Explore the vast range of titles available.

Pancreatic cancer is an aggressive disease that typically presents late with poor outcomes, indicating a pronounced need for early detection. In this study, we applied artificial intelligence methods to clinical data from 6 million patients (24,000 pancreatic cancer cases) in Denmark (Danish National Patient Registry (DNPR)) and from 3 million patients (3,900 cases) in the United States (US Veterans Affairs (US-VA)). We trained machine learning models on the sequence of disease codes in clinical histories and tested prediction of cancer occurrence within incremental time windows (CancerRiskNet). For cancer occurrence within 36 months, the performance of the best DNPR model has area under the receiver operating characteristic (AUROC) curve = 0.88 and decreases to AUROC (3m) = 0.83 when disease events within 3 months before cancer diagnosis are excluded from training, with an estimated relative risk of 59 for 1,000 highest-risk patients older than age 50 years. Cross-application of the Danish model to US-VA data had lower performance (AUROC = 0.71), and retraining was needed to improve performance (AUROC = 0.78, AUROC (3m) = 0.76). These results improve the ability to design realistic surveillance programs for patients at elevated risk, potentially benefiting lifespan and quality of life by early detection of this aggressive cancer. A deep learning algorithm using electronic health records from two large cohorts of patients predicts the risk of pancreatic cancer from pre-cancer disease trajectories up to 3 years in advance, showing promising performance in retrospective validation.

The CRISPR/Cas9 system has been proposed as a cure strategy for HIV. However, few published guide RNAs (gRNAs) are predicted to cleave the majority of HIV-1 viral quasispecies (vQS) observed within and among patients. We report the design of a novel pipeline to identify gRNAs that target HIV across a large number of infected individuals. Next generation sequencing (NGS) of LTRs from 269 HIV-1-infected samples in the Drexel CARES Cohort was used to select gRNAs with predicted broad-spectrum activity. In silico , D-LTR-P4-227913 (package of the top 4 gRNAs) accounted for all detectable genetic variation within the vQS of the 269 samples and the Los Alamos National Laboratory HIV database. In silico secondary structure analyses from NGS indicated extensive TAR stem-loop malformations predicted to inactivate proviral transcription, which was confirmed by reduced viral gene expression in TZM-bl or P4R5 cells. Similarly, a high sensitivity in vitro CRISPR/Cas9 cleavage assay showed that the top-ranked gRNA was the most effective at cleaving patient-derived HIV-1 LTRs from five patients. Furthermore, the D-LTR-P4-227913 was predicted to cleave a median of 96.1% of patient-derived sequences from other HIV subtypes. These results demonstrate that the gRNAs possess broad-spectrum cutting activity and could contribute to an HIV cure.

Even in the era of combination antiretroviral therapies used to combat human immunodeficiency virus type 1 (HIV-1) infection, up to 50 % of well-suppressed HIV-1-infected patients are still diagnosed with mild neurological deficits referred to as HIV-associated neurocognitive disorders (HAND). The multifactorial nature of HAND likely involves the HIV-1 accessory protein viral protein R (Vpr) as an agent of neuropathogenesis. To investigate the effect of naturally occurring variations in Vpr on HAND in well-suppressed HIV-1-infected patients, bioinformatic analyses were used to correlate peripheral blood-derived Vpr sequences with patient neurocognitive performance, as measured by comprehensive neuropsychological assessment and the resulting Global Deficit Score (GDS). Our studies revealed unique associations between GDS and the presence of specific amino acid changes in peripheral blood-derived Vpr sequences [neuropsychological impairment Vpr (niVpr) variants]. Amino acids N41 and A55 in the Vpr sequence were associated with more pronounced neurocognitive deficits (higher GDS). In contrast, amino acids I37 and S41 were connected to measurably lower GDS. All niVpr variants were also detected in DNA isolated from HIV-1-infected brain tissues. The implication of these results is that niVpr variants alter the genesis and/or progression of HAND through differences in Vpr-mediated effects in the peripheral blood and/or the brain.

Background HIV-1 entry is a receptor-mediated process directed by the interaction of the viral envelope with the host cell CD4 molecule and one of two co-receptors, CCR5 or CXCR4. The amino acid sequence of the third variable (V3) loop of the HIV-1 envelope is highly predictive of co-receptor utilization preference during entry, and machine learning predictive algorithms have been developed to characterize sequences as CCR5-utilizing (R5) or CXCR4-utilizing (X4). It was hypothesized that while the V3 loop is predominantly responsible for determining co-receptor binding, additional components of the HIV-1 genome may contribute to overall viral tropism and display sequence signatures associated with co-receptor utilization. Results The accessory protein Tat and the HlV-1 long terminal repeat (LTR) were analyzed with respect to genetic diversity and compared by Jensen–Shannon divergence which resulted in a correlation with both mean genetic diversity as well as the absolute difference in genetic diversity between R5- and X4-genome specific trends. As expected, the V3 domain of the gp120 protein was enriched with statistically divergent positions. Statistically divergent positions were also identified in Tat amino acid sequences within the transactivation and TAR-binding domains, and in nucleotide positions throughout the LTR. We further analyzed LTR sequences for putative transcription factor binding sites using the JASPAR transcription factor binding profile database and found several putative differences in transcription factor binding sites between R5 and X4 HIV-1 genomes, specifically identifying the C/EBP sites I and II, and Sp site III to differ with respect to sequence configuration for R5 and X4 LTRs. Conclusion These observations support the hypothesis that co-receptor utilization coincides with specific genetic signatures in HIV-1 Tat and the LTR, likely due to differing transcriptional regulatory mechanisms and selective pressures applied within specific cellular targets during the course of productive HIV-1 infection.

Vpr is an HIV-1 accessory protein that plays numerous roles during viral replication, and some of which are cell type dependent. To test the hypothesis that HIV-1 tropism extends beyond the envelope into the vpr gene, studies were performed to identify the associations between coreceptor usage and Vpr variation in HIV-1-infected patients. Colinear HIV-1 Env-V3 and Vpr amino acid sequences were obtained from the LANL HIV-1 sequence database and from well-suppressed patients in the Drexel/Temple Medicine CNS AIDS Research and Eradication Study (CARES) Cohort. Genotypic classification of Env-V3 sequences as X4 (CXCR4-utilizing) or R5 (CCR5-utilizing) was used to group colinear Vpr sequences. To reveal the sequences associated with a specific coreceptor usage genotype, Vpr amino acid sequences were assessed for amino acid diversity and Jensen-Shannon divergence between the two groups. Five amino acid alphabets were used to comprehensively examine the impact of amino acid substitutions involving side chains with similar physiochemical properties. Positions 36, 37, 41, 89, and 96 of Vpr were characterized by statistically significant divergence across multiple alphabets when X4 and R5 sequence groups were compared. In addition, consensus amino acid switches were found at positions 37 and 41 in comparisons of the R5 and X4 sequence populations. These results suggest an evolutionary link between Vpr and gp120 in HIV-1-infected patients.

The low fidelity of HIV-1 reverse transcriptase results in a high frequency of mutation and generates extensive genetic variation both within and between infected individuals. Importantly, many of these mutations can contribute to disease presentation and progression despite the overall efficacy of antiretroviral therapy (ART). Of particular interest to the studies herein was the relationship between HIV-1 genetic diversity, tropism, and the development of HIV-1 associated neurocognitive disorders (HAND), a comorbidity of increasing prevalence in the HIV-1 infected population. These studies explored two interrelated avenues of research: developing a strategy to identify genetic signatures of HIV-1 co-receptor utilization during entry that extends beyond proteins forming the viral envelope, and utilizing the genetic variation of the accessory HIV-1 protein Tat to assess the likelihood of neurocognitive impairment. Jensen-Shannon divergence was applied to compare multiple sequence alignments and identify signature positions that are most distinct between X4- and R5-utilizing HIV-1 genotypes or quasispecies, for the accessory proteins Tat and Vpr. Additionally, within the HIV-1 long terminal repeat (LTR), predicted transcription factor binding sites demonstrated statistically different binding affinity scores for X4 and R5 LTR sequences. Finally, we developed a statistical learning approach utilizing regularized logistic regression for the prediction of neurocognitive impairment, integrating patient-derived HIV-1 genetic information as well as standard clinical measurements such as CD4+ T cell count and viral load. Here, we demonstrated that genetic-based models of neurocognitive impairment can outperform models utilizing standard clinical data based on the analysis of receiver operating characteristic (ROC) curves. This performance provides strong support that the integration of next generation sequencing data of HIV-1 Tat and other genes may yield improvements to clinical screening tools, as well as allow the statistical inference of genetic variants associated with neurocognitive impairment. These studies have shown that the naturally occurring genetic variation that exists throughout the genome associates with coreceptor usage and hence cell phenotype. This may play a role in disease pathogenesis as macrophages are classically implicated in bringing virus to the central nervous system (CNS) and infection of the T cell compartment is classically related to worsening of disease progression. In line with this thought, viral proteins made from infected cells in the CNS, like Tat, may play a role in CNS dysfunction leading to impairment. Taken together, viral sequences found in patients may have the ability to be used as a biomarker for disease progression and neurocognitive impairment.

Pancreatic cancer is an aggressive disease that typically presents late with poor patient outcomes. There is a pronounced medical need for early detection of pancreatic cancer, which can be addressed by identifying high-risk populations. Here we apply artificial intelligence (AI) methods to a dataset of 6 million patient records with 24,000 pancreatic cancer cases in the Danish National Patient Registry (DNPR) and, for comparison, a dataset of three million records with 3,900 pancreatic cancer cases in the United States Department of Veterans Affairs (US-VA) healthcare system. In contrast to existing methods that do not use temporal information, we explicitly train machine learning models on the time sequence of diseases in patient clinical histories and test the ability to predict cancer occurrence in time intervals of 3 to 60 months after risk assessment. For cancer occurrence within 36 months, the performance of the best model (AUROC=0.88, DNPR), trained and tested on disease trajectories, exceeds that of a model without longitudinal information (AUROC=0.85, DNPR). Performance decreases when disease events within a 3 month window before cancer diagnosis are excluded from training (AUROC[3m]=0.83). Independent training and testing on the US-VA dataset reaches comparable performance (AUROC=0.78, AUROC[3m]=0.76). These results raise the state-of-the-art level of performance of cancer risk prediction on real-world data sets and provide support for the design of prediction-surveillance programs based on risk assessment in a large population followed by affordable surveillance of a relatively small number of patients at highest risk. Use of AI on real-world clinical records has the potential to shift focus from treatment of late-stage to early-stage cancer, benefiting patients by improving lifespan and quality of life.Competing Interest StatementS.B. has ownership in Intomics A/S, Hoba Therapeutics Aps, Novo Nordisk A/S, Lundbeck A/S, ALK Abello and managing board memberships in Proscion A/S and Intomics A/S. B.M.W. notes grant funding from Celgene and Eli Lilly; consulting fees from BioLineRx, Celgene, and GRAIL. A.R. is a co-founder and equity holder of Celsius Therapeutics, an equity holder in Immunitas, and was an SAB member of ThermoFisher Scientific, Syros Pharmaceuticals, Neogene Therapeutics and Asimov until July 31, 2020. C.S. is on the science advisory board of Cytoreason LTD. From August 1, 2020, A.R. is an employee of Genentech.Footnotes* External validation was carried out on the US-VA dataset. Clinical relevance of the model was supported by new plots (relative risk by patients at risk).