Cox model inference for relative hazard and pure risk from stratified weight-calibrated case-cohort data

The case-cohort design obtains complete covariate data only on cases and on a random sample (the subcohort) of the entire cohort. Subsequent publications described the use of stratification and weight calibration to increase efficiency of estimates of Cox model log-relative hazards, and there has been some work estimating pure risk. Yet there are few examples of these options in the medical literature, and we could not find programs currently online to analyze these various options. We therefore present a unified approach and R software to facilitate such analyses. We used influence functions adapted to the various design and analysis options together with variance calculations that take the two-phase sampling into account. This work clarifies when the widely used "robust" variance estimate of Barlow (Biometrics 50:1064-1072, 1994) is appropriate. The corresponding R software, CaseCohortCoxSurvival, facilitates analysis with and without stratification and/or weight calibration, for subcohort sampling with or without replacement. We also allow for phase-two data to be missing at random for stratified designs. We provide inference not only for log-relative hazards in the Cox model, but also for cumulative baseline hazards and covariate-specific pure risks. We hope these calculations and software will promote wider use of more efficient and principled design and analysis options for case-cohort studies.

Predicting absolute risk for a person with missing risk factors

We compared methods to project absolute risk, the probability of experiencing the outcome of interest in a given projection interval accommodating competing risks, for a person from the target population with missing predictors. Without missing data, a perfectly calibrated model gives unbiased absolute risk estimates in a new target population, even if the predictor distribution differs from the training data. However, if predictors are missing in target population members, a reference dataset with complete data is needed to impute them and to estimate absolute risk, conditional only on the observed predictors. If the predictor distributions of the reference data and the target population differ, this approach yields biased estimates. We compared the bias and mean squared error of absolute risk predictions for seven methods that assume predictors are missing at random (MAR). Some methods imputed individual missing predictors, others imputed linear predictor combinations (risk scores). Simulations were based on real breast cancer predictor distributions and outcome data. We also analyzed a real breast cancer dataset. The largest bias for all methods resulted from different predictor distributions of the reference and target populations. No method was unbiased in this situation. Surprisingly, violating the MAR assumption did not induce severe biases. Most multiple imputation methods performed similarly and were less biased (but more variable) than a method that used a single expected risk score. Our work shows the importance of selecting predictor reference datasets similar to the target population to reduce bias of absolute risk predictions with missing risk factors.

Exposure to polycyclic aromatic hydrocarbons, volatile organic compounds, and tobacco-specific nitrosamines and incidence of esophageal cancer

BACKGROUND

Studying carcinogens in tobacco and nontobacco sources may be key to understanding the pathogenesis and geographic distribution of esophageal cancer.

METHODS

The Golestan Cohort Study has been conducted since 2004 in a region with high rates of esophageal squamous cell carcinoma. For this nested study, the cases comprised of all incident cases by January 1, 2018; controls were matched to the case by age, sex, residence, time in cohort, and tobacco use. We measured urinary concentrations of 33 exposure biomarkers of nicotine, polycyclic aromatic hydrocarbons, volatile organic compounds, and tobacco-specific nitrosamines. We used conditional logistic regression to calculate odds ratios (ORs) and 95% confidence intervals for associations between the 90th vs the 10th percentiles of the biomarker concentrations and incident esophageal squamous cell carcinoma.

RESULTS

Among individuals who did not currently use tobacco (148 cases and 163 controls), 2 acrolein metabolites, 2 acrylonitrile metabolites, 1 propylene oxide metabolite, and one 1,3-butadiene metabolite were significantly associated with incident esophageal squamous cell carcinoma (adjusted odds ratios between 1.8 and 4.3). Among tobacco users (57 cases and 63 controls), metabolites of 2 other volatile organic compounds (styrene and xylene) were associated with esophageal squamous cell carcinoma (OR = 6.2 and 9.0, respectively). In tobacco users, 2 tobacco-specific nitrosamines (NNN and N'-Nitrosoanatabine) were also associated with esophageal squamous cell carcinoma. Suggestive associations were seen with some polycyclic aromatic hydrocarbons (especially 2-hydroxynaphthalene) in nonusers of tobacco products and other tobacco-specific nitrosamines in tobacco users.

CONCLUSION

These novel associations based on individual-level data and samples collected many years before cancer diagnosis, from a population without occupational exposure, have important public health implications.

Cohort profile: evaluation of immune response and household transmission of SARS-CoV-2 in Costa Rica: the RESPIRA study

PURPOSE

The RESPIRA cohort aims to describe the nature, magnitude, time course and efficacy of the immune response to SARS-CoV-2 infection and vaccination, population prevalence, and household transmission of COVID-19.

PARTICIPANTS

From November 2020, we selected age-stratified random samples of COVID-19 cases from Costa Rica confirmed by PCR. For each case, two population-based controls, matched on age, sex and census tract were recruited, supplemented with hospitalised cases and household contacts. Participants were interviewed and blood and saliva collected for antibodies and PCR tests. Participants will be followed for 2 years to assess antibody response and infection incidence.

FINDINGS TO DATE

Recruitment included 3860 individuals: 1150 COVID-19 cases, 1999 population controls and 719 household contacts from 304 index cases. The age and regional distribution of cases was as planned, including four age strata, 30% rural and 70% urban. The control cohort had similar sex, age and regional distribution as the cases according to the study design. Among the 1999 controls recruited, 6.8% reported at enrolment having had COVID-19 and an additional 12.5% had antibodies against SARS-CoV-2. Compliance with visits and specimens has been close to 70% during the first 18 months of follow-up. During the study, national vaccination was implemented and nearly 90% of our cohort participants were vaccinated during follow-up.

FUTURE PLANS

RESPIRA will enable multiple analyses, including population prevalence of infection, clinical, behavioural, immunological and genetic risk factors for SARS-CoV-2 acquisition and severity, and determinants of household transmission. We are conducting retrospective and prospective assessment of antibody levels, their determinants and their protective efficacy after infection and vaccination, the impact of long-COVID and a series of ancillary studies. Follow-up continues with bimonthly saliva collection for PCR testing and biannual blood collection for immune response analyses. Follow-up will be completed in early 2024.

TRIAL REGISTRATION NUMBER

NCT04537338.

Estimating the cumulative incidence of SARS-CoV-2 infection in Costa Rica: modelling seroprevalence data in a population-based cohort

Background

The true incidence of SARS-CoV-2 infection in Costa Rica was likely much higher than officially reported, because infection is often associated with mild symptoms and testing was limited by official guidelines and socio-economic factors.

Methods

Using serology to define natural infection, we developed a statistical model to estimate the true cumulative incidence of SARS-CoV-2 in Costa Rica early in the pandemic. We estimated seroprevalence from 2223 blood samples collected from November 2020 to October 2021 from 1976 population-based controls from the RESPIRA study. Samples were tested for antibodies against SARS-CoV-2 nucleocapsid and the receptor-binding-domain of the spike proteins. Using a generalized linear model, we estimated the ratio of true infections to officially reported cases. Applying these ratios to officially reported totals by age, sex, and geographic area, we estimated the true number of infections in the study area, where 70% of Costa Ricans reside. We adjusted the seroprevalence estimates for antibody decay over time, estimated from 1562 blood samples from 996 PCR-confirmed COVID-19 cases.

Findings

The estimated total proportion infected (ETPI) was 4.0 times higher than the officially reported total proportion infected (OTPI). By December 16th, 2021, the ETPI was 47% [42-52] while the OTPI was 12%. In children and adolescents, the ETPI was 11.0 times higher than the OTPI.

Interpretation

Our findings suggest that nearly half the population had been infected by the end of 2021. By the end of 2022, it is likely that a large majority of the population had been infected.

Funding

This work was sponsored and funded by the National Institute of Allergy and Infectious Diseases through the National Cancer Institute, the Science, Innovation, Technology and Telecommunications Ministry of Costa Rica, and Costa Rican Biomedical Research Agency-Fundacion INCIENSA (grant N/A).

Behavioral factors and SARS-CoV-2 transmission heterogeneity within a household cohort in Costa Rica

INTRODUCTION

Variability in household secondary attack rates and transmission risks factors of SARS-CoV-2 remain poorly understood.

METHODS

We conducted a household transmission study of SARS-CoV-2 in Costa Rica, with SARS-CoV-2 index cases selected from a larger prospective cohort study and their household contacts were enrolled. A total of 719 household contacts of 304 household index cases were enrolled from November 21, 2020, through July 31, 2021. Blood specimens were collected from contacts within 30-60 days of index case diagnosis; and serum was tested for presence of spike and nucleocapsid SARS-CoV-2 IgG antibodies. Evidence of SARS-CoV-2 prior infections among household contacts was defined based on the presence of both spike and nucleocapsid antibodies. We fitted a chain binomial model to the serologic data, to account for exogenous community infection risk and potential multi-generational transmissions within the household.

RESULTS

Overall seroprevalence was 53% (95% confidence interval (CI) 48-58%) among household contacts. The estimated household secondary attack rate is 34% (95% CI 5-75%). Mask wearing by the index case is associated with the household transmission risk reduction by 67% (adjusted odds ratio = 0.33 with 95% CI: 0.09-0.75) and not sharing bedroom with the index case is associated with the risk reduction of household transmission by 78% (adjusted odds ratio = 0.22 with 95% CI 0.10-0.41). The estimated distribution of household secondary attack rates is highly heterogeneous across index cases, with 30% of index cases being the source for 80% of secondary cases.

CONCLUSIONS

Modeling analysis suggests that behavioral factors are important drivers of the observed SARS-CoV-2 transmission heterogeneity within the household.

Circulating vitamin D and breast cancer risk: an international pooling project of 17 cohorts

Laboratory and animal research support a protective role for vitamin D in breast carcinogenesis, but epidemiologic studies have been inconclusive. To examine comprehensively the relationship of circulating 25-hydroxyvitamin D [25(OH)D] to subsequent breast cancer incidence, we harmonized and pooled participant-level data from 10 U.S. and 7 European prospective cohorts. Included were 10,484 invasive breast cancer cases and 12,953 matched controls. Median age (interdecile range) was 57 (42-68) years at blood collection and 63 (49-75) years at breast cancer diagnosis. Prediagnostic circulating 25(OH)D was either newly measured using a widely accepted immunoassay and laboratory or, if previously measured by the cohort, calibrated to this assay to permit using a common metric. Study-specific relative risks (RRs) for season-standardized 25(OH)D concentrations were estimated by conditional logistic regression and combined by random-effects models. Circulating 25(OH)D increased from a median of 22.6 nmol/L in consortium-wide decile 1 to 93.2 nmol/L in decile 10. Breast cancer risk in each decile was not statistically significantly different from risk in decile 5 in models adjusted for breast cancer risk factors, and no trend was apparent (P-trend = 0.64). Compared to women with sufficient 25(OH)D based on Institute of Medicine guidelines (50- < 62.5 nmol/L), RRs were not statistically significantly different at either low concentrations (< 20 nmol/L, 3% of controls) or high concentrations (100- < 125 nmol/L, 3% of controls; ≥ 125 nmol/L, 0.7% of controls). RR per 25 nmol/L increase in 25(OH)D was 0.99 [95% confidence intervaI (CI) 0.95-1.03]. Associations remained null across subgroups, including those defined by body mass index, physical activity, latitude, and season of blood collection. Although none of the associations by tumor characteristics reached statistical significance, suggestive inverse associations were seen for distant and triple negative tumors. Circulating 25(OH)D, comparably measured in 17 international cohorts and season-standardized, was not related to subsequent incidence of invasive breast cancer over a broad range in vitamin D status.

Tools for Contralateral Prophylactic Mastectomy Decision Making

PURPOSE

Women with unilateral breast cancer are increasingly opting for the removal of not only the involved breast, but also for the removal of the opposite uninvolved breast (contralateral prophylactic mastectomy [CPM]), although the risk of contralateral breast cancer (CBC) has decreased in recent years. Models to predict the absolute risk of CBC can help a woman decide whether to undergo CPM. Our objective is to illustrate that a better decision can be made if the patient and doctor also have estimates of the absolute risks of regional and distant recurrences and mortality from non-breast cancer causes.

MATERIALS AND METHODS

We based our analyses on two published models for CBC and published information on the hazards of regional and distant recurrences and non-breast cancer mortality. Assuming that CPM eliminates CBC but has no effect on other events, we calculated how much CPM reduces a woman's CBC risk and total risk from all these events for 10 hypothetical women with various subtypes of breast cancer and risk factors.

RESULTS

The risk of CBC and total risk vary greatly, depending on the breast cancer subtype. In some cases, a decision for or against CPM can be based on CBC risk alone, but in others, additional consideration of total risk may cause a woman to decline CPM.

CONCLUSION

There is a potential to develop more informative tools for deciding on CPM. Realizing this potential will require more and better data to validate existing models of absolute CBC risk and to characterize the hazards of regional and distant recurrences and deaths from non-breast cancer causes for women with various subtypes of breast cancers and risk factors.

Accommodating population differences when validating risk prediction models

Validation of risk prediction models in independent data provides a more rigorous assessment of model performance than internal assessment, for example, done by cross-validation in the data used for model development. However, several differences between the populations that gave rise to the training and the validation data can lead to seemingly poor performance of a risk model. In this paper we formalize the notions of "similarity" or "relatedness" of the training and validation data, and define reproducibility and transportability. We address the impact of different distributions of model predictors and differences in verifying the disease status or outcome on measures of calibration, accuracy and discrimination of a model. When individual level information from both the training and validation data sets is available, we propose and study weighted versions of the validation metrics that adjust for differences in the risk factor distributions and in outcome verification between the training and validation data to provide a more comprehensive assessment of model performance. We provide conditions on the risk model and the populations that gave rise to the training and validation data that ensure a model's reproducibility or transportability, and show how to check these conditions using weighted and unweighted performance measures. We illustrate the method by developing and validating a model that predicts the risk of developing prostate cancer using data from two large prostate cancer screening trials.

Discussion of "A formal causal interpretation of the case-crossover design" by Zach Shahn, Miguel A. Hernan, and James M. Robins

Shahn, Hernan, and Robins give conditions under which estimates from a case-crossover analysis converge to the desired causal relative risk times a bias factor, and they discuss conditions needed to have small bias. To simplify the problem, we discuss only two exposure times and rely on randomized exposure assignments, thereby avoiding the need for potential outcome notation. We identify many, but not all, of the conditions discussed by Shahn et al. in this simple analysis.

Performance of Cervical Screening a Decade Following HPV Vaccination: The Costa Rica Vaccine Trial

BACKGROUND

We investigated the impact of human papillomavirus (HPV) vaccination on the performance of cytology-based and HPV-based screening for detection of cervical precancer among women vaccinated as young adults and reaching screening age.

METHODS

A total of 4632 women aged 25-36 years from the Costa Rica HPV Vaccine Trial were included (2418 HPV-vaccinated as young adults and 2214 unvaccinated). We assessed the performance of cytology- and HPV-based cervical screening modalities in vaccinated and unvaccinated women to detect high-grade cervical precancers diagnosed over 4 years and the absolute risk of cumulative cervical precancers by screening results at entry.

RESULTS

We detected 95 cervical intraepithelial neoplasia grade 3 or worse (52 in unvaccinated and 43 in vaccinated women). HPV16/18/31/33/45 was predominant (69%) among unvaccinated participants, and HPV35/52/58/39/51/56/59/66/68 predominated (65%) among vaccinated participants. Sensitivity and specificity of cervical screening approaches were comparable between women vaccinated as young adults and unvaccinated women. Colposcopy referral rates were lower in the vaccinated group for HPV-based screening modalities, but the positive predictive value was comparable between the 2 groups.

CONCLUSIONS

Among women approaching screening ages, vaccinated as young adults, and with a history of intensive screening, the expected reduction in the positive predictive value of HPV testing, associated with dropping prevalence of HPV-associated lesions, was not observed. This is likely due to the presence of high-grade lesions associated with nonvaccine HPV types, which may be less likely to progress to cancer.

The Oral Microbiome and Lung Cancer Risk: An Analysis of 3 Prospective Cohort Studies

BACKGROUND

Previous studies suggested associations between the oral microbiome and lung cancer, but studies were predominantly cross-sectional and underpowered.

METHODS

Using a case-cohort design, 1306 incident lung cancer cases were identified in the Agricultural Health Study; National Institutes of Health-AARP Diet and Health Study; and Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Referent subcohorts were randomly selected by strata of age, sex, and smoking history. DNA was extracted from oral wash specimens using the DSP DNA Virus Pathogen kit, the 16S rRNA gene V4 region was amplified and sequenced, and bioinformatics were conducted using QIIME 2. Hazard ratios and 95% confidence intervals were calculated using weighted Cox proportional hazards models.

RESULTS

Higher alpha diversity was associated with lower lung cancer risk (Shannon index hazard ratio = 0.90, 95% confidence interval = 0.84 to 0.96). Specific principal component vectors of the microbial communities were also statistically significantly associated with lung cancer risk. After multiple testing adjustment, greater relative abundance of 3 genera and presence of 1 genus were associated with greater lung cancer risk, whereas presence of 3 genera were associated with lower risk. For example, every SD increase in Streptococcus abundance was associated with 1.14 times the risk of lung cancer (95% confidence interval = 1.06 to 1.22). Associations were strongest among squamous cell carcinoma cases and former smokers.

CONCLUSIONS

Multiple oral microbial measures were prospectively associated with lung cancer risk in 3 US cohort studies, with associations varying by smoking history and histologic subtype. The oral microbiome may offer new opportunities for lung cancer prevention.

fast.adonis: a computationally efficient non-parametric multivariate analysis of microbiome data for large-scale studies

Motivation

Nonparametric multivariate analysis has been widely used to identify variables associated with a dissimilarity matrix and to quantify their contribution. For very large studies ( n ≥ 5000 ) and many explanatory variables, existing software packages (e.g. adonis and adonis2 in vegan) are computationally intensive when conducting sequential multivariate analysis with permutations or bootstrapping. Moreover, for subjects from a complex sampling design, we need to adjust for sampling weights to derive an unbiased estimate.

Results

We implemented an R function fast.adonis to overcome these computational challenges in large-scale studies. fast.adonis generates results consistent with adonis/adonis2 but much faster. For complex sampling studies, fast.adonis integrates sampling weights algebraically to mimic the source population; thus, analysis can be completed very fast without requiring a large amount of memory.

Availability and implementation

fast.adonis is implemented using R and is publicly available at https://github.com/jennylsl/fast.adonis.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

Rationale and design of a double-blind randomized non-inferiority clinical trial to evaluate one or two doses of vaccine against human papillomavirus including an epidemiologic survey to estimate vaccine efficacy: The Costa Rica ESCUDDO trial

HPV vaccination of adolescent girls is the most effective measure to prevent cervical cancer. The World Health Organization recommends that adolescent girls receive two doses of vaccine but only a small proportion of girls from regions with the highest disease burden are vaccinated because of cost and logistical considerations. Our Costa Rica HPV Vaccine trial suggested that one dose of the bivalent HPV vaccine provides robust and lasting protection against persistent HPV infections for over a decade. Data from a post-licensure trial of the quadrivalent vaccine in India also suggested that a single dose may be effective in reducing cervical cancer risk. To formally compare one versus two doses of the bivalent and nonavalent HPV vaccines, we implemented a large, randomized, double-blind trial to investigate the non-inferiority of one compared to two vaccine doses in the prevention of new HPV16/18 infections that persist 6 or more months. Bivalent and nonavalent vaccines will be evaluated separately. The trial enrolled and randomized (1:1:1:1 to 1- and 2-dose arms of the bivalent and nonavalent vaccines) 20,330 girls 12 to 16 years old residing in Costa Rica. Trial participants are followed every 6 months for up to 5 years. We also aim to estimate vaccine efficacy by comparing the rates of 6 month persistent infection in unvaccinated women with the rates in the follow-up visits of trial participants. We included one survey of unvaccinated women at the start of the study (N = 4452) and will include another survey concomitant with follow up visits of trial participants at year 4.5 (planned N = 3000). Survey participants attend two visits 6 months appart. Herein, we present the rationale, design, and enrolled study population of the ESCUDDO trial. ClinicalTrials.gov Identifier: NCT03180034.

Associations between Biomarkers of Exposure and Lung Cancer Risk among Exclusive Cigarette Smokers in the Golestan Cohort Study

Biomarkers of tobacco exposure are known to be associated with disease risk but previous studies are limited in number and restricted to certain regions. We conducted a nested case-control study examining baseline levels and subsequent lung cancer incidence among current male exclusive cigarette smokers in the Golestan Cohort Study in Iran. We calculated geometric mean biomarker concentrations for 28 matched cases and 52 controls for the correlation of biomarker levels among controls and for adjusted odds' ratios (ORs) for lung cancer incidence by biomarker concentration, accounting for demographic characteristics, smoking quantity and duration, and opium use. Lung cancer cases had higher average levels of most biomarkers including total nicotine equivalents (TNE-2), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL), and 3-hydroxyfluorene (3-FLU). Many biomarkers correlated highly with one another including TNE-2 with NNAL and N-Acetyl-S-(2-cyanoethyl)-L-cysteine (2CYEMA), and N-Acetyl-S-(4-hydroxy-2-buten-1-yl)-L-cysteine (t4HBEMA) with N-Acetyl-S-(3-hydroxypropyl-1-methyl)-L-cysteine (3HMPMA) and N-Acetyl-S-(4-hydroxy-2-methyl-2-buten-1-yl)-L-cysteine (4HMBEMA). Lung cancer risk increased with concentration for several biomarkers, including TNE-2 (OR = 2.22, 95% CI = 1.03, 4.78) and NNN (OR = 2.44, 95% CI = 1.13, 5.27), and estimates were significant after further adjustment for demographic and smoking characteristics for 2CYEMA (OR = 2.17, 95% CI = 1.03, 4.55), N-Acetyl-S-(2-carbamoylethyl)-L-cysteine (2CAEMA) (OR = 2.14, 95% CI = 1.01, 4.55), and N-Acetyl-S-(2-hydroxypropyl)-L-cysteine (2HPMA) (OR = 2.85, 95% CI = 1.04, 7.81). Estimates were not significant with adjustment for opium use. Concentrations of many biomarkers were higher at the baseline for participants who subsequently developed lung cancer than among the matched controls. Odds of lung cancer were higher for several biomarkers including with adjustment for smoking exposure for some but not with adjustment for opium use.

Thoughts on "AIDS and COVID-19: A Tale of Two Pandemics and the Role of Statisticians" by Susan S. Ellenberg and Jeffrey S. Morris

Human immunodeficiency virus and Covid-19 (or SARS-CoV-2) differ in their incubation distributions and in their susceptibility to immunologic defense. These features affect our ability to predict the course of these epidemics and to control them.

Assessing risk model calibration with missing covariates

When validating a risk model in an independent cohort, some predictors may be missing for some subjects. Missingness can be unplanned or by design, as in case-cohort or nested case-control studies, in which some covariates are measured only in subsampled subjects. Weighting methods and imputation are used to handle missing data. We propose methods to increase the efficiency of weighting to assess calibration of a risk model (i.e. bias in model predictions), which is quantified by the ratio of the number of observed events, $\mathcal{O}$, to expected events, $\mathcal{E}$, computed from the model. We adjust known inverse probability weights by incorporating auxiliary information available for all cohort members. We use survey calibration that requires the weighted sum of the auxiliary statistics in the complete data subset to equal their sum in the full cohort. We show that a pseudo-risk estimate that approximates the actual risk value but uses only variables available for the entire cohort is an excellent auxiliary statistic to estimate $\mathcal{E}$. We derive analytic variance formulas for $\mathcal{O}/\mathcal{E}$ with adjusted weights. In simulations, weight adjustment with pseudo-risk was much more efficient than inverse probability weighting and yielded consistent estimates even when the pseudo-risk was a poor approximation. Multiple imputation was often efficient but yielded biased estimates when the imputation model was misspecified. Using these methods, we assessed calibration of an absolute risk model for second primary thyroid cancer in an independent cohort.

Power of Microbiome Beta-Diversity Analyses Based on Standard Reference Samples

A simple method to analyze microbiome beta-diversity computes mean beta-diversity distances from a test sample to standard reference samples. We used reference stool and nasal samples from the Human Microbiome Project and regressed an outcome on mean distances (2 degrees-of-freedom (df) test) or additionally on squares and cross-product of mean distances (5-df test). We compared the power of 2-df and 5-df tests with the microbiome regression-based kernel association test (MiRKAT). In simulations, MiRKAT had moderately greater power than the 2-df test for discriminating skin versus saliva and skin versus nasal samples, but differences were negligible for skin versus stool and stool versus nasal samples. The 2-df test had slightly greater power than MiRKAT for Dirichlet multinomial samples. In associating body mass index with beta-diversity in stool samples from the American Gut Project, the 5-df test yielded smaller P values than MiRKAT for most taxonomic levels and beta-diversity measures. Unlike procedures like MiRKAT that are based on the beta-diversity matrix, mean distances to reference samples can be analyzed with standard statistical tools and shared or meta-analyzed without sharing primary DNA data. Our data indicate that standard reference tests have power comparable to MiRKAT's (and to permutational multivariate analysis of variance), but more simulations and applications are needed to confirm this.

Weight calibration to improve efficiency for estimating pure risks from the additive hazards model with the nested case-control design

We study the efficiency of covariate-specific estimates of pure risk (one minus the survival function) when some covariates are only available for case-control samples nested in a cohort. We focus on the semiparametric additive hazards model in which the hazard function equals a baseline hazard plus a linear combination of covariates with either time-varying or time-invariant coefficients. A published approach uses the design-based inclusion probabilities to reweight the nested case-control data. We obtain more efficient estimates of pure risks by calibrating the design weights to data available in the entire cohort, for both time-varying and time-invariant covariate coefficients. We develop explicit variance formulas for the weight-calibrated estimates based on influence functions. Simulations show the improvement in precision by using weight calibration and confirm the consistency of variance estimators and the validity of inference based on asymptotic normality. Examples are provided using data from the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial Study (PLCO).

Efficacy of the bivalent HPV vaccine against HPV 16/18-associated precancer: long-term follow-up results from the Costa Rica Vaccine Trial

BACKGROUND

Oncogenic human papillomavirus (HPV) infections cause most cases of cervical cancer. Here, we report long-term follow-up results for the Costa Rica Vaccine Trial (publicly funded and initiated before licensure of the HPV vaccines), with the aim of assessing the efficacy of the bivalent HPV vaccine for preventing HPV 16/18-associated cervical intraepithelial neoplasia grade 2 or worse (CIN2+).

METHODS

Women aged 18-25 years were enrolled in a randomised, double-blind, controlled trial in Costa Rica, between June 28, 2004, and Dec 21, 2005, designed to assess the efficacy of a bivalent vaccine for the prevention of infection with HPV 16/18 and associated precancerous lesions at the cervix. Participants were randomly assigned (1:1) to receive an HPV 16/18 AS04-adjuvanted vaccine or control hepatitis A vaccine. Vaccines were administered intramuscularly in three 0·5 mL doses at 0, 1, and 6 months and participants were followed up annually for 4 years. After the blinded phase, women in the HPV vaccine group were invited to enrol in the long-term follow-up study, which extended follow-up for 7 additional years. The control group received HPV vaccine and was replaced with a new unvaccinated control group. Women were followed up every 2 years until year 11. Investigators and patients were aware of treatment allocation for the follow-up phase. At each visit, clinicians collected cervical cells from sexually active women for cytology and HPV testing. Women with abnormal cytology were referred to colposcopy, biopsy, and treatment as needed. Women with negative results at the last screening visit (year 11) exited the long-term follow-up study. The analytical cohort for vaccine efficacy included women who were HPV 16/18 DNA-negative at vaccination. The primary outcome of this analysis was defined as histopathologically confirmed CIN2+ or cervical intraepithelial neoplasia grade 3 or worse associated with HPV 16/18 cervical infection detected at colposcopy referral. We calculated vaccine efficacy by year and cumulatively. This long-term follow-up study is registered with ClinicalTrials.gov, NCT00867464.

FINDINGS

7466 women were enrolled in the Costa Rica Vaccine Trial; 3727 received the HPV vaccine and 3739 received the control vaccine. Between March 30, 2009, and July 5, 2012, 2635 women in the HPV vaccine group and 2836 women in the new unvaccinated control group were enrolled in the long-term follow-up study. 2635 women in the HPV vaccine group and 2677 women in the control group were included in the analysis cohort for years 0-4, and 2073 women from the HPV vaccine group and 2530 women from the new unvaccinated control group were included in the analysis cohort for years 7-11. Median follow-up time for the HPV group was 11·1 years (IQR 9·1-11·7), 4·6 years (4·3-5·3) for the original control group, and 6·2 years (5·5-6·9) for the new unvaccinated control group. At year 11, vaccine efficacy against incident HPV 16/18-associated CIN2+ was 100% (95% CI 89·2-100·0); 34 (1·5%) of 2233 unvaccinated women had a CIN2+ outcome compared with none of 1913 women in the HPV group. Cumulative vaccine efficacy against HPV 16/18-associated CIN2+ over the 11-year period was 97·4% (95% CI 88·0-99·6). Similar protection was observed against HPV 16/18-associated CIN3-specifically at year 11, vaccine efficacy was 100% (95% CI 78·8-100·0) and cumulative vaccine efficacy was 94·9% (73·7-99·4). During the long-term follow-up, no serious adverse events occurred that were deemed related to the HPV vaccine. The most common grade 3 or worse serious adverse events were pregnancy, puerperium, and perinatal conditions (in 255 [10%] of 2530 women in the unvaccinated control group and 201 [10%] of 2073 women in the HPV vaccine group). Four women in the unvaccinated control group and three in the HPV vaccine group died; no deaths were deemed to be related to the HPV vaccine.

INTERPRETATION

The bivalent HPV vaccine has high efficacy against HPV 16/18-associated precancer for more than a decade after initial vaccination, supporting the notion that invasive cervical cancer is preventable.

FUNDING

US National Cancer Institute.

Durability of Cross-Protection by Different Schedules of the Bivalent HPV Vaccine: The CVT Trial

BACKGROUND

The Costa Rica HPV Vaccine Trial has documented cross-protection of the bivalent HPV vaccine against HPV31/33/45 up to 7 years after vaccination, even with one dose of the vaccine. However, the durability of such protection remains unknown. Here, we evaluate the efficacy of different schedules of the vaccine against HPV31/33/45 out to 11 years postvaccination, expanding to other nontargeted HPV types.

METHODS

We compared the rates of HPV infection in vaccinated women with the rates in a comparable cohort of unvaccinated women. We estimated the average vaccine efficacy (VEavg) against incident infections and tested for a change in VE over time.

RESULTS

Among 3-dose women, we observed statistically significant cross-protection against HPV31/33/45 (VEavg = 64.4%, 95% confidence interval [CI] = 57.7% to 70.0%). Additionally, we observed borderline, statistically significant cross-protection against HPV35 (VEavg = 23.2%, 95% CI = 0.3% to 40.8%) and HPV58 (VEavg = 21.2%, 95% CI = 4.2% to 35.3%). There was no decrease in VE over time (two-sided Ptrend > .05 for HPV31, -33, -35, -45, and -58). As a benchmark, VEavg against HPV16/18 was 82.0% (95% CI = 77.3% to 85.7%). Among 1-dose women, we observed comparable efficacy against HPV31/33/45 (VEavg = 54.4%, 95% CI = 21.0% to 73.7%). Acquisition of nonprotected HPV types was similar between vaccinated and unvaccinated women, indicating that the difference in HPV infection rates was not attributable to differential genital HPV exposure.

CONCLUSIONS

Substantial cross-protection afforded by the bivalent vaccine against HPV31/33/45, and to a lesser extent, HPV35 and HPV58, was sustained and remained stable after 11 years postvaccination, reinforcing the notion that the bivalent vaccine is an effective option for protection against HPV-associated cancers.

Evaluation of Durability of a Single Dose of the Bivalent HPV Vaccine: The CVT Trial

BACKGROUND

The authors investigated the durability of vaccine efficacy (VE) against human papillomavirus (HPV)16 or 18 infections and antibody response among nonrandomly assigned women who received a single dose of the bivalent HPV vaccine compared with women who received multiple doses and unvaccinated women.

METHODS

HPV infections were compared between HPV16 or 18-vaccinated women aged 18 to 25 years who received one (N = 112), two (N = 62), or three (N = 1365) doses, and age- and geography-matched unvaccinated women (N = 1783) in the long-term follow-up of the Costa Rica HPV Vaccine Trial. Cervical HPV infections were measured at two study visits, approximately 9 and 11 years after initial HPV vaccination, using National Cancer Institute next-generation sequencing TypeSeq1 assay. VE and 95% confidence intervals (CIs) were estimated. HPV16 or 18 antibody levels were measured in all one- and two-dose women, and a subset of three-dose women, using a virus-like particle-based enzyme-linked immunosorbent assay (n = 448).

RESULTS

Median follow-up for the HPV-vaccinated group was 11.3 years (interquartile range = 10.9-11.7 years) and did not vary by dose group. VE against prevalent HPV16 or 18 infection was 80.2% (95% CI = 70.7% to 87.0%) among three-dose, 83.8% (95% CI = 19.5% to 99.2%) among two-dose, and 82.1% (95% CI = 40.2% to 97.0%) among single-dose women. HPV16 or 18 antibody levels did not qualitatively decline between years four and 11 regardless of the number of doses given, although one-dose titers continue to be statistically significantly lower compared with two- and three-dose titers.

CONCLUSION

More than a decade after HPV vaccination, single-dose VE against HPV16 or 18 infection remained high and HPV16 or 18 antibodies remained stable. A single dose of bivalent HPV vaccine may induce sufficiently durable protection that obviates the need for more doses.

Comparative Validation of Breast Cancer Risk Prediction Models and Projections for Future Risk Stratification

BACKGROUND

External validation of risk models is critical for risk-stratified breast cancer prevention. We used the Individualized Coherent Absolute Risk Estimation (iCARE) as a flexible tool for risk model development and comparative model validation and to make projections for population risk stratification.

METHODS

Performance of two recently developed models, one based on the Breast and Prostate Cancer Cohort Consortium analysis (iCARE-BPC3) and another based on a literature review (iCARE-Lit), were compared with two established models (Breast Cancer Risk Assessment Tool and International Breast Cancer Intervention Study Model) based on classical risk factors in a UK-based cohort of 64 874 white non-Hispanic women (863 patients) age 35-74 years. Risk projections in a target population of US white non-Hispanic women age 50-70 years assessed potential improvements in risk stratification by adding mammographic breast density (MD) and polygenic risk score (PRS).

RESULTS

The best calibrated models were iCARE-Lit (expected to observed number of cases [E/O] = 0.98, 95% confidence interval [CI] = 0.87 to 1.11) for women younger than 50 years, and iCARE-BPC3 (E/O = 1.00, 95% CI = 0.93 to 1.09) for women 50 years or older. Risk projections using iCARE-BPC3 indicated classical risk factors can identify approximately 500 000 women at moderate to high risk (>3% 5-year risk) in the target population. Addition of MD and a 313-variant PRS is expected to increase this number to approximately 3.5 million women, and among them, approximately 153 000 are expected to develop invasive breast cancer within 5 years.

CONCLUSIONS

iCARE models based on classical risk factors perform similarly to or better than BCRAT or IBIS in white non-Hispanic women. Addition of MD and PRS can lead to substantial improvements in risk stratification. However, these integrated models require independent prospective validation before broad clinical applications.

Robustness of risk-based allocation of resources for disease prevention

Risk models for disease incidence can be useful for allocating resources for disease prevention if risk assessment is not too expensive. Assume there is a preventive intervention that should be given to everyone, but preventive resources are limited. We optimize risk-based prevention strategies and investigate robustness to modeling assumptions. The optimal strategy defines the proportion of the population to be given risk assessment and who should be offered intervention. The optimal strategy depends on the ratio of available resources to resources needed to intervene on everyone, and on the ratio of the costs of risk assessment to intervention. Risk assessment is not recommended if it is too expensive. Preventive efficiency decreases with decreasing compliance to risk assessment or intervention. Risk measurement error has little effect nor does misspecification of the risk distribution. Ignoring population substructure has small effects on optimal prevention strategy but can lead to modest over- or under-spending. We give conditions under which ignoring population substructure has no effect on optimal strategy. Thus, a simple one-population model offers robust guidance on prevention strategy but requires data on available resources, costs of risk assessment and intervention, population risk distribution, and probabilities of acceptance of risk assessment and intervention.

Confidence intervals for the difference between two relative risks

We provide methods to estimate the confidence interval for the difference between two relative risks. Letting p₀, p₁, and p₂ be the probabilities of an event in three groups (i.e. control, treatment 1, treatment 2), our methods estimate a confidence interval for r = p₁/p₀ - p₂/p₀. We highlight that our methods can handle small sample sizes, covariates, and study populations from multiple strata. We specifically developed these methods for vaccine trials to estimate the difference between two vaccine efficacies, where VE₁ = 1 - p₁/p₀, VE₂ = 1 - p₂/p₀ and r = VE₂ - VE₁. We showcase our methods by using interim data from one of these trials to suggest that one dose of the human papillomavirus vaccine may be as efficacious as two doses of the vaccine.

Abstract A39: The human oral microbiota and risk of lung cancer: An analysis of three prospective cohort studies

Background: The oral microbiota may be associated with lung cancer risk through direct mechanisms, including infection, immune responses, and periodontal disease, and through indirect mechanisms such as the modification of the oral microbiota by tobacco. We conducted a case-cohort study nested within three US prospective cohort studies to evaluate the association between oral microbiota ascertained years before a cancer diagnosis and risk of lung cancer.

Methods: Incident lung cancer cases within the Agricultural Health Study (AHS; N=244), NIH-AARP Diet and Health Study (N=376), and the Prostate, Lung, Colorectum, and Ovarian Cancer Screening Trial (PLCO; N=700) who provided an oral wash sample were identified. The median time between oral sample collection and diagnosis was approximately 6.6 years, 3.4 years, and 4.5 years for AHS, NIH-AARP, and PLCO, respectively. A referent subcohort was randomly selected by strata of age, sex, and cigarette smoking history. We extracted DNA using the DSP DNA Virus Pathogen kit, and the V4 region of the 16S rRNA gene was PCR amplified and sequenced using the MiSeq. The sequencing data were processed using QIIME2 with the DADA2 plugin and we generated alpha and beta diversity metrics. Cox proportional hazards models were used to evaluate the hazard ratios (HR) and 95% confidence intervals (CI) for the association between the oral microbial measures and the risk of lung cancer with adjustment for known lung cancer risk factors, and estimates from the three cohorts were meta-analyzed.

Results: Increased alpha diversity was associated with decreased lung cancer risk, although only the association with the Shannon Index reached statistical significance (HR for 5th quintile versus 1st quintile 0.74; 95% CI 0.60, 0.92) with no evidence of between-study heterogeneity (p = 0.5968). Specific principal coordinate vectors from the beta diversity matrices were also significantly associated with lung cancer risk, suggesting differing bacterial communities between future lung cancer cases. When stratified by histologic subtypes, the inverse association with alpha diversity was restricted to squamous cell carcinoma, with all alpha diversity metrics reaching statistical significance (e.g., Faith’s phylogenetic diversity HR for 5th quintile versus 1st quintile 0.57; 95% CI: 0.37, 0.87). Similarly, when stratified by smoking history, the inverse association with alpha diversity was restricted to former smokers (e.g., observed species HR for 5th quintile versus 1st quintile 0.63; 95% CI: 0.44, 0.89).

Conclusions: In oral wash samples collected years before diagnosis, we found significant associations between both alpha and beta diversity metrics of the oral microbial communities and risk of lung cancer. Additional work is required to understand the associations by histologic subtype and smoking history.

Citation Format: Emily Vogtmann, Xing Hua, Guoqin Yu, Autumn Hullings, Yunhu Wan, Casey L Dagnall, Kristine Jones, Belynda D. Hicks, Amy Hutchinson, Shalabh Suman, Bin Zhu, Barry Graubard, Mitchell H. Gail, J. Gregory Caporaso, William Wheeler, Dale Sandler, Laura E. Beane Freeman, Linda Liao, Neal D. Freedman, Neil Caporaso, Rashmi Sinha, Jianxin Shi, Christian C Abnet. The human oral microbiota and risk of lung cancer: An analysis of three prospective cohort studies [abstract]. In: Proceedings of the AACR Special Conference on the Microbiome, Viruses, and Cancer; 2020 Feb 21-24; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2020;80(8 Suppl):Abstract nr A39.

The Need for Combined Assessment of Multiple Outcomes in Noninferiority Trials in Oncology

Noninferiority trials in oncology assess novel therapies with the potential for slightly worse recurrence or death outcomes (ie, the margin of noninferiority) than standard therapies. This poses a dilemma because, in the absence of potential health outcome advantages, these trials may not provide the treatment equipoise required for an ethical study. Any new treatment with the potential for slightly worse recurrence or death outcomes should have countervailing health outcome advantages, but these are rarely taken into account in the design of noninferiority trials. This article presents the argument that not only the potentially worse health outcomes but also the potential benefits of the novel therapy should be considered when designing, analyzing, and reporting noninferiority trials. Some approaches to study design and analysis that consider both primary and secondary end points are discussed, and reporting the joint distributions of end points for the novel and standard treatments is recommended.

Design choices for observational studies of the effect of exposure on disease incidence

The purpose of this paper is to help readers choose an appropriate observational study design for measuring an association between an exposure and disease incidence. We discuss cohort studies, sub-samples from cohorts (case-cohort and nested case-control designs), and population-based or hospital-based case-control studies. Appropriate study design is the foundation of a scientifically valid observational study. Mistakes in design are often irremediable. Key steps are understanding the scientific aims of the study and what is required to achieve them. Some designs will not yield the information required to realise the aims. The choice of design also depends on the availability of source populations and resources. Choosing an appropriate design requires balancing the pros and cons of various designs in view of study aims and practical constraints. We compare various cohort and case-control designs to estimate the effect of an exposure on disease incidence and mention how certain design features can reduce threats to study validity.

On Mendelian randomization analysis of case-control study

Mendelian randomization (MR) analysis uses genotypes as instruments to estimate the causal effect of an exposure in the presence of unobserved confounders. The existing MR methods focus on the data generated from prospective cohort studies. We develop a procedure for studying binary outcomes under a case-control design. The proposed procedure is built upon two working models commonly used for MR analyses and adopts a quasi-empirical likelihood framework to address the ascertainment bias from case-control sampling. We derive various approaches for estimating the causal effect and hypothesis testing under the empirical likelihood framework. We conduct extensive simulation studies to evaluate the proposed methods. We find that the proposed empirical likelihood estimate is less biased than the existing estimates. Among all the approaches considered, the Lagrange multiplier (LM) test has the highest power, and the confidence intervals derived from the LM test have the most accurate coverage. We illustrate the use of our method in MR analysis of prostate cancer case-control data with vitamin D level as exposure and three single nucleotide polymorphisms as instruments.

Using standard microbiome reference groups to simplify beta-diversity analyses and facilitate independent validation

Motivation

Comparisons of microbiome communities across populations are often based on pairwise distance measures (beta-diversity). Standard analyses (principal coordinate plots, permutation tests, kernel methods) require access to primary data if another investigator wants to add or compare independent data. We propose using standard reference measurements to simplify microbiome beta-diversity analyses, to make them more transparent, and to facilitate independent validation and comparisons across studies.

Results

Using stool and nasal reference sets from the Human Microbiome Project (HMP), we computed mean distances (actually Bray-Curtis or Pearson correlation dissimilarities) to each reference set for each new sample. Thus, each new sample has two mean distances that can be plotted and analyzed with classical statistical methods. To test the approach, we studied independent (not reference) HMP subjects. Simple Hotelling tests demonstrated statistically significant differences in mean distances to reference sets between all pairs of body sites (stool, skin, nasal, saliva and vagina) at the phylum, class, order, family and genus levels. Using the distance to a single reference set was usually sufficient, but using both reference sets always worked well. The use of reference sets simplifies standard analyses of beta-diversity and facilitates the independent validation and combining of such data because others can compute distances to the same reference sets. Moreover, standard statistical methods for survival analysis, logistic regression and other procedures can be applied to vectors of mean distances to reference sets, thereby greatly expanding the potential uses of beta-diversity information. More work is needed to identify the best reference sets for particular applications.

Availability and implementation

https://github.com/NCI-biostats/microbiome-fixed-reference.

Supplementary information

Supplementary data are available at Bioinformatics online.

Risk-Reducing Salpingo-Oophorectomy and Breast Cancer Risk Reduction in the Gynecologic Oncology Group Protocol-0199 (GOG-0199)

Background

Risk-reducing salpingo-oophorectomy (RRSO) has been associated with approximately 50% breast cancer risk reduction among women with a pathogenic variant in BRCA1 or BRCA2 (BRCA1/2), a finding that has recently been questioned.

Methods

We estimated incidence rates of breast cancer and all cancers combined during 5 years of follow-up among participants selecting RRSO or ovarian cancer screening (OCS) among women with a BRCA1/2 pathogenic variant or strong breast and/or ovarian cancer family history. Ovarian or fallopian tube or peritoneal cancer incidence rates were estimated for the OCS group. Breast cancer hazard ratios (HRs) for time-dependent RRSO were estimated using Cox regression with age time-scale (4943 and 4990 women-years in RRSO and OCS cohorts, respectively). All statistical tests were two-sided.

Results

The RRSO cohort included 925 participants, and 1453 participants were in the OCS cohort (381 underwent RRSO during follow-up), with 88 incident breast cancers diagnosed. Among BRCA1/2 pathogenic variant carriers, a non-statistically significant lower breast cancer incidence was observed in the RRSO compared with the OCS cohort (HR = 0.86, 95% confidence interval  = 0.45 to 1.67; P = .67). No difference was observed in the overall population or among subgroups stratified by prior breast cancer history or menopausal status. Seven fallopian tube and four ovarian cancers were prospectively diagnosed in the OCS cohort, and one primary peritoneal carcinoma occurred in the RRSO cohort.

Conclusions

These data suggest that RRSO might be associated with reduced breast cancer incidence among women with a BRCA1/2 pathogenic variant, although the effect, if present, is small. This evolving evidence warrants a thorough discussion regarding the impact of RRSO on breast cancer risk with women considering this intervention.

Effects of Nutrition Intervention on Total and Cancer Mortality: 25-Year Post-trial Follow-up of the 5.25-Year Linxian Nutrition Intervention Trial

Background

A beneficial effect of supplementation with selenium, vitamin E, and beta-carotene was observed on total and cancer mortality in a Chinese population, and it endured for 10 years postintervention, but longer durability is unknown.

Methods

A randomized, double-blind, placebo-controlled trial was conducted in Linxian, China, from 1986 to 1991; 29 584 residents age 40 to 69 years received daily supplementations based on a factorial design: Factors A (retinol/zinc), B (riboflavin/niacin), C (vitamin C/molybdenum), and/or D (selenium/vitamin E/beta-carotene), or placebo for 5.25 years, and followed for up 25 years. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the intervention effects on mortalities were estimated using Cox proportional hazards models.

Results

Through 2016, the interventions showed no effect on total mortality. The previously reported protective effect of Factor D against total mortality was lost 10 years postintervention. The protective effect of Factor D for gastric cancer was attenuated (HR = 0.93, 95% CI = 0.85 to 1.01), but a newly apparent protective effect against esophageal cancer was found for Factor B (HR = 0.92, 95% CI = 0.85 to 1.00, two-sided P = .04). Other protective/adverse associations were observed for cause-specific mortalities. Protective effects were found in people younger than age 55 years at baseline against non-upper gastrointestinal cancer death for Factor A (HR = 0.80, 95% CI = 0.69 to 0.92) and against death from stroke for Factor C (HR = 0.89, 95% CI = 0.82 to 0.96). In contrast, increased risk of esophageal cancer was found when the intervention began after age 55 years for Factors C (HR = 1.16, 95% CI = 1.04 to 1.30) and D (HR = 1.20, 95% CI = 1.07 to 1.34).

Conclusions

Multiyear nutrition intervention is unlikely to have a meaningful effect on mortality more than a decade after supplementation ends, even in a nutritionally deprived population. Whether sustained or repeat intervention would provide longer effects needs further investigation.

Breast Cancer Risk Model Requirements for Counseling, Prevention, and Screening

Background

Incorporation of polygenic risk scores and mammographic density into models to predict breast cancer incidence can increase discriminatory accuracy (area under the receiver operating characteristic curve [AUC]) from 0.6 for models based only on epidemiologic factors to 0.7. It is timely to assess what impact these improvements will have on individual counseling and on public health prevention and screening strategies, and to determine what further improvements are needed.

Methods

We studied various clinical and public health applications using a log-normal distribution of risk.

Results

Provided they are well calibrated, even risk models with AUCs of 0.6 to 0.7 provide useful perspective for individual counseling and for weighing the harms and benefits of preventive interventions in the clinic. At the population level, they are helpful for designing preventive intervention trials, for assessing reductions in absolute risk from reducing exposure to modifiable risk factors, and for resource allocation (although a higher AUC would be desirable for risk-based allocation). Other public health applications require higher AUCs that can only be achieved with risk predictors 1.6 to 8.8 times as strong as all those yet discovered combined. Such applications are preventing an appreciable proportion of population disease when employing a high-risk prevention strategy and deciding who should be screened for subclinical disease.

Conclusions

Current and foreseeable risk models are useful for counseling and some prevention activities, but given the daunting challenge of achieving, for example, an AUC of 0.8, considerable effort should be put into finding effective preventive interventions and screening strategies with fewer adverse effects.

Perspectives for Consideration in the Development of Microbial Cell Reference Materials

Microbiome measurement and analyses benefit greatly from incorporation of reference materials as controls. However, there are many points to consider in defining an ideal whole-cell reference material standard. Such a standard would embody all the diversity and measurement challenges present in real samples, would be completely characterized to provide "ground truth" data, and would be inexpensive and widely available. This ideal is, unfortunately, not readily attainable because of the diverse nature of different sequencing projects. Some applications may benefit most from highly complex reference materials, while others will value characterization or low expense more highly. The selection of appropriate microbial whole-cell reference materials to benchmark and validate microbial measurements should be considered carefully and may vary among specific applications. In this article, we describe a perspective on the development of whole-cell microbial reference materials for use in metagenomics analyses.

Effects of Helicobacter pylori treatment and vitamin and garlic supplementation on gastric cancer incidence and mortality: follow-up of a randomized intervention trial

OBJECTIVE

To assess the effects of Helicobacter pylori treatment, vitamin supplementation, and garlic supplementation in the prevention of gastric cancer.

DESIGN

Blinded randomized placebo controlled trial.

SETTING

Linqu County, Shandong province, China.

PARTICIPANTS

3365 residents of a high risk region for gastric cancer. 2258 participants seropositive for antibodies to H pylori were randomly assigned to H pylori treatment, vitamin supplementation, garlic supplementation, or their placebos in a 2×2×2 factorial design, and 1107 H pylori seronegative participants were randomly assigned to vitamin supplementation, garlic supplementation, or their placebos in a 2×2 factorial design.

INTERVENTIONS

H pylori treatment with amoxicillin and omeprazole for two weeks; vitamin (C, E, and selenium) and garlic (extract and oil) supplementation for 7.3 years (1995-2003).

MAIN OUTCOME MEASURES

Primary outcomes were cumulative incidence of gastric cancer identified through scheduled gastroscopies and active clinical follow-up through 2017, and deaths due to gastric cancer ascertained from death certificates and hospital records. Secondary outcomes were associations with other cause specific deaths, including cancers or cardiovascular disease.

RESULTS

151 incident cases of gastric cancer and 94 deaths from gastric cancer were identified during 1995-2017. A protective effect of H pylori treatment on gastric cancer incidence persisted 22 years post-intervention (odds ratio 0.48, 95% confidence interval 0.32 to 0.71). Incidence decreased significantly with vitamin supplementation but not with garlic supplementation (0.64, 0.46 to 0.91 and 0.81, 0.57 to 1.13, respectively). All three interventions showed significant reductions in gastric cancer mortality: fully adjusted hazard ratio for H pylori treatment was 0.62 (95% confidence interval 0.39 to 0.99), for vitamin supplementation was 0.48 (0.31 to 0.75), and for garlic supplementation was 0.66 (0.43 to 1.00). Effects of H pylori treatment on both gastric cancer incidence and mortality and of vitamin supplementation on gastric cancer mortality appeared early, but the effects of vitamin supplementation on gastric cancer incidence and of garlic supplementation only appeared later. No statistically significant associations were found between interventions and other cancers or cardiovascular disease.

CONCLUSIONS

H pylori treatment for two weeks and vitamin or garlic supplementation for seven years were associated with a statistically significant reduced risk of death due to gastric cancer for more than 22 years. H pylori treatment and vitamin supplementation were also associated with a statistically significantly reduced incidence of gastric cancer.

TRIAL REGISTRATION

ClinicalTrials.gov NCT00339768.

Estimating the decision curve and its precision from three study designs

The decision curve plots the net benefit ( N B ) of a risk model for making decisions over a range of risk thresholds, corresponding to different ratios of misclassification costs. We discuss three methods to estimate the decision curve, together with corresponding methods of inference and methods to compare two risk models at a given risk threshold. One method uses risks (R) and a binary event indicator (Y) on the entire validation cohort. This method makes no assumptions on how well-calibrated the risk model is nor on the incidence of disease in the population and is comparatively robust to model miscalibration. If one assumes that the model is well-calibrated, one can compute a much more precise estimate of N B based on risks R alone. However, if the risk model is miscalibrated, serious bias can result. Case-control data can also be used to estimate N B if the incidence (or prevalence) of the event ( Y = 1 ) is known. This strategy has comparable efficiency to using the full ( R , Y ) data, and its efficiency is only modestly less than that for the full ( R , Y ) data if the incidence is estimated from the mean of Y. We estimate variances using influence functions and propose a bootstrap procedure to obtain simultaneous confidence bands around the decision curve for a range of thresholds. The influence function approach to estimate variances can also be applied to cohorts derived from complex survey samples instead of simple random samples.

Prenatal diethylstilbestrol exposure and cancer risk in women

In the Diethylstilbestrol [DES] Combined Cohort Follow-up, the age- and calendar-year specific standardized incidence ratio [SIR] for clear cell adenocarcinoma [CCA] was 27.6 (95% confidence interval [CI] 7.51-70.6) for the exposed women. The SIR for breast cancer was 1.17 (95% CI 1.01-1.36) and the hazard ratio [HR] adjusted for birth year and cohort for comparison with the unexposed was 1.05 (95% CI 0.79-1.41). The SIR for pancreatic cancer was 2.43 (95% CI 1.21-4.34) and the adjusted HR for comparison with unexposed women was 7.16 (95% CI 0.84-61.5). There was little evidence of excess risk for other sites. There appeared to be a deficit in risk for endometrial cancer among the exposed (SIR 0.61; 95% CI 0.35-0.98), and an excess in the unexposed (SIR 1.55; 95% CI 0.95-2.40); the adjusted HR was 0.45 (95% CI 0.22-0.93) for the internal comparison. There was no overall excess cancer risk in exposed women compared with general population rates (1.06; 95% CI 0.95-1.17) or with unexposed participants (adjusted HR 1.03; 95% CI 0.84-1.25). These data do not support the suggestion that there is a diathesis of cancers in DES exposed female offspring The excess risk of breast and pancreatic cancers that we observed is concerning and warrants continued follow-up and mechanistic investigation. Environ. Mol. Mutagen. 60:395-403, 2019. Published 2017. This article is a US Government work and is in the public domain in the USA.

Design and analysis considerations for combining data from multiple biomarker studies

Pooling data from multiple studies improves estimation of exposure-disease associations through increased sample size. However, biomarker exposure measurements can vary substantially across laboratories and often require calibration to a reference assay prior to pooling. We develop two statistical methods for aggregating biomarker data from multiple studies: the full calibration method and the internalized method. The full calibration method calibrates all biomarker measurements regardless of the availability of reference laboratory measurements while the internalized method calibrates only non-reference laboratory measurements. We compare the performance of these two aggregation methods to two-stage methods. Furthermore, we compare the aggregated and two-stage methods when estimating the calibration curve from controls only or from a random sample of individuals from the study cohort. Our findings include the following: (1) Under random sampling for calibration, exposure effect estimates from the internalized method have a smaller mean squared error than those from the full calibration method. (2) Under the controls-only calibration design, the full calibration method yields effect estimates with the least bias. (3) The two-stage approaches produce average effect estimates that are similar to the full calibration method under a controls only calibration design and the internalized method under a random sample calibration design. We illustrate the methods in an application evaluating the relationship between circulating vitamin D levels and stroke risk in a pooling project of cohort studies.

Next steps in studying the human microbiome and health in prospective studies, Bethesda, MD, May 16-17, 2017

The National Cancer Institute (NCI) sponsored a 2-day workshop, "Next Steps in Studying the Human Microbiome and Health in Prospective Studies," in Bethesda, Maryland, May 16-17, 2017. The workshop brought together researchers in the field to discuss the challenges of conducting microbiome studies, including study design, collection and processing of samples, bioinformatics and statistical methods, publishing results, and ensuring reproducibility of published results. The presenters emphasized the great potential of microbiome research in understanding the etiology of cancer. This report summarizes the workshop and presents practical suggestions for conducting microbiome studies, from workshop presenters, moderators, and participants.

Circulating 25-hydroxyvitamin D up to 3 decades prior to diagnosis in relation to overall and organ-specific cancer survival

While vitamin D has been associated with improved overall cancer survival in some investigations, few have prospectively evaluated organ-specific survival. We examined the accepted biomarker of vitamin D status, serum 25-hydroxyvitamin D [25(OH)D], and cancer survival in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. Of 4616 cancer cases with measured serum 25(OH)D, 2884 died of their cancer during 28 years of follow-up and 1732 survived or died of other causes. Proportional hazards regression estimated hazard ratios (HR) and 95% confidence intervals (CI) for the association between pre-diagnostic 25(OH)D and overall and site-specific survival. Serum 25(OH)D was significantly lower among cases who subsequently died from their malignancy compared with those who did not (medians 34.7 vs. 36.5 nmol/L, respectively; p = 0.01). Higher 25(OH)D was associated with lower overall cancer mortality (HR = 0.76, 95% CI 0.67-0.85 for highest vs. lowest quintile, p-trend < 0.0001). Higher 25(OH)D was related to lower mortality from the following site-specific malignancies: prostate (HR = 0.74, 95% CI 0.55-1.01, p-trend = 0.005), kidney (HR = 0.59, 95% CI 0.35-0.98, p-trend = 0.28), and melanoma (HR = 0.39, 95% CI 0.20-0.78, p-trend = 0.01), but increased mortality from lung cancer (HR = 1.28, 95% CI 1.02-1.61, p-trend = 0.19). Improved survival was also suggested for head and neck, gastric, pancreatic, and liver cancers, though not statistically significantly, and case numbers for the latter two organ sites were small. Higher 25(OH)D status years prior to diagnosis was related to improved survival for overall and some site-specific cancers, associations that should be examined in other prospective populations that include women and other racial-ethnic groups.

Proportion of U.S. Trends in Breast Cancer Incidence Attributable to Long-term Changes in Risk Factor Distributions

Background: U.S. breast cancer incidence has been changing, as have distributions of risk factors, including body mass index (BMI), age at menarche, age at first live birth, and number of live births.Methods: Using data for U.S. women from large nationally representative surveys, we estimated risk factor distributions from 1980 to 2008. To estimate ecologic associations with breast cancer incidence, we fitted Poisson models to age- and calendar year-specific incidence data from the NCI's Surveillance, Epidemiology and End Results registries from 1980 to 2011. We then assessed the proportion of incidence attributable to specific risk factors by comparing incidence from models that only included age and calendar period as predictors with models that additionally included age- and cohort-specific categorized mean risk factors. Analyses were stratified by age and race.Results: Ecologic associations usually agreed with previous findings from analytic epidemiology. From 1980 to 2011, compared with the risk factor reference level, increased BMI was associated with 7.6% decreased incidence in women ages 40 to 44 and 2.6% increased incidence for women ages 55 to 59. Fewer births were associated with 22.2% and 3.99% increased incidence in women ages 40 to 44 and 55 to 59 years, respectively. Changes in age at menarche and age at first live birth in parous women did not significantly impact population incidence from 1980 to 2011.Conclusions: Changes in BMI and number of births since 1980 significantly impacted U.S. breast cancer incidence.Impact: Quantifying long-term impact of risk factor trends on incidence is important to understand the future breast cancer burden and inform prevention efforts. Cancer Epidemiol Biomarkers Prev; 27(10); 1214-22. ©2018 AACR.

Evidence for single-dose protection by the bivalent HPV vaccine-Review of the Costa Rica HPV vaccine trial and future research studies

The Costa Rica Vaccine Trial (CVT), a phase III randomized clinical trial, provided the initial data that one dose of the HPV vaccine could provide durable protection against HPV infection. Although the study design was to administer all participants three doses of HPV or control vaccine, 20% of women did not receive the three-dose regimens, mostly due to involuntary reasons unrelated to vaccination. In 2011, we reported that a single dose of the bivalent HPV vaccine could be as efficacious as three doses of the vaccine using the endpoint of persistent HPV infection accumulated over the first four years of the trial; findings independently confirmed in the GSK-sponsored PATRICIA trial. Antibody levels after one dose, although lower than levels elicited by three doses, were 9-times higher than levels elicited by natural infection. Importantly, levels remained essentially constant over at least seven years, suggesting that the observed protection provided by a single dose might be durable. Much work has been done to assure these non-randomized findings are valid. Yet, the group of recipients who received one dose of the bivalent HPV vaccine in the CVT and PATRICIA trials was small and not randomly selected nor blinded to the number of doses received. The next phase of research is to conduct a formal randomized, controlled trial to evaluate the protection afforded by a single dose of HPV vaccine. Complementary studies are in progress to bridge our findings to other populations, and to further document the long-term durability of antibody response following a single dose.

Excess Deaths Associated With Underweight, Overweight, and Obesity: An Evaluation of Potential Bias

As the prevalence of obesity has increased over time in the United States (1,2), concern over the association between body weight and excess mortality also increased. In 2005, an analysis of estimated excess deaths, relative to the normal weight category (body mass index [BMI] 18.5-24.9), that were associated with underweight (BMI less than 18.5), overweight (BMI 25.0-29.9), and obesity (BMI greater than or equal to 30) in U.S. adults in 2000 was published (3). Both underweight and obesity, particularly higher levels of obesity, were associated with increased mortality relative to the normal weight category. Obesity was estimated to be associated with 111,909 excess deaths (95% confidence interval [CI]: 53,754 to 170,064) in 2000 relative to the normal weight category, and underweight with 33,746 excess deaths (95% CI: 15,726 to 51,766). Overweight was associated with reduced mortality (-86,094 deaths; 95% CI: -161,223 to -10,966). This report evaluates several potential sources of bias in that analysis.

Design and statistical considerations for studies evaluating the efficacy of a single dose of the human papillomavirus (HPV) vaccine

Cervical cancer is a leading cause of cancer mortality in women worldwide. Human papillomavirus (HPV) types 16 and 18 cause about 70% of all cervical cancers. Clinical trials have demonstrated that three doses of either commercially available HPV vaccine, Cervarix ® or Gardasil ®, prevent most new HPV 16/18 infections and associated precancerous lesions. Based on evidence of immunological non-inferiority, 2-dose regimens have been licensed for adolescents in the United States, European Union, and elsewhere. However, if a single dose were effective, vaccine costs would be reduced substantially and the logistics of vaccination would be greatly simplified, enabling vaccination programs in developing countries. The National Cancer Institute (NCI) and the Agencia Costarricense de Investigaciones Biomédicas (ACIB) are conducting, with support from the Bill & Melinda Gates Foundation and the International Agency for Research on Cancer (IARC), a large 24,000 girl study to evaluate the efficacy of a 1-dose regimen. The first component of the study is a four-year non-inferiority trial comparing 1- to 2-dose regimens of the two licensed vaccines. The second component is an observational study that estimates the vaccine efficacy (VE) of each regimen by comparing the HPV infection rates in the trial arms to those in a contemporaneous survey group of unvaccinated girls. In this paper, we describe the design and statistical analysis for this study. We explain the advantage of defining non-inferiority on the absolute risk scale when the expected event rate is near 0 and, given this definition, suggest an approach to account for missing clinic visits. We then describe the problem of estimating VE in the absence of a randomized placebo arm and offer our solution.

Durability of Protection Afforded by Fewer Doses of the HPV16/18 Vaccine: The CVT Trial

Background

Previously, we demonstrated similar human papillomavirus (HPV)16/18 vaccine efficacy estimates and stable HPV16/18 antibody levels four years postvaccination in a nonrandomized analysis of women who received a varying number of doses of the bivalent HPV16/18 vaccine. Here we extend data to seven years following initial vaccination.

Methods

We evaluated HPV16/18-vaccinated women who received one (n = 134), two (n 0/1 = 193, n 0/6 = 79), or three doses (n = 2043) to a median of 6.9 years postvaccination. Cervical HPV DNA was measured with the SPF10- DEIA-LiPA PCR system; HPV16/18-specific antibody levels were measured using enzyme-linked immunosorbent assays (n = 486). Infection and immunological measures were compared across vaccine dose groups. Prevalent HPV infection at year 7 was also compared with an unvaccinated control group (UCG). All statistical tests were two-sided.

Results

Among women in the three-dose, two-dose 0/6 , two-dose 0/1 , and one-dose groups, cumulative incident HPV16/18 infection rates (No. of events/No. of individuals) were 4.3% (88/2036, 95% confidence interval [CI] = 3.5% to 5.3%), 3.8% (3/78, 95% CI = 1.0% to 10.1%), 3.6% (7/192, 95% CI = 1.6% to 7.1%), and 1.5% (2/133, 95% CI = 0.3% to 4.9%; P = 1.00, .85, .17 comparing the two-dose 0/6 , two-dose 0/1 , and one-dose groups to the three-dose group, respectively). The prevalence of other carcinogenic and noncarcinogenic HPV types, excluding HPV16/18/31/33/45, were high and not statistically different among all dose groups, indicating that the low incidence of HPV16/18 in the one- and two-dose groups was not due to lack of exposure. At seven years, 100% of participants in all dose groups remained HPV16 and HPV18 seropositive. A non-statistically significant decrease in the geometric mean of the HPV16 antibody levels between years 4 and 7 was observed among women in the three-dose group: -10.8% (95% CI = -25.3% to 6.6%); two-dose (0/6 months) group: -17.3% (95% CI = -39.3% to 12.8%), two-dose (0/1 month) group: -6.9% (95% CI = -22.1% to 11.2%), and one-dose group: -5.5% (95% CI = -29.7% to 27.0%); results were similar for HPV18.

Conclusions

At an average of seven years of follow-up, we observed similar low rates of HPV16/18 infections and slight, if any, decreases in HPV16/18 antibody levels by dose group.

Power and sample size for multivariate logistic modeling of unmatched case-control studies

Sample size calculations are needed to design and assess the feasibility of case-control studies. Although such calculations are readily available for simple case-control designs and univariate analyses, there is limited theory and software for multivariate unconditional logistic analysis of case-control data. Here we outline the theory needed to detect scalar exposure effects or scalar interactions while controlling for other covariates in logistic regression. Both analytical and simulation methods are presented, together with links to the corresponding software.

Comparison of approaches for incorporating new information into existing risk prediction models

We compare the calibration and variability of risk prediction models that were estimated using various approaches for combining information on new predictors, termed 'markers', with parameter information available for other variables from an earlier model, which was estimated from a large data source. We assess the performance of risk prediction models updated based on likelihood ratio (LR) approaches that incorporate dependence between new and old risk factors as well as approaches that assume independence ('naive Bayes' methods). We study the impact of estimating the LR by (i) fitting a single model to cases and non-cases when the distribution of the new markers is in the exponential family or (ii) fitting separate models to cases and non-cases. We also evaluate a new constrained maximum likelihood method. We study updating the risk prediction model when the new data arise from a cohort and extend available methods to accommodate updating when the new data source is a case-control study. To create realistic correlations between predictors, we also based simulations on real data on response to antiviral therapy for hepatitis C. From these studies, we recommend the LR method fit using a single model or constrained maximum likelihood. Copyright © 2016 John Wiley & Sons, Ltd.