Leuprolide and goserelin carry different adverse-event signatures, 12M reports show

Leuprolide and goserelin are both GnRH agonists, used to suppress ovarian function and to treat hormone-dependent cancers, and are often treated as interchangeable within their drug class. A pharmacovigilance study published this month in Medicine used two decades of FDA safety reporting data to test whether their real-world adverse-event patterns actually match [s1].

The design

Researchers extracted and standardized reports from the FDA Adverse Event Reporting System (FAERS) from the first quarter of 2004 through the third quarter of 2024 [s1]. Out of more than 12.4 million total adverse event reports in that window, 64,324 were leuprolide-related and 4,253 were goserelin-related [s1]. Signal detection — the statistical process used to flag whether a drug is reported alongside a particular adverse event more often than would be expected by chance — was performed using four separate methods: the reporting odds ratio, proportional reporting ratio, chi-square test, and Bayesian confidence propagation neural network [s1]. A signal was considered significant if it met specific thresholds across these methods, following STROBE reporting guidelines for observational studies [s1].

What it found

Both drugs showed strong safety signals for expected, mechanism-related effects: "prostatic specific antigen increased" (leuprolide chi-square = 84,009.65; goserelin chi-square = 820.31), "blood testosterone increased" (leuprolide chi-square = 22,845.52; goserelin chi-square = 290.23), and "prostatic specific antigen abnormal" (leuprolide chi-square = 51,615.98; goserelin chi-square = 6.46 information component) [s1]. Both drugs also showed consistent signals for "prostate cancer metastatic" and "hot flush" [s1].

Where the drugs diverged was in the broader pattern of reported adverse events. Leuprolide showed stronger associations with events in the "reproductive system and breast disorders" category, "neoplastic benign and malignant conditions," and procedure-related categories [s1]. Goserelin, by contrast, was more strongly linked to "endocrine disorders" [s1].

Reading a pharmacovigilance study correctly

The scale of this dataset — more than 12 million total reports, with tens of thousands specific to each drug — is a genuine strength for detecting patterns that smaller studies would miss. But pharmacovigilance databases like FAERS have a structural limitation that applies regardless of size: reports are submitted voluntarily by clinicians, patients, and manufacturers, not systematically collected from everyone taking the drug, so the data can't establish true incidence rates or prove causation — only that a reported event occurs more often alongside a given drug than statistical expectation would predict. The far larger volume of leuprolide reports (64,324 versus 4,253 for goserelin) likely partly reflects leuprolide's more widespread use and longer market history rather than a difference in underlying risk alone, which is part of why the analysis relies on disproportionality statistics — designed to account for reporting volume — rather than raw counts.

Why the distinction between the two drugs matters clinically

Because leuprolide and goserelin are frequently treated as interchangeable within the GnRH agonist class, the finding that their real-world safety signal patterns diverge — leuprolide toward reproductive, breast, and neoplastic-condition reports, goserelin toward endocrine ones — offers clinically relevant differentiation that class-level prescribing guidance doesn't typically capture [s1]. The study's authors describe this as supporting "individualized risk monitoring" between the two drugs rather than treating a GnRH agonist safety profile as uniform [s1].

What this doesn't establish

This is a retrospective disproportionality analysis of spontaneous reports, not a controlled comparison — patients weren't randomized to either drug, and confounding by indication (why a particular patient received leuprolide versus goserelin in the first place) cannot be ruled out. The analysis cannot establish which drug is safer overall, only that their reported adverse-event patterns differ in specific categories.

What to watch

Whether these signal differences are corroborated by prospective comparative studies, and whether they translate into any change in prescribing guidance between the two drugs for specific patient populations. This article is not medical advice.