Weight Gain in Adulthood and Risk of Developing Glucose Tolerance Disturbance : a Study of a Japanese-Brazilian Population

We examined the data from 530 subjects enrolled in a survey on the prevalence of diabetes in a Japanese-Brazilian population aged 40-79 years. Past self-reported and current weight values were analysed. Student t test was used to compare anthropometric measures between subjects with and without disturbance of glucose tolerance (DGT), hypertension and dyslipidemia. Point and interval estimates of the weight at 20 years-, ageand sex-adjusted odds ratios (OR) were obtained by logistic regression analysis to evaluate the relationship between these diseases and the percent weight gain. Subjects with DGT, hypertension or dyslipidemia tended to have higher BMI during adulthood and to gain more weight in a shorter interval of time. Also, they presented higher waist-to-hip ratio and plasma glucose and worse lipid profile. OR were consistent with associations between chronic diseases and percent weight gain. Trend test of OR indicated that the risk of developing DGT alone or combined with hypertension and abdominal obesity increased 2% and 15% by percent unit of gained weight, respectively, as compared with those subjects with stable weight. Weight gain and the rate by which this occurs during lifetime may confer increased risk of chronic diseases. We suggested that preventive measures against obesity, i.e. the maintenance of healthy body weight lifelong, are necessary to minimize the occurrence of these diseases, also among migrant populations such as the JapaneseBrazilians. J Epidemiol, 2000; 10: 103-110


INTRODUCTION
The presence of obesity strongly predisposes to type 2 diabetes mellitus (DM), hypertension, dyslipidemia and cardiovascular disease.According to World Health Organization (WHO) it is an epidemic disease due to the high prevalence observed in industrialized countries and its progressive increase particularly in developing countries u.On the other hand, the maintenance of healthy body weight lifelong represents an important measure to prevent manifestations of the metabolic syndrome, particularly DM and cardiovascular disease 1-4).
Nakayama et al s, in a study of the effect of body mass index (BMI) on morbidity and mortality in a Japanese population found higher age-adjusted prevalence rates of hypertension, glucose intolerance and hypercholesterolemia and also higher age-and sex-adjusted mortality rates from myocardial infarction and stroke, among subjects with BMI >_ 27 kg/m2, as compared with those subjects with BMI between 23 to 25 kg/m2.Fujimoto et al e commented that some non-Caucasian population are particularly susceptible to development of the metabolic syndrome and that lifestyle changes may play an important etiologic role.They postulated that this population when exposed to the Western environment is subjected to a genetic predisposition to weight gain and to concentrate fat in visceral area.
Changes on body weight are associated with ageing but the relationship between the magnitude and rate of weight gain over time with cardiovascular risk factors (DM, hypertension and dyslipidemia) are less understood.In spite of the expected reduction in cardiovascular morbidity and mortality due to the weight control, results from epidemiological studies have been controversial, since associations were demonstrated with both weight losses and gains.These discrepancies have been attributed, at least in part, to the different age groups considered.Changes in body weight of young adults reflect mainly fat mass reductions while in older adults the weight loss may be due to an unknown consumptive disease 7,8).
The aim of this study was to examine the association between body weight variations during adulthood and the presence of disturbances of glucose tolerance (DGT), hypertension and dyslipidemia, based on data obtained from a Japanese-Brazilian community.

Study population
Brazil has the largest Japanese population outside Japan, estimated in 1,288,000 individuals in 1987.Of these, 828,000 subjects were living in the State of Sao Paulo.Japanese colonisation in Bauru, State of Sao Paulo, began in 1914, specially in rural area, but only 11% of all Japanese migrants is still living in rural area.In 1992, a census of the population with Japanese ancestry living in Bauru was conducted as a part of the planning for the diabetes study.A particular cluster of individuals from one specific prefecture of Japan in Bauru was not observed, similarly to what is seen at national level.The total population identified was approximately 3,000 subjects being 12% Issei or first generation, 39% Nisei or second-generation, 30% Sansei or third-generation and 19% mixed individuals.In 1993, we conducted a population-based study in a representative sample of the Japanese-Brazilian population living in Bauru to know the prevalence of DM and other associated diseases among first and second-generation Japanese-Brazilians.Among 1,137 urban Japanese-Brazilians in the age group 40-79 years, all Issei (n=284) and a random sample of Nisei (n=467) were invited to participate in this study.The percentages of non-responders were 15.8% and 12.6% for Issei and Nisei, respectively.Details of this survey was reported elsewhere gtu.Out of 647 subjects initially included 117 Japanese-Brazilians were excluded from the present study due to incomplete information concerning the variables of main interest.The sample was composed by 530 subjects of both sexes (272 first-and 258 second-generation Japanese-Brazilians), aged 40 to 79 years.No difference was observed between people who participate or not considering the variables of main interest.

Data collection
After informed consent, subjects underwent a home interview using a standardised questionnaire applied by a trained interviewer.Demographic, socio-economic and nutritional data were obtained; they were also submitted to a clinical examination which provided anthropometric and blood pressure data.Weight at age 20, the highest and the lowest weights in adulthood were assessed by the questionnaire.Current body weight and height (measured in duplicate with approximation of 100 grams and 0.5 centimetres, respectively) were those values obtained at the baseline clinical examination in 1993.Blood pressure was measured 3 times after a 10-min seated rest, using random-zero sphygmomanometer and the average of the 2 last measures was calculated.Subjects with mean levels of systolic and diastolic blood pressure higher than 140/90 mmHg, respectively, were considered hypertensive as well as those taking antihypertensive agents.Blood specimens for laboratory analyses were taken in the morning after a 12-hour fast.Laboratory procedures included a 2-hour oral glucose tolerance test with 75 grams of glucose.WHO criteria 12) were used to diagnose subjects with normal, impaired glucose tolerance (IGT) or diabetes mellitus (DM).Subjects with IGT and DM were grouped for the purposes of this study and composed the "disturbed glucose tolerance" (DGT) grou p. Serum total cholesterol, triglycerides and high density lipoprotein (HDL) were measured using enzymatic methods; low density lipoprotein (LDL) was calculated from these measures.Dyslipidemia was defined by total cholesterol or triglyceride levels higher than 200 mg/dl, or HDL-cholesterol lower than 35 mg/d, or LDLcholesterol higher than 130 mg/dl for both sexes.

Analyses
Calculations of BMI and waist-to-hip ratio (WHR) were based on data collected in 1993.BMI was calculated as weight (kg) divided by height (meters) squared.Nutritional status of the participants were defined according to BMI categories proposed by WHO 1) (i.e.BMI < 18.5 Kg/m2; 18.5 to 24.9 Kg/m2; 25.0 to 29.9 Kg/m2; >_ 30.0 Kg/m2).WHR was calculated as the ratio of waist (measured at the umbilicus) to the hip (at the level of the trocanter major) circumferences.Abdominal obesity was defined in this study by WHR >_ 0.85 for women and 0.90 for men.Percent weight gain and loss were calculated as weight change (difference between measured baseline weight and recalled weight at age 20) divided by the weight at age 20, multiplied by 100.Rates of weight gain and loss were calculated as the ratios between the weight change and age variation (baseline age minus age 20).
Anthropometric values by categories of glucose tolerance , blood pressure status and lipid profile were compared using unpaired Student t test and data were expressed as mean and standard deviation.Point and 95 % confidence interval (95%CI) of the weight at 20 years-, age-and sex-adjusted odds ratios (OR) were obtained by unconditional logistic regression analysis to evaluate the relationship between the percentage of weight gained and odds of having these diseases.Percentage of gained weight was entered into models as quartiles.Less than 9% of weight gain from age 20 was the reference category and the other levels of exposure were: 9-16%, 17-28% and >_ 29%.Presence of effect modification by age was tested by the insertion of first-order interaction terms between age and percentage of weight gain into the logistic model but excluded in the final model due to non statistically significant (p > 0.05).Possible increases in OR for the presence of chronic diseases with the increases of the percentage of weight gain were verified using trend test.Four models were obtained considering the presence or absence of glucose tolerance disturbance, hypertension, dyslipidemia, or DGT plus hypertension plus abdominal obesity as dependent variable.Dyslipidemia was excluded from the last model due to its very high prevalence in the sample studied.Current mean values of BMI, WHR and laboratories data of subjects with and without DGT plus hypertension plus abdominal obesity were compared using unpaired Student t test 13).A p-value less than 0.05 was considered as statistically significant.Analyses were conducted using Stata software, version 5.0 14) .

RESULTS
Among the 530 Japanese-Brazilians enrolled in this study 49% (n = 260) were women (male/female ratio = 0.96) and 51% (n = 272) were younger than 60 years in 1993.Crude prevalence rates of DGT, hypertension, dyslipidemia and abdominal obesity were 34% (being 17% of DM and 17% of impaired glucose tolerance), 30%, 81% and 80%, respectively.Percentages of Japanese-Brazilians according to sex, BMI at age 20 and current BMI are shown in Figure 1.Higher proportion of women as compared with men was observed among those with BMI at age 20 lower than 18.5 Kg/m2 (15.8% vs. 7.6%, p < 0.05).At age 20, 12% of the participants of both sexes were considered overweight or obese (BMI >_ 25 Kg/m2) while in 1993 this proportion increased 3.3 fold (40%).Similar pattern was observed when subjects were grouped according to sex and the presence of DGT, hypertension and dyslipidemia (data not shown).Among men those with DGT showed higher mean BMI value in adulthood (27.6 vs. 24.9Kg/m2, p < 0.05) and were older (47 vs. 41 years, p < 0.05) than the normal glucose tolerant ones (Table 1).Similar results were noted among men with hypertension, whose mean BMI at age 20 was also higher than the observed value in normotensive men (22.6 vs. 21.8Kg/m2, p < 0.05).Dyslipidemic men presented lower rate of weight loss than those without dyslipidernia (0.12 vs. 0.20, p < 0.05).Among women, anthropometric variables differed mainly according to the presence of hypertension.Hypertensive women showed higher mean BMI at age 20 (   As expected, men with DGT, hypertension and abdominal obesity exhibited a worse profile with higher BMI, WHR, mean blood pressure, fasting and 2-hr plasma glucose and triglyceride levels and lower HDL cholesterol than the normal ones (Table 2).Similar results were observed among women.
Odds ratios of having chronic diseases showed associations between the presence of DGT, hypertension and the simultaneous presence of DGT and hypertension and abdominal obesity with the percentage of gained weight, particularly for values above 16% (Table 3).Trend tests showed that since 20 years of age the risk of developing DGT alone or this condition combined with hypertension and abdominal obesity increased 2% (OR: 1.02, 95% CI: 1.01 to 1.04) and 15% (OR: 1.15,95% CI: 1.08 to 1.24) by percent unit of gained weight, respectively, when compared with those who kept stable weight (reference quar-tile: 0 to 8%), independently on weight at 20 years, sex and age.As expected, baseline weight (weight at age 20) and current age were both important risk factors to the development of chronic diseases.

DISCUSSION
In this study, variables which reflect changes in body weight during adulthood were associated with the presence of DGT, hypertension and dyslipidemia in Japanese-Brazilians. Subjects of both sexes with chronic diseases tended to have higher values of BMI (at age 20, the highest and the lowest reported BMI), to gain more weight in a shorter period and to loose less weight at slower rate than the normal ones.In spite of this, some differences were not statistically significant.Similar results were found by Fujimoto et al 15) among Japanese-Americans.
Logistic regression analyses showed that DGT (alone or Table 2. Mean (standard deviation) of the current values for biological variables of Japanese-Brazilians according to sex and the simultaneous presence of disturbed glucose tolerance (DOT), hypertension and abdominal obesity (n = 83).Table 3.Odds ratios (OR) and 95% confidence interval (95%CI) for the presence of chronic diseases and percent of weight gain.a194 subjects with weight loss or incomplete information were excluded b p -value of trend test of the odds ratio for percent of weight gain = 0 .00c 63 subjects were with weight loss or incomplete information were excluded d Simultaneous presence of disturbed glucose tolerance and hypertension and abdominal obesity e p-value of trend test of the odds ratio for percent of weight gain = 0 .00combined with hypertension and abdominal obesity) was associated with weight gain.Our findings are in agreement with those from Colditz et al 3) and Sakurai et al 16).In a prospective study including 114,281 nurses aged 30 to 55 years, Colditz et al a found relative risks for diabetes of 1.9 and 2 .7 among women who gained 5.0 to 7.9 and 8 .0 to 10.9 kg after age 18 years, respectively, when compared with those with stable weight, independently of family history of diabetes .In a case-control study of 895 Japanese males living in Japan, Sakurai et al 16) found that the weight gain between 20 and 25 years of age was significantly and positively associated with the risk of type 2 diabetes mellitus, independent of current age, BMI at 20 years of age, and weight change within other age strata.
In addition, we observed a worse cardiovascular profile among those with DGT plus hypertension plus abdominal obesity.Similar results were recently found by Everson et al who studied the relationship between the gained weight and the risk of developing metabolic syndrome, among 2272 participants in the Kuopio Ischemic Heart Disease Risk Factors Study.They proposed that the weight gain over time would contribute to an insulin resistance state in middle adulthood, expressed by hyperinsulinemia, hypertension and dyslipidemia, which reinforce the hypothesis that the decreased insulin sensitivity precedes metabolic and hemodynamic disturbances of the metabolic syndrome.Several lines of evidence support the interpretation that insulin resistance occurs after weight gain.Weight gain can lead to hyperinsulinemia and is associated with glucose intolerance 17).
Generally, cross-sectional studies have limitations to establish cause-effect relationship between variables.In the present study, based on reported information about body weight in different life times, we may suggest that the weight gain preceded the occurrence of the chronic diseases studied.This hypothesis is reinforced by the observation that among the newly diagnosed glucose intolerant Japanese-Brazilians, mean percentage of gained weight since age 20 was significantly higher than that found among the subjects with normal glucose tolerance (23.8 vs. 20%, p < 0.05).
Ageing is associated with an increased incidence of diabetes mellitus and coronary diseases.Glucose intolerance and insulin resistance are more common in the elderly 18), but it remains controversial whether this metabolic deterioration is an inevitable consequence of "biological ageing" or the result of environmental or lifestyle variables such as physical inactivity and obesity which usually accompany age.Recent studies have shown that intervention in these modifiable environmental factors improved insulin sensitivity and glucose tolerance but not in insulin secretion 18).In our study, the association of ageing process with further developing of chronic diseases could be observed for all conditions, independently of sex, percentage of weight gain and weight at age 20.In contrast, in a recent study with 17,689 Japanese aged 19 to 88 years living in Japan, Koda et al 19) found positive relationship between changes in body weight and blood pressure in all age groups suggesting that ageing per se did not affect such relationship.
In 1976, the prevalence of obesity (BMI >_ 30 Kg/ml) reported in the Japanese population older than 20 years living in Japan was 0.7% for men and 2.8% for women 1).Later studies found rates of 1.8 and 2.6%, respectively, which were lower than those observed in the present study (7.5% for both sexes) and in the general Brazilian population (5.9% for men and 13.3% for women) 1).Body weight changes are expected with ageing and larger variations have been reported among diabetic subjects.Keesey and Hirvonen 20) found weight variations around 0.5% during a 6 to 10-week period or less for longer periods of time in healthy subjects and between 3.7-4.6% in diabetics.Therefore, the 3.3 fold increase observed on the prevalence of obesity in the Japanese-Brazilian community during adulthood suggests that such increase could not be explained only by the ageing process.We speculate it may be also explained by the acculturation process, particularly related to changes in nutritional and physical activity patterns, which have been demonstrated to shift these migrants to a high risk group for glucose tolerance disturbance and cardiovascular diseases as compared with Japanese living in Japan 9).
The possibility of inaccurate self-reported weight at age 20 could to be raised, since a tendency toward underreporting is known, particularly among obese subjects.However, several studies indicate that, in the lack of measured data, subjects' reports of their weight during young adulthood are reasonably accurate and can be reliably used in epidemiological investigations [21][22][23][24][25][26].In this study, the weight at 20 years of age was a risk factor for DGT and/or hypertension suggesting the need of maintenance of a healthy body weight throughout adult life.
In addition to percentage of gained weight, the time interval of weight changes seems also to be important Although this study did not provide ideal information to quantify this fact, variables such as percentages of gained and lost weight, rate of weight gain and loss were used in an attempt to set pattern.Using these variables, it was assumed that weight gain (or loss) occurred gradually and constantly over time.The combined utilisation of data concerning BMI at age 20, highest and lowest BMI in adulthood, current BMI and ages revealed that 37 subjects gained weight during the whole period, 209 gained weight only during the early adult life loosing weight right after and 26 subjects initially lost but gained weight further (data not shown).Therefore, fluctuations on body composition were not reflected by the variables -percentages of gained and lost weight, rate of weight gain and loss.Additionally, the option for using current BMI to estimate such variables could have underestimated the effect found herein, since the current weight could be influenced by the subjects' awareness concerning the presence of diseases whose treatment includes dietary modifications.If this was the case, it would reduce the magnitude of the differences found.
The results of this study, in agreement with others 2. 3 7. 8 16.293333333c indicate that preventive measures against obesity and consequently maintenance of a constant body weight throughout adult life are necessary to minimize the occurrence of chronic diseases included in the spectrum of the metabolic syndrome,

< 18 .Figure 1 .
Figure 1.Percentage of Japanese-Brazilians according current body mass index (Kg/m-) and at 20 years of age by sex .

Table 1 .
Mean (standard deviation) of anthropometric variables in Japanese-Brazilians by sex and presence of chronic diseases.a BMI: body mass index b RWG (RWL): Rate of weight gain (loss): no included 89 Japanese-Brazilians with no weight modification (50 women and 39 men) RWG (RWL)= (current weight -weight at 20 years of age) (current age -20) %WG (%WL): Percent of weight gain (loss): no included 89 Japanese-Brazilians with no weight modification (50 women and 39 men) %WG (%WL)= (current weight -weight at 20 years of age) * 100 (weight at 20 years of age) d p -value of Student t test rate and percentage of weight loss (0.08 vs. 0.16 Kg/year and 7 vs. 13%, p < 0.05, respectively).