IT’S NOT much to look at. Just a tiny vial filled with clear fluid. But if
its inventor is right, that fluid represents a new era in cancer treatment. It
is a personalised cancer vaccine.
Pramod Srivastava, a professor of medicine at the University of Connecticut,
believes that the body’s own immune system may ultimately prove to be the best
tool for fighting cancer. The mix of proteins within that vial is unique, made
in a matter of hours from the patient’s own tumour, and they are designed to
trigger an immune response which will destroy it. Already, Srivastava has cured
rats and mice of tumours, and early tests have prolonged the lives of people
with kidney cancer and melanoma. If the results are confirmed it will be the
culmination of a 20-year struggle during which Srivastava switched careers and
risked his reputation on a hypothesis almost everyone thought must be wrong.
As a graduate student at the Center for Cell and Molecular Biology in
Hyderabad in 1979, Srivastava found himself with time on his hands while he
waited for equipment for his project on yeast genetics to arrive. So he decided
to have a go at the unrelated field of cancer vaccines. Researchers had long
been searching for a marker: a surface molecule common to all tumour cells, but
absent from healthy ones. They would then be able to design a vaccine to prime
the immune system to attack tumours bearing this telltale antigen. Everyone knew
it had to be there, because an injection of dead tumour cells had been shown to
immunise mice against cancer. But injecting tumour cells into animals or people
failed to boost their immune systems enough to fight an established cancer. So
the hunt was on for the antigen itself, in the hope that it might generate a
stronger response.
Advertisement
By 1979, no one had found it, and Srivastava felt sure that everyone was
missing something vital. “I thought there must be tumour-specific antigens,” he
recalls. “There obviously are, or else you couldn’t immunise.” So he began
breaking open tumour cells taken from rats and separating out their contents by
column chromatography, which fractionates proteins and other chemicals according
to size or electric charge. Then he tested each fraction for its potency as a
cancer vaccine in lab rats. “I looked for a part that had the same activity as
the whole,” he says.
He soon got a result. “I first found one protein, and then four or five
others.” The search was going so well that he changed his thesis to focus on it.
But when he finally worked out what the proteins were, everything fell apart.
Instead of finding the elusive tumour-specific antigens, he had purified the
most run-of-the-mill proteins in all of biology. “It turned out to my horror
that they were all heat shock proteins.”
It was a huge disappointment. There is nothing tumour-specific about HSPs.
They are found in virtually every cell in the body, where they help amino acid
chains fold in the correct way to become working proteins. The unfolded chains
tend to stick to themselves, and HSPs stop them from clumping up like wads of
sticky tape before they have a chance to fold properly. They get their name from
a second function they have, which is to rescue proteins damaged by stresses
such as infection or extreme heat.
Faced with such an unpromising result, Srivastava gave up and eventually took
a faculty position in India as a yeast geneticist. But he couldn’t quite let go
of his graduate work. “Something about it kept tugging at me,” he says.
Something in the purified fractions had immunised his rats against tumours, and
if it wasn’t HSPs, what was it? In 1983, curiosity got the better of him and he
moved to the US to pick up the trail.
Having ruled out the possibility that the HSPs from tumour cells were
different in some way that alerted the immune system to the cancer, Srivastava
began to check whether there was anything else in the protein fractions. Perhaps
they weren’t pure after all. But standard techniques such as gel
electrophoresis, which separates proteins according to size, revealed no
impurities. By 1990, he still hadn’t found any contaminants.
“I thought if there was something I cannot see on the gels, it must be very
small,” Srivastava says. So he guessed there might be peptides—small
protein fragments—bound to the HSPs. Such peptides are a vital part of the
immune system’s ability to sniff out invaders. There are enzymes inside the cell
which routinely chop up proteins they encounter and display the peptides on the
cell surface. Passing lymphocytes sample these peptides and sound the alarm if
they detect something foreign—a piece of viral protein, for example.
Srivastava soon realised that this process could be the answer to why the
vaccines worked. He guessed that the HSPs in the vaccines used on the rats had
peptides from the tumour cells bound to them—peptides unique to the
tumour. If the immune system’s T cells sampled these tumour peptides and learned
to recognise them, he reasoned, they would be able to attack any cells that
display them on their surface—that is, the cancer itself.
It was quite a leap. People knew that HSPs associated with whole proteins,
but no one imagined they could bind to fragments and carry them around. When
Srivastava published his proposal in 1991, many of his colleagues viewed it as
wild speculation. “Everybody said I had no evidence for it,” he recalls.
It took him another two years to gather solid experimental evidence of
peptides associated with the HSPs. He found he was able to strip the peptides
off the HSPs with acid, but by themselves, the peptides were completely
ineffective as a vaccine. HSP on its own was also useless at immunising mice
against cancer. Only when the tumour peptides and the HSP were combined did the
vaccine work. But just then, in 1993, before Srivastava was able to publish his
findings, his grant from the National Institutes of Health ran out.
“By this time I knew what was going on,” he says. “Everything seemed within
reach, but I had no money.” The NIH renewed the grant later that year, but by
then Srivastava had already found some venture capital and started a company
called Antigenics, based in New York, with the specific aim of getting an HSP
cancer vaccine into general use. The company soon established a procedure for
producing a vaccine tailored to a given patient’s cancer. First patients have a
sugar-lump-sized piece of their tumour sent to a lab in Woburn, Massachusetts.
There, technicians pulverise the tissue and purify the HSPs with column
chromatography, in much the same way that Srivastava initially did. The vaccine
is ready in just 10 hours, although it must then be monitored for microbial
contamination for two weeks, which gives patients time to recover from the
biopsy surgery.
The beauty of this approach is that nothing needs to be known about the
tumour’s chemical profile: HSPs do all the work. Every cancer cell must contain
mutated proteins—the cell would be normal otherwise—but it doesn’t
matter if the set of proteins affected varies from one tumour to another. HSPs
pick up peptides from all the proteins in the cell, both variant and normal, and
deliver them to the immune system, which is already trained to ignore the normal
ones and focus on the unusual. So there’s no need to isolate and identify the
mutated proteins, and the vaccine should work even if there is no such thing as
the elusive tumour-specific antigen that Srivastava was looking for back in
1979.
In principle it would be better to work with just a couple of well-chosen
antigens, admits Antigenics’ CEO Garo Armen, but he says that such vaccines have
serious limitations. For example, the only known antigens which will alert the
immune system to all prostate cancers are tissue-specific rather than
tumour-specific. Training the immune system to respond to those antigens means
turning it loose on the entire tissue, not just the tumour. That might be
acceptable for prostate cancer, in which the entire prostate is usually removed
anyway, but what about skin cancer or breast cancer? In contrast, there should
be no type of cancer for which HSP vaccines will not work, says Armen.
But for Antigenics to get off the ground, the company’s scientists had to
know exactly how the HSPs were alerting the immune system to the tumour
peptides. All they had was an empirical result. Without understanding the
precise biological mechanism, the company wouldn’t be able to refine and develop
its vaccine.
By the mid-1990s, they had determined that the HSPs were stimulating two
types of response from the immune system, one broad and the other highly
specific. Other researchers had already demonstrated that bacterial HSPs
activate “natural killer” cells. These cells don’t target specific antigens, but
somehow detect and attack virus-infected cells and some cancers too. In 1997,
Srivastava found that destroying the natural killer cells in mice made them less
responsive to his vaccine. He concluded that this non-specific response was
crucial to the vaccine’s function.
Details of the more specific response emerged last August, when Srivastava
and his colleagues announced the discovery of an HSP receptor on dendritic
cells, which are a type of T-cell activator (Nature Immunology, vol 1,
p 151). Dendritic cells soak up the HSPs and transfer the peptides bound to them
to the cell surface, they found. Once on the surface, the peptides prime T cells
to kill any cell bearing those particular foreign peptides
(see Diagram).
Actually, the immune system weeds out cancerous cells continually throughout
our lives, as tumour cells bearing mutated peptides activate T cells directly
and are then annihilated. The cancers we see are the ones that evade this
quality control. They do so by secreting substances called cytokines, which
suppress the activation of T cells. Many cancer cells also delete their genes
for a protein called a costimulator. T cells require this protein in order to
become active once they have found a rogue peptide. Such tumours effectively
make themselves invisible to T cells.
Antigenics’ vaccine activates T cells against these cloaked tumours by
delivering the crucial antigens to the dendritic cells, which produce plenty of
costimulators and are out of the reach of the cytokines secreted locally by the
tumour. “We have taken peptide antigens from an immune-unfriendly environment
and put them into an immune-friendly environment,” says Armen. The activated T
cells can then start scouting the rest of the body and attacking the cancer
cells.
So far, it looks as though the approach works. Srivastava’s vaccine has
worked extremely well in mice and rats, and animals immunised with HSPs taken
from cultured sarcoma cells reject a graft of the same cells a few weeks later.
Since 1994, more than a dozen other labs have independently confirmed these
results.
Antigenics completed preliminary clinical trials of its vaccine against
kidney cancer and melanoma last year. Eight weekly injections of the vaccine
appeared to improve survival, and had no side effects. Of 42 patients with
late-stage kidney cancer, nearly half were still alive two years after receiving
a vaccine made from their own tumours. Normally only 15 per cent of such
patients would be expected to survive that long. In a similar trial on patients
with melanoma, 95 per cent were still alive 14 months after being vaccinated,
compared with an expected survival rate of only 40 or 50 per cent. Larger, more
carefully controlled trials on kidney cancer are now under way, with hundreds of
patients enrolled, and tests on melanoma are likely to begin soon. The results
should be known in about two years, Armen says.
Cancer immunologists are optimistic about the work. “I’m excited about that
approach,” says Richard Young, an HSP expert at the Whitehead Institute in
Cambridge, Massachusetts. “It has the advantage that each individual can have
their own tumours dissected so you treat with the most relevant antigens,” he
says.
Craig Slingluff, a cancer immunologist at the University of Virginia, agrees:
“The heat shock proteins are a very reasonable approach.” But, he points out,
Antigenics’ technique is like giving a crude plant extract instead of a purified
drug. To refine the therapy, Antigenics would need to know what the active
ingredients are—which it doesn’t.
Others raise the issue of regulatory approval. The US Food and Drug
Administration prefers precise atomic descriptions of new drugs and may baulk at
a cryptic mix of HSPs and peptides. But Armen anticipates no problem. “The FDA
has two basic rules, safety and efficacy. If you can deliver a product that is
safe and effective, you are home free,” he says.
Most experts agree that an effective cancer vaccine will probably be on the
market within ten years, and more than one approach is likely to succeed (see
“Closing in”). But Srivastava has even bigger plans. He’s discovered that while
small doses of HSPs prime the immune system to attack tumours, much larger doses
have the opposite effect—they persuade the immune system to ignore the
unusual antigens. Srivastava isn’t talking about the details yet, but he does
point out that the effect could be used to treat diseases such as rheumatoid
arthritis and diabetes, which occur when the immune system attacks the body’s
healthy cells. He’s already cured mouse versions of diabetes and multiple
sclerosis, and hopes to start human trials in a year or two.
The race for the first successful cancer vaccine has dozens of entrants,
nearly all focusing on one crucial component of the immune
system—dendritic cells (DCs). These cells show fragments of rogue proteins
to T cells, which then seek out and kill any cells with the same marker.
• Dendreon Corporation of Seattle has a prostate-cancer treatment in the
final phase of clinical trials. Immature DCs from the patient are cultured with
a protein found only in the prostate. When reintroduced into the patient, the
dendritic cells direct T cells to attack all prostate cells, healthy or
cancerous. This is no more drastic than the customary surgery to remove
cancerous prostates, and has the advantage of killing any cancer cells that may
have spread.
• Therion Biologics Corporation in Cambridge, Massachusetts, puts genes for
antigens associated with prostate cancer, breast cancer or melanoma into
viruses. When the viruses are injected into a patient, they infect dendritic
cells and produce a strong immune response against the tumour cells, the company
reports.
• StressGen, in Victoria, British Columbia, hooks bacterial heat shock
proteins to a protein from human papillomavirus, the virus responsible for
cervical and anal cancer. The bacterial proteins are so similar to their human
equivalents that they still guide antigens into the dendritic cells. Some
patients with precancerous lesions showed significant improvement in early
clinical trials last year.
• Genzyme Molecular Oncology, with its headquarters near Boston, fuses tumour
cells with dendritic cells and introduces the hybrid cells back into the
patient. The fused cells act just like dendritic cells and display the complete
antigenic profile of the tumour.