Background
Blackwater fever (BWF), inspired by the French “fièvre bilieuse mélanurique” [
1‐
3] is the term usually applied to a febrile syndrome with intermittent passage of dark-red to black coloured urine in the context of malaria. Additional features include minimal or absent parasitaemia on peripheral blood smear and variable concurrent symptoms of renal failure, circulatory compromise, pallor, icterus, nausea, vomiting, and epigastric pain [
4,
5]. Confusingly, the term “blackwater” applies only to the typical appearance of haemoglobinuria. Thus whilst the clinical syndrome of haemoglobinuria accompanied by fever was recorded in the time of Hippocrates [
6], its strong association with malaria was not recognized until the 19
th century.
BWF is predominantly associated with
Plasmodium falciparum infection [
7], but cases have been documented in association with
Plasmodium vivax [
8] (alone or in a mixed
P. falciparum infection) [
5],
Plasmodium knowlesi [
9,
10] and
Plasmodium malariae [
5]. Over the twentieth century a clear variation in the number of BWF cases was observed with virtual disappearance of the syndrome in the early 1950s followed by a resurgence of reports from the 1990s [
5,
11‐
15]. These variations may have been associated with changes in first-line anti-malarial therapy. Cases fell as chloroquine replaced quinine after the Second World War, then appeared to rise as quinine and other amino-alcohol drugs were recycled in response to chloroquine resistance [
13,
15]. Mortality has also fallen over time with reports from the first half of the twentieth century suggesting 25–30% mortality [
16] whilst rates reported in later decades are substantially lower [
11,
13].
The cause of massive intravascular haemolysis associated with a low malaria parasite burden is incompletely understood. Mice infected with a lethal strain of
Plasmodium yoelii can develop haemoglobinuria, renal failure and hepatosplenomegaly and these have a far lower parasite burden at death than non-haemoglobinuria mice [
17]. In man, it has been suggested that anti-malarial therapy could be a cause or contributor to the syndrome of BWF. The mechanism by which G6PD-deficient patients develop intra-vascular haemolysis as a response to primaquine-induced oxidative stress is well understood [
18]. Stephens [
16] summarized multiple reports associating quinine with BWF in patients with severe malaria independently of their G6PD status. However, despite the high oxidative potential and generalized use of artemisinin (ART) compounds [
19] there is little evidence so far of their association with BWF, either alone or as part of artemisinin-based combination therapy (ACT) [
11,
20‐
24]. It is worth noting that post-artesunate delayed haemolysis (PADH) may occur in up to 27% of patients with anaemia post ART for severe malaria [
25‐
27]. However, this is an extra-vascular haemolysis predicted by re-entry from the spleen to the circulation of once-infected, microscopically pitted erythrocytes with short half-lives [
27]. Finally, BWF has been observed with other amino-alcohol drugs such as halofantrine [
12,
13,
28‐
30], mefloquine [
12,
13,
31,
32] and lumefantrine, a related aryl-amino-alcohol compound [
20,
33,
34]. To date there is little evidence of BWF in association with chloroquine [
5,
35‐
38] despite its extensive usage, nor with its piperaquine derivative [
21,
37,
38].
The present report describes an unusual case of a returning traveller who developed classic BWF with persistent parasitaemia demonstrated by PCR despite repeatedly negative microscopy after ACT. This leads to a re-examination of the pathogenesis of BWF.
Discussion
The present study documents persistence of parasitaemia detected by PCR in a returning traveller with an illness consistent with blackwater fever based on documented P. falciparum malaria, paroxysms of intravascular haemolysis with severe anaemia, exclusion of alternative cause and a history of exposure to arthemether in combination with the aryl-amino-alcohol drug lumefantrine. Other severe manifestations of BWF such as renal failure were absent. Low or absent levels of parasitaemia are known to be associated with BWF at presentation, but to the authors’ knowledge this is the first case where ongoing submicroscopic parasitaemia was detected in the later stages of illness in the face of negative microscopy.
Since the development of PCR as a research tool for Malaria [
46], molecular methods are increasingly finding their place in the diagnosis and management of malaria and become widely used for malaria surveillance [
40,
47‐
50]. In the case reported here, PCR detected
P. falciparum when multiple blood films were negative. Although in most cases it is possible for well-trained microscopists to detect and identify parasites in PCR-positive specimens, PCR remains more sensitive than microscopy and a useful tool to assess asymptomatic patients [
49,
50] or those with ongoing symptoms but low or null results from microscopy, [
4,
14]. In the present case an unexpected PCR result generated two questions. Firstly, was submicroscopic parasitaemia contributing to ongoing symptoms and secondly, what were the treatment and prognostic implications?
It is fair to say that decades of epidemiologic, clinical and laboratory observation of BWF have produced threads of reasoning which do not lead conclusively to a unifying mechanism of disease and which appear superficially contradictory. Amongst colonial servicemen in West Africa, BWF incidence peaked after 2–3 years of service, usually after several bouts of malaria treated with quinine [
16,
35]. To early researchers this implied an immune-pathologic mechanism whereby immune memory led to massive haemolysis upon repeat infection [
16,
35,
51‐
53]. Whilst this may explain the scanty parasitaemia and high levels of
P. falciparum antibody seen in many cases [
5,
13,
33] the direct Coombs reaction in BWF is variable [
11] and in at least one case of BWF associated with renal failure a renal biopsy did not show immune complex deposition [
54].
Oxidative stress has been advocated as one of the major contributor to the manifestations of BWF in in South Vietnam [
11]. The prevalence of G6PD deficiency amongst sufferers in this region is high and a small number have G6PD mutants of diminished function in the presence of apparently normal G6PD levels [
55]. The metabolites of quinine (the predominant drug used in the report of Chau et al. [
11]) may exert oxidative stress under particular conditions. This hypothesis finds additional support with the acute haemolysis observed in G6PD deficient patients exposed to drugs like primaquine. Artemisinin derivatives have an endoperoxide bridge that can release free radicals when reacting with iron, which may be relevant to recent cases of BWF in patients treated with ACT [
19] if oxidative stress plays a major role in driving this illness.
An alternate possibility is that anti-malarial drugs and the effects of immunity are epidemiologic bystanders to the true cause of BWF simply because most patients with BWF are at chronic risk (therefore immune or partially immune) and have had a de facto indication for drug treatment. There is now some evidence for this proposition (e.g. in murine models [
17]). In humans,
P. falciparum exerts a complex array of effects on RBC function including changes in RBC size, deformability, endothelial adhesion and upregulation of particular outer membrane proteins including the ring-infected erythrocyte surface antigen (RESA) [
56]. In acute
P. falciparum malaria the RESA is present on non-parazited RBCs, suggesting a circulating population of RBC “survivors” from which parasites have been removed by the spleen. This is thought to be why the fall in haematocrit seen in
falciparum malaria is less than it would be predicted by the number of parasitized RBCs. In some circumstances it may be that once-infected erythrocytes (o-iE) become fatally predisposed to non-immune oxidative haemolysis. This may in turn be driven by anti-malarial drug therapy, in this case artemether with lumefantrine. This hypothesis offers the tantalizing possibility that BWF could be a pre-ordained event in a small number of patients who control an initial parasitaemia at the expense of RBC predisposition to non-immune intravascular lysis that mimics or is caused by oxidative stress.
The persistence of PCR detectable parasitaemia after ACT is often observed in asymptomatic children from endemic areas [
57‐
62]. After AL, Beshir et al. [
57] detected residual parasitaemia at day 3 in 31.8% of the children enrolled in a study children treated with AL in Kenya. This was associated with a significantly longer duration of gametocyte carriage, increase in risk of malaria recurrence and higher parasite burden in infected mosquitos. Amongst AL-treated children in western Kenya, Sawa et al. [
59] reported a 20% risk of recurrent parasitaemia at 42 days of follow-up with mean gametocyte carriage (determined by PCR) of 5.5 days. Other reports have clarified that PCR assays detect only viable parasites [
63], and indicate that gametocyte carriage may be highly persistent [
63‐
65]. This was thought to be the most likely explanation for the positive PCR result in the present case. However, here, this assumption has been ruled out by PCR since neither of the two gametocyte-specific RT-PCR assays were positive. The dormant asexual stages that appear after ART treatment might also account for this persistent parasitaemia as they can remain in dormancy for up to 20 days in vitro [
66]. This strategy allows them to escape the rapid action of ART and contributes to the delay in parasite clearance observed in ART resistant strains [
62,
66,
67]. When returning to activity, previously dormant stages are killed by the long-lasting ART partner drug in ACT. The treatment and prognostic implications of submicroscopic populations of
Plasmodium in BWF remain unclear and it is unclear whether clinicians should act on a similar result, given that in the present report, the patient was well until day 42 after treatment guided using traditional end-points.
A weakness in this present case is the inability to determine with certainty when the patient received effective anti-malarial therapy. The impression of treating clinicians at the time of presentation in Singapore was that the anti-malarial therapy obtained from a rural chemist in Ghana attenuated but did not resolve the patient’s symptoms. This, together with the typical appearance of blackwater symptoms and the 0.8% parasitaemia reported from a clinic in Accra was taken to mean that the first course of therapy had been
effective at reducing a far more significant burden of
P. falciparum parasitaemia. A recent study for the artemisinin-based combination therapy consortium drug quality programme (ACTc-DQP) found that although no drug sold as ACT in Ghana was found to be fake (or to lack an artemisinin partner drug), 37% of samples were classified as substandard [
68]. Whilst this raises the possibility that one or both courses of AL administered to the patient in Ghana contained insufficient or degraded drug, the possibility that the patient received
no therapy until re-treated in Accra appears unlikely, and thus supports the timeline as presented.
The varying hypotheses presented above show that the cause of BWF remains to be fully understood. The authors have been unable to identify literature suggesting a direct link from persistent submicroscopic parasitaemia to clinical BWF and suggest that PCR methods should be undertaken prospectively in cases of BWF and treated controls to determine the frequency of this finding which, if universal, may point to an as yet unidentified mechanism involving the organism itself rather than simply the human response to infection and treatment. Recent reports of a rising incidence of BWF in endemic areas following the introduction of ACT [
69] suggest that this research should be a priority.
Authors’ contributions
PJH, CHN and JH managed the patient in hospital. JMC performed the morphological and molecular studies. PJH and JMC drafted the manuscript. RTPVL approved the study. All authors read and approved the final manuscript.