Increase of Oxidative Stress is Positively Associated with High-Sensitivity C-Reactive Protein in Women Using Combined Oral Contraceptives: An Observational Study


 Background

Information concerning oxidative stress and low-grade inflammation in oral contraception users is scarce. We investigated relationships of oxidative stress and high-sensitivity C-reactive protein (hsCRP) in women by oral combined contraceptive (OC) use.
Methods

Caucasian Italian healthy non-obese women (n = 290; 100 OC-users; 190 non-OC-users; mean age 23.2 ± 4.7 years) were analyzed. Blood hydroperoxides, as oxidative stress biomarkers, were assessed by Free Oxygen Radical Test (FORT), and serum hsCRP was determined by routine methods.
Results

Oxidative stress levels were elevated (≥ 400 FORT Units) in 77.0% and 15.8% of OC-users and non-OC-users, respectively (odds ratio (OR) = 209, 95%CI = 60.9-715.4, p < 0.001); hsCRP levels ≥ 2.0 mg/L were found in 41.0% and 9.5% of OC-users and non-users, respectively (OR = 6.6, 95%CI 3.5–12.4, p < 0.001). Continuous hydroperoxides values in OC-users were 1.7-fold higher compared to non-users (median 472 versus 270 FORT Units, p < 0.001); OC-users had 3.7-fold higher hsCRP values compared to non-OC-users (median 1.31 versus 0.35 mg/L, p < 0.001). Hydroperoxides were strongly positively correlated to hsCRP in all women (rs=0.622, p < 0.001), in OC-users (rs=0.442, p < 0.001), and in non-OC-users (rs=0.426, p < 0.001). Women with hydroperoxides ≥ 400 FORT Units were 8 times as likely to have hsCRP ≥ 2 mg/L. In non-OC-users only, hydroperoxides values were positively correlated with weight and body mass index, but negatively correlated with red meat, fish and chocolate consumption.
Conclusions

Our cross-sectional observational study is the first finding a strong positive correlation of serum hydroperoxides with hsCRP, a marker of inflammation. Further research is needed to elucidate the potential role of these two biomarkers in OC-use side-effects, like thromboembolism.

Progress in the understanding of mechanisms inducing oxidative stress and hsCRP in pre-menopausal women can furnish insights into the regulatory mechanisms underlying oxidative stress and chronic lowgrade in ammation induced damages and could promote further studies concerning the lifestyle and alimentary needs of young women OC-users, as these two serum biomarkers levels are potentially modi able.

Study objectives and aims
The general aim of the study was to investigate the serum hydroperoxides (as indicator of oxidative stress) and hsCRP (as marker of chronic low-grade in ammation) levels in young women according to OC use, accounting for the effects of the alimentary habits and lifestyles. More speci cally, we hypothesized that OC use might have some effect on continuous and strati ed hydroperoxides and hsCRP levels.

Methods
Design and setting of the study A cross-sectional observational study was conducted on healthy Italian adult women enrolled in the years from 2017 to 2019.
All participants attended once the Clinical Biochemistry laboratory of the Medical Department (University of Udine, Italy) early in the morning (8-10 am) and after 12 h fast; a nger capillary and a venous blood withdrawal (23,24) was thus taken on seated subjects (17,41). The day before, volunteers were asked to refrain from alcohol and supplement consumption as well as from physical exercise. Menstrual cycle phase was not accounted for; however, menstruation bleeding days were avoided (days 1 -7 of menstrual cycle) because during the latter days oral contraception consists of drug discontinuation or placebo use. Clinical, demographic, and habit data were blinded to personnel executing the blood collection and samples measurements.

Population
Women (age range 18-46 years) were recruited consecutively; announcements at Udine University were used to enroll the volunteers.
All study participants lled questionnaires and volunteered for a venous and a nger capillary blood withdrawal. The inclusion criteria were: a) no acute (current/recent infection) or chronic disease such as diabetes mellitus, autoimmune diseases, thyroid or cardiovascular disease, or any tumor (23,24); b) absence of pain from whatever cause (42); c) no assumption of anti-in ammatory drugs or antibiotics in the last 2 weeks (23,24), d) no hormonal drugs other than monophasic combined contraceptive pills (23,24), e) not lactating, pregnant, or postmenopausal (43); f) no present or past menstrual dysfunction (44); and h) no major sleeping disorders (17). Eligible women for inclusion in the OC-users group were those using a monophasic combined oral contraceptive for at least 3 months, whereas for inclusion in the non-OC-users control group, women had never used hormonal contraception or had discontinued since at least 3 months any hormonal treatment (17,24). Three-hundred women were examined for eligibility; at the time of blood withdrawal, 10 women were excluded due to acute disease potentially affecting the in ammatory status. A total of 290 participants were thus included in the study (n = 100 OC-users and n = 190 non-OCusers).
Demographic factors, medical history and lifestyle habits (including smoking) were collected by a questionnaire. Moreover, a 2 weeks long diary allowed to collect alimentary habits in the form of daily number of servings as described in (17). Alimentary supplements consisted of a wide variety of commercial products containing variable amounts of single or mixed compounds, assumed by women occasionally, intermittently, or continually. Thus, alimentary supplements were not categorized into speci c subgroups, and a categorical variable (yes or no) designed "supplement use" was employed for statistical analysis.
Body mass index (BMI) was calculated as body mass (kg) divided by the square of stature (m).

Ethical considerations
Written informed consent was obtained from each participant before enrolment and no compensation or incentive was paid to the participants for this study. The study was approved by the Institutional Ethical Committee of the Department of Medicine (University of Udine), and was conducted according to the Declaration of Helsinki.
Oxidative stress assessment Blood hydroperoxides (mainly consisting of lipid hydroperoxides; 17,41) were determined in 20 µL of capillary blood using the Free Oxygen Radical Test (FORT assay; Callegari, Parma, Italy), a 6 minute long colorimetric assay based on the ability of transition metals to catalyze the breakdown of hydroperoxides (ROOH) into radicals, according to the Fenton reaction (41). The FORT assay has been validated to assess oxidative stress in human blood by various recent studies (17,30,41,45). Results were expressed as FORT Units, whereby 1 FORT Unit corresponded to 0.26 mg/L H 2 O 2 (41). Variations of the intra-and inter-assay were both <5.0%. The range limits of measurable values were ≤160 and ≥600 FORT Units. Four non-OCusers had FORT values ≤160 and were set at 160 FORT Units, whereby 9 OC-users had ≥600 FORT Units and were set at 600 FORT Units for statistical calculations.
The threshold of 310 FORT Units was used as high cut-off level of hydroperoxides (17), and a value of 400 FORT Units was employed as very high threshold level of oxidative stress (46). The FORT assays has been validated for clinical use in women and has the advantage to be easily used in outpatient facilities (30,45). Lewis, Newell (45) determined a value for individual biological variations of the FORT assay equal to 5.0%.

C-reactive protein measurement
Just after collection, venous blood was processed for serum separation as already described (24).
Serum hsCRP was measured on Olympus AU5400 biochemistry analyzer (Olympus Diagnostic Systems Group, USA) using an ultra-sensitive immunoturbidimetric assay (Olympus System CRP latex). Detection limit of the assay was ≤0.1 mg/L. Samples with values below the detection limit were set at 0.1 mg/L for statistical calculations. The intra-and inter-assay variations were 1.1%, and 4.0%, respectively. Strati cation of hsCRP levels were performed as described (23,37,39).

Statistical Analysis
Normality of distribution of variables was assessed by the Kolmogorov-Smirnov test. Gaussian distributed data were shown as mean ± standard deviation (SD). The FORT and hsCRP data were not normally distributed, thus, their values were shown as median and interquartile (IQR, 25 th to 75 th percentile) and nonparametric tests were used. Continuous variables were compared by the Mann-Whitney U-test. The comparison of proportions between groups was assessed by Pearson's c 2 -test or Fisher's exact test, as appropriate; speci cally, Fisher's test was used for variables "nulliparity" and "hsCRP >10 mg/L", for all other categorical variables Pearson's test was appropriate. Crude odds ratios (ORs) and 95% con dence intervals (CIs) were evaluated for categorical variables. Logistic regression was performed to evaluate the difference in oxidative stress between groups after adjustment for: age, BMI, smoking, supplement use, chocolate, fresh vegetable, fruits, red meat and sh servings per week. Bivariate relationships were evaluated by Spearman Rho test (r s ). All tests were 2-sided. Statistical signi cance was set at P values <0.05. At an alpha level of 0.01, we had over 95% power to detect a difference in the proportion cases with hsCRP ≥2 mg/L between the OC-users and the non-OC-users groups and to nd a correlation between hsCRP and FORT units with r s >0.4. Statistical analysis was performed using the Statistical Package for Social Sciences (SPSS for Windows, SPSS Inc., Chicago, IL, USA).

Results
On average, the 290 study women were 23.2±4.71 year-olds, had BMI 21.2±2.31 kg/m 2 , 97.9% were nulliparous, mostly enrolled at or employed by Udine University, 82.8% had University level education, 16.6% were smokers, and all had a middle-class socioeconomic status.
Main demographic and lifestyle characteristics and continuous values of oxidative stress and hsCRP of the 100 OC-users and the 190 non-OC-users are summarized in Table 1.
A normal weight range (BMI 18.0-25.0 kg/m 2 ) was found in 91.4% of women, 9 women were underweight (3.1%, BMI <18.0 kg/m 2 ), and 16 were overweight (5.5%, BMI >25.0 kg/m 2 ); none of the women, however, had a BMI ≥30.0 kg/m 2 . OC-users had slightly more frequently an University education (89.0%) than non-OC-users (79.5%), p=0.041 (Table 1). However, most of the studied parameters were not signi cantly different between OC-users and non-OC-users including smoking habits, coffee and nutritional supplement use. In our sample, women smoke on average less than 1 cigarette per day. Only one smoker-woman smoked 26 cigarettes per day, none of the other smokers smoked more than 15 cigarettes per day.  (5) including desogestrel, cyproterone, and drospirenone were used by 40.0% of OC-users. Table 2 shows continuous values of hydroperoxides (FORT Units) and hsCRP (mg/L) in each generation pill users. Use of second generation pill was associated with lower values of hydroperoxides than third (p=0.009) and fourth (p=0.008) generation; conversely, third and fourth generation did not differ.
Concentrations of hsCRP did not vary signi cantly among groups. Strati ed values of oxidative stress and hsCRP Table 3 shows the comparison of OC-users and non-OC-users according to categorized values of hydroperoxides and of hsCRP.  Two thresholds of elevated hydroperoxides levels (≥310 or ≥400 FORT Units) were compared with the corresponding low hydroperoxides levels (<310 or <400 FORT Units, respectively) in the whole group of 290 women.
OR, odds ratio; CI, con dence interval; p value by two-tailed chi-squared test.
Statistically signi cant differences are highlighted by bold characters.  week. Our ndings are consistent with the known antioxidant properties of some foods (47)(48)(49). It is to mention that in our study fruits consumption had a negative correlation with oxidative stress in all the 290 women, however, in the distinct subgroups of 100 OC-users and 190 non-OC-users such negative correlation did not reach statistical signi cance, thus, fruits effects will require further research (48).
We did not found oxidative stress association with smoking and supplement consumption in the study women. The number of cigarettes smoked, however, was very low (on average <1 cigarette/day), moreover participants were asked to avoid supplements use in the day before blood testing. Interestingly, a study (30) performed in post-menopausal women showed that folate administration, in the form of 5methyltetrahydrofolate (5-MTHF), can reduce blood oxidative stress as assessed by FORT values reduction. Particular effects of supplements and/or drugs will require speci c further research.
Elevated BMI is known to increase systemic oxidative stress in the general population (50), and in active adults (32). Con rmatory with other studies, we found that BMI was positively correlated to the FORT levels in non-OC-user women only, but not in OC-users. A similar nding was seen previously in female athletes (17). Apparently, the strong OC induction of oxidative stress cancelled bene ts deriving from reduced BMI.
Results of our current study performed on the general female population concur with a previous investigation evaluating oxidative stress in female athletes (17) founding that 42 OC-users had signi cantly higher hydroperoxides than 102 non-OC-users (OR=42, 95%CI=12-149, p<0.001). However, that study did not measured hsCRP (17).
A research conducted in 40-48 years old Belgian women found a signi cant increase of lipid peroxides in 209 OC-users compared to 119 non-users of contraception (22). A study comparing 32 OC-users with 30 non-OC-users found increased lipid peroxides (+176%, p<0.001) and oxidized LDLs (+145%, p<0.002) in the former group of women (20).
Interesting studies (21,31) investigating the time-course of hydroperoxide elevation in women users of a low estrogen dose pill containing drospirenone demonstrated that oxidative stress increased signi cantly after only one week of OC use, remained constantly elevated during OC use, and returned to basal levels within one week of OC discontinuation, thus suggesting a causative role of OC use in increasing oxidative stress (31).
Mechanisms leading to elevation of hydroperoxides by OC are still not de nitively characterized (31), however, some evidence point to oxidative hepatotoxicity of OC (12). P450 cytochromes (CYPs) catabolizing exogenous hormones can cause increased ROS production (51) and, in turn, hyper-production of free radicals could provoke depletion of antioxidant defenses such as depletion of reduced glutathione (31,33). However, the role of estrogens and progestogens in OC induced oxidative stress is still debated (20,31,52). An in vitro study showed that beta-estradiol treatment of cells resulted cytotoxic through oxidative stress provoking signi cant increase in lipid peroxidation (53).
By recent evidence tissue redox status is adequately re ected by redox blood biomarkers (54), thus, the increased oxidative stress measured in blood associated to OC use likely parallels increased free radicals also in several body organs (55).

C-reactive protein
In the present study, OC-use signi cantly increased all risky levels of hsCRP, while provoking a loss of the protective levels below 0.5 mg/L. Speci cally, OC-users were more likely to have hsCRP levels ≥1 mg/L (crude OR=7.49, adjusted OR=11.3); ≥2 mg/L (crude OR=6.64, adjusted OR=11.1); ≥3 mg/L (crude OR=6.71, adjusted OR=8.05) and ≥5 mg/L (crude OR=3.27, adjusted OR=3.07) than non-OC-users. Finally, OC-users were 9.95 times as likely to have hsCRP levels ≥10 mg/L. These results are consistent with previous studies performed in 77 third generation pill OC-users (23) and 53 OC-users athletes (24). Our results are consistent also with a large Danish study nding low-grade in ammation (hsCRP 3-10 mg/L) in 29.9% of OC-users compared to 7.9% in non-OC users (26).
The role of hsCRP attesting low-grade in ammation in women was highlighted by large studies: women who developed cardiovascular events had higher baseline hsCRP levels than control subjects, so that hsCRP was a strong independent risk factor for any vascular event (RR=4.8; 95%CI=2.3-10.1) and for myocardial infarction or stroke (RR=7.3; 95%CI=2.7-19.9) (56). Further studies con rmed the key role of chronic low-grade hsCRP in risk of future CVDs in women (39).
Recent evidence supports chronic in ammation (CRP values over 10 mg/L) as a mechanism of ovarian carcinogenesis (57). Interestingly, one study (57) showed that ever users of OC had an increased risk for ovarian cancer (OR=3.24, 95%CI=1.62-6.47). Regarding breast cancer (14), a recent review stated that safety of long-term OC use for BRCA1/2 mutation carriers is uncertain, thus, non-hormonal contraceptive methods should be discussed with those women (13). Notably, combined estrogen plus progestogen contraceptives are considered human carcinogens and classi ed in Group 1 by the International Agency for Research on Cancer (14) for the liver and bile duct, breast and uterine cervix cancer.
Combined oral contraceptives may affect the mediators of low-grade chronic in ammation with potential additive risk in women with polycystic ovary syndrome (PCOS); however clinical implications of OC use by PCOS patients need further studies (29).
In ammation (58) and oxidative stress (34) have been implicated in the etiology of depression and disturbed sleep (59); in turn, OC use has been associated with depression (60). More longitudinal research is needed to improve the understanding of mechanisms induced by in ammation in OC-users possibly affecting the psyconeurological pathways (61).

Correlation between oxidative stress and hsCRP
Overall, our data highlighted a strong positive correlation of hsCRP with oxidative stress (p<0.001). Women with oxidative stress over 400 FORT units were eight time as likely to have hsCRP ³2.0 mg/L. So far, there has been a limited focus on this relationship speci cally in women.
It is to highlight that causality cannot be inferred by our data, i.e., the design of our study does not allow to determine whether OC use directly increased ROS production (possibly via oxidative hepatotoxicity) that cause formation of hydroperoxides and then oxidative stress induces in ammation and/or whether OC use directly increase hsCRP that in turn provokes hydroperoxidation. It is to mention that the nuclear factors NF-kB/IkB are key molecular switches both in oxidative stress and in ammation pathways (36), thus, molecular mechanisms could activate at the same time oxidative stress and in ammation.
Many potential adverse effects are linked to elevation of oxidative stress and in ammation including thromboembolic events, endothelial damage, CVDs, and cancer (62,63). The biochemical pathways of oxidative stress and in ammation elevation in users of combined contraceptive pills have, however, to be still completely elucidated.
Impact of the composition of the contraceptive pill The progestin component of combined contraceptive pills may be important in determining side effects of OC use (2,5,19). Recent studies demonstrated that hormonal contraception can modify the redox status in the vasculature of women using combined contraceptive pills containing low doses of ethinyl-estradiol and progestin agents such as drospirenone (20,64) or norethisterone (18).
New generations of OC pills are characterized by lower estrogen content and by newer progestins, like desogestrel, gestodene, cyproterone, and drospirenone with lower androgenicity than past generation pills (52). They have been introduced to reduce severe adverse effects of OC use, especially thromboembolism, and other cardiovascular diseases (11). However, these new OC preparations are still associated with the risk of pulmonary embolism, myocardial infarction, thrombotic stroke and VTE (2,52). A meta-analysis of observational studies (6) found that all four generations of progestin were associated with an elevated risk of ischemic stroke, with a higher rst-ever ischemic stroke risk associated with current OC use compared with non-current OC use (overall summary OR=2. 47 The risk of VTE associated to OC use is of particular concern and has been recently investigated in a total of 10562 cases of thromboembolism (5 (5). Similarly, another study (65) found that the relative risk of VTE for combined oral contraceptives with 30-35 μg ethinyl-estradiol and gestodene, desogestrel, cyproterone acetate, or drospirenone were similar and about 50-80% higher than for combined oral contraceptives with levonorgestrel.
In our study we did not nd statically signi cant differences in hsCRP values between users of second, third and fourth generation pills. However, second generation pill containing levonorgestrel had lower FORT values than third and fourth generation pills. Notably the group of OC containing desogestrel, cyproterone, and drospirenone had higher hydroperoxides values compared to all other progestins. Whether increased risk of thromboembolism in OC-users is mediated by the increased oxidative stress and/or chronic lowgrade in ammation needs further investigations are necessary to assess this relevant issue.

Strengths and limitations
Limits of our study include: recruitment of young adult Caucasian females, and thus, results cannot be generalized to older women and/or to women with different ethnic backgrounds; only women taking monophasic combined contraceptive pills were included in the study, excluding other types of contraceptive drugs; OCs were heterogeneous in type and amount of hormonal components although the majorities were OCs of third generation; detailed data about composition and dosing of potentially antioxidant supplements like vitamin E, C, and beta-carotene were not registered (31,33). Finally, oxidative stress was evaluated by an assay measuring hydroperoxides (expression mainly of lipid peroxidation) (31,41,63), that constitutes only one of the possible indirect markers to assess oxidative stress status (66); however, the FORT assays has been validated for clinical oxidative stress evaluation (30,45).
Strengths of the present study include assessment of oxidative stress and hsCRP and several lifestyles and alimentary habits, the homogeneous ethnic group, the rather narrow age range of women, and strictly healthy subject inclusion.
Based on present data, increase of oxidative stress and hsCRP provoked by OC use poses the question of medical eligibility criteria for contraceptive use (67,68), in particular for women affected by major diseases like high risk of CVDs, having had cancer or with major immune/in ammatory diseases like multiple sclerosis (69), diabetes (41,70) or HIV infection (71,72).

Conclusions
This study adds to the existing evidence that OC use alters the oxidative homeostasis and modify lowgrade in ammatory status of young women. Our present ndings showing a strong positive relationship between oxidative stress and basal chronic in ammation have several potential implications. We found elevated oxidative stress in the vast majority of OC-users (77% with FORT ³400 Units), whereas hsCRP was considerably elevated in less than half the OC-users (41% with hsCRP ³2mg/L). Although both parameters can potentially concur in OC side effects like the increased risk of CVDs, cancer and other diseases, oxidative stress seems even more crucial than in ammation, in particular since there is some lack of consensus about the role of hsCRP in CVD risk (39).
We also investigated lifestyle and alimentary habits of study women. OC-users appear resilient to antioxidant food and low BMI protective effects. However, in our opinion, future studies on food supplements having antioxidant and anti-in ammatory roles (48,73,74) could be designed and especially targeted for OC-user women. Based on our data, concurrent use of pro-oxidant (74) and/or proin ammatory (75) drugs in addition to OC use should also be carefully taken into account, and will require future investigations. We think advisable that, at least in women with pathologic conditions and/or at high risk of developing a pathology, oxidative stress and hsCRP should be monitored over OC time use.

Declarations
Ethics approval and consent to participate Written informed consent was obtained from each participant before enrolment and no compensation or incentive was paid to the participants for this study. The study was approved by the Institutional Ethical Committee of the Department of Medicine (University of Udine), and was conducted according to the Declaration of Helsinki.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.