Oral Doxycycline Compared to Intravenous Ceftriaxone in the Treatment of Lyme Neuroborreliosis: A Multicenter, Equivalence, Randomized, Open-label Trial

Abstract

Background

Lyme neuroborreliosis (LNB) is often treated with intravenous ceftriaxone even if doxycycline is suggested to be noninferior to ceftriaxone. We evaluated the efficacy of oral doxycycline in comparison to ceftriaxone in the treatment of LNB.

Methods

Patients with neurological symptoms suggestive of LNB without other obvious reasons were recruited. The inclusion criteria were (1) production of Borrelia burgdorferi–specific antibodies in cerebrospinal fluid (CSF) or serum; (2) B. burgdorferi DNA in the CSF; or (3) an erythema migrans during the past 3 months. Participants were randomized in a 1:1 ratio to receive either oral doxycycline 100 mg twice daily for 4 weeks, or intravenous ceftriaxone 2 g daily for 3 weeks. The participants described their subjective condition with a visual analogue scale (VAS) from 0 to 10 (0 = normal; 10 = worst) before the treatment, and 4 and 12 months after the treatment. The primary outcome was the change in the VAS score at 12 months.

Results

Between 14 September 2012 and 28 December 2017, 210 adults with suspected LNB were assigned to receive doxycycline (n = 104) or ceftriaxone (n = 106). The per-protocol analysis comprised 82 patients with doxycycline and 84 patients with ceftriaxone. The mean change in the VAS score was −3.9 in the doxycycline group and −3.8 in the ceftriaxone group (mean difference, 0.17 [95% confidence interval, −.59 to .92], which is within the prespecified equivalence margins of −1 to 1 units). Participants in both groups improved equally.

Conclusions

Oral doxycycline is equally effective as intravenous ceftriaxone in the treatment of LNB.

Clinical Trials Registration

NCT01635530 and EudraCT 2012-000313-37.

Lyme neuroborreliosis, Borrelia burgdorferi, ceftriaxone, doxycycline, cerebrospinal fluid

Lyme borreliosis (LB) is the most common tick-borne infection in Europe and in the United States [1, 2]. It is caused by spirochetes of the Borrelia burgdorferi sensu lato complex (later: B. burgdorferi). In Finland, the overall incidence of LB was 118 per 100 000 population in 2014 [3]. Nervous system involvement is diagnosed in 3%–12% of the patients with LB [4]. Typical presentations of Lyme neuroborreliosis (LNB) are facial nerve palsy, meningoradiculitis, and Garin-Bujadoux-Bannwarth syndrome (painful radiculopathy with lymphocytic meningitis and peripheral neuropathy) [5–7]. LNB is often treated with intravenous ceftriaxone, although oral doxycycline has been suggested to be noninferior to ceftriaxone in previous trials [8, 9]. The aim of this study was to further confirm that oral doxycycline is equally effective as intravenous ceftriaxone in the treatment of LNB.

METHODS

Study Design and Participants

In this multicenter, open-label, randomized study, we recruited 210 patients from 2 hospitals in southern Finland, Turku University Hospital and Helsinki University Hospital.

Eligible participants were adults with neurological symptoms suggestive of LNB without other obvious diagnoses. The inclusion criteria for definite LNB were cerebrospinal fluid (CSF) pleocytosis (≥5 leukocytes/μL) and intrathecal production of B. burgdorferi–specific antibodies, or detection of B. burgdorferi DNA in the CSF. The inclusion criteria for possible LNB were B. burgdorferi–specific antibodies in serum, or an erythema migrans during the previous 3 months. The complete inclusion and exclusion criteria are shown in Supplementary Table 1.

Written informed consent was obtained from all study participants. Ethical approval was given by the National Committee on Medical Research Ethics. This trial is registered with ClinicalTrials.gov (NCT01635530) and EudraCT (2012-000313-37).

Randomization

Patients were randomly assigned in a 1:1 ratio to receive either doxycycline or ceftriaxone. The physician recruiting the patient contacted the secretary’s office of the Infectious Diseases Department of Turku University Hospital, where a voucher was blindly drawn assigning the patient to one of the groups. No stratification factors were used.

Procedures

Doxycycline (orally, 100 mg twice a day) was administered for 4 weeks, while the length of ceftriaxone treatment (intravenously, 2 g daily) was 3 weeks. Doxycycline was self-administered by the participants and ceftriaxone was usually administered as outpatient parental therapy.

The patients were evaluated by an infectious disease specialist or a neurologist at the beginning of the treatment, and at 4 and 12 months after the initiation of the treatment. The patients defined their subjective condition by a visual analogue scale (VAS) on a scale of 0–10 (0 = normal, 10 = worst) at every visit. CSF specimens were drawn at the first visit for the analysis of intrathecal B. burgdorferi antibody production, leukocyte count, and protein, lactate, and CXCL13 concentration. The analyses were repeated after 3 weeks if a patient had ≥50 leukocytes/μL in the first CSF specimen. If the CSF cell count had not decreased by ≥80% from the initial value, the treatment with doxycycline was extended to 8 weeks, and the treatment with ceftriaxone to 4 weeks. The patients without CSF collection at 3 weeks were contacted by phone to evaluate the condition of the patient and potential side effects of the therapy.

If the patient developed a new objective symptom (such as a new facial nerve palsy) during the treatment, the patients in the doxycycline group were treated with ceftriaxone for 3 weeks in parallel with doxycycline, and the patients in the ceftriaxone group were treated in parallel with doxycycline for 4 weeks. These patients with a treatment modification were not included in the per-protocol analysis.

Methods for B. burgdorferi serology, B. burgdorferi polymerase chain reaction (PCR), and CXCL13 analyses are described in page 3 of the Supplementary Materials.

Outcomes

The primary outcome was the subjective improvement of the patient, evaluated as a change in VAS score at 12 months after the treatment. The change in VAS score was defined as the difference between the VAS score at the beginning of the treatment and at 12 months. The additional primary outcome was VAS score at 12 months. The secondary outcomes included the difference between ceftriaxone and doxycycline groups in leukocyte count, protein, lactate, and CXCL13 concentrations in the CSF specimens collected after 3 weeks of treatment.

Statistical Analysis

According to null hypothesis, patients with doxycycline improved equally well as patients with ceftriaxone treatment. It was determined that the mean change of the VAS score should not differ by > 10% (1 unit) in order to be regarded equal in both groups. The equivalence margin was based on a consensus among specialists in infectious diseases and neurology. Applied statistical methods are described in page 4 of the Supplementary Materials.

RESULTS

Patient Characteristics

Between 14 September 2012 and 28 December 2017, approximately 300 adults with suspicion of LNB were screened for eligibility. Of these patients, 210 were recruited and randomly allocated to receive either doxycycline (n = 104) or ceftriaxone (n = 106). After final evaluation, 23 participants were excluded from the modified intention-to-treat (mITT) population, because LNB was considered an unlikely diagnosis (n = 11), or because the participant received an alternative diagnosis (n = 12). During antibiotic treatment and follow-up, there were 21 dropouts. In the per-protocol analysis, 82 (79%) participants from the doxycycline group and 84 (79%) from the ceftriaxone group were included (Figure 1).

Figure 1.

Trial profile. ^aAlternative diagnoses were polymyalgia rheumatica, seronegative rheumatoid arthritis, viral meningitis, 2 cases of tick-borne encephalitis, 2 cases of multiple sclerosis, mononucleosis, cytomegaloviremia, chronic headache, osteoarthritis, and mechanical cervical radiculopathy. ^bPatients whose symptoms evaluated with laboratory findings did not refer to LNB, but in whom no alternative diagnosis was discovered, are described as “LNB unlikely.” ^cTreatment modification means that the patient developed a new objective symptom typical or possible for LNB during first-line treatment (3 patients with new facial nerve palsy and 1 patient in the doxycycline group with urinary bladder dysfunction) or the symptoms intensified significantly after first-line treatment and a new antibiotic course to LNB was given during follow-up (1 patient in the ceftriaxone group). Abbreviations: LNB, Lyme neuroborreliosis; mITT, modified intention-to-treat.

Definite LNB was diagnosed in 99 patients, and possible LNB in 88 patients. Three patients were classified as definite LNB based on positive CSF B. burgdorferi PCR. Other 96 patients with definite LNB had pleocytosis and intrathecal production of B. burgdorferi–specific antibodies.

There were no substantial baseline differences between the study groups (Table 1). When baseline demographics were compared between excluded patients/dropouts (n = 44) and per-protocol population (n = 166), the excluded/dropouts were younger than per-protocol population (mean, 52 vs 57; P = .047), whereas no other statistically significant differences were observed (Supplementary Table 4). Signs and symptoms of the patients at recruitment are presented in Table 2. Baseline data, coexisting diseases and the symptoms of the patients divided into subgroups according to certainty of LNB are shown in Supplementary Tables 2, 3, and 5.

Table 1.

Baseline Demographics and Clinical Characteristics of the Modified Intention-to-Treat Population

Characteristic	Doxycycline (n = 94)	Ceftriaxone (n = 93)
Female sex, no. (%)	37 (39)	30 (32)
Age, y, mean (SD)	54 (15)	59 (14)
No. of coexisting diseases, median (IQR)	1 (0–3)	1 (0–3)
Recalled tick bites during the past year, no. (%)
One tick bite	16 (17)	18 (19)
Several tick bites	19 (20)	19 (20)
EM during the past year, no. (%)
Solitary	22 (23)	27 (29)
Multiple	3 (3)	3 (3)
No. of symptoms, median (IQR)	4 (3–6)	4 (3–5.5)
Duration of symptoms, wk, median (IQR)	4 (3.0–10.3)	6 (3.5–11.0)
Symptoms relieved partly before recruitment to the study, no. (%)	26 (28)	29 (31)
Possible late LNB with symptoms for >6 mo, no. (%)	9 (10)	8 (9)
Certainty of LNB, no. (%)
Definite LNB	51 (54)	48 (52)
Possible LNB	43 (46)	45 (48)
Glucocorticoid therapya for facial nerve palsy, no. (%)	31 (33)	35 (38)
VAS at the beginning, mean (SD)	5.0 (2.4)	5.1 (2.3)

Abbreviations: EM, erythema migrans; IQR, interquartile range; LNB, Lyme neuroborreliosis; mITT, modified intention-to-treat; SD, standard deviation; VAS, visual analogue scale.

^aTen-day course of glucocorticoid (prednisolone) 60 mg daily for 5 days, then decreasing the dose 10 mg daily for 5 days.

Table 2.

Objective Findings and Reported Symptoms in the Modified Intention-to-Treat Population at Recruitment

Finding or Symptom	Doxycycline (n = 94)	Ceftriaxone (n = 93)
Radiculitis	43 (46)	46 (49)
Facial nerve palsy	42 (45)	44 (47)
Garin-Bujadoux-Bannwarth syndromea	19 (20)	19 (20)
Neuritis	2 (2)	3 (3)
Peripheral nerve palsyb	5 (5)	1 (1)
Recurrens nerve palsy	0	2 (2)
Abducens nerve palsy	1 (1)	0
Diplopiac	6 (6)	4 (4)
Difficulty in swallowing	1 (1)	2 (2)
Hypacusis	6 (6)	4 (4)
Tinnitus	4 (4)	2 (2)
Altered sense of taste	7 (7)	2 (2)
Polyneuropathy	2 (2)	2 (2)
Guillain-Barré syndrome	1 (1)	0
Convulsions	0	1 (1)
Peripheral muscle weaknessd	11 (12)	8 (9)
Memory disturbance	6 (6)	7 (8)
Cognition disturbance	5 (5)	6 (6)
Tremor	4 (4)	4 (4)
Nausea	10 (11)	10 (11)
Vertigo	17 (18)	20 (22)
Lymphocytoma	1 (1)	2 (2)
Monoarthritis	0	1 (1)
Polyarthritis	1 (1)	0
Iritis	1 (1)	0
Headache	47 (50)	43 (46)
Neck-shoulder pain, unspecified	30 (32)	33 (35)
Neck stiffness	11 (12)	9 (10)
Myalgia	27 (29)	26 (28)
Arthralgia	26 (28)	15 (16)
Fatigue	26 (28)	37 (40)
Fever >38°C	13 (14)	10 (11)
Flulike symptoms	7 (7)	4 (4)
Decline in general condition	11 (12)	9 (10)
Weight loss >3 kge	3 (3)	5 (5)

Data are presented as no. (%).

^aGarin-Bujadoux-Bannwarth syndrome includes lymphocytic meningitis, radiculitis, and peripheral nerve palsy. The patients with Garin-Bujadoux-Bannwarth syndrome are also included in the rows of radiculitis (n = 38), facial nerve palsy (n = 37), and abducens nerve palsy with another peripheral nerve palsy (n = 1).

^bCranial nerve palsies and other peripheral nerve palsies are enrolled separately.

^cSymptom was transient, and no cranial nerve palsy was detected in clinical examination.

^dNonspecific muscle weakness with no findings in electroneuromyography.

^eMedian time was 7 weeks (interquartile range, 3.6–11.0 weeks).

Primary Outcome

In the per-protocol analysis, the mean change in the VAS score among patients in the doxycycline group was −3.9 (standard deviation [SD], 2.5), and −3.8 (SD, 2.4) in the ceftriaxone group. The mean difference between the groups was 0.17. With a 95% confidence interval (CI) difference (−.59 to .92), the groups improved equally (the 95% CI difference is within the predefined equivalence margins of −1 to 1). In the intention-to-treat (ITT) population, 31 of 210 VAS scores at 12 months were missing (19 patients with doxycycline and 12 patients with ceftriaxone). In line with the results of the per-protocol group, the mean change in the VAS score in the ITT population with doxycycline was −3.9 (SD, 2.5) and −3.8 (SD, 2.5) with ceftriaxone. The mean difference between the groups was 0.07. With a 95% CI difference (−.67 to .82), the doxycycline and ceftriaxone groups also in the ITT population improved equally. The change in the VAS scores is shown in Figure 2.

Figure 2.

Change in visual analogue scale (VAS) score during follow-up of the per-protocol population with doxycycline or ceftriaxone. The box represents the interquartile range (IQR), which contains the middle 50% of the records. The line across the box indicates the median. The whiskers show the minimum and maximum values not greater than 1.5 times the IQR. Outliers (o) are cases between 1.5 and 3 times the IQR, and extremes (*) are cases with values >3 times the IQR.

In the per-protocol population, the median VAS score at 12 months did not differ between the doxycycline and ceftriaxone groups, being 0 (interquartile range [IQR], 0–1.8) and 1 (IQR, 0–2.0), respectively (P = .343). In the ITT population, the median VAS score in patients with doxycycline was 0 (IQR, 0–1.9) and with ceftriaxone 1 (IQR, 0–2.1), with no statistically significant difference between the groups (P = .160). The VAS scores at initiation, at 4 months, at 12 months, and the change in the VAS scores in the ITT population, mITT population, and per-protocol population and also in different subgroups are shown in Supplementary Tables 6–10. Among the patients in the per-protocol population (n = 166), 81 (49%) patients reported a VAS score of 0 at 12 months, including 42 (51%) in the doxycycline group and 39 (46%) in the ceftriaxone group. In addition, 14 patients with a VAS score of 1–3 did not report any residual symptoms relating to LNB. Altogether, 95 of 166 (57%) patients in the per-protocol population (47 in the doxycycline group and 48 in the ceftriaxone group) did not have any LNB-related complaints at 12 months (P = .982). Residual complaints of the per-protocol population are presented in Supplementary Table 11.

CSF Findings

A CSF specimen was collected from 206 of 210 patients before the treatment. The CSF leukocyte count, and concentrations of protein, lactate, and CXCL13 were clearly higher in definite LNB patients (Table 3). CSF leukocytes were predominantly lymphocytes in every participant with pleocytosis.

Table 3.

Cerebrospinal Fluid Findings in the Modified Intention-to-Treat Population Before Treatment Initiation Divided Into Subgroups of Patients According to Certainty of Lyme Neuroborreliosis and Study Treatment

Specimen	Doxycycline (n = 94)	Ceftriaxone (n = 93)	Definite LNB + Doxycycline (n = 51)	Definite LNB + Ceftriaxone (n = 48)	Possible LNB + Doxycycline (n = 41)	Possible LNB + Ceftriaxone (n = 43)
Leukocytes/μL, median (IQR)	37 (4–103) n = 92	24 (2–93) n = 91	87 (35–203) n = 51	82 (42–215) n = 48	3 (0–21) n = 41	2 (1–8) n = 43
Protein level, g/L, median (IQR)	654 (436–1168) n = 90	613 (423–1055) n = 90	1.0 (0.7–1.5) n = 51	1.0 (0.6–1.6) n = 48	0.4 (0.3–0.6) n = 39	0.4 (0.3–0.5) n = 42
Lactate level, mmol/L, median (IQR)	1.9 (1.5–2.2) n = 72	1.8 (1.6–2.2) n = 77	2.1 (1.9–2.5) n = 38	2.2 (1.7–2.7) n = 39	1.7 (1.5–2.2) n = 34	1.7 (1.5–2.0) n = 38
CXCL13 concentration, pg/mL, median (IQR)	249 (<7.8–3132) n = 84	392 (<7.8–4340) n = 88	3020 (762–9353) n = 45	3303 (430–15 325) n = 46	<7.8 (< 7.8–85) n = 39	< 7.8 (<7.8–72) n = 42
Positive Borrelia burgdorferi PCR, no. (%)	12 (14) n = 83	12 (14) n = 85	12 (29) n = 45	12 (27) n = 44	0 n = 38	0 n = 41

Abbreviations: IQR, interquartile range; LNB, Lyme neuroborreliosis; n, specimens analyzed; PCR, polymerase chain reaction.

In the mITT population, the CSF leukocyte count was ≥50 cells/μL in 77 of 187 (41%) patients at the beginning of the antibiotic treatment. Three weeks after treatment initiation, a control CSF specimen was successfully collected from 64 of 77 (83%) patients (32 patients in the doxycycline group and 32 patients in the ceftriaxone group). Five patients were not included in the per-protocol analysis because of interruption (n = 3) or a modification (n = 2) in the study treatment. One patient had other severe diseases, which made the evaluation of VAS after 12 months impossible. Thus, in total 58 CSF control specimens collected at 3 weeks (29 samples from patients with doxycycline and 29 samples from patients with ceftriaxone) were included in the per-protocol analysis.

Forty-one of 64 (64%) patients had a decline in CSF leukocyte count <80%. Two of them could not continue the study treatment because of an allergic reaction. In the case of 9 patients, the prolongation of the antibiotic treatment was not performed by a mistake. Thus, according to the protocol, 15 of 94 (16%) patients received an 8-week course of doxycycline, and 15 of 93 (16%) patients a 4-week course of ceftriaxone. In addition, 1 patient with a CSF leukocyte count of 131 cells/μL before treatment refused the control lumbar puncture, and the doxycycline treatment was prolonged to 8 weeks. One patient in the ceftriaxone group with treatment prolongation was excluded from per-protocol analysis because of the treatment modification. No statistically significant difference in the VAS scores was detected between the patients with CSF cell count decrease of >80% and <80% (Supplementary Table 12). However, these analyses are underpowered due to small number of patients. Patients with a decline in the CSF leukocyte count of >80% had statistically significantly higher leukocyte count at the beginning than patients with a decline of <80% (median, 255 vs 95; P = .03).

Secondary Outcomes

There were no statistically significant differences in the secondary outcomes (CSF leukocyte count, protein, and lactate levels or CXCL13 concentration) at 3 weeks after treatment initiation between the treatment groups (Table 4). All CSF specimens tested for B. burgdorferi DNA at 3 weeks (n = 62) were negative.

Table 4.

Initial and Follow-up (3 Weeks After Treatment Initiation) Cerebrospinal Fluid (CSF) Specimen Laboratory Results of the Patients Who Had a CSF Leukocyte Count of >50 Cells/μL in the Initial Sample

Laboratory Specimen	Doxycycline (n = 29)	Ceftriaxone (n = 29)	P Value
CSF leukocytes/μL, median (IQR)
At initiation	108 (78–259)	130 (74–238)
After 3 wk	29 (19–55)	37 (21–58)	.287
CSF protein level, g/L, median (IQR)
At initiation	1.15 (0.89–1.53)	1.16 (0.78–1.70)
After 3 wk (nDOX = 28, nCRO = 28)	0.56 (0.45–0.78)	0.59 (0.49–0.75)	.954
CSF lactate level, mmol/L, median (IQR)
At initiation (nDOX = 25, nCRO = 22)	2.2 (2.0–3.1)	2.3 (1.8–2.5)
After 3 wk (nDOX = 25, nCRO = 27)	1.8 (1.5–2.2)	1.8 (1.6–2.0)	.876
CSF CXCL13 concentration, pg/mL, median (IQR)
At initiation (nDOX = 28, nCRO = 27)	4028 (991–11 694)	3950 (1280–16 600)
After 3 wk	132 (42–189)	120 (47–303)	.744

Difference between the groups after 3 weeks was evaluated using Mann-Whitney U test.

Abbreviations: CSF, cerebrospinal fluid; IQR, interquartile range; nCRO, number of specimens analyzed in ceftriaxone group, unless n = 29; nDOX, number of specimens analyzed in doxycycline group, unless n = 29.

Adverse Events

The treatment was discontinued in the case of 7 patients due to adverse events (2 patients treated with doxycycline and 5 patients treated with ceftriaxone). Five (4.8%) patients treated with doxycycline had an adverse event (rash, nausea, solar eczema). In the ceftriaxone group, 9 (8.5%) patients had an adverse event due to the treatment (rash, rise in liver enzyme levels, diarrhea, Clostridioides difficile infection, superficial thrombophlebitis due to the intravenous cannula). No clear Jarisch-Herxheimer reactions were noticed among the patients.

Discussion

In this randomized, open-label trial, 210 adults with suspected LNB were assigned to receive either oral doxycycline or intravenous ceftriaxone treatment. The major finding of the study was that LNB patients improved equally well with both treatment regimens. In addition, there was no statistically significant difference in the secondary outcomes (CSF leukocyte count, protein level, lactate level, and CXCL13 concentration) after 3 weeks of treatment initiation between the groups.

The initiative for this study was the different LNB treatment practices in various countries [10], and even in different hospitals within a country. For example, in Finland, the typical practice in LNB treatment at the beginning of the 2000s was 2 weeks of intravenous ceftriaxone followed by oral amoxicillin or doxycycline for 3 months. A later study showed no additional benefit of the prolonged oral treatment [11], and thereafter LNB patients have been treated with intravenous ceftriaxone for 3 weeks.

Complications related to intravenous treatment, such as thrombophlebitis and secondary infections [12], would be avoided by using oral antibiotics, which would also be more comfortable for the patients and cheaper for the healthcare system. However, the uncertainty concerning LNB treatment practices still prevails in many countries, although some health authorities recommend the use of oral rather than intravenous treatment due to the lower costs, smaller risks for the patient, and a higher degree of patient satisfaction [13, 14]. Indeed, there are studies suggesting the noninferiority of oral doxycycline in comparison to intravenous ceftriaxone [8, 9], although there are differences in the mechanisms of action, pharmacokinetics, and pharmacodynamics between doxycycline and ceftriaxone, such as ceftriaxone being bactericidal, while doxycycline is mainly bacteriostatic [15]. Importantly, also doxycycline, with a daily dose of 200 mg (either 200 mg once daily or 100 mg twice daily), penetrates through the blood-brain barrier, resulting in CSF antibiotic levels above the lowest, but not the highest, reported minimum inhibitory concentration for B. burgdorferi [16, 17]. On the other hand, CSF concentrations of doxycycline with the same dosing did not reach the minimum bactericidal concentrations for B. burgdorferi [16–19]. The European Federation of Neurological Societies (EFNS) recommends a doxycycline dose of 200 mg once a day, and the Infectious Diseases Society of America (IDSA) recommends 100 mg–200 mg twice a day for the treatment of LNB [20, 21].

In our study, LNB patients recovered equally in both treatment groups with an equal decline in the VAS scores. The results are in line with a previous LNB treatment trial [9]. In that study, a 14-day oral doxycycline treatment was not inferior to 14 days of intravenous ceftriaxone. However, 59% of the patients reported residual symptoms at 4 months after the treatment. In our study, with a longer duration of treatment, and with the final evaluation time point at 12 months, 43% of the patients reported residual symptoms. In another, open-label, nonrandomized study, LNB patients treated with 10–14 days of ceftriaxone or doxycycline, 79% of the patients with ceftriaxone, and 72% of patients with doxycycline were completely recovered at 6 months [8]; in this study the method used to evaluate the clinical improvement of the patients was not thoroughly reported. However, based on these 3 studies, it appears that the 2 drugs are equally effective in the treatment of LNB.

Another issue of controversy is the optimal duration of LNB treatment. There are results showing that patients do not benefit from treatment durations >28 days, but there are no class I trials with different treatment durations [11, 20, 22]. The EFNS guidelines recommend 14–21 days of treatment with either oral or intravenous antibiotics depending on the duration and manifestations of LNB [13, 20]. The length of antibiotic treatment recommended by IDSA guidelines ranges from 10 to 28 days [21]. In the present study, we chose to use 28 days of doxycycline and 21 days of ceftriaxone because in a previous study many patients had persistent symptoms after 14 days of treatment [9] and because patients with late LNB were also eligible for this trial. However, it seems that no obvious benefit was achieved from the longer treatments, and patients in this study had residual symptoms roughly as often as patients in a previous study with 14 days’ treatment [23].

The purpose of the control CSF specimen was to monitor the treatment response with laboratory parameters. We decided not, however, to expose all our patients to a control lumbar puncture because of the invasive nature of the procedure and because it can cause complications such as postpuncture headache [24]. In the per-protocol analysis of 58 CSF specimens, laboratory parameters including leukocyte count, protein level, and CXCL13 concentrations showed similar decline in both treatment groups, supporting the efficacy of both treatments. However, after 3 weeks of treatment, the CSF leukocyte count was still elevated in most the patients. We defined in the study protocol that a decline in the CSF leukocyte count by 80% or more would be a sign of clearance of the infection, and therefore our design included a longer treatment for those patients who did not meet this criterion. Indeed, 64% of the patients did not respond to the treatment with an 80% decline within the timeframe, which led to extension of doxycycline treatment to 8 weeks in 15 patients, and ceftriaxone treatment to 4 weeks in 15 patients. However, there was no statistically significant difference in VAS scores at 12 months between the different treatment durations, although these analyses were underpowered. Thus, it remains uncertain whether the patients had any additional benefit from the longer course of antibiotics.

Patients with possible LNB were also included in this study. Our definition of possible LNB was, however, different from the EFNS guidelines, but similar to the definition used in one of the previous studies [9]. Importantly, patients with neurological symptoms and B. burgdorferi antibodies in serum, but no CSF pleocytosis or antibodies, form a considerable group of patients with suspected LNB. It is estimated that approximately 25% of patients with facial nerve palsy due to LB do not have CSF pleocytosis [25, 26]. However, these patients are often treated with antimicrobials. Based on the results of the present study, it is obvious that possible LNB patients, as they were defined in this study, will also benefit from the treatment.

This study has some limitations. First, we used the change in the subjective VAS score as a measure of treatment outcome. However, in LNB, the subjective experience of improvement by the patient, which is easily evaluated using the VAS score, may be considered as the ultimate indicator of recovery, especially when we consider the multitude of posttreatment symptoms suggested to be associated with LNB. Our results show that when properly diagnosed LNB patients receive antibiotic treatment, most of them recover excellently. Second, we did not record the symptoms or adverse drug reactions with a structured form. The presenting symptoms at the beginning and the residual symptoms at the end of follow-up were gathered from the medical records. This might result in some inaccuracy in reporting of symptoms.

The results of the present study can be applied to everyday clinical practice, because the participants in the study were typical LNB patients. The results should encourage physicians to use oral doxycycline in the treatment of LNB instead of intravenous ceftriaxone. The benefits of doxycycline treatment when compared to ceftriaxone include affordability, simple administration, and avoidance of complications related to intravenous treatment. The optimal duration of treatment is still undetermined and needs further research [27], as well as the ideal treatment of late LNB and LNB with central nervous system involvement.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. E. K., M. J. K., J. P., J. H., and J. O. participated in the design of the study. M. J. K., A. L., J. H., and J. O. obtained funding for the study. E. K. was responsible for the data collection, the data analyses, and writing of the draft manuscript. S. H. performed the power analysis, designed the statistical plan, and supervised the analysis. M. K. was the other consultant biostatisticians. A. P. and J. H. participated in the laboratory data collection and analysis. E. K., M. J. K., J. P., L. A., A. L., U. H., P. J. K., T. N., T. F., T. H., M. M. V., J. V., and J. O. participated in the clinical part of the study. All authors participated in the revision of the manuscript, contributed to the intellectual content, and approved the final manuscript.

Acknowledgments. The authors acknowledge H. Tukkiniemi and T. Rantasalo for their help in the laboratory and E. Mattila, A. Hanttu, M. Kivelä-Rajamäki, I. Jääskeläinen, P. Kivelä, M. Eriksson, and M. Pitkäpaasi for their contribution in the follow-up of the study participants.

Financial support. This work is supported by the Turku University Hospital, Helsinki University Hospital, the Jane and Aatos Erkko Foundation, the Orion Research Foundation, and the Finnish Medical Foundation.

Potential conflicts of interest. E. K. reports a grant from the Orion Research Foundation and the Finnish Medical Foundation and congress travel funds from Merck Sharp & Dohme Corp. (MSD). M. J. K. reports a lecture fee from Orion and Pfizer. J. P. reports congress travel/lecture honoraria from AbbVie, Aducate, Boehringer-Ingelheim, Nutricia Medical, Pharmac, Pharmacy Learning Center, MSD, Novartis, Orion, Roche, and Sanofi-Aventis; and grants from Varsinais-Suomi Hospital District and Satakunta Hospital District. L. A. reports institutional research funding from BiogenIdec, Novartis, Roche, and Merck; congress and travel funds from Sanofi Genzyme, Teva, Merck, Biogen, Roche, and Novartis; and compensation for lectures and advising from Novartis, Sanofi Genzyme, Teva, Merck, Biogen, and Roche. U. H. reports congress travel funds from MSD, Pfizer, Gilead, and Grifols. P. J. K. reports travel grants from Janssen, Gilead, MSD, and Pfizer; and a stock ownership in the Orion company. T. F. reports congress travel funds from MSD, Gilead, GlaxoSmithKline, Janssen-Cilag, and Novartis. T. H. reports a travel grant and a lecture fee from MSD. A. P. reports a grant from the Orion Research Foundation. M. M.-V. reports 2 lecture fees from Pfizer and coverage for congress travel/accommodation expenses from Grifols, MSD, Janssen, Steripolar, and Pfizer. J. V.​ reports travel grants from MSD, Gilead, and the Nordic Society of Clinical Microbiology and Infectious Disease; lecture fees from MSD and the Finnish Medical Society Duodecim; and research grants from the Maud Kuistila Foundation, the Finnish Medical Foundation, Rauno ja Anne Puolimatka Foundation, and Competitive State Financing of the Expert Responsibility area of Turku University Hospital. J. H. is a part-time consultant for the diagnostic company Reagena (Toivala, Finland) and reports a congress travel honorarium from Pfizer. J. O. has been a scientific advisor (review panel or advisory committee) to Gilead Sciences Finland, GlaxoSmithKline, MSD Finland, and Unimedic Pharma AB; received lecture honoraria from Gilead Sciences Finland, GlaxoSmithKline, and MSD Finland; and received coverage for congress travel/accommodation expenses from Gilead, Grifols, and MSD. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

J. H. and J. O. contributed equally to this work.