MR spectroscopy readies role as cancer diagnosis, treatment tool

The idea of MR spectroscopy has been kicking around almost as long as MR itself. The allure is seductive: a noninvasive biopsy that quantifies in hard numbers the presence-or absence-of cancer.

This potential first surfaced almost 20 years ago at MR research conferences. But back then it was very long on the potential and short on the practical. Acquisitions were tedious to the extreme. And data, when finally acquired, took hours to process and match to images.

Improved electronics, pulse sequences, processing algorithms, and computing hardware have solved many of the early problems. Recent studies have proven the potential of MRS for the diagnosis and management of patients with prostate and breast cancer, as well as for the precise targeting of radiation to recurrent brain tumors.

Spectroscopy of the prostate can pinpoint the location of cancer. Breast MRS can dramatically reduce the need for needle biopsy. Brain spectra can help physicians differentiate between recurrent tumors and changes due to radiation treatments, just as it allows radiosurgeons to home in on cancer cells, extending the lives of patients with glioblastoma.

Siemens, Philips, and GE offer MR spectroscopy packages on their 1.5T and 3T systems. Each has automated acquisition and processing to increase ease of use and improve workflow. All believe their packages can contribute clinically valuable information for routine use.

But there’s still plenty of room for improvement. Peter B. Barker, Ph.D., a professor of radiology at Johns Hopkins University, makes no bones about the shortcomings of this technology. The methodology needs to be more robust. More analytical tools are needed. Radiologists have to be better educated in their use. And multicenter trials have to demonstrate diagnostic efficacy.

As an inherently quantitative technology, its numerical and graphical presentations must yet be validated against pathology and various stages of cancer development (staging cancer), said Keyvan Farahani, Ph.D., acting chief of the image-guided interventions branch of the Cancer Imaging Program in the National Cancer Institute. Standardized protocols are required to reproduce the quantitative diagnostic information across different subjects and different sites, he said.

When these hurdles are cleared, MRS still must win reimbursement. And even then, providers will need the motivation to apply the technology. Advocates would like to see a trigger, some seminal development that creates enthusiasm for MR spectroscopy.

“While there have been interesting research findings, there has not been the kind of ‘killer app’ that would justify its [widespread] use,” said Dr. Peter L. Choyke, director of the Molecular Imaging Program at the National Institutes of Health.

Researchers might not have found it yet, but some are getting close. Choyke is one of them.

PROSTATE POTENTIAL

Choyke and his NIH team have done much to advance the prospects of MR spectroscopy of the prostate. A recent multiparametric study brought MRS together with T2-weighted and dynamic contrast-enhanced MR imaging.

The team concluded that utilizing these several techniques at 3T enabled tumor detection “with reasonable sensitivity and specificity” in 70 patients whose prostate cancer had been proven with biopsy for peripheral zone tumors. But while underscoring the potential of the technology, the research exemplifies the underlying weakness of spectroscopy, as its clinical value is not in dispute so much as its practicality.

“MRS is more difficult to set up, more lengthy to acquire, and more complex to interpret than other MRI sequences,” Choyke said.

The University of California, San Francisco has tinkered with the protocols needed to make MR spectroscopy a clinical tool for assessing the prostate. Their work, like that of others, has leveraged 3T to generate the signal-to-noise ratio (SNR) needed for spectroscopy and other MR techniques, such as diffusion imaging and T2-mapping, to provide quantitative and functional information to improve sensitivity and specificity.

Early efforts using a body coil for excitation and a Medrad inflatable endorectal coil ran aground due to artifacts associated with sampling problems encountered during echoplanar imaging. The artifacts were traced back to the air-filled space created by the inflated endorectal coil. The UCSF team solved the problem by filling this space with a nontoxic fluid that creates an MR-friendly interface with the tissue.

Early work showed the potential of this approach in prostate cancer patients whose exams produced artifact-free, high-quality T2 maps and apparent diffusion coefficient (ADC) value maps in diffusion-weighted imaging and spot-on MR spectrographs. The reduced citrate and elevated choline levels seen in those spectrographs corresponded to biopsy-proven prostate cancer. John Kurhanewicz, Ph.D., and UCSF colleagues have since performed thousands of prostate MRS studies.

“Sequences that acquire the volume of interest have gotten better, along with suppression pulses that assure us we are getting what we are looking for,” said Kurhanewicz, a UCSF professor of radiology and biomedical imaging. “We can come up right to the edge of the prostate, which is critical because many clinically important cancers are right on the periphery of the gland.”

The coils for gathering those signals have also gotten better and 3T scanners, which are becoming increasingly common, deliver stronger SNR compared with the old standard bearer, 1.5T. Leveraging these advances, the UCSF team has generated results they will present at the International Society for Magnetic Resonance in Medicine meeting this month showing that prostate MRS studies can be done in five minutes.

As impressive as these results are, they only hint at what might be achieved as the technology continues to evolve. The greatest promise lies with the highest field strengths.

Research done by Leo L. Cheng, Ph.D., an assistant professor of radiology and pathology at Harvard, has used a 7T scanner to probe metabolites found in excised prostates. The ultimate goal of the research is to establish a case for MR spectroscopy as the means for replacing biopsies, which are not only invasive but inherently inaccurate. Prostate tumors may be confined to only a small part of the prostate, Cheng said. Biopsies, therefore, can return a false negative because the area sampled was healthy even when there is a malignancy only a short distance away.

Research performed on five cancerous prostate glands removed from patients showed that analyses done on metabolites in the excised glands could be cranked into a “malignancy index” that correlates well with the presence of cancer.

“As we analyze more and more tumors with spectroscopy, we should be able to define profiles that reflect specific clinical and pathological states, achieving a true needle-free MR biopsy, Cheng said.

Because the studies were done on a 7T whole-body clinical MR scanner at the Massachusetts General Hospital Martinos Center for Biomedical Imaging, the approach should be adaptable to patients, according to Cheng. If these efforts are successful, the next step will be to migrate the approach to 3T. Over the longer term, he expects that MR spectra will yield patterns or profiles that indicate the location of tumors in the prostate and even their aggressiveness.

EXAMINING THE BREAST

MRI is more sensitive than x-ray–based mammography in finding the signs of breast cancer. But it is also less specific. The result is a high number of false positives, leading to unnecessary biopsies. Adding MR spectroscopy could improve specificity, according to Dr. Lia Bartella, director of breast imaging at Eastside Diagnostic Imaging in New York City.

Research done by Bartella and colleagues at Memorial Sloan-Kettering Cancer Center has shown that adding MRS to breast MRI dramatically reduces the need for biopsy. MR spectroscopy adds just 10 minutes to a standard MR exam and provides a biochemical snapshot that many times indicates whether the lesion is cancerous.

Her work centers on nonmass enhancing breast lesions, which are characterized by enhancement of an area that is neither a mass nor a lump. It may extend over a large or small area. These lesions can occur in the presence of benign hormonal changes, but they can also indicate cancer. Biopsy is often required to make the determination. But MRS shows potential for eliminating at least some of that need.

At Sloan-Kettering, Bartella and colleagues found elevated levels of choline in 15 of 32 lesions. Twelve of the 15 were cancerous, giving MRS a specificity rate of 85% and sensitivity of 100%.

Research now under way at Sloan-Kettering may further document the diagnostic benefits of MR spectroscopy and even extend the reach of this modality as a predictor of patient response to therapy. One team is tacking an MRS pulse sequence on the end of conventional breast MR exams at 1.5T and 3T to see if spectra can be used to distinguish benign and malignant lesions. The MRS data will be compared with biopsy results.

Another team is performing MRS at 1.5T on patients with biopsy-proven breast cancer who have begun neoadjuvant chemotherapy before surgery to remove the tumor. The spectrographic data are being analyzed to determine whether MRS can predict therapeutic response.

DELVING INTO THE BRAIN

The skull can be a complicating factor in MR spectroscopy, and interfere with the acquisition of spectra. But easy accessibility, lack of motion, and minimal fat content make the brain a relatively easy target for the modality. The irony is that many times MR imaging can distinguish benign and malignant tumors on its own. The need for MRS as a diagnostic tool is diminished even more when you consider that brain biopsies are performed on all patients before therapy is ordered.

But MRS can come in handy when differentiating recurrent tumors from tissue damaged by radiation therapy. Inflammation and tumor cells look a lot alike on MR images. Assessing the levels of choline, creatine, and N-acetylaspartate with MRS can help tell the difference, according to research by Dr. Patrick N. Weybright.

The research was done five years ago as part of a fellowship at the University of Michigan Medical Center. Weybright, now a radiologist with Lancaster Radiology Associates in Lancaster, PA, would seem ideally positioned to move MR spectroscopy from academia into routine clinical use. But since completing his neuroradiology fellowship, he hasn’t found spectroscopy being used much in private practice. In four years of private practice, he says he has read only about 10 MRS studies.

“Most radiologists and neurosurgeons are not sufficiently confident in spectroscopic results that they are willing to hang their hat on them,” he said.

Complicating matters is the difficulty that goes with acquiring MR spectra, even from the brain. Interfaces with bone and cerebrospinal fluid can cause artifacts that take hours to work around, Weybright said. Despite these caveats, he remains optimistic that MRS eventually will prove to be a clinically useful tool.

“MRS is a very nifty way to assess the biochemical thumbprint of a tissue,” he said.

When it comes to interventions, the precision possible with MR spectroscopy can become a factor in managing patients. Researchers from University Hospitals Case Medical Center report promising results treating glioblastoma multiforme with a Gamma Knife precisely focused on the cancers using MR spectroscopy. MRS allowed the Case Medical research team to concentrate high-dose radiation on GBM regions that were more aggressive than others. The end result was a 3.7-month average increase in the life span of patients with this almost universally fatal disease.

Despite its successes and obvious potential as a diagnostic, prognostic, and therapeutic enabler, MR spectroscopy continues to skirt the periphery of medical practice. It simultaneously beckons would-be adopters with its clinical promise, just as it puts them off them with its complexity. Strong cases can be made for MRS in the brain, breast, and prostate.

“As a research [tool], MRS has provided unparalleled insights into disease,” Choyke said. “It has already improved our understanding of the origins and biochemistry of some disease states.”

Yet the need for specialized skills and for time to acquire and analyze the data have so far restricted its routine use. For MRS to take hold in the clinical mainstream, it needs an application for which its value is indisputable, Choyke said.

Prostate MRS may provide that. Kurhanewicz argues that the benefits coming from prostate MRS will be obvious within a couple of years. The UCSF team is working with GE Healthcare to commercialize a platform that will include fast sequences capable of acquiring spectral data in five minutes.

“We will need to validate [this platform] to show the world that this is push-button technology,” he said. “But then it will be a done deal and people will be picking it up and doing it all the time.”