Harvard tests an experimental therapy for Creutzfeldt-Jakob - critical summary review

There is a protein that exists normally in everyone's brain. For reasons science still cannot explain, it can fold into an abnormal shape — and once altered, it induces healthy proteins around it to change too. In a chain reaction, the deformed proteins pile up, destroy neurons and leave the brain tissue pocked with tiny cavities, which is why it is described as spongy. That is Creutzfeldt-Jakob disease (CJD): rare, neurodegenerative, progressive, and with no treatment able to halt its advance. Around 90% of patients die within a year of the first symptoms, according to the literature used by Brazil's Ministry of Health.

The name came into circulation in Brazil when the wife of the pilot and influencer Lito Sousa, 59, said on social media that he had been diagnosed with it. With the case came a misleading association — prions, mad cow disease — and, underneath it, a question medicine has never been able to answer: if everything begins with a protein, can you stop it from being produced?

A prion disease, but not mad cow

CJD has four known forms, and only one of them has anything to do with contaminated beef. The sporadic form accounts for about 85% of cases: it appears with no identifiable cause, no proven person-to-person transmission and, according to the Ministry of Health, no link to family history, meat consumption or international travel. The hereditary form covers 10% to 15% and comes from a mutation in the gene that codes for the prion protein. The iatrogenic form, under 1% of cases worldwide, is tied to specific medical procedures — contaminated neurosurgical instruments, cornea and dura mater transplants.

The fourth is the variant (vCJD), identified in the 1990s, mainly in the United Kingdom, among people exposed to products from cattle infected with bovine spongiform encephalopathy. "The variant of CJD is related to mad cow disease, but it is different from the sporadic form of CJD that we usually see in medical practice," says Diogo Haddad, head of the specialised neurology centre at Hospital Alemão Oswaldo Cruz. The clinical picture differs too: the variant tended to affect younger patients and to begin with psychiatric and behavioural signs, while the sporadic form starts with rapid cognitive and neurological deterioration. The Ministry of Health says no case of vCJD has been recorded in Brazil since 2005.

Why a disease this fast takes so long to identify

There is no initial symptom exclusive to CJD. Memory loss, behavioural changes, difficulty walking, loss of coordination and involuntary movements all appear in far more common conditions. The disease enters what neurologists call the differential diagnosis of rapidly progressive dementia — a list that also includes encephalitis, autoimmune diseases, other neurodegenerative conditions and metabolic disorders. Some of those have treatment, which is why they have to be ruled out first.

The speed is the clue. "Alzheimer's, the most classic one, can take years, sometimes decades, of symptom evolution. Creutzfeldt-Jakob generally sets in faster and has a much shorter survival," says Marcio Luiz Figueredo Balthazar, of the neurology department at Unicamp's medical school. Speed alone, though, does not confirm anything.

The investigation combines magnetic resonance imaging, which can show characteristic brain patterns, an electroencephalogram, and a lumbar puncture to analyse the cerebrospinal fluid. From that fluid comes the RT-QuIC, a test that detects the activity of abnormal prion proteins and that, in some patient studies, showed specificity close to 100% — a positive result makes prion disease very likely. Sensitivity varies, so some cases escape it. "This test helps with the diagnosis in most cases. The problem is that it is still not so easily available in Brazil," says Jerusa Smid, neurologist at the USP Hospital das Clínicas, the Emílio Ribas Institute of Infectology and Hospital Israelita Albert Einstein. Definitive confirmation still depends on neuropathological analysis of brain tissue, collected by biopsy or autopsy. That is why the family of the influencer described weeks of waiting: the hypothesis was raised late precisely because it is rare and because the early signs pointed elsewhere.

What the Brazilian numbers show — and what they hide

International estimates, including those used by the British public health service, put the disease at one to two cases per million people a year. Even in countries with well-structured surveillance, it shows up a handful of times a year. In the United States, researchers cited by Harvard Magazine estimate around 500 deaths annually.

CJD has been on Brazil's list of notifiable diseases since 2005. Between that year and 2021, the Ministry of Health received 1,576 notifications of suspected cases, of which 547 were confirmed and 457 discarded. The figure that matters most is the third one: 572 notifications, or 36.3% of the total, never got a final classification, and the ministry's bulletin also records missing information on clinical outcome in more than half the notifications analysed. That does not allow anyone to calculate how many cases go undiagnosed, but it does show the picture is incomplete. "We don't know why this disease is so rare, but it isn't as rare as the data we have in Brazil," says Smid.

Attacking the origin instead of the deposit

That is where the experimental therapy being tested at Harvard comes in. Instead of trying to remove the abnormal protein once it has built up in the brain, the strategy is to reduce how much of it the body produces in the first place. The tool is a small interfering RNA, or siRNA, a molecule that binds to and cuts the messenger RNA carrying the instructions for making the prion protein. "The protein is the problem here, but it's hard to hit it once it's already there, so we want to go upstream and hit the RNA molecule that makes the protein," says Eric Minikel, the trial's principal investigator.

The research did not come out of nowhere. Minikel and Sonia Vallabh left other careers after Vallabh's mother died of fatal familial insomnia, another prion disease, and Vallabh discovered she carried the same mutation. Today the couple co-direct the Prion Therapeutic Science programme at the Broad Institute, a US research centre.

The first phase of the clinical trial, run with Harvard, the Broad Institute and the University of Massachusetts and funded by NeuroNEXT, a programme of the US National Institutes of Health, began in April. It enrols 15 patients who already have symptoms, each receiving a dose of the siRNA by lumbar puncture, with later participants getting higher doses than earlier ones; another 15 people form a comparison group receiving no treatment. Completion is expected in August 2029, according to the ClinicalTrials registry. In mice, a single dose given after symptoms had already appeared reduced prion protein by 49% and increased survival time by 64%. That result is pre-clinical, published in the journal Nucleic Acids Research, and animal findings frequently fail to reproduce in people. A second front by the same group, using gene editing, halved the protein in mice brains and extended their lives by 52%, but human trials there are still years away. It was this first trial that Lito Sousa's wife said the family is queuing for, in a study that opens on 20 October.

What is known — and what is still unknown

What is established: CJD is a prion disease, not an infection you catch through everyday contact; the sporadic form, which is the vast majority, has no connection with diet, and no case of the variant tied to mad cow disease has been recorded in Brazil since 2005. It is rare, at one to two cases per million a year. There is no treatment that reverses or halts it — care is palliative, focused on symptoms, complications and support at home. And the clinical marker that raises suspicion is speed: memory loss combined with motor decline over weeks or a few months, rather than the years typical of Alzheimer's, is information worth taking to a doctor, because it is what triggers the investigation.

What remains open: nobody knows why the protein starts changing shape in the sporadic form, or why it happens to some people and not others. Nobody knows how many Brazilian cases never reach a diagnosis, though the more than a third of notifications without a final classification suggests the official count is short. And nobody knows whether reducing prion protein production is safe or useful in humans — the trial under way tests tolerability in 15 symptomatic patients, not efficacy, and its results are due only in 2029. The question at the centre of it is still unanswered: if the disease begins with a protein taking the wrong shape, does turning down its production change anything once the damage has started?