Worldwide, 40 million persons are disabled owing to filariasis-related morbidity, with 15 million suffering from lymphedema (LE) or elephantiasis and 25 million with hydrocele. Filarial LE is caused by infection with the lymphatic filarial nematodes, Wuchereria bancrofti, Brugia malayi, and Brugia timori and occurs most commonly in the legs. Patients experience a gradual and progressive development in the severity of clinical LE, graded 1–7 (Supplementary Figure 1), leading to severe disability, loss of productivity, and social stigmatization [1]. Patients with LE also experience “acute attacks” or acute dermatolymphangioadenitis, arising through secondary microbial infection acquired through lesions in the skin [2]. The goals of the Global Programme to Eliminate Lymphatic Filariasis (GPELF) are (1) to interrupt transmission by reduction of microfilaremia levels using mass drug administration (MDA) of filaricidal drugs and (2) to provide morbidity management for those who suffer from clinical manifestations associated with lymphatic filariasis (LF) [3–5]. The first goal has been successfully approached in areas having used diethylcarbamazine and albendazole for mass drug treatment for ≥5 years but has been met less successfully in Africa, where ivermectin and albendazole have been administered [6]. During the first 10 years of the GPELF, prevention of new morbidity has been impressive, with an estimated 22 million persons protected from LF infection and disease, accounting for economic savings of US $24.2 billion [4]. However, the goal to reduce existing morbidity associated with chronic filarial disease has not scaled up as rapidly as MDA, with only 33% of endemic countries introducing morbidity management in the first decade of the GPELF [1]. Current morbidity management strategies rely on improving hygiene and skin care of affected limbs, with limb elevation, exercise, and the use of topical antibiotics and antifungal creams, which reduces the frequency of acute attacks and can help arrest the development of LE [3]. We have shown elsewhere that 6 weeks of doxycycline results in a significant amelioration of LE severity in patients with active infection of W. bancrofti [7]. The present study was designed to address whether these benefits of a 6-week course of doxycycline (200 mg/d) could be extended to patients negative for circulating filarial antigen (CFA). In addition, we tested to determine whether a similar course of amoxicillin was also effective against LE. Our results demonstrate that doxycycline improves mild to moderate LE over 2 years. Importantly, CFA-negative patients also demonstrated significant improvement in LE, showing that the activity of doxycycline is not confined to patients with active infection. These results strongly promote the use of doxycycline as a new strategy for improved morbidity management of LF. MATERIALS AND METHODS The trial was approved by the Committee on Human Research, Publication and Ethics at the Kwame Nkrumah University of Science and Technology, Kumasi, Ghana. The study conformed to the principles of the Declaration of Helsinki 1964 (amended most recently in 2004). The trial registration number is ISRCTN 90861344. Participants and Study Area This trial was conducted in 21 endemic communities in the Nzema East and Ahanta West districts in the western region of Ghana [7–9], where MDA was started in 2001. Written informed consent was obtained from all participants. Eligible individuals were adults of both sexes aged 18–60 years with LE stage 1–5 according to the staging scheme by Dreyer [5], with a minimum body weight of 40 kg and no requirement for chronic medication. Exclusion criteria included LE stage 6 or 7; abnormal gamma-glutamyltransferase, alanine aminotransferase, or creatinine levels; pregnancy; breast-feeding; intolerance to doxycycline or amoxicillin; and alcohol or drug abuse. A clinician (S. M.) experienced in the symptoms of LE examined all consenting patients and performed the staging as well as ultrasonography at pretreatment and follow-up examinations. Randomization and Masking This was a randomized trial; the design was placebo controlled for the comparison of doxycycline and placebo, with an additional open amoxicillin arm. Randomization was carried out using computer-generated software. Blinding (masking) was ensured by exclusion of study personnel performing clinical or laboratory analyses from randomization or tablet distribution. To avoid an influence on results caused by the open amoxicillin arm, the personnel administering the treatment (A. B. D. and Y. M. D.) did not take part in follow-up examinations. A total of 162 patients were randomly allocated into 3 treatment arms: (1) amoxicillin, (2) doxycycline, and (3) placebo treatment, each with 54 patients. Before randomization, participants were stratified into those with and those without active infection (CFA positive and CFA negative). These 2 subgroups were separately randomized, resulting in 6 final treatment groups (Figure 1, Supplementary Figure 2A and B ). Figure 1. Participant flow. Flow chart of all assessed volunteers (circulating filarial antigen positive and negative). Abbreviations: CFA, circulating filarial antigen; USG, ultrasonography. Treatment and Follow-up Participants received doxycycline (2 100-mg capsules daily), matching placebo, or amoxicillin (2 500-mg tablets daily) for a total of 6 weeks. Treatment was directly observed daily. Adverse events were recorded in the case report forms, and patients were seen by the trial clinician (A. B. D.). Two rounds of single-dose ivermectin and albendazole were distributed during the clinical trial within the scope of the annual MDA. Follow-up time points were 3, 12, and 24 months after treatment onset. Foot Care Training and Monitoring of Hygiene Measures All patients received soap, towels, and plastic bowls for washing their legs and were trained in foot hygiene, according to the booklet “New Hope” for persons with LE [5]. LE Staging Staging of LE was performed according to the “Basic Lymphedema Management” guidelines established by G. Dreyer and colleagues [5] (Supplementary Figure 1). To allow comparisons of LE staging in patients with either one or both legs affected, the following approach was employed: (1) if only one leg showed LE, this leg was analyzed; (2) if both legs were affected, the leg with the lower stage is reported; (3) if both legs were affected showing equal stages, either the left or the right leg was analyzed after computer-generated randomization. Monitoring of Acute Attacks Acute attacks were defined as pain, lymph node swelling (femoral and/or inguinal), fever, and peeling of the skin on the affected leg after the resolving of the febrile attack (Supplementary Figure 3). The history regarding acute attacks was evaluated based on booklets kept by the patients and questionnaires filled out by the research team at every visit. Each study participant was asked in detail about the number and duration of attacks, fever, lymph node swelling, and peeling of the skin to differentiate general pain in affected legs from genuine acute attacks. Microfilaremia, Antigenemia and Vascular Endothelial Growth Factors For quantification of microfilariae in the blood the Whatman Nucleopore filter method was applied using 10-mL samples of night blood before treatment and at follow-up. Levels of W. bancrofti CFA were determined using the TropBio enzyme-linked immunosorbent assay (ELISA) test (TropBio) [7]. Levels of vascular endothelial growth factor (VEGF) C and soluble VEGF receptor (VEGFR) 3 were measured from plasma samples using Quantikine or DuoSet ELISA kits, respectively (R&D systems). Circumference Measurement of Legs Leg circumferences were determined using a tape measure. Measurements were performed at 10 cm from the large toe and 12, 20, and 30 cm from the sole of the foot, as described elsewhere [10, 11]. Averages of the 4 measurements were determined before treatment and at follow-up. Ultrasound Examinations of the Ankles Ultrasonography was performed between 2 and 7 PM using a SonoSite 180 PLUS hand-carried ultrasound system (SonoSite) equipped with a 38-mm 5–10-MHz linear-array transducer [12]. Patients were scanned sitting with stretched legs and feet perpendicular to the legs. The transducer was positioned at the top of the malleolus (ankle) and kept at a 90° angle to the skin surface in transverse sections. The head of tibia or fibula had to be visible, and the malleolus had to appear as a sharp line (Supplementary Figure 1) to permit reproducibility. The thickness of the tissue (subcutis, dermis, and epidermis) was measured from the malleolus to the skin surface. Lateral and medial malleoli of both legs were measured before treatment and at follow-up. Statistical Analysis To compare between the treatment arms, the intraindividual differences in staging or measurements between the respective follow-up and pretreatment findings were first calculated. Kruskal-Wallis tests followed by Mann-Whitney U tests or Fisher's exact test were used to compare between the treatment arms. The Wilcoxon signed rank test or the McNemar test were chosen for comparing pretreatment with follow-up results in one treatment arm. Correlations were done using the Spearman rank test. Differences between the treatment arms in the survival curve (Kaplan-Meier) were analyzed with the log-rank test. To confirm the results of the per-protocol analyses (patients who completed the treatment and were present for the respective follow-up visit) an intention-to-treat (ITT) analysis was carried out with all patients who started treatment (missing values were filled in using the "last observation carried forward" method). Analyses were done using PASW statistics 18.0 software (IBM) and SAS 9.2 software (SAS). RESULTS Participants From a recruitment pool of 205 patients with LE, 162 patients were stratified according to CFA status (Figure 1, Supplementary Figure 2A and B ). At the start of treatment, 21 patients abandoned participation, although they had explicitly agreed to participate and had been randomized, and 13 participants stated that they had not given correct information about age or breast-feeding on recruitment. Along with these 34 patients, another 9 dropped out during treatment for reasons unrelated to the study drugs. Baseline Data Of the 119 patients who completed the treatment, 46 (39%) were CFA positive and 73 (61%) were CFA negative. Most of the patients had LE stage 2 (44.5%) or 3 (47.9%). The mean age was 47.7 ± 10.8 years. CFA-negative patients had on average a longer history of LE development than CFA-positive patients (P = .029). There was no difference among the treatment arms regarding compliance with MDA (Table 1). Table 1. Baseline Data Variable Total Doxycycline Amoxicillin Placebo P Value Patients (male/female), No. 119 (34/85) 46 (10/36) 35 (9/26) 38 (15/23) .198a CFA positive 46 (17/29) 19 (5/14) 13 (6/7) 14 (6/8) .471a CFA negative 73 (17/56) 27 (5/22) 22 (3/19) 24 (9/15) .153a LE stage 1 2 (0/2) 1 (0/1) 1 (0/1) 0 LE stage 2 53 (16/37) 20 (5/15) 18 (4/14) 15 (7/8) LE stage 3 57 (15/42) 23 (4/19) 15 (5/10) 19 (6/13) LE stage 4 1 (0/1) 1 (0/1) 0 0 LE stage 5 6 (3/3) 1 (1/0) 1 (0/1) 4 (2/2) Age, mean ± SD, years 47.7 ± 10.8 46.1 ± 11.6 51.3 ± 9.1 46.3 ± 10.9 .045b CFA positive 49.7 ± 12.1 45.5 ± 13.9 56.0 ± 5.6 49.4 ± 12.1 .045b CFA negative 46.5 ± 9.8 46.5 ± 9.9 48.6 ± 9.7 44.5 ± 9.9 .304b Duration of LE, mean ± SD, years 13.8 ± 12.3 13.7 ± 11.8 14.6 ± 14.9 13.3 ± 10.5 .893b CFA positive 11.5 ± 12.7 9.3 ± 10.2 12.2 ± 16.5 13.7 ± 12.3 .475b CFA negative 15.3 ± 11.9 16.9 ± 11.9 16.0 ± 14.1 13.0 ± 9.5 .554b MDA/total, No. (%) 2006 87/117 (74) 31/46 (67.4) 28/34 (82.4) 28/37 (75.7) .329a 2007 85/115 (74) 30/44 (68.2) 27/34 (79.4) 28/37 (75.7) .535a 2008 75/109 (69) 25/41 (61.0) 24/31 (77.4) 26/37 (70.3) .347a Abbreviations: CFA, circulating filarial antigen; LE, lymphedema; MDA, mass drug administration (ivermectin plus albendazole); SD, standard deviation. a Fisher's exact test. b Kruskal-Wallis test. Primary Outcome Analysis LE Staging Figures 2 A–C show changes in LE stages before treatment compared with 12 and 24 months after treatment. The affected legs of patients in the doxycycline group reverted to a lower stage at 12 and 24 months (P = .002; Table 2), whereas they progressed to a higher stage in the amoxicillin and placebo groups (P = .012 and P = .001 respectively). Comparing all groups at 24 months (Figure 2 A), there was a difference between doxycycline and amoxicillin groups (P 0 denote an increase to higher LE stages. Abbreviation: LE, lymphedema. Figure 3. Kaplan-Meier curves showing occurrence of acute attacks after treatment end for each treatment arm. Arrows denote follow-up time points. The following significant differences between the 3 treatment arms were detected at 3 months: amoxicillin versus doxycycline (P = .018) and amoxicillin versus placebo (P = .007); at 12 months: doxycycline versus placebo (P = .012) and amoxicillin versus placebo (P = .007); and at 24 months: doxycycline versus placebo (P = .007). Figure 4. Reduction in skin thickness at the ankles, as analyzed by ultrasound. Box plots show differences in skin thickness at the ankles at 24 months compared to pretreatment (P = .001 for overall difference between the 3 treatment arms; Kruskal-Wallis test). Doxycycline was able to improve conditions in 36.6% of patients and halt progression in 58.5%; in 4.9% the legs became worse (Table 3). This improvement was significantly superior to that seen with amoxicillin or placebo (P 9 years old when LE begins) and pregnant women may be manageable in patients with LE. Tolerability and adverse events were the same in all treatment arms, which, together with our experience in all previous trials [7, 19–22], shows that doxycycline is well tolerated and safe. In conclusion, this trial clearly demonstrates that doxycycline is beneficial in reverting or halting the progression of early stages of filarial LE, regardless of whether there is still active infection. These findings lead us to recommend that individuals with filarial LE stage 1–3 should take a course of doxycycline (200 mg/d) for 6 weeks every other year, or maybe even yearly, and that doxycycline should be considered as a new tool to improve morbidity management of LE. Supplementary Data Supplementary materials are available at Clinical Infectious Diseases online (http://cid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author. Supplementary Data