To outline the characteristics and define appropriate management of chronic hepatosplenic suppurative brucellosis (CHSB), 905 patients with brucellosis were analyzed. Sixteen episodes of CHSB (14 in the liver and 2 in the spleen) were found in 15 patients. Six patients had had previous remote brucellosis. Twelve patients presented with systemic symptoms, and 12 with local symptoms. Cultures of blood samples yielded negative results in all cases except 1, and the results of cultures of pus specimens were positive for Brucella melitensis in only 2 cases. All patients showed calcium deposits surrounded by a hypodense area on computed tomography. Patients often had low titers of agglutinating antibody. In patients who were receiving conservative management, early response was successful in 50% and late response was successful in 33.3%. In the patients who underwent surgery and concomitant antibiotic therapy, early and late response was successful in 100%. Thus, CHSB mainly represents a local reactivation of previous brucellosis. Its diagnosis may be difficult to establish and surgery may be required to cure many patients.
Diffuse hepatic and splenic involvement is usually recorded during the course of human brucellosis, underlining the important role of these organs in the defense mechanisms against Brucella infections . Liver or spleen enlargement and/or mild nonspecific elevation of liver enzyme levels caused by nonspecific or granulomatous hepatitis may be detected in ∼50% of patients with brucellosis [2–10]. However, the observation of a hepatic or splenic abscess as a complication of acute brucellosis is exceptional [1, 11–15].
Nonetheless, the first reports of chronic hepatosplenic suppurative brucellosis (CHSB) date from the preantibiotic era [16–21]. In 1964, Spink  described this complication in 6 patients who had a prolonged, recurrent disease caused by Brucella suis. The particular capacity of this species of Brucella to originate caseous necrosis has been repeatedly emphasized [1, 11, 23]. Since then, there have been occasional reports of this complication, mainly in the Mediterranean countries in which brucellosis due to Brucella melitensis is endemic [24–48], but there is need for a general review of the current understanding of this entity in patients with brucellosis.
We describe CHSB, on the basis of data from a large series of patients with brucellosis who were followed during a long period. We analyze the pathogenesis and diagnostic problems of the disease and discuss the contribution of modern radiological techniques to therapeutic management.
Patients, Materials, and Methods
Patients. To identify subjects with suppurative brucellosis of the liver or spleen, we reviewed the data on 905 patients with brucellosis who were admitted to the Hospital de Bellvitge and the Hospital General Vall d'Hebrón, two 1000-bed teaching hospitals in Barcelona. This is an urban area with a large population of immigrants from other regions of Spain, with an incidence of brucellosis reported over the past 20 years ranging from 1 to 2/100,000 population, and considerably lower than the figure for Spain as a whole (12.48/100,000 population; range, 5.3–22.33/100,000 population). A total 643 patients were admitted to the Hospital de Bellvitge from 1974 through 1999 and were prospectively followed in accordance with a clinical protocol described elsewhere [49, 50]. The other 262 patients were identified prospectively among the patients admitted to the Hospital General Vall d'Hebrón from 1984 through 1999.
Diagnostic criteria. The diagnosis of brucellosis was based on the clinical findings characteristic of the disease and on either positive results of culture of blood samples for Brucella species (in all of the cases that we analyzed, the species was identified as B. melitensis, with the exception of 1 case, in which Brucella abortus was found) or a titer of antibodies, assessed by means of tube agglutination, of ⩾1 : 160. Initially, we analyzed blood cultures by use of the classic method of adding 5 mL of blood to a Ruiz Castañeda flask and incubating the mixture at 35°C for ⩾6 weeks; more recently, we have introduced the modern, automated culture nonradiometric Bactec system . Organisms were identified in accordance with the taxonomic criteria of the International Committee on Nomenclature of Bacteria . In addition to standard tube agglutination, the rose bengal plate agglutination test and Coombs' test for antibody to Brucella were also performed as per standard methods [53–56].
The diagnosis of liver or spleen abscess was made on the basis of suggestive clinical and radiological findings and/or the demonstration of pus, obtained by percutaneous aspiration or at surgery. Radiological studies included abdominal echographic examination in some patients, but they mostly consisted of CTs of the abdomen. In some cases, the cytological study of the material that was obtained by means of percutaneous aspiration or biopsy of liver specimens was used to rule out other illnesses. Pus samples were cultured in 5% horse blood agar and chocolate agar plates and in trypticase, by use of Ruiz Castañeda flasks and thioglycollate medium, for ⩾4 weeks. For the serological diagnosis of brucellosis in these patients, standard criteria were modified and a titer of ⩾1 : 80, determined by means of Coombs' test, was considered significant if other causes were ruled out. Pyogenic abscesses, tuberculosis, yersiniosis, hydatidosis, and amebiasis were all excluded by means of cultures in Lowenstein medium and specific serological tests.
Therapy and evaluation of outcome. The initial therapeutic approach was not standardized. Treatment included antibiotic therapy alone or in association with conservative or surgical drainage or resection, depending on clinical findings and the ward in which the patient was hospitalized.
To evaluate outcome, we used radiological CT findings and serological tests, in addition to clinical symptoms. Because criteria for cure are not well established, we analyzed the outcome by differentiating between early and late response, which we arbitrarily defined. Early response was considered to be effective if clinical findings disappeared during the first 6 months after the beginning of therapy. Late response was evaluated after at least 6 months of follow-up and was classified as “cure” if patients remained free of brucellosis-related symptoms and if no progression of radiological findings was observed, and as “failure” if these conditions were not met.
Patient characteristics. Among the 905 patients with brucellosis who were admitted to our hospitals during the study period, we found 15 patients with a total of 16 episodes of suppurative brucellosis of the liver (14 cases) and/or the spleen (2 cases). One patient developed 2 separate episodes: first a liver abscess and then a spleen abscess 14 years later. Eight patients were men and 7 were women, and the mean patient age was 49.5 years (range, 23–73 years).
Epidemiological aspects. Six patients from rural environments where brucellosis is endemic had previously had episodes of the disease, at 2 years, 20 years, 22 years (2 cases), 30 years, and 35 years prior to their admission during the study period. Five of them had no history of brucellosis-related complaints during this period, until the appearance of hepatosplenic abscesses; in the other patient, brucellosis relapse with pleural effusion was diagnosed twice at another hospital, at 11 years and 8 years prior to admission during the study period. The remaining 9 patients did not have prior episodes of suspected or diagnosed brucellosis. One had worked for many years handling dairy products, and 6 frequently consumed unpasteurized dairy products; no epidemiological factors were identified in the other 2 patients.
Clinical findings. In 12 episodes, the disease was manifested by pronounced systemic symptoms. Localized symptoms were present concomitantly in 7 patients and after an interval of 30 days in 1. In 4 patients, none were observed. In the remaining 4 episodes, onset was insidious, with localized symptoms alone. The time of the evolution of the disease at admission was 2–4 weeks in patients with 11 of the cases, 2 months in 1 case, and prolonged in the other 4 cases (6 months [1 case], 8 months [1 case], and 1 year [2 cases]). Initial diagnoses at admission were acute brucellosis (3 cases), septicemia (1 case), brucellar abscess (3 cases), pyogenic or tuberculous abscess (4 cases), hydatidosis (2 cases), and hepatic carcinoma (3 cases). Systemic and local findings are shown in table 1. The patient who had 2 episodes developed secondary amyloidosis, and 1 patient with spleen abscess showed concomitant chronic hepatitis that was not attributable to a specific etiologic agent.
Laboratory data. The mean erythrocyte sedimentation rate was 66 mm/h (range, 6–108 mm/h); for patients with 10 of the episodes, it was >75 mm/h. The mean hemoglobin level was 11.9 g/dL (range, 8.5–16.8 g/dL), and the mean WBC count was 9400 cells/mm3 (range, 2680–16,300 cells/mm3); for patients with 6 of the episodes, the WBC count was >10,000 cells/mm3. Liver tests were unrevealing. The following levels were found to be moderately elevated: alkaline phosphatase, in patients with 10 of the episodes; γ-glutamyl transpeptidases, in patients with 5 of the episodes; and alanine aminotransferases, in patients with 4 of the episodes. The bilirubin level was within the normal range in all cases but 1.
Microbiological data. Results of cultures of blood samples were negative in 15 episodes and positive in only 1 episode, in which B. melitensis was isolated. Cultures of pus specimens, obtained by puncture or drainage of the abscess, yielded positive results in 2 cases (B. melitensis) and negative results in the other 13 episodes. In 9 of these 13 cases, pus was collected before any antibiotic therapy was administered; in the other 4 cases, patients had received antibiotic therapy for 2 weeks (2 cases), 6 weeks (1 case), and 10 weeks (1 case).
All patients had positive results of serological tests. However, at admission, the rose bengal test results were negative in 5 episodes (31%), and titers obtained by means of tube agglutination testing were usually low (in only 2 cases were they ⩾1/160). Therefore, the diagnosis of brucellosis was not made in 2 cases, and its diagnosis was delayed in 3 others, in which seroconversion was detected. In contrast, anti-Brucella titers, assessed by means of Coombs' test, were ⩾1/320 at admission in 12 (86%) of 14 cases. A decrease of ⩾2 dilutions was observed in ⩾1 serological test result over time in 10 cases.
Radiological findings. Localized lesions were detected by means of 99mTc-labeled sulfur colloid scanning of the liver (1 case) and abdominal CT examination in the other 15 episodes, with sizes that ranged from 1.5 cm to 9 cm. Ultrasonographic examination was used for 2 patients to follow the outcome. Multiple lesions were detected in 1 of the 14 hepatic abscess episodes and in 1 of the 2 splenic abscess episodes. In all cases, a calcium density was found in the middle or in the periphery of lesions at the time of admission; these calcium densities were mostly>1 cm in size, and some had a snowflake appearance (figures 1–3). In 3 patients, they had been detected by chance on plain film radiography, several years before the onset of CHSB symptoms. A hypodense area of varying intensity around calcium densities was detected in all cases, but sometimes only after contrast examinations (figures 1–3).
Histological findings. Liver or spleen tissues from 8 patients were available for histological examination. Two patients, in whom nonspecific inflammatory cellular infiltrates were observed, had nonspecific hepatitis diagnosed. In 5 other patients, hepatic granulomas were detected, along with clusters of mononuclear cells, lymphocytes, plasmatic cells, multinucleated giant cells, and fibrous tissue; in 2 of these patients, central necrosis was observed. In the remaining patient, who underwent splenectomy, fibrous hyaline nodules with caseiform pus were found.
Therapy and outcome. A variety of initial therapeutic approaches were applied during the prolonged period of study. Seven patients were initially managed with antibiotic therapy alone, 3 patients underwent conservative nonsurgical drainage, and in the remaining 6 episodes, initial surgical therapy was done. Except for 1 patient who had an episode each of liver abscess and spleen abscess and in whom brucellosis was not diagnosed at that time, all patients received different schedules of concomitant antibiotic therapy; most of them received a combination of 1g (750 mg in patients >50 years) of im streptomycin for 15 days and 100 mg of oral doxycycline b.i.d. for variable periods of time (table 2). Follow-up after therapy was prolonged, ranging from 6 months to 19 years (mean duration of follow-up, 5.2 years).
Among the 7 patients who were initially managed with antibiotic therapy alone, early response was considered a failure in 4 patients (cases 1–3 and 7) who had persistent fever and/or local symptoms at 2 weeks (2 cases), 6 weeks, and 10 weeks after beginning antibiotic therapy (figure 1); early response was considered successful in the other 3 patients (42%; cases 4–6). Late response in these 3 patients with successful early response was nonevaluable in 1 patient (case 6) who had a follow-up of only 6 months; it was considered a failure in 1 patient who had persistent complaints with regard to the upper left abdomen and in whom the CT of the spleen showed an increase of radiological hypodensity around the calcium density over time (case 5; figure 2); it was considered a cure in 1 symptom-free patient with persistent radiological hypodensity (case 4).
Four patients were treated with conservative nonsurgical drainage of pus; for 3 of them, it was the initial therapeutic approach (cases 8–10), and for 1, treatment occurred after failure of antibiotic therapy alone (case 7). Early response was considered a failure in 1 patient, after a single percutaneous puncture, who relapsed 10 days after finishing antibiotic therapy (case 8), and it was considered a success in the other 3 after serial percutaneous puncture (case 9) or pigtail catheter drainage (cases 7 and 10). Late response was considered a failure in 1 patient who relapsed with a suppurative fistula 1 year after stopping the treatment and who required surgical debridement (case 10), and a cure in the other 2 (cases 7 and 9). Overall, among patients who were receiving antibiotic therapy alone or in conjunction with nonsurgical drainage, early response was considered successful in 5 cases (50%) and late response was considered a cure in 3 (33.3%) of 9 evaluable cases.
Eleven patients underwent surgery; for 6 patients, it was the initial therapeutic approach (cases 11–16), and for 5, it was a second therapy because of previous failure (cases 1–3, 8, and 10). Five of these patients underwent liver resection, 1 underwent splenectomy, and 5 underwent surgical debridement. Early response was good for all 11 patients. Late response was nonevaluable for 1 patient (case 3) who had a follow-up of only 6 months, and good for the 8 patients who received concomitant antibiotic therapy (cases 1, 2, 8, 10, and 13–16). The patient with a liver abscess (case 11) and a spleen abscess 13 years later (case 12), who did not receive antibiotic therapy, developed further secondary amyloidosis.
Human brucellosis may present protean clinical manifestations that require a broad differential diagnosis, including many infectious and noninfectious diseases [1, 7–10]. If infection is suspected, the diagnosis can be made easily by isolating the microorganism or by performance of serological tests. However, these diagnostic criteria may not be appropriate for chronic forms of the disease that evolve during long periods of time [20, 22, 28, 43].
In a large series of patients with brucellosis, we identified 15 patients who had a total of 16 cases of hepatic or splenic abscesses and who were considered to have CHSB (1.6%). This incidence is greater than that expected in a general population with brucellosis; it probably occurred because our hospitals work as reference centers for complicated diseases. However, our data indicate that this complication may be more common than is generally supposed. To our knowledge, only ∼30 anecdotal reports of cases have appeared in the literature in the past 40 years [20–22, 24–48].
The detection of calcium densities in the liver or spleen abscesses is a constant characteristic of CHSB and reveals the chronic nature of the disease. Half of our patients had experienced brucellosis many years before, and most had been free of related symptoms for prolonged periods until the development of the hepatosplenic abscess. In the remaining patients (that is, those without previous brucellosis), calcium deposits were always already apparent when the first clinical findings of CHSB were observed. As others have indicated [27, 30, 41, 43], we believe that this strongly suggests that CHSB corresponds to a local hepatic or splenic reactivation of a previous brucellosis, as is usually the case in patients with tuberculosis. Accordingly, patients with CHSB were older than those in our general series of patients with brucellosis. With regard to possible risk factors, 2 reports [37, 44] have associated pregnancy with the reactivation of a latent brucellar liver granuloma, but we found no association between any particular factor and reactivation of the disease in our patients. Although relapses are characteristic in patients with brucellosis, they rarely develop >1 year after antibiotic therapy . A variety of chronic focal forms of the disease that had evolved over years or reactivated after long dormant periods were reported in the preantibiotic era [20, 30, 57]; however, these forms of local reactivated brucellosis are only rarely diagnosed today, unless the diagnosis includes what we describe as CHSB. None of our patients had acute liver or spleen abscesses, and patients with CHSB did not report previous clinical features of local hepatic or splenic involvement. Only a few of the hepatosplenic abscesses reported in the literature have developed in the course of acute brucellosis [12–15].
The clinical presentation of CHSB may be as an insidious and prolonged local disease, but relevant systemic symptoms are usually present. Patients with CHSB do not tend to present with leukopenia and relative lymphocytosis, as often occurs in patients with brucellosis , and biochemical tests of the liver detect only slight abnormalities. Initial diagnosis of CHSB was suspected in ∼20% of our cases. In some cases, hepatosplenic involvement can easily be overlooked; in others, this local disease may be confused with liver carcinoma, hydatidosis, pyogenic or amebic abscesses, and granulomatous infections, such as tuberculosis and histoplasmosis.
Despite the frequency of high temperature, results of blood cultures were negative in most of our cases and in other reports as well . This finding contrasted markedly with our previous experience with regard to human brucellosis, in which we recovered B. melitensis in 78% of cases in the general series of patients and in 90% of those with high temperature . The frequency of negative blood culture results is an illustrative finding of a CHSB corresponding to a local reactivated form of the disease rather than to a primary brucellosis. The results of cultures of caseiform pus were positive for B. melitensis in only 2 of our cases, even though we usually used specific methods to recover Brucella strains and even though most patients had not received prior antibiotic therapy. We could assume that CHSB in our patients was due to B. melitensis, because this species is responsible for 98% of cases of brucellosis in Spain . Similar observations have been made in other reported cases of brucellosis that are supposedly due to B. melitensis, in which the microorganism was rarely recovered , which suggests that the inoculum of the microorganism in the pus is very low. In contrast, B. suis is often isolated when this microorganism is involved [20, 22, 28].
The diagnosis of brucellosis in patients with CHSB may be misleading, given that serological titers, as assessed by means of tube agglutination testing, are usually low [22, 28, 30, 31, 43]; indeed, in 5 of our patients, this diagnosis was delayed or, in fact, was not even made because of the negative agglutination test results at admission. The anti-Brucella Coombs' test is especially useful, although the titers may not be very high. Standard serological criteria should be modified for these patients, and any positive result should be carefully evaluated. This serological behavior is consistent with the idea of a highly prolonged disease, in which the secondary serological response involves IgG and IgA but not IgM. Among the classic tests, the Coombs' test is the best suited to detection of this response [59, 60].
CT is useful for the diagnosis and management of CHSB. Radiological findings are considered characteristic and showed a hypodense lesion of variable size around or adjacent to a calcium density with a snowflake appearance . Although thinly lamellated calcification and a halo effect have been reported as typical [21, 22, 24], they were not observed in our patients. Of interest, this hypodensity may not be obvious on noncontrast CT; it may be interpreted as nonactive calcification, thus leading to misdiagnosis, as occurred in 1 of our patients (figure 1). Multiple lesions—not always in the same state of reactivation—are detected in ∼20% of patients [11, 22, 28, 30, 31]. Tuberculosis and histoplasmosis may be considered in the differential diagnosis [21, 24, 30, 31], but histoplasmosis does not exist in Spain, and no patients with tuberculous hepatic abscess were identified among the 4000 patients with tuberculosis who were prospectively followed in one of our hospitals since 1980 (R. Vidal, Hospital General Vall d'Hebrón, personal communication). Hydatidosis usually presents as calcification, but it can be distinguished by its characteristic thin rim of outer calcification and by the absence of a lamellar or snowflake appearance .
Some authors have reported good therapeutic results with administration of antibiotic therapy alone for several months [37, 41, 43]. However, this therapeutic approach achieved a good early response in only 20%–40% of cases, because systemic or local symptoms persisted and drainage of abscesses was required. The most appropriate antibiotic combination is not known, but the combination of doxycycline and an aminoglycoside is probably best, and the addition of rifampin may be a reasonable option. The combination of medical therapy and conservative drainage of pus by means of a pigtail or similar catheter has been reported to have a successful outcome in several cases of brucellar liver abscess [40, 46]; this is presently considered the standard means of management of pyogenic liver abscess . Three of our patients had a good early response to conservative drainage techniques as well, but 1 patient developed an early relapse. Therefore, the evacuation of pus may be ineffective in some of these cases because of the thickness of the caseiform pus that is usually found [41, 42]. In any case, conservative management of CHSB does not ensure the definitive cure of the disease; 2 of our patients (1 of whom was described elsewhere ) who were initially considered cured developed late failure ⩾1 year after termination of treatment. The most appropriate duration of antibiotic therapy for patients who are receiving conservative management is not well established, but it seems reasonable to prolong oral therapy for several months. The evaluation of late outcome for these patients is complicated, because radiological findings that show calcium densities with varying levels of hypodensity persist over time in most cases, even while patients may remain symptom free (figure 3). In addition, anti-Brucella antibodies, as assessed by use of Coombs' test, decrease over time, but the test often continues to yield positive results with high titers for many years, which raises doubts about the complete eradication of the disease. In contrast, surgical resection of liver abscesses (or splenectomy, in cases of spleen abscess) provide total cure of CHSB, and a shorter duration of antibiotic therapy of 8 weeks may be sufficient.
We conclude that CHSB is a local late reactivation of brucellosis, that it is more common than is usually believed, and that it is often caused by B. melitensis. Its diagnosis may be difficult because the isolation of the microorganism is unusual, serological agglutinating titers are low, and the initial detection of liver or spleen abscesses may be misleading. A conservative therapeutic approach, with use of antibiotic therapy alone or associated with nonsurgical drainage, may be an initial option, but a prolonged therapy over several months is required and a careful follow-up is mandatory because the complete cure of the disease cannot be guaranteed. Otherwise, if no rapid clinical response is observed, surgical resection is recommended, because more complete eradication of the disease is warranted.