False notes

The price of sound science is eternal vigilance.

Written by      

In May 2005, Hwang Woo-suk, professor of biotechnology at Seoul National University, announced in the journal Science that he had created 11 human embryonic stem cells, using 185 eggs. His work was everywhere hailed as a biotech breakthrough, and he became a hero in South Korea, which promptly appointed him head of its new World Stem Cell Hub.

But others were sceptical: creating human stem cells was widely thought impossible because of the complexity of our primate genome. So they did a bit of digging.

Soon came a series of accusations from contemporaries, who voiced substantial doubts. After a tip-off, television channel MBC analysed five of Hwang’s cell-line samples, along with corresponding tissue samples, and reported that the DNA in one cell line didn’t match the tissue, raising the possibility that the stem-cell lines hadn’t been cloned from the stated patients.

About then, Hwang’s co-authors began deserting, seeking to have their names taken off the Science paper, citing “problematic information”. In December that year, a Seoul National University investigation found no evidence that his stem cells came from patient-specific clones. At least nine of 11 lines in the Science paper weren’t what he claimed. They simply matched cells from normal embryos created by in vitro fertilisation.

In a press conference in January 2006, Hwang admitted his data was faked: “I feel so miserable that it’s difficult even to say sorry.” All the same, he blamed junior researchers for the malfeasance.

Japanese bone scientist Yoshihiro Sato published more than 200 papers before his death in 2016. Three researchers from the University of Auckland joined forces with a University of Aberdeen clinical nutritionist in 2012 to expose Sato’s fabrications, focusing on his 33 clinical trials.

In October 2009, he was handed a suspended two-year prison sentence. He was cleared of the main charge of fraud, but was convicted of embezzling research funds and illegally buying human eggs for his research. It was the nadir of a long fall from grace.

Science is overwhelmingly a trust exercise. We expect that researchers will obey the rules of the scientific method, and respect the ethics of their institution itself. Scientists are held up as dispassionate observers who keep their personal ambitions, agendas and biases in check out of reverence for the sanctity of the data.

That may be too much to expect. A 2002 survey of more than 3000 researchers in the United States found that fully a third had bent the rules by fabricating data, failing to declare conflicts of interest, plagiarising, or excising contradictory data. In theory, science is self-regulating: before a study can be published in a scientific journal, its methodology, results and references are checked by other scientists who work in the same field. Once published, other researchers will sometimes try to replicate the study’s findings themselves, just to make doubly sure.

All the same, shonky studies still make it into print, in surprising volumes. When a journal discovers it has published a questionable paper, it can seek corrections, or it can retract it altogether. In October, the science blog Retraction Watch released an online database of more than 18,000 retracted papers going back to the 1970s.

Some retractions are simply down to honest mistakes, but more than half of withdrawals were triggered by dishonest or unethical practices. (Retraction Watch currently logs around 500 retractions a year.)

In 2017, German academic press Springer, which publishes more than 3000 scientific titles, retracted 107 papers from Tumor Biology in a single day when it found they’d been championed by fake peer reviews—scientists approving their own papers while posing as someone else.

According to Retraction Watch, Springer has pulled more than 500 papers since 2012, the vast bulk of them from China. A subsequent Chinese government investigation saw hundreds of researchers expelled from their institutions.

So, is science broken? No: between two and three million papers are published every year in nearly 30,000 journals, but only some 0.02 per cent of them end up withdrawn.

That rogue scientists are being caught in greater numbers is, in fact, a sign of increased efforts to reproduce the results of suspect papers, and sharper oversight from editors.

It’s also down to a new level of scrutiny made possible only by sophisticated technology. This year, forensic researchers led by microbiologist Elisabeth Bik, of biotech company uBiome, subjected 960 papers published in the journal Molecular and Cellular Biology to scanning software, looking for duplicated and doctored images. They found 59 contained inappropriately duplicated images, and while some were innocent glitches, one in ten of the papers ended up being retracted.

“If this proportion is representative,” Bik told Science Alert, “then as many as 35,000 papers in the literature are candidates for retraction due to image duplication.”

Why do they do it? Retraction Watch found that fully a quarter of all fraudulent papers were the work of just 500 authors, and that implies some personality disorder at play, according to Cristy McGoff, former director of the research integrity office at the University of North Carolina at Greensboro. “The environment of research advancement contains levels of ego, competitive behaviours, and the need to be respected both by peers and students,” she told Retraction Watch. “I think there are levels of narcissism, and even sociopathy, within some cases.”

Reassuringly, the nemesis of fraudulent scientists is other scientists. In fact, three Auckland researchers were instrumental in uncovering one of the biggest frauds in research history. Japanese stem-cell scientist Yoshihiro Sato published more than 200 papers between 1980 and 2015, many of them on how to reduce bone fractures.

Of Sato’s 33 clinical trials, 12 have been widely cited. Some have changed the outcome of meta-analyses or been incorporated in treatment guidelines. New clinical studies were launched based on Sato’s data.

In one 2005 Neurology paper, Sato claimed a drug called risedronate reduced the risk of hip fractures in female stroke victims by a remarkable 86 per cent. He claimed to have conducted 33 clinical trials, on a total of 5894 patients. But somehow, Sato completed them with phenomenal speed: commentators marvelled at how he amassed large cohorts of study subjects in a fraction of the time it took his contemporaries—in a study of women with Alzheimer’s disease, he apparently recruited a staggering 500 patients in just two months.

What’s more, noted concerned colleagues, Sato’s studies almost always found “extremely large effects with significant results”. Mark Bolland, a clinical epidemiologist at the University of Auckland, was one such sceptic: “There is nothing that I can think of that produces a 70 to 80 per cent reduction in hip fractures, yet Sato was able to do it consistently in all his trials,” he told Science.

When Bolland analysed more than 500 variables that comprised the characteristics of Sato’s patients, across all 33 trials, he found an improbable uniformity. “That just shouldn’t happen,” he says. “The randomised groups were incredibly similar.” The data were simply too perfect to reflect real life.

Sato had fabricated his reports, but it took years of petitioning before influential journals agreed to retract them. The damage was already done: meta-analyses had included Sato’s results, thus contaminating their own. Health societies revised medical guidelines on the strength of his bogus papers. Thousands of dollars were wasted on fruitless attempts by researchers to replicate his results.

Sato committed suicide in 2016. Hwang Woo-suk has spent decades trying to wash out “a big stain”, even as he continues to insist that he really did extract a stem cell from a cloned human embryo. He has funded a new genetics lab with revenues from cloned pets—US $100,000 a doppelgänger—and in 2018, he announced he would bring an extinct ice-age horse back to life. It’s a claim that has the science community wondering if it’s history, not histones, about to repeat.

More by