Explore the vast range of titles available.

Background Cancer surveillance researchers analyze incidence or mortality rates jointly indexed by age group and calendar period using age-period-cohort models. Many studies consider age- and period-specific rates in two or more strata defined by sex, race/ethnicity, etc. A comprehensive characterization of trends and patterns within each stratum can be obtained using age-period-cohort (APC) estimable functions (EF). However, currently available approaches for joint analysis and synthesis of EF are limited. Methods We develop a new method called Comparative Age-Period-Cohort Analysis to quantify similarities and differences of EF across strata. Comparative Analysis identifies whether the stratum-specific hazard rates are proportional by age, period, or cohort. Results Proportionality imposes natural constraints on the EF that can be exploited to gain efficiency and simplify the interpretation of the data. Comparative Analysis can also identify differences or diversity in proportional relationships between subsets of strata (“pattern heterogeneity”). We present three examples using cancer incidence from the United States Surveillance, Epidemiology, and End Results Program: non-malignant meningioma by sex; multiple myeloma among men stratified by race/ethnicity; and in situ melanoma by anatomic site among white women. Conclusions For studies of cancer rates with from two through to around 10 strata, which covers many outstanding questions in cancer surveillance research, our new method provides a comprehensive, coherent, and reproducible approach for joint analysis and synthesis of age-period-cohort estimable functions.

Reduction of premature mortality is a UN Sustainable Development Goal. Unlike other high-income countries, age-adjusted mortality in the USA plateaued in 2010 and increased slightly in 2015, possibly because of rising premature mortality. We aimed to analyse trends in mortality in the USA between 1999 and 2014 in people aged 25–64 years by age group, sex, and race and ethnicity, and to identify specific causes of death underlying the temporal trends. For this analysis, we used cause-of-death and demographic data from death certificates from the US National Center for Health Statistics, and population estimates from the US Census Bureau. We estimated annual percentage changes in mortality using age-period-cohort models. Age-standardised excess deaths were estimated for 2000 to 2014 as observed deaths minus expected deaths (estimated from 1999 mortality rates). Between 1999 and 2014, premature mortality increased in white individuals and in American Indians and Alaska Natives. Increases were highest in women and those aged 25–30 years. Among 30-year-olds, annual mortality increases were 2·3% (95% CI 2·1–2·4) for white women, 0·6% (0·5–0·7) for white men, and 4·3% (3·5–5·0) and 1·9% (1·3–2·5), respectively, for American Indian and Alaska Native women and men. These increases were mainly attributable to accidental deaths (primarily drug poisonings), chronic liver disease and cirrhosis, and suicide. Among individuals aged 25–49 years, an estimated 111 000 excess premature deaths occurred in white individuals and 6600 in American Indians and Alaska Natives during 2000–14. By contrast, premature mortality decreased substantially across all age groups in Hispanic individuals (up to 3·2% per year), black individuals (up to 3·9% per year), and Asians and Pacific Islanders (up to 2·6% per year), mainly because of declines in HIV, cancer, and heart disease deaths, resulting in an estimated 112 000 fewer deaths in Hispanic individuals, 311 000 fewer deaths in black individuals, and 34 000 fewer deaths in Asians and Pacific Islanders aged 25–64 years. During 2011–14, American Indians and Alaska Natives had the highest premature mortality, followed by black individuals. Important public health successes, including HIV treatment and smoking cessation, have contributed to declining premature mortality in Hispanic individuals, black individuals, and Asians and Pacific Islanders. However, this progress has largely been negated in young and middle-aged (25–49 years) white individuals, and American Indians and Alaska Natives, primarily because of potentially avoidable causes such as drug poisonings, suicide, and chronic liver disease and cirrhosis. The magnitude of annual mortality increases in the USA is extremely unusual in high-income countries, and a rapid public health response is needed to avert further premature deaths. US National Cancer Institute Intramural Research Program.

Random forests are a popular type of machine learning model, which are relatively robust to overfitting, unlike some other machine learning models, and adequately capture non-linear relationships between an outcome of interest and multiple independent variables. There are relatively few adjustable hyperparameters in the standard random forest models, among them the minimum size of the terminal nodes on each tree. The usual stopping rule, as proposed by Breiman, stops tree expansion by limiting the size of the parent nodes, so that a node cannot be split if it has less than a specified number of observations. Recently an alternative stopping criterion has been proposed, stopping tree expansion so that all terminal nodes have at least a minimum number of observations. The present paper proposes three generalisations of this idea, limiting the growth in regression random forests, based on the variance, range, or inter-centile range. The new approaches are applied to diabetes data obtained from the National Health and Nutrition Examination Survey and four other datasets (Tasmanian Abalone data, Boston Housing crime rate data, Los Angeles ozone concentration data, MIT servo data). Empirical analysis presented herein demonstrate that the new stopping rules yield competitive mean square prediction error to standard random forest models. In general, use of the intercentile range statistic to control tree expansion yields much less variation in mean square prediction error, and mean square prediction error is also closer to the optimal. The Fortran code developed is provided in the Supplementary Material.

The prevalence of class III obesity (body mass index [BMI]≥40 kg/m2) has increased dramatically in several countries and currently affects 6% of adults in the US, with uncertain impact on the risks of illness and death. Using data from a large pooled study, we evaluated the risk of death, overall and due to a wide range of causes, and years of life expectancy lost associated with class III obesity. In a pooled analysis of 20 prospective studies from the United States, Sweden, and Australia, we estimated sex- and age-adjusted total and cause-specific mortality rates (deaths per 100,000 persons per year) and multivariable-adjusted hazard ratios for adults, aged 19-83 y at baseline, classified as obese class III (BMI 40.0-59.9 kg/m2) compared with those classified as normal weight (BMI 18.5-24.9 kg/m2). Participants reporting ever smoking cigarettes or a history of chronic disease (heart disease, cancer, stroke, or emphysema) on baseline questionnaires were excluded. Among 9,564 class III obesity participants, mortality rates were 856.0 in men and 663.0 in women during the study period (1976-2009). Among 304,011 normal-weight participants, rates were 346.7 and 280.5 in men and women, respectively. Deaths from heart disease contributed largely to the excess rates in the class III obesity group (rate differences = 238.9 and 132.8 in men and women, respectively), followed by deaths from cancer (rate differences = 36.7 and 62.3 in men and women, respectively) and diabetes (rate differences = 51.2 and 29.2 in men and women, respectively). Within the class III obesity range, multivariable-adjusted hazard ratios for total deaths and deaths due to heart disease, cancer, diabetes, nephritis/nephrotic syndrome/nephrosis, chronic lower respiratory disease, and influenza/pneumonia increased with increasing BMI. Compared with normal-weight BMI, a BMI of 40-44.9, 45-49.9, 50-54.9, and 55-59.9 kg/m2 was associated with an estimated 6.5 (95% CI: 5.7-7.3), 8.9 (95% CI: 7.4-10.4), 9.8 (95% CI: 7.4-12.2), and 13.7 (95% CI: 10.5-16.9) y of life lost. A limitation was that BMI was mainly ascertained by self-report. Class III obesity is associated with substantially elevated rates of total mortality, with most of the excess deaths due to heart disease, cancer, and diabetes, and major reductions in life expectancy compared with normal weight. Please see later in the article for the Editors' Summary.

Assessment of cancer incidence trends within the U.S. have mostly relied upon Surveillance, Epidemiology, and End Results (SEER) data, with implicit inference that such is representative of the general population. However, many cancer policy decisions are based at a more granular level. To help inform such, analyses of regional cancer incidence data are needed. Leveraging the unique resource of National Program of Cancer Registries (NPCR)-SEER, we assessed whether regional rates and trends of esophageal cancer significantly deviated from national estimates. From NPCR-SEER, we extracted cancer case counts and populations for whites aged 45-84 years by calendar year, histology, sex, and census region for the period 1999-2008. We calculated age-standardized incidence rates (ASRs), annual percent changes (APCs), and male-to-female incidence rate ratios (IRRs). This analysis included 65,823 esophageal adenocarcinomas and 27,094 esophageal squamous cell carcinomas diagnosed during 778 million person-years. We observed significant geographic variability in incidence rates and trends, especially for esophageal adenocarcinomas in males: ASRs were highest in the Northeast (17.7 per 100,000) and Midwest (18.1). Both were significantly higher than the national estimate (16.0). In addition, the Northeast APC was 62% higher than the national estimate (3.19% vs. 1.97%). Lastly, IRRs remained fairly constant across calendar time, despite changes in incidence rates. Significant regional variations in esophageal cancer incidence trends exist in the U.S. Stable IRRs may indicate the predominant factors affecting incidence rates are similar in men and women.

There are striking differences in breast cancer incidence between Asian and western women. Rates vary substantially within Asia also, with Mongolia's even lower than China's. These profound differences have been speculated to be due in part to diet, mediated by circulating hormone concentrations. Sex steroid hormone concentrations were measured in women living in Ulaanbaatar, Mongolia and the United Kingdom (U.K.). Diet was obtained by interview and national survey data. Mean hormone differences were compared by country, and systematic variation by number of days since last menstrual period was modeled and adjusted for age and parity; difference in overall area under the curves was assessed. The diet in Mongolia was higher in meat and dairy than in the U.K. Mean testosterone concentrations were 18.5% lower (p<0.0001) while estradiol concentrations were 19.1% higher (p = 0.02) in Mongolian than British women, adjusted for age and parity. Progesterone was almost 50% higher in Mongolian women (p = 0.04), particularly during the follicular phase and early luteal surge. Hormone concentrations generally were similar in Mongolian women born in Ulaanbaatar compared with those born in rural areas, although there was a decreasing progesterone trend by degree of westernization (rural Mongolia; urban Mongolia; U.K.). Mean hormone differences were similar when restricted to parous women, and with further adjustment for body mass index, height, and smoking status. These data augment accumulating evidence that circulating estrogens are unlikely to explain reduced breast cancer rates in Asia compared with the west, and suggest casting a wider net with respect to biomarkers. Lower testosterone and higher progesterone in Mongolian women raise the possibility that these hormones may be important to consider. In addition, the almost exclusive dietary reliance of Mongolians on meat and dairy argues against beneficial effects of a low-fat diet on circulating hormones explaining international breast cancer differences.

Epidemiologic studies are reporting associations between lead exposure and human cancers. A polymorphism in the 5-aminolevulinic acid dehydratase (ALAD) gene affects lead toxicokinetics and may modify the adverse effects of lead. The objective of this study was to evaluate single-nucleotide polymorphisms (SNPs) tagging the ALAD region among renal cancer cases and controls to determine whether genetic variation alters the relationship between lead and renal cancer. Occupational exposure to lead and risk of cancer was examined in a case-control study of renal cell carcinoma (RCC). Comprehensive analysis of variation across the ALAD gene was assessed using a tagging SNP approach among 987 cases and 1298 controls. Occupational lead exposure was estimated using questionnaire-based exposure assessment and expert review. Odds ratios (OR) and 95% confidence intervals (CI) were calculated using logistic regression. The adjusted risk associated with the ALAD variant rs8177796(CT/TT) was increased (OR = 1.35, 95%CI = 1.05-1.73, p-value = 0.02) when compared to the major allele, regardless of lead exposure. Joint effects of lead and ALAD rs2761016 suggest an increased RCC risk for the homozygous wild-type and heterozygous alleles ((GG)OR = 2.68, 95%CI = 1.17-6.12, p = 0.01; (GA)OR = 1.79, 95%CI = 1.06-3.04 with an interaction approaching significance (p(int) = 0.06). No significant modification in RCC risk was observed for the functional variant rs1800435(K68N). Haplotype analysis identified a region associated with risk supporting tagging SNP results. A common genetic variation in ALAD may alter the risk of RCC overall, and among individuals occupationally exposed to lead. Further work in larger exposed populations is warranted to determine if ALAD modifies RCC risk associated with lead exposure.

Breast cancer incidence rates vary according to estrogen receptor expression (ER) and histopathology. We hypothesized that annual mortality rates from breast cancer after initial diagnosis (hazard rates) might also vary by ER and histopathology. We accessioned the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER, 1992-2002) program to estimate hazard rates according to ER (positive and negative) and histopathology (duct, tubular, lobular, medullary, inflammatory, papillary, and mucinous types). We used spline functions to model hazard rates free of strongly parametric assumptions for ER negative and positive cases overall and by histopathology. Hazard rates for ER negative and ER positive cases were distinct and non-proportional. At 17 months, ER negative hazard rates peaked at 7.5% per year (95% CI, 7.3-7.8% per year) then declined, whereas ER positive hazard rates lacked a sharp early peak and were comparatively constant at 1.5-2% per year. Falling ER negative and constant ER positive hazard rates crossed at 7 years; after which, prognosis was better for ER negative cases. Among ER positive and negative cases, there were proportional and non-proportional hazards according to histopathologic type, but the two basic ER-associated patterns were maintained. Hazard rates differed quantitatively and qualitatively according to ER and histopathology. These large-scale population-based results seem consistent with genomic studies, demonstrating two main classes of breast cancers with distinct prognoses according to ER expression.

Folate and one-carbon metabolism are linked to cancer risk through their integral role in DNA synthesis and methylation. Variation in one-carbon metabolism genes, particularly MTHFR, has been associated with risk of a number of cancers in epidemiologic studies, but little is known regarding renal cancer. Tag single nucleotide polymorphisms (SNPs) selected to produce high genomic coverage of 13 gene regions of one-carbon metabolism (ALDH1L1, BHMT, CBS, FOLR1, MTHFR, MTR, MTRR, SHMT1, SLC19A1, TYMS) and the closely associated glutathione synthesis pathway (CTH, GGH, GSS) were genotyped for 777 renal cell carcinoma (RCC) cases and 1,035 controls in the Central and Eastern European Renal Cancer case-control study. Associations of individual SNPs (n = 163) with RCC risk were calculated using unconditional logistic regression adjusted for age, sex and study center. Minimum p-value permutation (Min-P) tests were used to identify gene regions associated with risk, and haplotypes were evaluated within these genes. The strongest associations with RCC risk were observed for SLC19A1 (P(min-P) = 0.03) and MTHFR (P(min-P) = 0.13). A haplotype consisting of four SNPs in SLC19A1 (rs12483553, rs2838950, rs2838951, and rs17004785) was associated with a 37% increased risk (p = 0.02), and exploratory stratified analysis suggested the association was only significant among those in the lowest tertile of vegetable intake. To our knowledge, this is the first study to comprehensively examine variation in one-carbon metabolism genes in relation to RCC risk. We identified a novel association with SLC19A1, which is important for transport of folate into cells. Replication in other populations is required to confirm these findings.

BackgroundCostello syndrome (CS) is a cancer-predisposition disorder caused by germline pathogenic variants in HRAS. We conducted a systematic review using case reports and case series to characterise cancer risk in CS.MethodsWe conducted a systematic review to identify CS cases to create a retrospective cohort. We tested genotype–phenotype correlations and calculated cumulative incidence and hazard rates (HR) for cancer and cancer-free death, standardised incidence rates (SIR) and survival after cancer.ResultsThis study includes 234 publications reporting 621 patients from 35 countries. Over nine percent had cancer, including rhabdomyosarcoma, bladder, and neuroblastoma. The rate of cancer and death associated with p.Gly12Ser were lower when compared to all other variants (P < 0.05). Higher mortality for p.Gly12Cys, p.Gly12Asp, p.Gly12Val and p.Gly60Val and higher malignancy rate for p.Gly12Ala were confirmed (P < 0.05). Cumulative incidence by age 20 was 13% (cancer) and 11% (cancer-free death). HR (death) was 3–4% until age 3. Statistically significant SIRs were found for rhabdomyosarcoma (SIR = 1240), bladder (SIR = 1971), and neuroblastoma (SIR = 60). Survival after cancer appeared reduced.ConclusionsThis is the largest investigation of cancer in CS to date. The high incidence and SIR values found to highlight the need for rigorous surveillance and evidence-based guidelines for this high-risk population.