Michael R. Kosorok

Nutritional Benefits of Neonatal Screening for Cystic Fibrosis

BACKGROUND Many patients with cystic fibrosis are malnourished at the time of diagnosis. Whether newborn screening and early treatment may prevent the development of a nutritional deficiency is not known. METHODS We compared the nutritional status of patients with cystic fibrosis identified by neonatal screening or by standard diagnostic methods. A total of 650,341 newborn infants were screened by measuring immunoreactive trypsinogen on dried blood spots (from April 1985 through June 1991) or by combining the trypsinogen test with DNA analysis (from July 1991 through June 1994). Of 325,171 infants assigned to an early-diagnosis group, cystic fibrosis was diagnosed in 74 infants, including 5 with negative screening tests. Excluding infants with meconium ileus, we evaluated nutritional status for up to 10 years by anthropometric and biochemical methods in 56 of the infants who received an early diagnosis and in 40 of the infants in whom the diagnosis was made by standard methods (the control group). Pancreatic insufficiency was managed with nutritional interventions that included high-calorie diets, pancreatic-enzyme therapy, and fat-soluble vitamin supplements. RESULTS The diagnosis of cystic fibrosis was confirmed by a positive sweat test at a younger age in the early-diagnosis group than in the control group (mean age, 12 vs. 72 weeks). At the time of diagnosis, the early-diagnosis group had significantly higher height and weight percentiles and a higher head-circumference percentile (52nd, vs. 32nd in the control group; P=0.003). The early-diagnosis group also had significantly higher anthropometric indexes during the follow-up period, especially the children with pancreatic insufficiency and those who were homozygous for the deltaF508 mutation. CONCLUSIONS Neonatal screening provides the opportunity to prevent malnutrition in infants with cystic fibrosis.

Acceleration of lung disease in children with cystic fibrosis after Pseudomonas aeruginosa acquisition

As part of the ongoing Wisconsin Cystic Fibrosis (CF) Neonatal Screening Project, we had the unique opportunity to study the longitudinal relationship between Pseudomonas aeruginosa (Pa) acquisition and infection and developing lung disease in children with CF. The primary objective was to determine whether acquisition of Pa was associated with a measurable change in the progression of lung disease. Two outcome measures were used to study 56 patients who were diagnosed through newborn screening: 1) Wisconsin additive chest radiograph score (WCXR), based on the average of scores from a pulmonologist and a radiologist, and 2) the highest forced expired volume in 1 sec (FEV1)/forced vital capacity (FVC) ratio. We used two measures of Pa acquisition: 1) time of first positive protocol‐determined oropharyngeal (with cough) culture, and 2) the magnitude of antibody titer detected by ELISA assays, using as antigen a crude cell lysate, purified exotoxin A, or an elastase toxoid prepared from three Pa strains. Other predictor variables included age, pancreatic status, height‐for age, and weight‐for‐age‐percentiles.

Estimating Individualized Treatment Rules Using Outcome Weighted Learning

There is increasing interest in discovering individualized treatment rules (ITRs) for patients who have heterogeneous responses to treatment. In particular, one aims to find an optimal ITR that is a deterministic function of patient-specific characteristics maximizing expected clinical outcome. In this article, we first show that estimating such an optimal treatment rule is equivalent to a classification problem where each subject is weighted proportional to his or her clinical outcome. We then propose an outcome weighted learning approach based on the support vector machine framework. We show that the resulting estimator of the treatment rule is consistent. We further obtain a finite sample bound for the difference between the expected outcome using the estimated ITR and that of the optimal treatment rule. The performance of the proposed approach is demonstrated via simulation studies and an analysis of chronic depression data.

Cardiac Resynchronization Therapy Reduces the Risk of Hospitalizations in Patients With Advanced Heart Failure Results From the Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) Trial

Background— In the Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) trial, 1520 patients with advanced heart failure were assigned in a 1:2:2 ratio to optimal pharmacological therapy or optimal pharmacological therapy plus cardiac resynchronization therapy (CRT-P) or CRT with defibrillator (CRT-D). Use of CRT-P and CRT-D was associated with a significant reduction in combined risk of death or all-cause hospitalizations. Because mortality also was significantly reduced (optimal pharmacological therapy versus CRT-D only), an assessment of the true reduction in hospitalization rates must consider the competing risk of death and varying follow-up times. Methods and Results— To overcome the challenges of comparing treatment groups, we used a nonparametric test of right-censored recurrent events that accounts for multiple hospital admissions, differential follow-up time between treatment groups, and death as a competing risk. An end-point committee adjudicated and classified all hospitalizations. Compared with optimal pharmacological therapy, CRT-P and CRT-D were associated with a 21% and 25% reduction in all-cause, 34% and 37% reduction in cardiac, and 44% and 41% reduction in heart failure hospital admissions per patient-year of follow-up, respectively. Similar reductions were seen in hospitalization days per patient-year. The reduction in hospitalization rate for heart failure in the CRT groups appeared within days of randomization and remained sustained. Noncardiac hospitalization rates were not different between groups. Conclusion— Use of CRT with or without a defibrillator in advanced heart failure patients was associated with marked reductions in all-cause, cardiac, and heart failure hospitalization rates in an analysis that accounted for the competing risk of mortality and unequal follow-up time.

Risk of persistent growth impairment after alternate-day prednisone treatment in children with cystic fibrosis

BACKGROUND It is uncertain whether the growth impairment that occurs in children during long-term treatment with glucocorticoids persists after the medication is discontinued and ultimately affects adult height. METHODS We evaluated growth six to seven years after alternate-day treatment with prednisone had been discontinued in 224 children 6 to 14 years of age with cystic fibrosis who had participated in a multicenter trial of this therapy from 1986 through 1991. Of the children, 151 had been randomly assigned to receive prednisone (either 1 or 2 mg per kilogram of body weight) and 73 to receive placebo. We obtained data on growth up to 1997 from the Cystic Fibrosis Foundation Patient Registry and standardized the data to sex- and age-specific norms from the National Center for Health Statistics. We used z scores to compare growth patterns among treatment groups. RESULTS In 1997, 68 percent of the patients were 18 years of age or older. The z scores for height declined during prednisone therapy; catch-up growth began two years after treatment with prednisone was discontinued. Among the boys, the z scores for height in those treated with prednisone remained lower than the scores for those who received placebo (P=0.02). The mean heights for boys 18 years of age or older were 4 cm less in the prednisone groups than in the placebo group, an equivalent of 13 percentile points (P=0.03). Among the girls, differences in height between those who were treated with prednisone and those who received placebo were no longer present two to three years after prednisone therapy was discontinued. CONCLUSIONS Among children with cystic fibrosis who have received alternate-day treatment with prednisone, boys, but not girls, have persistent growth impairment after treatment is discontinued.

Relationships among specific viral pathogens, virus-induced interleukin-8, and respiratory symptoms in infancy.

Both virus‐mediated damage to airway tissues and induction of pro‐inflammatory cytokines such as interleukin‐8 (IL‐8) could contribute to symptom severity during viral respiratory infections in children. To test the hypothesis that IL‐8 contributes to the pathogenesis of respiratory symptoms during naturally acquired respiratory viral infections in children, nasal wash samples collected from infants with acute viral infections (n = 198) or from healthy uninfected infants (n = 31) were analysed for IL‐8. Nasal wash IL‐8 was positively related to age in uninfected children (rs = 0.36, p < 0.05). Respiratory syncytial virus (RSV) infection caused more severe respiratory symptoms compared to infections with influenza A, parainfluenza viruses, or rhinoviruses. In addition, RSV, parainfluenza and rhinovirus infections increased levels of IL‐8 in nasal lavage fluid, and there were some differences in the ability of the viruses to induce IL‐8 production (RSV>influenza, p < 0.05). Finally, there were significant correlations between nasal wash IL‐8 levels and symptom scores during infections with rhinovirus (rs = 0.56, p < 0.001) or influenza A (rs = 0.45, p < 0.05), but not with parainfluenza virus or RSV. These findings provide evidence of a close relationship between the generation of IL‐8 and symptoms during acute community‐acquired infections with rhinovirus or influenza A. In contrast, for RSV and parainfluenza infections, factors in addition to IL‐8 production appear to contribute to the generation of clinical symptoms.

Reproducibility of a scoring system for computed tomography scanning in cystic fibrosis.

Introduction Computerized tomography (CT) scanning shows promise as an outcome surrogate for cystic fibrosis (CF) lung disease progression. The scoring system used to convert the CT image to numeric data is an essential determinant of the performance of CT scanning. Methods Three radiologists independently scored 16 high-resolution CT scans performed on children in the Wisconsin CF Neonatal Screening Project. The test scans were selected to provide a broad range of disease severity. The scoring system provided subscores for the presence and severity of 5 findings of CF lung disease. The sum of the subscores provided a total score. The CT scans were then read again by each of the radiologists at least 11 months later. Using Mixed Effects Linear Model Analysis, the sources of error (scan-to-scan variation, interrater variance, and intrarater variance) were calculated. Results For the total score, the scan-to-scan variation was 14.48, interrater variance was 0.28, and intrarater variance was 0.45, with an overall reproducibility of 95%. The square root of scan-to-scan variance, a measure of sensitivity, was 3.81. Evaluation of the subscores showed higher reproducibility for bronchiectasis and hyperinflation (95% and 88%, respectively). The bronchiectasis score was more sensitive than the air-trapping score (1.46 vs. 0.89). Discussion This system was developed to provide a reproducible method that could be used to evaluate the lobar location, severity, and extent of a broad spectrum of CT features of CF lung disease, especially in children. This study demonstrates that the overall score is both sensitive to variation in the severity of lung disease and reproducible.

Restricted activity days among older adults

OBJECTIVES The number of restricted activity days experienced by an individual in the course of a year is an important measure of functional well-being, particularly for older adults. We sought to determine multivariate associations between restricted activity days and various health conditions. METHODS We used data from the 1984 Supplement on Aging of the National Health Interview Survey to estimate the relationship between restricted activity days and age, gender, and the presence or absence of selected chronic conditions and falls for all noninstitutionalized people aged 65 years and over. Chronic conditions and falls accounted for most of the variance in the model. RESULTS Of an annual average of 31 restricted activity days, 6 days were associated with falls; 4 days with heart disease; 4 days with arthritis and rheumatism; 2 days each with high blood pressure, cerebrovascular disease, and visual impairment; and 1 day each with atherosclerosis, diabetes, major malignancies, and osteoporosis. CONCLUSIONS These results can be used in estimating the potential impact of health promotion programs on the health status of noninstitutionalized older adults.

Detection of gene pathways with predictive power for breast cancer prognosis

BackgroundPrognosis is of critical interest in breast cancer research. Biomedical studies suggest that genomic measurements may have independent predictive power for prognosis. Gene profiling studies have been conducted to search for predictive genomic measurements. Genes have the inherent pathway structure, where pathways are composed of multiple genes with coordinated functions. The goal of this study is to identify gene pathways with predictive power for breast cancer prognosis. Since our goal is fundamentally different from that of existing studies, a new pathway analysis method is proposed.ResultsThe new method advances beyond existing alternatives along the following aspects. First, it can assess the predictive power of gene pathways, whereas existing methods tend to focus on model fitting accuracy only. Second, it can account for the joint effects of multiple genes in a pathway, whereas existing methods tend to focus on the marginal effects of genes. Third, it can accommodate multiple heterogeneous datasets, whereas existing methods analyze a single dataset only. We analyze four breast cancer prognosis studies and identify 97 pathways with significant predictive power for prognosis. Important pathways missed by alternative methods are identified.ConclusionsThe proposed method provides a useful alternative to existing pathway analysis methods. Identified pathways can provide further insights into breast cancer prognosis.

Reinforcement learning design for cancer clinical trials

We develop reinforcement learning trials for discovering individualized treatment regimens for life-threatening diseases such as cancer. A temporal-difference learning method called Q-learning is utilized that involves learning an optimal policy from a single training set of finite longitudinal patient trajectories. Approximating the Q-function with time-indexed parameters can be achieved by using support vector regression or extremely randomized trees. Within this framework, we demonstrate that the procedure can extract optimal strategies directly from clinical data without relying on the identification of any accurate mathematical models, unlike approaches based on adaptive design. We show that reinforcement learning has tremendous potential in clinical research because it can select actions that improve outcomes by taking into account delayed effects even when the relationship between actions and outcomes is not fully known. To support our claims, the methodologys practical utility is illustrated in a simulation analysis. In the immediate future, we will apply this general strategy to studying and identifying new treatments for advanced metastatic stage IIIB/IV non-small cell lung cancer, which usually includes multiple lines of chemotherapy treatment. Moreover, there is significant potential of the proposed methodology for developing personalized treatment strategies in other cancers, in cystic fibrosis, and in other life-threatening diseases.