Coté ML, Kardia SLR, Wenzlaff AS, Ruckdeschel JC, Schwartz AG. Risk of Lung Cancer Among White and Black Relatives of Individuals With Early-Onset Lung Cancer. JAMA. 2005;293(24):3036-3042. doi:10.1001/jama.293.24.3036
Author Affiliations: Population Studies and
Prevention Program, Karmanos Cancer Institute at Wayne State University, Detroit,
Mich (Drs Coté, Ruckdeschel, and Schwartz and Ms Wenzlaff) and University
of Michigan School of Public Health, Ann Arbor (Dr Kardia).
Context Evidence exists that lung cancer aggregates in families and recent findings
of a chromosomal region linked to lung cancer susceptibility support a genetic
component to risk. Family studies of early-onset lung cancer patients offer
a unique opportunity to evaluate lifetime risk of lung cancer in relatives.
Objective To measure lung cancer aggregation and estimate lifetime risk among
relatives of early-onset cases and population-based controls.
Design and Setting Familial aggregation and cumulative risk estimates from interview data
of incident cases and concurrently ascertained controls between 1990 and 2003
in metropolitan Detroit, Mich.
Participants The study included 7576 biological mothers, fathers, and siblings of
692 early-onset cases and 773 frequency-matched controls. One third of the
population was black.
Main Outcome Measures Cumulative lifetime risk of lung cancer, stratified by race and smoking
behavior in relatives of early-onset cases and controls.
Results Smokers with a family history of early-onset lung cancer in a first-degree
relative had a higher risk of developing lung cancer with increasing age than
smokers without a family history. An increase in risk occurs after age 60
years in these individuals, with 17.1% (SE 2.4%) of white case relatives and
25.1% (SE 5.8%) of black case relatives diagnosed with lung cancer by age
70 years. Relatives of black cases were at statistically significant increased
risk of lung cancer compared with relatives of white cases (odds ratio, 2.07,
95% confidence interval, 1.29-3.32) after adjusting for age, sex, pack-years,
pneumonia, and chronic obstructive lung disease.
Conclusions First-degree relatives of black individuals with early-onset lung cancer
have greater risk of lung cancer than their white counterparts, and these
risks are further amplified by cigarette smoking. These data provide estimates
of lung cancer risk that can be used to offer counseling to family members
of patients with early-onset lung cancer.
Cigarette smoking has long been established as the major risk factor
for lung cancer in the general population.1 However,
familial aggregation of disease, even after adjusting for smoking habits,
has also been identified.2- 4 Greatest
risk is seen in families with early-onset disease compared with those whose
onset of lung cancer occurred at older ages.5- 8 The
recent identification of a susceptibility region on chromosome 6 provides
more evidence that lung cancer development is not only a disease of environmental
origins (ie, tobacco smoking) but that a genetic predisposition to lung cancer
also exists.9 However, discovery of the specific
germline mutations and development of a predictive genetic test for lung cancer
are still years off.
In the interim, approximately 173 770 new diagnoses were estimated
to have occurred in the United States in 2004 alone.10 Many
of these individuals will seek clinical advice for what this diagnosis means
for their relatives. This study quantifies risk of lung cancer across a lifetime
by race, smoking status, and family history of early-onset lung cancer, which
could be used to identify high-risk individuals and to counsel families with
a history of early-onset lung cancer.
Families were identified through population-based lung cancer cases
diagnosed before age 50 years (case probands) and population-based controls
younger than 50 years (control probands). Eligible cases were identified through
the Metropolitan Detroit Cancer Surveillance System (MDCSS), a participant
in the National Cancer Institute’s Surveillance, Epidemiology, and End
Results program. Cases were defined as those with a primary neoplasm of the
lung or bronchus (invasive behavior, International Classification
of Diseases, Ninth Revision (ICD-9) codes C34.0-34.9, any histology
excluding 8240-8245, 8800-8991, 9140, 9590-9595, and 9670-9717), residing
in Wayne, Macomb, or Oakland counties of Michigan at the time of diagnosis.
Population-based controls in the same age ranges were ascertained concurrently
via random-digit dialing. Case and control probands were frequency matched
by (self-reported by choosing from a list) race, sex, 5-year age group, and
county of residence. All analyses were restricted to white and black individuals.
Data for 3663 family members of cases (mean of 5.3 per case proband) and 3913
family members of controls (mean of 5.1 per control proband) were included
in these analyses.
This protocol was approved by the local institutional review board.
Written informed consent was mailed to probands. If signed consent forms were
not returned prior to initiation of the interview, verbal consent was obtained
from all probands at the start of the interview. Risk-factor information for
first-degree relatives (biological parents, siblings, children) was obtained
from probands or their proxies. Because lung cancer is almost nonexistent
in children, only parents and siblings 18 years and older were included in
these analyses. Smoking status, number of cigarettes smoked per day, and years
of smoking were collected for each family member. Past medical history for
each relative included asking if a physician had ever diagnosed any of the
following conditions: asthma, allergies, chronic bronchitis, emphysema, pneumonia,
chronic obstructive pulmonary disease (COPD), tuberculosis, or cancer.
Case and control probands were not included in the analysis. Distributions
of categorical variables were compared between case and control relatives
using χ2 tests and means of continuous variables were compared
using t tests. Kaplan-Meier survival analysis was
used to estimate the lifetime risks and age at diagnosis distribution in relatives
of the probands. Estimates of cumulative risk of lung cancer for relatives
were calculated with stratification by race and reported pack-years of tobacco
smoking. Kaplan-Meier survival analysis was also used to estimate the cumulative
risk by pack-year history at diagnosis for relatives of cases and controls,
stratified by race. P values from log-rank tests
of equality were used to assess statistically significant differences between
To determine whether familial risk of lung cancer was present after
adjustment for risk factors among relatives, unconditional logistic regression
models using generalized estimating equations (GEEs), that take into account
family correlation structures were used to estimate odds ratios and P values.11 The final multivariable
model used in all regression analyses for case and control relatives included
variables found to be significant (α =.05) in the multivariable
models: age, race, sex, pack-years, history of chronic obstructive lung disease
(for analysis, the variable included any report of COPD, chronic bronchitis,
or emphysema), and history of pneumonia, as well as family history of early-onset
lung cancer. Risk in black case relatives compared with white case relatives
was evaluated using an unconditional logistic regression GEE model, adjusting
for age, sex, pack-years, history of chronic obstructive lung disease, and
history of pneumonia for each relative. SAS version 8.02 was used for all
Kaplan-Meier and GEE analyses.12
Cumulative risk estimates and odds ratios (ORs) were based on family
data provided by 692 early-onset cases and 773 controls. Among case probands,
311 (44.9%) had adenocarcinomas, 90 (13.0%) had small cell lung cancers, 86
(12.4%) had squamous cell carcinoma, 65 (9.4%) had large cell carcinoma, and
the remaining 27 (3.5%) cases were non–small cell not specified, or
113 (14.6%) other histologic types. In this study group, 13.6% of cases reported
having at least 1 first-degree relative with lung cancer. Only 7.9% of controls
reported having a first-degree relative with lung cancer. Family history of
lung cancer did not vary by the proband’s histologic type (P = .85).
Characteristics of family members are reported in Table 1. Data were available for 95.8% of case mothers and 94.2%
of case fathers. Similarly, information was obtained for 96.6% of control
mothers and 95.0% of control fathers. Information was available for 98.7%
of case siblings and 98.0% of control siblings. Mother:father and sister:brother
ratios did not differ by race. Case mothers and siblings were 1.1 and 2.1
years older at age of death or age at interview, respectively, compared with
control mothers and siblings (P = .03 and P<.001, respectively). This age difference was not seen
in fathers. Approximately two thirds of case and control family members were
white. Diagnoses of emphysema (P = .001),
pneumonia (P = .002), and tuberculosis
(P = .01) were reported more often for
case vs control relatives. An allergy diagnosis (P = .03)
was reported more often for control relatives. Case family members were more
likely to be ever smokers and have higher mean pack-years of smoking compared
with control family members (P < .001).
The strength of familial aggregation of lung cancer was determined by
adjusting for age, race, sex, pack-years, history of chronic obstructive lung
disease, and pneumonia for each relative in this population. Risk of lung
cancer was 1.91-fold greater in first-degree relatives of early-onset cases
vs relatives in the control population (95% confidence interval [CI], 1.33-2.73; Table 2). No statistically significant increase
in risk was seen in relatives related to a nonsmoking, early-onset case after
adjusting for age and pack-years of smoking in the relatives (OR, 1.08; 95%
CI, 0.27-4.38). Risk was nearly 2-fold higher in relatives of ever smoking,
early-onset probands vs relatives of ever-smoking control probands after adjustment
(OR, 1.98; 95% CI, 1.31-3.00). A 5-fold increased risk was identified in white
siblings of cases (95% CI, 1.06-25.64) with similar risk estimates between
sexes (data not shown). Although this finding is statistically significant,
it is based on a small number of reported lung cancer diagnoses. Risk of lung
cancer among first-degree relatives of black cases was 3.23-fold higher than
that of relatives of black controls after adjustment (95% CI, 1.72-6.07) and
was similar for parents and siblings. Relatives of black cases were at statistically
significant increased risk of lung cancer vs relatives of white cases (OR,
2.07; 95% CI, 1.29-3.32; data not shown).
Cumulative risk estimates for first-degree relatives, stratified by
race and pack-years of cigarette smoking are shown in Table 3 and Figure 1. As expected,
greatest risk was seen in case relatives with extensive smoking histories.
In smoking relatives who smoked more than 30 pack-years, 5.6% (SE 1.2%) of
white case relatives and 10.6% (SE 3.2%) of black case relatives were diagnosed
with lung cancer by age 60 years. By age 70 years, risk in these groups had
increased to 17.1% (SE 2.4%) in white case relatives and 25.1% (SE 5.8%) in
black case relatives as shown in Table 3 and Figure 1, which illustrates that in both white
and black control relatives who smoke, going from 30 or fewer pack-years of
exposure to more than 30 pack-years of exposure increases risk primarily after
age 65 years. Being related to a case compared with being related to a control
increases lung cancer risk across most age groups. Figure 2 illustrates the cumulative risk of lung cancer increasing
as pack-years of smoking accumulate. Black individuals who are related to
a case were at highest risk, exceeding risk in black individuals related to
a control (P<.001) and exceeding risk in white
individuals related to a case (P<.001). White
relatives of cases and controls have similar risks at lower levels of pack-year
exposure. Risk diverges at approximately 80 pack-years, and elevated risk
is seen among those related to a case compared with control relatives although
there was not a statistically significant difference across all pack-years
of exposure (P = .22).
In the clinical setting, family members of patients with lung cancer
often recognize cigarette smoking as a contributing factor to lung cancer
development but are unaware of the potential heritable component of this disease.
Recent media coverage of the discovery of a chromosomal region linked to familial
lung cancer by the Genetic Epidemiology of Lung Cancer Consortium coupled
with increasing interest in genetic testing for cancers will result in more
requests by relatives for estimates of their own risk. The results of this
study represent the largest population-based sample of case families with
early-onset lung cancer to date and provide clinically relevant lung cancer
risk estimates for white and black families.
Risk of lung cancer varies significantly by race. From 1990 through
2001, the average annual age-adjusted incidence rate of lung cancer in white
men was 88.1 per 100 000 while the average annual age-adjusted rate in
black men was 131.0 per 100 000.13 Average
incidence rates for black and white women during this time do not show as
large a disparity, at 54.5 and 52.2 per 100 000 per year, respectively.13 Our findings demonstrate that lung cancer risk associated
with family history of lung cancer is stronger in blacks than in whites. This
finding could be the result of a higher degree of underlying susceptibility
or aggregation of unmeasured risk factors for lung cancer in black families.
Familial aggregation of smoking is well established.14,15 In
our analysis, when models were created without adjusting for smoking in each
relative (data not shown), the estimated ORs were approximately 20% higher
than in the smoking-adjusted models. A strength of this study was the ability
to adjust for each relative’s smoking exposure, so familial aggregation
of smoking habits is less likely to be driving the findings reported herein.
In our study, smoking habits among relatives were examined by race. Overall,
the mean pack-years of smoking reported for black relatives was 26.3 (95%
CI, 24.8-27.8). In white relatives, the mean pack-years of smoking reported
was significantly greater at 36.9 (95% CI, 35.6-38.2). These data suggest
that black individuals may be more susceptible to lung carcinogens, have different
tobacco consumption patterns than their white counterparts that confer increased
risk, or have other risk factors that aggregate in families that have not
yet been identified. It is plausible that these factors are genetic in nature
and seen more frequently in the black population.
The analytical approach used capitalized on the availability of risk
factor data for relatives. For comparison, we also analyzed our data using
a traditional case-control analysis of the early-onset probands in our population.
We found a 1.7-fold increased risk of lung cancer for individuals who had
a first-degree relative with lung cancer compared with those without a family
history of lung cancer after adjusting for race, age, sex, and pack-years
of smoking (OR, 1.71; 95% CI, 1.16-2.52; data not shown). When stratified
by race, black cases in our population were nearly 3-fold more likely to have
a family history of lung cancer compared with black controls, after adjusting
for age, sex, and pack-years of smoking (OR, 2.86; 95% CI, 1.43-5.72; data
not shown). In our study, familial risk in the black population was higher
than that in the white population (OR, 1.29; 95% CI, 0.80-2.08; data not shown).
These findings provide further evidence that lung cancer aggregates in families
and that aggregation is stronger in blacks.
In our early-onset population, no increase in risk was seen in family
members of nonsmoking probands after adjusting for age and pack-years of smoking
for each relative. Familial risk of lung cancer among nonsmokers and their
relatives has been described in several studies. In a population-based study
of nonsmokers in Detroit, Mich, Schwartz et al16 reported
that relatives related to a nonsmoking proband diagnosed between ages 40 and
59 years were at 6.1-fold increased risk after adjusting for smoking, occupational,
and medical history in each family member (OR, 6.1; 95% CI, 1.1-33.4).16 However, these findings have not been replicated
in other populations. A second population-based study examined lung cancer
risk in relatives of both never and former smokers who had lung cancer and
did not identify significantly increased risk in fathers (OR, 1.85; 95% CI,
0.80-4.33) or mothers (OR, 1.12; 95% CI, 0.22-5.60) of these never and former
smoking cases after adjustment for age and smoking status in each family member.17 A hospital-based study from Texas also reported no
evidence for increased risk of lung cancer among relatives of never-smokers,
after adjusting for age and smoking status of the proband and relative (OR,
0.86; 95% CI, 0.31-2.38).18 It is difficult
to compare results across studies with various inclusion criteria and study
methods and limited sample sizes. This is an area that requires further study.
Chronic obstructive pulmonary disease has been shown to be associated
with an increased risk of lung cancer independent of smoking history19,20 and familial aggregation has been
observed.21- 23 In
our population, aggregation of chronic obstructive lung disease with early-onset
lung cancer was identified in white relatives (OR, 1.48; 95% CI, 1.11-1.97)
but not in black relatives (OR, 0.76; 95% CI, 0.45-1.26) after adjusting for
age, sex, pack-years and history of pneumonia. Chronic obstructive lung disease
was not prevalent enough in these relatives to calculate risk of lung cancer
by disease status; however, it is possible that risk of lung cancer is even
greater for those with a personal history of COPD and that underlying susceptibility
varies by race. The associations between COPD, smoking, family history, and
race need to be evaluated further.
Our study has several limitations. We were unable to contact 37.2% of
case probands identified during this period. However, there were no significant
differences between participants and those who did not participate, stratified
by race, age at diagnosis, sex, or histology. Data for both case and control
family members were obtained from the proband or a proxy, due to the prohibitive
cost of interviewing each family member. Proxy reporting has been shown to
be accurate for cancers,24,25 lung
disease,26 and cigarette consumption,27 the types of data analyzed in this study. Lung cancer
lifetime risk estimates provided by the Surveillance, Epidemiology, and End
Results program are similar to the cumulative risks of lung cancer reported
in our control relatives; therefore, it is unlikely that our study controls
are underreporting family history of lung cancer.13
Two of the most important variables, age and pack-years of cigarette
smoking, were examined to determine the amount of missing data. Ages were
missing for 1.9% of control relatives, including 1 relative with lung cancer,
and 3.2% of case relatives, including 6 relatives with lung cancer. Exclusion
of individuals with missing ages may bias the Kaplan-Meier estimates and the
resulting age at onset distribution.28 To determine
whether the missing ages in our sample biased the results of this study, 3
data sets were created with imputed ages appropriate for each relationship
and race strata. The resulting analyses using the imputed data sets did not
appreciably alter the findings; thus, the original data set (without age imputation)
Pack-year information was unavailable for 484 case relatives (13.2%)
and 364 control relatives (9.3%). Black men, regardless of relationship to
a case or control, were more likely to have missing pack-year information.
Individuals with missing information were excluded from analyses, thereby
dropping 22 lung cancers among family members of cases (13 of which were black
men) and 5 lung cancers in control families (2 of which were black men). Therefore,
risk estimates are likely to be conservative. When data were reanalyzed using
a smoking indicator with fewer missing data (number of cigarettes per day
or years of smoking) instead of pack-years, the findings remained essentially
the same. Because risk estimates are based on contributions of each relative,
stratified by race and adjusting for their age, sex, and pack-years of smoking,
and not on an aggregate family history measure, differential missing data
by race or sex should not bias these estimates.
Overreporting of lung cancers, particularly for case relatives may also
account for the magnitude of our findings. Therefore, attempts were made to
verify the lung cancers reported among family members. We identified a subset
of 115 case and control relatives with lung cancer for whom we had name, year
of birth, and year of death information in an attempt to verify lung cancer
status through the MDCSS and Michigan death certificates. Seventy-two (62.6%)
were confirmed through these mechanisms. Lack of unique identifying information
(ie, social security numbers) and recent institutional review board modifications
restricting the reporting of certain information without the family member’s
consent made this process difficult. When analysis was performed including
only confirmed lung cancer cases, the findings were essentially the same.
The findings presented herein are applicable to first-degree relatives
of early-onset lung cancer cases. Early-onset cases (<50 years at diagnosis)
represent 6.7% of lung cancers diagnosed in the United States.13 The
focus of this study was on early-onset disease because there is a greater
likelihood of a genetic component to risk in this group. Risk estimates presented
herein, therefore, may not be applicable for counseling families with a history
of later-onset lung cancer.
Familial aggregation of lung cancer was demonstrated in families identified
through early-onset lung cancer cases, after other known risk factors were
adjusted for in the analysis. These findings are similar to previous reports3,4,29 but are the first to
report risks for African American families. Additionally, Table 3 and Figure 1 and Figure 2 can be used by clinicians to counsel
relatives of cases about their risk of developing lung cancer based on family
history of early-onset disease and personal tobacco use.
Family history assessment should be included when evaluating smokers
or those presenting with symptoms consistent with lung disease. Further characterization
of high-risk individuals is important to provide clinicians with counseling
tools and to enhance the effectiveness of screening programs. Family history
of early-onset lung cancer in a first-degree relative should be considered
a risk factor in other relatives older than 18 years. As knowledge about risk
factors (other than tobacco use) for lung cancer increases, physicians may
be more likely to consider lung cancer as a differential diagnosis in their
young patients. Earlier diagnosis and intervention may reduce mortality and
morbidity in this population. Ongoing trials should evaluate the usefulness
of screening modalities among those with a family history of early-onset lung
Corresponding Author: Michele L. Coté,
PhD, Epidemiology, 110 E Warren Ave, Detroit, MI 48201 (firstname.lastname@example.org).
Author Contributions: Dr Coté had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Coté, Schwartz.
Acquisition of data: Coté, Schwartz.
Analysis and interpretation of data: Coté,
Kardia, Wenzlaff, Ruckdeschel, Schwartz.
Drafting of the manuscript: Coté, Kardia,
Critical revision of the manuscript for important
intellectual content: Coté, Kardia, Wenzlaff, Ruckdeschel, Schwartz.
Statistical analysis: Coté, Kardia,
Obtained funding: Schwartz.
Administrative, technical, or material support:
Study supervision: Schwartz.
Financial Disclosures: None reported.
Funding/Support: This work was supported by
National Cancer Institute grant RO1-CA60691 and contract NO1 CN65064 (Dr Schwartz).
Role of the Sponsor: The National Cancer Institute
played no role in the design and conduct of the study; collection, management,
analysis, and interpretation of the data; and preparation, review, or approval
of the manuscript.