Chair of the Department of Biostatistics
Professor of Biostatistics
Prior to NYU, Dr. Betensky was Professor of Biostatistics at the Harvard T.H. Chan School of Public Health. She was director of the Harvard Catalyst (Clinical and Translational Science Award) Biostatistics Program; director of the Data and Statistics Core for the Massachusetts Alzheimer’s Disease Research Center; and director of the Biostatistics Neurology Core at Massachusetts General Hospital. Previously, she was the Biostatistics Program Leader for the Dana-Farber/Harvard Cancer Center.
Dr. Betensky’s research focuses on methods for the analysis of censored and truncated outcomes and covariates, which frequently arise from the subsampling of cohort studies. She has a long-time interest in clinical trials, and has written on the evaluation of biomarkers and the use and interpretation of p-values. She has collaborated extensively in studies in neurologic diseases, and serves as statistical editor for Annals of Neurology.
Dr. Betensky was awarded, and directed for 15 years, an NIH T32 training program in neurostatistics and neuroepidemiology for pre- and post-doctoral students in biostatistics and epidemiology and for clinician-scientists. She previously directed Harvard’s Biostatistics programs to promote and support diversity at all levels in the field of quantitative public health. She was also a member of the BMRD Study Section for review of NIH statistical methodology grants; on committees for the Institute of Medicine; and a co-chair of the technical advisory committee for the scientific registry of transplant recipients.
Dr. Betensky an elected Fellow of the American Statistical Association and of the International Statistical Institute, and is a past recipient of the Spiegelman Award from the American Public Health Association. She currently serves as a member of the Board of Scientific Counselors for Clinical Science and Epidemiology at the National Cancer Institute.
AB, Mathematics, Harvard University, Cambridge, MAPhD, Statistics, Stanford University, Stanford, CA
Association of Amyloid and Tau with Cognition in Preclinical Alzheimer Disease: A Longitudinal StudyHanseeuw, B. J., Betensky, R., Jacobs, H. I., Schultz, A. P., Sepulcre, J., Becker, J. A., Cosio, D. M., Farrell, M., Quiroz, Y. T., Mormino, E. C., Buckley, R. F., Papp, K. V., Amariglio, R. A., Dewachter, I., Ivanoiu, A., Huijbers, W., Hedden, T., Marshall, G. A., Chhatwal, J. P., Rentz, D. M., Sperling, R. A., & Johnson, K.
Journal titleJAMA NeurologyImportance: Positron emission tomography (PET) imaging now allows in vivo visualization of both neuropathologic hallmarks of Alzheimer disease (AD): amyloid-β (Aβ) plaques and tau neurofibrillary tangles. Observing their progressive accumulation in the brains of clinically normal older adults is critically important to understand the pathophysiologic cascade leading to AD and to inform the choice of outcome measures in prevention trials. Objective: To assess the associations among Aβ, tau, and cognition, measured during different observation periods for 7 years. Design, Setting, and Participants: Prospective cohort study conducted between 2010 and 2017 at the Harvard Aging Brain Study, Boston, Massachusetts. The study enrolled 279 clinically normal participants. An additional 90 individuals were approached but declined the study or did not meet the inclusion criteria. In this report, we analyzed data from 60 participants who had multiple Aβ and tau PET observations available on October 31, 2017. Main Outcomes and Measures: A median of 3 Pittsburgh compound B-PET (Aβ, 2010-2017) and 2 flortaucipir-PET (tau, 2013-2017) images were collected. We used initial PET and slope data, assessing the rates of change in Aβ and tau, to measure cognitive changes. Cognition was evaluated annually using the Preclinical Alzheimer Cognitive Composite (2010-2017). Annual consensus meetings evaluated progression to mild cognitive impairment. Results: Of the 60 participants, 35 were women (58%) and 25 were men (42%); median age at inclusion was 73 years (range, 65-85 years). Seventeen participants (28%) exhibited an initial high Aβ burden. An antecedent rise in Aβ was associated with subsequent changes in tau (1.07 flortaucipir standardized uptake value ratios [SUVr]/PiB-SUVr; 95% CI, 0.13-3.46; P =.02). Tau changes were associated with cognitive changes (-3.28 z scores/SUVR; 95% CI, -6.67 to -0.91; P =.001), covarying baseline Aβ and tau. Tau changes were greater in the participants who progressed to mild cognitive impairment (n = 6) than in those who did not (n = 11; 0.05 SUVr per year; 95% CI, 0.03-0.07; P =.001). A serial mediation model demonstrated that the association between initial Aβ and final cognition, measured 7 years later, was mediated by successive changes in Aβ and tau. Conclusions and Relevance: We identified sequential changes in normal older adults, from Aβ to tau to cognition, after which the participants with high Aβ with greater tau increase met clinical criteria for mild cognitive impairment. These findings highlight the importance of repeated tau-PET observations to track disease progression and the importance of repeated amyloid-PET observations to detect the earliest AD pathologic changes.
Genetic overlap between vascular pathologies and Alzheimer's dementia and potential causal mechanismsLin, Y. F., Smith, A. V., Aspelund, T., Betensky, R., Smoller, J. W., Gudnason, V., Launer, L. J., & Blacker, D.
Journal titleAlzheimer's and Dementia
Page(s)65-75Introduction: We sought to examine the genetic overlap between vascular pathologies and Alzheimer's disease (AD) dementia, and the potential mediating role of vascular pathologies between AD-related genetic variants and late-life cognition. Methods: For 2907 stroke-free older individuals, we examined the association of polygenic risk scores for AD dementia (ADPRSs) with vascular pathologies and with cognition. Mediation analyses addressed whether association between ADPRSs and cognition was mediated by a vascular pathology. Results: ADPRSs were associated with lobar cerebral microbleeds, white matter lesion load, and coronary artery calcification, mostly explained by single nucleotide polymorphisms in the 19q13 region. The effect of ADPRSs on cognition was partially but significantly mediated by cerebral microbleeds, white matter lesions, and coronary artery calcification. Discussion: Our findings provide evidence for genetic overlap, mostly due to apolipoprotein E (APOE) gene, between vascular pathologies and AD dementia. The association between AD polygenic risk and late-life cognition is mediated in part via effects on vascular pathologies.
Inverse probability weighting methods for Cox regression with right-truncated dataVakulenko-Lagun, B., Mandel, M., & Betensky, R. A.
Journal titleBiometricsRight-truncated data arise when observations are ascertained retrospectively, and only subjects who experience the event of interest by the time of sampling are selected. Such a selection scheme, without adjustment, leads to biased estimation of covariate effects in the Cox proportional hazards model. The existing methods for fitting the Cox model to right-truncated data, which are based on the maximization of the likelihood or solving estimating equations with respect to both the baseline hazard function and the covariate effects, are numerically challenging. We consider two alternative simple methods based on inverse probability weighting (IPW) estimating equations, which allow consistent estimation of covariate effects under a positivity assumption and avoid estimation of baseline hazards. We discuss problems of identifiability and consistency that arise when positivity does not hold and show that although the partial tests for null effects based on these IPW methods can be used in some settings even in the absence of positivity, they are not valid in general. We propose adjusted estimating equations that incorporate the probability of observation when it is known from external sources, which results in consistent estimation. We compare the methods in simulations and apply them to the analyses of human immunodeficiency virus latency.
Nonidentifiability in the presence of factorization for truncated data.Vakulenko-Lagun, B., Qian, J., Chiou, S. H., & Betensky, R. A.
Page(s)724-731A time to event, |$X$| , is left-truncated by |$T$| if |$X$| can be observed only if |$T<X$|. This often results in oversampling of large values of |$X$| , and necessitates adjustment of estimation procedures to avoid bias. Simple risk-set adjustments can be made to standard risk-set-based estimators to accommodate left truncation when |$T$| and |$X$| are quasi-independent. We derive a weaker factorization condition for the conditional distribution of |$T$| given |$X$| in the observable region that permits risk-set adjustment for estimation of the distribution of |$X$| , but not of the distribution of |$T$|. Quasi-independence results when the analogous factorization condition for |$X$| given |$T$| holds also, in which case the distributions of |$X$| and |$T$| are easily estimated. While we can test for factorization, if the test does not reject, we cannot identify which factorization condition holds, or whether quasi-independence holds. Hence we require an unverifiable assumption in order to estimate the distribution of |$X$| or |$T$| based on truncated data. This contrasts with the common understanding that truncation is different from censoring in requiring no unverifiable assumptions for estimation. We illustrate these concepts through a simulation of left-truncated and right-censored data.
Serum levels of 25-hydroxyvitamin D at diagnosis are not associated with overall survival in esophageal adenocarcinomaLoehrer, E., Betensky, R., Giovannucci, E., Su, L., Shafer, A., Hollis, B. W., & Christiani, D. C.
Journal titleCancer Epidemiology Biomarkers and Prevention
Page(s)1379-1387Background: Higher levels of circulating 25-hydroxyvitamin D [25(OH)D] are associated with longer survival in several cancers, but the results have differed across cancer sites. The association between serum 25(OH)D levels and overall survival (OS) time in esophageal adenocarcinoma remains unclear. Methods: We utilized serum samples from 476 patients with primary esophageal adenocarcinoma, recruited from Massachusetts General Hospital (Boston, MA) between 1999 and 2015. We used log-rank tests to test the difference in survival curves across quartiles of 25(OH)D levels and extended Cox modeling to estimate adjusted HRs. We tested for interactions between clinical stage or BMI on the association between 25(OH)D and OS. We additionally performed sensitivity analyses to determine whether race or timing of blood draw (relative to treatment) affected these results. Results: We found no evidence that survival differed across quartiles of 25(OH)D (log rank P = 0.48). Adjusting for confounders, we found no evidence that the hazard of death among the highest quartile of 25(OH)D (quartile 1) differed from any other quartile [quartile 2 HR = 0.90, 95% confidence interval (CI), 0.67-1.23; quartile 3 HR = 1.03, 95% CI, 0.76- 1.38; quartile 4 (lowest) HR = 0.98, 95% CI, 0.72-1.33]. Sensitivity analyses yielded consistent results when accounting for race or time between diagnosis and blood draw. Moreover, we did not find evidence of interaction between 25(OH)D and clinical stage or BMI on OS. Conclusions: Serum level of 25(OH)D near time of diagnosis was not associated with OS in patients with esophageal adenocarcinoma. Impact: Screening 25(OH)D levels among patients with esophageal adenocarcinoma at diagnosis is not clinically relevant to their cancer prognosis based on present evidence.
The impact of amyloid-beta and tau on prospective cognitive decline in older individualsSperling, R. A., Mormino, E. C., Schultz, A. P., Betensky, R., Papp, K. V., Amariglio, R. E., Hanseeuw, B. J., Buckley, R., Chhatwal, J., Hedden, T., Marshall, G. A., Quiroz, Y. T., Donovan, N. J., Jackson, J., Gatchel, J. R., Rabin, J. S., Jacobs, H., Yang, H. S., Properzi, M., Kirn, D. R., Rentz, D. M., & Johnson, K. A.
Journal titleAnnals of Neurology
Page(s)181-193Objectives: Amyloid-beta (Aβ) and tau pathologies are commonly observed among clinically normal older individuals at postmortem and can now be detected with in vivo neuroimaging. The association and interaction of these proteinopathies with prospective cognitive decline in normal aging and preclinical Alzheimer's disease (AD) remains to be fully elucidated. Methods: One hundred thirty-seven older individuals (age = 76.3 ± 6.22 years) participating in the Harvard Aging Brain Study underwent Aβ (11C-Pittsburgh compound B) and tau (18F-flortaucipir) positron emission tomography (PET) with prospective neuropsychological assessments following PET imaging (mean number of cognitive visits = 2.8 ± 1.1). Tau and Aβ PET measures were assessed in regions of interest (ROIs) as well as vertex-wise map analyses. Cognitive change was evaluated with Memory and Executive Function composites. Results: Higher levels of Aβ and tau were both associated with greater memory decline, but not with change in executive function. Higher cortical Aβ was associated with higher tau levels in all ROIs, independent of age, and very elevated levels of tau were observed primarily in clinically normal with elevated Aβ. A significant interaction between tau and Aβ was observed in both ROI and map-level analyses, such that rapid prospective memory decline was observed in participants who had high levels of both pathologies. Interpretation: Our results are consistent with the supposition that both Aβ and tau are necessary for memory decline in the preclinical stages of AD. These findings may be relevant for disambiguating aging and early cognitive manifestations of AD, and to inform secondary prevention trials in preclinical AD. Ann Neurol 2019;00:1–3 ANN NEUROL 2019;85:181–193.
The missing indicator approach for censored covariates subject to limit of detection in logistic regression modelsChiou, S. H., Betensky, R. A., & Balasubramanian, R.
Journal titleAnnals of Epidemiology
Page(s)57-64Purpose: In several biomedical studies, one or more exposures of interest may be subject to nonrandom missingness because of the failure of the measurement assay at levels below its limit of detection. This issue is commonly encountered in studies of the metabolome using tandem mass spectrometry–based technologies. Owing to a large number of metabolites measured in these studies, preserving statistical power is of utmost interest. In this article, we evaluate the small sample properties of the missing indicator approach in logistic and conditional logistic regression models. Methods: For nested case-control or matched case control study designs, we evaluate the bias, power, and type I error associated with the missing indicator method using simulation. We compare the missing indicator approach to complete case analysis and several imputation approaches. Results: We show that under a variety of settings, the missing indicator approach outperforms complete case analysis and other imputation approaches with regard to bias, mean squared error, and power. Conclusions: For nested case-control and matched study designs of modest sample sizes, the missing indicator model minimizes loss of information and thus provides an attractive alternative to the oft-used complete case analysis and other imputation approaches.
The p-Value Requires Context, Not a ThresholdBetensky, R.
Journal titleAmerican Statistician
Page(s)115-117It is widely recognized by statisticians, though not as widely by other researchers, that the p-value cannot be interpreted in isolation, but rather must be considered in the context of certain features of the design and substantive application, such as sample size and meaningful effect size. I consider the setting of the normal mean and highlight the information contained in the p-value in conjunction with the sample size and meaningful effect size. The p-value and sample size jointly yield 95% confidence bounds for the effect of interest, which can be compared to the predetermined meaningful effect size to make inferences about the true effect. I provide simple examples to demonstrate that although the p-value is calculated under the null hypothesis, and thus seemingly may be divorced from the features of the study from which it arises, its interpretation as a measure of evidence requires its contextualization within the study. This implies that any proposal for improved use of the p-value as a measure of the strength of evidence cannot simply be a change to the threshold for significance.
An optimal Wilcoxon–Mann–Whitney test of mortality and a continuous outcomeMatsouaka, R. A., Singhal, A. B., & Betensky, R.
Journal titleStatistical Methods in Medical Research
Page(s)2384-2400We consider a two-group randomized clinical trial, where mortality affects the assessment of a follow-up continuous outcome. Using the worst-rank composite endpoint, we develop a weighted Wilcoxon–Mann–Whitney test statistic to analyze the data. We determine the optimal weights for the Wilcoxon–Mann–Whitney test statistic that maximize its power. We derive a formula for its power and demonstrate its accuracy in simulations. Finally, we apply the method to data from an acute ischemic stroke clinical trial of normobaric oxygen therapy.
Association of anxiety with subcortical amyloidosis in cognitively normal older adultsHanseeuw, B. J., Jonas, V., Jackson, J., Betensky, R., Rentz, D. M., Johnson, K. A., Sperling, R. A., & Donovan, N. J.
Journal titleMolecular PsychiatryLate-life anxiety has been associated with increased progression from normal cognition to amnestic MCI, suggesting that anxiety may be a neuropsychiatric symptom of Alzheimer’s disease (AD) pathological changes and a possible marker of anatomical progression in preclinical AD. This study examined whether cortical or subcortical amyloidosis, indicating earlier or later stages of preclinical AD, was associated with greater self-reported anxiety among 118 cognitively normal volunteers, aged 65–90 years, and whether this association was stronger in APOEε4 carriers. Participants underwent Pittsburgh Compound B Positron Emission Tomography (PiB-PET) to assess fibrillar amyloid-β burden in cortical and subcortical regions, and measurement of anxiety using the Hospital Anxiety and Depression Scale-anxiety subscale. Higher PiB-PET measures in the subcortex (striatum, amygdala, and thalamus), but not in the cortex, were associated with greater anxiety, adjusting for demographics, cognition, and depression. Findings were similar using a cortico-striatal staging system and continuous PET measurements. Anxiety was highest in APOEε4 carriers with subcortical amyloidosis. This work supports in vivo staging of amyloid-β deposition in both cortical and subcortical regions as a promising approach to the study of neuropsychiatric symptoms such as anxiety in cognitively normal older individuals. Elevated anxiety symptoms in combination with high-risk biological factors such as APOEε4 and subcortical amyloid-β may identify participants closest to MCI for secondary prevention trials.
Biomarker validation with an imperfect reference: Issues and boundsEmerson, S. C., Waikar, S. S., Fuentes, C., Bonventre, J. V., & Betensky, R.
Journal titleStatistical Methods in Medical Research
Page(s)2933-2945Motivated by the goal of evaluating a biomarker for acute kidney injury, we consider the problem of assessing operating characteristics for a new biomarker when a true gold standard for disease status is unavailable. In this case, the biomarker is typically compared to another imperfect reference test, and this comparison is used to estimate the performance of the new biomarker. However, errors made by the reference test can bias assessment of the new test. Analysis methods like latent class analysis have been proposed to address this issue, generally employing some strong and unverifiable assumptions regarding the relationship between the new biomarker and the reference test. We investigate the conditional independence assumption that is present in many such approaches and show that for a given set of observed data, conditional independence is only possible for a restricted range of disease prevalence values. We explore the information content of the comparison between the new biomarker and the reference test, and give bounds for the true sensitivity and specificity of the new test when operating characteristics for the reference test are known. We demonstrate that in some cases these bounds may be tight enough to provide useful information, but in other cases these bounds may be quite wide.
Correction to: Association of anxiety with subcortical amyloidosis in cognitively normal older adults (Molecular Psychiatry, (2018), 10.1038/s41380-018-0214-2)Hanseeuw, B. J., Jonas, V., Jackson, J., Betensky, R., Rentz, D. M., Johnson, K. A., Sperling, R. A., & Donovan, N. J.
Journal titleMolecular PsychiatryThis Article was originally published under Nature Research’s License to Publish, but has now been made available under a CC BY 4.0 license. The PDF and HTML versions of the Article have been modified accordingly.
Design and analysis considerations for combining data from multiple biomarker studiesSloan, A., Song, Y., Gail, M. H., Betensky, R., Rosner, B., Ziegler, R. G., Smith-Warner, S. A., & Wang, M.
Journal titleStatistics in MedicinePooling data from multiple studies improves estimation of exposure-disease associations through increased sample size. However, biomarker exposure measurements can vary substantially across laboratories and often require calibration to a reference assay prior to pooling. We develop two statistical methods for aggregating biomarker data from multiple studies: the full calibration method and the internalized method. The full calibration method calibrates all biomarker measurements regardless of the availability of reference laboratory measurements while the internalized method calibrates only non-reference laboratory measurements. We compare the performance of these two aggregation methods to two-stage methods. Furthermore, we compare the aggregated and two-stage methods when estimating the calibration curve from controls only or from a random sample of individuals from the study cohort. Our findings include the following: (1) Under random sampling for calibration, exposure effect estimates from the internalized method have a smaller mean squared error than those from the full calibration method. (2) Under the controls-only calibration design, the full calibration method yields effect estimates with the least bias. (3) The two-stage approaches produce average effect estimates that are similar to the full calibration method under a controls only calibration design and the internalized method under a random sample calibration design. We illustrate the methods in an application evaluating the relationship between circulating vitamin D levels and stroke risk in a pooling project of cohort studies.
Hypothesis Tests for Neyman's Bias in Case–Control StudiesSwanson, D. M., Anderson, C. D., & Betensky, R.
Journal titleJournal of Applied Statistics
Page(s)1956-1977Survival bias is a long recognized problem in case–control studies, and many varieties of bias can come under this umbrella term. We focus on one of them, termed Neyman's bias or ‘prevalence–incidence bias’. It occurs in case–control studies when exposure affects both disease and disease-induced mortality, and we give a formula for the observed, biased odds ratio under such conditions. We compare our result with previous investigations into this phenomenon and consider models under which this bias may or may not be important. Finally, we propose three hypothesis tests to identify when Neyman's bias may be present in case–control studies. We apply these tests to three data sets, one of stroke mortality, another of brain tumors, and the last of atrial fibrillation, and find some evidence of Neyman's bias in the former two cases, but not the last case.
Immunophenotyping of pediatric brain tumors: correlating immune infiltrate with histology, mutational load, and survival and assessing clonal T cell responsePlant, A. S., Koyama, S., Sinai, C., Solomon, I. H., Griffin, G. K., Ligon, K. L., Bandopadhayay, P., Betensky, R., Emerson, R., Dranoff, G., Kieran, M. W., & Ritz, J.
Journal titleJournal of Neuro-Oncology
Page(s)269-278There is little known regarding the immune infiltrate present in pediatric brain tumors and how this compares to what is known about histologically similar adult tumors and its correlation with survival. Here, we provide a descriptive analysis of the immune infiltrate of 22 fresh pediatric brain tumor tissue samples of mixed diagnoses and 40 peripheral blood samples. Samples were analyzed using a flow cytometry panel containing markers for immune cell subtypes, costimulatory markers, inhibitory signals, and markers of activation. This was compared to the standard method of immunohistochemistry (IHC) for immune markers for 89 primary pediatric brain tumors. Both flow cytometry and IHC data did not correlate with the grade of tumor or mutational load and IHC data was not significantly associated with survival for either low grade or high grade gliomas. There is a trend towards a more immunosuppressive phenotype in higher grade tumors with more regulatory T cells present in these tumor types. Both PD1 and PDL1 were present in only a small percentage of the tumor infiltrate. T cell receptor sequencing revealed up to 10% clonality of T cells in tumor infiltrates and no significant difference in clonality between low and high grade gliomas. We have shown the immune infiltrate of pediatric brain tumors does not appear to correlate with grade or survival for a small sample of patients. Further research and larger studies are needed to fully understand the interaction of pediatric brain tumors and the immune system.
Integration of risk factors for Parkinson disease in 2 large longitudinal cohortsKim, I. Y., O’Reilly, Éilis J., Hughes, K. C., Gao, X., Schwarzschild, M. A., Hannan, M. T., Betensky, R., & Ascherio, A.
Page(s)e1646-e1653OBJECTIVE: To prospectively examine how selected lifestyle factors and family history of Parkinson disease (PD) combine to determine overall PD risk. METHODS: We derived risk scores among 69,968 women in the Nurses' Health Study (NHS) (1984-2012) and 45,830 men in the Health Professionals Follow-up Study (HPFS) (1986-2012). Risk scores were computed for each individual based on the following factors previously associated with PD risk: total caffeine intake, smoking, physical activity, and family history of PD for the NHS, and additionally total flavonoid intake and dietary urate index for the HPFS. Hazard ratios were estimated using Cox proportional hazards models. In addition, we performed tests of interactions on both the multiplicative and additive scale between pairs of risk factors. RESULTS: We documented 1,117 incident PD cases during follow-up. The adjusted hazard ratios comparing individuals in the highest category of the reduced risk score to those in the lowest category were 0.33 (95% confidence interval: 0.21, 0.49; ptrend < 0.0001) in the NHS and 0.18 (95% confidence interval: 0.10, 0.32; ptrend < 0.0001) in the HPFS. Results were similar when applying the risk scores computed by summing the predictors weighted by the log of their individual effect sizes on PD risk in these cohorts. Additive interaction was present between no family history of PD and caffeine in men and between caffeine and physical activity in women. CONCLUSIONS: Our results suggest that known protective factors for PD tend to have additive or superadditive effects, so that PD risk is very low in individuals with multiple protective risk factors.
Interaction between caffeine and polymorphisms of glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) and cytochrome P450 1A2 (CYP1A2) on Parkinson's disease riskKim, I. Y., O’Reilly, Éilis J., Hughes, K. C., Gao, X., Schwarzschild, M. A., McCullough, M. L., Hannan, M. T., Betensky, R., & Ascherio, A.
Journal titleMovement Disorders
Page(s)414-420Background: Caffeine intake has been inversely associated with Parkinson's disease (PD) risk. This relationship may be modified by polymorphisms of glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) and cytochrome P450 1A2 (CYP1A2), but the results of previous studies have been inconsistent. Method: We examined the interaction of caffeine intake with GRIN2A-rs4998386 and CYP1A2-rs762551 polymorphisms in influencing PD risk among 829 incident cases of PD and 2,754 matched controls selected among participants in the following 3 large prospective ongoing cohorts: the Nurses' Health Study, the Health Professionals' Follow-up Study, and the Cancer Prevention Study II Nutrition Cohort. Matching factors included cohort, birth year, source of DNA, date of DNA collection, and race. Relative risks and 95% confidence intervals were estimated using conditional logistic models. Interactions were tested both on the multiplicative scale and on the additive scale. Results: Overall, caffeine intake was associated with a lower PD risk (adjusted relative risk for highest versus lowest tertile = 0.70; 95% confidence interval, 0.57-0.86; p <.001). In analyses stratified by the GRIN2A-rs4998386 genotype, the multivariable-adjusted relative risk of PD comparing the highest to the lowest tertile of caffeine was 0.69 (95% confidence interval, 0.55-0.88; p <.01) among individuals homozygous for the C allele, and 0.85 (95% confidence interval, 0.55-1.32; p =.47; pRERI =.43) among carriers for the T allele. Interactions between caffeine and GRIN2A were not significant in either the multiplicative or additive scales. We also did not observe significant interactions for CYP1A2-rs762551 and incident PD risk. Conclusion: Our findings do not support the hypothesis of an interaction between the GRIN2A-rs4998386 or CYP1A2-rs762551 polymorphism and caffeine intake in determining PD risk.
Intravenous thrombolysis in unwitnessed stroke onset: MR WITNESS trial resultsFailed generating bibliography.Abstract
Journal titleAnnals of Neurology
Page(s)980-993Objective: Most acute ischemic stroke (AIS) patients with unwitnessed symptom onset are ineligible for intravenous thrombolysis due to timing alone. Lesion evolution on fluid-attenuated inversion recovery (FLAIR) magnetic resonance imaging (MRI) correlates with stroke duration, and quantitative mismatch of diffusion-weighted MRI with FLAIR (qDFM) might indicate stroke duration within guideline-recommended thrombolysis. We tested whether intravenous thrombolysis ≤4.5 hours from the time of symptom discovery is safe in patients with qDFM in an open-label, phase 2a, prospective study (NCT01282242). Methods: Patients aged 18 to 85 years with AIS of unwitnessed onset at 4.5 to 24 hours since they were last known to be well, treatable within 4.5 hours of symptom discovery with intravenous alteplase (0.9mg/kg), and presenting with qDFM were screened across 14 hospitals. The primary outcome was the risk of symptomatic intracranial hemorrhage (sICH) with preplanned stopping rules. Secondary outcomes included symptomatic brain edema risk, and functional outcomes of 90-day modified Rankin Scale (mRS). Results: Eighty subjects were enrolled between January 31, 2011 and October 4, 2015 and treated with alteplase at median 11.2 hours (IQR = 9.5–13.3) from when they were last known to be well. There was 1 sICH (1.3%) and 3 cases of symptomatic edema (3.8%). At 90 days, 39% of subjects achieved mRS = 0–1, as did 48% of subjects who had vessel imaging and were without large vessel occlusions. Interpretation: Intravenous thrombolysis within 4.5 hours of symptom discovery in patients with unwitnessed stroke selected by qDFM, who are beyond the recommended time windows, is safe. A randomized trial testing efficacy using qDFM appears feasible and is warranted in patients without large vessel occlusions. Ann Neurol 2018;83:980–993.
Inverse probability weighted Cox regression for doubly truncated dataMandel, M., De Uña-Álvarez, J., Simon, D. K., & Betensky, R.
Page(s)481-487Doubly truncated data arise when event times are observed only if they fall within subject-specific, possibly random, intervals. While non-parametric methods for survivor function estimation using doubly truncated data have been intensively studied, only a few methods for fitting regression models have been suggested, and only for a limited number of covariates. In this article, we present a method to fit the Cox regression model to doubly truncated data with multiple discrete and continuous covariates, and describe how to implement it using existing software. The approach is used to study the association between candidate single nucleotide polymorphisms and age of onset of Parkinson's disease.
Multicrossover Randomized Controlled Trial Designs in Alzheimer DiseaseArnold, S. E., & Betensky, R.
Journal titleAnnals of Neurology
Neuronal calcineurin transcriptional targets parallel changes observed in Alzheimer disease brainHopp, S. C., Bihlmeyer, N. A., Corradi, J. P., Vanderburg, C., Cacace, A. M., Das, S., Clark, T. W., Betensky, R., Hyman, B. T., & Hudry, E.
Journal titleJournal of Neurochemistry
Page(s)24-39Synaptic dysfunction and loss are core pathological features in Alzheimer disease (AD). In the vicinity of amyloid-β plaques in animal models, synaptic toxicity occurs and is associated with chronic activation of the phosphatase calcineurin (CN). Indeed, pharmacological inhibition of CN blocks amyloid-β synaptotoxicity. We therefore hypothesized that CN-mediated transcriptional changes may contribute to AD neuropathology and tested this by examining the impact of CN over-expression on neuronal gene expression in vivo. We found dramatic transcriptional down-regulation, especially of synaptic mRNAs, in neurons chronically exposed to CN activation. Importantly, the transcriptional profile parallels the changes in human AD tissue. Bioinformatics analyses suggest that both nuclear factor of activated T cells and numerous microRNAs may all be impacted by CN, and parallel findings are observed in AD. These data and analyses support the hypothesis that at least part of the synaptic failure characterizing AD may result from aberrant CN activation leading to down-regulation of synaptic genes, potentially via activation of specific transcription factors and expression of repressive microRNAs. Open Practices: Open Science: This manuscript was awarded with the Open Materials Badge. For more information see: https://cos.io/our-services/open-science-badges/. (Figure presented.). Read the Editorial Highlight for this article on page 8.
Permutation tests for general dependent truncationFailed generating bibliography.Abstract
Journal titleComputational Statistics and Data Analysis
Page(s)308-324Truncated survival data arise when the event time is observed only if it falls within a subject-specific region, known as the truncation set. Left-truncated data arise when there is delayed entry into a study, such that subjects are included only if their event time exceeds some other time. Quasi-independence of truncation and failure refers to factorization of their joint density in the observable region. Under quasi-independence, standard methods for survival data such as the Kaplan–Meier estimator and Cox regression can be applied after simple adjustments to the risk sets. Unlike the requisite assumption of independent censoring, quasi-independence can be tested, e.g., using a conditional Kendall's tau test. Current methods for testing for quasi-independence are powerful for monotone alternatives. Nonetheless, it is essential to detect any kind of deviation from quasi-independence so as not to report a biased Kaplan–Meier estimator or regression effect, which would arise from applying the simple risk set adjustment when dependence holds. Nonparametric, minimum p-value tests that are powerful against non-monotone alternatives are developed to offer protection against erroneous assumptions of quasi-independence. The use of conditional and unconditional methods of permutation for evaluation of the proposed tests is investigated in simulation studies. The proposed tests are applied to a study on the cognitive and functional decline in aging.
PET staging of amyloidosis using striatumFailed generating bibliography.Abstract
Journal titleAlzheimer's and Dementia
Page(s)1281-1292Introduction: Amyloid positron emission tomography (PET) data are commonly expressed as binary measures of cortical deposition. However, not all individuals with high cortical amyloid will experience rapid cognitive decline. Motivated by postmortem data, we evaluated a three-stage PET classification: low cortical; high cortical, low striatal; and high cortical, high striatal amyloid; hypothesizing this model could better reflect Alzheimer's dementia progression than a model based only on cortical measures. Methods: We classified PET data from 1433 participants (646 normal, 574 mild cognitive impairment, and 213 AD), explored the successive involvement of cortex and striatum using 3-year follow-up PET data, and evaluated the associations between PET stages, hippocampal volumes, and cognition. Results: Follow-up data indicated that PET detects amyloid first in cortex and then in striatum. Our three-category staging including striatum better predicted hippocampal volumes and subsequent cognition than a three-category staging including only cortical amyloid. Discussion: PET can evaluate amyloid expansion from cortex to subcortex. Using striatal signal as a marker of advanced amyloidosis may increase predictive power in Alzheimer's dementia research.
Polygenic pleiotropy and potential causal relationships between educational attainment, neurobiological profile, and positive psychotic symptomsLin, Y. F., Chen, C. Y., Öngür, D., Betensky, R., Smoller, J. W., Blacker, D., & Hall, M. H.
Journal titleTranslational Psychiatry
Issue1Event-related potential (ERP) components have been used to assess cognitive functions in patients with psychotic illness. Evidence suggests that among patients with psychosis there is a distinct heritable neurophysiologic phenotypic subtype captured by impairments across a range of ERP measures. In this study, we investigated the genetic basis of this "globally impaired" ERP cluster and its relationship to psychosis and cognitive abilities. We applied K-means clustering to six ERP measures to re-derive the globally impaired (n = 60) and the non-globally impaired ERP clusters (n = 323) in a sample of cases with schizophrenia (SCZ = 136) or bipolar disorder (BPD = 121) and healthy controls (n = 126). We used genome-wide association study (GWAS) results for SCZ, BPD, college completion, and childhood intelligence as the discovery datasets to derive polygenic risk scores (PRS) in our study sample and tested their associations with globally impaired ERP. We conducted mediation analyses to estimate the proportion of each PRS effect on severity of psychotic symptoms that is mediated through membership in the globally impaired ERP. Individuals with globally impaired ERP had significantly higher PANSS-positive scores (β = 3.95, P = 0.005). The SCZ-PRS was nominally associated with globally impaired ERP (unadjusted P = 0.01; R 2 = 3.07%). We also found a significant positive association between the college-PRS and globally impaired ERP (FDR-corrected P = 0.004; R 2 = 6.15%). The effect of college-PRS on PANSS positivity was almost entirely (97.1%) mediated through globally impaired ERP. These results suggest that the globally impaired ERP phenotype may represent some aspects of brain physiology on the path between genetic influences on educational attainment and psychotic symptoms.
Probing tumor microenvironment in patients with newly diagnosed glioblastoma during chemoradiation and adjuvant temozolomide with functional MRIIna Ly, K., Vakulenko-Lagun, B., Emblem, K. E., Ou, Y., Da, X., Betensky, R., Kalpathy-Cramer, J., Duda, D. G., Jain, R. K., Chi, A. S., Plotkin, S. R., Batchelor, T. T., Sorensen, G., Rosen, B. R., & Gerstner, E. R.
Journal titleScientific Reports
Issue1Functional MRI may identify critical windows of opportunity for drug delivery and distinguish between early treatment responders and non-responders. Using diffusion-weighted, dynamic contrast-enhanced, and dynamic susceptibility contrast MRI, as well as pro-angiogenic and pro-inflammatory blood markers, we prospectively studied the physiologic tumor-related changes in fourteen newly diagnosed glioblastoma patients during standard therapy. 153 MRI scans and blood collection were performed before chemoradiation (baseline), weekly during chemoradiation (week 1–6), monthly before each cycle of adjuvant temozolomide (pre-C1-C6), and after cycle 6. The apparent diffusion coefficient, volume transfer coefficient (Ktrans), and relative cerebral blood volume (rCBV) and flow (rCBF) were calculated within the tumor and edema regions and compared to baseline. Cox regression analysis was used to assess the effect of clinical variables, imaging, and blood markers on progression-free (PFS) and overall survival (OS). After controlling for additional covariates, high baseline rCBV and rCBF within the edema region were associated with worse PFS (microvessel rCBF: HR = 7.849, p = 0.044; panvessel rCBV: HR = 3.763, p = 0.032; panvessel rCBF: HR = 3.984; p = 0.049). The same applied to high week 5 and pre-C1 Ktrans within the tumor region (week 5 Ktrans: HR = 1.038, p = 0.003; pre-C1 Ktrans: HR = 1.029, p = 0.004). Elevated week 6 VEGF levels were associated with worse OS (HR = 1.034; p = 0.004). Our findings suggest a role for rCBV and rCBF at baseline and Ktrans and VEGF levels during treatment as markers of response. Functional imaging changes can differ substantially between tumor and edema regions, highlighting the variable biologic and vascular state of tumor microenvironment during therapy.