Recreational Exercise Before and During Pregnancy in Relation to Plasma C-Reactive Protein Concentrations in Pregnant Women.

The Norwegian Mother and Child Cohort Study (MoBa) is a prospective population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health. 19 Over 90,000 participants were recruited from all over Norway from 1999 -2008, and 39 % of invited women consented to participate. They received a postal invitation and the baseline questionnaire in pregnancy weeks 13 -17. 21 Maternal blood samples were collected when they had an ultrasound scan in week 17-18 of pregnancy. 22 The Regional Committee for Medical Research and the Norwegian Data Inspectorate approved the study and informed consent was obtained from each participant. A nested case-control study of the MoBa mothers was done to evaluate the relation of environmental contaminants to fertility. Mothers who enrolled in 2003 and 2004 and who delivered a live born child were eligible. Among eligible women, we randomly selected 400 who planned their pregnancy and were sub-fecund (i.e. who reported time to pregnancy greater than 12 months). Additionally, we randomly selected 550 cohort members who reported a time to pregnancy of any duration. 23 For the present study, we only used data from these 550 subjects. The baseline questionnaire inquired about frequencies (times/week) of 14 recreational exercises in the last 3 months before pregnancy, and separately during pregnancy. For each period, a metabolic equivalent task (MET) score was assigned to each exercise; 24 intensity of exercise was defined based on MET score. Specifically, low-intensity exercise (METs < 3) included walking; moderate-intensity exercise (METs 3 - 6) included brisk walking, bicycling, fitness training, special gymnastics/aerobics, aerobics/gymnastics dance without running, folk dancing, ball sports, swimming, riding, and other (Pilates and strength training); and vigorous-intensity exercise (METs > 6) included skiing, running/jogging/orienteering, and aerobics/gymnastics/dance with running. For recreational exercise before and during pregnancy, the sum over exercises of all intensities was calculated (total exercise in MET-hours/week) separately. The baseline questionnaire did not collect information about the time spent on each recreational exercise. We used the reported median time spent on each recreational exercise, ascertained in a preliminary version of the questionnaire (N = 2,555), to impute the duration of each recreational exercise in the present study. The approach has been validated in a previous MoBa study by comparison with motion sensor measurements. 25 Using these values, we calculated a score for vigorous-intensity exercise in MET-hours/week as the sum of METs for each vigorous exercise × hours spent × frequency (times/week). MET-hours/week of low- and moderate-intensity exercise was calculated in a similar way. CRP levels are not affected by food intake and have almost no circadian variation, thus it was not necessary to obtain fasting blood samples. 26 We measured CRP concentrations in non-fasting plasma drawn around 17 weeks of gestation by a high-sensitivity turbidimetric immunoassay using CRP Ultra Wide Range Reagent kit (Genzyme Diagnostics, Framingham, MA) on an Olympus AU400e Clinical Chemistry Analyzer (Olympus America, Inc., Irvin, TX). The detection limit was 0.05 mg/L. The coefficient of variation was < 6 % (intra-assay) and < 7 % (inter-assay). Elevated CRP concentrations were defined as ≥ 8 mg/L (~78 th percentile), a category associated with adverse pregnancy outcomes in a previous study. 27 CRP concentrations were naturally log transformed to obtain an approximately normal distribution and geometric means were used to describe the distribution. Means (standard deviations, SDs) and medians (inter-quartile ranges, IQRs) were used to describe the total recreational exercise and intensity of recreational exercise both before and during pregnancy. The correlation between total recreational exercise before pregnancy and that during pregnancy was evaluated by the Spearman correlation coefficient. We used linear regression models to investigate the association of total recreational exercise and naturally log transformed CRP levels, and logistic regression models to estimate odds ratios (ORs) for association between exercise and having an elevated CRP (CRP ≥ 8 mg/L). Exercise before and during pregnancy was analyzed in the models separately and after mutual adjustment. In addition, we calculated change in exercise as total recreational exercise before pregnancy minus exercise during pregnancy (MET-hours/week); we also fit a model with a term for exercise before pregnancy and a term for average exercise across the two periods. To express the coefficient (β) for exercise as a mean percent change in original units of CRP for total recreational exercise change (MET-hours/week), β was converted to 100 × [exp (β) – 1]. 28 Influential observations were also identified in the regression models and were excluded before refitting the models. Results were quite similar and therefore are not shown. We included women's age and gestational week at blood draw in the models a priori, and considered other covariates if they were statistically related to both exercise and CRP levels in the present data (linear regression model, P < 0.1). The covariates that met the criteria for adjustment were as follows: women's age, education level, gestational week at blood draw, and smoking during this pregnancy (yes/no), which were obtained from the baseline questionnaire; body mass index (BMI, kg/m 2 ) which was calculated as pre-pregnancy body weight (kg) divided by the square of body height (m 2 ); fruit consumption (categorized as < 4 and ≥ 4 servings/day) which was reported on MoBa food frequency questionnaire at about week 22 of pregnancy; 29 and energy intake (kcal/day) which was calculated based on the food frequency questionnaire. Covariates that were considered for inclusion but did not meet the criteria were: diseases during pregnancy (throat infection, ear infection, pneumonia, asthma, diabetes, heart disease, hypertension, arthritis, endometriosis, urinary tract infections, kidney infection, cancer and depression); and medications taken during pregnancy (antibiotics, antiplatelet agents, lipid lowering agents, anti-diabetic agents, estrogens, β-adrenoreceptor antagonists, antioxidants, inhibitors of renin-angiotensin system, and calcium channel antagonists). The questionnaire ascertained only two things about occupational activity during pregnancy: the degree to which the work was physically heavy, and how often they lifted anything weighing more than 10 kg. Neither of these was related to CRP levels, and thus neither was further considered in the models. To assess if exercise intensity was related to CRP levels, we estimated coefficients for low-, moderate, and vigorous-intensity exercise (MET-hours/week) in the same linear model and ORs in the same logistic regression model. The estimates for each one represents the “effect”of exercise of a given intensity while holding exercise from the other two intensities constant. All analyses were performed using SAS (9.2, SAS Institute Inc, Cary, NC).

There were 537 women with measurements of both recreational exercise and plasma CRP concentrations. The median (IQR) of CRP concentrations during pregnancy was 4.0 (2.6, 7.3) mg/L. There were 119 women (22.2%) with CRP concentrations ≥ 8 mg/L ( Table 1 ). Ninety six percent of the women exercised during pregnancy (512/537). Table 2 shows the mean (SD) and median (IQR) of recreational exercise. Total recreation exercise before and during pregnancy were correlated (Spearman r = 0.6, P < 0.001), however, women reduced their exercise frequency, energy expenditure and intensity of moderate and vigorous exercise after becoming pregnant; for example, women who reported inactive increased from 2 % to 4 %. The mean total recreational exercise was reduced from 10.9 (SD=8.0) to 6.8 (SD=6.1) MET-hours/week. In crude models, total recreational exercise, both before and during pregnancy, was inversely associated with CRP levels when models were fit separately (results not shown). Adjustment for covariates attenuated the magnitude of the associations in both models and only exercise before pregnancy was associated with CRP levels significantly (Model A, Table 3 ). The CRP levels were 1.0% lower (95 % CI: -1.9 %, -0.2 %) with each 1 MET-hour/week of total recreational exercise before pregnancy. Additional adjustment for exercise during pregnancy made exercise before pregnancy lose significance, probably due to the correlation between them (Model B, Table 3 ). The estimate for exercise before pregnancy did not change much with additional adjustment for change in exercise across two periods (Model C, Table 3 ), but appears to be insignificant with adjustment for the average exercise across the two periods (Model D, Table 3 ). The estimate for change in activity in model C was not statistically significant (Adjusted β = -0.003; 95% CI = -0.017, 0.010). Neither total recreational exercise before or during pregnancy were significantly associated with risk of having an elevated CRP level ( Table 3 ). We next investigated whether the association with CRP levels varied according to intensity of exercise before pregnancy ( Table 4 ). Vigorous-intensity exercise showed the strongest association with CRP levels, compared to low- and moderate-intensity exercise. The CRP concentrations was 1.7% lower (95 % CI: -3.2 %, -0.1 %) with each 1 MET-hour/week of vigorous-intensity exercise before pregnancy. Additionally, vigorous-intensity exercise before pregnancy also showed a suggestion of benefit to reduce elevated CRP levels. The adjusted odds of having an elevated CRP level were 9 % lower with each 1 MET-hour/week of vigorous-intensity exercise (OR: 0.91, [95% CI: 0.84, 0.98]). Both estimates did not change much with additional adjustment for change in exercise across the two periods.

Our study found that women who were more active before pregnancy had lower CRP levels during pregnancy than women who were less active. In addition, vigorous-intensity exercise was appeared to be particularly beneficial. As noted above, a systematic inflammatory response occurs in a normal pregnancy 18 . In addition, low-grade subclinical inflammation, reflected by CRP, is associated with oxidative stress 30 and vascular endothelial dysfunction 31 , which are implicated in the pathophysiology of preeclampsia 32 and gestational diabetes 33 . Women with preeclampsia 31 or gestational diabetes 5 demonstrate a heightened inflammatory response. If inflammation mediates the development or progression of these adverse pregnancy outcomes, then exercise may be protective through its anti-inflammatory effect. As noted earlier, the relationship between exercise and risk of preeclampsia has been investigated and the results are suggestive of a protective effect. Studies also found that exercise reduced the risk of gestational diabetes, although some lacked statistical significance. 34 The inverse association of exercise with CRP levels has previously been demonstrated among non-pregnant women and adult men; our study showed that the association also held in pregnant women. In one of the previous studies on physical activity in pregnancy and CRP levels, an accelerometer was used to measure activity in a random sample of 206 U.S. women. The results were similar to those in the present study, though the role of prepregnancy activity was not assessed. 17 The other study with relevant data in pregnant women showed a crude association similar to that found here, but because the focus was on another topic, adjusted estimates were not presented. 16 Our results suggested that compared with light or moderate exercise, vigorous exercise was most strongly related to lower CRP in pregnant women, a finding like that reported by Loprinzi et al. 17 This is consistent with some studies among non-pregnant women that found moderate- to vigorous-intensity exercise was inversely associated with CRP level. 35 - 37 In studies on exercise during pregnancy and preeclampsia and gestational diabetes, the protective effect also seems to be particularly strong for vigorous exercise. 38 , 39 Despite substantial evidence for a relationship between exercise and CRP among non-pregnant adults, an effect of exercise has not been consistently observed in randomized controlled trials. 40 , 41 , 42 , 43 , 44 , 45 Nonetheless, in a meta-analysis of randomized clinical trials that was based on 323 subjects, a 3% reduction in CRP was observed after relatively short-term exercise programs, though this did not reach statistical significance. 45 The subjects in the meta-analysis, and in many of the more recent randomized clinical trials, tended to be older, overweight, and afflicted with comorbidities. A longer period of exercise among younger, healthier subjects might have shown a statistically significant effect, especially if a larger sample size had been employed. Clinical trials among pregnant women have not, as yet, been carried out. The exact mechanism by which exercise might reduce CRP levels is not clear. However, Jung et al. found that exercise reduced levels of TNF-α and IL-6, which stimulate CRP release by hepatocytes. 46 In addition, exercise increases the levels of the anti-inflammatory cytokines IL-1 and IL-10, which diminish CRP production. 46 , 47 Several limitations in this study should be considered. We applied a single MET score from the compendium 24 to all subjects, e.g., we assigned a MET score of 6 to bicycling, but some women may perform it at a much higher intensity. The compendium-established MET did not take pregnancy into account, adding imprecision to the energy expenditure estimates. However, in spite of the possible misclassification, the questions used for exercise assessment in MoBa have been shown to be valid for ranking pregnant women according to recreational exercise level. 25 We were unable to determine if the reduction in CRP levels was due to long-term exercise or more recent exercise. In non-pregnant young women, a study reported that a 16-week aerobic exercise program significantly decreased levels of CRP 48 . Unfortunately, with the relatively small number of subjects, we were unable to evaluate interaction between exercise 3 months before pregnancy and exercise during early pregnancy. Neither were we able to further assess the dose-response effect of vigorous exercise on CRP levels, due to the relatively small number of vigorous exercisers. Because the women in our study were especially active, we had little statistical power to evaluate the effect of being sedentary. Finally, the single measurement of CRP concentration in this study may not precisely reflect inflammation status during the period of early pregnancy. Despite the limitations, our study had some strengths. Our subjects were from a well-defined observational cohort, the sample size was relatively large, and data on recreational exercise were available for both before and during pregnancy. Additionally, we were able to consider and adjust for a large number of potentially confounding covariates in the data analysis. In summary, we identified an association between self-reported recreational exercise before pregnancy and CRP levels among pregnant women. Furthermore, the data suggest that vigorous-intensity exercise may have the strongest effect. Whether exercise before or during pregnancy has the greatest effect on CRP in pregnancy deserves further study, especially given the implications of this issue for the design of clinical trials.