|Home | About | Journals | Submit | Contact Us | Français|
Soluble cytokine receptors and receptor antagonist of proinflammatory cytokines can modify cytokine signaling and may affect cancer risk.
In a case-cohort study nested within the Women’s Health Initiative cohort of postmenopausal women, we assessed the associations of plasma levels of IL-1 receptor antagonist (IL-1Ra) and the soluble receptors of IL-1 (sIL-1R2), IL-6 (sIL-6R and sgp130), and TNF (sTNFR1 and sTNFR2) with risk of colorectal cancer in 433 cases and 821 subcohort subjects. Baseline levels of estradiol, insulin, leptin, IL-6, and TNF-α measured previously were also available for data analysis.
After adjusting for significant covariates – including age, race, smoking, colonoscopy history, waist circumference, and levels of estrogen, insulin, and leptin – relatively high levels of sIL-6R and sIL-1R2 were associated with reduced colorectal cancer risk [hazard ratios comparing extreme quartiles (HRQ4-Q1) for sIL-6R = 0.56, 95% CI = 0.38–0.83; HRQ4-Q1 for sIL-1R2 = 0.44, 95% CI = 0.29–0.67]. The associations with IL-1Ra, sgp130, sTNFR1, and sTNFR2 were null. The inverse association of sIL-1R2 with colorectal cancer risk persisted in cases diagnosed ≤5 and >5 years from baseline blood draw; the association with sIL-6R, however, was not evident in the latter group, possibly indicating that relatively low levels of sIL-6R in cases might be due to undiagnosed cancer at the time of blood draw.
High circulating levels of sIL-1R2 may be protective against colorectal carcinogenesis and/or be a marker of reduced risk for the disease.
sIL-1R2 has potential to be a chemopreventive and/or immunotherapeutic agent in inflammation-related diseases.
Experimental studies have shown that potent proinflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-1β, have oncogenic effects (1, 2). However, soluble cytokine receptors or receptor antagonists of these cytokines can modify cytokine signaling and may also affect cancer risk (3, 4). This case-cohort study nested within the Women’s Health Initiative Observational Study (WHI-OS) assessed circulating levels of the soluble cytokine receptors and receptor antagonist of these three potent inflammatory cytokines for their associations with risk of colorectal cancer (CRC), an inflammation-associated disease (5–7). Specifically, IL-1 receptor antagonist (IL-1Ra) and the soluble receptors of IL-1 (sIL-1R2), IL-6 (sIL-6R and sgp130), and TNF (sTNFR1 and sTNFR2) were examined. With the exception of sTNFR2 (8), none of these analytes have ever been studied for their CRC associations.
Soluble cytokine receptors can be formed by proteolytic cleavage of cell-surface receptors or by alternative splicing of mRNA that deletes the transmembrane domain of membrane-associated receptors (3). Many soluble cytokine receptors (e.g., sIL-1R2, sTNFR1, and sTNFR2) may act as decoy receptors, block binding of the ligands to cognate functional membrane receptors, and hence inhibit cytokine signaling (4, 9–11). The biological effects of IL-6 soluble receptors are more complex. IL-6 signaling requires the interaction of IL-6 with its receptor complex consisting of IL-6R and the signal transducing protein gp130. While gp130 is ubiquitously expressed, IL-6R is not present on all cells. Nevertheless, IL-6R has a soluble form sIL-6R that binds to IL-6 to form a soluble complex, which binds to cell-surface gp130 on cells that lack the membrane bound IL-6R and initiates signaling. As such, sIL-6R is thought to be an IL-6 agonist enhancing IL-6 signaling (9). However, there is also a soluble form of gp130. IL-6 can be trapped in the soluble ternary complex with sIL-6R and sgp130, resulting in inhibition of IL-6 signaling (12). IL-1Ra is a naturally occurring receptor antagonist that binds to, without activating, the membrane-bound IL-1 receptors and hence competitively blocks IL-1 binding to its cell-surface receptors (4, 13).
The six analytes were chosen because of their biological activities that they could modify signaling of the potent inflammatory cytokines. This study tested the hypothesis that the levels of IL-1Ra, sIL-1R2, sgp130, sTNFR1, and sTNFR2 would be inversely associated with CRC risk, whereas sIL-6R might have the opposite effect. Moreover, we had previously examined the associations of the ligands, IL-6 and TNF-α, with CRC risk on the same cases and subcohort subjects (7); this present study would have the opportunity to assess how the cytokine ligands and their soluble receptors might act together to affect CRC risk.
The WHI-OS is a prospective study of 93,676 postmenopausal women 50–79 years of age at recruitment between 1993 to 1998 in the United States (14). At baseline, participants completed epidemiologic questionnaires, and morning, fasting blood samples were collected, centrifuged, frozen on-site at −80°C, and later shipped to the central specimen repository. Diagnosis of CRC was ascertained by annual self-administered questionnaires and confirmed through centralized review of medical records.
This CRC study was a component of a case-cohort study in which three cancer outcomes (breast, colorectum, and endometrium) were previously examined for their associations with estradiol, insulin, IL-6, TNF-α, leptin, and other adipokines (7, 15–17). There were 496 cases who had developed an incident primary CRC by June 2004, after excluding those diagnosed during the first year of follow up. A subcohort of 892 women was randomly sampled from the WHI-OS participants who had more than 12 months of follow-up and had no history of breast, colorectal, or endometrial cancer at 12 months after study enrollment, regardless of their cancer outcome thereafter. Of these, 433 cases and 821 subcohort members had baseline plasma samples available for the present study and did not have diabetes treatment at the time of blood draw (which would alter levels of insulin and leptin that were included in data analyses described later). There were 353 cases of colon cancer, 78 cases of rectosigmoid junction or rectal cancer, and 2 cases with unknown location of the CRC; 96% of the cases were diagnosed as adenocarcinoma.
The six analytes in EDTA plasma samples were assayed by the Millipore kits (EMD Millipore, Billerica, MA). The Milliplex Human Cytokine/Chemokine Panel was used to measure IL-1Ra; sIL-6R, sgp130, sIL-1R2, sTNFR1, and sTNFR2 were measured in a multiplex assay using the Milliplex Human Soluble Cytokine Receptor Panel. The Milliplex assays utilized the Luminex’s color-coded bead-based technology to achieve multiplexing (18).
The analytes were detectable in 99% to 100% of the samples (the lowest detectability was 99.2% for IL-1Ra). The inter-assay coefficients of variation (CVs) in our laboratory, which were determined from four control samples inserted into each of the 42 assay plates run over a period of time by the same technician, were 4.5% for sTNFR2, 5.9% for sTNFR1, 5.5% for sIL-6R, 6.6% for sgp130, 7.1% for sIL-1R2, and 10.7% for IL-1Ra. The 3-year intraclass correlation coefficients (ICC) of the six analytes were estimated from two previous studies. Based on three independent plasma samples collected over a 3-year period (baseline, year 1, and year 3) from each of 17 healthy women (19), the ICCs were 0.52 for sIL-6R, 0.63 for sgp130, 0.65 for sTNFR1, and 0.78 for sIL-1R2 (20). In another study, plasma levels of sTNFR2 and IL-1Ra at baseline and year 3 were measured in 148 subjects randomized to the placebo group of the Aspirin/Folate Polyp Prevention Trial (21). The 3-year ICCs were 0.56 for IL-1Ra and 0.72 for sTNFR2 (unpublished data). These ICC data suggest that a single measurement of circulating levels of the analytes under study in the baseline blood sample reflects an individual’s average levels over time.
From previous studies of this case-cohort study population, we found circulating levels of insulin, leptin, and estradiol to be significantly positively associated with CRC risk in multivariable analyses (7, 16). These variables were included in the data analyses described here. Although IL-6 and TNF-α were not significant risk factors in prior analyses, they were also included in data analyses, because they are ligands of the soluble receptors under study.
In univariable analyses, the prevalences of established CRC risk factors and analyte levels in quartiles between cases and subcohort subjects were compared using chi-square test. In the subcohort, we evaluated whether various CRC risk factors were associated with the analyte levels using Kruskal Wallis tests, and correlations among the analytes were estimated by Spearman rank correlation coefficients.
Multivariable analyses were conducted using Cox proportional hazard regression with robust variance estimation using the Self-Prentice method, which accounts for the case-cohort design in which cases may arise outside of or within the subcohort (22, 23). A base model was first developed to retain only the baseline covariates age (continuous), race (Whites vs. others), and CRC risk factors that were significant in multivariable analyses in our study population – smoking status (never, former, or current), ever had a colonoscopy, and estrogen level in four categories [serum estradiol in tertiles among women who were not using hormone therapy (<8, 8–13.9, or ≥14 pg/mL) or women using hormone therapy at baseline], waist circumference (continuous), insulin level (continuous), and leptin level (in quartiles due to its non-linear relationship with CRC risk). CRC risk factors that were not statistically significant in multivariable modeling were excluded from the base model (e.g., use of NSAIDs and alcohol, physical activity, family history of CRC, and folate and red meat intakes per day). We then assessed the hazard ratio (HR) for the association of each of the six analytes with CRC risk adjusting for the base model covariates. To be robust to any non-linear effect, all six analytes were categorized and analyzed in quartiles. Trend tests were performed using Wald tests associated with fitting the quartile categories as continuous variables in the regression model.
Although CRC cases diagnosed within the first year of follow-up were already excluded, sensitivity analyses were conducted to further examine whether reverse-causality might explain the analyte-disease association. Cases were separated into two groups defined by number of years from baseline recruitment to case diagnosis (>1 to 5 years or > 5 years). Subcohort members who had >1 year or > 5 years of follow-up were included in the corresponding 2 groups, respectively.
Table 1 shows the demographic factors and established risk factors for CRC in the cases and the subcohort. Briefly, as compared to the subcohort subjects, cases were older, more likely to be smokers, less physically active, and had higher BMI and waist circumference, with the latter having a stronger association with CRC than BMI. Cases were less likely to have had a colonoscopy, to have ever used oral contraceptives, and to be a hormone therapy user at baseline. Cases also had higher circulating levels of insulin and leptin and were more likely to have a moderately higher level of estradiol than the subcohort subjects. In terms of the six analytes under study, cases had lower plasma levels of sIL-6R and IL-1R2, but higher levels of sTNFR1 and sTNFR2 than the subcohort subjects. Table 1 also shows that the soluble receptors of IL-6 and TNF-α were circulating at a much higher concentration than their ligands (ng/mL vs. pg/mL).
Several of the demographic and lifestyle variables in Table 1 remained to be associated with CRC risk in multivariable analyses. Associations of the six analytes with these significant baseline risk factors for CRC are shown in Table 2. Sgp130, sTNFR1, and sTNFR2 increased with age. None was related to cigarette smoking. Use of hormone therapy was associated with reduced levels of all six analytes. Greater adiposity was associated with higher levels of IL-1Ra, sTNFR1, and sTNFR2. Concentrations of sIL-1R2, IL-1Ra, sTNFR1, and sTNFR2 increased with circulating levels of obesity-related factors (insulin and leptin).
Table 3 shows the correlations among the analytes and the ligands IL-6 and TNF-α. The highest correlations were between sTNFR1 and sTNFR2 (r = 0.51) and between TNF-α and sTNFR2 (r = 0.43). The ratios of soluble receptors to their ligands (e.g., sIL-6R/IL-6) were not meaningful indices, because they merely reflected the ligand levels and were highly correlated with them (|r| > 0.80).
The results of age-adjusted as well as multivariable adjusted analyses are shown in Table 4. After adjusting for the significant covariates in the base model (including age, race, smoking status, ever had colonoscopy, estrogen level, waist circumference, insulin level, and leptin level), two soluble cytokine receptors, sIL-6R and sIL-1R2, were inversely associated with CRC risk, with hazard ratios comparing extreme quartiles (HRQ4-Q1) of 0.56 for sIL-6R [95% confidence interval (CI) = 0.38 – 0.83, Ptrend = 0.007] and 0.44 for sIL-1R2 (95% CI = 0.29 – 0.67, Ptrend = 0.0004); sTNFR1 and sTNRF2 were no longer significantly associated with CRC risk. When all the 6 analytes and the ligands IL-6 and TNF-α were analyzed simultaneously, sIL-6R and sIL-1R2 remained significant (Table 4). In a saturated model, we adjusted for all the established risk factors, by including covariates that were not statistically associated with CRC into the base model (e.g., use of NSAIDs and alcohol, physical activity, family history of CRC, and folate and red meat intakes per day), and similar results were obtained (data not shown).
There were no significant interactions among the soluble cytokine receptors, receptor antagonist, IL-6, and TNF-α (data not shown), or between the study analytes and CRC risk factors (e.g., waist circumference, insulin, and hormone therapy, etc.). Results were similar when data were stratified by NSAID status at baseline (used regularly for ≥2 weeks or not) or when the 78 rectosigmoid junction or rectal cancer cases were excluded (data not shown).
When CRC cases were stratified by the number of years from baseline blood collection to case diagnosis in a sensitivity analysis (Table 5), sIL-6R was inversely associated with CRC risk only in those diagnosed between >1 to 5 years of baseline, but not in the cases diagnosed >5 years after baseline. In contrast, the inverse association between sIL-1R2 and CRC risk persisted regardless of the year of diagnosis (the last case was diagnosed 8.2 years after baseline in the study population reported here).
In this study, we found high levels of sIL-1R2, but not IL-1Ra, to be associated with a reduced risk of CRC. IL-1 signaling can be inhibited by its receptor antagonist (IL-1Ra) and soluble type 1 and type 2 IL-1 receptors (sIL-1R1 and sIL-1R2) (4, 13). We measured sIL-1R2 instead of sIL-1R1, because sIL-1R2 is the dominant soluble receptor and has greater affinity for IL-1 than sIL-R1 (4, 24). sIL-1R2 functions as a molecular decoy that prevents interaction of IL-1 with the signal-transducing type I receptor. Our finding of the inverse association between CRC risk and sIL-1R2 is consistent with the results of a study on Crohn’s disease, an inflammatory bowel disease associated with high risk of CRC, in which both circulating and mucosal sIL-1R2 levels were significantly higher in healthy controls than patients and treatment with corticosteroids induced a significant increase in sIL-1R2 (25). In accord with our present findings, in our previous study of patients with a history of colorectal adenoma, we did not observe any effects of IL-1Ra on adenoma recurrence (21).
Laboratory studies have shown that a very high concentration of sIL-1R2 relative to IL-1 is required to block IL-1 biological activity, since affinity of the soluble receptor is generally weaker than that of the membrane-bound receptor (4). We did not measure IL-1β in this study, because of its low circulating level. In a pilot study of 25 EDTA plasma samples from postmenopausal women, 36% of the samples had IL-1β levels below the assay limit of detection of 0.06 pg/mL (Milliplex High Sensitivity Human Cytokine Panel), and the median level among samples with a detectable level was 1.2 pg/mL. On the other hand, in the subcohort of this study, the median sIL-1R2 level was 7,385 pg/mL. It then appears that the circulating concentration of sIL-1R2 greatly exceeds that of IL-1β. As such, the ratio of sIL-1R2 to IL-1β levels, even if data on IL-1β were available in our study, would not be a useful indicator for the level of free IL-1β in circulation.
Although we found that a low level of sIL-6R was associated with increased CRC risk, this inverse association was only seen in the cases diagnosed in the first 5 years, suggesting the relatively low levels of sIL-6R in the baseline blood samples of cases might have arisen as a result of undiagnosed colorectal neoplasia at the time of the blood draw. A previous study of colorectal tumor tissue also indicated that sIL-6R level could be a marker for tumor growth (26). Specifically, this study showed that a low level of sIL-6R expression in tumor correlated with increased IL-6 expression and with disease progression, inferring consumption of sIL-6R by increased binding with IL-6 in the cancer stroma may favor tumor growth (26).
We did not find any associations of sTNFR1 and sTNFR2 with CRC. Similarly, there were no effects of sTNFR2 on adenoma recurrence in our previous study of patients with a history of colorectal adenoma (21). Contrarily, the Nurses’ Health Study found that increased sTNFR2 levels were associated with CRC risk (8). One possible explanation for this discrepancy is the fact that the assays used for sTNFR2 were different between this study and those used in the Nurses’ Health Study.
Similar to the situation of sIL-1R2 and IL-1β, the circulating concentrations of soluble receptors of IL-6 and TNF-α were several thousand-folds greater than those of the ligands (ng vs. pg). As such, their ratios (e.g., sIL-6R/IL-6 or sTNFR1/TNF-α) were not meaningful indices to make any biological inferences. Although one of the study goals was to examine how the cytokine ligands and their soluble receptors might act together to affect CRC risk, we could not assess this effectively. Nevertheless, we did not observe any interactive effects between the soluble receptors and their ligands on CRC risk.
Our study has other limitations. The mechanisms of regulation of circulating sIL-1R2 are unclear. The inverse association between sIL-1R2 and CRC risk might have been confounded by unmeasured protective factors for CRC that stimulate the release of sIL-1R2. Moreover, circulating levels of the soluble cytokine receptors and receptor antagonist may not reflect tissue levels. Finally, our results are not necessarily generalizable to men or to pre-menopausal women.
It is unlikely that our results were confounded by NSAID use, although close to 40% of the study population had used an NSAID regularly for two weeks or more at baseline. NSAID use was not associated with CRC risk in multivariate analyses in our study population. Even when NSAID use was added into the regression model, similar results were obtained. Moreover, there is no evidence in the literature to indicate that NSAID use affects the levels of soluble cytokine receptors and receptor antagonists. In fact, our previous data from an aspirin clinical trial demonstrated that low-dose aspirin had no effects on the circulating levels of sTNFR2 and IL-1Ra (21).
In summary, our data suggest that high circulating levels of sIL-1R2 may be protective against colorectal carcinogenesis or be a marker for reduced CRC risk. Further investigations of this soluble cytokine receptor are warranted for its potential as a risk prediction marker or as an immunologic agent for chemoprevention and therapy of CRC.
This work was supported by grant R01 CA122654 awarded to G Ho from the National Cancer Institute, National Institutes of Health. The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C. The short list of WHI centers and investigators can be found online at https://cleo.whi.org/researchers/Documents%20%20Write%20a%20Paper/WHI%20Investigator%20Short%20List.pdf.
The authors thank Elaine Cornell and Danielle Parent for conducting the laboratory measurements, and Dan Wang for assistance in data analyses.
Conflict of interest disclosure statement: Authors of this manuscript do not have any relationships that they believe could be construed as resulting in an actual, potential, or perceived conflict of interest with regard to the manuscript submitted for review.