DNA methylation and gene expression as determinants of genome-wide cell-free DNA fragmentation.

Circulating cell-free DNA (cfDNA) is emerging as an avenue for cancer detection, but the characteristics of cfDNA fragmentation in the blood are poorly understood. We evaluate the effect of DNA methylation and gene expression on genome-wide cfDNA fragmentation through analysis of 969 individuals. cfDNA fragment ends more frequently contained CCs or CGs, and fragments ending with CGs or CCGs are enriched or depleted, respectively, at methylated CpG positions. Higher levels and larger sizes of cfDNA fragments are associated with CpG methylation and reduced gene expression. These effects are validated in mice with isogenic tumors with or without the mutant IDH1, and are associated with genome-wide changes in cfDNA fragmentation in patients with cancer. Tumor-related hypomethylation and increased gene expression are associated with decrease in cfDNA fragment size that may explain smaller cfDNA fragments in human cancers. These results provide a connection between epigenetic changes and cfDNA fragmentation with implications for disease detection.

Genome-wide repeat landscapes in cancer and cell-free DNA.

Genetic changes in repetitive sequences are a hallmark of cancer and other diseases, but characterizing these has been challenging using standard sequencing approaches. We developed a de novo kmer finding approach, called ARTEMIS (Analysis of RepeaT EleMents in dISease), to identify repeat elements from whole-genome sequencing. Using this method, we analyzed 1.2 billion kmers in 2837 tissue and plasma samples from 1975 patients, including those with lung, breast, colorectal, ovarian, liver, gastric, head and neck, bladder, cervical, thyroid, or prostate cancer. We identified tumor-specific changes in these patients in 1280 repeat element types from the LINE, SINE, LTR, transposable element, and human satellite families. These included changes to known repeats and 820 elements that were not previously known to be altered in human cancer. Repeat elements were enriched in regions of driver genes, and their representation was altered by structural changes and epigenetic states. Machine learning analyses of genome-wide repeat landscapes and fragmentation profiles in cfDNA detected patients with early-stage lung or liver cancer in cross-validated and externally validated cohorts. In addition, these repeat landscapes could be used to noninvasively identify the tissue of origin of tumors. These analyses reveal widespread changes in repeat landscapes of human cancers and provide an approach for their detection and characterization that could benefit early detection and disease monitoring of patients with cancer.

Single-molecule genome-wide mutation profiles of cell-free DNA for non-invasive detection of cancer.

Somatic mutations are a hallmark of tumorigenesis and may be useful for non-invasive diagnosis of cancer. We analyzed whole-genome sequencing data from 2,511 individuals in the Pan-Cancer Analysis of Whole Genomes (PCAWG) study as well as 489 individuals from four prospective cohorts and found distinct regional mutation type-specific frequencies in tissue and cell-free DNA from patients with cancer that were associated with replication timing and other chromatin features. A machine-learning model using genome-wide mutational profiles combined with other features and followed by CT imaging detected >90% of patients with lung cancer, including those with stage I and II disease. The fixed model was validated in an independent cohort, detected patients with cancer earlier than standard approaches and could be used to monitor response to therapy. This approach lays the groundwork for non-invasive cancer detection using genome-wide mutation features that may facilitate cancer screening and monitoring.

H2Bub1 loss is an early contributor to clear cell ovarian cancer progression.

Epigenetic aberrations, including posttranslational modifications of core histones, are major contributors to cancer. Here, we define the status of histone H2B monoubiquitylation (H2Bub1) in clear cell ovarian carcinoma (CCOC), low-grade serous carcinoma, and endometrioid carcinomas. We report that clear cell carcinomas exhibited profound loss, with nearly all cases showing low or negative H2Bub1 expression. Moreover, we found that H2Bub1 loss occurred in endometriosis and atypical endometriosis, which are established precursors to CCOCs. To examine whether dysregulation of a specific E3 ligase contributes to the loss of H2Bub1, we explored expression of ring finger protein 40 (RNF40), ARID1A, and UBR7 in the same case cohort. Loss of RNF40 was significantly and profoundly correlated with loss of H2Bub1. Using genome-wide DNA methylation profiles of 230 patients with CCOC, we identified hypermethylation of RNF40 in CCOC as a likely mechanism underlying the loss of H2Bub1. Finally, we demonstrated that H2Bub1 depletion promoted cell proliferation and clonogenicity in an endometriosis cell line. Collectively, our results indicate that H2Bub1 plays a tumor-suppressive role in CCOCs and that its loss contributes to disease progression.

Detecting Liver Cancer Using Cell-Free DNA Fragmentomes.

Liver cancer is a major cause of cancer mortality worldwide. Screening individuals at high risk, including those with cirrhosis and viral hepatitis, provides an avenue for improved survival, but current screening methods are inadequate. In this study, we used whole-genome cell-free DNA (cfDNA) fragmentome analyses to evaluate 724 individuals from the United States, the European Union, or Hong Kong with hepatocellular carcinoma (HCC) or who were at average or high-risk for HCC. Using a machine learning model that incorporated multifeature fragmentome data, the sensitivity for detecting cancer was 88% in an average-risk population at 98% specificity and 85% among high-risk individuals at 80% specificity. We validated these results in an independent population. cfDNA fragmentation changes reflected genomic and chromatin changes in liver cancer, including from transcription factor binding sites. These findings provide a biological basis for changes in cfDNA fragmentation in patients with liver cancer and provide an accessible approach for noninvasive cancer detection. SIGNIFICANCE: There is a great need for accessible and sensitive screening approaches for HCC worldwide. We have developed an approach for examining genome-wide cfDNA fragmentation features to provide a high-performing and cost-effective approach for liver cancer detection. See related commentary Rolfo and Russo, p. 532. This article is highlighted in the In This Issue feature, p. 517.

Age and Racial Disparities in Telemedicine Utilization in an Academic Orthopedic Surgery Department.

Introduction: The COVID-19 pandemic has highlighted significant racial and age-related health disparities. In response to pandemic-related restrictions, orthopedic surgery departments have expanded telemedicine use. We analyzed data from a tertiary care institute during the pandemic to understand potential racial and age-based disparities in access to care and telemedicine utilization. Materials and Methods: Data on patient race and age, and numbers of telemedicine visits, in-person office visits, and types of telemedicine were extracted for time periods during and preceding the pandemic. We calculated odds ratios for visit occurrence and type across race and age groups. Results: Patients ages 27-54 were 1.3 (95% confidence interval [CI] 1.1-1.4, p < 0.01) and 1.2 (95% CI 1.0-1.3, p < 0.05) times more likely to be seen than patients <27 during the pandemic, versus the 2019 and 2020 controls. Patients 54-82 were 1.3 (95% CI 1.1-1.5, p < 0.001) times more likely to be seen than patients <27 during the pandemic versus the 2019 control. Patients 27-54, 54-82, and 82+, respectively, were 3.3 (95% CI 2.6-4.2, p < 1e-20), 3.5 (95% CI 2.8-4.4, p < 1e-24), and 1.9 (95% CI 1.1-3.4, p < 0.05) times more likely to be seen by telemedicine than patients <27. Among pandemic telemedicine appointments, Black patients were 1.5 (95% CI 1.2-1.9, p < 1e-3) times more likely to be seen by audio-only telemedicine than White patients, as compared with video telemedicine. Conclusions: Telemedicine access barriers must be reduced to ensure that disparities during the pandemic do not persist.

Factors Driving Rapid Adoption of Telemedicine in an Academic Orthopedic Surgery Department.

Introduction: With the COVID-19 epidemic ever-expanding, nonemergent access to health care resources has been reduced to decrease the exposure for patients and health care providers. Alternatives to in-office outpatient medical evaluations are necessary. We aimed to analyze how quickly orthopedic surgery providers at a large academic institution adopted telemedicine, and identify any factors that were associated with earlier or "faster" telemedicine adoption. Methods: We analyzed the telemedicine activity of 39 providers within the Department of Orthopedic Surgery between March 16, 2020, and May 30, 2020, and constructed logistic regression models to identify characteristics with significant association to earlier or faster telemedicine adoption. Results: No significant predictors of percentage of visits conducted via telemedicine were found. However, increased experience and practice at multiple locations was associated with slower telemedicine adoption time, while Professor level academic rank was associated with a faster time to achieving 10% of pre-COVID visit volumes via telemedicine. Higher pre-COVID visit volumes were also significantly associated with faster telemedicine adoption. Demographic factors, including, age, gender, practice locations, academic degrees, pediatric specialty, and use of physician assistants/nurse practitioners, were not found to have significant associations with telemedicine use. Conclusions: These results indicate that telemedicine has an important role to play within academic orthopedic surgery practices, with a wide and diverse range of orthopedic surgery providers choosing to utilize it during the COVID-19 pandemic. Given the rapid expansion and urgency driving the adoption of telemedicine, these results illustrate the importance of considering provider-side characteristics in ensuring that providers are well equipped to utilize telemedicine.

SWIFT: A deep learning approach to prediction of hypoxemic events in critically-Ill patients using SpO2 waveform prediction.

Hypoxemia is a significant driver of mortality and poor clinical outcomes in conditions such as brain injury and cardiac arrest in critically ill patients, including COVID-19 patients. Given the host of negative clinical outcomes attributed to hypoxemia, identifying patients likely to experience hypoxemia would offer valuable opportunities for early and thus more effective intervention. We present SWIFT (SpO2 Waveform ICU Forecasting Technique), a deep learning model that predicts blood oxygen saturation (SpO2) waveforms 5 and 30 minutes in the future using only prior SpO2 values as inputs. When tested on novel data, SWIFT predicts more than 80% and 60% of hypoxemic events in critically ill and COVID-19 patients, respectively. SWIFT also predicts SpO2 waveforms with average MSE below .0007. SWIFT predicts both occurrence and magnitude of potential hypoxemic events 30 minutes in the future, allowing it to be used to inform clinical interventions, patient triaging, and optimal resource allocation. SWIFT may be used in clinical decision support systems to inform the management of critically ill patients during the COVID-19 pandemic and beyond.

Detection and characterization of lung cancer using cell-free DNA fragmentomes.

Non-invasive approaches for cell-free DNA (cfDNA) assessment provide an opportunity for cancer detection and intervention. Here, we use a machine learning model for detecting tumor-derived cfDNA through genome-wide analyses of cfDNA fragmentation in a prospective study of 365 individuals at risk for lung cancer. We validate the cancer detection model using an independent cohort of 385 non-cancer individuals and 46 lung cancer patients. Combining fragmentation features, clinical risk factors, and CEA levels, followed by CT imaging, detected 94% of patients with cancer across stages and subtypes, including 91% of stage I/II and 96% of stage III/IV, at 80% specificity. Genome-wide fragmentation profiles across ~13,000 ASCL1 transcription factor binding sites distinguished individuals with small cell lung cancer from those with non-small cell lung cancer with high accuracy (AUC = 0.98). A higher fragmentation score represented an independent prognostic indicator of survival. This approach provides a facile avenue for non-invasive detection of lung cancer.

SWIFT: A Deep Learning Approach to Prediction of Hypoxemic Events in Critically-Ill Patients Using SpO 2 Waveform Prediction.

Hypoxemia is a significant driver of mortality and poor clinical outcomes in conditions such as brain injury and cardiac arrest in critically ill patients, including COVID-19 patients. Given the host of negative clinical outcomes attributed to hypoxemia, identifying patients likely to experience hypoxemia would offer valuable opportunities for early and thus more effective intervention. We present SWIFT (SpO 2 W aveform I CU F orecasting T echnique), a deep learning model that predicts blood oxygen saturation (SpO 2 ) waveforms 5 and 30 minutes in the future using only prior SpO 2 values as inputs. When tested on novel data, SWIFT predicts more than 80% and 60% of hypoxemic events in critically ill and COVID-19 patients, respectively. SWIFT also predicts SpO 2 waveforms with average MSE below .0007. SWIFT provides information on both occurrence and magnitude of potential hypoxemic events 30 minutes in advance, allowing it to be used to inform clinical interventions, patient triaging, and optimal resource allocation. SWIFT may be used in clinical decision support systems to inform the management of critically ill patients during the COVID-19 pandemic and beyond.

A natural language processing and deep learning approach to identify child abuse from pediatric electronic medical records.

Child physical abuse is a leading cause of traumatic injury and death in children. In 2017, child abuse was responsible for 1688 fatalities in the United States, of 3.5 million children referred to Child Protection Services and 674,000 substantiated victims. While large referral hospitals maintain teams trained in Child Abuse Pediatrics, smaller community hospitals often do not have such dedicated resources to evaluate patients for potential abuse. Moreover, identification of abuse has a low margin of error, as false positive identifications lead to unwarranted separations, while false negatives allow dangerous situations to continue. This context makes the consistent detection of and response to abuse difficult, particularly given subtle signs in young, non-verbal patients. Here, we describe the development of artificial intelligence algorithms that use unstructured free-text in the electronic medical record-including notes from physicians, nurses, and social workers-to identify children who are suspected victims of physical abuse. Importantly, only the notes from time of first encounter (e.g.: birth, routine visit, sickness) to the last record before child protection team involvement were used. This allowed us to develop an algorithm using only information available prior to referral to the specialized child protection team. The study was performed in a multi-center referral pediatric hospital on patients screened for abuse within five different locations between 2015 and 2019. Of 1123 patients, 867 records were available after data cleaning and processing, and 55% were abuse-positive as determined by a multi-disciplinary team of clinical professionals. These electronic medical records were encoded with three natural language processing (NLP) algorithms-Bag of Words (BOW), Word Embeddings (WE), and Rules-Based (RB)-and used to train multiple neural network architectures. The BOW and WE encodings utilize the full free-text, while RB selects crucial phrases as identified by physicians. The best architecture was selected by average classification accuracy for the best performing model from each train-test split of a cross-validation experiment. Natural language processing coupled with neural networks detected cases of likely child abuse using only information available to clinicians prior to child protection team referral with average accuracy of 0.90±0.02 and average area under the receiver operator characteristic curve (ROC-AUC) 0.93±0.02 for the best performing Bag of Words models. The best performing rules-based models achieved average accuracy of 0.77±0.04 and average ROC-AUC 0.81±0.05, while a Word Embeddings strategy was severely limited by lack of representative embeddings. Importantly, the best performing model had a false positive rate of 8%, as compared to rates of 20% or higher in previously reported studies. This artificial intelligence approach can help screen patients for whom an abuse concern exists and streamline the identification of patients who may benefit from referral to a child protection team. Furthermore, this approach could be applied to develop computer-aided-diagnosis platforms for the challenging and often intractable problem of reliably identifying pediatric patients suffering from physical abuse.