The Conception of FDG-PET Imaging







Abass Alavi, MD, Martin Reivich, MD







The concept of emission and transmission tomography was introduced by David


Kuhl and Roy Edwards in the late 1950's, which later led to the design and


construction of several tomographic instruments at the University of


Pennsylvania. These machines were able to successfully map regional


distribution of radionuclides such as 99mTechnetium as tomographic images.


The instruments built at Penn were designed to detect single gamma emitters


and therefore their research and clinical applications were limited to the


investigation of simple functions like breakdowns in blood-brain barrier in


disorders such as brain tumors and cerebral infarcts. The instruments


manufactured in the late 60's and the early 70's were also designed to


image only the brain and not the other organs, which was dictated by the


technical difficulties that were encountered at the time. Collaboration


between investigators from Nuclear Medicine and the Cerebrovascular Center


at the University of Pennsylvania (directed by Martin Reivich) resulted in


great interest in quantitative measurement of regional cerebral function


such as blood flow and blood volume. Although these attempts were


successfully implemented, it became clear that synthesizing biologically


important compounds with single gamma emitting radionuclides, like


technetium and iodine, was a major challenge at the time and therefore


other avenues were to be explored to overcome these limitations.







By the early 1970s, Louis Sokoloff et al from the National Institutes of


Health (NIH) and Martin Reivich from the University of Pennsylvania had


clearly shown that the beta-emitting 14C-deoxyglucose (DG) could be


successfully utilized to map regional brain metabolism, which was later


proven to correlate well with local function. These investigators were


able to show that DG crosses the blood-brain barrier and is phosphorylated


by the hexokinse system to DG-6-phosphate similarly to glucose. However,


in contrast to glucose-6-phosphate, which further metabolizes to C02 and


H20, DG-6-phosphate remains intact in the tissue for an extended period of


time.  This unique metabolic behavior makes radiolabeled deoxyglucose an


excellent candidate for mapping regional function in the brain and other


organs. Since 14C is a beta emitting radionuclide, optimal assessment of


its distribution could be revealed by a technique called autoradiography:


in animal experiments, 40-45 minutes following the intravenous


administration of 14C-DG, slices of the brain were exposed to radiographic


films to reveal the beta particles emitted for a period of time. The film


was then processed to capture the regional distribution of the compound


with exquisite detail. Following successful demonstration of 14C-DG as a


metabolic tracer, collaboration between investigators from the NIH and Penn


resulted in defining and measuring parameters that are essential for


calculating regional metabolic rates for glucose in various structures in


the brain.







By the early 70's, this powerful research technique had been adopted


worldwide as an important research tool for the assessment of regional


brain function in a variety of physiological and pathological states in


different animal models.







Increasingly, it became clear that employing the DG method as a


non-invasive methodology for the investigation of brain function in healthy


and diseased states in man would substantially advance our knowledge of


neuropsychiatric disorders.







In late 1973, the year x-ray computed tomography was introduced by


Hounsfield, which proved to be an extraordinary structural imaging


technique, Martin Reivich, Director of the Cerebrovascular Research Center,


David Kuhl, Director of Nuclear Medicine at the time, and Abass Alavi, a


junior staff in nuclear medicine (all at the Hospital of the University of


Pennsylvania) discussed the possibility of labeling DG with a gamma


emitting radionuclide for in-vivo imaging by an appropriate instrument. It


was clear to these investigators that only positron emitting radionuclides


would be suitable for this purpose. They consulted Alfred Wolf, an organic


chemist at Brookhaven National Laboratory (BNL) who had developed a great


interest in synthesizing positron-emitting compounds, for selecting an


appropriate label for DG. At a joint meeting of investigators from BNL and


Penn in December 1973, Al Wolf suggested that 18-Fluorine rather than 11


Carbon should be pursued as an appropriate option, because of its


relatively long half life and its low positron energy. The long life of


18F was also attractive for shipment of the compound from BNL to Penn where


the group had planned to conduct the first tomographic studies in man. At


the conclusion of the meeting, Al Wolf expressed his and his group's great


desire to work on this project and made it clear that he wanted to be a


close research collaborator with Penn investigators after the synthesis had


been accomplished.







In the ensuing 2 years NIH funding was secured, which resulted in


establishing a PET center at Penn to initiate this project and, in the


meantime, Tatsuo Ido had joined Wolf's lab as a visiting post-doctoral


fellow from Japan and was assigned to this project. Dr. Ido became the


author of the first paper describing the synthesis of this compound. By


1975, FDG was successfully synthesized at BNL and although the initial


yield was low, it was sufficient to plan for human studies. The BNL group


also was able to synthesize Carbon-14 FDG, which was shown by Sokoloff and


colleagues to have a similar behavior to that of Carbon-14 DG in in-vivo


experiments carried out at NIH. In addition, all the required steps were


taken to make certain the product could be safely prepared for human


studies. Soon thereafter, an Investigational New Drug application was


filed with the FDA in preparation for the administration of FDG to humans.







Investigators at Penn, in anticipation of human experiments with a positron


emitting radionuclide, had assembled a set of high-energy collimators to


equip the Mark IV scanner (designed and built at the University of


Pennsylvania), which until then was only capable of imaging low energy


radionuclides such as 99mTechnetium and 123I, to be able to image the


511Kev gamma rays emitted as a result of positron decay. By mid summer 1976


researchers from both institutions decided that the time had come to plan


the first human studies at Penn. In August 1976, two normal volunteers


each received a dose of FDG which was shown to concentrate in the brain by


utilizing only one of the two gamma rays emitted from the annihilation of


positron particles (instead of detecting the two gamma rays as a


co-incident event). (Fig1) This was a gratifying outcome for


investigators from both labs who had worked so tirelessly over the


proceeding years to achieve this goal, namely to perform the first images


of cerebral glucose metabolism in man. The quality of the images generated


was poor and is not comparable to that of scans acquired and reconstructed


with modern instruments (Fig2). A whole body image of the FDG distribution


was also obtained in one subject by using a dual head Ohio-Nuclear Scanner


(Ohio Nuclear, Cleveland, Ohio) which was also equipped with high-energy


collimators for Sr85 bone studies. (Fig 3) Uptake of FDG in the heart and


significant renal excretion of the compound was demonstrated on this first


human whole body study. Obviously, the quality of whole body images with


FDG is substantially enhanced by employing instruments that are optimally


designed for this purpose (Fig4).







Simultaneous with the development at the University of Pennsylvania and


BNL, investigators at Washington University, directed by Michel


TerPogossian and in collaboration with Michael Phelps and Edward Hoffman,


had developed the first successful Positron Emission Tomography (PET)


machine for optimal imaging of positron emitting radionuclides in man.


Soon thereafter, Gerd Muehelenner at Searle Radiographics (later purchased


by Siemens) successfully demonstrated the feasibility of employing two


opposing scintillation cameras as co-incident detectors to image position


emitting radionuclides. This approach was later perfected when he was a


faculty member at Penn. In mid 1976, David Kuhl was recruited by UCLA as


the Director of nuclear medicine at that institution and by the fall of


that year he had assembled on outstanding team of investigators to further


explore the potential application of PET in CNS and other organ disorders.


At that time, UCLA was one of a few centers with a functioning cyclotron in


a medical environment. Shortly, a PET scanner (designed and built based


upon principles established by the Washington University's PETT III


scanner) was installed at UCLA which for the first time allowed


investigators at that institution to image FDG uptake (the synthesis scheme


was established with assistance from the BNL group) in the brain with an


optimal instrument. This group headed by David Kuhl was able to


demonstrate the ability of FDG-PET imaging in mapping cerebral function in


a variety of physiological and pathological states. In the meantime, The


PETT III scanner, which was designed and manufactured at Washington


University, was transferred to BNL for conducting human studies at that


institution. The next phase of collaboration between BNL and University of


Pennsylvania investigation was initiated when Al Wolf requested the


research team from the latter institution to conduct FDG based CNS projects


at BNL. Every other week a research team, directed by Abass Alavi,


traveled to BNL by car and by plane to perform several interesting and


important research projects in normal volunteers and later in patients. By


1979, the group at Penn, directed by Martin Reivich and Abass Alavi, had


established a PET center independent of BNL but collaboration between the


two institutions continued for more than a decade.







The extraordinary power of functional imaging as evidenced by the FDG-PET


technique generated a great deal of interest in the scientific community


which later led the NIH to establish several centers which included the


University of Michigan, Johns Hopkins, Washington University and the NIH


Campus at Bethesda (in addition to Penn and UCLA) which expanded the domain


of research beyond what had been achieved with FDG.







Based on an observation made by Warburg in the 1930's that malignant cells


utilize glucose preferentially over other substrates, Som and colleagues at


BNL were able to demonstrate substantial concentration of FDG in tumor


models in animals based on these principles and for the first time, FDG was


used by Dr. Dichiro and colleagues at the NIH to investigate metabolic


activities of brain tumors in man at diagnosis and following treatment.


They were able to demonstrate that the degree of FDG uptake correlated with


the grade of the tumor and also it was a predictor of outcome at diagnosis.


More importantly, it was noted that FDG-PET imaging was superior to


contrast-enhanced CT and MRI in differentiating recurrent tumors from


radiation necrosis. Investigators from Penn (Jane and Abass Alavi) further


confirmed these early observations and since the mid-80's FDG-PET imaging


has been widely used to examine patients with brain tumors specifically for


the diagnosis of recurrent brain malignancies.







During most of the 80's, performance of whole body imaging with PET was


validated, and, by the early 90's, its application as an effective modality


was realized for this purpose. Investigators from UCLA and later from the


Universities of Duke, Michigan, Nebraska, and Heidelberg were among the


pioneers in demonstrating the efficacy of FDG-PET imaging in the management


of patients with a variety of malignancies. These included diagnosis,


staging, monitoring treatment and detecting recurrence of a variety of


tumors. The role of FDG-PET in differentiating benign from malignant


nodules as a standard and as the study of choice is unchallenged at this


time. This imaging technique has substantially simplified the management


of patients with solitary pulmonary nodules and staging patients with lung


cancer with high accuracy. Detection of recurrent tumors by FDG-PET


imaging following surgery for colon cancer as evidenced by elevated serum


Carcino, Embryonic Antigen (CEA) levels has been revolutionary, since in


the majority of these patients, CT and other anatomic imaging techniques


fail to demonstrate the sites of disease. FDG-PET imaging is not only very


cost effective in this setting, but is of great importance in providing an


answer for a difficult clinical problem and has been well accepted by the


clinical oncologists.







FDG-PET imaging may completely replace other imaging techniques for the


initial staging, restaging, and monitoring effects of treatment in patients


with Hodgkin's and non-Hodgkin's lymphomas. The extraordinary sensitivity


and specificity of FDG-PET imaging allows detection of disease activity in


the lymph nodes and other organs with great precision. It is conceivable


that in the near future, this modality will be used as the study of choice


in the management of patients with lymphomas.







Similar statements can be made about the application of FDG-PET to other


malignancies including head and neck tumors, breast cancer, melanoma,


ovarian cancer, mesothelioma and possibly thyroid cancer and genito-urinary


(GU) tumors.







Although FDG-PET imaging can play a role in the diagnosis of cancer, its


major contribution has been in the accurate staging of cancer, in the


assessment of the effectiveness of therapy and above all, in detecting


recurrence following medical, radiation, and surgical therapies. In the


latter settings, changes due to such treatments render structural


techniques incapable of providing a definitive answer about the disease


activity in many occasions.







FDG-PET imaging as an effective technique for the assessment of myocardial


viability is well established and in fact is considered as the gold


standard for this purpose. However, because of successes of conventional


imaging with single emitting radiopharmaceuticals, only in limited


circumstances is PET requested for determination of myocardial viability.







Increasingly, FDG-PET imaging is being used to detect suspected orthopedic


infections, as in failed prosthesis, complicated fractures and


osteomyelitis. Detection of inflammatory processes including sarcoidosis,


regional ileitis and arthritis may allow applications of FDG-PET imaging


for these challenging disorders and further enhance its clinical utility.







Also, early data from our laboratory and Sloan Kettering Memorial Hospital


demonstrate the feasibility of this technique in detecting atherosclerosis


which, following validation, may become an added domain for this exciting


modality.







What are the challenges ahead as we are witnessing the rapid expansion of


FDG-PET to the day-to-day practice of medicine around the country and the


world? The major difficulty that is being encountered by almost every


group is the tremendous shortage of personnel who are properly trained in


the discipline and are competent in performing various tasks that are


associated with optimal use of this technology. This applies to several


categories of professionals whose contributions have been essential for the


successful evolution of PET over the past 25 years. There is a great


shortage of technical staff to manage cyclotron facilities, produce


radionuclides, synthesize compounds and perform required quality control


for human studies. There is a great need for nuclear medicine


technologists who are optimally trained to perform these studies.







Above all, almost non-existent are experienced and competent physicians who


can provide this service in a competent manner in the community are very


few and in great demand. Training in diagnostic radiology and conventional


nuclear medicine is inadequate for interpretation of these complex and


artifact prone studies. It may not be an exaggeration to consider FDG-PET


images as some of the most difficult in the diagnostic discipline. It is


not uncommon to spend 20 to 30 minutes of time to examine a case so that


one can confidently render an accurate assessment of the findings portrayed


on the scan. Knowledge of cross sectional anatomy is helpful but not


essential for optimal interpretation of FDG-PET images. There is a


misconception that these studies must be interpreted by a radiologist,


rather than by a competent nuclear medicine physician who is not fully


trained in cross sectional anatomic imaging techniques. Currently, the


majority of cases around the world are adequately and competently


interpreted by non-radiologists and this trend will continue in the


foreseeable future. In fact, most PET specialists have mastered skills in


comparing FDG-PET images with other diagnostic techniques when such


comparison have been necessary.







It is becoming quite clear that acquiring optimal skills to interpret


FDG-PET images will require at least 6 months of full-time training in this


discipline. Fellowships for a period of 1-2 weeks as a certificate of


competence is unjustified and in fact will be a disservice to the medical


community if adopted by future practitioners. It would be unfortunate if


such practices tarnish the image of the discipline as an effective modality.







Finally, it is also quite evident that this type of service should be


provided by centers with active cancer programs. Such centers should be


able to refer at least 2-3 patients to the PET facility on a daily and


routine basis for financial viability of the operation of this complex


technology. We are of the belief that fixed sites are preferable to mobile


arrangements for this purpose. Fixed sites provide the continuity of


operations on a routine basis which translate into technically optimal


operation. The future of mobile units for this modality may be


questionable at this time. Surprisingly, FDG supply has adequately kept up


with rapid expansion of this technique to the medical community and does


not appear to be a source of difficulty at this time. It is not clear


whether future regulatory issues will limit its distribution within and


across states.







In conclusion, increasingly FDG-PET imaging is increasingly playing a major


role in management of patients with a variety of disorders, especially


those with cancer. This modality provides an exciting opportunity to the


imaging specialists, which is also associated with serious challenges. The


imaging community must make every effort to be certain that this modality


is appropriately utilized and managed so that in the end, patients with


difficult and challenging problems, will benefit from its capabilities. It


is important to note that FDG-PET imaging has made an everlasting impact on


our field and, in fact, its routine use and acceptance by the medical


community at large has rejuvenated the specialty into a powerful discipline


as we entered into the 21st century. It may not be an exaggeration to


project that, in the near future, the number of FDG scans performed in most


active nuclear medicine services will exceed that of all other procedures


peformed in most laboratories. It is therefore quite fitting to applaud Dr.


Henry Wagner for naming FDG as the "Molecule of the 20th Century" because


of its unparalleled and unique impact on the evolution of the field of


nuclear medicine.







 Legends







Fig. 1







First whole brain (planar) and tomographic FDG images of brain function of


a normal volunteer reveal high concentration of the agent in cortical and


subcortical gray matter. These images were acquired utilizing only one of


the two 511 KeV gamma rays emitted from the positron decay (instead of


coincidence detection of with appropriately designed modern instruments).


The whole brain scan was generated by an Ohio Nuclear Rectilinear scanner


while tomographic images were obtained with the Mark IV scanner, which was


designed to examine CNS disorders.







Fig. 2







Comparable FDG images of the brain obtained with Mark IV in Aug. (image on


the left) and with recently 1976 designed FDG-PET brain scanner, acquired


in June 2001, clearly demonstrate substantial technical improvements in


capabilities of instruments introduced over the past 25 years.







Fig. 3







Total body images were obtained following the acquisition of tomographic


studies utilizing a dual head Ohio Nuclear Rectilinear scanner. This


scanner was equipped with high energy collimators for performing Sr 85 bone


scans. These first whole body FDG images revealed uptake of the compound in


the heart (in addition to the brain) and significant excretion through the


kidneys.







Fig. 4







Whole body FDG-PET projection images of a patient with a lymphoma reveal a


great deal of detail about various structures included in the field of


view. The high quality of these images demonstrates a substantial


improvement in PET instrumentation, which has further enhanced effective


utilization of FDG in a multitude of malignant and non-malignant disorders.