Treatment of Q Fever Endocarditis:  Comparison of 2 Regimens Containing Doxycycline and Ofloxacin or Hydroxychloroquine FREE

Q FEVER, a rickettsiosis caused by Coxiella burnetii, was originally described by Derrick in 1937.¹ This zoonosis is widespread throughout the world and can present as either acute or chronic disease. Endocarditis is the most serious and often fatal form of chronic Q fever, and presents unspecifically as a blood culture–negative endocarditis. Diagnosis relies on nonspecific cardiac findings, evidence of an infectious or inflammatory process, and detection of specific antibodies. Coxiella burnetii is a strict intracellular gram-negative bacterium. It lives and multiplies in the phagolysosomes of infected cells at pH 4.8.^2- 3Coxiella burnetii exhibits a phase variation (phases I and II), which is extremely valuable for diagnosis. The serologic response of patients with acute Q fever is mainly directed against phase II, whereas patients with chronic Q fever exhibit a serologic response to phases I and II. However, because of its intracellular position, C burnetii is difficult to eradicate, explaining why—when the disease was described during the 1960s and 1970s—relapses and even death occurred frequently. In some series,⁴ two thirds of the patients died. Since that time, the role of combined and prolonged antibiotic drug therapy for these patients has been emphasized.^5- 7 The current recommended treatment duration is from 3 years to lifetime, even when quinolone is added to the reference tetracycline therapy. This combination treatment enhances patient survival,⁶ but the treatment of Q fever endocarditis is now the longest of any bacterial disease, including leprosy and other mycobacterial diseases.⁸ By testing C burnetii antibiotic susceptibility in a cellular model of infection, it was previously shown that intracellular pH may significantly diminish the bactericidal potency of antibiotic drugs.^9- 10 This bactericidal effect of doxycycline was restored by incubating cells with chloroquine, 1 mg/mL, which acted as a lysosomotropic alkalinizing agent and raised the pH of the phagolysosome from 4.8 to 5.7,¹¹ the optimal doxycycline activity being obtained at pH 6.6. The purpose of the present study was to assess the clinical efficacy of the combination of doxycyline and hydroxychloroquine on the treatment of Q fever endocarditis compared with a previously initiated regimen of doxycyline and ofloxacin.^1,6

Patients were treated from January 1987 to May 1991 with doxycycline and ofloxacin and from May 1991 to December 1997 with doxycycline and hydroxychloroquine regimens. Diagnosis was performed by confirmation of C burnetii infection in patients with chronic endocarditis based on previously published criteria.¹² An IgG titer of 1/1600 and an IgA titer of 1/100 directed against phase I of C burnetii were required for diagnosis. By using the Duke criteria modified for C burnetii, all these patients had a definite diagnosis of endocarditis.¹³ Treatment was prescribed by 1 of us (D.R.) after receiving patient information and consent.

Each patient initially received 1 tablet of doxycycline (100 mg) twice daily and 1 tablet of ofloxacin (200 mg) 3 times daily. In the study's second stage, patients received doxycycline at the same dose and 1 tablet of hydroxychloroquine (200 mg) 3 times a day. Doses of hydroxychloroquine were adapted after monthly plasmatic level evaluation to maintain a concentration between 0.8 and 1.2 µg/mL. Patients were made aware of the risks associated with overexposure to the sun and that they should wear gloves and a hat and use protective sunscreens. Patients with an evolutive immunocompromised disease such as cancer or acquired immunodeficiency syndrome were excluded from the study.

All patients were monitored monthly with a physical examination, blood cell count, transaminase dosage, serologic examination, and hydroxychloroquine plasma level determination. When treatment was stopped, blood cultures were obtained for C burnetii and other organisms. A cardiologist examined each patient 2 to 4 times a year, and an ophthalmologist (J.M.R.) performed a flash electroretinogram and a fundus examination twice a year to detect chloroquine accumulation.¹⁴ Patients were considered cured when IgG antibodies to C burnetii phase I were lower than 1/800 and IgM and IgA antibodies were lower than 1/50.¹²

When these results were obtained, antibiotic drug treatment was stopped and the patients were further monitored by clinical examination and blood sampling, including blood cell count, serologic examination, and blood culture performed for C burnetii and other organisms.¹⁵ Patients were examined monthly for the first 6 months, every 3 months for a further year, and subsequently every 6 months. The patients had an echocardiographic examination twice a year for 2 years.

Measurement of hydroxychloroquine blood levels was performed by liquid chromatography as previously described¹⁶ using a 25-cm Si 60,5 lichrospher column (Merck, Darmstadt, Germany). The mobile phase consisted of a mixture of 20% acetonitrite and ammoniac ethanol (92/8 by volume). Separation was performed by means of a gradient elution. Hydroxychloroquine and quinine, the internal standard, were extracted in an alkaline medium. After mixing for 15 minutes, the organic phase was recovered and evaporated. The residue was removed by acetonitrile, 50 µL, and 20 µL was injected into the chromatograph. The standard range was 0.5 to 10.0 µg/mL. The limit of detection was 0.2 µg/mL.

An indirect fluorescent antibody assay was used in this study. Antigens of C burnetii phase I and phase II (nine mile strain) were placed on 30-spot slides (Dynex, Cel-Line Associates Inc, New Field, NJ) and fixed in cold acetone for 20 minutes. The serum samples were diluted in phosphate-buffered saline solution (PBS) from 1:25 to 1:1600. After the serum samples were overlayed onto the antigen dots, slides were incubated for 30 minutes at 37°C in a moist chamber, then washed 3 times in PBS. Fluorescein-conjugated antiserum IgG (γ-chain specific), IgM (µ-chain specific), or IgA (α-chain specific) (Institut Pasteur Production, Paris, France) diluted 1:200 in PBS was then overlayed onto dried slides and were again incubated for 30 minutes in a moist chamber at 37°C. After washing in PBS, the slides were air dried and mounted with glycerol. The slides were examined under an epifluorescence microscope (magnification ×400; Nikon, Tokyo, Japan). Before determining the levels of IgM and IgA, the serum IgG was removed using a rheumatoid factor adsorbent (Calbiochem-Behring, Marburg, Germany). This factor was diluted in water (1:5), mixed with the serum diluted in PBS (1:5), and centrifuged.

Isolation of C burnetii by cell culture was performed as previously reported.¹⁵ Briefly, culturing was performed by the shell-vial assay with human embryonic lung fibroblasts. Three confluent shell-vials were inoculated for each blood sample and were incubated for 6 and 14 days at 37°C under a 5% carbon dioxide atmosphere. Detection of C burnetii was carried out directly with the shell-vial assay by direct immunofluorescence labeling with a rabbit anti–C burnetii polyclonal antibody. If immunofluorescence test results were negative after 6 and 14 days of incubation, the culture was negative. If positive, the culture was then subcultured to yield the isolate.

A patient was considered to be definitely cured only after an incident-free 6 months of follow-up. A patient was considered as relapsing if an unexplained cardiac abnormality (valvular dysfunction or echocardiographic change), an inflammatory syndrome, hepatitis, or a 4-fold increase of IgG or IgA antibodies to C burnetii was observed. When valvular surgery was performed, samples of valve tissue were inoculated on cell tissue culture in an attempt to isolate C burnetii¹⁵ and were studied by immunohistochemistry.¹⁷

Data were computerized using database and spreadsheet programs (Microsoft Access and Excel, Microsoft Corp, Redmond, Wash). Statistical tests were performed using a software program (Epi-Info, Centers for Disease Control and Prevention, Atlanta, Ga). Proportions were compared using 1-tailed Fisher exact tests, whereas means were compared using Student t tests. A difference was considered significant at P<.05.

Twenty-six men and 9 women were included in the study. Fourteen patients were treated with a combination of doxycycline and ofloxacin (Table 1). One patient died and 7 relapsed and were subsequently treated with the same regimen (3 patients) or were switched to the new regimen (4 patients). Five patients were cured using this regimen only. A total of 21 treatments using this regimen were prescribed. All the patients experienced photosensitivity during the summer after beginning treatment. One patient had an irreversible cutaneous pigmentation attributable to doxycycline use.¹⁸ No diarrhea or joint pain was reported. The patient who died was still receiving antibiotic drug treatment when he had a valvular replacement and died postoperatively of cardiac incompetency. The 7 patients who relapsed had an increase of antibody titers associated with clinical signs within 6 months of treatment interruption. These failures were observed only during the early stages of the regimen. Subsequently, stable low titer of anti–phase I antibodies had to be observed before therapy was stopped (IgG, <800; IgA, <50). The mean treatment duration for patients considered cured was 55 months (median, 60 months). No patient relapsed when treated for longer than 4 years, and only 1 patient relapsed after 3 years of treatment. However, 8 patients are still being treated or were switched to a doxycycline and hydroxychloroquine regimen (Table 1).

Twenty-one patients were treated with a combination of doxycycline and hydroxychloroquine (Table 2). One patient died 30 months after completing treatment: he underwent mitral valve replacement and died postoperatively of cardiac failure, although the valve was sterile. All patients experienced photosensitivity, mainly on the hands and nose. One patient had an irreversible cutaneous pigmentation caused by doxycyline treatment, and 1 patient had an infraclinical retinal accumulation of hydroxychloroquine that led to withdrawal of this drug (patient 3). However, because the treatment was completed, it did not change the course of the therapy. Two patients had decreased serum IgG (anti–phase I) antibody titers to less than 400 within 10 months of therapy despite high initial antibody levels and consequently discontinued treatment after 1 year. They both relapsed within the first 6 months of follow-up. After these cases, we decided to treat all patients for 18 months or longer, and none of the 15 patients treated subsequently relapsed. The 2 patients who relapsed were treated for a further 2 years. Among 21 patients, they were the only 2 who had a significantly lower CD4⁺/CD8⁺ cell ratio.¹⁹ Hydroxychloroquine plasma levels were determined to maintain a steady concentration of 1.0±0.2 µg/mL. Six patients did not require any change and received 600 mg for the duration. One patient, a man weighing 115 kg (patient 5), had low levels of hydroxychloroquine; after 3 months, the dosage was increased to 800 mg/d. In the 16 remaining patients, the dosage was decreased to 400 mg/d within a mean of 6 months (range, 2-13 months). Nine patients required a second dosage reduction to 150 mg within a mean of 3.4 months (range, 2-7 months). The mean treatment duration for cured patients was 31 months, with a median of 26 months. Two patients are still undergoing treatment. Only 2 patients have been treated for longer than 4 years.

The 2 regimens are identical in terms of mortality, valve surgery, and tolerance (Table 3) but significantly different in terms of relapse, which is much more frequent with doxycycline and quinolone combination treatment. The mean hydroxychloroquine regimen duration was also significantly lower by 2 years (Table 3). Moreover, patients treated with doxycycline plus ofloxacin for more than 48 months are still undergoing treatment, and the regimens of 4 patients have been altered to doxycycline plus hydroxychloroquine, and 3 of them are now cured.

We tried to compare the evolution of anti–phase I antibody levels, but the initial levels varied from 1600 to 256,000 for IgG and IgA. With our relatively small number of patients, the curves were hardly interpretable because of the size of the SD. However, the anti–phase I IgM antibodies were the first to reach a titer of less than 1:50, and this significantly more rapidly (6 months) with the doxycycline plus hydroxychloroquine regimen than with the doxycyxline plus ofloxacin regimen (30 months). As for IgG and IgA, no significant differences were noted.

Chronic endocarditis is the most serious complication of infection with C burnetii, often leading to life-threatening conditions.^1,20 Most often, Q fever endocarditis develops in patients with abnormal valves or blood vessels²¹ and involves mitral or aortic valves. Relapses are frequent despite prolonged antimicrobial chemotherapy, and mortality rates can exceed 60%.^22- 23 The antibiotic drug of choice for treatment is still to be determined. The β-lactams, used either alone or in combination with aminoglycosides, are not effective.^6,24 Since the observation of in vitro susceptibility of C burnetii to oxytetracycline (Terramycin) treatment,^25- 26 tetracycline compounds have been the mainstay in chronic Q fever treatment. Tetracycline, used either alone or in conjunction with valve replacement, has often been effective in controlling Q fever infections.^22,27- 28 The antibiotic drug regimen currently used is tetracycline, 2 g/d for several months then 1 g/d for at least 1 year. Surgery is undertaken when heart failure occurs. However, patients often relapse, and some die owing to irreversible damage of the heart valves. Therefore, combination therapy with tetracycline and other antibiotic drugs has been proposed. Turck et al²⁸ described 16 patients with Q fever endocarditis treated with tetracycline alone, lincomycin alone, or a combination of the 2 compounds. The regimen combining lincomycin, 2 g, and tetracycline, 1 g, daily was suggested to be most effective. Combinations of tetracycline and co-trimoxazole were evaluated because co-trimoxazole was reported^23,29- 31 to be effective when used alone. The antibiotic drug regimen used was tetracycline, 2 g, combined with co-trimoxazole (trimethoprim, 320 mg, and sulfamethoxazole, 1600 mg) daily. Some researchers^32- 33 reported a favorable response to treatment with chloramphenicol when combined with tetracycline, whereas others^28,32 reported chloramphenicol not to be effective when used either alone or combined with tetracycline. Rifampin use had promising results when combined with tetracycline.^4,33 Subramanya et al³⁴ reported the complete failure of the combination of rifampin and co-trimoxazole to treat C burnetii infection in a patient who had adverse toxic effects with tetracycline treatment. A 6-month course of doxycycline plus rifampin, followed by doxycycline treatment alone, seemed to be more effective than doxycycline treatment alone.²⁴ However, rifampin use may be contraindicated in patients with mechanical valves who are being treated with anticoagulant agents.

More recently, the demonstration of in vitro susceptibility of C burnetii to the newly available fluoroquinolones prompted the use of these compounds to treat chronic Q fever.³⁵ Haldane et al³⁶ and Ellis et al³⁷ carried out a compilation of the 7 reports on the antibiotic drug efficacy in treating the disease; however, because of the lack of follow-up and the small size of each series, we chose only to compare the failure and death ratios within the different regimens (Table 4). Monotherapy using tetracycline or doxycycline leads to a mortality rate higher than 50%. Taking a regimen associating tetracycline compounds to lincomycin or clindamycin, one third of the patients died. The association of co-trimoxazole is controversial, but it is significantly less efficient than the association of either quinolones or hydroxychloroquine to doxycycline (Table 4). These 2 last regimens had a low mortality rate of approximately 5%, which is lower than that observed in other prosthetic valve endocarditis. The mean death ratio in prosthetic valve endocarditis caused by other bacteria is 50%.³⁸ These observations show the dramatic improvement in the prognosis of Q fever endocarditis, with the death ratio being low for the past 10 years since these new regimens were initiated compared with the series reported by Raoult et al,⁴ in which 6 (40%) of 15 patients died. This progress has previously been reported,⁶ although the treatment duration, when quinolones were used, remained long (between 3 years and lifetime). This protracted regimen is required because of the inhibition of the bactericidal effect of tetracycline and quinolone by the low pH of the host cell phagolysosome in which C burnetii resides.⁹ However, it has now been shown that the addition of chloroquine to the regimen restores the bactericidal effect of doxycycline at a 1-µg/mL concentration.^9- 10 Hydroxychloroquine is widely used in rheumatic practice at a dose of 600 mg/d.³⁹ The patient's daily dose was adjusted to maintain the hydroxychloroquine plasmatic concentration at 1 µg/mL. This was achieved by administration of a daily dose of 150 to 800 mg. The adverse effects of this compound consist of photosensitivity and retinal accumulation. We assessed the condition of the retina in all patients throughout the trial but were obliged to discontinue treatment in only 1 patient. It is likely that even 5 years of treatment will be well tolerated. Doxycycline, hydroxychloroquine, and ofloxacin use all have a photohypersensitivity risk, which was experienced by all our patients. Despite repeated reminders, the first summer was difficult for all patients; however, most were subsequently more careful. Hypersensitivity to sunlight was therefore the major complication with both regimens, although 1 patient with irreversible skin pigmentation was observed in both groups.

Comparing the 2 regimens, treatment with doxycycline and hydroxychloroquine was more effective in terms of the percentage of relapses and the treatment duration. We think that doxycycline and ofloxacin should be prescribed for at least 4 years, whereas doxycycline and hydroxychloroquine should not be prescribed for shorter than 112 years or longer than 4 years.

Accepted for publication April 23, 1998.

We thank Richard Birtles, PhD, for reviewing the manuscript.

Reprints: Didier Raoult, MD, PhD, Unité des Rickettsies, CNRS-UPRESA 6020 Université de la Méditerranée, Faculté de Médecine, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 05, France (e-mail: Didier.Raoult@medecine.univ-mrs.fr).