Search

Supplementary Table 1. Sample Size of each participated study, by case-control status and genotype platform. Supplementary Table 2. Association between fourteen environmental factors and the risk of breast cancer. Supplementary Table 3. Interactions between genes and fourteen environmental factors.

Genome-wide association studies (GWAS) have identified more than 200 susceptibility loci for breast cancer, but these variants explain less than a fifth of the disease risk. Although gene–environment interactions have been proposed to account for some of the remaining heritability, few studies have empirically assessed this.We obtained genotype and risk factor data from 46,060 cases and 47,929 controls of European ancestry from population-based studies within the Breast Cancer Association Consortium (BCAC). We built gene expression prediction models for 4,864 genes with a significant (P < 0.01) heritable component using the transcriptome and genotype data from the Genotype-Tissue Expression (GTEx) project. We leveraged predicted gene expression information to investigate the interactions between gene-centric genetic variation and 14 established risk factors in association with breast cancer risk, using a mixed-effects score test.After adjusting for number of tests using Bonferroni correction, no interaction remained statistically significant. The strongest interaction observed was between the predicted expression of the C13orf45 gene and age at first full-term pregnancy (PGXE = 4.44 × 10−6).In this transcriptome-informed genome-wide gene–environment interaction study of breast cancer, we found no strong support for the role of gene expression in modifying the associations between established risk factors and breast cancer risk.Our study suggests a limited role of gene–environment interactions in breast cancer risk.

Funding and acknowledgements.

Quantile-Quantile plot (Q-Q plot) of the aMiSTi p-values for each set of the GxE interactions.

Glioma is an aggressive neoplasm of the brain with poorly understood etiology. A limited number of pathogens have been examined as glioma risk factors, but data from prospective studies with infection status determined before disease are lacking. Herpesviruses comprise a large family of DNA viruses that infect humans and are linked to a range of chronic diseases. We conducted a prospective evaluation of the association between antibody to six human herpesviruses and glioma risk in the Janus Serum Bank (Janus) and the Cancer Prevention Study-II (CPS-II). In Janus and CPS-II, the risk for glioma was not related to seroprevalence of herpes simplex virus-1, varicella zoster virus, or human herpes viruses 6A or 6B. In Janus, seropositivity to either the Epstein Barr virus (EBV) EA[D] or VCAp18 antigen was associated with a lower risk of glioma (ORs: 0.55 [95% CI 0.32-0.94] and 0.57 [95% CI 0.38-0.85]). This inverse association was consistent by histologic subtype and was observed for gliomas diagnosed up to two decades following antibody measurement. In Janus, seropositivity to at least one of three examined cytomegalovirus (CMV) antigens (pp150, pp52, pp28) was associated with an increased risk of nonglioblastoma (OR: 2.08 [95% CI 1.07-4.03]). This association was limited to tumors diagnosed within 12 years of antibody measurement. In summary, we report evidence of an inverse association between exposure to EBV and glioma. We further report that CMV exposure may be related to a higher likelihood of the nonglioblastoma subtype.

PURPOSE The prevalence of germline pathogenic variants (PVs) in established breast cancer predisposition genes in women in the general population over age 65 years is not well-defined. However, testing guidelines suggest that women diagnosed with breast cancer over age 65 years might have < 2.5% likelihood of a PV in a high-penetrance gene. This study aimed to establish the frequency of PVs and remaining risks of breast cancer for each gene in women over age 65 years. METHODS A total of 26,707 women over age 65 years from population-based studies (51.5% with breast cancer and 48.5% unaffected) were tested for PVs in germline predisposition gene. Frequencies of PVs and associations between PVs in each gene and breast cancer were assessed, and remaining lifetime breast cancer risks were estimated for non-Hispanic White women with PVs. RESULTS The frequency of PVs in predisposition genes was 3.18% for women with breast cancer and 1.48% for unaffected women over age 65 years. PVs in BRCA1, BRCA2, and PALB2 were found in 3.42% of women diagnosed with estrogen receptor (ER)–negative, 1.0% with ER-positive, and 3.01% with triple-negative breast cancer. Frequencies of PVs were lower among women with no first-degree relatives with breast cancer. PVs in CHEK2, PALB2, BRCA2, and BRCA1 were associated with increased risks (odds ratio = 2.9-4.0) of breast cancer. Remaining lifetime risks of breast cancer were ≥ 15% for those with PVs in BRCA1, BRCA2, and PALB2. CONCLUSION This study suggests that all women diagnosed with triple-negative breast cancer or ER-negative breast cancer should receive genetic testing and that women over age 65 years with BRCA1 and BRCA2 PVs and perhaps with PALB2 and CHEK2 PVs should be considered for magnetic resonance imaging screening.

Detailed knowledge of phylogeography is important for control of mosquito species involved in transmission of human infectious diseases.Anopheles messeaeis a geographically widespread and genetically diverse dominant vector of malaria in Eurasia. A closely related sibling species,An. daciae, was distinguished fromAn. messeaebased on a few nucleotide differences in its ribosomal DNA. However, the mechanisms of speciation and their evolutionary histories are poorly understood. Here, we performed a large-scale population genetics analysis of 3694 mosquitos from Eurasia to understand the species divergence, diversity, and population structure using the Internal Transcribed Spacer 2 of ribosomal DNA for species identification and frequencies of 11 polymorphic chromosomal inversions as genetic markers. The study revealed striking differences in the geographical distribution of the sibling species. The largest genetic differences betweenAn. messeaeandAn. daciaewere detected in the X sex chromosome suggesting that this chromosome plays a role in speciation. The frequencies of autosomal inversions differed significantly between the species, strongly supporting a restricted gene flow. The clinal variability of some inversion frequencies was revealed in both species implicating their possible involvement in climate adaptations. Statistical analysis of inversion polymorphism clearly distinguished two clusters associated with the two species and demonstrated much higher genetic diversity withinAn. messeae. Overall, the frequencies of hybrids in all locations were extremely low with the exception of several southeastern populations, where putative hybrids were abundant. Thus, the pattern of genetic differentiation implies dramatic differences in geographic distribution, population structure, and evolutionary histories of the sibling speciesAn. messeaeandAn. daciae.

BackgroundUnderstanding the evolutionary relationships between closely related taxa is important for mosquitoes that transmit human diseases. Six out of 41 dominant malaria vectors in the world belong to the Maculipennis group, which is subdivided into two North American subgroups (Freeborni and Quadrimaculatus), and one Eurasian (Maculipennis) subgroup. Although previous studies considered the Nearctic subgroups as ancestral, details about their relationship with the Palearctic subgroup, and their migration times and routes from North America to Eurasia remain controversial. The Eurasian species An. beklemishevi is currently included in the North American Quadrimaculatus subgroup adding to the uncertainties in mosquito systematics.ResultsTo reconstruct historic relationships between the North American and Eurasian mosquitoes, we conducted a phylogenomic analysis of 11 Palearctic and 2 Nearctic species based on 1271 orthologous genes using their transcriptomic or genomic sequences. The analysis indicated that the Palearctic species An. beklemishevi clusters together with other Eurasian species and represents a basal lineage among them. Also, An. beklemishevi is related more closely to An. freeborni, which inhabits the Western United States, rather than to An. quadrimaculatus, a species from the Eastern United States. The time-calibrated tree suggests a migration of mosquitoes in the Maculipennis group from North America to Eurasia about 20-25 million years ago through the Bering Land Bridge. A Hybridcheck analysis demonstrated highly significant signatures of introgression events between allopatric species An. labranchiae and An. beklemishevi. The analysis also identified ancestral introgression events between An. sacharovi and its Nearctic relative An. freeborni despite their current geographic isolation.ConclusionsOur phylogenomic analyses reveal migration routes and adaptive radiation timing of Holarctic malaria vectors and strongly support inclusion of An. beklemishevi into the Maculipennis subgroup. The vectorial capacity and the ability to diapause during winter evolved multiple times in Maculipennis evolution. Detailed knowledge of the evolutionary history in the Maculipennis subgroup will help us better understand the current and future patterns of malaria transmission in Eurasia.

BACKGROUND: Population-based estimates of the risk of breast cancer associated with germline pathogenic variants in cancer-predisposition genes are critically needed for risk assessment and management in women with inherited pathogenic variants. METHODS: In a population-based case–control study, we performed sequencing using a custom multigene amplicon-based panel to identify germline pathogenic variants in 28 cancer-predisposition genes among 32,247 women with breast cancer (case patients) and 32,544 unaffected women (controls) from population-based studies in the Cancer Risk Estimates Related to Susceptibility (CARRIERS) consortium. Associations between pathogenic variants in each gene and the risk of breast cancer were assessed. RESULTS: Pathogenic variants in 12 established breast cancer–predisposition genes were detected in 5.03% of case patients and in 1.63% of controls. Pathogenic variants in BRCA1 and BRCA2 were associated with a high risk of breast cancer, with odds ratios of 7.62 (95% confidence interval [CI], 5.33 to 11.27) and 5.23 (95% CI, 4.09 to 6.77), respectively. Pathogenic variants in PALB2 were associated with a moderate risk (odds ratio, 3.83; 95% CI, 2.68 to 5.63). Pathogenic variants in BARD1, RAD51C, and RAD51D were associated with increased risks of estrogen receptor–negative breast cancer and triple-negative breast cancer, whereas pathogenic variants in ATM, CDH1, and CHEK2 were associated with an increased risk of estrogen receptor–positive breast cancer. Pathogenic variants in 16 candidate breast cancer–predisposition genes, including the c.657_661del5 founder pathogenic variant in NBN, were not associated with an increased risk of breast cancer. CONCLUSIONS: This study provides estimates of the prevalence and risk of breast cancer associated with pathogenic variants in known breast cancer–predisposition genes in the U.S. population. These estimates can inform cancer testing and screening and improve clinical management strategies for women in the general population with inherited pathogenic variants in these genes. (Funded by the National Institutes of Health and the Breast Cancer Research Foundation.)

Genome-wide association studies (GWAS) have identified more than 200 susceptibility loci for breast cancer, but these variants explain less than a fifth of the disease risk. Although gene–environment interactions have been proposed to account for some of the remaining heritability, few studies have empirically assessed this. We obtained genotype and risk factor data from 46,060 cases and 47,929 controls of European ancestry from population-based studies within the Breast Cancer Association Consortium (BCAC). We built gene expression prediction models for 4,864 genes with a significant (P < 0.01) heritable component using the transcriptome and genotype data from the Genotype-Tissue Expression (GTEx) project. We leveraged predicted gene expression information to investigate the interactions between gene-centric genetic variation and 14 established risk factors in association with breast cancer risk, using a mixed-effects score test. After adjusting for number of tests using Bonferroni correction, no interaction remained statistically significant. The strongest interaction observed was between the predicted expression of the C13orf45 gene and age at first full-term pregnancy (PGXE = 4.44 × 10−6). In this transcriptome-informed genome-wide gene–environment interaction study of breast cancer, we found no strong support for the role of gene expression in modifying the associations between established risk factors and breast cancer risk. Our study suggests a limited role of gene–environment interactions in breast cancer risk.

Cadmium and lead are persistent environmental toxins that are known or probable carcinogens, based on evidence for causality for nonhematologic cancers. Associations of these metals with risk of non-Hodgkin lymphoma (NHL) and multiple myeloma (MM) are unknown but biologically plausible. To examine the associations of circulating levels of lead and cadmium exposure with risk of B-cell NHL (B-NHL) and multiple myeloma, we conducted a nested case-control study among 299 incident B-cell NHLs and 76 MM cases within the Cancer Prevention Study-II Nutrition Cohort (CPS-II NC). Each case was incidence-density matched to two eligible controls on age, race, sex and blood draw date. Conditional logistic regression was used to estimate relative risks (RR) and 95% confidence intervals (CI) for lymphoid malignancies overall and stratified by subtype. We observed a significant positive association between high erythrocyte lead concentration and risk of lymphoid malignancies overall (RR = 1.16, 95% CI: 1.02-1.33 per 17.6 μg/L (1 standard deviation [SD])) and follicular lymphoma in particular (RR = 1.80, 95% CI: 1.15-2.80 per SD). In contrast, there was no association between erythrocyte cadmium and risk of B-NHL (RR = 0.89, 95% CI: 0.75-1.06 per 0.37 μg/L [1 SD]), or any B-NHL subtypes; but a strong inverse association with MM risk (RR = 0.59, 95% CI: 0.38-0.89, per SD). Results from our study suggest a positive association between erythrocyte lead level and risk of lymphoid malignancies and a possible inverse association between cadmium and myeloma. Additional research is needed to confirm and further explore these findings.

Nighttime light exposure may increase cancer risk by disrupting the circadian system. However, there is no well-established survey method for measuring ambient light. In the Cancer Prevention Study-3, 732 men and women answered a light survey based on seven environments. The light environment in the past year was assessed twice, one year apart, and four one-week diaries were collected between the annual surveys. A total of 170 participants wore a meter to measure photopic illuminance and circadian stimulus (CS). Illuminance and CS values were estimated for lighting environments from measured values and evaluated with a cross validation approach. The kappas for self-reported light environment comparing the two annual surveys were 0.61 on workdays and 0.49 on non-workdays. Kappas comparing the annual survey to weekly diaries were 0.71 and 0.57 for work and non-workdays, respectively. Agreement was highest for reporting of darkness (95.3%), non-residential light (86.5%), and household light (75.6%) on workdays. Measured illuminance and CS identified three peaks of light (darkness, indoor lighting, and outdoor daytime light). Estimated illuminance and CS were correlated with the measured values overall (r = 0.77 and r = 0.67, respectively) but were less correlated within each light environment (r = 0.23–0.43). The survey has good validity to assess ambient light for studies of human health.

Haldane’s rule of speciation states that sterility or inviability affects the heterogametic sex of inter-species hybrids. Darwin’s corollary to Haldane’s rule implies that there are asymmetric phenotypes in inter-species hybrids from reciprocal crosses. Studying the phenotypes of F1 hybrids among closely related species of malaria mosquitoes can assist researchers in identifying the genetic factors and molecular mechanisms of speciation. To characterize phenotypes of sterile hybrid males in the Anopheles gambiae complex, we performed crosses between laboratory strains of An. merus and either An. gambiae or An. coluzzii. The reproductive tracts had normal external morphology in hybrid males from crosses between female An. merus and male An. gambiae or An. coluzzii. Despite being sterile, these males could copulate with females for a normal period of time and could transfer a mating plug to induce female oviposition and monogamy. In contrast, the entire reproductive tracts in hybrid males from crosses between female An. gambiae or An. coluzzii and male An. merus were severely underdeveloped. These males had atrophic testes and reduced somatic organs of the reproductive system including male accessary glands and ejaculatory duct. In addition, hybrid males with underdeveloped reproductive tracts displayed a shorter copulation time with females and failed to induce female oviposition and monogamy due to their inability to form and transfer a plug to females during mating. The asymmetry of the phenotypes associated with hybrid male sterility suggests that different genetic factors and molecular mechanisms are responsible for reproductive isolation in reciprocal crosses among species of the An. gambiae complex.

INTRODUCTION Several urban neighborhoods in Philadelphia, Pennsylvania, have a history of soil, household lead paint, and potential lead-emitting industry contamination. OBJECTIVES To (1) describe blood lead levels (BLLs) in target neighborhoods, (2) identify risk factors and sources of lead exposure, (3) describe household environmental lead levels, and (4) compare results with existing data. METHODS A simple, random, cross-sectional sampling strategy was used to enroll children 8 years or younger living in selected Philadelphia neighborhoods with a history of lead-emitting industry during July 2014. Geometric mean of child BLLs and prevalence of BLLs of 5 μg/dL or more were calculated. Linear and logistic regression analyses were used to ascertain risk factors for elevated BLLs. RESULTS Among 104 children tested for blood lead, 13 (12.4%; 95% confidence interval [CI], 7.5-20.2) had BLLs of 5 μg/dL or more. The geometric mean BLL was 2.0 μg/dL (95% CI, 1.7-2.3 μg/dL). Higher geometric mean BLLs were significantly associated with front door entryway dust lead content, residence built prior to 1900, and a child currently or ever receiving Medicaid. Seventy-one percent of households exceeded the screening level for soil, 25% had an elevated front door floor dust lead level, 28% had an elevated child play area floor dust lead level, and 14% had an elevated interior window dust lead level. Children in households with 2 to 3 elevated environmental lead samples were more likely to have BLLs of 5 μg/dL or more. A spatial relationship between household proximity to historic lead-emitting facilities and child BLL was not identified. CONCLUSION Entryway floor dust lead levels were strongly associated with blood lead levels in participants. Results underscore the importance to make housing lead safe by addressing all lead hazards in and around the home. Reduction of child lead exposure is crucial, and continued blood lead surveillance, testing, and inspection of homes of children with BLLs of 5 μg/dL or more to identify and control lead sources are recommended. Pediatric health care providers can be especially vigilant screening Medicaid-eligible/enrolled children and children living in very old housing.

Glioma is an aggressive primary tumor of the brain with a poorly understood etiology. We studied the association of 4 human polyomaviruses (HPyV)—JC virus (JCV), BK virus (BKV), human polyomavirus 6 (HPyV6), and Merkel cell polyomavirus (MCPyV) with glioma risk within the Cancer Prevention Study II in the US (CPS-II) and the Janus Serum Bank in Norway. Cohort participants subsequently diagnosed with glioma from the CPS-II (n = 37) and Janus Serum Bank (n = 323), a median of 6.9 and 15.4 years after blood collection, respectively, were matched to individual controls on age, sex, and date of blood draw. Serum antibodies to the major viral capsid protein (VP1) were used to establish infection history for each polyomavirus. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using conditional logistic regression. In the Janus Serum Bank, MCPyV infection was associated with a higher risk of glioma overall (OR: 1.56; 95% CI 1.10, 2.19). A modest, nonsignificant positive association with MCPyV infection was also observed in CPS-II (OR: 1.29; 95% CI 0.54, 3.08). In both cohorts, glioma risk was not significantly related to infection with JCV, BKV or HPyV6. The present study suggests that MCPyV infection may increase glioma risk.

Background The malaria mosquito Anopheles punctipennis, a widely distributed species in North America, is capable of transmitting human malaria and is actively involved in the transmission of the ungulate malaria parasite Plasmodium odocoilei. However, molecular diagnostic tools based on Internal Transcribed Spacer 2 (ITS2) of ribosomal DNA are lacking for this species. Anopheles punctipennis is a former member of the Anopheles maculipennis complex but its systematic position remains unclear. Methods In this study, ITS2 sequences were obtained from 276 An. punctipennis specimens collected in the eastern and midwestern United States and a simple and robust Restriction Fragment Length Polymorphism approach for species identification was developed. The maximum-likelihood phylogenetic tree was constructed based on ITS2 sequences available through this study and from GenBank for 20 species of Anopheles. Results The analysis demonstrated a consistent ITS2 sequence length and showed no indications of intragenomic variation among the samples based on ITS2, suggesting that An. punctipennis represents a single species in the studied geographic locations. In this study, An. punctipennis was found in urban, rural, and forest settings, suggesting its potential broad role in pathogen transmission. Phylogeny based on ITS2 sequence comparison demonstrated the close relationship of this species with other members of the Maculipennis group. Conclusions This study developed molecular tools based on ITS2 sequences for the malaria vector An. punctipennis and clarified the phylogenetic position of the species within the Maculipennis group.

Introduction Muscle-strengthening activity (MSA) has beneficial effects on hypertension, glucose homeostasis, and other health conditions; however, its association with mortality is not as well understood. Methods We analyzed data from the Cancer Prevention Study-II Nutrition Cohort (data collection 1982–2014), a prospective US cohort that consisted of 72,462 men and women who were free of major chronic diseases; 18,034 of the cohort died during 13 years of follow-up (2001–2014). We used Cox proportional hazards modeling, controlling for various potential confounding factors, to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for MSA (none, >0 to 0 to 0 to

Toxoplasma gondii (T gondii) is a common parasite that shows affinity to neural tissue and may lead to the formation of cysts in the brain. Previous epidemiologic studies have suggested an association between glioma and increased prevalence of T gondii infection, but prospective studies are lacking. Therefore, we examined the association between prediagnostic T gondii antibodies and risk of glioma in two prospective cohorts using a nested case-control study design. Cases and matched controls were selected from the American Cancer Society's Cancer Prevention Study-II Nutrition Cohort (CPSII-NC) (n = 37 cases and 74 controls) and the Norwegian Cancer Registry's Janus Serum Bank (Janus) (n = 323 cases and 323 controls). Blood samples collected prior to diagnosis were analyzed for antibodies to two T gondii surface antigens (p22 and sag-1), with individuals considered seropositive if antibodies to either antigen were detected. Conditional logistic regression was used to calculate odds ratios (OR) and 95% confidence intervals (95% CI) for each cohort. In both cohorts, a suggestive increase in glioma risk was observed among those infected with T gondii (OR: 2.70; 95% CI: 0.96-7.62 for CPSII-NC; OR: 1.32, 95% CI: 0.85-2.07 for Janus), particularly among participants with high antibody titers specific to the sag-1 antigen (CPSII-NC OR: 3.35, 95% CI: 0.99-11.38; Janus OR: 1.79, 95% CI: 1.02-3.14). Our findings provide the first prospective evidence of an association between T gondii infection and risk of glioma. Further studies with larger case numbers are needed to confirm a potential etiologic role for T gondii in glioma.

Background Height and weight are commonly used metrics in epidemiologic studies to calculate body mass index. Large cohort studies generally assess height and weight by self-report rather than by measurement. The aim of this study was to assess the validity of self-reported height and weight in the Cancer Prevention Study-3 (CPS-3), a large, nationwide cohort recruited by the American Cancer Society between 2006–2013. Methods In a subset of CPS-3 participants (n = 2,643), weight and height were assessed at the same time via self-report and in-person measurement. BMI was calculated and classified underweight (

Cadmium and lead have been classified as carcinogens by the International Agency for Research on Cancer. However, their associations with breast cancer risk are unknown despite their persistence in ...

Background: The glycemic potential and energy density (ED) of diet may influence endometrial cancer risk. Although glycemic load (GL) is considered a probable risk factor for endometrial cancer, no studies have evaluated the association of total dietary ED with risk. Methods: We evaluated associations of ED, GL, and glycemic index (GI) with postmenopausal endometrial cancer incidence. Analyses included 30,997 postmenopausal women from the Cancer Prevention Study II Nutrition Cohort with no previous history of cancer or diabetes, who provided information on diet, lifestyle, and medical history in 1999 and were followed for cancer incidence through June 2013. Multivariable-adjusted HRs and 95% confidence intervals were estimated for quartiles (Q) of total dietary ED, GL, and GI in relation to endometrial cancer incidence using Cox proportional hazards regression models. Results: During a median follow-up time of 13.6 years, 425 endometrial cancer cases were identified. Median dietary ED was 1.5 kcal/g [interquartile range (IQR) = 1.3–1.7 kcal/g]. Median (IQR) GL and GI were 113.7 (100.5–126.8) and 52.5 (50.4–54.5), respectively. After adjustment for age, use of hormone replacement therapy, physical activity, and body mass index (kg/m2), neither ED, GL, nor GI were associated with endometrial cancer risk. Conclusions: We found no associations of ED, GL, or GI with endometrial cancer risk. Impact: These results do not support an association between total dietary ED, GL, or GI and risk of postmenopausal endometrial cancer. Cancer Epidemiol Biomarkers Prev; 27(1); 113–5. ©2017 AACR.

BACKGROUND: Fecally contaminated drinking water is believed to be a major contributor to the global burden of diarrheal disease and a leading cause of mortality among young children. However, recent systematic reviews and results from blinded studies of water quality interventions have raised questions about the risk associated with fecally contaminated water, particularly as measured by thermotolerant coliform (TTC) bacteria, a WHO-approved indicator of drinking water quality. OBJECTIVES: We investigated the association between TTC in drinking water and diarrhea using data from seven previous studies. METHODS: We obtained individual-level data from available field studies that measured TTC levels in household-drinking water and reported prevalence of diarrhea among household members during the days prior to the visit. RESULTS: The combined data set included diarrhea prevalence for 26,518 individuals and 8,000 water samples from 4,017 households, yielding 45,052 observations. The odds of diarrhea increased for each log10 increase in TTC/100 mL by 18% (95% CI: 11, 26%) for children < 5 years old and 12% (95% CI: 8, 18%) for all ages. For all ages, the odds of diarrhea increased by 21%, 35% and 49% for those whose household water samples were from 11-100, 101-1,000, and > 1,000 TTC/100 mL, respectively compared to < 1 TTC/100 mL. We found no evidence of increased odds of diarrhea with contamination levels below 11 TTC/100 mL, either in adults or children. CONCLUSIONS: Our analysis of individual-level data shows increased risk of diarrhea with increasing levels of TTC in drinking water. These results suggest an association between fecally contaminated water and diarrheal disease and provides support for health-based targets for levels of TTC in drinking water and for interventions to improve drinking water quality to prevent diarrhea. CITATION: Hodge J, Chang HH, Boisson S, Collin SM, Peletz R, Clasen T. 2016. Assessing the association between thermotolerant coliforms in drinking water and diarrhea: an analysis of individual level data from multiple studies. Environ Health Perspect 124:1560-1567; http://dx.doi.org/10.1289/EHP156.

Background: Brain and other nervous system cancers are relatively rare but remain the eighth most common cancer among persons aged 40 years and older. Gliomas are the most common type of primary brain cancer, have poor overall survival, and their etiology remains unclear. Toxoplasma gondii (T. gondii) is a common protozoan parasite and has been reported to cause glioma-like tumors in animals. In humans, the infection shows affinity to neural tissue and may lead to local necrosis and inflammation, and the formation of cysts in the brain from which infection may be reactivated according to host immunologic status. Previous ecologic and case-control studies have suggested an association with increased risk of gliomas in humans infected with T. gondii, but retrospective collection of exposure information from brain tumor patients is challenging. In this study, we assessed the relation between prediagnostic T. gondii antibodies and risk of glioma in two prospective cohorts. Methods: Cases and matched controls were selected from the American Cancer Society's Cancer Prevention Study-II (CPSII) and the Norwegian Cancer Registry's Janus Serum Bank (Janus). To avoid potential effects of preclinical disease on serum antibody titers, the study was restricted to cases with blood collected a minimum of 3 years (in CPSII) or 5 years (in Janus) prior to glioma diagnosis. Using a multiplex serological assay, blood samples were analyzed simultaneously for antibodies to two surface antigens (p22 and sag1) specific to T. gondii which represent acute parasitic infection. Samples were considered positive for T. gondii infection if either antibody was positive, defined using assay-specific cutoffs unique to each antigen. Conditional logistic regression was used to calculate odds ratios for each cohort separately. Results: A total of 37 cases and 74 controls were included from CPSII while 323 cases and 323 controls were included from Janus. The median age at blood draw was 71 years in CPSII and 41.1 years in Janus. In both cohorts, there was a suggestive increased risk of glioma among those infected with T. gondii. The association was strongest for gliomas diagnosed in approximately the first decade after blood collection in both Janus (OR: 2.50; 95% CI: 0.97-6.44, p = 0.058) and CPSII (OR: 2.71; 95% CI: 0.95-7.62, p = 0.060), with similar results observed in high grade glioblastoma (CPSII N = 27, OR: 2.31, 95% CI: 0.71-7.47; Janus N = 61, OR: 2.25, 95% CI: 0.70-7.31) and lower grade gliomas (Janus N = 47, OR: 3.00, 95% CI: 0.61-15). Conclusions: Blood antibody levels from two prospective cohorts indicate a possible association between T. gondii infection and increased risk of glioma. The association was strongest in the first decade after antibody measurement. The consistency, magnitude, and direction of the association support the need for pooled analyses with larger case numbers for definitive investigation. Citation Format: James M. Hodge, Anna E. Coghill, Youngchul Kim, Noemi Bender, Stephanie Smith-Warner, Tom K. Grimsrud, Susan M. Gapstur, Tim Waterboer, Lauren R. Teras, Kathleen M. Egan. Association of Toxoplasma gondii antibodies in blood and risk of glioma [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1051.

BACKGROUND: Local, state, and national childhood blood lead surveillance is based on healthcare providers and clinical laboratories reporting test results to public health departments. Increased interest in detecting blood lead level (BLL) patterns and changes of potential public health significance in a timely manner has highlighted the need for surveillance systems to rapidly detect and investigate these events. OBJECTIVE: Decrease the time to detect changes in surveillance patterns by using an alerting algorithm developed and assessed through historical child blood lead surveillance data analysis. METHODS: We applied geographic and temporal data-aggregation strategies on childhood blood lead surveillance data and developed a novel alerting algorithm. The alerting algorithm employed a modified cumulative summary/Shewhart algorithm, initially applied on 113 months of data from two jurisdictions with a known increase in the proportion of children 5 µg/dl. RESULTS: Alert signals retrospectively identified time periods in two jurisdictions where a known change in the proportion of children =5 µg/dl occurred. Additionally, we identified alert signals among six of the 18 (33%) randomly selected counties assessed where no previously known or suspected pattern changes existed. CONCLUSION: The modified cumulative summary/Shewhart algorithm provides a framework for enhanced blood lead surveillance by identifying changes in the proportion of children with BLLs >=5 µg/dl. The algorithm has the potential to alert public health officials to changes requiring further important public health investigation.

Household water treatment (HWT) provides a means for vulnerable populations to take charge of their own drinking water quality as they patiently wait for the pipe to finally reach them. In many low-income countries, however, promoters have not succeeded in scaling up the intervention among the target population or securing its consistent and sustained use. Carbon financing can provide the funding for reaching targeted populations with effective HWT solutions and the incentives to ensure their long-term uptake. Nevertheless, programs have been criticized because they do not actually reduce carbon emissions. We summarize the background and operation of carbon financing of HWT interventions, including the controversial construct of "suppressed demand". We agree that these programs have limited potential to reduce greenhouse gas emissions and that their characterization of trading "carbon for water" is misleading. Nevertheless, we show that the Kyoto Protocol expressly encouraged the use of suppressed demand as a means of allowing low-income countries to benefit from carbon financing provided it is used to advance development priorities such as health. We conclude by recommending changes to existing criteria for eligible HWT programs that will help ensure that they meet the conditions of microbiological effectiveness and actual use that will improve their potential for health gains.

Digital pathology images potentially contain novel patterns that may be perceived by modern deep learning models, but not humans. Prior unsupervised pattern recognition approaches have been used to reveal prognostically-relevant subtypes of glioblastoma (PMID: 28984190) and breast density segmentation (PMID: 26915120), and may complement supervised machine learning models trained using labeled data. In the Cancer Prevention Study II (CPS-II) cohort (PMID: 12015775), high-resolution, digitized hemotoxylin and eosin diagnostic slides are available for approximately 1,700 breast cancer cases providing an opportunity to perform unsupervised pattern recognition image analysis for epidemiologic breast cancer studies. Given the size of the dataset and complexity of the models, we constructed an end-to-end analytical pipeline, including preprocessing, feature engineering, and clustering, using cloud-based technologies that enable analysis at scale. Prior to training the unsupervised models, we faced issues converting raw images with open-source software. Specifically, OpenSlides could not open the Leica Versa SCN files due to their proprietary format while BioFormats inverted colors. To fix these issues, we altered the BioFormats library to successfully convert the files into a TIFF format. Since this issue likely affects other researchers, we are in discussions to provide the fix under a public license. TIFF formatted images were then denoised through color normalization to reduce hue variance and artifact detection to remove unwanted features such as pathologist annotations. Due to the computational complexity of analyzing the full image, images were padded with white space to ensure divisibility and broken into nine tiles of a predefined size. To further reduce computation time, uninformative tiles were filtered based on a predetermined threshold of artifact and white space composition. The remaining tiles were input to the unsupervised models. We used convolutional autoencoders, specifically a modified VGG-16 model without pretrained weights and a deep embedded clustering algorithm. These models learn representations of the images called ‘feature vectors’ and encode the images’ salient patterns. The final model was chosen based on iterative testing on a subsample of 100 images (N=21,472 tiles) and performance comparison of various VGG-inspired autoencoders. The feature vectors were clustered by K-means to summarize the information in a format suitable for statistical analyses. Our initial results show that the system captures macro-scale tissue patterns at lower magnifications (1x and 5x) and produces clusters that can be integrated into epidemiological studies of breast cancer etiology and prognosis. Citation Format: Jacob L. Evans, William Seo, Mary Macheski-Preston, Michelle Fritz, Samantha Puvanesarajah, James Hodge, Ted Gansler, Susan Gapstur, Mia M. Gaudet, Michelle Yi. A scalable, cloud-based, unsupervised deep learning system for identification, extraction, and summarization of potentially imperceptible patterns in whole-slide images of breast cancer tissue [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1635.

Background: Machine learning (ML) methods are becoming more feasible for use in clinical and epidemiologic research of breast cancer, particularly when characterizing histopathology. Compared to supervised ML methods, unsupervised approaches represent an opportunity to distinguish features heretofore unknown. The purpose of this study was to use unsupervised deep learning methods to identify histopathological features in diagnostic breast cancer hematoxylin and eosin (H&E) slides that are associated with clinical characteristics and patient outcomes. Methods: One H&E slide was scanned (Leica Biosystems Aperio Versa scanner) at 20x magnification for each of 1,716 women diagnosed with breast cancer from the Cancer Prevention Study-II Nutrition Cohort. In the pre-processing phase, the scanned images underwent color normalization, artifact detection, and tiling. We then used an un-pretrained VGG16 autoencoder with data augmentation for feature learning and extraction from tiles. These features were two-tiered clustered using the K-means algorithm. Each tile was assigned the cluster with the highest probability. The tiles were reassembled into whole slide images. For each slide, the proportion of tiles in each cluster was calculated. We will associate clusters with clinical features and 5- and 10-year breast cancer-specific survival using multivariable logistic and Cox proportional hazards regression models, respectively. Results: Mean age at baseline enrollment (1992-1993) and breast cancer diagnosis for the cases was 60.6 years (SD=6.0) and 71.5 years (SD=7.0), respectively. The majority of cancer diagnoses occurred after 1999 (79%) and 81% of women included were diagnosed invasive breast cancer. The final pipeline for the full set of images is currently being built. Preliminary runs at the 1x magnification level with 100 cases (N=21,472 tiles) have shown clustering based on macro-level features such as adipose, stromal and epithelial content. Second-tier clustering (clustering within clusters) shows further delineation of groups within clusters of interest (i.e. epithelial-cell rich regions). The final output with all 1,716 slides will be based on analysis at the 5x magnification level. Discussion: We expect that some histopathological features identified by ML models will be associated with conventional pathology features, clinical features, and breast cancer-specific survival. Utilization of ML methods for analyzing histology slides provides additional data that can be integrated into epidemiological studies. Future directions include analyzing images at higher magnifications (10x or 20x) and assessing the association between ML histopathological characteristics and breast cancer risk factors and incorporating these characteristics into prognostic models. Citation Format: Samantha Puvanesarajah, James M. Hodge, Jacob L. Evans, William Seo, Michelle Yi, Michelle M. Fritz, Mary Macheski-Preston, Ted Gansler, Susan M. Gapstur, Mia M. Gaudet. Unsupervised deep-learning to identify histopathological features among breast cancers in the Cancer Prevention Study-II Nutrition Cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2417.

Development professionals exist in the gap between their institutions and their missions and benefactors and their values systems, providing a vital connection. Along with their benefactors, they too can be transformed in the development process and become more and more human through living well-examined lives.