Dienogest use and the risk of breast and gynecologic cancers: A nationwide population-based study.

1
INTRODUCTION
Endometriosis has a high recurrence rate even after appropriate surgical intervention, making long‐term pharmacotherapy—focused on pain reduction and recurrence prevention—the cornerstone of management. It is therefore essential to understand clearly the correct use and potential adverse effects of each therapeutic agent. Traditional pharmacologic options comprise gonadotropin‐releasing hormone (GnRH) agonists, danazol, and combined oral contraceptives, but their adverse‐effect profiles limit long‐term use.
1
In recent years, the use of dienogest for endometriosis has steadily increased, with many patients requiring long‐term therapy. Dienogest is a fourth‐generation, orally active progestin with highly selective binding to the progesterone receptor. As a selective progestin, it uniquely combines the pharmacologic properties of both progesterone derivatives and 19‐norprogestins, offering distinct advantages for the treatment of endometriosis. It has been approved for endometriosis treatment in more than 70 countries across Europe and Asia.
2
It was first approved by the Ministry of Food and Drug Safety of the Republic of Korea in 2011 and released for use in February 2013.
3

Women with endometriosis show a higher risk of ovarian cancer and, minimally, breast cancer.
4
Through a variety of intricate mechanisms, progestogens can affect the risk of breast, endometrial, and tubo‐ovarian cancers. Epidemiologic and clinical evidence indicate that uncontrolled progesterone action in breast cancer progenitor cells can contribute to the development of breast cancer.
5
Although they were once believed to protect against tubo‐ovarian cancer, recent research suggests a more nuanced relationship. Both stimulatory and inhibitory effects on cancer cell growth have been proposed.
6
Factors including progestogen type, dosage, and duration of use, as well as individual susceptibility, may play crucial roles in determining overall cancer risk.
5

Given that endometriosis is associated with various gynecologic cancers and prolonged progestin use can raise concerns about potential malignancies, both patients and healthcare providers may feel uneasy about extended medication use. These worries are particularly relevant in the context of breast cancer, where progestin use in postmenopausal hormone therapy has been consistently linked to an increased risk.
7
Dienogest has strong progestogenic effects. In addition, dienogest demonstrated inhibitory effects on the growth of vascular endothelial cells, suggesting potential preventive applications for breast and endometrial cancers, as well as endometriosis.
8
Epidemiologic studies assessing the potential risk reduction of breast or gynecologic cancers among dienogest users are limited. This study aimed to examine the association between dienogest use and the risk of breast and gynecologic cancers at a national level.

2
MATERIALS AND METHODS
2.1
Data
We conducted a nationwide retrospective observational study using the Korean Health Insurance Review & Assessment Service (HIRA) database. The HIRA database contains individual demographics and detailed national health insurance service‐reimbursed healthcare utilization records for over 99% of the population. The protocol of this study was specified according to the target trial emulation framework to estimate the effect of dienogest on risk for cancer (Table S1). We constructed a cohort of Korean women aged 20–49 years using data from the HIRA database from January 1, 2012, to December 31, 2023. The target population included individuals diagnosed with endometriosis (International Classification of Diseases 10th revision [ICD‐10] code N80). To assess the association between dienogest use and sex hormone‐dependent cancers, GnRH agonist users were selected as the active control group, because GnRH agonists have been the treatment of choice for postoperative hormonal suppression in endometriosis patients. We assumed that individuals receiving GnRH agonists exhibited a comparable severity of endometriosis that warranted postoperative hormonal suppression, implying that their baseline risks for breast and gynecologic cancer at medication initiation were similar to those of dienogest users. Individuals with a history of corresponding cancers before medication initiation were excluded—each criterion applied separately in the analyses. Those with a history of bilateral oophorectomy were omitted from the tubo‐ovarian cancer analysis, while individuals with a history of hysterectomy were excluded from the endometrial cancer analysis.

2.2
Ascertainment of exposure
Women who received GnRH agonists or dienogest were identified based on prescription records. The GnRH agonist group were defined as those who received any of the 11 injectable GnRH agonists available in the country during the study period (Table S2). The list to identify dienogest use included a single form of oral dienogest medication (Visanne®; Bayer AG, Berlin, Germany). Exposure to dienogest was defined as dispensation of dienogest (2 mg × 28 tablets) for a duration of 6 months or longer, to minimize exposure misclassification. This threshold reflects the minimum duration employed in pivotal long‐term trials of dienogest treatment and previous observations of increased risk of estrogen receptor‐positive breast cancer with more than 6 months of use of combined hormone therapy.
9
For endometrial cancer, the time required for the resolution of cancer following progesterone therapy varies widely, whereas most reported more than 3–6 months of progesterone administration is required for clinical response in endometrium.
10
Significant associations with ovarian cancer were observed with 6 months or more of hormone therapy.
11

2.3
Outcomes
Individuals with corresponding ICD‐10 codes were identified to have incident breast (C50, D05), endometrium (C54.1), and tubo‐ovarian (C56, C570) cancer. Among those without any pre‐existent cancer before the initiation of GnRH agonist or dienogest, women who had outpatient visits at least three times or at least one hospitalization within the first 365 days of the diagnosis of the same cancer were defined as newly diagnosed patients according to the validated method.
12
To exclude cases with pre‐existing cancer, we removed those whose diagnosis was made within 2 years of prescription because of the possibility of undiagnosed cancer at the time of prescription. This approach aligns with national guidelines for cancer screening, which recommend a clinical examination or mammography every 2 years.
13

2.4
Statistical analysis
Summary statistics were calculated to compare the characteristics of the GnRH agonist and dienogest groups. Cumulative incidences of each cancer were assessed in both groups. We used inverse probability of treatment weighting (IPTW) based on the propensity score (PS) to control for confounding by balancing the two groups at baseline. Our explanatory model included age, location of the healthcare facility (inside or outside the Seoul metropolitan area), the year and month of medication initiation, history of gynecologic surgery, pre‐existing conditions, Charlson comorbidity index (CCI, ≥3 or <3), and the risk of corresponding cancer within 2 years of starting medication. We also accounted for history of gynecologic surgery and any pre‐existing breast or gynecologic conditions, evaluated separately for each cancer outcome. The PS was calculated conditional on the covariates specified by cancer types. Incidence rates of cancers per person‐year were computed in the matched samples. Observation was censored at the time of (1) cancer diagnosis, (2) the last outpatient or inpatient visit recorded in the data, and (3) initiation of dienogest medication for GnRH agonist users or initiation of GnRH medication for dienogest users, whichever occurred first.
We compared the GnRH agonist and dienogest users with weighted measures for the three sets of study populations. Absolute standardized differences between the two groups less than 0.1 were considered to be indicative of balanced risk. Then we used Cox proportional hazards regression to estimate adjusted hazard ratios (aHR) and 95% confidence intervals (CI) of breast, uterine, and tubo‐ovarian cancers using a PS‐IPTW applied dataset. The explanatory models included covariates separately specified for each cancer. To test the robustness of our findings, we examined the dose–response relationship of dienogest across three types of cancer. Adjusted HRs were calculated stratified by duration of dienogest prescription: 0.5–1.5, 1.5–2.5, and over 2.5 years. The GnRH agonist group was used as the control across all three durations. SAS version 9.4 (SAS Institute, Cary, NC, USA) was used in all statistical analyses.

3
RESULTS
Among 1 887 957 women with endometriosis aged 20–49 years, 181 884 (9.6%) received GnRH agonist injections and 94 697 (5.0%) were prescribed dienogest during the study period (Figure 1). For the breast cancer analysis, we excluded 336 patients diagnosed with breast cancer before medication initiation and 80 011 patients with a dienogest prescription duration of fewer than 180 days. This resulted in a final study population of 181 587 GnRH agonist users and 14 647 dienogest users. For the endometrial cancer analysis, patients who underwent hysterectomy before or within 2 years of initiating medication (n = 15 038), as well as those with dienogest use for fewer than 180 days (n = 75 817), were excluded. The final cohort included 171 530 GnRH agonist users and 14 196 dienogest users. In the tubo‐ovarian cancer analysis, we excluded patients who had undergone bilateral oophorectomy before or within 2 years of medication initiation (n = 70 856), along with those who had received dienogest for fewer than 180 days (n = 48 006), resulting in 149 709 GnRH agonist users and 8010 dienogest users.
In the cohort constructed for breast cancer risk analysis, dienogest users were more likely to be older (30.3 years in dienogest versus 27.6 years in GnRH agonist), treated in health institutions outside the capital area (42.4% vs. 40.6%), have a history of bilateral oophorectomy (47.0% vs. 18.6%), and develop breast cancer within 2 years of medication initiation (0.2% vs. 0.4%), compared with GnRH agonist users. GnRH users showed higher prevalence of hysterectomy than dienogest users (6.8% in GnRH agonist vs. 3.9% in dienogest). This pattern of differences was generally consistent across the cohorts constructed for the analyses of endometrial and tubo‐ovarian cancers. For example, the risk of developing endometrial and tubo‐ovarian cancers withing 2 years of medication initiation was higher among dienogest users in the respective analysis cohort. After applying PS‐IPTW, baseline characteristics between the two groups were well balanced (Table 1; Tables S3 and S4). In the cohort for breast cancer analysis, the median duration of use was 55–56 days for GnRH agonists and 364 days for dienogest (Table S5). Observation time was generally shorter in dienogest users, with medians of 1167 days for GnRH agonist users and 1036 days for dienogest users.
Overall incidence of breast cancer in the study population was 102 per 100 000 person‐years. A total of 716 cases of breast cancer were identified among GnRH agonist users over 698 652 person‐years, yielding an incidence rate of 102 per 100 000 person‐years (95% CI 95–110; Table 2). Forty‐seven cases were observed among dienogest users over 48 677 person‐years, with an incidence rate of 97 per 100 000 person‐years (95% CI 73–129). The incidence rate ratio (IRR) was 0.94 (95% CI 0.70–1.27), and the adjusted aHR was 1.01 (95% CI 0.75–1.37).
Endometrial cancer occurred at a rate of 13 per 100 000 person‐years. Eighty‐six cases of endometrial cancer occurred in the GnRH agonist group (653 550 person‐years; incidence rate 13 per 100 000 person‐years, 95% CI 11–16), and eight cases occurred in the dienogest group (46 240 person‐years; incidence rate 17 per 100 000 person‐years, 95% CI 8–34). The IRR was 1.28 (95% CI 0.61–2.66) and the aHR was 0.84 (95% CI 0.40–1.77).
The incidence of tubo‐ovarian cancer was 15 per 100 000 person‐years. Tubo‐ovarian cancer occurred in 80 GnRH agonist users (529 868 person‐years; incidence rate 15 per 100 000 person‐years, 95% CI 12–19) and in four dienogest users (25 428 person‐years; incidence rate 13 per 100 000 person‐years, 95% CI 4–38). The IRR was 0.85 (95% CI 0.28–2.58) and the aHR was 0.92 (95% CI 0.30–2.80).
In the stratified analyses, the aHR of breast cancer was lower among individuals with 0.5–1.5 years of use (0.65, 95% CI 0.09–0.99; Table 3). For longer durations of dienogest use, the direction of association varied across strata, and the confidence intervals were imprecise. In the endometrial cancer analysis, the direction of association was similarly inconsistent across duration categories, with no incident cases observed among individuals with more than 2.5 years of dienogest use. A comparable pattern of inconsistent associations was also observed in the analysis of tubo‐ovarian cancer.

4
DISCUSSION
Using GnRH agonist users as an active comparator, we observed similar cancer incidence across the two groups, and this null association remained after prolonged dienogest treatment, except for breast cancer. Given the paucity of long‐term safety data on progestin therapy, our findings provide an important insight on the impact of dienogest on the risk of breast and gynecologic cancers among endometriosis patients.
Our findings support the oncologic safety of dienogest with respect to breast cancer. Although earlier observational work linked progestin‐based hormone therapy to higher breast cancer risk, the present null association suggests that dienogest may differ pharmacodynamically from other progestins. In vitro and human data indicate that, even at 20 mg daily for 6 months, dienogest reduces breast glandular volume and abnormal architecture without markedly affecting adipose tissue.
14
Effects that could counter pro‐proliferative signaling in mammary epithelium.
15
A study from Taiwan reported no association between the duration of dienogest use and abnormal mammographic findings.
16
The use of GnRH agonists as an active comparator—agents that may themselves lower breast cancer risk—together with relatively brief exposure durations in our cohort, is likely to have attenuated any detectable protective effect.
In contrast to the protection afforded by prolonged combined estrogen‐progestin therapy against endometrial cancer in postmenopausal women, we observed a near‐null association between dienogest and endometrial cancer. Our study population was younger, premenopausal, and the median exposure was relatively short, whereas 9 years or longer of continuous progestin is needed to reduce endometrial cancer risk.
17
In addition, endometriosis and its attendant hormonal milieu may modify responsiveness to progestin therapy. Clarifying whether dienogest offers endometrial protection beyond that demonstrated for other progestins will require cohorts with longer follow up.
The close‐to‐null estimate for tubo‐ovarian cancer accords with several studies reporting no effect of progestin therapy.
18
Although progestins can reduce tubo‐ovarian tumorigenesis by suppressing ovulation and altering the intra‐pelvic hormonal environment,
19
such benefits may depend on cumulative dose and indication. Doses used for endometriosis, coupled with brief exposure, may be insufficient to influence malignant transformation in an organ that is already subject to a four‐fold cancer excess among women with endometriosis.
20
Our findings may reflect differences in progestin type, dosing regimens, and patient populations, as dienogest is typically prescribed for conditions like endometriosis rather than solely for ovulation suppression. Moreover, the relatively short duration of exposure and underlying indication for dienogest use in our cohort may limit its potential effect on reducing tubo‐ovarian cancer risk. Further work should examine whether higher cumulative exposure or alternative progestin formulations modulate this risk.
Compared with the general Korean female population (132.8 per 100 000),
21
breast cancer incidence among our endometriosis patients treated with GnRH agonists or dienogest was lower. Previous epidemiologic studies have been mixed: one 10‐year follow up found no difference in breast cancer risk between women with endometriosis and matched controls,
22
but another reported a 44% higher risk among endometriosis patients.
23
The reduced incidence in our cohort likely reflects the younger age of those receiving hormonal therapy, who inherently have a lower baseline breast cancer risk.
By contrast, endometrial and ovarian cancer rates were higher in our cohort than in the general population (5.7 and 6.3 per 100 000 women‐years, respectively, among Korean women during the same period).
24
This aligns with earlier reports linking endometriosis to both endometrial and ovarian malignancies.
23
In particular, “endometriosis‐associated ovarian cancer” refers to the endometrioid and clear‐cell subtypes. Although this positive association is well documented, direct malignant transformation of endometriotic lesions is rare (<1%), suggesting that endometriosis may coexist with—or share pathogenic pathways in—a subset of ovarian cancers rather than serve as the precursor for most tumors.
25

Several limitations warrant caution. First, median dienogest exposure was short, potentially underestimating long‐term effects. Second, unmeasured confounding is possible because progesterone receptor positivity of the tumor, lifestyle factors, family history, and information on histologic subtypes were unavailable in the HIRA data. For example, hormone‐receptor positivity could have affected both clinicians' choice of dienogest and the risk of breast or endometrial cancer. Incorporating receptor status and histologic subtype into future analyses would enable more precise confounding control and risk estimation. Third, the wide confidence intervals for endometrial cancer reflect low event counts and indicate that larger populations or longer follow up are necessary. Finally, using GnRH agonists as the comparator precludes estimation of the absolute risk difference between dienogest and no hormone therapy; future studies incorporating non‐hormonal comparators would better delineate the net benefit–risk profile of dienogest.
In conclusion, we observed no increased risk of breast, endometrial, or tubo‐ovarian cancers among dienogest users based on large real‐world data. This suggests that dienogest therapy can be considered as a safe option for managing endometriosis without substantially increasing the likelihood of breast cancer.

AUTHOR CONTRIBUTIONS
HJ conducted primary analyses and drafted the manuscript. SK supported the refinement of the study protocol and reviewed the manuscript. HP and JYS reviewed and revised the manuscript for important intellectual content. SAC acquired the data, designed the study protocol, and revised the manuscript.

CONFLICT OF INTEREST STATEMENT
The authors have no conflicts of interest.

Supporting information

Table S1. Specification of the study protocol components according to the target trial emulation framework comparing dienogest and GnRH agonist on the risk for breast, endometrial, and tubo‐ovarian cancers using observational data.

Table S2. The 10th revision of the International Classification of Diseases codes, drug code based on main ingredient, and treatment codes used for identification of endometriosis, exposure to GnRH agonist or dienogest, history of surgeries, and cancer incidence; Health Insurance Review & Assessment Service (HIRA) database, 2012–2023.

Table S3. Demographic and clinical characteristics of GnRH agonist and dienogest users for the analysis of endometrial cancer; Health Insurance Review & Assessment Service (HIRA) database, 2012–2023.

Table S4. Demographic and clinical characteristics of GnRH agonist and dienogest users for the analysis of tubo‐ovarian cancer; Health Insurance Review & Assessment Service (HIRA) database, 2012–2023.

Table S5. Duration of medication and observation in the GnRH agonist and dienogest users in the three cohorts for the analysis of breast, endometrial, and tubo‐ovarian cancers; Health Insurance Review & Assessment Service (HIRA) database, 2012–2023.