The investigation into faked AIDS vaccine tests “followed an atypical path compared to most research misconduct cases,” an Iowa State University administrator wrote in a report last October.
The case “began with proof of research misconduct and only after considerable effort was the responsible party identified,” Charlotte Bronson, Iowa State University’s associate vice president for research and research integrity officer. Multiple researchers handled samples that later proved questionable. The research also hadn’t been published, at which point other scientists may have questioned the results.
The investigation, which led ISU scientist Dong-Pyou Han to resign, played out over nearly 10 months, starting a year ago this week. It actually began at Case Western Reserve University in Cleveland, where Han and the leader of his lab group, Michael Cho, worked before Iowa State recruited Cho.
The investigation exonerated Cho and other members of his lab, but it doesn’t erase their embarrassment and chagrin at having pursued a dead lead – and at facing questions about the federal financing that supported the research.
This is the fourth post I’ve written about Cho and Han. I didn’t expect to follow the thread this deeply, but the documents I’ve obtained in the course of my inquiries have been fascinating. They offer a glimpse into how ISU tied the allegations to Han – and into the research Cho’s group conducts.
To summarize, Han admitted adding human antibodies, called immunoglobulin G or IgG, to samples of blood drawn from rabbits that had been inoculated with possible candidates for an AIDS vaccine.
Cho’s research has focused on glycoproteins – proteins with attached carbohydrate molecules, abbreviated as gp – found on the outer membrane of HIV, the virus that causes AIDS. The idea is that giving people doses of the glycoproteins – or doses of benign bacteria engineered to produce the glycoproteins – will trigger an immune response, creating antibodies that would then “neutralize” HIV to ward off infection.
The Cho group’s vaccine candidates are gp120, gp41 and a gp41 segment labeled gp41-54Q. Rabbits were inoculated with glycoproteins or with a version of lactobacillus, a common bacterium Cho’s group genetically modified to produce the glycoproteins.
In her report to David Oliver, then ISU’s interim vice president for research and economic development, Bronson lays out the evidence that led her and Cho to Han.
As I guessed in an earlier post, one of the labs collaborating on the project discovered the fraud. The researcher’s name and location are redacted, but Bronson says that person notified Cho after finding human antibodies in rabbit blood samples the ISU lab had provided.
On January 23, 2013, Cho informed Bronson, as the ISU research integrity officer (RIO), about the “inconsistency in data.” Cho indicated he was looking into it and would inform Bronson if the problem looked like it was due to misconduct.
It’s somewhat surprising Cho would notify ISU administrators about this almost immediately rather than doing some checking first. Perhaps the presence of human antibodies was so far outside the realm of possible error it raised a red flag. In any case, it was wise.
On Valentine’s Day, Cho asked another collaborator to check samples of rabbit sera (blood and blood products) from his lab. Of 81 samples, 17 had neutralized a broad range of HIV “and all (but only) those 17 had been spiked with human” immunoglobulin G, Bronson writes. Han was one of five people who could have handled and/or had access to all 17.
Bronson’s “assessment was that, based on the evidence, the discrepancy constituted research misconduct.” She informed the Office of Research Integrity (ORI) at the U.S. Department of Health and Human Services that ISU was investigating misconduct involving NIH-funded research. The ORI opened a case number.
Cho and Bronson met over the spring and summer to discuss the case and who might be responsible. “Since the samples were being handled by multiple individuals and there was no way to meaningfully use cameras to watch the spiking being done, it became necessary to perform multiple tests … to see if spiking or other falsification would recur and thus allow us to identify the respondent,” Bronson reported.
It’s not clear what this means. Were Cho and Bronson setting a trap? Two pieces of the account make it at least appear so.
In evidence demonstrating that Han manipulated data from experiments, Bronson writes that Han gave Cho test data showing virus-neutralizing activity. But, Bronson writes, the samples Cho gave Han were known to have no neutralizing activity and no HIV antibodies.
Han also appears to have blundered when spiking rabbit sera. In the evidence for another allegation, Bronson says antibodies to gp120 were found “in sera of rabbits confirmed to not have been inoculated with the gp120 glycoprotein or to have gp120 antibodies.” (Emphasis added.) Other tests also found gp120 antibodies in samples from rabbits inoculated with gp41-54 – but NOT gp120.
On August 1, Cho identified Han as the likely culprit. A week later, Bronson and Dawn Bratsch-Prince, associate provost for faculty, met with Han and placed him on administrative leave, with no unescorted access to the College of Veterinary Medicine. His keys, pass cards and university identification were confiscated and his access to ISU computers was terminated.
ISU information technology workers also seized Han’s computers and backups, including a Dell Instrument Controller used with the testing apparatus in Cho’s laboratory.
Frozen mouse and rabbit serum samples related to the gp41-54 and gp41-64 projects were removed from lab freezers. ISU administrators also seized notebooks and files.
Han admitted the misconduct on August 22. He wrote his written admission and resignation the next month and signed it in a meeting with Bronson and Bratsch-Prince on Oct. 3.
Han’s admission includes spreadsheets detailing which samples he contaminated and the data he altered, including fever charts of containing faked data on antibody activity.
He also listed one poster and two oral presentations Cho delivered based on the false results. In her report, Bronson recommended – and Oliver concurred – that Han “retract and remove from the web, to the extent possible, his published poster and oral presentations reporting the falsified data.”
Cho also was to correct the research record by either removing links to his presentations based on the faked data or adding a disclaimer to them.
One poster and two presentations got the most attention – and at least two were still available this month, weeks after Oliver directed their removal.
A poster presentation in Banff, Alberta, apparently isn’t available on the Web. But the title and abstract (pages 22-23) are still available for Cho’s talk at the AIDS Vaccine 2011 meeting in Bangkok, Thailand.
Cho made four other presentations – two in Korea, one in China and one in Rockville, Md. – at other meetings. Though they included falsified data, Han was not an author or coauthor. The abstract is still available for a 2011 Korean meeting (page 50); Han reports an abstract also was available for the second Korean meeting, but I haven’t found it. It appears no abstracts were on the Web for the other two talks.
In fairness, it’s probably impossible to remove some abstracts, given that they’re buried in PDFs containing all the conference abstracts. And Oliver, in his statement accepting the report, commends Cho “for his already considerable progress” in informing NIH, collaborators and colleagues that the data are unreliable.
It also appears the phony results were included in three research grant proposals, although only one was approved. I haven’t been able to decipher which grant that is or how much money it provided.
In any case, as I noted before, trying to unwind this or other grants would be a nightmare – and probably overly punitive. It does call into question disbursement of any unallocated money, however. ISU and NIH are reportedly discussing this.