PET Project

There’s an undeniable undertone of urgency as Jacob Hooker guides me through the suite of labs at the Athinoula A. Martinos Center for Biomedical Imaging in Boston. Hooker, the center’s director of radiochemistry, seems remarkably calm while chemists in lab coats dodge us to go about their work. A Geiger counter steadily beeping only heightens the sense that no time is wasted here.

These labs, part of Massachusetts General Hospital and Harvard Medical School, are dedicated to making compounds for positron emission tomography, or PET. PET imaging uses radiotracers—short-lived, positron-emitting radioactive isotopes—to create a picture of particular parts of the brain or body.

Next to the rooms where the radiochemists go about their business, there is another room that houses a modest-looking metal machine encased in a massive concrete and lead shield that looks a little like an igloo. This is a cyclotron. Here radioisotopes are made, for example, by bombarding ¹⁸O-enriched water with protons to make ¹⁸F^–. Thin tubes run from this room though the ceiling to the adjacent lab, delivering the just-synthesized radioactive compounds into large cabinets lined in lead. Any subsequent chemistry is done in these cabinets via automated processes.

Today, Hooker explains, the scientists are making a neuroimaging radiotracer that contains carbon-11, an isotope that loses half of its radioactive potency every 20 minutes. “All this has to be magically timed,” he says. “With carbon we try to hit our marks plus or minus two minutes.” When people first start working with ¹¹C, the pace seems incredibly fast, Hooker notes. “But the world slows down for you, and then isotopes like fluorine-18 seem to give you tons of time.”

¹⁸F has a 110-minute half-life, so molecules that incorporate this isotope are attractive for many kinds of medical and drug studies. For example, the radiotracer ¹⁸F-FDG, or 2-deoxy-2-[¹⁸F]fluoro-d-glucose, is a clinical tool used for cancer diagnosis.

Among synthetic organic chemists, ¹⁸F has become a hot topic lately. With the recent renaissance in fluorination chemistry has come the promise that some of the new methods for tacking fluorine atoms onto small molecules might lead to new ways of doing the same with ¹⁸F. Such synthetic tools could, in turn, lead to new radiotracers for diagnostic use, pharmaceutical research, and better ways of making existing tracers.

But making the leap from “cold,” or nonradioactive, fluorinations to radioactive fluorinations isn’t as simple as switching on a cyclotron.

“Just because we can do fluorination with fluoride does not mean we have a good ¹⁸F-fluorination method yet,” says Tobias Ritter, a chemistry professor at Harvard University who collaborates with Hooker. “When you think about chemistry and isotope effects, ¹⁸F should really behave more or less the same as ¹⁹F, and rigorously that’s true. But the conditions that you need to use when you do ¹⁸F chemistry are so dramatically different that ¹⁸F chemistry often does not have much to do with ¹⁹F chemistry.”

To begin with, there is the matter of time. Because the radioisotope is quickly decaying, all the steps that occur once the ¹⁸F is made, including synthesis, separation, and purification, must take as little time as possible. Ideally, everything can be done within an hour.

Another thing that is vastly different is stoichiometry. Normally ¹⁸F^– is present in very minute quantities compared with the catalyst and substrate. That’s very different from most cold catalytic fluorinations, which use an excess of fluoride.

The most practical ¹⁸F-fluorinating agents use ¹⁸F^– rather than ¹⁸F⁺ reagents as a fluorine source. That’s because ¹⁸F^– is generated from ¹⁸O-enriched water, while ¹⁸F⁺ reagents come from radioactive F₂, a gas most PET centers don’t use. Also, because ¹⁸F^– is generated from ¹⁸O-enriched water, it’s tough to exclude moisture from the reaction. Thus, any fluorination catalysts that are supersensitive to water would be bound to fail if they used ¹⁸F^– as a reagent.

A key challenge in designing ¹⁸F-fluorination reactions is to make them operationally simple. “The field has made good progress in late-stage fluorination, but we have not yet made very good progress in having such a good reaction that the people who are actually going to do the imaging care enough to adopt it,” Ritter says. “That’s a lesson that we learned the hard way.”

In 2011, Ritter, Hooker, and colleagues published an ¹⁸F-fluorination method in Science (DOI: 10.1126/science.1212625). The reaction uses a palladium-¹⁸F complex to fluorinate an aryl-palladium complex. “This was all very nice chemistry, but it is operationally not simple enough to make a real impact in ¹⁸F PET chemistry,” Ritter observes. “So just because we can do it with ¹⁸F does not mean that it is going to be used.” To reduce barriers to usability, “we really need to have a dump and stir reaction,” he says.

Ritter and Hooker therefore worked to develop simpler ¹⁸F-fluorinations with a nickel complex. Working in concert with an oxidant and aqueous ¹⁸F^–, this catalyst appends radioactive fluoride to aryl groups.

The researchers recently used the reaction to prepare the radiotracer [¹⁸F]MDL100907 (ACS Chem. Neurosci. 2014, DOI: 10.1021/cn500078e). In a study with nonhuman primates, they compared this compound to [¹¹C]MDL100907, a compound that’s been used to study serotonin receptors. MDL100907 contains a fluorine atom, but until now it’s been tough to radiolabel that spot. The ¹⁸F compound, the researchers say, has effectively equivalent distribution and a longer half-life, making it better for many types of studies.

“Today people label where they can label instead of where they should label,” says Véronique Gouverneur, a chemistry professor at the University of Oxford who has been at the forefront of translating catalytic late-stage fluorination reactions to reactions with ¹⁸F. “Hopefully radiochemists will have a fuller toolbox of reactions to label whatever they want wherever they want.”

Gouverneur’s group has reported many reactions that she hopes will add to this toolbox. In one recent example, she explains, they wanted to streamline access to fluorinated arenes from readily available aryl boronic esters. The reaction uses a copper complex that is easy to handle, air stable, and commercially available, as well as a fluoride source that comes directly from the cyclotron (Angew. Chem. Int. Ed. 2014, DOI: 10.1002/anie.201404436).

The researchers used the reaction to prepare 6-[¹⁸F]fluoro-l-dihydroxyphenylalanine, or [¹⁸F]FDOPA, a radiotracer with potentially broad use in neurology. In the clinic, the compound is made from radioactive fluorine gas, which not all PET centers have access to. “The reaction that we developed is logistically simple,” Gouverneur says. “I hope that the chemistry is going to be used broadly because the precursors and reagents don’t require expert handling.”

A few years ago, when Princeton University chemistry professor Abigail G. Doyle developed an enantioselective ring opening reaction of epoxides using fluoride, she initially thought she’d leave the radioactive translation to another group. But after chatting with Harvard’s Ritter, she changed her mind. “Tobias said, ‘No one is going to do it unless you show them how to because it’s not trivial to translate cold fluorination to radiofluorination,’ ” Doyle recalls.

So she teamed up with Hank F. Kung, a radiochemist at the University of Pennsylvania, to make the cold chemistry hot (J. Am. Chem. Soc. 2014, DOI: 10.1021/ja5025645). The group’s big challenge, she says, was figuring out how to incorporate ¹⁸F into the system. They knew that the active catalyst was a (salen)cobalt fluoride, so they had to figure out a way to make that material quickly from the radioactive fluoride that comes from the cyclotron.

“The solution that my student came up with, which I think is really quite straightforward and simple—and therefore quite powerful—was to tweak the way radiochemists make K¹⁸F,” Doyle says. This involves passing aqueous ¹⁸F^– through a quaternary ammonium anion exchange cartridge. The resin causes ¹⁸F^– to localize on the cartridge via anion exchange. Radiochemists can then pass potassium carbonate through that cartridge to elute off K¹⁸F.

“What Thomas found was that we can pass a solution of (salen)cobalt tosylate through that same cartridge and elute what we propose is the (salen)cobalt-¹⁸F complex,” Doyle says. The automation, she points out, is already established. “All you have to do is change out the solution of potassium carbonate for (salen)cobalt tosylate, so it should be something that nonexperts can perform readily.”

Indeed, the researchers were able to conduct the fluorination in a remote-controlled hot cell to prepare ¹⁸F-labeled fluoromisonidazole, or [¹⁸F]FMISO, a radiotracer for imaging tumor hypoxia.

John T. Groves, Doyle’s colleague in the Princeton chemistry department, also uses the ion exchange cartridge technique to make a different transition-metal fluoride catalyst. This compound, which is made from a (salen)manganese tosylate catalyst, fluorinates benzylic C–H bonds (J. Am. Chem. Soc. 2014, DOI: 10.1021/ja5039819). Groves’s group developed the reaction in collaboration with Hooker’s. The reaction, Groves says, tolerates a number of functional groups and can be used to fluorinate molecules of medium complexity. They use the reaction to append ¹⁸F to a number of drug molecules.

“When you discover something like we had, it really is of interest to see if you can make it useful to somebody,” Groves says. “There are only a handful of molecules that are approved by the Food & Drug Administration for PET imaging. I think that the field is on the cusp of making PET more practical because new imaging agents can be made and evaluated more quickly. That’s a key thing. It’s especially key if you can fluorinate C–H bonds because you can just take a molecule off the shelf, put ¹⁸F in that, and test it.”

Hooker agrees that throughput is key to boosting the number of radiotracers available to scientists. More fluorination reactions with ¹⁸F can only help scientists in their search.

“You’re going to fail a lot,” he says. “Most things you label are going to nonspecifically accumulate, have poor penetration, or fail because of metabolism. Just like in drug discovery, throughput matters. I need to be able to radiolabel lots of molecules very efficiently. But I need to be able to screen them in preclinical imaging efficiently too. So if you can make molecules efficiently and image them efficiently, you’re going to get lucky.” ◾