After a decade-long trickle of antibody-drug conjugates (ADCs), the US Food and Drug Administration in 2019 approved three new ADCs to treat various cancers. That burst of activity brought the total number of FDA-approved ADCs on the market to seven, a definite uptick to end the decade.
ADCs are antibodies connected to cell-killing, small-molecule payloads via chemical linkers. To fight cancer, the antibodies hook up to proteins on the surfaces of tumor cells. Once the ADCs are there, the tumor cells engulf the ADCs; rip up the linkers via chemical, enzymatic, or biological processes; and release the small-molecule payloads. The payloads then kill the tumor cells. These therapies were conceived to act like guided missiles, delivering cancer-killing drugs exclusively to cancerous cells and sparing healthy cells.
The recent flurry of approvals portends a comeback for ADCs, many of which did not fare well in the clinic in the past decade. Companies have learned from past setbacks and made chemical advances that are enabling a new generation of more finely tuned ADCs. Some deliver more payload molecules on each ADC with the hopes of drilling deeper into solid tumors. Others use more-potent payload molecules. Still others connect the antibody to the linker with greater precision.
And while earlier failures prompted some companies to abandon the therapeutic approach, others still see enormous potential. There are 89 ADCs from dozens of companies currently in human clinical trials, according to a study that Beacon Targeted Therapies compiled for the news agency Reuters earlier this year.
There are a lot of moving parts in an ADC, and there’s no generic formula for success, says Andy Polson, who works on the therapy at Genentech. How the antibody chemically connects to the drug, the number of drugs on each antibody, the stability of the chemical linker, and whether the payload works in a particular tumor type are all important facets to the technology, he explains.
Balancing all those needs when creating an ADC, Polson says, is a relatively large drug development undertaking. Although we typically think of ADCs as guided missiles, he says, that’s only partially true. “What you’re really doing is preferentially delivering it to the tumor and causing greater accumulation in the tumor than you do elsewhere in the body. It’s better than systemic chemotherapy, but still—you’re putting a very potent drug into the body, and that’s going to have side effects.”
David Satijn, vice president of new antibody products at Genmab, which has paired ADCs with its own antibody products, agrees. “We want to kill tumor cells, but we don’t want to harm the nontumor cells, and at the moment, that’s still a challenge,” he says. “So, in a nutshell, we’re looking for better drug linkers and drug payloads.”
Penelope Drake, director of R&D at Catalent Biologics, says a lot of ADCs showed promise in shrinking tumors in mice. “But the problem was that didn’t translate to efficacy in the clinic,” she says. It’s likely that those ADCs didn’t fare well in people because of dose-limiting toxicities. “They just couldn’t give enough ADC to achieve efficacy,” she says.
“I think there was a general sense, maybe a naivete, that you could take an antibody, put a toxic compound on it, and, lo and behold, magic—it was going to be a drug,” says John Lambert, who spent 30 years working on ADC technology at ImmunoGen and is now an independent consultant. But early on, scientists chose antibodies and protein targets without considering the whole ADC. “It’s not as simple as sticking a potent payload on something and it will work,” Lambert says. “Now I think people are learning from the compounds that don’t make it that one has to be thoughtful about the biology end of the ADC as well as the payload end of the ADC.”
Of the three ADCs approved in 2019, Genentech’s Polivy and Astellas Pharma and Seattle Genetics’ Padcev use the payload-linker combo from one of the early ADC successes: Adcetris, from Seattle Genetics and Takeda Pharmaceutical. All three drugs feature a linker that contains the amino acid combination of valine and citrulline, which gets cleaved by cathepsin enzymes in tumor cells. This linker tethers an antibody to a powerful therapeutic called monomethyl auristatin E.
Only Daiichi Sankyo and AstraZeneca’s Enhertu features a novel payload and linker. Enhertu’s payload is the topoisomerase I inhibitor DXd. DXd is a more potent analog of SN-38, the active metabolite of the systemic chemotherapy drug irinotecan. A tetrapeptide-based linker connects DXd to an antibody that targets HER2, a protein sometimes overexpressed by breast cancer and other tumor cells. Each Enhertu ADC carries about eight DXd moieties, whereas most ADCs typically carry two to four payload molecules.
The tetrapeptide-based linker is designed to be broken down by cathepsins and other enzymes that can run amok in tumor cells. “This means that Enhertu acts like a Trojan horse—it moves undetected until it’s transported into the tumor cell, at which point the warhead is released,” explains Mika Sovak, who leads AstraZeneca’s Enhertu effort.
Sovak also believes that the linker stability and the ability to load up the antibody with cytotoxic molecules could have another advantage: once DXd is incorporated into a cancer cell, it can diffuse out of that cell and kill neighboring cancer cells, even if they do not express HER2. This phenomenon is known as the bystander effect. “This is particularly exciting in HER2-expressing tumors, such as breast cancer, where not all cells within a tumor express HER2,” she says.
Scientists at Immunomedics have similarly loaded their ADC IMMU-132 (sacituzumab govitecan) with about eight SN-38 molecules in hopes of killing tumor cells via the bystander effect. After an initial rejection, IMMU-132 was resubmitted to the FDA for approval as a third-line treatment for late-stage, metastatic triple-negative breast cancer late last year.
The SN-38 payload is significantly less toxic than payloads in FDA-approved ADCs, points out Serengulam V. Govindan, Immunomedics’ senior director of chemistry. “That gives us some leeway in how much ADC can be dosed into the patient,” he says. “We can really dose them at a higher level than the current agents.”
Govindan says Immunomedics’ payload-linker combination occupies a niche in the ADC arena. A hydrolysable carbonate linker connects the SN-38 payload at one end. This connection breaks under acidic or basic conditions. A lysine in the linker prevents the ADCs from aggregating, and a hydrophilic polyethylene glycol (PEG) unit boosts the payload-laden ADCs’ solubility.
Getting a linker with the desired robustness was initially a challenge, Govindan says. So the researchers built most of the linker and put a terminal azide on one end. They designed a maleimide portion that contains an acetylene. To assemble the final linker-payload combo, they used click chemistry to stitch the two together.
While Immunomedics is banking on a less potent payload, ADC Therapeutics hopes its highly potent payload will be the key to success. For example, one of its ADCs, ADCT-402 (loncastuximab tesirine), features a pyrrolobenzodiazepine (PBD) dimer payload. PBD dimers were originally developed by Spirogen, which was acquired by AstraZeneca in 2013.
What is special about PBD dimers is that they cross-link the cancer cell DNA in its minor groove, gluing itself between a guanine in one strand of the double helix and a guanine in the other strand, explains ADC Therapeutics CEO Chris Martin. “They bind in a way that does not distort the DNA helix. This is particularly important because by not distorting the helix and not protruding from the helix, they avoid detection by the machinery that leads to excision and mismatch repair.”
Once the PBD dimer forms a cross-link in the minor groove of a cancer cell’s DNA, Martin says, it can sit there undetected for weeks until the cell divides. When the cell does divide, replication stalls, and the cell dies. The dimer’s high potency means that each antibody needs to carry only two payload molecules. ADCT-402 is in Phase II clinical trials to treat relapsed or refractory diffuse large B-cell lymphoma.
While the payload is an ADC’s weapon for killing cancer, chemical design of the linker can help wield that weapon more effectively. Some of these covalent tethers are designed to be cleaved by specific chemical conditions or enzymes in cancer cells. Others are built to withstand breakdown until enzymes in the tumor cells’ lysosome rip them apart and release the payload. Linkers also give ADC makers the opportunity to add groups that balance the hydrophobicity common to many payload molecules. This hydrophobicity can make ADCs clump together in a way that makes the body’s immune system recognize them as foreign and clears them before they can reach cancer cells.
More than a decade ago, ADC makers realized that linkers could be unstable in plasma, particularly at the spot where the antibody connected to the linker, says Robert Lyon, senior director of protein sciences at Seattle Genetics. Researchers commonly hooked together these two units by using a Michael addition to attach a thiol from a cysteine residue on the antibody to a maleimide ring on the linker.
“We now understand that the resulting bond is susceptible to a slow elimination,” Lyon says. What would happen is that the C–S bond attaching the antibody to the linker would break, regenerating the maleimide ring that was the linker’s reactive handle. When this happens in the bloodstream, the linker-payload combo can subsequently react with cysteine residues on other proteins or biological thiols like the antioxidant glutathione.
But Lyon says that he and his colleagues realized that a different chemical reaction was also taking place at this attachment point on the ADC: a hydrolysis reaction can spring open the succinimide ring that results from the thiol-maleimide reaction. “When this occurs, the linker can no longer undergo the elimination pathway and is essentially completely stable—so ring hydrolysis leads to stability.”
With this in mind, Seattle Genetics’ new generation of linkers incorporate a primary amine adjacent to the maleimide, which promotes a much faster hydrolysis reaction. “Using this linker, we can prepare ADCs in which the succinimide is completely hydrolyzed during the conjugation process so that the bond is completely stable when the ADC is dosed,” Lyon says.
Another feature to Seattle Genetics’ new generation of linkers is that they replace the relatively hydrophobic valine-citrulline peptide portion of their linker, which is where the linker is cleaved when it enters the cancer cell, with a hydrophilic glucuronic acid moiety. Lysosomal compartments of cells have β-glucuronidase enzymes that can remove the glucuronic acid and release the small-molecule payload.
Lyon says that while preclinical pharmacokinetic studies show that these changes improve the ADC’s properties, they also needed to add a PEG unit to offset the hydrophobic payload. “By changing the length of this PEG between 0 and 12 units, we could tune the degree of nonspecific ADC uptake and thus tune its pharmacokinetics and off-target toxicity,” Lyon says.
Lyon adds that linker stability can also be improved by changing the site on the antibody that connects to the linker. “By engineering cysteine residues at different positions, the reaction kinetics for both the maleimide elimination and succinimide hydrolysis pathways can vary considerably,” he says.
Tweaking the sites where the linker and the payload attach to the antibody has been an area where many ADC makers are working. Except for Enhertu, all the FDA-approved ADCs are heterogeneous: the number of linker-payload groups attached to each antibody varies. Creating homogeneous ADCs, in which each antibody connects to the same number of payload-linker groups, has been an area of intense research.
ADC makers initially turned to lysine or cysteine groups on the antibody to attach the linker and payload. But they are increasingly engineering specific sites onto antibodies to make those connections. These site-specific, homogeneous ADCs have the potential to perform better than heterogeneous ADCs, says Catalent Biologics’ Drake. That’s because the homogeneous ADC is just one molecule rather than a group of molecules, so the biophysical properties are more uniform.
Catalent developed SMARTag technology to create site-specific homogeneous ADCs. SMARTag was originally developed at Redwood Bioscience, which was acquired by Catalent Pharma Solutions in 2014. The method creates sites for attaching linker-payload groups to antibodies by encoding a string of six amino acids into the antibody. As the antibody is made, a formylglycine-generating enzyme recognizes that string and converts its cysteine into an aldehyde-containing formylglycine.
“That’s the handle for our chemistry,” Drake says. By using that aldehyde, Drake explains, scientists can forge a C–C bond between the antibody and the linker-payload group. Triphase Accelerator recently licensed an ADC made using this technology, called TRPH-222. It’s in Phase 1 clinical trials for non-Hodgkin’s lymphoma.
As data for that drug and others in the pipeline emerge, the field will start to understand whether these chemical advances can make ADCs more effective.
“ADCs do seem to be making a comeback,” Drake says. “It’s a good modality, but it’s a complicated one, so there’s been a lot of learning. I think eventually we’ll be seeing more and more approvals. Newer technologies are going to be helpful to overcome some of the problems that led to clinical failures.”