Late last year, a group of whistleblowers submitted a report to the National Institutes of Health that questioned the integrity of a celebrated USC neuroscientist’s research and the safety of an experimental stroke treatment his company was developing.

NIH has since paused clinical trials for 3K3A-APC, a stroke drug sponsored by ZZ Biotech, a Houston-based company co-founded by Berislav V. Zlokovic, professor and chair of the department of physiology and neuroscience at the Keck School of Medicine of USC.

Three of Zlokovic’s research papers have been retracted by the journal that published them because of problems with their data or images. Journals have issued corrections for seven more papers in which Zlokovic is the only common author, with one receiving a second correction after the new supplied data were found to have problems as well.

For an 11th paper co-authored by Zlokovic the journal Nature Medicine issued an expression of concern, a note journals append to articles when they have reason to believe there may be a problem with the paper but have not conclusively proven so. Since Zlokovic and his co-authors no longer had the original data for one of the questioned figures, the editors wrote, “[r]eaders are therefore alerted to interpret these results with caution.”

“It’s quite unusual to see this volume of retractions, corrections and expressions of concern, especially in high-tier influential papers,” said Dr. Matthew Schrag, an assistant professor of neurology at Vanderbilt who co-authored the whistleblower report independently of his work at the university.

Both Zlokovic and representatives for USC declined to comment, citing an ongoing review initiated in the wake of the allegations, which were first reported in the journal Science.

“USC takes any allegations of research integrity very seriously,” the university said in a statement. “Consistent with federal regulations and USC policies, this review must be kept confidential.”

Zlokovic “remains committed to cooperating with and respecting that process, although it is unfortunately required due to allegations that are based on incorrect information and faulty premises,” his attorney Alfredo X. Jarrin wrote in an email.

Regarding the articles, “corrections and retractions are a normal and necessary part of the scientific post-publication process,” Jarrin wrote.

Authors of the whistleblower report and academic integrity experts challenged that assertion.

“If these are honest errors, then the authors should be able to show the actual original data,” said Elisabeth Bik, a microbiologist and scientific integrity consultant who co-wrote the whistleblower report. “It is totally human to make errors, but there are a lot of errors found in these papers. And some of the findings are suggestive of image manipulation.”

Given the staid pace of academic publishing, publishing this many corrections and retractions only a few months after the initial concerns were raised “is, bizarrely, pretty quick,” said Ivan Oransky, co-founder of Retraction Watch.

The whistleblower report submitted to NIH identified allegedly doctored images and data in 35 research papers in which Zlokovic was the sole common author.

“There had been rumblings about things not being reproducible [in Zlokovic’s research] for quite some time,” Schrag said. “The real motivation to speak publicly is that some of his work reached a stage where it was being used to justify clinical trials. And I think that when you have data that may be unreliable as the foundation for that kind of an experiment, the stakes are just so much higher. You’re talking about patients who are often at the most vulnerable medical moment of their life.”

Over the years, Zlokovic has created several biotech companies aimed at commercializing his scientific work. In 2007, he co-founded ZZ Biotech, which has been working to gain federal approval of 3K3A-APC.

The drug is intended to minimize the bleeding and subsequent brain damage that can occur after an ischemic stroke, in which a blood clot forms in an artery leading to the brain.

In 2022, USC’s Keck School of Medicine received from NIH the first $4 million of a planned $30-million grant to conduct Phase III trials of the experimental stroke treatment on 1,400 people.

In Phase II of the trial, which was published in 2018 and called Rhapsody, six of the 66 patients who received 3K3A-APC died in the first week after their stroke, compared to one person among the 44 patients who got a placebo. Patients who received the drug also tended to report more disability 90 days after their stroke than those who got the placebo. The differences between the two groups were not statistically significant and could have been due to chance, and the death rate for patients in both groups evened out one month after the initial stroke.

“The statements that there is a risk in this trial is false,” said Patrick Lyden, a USC neurologist and stroke expert who was employed by Cedars-Sinai at the time of the trial. Zlokovic worked with Lyden as a co-investigator on the study.

One correction has been issued to the paper describing the Phase II results, fixing an extra line in a data table that shifted some numbers to the wrong columns. “This mistake is mine. It’s not anybody else’s. I didn’t catch it in multiple readings,” Lyden said, adding that he noticed the error and was already working on the correction when the journal contacted him about it.

He disputed that the trial represented any undue risk to patients.

“I believe it’s safe, especially when you consider that the purpose of Rhapsody was to find a dose — the maximum dose — that was tolerated by the patients without risk, and the Rhapsody trial succeeded in doing that. We did not find any dose that was too high to limit proceeding to Phase III. It’s time to proceed with Phase III.”

Schrag stressed that the whistleblowers did not find evidence of manipulated data in the report from the Phase II trial. But given the errors and alleged data manipulation in Zlokovic’s earlier work, he said, it’s appropriate to scrutinize a clinical trial that would administer the product of his research to people in life-threatening situations.

In the Phase II data, “there’s a coherent pattern of [patient] outcomes trending in the wrong direction. There’s a signal in early mortality … there’s a trend toward worse disability numbers” for patients who received the drug instead of a placebo, he said.

None are “conclusive proof of harm,” he said. But “when you’re seeing a red flag or a trend in the clinical trial, I would tend to give that more weight in the setting of serious ethical concerns around the pre-clinical data.”

The NIH paused the clinical trial in November, and it remains on hold, said Dr. Pooja Khatr, principal investigator of the NIH StrokeNet National Coordinating Center. Khatr declined to comment on the pause or the trial’s future, referring further questions to USC and NIH.

The NIH Office of Extramural Research declined to discuss Rhapsody or Zlokovic, citing confidentiality regarding grant deliberations.

ZZ Biotech Chief Executive Kent Pryor, who in 2022 called the drug “a potential game-changer,” said he had no comment or information on the halted trial.

Zlokovic is a leading researcher on the blood-brain barrier, with particular interest in its role in stroke and dementia. He received his medical degree and doctorate in physiology at the University of Belgrade and joined the faculty at USC’s Keck School of Medicine after several fellowships in London. A polyglot and amateur opera singer, Zlokovic left USC and spent 11 years at the University of Rochester before returning in 2011. He was appointed director of USC’s Zilkha Neurogenetic Institute the following year.

A USC spokesperson confirmed that Zlokovic has retained his titles as department chair and director of the Zilkha institute.

Power is invisible, but its effects can be seen everywhere — especially in the health records of active duty military personnel.

By examining details of 1.5 million emergency room visits at U.S. military hospitals nationwide, researchers found that doctors invested significantly more resources in patients who outranked them than in patients of equal or lesser rank. The additional clinical effort devoted to powerful patients came at the expense of junior patients, who received worse care and were more likely to become seriously ill.

Military rank wasn’t the only form of power that translated into inequitable treatment. The researchers documented that patients fared better when they shared the same race or gender as their doctor, a pattern that tended to favor white men and caused Black patients in particular to be shortchanged by their physicians.

The results were published Thursday in the journal Science.

The findings have implications far beyond the realm of the military, said Manasvini Singh, a health and behavioral economist at Carnegie Mellon University who conducted the research with Stephen D. Schwab, an organizational health economist at the University of Texas at San Antonio.

For instance, they can help explain why Black students do better in school when they are taught by Black teachers, and why Black defendants get more even-handed treatment from Black judges.

“We think our results speak to many settings,” Singh said.

The disparities wrought by power imbalances are easy to spot but difficult to study in real-world scenarios.

“It’s just hard to measure power,” Singh said. “It’s abstract, it’s complicated.”

That’s where the military health records come in.

The Military Health System operates 51 hospitals across the country. The doctors who staff them are active-duty personnel, as are many of the patients they treat. Comparing their ranks gave Singh and Schwab a handy way to gauge the power differential between physicians and the people in their care.

The researchers restricted their analysis to patients who sought treatment in emergency departments, where patients are randomly assigned to doctors. That randomness made it easier to measure how power influenced the treatment patients received.

To further isolate the effects of power, the researchers made comparisons between patients of the same rank. If they happened to outrank their doctor, they were considered a “high-power” patient. If not, they were classified as a “low-power” patient.

The medical records showed that doctors put 3.6% more effort into treating high-power patients than low-power ones. They also utilized significantly more resources such as clinical tests, scans and procedures, according to the study.

Those extra resources translated into better care: High-power patients were 15% less likely to become sick enough to be admitted to the hospital over the next 30 days.

To see if they could replicate their results, Singh and Schwab narrowed their focus to doctors who treated patients within a one-year period before or after the patients were promoted to a higher rank. The researchers found that doctors devoted 1% more effort to patients post-promotion, as well as more medical resources. Those differences may have been small, but they were statistically significant, Schwab said.

Next, the pair considered what happened to low-power patients while high-power patients were getting extra attention. One hypothesis was that ordering additional tests for one patient might prompt doctors to order the same tests for everyone they treated that day. It was also possible that the decisions doctors made for their high-power patients had no bearing on their other patients.

Neither turned out to be the case. Instead, the added effort spent on high-power patients was siphoned away from low-power patients, who got 1.9% less effort from their doctors. On top of that, their risk of needing to return to the ER or be admitted to the hospital over the following 30 days increased by 3.4%, the researchers found.

“The powerful unwittingly ‘steal’ resources from less-powerful individuals,” Schwab and Singh wrote.

Outside the military, doctors and patients can’t use official rank to measure their power relative to each other, but they do contend with the effects of race and gender. That led the researchers to investigate whether the physicians in their study treated patients differently if they shared these attributes.

White doctors devoted more effort to white patients than to Black patients across the board, the researchers found. The gap was the same regardless of whether the doctor had a higher or lower rank than the patient.

However, white doctors increased their effort for high-power patients by the same amount regardless of race. As a result, white doctors treated high-power Black patients the same, on average, as low-power white patients.

The story was different for Black doctors. When they outranked their patients, they gave essentially the same amount of effort to everyone. But on the rare occasions when they encountered a higher-ranked Black patient, the amount by which they dialed up their efforts was more than 17 times greater than it was when they treated a higher-ranked white patient.

It’s not clear what accounted for this “off-the-charts effort,” the researchers wrote. They speculated that since Black service members were underrepresented among the pool of high-power patients, Black doctors were particularly attuned to their status.

The effects of gender were more difficult to ascertain, since biology dictates that men and women require different kinds of care.

Both male and female doctors invested the most effort in female patients who outranked them. But male doctors upgraded their care for high-power patients of both genders to a much greater extent than female doctors. And unlike female doctors, male doctors devoted more effort to female patients across the board.

Finally, the researchers wondered whether doctors gave preferential treatment to high-power patients because of their elevated status or because those patients had the authority to make trouble if they were unsatisfied with their care. To make inferences about this, they compared the treatment of retirees (who retained their status but had given up their authority) to the treatment of active-duty patients (who still had both).

Schwab and Singh found that high-power patients continued to elicit extra effort from doctors for up to five years after they retired, suggesting that status was an important factor.

“I think it’s really, really cool that even after retirement, you still have these effects,” said Joe C. Magee, a professor of management and organization at the NYU Stern School of Business who studies the role of hierarchy. He sees that as a strong sign that status was driving doctors’ decisions all along.

“What these folks are able to show is that it has real health consequences,” Magee said.

Eric Anicich, a professor of management and organization at the USC Marshall School of Business, called the study “impressive” and the findings “important.”

Although a 3.5% increase or a 1.9% decrease in physician effort may seem small, their cumulative impact is meaningful, especially when it comes to something as critically important as healthcare, he said.

The inequities documented in the study aren’t unique to doctors or to the armed forces, Schwab and Singh said. The mathematical model they developed to describe the behavior in military emergency rooms also helps explain why people in all kinds of situations give preferential treatment to people who look like them: It may help minimize the effects of societal disparities.

In a commentary that accompanies the study, Laura Nimmon of the University of British Columbia’s Centre for Health Education Scholarship wrote that “the ephemeral and unobservable nature of power has made it profoundly difficult to study.” But she said it’s worth the effort to make sure doctors wield their power more fairly.

The disparities reported by Schwab and Singh are “of serious concern to society at large,” she wrote.

As health officials turn increasingly toward wastewater testing as a means of tracking the spread of H5N1 bird flu among U.S. dairy herds, some researchers are raising questions about the effectiveness of the sewage assays.

Although the U.S. Centers for Disease Control and Prevention says current testing is standardized and will detect bird flu, some researchers voiced skepticism.

“Right now we are using these sort of broad tests” to test for influenza A viruses in wastewater, said epidemiologist Denis Nash, referring to a category of viruses that includes normal human flu and the bird flu that is circulating in dairy cattle, wild birds, and domestic poultry.

“It’s possible there are some locations around the country where the primers being used in these tests … might not work for H5N1,” said Nash, distinguished professor of epidemiology and executive director of City University of New York’s Institute for Implementation Science in Population Health.

The reason for this is that the tests most commonly used — polymerase chain reaction, or PCR, tests — are designed to detect genetic material from a specific organism, such as a flu virus.

But in order for them to identify the virus, they must be “primed” to know what they are looking for. Depending on what part of the virus researchers are looking for, they may not identify the bird flu subtype.

There are two common human influenza A viruses: H1N1 and H3N2. The “H” stands for hemagglutinin, which is an identifiable protein in the virus. The “N” stands for neuraminidase.

The bird flu, on the other hand, is also an influenza A virus. But it has the subtype H5N1.

That means that while the human and avian flu virus share the N1 signal, they don’t share an H.

If a test is designed to look for only the H1 and H3 as indicators of influenza A virus, they’re going to miss the bird flu.

Marc Johnson, a professor of molecular microbiology and immunology at the University of Missouri, said he doesn’t think that’s too likely. He said the generic panels that most labs use will capture H1, H3 and H5.

He said while his lab specifically looks for H1 and H3, “I think we may be the only ones doing that.”

It’s been just in the last few years that health officials have started using wastewater as a sentinel for community health.

Alexandria Boehm, professor of civil and environmental engineering at Stanford University and principal investigator and program director for WastewaterSCAN, said wastewater surveillance really got going during the pandemic. It’s become a routine way to look for hundreds if not thousands of viruses and other pathogens in municipal wastewater.

“Three years or four years ago, no one was doing it,” said Boehm, who collaborates with a network of researchers at labs at Stanford, Emory University and Alphabet Inc.’s life sciences research organization. “It sort of evolved in response to the pandemic and has continued to evolve.”

Since late March, when the bird flu was first reported in Texas dairy cattle, researchers and public health officials have been combing through wastewater samples. Most are using the influenza A tests they had already built into their systems — most of which were designed to detect human flu viruses, not bird flu.

On Tuesday, the CDC released its own dashboard showing wastewater sites where it has detected influenza A in the last two weeks.

Displaying a network of more than 650 sites across the nation, there were only three sites — in Florida, Illinois and Kansas — where levels of influenza A were considered high enough to warrant further agency investigation. There were more than 400 where data were insufficient to allow a determination.

Jonathan Yoder, deputy director of the CDC’s Division of Infectious Disease Readiness and Innovation, said those sites were deemed to have insufficient data because testing hasn’t been in place long enough, or there may not have been enough positive influenza A samples to include.

Asked if some of the tests being used could miss bird flu because of the way they were designed, he said: “We don’t have any evidence of that. It does seem like we’re at at a broad enough level that we don’t have any evidence that we would not pick up H5.”

He also said the tests were standardized across the network.

“I’m pretty sure that it’s the same assay being used at all the sites,” he said. “They’re all based on … what the CDC has published as a clinical assay for for influenza A, so it’s based on clinical tests.”

But there are discrepancies between the CDC’s findings and others’.

Earlier this week, a team of scientists from Baylor College of Medicine, the University of Texas Health Science Center at Houston, the Texas Epidemic Health Institute and the El Paso Water Utility, published a report showing high levels of bird flu from wastewater in nine Texas cities. Their data show that H5N1 is the dominant form of influenza A swirling in these Texas towns’ wastewater.

But unlike other research teams, including the CDC, they used an “agnostic” approach known as hybrid-capture sequencing.

“So it’s not just targeting one virus or one of several viruses,” as one does with PCR testing, said Eric Boerwinkle, dean of the UTHealth Houston School of Public Health and a member of the Texas team. “We’re actually in a very complex mixture, which is wastewater, pulling down viruses and sequencing them.”

“What’s critical here is it’s very specific to H5N1,” he said, noting they’d been doing this kind of testing for approximately two years, and hadn’t ever seen H5N1 before the middle of March.

Blake Hanson, an assistant professor at the University of Texas Health Science Center at Houston School of Public Health and a member of the Texas wastewater team, agreed, saying that PCR-based methods are “exquisite” and “highly accurate.”

“But we have the ability to look at the representation of the entire genome, not just a marker component of it. And so that has allowed us to look at H5N1, differentiate it from some of our seasonal fluids like H1N1 and H3N2,” he said. “It’s what gave us high confidence that it is entirely H5N1, whereas the other papers are using a part of the H5 gene as a marker for H5.”

Boerwinkle and Hanson underscored that while they could identify H5N1 in the wastewater, they cannot tell where it came from.

“Texas is really a confluence of a couple of different flyways for migratory birds, and Texas is also an agricultural state, despite having quite large cities,” Boerwinkle said. “It’s probably correct that if you had to put your dime and gamble what was happening, it’s probably coming from not just one source but from multiple sources. We have no reason to think that one source is more likely any one of those things.”

But they are pretty confident it’s not coming from people.

“Because we are looking at the entirety of the genome, when we look at the single human H5N1 case, the genomic sequence … has a hallmark amino acid change … compared to all of the cattle from that same time point,” Hanson said. “We do not see that hallmark amino acid present in any of our sequencing data. And we’ve looked very carefully for that, which gives us some confidence that we’re not seeing human-human transmission.”

The Texas’ team approach was really exciting, said Devabhaktuni Srikrishna, the CEO and founder of PatientKnowHow.com, noting it exhibited “proof of principle” for employing this kind of metagenomic testing protocol for wastewater and air.

He said government agencies, private companies and academics have been searching for a reliable way to test for thousands of microscopic organisms — such as pathogens — quickly, reliably and at low cost.

“They showed it can be done,” he said.