Can incorporating genotyping data into efficacy estimators improve efficiency of early phase malaria vaccine trials?

BACKGROUND

Early phase malaria vaccine field trials typically measure malaria infection by PCR or thick blood smear microscopy performed on serially sampled blood. Vaccine efficacy (VE) is the proportion reduction in an endpoint due to vaccination and is often calculated as VEHR = 1-hazard ratio or VERR = 1-risk ratio. Genotyping information can distinguish different clones and distinguish multiple infections over time, potentially increasing statistical power. This paper investigates two alternative VE endpoints incorporating genotyping information: VEmolFOI, the vaccine-induced proportion reduction in incidence of new clones acquired over time, and VEC, the vaccine-induced proportion reduction in mean number of infecting clones per exposure.

METHODS

Power of VEmolFOI and VEC was compared to that of VEHR and VERR by simulations and analytic derivations, and the four VE methods were applied to three data sets: a Phase 3 trial of RTS,S malaria vaccine in 6912 African infants, a Phase 2 trial of PfSPZ Vaccine in 80 Burkina Faso adults, and a trial comparing Plasmodium vivax incidence in 466 Papua New Guinean children after receiving chloroquine + artemether lumefantrine with or without primaquine (as these VE methods can also quantify effects of other prevention measures). By destroying hibernating liver-stage P. vivax, primaquine reduces subsequent reactivations after treatment completion.

RESULTS

In the trial of RTS,S vaccine, a significantly reduced number of clones at first infection was observed, but this was not the case in trials of PfSPZ Vaccine or primaquine, although the PfSPZ trial lacked power to show a reduction. Resampling smaller data sets from the large RTS,S trial to simulate phase 2 trials showed modest power gains from VEC compared to VEHR for data like those from RTS,S, but VEC is less powerful than VEHR for trials in which the number of clones at first infection is not reduced. VEmolFOI was most powerful in model-based simulations, but only the primaquine trial collected enough serial samples to precisely estimate VEmolFOI. The primaquine VEmolFOI estimate decreased after most control arm liver-stage infections reactivated (which mathematically resembles a waning vaccine), preventing VEmolFOI from improving power.

CONCLUSIONS

The power gain from the genotyping methods depends on the context. Because input parameters for early phase power calculations are often uncertain, these estimators are not recommended as primary endpoints for small trials unless supported by targeted data analysis.

TRIAL REGISTRATIONS

NCT00866619, NCT02663700, NCT02143934.

Can incorporating genotyping data into efficacy estimators improve efficiency of early phase malaria vaccine trials?

Background

Early phase malaria vaccine field trials typically measure malaria infection by PCR or thick blood smear microscopy performed on serially sampled blood. Vaccine efficacy (VE) is the proportion reduction in an endpoint due to vaccination and is often calculated as V E H R = 1 - hazard ratio or V E R R = 1 - risk ratio. Genotyping information can distinguish different clones and distinguish multiple infections over time, potentially increasing statistical power. This paper investigates two alternative VE endpoints incorporating genotyping information: V E m o l F O I , the vaccine-induced proportion reduction in incidence of new clones acquired over time, and V E C , the vaccine-induced proportion reduction in mean number of infecting clones per exposure.

Methods

We used simulations and analytic derivations to compare power of these methods to V E H R and V E R R and applied them to three data sets: a Phase 3 trial of RTS,S malaria vaccine in 6912 African infants, a Phase 2 trial of PfSPZ Vaccine in 80 Burkina Faso adults, and a trial comparing Plasmodium vivax incidence in 466 Papua New Guinean children after receiving chloroquine + artemether lumefantrine with or without primaquine (as these VE methods can also quantify effects of other prevention measures). By destroying hibernating liver-stage P. vivax, primaquine reduces subsequent reactivations after treatment completion.

Results

The RTS,S vaccine significantly reduced the number of clones at first infection, but PfSPZ vaccine and primaquine did not. Resampling smaller data sets from the large RTS,S trial to simulate phase 2 trials showed modest power gains from V E C compared to V E H R for data like RTS,S, but V E C is less powerful than V E H R for vaccines which do not reduce the number of clones at first infection. V E m o l F O I was most powerful in model-based simulations, but only the primaquine trial collected enough serial samples to precisely estimate V E m o l F O I . The primaquine V E m o l F O I estimate decreased after most control arm liver-stage infections reactivated (which mathematically resembles a waning vaccine), preventing V E m o l F O I from improving power.

Conclusions

The power gain from the genotyping methods depends on the context. Because input parameters for early phase power calculations are often uncertain, we recommend against these estimators as primary endpoints for small trials unless supported by targeted data analysis.

Trial registrations

NCT00866619, NCT02663700, NCT02143934.

Vaccine efficacy at a point in time

Vaccine trials are generally designed to assess efficacy on clinical disease. The vaccine effect on infection, while important both as a proxy for transmission and to describe a vaccine's entire effects, requires frequent (e.g., twice a week) longitudinal sampling to capture all infections. Such sampling may not always be feasible. A logistically easy approach is to collect a sample to test for infection at a regularly scheduled visit. Such point or cross-sectional sampling does not permit estimation of classic vaccine efficacy on infection, as long duration infections are sampled with higher probability. Building on work by Rinta-Kokko and others (2009) and Lipsitch and Kahn (2021), we evaluate proxies of the vaccine effect on transmission at a point in time; the vaccine efficacy on prevalent infection and on prevalent viral load, VE$_{\rm PI}$ and VE$_{\rm PVL}$, respectively. Longer infections with higher viral loads should have more transmission potential and prevalent vaccine efficacy naturally captures this aspect. We demonstrate how these parameters obtain from an underlying proportional hazards model for infection and allow for waning efficacy on infection, duration, and viral load. We estimate these parameters based on regression models with either repeated cross-sectional sampling or frequent longitudinal sampling. We evaluate the methods by simulation and analyze a phase III vaccine trial with polymerase chain reaction (PCR) cross-sectional sampling for subclinical infection.

A note on familywise error rate for a primary and secondary endpoint

Hung, Wang, and O'Neill (2007) considered the problem of controlling the type I error rate for a primary and secondary endpoint in a clinical trial using a gatekeeping approach in which the secondary endpoint is tested only if the primary endpoint crosses its monitoring boundary. They considered a two-look trial and showed by simulation that the naive method of testing the secondary endpoint at full level α at the time the primary endpoint reaches statistical significance does not control the familywise error rate at level α. Tamhane et al. (2010) derived analytic expressions for familywise error rate and power and confirmed the inflated error rate of the naive approach. Nonetheless, many people mistakenly believe that the closure principle can be used to prove that the naive procedure controls the familywise error rate. The purpose of this note is to explain in greater detail why there is a problem with the naive approach and show that the degree of alpha inflation can be as high as that of unadjusted monitoring of a single endpoint. This article is protected by copyright. All rights reserved.

Joint testing of overall and simple effects for the two-by-two factorial trial design

BACKGROUND/AIMS

The two-by-two factorial design randomizes participants to receive treatment A alone, treatment B alone, both treatments A and B(AB), or neither treatment (C). When the combined effect of A and B is less than the sum of the A and B effects, called a subadditive interaction, there can be low power to detect the A effect using an overall test, that is, factorial analysis, which compares the A and AB groups to the C and B groups. Such an interaction may have occurred in the Action to Control Cardiovascular Risk in Diabetes blood pressure trial (ACCORD BP) which simultaneously randomized participants to receive intensive or standard blood pressure, control and intensive or standard glycemic control. For the primary outcome of major cardiovascular event, the overall test for efficacy of intensive blood pressure control was nonsignificant. In such an instance, simple effect tests of A versus C and B versus C may be useful since they are not affected by a subadditive interaction, but they can have lower power since they use half the participants of the overall trial. We investigate multiple testing procedures which exploit the overall tests' sample size advantage and the simple tests' robustness to a potential interaction.

METHODS

In the time-to-event setting, we use the stratified and ordinary logrank statistics' asymptotic means to calculate the power of the overall and simple tests under various scenarios. We consider the A and B research questions to be unrelated and allocate 0.05 significance level to each. For each question, we investigate three multiple testing procedures which allocate the type 1 error in different proportions for the overall and simple effects as well as the AB effect. The Equal Allocation 3 procedure allocates equal type 1 error to each of the three effects, the Proportional Allocation 2 procedure allocates 2/3 of the type 1 error to the overall A (respectively, B) effect and the remaining type 1 error to the AB effect, and the Equal Allocation 2 procedure allocates equal amounts to the simple A (respectively, B) and AB effects. These procedures are applied to ACCORD BP.

RESULTS

Across various scenarios, Equal Allocation 3 had robust power for detecting a true effect. For ACCORD BP, all three procedures would have detected a benefit of intensive glycemia control.

CONCLUSIONS

When there is no interaction, Equal Allocation 3 has less power than a factorial analysis. However, Equal Allocation 3 often has greater power when there is an interaction. The R package factorial2x2 can be used to explore the power gain or loss for different scenarios.

Predictive cluster level surrogacy in the presence of interference

Surrogate evaluation is a difficult problem that is made more so by the presence of interference. Our proposed procedure can allow for relatively easy evaluation of surrogates for indirect or spill-over clinical effects at the cluster level. Our definition of surrogacy is based on the causal-association paradigm (Joffe and Greene, 2009. Related causal frameworks for surrogate outcomes. Biometrics65, 530-538), under which surrogates are evaluated by the strength of the association between a causal treatment effect on the clinical outcome and a causal treatment effect on the candidate surrogate. Hudgens and Halloran (2008, Toward causal inference with interference. Journal of the American Statistical Association103, 832-842) introduced estimators that can be used for many of the marginal causal estimands of interest in the presence of interference. We extend these to consider surrogates for not just direct effects, but indirect and total effects at the cluster level. We suggest existing estimators that can be used to evaluate biomarkers under our proposed definition of surrogacy. In our motivating setting of a transmission blocking malaria vaccine, there is expected to be no direct protection to those vaccinated and predictive surrogates are urgently needed. We use a set of simulated data examples based on the proposed Phase IIb/III trial design of transmission blocking malaria vaccine to demonstrate how our definition, proposed criteria and procedure can be used to identify biomarkers as predictive cluster level surrogates in the presence of interference on the clinical outcome.

Estimation of infection density and epidemic size of COVID-19 using the back-calculation algorithm

The novel coronavirus (COVID-19) is continuing its spread across the world, claiming more than 160,000 lives and sickening more than 2,400,000 people as of April 21, 2020. Early research has reported a basic reproduction number (R0) between 2.2 to 3.6, implying that the majority of the population is at risk of infection if no intervention measures were undertaken. The true size of the COVID-19 epidemic remains unknown, as a significant proportion of infected individuals only exhibit mild symptoms or are even asymptomatic. A timely assessment of the evolving epidemic size is crucial for resource allocation and triage decisions. In this article, we modify the back-calculation algorithm to obtain a lower bound estimate of the number of COVID-19 infected persons in China in and outside the Hubei province. We estimate the infection density among infected and show that the drastic control measures enforced throughout China following the lockdown of Wuhan City effectively slowed down the spread of the disease in two weeks. We also investigate the COVID-19 epidemic size in South Korea and find a similar effect of its "test, trace, isolate, and treat" strategy. Our findings are expected to provide guidelines and enlightenment for surveillance and control activities of COVID-19 in other countries around the world.

Immune correlates analysis using vaccinees from test negative designs

Determining the effect of vaccine-induced immune response on disease risk is an important goal of vaccinology. Typically, immune correlates analyses are conducted prospectively with immune response measured shortly after vaccination and subsequent disease status regressed on immune response. In outbreaks and rare disease settings, collecting samples from all vaccinees is not feasible. The test negative design is a retrospective design used to measure vaccine efficacy where symptomatic individuals who present at a clinic are assessed for relevant disease (cases) or some other disease (controls) and vaccination status ascertained. This article proposes that test negative vaccinees have immune response to vaccine assessed both for relevant (e.g., Ebola) and irrelevant (e.g., vector) proteins. If the latter immune response is unaffected by active (Ebola) infection, and is correlated with the relevant immune response, it can serve as a proxy for the immune response of interest proximal to infection. We show that logistic regression using imputed immune response as the covariate and case disease as outcome can estimate the prospective immune response slope and detail the assumptions needed for unbiased inference. The method is evaluated by simulation under various scenarios including constant and decaying immune response. A simulated dataset motivated by ring vaccination for an ongoing Ebola outbreak is analyzed.

Infection Density and Epidemic Size of COVID-19 in China outside the Hubei province

The novel coronavirus (COVID-19) has spread to almost all countries in the world, claiming more than 160,000 lives and sickening more than 2,400,000 people by April 21, 2020. There has been research showing that on average, each infected person spreads the infection to more than two persons. Therefore the majority of the population is at risk of infection if no intervention measures were undertaken. The true size of the COVID-19 epidemic remains unknown, as a significant proportion of infected individuals only exhibit mild symptoms or are even asymptomatic. A timely assessment of the evolving epidemic size is crucial for resource allocation and triage decisions. In this article, we modify the back-calculation algorithm to obtain a lower bound estimate of the number of COVID-19 infected persons in China outside the Hubei province. We estimate the infection density among infected and show that the drastic control measures enforced throughout China following the lockdown of Wuhan City effectively slowed down the spread of the disease in two weeks. Our findings from China are expected to provide guidelines and enlightenment for surveillance and control activities of COVID-19 in other countries around the world.

A unified evaluation of differential vaccine efficacy

Many infectious diseases are well prevented by proper vaccination. However, when a vaccine is not completely efficacious, there is great interest in determining how the vaccine effect differs in subgroups conditional on measured immune responses postvaccination and also according to the type of infecting agent (eg, strain of a virus). The former is often called correlate of protection (CoP) analysis, while the latter has been called sieve analysis. We propose a unified framework for simultaneously assessing CoP and sieve effects of a vaccine in a large Phase III randomized trial. We use flexible parametric models treating times to infection from different agents as competing risks and estimated maximum likelihood to fit the models. The parametric models under competing risks allow for estimation of both cumulative incidence-based contrasts and instantaneous rates. We outline the assumptions with which we can link the observable data to the causal contrasts of interest, propose hypothesis testing procedures, and evaluate our proposed methods in an extensive simulation study.

On Causal Inferences for Personalized Medicine: How Hidden Causal Assumptions Led to Erroneous Causal Claims About the D-Value

Personalized medicine asks if a new treatment will help a particular patient, rather than if it improves the average response in a population. Without a causal model to distinguish these questions, interpretational mistakes arise. These mistakes are seen in an article by Demidenko [2016] that recommends the "D-value," which is the probability that a randomly chosen person from the new-treatment group has a higher value for the outcome than a randomly chosen person from the control-treatment group. The abstract states "The D-value has a clear interpretation as the proportion of patients who get worse after the treatment" with similar assertions appearing later. We show these statements are incorrect because they require assumptions about the potential outcomes which are neither testable in randomized experiments nor plausible in general. The D-value will not equal the proportion of patients who get worse after treatment if (as expected) those outcomes are correlated. Independence of potential outcomes is unrealistic and eliminates any personalized treatment effects; with dependence, the D-value can even imply treatment is better than control even though most patients are harmed by the treatment. Thus, D-values are misleading for personalized medicine. To prevent misunderstandings, we advise incorporating causal models into basic statistics education.

Causal estimands and confidence intervals associated with Wilcoxon-Mann-Whitney tests in randomized experiments

Although the P value from a Wilcoxon-Mann-Whitney test is used often with randomized experiments, it is rarely accompanied with a causal effect estimate and its confidence interval. The natural parameter for the Wilcoxon-Mann-Whitney test is the Mann-Whitney parameter, ϕ, which measures the probability that a randomly selected individual in the treatment arm will have a larger response than a randomly selected individual in the control arm (plus an adjustment for ties). We show that the Mann-Whitney parameter may be framed as a causal parameter and show that it is not equal to a closely related and nonidentifiable causal effect, ψ, the probability that a randomly selected individual will have a larger response under treatment than under control (plus an adjustment for ties). We review the paradox, first expressed by Hand, that the ψ parameter may imply that the treatment is worse (or better) than control, while the Mann-Whitney parameter shows the opposite. Unlike the Mann-Whitney parameter, ψ is nonidentifiable from a randomized experiment. We review some nonparametric assumptions that rule out Hand's paradox through bounds on ψ and use bootstrap methods to make inferences on those bounds. We explore the relationship of the proportional odds parameter to Hand's paradox, showing that the paradox may occur for proportional odds parameters between 1/9 and 9. Thus, large effects are needed to ensure that if treatment appears better by the Mann-Whitney parameter, then treatment improves responses in most individuals. We demonstrate these issues using a vaccine trial.

A boundary-optimized rejection region test for the two-sample binomial problem

Testing the equality of 2 proportions for a control group versus a treatment group is a well-researched statistical problem. In some settings, there may be strong historical data that allow one to reliably expect that the control proportion is one, or nearly so. While one-sample tests or comparisons to historical controls could be used, neither can rigorously control the type I error rate in the event the true control rate changes. In this work, we propose an unconditional exact test that exploits the historical information while controlling the type I error rate. We sequentially construct a rejection region by first maximizing the rejection region in the space where all controls have an event, subject to the constraint that our type I error rate does not exceed α for any true event rate; then with any remaining α we maximize the additional rejection region in the space where one control avoids the event, and so on. When the true control event rate is one, our test is the most powerful nonrandomized test for all points in the alternative space. When the true control event rate is nearly one, we demonstrate that our test has equal or higher mean power, averaging over the alternative space, than a variety of well-known tests. For the comparison of 4 controls and 4 treated subjects, our proposed test has higher power than all comparator tests. We demonstrate the properties of our proposed test by simulation and use our method to design a malaria vaccine trial.

Impact of Intravenous Immunoglobulin on Survival in Necrotizing Fasciitis With Vasopressor-Dependent Shock: A Propensity Score-Matched Analysis From 130 US Hospitals

Background

Shock frequently complicates necrotizing fasciitis (NF) caused by group A Streptococcus (GAS) or Staphylococcus aureus. Intravenous immunoglobulin (IVIG) is sometimes administered for presumptive toxic shock syndrome (TSS), but its frequency of use and efficacy are unclear.

Methods

Adult patients with NF and vasopressor-dependent shock undergoing surgical debridement from 2010 to 2014 were identified at 130 US hospitals. IVIG cases were propensity-matched and risk-adjusted. The primary outcome was in-hospital mortality and the secondary outcome was median length of stay (LOS).

Results

Of 4127 cases of debrided NF with shock at 121 centers, only 164 patients (4%) at 61 centers received IVIG. IVIG subjects were younger with lower comorbidity indices, but higher illness severity. Clindamycin and vasopressor intensity were higher among IVIG cases, as was coding for TSS and GAS. In-hospital mortality did not differ between matched IVIG and non-IVIG groups (crude mortality, 27.3% vs 23.6%; adjusted odds ratio, 1.00 [95% confidence interval, .55-1.83]; P = .99). Early IVIG (≤2 days) did not alter this effect (P = .99). Among patients coded for TSS, GAS, and/or S. aureus, IVIG use was still unusual (59/868 [6.8%]) and lacked benefit (P = .63). Median LOS was similar between IVIG and non-IVIG groups (26 [13-49] vs 26 [11-43]; P = .84). Positive predictive values for identifying true NF and debridement among IVIG cases using our algorithms were 97% and 89%, respectively, based on records review at 4 hospitals.

Conclusions

Adjunctive IVIG was administered infrequently in NF with shock and had no apparent impact on mortality or hospital LOS beyond that achieved with debridement and antibiotics.

Non-inferiority tests for anti-infective drugs using control group quantiles

BACKGROUND/AIMS

In testing for non-inferiority of anti-infective drugs, the primary endpoint is often the difference in the proportion of failures between the test and control group at a landmark time. The landmark time is chosen to approximately correspond to the qth historic quantile of the control group, and the non-inferiority margin is selected to be reasonable for the target level q. For designing these studies, a troubling issue is that the landmark time must be pre-specified, but there is no guarantee that the proportion of control failures at the landmark time will be close to the target level q. If the landmark time is far from the target control quantile, then the pre-specified non-inferiority margin may not longer be reasonable. Exact variable margin tests have been developed by Röhmel and Kieser to address this problem, but these tests can have poor power if the observed control failure rate at the landmark time is far from its historic value.

METHODS

We develop a new variable margin non-inferiority test where we continue sampling until a pre-specified proportion of failures, q, have occurred in the control group, where q is the target quantile level. The test does not require any assumptions on the failure time distributions, and hence, no knowledge of the true [Formula: see text] control quantile for the study is needed.

RESULTS

Our new test is exact and has power comparable to (or greater than) its competitors when the true control quantile from the study equals (or differs moderately from) its historic value. Our nivm R package performs the test and gives confidence intervals on the difference in failure rates at the true target control quantile. The tests can be applied to time to cure or other numeric variables as well.

CONCLUSION

A substantial proportion of new anti-infective drugs being developed use non-inferiority tests in their development, and typically, a pre-specified landmark time and its associated difference margin are set at the design stage to match a specific target control quantile. If through changing standard of care or selection of a different population the target quantile for the control group changes from its historic value, then the appropriateness of the pre-specified margin at the landmark time may be questionable. Our proposed test avoids this problem by sampling until a pre-specified proportion of the controls have failed.

Random effects and latent processes approaches for analyzing binary longitudinal data with missingness: a comparison of approaches using opiate clinical trial data

The analysis of longitudinal data with non-ignorable missingness remains an active area in biostatistics research. This article discusses various random effects and latent process models which have been proposed for analyzing longitudinal binary data subject to both non-ignorable intermittent missing data and dropout. These models account for non-ignorable missingness by introducing random effects or a latent process which is shared between the response model and the model for the missing-data mechanism. We discuss various random effects and latent processes approaches and compare these approaches with analyses from an opiate clinical trial data set, which had high proportion of intermittent missingness and dropout. We also compare these random effect and latent process approaches with other methods for accounting for non-ignorable missingness using this data set.

Recurrent event data analysis with intermittently observed time-varying covariates

Although recurrent event data analysis is a rapidly evolving area of research, rigorous studies on estimation of the effects of intermittently observed time-varying covariates on the risk of recurrent events have been lacking. Existing methods for analyzing recurrent event data usually require that the covariate processes are observed throughout the entire follow-up period. However, covariates are often observed periodically rather than continuously. We propose a novel semiparametric estimator for the regression parameters in the popular proportional rate model. The proposed estimator is based on an estimated score function where we kernel smooth the mean covariate process. We show that the proposed semiparametric estimator is asymptotically unbiased, normally distributed, and derives the asymptotic variance. Simulation studies are conducted to compare the performance of the proposed estimator and the simple methods carrying forward the last covariates. The different methods are applied to an observational study designed to assess the effect of group A streptococcus on pharyngitis among school children in India. Copyright © 2016 John Wiley & Sons, Ltd.

Implementation of an Ebola virus disease vaccine clinical trial during the Ebola epidemic in Liberia: Design, procedures, and challenges

The index case of the Ebola virus disease epidemic in West Africa is believed to have originated in Guinea. By June 2014, Guinea, Liberia, and Sierra Leone were in the midst of a full-blown and complex global health emergency. The devastating effects of this Ebola epidemic in West Africa put the global health response in acute focus for urgent international interventions. Accordingly, in October 2014, a World Health Organization high-level meeting endorsed the concept of a phase 2/3 clinical trial in Liberia to study Ebola vaccines. As a follow-up to the global response, in November 2014, the Government of Liberia and the US Government signed an agreement to form a research partnership to investigate Ebola and to assess intervention strategies for treating, controlling, and preventing the disease in Liberia. This agreement led to the establishment of the Joint Liberia-US Partnership for Research on Ebola Virus in Liberia as the beginning of a long-term collaborative partnership in clinical research between the two countries. In this article, we discuss the methodology and related challenges associated with the implementation of the Ebola vaccines clinical trial, based on a double-blinded randomized controlled trial, in Liberia.

Vaccine design via nonnegative lasso-based variable selection

There are many different strains of malaria parasites, each represented by a unique sequence of amino acids. A desirable vaccine would match the amino acid sequence of the parasite antigen. Because of the three-dimensional structure of protein, not all sites in the amino acid sequence participate in the binding between the vaccine-induced antibody and the parasite antigen. Nor do all sites have equal importance. In this work, we apply a nonnegative lasso-based variable selection to identify the 'important' amino acid sites and evaluate their relative importance. We then define a metric, the functional coverage, to measure the 'effective' matching in the amino acid sequence between the vaccine and the parasite. With the variable selection procedure, development of a vaccine needs only to target the important sites, and the potential effectiveness of a vaccine candidate is reflected by the functional coverage. Published 2015. This article is a U.S. Government work and is in the public domain in the USA.

Estimation of mean response via effective balancing score

We introduce effective balancing scores for estimation of the mean response under a missing at random mechanism. Unlike conventional balancing scores, the effective balancing scores are constructed via dimension reduction free of model specification. Three types of effective balancing scores are introduced: those that carry the covariate information about the missingness, the response, or both. They lead to consistent estimation with little or no loss in efficiency. Compared to existing estimators, the effective balancing score based estimator relieves the burden of model specification and is the most robust. It is a near-automatic procedure which is most appealing when high dimensional covariates are involved. We investigate both the asymptotic and the numerical properties, and demonstrate the proposed method in a study on Human Immunodeficiency Virus disease.

A live attenuated influenza A(H5N1) vaccine induces long-term immunity in the absence of a primary antibody response

BACKGROUND

Highly pathogenic avian influenza A(H5N1) causes severe infections in humans. We generated 2 influenza A(H5N1) live attenuated influenza vaccines for pandemic use (pLAIVs), but they failed to elicit a primary immune response. Our objective was to determine whether the vaccines primed or established long-lasting immunity that could be detected by administration of inactivated subvirion influenza A(H5N1) vaccine (ISIV).

METHODS

The following groups were invited to participate in the study: persons who previously received influenza A(H5N1) pLAIV; persons who previously received an irrelevant influenza A(H7N3) pLAIV; and community members who were naive to influenza A(H5N1) and LAIV. LAIV-experienced subjects received a single 45-μg dose of influenza A(H5N1) ISIV. Influenza A(H5N1)- and LAIV-naive subjects received either 1 or 2 doses of ISIV.

RESULTS

In subjects who had previously received antigenically matched influenza A(H5N1) pLAIV followed by 1 dose of ISIV compared with those who were naive to influenza A(H5N1) and LAIV and received 2 doses of ISIV, we observed an increased frequency of antibody response (82% vs 50%, by the hemagglutination inhibition assay) and a significantly higher antibody titer (112 vs 76; P = .04). The affinity of antibody and breadth of cross-clade neutralization was also enhanced in influenza A(H5N1) pLAIV-primed subjects.

CONCLUSIONS

ISIV administration unmasked long-lasting immunity in influenza A(H5N1) pLAIV recipients, with a rapid, high-titer, high-quality antibody response that was broadly cross-reactive across several influenza A(H5N1) clades.

CLINICAL TRIALS REGISTRATION

NCT01109329.

Anthrax vaccine-induced antibodies provide cross-species prediction of survival to aerosol challenge

Because clinical trials to assess the efficacy of vaccines against anthrax are not ethical or feasible, licensure for new anthrax vaccines will likely involve the Food and Drug Administration's "Animal Rule," a set of regulations that allow approval of products based on efficacy data only in animals combined with immunogenicity and safety data in animals and humans. U.S. government-sponsored animal studies have shown anthrax vaccine efficacy in a variety of settings. We examined data from 21 of those studies to determine whether an immunological bridge based on lethal toxin neutralization activity assay (TNA) can predict survival against an inhalation anthrax challenge within and across species and genera. The 21 studies were classified into 11 different settings, each of which had the same animal species, vaccine type and formulation, vaccination schedule, time of TNA measurement, and challenge time. Logistic regression models determined the contribution of vaccine dilution dose and TNA on prediction of survival. For most settings, logistic models using only TNA explained more than 75% of the survival effect of the models with dose additionally included. Cross-species survival predictions using TNA were compared to the actual survival and shown to have good agreement (Cohen's κ ranged from 0.55 to 0.78). In one study design, cynomolgus macaque data predicted 78.6% survival in rhesus macaques (actual survival, 83.0%) and 72.6% in rabbits (actual survival, 64.6%). These data add support for the use of TNA as an immunological bridge between species to extrapolate data in animals to predict anthrax vaccine effectiveness in humans.

A valid formulation of the analysis of noninferiority trials under random effects meta-analysis

A noninferiority (NI) trial is sometimes employed to show efficacy of a new treatment when it is unethical to randomize current patients to placebo because of the established efficacy of a standard treatment. Under this framework, if the NI trial determines that the treatment advantage of the standard to the new drug (i.e. S-N) is less than the historic advantage of the standard to placebo (S-P), then the efficacy of the new treatment (N-P) is established indirectly. We explicitly combine information from the NI trial with estimates from a random effects model, allowing study-to-study variability in k historic trials. Existing methods under random effects, such as the synthesis method, fail to account for the variability of the true standard versus placebo effect in the NI trial. Our method effectively uses a prediction interval for the missing standard versus placebo effect rather than a confidence interval of the mean. The consequences are to increase the variance of the synthesis method by incorporating a prediction variance term and to approximate the null distribution of the new statistic with a t with k-1 degrees of freedom instead of the standard normal. Thus, it is harder to conclude NI of the new to (predicted) placebo, compared with traditional methods, especially when k is small or when between study variability is large. When the between study variances are nonzero, we demonstrate substantial Type I error rate inflation with conventional approaches; simulations suggest that the new procedure has only modest inflation, and it is very conservative when between study variances are zero. An example is used to illustrate practical issues.

A maximum pseudo-profile likelihood estimator for the Cox model under length-biased sampling

This paper considers semiparametric estimation of the Cox proportional hazards model for right-censored and length-biased data arising from prevalent sampling. To exploit the special structure of length-biased sampling, we propose a maximum pseudo-profile likelihood estimator, which can handle time-dependent covariates and is consistent under covariate-dependent censoring. Simulation studies show that the proposed estimator is more efficient than its competitors. A data analysis illustrates the methods and theory.

Development of functional and molecular correlates of vaccine-induced protection for a model intracellular pathogen, F. tularensis LVS

In contrast with common human infections for which vaccine efficacy can be evaluated directly in field studies, alternative strategies are needed to evaluate efficacy for slowly developing or sporadic diseases like tularemia. For diseases such as these caused by intracellular bacteria, serological measures of antibodies are generally not predictive. Here, we used vaccines varying in efficacy to explore development of clinically useful correlates of protection for intracellular bacteria, using Francisella tularensis as an experimental model. F. tularensis is an intracellular bacterium classified as Category A bioterrorism agent which causes tularemia. The primary vaccine candidate in the U.S., called Live Vaccine Strain (LVS), has been the subject of ongoing clinical studies; however, safety and efficacy are not well established, and LVS is not licensed by the U.S. FDA. Using a mouse model, we compared the in vivo efficacy of a panel of qualitatively different Francisella vaccine candidates, the in vitro functional activity of immune lymphocytes derived from vaccinated mice, and relative gene expression in immune lymphocytes. Integrated analyses showed that the hierarchy of protection in vivo engendered by qualitatively different vaccines was reflected by the degree of lymphocytes' in vitro activity in controlling the intramacrophage growth of Francisella. Thus, this assay may be a functional correlate. Further, the strength of protection was significantly related to the degree of up-regulation of expression of a panel of genes in cells recovered from the assay. These included IFN-γ, IL-6, IL-12Rβ2, T-bet, SOCS-1, and IL-18bp. Taken together, the results indicate that an in vitro assay that detects control of bacterial growth, and/or a selected panel of mediators, may ultimately be developed to predict the outcome of vaccine efficacy and to complement clinical trials. The overall approach may be applicable to intracellular pathogens in general.

Diseases such as tuberculosis (caused by Mycobacterium tuberculosis) or tularemia (caused by Francisella tularensis) result from infections by microbes that live within cells of a person's body. New vaccines are being developed against such intracellular pathogens, but some will be difficult to test, because disease takes a long time to develop (e.g., tuberculosis) or because outbreaks are unpredictable (e.g., tularemia). Usually such infections are controlled by activities of T cells. However, there are no accepted measures of T cell function that reliably predict vaccine-induced protection. We studied two new ways to do so. We used a group of vaccine candidates against tularemia that stimulated good, fair, or poor protection of mice against Francisella challenge. We then measured whether Francisella–immune cells from vaccinated mice controlled the growth of bacteria inside cells, and/or whether the expression of immune genes in Francisella–immune cells was increased. We found that the degree of protection was matched by the degree of the cells' function in controlling intramacrophage bacterial growth. Further, the degree was predicted by relative amounts of gene expression for several immune mediators. Thus the two new options explored here may help predict protection, without waiting for the onset of disease.

Empirical Likelihood-Based Estimation of the Treatment Effect in a Pretest-Posttest Study

The pretest-posttest study design is commonly used in medical and social science research to assess the effect of a treatment or an intervention. Recently, interest has been rising in developing inference procedures that improve efficiency while relaxing assumptions used in the pretest-posttest data analysis, especially when the posttest measurement might be missing. In this article we propose a semiparametric estimation procedure based on empirical likelihood (EL) that incorporates the common baseline covariate information to improve efficiency. The proposed method also yields an asymptotically unbiased estimate of the response distribution. Thus functions of the response distribution, such as the median, can be estimated straightforwardly, and the EL method can provide a more appealing estimate of the treatment effect for skewed data. We show that, compared with existing methods, the proposed EL estimator has appealing theoretical properties, especially when the working model for the underlying relationship between the pretest and posttest measurements is misspecified. A series of simulation studies demonstrates that the EL-based estimator outperforms its competitors when the working model is misspecified and the data are missing at random. We illustrate the methods by analyzing data from an AIDS clinical trial (ACTG 175).

Dimension reduced kernel estimation for distribution function with incomplete data

This work focuses on the estimation of distribution functions with incomplete data, where the variable of interest Y has ignorable missingness but the covariate X is always observed. When X is high dimensional, parametric approaches to incorporate X - information is encumbered by the risk of model misspecification and nonparametric approaches by the curse of dimensionality. We propose a semiparametric approach, which is developed under a nonparametric kernel regression framework, but with a parametric working index to condense the high dimensional X - information for reduced dimension. This kernel dimension reduction estimator has double robustness to model misspecification and is most efficient if the working index adequately conveys the X - information about the distribution of Y. Numerical studies indicate better performance of the semiparametric estimator over its parametric and nonparametric counterparts. We apply the kernel dimension reduction estimation to an HIV study for the effect of antiretroviral therapy on HIV virologic suppression.

Interferon-alpha produces significant decreases in HIV load

A randomized, controlled clinical trial was started in the pre-HAART era to compare the efficacy of zidovudine (AZT) and interferon-alpha (IFN-alpha) either alone or in combination to reduce HIV viremia, maintain CD4(+) cell count, and decrease time to AIDS progression and death. The purpose of the current study was to compare the effects of AZT and IFN on HIV load using modern technology. One hundred and eighty patients with CD4(+) counts above 500 cells/mm(3) were randomized to receive AZT alone, IFN-alpha alone, or AZT and IFN-alpha in combination. CD4(+) cell count and HIV load at baseline and at the last follow-up visit were compared, and time to AIDS or death was calculated by treatment group. At a mean follow-up of 45 weeks, the mean change in log HIV RNA was -0.06 for the AZT alone group, -0.47 for the AZT plus IFN-alpha group (P = 0.01 versus AZT group), and -0.35 for the IFN-alpha alone group (P = 0.02 versus AZT group). There was no significant difference among groups in change in total CD4(+) count or in time to AIDS or death. Since treatment with IFN-alpha produces significant decreases in HIV load, additional studies of IFN-alpha as part of a combination regimen are warranted.

Lack of effect from a previous single dose of nevirapine on virologic and immunologic responses after 6 months of antiretroviral regimens containing either efavirenz or lopinavir-ritonavir

STUDY OBJECTIVE

To evaluate the effect of a previous single dose of nevirapine given to prevent mother-to-child transmission of human immunodeficiency virus (HIV) on virologic and immunologic measures after months of an antiretroviral regimen containing either efavirenz or lopinavir-ritonavir.

DESIGN

Retrospective subgroup analysis of data from the Phidisa II trial.

SETTING

Six South African research clinics. Patients. A total of 394 women with HIV who completed 6 months of combination antiretroviral regimen containing either efavirenz or lopinavirritonavir as part of the Phidisa II trial.

MEASUREMENTS AND MAIN RESULTS

During the screening process for the Phidisa II study, 478 women were asked about previous nevirapine use: 392 women (82%) were nevirapine naïve, and 86 (18%) had received nevirapine. During the study, patients received either an efavirenz-based or lopinavir-ritonavir- based antiretroviral regimen. After 6 months of treatment, virologic (HIV RNA levels) and immunologic (CD4(+) cell count) responses were measured. These data were compared between women with or without previous nevirapine exposure, and between women who received efavirenz versus lopinavirritonavir. After 6 months of treatment, 394 women (324 nevirapine naïve, 70 exposed to nevirapine) had follow-up HIV RNA results. Two hundred twenty-seven (70.1%) of the nevirapine-naïve patients and 48 (68.6%) of the nevirapine-exposed patients achieved HIV RNA levels lower than 400 copies/ml (p=0.89), with CD4(+) cell count increases of 115.5 and 120.4 cells/mm(3), respectively (p=0. 7). Among the nevirapine-exposed women, 27 (75%) of 36 efavirenz-treated and 21 (61.8%) of 34 lopinavir-ritonavir-treated patients had HIV RNA levels lower than 400 copies/ml at months (p=0.31).

CONCLUSION

In this retrospective analysis of a small cohort, previous exposure to a single dose of nevirapine did not affect virologic outcomes after 6 months of either an efavirenz-based or lopinavir-ritonavir-based antiretroviral regimen. As efavirenz is one of the first-line combination antiretroviral therapies administered in Africa, it remains an option for women who received single-dose nevirapine.

A model checking method for the proportional hazards model with recurrent gap time data

Recurrent events are the natural outcome in many medical and epidemiology studies. To assess covariate effects on the gaps between consecutive recurrent events, the Cox proportional hazards model is frequently employed in data analysis. The validity of statistical inference, however, depends on the appropriateness of the Cox model. In this paper, we propose a class of graphical techniques and formal tests for checking the Cox model with recurrent gap time data. The building block of our model checking method is an averaged martingale-like process, based on which a class of multiparameter stochastic processes is proposed. This maneuver is very general and can be used to assess different aspects of model fit. Numerical simulations are conducted to examine finite-sample performance, and the proposed model checking techniques are illustrated with data from the Danish Psychiatric Central Register.

Semiparametric dimension reduction estimation for mean response with missing data

Model misspecification can be a concern for high-dimensional data. Nonparametric regression obviates model specification but is impeded by the curse of dimensionality. This paper focuses on the estimation of the marginal mean response when there is missingness in the response and multiple covariates are available. We propose estimating the mean response through nonparametric functional estimation, where the dimension is reduced by a parametric working index. The proposed semiparametric estimator is robust to model misspecification: it is consistent for any working index if the missing mechanism of the response is known or correctly specified up to unknown parameters; even with misspecification in the missing mechanism, it is consistent so long as the working index can recover E(Y  |   X), the conditional mean response given the covariates. In addition, when the missing mechanism is correctly specified, the semiparametric estimator attains the optimal efficiency if E(Y  |   X) is recoverable through the working index. Robustness and efficiency of the proposed estimator is further investigated by simulations. We apply the proposed method to a clinical trial for HIV.

Immunologic and virologic events in early HIV infection predict subsequent rate of progression

BACKGROUND

Variability in human immunodeficiency virus (HIV) disease progression cannot be fully predicted by CD4(+) T cell counts or viral load (VL). Because central memory T (T(CM)) cells play a critical role in the pathogenesis of simian immunodeficiency virus disease, we hypothesized that quantifying these cells in early HIV infection could provide prognostic information.

METHODS

We measured expression of CD45RO, chemokine (C-C motif) receptor (CCR) 5, CCR7, CD27, and CD28 to enumerate naive and memory subsets in samples from recently infected individuals. We also quantified proliferation, CD127 expression, and cell-associated VL. Disease progression was compared across subgroups defined by these measurements, using Kaplan-Meier survival curves and multivariate Cox proportional hazards regression.

RESULTS

Four hundred sixty-six subjects contributed 101 events. The proportion or absolute count of T(CM) cells did not correlate with disease progression, defined as the time to AIDS or death. However, significant associations were observed for proliferation within CD4(+) or CD8(+) T cells, loss of naive or CD127(+) memory CD8(+) T cells, and CD4(+) T cell-associated VL.

CONCLUSIONS

Our results demonstrate that the extent of the immunopathogenesis established early in HIV infection predicts the course of future disease. Because antiretroviral drug treatment reverses such defects in part, our study provides mechanistic clues to why early use of antiretrovirals may prove beneficial.

Neutralizing antibody titers conferring protection to macaques from a simian/human immunodeficiency virus challenge using the TZM-bl assay

We previously reported that passive transfer of polyclonal neutralizing antibodies (NAbs) sufficient to generate a titer of 1:38 in the plasma would confer sterilizing protection to 99% of macaques challenged intravenously with 75 TCID(50) of SHIV(DH12). Neutralizing activity in that study was measured in an MT4 cell assay in which infection was completely blocked (EC(100)). In the current study, the TZM-bl system was used to measure EC(50) neutralizing titers in several of the same macaque plasma samples and the relationship between these titers and in vivo protection was determined. The antiviral EC(50) NAb titers measured in individual plasma samples were higher than those previously obtained in the MT4 system. Furthermore, the geometric mean EC(50) NAb titers against pseudotyped SHIV(DH12) were 33-fold greater than the EC(100) titers measured in the MT4 cell assay against the replication-competent SHIV(DH12) inoculated into animals. An augmented probit regression model was used to generate curves relating TZM-bl EC(50) NAb titers and protection from a virus challenge; estimated titers conferring various levels of protection were then determined. In TZM-bl assays using pseudotyped SHIV(DH12), representative percent in vivo protection/estimated EC(50) titers were 99%/1:4467, 90%/1:1175, 80%/1:676, 50%/1:234, and 33%/1:141. Because it is likely that contributions from other arms of the immune system will contribute to vaccine-induced control, the range of EC(50) NAb titers we have derived may be more informative for evaluating the protective value of NAb activity from TZM-bl assays.

The effect of continuous versus pericycle antiretroviral therapy on IL-2 responsiveness

BACKGROUND

Intermittent administration of interleukin-2 (IL-2) to human immunodeficiency virus (HIV)- infected patients on antiretroviral therapy (ART) is capable of inducing significant increases in CD4 T cell counts as a result of increased T cell survival and decreased cell turnover. However, its role in the setting of ART interruptions (STI) is less well characterized. We sought to compare the effect of continuous (C) versus intermittent (P) ART on CD4 responses in patients undergoing IL-2 therapy.

METHODS

CD4 cell responses were compared in 25 patients who underwent IL-2 therapy during periods of continuous ART (n = 90 cycles) as well as during STI (n = 45 cycles). During STI, patients resumed ART for only 10 days surrounding each IL-2 cycle.

RESULTS

C cycles resulted in a significantly greater CD4 gain than P cycles (Delta156 cells/microL, 95% CI = 68-243). In multivariate analyses, baseline CD4/CD25 expression and treatment arm remained strong predictors of CD4 gain while CD8/CD38+, CD8/DR+, and CD4 Ki67+ phenotype were not predictive.

CONCLUSIONS

Continuous ART was associated with a statistically significantly greater CD4 cell response to IL-2 therapy than was intermittent ART. These observations may have important implications for the appropriate integration of IL-2 therapy into STI strategies.

Accounting for variability in sample size estimation with applications to nonadherence and estimation of variance and effect size

We consider sample size calculations for testing differences in means between two samples and allowing for different variances in the two groups. Typically, the power functions depend on the sample size and a set of parameters assumed known, and the sample size needed to obtain a prespecified power is calculated. Here, we account for two sources of variability: we allow the sample size in the power function to be a stochastic variable, and we consider estimating the parameters from preliminary data. An example of the first source of variability is nonadherence (noncompliance). We assume that the proportion of subjects who will adhere to their treatment regimen is not known before the study, but that the proportion is a stochastic variable with a known distribution. Under this assumption, we develop simple closed form sample size calculations based on asymptotic normality. The second source of variability is in parameter estimates that are estimated from prior data. For example, we account for variability in estimating the variance of the normal response from existing data which are assumed to have the same variance as the study for which we are calculating the sample size. We show that we can account for the variability of the variance estimate by simply using a slightly larger nominal power in the usual sample size calculation, which we call the calibrated power. We show that the calculation of the calibrated power depends only on the sample size of the existing data, and we give a table of calibrated power by sample size. Further, we consider the calculation of the sample size in the rarer situation where we account for the variability in estimating the standardized effect size from some existing data. This latter situation, as well as several of the previous ones, is motivated by sample size calculations for a Phase II trial of a malaria vaccine candidate.

The Accelerated Biased Coin Up-and-Down Design in Phase I Trials

The biased coin up-and-down design (BCD) is used to allocate doses in phase I clinical trials. The BCD requires that the treatment response or the toxicity evaluation is observed quickly. In trials with a long treatment evaluation, the BCD will lead to long trial duration because a new patient cannot be enrolled until the preceding patient has completed the evaluation period. We propose a simple method to modify the BCD that will reduce the trial duration without significantly affecting the estimate of the target dose. The idea is to allocate a dose to each patient as he or she arrives based on the toxicity information of the last completed subject. This allows multiple patients to be concurrently under evaluation. A simulation study shows that this modification does not adversely affect the precision of the recommended dose, as estimated by isotonic regression, but it does significantly reduce the total time to complete the study.

Genetic polymorphisms of eosinophil-derived neurotoxin and eosinophil cationic protein in tropical pulmonary eosinophilia

Because eosinophil-derived neurotoxin (EDN) and eosinophil cationic protein (ECP) are critical in the pathogenesis of tropical pulmonary eosinophilia (TPE), we analyzed genetic polymorphisms of both in 181 individuals from southern India with varying clinical manifestations of Wuchereria bancrofti infection (including 26 with TPE). Using haplotype frequency analysis, we identified four known (of nine) and two novel haplotypes for EDN (1, 2, 7, 8, 10, and 11). For ECP, five (of seven known) haplotypes (1-5) were identified. Although we found no significant association between frequencies of EDN and ECP polymorphisms and TPE development, we observed a unique pattern of EDN and ECP polymorphism distribution among this population. Genotype TT at locus 1088 of ECP in one TPE patient was not observed in any other clinical group. Although the EDN and ECP polymorphisms appear unlikely to be associated with the development of TPE, further analyses will be more definitive.

Effect of chronic cytokine therapy on clonal dynamics in nonhuman primates

Hematopoietic cytokines such as filgrastim are used extensively to stimulate granulocyte production or to mobilize hematopoietic progenitors into the circulation; however, their effect on more primitive hematopoietic progenitor and stem cells in vivo is unknown, particularly in large animals or humans. In particular, there is concern that chronic therapy with cytokines could result in stem cell exhaustion or clonal dominance; however, direct assessment of the dynamics of individual stem and progenitor cell clones in vivo has not been previously reported. A number of models can be proposed regarding the mechanisms by which the marrow responds to cytokine stimulation, including recruitment of previously quiescent clones, stimulation of proliferation of already active clones, or prevention of apoptosis of more mature progenitors from all clones. Using retroviral marking and comprehensive insertion site tracking of individual stem and progenitor cell clones in 2 rhesus macaques, we analyzed the effect of chronic administration of granulocyte colony-stimulating factor (G-CSF), or a combination of G-CSF plus stem cell factor (SCF). The overall number of contributing clones remained constant, and the relative output from each clone did not change significantly during or following cytokine treatments. These results suggest that individual transduced stem or progenitor cells can contribute to hematopoiesis for prolonged periods, with no evidence for an effect of G-CSF or G-CSF/SCF on the number, the lifespan, or the relative activity of individual stem or progenitor cell clones. These relevant large animal studies are reassuring regarding clinical applications of cytokines and provide new insights into their mechanisms of action.

Induction and maintenance therapy with intermittent interleukin-2 in HIV-1 infection

Studies establishing that intermittent subcutaneous interleukin-2 (IL-2) therapy can lead to substantial CD4 cell increases in many HIV-infected patients have generally been of limited duration. We studied 77 patients participating in active longitudinal studies of subcutaneous IL-2 therapy at our center in order to determine the long-term feasibility of this approach. Following initial induction, patients in each trial were eligible to receive intermittent 5-day cycles of subcutaneous IL-2 treatment at individualized doses and frequencies capable of maintaining CD4 counts at postinduction levels. The mean duration of study participation to date is 5.9 years (range, 1.0-9.3 years). Mean baseline CD4 cell count and CD4 percent values of 0.521 × 109/L (521 cells/μL) and 27% have risen to 1.005 × 109/L (1005 cells/μL) and 38%, respectively, at 90 months. The mean number of subcutaneous IL-2 cycles required to achieve and maintain these increases was 10 cycles (range, 3-29 cycles), and the current mean interval of cycling required to maintain these elevations is 39 months (median, 35 months; range, 2-91 months). We conclude that subcutaneous IL-2 therapy is capable of maintaining CD4 cell increases for an extended period using a remarkably low frequency of intermittent cycling. These observations may contribute to patients' acceptance of subcutaneous IL-2 as a favorable long-term treatment strategy. (Blood. 2004;103:3282-3286)

Comparing HLA antigen frequencies between two groups of patients

For diseases that involve the immune system, the alleles of the human leukocyte antigen (HLA) complex can play a major role. For example, if responsiveness to therapy is immunologically mediated, one would think that responders and non-responders might tend to have different HLA alleles. However, comparing the frequencies between the two groups of patients at each allele can introduce a substantial multiple comparisons problem as the number of alleles is large. This paper proposes an efficient two-stage procedure for identifying alleles that may mediate response. In the first-stage, the distribution of all alleles for the patients are compared to a reference population and a few alleles are selected. These candidate alleles are then compared between the two groups of patients using a modest Bonferroni correction. The two-stage procedure strongly controls the type I error rate as the first-stage selection is statistically independent of the second-stage tests. We analyse a cohort of patients with bone marrow failure who are classified as responders or non-responders to immunosuppressive therapy. Published in 2003 by John Wiley & Sons, Ltd.

Regression analysis based on pairwise ordering of patients' clinical histories

When a medical treatment influences a variety of outcomes, describing the global effect of treatment can be difficult. Traditional approaches specify how treatment affects each separate outcome. This can be done with separate models for each outcome, or by using a combined multivariate model. Describing the overall effect of a treatment thus requires combining these separate effects in some fashion and can be difficult to explain. In this paper, I specify a regression model for use with multiple outcomes where the outcome histories for each pair of patients are ranked. Pairs of patients with different lengths of follow-up are evaluated solely over the common follow-up interval. The logit of the probability that the outcome for patient i is better than that of patient j is assumed to depend on a linear function of the difference of the covariate vectors (for example, treatment indicators) for persons i and j. Thus covariates directly affect the entire clinical history, rather than directly affecting specific outcomes that comprise the history. The idea is that ranking outcomes is more relevant and interpretable than statistically combining separate effects. An estimating equations approach for estimation is described and an example of a clinical trial involving patients with heart failure is provided.

A latent autoregressive model for longitudinal binary data subject to informative missingness

Longitudinal clinical trials often collect long sequences of binary data. Our application is a recent clinical trial in opiate addicts that examined the effect of a new treatment on repeated binary urine tests to assess opiate use over an extended follow-up. The dataset had two sources of missingness: dropout and intermittent missing observations. The primary endpoint of the study was comparing the marginal probability of a positive urine test over follow-up across treatment arms. We present a latent autoregressive model for longitudinal binary data subject to informative missingness. In this model, a Gaussian autoregressive process is shared between the binary response and missing-data processes, thereby inducing informative missingness. Our approach extends the work of others who have developed models that link the various processes through a shared random effect but do not allow for autocorrelation. We discuss parameter estimation using Monte Carlo EM and demonstrate through simulations that incorporating within-subject autocorrelation through a latent autoregressive process can be very important when longitudinal binary data is subject to informative missingness. We illustrate our new methodology using the opiate clinical trial data.

Parametric and semiparametric approaches to testing for seasonal trend in serial count data

We present two tests for seasonal trend in monthly incidence data. The first approach uses a penalized likelihood to choose the number of harmonic terms to include in a parametric harmonic model (which includes time trends and autogression as well as seasonal harmonic terms) and then tests for seasonality using a parametric bootstrap test. The second approach uses a semiparametric regression model to test for seasonal trend. In the semiparametric model, the seasonal pattern is modeled nonparametrically, parametric terms are included for autoregressive effects and a linear time trend, and a parametric bootstrap test is used to test for seasonality. For both procedures, a null distribution is generated under a null Poisson model with time trends and autoregression parameters. We apply the methods to skin melanoma incidence rates collected by the surveillance, epidemiology, and end results (SEER) program of the National Cancer Institute, and perform simulation studies to evaluate the type I error rate and power for the two procedures. These simulations suggest that both procedures are alpha-level procedures. In addition, the harmonic model/bootstrap test had similar or larger power than the semiparametric model/bootstrap test for a wide range of alternatives, and the harmonic model/bootstrap test is much easier to implement. Thus, we recommend the harmonic model/bootstrap test for the analysis of seasonal incidence data.

Testing for treatment and interaction effects in semi-parametric analysis of covariance

Semi-parametric regression models assume that the effects of covariates on the mean response are additive. We propose a test of additivity when there is one continuous covariate and a group indicator. At p fixed points, the differences of the within-group kernel estimates of the means are calculated, and the likelihood ratio test that the p differences have a constant mean is formed. The kernel bandwidth and the location of the p fixed points are chosen to give the test good power. Performance of the proposed test is compared with parametric and non-parametric tests of additivity. Published in 2001 by John Wiley & Sons, Ltd.