MIT Research News' Journal

Study offers guidance for targeting residual ovarian tumors

Most women diagnosed with ovarian cancer undergo surgery to remove as many of the tumors as possible. However, it is usually impossible to eliminate all of the cancer cells because they have spread throughout the abdomen. Surgery is therefore followed by 18 weeks of chemotherapy.

Delivering chemotherapy drugs directly to the abdomen through a catheter offers better results than other methods, but this regimen suffers from significant complications, and many patients are unable to complete it.

MIT researchers who are working on an implantable device that could make intraperitoneal chemotherapy more bearable have published a new study that offers insight into how to improve chemotherapy strategies for ovarian cancer, and how to determine which patients would be most likely to benefit from this device.

“As we entered into this project, our question was how do we get the same beneficial outcomes and reduce all the side effects?” says Michael Cima, the David H. Koch Professor of Engineering in the Department of Materials Science and Engineering, a member of MIT’s Koch Institute for Integrative Cancer Research, and the senior author of the study.

The findings suggest that the outcome of initial surgery plays a key role in the effectiveness of subsequent intraperitoneal chemotherapy. Cisplatin, one of the most commonly used drugs, effectively treats very tiny tumor cell clusters when it is delivered continuously or as a single large dose. But the researchers found that for larger tumor cell clusters, continuous delivery of cisplatin, at higher doses than are tolerable with the current periodic chemotherapy method, was more effective. The device they are developing would make delivery of such higher, continuous doses possible.

Laura Tanenbaum, who recently received her PhD from the Harvard-MIT Program in Health Sciences and Technology (HST), and HST PhD student Aikaterini Mantzavinou are the lead authors of the paper, which appears in the journal Gynecologic Oncology. Other authors are HST PhD student Kriti Subramanyam and Massachusetts General Hospital gynecologic oncologist Marcela del Carmen.

Targeting residual tumors

Ovarian cancer is usually not detected until the cancer has reached an advanced stage, with metastases covering organs throughout the peritoneal cavity, including the liver, bladder, and intestines. After surgery, known as “tumor debulking,” patients receive two types of chemotherapy to treat tumors left behind: intravenous delivery of paclitaxel and intravenous or intraperitoneal delivery of a platinum drug such as cisplatin.

Intraperitoneal chemotherapy is pumped directly into the abdomen through a catheter every three weeks for a total of six cycles. This allows cisplatin to come in direct contact with the residual tumors, which has been shown to be more effective than intravenous delivery but is not tolerable for many patients. “It’s painful and the catheter can be a site of local infections,” Cima says.

Several years ago, Cima and colleagues set out to develop an implantable device that could deliver cisplatin into the abdomen without all of the side effects produced by the catheter and the large, repeated cisplatin doses. As this project was neither traditional clinical research nor fundamental scientific research, enabling support came from the Koch Institute’s Frontier Research Program and then later on, the Bridge Project, a partnership between MIT’s Koch Institute and the Dana-Farber/Harvard Cancer Center.

Their device is made of a drug-loaded polymer that could be inserted at the beginning of treatment and remain in place for the full treatment course, then removed with minimally invasive surgery. The researchers have tested proof-of-concept devices in mice and are now developing a version that could be tested in humans, though more animal studies are needed before human trials can begin.

In their new study, the researchers set out to investigate how the size of the residual tumors would affect their response to continuous, low-dose cisplatin delivery. They believed that size would play some role because once tumors reach a certain size, the drug may not be able to penetrate all the way into the inner core of the tumors.

To test this hypothesis, the researchers grew spherical ovarian cancer cell clusters 100 or 200 microns in diameter in a lab dish and exposed them to varying doses of cisplatin. Continuous, low-dose cisplatin delivery, similar to what tumors would receive from an implanted device, was as effective against 100-micron tumor spheroids as a single high dose, similar to that delivered by a catheter.

However, by increasing the continuous cisplatin dose, the researchers found that they could treat the larger 200-micron spheroids more effectively than they could with the single high dose. This increased dose could be delivered using an implantable device, but it would not be tolerable for patients if given through an abdominal catheter.

Better treatment strategies

Cima believes that the findings may also help to explain some of the preliminary results of a recent, large clinical trial in which doctors found that intraperitoneal cisplatin delivery was no more effective than intravenous chemotherapy alone. This contradicted previous findings from smaller studies, indicating that cisplatin delivery by catheter improved patient survival.

In the newer trial, conducted at about 500 treatment centers, surgeons admitted patients to the study based on size estimates of the tumors remaining after surgery. However, Cima says, this subjective evaluation may have resulted in patients entering the study whose tumors were too large to be helped by the current intraperitoneal therapy.

This points to the importance of both developing a good method for screening patients before future trials begin, to make sure they are likely to benefit from the treatment, and devising new strategies to help surgeons remove as many tumors as possible, Cima says.

The research was funded by the Koch Institute-Dana Farber/Harvard Cancer Center Bridge Project, the Koch Institute Frontier Research Program through the Kathy and Curt Marble Cancer Research Fund, the MIT Tata Center for Technology and Design, and the Koch Institute Support (core) Grant from the National Cancer Institute.

Using Bitcoin to prevent identity theft

A reaction to the 2008 financial crisis, Bitcoin is a digital-currency scheme designed to wrest control of the monetary system from central banks. With Bitcoin, anyone can mint money, provided he or she can complete a complex computation quickly enough. Through a set of clever protocols, that computational hurdle prevents the system from being coopted by malicious hackers.

At the IEEE Symposium on Security and Privacy this week, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory are presenting a new system that uses Bitcoin’s security machinery to defend against online identity theft.

“Our paper is about using Bitcoin to prevent online services from getting away with lying,” says Alin Tomescu, a graduate student in electrical engineering and computer science and first author on the paper. “When you build systems that are distributed and send each other digital signatures, for instance, those systems can be compromised, and they can lie. They can say one thing to one person and one thing to another. And we want to prevent that.”

An attacker who hacked a public-key encryption system, for instance, might “certify” — or cryptographically assert the validity of — a false encryption key, to trick users into revealing secret information. But it couldn’t also decertify the true key without setting off alarms, so there would be two keys in circulation bearing certification from the same authority. The new system, which Tomescu developed together with his thesis advisor, Srini Devadas, the Edwin Sibley Webster Professor of Electrical Engineering and Computer Science at MIT, defends against such “equivocation.”

Because Bitcoin is completely decentralized, the only thing ensuring its reliability is a massive public log — referred to as the blockchain — of every Bitcoin transaction conducted since the system was first introduced in 2009. Earlier systems have used the Bitcoin machinery to guard against equivocation, but for verification, they required the download of the entire blockchain, which is 110 gigabytes and growing hourly. Tomescu and Devadas’ system, by contrast, requires the download of only about 40 megabytes of data, so it could run on a smartphone.

Striking paydirt

Extending the blockchain is integral to the process of minting — or in Bitcoin terminology, “mining” — new bitcoins. The mining process is built around a mathematical function, called a one-way hash function, that takes three inputs: the last log entry in the blockchain; a new blockchain entry, in which the miner awards him- or herself a fixed number of new bitcoins (currently 12.5); and an integer. The output of the function is a string of 1s and 0s.

Mining consists of trying to find a value for the input integer that results in an output string with a prescribed number of leading 0s — currently about 72. There’s no way to do this except to try out lots of options, and even with a huge bank of servers churning away in the cloud the process typically takes about 10 minutes. And it’s a race: Adding a new entry — or “block” — to the blockchain invalidates the most recent work of all other miners, who now have to start over using the newly added block as an input.

In addition to assigning the winning miner the latest quota of bitcoins, a new block in the blockchain also records recent transactions by Bitcoin users. Roughly 100,000 commercial vendors in the real world now accept payment in bitcoins. To verify a payment, the payer and vendor simply broadcast a record of their transaction to the Bitcoin network. Miners add the transaction to the blocks they’re working on, and when the transaction shows up in the blockchain, it’s a matter of public record.

The transaction record also has room for an 80-character text annotation. Eighty characters isn’t enough to record, say, all the public keys certified by a public-key cryptography system. But it is enough to record a cryptographic signature verifying that a certification elsewhere on the Internet is legitimate.

Previous schemes for preventing equivocation simply stored such signatures in the annotations of transaction records. Bitcoin’s existing security structure prevents tampering with the signatures.

But verifying that a Web service using those schemes wasn’t equivocating required examining every transaction in every block of the blockchain — or at least, every block added since the service first used the scheme to certify a public assertion. It’s that verification process that Tomescu and Devadas have refined.

Efficient audits

“Our idea is so simple — it’s embarrassingly simple,” Tomescu says. The central requirement of Bitcoin is that no one can spend the same bitcoin in more than one place, and the system has cryptographic protocols in place to prevent that from happening.

So Tomescu and Devadas’s system — called Catena — simply adds the requirement that every Bitcoin transaction that logs a public assertion must involve an actual bitcoin transfer. Users may simply transfer the bitcoin to themselves, but that precludes the possibility of transferring the bitcoin to anyone else in the same block of the blockchain. Consequently, it also precludes equivocation within the block.

To prevent equivocation between blocks, it’s still necessary to confirm that the bitcoin that the Catena user spends in one block is the same one that it spent the last time it made a public assertion. But again, because the ability to verify a bitcoin’s chain of custody is so central to the success of the whole Bitcoin system, this is relatively easy to do. People who want to use Catena to audit all the public assertions of a given Web service still need to download information from every block of the blockchain. But they need to download only a small cryptographic proof — about 600 bytes — for each block, rather than the block’s full megabyte of data.

“The abstraction that the paper lays out is a really good idea — the idea of making it possible to create, you might say, smaller blockchains or linked lists within a blockchain specific to a particular account or a particular object,” says Bryan Ford, an associate professor of computer science at the Swiss Federal Institute of Technology in Lausanne. “It’s very cool, nice, clean, useful primitive, clearly explained. It’s very synergistic with an idea we’ve been working on, which creates an efficiently traversable timeline, which we call a skip chain, meaning a timeline you can skip around on arbitrarily forward and back, where from any point you can verify any other point in the timeline very efficiently.”

“If you can eliminate the possibility of equivocation, it becomes easier to secure many algorithms,” he adds. “It’s a generally important problem.”