SEARCH

SEARCH BY CITATION

Keywords:

  • body mass index;
  • obesity;
  • underweight;
  • Asians;
  • colorectal cancer

Abstract

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

BACKGROUND:

The authors chose to examine the association between body mass index (BMI) and incident colorectal cancer across the spectrum of BMI, including underweight persons, because detailed prospective cohort data on this topic in Asians is scarce, as is data on underweight persons (BMI, <18.5 kg/m2) in any population.

METHODS:

Analysis of the Singapore Chinese Health Study included 51,251 men and women aged 45-74 years enrolled in 1993-1998 and followed through 2007. Incident cancer cases and deaths among cohort members were identified through record linkage, and 980 cases were identified. Cox regression models were used to investigate the association of baseline BMI with risk of incident colorectal cancer during a mean of 11.5 years of follow-up.

RESULTS:

A significant, U-shaped, quadratic association was observed between BMI and colon cancer risk, with increased risk in BMIs ≥27.5 and <18.5 kg/m2. The association was more pronounced in never smokers and most prominent when further limiting the sample to those free of diabetes and cases with longer than 5 years of follow-up. Localized cases had a more pronounced association in BMIs ≥27.5, whereas advanced cases had a more pronounced association in BMIs <18.5 kg/m2. No association was found in relation to rectal cancer risk. The association was also stronger among patients aged 65 years and older.

CONCLUSIONS:

BMI displays a U-shaped, quadratic association with colon cancer risk in this Chinese population in Southeast Asia. Cancer 2011;. © 2011 American Cancer Society.

Singapore is historically a low-risk country for colorectal cancer. However, rates of colorectal cancer have doubled among Singapore Chinese in the last 3 decades. During 1968-1972, the age-standardized (world population) rates in Singaporean Chinese men and women were 22.2 and 16.6 per 100,000, respectively. The corresponding rates for 1993-1997 were 45.0 and 32.1 per 100,000, which are higher than comparable rates in US whites (42.2 and 29.5 per 100,000, respectively).1 This increase is primarily due to increasing incidence of colon cancer.1, 2 Incidence of colorectal cancer has been increasing in other economically developed parts of Asia including Hong Kong, Japan, and Korea.3, 4 Excess body weight is considered a risk factor for colorectal cancer in Western populations5; and Asian populations appear to have higher body fat percentages at lower levels of body mass index (BMI kg/m2) relative to Western populations.6, 7 However, detailed prospective studies examining the full spectrum of BMI in relation to risk of colorectal cancer in Asians are scarce, as are any meaningful assessments of the low end of the BMI spectrum from any population. Recently, 4 studies examined the question in Asians; however, the results were not conclusive.8-12 Continued investigation will contribute potential insight into the etiology of colorectal cancer and further understanding of a healthful BMI range in Asians.13

The Singapore Chinese Health Study (SCHS) is a prospective cohort investigation of more than 63,000 Chinese men and women in Singapore. The SCHS provides a unique Asian population relative to other studies on this topic with its ability to adjust more thoroughly for confounding variables. We aimed to perform a comprehensive investigation of the relation between BMI and risk of colorectal cancer in this middle-aged and older Chinese population in Singapore.

MATERIALS AND METHODS

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

Study Population

The design of the Singapore Chinese Health Study has been previously described.14 Briefly, the cohort was drawn from men and women, aged 45 to 74 years at enrollment, who belonged to 2 major dialect groups (Hokkien and Cantonese) of Chinese in Singapore. Between April 1993 and December 1998, 63,257 individuals completed an in-person interview that included questions on demographics, educational attainment, height, weight, use of tobacco and alcohol, usual physical activity, menstrual and reproductive history (women only), medical history, family history of cancer and a 165-item food frequency section assessing usual dietary intake of the previous year. For this study, we excluded 1936 subjects of the original 63,257 participants with a history of invasive cancer (except nonmelanoma skin cancer) or superficial, papillary bladder cancer at baseline because they did not meet study inclusion criteria, and we excluded an additional 10,070 participants who were missing either or both height and weight measures. Hence, the present study included 51,251 participants. Participants excluded because of missing BMI (N = 10,070) were not materially different across the noted demographic and lifestyle characteristics compared with participants with full data who were included in the analysis. The institutional review boards at the National University of Singapore and the University of Minnesota approved this study. Written informed consent was obtained from all participants.

Exposure Assessment

Self-reported height and weight were collected at the baseline interview. BMI was calculated as weight (kg) divided by height squared (m). Self-report of body weight has been shown to be highly valid across many populations,15 as well as specifically in Asians.16 Age was defined as age in years at the time of the baseline examination. Education was categorized into no formal education, primary school, and secondary school or above. Cigarette smoking was classified into never smoker, light, and heavy smoker as described previously.17 The “heavy” smokers were those who started to smoke before 15 years of age and smoked 13 or more cigarettes per day; all remaining ever smokers were defined as light smokers. A history of physician-diagnosed diabetes (yes vs no) was included as well.18

A semiquantitative food-frequency questionnaire specifically developed for this population to assess 165 commonly consumed food items was administered during the baseline interview assessing usual dietary intake of the previous year. The questionnaire has subsequently been validated against a series of 24-hour dietary recall interviews14 as well as selected biomarker studies.19, 20 Dietary patterns were derived for this study population by using principal component analysis as described previously.21 A vegetable, fruit, and soy-rich pattern characterized by high intake of those respective foods was included as a covariate. Frequency of alcohol intake as the summation of beer, rice wine, other wine, and hard liquor was considered to be an individual variable, and persons were grouped as nondrinker/monthly drinker, weekly drinker, and daily drinker.

Physical activity was assessed by using 8 continuous categories ranging from never to 31 hours or more in an average week spent doing strenuous sports (eg, jogging, bicycling on hills, tennis, squash, swimming laps, or aerobics), vigorous work (eg, moving heavy furniture, loading or unloading trucks, shoveling, or equivalent manual labor), and moderate activities (eg, brisk walking, bowling, bicycling on level ground, tai chi, and chi kung). For the present analysis, participants reporting any strenuous sporting activity or vigorous work were combined and compared with those reporting no activity because of low levels of activity. Usual sleep duration was assessed by asking participants the following question, “On the average, during the last year, how many hours in a day did you sleep?” Response categories were 5 hours or less, 6 hours, 7 hours, 8 hours, 9 hours, and 10 hours or more.

Case Ascertainment

Identification of incident colorectal cancer cases and deaths among cohort members were accomplished by record-linkage analysis of the cohort database with respective databases from the population-based Singapore Cancer Registry and the Singapore Registry of Births and Deaths. The nationwide cancer registry has been in place since 1968 and has been shown to be comprehensive in its recording of cancer cases.1 As of April 2008, only 27 cases were known to be lost to follow-up because of migration out of Singapore. As of December 31, 2007, 980 participants included in this study had developed invasive colorectal cancer (596 colon cancers and 384 rectal cancers). In the analysis, rectal cancers included cancers of the rectosigmoid junction. The diagnoses of 951 (97.0 %) colorectal cancer cases were confirmed histologically, and the remaining cases were diagnosed clinically (n = 23) or identified through death certificates only (n = 6).

Statistical Analysis

Study participants were initially grouped according to 8 categories of BMI, as reported at the baseline interview (<18.5 kg/m2, 18.5-19.9 kg/m2, 20.0-21.4 kg/m2, 21.5-22.9 kg/m2, 23.0-24.4 kg/m2, 24.5-25.9 kg/m2, 26.0-27.4 kg/m2, and ≥27.5 kg/m2). These categories were created to allow a detailed examination of the association between BMI and colorectal cancer based on the distribution of BMI in the study population with the consideration of BMI cutpoints recommended by the World Health Organization (WHO) Working Group for Asian populations (Underweight, <18.5 kg/m2; Normal weight, 18.5-22.9 kg/m2; Overweight, 23.0-27.4 kg/m2; Obese, ≥27.5 kg/m2).13 We collapsed these categories further on the basis of similar age- and sex-standardized cancer rates (<18.5 kg/m2, 18.5-21.4 kg/m2, 21.5-24.4 kg/m2, 24.5-27.4 kg/m2, and ≥ 27.5 kg/m2) to improve the precision of the estimates and allow more detailed examination through interaction and stratification. This approach did not alter the results. For each study subject, person years were counted from the date of baseline interview to the date of cancer diagnosis, the date of death, date of last contact (for the few subjects who migrated out of Singapore), or December 31, 2007, whichever occurred first. Baseline characteristics were calculated for participants across each category of BMI. Age- and sex-standardized cancer rates were calculated using the person-year weight of the entire cohort during follow-up by the following age categories: <50 years, 50-54 years, 55-59 years, 60-64 years, 65-69 years, and ≥70 years.

Proportional hazards (Cox) regression methods were used to examine the associations between BMI and risk of colorectal cancer. All regression analyses were conducted by using SAS statistical software version 9.1 (SAS institute, Cary, North Carolina). We estimated the hazard ratio (HR) of colorectal cancer for levels of BMI and the corresponding 95% confidence interval (CI). There was no evidence that proportional hazards assumptions were violated as indicated by the lack of significant interaction between BMI and a function of survival time in the models. The referent BMI category was chosen on the basis of the lowest age- and sex-standardized cancer rate. Our primary Cox regression model included the following covariates: Age (<50 years, 50-54 years, 55-59 years, 60-64 years, and ≥65 years), sex, year of interview (1993-95 and 1996-98), dialect (Hokkien vs Cantonese), level of education (no formal schooling, primary school, secondary school, or above), baseline physician diagnosed diabetes mellitus (yes vs no), familial history of colorectal cancer (yes vs no), smoking (never smokers, light smokers, heavy smokers), alcohol intake (never/occasionally, weekly, daily), any vigorous work or strenuous physical activity (yes vs no), sleep (<6 or ≥9 hours per night vs 6-8 hours per night), quintile of vegetable, fruit, and soy-rich dietary pattern score, and total energy intake (Kcal/day). Hormone replacement use by women in the cohort was low (6%), as was use of vitamin supplements (6%) and use of nonsteroidal anti-inflammatory drugs (1%) regardless of sex.

The presence of a linear or quadratic (the U- or J-shaped curve) BMI-cancer association was examined by including a linear or linear and quadratic terms of the median BMI value within each BMI category in the Cox regression models. Separate product terms of smoking, age, and sex with BMI were included in Cox regression models to examine potential interactions. We also carried out a stratified analysis based on ever versus never smoking because the distribution of BMI and cancer rates differed between ever versus never smokers. To reduce potential bias due to pre-existing disease or illness-related weight loss, as well as accounting for the anorectic effects of cigarette smoking,22, 23 an analysis excluding subjects with reported history of diabetes at baseline, history of smoking, and colorectal cancer incidence within the first 5 years after enrollment was completed. Lastly, we examined the association by localized versus advanced cancer as determined by the Dukes staging system (A and B vs C and D).

RESULTS

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

The distribution of selected demographic and lifestyle factors by BMI are presented in Table 1. Of 51,251 participants included in the analysis, 980 developed colorectal cancers. The study population included 27,881 women and 431 cases. The mean age (and standard deviation [SD]) at entry into the study was 55.9 (SD, 7.9) years with mean follow-up time of 11.5 (SD, 2.9) years. The mean age at baseline for those diagnosed with cancer was 60.5 (SD, 7.8) years with mean follow-up time of 7.0 (SD, 3.5) years. The mean age at baseline was 55.8 (SD, 7.9) years with mean follow-up time of 11.6 (SD, 2.8) years for those who were not diagnosed with cancer.

Table 1. Participant Characteristics According to Body Mass Index (BMI, kg/m2) at Baseline
Characteristics<18.518.5-21.421.5-24.424.5-27.4≥27.5Ptrend
  • a

    Dietary pattern (top 20%) = dietary pattern score of greatest conformity in the population to a diet rich in vegetables, fruit, and soy foods.

  • b

    Any vigorous work or strenuous physical activity.

No.385013,70517,73010,6505316<.0001
Age57.255.955.755.955.8<.0001
Sex, % women51.854.254.053.559.6<.0001
Education, % secondary29.734.033.229.825.3<.0001
Diabetes mellitus, %4.16.39.011.214.0<.0001
Ever smoked, %41.332.528.827.626.6<.0001
Alcohol, % daily5.84.33.12.52.3<.0001
Dietary pattern, top 20%a18.920.220.020.120.0.47
Energy intake, kcal15421586158215811564.82
Any physical activity, %b12.815.416.215.313.0.30
Sleep, h7.07.07.07.07.0.10

Relative risks for colorectal cancer by BMI are given in Table 2 for all cases and by site. Individuals with BMIs of 21.5-24.4 kg/m2 had the lowest age- and sex-standardized incidence rate of colorectal cancer and colon cancer as well, whereas the lowest incidence rates of rectal cancer were seen in underweight (BMIs of <18.5 kg/m2) or obese persons (BMIs of ≥27.5 kg/m2). BMIs of ≥27.5 kg/m2 were associated with an increased risk of colorectal cancer. This association was driven by colon cancer, which had a strong, positive association with BMIs ≥27.5 kg/m2 and a suggestive increase in risk in BMIs <18.5 kg/m2 for an overall quadratic, U-shaped association (P = .014). There was no association between BMI and rectal cancer.

Table 2. Standardized Colorectal Cancer Rate and Hazard Ratios According to Body Mass Index (BMI, kg/m2) in the Singapore Chinese Health Study
 <18.518.5-21.421.5-24.424.5-27.4≥27.5Ptrend
 LinearQuadratic
  • Model was adjusted for age, sex, year of enrollment, dialect, education, diabetes status, familial history of cancer, smoking, alcohol intake, dietary pattern score, physical activity, sleep and energy intake.

  • Ptrend indicates value for linear and quadratic term (U- or J-shaped) using median of body mass index (BMI) in each category as continuous variable; HR, hazard ratio; CI, confidence interval.

  • a

    aStandardized rate is the age and sex standardized cancer rate per 100,000 person years using person year time, age, and sex distributions of he Singapore Chinese Health Study.

Colorectal cancer
 No. cases/N76/3850273/13,705318/17,730199/10,650114/5316  
 Standardized ratea177174156165188  
 HR (95% CI)1.03 (0.80-1.32)1.12 (0.95-1.31)1.01.05 (0.88-1.25)1.25 (1.01-1.55).44.23
Colon cancer
 No. cases5116218112379  
 Standardized ratea11910389103130  
 HR (95% CI)1.23 (0.90-1.68)1.17 (0.95-1.45)1.01.12 (0.89-1.43)1.48 (1.13-1.92).31.014
Rectal cancer
 No. cases251111377635  
 Standardized ratea5871676358  
 HR (95% CI)0.77 (0.50-1.19)1.04 (0.81-1.34)1.00.95 (0.71-1.25)0.93 (0.64-1.36).92.29

Relative risks for colorectal cancer by BMI stratified by smoking status are presented in Table 3. Overall, ever smokers had a higher incidence rate of colorectal cancer than never smokers. Among never smokers, there was a stronger quadratic U-shaped association between BMI and colon cancer. Compared with a BMI of 21.5-24.4 kg/m2, HR of colon cancer for BMI <18.5 kg/m2 and BMI ≥27.5 kg/m2 were 1.47 (95% CI, 0.99-2.20) and 1.59 (95% CI, 1.16-2.16), respectively (Pquadratic = .004). However, there was no association between BMI and rectal cancer. Among ever smokers, BMI was not associated with either colon or rectal cancer risk. Adjusting for duration and amount of cigarettes smoked did not alter the null association between BMI and cancer risk.

Table 3. Standardized Colorectal Cancer Rate and Hazard Ratios According to Body Mass Index (BMI, kg/m2) Stratified by Smoking in the Singapore Chinese Health Study
 Never-Smokers N=35,766Ptrend
 <18.518.5-21.421.5-24.424.5-27.4≥27.5LinearQuadratic
  • Model was adjusted for age, sex, year of enrollment, dialect, education, diabetes status, familial history of cancer, smoking, alcohol intake, dietary pattern score, physical activity, sleep and energy intake.

  • Ptrend indicates value for linear and quadratic term (U- or J-shaped) using median of body mass index in each category as continuous variable; HR, hazard ratio; CI, confidence interval.

  • a

    Standardized rate is the age and sex standardized cancer rate per 100,000 person years using person year time, age, and sex distributions of he Singapore Chinese Health Study.

Colorectal cancer
 No. cases/N39/2260153/9255189/12,632128/771680/3903  
 Standardized ratea149141128143177  
 HR (95% CI)1.17 (0.83-1.66)1.16 (0.94-1.44)1.01.08 (0.87-1.36)1.35 (1.04-1.76).46.056
Colon cancer
 No. cases301031158459  
 Standardized ratea115957894131  
 HR (95% CI)1.47 (0.99-2.20)1.29 (0.99-1.68)1.01.15 (0.87-1.53)1.59 (1.16-2.16).58.004
Rectal cancer
 No. cases950744421  
 Standardized ratea3446504946  
 HR (95% CI)0.70 (0.35-1.40)0.96 (0.67-1.38)1.00.98 (0.67-1.42)0.96 (0.59-1.56).63.42
Ever-Smokers N=15,485
Colorectal cancer
 No. cases/N37/1590120/4450129/509871/293434/1413  
 Standardized ratea221248230229219  
 HR (95% CI)0.87 (0.60-1.26)1.04 (0.81-1.33)1.00.98 (0.73-1.31)1.08 (0.74-1.58).61.78
Colon cancer
 No. cases2159663920  
 Standardized ratea126122118127129  
 HR (95% CI)0.94 (0.57-1.54)0.99 (0.69-1.41)1.01.06 (0.71-1.57)1.26 (0.76-2.08).35.71
Rectal cancer
 No. cases1661633214  
 Standardized ratea9612611210290  
 HR (95% CI)0.80 (0.46-1.40)1.09 (0.76-1.55)1.00.89 (0.58-1.37)0.90 (0.50-1.61).80.41

We examined the BMI/cancer association separately in men versus women. There was no evidence the association differed by sex (P = .60). Men did have higher rates per 100,000 person years follow-up across the spectrum of BMI compared with women (colon cancer rates men, 151, 113, 105, 120, 138; colon cancer rates women, 91, 95, 76, 80, 125) for BMI groups (<18.5 kg/m2, 18.5-21.4 kg/m2, 21.5-24.4 kg/m2, 24.5-27.4 kg/m2, and ≥ 27.5 kg/m2).

Because BMI is a less reliable marker of adiposity because of differential loss of muscle and bone mass in persons aged 65 years and older,23, 24 we examined the BMI/cancer association separately among subjects younger than 65 years versus those age 65 years or older. In 42,340 participants aged <65 years (384 colon cancer cases), an increased risk was observed in BMIs ≥27.5 kg/m2, (HR, 1.40; 95% CI, 1.01-1.93), and there was suggestive increase in BMIs <18.5 kg/m2 (HR, 1.25; 95% CI, 0.85-1.83) (Pquadratic = .05). In 8911 (212 colon cancer cases) participants aged ≥65 years, a stronger risk was observed in BMIs ≥27.5 kg/m2, (HR, 1.66; 95% CI, 1.05-2.62), and a similar suggestive risk was observed in BMIs <18.5 kg/m2 (HR, 1.21; 95% CI, 0.71-2.06). There was no evidence the BMI/cancer association differed between the 2 age groups (P = .37). Greater rates of cancer also were observed in those aged ≥65 years (data not reported).

Figure 1 presents the results of a model examining colon cancer risk in participants who never smoked, were free of diabetes at baseline, and did not develop colon cancer in the first 5 years of follow-up. BMIs <18.5 kg/m2 and ≥27.5 kg/m2 were associated with an increased risk of colon cancer, and the point estimates were strengthened compared with the main analysis with a stronger quadratic U-shaped association (P = .002). Figure 2 presents data for the same set of participants as Figure 1 and considers localized versus advanced colon cancers. There is a significant U-shaped quadratic association in both groups with more pronounced point estimates for BMIs ≥27.5 kg/m2 for localized colon cancer and BMIs <18.5 kg/m2 for advanced colon cancer.

thumbnail image

Figure 1. Displayed are hazard ratios of colon cancer in never smokers, without reported diabetes, and excluding persons with less than 5 years of follow-up time (N = 31,851 and 236 cases colon cancer). Model is adjusted for age, sex, year of enrollment, dialect, education, familial history of cancer, alcohol intake, dietary pattern score, physical activity, sleep, and energy intake. Points represent hazard ratio (HR) point estimate, and error bars represent 95% confidence intervals. Pquadratic for association is.0016. Respective case counts for colon cancer by BMI are BMI, <18.5 (21 cancer cases); BMI, 18.5-21.4 (65 cancer cases); BMI, 21.5-24.4 (67 cancer cases); BMI, 24.5-27.4 (48 cancer cases); BMI, ≥27.5 (37 cancer cases).

Download figure to PowerPoint

thumbnail image

Figure 2. Hazard ratios of colon cancer in never smokers, without reported diabetes, and excluding persons with less than 5 years of follow-up time are stratified by staging level (local vs advanced) of cases (Duke A or B for local and Duke C or D for advanced). Duke A or B includes N = 31,670 and 128 cases localized colon cancer. Duke C or D includes N = 31,706 and 120 cases advanced colon cancer. Model is adjusted for age, sex, year of enrollment, dialect, education, familial history of cancer, alcohol intake, dietary pattern score, physical activity, sleep, and energy intake. Points represent hazard ratio (HR) point estimate, and error bars represent 95% confidence intervals. Pquadratic for association is.003 for localized and P = .045 for advanced.

Download figure to PowerPoint

Last, we examined weight change in relation to colon cancer in 27,602 (159 cases of colon cancer) never smokers, without a history of diabetes and free of cancer at follow-up I (interviewed 1999-2004), who also provided a weight at follow-up I. We did not find significant weight gain (≥ a gain of 10% body weight) or loss (≥ a loss of 10% body weight) during approximately 6 years to be associated with colon cancer incidence (data not presented). On the other hand, the U-shaped association between baseline BMI and colon cancer risk persisted (Pquadratic = .0003; HR, 1.88; 95% CI, 1.04-3.39; BMI, <18.5 kg/m2) and (HR, 2.23; 95% CI, 1.39-3.59; BMI, ≥27.5 kg/m2) after further adjustment for percentage of weight change.

DISCUSSION

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

In this large prospective cohort of Chinese men and women in Singapore, we observed a quadratic U-shaped association between BMI and incidence of colon cancer with increased risk at BMIs ≥27.5 kg/m2 and BMIs <18.5 kg/m2. The association was more pronounced in never smokers and most prominent when we further limited the sample to those free of diabetes and cases with more than 5 years of follow-up. The quadratic association held when considering localized versus advanced cases with more pronounced associations in BMIs ≥27.5 kg/m2 for localized cases and BMIs <18.5 kg/m2 for advanced cases. There was no association between BMI and rectal cancer. Stratified analyses by sex and age provided no evidence of differential risk, although men had greater rates of colorectal cancer, and rates were significantly greater in those aged 65 years or more.

Our results are not consistent with the small number of prospective studies on this topic in Asian populations. In approximately 29,000 Japanese men and women, a significant association in men (RR, 2.11; 95% CI, 1.26-3.53) was observed for colon cancer in comparison with the top tertile of BMI (>23.6 kg/m2) to the first tertile (<21.6 kg/m2), whereas no association was observed in women or either sex with rectal cancer.12 No comparison with underweight BMIs (<18.5 kg/m2) was reported. In another Japanese cohort, BMIs >25.0 kg/m2 were associated with an increased risk of colon cancer in men11 relative to BMIs <25 kg/m2. No association was observed in women for colon cancer or in rectal cancer for either sex. Furthermore, BMIs <25 kg/m2 were not categorized or specifically examined in relation to colon cancer. Kuriyama et al examined more than 27,000 middle-aged Japanese women and men and found no significant association between BMI and colon or rectal cancer.9 In this Japanese study, the BMI data were grouped and analyzed according to Western cutpoints, and the majority of participants as well as cases were found in the referent category (BMI, 18.5-24.9 kg/m2) giving small case numbers in the specific cancer sites in higher BMIs. Furthermore, almost 1600 participants were excluded with a BMI <18.5 kg/m2, so the low end of the BMI spectrum was not examined.

Two different studies using data from a national insurance registry in Korea examined the association in men,10 and in both sexes.8 Oh et al10 observed increased risks for colon cancer in BMIs of 23-26.9 kg/m2 and ≥30 kg/m2 relative to BMIs of 18.5-23.0 kg/m2 and no association in BMIs <18.5 kg/m2. When the analysis was limited to never smokers, the association for colon cancer was only significant in BMIs of 25-26.9 kg/m2. Never smokers with a BMI >27.0 kg/m2 also were at increased risk for rectal cancer. A possible reason the findings differ relative to our study is that the age range in the Korean study was more heterogeneous because men aged 20 years and older were included. Thus, a number of participants were younger and had very low risk of colon or rectal cancer. They also likely had a general difference in body composition at the same BMI level from middle-aged and older men who were at greater risk for the cancer. Furthermore, the authors report it was not clear whether cancer ascertainment was complete. If cancer ascertainment differed by level of BMI, this could influence the results. Jee et al8 observed inverse associations for colon cancer in men aged 30-95 years at BMIs <20.0 kg/m2 and between 20-22.9 kg/m2 and positive associations at BMIs of 25-29.9 kg/m2 and ≥30 kg/m2 compared with BMIs between 23-24.9 kg/m2 after adjustment for age and smoking habits. An inverse association for rectal cancer in men was observed in BMIs <20.0 kg/m2 and 20-22.9 kg/m2. In the same analysis in women,8 there was no association between BMI and rectal cancer, and there was an inverse association with colon cancer in BMIs <20.0 kg/m2 and between 20-22.9 kg/m2. BMIs <18.5 kg/m2 were not directly examined because of categorical grouping. Similar to the other Korean-based population study, the wide age range may be a reason for the different shape and relation between BMI and colorectal cancer of our study population from those of Korean study populations.

Numerous studies examined BMI in relation to incident colorectal cancer in Western populations. A recent meta-analysis25 on this topic presented a monotonic increase in risk of colon cancer with increasing BMIs >23 kg/m2 relative to BMIs <23 kg/m2. Of interest, the magnitude of risk estimate for colon cancer for obesity (BMI, ≥30 kg/m2) was similar to that for BMIs ≥27.5 kg/m2 in Chinese Singaporeans. The meta-analysis did not present the association between underweight (BMI, 18.5 kg/m2) and colon cancer risk because the lowest BMI category presented was <23.0 kg/m2. Similar to the studies in Asian populations, the majority of studies conducted in Western populations that were included in the meta-analysis25 also combined <18.5 kg/m2 with normal BMI range (<23 kg/m2) or excluded them from their analyses. Thus, the data do not refute or support any association of underweight or low BMI with colon cancer.

The exact mechanisms that link excess weight with colon cancer are uncertain. The majority of research points toward involvement of the insulin and insulin-like growth factor axis. Briefly, the insulin-cancer hypothesis postulates that chronic hyperinsulinemia decreases concentrations of IGF-binding protein-1 and IGF-binding protein-2, which leads to higher bioavailability or free IGF-1 with cellular changes favoring tumor formation.26 Circulating total IGF-1, a major determinant of free IGF-1 concentrations, is considered a risk factor for colorectal cancer as well.27, 28 However, total IGF-1 levels are lower in underweight and overweight persons relative to normal weight persons.26, 29 Sex hormones and adipokines also have received significant attention for their potential role mechanistically between excess weight and cancer.30 Indeed, there are important limitations to consider with these hypotheses as well as other hypotheses and developing areas of knowledge.30

Similarly, the mechanisms that may link underweight with an increased risk of colon cancer are unclear. Leanness as a risk factor for cancer of different organs has been noted.31, 32 In accordance with the leanness-cancer hypothesis is oxidative DNA stress.33 In a longitudinal study of BMI and a marker of oxidative DNA damage, decreasing levels of BMI, regardless of smoking status, were associated with significantly increased levels of DNA oxidative damage.34 Low-grade inflammation has been found to occur in persons with low BMIs35 and has been noted as a risk factor for colon cancer independent of BMI36 and as a stronger risk factor in lean versus heavy persons.37 The ability of low-grade inflammation to compromise the immune system and thus affect cancer risk is another consideration.38 Cancer screening practices are another consideration as data have shown underweight persons are more likely to delay screening relative to normal-weight persons.39 Cultural hurdles to screening in this population may be an even larger underlying issue.3

Strengths of our study include the assumption that colorectal cancer case ascertainment was complete given that Singapore is a small city-state where there is thorough specialized medical care available, and the nationwide cancer registry has been in place since 1968 and has been shown to be comprehensive in its recording of cases.1 Furthermore, in our comprehensive analysis, we were able to account for many known potential modifiers and confounders of the association between BMI and colorectal cancer. A large sample size and ample amount of events combined with a long follow-up time are other strengths as is the ability to account for local versus advanced case status.

Limitations include the use of self-reported height, weight, and other demographic and lifestyle data. Additional data on other measures of body habitus may complement BMI in this population and contribute to further understanding. As well, multiple assessments of relative weight may offer further insight on the topic. Furthermore, despite thorough adjustment for smoking, alcohol, dietary patterns, activity, and sleep, residual confounding and unmeasured confounding need to be considered in the interpretation. We also did not have information on cohort-participant cancer-screening practices.

In summary, we found a U-shaped, quadratic association between BMI and colon cancer and no association of BMI with rectal cancer. The colon cancer association did not differ by sex or age, but greater rates were observed in men and in those aged 65 years and older. The association was more pronounced in never smokers, and the association in ever smokers was null. Smokers, however, did have greater rates of colorectal cancer and a differing distribution of BMI. Furthermore, localized cases had a more pronounced association in BMIs ≥27.5 kg/m2, whereas advanced cases had a more pronounced association in BMIs <18.5 kg/m2. Our comprehensive study is not in general agreement with other prospective studies in Asian populations in terms of the range of BMI where risk begins to increase; however, different methods and analytic approaches in the few other studies preclude any firm conclusions. Our results do generally align with Western populations showing an increased risk of colon cancer with population-specific obesity; however, our findings of an increased risk with underweight are novel. Further research is needed to examine and understand the association between BMI, colon, and rectal cancers in specific Asian populations and any potential mechanisms involved. Our study especially highlights the need for further investigation into the association between underweight status and colon cancer.

CONFLICT OF INTEREST DISCLOSURES

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

This study was supported by National Institutes of Health grants NCI RO1 CA055069, R35 CA053890, R01 CA080205, R01 CA098497, and R01 CA144034.

REFERENCES

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