Introduction

Erythema migrans (EM) is the most common manifestation of early Lyme borreliosis (LB), caused by the tick-borne spirochete Borrelia burgdorferi sensu lato (s.l.) [1]. Manifestations of LB usually resolve without sequelae when well-studied, safe and effective antibiotic regimens are used [1, 2]. Still, in a small minority of cases, post-LB symptoms such as malaise, fatigue, neurocognitive dysfunction and diffuse pain, may persist for months to years despite otherwise microbiologically successful antibiotic treatment [1–3]. Adjunctive treatment targeting immune/inflammatory and metabolic responses to Borrelia infection could conceivably improve outcomes in this subset of patients.

Statins are widely used to treat hypercholesterolemia and reduce the risk of cardiovascular and cerebrovascular diseases [4]. They have a generally acceptable safety profile and can be cost-effective [4, 5]. Statins inhibit 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGR), a major regulatory and rate limiting enzyme involved in the mevalonic pathway [8]. In eukaryotes, the mevalonic pathway is involved in biosynthesis of cholesterol and other isoprenoids that are important for cellular composition, interactions, and messaging [6]. In addition to lowering lipids, statins have pleiotropic effects reflecting anti-inflammatory, immunomodulatory, and bactericidal properties [7]. B. burgdorferi s.l. has a functional homolog of the eukaryotic HMGR-I (HMGR-II) [8–10], but biosynthesis of cholesterol through the borrelial mevalonic pathway does not occur and therefore host-derived cholesterol is required for growth [11]. Simvastatin and lovastatin were shown to inhibit borrelial growth under in vitro conditions by inhibiting HMGR [8]. In a murine model of LB, statins were shown to reduce bacterial load and alter the immune response to favor clearance of borreliae [12]. It is not known whether the anti-borrelial effects of statins found in in vitro conditions and in animal models are applicable to humans, or if hyperlipidemia is associated with progression of borrelial infection in humans. Statin pretreatment was recently found to not be associated with clinical manifestations, laboratory test results, and outcomes in patients with Bannwarth’s syndrome [13].

Our study investigated whether statin use is associated with clinical and microbiologic characteristics or long-term outcome in patients treated with antibiotics for early LB manifesting as EM.

Materials and methods

Results

Patients’ characteristics at enrollment

Among the 1520 enrolled patients, 122 (8.0%) were receiving statins at enrollment and continued to do so during follow-up, while 1398 (92.0%) patients were not (Table 1). Patients’ pretreatment characteristics according to statin use are shown in Table 1. The majority (1325, 87.2%) of patients presented with solitary EM while 195 (12.8%) patients had disseminated disease manifesting as multiple EM. There was no significant difference in the frequency of multiple EM in patients taking statins compared to those not on statins (10/122, 8.3% vs. 185/1398, 13.2%; p = 0.146).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clinical and microbiologic characteristics of patients with erythema migrans at enrollment according to statin use.

https://doi.org/10.1371/journal.pone.0261194.t001

Statin-users were older (Table 1) and more likely to report having comorbidities other than hyperlipidemia, such as hypertension, osteoporosis, diabetes mellitus, thyroid disease, cardiac arrhythmias, psychiatric disorders, ischemic heart disease, osteoarthritis, or asthma (107/122, 87.7% vs. 523/1398, 37.4%; p < 0.001). Overall, patients with solitary EM reported LB-associated constitutional symptoms at enrollment less often than patients with multiple EM (361/1325, 27.2% vs. 92/195, 47.2%; p < 0.001). The association between statin use and reported LB-associated constitutional symptoms was not significant. Also, local progression of borrelial infection as assessed by skin erythema size at enrollment was comparable regardless of statin use (Table 1).

Treatment outcome according to statin use

Twenty-seven of 1520 (1.8%) patients missed the 14-day follow up, with increasing losses to follow-up at subsequent visits (Table 2). The median time to resolution of EM after starting antibiotic treatment was comparable in patients taking statins and in those not taking statins (median 7 days, IQR 4–15 vs. median 7 days, IQR 4–15; p = 0.9). The majority (≥76.5%) of patients showed complete recovery from 2 months onward, returning to their pre-LB health status. The proportion of patients with incomplete recovery, represented predominantly by the presence of post-LB symptoms, steadily decreased during follow-up regardless of statin use, except at the 2-month visit, when differences between the two groups reached significance (Table 2). At 12 months, 76/1183 (6.4%) patients showed incomplete recovery. When the outcome was evaluated at the last evaluable visit, the proportion of patients with incomplete recovery was similarly low (113/1516, 7.5%). In the multiple logistic regression model, the odds for incomplete recovery were higher for women, for patients with multiple EM, and in those reporting LB-associated constitutional symptoms at enrollment. (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Number (%) of patients with erythema migrans who had incomplete recovery at follow-up visits according to using or not using statins.

https://doi.org/10.1371/journal.pone.0261194.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between clinical characteristics at enrollment and incomplete recovery.

https://doi.org/10.1371/journal.pone.0261194.t003

Discussion

In our study of European patients with EM, predominantly infected with B. afzelii, statin use was not associated with local progression, dissemination of infection or disease severity as assessed by the size of skin EM, frequency of multiple EM, or presence of LB-associated symptoms, nor with serologic response and borreliae culture positivity, or long-term outcome as assessed by the rate of incomplete recovery. These findings do not support the hypothesis that statins are beneficial in this clinical setting.

Some observational studies suggest that statins may be beneficial in various infections, while the results of a meta-analysis of randomized placebo-controlled trials showed no obvious beneficial effect of statins on risk of infection or infection-related mortality [25]. We are not aware of any clinical study of the effect of statin use on the progression and outcome of early LB, manifesting as EM. However, it was recently reported that statin use is not associated with clinical manifestations, laboratory test results, and outcome in patients with Bannwarth’s syndrome [13], which is consistent with our results. Experimental studies, on the other hand, have shown that simvastatin and lovastatin have an inhibitory effect on borrelial growth in vitro [8] as well as promote clearance of borreliae in a murine model of LB [12].

Statins may affect pathogenesis of LB through complex lipid interactions and exchanges between spirochete and host cells involving antigenic borrelial outer membrane cholesterol-glycolipids and other components. These processes may alter borrelial immunogenicity and therefore host immune responses [11].

There is a generally pro-inflammatory cytokine profile in acute LB in murine models and in humans, thought to be important for disease clearance [12, 26]. Persistence of symptoms after treatment of EM may be associated with less intense pre-treatment pro-inflammatory cytokine responses in the skin [27]. However, down-regulation of an initially robust immune response at the proper time appears to be important for optimal resolution of infection [28]. In a murine model, lovastatin, but not simvastatin, was associated with decreased total IgM and IgG and modified cytokine responses to borrelial infection reflecting both T_H1 and T_H2 immune postures [12]. These responses may have contributed to lovastatin-associated borrelial clearance in that model.

The differential pathophysiology that might explain the divergence of results using statins in in vitro, murine model and observational human studies, such as ours, is certainly highly complex. Up regulation of borrelial enzymes, including HMGR-II, under certain conditions (e.g., changes in temperature, pH and increases in extracellular acetate) and some decrease of in vitro anti-borrelial effect of simvastatin and lovastatin in experimental studies with HMGR overexpression [8], may play a role among other factors. The binding affinity of statins on HMGR-I present in humans and mice is up to almost 10⁴ fold higher than for HMGR-II found in prokaryotes, including B. burgdorferi [29], suggesting a considerable difference in statin inhibitory effect between host and pathogen [10, 30]. Given this difference, it is unclear that the beneficial effects of statins seen in experimental animal models are primarily mediated by their inhibitory effect on borrelial HMGR-II enzymes. Since B. burgdorferi is unable to synthesize cholesterol and require cholesterol from their environment [11], it is possible that a statin-associated decrease in host serum cholesterol concentration through HMGR-I inhibition may play a role in attenuating LB infection in murine models. Basal cholesterol levels are lower in mice than in humans [31, 32], among many other metabolic and immunologic differences. In one experimental model, no information was provided on the concentrations of statins and cholesterol in the blood of treated mice [12], nor did we have this information in our study subjects. It is interesting that in another murine model, significant hyperlipidemia associated with certain genetic mutations, such as those affecting cholesterol transport, can lead to more severe presentations of LB [32]. It is not known if a specific level of host cholesterol might be associated with inhibition of borrelial growth; however, the statin doses required to decrease host cholesterol levels such that borrelial growth would be inhibited may be toxic to humans. In vitro and murine studies suggest that supra-therapeutic and potentially toxic levels of statins are required to produce antibacterial effects in certain bacterial strains, limiting the clinical value of these findings [12, 30, 33–35].

Some of the statins used in our study may have been more or less effective in inhibiting borrelial growth than simvastatin and lovastatin used in animal studies of infection with B. burgdorferi. Only 26.6% of patients in our study were taking one of these two statins. Different statins have been shown to have variable anti-bacterial effects depending on the pathogen studied [34].

In our study, patients taking statins were older and more frequently reported other comorbidities, besides hyperlipidemia. This is consistent with the known increase in the burden of comorbidities with age [36]. Although one study did not find an association between higher age and outcomes in patients with EM [37], it was performed in an American population infected with B. burgdorferi and the stratification by age differed from ours, limiting its applicability to our study. A previous European study showed that the presence of LB-associated symptoms at enrollment was the strongest predictor of incomplete recovery in patients with EM, but also that the probability of incomplete recovery increased in older age groups [24]. That study did not find a significant association between comorbidities and clinical outcome suggesting a possible decoupling of age and comorbidities with respect to LB outcome. Furthermore, the use of statins was not detailed, nor was the concentration of lipids measured. In the present study, older patients did not have statistically significant higher odds for incomplete recovery. Although we adjusted the analysis accordingly in the multiple logistic regression model, the observational nature of our study may have introduced a bias that could have masked a positive statin effect in older patients who might otherwise have fared worse. A randomized controlled trial could avoid this pitfall and explore this possibility, but it is unclear how such a trial could be constructed given the semi-random nature of borrelial infection and the probable need for a long duration of prior statin treatment to obtain benefits [38, 39]. Indeed, a study that examined the benefits of de novo treatment of infection with statins did not show favorable results [39].

Several other limitations should be considered when interpreting our findings. First, we did not measure serum concentrations of statins, lipids or inflammatory mediators in patients. This might have allowed for a more specific assessment of the association between statin use and borrelial infection in humans. Second, we did not adjust the analysis for the type and dose of statin used. The number of patients receiving a specific dose of a given statin in our study was too small to allow statistical analysis of the association between the type and dose of statin used and selected B. afzelii infection characteristics. This is potentially significant, as the greatest benefit may be obtained by matching specific statins and doses with specific infecting agents [34, 40, 41]. Third, patients with multiple EM were over-represented in the analysis, but we recognized this potential bias and adjusted the analysis accordingly. Fourth, our results may relate only to geographic areas where LB is predominantly caused by B. afzelii. Fifth, we did not investigate the potential role of statin therapy in various other clinical manifestations of LB. The immunologic, inflammatory, and metabolic mechanisms expressed in other stages and manifestations of LB may be differentially sensitive to statin therapy.

Despite these limitations, we believe that a particular strength of our study was the use of prospectively collected data by infectious disease specialists at a single center with decades of clinical experience and systematic follow-up. Data analysis using a large number of patients with well-characterized manifestations of LB allowed a meaningful comparison of clinical and microbiologic characteristics according to statin use.

In conclusion, statin use was not significantly associated with disease severity, Borrelia culture positivity, or long-term outcome in patients with early LB, manifesting as EM and predominantly caused by B. afzelii. Future adequately powered and more controlled studies would be needed to clarify whether different doses and types of statins have therapeutic benefit in patients infected with specific Borrelia species who present with various manifestations of LB.