Thyroid Cancer Risk and the Next Generation of Diabetes Drugs: What Can We Learn from Incretin-Based Therapies?

Incretin-based therapies, including glucagon-like peptide-1 receptor agonists (GLP-1RAs) and dipeptidyl peptidase-4 (DPP-4) inhibitors, have transformed the management of type 2 diabetes. While their benefits for glycemic control are well established, early preclinical studies raised concerns about potential safety issues related to thyroid malignancies, particularly medullary thyroid carcinoma (MTC), highlighting the need for comprehensive epidemiological evaluation. These safety considerations extend beyond immediate clinical practice, carrying significant implications for the ongoing development and use of next-generation incretin-based therapeutic agents.

Recently, Bea et al. [1] conducted a comprehensive cohort study using the Korean National Health Insurance Database, encompassing 21,722 GLP-1RA users and 904,300 DPP-4 inhibitor users, with sodium-glucose cotransporter-2 (SGLT2) inhibitors serving as the reference comparator group. Their findings indicated that neither GLP-1RAs (hazard ratio [HR], 0.98; 95% confidence interval [CI], 0.62 to 1.53) nor DPP-4 inhibitors (HR, 0.95; 95% CI, 0.79 to 1.14) were associated with an increased risk of thyroid cancer compared to SGLT2 inhibitors. These results were robust across multiple analytical approaches, including subgroup analyses, sensitivity analyses, and specific assessments of MTC cases, thereby strengthening the validity of their conclusions.

These findings align with converging evidence from large-scale international studies across diverse healthcare systems (Table 1). The multinational investigation by Pasternak et al. [2], utilizing data from six countries, found no increased risk of thyroid cancer with GLP-1RA use (HR, 0.81; 95% CI, 0.59 to 1.12) among 98,147 GLP-1RA users and 2,488,303 DPP-4 inhibitor users. Similarly, a multicenter study by Baxter et al. [3], published in Thyroid, reported no elevated risk and additionally identified a possible protective trend among women. Brito et al. [4] analyzed United States electronic health records and demonstrated no significant difference in thyroid cancer risk between GLP-1RA and DPP-4 inhibitor users (HR, 0.96; 95% CI, 0.67 to 1.36). Most recently, a large United States claims-based analysis by Morales et al. [5] supported these safety findings, although it noted a transient elevation in risk during the initial treatment year, which the investigators attributed to heightened medical surveillance rather than a true biological effect.

Table 1.

Comparative Summary of Major Population-Based Studies Evaluating Thyroid Cancer Risk with Incretin-Based Therapies

Study	Population	GLP-1RA users	DPP-4i users	Comparator	HR (95% CI)	Median follow-up, yr
Bea et al. (2024) [1]	Korea (NHID)	21,722	904,300	SGLT2i	GLP-1RA: 0.98 (0.62–1.53); DPP-4i: 0.95 (0.79–1.14)	~3.5
Pasternak et al. (2024) [2]	Scandinavia, Taiwan, Korea	98,147	2,488,303	DPP-4i	GLP-1RA: 0.81 (0.59–1.12)	~2–3
Baxter et al. (2025) [3]	Multinational	48,325	210,250	DPP-4i	GLP-1RA: 0.85 (0.63–1.17)	~3
Brito et al. (2025) [4]	USA	28,717	120,907	DPP-4i	GLP-1RA: 0.96 (0.67–1.36)	~2.4
Morales et al. (2025) [5]	USA	460,032	2,055,583	SGLT2i	GLP-1RA: 0.83 (0.57–1.27)	~2–3
				DPP-4i	GLP-1RA: 0.78 (0.60–1.01)

GLP-1RA, glucagon-like peptide-1 receptor agonist; DPP-4i, dipeptidyl peptidase-4 inhibitor; HR, hazard ratio; CI, confidence interval; NHID, National Health Insurance Database; SGLT2i, sodium-glucose cotransporter-2 inhibitor.

A critical additional consideration is the mechanistic basis underlying these epidemiological findings, which reveals marked species-specific differences in GLP-1 receptor biology and thyroid cellular responses. Preclinical data suggesting increased thyroid cancer risk stemmed primarily from rodent studies using high, non-physiological doses of GLP-1 analogs, wherein GLP-1RAs stimulated calcitonin gene expression, calcitonin release, and C-cell hyperplasia, resulting in an elevated risk of MTC [6,7]. However, human thyroid C-cell lesions exhibit much lower GLP-1 receptor expression than those of rodents, and primary cultures of human thyroid tissues show no functional response to GLP-1RA stimulation, indicating minimal functional GLP-1 receptor activity [8]. This mechanistic disparity provides a biological rationale for the discordance between preclinical safety signals and human epidemiological data. The species-specific differences in GLP-1 receptor expression and functional responsiveness underscore the importance of interpreting preclinical safety signals within the appropriate clinical context before extrapolating to human risk.

Moreover, increased clinical surveillance and diagnostic vigilance among patients receiving GLP-1RAs and DPP-4 inhibitors—prompted by early safety concerns—may have contributed to higher detection rates of asymptomatic or clinically insignificant thyroid nodules, potentially resulting in overdiagnosis. Recognizing and addressing this potential diagnostic bias is essential for accurate interpretation of epidemiological data and for minimizing unnecessary patient anxiety and unwarranted clinical interventions.

Collectively, these large-scale, population-based analyses from diverse geographic and ethnic backgrounds reinforce the conclusion that incretin-based therapies do not confer a clinically meaningful increase in thyroid cancer risk. Importantly, the methodological rigor of these studies, including propensity score matching, extensive control for confounding factors, and thorough sensitivity analyses, increases the reliability of their findings. Nonetheless, ongoing vigilance through longer-term follow-up studies remains crucial to confirm these safety profiles over extended treatment periods, especially given the potential for surveillance bias and overdiagnosis in the setting of intensified clinical monitoring [9].

Clinicians can find reassurance in the consistency of the emerging evidence supporting the thyroid cancer safety of incretin-based therapies. The convergence of data across multiple healthcare systems, study designs, and patient populations provides substantial confidence in these safety profiles. Nevertheless, individualized risk assessment remains warranted, especially for patients with specific risk factors such as familial thyroid cancer syndromes, a personal history of thyroid disease, or genetic predispositions to endocrine malignancies. Clinical decision-making should incorporate these patient-specific factors alongside the broader population-level safety evidence.

Finally, these findings highlight broader implications for drug development and regulatory oversight in the rapidly evolving field of diabetes therapeutics. As next-generation diabetes medications are developed, preclinical oncologic signals, especially those derived from rodent models, must be evaluated in light of human-specific biological plausibility and validated through high-quality, real-world evidence. The integration of long-term cancer surveillance, comprehensive post-marketing safety monitoring, and translational studies bridging laboratory findings with clinical outcomes will be essential to ensure the safe and responsible advancement of metabolic disease therapeutics.