American Journal of Epidemiology Vol. 154, No. 7 : 618-623
Copyright © 2001 by The Johns Hopkins University School of Hygiene and Public Health
ORIGINAL CONTRIBUTIONS |
Relation of Childhood Height and Later Risk of Breast Cancer
1 From the Division of Research, Kaiser Permanente, 3505 Broadway, Oakland CA 94611–5714 (e-mail: ljh{at}dor.kaiser.org)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The authors sought to examine the hypothesis that girls who were relatively tall during the prepubescent period (indicative of an affluent diet and good general health) were at increased risk of subsequent breast cancer. They conducted a case-control study of 214 long-term members who were diagnosed with breast cancer during 1973–1995 and who were age 12 years or younger when they first joined Kaiser Permanente and of 214 appropriately matched controls. Information was obtained from the medical records. While the authors observed the expected association of adult height with risk of breast cancer (height at age 15–18 years, tall-for-age vs. short-for-age: odds ratio = 2.2, 95% confidence interval: 1.1, 4.3), the association was no stronger earlier in life (height at age 9–11 years: odds ratio = 1.0, 95% confidence interval: 0.5, 1.8). The study does not support a relation between pubertal skeletal growth and adult risk of breast cancer. However, it is limited by the inclusion of few postmenopausal women.
anthropometry; breast neoplasms; nutrition; risk factors
Abbreviations: CI, confidence interval; OR, odds ratio
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The manyfold increase in breast cancer risk with migration from low-risk to high-risk countries and the several-generation period over which that increase takes place have led epidemiologists to search for an etiologic agent that is environmental in nature and changes slowly over generations (1
). Attempts to show that diet explains the international and secular pattern of breast cancer risk have given evidence of only a modest relation, if any (2). However, past diet is difficult to measure, and some have argued that, in any case, it is childhood diet that sets the stage for the higher risk observed in women who live in affluent countries (3–6). An affluent-type diet affects the hormonal profile and triggers early puberty, is thought to reduce the age at which the height potential is achieved, and increases the height potential overall (7). Kaiser Permanente, a health maintenance organization, has provided medical care in northern California since 1946. Some members who joined the health plan as children in the 1940s and 1950s were still enrolled in the plan, with or without interruptions in their membership 20, 30, and even 40 years later, when they were diagnosed with breast cancer. This afforded us the opportunity to use information on childhood height recorded prospectively in the medical record to examine the hypothesis that greater prepubescent height is associated with greater adult breast cancer risk.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study population
The cases included female Kaiser Permanente members diagnosed with epithelial breast cancer (International Classification of Diseases for Oncology site code C50, histology codes 802–857) (8
) during the period 1974–1995. They were ascertained through the Northern California Cancer Center, which has reported incident cancer cases occurring in the seven-county Bay Area to the Surveillance, Epidemiology, and End Results program since 1973, or through the Kaiser Permanente Regional Cancer Registry, which has reported incident cases diagnosed among members of the 23-county northern California service area since 1988 directly to the Northern California Cancer Registry in conjunction with state reporting requirements. In addition, cases must have first joined Kaiser Permanente when they were age 12 years or younger to ensure that some information on childhood height could be obtained. Interruptions in membership before and after age 12 years were permitted. Of the 391 cases initially identified, 177 were ineligible at chart review. Of these, 164 were older than age 12 years upon entry into the plan, with this large number resulting from our use of the medical record number as a proxy for year of entry into the health plan. We designed this approach to be sensitive but not specific. In addition, two ineligible cases had been treated but not diagnosed at Kaiser Permanente, three had a history of breast cancer before 1974, seven had inaccessible charts, and one had not really developed breast cancer.
The controls included women who were members of Kaiser Permanente during the period when incident cancer cases were ascertained (1973–1995) and, like the cases, first joined Kaiser Permanente when they were age 12 years or younger. One control was selected for each eligible case, and the controls were frequency matched to the cases on the following: year of birth (1934–1939, 1940–1944, 1945–1949, 1950–1954, and 1955–1962) and age at entry in Kaiser Permanente (ages 0–4, 5–8, and 9–12 years). In addition, controls were assigned a reference date at which they were active members of the health plan, so that the distribution of reference dates for the controls corresponded to the distribution of diagnosis dates for the cases. There were 214 eligible and 119 ineligible controls, with all of the latter being older than age 12 years at entry into the plan.
Data collection
Information on height, weight, demographic characteristics, menstrual and reproductive characteristics, and benign breast disease was obtained from the medical record by a single trained medical record analyst. The medical record was reviewed from the first chart date to the diagnosis or reference date. However, a family history of cancer was not ascertained if the note was recorded within the 6 months preceding the reference date because the case's chart might have more complete information on family history once breast cancer was suspected (even though it had not yet been diagnosed). Every measurement of height and weight recorded in the medical record through age 24 years was ascertained. We sought to obtain the heights of the mother and father, but the father's height was missing for 106 of the cases.
Data analysis
For each subject, we computed height and weight at each year of age as well as the age of the measurement to one-tenth of a year (e.g., 8.5 years). If a subject had multiple measurements at age 8 years, for example, one at age 8.2 years and one at age 8.6 years, we computed the mean of the measurements as well as the mean of the ages.
The body mass index was computed as weight (kg)/height (m2). The Pearson correlation coefficients of age-specific height with age at menarche, adult height, and mean parental height were estimated. The growth curves of the cases and controls were plotted and visually inspected. Age-specific height and body mass index were coded into tertiles by using cutpoints based on the distribution in the controls. There were relatively large numbers of subjects without a measurement in any particular year of age, and we put these subjects into a separate exposure category, "missing." In additional analyses, we increased the sample size in each category by grouping girls who had at least one height measurement during the age intervals 3–5, 6–8...15–18 years. For example, if a girl had been measured only at age 4 years, her height at that age was compared with the tertiles of height among the controls who were aged 4 years at measurement, and she was categorized as short, medium, or tall. A girl who was measured only at age 3 years was compared with the tertiles of height among controls who were age 3 years at measurement. Thus, a girl categorized as short in the age group 3–5 years could have been in the lower tertile at age 3, 4, or 5 years.
The distributions of height and body mass index, both as continuous variables and as tertiles (with missing values being treated as an additional category), were compared between cases and controls in contingency tables and in logistic regression analysis using the Statistical Analysis System (9). Maximum likelihood methods were used to estimate the odds ratios for the associations of height and body mass index with breast cancer risk at each year of age after adjustment for confounding factors, and the 95 percent confidence intervals were calculated using standard errors.
The following were examined as potential confounding factors through adjustment: the matching variables birth year and age at entry, and marital status, alcohol use, race, parity, age at first birth, and menopausal status. Maternal breast cancer history was examined as a confounding factor by excluding women who reported a family history. None of these variables confounded the relations. In addition, we sought to examine whether age-specific height in early adolescence was associated with breast cancer risk strictly through an effect on age at menarche and whether deviation from mother's height, a partial measure of genetic height potential, was associated with breast cancer risk. We attempted to examine the association of height 1 or 2 years prior to the onset of menses with risk, but the available sample size was too small (because of missing values for height during this brief period) to permit meaningful interpretation of the results. We also sought to examine the relation separately in pre- and postmenopausal women, but had an adequate sample to examine premenopausal women only.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Because of the eligibility requirements concerning age at entry and year of entry, the subjects were relatively young at diagnosis (table 1), with 83 percent being aged 45 years or younger. More than half of the cases were diagnosed in the 1990s. Three-quarters of the cases and controls were White, 30 percent were nulliparous on their reference date, and the great majority had not reached menopause. The mean age at menarche was 12.3 years for the cases and 12.4 years for the controls. In addition, for many variables, there were differences in the amount of missing information between cases and controls. Ten percent of the subjects did not have a single height measurement recorded in the chart, 15 percent had only one, 25 percent had two or three measurements, and 50 percent had four or more measurements during the age period 3–17 years.
|
The number of cases and controls with a height measurement at each age is shown in table 2. Of the 214 cases and 214 controls in the study, 84 cases and 76 controls had a measurement of height at age 13 years. The pattern of missing height data was comparable between the cases and controls. In both cases and controls with a height measurement, the mean age of measurement was invariably the midpoint of the year, indicating the random distribution of medical visits during the year after the birthday. Thus, there was no difference in age between cases and controls to the first decimal place (table 2).
|
In the first decade of life, there was little consistent evidence for an association of height with risk of breast cancer (table 3). In the age group 12–14 years, girls who were relatively tall for their age were at 1.7-fold greater risk of breast cancer (95 percent confidence interval (CI): 1.1, 2.8), while in the age group 15–18 years, the increased risk was 2.2-fold (95 percent CI: 1.1, 4.3). The breast cancer risk of subjects without a height measurement was not notably different from the risk of subjects with one. We also included age at menarche in the model to determine whether there was an effect of childhood height on breast cancer risk above and beyond the association of age at menarche with breast cancer risk, but we observed no meaningful difference in the odds ratios (ages 9–11 years compared with short girls: medium height, odds ratio (OR) = 0.9, 95 percent CI: 0.5, 1.7; tall, OR = 1.0, 95 percent CI: 0.5, 1.8; missing, OR = 0.9, 95 percent CI: 0.5, 1.5).
|
Adjustment for maternal height did not noticeably affect these findings. Body mass index also was not associated with subsequent risk of breast cancer (age 10 years: OR = 0.9, 95 percent CI: 0.7, 1.1) (results at other ages not shown), nor did the odds ratio differ when the analysis was restricted to premenopausal women.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The strength of our study lies in our use of prospectively recorded data on childhood height and weight, age at menarche, and parental height for 214 breast cancer cases and 214 controls.
However, the study has several limitations. We observed a case-control difference in the level of missing information on covariates. This difference was particularly noticeable for variables measuring events in adulthood, such as marital status, alcohol use, and menopausal status. We believe through indirect evidence that the case-control difference in the level of missing information can be attributed to earlier identification of breast cancer among women who are relatively heavy utilizers of the health plan and who would have undergone more frequent breast cancer screening. These women would also have had greater opportunity to provide information on their covariate levels. Evidence for this comes from the comparison of the proportion of ductal carcinoma in situ in our sample (20 percent) relative to the proportion in the entire health plan (15 percent) during the period 1974– 1995 (10).
The growth characteristics of our sample are roughly similar to a representative sample of children residing in Berkeley, California, as described by Thissen et al. (11) in the Berkeley Growth Study. Growth characteristically has prepubertal and adolescent components, each of which is driven by genetic and environmental factors, the primary factors being nutrition and disease (12). Adult height is largely driven by prepubescent growth (13). Early nutrition influences the balance of hormones (both growth hormones and sex hormones) important in growth and development, and these hormones drive prepubertal growth, the timing of the adolescent growth spurt, and the onset of puberty (14–16). Indeed, an affluent diet drives rapid growth and early puberty, and age-specific height can be taken as a monitor of nutritional status (17).
It has been hypothesized that childhood nutrition increases breast cancer risk by changing the hormonal profile. Thus, the increasing incidence rates of breast cancer observed in second- and third-generation migrants from Asia to the United States, as well as the secular trend in breast cancer risk among Asian residents, are postulated to result from improved childhood nutrition (18).
We observed little evidence for an association of childhood height with risk of breast cancer beyond that expected based on the relation of childhood height with adult height and the known relation of adult height with risk of breast cancer (19–21). The strongest association of age-specific height with breast cancer risk did not occur with prepubertal height, as we had hypothesized, but rather with height in the latest years of childhood (15–18 years).
Only one other study has ascertained childhood exposures prospectively. Le Marchand et al. (22) observed that Hawaiian girls of Japanese descent aged 10–14 years born from 1918 to 1943 and at the highest tertile of body mass index were at 70 percent (p < 0.05) lower risk of premenopausal breast cancer relative to girls in the lowest tertile. They observed no important associations with height. Li et al. (23), using recalled height in a study of 747 cases, observed that women who reached their adult height at age 18 years or older were at 30 percent (95 percent CI: 0.5, 1.0) lower risk than women who had reached their adult height at age 13 years or younger. We were unable to confirm their findings because of our limited sample size. In the Nurses' Health Study, peak height velocity in the highest quintile relative to the lowest, estimated from recalled age at menarche, recalled body fatness at 10 years of age, and adult stature, was associated with a 30 percent increased risk of breast cancer (24).
Three studies have observed associations of premenopausal breast cancer with recalled adolescent body mass index (25–27), but the results have conflicted, with one indicating a direct association, the second indicating an inverse association, and the third indicating a U-shaped association. In addition, two studies have reported associations with recalled adolescent diet (27, 28).
Our main result--that prepubertal height was not associated with breast cancer risk beyond its association with adult height--is not adequate to refute the hypothesis that nutritional status early in life affects the risk of breast cancer. Because the girls included in our study were born to relative prosperity in northern California during 1934–1963, there may have been too little diet-related variation in age- specific height in our population. Thus, the coefficient of variation for height at the age of peak height velocity (10 years) was only 5 percent. Age-specific height shows greater variation across time and internationally, so we cannot rule out that nutritional status, as measured by age-specific height, would correlate with the international pattern in breast cancer risk.
|
NOTES |
|---|
(Correspondence to Dr. Herrinton at this address).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
- Stanford JL, Herrinton LJ, Schwartz SM, et al. Breast cancer incidence in Asian migrants to the United States and their descendants. Epidemiology 1995;6:181–3.[Web of Science][Medline]
- Hunter DJ, Willett WC. Nutrition and breast cancer. Cancer Causes Control 1996;7:56–68.[Web of Science][Medline]
- Apter D. Hormonal events during female puberty in relation to breast cancer risk. Eur J Cancer Prev 1996;5:476–82.[Web of Science][Medline]
- de Waard F, Trichopoulos D. A unifying concept of the aetiology of breast cancer. Int J Cancer 1988;41:666–9.[Web of Science][Medline]
- Tretli S, Gaard M. Lifestyle changes during adolescence and risk of breast cancer: an ecologic study of the effect of World War II in Norway. Cancer Causes Control 1996;7:507–12.[Web of Science][Medline]
- Stoll BA. Western diet, early puberty, and breast cancer risk. Breast Cancer Res Treat 1998;49:187–93.[Web of Science][Medline]
- Micozzi MS. Functional consequences from varying patterns of growth and maturation during adolescence. Horm Res 1993;39(suppl 3):49–58.
- Percy C, Van Holten V, Muir C, eds. International classification of diseases for oncology. Geneva, Switzerland: World Health Organization, 1990.
- SAS Institute, Inc. SAS/STAT users guide. Version 6, fourth ed. Vol 2. Cary NC: SAS Institute, Inc, 1990.
-
Husson G, Herrinton LJ. How accurately does the medical record capture maternal history of cancer? Cancer Epidemiol Biomarkers Prev 2000;9:765–8.
[Abstract/Free Full Text] - Thissen D, Bock RD, Wainer H, et al. Individual growth in stature: a comparison of four growth studies in the U.S.A. Ann Hum Biol 1976;3:529–42.[Web of Science][Medline]
- Karlberg J, Fryer JG, Engstrom I, et al. Analysis of linear growth using a mathematical model. II. From 3 to 21 years of age. Acta Paediatr Scand Suppl 1987;337:12–29.[Medline]
- Proos LA. Anthropometry in adolescence--secular trends, adoption, ethnic and environmental differences. Horm Res 1993;39(suppl 3):18–24.
- Caufriez A. The pubertal spurt: effects of sex steroids on growth hormone and insulin-like growth factor I. Eur J Obstet Gynecol Reprod Biol 1997;71:215–17.[Web of Science][Medline]
-
Merimee TJ, Russell B, Quinn S, et al. Hormone and receptor studies: relationship to linear growth in childhood and puberty. J Clin Endocrinol Metab 1991;73:1031–7.
[Abstract/Free Full Text] - Stoll BA, Vatten LJ, Svinnsland S. Does early physical maturity influence breast cancer risk? Acta Oncol 1994;33:171–6.[Web of Science][Medline]
- Tanner JM. Growth as a monitor of nutritional status. Proc Nutr Soc 1976;35:315–22.[Web of Science][Medline]
- Murata M, Hibi I. Nutrition and the secular trend of growth. Horm Res 1992;38(suppl 1):89–96.
- Baanders AN, de Waard F. Breast cancer in Europe: the importance of factors operating at an early age. Eur J Cancer Prev 1992;1:285–91.[Medline]
- Brinton LA, Swanson CA. Height and weight at various ages and risk of breast cancer. Ann Epidemiol 1992;2:597–609.[Medline]
-
Swanson CA, Coates RJ, Schoenberg JB, et al. Body size and breast cancer risk among women under age 45 years. Am J Epidemiol 1996;143:698–706.
[Abstract/Free Full Text] -
Le Marchand L, Kolonel LN, Earle ME, et al. Body size at different periods of life and breast cancer risk. Am J Epidemiol 1988;128:137–52.
[Abstract/Free Full Text] - Li CI, Malone KE, White E, et al. Age when maximum height is reached as a risk factors for breast cancer among young U.S. women. Epidemiology 1997;8:559–65.[Web of Science][Medline]
- Berkey CS, Frazier AL, Gardner JD, et al. Adolescence and breast carcinoma risk. Cancer 1999;85:2400–9.[Web of Science][Medline]
- Barnes-Josiah D, Potter JD, Sellers TA, et al. Early body size and subsequent weight gain as predictors of breast cancer incidence. Cancer Causes Control 1995;6:112–18.[Web of Science][Medline]
- Coates RJ, Uhler RJ, Hall HI, et al. Risk of breast cancer in young women in relation to body size and weight gain in adolescence and early adulthood. Br J Cancer 1999;81:167–74.[Web of Science][Medline]
-
Pryor M, Slattery ML, Robison LM, et al. Adolescent diet and breast cancer in Utah. Cancer Res 1989;49:2161–7.
[Abstract/Free Full Text] -
Potischman N, Weiss HA, Swanson CA, et al. Diet during adolescence and risk of breast cancer among young women. J Natl Cancer Inst 1998;90:226–33.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Ahlgren, M. Melbye, J. Wohlfahrt, and T. I. A. Sorensen Growth Patterns and the Risk of Breast Cancer in Women N. Engl. J. Med., October 14, 2004; 351(16): 1619 - 1626. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. L. De Stavola, I. dos Santos Silva, V. McCormack, R. J. Hardy, D. J. Kuh, and M. E. J. Wadsworth Childhood Growth and Breast Cancer Am. J. Epidemiol., April 1, 2004; 159(7): 671 - 682. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||