Factors Associated with Inflammation Markers, a Cross-Sectional Analysis

Abstract

Epidemiological studies have reported associations between circulating inflammation markers and risk of chronic diseases. It is of interest to examine whether risk factors for these diseases are associated with inflammation. We conducted a cross-sectional analysis to evaluate whether reproductive and lifestyle factors and circulating vitamin D were associated with inflammation markers, including C-reactive protein, cytokines (IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, IL-13, TNFα), and cytokine modulators (IL-1RA, sIL-lRII, sIL-2Ra, sIL-4R, sIL-6R, sTNF-R1/R2), among 616 healthy women. We confirmed associations of several inflammation markers with age and BMI. We also observed significantly higher levels of certain inflammation markers in postmenopausal versus premenopausal women (TNFα, sIL-1RII, sIL-2Ra), with increasing parity (IL-12p40), and with higher circulating 25(OH) vitamin D (IL-13) and lower levels among current users of non-steroidal anti-inflammatory drugs (NSAIDs) (IL-1β, IL-2, IL-10, IL-12p70, and IL-12p40), current smokers (IL-4, IL-13, IL-12p40), and women with a family history of breast or ovarian cancer (IL-4, IL-10, IL-13). Our findings suggest that risk factors for chronic diseases (age, BMI, menopausal status, parity, NSAID use, family history of breast and ovarian cancer, and smoking) are associated with inflammation markers in healthy women.

1. Introduction

Several studies have shown that elevated inflammation markers, primarily C-reactive protein (CRP), interleukin (IL)-6, and tumor necrosis factor (TNFα), are associated with increased risk of cardiovascular disease, type II diabetes, and other chronic conditions, including cancer [1–9]. We recently conducted a case-control study of ovarian cancer nested in three prospective cohorts, and found that IL-2, IL-4, IL-6, IL-12, and IL-13 were associated with risk [10]. Several inflammation-related markers, including IL-2, IL-5, TNFβ, interferon γ (IFNγ), ICAM, soluble IL-2 receptor (sIL-2R), and soluble TNFα receptor 1 (sTNF-R1), have been found to be positively, and IL-13 inversely, associated with subsequent risk of non-Hodgkin lymphoma (NHL) [11–13]. Given the potential role of inflammation in chronic disease, it is of interest to identify factors that contribute to differences in levels of inflammation markers among healthy people.

The emphasis of previous studies has generally been on the association between a limited number of inflammation markers (usually CRP, IL-6, TNFα) and general lifestyle and/or cardiovascular risk factors [1, 14,15]. The focus of the present study was to assess whether reproductive and lifestyle factors and circulating vitamin D, which is immunomodulatory and has anti-inflammatory properties in vitro [16, 17], are associated with CRP, cytokines (IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p40, IL-12p70, IL-13, TNFα) and cytokine modulators/soluble receptors (IL-1 receptor antagonist (Ra), soluble (s)IL-1RII, sIL-2Ra, sIL-4R, sIL-6R, and sTNF-R1/R2) in healthy women. We selected inflammation markers that showed adequate temporal reproducibility (ICC ≥ 0.5) over a 2–3 year period in preliminary reproducibility studies, thus suggesting that a single inflammation marker measurement can be used to rank women according to their average level [18–20].

2. Materials and Methods

2.1 Study Subjects

For these cross-sectional analyses, we included healthy controls from two nested case-control studies in which inflammation markers have been measured: 1) a study of ovarian cancer in three prospective cohorts: the New York University Women’s Health Study (NYUWHS-OVCA), the Northern Sweden Health and Disease Study (NSHDS), and ORDET in Italy [10]; and 2) a study of non-Hodgkin lymphoma in the NYUWHS cohort (NYUWHS-NHL) [11]. The parent cohorts have been described in detail previously [21–24]. Controls were selected to match cases on age (± 6 months), date of blood sampling (± 3 months), and parent cohort (NYUWHS, NSHDS, or ORDET). Controls also had to be alive and free of cancer at the time of the matched case’s cancer diagnosis. Participants were not eligible for inclusion if they were using hormone replacement therapy (HRT) or oral contraceptives (OC) at the time of blood sampling. Up to two controls per case were selected at random from cohort members who met the above criteria. We included all 616 controls from the nested case-control studies in the present cross-sectional analysis (163 controls from NYUWHS-OVCA, 82 from ORDET, 187 from NSHDS, and 184 from NYUWHS-NHL).

Self-reported questionnaires were used to collect data on lifestyle and reproductive variables and height and weight for the NYUWHS and NSHDS subjects. ORDET participants completed in-person interviews with a nurse who also measured height and weight. Although questionnaires varied according to cohort, harmonization of variables was usually straightforward. Participants were considered to have a family history of breast or ovarian cancer only if cancer was reported for a first degree relative (i.e. mother, sister, or daughter). The NSHDS questionnaire asked about “current use” of NSAIDs whereas the NYUWHS questionnaire asked about NSAID use over the previous four weeks. To harmonize NYUWHS and NSHDS data, we restricted NSAID use to use the day of, or one day prior to, blood sampling for the NYUWHS participants, which is more similar to the data that was collected from NSHDS participants. This cut-off is also justified because the half-life of most NSAIDs is less than two days in circulation [25]. ORDET did not collect information on NSAIDs.

2.2 Laboratory Methods

Luminex multiplex bead-based technology was used for measurement of inflammation markers in serum (NYUWHS and ORDET) and EDTA plasma (NSHDS) [26]. The assay kits and procedures and coefficients of variation (CVs) have been described previously in our reproducibility and case-control studies [10, 11,19, 20].

25(OH)D was measured using a direct, competitive chemiluminescence immunoassay (DiaSorin LIAISON 25 OH Vitamin D TOTAL Assay) for the NYUWHS-OVCA and NYUWHS-NHL serum samples. NSHDS plasma 25(OH)D was measured using a gamma-B 25-hydroxy vitamin D radioimmunoassay (Immunodiagnostic Systems, Inc.). Details about the 25(OH)D measurements have been described previously [27–29].

2.3 Statistical Methods

Cytokine measurements that were below the lower limit of detection (LLD) were imputed. For eleven markers with less than 5% of the measurements below the lower limit of detection (LLD), we assigned a value equal to the midpoint between the LLD and zero. For seven markers with 5% or more values below the LLD (IL-1β, IL-2, IL-4, IL-5, IL-12p70, IL-13, IL-1RA had 17–25% values below the lower limit of detection), we used a maximum likelihood estimation procedure developed by Lubin et al. to perform multiple imputation in the presence of detection limits [30]. For the NHL study, sIL-1RII and sIL-4R were not measured, sIL-2Ra was measured with a different assay, and sIL-6R measurements were outside of the detection limits of the assay, thus NHL samples are not included in the analysis of these four markers. Cytokine values were log-transformed to reduce departures from the normal distribution.

We estimated geometric means for each cytokine adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, NSHDS, ORDET), age (continuous), and BMI (continuous, log₂-transformed) within categories of a number of descriptive variables: age (adjusted for study and BMI only), BMI (adjusted for study and age only), menopausal status (pre/post), number of full term pregnancies, smoking status at blood sampling (current/former/never), first degree family history of breast or ovarian cancer (yes/no), use of NSAIDS at blood sampling (yes/no), use of vitamin supplements at time of blood sampling (yes/no), and ever use of OCs (yes/no). Multivariate regression analyses were conducted to assess which variables were independently associated with each cytokine. Partial Pearson correlations were estimated between log-transformed cytokines and cytokine modulators, adjusting for study, age and BMI. Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC). All statistical tests were two-sided and P-values < 0.05 were considered significant.

The Institutional Review Board of New York University School of Medicine, the Ethical Review Board of the National Cancer Institute of Milan (Italy), the Regional Ethical Committee of the University of Umeå, Sweden, and the Swedish Data Inspection Board reviewed and approved this study.

3. Results

Characteristics of the study subjects are shown in Table 1. The age range at blood sampling was 30–75 years (mean = 55 years), and 66% of the women were post-menopausal. The median BMI was about 25kg/m². Thirty-five percent of women had previously used oral contraceptives, 23% were nulliparous, and 82% of parous women had 2 or more children. Eighteen percent of women reported having a first-degree family history of breast or ovarian cancer. At the time of blood sampling, 18% were current smokers, 17% were taking NSAIDs, and 42% were using multivitamins.

Table 1.

Descriptive Characteristics of Study Subjects

Characteristic	N (%)
Study
NYUWHS- OVCA	163 (26.5)
NYUWHS- NHL	184 (29.9)
ORDET	82 (13.3)
NSHDS	187 (30.4)
Age at blood sampling, y
30–45	123 (20.0)
46–50	74 (12.0)
51–55	101 (16.4)
56–60	144 (23.4)
61–75	173 (28.1)
Missing	1
Body Mass Index, kg/m2
<25	303 (51.4)
25–29	166 (28.1)
30–34	89 (15.1)
≥35	32 (5.4)
Missing	26
Menopausal Status at Baseline
Premenopausal	212 (34.5)
Postmenopausal	403 (65.5)
Missing	1
Parity
Nulliparous	124 (23.1)
1	74 (13.8)
2	181 (33.7)
3	104 (19.4)
4 or more	54 (10.1)
Missing	79
Family history of breast/ovarian cancera
No	413 (82.3)
Yes	89 (17.7)
Missing	114
Use of NSAIDs at blood sampling a
No	446 (83.5)
Yes	88 (16.5)
Missing	82
Smoking status at blood sampling
Never	287 (56.0)
Current	92 (18.0)
Former	133 (26.0)
Missing	104
Use of vitamins at blood sampling a
No	308 (58.0)
Yes	223 (42.0)
Missing	85
Past use of oral contraceptives
Never	344 (64.5)
Ever	189 (35.5)
Missing	83
25(OH)D status b
≤ 50 nmol/L (“insufficient”)	171 (62.9)
> 50 nmol/L (“sufficient”)	101 (37.1)
Missing	344

Note: NSAIDs: non-steroidal anti-inflammatory drugs, 25(OH)D: 25-hydroxy vitamin D

^aHigh proportion of missing data because this variable was not available from ORDET (n=82) subjects.

^bHigh proportion of missing data because this variable was not available from ORDET (n=82) subjects and from subjects who were not included in previous studies on 25(OH)D.

Cytokine Distributions by Subject Characteristics

Age-, BMI-, and study-adjusted geometric means and 95% confidence intervals (CI) for selected characteristics are shown in Tables 2 and 3 for cytokines and cytokine modulators, respectively. Table 4 shows beta estimates from multivariate models controlling for all characteristics significantly associated with each marker.

Table 2.

Cytokines: Adjusted Geometric Means and 95% Confidence Intervals Among Categories of Selected Characteristics

	Geometric Mean (95% Confidence Interval)
Cytokines	CRP (mg/L)	IL-1β	IL-2	IL-4	IL-5	IL-6	IL-10	IL-12 p70	IL-13	TNFα
N above LLD a	616	487	500	511	466	597	609	486	503	616
N imputed: MI	0	129	116	105	150	0	0	130	113	0
N imputed: LLD/2 Characteristics	0	0	0	0	0	19	7	0		0
Age at blood sampling, y b
30–45	0.9(0.7, 1.2)	0.7(0.6,0.9)	5.5(4.3, 6.9)	15.0(10.3, 21.8)	0.3(0.2,0.4)	4.6(3.6, 5.9)	6.2 (5.0, 7.7)	1.3(0.9, 1.7)	7.5 (5.6, 9.9)	2.8(2.5, 3.1)
46–50	1.5 (1.1, 1.9)	0.9 (0.6, 1.2)	7.0(5.1,9.4)	26.8(16.3,43.6)	0.4(0.3,0.5)	7.3(5.2, 10)	6.0(4.4, 7.9)	1.4(0.9, 2.1)	11.5(7.9, 16.4)	2.9(2.6, 3.3)
51–55	1.2 (0.9, 1.5)	0.7 (0.5, 0.9)	4.8 (3.7, 6.3)	15.7(10.3,23.8)	0.3 (0.2, 0.4)	5.2(3.9,6.8)	5.8 (4.5,7.3)	1.1 (0.7, 1.6)	8.1(5.9, 11.0)	3.4(3.0, 3.7)
56–60	1.1 (0.9, 1.3)	0.8 (0.6, 0.9)	5.4(4.3, 6.7)	13.3 (9.4, 18.9)	0.3 (0.2, 0.4)	5.2 (4.1,6.6)	6.1 (4.9, 7.4)	1.3 (0.9, 1.7)	6.6(5.0,8.5)	3.6(3.3, 3.9)
61–75	1.4(1.2, 1.7)	0.7 (0.6, 0.9)	5.2 (4.2, 6.3)	22.1(16.1,30.4)	0.4(0.3,0.5)	8.2 (6.6, 10)	5.6(4.6, 6.8)	1.1 (0.8, 1.5)	8.5 (6.7, 10.8)	3.5 (3.3, 3.8)
P-trend	0.05	0.57	0.47	0.53	0.65	0.01	0.59	0.50	0.80	0.00
Body Mass Index, kg/m2 c
<25	0.8 (0.6, 0.9)	0.8 (0.7, 0.9)	5.7(4.9, 6.6)	17.4(13.7, 22)	0.3(0.3,0.4)	5.6(4.8, 6.5)	6.1(5.3,6.9)	1.3(1.1, 1.6)	8.1(6.7,9.6)	3.1(2.9, 3.3)
25–29	1.3 (1.1, 1.6)	0.6(0.5,0.8)	4.7 (3.8, 5.7)	21.6(15.7, 29.4)	0.4(0.3,0.5)	6.9 (5.6, 8.5)	6.2(5.1,7.4)	1.2(0.9, 1.6)	8.5 (6.7, 10.7)	3.4(3.2, 3.7)
30–34	2.0(1.6, 2.5)	0.6(0.4,0.8)	4.8 (3.6, 6.3)	11.7 (7.4, 18.1)	0.2 (0.1, 0.3)	5.0(3.7, 6.7)	4.7 (3.6, 6.1)	0.8 (0.4, 1.2)	6.5 (4.6, 9.1)	3.4(3.0, 3.7)
≥35	4.7 (3.5, 6.2)	1.2 (0.8, 1.7)	9.2 (5.9, 14)	19.2 (9.2,39.1)	0.3 (0.1, 0.5)	8.7 (5.4, 13.6)	6.9 (4.5, 10.4)	1.4(0.7, 2.5)	9.7 (5.6, 16.3)	4.2 (3.6, 5)
P-trend	0.00	0.97	0.83	0.54	0.40	0.33	0.52	0.22	0.81	0.00
P-value ≥ 30 vs. <30	0.00	0.76	0.62	0.11	0.04	0.80	0.21	0.14	0.44	0.05
Menopausal Status d
Premenopausal	1.2 (0.9, 1.5)	0.7(0.5,0.9)	6.1(4.6, 7.9)	22.1(14.4, 33.5)	0.3(0.2,0.5)	7.3(5.5,9.6)	5.4(4.2, 6.9)	0.9(0.5, 1.3)	9.6(7.0, 13.1)	2.8(2.5, 3.2)
Postmenopausal	1.2 (1.0, 1.4)	0.8 (0.7, 0.9)	5.1 (4.3, 6.0)	15.4(11.8, 20.2)	0.4(0.3,0.4)	5.4 (4.5, 6.4)	6.2(5.3,7.3)	1.4(1.1, 1.7)	7.3 (5.9,8.9)	3.5(3.3, 3.8)
P-value	0.95	0.36	0.34	0.24	0.72	0.13	0.42	0.12	0.23	0.00
Parity d
Nulliparous	1.1(0.9, 1.4)	0.8(0.6, 1.0)	6.0(4.7, 7.7)	18.9(12.8, 27.6)	0.3(0.2,0.4)	6.9 (5.4, 8.8)	5.6(4.5, 7.0)	1.1(0.7, 1.5)	8.6(6.4, 11.4)	3.4(3.1, 3.7)
1	1.3 (1.0, 1.7)	0.8 (0.6, 1.1)	5.1 (3.7, 6.9)	17.0(10.5, 27.2)	0.3 (0.2, 0.4)	5.7 (4.1,7.7)	6.4 (4.8, 8.4)	1.4(0.9, 2.1)	8.1 (5.6, 11.5)	3.0(2.7, 3.4)
2	1.0(0.9, 1.3)	0.8 (0.7, 1.0)	5.6(4.6, 6.8)	13.2 (9.6, 17.9)	0.3 (0.3, 0.4)	5.0(4.0, 6.1)	6.4(5.4, 7.7)	1.5 (1.1, 1.9)	7.4(5.9,9.4)	3.4(3.1, 3.6)
3	1.3 (1.0, 1.6)	0.6(0.5,0.8)	4.7 (3.6, 6.2)	20.7(13.8,30.7)	0.3 (0.2, 0.5)	6.5 (5.0, 8.4)	5.1 (4, 6.5)	1.0(0.6, 1.4)	7.6(5.5, 10.2)	3.2 (2.9, 3.5)
4 or more	1.3(0.9, 1.7)	0.6(0.4, 0.9)	4.9(3.3, 7)	16(9, 27.8)	0.3(0.1,0.4)	5.1 (3.4, 7.3)	5.7 (4.1,7.9)	0.8(0.4, 1.4)	6.7 (4.2, 10.2)	3.2(2.8, 3.6)
P-trend e	0.63	0.17	0.70	0.53	0.87	0.71	0.31	0.07	0.61	0.87
P-value parous/nulliparous	0.83	0.85	0.37	0.80	0.45	0.29	0.52	0.52	0.66	0.63
Family history of breast/ovarian cancerd
No	1.5 (1.3, 1.6)	0.8 (0.7,0.9)	5.4(4.7, 6.2)	19.7(15.9, 24.2)	0.4(0.3,0.5)	7.1(6.2,8.1)	6.4(5.6, 7.2)	1.4(1.1, 1.6)	8.8 (7.6, 10.2)	3.4(3.3, 3.6)
Yes	1.6(1.2, 2)	0.6(0.4,0.8)	4.7(3.5,6.3)	11.7(7.2, 18.5)	0.3 (0.2, 0.4)	5.5(4,7.3)	4.4(3.3, 5.8)	1.1(0.6, 1.6)	5.6(4, 7.9)	3.3(2.9, 3.7)
P-value	0.59	0.10	0.42	0.04	0.13	0.11	0.02	0.28	0.02	0.49
Use of NSAIDs at blood sampling d
No	1.4(1.3, 1.6)	0.8 (0.7, 0.9)	6.0(5.3, 6.8)	18.7(15.2, 22.9)	0.4(0.3,0.5)	6.9 (6.0, 7.8)	6.4(5.7, 7.2)	1.4(1.2, 1.7)	8.8 (7.5, 10.2)	3.4(3.2, 3.6)
Yes	1.2 (0.9, 1.5)	0.6(0.4,0.8)	4.2 (3.1, 5.6)	18 (11.2, 28.4)	0.3 (0.1, 0.4)	6.4(4.7,8.6)	4.3(3.2,5.7)	0.7(0.4, 1.1)	7.3 (5.2, 10.2)	3(2.7,3.4)
P-value	0.13	0.04	0.03	0.88	0.10	0.71	0.01	0.01	0.34	0.07
Smoking status at blood sampling d
Never	1.3 (1.1, 1.5)	0.7(0.6,0.8)	5.2(4.4, 6.0)	22.9(17.8, 29.3)	0.4(0.3,0.5)	6.6(5.6, 7.8)	6.5 (5.6, 7.5)	1.2(0.9, 1.4)	8.6(7.1, 10.3)	3.2(3.0, 3.4)
Former	1.2 (1.0, 1.5)	0.7 (0.5, 0.8)	4.7 (3.7, 5.9)	12.5 (8.5, 18.1)	0.3(0.2,0.4)	5.0(3.9, 6.5)	5.0(4.0, 6.2)	1.0(0.7, 1.4)	7.0(5.3,9.2)	3.5(3.2, 3.8)
Current	1.6(1.2, 1.9)	0.7 (0.5, 0.9)	4.4(3.3, 5.8)	10.1(6.3, 15.9)	0.3 (0.1, 0.4)	5.3(3.9,7.1)	5.1(3.9,6.6)	1.4(1.0, 2.0)	5.0(3.5, 7.0)	3.4(3.1, 3.8)
P-value current vs. never	0.25	0.70	0.36	0.00	0.07	0.23	0.16	0.30	0.01	0.16
P-value former vs. never	0.69	0.63	0.48	0.02	0.27	0.12	0.05	0.58	0.30	0.13
25(OH)D d
≤ 50 nmol/L	1.1 (0.8, 1.3)	0.8 (0.7, 1.0)	6.2(5.0, 7.7)	16.4(11.6, 22.9)	0.3(0.2,0.4)	6.0(4.8, 7.5)	5.7 (4.6, 7.0)	1.3(0.9, 1.7)	8.5 (6.6, 10.8)	3.4(3.1, 3.7)
> 50 nmol/L	1.2 (0.9, 1.6)	1.0(0.7, 1.2)	8.4(6.4, 10.9)	25.7 (16.6, 39.5)	0.5 (0.3, 0.6)	8.2(6.2, 10.9)	7.8 (6.0, 10)	1.4(1.0, 2.0)	13.9(10.1, 18.9)	3.6(3.2,4.0)
P-value	0.49	0.39	0.11	0.12	0.10	0.10	0.08	0.58	0.02	0.42

Note: NSAIDs: non-steroidal anti-inflammatory drugs, 25(OH)D: 25-hydroxy vitamin D, CRP: C-reactive protein, IL: interleukin, TNF: tumor necrosis factor All cytokine units are pg/mL unless otherwise specified.

^aNumber of women for whom the biomarker was above the LLD (fewer women may be included in analyses of characteristics with missing data).

^bModel adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS) and BMI

^cModel adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS) and age at sampling

^dModels adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS), age at sampling, and BMI

^eP-value for trend among parous women

Table 3.

Cytokine Modulators: Adjusted Geometric Means and 95% Confidence Intervals Among Categories of Selected Characteristics

	Geometric Mean (95% Confidence Interval)
Cytokine Modulators	IL-1RA	sIL-1Rlle	sIL-2Rae	sIL-4Re	sIL-6Re	IL-12p40	sTNF-R1	sTNF-R2
N above LLD a	480	429	427	428	411	615	616	615
N imputed: MI	136	0	0	0	0	0	0	0
N imputed: LLD/2 Characteristics	0	0	2	1	0	1	0	1
Age at blood sampling, y b
30–45	400 (291, 551)	4800 (4347, 5299)	478 (430, 532)	699 (639, 765)	49307(43823, 55477)	161(141, 184)	1141(1060, 1228)	673(617,735)
46–50	538 (358, 809)	4870(4265, 5561)	491 (426, 566)	666(590, 751)	59981 (51303, 70128)	188(157, 224)	1204(1091, 1328)	737 (656, 828)
51–55	276(194, 393)	5099 (4590, 5664)	490 (438, 549)	635 (577, 698)	55775(49104, 63352)	162 (140, 188)	1214(1118, 1319)	751(681,828)
56–60	323(241,431)	5566(5063, 6119)	554(500, 613)	611(561,666)	58031(51719,65112)	177 (156, 200)	1302(1216, 1394)	768 (709, 832)
60–75	345 (262, 455)	5970(5451, 6539)	640 (580, 705)	640 (589, 695)	62082(55557, 69374)	173 (154, 194)	1478(1387, 1575)	863 (800, 931)
P-trend	0.21	0.00	0.00	0.08	0.01	0.57	0.00	0.00
Body Mass Index, kg/m2c
<25	387(317,472)	4993 (4683, 5324)	512 (477, 549)	646 (610, 686)	52715 (48789, 56958)	176(162, 192)	1262(1205, 1322)	739 (700, 781)
25–29	251(192,328)	5622 (5174, 6108)	558 (510, 611)	645 (598, 696)	59982 (54286, 66275)	161 (144, 181)	1220(1146, 1298)	806 (748, 867)
30–34	448 (313, 642)	5748 (5146, 6420)	560(496, 631)	655(592,725)	60807 (53187, 69520)	162(139, 188)	1381(1269, 1504)	757(685,837)
≥35	506 (278, 919)	5478 (4438, 6761)	589 (468, 740)	645 (532, 782)	69546(53457,90478)	204(158, 264)	1632(1416, 1882)	839 (709, 993)
P-trend	0.62	0.02	0.08	0.90	0.01	0.87	0.00	0.15
P-value ≥30 vs. <30	0.06	0.13	0.27	0.84	0.08	0.93	0.00	0.69
Menopausal status at blood sampling d
Premenopausal	307 (215, 439)	4592(4126,5110)	460(410, 516)	649 (588, 715)	54014 (47501, 61420)	179 (154, 208)	1274(1172, 1385)	742 (672, 819)
Postmenopausal	381 (302, 481)	5783 (5383, 6213)	588 (544, 635)	647 (606, 691)	58594 (53641, 64004)	167(152, 183)	1290(1224, 1360)	779(732,829)
P-value	0.41	0.00	0.00	0.98	0.39	0.51	0.83	0.50
Parity d
Nulliparous	309 (221, 432)	5493 (4863, 6206)	524 (458, 600)	652 (583, 728)	58354 (50588, 67312)	163(142, 188)	1235(1144, 1334)	795(725,872)
1	311 (205, 470)	5692 (4988, 6495)	494 (427, 572)	625 (554, 705)	57722 (49330, 67542)	156(131, 185)	1230(1119, 1353)	730(651,818)
2	354 (272, 461)	5174(4772, 5609)	545 (499, 596)	673 (625, 724)	57052 (51789, 62850)	168 (150, 188)	1303(1225, 1385)	747 (694, 804)
3	348 (245, 493)	4995(4475, 5577)	541 (479, 611)	621 (561, 686)	55593 (48653, 63522)	171(148, 198)	1324(1221, 1435)	759 (689, 836)
4 or more	417 (250, 693)	4880(4191, 5682)	566 (479, 670)	603 (525, 693)	53713 (44878, 64288)	221 (180, 271)	1287(1150, 1440)	772 (674, 883)
P-trendf	0.47	0.11	0.28	0.41	0.58	0.03	0.49	0.56
P-value parous/nulliparous	0.37	0.46	0.79	0.82	0.63	0.50	0.30	0.44
Family history of breast/ovarian cancer d, e
No	345 (291, 408)	5505 (5181, 5849)	575 (538, 614)	676 (641, 714)	56180 (52631, 59969)	182(169, 196)	1323 (1270, 1377)	785 (747, 824)
Yes	330(231,471)	5507 (4763, 6368)	525 (448, 615)	734 (644, 836)	54512 (46584, 63790)	172(147,202)	1361(1248, 1485)	808 (728, 897)
P-value	0.83	1.00	0.30	0.26	0.73	0.53	0.56	0.62
Use of NSAIDs at blood sampling d, e
No	357 (304, 420)	5609 (5306, 5929)	560 (527, 595)	680 (647, 715)	54106 (50914, 57499)	184(171, 198)	1336(1284, 1390)	783(747,821)
Yes	332 (229, 483)	4953(4242,5783)	531 (448, 630)	655 (569, 754)	55004 (46325, 65310)	154(131, 180)	1257(1151, 1373)	771 (694, 857)
P-value	0.73	0.14	0.57	0.63	0.86	0.05	0.22	0.80
Smoking status at blood sampling d
Never	358 (290, 442)	5167(4837,5519)	514 (481, 550)	665 (628, 703)	54126(49814, 58812)	179(163, 196)	1258(1195, 1323)	743 (701, 789)
Former	324 (237, 443)	5397(4855,5999)	568 (510, 633)	624(570, 682)	60654 (52942, 69491)	183 (160, 210)	1307(1213, 1409)	869 (796, 948)
Current	337 (233, 488)	5428 (4829, 6101)	585 (519, 660)	689 (624, 761)	59790 (51297, 69689)	147(125, 173)	1310(1198, 1432)	771(695,855)
P-value current vs. never	0.64	0.48	0.07	0.61	0.35	0.05	0.47	0.49
P-value former vs. never	0.70	0.46	0.14	0.25	0.21	0.72	0.36	0.00
25(OH)Dd
≤ 50 nmol/L	329 (248, 437)	5558 (5071, 6092)	541 (494, 593)	663 (610, 720)	53340 (48668, 58461)	167(149, 188)	1304(1215, 1399)	743 (684, 806)
> 50 nmol/L	515(357,743)	5318 (4622, 6119)	559 (486, 642)	678 (598, 769)	60096 (52202, 69184)	193 (166, 225)	1361(1242, 1493)	772 (694, 859)
P-value	0.07	0.61	0.72	0.77	0.17	0.15	0.48	0.59

Note: NSAIDs: non-steroidal anti-inflammatory drugs, 25(OH)D: 25-hydroxy vitamin D, IL: interleukin, RA: receptor antagonist, Ra: receptor alpha, R: receptor, TNF: tumor necrosis factor All cytokine units are pg/mL.

^aNumber of women for whom the biomarker was above the LLD (fewer women may be included in analyses of characteristics with missing data).

^bModel adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS) and BMI

^cModel adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS) and age at sampling

^dModels adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS), age at sampling, and BMI

^esIL-1RII and sIL-4R were not measured in the NYUWHS-NHL study (n=184 subjects). Because sIL-2Ra was measured in the NYUWHS-NHL study using a different manufacturer’s kit and protocol, the NYUWHS-NHL subjects are not included for this marker. sIL-6R values were above the upper limit of detection for 90% of the NYUWHS-NHL samples, thus the NHL subjects are not included for this marker.

^fP-value for trend among parous women

Table 4.

Multivariate Regression Coefficient Estimates for Factors Independently Associated with Cytokines and Cytokine Modulators

Cytokines and Modulators	Beta coefficient (standard error), P-value
	Age	BMI	Menopausal statusa	Family history of breast or ovarian cancer	Current use of NSAIDs	Current vs. never smokerb	Former vs. never smokerb	Current vs. never/former smoker
CRP	0.01 (0.005) P = 0.04	1.84 (0.17) P < 0.0001
IL-1β					−0.21 (0.10) P = 0.04
IL-2					−0.44 (0.19) P = 0.02
IL-4						−1.09 (0.34) P = 0.001	−0.88(0.31) P = 0.004
IL-5		−0.13 (0.06) P = 0.04c
IL-6	0.02 (0.09) P = 0.01						−0.39(0.19) P = 0.04
IL-10				−0.43 (0.18) P = 0.02	−0.50 (0.19) P = 0.01
IL-12p70					−0.52 (0.18) P = 0.004
IL-13						−0.65 (0.24) P = 0.01
TNFα	0.001 (0.005) P = 0.77	0.41 (0.10) P < 0.0001	0.24 (0.08) P = 0.004
sIL-1RII	−0.002 (0.006) P = 0.72	0.30 (0.15) P = 0.04	0.33 (0.11) P = 0.003
sIL-2Ra	0.001 (0.007) P = 0.88	0.39 (0.16) P = 0.02	0.35 (0.12) P = 0.004
sIL-6R	0.01 (0.005) P = 0.02	0.42 (0.18) P = 0.02
IL-12p40								−0.26(0.12) P = 0.04
sTNF-R1	0.02 (0.003) P < 0.0001	0.37 (0.10) P = 0.001
sTNF-R2	0.02 (0.004) P < 0.0001						0.19 (0.08) P = 0.02

Note: NSAIDs: non-steroidal anti-inflammatory drugs, CRP: C-reactive protein, IL: interleukin, Ra: receptor alpha, R: receptor, TNF: tumor necrosis factor

All models adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS); BMI was log₂-transformed, cytokines and cytokine modulators were log₂+1 transformed, and age, BMI, and inflammation markers were modeled on the continuous scale unless otherwise noted. IL-1RA and sIL-4R are not included in the table because there were no significant predictors.

^aModels adjusted for menopausal status are also adjusted for age even if it was not statistically significant.

^bModels evaluating smoking status include a categorical variable with levels for current, former, and never smokers (as shown in Tables 2 and 3), though betas are only shown for the associations that were statistically significant.

^cThe association with BMI was significant for BMI ≥ 30kg/m² vs. BMI < 30kg/m², but not for BMI on the continuous log scale.

Age and Menopausal Status

There was a significant increasing trend across age groups for CRP, IL-6, and TNFα (Table 2) and for sIL-1RII, sIL-2Ra, sIL-6R, sTNF-R1, and sTNF-R2 (Table 3). After adjusting for menopausal status (Table 4), age was no longer significantly associated with TNFα, sIL-1RII, and sIL-2Ra; these three markers were significantly higher in postmenopausal women (Tables 2–4).

BMI

As shown in Tables 2 and 3, age-adjusted inflammation marker levels increased significantly with increasing BMI for five markers: CRP, TNFα, sIL-1RII, sIL-6R, and sTNF-R1. Though other cytokines did not show a significant trend with increasing BMI, levels were highest among obese women (BMI > 35 kg/m²) for several markers.

Parity

IL-12p40 showed a significant increasing trend with parity (Table 3), though this association was no longer significant after adjustment for smoking (Table 4).

Family History of Breast or Ovarian Cancer

Several markers were significantly lower in women with a family history of breast or ovarian cancer: IL-4, IL-10, and IL-13 (Table 2). However, the association between family history of breast and ovarian cancer was no longer significant for IL-4 or IL-13 after controlling for smoking (Table 4).

Current use of NSAIDs

Current users of NSAIDs had significantly lower levels of IL-1β, IL-2, IL-10, IL-12p70, and IL-12p40 (Tables 2 and 3). However, NSAID use was no longer associated with IL-12p40 after controlling for smoking (Table 4).

Smoking Status

Current smokers had significantly lower levels of IL-4, IL-13, and IL-12p40 (Tables 2–4). Former smokers had significantly lower levels of IL-4 and IL-10 (Table 2) and higher levels of sTNF-R2 (Table 3). However, IL-10 was not associated with smoking in multivariate models, while IL-6 was lower among former smokers in the multivariate model adjusted for age (Table 4).

Vitamin D (25(OH)D)

Women with 25(OH)D levels above 50 nmol/L had generally higher levels of inflammatory markers than did women with lower 25(OH)D levels, but the difference was only statistically significant for IL-13 (Tables 2 and 3), and this difference was no longer significant after adjustment for smoking (Table 4).

Other Variables

We did not observe any significant differences in cytokine levels related to ever use of oral contraceptives or current use of multivitamins. Thus, these characteristics are not shown in Tables 2–4.

Cytokine Correlations

Pearson correlation coefficients adjusted for age, BMI, and study are shown in Table 5. CRP was weakly correlated with the cytokine IL-6 and the cytokine modulator sIL-2Ra (r~ 0.1–0.2). Cytokines were significantly correlated with all other cytokines. The strongest correlations (r > 0.6) were observed between IL-1β and IL-2, IL-10 and IL-12p70, and among the following markers: IL-4, IL-5, IL-6, and IL-13. Correlations between cytokines and cytokine modulators were weak for most pairs (r <0.4), though moderate correlations (r~0.5) were observed between IL-12p40 and two modulators: IL-1RAandsTNF-R1.

Table 5.

Adjusted Pearson Correlation Coefficients for Cytokines and Cytokine Modulators^a

	CRP	IL-1β	IL-2	IL-4	IL-5	IL-6	IL-10	IL-12 p70	IL-13	TNFα	IL-1RA	sIL-1R ll	sIL-2Ra	sIL-4R	sIL-6R	IL-12 p40	sTNF-R1
IL-1β	0.02
IL-2	0.03	0.77***
IL-4	0.06	0.20***	0.22***
IL-5	0.01	0.22***	0.21***	0.58***
IL-6	0.12*	0.29***	0.28***	0.84***	0.62***
IL-10	0.07	0.36***	0.32***	0.11*	0.37***	0.21***
IL-12 p70	−0.02	0.49***	0.45***	0.14**	0.35***	0.35***	0.66***
IL-13	0.00	0.31***	0.27***	0.65***	0.43***	0.68***	0.36***	0.41***
TNFα	0.07	0.30***	0.23***	0.12*	0.19***	0.21***	0.34***	0.34***	0.19***
IL-1RA	0.03	0.35***	0.36***	0.05	0.04	0.04	0.06	0.11*	0.06	−0.03
sIL-1Rll	0.12	−0.02	−0.01	0.07	0.10	0.05	0.07	0.01	0.04	0.12	−0.06
sIL-2Ra	0.19**	0.03	0.02	0.06	0.05	0.04	0.11	−0.03	−0.01	0.15*	0.00	0.41***
sIL-4R	−0.05	0.18**	0.21***	0.13*	0.13*	0.18**	0.09	0.25***	0.11	0.09	0.09	0.20***	0.33***
sIL-6R	0.05	−0.02	−0.04	−0.01	−0.03	−0.01	−0.08	−0.10	−0.01	0.07	−0.15*	0.01	−0.01	−0.08
IL-12p40	0.04	0.29***	0.26***	0.08	0.07	0.04	0.18***	0.12*	0.11*	0.14**	0.56***	−0.04	0.18**	0.05	−0.10
sTNF-R1	0.10	0.19***	0.16***	0.01	0.01	0.03	0.09	0.03	0.02	0.11*	0.46***	0.00	0.25***	0.10	−0.08	0.57***
sTNF-R2	0.10	−0.06	−0.08	−0.07	0.02	−0.05	0.02	−0.07	−0.06	0.11*	−0.05	0.04	0.16**	−0.10	0.14*	0.08	0.18***

Note: CRP: C-reactive protein, IL: interleukin, RA: receptor antagonist, Ra: receptor alpha, R: receptor, TNF: tumor necrosis factor. Cytokines and cytokine modulators were log transformed.

Coefficients with * are significant at P< 0.01, with ** at P<0.001, and with *** at P<0.0001.

Correlations are based on a different number of subjects for each pair of markers, ranging from 390 to 616 subjects.

^aAll models adjusted for study (NYUWHS-OVCA, NYUWHS-NHL, ORDET, NSHDS), age at blood sampling, and body mass index.

4. Discussion

Our results support three previous studies that observed higher levels of the pro-inflammatory cytokine TNFα among post-menopausal women [31–33], although others did not observe such a difference [34, 35], or observed lower levels [36]. Two soluble cytokine receptors, sIL-1RII and sIL-2Ra were also higher among post- vs. pre-menopausal women. The associations were apparent after adjustment for age and BMI, which suggests that these markers may be influenced by sex hormones.

Increasing parity was associated with higher levels IL-12p40, although this association was no longer statistically significant after adjustment for smoking. We are unaware of previous reports that have evaluated the association between previous pregnancies and inflammation markers. However, a microarray study of normal breast tissue found that several inflammation-associated genes were upregulated in both recently (0–2 years since pregnancy) and distantly (5–10 years since pregnancy) parous vs. nulliparous women [37]. However, despite adjustment for age and BMI, we cannot exclude the possibility that the association may have been confounded by other factors, including age at each pregnancy and time since last pregnancy.

Women with a first degree family history of breast or ovarian cancer had lower levels of several cytokines. While inflammation marker levels are not likely to be directly related to family history, this risk factor may be associated with adoption of healthy lifestyle elements.

Current use of NSAIDs was inversely associated with a number of inflammatory markers (IL-1β, IL-2, IL-10 and IL-12p70), in line with our expectations. However, NSAID use was not associated with levels of CRP. Prior reports on the association between NSAID use and CRP are conflicting [38–43], which may be due to differences in duration, types, and doses of NSAID evaluated, but also because some studies were limited in sample size [38–40, 42], included subjects with existing chronic disease [43], or reported associations for regular users who were asked to abstain from use prior to blood sampling [41].

Somewhat contrary to our expectations, smoking status was not significantly associated with CRP, IL-6, TNFα or other pro-inflammatory cytokines. In our study, current smokers generally had lower levels of cytokines than non-smokers, though these associations were only statistically significant for the anti-inflammatory markers, IL-4, IL-13 and IL-12p40. Lower levels of these cytokines among smokers vs. non-smokers were observed in all four sub-studies (NYUWHS-OVCA, NYUWHS-NHL, ORDET, and NSHDS). Previous studies on current smoking and cytokines among healthy individuals have found positive [34, 44–50], inverse [51, 52], and null [49, 52–55] associations. Inconsistent findings may be a result of incomplete control for lifetime exposure to cigarette smoke (e.g. pack-years), which may be a more relevant measure of exposure than current smoking status.

CRP was not correlated with most cytokines. Given the role of IL-6, and to a lesser extent TNFα and IL-1β, in the induction of CRP, we expected to observe moderate to strong correlations of CRP with these cytokines. In models adjusted for study and age (but not BMI), CRP was positively correlated with TNFα (r = 0.1, P < 0.005) and IL-6 (r= 0.1, P=0.001), but not IL-1β. Others have found a moderate correlation between CRP and IL-6, some of which adjusted for BMI, with most estimating a correlation coefficient between 0.3–0.5 [2, 14, 34, 41, 50, 56, 57]. However, data from the Women’s Health Initiative trial suggests that there could be IL-6-independent pathways for regulating CRP, as CRP and IL-6 had different associations with several cardiovascular risk factors: HRT use, alcohol use, and exercise [44]. For example, the consistent association between CRP and HRT [58] vs. the weak or null association between IL-6 and HRT, may be due to direct (IL-6-independent) hepatic induction of CRP [44].

The observed significant correlation between TNFα, IL-6, and IL-1β was expected due to the regulatory inter-relationship of these cytokines [34, 41, 56, 59, 60]. Cytokines were correlated with each of the other cytokines, and less so with cytokine modulators. In physiological conditions, elevations in pro-inflammatory cytokines trigger elevations in anti-inflammatory cytokines to resolve the inflammatory response. Thus, we expected all pro- and anti-inflammatory cytokines to be positively correlated with each other in healthy women. However, since all the cytokines were measured using the same assay kit, we cannot rule out the possibility that imperfect antibody specificity could have contributed to the strength of the correlations. The relationship between cytokines and their modulators is complex, because soluble cytokine receptors can act as both cytokine agonists and/or antagonists [61]. Possible reasons for the lack of correlation between cytokines and their modulators have been suggested by others, such as the longer half-life of soluble receptors in circulation and/or the inability to detect all unbound and bound forms of cytokines with immunoassays [62–64].

The lack of association between cytokines and 25(OH)D, the best representation of an individual’s vitamin D status, is in agreement with a study of vitamin D supplementation in overweight and obese individuals that did not find an association between circulating 25(OH)D and CRP or cytokines (including IL-2, IL-4, IL-5, IL-10, IL-12, IL-13) [52]. We found some evidence of a positive association for IL-13 and 25(OH)D, though the association was not significant in multivariate analyses. Others have reported a lack of association between 25(OH)D and several inflammation markers (CRP, IL-1β, IL-6, IL-10, SIL-2R, TNFα, sTNF-R1, and sTNF-R2) [65–70], though one small study found an inverse association between 25(OH)D and TNFα [65] and two others found that vitamin D supplementation stabilized [71] or reduced [72] TNFα among patients with congestive heart failure or obesity, respectively. We note that the median 25(OH)D level in our study group (median: 48 nmoI/L, interquartile range: 25, 81 nmoI/L) is considered to be in the “insufficient” range (~<50 nmoI/L) based on some recommendations [73, 74].

This study has a number of strengths. First, participants were healthy and not using HRT or OCs for at least 6 months prior to blood sampling, thus minimizing the effects of existing disease and exogenous hormones on cytokine levels. Second, we evaluated associations between risk factors and a large number of cytokines and cytokine modulators which have not been evaluated in previous studies. Third, all inflammation marker measurements were performed in the same laboratory, which minimizes assay variability and allows individuals to be ranked relative to others.

Our study also has some limitations. First, questionnaires differed between cohorts. However, results were usually consistent across cohorts, providing confidence in our findings. Second, it is possible that measured levels of a single cytokine are not reflective of their biologically relevant concentration. This may be due to assay limitations in quantification of absolute levels and/or the presence of circulating cytokine inhibitors, including soluble receptors and other binding proteins. Finally, the exploratory nature of our analyses resulted in a substantial number of statistical tests, thus it is likely that some associations were significant due to chance. However, we did not adjust for multiple comparisons because our goal was to identify general patterns of association that may explain risk factor associations and to aid in the selection of potential confounders for future studies of cytokines and chronic disease risk. The associations observed here should be evaluated in independent studies.

In conclusion, we observed the expected relationships between inflammation markers and age and BMI. Higher levels of certain markers were observed among postmenopausal vs. premenopausal women (TNFα, sIL-1RII, and sIL-2Ra) and with increasing parity (IL-12p40). Lower levels were observed among current versus non-users of NSAIDs (IL-10, IL-β, IL-10, IL-12p70, and IL-12p40), women with a family history of breast or ovarian cancer (IL-4, IL-10, and IL-13), and current versus non-smokers (IL-4, IL-13, and IL-12p40). These findings suggest that one mechanism underlying the relationship between reproductive and lifestyle factors and chronic diseases may involve inflammation mediators.

Highlights.

• We assessed potential determinants of inflammation markers among 616 healthy women.

• Observed associations between lifestyle and reproductive factors and inflammation markers.

• Associations: age, BMI, menopause, parity, NSAIDs, family history of cancer, smoking.

Abbreviations

25(OH)D

25-hydroxyvitamin D

BMI

body mass index

CV

coefficient of variation

CI

confidence interval

CRP

C-reactive protein

ORDET

diet and hormones in the etiology of cancer study

EDTA

ethylenediaminetetraacetic acid

HRT

hormone replacement therapy

ICAM

inter-cellular adhesion molecule

IFN

interferon

IL

interleukin

IL-1Ra

interleukin-1 receptor antagonist

ICC

intra-class correlation coefficient

LLD

lower limit of detection

MET

metabolic equivalent tasks

NYUWHS

New York University Women’s Health Study

NHL

non-Hodgkin lymphomas

NSAID

non-steroidal anti-inflammatory drug

NSHDS

Northern Sweden Health and Disease Study

OC

oral contraceptives

PBMC

peripheral blood mononuclear cell

R- suffix

receptor

s -prefix

soluble

sIL-1RII

soluble IL-1 receptor

sIL-2Ra

soluble IL-2 receptor alpha

sIL-4R

soluble IL-4 receptor

sIL-6R

soluble IL-6 receptor

TNF

tumor necrosis factor