The Association Between Oncology Outreach and Timely Treatment for Rural Patients with Breast Cancer: A Claims-Based Approach

BACKGROUND

Oncology outreach is a common strategy for increasing rural access to cancer care, where traveling oncologists commute across healthcare settings to extend specialized care. Examining the extent to which physician outreach is associated with timely treatment for rural patients is critical for informing outreach strategies.

METHODS

We identified a 100% fee-for-service sample of incident breast cancer patients from 2015 to 2020 Medicare claims and apportioned them into surgery and adjuvant therapy cohorts based on treatment history. We defined an outreach visit as the provision of care by a traveling oncologist at a clinic outside of their primary hospital service area. We used hierarchical logistic regression to examine the associations between patient receipt of preoperative care at an outreach visit (preoperative outreach) and > 60-day surgical delay, and patient receipt of postoperative care at an outreach visit (postoperative outreach) and > 60-day adjuvant delay.

RESULTS

We identified 30,337 rural-residing patients who received breast cancer surgery, of whom 4071 (13.4%) experienced surgical delay. Among surgical patients, 14,501 received adjuvant therapy, of whom 2943 (20.3%) experienced adjuvant delay. In adjusted analysis, we found that patient receipt of preoperative outreach was associated with reduced odds of surgical delay (odds ratio [OR] 0.75, 95% confidence interval [CI] 0.61-0.91); however, we found no association between patient receipt of postoperative outreach and adjuvant delay (OR 1.04, 95% CI 0.85-1.25).

CONCLUSIONS

Our findings indicate that preoperative outreach is protective against surgical delay. The traveling oncologists who enable such outreach may play an integral role in catalyzing the coordination and timeliness of patient-centered care.

Estimating the impact of physician risky-prescribing on the network structure underlying physician shared-patient relationships

Social network analysis and shared-patient physician networks have become effective ways of studying physician collaborations. Assortative mixing or "homophily" is the network phenomenon whereby the propensity for similar individuals to form ties is greater than for dissimilar individuals. Motivated by the public health concern of risky-prescribing among older patients in the United States, we develop network models and tests involving novel network measures to study whether there is evidence of geographic homophily in prescribing and deprescribing in the specific shared-patient network of physicians linked to the US state of Ohio in 2014. Evidence of homophily in risky-prescribing would imply that prescribing behaviors help shape physician networks and could inform interventions to reduce risky-prescribing (e.g., should interventions target groups of physicians or select physicians at random). Furthermore, if such effects varied depending on the structural features of a physician's position in the network (e.g., by whether or not they are involved in cliques - groups of actors that are fully connected to each other - such as closed triangles in the case of three actors), this would further strengthen the case for targeting of select physicians for interventions. Using accompanying Medicare Part D data, we converted patient longitudinal prescription receipts into novel measures of the intensity of each physician's risky-prescribing. Exponential random graph models were used to simultaneously estimate the importance of homophily in prescribing and deprescribing in the network beyond the characteristics of physician specialty (or other metadata) and network-derived features. In addition, novel network measures were introduced to allow homophily to be characterized in relation to specific triadic (three-actor) structural configurations in the network with associated non-parametric randomization tests to evaluate their statistical significance in the network against the null hypothesis of no such phenomena. We found physician homophily in prescribing and deprescribing in both the state-wide and multiple HRR sub-networks, and that the level of homophily varied across HRRs. We also found that physicians exhibited within-triad homophily in risky-prescribing, with the prevalence of homophilic triads significantly higher than expected by chance absent homophily. These results may explain why communities of prescribers emerge and evolve, helping to justify group-level prescriber interventions. The methodology could be applied to arbitrary shared-patient networks and even more generally to other kinds of network data that underlies other kinds of social phenomena.

Disparities in Access to Multidisciplinary Cancer Consultations and Treatment for Patients With Early-Stage Non-Small Cell Lung Cancer: A SEER-Medicare Analysis

PURPOSE

Disparities in access to a multidisciplinary cancer consultation (MDCc) persist, and the role of physician relationships remains understudied. This study examined the extent to which multilevel factors, including patient characteristics and patient-sharing network measures reflecting the structure of physician relationships, are associated with an MDCc and receipt of stereotactic body radiation therapy versus surgery among patients with early-stage non-small cell lung cancer (NSCLC).

METHODS AND MATERIALS

In this cross-sectional study, we analyzed Surveillance, Epidemiology, and End Results (SEER)-Medicare data for patients diagnosed with stage I-IIA NSCLC from 2016 to 2017. We assembled patient-sharing networks and identified cancer specialists who were locally unique for their specialty, herein referred to as linchpins. The proportion of linchpin cancer specialists for each hospital referral region (HRR) was calculated as a network-based measure of specialist scarcity. We used multilevel multinomial logistic regression to estimate associations between study variables and receipt of an MDCc and multilevel logistic regression to examine the relationship between patient receipt of an MDCc and initial treatment.

RESULTS

Our study included 6120 patients with stage I-IIA NSCLC, of whom 751 (12.3%) received an MDCc, 1729 (28.3%) consulted only a radiation oncologist, 2010 (32.8%) consulted only a surgeon, and 1630 (26.6%) consulted neither specialist within 2 months of diagnosis. Compared with patients residing in an HRR with a low proportion of linchpin surgeons, those residing in an HRR with a high proportion of linchpin surgeons had a 2.99 (95% CI, 1.87-4.78) greater relative risk of consulting only a radiation oncologist versus receiving an MDCc and a 2.70 (95% CI, 1.68-4.35) greater relative risk of consulting neither specialist versus receiving an MDCc. Patients who received an MDCc were 5.32 times (95% CI, 4.27-6.63) more likely to receive stereotactic body radiation therapy versus surgery.

CONCLUSIONS

Physician networks are associated with receipt of an MDCc and treatment, underscoring the potential for leveraging patient-sharing network analysis to improve access to lung cancer care.

Broadband Data Forensics: Spatiotemporal Variations of the Download/Upload Speed Metric Commonly Used to Evaluate Potential Telehealth Accessibility

Introduction: The complicated task of evaluating potential telehealth access begins with the metrics and supporting datasets that seek toevaluate the presence and durability of broadband connections in a community. Broadband download/upload speeds are one of the popular metrics used to measure potential telehealth access, which is critical to health equity. An understanding of the limitations of these measures is important for drawing conclusions about the reality of the digital divide in telehealth access. The objective of this study was to assess spatiotemporal variations in broadband download/upload speeds. Method: We analyzed a sample of data from the Speedtest Intelligence Portal provided through the Ookla for Good initiative. Results: We found that variation is inherent across the states of Vermont, New Hampshire, Louisiana, and Utah. Conclusions: The variation suggests that when single measures of download/upload speeds are used to evaluate telehealth accessibility they may be masking the true magnitude of the digital divide.

Rural-urban disparities in health care delivery for children with medical complexity and moderating effects of payer, disability, and community poverty

PURPOSE

Children with medical complexity (CMC) may be at increased risk of rural-urban disparities in health care delivery given their multifaceted health care needs, but these disparities are poorly understood. This study evaluated rural-urban disparities in health care delivery to CMC and determined whether Medicaid coverage, co-occurring disability, and community poverty modified the effects of rurality on care delivery.

METHODS

This retrospective cohort study of 2012-2017 all-payer claims data from Colorado, Massachusetts, and New Hampshire included CMC <18 years. Health care delivery measures (ambulatory clinic visits, emergency department visits, acute care hospitalizations, total hospital days, and receipt of post-acute care) were compared for rural- versus urban-residing CMC in multivariable regression models, following established methods to evaluate effect modification.

FINDINGS

Of 112,475 CMC, 7307 (6.5%) were rural residing and 105,168 (93.5%) were urban residing. A total of 68.9% had Medicaid coverage, 33.9% had a disability, and 39.7% lived in communities with >20% child poverty. In adjusted analyses, rural-residing CMC received significantly fewer ambulatory visits (risk ratio [RR] = 0.95, 95% confidence interval [CI]: 0.94-0.96), more emergency visits (RR = 1.12, 95% CI: 1.08-1.16), and fewer hospitalization days (RR = 0.90, 95% CI = 0.85-0.96). The estimated modification effects of rural residence by Medicaid coverage, disability, and community poverty were each statistically significant. Differences in the odds of having a hospitalization and receiving post-acute care did not persist after incorporating sociodemographic and clinical characteristics and interaction effects.

CONCLUSIONS

Rural- and urban-residing CMC differed in their receipt of health care, and Medicaid coverage, co-occurring disabilities, and community poverty modified several of these effects. These modifying effects should be considered in clinical and policy initiatives to ensure that such initiatives do not widen rural-urban disparities.

Characterizing the Traveling Oncology Workforce and Its Influence on Patient Travel Burden: A Claims-Based Approach

PURPOSE

Oncology outreach is a common strategy for extending cancer care to rural patients. However, a nationwide characterization of the traveling workforce that enables this outreach is lacking, and the extent to which outreach reduces travel burden for rural patients is unknown.

METHODS

This cross-sectional study analyzed a rural (nonurban) subset of a 100% fee-for-service sample of 355,139 Medicare beneficiaries with incident breast, colorectal, and lung cancers. Surgical, medical, and radiation oncologists were linked to patients using Part B claims, and traveling oncologists were identified by observing hospital service area (HSA) transition patterns. We defined oncology outreach as the provision of cancer care by a traveling oncologist outside of their primary HSA. We used hierarchical gamma regression models to examine the separate associations between patient receipt of oncology outreach and one-way patient travel times to chemotherapy, radiotherapy, and surgery.

RESULTS

On average, 9,935 of 39,960 oncologists conducted annual outreach, where 57.8% traveled with low frequency (0-1 outreach visits/mo), 21.1% with medium frequency (1-3 outreach visits/mo), and 21.1% with high frequency (>3 outreach visits/mo). Oncologists provided surgery, radiotherapy, and chemotherapy to 51,715, 27,120, and 5,874 rural beneficiaries, respectively, of whom 2.5%, 6.9%, and 3.6% received oncology outreach. Rural patients who received oncology outreach traveled 16% (95% CI, 11 to 21) and 11% (95% CI, 9 to 13) less minutes to chemotherapy and radiotherapy than those who did not receive oncology outreach, corresponding to expected one-way savings of 15.9 (95% CI, 15.5 to 16.4) and 11.9 (95% CI, 11.7 to 12.2) minutes, respectively.

CONCLUSION

Our study introduces a novel claims-based approach for tracking the nationwide traveling oncology workforce and supports oncology outreach as an effective means for improving rural access to cancer care.

Association between a network-based physician linchpin score and cancer patient mortality: a SEER-Medicare analysis

BACKGROUND

Patients with cancer frequently require multidisciplinary teams for optimal cancer outcomes. Network analysis can capture relationships among cancer specialists, and we developed a novel physician linchpin score to characterize "linchpin" physicians whose peers have fewer ties to other physicians of the same oncologic specialty. Our study examined whether being treated by a linchpin physician was associated with worse survival.

METHODS

In this cross-sectional study, we analyzed Surveillance, Epidemiology, and End Results-Medicare data for patients diagnosed with stage I to III non-small cell lung cancer or colorectal cancer (CRC) in 2016-2017. We assembled patient-sharing networks and calculated linchpin scores for medical oncologists, radiation oncologists, and surgeons. Physicians were considered linchpins if their linchpin score was within the top 15% for their specialty. We used Cox proportional hazards models to examine associations between being treated by a linchpin physician and survival, with a 2-year follow-up period.

RESULTS

The study cohort included 10 081 patients with non-small cell lung cancer and 9036 patients with CRC. Patients with lung cancer treated by a linchpin radiation oncologist had a 17% (95% confidence interval = 1.04 to 1.32) greater hazard of mortality, and similar trends were observed for linchpin medical oncologists. Patients with CRC treated by a linchpin surgeon had a 22% (95% confidence interval = 1.03 to 1.43) greater hazard of mortality.

CONCLUSIONS

In an analysis of Medicare beneficiaries with nonmetastatic lung cancer or CRC, those treated by linchpin physicians often experienced worse survival. Efforts to improve outcomes can use network analysis to identify areas with reduced access to multidisciplinary specialists.

Comparison of US Oncologist Rurality by Practice Setting and Patients Served

Importance

Studies of the oncology workforce most often classify physician rurality by their practice location, but this could miss the true extent of physicians involved in rural cancer care.

Objective

To compare a method for identifying oncology physicians involved in rural cancer care that uses the proportion of rural patients served with the standard method based on practice location.

Design, Setting, and Participants

This cross-sectional study used retrospective Centers for Medicare & Medicaid Services encounter data on medical oncologists, radiation oncologists, and surgeons treating Medicare beneficiaries diagnosed with breast, colorectal, or lung cancer from January 1 to December 31, 2019. Data were analyzed from May to September 2023.

Main Outcomes and Measures

The standard method of classifying oncologist physician rurality based on practice location was compared with a novel method of classification based on proportion of rural patients served.

Results

The study included 27 870 oncology physicians (71.3% male), of whom 835 (3.0%) practiced in a rural location. Physicians practicing in a rural location treated a high proportion of rural patients (median, 50.0% [IQR, 16.7%-100%]). When considering the rurality of physicians' patient panels, 5123 physicians (18.4%) whose patient panel included at least 20% rural patients, 3199 (11.5%) with at least 33% rural patients, and 1996 (7.2%) with at least 50% rural patients were identified. Using a physician's patient panel to classify physician rurality revealed a higher number and greater spread of oncology physicians involved in rural cancer care in the US than the standard method, while maintaining high performance (area under the curve, 0.857) and fair concordance (κ, 0.346; 95% CI, 0.323-0.369) with the method based on practice setting.

Conclusions and Relevance

In this cross-sectional study, classifying oncologist rurality by the proportion of rural patients served identified more oncology physicians treating patients living in rural areas than the standard method of practice location and may more accurately capture the rural cancer physician workforce, as many hospitals have historically been located in more urban areas. This new method may be used to improve future studies of rural cancer care delivery.

Dissecting patterns and predictors of interhospital transfers for patients with brain metastasis

OBJECTIVE

Interhospital transfers in the acute setting may contribute to high cost, patient inconvenience, and delayed treatment. The authors sought to understand patterns and predictors in the transfer of brain metastasis patients after emergency department (ED) encounter.

METHODS

The authors analyzed 3037 patients with brain metastasis who presented to the ED in Massachusetts and were included in the Healthcare Cost and Utilization Project State Inpatient Database and State Emergency Department Database in 2018 and 2019.

RESULTS

The authors found that 6.9% of brain metastasis patients who presented to the ED were transferred to another facility, either directly or indirectly after admission. The sending EDs were more likely to be nonteaching hospitals without neurosurgery and radiation oncology services (p < 0.01). Transferred patients were more likely to present with neurological symptoms compared to those admitted or discharged (p < 0.01). Among those transferred, approximately 30% did not undergo a significant procedure after transfer and approximately 10% were discharged within 3 days, in addition to not undergoing significant interventions. In total, 74% of transferred patients were sent to a facility significantly farther (> 3 miles) than the nearest facility with neurosurgery and radiation oncology services. Further distance transfers were not associated with improvements in 30-day readmission rate (OR [95% CI] 0.64 [0.30-1.34] for 15-30 miles; OR [95% CI] 0.73 [0.37-1.46] for > 30 miles), 90-day readmission rate (OR [95% CI] 0.50 [0.18-1.28] for 15-30 miles; OR [95% CI] 0.53 [0.18-1.51] for > 30 miles), and length of stay (OR [95% CI] 1.21 days [0.94-1.29] for both 15-30 miles and > 30 miles) compared to close-distance transfers.

CONCLUSIONS

The authors identified a notable proportion of transfers without subsequent significant intervention or appreciable medical management. This may reflect ED physician discomfort with the neurological symptoms of brain metastasis. Many patients were also transferred to hospitals distant from their point of origin and demonstrated no differences in readmission rates and length of stay.

Do Physician Patient Sharing Networks Impact Disparities in Access to Multidisciplinary Cancer Consultation for Early-Stage Lung Cancer? A SEER-Medicare Network Analysis

PURPOSE/OBJECTIVE(S)

To examine early-stage non-small cell lung cancer (NSCLC) patient and physician network characteristics associated with receipt of multidisciplinary cancer consultations (MDCCs), defined as an encounter with both a surgeon and a radiation oncologist after diagnosis. We tested the hypothesis that physician roles in patient-sharing networks can modify disparities in access to surgical and radiotherapy care. Specifically, we investigated the role of "linchpin physicians" using a novel network measure that assesses the extent to which a physician is the sole type of specialist in their physician patient sharing network, and the association of these physicians on patterns of care.

MATERIALS/METHODS

The study cohort included patients diagnosed with stage I/IIA NSCLC in 2016-17 using Surveillance, Epidemiology and End Results (SEER)-Medicare data. We assembled a physician patient-sharing network using all NSCLC patient encounters between 3 months prior to and 12 months following diagnosis. We weighted physicians' patient-sharing ties by the number of shared patients and calculated physician's linchpin scores as a continuous variable measuring the extent to which a physicians' peers in the network lack ties to others of the same specialty as the focal physician. Physicians in the top 15%ile were considered a linchpin specialist. We then derived the proportion of linchpin surgeons or radiation oncologists per hospital referral region (HRR). We assessed patient characteristics including age, sex, race, ethnicity, cancer stage, comorbidity index, rurality, and socioeconomic status (SES). The study outcome was a surgical, radiation oncology, or MDCC in the 2 months following diagnosis. We then performed multivariable logistic regressions (MLR) to assess whether linchpins modified disparities in MDCC.

RESULTS

Of 6,148 patients in the study, 12% received a MDCC, 28% only saw a radiation oncologist, 33% only saw a surgeon, and 27% saw neither for 2 months after diagnosis. On MLR, Hispanic patients were significantly less likely than White patients to receive consultations with a radiation oncologist, while patients in lowest SES quintile were significantly less likely to receive a surgical consultation than those in the highest SES. The proportion of linchpin surgeons and radiation oncologists per HRR was negatively associated with receipt of MDCC (OR low vs high [95% CI] = 0.75 [0.66 to 0.86]; 0.70 [0.61 to 0.80, respectively), but our regressions did not show that any demographic or SES disparities were modified by linchpin specialists.

CONCLUSION

Linchpin specialists do not impact known disparities in access to cancer consultations. Identifying additional system-level factors that contribute to disparities in cancer consultations may lead to policy insights aimed at reducing unwarranted variation in care.

Breaking membrane barriers to neutralize E. coli and K. pneumoniae virulence with PEGylated branched polyethylenimine

Bacterial infections caused by Gram-negative pathogens, such as those in the family Enterobacteriaceae, are among the most difficult to treat because effective therapeutic options are either very limited or non-existent. This raises serious concern regarding the emergence and spread of multi-drug resistant (MDR) pathogens in the community setting; and thus, creates the need for discovery efforts and/or early-stage development of novel therapies for infections. Our work is directed towards branched polyethylenimine (BPEI) modified with polyethylene glycol (PEG) as a strategy for targeting virulence from Gram-negative bacterial pathogens. Here, we neutralize lipopolysaccharide (LPS) as a barrier to the influx of antibiotics. Data demonstrate that the β-lactam antibiotic oxacillin, generally regarded as ineffective against Gram-negative bacteria, can be potentiated by 600 Da BPEI to kill some Escherichia coli and some Klebsiella pneumoniae. Modification of 600 Da BPEI with polyethylene glycol (PEG) could increase drug safety and improves potentiation activity. The ability to use the Gram-positive agent, oxacillin, against Gram-negative pathogens could expand the capability to deliver effective treatments that simplify, reduce, or eliminate some complicated treatment regimens.

The interaction of rurality and rare cancers for travel time to cancer care

PURPOSE

Geographic access to cancer care is known to significantly impact utilization and outcomes. Longer travel times have negative impacts for patients requiring highly specialized care, such as for rare cancers, and for those in rural areas. Scant population-based research informs geographic access to care for rare cancers and whether rurality impacts that access.

METHODS

Using Medicare data (2014-2015), we identified prevalent cancers and cancer-directed surgeries, chemotherapy, and radiation. We classified cancers as rare (incidence <6/100,000/year) or common (incidence ≥6/100,000/year) using previously published thresholds and categorized rurality from ZIP code of beneficiary residence. We estimated travel time between beneficiaries and providers for each service based on ZIP code. Descriptive statistics summarized travel time by rare versus common cancers, service type, and rurality.

FINDINGS

We included 1,169,761 Medicare beneficiaries (21.9% in nonmetropolitan areas), 87,399; 7.5% had rare cancers, with 9,133,003 cancer-directed services. Travel times for cancer services ranged from approximately 29 minutes (25th percentile) to 68 minutes (75th percentile). Travel times were similar for rare and common cancers overall (median: 45 vs 43 minutes) but differed by service type; 13.4% of surgeries were >2 hours away for rare cancers, compared to 8.3% for common cancers. Increasing rurality disproportionately increased travel time to surgical care for rare compared to common cancers.

CONCLUSIONS

Travel times to cancer services are longest for surgery, especially among rural residents, yet not markedly longer overall between rare versus common cancers. Understanding geographic access to cancer care for patients with rare cancers is important to delivering specialized care.

Telehealth Use Following COVID-19 Within Patient-Sharing Physician Networks at a Rural Comprehensive Cancer Center: Cross-sectional Analysis

BACKGROUND

In response to the COVID-19 pandemic, cancer centers rapidly adopted telehealth to deliver care remotely. Telehealth will likely remain a model of care for years to come and may not only affect the way oncologists deliver care to their own patients but also the physicians with whom they share patients.

OBJECTIVE

This study aimed to examine oncologist characteristics associated with telehealth use and compare patient-sharing networks before and after the COVID-19 pandemic in a rural catchment area with a particular focus on the ties between physicians at the comprehensive cancer center and regional facilities.

METHODS

In this retrospective observational study, we obtained deidentified electronic health record data for individuals diagnosed with breast, colorectal, or lung cancer at Dartmouth Health in New Hampshire from 2018-2020. Hierarchical logistic regression was used to identify physician factors associated with telehealth encounters post COVID-19. Patient-sharing networks for each cancer type before and post COVID-19 were characterized with global network measures. Exponential-family random graph models were performed to estimate homophily terms for the likelihood of ties existing between physicians colocated at the hub comprehensive cancer center.

RESULTS

Of the 12,559 encounters between patients and oncologists post COVID-19, 1228 (9.8%) were via telehealth. Patient encounters with breast oncologists who practiced at the hub hospital were over twice as likely to occur via telehealth compared to encounters with oncologists who practiced in regional facilities (odds ratio 2.2, 95% CI 1.17-4.15; P=.01). Patient encounters with oncologists who practiced in multiple locations were less likely to occur via telehealth, and this association was statistically significant for lung cancer care (odds ratio 0.26, 95% CI 0.09-0.76; P=.01). We observed an increase in ties between oncologists at the hub hospital and oncologists at regional facilities in the lung cancer network post COVID-19 compared to before COVID-19 (93/318, 29.3%, vs 79/370, 21.6%, respectively), which was also reflected in the lower homophily coefficients post COVID-19 compared to before COVID-19 for physicians being colocated at the hub hospital (estimate: 1.92, 95% CI 1.46-2.51, vs 2.45, 95% CI 1.98-3.02). There were no significant differences observed in breast cancer or colorectal cancer networks.

CONCLUSIONS

Telehealth use and associated changes to patient-sharing patterns associated with telehealth varied by cancer type, suggesting disparate approaches for integrating telehealth across clinical groups within this health system. The limited changes to the patient-sharing patterns between oncologists at the hub hospital and regional facilities suggest that telehealth was less likely to create new referral patterns between these types of facilities and rather replace care that would otherwise have been delivered in person. However, this study was limited to the 2 years immediately following the initial outbreak of COVID-19, and longer-term follow-up may uncover delayed effects that were not observed in this study period.

Use of a Novel Network-Based Linchpin Score to Characterize Accessibility to the Oncology Physician Workforce in the United States

Importance

Physician headcounts provide useful information about the cancer care delivery workforce; however, efforts to track the oncology workforce would benefit from new measures that capture how essential a physician is for meeting the multidisciplinary cancer care needs of the region. Physicians are considered linchpins when fewer of their peers are connected to other physicians of the same specialty as the focal physician. Because they are locally unique for their specialty, these physicians' networks may be particularly vulnerable to their removal from the network (eg, through relocation or retirement).

Objective

To examine a novel network-based physician linchpin score within nationwide cancer patient-sharing networks and explore variation in network vulnerability across hospital referral regions (HRRs).

Design, Setting, and Participants

This cross-sectional study analyzed fee-for-service Medicare claims and included Medicare beneficiaries with an incident diagnosis of breast, colorectal, or lung cancer from 2016 to 2018 and their treating physicians. Data were analyzed from March 2022 to October 2022.

Exposures

Physician characteristics assessed were specialty, rurality, and Census region. HRR variables assessed include sociodemographic and socioeconomic characteristics and use of cancer services.

Main Outcomes and Measures

Oncologist linchpin score, which examined the extent to which a physician's peers were connected to other physicians of the same specialty as the focal physician. Network vulnerability, which distinguished HRRs with more linchpin oncologists than expected based on oncologist density. χ2 and Fisher exact tests were used to examine relationships between oncologist characteristics and linchpin score. Spearman rank correlation coefficient (ρ) was used to measure the strength and direction of relationships between HRR network vulnerability, oncologist density, population sociodemographic and socioeconomic characteristics, and cancer service use.

Results

The study cohort comprised 308 714 patients with breast, colorectal, or lung cancer. The study cohort of 308 714 patients included 161 206 (52.2%) patients with breast cancer, 76 604 (24.8%) patients with colorectal cancer, and 70 904 (23.0%) patients with lung cancer. In our sample, 272 425 patients (88%) were White, and 238 603 patients (77%) lived in metropolitan areas. The cancer patient-sharing network included 7221 medical oncologists and 3573 radiation oncologists. HRRs with more vulnerable networks for medical oncology had a higher percentage of beneficiaries eligible for Medicaid (ρ, 0.19; 95% CI, 0.08 to 0.29). HRRs with more vulnerable networks for radiation oncology had a higher percentage of beneficiaries living in poverty (ρ, 0.17; 95% CI, 0.06 to 0.27), and a higher percentage of beneficiaries eligible for Medicaid (ρ, 0.21; 95% CI, 0.09 to 0.31), and lower rates of cohort patients receiving radiation therapy (ρ, -0.18; 95% CI, -0.28 to -0.06; P = .003). The was no association between network vulnerability for medical oncology and percent of cohort patients receiving chemotherapy (ρ, -0.03; 95% CI, -0.15 to 0.08).

Conclusions and Relevance

This study found that patient-sharing network vulnerability was associated with poverty and lower rates of radiation therapy. Health policy strategies for addressing network vulnerability may improve access to interdisciplinary care and reduce treatment disparities.

Using passive sensor data to probe associations of social structure with changes in personality: A synthesis of network analysis and machine learning

Social network analysis (SNA) is an increasingly popular and effective tool for modeling psychological phenomena. Through application to the personality literature, social networks, in conjunction with passive, non-invasive sensing technologies, have begun to offer powerful insight into personality state variability. Resultant constructions of social networks can be utilized alongside machine learning-based frameworks to uniquely model personality states. Accordingly, this work leverages data from a previously published study to combine passively collected wearable sensor information on face-to-face, workplace social interactions with ecological momentary assessments of personality state. Data from 54 individuals across six weeks was used to explore the relative importance of 26 unique structural and nodal social network features in predicting individual changes in each of the Big Five (5F) personality states. Changes in personality state were operationalized by calculating the weekly root mean square of successive differences (RMSSD) in 5F state scores measured daily via self-report. Using only SNA-derived features from wearable sensor data, boosted tree-based machine learning models explained, on average, approximately 28-30% of the variance in individual personality state change. Model introspection implicated egocentric features as the most influential predictors across 5F-specific models, with network efficiency, constraint, and effective size measures among the most important. Feature importance profiles for each 5F model partially echoed previous empirical findings. Results support future efforts focusing on egocentric components of SNA and suggest particular investment in exploring efficiency measures to model personality fluctuations within the workplace setting.

Rural-Urban Differences in Breast Cancer Surgical Delays in Medicare Beneficiaries

BACKGROUND

Delays between breast cancer diagnosis and surgery are associated with worsened survival. Delays are more common in urban-residing patients, although factors specific to surgical delays among rural and urban patients are not well understood.

METHODS

We used a 100% sample of fee-for-service Medicare claims during 2007-2014 to identify 238,491 women diagnosed with early-stage breast cancer undergoing initial surgery and assessed whether they experienced biopsy-to-surgery intervals > 90 days. We employed multilevel regression to identify associations between delays and patient, regional, and surgeon characteristics, both in combined analyses and stratified by rurality of patient residence.

RESULTS

Delays were more prevalent among urban patients (2.5%) than rural patients (1.9%). Rural patients with medium- or high-volume surgeons had lower odds of delay than patients with low-volume surgeons (odds ratio [OR] = 0.71, 95% confidence interval [CI] = 0.58-0.88; OR = 0.74, 95% CI = 0.61-0.90). Rural patients whose surgeon operated at ≥ 3 hospitals were more likely to experience delays (OR = 1.29, 95% CI = 1.01-1.64, Ref: 1 hospital). Patient driving times ≥ 1 h were associated with delays among urban patients only. Age, black race, Hispanic ethnicity, multimorbidity, and academic/specialty hospital status were associated with delays.

CONCLUSIONS

Sociodemographic, geographic, surgeon, and facility factors have distinct associations with > 90-day delays to initial breast cancer surgery. Interventions to improve timeliness of breast cancer surgery may have disparate impacts on vulnerable populations by rural-urban status.

Prevalence of Children With Medical Complexity and Associations With Health Care Utilization and In-Hospital Mortality

Importance

Children with medical complexity (CMC) have substantial health care needs and frequently experience poor health care quality. Understanding the population prevalence and associated health care needs can inform clinical and public health initiatives.

Objective

To estimate the prevalence of CMC using open-source pediatric algorithms, evaluate performance of these algorithms in predicting health care utilization and in-hospital mortality, and identify associations between medical complexity as defined by these algorithms and clinical outcomes.

Design, Setting, and Participants

This retrospective cohort study used all-payer claims data from Colorado, Massachusetts, and New Hampshire from 2012 through 2017. Children and adolescents younger than 18 years residing in these states were included if they had 12 months or longer of enrollment in a participating health care plan. Analyses were conducted from March 12, 2021, to January 7, 2022.

Exposures

The pediatric Complex Chronic Condition Classification System, Pediatric Medical Complexity Algorithm, and Children With Disabilities Algorithm were applied to 3 years of data to identify children with complex and disabling conditions, first in their original form and then using more conservative criteria that required multiple health care claims or involvement of 3 or more body systems.

Main Outcomes and Measures

Primary outcomes, examined over 2 years, included in-hospital mortality and a composite measure of health care services, including specialized therapies, specialized medical equipment, and inpatient care. Outcomes were modeled using logistic regression. Model performance was evaluated using C statistics, sensitivity, and specificity.

Results

Of 1 936 957 children, 48.4% were female, 87.8% resided in urban core areas, and 45.1% had government-sponsored insurance as their only primary payer. Depending on the algorithm and coding criteria applied, 0.67% to 11.44% were identified as CMC. All 3 algorithms had adequate discriminative ability, sensitivity, and specificity to predict in-hospital mortality and composite health care services (C statistic = 0.76 [95% CI, 0.73-0.80] to 0.81 [95% CI, 0.78-0.84] for mortality and 0.77 [95% CI, 0.76-0.77] to 0.80 [95% CI, 0.79-0.80] for composite health care services). Across algorithms, CMC had significantly greater odds of mortality (adjusted odds ratio [aOR], 9.97; 95% CI, 7.70-12.89; to aOR, 69.35; 95% CI, 52.52-91.57) and composite health care services (aOR, 4.59; 95% CI, 4.44-4.73; to aOR, 18.87; 95% CI, 17.87-19.93) than children not identified as CMC.

Conclusions and Relevance

In this study, open-source algorithms identified different cohorts of CMC in terms of prevalence and magnitude of risk, but all predicted increased health care utilization and in-hospital mortality. These results can inform research, programs, and policies for CMC.

Characterizing node-negative non-small cell lung cancer patients with similarity networks: A CancerLinQ Discovery analysis

e20585

Background: Patients with node-negative non-small cell lung cancer (NSCLC) whose tumors are completely resected account for approximately 17% of all patients with lung cancer, and disease recurrence occurs in approximately 1 in 5 of these patients. The lack of consensus on factors associated with risk of disease recurrence among node-negative NSCLC patients is a significant barrier to applying precision medicine strategies in this patient population. Clustering similar patients based on distances between various features of data is an emerging topic in precision medicine. Patient similarity networks represent a new model for clustering patients based on heterogeneous data, whereby any data type is converted into a similarity network by defining a similarity measure. The objective of this study was to examine the utility of patient similarity networks to identify NSCLC patients at higher risk of adverse outcomes. Methods: We conducted a retrospective, observational study of 6,020 node-negative NSCLC patients with an initial diagnosis in 2011-2014 in the CancerLinQ Discovery database. A patient similarity network was assembled based on comorbidities and network communities of patients with similar comorbidities at diagnosis were identified. Using Cox proportional-hazards modeling, we examined the extent to which patient age, sex, race, ethnicity, and network community predicted 2-year disease recurrence and 2-year mortality. Results: In the adjusted analyses, patients in the network community enriched for renal disease and congestive heart failure had an 83% increased risk of mortality (95% CI = 1.39-2.41). Patients in the network community enriched for pulmonary disease had a 37% increased risk of mortality (95% CI = 1.06-1.74) yet a lower risk of recurrence (HR = 0.5, 95% CI = 0.34-0.75). After adjusting for comorbidity network community, male patients had a 14% increased risk of mortality (95%CI = 1.02-1.28) and a 21% increased risk of recurrence (95% CI = 1.05-1.40) and black patients had a lower mortality risk (HR = 0.71, 95% CI = 0.58-0.86). Conclusions: Future studies applying patient similar networks to integrate additional diverse and high dimensional data types may afford more clarity in assigning risk of adverse outcomes.

MethylSPWNet and MethylCapsNet: Biologically Motivated Organization of DNAm Neural Networks, Inspired by Capsule Networks

DNA methylation (DNAm) alterations have been heavily implicated in carcinogenesis and the pathophysiology of diseases through upstream regulation of gene expression. DNAm deep-learning approaches are able to capture features associated with aging, cell type, and disease progression, but lack incorporation of prior biological knowledge. Here, we present modular, user-friendly deep-learning methodology and software, MethylCapsNet and MethylSPWNet, that group CpGs into biologically relevant capsules-such as gene promoter context, CpG island relationship, or user-defined groupings-and relate them to diagnostic and prognostic outcomes. We demonstrate these models' utility on 3,897 individuals in the classification of central nervous system (CNS) tumors. MethylCapsNet and MethylSPWNet provide an opportunity to increase DNAm deep-learning analyses' interpretability by enabling a flexible organization of DNAm data into biologically relevant capsules.

A measure of local uniqueness to identify linchpins in a social network with node attributes

Network centrality measures assign importance to influential or key nodes in a network based on the topological structure of the underlying adjacency matrix. In this work, we define the importance of a node in a network as being dependent on whether it is the only one of its kind among its neighbors' ties. We introduce linchpin score, a measure of local uniqueness used to identify important nodes by assessing both network structure and a node attribute. We explore linchpin score by attribute type and examine relationships between linchpin score and other established network centrality measures (degree, betweenness, closeness, and eigenvector centrality). To assess the utility of this measure in a real-world application, we measured the linchpin score of physicians in patient-sharing networks to identify and characterize important physicians based on being locally unique for their specialty. We hypothesized that linchpin score would identify indispensable physicians who would not be easily replaced by another physician of their specialty type if they were to be removed from the network. We explored differences in rural and urban physicians by linchpin score compared with other network centrality measures in patient-sharing networks representing the 306 hospital referral regions in the United States. We show that linchpin score is uniquely able to make the distinction that rural specialists, but not rural general practitioners, are indispensable for rural patient care. Linchpin score reveals a novel aspect of network importance that can provide important insight into the vulnerability of health care provider networks. More broadly, applications of linchpin score may be relevant for the analysis of social networks where interdisciplinary collaboration is important.

Infant Mortality Associated With Prenatal Opioid Exposure

IMPORTANCE

Knowledge of health outcomes among opioid-exposed infants is limited, particularly for those not diagnosed with neonatal opioid withdrawal syndrome (NOWS).

OBJECTIVES

To describe infant mortality among opioid-exposed infants and identify how mortality risk differs in opioid-exposed infants with and without a diagnosis of NOWS compared with infants without opioid exposure.

DESIGN, SETTING, AND PARTICIPANTS

A retrospective cohort study of maternal-infant dyads was conducted, linking health care claims with vital records for births from January 1, 2010, to December 31, 2014, with follow-up of infants until age 1 year (through 2015). Maternal-infant dyads were included if the infant was born in Texas at 22 to 43 weeks' gestational age to a woman aged 15 to 44 years insured by Texas Medicaid. Data analysis was performed from May 2019 to October 2020.

EXPOSURE

The primary exposure was prenatal opioid exposure, with infants stratified by the presence or absence of a diagnosis of NOWS during the birth hospitalization.

MAIN OUTCOMES AND MEASURES

Risk of infant mortality (death at age <365 days) was examined using Kaplan-Meier and log-rank tests. A series of logistic regression models was estimated to determine associations between prenatal opioid exposure and mortality, adjusting for maternal and neonatal characteristics and clustering infants at the maternal level to account for statistical dependence owing to multiple births during the study period.

RESULTS

Among 1 129 032 maternal-infant dyads, 7207 had prenatal opioid exposure, including 4238 diagnosed with NOWS (mean [SD] birth weight, 2851 [624] g) and 2969 not diagnosed with NOWS (mean [SD] birth weight, 2971 [639] g). Infant mortality was 20 per 1000 live births for opioid-exposed infants not diagnosed with NOWS, 11 per 1000 live births for infants with NOWS, and 6 per 1000 live births in the reference group (P < .001). After adjusting for maternal and neonatal characteristics, mortality in infants with a NOWS diagnosis was not significantly different from the reference population (odds ratio, 0.82; 95% CI, 0.58-1.14). In contrast, the odds of mortality in opioid-exposed infants not diagnosed with NOWS was 72% greater than the reference population (odds ratio, 1.72; 95% CI, 1.25-2.37).

CONCLUSIONS AND RELEVANCE

In this study, opioid-exposed infants appeared to be at increased risk of mortality, and the treatments and supports provided to those diagnosed with NOWS may be protective. Interventions to support opioid-exposed maternal-infant dyads are warranted, regardless of the perceived severity of neonatal opioid withdrawal.

Dual-Function Potentiation by PEG-BPEI Restores Activity of Carbapenems and Penicillins against Carbapenem-Resistant Enterobacteriaceae

The rise of life-threatening carbapenem-resistant Enterobacteriaceae (CRE) infections has become a critical medical threat. Some of the most dangerous CRE bacteria can produce enzymes that degrade a wide range of antibiotics, including carbapenems and β-lactams. Infections by CRE have a high mortality rate, and survivors can have severe morbidity from treatment with toxic last-resort antibiotics. CRE have mobile genetic elements that transfer resistance genes to other species. These bacteria also circulate throughout the healthcare system. The mobility and spread of CRE need to be curtailed, but these goals are impeded by having few agents that target a limited range of pathogenic CRE species. Against CRE possessing the metallo-β-lactamase NDM-1, Klebsiella pneumoniae ATCC BAA-2146 and Escherichia coli ATCC BAA-2452, the potentiation of meropenem and imipenem is possible with low-molecular weight branched polyethylenimine (600 Da BPEI) and its poly(ethylene glycol) (PEG)ylated derivative (PEG-BPEI) that has a low in vivo toxicity. The mechanism of action is elucidated with fluorescence assays of drug influx and isothermal calorimetry data showing the chelation of essential Zn2+ ions. These results suggested that 600 Da BPEI and PEG-BPEI may also improve the uptake of antibiotics and β-lactamase inhibitors. Indeed, the CRE E. coli strain is rendered susceptible to the combination of piperacillin and tazobactam. These results expand the possible utility of 600 Da BPEI potentiators, where previously we have demonstrated the ability to improve antibiotic efficacy against antibiotic resistant clinical isolates of Pseudomonas aeruginosa, Staphylococcus aureus, and Staphylococcus epidermidis.

Differences in Uptake of Low-Dose CT Scan for Lung Cancer among White and Black Adult Smokers in the United States-2017

To study racial/ethnic differences in the utilization of low-dose computerized tomography (LDCT) scan for lung cancer among adult smokers. Cross-sectional data (n=2,640) of adults aged 55-74, were from the 2017 Behavioral Risk Factor Surveillance System, Lung Cancer Screening module. Weighted, multivariable logistic regression was conducted. Most, 70.9%, were White and 52.2% male. About 16.0 % reported receiving LDCT scan in the past 12 months, 12.0% of Blacks and 17.4% of Whites. More Whites (55.0%) had ≥30 pack-years smoking history than Blacks (20%). Blacks had lower odds, .52 (CI: 0.28-0.96) of receiving LDCT scan than Whites. The odds of receiving LDCT scan were higher for those who were male, who tried to quit smoking in the past year, and for those with more education, health insurance, high blood pressure, lung disease, or cancer history (other than skin or lung cancer). This study suggests racial differences in the use of LDCT scan.

Prediction of an outcome using NETwork Clusters (NET-C)

Birth weight is a key consequence of environmental exposures and metabolic alterations and can influence lifelong health. While a number of methods have been used to examine associations of trace element (including essential nutrients and toxic metals) concentrations or metabolite concentrations with a health outcome, birth weight, studies evaluating how the coexistence of these factors impacts birth weight are extremely limited. Here, we present a novel algorithm NETwork Clusters (NET-C), to improve the prediction of outcome by considering the interactions of features in the network and then apply this method to predict birth weight by jointly modelling trace element and cord blood metabolite data. Specifically, by using trace element and/or metabolite subnetworks as groups, we apply group lasso to estimate birth weight. We conducted statistical simulation studies to examine how both sample size and correlations between grouped features and the outcome affect prediction performance. We showed that in terms of prediction error, our proposed method outperformed other methods such as (a) group lasso with groups defined by hierarchical clustering, (b) random forest regression and (c) neural networks. We applied our method to data ascertained as part of the New Hampshire Birth Cohort Study on trace elements, metabolites and birth outcomes, adjusting for other covariates such as maternal body mass index (BMI) and enrollment age. Our proposed method can be applied to a variety of similarly structured high-dimensional datasets to predict health outcomes.

Surgeon and medical oncologist peer network effects on the uptake of the 21-gene breast cancer recurrence score assay

BACKGROUND

Drivers behind the adoption of gene expression profiling in breast cancer oncology have been shown to include exposure to physician colleagues' use of a given genomic test. We examined adoption of the Oncotype DX 21-gene breast cancer recurrence score assay (ODX) in the United States after its incorporation into clinical guidelines. The influence of patient-sharing ties and co-location with prior adopters and the role of these potential exposures across medical specialties on peers' adoption of the test were examined.

METHODS

We conducted a retrospective cohort study of women with incident breast cancer using a 100% sample of fee-for-service Medicare enrollee claims over 2008-2011. Peer networks connecting medical oncologists and surgeons treating these patients were constructed using patient-sharing and geographic co-location. The impact of peer connections on the adoption of ODX by physicians and testing of patients was modeled with multivariable hierarchical regression.

RESULTS

Altogether, 156,229 women identified with incident breast cancer met criteria for cohort inclusion. A total of 7689 ODX prescribing physicians were identified. Co-location with medical oncologists who adopted the test in the early period (2008-2009) was associated with a 1.38-fold increase in the odds of a medical oncologist adopting ODX in 2010-2011 (95% CI = 1.04-1.83), as was co-location with early-adopting surgeons (odds ratio [OR] = 1.25, 95% CI = 1.00-1.58). Patients whose primary medical oncologist was linked to an early-adopting surgeon through co-location (OR = 1.17, 95% CI = 1.04-1.32) or both patient-sharing and co-location (OR = 1.17, 95% CI = 1.03-1.34) were more likely to receive ODX.

CONCLUSIONS

Exposure to surgeon early adopters through peer networks and co-location was predictive of ODX uptake by medical oncologists and testing of patients. Interventions focused on the role of surgeons in molecular testing may improve the implementation of best practices in breast cancer care.

PEGylation of Polyethylenimine Lowers Acute Toxicity while Retaining Anti-Biofilm and β-Lactam Potentiation Properties against Antibiotic-Resistant Pathogens

Bacterial biofilms, often impenetrable to antibiotic medications, are a leading cause of poor wound healing. The prognosis is worse for wounds with biofilms of antimicrobial-resistant (AMR) bacteria, such as methicillin-resistant Staphylococcus aureus (MRSA), methicillin-resistant S. epidermidis (MRSE), and multi-drug resistant Pseudomonas aeruginosa (MDR-PA). Resistance hinders initial treatment of standard-of-care antibiotics. The persistence of MRSA, MRSE, and/or MDR-PA often allows acute infections to become chronic wound infections. The water-soluble hydrophilic properties of low-molecular-weight (600 Da) branched polyethylenimine (600 Da BPEI) enable easy drug delivery to directly attack AMR and biofilms in the wound environment as a topical agent for wound treatment. To mitigate toxicity issues, we have modified 600 Da BPEI with polyethylene glycol (PEG) in a straightforward one-step reaction. The PEG-BPEI molecules disable β-lactam resistance in MRSA, MRSE, and MDR-PA while also having the ability to dissolve established biofilms. PEG-BPEI accomplishes these tasks independently, resulting in a multifunction potentiation agent. We envision wound treatment with antibiotics given topically, orally, or intravenously in which external application of PEG-BPEIs disables biofilms and resistance mechanisms. In the absence of a robust pipeline of new drugs, existing drugs and regimens must be re-evaluated as combination(s) with potentiators. The PEGylation of 600 Da BPEI provides new opportunities to meet this goal with a single compound whose multifunction properties are retained while lowering acute toxicity.

Patient and physician factors associated with Oncotype DX and adjuvant chemotherapy utilization for breast cancer patients in New Hampshire, 2010-2016

BACKGROUND

Oncotype DX® (ODX) is used to assess risk of disease recurrence in hormone receptor positive, HER2-negative breast cancer and to guide decisions regarding adjuvant chemotherapy. Little is known about how physician factors impact treatment decisions. The purpose of this study was to examine patient and physician factors associated with ODX testing and adjuvant chemotherapy for breast cancer patients in New Hampshire.

METHODS

We examined New Hampshire State Cancer Registry data on 5630 female breast cancer patients diagnosed from 2010 to 2016. We performed unadjusted and adjusted hierarchical logistic regression to identify factors associated with a patient's receipt of ODX, being recommended and receiving chemotherapy, and refusing chemotherapy. We calculated intraclass correlation coefficients (ICCs) to examine the proportion of variance in clinical decisions explained by between-physician and between-hospital variation.

RESULTS

Over the study period, 1512 breast cancer patients received ODX. After adjustment for patient and tumor characteristics, we found that patients seen by a male medical oncologist were less likely to be recommended chemotherapy following ODX (OR = 0.50 (95% CI = 0.34-0.74), p < 0.01). Medical oncologists with more clinical experience (reference: less than 10 years) were more likely to recommend chemotherapy (20-29 years: OR = 4.05 (95% CI = 1.57-10.43), p < 0.01; > 29 years: OR = 4.48 (95% CI = 1.68-11.95), p < 0.01). A substantial amount of the variation in receiving chemotherapy was due to variation between physicians, particularly among low risk patients (ICC = 0.33).

CONCLUSIONS

In addition to patient clinicopathologic characteristics, physician gender and clinical experience were associated with chemotherapy treatment following ODX testing. The significant variation between physicians indicates the potential for interventions to reduce variation in care.

Abstract A12: Evaluating physician network-based measures of care coordination through the lens of the patient experience

There is significant promise in analyzing physician patient-sharing networks to construct indirect measures of care coordination by defining relationships between physicians based on shared patients. This approach has the potential to uncover how otherwise latent organizational aspects of health care systems impact patient outcomes. A key challenge to this approach is how the various measures used to describe these networks infer the complex realities of health care delivery. The extent of patient-sharing relationships within physician networks has been associated with care utilization, cost of care, and some measures of care quality. While this approach has been validated to signal true professional relationships between physicians from the physicians’ perspective, it is unknown whether these measures reflect patients’ perceptions of their own care coordination. The objective of this study is to evaluate the associations between network-based measures of care coordination and patient-reported experience measures. We linked three publicly available data sources released by Centers for Medicare and Medicaid Services (CMS) that include national data on physician patient sharing in 2015, physician participation in a group practice, and patient-reported quality performance for group practices based on responses to the Consumer Assessment of Healthcare Providers and Systems (CAHPS) Survey in 2016. The outcomes of interest were patient-reported experience measures reflecting “Between visit communication,” “Clinicians working together for your care,” and “Getting timely care, appointments, and information” (CAHPS). The predictor variables of interests were physician group practice density (the number of physician pairs who share patients adjusting for total number of physician pairs) and clustering (the extent to which sets of three physicians all share patients). There were 476 groups that had patient-reported measures available. Patients’ perception of “Clinicians working together for your care” was significantly positively associated with both physician group practice density (p=0.02) and clustering (p=0.006). Physician group practice clustering was also significantly positively associated with “Getting timely care, appointments, and information” (p=0.005). This work suggests that network-based measures of care coordination are associated with some patient-reported experience measures. Evaluating and intervening on patient-sharing networks may provide novel strategies for initiatives aimed at improving quality of care and the patient experience.

Citation Format: Erika L. Moen, Julie P. Bynum. Evaluating physician network-based measures of care coordination through the lens of the patient experience [abstract]. In: Proceedings of the AACR Special Conference on Modernizing Population Sciences in the Digital Age; 2019 Feb 19-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(9 Suppl):Abstract nr A12.

Expanding the Spectrum of Antibiotics Capable of Killing Multidrug-Resistant Staphylococcus aureus and Pseudomonas aeruginosa

Infections from antibiotic-resistant Staphylococcus aureus and Pseudomonas aeruginosa are a serious threat because reduced antibiotic efficacy complicates treatment decisions and prolongs the disease state in many patients. To expand the arsenal of treatments against antimicrobial-resistant (AMR) pathogens, 600-Da branched polyethylenimine (BPEI) can overcome antibiotic resistance mechanisms and potentiate β-lactam antibiotics against Gram-positive bacteria. BPEI binds cell-wall teichoic acids and disables resistance factors from penicillin binding proteins PBP2a and PBP4. This study describes a new mechanism of action for BPEI potentiation of antibiotics generally regarded as agents effective against Gram-positive pathogens but not Gram-negative bacteria. 600-Da BPEI is able to reduce the barriers to drug influx and facilitate the uptake of a non-β-lactam co-drug, erythromycin, which targets the intracellular machinery. Also, BPEI can suppress production of the cytokine interleukin IL-8 by human epithelial keratinocytes. This enables BPEI to function as a broad-spectrum antibiotic potentiator, and expands the opportunities to improve drug design, antibiotic development, and therapeutic approaches against pathogenic bacteria, especially for wound care.

Overcoming Multidrug Resistance and Biofilms of Pseudomonas aeruginosa with a Single Dual-Function Potentiator of β-Lactams

Clinicians prescribe hundreds of millions of β-lactam antibiotics to treat the majority of patients presenting with bacterial infections. Patient outcomes are positive unless resistant bacteria, such as Pseudomonas aeruginosa (P. aeruginosa), are present. P. aeruginosa has both intrinsic and acquired antibiotic resistance, making clinical management of infection a real challenge, particularly when these bacteria are sequestered in biofilms. These problems would be alleviated if, upon the initial presentation of bacterial infection symptoms, clinicians were able to administer an antibiotic that kills both susceptible and otherwise resistant bacteria and eradicates biofilms. As the most common class of antibiotics, β-lactams could be used in a new drug if the leading causes of β-lactam antibiotic resistance, permeation barriers from lipopolysaccharide, efflux pumps, and β-lactamase enzymes, were also defeated. Against P. aeruginosa and their biofilms, the potency of β-lactam antibiotics is restored with 600 Da branched polyethylenimine (600 Da BPEI). Checkerboard assays using microtiter plates demonstrate the potentiation of piperacillin, cefepime, Meropenem, and erythromycin antibiotics. Growth curves demonstrate that only a combination of 600 Da BPEI and piperacillin produces growth inhibition against antibiotic resistant P. aeruginosa. Scanning electron microscopy (SEM) was used to confirm that the combination treatment leads to abnormal P. aeruginosa morphology. Data collected with isothermal titration calorimetry and fluorescence spectroscopy demonstrate a mechanism of action in which potentiation at low concentrations of 600 Da BPEI reduces diffusion barriers from lipopolysaccharides without disrupting the outer membrane itself. Coupled with the ability to overcome a reduction in antibiotic activity created by biofilm exopolymers, targeting anionic sites on lipopolysaccharides and biofilm exopolysaccharides with the same compound provides new opportunities to counter the rise of multidrug-resistant infections.

Modeling peer effect modification by network strength: The diffusion of implantable cardioverter defibrillators in the US hospital network

We develop methodology that allows peer effects (also referred to as social influence and contagion) to be modified by the structural importance of the focal actor's position in the network. The methodology is first developed for a single peer effect and then extended to simultaneously model multiple peer-effects and their modifications by the structural importance of the focal actor. This work is motivated by the diffusion of implantable cardioverter defibrillators (ICDs) in patients with congestive heart failure across a cardiovascular disease patient-sharing network of United States hospitals. We apply the general methodology to estimate peer effects for the adoption of capability to implant ICDs, the number of ICD implants performed by hospitals that are capable, and the number of patients referred to other hospitals by noncapable hospitals. Applying our novel methodology to study ICD diffusion across hospitals, we find evidence that exposure to ICD-capable peer hospitals is strongly associated with the chance a hospital becomes ICD-capable and that the direction and magnitude of the association is extensively modified by the strength of that hospital's position in the network, even after controlling for effects of geography. Therefore, interhospital networks, rather than geography per se, may explain key patterns of regional variations in healthcare utilization.

Low-Molecular-Weight Branched Polyethylenimine Potentiates Ampicillin against MRSA Biofilms

Methicillin-resistant Staphylococcus aureus (MRSA) infections pose a serious threat worldwide. MRSA is the predominant species isolated from medical-device-related biofilm infections and chronic wounds. Its ability to form biofilms grants it resistance to almost all antibiotics on the market. Answering the call for alternative treatments, our lab has been investigating the efficacy of 600 Da branched polyethylenimine (BPEI) as a β-lactam potentiator against bacterial biofilms. Our previous study showed promise against methicillin-resistant Staphylococcus epidermidis biofilms. This study extends our previous findings to eradicate a more virulent pathogen: MRSA biofilms. Microtiter minimum biofilm eradication concentration models, crystal violet assays, and electron microscopy images show synergistic effects between BPEI and ampicillin as a two-step mechanism: step one is the removal of the extracellular polymeric substances (EPS) to expose individual bacteria targets, and step two involves electrostatic interaction of BPEI with anionic teichoic acid in the cell wall to potentiate the antibiotic.

Artificial microbiome heterogeneity spurs six practical action themes and examples to increase study power-driven reproducibility

With >70,000 yearly publications using mouse data, mouse models represent the best engrained research system to address numerous biological questions across all fields of science. Concerns of poor study and microbiome reproducibility also abound in the literature. Despite the well-known, negative-effects of data clustering on interpretation and study power, it is unclear why scientists often house >4 mice/cage during experiments, instead of ≤2. We hypothesized that this high animal-cage-density  practice abounds in published literature because more mice/cage could be perceived as a strategy to reduce housing costs. Among other sources of 'artificial' confounding, including cyclical oscillations of the 'dirty-cage/excrement microbiome', we ranked by priority the heterogeneity of modern husbandry practices/perceptions across three professional organizations that we surveyed in the USA. Data integration (scoping-reviews, professional-surveys, expert-opinion, and 'implementability-score-statistics') identified Six-Actionable Recommendation Themes (SART) as a framework to re-launch emerging protocols and intuitive statistical strategies to use/increase study power. 'Cost-vs-science' discordance was a major aspect explaining heterogeneity, and scientists' reluctance to change. With a 'housing-density cost-calculator-simulator' and fully-annotated statistical examples/code, this themed-framework streamlines the rapid analysis of cage-clustered-data and promotes the use of 'study-power-statistics' to self-monitor the success/reproducibility of basic and translational research. Examples are provided to help scientists document analysis for study power-based sample size estimations using preclinical mouse data to support translational clinical trials, as requested in NIH/similar grants or publications.

An Integrated Gaussian Graphical Model to evaluate the impact of exposures on metabolic networks

Examining the effects of exogenous exposures on complex metabolic processes poses the unique challenge of identifying interactions among a large number of metabolites. Recent progress in the quantification of the metabolome through mass spectrometry (MS) and nuclear magnetic resonance (NMR) has given rise to high-dimensional biomedical data of specific metabolites that can be leveraged to study their effects in humans. These metabolic interactions can be evaluated using probabilistic graphical models (PGMs), which define conditional dependence and independence between components within and between heterogeneous biomedical datasets. This method allows for the detection and recovery of valuable but latent information that cannot be easily detected by other currently existing methods. Here, we develop a PGM method, referred to as an "Integrated Gaussian Graphical Model (IGGM)", to incorporate exposure concentrations of seven trace elements-arsenic (As), lead (Pb), mercury (Hg), cadmium (Cd), zinc (Zn), selenium (Se) and copper (Cu-into metabolic networks. We first conducted a simulation study demonstrating that the integration of trace elements into metabolomics data can improve the accuracy of detecting latent interactions of metabolites impacted by exposure in the network. We tested parameters such as sample size and the number of neighboring metabolites of a chosen trace element for their impact on the accuracy of detecting metabolite interactions. We then applied this method to measurements of cord blood plasma metabolites and placental trace elements collected from newborns in the New Hampshire Birth Cohort Study (NHBCS). We found that our approach can identify latent interactions among metabolites that are related to trace element concentrations. Application to similarly structured data may contribute to our understanding of the complex interplay between exposure-related metabolic interactions that are important for human health.

BPEI-Induced Delocalization of PBP4 Potentiates β-Lactams against MRSA

With its high morbidity rate and increasing resistance to treatment, methicillin-resistant Staphylococcus aureus (MRSA) is a grave concern in the medical field. In methicillin-susceptible strains, β-lactam antibiotics disable the penicillin binding proteins (PBPs) that cross-link the bacterial cell wall. However, methicillin-resistant strains have PBP2a and PBP4, which continue enzymatic activity in the presence of β-lactam antibiotics. The activity of PBP2a and PBP4 is linked to the presence of wall teichoic acid (WTA); thus, WTA has emerged as a target for antibiotic drug discovery. In this work, we disable WTA in situ using its anionic phosphodiester backbone to attract cationic branched polyethylenimine (BPEI). Data show that BPEI removes β-lactam resistance in common MRSA strains and clinical isolates. Fluorescence microscopy was used to investigate this mechanism of action. The results indicate that BPEI prevents the localization of PBP4 to the cell division septum, thereby changing the cellular morphology and inhibiting cell division. Although PBP4 is not required for septum formation, proper cell division and morphology require WTA; BPEI prevents this essential function. The combination of BPEI and β-lactams is bactericidal and synergistic. Because BPEI allows us to study the role of WTA in the cell wall without genetic mutation or altered translocation of biomolecules and/or their precursors, this approach can help revise existing paradigms regarding the role of WTA in prokaryotic biochemistry at every growth stage.

Antibiofilm Synergy of β-Lactams and Branched Polyethylenimine against Methicillin-Resistant Staphylococcus epidermidis

Microbial biofilms are ubiquitous in nature, and they pose a serious threat to public health. Staphylococcus epidermidis is the most common clinical isolate from healthcare- and medical device-related biofilm infections. No antibiotic currently on the market can eradicate pathogenic biofilms, which contain complex defense mechanisms composed of slimelike extracellular polymeric substances. Understanding the need to develop alternative approaches, we examine 600 Da branched polyethylenimine (BPEI) against methicillin-resistant Staphylococcus epidermidis (MRSE) biofilms. Here, a microtiter biofilm model is used to test the synergistic effects between the two components of our combination treatment: BPEI and β-lactam antibiotics. Electron microscopy was used to confirm the growth of MRSE biofilms from the model. Minimum biofilm eradication concentration assays, crystal violet assays, and biofilm kill curves suggest that BPEI exhibits antibiofilm activity and can potentiate β-lactams to eradicate MRSE biofilms.

Physician network position and patient outcomes following implantable cardioverter defibrillator therapy

OBJECTIVE

To evaluate two novel measures of physician network centrality and their associations with implantable cardioverter defibrillator (ICD) procedure volume and health outcomes.

DATA SOURCES

Medicare claims and the National Cardiovascular Data Registry data from 2007 to 2011.

STUDY DESIGN

We constructed a national cardiovascular disease patient-sharing physician network and used network analysis to characterize physician network centrality with two measures: within-hospital degree centrality (number of connections within a hospital) and across-hospital degree centrality (number of connections across hospitals). The primary outcome was risk-adjusted 2-year case fatality. Hierarchical logistic regression estimated the effects of physician's within-hospital and across-hospital degree centrality on case fatality. We included 105 109 ICD therapy patients and 3474 ICD implanting physicians in our analyses.

PRINCIPAL FINDINGS

After controlling for other physician and hospital characteristics, we observed greater risk-adjusted case fatality among patients treated by physicians in the highest across-hospital degree tertile compared to lowest tertile (OR [95% CI] = 1.10 [1.04-1.16], P = 0.001) and lowest tertile volume physicians compared with highest volume (OR [95% CI] = 0.90 [0.84-0.95], P < 0.001). Physician's within-hospital degree tertile was inversely associated with case fatality but not statistically significant.

CONCLUSIONS

Degree centrality measures capture information independent of procedure volume and raise questions about the quality of physicians with networks that predict worse health outcomes.

Evaluating Breast Cancer Care Coordination at a Rural National Cancer Institute Comprehensive Cancer Center Using Network Analysis and Geospatial Methods

BACKGROUND

Variation in cancer care coordination may affect care quality and patient outcomes. We sought to characterize the impact of geographic access to and dispersion of cancer care providers on variation in care coordination.

METHODS

Using electronic health record data from 2,507 women diagnosed with breast cancer at a National Cancer Institute Comprehensive Cancer Center from April 2011 to September 2015, a breast cancer patient-sharing physician network was constructed. Patient "care networks" represent the subnetworks of physicians with whom the focal patient had a clinical encounter. Patient care networks were analyzed to generate two measures of care coordination, care density (ratio of observed vs. potential connections between physicians), and clustering (extent to which physicians form connected triangles).

RESULTS

The breast cancer physician network included 667 physicians. On average, the physicians shared patients with 12 other physicians. Patients saw an average of 8 physicians during active treatment. In multivariable models adjusting for patient sociodemographic and clinical characteristics, we observed that greater travel burden (>2 hours) and lower geographic dispersion were associated with higher care density (P < 0.05 and P < 0.001, respectively) but lower care network clustering (P < 0.05).

CONCLUSIONS

Variation in network-based measures of care coordination is partially explained by patient travel burden and geographic dispersion of care.

IMPACT

Improved understanding of factors driving variation in patient care networks may identify patients at risk of receiving poorly coordinated cancer care.

Cationic Branched Polyethylenimine (BPEI) Disables Antibiotic Resistance in Methicillin-Resistant Staphylococcus epidermidis (MRSE)

Staphylococcus epidermidis is one of the most prevalent prokaryotic species on human skin and mucosal membranes that constitute the commensal flora. S. epidermidis has become one of the most common causes of primary bacteremia. Infections are difficult to diagnose because the pathogen has natural niches on human skin and the ability to adhere to inanimate surfaces via biofilms. Alarmingly, S. epidermidis has acquired resistance to many antibiotics, which presents a danger to human health. Known as methicillin-resistant S. epidermidis (MRSE), most clinical isolates of MRSE in North America exhibit β-lactam resistance primarily due to the presence of mecA, a gene that bestows β-lactam antibiotic resistance in a manner similar to methicillin-resistant Staphylococcus aureus (MRSA). MecA encodes for expression of penicillin-binding protein 2a (PBP2a), which is absent in β-lactam susceptible strains of S. epidermidis. We can disable this resistance factor in MRSE with 600-Da branched polyethylenimine (BPEI). Cationic BPEI targets anionic wall teichoic acid (WTA), an essential cofactor for proper functioning of PBP2a. We found that BPEI synergizes the activity of β-lactam antibiotics against MRSE. Growth curves suggest that the combination of BPEI and oxacillin is bactericidal. Electron micrographs indicate abnormalities in the cellular septa and cell walls of treated samples. Therefore, first-line clinical treatments can be effective against MRSE when used in combination with BPEI.

Nuclear Excluded Autism-Associated Phosphatase and Tensin Homolog Mutations Dysregulate Neuronal Growth

BACKGROUND

Phosphatase and tensin homolog (PTEN) negatively regulates downstream protein kinase B signaling, resulting in decreased cellular growth and proliferation. PTEN is mutated in a subset of children with autism spectrum disorder (ASD); however, the mechanism by which specific point mutations alter PTEN function is largely unknown. Here, we assessed how ASD-associated single-nucleotide variations in PTEN (ASD-PTEN) affect function.

METHODS

We used viral-mediated molecular substitution of human PTEN into Pten knockout mouse neurons and assessed neuronal morphology to determine the functional impact of ASD-PTEN. We employed molecular cloning to examine how PTEN's stability, subcellular localization, and catalytic activity affect neuronal growth.

RESULTS

We identified a set of ASD-PTEN mutations displaying altered lipid phosphatase function and subcellular localization. We demonstrated that wild-type PTEN can rescue the neuronal hypertrophy, while PTEN H93R, F241S, D252G, W274L, N276S, and D326N failed to rescue this hypertrophy. A subset of these mutations lacked nuclear localization, prompting us to examine the role of nuclear PTEN in regulating neuronal growth. We found that nuclear PTEN alone is sufficient to regulate soma size. Furthermore, forced localization of the D252G and W274L mutations into the nucleus partially restores regulation of soma size.

CONCLUSIONS

ASD-PTEN mutations display decreased stability, catalytic activity, and/or altered subcellular localization. Mutations lacking nuclear localization uncover a novel mechanism whereby lipid phosphatase activity in the nucleus can regulate mammalian target of rapamycin signaling and neuronal growth.

Social and Structural Determinants of Smoking Status and Quit Attempts Among Adults Living in 12 US States, 2015

PURPOSE

This study examined the relationships among intermediary determinants, structural determinants, and adult smoking status and quit attempts.

DESIGN

Secondary data analysis of cross-sectional data.

SETTING

Data come from the 2015 Behavioral Risk Factor Surveillance System (BRFSS), Social Context module. A national, representative sample from 12 US states (Alabama, Arkansas, Delaware, District of Columbia, Georgia, Louisiana, Minnesota, Mississippi, Missouri, Rhode Island, Tennessee, and Utah).

PARTICIPANTS

A total of 64 053 noninstitutionalized US adults aged ≥18 years.

MEASURES

Smoking status and quit attempts were outcome variables. Individual-level structural determinants (age, sex, race/ethnicity, marital status, education, and employment status) and intermediary determinants (housing insecurity, food insecurity, health insurance, binge drinking, and general health mental health) from BRFSS.

ANALYSIS

Weighted multivariate, multinomial logistic regression.

RESULTS

Current smoking was greater among men, respondents aged between 35 to 64 and 55 to 64, adults who reported food insecurity, housing insecurity, frequent mental distress, binge drinking, and who were unemployed. Current smokers had higher odds of making quit attempts in the past 12 months if they were non-Hispanic Black, graduated college, and reported food and housing insecurity.

CONCLUSION

Multifaceted smoking cessation interventions that address food and housing needs also incorporate screening for potential comorbidities such as mental distress and/or hazardous alcohol use and may be needed to enhance smoking cessation rates among racially diverse adults.

Assessing Variation in Implantable Cardioverter Defibrillator Therapy Guideline Adherence With Physician and Hospital Patient-sharing Networks

BACKGROUND

Implantable cardioverter defibrillator (ICD) therapy is used for primary prevention of death among people with heart failure, and new evidence in 2005 on its effectiveness changed practice guidelines in the United States.

OBJECTIVES

The objective of this study is to examine how the connectedness of physicians and hospitals, measured using network analysis, relates to guideline-consistent ICD implantation.

RESEARCH DESIGN

We constructed physician and hospital networks for cardiovascular disease. Physicians were linked if they shared cardiovascular disease patients; these links were aggregated by hospital affiliation to construct a hospital network.

SUBJECTS

Medicare beneficiaries who underwent ICD therapy for primary prevention from 2007 to 2011.

MEASURES

The clinical outcome of interest was guideline-consistent ICD implantation, calculated using the National Cardiovascular Data Registry. The exposure variables of interest were the network measures of the ICD surgeon, the referring hospital, and the hospital where the ICD surgery occurred.

RESULTS

We focused on patients who were referred between hospitals for ICD implantation because they were more likely influenced by the hospital network (n=28,179). Patients were less likely to meet guidelines if their referring hospital had more connections to other hospitals (OR, 0.49; 95% confidence interval, 0.25-0.96) and more likely to meet guidelines if their ICD surgery hospital had more connections (OR, 1.61; 95% confidence interval, 0.98-2.64). The ICD surgeon's network measures were not associated with guideline-consistent implantation.

CONCLUSIONS

Associations between the hospital network measures and guideline adherence suggests new approaches to better disseminate clinical guidelines across health systems.

Penalized estimation of sparse concentration matrices based on prior knowledge with applications to placenta elemental data

Identifying patterns of association or dependency among high-dimensional biological datasets with sparse precision matrices remains a challenge. In this paper, we introduce a weighted sparse Gaussian graphical model that can incorporate prior knowledge to infer the structure of the network of trace element concentrations, including essential elements as well as toxic metals and metaloids measured in the human placentas. We present the weighted L1 penalized regularization procedure for estimating the sparse precision matrix in the setting of Gaussian graphical models. First, we use simulation models to demonstrate that the proposed method yields a better estimate of the precision matrix than the procedures that fail to account for the prior knowledge of the network structure. Then, we apply this method to estimate sparse element concentration matrices of placental biopsies from the New Hampshire Birth Cohort Study. The chemical architecture for elements is complex; thus, the method proposed herein was applied to infer the dependency structures of the elements using prior knowledge of their biological roles.

Targeting Wall Teichoic Acid in Situ with Branched Polyethylenimine Potentiates β-Lactam Efficacy against MRSA

Methicillin-resistant Staphylococcus aureus (MRSA) is a medical concern. Here, we show that branched polyethylenimine (BPEI), a nontoxic, cationic polymer, restores MRSA's susceptibility to β-lactam antibiotics. Checkerboard assays with MRSA demonstrated synergy between BPEI and β-lactam antibiotics. A time-killing curve showed BPEI to be bactericidal in combination with oxacillin. BPEI did not potentiate efficacy with vancomycin, chloramphenicol, or linezolid. When exposed to BPEI, MRSA increased in size and had difficulty forming septa. BPEI electrostatically binds to wall teichoic acid (WTA), a cell wall anionic polymer of Gram-positive bacteria that is important for localization of certain cell wall proteins. Lack of potentiation in a WTA knockout mutant supports the WTA-based mechanism. These data suggest that BPEI may prevent proper localization of cell wall machinery by binding to WTA; leading to cell death when administered in combination with β-lactam antibiotics. Negligible in vitro toxicity suggests the combination could be a viable treatment option.

An analysis of patient-sharing physician networks and implantable cardioverter defibrillator therapy

The application of social network analysis to the organization of healthcare delivery is a relatively new area of research that may not be familiar to health services statisticians and other methodologists. We present a methodological introduction to social network analysis with a case study of physicians' adherence to clinical guidelines regarding use of implantable cardioverter defibrillators (ICDs) for the prevention of sudden cardiac death. We focus on two hospital referral regions (HRRs) in Indiana, Gary and South Bend, characterized by different rates of evidence-based ICD use (86% and 66%, respectively). Using Medicare Part B claims, we construct a network of physicians who care for cardiovascular disease patients based on patient-sharing relationships. Approaches for weighting physician dyads and aggregating physician dyads by hospital are discussed. Then, we obtain a set of weighted network statistics for the positions of hospitals in their referral region, global statistics for the physician network within each hospital, and of the network positions of individual physicians within hospitals, providing the mathematical specification and sociological intuition underlying each measure. We find that adjusting for network measures can reduce the observed differences between referral regions for evidence-based ICD therapy. This study supports previous reports on how variation in physician network structure relates to utilization of care, and motivates future work using physician network measures to examine variation in evidence-based medicine.

Analyzing Clustered Data: Why and How to Account for Multiple Observations Nested within a Study Participant?

A conventional study design among medical and biological experimentalists involves collecting multiple measurements from a study subject. For example, experiments utilizing mouse models in neuroscience often involve collecting multiple neuron measurements per mouse to increase the number of observations without requiring a large number of mice. This leads to a form of statistical dependence referred to as clustering. Inappropriate analyses of clustered data have resulted in several recent critiques of neuroscience research that suggest the bar for statistical analyses within the field is set too low. We compare naïve analytical approaches to marginal, fixed-effect, and mixed-effect models and provide guidelines for when each of these models is most appropriate based on study design. We demonstrate the influence of clustering on a between-mouse treatment effect, a within-mouse treatment effect, and an interaction effect between the two. Our analyses demonstrate that these statistical approaches can give substantially different results, primarily when the analyses include a between-mouse treatment effect. In a novel analysis from a neuroscience perspective, we also refine the mixed-effect approach through the inclusion of an aggregate mouse-level counterpart to a within-mouse (neuron level) treatment as an additional predictor by adapting an advanced modeling technique that has been used in social science research and show that this yields more informative results. Based on these findings, we emphasize the importance of appropriate analyses of clustered data, and we aim for this work to serve as a resource for when one is deciding which approach will work best for a given study.

New themes in the biological functions of 5-methylcytosine and 5-hydroxymethylcytosine

5-methylcytosine (5-mC) and 5-hydroxymethylcytosine (5-hmC) play a critical role in development and normal physiology. Alterations in 5-mC and 5-hmC patterns are common events in hematopoietic neoplasms. In this review, we begin by emphasizing the importance of 5-mC, 5-hmC, and their enzymatic modifiers in hematological malignancies. Then, we discuss the functions of 5-mC and 5-hmC at distinct genic contexts, including promoter regions, gene bodies, intron-exon boundaries, alternative promoters, and intragenic microRNAs. Recent advances in technology have allowed for the study of 5-mC and 5-hmC independently and specifically permitting distinction between the bases that show them to have transcriptional effects that vary by their location relative to gene structure. We extend these observations to their functions at enhancers and transcription factor binding sites. We discuss dietary influences on 5-mC and 5-hmC levels and summarize the literature on the effects of folate and vitamin C on 5-mC and 5-hmC, respectively. Finally, we discuss how these new themes in the functions of 5-mC and 5-hmC will likely influence the broader research field of epigenetics.

Characterization of CpG sites that escape methylation on the inactive human X-chromosome

In many whole genome studies of gene expression or modified cytosines, data from probes localized to the X-chromosome are removed from analyses due to gender bias. Previously, we observed population differences in cytosine modifications between Caucasian and African lymphoblastoid cell lines (LCLs) on the autosomes using whole genome arrays to measure modified cytosines. DNA methylation plays a critical role in establishment and maintenance of X-chromosome inactivation in females. Therefore, we reasoned that by investigating cytosine modification patterns specifically on the X-chromosome, we could obtain valuable information about a chromosome that is often disregarded in genome-wide analyses. We investigated population differences in cytosine modification patterns along the X-chromosome between Caucasian and African LCLs and identified novel sites that escape methylation on the inactive X-chromosome (Xi) in females. We characterized the chromatin state of these loci by incorporating the extensive histone modification ChIP-seq data generated by ENCODE. To explore the relationship between DNA and histone modifications further, we hypothesized that BRD4, a protein that binds acetylated histones, could be preventing some sites from becoming de novo methylated. To test this, we treated 4 female LCLs with JQ1, a small molecule inhibitor of BRD4, but found that JQ1 treatment induced minor changes in cytosine modification levels, and the majority of sites escaping methylation on the Xi remained unmethylated. This suggests that other epigenetic mechanisms or transcription factors are likely playing a larger role in protecting these sites from de novo methylation on the Xi.

Abstract 5461: The role of gene body cytosine modifications in MGMT expression and sensitivity to temozolomide

The goal of precision medicine is to use patient-specific markers to design individualized treatment strategies. Promoter methylation of O6-methylguanine-DNA methyltransferase (MGMT), which encodes for the DNA repair protein MGMT, has been shown to predict glioblastoma patient survival after treatment with temozolomide. Unfortunately, there is no approved therapeutic alternative for patients whose tumors have an unmethylated MGMT promoter and are unlikely to respond to temozolomide treatment. Promoter methylation status of MGMT is used to predict expression levels of MGMT, despite some observed discordance between promoter methylation and protein levels. We hypothesized that incorporation of MGMT gene body cytosine modification levels in models of temozolomide response may lead to better prediction of MGMT expression levels and, more importantly, improved methods of prediction for glioblastoma patient response to temozolomide. Using 91 human glioblastoma samples from The Cancer Genome Atlas, we observed significant variation in MGMT expression levels in patients with an unmethylated promoter, with higher levels of gene body cytosine modification correlating with higher gene expression levels. Furthermore, inducing hypomethylation across the MGMT gene body with decitabine resulted in decreased levels of MGMT gene expression in lymphoblastoid and glioblastoma cell lines. These data indicate an important functional role for gene body cytosine modifications in maintaining gene expression levels. We reasoned that the decrease in MGMT expression induced by decitabine might render resistant glioblastoma cell lines more sensitive to temozolomide. Indeed, MGMT-expressing glioblastoma cell lines pre-treated with decitabine became significantly more sensitive to temozolomide-induced cytotoxicity. Overall, our results demonstrate a functional role for gene body cytosine modification in regulating gene expression of MGMT and suggest that patients with an unmethylated MGMT promoter could benefit from pre-treatment with decitabine to increase tumor sensitivity to temozolomide.

Citation Format: Erika L. Moen, Amy L. Stark, Wei Zhang, M. Eileen Dolan, Lucy A. Godley. The role of gene body cytosine modifications in MGMT expression and sensitivity to temozolomide. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 5461. doi:10.1158/1538-7445.AM2014-5461