Statin Use and Site-Specific Cancer Risk: A Nationwide Cohort Study Using Korean Health Insurance Data

Abstract

Objective

Statins are used for cardiovascular prevention and may exert anticancer effects by modulating the mevalonate pathway and tumor microenvironment. However, population-based evidence in Koreans remains limited. This study evaluated the association between statin use and cancer risk using Korean nationwide health data.

Methods

This population-based cohort study used the Korean National Health Insurance Service database. Individuals who underwent health screening in 2009–2010 were followed until 2019. Statin users referred to individuals prescribed statins for ≥6 months, whereas non-users had no statin exposure. Incident cancers were identified using International Classification of Diseases, 10th Revision, codes. Multivariable Cox proportional hazards models were used to estimate adjusted hazard ratios (aHRs).

Results

Among 232,133 participants (67,064 statin users, 165,069 non-users), the incidence rate for all malignancies was 10.60 per 1,000 person-years (PY) in statin users and 10.18 per 1,000 PY in non-users. After adjusting for confounders, statin use was associated with a modest but significant reduction in overall cancer risk (aHR, 0.96; 95% confidence interval [CI], 0.93–1.00; p=0.031). Site-specific analyses demonstrated reduced risk of liver cancer (aHR, 0.52; 95% CI, 0.44–0.61; p<0.001) among statin users. Conversely, thyroid cancer incidence was higher among statin users, and statin use was markedly associated with increased thyroid cancer risk (aHR, 1.19; 95% CI, 1.07–1.33; p=0.001).

Conclusion

In this nationwide Korean cohort, statin use was associated with a reduced risk of liver cancer but a significantly increased risk of thyroid cancer. These findings suggest organ-specific effects of statins on carcinogenesis and warrant further mechanistic investigation.

INTRODUCTION

Statins are lipid-lowering agents that reduce cholesterol synthesis by inhibiting 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase.¹ By inhibiting HMG-CoA reductase, the rate-limiting enzyme in cholesterol biosynthesis that catalyzes the conversion of HMG-CoA to mevalonate, statins effectively mitigate endogenous cholesterol synthesis. Statins are currently the first-line pharmacologic treatment for dyslipidemia and are most widely prescribed for the prevention and treatment of atherosclerotic cardiovascular disease.²,³,⁴ Beyond their lipid-lowering effects, statins exert pleiotropic effects, including anti-inflammatory activity, modulation of immune cell function, and independent improvement of endothelial function.⁵,⁶

Moreover, recent preclinical evidence suggests that the inhibition of the mevalonate pathway by statins may also suppress cancer development and progression.⁷ Statin-induced depletion of geranylgeranyl pyrophosphate and downregulation of B-cell lymphoma 2 family proteins have been reported to induce tumor-specific apoptosis and inhibit cancer cell proliferation.⁸ In addition, statins have been reported to induce ferroptosis—an iron-dependent form of regulated cell death—through regulation of glutathione peroxidase 4 and FSP1/CoQ10/NAD(P)H axes via the mevalonate pathway.⁹ Beyond these direct anticancer effects, accumulating evidence indicates that statins can modulate the tumor microenvironment, thereby exerting additional antitumor effects.¹⁰

Based on these mechanistic insights, several population-based observational studies have evaluated the potential anticancer effects of statins. A large Danish observational study with 15 years of follow-up reported reduced cancer-related mortality across multiple cancer types (13 cancer types) among statin users.¹¹ Similarly, a meta-analysis involving 1,111,407 patients with cancer demonstrated that statin use was associated with reductions in overall and cancer-related mortality by approximately 30% and 40%, respectively.¹²

Although preclinical and clinical evidence supporting the anticancer effect of statins has rapidly accumulated in recent years, comprehensive cohort studies evaluating these effects in Korean populations remain limited. Given the differences in genetic background, lifestyle, and environmental factors between the Korean and Western populations, population-specific evidence derived from Korean cohorts is essential. Therefore, this study aimed to evaluate the association between statin use and cancer risk in a Korean cohort using the National Health Insurance Service (NHIS) database. Specifically, the study population was stratified by age, sex, and comorbidity status to assess overall cancer incidence, followed by analyses of organ-specific and individual cancer types.

MATERIALS AND METHODS

1. Study population

The study population was derived from the NHIS-National Health Screening Cohort database. We included participants aged 40–79 years who underwent a general health screening provided by the NHIS. The NHIS database comprehensively captures healthcare service utilization across Korea and includes biennial general health checkup data, making it a robust resource for population-based research. This study included individuals who underwent a general health examination in 2009 and were followed until December 31, 2019, to evaluate the association between statin use and the incidence of cancer.

The initial study population comprised 514,866 individuals who participated in the 2009 health examination. Among these, participants with complete health screening data, including low-density lipoprotein cholesterol (LDL-C) measurements, recorded between 2009 and 2010 were selected. After applying predefined exclusion criteria, 232,133 individuals were included in the final analytic cohort (Fig. 1).

Fig. 1. Flowchart of the study population selection process.

NHIS-HEALS, National Health Insurance Service-National Health Screening Cohort; LDL, low-density lipoprotein; BMI, body mass index; eGFR, estimated glomerular filtration rate; SGPT, serum glutamic-pyruvic transaminase (alanine aminotransferase).

This study used the National Health Information Database provided by the NHIS. The study protocol was approved by the Institutional Review Board (IRB) of Asan Medical Center (IRB No. 2025-0391). The NHIS approved the use of data for this research (NHIS-REQ No. REQ2025041341). The requirement for informed consent was waived by the IRB as the database consists of de-identified secondary data for research purposes. This study was conducted and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

2. Definition of statin use

Statin exposure was defined in 2 categories: index-period statin users and follow-up statin initiators. Index-period statin users included individuals who used statins for at least 6 months during the index period (2009–2010). The mean ± standard deviation (SD) duration of statin use in this group was 367.7±275.9 days, with a median duration of 330 (interquartile range [IQR], 103–630) days. Follow-up statin initiators were defined as individuals who newly initiated statin therapy during the follow-up period (2011–2018) and used statins for at least 6 months. These individuals were included in the analysis as a time-dependent exposure. The median time from cohort entry to statin initiation was 3.21 (IQR, 1.31–5.19) years. This stratification of statin exposure was implemented to account for the substantial number of participants who initiated statin therapy after the index period, which could otherwise bias effect estimates.

Overall, the treatment group (statin users) included individuals who continuously used statins—such as atorvastatin, rosuvastatin, simvastatin, pitavastatin, pravastatin, and lovastatin—for at least 6 months during the study period. Statin use was calculated using prescription claims data from the NHIS database.

The control group (statin non-users) comprised individuals with no record of statin prescriptions throughout the study period. Subgroup analyses by statin type or dosage were not performed.

3. Outcome

The primary outcome was the incidence of malignant neoplasms, identified using the International Classification of Diseases, 10th Revision (ICD-10), codes. Overall cancer incidence (all cancers) was first assessed, followed by subgroup analyses according to anatomical site and individual cancer types. Incident cancer was defined as the first occurrence of any malignant neoplasm (ICD-10 codes C00–C96) recorded in the NHIS claims database. Participants were followed from the index date until the first cancer diagnosis or December 31, 2019, whichever occurred first. Secondary outcomes included site-specific cancer incidence, with individual cancer types categorized into 10 groups based on their corresponding ICD-10 codes.

4. Covariates

Covariates were obtained from baseline health examination data and included age, sex, household income, Charlson comorbidity index (CCI), hypertension, diabetes mellitus (DM), body mass index (BMI), smoking status, heavy alcohol consumption, regular physical activity, and baseline LDL-C levels. Household income was categorized into five quartiles, with the lowest quartile serving as the reference group. The CCI was categorized as 0, 1, 2, or ≥3, and comorbid conditions included hypertension and DM. Smoking status was classified as never, former, or current smoker. BMI was categorized as 18.5–22.9, 23–24.9, 25.0–29.9, and ≥30 kg/m². Baseline LDL-C was dichotomized as <160 mg/dL or ≥160 mg/dL. All multivariable models, including overall and subgroup analyses, were adjusted for these covariates.

5. Statistical analysis

Baseline characteristics were compared between statin users and non-users, and both overall and site-specific cancer incidence, as well as cumulative incidence, were evaluated according to statin exposure. Continuous variables are presented as mean ± SD, and categorical variables as proportions (%). A 2-sided p-value <0.05 was considered significant.

The association between statin use and cancer risk was assessed using the Cox proportional hazards model, with hazard ratios (HRs) and 95% confidence intervals (CIs) estimated. To rigorously control for potential confounding, we employed multivariable-adjusted time-dependent Cox proportional hazards models. This approach allowed us to account for changes in statin use over time, which provides a more dynamic and accurate estimation of exposure than baseline-fixed models. Multivariable models were adjusted for age, sex, BMI, smoking status, alcohol consumption, physical activity, household income, CCI, hypertension, DM, and baseline LDL-C levels. In the primary analysis, individuals who used statins for ≥6 months during the index period were compared with statin non-users, and all covariates were corrected. To account for statin initiation during the follow-up period, a time-dependent Cox proportional hazards model was applied, treating statin use as a time-varying covariate to accurately reflect changes in exposure status over time. Subgroup analyses were subsequently performed to assess the robustness of the findings.

RESULTS

1. Baseline characteristics of the entire cohort

Baseline characteristics of the study population according to statin use are presented in Table 1. Among the 232,133 participants, 71.1% (n=165,069) were statin non-users and 28.9% (n=67,064) were statin users. Compared with non-users, statin users had a higher prevalence of hypertension (40.5% vs. 21.4%) and DM (15.4% vs. 6.2%), as well as a greater comorbidity burden, as reflected by higher CCI scores. With respect to lipid profiles, the majority of participants in both groups had baseline LDL-C <160 mg/dL; however, this proportion was higher among non-users (93.4% vs. 81.2%). Conversely, LDL-C levels ≥160 mg/dL were more prevalent among statin users than among non-users (18.8% vs. 6.6%). Other baseline characteristics were generally comparable between the two groups.

Table 1. Baseline characteristics of the study population according to statin use.

Variables		Total	Non-user	User	p-value
Number		232,133	165,069	67,064
Age (yr)		57.8±8.5	57.5±8.6	58.6±8.2	<0.001
Sex (male)		127,948 (55.1)	95,395 (57.8)	32,553 (48.5)	<0.001
BMI (kg/m2)		23.9±2.7	23.7±2.7	24.5±2.7	<0.001
WC (cm)		81.6±7.9	81.1±7.9	82.7±8.0	<0.001
SBP (mmHg)		124.6±15.2	123.5±15.0	127.3±15.4	<0.001
DBP (mmHg)		77.5±10.0	76.9±9.9	78.8±10.1	<0.001
Smoking					<0.001
	None	146,372 (63.1)	102,106 (61.9)	44,266 (66)
	Past	41,116 (17.7)	30,129 (18.3)	10,987 (16.4)
	Current	40,810 (17.6)	30,025 (18.2)	10,785 (16.1)
Alcohol consumption					<0.001
	None	133,816 (57.6)	92,809 (56.2)	41,007 (61.1)
	Mild to moderate	53,569 (23.1)	39,577 (24)	13,992 (20.9)
	Heavy	43,085 (18.6)	31,553 (19.1)	11,532 (17.2)
Regular physical activity					0.0303
	None	179,669 (77.4)	127,861 (77.5)	51,808 (77.3)
	Yes	51,311 (22.1)	36,427 (22.1)	14,884 (22.2)
Household income					<0.001
	1st quartile and medical aid	32,384 (14)	22,534 (13.7)	9,850 (14.7)
	2nd quartile	31,093 (13.4)	21,997 (13.3)	9,096 (13.6)
	3rd quartile	37,578 (16.2)	26,536 (16.1)	11,042 (16.5)
	4th quartile	48,517 (20.9)	34,166 (20.7)	14,351 (21.4)
	5th quartile	82,561 (35.6)	59,836 (36.2)	22,725 (33.9)
Hypertension		62,502 (26.9)	35,346 (21.4)	27,156 (40.5)	<0.001
Diabetes mellitus		20,511 (8.8)	10,172 (6.2)	10,339 (15.4)	<0.001
Chronic kidney disease		191 (0.1)	106 (0.1)	85 (0.1)	<0.001
Charlson comorbidity index					<0.001
	0	126,723 (54.6)	97,263 (58.9)	29,460 (43.9)
	1	61,705 (26.6)	42,148 (25.5)	19,557 (29.2)
	2	26,062 (11.2)	16,158 (9.8)	9,904 (14.8)
	3+	17,643 (7.6)	9,500 (5.8)	8,143 (12.1)
LDL-C (mg/dL)		119.6±35.0	114.9±33.5	131.2±35.9	<0.001
TC (mg/dL)		199.9±34.2	193.9±31.9	214.5±35.2	<0.001
HDL-C (mg/dL)		54.5±21.4	54.9±21.4	53.8±21.5	<0.001
TG (mg/dL)		132.3±79.6	124.8±73.6	150.7±90.1	<0.001
FPG (mg/dL)		99±22.3	97.2±19.5	103.4±27.5	<0.001
Hemoglobin (g/dL)		13.9±1.5	13.9±1.5	13.9±1.5	<0.001
ALT (U/L)		24.1±14.7	23.6±14.4	25.2±15.3	<0.001
Creatinine (mg/dL)		0.9±0.2	0.9±0.2	0.9±0.2	<0.001

Values are presented as mean ± standard deviation for continuous variables or number (%) for categorical variables.

BMI, body mass index; WC, waist circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; CCI, Charlson comorbidity index; TG, triglyceride; FPG, fasting plasma glucose; ALT, alanine aminotransferase.

2. Association between statin use and cancer risk

During the follow-up period, a total of 20,184 incident cancer cases were identified. In the unadjusted analysis, statin use was not significantly associated with overall cancer incidence. However, after adjusting for potential confounders, statin use was associated with a modest but statistically significant reduction in overall cancer risk (adjusted hazard ratio [aHR], 0.96; 95% CI, 0.93–1.00; p=0.031).

In site-specific analyses (Table 2), statin use was significantly associated with a reduced risk of liver cancer (aHR, 0.52; 95% CI, 0.44–0.61; p<0.001). In contrast, statin use was associated with an increased risk of thyroid cancer (aHR, 1.19; 95% CI, 1.07–1.33; p=0.001). No significant associations were observed for stomach, colorectal, pancreatic, lung, kidney, and bladder cancers.

Table 2. IRs and HRs for overall and site-specific cancers according to statin use.

Cancer type	IR in non-users	IR in users	Crude HR (95% CI)	p-value	Age and sex-adjusted HR (95% CI)	p-value	Multivariate-adjusted HR (95% CI)	p-value
All malignancy	10.18	10.60	1.033 (0.997–1.071)	0.072	0.995 (0.960–1.031)	0.772	0.960 (0.925–0.996)	0.031
Stomach	1.79	1.79	1.005 (0.923–1.095)	0.901	1.004 (0.922–1.094)	0.924	0.977 (0.894–1.067)	0.601
Colorectal	1.55	1.56	1.040 (0.949–1.139)	0.404	1.010 (0.922–1.107)	0.834	0.977 (0.888–1.075)	0.632
Liver	0.78	0.47	0.585 (0.500–0.685)	<0.001	0.607 (0.518–0.711)	<0.001	0.520 (0.442–0.612)	<0.001
Bile duct and GB	0.58	0.69	1.130 (0.984–1.299)	0.084	1.056 (0.919–1.214)	0.441	0.999 (0.865–1.155)	0.994
Pancreas	0.51	0.54	0.976 (0.836–1.139)	0.757	0.910 (0.779–1.062)	0.232	0.854 (0.726–1.003)	0.054
Lung	1.57	1.81	1.069 (0.982–1.165)	0.125	1.042 (0.957–1.136)	0.343	1.031 (0.943–1.127)	0.499
Kidney	0.27	0.35	1.170 (0.960–1.425)	0.119	1.189 (0.975–1.450)	0.087	1.067 (0.867–1.311)	0.542
Bladder	0.40	0.51	1.170 (1.000–1.384)	0.050	1.187 (1.008–1.397)	0.040	1.124 (0.948–1.332)	0.178
Thyroid	1.18	1.31	1.332 (1.204–1.475)	<0.001	1.276 (1.152–1.413)	<0.001	1.194 (1.072–1.329)	0.001

The multivariate-adjusted model was adjusted for age, sex, household income, Charlson comorbidity index, hypertension, diabetes mellitus, body mass index, smoking status, alcohol consumption, physical activity, and baseline low-density lipoprotein cholesterol levels.

IR, incidence rate (per 1,000 person-years); HR, hazard ratio; CI, confidence interval; GB, gallbladder.

3. Subgroup analysis

Subgroup analyses were conducted according to sex and baseline LDL-C levels (Fig. 2). The association between statin use and overall cancer risk was consistent across sexes, with no evidence of effect modification (p for interaction=0.316). Similarly, the reduced risk of liver cancer and increased risk of thyroid cancer associated with statin use were consistent in males and females (p for interaction=0.973 and 0.699, respectively). However, a significant interaction by sex was observed for pancreatic cancer, with a significant risk reduction observed only among males (aHR, 0.72; 95% CI, 0.58–0.91; p=0.005; p for interaction=0.032).

Fig. 2. Subgroup analyses of the association between statin use and site-specific cancer risk stratified by sex and baseline LDL-C levels. (A) Subgroup analysis by sex (male vs. female). (B) Subgroup analysis by baseline LDL-C levels (<160 mg/dL vs. ≥160 mg/dL). Forest plots show the multivariate-adjusted HRs (squares) with 95% CIs (horizontal lines). Statin non-users served as the reference group.

GB, gallbladder; IR, incidence rate (per 1,000 person-years); HR, hazard ratio; CI, confidence interval; LDL-C, low-density lipoprotein cholesterol; BMI, body mass index.

In subgroup analyses stratified by baseline LDL-C levels, statin use was associated with a significantly reduced risk of overall cancer among participants with baseline LDL-C <160 mg/dL (aHR, 0.95; 95% CI, 0.92–0.99; p=0.018), whereas no significant association was observed among those with LDL-C ≥160 mg/dL (p for interaction=0.320). In addition, both liver and pancreatic cancer risks were significantly lower among participants with LDL-C <160 mg/dL, with no significant associations observed in the higher LDL-C group (liver cancer: aHR, 0.51; 95% CI, 0.43–0.61; p<0.001; p for interaction=0.508; pancreatic cancer: aHR, 0.82; 95% CI, 0.69–0.98; p=0.031; p for interaction=0.293). For thyroid cancer, a significant interaction by LDL-C level was observed (p for interaction=0.030). Statin use was associated with a marked increase in the risk of thyroid cancer in both LDL-C strata, with a stronger association observed among participants with LDL-C ≥160 mg/dL (aHR, 1.52; 95% CI, 1.19–1.93; p=0.001) than those with LDL-C <160 mg/dL (aHR, 1.13; 95% CI, 1.01–1.28; p=0.040).

DISCUSSION

This large-scale, nationwide cohort study using the Korean NHIS database investigated the association between statin use and the risk of site-specific cancers. The findings demonstrate that statin use is associated with a divergent risk profile depending on the cancer type. Specifically, statin use was associated with a significant reduction in liver cancer risk and a marginally significant reduction in pancreatic cancer risk, particularly among men. In contrast, an increased risk of thyroid cancer was observed among statin users. After adjusting for potential confounders, statin use was also associated with a modest but statistically significant reduction in overall cancer incidence.

A major strength of this study is its longitudinal population-based cohort design, which distinguishes it from most existing literature that relies on case–control studies.¹³,¹⁴ Notably, case–control designs are inherently vulnerable to recall bias and challenges in establishing temporal causality.¹⁵,¹⁶ In contrast, our nationwide population-based cohort allowed for the assessment of statin exposure before cancer diagnosis and enabled comprehensive adjustment for key confounders, including age, BMI, smoking status, and alcohol consumption, thereby providing a higher level of evidence. Moreover, rather than focusing on a single malignancy, the present study provides an integrated evaluation of statin use across multiple cancer types within the same population, allowing direct comparison of organ-specific effects—most notably reduced risk in liver cancer and increased risk in thyroid cancer—within a consistent analytical framework.

The most pronounced association observed was the substantial reduction in liver cancer risk among statin users (aHR, 0.52). This finding is consistent with the results of previous meta-analyses and cohort studies supporting the chemopreventive potential of statins in hepatocellular carcinoma (HCC). For instance, a 2020 meta-analysis of observational studies reported a 46% reduction in HCC risk among statin users (risk ratio, 0.54).¹⁷ Similarly, a recent nationwide cohort study from Taiwan involving patients with heart failure demonstrated a significant reduction in liver cancer risk regardless of statin lipophilicity.¹⁸ The concordance between our findings (HR, 0.52) and these global estimates suggests that the inhibition of the mevalonate pathway and attenuation of hepatic inflammation may underlie the protective effect of statins on liver carcinogenesis.

Interestingly, the protective effect of statins against pancreatic cancer was confined to men (aHR, 0.72; 95% CI, 0.58–0.91) and was not observed in women. This sex-specific difference may be attributed to several factors. First, the substantially higher prevalence of smoking and heavy drinking among men in the Korean population might have enhanced the detectable anti-inflammatory and antioxidant effects of statins.¹⁹ Recent epidemiological evidence indicates a stark gender gap in smoking rates in Korea, with a prevalence of 30.0% to 35.7% in men compared to only 5.0% to 6.7% in women.¹⁹ Given that smoking is a potent, dose-dependent risk factor for pancreatic cancer, the ‘pleiotropic’ benefits of statins in modulating the tumor microenvironment may be more pronounced in this high-risk group.²⁰,²¹ Second, sexual dimorphism in pancreatic carcinogenesis and the sex hormones present in systemic circulation can have an impact on both tumor development and the tumor response to therapy in pancreatic cancer.²² Lastly, the lower absolute incidence rate of pancreatic cancer in women may have reduced the statistical power to reach significance.²³ Future studies are warranted to further elucidate the sex-specific mechanisms underlying the chemopreventive effects of statins in pancreatic cancer. This sex-specific association is supported by prior evidence, including a recent Japanese cohort study that reported a significantly reduced pancreatic cancer incidence among statin users (HR, 0.84) and a large case–control study that identified a protective effect confined to men.¹⁴,²⁴ Our findings support this protective trend and suggest that the male sex-specific benefit may be linked to the modulation of higher baseline risk factors, such as smoking status, which are more prevalent in male populations.

Conversely, this study identified a significant positive association between statin use and thyroid cancer incidence (aHR, 1.19). This finding contrasts with several previous reports that suggested null or even protective effects of statins on thyroid cancer risk.²⁵,²⁶ This discrepancy likely arises from differences in study design and exposure assessment. While the previous studies utilized a case-control design—which is inherently limited in establishing temporal causality—our study adopted a longitudinal cohort approach using a time-dependent Cox model. Although this discrepancy has often been attributed to surveillance bias (detection bias)—given that statin users undergo more frequent medical evaluations, including thyroid ultrasonography—this explanation has inherent limitations. If increased surveillance were the primary determinant, a parallel rise in the diagnosis of other malignancies would be expected. However, our findings demonstrated an overall reduction in cancer risk and site-specific decreases in liver cancer, indicating that the observed increase is highly specific to thyroid cancer and unlikely to be explained solely by enhanced screening. Emerging evidence points to a more complex biological relationship between lipid metabolism and thyroid carcinogenesis. Recent epidemiological studies have reported inverse associations between serum cholesterol levels and thyroid cancer risk, with lower total cholesterol or LDL-C levels associated with increased incidence or aggressiveness of thyroid malignancy.²⁷ In addition, a nationwide population-based study in Korea reported that persistently low high-density lipoprotein-cholesterol levels were associated with an increased risk of thyroid cancer.²⁸ Given that statins potently alter lipid profiles, it is biologically plausible that changes in the lipid microenvironment or downstream metabolic signaling may paradoxically promote thyroid tumorigenesis. Moreover, statin-associated effects, such as increased insulin resistance—a recognized risk factor for thyroid cancer—may further contribute to this association.²⁵ Collectively, these findings suggest that the increased thyroid cancer risk observed in this study may likely reflect a combination of heightened diagnostic intensity and potential lipid-mediated biological mechanisms.

Despite these strengths, this study has some limitations. First, the observational analysis of administrative claims data hinders making causal inferences. Although we used multivariable-adjusted models with a wide array of covariates to minimize confounding, the possibility of residual confounding remains, as this was an observational study. Further studies utilizing propensity score-matched cohorts or randomized controlled designs could provide additional validation of our findings. Second, statin exposure was ascertained from prescription records, which may not fully capture medication adherence, potentially resulting in exposure misclassification. Third, cumulative dose–response relationships were not assessed, and lipophilic and hydrophilic statins were not differentiated, limiting pharmacological mechanistic interpretation. Fourth, information on potential unmeasured confounders, such as family history of cancer and detailed dietary factors, was unavailable in the NHIS database. Fifth, our study excluded individuals with BMI <18.5 kg/m² to ensure internal validity and reduce reverse causality; however, this might limit the generalizability of our results to the underweight population. Finally, although we carefully considered surveillance bias in interpreting thyroid cancer risk, its precise contribution relative to true biological incidence cannot be fully quantified in a retrospective context.

In conclusion, this large-scale nationwide cohort study provides robust evidence that statin use is associated with a divergent cancer risk profile in the Korean population, characterized by reduced risks of liver cancers and an increased risk of thyroid cancer. These findings underscore the complexity and organ specificity of statins’ pleiotropic effects. While the observed reduction in liver cancer supports the chemopreventive potential of statins, the increased risk of thyroid cancer warrants cautious interpretation and further investigation. Future prospective studies and mechanistic research are needed to clarify the biological pathways underlying these divergent associations and refine clinical guidelines for statin therapy in cancer-specific contexts.