Human Research Violations By UCSD Eye Doctor Showcase A National Problem
Tens of millions of people have volunteered their time and bodies to help create breakthroughs in medicine. You see the results with the pain relievers in your medicine cabinet, the vaccines that protect you from disease, the pacemakers that keep your heart beating and the innovations happening now with stem cells.
Yet the systems meant to protect those volunteers from harm are far from perfect, and research violations by Dr. Kang Zhang, an eye doctor at the University of California San Diego, show just how easily that well-intentioned framework can collapse.
Zhang is the chief of eye genetics at UCSD and has a lab named after him at the university. He receives millions of dollars in federal grants and presents at symposiums around the world.
A few years ago, he helped develop a way to remove cataracts from infants and regenerate their lenses using their own stem cells. He also built a tool that scanned over a million patient records and diagnosed illnesses with more than 90% accuracy.
But several of Zhang’s studies were riddled with violations of basic human research standards. A U.S. Food and Drug Administration warning in 2017 and a UCSD audit that followed reveal a pattern that put patients in harm’s way for years.
The 56-year-old doctor enrolled people he shouldn’t have for his medical trials, failed to document what happened to 25 units of a study drug, performed HIV tests on participants without their permission, kept poor records on his patients and didn’t complete necessary ethics training for a stem cell study.
When asked by the FDA to create a plan to prevent more violations from happening in the future, Zhang didn’t provide one.
“There ought to be some serious penalties for this sort of thing,” said Spencer Hey, a Harvard Medical School expert in biomedical research ethics.
inewsource reached out to Zhang, the director of the eye institute where he works and the director of UCSD’s human research protection program for interviews. In response, UCSD sent a statement that said the university had “implemented a comprehensive management plan to address these issues” and suspended Zhang indefinitely from serving as a primary researcher overseeing human research studies at UCSD. He may continue to apply for federal grants, publish in medical journals and train the next generation of scientists.
UCSD later told inewsource, “Zhang’s research had undergone multiple audits since 2012,” which prompted his suspension. When asked if that meant the university had known about Zhang’s violations for five years before taking action, a UCSD spokeswoman would not comment further.
After speaking with five ethics experts for this story, it appears Zhang’s violations — and what happened after they were discovered — are symptomatic of larger problems impacting the field of human subject research in the U.S. Those include patchwork oversight and poor communication between watchdog agencies, a lack of transparency and dialogue with the public, and a combination of money and prestige that sometimes safeguard an institution’s reputation more than patient welfare.
“Science is accelerating always,” said Stacey Springs, Harvard University’s research integrity officer. “But it feels that it’s accelerating in ways that are pushing on regulatory compliance like never before. So we really need to focus now, and learn from each other, and apply these best practices – because it's coming fast.”
An early warning
Zhang received his medical degree and doctorate in genetics from Harvard, then taught at Johns Hopkins University and the University of Utah before founding UCSD’s Institute for Genomic Medicine in 2009.
His accomplishments have landed him on CBS’ “60 Minutes” and in The New York Times, The Wall Street Journal and the Los Angeles Times. He’s received dozens of honors from national and international associations and universities, published or co-authored more than 100 peer-reviewed manuscripts in top journals and recruited human research subjects from around the world – including from the San Diego VA and UCSD’s Shiley Eye Institute on the La Jolla campus.
It was during Zhang’s time at the institute, in the summer of 2016, when the FDA inspected one of his ongoing human trials to ensure the “rights, safety, and welfare” of his patients were protected.
For five years, Zhang had been testing a drug to reverse the effects of a common age-related eye disease. He received approval to enroll patients whose vision had already started to decline – to see if the drug could restore their sight.
Zhang’s research team injected ranibizumab once a month into each test subject’s eyes, 12 times total for each patient. The drug can produce side effects that include eye haemorrhages, pain, inflammation and spots in a field of vision. In rare cases, it can prompt serious cataracts or blindness.
The FDA letter prompted UCSD to suspend enrollment in all of Zhang’s active research projects at the time, pending the results of an internal audit.
Amy Caruso Brown is an assistant professor of bioethics, humanities and pediatrics at New York’s Upstate Medical University. She also is a member of an institutional review board – a safety committee that approves and oversees projects like Zhang’s.
Brown spoke to inewsource after reading the UCSD audit and said, “I have not seen this number of issues in the five years that I've been on an IRB.”Twelve people had participated in the study by the time the FDA stepped in and found five of them were ineligible because they didn’t have the vision problems Zhang outlined for participants. Another patient’s eyesight wasn’t correctly evaluated before the person was injected with the drug.
“If it had been one out of a hundred, we could probably chalk that up to an error that doesn’t reflect a pattern of misconduct,” said Michael Carome, a former associate director at the U.S. Office for Human Research Protections, one of many federal agencies that protects human research subjects.
“But to enroll half the subjects not meeting enrollment criteria – that is more than just an occasional error. That suggests something systematically wrong with how they’re doing the research,” said Carome, who spent years investigating these kind of violations while at the agency.
He left the Office for Human Research Protections in 2010 during a decade of decline, when the office all but stopped using its enforcement tools in favor of “a more friendly approach toward institutions,” he said.
Carome is now a director of the health research group at Public Citizen, a consumer advocacy nonprofit based in Washington, D.C.
He told inewsource that research on otherwise healthy patients needs to be performed carefully, because they aren’t sick enough to justify taking chances.
“Both from a scientific standpoint and ethical human subjects standpoint, not complying with the enrollment criteria is a big deal,” Carome said, calling the guidelines “crucial in terms of ensuring that human subjects are protected.”
The FDA agreed. It issued Zhang a warning letter in January 2017 that called out his use of ineligible patients and his failure to perform required screenings and procedures, poor recordkeeping and lack of documentation about what happened to 25 units of the unused study drug (which Zhang said were destroyed).
Though not mentioned in the letter, the UCSD audit that followed said Zhang also enrolled patients while the study was suspended.
The FDA uses warning letters to document serious research problems and mandate corrective actions. Three times in the letter, it told Zhang his actions raised “concerns about the validity and integrity of the data collected,” and three times it told Zhang he didn’t have an adequate plan to keep his patients safe moving forward.
“We are unable to undertake an informed evaluation of your written response because you did not provide a corrective action plan that, if properly carried out, would prevent this type of violation in the future,” the FDA wrote.
The study was eventually shut down, and inewsource could find no articles published based on the research.
‘Major league science’
Zhang’s work at UCSD should be viewed in context.
He is one of more than 1,600 faculty members in the schools of medicine and pharmacy, one part of a healthcare system at a university ranked among the top research institutions in the country.
UCSD secured $1.2 billion in sponsored research support in 2018 – with $686 million going toward UC Health Sciences – and had more than 7,000 patients participating in clinical trials. Its scientists have made breakthroughs in diabetes research, understanding cancer genes, identifying early signs of autism and treating Alzheimer’s disease. It counts 16 Nobel laureates among current and former faculty.
All of that makes UCSD’s investigation of Zhang unique. The university has published 249 internal audits since July 2010, and the Zhang report is the only one inewsource could find specific to an individual researcher.
Auditors reviewed Zhang’s training records, enrollment logs, regulatory binders and files for ongoing projects that had enrolled human research patients. They found problems everywhere they looked: Zhang failed to get the proper consent from all patients; didn’t report problems to UCSD’s institutional review board; lost documents; kept inaccurate records; wrongly billed patients; and didn’t complete the training required to work with human embryonic stem cells.
In one study, Zhang’s staff tested patients’ blood for HIV and AIDS without telling them, against federal policy.
“He’s lucky there weren’t any major patient harms,” Brown said. ”But if you act like this all the time, eventually you will hurt someone.”
The auditors found Zhang’s actions may have “negatively impacted the rights, welfare, and safety of human subjects in clinical research.” One of his studies, sponsored by the California Institute for Regenerative Medicine, collected tissue from donors with blinding eye disease for a stem cell bank. California voters created the institute in 2004 to fund this type of research.
The study’s rules stated no one under 50 years old was allowed to enroll. UCSD audited files for 50 of the more than 400 patients and found seven were too young, including a minor.
Carome, the former federal director, laughed when he read that section.
“That’s not a subtle mistake,” he said.
Another finding noted the importance of “credible and valid data” when describing a study that was missing 25 of 50 patient progress reports.
“This isn’t minor league stuff. This is major league science,” said Hey, the Harvard bioethicist, who added, “this sort of pattern should raise questions about the validity of (Zhang’s) published work.”
Zhang did publish a correction in January, but not for a study included in the audit. It involved gene editing in animals, and Zhang’s paper said UCSD supervised and approved the research. That wasn’t true – it was overseen by a university and medical center in China.
inewsource couldn’t find any published articles based on the six studies reviewed by the FDA and UCSD. Yet during the audits and study suspensions, Zhang is listed as having continued a genetics research project at the San Diego VA, which did result in 10 published articles.
UCSD doctors often work as attending physi cians at the San Diego VA and share funding, research samples, data and lab space. The VA Hospital is less than a mile from the Shiley Eye Institute and on the same campus. Yet the VA said it was never notified of the UCSD or FDA reports until inewsource asked about them in March.
A VA spokeswoman would not answer questions about whether Zhang was still practicing at the VA, enrolling patients in trials or proposing new research at the institution. The San Diego VA has one of the largest research programs in the national VA network.
Nor were other related federal or state regulators notified. That includes the federal Office for Human Research Protections, which protects research subjects from harm; the National Institutes of Health, which funds many of Zhang’s studies; the Office of Research Integrity, which oversees research misconduct cases; or the California State Medical Board, which gave Zhang a license to practice medicine.
“Communication is fraught with complexity in compliance matters,” said Springs, Harvard’s research integrity officer. Things sometimes happen on a need-to-know basis, she said, and confidentiality plays a big part.
But, she said, “If there are active studies and participants are in danger, the confidentiality is out the window.”
Symptoms of a larger problem
Picture the framework for protecting research subjects as a house.
The foundation consists of a study, planned with sound scientific and ethical principles, and a responsible, ethical researcher. If and when things go wrong or change, they are communicated and addressed immediately. That didn’t happen in this case.
Institutional review boards, the first floor, are often composed of expert volunteers who spend countless hours poring over hundreds of pages of research protocols, guidelines and regulations while also working their regular jobs. They rely on researchers to keep them updated, alert them to problems and speak the truth, but they may also have the power and responsibility to audit ongoing studies. It’s often a proactive system, but it wasn’t in Zhang’s case.
And the institutional review board system is “vulnerable to unethical manipulation, particularly by companies or individuals who intend to abuse the system or to commit fraud,” according to an undercover federal investigation from 2009.
Higher-ups at an institution – the second floor – may have reasons and methods to keep violations quiet and away from public scrutiny. The Zhang audit is a perfect example: It was published more than two years ago but didn’t reach the VA across campus, never made the news, and likely wouldn’t have shocked anyone who stumbled upon it because it never named Zhang as the researcher under scrutiny. The only place his name appears is on the report’s cover page – as one of 16 people copied on its transmittal within the UC system, including UCSD’s chief ethics and compliance officer, vice chancellors, various directors and others.
“It doesn’t look good for the university to be calling out these high-profile faculty for these kinds of violations and making a big deal out of it and applying sanctions, because that’s not the kind of attention you want to draw to your researchers,” Harvard’s Hey said.
Problems with internal probes
An opinion article published by three ethics experts last year in The Journal of the American Medical Association said when internal investigations are completed, the reports are often not standardized, not peer-reviewed, have limited oversight and contain conflicts of interest. “Even when institutions act, the information they release to the public is often limited and unhelpful,” they wrote.
Calling out lucrative researchers can also result in lost funding. Last fiscal year, federal agencies including the National Institutes of Health, National Science Foundation and U.S. Department of Defense gave UCSD $681 million in research funding. UCSD prides itself on how much grant money flows its way each year and has defended that stream aggressively: It sued the University of Southern California in 2015 for poaching one of its most lucrative researchers.
And as quickly as those funding agencies give, they can take away – or even require repayment if serious violations happened using taxpayer dollars.
At the very top of the house, the roof is made up of agencies that regulate human subject protection across the country. The big one – the Office for Human Research Protections – protects those involved in research funded or conducted by the U.S. Health and Human Services Department. It can investigate allegations of wrongdoing but often does not, choosing instead to refer investigations back to the institutions themselves. It can also revoke an entire institution’s ability to perform human research, though that hasn’t happened since 2007.
At Harvard, Springs said she often uses these agencies as a lever in talks with researchers to get them to understand and follow rules. She said the patchwork of institutional, academic and statewide regulations can be messy, so there’s a hope that the feds will act consistently, “because they really are the heavy in these conversations.”
But, she added, “Then they aren't predictable, and you're like, ‘Wow, OK. So they have these enforcement mechanisms and they're not using them. Or they're using them selectively.’”
The Office for Human Research Protections never took action in the Zhang case, which isn’t surprising for two reasons. One, they weren’t informed by the FDA or UCSD of his violations. Two, the agency has drastically cut down on enforcement over the past decade.
For example, it charged institutions with investigating misconduct allegations 94 times in 2002. In 2015, it did that three times. For the same period, the agency went from issuing 146 “determination letters” – an important tool for communicating findings of misconduct – to issuing five.
It’s not due to a lack of funding or a drop in complaints. In fact, institutional misconduct reports sent to the office jumped 400 percent from 2002 to 2014.
Carome said his former agency’s current leadership is “less interested in issuing harsh findings and embarrassing institutions.”
The problem with that, he said, is enforcement actions are “one of the more important tools the office has to change behavior.”
The Office of Research Integrity, another federal agency that oversees research misconduct investigations, has also been criticized for slowing its enforcement. It recently went nearly 10 months without issuing a single finding of research misconduct, though the agency has since stepped up its actions.
One medical ethicist quoted in an industry publication in 2017 called the lack of oversight by the Office of Research Integrity and the Office for Human Research Protections “nothing short of appalling.”
Even if oversight agencies were operating at full speed, a lack of transparency and data sharing would still be major flaws in safeguarding patients that could rip the roof apart.
Compliance data at the Office for Human Research Protections is kept in-house and offline, and getting at it requires a public records request, weeks to months of wait time and then skilled data analysis. The California Medical Board publishes doctor information online, but it’s more concerned with things like medical malpractice judgments, physician substance abuse and negligence in the course of routine health care than monitoring human research and clinical trials. Its database isn’t designed to incorporate audit findings, FDA warning letters or federal databases.
Academic investigations – at least in California – are kept far from Google’s reach, and typically require a public records request and the knowledge they exist. The same goes for institutional review board investigations, reviews and audits.
And none of these systems communicate with each other in any meaningful way.
Carome added one more problem plaguing the field: “Much noncompliance goes undetected or unnoticed.”
Zhang’s audit likely would have gone unnoticed if inewsource weren’t digging into the risks associated with human research, yet Springs said this may prove an opportunity for UCSD.
“When these cases emerge in the press, or people come to hear about them in some way, it highlights our failures – and that should happen,” she said.
But even then, Springs said, it doesn’t always “turn into constructive discussion and dialogue around how we can fix it.”
Audits like Zhang’s are often unintelligible to the community, to patients enrolled and even to academics, even though they are the people who should be providing feedback and criticism, she said.
“These are opportunities where we can actually promote transparency,” she said, “and say, ‘Hey, this is what happened. This is how we approached it,’ and model effective dialogue with communities on how we can work better on this.”
Correction: An earlier version of this story cited a 2017 industry publication that said the Office of Research Integrity had gone a year without making a research misconduct finding. Since the story published, the office has clarified that it went nine months and 21 days without a finding of research misconduct between 2016 and 2017.