About the Incretin Atlas
A living, openly licensed database of trial results for GLP-1 receptor agonists, dual and triple agonists and amylin analogues, with a primary source and a verbatim quote behind every number.
Who
The atlas is built and edited by Medical Adviser's Group (MAG), a medical-communication and evidence agency. Editor responsible: Yakov Pakhomov, MD, PhD, Medical Director. Contact: yakov@mdwrt.com (yakov@mdwrt.com).
MAG provides medical-communication and evidence services to pharmaceutical companies, including manufacturers of medicines shown in the atlas. See Independence and funding.
What the atlas is
A database of records, not a text. Each record is one trial arm, population, duration, estimand, value, primary source and verification status. The main page draws interactive maps from the database; the molecule pages and trial pages show the same records as plain tables with stable URLs, so they can be cited and linked. The whole site is regenerated from the database on every update.
Scope
- Molecules: approved and investigational incretin-based therapies (GLP-1, GIP/GLP-1, GLP-1/glucagon, GIP/GLP-1/glucagon, GLP-1 + amylin) and amylin analogues, worldwide, including China.
- Trials: phase 2–3 trials in obesity and type 2 diabetes with weight, HbA1c and tolerability data; cardiovascular, kidney, heart-failure, liver, sleep-apnoea, osteoarthritis and neurodegeneration outcome trials; registered trials from ClinicalTrials.gov.
- Regulatory status: FDA (DailyMed labels), EMA, NMPA and other national registers, reconciled at each update.
Method and update rules
Update pipeline
- Collect. PubMed (daily monitoring), ClinicalTrials.gov (weekly sweep: status changes, new trials, posted results), company press releases, congresses (EASD, ADA, ObesityWeek, ESC), regulators (FDA, EMA, national registers).
- Extract. Two independent LLM extractions under the SR Workbench protocol; every value is stored with a verbatim quote and its location in the source.
- Verify. Discrepancies between the two extractions, and every new record, go to a review queue approved by a human editor; record status moves from «seed» to «verified».
- Publish. Maps and pages are rebuilt, an entry is added to the update feed, subscribers receive an e-mail. Data snapshots are deposited on Zenodo.
Editorial rules
- A peer-reviewed publication replaces a company release or a congress abstract as soon as it appears.
- Estimands are kept separate: «efficacy» (on-treatment, hypothetical) and «treatment-policy» (intention-to-treat) values are never mixed in one comparison; the estimand is shown next to each value.
- Safety definitions are strict: «AE discontinuation» means stopping the study drug for any adverse event; narrower figures stay in the notes. A reported zero in a small arm is not proof of zero risk.
- Time point: the primary end point of each trial, or the end of randomised treatment when the paper reports it with the same estimand, so that weight and safety cover the same period.
- Post hoc analyses, meta-analyses, real-world evidence and financial news are not entered as trial records.
- Cross-trial comparisons are a guide, not proof: populations, duration, estimand and baseline BMI differ between trials.
Corrections and how to report an error
Every value links to its primary source. If you find a discrepancy, a missing trial or an outdated status, write to yakov@mdwrt.com with the page URL, the value in question and the source. Corrections from any reader, manufacturers included, go through the same review queue as new data; accepted corrections appear in the update feed with the date. Readers can also attach a paper, poster or slides through the help-wanted list.
Independence and funding
The atlas is developed by Medical Adviser's Group and funded from MAG's own resources. No pharmaceutical company funds, reviews or approves its content, and no company sees data before publication. MAG provides medical-communication and evidence services to pharmaceutical companies, including manufacturers of medicines shown here; for this reason every value carries a link to its primary source and a verbatim quote, and corrections from any reader go through the same editorial process. The atlas is not an advertisement of medicinal products; data belong to the authors of the cited works.
Data and licence
The database (CSV and JSON) is deposited on Zenodo: DOI 10.5281/zenodo.23473913 (concept DOI, all versions; latest snapshot 19 October 2026). Data are released under CC BY 4.0: reuse with attribution to Medical Adviser's Group and to the primary publications. Page text and figures may be quoted with attribution.
How to cite
BibTeX
Please cite the atlas when you reuse its figures or numbers, and cite the primary publication for any single result.
Recent updates
- 2026-11-12 — Orforglipron ATTAIN-2 on the main maps; pemvidutide in MASH (IMPACT); regulatory and pipeline updates
- 2026-11-10 — New papers: retatrutide in knee osteoarthritis (NEJM), orforglipron in sleep apnoea (JAMA), efinopegdutide in MASH (Lancet)
- 2026-11-09 — SYNCHRONIZE-CVOT: survodutide shows cardiovascular safety (Lancet)
- 2026-11-02 — New data: retatrutide vs semaglutide 2 mg in type 2 diabetes (TRANSCEND-T2D-2); orforglipron in Japan (ATTAIN-J)
- 2026-10-26 — MariTide phase 2: both cohorts; safety audit completed for every arm
- 2026-10-26 — New data: retatrutide in sleep apnoea, aleniglipron phase 2b, HRS-7535 phase 2
- 2026-10-26 — Safety data audit: strict definitions, zeros shown with their uncertainty
- 2026-10-20 — Pipeline: nine development stages brought in line with ClinicalTrials.gov
- 2026-10-19 — Mazdutide: GLORY-1, GLORY-2 and DREAMS-2 now from the publications
- 2026-10-19 — Map 6: anchored indirect comparisons from 28 trial publications
- 2026-10-19 — ACHIEVE-2: oral orforglipron versus dapagliflozin; weekly registry update
- 2026-10-14 — CagriSema vs tirzepatide (REIMAGINE 5), EloraTZP, enicepatide in T2D, BGM0504 phase 2, ACHIEVE-4 arms
Full feed with the list of changes: main page, Updates.
For search engines and language models
All pages are static HTML with tables and schema.org markup (Dataset, MedicalTrial, ScholarlyArticle). Machine-readable indexes: sitemap.xml, llms.txt (page list) and llms-full.txt (all molecule and trial records as plain text). Crawling by search engines and AI crawlers is allowed (robots.txt).