The Next GLP-1 Wave: Oral Pills, Dual Agonists, and Nanofiber Depots

The biohacker reflex is to treat each new GLP-1 headline as a product announcement. The more useful frame is mechanistic: each candidate is a different bet on how to make a finicky 30-amino-acid hormone behave like a drug. Semaglutide solved the half-life problem with a fatty-acid tether. Tirzepatide added a second receptor. The candidates surveyed here push further — shrinking the molecule, extending the depot, and stress-testing the class outside its original Western trial populations.

None of this is settled. The evidence rating across the pipeline is best read as moderate: signals are coherent and the mechanisms plausible, but several of the most interesting datasets are early-phase, preclinical, or retrospective. That is exactly the stage at which careful reporting matters more than enthusiasm.

Peptides hate being swallowed. The gut treats them like lunch. That is why the first oral semaglutide formulation leans on an absorption enhancer and a strict empty-stomach protocol, and why most of the field has stayed injectable. A small molecule that hits the GLP-1 receptor with peptide-like potency would change the calculus entirely.

That is the bet behind AZD5004 (ECC5004), profiled in non-clinical models and a first-in-human study published in Diabetes, Obesity and Metabolism. In cell assays, the compound bound the human GLP-1 receptor with an IC50 of 2.4 nM and triggered cAMP signaling without recruiting β-arrestin-2 — a biased-agonism profile that some researchers believe may track with better tolerability. In a phase 1 trial of healthy volunteers (single doses 1–300 mg) and type 2 diabetes patients (28 days of daily dosing at 5, 10, 30, and 50 mg), the molecule was well tolerated with no serious adverse events, and produced dose-dependent reductions in glucose and body weight.

The caveats are the usual ones for phase 1: small numbers, short duration, healthy-volunteer dominance. But the readout matters because the bottleneck for oral peptides has always been bioavailability, and a small molecule routes around that problem entirely.

Tirzepatide's pitch has always been mechanistic: agonize GLP-1 and GIP simultaneously and you get larger weight-loss numbers than GLP-1 monotherapy. The marketing got ahead of the data for a while. The data is catching up.

A 2024 systematic review in Current Drug Safety, following PRISMA methodology across PubMed, Embase, and Cochrane, pulled together seven phase 3 trials of tirzepatide. The authors report significant, dose-dependent reductions in body weight and improvements in metabolic markers across the included studies, with the drug well tolerated and hypoglycemia rates low. Subgroups with higher baseline BMI saw greater absolute benefit — a pattern consistent with the rest of the class.

Two honest qualifiers. First, a systematic review inherits the limits of its inputs, and phase 3 obesity trials are not famous for long follow-up. The authors themselves call for long-term studies. Second, head-to-head durability against semaglutide outside of SURMOUNT-style endpoints remains an open question. But for a dual-agonist concept that not long ago lived mostly in slide decks, seven phase 3 readouts is a meaningful base of evidence.

One of the loudest critiques of the GLP-1 evidence base is that pivotal trials skewed toward heavier Western patients. The leaner metabolic phenotype common in East Asian populations — with different insulin-secretion dynamics and a lower BMI distribution — was underrepresented. Whether the drugs perform similarly there is a question real-world data is starting to answer.

A retrospective propensity-score-matched analysis from Keio University School of Medicine, published in Diabetology International, compared 313 Japanese type 2 diabetes patients on oral semaglutide with 11,239 untreated controls. At 180 days, the semaglutide group showed significantly better HbA1c reduction and weight loss, along with improvements in systolic blood pressure, LDL cholesterol, and liver function. In the subgroup who achieved ≥3% weight loss, HbA1c improvements were superior to controls.

Retrospective data is retrospective data — residual confounding survives even careful matching, and a single-center cohort is not a global verdict. Still, the directionality is consistent with the trial evidence, and it begins to address the phenotype gap the original studies left open.

The most futuristic entry in this survey is also the earliest. Researchers reporting in ACS Nano describe an engineered supramolecular nanofiber depot built by fusing a self-assembling peptide motif onto a GLP-1 receptor agonist. The hybrid molecule forms a hydrogel at the injection site and releases drug slowly as the nanofibers disassemble.

In vitro, the depot sustained release for multiple weeks. In a rat model of type 2 diabetes, a single injection produced detectable serum drug concentrations for at least 40 days, reduced blood glucose, and limited weight gain — performance the authors describe as comparing favorably to daily semaglutide dosing in the same model. The platform is modular: in principle, the same prosthetic motif could be appended to other peptide therapeutics.

This is preclinical. Rat pharmacokinetics do not translate cleanly to human dosing intervals, immunogenicity for self-assembling peptide constructs is its own chapter, and no IND has been announced from the supplied evidence. But if even a fraction of the duration carries forward, the meaningful unit of GLP-1 therapy could shift from a week to a month or a season.

Put the four candidates side by side and a rough map emerges. The near term belongs to incremental wins — better real-world evidence in populations the original trials underweighted, and continued accumulation of phase 3 data on dual agonism. The mid term may belong to the small-molecule oral, which would reframe GLP-1 therapy as something closer to a daily statin than a weekly ritual. The long term, if the chemistry holds, may belong to depots that turn weekly injections into a quarterly appointment.

None of those futures is guaranteed. All of them are now legible enough in the literature to be worth tracking carefully — and skeptically — as the next readouts arrive.