Abstract
Context
The functional status of organs, such as the liver, involved in IGF-1 signaling pathways influences circulating levels of IGF-1 and hence its relationship to risk of chronic disease and mortality, yet this has received limited attention.
Objective
To examine the relationship between IGF-1 and risk of morbidity and mortality from cancer, cardiovascular diseases (CVD), and all causes, accounting for liver function.
Methods
This study was a case-cohort design nested within EPIC-Heidelberg. IGF-1 was measured in 7461 stored serum samples collected from 1994 to 1998. Median follow-up for incident mortality events was 17.5 years. The case-cohort included a subcohort of 1810 men and 1890 women, in addition to 1668 incident cases of cancer (623 breast, 577 prostate, 202 lung, and 268 colorectal), and 1428 cases of CVD (707 myocardial infarctions and 723 strokes) and 2441 cases of death.
Results
Higher IGF-1 levels showed direct associations with risks of breast (1.25; 95% CI [1.06-1.47]) and prostate (1.31; [1.09-1.57]) cancers. Restricted cubic splines plots and models including IGF-1 as quintiles revealed a U-shaped relationship between the biomarker and mortality. Participants with the lowest and the highest levels of IGF-1 experienced higher hazards of mortality from cancer, CVD, and all causes. The U-shaped form of the relationship persisted but was attenuated in analyses including only participants without any indications of liver dysfunction.
Conclusion
This large population-based prospective study showed that both individuals with lowest and highest levels of circulating IGF-1 were at increased risk of deaths from cancer, CVD, and all causes. For individuals with low IGF-1, the excess risks of death were more pronounced among individuals with liver cancer and cirrhosis but were also present among individuals without elevated liver enzymes.
Keywords: IGF-1, liver function, liver enzymes, cardiovascular diseases, cancer, mortality
Insulin-like growth factor 1 (IGF-1) has been extensively investigated as a biomarker of chronic disease risk and mortality. Although most of the circulating IGF-1 is produced by the liver, growth hormone (GH)–stimulated local production of IGF-1 occurs in various tissues in both paracrine/autocrine fashion (1). IGF-1 plays essential roles in growth, development, and metabolism, is recognized to regulate aging and also implicated in the development of several age-related diseases (2). Circulating levels of IGF-1 vary across the lifespan, peaking in the second decade of life—a period marked by accelerated linear growth and cell proliferation, and rapidly declining until the sixth decade, beyond which it plateaus (3, 4). In animal models, reduced secretion of IGF-1 is associated with delayed onset of cognition impairments, glucose dysregulation, cancer, and extended lifespan (5). In contrast, overexpression of IGF-1 in mouse models resulted in accelerated onset of age-related conditions and reduced lifespan (6, 7).
Scientific controversy remains about whether reduced or elevated levels of IGF-1 are beneficial or detrimental to human health. Individuals with genetic mutations affecting the function of the GH/IGF-1 pathway, such as individuals with Laron syndrome, who lack functional growth hormone receptors (GHR) because of a mutation in the GHR gene, serve as models for investigating the role of IGF-1 in development of disease. Studies show that individuals with Laron syndrome have a reduced risk of developing cancer (8, 9), may have better insulin sensitivity, and are at lower risk of diabetes (9) but may be at increased risk of stroke and cardiovascular diseases (CVD) (10). On the contrary, individuals with elevated GH/IGF-1 signaling, such as patients with acromegaly, exhibit increased risk of several adverse outcomes—mainly cardiovascular and neoplastic complications (11). It is hypothesized that IGF-1 promotes tumorigenesis by activating pathways that promote cell growth and proliferation and inhibit apoptosis (12). There is strong evidence for the association between circulating levels of IGF-1 levels with the development of cancer, particularly the incidence of prostate (13) and breast cancer (14-17). Associations with incidence of cancers of the lung and colorectal have also been suggested in some studies but remain equivocal (18-21), while an analysis of the UK Biobank found both positive and negative associations for cancers at different sites (16).
Earlier investigations of the relationship between circulating IGF-1 and mortality from cancers, CVD, and from all causes have shown inconsistent findings, with studies reporting both positive (22-24) and negative associations (25-27), and sometimes null results (28, 29). However, 2 of the largest recent studies using the UK Biobank concluded that IGF-1 had a U-shaped relationship with disease risk and mortality endpoints (30, 31). Some earlier studies had also suggested a U-shaped relationship between IGF-1 and mortality (32, 33).
A key limitation of most of the epidemiological studies on this subject, perhaps contributing to the inconsistency of findings, is the failure to account for the functioning and health status of relevant organs, which may affect both the circulating levels of IGF-1 and the effect of IGF-1 on disease development and progression (2). Of particular interest is the health status of the liver, which secretes the bulk of the IGF-1 that is in circulation. Studies show that serum levels of IGF-1 are correlated with the extent of hepatocellular function (34). Decreased IGF-1 production from the liver may be a result of impaired liver function (35). In addition, impaired liver function may in itself be a risk factor for mortality. Circulating levels of liver enzymes are routinely used to test liver function and identify individuals with liver injury (36).
Here, using liver enzymes as indicators of liver function, we examine the associations between IGF-1 and incidence of cancers and CVD and mortality due to cancer, cardiovascular causes, and all causes in all study participants and separately for individuals with and without indications for impaired liver function. The study uses a case-cohort study design nested within the European Prospective Investigation into Cancer and Nutrition (EPIC) Heidelberg Study.
Methods
EPIC-Heidelberg Cohort
EPIC-Heidelberg is one of the study centers for the European Prospective Investigation into Cancer and Nutrition (EPIC) study, which recruited more than half a million (510 000) participants across 10 European countries to investigate the associations between diet, metabolic, and lifestyle factors and the risk of cancer and other chronic diseases (37). In the EPIC-Heidelberg cohort, a total of 25 540 women aged 35 to 65 and men aged 40 to 65 years were recruited between 1994 and 1998 from the local general population in Heidelberg and surrounding municipalities. At baseline, data about participants’ health status, lifestyle, socioeconomic status, and reproductive history were collected using extensive self-administered questionnaires and interviews, and anthropometric indices were measured. On the day of recruitment, a blood sample was drawn, and kept for a maximum of 24 hours at +4 °C to +10 °C until centrifugation and further processing. Samples were aliquoted into fractions of plasma, serum, erythrocytes, and buffy coat and stored under liquid nitrogen at −196 °C. All participants provided informed consent on their time of enrollment.
Prospective Ascertainment of Outcomes
In EPIC-Heidelberg, incident CVD and cancer occurrences were prospectively ascertained through active follow-up among study subjects and their next of kin, combined with linkages to hospitalization records, and cancer and pathology registries. Mortality outcomes were ascertained from death certificates which were collected from mortality registries. All cases were validated and coded by a study physician based on medical records.
Case-cohort Sampling
We used the case-cohort design because it allows investigation of several different outcomes and enables efficient use of resources such as biological samples. The case-cohort study is a cost-efficient design because measurement of IGF-1 and other biomarkers for all the participants of the EPIC-Heidelberg cohort would have been very expensive. Instead, the case-cohort design allows for measurement of the biomarkers for all the cases of interest and for a representative sample of the study population (subcohort), with minimal losses of statistical power. The subcohort—a representative sample of the full EPIC-Heidelberg cohort—was selected in a 2-step process. The first step consisted of selection of a 10% random sample of the entire EPIC-Heidelberg cohort and the second step consisted of selection of a further random 10% sample of the EPIC cohort participants older than 50 years at recruitment and who were not in the initial subcohort. The second step oversampled older participants because the incidence of cancers and cardiovascular events are correlated with chronological age and mostly occur in this age group. The case-cohort further included all verified incident cases of cancer of the breast (International Classification of Diseases [ICD]-10: C50, 685 cases), prostate (ICD-10: C61, 597 cases), lung (ICD-10: C34, 219 cases), and colorectal (ICD-10: C18–20, 284 cases), as well as incident cases of myocardial infarction (MI) (ICD-10: I21, 774 cases) and stroke (ICD-10: I60, I61, I63, I64, 798 cases), diagnosed up to the end of December 2014, as well as all deaths that were recorded until the end of 2014. In addition to mortality from all causes, we considered mortality due to specific causes: cancer (C00–D48), diseases of the circulatory system (I00–I99), diseases of the respiratory system (J00–J99), and diseases of the digestive system (K00–K93), and mortality due to external causes of morbidity and mortality (V01–Y98). Figure 1 shows the graphical representation of the sampling process.
Figure 1.
Graphical representation of the selection process of the EPIC-Heidelberg case-cohort and the study participants. The case-cohort was composed of all cases of cardiovascular diseases, 4 major cancers (lung, colorectal, prostate, and breast) and all deaths that occurred until end of 2014 and a subcohort (a representative sample of the EPIC-Heidelberg cohort). The subcohort was selected in 2 steps. A 10% random sample of all EPIC-Heidelberg participants (n = 2691) in the first step and further 10% sample of the participants who were older than 50 years at recruitment and not sampled in step 1 (n = 1103).
Laboratory Methods
Total IGF-1 was measured in serum samples using the immunoradiometric assay A15729 from Immunotech/Beckman Coulter (Heidelberg, Germany). The manufacturers’ protocol was used for all samples. Samples were removed from long-term storage at −196 °C and temporarily stored at −80 °C prior to thawing and biomarker measurements. Once at room temperature, samples were treated with the provided dissociation buffer with shaking. To the antibody-coated tubes the radioactive tracer was added, followed by the calibrator, control, or samples. Tubes were then incubated at 23 °C for 1 hour with shaking. Liquid was removed and the tubes washed twice with the provided wash buffer. Liquid was aspirated and the tubes dried. Counts were measured using a Wizard gamma counter (Perkin Elmer, Rodgau, Germany). Quality controls (QC) were included in each batch in duplicate. The coefficient of variation (CV%) for inter-batch QCs was 3.38% and between batches was 13.7%. All samples were anonymized/blinded during the laboratory analyses. Laboratory methods used to measure liver enzymes have been previously described (36). Measurement of IGF-1 was performed in 2021.
Liver Health and Function
We defined liver function at blood collection based on the following criteria:
Participants who were in the top 5th percentiles of any of the 4 liver enzymes based on sex-specific distributions in the subcohort: alkaline phosphatase (ALP), gamma-glutamyltransferase (GGT), alanine aminotransferase (ALT), or aspartate transaminase (AST). The corresponding 95th percentile sex-specific cutoffs for the liver enzymes were 113 ng/mL and 112 ng/mL for ALP, 144 ng/mL and 68 ng/mL for GGT, 50 ng/mL and 35 ng/mL for AST, 73 ng/mL and 46 ng/mL in men and women respectively
Participants who self-reported a past medical diagnosis of chronic liver insufficiency, liver dysfunction, or hepatitis or self-reported a diagnosis of liver dysfunction/insufficiency during follow-up rounds until end of 2014 (n = 525)
Participants diagnosed with liver cancer until end of 2014 (n = 56)
Participants who died from liver cancer until end of 2014 (n = 43)
Participants who died from cirrhosis during follow-up (n = 65)
Based on these criteria, 1553 participants were considered to have impaired liver function.
Statistical Analyses
The case-cohort included 7783 and IGF-1 measurements were available for 7461 (including 3700 in the subcohort) participants who were included in the analyses. Participants’ characteristics at baseline were described separately for men and women in the subcohort. We performed an analysis of variance separately for men and women to compute marginal means and tests for differences in the mean concentrations of IGF-1 by age group at blood collection, body mass index (BMI) category, level of physical activity, smoking status, self-reported diabetes and hypertension status, and by percentiles of liver enzymes.
Prospective associations between circulating levels of IGF-1 and risks of breast, prostate, lung, colorectal cancers, MI, stroke, CVD, and mortality (all-cause and cause-specific) were assessed using cause-specific Cox proportional hazards models. Inverse subcohort sampling probability (ISSP) (38) weighting was used to account for the case-cohort design, and for the oversampling of older participants according to the case-cohort sampling scheme; that is, participants aged 50 or younger were assigned a 10% probability, and those aged above 50 a 19% probability (10% given they were not drawn in the first selection step (a 90% probability): 10% + (10% × 90%)). Cause-specific hazard ratios (HR) and their 95% CI were obtained for causes of death, any first occurrence of incident cancer (ie, considering the earlier occurrence of any other cancer type as competing events, with the exception of non-melanoma skin cancer), or incident cardiovascular event (where stroke and MI were considered as mutually competing events).
We built a model adjusted for age as timescale and sex (except for breast and prostate cancer), and additionally adjusted for potential confounders, including BMI (quintiles), smoking status (never, long-time quitters, short-time quitters, current light, and current heavy smokers), and physical activity (inactive, moderately inactive, moderately active, active). All confounders were completely known with no missing data. In all Cox models, age was the underlying timescale, and all models were additionally stratified by age at recruitment (5-year category) to account for a potential birth-cohort effect. IGF-1 was used both as categorical (sex-specific quintiles based on the distribution in the subcohort, with the middle (Q3) as reference group), and continuous (after a log-2 based transformation; HRs were therefore interpreted as the relative hazard for a doubling of the biomarker concentration). We performed stratified analyses for all participants and separately for participants with and without impaired liver function. We checked the proportional hazards assumption using an extended version of the Schoenfeld residuals (39), and tests for linear trends were based on the median of each quintile modeled as a continuous variable. To graphically visualize dose-response associations and assess log-linearity, we fitted Cox models with restricted cubic splines (3 knots).
Sensitivity Analyses
We explored the impact of using less conservative thresholds of 90th percentiles for liver enzymes as indicators for hepatic dysfunction. The corresponding cutoff points for the liver enzymes were as follows: 100 ng/mL and 98 ng/mL for ALP, 101 ng/mL and 47 ng/mL for GGT, 58 ng/mL and 36 ng/mL for ALT, and 40 ng/mL and 30 ng/mL for AST, in men and women, respectively. We also performed stratified analyses by sex and by age at blood draw (younger or older than 55 years) because some studies have reported sex and age group differences in the relationship between men and women and younger and older participants. To examine whether the predictive power of IGF-1 on mortality outcomes wanes over time, we performed analyses stratified by length of follow-up comparing the associations in the first 10 years of follow-up with associations with longer follow-up. Tests were two-sided and P values less than .05 were considered statistically significant. All analyses were performed using SAS V.9.4 (SAS Institute).
Results
Characteristics of the Study Population
This case-cohort analysis included a subcohort of 1810 men and 1890 women, in addition to 1668 cases of cancer (623 breast, 577 prostate, 202 lung, and 268 colorectal cancers), 1428 cases of CVD (707 MIs and 723 strokes), and 2441 deaths, ascertained until end of December 2014. Among the 2441 cases of death, 43.7% were due to cancer (n = 1066), 24.7% to circulatory diseases (n = 602), 3.5% to respiratory diseases (n = 85), and 4.6% to digestive diseases (n = 112). Among deaths from digestive diseases, 58% (n = 65) were due to liver cirrhosis.
As shown in Table 1, and by study design, women in the subcohort were slightly younger at baseline recruitment (range, 35-66 years; average 51.3 years) than men (range, 40-65; average 54 years). Both women and men in the subcohort were slightly overweight, with average BMI of 25.5 and 26.9 kg/m2, respectively. Women were more likely to be physically active than men: 55.8% of women were at least moderately active, compared with 48% of the men. In the subcohort, a larger proportion in women than in men reported having never smoked: half of women in the subcohort and 31% of men. About one-third of the men and half of the women self-reported to be hypertensive.
Table 1.
Baseline characteristics of the study population, EPIC-Heidelberg subcohort (n = 3700)
| Characteristic | Men (n = 1810) | Women (n = 1890) | Total (n = 3700) |
|---|---|---|---|
| Age at blood collection (years) | |||
| Median (Min, Max) | 54 (40.3, 65.4) | 51.3 (35.2, 66) | 52.7 (35.2, 66) |
| 35-<40 | 242 (12.8) | 242 (6.5) | |
| 40-<45 | 259 (14.3) | 269 (14.2) | 528 (14.3) |
| 45-<50 | 216 (11.9) | 248 (13.1) | 464 (12.5) |
| 50-<55 | 447 (24.7) | 412 (21.8) | 859 (23.2) |
| 55-<60 | 478 (26.4) | 360 (19) | 838 (22.6) |
| 60-65 | 410 (22.7) | 359 (19) | 769 (20.8) |
| Body mass index (kg/m2) | |||
| Mean ± SD | 26.9 ± 3.6 | 25.5 ± 4.7 | 26.2 ± 4.3 |
| Underweight (<18.5) | 6 (0.3) | 32 (1.7) | 38 (1) |
| Normal (18.5 to <25) | 549 (30.3) | 983 (52) | 1532 (41.4) |
| Overweight (25 to <30) | 933 (51.5) | 579 (30.6) | 1512 (40.9) |
| Obese (≥30) | 322 (17.8) | 296 (15.7) | 618 (16.7) |
| Smoking status | |||
| Never | 563 (31.1) | 990 (52.4) | 1553 (42) |
| Long-time quitter (>10 y) | 647 (35.7) | 363 (19.2) | 1010 (27.3) |
| Short-time quitter (=10 y) | 200 (11) | 149 (7.9) | 349 (9.4) |
| Current, light (=10 cig/d) | 102 (5.6) | 171 (9) | 273 (7.4) |
| Current, heavy (>10 cig/d) | 298 (16.5) | 217 (11.5) | 515 (13.9) |
| Physical activity level | |||
| Inactive | 411 (22.7) | 297 (15.7) | 708 (19.1) |
| Moderately inactive | 530 (29.3) | 538 (28.5) | 1068 (28.9) |
| Moderately active | 659 (36.4) | 900 (47.6) | 1559 (42.1) |
| Active | 210 (11.6) | 155 (8.2) | 365 (9.9) |
| Diabetes | |||
| No | 1705 (94.2) | 1850 (97.9) | 3555 (96.1) |
| Yes | 105 (5.8) | 40 (2.1) | 145 (3.9) |
| Hypertension | |||
| No | 1143 (63.1) | 1391 (73.6) | 2534 (68.5) |
| Yes | 667 (36.9) | 499 (26.4) | 1166 (31.5) |
| Menopausal status | |||
| Premenopausal | 650 (34.4) | ||
| Postmenopausal | 885 (46.8) | ||
| Perimenopausal | 297 (15.7) | ||
| Bilateral ovariectomy | 58 (3.1) | ||
| IGF-1 | |||
| Mean ± SD | 123.7 ± 48.1 | 124.8 ± 56.4 | 124.3 ± 52.5 |
IGF-1 Levels
The mean IGF-1 levels at baseline were 123.7 ± 48.1 ng/mL in men and 124.8 ± 56.4 ng/dL in women (Table 1). Table 2 shows means of IGF-1 in men and women by strata of age, BMI, physical activity, smoking status, self-reported diabetes and hypertension status, percentile (<95th percentile vs ≥95th percentile) of liver enzymes, and by menopausal status in women. In both men and women, mean IGF-1 levels lower among older participants at blood collection (P < .001), and was also lower among participants with higher BMI. In men but not in women, IGF-1 was lowest among those who were most physically active (P = .028), and among smokers (P = .002). In women but not in men, IGF-1 levels were higher among those who reported to be hypertensive (P = .029). The levels of IGF-1 were also lower among participants in the top 5th percentiles of the liver enzymes GGT, AST, and ALT in both sexes; however, for participants in the top 5th percentile of ALP, the IGF-1 levels were only significantly lower in men (Table 2).
Table 2.
Age-adjusted mean levels of IGF-1 across different strata of the subcohort, EPIC-Heidelberg subcohort (n = 3700)
| Men | Women | |||||
|---|---|---|---|---|---|---|
| Variables | n | Mean ± SE | P value | n | Mean ± SE | P value |
| Age group (years) | ||||||
| 35 to <40 | — | 242 | 165.1 ± 3.3 | <.001 | ||
| 40 to <45 | 259 | 145.4 ± 2.9 | <.001 | 269 | 149.9 ± 3.1 | |
| 45 to <50 | 216 | 135.7 ± 3.2 | 248 | 147.8 ± 3.2 | ||
| 50 to <55 | 447 | 121.2 ± 2.2 | 412 | 111.6 ± 2.5 | ||
| 55 to <60 | 478 | 115 ± 2.1 | 360 | 103.7 ± 2.7 | ||
| Older than 60 | 410 | 116.5 ± 2.3 | 359 | 99.3 ± 2.7 | ||
| BMI (kg/m2) | ||||||
| Normal (18 to <25) | 549 | 127.8 ± 2 | <.001 | 983 | 128.7 ± 1.7 | .001 |
| Overweight (25 to <30) | 933 | 125.6 ± 1.5 | 579 | 121.8 ± 2.1 | ||
| Obese (≥30) | 322 | 111.7 ± 2.6 | 296 | 117.2 ± 3 | ||
| Physical activity level | ||||||
| <3.5 hours/week | 411 | 127.1 ± 2.3 | .028 | 297 | 121 ± 3 | .066 |
| 3.5–7 hours/week | 530 | 120.3 ± 2 | 538 | 126.9 ± 2.2 | ||
| 7–14 hours/week | 659 | 126 ± 1.9 | 900 | 126.2 ± 1.7 | ||
| >14 hours/week | 210 | 118.4 ± 3.3 | 155 | 116.4 ± 4.1 | ||
| Smoking status | ||||||
| Never | 563 | 129.1 ± 2 | .002 | 990 | 126 ± 1.6 | .889 |
| Quitter (10+ years) | 647 | 124.5 ± 1.9 | 363 | 123.4 ± 2.7 | ||
| Quitter (<10 years) | 200 | 118.2 ± 3.3 | 149 | 122.9 ± 4.2 | ||
| Smoker (10+ cigs) | 298 | 116.8 ± 2.7 | 217 | 123.9 ± 3.5 | ||
| Smoker (<10 cigs) | 102 | 119.3 ± 4.6 | 171 | 123.5 ± 3.9 | ||
| Self-reported diabetes | ||||||
| No | 1705 | 124.1 ± 1.1 | .125 | 1850 | 125 ± 1.2 | .31 |
| Yes | 105 | 116.8 ± 4.6 | 40 | 116.6 ± 8.1 | ||
| Self-reported hepatitis | ||||||
| No | 1708 | 123.5 ± 1.1 | .519 | 1776 | 124.8 ± 1.2 | .888 |
| Yes | 102 | 126.6 ± 4.7 | 114 | 125.4 ± 4.8 | ||
| Self-reported hypertension | ||||||
| No | 1143 | 125 ± 1.4 | .125 | 1391 | 126.4 ± 1.4 | .029 |
| Yes | 667 | 121.4 ± 1.8 | 499 | 120.3 ± 2.4 | ||
| Self-reported liver dysfunction | ||||||
| No | 1779 | 123.8 ± 1.1 | .363 | 1864 | 124.8 ± 1.2 | .889 |
| Yes | 31 | 116.1 ± 8.5 | 26 | 123.4 ± 10.1 | ||
| Menopausal status | ||||||
| Perimenopausal | 58 | 114.1 ± 7 | .147 | |||
| Postmenopausal | 885 | 124.5 ± 2.5 | ||||
| Premenopausal | 650 | 128.1 ± 3.2 | ||||
| Surgical premenopausal | 297 | 120.7 ± 3 | ||||
| ALP | ||||||
| <95th percentile | 1722 | 124.4 ± 1.1 | .003 | 1802 | 125 ± 1.2 | .479 |
| ≥95th percentile | 88 | 109.1 ± 5 | 88 | 121 ± 5.5 | ||
| GGT | ||||||
| <95th percentile | 1660 | 126 ± 1.1 | <.001 | 1858 | 125.1 ± 1.2 | .022 |
| ≥95th percentile | 150 | 98.6 ± 3.8 | 32 | 104.2 ± 9.1 | ||
| AST | ||||||
| <95th percentile | 1675 | 125.3 ± 1.1 | <.001 | 1853 | 125.4 ± 1.2 | <.001 |
| ≥95th percentile | 135 | 103.1 ± 4 | 37 | 93.6 ± 8.4 | ||
| ALT | ||||||
| <95th percentile | 1675 | 124.9 ± 1.1 | <.001 | 1848 | 125.2 ± 1.2 | .014 |
| ≥95th percentile | 135 | 108.1 ± 4 | 42 | 105.6 ± 7.9 | ||
Abbreviations: BMI, body mass index; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase.
IGF-1 and Cancer and Cardiovascular Disease
Table 3 shows the prospective associations between circulating levels of IGF-1 and risks of cancer and CVD incidence and mortality as well as risk of mortality from other causes of death. Higher IGF-1 levels showed direct associations with risks of breast (HR = 1.25; 95% CI [1.06-1.47]) and prostate (1.31 [1.09-1.57]) cancers after adjusting for potential confounders. There was no evidence of an association between levels of IGF-1 and risks of colorectal cancer. Although there was no overall association between IGF-1 and lung cancer incidence, participants in the lowest quintile of IGF-1 were at higher hazards of lung cancer incidence (1.77 [1.09-2.86]), and hazard ratios across other quintiles suggested a U-shaped relationship. Regarding CVD, an inverse association was observed between IGF-1 and the risk of stroke (0.85 [0.74-0.97]). This association was observed using the continuous IGF-1 variable, independently of the adjustment covariates. When IGF-1 was modeled as a categorical variable (quintiles), the inverse association was not statistically significant for any stratum but there was consistent and significant trend (P for trend = .048). There was no evidence of statistically significant associations with risks of MI or overall CVD.
Table 3.
Associations between IGF-1 and risk of and mortality from cancers, cardiovascular diseases, and all causes in EPIC-Heidelberg case-cohort (n = 7461)
| Quintiles of IGF-1 | P for trend | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Outcome | Q1 | Q2 | Q3 | Q4 | Q5 | Q1-Q5 | Q1-Q3 | Q3-Q5 | Continuous | |
| Incidence | ||||||||||
| Breast | N | 105 | 106 | 125 | 139 | 148 | 623 | |||
| HR (95% CI) | 0.86 (0.64-1.16) | 0.85 (0.64-1.14) | 1.00 Ref | 1.14 (0.86-1.51) | 1.22 (0.92-1.63) | .009 | .463 | .210 | 1.25 (1.06-1.47) | |
| Prostate | N | 97 | 109 | 117 | 128 | 126 | 577 | |||
| HR (95% CI) | 0.76 (0.56-1.04) | 0.92 (0.67-1.24) | 1.00 Ref | 1.16 (0.86-1.57) | 1.24 (0.92-1.67) | .001 | .107 | .178 | 1.31 (1.09-1.57) | |
| Lung | N | 51 | 43 | 32 | 37 | 39 | 202 | |||
| HR (95% CI) | 1.77 (1.09-2.86) | 1.42 (0.87-2.32) | 1.00 Ref | 1.32 (0.79-2.2) | 1.43 (0.85-2.41) | .418 | .020 | .206 | 0.82 (0.63-1.08) | |
| Colon | N | 64 | 56 | 48 | 58 | 42 | 268 | |||
| HR (95% CI) | 1.19 (0.8-1.77) | 1.15 (0.76-1.72) | 1.00 Ref | 1.35 (0.9-2.02) | 1.07 (0.69-1.66) | .867 | .335 | .892 | 1.04 (0.85-1.28) | |
| CVD | N | 337 | 296 | 277 | 257 | 261 | 1428 | |||
| HR (95% CI) | 1.12 (0.92-1.36) | 1.04 (0.85-1.27) | 1.00 Ref | 0.99 (0.81-1.22) | 1.08 (0.87-1.34) | .708 | .200 | .356 | 0.95 (0.85-1.07) | |
| MI | N | 151 | 140 | 140 | 128 | 148 | 707 | |||
| HR (95% CI) | 0.99 (0.76-1.29) | 0.97 (0.74-1.27) | 1.00 Ref | 0.97 (0.74-1.28) | 1.21 (0.92-1.59) | .166 | .909 | .118 | 1.09 (0.93-1.28) | |
| Stroke | N | 187 | 157 | 137 | 129 | 113 | 723 | |||
| HR (95% CI) | 1.26 (0.98-1.63) | 1.12 (0.86-1.44) | 1.00 Ref | 1.02 (0.78-1.33) | 0.97 (0.74-1.29) | .048 | .060 | .902 | 0.85 (0.74-0.97) | |
| Mortality | ||||||||||
| All | N | 704 | 510 | 413 | 410 | 404 | 2441 | |||
| HR (95% CI) | 1.56 (1.31-1.85) | 1.19 (0.99-1.42) | 1.00 Ref | 1.13 (0.93-1.36) | 1.24 (1.02-1.5) | .008 | .000 | .033 | 0.82 (0.74-0.9) | |
| Cancers | N | 288 | 224 | 184 | 182 | 188 | 1066 | |||
| HR (95% CI) | 1.49 (1.2-1.85) | 1.2 (0.96-1.5) | 1.00 Ref | 1.11 (0.87-1.4) | 1.25 (0.98-1.59) | .097 | .000 | .076 | 0.86 (0.76-0.97) | |
| Liver | N | 21 | 10 | 4 | 5 | 3 | 43 | |||
| HR (95% CI) | 4.6 (1.53-13.81) | 2.34 (0.73-7.56) | 1.00 Ref | 1.42 (0.37-5.45) | 0.91 (0.19-4.44) | .004 | .002 | .747 | 0.31 (0.18-0.53) | |
| Other cancers | N | 267 | 214 | 180 | 177 | 185 | 1023 | |||
| HR (95% CI) | 1.42 (1.13-1.77) | 1.17 (0.94-1.47) | 1.00 Ref | 1.1 (0.87-1.4) | 1.26 (0.99-1.6) | .263 | .002 | .065 | 0.9 (0.8-1.02) | |
| Respiratory disease | N | 21 | 17 | 20 | 19 | 8 | 85 | |||
| HR (95% CI) | 0.94 (0.46-1.9) | 0.78 (0.37-1.67) | 1.00 Ref | 1.15 (0.56-2.37) | 0.62 (0.25-1.57) | .730 | .790 | .338 | 0.98 (0.65-1.48) | |
| Digestive disease | N | 49 | 20 | 19 | 18 | 6 | 112 | |||
| HR (95% CI) | 2.42 (1.39-4.22) | 0.98 (0.51-1.9) | 1.00 Ref | 1.01 (0.51-2.03) | 0.37 (0.14-0.96) | <.001 | <.001 | .018 | 0.35 (0.25-0.49) | |
| Cirrhosis | N | 32 | 12 | 10 | 7 | 4 | 65 | |||
| HR (95% CI) | 3.22 (1.55-6.68) | 1.12 (0.46-2.69) | 1.00 Ref | 0.69 (0.25-1.88) | 0.42 (0.13-1.41) | <.001 | <.001 | .163 | 0.23 (0.15-0.37) | |
| Other digestive | N | 17 | 8 | 9 | 11 | 2 | 47 | |||
| HR (95% CI) | 1.63 (0.69-3.83) | 0.83 (0.32-2.18) | 1.00 Ref | 1.47 (0.56-3.85) | 0.28 (0.06-1.28) | .029 | .212 | .033 | 0.61 (0.39-0.96) | |
| Circulatory | N | 174 | 113 | 107 | 108 | 100 | 602 | |||
| HR (95% CI) | 1.32 (0.99-1.75) | 0.96 (0.71-1.3) | 1.00 Ref | 1.15 (0.84-1.56) | 1.21 (0.88-1.68) | .782 | .022 | .297 | 0.92 (0.78-1.08) | |
| External causes | N | 31 | 20 | 17 | 14 | 19 | 101 | |||
| HR (95% CI) | 2.06 (1.1-3.87) | 1.18 (0.61-2.31) | 1.00 Ref | 0.84 (0.41-1.73) | 1.15 (0.59-2.23) | .070 | .020 | .458 | 0.7 (0.48-1.03) | |
Quintiles are based on the sex-specific distribution of IGF-1 in the subcohort.
The models were adjusted for age as timescale and sex (except for breast and prostate cancer), body mass index (quintiles), smoking status (never, long-time quitters, short-time quitters, current light, and current heavy smokers), and physical activity (inactive, moderately inactive, moderately active, active).
Continuous HR for 1 unit increment in log-2 based IGF-1 = change in hazard associated with a doubling of IGF-1 concentration.
HRs were corrected to match case-cohort design using inverse subcohort sampling probability weighting (ISSP).
Abbreviations: HR, hazard ratio, IGF-1, insulin-like factor 1.
IGF-1 and Mortality
Associations between IGF-1 and risks of all-cause and cause-specific mortality are also presented in Table 3. Overall, circulating levels of IGF-1 were inversely associated with the risk of death from all causes (HR for a doubling in IGF-1 levels = 0.82 [95% CI: 0.74, 0.90]), cancers (0.86 [0.76-0.97]), and digestive diseases (0.35 [0.25-0.49]) after adjusting for potential confounders. However, modeling IGF-1 as quintiles and plotting the dose-response curves using restricted cubic splines (Fig. 2 and Table 3) revealed U-shaped relationship between the biomarker and mortality such that both participants with lowest and highest levels of IGF-1 experienced higher hazards of mortality from cancer, CVD, and all causes. There was an inverse relationship between IGF-1 levels and mortality from digestive diseases (0.35 [0.25-0.49]), with an even stronger association for mortality due to cirrhosis (0.23 [0.15-0.37]). Regarding mortality from cardiovascular causes, although hazards of mortality were higher among participants at both extremes of the IGF-1 distribution, they did not reach statistical significance. However, the hazard ratios across the quintiles as well as dose-response plots suggested a U-shaped relationship, with a significant trend observed for hazards of mortality between Q1 and Q3 (Table 3 and Fig. 2).
Figure 2.
Dose-response relationships between IGF-1 and cancer, cardiovascular, and total mortality in all participants and in participants with and without liver conditions. The dashed lines indicate 95% CI. **Liver conditions include; elevated (above 95th percentile of subcohort distribution) liver enzymes, liver cirrhosis, liver cancer, and self-reports of liver insufficiency, hepatitis, and liver dysfunction.
The U-shaped form of the relationship persisted but attenuated in analyses that included only participants without any indications of liver dysfunction. For example, compared with participants in the middle quintile of IGF-1 distribution, participants in the lowest quintile experienced hazards of total and cancer mortality that were 1.56- and 1.49-fold higher, which were attenuated to 1.37- and 1.35-fold higher respectively, while remaining statistically significant (Table 4). On the other hand, the hazards of total and cancer related mortality were further increased among those in the lowest quintile of IGF-1 compared with those in Q3 in analyses that included only participants with elevated liver enzymes and liver-related morbidities. Among participants with liver dysfunction–associated conditions, participants in the lowest quintile (Q1) of IGF-1 were at higher hazards of lung cancer incidence and of mortality compared to participants in the third quintile. In particular, these participants experienced particularly high hazards of deaths due to liver cancer (3.36 [1.11-10.21]) and cirrhosis (2.78 [1.23-6.25]) (Table 5).
Table 4.
Associations between IGF-1 and risk of and mortality from cancers, cardiovascular diseases, and all causes in the EPIC-Heidelberg case-cohort among participants without liver-related conditions (n = 5911)
| Quintiles of IGF-1 | P for trend | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Outcome | Q1 | Q2 | Q3 | Q4 | Q5 | Q1-Q5 | Q1-Q3 | Q3-Q5 | Continuous | |
| Incidence | ||||||||||
| Breast | N | 77 | 91 | 102 | 118 | 128 | 516 | |||
| HR (95% CI) | 0.78 (0.56-1.1) | 0.9 (0.65-1.24) | 1.00 Ref | 1.13 (0.83-1.54) | 1.24 (0.91-1.71) | .007 | .213 | .255 | 1.27 (1.07-1.51) | |
| Prostate | N | 69 | 84 | 96 | 107 | 109 | 465 | |||
| HR (95% CI) | 0.74 (0.52-1.05) | 0.93 (0.66-1.32) | 1.00 Ref | 1.16 (0.84-1.62) | 1.25 (0.9-1.74) | .002 | .119 | .235 | 1.35 (1.1-1.66) | |
| Lung | N | 39 | 33 | 30 | 30 | 33 | 165 | |||
| HR (95% CI) | 1.5 (0.89-2.53) | 1.17 (0.68-2) | 1.00 Ref | 1.07 (0.62-1.87) | 1.18 (0.68-2.05) | .428 | .151 | .566 | 0.81 (0.6-1.1) | |
| Colon | N | 48 | 46 | 39 | 55 | 36 | 224 | |||
| HR (95% CI) | 1.22 (0.78-1.92) | 1.22 (0.78-1.92) | 1.00 Ref | 1.54 (0.99-2.39) | 1.06 (0.65-1.73) | .879 | .352 | .691 | 1.1 (0.87-1.38) | |
| CVD | N | 250 | 227 | 222 | 208 | 212 | 1119 | |||
| HR (95% CI) | 1.11 (0.88-1.4) | 1.04 (0.82-1.3) | 1.00 Ref | 0.97 (0.77-1.22) | 1.01 (0.79-1.28) | .383 | .274 | .800 | 0.93 (0.81-1.05) | |
| MI | N | 115 | 113 | 110 | 104 | 115 | 557 | |||
| HR (95% CI) | 1.04 (0.76-1.4) | 1.04 (0.77-1.41) | 1.00 Ref | 0.98 (0.72-1.33) | 1.1 (0.8-1.5) | .817 | .704 | .456 | 1.01 (0.84-1.2) | |
| Stroke | N | 136 | 114 | 112 | 104 | 97 | 563 | |||
| HR (95% CI) | 1.21 (0.91-1.61) | 1.03 (0.77-1.38) | 1.00 Ref | 0.97 (0.72-1.31) | 0.95 (0.7-1.29) | .120 | .182 | .816 | 0.86 (0.74-1) | |
| Mortality | ||||||||||
| All | N | 453 | 369 | 322 | 312 | 332 | 1788 | |||
| HR (95% CI) | 1.37 (1.12-1.68) | 1.17 (0.95-1.44) | 1.00 Ref | 1.08 (0.88-1.33) | 1.23 (0.99-1.54) | .263 | .002 | .051 | 0.92 (0.82-1.03) | |
| Cancersa | N | 195 | 172 | 146 | 145 | 161 | 819 | |||
| HR (95% CI) | 1.35 (1.05-1.73) | 1.21 (0.94-1.56) | 1.00 Ref | 1.08 (0.83-1.41) | 1.27 (0.98-1.66) | .549 | .022 | .069 | 0.94 (0.82-1.09) | |
| Respiratory | N | 13 | 13 | 16 | 15 | 6 | 63 | |||
| HR (95% CI) | 0.78 (0.34-1.79) | 0.84 (0.35-2.02) | 1.00 Ref | 1.24 (0.55-2.77) | 0.56 (0.2-1.58) | .921 | .569 | .294 | 1.11 (0.71-1.75) | |
| Digestiveb | N | 12 | 6 | 7 | 6 | 1 | 32 | |||
| HR (95% CI) | 1.54 (0.56-4.26) | 0.8 (0.27-2.34) | 1.00 Ref | 0.97 (0.29-3.23) | 0.14 (0.02-1.05) | .010 | .312 | .014 | 0.59 (0.37-0.93) | |
| Circulatory | N | 128 | 82 | 84 | 79 | 85 | 458 | |||
| HR (95% CI) | 1.35 (0.98-1.87) | 0.97 (0.68-1.37) | 1.00 Ref | 1.06 (0.75-1.52) | 1.23 (0.86-1.78) | .693 | .029 | .285 | 0.9 (0.75-1.1) | |
| External causes | N | 19 | 12 | 14 | 11 | 16 | 72 | |||
| HR (95% CI) | 1.58 (0.76-3.3) | 0.87 (0.4-1.92) | 1.00 Ref | 0.77 (0.34-1.74) | 1.13 (0.55-2.35) | .460 | .220 | .632 | 0.99 (0.62-1.58) | |
Quintiles are based on the sex-specific distribution of IGF-1 in the subcohort.
The models were adjusted for age as timescale and sex (except for breast and prostate cancer), BMI (quintiles), smoking status (never, long-time quitters, short-time quitters, current light, and current heavy smokers), and physical activity (inactive, moderately inactive, moderately active, active).
Continuous HR for 1 unit increment in log-2 based IGF-1 = change in hazard associated with a doubling of IGF-1 concentration.
HRs were corrected to match case-cohort design using inverse subcohort sampling probability weighting (ISSP).
Abbreviations: CVD, cardiovascular diseases; HR, hazard ratio; MI, myocardial infarction.
Excluding liver cancer.
Excluding deaths from cirrhosis.
Table 5.
Associations between IGF-1 and risk of and mortality from cancers, cardiovascular diseases, and all causes in the EPIC-Heidelberg case-cohort among participants with liver-related conditions (n = 1553)
| Quintiles of IGF-1 | P for trend | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Outcome | Q1 | Q2 | Q3 | Q4 | Q5 | Q1-Q5 | Q1-Q3 | Q3-Q5 | Continuous | |
| Incidence | ||||||||||
| Breast | N | 28 | 15 | 23 | 21 | 20 | 107 | |||
| HR (95% CI) | 1.15 (0.56-2.35) | 0.58 (0.27-1.22) | 1.00 Ref | 1.11 (0.53-2.33) | 1.05 (0.47-2.35) | .760 | .526 | .829 | 1.13 (0.73-1.74) | |
| Prostate | N | 28 | 25 | 21 | 21 | 17 | 112 | |||
| HR (95% CI) | 0.78 (0.39-1.56) | 0.82 (0.4-1.68) | 1.00 Ref | 1.15 (0.52-2.55) | 1.33 (0.62-2.84) | .115 | .734 | .536 | 1.3 (0.85-1.97) | |
| Lung | N | 12 | 10 | 2 | 7 | 6 | 37 | |||
| HR (95% CI) | 5.95 (1.01-34.99) | 6.42 (0.98-41.97) | 1.00 Ref | 3.85 (0.65-22.67) | 6.89 (0.83-57.09) | .807 | .033 | .143 | 0.77 (0.39-1.54) | |
| Colon | N | 16 | 10 | 9 | 3 | 6 | 44 | |||
| HR (95% CI) | 0.98 (0.41-2.35) | 0.85 (0.32-2.27) | 1.00 Ref | 0.34 (0.09-1.33) | 1.04 (0.33-3.21) | .656 | .828 | .758 | 0.76 (0.45-1.28) | |
| CVD | N | 87 | 69 | 55 | 49 | 49 | 309 | |||
| HR (95% CI) | 1.11 (0.73-1.7) | 1.01 (0.65-1.57) | 1.00 Ref | 1.05 (0.64-1.73) | 1.38 (0.83-2.3) | .470 | .574 | .149 | 1.06 (0.82-1.37) | |
| MI | N | 36 | 27 | 30 | 24 | 33 | 150 | |||
| HR (95% CI) | 0.85 (0.49-1.48) | 0.76 (0.42-1.38) | 1.00 Ref | 0.96 (0.52-1.76) | 1.74 (0.94-3.22) | .021 | .597 | .051 | 1.39 (0.96-2) | |
| Stroke | N | 51 | 43 | 25 | 25 | 16 | 160 | |||
| HR (95% CI) | 1.42 (0.8-2.51) | 1.33 (0.75-2.34) | 1.00 Ref | 1.22 (0.62-2.41) | 0.98 (0.48-1.99) | .254 | .225 | .997 | 0.84 (0.61-1.14) | |
| Mortality | ||||||||||
| All | N | 251 | 141 | 91 | 98 | 72 | 653 | |||
| HR (95% CI) | 2.02 (1.4-2.91) | 1.21 (0.82-1.79) | 1.00 Ref | 1.28 (0.83-1.97) | 1.36 (0.87-2.11) | .012 | .000 | .169 | 0.65 (0.53-0.8) | |
| Cancers | N | 93 | 52 | 38 | 37 | 27 | 247 | |||
| HR (95% CI) | 1.82 (1.14-2.9) | 1.13 (0.68-1.86) | 1.00 Ref | 1.21 (0.69-2.11) | 1.15 (0.63-2.08) | .054 | .007 | .808 | 0.69 (0.53-0.89) | |
| Liver cancer | N | 21 | 10 | 4 | 5 | 3 | 43 | |||
| HR (95% CI) | 3.36 (1.11-10.21) | 1.86 (0.57-6.02) | 1.00 Ref | 1.51 (0.37-6.27) | 1.12 (0.22-5.62) | .051 | .023 | .703 | 0.41 (0.23-0.73) | |
| Other cancers** | N | 72 | 42 | 34 | 32 | 24 | 204 | |||
| HR (95% CI) | 1.62 (0.99-2.66) | 1.04 (0.61-1.78) | 1.00 Ref | 1.17 (0.65-2.1) | 1.13 (0.61-2.11) | .180 | .040 | .866 | 0.77 (0.59-1.01) | |
| Respiratory | N | 8 | 4 | 4 | 4 | 2 | 22 | |||
| HR (95% CI) | 1.72 (0.3-9.79) | 0.86 (0.12-6.1) | 1.00 Ref | 1.12 (0.17-7.39) | 1.39 (0.19-10.49) | .724 | .622 | .997 | 0.66 (0.23-1.84) | |
| Digestive | N | 37 | 14 | 12 | 12 | 5 | 80 | |||
| HR (95% CI) | 2.61 (1.25-5.48) | 1.01 (0.42-2.4) | 1.00 Ref | 1.31 (0.51-3.4) | 0.71 (0.22-2.32) | .007 | .004 | .423 | 0.34 (0.21-0.55) | |
| Liver cirrhosis | N | 32 | 12 | 10 | 7 | 4 | 65 | |||
| HR (95% CI) | 2.78 (1.23-6.25) | 1.08 (0.42-2.78) | 1.00 Ref | 0.89 (0.29-2.69) | 0.69 (0.19-2.52) | .003 | .006 | .467 | 0.3 (0.18-0.5) | |
| Circulatory | N | 46 | 31 | 23 | 29 | 15 | 144 | |||
| HR (95% CI) | 1.24 (0.67-2.3) | 0.94 (0.49-1.82) | 1.00 Ref | 1.32 (0.67-2.62) | 1.22 (0.55-2.73) | .870 | .495 | .738 | 0.95 (0.68-1.34) | |
| External causes, eg, accidents | N | 12 | 8 | 3 | 3 | 3 | 29 | |||
| HR (95% CI) | 3.6 (0.89-14.61) | 2.19 (0.52-9.29) | 1.00 Ref | 0.91 (0.17-4.86) | 1.07 (0.19-6.11) | .047 | .061 | .765 | 0.34 (0.17-0.68) | |
Quintiles are based on the sex-specific distribution of IGF-1 in the subcohort.
The models were adjusted for age as timescale and sex (except for breast and prostate cancer), body mass index (quintiles), smoking status (never, long-time quitters, short-time quitters, current light, and current heavy smokers), and physical activity (inactive, moderately inactive, moderately active, active).
Continuous HR for 1 unit increment in log-2 based IGF-1 = change in hazard associated with a doubling of IGF-1 concentration.
HRs were corrected to match case-cohort design using inverse subcohort sampling probability weighting (ISSP).
Abbreviations: CVD, cardiovascular diseases; HR, hazard ratio; MI, myocardial infarction.
To examine heterogeneity by sex and age, which has been previously reported, we performed some sensitivity analyses stratifying by these factors. Age at blood collection was categorized into 2 groups, younger than or older than 55 years. We found that the U-shaped relationship between IGF-1 and total and cancer mortality was apparent for both sexes and both age groups (Supplementary Figs. S1 and S2) (40). For CVD mortality, the U-shaped relationship was observed only in men and among participants who were at least 55 years at blood collection. We also explored the effect of using less conservative thresholds for liver enzymes (ie, using 90th percentile instead of 95% as cutoffs) and our conclusion remained unchanged (Supplementary Tables S1 and S2) (40). In analyses stratified by length of follow-up (<10 years vs ≥10 years), we found that the U-shape of the relationship between IGF-1 and cancer, cardiovascular and total mortality persisted both in the first 10 years following blood draw and in follow-up periods beyond 10 years (Supplementary Fig. S3) (40).
Discussion
In this large prospective study, we found that higher circulating levels of IGF-1 were associated with increased risk of breast and prostate cancer incidence, whereas individuals in the lowest quintile of IGF-1 levels were at increased risk of incident lung cancer. Individuals in the lowest quintile of IGF-1 experienced increased hazards of mortality from digestive diseases, especially liver cirrhosis. We found that circulating levels IGF-1 exhibited a U-shaped relationship with all-cause, cancer, and cardiovascular mortality. For cardiovascular mortality, the U-shaped relationship was more defined in men and in participants who were older than 55 years at the time of blood collection. Stratified analyses revealed that the risk of mortality among participants with low IGF-1 levels were more pronounced among participants with indications of liver diseases and dysfunction.
Earlier studies on the relationship between IGF-1 and mortality produced conflicting findings, with some of the largest studies and recent meta-analyses showing a U-shaped relationship (30, 31, 41). Our findings add to this growing evidence and emerging consensus of a U-shaped relationship between the IGF-1 and mortality outcomes. One concept that has been advanced as relevant for the understanding of the U-shaped nature of the relationship between IGF-1 and mortality relationship is “antagonistic pleiotropy,” which posits that some factors may have been evolutionary selected due to their usefulness in growth and reproduction but may be detrimental in old age (2, 30). This theory suggests that IGF-1 may be beneficial in younger individuals but detrimental in older persons and hence previous conflicting findings reflected the segment of population from which participants were drawn (30). In our study, the U-shaped relationship was observed both among individuals who were aged younger than 55 years and those who were older. However, our study lacked a wider age range as it included participants who were at least 35 years of age and younger than 66 years at the time of collecting the blood sample. Others have suggested that the U-shaped associations with all-cause mortality were observed because of the opposing effects on cancer and cardiovascular mortality (42). This explanation is, however, unlikely as we and others have reported a U-shaped relationship both with cardiovascular (32) and cancer mortality (33).
Another potential explanation, particularly regarding low levels IGF-1 and increased risk of mortality was that of reverse causation. It is probable that individuals with underlying preclinical conditions, particularly those affecting the liver, have low circulating levels of IGF-1 and that their underlying diseases were responsible for the observed increased risk of mortality. We explored this question by using liver enzymes and other liver-related conditions, such as hepatitis, liver cancer, and liver cirrhosis as markers of liver function and health status. We found that, among participants with low IGF-1, the risks of cancer and all-cause mortality were substantially attenuated in analyses that included participants without indications of liver diseases but were elevated in analyses restricted to participants with liver conditions. Our findings suggest that the health and functional status of the liver may explain some but not all of the mortality risk associated with low levels of IGF-1.
Another important factor in understanding the relationship between IGF-1 and mortality is nutritional status, in particular, protein intake. Earlier studies among EPIC participants reported associations between IGF-1 levels and intake of vitamins B6 and B2, proteins, milk, potassium, calcium, magnesium, and phosphorous (43, 44). In our recent analyses, we found that animal intake was associated with increased risk of mortality, largely driven by increased risk of cardiovascular mortality (45). Others analyzed the National Health and Nutrition Examination Survey (NHANES) and reported that protein intake is associated with increased risk of total mortality and that IGF-1 moderates this association (46). In particular, intake of carbohydrates, dairy products, and B vitamins have been implicated to be associated with IGF-1 levels (41). Whether IGF-1 moderates the relationship between protein intake/overall nutritional status and cancer, cardiovascular, and total mortality endpoints in the context of the EPIC-Heidelberg cohort is a subject of ongoing research.
An interesting finding was the increased risk of lung cancer incidence among participants with low circulating IGF-1 levels, with the risk even higher among participants with indications of liver damage. A previous meta-analysis concluded that there was no association between IGF-1 and risk of lung cancer incidence (18). On the other hand, analysis of the UK Biobank showed that IGF-1 was associated with increased risk of some cancers as well as reduced risk of other cancers including lung cancer (16). An inverse association between IGF-1 and lung cancer incidence is counterintuitive because increased levels of IGF-1 are expected to promote cell proliferation and inflammation and believed to promote tumor development (hence an increased risk of cancer). It is possible that residual confounding by smoking intensity may explain this unexpected observation, such that individuals with very low IGF-1 levels are those with intense smoking behaviors. In this study, mean levels of IGF-1 were slightly lower among male smokers.
The main strength of our study was the ability to use liver enzymes and questionnaire information about prevalent and incident liver-related conditions to account for liver function in assessing the association between IGF-1 and morbidity and mortality outcomes. Our study was also one of the largest with regard to number of mortality endpoints with a long median follow-up period of more than 15 years to investigate the relationship between IGF-1 and several outcomes, including incidence of lung cancer and mortality from digestive diseases—which have been rarely studied. Our study also had some limitations. First, despite the large number of mortality endpoints, we had insufficient numbers of some outcomes to allow for detailed stratified analysis. For instance, it would have been interesting to perform subgroup analyses among participants with low IGF-1 who developed lung cancer to evaluate the effects of smoking on the observed associations.
Despite using extensive criteria to identify individuals with likely reduced liver health at the time of blood collection, it is likely that our criteria misclassified some individuals. For instance, not all individuals with liver diseases exhibit changes in their liver enzyme profiles (47). In analyses among participants without indications of reduced liver health at baseline, such misclassification might have resulted in overestimation of hazards among individuals with low IGF-1 and an underestimation of hazards among those with likely ill liver function. However, it is unlikely that such misclassification was substantial enough to alter our major conclusions. Relatedly, there are no clearly defined cutoffs of liver enzymes for identifying individuals with latent liver dysfunction. We therefore explored using different thresholds of 95th and 90th percentiles of liver enzymes distribution in the subcohort, to define elevated liver enzymes, with similar conclusions. Lastly, despite our extensive efforts to adjust our analyses for known and potential confounders of the relationship between IGF-1 and disease and mortality, it is still possible that unmeasured confounding may explain the observed associations. In particular, we used IGF-1 as a surrogate of free IGF-1, the more biologically active form, because good and validated assays for free IGF-1 and IGF-1 bioactivity are lacking. We also did not have data on biomarkers of liver functions and on medications that could have modified the IGF-1 and liver enzyme profiles of some participants at baseline.
In conclusion, in this large prospective study, we found that IGF-1 exhibited a U-shaped relationship with cancer, cardiovascular, and all-cause mortality such that both individuals with the lowest and the highest circulating levels of IGF-1 were at increased risk of death from these causes. The risk of mortality among participants with low IGF-1 relationship was attenuated but persisted after excluding participants with elevated liver enzymes and other liver conditions, suggesting that ill liver health may explain some but not all risk observed in participants with low IGF-1 levels.
Acknowledgments
The authors are grateful for the continuous participation of the EPIC-Heidelberg cohort participants, without whose commitment this work would not have been possible.
Abbreviations
- ALP
alkaline phosphatase
- ALT
alanine aminotransferase
- AST
aspartate aminotransferase
- BMI
body mass index
- EPIC
European Prospective Investigation into Cancer and Nutrition
- GH
growth hormone
- GHR
growth hormone receptor
- GGT
gamma-glutamyltransferase
- HDL-C
high-density lipoprotein cholesterol
- HR
hazard ratio
- IGF-1
insulin-like growth factor 1
- MI
myocardial infarction
Contributor Information
Trasias Mukama, Division of Cancer Epidemiology, German Cancer Research Center, DKFZ, 69120 Heidelberg, Germany.
Bernard Srour, Nutritional Epidemiology Research Team (EREN), Sorbonne Paris Nord University, Inserm U1153, Inrae U1125, Cnam, Epidemiology and Statistics Research Center–University of Paris-Cité (CRESS), 93017 Bobigny, France.
Theron Johnson, Division of Cancer Epidemiology, German Cancer Research Center, DKFZ, 69120 Heidelberg, Germany.
Verena Katzke, Division of Cancer Epidemiology, German Cancer Research Center, DKFZ, 69120 Heidelberg, Germany.
Rudolf Kaaks, Division of Cancer Epidemiology, German Cancer Research Center, DKFZ, 69120 Heidelberg, Germany.
Funding
The project was supported by the Helmholtz Association's Initiative on Aging and Metabolic Programming (AMPro ZT-0026).
Author Contributions
R.K., T.M., and V.K. contributed to the conception of the study. T.J. organized the sample handling and laboratory analyses. T.M. performed statistical analyses. T.M. drafted the manuscript, which all authors reviewed and revised. All authors reviewed and approved the final manuscript.
Ethics Approval and Consent to Participate
This project is covered by the ethical approval for the EPIC-Heidelberg cohort (Ethical Committee of the Medical Faculty Heidelberg, reference number 13/94).
Disclosures
The authors have no competing interests to disclose.
Data Availability
The EPIC project was launched in the 1990s. Unlike in new studies that we run today, public access to data from the EPIC population was not part of the study protocol at that time. Thus, the data protection statement and informed consent of the EPIC participants do not cover the provision of data in public repositories. Nevertheless, we are open to providing our dataset upon request for (a) statistical validation by reviewers and (b) pooling projects under clearly defined and secure conditions and based on valid data transfer agreements.
References
- 1. Liu JL, Yakar S, LeRoith D. Conditional knockout of mouse insulin-like growth factor-1 gene using the Cre/loxP system. Proc Soc Exp Biol Med. 2000;223(4):344‐351. [DOI] [PubMed] [Google Scholar]
- 2. Milman S, Huffman DM, Barzilai N. The somatotropic axis in human aging: framework for the current state of knowledge and future research. Cell Metab. 2016;23(6):980‐989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Yamamoto H, Sohmiya M, Oka N, Kato Y. Effects of aging and sex on plasma insulin-like growth factor I (IGF-I) levels in normal adults. Acta Endocrinol (Copenh). 1991;124(5):497‐500. [DOI] [PubMed] [Google Scholar]
- 4. Juul A. Serum levels of insulin-like growth factor I and its binding proteins in health and disease. Growth Horm IGF Res. 2003;13(4):113‐170. [DOI] [PubMed] [Google Scholar]
- 5. Bartke A, List EO, Kopchick JJ. The somatotropic axis and aging: benefits of endocrine defects. Growth Horm IGF Res. 2016;27:41‐45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ashpole NM, Logan S, Yabluchanskiy A, et al. IGF-1 has sexually dimorphic, pleiotropic, and time-dependent effects on healthspan, pathology, and lifespan. Geroscience. 2017;39(2):129‐145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Sonntag WE, Csiszar A, deCabo R, Ferrucci L, Ungvari Z. Diverse roles of growth hormone and insulin-like growth factor-1 in mammalian aging: progress and controversies. J Gerontol A Biol Sci Med Sci. 2012;67(6):587‐598. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Steuerman R, Shevah O, Laron Z. Congenital IGF1 deficiency tends to confer protection against post-natal development of malignancies. Eur J Endocrinol. 2011;164(4):485‐489. [DOI] [PubMed] [Google Scholar]
- 9. Guevara-Aguirre J, Balasubramanian P, Guevara-Aguirre M, et al. Growth hormone receptor deficiency is associated with a major reduction in pro-aging signaling, cancer, and diabetes in humans. Sci Transl Med. 2011;3(70):70ra13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ungvari Z, Csiszar A. The emerging role of IGF-1 deficiency in cardiovascular aging: recent advances. J Gerontol A Biol Sci Med Sci. 2012;67(6):599‐610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Pivonello R, Auriemma RS, Grasso LF, et al. Complications of acromegaly: cardiovascular, respiratory and metabolic comorbidities. Pituitary. 2017;20(1):46‐62. [DOI] [PubMed] [Google Scholar]
- 12. Novosyadlyy R, Leroith D. Insulin-like growth factors and insulin: at the crossroad between tumor development and longevity. J Gerontol A Biol Sci Med Sci. 2012;67(6):640‐651. [DOI] [PubMed] [Google Scholar]
- 13. Travis RC, Appleby PN, Martin RM, et al. A meta-analysis of individual participant data reveals an association between circulating levels of IGF-I and prostate cancer risk. Cancer Res. 2016;76(8):2288‐2300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Key TJ, Appleby PN, Reeves GK, Roddam AW. Insulin-like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: pooled individual data analysis of 17 prospective studies. Lancet Oncol. 2010;11(6):530‐542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Murphy N, Knuppel A, Papadimitriou N, et al. Insulin-like growth factor-1, insulin-like growth factor-binding protein-3, and breast cancer risk: observational and Mendelian randomization analyses with ∼430 000 women. Ann Oncol. 2020;31(5):641‐649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Qian F, Huo D. Circulating insulin-like growth factor-1 and risk of total and 19 site-specific cancers: cohort study analyses from the UK biobank. Cancer Epidemiol Biomarkers Prev. 2020;29(11):2332‐2342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kaaks R, Johnson T, Tikk K, et al. Insulin-like growth factor I and risk of breast cancer by age and hormone receptor status-a prospective study within the EPIC cohort. Int J Cancer. 2014;134(11):2683‐2690. [DOI] [PubMed] [Google Scholar]
- 18. Cao H, Wang G, Meng L, et al. Association between circulating levels of IGF-1 and IGFBP-3 and lung cancer risk: a meta-analysis. PLoS One. 2012;7(11):e49884. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Yoon YS, Keum N, Zhang X, Cho E, Giovannucci EL. Circulating levels of IGF-1, IGFBP-3, and IGF-1/IGFBP-3 molar ratio and colorectal adenomas: a meta-analysis. Cancer Epidemiol. 2015;39(6):1026‐1035. [DOI] [PubMed] [Google Scholar]
- 20. Rinaldi S, Cleveland R, Norat T, et al. Serum levels of IGF-I, IGFBP-3 and colorectal cancer risk: results from the EPIC cohort, plus a meta-analysis of prospective studies. Int J Cancer. 2010;126(7):1702‐1715. [DOI] [PubMed] [Google Scholar]
- 21. Murphy N, Carreras-Torres R, Song M, et al. Circulating levels of insulin-like growth factor 1 and insulin-like growth factor binding protein 3 associate with risk of colorectal cancer based on serologic and Mendelian randomization analyses. Gastroenterology. 2020;158(5):1300‐1312.e20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zhang WB, Aleksic S, Gao T, et al. Insulin-like growth factor-1 and IGF binding proteins predict all-cause mortality and morbidity in older adults. Cells. 2020;9(6):1368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. van der Spoel E, Rozing M, Houwing-Duistermaat J, et al. Association analysis of insulin-like growth factor-1 axis parameters with survival and functional status in nonagenarians of the Leiden longevity study. Aging. 2015;7(11):956‐963. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Milman S, Atzmon G, Huffman D, et al. Low insulin-like growth factor-1 level predicts survival in humans with exceptional longevity. Aging Cell. 2014;13(4):769‐771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Schutte A, Conti E, Mels C, et al. Attenuated IGF-1 predicts all-cause and cardiovascular mortality in a black population: a five-year prospective study. Eur J Prev Cardiol. 2016;23(16):1690‐1699. [DOI] [PubMed] [Google Scholar]
- 26. Bourron O, Le Bouc Y, Berard L, et al. Impact of age-adjusted insulin-like growth factor 1 on major cardiovascular events after acute myocardial infarction: results from the fast-MI registry. J Clin Endocrinol Metab. 2015;100(5):1879‐1886. [DOI] [PubMed] [Google Scholar]
- 27. Brugts M, van den Beld A, Hofland L, et al. Low circulating insulin-like growth factor I bioactivity in elderly men is associated with increased mortality. J Clin Endocrinol Metab. 2008;93(7):2515‐2522. [DOI] [PubMed] [Google Scholar]
- 28. Kaplan R, Buzkova P, Cappola A, et al. Decline in circulating insulin-like growth factors and mortality in older adults: cardiovascular health study all-stars study. J Clin Endocrinol Metab. 2012;97(6):1970‐1976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Hu D, Pawlikowska L, Kanaya A, et al. Serum insulin-like growth factor-1 binding proteins 1 and 2 and mortality in older adults: the health, aging, and body composition study. J Am Geriatr Soc. 2009;57(7):1213‐1218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Zhang WB, Ye K, Barzilai N, Milman S. The antagonistic pleiotropy of insulin-like growth factor 1. Aging Cell. 2021;20(9):e13443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Xie Y, Huang C, Zhu X, et al. Association between circulating insulin-like growth factor 1 and risk of all-cause and cause-specific mortality. Eur J Endocrinol. 2021;185(5):681‐689. [DOI] [PubMed] [Google Scholar]
- 32. Burgers AM, Biermasz NR, Schoones JW, et al. Meta-analysis and dose-response metaregression: circulating insulin-like growth factor I (IGF-I) and mortality. J Clin Endocrinol Metab. 2011;96(9):2912‐2920. [DOI] [PubMed] [Google Scholar]
- 33. Svensson J, Carlzon D, Petzold M, et al. Both low and high serum IGF-I levels associate with cancer mortality in older men. J Clin Endocrinol Metab. 2012;97(12):4623‐4630. [DOI] [PubMed] [Google Scholar]
- 34. Adamek A, Kasprzak A. Insulin-like growth factor (IGF) system in liver diseases. Int J Mol Sci. 2018;19(5):1308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Takahashi Y. The role of growth hormone and insulin-like growth factor-I in the liver. Int J Mol Sci. 2017;18(7):1447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Katzke V, Johnson T, Sookthai D, Hüsing A, Kühn T, Kaaks R. Circulating liver enzymes and risks of chronic diseases and mortality in the prospective EPIC-Heidelberg case-cohort study. BMJ Open. 2020;10(3):e033532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Riboli E, Hunt KJ, Slimani N, et al. European Prospective Investigation into Cancer and Nutrition (EPIC): study populations and data collection. Public Health Nutr. 2002;5(6b):1113‐1124. [DOI] [PubMed] [Google Scholar]
- 38. Kulathinal S, Karvanen J, Saarela O, Kuulasmaa K. Case-cohort design in practice - experiences from the MORGAM project. Epidemiol Perspect Innov. 2007;4(1):15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Xue X, Xie X, Gunter M, et al. Testing the proportional hazards assumption in case-cohort analysis. BMC Med Res Methodol. 2013;13(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Mukama T, Srour B, Johnson T, Katzke V, Kaaks R. IGF-1 and risk of morbidity and mortality from cancer, cardiovascular diseases, and all-causes in EPIC – Heidelberg: a case-cohort study - supplementary files. Mendeley Data. Published March 23, 2023;V1. 10.17632/cgwjjk35yg.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Rahmani J, Montesanto A, Giovannucci E, et al. Association between IGF-1 levels ranges and all-cause mortality: a meta-analysis. Aging Cell. 2022;21(2):e13540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Janssen JAMJL. IGF-I and the endocrinology of aging. Curr Opin Endocr Metab Res. 2019;5:1‐6. [Google Scholar]
- 43. Norat T, Dossus L, Rinaldi S, et al. Diet, serum insulin-like growth factor-I and IGF-binding protein-3 in European women. Eur J Clin Nutr. 2007;61(1):91‐98. [DOI] [PubMed] [Google Scholar]
- 44. Crowe FL, Key TJ, Allen NE, et al. The association between diet and serum concentrations of IGF-I, IGFBP-1, IGFBP-2, and IGFBP-3 in the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev. 2009;18(5):1333‐1340. [DOI] [PubMed] [Google Scholar]
- 45. Bajracharya R, Katzke V, Mukama T, Kaaks R. Effect of iso-caloric substitution of animal protein for other macro nutrients on risk of overall, cardiovascular and cancer mortality: prospective evaluation in EPIC-Heidelberg cohort and systematic review. Nutrients. 2023;15(3):794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Levine ME, Suarez JA, Brandhorst S, et al. Low protein intake is associated with a major reduction in IGF-1, cancer, and overall mortality in the 65 and younger but not older population. Cell Metab. 2014;19(3):407‐417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Lominadze Z, Kallwitz ER. Misconception: you can't have liver disease with normal liver chemistries. Clin Liver Dis (Hoboken). 2018;12(4):96‐99. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The EPIC project was launched in the 1990s. Unlike in new studies that we run today, public access to data from the EPIC population was not part of the study protocol at that time. Thus, the data protection statement and informed consent of the EPIC participants do not cover the provision of data in public repositories. Nevertheless, we are open to providing our dataset upon request for (a) statistical validation by reviewers and (b) pooling projects under clearly defined and secure conditions and based on valid data transfer agreements.