A systematic review of the clinical presentation, treatment and relapse characteristics of human Plasmodium ovale malaria

Screening, selection and data extraction were performed independently by the first and the second author. Disagreements and uncertainties at any stage of the process were discussed and resolved by consensus. If needed, the last author was consulted for a final decision. Only English, German and French articles were included in this analysis unless there was clear indication for relevant information of publications in other languages. Full texts of potentially relevant articles were obtained and articles from other sources were included in the pool of articles. Articles were matched to three different categories: treatment, severity and relapse.

Plasmodium ovale mono-infection of a human subject, defined by diagnosis based on microscopy and/or polymerase chain reaction (PCR) was a strict eligibility criterion for all categories. Additionally, separate definitions as follows were applicable for each category. Case reports were not used for treatment evaluation, otherwise there were no inclusion restrictions regarding types of studies. For in vitro studies, only assays with interpretable results were considered. Severe P. ovale malaria was determined on the basis of the 2014 WHO criteria for severe falciparum malaria [14] and other serious or life threatening clinical conditions as defined by the authors. As parasite counts are generally lower in severe P. ovale than in severe P. falciparum malaria [7] no threshold was determined for parasitaemia. A relapse was defined as a reappearing P. ovale parasitaemia following an initially diagnosed and adequately treated P. ovale “primary infection” (regardless of 8-aminoquinoline application) and subsequent permanent residence in a non-endemic country. The term “primary infection” was used to describe the first reported P. ovale infection in the article, which was adequately treated (important for malaria infection studies, where patients were mostly not treated in case of self-limiting infection). The period between primary infection and relapse and two relapse events, respectively, was counted as time between the date of treatment and the first mentioned date of reappearance. In order not to confuse delayed primary attacks with relapses, articles where the primary infection was not explicitly stated to have been a P. ovale infection were excluded.

Primary outcomes were adequate clinical and parasitological response on day 28, frequency of severe complications and the number of reported true relapses. Secondary outcomes were to obtain relapse characteristics, treatment regimen used, parasite clearance time (PCT), fever clearance time (FCT) and treatment outcome.

References were compiled in EndNote X6 (Thomson Reuters) and extracted data was collected in a standardized Microsoft^® Excel^® 2013 datasheet. Descriptive statistics were performed using IBM^® SPSS^® Statistics 20. Applicable risk of bias was assessed in applicable studies using the Cochrane collaboration’s tool for assessing risk of bias in combination with the methods guide for comparative effectiveness reviews [15, 16]. To assess overall quality of reporting an evaluation tool was created uniformly for all included study designs following the study quality assessment of the case series studies tool of the National Institute of Health [17].

The search yielded 3454 publications. After elimination of duplicates and screening of available titles and abstracts for relevance, 212 articles were selected for full review (Fig. 1). Two articles were added from other sources and a total of 36 articles met the required criteria. Of this pool, two articles reporting severe cases were excluded due to incomplete data and one because of double reporting, leaving 33 articles for data extraction.

No report was rated as having a low risk of bias due to the underlying study design. There were many case reports which made the systematic review especially vulnerable to selection bias and publication bias. To deal with publication bias, results from Clinical Trial Registers were included. As data were not used for a meta-analysis, missing data items did not influence individual risk of bias assessment. Individual risk of bias within studies as well as completeness of reporting are given in the Additional files 1, 2 and 3.

The literature search yielded five prospective studies evaluating treatment for P. ovale in a total of 58 participants. Baseline characteristics are outlined in Table 1. One study was conducted in Indonesia [18], two in Cameroon [19, 20], one in Gabon [21] and one in France on returnees from sub-Saharan Africa [22]. One trial was exclusively designed for P. ovale infected individuals [19]. Artesunate, atovaquone, chloroquine, mefloquine and pyronaridine were used as study drugs. Two prospective clinical trials with 13 participants in total chose chloroquine as study drugs [18, 21]. In general, sample sizes were small and control groups were missing in all 5 prospective studies. In fact, the largest study recruiting 30 patients evaluated artesunate therapy. Although the authors classified it as randomized trial, neither a placebo group nor a second treatment arm were described [19].

The longest follow-up period was 28 days, therefore, treatment success could not be obtained for days 42 and 63. Besides Siswantoro et al. (eight male, three female) [18], no publication reported the participants’ sex. For further details see Table 2. Two clinical trials additionally performed in vitro drug sensitivity testing. Interpretable assays showed no resistances of P. ovale against amodiaquine, artesunate, chloroquine, mefloquine, piperaquine or pyronaridine [18, 20].

Twenty two cases of severe P. ovale malaria were identified in scientific literature. Nigeria was the most commonly reported place of potential infection in travel histories (4 times) followed by Ghana, Cameroon and the Democratic Republic of Congo (3 times), and Ivory Coast and Niger (twice). The only non-African country reporting a complicated disease course was India (once). Mean age was 35.8 ± 13.6 years standard deviation (SD), with a range from 17 to 75 years. Fourteen cases were male, six female, for two sex was unknown. In Table 3, baseline characteristics are displayed in more detail. In 15 cases, P. ovale was diagnosed by microscopy. Seven patients were diagnosed by microscopy and PCR, out of which 2 cases were microscopically negative with a positive PCR result [24]. Species specific PCR was performed for four cases. Two were positive for P. ovale curtisi [8, 24], 1 for P. ovale wallikeri and for 1 species differentiation could not be deducted from the article [24, 25].

For the 22 patients with severe clinical conditions, 15 different features of severity could be identified. Taking the patients together, 35 severe conditions were reported. Acute respiratory distress syndrome (ARDS) was reported in five patients and therefore was the most prevalent severe condition. It was followed by anaemia with a hemoglobin level <7 g/dl, and pulmonary edema which occurred in 4 patients. 5 of the reported cases died and 3 patients had organic sequelae, however, 64% of the reported cases (n = 14) survived without sequelae. The majority of deaths occurred following onset of ARDS. Further details are displayed in Table 4.

Besides the clinically severe cases described above, two independent cases of congenital P. ovale malaria were identified presenting with severe anaemia [26, 27]. The two mothers (both secundigravidae) had resided in an African country prior to birth but gave birth to their children in Europe and also remained there during the observation period. Both had a history of treated malaria of unknown species. The respective children were delivered by Cesarean, one because of a treated human immunodeficiency virus (HIV) infection of the mother, the other one as an emergency cesarean section. Being healthy at birth, malaria was diagnosed 5 and 3 weeks post-partum. Detailed information is presented in Table 5.

From the description of P. ovale as distinct species in 1922 up to 2015 a total of 18 cases with potentially relapsing P. ovale parasitaemia according to the inclusion criteria applied for this systematic review were reported in scientific literature. These patients were described to have experienced a total of 28 potential relapse events. 4 cases (22%) occurred in tourists, 14 (78%) in malaria infection studies. Sex was specified in 44% of the patients, all of them were male. Fever was mentioned in five episodes, other clinical information about relapse characteristics was missing. The most commonly used drugs to treat primary infections and relapses were chloroquine and quinine sulfate. Median time between primary infections and first potential relapses was 17 weeks (min–max 2–60 weeks). The median time between first and second potential relapse was also 17 weeks, ranging from 5 to 72 weeks. The time between second and third relapse was not reported. Six relapses occurred despite previous primaquine treatment. Eight individuals presented with two relapses and one individual relapsed three times. Details can be found in Tables 6 and 7.

Chloroquine has been the recommended treatment for P. ovale malaria for many years. In the latest guideline for the treatment of malaria, the WHO strongly recommends to treat P. ovale and other non-falciparum Plasmodium species with artemisinin-based combination therapy or chloroquine on the basis of “high-quality evidence”. Following elaborations of underlying studies in the WHO guideline however rather break this down to experience [38]. In this systematic review, no high-quality studies supporting current treatment recommendations were identified. Not a single randomized controlled clinical trial on P. ovale malaria has been published in scientific literature. This finding is supported by a report by Visser et al. [37]. Although chloroquine has been tested in small prospective uncontrolled trials, one might question whether this small number of participants and a lack of control groups in all studies provide enough evidence for unequivocal treatment recommendations. Summing up all published reports and clinical experience, it becomes evident that anti-malarial drugs employed for P. falciparum are also effective for P. ovale. However, scientifically sound evidence for this is currently missing.

In 1932, James and coworkers stated that it was unlikely that another malignant species besides P. falciparum would be discovered [39]. Since then Plasmodium knowlesi was found to be infective for humans leading to life-threatening quotidian malaria. Also the previously considered benign malaria species P. malariae, P. vivax and P. ovale were reported to cause severe disease and even death in a small minority of patients. To date little is known on the specific pathogenesis of severe diseases in these non-falciparum malarias. The results of this systematic review support this understanding.

It is of interest that ARDS was the main feature of severe disease in P. ovale malaria as it was described in returning travellers with P. vivax malaria [40]. The potential coincidence that the two patients with a history of tuberculosis 10 years and more ago both died from ARDS raises the question whether a preexisting pulmonary condition may be a risk factor for respiratory complications of P. ovale infection [41, 42]. Anaemia was also reported as a feature of severe P. ovale malaria, however due to its multifactorial aetiology it is difficult to attribute this with confidence to P. ovale infection. Nevertheless it has been reported concordantly in paediatric patients with P. vivax infection in Asia [43].

An important limitation in the description of severe cases of P. ovale infection is the only partly performed molecular assessment of blood samples. Although light microscopy forms the current gold standard for malaria diagnostics, its sensitivity is inferior to most molecular methods. Additionally, species determination and distinction, especially between P. ovale and P. vivax can be challenging most notably in low parasitaemic smears [44, 45]. It is, therefore, not possible to entirely exclude the possibility of coinfection with other Plasmodium species including P. falciparum in these cases.

Congenital malaria is a rare finding in non-endemic countries. Even more surprising was the identification of two cases of congenital P. ovale malaria with severe anaemia in Europe. Both mothers had been living in an endemic country in the past. Interestingly, one of the infants was born to a HIV positive mother. An association between HIV and the incidence of P. falciparum in pregnancy has already been shown [46] and it might be speculated that the same is true for P. ovale. In 2008, Vottier et al. reported another congenital P. ovale infection transmitted by an HIV positive mother which was however not severe [47].

Although the concept of hypnozoite-induced relapse in all tertian malarias seemed as a unanimous concept until recently, molecular evidence supportive for this model is scarce. A recent experimental study in mice engrafted with human hepatocytes observed uninucleate parasitic structures measuring ~5 µm (day 8) and late schizonts (day 21) after P. ovale sporozoite inoculation [48, 49]. The description of these histological structures resembles the findings of Krotoski described for Plasmodium cynomolgi bastianelli in Rhesus monkeys (average diameter 4.5 µm) and for P. vivax in chimpanzees (approximately 4–5 µm diameter) [50, 51]. However, this analogy does not constitute proof that these uninucleate structures truly represent hypnozoites or rather retarded forms. Furthermore it does not provide evidence for these structures to cause relapse events [48]. Based on this lack of firm experimental evidence and the scarcity of clinical reports a recent perspective article challenged the current concept of P. ovale relapse caused by liver hypnozoites proposing a gradual dormancy concept [52].

The presence of dormancies as such can be assumed as data from malaria elimination settings suggests their important role for sustained malaria transmission, along with P. vivax [53].

Plasmodium ovale hypnozoites have not yet been unequivocally demonstrated in the human host. As evidenced by this systematic review a total of 18 reported cases of P. ovale relapse in nearly 100 years do not provide solid evidence for the current relapse theory. On the other hand, experiments and malaria treatment of neurosyphilis patients have shown that in case of repetitive inoculation with the same strain, immunity to this homologous challenge develops fast and subsequent infections remain often asymptomatic [54, 55]. Hence, it may be speculated that a true relapse may lead to mitigated symptoms or may even be sub-clinical.

In this context, it is of interest to note that six potential relapses occurred despite previous primaquine treatment. However, intake of primaquine has not been evaluated in these patients.

Historically, the concept of treatment of P. ovale relapses with an 8-aminoquinoline is based on the observation that quinine and pamaquine (the first synthetic 8-aminoquinoline) together were more effective in the treatment of certain malaria cases than quinine alone. When Sinton and Bird observed that pamaquine reduced the relapse rate of P. vivax malaria [56] several 8-aminoquinoline derivatives were synthesized and tested for this purpose. Primaquine finally showed a higher anti-relapse effect than pamaquine with reduced toxicity among the most promising substances, but effectiveness for P. ovale relapses has since then only been presumed and never demonstrated [57]. Importantly, from a methodological point of view, to prove the effectiveness of a medication it is necessary to first unequivocally demonstrate the existence of the condition to be treated—in this case hypnozoite-induced relapse.

Richter et al. questioned the existence of relapses in P. ovale in a review in 2010. They stated that “it may be difficult to differentiate a true relapse from a primary malaria attack with a long latency” [10]. To overcome that difficulty, the analysis was restricted to cases, which did not reside in a malaria endemic area between the occurrences of primary infection and relapse. In addition, the species of the primary infection had to be explicitly mentioned to be P. ovale and treated with anti-malarial chemotherapy. Comparing the results of this systematic review with those of Richter et al. [52], these strict criteria are the main reason why even fewer cases of potential relapses were observed here.

Finally, only one potential case of relapse that was investigated with molecular methods could be identified in the literature [58, 59]. As this case occurred in an endemic area, the report did not fit the criteria of this systematic review and was therefore not included in the primary analyses. After personal communication with one of the authors (Fuehrer) the confirmation for this potential relapse case was based on the sequence homology of partial cox1, SSU rRNA, and porbp2 loci between the primary and the potential relapse isolate [60]. These markers are usually not used for intraspecific distinction but for differentiation between the species. The multigene approach, however, enhances the significance of the result. In summary, the identification of highly sensitive genetic markers or techniques that can discriminate between hypnozoite induced relapse and other sources of recurrent infections is still a work in progress.

Limitations of this systematic review are the low strength of evidence of the included studies based on their study design. At the same time, they form the only available evidence to address the review questions and form the basis of current recommendations.