Sustainable production of orphan radionuclides at Wisconsin

Abstract

Over a hundred proton-induced reactions have been studied at the University of Wisconsin Medical Physics department since the installation of the first CTI RDS 112 in 1985. The focus has been to measure thick target yields at 11 MeV, in an effort to concentrate on the practical production of positron emitting radionuclides that have favorable decay characteristics, high yields and the potential for labeling pivotal biological tracers. This review covers our recent advances to scale-up the production of the heavy halogens and transition metals as feed-stock for non-conventional PET tracers that are currently attracting increased attention in oncology.

Positron emission tomography (PET) extends the promise of quantitatively imaging trace levels of pivotal bio-compounds, ranging in simplicity from oxygen to potent neural ligands implicated in the most subtle of disease states. This broad brush is possible because of the chemical versatility of the “conventional” radionuclides that act as feedstock in the synthesis of the PET radiopharmaceutical. The quartet, ¹⁸F, ¹¹C, ¹³N and ¹⁵O, have long been regarded as the major building blocks, listed in descending order of half lives and, not surprisingly, the number of ligands that have been successfully developed.

This simplistic view overlooks the hundreds of other positron emitting radionuclides that span the proton-rich terraces above the valley of β-stability, from ¹⁰C to the rare earths. Viewed from this northwest landscape, an increasing number of candidates emerge with β⁺ branching ratios, half lives and attractive chemistry that warrant serious re-consideration to label biological tracers, even if bearing the sobriquet of “non-conventional” or “orphan” radionuclides. In particular, the heavier halogens ^34mCl, ⁷⁶Br and ¹²⁴I, as well as a number of transition metals, are poised to fill a significant role in labeling diagnostic agents directed toward the study of neoplastic disease.

Resources

Accelerators

The “heavy steel” of any PET site is the accelerator, which at the University of Wisconsin has employed in sequence: an EN tandem (installed in 1957; 12 MeV protons and deuterons; 10 μA; UW Physics Department), the first CTI RDS 112 (1985; 11 MeV protons; 50 μA; UW Medical Physics Department) and, most recently, an NEC 9SDH-2 tandem (1999; 6 MeV protons and deuterons; 120 μA; UW Keck Neuroscience Laboratory). All of three accelerators are bunkered with full access to targets on multiple beam-lines that allow target development and telemetry in unrestrained geometry. This paper will discuss work done primarily on the 11 MeV proton cyclotron over the past two decades, investigating the yields and applications of the positron emitters between 6≤Z≤53.

Targets

Liquid targets include: a high-pressure niobium gridded target¹ for [¹⁸F]fluoride; a flow-through [¹³N]ammonia target;² and a He-sparged cell for the steady state production of the noble gas pair ⁷⁹Kr/^81mKr from the irradiation of bromoform. Various gas targets produce ¹⁸F₂ from ¹⁸O₂ by the “two-shoot” technique,³ and a number of aluminum, stainless and niobium flow-through targets for the production of ¹⁴O₂, ¹⁵O₂ and ¹¹CH₄.

About a dozen specialty targets have been developed to irradiate solid materials, with the greatest challenge being the production of ¹⁰CO₂ for steady state inhalation to image regional blood flow. The absence of recirculation with ¹⁰CO₂ (t_1/2=19 s) is evident in Figure 1. This collage shows the equilibrium distribution of activity during constant inhalation of the four radio-isotopic forms of carbon dioxide, cognizant of the fact that label exchange results in the immediate transfer of labeled oxygen to water. While ¹⁰CO₂ has found modest clinical appeal,⁴ its production demands serious attention to the irradiation of helium-swept, incandescent ¹⁰B₂¹⁸O₃ targets (Figure 2). At the extreme limit, a vertical switching magnet was constructed to focus a downward-directed beam onto the most corrosive of the molten targets, constrained by gravity⁵ from burning their way into the cyclotron interior. These hard-won lessons, learned in the crucible of ¹⁰CO₂ production, bear fruit in the more relaxed bombardment conditions leading to ¹²⁴I and the transition metals.

Figure 1.—

The sensitivity of the steady state technique for the measurement of regional blood flow is determined between the “impedance matching” of mass specific flow f (mL/s/gram of tissue = s⁻¹) to the mean decay rate λ=ln 2/t_1/2. The optimal half life is 71 s for grey matter flow, where a 10% variation in f would result in a 5% variation in the equilibrium activity, slightly displaced to the right when accounting for the lung-to-brain arterial delay c_a(t). The PET steady state flow images of the normal subjects reflect the increased contrast at the shorter lifetime CO₂ tracers.

Figure 2.—

The ¹⁰B₂O₃ target with He-swept ¹⁰CO₂ represents the extreme case of handling molten materials in the beam strike. Adjusting the helium flow, beam and thermal cantilevering determines the target temperature, which optimizes the diffusion and transport of the 19-s product.

Telemetry

To irradiate a costly target at kW power levels without constantly watching the immediate consequences is an invitation to disaster. The visible signature of a nuclear reaction is the outgoing, penetrating radiation, namely neutrons and gammas. The prompt γ spectrum is dominated by inelastic scattering during irradiation, which at least confirms that the correct target is receiving beam. Conversely, for target materials electroplated onto gold blanks, the sudden appearance of the signature 497 keV γ from ¹⁹⁷Au(p,p’) is a warning of a thin, or flaked target. The concurrent neutron flux, absorbed in the bunker walls, gives rise to an overwhelming γ background, necessitating massive collimation of the γ detector to restrict the solid angle to the target only. Finally, the γ spectrum is further confounded by neutron activation of the γ detector itself. A germanium detector would only last minutes in the fast neutron flux, while NaI(Tl) quickly activates with ²⁴Na (isomeric and ground state) and ¹²⁸I. Bismuth germanate and yttrium aluminum perovskite do not activate, but energy resolution is sacrificed. Both LSO and the lanthanum trihalides (LaX₃ (Ce); X=Cl, Br) have excellent energy resolution and speed needed in the high γ flux. The bromide is favored for energy resolution, but suffers from modest neutron activation to ⁸⁰Br that requires correction of the γ signal time course. After a protracted search, two γ detectors are employed. The first is a 4×4 cm LaBr₃ detector, laser aligned within its 200 kg lead collimator at 5 m to view the 1 μ-steradian target field (Eγ>1 MeV). The second is a simple plastic scintillator mounted on a photomultiplier tube operated in the current mode. By consistently monitoring these γ signals, a wide variety of target problems can be diagnosed and remedied.

The neutron flux is sensed with a BF₃ “long counter” at several meters from the target, chosen for its excellent n/γ discrimination. Uranium-235 fission counters have similar favorable γ insensitivity, but seem to have an operational lifetime only of the order of a year. The geometry and 10⁻⁴ sensitivity of the long counter result in 10⁴ cps counting rates at realistic bombardment conditions, making the neutron flux the primary signature of the irradiation progress. Crowbars, vapor voids and drifting beam over target thickness inhomogeneities are all immediately apparent in a glance at the display of the neutron and gamma flux, logged at 10-s intervals. Furthermore, the zero background makes the neutron counting rate n the perfect input signal for the leaky integration of the difference equation ΔN/Δt = n-λN,, describing the incremental buildup of the product N, proportional to the neutron counting rate less the decay. Gammas, neutrons and the leaky integrator are displayed on the Labview interface control page that forms the new operating system for our legacy RDS 112.⁶

The raw neutron and γ flux is shown in Figure 3, normalized to 1 μA of 11 MeV protons on such thick targets as elemental ¹⁹⁷Au, ¹⁴N₂, ^6lNi, ⁸⁹Y, ⁶⁴Ni, ^natKr, and ¹⁸O₂. The advantage of choosing a target backing substrate and slit material from high-Z materials (Au, Ta) is obvious in terms of reducing neutron backgrounds, that would otherwise obscure the desired target signal. Nonetheless, the ratio of the neutron flux from irradiating highly enriched [¹⁸O]-water to natural water is only a factor of three. Given that 11 MeV protons on [¹⁶O]-water cannot result in neutrons, the failure to observe a factor of 500 expected from the natural abundances alerts us to the neutron background from beam lost on slits, foils and other internal structures.

Figure 3.—

A linear plot (in arbitrary units) of the relative γ to neutron flux rates per unit beam current, demonstrating the “signature” of various thick target substrates under 11 MeV proton irradiation. Elements without superscripts denote the naturally occurring abundance.

Methodology

Five unconventional PET radionuclides with longer half-lives have captured the interest of researchers at Wisconsin (Table I), along with the reaction, half-life, target chosen, measured end-of-saturated-bombardment yield extrapolated to a thick target of the stated target compound, at 100% enrichment, a representative application and local references.

Table I.—

Five unconventional positron emission tomography radionuclides with longer half-lives.

Reaction	t1/2	Target	Yield (mCi/μA)	Application	Reference
61Ni(p,n)61Cu	3.4 h	61Ni on Au	105	Preclinical Cu-ATSM	7
64Ni(p,n)64Cu	12.7 h	64Ni on Au	181	Large molecules	8
86Sr(p,n)86Y	14.7 h	86SrCO3	27	PET surrogate for 90Y	9–12
natY(p,n)89Zr	78 h	natY-foil	100	Peptide labeling	13
124Te(p,n)124I	4.2 d	Al2124Te3	33	Many	14, 15

Copper isotopes

The relative dosimetry, logistics and economics of the various copper isotopes have been well described in the literature,¹⁶ with 4 isotopes, ⁶⁰Cu, ^6lCu, ⁶²Cu and ⁶⁴Cu contending for clinical applications. Copper-64, with its 12.7 h half-life stands alone for labeling ligands with protracted blood clearance, or for national distribution by commercial carriers. This same lifetime confounds its use in sequential patient studies, or in the case of veterinary patients, impeding the immediate release of the pet to its owner following the study. Copper-61 has more favorable dosimetry and positron imaging characteristics, and has found a preclinical role in the imaging of tumor hypoxia responding to therapy. Our experience shows that more than an hour is needed for the tumor distribution of Cu-ATSM to stabilize, ruling out the shorter-lived ⁶⁰Cu or generator-accessible ⁶²Cu.

The close race between ^6lCu and ⁶⁴Cu to label Cu-ATSM hinges on a second tier of considerations: commercial availability, flexibility of patient scheduling, cyclotron loading, and coupling to fixed distribution schedules. If these considerations dictate that ^6lCu-ATSM is favored, then three entrance channels present themselves at our site: ^6lNi(p,n)⁶¹Cu on enriched ^6lNi, ⁶⁴Zn(p,α)⁶¹Cu or ⁶⁰Ni(d,n)^6lCu at 6 MeV deuteron energy on the Keck Lab tandem, with yields in the ratio of 100:5:4.

The target preparation of both the enriched nickel targets are identical, and both tightened by the $17/mg cost of the enriched elemental materials. The metals are electro-plated⁷ onto gold discs (99.999%), irradiated, digested, and separated as the chloro-adducts. Nickel elutes in 6N HCl, Co-isotopes in 4N HCl and Cu-isotopes in 0.1N HCl. The nickel is recycled with 95±2% mass recovery, and the copper isotopes are collected for labeling and distribution. TETA-titration shows that ⁶⁴CuCl₂ has a specific activity of the order of 19 Ci/μMole, approaching 10% of the carrier-free level. Over the past 2 years, our 20 Ci production has been shared with colleagues at other institutions, with 10 Ci leaving our dock.

Finally, in the case of microspheres for selective internal radiation therapy (SIRT), one of the authors (R. G. S.) has developed a technique to employ ⁶⁴Cu-labeling, which allowing quantitative dosimetry to the intra-arterial administration of co-injected ⁹⁰Y-SIR-spheres™ for the treatment of hepatic carcinoma.

Yttrium and zirconium isotopes

The mushrooming application of SIRT, based both on capillary-occlusive ⁹⁰Y-labeled microspheres and targeted large molecules, demands that PET bring its quantitative, non-invasive potential to bear on the task of imaging the spatial heterogeneity of the treatment. While considered a pure β- emitter, ⁹⁰Y actually has a 32 ppm positron branch from the E₀ decay of the first excited state of ⁹⁰Zr,¹⁷ permitting a crude PET imaging through the overwhelming bremsstrahlung. Far better is to co-inject the surrogate ⁸⁶Y-agent, or if need be, ⁸⁹Zr-agent with a matching half-life, if the different oxidation states do not seriously affect the chelation.

Early attempts to irradiate metallic strontium targets at high power led to serious problems with target materials and long-lived yttrium products sputtered into the cyclotron interior. The carbonates and oxides of strontium allow 10–20 μ-amp currents if careful attention is paid to removal of beam power with both water cooling below and turbulent helium flow over the of the beam strike on the target surface.⁹ Separation of the ⁸⁶Y from the alkaline ⁸⁶Sr²⁺ digest makes use of the remarkable affinity of the yttrium hydroxide to filter paper.^{18, 19}

The production of ⁸⁹Zr has the advantage of starting from the naturally occurring ⁸⁹Y target material, obviating the need for any target reclamation. Straightforward dissolution and ion exchange chromatography suffices to separate Y³⁺ from ⁸⁹Zr⁴⁺, as shown by X-ray fluorescence²⁰ and ICP-MS. While both ⁸⁶Y and ⁸⁹Zr present additional gammas, ⁸⁹Zr is favored because its 909 keV γ is delayed by seconds in its isomeric decay, falling far outside the coincidence window in PET imaging.

Iodine-124

Chemists love iodine isotopes, with their labeling versatility under gentle conditions. Iodine-124, once considered a noxious contaminant of ¹²³I from the irradiation of tellurium targets, is now embraced as the poster child of PET for peptide and protein labeling. Alpha irradiation of natural antimony would provide ¹²⁴I for animal studies, albeit contaminated with ¹²⁶I, but few cyclotrons with good α beams still remain. The proton irradiation of ¹²⁴Te as a TeO₂/Al₂O₃ glass at shallow beam incidence is widely used,^{21, 22} but is beam-limited by the volatility of both the tellurium and iodine. The binary compound Al₂Te₃ has both a higher melting point and a greater effective fraction of tellurium than other candidates, and can be made from the elements with an exothermic reaction under argon with an enthalpy of formation of 267 kJ/mole. With ¹²⁴Te, the black glassy aluminum telluride^{14, 23} in a target fixture similar to Figure 2 permits beam currents up to 20 μA at 11 MeV, resulting in 173 μCi/μA-hr. Dry distillation at 910 °C in a gentle steam of dry air deposits the ¹²⁴I downstream on a cooled platinum wire, from which it can be quantitatively harvested in weak base. A commercial system is under development (Comecer SpA, Castel Bolgnese, Ravenna, Italy).

Results and conclusions

The decision to scale up the local production of the five radioisotopes in Table I came about when the expected benefits outweighed the costs. At the tipping point, the benefits are potential, belonging to the realm of the future, while the costs are real and immediate. In particular, the concept of opportunity cost, the value of an opportunity forgone, dominates both sides of the balance: to wait for some commercial supplier to provide ⁶⁴Cu on a nationwide level means forgoing, delaying research opportunities that need it now. On the other hand, to commit the resources for enriched ⁶⁴Ni, gold blanks, Ultrex reagents and large blocks of beam time means that other cyclotron projects will be left undone. Opportunity cost accounting is notoriously fickle, but a reasonable value for beam time can be estimated by its worth in the likely alternate enterprise of FDG production, easily justifying a $300/h figure. The case for pursuing local ⁶⁴Cu production is the most obvious, given the mushrooming interest in labeling a wide variety of large molecules, a half-life well matched to national distribution of the excess, and the large cross section even at 11 MeV that permits Curie levels to be made during the off hours.

The regular production of ⁶⁴Cu starts each Monday morning, with 5 local research groups able to count on tens of millicuries on Monday, a few millicuries on Tuesday and microcuries throughout the rest of the week. MicroPET (P4 and Inveon) sensitivities are so great that targeted tracers bearing tens of microcries suffice in the nude mouse models. Generally half of our ⁶⁴Cu production is shipped by FedEx overnight priority to several collegial institutions, which provide us with valuable feedback as we improve our specific activity. This ratio of ⁶⁴Cu to cold copper has now reached almost 1/10^th of the carrier free level shown by TETA-titration, and our distant colleagues are reporting much improved labeling of their nano-mole levels of ligand.

The production of non-conventional radionuclides on the UW RDS 112 started out as a shotgun reconnaissance of the proton-rich side of the valley of β stability, crudely determining what was feasible with 11 MeV protons before succumbing to the Coulomb barrier in the rare earths. This search matured into a fertile field for a half dozen PhD dissertations, each improving and extending the reach of our legacy cyclotron. Now more than a simple survival strategy, the production of these orphan isotopes opens the doorway to a sustainable future: a seasoned cyclotron with a new brain, with the will and capacity to provide a significant fraction of the national needs.