Marginal false discovery rate control for likelihood-based penalized regression models.

The popularity of penalized regression in high-dimensional data analysis has led to a demand for new inferential tools for these models. False discovery rate control is widely used in high-dimensional hypothesis testing, but has only recently been considered in the context of penalized regression. Almost all of this work, however, has focused on lasso-penalized linear regression. In this paper, we derive a general method for controlling the marginal false discovery rate that can be applied to any penalized likelihood-based model, such as logistic regression and Cox regression. Our approach is fast, flexible and can be used with a variety of penalty functions including lasso, elastic net, MCP, and MNet. We derive theoretical results under which the proposed method is valid, and use simulation studies to demonstrate that the approach is reasonably robust, albeit slightly conservative, when these assumptions are violated. Despite being conservative, we show that our method often offers more power to select causally important features than existing approaches. Finally, the practical utility of the method is demonstrated on gene expression datasets with binary and time-to-event outcomes.

Late Reinfection May Recur More Than 5 Years After Reimplantation of THA and TKA: Analysis of Pathogen Factors.

BACKGROUND: Two-stage reimplantation has consistently yielded high rates of success for patients with chronic prosthetic joint infection, although results more than 5 years after reimplantation are not commonly reported. Numerous factors may contribute to the risk of reinfection, although these factors-as well as the at-risk period after reimplantation-are not well characterized. QUESTIONS/PURPOSES: (1) What is the risk of reinfection after reimplantation for prosthetic joint infection at a minimum of 5 years? (2) Is the bacteriology of the index infection associated with late reinfection? (3) Is the presence of bacteria at the time of reimplantation associated with late reinfection? METHODS: Between 1995 and 2010, we performed 97 two-stage revisions in 93 patients for prosthetic joint infection of the hip or knee, and all are included in this retrospective study. During that time, the indications for this procedure generally were (1) infections occurring more than 3 months after the index arthroplasty; and (2) more acute infections associated with prosthetic loosening or resistant organisms. One patient (1%) was lost to followup; all others have a minimum of 5 years of followup (mean, 11 years; range, 5-20 years) and all living patients have been seen within the last 2 years. Patients were considered free from infection if they did not have pain at rest or constitutional symptoms such as fever, chills, or malaise. The patients' bacteriology and resistance patterns of these organisms were observed with respect to recurrence of infection. Odds ratios and Fisher's exact test were performed to analyze the data. The incidence of reinfection was determined using cumulative incidence methods that considered death as a competing event. RESULTS: Reinfection occurred in 12 of the 97 joints resulting in implant revision. The estimated 10-year cumulative incidence of infection was 14% (95% confidence interval [CI], 7%-23%) and incidence of infection from the same organism was 5% (95% CI, 1%-11%). Five occurred early or within 2 years and three were resistant pathogens (methicillin-resistant Staphylococcus aureus, methicillin-resistant Staphylococcus epidermidis, or vancomycin-resistant Enterococcus). Seven late hematogenous infections occurred and all were > 4 years after reimplantation and involved nonresistant organisms. Three of the five (60%) early infections were caused by resistant bacteria, whereas all seven late infections were caused by different organisms or a combination of different organisms than were isolated in the original infection. The early reinfections were more often caused by resistant organisms, whereas late infections involved different organisms than were isolated in the original infection and none involved resistant organisms. With the numbers available, we found no difference between patients in whom bacteria were detected at the time of reimplantation and those in whom cultures were negative in terms of the risk of reinfection 5 years after reimplantation (18.6% [18 of 97] versus 81.4% [79 of 97], odds ratio 1.56 [95% CI, 0.38-6.44]; p = 0.54); however, with only 93 patients, we may have been underpowered to make this analysis. CONCLUSIONS: In our study, resistant organisms were more often associated with early reinfection, whereas late failures were more commonly associated with new pathogens. We believe the most important finding in our study is that substantial risk of late infection remains even among patients who seemed free from infection 2 years after reimplantation for prosthetic joint infections of the hip or knee. This highlights the importance of educating our patients about the ongoing risk of prosthetic joint infection. LEVEL OF EVIDENCE: Level III, therapeutic study.

Impact of cannabis and low alcohol concentration on divided attention tasks during driving.

OBJECTIVE: To assess divided-attention performance when driving under the influence of cannabis with and without alcohol. Three divided-attention tasks were performed following administration of placebo, cannabis, and/or alcohol. METHODS: Healthy adult cannabis users participated in 6 sessions, receiving combinations of cannabis (placebo/low-THC/high-THC) and alcohol (placebo/active) in randomized order, separated by washout periods of ≥1 week. At 0.5 hours post-dosing, participants performed simulator drives in the University of Iowa National Advanced Driving Simulator (NADS-1), a full vehicle cab simulator with a 360° horizontal field of view and motion base that provides realistic feedback. Drives contained repeated instances of three tasks: a side-mirror task (reaction to a triangle appearing in the side-mirrors), an artist-search task (select a specified artist from a navigable menu on the vehicle's console), and a message-reading task (read aloud a message displayed on the console). Blood THC and breath alcohol concentration (BrAC) were interpolated using individual power curves from samples collected approximately 0.17, 0.42, 1.4, and 2.3 hours post-dose. Driving measures during tasks were compared to equal-duration control periods occurring just prior to the task. Performance shifts, task completion, and lane departures were modeled relative to blood THC and BrAC using mixed-effects regression models. RESULTS: Each 1 µg/L increase in blood THC concentration predicted increased odds of failing to complete the artist-search task (OR: 1.05, 95% CI: 1.01-1.11, p = 0.046), increased odds of selecting at least one incorrect response (OR: 1.05, 95% CI: 1.00-1.09, p = 0.041), declines in speed during the side-mirror task (0.005 m/s, 95% CI: 0.001-0.009, p = 0.023), and longer lane departure durations during the artist-search task (0.74% of task-period, 95% CI: 0.12-1.36 p = 0.020). BrAC (approximately 0.05%) was not associated with task performance, though each 0.01 g/210 L increase predicted longer departure durations during the side-mirror task (1.41% of task-period, 95% CI: 0.08-2.76, p = 0.040) and increased standard deviation of lane position in the message-reading task (0.61 cm, 95% CI: 0.14-1.08, p = 0.011). CONCLUSIONS: With increasing medical and legal cannabis use, understanding the impact of acute cannabis use on driving performance, including divided-attention, is essential. These data indicate that impaired divided-attention performance is a safety concern.

Histone deacetylases and NF-kB signaling coordinate expression of CX3CL1 in epithelial cells in response to microbial challenge by suppressing miR-424 and miR-503.

The NF-kB pathway is key to epithelial immune defense and has been implicated in secretion of antimicrobial peptides, release of cytokines/chemokines to mobilize immune effector cells, and activation of adaptive immunity. The expression of many inflammatory genes following infection involves the remodeling of the chromatin structure. We reported here that histone deacetylases (HDACs) and NF-kB signaling coordinate expression of CX3CL1 in epithelial cells following Cryptosporidium parvum infection. Upregulation of CX3CL1 was detected in cultured human biliary epithelial cells following infection. Expression of miR-424 and miR-503 was downregulated, and was involved in the induction of CX3CL1 in infected cells. C. parvum infection suppressed transcription of the mir-424-503 gene in a NF-kB- and HDAC-dependent manner. Increased promoter recruitment of NF-kB p50 and HDACs, and decreased promoter H3 acetylation associated with the mir-424-503 gene were observed in infected cells. Upregulation of CX3CL1 in biliary epithelial cells and increased infiltration of CX3CR1(+) cells were detected during C. parvum infection in vivo. Induction of CX3CL1 and downregulation of miR-424 and miR-503 were also detected in epithelial cells in response to LPS stimulation. The above results indicate that HDACs and NF-kB signaling coordinate epithelial expression of CX3CL1 to promote mucosal antimicrobial defense through suppression of the mir-424-503 gene.