Fighting disease
with the power of cells.

Unless the severed wiring is reconnected exactly as it was, there can be no recovery from spinal cord injury—decades of research have been shaking this premise. Biology that grows and guides; engineering that wakes dormant circuits and trains them. A 2019 review in Nature Medicine lays these two toolboxes side by side and argues that combinations should be designed not as simple addition but around the interactions between mechanisms and their timing. Dr. Exotaro reads it, adding what has happened since, up to 2026.

mRNA vaccines contain no adjuvant, yet immunity is strong. A 2025 npj Vaccines study: empty LNPs with no mRNA switched on innate immunity via the ionizable lipid and TLR4, and adding mRNA added nothing.

The 'encapsulation efficiency 94%' on an LNP spec sheet counts mRNA, not particles. A 2022 Nature Communications paper counted one particle at a time: mean 2.80 copies, mode 2, 77% carrying none.

An 'extra-large extracellular vesicle' (XLEV), far bigger than an exosome, is released from the region around the cell nucleus loaded with sonic hedgehog, the molecule that shapes the body. The work comes from the University of Tokyo. The surprise was that what pushed that release was the 'off' form of a protein called Rab18 — a logic that contrasts with conventional models, which have emphasized the GTP-bound form.

A 2022 formulation paper tackles one thing: how to give an LNP its address label. Direct surface coupling swelled the particles to 153 nm; inserting after dialysis held them to 93 nm.

Where inside a tumor do dendritic cells, the command centers of immunity, gather, and who are their neighbors? This article explains a preprint, not yet peer reviewed, that brings together single-cell analysis of 28 human and 12 mouse datasets, spatial transcriptomics of human colorectal cancer, and multiplexed imaging of more than 1.4 million cells in mouse melanoma. After immunotherapy, the detectable changes in the transcriptional states of dendritic cells were minimal, whereas their spatial arrangement clearly changed.

A review sketching a blueprint for immuno-exosomes in hepatocellular carcinoma. Yet it names neither the target antigen nor any approved comparator drug. The authors themselves call it speculative.

Rats had their spinal cords hemisected at two separated levels—the right side at T7 and the left side at T12—and received a single intravenous dose of bone-marrow-derived mesenchymal stem cells the day after injury. At 28 days the motor score was 12.2 versus 5.4. The nerve pathways drawn from nine hours of 7-tesla MRI on the excised cords had built a detour to the outside at the rostral lesion. At the caudal lesion, however, what stood out was not a lateral detour but a route crossing the midline to the opposite side. Dr. Exotaro reads a preclinical study.

Small extracellular vesicles (sEVs) from human bone marrow MSCs, delivered to spinal cord–injured rats either as daily injections or as a continuous infusion from an osmotic pump. At the same total dose of 9.2 μg, only the group infused continuously for six days climbed to a BBB score of 12.9±0.9, while the same amount split into six injections stalled at 8.6±0.9. Doubling the injected dose changed nothing against the three-day schedule, and collagen 1 at the lesion was also significantly lower under continuous infusion than under injections of the same total amount. Dr. Exotaro reads a preclinical rat study.

Extracted wisdom teeth and baby teeth—this paper takes on spinal cord injury with exosomes from stem cells harvested from teeth that would normally be thrown away. The authors compared 3 kinds of stem cells on the same footing and chose dental pulp, narrowed 727 proteins down to a candidate called PSMD1, a “part of the proteasome,” and report that it tipped microglia toward the M2 side. But the treatment was tried only in cultured cells and mice. Dr. Exotaro reads through this preclinical study.

A review that takes on insulin resistance with gene-modified MSC exosomes loaded with miR-21 and SIRT3. But it contains not a single experiment by the authors themselves, prints not one p-value, dose, or n, and cites not one clinical trial targeting insulin resistance in humans. Dr. Exotaro reads through a review that stays preclinical.

An original research paper that ran human bone marrow–derived MSCs continuously in a hollow fiber bioreactor for about 40 days and kept harvesting the small extracellular vesicles (sEVs) that came out, four times a week. The paper reports that particle size, markers, and yield were stable (no numerical value for particle size appears in the main text), while mass spectrometry found 1,618 proteins differentially altered between Phase 3 and Phase 1. It is entirely a story from inside a culture device, and whether the later vesicles work was never measured once.

We read a review that promotes the "young" mesenchymal stem cells harvested from the umbilical cord jelly thrown away at every birth, along with their secreted products, as the trump card of post-stroke regenerative medicine. There are no new data, and no literature search procedure is written down. And among its 285 references, there was not a single clinical trial that gave stroke patients material explicitly identified as Wharton's jelly-derived.

1.2 billion allogeneic MAPC (MultiStem) IV 24–48 h after stroke. MASTERS (double-blind, 137 randomised, 33 UK/US sites) was safe, but the day-90 primary endpoint failed in the 126 analysed (OR 1.08, p=0.83). Plus the 9 years since.

At eight UK sites, 20 million allogeneic neural stem cells (CTX0E03) were injected into the putamen of 23 stroke patients by stereotactic neurosurgery. The primary endpoint was not met, the randomized trial meant to confirm the findings stopped at 15 patients, and the developer went through administration and delisted. Here is what a single-arm study can and cannot say, read straight from the numbers.

An average of 22 months after stroke, paralysis that should no longer move. A phase 1/2a trial implanting genetically modified allogeneic bone marrow MSCs (SB623) around the infarct met its primary endpoint in 18 patients over two years. But there is no control group. Two months later the phase 2b ACTIsSIMA, sham-controlled in 163 patients, missed its primary endpoint (p = 0.6743). The same cells reached approval not in stroke but in the parallel traumatic brain injury trial.

IL-10, the cytokine that calms inflammation, never quite became a drug: it is unstable and clears from the body fast. A Michigan State University team harvested extracellular vesicles from cells made to overexpress IL-10 with minicircle DNA, and shows they are more stable than recombinant IL-10 and quiet inflammatory macrophages. But naïve vesicles, with no IL-10 gene introduced, worked too. Every experiment is in cultured cells.

ALMB-0166, the first Cx43-hemichannel antibody, given as a single dose to 24 SCI patients: adverse events matched placebo, 1200/2400mg improved scores, two reached AIS E. A small, exploratory result.

A narrative review of knee injections, corticosteroids to exosomes: an RCT tying repeated triamcinolone to cartilage loss, and the evidence-maturity gap across PRP, orthobiologics, MSCs, exosomes.

Stroke, TBI, and Alzheimer's/Parkinson's differ in cause but converge on the same breakdown pathways. Why don't NGF and BDNF work in humans? A Russian team reads the delivery barrier from six trials.