Reflections on the Retraction of the Time-of-day Immunotherapy Trial

In February, Nature Medicine published a trial that seemed too good to be true. It was welcomed with open arms. Viral is an understatement. Its Altmetric score is over 800, or what we consider “water cooler” talk for a scientific study.

Eric Topol, a key opinion leader, called it the “best proof” of the concept to date.

STAT News cautiously amplified the findings.

Patients seized on the hope, calling their clinics to reschedule infusions to the morning.

The academy moved quickly, too. In just three months, the study was cited 22 times. That’s remarkable given that most papers take many months to move from conception to publication.

But there was just one problem. The trial wasn’t true. After an editorial investigation, Nature Medicine retracted the study on June 24^th.

Some celebrated the retraction as proof that science will self-correct. Don’t get me wrong. I do agree that, over time, science bends toward the truth. Sadly, untruths can last years, sometimes decades or even centuries. But eventually we move past hype, fads, and politics.

Others celebrated the editors for their swift action, retracting the study only months after publication. Again, I believe we should incentivize this behavior, including encouraging authors to come forward and admit mistakes that might otherwise be buried—even when the errors are genuine—for fear of reputational harm.

Is this an example of science self-correcting, led by heroic editors who did the right thing?

I first learned of the retraction from an email by Rebecca Robbins, a New York Times journalist, reaching out for comment based on my earlier AAOx2 post in February, apropos of the situation, titled, “Retract Until Verified.”

I wish I could say I got lucky. Or that I have some paranormal psychic power. Or that I possess prodigy-level critical appraisal skills.

The truth is much simpler, which makes me profoundly sad for the entire research enterprise: from funders to authors to editors to peer reviewers to the scientific public to experts to the media. Countless responsible entities were asleep at the wheel. Or maybe they were never awake to begin with.

Either way, it’s just another tale in a very large book breaking at the spine about the sad state of science.

I read the study.

Seriously, I don’t think many make it past the tweet, let alone the abstract.

The top-line findings were extraordinary—turning the knob on the clock had the same effect as the drug itself compared with placebo. As Reverend Bayes would say, extraordinary claims require extraordinary evidence.

Yet, the more I read, the more suspicious I became. No treatment-related adverse effects, no loss to follow-up, and a large, clean separation in infusion times that defies the messiness of routine clinical practice.

These red flags were compounded by many discrepancies surfaced by other unsung heroes, namely @houndcl (why is their Twitter account suspended?!?!).

The only plausible conclusion I could muster was some major unseen clerical error or, far more likely, plain fraud.

Among the 22 citations accrued in just 3 months is a meta-analysis published in another high-profile journal, JAMA Network Open, on time-of-day immunotherapy administration for advanced cancers.

The authors included 29 studies, but just one was a randomized controlled trial rated by two independent reviewers as having a “low risk of bias.” They conclude:

Although evidence from RCTs is currently limited to [lung cancer], our findings highlight the potential relevance of administration timing as a modifiable factor in cancer immunotherapy.

Will these 22 studies issue notices of error? Or, in the case of the meta-analysis, a retraction, given that the retracted study played an outsized role as the sole RCT included?

I wouldn’t hold my breath.

And if there were already 22 citations in just 3 months, I can assure you there are many more studies in peer review or production just waiting to add to the citation count. Some may scrub the reference, but I doubt they’ll abandon their fervor for circadian immunobiology in cancer therapy. More enthusiasm will yield more grants, more papers, and more press releases.

The retraction is not a death knell for the idea that circadian rhythms matter to immunobiology. But we should reserve our enthusiasm for time-of-day immunotherapy administration when the field rests largely on animal studies, confounded observational data, and implausible pharmacology for drugs whose effects unfold over weeks, not minutes or hours.

There is always room for the dreaded phrase, “future research is needed.” Maybe when we arrive at that future, things will be different.

But that seems hopeful. Sorry, Biggie Smalls, it’s the Same Song. Things have not Done Changed.