Figures in cancer paper at root of newly failed compound called into question

How much role did a potentially problematic paper play in the demise of a once-promising compound?

Researchers are questioning the validity of a high-profile article, published by Nature in 2006. Although the letter is 12 years old, the concerns have current implications: It was among the early evidence used to develop a cancer compound that recently failed a number of clinical trials.

It’s unclear whether the problems with the paper — if validated — could have contributed to the compound’s demise. But an outside expert has some thoughts — and so do image experts and multiple external reports, including one released this month, which agree the concerns about the figures have merit. (The first author’s ex-husband isn’t too happy with the article, either.)

The letter, whose first and last authors were based at Stanford University, suggested targeting the enzyme lysyl oxidase (LOX) could prevent and treat cancer metastases. Less than two years after publication, the paper was among the more than 400 references cited in a patent application to develop inhibitors of the protein, assigned to Gilead Biologics. In 2015, PubPeer users began discussing figures in the paper, suggesting some of the images showed signs of manipulation. In late 2016, Gilead declared it was abandoning the LOX-inhibitor compound it had developed, simtuzumab, after it had failed multiple clinical trials for different indications, including cancer, lung disease, and liver problems.

A spokesperson for Nature told us:

We are following an established process to investigate the issues. However, that process is continuing and we do not comment on individual ongoing investigations.

Last author Amato J Giaccia at Stanford declined to comment, as did first author Janine Erler, who now runs a lab at the Biotech Research and Innovation Centre at the University of Copenhagen.

“Lysyl oxidase is essential for hypoxia-induced metastasis” has been cited 794 times since it was published in 2006, according to Clarivate Analytics’ Web of Science.

“There are sufficient concerns”

We consulted numerous image experts about the image concerns raised on PubPeer. All said the concerns raised on PubPeer have some merit.

We spoke with Mike Rossner of Image Data Integrity who has become known as a “manipulation detective” after he instituted a policy at the Journal of Cell Biology of screening images in accepted manuscripts for signs of manipulation. Rossner told us:

In my opinion, there are sufficient concerns about the published images in Fig.4C and Suppl. Fig.1A to warrant examination of the original data.

James Longden, a researcher who collaborated with Ehrler while working in the lab of Rune Linding at the University of Copenhagen, told us he paid approximately 150 euros for an image analysis of the 2006 paper after he developed concerns about her work. The report corroborates Rossner’s concerns, noting potential duplications and splicing.

Rune Linding — also Erler’s ex-husband — contacted us independently about concerns regarding the paper. He sent us an analysis of the paper by the HEADT Centre, which analyzes data manipulations. That analysis also found signs of duplication, and concluded:

The HEADT Centre research integrity group strongly supports a closer investigation of the case and encourages all of the involved parties to assess and evaluate the indicated inconsistencies and to carry out a more thorough review of the case.

Linding — who said he and Erler are involved in a custody battle — also forwarded emails he sent to the U.S. Office of Research Integrity (ORI), in which a representative of the U.S. Department of Health and Human Services (under which the ORI falls) says the agency will review the allegations on PubPeer that fall under the ORI’s jurisdiction. Linding forwarded correspondence he had with Stanford, including a July 2017 email in which a representative informed him a dean had appointed a committee to assess the allegations, but determined that no further inquiry was warranted — given the age of the article, the fact the complainants were anonymous, and other factors.

A representative of Stanford could “neither confirm nor deny an investigation.”

If the images are, in fact, problematic, how much of an influence might that have had on the failure of the compound? A spokesperson for Gilead declined to comment on the allegations about the paper, nor its role in the decision to develop simtuzumab. He also declined to reveal how much the company had invested in the compound before abandoning it.

Even if the 2006 Nature letter is problematic in some way, that likely would have had little impact on Gilead’s decision to develop the compound, and its subsequent failure, according to Steven D. Nathan, director of the Advanced Lung Disease Program and director of the Lung Transplant Program at Inova Fairfax Hospital. “Obviously anything that’s fraudulent might affect people’s interpretation of [any potential compound’s success],” said Nathan. But “I highly doubt the company developed this based on one study,” he added. “I think it probably would have been developed anyway.”

Simtuzumab was the focus of multiple clinical trials; the Phase 2 trial in idiopathic pulmonary fibrosis enrolled 544 patients before it was terminated; a Phase 2 trial in a form of cirrhosis enrolled 259 patients. A Phase 2 trial of the compound in 266 patients with colorectal cancer has not yet posted its results on clinicaltrials.gov, as has a trial of 250 patients with pancreatic cancer.

Although it’s always disappointing to see once-promising compounds fail in the clinic, Nathan said that he wasn’t surprised to see simtuzumab didn’t help with idiopathic pulmonary fibrosis, even though the LOX pathway is known to be important in the disease:

Closing one door on the disease might just mean the disease uses other escape mechanisms and pathways.