Course and Outcome of Erythema Migrans in Pregnant Women

Information on Lyme borreliosis (LB) during pregnancy is limited. In the present study, the course and outcome of erythema migrans (EM) in 304 pregnant women, diagnosed in the period 1990–2015, was assessed and compared with that in age-matched non-pregnant women. The frequency of unfavorable outcome of pregnancies was also evaluated. The pregnant women reported constitutional symptoms less frequently than the non-pregnant women (22.4% vs. 37.2%, p < 0.001). Pregnant women diagnosed with EM later during pregnancy had a lower probability of reporting constitutional symptoms (odds ratio = 0.97 for 1-week difference in gestation week at diagnosis of EM, 95% CI: 0.94–0.99, p = 0.02). The outcome of pregnancy was unfavorable in 42/304 (13.8%) patients: preterm birth in 22/42 (52.4%), fetal/perinatal death in 10/42 (23.8%), and/or anomalies in 15/42 (35.7%). Several patients had potential explanation(s) for the unfavorable outcome. In conclusion, the course of early LB during pregnancy is milder than in age-matched non-pregnant women. The outcome of pregnancy with the treatment approach used in the present study (i.v. ceftriaxone 2 g once daily for 14 days) is favorable.


Introduction
Lyme borreliosis (LB) usually presents as the skin lesion erythema migrans (EM). The lesion, which is the result of tick bite inoculation of Borrelia burgdorferi sensu lato (s.l.) into the skin, develops early in the course of the disease. The causative agent can disseminate in some patients, resulting in secondary skin lesions and involvement of the nervous system, joints, heart, and/or eye [1].
Information on LB during pregnancy is limited. According to general belief, there are no differences in the course of the disease in pregnant and non-pregnant women. However, a PubMed literature search has found no straightforward data on the course of the infection in pregnant women, and information on the outcome of their pregnancies is limited .
The aim of our study was to evaluate and compare the course and outcome of early LB in non-pregnant and pregnant women and to assess the outcomes of the pregnancies.

Laboratory and Microbiological Evaluation
Basic laboratory tests were performed at the first visit and two weeks later. Serum antibodies to B. burgdorferi s.l. were determined at baseline and at follow-up examinations using either an indirect immunofluorescent assay with a local isolate of B. afzelii as antigen [25] or an indirect chemiluminescence immunoassay (LIAISON ® ) with antigens OspC and VlsE for the detection of IgM antibodies and VlsE for IgG antibodies.
As of 1994, a sample of citrated blood (5 mL until 2000, 9 mL subsequently) was taken at presentation for cultivation of borreliae in modified Kelly Pettenkofer medium [26]. Isolates were identified to species level using pulsed-field gel electrophoresis after MluI restriction of genomic DNA or by PCR-based restriction fragment length polymorphism of the intergenic region [27,28].

Treatment
At the first visit, patients received either i.v. ceftriaxone 2 g once daily, i.v. penicillin G 10,000,000 units twice daily, or oral phenoxymethylpenicillin 1 g three times per day. The duration of antibiotic treatment was 14 days.

Assessment of Outcome
The course and outcome of early LB in the pregnant women and the frequency of unfavorable outcome of their pregnancies (appraised by fetal death, pre-term birth, offspring malformations) were evaluated. The medical history was taken at each visit, including information on the presence of subjective symptoms, and patients were examined clinically. A gynecologist regularly monitored the course of gestation. At the first visit after delivery, detailed information about the birth and the infant was collected. A pediatrician monitored the babies at birth and after 6 months; however, several babies had more frequent and/or longer follow-ups.

Control Group
The course and outcome of EM in pregnant women was compared with that in age-matched non-pregnant woman diagnosed with EM at our institution in the same year.

Statistical Analysis
The characteristics of the pregnant women and the control group were compared using the Mann-Whitney test for numeric covariates and Fisher's exact test for categorical covariates. Categorical variables were summarized with frequencies and percentages and 95% confidence intervals (CI), numeric variables with medians and interquartile ranges. To control for false positives, the p values shown in Table 1 were adjusted using a multivariate permutation procedure [29]. The association between gestation week at diagnosis of EM and the presence of constitutional symptoms was investigated using a logistic regression model. A possible departure from the assumption of a linear association between gestation week at which EM was diagnosed and the logit of probability of reporting the symptoms was assessed by fitting an additional model that included a restricted cubic spline transformation (with five knots) of the covariate.
The risk of an adverse outcome of pregnancy varies with week of gestation, and this was used as the timescale. We compared the risk for women in the same gestation week with respect to gestation week at diagnosis of EM. Four additional preselected possible confounders (age, duration of EM at diagnosis in weeks, multiple EM or borreliae isolated from blood, presence of constitutional symptoms) were included in univariable and multivariable Cox regression models.
To account for a possible selection bias of the pregnant patients, 39 patients enrolled in the first eight weeks of pregnancy were included in the risk set only after week 9 of their pregnancy.
R statistical language was used for all the analyses [30].

Ethical Considerations
The study was performed according to the Declaration of Helsinki and was approved by the Medical Ethics Committee of the Ministry of Health of the Republic of Slovenia (35/04/09). The Ethics Committee waived the need for written informed consent.

Results
Among 14,010 patients diagnosed with EM at our institution during a 26-year period, 307 were pregnant, and, for 304 (99%), the natural outcome of the pregnancy was available; 105 of these 304 patients have been reported previously [18,20,23].

Pre-Treatment Characteristics
In the comparison of basic clinical characteristics of pregnant and non-pregnant women with EM, findings were analogous for the majority of parameters, with the exceptions that the pregnant women less often had a ring-like EM (42.4% vs. 55.3%, p = 0.002), less often had EM located on the trunk (14.1% vs. 24.0%, p = 0.009), and less often reported constitutional symptoms (22.4% vs. 37.2%, p < 0.001). When the analyses were adjusted for multiple comparisons, the differences remained significant for ring-like EM and the presence of constitutional symptoms (Table 1). Patients who were diagnosed in the later stages of pregnancy had a lower probability of reporting constitutional symptoms (OR = 0.97 for 1-week difference in gestation week at diagnosis of EM, 95% CI: 0.94-0.99, p = 0.02). Our data do not prove a non-linear association between gestation week and the logit of probability (p = 0.20). When including the non-linear term, the estimated prevalence of constitutional symptoms decreased very markedly only in the first 20 weeks of pregnancy but afterwards remained low ( Figure 1). In both groups of women, the proportion of patients with borrelial IgM and/or IgG serum antibodies at presentation (63/295, 21.4% vs. 72/289, 24.9%), the isolation rate of borreliae from blood (8/216, 3.7% vs. 4/187, 2.1%), and the ratio between the isolated Borrelia species (six B. afzelii and two B. garinii vs. three B. afzelii and one B. garinii) were similar.

Outcome of Pregnancies
The outcome of pregnancy was unfavorable in 42/304 (13.8%) patients: preterm birth in 22/42 (52.4%), fetal/perinatal death in 10/42 (23.8%), and/or anomalies detected at birth or within the first year of life in 15/42 (35.7%). Several patients had a potential explanation for the unfavorable outcome (Table 2). At a significance level of alpha = 0.01, none of the chosen factors were associated with a higher risk of unfavorable outcome (Table 3).

Discussion
Several mechanical and pathophysiologic changes occur during pregnancy, and immune adaptations develop to accommodate the fetus [31]. As pregnancy progresses, hormone levels (estradiol, progesterone) increase markedly in association with several immunologic changes, including a shift from Th1 to Th2 immunity, which results in decreasing robustness of cell mediated immunity [32]. Consequently, the acquisition, clinical presentation, and course of infectious diseases in pregnant women may be altered [33]. Although clinical findings indicate that the course of most infections in pregnant and non-pregnant women are similar, some diseases (for example, listeriosis and malaria) are more frequent during gestation, and several (such as influenza, hepatitis E, herpes simplex virus infections, malaria) are more severe in pregnant women [31]. While it is well known that some non-infectious diseases, such as multiple sclerosis, have a milder course during pregnancy [34], no infectious disease has been reported to have a less severe course during gestation. There is a concept that pregnancy does not affect the course and outcome of early LB; however, studies that directly compare the course of LB in pregnant and in non-pregnant women are lacking.
Our study has shown that the majority of basic clinical and epidemiologic characteristics of EM before treatment with antibiotics were analogous in the two groups of women and consonant with previous findings in Slovenian patients with EM [35][36][37][38][39][40][41][42] and that the outcome after antibiotic treatment was excellent regardless of pregnancy. No subsequent objective manifestations of LB were established in either of the two groups, and the proportion of patients with symptoms at follow-up visits was even lower than found in other recent studies from Slovenia [37][38][39][40][41][42], possibly because only young, previously healthy patients were included in the present study.
Nevertheless, there were also several differences. We do not have a reliable explanation for the observation that the pregnant women less often had ring-like EM despite similar duration of the skin lesion before treatment, but we stress that the findings in our control group are in agreement with previous reports in Slovenian patients with EM [35][36][37][38][39][40][41][42]. Furthermore, the proportion of reported constitutional symptoms accompanying EM was lower in the pregnant women, indicating that the course of EM during pregnancy was milder than in the age-matched non-pregnant women ( Table 1, Figure 1), as also shown in previous reports on EM from the same region [35][36][37][38][39][40][41][42]. Our results indicate that the probability of reporting constitutional symptoms systematically decreases with gestation week at diagnosis of EM (EM was diagnosed a median 7 days after the appearance of the skin lesion) and that women infected during the later stages of pregnancy report fewer constitutional symptoms compared with those infected during the early phases of pregnancy, who are more similar to non-pregnant women.
Since B. burgdorferi s.l. does not produce its own toxins or extracellular matrix-degrading proteases, most manifestations of LB result from inflammation generated by the host immune response to the spirochete [43]. Thus, fewer symptoms, as found in the present study of pregnant women with EM, may be associated with lower levels of inflammation. This assumption parallels findings in animal models where, for example, Lyme arthritis in pregnant mice was less severe. The amelioration of arthritis was associated with a shift in inflammatory responses-that is, the down-regulation of Th1 responses, most likely via the progesterone-mediated up-regulation of Th2 cytokine production, resulting in the reduction in pathogenic inflammatory responses during gestation in the mice [44]. In addition, our previous work has shown that higher numbers of constitutional symptoms in EM patients are associated with greater Th1 or Th17-associated cytokine responses, which is consistent with findings that immune responses shift towards a Th2 response in the course of pregnancy [45].
Information on the outcomes of pregnancy in women who develop LB during gestation is limited [46]. Findings of a PubMed literature search for the period of 1985 to January 2020 are shown in Table 4. The search found several individual case reports and a few series on unfavorable outcome of pregnancy in patients who were treated or not treated with antibiotics, the large majority of studies being from the two decades after recognition of LB . The described cases show no uniform pattern of abnormalities. In the majority of cases, only a temporal association with maternal LB was described but no causal relationship was confirmed or searched for, and, in some articles, the proof of borrelial infection was imprecise by present standards. In addition, in patients in whom spirochetes were identified microscopically or by culture of placenta or autopsied infants, no signs of inflammation, granuloma formation, or necrosis in the affected tissues were detected. Moreover, Mather et al. reported the absence of the transplacental transmission of Lyme disease spirochetes from reservoir mice to their offspring [47], though the human placenta could be different to that of the mouse. Thus, the association between maternal borrelial infection and unfavorable outcome of pregnancy remains unclear, and the risk of adverse outcomes from maternal Lyme borreliosis has been interpreted to be negligible.
In the present study, 42/304 (13.8%) patients had an unfavorable outcome of pregnancy. However, several of these patients had a potential explanation for the unfavorable outcome (Table 2), and none of the tested parameters were associated with unfavorable pregnancy outcome (Table 3). Although our multivariable analyses showed an association between the week of pregnancy in which EM was diagnosed and unfavorable outcome, suggesting that patients infected earlier during pregnancy might have a higher risk of such an outcome, the diminishment of the odds of unfavorable outcome with the duration of pregnancy is an expected finding that is valid for the overall population of pregnant women. Thus, unfavorable outcome of pregnancy in women who had LB during pregnancy does not in any way signify that borrelia infection was the cause of the adverse result. Furthermore, the frequency of unfavorable outcome of pregnancy as found in the present study is comparable to the findings described in the literature for the general population, with the exception that the miscarriage rate in our study (11.6%) is somewhat lower than in the majority of previous reports (15-20%) [48][49][50][51] yet still in agreement with a Swedish study [51]. The findings of the present study are also in accord with the pregnancy outcomes reported for the general Slovenian population in the corresponding time frame [52], including miscarriage rate (11.6% in the present study versus 12% in the general population), preterm birth (7.2% versus 6.1%), and anomalies detected at birth (2.6% versus 1.7%). However, we would like to stress that we did not perform a cohort study, where all women enter the study at the beginning of their pregnancy, and therefore we did not attempt to analyze the approximate prevalence of unfavorable outcomes by trimester or to compare them with values observed in the overall population of pregnant women. In our study, pregnant women were enrolled at the gestation week when they were diagnosed with EM and left the study at delivery or at the gestation week of an unfavorable event. Consequently, women experiencing abortion or early delivery were less likely to be included in the sample of pregnant women diagnosed with EM compared with those who did not experience such unfavorable events.     All but five pregnant women were treated for their EM with i.v. ceftriaxone for 14 days. According to current knowledge, this is clearly overtreatment of EM in non-pregnant and possibly also in pregnant women. However, 30 years ago, the understanding of LB was somewhat rudimentary, and information on the course and outcome of pregnancy in patients with early LB was limited to individual case reports, several of them indicating unfavorable outcomes after treatment of EM with oral antibiotics [3,10]. At that time, we decided on a treatment protocol that would achieve high enough levels of antibiotics not only in skin but also in the placenta and fetus; the decision being based on the premise that damage to the fetus probably results from the direct dissemination of borreliae or indirectly through damage to the placenta. In the years since, a concept has been developed in which the outcome of pregnancy with this treatment approach is similar to that in the general population [20]. However, because cases of LB during gestation are not numerous, we did not alter our approach and have been waiting to see whether studies from other research groups would confirm that the same outcome would follow oral antibiotic treatment, as recommended for EM in the non-pregnant population. A report published in 2010 (the most recent available information) states that the proportion of unfavorable pregnancy outcomes in patients with LB was the highest in patients who received no antibiotic treatment for their LB, followed by those who received oral antibiotics, and it was the lowest in patients treated with parenteral antibiotics [22]. The report, although it has several methodologic drawbacks, has influenced us to not change our treatment approach. The consequence of this "wait and watch" tactic is that we know that the outcome of pregnancies is relatively favorable using our treatment protocol, but we do not know whether the same results could be obtained with oral treatment, as is usually recommended for EM. Furthermore, since we did not demonstrate the direct detection of borreliae in fetal tissue or umbilical blood, etc., which is a substantial limitation of the present study, we do not know whether a relatively favorable outcome of pregnancy is the result of our efficacious antibiotic treatment or a consequence of very rare or perhaps even non-existent borrelial involvement in the offspring.

Conclusions
The course of early LB during pregnancy is milder than in age-matched non-pregnant women. The smaller proportion of pregnant patients reporting constitutional symptoms at the time of EM diagnosis might be the result of immunologic changes during gestation.
The outcome of pregnancy with the treatment approach used in the present study (i.v. ceftriaxone 2 g once daily for 14 days) is favorable. Multivariable analyses showed that patients who develop EM in the early stages of pregnancy might have a higher risk of unfavorable outcome. Funding: This work was supported by the Slovenian Research Agency, grant number P3-0296. The funding source had no role in the study design, data collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.