Search tips
Search criteria

Results 1-25 (49)

Clipboard (0)

Select a Filter Below

more »
Year of Publication
author:("Xue, xiaoyan")
1.  Use of the Dye Stain Assay and Ultraviolet Light Test for Assessing Vaginal Insertion of Placebo-filled Applicators Before and After Sex 
Sexually transmitted diseases  2013;40(12):939-943.
Applicator dye staining and ultraviolet (UV) light have been used in trials to measure adherence, but not in the setting of before and after sex gel dosing (BAT-24). This study was designed to determine if semen or pre-sex gel dosing impacts the sensitivity and specificity of a dye stain assay (DSA) for measuring vaginal insertion of placebo-filled applicators with BAT-24 dosing.
Healthy monogamous couples received Microlax®-type applicators filled with hydroxyethylcelluose placebo gel. Women were instructed to vaginally insert one dose of gel before and a second dose after sex and to return applicators within 48 hours after sex. Applicators were stained to detect semen followed by UV then DSA and scored by two readers. Positive and negative controls were randomly included in applicator batches.
Fifteen couples completed the study. Each female returned at least six applicators over a 30-day period. The sensitivity for insertion of post-sex applicators was higher for UV (97%) compared to DSA (90%) and the specificity was similar (≥96%). For pre-sex applicators, the sensitivity and specificity were higher for DSA (100%) compared to UV testing (87% sensitivity, 96% specificity). Among returned post-sex applicators, 95% tested positive by UV compared to 87% by DSA. Agreement between readers was significantly better on the pre-sex applicators for DSA than for UV and for post-sex readings agreement was less than half that for UV, although the results were not statistically significant.
Applicator tests are feasible for measuring adherence in trials with gel dosing before and after sex.
PMCID: PMC4143393  PMID: 24220355
2.  Insulin, Insulin-like Growth Factor-I, Endogenous Estradiol, and Risk of Colorectal Cancer in Postmenopausal Women 
Cancer research  2008;68(1):329-337.
Obesity is a risk factor for colorectal cancer, and hyperinsulinemia, a common condition in obese patients, may underlie this relationship. Insulin, in addition to its metabolic effects, has promitotic and antiapoptotic activity that may be tumorigenic. Insulin-like growth factor (IGF)-I, a related hormone, shares sequence homology with insulin, and has even stronger mitogenic effects. However, few prospective colorectal cancer studies directly measured fasting insulin, and none evaluated free IGF-I, or endogenous estradiol, a potential cofactor in postmenopausal women. Therefore, we conducted a case-cohort investigation of colorectal cancer among nondiabetic subjects enrolled in the Women’s Health Initiative Observational Study, a prospective cohort of 93,676 postmenopausal women. Fasting baseline serum specimens from all incident colorectal cancer cases (n = 438) and a random subcohort (n = 816) of Women’s Health Initiative Observational Study subjects were tested for insulin, glucose, total IGF-I, free IGF-I, IGF binding protein-3, and estradiol. Comparing extreme quartiles, insulin [hazard ratio (HR)q4–q1, 1.73; 95% confidence interval (CI), 1.16–2.57; ptrend = 0.005], waist circumference (HRq4–q1, 1.82; 95% CI, 1.22–2.70; ptrend = 0.001), and free IGF-I (HRq4–q1, 1.35; 95% CI, 0.92–1.98; Ptrend = 0.05) were each associated with colorectal cancer incidence in multivariate models. However, these associations each became nonsignificant when adjusted for one another. Endogenous estradiol levels, in contrast, were positively associated with risk of colorectal cancer (HR comparing high versus low levels, 1.53; 95% CI, 1.05–2.22), even after control for insulin, free IGF-I, and waist circumference. These data suggest the existence of at least two independent biological pathways that are related to colorectal cancer: one that involves endogenous estradiol, and a second pathway broadly associated with obesity, hyperinsulinemia, and free IGF-I.
PMCID: PMC4225702  PMID: 18172327
3.  Circulating Soluble Cytokine Receptors and Colorectal Cancer Risk 
Soluble cytokine receptors and receptor antagonist of proinflammatory cytokines can modify cytokine signaling and may affect cancer risk.
In a case-cohort study nested within the Women’s Health Initiative cohort of postmenopausal women, we assessed the associations of plasma levels of IL-1 receptor antagonist (IL-1Ra) and the soluble receptors of IL-1 (sIL-1R2), IL-6 (sIL-6R and sgp130), and TNF (sTNFR1 and sTNFR2) with risk of colorectal cancer in 433 cases and 821 subcohort subjects. Baseline levels of estradiol, insulin, leptin, IL-6, and TNF-α measured previously were also available for data analysis.
After adjusting for significant covariates – including age, race, smoking, colonoscopy history, waist circumference, and levels of estrogen, insulin, and leptin – relatively high levels of sIL-6R and sIL-1R2 were associated with reduced colorectal cancer risk [hazard ratios comparing extreme quartiles (HRQ4-Q1) for sIL-6R = 0.56, 95% CI = 0.38–0.83; HRQ4-Q1 for sIL-1R2 = 0.44, 95% CI = 0.29–0.67]. The associations with IL-1Ra, sgp130, sTNFR1, and sTNFR2 were null. The inverse association of sIL-1R2 with colorectal cancer risk persisted in cases diagnosed ≤5 and >5 years from baseline blood draw; the association with sIL-6R, however, was not evident in the latter group, possibly indicating that relatively low levels of sIL-6R in cases might be due to undiagnosed cancer at the time of blood draw.
High circulating levels of sIL-1R2 may be protective against colorectal carcinogenesis and/or be a marker of reduced risk for the disease.
sIL-1R2 has potential to be a chemopreventive and/or immunotherapeutic agent in inflammation-related diseases.
PMCID: PMC3947182  PMID: 24192010
soluble cytokine receptor; receptor antagonist; colorectal cancer; IL-1; IL-6; TNF
4.  Power and sample size calculations for evaluating mediation effects in longitudinal studies 
Statistical methods in medical research  2012;10.1177/0962280212465163.
Current methods of power and sample size calculations for the design of longitudinal studies to evaluate mediation effects are mostly based on simulation studies and do not provide closed form formulae. A further challenge due to the longitudinal study design is the consideration of missing data, which almost always occur in longitudinal studies due to staggered entry or drop out. In this paper, we consider the product of coefficients as a measure for the longitudinal mediation effect and evaluate three methods for testing the hypothesis on the longitudinal mediation effect: the joint significant test, the normal approximation and the test of b methods. Formulae for power and sample size calculations are provided under each method while taking into account missing data. Performance of the three methods under limited sample size are examined using simulation studies. An example from the Einstein Aging Study (EAS) is provided to illustrate the methods.
PMCID: PMC3883797  PMID: 23221975
drop out; joint significance test; linear mixed effects model; missing data; power analysis; product of coefficients
5.  Treatment-related changes in serum lipids and inflammation: clinical relevance remains unclear. Analyses from the Women's Interagency HIV Study 
AIDS (London, England)  2013;27(9):1516-1519.
Among 127 HIV-infected women, the magnitude of HDLc increases after HAART initiation predicted the magnitude of concurrent decreases in inflammation biomarkers. After HAART initiation, changes in LDLc and inflammation were unrelated. In the same population, predicted risk of coronary heart disease based upon levels of standard clinical risk factors was similar before and after HAART treatment. Thus, it remains unknown whether short-term treatment-related changes in standard risk factors may appreciably change risk of CVD.
PMCID: PMC3909663  PMID: 23435295
lipids; HAART; HIV infection; inflammation
6.  Improvement in Stroke Risk Prediction: Role of c-reactive protein (CRP) and Lipoprotein-Associated Phospholipase A2 (Lp-PLA2) in the Women’s Health Initiative 
Background and Purpose
Classification of risk of ischemic stroke is important for medical care and public health reasons. Whether addition of biomarkers adds to predictive power of the Framingham Stroke Risk or other traditional risk factors has not been studied in older women.
The Hormones and Biomarkers Predicting Stroke (HaBPS) Study is a case-control study of blood biomarkers assayed in 972 ischemic stroke cases and 972 controls, nested in the Women’s Health Initiative Observational Study of 93,676 postmenopausal women followed for an average of 8 years. We evaluated additive predictive value of two commercially available biomarkers: c-reactive protein (CRP) and Lipoprotein-Associated Phospholipase A2 (Lp-PLA2) to determine if they added to risk prediction by the Framingham Stroke Risk Score (FSRS) or by traditional risk factors (TRF) which included lipids and other variables not included in the FSRS. As measures of additive predictive value, we used the c-statistic, Net Reclassification Improvement (NRI), category-less NRI, and Integrated Discrimination Improvement Index (IDI).
Addition of CRP to Framingham risk models or additional traditional risk factors overall modestly improved prediction of ischemic stroke and resulted in overall NRI of 6.3%, (case NRI=3.9%, control NRI=2.4%) .In particular, hs-CRP was useful in prediction of cardioembolic strokes (NRI=12.0%; 95%CI: 4.3-19.6%) and in strokes occurring in less than 3 years (NRI=7.9%, 95%CI: 0.8-14.9%). Lp-PLA2 was useful in risk prediction of large artery strokes (NRI=19.8%, 95%CI: 7.4 -32.1%) and in early strokes (NRI=5.8%, 95%CI: 0.4-11.2%).
CRP and Lp-PLA2 can improve prediction of certain subtypes of ischemic stroke in older women, over the Framingham stroke risk model and traditional risk factors, and may help to guide surveillance and treatment of women at risk.
PMCID: PMC3556354  PMID: 23088183
7.  Sample size requirements to detect an intervention by time interaction in longitudinal cluster randomized clinical trials with random slopes 
In longitudinal cluster randomized clinical trials (cluster-RCT), subjects are nested within a higher level unit such as clinics and are evaluated for outcome repeatedly over the study period. This study design results in a three level hierarchical data structure. When the primary goal is to test the hypothesis that an intervention has an effect on the rate of change in the outcome over time and the between-subject variation in slopes is substantial, the subject-specific slopes are often modeled as random coefficients in a mixed-effects linear model. In this paper, we propose approaches for determining the samples size for each level of a 3-level hierarchical trial design based on ordinary least squares (OLS) estimates for detecting a difference in mean slopes between two intervention groups when the slopes are modeled as random. Notably, the sample size is not a function of the variances of either the second or the third level random intercepts and depends on the number of second and third level data units only through their product. Simulation results indicate that the OLS-based power and sample sizes are virtually identical to the empirical maximum likelihood based estimates even with varying cluster sizes. Sample sizes for random versus fixed slope models are also compared. The effects of the variance of the random slope on the sample size determinations are shown to be enormous. Therefore, when between-subject variations in outcome trends are anticipated to be significant, sample size determinations based on a fixed slope model can result in a seriously underpowered study.
PMCID: PMC3580878  PMID: 23459110
longitudinal cluster RCT; three level data; power; sample size; random slope; effect size
8.  Relation of HLA Class I and II Supertypes with Spontaneous Clearance of Hepatitis C Virus 
Genes and immunity  2013;14(5):330-335.
Human leukocyte antigen (HLA) genotype has been associated with probability of spontaneous clearance of hepatitis C virus (HCV). However, no prior studies have examined whether this relationship may be further characterized by grouping HLA alleles according to their supertypes, defined by their binding capacities. There is debate regarding the most appropriate method to define supertypes. Therefore, previously reported HLA supertypes (46 class I and 25 class II) were assessed for their relation with HCV clearance in a population of 758 HCV-seropositive women. Two HLA class II supertypes were significant in multivariable models that included: (i) supertypes with significant or borderline associations with HCV clearance after adjustment for multiple tests, and (ii) individual HLA alleles not part of these supertypes, but associated with HCV clearance in our prior study in this population. Specifically, supertype DRB3 (prevalence ratio (PR)=0.4; p=0.004) was associated with HCV persistence while DR8 (PR=1.8; p=0.01) was associated with HCV clearance. Two individual alleles (B*57:01 and C*01:02) associated with HCV clearance in our prior study became non-significant in analysis that included supertypes while B*57:03 (PR=1.9; p=0.008) and DRB1*07:01 (PR=1.7; p=0.005) retained significance. These data provide epidemiologic support for the significance of HLA supertypes in relation to HCV clearance.
PMCID: PMC3723800  PMID: 23636221
hepatitis C virus; HLA; human leukocyte antigen; supertype
9.  Estimation and inference on correlations between biomarkers with repeated measures and left-censoring due to minimum detection levels 
Statistics in medicine  2012;31(21):10.1002/sim.5371.
Statistical approaches for estimating and drawing inference on the correlation between two biomarkers which are repeatedly assessed over time and subject to left-censoring due to minimum detection levels are lacking. We propose a linear mixed-effects model and estimate the parameters with the Monte Carlo Expectation Maximization (MCEM) method. Inferences regarding the model parameters and the correlation between the biomarkers are performed by applying Louis’s method and the delta method. Simulation studies were conducted to compare the proposed MCEM method with existing methods including the MLE method, the multiple imputation (MI) method, and two widely used ad hoc approaches: replacing the censored values with the detection limit (DL) or with half of the detection limit (HDL). The results show that the performance of the MCEM with respect to relative bias and coverage probability for the 95% confidence interval is superior to the DL and HDL approaches and exceeds that of the MI method at medium to high levels of censoring, and the standard error estimates from the MCEM method are close to ideal. The MLE method can estimate the parameters accurately; however, a non-positive definite information matrix can occur so that the variances are not estimable. These five methods are illustrated with data from a longitudinal HIV study to estimate and draw inference on the correlation between HIV RNA levels measured in plasma and in cervical secretions at multiple time points.
PMCID: PMC3875381  PMID: 22714546
information matrix; longitudinal data; mixed-effects; monte carlo expectation maximization
10.  Association of subclinical atherosclerosis with lipid levels amongst antiretroviral-treated and untreated HIV-infected women in the Women’s Interagency HIV Study 
Atherosclerosis  2012;225(2):408-411.
We examined serum lipids in association with carotid artery intima-media thickness (CIMT) in HIV-infected and HIV-uninfected women.
In 2003–4, among 1827 Women’s Interagency HIV Study participants, we measured CIMT and lipids (high-density lipoprotein cholesterol [HDL-c], low-density lipoprotein cholesterol [LDL-c], total cholesterol [TC], non-HDL-c). A subset of 520 treated HIV-infected women had pre-1997 lipid measures. We used multivariable linear regression to examine associations between lipids and CIMT.
In HIV-uninfected women, higher TC, LDL-c and non-HDL-c were associated with increased CIMT. Among HIV-infected women, associations of lipids with CIMT were observed in treated but not untreated women. Among the HIV-infected women treated in 2003–4, CIMT was associated both with lipids measured a decade earlier in infection, and with late lipid measurements.
Among HIV-infected women, hyperlipidemia is most strongly associated with subclinical atherosclerosis in treated women. Among treated women, the association appeared strongest early in the disease course.
PMCID: PMC3696584  PMID: 23089369
cardiovascular diseases; carotid arteries; HAART; HIV; lipids
11.  Adipokines Linking Obesity with Colorectal Cancer Risk in Postmenopausal Women 
Cancer research  2012;72(12):3029-3037.
Mechanistic associations between obesity and colorectal cancer remain unclear. In this study, we investigated whether adipokines are risk factors for colorectal cancer and whether they may mediate its association with obesity. In a case–cohort study nested within the Women’s Health Initiative cohort of postmenopausal women, baseline plasma samples from 457 colorectal cancer cases and 841 subcohort subjects were assayed for seven adipokines—adiponectin, leptin, plasminogen activator inhibitor-1 (PAI-1), resistin, hepatocyte growth factor, interleukin-6 (IL-6), and TNF-α. Serum insulin and estradiol values measured previously were also available for data analysis. After adjusting for age, race, smoking, colonoscopy history, and estrogen level, a low level of antiinflammatory adiponectin and high levels of proinflammatory leptin, PAI-1, and IL-6 were associated with increased colorectal cancer risk, though only leptin remained significant after further adjustment for insulin [HRs comparing extreme quartiles (HRQ4–Q1), 1.84; 95% CI, 1.17–2.90]. Mediation analyses showed that leptin and insulin partially explained the association between waist circumference and colorectal cancer and attenuated it by 25% and 37%, respectively, with insulin being a significant mediator (P = 0.041). Our findings support the conclusion that adipokines involved in inflammation are associated with colorectal cancer risk, but that their effects may be mediated mostly by insulin, with leptin exerting an independent effect. Hyperinsulinemia and hyperleptinemia may therefore partially explain the adiposity association with colorectal cancer in postmenopausal women.
PMCID: PMC3790260  PMID: 22511581
12.  Cervicovaginal HPV Infection Before and After Hysterectomy: Evidence of Different Tissue Tropism for Oncogenic and Non-Oncogenic HPV Types in a Cohort of HIV-positive and HIV-negative Women 
Human papillomavirus (HPV) is detected in nearly all cervical cancers and approximately half of vaginal cancers. However, vaginal cancer is an order of magnitude less common than cervical cancer, not only in the general population but also among women with HIV/AIDS. It is interesting therefore that recent studies found that HPV was common in both normal vaginal and cervical tissue, with higher prevalence of non-oncogenic HPV types in the vagina. In the current investigation, we prospectively examined HPV infection in 86 HIV-positive and 17 HIV-negative women who underwent hysterectomy during follow-up in a longitudinal cohort. Cervicovaginal lavage specimens were obtained semi-annually and tested for HPV DNA by PCR. To address possible selection biases associated with having a hysterectomy, subjects acted as their own comparison group – before versus after hysterectomy. The average HPV prevalence was higher in HIV-positive than HIV-negative women both before (59% versus 12%; P<0.001) and after hysterectomy (56% versus 6%; P<0.001). Multivariate random effects models (within-individual comparisons) demonstrated significantly lower HPV prevalence (odds ratio [OR]=0.71; 95% confidence interval [CI]=0.59-0.85) after hysterectomy. The association of HPV prevalence with hysterectomy was similar among HIV-positive and HIV-negative women. However, hysterectomy had greater effects on oncogenic (OR=0.48; 95%CI=0.35-0.66) than non-oncogenic HPV types (OR=0.89; 95%CI=0.71-1.11; Pinteraction=0.002). Overall, we observed greater reductions in oncogenic than non-oncogenic HPV prevalence following hysterectomy. If correct, these data could suggest that oncogenic HPV have greater tropism for cervical compared with vaginal epithelium, consistent with the lower incidence of vaginal than cervical cancer.
PMCID: PMC3321069  PMID: 22120980
vaginal; HPV; hysterectomy; viral tropism; HIV
13.  A Prospective Investigation of Coffee Drinking and Endometrial Cancer Incidence 
Coffee drinking may be associated with reduced risk of endometrial cancer; however, prospective data are limited. Further, it is not clear whether any association between coffee and endometrial cancer differs according to coffee caffeine content. The association of coffee drinking with incidence of endometrial cancer was evaluated among 226,732 women, aged 50–71, enrolled in the NIH-AARP Diet and Health Study who completed a baseline epidemiologic questionnaire. Following a mean 9.3 years of follow-up, data were available for 1,486 incident endometrial cancer cases. Cox proportional hazards models were used to estimate associations of coffee with endometrial cancer incidence. Sub-group analyses were performed according to smoking status, hormone therapy use (HT) and body habitus. Coffee drinking was inversely related to incidence of endometrial cancer (Hazard Ratio [HR] comparing drinking of >3 cups/day versus no cups=0.64, 95%CI, 0.51–0.80; Ptrend= 0.0004). The association of coffee with endometrial cancer risk was apparent for consumption of both regular (HR per cup= 0.90, 95%CI, 0.86–0.95) and decaffeinated coffee (HR per cup=0.93, 95%CI, 0.87–0.99). The relation of coffee with endometrial cancer incidence varied significantly by HT use (Pinteraction=0.03) with an association only apparent among HT-never users (HR comparing drinking >3 cups/day versus no cups= 0.54, 95%CI, 0.41–0.72; Ptrend=0.0005). Endometrial cancer incidence appears to be reduced among women that habitually drink coffee, an association that does not differ according to caffeine content.
PMCID: PMC3288610  PMID: 22021096
14.  Detectability and reproducibility of plasma levels of chemokines and soluble receptors☆ 
Results in Immunology  2013;3:79-84.
Background: Multiplex assays are available to measure an array of circulating chemokines, soluble cytokine receptors and growth factors. However, there is limited information regarding whether these analytes are suitable for large-scale epidemiological studies to assess their relationships with chronic diseases, including cancer.
Methods: We examined detectability, assay repeatability, and 3-year within-subject reproducibility of plasma levels of 25 chemokines and 11 soluble receptors of cytokines and growth factors selected from the Human Millipore Panels. Plasma samples were obtained from 36 men (average age 62 years) and 17 women (average age 32 years) who participated in two epidemiological studies. Inter-assay and within-subject reproducibility were assessed by intraclass correlation coefficients (ICC).
Results: All analytes, except lymphotactin (47% detectability), were detectable in >90% of plasma samples. Inter-assay reproducibility for all analytes in 36 men tested three times on separate days were good to excellent (ICCs: 0.71–1.00). Within-subject reproducibility in 17 women sampled three times in three years were excellent (ICC ≥ 0.75) for five chemokines (eotaxin, fractalkine, 6Ckine, eotaxin 3, and SDF-1α+β) and three soluble receptors (sIL-1R2, sIL-4R and sVEGFR2); ICCs were fair to good (0.4 ≤ ICC < 0.75) for 15 chemokines and eight soluble receptors. However, five chemokines (GRO, IP-10, MIP-1β, BCA-1, and MIP-3α) had ICC < 0.4, suggesting biological variability.
Conclusion: Multiplex assays for plasma levels of selected chemokines and soluble receptors showed good to excellent assay detectability and repeatability. Most analytes also had good 3-year within-subject reproducibility, indicating that a single measurement of these analytes may be used to assess biomarker-disease associations.
PMCID: PMC3908322  PMID: 24600562
Chemokines; Soluble receptors; Within-subject variability; Biomarker; Limit of detection
15.  Potential cardiovascular disease risk markers among HIV-infected women initiating antiretroviral treatment 
Inflammation and hemostasis perturbation may be involved in vascular complications of HIV infection. We examined atherogenic biomarkers and subclinical atherosclerosis in HIV-infected adults before and after beginning highly-active antiretroviral therapy (HAART).
In the Women's Interagency HIV Study (WIHS), 127 HIV-infected women studied pre- and post-HAART were matched to HIV-uninfected controls. Six semi-annual measurements of soluble CD14, tumor necrosis factor (TNF)-alpha, soluble interleukin (IL)-2 receptor, IL-6, IL-10, monocyte chemoattractant protein (MCP)-1, D-dimer, and fibrinogen were obtained. Carotid artery intima-media thickness (CIMT) was measured by B-mode ultrasound.
Relative to HIV-uninfected controls, HAART-naïve HIV-infected women had elevated levels of soluble CD14 (1945 vs 1662 ng/mL, Wilcoxon signed rank P<0.0001), TNF-alpha (6.3 vs 3.4 pg/mL, P<0.0001), soluble IL-2 receptor (1587 vs 949 pg/mL, P<0.0001), IL-10 (3.3 vs 1.9 pg/mL, P<0.0001), MCP-1 (190 vs 163 pg/mL, P<0.0001) and D-dimer (0.43 vs 0.31 µg/mL, P<0.01). Elevated biomarker levels declined after HAART. While most biomarkers normalized to HIV-uninfected levels, in women on effective HAART, TNF-alpha levels remained elevated compared to HIV-uninfected women (+0.8 pg/mL, P=0.0002). Higher post-HAART levels of soluble IL-2 receptor (P=0.02), IL-6 (P=0.05), and D-dimer (P=0.03) were associated with increased CIMT.
Untreated HIV infection is associated with abnormal hemostasis (e.g., D-dimer), and pro-atherogenic (e.g., TNF-alpha) and anti-atherogenic (e.g., IL-10) inflammatory markers. HAART reduces most inflammatory mediators to HIV-uninfected levels. Increased inflammation and hemostasis are associated with subclinical atherosclerosis in recently treated women. These findings have potential implications for long-term risk of cardiovascular disease in HIV-infected patients, even with effective therapy.
PMCID: PMC3400505  PMID: 22592585
antiretroviral therapy; cardiovascular diseases; cytokines; hemostasis; HIV; inflammation
16.  A Comparison of the Polytomous Logistic Regression and Joint Cox Proportional Hazards Models for Evaluating Multiple Disease Subtypes in Prospective Cohort Studies 
Polytomous logistic regression models are commonly used in case-control studies of cancer to directly compare the risks associated with an exposure variable across multiple cancer subtypes. However, the validity, accuracy and efficiency of this approach for prospective cohort studies have not been formally evaluated.
We investigated the performance of the polytomous logistic regression model and compared it to an alternative approach based on a joint Cox proportional hazards model using simulation studies. We then applied both methods to a prospective cohort study to assess whether the association of breast cancer with body size differs according to estrogen and progesterone receptor-defined subtypes.
Our simulations showed that the polytomous logistic regression model but not the joint Cox regression model yielded biased results in comparing exposure and disease subtype associations when the baseline hazards for different disease subtypes are non-proportional. For this reason, an analysis of a real data set was based on the joint Cox proportional hazards model and showed that body size has a significantly greater association with estrogen and progesterone positive breast cancer than with other subtypes.
Because of the limitations of the polytomous logistic regression model for the comparison of exposure-disease associations across disease subtypes, the joint Cox proportional hazards model is recommended over the polytomous logistic regression model in prospective cohort studies.
The paper will promote the use of the joint Cox model in a prospective cohort study. Examples of SAS and S-plus programming codes are provided to facilitate use by non-statisticians.
PMCID: PMC3565022  PMID: 23292084
time to event data; proportional hazards functions; robust variance; competing risk
17.  Testing the proportional hazards assumption in case-cohort analysis 
Case-cohort studies have become common in epidemiological studies of rare disease, with Cox regression models the principal method used in their analysis. However, no appropriate procedures to assess the assumption of proportional hazards of case-cohort Cox models have been proposed.
We extended the correlation test based on Schoenfeld residuals, an approach used to evaluate the proportionality of hazards in standard Cox models. Specifically, pseudolikelihood functions were used to define “case-cohort Schoenfeld residuals”, and then the correlation of these residuals with each of three functions of event time (i.e., the event time itself, rank order, Kaplan-Meier estimates) was determined. The performances of the proposed tests were examined using simulation studies. We then applied these methods to data from a previously published case-cohort investigation of the insulin/IGF-axis and colorectal cancer.
Simulation studies showed that each of the three correlation tests accurately detected non-proportionality. Application of the proposed tests to the example case-cohort investigation dataset showed that the Cox proportional hazards assumption was not satisfied for certain exposure variables in that study, an issue we addressed through use of available, alternative analytical approaches.
The proposed correlation tests provide a simple and accurate approach for testing the proportional hazards assumption of Cox models in case-cohort analysis. Evaluation of the proportional hazards assumption is essential since its violation raises questions regarding the validity of Cox model results which, if unrecognized, could result in the publication of erroneous scientific findings.
PMCID: PMC3710085  PMID: 23834739
Proportional hazards; Schoenfeld residuals; Case-cohort studies; Cox models
18.  Cytomegalovirus Immunoglobulin G Antibody Is Associated With Subclinical Carotid Artery Disease Among HIV-Infected Women 
The Journal of Infectious Diseases  2012;205(12):1788-1796.
Background. Cytomegalovirus (CMV) infection has been implicated in immune activation and accelerated progression of immunodeficiency from human immunodeficiency virus (HIV) coinfection. We hypothesized that CMV is associated with vascular disease in HIV-infected adults.
Methods. In the Women's Interagency HIV Study, we studied 601 HIV-infected and 90 HIV-uninfected participants. We assessed the association of CMV immunoglobulin G (IgG) level with carotid artery intima-media thickness, carotid artery distensibility, Young's elastic modulus, and blood pressures. Multivariable models adjusted for age, race/ethnicity, smoking, diabetes, and body mass index.
Results. Mean CMV IgG levels were higher in HIV-infected women compared with HIV-uninfected women (P < .01). Among HIV-infected women, higher CMV IgG level was associated with decreased carotid artery distensibility (P < .01) and increased Young's modulus (P = .02). Higher CMV IgG antibody level was associated with increased prevalence of carotid artery lesions among HIV-infected women who achieved HIV suppression on antiretroviral therapy, but not among viremic or untreated HIV-infected women. Adjustment for Epstein–Barr virus antibody levels and C-reactive protein levels had no effect on the associations between CMV IgG levels and vascular parameters.
Conclusions. Cytomegalovirus antibody titers are increased in HIV-infected women and associated with subclinical cardiovascular disease. Host responses to CMV may be abnormal in HIV infection and associated with clinical disease.
PMCID: PMC3415890  PMID: 22492856
19.  Risk of Cervical Pre-Cancer and Cancer Among HIV-Infected Women With Normal Cervical Cytology and No Evidence of Oncogenic HPV Infection 
U.S. cervical cancer screening guidelines for HIV-uninfected women 30 years of age and older have recently been revised, increasing the suggested interval between Pap tests from three years to five years among those with normal cervical cytology (the Pap test) who test negative for oncogenic human papillomavirus (HPV). Whether a three-year or five-year screening interval might be used in HIV-infected women who are cytologically normal and oncogenic HPV-negative is unknown.
To determine the risk of cervical pre-cancer or cancer defined cytologically (high-grade squamous intraepithelial lesions or greater [HSIL+]) or histologically (cervical intraepithelial neoplasia 2 or greater [CIN-2+]), as two separate endpoints, in HIV-infected women and HIV-uninfected women who at baseline had a normal Pap test and were negative for oncogenic HPV.
Design, Setting and Participants
Participants included 420 HIV-infected women and 279 HIV-uninfected women with normal cervical cytology at their enrollment in a multi-institutional cohort, between October 1, 2001 and September 30, 2002, with follow-up through April 30, 2011. Clinical sites were in the Bronx, Brooklyn, Chicago, Los Angeles, San Francisco, and Washington, DC. Semi-annual visits included Pap testing and, if indicated, cervical biopsy. Cervicovaginal lavage specimens from enrollment were tested for HPV DNA using PCR. The primary analysis was truncated at five years of follow-up.
Main Outcome Measure
The five-year cumulative incidence of cervical pre-cancer and cancer.
No oncogenic HPV was detected in 369 (88%; 95% CI, 84%-91%) of the HIV-infected women and 255 (91%; 95% CI, 88%-94%) of the HIV-uninfected women with normal cervical cytology at enrollment. Among these oncogenic HPV-negative women two cases of HSIL+ were observed; an HIV-uninfected woman and an HIV-infected woman with a CD4 cell count of 500/μL or greater. Histologic data were obtained from four of the six sites. There were six cases of CIN-2+ in N=145 HIV-uninfected women (cumulative incidence = 5% [95% CI, 1%-8%]) and nine cases in N=219 HIV-infected women (cumulative incidence = 5% [95% CI, 2%-8%]). This included one case of CIN-2+ in N=44 oncogenic HPV-negative HIV-infected women with CD4 cell counts less than 350/μL (cumulative incidence = 2% [95% CI, 0%-7%]), one case in N=47 women with CD4 cell counts of 350 to 499/μL (cumulative incidence = 2% [95% CI, 0%-7%]), and seven cases in N=128 women with CD4 cell counts of 500/μL or greater (cumulative incidence = 6% [95% CI, 2%-10%]). One HIV-infected and one HIV-uninfected woman had CIN-3, but none had cancer.
The five-year cumulative incidence of HSIL+ and CIN-2+ was similar in HIV-infected women and HIV-uninfected women who were cytologically normal and oncogenic HPV-negative at enrollment.
PMCID: PMC3556987  PMID: 22820789
20.  Additive Hazard Regression Models: An Application to the Natural History of Human Papillomavirus 
There are several statistical methods for time-to-event analysis, among which is the Cox proportional hazards model that is most commonly used. However, when the absolute change in risk, instead of the risk ratio, is of primary interest or when the proportional hazard assumption for the Cox proportional hazards model is violated, an additive hazard regression model may be more appropriate. In this paper, we give an overview of this approach and then apply a semiparametric as well as a nonparametric additive model to a data set from a study of the natural history of human papillomavirus (HPV) in HIV-positive and HIV-negative women. The results from the semiparametric model indicated on average an additional 14 oncogenic HPV infections per 100 woman-years related to CD4 count < 200 relative to HIV-negative women, and those from the nonparametric additive model showed an additional 40 oncogenic HPV infections per 100 women over 5 years of followup, while the estimated hazard ratio in the Cox model was 3.82. Although the Cox model can provide a better understanding of the exposure disease association, the additive model is often more useful for public health planning and intervention.
PMCID: PMC3569891  PMID: 23424606
21.  The Relationship Between Specific Cognitive Functions and Falls in Aging 
Neuropsychology  2007;21(5):540-548.
The current study examined the relationship between cognitive function and falls in elders who did not meet criteria for dementia or Mild Cognitive Impairment (n=172). To address limitations of previous research, associations between cognitive function and falls controlled for the confounding effects of gait measures and other risk factors. A neuropsychological test battery was submitted to factor analysis yielding three orthogonal factors (verbal IQ, Speed/Executive Attention, Memory). Single and recurrent falls within the last 12 months were evaluated. We hypothesized that Speed/Executive Attention would be associated with falls. Additionally, we assessed whether associations between different cognitive functions and falls varied depending on whether single or recurrent falls were examined. Multivariate logistic regressions showed that worse scores on Speed/Executive Attention were associated with increased single and recurrent falls. Worse scores on Verbal IQ were related only to increased recurrent falls. Memory was not associated with either single or recurrent falls. These findings are relevant to risk assessment and prevention of falls, and point to possible shared neural substrate of cognitive and motor function.
PMCID: PMC3476056  PMID: 17784802
cognition; falls; aging
22.  Quantitative assessment of invasive mena isoforms (Menacalc) as an independent prognostic marker in breast cancer 
Breast Cancer Research : BCR  2012;14(5):R124.
Mena, an Ena/VASP protein family member, is a key actin regulatory protein. Mena is up-regulated in breast cancers and promotes invasion and motility of tumor cells. Mena has multiple splice variants, including Mena invasive (MenaINV) and Mena11a, which are expressed in invasive or non-invasive tumor cells, respectively. We developed a multiplex quantitative immunofluorescence (MQIF) approach to assess the fraction of Mena lacking 11a sequence as a method to infer the presence of invasive tumor cells represented as total Mena minus Mena11a (called Menacalc) and determined its association with metastasis in breast cancer.
The MQIF method was applied to two independent primary breast cancer cohorts (Cohort 1 with 501 and Cohort 2 with 296 patients) using antibodies against Mena and its isoform, Mena11a. Menacalc was determined for each patient and assessed for association with risk of disease-specific death.
Total Mena or Mena11a isoform expression failed to show any statistically significant association with outcome in either cohort. However, assessment of Menacalc showed that relatively high levels of this biomarker is associated with poor outcome in two independent breast cancer cohorts (log rank P = 0.0004 for Cohort 1 and 0.0321 for Cohort 2). Multivariate analysis on combined cohorts revealed that high Menacalc is associated with poor outcome, independent of age, node status, receptor status and tumor size.
High Menacalc levels identify a subgroup of breast cancer patients with poor disease-specific survival, suggesting that Menacalc may serve as a biomarker for metastasis.
PMCID: PMC3962029  PMID: 22971274
23.  Iron homeostasis and distal colorectal adenoma risk in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial 
Red meat consumption has been positively associated with colorectal cancer; however, the biologic mechanism underlying this relationship is not understood. Red meat is a major source of iron, which may play a role in colorectal carcinogenesis via increased crypt cell proliferation, cytotoxicity, and endogenous N-nitrosation. In a nested case-control study within the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, we prospectively evaluated multiple iron exposure parameters, including dietary intake and serum measures of iron, ferritin, transferrin, total iron binding capacity (TIBC), and unsaturated iron binding capacity (UIBC) in relation to incident colorectal adenoma in 356 cases and 396 matched, polyp-free controls. We also investigated variation in eight key genes involved in iron homeostasis in relation to colorectal adenoma in an additional series totaling 1,126 cases and 1,173 matched controls. We observed a positive association between red meat intake and colorectal adenoma (odds ratio comparing extreme quartiles [ORq4-q1] = 1.59, 95% confidence interval [CI]: 1.02-2.49, P-trend = 0.03). Serum TIBC and UIBC were inversely associated with colorectal adenoma (ORq4-q1 = 0.57, 95% CI: 0.37-0.88, P-trend = 0.03; and ORq4-q1 = 0.62, 95% CI: 0.40-0.95, P-trend = 0.04, respectively). Colorectal adenoma was not associated with serum ferritin, iron, or transferrin saturation, or with polymorphisms in genes involved in iron homeostasis. Serum TIBC and UIBC, parameters which have a reciprocal relationship with overall iron load, were inversely related to colorectal adenoma, suggesting that individuals with lower iron status have a reduced risk of colorectal adenoma.
PMCID: PMC3168068  PMID: 21685236
Diet; meat; iron; colorectal; adenoma; cancer
24.  T cell activation predicts carotid artery stiffness among HIV-infected women 
Atherosclerosis  2011;217(1):207-213.
HIV disease is associated with increased arterial stiffness, which may be related to inflammation provoked by HIV-related immune perturbation. We assessed the association of T cell markers of immune activation and immunosenescence with carotid artery stiffness among HIV-infected women.
Among 114 HIV-infected and 43 HIV-uninfected women, we measured CD4+ and CD8+ T cell populations expressing activation (CD38+HLA-DR+) and senescence (CD28-CD57+) markers. We then related these measures of immune status with parameters of carotid artery stiffness, including decreased distensibility, and increased Young’s elastic modulus, as assessed by B-mode ultrasound.
HIV infection was associated with increased CD4+ T cell activation, CD8+ T cell activation and CD8+ T cell senescence. Among HIV-infected women, adjusted for age, HIV medications, and vascular risk factors, higher CD4+CD38+HLA-DR+ T cell frequency was associated with decreased carotid artery distensibility (β= −2.00, 95% confidence interval [CI]= −3.86,−0.14, P=0.04) and increased Young’s modulus (β=1.00, 95% CI=0.03,1.97, P=0.04). These associations were affected little by further adjustment for CD4+ T cell count and viral load. Among HIV-infected women, higher frequencies of immunosenescent T cells, including CD4+CD28-CD57+ and CD8+CD28-CD57+ T cells, were also associated with decreased arterial distensibility. Among HIV-uninfected women, frequencies of activated or senescent T cells were not significantly associated with measures of carotid stiffness.
T cell activation and senescence are associated with arterial stiffness, suggesting that pro-inflammatory populations of T cells may produce functional or structural vascular changes in HIV-infected women.
PMCID: PMC3139014  PMID: 21492857
25.  Transient Versus Persistent Fear of Falling in Community-Dwelling Older Adults: Incidence and Risk Factors 
To investigate the incidence of fear of falling (FOF) and the risk factors associated with transient versus persistent FOF in community-dwelling older adults.
Prospective cohort study.
Bronx County, New York.
Three hundred eighty participants without FOF at baseline in the Einstein Aging Study aged 70 and older.
FOF was assessed at baseline and during follow-up interviews at 2- to 3-month intervals for a minimum 2 years. Incident FOF was classified as transient or persistent FOF. Transient FOF was defined as new-onset FOF reported at only one interview, and persistent FOF was FOF reported at two or more interviews over a 2-year period.
Twenty-four-month cumulative incidence of incident FOF was 45.4%, with 60.0% of FOF being persistent. Predictors of incident FOF included female sex (adjusted hazard ratio (aHR) = 1.55, 95% confidence interval (CI) = 1.08–2.23), depressive symptoms (aHR = 1.16, 95% CI = 1.07–1.26), falls (aHR = 1.50, 95% CI = 1.01–2.21), and clinical gait abnormality (aHR = 2.07, 95% CI = 1.42–3.01). The proportion of participants with incident FOF increased linearly with increasing number of risk factors. Predictors for transient and persistent FOF were depressive symptoms and clinical gait abnormality. Female sex and previous falls were predictors of persistent but not transient FOF.
FOF status in older adults may change over time, with shared and distinct risk factors for persistent and transient FOF. Understanding the dynamic nature of FOF and these risk factors will help identify high-risk groups and design future intervention studies.
PMCID: PMC3298667  PMID: 21718266
fear of falling; risk factors; older adults

Results 1-25 (49)