Biomedical research, not unlike any other research, should be driven ideally by the intention to solve a problem. In basic science, the aim of good biomedical research is to understand the principles underlying physiology in health and disease, as a pre-requisite to identify targets, tools, and mechanisms suitable for a possible intervention. When it comes to clinical medicine, research—apart from purely epidemiological investigations—typically aims to satisfy so-called “unmet needs” for better diagnosis, prevention, or therapy of diseases.

Nuclear medicine, the opera of the sciences, nowadays combines the possibility to quantitatively investigate—once a suitable radiopharmaceutical tracer is available—virtually any physiological process, and to translate such diagnostic method into a therapeutic intervention by simple isotope exchange. The basis of the signal, pharmacological interactions of a tracer with its molecular target, irrespective of whether receptor, transport system, or enzyme, combined with radioactive decay, the most sensitive, reproducible, and quantifiable detection system in nature, allows in connection with current SPECT and PET camera systems to detect, measure, and analyze physiology of the host in health and disease on one hand, as well as pharmacokinetics and pharmacology of the radiopharmaceutical on the other.

Nuclear medicine, though, has been in its beginnings a purely therapeutic discipline, when Saul Hertz in 1941 first used artificially manufactured 130Iodine to treat hyperthyroidism [1, 2], followed in 1946 by Samuel Seidlin treating metastasized thyroid cancer with 131Iodine [3].

Only in 1951 it became possible to actually localize radioactivity inside the body using the “rectilinear scanner” invented by Benedict Cassen, a simple scintillation counter moving linearly over the body allowing to correlate count rates to two-dimensional anatomical coordinates at the body surface [4]. To the present day, therefore, camera systems in nuclear medicine continue to be referred to as “scanners”.

The unique feature of nuclear medicine, providing its right to exist as an independent diagnostic discipline besides radiology, is the possibility to investigate physiological processes, rather than only anatomical structures. What is now generally known as molecular imaging, exemplified by standard techniques such as functional uptake, perfusion, or excretion studies, required the introduction of time, as the fourth dimension into signal acquisition. In order to achieve this, the detection of signal changes over time, as well as a sufficiently large detector area, allowing to detect regional variances in radioactivity in a given field of view was necessary. In 1958, Hal Anger solved the problem, to first “focus” photons using a parallel collimator, and then detect their presence using scintillation crystals. An intelligent array of photomultipliers enabled an unprecedented localization of photons with sufficient timely and spatial resolution, creating an image of the underlying object in the detector plane [5]. This device, today known in honor of its inventor as Anger camera, enabled for the first time dynamic imaging studies of functional phenomena in vivo. Nuclear medicine found its feet. To the present day, Anger’s basic concept of photon detection by scintillation crystals coupled to photomultipliers remains the standard layout of current SPECT and PET systems.

When represented in a two-dimensional detector plane, though, photons originating in different planes of the body cannot be differentiated. For a more appropriate anatomical allocation of the photon signal, the resolution of the third dimension was required. The physician David Kuhl and the engineer Roy Edwards pioneered tomographic medical imaging, introducing the concept of “rotational scanning” in 1958 [6]. Their work involved the development of basic methodologies for acquisition, reconstruction, and display of data, acquired with multiple rotating detectors, allowing to generate a three-dimensional image. It took until 1964, when sufficient computational power became available, until the first human SPECT scans of brain and body could be performed [7]. Computer-filtered back projection replaced the initial optical back projection.

Only in 1971 X-ray computed tomography was introduced by Godfrey Hounsfield also for morphological imaging, for which he received a Nobel prize in 1979 [8]. It is unclear, whether or not Hounsfield knew Kuhl’s and Edward’s prior work.

On their way through the tissue, photons undergo attenuation and scatter, which both degrade image quality, and hence the precision of anatomical allocation of a photon source in the body, even when using three-dimensional SPECT imaging. Photons originating from annihilation of a positron with an electron, though, have the peculiarity to occur as pairs, leaving the spot of annihilation in an angle of 180° with a high energy of 511 MeV. These features allow localization of an annihilation event in the body based on simultaneous – “coincident” – activation of opposite scintillation detectors, which avoids the need for “focusing” by collimation, as in SPECT, a procedure, which excludes 99.99% of available photons in the body from contributing to image generation. Accordingly, positron emission tomography (PET) has a much higher sensitivity, better spatial resolution, and quantitative accuracy, compared to SPECT. The basic scanner layout for tomographic positron imaging by coincidence detection was—again – initially conceived by David Kuhl in the 1960s.

In 1973, the first clinically used PET scanner was built at the UCLA by Edward Hoffman, Michael Ter-Pogossian, and Michael Phelps, initially for brain imaging only, followed by a first whole-body scanner in 1977 [9]. With this achievement, PET was established as a general methodology to non-invasively image and quantitatively measure physiological phenomena in vivo. To fully exploit the potential of the new methodology, dedicated positron-emitting radiopharmaceuticals were needed, as well as methodologies, to analyze and interpret the now truly four-dimensional complex data sets, which then consumed the performance of the most powerful computer systems available at the time, a likable feature nuclear medicine has retained to the present day. Two handful of PET institutions in Japan, Europe, and the USA developed into the drivers of method development which attracted the most talented, creatively thinking scientists from any discipline and all over the world. Not surprisingly most of these institutions were run by scientists, mostly physicists and chemists, rather than physicians, who tend—often hindered by a widespread professional conceit—to be less able to create truly interdisciplinary teams at eye level. Where until the 1990s nuclear medicine was often condescendingly referred to as “Unclear Medicine”, holistically thinking scientific visionaries like Terry Jones, Michael Welsh, Bengt Langström, Jun Hatazawa, Adriaan Lammertsma, André Luxen, or Richard Baum have established today’s perception of nuclear medicine as the embodiment of precision medicine per se.

For imaging of physiology, radiolabeled tracer molecules—rather than just elemental radionuclides such as 131iodine, 133xenon, 67gallium, or 201thallium—were required. The basic methodologies to introduce radionuclides into organic molecules, though, had yet to be developed. Alfred P. Wolf at Brookhaven, together with his co-workers Tatsuo Ido, Joanna Fowler, Michael Welsh, and Gerhard Stöcklin developed the first methods to introduce 14carbon [10], 11carbon [11] and later 18fluorine into chemical syntheses [12], which all of a sudden enabled covalent radiolabeling of any organic molecule—at least theoretically.

These methods were soon widely used, in order to explore derivatives of simple biomolecules as potential tracers, to address the carbohydrate, the nucleic acid, and protein metabolism. Ido, Fowler, and Wolf in 1976 first synthesized 18F-fluorodesoxy-glucose (FDG) [13], Hiroshi Fukuda and Ren Iwata explored 18F-fluorodesoxy-mannose [14], Anthony Shields and John Grierson explored 18F-fluoro-thymidine [15], while Stöcklin [16] Kiichi Ishiwata, and others explored amino acids, from which 18F-fluoroethyl-tyrosine (FET) [17] and 18F-fluoro-DOPA [18] made their way into clinical medicine.

On the other side, the possibility to label pharmacologically active substances, in order to study drug biodistribution, metabolism, and more importantly, receptor occupancy in vivo was soon perceived. Marieannik and Bernard Mazière in Orsay, Henry N. Wagner, J. James Frost, and Robert Dannels in Baltimore, Philipp Elsinga and Aren van Waarde in Groningen, Joanna Fowler, Nora Volkow, and Stephen Dewie in Brookhaven, Lars Farde, Christer Halldin, and Bengt Langström in Sweden, Kazuhiko Yanai and Tatsuhaki Watanabe in Sendai or Olof Solin in Turku all deserve credit for having developed and introduced the basic methodologies for PET in vivo pharmacology.

The analysis of four-dimensional functional imaging data required new methods. The neuroscientist Albert Gjedde and the mathematician Clifford Patlak developed independently (1981 and 1983) graphical methods to analyze the pharmacokinetics of tracers involving irreversible uptake, such as FDG, now known as Gjedde-Patlak [19, 20] plot. For tracers binding reversibly to receptors and enzymes – as most pharmacologically active substances do – Jean Logan from Joanna Fowler’s group found in 1990 a graphical method to estimate the distribution volume from plasma activity curves, which allows to determine receptor occupancy. This method is now referred to as the Logan plot [21].

In order to comprehensively characterize the biological behavior of a physiological substrate or a drug, not only the binding characteristics to their receptors in vivo, i.e., binding constants, are of interest, but also pharmacokinetic properties, such as biodistribution, penetration of the blood-brain barrier, plasma protein binding, the relationship of plasma concentration and receptor occupancy, as well as kinetics and pattern of metabolization and excretion. Once a positron-emitter-labeled analog is available, all these questions can be addressed in a single experiment. Since in particular metabolic pathways may differ considerably between animal models and humans, it is of high interest to get early information from the human target species.

Mats Bergström and Bengt Langström from Uppsala were the first to suggest in 2003 the concept of PET microdosing for the development of new drugs [22]. Microdosing means the administration of less than 100 μg or 30 nMol of a substance, which is generally assumed to be pharmacologically inactive. As the mass dose of PET tracers is nearly always below this margin, new compounds can mostly be administered to humans in the context of PET microdosing studies, also referred to as phase 0 studies, with a significantly reduced toxicological characterization. Such a method allows to verify expected substance properties, and on the other hand to recognize possible development roadblocks, such as unexpectedly high protein binding, fast metabolization, or lacking brain uptake, early on during the development process. Nowadays, many new drug candidates intended for CNS indications and beyond contain fluorine atoms, in order to allow a quantitative characterization of pharmacology in vivo using chemically identical 18F-analogs in PET microdosing studies.

In contrast, for routine diagnostics in clinical medicine, a universal PET tracer is desirable, which allows to detect physiological abnormalities in the body, without the need for advanced data processing. In other words, producing a foolproof signal. Such a tracer is 18F-FDG. In 1976, the physician Abass Alavi was the first to administer 18F-FDG, synthesized by Joanna Fowler, to two healthy human volunteers, which were scanned on the UCLA scanner [23].

Today, 18F-FDG is the by far most frequently used PET radiopharmaceutical, accounting for more than 90% of an estimated more than four million PET scans conducted around the world annually. This broad clinical adoption of PET imaging would not have been possible without a remarkable industrialization of all aspects of production and distribution for the very short-lived PET radiopharmaceuticals.

Introducing a robust stereospecific, high-yield radiosynthesis for carrier-free FDG, Kurt Hamacher in 1986 laid the ground for this development. The method—using an aminopolyether, the legendary Kryptofix 222, as phase transfer catalyst for 18F—has revolutionized the preparation of 18F-labeled tracers in general, and laid the ground for commercial high-volume routine preparation of PET radiopharmaceuticals [24]. The high radiation exposure associated with the handling of positron-emitters had early on stimulated the automated production of PET radiopharmaceuticals. In 1986, JW Brodack, Michael Kilbourn, and Michael Welsh were the first to use an adapted commercial lab automation system for the preparation of a PET radiopharmaceutical [25]. In the 1990s, Bruno Nebeling, Jean-Luc Morelle, and later Vincent Tadino designed the first dedicated automated synthesizer modules for 18F- and 11C-labeling reactions, and commercialized these successfully. In parallel, new commercial players were created to make available materials for isotope production, pharmaceutical grade chemicals for GMP radiosynthesis, as well as radiopharmacy networks, able to provide reliable manufacturing and distribution of PET radiopharmaceuticals, opening the possibility to operate clinical PET imaging sites without an own radiopharmacy. Today manufacturing PET radiopharmaceuticals is a several hundred-million-euro business, with strong competition, which has led to a remarkably complete geographical coverage of tracer supply in the developed countries.

The production of the required positron-emitting radionuclides does—in contrast to most single photon emitters—not require a nuclear reactor as neutron or proton beam source, which involves always a massive investment, typically only amenable to governments or large monopolistic utility companies, but is possible using smaller scale particle accelerators, the cyclotrons.

The basic design of the cyclotron was conceived by Leo Szilard already in 1929 [26], and reduced to practice independently by Ernest Lawrence in 1931 in Berkeley [27], who in 1939 received a Nobel prize for it, 1935 by Nishina and Nishikawa in Japan, 1937 by George Gamov and Igor Kurtchatov in Russia, and 1943 by Walther Bothe and Wolfgang Gentner in Germany. The times and the names indicate that the primary motivation for their research was not at all medical in nature. Nevertheless, having experienced the non-medical applications of their work, most of these brilliant physicists, many of which Nobel laureates, became driving forces to establish peaceful applications of radioactivity.

In this spirit, the Brookhaven National Laboratory (BNL) was founded in 1947, which stimulated the establishment of similar nuclear research institutions in many countries around the world, and initially educated many of their leading scientists. At the international level, the IAEA with its mission “atoms for peace” was created in 1957.

With the increasing installation of PET scanners in academia, the need for positron-emitting radionuclides ideally produced in the vicinity of the scanner, grew. Rather than large-scale research installations, small self-shielded cyclotrons, amenable to medical institutions, were needed. Newly created companies, started by academic physicists active in the field served this need. In 1983 Michael Phelps, together with Ronald Nutt, and Terry Douglass founded CTI, which—besides cyclotrons – later also manufactured PET scanners, and considerably contributed to the clinical establishment of the methodology. CTI was acquired by SIEMENS in 2005. In 1986, Yves Jongen of Louvain founded IBA, which remains a leading independent manufacturer of medical cyclotrons.

To the present day, however, the mainstay of diagnostic procedures in nuclear medicine is being conducted using 99mTechnetium as a radiolabel. From an estimated 40 million annual diagnostic procedures globally, approximately 85% involve 99mTc-labeled radiopharmaceuticals. Ironically, the development of this radionuclide was only by chance.

When Walt Tucker and Margaret Greene, chemists at BNL, tried in the late 1950s to chromatographically isolate 132Iodine – which they believed might be favorable for diagnostic procedures in view of a 2-h short half-life – from 132Tellurium out of reactor fission products, they found it to be contaminated with 99Molybdenum, which decayed to 99mTechnetium. Due to the chemical similarity of the 132Te/132I and 99Mo/99mTc nuclide pairs, they could recycle their methodology to create the first 99Mo/99mTc generator, which they nicknamed a “moly cow”, since the mother nuclide 99Mo stays immobilized on the alumina generator column, from which pure 99mTc can be eluted by physiological saline for further use [28]. When trying to apply for a patent, the patent office replied visionarily, “We are not aware of a potential market for 99mTc great enough to encourage one to undertake the risk of patenting”.

Powell “Jim” Richards, though, the head of isotope production at BNL recognized, that 99mTc had—compared to all other accessible nuclides at the time—by far the best physical properties for medical imaging with the just invented Anger camera. The photon emission of 140 keV had a sufficient tissue penetration, and was at the same time low enough, to allow efficient collimation for “focusing”. The comparatively short half-life minimized radiation exposure for the patient, and in addition, the difference in half-lives between parent and daughter nuclide (66 vs. 6 h) allowed to ship the generator to hospitals, allowing to generate the radionuclide for diagnostic practice on the spot, which could create accessibility.

Richards and Suresh Srivastava, who later substantially refined the generator technology and technetium labeling [29], started lobbying the medical and scientific community for the nuclide. Richards first presented on 99mTc in 1960 at the seventh Electronic and Nuclear Symposium in Rome. On his way, he met Paul V. Harper, from the newly founded Argonne Cancer Research Hospital in Chicago, who ordered in 1961 the first 99Mo/99mTc generator from BNL. He introduced 99mTc for blood flow measurements of liver and kidney. In the same year Harper could also demonstrate the use of 99mTc for imaging thyroid and brain tumors, a more than welcome alternative to pneumoencephalography in the pre-CT era [30]. His methods—obviously satisfying an unmet need—were soon widely adopted. By 1967 BNL had to transfer the production of the 99Mo/99mTc generator to commercial providers, able to industrially scale manufacturing, in order to keep up with the rising clinical demand. Today, the key role of the trio 99mTc, Richards, and Harper for establishing nuclear medicine as a medical specialty in its own right is widely recognized.

In order to extend the use of the physically favorable 99mTechnetium to biological targets, not addressable by virtue of perfusion or their avidity for the iodine-like properties of the element, Richards started in the mid-1960s his search for ways to use technetium 99mTc as a radiolabel for more complex tracers. As metals do not form covalent bonds, complexation agents were required, allowing to bind the radiometal to pharmacophores, intended to bind to a biological target. However, the chemistry proved to be tricky. Only in 1970 William Eckelman and Richards succeeded to identify DTPA as a universal complexing agent for 99mTechnetium, which was subsequently used to radiolabel not only multiple pharmacophores with 99mTechnetium, but also with other radiometals such as 111Indium [31].

The general possibility to label pharmacophores with radiometals stimulated the research to enable internal radiotherapy with beta-emitting radiometals, extending the principles of 131I-radioiodine therapy, to targets not addressable with elemental 131I.

Donald Hnatowich, as well as Sally and Gerald DeNardo in the 1980s, were among the first to consider 90Yttrium as an alternative to 131I for labeling pharmacophores for therapeutic purposes [32]. After initial dissatisfactory attempts with DTPA, which produced unstable complexes, Shrikant Deshpande and Sally Denardo in 1989 finally identified DOTA as a suitable chelator for 90Y [33].

DOTA offered for the first time the possibility to form stable complexes not only with 90Y, but also with other tri-valent radiometals such as 177Lutetium, a beta-emitter with shorter particle path length in tissue, featuring an imageable gamma emission, alpha-emitters such as 225Actinium, and at the same time diagnostic nuclides for SPECT and PET, like 111Indium, 68Gallium, or 64Copper.

Only with the introduction of DOTA, as a multivalent chelator suitable to complex diagnostic and therapeutic nuclides alike, nuclear medicine finally became technically able to unfold its full theranostic potential. Pharmacophores as diverse as monoclonal antibodies, peptides, small molecules, or nanobodies have since been labeled for PET and SPECT imaging, as well as for therapeutic administration with either beta- or alpha-emitters. Theoretically, any given pharmacophore can—by exchange of the radiolabel—be multiply used as a targeting agent to diagnostically identify target expression, to measure target engagement, and then to plan and administer a therapeutic intervention, using a therapeutic payload.

The general access to today’s most popular theranostic nuclide pair 68Ga and 177Lu, both efficiently complexed by DOTA, was paved by Frank Rösch and Konstantin Zhernosekov from Mainz, who translated what Tucker and Green had done for 99mTc into the PET world, making available the 68Ge/68Ga generator [34], allowing on-site production of positron-labeled tracers without a cyclotron, and devising – together with Nicolai A. Lebedev from Dubna—an efficient and reliable way to produce carrier-free 177Lu, today’s standard therapy radionuclide [35].

David Goldenberg is credited to have first used antibodies as targeting agents in 1977 [36]. It took until the late 1990s that peptide ligands were explored for diagnosis and treatment by many groups. The somatostatin receptor system, well known from established peptide therapeutics for neuroendocrine tumors, served as a model to pave the way for many other receptor systems, today addressed by most diverse peptide therapeutics under development. Claude Reubi, Helmut Maecke, and Marion de Jong paved the way for pharmacology and chemistry. Jan Müller-Brand in Basel, Dik Kwekkeboom and Eric Krenning in Rotterdam, as well as Richard Baum in Bad Berka were among the earliest clinical adopters of the method [37,38,39], and later continued the development of this new therapy modality in academia for the benefit of their patients on their own, in the absence of any industrial interest for many years.

Thanks to important progress in instrumentation, it is now possible to also quantitatively reconstruct SPECT, yielding information in terms of Bq/mL tissue, which is a prerequisite for accurate detection and dosimetry of therapeutic nuclides in vivo, most of which are single photon emitters. Bruce Hasegawa is credited for having first suggested to combine SPECT and CT imaging to acquire simultaneous SPECT/CT, allowing for voxel-based scatter and attenuation correction [40]. Hidehiro Iida from Osaka [41], as well as Dale Bailey from Sydney [42] practically implemented the absolute quantification of SPECT images (“QSPECT”) in the years 2000, with which it is now possible not only to image and localize, but also to quantitatively measure the radiation absorbed dose, conveyed by a therapeutic radiopharmaceutical to a tumor in Gray (Gy), as established in external field radiation therapy. It took until the years 2010, though, that the major camera manufacturers started implementing this methodology into their SPECT scanners, as they feared, their PET camera business—much higher priced in reason of the possibility to quantify the image data—might suffer.

With these achievements the basic technical toolbox of nuclear medicine as a universal diagnostic and therapeutic discipline was complete.

Nevertheless, PET instrumentation has since seen quantum leaps in sensitivity and resolution. The introduction of simultaneous PET/CT by Thomas Beyer and David Townsend in 2000 has increased the sensitivity of PET by a factor of 40 compared to the initial UCLA instrument [43]. Time-of-flight detection further improved the signal-to-noise ratio of the PET images [44]. The development of a total body PET scanner, driven by Simon Cherry and colleagues, now allows to further increase the sensitivity of PET by a factor of 100, allowing to reduce activity doses of diagnostic radiopharmaceuticals accordingly, without any loss in image quality [45]. Provided, however, that the budgets involved in health care—not only for instrumentation—will not prohibit a wider adoption.

Last but not least: PET/MRI. Its development has been much more of a technical challenge than the PET/CT, considering the need to harmonize mutually incompatible magnetic and scintillation detector systems in a very small space, and to come up with attenuation correction methods based on an MRI, rather than a CT image, to which Bernd Pichler from Tübingen provided key contributions. Now, after years of enthusiastic search by the scientific community for the “killer application”, with series of own symposia conducted in the pre-pandemic era, the unique clinical value of PET/MRI becomes increasingly clear in situations, where a detailed morphological or perfusion information are needed in addition to molecular information [46].

With the complete methodology of nuclear medicine available today, we have become able to read the book of life, at least its physiological basis. We may need artificial intelligence to master the flood of information and to decipher its meaning in the future, but we will always need academic teachers, and great humans, able to welcome new members to the community, and spread the flame of curiosity.

Thank you, Richard.