Bio


I am an MD from Chung Shan Medical University, Taiwan. Before coming to Stanford, I obtained my MS degree in epidemiology at Harvard T.H. Chan School of Public Health, Boston, MA, where I completed graduate training in clinical, pharmacologic, and genetic epidemiology, and pursued advanced skills in biostatistics and causal inference.

My past research focused on real-world epidemiology studies using patient registries and national health insurance databases to elucidate the predictors or risk factors of immunologic diseases. For my graduate study, I conducted pharmacoepidemiology studies using electronic health record (EHR) data to elucidate the predictors of anti-drug antibodies development and its correlation to autoimmunity, to identify the generation of immunogenicity that may impact the effectiveness of monoclonal antibody therapies in individuals with autoimmune diseases. I gained experience in genetic data manipulation to investigate polymorphisms in response to monoclonal antibody therapies in asthma patients.

At Stanford, I am involved in research on the identification of molecular determinants of cardiometabolic diseases.

Professional Education


  • Master of Science, Harvard University, Boston, MA, Epidemiology (2023)
  • Doctor of Medicine, Chung Shan Medical University, Taichung, Taiwan, Medicine (2021)

Stanford Advisors


All Publications


  • Plasma proteomics and carotid intima-media thickness in the UK biobank cohort. Frontiers in cardiovascular medicine Chen, M. L., Kho, P. F., Guarischi-Sousa, R., Zhou, J., Panyard, D. J., Azizi, Z., Gupte, T., Watson, K., Abbasi, F., Assimes, T. L. 2024; 11: 1478600

    Abstract

    Ultrasound derived carotid intima-media thickness (cIMT) is valuable for cardiovascular risk stratification. We assessed the relative importance of traditional atherosclerosis risk factors and plasma proteins in predicting cIMT measured nearly a decade later.We examined 6,136 UK Biobank participants with 1,461 proteins profiled using the proximity extension assay applied to their baseline blood draw who subsequently underwent a cIMT measurement. We implemented linear regression, stepwise Akaike Information Criterion-based, and the least absolute shrinkage and selection operator (LASSO) models to identify potential proteomic as well as non-proteomic predictors. We evaluated our model performance using the proportion variance explained (R 2).The mean time from baseline assessment to cIMT measurement was 9.2 years. Age, blood pressure, and anthropometric related variables were the strongest predictors of cIMT with fat-free mass index of the truncal region being the strongest predictor among adiposity measurements. A LASSO model incorporating variables including age, assessment center, genetic risk factors, smoking, blood pressure, trunk fat-free mass index, apolipoprotein B, and Townsend deprivation index combined with 97 proteins achieved the highest R 2 (0.308, 95% C.I. 0.274, 0.341). In contrast, models built with proteins alone or non-proteomic variables alone explained a notably lower R 2 (0.261, 0.228-0.294 and 0.260, 0.226-0.293, respectively). Chromogranin b (CHGB), Cystatin-M/E (CST6), leptin (LEP), and prolargin (PRELP) were the proteins consistently selected across all models.Plasma proteins add to the clinical and genetic risk factors in predicting a cIMT measurement. Our findings implicate blood pressure and extracellular matrix-related proteins in cIMT pathophysiology.

    View details for DOI 10.3389/fcvm.2024.1478600

    View details for PubMedID 39416432

    View details for PubMedCentralID PMC11480011

  • A plasma proteomic signature for atherosclerotic cardiovascular disease risk prediction in the UK Biobank cohort. medRxiv : the preprint server for health sciences Gupte, T. P., Azizi, Z., Kho, P. F., Zhou, J., Chen, M., Panyard, D. J., Guarischi-Sousa, R., Hilliard, A. T., Sharma, D., Watson, K., Abbasi, F., Tsao, P. S., Clarke, S. L., Assimes, T. L. 2024

    Abstract

    Background: While risk stratification for atherosclerotic cardiovascular disease (ASCVD) is essential for primary prevention, current clinical risk algorithms demonstrate variability and leave room for further improvement. The plasma proteome holds promise as a future diagnostic and prognostic tool that can accurately reflect complex human traits and disease processes. We assessed the ability of plasma proteins to predict ASCVD.Method: Clinical, genetic, and high-throughput plasma proteomic data were analyzed for association with ASCVD in a cohort of 41,650 UK Biobank participants. Selected features for analysis included clinical variables such as a UK-based cardiovascular clinical risk score (QRISK3) and lipid levels, 36 polygenic risk scores (PRSs), and Olink protein expression data of 2,920 proteins. We used least absolute shrinkage and selection operator (LASSO) regression to select features and compared area under the curve (AUC) statistics between data types. Randomized LASSO regression with a stability selection algorithm identified a smaller set of more robustly associated proteins. The benefit of plasma proteins over standard clinical variables, the QRISK3 score, and PRSs was evaluated through the derivation of Delta AUC values. We also assessed the incremental gain in model performance using proteomic datasets with varying numbers of proteins. To identify potential causal proteins for ASCVD, we conducted a two-sample Mendelian randomization (MR) analysis.Result: The mean age of our cohort was 56.0 years, 60.3% were female, and 9.8% developed incident ASCVD over a median follow-up of 6.9 years. A protein-only LASSO model selected 294 proteins and returned an AUC of 0.723 (95% CI 0.708-0.737). A clinical variable and PRS-only LASSO model selected 4 clinical variables and 20 PRSs and achieved an AUC of 0.726 (95% CI 0.712-0.741). The addition of the full proteomic dataset to clinical variables and PRSs resulted in a Delta AUC of 0.010 (95% CI 0.003-0.018). Fifteen proteins selected by a stability selection algorithm offered improvement in ASCVD prediction over the QRISK3 risk score [Delta AUC: 0.013 (95% CI 0.005-0.021)]. Filtered and clustered versions of the full proteomic dataset (consisting of 600-1,500 proteins) performed comparably to the full dataset for ASCVD prediction. Using MR, we identified 11 proteins as potentially causal for ASCVD.Conclusion: A plasma proteomic signature performs well for incident ASCVD prediction but only modestly improves prediction over clinical and genetic factors. Further studies are warranted to better elucidate the clinical utility of this signature in predicting the risk of ASCVD over the standard practice of using the QRISK3 score.

    View details for DOI 10.1101/2024.09.13.24313652

    View details for PubMedID 39314942

  • Plasma proteomic signatures for type 2 diabetes mellitus and related traits in the UK Biobank cohort. medRxiv : the preprint server for health sciences Gupte, T. P., Azizi, Z., Kho, P. F., Zhou, J., Nzenkue, K., Chen, M., Panyard, D. J., Guarischi-Sousa, R., Hilliard, A. T., Sharma, D., Watson, K., Abbasi, F., Tsao, P. S., Clarke, S. L., Assimes, T. L. 2024

    Abstract

    Aims/hypothesis: The plasma proteome holds promise as a diagnostic and prognostic tool that can accurately reflect complex human traits and disease processes. We assessed the ability of plasma proteins to predict type 2 diabetes mellitus (T2DM) and related traits.Methods: Clinical, genetic, and high-throughput proteomic data from three subcohorts of UK Biobank participants were analyzed for association with dual-energy x-ray absorptiometry (DXA) derived truncal fat (in the adiposity subcohort), estimated maximum oxygen consumption (VO 2 max) (in the fitness subcohort), and incident T2DM (in the T2DM subcohort). We used least absolute shrinkage and selection operator (LASSO) regression to assess the relative ability of non-proteomic and proteomic variables to associate with each trait by comparing variance explained (R 2 ) and area under the curve (AUC) statistics between data types. Stability selection with randomized LASSO regression identified the most robustly associated proteins for each trait. The benefit of proteomic signatures (PSs) over QDiabetes, a T2DM clinical risk score, was evaluated through the derivation of delta (Delta) AUC values. We also assessed the incremental gain in model performance metrics using proteomic datasets with varying numbers of proteins. A series of two-sample Mendelian randomization (MR) analyses were conducted to identify potentially causal proteins for adiposity, fitness, and T2DM.Results: Across all three subcohorts, the mean age was 56.7 years and 54.9% were female. In the T2DM subcohort, 5.8% developed incident T2DM over a median follow-up of 7.6 years. LASSO-derived PSs increased the R 2 of truncal fat and VO 2 max over clinical and genetic factors by 0.074 and 0.057, respectively. We observed a similar improvement in T2DM prediction over the QDiabetes score [Delta AUC: 0.016 (95% CI 0.008, 0.024)] when using a robust PS derived strictly from the T2DM outcome versus a model further augmented with non-overlapping proteins associated with adiposity and fitness. A small number of proteins (29 for truncal adiposity, 18 for VO2max, and 26 for T2DM) identified by stability selection algorithms offered most of the improvement in prediction of each outcome. Filtered and clustered versions of the full proteomic dataset supplied by the UK Biobank (ranging between 600-1,500 proteins) performed comparably to the full dataset for T2DM prediction. Using MR, we identified 4 proteins as potentially causal for adiposity, 1 as potentially causal for fitness, and 4 as potentially causal for T2DM.Conclusions/Interpretation: Plasma PSs modestly improve the prediction of incident T2DM over that possible with clinical and genetic factors. Further studies are warranted to better elucidate the clinical utility of these signatures in predicting the risk of T2DM over the standard practice of using the QDiabetes score. Candidate causally associated proteins identified through MR deserve further study as potential novel therapeutic targets for T2DM.

    View details for DOI 10.1101/2024.09.13.24313501

    View details for PubMedID 39314935

  • Associations between accurate measures of adiposity and fitness, blood proteins, and insulin sensitivity among South Asians and Europeans. medRxiv : the preprint server for health sciences Kho, P. F., Stell, L., Jimenez, S., Zanetti, D., Panyard, D. J., Watson, K. L., Sarraju, A., Chen, M. L., Lind, L., Petrie, J. R., Chan, K. N., Fonda, H., Kent, K., Myers, J. N., Palaniappan, L., Abbasi, F., Assimes, T. L. 2024

    Abstract

    South Asians (SAs) may possess a unique predisposition to insulin resistance (IR). We explored this possibility by investigating the relationship between 'gold standard' measures of adiposity, fitness, selected proteomic biomarkers, and insulin sensitivity among a cohort of SAs and Europeans (EURs).A total of 46 SAs and 41 EURs completed 'conventional' (lifestyle questionnaires, standard physical exam) as well as 'gold standard' (dual energy X-ray absorptiometry scan, cardiopulmonary exercise test, and insulin suppression test) assessments of adiposity, fitness, and insulin sensitivity. In a subset of 28 SAs and 36 EURs, we also measured the blood-levels of eleven IR-related proteins. We conducted Spearman correlation to identify correlates of steady-state plasma glucose (SSPG) derived from the insulin suppression test, followed by multivariable linear regression analyses of SSPG, adjusting for age, sex and ancestral group.Sixteen of 30 measures significantly associated with SSPG, including one conventional and eight gold standard measures of adiposity, one conventional and one gold standard measure of fitness, and five proteins. Multivariable regressions revealed that gold standard measures and plasma proteins attenuated ancestral group differences in IR, suggesting their potential utility in assessing IR, especially among SAs.Ancestral group differences in IR may be explained by accurate measures of adiposity and fitness, with specific proteins possibly serving as useful surrogates for these measures, particularly for SAs.

    View details for DOI 10.1101/2024.09.06.24313199

    View details for PubMedID 39281745

    View details for PubMedCentralID PMC11398600

  • PLASMA PROTEOMICS AND VISCERAL ADIPOSE TISSUE VOLUME: A MACHINE LEARNING ANALYSIS OF INTERACTION BETWEEN BIOMARKERS, SOCIO-BEHAVIORAL, AND FITNESS FACTORS IN UK BIOBANK Azizi, Z., Gupte, T., Kho, P., Nzenkue, K., Zhou, J., Guarischi-Sousa, R., Panyard, D., Chen, M., Abbasi, F., Clarke, S., Tsao, P., Assimes, T. L. ELSEVIER SCIENCE INC. 2024: 1699
  • Incidence of Anti-Drug Antibodies to Monoclonal Antibodies in Asthma: A Systematic Review and Meta-Analysis JOURNAL OF ALLERGY AND CLINICAL IMMUNOLOGY-IN PRACTICE Chen, M., Nopsopon, T., Akenroye, A. 2023; 11 (5): 1475-+

    Abstract

    Antidrug antibodies (ADAs) may worsen the efficacy and safety of biologics. However, little is known about the incidence of ADAs associated with the 6 biologics approved for the treatment of asthma in the United States.To elucidate the incidence of ADAs and their impact on reported clinical outcomes.Systematic review and meta-analyses of randomized controlled trials, open-label extension studies, and nonrandomized studies of biologics in patients with asthma indexed in PubMed, Embase, and CENTRAL between January 1, 2000, and July 9, 2022, were carried out. The primary outcomes were treatment-emergent ADAs (incidence) and ADA prevalence.A total of 46 studies met the eligibility criteria. ADA incidence over follow-up was 2.91% (95% CI, 1.60-4.55) and was highest in the benralizumab studies (8.35%), with a risk ratio of 4.9 (2.69-8.92) when compared with placebo, and lowest in the omalizumab studies (0.00%). Incidence was 7.61% in the dupilumab studies, 4.39% in reslizumab, 3.63% in mepolizumab, and 1.12% in the tezepelumab studies. Incidence of neutralizing antibodies was 0.00% to 10.74% and was highest for benralizumab (7.12%). Incidence of neutralizing antibodies was higher in the benralizumab every 8 weeks (8.17%) versus every 4 weeks arms (5.81%). Results were consistent in subgroup analyses by study type and length of follow-up.Approximately 2.9% of individuals in the included studies developed ADAs over study follow-up period. The incidence was highest in the benralizumab group and lowest in the omalizumab group. The subcutaneous route and longer dosing intervals were associated with higher ADA development.

    View details for DOI 10.1016/j.jaip.2022.12.046

    View details for Web of Science ID 001025725300001

    View details for PubMedID 36716995

  • Comparative efficacy of tezepelumab to mepolizumab, benralizumab, and dupilumab in eosinophilic asthma: A Bayesian network meta-analysis JOURNAL OF ALLERGY AND CLINICAL IMMUNOLOGY Nopsopon, T., Lassiter, G., Chen, M., Alexander, C., Keet, C., Hong, H., Akenroye, A. 2023; 151 (3): 747-755

    Abstract

    It is unclear how the efficacy of tezepelumab, approved for the treatment of type 2 high and low asthma, compares to the efficacy of other biologics for type 2-high asthma.We sought to conduct an indirect comparison of tezepelumab to dupilumab, benralizumab, and mepolizumab in the treatment of eosinophilic asthma.The investigators conducted a systematic review and Bayesian network meta-analyses. They identified randomized controlled trials indexed in PubMed, Embase, or Cochrane Central Register of Controlled Trials (CENTRAL) between January 1, 2000, and August 12, 2022. Outcomes included exacerbation rates, prebronchodilator FEV1, and the Asthma Control Questionnaire.Ten randomized controlled trials (n = 9201) met eligibility. Tezepelumab (relative risk: 0.63; 95% credible interval [CI]: 0.46-0.86) was associated with significantly lower exacerbation rates than benralizumab and larger improvements in FEV1 compared to mepolizumab (mean difference [MD]: 66; 95% CI: -33 to 170) and benralizumab (MD: 62; 95% CI: -22 to 150), though the 95% CI crossed the null value of 0. Mepolizumab improved the Asthma Control Questionnaire score the most, but this improvement was not significantly different from that of tezepelumab (tezepelumab vs mepolizumab; MD: 0.14; 95% CI: -0.10 to 0.38). For efficacy by clinically important thresholds, tezepelumab, mepolizumab, and dupilumab achieved a >99% probability of reducing exacerbation rates by ≥50% compared to placebo, but benralizumab had only a 66% probability of doing so. Tezepelumab and dupilumab had a probability of 1.00 of improving prebronchodilator FEV1 by ≥100 mL above placebo. Compared to mepolizumab, dupilumab had >90% chance for improving FEV1 by ≥50 mL, but none of the differences between biologics exceeded 100 mL.In individuals with eosinophilic asthma, tezepelumab and dupilumab were associated with greater improvements (although below clinical thresholds) in exacerbation rates and lung function than benralizumab or mepolizumab.

    View details for DOI 10.1016/j.jaci.2022.11.021

    View details for Web of Science ID 001025005600001

    View details for PubMedID 36538979

    View details for PubMedCentralID PMC9992307