Individual risk prediction: Comparing random forests with Cox proportional-hazards model by a simulation study

With big data becoming widely available in healthcare, machine learning algorithms such as random forest (RF) that ignores time-to-event information and random survival forest (RSF) that handles right-censored data are used for individual risk prediction alternatively to the Cox proportional hazards (Cox-PH) model. We aimed to systematically compare RF and RSF with Cox-PH. RSF with three split criteria [log-rank (RSF-LR), log-rank score (RSF-LRS), maximally selected rank statistics (RSF-MSR)]; RF, Cox-PH, and Cox-PH with splines (Cox-S) were evaluated through a simulation study based on real data. One hundred eighty scenarios were investigated assuming different associations between the predictors and the outcome (linear/linear and interactions/nonlinear/nonlinear and interactions), training sample sizes (500/1000/5000), censoring rates (50%/75%/93%), hazard functions (increasing/decreasing/constant), and number of predictors (seven, 15 including noise variables). Methods' performance was evaluated with time-dependent area under curve and integrated Brier score. In all scenarios, RF had the worst performance. In scenarios with a low number of events (⩽70), Cox-PH was at least noninferior to RSF, whereas under linearity assumption it outperformed RSF. Under the presence of interactions, RSF performed better than Cox-PH as the number of events increased whereas Cox-S reached at least similar performance with RSF under nonlinear effects. RSF-LRS performed slightly worse than RSF-LR and RSF-MSR when including noise variables and interaction effects. When applied to real data, models incorporating survival time performed better. Although RSF algorithms are a promising alternative to conventional Cox-PH as data complexity increases, they require a higher number of events for training. In time-to-event analysis, algorithms that consider survival time should be used.     

Performance of the marginal structural models under various scenarios of incomplete marker's values: A simulation study

Marginal structural models (MSMs) have been proposed for estimating a treatment's effect, in the presence of time‐dependent confounding. We aimed to evaluate the performance of the Cox MSM in the presence of missing data and to explore methods to adjust for missingness. We simulated data with a continuous time‐dependent confounder and a binary treatment. We explored two classes of missing data: (i) missed visits, which resemble clinical cohort studies; (ii) missing confounder's values, which correspond to interval cohort studies. Missing data were generated under various mechanisms. In the first class, the source of the bias was the extreme treatment weights. Truncation or normalization improved estimation. Therefore, particular attention must be paid to the distribution of weights, and truncation or normalization should be applied if extreme weights are noticed. In the second case, bias was due to the misspecification of the treatment model. Last observation carried forward (LOCF), multiple imputation (MI), and inverse probability of missingness weighting (IPMW) were used to correct for the missingness. We found that alternatives, especially the IPMW method, perform better than the classic LOCF method. Nevertheless, in situations with high marker's variance and rarely recorded measurements none of the examined method adequately corrected the bias.     

New horizons in schizophrenia treatment: autophagy protection is coupled with behavioral improvements in a mouse model of schizophrenia

Autophagy plays a key role in the pathophysiology of schizophrenia as manifested by a 40% decrease in BECN1/Beclin 1 mRNA in postmortem hippocampal tissues relative to controls. This decrease was coupled with the deregulation of the essential ADNP (activity-dependent neuroprotector homeobox), a binding partner of MAP1LC3B/LC3B (microtubule-associated protein 1 light chain 3 β) another major constituent of autophagy. The drug candidate NAP (davunetide), a peptide fragment from ADNP, enhanced the ADNP-LC3B interaction. Parallel genetic studies have linked allelic variation in the gene encoding MAP6/STOP (microtubule-associated protein 6) to schizophrenia, along with altered MAP6/STOP protein expression in the schizophrenic brain and schizophrenic-like behaviors in Map6-deficient mice. In this study, for the first time, we reveal significant decreases in hippocampal Becn1 mRNA and reversal by NAP but not by the antipsychotic clozapine (CLZ) in Map6-deficient (Map6^+/−) mice. Normalization of Becn1 expression by NAP was coupled with behavioral protection against hyperlocomotion and cognitive deficits measured in the object recognition test. CLZ reduced hyperlocomotion below control levels and did not significantly affect object recognition. The combination of CLZ and NAP resulted in normalized outcome behaviors. Phase II clinical studies have shown NAP-dependent augmentation of functional activities of daily living coupled with brain protection. The current studies provide a new mechanistic pathway and a novel avenue for drug development.     

Quantitation of human immunodeficiency virus type 1 DNA forms with the second template switch in peripheral blood cells predicts disease progression independently of plasma RNA load

There are several forms of human immunodeficiency virus type 1 (HIV-1) DNA in peripheral blood T cells and lymph nodes in untreated HIV-1-infected individuals and in patients whose plasma HIV-1 RNA levels are suppressed by long-term combination antiretroviral therapy. However, it remains to be established whether the concentration of HIV-1 DNA in cells predicts the clinical outcome of HIV-1 infection. In this report, we measured the concentration of HIV-1 DNA forms which has undergone the second template switch (STS DNA) and 2-long-terminal-repeat DNA circles in peripheral blood mononuclear cell (PBMC) samples. To do this, we used molecular-beacon-based real-time PCR assays and studied 130 patients with hemophilia in the Multicenter Hemophilia Cohort Study. We assessed the influence of baseline HIV-1 STS DNA levels on the progression of HIV-1 disease in the absence of combination antiretroviral therapy by Kaplan-Meier and Cox regression analysis. Among the patients who progressed to AIDS, the median levels (interquartile ranges) of STS HIV-1 DNA in PBMC were significantly higher than those of patients who remained AIDS free during the 16 years of follow-up (1,017 [235 to 6,059] and 286 [31 to 732] copies per 10(6) PBMC, respectively; P < 0.0001). Rates of progression to death and development of AIDS varied significantly (log rank P < 0.001) by quartile distribution of HIV-1 STS DNA levels. After adjustment for age at seroconversion, baseline CD4(+) T-cell counts, plasma viral load, and T-cell-receptor excision circles, the relative hazards (RH) of death and AIDS were significantly increased with higher HIV-1 STS DNA levels (adjusted RH, 1.84 [95% confidence interval (CI), 1.30 to 2.59] and 2.62 [95% CI, 1.75 to 3.93] per 10-fold increase per 10(6) PBMC, respectively). HIV-1 STS DNA levels in each individual remained steady in longitudinal PBMC samples during 16 years of follow-up. Our findings show that the concentration of HIV-1 STS DNA in PBMC complements the HIV-1 RNA load in plasma in predicting the clinical outcome of HIV-1 disease. This parameter may have important implications for understanding the virological response to combination antiretroviral therapy.     

Virologic and Immunologic Response to cART by HIV-1 Subtype in the CASCADE Collaboration

Background

We aimed to compare rates of virologic response and CD4 changes after combination antiretroviral (cART) initiation in individuals infected with B and specific non-B HIV subtypes.

Methods

Using CASCADE data we analyzed HIV-RNA and CD4 counts for persons infected ≥1996, ≥15 years of age. We used survival and longitudinal modeling to estimate probabilities of virologic response (confirmed HIV-RNA <500 c/ml), and failure (HIV-RNA>500 c/ml at 6 months or ≥1000 c/ml following response) and CD4 increase after cART initiation.

Results

2003 (1706 B, 142 CRF02_AG, 55 A, 53 C, 47 CRF01_AE) seroconverters were included in analysis. There was no evidence of subtype effect overall for response or failure (p = 0.075 and 0.317, respectively) although there was a suggestion that those infected with subtypes CRF01_AE and A responded sooner than those with subtype B infection [HR (95% CI):1.37 (1.01–1.86) and 1.29 (0.96–1.72), respectively]. Rates of CD4 increase were similar in all subtypes except subtype A, which tended to have lower initial, but faster long-term, increases.

Conclusions

Virologic and immunologic response to cART was similar across all studied subtypes but statistical power was limited by the rarity of some non-B subtypes. Current antiretroviral agents seem to have similar efficacy in subtype B and most widely encountered non-B infections in high-income countries.     

Differences in HIV natural history among African and non-African seroconverters in Europe and seroconverters in sub-Saharan Africa

Introduction

It is unknown whether HIV treatment guidelines, based on resource-rich country cohorts, are applicable to African populations.

Methods

We estimated CD4 cell loss in ART-naïve, AIDS-free individuals using mixed models allowing for random intercept and slope, and time from seroconversion to clinical AIDS, death and antiretroviral therapy (ART) initiation by survival methods. Using CASCADE data from 20 European and 3 sub-Saharan African (SSA) cohorts of heterosexually-infected individuals, aged ≥15 years, infected ≥2000, we compared estimates between non-African Europeans, Africans in Europe, and Africans in SSA.

Results

Of 1,959 (913 non-Africans, 302 Europeans - African origin, 744 SSA), two-thirds were female; median age at seroconversion was 31 years. Individuals in SSA progressed faster to clinical AIDS but not to death or non-TB AIDS. They also initiated ART later than Europeans and at lower CD4 cell counts. In adjusted models, Africans (especially from Europe) had lower CD4 counts at seroconversion and slower CD4 decline than non-African Europeans. Median (95% CI) CD4 count at seroconversion for a 15–29 year old woman was 607 (588–627) (non-African European), 469 (442–497) (European - African origin) and 570 (551–589) (SSA) cells/µL with respective CD4 decline during the first 4 years of 259 (228–289), 155 (110–200), and 199 (174–224) cells/µL (p<0.01).

Discussion

Despite differences in CD4 cell count evolution, death and non-TB AIDS rates were similar across study groups. It is therefore prudent to apply current ART guidelines from resource-rich countries to African populations.     

Short-term effects of ambient sulphur dioxide and particulate matter on mortality in 12 European cities: results from time series data from the APHEA project. Air Pollution and Health: a European Approach.

OBJECTIVES: To carry out a prospective combined quantitative analysis of the associations between all cause mortality and ambient particulate matter and sulphur dioxide. DESIGN: Analysis of time series data on daily number of deaths from all causes and concentrations of sulphur dioxide and particulate matter (measured as black smoke or particles smaller than 10 microns in diameter (PM10)) and potential confounders. SETTING: 12 European cities in the APHEA project (Air Pollution and Health: a European Approach). MAIN OUTCOME MEASURE: Relative risk of death. RESULTS: In western European cities it was found that an increase of 50 micrograms/m3 in sulphur dioxide or black smoke was associated with a 3% (95% confidence interval 2% to 4%) increase in daily mortality and the corresponding figure for PM10 was 2% (1% to 3%). In central eastern European cities the increase in mortality associated with a 50 micrograms/m3 change in sulphur dioxide was 0.8% (-0.1% to 2.4%) and in black smoke 0.6% (0.1% to 1.1%). Cumulative effects of prolonged (two to four days) exposure to air pollutants resulted in estimates comparable with the one day effects. The effects of both pollutants were stronger during the summer and were mutually independent. CONCLUSIONS: The internal consistency of the results in western European cities with wide differences in climate and environmental conditions suggest that these associations may be causal. The long term health impact of these effects is uncertain, but today''s relatively low levels of sulphur dioxide and particles still have detectable short term effects on health and further reductions in air pollution are advisable.