Critics challenge data showing key lipids can curb inflammation

For 3 decades, biochemist Charles Serhan at Brigham and Women’s Hospital has been the torch bearer for “resolution immunology”—a research field he pioneered after discovering the first molecules that seem to tamp down, or resolve, inflammation, the body’s rapid cellular and chemical response to injuries, infections, and even allergens. Ever since, research on these stop signals, formally known as specialized pro-resolving mediators (SPMs), has exploded.

Scientists around the world have implicated multiple classes of these lipid-based molecules in all manner of diseases. A search on PubMed for “lipoxin,” the first SPM identified, turns up more than 2200 papers, and at least three biotech companies are planning clinical trials with synthetic molecules meant to restore or boost the body’s natural ability to end inflammation, which can wreak havoc if prolonged.

But what if a large chunk of the research on SPMs is based on flawed science?

Last month, in a paper in Frontiers in Pharmacology, an international group of 18 scientists including specialists in lipid analysis argued that the studies identifying these molecules in people typically detect them at levels so low they are unlikely to play a major role in resolving inflammation. The critics also found that most studies of human samples merely correlate SPM levels with a phase during the arclike time course of inflammation; data proving SPMs actually resolve it are weak to nonexistent, the authors argue.

The paper’s authors offered an even more damning critique: Protocols developed by Serhan and his collaborators to detect SPMs in body fluids or tissue samples don’t conform to accepted standards, they say. What is provided as proof of SPMs in many papers should actually be interpreted as mere “noise,” they argue. “You cannot ignore the fact that there is a growing number of analytical chemists who cannot detect these things,” says Nils Helge Schebb, chair of food chemistry at the University of Wuppertal and lead author.

Even some scientists who didn’t join the broadside have struggled. Immunology researcher Derek Gilroy of University College London, who has collaborated with Serhan but is now moving his lab away from SPMs, tells Science he has had a “very negative experience working with these things and seen some data that makes me question whether these things are real.”

The concerns over the detection methods, Science has learned from several sources, have prompted investigations at Harvard Medical School, the parent organization of Serhan’s hospital, and at Queen Mary University of London (QMUL), where molecular pharmacologist Jesmond Dalli, a former Serhan postdoc, now studies lipid mediators.

Neither university would confirm the probes to Science. The criticisms in the Frontiers commentary are “not true,” Serhan said in an email. “Many others have indeed detected SPMs in human samples.” A single critical paper compared with thousands on SPMs “is not a valid comparison of the facts. It’s not clear to me what the goal of their ‘review’ is because it’s totally wrong.”

“I am baffled,” adds QMUL immunepharmacologist Mauro Perretti, who studies SPMs independently of Dalli but also collaborates with Serhan. “SPMs exist, they have been characterized in multiple ways, they have been synthesized, and they are active.” SPM proponents also note that animal studies and one human trial have already yielded hints that synthetic versions of the molecules work as advertised.

Searching for signals

Serhan has been studying inflammation since his postdoc years at the Karolinska Institute. In 1984, he discovered a novel fatty acid made by the body. Over time, he and his collaborators piled up evidence that the molecule, which they dubbed lipoxin, could shut down inflammation. At the time researchers believed inflammation just dissipated, like a crowd after an event. Serhan argued instead that lipoxin and other molecules actively end the response, like riot police arriving to disperse the crowd. Since then, his lab and others around the world have found and characterized more than 20 other types of lipid-based biochemical that form four SPM classes—lipoxins, resolvins, maresins, and protectins.

Schebb began to look for SPMs about 6 years ago but had trouble detecting any of them in human samples. He visited Dalli’s lab, looking for advice, but says “they didn’t teach me anything that helped.” Schebb soon learned that several other scientists also had trouble detecting SPMs, and he became increasingly skeptical of the research.

Enter Valerie O’Donnell, a lipid biochemist at Cardiff University. About 3 years ago, when reviewing papers on SPMs, O’Donnell noticed a few “strange” looking figures showing the results of liquid chromatography mass spectrometry (LCMS), which separates a sample’s molecular contents into ions with distinct molecular masses, producing a chart called a chromatogram. These diagrams are supposed to show a sharp peak representing a detected molecule. But when O’Donnell looked at high-resolution images of chromatograms said to show SPMs, the figures didn’t look like “real data,” she says.

She consulted Ian Blair and Garret FitzGerald, both at the University of Pennsylvania, and Robert Murphy at the University of Colorado, Denver, who also study lipids, and together they analyzed other published SPM papers. They found at least 70 papers between the Serhan and Dalli research groups with chromatograms that they considered suspicious.

They took some of their criticism public last year. In a preprint posted on 8 December 2021, the same group together with other specialists in lipid analysis highlighted a paper published by Dalli’s group in 2020 that identified a lipoxin, a resolvin, and a maresin in the serum of people with early rheumatoid arthritis. The paper went on to suggest the SPMs could be used as biomarkers to assess patients’ response to medications. But the critics said Dalli was not setting a limit of detection—a value representing the lowest measurable concentration that provides high confidence a molecule exists. Instead, they say, he used a different method that did not comply with accepted standards for detecting biomolecules. When O’Donnell and colleagues applied the criteria described in the paper to inert methanol and buffer fluid, they found it indicated the presence of lipids where clearly none existed.

O’Donnell and her colleagues concluded that Serhan and Dalli often appeared to quantify lipids by integrating what the critics would consider “noise.” “We had never seen mass spectrometry quantitation performed this way before,” Blair says.

In a rebuttal posted this week as a preprint on bioRxiv, Dalli, Serhan, and a colleague defend their methods by reanalyzing some of their LCMS data using the limit-of-detection criteria suggested in the preprint. The outcome confirms his group’s original results, Dalli says. Dalli also notes that an analysis of blank samples on his lab’s mass spectrometry instruments did not yield false signs of SPMs. Dalli canceled an interview after Science contacted QMUL about the dispute but answered questions by email, calling the conclusions of the preprint “incorrect.” He added: “I really cannot understand why someone would feel they need to go to these lengths to attack me and my colleagues.”

No resolution

With the Frontiers paper, the critics have now broadened their case against the SPM work. But if they are right about the flaws in the analyses, how did these papers pass peer review and get published? FitzGerald says there are very few experts in mass spectrometry of low abundance lipids. Consequently, submitted papers about SPMs are usually sent just to experts in the particular disease being studied, who do not necessarily understand the intricacies of mass spectrometry. The critics also note that journals prefer to publish positive results, not negative detections.

Hardly any of the positive results come from labs without a tie to Serhan, FitzGerald says, and only a handful of labs, including Serhan’s, are responsible for the analytical chemistry. “We are motivated to get the literature corrected,” says FitzGerald, whose own doubts about SPMs date back to 2015, when his group found no relationship between inflammation and the formation of SPMs in the urine or plasma of healthy volunteers. “I had satisfied myself that there was nothing to this story, so we just stopped working on it,” he says.

FitzGerald and the other authors on the Frontiers paper aren’t arguing that lipoxins, resolvins, and other putative SPMs don’t exist at all. Some researchers have detected them in people suffering from COVID-19 or septic shock, for example. But they don’t find the signals in “most biological samples,” Schebb says, or see compelling evidence they resolve inflammation.

Serhan counters that just because other scientists have trouble detecting the biochemical in human samples doesn’t make them irrelevant. “These mediators are made in the local environment and by the time they get out in the blood or urine they are on their way out.” There is no reason why they would exist in such samples at high levels, he says.

It is not just an academic debate. Animal studies that indicate SPMs can calm inflammation are paving the way toward clinical tests in people. A phase 1 trial of a lipoxin-based mouthwash published last year indicated the drug was safe in people with periodontal disease and also showed some hints of efficacy. Thetis Pharmaceuticals is planning trials of a synthetic resolvin for cases of inflammatory bowel disease and cancer. Two other biotechnology companies, OSE Therapeutics and ResoTher Pharma, have SPM-based drugs in their pipelines. (Several of the sources quoted in this article, including Serhan, Dalli, and the critics, are involved with the commercial development of SPMs or receive funding from companies developing other types of anti-inflammatory lipids.)

A single phase 1 trial designed to show safety doesn’t prove anything about efficacy, FitzGerald says. But, he concedes, “It is possible that these chemical entities given at high concentrations can modulate the inflammatory response.” Even if SPMs don’t naturally resolve inflammation, it may not matter for drug developers if synthetic versions still do so.

Serhan is sticking to his guns: “I have had independent confirmation of all the work. People can buy synthetic SPMs from several companies, they can use them for detection or functional studies. That’s why there are thousands of publications. If I was trying to pull the wool over people’s eyes, I don’t think you’d see positive results come out of a clinical trial.”

Scientists from both camps are scheduled to give talks at a workshop in late June. But no one expects an easy resolution to this highly inflamed dispute.