Staging and clean room: Constructs designed to facilitate transparency and reduce bias in comparative analyses of real-world data.

PURPOSE: We describe constructs designed to protect the integrity of the results from comparative analyses using real-world data (RWD): staging and clean room. METHODS: Staging involves performing sequential preliminary analyses and evaluating the population size available and potential bias before conducting comparative analyses. A clean room involves restricted access to data and preliminary results, policies governing exploratory analyses and protocol deviations, and audit trail. These constructs are intended to allow decisions about protocol deviations, such as changes to design or model specification, to be made without knowledge of how they might affect subsequent analyses. We describe an example for implementing staging with a clean room. RESULTS: Stage 1 may involve selecting a data source, developing and registering a protocol, establishing a clean room, and applying inclusion/exclusion criteria. Stage 2 may involve attempting to achieve covariate balance, often through propensity score models. Stage 3 may involve evaluating the presence of residual confounding using negative control outcomes. After each stage, check points may be implemented when a team of statisticians, epidemiologists and clinicians masked to how their decisions may affect study outcomes, reviews the results. This review team may be tasked with making recommendations for protocol deviations to address study precision or bias. They may recommend proceeding to the next stage, conducting additional analyses to address bias, or terminating the study. Stage 4 may involve conducting the comparative analyses. CONCLUSIONS: The staging and clean room constructs are intended to protect the integrity and enhance confidence in the results of analyses of RWD.

Circadian pathway genes in relation to glioma risk and outcome.

PURPOSE: There is growing evidence that circadian disruption may alter risk and aggressiveness of cancer. We evaluated common genetic variants in the circadian gene pathway for associations with glioma risk and patient outcome in a US clinic-based case-control study. METHODS: Subjects were genotyped for 17 candidate single nucleotide polymorphisms in ARNTL, CRY1, CRY2, CSNK1E, KLHL30, NPAS2, PER1, PER3, CLOCK, and MYRIP. Unconditional logistic regression was used to estimate age and gender-adjusted odds ratios (OR) and 95 % confidence intervals (CI) for glioma risk under three inheritance models (additive, dominant, and recessive). Proportional hazards regression was used to estimate hazard ratios for glioma-related death among 441 patients with high-grade tumors. Survival associations were validated using The Cancer Genome Atlas (TCGA) dataset. RESULTS: A variant in PER1 (rs2289591) was significantly associated with overall glioma risk (per variant allele OR 0.80; 95 % CI 0.66-0.97; p trend = 0.027). The variant allele for CLOCK rs11133391 under a recessive model increased risk of oligodendroglioma (OR 2.41; 95 % CI 1.31-4.42; p = 0.005), though not other glioma subtypes (p for heterogeneity = 0.0033). The association remained significant after false discovery rate adjustment (p = 0.008). Differential associations by gender were observed for MYRIP rs6599077 and CSNK1E rs1534891 though differences were not significant after adjustment for multiple testing. No consistent mortality associations were identified. Several of the examined genes exhibited differential expression in glioblastoma multiforme versus normal brain in TCGA data (MYRIP, ARNTL, CRY1, KLHL30, PER1, CLOCK, and PER3), and expression of NPAS2 was significantly associated with a poor patient outcome in TCGA patients. CONCLUSION: This exploratory analysis provides some evidence supporting a role for circadian genes in the onset of glioma and possibly the outcome of glioma.

Anthropometric factors in relation to risk of glioma.

INTRODUCTION: Increased height and greater adiposity have been linked to an increased risk of many cancer types, though few large studies have examined these associations in glioma. We examined body weight and height as potential risk factors for glioma in a large US-based case-control study. METHODS: The analysis included 1,111 glioma cases and 1,096 community controls. In a structured interview, participants reported their height and weight at 21 years of age, lowest and highest weight in adulthood, and weight 1-5 years in the past. RESULTS: Being underweight at age 21 (BMI < 18.5 kg/m(2)) was inversely associated with the risk of glioma development. This protective association was observed in both men and women, but reached statistical significance in women only (multivariate OR 0.68; 95 % CI 0.48, 0.96). When BMI at age 21 was assessed as a continuous variate, a small but significant increase in risk was observed per unit increase in kg/m(2) (OR 1.04; 95 % CI 1.02, 1.07). Adult height, recent body weight, and weight change in adulthood were not associated with glioma risk. All results were similar among never smokers and were consistent after stratifying by glioma subtype. CONCLUSION: The present data suggest that a low body weight in early adulthood is associated with a reduced risk of glioma later in life. Results are consistent with previous studies in showing no material association of glioma risk with usual adult body weight. The present study does not support any association of adult stature with glioma risk.

Toenail iron, genetic determinants of iron status, and the risk of glioma.

PURPOSE: Iron is essential for oxygen transport and oxidative metabolism; however, elevated iron stores can trigger overproduction of reactive oxygen species and induce DNA damage. Little is known about the association between body iron stores and glioma risk. This study examined the associations of iron levels measured in toenails and genetic variants linked to body iron stores with risk of glioma in a clinic-based case-control study. METHODS: Samples were collected a median of 24 days following glioma diagnosis in the cases (10th-90th percentile, range: 10-44 days). Nail iron levels were measured in 300 cases and 300 controls using neutron activation analysis. A total of 24 genetic variants associated with iron status were genotyped in 622 cases and 628 controls. Logistic regression was used to estimate odds ratios (OR) and 95 % confidence intervals (CI) for glioma risk according to toenail iron and the examined genotypes. RESULTS: No association was observed between toenail iron and glioma risk when restricting to cases with nails collected within ~3 weeks of diagnosis (OR = 0.93; 95 % CI 0.46, 1.87 comparing those with high (≥14 µg/g) vs. low (<6 µg/g) iron levels). In contrast, an inverse association with increasing iron was observed after restricting to cases with a delay of 3 weeks or greater (OR = 0.42; 95 % CI 0.19, 0.95), reflecting potentially insidious effects of advancing disease on iron levels among the cases. No associations were observed for any of the examined genetic variants. CONCLUSION: The results do not support a role for body iron stores as a determinant of glioma risk.

Prediagnostic body weight and survival in high grade glioma.

Greater adiposity has been linked to an increased risk and/or poorer survival in a variety of cancers. We examined whether prediagnostic body weight 1-5 years prior to diagnosis is associated with survival in patients with high grade glioma. The analysis was based on a series of patients with high-grade glioma (N = 853) enrolled in a US-based multicenter case-control study. Subjects reported height and weight 1-5 years prior to interview and at age 21. BMI was categorized according to WHO criteria as underweight (BMI <18.5 kg/m(2)), normal weight (BMI 18.5-24.9 kg/m(2)), overweight (BMI 25-29.9 kg/m(2)) and obese (BMI ≥30 kg/m(2)). Proportional hazards regression was used to estimate hazard ratios (HR) and 95 % confidence intervals (CIs) for glioma-related death according to body mass index (BMI, kg/m(2)). Overall survival was reduced among patients underweight (median survival: 12.0 months) or obese (median: 13.6 months) when compared to patients of normal weight (median: 17.5 months) prior to glioma diagnosis (p = 0.004). In a multivariate model controlling for other prognostic factors, an excess mortality was observed in patients reporting obese body weights 1-5 years prior to study interview when compared to patients with a normal BMI (HR = 1.32; 95 % CI 1.04-1.68). Consistent patterns of association with excess body weight were observed in men and women, and all findings were similar regardless of treatment for glioma. A lower than optimal body weight was associated with a nonsignificant excess mortality in multivariate analysis. Premorbid obesity was significantly associated with a poor patient outcome independent of treatment and established prognostic factors. Excess body weight may be an adverse prognostic factor in glioma, a relationship observed across a spectrum of cancer types. The current findings linking prediagnostic body weight with mortality in high-grade glioma warrant further research.

Early life exposures and the risk of adult glioma.

Exposure to common infections in early life may stimulate immune development and reduce the risk for developing cancer. Birth order and family size are proxies for the timing of exposure to childhood infections with several studies showing a reduced risk of glioma associated with a higher order of birth (and presumed younger age at infection). The aim of this study was to examine whether birth order, family size, and other early life exposures are associated with the risk of glioma in adults using data collected in a large clinic-based US case-control study including 889 glioma cases and 903 community controls. A structured interviewer-administered questionnaire was used to collect information on family structure, childhood exposures and other potential risk factors. Logistic regression was used to calculate odds ratios (OR) and corresponding 95% confidence intervals (CI) for the association between early life factors and glioma risk. Persons having any siblings were at significantly lower risk for glioma when compared to those reporting no siblings (OR=0.64; 95% CI 0.44-0.93; p=0.020). Compared to first-borns, individuals with older siblings had a significantly lower risk (OR=0.75; 95% CI 0.61-0.91; p=0.004). Birth weight, having been breast fed in infancy, and season of birth were not associated with glioma risk. The current findings lend further support to a growing body of evidence that early exposure to childhood infections reduces the risk of glioma onset in children and adults.

Rare TP53 genetic variant associated with glioma risk and outcome.

Validation of a recent finding linking a rare variant in TP53 to the risk of glioma, the most common primary brain tumour, is reported here. This study genotyped the single nucleotide polymorphism (SNP) rs78378222 in 566 glioma cases and 603 controls. The variant 'C' allele (with an allelic frequency of 1.1% in controls) was associated with a 3.5-fold excess in glioma risk (odds ratio 3.54; p=0.0001). Variant carriers had significantly improved survival (hazard ratio 0.52; p=0.009) when compared to non-carriers. The rs78378222 SNP is the first confirmed rare susceptibility variant in glioma. Results may shed light on the aetiology and progression of these tumours.

An exploratory analysis of common genetic variants in the vitamin D pathway including genome-wide associated variants in relation to glioma risk and outcome.

PURPOSE: Experimental and epidemiological evidence shows a beneficial role of vitamin D in cancer. In vitro evidence is consistent with a similar protective function in glioma; however, no study has yet examined the potential role of vitamin D in glioma. METHODS: We evaluated the association between common genetic variants in the vitamin D pathway and glioma risk and patient outcome in 622 newly diagnosed glioma cases and 628 healthy controls enrolled in a clinic-based case-control study. Subjects were genotyped for 7 candidate and tagging single nucleotide polymorphisms in the vitamin D receptor and 8 additional variants in NADSYN1, GC, CYP24A1, CYP2R1, and C10ORF88 linked in genome-wide association studies to serum concentrations of vitamin D. Unconditional logistic regression was used to estimate age- and gender-adjusted odds ratios and 95 % confidence intervals for glioma risk according to vitamin D genotypes. Proportional hazards regression was used to estimate hazard ratios for glioma-related death among 320 patients diagnosed with high-grade tumors. P values were uncorrected for multiple comparisons. RESULTS: Risk of astrocytic tumors was associated with variant alleles in rs3829251 (NADSYN1), rs10741657 (CYP2R1), rs2228570 (Fok1, VDR), and rs731236 (Taq1, VDR). No risk associations were found among oligodendroglial tumors. Survival associations were observed according to variant status for rs1544410 (Bsm1, VDR) and rs6013897 (CYP24A1). CONCLUSION: This exploratory analysis provides limited evidence of a role for genetic variation in vitamin D pathway genes with glioma risk and survival.

Cancer susceptibility variants and the risk of adult glioma in a US case-control study.

Malignant gliomas are the most common and deadly brain tumors. Although their etiology remains elusive, recent studies have narrowed the search for genetic loci that influence risk. We examined variants implicated in recent cancer genome-wide association studies (GWAS) for associations with glioma risk in a US case-control study. Cases were identified from neurosurgical and neuro-oncology clinics at major academic centers in the Southeastern US. Controls were identified from the community or were friends or other associates of cases. We examined a total of 191 susceptibility variants in genes identified in published cancer GWAS including glioma. A total of 639 glioma cases and 649 controls, all Caucasian, were included in analysis. Cases were enrolled a median of 1 month following diagnosis. Among glioma GWAS-identified variants, we detected associations in CDKN2B, RTEL1, TERT and PHLDB1, whereas we did not find overall associations for CCDC26. Results showed clear heterogeneity according to histologic subtypes of glioma, with TERT and RTEL variants a feature of astrocytic tumors and glioblastoma (GBM), CCDC26 and PHLDB1 variants a feature of astrocytic and oligodendroglial tumors, and CDKN2B variants most prominent in GBM. No examined variant in other cancer GWAS was found to be related to risk after adjustment for multiple comparisons. These results suggest that GWAS-identified SNPs in glioma mark different molecular etiologies in glioma. Stratification by broad histological subgroups may shed light on molecular mechanisms and assist in the discovery of novel loci in future studies of genetic susceptibility variants in glioma.

A functional polymorphism in the pre-miR-146a gene is associated with risk and prognosis in adult glioma.

MicroRNAs (miRNAs) are non-coding RNAs that function as post-transcriptional regulators of tumor suppressors and oncogenes. Single nucleotide polymorphisms (SNPs) in miRNAs may contribute to carcinogenesis by altering expression of miRNAs and their targets. A G>C polymorphism (rs2910164) in the miR-146a precursor sequence leads to a functional change associated with the risk for numerous malignancies. A role for this SNP in glioma pathogenesis has not yet been examined. We investigated whether rs2910164 genotypes influence glioma risk and prognosis in a multi-center case-control study comprised of 593 Caucasian glioma cases and 614 community-based controls. Unconditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for rs2910164 genotypes according to case status. Cox proportional hazards regression modeling was used to estimate hazards ratios (HR) and 95% CIs according to genotype among glioblastomas, the most lethal glioma subtype. An increased glioma risk was observed among rs2910164 minor allele (C) carriers (per allele OR (95% CI) = 1.22 (1.01-1.46, p (trend) = 0.039)). The association was stronger among older subjects carrying at least one copy of the C allele (OR (95% CI) = 1.38 (1.04-1.83, P = 0.026). Mortality was increased among minor allele carriers (HR (95% CI) = 1.33 (1.03-1.72, P = 0.029)), with the association largely restricted to females (HR (95% CI) = 2.02 (1.28-3.17, P = 0.002)). We provide novel data suggesting rs2910164 genotype may contribute to glioma susceptibility and outcome. Future studies are warranted to replicate these findings and characterize mechanisms underlying these associations.

Brain tumor risk according to germ-line variation in the MLLT10 locus.

Genome-wide association studies have recently identified a cancer susceptibility locus at 10p12 mapping to MLLT10 associated with the onset of diverse tumors. We genotyped two tightly linked single-nucleotide polymorphisms (SNPs) at MLLT10 associated with meningioma (rs12770228) or ovarian cancer (rs1243180), and tested for associations among 295 meningioma cases, 606 glioma cases and 646 noncancer controls, all of European descent. The variant 'A' allele in MLLT10 rs12770228 was associated with an increased risk of meningioma (per allele odds ratio: 1.25; 95% confidence interval: 1.02, 1.53; P=0.031). Similar associations were observed for rs1243180. MLLT10 variants were unrelated to glioma. Functional investigation identified 22 candidate functional SNPs mapping to this region. The present study further validates 10p12 as a meningioma risk locus.

SWI/SNF gene variants and glioma risk and outcome.

BACKGROUND: The human SWItch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complex plays essential roles in a variety of cellular processes and has been implicated in human cancer. However, the role of germline genetic variants in this complex in relation to cancer risk is not well studied. METHODS: We assessed the association of 16 variants in the catalytic subunits (SMARCA2 and SMARCA4) of the SWI/SNF complex with the risk of glioma subtypes (lower grade astrocytoma, oligodendroglioma and glioblastoma [GBM]) and with mortality from high-grade tumors (GBM) in a multicenter US case-control study that included 561 cases and 574 controls. Associations were estimated with odds ratios (OR, for risk) or hazards ratios (HR, for mortality) with 95% confidence intervals (CI). False discovery rate (FDR-q) was used to control for multiple testing in risk associations. RESULTS: None of the investigated SNPs was associated with overall glioma risk. However, analyses according to histological subtypes revealed a statistically significant increased risk of oligodendroglioma in association with SMARCA2 rs2296212 (OR = 4.05, 95% CI = 1.11-14.80, P = 0.030, q = 0.08) and rs4741651 (OR = 4.68, 95% CI = 1.43-15.30, P = 0.011, q = 0.08) and SMARCA4 rs11672232 (OR = 1.90, 95% CI = 1.01-3.58, P = 0.048, q = 0.08) and rs12232780 (OR = 2.14, 95% CI = 1.06-4.33, P = 0.035, q = 0.08). No significant risk associations were observed for GBM or lower grade astrocytoma. Suggestive associations with GBM mortality were not validated in the Cancer Genome Atlas. CONCLUSION: Our findings suggest that genetic variants in SMARCA2 and SMARCA4 influence the risk of oligodendroglioma. Further research is warranted on the SWI/SNF complex genes and epigenetic mechanisms more generally in the development of glioma in adults.

SSBP2 variants are associated with survival in glioblastoma patients.

PURPOSE: Glioblastoma is a devastating, incurable disease with few known prognostic factors. Here, we present the first genome-wide survival and validation study for glioblastoma. EXPERIMENTAL DESIGN: Cox regressions for survival with 314,635 inherited autosomal single-nucleotide polymorphisms (SNP) among 315 San Francisco Adult Glioma Study patients for discovery and three independent validation data sets [87 Mayo Clinic, 232 glioma patients recruited from several medical centers in Southeastern United States (GliomaSE), and 115 The Cancer Genome Atlas patients] were used to identify SNPs associated with overall survival for Caucasian glioblastoma patients treated with the current standard of care, resection, radiation, and temozolomide (total n = 749). Tumor expression of the gene that contained the identified prognostic SNP was examined in three separate data sets (total n = 619). Genotype imputation was used to estimate hazard ratios (HR) for SNPs that had not been directly genotyped. RESULTS: From the discovery and validation analyses, we identified a variant in single-stranded DNA-binding protein 2 (SSBP2) on 5q14.1 associated with overall survival in combined analyses (HR, 1.64; P = 1.3 × 10(-6)). Expression of SSBP2 in tumors from three independent data sets also was significantly related to patient survival (P = 5.3 × 10(-4)). Using genotype imputation, the SSBP2 SNP rs17296479 had the strongest statistically significant genome-wide association with poorer overall patient survival (HR, 1.79; 95% CI, 1.45-2.22; P = 1.0 × 10(-7)). CONCLUSION: The minor allele of SSBP2 SNP rs17296479 and the increased tumor expression of SSBP2 were statistically significantly associated with poorer overall survival among glioblastoma patients. With further confirmation, previously unrecognized inherited variations influencing survival may warrant inclusion in clinical trials to improve randomization. Unaccounted for genetic influence on survival could produce unwanted bias in such studies.