Estimation and inference on correlations between biomarkers with repeated measures and left-censoring due to minimum detection levels

Generalized linear mixed model for binary outcomes when covariates are subject to measurement errors and detection limits

Longitudinal measurement of biomarkers is important in determining risk factors for binary endpoints such as infection or disease. However, biomarkers are subject to measurement error, and some are also subject to left-censoring due to a lower limit of detection. Statistical methods to address these issues are few. We herein propose a generalized linear mixed model and estimate the model parameters using the Monte Carlo Newton-Raphson (MCNR) method. Inferences regarding the parameters are made by applying Louis's method and the delta method. Simulation studies were conducted to compare the proposed MCNR method with existing methods including the maximum likelihood (ML) method and the ad hoc approach of replacing the left-censored values with half of the detection limit (HDL). The results showed that the performance of the MCNR method is superior to ML and HDL with respect to the empirical standard error, as well as the coverage probability for the 95% confidence interval. The HDL method uses an incorrect imputation method, and the computation is constrained by the number of quadrature points; while the ML method also suffers from the constrain for the number of quadrature points, the MCNR method does not have this limitation and approximates the likelihood function better than the other methods. The improvement of the MCNR method is further illustrated with real-world data from a longitudinal study of local cervicovaginal HIV viral load and its effects on oncogenic HPV detection in HIV-positive women.

Modeling the Kinetics of Serum Glial Fibrillary Acidic Protein, Ubiquitin Carboxyl-Terminal Hydrolase-L1, and S100B Concentrations in Patients with Traumatic Brain Injury

Glial fibrillary acidic protein (GFAP), ubiquitin carboxyl-terminal hydrolase-L1 (UCH-L1), and S100B have been shown to be predictive of patients with brain injury. Kinetics of these biomarkers in injured humans have not been extensively examined. This prospective multi-center study included patients with mild-to-moderate traumatic brain injury. Blood samples obtained at enrollment and every 6 h up to 24 h post-injury were assayed for GFAP, UCH-L1, and S100B. Random effects models examined changes in the biomarkers' level over time. A total of 167 patients were enrolled; mean age was 46.0 ± 17.8, 61.1% were male, 143 (85.6%) had a Glasgow Coma Scale score of 15, and 33 (19.8%) had a positive head computed tomography (CT) scan. Baseline median biomarker concentrations for all three were higher among CT-positive patients (p < 0.0001) but GFAP was the only biomarker that significantly increased over time among CT-positive patients relative to CT-negative patients (log transformed values 0.037; 95% confidence interval 0.02, 0.05; p < 0.001), indicating a 3.7% per hour rise in GFAP concentration. There was no significant increase in either UCH-L1 or S100B in CT-positive patients (p = 0.15 and p = 0.47, respectively). GFAP concentrations increased 3.7% per hour among CT-positive patients whereas neither UCH-L1 nor S100B increased, compared with CT-negative patients. The kinetics and temporal profile of GFAP suggest it may be a more robust biomarker to detect patients with positive CT findings, particularly at later post-injury times. Further study is needed to determine if GFAP is a useful test to follow throughout a patient's clinical course.

Genital tract HIV RNA levels and their associations with human papillomavirus infection and risk of cervical precancer

OBJECTIVE

Plasma HIV RNA levels have been associated with the risk of human papillomavirus (HPV) and cervical neoplasia in HIV-seropositive women. However, little is known regarding local genital tract HIV RNA levels and their relation with cervical HPV and neoplasia.

DESIGN/METHODS

In an HIV-seropositive women's cohort with semiannual follow-up, we conducted a nested case-control study of genital tract HIV RNA levels and their relation with incident high-grade squamous intraepithelial lesions (HSIL) subclassified as severe (severe HSIL), as provided for under the Bethesda 2001 classification system. Specifically, 66 incidents of severe HSIL were matched to 130 controls by age, CD4 count, highly active antiretroviral therapy use, and other factors. We also studied HPV prevalence, incident detection, and persistence in a random sample of 250 subjects.

RESULTS

Risk of severe HSIL was associated with genital tract HIV RNA levels (odds ratio comparing HIV RNA ≥ the median among women with detectable levels versus undetectable, 2.96; 95% confidence interval: 0.99 to 8.84; Ptrend = 0.03). However, this association became nonsignificant (Ptrend = 0.51) after adjustment for plasma HIV RNA levels. There was also no association between genital tract HIV RNA levels and the prevalence of any HPV or oncogenic HPV. However, the incident detection of any HPV (Ptrend = 0.02) and persistence of oncogenic HPV (Ptrend = 0.04) were associated with genital tract HIV RNA levels, after controlling plasma HIV RNA levels.

CONCLUSIONS

These prospective data suggest that genital tract HIV RNA levels are not a significant independent risk factor for cervical precancer in HIV-seropositive women, but they leave open the possibility that they may modestly influence HPV infection, an early stage of cervical tumorigenesis.