Time Trend and Geographic Patterns of Lung Adenocarcinoma in the United States, 1973-2002

Substantial increases in the incidence rate of adenocarcinoma of the lung (ADL) were reported during the last several decades. ADL surpassed squamous cell lung carcinoma in Connecticut in the 1980s. Devesa et al. (1) reported that through 1997, ADL incidence increased in virtually all areas of the world, with the increases among men exceeding 50% in many parts of Europe.

It has been hypothesized that the increasing trend of adenocarcinoma is mainly due to the dissemination of low-tar filter cigarettes (2-5). Smoke from low-yield filter-tipped cigarettes is inhaled more deeply than smoke from earlier unfiltered cigarettes and releases higher concentrations of nitrosamines. Inhalation transports tobacco-specific carcinogens more distally toward the bronchoalveolar junction where adenocarcinomas often arise. An alternative hypothesis may be the air pollution that stems from industrialization and urbanization. Pope III et al. (6) reported that long-term elevated fine particulate air pollution exposures were associated with significant increases in lung cancer mortality based on data from a prospective study. Similar results were reported recently by Vineis et al. (7) who conducted a nested case-control study in 10 European countries. They found a 30% increase in the risk of developing lung cancer for those who were exposed to nitrogen dioxide (NO₂) at levels >30 μg/m³ compared with those who were exposed to NO₂ at levels <30 μg/m³. However, neither of these studies stratified lung cancer data by histologic types.

We conducted this study to determine which of these two factors is the major contributor for the sharp increase of ADL. The relationships between the distribution patterns of ADL and the distribution patterns of air pollution or low-tar cigarette consumption provide epidemiologic evidence to support either the air pollution hypothesis or the low-tar cigarette hypothesis.

We obtained ADL incidence data from the Surveillance, Epidemiology, and End Results (SEER) Program of the U.S. National Cancer Institute. The data covered about 10% of U.S. population in nine standard SEER regions. It includes the states of Connecticut, Hawaii, Iowa, New Mexico, and Utah, as well as the metropolitan areas of Atlanta (GA), Detroit (MI), San Francisco/Oakland (CA), and Seattle/Puget Sound (WA). We describe time trends in the age-adjusted incidence rates of ADL (ICD-O codes 8140, 8211, 8230-8231, 8250-8260, 8323, 8480-8490, 8550-8560, and 8570-8572) and squamous cell lung carcinoma (ICD-O codes 8050-8076) for the years 1973-2002 because the SEER data are available for this period only while we conducted this study. However, we can describe the trend of ADL incidence for Connecticut from 1960 because we can estimate ADL incidence rates for 1960, 1965, and 1970 based on reports from Connecticut (8).

To assess whether the patterns of ADL incidence are associated with the use of low-tar cigarettes (<15 mg/cigarette) in the United States, we estimated the total low-tar cigarette consumption in the United States by year. This consumption was based on the market share of low-tar cigarettes and total cigarette sales in each year. These data were obtained from the Federal Trade Commission (9).

To estimate air pollution, we collected emission data of seven pollutants including particulate₁₀, particulate_2.5, nitrogen oxides (NO_x), volatile organic components, sulfur dioxide (SO₂), lead, and ammonia emission levels from Environmental Protection Agency (EPA; refs. 10, 11). Because the data for particulate_2.5 and ammonia are not available before 1990, we did not describe trends for these two components but estimated the trends of the remaining five pollutants.

To compare the long-term changes of ADL incidence, low-tar cigarette consumption, and NO_x emissions, we converted the levels for each of them into an index so that we can measure the change in the same scale. The index for different years was calculated using the level of each year divided by its own level in 1985 for ADL incidence, low-tar cigarette consumption, and NO_x emission level.

Figure 1 shows the secular trends of different histologic types of lung cancer.

We can see the peak of incidence curves for squamous cell lung carcinoma and small-cell lung carcinoma decrease in the period from 1978 to 1986, but the peak of incidence curve for ADL is at the end of 1990s, about 15 years later than for squamous cell lung carcinoma and small-cell lung carcinoma. After 1986, squamous cell lung carcinoma and small-cell lung carcinoma substantially decline but ADL keeps an increasing trend until 1998.

Table 1 describes the comparison of the changes in indexes for ADL incidence, low-tar cigarette consumption, and NO_x emissions by year. We selected NO_x as an indicator for air pollution because among trends of the five air pollutants, the one with a pattern most similar to the ADL incidence trend is the trend of NO_x, which had a 3.7-fold higher increase at the peak in 1980 compared with 1940, leveled off after 1980, and then declined. The CO, SO₂, and volatile organic component trends showed only 50% increases, which is much smaller compared with the increase of NO_x. These three components declined nearly 30 years earlier than the ADL incidence. No data for Pb level are available until 1970, and there is a substantial decline in Pb emission since 1970. In addition to the similarity in trend pattern with ADL, the total emissions of NO_x reflect the level of urbanization and industrialization in the past decade because the major source of NO_x is combustion from vehicles and industrial activity. Therefore, we used NO_x as indicator of air pollution, which includes nitrogen dioxide (NO₂), nitrogen oxide (NO), and possibly small amounts of other compounds of nitrogen and oxygen. The harmful part of NO_x is NO₂.

Figure 2 provides a graphical presentation of these data. The increase trend of NO_x emissions in Connecticut paralleled with the increase of ADL incidence but came about 10 to 15 years earlier than ADL curve. The NO_x emission declined after 1980, which is about 18 years earlier than the decline of ADL for both the Connecticut curve and the curve for SEER sites, which began a declining trend in 1998. There is no ADL incidence available for 1950. However, we used two reports to estimate this rate in 1950: lung cancer incidence rate and the percentage of ADL in lung cancers. The percentage estimation is based on the best available data reported by Wynder and Graham (12). This study involved hospitals from seven states, New York, and District of Columbia. The percentage of ADL was 6.4% in male lung cancer patients in 1950. The lung cancer incidence data in 1950 were reported by the National Cancer Institute (13), covering Atlanta, Connecticut, Detroit, Iowa, and San Francisco-Oakland. The ADL incidence estimated in this way for the year 1950 is 2/100,000, which can be converted to indices of 10 and 8 for SEER data and Connecticut data, respectively. Although we did not put this estimation in Fig. 2 because it is neither Connecticut data nor SEER data, it is the best available data for our reference. Between 1950 and 1960, there is not any known achievement in diagnosis or changed standard of diagnosis for ADL. Therefore, it can be compared with the data in 1960. It is much lower than the level in 1960 for both Connecticut and SEER data and suggests that a substantial increase of ADL incidence started in 1950s. This is much earlier than the quick increase in low-tar cigarette consumption, which began in 1972 when the market share of low-tar cigarettes sharply increased to 6.6%. The second finding is that the peak in NO_x emissions decreases after the year 1980. This is about 20 years earlier than the decrease trend of ADL incidence rates. In comparison, the low-tar cigarette consumption keeps increasing. Although smoking rates have substantially declined in the last 40 years, the market share of low-tar cigarette increased at the same time. Therefore, it is reasonable to observe that the low-tar cigarette consumption per capita in 2000 is higher than that in 1980.

Of the nine sites in SEER data per EPA NO_x emission map (14), only Utah and New Mexico were in low NO_x emission areas. We combined ADL data from these two areas to compare with the seven other high NO_x emission sites.

Figure 3 shows that the ADL incidence in high NO_x emission areas is much higher than in NO_x low emission areas. We cannot contribute the difference directly to the difference in levels of NO_x because Utah and New Mexico are both characterized as the low NO_x emission areas and low smoking prevalence areas. Utah and New Mexico were ranked the 1st and 14th lowest smoking prevalence state in the United States (15). However, we can estimate the potential effect of difference in smoking prevalence by comparing the squamous cell lung carcinoma incidence because the squamous cell lung carcinoma is well known to have high relative risk due to smoking. The relative risks of cigarette smoking for squamous cell lung carcinoma and ADL are 18.9 and 4.8, respectively (16); therefore, we can expect that if the difference in incidence between high and low exposure areas is due to higher smoking rate alone, the difference would be much greater for squamous cell lung carcinoma than for ADL. However, the data show the opposite. We reviewed the difference shown in Fig. 3 and compared it with that of squamous cell lung carcinoma data, and found that the corresponding difference for squamous cell lung carcinoma is significantly smaller than that for ADL (P < 0.0005). This fact suggests that the large difference in ADL incidence between these two areas is not solely due to smoking. There should be a strong factor besides smoking that is different in these two areas.

Figure 4 shows the incidence trend by race for men. There was no difference in the incidence rates of ADL between White men and Black men in 1973, but the incidence increases much more in Black men than in White men thereafter. By 1992, it was >50% higher in Black men than in White men. A steady decline of ADL incidence started in 1996 for Black and in 1999 for White. Compared with the incidence in 1992, the incidence in 2002 decreased 38.4% for Black and 18.6% for White.

Whether the steep increase in ADL incidence in the past several decades is real or an artifact associated with technologic advances in diagnosis is an issue to be considered. The technologic advances include bronchoscopy (1968), thin-needle aspiration, computed tomography scans, and improved stains for mucin, all introduced in the 1980s (17). However, these advances cannot explain the observed increased incidence in ADL because it occurred before the technology advances. Moreover, the incidence of ADL in males was about 2.5-fold higher than in females in 1973, but this ratio continuously declined and was down to 1.3 in the year 2000. There is no reason to believe that diagnostic advances were greater for one gender than another. Of course, diagnostic advances cannot explain the considerable decrease in ADL incidence rate after 1998. Therefore, the major part of the substantial changes in ADL incidence is real.

One hypothesis explaining the change in ADL incidence in the last half of the century is combustion-source ambient air pollution. We reviewed the temporal trends for particulate₁₀, NO_x, volatile organic component, SO₂, and lead emission levels. The temporal trend of total NO_x emissions, an increase followed by a decline, was pretty similar to the temporal trend of ADL, but about 20 years earlier than the changes in ADL incidence. Twenty years is a reasonable time for an ADL induction period. We did not find any other components that have similar temporal trend. NO_x form when fuel is burned at high temperatures, as in a combustion process. The primary manmade sources of NO_x are motor vehicles (55%), electric utilities (22%), and the rest is from other industrial, commercial, and residential sources that burn fuels (18). Since the enactment of Clean Air Act in the United States in 1970 and the amendment of the Clean Air Act in 1990, the NO_x emission by automobiles and industry has leveled off and then substantially declined.

In contrast, the temporal trend argues against the low-tar cigarette hypothesis for two reasons. First, increases in ADL occurred before the nationwide use of low-tar cigarettes. Increases in ADL incidence started in 1950, but the sales-weighted average tar level was still >37 mg in 1956 (19). The market share of cigarettes with tar <15 mg was only 3.6% in 1970 (20), but about half of the total percent increase in ADL had been completed at that time. Second, there is no downward trend in consumption of low-tar cigarettes before the declining trend of ADL incidence rates that started in 1998.

In addition to the temporal trend, we checked the distribution pattern of ADL incidence by place and population. Nearly all the metropolitan areas in SEER program showed high NO_x emissions in EPA data (21), except for the states of Utah and New Mexico, where low NO_x emissions were found. To remove the influence of low cigarette smoking rate in low NO_x emission areas, we compared the incidence ratio of high to low NO_x areas for ADL and squamous cell lung carcinoma separately. It is reasonable to expect that the difference in incidence of squamous cell lung carcinoma between high and low exposure areas would be much larger than the corresponding difference for ADL because the relative risk of smoking for squamous cell lung carcinoma is 18.9, which is much higher than the relative risk for ADL, which is only 4.8. However, the data showed the opposite, which indicates that a strong factor other than smoking is involved in high NO_x emissions areas that can explain the unexpected high ADL incidence. Most likely, this factor is air pollution represented by high NO_x emission levels. We found that lung cancer incidence rates of 44 states in 1960 are positively correlated to the vehicle density in these states in 1950 (data are not reported here). These data also support the NO_x hypothesis because automobile exhaustion is the number one source of NO_x emission.

The air pollution hypothesis may also explain the racial difference. Blacks are more likely than Whites to be living in census tracts with higher total modeled air toxics in every large metropolitan area in the United States (22). Moreover, the indoor levels of NO₂ are often higher than the outdoor levels, and the NO₂ level is related to the size of the house (23). For a house bigger than 2,500 ft², the highest level of NO₂ is 30 ppb, but for a house smaller than 1,000 ft², the highest level of NO₂ is 170 ppb. The percent of families below poverty level is 19.3% and 7.1% for Black alone and White alone, respectively (24); therefore, we can believe the percentage of Black families with residences smaller than 1,000 ft² is greater than that of Whites, and therefore more Blacks may exposed to higher NO₂ level than Whites do. Moreover, residence block groups in the lowest quartile of median family income were thrice more likely to have high traffic density than block groups in the highest income quartile (25). This discrepancy in exposure to air toxins supports the air pollution hypothesis and may explain or partly explain why the incidence of ADL in Black males was much higher than in White males. In contrast, the hypothesis of dissemination of low-tar cigarettes cannot explain why ADL incidence is much higher in Blacks. According to the data in Surgeon General Report (1981), “White smokers choose lower tar products approximately twice as frequently as do Black smokers of the same sex” (26).

The percent of ADL of all the histologic types of lung cancer among women is remarkably higher than among men. In addition to differences in genetic background, the NO_x hypothesis may help to explain different proportions by gender because indoor NO_x is produced by fuel burning stoves, furnaces, fireplaces, heaters, water heaters, and dryers and all other combustion appliances. If we consider both cigarette smoking and indoor air pollution as causes, men have higher level of cigarette smoking exposure but women may have relatively higher level of exposure to this indoor air pollution compared with men over the past several decades.

Another possible explanation for the ADL trend may be environmental tobacco smoking (ETS). In recent years, stricter regulations have been enacted that reduced ETS exposure. These changes may partially explain the recent down trend of ADL incidence. However, based on the nationwide Current Population Survey, only four states reported smoke-free public areas in 1992, and 32 states did in 1995 (27). If the policies restricting ETS were enacted in most states during 1990s, it is then less likely to be the cause of downward trend of ADL after in 1998 because the years between the policies and the downward trends are too short for the expected induction period. Moreover, the total lung cancer cases attributed to ETS is about 3,000/y in the United States, which accounts for <2% of the 172,570 new lung cancer cases a year (28). If the decline in the number of lung cancer cases due to declined exposure to ETS is 5% per year, we would expect <0.08% changes in total number of lung cancer cases per year. This number is therefore too small to explain the remarkable change in ADL incidence. In fact, for the period 1998 to 2003, the incidence rate of ADL declined by 14% for men and 8% for women. Such large reductions could not be explained by a decline in the number of people who were exposed to ETS. However, ETS may play a relatively minor role in causing ADL.

Other reported possible causes for ADL include exposure to oil during high-temperature cooking in Asia (29-31) as well as wood smoke exposure in Mexican women (32).

Further evidence supporting the NO_x hypothesis comes from an epidemiologic study of migrant Jews in Israel. Among Jews born in North Africa or Europe, the age-adjusted incidence rates of ADL increased ≥2-fold from 1962 to 1982. The rates did not increase at all among those born in Turkey (33). This finding may reflect the less urbanized, less industrialized environment and lower number of automobiles in Turkey.

Our study is an ecological study and therefore has a major limitation: the data of exposure and disease obtained from populations cannot be linked to individuals. It caused a failure to control confounding factors like smoking. However, unlike most ecological studies, we used a three-dimensional analysis of time-place-population in this study. If we calculate a correlation coefficient from only one dimension, like most ecological studies do, it is possible that many factors can show similar correlation. It is less likely that one factor will show the same relation in time, place, and population dimensions simultaneously. Moreover, the results in this study are consistent with some case-control and cohort studies. This result is consistent with the result of a case-control study conducted by Nyberg et al. (34). They found that for an induction time of 21 to 30 years in those with the highest exposure to NO₂, there is an estimated 44% increase in risk of developing lung cancer (relative risk, 1.44; 95% confidence interval, 1.05-1.99) compared with those who experienced the lowest levels, adjusting for age, year, smoking habits, radon exposure, and occupational exposures known to be associated with lung cancer. In addition, our result is consistent with the results reported by a prospective study and the case control study mentioned in Introduction. Vineis et al. (7) conducted a nested case-control study in 10 European countries and found a 30% increase in the risk of developing lung cancer (relative risk, 1.30; 95% confidence interval, 1.02-1.66) for those who were exposed to NO₂ at levels >30 μg/m³ compared with those who were exposed to NO₂ at levels <30 μg/m³. Pope III et al. (6), based on a prospective mortality study involving 1.2 million adults, reported that each 10-μg/m³ elevation in fine particulate air pollution was associated with an ∼8% increased risk of lung cancer mortality.

It is not clear whether NO₂ is only an indicator of some real causes or it is the major cause. Further studies are warranted.

The temporal and geographic patterns of ADL and its incidence patterns in different racial groups support the air pollution hypothesis. Long-term exposure to some components of polluted air, especially NO_x, may play the major role in the increase in ADL over the last 50 years. There is an urgent need to conduct further studies to determine whether there is a causal relation between long-term, low-dose exposure to NO_x and ADL because both the current guideline standards of 21 ppb (40 μg/m³) proposed by the WHO (35) and the National Ambient Air Quality Standards of 53 ppb (100 μg/m³; ref. 36) proposed in the United States are far above the level used in Vineis study, which suggested a great increase in risk of lung cancer when exposed to NO₂ for a long time at levels >30 μg/m³. Although cigarette smoking is a cause of ADL, it may play a lesser role than air pollution in causing the substantial change in ADL incidence rates over the last several decades.