Abstract

Background:

Epidemiologic evidence on the association between infertility and thyroid cancer risk remains inconsistent. We conducted a systematic review and meta-analysis to synthesize available observational evidence and quantify this association.

Methods:

We systematically searched PubMed, Web of Science, and Scopus from inception to February 10, 2026, to identify observational studies evaluating the association between infertility and thyroid cancer risk. Eligible cohort and case–control studies reporting relative effect estimates with 95% confidence intervals (CIs) were included. Summary risk estimates were pooled using random-effects models. Between-study heterogeneity was assessed using the I² statistic. Prespecified subgroup analyses were conducted by sex and study design. Sensitivity analyses and assessments of publication bias were also performed.

Results:

Nine studies met the inclusion criteria, including seven cohort studies and two case–control studies, comprising a total of 4,563,090 participants. In the pooled analysis, infertility was associated with a significantly higher risk of thyroid cancer (relative risk [RR], 1.37; 95% CI, 1.15–1.63), with substantial heterogeneity across studies (I² = 81.9%). Sex-stratified analyses suggested a stronger association among men (RR, 1.53; 95% CI, 1.43–1.65) than among women (RR, 1.31; 95% CI, 1.04–1.66), although the test for interaction was not statistically significant. The findings were robust across sensitivity analyses. Evidence of small-study effects was observed; however, the association remained statistically significant after adjustment using the trim-and-fill method.

Conclusions:

This study suggests that infertility is associated with an increased risk of thyroid cancer. Although substantial heterogeneity was observed, the association remained consistent across multiple sensitivity analyses. However, further well-designed research is needed to verify and clarify this association and its underlying mechanisms.

1 Introduction

Thyroid cancer is one of the most common endocrine malignancies worldwide and disproportionately affects women, with incidence peaking during the reproductive years (1). Although ionizing radiation and genetic susceptibility are well-established risk factors, the sustained rise in thyroid cancer incidence over recent decades has renewed interest in the role of reproductive and endocrine influences (2–4). In parallel, infertility has become an increasingly prevalent public health issue, affecting an estimated 9–18% of couples globally, alongside expanding use of ovulation induction and assisted reproductive technologies (ART) (5). These converging trends have raised concern that infertility itself—and the hormonal disturbances inherent to both its underlying biology and treatment—may be associated with altered thyroid cancer risk (4, 6).

Biological plausibility for such an association is supported by shared endocrine pathways. Estrogen receptors are expressed in normal and malignant thyroid tissue, and experimental and clinical evidence suggests that estrogen signaling may promote thyroid cell proliferation and tumorigenesis (7, 8). Fertility treatments can also induce transient but substantial alterations in estrogen, progesterone, gonadotropins, and thyroid-stimulating hormone, potentially influencing thyroid growth and carcinogenic processes (9). Moreover, infertility frequently coexists with autoimmune and metabolic conditions that may independently modify thyroid cancer risk, complicating etiologic interpretation (10).

Despite these mechanistic considerations, epidemiologic evidence remains inconsistent. Some population-based cohort studies have reported higher thyroid cancer incidence among women with infertility. For instance, a large Taiwanese cohort observed an adjusted incidence rate ratio (IRR) of 1.80 overall, with a more pronounced excess after seven years of follow-up (IRR 4.39) (11). However, other observational studies have reported null or attenuated associations, underscoring substantial heterogeneity in findings (12–14).

Previous meta-analyses have largely focused on fertility treatments rather than infertility itself. A 2018 meta-analysis reported an increased thyroid cancer risk among infertile women exposed to fertility drugs, particularly clomiphene citrate (15), while more recent syntheses yielded similar findings but found no clear association for in vitro fertilization or ART (16). These analyses, however, often conflate infertility diagnoses with pharmacologic or procedural exposures, limiting inference regarding the independent role of infertility.

Given the growing prevalence of infertility and the rising global burden of thyroid cancer, a focused synthesis is warranted. Accordingly, we conducted a systematic review and meta-analysis of observational studies to quantify the association between infertility and thyroid cancer risk, explore potential sex-specific differences, and assess between-study heterogeneity.

2 Methods2.1 Study design and eligibility criteria

This systematic review and meta-analysis were conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (17). The study focused on observational evidence examining the association between infertility and thyroid cancer risk, structured according to the PICOS framework: (1) Population (P): Adults of reproductive age (both men and women) with infertility, defined according to study-specific criteria, including clinical diagnosis, self-reported infertility, or documented fertility treatment. (2) Exposure (E): Infertility or fertility-related exposures, including primary or secondary infertility, regardless of etiology. (3) Comparator (C): Individuals without infertility or those not exposed to fertility-related factors. (4) Outcome (O): Primary outcome was incidence of thyroid cancer (all histologic subtypes). Secondary outcomes included thyroid cancer subtypes or sex-specific associations where reported. (5) Study design (S): Observational studies, including cohort, case-control, and nested case-control studies reporting risk estimates (RR, OR, HR) with 95% confidence intervals (CIs) or sufficient data for calculation. We excluded: studies without a comparison group; studies focusing exclusively on infertility treatments (e.g., assisted reproductive technology) without a clear infertility definition; reviews, editorials, case reports, conference abstracts without sufficient data; studies lacking effect estimates or sufficient data to calculate them.

2.2 Literature search, data extraction, and quality assessment

We performed a comprehensive literature search of PubMed, Web of Science, and Scopus from database inception to February 10, 2026. Search terms combined keywords and MeSH terms for “infertility” and “thyroid cancer,” including relevant synonyms (e.g., “subfertility,” “sterility,” “thyroid carcinoma,” “thyroid neoplasm”). No language restrictions were applied. Reference lists of included studies and relevant reviews were screened to identify additional eligible studies. The detailed search strategies for each database are provided in Supplementary Table 1.

Two reviewers (XQ and XT) independently screened titles, abstracts, and full texts, with disagreements resolved through discussion or consultation with a third reviewer (JY). Data extracted included author, year, country, study design, sample size, participant characteristics, infertility definition, outcome type, effect estimates, and adjusted confounders. Study quality was assessed using the Newcastle–Ottawa Scale (NOS) (18), which evaluates selection, comparability, and exposure/outcome assessment. Studies scoring ≥7 was considered high quality.

2.3 Statistical analysis

Effect estimates were harmonized as RRs for meta-analysis. Pooled RRs and 95% CIs were calculated using random-effects models (DerSimonian–Laird), with fixed-effect models applied for sensitivity analyses. Heterogeneity was assessed using Cochran’s Q test and the I² statistic (I² >50% indicating substantial heterogeneity) (19). Subgroup analyses were performed by sex and study design. Sensitivity analyses included leave-one-out and influence analyses (20). Publication bias was assessed using funnel plots, Egger’s and Begg’s regression test (21), and the trim-and-fill method (22). To explore potential sources of heterogeneity, we conducted subgroup analyses according to the definition of infertility used in the original studies (reproductive/biological definition, medical record–based diagnosis, self-reported infertility, or not reported). Moreover, a multivariable meta-regression analysis was conducted to examined several factors that could contribute to variability across studies, including sex distribution, study design, geographic region, and the definition of infertility. All analyses were performed using R (version 4.3.0) with the meta and metafor packages. Two-sided p-values <0.05 were considered statistically significant.

3 Results3.1 Study selection and characteristics

The literature search identified 655 records, of which 9 studies met the inclusion criteria, involving 4,563,090 participants (11–14, 23–27) (Figure 1). These included 7 cohort studies (5 retrospective (11, 13, 23, 24, 26) and 2 prospective (12, 25)) and 2 case-control studies (14, 27). Overall, the studies were conducted predominantly in the United States (n = 6) (12, 13, 23, 24, 26, 27), with additional studies from Europe (n = 2) (14, 25) and China (n = 1) (11). Across studies, infertility was defined variably, including clinical diagnosis, self-reported infertility, or exposure to fertility treatment. Most studies adjusted for age and parity (8 of 9 studies) (11–14, 24–27), although adjustment for hormonal factors, autoimmune thyroid disease, and family history was inconsistent (Table 1). All included studies were rated as high quality according to the Newcastle–Ottawa Scale (score ≥7; Tables 2, 3).

Flow diagram of study selection for the meta-analysis of infertility and thyroid cancer risk. Numbers indicate the studies identified, screened, assessed for eligibility, and included at each stage.

AuthorYearCountryStudy designSample sizeFollow upAgeInfertility typeInfertility diagnosis criteriaThyroid cancer typeEffect estimate (95% CI)Covariates adjustedRamsay JM, et al (23)2024USARetrospective cohort426 azoospermia men and their families, and 3,105 fertile men and their families51 yearsMedian 40-49 yearsMale infertility (Azoospermia)sperm concentration of 0 million/mlAll TC (80 cases in infertile group and 396 cases in non-infertile group)HR 1.54 (1.21-1.97)/Wang S, et al (12)2023USAProspective cohort26,208 infertile women and 76,872 non-infertile womeninfertile women (613,813 person-years) and non-infertile women (1,535,572 person-years)Median 35 yearsFemale infertility (28.8% due to ovulatory disorders)failure to conceive after one year of tryingAll TC (142 cases in infertile group and 327 cases in non-infertile group)HR 1.06 (0.86-1.31)Age, family history of cancer, race, BMI, age at menarche, oral contraceptive use before age 18, time-varying smoking status, physical activity, AHEI, marital status, recent health seeking behavior, hormonal therapy, gravidity, time-varying gravidity and age at first pregnancyMurugappan G, et al (24)2019USARetrospective cohort64,345 infertile women and 3,128,345 non-infertile womeninfertile women (3.8 ± 3.3 years) and non-infertile women (3.9 ± 3.3 years)infertile group (34.0 ± 5.7 years) and non-infertile group (32.7 ± 7.4 years)Female infertilityWomen were considered to have infertility if they had a diagnosis of infertility (ICD-9: 628.x, 614.6, V26.89; ICD-10: E23.0, N73.6, N97.x, Z31.81), underwent fertility testing (diagnosis codes V26.21, Z31.41), or received fertility treatment, such as a hysterosalpingogram (HSG, CPT 74740).All TC (138 cases in infertile group and 4869 cases in non-infertile group)HR 1.29 (1.09-1.53)Age at index date, index year, nulliparity, race, number of visits per year and highest level of educationDing DC, et al (11)2019ChinaRetrospective cohort13,356 infertile women and 53,424 non-infertile womeninfertile women (94809 person-years) and non-infertile women (372944 person-years)infertile group (31.1 ± 5.26 years) and non-infertile group (31.1 ± 5.60 years)Female infertilityinfertility was identified by ICD9: 628.x, 614.6, V26.89All TC (27 cases in infertile group and 57 cases in non-infertile group)IRR 1.80 (1.70-1.92)Age, comorbidity, and medication useZamora-Ros R, et al (25)2015FranceProspective cohort345,157 womeninfertile women (76,530 person-years) and non-infertile women (1,968,705 person-years)Mean 51 yearsFemale infertilityself-reportDifferentiated TC (24 cases in infertile group and 327 cases in non-infertile group)HR 1.70 (1.12-2.60)Age and countryEisenberg ML (26),2015USARetrospective cohort76083 infertile men and 760830 non-infertile meninfertile men (277,703 person-years) and not report for non-infertile groupinfertile group (35.08 ± 5.89 years)Male infertilityinfertility was identified by ICD9: 606.x, V26.21 or by the presence on any claim of a procedure code (CPT) for fertility testing or semen analysis/semen preparation (89300, 89310, 89320, 89321, 89322, 89325, 89329, 89330, 89331).All TC (27 cases in infertile group and 170 cases in non-infertile group)HR 1.52 (1.01-2.30)Age, year of evaluation, comorbidity and follow up timeBrinton LA, et al (13)2005USARetrospective cohort8422 infertile women and External controlinfertile women (155527 person-years)Median 30 yearsFemale infertilityMedical recordAll TC (18 cases observed in infertile group and 18.1 cases expected in general population)SIR 0.99 (0.60-1.60)Age, race, and calendar yearNegri E, et al (14)1999ItalyCase-control2247 female cases of TC (80% papillary) and 3699 control women/Not reportFemale infertilityNot reportAll TCOR 1.20 (0.90-1.60)Age and history of radiationMcTiernan A, et al (27)1987USACase-control182 female cases of TC and 389 control women/Not reportFemale infertilityself-reportAll TCOR 0.81 (0.45-3.10)Age and weight

Key characteristics of studies included in the meta-analysis of infertility and thyroid cancer (TC) incidence.

AuthorYearSelectionComparabilityOutcomeScoreRepresentativeness of exposed cohortSelection of non-exposed cohortExposure AscertainmentOutcome present at start of studyStudy controls for ageStudy controls for any additional important factorAssessment of OutcomeLength of follow-upAdequacy of follow-upRamsay JM, et al (23)2024★★★★★★★★8Wang S, et al (12)2023★★★★★★★★★9Murugappan G, et al (24)2019★★★★★★★★★9Ding DC, et al (11)2019★★★★★★★★★9Zamora-Ros R, et al (25)2015★★★★★★★★★9Eisenberg ML (26),2015★★★★★★★★★9Brinton LA, et al (13)2005★★★★★★★7

Quality assessment of included cohort studies by Newcastle-Ottawa Scale.

Cohort Studies: Selection ① Representativeness of the exposed cohort★ Selection of the non-exposed cohort★ ② Ascertainment of exposure★ ③ Demonstration that outcome of interest was not present at start of study★. Comparability ① Comparability of cohorts on the basis of the design or analysis★★. Outcome ① Assessment of outcome★ ② Was follow-up long enough for outcomes to occur★ ③ Adequacy of follow up of cohorts★.

AuthorYearSelectionComparabilityExposureScoreAdequacy of the case definitionRepresentativeness of casesChoice of controlsDefinition of controlStudy controls for ageStudy controls for any additional important factorExposure assessmentThe method of exposure assessmentNon-response rateNegri E, et al (14)1999★★★★★★★★8McTiernan A, et al (27)1987★★★★★★★★8

Quality assessment of included case-control studies by Newcastle-Ottawa Scale.

Case-control Studies: Selection ① Adequacy of the case definition★ ② Representativeness of cases★ ③ Choice of controls★ ④ Definition of control★. Comparability ① Comparability of case-controls on the basis of the design or analysis★★. Exposure ① Exposure assessment★ ② The method of exposure assessment★ ③ Non-response rate★.

3.2 Association between infertility and thyroid cancer risk

In pooled analyses using a random-effects model, infertility was associated with a 37% increased risk of thyroid cancer (RR, 1.37; 95% CI, 1.15–1.63; Figure 2). Between-study heterogeneity was substantial (I² = 81.9%; P < 0.001). Results were similar under a fixed-effect model. In sex-stratified analyses, the association appeared stronger among men (RR, 1.53; 95% CI, 1.43–1.65) than among women (RR, 1.31; 95% CI, 1.04–1.66), although the test for interaction was not statistically significant (Figure 2). When stratified by study design, retrospective cohort studies yielded a pooled RR of 1.48 (95% CI, 1.15–1.90). The two prospective cohort studies showed an imprecise estimate (RR, 1.29; 95% CI, 0.07–25.09). Case-control studies demonstrated a pooled OR of 1.16 (95% CI, 0.30–4.56), with wide confidence intervals and no statistically significant association (Figure 3). Subgroup analyses according to infertility definition are presented in Supplementary Figure 1. Although effect sizes varied slightly across categories, the direction of the association was broadly consistent. The test for subgroup differences was not statistically significant under the random-effects model (p = 0.7994).

Forest plot of the association between infertility and thyroid cancer risk, stratified by male and female infertility. Diamonds represent pooled estimates from common-effect and random-effects models, with widths indicating 95% confidence intervals (CIs). Heterogeneity within subgroups and overall was assessed using I² and Cochran’s Q. Subgroup differences between male and female infertility were evaluated under both models.

Forest plot of the association between infertility and thyroid cancer risk, stratified by study design. Diamonds indicate pooled estimates from common-effect and random-effects models, with widths representing 95% confidence intervals (CIs). Subgroup and overall heterogeneity were assessed using I² and Cochran’s Q. Differences across study designs were evaluated under both models.

3.3 Sensitivity analyses and publication bias

Leave-one-out analyses indicated that exclusion of any single study did not materially alter the pooled estimate (RR range, 1.27–1.68), supporting the robustness of the primary findings (Figure 4). Visual inspection of the funnel plot suggested possible asymmetry (Figure 5A). Egger’s regression test indicated evidence of small-study effects (t = −2.83; P = 0.026), whereas Begg’s test did not detect significant asymmetry (z = 0.42; P = 0.68). Using the trim-and-fill method, five potentially missing studies were imputed (Figure 5B). The adjusted pooled estimate remained statistically significant (RR, 1.73; 95% CI, 1.35–2.22; P = 0.0004). However, between-study heterogeneity remained substantial (I² = 87.7%; 95% CI, 81.1%–92.0%).

Leave-one-out sensitivity analysis of infertility and thyroid cancer risk. Each line shows the 95% confidence interval (CI) of the pooled risk ratio (RR) after excluding the study on the y-axis, with points representing the pooled RR. The red dashed line shows the pooled RR including all studies, and the black dotted line indicates RR = 1.

Funnel plots evaluating publication bias in the meta-analysis of infertility and thyroid cancer risk. The left panel shows study-specific risk ratios (RRs) plotted against their standard errors (A), and the right panel shows the trim-and-fill–adjusted plot (B). The vertical dashed line indicates the pooled effect, and diagonal lines represent pseudo 95% confidence limits. The trim-and-fill method assesses the potential impact of small-study effects and publication bias.

3.4 Meta-regression

Female infertility was not significantly associated with thyroid cancer risk compared with male infertility (β = −0.1923, 95% CI: −3.0544 to 2.6698; P = 0.5501; Supplementary Table 2). No significant effect modification was observed by geographic region, with studies from East Asia (β = 0.3614, 95% CI: −0.7496 to 1.4724; P = 0.1511) and West Europe (β = 0.4327, 95% CI: −3.1695 to 4.0349; P = 0.6920) showing no differences compared with North America. Similarly, study design did not influence the association, as prospective cohort (β = −0.1812, 95% CI: −3.7234 to 3.3611; P = 0.6331) and case-control studies (β = −0.4898, 95% CI: −5.3543 to 3.3746; P = 0.4223) were comparable to retrospective cohorts. The definition of infertility was also not associated with differences in effect estimates, with medical record–based and self-reported measures yielding similar results to reproductive/biological definitions. Overall, these findings indicate that sex, region, study design, and infertility definition did not significantly explain between-study heterogeneity.

4 Discussion

In this systematic review and meta-analysis of nine observational studies including over 4.5 million participants, we found that infertility was associated with a 37% increased risk of thyroid cancer. Although between-study heterogeneity was substantial, the positive association persisted across sensitivity analyses and remained statistically significant after adjustment for potential small-study effects. Our pooled estimate helps contextualize and reconcile the heterogeneous findings reported by individual studies. Several large cohort studies have suggested an elevated thyroid cancer risk among infertile individuals (11, 23–26), whereas others reported null or modest associations (12–14, 27). By synthesizing available evidence, our analysis provides a more precise and stable estimate, suggesting that infertility itself—rather than isolated findings in specific populations—may be associated with increased thyroid cancer incidence. Importantly, this association persisted despite substantial heterogeneity and evidence of small-study effects. The trim-and-fill–adjusted estimate remained statistically significant and even strengthened, indicating that the observed association is unlikely to be entirely explained by publication bias.

Our results extend prior literature by focusing specifically on infertility rather than fertility treatment exposure alone. Previous meta-analyses primarily evaluated fertility drugs—particularl