Inflammatory markers in women with reported benign gynecologic pathology: An analysis of the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial

Abstract

Background:

Associations between benign gynecologic pathologies and circulating inflammatory markers are unknown. Our goal was to evaluate self-reported history of benign gynecologic pathology and subsequent alterations in systemic inflammation.

Methods:

Using nested case-control studies from the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, study-specific associations between self-reported history of benign ovarian cysts, uterine fibroids, and endometriosis with inflammatory marker concentrations were evaluated using logistic regression and combined using meta-analysis. Inflammatory markers associated with individual benign pathologies were mutually adjusted for one another to evaluate independent associations.

Results:

Compared to women without a self-reported history of the pathology evaluated, benign ovarian cysts were associated with increased PAI-1 (OR [95% CI] 6.24 [2.53–15.39], P<0.001) and TGF-β1 (3.79 [1.62–8.86], P=0.002) and decreased BCA-1 (0.38 [0.19–0.73], P=0.004). Uterine fibroids were associated with decreased CXCL11 (0.37 [0.22–0.63], P<0.001) and VEGFR3 (0.40 [0.24–0.65], P<0.001). Endometriosis was associated with increased SIL-4R (4.75 [1.84–12.26], P=0.001).

Conclusions:

Self-reported history of benign gynecologic pathologies were associated with alterations in inflammatory markers that have been previously linked to cancer risk. Understanding interactions between benign gynecologic pathologies and the systemic immune system may help inform disease risk later in life.

INTRODUCTION

Chronic systemic and local inflammation is associated with cancer risk.¹ The body’s immune reaction is tightly linked with the development of many cancers, including endometrial and ovarian cancer.^2–4 Prior studies using the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial data found that PAI-2 and VEGF-A are positively associated with endometrial cancer risk while CCL3, IL13, IL21, IL1b and IL23 are inversely associated with risk.² Additionally, CRP, IL1a, IL8 and TNF-α have been positively associated with ovarian cancer risk.⁴ Many chronic inflammatory states are linked to increases in cancer risk by fostering cell proliferation through growth factors, inflammatory cells, cytokines and DNA damage.⁵ In the case of endometriosis, large cohort studies have confirmed that endometriosis is an independent risk factor for ovarian cancer. However, associations of ovarian cancer with fibroids and other benign pathologies are less clear.^6,7 Often the organ affected by inflammation is most likely to develop cancer, but systemic inflammation may affect cancer development at multiple sites. It is worth understanding whether a self-reported history of certain benign conditions are associated with inflammatory markers known to be linked to increased cancer risk in order to identify risk-factors for women throughout their life.

Benign gynecologic pathologies are very common in the US and have morbidity due to chronic pelvic pain and treatments that may involve surgical intervention. An estimated 20–50% of women have symptomatic uterine fibroids and they are the leading indication for hysterectomy in the US.^8,9 Fibroids have been shown to be influenced by the tumor microenvironment in which inflammation plays a prominent role.¹⁰ Inflammatory modulators favoring the upregulation of chronic inflammatory pathway, as opposed to acute inflammatory pathways, have been shown to locally influence the proliferation, fibrosis and angiogenesis that sustain fibroid growth and formation.^11–13

Prevalence estimates of endometriosis are complicated by the requirement for surgical visualization for diagnosis; population estimates range from 2–43% among asymptomatic women.^14–16 A meta-analysis of genome-wide association studies has shown common genetic variants in endometriosis.¹⁷ Endometriosis is a disease of chronic inflammation caused by ectopic endometrial implants and stroma outside the uterine cavity. Implants are commonly found on the ovaries and peritoneum but have also been shown to spread systemically to distant locations such as the lungs.¹⁸ The chronic inflammatory effects of endometriosis have been linked to a hyperalgesic state, myofascial pain and altered brain chemistry correlated with pain intensity.^19,20 It is not clear whether the local inflammatory effects of these conditions are associated with long-term systemic changes in circulating inflammatory markers among women who have had these conditions.

An estimated 14% of postmenopausal women have benign ovarian cysts.²¹ The pathophysiology of benign ovarian cysts is not fully understood, and they may often be multifactorial arising from endometriosis, polycystic ovarian syndrome, pelvic inflammatory disease, and/or functional cysts. They also vary widely in their pathology including follicular cysts, corpus luteum cysts, dermoid cysts, cystadenomas and endometriomas.²² The role of systemic inflammation in benign ovarian cysts has not been well studied to date.

This study utilized baseline questionnaire data collected in the PLCO Cancer Screening Trial that captured information on self-reported history of benign gynecologic pathology and serum measurements on inflammatory markers to determine relationships between history of these pathologies and long-term alterations in systemic inflammatory markers.

METHODS

Study Design

The PLCO Cancer screening trial recruited approximately 155,000 subjects (78,216 women) 55–74 years of age between 1993 and 2001 from ten cities from the general population and randomized to the screening or non-screening arm of the study.²³ Screening-arm subjects provided blood samples at baseline and five subsequent annual medical examinations.²⁴ Institutional review boards of the US National Cancer Institute and the ten study centers approved the trial, and all participants provided written informed consent.

We utilized data from a series of nested case-control studies within the PLCO trial that measured pre-diagnostic circulating inflammatory markers using the same laboratory/platform (screening arm, Figure 1). Self-reported history of benign gynecologic pathology (yes/no) was used to determine individuals with a history of uterine fibroids, benign ovarian cyst and/or endometriosis. Eligibility criteria for the nested case-control studies included the availability of an unthawed serum sample, consent to biochemical studies, completion of the baseline questionnaire, and no history of cancer (other than non-melanoma skin cancer) prior to participation in the study. The ovarian cancer study utilized serum specimens collected at either the time of the baseline questionnaire, or at a single visit post-baseline. All other studies used a specimen collected at the same time as the original assessment and baseline questionnaire. Serum samples for all participants were selected from a time at which no participants had cancer or a history of cancer and all questionnaire-derived data were from the baseline questionnaire. Cases were individuals who did not have cancer at study initiation/time of inflammatory marker blood-draw, but went on to develop the cancer type specific to that nested study during follow-up. Controls were matched to cases based on criteria for the individual nested case-control studies, but were to be free of the cancer of interest at the time of the matched cases’ diagnosis. The baseline questionnaire responses were then used to identify all reports of benign gynecologic pathology (uterine fibroids, benign ovarian cysts, endometriosis) among the female cases and controls of the six nested case-control studies.

Figure 1.

Study design utilizing data from PLCO Screening Arm (inflammatory marker blood draw) and baseline questionnaire. ^aIn this figure cases are individuals who did not have cancer at study initiation/time of inflammatory marker blood-draw, but went on to develop the cancer type specific to that nested study during follow-up. Controls were matched to cases based on criteria for the individual nested case-control studies, but were to be free of the cancer of interest at the time of the matched cases’ diagnosis. The baseline questionnaire responses were then used to identify all reports of benign gynecologic pathology (uterine fibroids, benign ovarian cysts, endometriosis) among the female cases and controls of the six nested case-control studies.

Associations between inflammatory markers and history of uterine fibroids, benign ovarian cysts, and endometriosis were analyzed via a meta-analysis of data from six case-control studies (i.e. studies of colorectal cancer [208 case and 213 matched control], endometrial cancer [256 case and 245 matched control], lung cancer [150 case and 189 matched control], lung cancer II [169 case and 215 matched control], Non-Hodgkin Lymphoma (NHL) [105 case and 105 matched control], and ovarian cancer [126 case and 122 matched control]) that were previously conducted in the screening arm as described above.²⁴ Details on the exclusion criteria, matching factors and inflammation markers measured in these studies are presented in Table S1. For the current analysis, study participants were limited to non-Hispanic white women without a personal history of cancer prior to randomization. This was due to the inability to estimate propensity scores for other groups of women in whom the data was too sparse. We excluded women with missing data on benign gynecologic pathologies [uterine fibroids (2.5%), benign ovarian cysts (2.7%), endometriosis (3.0%)] or information on adjustment factors [cigarette smoking status (0%), oral contraceptive use (0.3%), postmenopausal hormone use (0.6%), Body Mass Index (BMI, 1.2%)].

Laboratory methods

Circulating levels of immune and inflammation markers were measured, including cytokines, chemokines, growth factors, and soluble products of immune activation (Table S1). Serum specimens were selected from the baseline blood draw for all studies except the ovarian study where 11.4% of samples selected were measured at baseline and the remaining at follow-up visits: 18.1% year 1, 26.2% year 2, 12.8% year 4, and 31.5% year 5. However, all ovarian cancer samples were selected prior to the development of ovarian cancer and collection time was spaced to ensure a relatively equal distribution of specimens between 2 and 14 years prior to eventual cancer diagnosis.² Samples (processed at 1200xg for 15 minutes, frozen within two hours of collection, stored at −70 °C) were used to measure circulating levels of 92 markers (57 markers in the colorectal cancer study, 57 endometrial cancer study, 77 lung cancer study, 51 lung cancer II study, 83 NHL study and 60 ovarian cancer study). These markers were selected based on methodologic work that evaluated the performance and reproducibility of multiplexed assays for measurement of inflammation markers in serum.²⁵ Most markers were measured using Luminex bead-based assays (EMD Millipore, Inc., Billerica, MA). TGF-beta, was measured using a quantitative sandwich enzyme immunoassay (R&D Systems, Minneapolis, MN). Concentrations were calculated using either a four- or five-parameter standard curve. Blinded quality control specimens were included across study batches to assess laboratory drift. Cancer cases and matched controls were included in the same analytic batch. Samples were assayed in duplicate and averaged to calculate concentrations. To evaluate assay performance, a replicate sample from a quality control (QC) pool was included in each batch. To evaluate assay reproducibility coefficients-of-variation (CV) and intraclass correlation coefficients (ICC) of log-transformed marker values were calculated from blinded duplicates in all studies. Logtransformed ICCs were greater than 0.8 in 91%, 91%, 86% and 78% of evaluable markers in the lung, NHL, endometrium/colorectal and ovary studies, respectively.^17,26 Additional information on quality control measurements can be found elsewhere.^17,19,27,28

Statistical Analysis

We developed propensity-score adjusted sampling weights for each nested case-control study to ensure that the analysis accounted for the inclusion/exclusion criteria and sampling plan.^29,30 The derivation of the propensity score approach has been previously described.²⁶ In brief, each participant was assigned a specific sampling weight based on the study in which they were included. For participants included in multiple studies, they were assigned a study-specific sampling weight and included in the study-specific analysis (8.4% of samples included in multiple studies, i.e. shared controls in colorectal and endometrial studies). The sampling weights enabled us to include all participants with marker data (including cancer cases that were cancer-free at blood draw) and made our analysis as representative as possible of the non-Hispanic white females in the PLCO screening arm.

Associations were analyzed by yes/no response to history of benign pathology (uterine fibroids, benign ovarian cysts, endometriosis). We excluded studies from analysis with fewer than 20 cases for an exposure. Therefore, no studies were excluded for uterine fibroids, the ovarian study was excluded for benign ovarian cysts, and the NHL and ovarian studies were excluded for endometriosis. While the endometrial study only had 18 individuals with reported endometriosis, this study shared controls and was assayed at the same time as the colorectal study. Thus, for endometriosis we combined the endometrial and colorectal study populations for analysis. For the 120 individuals who were controls for both the endometrial and colorectal studies we used the colorectal weights for analysis.

Marker levels were categorized into groups based on the proportion of individuals with measurements above the lower limit of detection (LLOD) as follows: markers with greater than 70% of individuals with measurements above the LLOD (colorectal n=44, endometrial n=45, lung n=39, lung II n=37, NHL n=41, ovarian n=27) were categorized into two groups (above or below the median value of the detectable samples) where individuals with non-detectable samples were included with those below the median; markers with 30–70% of individuals with measurements above the LLOD (colorectal n=10, endometrial n=8, lung n=9, lung II n=6, NHL n=10, ovarian n=8) were categorized into two groups [detectable (>LLOD) vs. non-detectable (≤LLOD)]; markers with less than 30% of individuals above the LLOD (colorectal n=3, endometrial n=4, lung n= 29, lung II n=8, NHL n=32, ovarian n=24) were excluded from analysis (See Table S2).

Multivariable adjusted unconditional logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between the serum inflammation markers and self-reported benign pathology history by study. All models were adjusted for potential confounding factors: duration of oral contraceptive use (never, ≤1–5 years, 6+ years), duration of menopausal hormone therapy use (never, ≤1–5 years, 6+ years), cigarette smoking status (never, former, current) and body mass index (BMI; <25, 25–29.9, 30+ kg/m²). Additionally, matching factors varied for each study and were included for adjustment. Study-specific parameter estimates and standard errors were used to perform a random-effects meta-analysis. Sensitivity analyses were conducted limiting to markers with 70% or greater above the LLOD. We also conducted analyses with mutual adjustment for statistically significant markers and with additional adjustment for ibuprofen use as this is a common treatment for pain due to benign gynecologic pathologies and is an anti-inflammatory agent.^1,11 Markers for which results were inconsistent based on heterogeneity I² > 80 across the studies were excluded (uterine fibroids n=5, benign ovarian cysts n=9, endometriosis n=4). Statistical significance was defined as a p-value < 0.05. We applied the false discovery rate (FDR) to account for testing 60 markers separately for uterine fibroids, 55 markers for benign ovarian cysts, and 52 markers for endometriosis.^31,32 Data analyses were conducted in SAS (version 9.4), STATA (version 16).

RESULTS

The distribution of selected characteristics of the studies analyzed are summarized in Table 1. In general, participants’ characteristics were similar across the studies included. Additionally, similar rates and duration of both oral contraceptive and postmenopausal-hormone use were observed across studies.

Table 1.

Characteristics of Study Population

	Colorectal Cancer, N (%)*		Endometrial Cancer, N (%]		Lung Cancer, N (%)		Lung Cancer II, N(%)		Non-Hodgkin Lymphoma, N (%)		Ovarian Cancer, N (%}
Total Included in Study, N		421		501		339		384		210		248
Benign Gynecologic Pathology	Cases	Matched controls	Cases	Matched controls	Cases	Matched controls	Cases	Matched controls	Cases	Matched controls	Cases	Matched controls
Uterine Fibroids, N (%)	43(20.57}	37(18.71)	35(13.38)	24(10.06)	24(15.39)	24(18.13)	34(40.31)	30(18,82)	25(21.54)	19(21.09)	23(17.91)	26(21.19}
Benign Ovarian Cysts, N (%)	24(10.42}	21(10.95)	19(6.39)	21(7.84)	20(14.15)	21(11.42)	16(8.02)	30(12.29)	9(8.64)	13(14.63)	10(9.5)	6(4.93)
Endometriosis, N (%)	12(5.67)	12(4.67)	9(3.91)	9(3.94)	7(4.02)	13(7.92)	11(4.29)	18(4.9)	5(4.06)	5(5,5)	10(9.11)	9(7.73)
BMI Category
<25	83(40.54}	92(42.79)	73(30.23}	111(45.67)	70(50.59)	84(38.17)	93(67.14)	94(44.99)	46(40.97)	44(40.47)	60(42.74)	51(41.03}
25 to 30	72(34.31}	75(32.67)	79(30.49)	86(33.93)	52(31.19)	69(36.39)	51(21.32)	79(34.58)	42(41.05)	37(36.94)	43(41.52)	44(37.03}
>35	53(25.15}	46(24.54)	104(39.28)	48(20.4)	28(1822)	36(25.44)	25(11.54)	42(20.44)	17(17.98)	24(22.58)	23(15.74)	27(21.94}
OC Duration of Use
Never	117(57.08)	113(47.63)	148(S4.79)	110(43.53)	75(49,88)	88(41.48)	74(59.5)	105(49.96)	52(46.4)	55(49.7)	64(52.97)	58(44.12}
SI to 5 years	56(26.49)	56(30.66)	79(31.83}	77(32.37)	44(28.04)	59(29.13)	58(25.59)	63(30)	30(33.21)	37(39.35)	42(33.56)	38(32.66)
> 6 years	35(16.43)	44(21.71)	29(13.37}	58(24.1)	31(22.09)	42(29.39)	37(14.91)	47(20.04)	23(20.4)	13(10.95)	20(13.47)	26(23.21}
Menopausal Hormone Duration of Use
Never	70(36.18)	67(28.95)	100(39.9]	90(37.57)	64(38.81)	77(35.48)	53(47.05)	67(31.28)	42(45.04)	42(40.02)	29(22.63)	48(37.25)
£1 to 5 years	64(30.01}	65(32.4)	76(28.91)	85(34.37)	43(31.18)	46(29.88)	50(21.27)	68(32.09)	24(20.18)	29(29.35)	43(36.73)	38(32.04)
> 6 years	74(33.8)	81(38.65)	80(31.19)	70(28.06)	43(30.01)	66(34.65)	66(31.68)	80(36.64)	39(34.78)	34(30.64)	54(40.64)	36(30.7)
Cigarette Smoking Status
Never	121(54.45)	122(57.9)	161(50.3}	139(57.85)	21(10.85)	55(57.18)	24(34.39)	62(56.05)	70(52.24)	61(56.83)	67(53.81)	80(56.2)
Current	11(8.07)	11(5.65)	13(10.45)	13(6.96)	49(42.3)	53(8.39)	76(31.14)	67(8.61)	8(8.77)	13(7.63)	10(7.19)	11(9.16)
Former	76(37.48)	80(36.45)	82(39.25)	93(35.19)	80(46,86)	81(34.43)	69(34.47)	86(35.34)	27(38,99)	31(35.54)	49(39)	31(34.64)

^aAll percentages are using sample weights for each individual.

Seven of 60 inflammatory markers were associated with prevalence of uterine fibroids. Markers that were positively associated with uterine fibroids were IL-33, CCL2 and IL-16, whereas CXCL11, VEGFR3, CCL27 and CCL21 were inversely associated with uterine fibroids (Table 2). In analyses limited to markers with 70% of individuals above the LLOD we observed consistent associations between uterine fibroids and CCL2 (OR 1.85, CI 1.14–2.98, P=0.012, I²=0), CXCL11 (OR 0.37, CI 0.22–0.63, P<0.001, I²=0), VEGFR3 (OR 0.40 CI 0.24–0.65, P<0.001, I²=0), CCL27 (OR 0.56, CI 0.33–0.93, P=0.025) and CCL21 (OR 0.57, CI 0.34–0.96, P=0.033, I²=0). CCXCLII and VEGFR3 were associated with uterine fibroids at FDR less than 0.05 (Table S3).

Table 2.

Associations between self-reported benign pathologies and inflammatory markers.

UTERINE FIBROIDS
Marker	Studies Included	ORa	95% CIb	Weight (%)
Interleukin 33 (IL-33)	Colorectalc	2.41	(1.04, 5.58)	100.00
	Total	2.41	(1.04, 5.58)	P = 0.039	I2 = n/a
Chemokine Ligand 2 (CCL2)	Colorectal	1.90	(0.81, 4.47)	23.41
	Endometrial	1.37	(0.54, 3.45)	22.04
	Lung	0.79	(0.19, 3.23)	14.63
	Second Lung	3.16	(0.87, 11.42)	16.22
	NHL	1.92	(0.60, 25.64)	10.03
	Ovarian	2.86	(0.64, 12.70)	13.67
	Total	1.85	(1.14, 2.98)	P = 0.012	I2 = 0
Interleukin 16 (IL-16)	Colorectalc	1.66	(0.65, 4.19)	13.87
	Endometrialc	1.30	(0.54, 3.17)	14.74
	Lungc	2.49	(0.58, 10.67)	6.93
	Second Lungc	1.53	(0.39, 6.01)	7.67
	NHL	2.51	(0.60, 10.58)	7.09
	Total	1.68	(1.01, 2.78)	P = 0.045	I2 = 0
Chemokine Ligand 11 (CXCL11)d	Colorectal	0.42	(0.16, 1.10)	23.99
	Endometrial	0.28	(0.09, 0.83)	21.96
	Lung	0.25	(0.07, 0.91)	19.21
	Second Lung	0.31	(0.09, 1.04)	20.41
	NHL	1.36	(0.25, 7.50)	14.43
	Total	0.37	(0.22, 0.63)	P < 0.001	I2 = 0
Soluble vascular endothelial growth factor receptor 3 (SVEGFR3)d	Colorectal	0.37	(0.14, 0.96)	8.45
	Endometrial	0.26	(0.09, 0.78)	7.02
	Lung	0.05	(0.00, 0.89)	1.47
	Second Lung	0.56	(0.19, 1.61)	7.42
	NHL	0.73	(0.18, 2.92)	5.07
	Ovarian	0.45	(0.12, 1.67)	5.54
	Total	0.40	(0.24, 0.65)	P < 0.001	I2 = 0
Chemokine Ligand 27 (CCL27)	Colorectal	0.34	(0.13, 0.90)	23.11
	Endometrial	0.50	(0.17, 1.47)	22.06
	Lung	0.65	(0.17, 2.44)	18.21
	Second Lung	0.72	(0.26, 2.00)	22.52
	NHL	1.21	(0.22, 6.57)	14.11
	Total	0.56	(0.33, 0.93)	P = 0.025	I2 = 0
Chemokine Ligand 21 (CCL21)	Colorectal	0.41	(0.16, 1.07)	13.28
	Endometrial	0.54	(0.21, 1.39)	13.52
	Lung	0.57	(0.15, 2.19)	7.96
	Second Lung	0.72	(0.24, 2.14)	11.00
	NHL	1.54	(0.21, 11.45)	3.94
	Total	0.57	(0.34, 0.96)	P = 0.033	I2 = 0
BENIGN OVARIAN CYST
Marker	Study	OR	95% CI	Weight
Plasminogen activator	Colorectal	6.81	(1.82, 25.50)	48.80
inhibitor-1 (PAI-1)d	Endometrial	5.79	(1.68, 19.91)	51.20
	Total	6.24	(2.53, 15.39)	P < 0.001	I2 = 0
Transforming growth factor beta 1 (TGF-β1)d	Colorectal	5.53	(1.12, 8.89)	43.89
	Endometrial	3.16	(1.62, 8.86)	56.11
	Total	3.79	(1.62, 8.86)	P = 0.002	I2 = 0
Serum Amyloid P (SAP)	Colorectal	1.34	(0.39, 4.62)	27.10
	Endometrial	5.01	(1.34, 18.75)	25.97
	Lung	1.22	(0.23, 6.50)	21.18
	Second Lung	4.55	(1.20, 17.27)	25.75
	Total	2.58	(1.22, 5.45)	P = 0.013	I2 = 16.3
Soluble vascular endothelial growth factor receptor 2 (SVEGFR2)	Colorectal	1.68	(0.59, 4.84)	23.84
	Endometrial	2.33	(0.67, 8.12)	21.45
	Lung	1.04	(0.30, 3.57)	21.69
	Second Lung	5.92	(2.58, 22.16)	20.62
	NHL	1.74	(0.19, 15.78)	12.40
	Total	2.07	(1.15, 3.23)	P = 0.015	I2 = 1.8
B-Cell Attracting Chemokine 1 (BCA-1)d	Colorectal	0.86	(0.30, 2.50)	22.90
	Endometrial	0.44	(0.17, 1.12)	24.45
	Lung	0.27	(0.06, 1.12)	18.80
	Second Lung	0.15	(0.05, 0.47)	22.20
	NHL	0.46	(0.05, 4.37)	11.65
	Total	0.38	(0.19, 0.73)	P = 0.004	I2 = 30.6
Chemokine Ligand 9 (CCL9)	Colorectal	0.35	(0.09, 1.31)	22.19
	Endometrial	0.35	(0.12, 1.06)	24.87
	Lung	0.58	(0.18, 1.88)	23.91
	Second Lung	1.40	(0.36, 5.41)	21.80
	NHL	0.20	(0.01, 5.77)	7.23
	Total	0.52	(0.28, 0.94)	P = 0.032	I2 = 0
ENDOMETRIOSIS
Marker	Study	OR	95% CI	Weight
Soluble Interleukin 4 Receptor (SIL-4R)d	Endometrial/Colorectal	5.95	(1.21, 29.24)	34.51
	Lung	3.39	(0.45, 25.47)	29.25
	Second Lung	4.69	(1.09, 20.09)	36.25
	Total	4.75	(1.84, 12.26)	P = 0.001	I2 = 0
Chemokine Ligand 11 (CXCL11)	Endometrial/Colorectal	0.52	(0.18, 1.51)	38.49
	Lung	0.16	(0.02, 1.02)	29.03
	Second Lung	0.14	(0.03, 0.66)	32.48
	Total	0.27	(0.10, 0.71)	P = 0.008	I2 = 26.4
Chemokine Ligand 22 (CCL22)	Endometrial/Colorectal	0.42	(0.12, 1.51)	41.27
	Lung	0.15	(0.01, 2.46)	22.79
	Second Lung	0.26	(0.05, 1.37)	35.94
	Total	0.32	(0.12, 0.82)	P = 0.019	I2 = 0

^aOdds Ratio (OR).

^bConfidence Interval (CI).

^cIndicates studies for which 30–70% of the values were below the lower limit of detection (LLOD) and were therefore classified as detect/nondetect based on whether or not they were above or below the LLOD.

^dThe q value (estimated probability of a false discovery) has a value less than 0.1.

Six of 55 inflammatory markers were associated with prevalence of benign ovarian cysts. PAI-1 (OR 6.24, CI 2.53–15.39, P<0.001, I²=0), TGF-β1 (OR 3.79, CI 1.62–8.86, P=0.002, I²=0), SAP (OR 2.58, CI 1.22–5.45, P=0.013, I²=0) and VEGFR2 (OR 2.07, CI 1.15–3.73, P=0.015, I²=0) were positively associated with benign ovarian cysts, while BCA-1 (OR 0.38, CI 0.19–0.73, P=0.004, I²=30.6), and CCL9 (OR 0.52, CI 0.28–0.94, P=0.032, I²=0) were inversely associated with benign ovarian cysts (Table 2). All markers associated with benign ovarian cysts had 70% of values above the LLOD. Associations between PAI-1, TGF-β1, and BCA-1 and benign ovarian cysts remained after correction for multiple comparisons (FDR≤0.06); q-values for the other marker associations with benign ovarian cysts were >0.14 (Table S3).

Three of 52 markers were associated with prevalence of endometriosis. SIL-4R was positively associated with endometriosis (OR 4.75, CI 1.84–12.26, P=0.001, I²=0), while CXCL11 (OR 0.27, CI 0.1–0.71, P=0.008, I²=26.4) and CCL22 (OR 0.32, CI 0.12–0.83, P=0.019, I²=0) were inversely associated with endometriosis (Table 2). All markers associated with endometriosis had 70% of values above the LLOD. No values for SIL-4R were below the LLOD and SIL-4R was associated with endometriosis at FDR=0.05; q-values for the other marker associations with endometriosis were >0.2 (Table S3). Additional adjustment for regular non-steroidal anti-inflammatory (NSAID) use (one or more pill per week), a common treatment for pain caused by benign gynecologic pathologies, did not substantially alter inflammatory marker associations for any of the benign reproductive morbidities evaluated (results not shown). Markers remained associated with outcomes in models mutually adjusted for other statistically significant markers (results not shown).

DISCUSSION

In an investigation of the association of self-reported history of benign gynecologic pathologies on systemic inflammatory marker levels in 55–74-year-old women, we found differences in long-term levels of several immune markers between those with and without history of these pathologies. Important in the consideration of these inflammatory markers is their known associations with different types of cancers.

The marker with the strongest association was PAI-1 with prevalence of benign ovarian cysts. The marker PAI-1, also known as serine proteinase inhibitor (serpin-E1), is the principal inhibitor of tissue plasminogen activator and urokinase, which therefore inhibits fibrinolysis, inducing tumor vascularization to promote cell dissemination and tumor metastasis.³³ Compared to women with no history of benign ovarian cysts, PAI-1 was higher in women with a positive history of benign ovarian cysts. This marker is also elevated in women who develop endometrial cancer, as well as a prognostic indicator for women with breast cancer.^17,34 The association of benign ovarian cysts with endometrial cancer has not been studied previously; however, the elevation of inflammatory markers in women with a history of ovarian cysts and at increased risk of endometrial cancer, may inform a potential link between these pathologies. Thus, we need to carefully evaluate whether women with a history of benign ovarian cysts are at increased risk for endometrial cancer. BCA-1, also known as CXCL13, is a chemokine involved in the recruitment of B cells.³⁵ BCA-1 showed a greatly reduced level in women with a history of benign ovarian cysts compared to women without. Elevated somatic gene expression of BCA-1 in serous ovarian cancer was associated with better overall survival compared to wild-type tumors.³⁵ However, another recent study showed an increased risk of epithelial ovarian cancer in women with elevated levels of BCA-1.³⁶ These relationships suggest there may be a systemic effect of ovarian pathologies on the immunoregulation of BCA-1.

There were multiple markers that showed an association with self-reported history of uterine fibroids. While levels of CXCL11 and VEGFR3 in the serum have not been shown to be associated with alterations in cancer risk, CXCL11, which is involved in T cell signaling and angiogenesis,³⁷ is increased in women with hormone receptor positive (HR+) breast cancer compared to HR- cancers and healthy controls.³⁸ Also, VEGFR3, a tyrosine kinase receptor thought to be involved in lymphangiogenesis and maintenance of lymphatic endothelium,³⁹ is elevated in patients with esophageal squamous cell carcinoma.⁴⁰ Interestingly, VEGFR3 has also been shown to have higher expression in the arterioles of women with idiopathic menorrhagia.⁴¹ Since uterine fibroids are also associated with menorrhagia, it is interesting that women with fibroids had lower systemic levels of VEGFR3 compared to women without fibroids.^42–44 Alterations in SIL-4R, which was associated with a self-reported history of endometriosis, has not been specifically studied in regards to alterations in cancer risk. Understanding interactions between benign gynecologic pathologies and the systemic immune system may help inform disease risk later in life for these women.

To further evaluate the relationships between benign gynecologic pathologies and the systemic immune system it would be worthwhile to conduct a large prospective cohort study in which we could evaluate immune markers, that includes sample collection and measurement prior to diagnosis of these benign pathologies as well as post-diagnosis/post-treatment. Understanding whether or not inflammatory markers are elevated prior to diagnosis would help us determine whether the benign pathologies are leading to long-term systemic inflammation or if the inflammatory state of the individual plays a role in the development of these pathologies as well as cancer risk.

This study is limited by the temporal relationship between history of benign pathology and the time of blood draw. The time at which women had been diagnosed with a benign pathology was unknown. All women included were post-menopausal at the time of blood draw, whereas many benign gynecologic pathologies present themselves prior to menopause.^1,11 We were interested in evaluating the long-term associations between reported benign pathologies and the immune system and therefore it was reasonable to utilize data from post-menopausal women who had previously been diagnosed with a benign gynecologic pathology. Finally, we focused on interpreting the results of our study based on markers with the strongest associations (largest effect measures) and that remained after considering multiple comparisons. This study is limited by self-report of gynecologic conditions. This led to a lower reported prevalence of these conditions in the study than would be expected in the general population.^8,14,21 This is likely due to reporting of benign pathologies by women who were more severely symptomatic or underwent intervention whereas women who were asymptomatic or minimally symptomatic may have been less likely to report this history. It would be worthwhile in the future to produce a study that graded the symptomatology and need for intervention in order to assess how disease severity affects immune dysregulation. It is also possible that women with these conditions required surgical intervention and may have underwent oophorectomy or other procedures that led to premature menopause and alteration of immune state. Additionally, most cancer in women occurs after menopause;⁴⁵ therefore, it is important to understand how risk factors earlier in life inform future disease risk for this population.

Another important limitation to this study is the restriction to white non-Hispanic women. It has been shown that black women have a higher lifetime prevalence and more severe symptoms of uterine fibroids as compared to white women.⁴⁶ It is unknown whether these more severe disease states are due to genetic or environmental exposures, but evidence has shown that high stress environments may lead to immune dysregulation.⁴⁷ The relationship between altered inflammatory markers and benign gynecologic pathologies may be particularly interesting in this population and thus additional studies are needed to further evaluate this association.

CONCLUSIONS

This study is the most comprehensive evaluation of pro- and anti-inflammatory markers in women with a self-reported history of benign gynecologic pathologies and highlights the potential long-term effects on the systemic immune system from conditions localized to the uterus and ovaries. Importantly, prevalence of benign ovarian cysts was associated with inflammatory markers that have been linked to alterations in cancer risk^17,36 and may help inform future disease risk in postmenopausal women.