Dongfeng Wu

Sojourn time and lead time projection in lung cancer screening.

OBJECTIVES We investigate screening sensitivity, transition probability and sojourn time in lung cancer screening for male heavy smokers using the Mayo Lung Project data. We also estimate the lead time distribution, its property, and the projected effect of taking regular chest X-rays for lung cancer detection. METHODS We apply the statistical method developed by Wu et al. [1] using the Mayo Lung Project (MLP) data, to make Bayesian inference for the screening test sensitivity, the age-dependent transition probability from disease-free to preclinical state, and the sojourn time distribution, for male heavy smokers in a periodic screening program. We then apply the statistical method developed by Wu et al. [2] using the Bayesian posterior samples from the MLP data to make inference for the lead time, the time of diagnosis advanced by screening for male heavy smokers. The lead time is distributed as a mixture of a point mass at zero and a piecewise continuous distribution, which corresponds to the probability of no-early-detection, and the probability distribution of the early diagnosis time. We present estimates of these two measures for male heavy smokers by simulations. RESULTS The posterior sensitivity is almost symmetric, with posterior mean 0.89, and posterior median 0.91; the 95% highest posterior density (HPD) interval is (0.72, 0.98). The posterior mean sojourn time is 2.24 years, with a posterior median of 2.20 years for male heavy smokers. The 95% HPD interval for the mean sojourn time is (1.57, 3.35) years. The age-dependent transition probability is not a monotone function of age; it has a single maximum at age 68. The mean lead time increases as the screening time interval decreases. The standard error of the lead time also increases as the screening time interval decreases. CONCLUSION Although the mean sojourn time for male heavy smokers is longer than expected, the predictive estimation of the lead time is much shorter. This may provide policy makers important information on the effectiveness of the chest X-rays and sputum cytology in lung cancer early detection.

Preliminary Assessment of a School-Based Healthy Lifestyle Intervention Among Rural Elementary School Children

BACKGROUND Childhood obesity has become a national public health crisis in America. Physical inactivity and unhealthy eating behaviors may contribute to the childhood obesity epidemic. School-based healthy lifestyle interventions play a promising role in preventing and controlling childhood obesity. A comprehensive school-based healthy lifestyle intervention was implemented in 4 rural elementary schools in Kentucky. METHODS The intervention included 4 goals: improving physical education, health education, family/community involvement, and school wellness policies. Childrens physical activity was assessed by pedometer, and nutrition was assessed by a previous day recall survey in January (baseline), February (t1), March (t2), April (t3), and May (t4) of 2011. RESULTS The intervention had significant effects on increasing the percentages of children meeting physical activity (1% vs 5%, p < .01) and nutrition (15% vs 26%, p < .01) recommendations. The effects of the intervention on physical activity and nutrition depended on school, grade, and age of the children. There was an increasing linear trend of physical activity and an increasing quadratic trend of nutrition over time among children. CONCLUSIONS The intervention had beneficial effects in improving healthy behaviors among children. Further studies are needed to assess its long-term effects and cost-effectiveness.

Estimating key parameters in periodic breast cancer screening—Application to the Canadian National Breast Screening Study data

PROBLEM STATEMENT Breast cancer screening in women of younger age has been controversial. The screening sensitivities, transition probabilities and sojourn time distributions are estimated for females aged 40-49 years and 50-59 years separately, using the Canadian National Breast Screening Study (CNBSS) data. The purpose is to estimate the lead time distribution and the probability of not detecting the cancer early. APPROACH Within the 40-49-year-old and 50-59-year-old cohorts separately, the age-independent statistical model was applied. Bayesian estimators along with 95% highest probability density (HPD) credible intervals (CI) were calculated. Bayesian hypothesis testing was used to compare the parameter estimates of the two cohorts. The lead time density was also estimated for both the 40-49 and 50-59-year-old cohorts. RESULTS The screening sensitivity, transition probability of the disease, and mean sojourn time were all found to increase with age. For the 40-49-year-old and 50-59-year-old cohorts, the posterior mean sensitivities were 0.70 (95% HPD-CI: 0.46, 0.93) and 0.77 (0.61, 0.93), respectively. The posterior mean transition probabilities were 0.0023 (0.0018, 0.0027) and 0.0031 (0.0024, 0.0038), while the posterior mean sojourn times were 2.55 (1.56, 4.26) years and 3.15 (2.12, 4.96) years. Bayes factors for the ratio of posterior probabilities that the respective parameter was larger vs. smaller in the 50-59-year-old cohort were estimated to be 2.09, 40.8 and 3.0 for the sensitivity, transition probability, and mean sojourn time, respectively. All three Bayes factors were larger than two, indicating greater than 2:1 odds in favor of the hypothesis that each of these parameters was greater in the 50-59-year-old cohort. The estimated mean lead times were 0.83 years and 0.96 years if the two cohorts were screened annually. CONCLUSIONS The increase in sensitivity corresponds to an increase in the mean sojourn time. Breast cancer in younger women is more difficult to detect by screening tests and is more aggressive than breast cancer in older women. Women aged 50-59 tend to benefit more from screening compared with women aged 40-49.

A projection of benefits due to fecal occult blood test for colorectal cancer.

OBJECTIVES A prospective study to estimate benefits due to fecal occult blood tests for colorectal cancer are carried out for both males and females, under different screening frequencies. METHODS We apply the statistical method developed by Wu et al. (2007) [1] using the Minnesota colorectal cancer study group data, to make Bayesian inference for the lead time, the time of diagnosis advanced by screening for both male and female participants in a periodic screening program. The lead time is distributed as a mixture of a point mass at zero and a piecewise continuous distribution. The two parts of the mixture correspond to two aspects of the screening: the probability of no benefit, or the percentage of interval cases; and the probability distribution of the early diagnosis time. We present estimates of these two measures for males and females by simulation studies using the Minnesota study group data. We also provide the mean, mode, variance, and density curve of the programs lead time by gender. This may provide policy makers important information on the effectiveness of the FOBT screening in colorectal cancer early detection. RESULTS The mean lead time increases as the screening time interval decreases for both males and females. The standard error of the lead time also increases as the screening time interval decreases for both genders. Females seem get more benefit than males, in that females usually have a longer lead time than males if both take the test at the same time interval and the lead time mode for females is greater than that of males in the same screening time interval. CONCLUSION According to the predictive estimation of the lead time distribution, to guarantee a 90% chance of early detection, it seems necessary for the males to take the fecal occult blood test every 9 months, while the females can take it annually.

A comparison of per sample global scaling and per gene normalization methods for differential expression analysis of RNA-seq data

Normalization is an essential step with considerable impact on high-throughput RNA sequencing (RNA-seq) data analysis. Although there are numerous methods for read count normalization, it remains a challenge to choose an optimal method due to multiple factors contributing to read count variability that affects the overall sensitivity and specificity. In order to properly determine the most appropriate normalization methods, it is critical to compare the performance and shortcomings of a representative set of normalization routines based on different dataset characteristics. Therefore, we set out to evaluate the performance of the commonly used methods (DESeq, TMM-edgeR, FPKM-CuffDiff, TC, Med UQ and FQ) and two new methods we propose: Med-pgQ2 and UQ-pgQ2 (per-gene normalization after per-sample median or upper-quartile global scaling). Our per-gene normalization approach allows for comparisons between conditions based on similar count levels. Using the benchmark Microarray Quality Control Project (MAQC) and simulated datasets, we performed differential gene expression analysis to evaluate these methods. When evaluating MAQC2 with two replicates, we observed that Med-pgQ2 and UQ-pgQ2 achieved a slightly higher area under the Receiver Operating Characteristic Curve (AUC), a specificity rate > 85%, the detection power > 92% and an actual false discovery rate (FDR) under 0.06 given the nominal FDR (≤0.05). Although the top commonly used methods (DESeq and TMM-edgeR) yield a higher power (>93%) for MAQC2 data, they trade off with a reduced specificity (<70%) and a slightly higher actual FDR than our proposed methods. In addition, the results from an analysis based on the qualitative characteristics of sample distribution for MAQC2 and human breast cancer datasets show that only our gene-wise normalization methods corrected data skewed towards lower read counts. However, when we evaluated MAQC3 with less variation in five replicates, all methods performed similarly. Thus, our proposed Med-pgQ2 and UQ-pgQ2 methods perform slightly better for differential gene analysis of RNA-seq data skewed towards lowly expressed read counts with high variation by improving specificity while maintaining a good detection power with a control of the nominal FDR level.

The lead time distribution when lifetime is subject to competing risks in cancer screening.

This paper extends the previous probability model for the distribution of lead time in periodic cancer screening exams, namely, in that the lifetime T is treated as a random variable, instead of a fixed value. Hence the number of screens for a given individual is a random variable as well. We use the actuarial life table from the Social Security Administration to obtain the lifetime distribution, and then use this information to project the lead time distribution for someone with a future screening schedule. Simulation studies using the HIP study group data provide estimates of the lead time under different screening frequencies. The projected lead time has two components: a point mass at zero (corresponding to interval cases detected between screening exams) and a continuous probability density. We present estimates of the projected lead time for participants in a breast cancer screening program. The model is more realistic and can inform optimal screening frequency. This study focuses on breast cancer screening, but is applicable to other kinds of cancer screening also.

Classification of clinical outcomes using high-throughput informatics: Part 1 - nonparametric method reviews

It is widely recognized that many cancer therapies are effective only for a subset of patients. However clinical studies are most often powered to detect an overall treatment effect. To address this issue, classification methods are increasingly being used to predict a subset of patients which respond differently to treatment. This study begins with a brief history of classifi- cation methods with an emphasis on applications involving melanoma. Nonparametric methods suitable for predicting subsets of patients responding differently to treatment are then reviewed. Each method has different ways of incorporating continuous, categorical, clinical and high-throughput covariates. More recent methods have built-in dimension reduction methods for high throughput data. Pre-validation is one method of assessing the added value of high-throughput data to clinical covariates. The way in which treatment interactions are incorporated is important if the goal is to predict a subset of patients which respond differently to treatment. For nonparametric methods, distance measures specific to the method are used to make classification decisions. Approaches are outlined which employ these distances to measure treatment interactions. It is hoped that this study will stimulate more development of nonparametric methods to predict subsets of patients responding differently to treatment.

Bayesian lead time estimation for the Johns Hopkins Lung Project data.

Problem statement: Lung cancer screening using X-rays has been controversial for many years. A major concern is whether lung cancer screening really brings any survival benefits, which depends on effective treatment after early detection. The problem was analyzed from a different point of view and estimates were presented of the projected lead time for participants in a lung cancer screening program using the Johns Hopkins Lung Project (JHLP) data. Method: The newly developed method of lead time estimation was applied where the lifetime T was treated as a random variable rather than a fixed value, resulting in the number of future screenings for a given individual is a random variable. Using the actuarial life table available from the United States Social Security Administration, the lifetime distribution was first obtained, then the lead time distribution was projected using the JHLP data. Results: The data analysis with the JHLP data shows that, for a male heavy smoker with initial screening ages at 50, 60, and 70, the probability of no-early-detection with semiannual screens will be 32.16%, 32.45%, and 33.17%, respectively; while the mean lead time is 1.36, 1.33 and 1.23 years. The probability of no-early-detection increases monotonically when the screening interval increases, and it increases slightly as the initial age increases for the same screening interval. The mean lead time and its standard error decrease when the screening interval increases for all age groups, and both decrease when initial age increases with the same screening interval. Conclusion: The overall mean lead time estimated with a random lifetime T is slightly less than that with a fixed value of T. This result is hoped to be of benefit to improve current screening programs.

Efficacy of Dual Lung Cancer Screening by Chest X-Ray and SputumCytology Using Johns Hopkins Lung Project Data

We investigate the efficacy of the dual-screening, annual chest X-ray and 4-monthly sputum cytology, for lung cancer detection using the Johns Hopkins Lung Project (JHLP) data. An advanced statistical approach to testing the dependency of dual diagnostic tests, chest X-ray and sputum cytology is applied for lung cancer detection. To achieve this, the screening sensitivities of X-ray only, cytology only, and dual-screening, the overall transition probability, the sojourn time, and the mean lead time were estimated from JHLP data. A derivative-free global optimization method, Particle Swarm Optimization (PSO), was further employed in the procedure to avoid being stuck at a local optimum or the boundary. Our analysis shows that the overall sensitivity of lung cancer screening is 85.34% with chest X-ray of 79.93% and 4-monthly sputum cytology of 26.98%, resulting in a small correlation coefficient of -0.0039, which is not significantly different from zero. As a result, the dual procedure improves the overall screening sensitivity up to ~5% in lung cancer detection.

Evaluating Long-Term Outcomes via Computed Tomography in LungCancer Screening

Objectives: Future outcomes of computed tomography in lung cancer screening were evaluated using recently derived probability formula in the disease progressive model, and the recently completed National Lung Screening Trial computed tomography (NLST-CT) data. Methods: Every participant in a screening program would fall into one of the four disjoint groups eventually: symptom-free-life, no-early-detection, true-early-detection and overdiagnosis, depending on whether he/she would be diagnosed with cancer and whether symptoms would have appeared before death. The probability of each outcome was a function of an individual’s current age, past and future screening frequency and the three key parameters: screening sensitivity, sojourn time and time in the disease-free state. The predictive probability was estimated for people with and without screening histories. Percentage of over-diagnosis among the screen-detected cases was also presented with human lifetime as a random variable. Results: The probability of heavy smokers to live a lung-cancer-free life would depend on their current age; it was about 80%, 86% and 94% for the 60, 70, and 80 years old respectively. The probabilities of no-early detection and true-early-detection were determined by the future screening interval and the current age: the probability of no-earlydetection would increase to about three times if the future screening interval changes from annual to biennial; while the probability of true-early-detection would decrease to about 75% if the future screening interval changes from annual to biennial. The probability of over-diagnosis among the screen-detected was increasing as people aging: ~3%, 5% and 9% for the 60, 70, and 80 years old correspondingly; this probability decreases slightly when the historic screening interval increases. Conclusion: This research provided the estimated probabilities of the four outcomes in the future and the percentage of overdiagnosis among the screen-detected cases. It provided a practical approach on the evaluation of long-term outcomes via CT in lung cancer screening.