A Mayo Clinic meta-analysis of published studies shows that gene array-based profiling is no better than conventional histological methods at prognosing lung cancer.
“There are many studies currently ongoing using microarrays, but many of them may not be useful clinically,” Zhifu Sun, a research associate in the Department of Health Sciences Research at the Mayo Clinic, told Pharmacogenomics Reporter last week.
“If we find something that is just a surrogate of the same thing that we currently measure, [then] that may be not very useful,” he said. “The new measure should be even better than the current one.”
Sun was the main author in the study, “Gene Expression Profiling on Lung Cancer Outcome Prediction: Present Clinical Value and Future Premise,” which reviewed recently published outcomes studies in lung cancer using microarray technology and found that gene expression profiling is “not quite perfected.” The studies used technologies such as cDNA arrays and Affymetrix’s GeneChip platform. Sun’s report appears in the November issue of Cancer Epidemiology.
Sun discussed the role of tumor cell type in lung cancer prognosis to note that microarray technology doesn’t necessary add value to cancer treatment over conventional methods.
According to Sun, oncologists can use histological predictors and microarray technology to determine tumor cell type. Therefore, in lung cancer, molecular profiling may be a surrogate to, but not necessarily better than, conventional methods. “They measure the same thing but from different angles,” said Sun. “So, they are redundant. The microarray is not a cheap test. It costs a lot right now.”
The cost effectiveness of genetic tests is currently being debated by disparate groups within health care. For instance, one collaborative study between the Mayo Clinic and the pharmacy benefit manager Medco is looking into the pharmacoeconomics of genetic testing in warfarin therapy [see PGx Reporter 12-06-06]. The Medco/Mayo alliance comes on the heels of a Harvard Partners Center for Genetics and Genomics study investigating the clinical and economic utility of using genetic data in warfarin therapy [see PGx Reporter 11-15-06].
Cost effectiveness of genetic tests may be a particularly meaningful consideration in the oncology arena, given the hefty price tag of cancer therapies.
The Problem With Microarrays
Sun’s Mayo study notes that “when conventional predictors of age, gender, stage, cell type, and tumor grade are considered collectively, the predictive advantage of the gene expression profile diminishes.”
The study authors conclude that outcome prediction from gene expression signatures can be explained by conventional predictors, “particularly histologic subtype and grade of differentiation.”
“Any new technique that does not significantly outperform less expensive and easily conducted approaches is less likely to be useful in clinical practice,” the authors said.
In their review, Sun and Yang identified several problems with gene expression profiling, including: variability in the accuracy of gene expression-based outcome prediction; a lack of independent validation; overlap between gene expression profiling and pathological prediction of clinical outcomes; a preference of current technology toward highly expressed genes related to tumor differentiation, not clinical outcomes; and overlooking genes expressed at low levels, which may be more clinically meaningful.
When comparing gene expression profiling to conventional, histological methods, the study authors found that studies using gene-based signature panels achieved “very good” prediction for patient survival but did not outperform the prediction using the five conventional variables of age, gender, stage, cell type and tumor grade.
“The microarray is not a cheap test. It costs a lot right now.”
“Right now survival is impacted by many factors,” Sun said. “In the outcomes you have to consider whether the patient dies not from the disease but for other reasons. Then you have to remove the patient. That means that comorbidity of the patient is the reason for the death, not the disease. So you have to consider that when you use molecular profiling.”
The variability of data from gene expression profiling is another issue plaguing microarray technology.
“The main thing is the reproducibility because if you look at the literature there are very few overlaps from different labs,” Sun said. “That’s the major barrier to moving in this direction [of molecular profiling]. If you have to test this kind of marker in the clinic you have to come up with some type of consistency. … The variability may come from the different labs or from the different set of patients.”
The study authors ultimately recommend several remedies to improve the clinical utility of gene panel testing. They encourage medical scientists working with gene expression profiling to: (1) form a clearly defined study aim, (2) lay out and compare alternative study designs, (3) select tumor samples in terms of size, quality, and unambiguous clinical outcomes, (4) provide clinical relevant interpretation from the study results, and (5) understand the limitations of microarray testing.
“Be fully aware of the limitations of DNA microarray [testing] and of what you are expecting for a chosen platform,” the study authors suggest. “Particularly, [the] DNA microarray is prone to various sources of variations, and genes at low expression are less reliably detected if possible at all.”
In Sun’s view, “lung cancer is a major problem because it’s the number one killer in terms of cancer diseases,” but “has very few reliable predictor or prognosis factors to stratify the patients in the clinic to customize treatment.”
The study, which Sun conducted along with fellow Mayo Clinic researcher Ping Yang, was funded by the National Institutes of Health, the National Cancer Institute and the Mayo Clinic. According to Sun, the study reviewed approximately 25 studies published between 2000 and early 2006. Most of the studies were conducted in academia and not by industry.