Abstract
Background:
The major histocompatibility complex class I chain-related protein A (MICA) and protein B (MICB) participate in tumor immunosurveillance and may be important in colorectal cancer (CRC), but have not been examined in CRC development.
Methods:
sMICA and sMICB blood levels were measured by SomaScan in Visit 2 (1990–92, baseline) and Visit 3 (1993–95) samples in cancer-free participants in the Atherosclerosis Risk in Communities (ARIC) study. We selected rs1051792, rs1063635, rs2516448, rs3763288, rs1131896, rs2596542, and rs2395029 that were located in or in the vicinity of MICA or MICB and were associated with cancer or autoimmune diseases in published studies. SNPs were genotyped by the Affymetrix Genome-Wide Human SNP Array. We applied linear and Cox proportional hazards regressions to examine the associations of pre-selected SNPs with sMICA and sMICB levels and CRC risk (236 CRCs, 8,609 participants) and of sMICA and sMICB levels with CRC risk (312 CRCs, 10,834 participants). In genetic analyses, estimates adjusted for ancestry markers were meta-analyzed.
Results:
Rs1051792-A, rs1063635-A, rs2516448-C, rs3763288-A, rs2596542-T, and rs2395029-G were significantly associated with decreased sMICA levels. Rs2395029-G, in the vicinity of MICA and MICB, was also associated with increased sMICB levels. Rs2596542-T was significantly associated with decreased CRC risk. Lower sMICA levels were associated with lower CRC risk in males (HR=0.68, 95%CI 0.49–0.96) but not in females (p-interaction=0.08).
Conclusions:
Rs2596542-T associated with lower sMICA levels was associated with decreased CRC risk. Lower sMICA levels were associated with lower CRC risk in males.
Impact:
These findings support an importance of immunosurveillance in CRC.
Keywords: colorectal cancer risk, immunosurveillance
Introduction
Colorectal cancer (CRC) is the third leading cause of cancer death and the third most common cancer in men and women in the U.S (1). One of the mechanisms affecting cancer development, including CRC, is tumor immunosurveillance, which allows the host immune system to identify and clear tumor cells (2–4). Therefore, understanding the interaction between colorectal tumor cells and the immune response may inform interception strategies.
Major histocompatibility complex class I-like molecules (MICA and MICB) are transmembrane proteins that may be important in tumor immunosurveillance (5). MICA and MICB serve as major ligands for natural-killer group 2, member D (NKG2D) receptors (6,7). NKG2D, which are present on natural killer (NK) cells, can recognize MICA and MICB expressed on the surface of abnormal cells and trigger NK killing (8–10). In line with this, two studies of CRC patients from China (N=863 in the primary cohort and N=556 in the validation cohort) (11) and from the United Kingdom (N = 449) (12) found that higher MICA and MICB expression in their tumors was associated with better survival. However, a smaller study of 182 CRC tumors from GSE41258 and 65 CRC tumors from GSE29621 found that higher MICA expression was associated with worse prognosis (13).
Cancer cells may escape NK cell immune surveillance by releasing MICA and MICB from the cell surface into a soluble form (sMICA and sMICB) and can be detected in blood (14,15). Many studies have investigated whether sMICA and sMICB are linked to cancer survival; a meta-analysis of 13 studies of sMICA or sMICB examined together showed that their high levels were associated with a poorer overall survival of cancer patients, although that meta-analysis did not include individuals with CRC (9). In addition, several studies have examined associations of MICA and MICB polymorphisms with cancer risk (16–22). To our knowledge, no studies have examined the associations of MICA and MICB polymorphisms or sMICA and sMICB levels with CRC risk.
Given the important role of MICA and MICB in tumor immunosurveillance and highly polymorphic nature of the MICA and MICB genes (9,18,23–26), we examined whether MICA and MICB single nucleotide polymorphisms (SNPs) and sMICA and sMICB blood levels were associated with CRC risk. We selected seven SNPs that were located either in or in the vicinity of MICA or MICB and were previously associated with risk of different types of cancers, including breast (16), hepatocellular (19,20,27), and cervical (21,22), or autoimmune diseases, such as rheumatoid arthritis (28), Takayasu arteritis (29), and psoriasis (30) (Supplemental Table 1). Of those SNPs, rs2516448 is approximately 7.5 kb downstream of MICA, and rs2395029 is approximately 50 kb downstream of MICA and 35 kb upstream of MICB. All the other pre-selected SNPs are located in MICA (31). Using the Atherosclerosis Risk in Communities (ARIC) prospective cohort, we examined: 1) the associations between the pre-selected SNPs and protein levels (sMICA or sMICB) (Aim 1), and 2) the associations of the pre-selected SNPs and sMICA and sMICB levels with CRC risk (Aim 2). We hypothesized that the pre-selected SNPs associated with lower levels of sMICA and sMICB were associated with a lower CRC risk.
Materials and Methods
Study population
The ARIC study (RRID:SCR_021769) is a prospective cohort initiated in 1987 (32,33). In 1987–89 (Visit 1), 15,792 volunteers aged 45–64 were recruited from four study centers -- Maryland, Minnesota, Mississippi, and North Carolina. Participants in Maryland and Minnesota were primarily White, and the recruitment in Mississippi was restricted to Black residents. The ARIC study was approved by institutional review boards at each participating center, and all study participants provided written informed consent. Thus far, nine visits have been completed (32). Additionally, ARIC participants have received follow-up telephone calls annually in 1987–2012 and semi-annually after 2012, with response rates of 90%−99% for the annual follow-up calls and 83%−90% for semi-annual follow-up calls among living participants who have not withdrawn consent to be contacted (33).
Ascertainment of CRC cases and death
Incident CRC cases were ascertained through 2015 via linkage with state cancer registries in Maryland, Minnesota, Mississippi, and North Carolina. These records were supplemented by the abstraction of medical records and hospital discharge codes (33). Deaths were identified through annual (semi-annual since 2012) follow-up telephone calls to participants or their proxies, surveillance of local hospitals, state records and linkage to the National Death Index.
Blood collection
The ARIC protocol for plasma sample collection, processing, and storage was designed to minimize the spontaneous biochemical reactions after blood collection and is consistent with the recommended practice for proteomics data analysis in epidemiological studies (34–36). After venipuncture, blood samples were put immediately in an ice water bath and centrifuged at 15–25 °C within 10 min of venipuncture. Then, the aliquots were stored at −80 °C within 90 min of venipuncture and were never thawed before this analysis (https://sites.cscc.unc.edu/aric/sites/default/files/public/manuals/Blood_Collection_and_Processing.2_7.pdf).
Measurement of sMICA and sMICB levels and quality control
We examined sMICA and sMICB levels measured in EDTA-plasma samples collected at Visit 2 (1990–92) and Visit 3 (1993–95). Samples were analyzed using a SOMAmer (Slow Off-rate Modified Aptamers)-based capture array called SomaScan® by Somalogic, Inc. (Boulder, CO, USA) (37–40). The description of the SomaScan assay and the data normalization process have been described in published papers (36,40,41). In ARIC, the median split sample reliability coefficients were 0.93 at Visit 2 and 0.88 at Visit 3 after excluding proteins with a Bland-Altman coefficient of variation greater than 50% or a variance of less than 0.01 on the log scale, or proteins binding to mouse Fc-fusion, contaminants, or non-proteins (42). sMICA and sMICB levels were measured in relative fluorescent units (RFU) and were log2 transformed to correct for skewness.
Genotyping of MICA and MICB genetic variants
The SNPs were genotyped using the Affymetrix Genome-Wide Human SNP Array 6.0 and analyzed with the Birdseed variant-calling algorithm (43). Measured SNPs used for imputation were restricted to SNPs with minor allele frequencies >1%, call rate >95%, and Hardy–Weinberg equilibrium P-values >0.00001 (44). Individuals with cryptic relatedness, defined as an identity-by-state distance >0.8, generated using PLINK, were also excluded (45). Genotypes were then imputed on the Michigan Imputation Server to the TOPMed reference panel R1, and the resulting dosage values were used in subsequent analyses (46,47). Ancestry markers, principal components (PCs) based on the GWAS data, were generated by EIGENSTRAT (48) to reflect the population structure or genetic ancestry of the ARIC participants.
Assessment of other participants’ characteristics
Other variables of interest that were collected at Visits 1–3 included demographic characteristics and CRC risk factors, namely age, sex, race, study center, cigarette smoking, height, weight, hormone replacement therapy (females only), aspirin use, and diabetes status (49). Information on smoking, use of medications and the measures of height and weight were collected at each visit. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared at each visit. The detailed procedures for assessing pack-years of smoking have been published (50). Diabetes mellitus was defined as fasting glucose ≥126 mg/dL, non-fasting glucose ≥200 mg/dL, treatment for diabetes mellitus, or self-reported physician diagnosis of diabetes. Another variable of interest was estimated glomerular filtration rate (eGFR) because it was associated with different SomaScan protein levels in a pilot ARIC study, and it was associated with sMICA and sMICB levels in our analysis. eGFR (ml/min per 1.73 m2) at Visits 2 and 3 was calculated based on serum creatinine and Cystatin C by incorporating age and sex (51). We were also interested in the inflammatory and immune biomarkers, C-reactive protein (CRP) and beta-2 microglobulin (B2M), that were measured in blood samples collected at Visit 2 (52,53).
Statistical analysis
The meta-analyses were performed using R statistical software, version 4.1.2, package “metafor”. All the other analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC). In all analyses, a p-value < 0.05 was considered statistically significant, because we had pre-specified hypotheses for each biomarker and each SNP (54). Dosage values of the pre-selected SNPs were modeled as continuous variables unless stated otherwise.
We excluded participants with missing sMICA and sMICB levels at Visit 2 (N=2,550), participants with race other than White or Black (N=37), and those who reported prevalent cancer at Visit 2 (N=927), resulting in 10,834 White and Black participants without a history of cancer at Visit 2. Among those participants, 312 developed CRC through the end of 2015. For the analyses involving SNPs, we additionally excluded 1,882 participants who had missing SNPs information, resulting in 8,952 participants (253 incident CRC cases through the end of 2015). Accounting for ancestry markers in the analyses of SNPs led to an additional exclusion of 343 participants who had missing information on ancestry markers, resulting in 8,609 participants (236 incident CRC cases through the end of 2015) (Figure 1).
Figure 1. Study population.
The gray boxes show the participants excluded from the study.
Demographic and lifestyle/medical characteristics at Visit 2 were examined across quartiles of Visit 2 sMICA and sMICB levels as mean (standard deviation (SD)) or percentage (%). The Pearson correlation coefficient was calculated between sMICA and sMICB levels.
For Aim 1, we used linear regression to assess associations between the pre-selected SNPs and corresponding sMICA and sMICB levels (modeled as continuous variables) at Visit 2. For Aim 2, we used Cox proportional hazards regression to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for incident CRC in relation to the pre-selected SNPs and Visit 2 sMICA and sMICB levels (modeled as continuous variables per 1 RFU decrease, and in quartiles). For each participant, person-years were determined from Visit 2 date until the diagnosis of CRC or another cancer, death, loss to follow-up, or administrative censoring on December 31, 2015, whichever occurred first. For both aims 1 and 2, we adjusted for age, sex, and joint terms for race and study center (Black participants from Mississippi; Black participants from any of the other field centers; White participants from Maryland or North Carolina; and White participants from Minnesota), education, BMI, cigarette smoking, pack-years of smoking, hormone replacement therapy (in women), aspirin use, diabetes status, and eGFR at Visit 2. We adjusted for these variables because they were associated with sMICA or sMICB levels and with CRC risk. In addition, we tested for B2M and CRP as potential confounders, but they changed the estimate by less than 2% and were not included into the model. Finally, we included both sMICA and sMICB levels simultaneously in the model when examining the association with CRC risk.
The analyses of pre-selected SNPs with corresponding sMICA and sMICB levels and risk of CRC were further adjusted for ancestry markers (two ancestry-specific PCs for White individuals and four ancestry-specific PCs for Black individuals). We used fixed effects model to meta-analyze the race-specific estimates if the estimates were not statistically significantly different from each other (p-heterogeneity > 0.05). We used random effects model to meta-analyze the race-specific estimates if the estimates were statistically significantly different from each other (p-heterogeneity ≤ 0.05) but were in the same direction. We did not meta-analyze the race-specific estimates if the estimates were in opposite directions. For the analyses of seven pre-selected SNPs and CRC risk, we accounted for the multiple testing using the False Discovery Rate (FDR) p-values.
In both aims, we also examined whether sex modified these associations since immune response may differ by sex (55,56). We also examined whether race or diabetes modified these associations because Black individuals had a higher CRC incidence (57), and individuals with diabetes might have a higher CRC risk versus those without diabetes (58).
Exploratory and sensitivity analyses
We conducted several exploratory analyses. In the exploratory analyses of pre-selected SNPs and CRC risk, we examined if the associations between the pre-selected SNPs and CRC risk were mediated by sMICA or sMICB levels, by including sMICA or sMICB levels to the models. Further, we examined the associations of pre-selected SNPs and sMICA and sMICB levels with CRC risk stratified by the family history of CRC and by subtypes of CRC – proximal colon cancer, distal colon cancer, or rectal cancer.
We also conducted two sensitivity analyses. Because we have a relatively small number of participants who had information on both SNPs and ancestry markers (N = 8,609) compared to all participants with SNPs’ information (N = 8,952), to increase the sample size, we also examined these associations without adjustment for ancestry markers in all the 8,952 participants with SNPs’ data. In this sensitivity analysis, for rs1131896 and sMICA levels, we reported the estimates for White and Black participants instead of overall, because these race-specific estimates were in opposite directions. In the other sensitivity analysis, we calculated sMICA and sMICB levels as the mean of their levels at Visits 2 and 3. The mean was used to reflect the usual levels of these proteins. We repeated the main analyses using the mean of protein levels instead of protein levels at Visit 2. For this sensitivity analysis, the follow up of participants started at Visit 3.
Data availability
The data used in this study are available from the corresponding author upon reasonable request (https://sites.cscc.unc.edu/aric/pubs-policies-and-forms-pg).
Results
The study included 10,834 White and Black participants without a history of cancer at Visit 2 who were followed up for 199,821 person-years. Among them, 312 incident CRC cases occurred until 2015. Those with lower sMICA levels at Visit 2 were less likely to be White and tended to have higher eGFR and lower B2M. Those in the highest quartile of sMICA levels were more likely to have diabetes and were less likely to be current smokers and ever users of hormone replacement therapy. Those with lower Visit 2 sMICB levels were less likely to be male and White, were more likely to be current smokers, and were more likely to have higher eGFR and lower B2M (Table 1). sMICA and sMICB levels were not correlated (r=0.01, p=0.24).
Table 1.
Baseline characteristics across quartiles of sMICA/sMICB levelsa of 10,834 participants in the ARIC study at Visit 2 (1990–1992)
| sMICA level quartiles | sMICB level quartiles | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
|
||||||||||
| Q1 (N = 2,708) | Q2 (N = 2,708) | Q3 (N = 2,711) | Q4 (N = 2,707) | p-value | Q1 (N =2,706) | Q2 (N = 2,708) | Q3 (N = 2,712) | Q4 (N = 2,708) | p-value | |
| Range of sMICA/sMICB levels | 6.49 – 7.68 | 7.69 – 8.00 | 8.01 – 8.31 | 8.32 – 11.13 | 7.14 – 8.39 | 8.40 – 8.63 | 8.64 – 8.92 | 8.93 – 12.16 | ||
|
| ||||||||||
| Demographic | ||||||||||
| Age, years (SD) | 56.7 (5.7) | 56.5 (5.7) | 57.0 (5.8) | 57.6 (5.6) | <0.0001 | 56.2 (5.6) | 56.9 (5.6) | 57.1 (5.7) | 57.6 (5.8) | <0.0001 |
| Male (%) | 44.9 | 43.7 | 46.3 | 46.0 | 0.21 | 39.2 | 44.0 | 47.9 | 49.8 | <0.0001 |
| White (%) | 69.0 | 73.6 | 78.6 | 81.5 | <0.0001 | 61.8 | 74.2 | 82.5 | 84.3 | <0.0001 |
| Education (%) | ||||||||||
| Less than high school | 22.8 | 21.9 | 20.6 | 22.7 | 0.34 | 24.3 | 22.3 | 19.1 | 22.3 | 0.0005 |
| High school equivalent | 40.8 | 42.3 | 42.5 | 42.4 | 39.7 | 42.2 | 43.7 | 42.4 | ||
| Greater than high school | 36.3 | 35.8 | 36.9 | 34.9 | 35.9 | 35.5 | 37.2 | 35.3 | ||
| Lifestyle/Medical factors | ||||||||||
| BMI, kg/m2 (SD) | 27.9 (5.3) | 27.8 (5.2) | 28.0 (5.4) | 28.2 (5.5) | 0.01 | 28.3 (5.4) | 28.2 (5.5) | 27.6 (5.2) | 27.8 (5.3) | <0.0001 |
| Smoking status | ||||||||||
| Current smoker | 23.4 | 24.7 | 22.4 | 18.9 | <0.0001 | 25.8 | 23.1 | 20.8 | 19.8 | <0.0001 |
| Former smoker | 37.6 | 37.6 | 37.7 | 38.8 | 33.3 | 36.8 | 39.6 | 42.0 | ||
| Never smoker | 39.0 | 37.7 | 39.9 | 42.4 | 41.0 | 40.1 | 39.7 | 38.3 | ||
| Pack-years of smoking among current and former smokers, pack-years (SD) | 27.53 (22.26) | 27.97 (22.05) | 29.12 (20.98) | 29.68 (23.10) | 0.03 | 28.7 (21.8) | 27.0 (20.7) | 28.9 (22.3) | 29.6 (23.5) | 0.0101 |
| Diabetes (%) | 13.3 | 12.3 | 15.6 | 19.6 | <0.0001 | 14.8 | 14.2 | 15.2 | 16.6 | 0.08 |
| Aspirin use in the past two weeks (%) | 50.3 | 50.0 | 52.3 | 51.3 | 0.29 | 49.1 | 50.9 | 52.7 | 51.3 | 0.06 |
| Ever users of hormone replacement therapy among females (%) | 44.5 | 49.1 | 44.6 | 41.1 | 0.0002 | 45.3 | 45.2 | 46.0 | 42.9 | 0.39 |
| eGFR, mL/min/1.73 m2 (SD) | 99.2 (16.4) | 99.3 (16.4) | 98.2 (16.4) | 96.3 (16.9) | <0.0001 | 99.8 (16.2) | 99.2 (15.9) | 98.1 (16.1) | 95.8 (17.8) | <0.0001 |
| CRP, mg/L (SD) | 4.4 (8.3) | 4.4 (6.8) | 4.3 (6.9) | 4.5 (7.9) | 0.72 | 4.6 (7.5) | 4.3 (7.5) | 4.4 (7.7) | 4.5 (7.1) | 0.26 |
| B2M, mg/L (SD) | 1.9 (1.5) | 1.9 (1.2) | 2.0 (1.0) | 2.1 (1.3) | <0.0001 | 1.9 (0.6) | 1.9 (0.7) | 2.0 (1.3) | 2.1 (1.9) | <0.0001 |
sMICA and sMICB levels were log2 transformed and were measured in relative fluorescent units (RFU).
Associations between the pre-selected SNPs and corresponding sMICA and sMICB levels at Visit 2
In the main analysis that meta-analyzed the race-specific estimates, six of seven pre-selected SNPs, rs1051792-A, rs1063635-A, rs2516448-C, rs3763288-A, rs2596542-T, and rs2395029-G, were associated with lower sMICA levels (Table 2). Rs1131896-A was associated with lower sMICA levels among White participants, but with higher sMICA levels among Black participants (p-heterogeneity < 0.01) (Table 2 and Supplemental Table 2). For two other SNPs, rs2516448-C and rs3763288-A, the associations with sMICA levels were statistically different in White and Black participants (p-heterogeneity < 0.05) (Supplemental Table 2), although the direction of the associations in White and Black participants were the same. Rs2395029-G was also associated with higher sMICB levels (Table 2), and this association was significantly different across race groups (p-heterogeneity = 0.01) (Supplemental Table 2), although the associations were in the same direction. Sex did not statistically modify any associations between pre-selected SNPs and corresponding sMICA or sMICB levels, while diabetes statistically modified the associations of rs1051792 and rs2596542 with sMICA levels, but the associations were in the same direction for those with and without diabetes (Supplemental Table 2).
Table 2.
Associations of pre-selected SNPs with Visit 2 sMICA and sMICB levelsa and with CRC risk, ARIC (1990–2015)
| SNPS | Reference/Effect allele | Association with sMICA or sMICB | Association with CRC | ||||||
|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||
| sMICA or sMICB | Race group | Regression coefficient (95% CI)b | Meta-analysis Regression coefficient | P-value | HR (95% CI)a | Meta-analysis HR | FDR P-valuec | ||
|
| |||||||||
| rs1051792 | G/A | sMICA | White | −0.50 (−0.52, −0.49) | −0.51 (−0.52, −0.40) | <0.0001 | 0.86 (0.67, 1.11) | 0.82 (0.66, 1.01) | 0.13 |
| Black | −0.53 (−0.55, −0.50) | 0.71 (0.48, 1.06) | |||||||
| rs1131896d | G/A | sMICA | White | −0.10 (−0.12, −0.08) | -- | -- | 1.06 (0.83, 1.37) | 1.06 (0.85, 1.31) | 0.71 |
| Black | 0.23 (0.19, 0.26) | 1.02 (0.68, 1.54) | |||||||
| rs1063635 | G/A | sMIA | White | −0.39 (−0.41, −0.38) | −0.39 (−0.40, −0.38) | <0.0001 | 0.93 (0.74, 1.16) | 0.87 (0.72, 1.06) | 0.24 |
| Black | −0.39 (−0.42, −0.35) | 0.72 (0.49, 1.07) | |||||||
| rs2516448 | T/C | sMICA | White | −0.48 (−0.50, −0.47) | −0.42 (−0.54, −0.30) | <0.0001 | 0.90 (0.72, 1.13) | 0.85 (0.70, 1.03) | 0.19 |
| Black | −0.36 (−0.40, −0.33) | 0.72 (0.49, 1.07) | |||||||
| rs3763288 | G/A | sMICA | White | −0.37 (−0.40, −0.33) | −0.31 (−0.46, −0.15) | <0.0001 | 0.93 (0.60, 1.42) | 0.92 (0.61, 1.41) | 0.71 |
| Black | −0.21 (−0.36, −0.06) | 0.78 (0.10, 5.81) | |||||||
| rs2596542 | C/T | sMICA | White | −0.39 (−0.41, −0.37) | −0.40 (−0.41, −0.38) | <0.0001 | 0.72 (0.56, 0.92) | 0.75 (0.61, 0.92) | 0.04 |
| Black | −0.41 (−0.44, −0.38) | 0.82 (0.55, 1.18) | |||||||
| rs2395029 | T/G | sMICA | White | −0.36 (−0.41, −0.31) | −0.36 (−0.41, −0.31) | <0.0001 | White: 0.43 (0.18, 1.03) Black: unknown |
unknown | 0.13 |
| Black | −0.35 (−0.58, −0.12) | ||||||||
| sMICB | White | 0.81 (0.76, 0.86) | 0.70 (0.43, 0.96) | <0.0001 | |||||
| Black | 0.54 (0.33, 0.75) | ||||||||
sMICA and sMICB levels were measured in relative fluorescent units (RFU).
Regression coefficient and HR per 1 effect allele and model adjusted for age, sex, ancestry markers, study center, education, BMI, hormone replacement therapy (in women), aspirin use, smoking status, pack-years of smoking, diabetes status, and eGFR.
False discovery rate (FDR) p-value was calculated as the multiple testing of the seven pre-selected SNPs.
The race-specific estimates for the association between rs1131896 and sMICA levels were not meta-analyzed because the race-specific estimates were in opposite directions.
Associations of the pre-selected SNPs and of sMICA and sMICB levels at Visit 2 with subsequent CRC risk
In the main analysis that meta-analyzed the race-specific estimates, only rs2596542-T was significantly associated with a lower CRC risk (Table 2). Sex, race, or diabetes did not statistically modify this association (Supplemental Table 3). Another pre-selected SNP, rs1051792-A, was associated with a lower CRC risk in males but not females (p-interaction = 0.29) (Supplemental Table 3). There were no associations of sMICA and sMICB levels (modeled as continuous variables per 1 RFU decrease, or in quartiles) with CRC risk (Table 3 and Supplementary Table 4). However, lower sMICA levels appeared to be associated with a decreased CRC risk in males (HR=0.68, 95% CI 0.49–0.96) but not in females (p-interaction=0.08) (Table 3). Race and diabetes status at Visit 2 did not statistically modify the associations of sMICA and sMICB levels with CRC risk (Table 3). When both sMICA and sMICB levels were included in the same model, the HR (95% CI) for CRC risk associated with sMICA was 0.83 (0.66, 1.05) and with sMICB was 1.10 (0.87, 1.39). These HRs were the same as the results for CRC risk when sMICA and sMICB levels were examined individually.
Table 3.
Cox Regression for the associations of Visit 2 sMICA and sMICB levels (continuous variables per 1 relative fluorescent unit (RFU) decrease) with CRC risk in all participants and stratified by sex, race, or diabetes status; ARIC (1990–2015)
| N of incident CRC cases | Total person-years | HR (95% CI)a | ||
|---|---|---|---|---|
|
| ||||
| sMICA level | sMICB level | |||
|
| ||||
| All participants | ||||
|
| ||||
| 312 | 199,821 | 0.83 (0.66, 1.05) | 1.10 (0.87, 1.39) | |
|
| ||||
| Stratified by sex | ||||
|
| ||||
| Male | 161 | 83,933 | 0.68 (0.49, 0.96) | 0.98 (0.72, 1.33) |
| Female | 151 | 115,888 | 0.99 (0.71, 1.38) | 1.29 (0.90, 1.85) |
| P-interaction | 0.08 | 0.17 | ||
|
| ||||
| Stratified by race | ||||
|
| ||||
| White | 221 | 153,954 | 0.84 (0.64, 1.12) | 1.20 (0.91, 1.58) |
| Black | 91 | 45,867 | 0.80 (0.51, 1.25) | 0.87 (0.55, 1.36) |
| P-interaction | 0.97 | 0.37 | ||
|
| ||||
| Stratified by diabetes status | ||||
|
| ||||
| Yes | 59 | 25,983 | 0.79 (0.48, 1.31) | 0.83 (0.53, 1.31) |
| No | 253 | 173,838 | 0.83 (0.64, 1.08) | 1.20 (0.91, 1.58) |
| P-interaction | 0.86 | 0.18 | ||
Model was adjusted for age, sex, joint terms for race and study center (Black participants from Mississippi; Black participants from any of the other field centers; White participants from Maryland or North Carolina; and White participants from Minnesota), education, BMI, hormone replacement therapy (in women), aspirin use, smoking, pack-years of smoking, diabetes status, and eGFR.
Exploratory analyses for CRC risk
In the exploratory analysis of pre-selected SNPs and CRC risk, the associations did not change markedly (less than or equal to 10%) after additional adjustment for sMICA or sMICB levels (Supplemental Table 5). Therefore, the impact of the pre-selected SNPs on the development of CRC is mainly not through sMICA or sMICB levels. In the exploratory analysis of risk of subtypes of CRC, most of the estimates for the risk of proximal colon cancer and distal colon cancer were similar (Supplemental Table 6). The estimates for the risk of rectal cancer appeared to be different for sMICB levels and several pre-selected SNPs, including a stronger inverse association for rs2596542 with risk of rectal cancer (HR = 0.49, 95% CI 0.26–0.93) than with risk of CRC (HR = 0.75, 95% CI 0.61–0.92), but the number of rectal cancer cases was small (≤44 across different analyses) (Supplemental Table 6). In the analysis stratified by family history of CRC, family history did not modify the associations of sMICA or sMICB levels or pre-selected SNPs with CRC risk (p-interactions = 0.37–0.97) (Supplemental Table 7).
Sensitivity analyses
In the sensitivity analysis of pre-selected SNPs with sMICA and sMICB levels without adjustment for ancestry markers, the associations between all pre-selected SNPs and corresponding sMICA and sMICB levels were similar to the associations in the main analysis (Supplemental Table 8). In the sensitivity analysis of CRC risk, without adjustment for ancestry markers, two additional SNPs, rs1051792-A and rs2395029-G, examined individually, were significantly associated with a lower CRC risk (Supplemental Table 8). The difference between the findings in the main and sensitivity analyses for rs2395029-G may be because the sensitivity analysis included both White and Black participants while Black participants were excluded in the main analysis because of a small number of Black participants with one or two effect alleles. For rs1051792-A and CRC risk, it is possible that the association was not significant in the main analysis due to a smaller sample size of 8,609 participants compared to 8,952 participants in the sensitivity analysis or just due to chance, but this should be tested in larger studies. The associations of pre-selected SNPs with the mean of sMICA and sMICB levels at Visits 2 and 3 (Supplementary Table 9) and the associations of the mean of sMICA and sMICB levels at Visits 2 and 3 with CRC risk (Supplementary Table 10) were similar to the findings in our main analysis, respectively.
Discussion
In this large prospective study of White and Black individuals, we found associations between the pre-selected SNPs, located in or in the vicinity of the MICA or MICB genes, and corresponding sMICA and sMICB levels. We found that rs2596542-T that was significantly associated with decreased sMICA levels was also significantly associated with a decreased CRC risk. Although no association was observed for sMICA or sMICB levels with CRC risk overall, lower sMICA levels were associated with a decreased risk of CRC in males but not females.
The interaction between MICA and MICB and receptor NKG2D on the NK cells is essential for the activation of NK cells against tumor cells (9,59). NKG2D can recognize MICA and MICB expressed on the tumor cell surface and trigger the activation of NK cells to target and eliminate tumor cells (8–10). In agreement with this, two previous studies reported that higher MICA and MICB expression in CRC cells was associated with a favorable outcome (11,12). One of these studies found that a lower versus higher MICA expression in 449 colorectal tumor samples was associated with worse disease-specific survival: HR =1.50, 95%CI 1.14–1.97 (12). Likewise, the second study reported better overall survival in CRC patients with higher MICB expression (N=863 in the primary cohort and N=556 in the validation cohort): for those with higher versus lower MICB expression, HR was 0.74 (95% CI 0.59–0.92) in the primary cohort and HR was 0.70 (95% CI 0.51–0.96) in the validation cohort (11). However, cancer cells may escape the immune response involving NKG2D ligands through the release of MICA and MICB via proteolytic cleavage into circulation (59,60). Circulating sMICA and sMICB can bind to NKG2D receptor and induce its downregulation and therefore decrease immune response (5,61,62). In agreement with this mechanism, a meta-analysis of 13 studies of several cancer types combined, reported that high levels of sMICA or sMICB were associated with a poorer overall survival: HR=1.65, 95%CI 1.42–1.92 compared to the low levels (9).
In our study, lower sMICA levels were associated with a decreased CRC risk only among males. Although the exact explanation for this difference is unknown, the absence of the association in females may be explained by the stronger immune response and higher activity of NK cells in females so that NK cells can exert their cytolytic effect at any levels of sMICA and the sMICA does not impact CRC (55,56). However, the p-interaction between sex and sMICA levels was 0.08, and the possibility of the findings by chance cannot be excluded.
Rs2596542 (C/T), a well-studied polymorphism, is located in the promoter region of MICA (63), and may initiate and promote gene expression. In our study, we found that rs2596542-T was statistically significantly associated with lower sMICA levels in cancer-free participants at Visit 2. Our result is in agreement with other studies that observed lower sMICA levels in healthy controls with rs2596542-T (19). Also, we found that the rs2596542-T was significantly associated with a lower CRC risk, which is in line with this SNP’s association with lower sMICA levels. No other studies examined this SNP and CRC risk; however, previous studies investigated the association between this SNP and hepatocellular carcinoma risk. A meta-analysis of 11 studies including two European studies, reported a positive association in Whites (CC vs. CT+TT: OR=1.56, 95% CI 1.12–2.18) (20), i.e., rs2596542-T was associated with lower hepatocellular carcinoma risk, which is consistent with our findings. Of note, in our study, the association between rs2596542-T and CRC risk was inverse in both White (HR=0.72, 95% CI 0.56–0.92) and Black (HR=0.82, 95% CI 0.55–1.18) participants.
In our study, we also examined a functional MICA-129 polymorphism (rs1051792) that causes a replacement of valine (Val) by methionine (Met) in the position 129 (the change from G to A) (16,64). MICA-129Met variant (rs1051792-A) has a higher affinity of binding MICA to NKG2D receptor than MICA-129Val (rs1051792-G) (64,65). MICA-129Met is more prone to shedding than MICA-129Val (24,25). Thus, we expected a greater shedding of MICA-129Met; however, in our study, those with MICA-129Met (rs1051792-A) had significant lower mean sMICA levels in both main and sensitivity analyses and a significant lower CRC risk in the sensitivity analysis. Although the higher shedding and lower blood levels seem contradictory, previous studies (24,66) also observed this phenomenon: a study of 552 patients infected with hepatitis B virus and 418 healthy controls found that MICA-129Val (rs1051792-G) was associated with higher sMICA levels in both patients and controls (19); a study of 73 patients diagnosed with oral squamous cell carcinoma and 149 healthy controls reported significantly higher sMICA levels in both cases and controls with MICA-129Val/Val versus other genotypes (24) and another study of 91 patients diagnosed with multiple myeloma reported higher sMICA levels in patients with MICA-129Val/Val versus other genotypes (66). These findings were explained by intracellular retention among those with MICA-129Met as observed in transfected cells (67). Besides the reported association between rs1051792 and sMICA levels, rs1051792-A (MICA-129Met) was associated with a lower breast cancer risk in a case-control study of 192 breast cancer patients and 205 age-matched healthy controls from Tunisia (A vs. G: OR=0.61, 95% CI 0.44–0.84) (16). However, there have been no reports of studies examining this SNP and CRC risk. In our study, rs1051792-A was significantly associated with decreased CRC risk only in the analysis that did not adjust for ancestry markers (HR=0.78, 95% CI 0.63–0.96; 253 CRC cases in 8,952 participants) but not in the meta-analysis that pooled the race-specific estimates adjusted for ancestry markers (HR=0.82, 95% CI 0.66–1.01); however, the latter sample size was smaller.
The other SNP under study, rs2395029, is approximately 50 kb downstream of MICA and 35 kb upstream from MICB (31). This SNP was found to be associated with psoriasis, an autoimmune disease (30), but there were no reports of studies examining this SNP and cancer. In our study, we found that rs2395029-G was associated with higher sMICB levels, but with lower sMICA levels. HRs (95% CIs) for the association between this SNP and colorectal cancer risk were 0.43 (0.18–1.03) with and 0.36 (0.15–0.88) without adjustment for ancestry markers. Future larger studies will be needed to test our results for rs1051792 and rs2395029 with CRC risk.
Two other SNPs, rs1131896 and rs1063635, which were previously associated with hepatocellular carcinoma risk (19), were not associated with CRC risk in our study. Our results are similar to those in a previous study that did not detect different distributions of these SNPs’ genotypes among gastric cancer patients and healthy controls (18). In addition, another pre-selected SNP, rs2516448, was previously associated with cervical cancer risk (21,22), and pre-selected rs3763288 was associated with rheumatoid arthritis and Takayasu arteritis (28,29). However, we did not detect any associations between these SNPs and CRC risk.
To the best of our knowledge, our study is the first to report on the associations between MICA and MICB polymorphisms with sMICA and sMICB levels measured by SomaScan assay. Only a few studies, including two studies of individuals from Europe (24,66), one study of Vietnamese (19), and a previous study (mainly White participants) conducted by our group (17), examined the associations of rs1051792, rs1131896, and rs2596542 with sMICA levels measured by other assays, such as Luminex Bead-based assay or ELISA. The associations of rs1051792 and rs2596542 with sMICA levels reported in our current study and previous studies were in the same direction (17,19,24,66). However, for rs1131896 and sMICA levels, we found that the associations were in opposite directions for White and Black participants; and the direction of association reported in the previous study of Vietnamese (19) was the same as the association in Black participants in our study. Additional studies are needed to validate our results. Our study is also the first to document associations of MICA SNP, rs2596542, with CRC risk. The association with rs2596542 confirmed our hypothesis that rs2596542-T associated with lower sMICA levels was also associated with a lower CRC risk.
The strengths of this population-based study include the prospective design with over 20 years of follow-up, adjudicated CRC incidence, the large community-based sample of White and Black individuals, and the availability of protein levels and genetic variants in the same study. Our study may have some limitations. First, the possibility of protein degradation during long-term storage cannot be excluded. However, the blood samples were frozen right after their collection and have never been thawed reducing the possibility of degradation. In addition, ARIC samples stored for different time showed similar performance as indicated by the similar coefficient of variation (CV) for the split samples collected at Visit 2 and Visit 5 (2011–13) (CV = 6% at Visit 2 and 7% at Visit 5). Second, SomaScan provide relative quantification instead of absolute quantification, but SomaScan measurements are able to detect very low protein levels (36,68). Third, in our study, there are no participants diagnosed before age of 50; therefore, we cannot compare associations in early-onset and late-onset CRC. Last, in our study, we have low effect allele frequencies for some pre-selected SNPs. This would limit the power for detecting association as well as effect modification.
In conclusion, rs2596542-T was associated with a lower risk of CRC in our cohort. While sMICA and sMICB levels were not associated with CRC risk overall, lower sMICA was statistically significantly associated with a decreased CRC risk in males. Further studies are warranted to validate our findings and examine these associations in larger or pooled prospective studies. These findings are important to elucidate the role of immunosurveillance in CRC development, and potentially could lead to novel interception strategies.
Supplementary Material
Acknowledgement
The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute; National Institutes of Health; Department of Health and Human Services, under Contract nos. (75N92022D00001, 75N92022D00002, 75N92022D00003, 75N92022D00004, 75N92022D00005). This work was also supported by grants R01HL087641 and R01HL086694; National Human Genome Research Institute contract U01HG004402; and National Institutes of Health contract HHSN268200625226C. Infrastructure was partly supported by Grant Number UL1RR025005, a component of the National Institutes of Health and NIH Roadmap for Medical Research. SomaLogic Inc. conducted the SomaScan assays in exchange for use of ARIC data. This work was supported in part by NIH/NHLBI grant R01 HL134320. Studies on cancer in ARIC are also supported by the National Cancer Institute (U01 CA164975, E.A. Platz). Cancer incidence data have been provided by the Maryland Cancer Registry, Center for Cancer Surveillance and Control, Department of Mental Health and Hygiene, 201 W. Preston Street, Room 400, Baltimore, MD 21201. We acknowledge the State of Maryland, the Maryland Cigarette Restitution Fund, and the National Program of Cancer Registries (NPCR) of the Centers for Disease Control and Prevention (CDC) for the funds that helped support the availability of the cancer registry data. The authors thank the staff and participants of the ARIC study for their important contributions. This study was also supported by R03CA249461 (A.E. Prizment) and the Masonic Cancer Center, University of Minnesota (via Mezin-Koats Colon Cancer Research Award and the Forster Family Chair funds, A.E. Prizment). The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Abbreviations used:
- ARIC
Atherosclerosis Risk in Communities
- CRC
Colorectal cancer
- MICA
Major histocompatibility complex class I chain-related protein A
- MICB
Major histocompatibility complex class I chain-related protein B
- sMICA
Soluble-MICA
- sMICB
Soluble-MICB
- NKG2D
Natural-killer group 2, member D
- NK
Natural killer
- SNP
Single nucleotide polymorphism
- BMI
Body mass index
- PC
Principal component
- eGFR
Estimated glomerular filtration rate
- CRP
C-reactive protein
- B2M
Beta-2 microglobulin
- SD
Standard deviation
- CI
Confidence interval
- HR
Hazard ratio
- RFU
Relative fluorescent unit
Footnotes
Disclaimer
Dr. Gomez contributed to this article as an employee of the University of Mississippi Medical Center. Currently, he is employed at the National Institutes of Health. The views expressed are his own and do not necessarily represent the views of the National Institutes of Health or the US Government.
Conflict of interest statement: The authors declare no potential conflicts of interest.
References
- 1.Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2022. CA Cancer J Clin 2022;72(1):7–33 doi 10.3322/caac.21708. [DOI] [PubMed] [Google Scholar]
- 2.Smyth MJ, Dunn GP, Schreiber RD. Cancer immunosurveillance and immunoediting: the roles of immunity in suppressing tumor development and shaping tumor immunogenicity. Adv Immunol 2006;90:1–50 doi 10.1016/S0065-2776(06)90001-7. [DOI] [PubMed] [Google Scholar]
- 3.Zitvogel L, Tesniere A, Kroemer G. Cancer despite immunosurveillance: immunoselection and immunosubversion. Nat Rev Immunol 2006;6(10):715–27 doi 10.1038/nri1936. [DOI] [PubMed] [Google Scholar]
- 4.Dunn GP, Old LJ, Schreiber RD. The immunobiology of cancer immunosurveillance and immunoediting. Immunity 2004;21(2):137–48 doi 10.1016/j.immuni.2004.07.017. [DOI] [PubMed] [Google Scholar]
- 5.Tamaki S, Kawakami M, Ishitani A, Kawashima W, Kasuda S, Yamanaka Y, et al. Soluble MICB serum levels correlate with disease stage and survival rate in patients with oral squamous cell carcinoma. Anticancer Res 2010;30(10):4097–101. [PubMed] [Google Scholar]
- 6.Groh V, Bahram S, Bauer S, Herman A, Beauchamp M, Spies T. Cell stress-regulated human major histocompatibility complex class I gene expressed in gastrointestinal epithelium. Proc Natl Acad Sci U S A 1996;93(22):12445–50 doi 10.1073/pnas.93.22.12445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Groh V, Steinle A, Bauer S, Spies T. Recognition of stress-induced MHC molecules by intestinal epithelial gammadelta T cells. Science 1998;279(5357):1737–40 doi 10.1126/science.279.5357.1737. [DOI] [PubMed] [Google Scholar]
- 8.Bauer S, Groh V, Wu J, Steinle A, Phillips JH, Lanier LL, et al. Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA. Science 1999;285(5428):727–9 doi 10.1126/science.285.5428.727. [DOI] [PubMed] [Google Scholar]
- 9.Zhao Y, Chen N, Yu Y, Zhou L, Niu C, Liu Y, et al. Prognostic value of MICA/B in cancers: a systematic review and meta-analysis. Oncotarget 2017;8(56):96384–95 doi 10.18632/oncotarget.21466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Fuertes MB, Domaica CI, Zwirner NW. Leveraging NKG2D Ligands in Immuno-Oncology. Front Immunol 2021;12:713158 doi 10.3389/fimmu.2021.713158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Feng Q, Yu S, Mao Y, Ji M, Wei Y, He G, et al. High MICB expression as a biomarker for good prognosis of colorectal cancer. J Cancer Res Clin Oncol 2020;146(6):1405–13 doi 10.1007/s00432-020-03159-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Watson NF, Spendlove I, Madjd Z, McGilvray R, Green AR, Ellis IO, et al. Expression of the stress-related MHC class I chain-related protein MICA is an indicator of good prognosis in colorectal cancer patients. Int J Cancer 2006;118(6):1445–52 doi 10.1002/ijc.21510. [DOI] [PubMed] [Google Scholar]
- 13.Espinoza I, Agarwal S, Sakiyama M, Shenoy V, Orr WS, Diffalha SA, et al. Expression of MHC class I polypeptide-related sequence A (MICA) in colorectal cancer. Front Biosci (Landmark Ed) 2021;26(10):765–76 doi 10.52586/4986. [DOI] [PubMed] [Google Scholar]
- 14.Chitadze G, Lettau M, Bhat J, Wesch D, Steinle A, Furst D, et al. Shedding of endogenous MHC class I-related chain molecules A and B from different human tumor entities: heterogeneous involvement of the “a disintegrin and metalloproteases” 10 and 17. Int J Cancer 2013;133(7):1557–66 doi 10.1002/ijc.28174. [DOI] [PubMed] [Google Scholar]
- 15.Holdenrieder S, Stieber P, Peterfi A, Nagel D, Steinle A, Salih HR. Soluble MICA in malignant diseases. Int J Cancer 2006;118(3):684–7 doi 10.1002/ijc.21382. [DOI] [PubMed] [Google Scholar]
- 16.Ouni N, Ben Chaaben A, Kablouti G, Lajnef M, Ayari F, Abaza H, et al. MICA-129Met/Val Polymorphism Is Associated with Early-Onset Breast Cancer Risk. Immunol Invest 2017;46(6):603–14 doi 10.1080/08820139.2017.1336175. [DOI] [PubMed] [Google Scholar]
- 17.Onyeaghala G, Lane J, Pankratz N, Nelson HH, Thyagarajan B, Walcheck B, et al. Association between MICA polymorphisms, s-MICA levels, and pancreatic cancer risk in a population-based case-control study. PLoS One 2019;14(6):e0217868 doi 10.1371/journal.pone.0217868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Toledo-Stuardo K, Ribeiro CH, Canals A, Morales M, Gárate V, Rodríguez-Siza J, et al. Major Histocompatibility Complex Class I-Related Chain A (MICA) Allelic Variants Associate With Susceptibility and Prognosis of Gastric Cancer. Front Immunol 2021;12:645528 doi 10.3389/fimmu.2021.645528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tong HV, Toan NL, Song LH, Bock CT, Kremsner PG, Velavan TP. Hepatitis B virus-induced hepatocellular carcinoma: functional roles of MICA variants. J Viral Hepat 2013;20(10):687–98 doi 10.1111/jvh.12089. [DOI] [PubMed] [Google Scholar]
- 20.Wang H, Cao H, Xu Z, Wang D, Zeng Y. SNP rs2596542G>A in MICA is associated with risk of hepatocellular carcinoma: a meta-analysis. Biosci Rep 2019;39(5) doi 10.1042/BSR20181400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Chen D, Juko-Pecirep I, Hammer J, Ivansson E, Enroth S, Gustavsson I, et al. Genome-wide association study of susceptibility loci for cervical cancer. J Natl Cancer Inst 2013;105(9):624–33 doi 10.1093/jnci/djt051. [DOI] [PubMed] [Google Scholar]
- 22.Chen D, Hammer J, Lindquist D, Idahl A, Gyllensten U. A variant upstream of HLA-DRB1 and multiple variants in MICA influence susceptibility to cervical cancer in a Swedish population. Cancer Med 2014;3(1):190–8 doi 10.1002/cam4.183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ji M, Wang J, Yuan L, Zhang Y, Zhang J, Dong W, et al. MICA polymorphisms and cancer risk: a meta-analysis. Int J Clin Exp Med 2015;8(1):818–26. [PMC free article] [PubMed] [Google Scholar]
- 24.Ivanova M, Al Hadra B, Yordanov S, Lesichkova S, Stoyanov H, Shivarov V, et al. Associations of high-resolution-typing-defined MICA and MICB polymorphisms, and the levels of soluble MICA and MICB with Oral Squamous Cell Carcinoma in Bulgarian patients. J Oral Pathol Med 2021;50(8):758–65 doi 10.1111/jop.13185. [DOI] [PubMed] [Google Scholar]
- 25.Isernhagen A, Malzahn D, Bickeböller H, Dressel R. Impact of the MICA-129Met/Val Dimorphism on NKG2D-Mediated Biological Functions and Disease Risks. Front Immunol 2016;7:588 doi 10.3389/fimmu.2016.00588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Yang X, Kuang S, Wang L, Wei Y. MHC class I chain-related A: Polymorphism, regulation and therapeutic value in cancer. Biomed Pharmacother 2018;103:111–7 doi 10.1016/j.biopha.2018.03.177. [DOI] [PubMed] [Google Scholar]
- 27.Mohamed AA, Elsaid OM, Amer EA, Elosaily HH, Sleem MI, Gerges SS, et al. Clinical significance of SNP (rs2596542) in histocompatibility complex class I-related gene A promoter region among hepatitis C virus related hepatocellular carcinoma cases. J Adv Res 2017;8(4):343–9 doi 10.1016/j.jare.2017.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kirsten H, Petit-Teixeira E, Scholz M, Hasenclever D, Hantmann H, Heider D, et al. Association of MICA with rheumatoid arthritis independent of known HLA-DRB1 risk alleles in a family-based and a case control study. Arthritis Res Ther 2009;11(3):R60 doi 10.1186/ar2683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Wen X, Chen S, Li J, Li Y, Li L, Wu Z, et al. Association between genetic variants in the human leukocyte antigen-B/MICA and Takayasu arteritis in Chinese Han population. Int J Rheum Dis 2018;21(1):271–7 doi 10.1111/1756-185X.13012. [DOI] [PubMed] [Google Scholar]
- 30.Liu Y, Helms C, Liao W, Zaba LC, Duan S, Gardner J, et al. A genome-wide association study of psoriasis and psoriatic arthritis identifies new disease loci. PLoS Genet 2008;4(3):e1000041 doi 10.1371/journal.pgen.1000041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Limou S, Le Clerc S, Coulonges C, Carpentier W, Dina C, Delaneau O, et al. Genomewide association study of an AIDS-nonprogression cohort emphasizes the role played by HLA genes (ANRS Genomewide Association Study 02). J Infect Dis 2009;199(3):419–26 doi 10.1086/596067. [DOI] [PubMed] [Google Scholar]
- 32.Wright JD, Folsom AR, Coresh J, Sharrett AR, Couper D, Wagenknecht LE, et al. The ARIC (Atherosclerosis Risk In Communities) Study: JACC Focus Seminar 3/8. J Am Coll Cardiol 2021;77(23):2939–59 doi 10.1016/j.jacc.2021.04.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Joshu CE, Barber JR, Coresh J, Couper DJ, Mosley TH, Vitolins MZ, et al. Enhancing the Infrastructure of the Atherosclerosis Risk in Communities (ARIC) Study for Cancer Epidemiology Research: ARIC Cancer. Cancer Epidemiol Biomarkers Prev 2018;27(3):295–305 doi 10.1158/1055-9965.EPI-17-0696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Tworoger SS, Hankinson SE. Collection, processing, and storage of biological samples in epidemiologic studies: sex hormones, carotenoids, inflammatory markers, and proteomics as examples. Cancer Epidemiol Biomarkers Prev 2006;15(9):1578–81 doi 10.1158/1055-9965.epi-06-0629. [DOI] [PubMed] [Google Scholar]
- 35.Rai AJ, Gelfand CA, Haywood BC, Warunek DJ, Yi J, Schuchard MD, et al. HUPO Plasma Proteome Project specimen collection and handling: towards the standardization of parameters for plasma proteome samples. Proteomics 2005;5(13):3262–77 doi 10.1002/pmic.200401245. [DOI] [PubMed] [Google Scholar]
- 36.Tin A, Yu B, Ma J, Masushita K, Daya N, Hoogeveen RC, et al. Reproducibility and Variability of Protein Analytes Measured Using a Multiplexed Modified Aptamer Assay. J Appl Lab Med 2019;4(1):30–9 doi 10.1373/jalm.2018.027086. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Gold L, Ayers D, Bertino J, Bock C, Bock A, Brody EN, et al. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS One 2010;5(12):e15004 doi 10.1371/journal.pone.0015004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Gold L, Walker JJ, Wilcox SK, Williams S. Advances in human proteomics at high scale with the SOMAscan proteomics platform. New biotechnology 2012;29(5):543–9 doi 10.1016/j.nbt.2011.11.016. [DOI] [PubMed] [Google Scholar]
- 39.Kim CH, Tworoger SS, Stampfer MJ, Dillon ST, Gu X, Sawyer SJ, et al. Stability and reproducibility of proteomic profiles measured with an aptamer-based platform. Sci Rep 2018;8(1):8382 doi 10.1038/s41598-018-26640-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Candia J, Cheung F, Kotliarov Y, Fantoni G, Sellers B, Griesman T, et al. Assessment of Variability in the SOMAscan Assay. Sci Rep 2017;7(1):14248 doi 10.1038/s41598-017-14755-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Candia J, Daya GN, Tanaka T, Ferrucci L, Walker KA. Assessment of variability in the plasma 7k SomaScan proteomics assay. Sci Rep 2022;12(1):17147 doi 10.1038/s41598-022-22116-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Rooney MR, Chen J, Ballantyne CM, Hoogeveen RC, Tang O, Grams ME, et al. Comparison of Proteomic Measurements Across Platforms in the Atherosclerosis Risk in Communities (ARIC) Study. Clin Chem 2023;69(1):68–79 doi 10.1093/clinchem/hvac186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Delaneau O, Marchini J, Zagury JF. A linear complexity phasing method for thousands of genomes. Nat Methods 2011;9(2):179–81 doi 10.1038/nmeth.1785. [DOI] [PubMed] [Google Scholar]
- 44.Ellinor PT, Lunetta KL, Albert CM, Glazer NL, Ritchie MD, Smith AV, et al. Meta-analysis identifies six new susceptibility loci for atrial fibrillation. Nat Genet 2012;44(6):670–5 doi 10.1038/ng.2261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81(3):559–75 doi 10.1086/519795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Howie BN, Donnelly P, Marchini J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet 2009;5(6):e1000529 doi 10.1371/journal.pgen.1000529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Kowalski MH, Qian H, Hou Z, Rosen JD, Tapia AL, Shan Y, et al. Use of >100,000 NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium whole genome sequences improves imputation quality and detection of rare variant associations in admixed African and Hispanic/Latino populations. PLoS Genet 2019;15(12):e1008500 doi 10.1371/journal.pgen.1008500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 2006;38(8):904–9 doi 10.1038/ng1847. [DOI] [PubMed] [Google Scholar]
- 49.The Atherosclerosis Risk in Communities (ARIC) Study: design and objectives. The ARIC investigators. Am J Epidemiol 1989;129(4):687–702. [PubMed] [Google Scholar]
- 50.Polter EJ, Onyeaghala G, Lutsey PL, Folsom AR, Joshu CE, Platz EA, et al. Prospective Association of Serum and Dietary Magnesium with Colorectal Cancer Incidence. Cancer Epidemiol Biomarkers Prev 2019;28(8):1292–9 doi 10.1158/1055-9965.EPI-18-1300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Inker LA, Eneanya ND, Coresh J, Tighiouart H, Wang D, Sang Y, et al. New Creatinine- and Cystatin C-Based Equations to Estimate GFR without Race. N Engl J Med 2021;385(19):1737–49 doi 10.1056/NEJMoa2102953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Walker KA, Windham BG, Power MC, Hoogeveen RC, Folsom AR, Ballantyne CM, et al. The association of mid-to late-life systemic inflammation with white matter structure in older adults: The Atherosclerosis Risk in Communities Study. Neurobiol Aging 2018;68:26–33 doi 10.1016/j.neurobiolaging.2018.03.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Prizment AE, Linabery AM, Lutsey PL, Selvin E, Nelson HH, Folsom AR, et al. Circulating Beta-2 Microglobulin and Risk of Cancer: The Atherosclerosis Risk in Communities Study (ARIC). Cancer Epidemiol Biomarkers Prev 2016;25(4):657–64 doi 10.1158/1055-9965.EPI-15-0849. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Rubin M. When to adjust alpha during multiple testing: a consideration of disjunction, conjunction, and individual testing. Synthese 2021;199(3):10969–1000 doi 10.1007/s11229-021-03276-4. [DOI] [Google Scholar]
- 55.Klein SL, Flanagan KL. Sex differences in immune responses. Nat Rev Immunol 2016;16(10):626–38 doi 10.1038/nri.2016.90. [DOI] [PubMed] [Google Scholar]
- 56.Márquez EJ, Chung CH, Marches R, Rossi RJ, Nehar-Belaid D, Eroglu A, et al. Sexual-dimorphism in human immune system aging. Nat Commun 2020;11(1):751 doi 10.1038/s41467-020-14396-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Augustus GJ, Ellis NA. Colorectal Cancer Disparity in African Americans: Risk Factors and Carcinogenic Mechanisms. Am J Pathol 2018;188(2):291–303 doi 10.1016/j.ajpath.2017.07.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.González N, Prieto I, Del Puerto-Nevado L, Portal-Nuñez S, Ardura JA, Corton M, et al. 2017 update on the relationship between diabetes and colorectal cancer: epidemiology, potential molecular mechanisms and therapeutic implications. Oncotarget 2017;8(11):18456–85 doi 10.18632/oncotarget.14472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Kucuk B, Yilmaz E, Cacan E. Expression profiles of Natural Killer Group 2D Ligands (NGK2DLs) in colorectal carcinoma and changes in response to chemotherapeutic agents. Mol Biol Rep 2021;48(5):3999–4008 doi 10.1007/s11033-021-06404-y. [DOI] [PubMed] [Google Scholar]
- 60.Chitadze G, Bhat J, Lettau M, Janssen O, Kabelitz D. Generation of soluble NKG2D ligands: proteolytic cleavage, exosome secretion and functional implications. Scand J Immunol 2013;78(2):120–9 doi 10.1111/sji.12072. [DOI] [PubMed] [Google Scholar]
- 61.Zhao YK, Jia CM, Yuan GJ, Liu W, Qiu Y, Zhu QG. Expression and clinical value of the soluble major histocompatibility complex class I-related chain A molecule in the serum of patients with renal tumors. Genet Mol Res 2015;14(2):7233–40 doi 10.4238/2015.June.29.16. [DOI] [PubMed] [Google Scholar]
- 62.Groh V, Wu J, Yee C, Spies T. Tumour-derived soluble MIC ligands impair expression of NKG2D and T-cell activation. Nature 2002;419(6908):734–8 doi 10.1038/nature01112. [DOI] [PubMed] [Google Scholar]
- 63.Kumar V, Yi Lo PH, Sawai H, Kato N, Takahashi A, Deng Z, et al. Soluble MICA and a MICA variation as possible prognostic biomarkers for HBV-induced hepatocellular carcinoma. PLoS One 2012;7(9):e44743 doi 10.1371/journal.pone.0044743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 64.Iwaszko M, Świerkot J, Dratwa M, Wysoczańska B, Korman L, Bugaj B, et al. Association of MICA-129Met/Val polymorphism with clinical outcome of anti-TNF therapy and MICA serum levels in patients with rheumatoid arthritis. Pharmacogenomics J 2020;20(6):760–9 doi 10.1038/s41397-020-0164-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Steinle A, Li P, Morris DL, Groh V, Lanier LL, Strong RK, et al. Interactions of human NKG2D with its ligands MICA, MICB, and homologs of the mouse RAE-1 protein family. Immunogenetics 2001;53(4):279–87 doi 10.1007/s002510100325. [DOI] [PubMed] [Google Scholar]
- 66.Zingoni A, Vulpis E, Cecere F, Amendola MG, Fuerst D, Saribekyan T, et al. MICA-129 Dimorphism and Soluble MICA Are Associated With the Progression of Multiple Myeloma. Front Immunol 2018;9:926 doi 10.3389/fimmu.2018.00926. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Isernhagen A, Schilling D, Monecke S, Shah P, Elsner L, Walter L, et al. The MICA-129Met/Val dimorphism affects plasma membrane expression and shedding of the NKG2D ligand MICA. Immunogenetics 2016;68(2):109–23 doi 10.1007/s00251-015-0884-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 68.Sathyan S, Ayers E, Gao T, Weiss EF, Milman S, Verghese J, et al. Plasma proteomic profile of age, health span, and all-cause mortality in older adults. Aging Cell 2020;19(11):e13250 doi 10.1111/acel.13250. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data used in this study are available from the corresponding author upon reasonable request (https://sites.cscc.unc.edu/aric/pubs-policies-and-forms-pg).