When I put out a call for promising young scientists to be part of Forbes’ annual 30 Under 30 project, no name came up more than that of Jack Andraka, the 16-year-old who created a prototype cancer diagnostic test and won the top $75,000 prize at Intel’s annual high school science fair.
And no wonder. For two years, Andraka has been everywhere. The Intel win resulted in a barrage of media appearances on 60 Minutes, which said the test “might save countless lives,” ABC World News Tonight, and Forbes. He was honored as a “Champion of Change” at the White House and was Michelle Obama’s guest during the 2013 State of the Union Address. He has given many TEDx talks, the most prominent of which has generated more than 1 million views. These presentations have consistently portrayed Andraka’s achievements as Earth-shaking. A Forbes article, for instance, described the test as a “breakthrough.” Smithsonian called him “The Teen Prodigy Of Pancreatic Cancer.”
But I decided not to include Andraka on the list, overriding the recommendation of an expert judging panel, because the work was not yet published in a peer-reviewed scientific journal. It is by published work that scientists are judged. I still think this was the right decision. In fact, when Andraka volunteered to share a draft of a paper that he does plan to submit to a scientific journal, my concerns deepened.
George Church of Harvard University, one of the expert judges for the 30 Under 30 project and one of the fathers of next-generation DNA sequencing, reviewed the paper and thought that many of the key claims that had been made of Andraka’s sensor – for instance, that it was 168 times faster, 26,667 times less expensive, and 400 times more sensitive than existing technologies – may not hold up, or at least require more work to be proven. I sent the paper to five more top scientists, who reviewed it, for the most part, anonymously. Most saw holes and said the results did not match the glowing accounts reported, well everywhere.
The consensus: Andraka’s sensor is a probably a publishable piece of science that could eventually appear in a journal, and was a remarkable achievement for a high school student. But it falls far short of changing science and is only a small step toward developing a workable cancer diagnostic.
“While promising the project really was extremely preliminary and is by no means as sophisticated as some stories say it is,” Andraka writes me via email. “I realize that in retrospect that it was just a high school science fair project and it was a proof of concept experiment and initially I thought that it could get on the market in 1-2 years however I’ve learned so much over the course of this journey and realized that it will in reality take a lot longer than this.” A lot longer.
Underestimating Current Technology
Here is what Andraka did, according to his draft paper: he created a device by dipping filter paper into a solution of carbon nanotubes coated with antibodies for a specific protein that can be elevated in patients with pancreatic cancer. He proved that the device can detect the presence of that protein, mesothelin, in cell culture solutions. He also showed that the sensor could tell the difference between the blood of mice that had human pancreatic tumors grafted onto them and those that did not.
So far, so good. But what about all those claims that this new technique is faster, better, and cheaper than the existing technology, the enzyme-linked immunosorbent assay (ELISA)? This 60-year-old technology uses antibodies linked to color-changing enzymes. If the substance you’re looking for is present, the color shows up when you add a third chemical.
Andraka’s “168 times faster, 26,667 times less expensive, and 400 times more sensitive” figures are based on a comparison with ELISA. But Church saw problems with the way that Andraka characterizes the ELISA test.
Test speed: Andraka says he compared the speed of his test to the amount of time he spent trying to get results from an ELISA kit he ordered online: 14 hours. But usually a modern ELISA test takes 1 hour.
Test cost: Andraka is comparing the commercial cost for a test – including the manufacturer’s profit and overhead – to his own materials cost. That’s not a fair comparison. He says the only mesothelin test that he found cost $912 per kit. But other ELISA tests are for sale online for $400 per 60 tests or $600 for 96 tests – in other words, about $6.50 per test run. That still compares favorably to Andraka’s $3 per 10 tests, but remember that there would also be a commercial markup if a company decided to sell his tests.
Sensitivity: Church felt that the standard deviations on Andraka’s sensitivity measurements were too big to say that his test is much more sensitive than ELISA. Others said that it’s not clear how much increased sensitivity a new test would need.
Is it the right protein? Another question: Is Andraka even testing for a protein that will effectively identify pancreatic cancer? Doctors use a blood protein called CA19-9 to track whether a pancreatic tumor is growing, or whether it will come back. But they don’t use CA19-9 for screening, because it would miss too many cancers and, more importantly, identify pancreatic cancer in patients who don’t have it. That could lead to unnecessary CAT scans, or even to unnecessary surgeries to remove tumors that don’t exist.
Andraka asserts that mesothelin would be a better protein to test for. But one of my anonymous reviewers pointed to a paper published in HPB, the journal of the International Hepato-Pancreato-Biliary Association. It compared mesothelin to CA19-9 as a marker for pancreatic cancer, and found mesothelin was less useful. The paper expressed hope that lots of proteins, used together, would enable the development of a prostate cancer screening test. It’s not certain that Andraka’s test, no matter how sensitive it is, could become widely used just by detecting mesothelin. (His mentor at Johns Hopkins has suggested that other proteins would need to be added in news articles.) These are all questions Andraka would have had to answer in the peer-review process, perhaps leading him to modify his claims.
The Difficulties For Diagnostic Tests
In general, cancer doctors are becoming skeptical about screening tests. There has been controversy about using PSA screening to detect prostate cancer, and mammography to detect breast cancer in young woman. The reason is that the harms of screening – the extra CAT scans and incumbent radiation, the extra biopsies, even the removal of tumors that would never prove deadly – can in aggregate be worse than the cancer being treated.
This is a bigger issue in pancreatic cancer, because pancreatic tumors are hard to detect and the pancreas is nestled deep in the abdomen, beneath other, fragile organs.
Susan Desmond-Hellmann, the chancellor at University of California, San Francisco, reviewed the paper and wrote:
The larger issue today is the increasing understanding of the limitations of ‘early detection’. For a pancreatic cancer screening test to have clinical utility one would need to believe that the cancers detected would be early stage and resectable and it would be rare to operate on someone and not find a cancer (pancreatic surgery is very difficult, not as straightforward as breast or prostate surgery).
So the manuscript is compelling, but the road to an approved screening test is likely to be long, difficult and expensive.
Andraka’s draft paper does not include tests of the sensor in human blood samples. There have been no tests of it in human beings. Andraka’s test can tell a mouse with a human tumor growing on it from one without. But can it tell the blood of someone with an inflamed pancreas or gallstones from another person with cancer? After all that work is done, there will still be a need for prospective clinical trials, the ultimate test in which people are randomized to get the test – or not. At the end of the day, does getting the test make you healthier? There’s a long path between building a tool to detect a protein in the blood — which is what Andraka started to do — and creating a diagnostic test. We’re doing him no favors by pretending otherwise.
Reality Check: Medicine Is Super-Hard
The two Forbes 30 Under 30 judges who voted that we should highlight Andraka were Mikael Dolsten, the head of R&D at Pfizer, and Daniel Kraft, the executive director at FutureMed. Both told me that they thought his example was encouraging other young people. I hope so, but I don’t think it is right to honor Andraka for that reason. People who get involved in medical research need to know how hard it is — that not only are there thousands of failures, but that the successes seem to turn into failures too. And that’s why I want to see published work for scientists on the 30 Under 30.
More than that, I think it might have been better for Andraka, and for his test, if he had gone through the traditional process of publishing and presenting his work to other, hypercritical but friendly scientists before he became a media star. A scientific paper should have come before the TED Talks, and before the adulation. He says he’s looking for a big diagnostics company to license and develop the test; I’m sure the companies he’s talking to would feel much more comfortable if they were looking at published research. Instead, it seems to me that one writer after another has just taken the claims he made while trying to win a high school science fair and inflated them to fill a global stage. In Andraka’s own words, it has been a “media circus.” Andraka told me: “It was my very first time in the laboratory and I was just so excited to be able to do serious scientific research and talk about my research in the media.”
By taking a teenager’s excitement and using it to turn him into a folk hero, TED and many, many media organizations including my own have provided false hope to cancer patients and given the general population a distorted view of how medical science works. People might like to believe that cancer researchers have missed something obvious, and that we just need a high school student to come in and create a breakthrough. We need that super-smart, ambitious high school student working on cancer, but this is a disease that’s literally written into our DNA. (60 Minutes and Smithsonian, at least, did better, cautioning that it would be many years and take many clinical trials before the test could be used.)
In one of the first stories about Andraka, Forbes Managing Editor Bruce Upbin asked: “Wait, Did This 15-Year-Old From Maryland Just Change Cancer Treatment?” Nineteen months later, I feel safe answering: No, he didn’t. And I think it’s unfair to him, and to the work he did do, that we expected him to. Because what he did — creating a cool biosensor while still in high school — was pretty neat on its own.