A large cohort study using nationwide data from Denmark, Norway, and Sweden found that glucagon-like peptide 1 receptor agonist treatment was not associated with a substantially increased risk of thyroid cancer over a mean follow-up of 3.9 years.
The study, published in BMJ, used an active-comparator new-user design to minimize risks of confounding and time-related biases. Patients aged 18 to 84 years who were new users of glucagon-like peptide 1 (GLP1) receptor agonists or dipeptidyl peptidase 4 (DPP4) inhibitors were eligible for inclusion. Patients were excluded if they had thyroid cancer at any time before cohort entry, any other cancer in the previous year, end-stage illness, human immunodeficiency virus infection, drug or alcohol misuse in the previous year, major pancreatic disease, genetic syndromes associated with thyroid cancer, and no health care contact in the previous year.
The primary outcome of the study was thyroid cancer identified from nationwide cancer registers. Cox regression was used to estimate hazard ratios, controlling for potential confounders with propensity score weighting. The propensity score was estimated using logistic regression as the probability of starting a GLP1 receptor agonist versus a DPP4 inhibitor, conditional on a number of variables, including sociodemographic characteristics, medical history, other diabetes treatment, and markers of health care use.
The most common individual GLP1 receptor agonist used by patients in the analysis was liraglutide (57.3%), followed by semaglutide (32.9%), dulaglutide (4.9%), exenatide (4.1%), and lixisenatide (0.9%). The mean follow-up time among patients who started GLP1 receptor agonists was 3.9 years (standard deviation, 3.5 years), with 25% of patients followed for 6.1 years or longer. The mean follow-up time varied by specific drug: liraglutide, 5.2 years; semaglutide, 1.1 years; dulaglutide, 2.9 years; exenatide, 9.1 years; and lixisenatide, 4.0 years.
The study included 145,410 patients receiving GLP1 receptor agonists and 291,667 patients taking DPP4 inhibitors. The incidence rates of thyroid cancer were 1.33 and 1.46 per 10,000 person-years in the GLP1 receptor agonist and DPP4 inhibitor groups, respectively. The hazard ratio (HR) for thyroid cancer with GLP1 receptor agonist use was 0.93 (95% confidence interval [CI], 0.66-1.31), and the rate difference was −0.13 (95% CI, −0.61 to 0.36) events per 10,000 person-years.
In additional analyses, no significant increases in risk of any thyroid cancer subtypes, including papillary (HR, 0.92; 95% CI, 0.61-1.39), follicular (HR, 0.99; 95% CI, 0.47-2.08), medullary (HR, 1.19; 95% CI, 0.37-3.86), and other types (HR, 1.51; 95% CI, 0.44-5.20), were identified with GLP1 receptor agonist use. However, the number of events for thyroid cancer subtypes other than papillary was relatively small, resulting in imprecise estimates.
The study's findings were robust in several additional analyses, including when an alternative comparator group of sodium-glucose cotransporter 2 (SGLT2) inhibitors was used. In this analysis, the HR for thyroid cancer comparing GLP1 receptor agonists with SGLT2 inhibitors was 1.16 (95% CI, 0.65-2.05).
The authors concluded that while the study cannot exclude a small increase in risk, the upper limit of the CI in the main analysis comparing GLP1 receptor agonists with DPP4 inhibitors was consistent with no more than a 31% increase in relative risk, which translates to no more than 0.36 excess events per 10,000 person-years.
Full statement about competing interests available in the study.
