Letters to the Editor

Letters to the Editor Body mass index and body fat In the recent article by Guo et al (1) and Bray s recent editorial (2), little reference to or discussion of body fat is made. In the Thousand Families Cohort Study (3), body mass index (BMI) at 9 y of age correlated significantly with BMI at 50 y of age but not with the percentage of body fat estimated from bioelectrical impedance measurements. Only children who were identified as being obese (BMI > 90th percentile) at 13 y of age had an increased risk of obesity (percentage body fat) as adults. I think that Guo et al could have expanded the discussion portion of their article and cited a few references dealing with the association of BMI and excess body fat with the health risks related thereto. Owen & Owen Ltd 24216 North 82nd Place Scottsdale, AZ 85255-2811 George M Owen 1. Guo SS, Wu W, Chumlea CC, Roche AF. Predicting overweight and obesity in adulthood from body mass index values in childhood and adolescence. Am J Clin Nutr 2002;76:653 8. 2. Bray GA. Predicting obesity in adults from childhood and adolescent weight. Am J Clin Nutr 2002;76:497 8. 3. Wright CM, Parker L, Lamont D, Craft AW. Implications of childhood obesity for adult health: findings from thousand families cohort study. BMJ 2001;323:1280 4. Reply to GM Owen We appreciate the informative comments by Owen regarding our recently published article (1). The body mass index (BMI; in kg/m 2 ) is very useful in identifying persons at risk of overweight, obesity, and subsequent morbidity and mortality. As a result of this utility and the increased incidence of overweight and obesity among children (2), BMI was included in the Centers for Disease Control and Prevention s revised growth charts for US children (3). Because of the large collections of stature and weight data in many surveys, BMI is the variable most often selected for analysis. This is especially true because of the limited availability of measured body-composition values in large national surveys. In our study (1), we used data from participants in the Fels Longitudinal Study from as early as the 1950s and were thus limited to measures of stature, weight, and BMI. The point raised by Owen regarding the association of BMI with total and percentage body fat is important. There is a considerable lack of data relating BMI to body-composition measures in children, adults, and the elderly. We previously reported associations between BMI and total and percentage body fat from more recent underwater weighing data from participants in the Fels Longitudinal Study (4). Additional similar information from other investigators would go a long way in clarifying the use of BMI in quantifying and qualifying the relations between overweight and obesity and measures of total and percentage body fat and the associated health risks. Note that the relation between BMI and percentage body fat in the study (5) referenced by Owen had used estimates of body fatness derived from bioelectrical impedance analysis, which has a high error rate (6). Our recent study (1) is an update of similar analyses, which were presented earlier (7). The focus of our more recent study (1) was on the relations between BMI values during childhood (derived from the Centers for Disease Control and Prevention s revised BMI growth charts) and the risk of high BMIs (> 25 and > 30) in adulthood. Our results indicate that BMI values during childhood can predict the risk of overweight and obesity in adulthood. The identification of these risk factors is important so that children at risk of becoming overweight or obese can be treated to prevent these conditions in adulthood. Because this was our focus, we did not consider the association of BMI with body fatness to be relevant within the framework of our study, especially because no body-composition data were included in the analysis. Clearly, more information is needed on the relation between indexes such as BMI and direct measures of body composition at all ages so that associations between these variables can be interpreted clearly. Department of Community Health Wright State University School of Medicine Dayton, OH E-mail: shumei.sun@wright.edu Shumei S Sun Wei Wu William Cameron Chumlea Alex F Roche 1. Guo SS, Wu W, Chumlea WC, Roche AF. Predicting overweight and obesity in adulthood from body mass index values in childhood and adolescence. Am J Clin Nutr 2002;76:653 8. 2. Troiano RP, Flegal KM, Kuczmarski RJ, Campbell SM, Johnson CL. Overweight prevalence and trends for children and adolescents: the 348 Am J Clin Nutr 2003;78:348 52. Printed in USA. 2003 American Society for Clinical Nutrition

LETTERS TO THE EDITOR 349 National Health and Nutrition Examination Surveys, 1963 to 1991. Arch Pediatr Adolesc Med 1995;149:1085 91. 3. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, et al. CDC growth charts: United States. Advance Data 2000;314:1 28. 4. Roche AF. Growth, maturation and body composition: the Fels Longitudinal Study 1929 1991. Cambridge, United Kingdom: Cambridge University Press, 1992. 5. Wright CM, Parker L, Lamont D, Craft AW. Implications of childhood obesity for adult health: findings from thousand families cohort study. BMJ 2001;323:1280 4. 6. Guo SS, Chumlea WC. Statistical methods for the development and testing of predictive equations. In: Roche AF, Heymsfield SB, Lohman TG, eds. Human body composition: methods and findings. Champaign, IL: Human Kinetic Press, 1996:191 202. 7. Guo SS, Chumlea WC, Roche AF, Gardner JD, Siervogel RM. The predictive value of childhood body mass index values for overweight at age 35 years. Am J Clin Nutr 1994;59:810 9. Alternate Healthy Eating Index The use of the ratio of polyunsaturated to saturated fatty acids in the Alternate Healthy Eating Index of McCullough et al (1) seems surprising. In an international observational study, Keys et al (2) reported that death rates were unrelated to dietary energy percentage from polyunsaturated fatty acids but strikingly related to the ratio of monounsaturated to saturated fatty acids. The Lyon Diet Heart Study (3) reported that ratios of polyunsaturated to saturated fatty acids at 1 4 y of follow-up did not differ significantly between the control subjects and the experimental subjects (0.69 and 0.65, respectively). In contrast, the concentrations of n 3 and n 6 fatty acids in the experimental subjects were 3 times and about two-thirds, respectively, those in the control subjects. In addition, the concentrations of saturated fatty acids and monounsaturated fatty acids in the experimental subjects were 29% lower and 25% higher, respectively, than those in the control subjects. The most striking comparison was for the ratio of n 6 ton 3 fatty acids (19.6 for the control subjects compared with 4.4 for the experimental subjects). Other studies, most recently that of Albert et al (4), have pointed strongly to the merits of increasing the intake of n 3 fatty acids, at least for populations who consume a Western diet. Thus, McCullough et al may have substantially underestimated the health, survival, and medical-cost benefits of prudent eating. 98 Clinton Avenue Montclair, NJ 07042 E-mail: granddaddy3@msn.com Jetson E Lincoln 1. McCullough ML, Feskanich D, Stampfer MJ, et al. Diet quality and major chronic disease risk in men and women: moving toward improved dietary guidance. Am J Clin Nutr 2002;76:1261 71. 2. Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year death rate in the Seven Countries Study. Am J Epidemiol 1986;124:903 15. 3. de Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean alphalinolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet 1994;343:1454 9. 4. Albert CM, Campos H, Stampfer MJ, et al. Blood levels of long-chain n 3 fatty acids and the risk of sudden death. N Engl J Med 2002; 346:1113 8. Reply to JE Lincoln In his letter, Lincoln suggests that the Alternate Healthy Eating Index (AHEI) (1) would have more strongly predicted chronic disease risk had more specific attention been paid to fatty acids, namely, monounsaturated fatty acids and n 3 polyunsaturated fatty acids. In designing the index, we included components with well-established relations to disease, including trans fatty acids and the ratio of polyunsaturated to saturated fatty acids, and gave credit for the consumption of fish and nuts (both rich sources of n 3 fatty acids). The intakes of both n 3 and n 6 polyunsaturated fatty acids were substantially (P < 0.001) higher in the fifth quintile of the AHEI than in the first quintile (Tables 1 and 2). Ecologic studies, such as the study by Keys et al (2) that is cited by Lincoln, are useful in generating hypotheses but are not confirmatory, because of the great potential for uncontrolled confounding (3). Nevertheless, monounsaturated fatty acids have been found to have beneficial effects on both blood lipids (4) and disease risk (5). Currently, monounsaturated fat intake in the United States is mainly from beef, margarine, and baked goods (6) rather than from olive oil, which is still not commonly used in the United States. We avoided overemphasizing these foods because they contain other potentially adverse components. Specific polyunsaturated fatty acids have different metabolic effects (7), but both n 3 and n 6 fatty acids have protective associations with blood lipids and cardiovascular disease risk (5, 8). In the Lyon Diet Heart Study (9), subjects were given defined diets to test specific hypotheses. Therefore, the concentrations of fatty acids in their blood after the intervention reflected the diet and the hypotheses being tested. The experimental subjects had significantly higher oleic acid concentrations and significantly TABLE 1 Unsaturated fatty acid intakes by quintile (Q) of Alternate Healthy Eating Index score in 38615 men who participated in the Health Professionals Follow-up Study 1 Q1 Q2 Q3 Q4 Q5 P:S 0.3 0.4 0.5 0.5 0.7 PUFAs (g/d) 10.2 12.0 13.3 14.5 16.0 MUFAs (g/d) 25.7 27.1 28.1 28.6 27.7 Linoleic acid (g/d) 8.8 10.4 11.5 12.6 13.8 Linolenic acid (g/d) 0.9 1.1 1.1 1.2 1.3 1 P:S, polyunsaturated-to-saturated fatty acid ratio; PUFAs, polyunsaturated fatty acids; MUFAs, monounsaturated fatty acids. For all of the variables, P for trend < 0.001.

350 LETTERS TO THE EDITOR TABLE 2 Unsaturated fatty acid intakes by quintile (Q) of Alternate Healthy Eating Index score in 67271 women who participated in the Nurses Health Study 1 Q1 Q2 Q3 Q4 Q5 P:S 0.4 0.4 0.4 0.5 0.6 PUFAs (g/d) 10.5 11.9 12.8 13.7 15.4 MUFAs (g/d) 23.1 24.3 24.8 25.3 25.7 Linoleic acid (g/d) 9.1 10.3 11.1 11.8 13.3 Linolenic acid (g/d) 1.0 1.1 1.2 1.2 1.3 1 P:S, polyunsaturated-to-saturated fatty acid ratio; PUFAs, polyunsaturated fatty acids; MUFAs, monounsaturated fatty acids. For all of the variables, P for trend < 0.001. lower linoleic acid concentrations than did the control subjects. The experimental subjects also had significantly higher concentrations of linolenic and eicosapentaenoic acids than did the control subjects. Because the study was not designed to test the n 6 hypothesis, linoleic acid concentrations (and thus the ratio of polyunsaturated to saturated fatty acids) were not expected to differ between the experimental and control groups. In the freeliving populations that we studied, the ratio of polyunsaturated to saturated fatty acids captured healthy trends in ingested lipids and was strongly associated with the risk of ischemic heart disease (10). Inclusion of -linolenic acid in the AHEI score might have improved our prediction of cardiovascular disease, but again, we chose to include only those factors with longer, more established relations. As we learn more about the relation between diet and health, scores such as the AHEI can continue to be refined and improved, and diet patterns being recommended to the US public can become more precise. For now, advice that emphasizes the intake of unsaturated fats and the restriction of saturated and trans fats and that encourages the consumption of fish, nuts, and whole grains clearly represents an improvement in recommendations to reduce chronic disease risk. Epidemiology and Surveillance Research American Cancer Society 1599 Clifton Road, NE Atlanta, GA 30329 E-mail: marji.mccullough@cancer.org Department of Nutrition Harvard School of Public Health 665 Huntington Avenue Boston, MA 02115 Marjorie L McCullough Walter C Willett 1. McCullough ML, Feskanich D, Stampfer MJ, et al. Diet quality and major chronic disease risk in men and women: moving toward improved dietary guidance. Am J Clin Nutr 2002;76:1261 71. 2. Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year death rate in the Seven Countries Study. Am J Epidemiol 1986;124:903 15. 3. Willett WC. Nutritional epidemiology. 2nd ed. New York: Oxford University Press, 1998. 4. Mensink RP, Katan MB. Effect of monounsaturated fatty acids versus complex carbohydrates on high-density lipoprotein in healthy men and women. Lancet 1987;1:122 5. 5. Hu FB, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med 1997;337:1491 9. 6. Subar AF, Krebs-Smith SM, Cook A, Kahle LL. Dietary sources of nutrients among US adults, 1989 to 1991. J Am Diet Assoc 1998;98:537 47. 7. Ulbricht TLV, Southgate DAT. Coronary heart disease: seven dietary factors. Lancet 1991;338:985 92. 8. Mensink RP, Katan MB. Effect of dietary fatty acids on serum lipids and lipoproteins: a meta-analysis of 27 trials. Arterioscler Thromb 1992;12:911 9. 9. de Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean alphalinolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet 1994;343:1454 9. 10. Hu FB, Stampfer MJ, Manson JE, et al. Dietary saturated fats and their food sources in relation to the risk of coronary heart disease in women. Am J Clin Nutr 1999;70:1001 8. Bone mineral content, not bone mineral density, is the correct bone measure for growth studies In a recent issue of the Journal, Lehtonen-Veromaa et al (1) reported important findings with respect to the apparent effect of basal vitamin D status on the attainment of peak bone mass in peripubertal girls living at a latitude at which solar vitamin D synthesis in the skin is minimal and in a country () in which vitamin D fortification of milk is so low as to be nearly negligible. The importance of their findings, although statistically significant, is minimized by an unfortunate choice of outcome variable, ie, areal bone mineral density (BMD). This is precisely the wrong measurement during growth, because it factors out most of the component of bone accumulation that is associated with change in bone size. What is important in a growth experiment is bone mass (measured as bone mineral content, BMC), not bone density. The authors title captures that truism, even if their data do not. The positive correlation between 25-hydroxyvitamin D [25(OH)D] and BMD gain, depicted in Figure 1 in the article (1), could have been due to the fact that vertebral size was expanding more in the girls with low 25(OH)D values than in the girls with higher 25(OH)D values. With a larger denominator, the BMD value would have increased less. I doubt that that is the case, but without the critical data there is no way to tell. BMD should never be used in a growth study (2, 3). There is no mechanical reason why true density should change appreciably with growth, and Matkovic et al (4) showed that, in fact, it did not. Bone mineral apparent density is an even less appropriate measure under these circumstances, because it represents an empirical method of attempting to adjust for differences in the third dimension that BMD does capture. My guess is that the reported change in BMD in these girls was almost surely less than the actual change in bone mass, and that, had the authors used BMC as their outcome variable, the association with vitamin D would have been even stronger than the association they report. What Lehtonen-Veromaa et al should provide are the BMC and area values (and their corresponding 3-y changes). Only these are capable of capturing the variable in the

LETTERS TO THE EDITOR 351 title of their paper (bone mass). One hopes that they will make these important data available to the readers of the Journal. Creighton University 601 North 30th Street, Suite 4841 Omaha, NE 68131 E-mail: rheaney@creighton.edu Robert P Heaney 1. Lehtonen-Veromaa MKM, Möttönen TT, Nuotio IO, Irjala KMA, Leino AE, Viikari JSA. Vitamin D and attainment of peak bone mass among peripubertal Finnish girls: a 3-y prospective study. Am J Clin Nutr 2002;76:1446 53. 2. Prentice A, Parsons TJ, Cole TJ. Uncritical use of bone mineral density in absorptiometry may lead to size-related artifacts in the identification of bone mineral determinants. Am J Clin Nutr 1994;60: 837 42. 3. Heaney RP. Design considerations for clinical investigations of osteoporosis. In: Marcus R, Kelsey J, Feldman D, eds. Osteoporosis. 2nd ed, vol 2. San Diego: Academic Press, 2001:513 32. 4. Matkovic V, Jelic T, Wardlaw GM, et al. Timing of peak bone mass in Caucasian females and its implication for the prevention of osteoporosis. J Clin Invest 1994;93:799 808. Reply to RP Heaney We thank RP Heaney for his interest in our work and for generating the opportunity for additional discussion and further analysis of our results concerning vitamin D and the attainment of peak bone mass in growing girls as measured with the use of new statistical models (1). We agree that it is evident that bone mineral density (BMD) reflects not only bone density but also bone size and that the outcome variable bone mineral content (BMC) may generally reflect peak bone mass better than does BMD. However, our main results regarding the changes in the BMD of the lumbar spine in the girls who were in the same phase of growth (decelerating with age) and sexual maturation but who had different vitamin D status were convincing. The changes in BMD in growing girls were controlled accurately by the use of several covariates (ie, baseline reproductive years, baseline bone mineral values, increases in height and weight, mean intake of calcium, and mean amount of physical activity during the study) to adjust the changes in bone size. However, it is true that the use of a method with BMC as a dependent variable and with adjustment for bone area (BA) is an interesting way of avoiding pitfalls in the assessment of real changes in bone density in the growing bone (2). Thus, we reanalyzed our main results with the use of this recommended principle. In our 3-y prospective study of 171 peripubertal girls, the correlation between the 3-y changes ( ) in BMD and BMC was highly significant (r = 0.969), and the degree of relation between baseline 25-hydroxyvitamin D [25(OH)D] and BMD (r = 0.35, P < 0.001) was quite similar to that between baseline 25(OH)D and BMC (r = 0.33, P < 0.001). The mean (± SD) crude values of 3-y BMC in the lumbar spine were significantly different in the vitamin D tertiles (11.0 ± 8.0, 10.9 ± 9.6, and 16.1 ± 7.9 g, respectively; P = 0.006), whereas the BA did not differ significantly between the vitamin D tertiles. In the girls with advanced sexual maturation at baseline (n = 129), the difference in 3-y BMC accumulation from baseline (adjusted for baseline reproductive years, baseline bone mineral values, BA, increases in height and weight, mean intake of calcium, and mean amount of physical activity) between the girls with severe hypovitaminosis D [25(OH)D concentration < 20 nmol/l] and those with normal vitamin D status [25(OH)D concentration 37.5 nmol/l] was 6.4% (P = 0.007) in the lumbar spine. In addition, when this method was used (Table 1), BMC was 1.839 g greater (95% CI: 0.436 g, 3.242 g) in the highest vitamin D tertile than in the lowest tertile. BMC values obtained after various other adjustments are also given in Table 1. These values for the femoral neck did not differ significantly, except when BMC values were adjusted only for increases in BA, height, TABLE 1 Three-year changes ( ) in the bone mineral density (BMD) and the bone mineral content (BMC) of the lumbar spine (L1 L4) and at the femoral neck (FN) analyzed after various adjustments in peripubertal girls with advanced sexual maturation (n = 129) according to tertiles of baseline serum 25- hydroxyvitamin D [25(OH)D] concentration Serum 25(OH)D tertiles Lowest Middle Highest Difference between (19.2 ± 5.1 nmol/l) 1 (30.2 ± 2.5 nmol/l) (45.1 ± 8.2 nmol/l) highest and lowest tertiles (n = 46) (n = 38) (n = 45) BMD, L1 L4 (g/cm 2 ) 2 0.111 ± 0.007 3 0.118 ± 0.008 0.140 ± 0.007 0.029 (0.003, 0.054) 4 BMC, L1 L4 (g) 5 12.067 ± 0.401 12.401 ± 0.436 13.906 ± 0.399 1.839 (0.436, 3.242) BMC, L1 L4 (g) 6 12.132 ± 0.403 12.322 ± 0.437 13.906 ± 0.402 1.774 (0.363, 3.185) BMC, L1 L4 (g) 7 11.993 ± 0.403 12.378 ± 0.446 14.000 ± 0.408 2.007 (0.608, 3.406) BMC, FN (g) 7 0.524 ± 0.045 0.581 ± 0.050 0.686 ± 0.046 0.162 (0.0029, 0.322) 1 x ± SE. 2 Adjusted for baseline reproductive year, baseline value of BMD, increases in height and weight, mean intake of calcium, and mean amount of physical activity. 3 x ± SE relative to baseline. 4 x ; 95% CI in parentheses. 5 Adjusted for baseline reproductive year; baseline value of BMC; increases in bone area, height, and weight; mean intake of calcium; and mean amount of physical activity. 6 Adjusted for increases in bone area, height, and weight; mean intake of calcium; and mean amount of physical activity. 7 Adjusted for increases in bone area, height, and weight.

352 LETTERS TO THE EDITOR and weight. In addition, the 3-y adjusted BMC values for lumbar spine or femoral neck among the less mature girls did not differ significantly by 25(OH)D tertile. These results concerning BMC are quite comparable to the original results obtained with the use of BMD as the outcome variable. Perhaps after this analysis, the title of our original paper may be accepted as reflecting the substance of the study reported. Paavo Nurmi Centre Sport and Exercise Medicine Unit Department of Physiology University of Marjo KM Lehtonen-Veromaa Department of Medicine Central Laboratory Department of Clinical Chemistry Ilpo O Nuotio Jorma SA Viikari Kerttu MA Irjala Aila E Leino Division of Rheumatology Department of Medicine Paimio Hospital FIN-21540 Paimio E-mail: timo.mottonen@tyks.fi Timo T Möttönen 1. Lehtonen-Veromaa MKM, Möttönen TT, Nuotio IO, Irjala KMA, Leino AE, Viikari JSA. Vitamin D and attainment of peak bone mass among peripubertal Finnish girls: a 3-y prospective study. Am J Clin Nutr 2002;76:1446 53. 2. Prentice A, Parsons TJ, Cole TJ. Uncritical use of bone mineral density in absorptiometry may lead to size-related artifacts in the identification of bone mineral determinants. Am J Clin Nutr 1994;60: 837 42.