| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Social Medicine (J.S., G.D.S., Y.B.-S.), University of Bristol, Bristol BS8 2PR, United Kingdom; Clinical Science South Bristol (J.H.), University of Bristol, Bristol Royal Infirmary, Bristol BS2 8HW, United Kingdom; and Centre for Paediatric Epidemiology and Biostatistics (T.J.C.), University College London Institute of Child Health, London WC1N 1EH, United Kingdom
Address all correspondence and requests for reprints to: Dr. Jat Sandhu, Department of Social Medicine, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, United Kingdom. E-mail: jat.sandhu{at}bristol.ac.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The objective of the study was to examine the association between timing of puberty and adulthood serum IGFs (IGF-I and IGF binding protein-3).
Design: This was a retrospective cohort study.
Setting: Male pupils who attended a single school in Southern England were part of the study.
Participants: Participants in the study were a cohort of 1028 men born between 1927 and 1956 with anthropometric measures between 9 and 18 yr and adulthood serum IGF levels.
Main Outcome Measure: The study measured serum IGF-I and IGF binding protein-3 at mean age 63 yr.
Results: Age at peak height velocity (APHV) was inversely associated with adult IGF-I levels. IGF-I decreased by 3.7 ng/ml (95% confidence interval 1.0–6.4, P = 0.007) for each SD increase in APHV. Prepubertal childhood height and body mass index were both inversely associated with APHV (P trend < 0.001). APHV was positively associated with adult height and inversely associated with adult body mass index. Adjustment for childhood, adult anthropometry, and other lifestyle factors did not substantially alter the association between APHV and adult IGF-I.
Conclusions: This is the first study to document an association between timing of puberty and adult IGF-I levels. A better understanding of life course determinants of the IGF system may provide new insights into disease etiology and primary prevention.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
It has been suggested that cells with accelerated rates of division and proliferation are predisposed to the development of cancer (7). Because IGFs are important regulators of these activities, the role of the IGF family in cancer development and progression is therefore at the very least theoretically plausible (8). Some studies have demonstrated elevated serum levels of IGF-I to be associated with increased risk of prostate cancer (9, 10), breast cancer in premenopausal women (11), and colorectal cancer (12).
Dietary manipulation is known to alter IGF levels; restriction decreases the serum concentration of IGF-I in both humans and animals, whereas excess calorie intake may increase serum IGF-I in humans (13). Dietary restriction reduces the incidence of cancer in many animal models (14, 15, 16). We have previously replicated these observations in a human population using the Boyd-Orr cohort, in which an increase in 1 MJ/d of calorie intake during childhood was associated with a 20% increase in non-smoking-related adult cancers (17). Dietary intake will also be one of several factors that determine childhood growth and the timing of puberty (18). We have shown that a 1 SD increase in leg length, a marker of prepubertal growth, was associated with a more than doubling of risk in hormone-dependent cancers (19). Puberty is also clearly a critical period in human development with several lines of evidence indicating that changes at puberty have long-term implications, particularly for determining the subsequent development of cancers such as those of the breast and prostate in later life (20, 21, 22, 23, 24). A recent study of 117,415 Danish women found age at peak growth to be inversely associated with breast cancer risk (25).
Circulating IGF-I levels vary throughout life, increasing from birth to a pubertal peak before declining steadily after age 20 yr (26). However, almost nothing is known about whether growth in childhood and the timing of puberty have any long-term influences on adult IGF levels. We hypothesized that age at peak height velocity (APHV), a proxy marker for the timing of puberty, would be inversely associated with adult IGF levels and that this may provide the biological pathway linking previous studies (27) of anthropometry and certain types of cancer.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The study population for this historical cohort follow-up comprised 3175 former male students of Christ’s Hospital (CH) born between 1927 and 1956. Students born in these years attended between 1936 and 1969 during which regular measures of height and weight were conducted. Dr. Gerald Edward Friend, an influential figure in the field of schoolchildren’s nutrition (28, 29), was Medical Officer at CH from 1913 to 1946. He implemented a regimen of regular growth monitoring that was maintained within the school for decades. The regimen of twice-a-term height measurements and three-times-a-term weight measurement implemented since 1918 provides what are considered remarkably good data for the period (30). Growth record cards with these measures were available and formed the basis of defining this cohort.
A combination of the CH alumni network and the National Health Service, U.K. Central Register was used to trace the cohort and invite them to take part in the study. Participants were sent a detailed self-administered postal questionnaire that covered participant’s medical history, past and current lifestyle factors, current and recalled anthropometric measures (including self-reported leg length with provided tape measure), family history, self-rated health, and sociodemographic details. Childhood socioeconomic position (SEP) was derived from the occupation of the participant’s father, recorded at school entry, to code a five-category occupational social class based on contemporary occupations of the time.
Cohort members were also invited to attend their local general practice so that a practice nurse could take a blood sample and measure their height and weight. Blood was initially stored in a refrigerator at 5 C before being mailed to the researchers in Bristol. All received specimens were processed using a standard protocol that involved centrifugation, separation, aliquoting, and storage in a –80 C freezer (31).
Ethics approval
The South West Multi-Centre Research Ethics approved tracing of the cohort and the follow-up study, as did the data custodians (CH).
Growth data
The cohort members had on average 54 (SD±16) anthropometry measures during their time at CH (ages 9–18 yr). Given the frequency of these measurements, it was expected that some data errors would have occurred. Therefore, we applied a natural cubic smoothing spline to each individual’s growth curve to reduce noise due to errors. Spline smoothing trades smoothness of the fitted function against conformity to the data by pasting together cubic polynomials so that at the knots the curve has continuous first derivatives (32). The first derivative of the smoothed growth trajectory yields the velocity curve.
Smoothed height and weight data were used to calculate body mass index (BMI), and all growth measures were transformed to an internally derived age-specific z-score. We also derived the timing of maximum pubertal growth known as the age at peak height velocity (APHV). This provides an indication of the timing of puberty and was obtained from the individual velocity curves. This method for determining the timing of puberty is used frequently and is regarded as reliable (33, 34). Because the degree of smoothing can alter the shape of the velocity curve and hence the derived APHV, we undertook a sensitivity analysis by repeating the above process but reducing the degree of smoothing to allow for a more complex curve shape and a better-defined peak on the height velocity curve.
IGF-I and IGF binding protein (IGFBP)-3 assays
Serum levels of IGF-I were determined by RIA using a monoclonal antibody (Blood Products, Elstree, Hertfordshire, UK) and recombinant peptide (Pharmacia, Stockholm, Sweden) for standard and tracer after iodination using the chloramine-T method. Samples were analyzed after acid-acetone extraction to remove the IGFBPs with an excess of IGF-II added to the extract to saturate any residual binding proteins (35) (intraassay coefficient of variation 8%, interassay coefficient of variation 12%). Serum levels of IGFBP-3 were determined by RIA using an in-house polyclonal antibody raised against recombinant nonglycosylated IGFBP-3. The assay was calibrated against recombinant glycosylated IGFBP-3 (Dr. C. Maack, Celltrix, Santa Clara, CA). Samples were analyzed after a 1:100 dilution with antibody used at a final dilution of 1:20,000 (35) (intraassay coefficient of variation 3.8%, interassay coefficient of variation 14%).
Only samples that were received within 4 d of venesection were used in the analysis. There was no evidence that values of IGF-I and IGFBP-3 were related to the delay from taking the sample and receipt (P = 0.74 and P = 0.37, respectively) (31).
Statistical analysis
An initial assessment was made of the baseline characteristics available for all cohort members (year of birth, age at entry/leaving CH) by follow-up status (National Health Service, U.K. Central Register trace, questionnaire response, clinic attendance). For continuous variables Student’s t test was used to test the difference between two groups, whereas categorical variables were compared using cross-tabulations and 
2 tests.
Multiple linear regression was used to assess the independent association of APHV with childhood (social class, height, and BMI at entry) and adulthood (social class, leg length, height, BMI) measures. Additionally, the association of anthropometry, SEP, and adult lifestyle measures with circulating serum IGF levels in adulthood was examined. We included an individual’s year of birth in all analyses to control for secular trends in growth. The adjusted means, 95% confidence intervals (CIs), and P values for linear trend are presented.
We conceptualized, a priori, five potential pathways linking APHV and adult IGF levels (Fig. 1
). The first is a direct causal pathway (a) whereby the timing of puberty has a long-term effect on the GH-IGF axis so that this is either up- or down-regulated into adulthood. The second pathway (b) suggests that any observed association is secondary to prepubertal IGF levels, measured indirectly through prepubertal anthropometry and adult leg length. Other studies have shown a positive association between prepubertal height and cross-sectional IGF-I levels (36, 37). Higher prepubertal IGF levels will increase childhood growth and trigger an earlier onset of puberty as well as be associated with higher adult IGF levels assuming IGF levels track over the life course. This second pathway may be determined by environmental factors, such as breast-feeding and diet (38), infections, etc., acting in early life (c), which we measured indirectly by childhood SEP. It is known that worse childhood SEP for a group of this age would have been associated with more overcrowding, less sanitary conditions, increased risk of Helicobacter pylori infection, and slower childhood growth (39, 40, 41, 42). Childhood SEP may confound the association because it may have a role in determining childhood IGF levels and hence the timing of puberty and adult SEP and other adult lifestyle factors, such as smoking or alcohol consumption, that may influence adult IGF levels (d). Finally, there may be genetic factors (e) that either directly or indirectly determine childhood IGF, the timing of puberty, and adult IGF levels. We had no direct measures of genetic factors and therefore used a crude proxy based on adult anthropometry. Adult height and leg length are clearly related to both genetic and environmental factors; however, because all the children in this cohort shared similar environments while boarding at the school, there would be less environmental heterogeneity than that found in general populations, and hence, the relative importance of genetic factors will be enhanced. Both twin and familial studies suggest that genetic influences on height are profound and explain the largest component of variance (43, 44).
|
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Eighty-two percent (n = 2593) of cohort members were successfully traced. Untraced subjects were born later (mean year of birth 1939 for traced vs. 1944 for untraced, P < 0.001) and left CH at a younger age (17.4 yr for traced vs. 16.6 yr for untraced, P < 0.01). Some of these participants (285) were known to have died so that we contacted only 2308 subjects (89%). Sixty-six percent (n = 1520) agreed to participate by completing the questionnaire, 5% (n = 126) declined to participate, and 29% (n = 662) did not respond. Like the untraced subjects, nonparticipants and nonresponders tended to have been born later (mean year of birth 1938 for participants vs. 1940 for nonparticipants and nonresponders, P < 0.001) and left school slightly younger than participants (age at leaving school for participants 17.6 yr vs. 17.3 yr for nonparticipants and nonresponders, P < 0.01).
Sixty-eight percent (n = 1028) of participants who completed the questionnaire also consented to visit their general practice clinic. Participants who declined to attend were not significantly different from clinic attendees for age at entry (attendees 10.6 yr vs. nonattendees 10.7 yr, P = 0.2) and school exit (attendees 17.5 yr vs. nonattendees 17.5 yr, P = 0.7), although they were born slightly later (mean year of birth 1938 for attendees vs. 1939 for nonattendees, P = 0.03).
Early growth data were of sufficient quality to determine APHV for 99.7% of participants (Table 1), but there were far more missing data for adult measures.
|
Timing of puberty as measured by APHV was significantly inversely associated with BMI both at school entry and in adulthood as measured in the clinic. However, it showed different relationships with height so that it was inversely associated with childhood height but positively associated with adult height and leg length measurement (Table 2). SEP in childhood or adulthood was not associated with APHV.
|
Adult IGF was unrelated with child anthropometry, whereas IGF-I was inversely associated with both leg length and BMI in adults. Leg length was also inversely associated with IGFBP-3, but adult BMI was positively associated with IGFBP-3 so that there was a strong inverse association between adult BMI and the molar ratio (P < 0.001) (Table 3). Neither childhood nor adult social class showed any association with IGF-I, IGFBP-3, or the molar ratio.
|
Never- and ex-occasional smokers showed higher IGF-I levels than current or ex-smokers (P value for trend = 0.04). Among alcohol drinkers there were strong inverse trends for both IGF-I and the molar ratio so that heavier drinkers had lower IGF levels (P value for trend = 0.003 and 0.002, respectively). No pattern was seen with milk consumption.
Timing of puberty and adult serum IGF levels
An inverse association between timing of puberty and serum IGF measures was seen (Fig. 2). The mean adjusted serum IGFBP-3 level exhibited a weak trend across quartiles of APHV, whereas a more robust trend was seen with mean adjusted serum IGF-I levels.
|
). Adjustment for childhood anthropometry (model b) as a proxy measure of prepubertal IGF, if anything, strengthened the effect estimate of APHV on adult IGF-I (4.3 ng/ml, 95% CI 1.5–7.1, P = 0.003). Adjustment for childhood SEP (model c) made little difference (3.9 ng/ml, 95% CI 0.9–6.9, P = 0.01). Because this variable was missing for 19% of subjects, we did not include it in the subsequent models to preserve statistical power. Further adjustment for adult risk factors (model d) and adult anthropometry (model e) as crude measures of genetic influences on adult IGF levels attenuated the associations toward the null.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Evidence from our previous report of early childhood milk supplementation (46) and that from the Dutch famine study (47) indicates that childhood exposures can lead to long-term resetting of the GH-IGF axis. These findings support our assertion that exposures that affect the timing of puberty could also cause a resetting of this axis. Therefore, puberty may be yet another critical period for programming of the axis with potential consequences on chronic disorders later in life.
Recent work by Ahlgren et al. (25), who reported that a lower age at peak growth was associated with increased risk for breast cancer, is consistent with our observation. A similar association has also been found for earlier age of menarche (22, 48). These observations reflect the potential importance of pubertal timing on disease etiology and suggest that circulating IGF-I levels may be one of the pathophysiological pathways that increase the risk of several common cancers (8).
The main determinant of adult height after adjustment for prepubertal growth is likely to be the timing, duration, and intensity of the pubertal growth spurt, suggesting that pubertal influences may also determine IGF-I levels around puberty. Although not all studies show an association between timing of peak velocity and adult height (49), a large Swedish population-based cohort, with repeat measures from birth to 18 yr, found that final height was strongly predicted by APHV (50).
The strong associations between height and IGF-I in children and adolescents are not usually replicated in adults (9, 51). We did not observe an association between childhood or adult height and IGF-I levels. This is similar to other studies with childhood measures (52, 53). We did find a modest inverse trend with leg length for IGF-I and IGFBP-3. One must be careful not to overinterpret this observation (P = 0.04), but it is internally consistent because taller children would have had an earlier puberty resulting in a shorter period of prepubertal growth and hence possibly shorter legs in adulthood. This observation appears to contradict our previous finding that an increase in leg length in the Boyd Orr cohort was associated with an increased risk of cancer (19); however, this was not confirmed in the Caerphilly cohort study using a measure of adult leg length, although this study had limited statistical power. Leg length measured in childhood is not the same as final attained leg length, which is determined both by childhood growth and the timing and duration of the pubertal growth spurt (27). However, in the Boyd Orr cohort, there was a moderately strong correlation (0.61) between childhood and adult leg length (Martin, R., personal communication).
Study strengths and limitations
The main strength of our study is the availability of detailed anthropometric data in adolescence on a relatively large cohort with long follow-up time and good participation rate for provision of blood specimens. By examining serum levels of IGF in adulthood, 50 yr after biological maturation, the impact of early and late puberty timing in adolescence could be assessed. Few studies have such regular measures over 6–10 yr to enable estimation of APHV. The adolescent growth spurt was determined as an indicator of somatic maturation. The frequent measurements of height resulted in our being able to estimate the APHV with precision; this was found to be consistent with self-reported age at shaving initiation, a late-stage puberty event (data not shown). However, even with such detailed data, problems can arise because the degree of smoothing can result in different velocity profiles, which occasionally produced divergent results in the APHV, depending on whether less or more smoothing was used. Our results using the less smoothed data did produce weaker effect estimates. It was not possible for us to determine which was the more valid method, although qualitatively our results were not altered.
We were fortunate to have reasonably good data on many childhood and adult potential confounders. We did not have detailed dietary data or any specific measures of genetic influences on IGF levels, although we hope to test genetic factors more directly in the future by using stored DNA from consenting participants. We do not as yet know whether our measure of APHV will predict risk of specific cancers because the cohort members are still relatively young and we have had too few events up to now. Our male cohort members are more likely to have higher adult SEP than a random population sample. One must therefore be cautious in generalizing our results to the general population.
Implications
The association between APHV and adult IGF levels may indicate tracking of IGF from puberty into adulthood as a consequence of the GH-IGF axis being programmed during a sensitive period of growth and development. Ours is the first study to document such an association, and ideally future studies will collect repeat IGF levels in childhood, peri- and postpuberty, and adulthood to confirm our findings. The fact that adult IGF-I, some 50 yr after puberty, is still associated with the timing of puberty is in many ways remarkable. The slowly growing body of empirical evidence suggests that there may be important life course determinants of the IGF system acting in the postnatal (46) and childhood period (47). A better understanding of the association between early growth patterns and the risk of cancer could improve our knowledge of the mechanisms of the disease and offer opportunities for prevention as well as predict secular trends in cancer incidence, given the marked changes in height and weight seen over the last 50 yr (54, 55).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
The authors have nothing to declare.
First Published Online May 23, 2006
Abbreviations: APHV, Age at peak height velocity; BMI, body mass index; CH, Christ’s Hospital; CI, confidence interval; IGFBP, IGF binding protein; SEP, socioeconomic position.
Received October 21, 2005.
Accepted May 15, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K. Casazza, P. B. Higgins, J. R. Fernandez, M. I. Goran, and B. A. Gower Longitudinal Analysis of the Insulin-Like Growth Factor System in African-American and European American Children and Adolescents J. Clin. Endocrinol. Metab., December 1, 2008; 93(12): 4917 - 4923. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. J. Cole, M. Cortina-Borja, J. Sandhu, F. P. Kelly, and H. Pan Nonlinear growth generates age changes in the moments of the frequency distribution: the example of height in puberty Biostat., January 1, 2008; 9(1): 159 - 171. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Troisi, N. Potischman, and R. N. Hoover Exploring the Underlying Hormonal Mechanisms of Prenatal Risk Factors for Breast Cancer: A Review and Commentary Cancer Epidemiol. Biomarkers Prev., September 1, 2007; 16(9): 1700 - 1712. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Ben-Shlomo Rising to the challenges and opportunities of life course epidemiology Int. J. Epidemiol., June 1, 2007; 36(3): 481 - 483. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
