Abstract
Objective
This study aimed to evaluate the association between hormonal therapies and bone adverse events using the Food & Drug Administration Adverse Event Reporting System (FAERS) and to explore possible molecular mechanisms.
Methods
FAERS data were analyzed for adverse events related to five hormonal therapy drug categories, and disproportionality analysis was used to identify significant adverse events. Transcriptomic data from Gene Expression Omnibus datasets (GSE147271 and GSE20181) were analyzed to identify bone-related pathways and differentially expressed genes.
Results
Overall, 57 significant bone-related signals, including 22 Important Medical Events, were identified, most commonly fractures at various sites, osteoporosis, and bone metastases associated with estrogen receptor-targeted drugs and aromatase inhibitors. Estrogen-related adverse events typically occurred after 6 months, whereas androgen-related events appeared earlier. Transcriptomic analysis identified FOS, JUN, COL1A1, and IGF1 as key genes, implicating the Janus kinase signaling pathway in bone injury.
Conclusion
This study demonstrates a strong association between hormonal therapy drugs and bone-related adverse events, particularly fractures and bone cancers. It emphasizes the importance of monitoring bone health and suggests the Janus kinase signaling pathway as a potential therapeutic target for mitigating bone-related adverse events.
Keywords
Introduction
Hormonal therapy targets hormone pathways to treat conditions driven by hormonal imbalances, including thyroid dysfunction, polycystic ovary syndrome (PCOS), and menopausal symptoms, offering significant potential in personalized medicine.1–3 In oncology, hormonal therapy is pivotal in managing hormone-sensitive malignancies, such as breast and prostate cancers, by modulating hormone activity to inhibit tumor progression and reduce recurrence.
Breast cancer is the most common malignancy among women worldwide, and surgery remains the primary treatment modality, often requiring combination with systemic therapies. 4 Estrogen significantly contributes to breast cancer pathophysiology by promoting cellular proliferation, migration, and metabolic adaptation in low-glucose environments. 5 Additionally, high estrogen activity is correlated with reduced immune surveillance, suggesting its role in fostering an immunosuppressive tumor microenvironment. 6 Prostate cancer was the second most commonly diagnosed cancer among men in 2022, following lung cancer. 7 Its progression depends on androgen receptor signaling, with androgens such as testosterone and dihydrotestosterone (DHT) promoting prostate cancer cell growth and proliferation. 8
Given the integral role of sex hormones in these cancers, hormonal regulation therapies are widely utilized. Breast cancer treatment primarily employs two drug classes: estrogen receptor-targeted agents and aromatase inhibitors. Estrogen receptor-targeted agents include selective estrogen receptor modulators (SERMs), such as tamoxifen, which competitively bind to estrogen receptors, 9 and selective estrogen receptor degraders (SERDs), such as fulvestrant, which promote receptor degradation. 10 Aromatase inhibitors, such as letrozole, reduce estrogen synthesis by inhibiting aromatase activity. 11 Prostate cancer therapy includes androgen receptor antagonists, which competitively inhibit receptor activity, and abiraterone, which suppresses androgen biosynthesis by targeting CYP17A1.12,13 Second-generation androgen receptor antagonists, such as enzalutamide and darolutamide, exhibit higher receptor affinity than earlier generations. 12
Moreover, the synthesis of androgens and estrogens is regulated by the hypothalamic–pituitary–gonadal axis, making the targeting of gonadotropin-releasing hormone (GnRH) the most prevalent therapeutic approach currently employed. GnRH agonists, such as leuprolide, initially cause a transient rise in sex hormone levels. However, long-term treatment desensitizes GnRH receptors, leading to suppressed hormone production. 14 GnRH antagonists, such as degarelix, are primarily used in prostate cancer treatment to directly block GnRH receptors, rapidly reducing testosterone levels. 15
Hormonal therapy drugs targeting estrogen or androgen pathways can cause a wide range of adverse events (AEs), including cardiovascular complications, gastrointestinal symptoms, neurological impairments, and musculoskeletal issues.16–18 Bone-related AEs are particularly notable with hormonal therapy, chemotherapy, and radiotherapy. Considering that patients with breast and prostate cancer are predominantly older adults,19,20 bone-related AEs, although rarely fatal, significantly affect quality of life. Thus, comprehensively summarizing these events and identifying strategies to mitigate their impact remain essential.
This study analyzes bone-related AEs in the Food & Drug Administration (FDA) Adverse Event Reporting System (FAERS) using demographic, disproportionality, and onset-time analyses to compare risks across hormonal therapy drugs. Omics data from patients treated with estrogen-related drugs were further used for gene enrichment and differential expression analyses to explore molecular mechanisms. The findings provide insights into hormonal therapy-induced AEs, support improved treatments, and inform future clinical practice.
Methods
Cases and drugs included
All data were sourced from the FAERS database, a centralized, computerized public repository utilized by the FDA and pharmacovigilance experts to collect and monitor reports of AEs associated with drugs and biological products. The data span Quarter 1 (Q1) of 2004 to Q2 of 2024. Following FDA protocols, we performed deduplication using the PRIMARYID, CASEID, and FDA_DT fields from the DEMO file. Notably, as a spontaneous reporting system, the FAERS database is subject to reporting bias, underreporting, incomplete records, and potential confounding, which may affect signal detection.
Standardized drug names were obtained from the FDA. Using these names, we extracted relevant search terms for estrogen receptor-targeted drugs, aromatase inhibitors, androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists from the DRUGNAME field in the FAERS database's DRUG file (Table S1). Only reports in which the drug was identified as the “Primary Suspect” (PS) were included.
Classification of AEs
The Medical Dictionary for Regulatory Activities (MedDRA) is a standardized medical terminology system with a five-level hierarchical structure. This study utilized two of these levels: High-Level Group Term (HLGT), to identify the affected anatomical systems, and Preferred Term (PT), to provide detailed descriptions of specific AEs. Disproportionality analysis was applied at both levels to assess potential associations between AEs and the use of five hormonal therapy drug categories.
This study employed four commonly used disproportionality analysis methods—reporting odds ratio (ROR), proportional reporting ratio (PRR), Bayesian confidence propagation neural network, and empirical Bayesian geometric mean (EBGM)—to detect potential drug–AE associations (Table S2). Only AEs with positive results across all four methods were included.
Positive PTs were further categorized based on their classification as Important Medical Events (IMEs) to evaluate clinical significance. IMEs are AEs that, while not immediately life-threatening or associated with severe outcomes such as hospitalization or death, warrant attention because of their potential to signal serious underlying risks. This study utilized the European Medicines Agency (EMA) IME list (version 27.0) to categorize reported AEs.
Case-by-case analysis
We conducted post-hoc case-by-case analyses of 326, 3448, and 88 patients using one of the five hormonal therapy drug categories as the PS drug and experiencing bone cancer, bone metastases, or metastatic bone cancer, respectively, after excluding other potential causes. Basic demographics, concomitant medications, and indications were summarized. A similar analysis was performed for 110 patients with pathological fractures.
Analysis of transcriptome data
The analysis utilized two datasets from the Gene Expression Omnibus (GEO) database: GSE147271 and GSE20181. GSE147271 included 102 samples from core needle biopsies of breast cancer tumors, comprising 62 pretreatment and 40 post-treatment samples following tamoxifen (an estrogen receptor-targeted drug) administration. 21 GSE20181 documented gene expression in breast tumor biopsies from patients with breast cancer treated with letrozole (an aromatase inhibitor), with 58 pretreatment and 60 post-treatment samples. 22
Differentially expressed genes (DEGs) were visualized using volcano plots. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses identified key biological processes and functions. A protein–protein interaction (PPI) network was constructed to examine interactions among the encoded proteins. Additionally, 30 bone-related DEGs were analyzed to investigate potential medication-induced bone-related changes and explore underlying transcriptomic mechanisms. Data processing and statistical analyses were performed using R (version 4.3.3). ChatGPT (OpenAI) was used solely for language polishing to improve readability and grammar. It was not used for data analysis, interpretation, or generation of scientific content.
Results
Descriptive analysis
Figure S1 illustrates the data filtering process based on the FAERS database. The baseline characteristics of patients using the five hormonal therapy drug categories from Q1 2004 to Q2 2024 are summarized in Table 1, with patient counts of 22,446, 41,217, 59,544, 32,003, and 84,566, respectively. Females predominated among users of estrogen receptor-targeted drugs and aromatase inhibitors, whereas males constituted the majority among users of androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists (64.6%). Notably, some males used estrogen-related drugs, including estrogen receptor-targeted drugs and aromatase inhibitors, whereas some females used androgen-related drugs, including androgen receptor antagonists and abiraterone. This may reflect potential uses such as boosting testosterone in men with late-onset hypogonadism or blocking androgen signaling in women with conditions such as PCOS and breast cancer.23–25
Characteristics of reports associated with hormonal therapy drugs from 2004Q1 to 2024Q2.
Reporter: healthcare professionals including reporters such as physicians and pharmacists; nonhealthcare professionals including reporters such as consumer and lawyer.
In terms of age distribution, users of the two estrogen-related drug categories were evenly distributed between the 18 and 65 (27.7%, N = 6228, and 29.7%, N = 12,237) and 65–85 (24.6%, N = 5528, and 29.9%, N = 12,314) age groups. In contrast, the other three drug categories, including androgen-related drugs and GnRH agonists and antagonists, had a higher proportion of users aged 65–85 years (40.0%, N = 23,815; 39.1%, N = 12,523; and 26.6%, N = 22,454, respectively). Additionally, it is noteworthy that death was a significant outcome among users of all five hormonal therapy drug categories, with proportions of 10.2% (N = 2279), 7.2% (N = 2963), 20.0% (N = 11,884), 21.5% (N = 6891), and 16.3% (N = 13,809), respectively.
Figure 1(a) shows AE reports for hormonal therapy drugs from Q1 2004 to Q2 2024. Reports were relatively stable before 2013 and then rose sharply during 2013–2018, with androgen receptor antagonists increasing most rapidly and peaking in 2017. After 2018, reports for estrogen receptor-targeted drugs, aromatase inhibitors, and GnRH agonists and antagonists continued to increase, peaking in 2023; since 2019, GnRH drugs have been the most frequently reported. Figure 1(b) shows that, despite the absolute growth shown in Figure 1(a), proportional AE reports changed little after 2013: aromatase inhibitors dominated early (peak 0.73% in 2009), GnRH drugs rose steadily to 0.83% in 2024, and the other categories stabilized after 2019. Figure 1(c) displays a U-shaped AE onset pattern, with most AEs occurring within 0–30 days or >360 days after treatment initiation and fewer occurring around 180 days.
Comprehensive analysis of target drugs’ data. (a) Frequency of target drugs’ AEs in the FAERS database (2004Q1–2024Q2). (b) Number of target drugs’ cases in the FAERS database (2004Q1–2024Q2). (c) Frequency of AEs occurring during different periods.
Classification and screening of positive signals
HLGTs flagged as positive by all four methods, along with their preliminary classifications, are visualized in Figure 2. The positive HLGTs shown in Figure 2 and Table S3 are categorized into three groups: musculoskeletal disorders, reproductive and breast disorders, and others. HLGTs classified as “others” are diverse but account for a smaller overall proportion, predominantly falling into eye- and vascular-related categories within the estrogen receptor-targeted drugs group. HLGTs under reproductive and breast disorders, such as breast therapeutic procedures and Ureteric disorders, were expected because of their relevance to hormonal therapy. In contrast, positive HLGTs under musculoskeletal disorders warrant further investigation, as they were underrepresented in comparative studies across drug categories.
Disproportionality analysis signal values and HLGT classifications of positive HLGTs associated with target drugs. (a) Disproportionality analysis signal values and HLGT classifications of positive HLGTs associated with estrogen receptor-targeted drugs. (b) Disproportionality analysis signal values and HLGT classification of positive HLGTs associated with androgen receptor antagonists. (c) Disproportionality analysis signal values and HLGT classifications of positive HLGTs associated with abiraterone. (d) Disproportionality analysis signal values and HLGT classifications of positive HLGTs associated with aromatase inhibitors. (e) Disproportionality analysis signal values and HLGT classifications of positive HLGTs associated with GnRH agonists and antagonists.
Among users of estrogen receptor-targeted drugs, four positive HLGTs related to musculoskeletal disorders were identified: fractures, skeletal neoplasms malignant and unspecified, bone and joint injuries, and musculoskeletal and soft tissue investigations (excluding enzyme tests). Similarly, among users of aromatase inhibitors, four positive HLGTs related to musculoskeletal disorders were identified: fractures, skeletal neoplasms malignant and unspecified, tendon, ligament and cartilage disorders, and bone disorders (excluding congenital disorders and fractures). For androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists, only skeletal neoplasms (malignant and unspecified) was flagged.
Single-drug AEs
We analyzed AEs at the PT level for the five hormonal therapy drug categories, focusing on the top 10 HLGTs with the greatest numbers of positive PTs, as shown in Figure S2(a) to (e). Bone-related PTs were prominent for estrogen receptor-targeted drugs and aromatase inhibitors but less so for androgen receptor antagonists and abiraterone. Figure S2(f), as a supplement to Figure 2, visually highlights the minority status of positive bone-related HLGTs among androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists.
A total of 448 bone-related PTs were identified based on the MedDRA classification (Table S4). Disproportionality analysis yielded 36, 39, 7, 8, and 10 positive bone-related PTs for estrogen receptor-targeted drugs, aromatase inhibitors, androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists, respectively. For categories with more than 20 positive PTs, the top 20 were analyzed; otherwise, all PTs were included, as shown in Figure 3. Notably, androgen receptor antagonists and GnRH agonists and antagonists showed the highest ROR values for bone cancer and Epiphyses premature fusion, respectively. Abiraterone exhibited generally low frequencies of positive bone-related PTs across categories.
Reporting odds ratios for the top bone-related AEs associated with target drugs. (a) Reporting odds ratios for the top 20 bone-related AEs associated with estrogen receptor-targeted drugs. (b) Reporting odds ratios for the top 20 bone-related AEs associated with aromatase inhibitors. (c) Reporting odds ratios for the top 7 bone-related AEs associated with androgen receptor antagonists. (d) Reporting odds ratios for the top 8 bone-related AEs associated with abiraterone. (e) Reporting odds ratios for the top 10 bone-related AEs associated with GnRH agonists and antagonists.
In total, 57 positive bone-related PT level AEs were identified across the five drug categories, including 22 listed as IMEs by the EMA (Figure 4(a)). These included 12 types of fractures in different body regions, such as thoracic vertebral fracture, spinal compression fracture, and femoral neck fracture. Notably, three PTs—bone cancer metastatic, bone cancer, and metastases to bone—were positive across all five drug categories. These clinically significant AEs highlight the need for close monitoring and underscore the importance of assessing their associations with hormonal therapy drugs.
Heatmaps of bone-related IMEs and fracture types associated with hormonal therapy drugs. (a) Heatmap of bone-related AEs identified as IMEs associated with target drugs. (b) Heatmap of 12 fracture types identified as IMEs associated with target drugs. The heatmaps display disproportionality analysis signal values for estrogen receptor-targeted drugs, aromatase inhibitors, androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists. The five drug categories are presented in separate sections separated by blank spaces. Darker colors indicate stronger positive signals.
Analysis of the induction and treatment time of AEs
We analyzed the mean treatment duration and time to AE onset in patients with bone-related AEs, using 21,159, 52,701, 51,404, 15,873, and 69,943 records for estrogen receptor-targeted drugs, aromatase inhibitors, androgen receptor antagonists, abiraterone, and GnRH agonists and antagonists, respectively. Figure S3 shows the average onset times and treatment durations for the AEs identified in Figure 3, excluding cases without time records. Users of estrogen-related drugs had longer average onset times and treatment durations than users of androgen-related drugs. For instance, metastases to bone had onset times of 330, 585, 208, and 145 days and treatment durations of 331, 617, 222, and 167 days for estrogen receptor-targeted drugs, aromatase inhibitors, androgen receptor antagonists, and abiraterone, respectively. GnRH agonists and antagonists generally showed intermediate values between these two groups. While estrogen receptor-targeted drugs and aromatase inhibitors showed the longest durations and the greatest number of bone-related AEs, certain AEs, such as bone lesion and joint lock, had shorter onset times. These findings stress the need for long-term bone health monitoring and prompt management of acute AEs.
Risk management plan-driven case-by-case assessment
We conducted a case-by-case analysis (Tables S5–S7) summarizing the basic demographic information, medication use, and indications for patients experiencing bone cancer, metastases to bone, and bone cancer metastatic. Among patients using estrogen receptor-targeted drugs and aromatase inhibitors as PS drugs, the majority were female, with the primary indications being breast cancer and breast cancer metastatic, except for a few cases such as primaryid 102407202 and 200936531. Case 102407202 involved a 6-year-old male with indications of breast cancer male and Migraine, while case 200936531 involved a 41-year-old female with Intraductal proliferative breast lesion.
For androgen receptor antagonists and abiraterone, all PS drug cases were male except for primaryid 103471581, with PT Prostate cancer as the main indication regardless of sex. GnRH agonists and antagonists were mostly used by males with prostate cancer, whereas some female users had breast cancer. Although a few patients in each category reported bone-related cancers before treatment, most reports of metastases to bone occurred after treatment initiation, generally more than 6 months later, except among abiraterone users. Although causality cannot be established, these observations raise the possibility that hormonal therapies may be associated with subsequent bone metastases in patients with primary breast and prostate cancer. Patients using estrogen-related drugs were generally younger than those using androgen-related drugs, with average ages among patients with metastases to bone of 55 and 59 years compared with 71 and 73 years.
Concomitant medication analysis revealed that some users of GnRH agonists and antagonists were also treated with androgen receptor antagonists such as enzalutamide, and some estrogen receptor-targeted drug users received palbociclib or denosumab, indicating both the osteoporosis risk associated with hormonal therapy and possible enhanced therapeutic effects when combined with these agents.
Positive signals at different fracture sites
After screening for IMEs, 12 types of fractures occurring at different sites and under various conditions were identified. Associations between these fractures and the five hormonal therapy drug categories are summarized in Figure 4(b). Estrogen-related drugs showed stronger associations with fractures and more positive signals than androgen-related drugs and GnRH agonists and antagonists. Among estrogen receptor-targeted drugs, Osteoporotic fracture had the highest signal value (e.g. ROR = 27.57 (21.25–35.76)), whereas ilium fracture ranked highest for aromatase inhibitors (ROR = 10.92 (4.84–24.63)).
Four significant and representative fractures—thoracic vertebral fracture, spinal compression fracture, femoral neck fracture, and osteoporotic fracture—were prioritized. All four showed positive associations with estrogen receptor-targeted drugs (e.g. Thoracic vertebral fracture: ROR = 12.11 (8.93–16.43)), whereas for aromatase inhibitors, Spinal compression fracture had positive signals only for the ROR and PRR. Similarly, spinal fractures showed a stronger association with estrogen receptor-targeted drugs, particularly spine-related fractures. However, aromatase inhibitors were most strongly associated with sternal fracture (ROR = 3.33 (1.84–6.03)) and ilium fracture (ROR = 10.92 (4.84–24.63)) among the five drug categories.
Pathological fracture, often linked to underlying bone lesions and occurring at various skeletal sites, was positively associated with estrogen receptor-targeted drugs (ROR = 5.16 (3.64–7.30)), aromatase inhibitors (ROR = 4.39 (3.39–5.69)), and abiraterone (ROR = 3.79 (2.44–5.88)). A case-by-case analysis in Table S8 showed that although some patients reported pre-existing bone-related AEs such as Osteoporosis or Osteopenia, most had breast or prostate cancer as the primary indication without prior bone-related AEs. Notably, Pathological fracture was frequently reported alongside post-treatment bone-related AEs, such as bone disorder, bone density decreased, and Osteolysis, suggesting a potential drug-induced contribution to its development.
Differential expression of bone-related genes
In the GEO datasets GSE147271 and GSE20181, we identified 233 and 110 DEGs, respectively, using criteria of |logFC| > 1 and padj < 0.05. Eight common DEGs—FOSB, MFAP4, DUSP1, CYR61, FOS, HLF, DPT, and ADAMTS1—were identified, as shown in Figure 5(a). GO and KEGG enrichment analyses of the combined DEGs (Figure 5(b) and (c)) revealed significant pathways, including response to reactive oxygen species, collagen-containing extracellular matrix, and extracellular matrix structural constituent. KEGG analysis highlighted involvement in pathways such as PPAR signaling, PI3K–Akt signaling, and cytoskeleton regulation in muscle cells. The PPI network (Figure 5(d)) identified hub genes such as albumin (ALB), JUN, FOS, COL1A1, and IGF1. Figures S4 and S5 depict the individual DEGs in the two gene sets, respectively.
Transcriptomic analysis of two gene sets, GSE147271 (treated with tamoxifen) and GSE20181 (treated with letrozole). (a) Intersection of DEGs between the two gene sets. (b) GO enrichment analysis of the union of DEGs. (c) KEGG enrichment analysis of the union of DEGs. (d) PPI network analysis of the union of DEGs. (e) Heatmap of the expression of 20 bone-related genes in the pretreatment and treatment groups of GSE147271. (f) Heatmap of the expression of 20 bone-related genes in the pretreatment and treatment groups of GSE20181.
Additionally, we selected 30 genes from the DEGs identified in at least one dataset that were either identified in pathways related to bone metabolism or found to be differentially expressed in bone-related conditions. Heatmaps of their expression profiles across samples are shown in Figure 5(e) and (f). In the tamoxifen-treated GSE147271 dataset, the most significantly altered bone-related genes were NR4A1, DUSP1, FOS, JUN, and EGR2. In the letrozole-treated GSE20181 dataset, key DEGs included FOSB, S100A7, FOS, TFF1, and DUSP1.
Interestingly, all DEGs in the tamoxifen group showed increased expression relative to the pretreatment group. Conversely, in the letrozole group, genes such as S100A7, TFF1, S100A8, MMP1, and SERPINA1 exhibited decreased expression compared with the pretreatment group. These findings suggest distinct transcriptional responses to tamoxifen and letrozole, particularly in bone-related pathways.
Discussion
The FAERS data revealed a strong correlation between the sex distribution of hormonal therapy drug users and the targeted sex hormones. GnRH agonists and antagonists were predominantly used by male patients. Age distributions also differed: users of estrogen-related drugs were mostly aged 18–65 years, whereas users of androgen-related drugs were concentrated in the 65–85 years age range. Trends in AEs showed an increase in those linked to estrogen-related drugs over recent years, while AEs related to androgen receptor antagonists declined. In contrast, AEs associated with GnRH agonists and antagonists rose significantly after 2018, likely because of the overlapping target populations of androgen receptor antagonists and GnRH agonists and antagonists, as well as the replacement of first-generation androgen receptor antagonists with second-generation drugs during that period. 26 Notably, bone-related AEs consistently accounted for a significant proportion of events across all drug categories. However, as shown in Table 1, missing data were substantial for age (up to 50%) and sex (up to 12.1%), which may bias descriptive and subgroup analyses. Nevertheless, the consistent positive signals from disproportionality analyses support the robustness of the main findings.
Since the discovery in the 1940s that estrogen deficiency is associated with bone loss and osteoporosis, the critical role of estrogen in bone health has been widely recognized. 27 Subsequent studies have also highlighted the effects of androgens, such as testosterone and its precursors, which promote the proliferation of osteoblast precursors and guide their differentiation into osteoblasts in humans and mice. 28
Further research has focused on the contributions of both sex hormones to bone metabolism. In a study of elderly men, statistical analyses revealed that while both estrogen and testosterone maintain serum osteocalcin levels—a marker of bone formation—estrogen accounted for 70% of the influence on bone resorption, whereas testosterone contributed less than 30%. 29 Additionally, in men, circulating estradiol levels were found to have a stronger correlation with bone mineral density than circulating testosterone levels. 30 Consistent with these findings, estrogen-related drugs and aromatase inhibitors were associated with a higher frequency and broader spectrum of bone-related AEs than the other hormonal therapy categories. These two drug categories are associated with increased bone fragility and a broader range of joint-related conditions, emphasizing the need for vigilant monitoring of bone health, particularly in patients receiving estrogen-related therapies.
Using the IME list to identify significant AEs, this study found fractures to be the most relevant, with varying risks depending on the anatomical location and the specific hormonal therapy drug used. Research has shown that estrogen deficiency is more likely to cause bone loss in trabecular bone than in cortical bone. 31 A recent study further demonstrated that bone loss in estrogen-deficient mice can be predicted by peak bone microstructure, particularly trabecular thickness, 32 providing additional evidence for this phenomenon. Interestingly, the two primary forms of estrogen receptors, ERα and ERβ, are predominantly expressed in cortical and trabecular bone, respectively, 33 suggesting that receptor-specific mechanisms may underlie these variations in bone loss.
In general, trabecular bone has a lower calcium content and higher water content than cortical bone and is primarily located in the vertebral body, whereas cortical bone is characteristic of regions such as the hip. 34 This observation aligns with the findings related to estrogen-related drugs in this study: fractures involving structures with high trabecular bone content, such as spinal, thoracic vertebral, and lumbar vertebral fractures, exhibited more positive signals and stronger associations than fractures involving hip-related cortical bone, such as hip and ilium fractures. Furthermore, as most users of estrogen-related drugs in this study were elderly women, whose trabecular bone loss exceeds cortical bone loss, 35 the relative depletion of trabecular bone likely increases susceptibility to estrogen deficiency-induced bone loss.
To explore the bone-related pathways affected by hormonal therapy drugs at the gene expression level, and given that estrogen-related drugs were more likely to cause bone-related AEs, we conducted differential gene expression analyses in two cohorts treated with estrogen-related drugs from the GEO database. Key genes with notable fold changes were identified, with bone-related genes comprising a significant proportion and occupying central positions in the PPI network. Among the five highest-scoring genes, four were bone-related (excluding ALB).
For example, FOS is frequently associated with rearrangements observed in osteoid osteomas and osteoblastomas, indicating a potential link between these drugs and an elevated risk of bone-related tumors. 36 JUN plays a role in the microRNA-98/DUSP1/Janus kinase (JNK) axis, which has been implicated in bone loss associated with postmenopausal osteoporosis. 37 IGF-1 is known to promote cell differentiation and proliferation, key processes in tissue growth and remodeling. 38 Consistent with this biological role, serum IGF-1 levels are correlated with cortical bone size and density. 39 COL1A1, a key structural protein, is critical in osteogenesis imperfecta because of its impact on collagen quantity and structure. 40 Notably, all four bone-related genes—FOS, JUN, IGF-1, and COL1A1—are directly or indirectly connected to the JNK signaling pathway.41–43 These results highlight specific directions for future research into the mechanisms underlying bone-related AEs induced by hormonal therapy drugs. From a clinical translation perspective, targeting the JNK signaling pathway, together with established bone-protective agents (e.g. bisphosphonates or denosumab), may represent a potential strategy for mitigating hormonal therapy-induced bone AEs. However, this hypothesis requires further validation through preclinical models and prospective clinical studies.
This study has several limitations. The FAERS database is based on a spontaneous reporting system and is therefore subject to reporting biases, underreporting, incomplete records, and temporal reporting fluctuations such as the Weber effect, which may have led to overestimation or underestimation of certain bone-related AE signals. In particular, recently marketed drugs may show inflated disproportionality signals during the early postmarketing period. As an observational analysis, this study cannot reveal underlying mechanisms, and rigorous adjustment for confounding factors, including age, baseline bone disease, and concomitant medications, was not feasible because of the incomplete clinical information available in the FAERS database. In addition, the interpretation of bone metastasis signals is limited by the inability to differentiate drug-induced metastases from natural tumor progression. Moreover, the transcriptomic datasets were derived from breast cancer tissues rather than bone tissues or bone-related cell models. Therefore, the identified DEGs and pathways should be regarded as reflecting tumor-associated or systemic responses to hormonal therapy, and their relevance to direct molecular alterations in bone microenvironments remains inferential and requires validation in bone-specific experimental systems.
Despite these constraints, previous FAERS studies focused exclusively on single drug classes such as aromatase inhibitors and specific outcomes such as fractures 44 or osteoporosis, 45 reporting stronger associations for anastrozole and letrozole than for exemestane. In contrast, our study expanded the analysis by simultaneously comparing five hormonal therapy categories, identifying 57 positive bone-related signals, including 22 IMEs, and incorporating time-to-onset and case-by-case assessments. In addition, transcriptomic analyses suggested the potential involvement of the JNK pathway and several key genes in hormonal therapy-associated bone-related AEs. Future work should incorporate more diverse pharmacovigilance data, mechanistic validation, and bone-specific experimental models to further evaluate the biological and clinical relevance of these findings.
Conclusion
In conclusion, this study identified significant associations between hormonal therapy drugs and bone-related AEs in the FAERS database, particularly fractures and skeletal complications associated with estrogen-related therapies. Transcriptomic analyses suggested that genes including JUN, FOS, IGF1, and COL1A1, as well as the JNK signaling pathway, may be involved in these bone-related effects. These findings highlight the importance of bone health monitoring during hormonal therapy, especially in patients with pre-existing skeletal risk factors. Future studies using bone-specific experimental models and prospective clinical data are needed to further validate the underlying mechanisms and clinical relevance of these findings.
Supplemental Material
sj-docx-1-imr-10.1177_03000605261463660 - Supplemental material for Bone-related adverse events of hormonal therapy: A pharmacovigilance study based on the Food & Drug Administration Adverse Event Reporting System
Supplemental material, sj-docx-1-imr-10.1177_03000605261463660 for Bone-related adverse events of hormonal therapy: A pharmacovigilance study based on the Food & Drug Administration Adverse Event Reporting System by Yantian Wang, Anqi Li, Jiafu Wang and Min Xie in Journal of International Medical Research
Supplemental Material
sj-docx-2-imr-10.1177_03000605261463660 - Supplemental material for Bone-related adverse events of hormonal therapy: A pharmacovigilance study based on the Food & Drug Administration Adverse Event Reporting System
Supplemental material, sj-docx-2-imr-10.1177_03000605261463660 for Bone-related adverse events of hormonal therapy: A pharmacovigilance study based on the Food & Drug Administration Adverse Event Reporting System by Yantian Wang, Anqi Li, Jiafu Wang and Min Xie in Journal of International Medical Research
Footnotes
Acknowledgments
The authors would like to thank the FDA Adverse Event Reporting System and the providers of the GSE147271 and GSE20181 datasets for their contributions. We also thank ChatGPT for its assistance with language polishing and improving readability.
Ethics approval and consent to participate
Not applicable.
Patient consent
Not applicable.
Author contributions
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Data availability statement
The data used in this study are publicly available and may be reused without restriction under an open license. The datasets analyzed during the current study are available from the corresponding author upon reasonable request.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
