April 22 - 23, 1999: Rockville, MD
On April 22 - 23, 1999 experts in clinical magnetic resonance spectroscopy met for two days at the DoubleTree Hotel in Rockville, Maryland to discuss the current state-of-the-art, potential future progress in, and benefit of, magnetic resonance spectroscopy (MRS) in clinical oncology. The specific goal of the focus group was to formulate recommendations that would assist the NCI in its programmatic decision making. Specifically, the focus group identified five areas that deserve renewed and increased attention by the NCI.
Below, these five areas are concisely described. Following that, an appendix is included that provides brief summaries of many of the presentations made by focus group participants, a roster of participants, and the meeting agenda.
Focus Group Recommendations
1. NMR Spectroscopy: Clinical Research and Multi-Center Studies
The goal of funding clinical MR spectroscopy studies is to facilitate introduction of NMR spectroscopy/imaging to well-designed clinical studies to determine their clinical utility. The introduction of this expensive and technically demanding but non-invasive method of monitoring tissue biochemistry and physiology will require academic, industrial and governmental cooperation and interaction. The necessity of industrial support [hardware specific for spectroscopic applications (decoupler, transmit/receive coils, etc.)], improved reliability of equipment and improvement in the speed of repairs, and stable system platforms is paramount and is discussed separately. The need for robust analytical techniques and standardization cannot be overemphasized and is also discussed in a different section. These are important for obtaining maximum data from a given study, and obtaining reliable multi-voxel data reproducibly and reliably. This section focuses on potential projects recommended to the National Cancer Institute for funding to facilitate the introduction of quantitative NMR measurements to oncology.
A potential mechanism for some of these funding requirements may fall under ACRIN or cooperative groups such as the Pediatric Brain Tumor Consortium (PBTC). This is a new mechanism and needs to be explored and possibly expanded further.
A. The NCI should support single institution studies designed to obtain pilot data. This research will test novel ideas and obtain preliminary data. This research falls into a "high risk" category, but with important potential gain. It thus requires a fast "turnaround time", and should provide ca. $100,000 to an investigator with a clinically related project that is "high risk, high yield". The R21 mechanism might be suitable but needs to be streamlined (both in length of application and response time).
B. The NCI should support rapid funding of "add-on" spectroscopic or imaging studies to ongoing clinical trials. This would allow accumulation of important preliminary data on technological feasibility, or on clinical questions in select populations. It would allow NMR investigators to take advantage of the infrastructure from the clinical trial (statistics, uniform patient population, data base etc.) and provide new data to the ongoing clinical trial. This could be coordinated through CTEP.
C. The NCI should support multi-center studies falling into two categories, those involving a small number of institutions (2 - 4) and those involving larger groups of institutions.
(i) Multi-center studies involving a small number of (2-4) institutions would have the dual goals of technology transfer and validating how "robust" the technique is. These studies would obtain preliminary data focused on a significant clinical question for subsequent larger studies. Examples of these include (but are not restricted to): directed biopsy, diagnosis (image guided), and treatment planning. Possible disease sites should include: (i) prostate - 1H NMR, (ii) brain tumors (pediatric and adult) - 1H NMR, (iii) breast - 1H NMR (e.g., choline), and (iv) lymphoma - 31P NMR [note, pharmacokinetics (such as with fluorinated pyrimidines), while not a "disease site", would fall into this category].
(ii) Multi-center studies involving a larger number of institutions (>4) would focus on collaborations that investigate significant clinical questions and obtain definitive data. These studies will include a short preparation period (~6 months) to ensure successful technology and uniformity of techniques, prior to obtaining "definitive data". Three areas have sufficient preliminary data to warrant consideration. The use of 31P MRS in continued investigation of the predictive value of the phosphomonoester region as an early marker of tumor response, the use of 1H MRS for study of prostate cancer, and the use of 1H MRS for evaluation of brain tumors
D. The NCI should support the further training of physicians in MRS. This will be required if the potential of these techniques is to be realized. This should focus on training individuals in the correct interpretation of NMR spectra, obtained with relatively complex techniques. Possible NCI programs/mechanisms for this activity are: the T32 program for fellows and house staff and The K23 program, with a Ph.D. spectroscopist as mentor.
2. Instrument Specifications and Relations with Commercial Manufactures
Clinical applications of MRS in cancer have been sorely compromised by inadequate availability and maintenance of MRS capabilities on clinical MRI/MRS scanners. The maintenance of robust MRS capabilities on scanners located in clinical environments is essential for ongoing clinical MRS research.
A The NCI should invite representatives from relevant MRI scanner and ancillary equipment manufacturers, and key MRS investigators to participate in a discussion to address the problem of establishing and maintaining MRS functionality on whole-body MRI/MRS scanners for the purpose of cancer research and clinical studies. Specifically, to discuss the maintenance of basic proton (1H) MRS functionality, broadband MRS functionality, the ability to image nuclei other-than-1H, state-of-the-art detector coils and associated hardware, and MRS data processing. Invitees should include major clinical scanner manufacturers (GE, Siemens, Philips, Picker, etc), small-bore MR instrument and third party MR equipment manufacturers (Varian, Bruker, SMIS, MEDRAD etc). The discussions should address the timely maintenance of the above capabilities in system upgrades as it affects ongoing NCI sponsored research and, especially, multi-center trials. The discussions should include possible mechanisms for NCI support of industry or partnerships for multi-center trials, as well as safety, legal, and proprietary issues associated with third party suppliers.
B. The NCI should provide programmatic means whereby manufacturers, and moreover, appropriately qualified scientists and engineers from the manufacturers' research and development effort, can be invited to participate directly in multi-institutional trials as partners. The focus group feels that a binding commitment from equipment manufacturers for maintaining MRS functionality over the course of a multi-year multi-institutional research study is essential to the success of such studies.
C. The NCI should develop funding mechanisms to directly support industrial development of technologies key to human MRS studies including state-of-the-art hardware (optimized coils, phased-arrays, non-1H quadrature detectors etc), software (pulse sequences etc), and data processing. The focus group recognizes that the development and inclusion of MRS is ideally addressed in a MRS product fully integrated into standard clinical commercial clinical scanners. The NCI MRS Focus group strongly recommends that the major commercial MRI scanner manufacturers embrace MRS in such a fully integrated product. This is the preferable solution for the long term but its realization faces an uphill battle given the reluctance to date of the major commercial MRI scanner manufactures to embrace MRS. However, the significance and potential impact of MRS in cancer that is evident from the data available so far urge that a short term solution be supported that will ensure the timely availability of clinical MRS in ongoing studies. NCI support of niche manufacturers and third party vendors may provide a viable means of addressing the need for state-of-the-art MRS technology in the short term.
3. High Field MR Systems for Cancer Research
In vivo MR spectroscopy is limited by sensitivity and spectral resolution. These limitations can be overcome by making available MR magnets that are at higher field strength than current clinical MR magnets possessing field strengths £ 1.5 tesla. With higher magnetic fields (³ 3 tesla), significant gains in sensitivity and spectral resolution are possible, and these improvements will have a major impact on clinical cancer research. Currently, access to high field human systems for cancer research is very limited.
A. The NCI should provide programmatic means to make available high field MR systems at NCI funded Cancer Centers as well as other institutions demonstrating an equivalent level of clinical cancer research. It is recommended that the magnetic fields of such high field systems be ³ 3 tesla.
B. The NCI should provide programmatic support for technical development (e.g., transmit/receive coils, pulse sequences, etc.) specific to the high magnetic field clinical research MRS applications.
4. Technical Development
Clinical MRS is a technically demanding modality, which in general has been only partially developed, supported and exploited by the manufacturers. It has also been difficult to obtain investigator initiated funding for the development of the software and hardware necessary to support the evolution and optimization of clinical MR studies.
A. Specific NCI programmatic funding should be allocated for the development of pulse sequences, data analysis, spectroscopic standardization, informatics technology, display strategies and methods for the integration of MRS data with the large range of other clinical information.
B. Likewise, NCI support should be directed toward the development of robust, high-speed shimming algorithms and multinuclear transmit/receive coils.
C. In addition we recommend the funding of specific training programs for clinical MRS at all levels. (See also, 1D above.)
5. Pre-clinical research
Applications of MRS to cancer have a history of strong and fruitful interactions between clinical and pre-clinical (basic) research. This has involved a two-way transfer of information: i.e. discoveries made in animal models have been used to formulate hypotheses underlying clinical research, and clinical observations have spawned important research into underlying biochemical mechanisms. For example, dramatic changes in phosphomonoester levels in response to therapy, first observed by 31P-MRS in humans, have been characterized in tumor xenograft models. Similarly, the consistent 1H-MRS visible difference in choline levels between malignant and normal brain, prostate and breast tissues has stimulated basic research into the metabolism of these compounds in cultured cells and tumors.
Current technical developments in MRS also place this technology at the cutting edge of non-invasively investigating the distribution and metabolism of conventional and novel therapeutic agents, such as transgene expression. Significant differences between humans and animal/cellular models must, however, be taken into account. The overall goals for MRS involving model systems are to help cancer prognosis and to improve cancer therapy. Toward this end we recommend NCI programmatic funding directed at the four fundamental, basic research areas described below.
A. The NCI should provide support for research directed toward the delineation of the mechanisms underlying metabolic processes observed by MRS in the clinic. Knowledge of the biochemistry underlying spectral changes is essential to the rational utilization of MRS in clinical practice. More profound basic knowledge could be used to improve the sensitivity or specificity of the clinical MRS exams.
B. The NCI should provide support for basic MRS research directed toward pharmacokinetics and pharmacodynamics and other biochemical methods of enhancing tumor response. Development of novel therapeutic agents requires accurate knowledge of their biodistribution and metabolism. This can involve the use of MRS markers as early surrogate endpoints, or actual monitoring of drugs, gene therapy or immunotherapy. MRS-visible metabolites can change early in response to successful therapy. Hence, monitoring tumor response with MRS can be used to predict the eventual effectiveness of the putative therapy. Additionally, the regional delivery of therapies can be directly monitored with MRS and MRI.
C The NCI should provide support for basic research directed toward development of novel MRS-visible markers of tumor-specific physiology or metabolism. There is a long history of developing exogenous markers of in vivo physiology and metabolism and this effort should continue. For example, indicators of phospholipid metabolism, pH, Ca2+, pO2 and compartment specific volume have all been successfully applied to tumor model systems.
D The NIH should provide support for the emerging application of exogenous markers or agents that recognize specific enzymes or gene expression products to generate MRS-visible signals. It is conceivable that methods could be developed to not only identify tumors at different stages, but also to interact with signaling pathways to abort carcinogenesis.
A.1. Summaries of Presentations by Focus Group Participants.
Jeffry R. Alger
University of California, Los Angeles
Magnetic Resonance Spectroscopy of Brain Cancer
The presentation summarized background and recent findings related to MRS characteristics of primary glioma and the clinical implementation of routine MRS scanning for glioma. The presentation did not consider metastatic infiltration of the brain; this condition has not been significantly evaluated with MRS. Rarer forms of primary glioma were also not discussed. The discussion of pediatric glioma was deferred to Dr. June Taylor.
Primary glioma represents a significant health problem. Each year, there are approximately 35,000 new cases of primary glioma diagnosed in the United States and 17,000 deaths are annually attributed to this condition. Over half the newly diagnosed cases are gliobastoma multiforme; these patients have a statistical life expectancy of about 9 months. There is epidemiological evidence supporting an increased incidence of primary glioma beginning in the 1960's. The histopathological assessment of readily identifiable microscopic features (pleomorphism, endothelial proliferation, mitoses and necrosis) within a standardized simplified classification system developed at the Mayo Clinic is used to grade primary glioma. The presently-used histopathological classification system is justified by its prognostic value; it predicts survival well. This emphasizes that diagnosis and prognosis are not independent of each other. Hence when new imaging technologies such as MRS are assessed, one can not arbitrarily separate their diagnostic and prognostic capabilities.
The first proton MRS study of human glioma appeared in 1989. The majority of subsequent papers and the clinical use emphasis has been on proton (1H) MRS. Nevertheless, there are some examples of successful 31P MRS studies of human glioma. Technological factors have biased the progress in the direction of proton MRS. The 1989 paper addressed the possibility of obtaining a preoperative (noninvasive) diagnosis of tumor type and grade using proton MRS. Since this time there have been more that 60 publications reporting the results of single and multiple center studies of the MRS characteristics of glioma. The majority of the existing work points toward the ability of MRS to make preoperative diagnoses. Given that surgical resection is palliative for glioma and that a growing armamentarium of neurosurgical navigation tools now permits safe neurosurgery, patients having suspected glioma are rarely treated without surgical debulking or biopsy. The need for a noninvasive diagnostic procedure, such as MRS is thereby lessened. On the other hand, the glioma management represents an area of opportunity for MRS. MRS can provide valuable clues regarding when to initiate further palliative treatment and whether treatments are being effective in reducing tumor burden.
Paul A. Bottomley
Johns Hopkins University
Metabolic Imaging of Nuclei other than Hydrogen.
While our research has focused primarily on developing metabolite quantification and imaging technologies for the human heart, these technologies are also directly applicable to human cancer studies. Specifically, we have developed methods for quantifying absolute concentrations of the creatine kinase metabolites, phosphocreatine and adenosine triphosphate using phosphorus (31P) MRS and total creatine using 1H MRS, and of total sodium with sodium (23Na) MRI (1-3). An important focus is the development of techniques to directly image metabolites on a clinical 1.5 T research scanner (3). Technologies that have long been established and proven advantageous for clinical 1H MRI such as phased-arrays and quadrature detection are simply unavailable for MRS on many clinical scanners. Nor is MRI commercially available for non-1H nuclei. We have developed our own broadband phased-array system for use with a conventional narrowband 1H phased-array system, incorporating multi-channel RF mixers. This system was implemented in human cardiac studies with both 23Na and chemically-selective 31P phased-arrays (3). I show here some preliminary 3D 23Na MRI data (collaboration with C. Constantinides and M. Pomper) obtained indicating elevated sodium in a brain tumor, consistent with early work showing elevated sodium levels in cancerous tissue. The future of broadband MRS and standard detector coil and decoupling technologies on clinical research scanners is a significant concern for our work.
1. Bottomley PA, Atalar E, Weiss RG. Human cardiac high-energy phosphate metabolite concentrations by 1D-resolved NMR spectroscopy. Magn Reson Med 1996; 35: 664-670.
A. Bottomley PA, Weiss RG. Noninvasive MRS detection of localized creatine depletion in non-viable, infarcted myocardium. The Lancet 1998; 351: 714-718.
B. Lee RF, Giaquinto R, Constantinides, Souza, S, Weiss RG, Bottomley PA. A broadband phased-array system for direct phosphorus and sodium metabolic MRI on a clinical scanner. Magn Reson Med 1999 (submitted)
Zaver M. Bhujwalla
Johns Hopkins University
Magnetic Resonance Spectroscopy in Exerimental Models of Breast and Prostate Cancer
The role of MRS in Clinical Oncology was discussed in the context of data obtained by us (Oncology Section - MR Research, JHU Radiology) in experimental studies using human breast and prostate cancer models. These data further support the role of MRS in (i) Detection / Diagnosis / Prognosis and (ii) Therapy.
Patrick M. Colletti
University of Southern California
Spectroscopy with 5FU and Pharmacokinetic Imaging in Cancer
With regard to tumor, there are usually three general questions: (1)What is it? (Single Voxel MRS), (2) Where is it? (2D/3D MRS), (3) What is happening to it? (Sequential/Dynamic MRS).
For cancer treatment to be successful, targeting of chemotherapeutic drugs must be individually optimized. A dose that may be effective in one patient may be ineffective in another and toxic in a third, due in large part to tumor heterogeneity, both spatial and temporal. Noninvasive studies (functional imaging and pharmacokinetic imaging) are critical in measuring, in a given patient at a given stage of disease, both the pathophysiology of tumors as well as whether the chemotherapeutic agent will be delivered to that tumor mass at the right rate and at the right dose.
Fluorinated drugs, including but not limited to 5-FU, gemzar, FdUR, capecitabine, UFT, and many others, offer a unique opportunity for noninvasive studies using the high NMR sensitivity of 19F in the absence of naturally occurring fluorinated compounds. Integration of such noninvasive 19F measurements with pharmacokinetic methods allows, for the first time, direct measurements of tumoral pharmacokinetics of anticancer agents, and thereby, the prediction of response, dose optimization, quantitation of the effect of modulators, as well as mechanistic studies. Not only are such 19F measurements of great significance to the very large number of patients treated with fluorinated drugs, but they also serve as models for the expansion of such pharmacokinetic imaging studies to drugs where the nuclei to be measured would be 1H, 13C or others.
Jeffrey L. Evelhoch
Wayne State University
MultiCenter 31P Magnetic Resonance Spectroscopy Trial
A group of 6 institutions from the US (Duke University, Fox Chase Cancer Center, Memorial Sloan-Kettering Cancer Center, University of California at San Francisco, University of Pennsylvania, Wayne State University), 2 institutions from the UK (The Royal Marsden Hospital, St. George's Hospital Medical School) and 1 institution from The Netherlands (University Hospital Nijmegen) have been participating in a multi-center project to develop and implement 31P MRS in clinical oncology since 1995. The major goals of this project are to test the hypothesis that the in vivo31P MR spectrum can predict sensitivity or resistance to treatment in patients with non-Hodgkin's lymphoma, soft tissue sarcomas, or head and neck cancer by correlating the initial response to metabolic features in the baseline spectrum and to changes that occur in the spectrum early after the initiation of treatment.
During the first three years we received NCI support enabling us to implement uniform spectroscopic and imaging procedures, develop and distribute tumor-specific 1H/31P coils, establish protocols for transfer of imaging and spectroscopic data in a common format, establish procedures for quality control and quality assurance, and test spectral analysis and quantification. With these capabilities in place at all institutions and increased NCI support during the fourth and fifth years of this project, patient accrual has begun at an increased rate. Initial results for non-Hodgkin's lymphoma are very encouraging showing both a decrease in the ratio of phosphomonoesters to NTP (PME:NTP) in patients responding to treatment and, more exciting, a lower PME:NTP in naïve tumors prior to treatment.
University of Minnesota
Future Technical Prospects for Clinical MRS
The biochemical information and spatial resolution achievable with localized MRS are limited by the inherent low sensitivity of MRS and the low concentrations of metabolites in vivo. Significant gains in sensitivity and spectral resolution are possible with higher magnetic fields (³ 3 tesla), and such improvements will have a major impact on the usefulness of in vivo MRS in clinical cancer research. Recent results obtained from human studies at 4 tesla and animal studies at 9.4 tesla provide solid evidence of the benefits of high fields. Increases in sensitivity, chemical-shift dispersion, and spectral simplifications of coupled spin systems allow previously unresolved resonances to be readily observed and accurately quantified. For most applications, high field MRS will not be limited by FDA guidelines regarding the allowable specific absorption rates (SAR), since spectroscopy scans are usually performed using a low duty cycle (i.e., long repetition time, TR).
Further methodological developments are also needed to optimize sensitivity and to fully extract the wealth of physiologic information that high field MRS can offer. (1) RF transmit/receive coil developments. (2) Spectral editing methods to separate lactate from overlapping mobile lipid peaks in tumors. (3) Indirect (1H) detection methods which will maximize sensitivity in dynamic studies of the metabolism of isotopically-enriched substrates (e.g., 13C-labeled glucose). (4) Fast automated shimming of first and second order shims is essential to maximize spectral resolution and information content. (5) Broadband selective RF pulses that minimize chemical-shift-induced displacement of the volume of interest (VOI) in single-voxel spectroscopy.
Robert J. Gillies
University of Arizona
MRS of Cancer: Metabolic Basis
The major points of my presentation were that (1) MRS reveals information not obtainable by other techniques, (2) pre-clinical studies are important and (3) exogenous indicators of physiology are useful.
MRS reveals information not obtainable by other techniques because only compounds with short correlation times are visible. This is exemplified by the fact that alanine or lactate are often not completely visible in vivo, yet extracts of the same tissues reveal much higher levels. This suggests that these metabolites may be compartmentalized in a non-visible pool (e.g. bound to macromolecules). Another example is the observation of mobile lipids in cancer cells with high metabolic potential. These lipids correspond to small intracellular lipid droplets, which would not be distinguished by other techniques. MRS can be used to investigate the biophysical behavior of these droplets. Additionally, MRS is uniquely appropriate for in vivo longitudinal studies of metabolism since it is non-destructive. In many experiments the steady-state levels of metabolites can be monitored in at the same time as pathway flux, providing an unprecedented window on metabolic control.
Model systems include animals, cultured cells in bioreactors, and extracts of cells and tissues. These can be used to develop hypotheses for later application in the clinic, or can be used to follow up clinical observations with high resolution. For example, phosphomonoesters have long been shown to be altered in tumors relative to normal tissues, and to change in response to successful therapy. These observations have been followed up in animal models and in vitro to characterize the control of phosphomonoester metabolism.
A final, important application of MRS is the use of exogenous indicators of physiology and metaoblism. These have included phosphonoium choline, 2-deoxyglucose, 3-Amino propylphosphonate, Dimethyl methylphosphonate, Fluoro BAPTA, and Imidazoles (e.g. IEPA). These allow the observation of metabolic and physiologic parameters that are not accessible via non-invasive spectroscopy and increase the applicability of the technique in vitro and in vivo.
Jerry D. Glickson
University of Pennsylvania
Molecular Targets for MRS & "Smart MRI" in Oncology
Phosphorus-31 was the first nucleus studied by in vivo NMR spectroscopy of tumors. Detectable molecules include energy metabolites (nucleoside di- and triphosphates (NTP, NDP), phosphocreatine (PCr), inorganic phosphate (Pi)), pH indicators (Pi) and phospholipid metabolites (phospomonoesters (PME) and phosphodiesters (PDE)). This phase of MRS cancer research has now progressed to clinical evaluation. Proton spectroscopy has been used to detect lactate, choline metabolites (Cho), creatine and PCr, citrate (in the prostate), N-acetylaspartate (NAA; in the brain) and various amino acids (e.g., alanine, glutamate and taurine). This research has reached a mature state with brain tumors, but is still in the developmental stage for detection of tumors outside the brain. Other nuclei need to be explored at natural abundance.
The third most sensitive naturally occurring NMR is 23Na. Of particular interest to cancer are reports originating from Ephraim Racker that tumors contain an aberrant Na+,K+ ATPase, leading to an increase in intracellular levels of 23Na. This could provide a basis for tumor diagnosis if appropriate methods can be developed for distinguishing between intracellular and extracellular sodium. In addition, 23Na NMR could provide a sensitive method for detecting edema and for monitoring response to therapy. Dr. Navin Bansal has been developing methods to distinguish between intracellular and extracellular sodium in tumors, using a sodium shift reagent TmDOTP and multiple quantum coherence transfer. The shift reagent displaces the resonance of extracellular sodium from the intracellular sodium peak in the single quantum spectrum; however, this agent cannot be used in the clinic because it modifies blood pressure. For clinical studies multiple quantum measurements permit approximate distinction between intracellular and extracellular compartments, the intracellular component remaining approximately constant in intensity, while the intracellular component changes in intensity with various interventions. Robert Lenkinski has obtained 23Na images of the human brain at 4T, which are comparable in quality to 1H brain images obtained in the early stages of development of 1H MRI.
Physiological indicators. Gillies' laboratory has pioneered the development of probes for monitoring extracellular pH. Their first agent was 3-aminopropylphosphonate (3APP); Pi served as an indicator of the approximate intracellular pH (since this metabolite occurs in both compartments at comparable concentrations, but the intracellular compartment usually comprises ~70% of the tumor volume). The key limitation of 3APP was the limited spatial resolution of 31P MRS. Much higher resolution has recently been obtained by Gillies' laboratory using imidazole derivatives, which are detectable by 1H MRS. Sha He and Ralph Mason have recently developed 6-fluoropyridoxamine (6FPAM), a fluorinated derivative of vitamin B6 detectable by 19F MRS. This agent simultaneously measures both intracellular and extracellular pH with sensitivity comparable to the imidazole analogs of Gillies. In highly necrotic tumors the extracellular volume may be comparable to or greater than the intracellular volume; consequently, Pi is no longer a good indicator of pH under these conditions. 6FPAM does not suffer from this limitation.
There is a need to develop indicators for monitoring various other physiological properties of tumors. These include the calcium concentration, since F-BAPTA has proven to be of limited utility, and the pO2, which is a critical determinant of tumor radiosensitivity. Meade's laboratory has developed an ingenious scheme for sensitive detection of calcium. They have synthesized caged chelates of Gd in which all seven ligation sites of the lanthanide are coordinated to the chelate. In the presence of calcium one arm of the chelate swings over to bind to the calcium ion, thereby exposing the Gd to water and producing efficient water relaxation that is readily detected by MRI. The method is capable of detecting free calcium concentrations in the micromolar range, i.e. the physiologically relevant range. Pefluorocarbons have been used as oxygen indicators, but have limited utility because they are not sensitive in the radiobiologically relevant range of hypoxia (
The Warburg Hypothesis, i.e. that tumors exhibit anomalously high levels of aerobic glycolysis, has been the foundation of tumor detection by PET using 2'-fluorodeoxyglucose (FDG). While this hypothesis has been questioned as a universal indicator of malignancy, it still has proven useful in detection of many tumors and metastases. Shulman's laboratory has demonstrated the power of 13C MRS measurements not only of glycolysis but of virtually all aspects of glucose metabolism, e.g. the pentose shunt, the glycogen pathway and the TCA cycle. While 13C-labeled glucose is expensive, the overall cost of such measurements is less than that of PET examinations, and the method can be implemented virtually on all clinical MRI instruments operating at 1.5 T or higher. This methodology needs to be implemented in the clinic and in animal models.
Gene Therapy Markers. Gene therapy is currently being explored as a novel and promising method for treatment of cancer. There is a need to develop markers of transfection consisting of proteins which produce products detectable by MRS or MRI that are expressed together with the target gene. One such agent, arginine kinase an enzyme that normally does not occur in mammals, has recently been developed by Walter and Sweeney for gene therapy of muscular dystrophy. This enzyme produces phosphoarginine whose resonance can be distinguished from PCr in the 31P spectrum. Phosphoarginine may serve as a marker gene for gene therapy of other diseases, including neoplasms such as rhabydomyosarcomas. Of particular current interest in oncology is the use of suicide gene therapy - transfection of tumors with genes that selectively sensitize them to cytotoxic agents. Drs. Ross and Bhujwalla will present data at this meeting on the use of cytosine deaminase to convert the prodrug fluorocytosine to the cytotoxic agent 5-fluorouracil (5FU). In this instance the cytoxic agent is acting as both a marker of gene expression and a drug for treatment of the neoplasm. Similar schemes are being implemented with herpes simplex virus thymidine kinase transduction of tumors to sensitize them to gancyclovir therapy, which can be monitored by PET with 18F-labeled gancyclovir.
Receptor Targeting. A number of tumors over express low density lipoprotein (LDL) receptors, presumably because of requirements for cholesterol and fatty acids in membrane synthesis. Since the classic studies of Goldstein and Brown elucidating how these receptors function, it has been recognized that the LDL receptor system could be used for the selective delivery of antineoplastic agents to tumors over expressing these receptors. This is accomplished by replacing the lipid core of LDL with lipids containing lipophilic antineoplastic agents. The same strategy could be used for the detection of these neoplasms by delivery of MRS detectable or MRI detectable molecules. Of particular interest is the possibility of replacing LDL by lipid vesicles to which the much smaller lipoprotein, apoE (which is available in a recombinant form), is attached. Incorporation of MRI/S detectable agents is much more easily accomplished in vesicles than in the lipid core of LDL. The LDL molecules or the apoE labeled vesicles bind to receptors on the cell surface, are internalized and incorporated into lysosomes, in which the LDL or apoE are hydrolyzed to amino acids, and the cholesterol esters and phospholipids are hydrolyzed to their constituent components. Any agents that are incorporated in the LDL or apoE/vesical are released. The receptors are recycled to the cell surface, where they bind to and transport more LDL and apoE/vesicle complexes into the cell. Our laboratory is developing this system for delivery of various agents detectable by MRI/S or optical imaging to tumor cells. This approach can be extended to other receptors. For example, Weissleder's laboratory is developing a method for delivery of MRI contrast agents to transferrin receptors, which are also over expressed by some tumors. Some agents can be delivered by a nonspecific delivery system. For example, molecular beacons, which utilize antisense technology to selectively identify cells with specific mRNA molecules may be delivered by virosomes--influenzae virus envelopes encapsulating the constructs. Gewirtz's laboratory has developed these agents. The development of specific and nonspecific methods for delivery of drugs, transfection agents and MRI/S contrast agents will go hand in hand with the development of these novel therapeutic agents.
Specific Enzyme Probes.Meade's laboratory has developed an extremely clever and potentially very useful method for detecting specific enzymes by MRI. Again they have used the caged Gd strategy, but in this instance the lanthanide is encapsulated not only by the chelator but also by a cap consisting of a substrate for a specific enzyme. For example, they have used a galactose cap to detect _-galactosidase (b-Gal). b-Gal hydroyzes the bonds holding the galactose cap in place, thereby exposing the Gd ion for coordination by water, whose relaxivity is thereby increased in proportion to the concentration of the enzyme. They have demonstrated the methods in Xenopus embryos, in which they have coexpressed b-Gal with the green fluorescent protein. Regions that expressed the b-Gal were detectable both by MRI and by fluorescence microscopy. They have also developed these probes for other enzymes, such as glucuronidase; proteinases could also be targeted. Delivery of such constructs could be accomplished by the vehicles we have described in the previous section. This research demonstrates the feasibility of developing novel contrast agents targeted at specific enzymes that occur in cancer cells. Development of such "smart contrast agents" needs to be encouraged.
Cell Tagging. Weissleder and also Koretzky and Ho have recently demonstrated that cells can be tagged with ferromagnetic iron particles, which serve as MRI tracers for monitoring the transport of these cells in the body. Such methods could be used to track the course of tumor metastasis or other critical pathways.
Molecular Switches. I wish to illustrate an important principle that is currently being developed for optical imaging, but which may eventually be implemented in MRI/S. Molecular beacons are molecules consisting of a loop of DNA or antisense DNA that is complementary to a specific target mRNA. At the ends of the loop are 'stems" consisting of short sequences of complementary DNA, which hold the ends of the loop together. Attached to the stems are linkers, one of which is attached to a fluorophore, the other to a quencher. The hydrogen bonds in the stem keep the fluorophore and quencher in close enough contact to quench all the fluorescence. Consequently, in the absence of the target mRNA no fluorescence is detected. When the target mRNA is present, the loop hybridizes with it, and because there are many more base pairs in the loop DNA-target mRNA hybrid than in the stem, the hydrogen bonds holding the stem together are broken, thereby separating the fluorophore from the quencher and producing detectable fluorescence. It is generally recognized that these molecular probes can be used to detect specific RNA sequences. NMR can, of course, never match the sensitivity of fluorescence spectroscopy; consequently, NMR could not be used for nucleic acid detection, but it might be used to detect more abundant enzymes, as Meade has recently demonstrated (see above). Now consider more carefully what the molecular beacon is doing. When it finds its target mRNA, the fluorophore emits a quantum of light energy that is detected by the observer. Suppose, however, that on the other side of fluorophore there was a molecule that could capture that quantum of light energy and convert it into chemical energy to produce superoxide or hydroxy radicals that were toxic to the cells, i.e. the fluorophore could activate an agent suitable for photodynamic therapy. We would then never see the fluorescence, but we might selectively kill the cells containing the target sequence of mRNA. Agents such as this might be targeted to mRNA for mutant tumor respressor or oncogenes, and might be used to selectively destroy not only tumors but preneoplastic cells. Now, this is only a dream, and it will require a great deal of resaearch to develop, but it provides a very attractive strategy for development of very specific antineoplastic agents. The analog for fluorescence quenching in MRS might be anti-ferromagnetic coupling, such as exists in plant ferrodoxins, and instead of targeting mRNA, one might target specific proteins in critical cell signalling pathways. The key principle we are proposing is to design agents that detect specific molecules that are modified in tumor progression, i..e. the "signatures of cancer," and generate signals that either lead to cell death or turn off cell replication. I would like to suggest to the NCI that it should encourage this type of research.
Fluorine Labeled Drugs: 5FU. The paradigm for MRS pharmacokinetics of antineoplastic agents is 5FU. Griffiths' laboratory first demonstrated in rats that 19F MRS could detect catabolism of this agent in the liver and its anabolism in tumors. They have further shown in animal models that response to this agent varies directly with the production of 5FU nucleotides in the tumor; however, these key molecular predictors of therapeutic response have only rarely been detected in the clinic (by Leach and colleagues, personal communication). In most clinical studies only the parent drug has been detected in the tumor. Wolf and coworkers have been conducting a pilot clinical study of patients with colorectal cancer having metastases in the liver. By measuring the half lives of 5FU in the liver (presumably in the tumor) they have divided the patients into two categories - trappers (half life > 20 minutes) and non-trappers (half life
Therefore, to quote a colleague of mine at this meeting (T.R. Brown), "the time is ripe" for a clinical trial of MRS monitoring 5FU therapy of colorectal cancer metastases to the liver. In addition, a great deal can be done to improve the already impressive accomplishments of this method. No localization was employed in monitoring these colorectal tumor metastases. Localization methods need to be perfected for such studies. Sensitivity also has been enhanced by using appropriate proton decoupling schemes and/or using higher magnetic fields. These improvements would facilitate detection of 5FU nucleotides that might prove much more reliable in predicting tumor responders.
Carbon-13 Labeled Drugs: Temozolomide. It is essential to implement the principles demonstrated with 5FU with other, more potent agents. Recognizing that labeling of non-fluorinated drugs with fluorine often modified their pharmacological activity, Artemov et al. demonstrated that it was possible to label and detect antineoplastic agents with 13C with little or no modification of pharmacological activity. We chose to demonstrate this principle on temozolomide, a new alkylating agent active against brain tumors and melanomas. Studies were performed on RIF-1 tumors subcutaneously implanted in mice. The drug was detected by 13C MRS with polarization-transfer from protons. Even higher sensitivity might be achieved by 1H detection of heteronuclear multiple quantum coherence, but at the time this study was performed, the indirect detection method produced excessive heating of the tumor. The mouse studies employed a substantially higher concentration of the drug than is used in the clinic. Studies in dogs have been proposed to determine if the drug can be detected at clinically relevant levels by using larger voxels.
Proton MRS Detection of Antineoplastic Agents.Cisplatin and its analogs are widely employed in cancer chemotherapy. The parent drug is used at too low a concentration for detection by MRS, but the organoplatin analogs are used in more appropriate concentrations. Iproplatin, one of these analogs, contains 12 equivalent methyl protons, which unfortunately coresonate with the methyl protons of lactic acid. However, He et al. demonstrated that it is possible to selectively detect this agent by using multiple quantum coherence transfer from the methine hydrogen on the isopropyl groups. While this agent is not widely used in the clinic, the same principle could be used to detect other cisplatin analogs such as carboplatin. This approach is more practical than labeling with 13C (see above) because it avoids the very difficult problem of custom synthesis of these agents under Good Manufacturing Practice conditions for clinical use. Therefore, we recommend that the NCI encourage the development of MRS methods for monitoring the pharmacokinetics of effective antineoplastic agents (such as cisplatin analogs) which can be detected in their native state without isotopic or fluorine labeling.
MRI Detection of Paramagnetic Antineoplastic Agents. It has long been recognized that porphyrins tend to selectively localize in tumors. In fact, this forms the basis of photodynamic therapy. The Texophryns are porphyrin-like agents that contain Gd coordinated to the "porphyrin" ring. They also have been proposed as photosensitizers in photodynamic therapy. At the meeting showed an image (provided by Dr. David Rosenthal, U. Penn) of a patient with brain metastases of a non-small cell lung carcinoma. No tumor was visible in the brain, but after administration of texophryn the tumor was clearly delineated; no other enhancement was detected in the brain. The photosensitizer was simultaneously serving as a tumor selective MRI contrast agent. Development of such tumor selective MRI contrast agents/therapeutic agents should be encouraged by the NCI
Jason A. Koutcher
Memorial Sloan Kettering Cancer Center
Magnetic Resonance Spectroscopy in Oncology at Memorial Sloan Kettering
Dr. Koutcher presented some of the ongoing magnetic resonance spectroscopy at Memorial Sloan Kettering Cancer Center (MSKCC). As part of the Multi-Institutional 31P NMR Tumor Consortium, the NMR group at MSKCC has been studying sarcomas, breast and head and neck tumors. Dr. Koutcher showed several spectra showing differences in responding and non-responding patients with sarcomas. These included early studies without decoupling, and more recent studies with decoupling. High quality spectra with resolution of PE and PC were shown. Non-responding patients often showed little change in the spectra over time. Spectra from normal breast tissue and advanced breast carcinomas were also presented.
Preoperative studies of osteogenic sarcoma of the lower extremity using Gd-DTPA uptake were also presented. After measuring the initial slope of the increase in signal intensity vs. time, the fraction of voxels with a slope less than a predetermined threshold was used to estimate the tumor necrotic fraction. Using this technique, data on 8 patients showed a good correlation between the "Huvos grade" and predicted necrotic fraction, with 7 out of 8 patients classified correctly.
1H NMR prostate and brain studies are also ongoing at MSKCC. The brain studies focus on primary lymphoma of the brain. The prostate studies are done using software obtained from the University of California at San Francisco (Dr. Kurhanewicz, Nelson and Vigneron) and are focusing on assessing response and residual tumor in "poor risk" patients undergoing pre-operative neoadjuvant chemotherapy (estramustine, carbo-platinum and paclitaxel) and hormones. Serial spectra were shown which displaying the effects of these drugs on 1H NMR spectra and pathologic correlation is ongoing. In conjunction with this research effort, studies of 1H NMR spectra of human xenografts to determine whether the spectra can monitor the effects of different therapies and predict response were also presented.
Since the interaction between animal/cell and clinical studies is considered essential for the further development of MRS in oncology, 2 potential applications were presented in animal tumor models. With the development of cytostatic anti-neoplastic agents for clinical trials ongoing, a method of assessing response is essential. 31P NMR spectra from murine mammary carcinomas were presented which were sued to assess the effect of Combretastatin A-4 phosphate on tumor metabolism, as a surrogate of blood flow. Within 30-60 minutes there was almost complete loss of nucleoside triphosphates and phosphocreatine. In contrast to previous studies evaluating the effect of this drug on tumor perfusion, significant recovery was present in 4 hours, suggesting that multiple doses or continuous infusion of the drug may be necessary. NMR spectroscopy is an ideal tool for studying the effects of metabolic inhibitors. The effect of a 3 drug combination (N-(phosphonacetyl)-L-aspartate (PALA), 6-methylmercaptopurine riboside (MMPR), and 6-aminonicotinamide (6AN)) on tumor metabolism was presented. The NMR data was used to determine the timing between the drugs and radiation based on when tumor metabolism was maximally inhibited. While the drugs alone induced no complete responses (CRs), and radiation only induced a single (1/20) CR which only lasted a few weeks, the combination of the drugs and radiation (administered when the NMR data demonstrated maximal metabolic inhibition), yielded a 65% CR rate and a 25 % durable (
A.2. Roster of Focus Group Participants.
Invitees and Meeting Organizers
A.3 National Cancer Institute MRS Focus Group Agenda
1H and 31P Magnetic Resonance Spectroscopy in Clinical Oncology
Day 01: Thursday, April 22, 1999
2:00 Opening Remarks
Dan Sullivan and Joe Ackerman
2:15 Biological Basis for MRS in Oncology
Robert Gillies (15 min)
Jerry Glickson (15 min)
- Background; accessible molecules; what might distinguish tumors fromnormal tissue and/or define classes of tumors and/or appropriate therapy; use of labeled drugs or metabolic markers; probing specific aspects of metabolism at pharmacologic doses.
3:00 Technical Overview of Clinical MRS
Sarah Nelson (15 min)
- 1H MRS in the clinic.
Cecil Charles (15 min)
- 31P MRS in the clinic.
3:45 Future Technical Prospects for Clinical MRS
Mike Garwood (15 min)
- High field, novel coils, new pulse designs.
Truman Brown (15 min)
- Data analysis.
5:00 Biology and Technology: Brief Contributed Presentations and Discussion
Attendees (5 - 10 min, overheads preferred)
- Contributions related to afternoon topics.
7:00 Working Dinner (Entire Focus Group)
Clinical MRS Results
Jeff Evelhoch (15 min)
- Cooperative NCI/U01 Group.
John Kurhanewicz (15 min)
- Prostate Cancer.
Robert Lenkinski (15 min)
- Breast Cancer.
Dan Spielman (15 min)
- Head and Neck Cancer.
Jeffrey Alger (15 min)
- Brain Cancer.
June Taylor (15)
- Pediatric Cancer.
9:00 Day 01 Concluding Remarks
Joseph Ackerman and Dan Sullivan
Day 02: Friday, April 23, 1999
8:30 Clinical MRS: Brief Contributed Presentations and Discussion
Attendees (5 - 10 min, overheads preferred)
- Contributions related to working dinner topics.
10:00 Group Discussion
- Is there justification for further substantive funding of clinical MRS in cancer for (i) detection, (ii) diagnosis, (iii) monitoring of therapy, and/or (iv) monitoring of recurrence? Nuclides of interest (1H/31P)?
- What are the acceptable minimum technical requirements?
- What barriers must still be overcome?
- What is the relationship between MRS and MRI in the cancer clinic?
(Is MRS of value as a stand-alone technique?)
12:00 Working Lunch (Small Breakout Groups)
- Formulation of Summary and Recommendations.
2:00 Conclusion of MRS Focus Group Mtg.: Summary and Recommendations
Joseph Ackerman, Dan Sullivan, and attendees
- Presentation and discussion regarding workshop conclusions.