SEARCH

SEARCH BY CITATION

Keywords:

  • colorectal cancer;
  • smoking;
  • cancer recurrence;
  • cancer survival;
  • adjuvant chemotherapy

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

BACKGROUND:

Cigarette smoking has been shown to increase the risk of developing colorectal cancer, particularly smoking early in life. Little is known about the impact of tobacco use on colon cancer recurrence among colon cancer survivors.

METHODS:

The authors prospectively collected lifetime smoking history from stage III colon cancer patients enrolled in a phase 3 trial via self-report questionnaires during and 6 months after completion of adjuvant chemotherapy. Smoking status was defined as never, current, or past. Lifetime pack-years were defined as number of lifetime packs of cigarettes. Patients were followed for recurrence or death.

RESULTS:

Data on smoking history were captured on 1045 patients with stage III colon cancer receiving adjuvant therapy (46% never smokers; 44% past; 10% current). The adjusted hazard ratio (HR) for disease-free survival (DFS) was 0.99 (95% confidence interval [CI], 0.70-1.41), 1.17 (95% CI 0.89-1.55), and 1.22 (95% CI 0.92-1.61) for lifetime pack-years 0-10, 10-20, and 20+, respectively, compared with never smoking (P = .16). In a preplanned exploratory analysis of smoking intensity early in life, the adjusted HR for 12+ pack-years before age 30 years for DFS was 1.37 (95% CI, 1.02-1.84) compared with never smoking (P = .04). The adjusted HR for DFS was 1.18 (95% CI, 0.92-1.50) for past smokers and 1.10 (95% CI, 0.73-1.64) for current smokers, compared with never smokers.

CONCLUSIONS:

Total tobacco usage early in life may be an important, independent prognostic factor of cancer recurrences and mortality in patients with stage III colon cancer. Cancer 2010. © 2010 American Cancer Society.

Tobacco use was estimated to cause approximately 170,000 cancer deaths in 2008 in the United States.1 Worldwide, colorectal cancer is the fourth most commonly diagnosed malignancy, with an estimated 1,023,152 new cases and 529,978 deaths in 2002.2 Nearly 150,000 persons were expected to be diagnosed with and 50,000 to die from colorectal cancer in 2008.1 Studies attribute 11% to 22% of these new cases to tobacco use.3-6 An association between tobacco use and development of premalignant colorectal adenomas has been established. In 1 study, smokers with a 20 to 40 cigarette pack-years history had a 5-fold increased risk of colorectal adenomas compared with nonsmokers.7 Smoking at a younger age and for longer intervals also conveys an increased risk of adenoma.3, 8

There is controversy in the literature regarding the impact of tobacco use on the incidence of colorectal cancer. This may be because of the short follow-up period in some studies.4, 7, 9 Recent studies have noted that the duration and total quantity of cigarette usage, particularly in smokers of at least 4 decades, is strongly associated with an increased incidence of colorectal cancer.7, 10, 11

No data exist regarding the impact of tobacco use on patients already diagnosed with colorectal cancer. The present analysis evaluates the impact of tobacco use—defined as smoking status, intensity, and duration—on colon cancer recurrence in patients with stage III colon cancer.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Study Population

The Cancer and Leukemia Group B (CALGB) led a multicenter, phase 3 trial (CALGB 89,803) in which 1264 patients with resected stage III adenocarcinoma of the colon were randomized to receive bolus 5-fluorouracil/leucovorin with or without irinotecan. Patients were also enrolled by the North Central Cancer Treatment Group, the National Cancer Institute of Canada Clinical Trials group, the Eastern Cooperative Oncology Group (ECOG), the Southwest Oncology Group, and the National Cancer Institute Cancer Trials Support Unit from May 21, 1999 through April 27, 2001. Patients with resected stage III adenocarcinoma of the colon without evidence of radiographic or surgical metastases and without prior radiation or chemotherapy were eligible for this study. Additional eligibility criteria included ECOG performance status of ≤2, normal bone marrow function and bilirubin, serum creatinine ≤1.5× the upper limit of normal, and age ≥18 years. Patients with rectal cancers were excluded. Study treatment commenced between Day 21 and Day 56 after surgical resection. Data and safety monitoring was conducted by the CALGB according to the CALGB policies and procedures. Participating subjects signed informed consent, independently approved at each site's institutional review board.

Tobacco Use Assessment

Early in the enrollment period, the protocol was amended to include a self-administered questionnaire capturing diet and lifestyle habits midway through adjuvant therapy (4 months after surgery; questionnaire 1) and again 6 months after completion of adjuvant therapy (14 months after surgery; questionnaire 2). On questionnaire 1, subjects were asked if they had smoked 20 packs of cigarettes or more in their lifetime. If they answered “yes, currently smoke” or “yes, smoked in past,” they answered the average number of cigarettes smoked per day at age intervals of age <15 years, age 15-19 years, age 20-29 years, age 30-39 years, age 40-49 years, age 50-59 years, and age 60+ years. In addition, for those who answered “yes, smoked in past, but quit,” they indicated the time since smoking cessation (<1 year, 1-2 years, 3-5 years, 6-9 years, 10+ years). Those who answered “no” were categorized as nonsmokers for questionnaire 1. On questionnaire 2, subjects were asked if they currently smoke cigarettes; if they answered “yes,” they indicated the number of cigarettes smoked per day (1-4, 5-14, 15-24, 25-34, 35-44, 45+). If they answered “no,” they were considered nonsmokers for questionnaire 2.

Smoking status was defined as never, current, or past based on data from questionnaire 1. Lifetime pack years were defined as the estimated number of packs of cigarettes smoked per year over the subjects' lifetime based on questionnaire 1. Pack-years were divided into quartiles, for overall pack-years (0, 0-10, 10-20, and 20+ pack-years), comparable to other similar studies in the literature.3-8, 10 Pack-years were further categorized as pack-year quartiles before age 30 years (0, 0-4, 4-12, 12+ pack-years) and after age 30 years (0, 0-1, 1-20, 20+ pack-years) to examine the impact of early smoking on outcomes as previously described.3, 8 Time since smoking cessation was defined by the median value of the time categories since last smoking (0.5, 1.5, 4, 7.5, and 10 years) based on questionnaire 1. In secondary analyses, we updated exposures based on data from the second questionnaire and did not observe an appreciable change in the results. As such, we present only data using the first questionnaire completed during adjuvant therapy.

Study Endpoints

The primary endpoint for these analyses was disease-free survival, defined as time from completion of questionnaire 1 to death from any cause or recurrence of colon cancer. Secondary endpoints were overall and recurrence-free survival. Overall survival is defined as time from completion of questionnaire 1 to death from any cause. Recurrence-free survival is defined as time from completion of questionnaire 1 to recurrence of colon cancer. For recurrence-free survival, subjects who died without evidence of colon cancer recurrence were censored at the time of their last visit to their clinician. Median time from study entry to questionnaire 1 completion was 3.5 months (95% range, 2.5-5.0 months). Survival endpoints were measured from completion of questionnaire 1 to event of interest, excluding events within the first 90 days to avoid biases of change in behavior because of an impending event.

Statistical Analyses

A detailed description of the statistical plan for the comparison of 5-fluorouracil and leucovorin to 5-fluorouracil, leucovorin, and irinotecan has been published.12 Because there was no difference in efficacy between the 2 treatments in the clinical trial, data for all patients were combined.12 In this secondary analysis, differences in survival outcome were calculated using Cox proportional hazards. The Kaplan-Meier method and log-rank analysis were used to graphically represent disease-free survival, overall survival, and recurrence-free survival by smoking status and intensity. The contribution of other variables was also entered into the model as fixed covariates (age at study entry, sex, body mass index at study entry, extent of penetration through the bowel wall, number of positive lymph nodes, and tumor differentiation). Covariates with missing values were coded with indicator variables in all adjusted models. Differences in baseline characteristics across smoking status and intensity groups were tested using the chi-square test of independence for dichotomous/categorical variables and the Wilcoxon test for continuous variables. Testing for linear trends across quartiles of smoking intensity and time since smoking cessation was performed by assigning median values for each quartile then modeling this value as a continuous variable. All statistical analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC). All P values presented are 2-sided, using a level of significance of .05 as statistically significant.

The sample size for the initial study was determined based on a comparison of the 2 chemotherapy regimens in the adjuvant setting for stage III colon cancer. In a post hoc calculation of power based on the known sample size and number of events for disease-free survival for tobacco use, we had >90% power to detect step-wise increases in hazard ratios (HRs) of 0.25 for each category of smoking intensity, compared with baseline 0 pack-year history, for the third and fourth category (10-20 pack years and >20 pack years). We had 80% power to detect a HR of 1.3 for past smokers and 1.5 for current smokers, both compared with never smokers.

Patient registration and clinical data collection were managed by the CALGB Statistical Center. All analyses were based on the study database frozen on March 7, 2007 and performed with the supervision of CALGB statisticians. By using Clark's C, the completeness of follow-up for this study was 83.25%.13 Applying Wu's modification, to adjust for unreported deaths, the assessment of the completeness was 85.0%.14

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Baseline Characteristics

Current smokers were more likely to be male, had lower body mass index (BMI), had more poorly differentiated tumors, and had deeper invasion into the bowel wall at diagnosis than past or never smokers (Table 1). Smokers did not differ from nonsmokers with regard to treatment arm, race, ECOG performance status, postoperative carcinoembryonic antigen (CEA) titers, or the presence of bowel perforation at diagnosis. In regard to baseline characteristics by smoking intensity, those with higher pack-years of smoking were older, were more likely to be male, and had worse performance status compared with never smokers but did not differ by treatment arm, race, postoperative CEA, or the presence of bowel perforation at diagnosis.

Table 1. Baseline Patient Characteristics by Smoking Status and Intensity
CharacteristicsSmoking StatusSmoking Intensity
Never (n=478)Current (n=107)Past (n=460)0 pyr (n=478)0-10 pyr (n=122)10-20 pyr (n=211)20+ pyr (n=205)
  • pyr indicates pack-years; 5-FU, 5-fluorouracil; LV, leucovorin; IFL, irinotecan, 5-FU, leucovorin; ECOG, Eastern Cooperative Oncology Group; PS, performance status; BMI, body mass index; CEA, carcinoembryonic antigen.

  • a

    P < .05.

  • b

    PS 0 indicates fully active; PS 1, restricted in physically strenuous activity but ambulatory and able to carry out light work; PS 2, ambulatory and capable of all self-care but unable to carry out any work activities, up and about >50% of waking hours.

  • c

    T1-2 indicates level of invasion through the bowel wall not beyond the muscle layer; T3-4, level of invasion through the bowel wall beyond the muscle layer.

Age, y       
 Median585764a58616065a
 Range24-8521-8326-8124-8529-8021-8140-83
Men, %457064a45586374a
Treatment arm, %       
 5-FU+LV51574951485052
 IFL49445149525048
Race, %       
 White87909087849293
 Black87681254
 Other5345433
ECOG PS, %b       
 075697775798268a
 1 or 225312325211832
BMI at diagnosis, %       
 Median272628a27282727
 Range17-5017-3816-5217-5017-4316-5016-52
Invasion through bowel wall (T stage), %c     
 T1-218615a18111415
 T3-482948582898684
Grade of tumor differentiation, %       
 Well/moderate796877a79777771
 Poor/undifferentiated21322321232329
No. of nodes sampled       
 Median131312a13121311
 Range1-613-681-681-612-521-681-41
No. of positive nodes       
 Median3323222
 Range1-290-141-231-291-140-181-23
Postoperative CEA level, %       
 ≤5 ng/mL93909293879294
 >5 ng/mL710871386
Clinical bowel obstruction at diagnosis, %     
 No81727681747675
 Yes19282419262425
Clinical bowel perforation, %       
 No97929597949594
 Yes3853656

Smoking Status

The median follow-up time from the time of completion of the first questionnaire was 5.3 years. In total, 363 of the 1045 patients experienced disease-free survival events (colon cancer recurrence and/or death), and 257 patients died from any cause (221 with cancer recurrence and 31 without evidence of recurrence at time of death). Current smokers had a nonsignificant increased HR of 1.30 (95% confidence interval [CI], 0.93-1.84) for colon cancer recurrence or death from any cause compared with never smokers in adjusted analyses (Table 2; Fig. 1). After adjustment for age, sex, number of positive lymph nodes, extent of invasion through bowel wall, tumor differentiation, BMI, and clinical bowel obstruction at diagnosis, the HR for disease-free survival was 1.10 (95% CI, 0.73-1.64). This nonsignificantly inferior disease-free survival seems to be driven by death from any cause. Compared with never smokers, current smokers had an adjusted HR for overall mortality of 1.38 (95% CI, 0.87-2.18), whereas there was no association with recurrence-free survival (HR, 0.90; 95% CI, 0.58-1.41). There were no significant differences in adjusted disease-free, recurrence-free, or overall survival for past smokers compared with never smokers.

thumbnail image

Figure 1. Disease-free and overall survival are shown by smoking status for never, current, and past smokers by time from study entry in years.

Download figure to PowerPoint

Table 2. Disease-Free, Overall, and Recurrence-Free Survival by Smoking Status
Smoking StatusNeverCurrentPast
  • HR indicates hazard ratio; CI, confidence interval.

  • a

    Adjusted for age, sex, number of positive lymph nodes, extent of invasion through bowel wall, tumor differentiation, body mass index, and clinical bowel obstruction at diagnosis.

Disease-free survival   
 No. of cases/at risk152/47842/107169/460
 HR (95% CI), unadjusted1.001.30 (0.93-1.84)1.19 (0.96-1.49)
 HR (95% CI), adjusteda1.001.10 (0.73-1.64)1.18 (0.92-1.50)
Overall survival   
 No. of cases/at risk105/47835/107117/460
 HR (95% CI), unadjusted1.001.65 (1.12-2.42)1.20 (0.93-1.56)
 HR (95% CI), adjusteda1.001.38 (0.87-2.18)1.17 (0.87-1.57)
Recurrence-free survival   
 No. of cases/at risk144/47834/107154/460
 HR (95% CI), unadjusted1.001.11 (0.76-1.61)1.15 (0.92-1.44)
 HR (95% CI), adjusteda1.000.90 (0.58-1.41)1.15 (0.89-1.48)

Our primary analyses were based on smoking status at the time of questionnaire 1. These results did not change when considering status at questionnaire 2 (6 months after completing adjuvant therapy; data not shown). Of 968 for whom we have data for smoking status during questionnaire 1 and questionnaire 2, 95% did not change their status, whereas 3% were current smokers on questionnaire 1 and then nonsmokers on questionnaire 2, and 2% were either past or never smokers and then reported to be current smokers.

Smoking Intensity

We examined whether there was a dose-response association between smoking intensity and colon cancer recurrences and overall mortality (Table 3; Fig. 2A). We initially considered total lifetime exposure using total pack-years of cigarettes. Subjects in the highest quartile of smoking intensity, 20+ pack-years, had an unadjusted HR for cancer recurrences or death from any cause of 1.34 (95% CI, 1.02-1.75), compared with never smokers. In multivariate analyses, this point estimate was attenuated to HR 1.21 (95% CI, 0.92-1.61). The major factor that led to the effect attenuation was sex. However, we did not detect effect modification by sex. In men, the HR for disease-free survival for 20+ pack-years compared with never smokers was 1.37 (95% CI, 0.98-1.92; P = .06), whereas in women, the HR was 1.17 (95% CI, 0.70-1.95; P = .55). The P for interaction by sex was .15. The inferior disease-free survival was primarily because of death from any cause rather than colon cancer recurrences. Compared with nonsmokers, subjects with >20 pack-years history had an adjusted HR for overall mortality of 1.23 (95% CI, 0.88-1.71), but an adjusted HR for colon cancer recurrences of 1.12 (95% CI, 0.83-1.51).

thumbnail image

Figure 2. Disease-free and overall survival are shown by smoking intensity (pack-years) by time from study entry in years: (A) disease-free survival by overall smoking intensity; (B) overall survival by time from study entry in years; (C) disease-free survival by smoking intensity before age 30 years; (D) overall survival by smoking intensity before age 30 years.

Download figure to PowerPoint

Table 3. Disease-Free, Overall, and Recurrence-Free Survival by Overall Smoking Intensity
Overall Smoking Intensity0 pyr0-10 pyr10-20 pyr20+ pyrPtrend
  • pyr indicates pack-years; HR, hazard ratio; CI, confidence interval.

  • Quartiles were computed via SAS v.9 (SAS Institute, Cary, NC).

  • a

    Adjusted for age, sex, number of positive lymph nodes, extent of invasion through bowel wall, tumor differentiation, body mass index, and clinical bowel obstruction at diagnosis.

Disease-free survival     
 No. of cases/at risk152/47840/12279/21184/205 
 HR (95% CI), unadjusted1.001.06 (0.75-1.50)1.26 (0.96-1.66)1.34 (1.02-1.75).03
 HR (95% CI), adjusteda1.000.99 (0.70-1.41)1.17 (0.89-1.55)1.21 (0.92-1.61).16
Overall survival     
 No. of cases/at risk105/47828/12256/21161/205 
 HR (95% CI), unadjusted1.001.08 (0.72-1.65)1.29 (0.93-1.78)1.44 (1.05-1.97).03
 HR (95% CI), adjusteda1.001.01 (0.67-1.54)1.19 (0.86-1.65)1.23 (0.88-1.71).22
Recurrence-free survival     
 No. of cases/at risk144/47837/12273/21171/205 
 HR (95% CI), unadjusted1.001.03 (0.72-1.48)1.22 (0.92-1.62)1.19 (0.89-1.58).23
 HR (95% CI), adjusteda1.000.97 (0.68-1.40)1.15 (0.86-1.53)1.12 (0.83-1.51).41

Because published studies suggest that smoking intensity before age 30 years may be the most predictive influence on colon cancer development,3, 5, 7-9 we tested the impact of smoking intensity before age 30 years and after age 30 years in subjects with established colon cancer. The only baseline characteristic that was significantly different across smoking intensity categories before age 30 years was sex, where the rates of 4-12 pack-years and 12 or more pack-years before age 30 were 20% and 22% in males and 13% and 7% in females, respectively. As with overall smoking intensity, increased pack-years of smoking were associated with worse outcomes in colon cancer survivors (Table 4; Fig. 2B). For smokers with 12 or more pack-years before age 30 years, the adjusted HR for disease-free survival was 1.37 (95% CI, 1.02-1.84; P trend = .04), compared with no smoking history before age 30 years. This association remained significant after further adjustment by smoking intensity after age 30 years (HR, 1.41; 95% CI, 1-1.97; P trend = .05). The point estimates for recurrence-free and overall survival were similar to those seen for disease-free survival for smoking before age 30 years, although they were more limited in power because of fewer events and were not statistically significant (Table 4).

Table 4. Disease-Free, Overall, and Recurrence-Free Survival by Smoking Intensity Before and After Age 30 Years
Characteristics0 pyr0-4 pyr4-12 pyr12+ pyrPtrend
  • pyr indicates pack-years; HR, hazard ratio; CI, confidence interval.

  • Quartiles were computed via SAS v.9 (SAS Institute, Cary, NC).

  • a

    Adjusted for age, sex, number of positive lymph nodes, extent of invasion through bowel wall, tumor differentiation, body mass index, and clinical bowel obstruction at diagnosis.

  • b

    Adjusted as above as well as for smoking intensity after age 30 years (in analyses of pre-age 30 years) or before age 30 years (in analyses of post-age 30 years).

Pack-years of smoking by age 30 years     
 Disease-free survival     
  No. of cases/at risk166/52059/17360/16771/160 
  HR (95% CI), unadjusted1.001.10 (0.82-1.48)1.17 (0.87-1.57)1.49 (1.13-1.97).005
  HR (95% CI), adjusteda1.001.06 (0.78-1.42)1.10 (0.81-1.49)1.37 (1.02-1.84).04
  HR (95% CI), adjustedb1.001.06 (0.78-1.38)1.12 (0.81-1.541.41 (1.00-1.97).05
 Overall survival     
  No. of cases/at risk118/52041/17345/16747/160 
  HR (95% CI), unadjusted1.001.07 (0.75-1.52)1.24 (0.88-1.75)1.40 (1.00-1.96).04
  HR (95% CI), adjusteda1.001.03 (0.72-1.47)1.13 (0.80-1.60)1.27 (0.89-1.80).17
  HR (95% CI), adjustedb1.001.02 (0.71-1.46)1.09 (0.75-1.59)1.20 (0.80-1.81).37
 Recurrence-free survival     
  No. of cases/at risk156/52054/17356/16760/160 
  HR (95% CI), unadjusted1.001.07 (0.78-1.45)1.16 (0.85-1.57)1.32 (0.98-1.78).06
  HR (95% CI), adjusteda1.001.04 (0.76-1.42)1.12 (0.82-1.52)1.29 (0.91-1.76).11
  HR (95% CI), adjustedb1.001.05 (0.77-1.43)1.17 (0.84-1.63)1.38 (0.96-1.98).08
Pack-years of smoking after age 30 years     
 Disease-free survival     
  No. of cases/at risk193/59955/14750/12958/145 
  HR (95% CI), unadjusted1.001.25 (0.92-1.68)1.31 (0.96-1.79)1.28 (0.95-1.71).12
  HR (95% CI), adjusteda1.001.08 (0.79-1.47)1.25 (0.91-1.71)1.20 (0.89-1.64).21
  HR (95% CI), adjustedb1.000.95 (0.67-1.34)1.11 (0.79-1.57)1.00 (0.69-1.46).85
 Overall survival     
  No. of cases/at risk130/59941/14735/12945/145 
  HR (95% CI), unadjusted1.001.39 (0.98-1.98)1.38 (0.95-2.00)1.50 (1.07-2.10).03
  HR (95% CI), adjusteda1.001.21 (0.84-1.72)1.29 (0.89-1.88)1.32 (0.93-1.88).14
  HR (95% CI), adjustedb1.001.16 (0.78-1.74)1.25 (0.84-1.88)1.26 (0.82-1.94).38
 Recurrence-free survival     
  No. of cases/at risk182/59950/14746/12948/145 
  HR (95% CI), unadjusted1.001.19 (0.87-1.63)1.27 (0.92-1.76)1.12 (0.81-1.54).49
  HR (95% CI), adjusteda1.001.07 (0.77-1.47)1.24 (0.89-1.72)1.11 (0.79-1.54).48
  HR (95% CI), adjustedb1.000.97 (0.68-1.40)1.14 (0.80-1.63)0.96 (0.64-1.45).96

Smoking intensity after age 30 years was not significantly associated with any of the survival endpoints (adjusted P trend = .21 for disease-free survival, .14 for overall survival, and .48 for recurrence-free survival) (Table 4).

Time From Smoking Cessation

The majority of past smokers reported cessation >10 years before completing the first questionnaire. For past smokers, compared with those who reported cessation of tobacco usage within 1 year of the first questionnaire, those who quit at least 10 years prior had an adjusted HR for disease-free survival of 0.93 (95% CI, 0.54-1.58). The P trend by number of years from cessation was .89. Similarly, overall survival and recurrence-free survival were not associated with number of years of cessation (adjusted P trends = .55 and .82, respectively).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

By using a cohort of patients who participated in CALGB 89,803, we evaluated the impact of smoking on colon cancer recurrence and survival in the patients who participated in this clinical trial. Neither smoking status nor time since smoking cessation appeared to have a statistically significant impact on the primary endpoint of these analyses, disease-free survival, or the secondary endpoints, overall survival or recurrence-free survival. In contrast, a dose-response association was noted for smoking intensity, particularly for the risk of death or recurrence in higher quartiles of pack-years smoked before age 30 years compared with nonsmokers.

The notion that a longer induction period of tobacco use may be associated with increased risk of colon cancer recurrence or death has not previously been demonstrated in a population already diagnosed with colon cancer. Giovannucci and colleagues, however, demonstrated that smoking during early adulthood was more predictive of an increased risk of adenomas than other smoking variables in 2 large cohort studies.3, 8 In a prospective study of self-reported tobacco use and risk of developing adenoma and colorectal cancer among 121,700 women participating in the Nurses Health Study, Giovannucci et al noted that increased smoking intensity before age 30 years was associated with a 2-fold increased risk of rectal cancer (95% CI, 1.24-3.42) and 23% increased risk of colon cancer. A long induction period (smoking before age 30 years) was associated with an increased risk of colorectal cancer after age 55 years (overall risk ratio [RR], 1.71; 95% CI, 1.27-1.31; P = .008). Similar results were noted in a companion study of 51,529 male health professionals in the Health Professionals Follow-up Study for smoking intensity before age 30 years, with a nearly 2-fold increased risk of colon cancer (RR, 1.96; 95% CI, 1.16-3.29; P trend = .007).

In the present analysis, smoking intensity, particularly before age 30 years, is associated with worse disease-free survival. It is plausible that a greater duration of exposure to tobacco leads to a more aggressive tumor. Adjustment for known clinical factors associated with more aggressive tumor behavior (extent of invasion through bowel wall, nodal stage, grade of differentiation, presence of clinical bowel obstruction or perforation at diagnosis) did not fully attenuate our finding. Thus, although smoking intensity early in life may lead to a more aggressive cancer, it does seem to be independent of other known prognostic factors.

Observational studies are limited in elucidating the mechanism by which tobacco use is associated with risk of premalignant adenomas and colorectal cancer. However, most studies suggest that smoking intensity in early adulthood may be associated with molecular changes early in the pathogenesis of colorectal cancer. One theory of mechanism is that a deletion of the gene GSTM1, which encodes enzymes responsible for detoxification of carcinogenic compounds found in tobacco, leads to an increased risk of colorectal cancer. However, this was not supported by a case-control pooled analysis of 6 studies.15 Slattery et al found a significant increase in the presence of microsatellite instability in colon tumors among subjects who smoked >1 pack per day compared with nonsmokers (odds ratio, 1.6; 95% CI, 1-2.5 for men; odds ratio, 2.2; 95% CI, 1.4-3.5 for women).16 A strong association between colon tumors with microsatellite instability and cigarette smoking was noted in a subset of heavy cigarette smokers (those who started smoking at an early age, smoked for >35 years, those who were current smokers, and those who stopped smoking <15 years before the diagnosis of colon cancer).

Smoking status and intensity are shown to be associated with an increase in genetic alterations in adenomas and sporadic colon cancers in some17-20 but not all15, 16, 21 studies. In a Dutch case-control study, Diergaarde et al noted an increase in p53 overexpression and the presence of transversion mutations in APC, K-ras, and p53 for ever smokers compared with never smokers among 176 case subjects compared with 249 controls.17 In that study, there was not a clear correlation of smoking status or intensity with the presence of microsatellite instability. Huang and colleagues reported significantly higher rates of p53 deletion mutations among smokers compared with nonsmokers.22 Recently, Limsui et al showed that smoking status and intensity was linked to specific colon cancer carcinogenesis pathways in a prospective cohort of nearly 42,000 Iowa women.23 There was a statistically significant risk of the presence of BRAF mutation or CpG island methylation phenotype for women associated with cumulative pack-years smoked or with current smoking status.

The present study uniquely evaluates the risk of colon cancer recurrence in a cohort of patients with stage III colon cancer. However, there are limitations that should be considered. These analyses are based on self-reporting of tobacco use history. Self-report of tobacco use has been shown to be accurate for current smokers. Approximately 20% of never smokers tend to misreport their tobacco exposure24; such measurement error would bias toward the null. Patients that enroll in randomized trials may differ from the population at large. However, the distribution of smoking status as well as past smoking intensity is consistent with other studies in noncancer populations.25, 26 Because the study included patients across North America, at community and academic centers, our findings should reflect the general population of stage III patients.

Gritz et al called for the inclusion of questions on smoking status in clinical trials to determine the impact of tobacco use on cancer treatment and outcome.27 The CALGB 89,803 smoking analysis allows for evaluation of such an impact in colon cancer recurrence. A dose-response for lifetime total of tobacco usage exists that may impact survival outcomes in patients with stage III colon cancer, particularly for longer induction periods of tobacco usage before age 30 years. Further research on the potential interactions between molecular markers and smoking history and confirmation of these findings with independent datasets are warranted.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

We thank Charles S. Fuchs, MD, MPH and Robert Mayer, MD of the Dana-Farber Cancer Institute for independently reviewing the manuscript.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

This work was supported by National Cancer Institute grants to CALGB (grant CA32291 to N.J.M., J.A.M.; grant CA33601 to D.N., D.H.; grant CA77651 to L.B.S.; grant CA47559 to R.G.), the North Central Cancer Treatment Group (grant CA35415 to P.S.), the National Cancer Institute of Canada (grant C015 to R.W.), the Southwest Oncology Group (grants CA38926, CA32101, CA46282 to A.H.), and ECOG (grant CA17145 to A.B.). Research for CALGB 89803 was also supported by grants from the National Cancer Institute (CA31946) to the Cancer and Leukemia Group B (Richard L. Schilsky, MD, Chairman), and to the CALGB Statistical Center (Stephen George, PhD, CA33601). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES