Cancer Causes & Control

, Volume 21, Issue 8, pp 1165–1170

Interleukin-22 genetic polymorphisms and risk of colon cancer


  • Cheryl L. Thompson
    • Department of Family Medicine-Research DivisionCase Western Reserve University/University Hospitals of Cleveland
    • Case Center for Transdisciplinary Research on Energetics and Cancer
  • Sarah J. Plummer
    • Department of Preventive MedicineUniversity of Southern California
  • Thomas C. Tucker
    • Cancer Control Program, University of Kentucky
  • Graham Casey
    • Case Center for Transdisciplinary Research on Energetics and Cancer
    • Department of Preventive MedicineUniversity of Southern California
    • Department of Family Medicine-Research DivisionCase Western Reserve University/University Hospitals of Cleveland
    • Case Center for Transdisciplinary Research on Energetics and Cancer
Original paper

DOI: 10.1007/s10552-010-9542-5

Cite this article as:
Thompson, C.L., Plummer, S.J., Tucker, T.C. et al. Cancer Causes Control (2010) 21: 1165. doi:10.1007/s10552-010-9542-5


Interleukin-22 (IL-22) is a member of the IL-10 family of anti-inflammatory cytokines that mediates epithelial immunity. IL-22 expression is enhanced in inflamed colon mucosa in individuals with inflammatory bowel disease. We carried out an association study to examine the hypothesis that common variation in the IL-22 gene is associated with risk of colon cancer. Seven tagging SNPs were genotyped in 561 colon cancer cases and 722 population controls. Information on lifestyle risk factors was collected via a self-administered questionnaire. The rs1179251 SNP conferred an estimated odds ratio (OR) of 1.46 (95% CI = 1.04–2.05) and 2.10 (95% CI = 0.66–6.66), respectively, for those heterozygous and homozygous for the G variant (padditive = 0.013) after adjustment for age, gender, and race; the OR assuming a dominant model was 1.50 (95% CI = 1.05–2.08, pdominant = 0.016). No other SNP was statistically significantly associated with colon cancer risk. Haplotype analysis found that one haplotype containing the rs1179251 G allele gave an estimated 52% increase in risk of colon cancer for individuals with at least one copy (OR = 1.52, 95% CI = 1.12–2.06, p = 0.0073). Our findings suggest that the rs1179251 SNP in IL-22 is associated with risk of colon cancer.


IL-22Colon cancerSNPAssociationCase–control study


Increasing evidence indicates that host immune response, cytokines, and related inflammatory and immune mediators participate in human carcinogenesis (reviewed in [1, 2]). Interleukin-22 (IL-22), a member of the IL-10 family, is a recently identified T-cell-derived cytokine that mediates epithelial immunity and mucosal tissue repair (reviewed in [3]). IL-22 binds to a heterodimeric receptor complex composed of IL-22 receptor 1 (IL-22R1) and IL-10 receptor 2 (IL-10R2) at the cell surface [4], resulting in activation of the JAK-STAT pathway, leading to phosphorylation of STAT3 and STAT1/STAT5 [5]. IL-22 has also been shown to activate ERK, JNK, and p38 MAPK [5]. Cellular IL-22 responsiveness is mainly determined by the expression of IL-22R1, which is expressed on a variety of cells of epithelial origin, including human colonic subepithelial myofibroblasts and colon carcinoma cells [6]. It has recently been shown that IL-22 expression is enhanced in inflamed colon mucosa in patients with inflammatory bowel disease [7] and has been shown to influence tumor growth via multiple pathways [8]. However, no epidemiological study has examined IL-22 expression or its genetic variation with cancer risk.

Although not extensively studied to date, evidence supporting a role of IL-22 in colon carcinogenesis is emerging [6]. Furthermore, because IL-22 stimulates signaling pathways that are involved in the regulation of cell growth, proliferation, and cell cycle control, it is plausible that IL-22 may influence the development of colon cancer and that the IL-22 gene is a candidate susceptibility gene for colon carcinogenesis. Therefore, we comprehensively evaluated the association of common inherited IL-22 genetic polymorphisms with colon cancer risk in a population-based case–control study.

Materials and methods

Study population

The study population consisted of 561 incident colon cancer cases and 721 population controls recruited between July 2003 and March 2007. This study population has been described in detail earlier [9]. Potential colon cancer cases were identified through the Surveillance, Epidemiology, and End Results (SEER) Kentucky Cancer Registry, which covers all residents of the State of Kentucky at the time of diagnosis. The cancer registry was queried every 3 months to identify all incident primary colon cancer cases reported within 6 months of diagnosis. We used random digital dialing to recruit a population sample of controls representative of the general Kentucky population. Controls were required to be 30 years of age or older and free of personal history of cancer other than skin cancer. For both cases and controls, we excluded those with self-reported inflammatory bowel diseases, family history of familial adenomatous polyposis (FAP), and hereditary non-polyposis colorectal cancer (HNPCC). The response rates were 72.2% for the cases and 62.5% for eligible controls.

Each participant donated a blood sample for genetic analysis and responded to a lifestyle risk factor questionnaire (RFQ) developed by the colon cancer family registry group ( Blood was drawn at a facility close to the participant’s home throughout the state of Kentucky and shipped, on ice, overnight to the research laboratory at Case Western Reserve University for processing. White blood cell fractions were stored at −80°C until DNA extraction. The RFQ collected detailed information on demographics, family history of colorectal cancer, lifestyle, and behavioral risk factors.

The study was approved by the Institutional Review Boards of the University of Kentucky, Lexington, Case Western Reserve University/University Hospitals of Cleveland, and University of Southern California. All participants provided written informed consent.

Lifestyle risk factors

Lifestyle variables were obtained through use of the self-administered risk factor questionnaire as described [9]. Body mass index (BMI) was calculated based on current weight in kilograms divided by height in meters squared. A positive family history of colorectal cancer was defined as reporting of knowledge of colorectal cancer in one or more first-degree relatives. Regular non-steroidal anti-inflammatory drug (NSAID) use was defined as reporting usage at least twice a week for 6 months or longer.

Tagging SNP selection and genotypes (IL-22)

Tagging SNPs were identified by querying the HapMap for all SNPs within IL-22 covering 5-kbp upstream and 5-kb downstream of the gene, with a minor allele frequency (MAF) of at least 5% in Caucasians, given that our population is predominantly of European descent (>94%). The Tagger program [10] was used to identify SNPs that “tag” the gene and capture all known common (≥5% MAF) variation in the gene with at least 80% correlation. No known non-synonymous coding or regulatory SNPs with a MAF of at least 5% had been reported in IL-22 in the HapMap Caucasian population. All SNPs identified via Tagger had been previously classified as non-coding.

DNA extraction was performed from previously frozen white blood cells using the Qiagen Biorobot EZ1 workstation. DNA extraction was successful in 100% of both cases and controls. Samples were genotyped using predesigned primer/probe sets (Applied Biosystems, Foster City, CA) and RealMasterMix Probe ROX (5 Prime, Hamburg, Germany) on an Applied Biosystems 7900HT according to manufacturer’s instructions. Lab personnel were blinded to case–control status, and samples were processed in a random fashion. Two percent of samples were repeated with 100% concordance of results. Statistical analysis was performed after all genotyping was complete.

Statistical analysis

We first examined the genotypic distribution for departure from Hardy–Weinberg proportions via chi-square goodness of fit statistic. This was computed separately for the cases and controls. We then evaluated the association of the genotypes of each SNP with colon cancer using three different unconditional logistic regression models. The first model included only the SNP (univariate model). The second included the SNP, age, gender, and race (adjusted model).

For each SNP, the more common variant was treated as the referent. Three possible genetic models for risk were evaluated—dominant (1 if have at least one non-reference allele, 0 otherwise), additive (2 if have two non-reference alleles, 1 if have one, 0 otherwise), or recessive (1 if are homozygous for the non-reference allele and 0 otherwise).

We addressed potential population substructure by limiting our analyses to those individuals with self-reported race of Caucasian. As was expected since this was approximately 94% of our sample, limiting our analyses did not change results or alter any conclusion. Therefore, only the results in the entire sample population are presented in full here.

Haplotypes were reconstructed using an expectation–maximization algorithm as implemented in DECIPHER in the S.A.G.E. software package ( using the entire sample population. Since all the participants had a probability of 95% or higher for each of the inferred haplotypes, we chose the most likely combination of haplotypes for each individual. All haplotypes with a population frequency of 5% or higher were included in the haplotype association analyses. Haplotypes were evaluated for association with colon cancer in unadjusted and adjusted logistic regression models as done for the individual SNPs. Each haplotype was coded as a dummy variable (0 or 1), representing the possession of at least one copy of that haplotype. All haplotypes were examined simultaneously in regression models with the most common haplotype as the referent.

All p-values presented here are two-sided, and an α = 0.05 was used to declare statistical significance. All analyses were completed using SAS software (version 9.1; SAS Institute, Inc., Cary, North Carolina).


Basic demographics are shown in Table 1. Our sample is predominantly (about 94%) Caucasian and closely mirrors the overall Kentucky population. Cases had a higher BMI, a higher proportion of a positive family history of colorectal cancer, and were older than the controls.
Table 1

Descriptive characteristics of study population


Cases (n = 560)

Controls (n = 722)


Age (SD)

62.9 (10.6)

58.4 (10.9)


Sex (%)




270 (48.2)

275 (38.1)



290 (51.8)

447 (61.9)


Race (%)




23 (4.1)

23 (3.2)



527 (94.3)

681 (94.5)



9 (1.6)

17 (2.4)


Family history of colorectal cancer (%)a




176 (34.7)

172 (25.4)



332 (65.4)

504 (74.6)


Regular non-steroidal anti-inflammatory drug use (%)b




218 (42.8)

311 (45.7)



292 (57.3)

307 (54.3)


Body mass index (SD)c

29.3 (6.1)

28.1 (6.3)


aFamily History defined as self-reporting of at least one first-degree relative with known colorectal cancer

bRegular non-steroidal anti-inflammatory drug (NSAID) use was defined as self-report use of aspirin or ibuprofen at least twice a week for 6 months or longer

cBody mass index (BMI) was calculated as current weight in kilograms divided by height in meters squared

One SNP in IL-22 (rs1179251) showed a positive association with colon cancer (Table 2). Having one or more copy of the risk allele (G) conferred an estimated 50% increase in colon cancer risk in the model adjusted for age, gender, and race (OR = 1.50, 95% CI = 1.11–2.02, pdominant = 0.0087). No other SNP was statistically significantly associated with colon cancer (Table 2). The SNP association results were almost identical when the analysis was restricted to Caucasians to minimize the effect of population stratification. In Caucasians only, having one or more copy of the risk allele showed a 45% increase in odds of colon cancer after adjustment for age and gender (OR = 1.45; 95% CI = 1.06–1.99; pdominant = 0.020, Supplementary Table S1). As with the full sample, when restricting our population to self-reported Caucasians, no other SNP was statistically significantly associated with colon cancer risk (Supplementary Table S1).
Table 2

Associations between IL-22 SNPs and colon cancer risk

SNP genotype


ORa (95% CI)


ORb (95% CI)





189 (33.9)/236 (32.7)






250 (44.8)/347 (48.1)

0.90 (0.70–1.16)


0.89 (0.69–1.15)



119 (21.3)/139 (19.3)

1.07 (0.78–1.46)


1.15 (0.84–1.60)


AC or CC vs. AA


0.95 (0.75–1.20)


0.96 (0.76–1.22)


CC vs. AC or AA


1.14 (0.86–1.50)


1.24 (0.93–1.65)





279 (50.1)/380 (52.6)






214 (38.4)/279 (38.6)

1.05 (0.83–1.32)


1.05 (0.83–1.34)



64 (11.5)/63 (8.7)

1.38 (0.95–2.03)


1.37 (0.92–2.02)


AA or TA vs. TT


1.11 (0.89–1.38)


1.11 (0.89–1.40)


AA vs. TA or TT


1.36 (0.94–1.96)


1.34 (0.91–1.95)





441 (78.9)/611 (84.6)






109 (19.5)/103 (14.3)

1.47 (1.09–1.97)


1.50 (1.10–2.04)



9 (1.2)/8 (1.1)

1.56 (0.60–4.07)


1.50 (0.55–4.08)


GG or CG vs. CC


1.47 (1.11–1.96)


1.50 (1.11–2.02)


GG vs. CG or CC


1.46 (0.56–3.81)


1.36 (0.50–3.69)





461 (82.9)/579 (80.2)






91 (16.4)/135 (18.7)

0.85 (0.63–1.13)


0.89 (0.66–1.21)



4 (0.7)/8 (1.1)

0.93 (0.19–2.10)


0.93 (0.27–3.19)


AA or TA vs. TT


0.83 (0.63–1.11)


0.89 (0.66–1.20)


AA vs. TA or TT


0.65 (0.19–2.16)


0.96 (0.28–3.25)





352 (63.1)/454 (62.9)






168 (30.1)/232 (32.1)

0.93 (0.73–1.19)


0.89 (0.69–1.15)



38 (6.8)/36 (5.0)

1.36 (0.85–2.19)


1.17 (0.72–1.91)


CC or TC vs. TT


0.99 (0.79–1.25)


0.93 (0.73–1.18)


CC vs. TC or TT


1.39 (0.87–2.23)


1.22 (0.76–1.98)





351 (63.0)/464 (64.3)






184 (33.1)/221 (30.6)

1.10 (0.87–1.40)


1.12 (0.87–1.43)



22 (4.0)/37 (5.1)

0.79 (0.46–1.36)


0.74 (0.42–1.30)


AA or GA vs. GG


1.06 (0.84–1.33)


1.06 (0.84–1.34)


AA vs. GA or GG


0.76 (0.44–1.31)


0.71 (0.41–1.24)





425 (76.4)/524 (72.6)






125 (22.5)/181 (25.1)

0.85 (0.66–1.11)


0.89 (0.68–1.12)



6 (1.1)/17 (2.4)

0.44 (0.17–1.11)


0.57 (0.22–1.50)


CC or TC vs. TT


0.82 (0.63–1.05)


0.86 (0.66–1.12)


CC vs. TC or TT


0.45 (0.18–1.16)


0.59 (0.22–1.54)


*p-value for additive genetic model (p for trend)

aUnadjusted; bAdjusted for age, gender, and race

Five haplotypes were estimated to have a population frequency of at least 5%. One haplotype, C-A-G-A-T-G-T, showed evidence of a statistically significantly increased risk of colon cancer in this population, with an approximately 50% increase in risk for individuals with at least one risk haplotype (OR = 1.54, 95% CI = 1.13–2.11, p = 0.0070, in the adjusted model) (Table 3). Repeating the haplotype analysis while restricting to Caucasians resulted in very similar results, with the C-A-G-A-T-G-T haplotype being significantly associated with colon cancer risk (OR = 1.48; 95% CI = 1.07–2.06; p = 0.018, Supplementary Table S2).
Table 3

Associations between IL-22 haplotypes and colon cancer risk

Haplotype (copies)

Cases (%)/controls (%)

ORa (95% CI)


ORb (95% CI)









242 (43.2)/301 (41.7)



252 (45.0)/332 (46.0)



66 (11.8)/89 (12.3)




1.08 (0.84–1.39)


1.02 (0.78–1.32)



354 (63.2)/456 (63.2)



168 (30.0)/230 (31.9)



38 (6.8)/36 (5.0)




1.08 (0.84–1.39)


1.08 (0.83–1.39)



357 (63.8)/468 (64.8)



182 (32.5)/217 (30.1)



21 (3.8)/37 (5.1)




0.89 (0.66–1.20)


0.94 (0.69–1.28)



465 (83.0)/580 (80.3)



91 (16.3)/134 (18.6)



4 (0.7)/8 (1.1)




1.51 (1.12–2.05)


1.54 (1.13–2.11)



448 (80.0)/617 (85.5)



105 (18.8)/99 (13.7)



7 (1.3)/6 (0.8)


Association of possession of at least one copy of the given rs1179246-rs1182844-rs1179251-rs17224704-rs2227513-rs2227484-rs1026786 inferred haplotypes in IL-22 with risk of colon cancer under two models: aUnadjusted, bAdjusted for age, gender, and race


IL-22 has shown promise as a potential regulator of carcinogenesis and tumor progression via multiple biological pathways [8]. IL-22 is a member of the IL-10 family that has been associated with response to microbial infection [11, 12] and suggested as a potential link between inflammation and immune response and cancer development (reviewed in [2]). SNPs in IL-10 have been associated with sporadic colon cancer risk [13]. Furthermore, studies have linked IL-22 with inducible nitric-oxide synthase (iNOS), a marker of colonic inflammation and promoter of colon cancer, expression in colon carcinoma cells [6]. We hypothesized that IL-22 represented a novel candidate susceptibility gene for colon cancer. Indeed, we found evidence for an association of the rs1179251 SNP with colon cancer. Haplotype analysis provided corroborating evidence. To our knowledge, this is the first study to report an association of SNPs in IL-22 with risk of colon cancer or any other cancers.

There are some limitations in our study. We used tagging SNPs in this study, which has the advantage of efficiently capturing the entire gene with minimal redundancy, but individual SNPs selected do not necessarily have any biological significance. In fact, the SNP (rs1179251) with the strongest statistical evidence of association with colon cancer is an intronic (non-coding) SNP. Further work is warranted to determine the functionality of this SNP or if it is a surrogate for nearby variants that may have biological consequences.

Our controls were recruited from the general population in the state of Kentucky; it is possible that some of the controls enrolled may have undiagnosed cancer or asymptomatic adenomatous polyps, an established precursor to colon cancer. We believe such a scenario is extremely unlikely for colon cancer. As for participants with undiagnosed adenomatous polyps, it is important to keep these individuals in the study as controls unless a subsequent diagnosis of colon cancer is made because automatically excluding these participants can lead to a bias away from the null [14]. Further, we have replicated the previously reported associations of genetic variants in SMAD7 [15] and the 8q24 genomic region [16] with risk of colon cancer in this study population, lending credibility to our present study.

Multiple testing can be an issue in genetic association studies when an agnostic approach is taken and a large number of independent genetic variants across the genome are evaluated. For this study, because of the focused, hypothesis-driven approach of the evaluation of the association of non-independent variants within a single gene with prior biological evidence of a likely association with colon cancer, we did not initially adjust for multiple testing. By reducing the type I error (false positives) of our study by adjusting for multiple comparisons, we would increase the type II error (false negatives) unnecessarily [17]. Application of the rapid adjustment for p-values for correlated tests method, as described in Conneely and Boehnke [18], resulted in an adjusted p-value of 0.11 for the rs1179251 SNP. Therefore, we cannot rule out the possibility that this finding is a false positive. Nevertheless, the biological plausibility for such an association and our suggestive findings warrant further investigation in independent populations.

Population stratification is a potential confounder in any genetic association study and can lead to false positives. While we did not formally analyze or adjust for population stratification, we repeated our statistical analyses after restricting to individuals of self-reported Caucasian race to minimize the effects of heterogeneity across ethnic populations. The results were very similar to analysis including the entire study population, suggesting that population stratification, if any, is unlikely to have biased our results. Unfortunately, small numbers do not allow us to repeat analysis when restricted to any other racial subpopulation. Replication in other ethnic populations will be necessary to generalize conclusions to these populations.

Our haplotype analyses showed a strong association with a single haplotype, which is the only haplotype containing the risk allele (G) of the rs1179251 SNP. Therefore, the rs1179251 SNP is likely to be driving this haplotype association. The region tagged by the rs1179251 SNP thus may harbor previously unidentified variances that with functional consequences. While confirmation of our results in other populations will be necessary, the results of this study are in line with other studies suggesting the potential role of this gene in tumor progression [6, 8], and support common inherited variation in IL-22 is associated with colon cancer risk.


Supported by an R25 and K07 training grants and an R01 research grant from the National Cancer Institute (R25 CA094186, K07 CA136758, R01 CA136726), the Damon Runyon Cancer Research Foundation, Clinical Investigator Award (CI-8), Case Center for Transdisciplinary Research on Energetics and Cancer (U54 CA116867), and National Cancer Institute K22 Award (K22 CA120545). The authors wish to thank Svetlana Zelenskiy for her assistance in constructing the tables.

Supplementary material

10552_2010_9542_MOESM1_ESM.doc (112 kb)
Supplementary material 1 (DOC 112 kb)

Copyright information

© Springer Science+Business Media B.V. 2010