Menarche is one of the longest known and well-established early-life risk factors for breast cancer [1]. However, menarche is the final milestone of a complex, integrated series of biological events - the pubertal transition. This includes adrenarche (onset of substantial adrenal androgen production), thelarche (onset of breast development), pubarche (onset of pubic hair growth), and finally menarche. The timing (or age) at each of these milestones, and the intervals between them (the tempo of the transition) (Figure 1), are also relevant to breast cancer risk. In particular, an earlier age at thelarche and a longer interval between thelarche and menarche have recently been associated with an increased risk in breast cancer [2]. It may be that molecular changes due to hormonal or environmental factors during the development of the mammary gland (thelarche) itself may be the operative mechanism beneath the menarche/ breast cancer association.

There have been numerous studies of international differences in age at menarche [3], but no study has compared the age and tempo of puberty (including adrenarche) among populations at varying risks of breast cancer. Breast cancer incidence in Bangladesh is five times lower than in the UK [4]. However, South Asians in Britain have a rate double that in South Asia [5]. We designed the ABBY Project (Adolescence among Bangladeshi and British Youth) [6], a migrant study of Bangladeshi girls in the UK, to test whether the growth environment (Bangladesh versus UK) and/or ethnicity (Bangladeshi versus white) were associated with the age and tempo of pubertal milestones. The environment in the UK differs greatly from that in Bangladesh, specifically with regard to nutrition, obesity, and exposure to infectious diseases and/or endocrine disrupting chemicals [7], all of which are likely to affect the age and tempo of pubertal milestones either independently or in concert with each other.

We compared the ages of adrenarche, thelarche, pubarche and menarche across groups of girls who differ by growth environment (Bangladesh versus UK) and ethnicity (Bangladeshi versus White). British-Bangladeshi girls begin and progress through puberty earlier and slower than girls living in Bangladesh, but not as early or slowly as white British girls. The differences in pubertal development were partly, but not entirely, explained by higher BMI among white British and British-Bangladeshi girls versus Bangladeshi girls, suggesting that nutritional and other environmental factors associated with the UK growth environment are associated with early, but prolonged, pubertal development.

We are the first to compare adrenarche (hormonally defined) between European and Asian populations, and the ages of adrenarche determined in this study are also consistent with other studies that report adrenarche to occur between ages 6 to 8 years in healthy children [10],[18],[19]. Contrary to expectations, we found no difference in age at menarche among Bangladeshi, British-Bangladeshi or white British girls. The median ages at menarche in our study were consistent with the current ages in national studies [20], which, in Western Europe, have remained between ages 12 to 13 years since the 1960s [21]. Previously, however, there was a secular decline in age at menarche in the early 20^th century [3],[22] that is largely attributed to concurrent improvements in general health, nutrition and living conditions [21],[23],[24]. Within Bangladesh, girls with higher socioeconomic status (SES) reach menarche earlier than girls with lower SES [25]. Since the Bangladeshi girls were recruited from middle class families with the means to migrate, their SES may explain the similarities in age at menarche between them and white British girls.

Our findings support secular observations of the age at thelarche continuing to decline [21]. The median ages at thelarche for each study population, even the Bangladeshi girls, were markedly earlier than the mean age at breast stage 2 (11 years) first reported for British girls in the classic Marshall and Tanner study [9]. Our findings are consistent with more recent studies that show breast development occurring between ages 9 to 11 years [26]–[32]. The white British girls in our study, however, reached thelarche particularly early compared with white American girls, but no earlier than black American girls [32],[33]. This discrepancy may be due to differences in study design, but high rates of overweight and obesity in our white population may also explain the particularly early age at thelarche.

The age at pubertal onset is strongly influenced by environmental factors. Increased body fat, possibly mediated by leptin, has been previously linked with an earlier pubertal onset [32],[34]. BMI accounted for about 80% of the variation between Bangladeshi and white British girls, suggesting that differences in dietary composition and quality [35]–[36] between ethnicities may alter the age at pubertal onset. BMI accounted for 55% of the difference in age at thelarche between Bangladeshi and British-Bangladeshi girls suggesting that other factors associated with the UK environment, such as less infectious disease or increased exposure to oestrogenic substances, are also driving the earlier pubertal onset in British-Bangladeshi girls. BMI was not independently associated with pubarche in our data, and variation in pubarche according to residency scale was not explained by BMI, suggesting that other environmental factors are implicated.

This study supports a model whereby most sexual development starts with adrenarche and then proceeds from thelarche to pubarche to menarche. In contrast, other studies have previously distinguished girls that progress through puberty along the thelarche pathway from those that progress through an adrenarche pathway [37]. Our study suggests that adrenarche precedes the physical manifestation of both thelarche and pubarche. The intervals between the age at adrenarche and the age at thelarche or pubarche may reflect underlying differences in the source of androgens and oestrogen production during sexual development.

A slower pubertal tempo, but faster juvenile tempo, was associated with increasing residency in the UK growth environment in our data. An earlier onset of thelarche, but slower progression through puberty, has also been observed over time within the US and Western Europe [23],[38]. Girls growing up in an industrialized growth environment could either be progressing through puberty at a steady, but slower tempo than before or pubertal development in industrialized countries has become more punctuated, meaning that some pubertal milestones appear abruptly and are disjointed from subsequent ones. Punctuated puberty may not reflect an activated hypothalamic-pituitary-ovarian axis, but rather peripheral production of oestrogens from adipose tissue [39] or exposure to endocrine disrupting chemicals [40]. This earlier and prolonged exposure to oestrogenic substances, regardless of source, may be implicated in the increasing rates of breast cancer observed in South Asian populations living in the UK [41],[42]. Alternatively, androgen exposure occurring between the onset of adrenarche and thelarche could be protective for the developing breast. The reduction in juvenile tempo with increasing residency in the UK may explain increasing breast cancer rates in South Asian migrants to the UK [41].

There are limitations that need to be considered when interpreting the findings reported here. First, the comparison of age at menarche across groups was limited by the small sample size in the older white British girls, which resulted in some instability in estimates. However, the consistency in median ages between status quo and recalled methods, suggests that our approach is robust. Secondly, while our cross-sectional findings should be reproduced in longitudinal studies, the migrant study design enabled us to compare migrants of the same ethnicity, in terms of two generations, with girls living in both the “home” and “host” country. Finally, we assessed pubertal staging using self-reported PDS because clinical examinations are not a viable method in a school setting. Although there can be low concordance between findings from physical examination and those from the PDS [43], another study comparing the PDS with both a physical examination and hormone measures found that self-reported breast development and menarche were more reliably related to hormones than the physical examination [44]. Indeed, in our previous study, DHEAS and urinary oestrogen concentrations were higher among girls that reported having secondary sex characteristics compared with girls who did not [6]. Level of BMI can lead to a systematic bias in self-reporting of pubertal staging [44], so it is possible that girls with higher BMI mistook lipomastia for the development of breasts. However, repeating the survival analyses using the progressive stage of breast stage 3 as the end point (as Euling et al. (2008) suggest) did not change the overall observed pattern of results. It has also been demonstrated that white adolescents overestimate their pubertal stage more often than non-white adolescents; however within the same Bangladeshi ethnic group, we observed earlier breast development in those living in the UK compared with those living in Bangladesh. Interestingly, there are cultural markers of thelarche within both Bangladeshi populations [6] so it is unlikely that one would overestimate breast development more than the other.

Our findings suggest that the growth environment might account for the ethnic differences in pubertal age and tempo seen across and within international populations. Bangladeshi girls who grew up in the UK progressed through puberty earlier, but more slowly, than girls who grew up in Sylhet, Bangladesh. BMI explained some, but not all, of these differences in pubertal age. Other environmental factors such as psychosocial stress and endocrine disrupting chemicals may also affect the age and tempo of pubertal milestones. When the tempo between adrenarche, thelarche and menarche are considered together and compared across Bangladeshi, migrant British-Bangladeshi, and white British girls, a clear and provocative pattern emerges: there is a longer period of androgen exposure in populations at low risk, equal periods of androgen and oestrogen exposure in populations at intermediate risk and a prolonged period of oestrogen exposure in populations at high risk. These findings support the inclusion of age of thelarche, a potentially new risk factor, in future breast cancer studies.