Charles P. Quesenberry

A Case–Control Study of Screening Sigmoidoscopy and Mortality from Colorectal Cancer

BACKGROUND The efficacy of sigmoidoscopic screening in reducing mortality from colorectal cancer remains uncertain. A randomized trial would be ideal for clarifying this issue but is very difficult to conduct. Case-control studies provide an alternative method of estimating the efficacy of screening sigmoidoscopy. METHODS Using data on the 261 members of the Kaiser Permanente Medical Care Program who died of cancer of the rectum or distal colon from 1971 to 1988, we examined the use of screening by rigid sigmoidoscopy during the 10 years before the diagnosis and compared it with the use of screening in 868 control subjects matched with the case subjects for age and sex. RESULTS Only 8.8 percent of the case subjects had undergone screening by sigmoidoscopy, as compared with 24.2 percent of the controls (matched odds ratio, 0.30; 95 percent confidence interval, 0.19 to 0.48). Adjustment for potential confounding factors increased the odds ratio to 0.41 (95 percent confidence interval, 0.25 to 0.69). The negative association was as strong when the most recent sigmoidoscopy was 9 to 10 years before diagnosis as it was when examinations were more recent. By contrast, for 268 subjects with fatal colon cancer above the reach of the sigmoidoscope and for 268 controls, the adjusted odds ratio was 0.96 (95 percent confidence interval, 0.61 to 1.50). The specificity of the negative association for cancer within the reach of the sigmoidoscope is consistent with a true efficacy of screening rather than a confounding by unmeasured selection factors. CONCLUSIONS Screening by sigmoidoscopy can reduce mortality from cancer of the rectum and distal colon. A screening once every 10 years may be nearly as efficacious as more frequent screening.

Hypoglycemic Episodes and Risk of Dementia in Older Patients With Type 2 Diabetes Mellitus

CONTEXT Although acute hypoglycemia may be associated with cognitive impairment in children with type 1 diabetes, no studies to date have evaluated whether hypoglycemia is a risk factor for dementia in older patients with type 2 diabetes. OBJECTIVE To determine if hypoglycemic episodes severe enough to require hospitalization are associated with an increased risk of dementia in a population of older patients with type 2 diabetes followed up for 27 years. DESIGN, SETTING, AND PATIENTS A longitudinal cohort study from 1980-2007 of 16,667 patients with a mean age of 65 years and type 2 diabetes who are members of an integrated health care delivery system in northern California. MAIN OUTCOME MEASURE Hypoglycemic events from 1980-2002 were collected and reviewed using hospital discharge and emergency department diagnoses. Cohort members with no prior diagnoses of dementia, mild cognitive impairment, or general memory complaints as of January 1, 2003, were followed up for a dementia diagnosis through January 15, 2007. Dementia risk was examined using Cox proportional hazard regression models, adjusted for age, sex, race/ethnicity, education, body mass index, duration of diabetes, 7-year mean glycated hemoglobin, diabetes treatment, duration of insulin use, hyperlipidemia, hypertension, cardiovascular disease, stroke, transient cerebral ischemia, and end-stage renal disease. RESULTS At least 1 episode of hypoglycemia was diagnosed in 1465 patients (8.8%) and dementia was diagnosed in 1822 patients (11%) during follow-up; 250 patients had both dementia and at least 1 episode of hypoglycemia (16.95%). Compared with patients with no hypoglycemia, patients with single or multiple episodes had a graded increase in risk with fully adjusted hazard ratios (HRs): for 1 episode (HR, 1.26; 95% confidence interval [CI], 1.10-1.49); 2 episodes (HR, 1.80; 95% CI, 1.37-2.36); and 3 or more episodes (HR, 1.94; 95% CI, 1.42-2.64). The attributable risk of dementia between individuals with and without a history of hypoglycemia was 2.39% per year (95% CI, 1.72%-3.01%). Results were not attenuated when medical utilization rates, length of health plan membership, or time since initial diabetes diagnosis were added to the model. When examining emergency department admissions for hypoglycemia for association with risk of dementia (535 episodes), results were similar (compared with patients with 0 episodes) with fully adjusted HRs: for 1 episode (HR, 1.42; 95% CI, 1.12-1.78) and for 2 or more episodes (HR, 2.36; 95% CI, 1.57-3.55). CONCLUSIONS Among older patients with type 2 diabetes, a history of severe hypoglycemic episodes was associated with a greater risk of dementia. Whether minor hypoglycemic episodes increase risk of dementia is unknown.

Obesity in middle age and future risk of dementia: a 27 year longitudinal population based study

Abstract Objective To evaluate any association between obesity in middle age, measured by body mass index and skinfold thickness, and risk of dementia later in life. Design Analysis of prospective data from a multiethnic population based cohort. Setting Kaiser Permanente Northern California Medical Group, a healthcare delivery organisation. Participants 10 276 men and women who underwent detailed health evaluations from 1964 to 1973 when they were aged 40-45 and who were still members of the health plan in 1994. Main outcome measures Diagnosis of dementia from January 1994 to April 2003. Time to diagnosis was analysed with Cox proportional hazard models adjusted for age, sex, race, education, smoking, alcohol use, marital status, diabetes, hypertension, hyperlipidaemia, stroke, and ischaemic heart disease. Results Dementia was diagnosed in 713 (6.9%) participants. Obese people (body mass index ≥ 30) had a 74% increased risk of dementia (hazard ratio 1.74, 95% confidence interval 1.34 to 2.26), while overweight people (body mass index 25.0-29.9) had a 35% greater risk of dementia (1.35, 1.14 to 1.60) compared with those of normal weight (body mass index 18.6-24.9). Compared with those in the lowest fifth, men and women in the highest fifth of the distribution of subscapular or tricep skinfold thickness had a 72% and 59% greater risk of dementia, respectively (1.72, 1.36 to 2.18, and 1.59, 1.24 to 2.04). Conclusions Obesity in middle age increases the risk of future dementia independently of comorbid conditions.

Risk of Bladder Cancer Among Diabetic Patients Treated With Pioglitazone: Interim report of a longitudinal cohort study

OBJECTIVE Some preclinical in vivo studies and limited human data suggest a possible increased risk of bladder cancer with pioglitazone therapy. This is an interim report of an ongoing cohort study examining the association between pioglitazone therapy and the risk of bladder cancer in patients with diabetes. RESEARCH DESIGN AND METHODS This study includes 193,099 patients in the Kaiser Permanente Northern California diabetes registry who were ≥40 years of age between 1997 and 2002. Those with prior bladder cancer were excluded. Ever use of each diabetes medication (defined as two or more prescriptions within 6 months) was treated as a time-dependent variable. Cox regression–generated hazard ratios (HRs) compared pioglitazone use with nonpioglitazone use adjusted for age, sex, race/ethnicity, diabetes medications, A1C, heart failure, household income, renal function, other bladder conditions, and smoking. RESULTS The group treated with pioglitazone comprised 30,173 patients. There were 90 cases of bladder cancer among pioglitazone users and 791 cases of bladder cancer among nonpioglitazone users. Overall, ever use of pioglitazone was not associated with risk of bladder cancer (HR 1.2 [95% CI 0.9–1.5]), with similar results in men and women (test for interaction P = 0.8). However, in the a priori category of >24 months of therapy, there was an increased risk (1.4 [1.03–2.0]). Ninety-five percent of cancers diagnosed among pioglitazone users were detected at early stage. CONCLUSIONS In this cohort of patients with diabetes, short-term use of pioglitazone was not associated with an increased incidence of bladder cancer, but use for more than 2 years was weakly associated with increased risk.

A population-based study of tumor gene expression and risk of breast cancer death among lymph node-negative patients

IntroductionThe Oncotype DX assay was recently reported to predict risk for distant recurrence among a clinical trial population of tamoxifen-treated patients with lymph node-negative, estrogen receptor (ER)-positive breast cancer. To confirm and extend these findings, we evaluated the performance of this 21-gene assay among node-negative patients from a community hospital setting.MethodsA case-control study was conducted among 4,964 Kaiser Permanente patients diagnosed with node-negative invasive breast cancer from 1985 to 1994 and not treated with adjuvant chemotherapy. Cases (n = 220) were patients who died from breast cancer. Controls (n = 570) were breast cancer patients who were individually matched to cases with respect to age, race, adjuvant tamoxifen, medical facility and diagnosis year, and were alive at the date of death of their matched case. Using an RT-PCR assay, archived tumor tissues were analyzed for expression levels of 16 cancer-related and five reference genes, and a summary risk score (the Recurrence Score) was calculated for each patient. Conditional logistic regression methods were used to estimate the association between risk of breast cancer death and Recurrence Score.ResultsAfter adjusting for tumor size and grade, the Recurrence Score was associated with risk of breast cancer death in ER-positive, tamoxifen-treated and -untreated patients (P = 0.003 and P = 0.03, respectively). At 10 years, the risks for breast cancer death in ER-positive, tamoxifen-treated patients were 2.8% (95% confidence interval [CI] 1.7–3.9%), 10.7% (95% CI 6.3–14.9%), and 15.5% (95% CI 7.6–22.8%) for those in the low, intermediate and high risk Recurrence Score groups, respectively. They were 6.2% (95% CI 4.5–7.9%), 17.8% (95% CI 11.8–23.3%), and 19.9% (95% CI 14.2–25.2%) for ER-positive patients not treated with tamoxifen. In both the tamoxifen-treated and -untreated groups, approximately 50% of patients had low risk Recurrence Score values.ConclusionIn this large, population-based study of lymph node-negative patients not treated with chemotherapy, the Recurrence Score was strongly associated with risk of breast cancer death among ER-positive, tamoxifen-treated and -untreated patients.

Variants in the CDKN2B and RTEL1 regions are associated with high-grade glioma susceptibility.

The causes of glioblastoma and other gliomas remain obscure. To discover new candidate genes influencing glioma susceptibility, we conducted a principal component–adjusted genome-wide association study (GWAS) of 275,895 autosomal variants among 692 adult high-grade glioma cases (622 from the San Francisco Adult Glioma Study (AGS) and 70 from the Cancer Genome Atlas (TCGA)) and 3,992 controls (602 from AGS and 3,390 from Illumina iControlDB (iControls)). For replication, we analyzed the 13 SNPs with P < 10−6 using independent data from 176 high-grade glioma cases and 174 controls from the Mayo Clinic. On 9p21, rs1412829 near CDKN2B had discovery P = 3.4 × 10−8, replication P = 0.0038 and combined P = 1.85 × 10−10. On 20q13.3, rs6010620 intronic to RTEL1 had discovery P = 1.5 × 10−7, replication P = 0.00035 and combined P = 3.40 × 10−9. For both SNPs, the direction of association was the same in discovery and replication phases.

An increase in the incidence of gestational diabetes mellitus: Northern California, 1991-2000.

OBJECTIVE: Women with gestational diabetes mellitus (GDM) and their offspring are at increased risk of developing diabetes. Although increases in diabetes prevalence have been reported in the United States, it is unknown whether this trend is also occurring for GDM. We examined trends in the yearly cumulative incidence of GDM between the years 1991 and 2000. METHODS: A cohort study of 267,051 pregnancies screened for GDM that occurred among members of the Northern California Kaiser Permanente Medical Care Program, representing 86.8% of all eligible pregnancies, was undertaken. RESULTS: GDM was identified in 14,175 pregnancies according to the diagnostic plasma glucose thresholds of the American Diabetes Association (96.5%) or the World Health Organization (3.5%). An additional 2,743 pregnant women with GDM were identified by a hospital discharge diagnosis. The women screened in 2000 were slightly older (mean [standard deviation] age 28.8 [6.0] years) than were those screened in 1991 (28.2 [5.7] years) and more likely to be from minority ethnic groups (51.4% versus 37.3% identified as African American, Asian, Hispanic, and other). The age- and ethnicity-adjusted yearly cumulative incidence of GDM increased steadily from 5.1% in 1991 to 7.4% in 1997 and leveled off through 2000 (6.9%). DISCUSSION: The observed increase in yearly cumulative incidence of GDM was independent of changes in age and ethnicity of the study population. A true increase in GDM incidence might reflect or contribute to the increases in the prevalence of diabetes and obesity. Coordinated efforts are needed to alter this trend and to prevent chronic diabetes in GDM patients and their offspring. LEVEL OF EVIDENCE: II-2

Stroke in Users of Low-Dose Oral Contraceptives

BACKGROUND Previous studies have linked the use of oral contraceptive agents to an increased risk of stroke, but those studies have been limited to oral contraceptives containing more estrogen than is now generally used. METHODS In a population-based, case-control study, we identified fatal and nonfatal strokes in female members of the California Kaiser Permanente Medical Care Program and who were 15 through 44 years of age. Matched controls were randomly selected from female members who had not had strokes. Information about the use of oral contraceptives (essentially limited to low-estrogen preparations) was obtained in interviews. RESULTS A total of 408 confirmed strokes occurred in a total of 1.1 million women during 3.6 million woman-years of observation. The incidence of stroke was thus 11.3 per 100,000 woman-years. On the basis of data from 295 women with stroke who were interviewed and their controls, the odds ratio for ischemic stroke among current users of oral contraceptives, as compared with former users and women who had never used such drugs, was 1.18 (95 percent confidence interval, 0.54 to 2.59) after adjustment for other risk factors for stroke. The adjusted odds ratio for hemorrhagic stroke was 1.14 (95 percent confidence interval, 0.60 to 2.16). With respect to the risk of hemorrhagic stroke, there was a positive interaction between the current use of oral contraceptives and smoking (odds ratio for women with both these factors, 3.64; 95 percent confidence interval, 0.95 to 13.87). CONCLUSIONS Stroke is rare among women of childbearing age. Low-estrogen oral-contraceptive preparations do not appear to increase the risk of stroke.

Body Mass Index in Midlife and Risk of Alzheimer Disease and Vascular Dementia

Prior work has suggested that obesity and overweight as measured by body mass index (BMI) increases risk of dementia. It is unknown if there is a difference in the risk of developing Alzheimer disease (AD) versus vascular dementia (VaD) associated with high body weight. The goal of this study was to examine the association between midlife BMI and risk of both AD and VaD an average of 36 years later in a large (N= 10,136) and diverse cohort of members of a health care delivery system. Participants aged 40-45 participated in health exams between 1964 and 1968. AD and VaD diagnoses were obtained from Neurology visits between January 1, 1994 and June 15, 2006. Those with diagnoses of general dementia from primary care providers were excluded from the study. BMI was analyzed in WHO categories of underweight, overweight and obese, as well as in subdivisions of WHO categories. All models were fully adjusted for age, education, race, sex, marital status, smoking, hyperlipidemia, hypertension, diabetes, ischemic heart disease and stroke. Cox proportional hazard models showed that compared to those with a normal BMI (18.5-24.9), those obese (BMI > or = 30) at midlife had a 3.10 fold increase in risk of AD (fully adjusted model, Hazard Ratio=3.10, 95% CI 2.19-4.38), and a five fold increase in risk of VaD (fully adjusted model, HR=5.01, 95% CI 2.98-8.43) while those overweight ( BMI > or = 25 and <30) had a two fold increase in risk of AD and VaD (fully adjusted model, HR=2.09, 95% CI 1.69-2.60 for AD and HR=1.95, 95% CI 1.29-2.96 for VaD). These data suggest that midlife BMI is strongly predictive of both AD and VaD, independent of stroke, cardiovascular and diabetes co morbidities. Future studies need to unveil the mechanisms between adiposity and excess risk of AD and VaD.

Insulin-like Growth Factors, Their Binding Proteins, and Prostate Cancer Risk: Analysis of Individual Patient Data from 12 Prospective Studies

Context Insulin-like growth factors (IGFs) and IGF binding proteins may be associated with some cancers. Contribution This reanalysis of individual patient data from 12 studies of the association between IGFs and IGF binding proteins and prostate cancer suggests that higher levels of serum IGF-I are associated with higher risk for prostate cancer. Caution The 12 studies varied in the types of patients they studied and in how they measured IGFs. Implication High IGF-I levels seem to be a risk factor for prostate cancer. The Editors Prostate cancer is one of the most common types of cancer in men, yet few risk factors for the disease, other than age, race, and a family history, have been established (1, 2). Insulin-like growth factors (IGFs) and their associated binding proteins (IGFBPs) have been the subject of many epidemiologic investigations of prostate cancer because they are known to help regulate cell proliferation, differentiation, and apoptosis (3). Although results from some, but not all, studies suggest an association between IGFs and IGFBPs and prostate cancer risk, there has been much uncertainty about its consistency and magnitude. A previous meta-analysis that included only 3 prospective studies suggested that high levels could be associated with more than a 2-fold increase in risk (4), although recent studies have suggested the risk is lower. Furthermore, given that these peptides are correlated with each other, uncertainty remains about any observed relationships. The individual studies are rarely large enough to allow proper mutual adjustment for these correlated factors, and they are insufficiently powered to investigate the consistency of their findings in key subgroups (for example, stage and grade of disease). Such analyses are important because studies have suggested that IGF-I might be more associated with advanced than with localized disease (5, 6). The Endogenous Hormones and Prostate Cancer Collaborative Group was established to conduct collaborative reanalyses of individual data from prospective studies on the relationships between circulating levels of sex hormones and IGFs and subsequent prostate cancer risk. Results for the sex hormones have been reported elsewhere and show no statistically significant relation between androgen or estrogen levels in men and the subsequent risk for prostate cancer (7). We report results for concentrations of IGFs and IGFBPs. Methods Participants The Endogenous Hormones and Prostate Cancer Collaborative Group is described in detail elsewhere (7). In brief, the group invited principal investigators of all studies, found by searching PubMed, Web of Science, and CancerLit, that provided data on circulating concentrations of sex steroids, IGFs or IGFBPs, and prostate cancer risk by using prospectively collected blood samples to join the collaboration. Thirteen studies collected data on circulating IGF concentrations and the subsequent risk for prostate cancer (5, 6, 820), of which 1 contributed only data on sex hormones (20). Eleven of the studies used a matched casecontrol design nested within a prospective cohort study (5, 6, 812, 16, 19) or a randomized trial (1315, 17). One study used a casecohort design (18) and was converted into a matched casecontrol design by randomly matching up to 3 control participants to each case patient by age at recruitment, time between blood collection and diagnosis, time of blood draw, and race. (Table 1 provides a full description of the studies and matching criteria used.) Most of the prospective studies were population-based, with the exception of 1 based on health plan members (9), 1 that recruited male health professionals (16), and 1 that was a combination of an intervention study and a monitoring study for cardiovascular disease (6, 10). Two of the randomized trials did not have prostate cancer as a primary end point (5, 8, 15); the other 2 were based within a screening trial (13) or were about treatment of prostate-specific antigen (PSA)detected prostate cancer (14). Table 1. Study Characteristics Individual participant data were available for age; height; weight; smoking status; alcohol consumption; marital status; socioeconomic status (assessed by educational achievement); race; concentrations of IGFs, IGFBPs, and endogenous sex steroids; and PSA level. Information sought about prostate cancer included date of diagnosis, stage and grade of disease, and method of case patient ascertainment. Some studies (5, 6, 8, 10, 16) published more than 1 article or performed assays at different times on the association between IGFs and prostate cancer risk, sometimes with different matched casecontrol sets, laboratory measurements, and durations of follow-up. For each study, we created a single data set in which each participant appeared only once. In our analysis, we treated any participant who appeared in a study as both a control participant and a case patient as a case patient only. We removed matched set identifiers, and we generated a series of strata (equivalent to matched sets) in which participants in each study were grouped according to age at recruitment (2-year age bands) and date of recruitment (by year), because these matching criteria were common to most studies (Table 1). The number of strata used in the collaborative analysis was slightly less than that of matched sets used in the original analyses. To ensure that this process did not introduce any bias, we checked that the results for each study, using the original matched sets, were the same as those using the strata described above. Tumors were classified as advanced if the tumor was described as extending beyond the prostate capsule (T3/T4), and/or there was lymph node involvement (N1/N2/N3), and/or there were distant metastases (M1); tumors were classified as localized if they were T0/T1/T2 and N0/NX and M0. We classified tumors as high-grade if they had a Gleason score of 7 or more or were moderately poorly or poorly differentiated; otherwise, they were classified as low-grade. Statistical Analysis We calculated partial correlation coefficients between log-transformed IGF and IGFBP concentrations among control participants, adjusted for age at blood collection (<50, 50 to 59, 60 to 69, or 70 years) and study. For each IGF and IGFBP, we categorized men into quintiles of IGF and IGFBP serum concentrations, with cut-points defined by the study-specific quintiles of the distribution within control participants. For studies with more than 1 publication or in which the serum assays were done at different times, resulting in different absolute levels of IGFs (5, 6, 8, 10, 16), we calculated cut-points separately for each substudy. We used a conditional logistic regression stratified by study, age at recruitment (2-year age bands), and date of recruitment (single year) as our main method of analysis. To provide a summary measure of risk, we calculated a linear trend by scoring the quintiles of the serum IGF or IGFBP concentrations as 0, 0.25, 0.5, 0.75, and 1. Under the assumption of linearity, a unit change in this trend variable is equivalent to the odds ratio (OR) comparing the highest with the lowest quintile. All results are unadjusted for participant characteristics, except for those controlled by the stratification variables. We examined the possible influence of 5 participant characteristics by adjusting the relevant conditional logistic regression models for body mass index (BMI) (<22.5, 22.5 to 24.9, 25.0 to 27.4, 27.5 to 29.9, or >30 kg/m2), marital status (married or cohabiting, or not married or cohabiting), educational status (did not attend college or university, or attended college or university), smoking (never, previous, or current), and alcohol consumption (<10 or 10 g/d). We excluded participants from the analysis if they had a missing value for the characteristic under examination. We assessed heterogeneity in linear trends among studies by using a chi-square statistic to test whether the study-specific ORs were statistically different from the overall OR (21). Heterogeneity among studies was also quantified by calculating the H and I 2 statistics (22). To test whether the linear trend OR estimates for each IGF and IGFBP varied according to case patient characteristics, we estimated a series of subsets for each characteristic: stage at diagnosis (localized or advanced), grade at diagnosis (low or high), year of diagnosis (before 1990, 1990 to 1994, or 1995 onward; these year cutoffs were chosen to attempt to reflect differences in the use of the PSA test for cancer detection), age at diagnosis (<60, 60 to 69, or 70 years), and time between blood collection and diagnosis (<3, 3 to 6, or 7 years). We excluded case patients from the analyses of stage and grade at diagnosis if the relevant information was not available. For each of these case patient characteristics, we calculated a heterogeneity chi-square statistic to assess whether the estimated ORs statistically differed from each other (21). To assess whether the OR estimate of the linear trend for each IGF or IGFBP varied according to PSA level at recruitment (<2 g/L or 2 g/L), we entered an interaction term into the conditional logistic regression model for each IGF or IGFBP, and we tested the statistical significance of the interaction term with a likelihood ratio test. Statistical significance was set at the 5% level. All statistical tests were 2-sided. All statistical analyses were done with Stata, version 9.0 (StataCorp, College Station, Texas). Results Table 1 shows the characteristics of the studies. The 12 prospective studies included approximately 3700 case patients with prostate cancer and 5200 control participants. Insulin-like growth factor I and IGFBP-III measurements were available for all and 3600 case patients, respectively. However, IGF-II and IGFBP-II measurements were available for only 379 and 419 case patients, respectively (Table 2). Mean age at blood collection