A network-based model to explore the role of testing in the epidemiological control of the COVID-19 pandemic

BACKGROUND

Testing is one of the most effective means to manage the COVID-19 pandemic. However, there is an upper bound on daily testing volume because of limited healthcare staff and working hours, as well as different testing methods, such as random testing and contact-tracking testing. In this study, a network-based epidemic transmission model combined with a testing mechanism was proposed to study the role of testing in epidemic control. The aim of this study was to determine how testing affects the spread of epidemics and the daily testing volume needed to control infectious diseases.

METHODS

We simulated the epidemic spread process on complex networks and introduced testing preferences to describe different testing strategies. Different networks were generated to represent social contact between individuals. An extended susceptible-exposed-infected-recovered (SEIR) epidemic model was adopted to simulate the spread of epidemics in these networks. The model establishes a testing preference of between 0 and 1; the larger the testing preference, the higher the testing priority for people in close contact with confirmed cases.

RESULTS

The numerical simulations revealed that the higher the priority for testing individuals in close contact with confirmed cases, the smaller the infection scale. In addition, the infection peak decreased with an increase in daily testing volume and increased as the testing start time was delayed. We also discovered that when testing and other measures were adopted, the daily testing volume required to keep the infection scale below 5% was reduced by more than 40% even if other measures only reduced individuals' infection probability by 10%. The proposed model was validated using COVID-19 testing data.

CONCLUSIONS

Although testing could effectively inhibit the spread of infectious diseases and epidemics, our results indicated that it requires a huge daily testing volume. Thus, it is highly recommended that testing be adopted in combination with measures such as wearing masks and social distancing to better manage infectious diseases. Our research contributes to understanding the role of testing in epidemic control and provides useful suggestions for the government and individuals in responding to epidemics.

An optomechanogram for assessment of the structural and mechanical properties of tissues

The structural and mechanical properties of tissue and the interplay between them play a critical role in tissue function. We introduce the optomechanogram, a combined quantitative and qualitative visualization of spatially co-registered measurements of the microstructural and micromechanical properties of any tissue. Our approach relies on the co-registration of two independent platforms, second-harmonic generation (SHG) microscopy for quantitative assessment of 3D collagen-fiber microstructural organization, and nanoindentation (NI) for local micromechanical properties. We experimentally validate our method by applying to uterine cervix tissue, which exhibits structural and mechanical complexity. We find statistically significant agreement between the micromechanical and microstructural data, and confirm that the distinct tissue regions are distinguishable using either the SHG or NI measurements. Our method could potentially be used for research in pregnancy maintenance, mechanobiological studies of tissues and their constitutive modeling and more generally for the optomechanical metrology of materials.

Clinical Frailty Scale and Biomarkers for Assessing Frailty in Elder Inpatients in China

OBJECTIVE

This study aimed to assess the feasibility of the Clinical Frailty Scale (CFS) and clinical biomarkers in assessing the frailty in elder inpatients in China.

DESIGN

The study was a cross-sectional study.

SETTING AND PARTICIPANTS

The study included 642 elder inpatients (295 females and 347 males) aged ≥65 years, from the Department of Geriatrics of Zhejiang Hospital between January 2018 and December 2019.

MEASUREMENTS

All participants underwent a comprehensive geriatric assessment and blood tests. Univariate and multivariate logistic regression was used to analyze the association between risk factors and frailty.

RESULTS

The average age of the participants was 82.72±8.06 years (range: 65-95 years) and the prevalence of frailty was 39.1% according to the CFS. Frail participants showed significantly lower short physical performance battery (SPPB), basic activities of daily living (ADL) and instrumental activities of daily living (IADL) scores (all p<0.001), and lower hemoglobin, total protein and albumin levels (all P<0.05) than nonfrail participants. Frail participants had higher CRP, D-dimer and fibrinogen levels than nonfrail participants (all p<0.05). Univariate logistic regression analysis showed a significant association between frailty and age, comorbidity, polypharmacy, fall history, SPPB, ADL, and IADL scores, D-dimer, fibrinogen, hemoglobin, total protein and albumin levels (all P<0.05). Multivariate logistic regression analysis indicated that age (odds ratio (OR), 95% confidence interval (CI)= 1.151(1.042-1.272), P=0.006), SPPB scores (OR, 95% CI=0.901(0.601-1.350), P<0.001), and D-dimer (OR, 95% CI=4.857(2.182-6.983), P<0.001), fibrinogen (OR, 95% CI=2.665(0.977-4.254), P<0.001), hemoglobin (OR, 95% CI=0.837(0.725-0.963), P= 0.044), and albumin (OR, 95% CI=0.860 (0.776-1.188), P<0.001) levels were independently associated with frailty in all participants.

CONCLUSION

Frailty in elder inpatients in China is characterized by older age, a lower SPPB scores, higher D-dimer and fibrinogen levels and lower hemoglobin and albumin levels. Functional decline and malnutrition may be the targets of frailty interventions.

Correlations of the expression of γδ T cells and their co-stimulatory molecules TIGIT, PD-1, ICOS and BTLA with PR and PIBF in the peripheral blood and decidual tissues of women with unexplained recurrent spontaneous abortion

Semi-allogeneic embryos are not rejected by the maternal immune system due to maternal-fetal immune tolerance. Progesterone (P) receptor (PR)-expressing γδ T cells are present in healthy pregnant women. In the presence of P, these cells secrete an immunomodulatory protein called progesterone-induced blocking factor (PIBF), which can facilitate immune escape and is important in preventing embryonic rejection. This work investigated the correlations of the expression of γδ T cells and their co-stimulatory molecules T cell immunoglobulin and ITIM domain (TIGIT), programmed cell death 1 (PD-1), inducible co-stimulator (ICOS) and B and T lymphocyte attenuator (BTLA) with progesterone receptor (PR) and progesterone-induced blocking factor (PIBF) in peripheral blood and decidual tissue in women with unexplained recurrent spontaneous abortion (URSA) and normal pregnant (NP) women. We confirmed that γδ T cell proportions and PIBF expression in the peripheral blood and decidua of URSA women decreased significantly, while PR expression in decidua decreased. However, TIGIT, PD-1, ICOS and BTLA expression in γδ T cells in peripheral blood did not change, while TIGIT and PD-1 expression in γδ T cells in decidua increased significantly. Under the action of PHA-P (10 µg/ml), co-blocking of TIGIT (15 µg/ml) and PD-1 (10 µg/ml) antibodies further induced γδ T cell proliferation, but PIBF levels in the culture medium supernatant did not change. At 10^-10 M P, γδ T cells proliferated significantly, and PIBF concentrations in the culture medium supernatant increased. γδ T cells co-cultured with P, TIGIT and PD-1 blocking antibodies showed the most significant proliferation, and PIBF concentrations in the culture medium supernatant were the highest. These results confirm that P is necessary for PIBF production. The TIGIT and PD-1 pathways participate in γδ T cell proliferation and activation and PIBF expression and play important roles in maintaining pregnancy.

A multi-modal virtual reality treadmill intervention for enhancing mobility and cognitive function in people with multiple sclerosis: Protocol for a randomized controlled trial

BACKGROUND

Gait and cognitive impairments are common in individuals with Multiple Sclerosis (MS) and can interfere with everyday function. Those with MS have difficulties executing cognitive tasks and walking simultaneously, a reflection of dual-task interference. Therefore, dual-task training may improve functional ambulation. Additionally, using technology such as virtual reality can provide personalized rehabilitation while mimicking real-world environments. The purpose of this randomized controlled trial is to establish the benefits of a combined cognitive-motor virtual reality training on MS symptoms compared to conventional treadmill training.

METHODS

This study will be a single-blinded, two arm RCT with a six-week intervention period. 144 people with MS will be randomized into a treadmill training alone group or treadmill training with virtual reality group. Both groups will receive 18 sessions of training while walking on a treadmill, with the virtual reality group receiving feedback from the virtual system. Primary outcome measures include dual-task gait speed and information processing speed, which will be measured prior to training, one-week post-training, and three months following training.

DISCUSSION

This study will provide insight into the ability of a multi-modal cognitive-motor intervention to reduce dual-task cost and to enhance information processing speed in those with MS. This is one of the first studies that is powered to understand whether targeted dual-task training can improve MS symptoms and increase functional ambulation. We anticipate that those in the virtual reality group will have a significantly greater increase in dual-task gait speed and information processing speed than those achieved via treadmill training alone.

Supercomputer-Based Ensemble Docking Drug Discovery Pipeline with Application to Covid-19

We present a supercomputer-driven pipeline for in-silico drug discovery using enhanced sampling molecular dynamics (MD) and ensemble docking. We also describe preliminary results obtained for 23 systems involving eight protein targets of the proteome of SARS CoV-2. THe MD performed is temperature replica-exchange enhanced sampling, making use of the massively parallel supercomputing on the SUMMIT supercomputer at Oak Ridge National Laboratory, with which more than 1ms of enhanced sampling MD can be generated per day. We have ensemble docked repurposing databases to ten configurations of each of the 23 SARS CoV-2 systems using AutoDock Vina. We also demonstrate that using Autodock-GPU on SUMMIT, it is possible to perform exhaustive docking of one billion compounds in under 24 hours. Finally, we discuss preliminary results and planned improvements to the pipeline, including the use of quantum mechanical (QM), machine learning, and AI methods to cluster MD trajectories and rescore docking poses.

Influence of an InGaN superlattice pre-layer on the performance of semi-polar (11-22) green LEDs grown on silicon

It is well-known that it is crucial to insert either a single InGaN underlayer or an InGaN superlattice (SLS) structure (both with low InN content) as a pre-layer prior to the growth of InGaN/GaN multiple quantum wells (MQWs) served as an active region for a light-emitting diode (LED). So far, this growth scheme has achieved a great success in the growth of III-nitride LEDs on c-plane substrates, but has not yet been applied in the growth of any other orientated III-nitride LEDs. In this paper, we have applied this growth scheme in the growth of semi-polar (11–22) green LEDs, and have investigated the impact of the SLS pre-layer on the optical performance of semi-polar (11–22) green LEDs grown on patterned (113) silicon substrates. Our results demonstrate that the semi-polar LEDs with the SLS pre-layer exhibit an improvement in both internal quantum efficiency and light output, which is similar to their c-plane counterparts. However, the performance improvement is not so significant as in the c-plane case. This is because the SLS pre-layer also introduces extra misfit dislocations for the semi-polar, but not the c-plane case, which act as non-radiative recombination centres.

Suppression of wood dust explosion by ultrafine magnesium hydroxide

The suppression effects of ultrafine Mg (OH)₂ powders with different particle sizes and mass fractions on explosion flame of wood dust are experimentally studied in a half-closed vertical experimental duct. Flame structures and characteristic parameters, including flame light emission images, propagation velocity, temperature, during the flame propagation of wood dust explosion are recorded by high-speed photography and fine thermocouple. Thermal decomposition behaviors of wood dust and Mg(OH)₂ powders are studied using synchronous thermal analyzer. Chemical structures of residual dust samples after the explosion are characterized by Fourier transform infrared spectroscopy and X-ray photoelectron spectroscopy. The experimental results show that explosion flame of wood dust is obviously suppressed by physical and chemical effects of Mg(OH)₂ powders, and the suppression effect of nano-Mg(OH)₂ is better than that of micron-Mg(OH)₂ under same mass fractions. By analyzing multiple characteristics of nano-powders, the advantages of nano-Mg(OH)₂ over micron-powders are further investigated.

Scenario-Entity Analysis based on an entity-relationship model: Revisiting crime reconstruction

For a crime case, the related physical evidence and information can be termed entities, and there exist different types of relationships between entities. Entity-relationship models connect numerous entities through different relationships, which is useful in crime reconstructions. However, two types of problems may occur that can mislead crime reconstructions in the real world. Specifically, important entities may not be collected and vital relationships may go undiscovered. In this paper, we used an approach based on an entity-relationship model to address these problems. We organized the related entities used to reconstruct crimes according to their physical properties and sorted the relationships between entities through temporal, spatial and logical dimensions. The proposed approach is called 'Scenario-Entity Analysis' (SEA), and it uses several steps for discovering entities and relationships. The SEA also provides a framework for associating events/scenarios with evidence, which is important for crime reconstructions. Using a combination of SEA and Bayesian networks, a three-layered Bayesian network was constructed for uncertainty reasoning. A knife-attack case is then presented to demonstrate the analytical process of SEA.

Genetic diversity and maternal origin of Northeast African and South American donkey populations

To investigate the mtDNA variation and origin of maternal lineages in South American donkeys and to reassess the domestication of donkeys in northeast Africa, we analyzed sequences (489 bp of the D-loop) from 323 domestic donkeys sampled from Peru, Brazil, Ethiopia and Egypt. Altogether, the 323 sequences displayed 53 different haplotypes (45 in Ethiopia, 14 in Egypt, eight in Peru and six in Brazil). Among the four populations, Egyptian donkeys possessed the highest haplotype diversity (0.910 ± 0.032), followed by Brazilian donkeys (0.879 ± 0.060). The Clade I haplotypes dominated in Peruvian donkeys (65%), whereas Clade II haplotypes dominated in Brazilian donkeys (67%). Estimates of F_ST values showed a high genetic differentiation between Peruvian and Brazilian donkey populations (F_ST  = 0.4066), which could be explained by the complex introduction history of South American donkeys. Phylogeographic analysis indicates that northeast Africa could be the most probable domestication center for Clade I donkeys. Analysis of molecular variance confirmed a weak genetic structure in domestic donkey populations among four continents (Europe, Asia, Africa and South America).

A reversed approach for simultaneous mandibular symphyseal split osteotomy and genioplasty

Performing a mandibular symphyseal split and genioplasty simultaneously and accurately is a technical challenge for the surgeon. The aim of this study was to validate a reversed approach for simultaneous symphyseal split and genioplasty. A cutting guide and a repositioning guide were designed and printed three-dimensionally in titanium. The symphyseal split and genioplasty were performed successfully. The accuracy of the technique appears to be appropriate for clinical application.

Abstract P2-14-30: Survival outcomes of breast conserving surgery versus mastectomy for ultrasound detected non-palpable breast cancer in hospital-based screening among Chinese women

Background: Some population-base studies have reported similar or improved survival for breast-conserving surgery (BCS) plus radiotherapy compared with mastectomy (Mx) in early breast cancer [PMID: 22373563, 27344114]. Among the screening detected early breast cancer, ultrasound (US) could detect more invasive non-palpable breast cancer (NPBC) with positive lymph nodes in hospital-based asymptomatic Chinese women, who could achieve comparable 10-year DFS and OS as mammography (MG)-detected NPBC [2016 SABCS P5-02-05, PMID: 27689334]. However, there is little data about the surgical outcomes of BCS verse Mx in the low-risk screening detected NPBC with US as the initial imaging test.

Methods: From 2001 to 2017, 6,423 consecutive asymptomatic women underwent mammography or ultrasound guided biopsy in Peking Union Medical College Hospital. Among them, 1130 NPBC including 914 US-detected and 216 MG-detected NPBC were diagnosed and treated. There were 349 (30.9%) patients underwent BCS including 286 (25.3%) patients received radiation therapy and 63 (5.6%) elderly patients (&gt;70 years) who did not. The clinicopathological features, treatment choice, 10-year disease-free survival (DFS) and overall survival (OS) were compared between breast conservingsurgery (BCS) versus mastectomy(Mx) in all NPBC and between the US-detected and MG-detected NPBC.

Result: Compared to those who received BCS, the 781 (69.1%) patients who underwent Mx had more cancers with relatively higher histologic grade (p=0.003), positive lymph node (18.8% vs 12.0%, p=0.005), ER-negative (22.5% vs 11.5%, p&lt;0.001), PR-negative (29.6% vs 16.3, p&lt;0.001), Her2-positive (16.3% vs 8.9%, p=0.001), and received chemotherapy (37.6% vs 28.7%, p=0.003). The breast conserving rates of US-NPBCwere higher than that of MG-NPBC (32.6% vs 23.6%, p=0.010), but the breast conserving rates were similar between ductal carcinoma in situ (DCIS) and invasive cancers. The 10-year DFS and OS were similar among BCS with radiation therapy, BCS without radiation therapy and Mx as well as among US-NPBC with BCS, US-NPBC with Mx, MG-NPBC with BCS and MG-NPBC with Mx. However, MG-NPBC with Mx had favorable 10-year DFS than that of MG-NPBC with BCS (p=0.041).

Table 1.Kaplan-Meier estimated 10-year DFS and OS of all NPBC§Patients (No.)NPBC GroupNumber (%)10-year DFS (%)P value10-year OSP valueAll NPBC (1130)BCS without Radiotherapy63 (5.6)85.00.10592.30.722 BCS with Radiotherapy286 (25.3)92.7 99.5  Mastectomy781 (69.1)93.2 98.7        All NPBC (1130)US+BCS298 (26.4)90.40.24896.30.542 US+Mx616 (64.5)92.4 98.4  MG+BCS51 (4.5)90.3 100.0  MG+Mx165 (14.6)96.1 100.0 § Kaplan-Meier survival curves would be displayed in the poster.

Conclusion: The 10-year DFS and OS of breast conserving surgery versus mastectomy were similar among all NPBC patients. As the current initial imaging test, US-detected NPBC patients would receive significantly more BCS compared to MG. There was no significant difference in surgical outcomes among BCS and Mx in US-detected NPBC. However, among MG-detected NPBC, patients with Mx reached a better DFS but a similar OS than those with BCS. The radiation therapy could be safely omitted in the elderly patients (&gt;70 years) with NPBC.

Citation Format: Yao R, Pan B, Xu Y, Zhou Y, Zhu Q, Zhang J, Wu H, Mao F, Lin Y, Shen S, Sun Q. Survival outcomes of breast conserving surgery versus mastectomy for ultrasound detected non-palpable breast cancer in hospital-based screening among Chinese women [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-14-30.

Abstract P1-02-02: Is ultrasound screening justified for non-palpable breast cancer in asymptomatic Chinese women: A real-world study based on long-term survival of consecutive cohort (2001-2017)

Background: Ultrasound (US) is an effective initial screening test for breast cancer both in Caucasian and Chinese women [PMID: 26712110, 26715161, and 25668012]. The real-world modality of breast cancer screening in the China is hospital-based screening among asymptomatic self-referred women. In our previous study, we showed that US and mammography (MG) detected non-palpable breast cancer (NPBC) had similar long-term survival and that US detected more invasive NPBC with positive lymph node [2016 SABCS P5-02-05, PMID: 27689334]. This study was to investigate whether these findings would be still true with more NPBC cases included and longer follow-up in the consecutive hospital cohort.

Methods: From 2001 to 2017, 5,264 asymptomatic women with positive (BI-RADS 4 and 5) initial screening US underwent biopsies in PUMC Hospital, and 914 US-NPBC in 883 women were diagnosed. Meanwhile, women without dense breasts (defined as BI-RADS category C and D) also received screening MG after physical examination and US. There were 1,159 patients with positive (BI-RADS 4 and 5) MG and normal US (BI-RADS 1, 2 and 3) underwent MG-guided biopsies and 216 MG-NPBC were diagnosed in 214 women. The clinicopathological characteristics and 10-year disease-free survival (DFS) and overall survival (OS) were reviewed and compared between the US-NPBC and MG-NPBC. Prognostic factors of NPBC were identified by univariate and multivariate Cox analysis.

Result: Compared to MG, US could detect more invasive (81.2% vs 48.6%, p&lt;0.001), lymph node positive (18.3% vs 10.2%, p&lt;0.001), stage II+III (21.7% vs 12.5%, p&lt;0.001) and low grade cancer (p=0.001).Between invasive US-NPBC and MG-NPBC, no significant difference was identified for lymph node status, TNM stage or subtype.US-NPBC received more breast conserving surgery (32.6% vs 24.1%, p&lt;0.001) and chemotherapy (37.5% vs 23.6%, p&lt;0.001). There was no significant difference in DFS or OS between US- vs MG-NPBC among ductal carcinoma in situ (DCIS), invasive and all NPBC. For the US-NPBC, the DFS factors included TNM stage and Hormone receptor status whereas OS-predictors were pN and subtype.

Table 1.Kaplan-Meier estimates of DFS and OS between US-NPBC and MG-NPBC§Patients (No.) 10-year DFS (%)P value10-year OS (%)P valueAllUS-NPBC (914)92.40.57098.20.143 MG-NPBC (216)94.7 100.0 DCISUS-NPBC (172)97.70.170100.0- MG-NBPC (111)95.3 100.0 InvasiveUS-NPBC (742)91.20.45897.90.251 MG-NPBC (105)94.4 100.0 § Kaplan-Meier survival curves between each two subgroups would be displayed in the poster.

Conclusion: Overall, US could detect more invasive NPBC patients with positive lymph node and advanced stage compared to MG, and screen invasive NPBC at similar TNM stage and subtype distribution as MG. US-NPBC patients received more breast conserving surgery and chemotherapy, and could achieve comparable 10-year DFS and OS as MG-detected NPBC. Hence US is justified in the real-world as the initial imaging modality in hospital-based screening Chinese women.

Citation Format: Yao R, Pan B, Xu Y, Zhou Y, Zhu Q, Zhang J, Wu H, Mao F, Lin Y, Shen S, Sun Q. Is ultrasound screening justified for non-palpable breast cancer in asymptomatic Chinese women: A real-world study based on long-term survival of consecutive cohort (2001-2017) [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P1-02-02.

Abstract PD6-10: Association between adherence to cardiovascular medications and cardiovascular events following a diagnosis of early stage breast cancer

Background: Studies show that patients diagnosed with early-stage breast cancer (BC) are more likely to die from cardiovascular disease (CVD) than BC. Adherence to CVD medications, such as statins and antihypertensives, is poor in BC survivors, particularly in the year following diagnosis. The impact of non-adherence to CVD medications on cardiovascular events in BC survivors is unknown.

Methods: Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked dataset, we evaluated patients with non-metastatic BC who were diagnosed between 2006-2014. Prescriptions were identified for the treatment of hypertension, hyperlipidemia and diabetes. The pre-cancer diagnosis study period for adherence was defined as 1 year prior to the diagnosis of cancer. The follow up adherence period was between years 1 and 2 following the diagnosis of cancer, so the BC treatment period was not included. Adherence was defined as a medication possession ratio of 380%. A CVD event was defined as an ischemic event or acute heart failure. Patients with a CVD event prior to diagnosis were excluded. Logistic regression was performed for each non-cancer condition to define factors associated with medication non-adherence. Cox regression was used to calculate the association between CVD medication adherence and time-to-subsequent cardiac events, adjusted for baseline factors. Cox regression was performed separately for each non-cancer condition.

Results: Among 23,080 women with BC in the cohort, 15,576 were adherent to at least one CVD medication prior to diagnosis, and of these, 2732 (17.5%) were non-adherent to at least one medication following treatment. Among the women adherent to medications prior to diagnosis, 19.2% were non-adherent to hypertension medications, 26.2% were non-adherent to cholesterol medications, and 30.6% were non-adherent to diabetes medications following the first year of BC treatment. Factors that were associated with non-adherence to anti-hypertensives included receipt of chemotherapy (OR 1.24, p&lt;0.001), other comorbidities (OR 1.34, p&lt;0.001), higher stage (OR 1.18, p &lt;0.001) and hormone receptor negative tumors (OR 1.15, p&lt;0.001). Similar factors were associated with non-adherence to cholesterol medications, whereas only stage and tumor type were associated with non-adherence to diabetes medications. Non-adherence to hypertension medications compared to adherence following diagnosis was associated with an increased risk of having a CVD event (HR 1.33, 95% CI 1.18-1.51, p&lt;0.001; 5-year cumulative incidence of 32% vs 26%, respectively, p&lt;0.001). Similar results were seen for adherence to cholesterol medications (HR 1.21, 95% CI 1.05-1.40, p=0.009) and diabetes medications (HR 1.31, 95% CI 1.09-1.56, p=0.003).

Conclusions:In summary, we found that a large proportion of women who were previously adherent to their medications to prevent CVD events prior to their breast cancer diagnosis were non-adherent following treatment. Of concern, non-adherence to any of these classes of medications resulted in an increased risk of having a cardiovascular event. Improving outcomes and reducing morbidity following a breast cancer diagnosis also requires focused attention on non-breast cancer conditions.

Citation Format: Hershman DL, Accordino M, Shen S, Buono D, Crew KD, Kalinsky K, Trivedi MS, Unger JM, Wright JD. Association between adherence to cardiovascular medications and cardiovascular events following a diagnosis of early stage breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD6-10.