Multiplex Testing of 10 Actionable Lung Cancer Drivers Used to Select Therapy

Multiplex testing of 10 actionable cancer drivers found these drivers in 64% of patients and helped physicians in selecting therapies (JAMA 2014;311:1998-2006). Subjects who received therapies targeted to the actionable drivers had a median survival of 3.5 years versus 2.4 years in patients with actionable drivers who did not receive genotype-directed therapy; however, the authors cautioned that randomized clinical trials are needed to determine whether therapies selected based upon cancer drivers improve survival.

The study involved 1,102 eligible patients with stage IV or recurrent adenocarcinomas of the lung treated at 14 Lung Cancer Mutation Consortium (LCMC) sites, of whom 1,007 were tested for at least one gene and 733 for 10 genes. The researchers tested for 10 genes described as oncogenic drivers crucial to cancer development and maintenance using any one of three methods: matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; multiplexed single nucleotide extension sequencing; or Sanger sequencing with peptide nucleic acid probes. All sites also performed sizing electrophoresis to detectERBB2 insertions and EGFR deletions. In addition to ERBB2 and EGFR, the authors also tested for KRASALKNRASBRAFPIK3CAMEK1AKT1 and MET.  
KRAS mutations were the most frequent, found in 25% of patients, followed by EGFR (17%),ALK (8%), ERBB2 (3%), and BRAF (2%), with others less than 1%. In all, 24 patients had tumors with two oncogenic drivers. Treating physicians made the decision whether to use therapies targeted at the cancer drivers. LCMC sites reported whether patients received driver-targeted therapy as well as their survival times.

Overall, 28% of patients received driver-targeted therapy. Median survival time ranged from 4.9 years in patients with actionable drivers other than EGFR- and ALK-positive tumors who received targeted therapy, to 1.5 years in those with drivers who did not receive targeted therapy. However, since the study was not designed specifically to determine whether using targeted therapies improved survival times, the authors called for randomized clinical trials to address this issue.

Metformin Users With Reduced Renal Function at Risk of Lactic Acidosis

Risk of lactic acidosis or elevated lactate concentrations in diabetics taking metformin was significantly associated with reduced renal function (Diabetes Care 2014; DOI:12.2337/dc13-3023). The findings support guideline recommendations to monitor renal function in metformin users and to adjust the dose if estimated glomerular filtration rate (eGFR) drops <60 mL/min/1.73 m2.

This retrospective cohort study involved 223,968 diabetic metformin users and 32,571 diabetic patients who had never used metformin, identified through the Clinical Practice Research Datalink (CPRD), which collects electronic health records from general practitioners in the United Kingdom. All patients had at least one prescription for a non-insulin antidiabetic drug and were at least 18 years old. Researchers obtained renal function data by reviewing lab test data and CPRD READ codes, a coded thesaurus of clinical terms like SNOMED. The researchers defined lactic acidosis or elevated lactate concentrations as either a CPRD READ code for lactic acidosis or a record of plasma lactate concentration >5 mmol/L.

The authors found an overall incidence rate of lactic acidosis or elevated lactate concentration of 7.4 events per 100,000 person-years among current metformin users versus 2.2 events per 100,000 person-years among non-users. The risk for lactic acidosis or elevated lactate in metformin users with most recent eGFR <60 mL/min/1.73 m2, was associated with an adjusted hazard ratio of 6.37 in comparison to metformin never-users. This risk was further accentuated in patients with impaired renal function who had been on higher doses of metformin ≥730g within the past year or who had a recent high daily dose >2 g, with adjusted hazard ratios of 11.8 and 13.0, respectively.

Galectin-3 Linked to Cardiovascular-Related, All-Cause Mortality

Higher levels of galectin-3 (Gal-3) are independently associated with all-cause and cardiovascular disease (CVD) mortality and add prognostic information beyond that provided by natriuretic peptides in community-dwelling individuals who were free of known CVD at baseline (Am Heart J 2014;164:674-82.e1). 
The findings come from the Rancho Bernardo Study, an ongoing, prospective, population-based study of the epidemiology of CVD and other chronic diseases. Data from follow-up study visits of 1,781 participants between 1992-1996 served as the baseline for the current analysis. Gal-3 measurements were available in 1,742, of whom 349 had a history of CVD and were excluded from the study. The researchers continued to follow the participants through 2009 via periodic clinic visits and mailed questionnaires.

At baseline, nearly two-thirds of participants were women, with a mean age of 70 years. The median Gal-3 level was 14.8 ng/mL, and women had higher levels than men, 15.3 ng/mL versus 13.7 ng/mL. The mean follow-up period was 11.0 years, during which 436 subjects died, 169 from CVD, and 151 developed incident coronary heart disease (CHD).

The researchers found in models that adjusted for traditional CVD risk factors Gal-3 was a significant predictor of CVD mortality and all-cause mortality, with hazard ratios of 1.30 and 1.12, respectively. In contrast, Gal-3 was not a significant predictor of CHD. Gal-3 remained a significant independent predictor of CVD mortality after further adjustments for N-terminal pro B-type natriuretic peptide.

New, Dynamic Estimation Model of HbA1c from Blood Glucose Meters

Researchers at the University of Virginia in Charlottesville and Sanofi-Aventis in Germany have developed a dynamic estimation model for tracking average glycemia (HbA1c) over time based on even infrequent self-monitored blood glucose (SMBG) data (Diabetes Technol Ther 2014;16:303-9). The model is not intended as a substitute for laboratory-based HbA1c testing, but rather as means for diabetics to conveniently track their average glucose levels. It is not computationally demanding, so would be readily usable in SMBG devices with limited processing power.

The model considers two types of SMBG measurements: fasting glucose measurements and seven-point profile glucose measurements. It uses fasting glucose readings, expected to occur about every day, to compute the base glucose exposure tracking function of the model. Less-frequent profile measurements enable calibration of the model’s base glucose exposure function to the patient’s glucose variability. The model arrives at an estimated HbA1c based on these two data inputs, the former updated with any incoming fasting SMBG measurement and the latter with any incoming seven-point profile.

The authors used a training data set from 379 subjects to estimate the model parameters. After the model was fixed, they applied it to a different test data set from 375 subjects. The investigators assessed accuracy of the model by computing a mean absolute deviation (MAD) and a mean absolute relative deviation (MARD) of estimated HbA1c from reference HbA1c measurements, as well as the correlation between estimated and actual HbA1c. MAD was 0.50, MARD was 6.7% and the r correlation between estimated and actual HbA1c was 0.76. Using an HbA1c error grid plot, 77.5% and 97.9% of all estimated HbA1c were within 10% and 20% of reference HbA1c, respectively.

The authors suggested the utility of the model would come in providing diabetics an ongoing estimate of HbA1c outside of HbA1c testing they undergo at set intervals, as a means of optimizing their day-to-day diabetes control.