GLP-1 Safety and Physiology: Thyroid Signal, Cancer Questions, Testosterone and the Weight-Loss Plateau
GLP-1 safety and physiology in research: the thyroid C-cell signal, cancer questions, the testosterone/HPG axis and why weight loss plateaus.
TL;DR: what the class-level evidence actually shows
The rodent thyroid C-cell signal behind the FDA boxed warning has not been resolved in humans: some large cohorts find no excess risk, others (including the two largest RCT meta-analyses) still detect one. Pancreatic cancer risk is consistently not elevated across meta-analyses with tens of thousands of patients. Testosterone, LH and FSH rise in obese hypogonadal men on GLP-1 receptor agonists, but the mechanism looks indirect (weight loss relieving the axis), not a direct hormonal action of the drug. The weight-loss plateau seen around 12-18 months is a measurable metabolic-adaptation effect (falling energy expenditure), not simply "the drug stopped working." Every finding below is class-level research data with a cited PMID, not treatment advice; nothing here is a human dosing instruction.
The GLP-1 receptor agonist class, from early exenatide through liraglutide, dulaglutide, semaglutide and the dual GIP/GLP-1 agonist tirzepatide, is now one of the most heavily studied drug families in metabolic research. That volume of research has produced two parallel narratives that rarely get reconciled in one place: a genuinely reassuring safety record on most endpoints, and a handful of unresolved signals that keep resurfacing in peer-reviewed meta-analyses. This article walks through five specific questions that come up constantly in the research literature: the thyroid C-cell / MTC signal, the broader cancer picture, the testosterone and HPG-axis data, why weight loss plateaus, and the chemistry that turned a 2-minute hormone into a once-weekly injectable. Every claim below is sourced to a PMID-indexed study; where the evidence is mixed, both sides are shown rather than the more comfortable one.
This is a research-literature summary for laboratory and research contexts. Nothing in this article is a treatment recommendation, a claim of disease cure, or a human dosing instruction. All products referenced are sold by PeptidesDirect strictly as research material, not for human consumption.
The thyroid C-cell signal: what the boxed warning is actually based on
Every major GLP-1 receptor agonist approved in the US and EU (semaglutide, liraglutide, dulaglutide, exenatide, and the dual-agonist tirzepatide) carries a class-wide contraindication for patients with a personal or family history of medullary thyroid carcinoma (MTC) or Multiple Endocrine Neoplasia type 2 (MEN2). In the US this appears as an FDA boxed warning; in the EU the equivalent restriction sits in the Summary of Product Characteristics (sections 4.3/4.4). The substance of the warning is identical on both sides of the Atlantic: screen for MTC/MEN2 history, monitor for thyroid symptoms.
The warning did not originate from a human trial. It came from rodent carcinogenicity studies, the two-year studies the FDA requires before approval, which showed that sustained pharmacologic GLP-1 receptor activation on thyroid parafollicular (C) cells drives C-cell hyperplasia progressing to adenoma and, at high and prolonged exposure, medullary thyroid carcinoma, in a dose- and duration-dependent pattern. The honest complication is that human relevance is genuinely uncertain: human C-cells express the GLP-1 receptor at far lower density than rodent C-cells, and the human thyroid contains proportionally far fewer C-cells than a rodent thyroid. The FDA label itself states that human relevance "has not been determined." That single sentence is the reason this question keeps generating new studies instead of settling.
The human evidence is split, not settled
A French national case-control study (Bezin et al., Diabetes Care 2023, PMID 36356111) analyzed 2,562 thyroid-cancer cases matched to 45,184 controls from 2006-2018 insurance data and found GLP-1RA use for 1-3 years was associated with an adjusted HR of 1.58 for thyroid cancer overall and 1.78 for MTC specifically. A 6-country cohort (Baxter et al., Thyroid 2025, PMID 39772758), 98,147 GLP-1RA users versus 2,488,303 DPP-4 inhibitor users, found a pooled HR of 0.81 (95 percent CI 0.59-1.12), no increased risk, but with median follow-up of only 1.8-3.0 years. Two independent RCT meta-analyses by the same research group reached the same direction twice: Silverii et al. 2024 (Diabetes Obes Metab, PMID 38018310, 26 trials with events) found a fixed-effect odds ratio of 1.52 (95 percent CI 1.01-2.29) that lost significance under random-effects modeling (OR 1.41, CI 0.91-2.17); the larger, more recent Silverii et al. 2025 analysis (50 trials, PMID 40437949) found OR 1.55 (CI 1.05-2.27), described as more pronounced in longer trials. A pharmacovigilance disproportionality study combining SEER registry data with FDA Adverse Event Reporting System reports (D'Arcy et al., Head Neck 2026, PMID 42003032) found a 50.7-fold elevated MTC reporting ratio, but this design is prone to major reporting and detection bias and is explicitly not an incidence estimate.
The most defensible summary, echoed in a 2023 narrative review (Lisco et al., Endocr Connect, PMID 37656509), is that a causal link remains speculative rather than confirmed or excluded. Current clinical guidance leans toward ordinary monitoring, personal/family MTC or MEN2 history screening and attention to thyroid nodules or neck symptoms, rather than routine calcitonin or ultrasound screening in unselected populations. A recurring limitation across the reassuring cohort studies is follow-up length: MTC is slow-growing and can take years to decades to become clinically apparent, and 1.8-3.0 years of observation may simply be too short a window to rule out a modest excess risk.
The broader cancer picture: mostly reassuring, with one unresolved outlier
Zooming out from the thyroid-specific question, the largest and most recent RCT meta-analysis of the class (Silverii et al., Diabetes Obes Metab 2025, PMID 40437949, 50 trials) found overall cancer risk was not elevated (Mantel-Haenszel OR 1.05, 95 percent CI 0.98-1.13). That is a meaningfully reassuring top-line number. But the same analysis, looking at cancer sites individually, found colorectal cancer risk elevated (OR 1.27, CI 1.03-1.57, more pronounced in longer trials) and uterine/endometrial cancer risk reduced in obesity trials (OR 0.24, CI 0.06-0.94), plausibly because weight loss reduces adipose aromatization of androgens to estrogen, lowering estrogen-driven endometrial proliferation.
Pancreatic cancer is the one site where the literature is genuinely convergent and reassuring. A 2019 meta-analysis with trial sequential analysis (Pinto et al., Sci Rep, PMID 30787365), pooling 12 RCTs and 36,397 patients over a mean 1.74 years, found no increased risk (OR 1.06, 95 percent CI 0.67-1.67) and the sequential analysis excluded a number-needed-to-harm below 1,000. A much larger 2026 analysis (Wali et al., Front Pharmacol, PMID 41743116), 93 RCTs and roughly 1.85 million participants, found overall gastrointestinal cancer risk was not elevated (HR 0.81, CI 0.68-0.96) with site-specific hazard ratios of 0.78 for pancreatic, 0.81 for colorectal, 0.74 for liver, and 0.85 for gastric cancer.
An honest tension the literature has not resolved
Silverii 2025 found colorectal cancer risk elevated (OR 1.27) in RCTs. Wali 2026 found colorectal cancer risk reduced (HR 0.81) in a much larger pooled dataset. Both are legitimate, recent, peer-reviewed meta-analyses looking at the same question with different trial sets, outcome ascertainment methods, and follow-up windows. This is a genuine open question in the literature, not a case where one number is "correct" and the other is outdated. A further limitation applies across nearly all of these studies: cancer was almost always a secondary or exploratory endpoint in the source trials, not a primary outcome the trials were statistically powered to detect.
Testosterone and the HPG axis: relief of suppression, not direct stimulation
A separate line of research asks whether GLP-1 receptor agonists affect the male hypothalamic-pituitary-gonadal (HPG) axis, since obesity itself is a well-documented suppressor of that axis. Obesity lowers hepatic SHBG production via hyperinsulinemia, increases adipose-tissue aromatization of testosterone to estradiol (which centrally suppresses GnRH/LH pulsatility), and blunts hypothalamic GnRH drive through leptin resistance and inflammatory signaling. Weight loss, by whatever mechanism achieves it, tends to reverse these effects.
Jensterle et al. (Endocr Connect 2019, PMID 30707677) ran a 16-week randomized trial in 30 men with obesity (mean BMI 41.2) and functional hypogonadism, comparing liraglutide 3.0 mg/day against transdermal testosterone gel. The liraglutide arm lost 7.9 plus/minus 3.8 kg (versus 0.9 plus/minus 4.5 kg on TRT) and total testosterone rose significantly (plus 2.6 plus/minus 3.5 nmol/L), alongside significant increases in both LH and FSH, meaning the gonadotropin axis was being released from suppression, not shut down further. Both study arms saw improved sexual function.
Izzi-Engbeaya et al. (J Clin Endocrinol Metab 2020, PMID 32052032) tested the opposite scenario: an 8-hour IV GLP-1 infusion in 18 healthy, lean men with normal gonadal function, no weight loss involved. Despite a 15 percent acute reduction in food intake (P equals 0.01), LH pulse frequency, LH area-under-curve, FSH, and testosterone were all unchanged. Taken together, these two studies point toward an indirect mechanism: GLP-1 receptor agonism does not appear to directly stimulate the hypothalamic-pituitary axis in men with a normally functioning system; the testosterone increases seen in obese hypogonadal men look like a downstream consequence of weight loss and improved insulin sensitivity, not a direct endocrine action of the drug class. Whether there is also some direct central effect specifically in obese or insulin-resistant states remains an open research question rather than a settled one.
A first-in-class dual GIP and GLP-1 receptor agonist, and one of the most extensively studied compounds in modern metabolic and weight-regulation research. Supplied as a lyophilised research peptide with a per-batch certificate of analysis, for laboratory and in-vitro use only.
First-ever triple-action weight management peptide targeting three receptors at once: GLP-1, GIP, and glucagon. Shown exceptional results in Phase 2 trials - up to 24% weight reduction. The most advanced metabolic peptide available.
Long-acting amylin analog studied for once-weekly satiety and appetite control. Phase 3 REDEFINE trials complete, NDA filed with FDA December 2025. A mechanism distinct from GLP-1 agonists.
Why weight loss plateaus around 12-18 months
One of the most consistent observations in longer GLP-1 trials is that weight loss decelerates and then plateaus, typically somewhere between 12 and 18 months, even with continued treatment. This is not a new phenomenon specific to GLP-1 agonists: it is a well-described feature of any sustained weight-loss intervention, formalized decades ago from the Minnesota semistarvation experiment data (Kozusko, Bull Math Biol 2001, PMID 11276532) as the "body weight setpoint" model, where the body reduces relative energy expenditure as weight falls, which mechanically slows further loss over time regardless of continued caloric restriction.
A 2016 review (Muller et al., Curr Obes Rep, PMID 27739007) quantifies this "adaptive thermogenesis": total energy expenditure falls roughly 10-20 percent below what is predicted purely from the loss of fat and fat-free mass. Mechanistically there appear to be two components with different timing: reduced resting energy expenditure early in a diet (tied to falling insulin and glycogen), and reduced non-resting, i.e. activity-related, energy expenditure later (tied to falling leptin and to improved skeletal-muscle mechanical efficiency, meaning the same physical activity simply burns fewer calories once a person is lighter and more efficient).
A 2025 mechanistic addition specific to GLP-1 treatment
A 2025 perspective piece (Wang et al., Cell Rep Med, PMID 40961927) proposes that GLP-1 receptor agonist treatment compounds ordinary adaptive thermogenesis in two ways: weight loss itself increases skeletal-muscle mechanical efficiency by roughly 25 percent per 10 percent of body weight lost, and GLP-1 receptor agonist treatment appears to independently reduce non-resting energy expenditure by an estimated 170 kcal per day, even when body weight has barely changed yet. The combination, a shrinking caloric deficit as the drug effect stabilizes plus a shrinking energy requirement from adaptive efficiency, is proposed as the mechanistic explanation for why the plateau appears where it does, and why weight tends to regain quickly once treatment stops.
Modified hGH fragment (177-191) studied for fat metabolism and lipolysis research. Interacts with beta-3 adrenergic receptors without growth-promoting effects.
GIP/GLP-1/Glucagon agonists and metabolic pathways
Half-life engineering: from a 2-minute hormone to a once-weekly injectable
None of the pharmacology above would be researchable as a once-weekly compound without a specific chemistry solution. Native GLP-1(7-36)amide has an intravenous half-life of only about 1.5-2 minutes: it is cleaved almost immediately by dipeptidyl peptidase-4 (DPP-4) and cleared renally. The foundational medicinal-chemistry work (Knudsen et al., J Med Chem 2000, PMID 10794683) established that attaching a fatty acid of at least C12 chain length to the peptide backbone confers strong, reversible albumin binding and markedly protracts its action, the acylation strategy that underlies the entire modern class.
Liraglutide uses a single C16 (palmitic) fatty acid attached via a gamma-glutamate linker, combined with an Arg34Lys substitution, giving a half-life of roughly 13 hours, compatible with once-daily dosing (Knudsen and Lau, Front Endocrinol 2019, PMID 31031702). Semaglutide goes further: a C18 fatty diacid attached through an AEEA-based hydrophilic linker at Lys26, combined with an Aib8 substitution for additional DPP-4 resistance (Lau et al., J Med Chem 2015, PMID 26308095). The resulting half-life is approximately 7 hours in rat, around 75 hours subcutaneously in minipig, and translates to roughly 165-184 hours, about one week, in humans, which is what makes once-weekly subcutaneous dosing possible (Knudsen and Lau, PMID 31031702). A 13-week trial in 44 healthy men (Kapitza et al., Adv Ther 2018, PMID 29536338) confirmed this week-long elimination profile and found near-identical pharmacokinetics, pharmacodynamics, and safety between Japanese and Caucasian subjects, meaning no ethnicity-based dose adjustment was indicated in that study population.
The engineering, not the hormone, is what changed
The extended duration of action across this drug class is an engineered property, reversible albumin binding via fatty-acid acylation plus protease-resistance substitutions, not an intrinsic feature of the native GLP-1 hormone. This is a useful distinction when reading marketing language: "once-weekly" describes the chemistry attached to the peptide, not some inherent long-acting quality of GLP-1 signaling itself.
Thyroid and cancer safety literature
Dual and triple receptor agonism research
Frequently asked questions
This article summarizes class-level peer-reviewed research literature for laboratory and research purposes only. It is not medical advice, not a treatment recommendation, and contains no human dosing instructions. All PeptidesDirect products referenced are sold exclusively as research material.
Research context for English-speaking buyers
Most of our English-speaking customers ship to the UK, Ireland, Malta or other English-as-second-language EU territories. The regulatory picture differs per country.
- Relevant authorities
- MHRA (UK, post-Brexit), HPRA (Ireland, EU-aligned), FDA Section 503A bulks list (US, restricted Cat 2 status of several peptides as of 2026)
- Customs and VAT
- EU shipments include 19% VAT; UK shipments after Brexit are now extra-EU and may attract UK VAT plus a handling fee at import
- Typical shipping window
- EU 2-4 working days, UK 4-7 working days, other international 7-14 working days, depending on customs
Research-grade peptides shipped from our EU warehouse are sold for laboratory use only and are not authorised for human or veterinary therapeutic application in any of the destination jurisdictions. US customers should be aware that the FDA Section 503A bulks list classification (and the April 2026 reclassification of twelve compounds) only governs compounding pharmacies, not direct-to-researcher imports for non-clinical work. UK buyers should declare the consignment on import and may be asked for a research justification by HMRC. We provide a CoA per batch identified by colour code rather than serial number; customs sometimes asks for this document when clearing the parcel.