September 2008: Volume 34, Number 9
Genes Predict Survival in Lung Cancer Patients
Prognostic models for early-stage lung cancer should include both gene expression and clinical parameters such as patient age, sex, and stage of disease, new research concludes. The retrospective study, published in Nature Medicine (2008;14:822–7), examined the validity of the various biomarkers and gene signatures used to classify patients with non-small-cell lung cancer. Data from 442 cancer patients from six institutions were collected and analyzed based on gene expression and patient characteristics. The investigators then divided the patients into four groups, based on the degree of survival associated with each. Two sets constituted training sets for development of predictive models; the other two sets were for validation. The investigators then modeled risk profiles by scoring and classifying each case based on one of the following methods: gene cluster expression summaries, expression of individual genes, principal components of gene expression, membership in clusters defined by gene expression or a vote of single-gene classifiers. According to the researchers, only the first and last methods showed statistically significant predictive value of survival.
Fetuin-A Levels Predict Incident Diabetes in Elderly
A retrospective case-cohort study reports that among participants age 70–79 years, the rate of incident diabetes was associated with higher levels of fetuin-A. The findings, reported in JAMA (2008;300:182–8), were based on a cohort of 3,075 participants recruited from the Health ABC Study. The researchers assessed Medicare participants for fasting blood glucose at the second, fourth, and sixth annual visits. A total of 500 patients were randomly selected for stratification by age and race. Of this group, 135 were diagnosed with incident diabetes during 6 years of follow-up. Participants in this subcohort were 50% female and 50% African American. Of 575 not stratified, 112 had incident diabetes. Levels of fetuin-A were divided into tertiles; tertile 1 (≤0.76 g/L) served as the reference range. Patients in tertile 2 (0.77-0.97 g/L), after adjustment for age, sex, race, visceral adiposity, PAI-1, adiponectin, leptin, TNF, and IL-6, had an odds ratio of 1.83 for incident diabetes. For patients in tertile 3 (> 0.97 g/L), the OR was 2.44 after adjustment for all such factors. These results suggest fetuin-A could be an important marker for detecting an increased risk for diabetes in elderly persons. In addition, “blockade of fetuin-A binding to the insulin receptor might be considered a novel therapeutic target for prevention or treatment of insulin-resistant states,” the authors wrote.
Thyrotropin Levels Linked to Alzheimer’s Disease in Women
According to new findings from the Framingham Study, low and high levels of thyrotropin are associated with increased risk of Alzheimer’s disease (AD) in women (Arch Intern Med 2008;168:1514–20). The investigators analyzed the relationship between baseline levels of thyrotropin from blood collected in 1977-1979 and the incidence of subsequent AD in 1,864 male and female participants. They found that women with the lowest (< 1.0 mU/L) and highest (>2.1-50.5 mU/L) levels of thyrotropin had more than double the incidence of AD compared to women with levels in the middle range (1.0-2.1 mU/L). The adjusted hazard ratio for AD in the lowest-tertile group was 2.39, compared with the middle-tertile patients, and 2.39 for the highest-tertile group. In men, however, there was no such relationship. There were 51 cases of AD among the 330 women in the low tertile, 25 cases of 313 in the middle tertile, and 45 cases out of 305 women in the high tertile. Mean patient age at baseline was 71 years in women and 70 years in men. The authors noted that there is no clear explanation for the association between low levels of thyrotropin and susceptibility to Alzheimer’s disease in women. It might be a consequence, rather than a cause, of Alzheimer’s disease, they wrote, in that neurodegeneration might cause the hypothalamus to secrete less thyrotropin or cause the pituitary gland to be less responsive to thyrotropin-releasing hormone.
Performance of Rapid HIV Tests in the ED
A study of an on-site screening test for HIV in the emergency department at Brigham and Women’s Hospital, Boston, found that an oral rapid test to detect the virus had a lower specificity than the investigators had expected and lower than that listed by the test’s manufacturer. The researchers, who noted that this result was consistent with earlier studies of such tests when tested in actual field settings, reported findings from 849 participants in the Annals of Internal Medicine (2008;149:153–60). Of these patients, 810 had a nonreactive result, and 39 had a reactive result. Of those with a reactive result, 31 consented to have the result confirmed; 26 of those 31 were found not to be infected with HIV. Based on their study, the investigators estimate the HIV test sensitivity to be 96.9% (95% CI, 95.7% to 98.1%) in this ED setting. According to the manufacturer, the sensitivity is 99%. The higher rate of false-positives than that indicated by the test manufacturer points to the need for confirmation of oral tests as quickly as possible—perhaps by making HIV RNA testing part of such confirmation testing, the investigators noted. Testing for the virus’s RNA by PCR amplification, which is used in blood banking, detects HIV at an earlier stage than standard antibody testing and may help catch more new infections.