INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: spirometry; epidemiologic studies; genetics

Tobacco smoking is the most important exogenous risk factor for the development of chronic obstructive pulmonary disease (COPD). Although only 15-20% of smokers develop clinically significant COPD, the risk of COPD is approximately two to three times as great in smokers who have a first-degree relative affected by COPD (1). This suggests that genetic factors may be important determinants of susceptibility to the development of COPD in smokers, although the only clearly established genetic risk factor for smoking-related COPD is homozygous deficiency of ₁-antitrypsin, which accounts for fewer than 2% of cases of COPD (2). The search for genetic risk factors for COPD is impeded by the late age of onset of the disease, which makes it difficult to identify affected family members in more than one generation, and by competing smoking-related comorbidities, such as coronary heart disease death. Because symptomatic COPD is preceded by years or decades of accelerated lung function decline, spirometric measures of lung function can be used as markers of preclinical disease in assessing both genetic and environmental risk factors for COPD. Previous studies suggest that familial factors account for 40-50% of the variability in cross-sectional lung function (5), but to our knowledge, no prior report has examined the heritability of longitudinal change in lung function. In the present study, we used spirometric data collected over several decades, in two generations of families participating in the Framingham Heart Study, to evaluate the heritability of longitudinal change in lung function.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sample

The Original Cohort of the Framingham Study was established between 1948 and 1952 as a random sample of adult residents of the town of Framingham, MA. Of the approximately 10,000 men and women aged 30-59 yr living in the town, 5,209 individuals were enrolled in the study. This group included 596 sibships and 1,644 husband-wife pairs. Reflecting the composition of Framingham in 1948, the cohort was > 99% white. Original Cohort members underwent periodic clinical examinations on a biennial basis. Between 1971 and 1975, the Framingham Study was expanded to include a second generation. This Offspring Cohort comprises 5,124 individuals, including 2,616 biologic children of the 1,644 Original Cohort husband-wife pairs and 898 biologic children of 378 other Original Cohort members with heart disease. Offspring Cohort members underwent periodic clinical examinations every 4 yr.

Biologic and spousal relationships were reported by the participants in both cohorts of the Framingham Study and extended pedigrees were constructed. For this analysis, those subjects with acceptable spirometry at both the baseline and follow-up time points and complete data for height, weight, and smoking status were used to obtain standardized residual change in lung function measures, as described below. This sample includes 5,162 individuals, of whom 2,623 were members of the Original Cohort and 2,539 were members of the Offspring Cohort. Of these 5,162 individuals, 4,314 individuals in 958 extended families had at least one biologic relative with lung function data, and therefore contributed to the heritability analysis.

For each subject, current smoking status was ascertained at each clinical examination. From these data, the interim packs/d of cigarette smoking between the baseline and follow-up spirometry was calculated. Weight and standing height were measured with shoes removed.

Spirometry

Spirometry from two points in time was used to estimate the annualized rate of decline in lung function. For members of the Original Cohort, the baseline examination was Exam 6 (conducted in 1960-1961) if spirometry data were available at that examination. If spirometry was available at Exam 5 but not Exam 6, then Exam 5 (1958-1959) was used for the baseline. The follow-up spirometry was performed at Exam 13 (1974-1975). Although spirometry was performed at other cycle examinations, these were chosen because of the high quality of spirometry performed during these cycles (13). For members of the Offspring Cohort, the baseline spirometry was performed at Exam 3 (1984-1987), and the follow-up spirometry at Exam 5 (1992-1995).

Spirometry at Original Cohort Examinations 5-6 was performed using a 13.5L Collins water-sealed bell spirometer. Each subject performed three forced vital capacity (FVC) maneuvers while standing. The FEV₁ and FVC were measured by hand from the chymograph tracings using the method of back extrapolation. Only the maneuver with the greatest FVC was recorded in the chart and available for this analysis. Although these maneuvers were performed well before the first American Thoracic Society guidelines for standardization of spirometry (14), a review of a sample of the spirograms from these examinations indicated that > 80% of the tracings met acceptability guidelines relating to smoothness of curve and forceful initial push (13). Raw values were multiplied by 1.10 to correct for BTPS (body temperature, ambient pressure, saturated with water), assuming a clinic temperature of 20° C.

Spirometry for the Original Cohort Examination 13 and Offspring Cohort Examination 3 was performed using a 6L Collins water-sealed bell spirometer connected to an Eagle II microprocessor. This provided automatic correction for BTPS, based on calibrations performed daily by the technicians. At Offspring Examination 5, the spirometer was interfaced with lung function software developed by S & M Instruments (Doylestown, PA) and adapted for use in epidemiologic studies, which provided real-time quality assurance information to the technician supervising the spirometry. FVC maneuvers were performed standing while wearing nose clips and repeated until at least three acceptable spirograms were obtained, up to a maximum of eight spirograms, using American Thoracic Society standards to identify acceptable spirograms (15). The largest FEV₁ and the largest FVC from among all acceptable maneuvers, and the FEV₁/FVC ratio from the maneuver with the largest sum of FEV₁ + FVC, were used in this analysis.

Analysis

Annualized change in FEV₁, FVC, and FEV₁/FVC ratio were obtained by subtracting the baseline measurement from the follow-up measurement, and dividing by the time in years between measurements. Predicted values for the change in each spirometric measure were obtained by regression of the annualized change on age (using the mean age during the follow-up interval), initial height, initial weight, change in weight during the follow-up period, and the initial value of the spirometric measure in lifetime nonsmokers. For each of the three measures of lung function, separate models were calculated for men and women within each of the two cohorts. To optimize the accuracy of the predictive models, all subjects with nonmissing data for initial and follow-up lung function and all independent variables were included in the models, even if they had no biologic relatives in the study with available spirometry measures (n = 5,162). Residual change in lung function was defined as the difference between the measured change in lung function and the change in lung function predicted by the regression equation. Standardized residual change in FEV₁ (srFEV₁), FVC (srFVC), and ratio (srratio) for each subject were calculated by dividing the residual change by the standard deviation of residual change in each sex- and cohort-specific group. The use of the standardized residuals as the phenotypic variables in the heritability analyses provided adjustment for sex effects and for both biologic and technical cohort effects on change in lung function. Heritability of the standardized residual lung function measures was estimated using variance components analysis (SOLAR version 1.6.6, Southwest Foundation for Biomedical Research). As the distribution of the standardized residuals was approximately normal, no further transformation was performed prior to variance components analysis. Interim pack-years of cigarette smoking and interim smoking cessation (as a dichotomous variable) were included as covariates in the analysis, in order to assess the proportion of variance due to this environmental exposure. Variance components analysis was performed on the entire sample, and in subsets of the sample that included only pairs concordant for smoking status.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The spirometry used in this analysis was performed during middle age in both the Original and Offspring Cohorts (Table 1). The mean interval between the baseline and follow-up spirometry was 14.1 yr in Original Cohort and 7.1 yr in Offspring Cohort subjects. Offspring subjects were taller and heavier than Original Cohort subjects, and were less likely to be current cigarette smokers, reflecting secular trends in body habitus and smoking behavior in the United States between the 1960s and the 1980s. Original Cohort subjects had lower unadjusted lung function (FEV₁ and FVC) compared with Offspring Cohort subjects. In part, this difference reflects both the shorter height and greater prevalence of smoking in the older generation, but may also reflect changes in spirometric techniques employed in the 1960s versus the 1980s.

The models used to calculate predicted change in FEV₁, FVC, and ratio for subjects in each cohort are given in Table 2. Older age was associated with a more rapid decline of all three measures. Greater height was associated with a slower decline of FEV₁ and FVC, and a slightly more rapid decline in ratio. Greater initial weight was weakly associated with more rapid decline in FEV₁ and FVC, as was interim weight gain. When these models were repeated with the inclusion of smokers and the addition of interim packs/d of smoking and interim smoking cessation as independent variables, the expected effects of smoking on change in lung function were observed. Cigarette smoking was strongly associated with a more rapid decline in lung function, approximately 10-13 ml/yr FEV₁, 8-11 ml/yr FVC, and 0.2-0.3 percent/yr FEV₁/FVC ratio for each pack/d of smoking. Interim smoking cessation was associated with a reduced rate of lung function decline, although this effect was not significant. The negative coefficients for initial lung function level reflect the effect of regression to the mean. Excluding initial lung function from these models had no meaningful effect on the subsequent analyses.

The results of the variance components analysis of standardized residual change in lung function are shown in Table 3. When all subjects were included in the analysis, the estimated heritability of 0.05 for srFEV₁ was trivial and not significantly different from zero (p = 0.06). The heritability estimates of 0.18 for srFVC (p < 0.0001) and 0.13 for srratio (p = 0.0001) were somewhat larger although still quite modest. The proportion of variance explained by interim smoking was also modest, at 3-6% of total variance in the three measures of lung function decline.

The above analysis was repeated using only those subjects concordant for smoking status during the interval between the baseline and follow-up spirometry. Subjects were considered concordant for interim nonsmoking if both relatives reported not smoking throughout the interval (n = 2,700). Pairs were considered concordant for interim active smoking if both relatives reported smoking  0.4 packs/d of cigarettes during the interim (n = 1,251). When the analysis was restricted to subjects who were concordant for interim smoking, the heritability estimates were substantially increased to 0.18 for srFEV₁ (p < 0.05) and to 0.39 for srFVC (p < 0.001). The heritability of srratio was essentially unchanged at 0.14. Within this group, there was a small residual effect of the intensity of interim smoking, which accounted for 1-5% of the variance in lung function decline. The estimated heritabilities of srFEV₁ and srFVC also increased when the analysis was restricted to subjects concordant for interim nonsmoking, although to a lesser extent than was seen in the smoking-concordant subjects. The estimated heritability of srratio, however, increased substantially to 0.29.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have previously demonstrated substantial familial correlation of cross-sectional lung function in the community-dwelling sample of middle-aged adults participating in the Framingham Heart Study (11). In the present study, we utilized data from these subjects to examine the heritability of longitudinal change in lung function, a trait that may be particularly relevant to the development of COPD. Within the entire study sample, there was evidence of only minimal heritability of rate of decline in FEV₁, FVC, and FEV₁/FVC ratio after adjustment for age and body habitus, with heritable factors accounting for an estimated 5-18% of the variance in rate of lung function decline. Approximately one-third of subjects smoked during the interval over which lung function was measured, and smoking accounted for an additional 3-6% of the variance in rate of lung function decline. When the analysis was restricted to subjects concordant for interim smoking habits, there was a substantial increase in the magnitude of the estimated heritability of each of the three measures of lung function decline, with the greatest increase in heritability observed for srFEV₁ and srFVC in interim smokers, and for srratio in interim nonsmokers.

Prior studies have demonstrated substantial familial correlation of cross-sectional level of spirometric lung function. Pedigree-based studies suggest that 20-60% of total phenotypic variance may be accounted for by familial factors (5). Estimates of the narrow-sense heritability of cross-sectional FEV₁ in these studies has ranged from 28 to 47%, substantially greater than our present estimates of heritability of decline in lung function. Twin studies have also generally shown evidence of a genetic contribution to cross-sectional lung function, although the range of heritability estimates from these studies ranges from near zero to almost 100% (16). In our prior evaluation of cross-sectional lung function in the Framingham Study cohorts, we found that the heritability estimates for cross-sectional level of FEV₁, FVC, and FEV₁/FVC ratio, after adjustment for age, body habitus, and smoking history, were 0.37, 0.53, and 0.25, respectively (unpublished data). The higher heritability estimates for cross-sectional measures of FEV₁ and FVC, compared with either cross-sectional FEV₁/ FVC ratio or longitudinal change in FEV₁, FVC, or ratio, may in part reflect the greater precision with which cross-sectional FEV₁ and FVC can be modeled on known predictors. In our own data, age, height, weight, and smoking history explain approximately 50% of the variance in cross-sectional FEV₁ and FVC in both men and women, but only approximately 20% of the variance in cross-sectional FEV₁/FVC ratio or in any of the longitudinal change measures.

It is likely, however, that true biologic differences in heritability of cross-sectional versus longitudinal lung function are important. The strong familial correlation of cross-sectional level of lung function is likely to reflect familial aggregation of factors influencing lung growth and development, which may be highly genetically regulated, whereas rate of lung function decline is likely to reflect the effects of environmental exposures, which are difficult to measure, and gene-environment interactions. Although the maximal lung size and function attained by young adulthood may well be relevant to the risk of developing COPD, most COPD occurs in later life; the accelerated decline of lung function over a period of many years during adulthood is critical to the pathogenesis of this condition. Although smokers have, on average, a rate of lung function decline that is approximately 50% greater than nonsmokers, many smokers experience normal rates of decline and a minority have markedly accelerated decline (1). Smoking relatives of patients with COPD are two to four times more likely to develop COPD than are smokers without affected relatives (3, 4, 20, 21). These data suggest that accelerated lung function decline leading to COPD should aggregate in families. Our present findings confirm such a familial aggregation of lung function decline, although the effect is modest.

One potential explanation for the relatively small magnitude of the observed heritability is that decline in lung function results from gene-environment interactions, requiring both a sufficient environmental exposure (e.g., cigarette smoke) and a genetically susceptible individual. Adjustment for smoking history using multiple linear regression analysis, which determines a sample-wide mean effect of smoking on lung function decline, fails to account for individual differences in susceptibility. Therefore, assessment of heritability of lung function decline among relatives discordant for smoking status is likely to underestimate the importance of genetic factors. This is consistent with our observation that the heritability of rate of decline in lung function among biologic relatives increases when the analysis is restricted to relative pairs concordant for smoking status.

If gene-environment interactions are indeed important in determining the rate of decline in lung function, genetic factors that confer susceptibility to the adverse pulmonary effects of cigarette smoke might have little or no influence on the rate of lung function decline in nonsmokers. If so, we would expect that heritability of lung function decline would be greater in pairs concordant for cigarette smoking than in pairs concordant for the absence of smoking. This appeared to be the case for decline in FEV₁ and FVC, although the opposite was observed for FEV₁/FVC ratio.

A potential limitation of our analysis relates to possible imprecision in spirometry data collected approximately 20 yr earlier in the Original Cohort than in the Offspring Cohort, prior to the publication of ATS guidelines for standardization of spirometry (14, 15). The use of cohort- and sex-specific regressions and standardization of the residual lung function measures minimizes the impact of cohort effects related to spirometric technique. As the spirometric measurements within each Cohort were made during a narrow window of time (generally all completed within a 2- to 3-yr period), and length of follow-up was quite uniform within each Cohort, no further adjustment for the time when the spirometry was actually performed, or the interval between initial and follow-up spirometry, was necessary.

Decline of lung function accelerates with advancing age, and is thus nonlinear over time, as reflected in the strong negative coefficient for age present in the predictive models (Table 2). The use of a two-point gradient to measure rate of decline may therefore reduce the efficiency of the heritability analysis; however, the magnitude of the nonlinear component of lung function decline is relatively small over a 7-14 yr interval. Its impact is further reduced by the very small variance in follow-up time within each cohort, resulting from the timing of scheduled follow-up visits to the Framingham Heart Study clinic. Although spirometry measures were made in each cohort at more than two time points, the methodology for performance of spirometry has changed repeatedly since the inception of the Framingham Heart Study in 1948, and data quality has fluctuated, particularly in the period prior to 1980 (13). We chose examinations that were well attended and at which the spirometry data were determined to be of particularly high quality (13). Despite the day-to-day variability in FEV₁ and FVC, the long interval between initial and follow-up spirometry ensures that the rate of decline in lung function is estimated with a good signal-to-noise ratio, and would be little improved by inclusion of interim spirometry measures in the determination of rate of decline (22).

Another potential limitation is the absence of data regarding other environmental exposures that might contribute to COPD risk, such as occupational dust exposure or environmental tobacco smoke. There was limited opportunity for environmental exposure to occupational dusts in the Framingham area during the period of this study, and the magnitude of the effects of environmental tobacco smoke is small relative to the effects of direct cigarette smoking (23). Nonetheless, even in the analyses stratified by interim smoking status, approximately 60-90% of the variance in our measures of lung function decline remained unexplained, and it is likely that some of this variance is due to the effects of unmeasured environmental exposures. Explaining this variance would improve the power of the heritability analysis. Because biologically related subjects of this study did not share households during the period of study, it is not possible to estimate the effects of unmeasured shared environmental exposures in this study.

In summary, among middle-aged biologic relatives in the Framingham Heart Study, the heritability of longitudinal lung function decline after adjustment for age, body habitus, and smoking status was quite modest. Heritability estimates increased when the analysis was limited to subjects concordant for smoking status. This suggests that genetic factors contribute modestly to the overall population variance in rate of lung function decline, and that gene-environment interactions contribute to this variance. Genetic linkage analysis in population-based studies may be fruitful in the study of cross-sectional lung function, which is strongly heritable; however, this approach appears to be suboptimal for the study of genetic mechanisms underlying accelerated lung function decline leading to COPD.