Prostate cancer (Prostate cancer, PCa) is the second most common malignant tumor among men worldwide and is also one of the important causes of cancer-related death. With the aging of the population and changes in lifestyle, the incidence of prostate cancer has continued to increase globally, especially in China, where the number of new and fatal cases has shown an increasing trend year by year in recent years, seriously threatening men’s health and causing a huge social and economic burden (1, 2). Epidemiological data show that prostate cancer is more common in people aged 65 years and above, and its risk is influenced by multiple factors, including genetics, age, hormone levels, lifestyle, etc. Although prostate-specific antigen (PSA) testing has played a certain role in early screening and diagnosis, its specificity and sensitivity are limited, and there is still a need to find more effective molecular biomarkers to achieve early diagnosis and treatment and reduce the disease burden (1, 2).

Insulin-like growth factor-1 (Insulin-like growth factor-1, IGF-I) is a polypeptide hormone secreted by the liver, its structure is similar to insulin, and it plays an important regulatory role in various physiological processes such as cell proliferation, differentiation, and apoptosis. IGF-I binds to its receptor (IGF1R), activates a series of downstream signaling pathways, promotes cell growth and survival, and inhibits programmed cell death. Existing studies have shown that IGF-I is not only essential in normal development and metabolism, but also participates in the occurrence and development of tumors, including promoting tumor cell growth and metastasis through mechanisms such as regulating the cell cycle, promoting angiogenesis, and enhancing anti-apoptotic ability (2–4). In prostate tissue, high expression of IGF-I and its receptor is considered to be closely related to the occurrence and progression of prostate cancer, and some genetic variations (such as SNPs in the 3’UTR region of IGF1R) may also regulate IGF-I signaling by affecting miRNA binding, thereby changing individual susceptibility (3).

Although a large number of epidemiological and basic studies support the correlation between IGF-I and the risk of prostate cancer, there is still considerable controversy regarding the conclusions about the relationship between the two. On the one hand, differences in IGF-I level measurement methods, sample size, follow-up time, and regional differences among different study populations lead to inconsistent results; on the other hand, some previous meta-analyses had limited sample sizes and ignored the heterogeneity of genetic background and environmental exposure (2, 3, 5). Whether there is a causal relationship between IGF-I and prostate cancer or merely a correlation phenomenon still awaits higher-quality evidence. In recent years, some large-sample cohort, Mendelian randomization, and functional studies have gradually revealed the molecular mechanisms of the IGF-I signaling axis in tumorigenesis, suggesting that it may become a potential target for tumor prevention and early screening, but there are still gaps in evidence and unclear mechanisms (1, 4, 5).

This study, based on epidemiological evidence of serum IGF-I levels and prostate cancer risk among different populations worldwide, combined with the latest large-sample, multicenter original studies and genetic epidemiological results, explores the application value of IGF-I as a potential biomarker in early screening and risk stratification of prostate cancer. It is hoped that this study will provide a more solid evidence base for clinical prevention, individualized screening, and molecular typing of prostate cancer, and contribute theoretical foundations and practical references to reduce the disease burden and improve patient outcomes.

This study was conducted in strict accordance with the reporting standards for systematic reviews and meta-analyses (PRISMA statement), and systematically searched international mainstream medical databases such as PubMed, Web of Science, and Embase. There were no restrictions on the time frame or language, and all relevant literature published since the establishment of each database was included. The search strategy combined subject terms and free words, with the main keywords including “IGF-I”, “insulin-like growth factor-1”, “prostate cancer”, “risk”, etc. (see Supplementary Table 1 for the specific search strategy). At the same time, the references of the included literature were traced to supplement any missing relevant studies. All literature retrieval, screening, and inclusion processes were independently carried out by four researchers, and any disagreements were resolved through discussion among all authors. This study has been registered on the PROSPERO platform, with the registration number (CRD420251174259).

Inclusion criteria:

Exclusion criteria:

Key information from the included studies was independently extracted by four researchers, including: first author, year of publication, country or region, study design type, sample size, age of participants, follow-up duration, mean serum IGF-I, effect values and their 95% confidence intervals (OR, RR or HR and 95% CI), as well as the main adjusted (corrected) factors. If data in the literature were incomplete, the authors were contacted or the supplementary tables were consulted to obtain the required information. In case of disagreement on data, all researchers reviewed and discussed to ensure the accuracy and consistency of the data.

The included non-randomized intervention studies were assessed for risk of bias using the internationally recognized Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool, which can systematically evaluate the risk of bias in each domain of non-randomized studies. Included Mendelian randomization studies were evaluated using the Risk Of Bias in Mendelian Randomization Studies (ROB-MR) tool, which is currently the most authoritative risk assessment method for Mendelian randomization studies internationally. All quality assessments were completed independently by two researchers and reviewed by the first author; any disagreements were resolved through discussion among all researchers to ensure the objectivity and consistency of the evaluation.

All analyses in this study used a random-effects model to combine the effect estimates from the included studies. Heterogeneity among studies was assessed using the Cochrane Q test and the I² statistic. Sensitivity analysis was performed to evaluate the robustness of the results by sequentially removing each study and recombining the effect estimates. To explore potential influencing factors, subgroup analyses were conducted according to variables such as year of publication, study type, sample size, age, and serum IGF-I, and meta-regression analysis was carried out to further assess the impact of related covariates on the results. Publication bias was assessed by visual inspection of funnel plots and quantitatively evaluated using Egger’s test and Begg’s test. All data analyses were performed using R software (version 4.4.3), and literature management and deduplication were performed using Zotero 6.0.37.

A total of 1,479 articles were retrieved in this study, and 1,062 remained after removing duplicates. After preliminary screening of titles and abstracts, 99 articles entered the full-text review stage. Among them, 10 articles were excluded because the full text could not be obtained. The remaining 89 articles entered the eligibility assessment stage and were further excluded for the following reasons: study subjects did not meet the inclusion criteria (such as non-adults, non-human studies, etc.); relevant outcome indicators were not reported; data were incomplete or could not be extracted. A total of 16 studies met the inclusion criteria and were included in the meta-analysis. For the specific screening process and reasons for exclusion at each stage, see Figure 1.

A total of 16 studies were included, comprising prospective nested case-control studies (NCC), case-control studies (CC), population-based case-control studies (PBCC), and Mendelian randomization studies (MR). The study subjects covered multiple countries and regions including the United States, Europe, the United Kingdom, and India. The sample sizes ranged from 50 cases to over 290,000 cases, with the maximum average follow-up time up to 18 years. The ages of participants in each study were mainly concentrated in middle-aged and elderly men, and the mean serum IGF-I levels varied greatly. All studies reported effect values for serum IGF-I and prostate cancer risk, along with their 95% confidence intervals. For the specific basic information of the included studies, see Table 1 and Supplementary Table 2.

The ROBINS-I tool was used to assess the risk of bias in the included non-randomized intervention studies. Most studies had a moderate risk in terms of confounding factors (Moderate risk), while the remaining domains were mostly assessed as low risk (Low risk). The overall risk of bias was mainly concentrated in the control of confounding factors. For the included Mendelian randomization studies, the ROB-MR tool was used for evaluation. The two studies were both assessed as having a moderate risk of bias (Moderate risk), mainly limited by instrument variable independence, sample overlap, and pleiotropy. For detailed evaluation results of each dimension in each study, see Supplementary Table 3. The results of the risk of bias assessment are shown in Figure 2.

A random-effects model was used to combine the analysis of the relationship between serum IGF-I levels and the risk of prostate cancer. The results showed that higher serum IGF-I levels were associated with an increased risk of prostate cancer, with a combined odds ratio (OR) of 1.10 (95% CI: 1.02–1.18, P = 0.0136), which was statistically significant. The heterogeneity test result was I² = 50.6%, and the Q test P = 0.0108, indicating a moderate degree of heterogeneity among the studies. For the effect values and confidence intervals of each study, see Figure 3.

To further explore the sources of heterogeneity between serum IGF-I and the risk of prostate cancer, this study conducted subgroup analyses on variables such as publication time, study design type, and sample size. The numerical results are presented in Table 2, and the corresponding subgroup forest plots are provided as Supplementary Figures 1–6.

In summary, subgroup analysis found that the association between high serum IGF-I and increased risk of prostate cancer was more prominent in studies published in the past decade and in NCC designs, but heterogeneity increased with the expansion of sample size and research background. It should be noted that although the main effect analysis reached statistical significance, the combined effects of some subgroups (such as large samples, certain study types) were not statistically significant. This phenomenon is relatively common in meta-analysis, mainly due to the reduction of sample size and statistical power after subgroup splitting, as well as the increased heterogeneity brought by large-sample or multicenter studies. It is worth noting that the effect direction of the vast majority of subgroups is consistent with the main analysis, suggesting that the overall association may truly exist, but the specific effect strength and consistency still need to be further verified by larger sample size and higher quality studies. The above results indicate that, although the overall evidence supports high IGF-I as a potential risk factor for prostate cancer, different study characteristics affect the effect size and consistency, and it is still necessary to combine sensitivity analysis and meta-regression to further explain the sources of heterogeneity.

To further explore the sources of heterogeneity in the association effect between serum IGF-I and the risk of prostate cancer, this study conducted meta-regression analyses using the mean age of study subjects and the mean serum IGF-I value as covariates (Figure 4).

When using the mean age of study subjects as a covariate for meta-regression analysis, the results showed that the explanatory power of the age variable on the combined effect among studies was extremely limited (R²=0.00%), with a regression coefficient of 0.0005 and P = 0.9526, which was not statistically significant. The overall model test (QM = 0.0035, P = 0.9526) was also not significant, suggesting that there was no significant linear relationship between the average age of different studies and the effect size of the association between IGF-I and prostate cancer risk. In addition, the residual heterogeneity remained high (I²=54.06%, QE = 29.90, P = 0.0079), indicating that the age variable could not effectively explain most of the effect differences between studies. In other words, the variation in average age among the included studies did not statistically significantly affect the association strength between IGF-I and prostate cancer risk. Therefore, the current evidence does not support average age as a main source of heterogeneity in this study.

The meta-regression analysis with mean serum IGF-I as a covariate also showed that its explanatory power on the combined effect among studies was likewise extremely limited (R²=0.00%), with a regression coefficient of -0.0002 and P = 0.8121, which was not statistically significant. The overall model test (QM = 0.0565, P = 0.8121) was also not significant, and the residual heterogeneity remained high (I²=53.11%), further suggesting that the mean IGF-I value was not a major factor affecting the effect differences. The above results indicate that no significant linear relationship was observed between changes in mean serum IGF-I across different studies and the combined effect on prostate cancer risk.

It is important to emphasize that meta-regression can only provide an exploratory analysis of the statistical relationship between covariates and effect values, and its explanatory power and robustness are limited, especially when the number of included studies is small. Although this study attempted to include major demographic and biological indicators that might influence the effect differences, the current analysis did not find that average age or mean IGF-I had a significant explanatory effect on heterogeneity. Combined with the results of the residual heterogeneity tests, this suggests that other potential covariates (such as region, ethnicity, study design, detection method, lifestyle, etc.) may be involved in influencing the effect differences between studies.

In summary, meta-regression analysis did not find that average age or mean serum IGF-I had significant explanatory power for effect heterogeneity. The main source of heterogeneity in this study remains unclear. It is recommended that future research expand the sample size, include more possible influencing factors, and use multivariate meta-regression and stratified sensitivity analysis to more comprehensively and systematically reveal the root causes of heterogeneity in the association between serum IGF-I and prostate cancer risk, providing a more solid theoretical basis for subsequent evidence-based research and molecular epidemiological prevention and control.

To evaluate the robustness of the combined results, this study performed leave-one-out sensitivity analysis on the 16 included studies. The results showed that, regardless of which study was excluded, the change in the combined effect size was minimal, and the results of the main effect analysis did not undergo substantial change (see Figure 5). This suggests that the results of this meta-analysis are robust and not significantly influenced by any single study.

To assess the risk of publication bias in the included studies, this study plotted a funnel plot and combined it with Egger’s test and Begg’s test for analysis (see Figure 6). The funnel plot showed that the distribution of study points was relatively symmetrical, with no obvious asymmetric clustering, suggesting a low risk of publication bias. The result of Egger’s test showed that the regression intercept z value was 1.8364, P = 0.0663, which did not reach statistical significance, indicating no significant bias in the funnel plot. The Begg’s test Kendall’s tau was 0.0500, P = 0.8248, also without statistical significance. This further supports that there is no obvious publication bias in the included literature as a whole. The robustness of the study conclusions is relatively good.

This meta-analysis evaluated the quality of evidence of the included studies according to the GRADE system. Taking into account study design, risk of bias, heterogeneity, precision, and publication bias, the overall quality of evidence was rated as moderate. For the specific evaluation process and results, see Supplementary Table 4.

This study systematically evaluated the relationship between serum IGF-I levels and the risk of prostate cancer. The combined analysis showed that there is a certain positive correlation between high serum IGF-I levels and prostate cancer risk, but meta-regression did not observe a clear dose-response relationship; that is, elevated IGF-I levels do not necessarily present a “gradient” phenomenon in which the risk increases linearly. Subgroup analyses also showed some discrepancies in the conclusions of studies involving different populations, detection methods, and adjustment for confounding factors. Overall, IGF-I has a certain epidemiological basis as a biomarker for prostate cancer risk, but its dose-effect relationship remains unclear.

Despite our efforts to adhere to strict inclusion criteria and analytical standards, several important limitations remain that may affect the robustness and interpretation of our findings:

(1) Lack of Dose-Response Data

Although we confirmed a positive association between elevated serum IGF-I and increased prostate cancer risk, most included studies only compared the highest versus lowest IGF-I categories or used broad groupings. Detailed continuous data (such as per 10 ng/ml or per SD increase) were rarely reported, preventing a clear dose-response analysis. This makes it difficult to determine whether risk increases linearly or if there is a threshold effect.

(2) Heterogeneity and Influence of Study Characteristics

The included studies originated from diverse countries and populations, with significant variation in age, follow-up duration, and lifestyle, as well as study design (nested case-control, case-control, Mendelian randomization), detection methods, grouping criteria, and statistical approaches. Such heterogeneity likely contributes to the observed differences in effect estimates and may mask true associations. Although we attempted to explore these sources via subgroup and meta-regression analyses, no significant effect modifiers were identified, and the primary sources of heterogeneity remain unclear.

(3) Inconsistent Confounding Adjustment and Limited Control

Confounder adjustment varied widely: some studies only adjusted for basic demographic factors, while others considered additional variables such as family history, hormone levels, comorbidities, or lifestyle factors. The lack of original individual-level data precluded unified adjustment and restricted our ability to conduct stratified or sensitivity analyses. This may have led to residual confounding, causing the true association to be over- or underestimated.

(4) Unavailable Original Data

Our analysis was based entirely on published aggregate data. The absence of original raw data limited our capacity to perform more granular analyses (such as individual patient data meta-analysis or fine subgrouping) and to fully explore the dose-response relationship and interactions with other risk factors.

(5) Potential Publication Bias

Although funnel plots and statistical tests (Egger’s and Begg’s) suggested no major publication bias, we cannot exclude the possibility that studies with negative or null results were less likely to be published. Such bias could distort the pooled effect size.

In summary, while we applied multidimensional analyses and sensitivity tests to enhance reliability, our conclusions remain subject to limitations related to insufficient dose-response data, study heterogeneity, incomplete confounder adjustment, unavailable original data, and potential publication bias. Therefore, results should be interpreted with caution. Future research should prioritize multicenter, high-quality prospective studies, integration and sharing of individual-level data, and standardized reporting to provide more robust and comparable evidence.

Although our study suggests that serum IGF-I is associated with prostate cancer risk, more robust evidence is still required before clinical application. Future research should focus on the following aspects:

(1) Conduct Large-Scale, Multicenter Prospective Cohort Studies

To further validate the association between IGF-I and prostate cancer risk, future studies should prioritize well-designed, multicenter, large-sample prospective cohorts involving diverse populations and subgroups. Such studies will improve statistical power and enable a more precise evaluation of risk across different ethnic, age, and clinical backgrounds.

(2) Promote Integration and Sharing of Individual-Level Data

The integration of individual participant data (IPD) from multiple cohorts will facilitate unified covariate adjustment, enable comprehensive subgroup analyses, and allow for the robust assessment of dose-response relationships between IGF-I levels and cancer risk.

(3) Clarify Dose-Response Relationships and Address Confounding

Future studies should aim to collect and report detailed continuous IGF-I data, which would enable quantitative analysis of dose-response patterns. Comprehensive confounder assessment and advanced statistical modeling (e.g., multivariable adjustment, interaction analysis) are also essential to elucidate the true causal relationship and underlying biological mechanisms.

(4)Integrate Multi-Omics and Clinical Data

Combining IGF-I with multi-omics approaches (genomics, proteomics, metabolomics) and detailed clinical information will help to clarify the molecular mechanisms of IGF-I in prostate carcinogenesis and evaluate its utility for early screening, risk stratification, and individualized prevention strategies.

By addressing these directions, future research can enhance the scientific rigor and translational value of IGF-I as a biomarker, ultimately contributing to more precise prevention, early detection, and personalized management of prostate cancer.

Epidemiological evidence has suggested a positive association between higher IGF-I levels and the risk of premenopausal breast cancer, as noted in previous large-scale prospective studies. The biological plausibility of these associations is supported by the established effects of IGF-I on cellular proliferation and inhibition of apoptosis. However, the magnitude and consistency of the association between IGF-I and cancer risk appear to differ according to cancer type, study design, and population characteristics. As discussed by Stattin et al., while IGF-I may act as a biomarker of susceptibility for multiple malignancies, further research is needed to clarify its clinical utility in risk prediction and prevention strategies across different cancer types (16).