HMS Theses and Dissertations

Benign prostatic hyperplasia (BPH) is one of the most commonly diagnosed chronic diseases in older men associated with a gradual deterioration of lower urinary tract symptoms (LUTS). 5-α reductase inhibitors (5ARIs) are among the most commonly used pharmacological treatments for BPH(1, 2). They act by inhibiting the enzyme 5α-reductase, promoting prostatic apoptosis and prostate size reduction in order to improve urinary symptoms. Although multiple clinical trials showed 5ARIs could reduce the risk of AUR or need for surgery(3-7), In practice, poor adherence to 5ARIs treatment remains to be a problem, which could lead to a reduced efficacy. The relationship between levels of adherence and long-term clinical outcomes has not been determined. Despite the fact that 5ARIs are widely used for benign prostatic hyperplasia, several preclinical studies have demonstrated that treatment with an anti-androgen such as 5ARIs could potentially decrease the incidence of bladder cancer(8, 9). Limited results from observational studies failed to reach a consensus on whether there was a difference in the incidence of bladder cancer between men who received 5ARIs and those who did not(10, 11). In our study we utilized big data generated from medical claims by the Partners Center for Population Health Evaluation and Research Unit to investigate the effect of 5ARIs therapy on men diagnosed with BPH. In the first part of the study, we focused on evaluation of drug adherence and long-term clinical outcomes. In the second part, we investigated if 5ARIs therapy affect on bladder cancer incidence.

Background Chronic stress associates with higher mortality among individuals with cancer; however, the mechanisms underlying this association remain incompletely understood. Quantification of amygdalar metabolic activity (AmygA) using 18F-fluorodexoyglucose positron emission tomography/computed tomography (18F-FDG-PET/CT) provides a measure of stress-related neurobiological activity, that associates with increased systemic inflammation and a higher risk of cardiovascular disease events. Accordingly, we sought to test the hypotheses, in individuals with active cancer, that higher AmygA at cancer staging associates with:1) increased inflammation, and 2) higher mortality. Methods We studied 240 patients with head and neck cancer (HNCA) at a single academic center who underwent 18F-FDG-PET/CT imaging as part of initial cancer staging. AmygA and metabolic activity of the bone marrow (a measure of leukopoiesis) were measured using validated methods. Image analysis, clinical data assessments and adjudication were performed by mutually blinded, independent teams. The primary outcome of interest, the association between AmygA and mortality, was determined using Cox proportional hazard models. The secondary outcome of interest was a composite of the mortality plus cancer progression. Findings Among individuals with HNCA (age 59±13 years; 30% female), 67 died over a median follow-up period of 3 years (IQR: 1·7-5·1). AmygA associated with heightened bone marrow activity, leukocytosis, CRP (P<0·05 each) and cancer progression. In adjusted analyses, AmygA associated with subsequent mortality (1·28, [1·01-1·61], P=0·039) and the association between AmygA and cancer mortality persisted when analyses were restricted to patients with advanced cancer stage (P<0·001). Individuals within the highest tertile of AmygA experienced a 2-fold higher mortality rate compared to others (P=0·01). Interpretation AmygA, quantified on routine 18F-FDG-PET/CT images obtained at cancer staging, independently and robustly predicts mortality via a mechanism which may, in part, be mediated through increased leukopoietic activity and inflammation and cancer progression. Future studies should test whether strategies that attenuate AmygA (or its downstream biological consequences) may improve cancer survival.

Project 1 Background: We assessed if machine learners outperform Cox-PH models in a high dimensional setting in the context of anti-PD-1 treatment for melanoma brain metastases as accurate prediction of overall survival/OS is critical to prioritize limited treatment options. Methods: In this retrospective cohort study, metastatic melanoma patients had radiological evidence of brain metastases (treated/untreated) at the time of anti-PD-1 (monotherapy/combination) immunotherapy (June-2014–September-2016). The primary outcome was OS. Five tasks were evaluated (Task-1: 22-features; Task-2: diagnosis specific-graded prognostic assessment/ds-GPA; Task-3: Recursive Partitioning Analysis/RPA; Task-4: ds-GPA plus 20-features; Task-5: RPA plus 21-features) using machine learners (Random Survival Forests/RSF, Boosting, Cox-Boost, Survival-Tree) and Cox-PH models based on test-set performance (5-fold cross-validated C-index) in a benchmark experiment followed by feature filtering and sequential forward selection (best performing learner). Results: The median OS was 59 days (IQR:13-97) for ds-GPA 0-1, 473 days (IQR:150-Not Reached/NR) for ds-GPA 2 and was NR for ds-GPA 3-4 (N=56 patients/33 deaths). RSF had the lowest mean rank and performed significantly better than Cox-PH (mean-rank: 1.3vs.4.3; Critical-Difference=2.73; p-value=0.023). Average C-index for Task-2/ds-GPA (0.867;95%CI:0.777-0.947) and Task-4 (0.878; 95%CI:0.760-0.961) feature-subset was comparable. Task-3/RPA performed poorly (C-index:0.664; 95%CI:0.560-0.779). Task-4 subset included ds-GPA, ≥4 extra-cranial metastases, prior surgical resection and stereotactic radiotherapy/SRT, neurological symptoms and spine metastases. Prior surgical resection and SRT predicted improved survival for ds-GPA>1. Conclusions and Impact: RSF had the best relative performance compared to Cox-PH models in a high dimensional setting. OS was prolonged for ds-GPA>1 with anti-PD-1 immunotherapy compared to historical estimates (surgery and/or radiation). Baseline dsGPA was the strongest predictor. Project 2 Background: Obesity paradox (improved survival in overweight/obese patients compared to normal weight) is well documented in cancer but not in metastatic melanoma. We characterized the relationship of Body Mass Index (BMI) with survival outcomes and explored complex interactions in the context of PD-1 blockade. Methods: Unresectable or metastatic melanoma patients who received at least one dose of pembrolizumab, nivolumab, or nivolumab plus ipilimumab (combination) from June- 2014-September-2016 were included in this retrospective cohort study. Overall Survival and Progression Free Survival were the main outcomes. Analysis was performed using Random Survival Forests (RSF) and multi-variable Cox Proportional Hazards models. Results: A “U” shaped relationship was noted for pre-treatment BMI where overweight/Class-I (25-<35kg/m2) obese patients had a lower risk of mortality (HR:0.26; 95%CI:0.1-0.71; p-value=0.008) and progressive disease (HR:0.43; 95%CI:0.19-0.95; pvalue:0.038) compared to normal-weight (18.5-<25kg/m2) (Total N=139). Exploration of interactions (RSF) showed that the association was predominantly driven by males; it was attenuated in patients with serum creatinine<0.9mg/dL, who were predominantly females. Findings were similar for anti-PD-1 monotherapy(n=79)/combination(n=60). Conclusions: The findings support the presence of “obesity paradox” where overweight/Class-I obesity was associated with substantially lower risk of mortality and progressive disease compared to normal weight; this association was driven predominantly by males who largely had higher serum creatinine levels (surrogate for skeletal muscle mass in the setting of metastatic disease). These findings suggest that instead of BMI alone, sarcopenia (low skeletal muscle mass) or direct measures of body mass composition may be more suitable predictors of survival in melanoma patients treated with PD-1 blockade.

OVERVIEW: During the curriculum of Master of Medical Sciences Program at Harvard Medical School, I have been focusing on two inferential goals through respiratory disease research—1) prediction and 2) causal inference—which are central pillars of research questions as clinical investigators. Bronchiolitis in infants and asthma in adults are two of the major respiratory diseases in industrialized nations. Indeed, bronchiolitis is the leading cause of infant hospitalization in the US, accounting for 107,000 infant hospitalizations each year. Besides, previous literature suggests that bronchiolitis during infancy is one of the major risk factors for subsequent asthma. Likewise, approximately 26 million Americans have asthma. Asthma exacerbations account for a substantial proportion of the personal and societal burden, leading to approximately 340,000 hospitalizations annually. Paper 1: The literature has demonstrated a high variability in bronchiolitis management across the nation, which is, at least partially, attributable to the variable clinical course in this population. Despite the development of prediction scoring models (e.g., logistic regression models), identifying the subgroup of infants with bronchiolitis who require higher acuity care (e.g., positive pressure ventilation, intensive care unit [ICU] admission) remains an important challenge. The difficulty and uncertainty of predicting acute severity—and, consequently, the appropriate level of care for infants with bronchiolitis—are reflected by the well-documented variability in inpatient management across the nation. Recently, machine learning models have gained increasing attention because of their advantages, such as the ability to incorporate high-order, nonlinear interactions between predictors and to yield more accurate and stable predictions. While machine learning approaches may enhance the predictive ability, little is known about their utility to predict acute severity outcomes in infants with bronchiolitis. We developed advanced machine learning models using the data from a well-characterized prospective cohort study. Paper 2: Previous research has demonstrated that patients with chronic asthma have a high incidence of cardiovascular events. Besides, the emerging evidence suggests that acute inflammatory processes (e.g., bacteremia, pneumonia, influenza virus infection, Streptococcal infection, and acute exacerbation of COPD) have been linked to acute cardiovascular outcomes. However, despite the clinical and research importance of asthma exacerbations, little is known about its acute effect on cardiovascular outcomes. Herein, using large population-based data of four diverse states in the US, I conducted a self-controlled case series study to investigate the acute causal effect of asthma exacerbation on cardiovascular events.

Outpatient cardiac rehabilitation programs provide supervised exercise training in conjunction with other secondary prevention interventions. They are designed to speed recovery from acute cardiovascular events such as myocardial infarction, coronary artery bypass grafting, percutaneous coronary intervention, angina, heart failure, heart transplant and after heart valve procedure to improve quality of life1. Aortic Stenosis is one of the most common types of valvular heart disease in elderly with reported prevalence as high as 12-26%2. The most common etiology of mitral stenosis is rheumatic heart disease; however, it is not common that the disease remains undiagnosed up until an advanced age. Functional mitral stenosis related to massive annular calcification and reduced leaflet excursion has been reported in 2.5-18.0% of elderly patients. Mitral regurgitation is the most frequent valvular heart disease in patients over the age of 65 years. Elderly patients account for about 40% of all patients with mitral regurgitation and 4.5% are over 80 years of age. Mitral valve diseases are increasing in prevalence and require valve replacement in severe diseases. Patients after mitral valve procedure do go to cardiac rehabilitation but minimal clinical studies have done to show magnitude of benefit for cardiac rehabilitation in this patient population. Randomized clinical trials have shown improvements (both physical and psychological) in patients undergoing cardiac rehabilitation after aortic valve replacement, so we compared mitral valve patients with them to see any benefits. We evaluated baseline exercise capacity and psychological well-being for mitral valve patients participating in cardiac rehabilitation and compare physical and psychological outcomes between mitral valve and aortic valve patients. In our 2nd paper we discussed importance of guideline directed medial therapy in cardiovascular diseases. As cardiac diseases are the most common cause of morbidity and mortality in USA, it is important to diagnose and treat effectively for secondary prevention of cardiovascular diseases. Patients with coronary artery disease & congestive heart failure need to be on guided medical therapy for recovery from primary event as well as for prevention of further cardiovascular events. Cardiac Rehabilitation programs have shown improvements in secondary prevention of cardiovascular events; however, no clinical studies are available to show prevalence of guideline directed medical therapy in this patient population. Also, we can identify predictors of better medication prevalence in patient with coronary artery disease and heart failure.

Background: Despite significant progress in primary prevention, the rate of MI has not declined in young adults. Objectives: We aimed to evaluate statin eligibility based on the 2013 American College of Cardiology / American Heart Association (ACC/AHA) guidelines for treatment of blood cholesterol and 2016 United States Preventive Services Task Force (USPSTF) recommendations for statin use in primary prevention in a cohort of adults who experienced a first-time myocardial infarction (MI) at a young age. Methods: The YOUNG-MI registry is a retrospective cohort study from two large academic centers which includes patients who experienced an MI at 50 years of age or younger. Diagnosis of Type 1-MI was adjudicated by study physicians. Pooled cohort risk equations (PCE) were used to estimate atherosclerotic cardiovascular disease (ASCVD) risk score based on data available prior to MI or at the time of presentation. Results: Of 1685 patients meeting inclusion criteria, 210 (12.5%) were on statin therapy prior to MI and were excluded. Among the remaining 1475 individuals, the median age was 45 years, there were 294 (20%) women, and 846 (57%) had STEMI. At least one cardiovascular risk factor was present in 1225 (83%) patients. The median 10-year ASCVD risk score of the cohort was 4.8% (interquartile range: 2.8, 8.0). Only 724 (49%) and 430 (29%) would have met criteria for statin eligibility per the 2013 ACC/AHA guidelines and 2016 USPSTF recommendations, respectively. This finding was even more pronounced in women, in whom 184 (63%) were not eligible for statins by either guideline, compared with 549 (46%) of men (p<0.001). Conclusions: The vast majority of adults who present with an MI at a young age would not have met current guideline-based treatment thresholds for statin therapy prior to their MI. These findings highlight the need for better risk assessment tools among young adults.

Aims: We sought to compare the risk factor profiles and outcomes between very young (age ≤40) and young (40< age ≤50) individuals who experienced a first-time MI at a young age. Methods: We evaluated all patients ≤ 50 years of age who were admitted with a type 1 MI to two large academic hospitals from 2000 to 2016. Risk factors were determined by review of electronic medical records. The primary outcomes of interest were all-cause and cardiovascular mortality. Results: Among 2097 consecutive young patients with MI, a total of 431 (20.5%) were ≤40 years of age. When compared with their older counterparts, very young patients had similar risk profiles with the exception of greater substance abuse (17.9% vs. 9.3%) and less hypertension (37.9% vs. 50.9%). There was a higher proportion of spontaneous coronary artery dissection in very young patients (3.1% vs. 1.1% p=0.003). Upon discharge, very young MI patients were less likely to be prescribed a statin (85.9% vs. 89.9% p=0.017) and aspirin (92% vs. 95.2%, p=0.011). Over a median follow-up of 11.2 years, very young MI patients had a similar risk of all-cause and CV mortality. Conclusions: Despite being on average 10 years younger and having a lower prevalence of hypertension, very young MI patients had similar one-year and long-term outcomes when compared to those who were aged 41 to 50 at the time of their index infarction. Our findings suggest the need for aggressive secondary prevention measures in very young patients who experience an MI.

Recent evidence has demonstrated association between dysbiosis of oral and gut microbiota with Cardiometabolic risk factors (CMR). However, the nature of dysbiosis and its implication on cardiometabolic risk phenotypes (obesity and hypertension) is not clear. In this longitudinal Kuwait Healthy Lifestyle Study (KHLS), I investigated whether changes in oral microbiota can predict changes in weight gain (Project-I), and development of hypertension (Project-II) in Kuwaiti adolescents. In investigating oral microbial changes with the development of obesity, we observed a higher Firmicutes-to-Bacteroidetes ratio, in overweight phenotype. Increased abundance of Firmicutes tends to digest otherwise indigestible polysaccharides – which play a key role in the pathogenesis of obesity. Besides, higher proportions of pro-inflammatory, Lipopolysaccharide (LPS) containing Prevotella sp., were observed in parallel with, a significant increase in CRP levels, in obese subjects, which is indicative of active systemic inflammation. We also observed that a reduction in nitrate-reducing (Actinomyces sp.), and nitrite-reducing (Streptococcus sp.), oral bacteria was associated with an increase in systolic, diastolic, and mean arterial blood pressure. In summary, we identified that changes in oral microbiota can predict the development of obesity and hypertension in adolescent children. Moreover, these changes in oral microbiota were modulated by changes in systemic inflammatory and metabolic pathways.

Purpose Biomarkers for early detection of disease progression in MBC are urgently needed. We explored on-treatment increase in ctDNA fraction as an early prognostic biomarker of subsequent disease progression.
Methods Eighty four paired samples from 70 patients with MBC were analyzed for levels of ctDNA, CEA, and CA 15-3. Baseline specimens were collected at the time of initiation of a new line of therapy and on-treatment specimens were obtained 4-12 weeks later. Each patient underwent a CT scan of chest, abdomen and pelvis on 4-16 weeks (on average 7 weeks) later. Sequencing panel of 73 genes was used to quantify ctDNA. Genomic progression was predefined as an increase ≥25% in total MAF from baseline to on-treatment, in patients with MBC. Progression Free Survival (PFS) was estimated using univariate and multivariate Cox proportional hazards models, adjusting for tumor hormonal status, age at metastatic diagnosis, time elapsed and number of prior therapies received. Results Tumor-specific genetic mutations were detected in 89.3% of baseline and 84.5% of on-treatment specimens. Median follow-up time was 14.5 months (1.5-35.9 months).Patients with total average MAF increase ≥25% were more than twice as likely to have subsequent radiologic progression (OR= 2.04, CI 1.75-2.38, p<0.0001) and significantly shorter PFS (median 4.7 vs. 8.7 months; HR=2.57, 95% CI 1.57-4.22, log-rank p=0.0001), adjusting for other prognostic variables in multivariable analysis. Rise in CA 15-3 levels was prognostic of radiologic progression but not PFS. CEA level increase was not prognostic of either outcome.
Conclusion Rise in ctDNA fraction as a surrogate of genomic progression during MBC treatment could be an independent biomarker of early disease progression before radiologic and clinical progression, and warrants further evaluation in additional studies.

Abstract 1 Background: To define the symptoms, signs and treatment outcomes in a population of older patients with Cushing’s disease (CD). Methods: We analyzed the clinical presentation and treatment outcomes in 45 CD patients older than 60 years, in comparison to 90 CD patients younger than 60, and a control group of 45 older patients with nonfunctioning pituitary adenomas. We reviewed preoperative clinical characteristics, medical comorbidities, imaging findings and endocrine testing as well as surgical and endocrine outcomes. Results: Older CD patients had significantly lower body mass index (BMI) (p = 0.031), were more likely to have muscle wasting (p = 0.006) and women were less likely to have hirsutism (p = 0.033). Older patients with CD had more medical comorbidities than younger patients, which correlated with a higher ASA grade (p < 0.001), but the surgical complication rates were similar in both groups. Surgical remission was achieved in 38/45 (84.4%) older patients and 78/90 (86.7%) younger patients (p = NS). Recurrent disease was more frequent in younger patients (19.2%) in comparison to older patients (2.6%, p = 0.019). Conclusions: Older patients with CD appear to have a distinct phenotype with a more catabolic picture, including a lower BMI and greater prevalence of muscle wasting. Surgical outcomes are similar without a significant age-related increase in complications. Abstract 2 Background: Cushing’s disease (CD) is most commonly caused by microadenomas, which at surgical exploration may not provide adequate tissue for pathologic diagnosis using standard techniques. We wished to determine the accuracy of intraoperative pathologic examination and whether the addition of intraoperative cytology increased the diagnostic yield. Methods: We reviewed the pathology reports from 403 operations on 341 patients with CD microadenomas from a single institution. The concordance rates of intraoperative diagnoses (cytology and frozen) with the final (paraffin section) pathological diagnosis was calculated. The overall pathologic confirmation of an adenoma (by either cytology, frozen or paraffin section) was compared with the result from a historical cohort (only using standard frozen section procedures but not intraoperative cytology) and the pooled results from a meta-analysis of previously published data. Results: The concordance rate between frozen section diagnosis and paraffin section histology was 390/403(96.8%), and the discordance rate was 13/403(3.2%). Intraoperative smears were prepared by the surgeon in 246 operations with an ambiguous or a minute tumor, when the surgeon suspected that there would be insufficient tissue for traditional frozen section and paraffin diagnosis. The concordance rate between cytological smear and paraffin section histology was 213/246(86.6%), and the discordance rate was 33/246(13.4%). In 54 cases (13.4%) with ultimate remission, pathologic confirmation was obtained only on intraoperative pathology (frozen section or cytology). Overall, pathologic confirmation was obtained in 326 operations (80.9%) by at least one pathological modality. The overall pathological confirmation rate of an adenoma was greater after the introduction of intraoperative cytology when compared to the historical control (67.1%, p = 0.015), and compared to the pooled rate of published data from the meta-analysis (72.1%, p < 0.001). Conclusion: Our findings suggest that intraoperative frozen and cytological section analyses during surgery for CD may be additional useful diagnostic tools for both neurosurgeons and neuropathologists.

Over the last two decades, various drugs have been approved for the treatment of osteoporosis, including alendronate, risedronate, ibandronate, zoledronic acid, estrogen replacement therapy, selective estrogen receptor modulators, denosumab, teriparatide, and abaloparatide. Comparative effectiveness evidence between different osteoporosis medications can provide valuable information for clinical and health policy decision-making. However, data on the comparative effectiveness in clinical settings are limited. Electronic medical records (EMR) database can provide ample longitudinal information on typical patients and is a useful data source for generating comparative effectiveness evidence. However, the feasibility of using EMR to do comparative effectiveness study of osteoporosis drugs has not been well examined before. There are several challenges that need to be addressed before EMR can be used to generate reliable evidence. First, most EMR databases include a relatively small number of patients, which hamper their ability to test the reduction in the risk of fracture between various anti-osteoporosis agents. Second, bone mineral density (BMD) changes, usually examined by dual-energy X-ray absorptiometry (DXA), is a well-recognized proxy of anti-osteoporosis efficacy, but the frequency of DXA test in routine clinical practice has not been well examined as well as its application as a comparative effectiveness endpoint. Third, drug data in most EMR databases are prescription data. Verification studies or manually chart review are typically needed to accurately define the drug exposure. In this project, we examined the feasibility of doing comparative effectiveness research (CER) using the EMR database. First, we examined the frequency and patterns of DXA tests in routine clinical practice and determined whether BMD data were rich enough for CER study. Second, we emulated a hypothetical RCT comparing BMD changes of teriparatide versus denosumab in patients switched from bisphosphonates to answer this clinically important question while conclusive RCT is not currently available. This project highlights the strengths and pitfalls of doing a CER study of osteoporosis medications using EMR. With the appropriate study design and analytical approaches, EMR can be a very useful data source for generating comparative effectiveness evidence of osteoporosis medications.

Tuberculosis (TB) is a deadly infectious disease caused by Mycobacterium tuberculosis (M. tuberculosis). Currently, it is the leading infectious disease killer in the world ranking above HIV/AIDS. Furthermore, the continued emergence and spread of drug-resistant forms of TB are increasingly a problem, causing increased mortality and global economic loss. One of the fundamental gaps that exists in the control of TB is the lack of rapid, ultra-sensitive, inexpensive, and easy-to-use diagnostic tools. The consequences of this setback are particularly acute in remote and low-income nations.

While there have been significant advances in the TB diagnostic development pipeline, control and management of TB remain a huge problem in many parts of the world. For instance, in remote and resource-limited settings, the lack of access to TB testing, coupled with high costs of tests, present significant barriers to TB elimination. Moreover, many existing TB diagnostic tests are known for their poor sensitivities, specificities, as well as lack of portability and ability to rapidly identify and discriminate among TB pathogens. Rapid and portable diagnostics that can both identify TB pathogens and provide drug susceptibility data in real-time would transform patient management and critical public health issues such as the current antibiotic resistance crisis. Their application would reach broadly from clinics and primary care offices to tertiary care hospitals, providing immediate guidance for therapeutic intervention thereby resulting in more prudent and appropriate use of antibiotics, in some cases with mortality benefit.

The objective of this thesis work is to explore the utility of the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-based technology in the rapid detection of TB and point-of-care applications. Here, we exploited the CRISPR-Cas-based RNA-detecting platform termed SHERLOCK (Specific High sensitivity Enzymatic Reporter unLOCKing) which utilizes a programmable RNAse (Cas13a) in the detection of genetic targets. We capitalize on targeting a multicopy mobile genetic element IS6110 to rapidly identify M. tuberculosis. Following in vitro transcription and deployment of Cas13a-CRISPR RNA complex, a robust, positive SHERLOCK signal was observed in the presence of fragmented and intact genomic DNA of the H37Rv strain. Ongoing experiments currently focus on developing multiplexing strategies to improve the Limit of Detection (LOD). In the context of patient care, it is envisioned that the CRISPR-based test would be applicable in multiple clinical settings, including latent TB screening in communities with high TB burden, as well as triage testing of high-risk patients according to WHO’s recommendations. Characteristics of the triage TB test would include low cost, rapid turnaround time, easy-to-use, high sensitivity and specificity (comparable to standard confirmatory testing), good stability in a broad range of temperature and humidity, versatility in testing non-sputum samples, and usability in peripheral health facilities.

The rate of autoimmune diseases is on the rise worldwide. Autoimmune thyroid disease is the most common autoimmune disease. No known cures have been found for autoimmune diseases; however, environmental factors, including dietary proteins and chemicals, may influence the expression of the disease by mechanisms of cross-reactivity. Immunologic cross-reactivity occurs when adaptive immune response against one antigen also occurs in another antigen that shares the amino acid sequence homology. The first manuscript in this thesis titled, “Immunological reactivity using monoclonal and polyclonal antibodies of autoimmune target sites with dietary proteins,” investigates the role of 204 purified dietary proteins for cross-reactivity in autoimmune target sites of the thyroid axis. Cross-reactivity can also occur when chemicals bind to proteins and induce a conformational change in the macromolecule, thereby creating new peptide structures that act as neoantigens. When epitopes of neoantigens share surface topology similar to other binding sites, there is a potential for development of cross-reactive interactions. The second manuscript in this thesis titled, “Reaction between antibodies to thyroid axis target sites and chemicals bound to human serum albumin,” investigates the role of commonly encountered chemicals in autoimmune thyroid disease target sites. The identification of immunological cross-reactivity in the thyroid axis target sites supports the understanding of how specific environmental triggers may interfere with thyroid hormone replacement, disrupt hormone metabolism, act as a potential triggering agent for thyroid disease, non-thyroid illness syndrome, and thyroid autoimmune diseases such as Hashimoto’s and Grave’s. Both manuscripts, followed by a conclusion and discussion of their combined role in the thyroid disease, are listed in this thesis.

Direct endoscopic necrosectomy (DEN) with lumen-apposing metal stents (LAMS) is increasingly used in the management of pancreatic walled-off necrosis (WON). Early identification of high-risk patients irresponsive to DEN with LAMS will facilitate closer postoperative surveillance and optimal clinical decision-making. However, the prediction study on successful DEN with LAMS for pancreatic WON is lacking. Meantime, there has been a global trend to perform endoscopic interventions on an outpatient basis, which has advantages of convenience, efficiency, economy, and satisfaction. Now most patients undergoing DEN with LAMS for WON are hospitalized due to perceived increased risk of adverse events. Whether outpatient DEN with LAMs is safe and feasible, and which patients could be performed in an outpatient setting remain unknown. In this project, we retrospectively reviewed and analyzed a prospectively maintained database, including 101 consecutive patients who underwent DEN with LAMS since the year (2014) LAMS arrived on the market. First, we explore predictors of successful DEN with LAMS for pancreatic WON. Second, we evaluate the feasibility of outpatient DEN with LAMS, and compare the safety and efficacy on an outpatient versus inpatient basis among a case-matched cohort. This project provided a relatively large cohort of patients with WON who exclusively underwent DEN with the novel LAMS. To the best of our knowledge, this is the first prediction study for successful DEN with LAMS, and the first comparative study of DEN with LAMS on outpatient versus inpatient basis. The project will facilitate clinical decision-making and prognostication prior to DEN, and provide a good reference to proper selection criteria for outpatient DEN.

Hepatitis C Virus (HCV) infection in a donor, recipient, or both affects deceased-donor kidney transplantation outcomes. Direct-acting antivirals (DAA) may influence those outcomes. The Organ Procurement and Transplantation Network (OPTN) data of adult first-time solitary deceased-donor kidney transplant recipients 1994-2019 were allocated into four groups by donor or recipient HCV infection. We performed patient survival (PS) and death-censored graft survival (DCGS) pairwise comparisons after propensity score matching to assess donors and/or recipients HCV infection effect. We stratified our study by DAA era to evaluate for potential effect modification. In the pre-DAA era, donor HCV infection decreased PS and DCGS in all recipients. However, recipient HCV infection impaired PS and DCGS only with uninfected donors. HCV dual-infection (donor plus recipient) group had worse PS and DCGS than the dual-uninfected. Donor HCV infection derived worse post-transplant outcomes than recipient HCV infection. The risk associated with HCV infection in donors and/or recipients was no longer statistically significant in the post-DAA era, except for impaired PS in dual-infected versus dual-uninfected. Kidney donor HCV infection negatively affected transplant outcomes in all recipients, while recipient HCV infection impaired outcomes only with uninfected donors. Early post-DAA era analysis showed no effect of donor or recipient HCV infection.

Esophagogastric cancer (EGC) is the 2nd leading cause of cancer-related death worldwide and have unmet need to explore new treatment options. In first phase II study Heat Shock Protein (HSP 90) inhibitor Ganetespib show common grade 3/4 AEs leucopenia (12%), fatigue (12%), diarrhea (8%), and elevated ALKP (8%). The ORR of 4% reflects the single patient of 26 who had a complete response. Median PFS and OS was 61 days, 94 days respectively.Study was terminated early due to insufficient evidence of single-agent activity. Manageable toxicity,durable CR and 2 minor responses suggest that there may be a subset of EG patients who could benefit from this drug. In the 2nd study we explored the clinicopathological features and outcome of patients with MET amplified EGC. METamp was seen in 28 (12%) of patients versus 205 (88%) patients without amplification. Of MET-amplified patients, 7 (25%) had a TP53 mutation, 5 (18%) had HER2 co-amplification, and 1 (3.6%) had a PIK3CA mutation. Progression-free survival to initial treatment for MET-amplified patients were (5.6 vs 9.1 months, p=0.013) and for those with metastatic disease at presentation (4.4 vs 7.9 months, p=0.035). Overall, patients with MET amplification had shorter overall survival (15.3 vs 24.6 months, p=0.014).

As the population ages, understanding the relationship between frailty, a syndrome involving multisystem impairment in functional recovery, and outcomes is increasingly important to accurately predict healthcare utilization and adverse outcomes. The addition of frailty to risk models is also important to ensure adequate risk-adjustment. National guidelines strongly recommend an objective evaluation of frailty to optimize patient selection, but the range of available measures raises issues with consistency. In clinical practice, frailty is not often measured due to the lack of consensus surrounding frailty assessment tools, and there are divergent prevalence estimates and effect sizes reported across different studies. Furthermore, the prospective collection of information on frailty is time-consuming and may not always be feasible. In addition, registries may misrepresent frailty status and prevalence by focusing on a limited definition of frailty, not including all hospitals, and defining frailty according to a single point in time. Administrative claims represent an alternative source of data by which frailty might be more easily assessed. In the first paper, we, therefore, focused on ICD-9 claims-based frailty index and aimed to show the role of frailty on long-term mortality in patients undergoing transcatheter valve therapies. However, we realized that based on ICD-9 based claims may not comprehensively quantify frailty across all patients due to limited available number of codes. Since the transition to ICD-10 on October 1, 2015, which contains nearly 5-fold (from 14,000 to 70,000) the number of available claims, the increased granularity of claims data now permits a more comprehensive assessment of conditions associated with frailty and allows a more detailed longitudinal record of how frailty influences risk. Thus, in the second paper, we used an ICD-10 based frailty index to evaluate frailty and measure the impact of frailty on more outcomes (not only mortality) such as long hospital stay and readmission in patients undergoing transcatheter valve therapies.

Paper 1 Genetic factors have been postulated to be involved in the etiology of diabetic peripheral neuropathy (DPN), but their identity remains mostly unknown. The aim of this study was to conduct a systematic search for genetic variants influencing DPN risk using two well-characterized cohorts. A genome-wide association study (GWAS) testing 6.8 M SNPs was conducted among participants of the ACCORD clinical trial. Included were 4,384 White cases with type 2 diabetes (T2D) and prevalent or incident DPN (defined as a Michigan Neuropathy Screening Instrument [MNSI] clinical exam score >2.0) and 784 White controls with T2D and no evidence of DPN at baseline or during follow-up. Replication of significant loci was sought among White subjects with T2D (791 DPN-positive cases and 158 DPN-negative controls) from the BARI 2D trial. Association between significant variants and gene expression in peripheral nerves was evaluated in the GTEx database. A cluster of 28 SNPs on chromosome 2q24 reached GWAS significance (P<5x10-8) in ACCORD. The minor allele of the lead SNP (rs13417783, minor allele frequency=0.14) decreased DPN odds by 36% (OR 0.64, 95% CI 0.55-0.74, P=1.9x10-9). This effect was not influenced by ACCORD treatment assignments (P for interaction=0.6) nor was mediated by an association with known DPN risk factors. This locus was successfully validated in BARI 2D (OR 0.57, 95% CI 0.42-0.80, P=9x10-4, summary P value=7.9x10-12). In GTEx, the minor, protective allele at this locus was associated with higher tibial nerve expression of an adjacent gene (SCN2A) coding for human voltage-gated sodium channel NaV1.2 (P=9x10-4). To conclude, we have identified and successfully validated a previously unknown locus with a powerful protective effect on the development of DPN in T2D. These results may provide novel insights into DPN pathogenesis and point to a potential target for novel interventions. Paper 2. Cardiovascular autonomic neuropathy (CAN) is an independent predictor of cardiovascular disease (CVD) morbidity and all-cause and CVD mortality in diabetes. In light of its significant heritability, we aimed to identify genetic predictors of CAN through an unbiased genome-wide association study (GWAS) among Whites in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) cohort, comprising individuals with Type 2 Diabetes (T2D) and high CVD risk. CAN was defined based on indices of heart rate variability derived from 10-s resting electrocardiograms, namely, as the lowest quartile of standard deviation of normally conducted R-R intervals (SDNN <7.813 ms) and highest quartile of QT index (QTI >104.32%). Applying logistic regression models, we conducted a GWAS testing 6.8 million common single nucleotide polymorphisms (SNPs), with cases (N=807) having CAN at baseline and/or during follow-up and controls (N=3,144) being without CAN at end of follow-up. The top signal, rs779142, on chromosome 1p36, was associated with 35% increased odds of CAN (OR=1.35 [1.20 – 1.51], P=1.9 ×10-7). Among incident cases (N=499), this SNP reached GWAS significance (OR=1.48 [1.29 – 1.69], P=1.7 ×10-8). In a time-to-event analysis, rs779142 was associated with increased risk of progression to CAN (HR=1.42 [1.25 – 1.61], P=1.8 ×10-8) – an effect that was independent of the glycemic control arm assignments within the trial (P for interaction = 0.35). Genes lying within 1Mb of the lead SNP included CAMTA1 – a calmodulin binding transcription factor affecting cardiac embryonal development, KCNAB2 – a potassium voltage-gated channel associated with fatal cardiac arrhythmia, and RNF207 – a RING finger protein regulating heart action potential duration. To conclude, we have identified a locus on 1p36 having a genome-wide significant effect on the development of incident CAN in T2D. Pending further studies, these findings may provide better mechanistic insights into CAN pathophysiology as well as potential new therapeutic targets against this complication.

Paper 1: Predictors of antepartum admission in diabetes Objective: Diabetes is associated with increased healthcare utilization and cost in pregnant women. We sought to investigate the predictors of antepartum hospitalization in pregnant women with diabetes. Methods: We performed a retrospective review of women with pre-gestational diabetes who received care in a multidisciplinary program for diabetes in pregnancy at Brigham and women’s hospital from 2006-2011. Our primary outcome was antepartum hospitalization, with a secondary measure of length of stay. We modeled the multivariate odds of hospitalization prior to delivery using logistic regression. Results: We identified 244 women of whom 103 (42.2%) were hospitalized antepartum. The median (IQR) length of stay was 2 days (+2 days). There was no difference in age, race or pre-pregnancy body mass index (BMI) between women who were admitted and those who were not. Women who were admitted had a higher first Hemoglobin A1c in pregnancy (median of 7.85 + 2.75% vs 6.55 + 1.48%). In a multivariate model, markers of preconception health including first hemoglobin A1c in pregnancy (adjusted odds ratio[aOR] 1.5, 95% CI 1.24-1.81, p<0.001), smoking (aOR 4.02, CI 1.25- 12.90, p=0.019) and chronic hypertension (aOR 2.05, 95% CI 1.03-4.05, p= 0.03) were associated with increased odds of antepartum admission. First hemoglobin A1c in pregnancy was also associated with increased length of stay prior to delivery (beta=0.32, 95% CI 1.15- 5.87, p=0.004) Conclusion: In our study, glycemic control, smoking and chronic hypertension prior to pregnancy were associated with antepartum hospitalization, highlighting the importance of preconception health for reproductive-aged women with diabetes. Paper 2:Knowledge and attitudes toward prenatal yoga among women with high risk pregnancies Objective: To evaluate the attitudes and knowledge of prenatal yoga and investigate barriers and facilitators to yoga participation in high risk pregnant women receiving prenatal care in an academic tertiary care center. Methods: We surveyed a convenience sample of women receiving prenatal care through the Maternal-Fetal Medicine practice at Brigham and Women’s Hospital. We classified participants as yoga experienced or yoga naïve depending on self-report. We compared differences between the two groups using the appropriate nonparametric tests and compared bivariate odds ratios for survey results using logistic regression. Results: Of the 100 respondents, 53% had practiced yoga previously. Women with yoga experience were older (age 34.9 +5.6 vs 31 +6 years, p=0.004), more likely to be college graduates (94% vs 68%, p=0.002), and more likely to be white (77% vs 47%, p=0.002) than women without previous yoga experience. Previous yoga experience was associated with agreement that yoga was safe during their current pregnancy (OR 5.9, 95% CI 1.9-17.7). Of the women surveyed, 56% agreed that they would like to attend a prenatal yoga class. In a multivariate model including age, race and education, previous yoga experience was the only significant predictor associated with willingness to participate in prenatal yoga classes during current pregnancy. (OR 3.1, 95% CI 1.1-8.6) Conclusion: Prior yoga experience was the strongest predictor of willingness to attend a prenatal yoga class in our population. Our results suggest that women with high risk pregnancies who may benefit from prenatal yoga interventions but lack prior yoga experience may need additional education to facilitate participation.

Abstract paper 1- Background: Preoperative dental screening (PDS) prior to valvular surgery has been adopted widely by cardiac surgery to prevent poor outcomes associated with infections of oral origin. However, there is limited evidence to guide PDS protocols. This study was designed to understand the differences and association of focused and comprehensive approaches to PDS with post valvular surgery 90-day mortality. Methods: A retrospective cohort analysis was performed on all adult patients who underwent elective valvular surgery and PDS at two academic medical centers (MGH and BWH) from January 2009 to December 2016. Patients with a history of intravenous drug abuse were excluded. At MGH (n=692), PDS involved a dental examination including full series of diagnostic radiographs and definitive therapy for teeth with active and/or chronic disease (comprehensive protocol). At BWH (n=1143), PDS involved a dental examination that included selective diagnostic radiographs limited to symptomatic teeth and teeth with evidence of active infection (e.g. asymptomatic tooth with associated swelling and suppuration) and intervention only when active signs of disease were observed (focused protocol). Univariate and multivariable analysis were performed to compare 90-day mortality after surgery with the two different approaches. Results: A total of 1,835 patients met the inclusion criteria and their charts were reviewed. At BWH 1097/1143 (96%) patients received dental clearance at the initial visit with 3.3% (n=38) receiving radiographs. Dental findings included carious teeth (n=197, 17.2%), root tips (n=135, 11.8%), periodontically hopeless teeth (n=27, 2.4%), and periapical infections (n=20, 1.7%); 16 patients (1.4%) were symptomatic. Extractions were performed in 38 patients (3.3%), and 4 (10.5%) had post-extraction complications including pain, bleeding and infection. At MGH 245/692 (35.5%) patients received dental clearance at the initial visit with 94.2% (n=652) receiving radiographs. Dental findings included carious teeth (n=250, 36.1%), root tips (n=118, 17.1%), periodontically hopeless teeth (n=48, 6.9%), and periapical infections (n=149, 21.5%); 40 patients (5.8%) were symptomatic. Extractions were performed in 151 patients (21.8%), and 15 (9.93%) had post-extraction complications including pain, bleeding and infection. There was no significant difference in 90-day mortality between BWH and MGH (n=114, 10% vs n=58, 8.4%, p=0.257). This remained unchanged in a multivariable model after adjusting for demographics and baseline comorbidities (OR focused vs comprehensive:1.32 (95%CI:0.91-1.93), (p=0.14). Conclusions: There was no difference in 90-day mortality after cardiac valve surgery between patients who underwent a focused versus comprehensive PDS. Further prospective randomized comparative studies are needed to validate and expand on these findings. Abstract paper 2- Background- Odontogenic infections/ poor oral hygiene is one of the risk factors for developing post-surgical valvular complications. The magnitude and the impact to which these infections affect the post-surgical outcome especially early mortality remains unclear. This study was designed to evaluate the association of individual dental characteristic with the 90-day mortality. Methodology- A retrospective cohort analysis was performed on all adult patients who underwent valvular surgery and preoperative dental screening at Massachusetts General Hospital from January 2009 to December 2016. All valve surgery and dental data were procured using STS data set and electronic health record system of MGH respectively. Patients with a self-reported history of intravenous drug use and systemic infections were excluded from the study. We performed univariate and multivariable analysis for all the dental characteristics using logistic regression model. Odds Ratios (OR) and 95% CI’s were reported, and the significance level was 5%. Result- A total of 692 patients met the inclusion criteria of which, 58 (8.4%) patients died within 90-days and 634 (91.6%) patients survived after the valve surgery (within 90 days). Among the 90-day morality group and survived patients, the mean (±SD) age was 73.9±13.3 years vs 68.1±14.2 years with more females (n=32, 55.2% vs n=243, 38.3%). Although dental characteristics including dental complaint (n=1, 1.7% vs n=39, 6.2%), total number of teeth present (19.7 ± 8.1 vs 21.1 ±7.8), complete denture (n=9, 15.5% vs n=64, 10.1%), partial denture (n=1, 1.7% vs n=43, 6.8%), carious teeth (n=18, 31% vs n=232, 36.5%), root tips (n=5, 8.6% vs n=113, 17.8%), total number of asymptomatic root tips (0.3 ± 0.9 vs 0.6 ± 1.6), endodontically treated teeth (n=26, 44.8% vs n=289, 45.5%), periodontally hopeless teeth (n=3, 5.2% vs n=45, 7.1%) and tenderness on percussion (n=2, 3.4% vs n=17, 2.7%) varied amongst the group, the univariate and multivariable analysis showed no significant association of dental/ oral health characteristics with 90-day mortality. Conclusion- There was no significant association noted between dental characteristics and 90-day mortality. Further large prospective comparative studies are needed to validate and expand these findings.