MOUNTAIN VIEW, Calif. — Google (GOOGL) employees, squeezed onto metal risers and standing in the back of a meeting room, erupted in cheers as newly arrived executive Andrew Conrad announced they would try to turn science fiction into reality: The tech giant had formed a biotech venture to create a futuristic device like Star Trek’s iconic “Tricorder” diagnostic wizard — and use it to cure cancer.
Conrad, recalled an employee who was present, displayed images on the room’s big screens showing nanoparticles tracking down cancer cells in the bloodstream and flashing signals to a Fitbit-style wristband. He promised a working prototype of the cancer early-detection device within six months.
That was three years ago. Recently departed employees said the prototype didn’t work as hoped, and the Tricorder project is floundering.
Tricorder is not the only misfire for Google’s ambitious and extravagantly funded biotech venture, now named Verily Life Sciences. It has announced three signature projects meant to transform medicine, and a STAT examination found that all of them are plagued by serious, if not fatal, scientific shortcomings, even as Verily has vigorously promoted their promise.
The Tricorder, as Conrad and others at Verily call the device, is “in the realm of not only science fiction, but beyond that — science fantasy,” said David Walt, a Tufts University chemistry professor and nanoscience expert who met with Verily scientists and engineers last year to share his concerns. “And I’m not sure it will ever be science reality.”
The company has also touted a glucose-sensing contact lens as a substitute for frequent blood tests on diabetics, but independent experts said it is scientifically dubious at best.
It claims a billion-dollar “Baseline” study of human health will define what it means to be healthy and help identify early signs of disease. But researchers said design weaknesses make these lofty goals far-fetched.
Largely through Verily, Google has positioned itself to be a giant in life sciences by marrying technology and big data with science to cure diseases that have, so far, defied the best minds. But its setbacks and prominent scientists’ skepticism call into question this vision of the future of medicine.
Verily insists it is forging ahead with the projects, though Conrad, the company’s CEO, and other executives are well aware of outside scientists’ criticism.
Conrad and other Verily leaders declined interview requests. But in a written response to questions from STAT, the company strongly defended its record, saying its projects were selected specifically because “they are inherently difficult. We, together with our partners, believe that we have technology, expertise, and insights that might make success attainable on very challenging projects.”
Still, Verily acknowledged, “As with all true innovation, some projects can and will fail.”
STAT previously documented a significant departure of top talent from Verily under its divisive leader, Conrad, and ethical and conflict-of-interest concerns regarding Baseline. In more recent interviews, some former employees said they had voiced doubts about projects Verily was pursuing and had been frozen out of decision-making, while their concerns were brushed aside.
Verily appears to be having more success with less world-changing projects, but it still chooses to showcase its most ambitious ones — perhaps, some critics suggest, to promote itself as a company poised to defeat disease.
“One needs to balance how much these toys are used mostly for marketing and for giving a sense of a company really working on something impressive — the brave new world — or if we’re talking about something that will have clear and immediate clinical impact,” said Dr. John Ioannidis, a professor of disease prevention at Stanford University. “The latter is very hard to imagine.”
It’s axiomatic in Silicon Valley’s tech companies that if the math and the coding can be done, the product can be made. But seven former Verily employees said the company’s leadership often seems not to grasp the reality that biology can be more complex and less predictable than computers.
They said Conrad, who has a PhD in anatomy and cell biology, applies the confident impatience of computer engineering, along with extravagant hype, to biotech ideas that demand rigorous peer review and years or decades of painstaking work.
Verily said in response to this criticism that it has hired “many seasoned and respected industry, academic, public health, and regulatory veterans who understand the complexity of biology and how long it takes to move from idea to device and/or therapy.”
But Chad Mirkin, a Northwestern University biosensor and nanotechnology expert who has reviewed public statements and patents on the Tricorder, questioned whether the company had really internalized the fact that in the life sciences, a concept without a well-vetted technological pathway or rationale rarely succeeds.
“That’s a type of Silicon Valley arrogance,” he said. “That isn’t how science works.”
STAR TREK vs. REALITY
Google’s leap into the life sciences was just three months old when Conrad announced the Tricorder to his new colleagues. The company hadn’t yet developed expertise to critically assess the project’s scientific merits, but he revealed no doubts.
It was pure Conrad, former Verily managers said: Boast now, build later.
The biotech venture was housed in Google X — the company’s incubator for radical projects to solve big problems — which had rolled out the self-driving car and Google Glass eyeglass computer. So its engineers understandably reasoned, “Why not a Tricorder?” one of the former managers said.
He and other former or current Google or Verily employees and contractors spoke to STAT on the condition they wouldn’t be identified because they signed nondisclosure agreements and wanted to protect ongoing relationships with the companies.
The self-driving car and other Google X projects were created in secret and vetted by experts for years before anyone outside the tight circle of inventors got so much as a peek. But with his characteristic informal charm, Conrad rolled out the Tricorder to tech reporters in 2014, describing its scientific basis as proven. He predicted that high-risk patients would begin wearing the device within a few years, followed by widespread adoption.
Conrad said a patient would swallow a pill full of magnetic nanoparticles engineered to grab on to tumor cells floating in the bloodstream and light up when they do. A wristband magnet would concentrate the particles and their captured cells inside adjacent veins, then periodically read their fluorescent signals. Conrad said most of the system operated seamlessly in the lab.
“We’ve done a lot, to be quite humble about it. Enough to give us great confidence that this is all likely to work,” he told Backchannel, an online journal.
The particles were so safe, he said, that animal testing could be skipped. Conrad provided no details beyond conceptual patents, and Verily scientists have published no scholarly papers on the device. Still, media coverage portrayed Google’s upstart biotech venture as the vanguard of a medical revolution.
Walt, the Tufts nanoscience expert, was skeptical when he saw the media coverage of the Tricorder, and offered to share his expertise with a friend, Jeff Huber, who was then a Google X executive. Huber, now head of Grail, a startup also focused on detecting early signs of cancer, invited Walt to visit Verily last August.
The Verily team listened carefully to Walt’s concerns during two hours of discussions, he said in a recent interview, and he came away impressed with the company’s scientists and engineers — but not with the Tricorder.
“What (Verily is) really good at is physical measurements — things like temperature, pulse rate, activity level,” Walt said. “They are not particularly good at … the chemical and the biological stuff.”
He ticked off challenges confronting the Tricorder team.
Accurately detecting incredibly rare tumor cells via remote sensors would require “a real transformation in our capabilities from where we are today,” he said. Getting nanoparticles to circulate for long periods would be tough, because the body directs them from the bloodstream into the liver or other organs.
Cancer cells also can absorb nanoparticles, obliterating their detection powers. Magnetic tracking might cause clots by altering blood flow, Walt said, and the Tricorder would require years of animal and human safety testing, contrary to Conrad’s assertions.
Walt also questioned whether the continuous monitoring Conrad proposes offers advantages over safer, promising alternatives, such as testing blood samples for early signs of cancer. A number of companies are racing to develop this liquid biopsy technology, though it, too, faces challenges. (Walt is scientific founder of Quanterix, a Massachusetts firm that builds related diagnostic tools.)
Stanford’s Ioannidis raised a different concern. “Screening asymptomatic people (for cancer) has met with so many failures and so many problems — including overdiagnosis,” he said. For example, a few stray cancer cells in the blood are thought to be common in people, but might pose no significant risk of future disease. Inevitable errors in test sensitivity — false-positive results — could suggest cancer when none is present, leading some patients to needlessly undergo treatments that can cause harmful side effects.
In its written statement, Verily sounded far more cautious about the Tricorder’s prospects than Conrad has been, saying the “very early-stage” project is “ambitious and difficult,” with “unsolved technical challenges. It’s our aspiration, with our partners, to solve these challenges, even if it takes years.”
In recent months, four former Verily employees said, the Tricorder has been seen internally more as a way to generate buzz than as a viable project.
Google cofounder Sergey Brin has said the glucose-sensing contact lens is the idea that inspired Verily’s formation, and the company features the product on its website as an example of small devices it’s building that “fit more easily into daily life.”
But a former Verily manager recently called the lens “slideware” — a Silicon Valley term for breakthroughs that exist only on PowerPoint images.
That’s not how the company framed the technology in a January 2014 announcement about the lens. “(W)e’ve completed multiple clinical research studies, which are helping to refine our prototype,” Chief Technical Officer Brian Otis said in a news release.
The claim suggested that the project was on the road to solving one of the most vexing problems faced by millions of diabetics: The frequent need to draw blood, then measure its glucose level, to calibrate diet and insulin intake. It looked like Google’s life sciences startup might reinvent the $10 billion diabetes monitoring market.
The company indeed produced a prototype, but it didn’t work, the former manager told STAT.
John Smith, a chemist and former chief scientific officer of the LifeScan division of Johnson & Johnson (JNJ), said the 2014 announcement created enormous buzz, but he greeted the news skeptically. He knew that University of Maryland researchers had announced glucose-sensing contacts in 2003, and that in 2009 other researchers applied for a patent on a lens similar to Verily’s, using integrated biosensors. Neither reached the market.
Smith has evaluated more than 30 “noninvasive” technologies that measure glucose from sweat, saliva, and tears. “I saw people working on this, and time after time after time, failing in the same ways or in entirely new ones,” he said in an interview. They all faced a problem no technical advance can overcome, Smith said. None of those fluids offers glucose readings that reflect the levels of glucose in blood.
Glucose levels in tears fluctuate with ambient temperature, humidity, and other factors, he said, calling its correlation with blood glucose readings “very, very poor.” A paper in the British Journal of Ophthalmology authoritatively demonstrated the problem in 1980.
Former employees said visiting scientists have warned Verily about the apparent fatal flaw with using tears to monitor glucose. But in response to STAT questions, Verily wrote that “Tear/blood glucose correlation is an open scientific/biological question, with conflicting data,” and that it is trying to improve on flawed tear-collection methods that might have skewed earlier findings.
“(T)here is always a new adherent who thinks all the measurements that went before were wrong,” Smith said in an email. “This kind of ‘faith-based science’ has proven to be very expensive, and should not come from companies like Verily; but then, cost does not seem to be an issue there.”
In July 2014, Conrad partnered with Alcon (ALC), a division of Novartis (NVS), to commercialize the lens. “We discussed the challenges in detail with Novartis/Alcon before launching our collaboration,” Verily’s statement noted.
Alcon spokesperson Kara Peterson said in an email that the glucose-sensing lens “continues to make steady progress and remains in the research phase.” She declined to comment further.
‘SPECK OF DUST’
The Baseline clinical study, slated to begin enrolling subjects this year, has the audacious goal of defining what it means to be healthy and finding very early signs of cancer and heart disease.
Baseline will help create “a new way of thinking about human systems biology,” Dr. Jessica Mega, who heads the project, said in an April interview. Ideally, she said, it will discover new markers of disease that can inform clinical studies to develop cures even before symptoms arise.
Over five years, it will collect a raft of information — including psychosocial, molecular, imaging, genetic, and microbiome data — from 10,000 people, some healthy, some with a history of heart disease or cancer. That seems like a big number, but experts on designing such studies said the project isn’t big enough to meet its goals.
Former Verily employees and contractors, and independent experts, pointed to a number of flaws. Among them, Baseline could become mired in trivial or false patterns found in the data.
“Biology is really complex. People operate with very different software codes,” Tufts’s Walt said, noting much of Baseline’s data would vary minute to minute due to subjects’ emotional stress, inflammation, and other factors.
Stanford’s Ioannidis said Verily has hired excellent people and has contracted with some of his Stanford colleagues to collaborate on Baseline, which might yield interesting ideas. But the study is too small, and five years too short a duration to uncover anything clearly meaningful about very early signs of cancer or heart disease. Healthy subjects would suffer few serious illnesses.
Compared with observational studies involving hundreds of thousands of subjects, Ioannidis called 10,000 “a speck of dust.”
David Hunter, acting dean at the Harvard T.H. Chan School of Public Health, agreed that Baseline’s scale was too small for its described goals. Hunter is an investigator for the influential Nurses’ Health Study, an examination of chronic disease in more than 275,000 women over the last four decades.
A former Verily manager said such observations echoed what several outside experts said in meetings with top Baseline officials as the project was being designed. At the time, Verily executives ignored or discounted the concerns, the former manager said.
Hunter and Ioannidis said they were supportive of Baseline as a potentially fascinating exercise that might generate interesting research questions. But they held out little hope that it would lead to broad clinical understanding or breakthroughs, despite Google’s world-class skill mining complex datasets with artificial intelligence software.
“We already have quite a lot of experience in the last 10 years working with multidimensional data. If you look at how much of that has moved to clinical application, it’s close to nil,” Ioannidis said. The few useful discoveries affected just handfuls of patients, he added.
Verily is trying — so far, without success — to gain corporate partners to underwrite the study in exchange for exclusive, early looks at the data to sift for novel drug-development prospects.
Baseline might “adapt and expand over time,” Verily said. “While we agree that larger, broader studies also have value, it is designed to collect a rich, deep … dataset that we hope will be leveraged by the broader life science community for years to come.”
SELLING GOOGLE’S BRAIN TRUST
As Verily’s most ambitious, futuristic, and talked-about projects stumble, some of its more modest ventures are quietly showing promise.
In April, Verily Chief Scientific Officer Vik Bajaj said Verb Surgical, a joint venture recently formed with Johnson & Johnson to compete in the crowded robotic surgery market, is working on novel imaging methods. These include ways to help surgeons reliably remove tumors by using fluorescence to highlight the cancer or the margins around it.
Dr. Ken Drazan, a former Johnson & Johnson executive who recently joined Grail, helped create Verb and praised Conrad’s management. “When (Conrad) sees that something isn’t working properly, he becomes maniacal about correcting it,” Drazan said. “Trying to create a life sciences company inside a software company is no trivial challenge. He’s probably one of the few people who has a reasonable shot at it.”
Verily is also working with Dexcom, maker of a glucose monitor that attaches to the abdomen, continuously samples a patient’s blood through a fine needle, and sends readings to an external device. Verily is shrinking the product for patient comfort, and improving data transmission and analysis.
Dexcom executive Jake Leach said a new monitor might be on the market in two or three years, thanks to Verily’s deft work.
And even troubled projects sometimes yield valuable secondary prospects. The electronics designed for the glucose-sensing contact lens could be deployed for autofocus lenses — another Verily project licensed by Alcon. According to Alcon, autofocus is “moving toward clinical trials.”
Former Verily managers and others connected with the company said Verily doesn’t need blockbusters to find paying customers — and can even sell some apparent duds — because it’s as much in the relationship business as in biotech. Verily owns the franchise for selling to life science companies the coveted links to Google’s brain trust and its Android operating system.
Pharmaceutical and device partners get an instant image boost among shareholders and market movers in an era of great expectations for big-data biotech. Some former Verily managers called this the primary motivation behind the Novartis/Alcon deal.
“Being associated with them and their ability to gain access to technology is key,” and “a big part” of Dexcom’s decision to pay Verily up to $100 million, Leach said. “It’s a pipeline to big Google.”
All the hype around Verily has led some scientists and biotech industry experts to dare compare Verily to Theranos, the troubled blood-testing firm whose CEO Elizabeth Holmes became a media star and billionaire on paper before a series of Wall Street Journal articles cast doubts on its core technology.
UC Berkeley business professor Jo-Ellen Pozner said most biotech firms operate stealthily at first, and when they court publicity, they generally “have some muscle to back it up.” Theranos and Verily are exceptions. Their high-profile, media-savvy leaders made big claims without peer-review validation or real proof.
Conrad, who has a sign on his desk saying “do epic shit,” told a tech reporter in a 2014 interview that his nascent firm was “punching way above our weight, and we may have a chance to turn this battleship of health care around.”
“I think about it as the vaporware culture,” Pozner said, using slang for announced technology products that don’t actually exist — a ploy often used to scare competing firms away from a market opportunity.
Verily said that it prefers to avoid early disclosures, but that sometimes external events force its hand, such as a patent filing that would have revealed elements of the Tricorder in December 2014.
“Part of the Silicon Valley ethos is about changing the world, about disruptive technology, about ignoring existing business models,” and “taking on grand challenges,” Pozner said.
“That’s admirable,” she said, but in Verily’s case, “it also feels pretty quixotic.”
Back in June 2013, when Conrad announced the Tricorder project to 500 cheering employees, he made a promise aimed at spurring his team to think bold. Whoever proposed the best life-sciences idea, Conrad said, would win a vacation at his lavish home in Hawaii. “Cure cancer, get a trip to Hawaii,” some joked afterward.
No prize has been awarded.