The objective of this study was to conduct a systematic review and analyze the association between tobacco smoking and CRC from published papers during the previous five years. All published cohort studies within the last five years using specific keywords were reviewed. The title and abstract of all available papers were reviewed and considered for eligibility inclusion.
Trang 1Corresponding author: Le Tran Ngoan, Hanoi Medical
University
Email: letngoan@hmu.edu.vn
Received: 15/4/2018
Accepted: 22/11/2018
THE ASSOCIATION BETWEEN TOBACCO SMOKING
AND COLORECTAL CANCER: A META ANALYSIS
1
Nguyen Thi Nga, 2 Pham Phuong Lien, 3 Khanpaseuth Sengngam, 4,5 Le Tran Ngoan
1 Vinh Medical University, Vietnam; 2 Hanoi University of Public Health, Vietnam;
3
National Institute of Public Health, Lao PDR;
4
International University of Health and Welfare, Japan; 5 Hanoi Medical University, Vietnam Cigarette smoking is recognized as the cause of a number of diseases including cancer, however, previ-ous findings of its relation to colorectal cancer (CRC) are inconsistent The objective of this study was to conduct a systematic review and analyze the association between tobacco smoking and CRC from published papers during the previous five years All published cohort studies within the last five years using specific keywords were reviewed The title and abstract of all available papers were reviewed and considered for eligibility inclusion The ln(HR) and se(ln(HR)) were estimated from the multivariable adjusted
HR and the 95% confidence interval (CI) was derived from published studies The random pooled multivariable adjusted HR and 95%CI was analyzed using STATA 10 There were 20 studies included for pooled analysis The test for heterogeneity yielded Q = 128.044 on 22 degrees of freedom (p = 0.000) Moment-based estimate of between studies variance = 0.021 HR = 1.16; CI (1.08 - 1.27), statistically significant, p < 0.01 We observed a significant positive association between tobacco smoking and the risk of colorectal cancer
Key words: CRC, cigarette smoking, cohort study, meta-analysis
I INTRODUCTION
Cigarette smoking has been recognized as
the cause of a number of diseases including
cancer [1] Annually, while active smoking kills
more than five million people, secondhand
smoking (SHS) causes the death of over
600,000 people worldwide [2] If the situation
is not controlled, deaths due to tobacco use
will reach eight million per year by 2030 The
vast majority of these deaths are projected to
occur in the developing world, including
Vietnam [3] Tobacco use was also associated
with a high burden of diseases from
non-communicable diseases such as cardiovascular
diseases and diabetes Despite these negative health effects, the prevalence of tobacco smoking has been increasing globally in recent years, particularly among youth in low and middle income countries [4 - 6] Colorectal cancer incidence and mortality has been the
approximately 1.4 million new cases and 694,000 deaths were estimated to have occurred in 2012 alone [7] Although the involvement of cigarette smoking in the development of colorectal cancer has been reported in some studies, evidence of the association between tobacco smoking and colorectal cancer risk is still unclear [8; 9] To our knowledge, no literature review has been conducted on the association between
Trang 2tobacco smoking and colorectal cancer since
our earlier review in 2013 We aim to review
the association between tobacco smoking and
CRC from published papers during the
previous five years
II METHODS
To further investigate the controversial
relationship between cigarette smoking and
CRC, we conducted a review of all published
cohort studies within the last five years The
search process was conducted in January
through August of 2017 using PubMed with
the keywords: (smoke OR cigarette OR
tobacco OR smoking) AND (Colon cancer OR
Rectum cancer OR colo-rectal cancer OR
colorectal cancer OR colorectum cancer OR
colon rectum cancer) AND cohort studies)
The studies were collected and handled in
two stages In the first stage, the title and
abstract of all collected researches were
reviewed Studies not related to cigarette
smoking and CRC were excluded Studies
matching the selection criteria were stored as
full text and were moved to the second stage
At this stage, we proceeded to read and check
the results and methodology of the studies
Studies related to the association of cigarette
smoking and CRC published from 2013 until
the present were selected For studies that
published data from the same cohort, we
chose only the most recent and complete
report for analysis
- Patients were prospectively recruited and
followed up
- Studies reported relative risk (RR) or
hazard ratios (HR) and their corresponding
95% confidence intervals (95% CIs) of CRC or
some other factors effecting CRC status by
different smoking categories
• At least one of the outcomes (colon, rectal, or CRC) was reported
Inclusion criteria
- Patients were prospectively recruited and followed up
- Studies reported relative risk (RR) or hazard ratios (HR) and their corresponding 95% confidence intervals (95% CIs) of CRC or some other factors effecting CRC status by different smoking categories
- At least one of the outcomes (colon, rectal, or CRC) was reported
Exclusion criteria
- Case-control design
- Studies that included hereditary CRC
disease, history of colorectal cancer, or previous bowel resection
- Full publication not written in English Data of all studies were extracted and arranged into a formation for analyzing and evaluating The characteristics extracted include:
- Basic information: Name of author, conducted year, published year, setting
- Detailed information: Subject, gender, person at risk, type of CRC
- Research results: Incidence or mortality, smoking category, cigarettes per day, smoking duration, pack-year, initiate age, RR, HR adjusted
The primary outcome of this study was the incidence of CRC (International Classification
of Disease [ICD] versions 7 - 9: 153 - 154; ICD 10: C18 - 21) Secondary outcomes included incidence of colonic cancer (ICD 7 - 9: 153; ICD 10: C18 - 19) and rectal cancer (ICD 7 - 9:
Trang 3154; ICD 10: C20 –21) The cancer diagnosis
was identified through hospital records,
pathology reports, or cancer registry All
studies used were published and data can be
used for researching purposes All the
information collected was kept confidential and
was only available for research purposes
Data synthesis and analysis
The ln(HR) and se(ln(HR)) were estimated
from the multivariable adjusted HR, 95% CI
derived from published prospective studies
The random pooled multivariable adjusted HR,
95% CI was analyzed using STATA 10
III RESULTS
We identified eligible 400 abstracts from
the initial literature search After screening and
excluding duplicate abstracts, 20 articles were
considered of interest and full texts were
retrieved for detailed evaluation The present
study included 20 cohort studies with data
from a total of 6.302.836 participants Six
populations, eight in the Asian Pacific
populations and six in European populations
(Figure 1) All studies were conducted and followed up between 1972 and 2013 Most of the articles were published in regional or world cancer magazines In 2016 and 2017, only one study was published while four were published in 2014 Six studies were published
in 2015 and eight were published in 2013 Sixteen studies included CRC, three studies included colon cancer only and one included only rectal cancer (Table 1) Five studies indicated cases of CRC deaths but only four studies described hazard ratios of colorectal cancer mortality for current smokers (Table 2)
In two studies of Ahmadi et al and Tao L et al,
colorectal cancer-specific mortality [10; 28] whereas two studies were not associated with colorectal cancer specific mortality [10; 15] In
a study of Jang B et al, multivariable-adjusted Cox proportional hazards regression models showed that smoking before diagnosis was associated with colorectal cancer-specific mortality (RR, 2.14; 95% CI, 1.50 to 3.07) and post-diagnosis smoking was associated with colorectal cancer-specific mortality (RR, 1.92; 95% CI, 1.15 to 3.21) [12]
Trang 4Records identified through
Records excluded (n = 362)
Records after duplicates removed
(n = 243)
Records screened (n = 400)
Studies included in qualita-tive synthesis (n = 20)
Studies included in quanti-tative synthesis (Meta-analysis), (n = 18)
Figure 1 Flowchart of systematic literature search and review for eligible studies
Table 1 The title and author in included studies
1
Behavioural and Metabolic Risk Factors for Mortality from
Colon and Rectum Cancer: Analysis of Data from the
Asia-Pacific Cohort Studies Collaboration [10]
Morrison DS et al
participants in the EPIC-PANACEA study [12]
Steins Bisschop BN
et al
Trang 5Number Name of studies and [source] Author
5 Fruit and vegetable intake and the risk of colorectal cancer:
6 Lifestyle factors associated with survival after colorectal
9 Proportion of Colon Cancer Attributable to Lifestyle in a
14
Associations between Environmental Exposures and Incident
Colorectal Cancer by ESR2 Protein Expression Level in a
Population-Based Cohort of Older Women [23]
Tillmans LS et al
17
Hypertension is an important predictor of recurrent colorectal
adenoma after screening colonoscopy with adenoma
polypec-tomy [26]
Lin CC et al
20
A Prospective Study of Duration of Smoking Cessation and
Colorectal Cancer Risk by Epigenetics-related Tumor
Classifi-cation[29]
Nishihara R et al
Trang 6Table 2 Hazard ratio of colorectal cancer mortality for smoking status
status
HR (Multivariate Adjusted)*
Never smokers = 1
*HR adjusted by many variables depending on the research including age, BMI, physical activ-ity, height, drink, smoke, cholesterol, diabetes and education were included in the sex and study stratified model
Table 3 HR of Colorectal Cancer Incidence for smoking status
Current smokers: HR = 1.28, (1.00 - 1.63) Former smokers: HR = 1.10, (0.97 - 1.24)
Male ever-smoker: HR = 1.08, (0.97 - 1.19)
current smokers
Ex-smokers: HR = 1.34(0.52 - 3.46) Current smokers: HR = 0.51(0.18 - 1.38)
Female ever-smokers: HR = 1.28, (1.11 - 1.48)
Trang 7There were 12 studies which indicated the
incidence of CRC but only 4 studies described
HR indicator for different types of smoking
status (Table 3)
Our study included 20 studies that met the
criteria, including 18 studies showing the
asso-ciation of smoking and colorectal cancer with
HR and RR, although each study presented
many different RR and HR indicators
Therefore, the author conducted a selection of
the lowest correlation indicators for the
combined study Of the 18 studies, 15 studies
included HR for both colorectal cancer in
general; one for colon (colon, proximal, distal);
one for result of three types of CRC (colon,
rectal and CRC) and one for gender (both for
Table 4 Combined analyses results of fixed and random methods
Method Pooled
estimation 95% confidence interval
analyses z-value p-value
Test for heterogeneity: Q = 128.044 on 22 degrees of freedom (p = 0.000) Moment-based
selecting 23 indices from 18 studies, the pooled estimation from the Random methods were simi-lar (statistically significant) After a meta-analysis of 23 indicators of 18 studies, it was concluded that smoking increases the risk of colorectal cancer
female and males) such that 23 variables were analyzed together In these 23 variables, there were two relatively low risk and statistically significant outcomes while there were seven
relationship Fourteen results suggested that smoking increases the risk of colorectal cancer These results were inconsistent and the data was put into the Stata analysis table This result evaluated the dispersion of data sets and it can
be seen that the studies had a high dispersion (p < 0.000) (Figure 2) Specifically, the results
of the analysis by the random method was included in pooled estimation, odds ratio (hazard ratio) = 1.16; CI (1.08 - 1.27) and p < 0.000, (Figure 2, Table 4)
Trang 8IV DISCUSSION
The research of David Stewart Morrison et
al and T Boy et al found no convincing
relationship between smoking and colorectal
cancer mortality [10] and the remaining three
studies found no association between current
smoking and survival in colorectal cancer
patients [30 - 32] A recent meta-analysis of
Liang et al reported that current smokers had
higher colorectal cancer mortality compared
with never-smokers, but the absence of any
significant association between former
smok-ers and colorectal cancer mortality or between
smoking and site specific cancer mortality
suggested that further research was needed
[33] The research of Baiyu Yang et al is one
of the largest studies of smoking and
colorec-tal cancer survival and the first study to prospectively collect both pre- and post-diagnosis smoking information In this cohort study of colorectal cancer survivors, smoking before or after cancer diagnosis was associ-ated with higher risk of mortality resulting from colorectal cancer [34] According to a recent meta-analysis from Walter et al, smoking is associated with poorer long-term prognosis after colorectal cancer diagnosis Specifically, the risk of all-cause mortality was higher for current smoking at all time points (HR, 1.26; 95% CI, 1.15 to 1.37) [35] We found a greater than two-fold risk of all-cause mortality for both pre- and post-diagnosis smoking compared with never smoking and lower though still
sta-Figure 2 Combined estimation of 23 studies
Trang 9tistically significant associations with both
pre- and post- diagnosis former smoking This
result is similar to the research of Tao L et al
[28] Only six other studies have examined the
association between smoking and colorectal
cancer–specific mortality[15; 36 - 40]; of these,
two studies with sample sizes comparable to
ours [37; 38] found current smoking to be
as-sociated with significantly higher colorectal
cancer–specific mortality, consistent with our
results However, the previous RRs were
lower than the RRs in our study, with
pre-diagnosis smoking associated with an RR of
1.30 in a study of patients with colorectal
can-cer in Washington state and an RR of 1.46
among patients with colon cancer in a large
US cohort [37; 38] Another study found a
greater than two-fold higher risk of colorectal
cancer–specific mortality comparing current
smokers with former or never-smokers
com-bined, and the remaining three studies found
no association between pre- and
post-diagnosis current and ever smoking with
colo-rectal cancer–specific mortality; however,
these analyses were based on relatively
smaller sample sizes [15; 36; 39; 40] The
study of Ali Ahmadi et al illustrated that
smok-ing increased the risk of death in these
pa-tients, which is consistent with a study in the
United States that reported smoking increased
the mortality risk after CRC diagnosis [37]
The colorectal cancer risk estimated for
smoking status from the study of Hurley et al
(HR = 1.28 for current smokers; HR = 1.10 for
former smokers) [13] is consistent with
find-ings from a number of recently published meta
-analysis on this topic in which summary
measures of risk have ranged from 1.12 to
1.26 for current smokers and 1.18 to 1.20 for
former smokers [41 - 45] The marginally lower
risk estimate for former smokers in the current study is likely a reflection of the fact that nearly half of the former smokers in our study popula-tion quit smoking more than 20 years before joining the cohort, by which time their risk ap-pears to no longer be elevated Interestingly, the most recent and one of the largest studies conducted to date reported no association between age at smoking initiation and colorec-tal cancer risk among members of the EPIC cohort [46] In a recent meta-analysis, Liang reported that for each 10-year delay in smok-ing initiation, there was a 4.4% reduction in risk ratios for colorectal cancer [33]
The degree to which smoking-related colo-rectal cancer risks are similar among men and women has been a matter of debate Initially, the preponderance of data seemed to suggest that the effect of smoking was either limited to,
or at least stronger, among men than among women [47] Explanations offered for this ap-parent difference have included both limita-tions in exposure potential (given the apparent long latency) as well as real sex-related bio-logic differences potentially arising from differ-ential interactions between smoking and pro-tective endogenous estrogens, body mass index, and/or abdominal adiposity [42] Two recent meta-analyses of prospective cohort studies on this topic reported that risks for cur-rent smoking continued to be higher among men than among women [42; 43], although only one found these differences to be statisti-cally significant at the 0.05 level [42] In con-trast, a meta-analyses that included both co-hort and case–control studies published during the same time period reported no evidence for differences in risk by sex [41] More recent findings from the European Prospective Inves-tigation Into Cancer [46]and the Cancer
Trang 10Pre-vention Study II [9], both of which reported no
differences in risk by sex, were not included in
these meta-analyses Regardless of whether
risks are higher in men than in women, there
is now convincing evidence that risks are
ap-parent in women Along with the elevated risks
found in this study and those reported among
the female participants in the EPIC and CPS-II
cohorts, elevated risks also have been
re-ported among members of the Norwegian
Women and Cancer Study [48] and the
Women's Health Initiative [49], both large
well-conducted prospective cohort studies among
women The Norwegian study, however, only
observed an effect for rectal but not colon
can-cer, a finding that also was reported among
members of the Canadian Breast Screening
Study over 10 years ago [50]
The meta-analysis of Botteri et al in 2008
which analyzed one hundred and six
observa-tional studies found that cigarette smoking is
significantly associated with colorectal cancer
incidence and mortality but the association
was stronger for cancer of the rectum than of
the colon [41]
The meta-analysis of Tsoi et al included 28
prospective cohort studies in 2009 showing
that smoking was associated with a
signifi-cantly increased risk of CRC Current smokers
had a modestly higher risk of CRC than never
smokers and former smokers still carried a
higher CRC risk than never smokers In
addi-tion, the associated risk was higher for men
and rectal cancers and the increased risk of
CRC was related to cigarettes per day, longer
years of smoking, or larger pack years [42]
Another meta-analysis including 103
co-hort studies of Huxley et al in 2009 indicated
that smoking may be a lifestyle factor
associ-ated with a significant increased risk of colo-rectal cancer [44] but the meta-analysis of Constance M Johnson et al in 2013 indicated that cigarette smoking was associated with moderately increased risk of CR (RR = 1.06, 95% CI: 1.03 - 1.08 for 5 pack- years) [51]
Limitations
To our knowledge, no literature review has been conducted on the association between tobacco smoking and colorectal cancer since our earlier review in 2013 The confidence in the effects estimates in review is affected by a number of limitations Indeed, we only de-scribed the results of the selected studies by using a sensitive search strategy in Pub Med and conducting screening and data extraction independently and in duplicate Most studies
do not give a precise percentage of the num-ber of smokers, and the groupings of smoking status are different and depend on the re-search questions asked by the authors Some of the studies that produce the HR index have been calibrated but are corrected
by different factors so we were not able to conduct a meta-analysis for all outcomes One reason was the high level of heterogeneity, as was the case for the quality of life outcome Another reason was that we could not pool several outcomes derived from the same study, different smoking status, duration and type of CRC The study results do not cover other life style factors and information on mo-lecular subtypes Additional studies of our find-ings include the need for further research on this topic by conducting more cohort studies to clearly determine the effects of smoking status
on the types and stages of colorectal cancer and the factors that can be combined