MIT researchers have built an AI-guided route to mRNA vaccines that can leave the freezer behind, and the clinical stakes are plain: a dose that survives heat is a dose that can reach patients in clinics without dependable refrigeration. Their formulations held full bioactivity after more than two months at 37 degrees Celsius and stayed stable at room temperature for as long as a year, the team reports in Nature Biotechnology.
Moderna’s and Pfizer-BioNTech’s COVID-19 vaccines are the best-known products of the mRNA platform, and cold-chain logistics remain the constraint that keeps it out of reach in low-resource settings. Removing that requirement is a clinical-access problem as much as a chemistry one.
The group left the lipid nanoparticles untouched. Instead it hunted for excipient blends, ordinary pharmaceutical ingredients drawn from sugars, salts and polymers, able to shield an existing LNP formulation while it is dried and then sits in storage.
That space is far too large to search by hand, so the team built AGENT, short for Algorithm-Guided Experimental design for lipid Nanoparticle Thermostabilization, which pairs high-throughput experiments with Bayesian optimization.
From nearly 50 FDA-approved excipients, AGENT arrived at optimized formulations in six iterations over one month, work the authors say normally consumes months or years.
The approach held up on two clinically relevant LNP compositions of the kind used in the Moderna and Pfizer-BioNTech shots, the study says. Animal testing pointed the same way: in rodents and nonhuman primates the results held. The dried versions drew immune responses that were noninferior to a freshly prepared injection, the authors report. The team also delivered mRNA vaccines with solid microneedle patches.
On the economics, the authors point to earlier projections: taking the cold chain out of the picture could trim storage spending by 71% to 86%, and waste at least halve.
