Because of the high prevalence of PI3K/Akt/mTOR pathway activation in GBM tumors,5
targeted therapies that specifically inhibit this signaling pathway are of high interest for incorporating into the treatment of patients with GBM. However, evaluation of response to such molecularly targeted treatments remains a challenge, because conventional imaging methods are often limited to the detection of anatomical changes. As a consequence, new metabolic biomarkers that inform on drug delivery and molecular target inhibition are critically needed. In this study, we evaluated the potential of 2 mechanistically linked biomarkers: PC as assessed by 31
P MRS and hyperpolarized lactate as assessed by hyperpolarized 13
To mimic the type of treatments currently in clinical trials and to address the generality of our observations, we monitored the effect of 2 inhibitors of the PI3K/Akt/mTOR pathway: the clinically relevant mTOR inhibitor everolimus and the PI3K inhibitor LY294002. Studies were performed in 3 GBM cells lines with distinct signature gene alterations, each of which leads to activation of the PI3K/Akt/mTOR pathway. Accordingly, on the basis of the recently published study from The Cancer Genome Atlas Research Network,5
the 3 cell lines studied collectively recapitulate the genetic make-up that would be expected in the majority of patients with GBM with regard to the PI3K/Akt/mTOR pathway.
Following signal inhibition, we observed a decrease in signaling downstream of mTOR and a decrease in HIF-1 transcriptional activity in all cell lines and with either treatment. These results are consistent with previous studies demonstrating a decrease in the translation of HIF-1α following inhibition of the PI3K signaling cascade.36,37
Our results are also consistent with previous reports of decreased HIF-1α levels in several cancer models following treatment with everolimus38
HIF-1 is known to regulate the expression of ChoKα23
and several glycolytic enzymes, including LDHA.24,25
We confirmed that the decrease in HIF-1 transcriptional activity following inhibition of the PI3K/Akt/mTOR pathway in our cell models was associated with a decrease in expression and activity of ChoK and LDH. Furthermore, the decrease in enzyme activities explains the reduction in PC and in hyperpolarized lactate that was observed in our cell lines following treatment with the PI3K or mTOR inhibitors. These observations are in line with a previous study demonstrating a decrease in PC following inhibition of HIF-1α in a colorectal cancer model.40
These findings also indicate that the 2 metabolic biomarkers, PC and hyperpolarized lactate, are mechanistically linked and can serve as complementary readouts of PI3K/Akt/mTOR signal inhibition upstream of HIF-1α.
Whereas HIF-1α expression is dependent on PI3K/Akt/mTOR signaling, the stability of HIF-1α depends on oxygen levels, and HIF-1α levels are typically higher under hypoxic conditions. Because our cell studies were performed under normoxic conditions (95% air/5% CO2
), it could be argued that our observations in cells will be obfuscated in the in vivo setting, wherein tumor cells are frequently hypoxic. However, in a parallel study, we recently showed that the observations described here are also applicable to the in vivo setting, at least for one of the cell lines investigated in this study.18
Specifically, we studied the effect of everolimus on GS-2 orthotopic GBM tumors in rats and demonstrated a significant decrease in the lactate-to-pyruvate ratio in treated animals when compared with controls. Furthermore, we demonstrated that this metabolic effect was mediated in vivo by a decrease in the transcriptional activity of HIF-1 and a decrease in LDHA expression, similar to findings in the current cell study. Of significance, the decrease in hyperpolarized lactate occurred in vivo over a week earlier than the effect on tumor size detected by standard MR anatomical imaging. Our finding that PC levels decrease following PI3K/Akt/mTOR signal inhibition is also consistent with our previous work in GBM tumors in which we observed a decrease in tCho using 1
H MRS in orthotopic U87 tumors.14
The drop in tCho is likely to have been a result of a decrease in PC, which is typically the main component of tCho in tumors.
As a control, we monitored the effect of temozolomide, a DNA-damaging agent commonly used to treat GBM2
but not expected to affect the PI3K/Akt/mTOR signaling pathway. Our results confirm that, in the cell lines examined, temozolomide treatment had no effects on P-4EBP1, CA-IX, and LDHA expression. Consistent with this, no differences in the levels of PC and hyperpolarized lactate were observed between control cells and temozolomide-treated cells for all 3 cell lines. Nonetheless, it should be noted that we have recently performed an in vivo study aimed at monitoring the effect of temozolomide in U87 orthotopic brain tumors in rats. In that study, a decrease in the pyruvate-to-lactate conversion following treatment was observed,41
and we have subsequently shown that this effect is specific to tumors that respond to treatment, while not being observed in tumors that express active O-6-methylguanine-DNA methyltransferase (MGMT), and therefore, do no respond to temozolomide.42
The underlying mechanism for this effect remains to be elucidated, but we have shown that it is not associated with a decrease in LDHA expression. A decrease in hyperpolarized lactate has also been observed in C6 gliomas following radiotherapy.43
Thus, whereas the drop in hyperpolarized lactate following signaling inhibition is not a specific metabolic effect associated with PI3K/Akt/mTOR inhibition, it nonetheless provides an informative biomarker of drug delivery and pathway inhibition by such targeted therapeutics.
From a clinical perspective, 1
H MRS is the spectroscopic method most widely implemented on clinical systems. It can be used to detect both tCho and lactate levels. As mentioned above, because tCho is comprised of Cho, PC, and GPC and because PC is typically the largest component of the tCho peak in cancer cells, 1
H MRS is likely to be as useful as 31
P MRS for detecting a decrease in PC and, thus, informing on the effect of treatment. However, although intrinsically less sensitive, 31
P MRS remains the method of choice for resolving phosphorylated compounds, such as PC, and approaches such as 1
H to 31
P polarization transfer have recently been developed to increase detection sensitivity at clinical field strength for potential use in clinical examinations.44
With regard to detection of lactate by 1
H MRS, this biomarker has been used in vivo for the diagnosis and evaluation of treatment response in preclinical and clinical settings, including with GBM.45,46
However, lactate detection by 1
H MRS presents several limitations. The lactate resonance overlaps with resonances of cerebral lipids, and the implementation of special editing methods is necessary to accurately measure lactate levels in vivo.47
Furthermore, the lactate level measured by 1
H MRS represents the total lactate pool, which is composed of both metabolically active and inactive lactate. Metabolically active lactate, which is lactate produced from pyruvate by live cells, can inform on the effects of treatment. The inactive lactate represents the pool of extracellular lactate, which is known to frequently accumulate in poorly vascularized necrotic regions and, thus, cannot reliably inform on response to treatment.48
In contrast to 1
H MRS, hyperpolarized 13
C MRS measures hyperpolarized lactate formed from exogenous hyperpolarized pyruvate. It therefore informs on the metabolism of live cells and can serve as a metabolic biomarker of treatment effects. Over the past few years, hyperpolarized 13
C MRS has shown tremendous potential for the diagnosis and follow-up of several types of cancer in vivo in preclinical models.13,18–20,33,49
This method is now in clinical trial in patients with prostate cancer at University of California, San Francisco (http://clinicaltrial.gov
) and is generating very promising results. Moving forward, this new imaging approach is therefore likely to be applied clinically to other cancer types and could serve as a noninvasive approach for monitoring response to novel PI3K/Akt/mTOR inhibitors.
In summary, we reported here that inhibition of the PI3K/Akt/mTOR signaling pathway can be monitored in GBM cells through the assessment of 2 noninvasive mechanistically linked metabolic MR biomarkers: PC and hyperpolarized lactate. Considering that the PI3K/Akt/mTOR pathway is activated in more than88% of GBM, such findings highlight the potential of MRS as a noninvasive imaging approach for informing on drug delivery and drug target modulation in patients with GBM who are treated with drugs that inhibit this pathway.