Clinical biostatistics in the 2020s

Analysis of RWD/RWE recommendations and guidelines for 2021. International experience on the example of USA

The regulatory aspects of the conduction of the real-world clinical data studies became a trend in the past few years. In 2021 FDA issued three projects of such documents. The article resumes the main objectives of these FDA documents. Analysed projects of FDA recommendations represents the international expertise in this field and may be used in the process of building local regulations.

Non-invasive monitoring and treat-to-target approach are cost-effective in patients with mild-moderate ulcerative colitis.

There are no data to assess the value associated with a treat-to-target (T2T) strategy based on tight control of mild-moderate ulcerative colitis (UC). To assess the cost-effectiveness of a T2T approach based on the normalisation of clinical signs and faecal calprotectin (FC) METHODS: A decision analytical Markov model was developed to compare T2T algorithm combining clinical symptoms and FC levels to define treatment response and the possible switch to the next treatment line (T2T-FC), and the reference strategy based only on symptoms. The model included five treatment lines and was conducted from the Italian national health service (NHS) perspective using a 3-year time horizon. The model calculated the incremental cost-effectiveness ratio as € per relapse avoided. Deterministic and probabilistic sensitivity analyses were conducted. The cost-effectiveness analysis produced an increased time spent by a patient in clinical remission and FC ≤ 100 level (+0.177 years; about 2months) and a decreasing number of relapses (-0.1937; -20.9%) per patient using a T2T-FC approach compared to only symptoms. Furthermore, the T2T-FC was associated with higher cost (+€1795). The ICER estimated was €9263 per relapse avoided. These results were confirmed by sensitivity analyses. T2T-FC approach resulted in a higher benefit for mild-moderate UC patients in terms of time in remission and incidence of relapse but was associated with higher costs. Clinical trials and real-world clinical studies are needed to provide additional data on the cost-benefit of this approach.

Predicting Post-Operative Complications with Wearables

Post-operative complications and hospital readmission are of great concern to surgical patients and health care providers. Wearable devices such as Fitbit wristbands enable long-term and non-intrusive monitoring of patients outside clinical environments. To build accurate predictive models based on wearable data, however, requires effective feature engineering to extract high-level features from time series data collected by the wearable sensors. This paper presents a pipeline for developing clinical predictive models based on wearable sensors. The core of the pipeline is a multi-level feature engineering framework for extracting high-level features from fine-grained time series data. The framework integrates a set of techniques tailored for noisy and incomplete wearable data collected in real-world clinical studies: (1) singular spectrum analysis for extracting high-level features from daily features over the course of the study; (2) a set of daily features that are resilient to missing data in wearable time series data; (3) a K-Nearest Neighbors (KNN) method for imputing short missing heart rate segments; (4) the integration of patients' clinical characteristics and wearable features. We evaluated the feature engineering approach and machine learning models in a clinical study involving 61 patients undergoing pancreatic surgery. Linear support vector machine (SVM) with integrated feature engineering achieved an AUROC of 0.8802 for predicting post-operative readmission or severe complications, which significantly outperformed the existing rule-based model used in clinical practice and other state-of-the-art feature engineering approaches.

The Optimal Management of Fistulizing Crohn's Disease: Evidence beyond Randomized Clinical Trials.

Fistulizing Crohn’s disease (FCD) remains the most challenging aspect of treating patients with CD. FCD can occur in up to 30% of patients with CD and may lead to significant disability and impaired quality of life. The optimal treatment strategies for FCD require a multidisciplinary approach, including a combined medical and surgical approach. The therapeutic options for FCD are limited due to sparse evidence from randomized clinical trials (RCTs). The current recommendations are mainly based on post hoc analysis from RCTs, real-world clinical studies and expert opinion. There is variation in everyday clinical practice amongst gastroenterologists and surgeons. The evidence for anti-tumor necrosis factor therapy is the strongest in the treatment of FCD. However, long-term fistula healing can be achieved in only 30–50% of patients. In recent years, emerging data in the advent of therapeutic modalities, including the use of new biologic agents, therapeutic drug monitoring, novel surgical methods and mesenchymal stem cell therapy, have been shown to improve outcomes in achieving fistula healing. This review summarizes the existing literature on current and emerging therapies to provide guidance beyond RCTs in managing FCD.

An Easy Approach to Understanding Acid-Base Balance in a Blood Buffer System

Understanding acid-base disorders using weak-acid concepts learned in general chemistry class is challenging for pre-nursing and pre-professional biology students enrolled in anatomy/physiology and biochemistry classes. We utilized a graphic seesaw model of carbonic acid-bicarbonate equilibrium using the Henderson-Hasselbalch (H-H) equation of a weak acid. We then used real-world clinical case studies for students to identify acid-base disorders and the appropriate compensatory responses of the lungs and kidneys. Students developed a working knowledge of how the bicarbonate blood buffer system maintains a physiological pH of 7.4 using a “seesaw” with metabolic [HCO3−] on one side, and respiratory PCO2 on the other at a ratio of 20:1 in the H-H equation. When the dysfunction of either the kidneys or lungs causes the seesaw to tip, homeostasis pH is disrupted, causing an acid-base disorder classified as metabolic or respiratory acidosis or alkalosis. The functioning organ can “level the seesaw” by compensating for the dysfunction of the opposite organ to regain homeostasis. Unlike traditional ways of explaining acid-base disorders, this graphic seesaw method is a simple and easy way to achieve understanding.

SHARP: enabling generation of real-world evidence on a pan-European scale to improve the lives of individuals with severe asthma.

Real-world evidence is important to help unravel unanswered problems in severe asthma and is valuable to better understand the patient experience and common clinical practice.The Severe Heterogeneous Asthma Registry, Patient-centred (SHARP) Clinical Research Collaboration is created as a network of national registries and severe asthma centres that work together to perform registry based real-world research and clinical studies on a pan-European scale.Such collaboration requires a new, innovative design to overcome the many issues that arise with large-scale data collection across national borders. SHARP has developed a platform that offers a federated analysis approach where national registry data are transformed and integrated into a common data model (CDM). The CDM then allows a local analysis of de-identified patient data and subsequent aggregate (meta-)analysis. To facilitate an easily accessible way to set up new registries, SHARP enables new registries to take part in a central database, based on already proven technology. Next to being economical, this linkage ensures data from different SHARP central members to be comparable.Technological advancements lead to an ever-expanding rate of patient data that will be collected; with the collective effort of the pan-European severe asthma research community SHARP hopes to ensure that they are well equipped to enter a new era of medical research, with the ultimate goal to positively impact the lives of patients with severe asthma.

MATHLA: a robust framework for HLA-peptide binding prediction integrating bidirectional LSTM and multiple head attention mechanism

BackgroundAccurate prediction of binding between class I human leukocyte antigen (HLA) and neoepitope is critical for target identification within personalized T-cell based immunotherapy. Many recent prediction tools developed upon the deep learning algorithms and mass spectrometry data have indeed showed improvement on the average predicting power for class I HLA-peptide interaction. However, their prediction performances show great variability over individual HLA alleles and peptides with different lengths, which is particularly the case for HLA-C alleles due to the limited amount of experimental data. To meet the increasing demand for attaining the most accurate HLA-peptide binding prediction for individual patient in the real-world clinical studies, more advanced deep learning framework with higher prediction accuracy for HLA-C alleles and longer peptides is highly desirable.ResultsWe present a pan-allele HLA-peptide binding prediction framework—MATHLA which integrates bi-directional long short-term memory network and multiple head attention mechanism. This model achieves better prediction accuracy in both fivefold cross-validation test and independent test dataset. In addition, this model is superior over existing tools regarding to the prediction accuracy for longer ligand ranging from 11 to 15 amino acids. Moreover, our model also shows a significant improvement for HLA-C-peptide-binding prediction. By investigating multiple-head attention weight scores, we depicted possible interaction patterns between three HLA I supergroups and their cognate peptides.ConclusionOur method demonstrates the necessity of further development of deep learning algorithm in improving and interpreting HLA-peptide binding prediction in parallel to increasing the amount of high-quality HLA ligandome data.