Developing a Cancer Digital Twin: Supervised Metastases Detection From Consecutive Structured Radiology Reports.

The development of digital cancer twins relies on the capture of high-resolution representations of individual cancer patients throughout the course of their treatment. Our research aims to improve the detection of metastatic disease over time from structured radiology reports by exposing prediction models to historical information. We demonstrate that Natural language processing (NLP) can generate better weak labels for semi-supervised classification of computed tomography (CT) reports when it is exposed to consecutive reports through a patient's treatment history. Around 714,454 structured radiology reports from Memorial Sloan Kettering Cancer Center adhering to a standardized departmental structured template were used for model development with a subset of the reports included for validation. To develop the models, a subset of the reports was curated for ground-truth: 7,732 total reports in the lung metastases dataset from 867 individual patients; 2,777 reports in the liver metastases dataset from 315 patients; and 4,107 reports in the adrenal metastases dataset from 404 patients. We use NLP to extract and encode important features from the structured text reports, which are then used to develop, train, and validate models. Three models-a simple convolutional neural network (CNN), a CNN augmented with an attention layer, and a recurrent neural network (RNN)-were developed to classify the type of metastatic disease and validated against the ground truth labels. The models use features from consecutive structured text radiology reports of a patient to predict the presence of metastatic disease in the reports. A single-report model, previously developed to analyze one report instead of multiple past reports, is included and the results from all four models are compared based on accuracy, precision, recall, and F1-score. The best model is used to label all 714,454 reports to generate metastases maps. Our results suggest that NLP models can extract cancer progression patterns from multiple consecutive reports and predict the presence of metastatic disease in multiple organs with higher performance when compared with a single-report-based prediction. It demonstrates a promising automated approach to label large numbers of radiology reports without involving human experts in a time- and cost-effective manner and enables tracking of cancer progression over time.

Oncologic Outcomes Associated With MRI-detected Extramural Venous Invasion (mrEMVI) in Rectal Cancer: A Systematic Review and Meta-analysis.

BACKGROUND: The role of MRI-detected EMVI (mrEMVI) as a reliable prognostic factor in rectal cancer has been emphasized in recent years but this finding remains underreported by many institutions. OBJECTIVE: This review aimed to demonstrate the importance of pre- and post-treatment MRI-detected EMVI as independent prognostic factors of adverse oncologic outcomes in patients undergoing neoadjuvant therapy followed by total mesorectal excision. METHODS: This review was designed using the PRISMA guidelines. The following electronic databases were searched from January 2002 to January 2020: CENTRAL, Ovid MEDLINE, PubMed, and Ovid Embase. Main outcomes included DFS and overall survival (OS). Other outcomes of interest comprised positive resection margin and synchronous metastases. RESULTS: Seventeen studies involving a total of 3821 patients were included for data synthesis. For preneoadjuvant treatment mrEMVI, pooled hazard ratio (HR) estimate for DFS was 2.30 (95% confidence intervals (CI) 1.54-3.44) for higher recurrence in mrEMVI-positive patients. mrEMVI-positive patients were found to have a lower OS with a pooled HR of 1.68 (95%CI 1.27-2.22). Pooled risk ratio for synchronous metastasis was 4.11 (95%CI 2.80-6.02) for mrEMVI-positivity. For postneoadjuvant treatment EMVI (ymrEMVI), positive status showed a lower DFS with a pooled HR of 2.04 (95%CI 1.55-2.69). Risk ratio of having a positive resection margin status was 2.95 (95%CI 1.75-4.98) for ymrEMVI-positive patients. CONCLUSIONS: This review showed that oncologic outcomes are significantly worse for both pre- and post-neoadjuvant treatment mrEMVI-positive patients. MRI-detected EMVI should be consistently reported in rectal cancer staging and may provide guidance for the targeted use of additional systemic therapy.

Long-term outcomes after seton placement for perianal fistulas with and without Crohn's disease.

AIM: Perianal sepsis in Crohn's disease (CD) fistulas is managed with antibiotics and surgical drainage; a noncutting seton is used for an identified transsphincteric fistula tract. The optimal management following seton placement for initial control of perianal sepsis remains to be determined. Our main aim was to assess the success rates of curative surgery, seton removal or long-term indwelling seton in patients with and without CD. METHOD: This was a retrospective cohort of consecutive patients with a perianal fistula treated with a noncutting seton between 2010 and 2019, including 83 CD patients and 94 patients without CD. Initial control of symptomatic perianal infection with a seton and subsequent healing and reintervention rates were compared between the three postseton management strategies. RESULTS: A total of 177 patients, 61% male and 83.1% with complex fistulas, were followed for a median of 23 months (interquartile range 11-40 months). Immunomodulatory treatment was used in 90.4% of CD patients after seton placement. Good initial control of perianal infection was achieved with a seton in CD and non-CD patients, at 92.9% and 96.7%, respectively (p = 0.11). Overall fistula healing or control for CD and non-CD patients was, respectively, 64% and 86% (p = 0.1) after curative surgery, 49% and 71% after seton removal (p = 0.21) and 58% and 50% with long-term seton placement (p = 0.72). Overall reintervention for recurrence was 83% in CD versus 53.1% in non-CD patients during the follow-up period (p = 0.002). CONCLUSION: Definitive surgery was possible in only a minority of CD patients. Long-term seton management was an effective option in patients with CD with acceptable improvement and recurrence rates.

Specific inhibition by hGRB10zeta of insulin-induced glycogen synthase activation: evidence for a novel signaling pathway.

Grb10 is a member of a family of adapter proteins that binds to tyrosine-phosphorylated receptors including the insulin receptor kinase (IRK). In this study recombinant adenovirus was used to over-express hGrb10zeta, a new Grb10 isoform, in primary rat hepatocytes and the consequences for insulin signaling were evaluated. Over-expression of hGrb10zeta resulted in 50% inhibition of insulin-stimulated IRK autophosphorylation and activation. Analysis of downstream events showed that hGrb10zeta over-expression specifically inhibits insulin-stimulated glycogen synthase (GS) activity and glycogen synthesis without affecting insulin-induced IRS1/2 phosphorylation, PI3-kinase activation, insulin like growth factor binding protein-1 (IGFBP-1) mRNA expression, and ERK1/2 MAP kinase activity. The classical pathway from PI3-kinase through Akt-PKB/GSK-3 leading to GS activation by insulin was also not affected by hGrb10zeta over-expression. These results indicate that hGrb10zeta inhibits a novel and presently unidentified insulin signaling pathway leading to GS activation in liver.

Interaction of the Grb10 adapter protein with the Raf1 and MEK1 kinases.

Grb10 and its close homologues Grb7 and Grb14, belong to a family of adapter proteins characterized by a proline-rich region, a central PH domain, and a carboxyl-terminal Src homology 2 (SH2) domain. Their interaction with a variety of activated tyrosine kinase receptors is well documented, but their actual function remains a mystery. The Grb10 SH2 domain was isolated from a two-hybrid screen using the MEK1 kinase as a bait. We show that this unusual SH2 domain interacts, in a phosphotyrosine-independent manner, with both the Raf1 and MEK1 kinases. Mutation of the MEK1 Thr-386 residue, which is phosphorylated by mitogen-activated protein kinase in vitro, reduces binding to Grb10 in a two-hybrid assay. Interaction of Grb10 with Raf1 is constitutive, while interaction between Grb10 and MEK1 needs insulin treatment of the cells and follows mitogen-activated protein kinase activation. Random mutagenesis of the SH2 domain demonstrated that the Arg-betaB5 and Asp-EF2 residues are necessary for binding to the epidermal growth factor and insulin receptors as well as to the two kinases. In addition, we show that a mutation in Ser-betaB7 affects binding only to the receptors, while a mutation in Thr-betaC5 abrogates binding only to MEK1. Finally, transfection of Grb10 genes with specific mutations in their SH2 domains induces apoptosis in HTC-IR and COS-7 cells. These effects can be competed by co-expression of the wild type protein, suggesting that these mutants act by sequestering necessary signaling components.

Gorilla and orangutan c-myc nucleotide sequences: inference on hominoid phylogeny.

The nucleotide sequences of the gorilla and orangutan myc loci have been determined by the dideoxy nucleotide method. As previously observed in the human and chimpanzee sequences, an open reading frame (ORF) of 188 codons overlapping exon 1 could be deduced from the gorilla sequence. However, no such ORF appeared in the orangutan sequence. The two sequences were aligned with those of human and chimpanzee as hominoids and of gibbon and marmoset as outgroups of hominoids. The branching order in the evolution of primates was inferred from these data by different methods: maximum parsimony and neighbor-joining. Our results support the view that the gorilla lineage branched off before the human and chimpanzee diverged and strengthen the hypothesis that chimpanzee and gorilla are more related to human than is orangutan.

The molecular cloning and sequence of the common marmoset interferon-gamma (MaIFN-gamma) gene.

A DNA fragment covering marmoset interferon gamma (MaIFN-gamma) was cloned from the DNA of peripheral blood leucocytes, sequenced and compared to its human IFN-gamma counterpart. The two nucleotide sequences were found to be highly homologous (90.3%). The position of the exons are directly comparable with those of the human IFN-gamma gene and follows the pattern of structural conservation found elsewhere. The present work makes available recombinant MaIFN-gamma for animal experiments.

Gibbon and marmoset c-myc nucleotide sequences.

The nucleotide sequences of the gibbon and marmoset myc loci have been determined by the dideoxy ribomethod. The number of mutations which occurred during evolution and the branches affected were deduced according to the principle of maximum parsimony, from a comparison with known mammal sequences. As previously observed for the human and chimpanzee myc genes, an Alu repeat belonging to subclass III was observed in the second intron of the gibbon myc gene. In contrast, no such element was found in the marmoset gene. Alignment of the Myc amino acid (aa) sequences provided clues for detecting which aa or which protein regions have been more heavily mutated. Conversely, some regions remained free of mutations and remained unchanged from mouse to human, most probably in connection with some important embedded property(ies). An intriguing feature of the human Myc protein is duplication of 50 aa out of 439. Strikingly, most of these aa remain unchanged in mouse, rat, cat, marmoset, gibbon, chimpanzee and human.