Blackwater fever (BWF) is a severe clinical syndrome, characterized by intravascular hemolysis, hemoglobinuria, and acute renal failure that is classically seen in European expatriates chronically exposed to Plasmodium falciparum and irregularly taking quinine. BWF virtually disappeared after 1950, when chloroquine superseded quinine. We report 21 cases of BWF seen in France from 1990 through 1999 in European expatriates who lived in sub-Saharan Africa. All patients had macroscopic hemoglobinuria, jaundice, and anemia. Acute renal failure occurred in 15 patients (71%), 7 of whom required dialysis. The presumed triggers of BWF were halofantrine (38%), quinine (24%), mefloquine (24%), and halofantrine or quinine (14%). Glucose-6-phosphate dehydrogenase (G6PD) activity was normal in the 14 patients who underwent this test. Low-level P. falciparum parasitemia was found in 8 patients. All 21 patients survived. Our data and 13 cases reported in the literature suggest a resurgence of classic BWF among Europeans living in Africa and a need to discuss attendant therapeutic implications.
Acute hemoglobinuria with passage of black urine associated with recent or concurrent Plasmodium falciparum infection is a well-known clinical syndrome, first described a century ago . Acute hemoglobinuria indicates massive intravascular hemolysis and can be caused by a variety of factors in patients with P. falciparum infestation, including classic blackwater fever (BWF) as defined by the World Health Organization (WHO) on the basis of classic descriptions , overwhelming malaria with a high level of parasitemia [2, 3], glucose-6-phosphate dehydrogenase (G6PD) deficiency-induced hemolysis [4, 5], and other causes of hemolysis . Unfortunately, the term “BWF” is commonly used to designate all of these conditions, a practice that leads to confusion and complicates analysis of the relevant literature.
The introduction in 1950 of chloroquine for use instead of quinine was associated with a dramatic drop in the incidence of BWF [1, 6]. In recent years, the development of resistance to chloroquine has led to the reintroduction of quinine and the introduction of mefloquine and halofantrine and has been associated with a reappearance of BWF [7, 8]. The objective of this article is to raise awareness among clinicians of the resurgence of BWF among long-term European expatriates in Africa, to discuss the potential pathogenic role of various amino-alcohol drugs, and to discuss therapeutic implications.
Patients and Methods
Definitions. We defined BWF on the basis of the 1990 WHO description: severe, acute intravascular hemolysis with hemoglobinuria and a dramatic fall in hemoglobin value, but scant or absent parasitemia, that occurred in a patient (a European expatriate) who had lived in an area of malarial endemicity for several years, during which amino-alcohol drugs (quinine, halofantrine, mefloquine) were taken in an irregular fashion for prophylaxis and treatment . Consequently, 2 cases of BWF in Africans were excluded.
Tropical splenomegaly syndrome was defined as chronic splenomegaly with asthenia and mild anemia that occurs in a patient living in an area of high malarial endemicity, with at least 3 of the 4 following criteria: (1) massive splenomegaly, (2) high titers of malarial antibodies, (3) high titers of total serum IgM, and (4) a dramatic clinical response to antimalarial agents [9, 10].
Patients. Twenty-one patients were included in our study from 1990 through 1999. Fourteen were recruited by 2 units (the Bichat-Claude Bernard and Pitié-Salpêtrière hospitals); 5 of these 14 cases have already been reported in letters [7, 8]. The remaining 7 cases were recruited by clinicians of the Corresponding Group, at 43 French hospital departments, who were asked to report any cases of BWF during the same period.
Clinical data. Medical records were reviewed retrospectively for information on medical history, physical findings, the course of the disease, its treatment, any complications, and the outcome.
Laboratory investigations. Blood tests were done for all or most patients to determine the following: hemoglobin level; leukocyte and platelet counts; routine biochemistry values; and liver function (including bilirubin, haptoglobin, and lactate dehydrogenase levels). Thick and thin peripheral blood smears also were done. Data from the following tests were available for only some patients: direct antiglobulin; G6PD deficiency; amino-alcohol-dependent antibodies to RBCs; and infectious disease detection. Antibodies to P. falciparum were sought with an indirect immunofluorescent assay for the intra-erythrocytic form of P. falciparum (negative at 1:32 dilution).
Treatment. Antimalarial agents, blood transfusions, and symptomatic treatments were administered at the discretion of the attending physicians. Usually, amino-alcohol drugs were rapidly withdrawn when it was suspected that they had triggered the BWF.
Background epidemiological data. Epidemiological data on malaria cases seen during our study period in French nationals who had been expatriates in Africa for ⩾6 consecutive months at the time malaria was diagnosed were obtained from the Centre National de Référence pour les Maladies d'Importation (CNRMI) [11, 12].
Statistics. Data are reported as means ± SE or as medians (ranges). Categorical variables were compared with the Fisher's exact test. Two-sided P values <.05 were considered significant.
Literature review. MEDLINE was searched for studies mentioning BWF and published from January 1989 through December 1999. Among the studies thus retrieved, we selected those reporting the classic form of BWF in white long-term expatriates.
Epidemiology. The most relevant data are listed in table 1. All 21 patients were white long-term expatriates in Africa who either were sent to France for treatment of BWF or developed BWF shortly after coming to France. Only 5 (24%) were women. Duration of residence in Africa was 18.1 ± 8.9 years (range, 4–37 years). The country of residence was the Central African Republic (8 patients), Ivory Coast (4), Cameroon (2), Guinea (2), or Gabon, Congo, Burkina, Benin, or Republic of Togo (1 patient each). Three patients (14%) had experienced 1 prior episode of BWF. All 21 patients had experienced several episodes of malaria. None used regular prophylactic medication. Seven patients (33%) had preexisting tropical splenomegaly. All 21 patients were treated with an amino-alcohol drug before the onset of BWF, for proven (n = 8) or suspected (n = 13) falciparum malaria. The median time from the last amino-alcohol dose to onset of BWF was 24 hours (range, 5–120 ha).
Background epidemiological data. During the study period, 17,248 malaria cases imported to France were reported to the CNRMI [11, 12]. Table 2 details the cases of falciparum malaria and BWF among French adult expatriates to Africa reported in France. To provide a more accurate picture of BWF in France during the study period, we added to our 21 cases 2 French cases that have been reported elsewhere [13, 14], yielding a total of 23 cases. No comparisons between the incidences of BWF and malaria in relation to quinine, halofantrine, and mefloquine were significant.
Clinical features at admission. Sixteen patients required initial admission to an intensive care unit for a median period of 12 days (range, 4–63 days). The Simplified Acute Physiology Score II  for these 16 patients was 36.8 ± 13.0. All 21 patients presented with macroscopic hemoglobinuria, fatigue, jaundice, and a characteristic slate-gray skin color. Moderate hypotension the required dopamine therapy occurred in 2 patients. Thirteen patients (62%) had nausea, often with vomiting, and 5 (24%) had diarrhea. Splenomegaly was present in 14 patients (67%), hepatomegaly in 11 (52%), and hepatosplenomegaly in 9 (43%). Ten patients (48%) had evidence of minor neurological dysfunction.
Laboratory features. The plasma haptoglobin level (available for 15 patients) was 15.0 ± 10.0 mg/dL (normal range, 97.0–258.0 mg/dL), the leukocyte count (17 patients) was 7.2 ± 3.8 × 103 cells/mm3, and the conjugated and unconjugated bilirubin levels (13 patients) were 6.3 ± 12.6 mg/dL and 4.6 ± 2.8 mg/dL, respectively.
Peripheral blood films were done for all patients: 4 patients had thin smears done in Africa and the remainder had both thin and thick smears done in France. In 8 cases, blood films done before the administration of amino-alcohol drug therapy was initiated revealed a few (⩽1%) erythrocytes parasitized with P. falciparum trophozoites. In the 13 remaining cases, results of blood films were negative, but all but 1 were done after the beginning of amino-alcohol therapy.
P. falciparum antibodies were sought in 12 patients, which included 6 with tropical splenomegaly. In all patients, the results of antibody titers were strongly positive (median, 11,392; range, 4096–48,600). Three patients had high titers of total serum IgM, and all of them had tropical splenomegaly.
Seven patients were tested for amino-alcohol-dependent antibodies to erythrocytes. The test showed halofantrinedependent antibodies in only 1 patient.
A number of tests were done to eliminate differential diagnoses. Test results for antibodies to HIV (n = 8), leptospirosis (n = 7), dengue virus (n = 2), hepatitis A (n = 2), cytomegalovirus (n = 2), Mycoplasma species (n = 2), and hantavirus (n = 1) were all negative. Of the 8 patients tested for hepatitis C antibodies, 2 had positive results. Test results for hepatitis B surface antigen were positive for 1 of 8 patients. Two of 4 patients had antibodies to Epstein-Barr virus, with a profile that suggested reactivation. False-positive results were noted for 4 patients (for VDRL and IgM antibody to cardiolipin, leptospirosis, and cytomegalovirus).
Treatment and outcome. During hospitalization the hemoglobin level dropped to a mean minimum value of 4.6 ± 1.2 g/dL. Transfusions of packed RBCs were needed by 18 of the 21 patients; the median number of units given was 9.5 (range, 2–37 units). Of the 15 patients with acute renal failure, 6 required hemodialysis and 1 needed peritoneal dialysis. The urinalysis suggested acute tubular necrosis, but full recovery occurred in every case. Two patients required mechanical ventilation. Nosocomial infections occurred in 5 patients. Four patients developed a lower urinary tract infection (due to Enterococcus faecalis [n = 2], Escherichia coli [n = 1], or Pseudomonas aeruginosa [n = 1]). The remaining patient had 3 infections: Klebsiella pneumoniae bacteremia, Candida albicans infection of a peritoneal dialysis catheter, and E. coli thrombophlebitis. Acute cholecystitis (acalculous in 3 cases) occurred in 4 patients, of whom 3 required surgery.
Seven patients received quinine after the onset of BWF. In 5 of these 7 cases, the presumed trigger of the BWF was halofantrine, and in 2 of these 5, quinine treatment worsened the symptoms. In the 2 remaining cases, the trigger was quinine itself; the condition of 1 of these patients worsened after further quinine therapy. Three patients received steroid therapy. All the patients with tropical splenomegaly received long-term treatment with chloroquine after the BWF episode. All 21 patients survived. The median duration of hospitalization was 20 days (range, 5–93 days).
We report 21 cases of BWF in European long-term expatriates in Africa whose clinical presentations closely matched the classic descriptions of BWF published in the early 20th century [1, 2]. We found 13 similar cases in the literature that occurred during the same period; 6 of the patients were Italian, 5 were Belgian, and 2 were French (table 3) [13, 14, 16–18]. These 34 cases reported during a 10-year period indicate a resurgence of BWF in European expatriates, since a review of the literature from the 30 previous years revealed only 5 comparable cases [19, 20]. Further evidence of a resurgence of BWF among European expatriates is that the current ratio of BWF cases over falciparum malaria cases (table 2) is comparable to that seen in similar populations during the first half of the 20th century . For instance, the incidence of BWF per 100 cases of malaria was about 1% (0.7%–1.2%) among West African Command personnel during 1941–1943 . During our study period, the incidence of BWF per 100 cases of malaria was about 2.6% (table 2).
However, this figure may be an overestimation, since it has been suggested that ∼50% of malaria cases go unreported , whereas high reporting rates are the rule for severe diseases such as classic BWF. These data suggest that the incidence of BWF in French expatriates and probably in European expatriates is comparable to that noted before 1945. We feel that this resurgence can be ascribed to the reintroduction of quinine (starting in 1986) and to the introduction of the related amino-alcohol drugs mefloquine (1987) and halofantrine (1988) in response to the development of chloroquine resistance in P. falciparum in Central and West Africa. Moreover, the dramatic decrease in the incidence of BWF when chloroquine superseded quinine in 1950 [1, 8, 9] and the reemergence of BWF following the reintroduction of quinine and introduction of mefloquine and halofantrine both strongly suggest that amino-alcohol drugs play a role in the etiology of BWF.
Quinine was the only presumed trigger in 29.5% of cases (10 of 34). In 26.5% (9 of 34) another amino-alcohol drug was taken with quinine, and the trigger may have been either or both. Halofantrine and mefloquine were the only presumed triggers in 26.5% (9) of 34 cases and 17.5% (6) of 34 cases, respectively, a finding that strongly indicates that both of these agents should be considered capable of triggering BWF. This effect is consistent with their chemical similarity with quinine .
Of great interest are the cases of 5 patients (2 in our study) who each had 2 episodes of BWF, triggered by different drugs (halofantrine-quinine, n = 1; quinine-mefloquine, n = 1; or quinine-halofantrine, n = 3) [17, 18]. These 5 cases support the theory that there is cross-reactivity between the 3 amino-alcohol drugs. Moreover, the clinical and laboratory features of BWF were the same no matter which drug was the trigger; this suggests that the 3 drugs precipitate BWF by similar mechanisms. Taken in aggregate, these data suggest that these 3 related amino-alcohol drugs are major contributors to classic BWF, although a study with a control group would be needed to confirm this possibility.
BWF is one of the complications that results from chronic exposure to P. falciparum. It is important to recognize it because its treatment is highly specific. The first clue is a history of a prolonged stay in an area where malaria is endemic. Second, the presentation of BWF is strikingly different from that of severe imported falciparum malaria, in which macroscopic hemoglobinuria is exceedingly rare , coma is common, anemia is initially mild, and the platelet count is very low [2, 3]. Nevertheless, this rare presentation must be distinguished early from BWF since severe malaria necessitates immediate intravenous injection of a loading dose of quinine.
Therefore, in a long-term European expatriate chronically exposed to P. falciparum, the presence of macroscopic hemoglobinuria should suggest BWF as the most likely diagnosis and should prompt questions about recent use of amino-alcohol drug therapy. In 11 patients tropical splenomegaly was probably present before the BWF episode. Although tropical splenomegaly differs from BWF, it is associated with many factors that increase the risk of BWF, a fact that explains the high proportion of patients with tropical splenomegaly in our study.
In our European patients, the mechanism of hemolysis remains unclear, since G6PD deficiency was not found in any of the 20 tested patients. This is not surprising, since the incidence of G6PD deficiency is extremely low among Europeans. G6PD deficiency may substantially contribute to hemolysis in cases of BWF that occur in non-European populations, however [4, 5]. In our study 14 patients were not tested for G6PD deficiency, but a high frequency of G6PD deficiency among them is highly unlikely. The direct antiglobulin test was positive at admission for 17 of the 25 tested patients. However, we believe it is unlikely that this fully explained the massive intravascular hemolysis. In our study, 6 of the 11 patients with positive test results had preexisting tropical splenomegaly, which is a well-known possible cause of a positive direct antiglobulin test. For the remaining 5 patients, direct antiglobulin test results were only slightly positive and became negative within 3–6 days, results that suggest an autoimmune mechanism was not the sole cause of the massive hemolysis.
A transiently positive result of a direct antiglobulin test in the context of BWF probably reflects nonspecific binding of Igs associated with malaria-induced polyclonal hypergammaglobulinemia . Another hypothetical factor is drug-induced hemolysis . Quinine can induce immune thrombocytopenia and hemolysis with the presence of multiple quinine-dependent antibodies . In our study, only 1 patient was positive for halofantrine-induced antibodies. Infectious diseases that have been reported to contribute to anemia and jaundice include septicemia , leptospirosis [4, 5], and hantavirus infection . Of our 21 patients, 3 had viral hepatitis and 2 had reactivated Epstein-Barr virus infection, which cannot explain the occurrence of BWF.
The exact pathogenesis of classic BWF remains unclear, and it is still difficult to determine the respective contributions to BWF of malariogenic conditions, amino-alcohol treatment, and human behaviors . The mechanism of the massive hemolytic response in the absence of high parasitemia may involve immune lysis of RBCs sensitized by quinine, halofantrine, or mefloquine in patients chronically exposed to P. falciparum . Although it is good clinical practice to seek evidence of G6PD deficiency and amino-alcohol-dependent antibodies, these 2 causes seem to be exceedingly rare in white Europeans.
Despite the initial severity of the disease, there were no deaths among our 21 patients, and only one of the other 13 patients died. Another recent study of patients with BWF that was less severe than in our patients also showed that the prognosis was favorable: the mortality rate was 2% . In contrast, the mortality rate associated with BWF was generally in the 25%–30% range in the early 20th century [1, 2]. This dramatic improvement in prognosis may be explained by advances in medical care, most notably the availability of initial intensive care and of hemodialysis.
There is no consensus with regard to antimalarial drug therapy during a BWF episode. The presumed trigger should be withdrawn immediately. Nevertheless, antimalarial treatment is required for patients with parasitemia and/or preexisting tropical splenomegaly. Because there is probably cross-reactivity between quinine, halofantrine, and mefloquine, appropriate antimalarial agents may include sulfadoxine-pyrimethamine, artemisinin derivatives, and atovaquone-proguanil [24, 25]. In a recent study in Thailand, BWF did not occur during treatment with artemisinin derivatives alone . After a BWF episode, it is reasonable to refrain from the administration of quinine, mefloquine, or halofantrine because of the risk of recurrence. Consequently, other molecules should be used for prophylaxis and curative treatment. More generally, the risk of BWF should be taken into account when quinine, halofantrine, or mefloquine is given as presumptive treatment for malaria in European long-term expatriates to Africa.
In conclusion, our study and literature review clearly suggest a resurgence of classic BWF in European long-term expatriates in Africa. Not only quinine but also halofantrine and mefloquine can induce BWF. Moreover, the reports of recurrences, sometimes triggered by different amino-alcohol drugs, suggest cross-reactivity between these related drugs. Therefore, the current resurgence of BWF may suggest the need for caution in administering these 3 amino-alcohol drugs to patients who are at risk for BWF.
M. Arsac (Centre Hospitalier Régional d'Orléans); L. Brinquin (Hôpital d'Instruction des Armées Val de Grâce, Paris); V. Caudwell (Hôpital Broussais, Paris); J. P. Coulaud, J. Le Bras, and O. Bouchaud (Hôpital Bichat-Claude Bernard, Paris); J. Delmont (Hôpital Félix Houphouet-Boigny, Marseille); C. Gabinski (Hôpital Saint-André, Bordeaux); F. Janbon, O. Jonquet, and J. J. Béraud (Centre Hospitalo Universitaire de Montpellier); M. F. Mamzer-Bruneel (Hôpital Necker-Enfants Malades, Paris); J. D. Tempe (Hôpital Hautepierre, Strasbourg, France); F. Bricaire and M. M. Thiebault (Hôpital de la Pitié Salpêtrière, Paris); and M. Wysocki (Institut Mutualiste Montsouris, Paris).
We thank Antoinette Wolfe and Annabel Vicente for providing technical support, Jean-Christophe Lucet for statistical assistance, and Fabrice Legros for supplying data from the CNRMI clinical database.