Thursday, September 5, 2024

Rushing To Publish - Then Launching A Startup - Why The Cancer 'Blood Test' Never Materialized

 

                                     Bacterium found in digestive tract tumors


Plenty of excitement erupted 4 years ago when a gung-ho team of researchers made a  stunning microbiology claim.  The claim?  Cancers(most of them) have unique microbial signatures that can allow tumors to be diagnosed with a blood test.  Not long after a prestigious journal published the research and more than 600 papers cited the study. At least a dozen groups based new work on the study and the scientists behind the claim actually launched a Wall Street startup to capitalize on the findings.

Well, as fate would have it, those microbiologists were likely the victims of hubris.vIt reminded one of the infamous Elizabeth Holmes and Theranos  -her healthcare startup that claimed to revolutionize the blood-testing industry.  That was by asserting a blood test could be done via a simple drop retrived from a pin prick. As opposed to having to get your vein stuck by a phlebotomist to with draw cubic centimeters  each time.

In the current case, the cancer detected from a blood draw, the paper making the claim was retracted in June.  This followed steady criticisms from other scientists who questioned the methodology and said the findings were invalid.  What led to the canceling?

The original 2020 Nature paper (featuring lead author Rob Knight, regarded as a pioneer in microbial analysis) had reported that 32 different cancer - including prostate and skin melanoma - harbored unique combinations of microbes (see graphic) that could be interpreted as a 'fingerprint', i.e. for each type. Hence, a blood test could theoretically allow physicians to use specific microbes as proxies to identify the cancers.

Didn't work out that way. This despite more than 17,000 samples from 10,000 patients as reported in the Wall Street Journal (Aug. 31, p. 1A).  The Knight study, as reported by the WSJ, was "the first to survey cancers across the human body and showed that microbes existed in combinations that were unique to each type of tumor."  The result eventually was a startup called Micronoma, to which the FDA actually gave a "breakthrough designation to a Micronoma device to test for blood cancer."

That  put the device on an accelerated track for approval. The problem was, independent researchers had begun raising alarms.  For one, some microbes flagged as components of cancer signatures weren't know to exist in humans, prompting further scrutiny. This led Steven Salzberg ( a computational biologist)  and his team at Johns Hopkins to analyze a handful of the cancer types and microbes. Their analysis (published in October 2023 in the journal mBio) found that "the near -perfect association between microbes and cancer types ....is simply put, a fiction." (WSJ Ibid.)

Among the errors, the Rob Knight team (at UC San Deigo) had "incorrectly deployed a genomic tool built by Salzberg's lab to  match tumor data to microbial sequences."  As Salzberg informed the WSJ: "It wasn't a close call. This data is completely wrong."

Incredibly, Knight continues to defend his work, despite the criticisms and the fact that the retraction of the original Nature paper cited Salzberg's arguments and furthermore "all the paper's authors agreed to its retraction."  

The moral of the story? Be sure before claiming x data shows y effect, that the relationship is strong enough to base a paper - or a startup  - on.  Meanwhile, Knight insists his work remains essentially valid. He points to a paper he and his co-authors published in the journal Oncogene in February that "acknowledge the critique and presented a new analysis they said validates their central claim."

Well, we will see in the coming months the degree to which that holds true, and what  - if any- blowback surfaces.  Let us say the jury is still out  but it does look more like Salzberg and his group are correct than that the Knight team is.

See Also:

Physics Is A "Troubled Field"? Not As Much as Biomedical Research And Psychology 


No comments:

Post a Comment