This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Parent, A.-S. Articles by Bourguignon, J.-P. Search for Related Content PubMed PubMed Citation Articles by Parent, A.-S. Articles by Bourguignon, J.-P.

The Timing of Normal Puberty and the Age Limits of Sexual Precocity: Variations around the World, Secular Trends, and Changes after Migration

Division of Ambulatory Pediatrics and Adolescent Medicine (A.-S.P., J.-P.B.), University of Liège, B-4000 Liège, Belgium; Department of Growth and Reproduction (G.T., A.J., N.E.S.), Copenhagen University Hospital, Rigshospitalet, DK-2100 Copenhagen Ø, Denmark; and Departments of Paediatrics and Physiology (J.T.), University of Turku, 20520 Turku, Finland

Correspondence: Address all correspondence and requests for reprints to: Jean-Pierre Bourguignon, M.D., Ph.D., Division of Ambulatory Pediatrics and Adolescent Medicine, University of Liège, Centre Hospitalier Universitaire Sart Tilman, B-4000 Liège, Belgium. E-mail: jpbourguignon{at}ulg.ac.be

	Abstract

Top Abstract I. Introduction II. Normal Puberty and... III. Precocious Puberty IV. Possible Mechanisms of... V. Conclusion and Future... References

 
During the past decade, possible advancement in timing of puberty has been reported in the United States. In addition, early pubertal development and an increased incidence of sexual precocity have been noticed in children, primarily girls, migrating for foreign adoption in several Western European countries. These observations are raising the issues of current differences and secular trends in timing of puberty in relation to ethnic, geographical, and socioeconomic background. None of these factors provide an unequivocal explanation for the earlier onset of puberty seen in the United States. In the formerly deprived migrating children, refeeding and catch-up growth may prime maturation. However, precocious puberty is seen also in some nondeprived migrating children. Attention has been paid to the changing milieu after migration, and recently, the possible role of endocrine- disrupting chemicals from the environment has been considered. These observations urge further study of the onset of puberty as a possible sensitive and early marker of the interactions between environmental conditions and genetic susceptibility that can influence physiological and pathological processes.

I. Introduction

II. Normal Puberty and Variations in Timing

A. Clinical indicators

B. Age references in well-off conditions

C. Age references in underprivileged conditions

D. Secular trends

III. Precocious Puberty

A. Age limits

B. Common etiologies

C. Children migrating from developing countries

IV. Possible Mechanisms of Variations in Timing of Puberty around the World and after Migration

A. Genetic factors (family, ethnicity, gender)

B. Intrauterine conditions

C. Nutrition

D. Other stresses

E. Light-darkness cycle and climatic conditions

F. Exposure to endocrine-disrupting chemicals (EDCs)

V. Conclusion and Future Research Directions

	I. Introduction

Top Abstract I. Introduction II. Normal Puberty and... III. Precocious Puberty IV. Possible Mechanisms of... V. Conclusion and Future... References

 
PUBERTY RESULTS FROM the awakening of a complex neuroendocrine machinery in which the primary mechanism is still unclear (1). A peculiarity of sexual maturation in the human species is the 4- to 5-yr physiological variation in age at onset of puberty that is observed among normal individuals despite relatively similar life conditions (2). This variability involves genetic factors, as indicated by the studies on heritability of menarcheal age (3), although the molecular determinants are yet to be identified. Other factors such as ethnicity, nutritional conditions, and secular trends have been shown to influence the physiological range in age at the onset of puberty (4, 5). In these conditions, the age limits used to define sexual precocity are necessarily subject to local assessment and regular revision. Whereas reference data seemed to have stabilized in most industrialized countries during the 1990s, two recent American studies (6, 7), which were reviewed by Lee et al. (8), highlighted an unexpected and unexplained advance in physiological age at the onset of breast development. These findings urged us to examine the current variations in timing of puberty around the world and the related age limits for sexual precocity. In addition, new aspects regarding the etiology of sexual precocity provided another reason to consider the differences in timing of puberty worldwide. Most patients with sexual precocity are diagnosed as idiopathic, and no causative process is found at brain imaging (9, 10). Recently, a special form of sexual precocity has been described in foreign children adopted from developing countries to Western Europe. The occurrence of early or precocious sexual maturation in such a unique situation of changing environment was originally described in Indian girls adopted in Sweden (11). Subsequently, similar observations were made in several European countries and involved children from various countries (12, 13, 14, 15, 16, 17, 18).

The aim of this review is to address the significance of early onset of puberty in children living in their home countries or migrating to other countries. Emphasis will be put on the differences in age limits for pubertal development between countries and ethnic groups. The possible etiologies for advancement in timing of puberty in some conditions will be discussed in the light of our current understanding of the mechanism of the onset of puberty.

	II. Normal Puberty and Variations in Timing

Top Abstract I. Introduction II. Normal Puberty and... III. Precocious Puberty IV. Possible Mechanisms of... V. Conclusion and Future... References

 
A. Clinical indicators
Although the concerns about sexual precocity and changes in timing of puberty appear to be much greater in girls than in boys, these issues must be addressed in both sexes, in a comprehensive and comparative perspective. In boys, the first sign of pubertal development is an increase in testicular volume above 3 ml, consistent with Tanner G2 stage (2, 9). This change, however, can be observed only by thorough evaluation at physical examination (2, 19). There is no reliable event that can be recalled to time male pubertal development when taking a medical history. In girls, the earliest manifestation of puberty is acceleration in growth velocity, which, however, is difficult to ascertain because it requires several accurate height measurements each year (20). The commonly used markers of the timing of female puberty are thelarche and menarche, of which different characteristics are summarized in Table 1. Thelarche is the first appearance of breast development defined as Tanner B2 stage (2). This relatively obvious pubertal sign might not be easily distinguished from fat tissue in slightly obese girls. Importantly, the appearance of pubic hair is dependent primarily on testicular androgens in boys and on adrenal androgens in girls. The developmental increase in adrenal androgen secretion is referred to as adrenarche and occurs independently of the pituitary- gonadal maturation or gonadarche (8, 21, 22). Thus, development of pubic hair in girls does not provide any information on pituitary-ovarian maturation. The evaluation of Tanner stages can be obtained by self-assessment or physician’s assessment. The self-assessment of pubertal characteristics has been shown to be nonreliable (23), although the answers of adolescents to a global question on development compared with their peers can be a valid approach (24). The assessment of Tanner stages by professionals provides more reliable information than self-assessment but involves significant variations between observers (23). The Tanner stages provide semiquantitative information with less accuracy than using the menarcheal age to assess the timing of pubertal development.

In the published literature on the timing of puberty, the data are given either as mean ± SD in some studies, assuming that the data are normally distributed, or as median and centiles in other studies. This issue is important because the mean and median are comparable only when the distribution of data is Gaussian (Fig. 1A). This can be the case in a well-off setting, as indicated by the symmetrical distribution of centiles of menarcheal age in North American girls (31). By contrast, an asymmetrical distribution of data with increased variability toward late ages (Fig. 1B) can be observed in an underprivileged setting (32). In these conditions, the calculated mean age will be greater than the median. An increased prevalence of early sexual maturation in a population such as migrating children can result from either a subset of fast maturers causing an asymmetrical distribution of ages at onset of puberty (Fig. 1C) or a shift of the whole population toward earlier timing of puberty (Fig. 1D), or both. The magnitude of the SD provides an index of variability in timing of puberty. For instance, during the 19th century (33), not only was the average menarcheal age later but also the SD was greater than today (Fig. 1E).

View larger version (20K): [in this window] [in a new window]   FIG. 1. Schematic representation of the distribution of timing of puberty (e.g., menarcheal age) in different conditions. The important parameters are the median and mean ages, the symmetry or asymmetry of distribution, and the variability as expressed by the SD values (–2 SD to +2 SD illustrated).

B. Age references in well-off conditions
In the United States, Mac Mahon (35) reported a mean menarcheal age of 12.8 yr in 1973. Very similar data (12.7 yr) were reported by Tanner and Davies (31) in 1985 with a median age of 10.9 yr for onset of breast development. This is very close to the data obtained in most European countries (Fig. 2). In 1997, however, a publication from the American Academy of Pediatrics-Pediatric Research in Office Settings (PROS) network reporting on more than 17,000 girls led to an adjustment in the above-mentioned limits in the United States (6). These authors found the mean age at B2 to be 10.0 yr in White American girls and 8.9 yr in African-American girls, with –2 SD limits of 6.3 and 5.0 yr, respectively. These findings generated comments on the possible overestimation of breast development because assessment was made through visual inspection only, whereas palpation may be required to distinguish between adipose and glandular tissue (8, 36). In addition, the data were not corrected for the difference in racial representation that resulted from only 10% of the participants in the PROS study being African-Americans (37). This, together with other possible biases such as the increased variance due to the involvement of multiple observers, raises the issue of whether or not the onset of puberty is shifted toward earlier ages in the United States (36, 37). In another large cross-sectional American study (the National Health and Nutrition Examination Survey, NHANES III), a similarly early median age of 9.7 yr at B2 was found in White Americans (7), although slightly less advanced median ages of 10.4 yr (White Americans) and 9.5 yr (Black Americans) were recently reported in subgroups from that cohort (38). The large NHANES III study, however, was not different from the PROS study in that it was cross-sectional and involved many different observers, without mention of the possible or systematic use of palpation to assess breast tissue development. Therefore, confirmation of those findings may warrant a carefully designed prospective longitudinal study involving a limited number of observers who use breast palpation to assess development. Interestingly, the mean menarcheal age found in the PROS study (6) did not show the same shift as age at the onset of breast development because age at menarche (12.9 yr in White Americans and 12.7 yr in Black Americans) was unchanged when compared with the data reported earlier (31, 35). Here also, a bias was possible in the PROS study where the subjects were aged 12 yr or less and mean menarcheal age was extrapolated by probit analysis. However, in the NHANES III study (7), which involved subjects aged up to 17 yr, menarche occurred in White Americans at an age (12.5 yr) similar to that extrapolated in the PROS study. The similar menarcheal ages observed in the United States recently (6, 7) and several decades earlier (31, 35) suggest that the secular trend toward earlier menarche has stopped. This issue was more specifically addressed in two recent studies based on the NHANES III data and the National Health Examination Survey (NHES) data (39, 40 ). In these two studies, the median menarcheal age in the U.S. girls studied around 1990 was 12.43 or 12.54 yr with a reduction of 0.34 yr in 30 yr or 0.21 yr in 25 yr, respectively. These differences are much less than that described for breast development in the same period. Is it possible that the secular trend toward earlier menarche has been underestimated in the two large cross-sectional American studies (6, 7)? Such a possibility might be consistent with the data from the Bogalusa Heart study, which reported on a smaller biracial cohort of 1082 girls studied in 1992–1994 (41). In this study, the mean menarcheal age was found to be 11.4 and 11.5 yr in African-American and White-American girls, respectively, which is much earlier than in the PROS and NHANES studies and indicates an important secular trend, because in 1978–1979, the menarcheal age was 12.3 and 12.2 yr, respectively, in the same area (41). Several factors might account for this discrepant observation. First, local environmental changes in the semirural area of Bogalusa may be greater than in the United States in general, resulting in faster increase in body fatness. Between 1974 and 1994, the prevalence of body mass index (BMI) greater than the 85th percentile among 5- to 14-yr-old subjects has increased by 22% (from 15 to 37%) in the Bogalusa study (42). This race- and gender-independent increase is much more than the 7% increase found using the same criteria in the NHES study between 1963 and 1991 in 6- to 17-yr-old subjects (43) and the 8% increase found in the adult U.S. population (44). Another bias in the report from the Bogalusa cohort (42) was that two different cohorts of subjects were studied at a 15-yr interval. The impact of such a biasing factor is likely because a recent analysis including a longitudinal subset of 2058 subjects showed that the median menarcheal age in the Bogalusa cohort was 12.6 yr in White Americans and 12.3 yr in Black Americans, with a secular reduction of 0.8 and 0.17 yr during the past 20 yr, respectively (45). In a recent longitudinal study of an equal proportion of White and Black Californian girls, similar mean menarcheal ages of 12.7 and 12.0 yr, respectively, were reported (46). In a cross-sectional study performed in the 1970s and involving 84% of White American girls, the mean menarcheal age was 12.2 yr (47). Thus, it is likely that the average menarcheal age has almost stabilized in the United States, whereas age at onset of breast development may have declined. As already mentioned, the increasing difference between trends in ages at B2 and menarche might involve an inverse correlation between age at onset of puberty and duration of puberty (28, 29, 30). However, a trend toward increasing time length between B2 and menarche is somewhat contradictory to the observation of a secular acceleration of the progression or tempo of pubertal development in Dutch and Swedish boys and girls (48). Alternatively, unidentified external factors might cause breast development to start earlier without affecting menarcheal age.

View larger version (32K): [in this window] [in a new window]   FIG. 2. Average (mean or median) ages at onset of breast development (B2) or menarche in different well-off populations around the world. The most recently published data were used to provide an update.

In the pioneering work of Marshall and Tanner (19), which provided age references for male pubertal development in 1970, the mean age for G2 stage was found to be 11.6 yr in the United Kingdom. Very similar data have been reported for the United States (11.5 yr) in 1985, Sweden (11.6 yr) in 1996, The Netherlands (11.5 yr) in 2001, and Switzerland (11.2 yr) in 1983 (31, 51, 57, 74). In a longitudinal Spanish study, the mean age at G2 was 12.3 yr (63). These stabilized references are in contrast to the lowered median age of 9.7 yr at G2 that has been reported from the American NHANES III study, which was performed in the period 1988–1994 (75). This study involved 25% of White Americans, 37% of Mexican Americans and 38% of African Americans with respective median ages at G2 of 10.1, 10.4, and 9.5 yr, a difference that was not significant (72). In a more recent subgroup study from the same cohort, similar ages were reported (38). Those data were obtained based on visual inspection without palpation of testicular volume or assessment of testicular size. As mentioned in an editorial comment by Reiter and Lee (76), that bias may be important because the increase in testicular volume or size is critical in the evaluation of the G2 stage of Tanner. The inaccuracy of visual assessment might also account for the one-stage variance between physicians involved in the study (76) and the surprisingly long time interval of 2 yr between G2 and P2 stages (75), whereas this interval was found to be 1 yr by others (51, 59). There are unfortunately no data available from comparable studies. In the NHES study, G2 data were not available (77). In a longitudinal study of 78 boys, Roche et al. (78) reported a mean age of 11.3 yr at G2, but the validity of these data may be limited because of self-assessment. Biro et al. (79) also performed a longitudinal study, but they used modified global stages of puberty that are not comparable. Thus, it is difficult to draw conclusions before we have additional data from a prospective study with assessment of testicular volume or size. The stable mean age at G5 (15.3 yr) in the NHANES III study is in contrast with the earlier age at G2. This would suggest that the tempo of male puberty is, in fact, decreasing. In some countries such as Hong Kong where the median ages at G2 and B2 are 11.4 and 9.8 yr, respectively (66, 80), the age at G2 in boys has not changed, whereas the age at B2 is declining. Although girls classically precede boys by an average of 0.8 yr for the onset of puberty (51, 53, 57), such a sexual dimorphism has not been found any longer in two recent American and German studies where identical ages at B2 and G2 were reported (6, 55, 75, 81). Such discrepancies point to the need to compare studies conducted using similar design and methods in different countries.

C. Age references in underprivileged conditions
In developing countries, inequalities related to socioeconomic status or life setting (urban vs. rural) are still prominent and might account for important variations in timing of puberty within and among countries (5, 65). Secular trends can still be observed in developing countries because of current changes in living standards. When subgroups of girls in well-off and underprivileged conditions are compared in some countries (Fig. 3), the impact of these differences in living standards is obvious (5, 68, 72, 73, 82, 83). The socioeconomic status of the study group is not always specified, and we can then assume an average living standard. In recent reports from Nigeria, Guatemala, or Colombia (32, 84, 85), later ages at menarche were observed compared with well-off girls from the neighboring countries of Chile and Venezuela (70, 71). In countries such as China and Senegal where girls living in underprivileged conditions have been studied, the mean menarcheal age is as great as 16.1 yr (86, 87). The variability in menarcheal age can be translated into the coefficient of variation (CV), which is the SD expressed as a percentage of the mean. The CV is greater in underprivileged girls from Cameroon (11.6%) than in privileged girls from the same country (8.2%) (72), although such a difference is not observed in South Africa (73). A low CV (9.2–10.0%) is obtained in other well-off conditions (6, 64, 69). Taken together, these data highlight the crucial role of socioeconomic and nutritional conditions in the timing of puberty. A similar conclusion was drawn in a recent study reviewing menarcheal age in 67 countries (88). These authors pointed out that, on a large scale, extrinsic factors were crucial: the effects of early involvement in physical activities on energy expenditure were emphasized in relation to frequent illiteracy rate.

View larger version (16K): [in this window] [in a new window]   FIG. 3. Average (mean or median) menarcheal age in different developing countries. The data are shown separately for different living conditions.

View larger version (22K): [in this window] [in a new window]   FIG. 4. Secular trend in menarcheal age in different countries around the world. A, Twin set of average data obtained after 1960 in different countries at several 10-yr intervals. B, Secular difference in menarcheal age calculated from those data. For the purpose of comparison, the average secular trend observed before 1960 in the United States and in Europe is also shown.

The variability in menarcheal age in France as reflected by the CV was greater in the mid-19th century (17.3%) than in the mid-20th century (10.0%) (33, 60). It appears, however, that between the 1930s and the 1960s in the United States, CV was 9–10% and did not change any longer, whereas the decline in average menarcheal age was still progressing (89). In China, where the secular trend in decline of menarcheal age was still important between the early 1970s and the early 1990s (Fig. 4), there was only a modest concomitant reduction in that CV from 11.9 to 9.8% (87). Such a dissociation between the mild changes in variability of age at menarche and the steady decline in average menarcheal age suggests that both parameters are influenced by different factors or differently by the same factors.

Some authors have looked into the secular trend dynamics using other markers such as the onset of breast development in girls and genital development in boys. In Sweden and the United Kingdom, the onset of breast development has been found to be slightly earlier in 1980 than in the 1960s or 1970s (48). In Denmark (97), the average age at onset of breast development even tended to increase between 1964 (10.6 yr) and 1990–1991 (11.2 yr). In Europe, there is no evidence of marked secular reduction in average age at B2 such as shown in the United States (6, 7). The age at G2 has shown some increase in The Netherlands during the last decades, whereas a decrease has been observed in Sweden (48). Although a methodological bias is possible due to the small number of boys in the G2 stage, such discrepant observations might point to some country-specific environmental changes. These observations are also raising important issues regarding the significance of the data in relation to the markers used to evaluate the timing of pubertal development as discussed above.

	III. Precocious Puberty

Top Abstract I. Introduction II. Normal Puberty and... III. Precocious Puberty IV. Possible Mechanisms of... V. Conclusion and Future... References

 
A. Age limits
On the basis of the above mentioned average ages at onset of pubertal development and assuming a Gaussian distribution in the normal population, abnormally precocious sexual development has been defined in Europe as less than 8 yr for the B2 stage in girls and less than 9 yr for the G2 stage in boys (10, 98, 99). These age limits, which are above the 99th centile, have been used for several decades and are still used currently. Similar age limits were thought to be relevant to the United States until the publication of the PROS study (6) prompted Kaplowitz and Oberfield (100) to recommend, on behalf of the Lawson Wilkins Pediatric Endocrine Society, to reset the age limit for nonphysiological onset of breast development at 7 yr in Caucasian girls and 6 in African-Americans. Some American colleagues, however, did not share the opinion of Kaplowitz and Oberfield and advocated a possible recruitment bias through pediatricians’ offices (36). In addition, the unexplained dissociation between earlier breast development and relatively unchanged menarcheal age was raising indirectly the issue of the reality of breast development estimated through visual inspection. The definition of appropriate age limits is crucial to restrict diagnostic evaluation and possible therapeutic intervention to children with abnormal precocious development (98, 99). In a recent review of 223 patients referred for sexual precocity occurring between 7 and 8 yr in white girls or 6 and 8 yr in black girls, Midyett et al. (101) found that 47% showed both breast and pubic hair development, 35% had bone age at more than 3 SD above chronological age, and 12% appeared to have a nonidiopathic form of sexual precocity. The authors concluded that occurrence of sex characteristics between 6 and 8 yr is not necessarily benign and may warrant diagnostic and therapeutic intervention (101). Thus, the question of age limit for sexual precocity does not have any definitive and unequivocal answer.

B. Common etiologies
Sexual precocity is classified as central, gonadotropin-dependent, i.e., driven by the central nervous system, when it results primarily from early hypothalamic-pituitary maturation (8, 99). Central precocious puberty represents four fifths of the total number of patients with precocious puberty (99, 102) and is much more frequently seen in girls than in boys. Idiopathic central precocious puberty is diagnosed when early pubertal development (including acceleration of growth and bone maturation) is associated with a pubertal pattern of gonadotropin secretion (increased LH secretion) and when there is no evidence of organic cause provided through history, physical examination, or brain imaging (9, 98). Among patients with central precocious puberty, the proportion of idiopathic forms varies between 58 and 96%, as reviewed recently (103, 104). This proportion is greater in girls than in boys, who show a higher prevalence of recognizable organic forms. As shown in Fig. 5, the female to male ratio found among the patients with central precocious puberty in the studies published between 1961 and 1990 was relatively similar, around 3:1 to 4:1 (102, 105, 106, 107, 108, 109, 110, 111, 112). In the same studies, the idiopathic to organic ratio was around 2:1, indicating that two thirds of the patients had idiopathic forms and one third had organic forms with identified neurogenic etiologies. In more recently published studies from four European countries as well as from the United States (10, 17, 18, 113, 114, 115), the proportion of female patients and idiopathic forms has clearly increased (Fig. 5). In a recently published American series, however, the sex ratio and etiological distribution were similar to the data from early studies despite the changing age limits for sexual precocity in the United States (116). It should be noted that this study reported on final height and involved patients diagnosed several years earlier and referred to the NIH with possible recruitment biases. The changing gender and etiological distribution of patients with precocious puberty in the United States could have resulted from the recently observed reduction in age at onset of puberty, before the age limits for sexual precocity were revised. Such an explanation, however, is unlikely in Europe where the physiological age at onset of puberty and the age limits for sexual precocity did not change during the past decade. An increased variability in timing of puberty without change in the mean ages could account for an increasing number of girls with idiopathic forms. Based on the diagnostic advances made during the past decades using nuclear magnetic resonance imaging, an increasing proportion of idiopathic forms would not have been expected. Another clue to the changing epidemiology of central precocious puberty could be the occurrence of new etiologies as suggested by the increasing proportion of migrating children who develop sexual precocity, as discussed below.

View larger version (33K): [in this window] [in a new window]   FIG. 5. Gender and etiological distribution of patients with central precocious puberty in the series published between 1961 and 2001 in relation to the year of publication and the country of study (DK, Denmark; F, France; UK, United Kingdom; NL, the Netherlands; B, Belgium). The number of patients and reference (in parentheses) are indicated at the top of each bar.

C. Children migrating from developing countries
After an initial report from Sweden in 1981 (118), sexual precocity has been reported in children migrating from developing countries to different European countries, primarily through international adoption (11, 12, 13, 14, 15, 16, 17, 18). So far, there is only one similar report from the United States published in abstract form (119). Some data from the published series are summarized in Table 2, and methodological aspects are listed in Table 3. In the original Swedish report and the subsequent Dutch, French, and American reports, cohorts of foreign adopted patients were studied, and the mean menarcheal age was evaluated through questionnaires sent out to the families (11, 13, 16, 119). In these studies, the absolute frequency of sexual precocity was undoubtedly increased (Table 3) but was variable among the studies between 13% (11) and 30% (119). In the Swedish cohort study, the menarcheal ages, although advanced, followed a relatively symmetrical pattern of distribution (11), such as expected in normal adolescents (120). This indicates that advanced puberty was a general feature of the whole cohort (Fig. 1D) instead of precocity being a feature of a particular subset (Fig. 1C). It is of note that the variability in menarcheal age (CV, 11.5–24.7%) in these cohorts (13, 16, 119) can be greater than in the nonmigrating girls living in well-off conditions. Several biases may have influenced the calculation of the frequency of sexual precocity. In the cohort studies, sexual precocity was defined as early menarche, which was reported in the questionnaire (11, 13, 16). Such a definition was different from the patient studies in which early breast development was the first criterion used to define precocity (14, 16, 17, 18). This distinction is important because thelarche and menarche may have a different significance (Table 1). India was the only country of origin in the Swedish study (11), whereas several different countries were involved in the other studies (13, 16, 119). The differences in country of origin were unlikely to bias the findings because, in three cohort studies, the average menarcheal age was advanced in comparison with data from the foster country and the country of origin as well (11, 13, 16). In the study by Oostdijk et al. (13), mean ages at menarche differed significantly, with the lowest observed in girls from India (11.2 yr) and the highest in girls adopted from South Korea (12.4 yr). In the Swedish study, birth date was uncertain in 28% of subjects, but similar data were obtained when the subsets with certain and uncertain birth date were compared (11). In a study on children adopted from China, 11% were thought to have an inaccurate birth date (121). Birth date uncertainty is a potential bias in all the studies, with possible differences among the countries of origin. This, however, is unlikely to account for false diagnosis of sexual precocity in most migrating children, although such a possibility should be kept in mind for individual patients. More important, the bias of parental concern about early sexual development of their adopted child may have influenced the decision to respond to the questionnaire, accounting for overrepresentation of early maturers in the responding families.

In the foreign children with sexual precocity, evidence of early hypothalamic-pituitary maturation and normal brain imaging has been provided, leading to the diagnosis of idiopathic central precocious puberty (14, 16, 17). In all the studies, sexual precocity was seen much more frequently in foreign girls than in boys (11, 12, 13, 14, 16, 17, 119). Such a sexual dimorphism might reflect the general and unexplained female preponderance of idiopathic central precocious puberty. However, a precise cause influencing specifically the female endocrine system cannot yet be excluded. The early hypotheses to explain sexual precocity in foreign adopted children after migration have incriminated the transition from an underprivileged to a privileged environment (11, 12). A new perspective has come from the recent Belgian report indicating that 12 patients from a total group of 145 were foreign children moving together with their original families without past history of deprivation (17). The observation of both nonadopted and adopted foreign children among the patients with precocious puberty was also found in Denmark (Table 2). The occurrence of precocious puberty in nonadopted migrating children suggests the possible role of factors related to migration and change of environment. Although the impact of former nutritional and emotional stress is less likely in children moving together with their original families, it is possible as well that such children are relieved from some stressful conditions after leaving their country.

The available data on actual or predicted adult height in migrating girls, including those with early or advanced puberty (11, 13, 14, 17), indicate that the average final height was close to or below the third centile of national references in the foster countries (Table 2). The interpretation of height data is difficult because of the possible role of ethnic differences, unknown short parental height, and intrauterine growth retardation (IUGR) in addition to precocious puberty. The observation of short adult stature prompted several clinicians to study the effects of combined GnRH agonist suppression of puberty and stimulation of growth with GH (125, 126). Pubertal growth and final height are obviously very important factors in relation to changes in pubertal timing and secular trends in the general population as well as in migrating children. These aspects, however, involve many factors beyond the control of pubertal timing and will not be discussed in the present paper.

	IV. Possible Mechanisms of Variations in Timing of Puberty around the World and after Migration

Top Abstract I. Introduction II. Normal Puberty and... III. Precocious Puberty IV. Possible Mechanisms of... V. Conclusion and Future... References

 
Puberty and reproduction are determined by changes in the secretion of the pituitary gonadotropins, LH and FSH, which are dependent on the frequency and amplitude of pulsatile GnRH neurosecretion from the hypothalamus (1, 9). The timing of puberty can be influenced by signals involving neurotransmitters and neuropeptides that originate in the hypothalamus, in addition to peripheral or gonadal signals. Signals linked to the environment such as nutrition, light, stressors, and endocrine disrupters might impinge on the hypothalamic signaling network directly or through peripheral signals (Fig. 6). In the following sections, we will discuss the possible contribution of hypothalamic, peripheral, and environmental signaling to the differences in timing of puberty around the world and the early activation of the hypothalamic-pituitary-gonadal system in migrating children. It is, however, virtually impossible to isolate the contribution of each signal because many of them are interrelated. Geographical differences might involve altitude, temperature, humidity, and lighting. An underprivileged life setting might involve nutritional problems, high energy expenditure, insufficient public health, individual diseases, large family size, and social and emotional injuries.

View larger version (28K): [in this window] [in a new window]   FIG. 6. Schematic illustration of the relative influences of hypothalamic, peripheral, and environmental signaling on the physiological variability in timing of puberty (A) or the occurrence of a subset with sexual precocity (B) or the shift of the whole study population toward early pubertal timing (C).

It is likely that a cascade of genes may determine variations in timing of pubertal onset. In the rat, manipulation of some homeobox-containing genes (OCT2, TTF-1) involved as regulators of downstream genes of neuropeptides stimulating GnRH secretion such as TGF, can result in alterations of the timing of puberty (141). We have not yet identified, however, the upstream genes that possibly mediate variations related to ethnic and familial patterns of development, as well as the striking sexual dimorphism in onset of puberty. Unraveling those genes will be critical because genetic control can account for most of the physiological variations in pubertal timing whereas peripheral and environmental signals could play relatively minor roles in that respect (Fig. 6A).

The ethnic or racial characteristics belong to the genetically determined factors and could contribute to the sexual precocity of foreign migrating children, in addition to environmental factors. In this respect, studies in migrating people provide an interesting model and put emphasis on environmental factors. In 1942, Ito (142) reported that, in Japanese girls born and reared in California, menarcheal age was more than 1.5 yr earlier than in Japanese girls born in California or Japan and reared in Japan. Although this study put emphasis on environmental factors, the dominant role of genetic factors was emphasized by others who found a similar menarcheal age in Turkish girls living in Germany in comparison with girls living in Turkey (143). Foreign adoption creates a unique situation with children from many different countries and races sharing a similar new setting, although many variables such as importance of former deprivation or insults and age at adoption still account for heterogeneity. In the Belgian experience, sexual precocity could not be related to any clear racial factor because the patients came from different countries all around the world and belonged to different races (17). The percentage of girls seen and treated for precocious puberty was relatively high (1.2%) among those coming from Colombia and India or Sri Lanka compared with other countries. There is a possible bias, however, because adoption from those countries has occurred for many years and at a relatively stable rate throughout the entire study period, whereas that from some other countries (Vietnam, China, Haiti) has started more recently and is still increasing. For instance, many patients with sexual precocity were seen from Haiti in the past few years but not formerly. In addition, differences among countries, such as absence of sexual precocity in girls adopted from Ethiopia or the Philippines, might involve racial and environmental influences, which will be difficult to isolate from each other. In Western Europe, we have relatively few children adopted from Eastern Europe, whereas these countries, particularly Russia and Romania, have provided a great number of children adopted in the United States (119, 123, 144) and among whom early puberty was also common (119, 123). Thus, although racial factors may play some role in the timing of puberty, such a role is relatively minor and unlikely to explain the variations seen in different countries around the world or in migrating children (89).

B. Intrauterine conditions
It has been proposed recently that the intrauterine milieu might influence physiological and pathological events occurring throughout life, although the mechanism of such a programming remains elusive (145). With respect to puberty and reproduction, low birth weight appears to be associated with precocious pubarche (adrenarche) and ovarian hyperandrogenism in the human female (146) and subfertility in the male (147). Evidence of central precocious puberty associated with IUGR has been provided in some patients with Russel-Silver syndrome (148) because, in a review of 148 observations with data on pubertal timing provided for 17 girls and eight boys, eight girls and one boy showed early puberty (149). More recently, precocious puberty was reported in some IUGR patients with maternal uniparental isodisomy of chromosome 14 (150, 151). It is unknown why some particular patients with this syndrome or Russel-Silver syndrome will enter puberty early. Some variations in menarcheal age can be related to low birth weight. In the United Kingdom, the age at menarche was 0.2 yr earlier in girls with birth weight below 2.85 kg compared with those weighing more than 3.75 kg (152). Among Spanish girls with early puberty (B2 between 8 and 9 yr), menarcheal age was 1 yr earlier in girls with a birth weight below 2.7 kg compared with the rest of the cohort (153). In a recent study from Israel, the mean menarcheal age of IUGR girls was found to be advanced by 1.3 yr vs. girls with birth weight appropriate for gestational age, due to increased prevalence of early puberty and reduced prevalence of delayed puberty (154). In a French study, however, IUGR was found to be associated with a pubertal delay averaging 0.8 yr in girls and 2.1 yr in boys, respectively (155). In the general population, some authors found no significant correlation between birth weight and menarcheal age (156), whereas others reported that thin newborns enter puberty earlier (157, 158). A sexual dimorphism in the relationship between birth weight and timing of puberty was observed in a limited group of 35 girls who showed pubertal age positively and significantly correlated with birth weight tertiles, whereas a trend toward a negative correlation was seen in 34 boys (159). The female predisposition to early onset of puberty in IUGR (149, 159) is consistent with the gender dimorphism seen in other etiological conditions. This suggests that factors linked with IUGR are superimposed to a more general mechanism making females prone to develop sexual precocity. Experimental data obtained in rats of both sexes, indicate that IUGR causes delayed puberty in both sexes, whereas early postnatal malnutrition does not affect timing of puberty in the female but causes a delay in the male (160). Thus, alterations in intrauterine conditions are possibly associated with disorders of puberty and reproduction, but their role in physiological variations in timing of puberty remains uncertain.

Could IUGR contribute to sexual precocity in adopted children migrating to Western Europe or the United States? In these children, the likelihood of increased incidence of IUGR is suggested by the Swedish cohort study in which 80% of 37 children with known birth weight were below 2.5 kg (161, 162). Unfortunately, information on birth weight and maternal risk factors for IUGR is lacking in the other studies. In those cases, we can only speculate until collaborative studies are set up between foster and home countries of adopted children. There is no evidence that, in the foreign adopted girls with precocious puberty, a putative IUGR can be linked with premature pubarche as observed in other female populations (146). By contrast, we reported that the development of pubic hair was less advanced in these patients compared with Belgian native girls with idiopathic or organic sexual precocity (163). This finding is consistent with the dissociation between adrenarche and gonadarche (21, 22). The delay in pubic hair development in migrating girls may involve particularly the subset of Indian girls, indicating a possible racial or geographical factor.

C. Nutrition
Among the factors linked with the living standards that account for the downward secular trend in timing of puberty and the differences between underprivileged and privileged settings, nutrition is likely to play a key role. This area appears to be most complex for the following reasons. 1) Nutrition involves both quantitative and qualitative aspects that have been rarely individualized in human studies. 2) Usually, the dietary status is assessed indirectly and incompletely through anthropometry. Weight and the related variables provide some information on storage of metabolic fuels, depending on individual differences in energy balance that may be genetically determined. 3) Reproduction has been more extensively studied than onset of puberty, whereas these two manifestations of pituitary-gonadal function can be differently affected by nutritional signaling. 4) The hypothalamic mechanism controlling food intake/energy balance and onset of puberty/reproduction involves common local and peripheral regulators, which are numerous and act through redundant pathways. 5) The contribution of differences in nutrition to disorders of puberty and reproduction does not mean that nutrition is a determinant of physiological variations in timing of puberty. In this section, we will concentrate on the possible involvement of nutrition in differences in timing of puberty around the world and in migrating children.

A direct relationship between body weight and the age at onset of puberty was suggested by Frisch and Revelle (164, 165) based on comparison between early and late maturers. Frisch et al. (166) concluded that a critical amount of body fat was needed for the onset of puberty. The maintenance of cycling in women has been estimated to require that at least 22% of body composition is fat (167, 168). In the amenorrheic running female monkey, ovulatory cycles are restored by increasing energy availability (169). However, the nutritional determinants of the ovulatory cycle can be different from the control of pubertal timing. The secular decline in menarcheal age occurs together with an opposite trend for regular cycling because, within a 25-yr period, the interval between menarche and regular cycling has increased from 1.9 to 3.0 yr and attains an interval of 5 yr in 21% of women instead of 9% formerly (170). These authors pointed out the possible influence of nutritional changes as well as differences in physical activities among female adolescents. Parallel to the delay in regular cycling, the prevalence of ovulatory disorders seems to increase, pointing to a possible environmental effect as well (171). Nutritional factors also play some role in polycystic ovary syndrome that can, however, predominantly involve genetic factors (172).

The Frisch and Revelle hypothesis has triggered a number of studies that confirmed (41, 156, 173, 174, 175) or did not confirm (28, 176, 177, 178) a significant relationship between menarcheal age and fat mass estimated through BMI or the sum of skinfold thickness or dual energy x-ray absorptiometry. It is debatable whether the Frisch and Revelle hypothesis could be relevant when only the physiological variations in body fatness are considered. As an example, girls with early menarche are more likely to be obese than those with late menarche (156), and, in comparison with nonobese girls, the average menarcheal age of obese girls was 9 months earlier in Japan (67) and 0.9 yr earlier in Thailand (179). However, the mechanisms involved in these pathological conditions may be different from those in normal subjects. Another difficult issue is the meaning of a significant correlation between fatness and menarcheal age. This may indicate a direct relation between fatness and menarche that can be either causal or consequential. Alternatively, the link between the two parameters can only be indirect because they share similar genetic determinants. In this respect, the recent study by Wang (180) is interesting because early sexual maturation is associated with an increased prevalence of fatness in girls and leanness in boys. Such a sexual dimorphism could involve genetic and/or endocrine factors. Several authors reported that early menarche was associated with an increased risk of obesity in adulthood (181, 182). Conversely, several studies suggested that childhood might be a critical period for weight to influence the timing of puberty because menarcheal age was inversely related to weight at 7 yr (152). Qing and Karlberg (183) reported that a gain in 1 U of BMI between 2 and 8 yr was associated with an advancement of age at the pubertal growth spurt reaching 0.6 yr in boys and 0.7 yr in girls. Davison et al. (184) reported that early onset of breast development by 9 yr could be weakly but significantly predicted by a higher percentage body fat at 5 and 7 yr. In this study, up to 14 and 35% of girls reached B2 stage at 7 and 9 yr, respectively, which was assessed, however, by visual inspection only. Kaprio et al. (3) suggested that the association between relative body weight and menarcheal age was primarily due to correlated genetic effects, whereas the two parameters were influenced by separate environmental correlates independent of each other. Karlberg (158) came to a similar conclusion about peak height velocity and menarche, which can occur simultaneously or at a time interval of 2 yr. They also emphasized the halt in secular trend in menarcheal age while height (and weight) are still increasing. It is tempting to conclude that the link between nutritional status and physiological variations in timing of puberty can be significant but is not particularly strong, suggesting that the relationship is indirect or partial and superseded by other factors.

Is there any possible link between the anthropometric data and the ethnic variations in pubertal timing? In some studies, only a minor advancement of 0.1–0.3 yr in menarcheal age was observed in African-American girls as compared with white American girls (35, 41). That advancement was found to be 0.4 yr in the NHANES III cohort (39) and reached 0.7 yr in the PROS study (6), which raised the question of whether racial differences in nutrition and weight for height play any role. Recently, Anderson et al. (40) found that the racial differences in menarcheal age were independent of differences in BMI. In another recent study, African-American girls were found to be fatter and sexually more mature than white American girls, a difference that was associated with increased levels of free bioavailable IGF-I (185). These data, however, are not comparable to the PROS study (6) because pubic hair, which was used as the marker of sexual maturation, may provide information on adrenarche but not gonadarche. There are additional ethnic differences, including higher insulin response to a glucose challenge among black vs. white subjects, lower resting energy expenditure, and perhaps less physical activity (186), without evidence, so far, of contribution to the racial differences in pubertal timing. Kaplowitz et al. (187) used the growth data from the PROS study (6) to calculate the SD scores (SDS) or Z scores of BMI. They found these scores to increase at the time of onset of breast development and concluded that the gain in BMI was predictive of onset of puberty and consistent with the racial difference in pubertal timing. Again, this conclusion raises the above debate about the causal or consequential nature of the observed correlation. In addition, there is a possible bias in these study conditions because BMI reference data from average maturers were used to calculate Z scores in early maturers. Likewise, in the recent report by Wang (180), the definition of overweight in early maturers using the BMI percentiles obtained in average maturers can involve a similar bias. If an increase in BMI occurs early as a possible consequence of early pubertal development, the reference data from average maturers provide inappropriately low values to calculate Z scores of BMI in early maturers. This may result in overestimation of BMI increase and can mistakenly lead to the conclusion that such an increase has a causal influence on pubertal onset.

Studies assessing directly the influence of dietary intake on the age at menarche have not shown any clear correlation between diets and timing of puberty. A weak but statistically significant correlation was found between early menarche and energy intake and expenditure (173). In contrast, high vs. low energy intake was associated with later menarcheal age in heavy Hispanic girls in California, which might have been caused by underreporting of the dietary intake in some groups (188). The protein source of food in early life could also influence the timing of puberty because a high animal vs. vegetable protein ratio at the ages of 3–5 yr is associated with early menarche, after controlling for body size (189). Phytoestrogens in the diet might have a role in the regulation of puberty both directly and indirectly (190). They interact with estrogen receptors and may have either agonistic or antagonistic effects, depending on the endogenous hormonal balance (191). In prepubertal boys who have very low endogenous estrogen levels, phytoestrogens might have some agonistic effects, whereas antagonistic effects would be expected when endogenous estrogen levels are higher. It was shown in the rat that the phytoestrogen coumestrol caused both reduced frequency of the electrophysiological hypothalamic discharges associated with pulsatile LH secretion as well as direct pituitary inhibition (192). In addition to their receptor effects, phytoestrogens affect the metabolism of hormones. They have been shown to inhibit aromatase and 17ß-hydroxysteroid dehydrogenase type 1 and type 5 enzymes (193, 194, 195, 196). Thus, the sum effect of phytoestrogens appears clearly antiestrogenic, which is also in accordance with animal experiments (197). A phytoestrogen-rich diet might therefore delay puberty as demonstrated by Berkey et al. (189). Such a conclusion may be relevant to the nutritional changes that can be involved in early puberty after migration because vegetable products such as soybean or cassava are major food components in developing countries but not in Western European societies (198).

Underfed children have delayed puberty (199), and, in many children adopted from foreign countries, malnutrition at arrival has been reported. The mean SDS of weight for height at arrival was –0.6 in the Swedish study (161, 162), and the weight deficit was –10.5% in the Italian study (14). In such conditions, a catch up in height and weight is commonly seen after arrival and precedes early sexual maturation. This has led to the hypothesis that hormonal events linked with catch-up growth may prime hypothalamic maturation and lead to puberty. This hypothalamic priming was thought to possibly occur in a critical age window because precocious menarche (<10 yr) was more common in Indian girls who arrived in Sweden between 3 and 6 yr than in those who arrived before 2 yr of age (11). The concept of a critical prepubertal period for nutritional priming of maturation is in agreement with observations in healthy boys and girls (152, 183, 184). Proos (200) reported that, in foreign adopted children, earlier menarche was correlated with late