Somatic alterations, metabolizing genes, and smoking in rectal cancer

Abstract

Cigarette smoking has been identified as a risk factor for rectal cancer. Our investigation evaluates associations between active and passive smoking and TP53, KRAS2, and BRAF V600E mutations, microsatellite instability (MSI), and CpG Island Methylator Phenotype (CIMP) in rectal tumors. We examine how genetic variants of GSTM1 and NAT2 alter these associations in a population-based, case-control study of 750 incident rectal cancer cases and 1201 controls. Detailed tobacco exposure data were collected in an extensive questionnaire. DNA from blood was examined for GSTM1 and NAT2 variants. Tumor DNA was assessed to determine TP53 (exons 5–8), KRAS2 (codons 12–13) and BRAF mutations, MSI (BAT26 and TGFβRII analysis), and CIMP (methylation of CpG islands in CDKN2A, MLH1, MINT1, MINT2, and MINT31). Cigarette smoking (>20 pack-years, relative to non-smokers) was associated with increased risk of TP53 mutations (OR=1.4, 95%CI 1.02–2.0), BRAF mutations (OR=4.2, 95%CI 1.3–14.2), and MSI (OR=5.7, 95%CI 1.1–29.8) in rectal tumors. Long-term environmental tobacco smoke (ETS) exposure of >10 hours/week was associated with increased risk of KRAS2 mutation (OR=1.5, 95%CI 1.04–2.2). All smoking indicators were suggestive of increased risk in CIMP+ rectal cancer. GSTM1 and NAT2 were generally not associated with rectal tumor alterations; however, we observed an interaction of ETS and NAT2 in TP53-mutated tumors (p<0.01). Our investigation shows active smoking is associated with increased risk of TP53, BRAF, and MSI+ in rectal tumors and is suggestive of increased risk of CIMP+ tumors. ETS may increase risk of KRAS2 mutations; association with TP53 mutations and ETS may be influenced by NAT2.

INTRODUCTION

Cigarette smoking, both active and passive, have been identified as risk factors for rectal cancer in case-control and cohort studies^1–6. Recent smokers and those that have smoked many pack-years of cigarettes have been estimated to have from a 50% increase to a doubling of risk of rectal cancer compared to non-smokers^1–4. Long-term exposure to environmental tobacco smoke (ETS) was reported to increase risk of rectal cancer by 50% in a large case-control investigation¹; other cohort studies have observed an association of ETS and rectal cancer in men⁴ and in women⁶.

The association between cigarette smoking and rectal cancer may be impacted by the ability to detoxify polycyclic aromatic hydrocarbons (PAH) generated from cigarette smoking. Genetic variants in glutathione S-transferases (e.g. GSTM1) or N-acetyltransferase (NAT2), Phase II metabolizing enzymes, may be important components of an individual’s ability to detoxify carcinogens^7–9. It has previously been shown that among recent cigarette smokers, the GSTM1 null phenotype was associated with an elevated risk of rectal cancer,^{1, 10} and we observed that the NAT2-imputed rapid acetylators were at slightly lower increased risk of rectal cancer than slow acetylators¹. Previous studies of colon cancer have reported the NAT2 rapid-acytelator imputed phenotype is associated with slightly higher risk for different measures of smoking status^{11, 12}.

It has been suggested that different types of TP53 mutations observed in Japanese and Western populations may result from different exposures and also may be influenced by germline polymorphisms involved in carcinogen metabolism¹³. In our previous investigation of colon cancer, cigarette smoking was associated with increased risk of TP53 transversion mutations among those who were GSTM1 positive¹⁴. We have also reported a direct association between smoking and microsatellite instability (MSI), CpG Island Methylator phenotype (CIMP) and BRAF V600E mutations in colon tumors^{15, 16}. In this population-based case-control study of 750 incident rectal cancer cases with tumor DNA and 1201 sex- and age-matched controls, we evaluate the relationship of cigarette smoking and ETS to somatic alterations (TP53 mutation, KRAS2 mutation, CIMP, BRAF V600E, and MSI) in rectal tumors. Given our previous work, we also assess whether genetic variants of GSTM1 and NAT2 metabolizing enzymes influence risk associated with specific types of rectal tumor mutations.

MATERIALS AND METHODS

Study population

Participants were from the Kaiser Permanente Medical Care Program (KPMCP) of Northern California and the state of Utah. Cases with a first primary tumor in the rectosigmoid junction or rectum were identified between May 1997 and May 2001. Case eligibility was determined by the Surveillance Epidemiology and End Results (SEER) Cancer Registries in Northern California and Utah. Case eligibility included: no previous colorectal tumor, between ages 30 and 79 years at diagnosis, English speaking, and mentally competent to complete the interview. Cases with known (as indicated on the pathology report) familial adenomatous polyposis, ulcerative colitis, or Crohn’s disease were not eligible. Using the same case eligibility criteria, controls were selected to match cases by sex and 5-year age cohort. Additionally, controls could not have had a previous colon or rectal cancer. The study response rate was 65.2% for cases and 65.3% for controls¹.

Data were collected by trained and certified interviewers using laptop computers. The self-reported race/ethnicity of study participants was 82% white and non-Hispanic, 7.6% Hispanic, 4.6% Asian, 4.1% African American, 0.7% Native American, and 1% multiple races/ethnicities. The referent period participants were asked to recall was the year that occurred two years before the date of case diagnosis or control selection.

Cigarette smoking history and long-term exposure to passive cigarette smoke both inside and outside the home (ETS) were collected¹. Participants were asked if they ever smoked on a regular bases, or at least 100 cigarettes in their lifetime. Those who answered ‘yes’ were asked the usual number of cigarettes smoked per day while smoking, ages started and stopped smoking, and total number of years smoked. Pack-years of cigarettes smoked was determined by multiplying the usual number of cigarettes smoked per day by total years of smoking cigarettes, and dividing by 20 or a pack of cigarettes. Exposure to ETS was obtained by asking the usual number of hours per week of exposure in the home and outside the home to cigarette smoke of others, for the referent period and 10 and 20 years ago¹. Other information obtained in the interview included physical activity performed over the past 20 years, reported height and weight during the referent year, long-term alcohol use as well as detailed alcohol consumption during the referent year, family history of cancer, use of aspirin and nonsteroidal anti-inflammatory drugs, and a detailed dietary intake.

Genetic analysis

Tumor DNA was obtained from paraffin-embedded tissue as described¹⁷. Tumors were characterized by their genetic profile that included: sequence data for exons 5 through 8, the mutation hotspots of the TP53 gene; sequence data for KRAS2 codons 12 and 13; methylation specific PCR of sodium bisulfite modified DNA for five CpG Island markers, CDKN2A (p16), MLH1 and methylated in tumors (MINT) 1, 2 and 31 ¹⁸; the V600E BRAF mutation by TaqMan assay¹⁹; and MSI by instability in the mononucleotide repeat BAT26 (for ~95% of tumors) and by a coding mononucleotide repeat in TGFβRII (for ~5% of tumors in which BAT26 failed).

Tumors with two or more methylated CpG islands were scored as CIMP+. At this time, there is no “consensus” as to the appropriate CpG island panel or method of detection to determine CIMP. However, we have used our panel to demonstrate significant relationships between CIMP and numerous clinicopathologic variables, including cigarette smoking and the BRAF V600E mutation, which were independent of microsatellite instability^{16, 18}. This work has also helped to support the legitimacy of the CIMP concept²⁰.

Our MSI assessment preceded the Bethesda consensus panel; however, BAT26 by itself is a very good measure of generalized instability ²¹, and we have shown high correlations between instability in BAT26 and TGFβRII and the Bethesda consensus panel^{22, 23}. In the assessment of MSI, germline DNA obtained from normal tissue in the paraffin blocks or from blood samples provided by study participants was used to verify that BAT26 and TGFβRII alterations were acquired rather than inherited.

Genotyping

DNA was extracted from blood drawn on study participants. The GSTM1 null genotype was detected using the PCR method described by Zhong, et al.²⁴, and three variants of NAT2 (481C>T, 590G>A, and 857A>G) were assessed using the method of Bell, et al.²⁵ and used to impute phenotype, as previously detailed¹. The PCR reaction for GSTM1 uses three primers. The P3 primer sequence is specific to GSTM1 and when used with P1 results in a 230-bp product. The P1-P2 primer pair is nonspecific for GSTM1 and results in a 157-bp product. The PCR products are run on a 2% agarose gel and stained with ethidium bromide. If only the 157-bp product is displayed, the sample is classified as GSTM1 null, whereas if both bands are present, it is classified as GSTM1 positive.

All three NAT2 variants could be identified from one PCR product and digested with three different restriction enzymes (481C>T, KpnI digest; 50G>A, TaqI digest, and 857A>G, BamHI digest). These three variants account for ~95% of the slow acetylation phenotype in Caucasians. Individuals with at least two variant alleles were classified as slow acetylators, while those with one or no variant alleles were classified as rapid acetylators.

Statistical analysis

Study participants who smoked within five years of the referent period were considered recent smokers; former smokers were those individuals who had smoked on a regular basis sometime during their life, but not within the past five years. The cutpoints for pack-years smoked and passive smoke exposure (averaged over the referent period and 10 and 20 years ago) were based on the distribution in controls¹. Associations were evaluated using an unconditional multiple polytomous logistic regression model and adjusting for age at diagnosis or selection, sex, race/ethnicity, center (KPMCP or Utah), and long-term alcohol use. Other potentially confounding variables, including physical activity level and body size, did not substantively impact the estimates and therefore were not included in the final models presented. In addition, passive smoke associations were adjusted for active cigarette smoking. SAS 9.1 was used for all analyses. P-values for trends for frequency categories of pack-years smoked and long-term ETS were calculated using a likelihood ratio test of a full model with categorical exposure category as a continuous variable compared to a restricted model without an exposure variable ²⁶. P-values for interactions were determined by comparing a full model including an ordinal multiplicative interaction term to a reduced model without an interaction term, using a likelihood ratio test.

RESULTS

The distributions of somatic mutations in rectal tumors by recent cigarette smoking status (never or regularly smoked cigarettes within 5 years of the referent year) and GSTM1 genotype and NAT2 imputed phenotype is shown in Table 1. TP53, KRAS2, CIMP, BRAF, and MSI mutation/alteration status was not associated with recent smoking, GSTM1 genotype or NAT2 phenotype in case subjects (chi-square test). In contrast to TP53 or KRAS2 mutations, MSI and BRAF mutations occur infrequently in rectal cancers, in only 2% and 3% of tumors, respectively²⁷. There were no significant differences in tumor distribution by gender or age at diagnosis/control selection (data not shown in table).

Table 1.

Distributions of types of mutations in rectal tumors and smoking,GSTM1,and NAT2 status

		% of	Recent smoker		Not recent			GSTM1 positive		GSTM1 null			NAT2 slow		NAT2 rapid
	N	cases1	N	(%)	N	(%)	p2	N	(%)	N	(%)	p2	N	(%)	N	(%)	p2
Control subjects	1201		189	(15.8)	1009	(84.2)		465	(46.0)	545	(54.0)		520	(52.7)	467	(47.3)
Case subjects	750		149	(19.9)	599	(80.1)	0.02	310	(49.0)	323	(51.0)	0.25	328	(54.6)	273	(45.4)	0.46
Cases with TP53 mutation	338	(48.2)	62	(18.4)	275	(81.6)	0.40	142	(48.6)	150	(51.4)	0.91	157	(56.5)	121	(43.5)	0.48
Cases with KRAS2 mutation	214	(28.9)	41	(19.3)	172	(80.8)	0.83	91	(50.3)	90	(49.7)	0.60	87	(51.2)	83	(48.8)	0.25
Cases with CIMP+  (≥2 of 5 markers)	74	(11.1)	16	(21.9)	57	(78.1)	0.42	36	(57.1)	27	(42.9)	0.15	27	(45.8)	32	(54.2)	0.09
Cases with BRAF mutation	18	(2.5)	5	(27.8)	13	(72.2)	0.41	9	(64.3)	5	(35.7)	0.24	5	(35.7)	9	(64.3)	0.15
Cases with MSI+	16	(2.2)	5	(31.3)	11	(68.8)	0.25	6	(42.9)	8	(57.1)	0.66	7	(50.0)	7	(50.0)	0.72

¹Percent of cases in which alteration status (TP53, KRAS2, CIMP, BRAF, or MSI) could be determined.

²Chi-square test of association; for types of rectal tumors, cases with a specified mutation type were compared to cases not having that mutation type.

In Table 2, risk of rectal cancer associated with TP53, KRAS2 and CIMP status and exposure to active smoking and ETS are shown. As previously reported, the highest category of pack-years smoked (>20years) was associated with an increased risk of rectal cancer overall ¹. An increased risk in smokers (pack-years smoked >20 years) was observed in tumors with a TP53 mutation. Similar to our previous report in colon tumors ¹⁴, we observed an increased risk of TP53 transversion mutations for highest level of pack-years smoked (>20 years), that was not observed for more common TP53 transition mutations (Table 3). Long-term ETS exposure of >10 hours/week, both inside and outside the home, was associated with a 1.5-fold increased risk of a KRAS2 mutation (Table 2). When restricted to only those individuals who were lifetime nonsmokers, a similar but statistically non-significant increased risk of KRAS2 mutation was observed in those exposed to long-term ETS ≤10 hours/week (OR 1.4, 95%CI 0.9, 2.4) and in those with exposure of >10 hours/week (OR 1.5, 95%CI 0.9, 2.6), compared to those not exposed (data not shown in table). Although active smoking was not associated with KRAS2 mutations overall, a significant doubling of risk of KRAS2 transition mutations, the majority occurring in the codon 12 mutation hotspot, in recent smokers (within 5 years of the referent year) was observed (Table 3). In contrast to TP53 and KRAS2 mutations that appear to be related to specific indicators of cigarette smoke exposure, a 50% increased risk was observed in CIMP+ tumors for all indicators of smoking; however, these associations were not statistically significant, given that CIs were imprecise due to small numbers. More than 20 pack-years of smoking was significantly associated with an increased risk of BRAF mutation and an increased risk of MSI+ rectal cancer compared to non-smokers. There were no independent associations of GSTM1 and NAT2 and type of rectal tumor mutation or alteration, with the exception of KRAS2 transversion mutations, in which a GSTM1 null genotype conferred a decreased risk compared to GSTM1 positive (Table 3).

Table 2.

Risk of rectal cancer associated with TP53 mutation, KRAS2 mutation, CIMP, BRAF V600E, and MSI and smoking exposure or metabolizing genes¹,²

			Overall				TP53 mutation				KRAS2 mutation				CIMP+				BRAF V600E mutation				MSI+
Exposure		Controls N	Cases N	OR	95% CI Low. Up.	p- trend	Cases N	OR	95% CI Low. Up.	p- trend	Cases N	OR	95% CI Low. Up.	p- trend	Cases N	OR	95% CI Low. Up.	p- trend	Cases N	OR	95% CI Low. Up.	p- trend	Cases N	OR	95% CI Low. Up.	p- trend
Cigarette	Never/former	1008	598	1.0	-Ref-		274	1.0	-Ref-		172	1.0	-Ref-		57	1.0	-Ref-		13	1.0	-Ref-		10	1.0	-Ref-
smoker	Recent <5yr	189	148	1.3	(1.0 , 1.6)		62	1.2	(0.8 , 1.6)		41	1.3	(0.8 , 1.9)		16	1.5	(0.8 , 2.8)		5	2.1	(0.7 , 6.3)		4	2.1	(0.6 , 7.3)
Pack-years  smoked	None	625	351	1.0	-Ref-		165	1.0	-Ref-		105	1.0	-Ref-		36	1.0	-Ref-		6	1.0	-Ref-		3	1.0	-Ref-
	≤20yr	299	153	0.9	(0.7 , 1.1)		69	0.8	(0.6 , 1.2)		48	0.9	(0.6 , 1.4)		11	0.6	(0.3 , 1.3)		3	1.2	(0.3 , 5.5)		6	4.9	(1.1 , 21.6)
	>20yr	271	241	1.6	(1.2 , 2.0)	<0.01	101	1.4	(1.0 , 2.0)	0.06	60	1.3	(0.9 , 1.9)	0.24	26	1.5	(0.8 , 2.8)	0.22	9	4.2	(1.3 , 14.2)	0.02	5	5.7	(1.1 , 29.8)	0.04
Long-term  ETS	None	408	210	1.0	-Ref-		109	1.0	-Ref-		57	1.0	-Ref-		18	1.0	-Ref-		4	1.0	-Ref-		5	1.0	-Ref-
	≤10 hr/wk	367	220	1.1	(0.9 , 1.4)		93	0.9	(0.7 , 1.3)		63	1.2	(0.8 , 1.8)		27	1.7	(0.9 , 3.1)		8	1.9	(0.6 , 6.7)		5	0.9	(0.3 , 3.4)
	>10 hr/wk	419	314	1.4	(1.1 , 1.7)	0.01	133	1.1	(0.8 , 1.5)	0.43	92	1.5	(1.04 , 2.2)	0.03	29	1.5	(0.8 , 2.8)	0.34	6	1.1	(0.3 , 4.4)	0.83	6	0.9	(0.2 , 3.2)	0.74
GSTM1	Positive	464	310	1.0	-Ref-		142	1.0	-Ref-		91	1.0	-Ref-		36	1.0	-Ref-		9	1.0	-Ref-		6	1.0	-Ref-
	Null	543	320	0.9	(0.7 , 1.1)		148	0.9	(0.7 , 1.1)		89	0.8	(0.6 , 1.2)		26	0.6	(0.4 , 1.0)		5	0.5	(0.2 , 1.4)		8	1.1	(0.4 , 3.3)
NAT2	Slow	519	325	1.0	-Ref-		155	1.0	-Ref-		86	1.0	-Ref-		26	1.0	-Ref-		5	1.0	-Ref-		7	1.0	-Ref-
	Rapid	465	273	0.9	(0.7 , 1.1)		121	0.8	(0.6 , 1.1)		83	1.1	(0.8 , 1.5)		32	1.4	(0.8 , 2.3)		9	2.0	(0.6 , 6.5)		7	1.2	(0.4 , 3.4)

¹Adjusted for age, sex, race, center, long-term alcohol use, and recent smoking status for long-term ETS, GSTM1, and NAT2.

²Number of controls and cases varies from Table 1 slightly due to missing values for one or more adjustment variables.

Table 3.

Risk of rectal cancer associated with TP53 and KRAS2 transition and transversion mutations and smoking exposure or metabolizing genes¹,²

			TP53 transitions				TP53 transversions				KRAS2 transitions				KRAS2 transversions
Exposure		Controls N	Cases N	OR	95% CI Lower Upper	p- trend	Cases N	OR	95% CI Lower Upper	p- trend	Cases N	OR	95% CI Lower Upper	p- trend	Cases N	OR	95% CI Lower Upper	p- trend
Cigarette smoker	Never/former	1008	203	1.0	-Ref-		53	1.0	-Ref-		104	1.0	-Ref-		68	1.0	-Ref-
	Recent <5yr	189	42	1.1	(0.7 , 1.6)		15	1.5	(0.8 , 2.9)		35	1.9	(1.2 , 2.9)		6	0.4	(0.2 , 1.0)
Pack-years smoked	None	625	121	1.0	-Ref-		31	1.0	-Ref-		65	1.0	-Ref-		40	1.0	-Ref-
	≤20yr	299	53	0.9	(0.6 , 1.3)		13	0.9	(0.5 , 1.8)		31	1.0	(0.6 , 1.6)		17	0.9	(0.5 , 1.6)
	>20yr	271	70	1.3	(0.9 , 1.9) 0.17	0.17	22	1.9	(1.01 , 3.4)	0.01	43	1.5	(0.9 , 2.4)	0.10	17	1.0	(0.5 , 1.9)	0.88
Long-term ETS	None	408	80	1.0	-Ref-		19	1.0	-Ref-		37	1.0	-Ref-		20	1.0	-Ref-
	≤10 hr/wk	367	71	0.9	(0.7 , 1.4)		19	1.1	(0.6 , 2.1)		35	1.0	(0.6 , 1.7)		28	1.6	(0.9 , 2.9)
	>10 hr/wk	419	93	1.1	(0.8 , 1.6) 0.58	0.58	30	1.5	(0.8 , 2.8)	0.23	67	1.6	(1.01 , 2.5)	0.03	5	1.4	(0.7 , 2.6)	0.32
GSTM1	Positive	464	109	1.0	-Ref-		21	1.0	-Ref-		53	1.0	-Ref-		38	1.0	-Ref-
	Null	543	105	0.8	(0.6 , 1.1)		35	1.4	(0.8 , 2.4)		65	1.1	(0.7 , 1.6)		24	0.5	(0.3 , 0.9)
NAT2	Slow	519	115	1.0	-Ref-		31	1.0	-Ref-		52	1.0	-Ref-		34	1.0	-Ref-
	Rapid	465	90	0.8	(0.6 , 1.2)		22	0.7	(0.4 , 1.3)		59	1.2	(0.8 , 1.8)		24	0.8	(0.5 , 1.4)

¹Adjusted for age, sex, race, center, long-term alcohol use, and recent smoking status for long-term ETS, GSTM1, and NAT2.

²Number of controls and cases varies from Table 1 slightly due to missing values for one or more adjustment variables.

Interaction models of active and passive smoking and GSTM1 or NAT2 in rectal tumors by alteration type are shown in Table 4 and Table 5, respectively. In Table 4, an interaction between recent smoking (within 5 years of the referent year) and imputed NAT2 rapid-acetylator phenotype was statistically non-significant in CIMP+ tumors (p=0.10); however, recent smokers had a 2.5-fold increased risk of CIMP+ tumors compared to non-recent smokers with a slow NAT2 phenotype, although the point estimate was imprecise due to small numbers. After adjusting for active smoking, individuals with long-term exposure to more than 10 hours of ETS per week (Table 5) had an increased risk of TP53 mutation among those with an imputed NAT2 slow-acetylator phenotype, but not among those with a rapid phenotype (p-interaction <0.01). Neither GSTM1 nor NAT2 significantly interacted with ETS to alter risk of other rectal tumor somatic alterations.

Table 4.

Risk of rectal cancer associated with TP53 mutation, KRAS2 mutation, CIMP and metabolizing genes, and interaction with recent cigarette smoking¹,²

	GSTM1								NAT2
	Positive				Null				Slow				Rapid
	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper
All tumor types
Never/former	390	250	1.0	-Ref-	458	247	0.8	(0.7 , 1.0)	439	271	1.0	-Ref-	392	202	0.8	(0.6 , 1.0)
Recent <5yr	74	60	1.2	(0.8 , 1.8)	85	73	1.3	(0.9 , 1.8)	80	54	1.0	(0.7 , 1.5)	73	71	1.5	(1.02 , 2.1)
p-interaction								0.38								0.04
TP53 mutation
Never/former	390	116	1.0	-Ref-	458	116	0.8	(0.6 , 1.1)	439	131	1.0	-Ref-	392	90	0.7	(0.5 , 0.99)
Recent <5yr	74	25	1.2	(0.7 , 1.9)	85	32	1.2	(0.8 , 2.0)	80	24	1.0	(0.6 , 1.6)	73	31	1.3	(0.8 , 2.1)
p-interaction								0.53								0.09
KRAS2 mutation
Never/former	390	73	1.0	-Ref-	458	116	0.9	(0.6 , 1.2)	439	72	1.0	-Ref-	392	63	0.9	(0.7 , 1.4)
Recent <5yr	74	18	1.2	(0.7 , 2.2)	85	32	1.0	(0.5 , 1.8)	80	14	1.0	(0.5 , 1.9)	73	20	1.6	(0.9 , 2.8)
p-interaction								0.83								0.22
CIMP+
Never/former	390	29	1.0	-Ref-	458	20	0.6	(0.3 , 1.0)	439	23	1.0	-Ref-	392	23	1.1	(0.6 , 2.0)
Recent <5yr	74	7	1.3	(0.5 , 3.2)	85	6	1.0	(0.4 , 2.5)	80	3	0.7	(0.2 , 2.5)	73	9	2.5	(1.1 , 6.0)
p-interaction								0.69								0.10

¹Adjusted for age, sex, race, center, and long-term alcohol use.

²Number of controls and cases varies from Table 1 slightly due to missing values for one or more adjustment variables.

Table 5.

Risk of rectal cancer associated with TP53 mutation, KRAS2 mutation, CIMP and metabolizing genes, and interaction with ETS¹,²

	GSTM1								NAT2
	Positive				Null				Slow				Rapid
	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper	Controls N	Cases N	OR	95% CI Lower Upper
All tumor types
No Exposure	164	91	1.0	-Ref-	165	82	0.9	(0.6 , 1.3)	174	83	1.0	-Ref-	147	80	1.1	(0.7 , 1.6)
≤10hr/wk	138	90	1.2	(0.8 , 1.7)	176	98	1.0	(0.7 , 1.4)	168	97	1.2	(0.8 , 1.7)	141	84	1.2	(0.8 , 1.7)
>10hr/wk	161	126	1.3	(0.9 , 1.8)	201	140	1.2	(0.8 , 1.6)	176	145	1.6	(1.1 , 2.2)	176	107	1.1	(0.8 , 1.6)
p-interaction								0.98								0.07
TP53 mutation
No Exposure	164	46	1.0	-Ref-	165	47	1.1	(0.6 , 1.6)	174	40	1.0	-Ref-	147	48	1.4	(0.8 , 2.2)
≤10hr/wk	138	37	1.0	(0.6 , 1.6)	176	45	0.9	(0.5 , 1.4)	168	44	1.1	(0.7 , 1.8)	141	33	0.9	(0.6 , 1.6)
>10hr/wk	161	58	1.2	(0.7 , 1.9)	201	56	0.9	(0.6 , 1.5)	176	71	1.6	(1.0 , 2.6)	176	39	0.8	(0.5 , 1.4)
p-interaction								0.37								<.01
KRAS2 mutation
No Exposure	164	27	1.0	-Ref-	165	20	0.7	(0.4 , 1.4)	174	19	1.0	-Ref-	147	22	1.4	(0.7 , 2.6)
≤10hr/wk	138	30	1.3	(0.7 , 2.3)	176	26	0.9	(0.5 , 1.6)	168	30	1.6	(0.9 , 2.9)	141	24	1.5	(0.8 , 2.9)
>10hr/wk	161	33	1.2	(0.7 , 2.1)	201	43	1.3	(0.7 , 2.2)	176	37	1.8	(1.0 , 3.3)	176	36	1.7	(0.9 , 3.2)
p-interaction								0.33								0.42
CIMP+
No Exposure	164	10	1.0	-Ref-	165	5	0.5	(0.2 , 1.5)	174	5	1.0	-Ref-	147	9	2.1	(0.7 , 6.4)
≤10hr/wk	138	14	1.7	(0.7 , 4.0)	176	9	0.8	(0.3 , 2.1)	168	10	2.1	(0.7 , 6.2)	141	12	3.0	(1.0 , 8.8)
>10hr/wk	161	12	1.2	(0.5 , 2.9)	201	12	0.9	(0.4 , 2.3)	176	11	2.1	(0.7 , 6.3)	176	11	2.1	(0.7 , 6.5)
p-interaction								0.49								0.34

¹Adjusted for age, sex, race, center, and long-term alcohol use, and recent cigarette smoking status.

²Number of controls and cases varies from Table 1 slightly due to missing values for one or more adjustment variables.

DISCUSSION

This large, population-based investigation is the first to examine associations of active smoking and exposure to ETS and somatic alterations in rectal cancers. Similar to our previous research in colon cancers^{16, 28}, active cigarette smoking was associated with an increased risk of TP53 mutation overall and less-common TP53 transversions in particular, BRAF V600E mutations, and MSI in rectal tumors. Although power was limited to detect statistically significant associations for CIMP+ in rectal tumors due to small numbers, our results for all indicators of smoking suggest that active smoking may modestly increase risk, similar to our findings for colon tumors¹⁶; additionally, ETS may increase risk of CIMP+ rectal tumors (passive smoking was not assessed in our previous study¹⁶). Long-term ETS exposure was associated with KRAS2 mutations in rectal tumors, and NAT2-imputed slow phenotype appeared to modify associations of rectal cancer and exposure to ETS in tumors with a TP53 mutation.

Similar to colon cancer, smoking appears to be related to CIMP+, BRAF V600E, and MSI+ in rectal cancer. However, in our previous study of colon tumors, we hypothesized that the effect on MSI+ tumors was due to the effect on CIMP, resulting from an association between smoking and CIMP+ tumors that also was observed in stable colon cancers. In this study, the association with MSI+ is probably not a consequence of an association with CIMP+, as the majority of rectal MSI+ tumors are not CIMP+ and do not exhibit MLH1 methylation, or BRAF mutations^{27, 29}. Indeed, this research suggests a non-CIMP pathway in Lynch-related rectal cancers which is affected by smoking. In our current study, it was previously determined that Lynch-associated cancers are over-represented in the small number of rectal tumors that exhibit MSI ²⁹. It has been reported by others that cigarette smoking increases risk of CRC tumors in patients with Lynch syndrome^{30, 31}. It should be noted in contrast to our previous study of colon cases, the numbers of rectal cases exhibiting BRAF mutation or MSI+ were small and consequently, confidence intervals for risks associated with smoking exposures, including statistically significant estimates for pack-years of smoking, were imprecise for these tumor subtypes (see Table 2).

It is plausible that the association between smoking and ETS and rectal tumor alterations might be modified by the ability to detoxify PAH generated from smoking cigarettes. Although associations between rectal cancer and genetic variants of GSTM1 and NAT2 are inconsistent^{1, 12, 32}, it was hypothesized that genetic variants of these Phase II-detoxifying enzymes may be important in conjunction with cigarette smoking and rectal tumor characteristics given the associations we observed with TP53, KRAS2, and BRAF mutations, CIMP, MSI, and exposure to active and passive cigarette smoke. Lilla et al. reported heterocyclic aromatic amines may play an important role in CRC and possibly exposure to ETS, and that NAT enzymes play a greater role in bioactivation, rather than in detoxification³³. In a large, prospective female cohort, van der Hel et al. reported that rapid acetylation did not increase CRC risk, but in combination with smoking risk was increased, compared to women who had a slow NAT2 imputed phenotype and never smoked; however, when stratified further by colon or rectal site, the increased risk for rectal cancer was not statistically significant¹². We previously reported an increased risk of rectal cancer with smoking cigarettes among those who were GSTM1 null relative to smokers with a GSTM1 positive genotype, and that men who were both GSTM1 positive and NAT2 rapid acetylators had no increased risk from cigarette smoking¹.

It is interesting to note associations with smoking observed for TP53 or KRAS2 mutations appear to differ between transition and transversion mutations (which are less common than transitions; see Table 3). Whereas >20 pack-years of smoking was more strongly associated with TP53 transversions, the majority comprised of G to T mutations, recent active smoking was significant for KRAS2 transition mutations only. A significant excess of TP53 G to T transversions in smoking-associated lung cancers compared to lung cancers in nonsmokers has been reported, with a similar trend observed in laryngeal, head and neck cancers³⁴. The Netherlands Cohort Study recently reported they observed no associations between smoking and risk for colorectal cancer with KRAS2-mutated tumors, and no clear associations with respect to KRAS2 transition and transversion mutations in 648 incident colorectal cancer cases ³⁵. The relationship between ETS and increased risk of KRAS2 was similar for both transition and transversion mutations, although the association was statistically significant for transition mutations only. Findings were similar for ETS exposure and rectal tumors with KRAS2 mutations or CIMP+ phenotype. This perhaps suggests that KRAS2+ and CIMP+ tumors, or tumors with both alterations (which generally do not exhibit BRAF mutation or MSI),^{18, 27} are predisposed to rectal cancer associated with a high level of long-term exposure to second-hand smoke.

It is possible genetic variants of metabolizing enzymes that modify environmental exposures may influence risk associated with having specific types of tumor mutations^{13, 14}. In this study, we further investigate potential interaction of smoking exposure with Phase II metabolizing enzymes and alteration sub-type in rectal cancer. We found no significant interaction of GSTM1 and smoking-related risk in rectal tumor alterations. However, our results suggest NAT2 phenotype may interact with ETS and risk of TP53 mutations, although no interaction was observed with regard to active smoking; therefore, additional studies are needed to confirm or disprove the influence of NAT2 in active and passive smoking-related risk of mutations in TP53. We recognize that the associations we observed in our investigation of specific study hypotheses may represent chance findings, because a number of comparisons were made; thus replication in other studies is warranted.

In conclusion, our investigation of somatic alterations and smoking in rectal cancer supports cigarette smoking as a risk factor for TP53 mutations, particularly transversions, as we previously observed in colon cancer. Our results suggest recent smoking may be a risk factor for KRAS2 transition mutations in rectal tumors. Although relatively rare in rectal cancers, smoking is associated with increased risk of BRAF and MSI+, and is suggestive of an increased risk of CIMP+ tumors, also similar to our findings in colon cancers. Additionally, KRAS2+ and CIMP+ tumors (or tumors with both alterations) may be predisposed to rectal cancer associated with a high level of long-term exposure to second-hand smoke. We generally observed no interaction of GSTM1 and NAT2 metabolizing gene variants and smoking-related risk of rectal tumor subtypes. Our results suggest NAT2 phenotype may interact with ETS to influence risk of TP53 mutations in rectal cancer; confirmatory studies are warranted.

Abbreviations

CIMP

CpG Island Methylator Phenotype

PCR

polymerase chain reaction

MSI

microsattelite instability

OR

odds ratio

CI

confidence interval