Acting FDA Regulator Fired After Rejecting Dangerous Drug

Written by Anish Koka MD (Cardiology) on July 6, 2026. Posted in Current News

The FDA fired its acting drug chief after she blocked approval of an immune-suppressing drug for children. Her replacement approved it four weeks later — without the independent expert review she had ordered

Here is what the record shows. Tracy Høeg (pictured), acting director of the FDA CDER division for six months, was fired.

Twenty-eight days later, a drug she had declined to approve because of concerns about safety — Teplizumab for Type 1 Diabetes — was approved without the scheduled FDA advisory committee by the official who replaced her.

Accompanying the announcement was a memo written by Mike Davis that detailed Høeg’s disagreement with approval of the drug that was not in line with FDA staff reviewers.

That’s the version of the story that’s being told, but piecing together what really happened reveals a story that should alarm all of us.

First, some background on the drug in question.

What the drug actually does

Type 1 Diabetes — a disease that primarily affects the young — is thought to be a result of an autoimmune reaction that destroys the beta cells in the pancreas responsible for producing insulin.

Teplizumab (brand name Tzield, manufactured by Sanofi) is a CD3-directed monoclonal antibody that suppresses the immune system’s attack on pancreatic beta cells to hopefully prevent the onset of type 1 diabetes.

It has been approved since 2022 for Stage 2 type 1 diabetes, where its documented benefit is delaying the onset of Stage 3 — the point at which patients require insulin — by a median of roughly six months.

That six-month number is different than what you’ll see if you just read the abstract of the randomized controlled trial that led to approval of the drug, which notes:

“The median time to the diagnosis of type 1 diabetes was 48.4 months in the teplizumab group and 24.4 months in the placebo group.”

Understanding that discrepancy is a good case study in the technically correct, but practically incomplete conclusion, and requires a not-too-deep dive into the paper and the supplementary data provided.

The Kaplan-Meier curve from the trial shows a drastic drop in Type 1 Diabetes in the treatment arm in the first 6 months compared to placebo. Pairing it with the annual hazard ratio table in the supplementary appendix confirms what’s happening.

Year 1 is dramatic. Only 3 of 44 teplizumab patients (6.8 percent) developed diabetes versus 14 of 32 placebo patients (43.8 percent). The interval HR is 0.129 — meaning teplizumab-treated patients had roughly an 87 percent lower risk of diagnosis that year.

Essentially the entire drug effect lives in the first 6 months.

Year 2 is the bombshell. Eight teplizumab patients (18.2 percent) developed diabetes versus only two placebo patients (6.3 percent). The interval HR is 1.8. Teplizumab patients were actually at nearly double the risk of diagnosis in year 2 compared to placebo.

The drug appears to have delayed onset in responders, but when the protection wore off, those patients progressed anyway in a concentrated burst. This is disease delay, not disease prevention. The people who were protected in year 1 didn’t escape diabetes; they just had it pushed into year 2.

Years 3, 4, and 5 show interval HRs of 0.58, 0.86, and 0.36 — all with confidence intervals that span 1.0 widely, which means the two groups are statistically indistinguishable from year 3 onward.

The other very interesting datapoint is that post-hoc analysis shows the drug’s benefit is concentrated in a subgroup definable by three biomarkers: HLA-DR3 negative, HLA-DR4 positive, ZnT8 antibody negative, and lower baseline C-peptide.

It’s generally wise not to read too much into subgroup analyses, but the differences here are stark with regards to who accrued the six-month delay to Type 1 DM diagnosis.

But the findings here are quite real, and quite exciting. I would view this as a proof of concept that paves the way for future therapies that would have an even longer effect. Approval for Stage 2 Type 1 Diabetes seems reasonable with the caveat that this was a very small trial and the FDA needs to have their antenna out for any signals of harm.

The current controversy: A new accelerated approval indication

The current debate was about accelerated approval for Stage 3 Type 1 diabetics, which is a different and significantly more aggressive claim. Stage 3 means the patient already has clinical diabetes and is already on insulin.

The question was whether teplizumab, given as two 12-day courses of intravenous teplizumab, can slow further decline when the cat may already be out of the bag.

The current FDA has concluded that answer is yes based on the data Sanofi presented them. Importantly, this new indication is based entirely on the PROTECT trial, which didn’t use clinical outcomes as an endpoint, and instead used a surrogate endpoint: C-peptide, a protein that serves as a proxy marker for endogenous insulin production.

The PROTECT trial showed teplizumab preserved C-peptide levels better than placebo. The problem is that the New England Journal of Medicine data from the same trial also showed no meaningful change in HbA1c over eight months to one year.

HbA1c is the actual clinical measure of blood sugar control — the number that determines how diabetic someone functionally is and what their long-term complication risk looks like. If the surrogate moves but a much closer clinical outcome surrogate like HbA1c does not, you are operating on mostly faith that the mechanism will eventually translate into some clinical benefit.

That is the explicit bet accelerated approval asks the public to make.

But, as the Protect Trial Discussion states there were “ no significant differences between the groups in the key secondary clinical end points — insulin dose, change in glycated hemoglobin, percentage of time in the target glucose range, and clinically important hypoglycemic events”, it did not appear that this product, at this stage of Type 1 diabetes, gets children off of insulin or improves their hemoglobin A1c.

Similar percentages of children were using an insulin pump at the end of the trial.

The detailed discussion about efficacy suggests that if there is a clinical benefit, it is minimal and must be contextualized with safety signals from this therapy. A drug that may work is perfectly reasonable to take if it’s safe.

Unfortunately, safety is always going to be challenging to ascertain in a rare disease with small trials because there simply aren’t enough people to discern rare but meaningful adverse event signals in the phase 3 trials.

This requires a working infrastructure at the FDA that tracks adverse events that come to light after approval of the drugs.

The safety signals

Teplizumab is an immune-suppressing drug. This is exactly what you want in a therapy for a disease driven by an exaggerated immune response.

But when you blunt the immune response to prevent pancreatic beta-cell destruction, you also blunt the immune response to everything else.

The clinical consequences of that tradeoff are documented in the trial data and in post-marketing reports, and they are considerably more alarming than the approval announcement language suggests.

Lymphopenia — a significant suppression of white blood cells — occurred in >70% of patients in prior trials. That is not a rare adverse event. That is just the modal outcome of Teplizumab infusions. But a little immune suppression isn’t so bad, right?

The problem with immune suppression is that it disrupts one of the fundamental mechanisms humans have to stay alive.

Nature is teeming with lots of very bad things, some that you can see, most that you cannot see, and if not for a functioning immune system very bad things can happen.

One of those very bad things relates to the ubiquitous EBV virus.

This is a relatively ubiquitous virus that is kept in check by your immune system and is harbored silently by most kids and adults. Every drug developer knows that any immune suppressing therapy runs the risk of EBV reactivation, and Teplizumab is no different.

In the initial paper that lead to approval of Teplizumab for Stage 2 Type 1 Diabetes (TN-10), the paper states:

“Anti-CD3 monoclonal antibody treatment has been associated with Epstein–Barr virus (EBV) reactivation.”

It then reports: of 30 trial participants who were EBV-seropositive at entry, eight had quantifiable EBV DNA in whole blood at weeks three to six after treatment — all eight were in the teplizumab group, zero in placebo.

One of those eight developed symptoms of pharyngitis, rhinorrhea, and cough on day 38. EBV DNA cleared between day 43 and day 134 (mean day 77).

In the PROTECT trial in Stage 3 Type 1 Diabetics, EBV reactivation was detected in eight patients treated with teplizumab and was asymptomatic in six of these patients. There was one new case of EBV infection in each treatment group. All events resolved without antiviral treatment.

The PROTECT trial required screening out any participant with “a clinically active infection with EBV, including but not limited to infectious mononucleosis, or an EBV viral load >10,000 copies/ml”.

High risk EBV patients were excluded from the trial and you still had eight reactivations out of 217 treated patients (3.7 percent), with one symptomatic case. This isn’t a red flag on its own. Every drug in the class of immunosuppressants is going to do this.

But the question neither of these trials or the FDA can answer is how often that reactivation may become fatal.

The Davis memo seems to allude to this in a bullet point :

“Another OSE review of all U.S. serious cases submitted to the FDA’s Adverse Event Reporting System (FAERS) in pediatric patients under 18 years of age from November 17, 2022, to March 9, 2026, identified no new safety signals, no increased severity of any labeled adverse events, and no deaths directly associated with teplizumab-mzwv.”

FAERS is a public database so I checked it. And I found one fatal case report. In a 30 year old.

The OSE FAERS review Davis mentions is scoped to pediatric patients under 18. Was that intentional?

FAERS report 25854795

This was a report submitted by Sanofi Aventis — a mandatory manufacturer report — and first received by the FDA on September 27, 2025. The patient was a 30-year-old male. He began a course of teplizumab on September 3, 2025, for what the report codes as “diabetes prophylaxis.”

The infusion course ended September 16. Eleven days later, Sanofi filed the death report with FDA. The clinical picture documented in the report is striking. Before the terminal cardiorespiratory arrest, the patient developed tonsillitis, pharyngeal swelling, tonsillar hypertrophy, lymphadenopathy, and splenomegaly — the constellation of findings seen in severe EBV lymphoproliferative disease.

Laboratory findings documented in the report include thrombocytopenia, neutropenia, lymphopenia, markedly elevated liver enzymes, elevated creatinine, and elevated lactic acid, indicating multi-organ involvement.

The terminal events coded are cardiorespiratory arrest, pulseless electrical activity, loss of consciousness, cyanosis, and unresponsive to stimuli. The report also codes “viral infection” as an adverse event. The reporter was a physician.

Notably, the term “Epstein-Barr virus reactivation” does not appear as a coded reaction in this report, even though the clinical syndrome — pharyngitis, tonsillar hypertrophy, lymphadenopathy, splenomegaly, cytopenias, hepatitis, and fatal multi-organ failure — is consistent with exactly that.

Whether EBV was confirmed by testing and simply not coded, or whether the connection was never pursued, cannot be determined from the public record.

What the report does show is that teplizumab was listed as the primary suspect drug by Sanofi’s own pharmacovigilance team.

What FAERS does not show — the malignancy link — and why that matters

Immunosuppression also increases malignancy risk primarily by impairing the immune system’s ability to recognize and destroy nascent cancer cells. There are also a number of cancer causing viruses (EBV is one of them) that can be reactivated by immune suppression.

Davis’s approval memo describes five cases of malignancy identified in teplizumab trial participants. None of these five malignancies appear anywhere in the public FAERS database, and no further details of those malignancies appear in the memo.

That reflects the problem with a passive reporting mechanism like FAERS — it undercounts.

The memo simply details that Høeg asked for a review of these cases by the Office of Surveillance and Epidemiology (OSE) and that a review was conducted by the FDA Oncology division.

Davis’s memo addresses the OSE conclusion directly — noting that “the OSE consult review concluded that, based on available information, it is not possible to implicate teplizumab-mzwv as having a direct effect on the development of reported neoplasms” — but is entirely silent on what the Oncology division found, despite the fact that the Davis memo lists an “Office of Oncology Drugs, Division of Hematologic Malignancies 2 consult dated January 9, 2026” among the documents reviewed.

The memo does not say the Oncology division concurred. It simply omits them. Given that the memo is otherwise exhaustive enough to name Høeg by name more than twenty times, that omission is not easily explained by oversight.

So we have a novel immunosuppressant drug that does not have clear clinical benefit and the lack of improvement in hemoglobin A1c suggests C peptide may not be a valid surrogate marker in stage 3 of the disease, potentially causes some rare, serious cancers, and based on my review of the FAERS report, the clinical picture — tonsillitis, lymphadenopathy, splenomegaly, multi-organ failure — is consistent with fatal EBV reactivation, a risk documented in both published teplizumab trials.

There were two paths to resolving the disagreement Høeg had with FDA internal review staff in the Endocrinology division — either overrule FDA staff as has happened before, or convene an AdCom to see what an independent group of experts recommended. Høeg chose the latter.

Mike Davis, her successor, disagreed, cancelled the AdCom, and approved the drug, noting that there were multiple mandated Post Marketing Requirement (PMR) studies ongoing both to confirm clinical benefit and assess risks of cytokine release syndrome, serious infections, hypersensitivity reactions, lymphoproliferative disorders (LPD) and malignancy.

It is unclear how many patients have been enrolled in this PMR and if it will be completed.

Again we are talking about a treatment for an indication with unclear clinical benefit based on a surrogate marker improvement when the surrogate has not been validated at this stage of the disease in Type 1 diabetes .

The question any parent facing this decision deserves answered is: what are the immune suppression malignancy risk I am accepting in exchange for the potential of not yet documented benefit?

That question will be answered in real-time, in children, outside a trial setting. And that assumes the FDA will actually follow through with the post marketing follow up requirements. Recent history suggests that is hardly a foregone conclusion.

What is troubling is that the official responsible for drug regulation who blocked this approval and ordered an independent expert review was fired.

Her replacement then cancelled the requested expert review and approved the drug despite some concerning safety signals and at least one Oncology expert opinion — not made public in the approval memo — that appears to have shared those concerns.

The facts that aren’t in dispute: The drug is given to children. It suppresses their immune systems. At least five individuals given the drug developed cancers that may be consistent with immune suppression.

The argument from the approval memo is that these cannot be definitively attributed to the drug. That may be correct. But “cannot definitively be attributed” is a substantially lower standard than “independently assessed as acceptable by a panel of outside experts” — which is what Høeg ordered and Davis cancelled.

The FDA’s regulatory legitimacy rests entirely on the premise that its approval decisions are scientific rather than political.

When the sequence of events is: official blocks drug, official is fired, drug is approved — the burden of proof shifts. The agency is no longer simply asking the public to trust the ‘science’.

It is asking the public to trust that the firing and the approval are unrelated.

Given everything I’ve detailed, that is a very substantial ask.

See more here substack.com

Bold emphasis added

Editor’s note: you could be forgiven for concluding that approving drugs that suppress the immune systems of children, causing some to die, might be helping the globalists with their plan to massively reduce the human population.

Please Donate Below To Support Our Ongoing Work To Defend The Scientific Method