Electrocardiogram-based mortality prediction in patients with COVID-19 using machine learning.

BACKGROUND AND PURPOSE: The electrocardiogram (ECG) is frequently obtained in the work-up of COVID-19 patients. So far, no study has evaluated whether ECG-based machine learning models have added value to predict in-hospital mortality specifically in COVID-19 patients. METHODS: Using data from the CAPACITY-COVID registry, we studied 882 patients admitted with COVID-19 across seven hospitals in the Netherlands. Raw format 12-lead ECGs recorded within 72 h of admission were studied. With data from five hospitals (n = 634), three models were developed: (a) a logistic regression baseline model using age and sex, (b) a least absolute shrinkage and selection operator (LASSO) model using age, sex and human annotated ECG features, and (c) a pre-trained deep neural network (DNN) using age, sex and the raw ECG waveforms. Data from two hospitals (n = 248) was used for external validation. RESULTS: Performances for models a, b and c were comparable with an area under the receiver operating curve of 0.73 (95% confidence interval [CI] 0.65-0.79), 0.76 (95% CI 0.68-0.82) and 0.77 (95% CI 0.70-0.83) respectively. Predictors of mortality in the LASSO model were age, low QRS voltage, ST depression, premature atrial complexes, sex, increased ventricular rate, and right bundle branch block. CONCLUSION: This study shows that the ECG-based prediction models could be helpful for the initial risk stratification of patients diagnosed with COVID-19, and that several ECG abnormalities are associated with in-hospital all-cause mortality of COVID-19 patients. Moreover, this proof-of-principle study shows that the use of pre-trained DNNs for ECG analysis does not underperform compared with time-consuming manual annotation of ECG features.

Targeted next-generation sequencing of commonly mutated genes in esophageal adenocarcinoma patients with long-term survival.

Survival of patients with esophageal adenocarcinoma remains poor and individual differences in prognosis remain unexplained. This study investigated whether gene mutations can explain why patients with high-risk (pT3-4, pN+) esophageal adenocarcinoma survive past 5 years after esophagectomy. Six long-term survivors (LTS) (≥5 years survival without recurrence) and six short-term survivors (STS) (<2 years survival due to recurrence) who underwent resection without neoadjuvant therapy for high-risk esophageal adenocarcinoma were included. Targeted next-generation sequencing of 16 genes related to esophageal adenocarcinoma was performed. Mutations were compared between the LTS and STS and described in comparison with literature. A total of 48 mutations in 10 genes were identified. In the LTS, the median number of mutated genes per sample was 5 (range: 0-5) and the samples together harbored 22 mutations in 8 genes: APC (n = 1), CDH11 (n = 2), CDKN2A (n = 2), FAT4 (n = 5), KRAS (n = 1), PTPRD (n = 1), TLR4 (n = 8), and TP53 (n = 2). The median number of mutated genes per sample in the STS was 4 (range: 1-8) and in total 26 mutations were found in six genes: CDH11 (n = 5), FAT4 (n = 7), SMAD4 (n = 1), SMARCA4 (n = 1), TLR4 (n = 7), and TP53 (n = 5). CDH11, CDKN2A, FAT4, TLR4, and TP53 were mutated in at least 2 LTS or STS, exceeding mutation rates in literature. Mutations across the LTS and STS were found in 10 of the 16 genes. The results warrant future studies to investigate a larger range of genes in a larger sample size. This may result in a panel with prognostic genes, to predict individual prognosis and to select effective individualized therapy for patients with esophageal adenocarcinoma.

Radiofrequency ablation of the pancreas: two-week follow-up in a porcine model.

OBJECTIVE: To determine the impact of radiofrequency ablation (RFA) in pancreas after two-week follow-up. BACKGROUND: RFA is a novel treatment strategy in patients with unresectable locally advanced pancreatic cancer. The histological effect and risk of postoperative complications has not been systematically addressed in an in-vivo animal model. METHODS: In a porcine model (n = 6), RFA was performed via laparotomy with previously determined optimal settings using a bipolar probe with 30 mm active length, at 30 W until a total energy of 15 KJ was administered. The probe was inserted in the pancreas at 10 mm distance from duodenum and portomesenteric vessels (PMV). RFA nearby duodenum was performed under continuous duodenal cooling using 100 ml/min saline of 5 °C. During two weeks the clinical condition was evaluated daily including blood analyses. After two weeks, total pancreatoduodenectomy was performed and the obtained tissue histopathologically assessed. RESULTS: No mortality occurred during or after RFA. Two animals had a serum amylase increase more than threefold the pre-intervention value without clinical manifestations. Histopathologic assessment showed total ablation within the ablation zone, with loss of normal pancreatic acinar cell outlines and necrosis. In one animal, focal necrosis of duodenal submucosa was seen and in another animal focal fibrosis in the muscular layer of the superior mesenteric vein without clinical manifestations. CONCLUSION: No major morbidity and no mortality was seen during a period of two weeks after RFA with previously validated RFA settings including duodenal cooling and 10 mm distance to PMV. Future clinical studies should confirm safety of RFA using the settings established here.

Very high risk of cancer in familial Peutz-Jeghers syndrome.

BACKGROUND & AIMS: The Peutz-Jeghers syndrome (PJS) is an autosomal dominant polyposis disorder with increased risk of multiple cancers, but literature estimates of risk vary. METHODS: We performed an individual patient meta-analysis to determine the relative risk (RR) of cancer in patients with PJS compared with the general population based on 210 individuals described in 6 publications. RESULTS: For patients with PJS, the RR for all cancers was 15.2 (95% confidence limits [CL], 2, 19). A statistically significant increase of RR was noted for esophagus (57; CL, 2.5, 557), stomach (213; CL, 96, 368), small intestine (520; CL, 220, 1306), colon (84; CL, 47, 137), pancreas (132; CL, 44, 261), lung (17.0; CL, 5.4, 39), breast (15.2; CL, 7.6, 27), uterus (16.0; CL, 1.9, 56), ovary (27; CL, 7.3, 68), but not testicular or cervical malignancies. Cumulative risk for all cancer was 93% from age 15 to 64 years old. CONCLUSIONS: Patients with PJS are at very high relative and absolute risk for gastrointestinal and nongastrointestinal cancers.

Progress in cancer genetics: lessons from pancreatic cancer.

BACKGROUND: In the near future advances in the molecular basis of cancer are expected to facilitate cancer diagnosis, to rationalize treatment, to facilitate screening, and to identify individuals requiring cancer prevention strategies. METHODS: The literature was reviewed concerning the genetic alterations that contribute to pancreatic cancer development. RESULTS: Virtually all pancreatic cancers have inactivation of the p16 pathway, and the majority inactivate the TGF beta/DPC4 and p53 tumor-suppressive pathways. Pancreatic cancers with mismatch repair deficiency have a characteristic histology and may have an improved prognosis. The recently discovered tumor suppressor genes, ALK-5, MKK4, and STK11 (the gene responsible for Peutz-Jeghers syndrome) are all targeted for mutation in a small proportion of sporadic pancreatic cancers. Germline mutations of the BRCA2 gene are present in 5-10% of patients with pancreatic cancer. Typically such patients do not have a family history of pancreatic cancer and are mistaken as patients with sporadic disease. Five to 10% of patients with pancreatic cancer have first-degree relatives that will develop pancreatic cancer. Some such families also have a family history of melanoma and harbor germline p16 mutations. However, the gene(s) responsible for much of the inherited predisposition to pancreatic cancer remain to be identified. CONCLUSION: Further advances in pancreatic cancer molecular genetics are needed to facilitate the development of molecular screening tests, to identify additional familial susceptibility genes, and to identify targets for rational therapeutic targeting.

The WAF1-mediated p53 growth-suppressor pathway is intact in the coronary arteries of heart transplant recipients.

It has been suggested that the interaction of cytomegalovirus (CMV) with the p53 tumor suppressor gene product plays a role in the development of coronary artery restenosis after angioplasty. CMV nucleic acids have been observed in the coronary arteries of allografted hearts, suggesting a possible role for the interaction of CMV with p53 in the development of accelerated graft arteriosclerosis in transplant recipients. Formalin-fixed, paraffin-embedded sections of coronary arteries from 19 transplanted hearts were immunostained for the p53 gene product using Target Unmasking Fluid (TUF)-mediated immunohistochemistry and the anti-p53 antibodies CM1 and DO7. Fresh-frozen sections of coronary arteries were also available from six of the 19 hearts, and these fresh-frozen sections were immunostained for the p53 gene product with the DO7 antibody and for WAF1 using the anti-WAF1 antibody EA10. Focal and weak staining for p53 was observed in smooth muscle and endothelial cells in two of 19 vessels, whereas the remaining 17 did not stain. CMV nucleic acids were previously shown in six of 13 of these hearts by in situ hybridization. The fresh-frozen sections of coronary arteries also did not stain for p53, but the smooth muscle cells in these vessels did stain intensely for WAF1. These results suggest three possibilities: (1) CMV-p53 interactions are not important in the development of accelerated graft arteriosclerosis; or (2) there is an interaction, but it is transient and not detectable at the time points examined in this study; or (3) there is an interaction, but binding of CMV to p53 leads to accelerated degradation of p53, as occurs with HPV-E6. The expression of WAF1 further suggests that the WAF1-mediated antiproliferative signal is intact in these vessels.

Sulindac induced regression of colorectal adenomas in familial adenomatous polyposis: evaluation of predictive factors.

BACKGROUND: Sulindac, a non-steroidal anti-inflammatory drug, causes regression of colorectal adenomas in patients with familial adenomatous polyposis (FAP) but the response is variable. Specific clinical factors predictive of sulindac induced regression have not been studied. METHODS: 22 patients with FAP were given sulindac 150 mg orally twice a day. Polyp number and size were determined before treatment and at three months. The relation of nine clinical factors to polyp regression (per cent of baseline polyp number after treatment) was evaluated by univariate and multivariate analysis. RESULTS: After three months of sulindac, polyp number had decreased to 45 per cent of baseline and polyp size to 50 per cent of baseline (p < 0.001 and p < 0.01, respectively). Univariate analysis showed greater polyp regression in older patients (p = 0.004), those with previous colectomy and ileorectal anastomosis (p = 0.001), and patients without identifiable mutation of the APC gene responsible for FAP (p = 0.05). With multivariate regression analysis, response to sulindac treatment was associated with previous subtotal colectomy. CONCLUSIONS: Sulindac treatment seems effective in producing regression of colorectal adenomas of FAP patients with previous subtotal colectomy regardless of baseline polyp number and size. Changed sulindac metabolism, reduced area of the target mucosa, or changed epithelial characteristics after ileorectal anastomosis may explain these findings.

Nasopharyngeal angiofibroma in patients with familial adenomatous polyposis.

Four patients with nasopharyngeal angiofibroma and familial adenomatous polyposis are reported here. Nasopharyngeal angiofibroma was 25 times more frequent in our patient population with familial adenomatous polyposis than in an age-matched hospital population. The association of these two rare conditions suggests that nasopharyngeal angiofibroma is an extracolonic manifestation of adenomatous polyposis. In addition, somatic mutation of the adenomatous polyposis coli gene, which causes adenomatous polyposis when mutated in the germline, could play a role in the pathogenesis of sporadic nasopharyngeal angiofibroma.

Hepatoblastoma, pigmented ocular fundus lesions and jaw lesions in Gardner syndrome.

Hepatoblastoma is a rare neoplasm of infants and children only recently documented in association with hereditary adenomatous polyposis of the colon [Kingston et al., 1983]. We report four children with hepatoblastoma from four unrelated families with Gardner syndrome (GS). One child, now 19 years old, survived after a resection of a hepatoblastoma in infancy and recently was found to have GS. He has an associated odontoma and pigmented ocular fundus lesions, both of which have been shown to be clinical markers of GS. Many individuals in these four GS families, both affected and at risk, have osteomatous jaw lesions and pigmented ocular fundus lesions. A search for colonic polyps should be made in families of infants and children with hepatoblastoma. If the child survives, he or she should be monitored for the later appearance of colonic polyps. The finding of jaw lesions and/or pigmented ocular fundus lesions in relatives at risk are indications of the possible presence of the GS gene.