NEW YORK (GenomeWeb) – Researchers at London's Institute of Cancer Research have identified a set of metabolomic markers in the plasma of cancer patients that change in response to treatment with PI3 kinase inhibitors.
The study, which was published this month in Molecular Cancer Therapeutics, identified 22 plasma metabolites that exhibited time- and dose-dependent changes upon treatment with the PI3K inhibitor pictilisib (GDC-0941), indicating that metabolomic measurements could be helpful in guiding cancer therapy, Florence Raynaud, an ICR researcher and the senior author on the paper, told GenomeWeb.
Raynaud helped lead development of pictilisib in collaboration with ICR spinout Piramed, which was acquired by Roche's Genentech for $160 million in 2008. Genentech, which participated in the MCT study, owns the license to the drug. Raynaud described the recent effort as "proof of concept" aimed at demonstrating that metabolite measurements could detect aspects of patients' responses to the therapy.
"When we started off, I don't think there was much interest or much belief that this would actually work," she said, noting that the large number of factors that can affect a person's metabolome make it challenging to pick out specifically drug-related changes.
For example, metabolites can vary with time of day or due to diet or exercise habits, Raynaud said. "There are so many confounding factors that are potentially around."
She and her colleagues hoped that PI3K might prove an amenable system for their study, given its central role in a variety of metabolic processes. "We thought, if it doesn't work with PI3 kinase, it won't work with anything."
The results, though, "were very clear," she said, and she and her colleagues have since successfully applied similar metabolomic approaches to other cancer targets. "It may not be appropriate for all targets, but so far everything we have looked at has given us a different metabolic signature," she said. "So we believe that this is an attractive strategy."
As with other liquid biopsy-type approaches, the method allows for easier and more frequent sampling of patients than can be done using conventional tumor biopsies, which, Raynaud noted, at best typically allow for one pre- and one post-treatment sampling.
"As a biomarker, I think it is one extra technique in the batter of assays that we have at our disposal," she said.
Raynaud and her colleagues first looked for metabolomic markers in human xenograft mouse models of prostate cancer and glioblastoma and in PTEN knockout mice. They collected blood samples from these mice two, eight, and 24 hours after treatment, using three to six mice for each time point.
The researchers performed an initial untargeted mass spec screen to identifying plasma metabolites showing significant changes in response to treatment. They then confirmed these identifications via targeted quantitation on a Waters TQ-S triple quadrupole using the AbsoluteIDQ p180 kit from Biocrates Life Sciences, which, according to the company, allows for quantitation of 188 endogenous metabolites from five different compound classes.
These measurements identified 26 metabolites that showed significant change in response to pictilisib treatment.
Raynaud and her colleagues then followed up on these findings in 41 advanced cancer patients participating in a Phase I dose escalation clinical trial of pictilisib. Looking at 24 of the 26 markers identified in the mouse study — they discarded two that they were unable to measure with sufficient reproducibility — the researchers found that 22 metabolites showed dose- and time-dependent changes following single dose pictilisib treatment on day one. Prior to the second dose eight days later, none of the metabolites were significantly changed from their baseline levels, indicating that they returned to normal levels after the drug left the patients' systems.
A number of the metabolites identified are linked to insulin resistance and the development of diabetes mellitus. As the researchers noted in the MCT paper, this could prove important from a pharmacologic perspective in that dose-dependent hyperglycemia and hyperinsulinemia have been reported as side effects in Phase I clinical trials for pictilisib and buparlisib, a PI3K inhibitor being developed by Novartis.
In fact, they noted, "the adverse event of hyperglycemia was the dose-limiting toxicity of [buparlisib]."
"I think these are relevant biomarkers that you could use as you would use any other biomarkers, as biomarkers of drug action or toxicity or potentially of sensitivity," Raynaud said. "There is a broad range of applications."
She noted that the metabolome's inherent variability remains a challenge, but added that she and other researchers are working to nail down the sources of variability so that they can take them into account.
"We have [studies] looking at the effects of normal variation on the levels of these metabolites throughout the day," Raynaud said. "So say, for example, you know that this particular metabolite goes up with food or with time of day, and you see [in your study] that it is going down [upon drug treatment.] Then there is no way it could be due to that [normal variation]."
" I think there is a way that you can maybe mitigate the effect of some of these variations when you control your conditions very well," she said. "It is something that can be used under very controlled conditions and by being extremely careful about how you interpret the data."
She added that metabolites are in many cases easier to measure than markers like plasma proteins, which "are intrinsically more unstable and variable."
"The proteomics field has really suffered from that," Raynaud said. On the other hand, she noted, "we have measured these [metabolites] for years. These assays are easy to validate and automate, which makes them reliable."
She cited a study that she and her colleagues participated in evaluating the performance of the AbsoluteIDQ p180 kit in different labs and across different mass spec platforms.
"We all have different mass spectrometers, but what we did was analyze a range of samples, and I think what we found was remarkable," Raynaud said. Around 80 percent of metabolites had less than 20 percent variation, she asid. And although there were differences across labs and platforms, overall, a large number of the metabolites were highly reproducible across and within laboratories.
"So you can really run the assays for these metabolites to a very high analytical standard," she said.