Abstract
Background: Multiplex assays are available to measure an array of circulating chemokines, soluble cytokine receptors and growth factors. However, there is limited information regarding whether these analytes are suitable for large-scale epidemiological studies to assess their relationships with chronic diseases, including cancer.
Methods: We examined detectability, assay repeatability, and 3-year within-subject reproducibility of plasma levels of 25 chemokines and 11 soluble receptors of cytokines and growth factors selected from the Human Millipore Panels. Plasma samples were obtained from 36 men (average age 62 years) and 17 women (average age 32 years) who participated in two epidemiological studies. Inter-assay and within-subject reproducibility were assessed by intraclass correlation coefficients (ICC).
Results: All analytes, except lymphotactin (47% detectability), were detectable in >90% of plasma samples. Inter-assay reproducibility for all analytes in 36 men tested three times on separate days were good to excellent (ICCs: 0.71–1.00). Within-subject reproducibility in 17 women sampled three times in three years were excellent (ICC ≥ 0.75) for five chemokines (eotaxin, fractalkine, 6Ckine, eotaxin 3, and SDF-1α+β) and three soluble receptors (sIL-1R2, sIL-4R and sVEGFR2); ICCs were fair to good (0.4 ≤ ICC < 0.75) for 15 chemokines and eight soluble receptors. However, five chemokines (GRO, IP-10, MIP-1β, BCA-1, and MIP-3α) had ICC < 0.4, suggesting biological variability.
Conclusion: Multiplex assays for plasma levels of selected chemokines and soluble receptors showed good to excellent assay detectability and repeatability. Most analytes also had good 3-year within-subject reproducibility, indicating that a single measurement of these analytes may be used to assess biomarker-disease associations.
Keywords: Chemokines, Soluble receptors, Within-subject variability, Biomarker, Limit of detection
1. Introduction
Increasing evidence suggests that chronic inflammation is associated with risk of several cancers [1–4]. Chronic inflammation may be involved in cancer initiation, promotion, and progression via several mechanisms. Activated inflammatory cells produce reactive oxygen and nitrogen species, which create a mutagenic environment enabling genome instability that leads to tumor initiation [5]. Growth factors and cytokines produced by the inflammatory cells further induce proliferation and suppress apoptosis of cancer cells and hence promote growth of primary tumors [4,6,7]. Inflammatory cells also produce various pro-inflammatory chemokines, a group of chemo-attractant cytokines which, in addition to recruiting leukocytes to infiltrate the local inflammatory site, may act as angiogenic factors or growth factors that further stimulate tumor progression [8–12].
However, to maintain tissue homeostasis, there are anti-inflammatory and anti-tumor immune mediators. For example, some chemokines are anti-inflammatory and induce infiltration of anti-tumor effectors (e.g., Th1 lymphocytes and natural killer cells) and/or have angiostatic functions [10,13]; soluble receptors of proinflammatory cytokines and growth factors could act as decoy receptors, block binding of their ligands to membrane receptors, and hence inhibit ligand signaling [14–16].
Despite the importance of host immunity in carcinogenesis, most epidemiologic studies have focused on cancer association with a few inflammation markers, mostly IL-6 and TNF-α [17–19]. There is a need to examine other pro- and anti-tumorigenic immune response markers in order to better understand their roles in cancer etiology. Such epidemiological studies can be facilitated by multiplex assays measuring an array of immune response mediators, such as cytokines, chemokines, soluble receptors of cytokines and growth factors, in plasma or serum. Before embarking on large prospective epidemiologic studies, it is important to examine assay performance of these analytes in terms of detectability and inter-assay reproducibility. It is also crucial to assess the within-person variability of immune response markers over time to determine whether the circulating levels of a marker are relatively stable, hence a single measurement of the marker can reflect an individual's level in the long run.
Several studies have evaluated the performance and/or within-subject reproducibility of circulating levels of cytokines, chemokines and soluble receptors in healthy individuals [20–28]. However, the majority of these studies have examined a small number of makers (e.g., IL-1Ra, IL-2, IL-5, IL-6 and TNF-α). Two studies assessed the within-subject variability of more than 20 immune response markers over a period of 2–3 years, but not the assay performance [24,28]. A large methodological study recently evaluated the assay performance of 116 inflammation, immune, and metabolic markers among 100 cancer-free participants in the Prostate, Lung, Colorectal and Ovarian Cancer (PLCO) Trial across different platforms and different samples: serum, heparin or EDTA [25]. Although this study showed that these biomarkers had good performance in plasma samples, they did not evaluate the within-subject reproducibility of these markers over time.
The goal of our study was to examine both assay performance (detectability and inter-assay reproducibility) and within-subject reproducibility over a 3-year period of plasma levels of 25 chemokines and 11 soluble receptors of cytokines and growth factors selected from the multiplex panels of Millipore (Billerica, MA). We focused on chemokines and soluble receptors rather than the traditional cytokines (e.g., interleukins and interferon), because the majority of cytokines have very low circulating levels in healthy individuals [25]. We report here that the chemokines and soluble receptors selected for this study have excellent detectability and inter-assay reproducibility, and the majority of them have good to excellent within-subject reproducibility over a 3-year period.
2. Materials and methods
2.1. Selection of chemokines and soluble receptors
The Millipore's Human Cytokine/Chemokine Panels I, II, and III (Billerica, MA) consist of a total of 76 analytes in three multiplex kits, and 46 of them are chemokines. The Milliplex Human Soluble Cytokine Receptor panel assays for 14 soluble receptors of cytokines and growth factors. We selected 25 chemokines and 11 soluble receptors from these four panels to be examined for this study, because detectable levels of these markers were reported in >75% of plasma samples from healthy individuals by either Millipore (Millipore personal communication) or by the methodological study of Chaturvedi et al. [25]. Our study objectives were to confirm detectability and assess inter-assay reproducibility and 3-year within-subject reproducibility of these immune response biomarkers.
2.2. Study design and laboratory assay
We conducted two separate studies to evaluate assay performance and 3-year within-subject reproducibility.
2.2.1. Study 1 for assay performance
EDTA plasma samples were obtained from a random sample of 36 men with no previous history of any cancer who were recruited as controls from general practitioners' offices and provided a single non-fasting blood sample as part of a prostate cancer case-control study conducted in the Bronx, New York from 1998 through 2000 [29]. Blood was collected in purple-top EDTA tubes and processed within an hour of blood draw at biorepository of the Albert Einstein College of Medicine. The EDTA plasma samples were stored in 2 mL aliquots at −75 °C. The plasma samples were assayed for the chemokines and soluble receptors three times, each on a separate day in the same laboratory by the same technician. Detectability of an immune response marker was defined as the proportion of samples with values above the assay limit of detection (LOD). Inter-assay intraclass correlation coefficient (ICC) was calculated for each marker to evaluate inter-assay reproducibility when the marker was measured repeatedly in the same blood sample. Mean age of the 36 men was 61.8 years (range: 46–79), and their racial/ethnic distribution was 55.6% African American, 19.4% Hispanic, and 25% others.
Of the 11 soluble receptors selected, the inter-assay reproducibility for sgp130, sIL-6R, sIL-1R2, and sTNFR1, as measured by coefficient of variation (CV), was previously evaluated, and therefore was not assessed in Study 1. For these soluble receptors, inter-assay CVs were obtained as follows: aliquots from the same four control samples were inserted in the assay plates testing 1478 study samples of a colorectal cancer study; 42 assay plates were run over several months in the same laboratory by the same technician. For each control sample, the four soluble receptors were measured 42 times, and the inter-assay CV for a soluble receptor was calculated as standard deviation divided by the mean of 42 measurements. The average inter-assay CVs over the four control samples were 6.6% for sgp130, 5.5% for sIL-6R, 7.1% for sIL-1R2, and 5.9% for sTNFR1 (unpublished data).
2.2.2. Study 2 for within-subject reproducibility
Three fasting blood samples were collected at baseline in 2002 and at the year-1 and year-3 visits from 17 women, who participated in the Bronx/Manhattan site (New York) of the multicenter Women's Interagency HIV study (WIHS) [30,31]. Within 6 h after collection, blood tubes were centrifuged at room temperature and citrate plasma aliquots (1 mL each) were stored at −70 °C. For this study, the samples were assayed for levels of the 25 chemokines and 11 soluble receptors. All three samples of an individual were assayed together on the same assay plate, avoiding the issue of calibration over time. For each marker, 3-year ICC was calculated to assess the within-subject reproducibility of marker level when the individual was sampled repeatedly over time. The 17 women were HIV-negative at baseline and follow-up visits, and did not have a history of hepatitis B or C infection [31]. They were on average 32.3 years of age at baseline (range: 19–47 years), and 65% were African-American.
2.2.3. Laboratory assays
Milliplex human cytokine/chemokine Panels I, II and III were used to measure plasma levels of the 25 chemokines, and the Milliplex human soluble cytokine receptor panel was used for the 11 soluble receptors (Millipore, Billerica, MA) according to manufacturer's instructions. These multiplex assays use the nonmagnetic polystyrene bead-based Luminex xMAP technology. Since Chaturvedi and colleagues [25] demonstrated that assays performance for the Milliplex human cytokine/chemokine panels was better in plasma in comparison to serum samples, we decided to run the assays for these two studies on plasma samples. All laboratory measurements were done at the Laboratory for Clinical Biochemistry Research at the University of Vermont according to the manufacturer's protocols. Study samples were tested in duplicate and the duplicate measurements were averaged for statistical analysis. The laboratory staff that conducted the experimental assays was blinded to the samples IDs and replicates. Studies were approved by the Institutional Review Board (IRB) of the Albert Einstein College of Medicine and written informed consents were obtained from participants in each of the two studies [29–31].
2.2.4. Statistical analysis
The distributions of plasma levels of the 25 chemokines were skewed, hence values were log-transformed, whereas the distributions of soluble receptors were normal. To be consistent in presenting the data, we report the medians and inter-quartile ranges (IQRs) for all the analytes in Tables 1 and 2. For samples with an undetectable level below the assay limit of detection (LOD), their values were imputed using ½×LOD of that analyte [32,33]. One-way random effects ANOVA models were used to calculate the inter-assay ICCs for assay reproducibility in Study 1, as well as the 3-year ICCs for within-subject reproducibility in Study 2. In this model, the total variance of a marker value (σT2) was separated into three components: (i) between-subject variance (σG2), (ii) intra-individual variance (σI2), and (iii) analytic residual variance (σA2). The proportion of total variance attributable to between-subject, intra-individual and analytic variation was annotated as RG, RI, and RA, respectively (e.g., RG = σG2 / σT2). Analytic variation was based on within-run variance from samples assayed in duplicate, hence RA (σA2 / σT2) tended to be negligible. We eliminated σA2 by using the average of the duplicate measurements. As such σT2 = σG2 + σI2. The ICC for each chemokine and soluble receptor was estimated by σG2 / σT2 [34]. We used mixed effects models to examine if analytes were also associated with women's characteristics including age at each visit, race (black vs. other), BMI, smoking status at baseline (current smoker vs. non-smoker) and white blood cell (WBC) counts (log-transformed value) at each blood draw in study 2. Since circulating levels of majority of chemokines and soluble receptors were not associated with women's characteristics or WBC counts, we present the unadjusted ICCs throughout the manuscript.
Table 1.
Inter-assay ICCs of measuring chemokines and soluble receptors in 36 samples tested three times on separate days.
| Minimum detectability (%) in 3 testsa | Median (IQRb) Day 1 (pg/mL) | Median (IQRb) Day 2 (pg/mL) | Median (IQRb) Day 3 (pg/mL) | Inter-assay ICC (95% CI) | |
|---|---|---|---|---|---|
| Milliplex human cytokine/chemokine Panel | |||||
| Panel Ic | |||||
| Eotaxin/CCL11 | 100 | 65.4 (52.1–84.5) | 66.8 (55.9–95.7) | 72.1 (58.5–103) | 0.96 (0.94–0.98) |
| Fractalkine/CX3CL1 | 94 | 88.6 (47.6–181) | 94.2 (58.0–207) | 119 (80.4–252) | 0.78 (0.65–0.87) |
| GRO/CXCL1 | 100 | 489 (337–722) | 551 (364–765) | 537 (361–791) | 0.98 (0.97–0.99) |
| IL-8/CXCL8 | 100 | 5.4 (2.9–9.9) | 4.9 (3.1–10.6) | 5.1 (2.9–11.4) | 0.95 (0.92–0.97) |
| IP-10/CXCL10 | 100 | 278 (212–355) | 320 (250–393) | 340 (248–419) | 0.94 (0.90–0.97) |
| MCP-1/CCL2 | 100 | 233 (205–336) | 253 (217–336) | 262 (224–340) | 0.96 (0.92–0.98) |
| MDC/CCL22 | 100 | 1027 (803–1250) | 1083 (853–1282) | 1190 (889–1467) | 0.87 (0.79–0.93) |
| MIP-1β/CCL4 | 100 | 27.2 (19.9–41.1) | 26.2 (18.6–32.8) | 31.3 (25.4–44.8) | 0.86 (0.78–0.92) |
| Panel IIc | |||||
| BCA-1/CXCL13 | 100 | 30.2 (22.0–38.5) | 28.6 (19.7–34.9) | 31.2 (21.6–39.3) | 0.95 (0.92–0.97) |
| CTACK/CCL27 | 100 | 620 (449–765) | 638 (499–810) | 568 (409–720) | 0.93 (0.88–0.96) |
| 6Ckin/CCL21 | 97 | 499 (290–593) | 481 (302–611) | 489 (325–622) | 0.98 (0.97–0.99) |
| Eotaxin 2/CCL24 | 100 | 376 (163–663) | 323 (167–608) | 363 (168–683) | 0.99 (0.98–0.99) |
| Eotaxin 3/CCL26 | 94 | 23.7 (8.6–31.3) | 16.2 (6.3–33.6) | 18.1 (7.6–31.1) | 0.86 (0.78–0.92) |
| ENA-78/CXCL5 | 100 | 440 (208–688) | 418 (194–678) | 454 (205–710) | 1.00 (0.99–1.00) |
| I-309/CCL1 | 97 | 2.4 (1.8–4.3) | 2.2 (1.6–4.6) | 2.3 (1.8–4.4) | 0.93 (0.88–0.96) |
| MCP-2/CCL8 | 100 | 17.2 (14.4–24.1) | 16.5 (13.9–22.3) | 14.4 (11.8–21.0) | 0.88 (0.81–0.93) |
| MIP-1d/CCL15 | 100 | 1901 (1310–2801) | 1721 (1304–2325) | 1967 (1443–2699) | 0.98 (0.97–0.99) |
| SDF-1α+β/CXCL12 | 100 | 2278 (1705–2874) | 2468 (1962–3093) | 2081 (1758–2527) | 0.96 (0.94–0.98) |
| TARC/CCL17 | 100 | 32.2 (18.3–41.8) | 33.2 (17.3–39.8) | 32.6 (18.5–40.7) | 0.98 (0.97–0.99) |
| Panel IIIc | |||||
| GCP-2/CXCL6 | 100 | 44.4 (32.3–51.7) | 44.0 (32.3–51.6) | 51.1 (39.2–58.8) | 0.94 (0.90–0.97) |
| I-TAC/CXCL11 | 100 | 34.0 (18.8–48.2) | 30.8 (19.4–46.0) | 37.0 (21.3–55.3) | 0.99 (0.98–0.99) |
| Lymphotactin/XCL1 | 47 | 13.7 (4.5–46.3) | 5.2 (5.1–34.3) | 17.8 (8.9–37.2) | 0.71 (0.56–0.83) |
| MIG/CXCL9 | 100 | 734 (397–1086) | 798 (402–1057) | 830 (468–1277) | 0.98 (0.97–0.99) |
| MIP-3α/CCL20 | 100 | 11.0 (7.0–14.5) | 12.1 (7.6–14.8) | 11.0 (6.8–16.0) | 0.98 (0.97–0.99) |
| MIP-3β/CCL19 | 100 | 70.5 (53.8–114) | 77.3 (59.6–108) | 87.7 (65.0–132) | 0.91(0.86–0.95) |
| Milliplex human soluble cytokine receptor Paneld | |||||
| sCD30 | 100 | 82.5 (39.3–150) | 68.2 (45.5–124) | 84.3 (37.8–180) | 0.83 (0.73–0.90) |
| sEGFR | 100 | 81,810 (72,797– 88,860) | 81,840 (71598–88,901) | 80,378 (73,298–90,038) | 0.95 (0.91–0.97) |
| sIL-2Rα | 100 | 675 (509–937) | 720 (525–980) | 659 (513–927) | 0.98 (0.97–0.99) |
| sIL-4R | 100 | 954 (822–1203) | 1140 (968–1405) | 1041 (89 –1257) | 0.97 (0.94–0.98) |
| sVEGFR1 | 100 | 1989 (1047–3243) | 1758 (867–2901) | 1463 (743–2181) | 0.93 (0.88–0.96) |
| sVEGFR2 | 100 | 16,881 (15,488–18,108) | 17,107 (15,193–18,774) | 16,450 (14,941–18,243) | 0.93 (0.89–0.96) |
| sVEGFR3 | 100 | 4328 (2674–8065) | 4016 (3002–5910) | 3598 (2216–7501) | 0.74 (0.60–0.85) |
Detectability is defined as the percent of 36 samples that had analyte levels above the assay limit of detection (LOD); the minimum detectability of the three tests in three separate days is reported here.
IQR— inter-quartile range.
For chemokines, the most commonly used name is listed first, followed by the name based on the new nomenclature.
The inter-assay reproducibility, as measured by CVs, for sgp130, sIL-1R2, sIL-6R, and sTNFR1 had previously been evaluated and not assessed in Study 1 (see Section 2).
Table 2.
Three-year ICCs of chemokines and soluble receptors among 17 women sampled at baseline, year 1 and year 3.
| Milliplex human cytokine/chemokine Panel | Baseline | Year 1 | Year 3 | 3-Year Overall | |
|---|---|---|---|---|---|
| Median (IQRa) pg/mL | Median (IQRa) pg/mL | Median (IQRa) pg/mL | ICC | 95% CI | |
| Panel Ib | |||||
| Eotaxin/CCL11 | 35.7 (23.3–60.9) | 37.1 (24.6–52.3) | 43.1 (25.1–89.1) | 0.76 | 0.56–0.90 |
| Fractalkine/CX3CL1 | 175 (119–201) | 190 (132–240) | 191 (132–253) | 0.77 | 0.57–0.90 |
| GRO/CXCL1 | 407 (345–521) | 388 (302–538) | 395 (330–559) | 0.21 | −0.07 to 0.54 |
| IL-8/CXCL8 | 396 (51–944) | 314 (79.7–1427) | 362 (40.7–865) | 0.44 | 0.14–0.71 |
| IP-10/CXCL10 | 74.2 (52.2–119) | 87.1 (72.0–124) | 91.2 (80.9–123) | 0.03 | −0.20 to 0.37 |
| MCP-1/CCL2 | 98.1 (57.4–162) | 110 (73.3–240) | 97.8 (72.6–143) | 0.40 | 0.11–0.69 |
| MDC/CCL22 | 607 (438–703) | 552 (414–853) | 619 (426–744) | 0.58 | 0.30–0.80 |
| MIP-1β/CCL4 | 15.2 (11.1–25.6) | 15.8 (12.7–19.8) | 20.4 (13.4–25.1) | 0.25 | −0.03 to 0.58 |
| Panel IIb | |||||
| BCA-1/CXCL13 | 20.7 (16.3–66.0) | 22.0 (15.4–27.7) | 23.2 (17.4–25.0) | 0.32 | 0.02–0.63 |
| CTACK/CCL27 | 137 (109–183) | 167 (133–208) | 161 (126–190) | 0.73 | 0.50–0.88 |
| 6Ckine/CCL21 | 469 (375–615) | 511 (400–683) | 463 (426–605) | 0.86 | 0.72–0.94 |
| Eotaxin 2/CCL24 | 93.8 (57.6–156) | 112 (81.9–220) | 138 (74.8–200) | 0.58 | 0.31–0.80 |
| Eotaxin 3/CCL26 | 35.1 (27.9–46.5) | 35.7 (24.6–45.8) | 32.7 (24.6–35.9) | 0.84 | 0.69–0.93 |
| ENA 78/CXCL5 | 458 (307–555) | 650 (282–947) | 654 (403–704) | 0.71 | 0.48 – 0.87 |
| I-309/CCL1 | 5.0 (4.1–5.9) | 5.7 (4.2–6.4) | 4.8 (3.9–5.5) | 0.67 | 0.42–0.85 |
| MCP-2/CCL8 | 18.6 (17.1–21.1) | 19.0 (16.5–22.2) | 18.2 (14.8–20.3) | 0.65 | 0.40–0.84 |
| MIP-1d/CCL15 | 829 (660–1152) | 1370 (760–1774) | 886 (825–1311) | 0.66 | 0.42–0.85 |
| SDF-1α+β/CXCL12 | 1186 (1073–1483) | 1265 (1102–1529) | 1414 (1186–1464) | 0.82 | 0.65–0.92 |
| TARC/CCL17 | 17.5 (16.1–26.0) | 21.5 (16.9–35.9) | 27.4 (18.1–32.6) | 0.52 | 0.24–0.77 |
| Panel IIIb | |||||
| GCP-2/CXCL6 | 22.9 (17.3 – 30.0) | 28.5 (20.7 – 36.3) | 30.7 (24.7 – 35.8) | 0.63 | 0.36 – 0.83 |
| I-TAC/CXCL11 | 27.8 (19.0–34.5) | 25.0 (13.6–41.4) | 36.3 (24.0– 43.3) | 0.52 | 0.24– 0.77 |
| Lymphotactin/XCL1 | 23.6 (7.4–30.9) | 23.6 (23.5–43.0) | 34.5 (23.6–43.0) | 0.58 | 0.30–0.80 |
| MIG/CXCL9 | 219 (138–278) | 194 (149–287) | 193 (162–326) | 0.43 | 0.13–0.71 |
| MIP-3α/CCL20 | 6.1 (4.6–9.4) | 10.5 (5.2–17.0) | 10.5 (7.5–14.3) | 0.05 | −0.19 to 0.39 |
| MIP-3β/CCL19 | 35.0 (26.0–47.9) | 40.5 (33.8–51.9) | 39.8 (32.7–52.1) | 0.46 | 0.17–0.73 |
| Milliplex human soluble cytokine receptor Panel | Median (IQRb) pg/mL | Median (IQRb) pg/mL | Median (IQRb) pg/mL | ICC | 95% CI |
| sCD30 | 230 (176–352) | 224 (144–288) | 276 (112–336) | 0.50 | 0.21–0.75 |
| sEGFR | 52,756 (48,014–58,685) | 54,594 (47,731–60,855) | 50,498 (45,873–54,982) | 0.42 | 0.12–0.70 |
| sgp130 | 99,703 (86,758–106,774) | 102,859 (84,526–123,977) | 100,214 (79,286–116,218) | 0.63 | 0.37–0.83 |
| sIL-1R2 | 5664 (4279–7609) | 5412 (4552–6904) | 4875 (4016–7482) | 0.78 | 0.58–0.90 |
| sIL-2Rα | 1091 (933–1212) | 1249 (832–1360) | 999 (759–1301) | 0.56 | 0.28–0.79 |
| sIL-4R | 1268 (1135–1423) | 1212 (1128–1524) | 1257 (1135–1372) | 0.86 | 0.72–0.94 |
| sIL-6R | 12,179 (9551–14,199) | 12,451 (9310–13,951) | 10,861 (9398–13,905) | 0.52 | 0.24–0.77 |
| sTNFR1 | 1072 (896–1217) | 1156 (957– 1382) | 1039 (951–1205) | 0.65 | 0.40–0.84 |
| sVEGFR1 | 8835 (7036–10,997) | 9126 (6561–10,997) | 9391 (4552–13,163) | 0.58 | 0.30–0.80 |
| sVEGFR2 | 10,300 (8664–12,430) | 10,909 (9596–12,192) | 10,065 (8897–11,425) | 0.79 | 0.59–0.91 |
| sVEGFR3 | 8717 (6180–13,542) | 8445 (5484–11,934) | 9579 (4611–12,738) | 0.71 | 0.48–0.87 |
IQR—inter-quartile range.
For chemokines, the most commonly used name is listed first, followed by the name based on the new nomenclature.
3. Results
Table 1 provides the results of Study 1, assessing detectability and inter-assay ICCs of the chemokines and soluble receptors assayed three times in separate days at the same laboratory. As observed, these analytes were detectable in all plasma samples with the exception of lymphotactin (XCL1), which was detectable in no more than 47% of samples from three separate tests. All analytes had good to excellent inter-assay ICCs ranging from 0.71 to 1.00.
The medians (IQR) of the chemokines and soluble receptors in 17 women sampled three times over a 3-year period (Study 2) as well as 3-year ICCs and 95% CIs for within-subject reproducibility are shown in Table 2. Lymphotactin also had a low detectability in Study 2, with 69% of the samples had levels above LOD. All the other analytes were detectable. As seen in Table 2, the majority of these analytes had good to excellent within-subject reproducibility (i.e. low variability). Five chemokines (eotaxin, eotaxin 3, fractalkine, 6Ckine, and SDF-1α+β) and three soluble receptors (sIL1-R2, sIL-4R and sVEGFR2) showed excellent within-subject reproducibility (ICC ≥ 0.75), 15 chemokines and eight soluble receptors had fair to good within-subject reproducibility (0.4 ≤ ICC < 0.75), whereas five chemokines (GRO, IP-10, MIP-1β, BCA-1, and MIP-3α) had poor reproducibility (ICC < 0.4).
4. Discussion
There has been a lot of interest in evaluating the role of circulating levels of cytokines, chemokines and soluble receptors in the etiology of chronic diseases, including cancer. However, there is paucity of information regarding the assay performance for these analytes and whether their levels are stable over time. Moreover, there are no established clinical reference values for these biomarkers that can be used in etiologic studies. For prospective epidemiologic studies, it is important to first demonstrate that these biomarkers (a) can be detectable in healthy individuals, (b) have excellent laboratory reproducibility, and (c) have reasonable reproducibility of measurements over a long period of time, which can indicate that a single measurement of these markers in plasma/serum can be used to assess their associations with chronic diseases..
In this study, we evaluated detectability, inter-assay reproducibility and three-year within-subject reproducibility of chemokines and soluble receptors selected from the Millipore panels. In Study 1, we confirmed the detectability of the selected chemokines (except lymphotactin) and soluble receptors. The high detectability of these analytes was expected, however, as these chemokines and soluble receptors were selected based on their detectability of >75% in plasma samples from healthy individuals in other laboratories (Millipore personal communication and the study of Chaturvedi et al. [25]. Four soluble receptors (sgp130, sIL-6R, sIL-1R2, and sTNFR1) were not evaluated in Study 1, because they had previously been examined in a colorectal cancer study conducted by our group, and all of them were detected in >99% of 1478 plasma samples (unpublished data). We used the detectability cutoff point of >75% to determine that an analyte is suitable to be examined as an exposure/predictor variable in epidemiological studies, because individuals with an undetectable level can be categorized in the lowest quartile group in data analyses. The multiplex assays for these chemokines and soluble receptors had shown good inter-assay reproducibility, with inter-assay ICCs >0.70 for all the 25 chemokines and 7 soluble receptors. For the four soluble receptors not evaluated in Study 1 (sgp130, sIL-6R, sIL-1R2, and sTNFR1), their inter-assay CVs obtained from our previous study ranged from 5.5% to 7.1%.
Although we found good laboratory assay performance for 24 chemokines (excluding lymphotactin) and all 11 soluble receptors considered in this study, not all of them are suitable for epidemiological studies based on their within-subject reproducibility. The 3-year ICCs were ≥ 0.4 for all soluble receptors and for 80% (20 out of 25) of the chemokines. However, five chemokines (GRO, IP-10, MIP-1β, BCA-1, and MIP-3α) had an ICC < 0.4, suggesting their circulating levels tend to vary from time to time; hence multiple serial samples over time may be necessary to reflect an individual's exposure status to these chemokines. The ICC cutoff point of ≥0.4, which we deemed as acceptable in this study, is comparable to the ICCs of some common epidemiological biomarkers, such as serum triglycerides (ICC = 0.5) [35] and estradiol (ICC = 0.45) [36].
There are a few studies that have assessed within-subject reproducibility of chemokines and soluble receptors [22,24,28]. Two studies [22,24] that utilized blood samples of 30 premenopausal and 35 postmenopausal women of the New York University Women's Health Study collected over 3 years, evaluated within-subject reproducibility of cytokines/chemokines, including fractalkine, MCP-1, MCP-2, MIP-1β, eotaxin, GRO-α, IP-10, sEGFR, sIL-6R and IL-8 that were assessed in our study. They reported good to excellent within-subject reproducibility (ICC > 0.60) for these chemokines with the exception of IL-8, which had a 3-year ICC of 0.02. [22,24]. However, in our study, we found overall fair to excellent within-subject reproducibility (ICC range: 0.40–0.77) for IL-8, eotaxin, fractalkine, MCP-1, and MCP-2, but poor reproducibility for, MIP-1β, GRO-α and IP-10 (ICC range: 0.03–0.25). In another study, Clendenen and colleagues [28] evaluated 1–3 years within-subject reproducibility (average two years apart) of 22 cytokines and soluble receptors among 18 female participants aged 42–62 years from the Northern Sweden Health and Disease study cohort. With regard to four analytes, which we also evaluated in our study, they reported a very good reproducibility for IL-8 (ICC = 0.86), sIL-2R (ICC = 0.86), and sIL-6R (ICC = 0.69), but low reproducibility for sTNFR1 (ICC = 0.31) [28]. In our study, all four analytes had fair to good reproducibility (ICC range 0.44–0.65). Some of the discrepancies observed between our results and those obtained from previous studies [22,24,28] could be due to differences in study population, women's demographic and lifestyle characteristics, length of follow-up period, blood collection in relation to timing of menstrual cycle or menopausal status as well as assay methods in each study. However, a previous study [24] showed no influence of the menstrual cycle or menopausal status of the women on the ICCs of cytokines or chemokines. Thus, it is unlikely that differences in our results could be due to variation in hormone levels.
Some studies have suggested that circulating levels of inflammatory markers could vary by subjects' demographic and lifestyle characteristics including age, gender, race, obesity, smoking, and physical exercise [37–39]. We examined whether circulating levels of chemokines and soluble receptors were associated with women's age, race, BMI, smoking status and white blood cell count (a surrogate marker of inflammation) in our Study 2 dataset. We did not see any major difference in within-subject 3-year ICC when we adjusted the model for these variables as fixed effects in mixed models. Thus we have presented the unadjusted ICCs values. Nevertheless, there were only 17 women in Study 2 and the lack of association could be due to small sample size.
In conclusion, of the 25 chemokines and 11 soluble receptors selected to be assessed in this study, 19 of the chemokines and all soluble receptors are suitable to be evaluated as exposure/predictor variables in epidemiological studies based on assay performance (detectability >75% and inter-assay reproducibility >0.70) and 3-year within-subject reproducibility (ICC ≥ 0.4). Lymphotactin has low detectability. Five chemokines (GRO, IP-10, MIP-1β, BCA-1, and MIP-3α) need to be assessed with caution in epidemiological studies due to their low 3-year within-subject reproducibility.
Acknowledgement
Some of the data that were used in the manuscript were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group at New York City/Bronx Consortium Site (Kathryn Anastos—Principal Investigator). Ilir Agalliu was supported in part by grant "MRSG-11–112–01-CNE" from the American Cancer Society.
Footnotes
This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
Reference
- 1.Balkwill F, Mantovani A. Inflammation and cancer: Back to Virchow? Lancet. 2001;357:539–545. doi: 10.1016/S0140-6736(00)04046-0. [DOI] [PubMed] [Google Scholar]
- 2.Coussens LM, Werb Z. Inflammation and cancer. Nature. 2002;420:860–867. doi: 10.1038/nature01322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.De Marzo AM, Platz EA, Sutcliffe S, Xu J, Gronberg H, Drake CG. Inflammation in prostate carcinogenesis. Nature Reviews Cancer. 2007;7:256–269. doi: 10.1038/nrc2090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Grivennikov SI, Greten FR, Karin M. Immunity, inflammation, and cancer. Cell. 2010;140:883–899. doi: 10.1016/j.cell.2010.01.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Mantovani A, Sica A. Macrophages, innate immunity and cancer: balance, tolerance, and diversity. Current Opinion in Immunology. 2010;22:231–237. doi: 10.1016/j.coi.2010.01.009. [DOI] [PubMed] [Google Scholar]
- 6.de Visser KE, Coussens LM. The inflammatory tumor microenvironment and its impact on cancer development. Contributions to Microbiology. 2006;13:118–137. doi: 10.1159/000092969. [DOI] [PubMed] [Google Scholar]
- 7.Gerber PA, Hippe A, Buhren BA, Muller A, Homey B. Chemokines in tumor-associated angiogenesis. Biological Chemistry. 2009;390:1213–1223. doi: 10.1515/BC.2009.144. [DOI] [PubMed] [Google Scholar]
- 8.Ben-Baruch A. Inflammation-associated immune suppression in cancer: the roles played by cytokines, chemokines and additional mediators. Seminars in Cancer Biology. 2006;16:38–52. doi: 10.1016/j.semcancer.2005.07.006. [DOI] [PubMed] [Google Scholar]
- 9.de Visser KE, Eichten A, Coussens LM. Paradoxical roles of the immune system during cancer development. Nature Reviews Cancer. 2006;6:24–37. doi: 10.1038/nrc1782. [DOI] [PubMed] [Google Scholar]
- 10.Rosenkilde MM, Schwartz TW. The chemokine system—a major regulator of angiogenesis in health and disease. APMIS. 2004;112:481–495. doi: 10.1111/j.1600-0463.2004.apm11207-0808.x. [DOI] [PubMed] [Google Scholar]
- 11.Culig Z. Cytokine disbalance in common human cancers. Biochimica et Biophysica Acta. 2011;1813:308–314. doi: 10.1016/j.bbamcr.2010.12.010. [DOI] [PubMed] [Google Scholar]
- 12.Wang D, Dubois RN, Richmond A. The role of chemokines in intestinal inflammation and cancer. Current Opinion in Pharmacology. 2009;9:688–696. doi: 10.1016/j.coph.2009.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ben-Baruch A. The multifaceted roles of chemokines in malignancy. Cancer and Metastasis Reviews. 2006;25:357–371. doi: 10.1007/s10555-006-9003-5. [DOI] [PubMed] [Google Scholar]
- 14.Scheller J, Ohnesorge N, Rose-John S. Interleukin-6 trans-signalling in chronic inflammation and cancer. Scandinavian Journal of Immunology. 2006;63:321–329. doi: 10.1111/j.1365-3083.2006.01750.x. [DOI] [PubMed] [Google Scholar]
- 15.Yang F, Jin C, Jiang YJ, Li J, Di Y. Potential role of soluble VEGFR-1 in antiangiogenesis therapy for cancer. Expert Review of Anticancer Therapy. 2011;11:541–549. doi: 10.1586/era.10.171. [DOI] [PubMed] [Google Scholar]
- 16.Lin WW, Karin M. A cytokine-mediated link between innate immunity, inflammation, and cancer. Journal of Clinical Investigation. 2007;117:1175–1183. doi: 10.1172/JCI31537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Heikkila K, Harris R, Lowe G, Rumley A, Yarnell JA, Gallacher J. Associations of circulating C-reactive protein and interleukin-6 with cancer risk: findings from two prospective cohorts and a meta-analysis. Cancer Causes Control. 2009;20:15–26. doi: 10.1007/s10552-008-9212-z. [DOI] [PubMed] [Google Scholar]
- 18.Pine SR, Mechanic LE, Enewold L, Chaturvedi AK, Katki HA, Zheng YL. Increased levels of circulating interleukin 6, interleukin 8, C-reactive protein, and risk of lung cancer. Journal of the National Cancer Institute. 2011;103:1112–1122. doi: 10.1093/jnci/djr216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang T, Rohan TE, Gunter MJ, Xue X, Wactawski-Wende J, Rajpathak SN. A prospective study of inflammation markers and endometrial cancer risk in postmenopausal hormone nonusers. Cancer Epidemiology, Biomarkersand Prevention. 2011;20:971–977. doi: 10.1158/1055-9965.EPI-10-1222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ho GY, Xue XN, Burk RD, Kaplan RC, Cornell E, Cushman M. Variability of serum levels of tumor necrosis factor-alpha, interleukin 6, and soluble interleukin 6 receptor over 2 years in young women. Cytokine. 2005;30:1–6. doi: 10.1016/j.cyto.2004.08.008. [DOI] [PubMed] [Google Scholar]
- 21.Wong HL, Pfeiffer RM, Fears TR, Vermeulen R, Ji S, Rabkin CS. Reproducibility and correlations of multiplex cytokine levels in asymptomatic persons. Cancer Epidemiology, Biomarkersand Prevention. 2008;17:3450–3456. doi: 10.1158/1055-9965.EPI-08-0311. [DOI] [PubMed] [Google Scholar]
- 22.Linkov F, Gu Y, Arslan AA, Liu M, Shore RE, Velikokhatnaya L. Reliability of tumor markers, chemokines, and metastasis-related molecules in serum. European Cytokine Network. 2009;20:21–26. doi: 10.1684/ecn.2009.0146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.de Jager W, Bourcier K, Rijkers GT, Prakken BJ, Seyfert-Margolis V. Prerequisites for cytokine measurements in clinical trials with multiplex immunoassays. BMC Immunology. 2009;10:52. doi: 10.1186/1471-2172-10-52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Gu Y, Zeleniuch-Jacquotte A, Linkov F, Koenig KL, Liu M, Velikokhatnaya L. Reproducibility of serum cytokines and growth factors. Cytokine. 2009;45:44–49. doi: 10.1016/j.cyto.2008.10.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chaturvedi AK, Kemp TJ, Pfeiffer RM, Biancotto A, Williams M, Munuo S. Evaluation of multiplexed cytokine and inflammation marker measurements: a methodologic study. Cancer Epidemiology, Biomarkersand Prevention. 2011;20:1902–1911. doi: 10.1158/1055-9965.EPI-11-0221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hofmann JN, Yu K, Bagni RK, Lan Q, Rothman N, Purdue MP. Intra-individual variability over time in serum cytokine levels among participants in the prostate, lung, colorectal, and ovarian cancer screening Trial. Cytokine. 2011;56:145–148. doi: 10.1016/j.cyto.2011.06.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Navarro SL, Brasky TM, Schwarz Y, Song X, Wang CY, Kristal AR. Reliability of serum biomarkers of inflammation from repeated measures in healthy individuals. Cancer Epidemiology, Biomarkersand Prevention. 2012;21(7):1167–1170. doi: 10.1158/1055-9965.EPI-12-0110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Clendenen TV, Arslan AA, Lokshin AE, Idahl A, Hallmans G, Koenig KL. Temporal reliability of cytokines and growth factors in EDTA plasma. BMC Research Notes. 2010;3:30. doi: 10.1186/1756-0500-3-302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Ho GY, Melman A, Liu SM, Li M, Yu H, Negassa A. Polymorphism of the insulin gene is associated with increased prostate cancer risk. British Journal of Cancer. 2003;88:263–269. doi: 10.1038/sj.bjc.6600747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P. The Women's Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998;9:117–125. [PubMed] [Google Scholar]
- 31.Kaplan RC, Ho GY, Xue X, Rajpathak S, Cushman M, Rohan TE. Within-individual stability of obesity-related biomarkers among women. Cancer Epidemiology, Biomarkersand Prevention. 2007;16:1291–1293. doi: 10.1158/1055-9965.EPI-06-1089. [DOI] [PubMed] [Google Scholar]
- 32.Helsel D. Less than obvious—statistical treatment of data below the detection limit. Environmental Science and Technology. 1990;24:1766–1774. [Google Scholar]
- 33.Hornung R, Reed LD. Estimation of average concentration in the presence of non-detectable values. Applied Occupational and Enviromental Hygiene. 1990;5:46–51. [Google Scholar]
- 34.Rosner B. Brooks Cole; 2011. Fundamentals of biostatistics; pp. 568–572. [chapter 12, Multisample inference] [Google Scholar]
- 35.Chiba K, Watanabe T, Ikeda M. Variability of serum high density lipoprotein cholesterol concentration in healthy subjects in a three year term. Journal of Epidemiology and Community Health. 1984;38:195–197. doi: 10.1136/jech.38.3.195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Missmer SA, Spiegelman D, Bertone-Johnson ER, Barbieri RL, Pollak MN, Hankinson SE. Reproducibility of plasma steroid hormones, prolactin, and insulin-like growth factor levels among premenopausal women over a 2- to 3-year period. Cancer Epidemiology, Biomarkersand Prevention. 2006;15:972–978. doi: 10.1158/1055-9965.EPI-05-0848. [DOI] [PubMed] [Google Scholar]
- 37.Dugue B, Leppanen E. Short-term variability in the concentration of serum interleukin-6 and its soluble receptor in subjectively healthy persons. Clinical Chemistry and Laboratory Medicine. 1998;36:323–325. doi: 10.1515/CCLM.1998.054. [DOI] [PubMed] [Google Scholar]
- 38.Sites CK, Toth MJ, Cushman M, L’Hommedieu GD, Tchernof A, Tracy RP. Menopause-related differences in inflammation markers and their relationship to body fat distribution and insulin-stimulated glucose disposal. Fertility and Sterility. 2002;77:128–135. doi: 10.1016/s0015-0282(01)02934-x. [DOI] [PubMed] [Google Scholar]
- 39.McIlhenny C, George WD, Doughty JC. A comparison of serum and plasma levels of vascular endothelial growth factor during the menstrual cycle in healthy female volunteers. British Journal of Cancer. 2002;86:1786–1789. doi: 10.1038/sj.bjc.6600322. [DOI] [PMC free article] [PubMed] [Google Scholar]