This is the largest prospective study to date on the two species of infection by imported P. ovale trying to overcome recruitment difficulties and limitations of previous retrospective studies.
On the results obtained, significant differences were found in the epidemiological characteristics of sex, ethnicity and type of patients between both groups. Ethnic differences may be the main confounding factor since there is a greater probability that African patients retain some semi-immunity against P. ovale spp. infection (excluding VFR travellers), which would generally result in less clinical and analytical involvement [20]. On the other hand, the difference in sex distribution and type of patients seems to be clearly related to the ethnic origin since in a subgroup analysis, 16 out of 18 (89%) of the patients with P. o. wallikeri who travelled by international cooperation were male and Caucasians, which would bias the characteristics of this group.
This distribution could be a result of chance alone or due to the fact that actually Caucasians patients are more susceptible to infection by wallikeri parasites. However, a previous retrospective study did not show these differences in sex, ethnicity or type of patient [5] and the few studies of imported ovale malaria reporting a significant number of patients showed either equality of sexes (but did not report on ethnicity) [13] or they were nearly all Chinese men who worked in African countries [10].
When performing the multivariate study adjusted by ethnicity to reduce this confounding factor, the number of patients with thrombocytopaenia and two of the analytical alterations remained significant. Thrombocytopaenia was more pronounced in P. o. wallikeri and INR values were higher in the wallikeri group. The comparation of the rest of analysis variables were not finally significant, over all those that indicate a greater degree of haemolysis, such as LDH, tBR or haemoglobin level.
In the case of thrombocytopaenia, these results confirm the main finding of a previous retrospective study where the only difference was more severe thrombocytopaenia in P. o. wallikeri infection [5]. Thrombocytopaenia is a common finding in patients with malaria of all Plasmodium species ranging from 24 to 94% of indicence, although spontaneous haemorrhages are infrequent and limited to very severe cases [21]. Also it is of note that thrombocytopaenia seems to be a very common finding in P. ovale spp. infection. Of the 79 patients analysed, 53 (67%) had under 150,000 platelets/µL, which corresponds to our previous series findings of 66.6% in 16 imported ovale infection patients [22] and a retrospective study where 26 out of 35 patients (75.3%) suffered from thrombocytopaenia [5]. Only one patient in each group showed under 50,000 platelets/µL again without any haemorrhagic feature.
The pathogenesis of thrombocytopaenia in malaria is not fully understood but is accepted that encompasses several different mechanisms that lead to increased platelet destruction or consumption, such as increased attachment to endothelium and adherence to Von Willebrand factor, clumping and agglutination of infected and uninfected erythrocytes, consumption into the coagulation process and haemolysis, increased diffuse platelet sequestration, decreased nitric oxide bioavailability, and immune complexes-mediated destruction [23].
Although thrombocytopaenia is not among the severity criteria of the WHO, evidence is accumulating that it can predict an adverse outcome, which seems to be driven by a greater severity of illness in P. falciparum or P. vivax [21, 24] and severe thrombocytopaenia can identify an increased risk of death from falciparum or vivax malaria [25]. Therefore, in the case of P. o. wallikeri it could indicate a greater intrinsic pathogenicity of this species compared to P. o. curtisi, although without evidence of any major significantly worse clinical criteria of severity.
Regarding the other significant finding of this study, the higher INR elevation in P. o. wallikeri compared to P. o. curtisi, this increase was always mild and in no case exceeded the value of 1.54 nor was related to any bleeding. On the other hand, the APTT values were similar in both groups. No other coagulation factors, fibrinogen levels, fibrin degradation products or d-dimer were measured that could argue in favour of a significant coagulation disorder.
It is well known that P. falciparum malaria is associated with significant coagulation activation. The increase in INR or its equivalent, prolongation of prothrombin time, is produced by the activation of the extrinsic pathway of the coagulation cascade and can lead to disseminated intravascular coagulation in the most severe cases of malaria. Plasmodium falciparum seems to trigger coagulation activation through multiple different pathways although the mechanisms involved are not well understood [26]. Much less information on coagulation is available in P. ovale spp. infection. In the case of INR, the finding of this study might support the hypothesis of a more pathogenic wallikeri parasite, but does not seem to pose a significant clinical difference or severity.
In the initial analysis, no differences were found in the latency period, but when the multivariate analysis was performed a significant shorter time appeared in P. o. wallikeri compared to P. o. curtisi infection, but only in Caucasians patients. This difference is more significant since only one of the seven Caucasians patients who suffered infection by P. o. curtisi had taken adequate prophylaxis versus 11 out of 22 (50%) of the wallikeri group, which should have lengthened their latency. The retrospective study by Nolder et al. in the UK also found significant shorter latency period in wallikeri cases but did not report the ethnicity or patient characteristics [13]. On the contrary, another study in Chinese people (ethnically homogeneous but with unknown previous exposure to malaria or chemoprophylaxis use) encompassing 109 P. o. curtisi and P. o. wallikeri [10] imported from Africa, did not find any difference in latency. In a previous retrospective study, only a trend to shorter wallikeri latency was found (p = 0.07) [5].
These results of different latency times in only Caucasians patients suggest a strong influence of the previous partial immunity to malaria in the time of symptoms of P. ovale spp. infection. Finally, note that in both species latencies greater than 1 year have been found as described previously [11].
A complete chemoprophylaxis was reported in 22.8% of ovale spp. infections and up to 35% in P. o. wallikeri, a high percentage if compared with data from large series of imported malaria as in USA 2014, where only 7.8% of travellers correctly took prophylaxis [11]. If analysed by Plasmodium species, these results are similar to the study by Nolder et al. which stated that the proportion of ovale malaria which occurred in patients reporting chemoprophylaxis use (33%) was significantly higher than for P. falciparum (6.4%) and P. vivax (23.7%) [13]. Also, in a USA and Israel study, up to 73% of imported P. ovale spp. used an effective prophylaxis [27]. Moreover, in a recent Spanish study considering only the subgroup of travellers with P. ovale/P. vivax malaria, these patients had also taken chemoprophylaxis significantly more frequently than those with non-P. ovale/P. vivax malaria (35.5% vs. 2.7%) [28]. These results reinforce the idea that current prophylaxis does not adequately act on P. ovale spp. infection, since it is ineffective against its hypnozoites and, therefore, does not prevent a delayed primary attack or true relapses from occurring when there is no prophylactic medication in blood.
Although the information cannot be completely reliable, over 40% of patients reported an episode of malaria in the previous 12 months. This could have been a primary infection by P. ovale, another ovale relapse, a mixed infection with other Plasmodium or simply a non-ovale malaria unrelated to the current episode. Since it is not feasible to obtain a previous diagnostic sample, it is impossible to know if the current episode represented a primary infection or a relapse by P. ovale. Primary infection could only be diagnosed in the cases of first malarial episode and very short latency period.
Regarding the underlaying conditions, it is worth focusing on diabetes mellitus and drepanocytosis. The prevalence of diabetes mellitus is increasing worldwide, especially in low income countries, as are found in sub-Saharan Africa [29]. The number of diabetics in this study is as high as 6.32% (5/79) in a population with a low average age, especially in wallikeri group. In a retrospective curtisi–wallikeri study up to 8.5% of patients suffered from diabetes [5] and in a previous study of imported P. ovale infection 3 out of 16 patients (18.7% with mean age 30.6 years) were also diabetics [22]. A recent Swedish study, reported diabetes in 3.5% of 937 patients with imported P. falciparum and 2.5% in 398 imported malaria in a Spanish Tropical Medicine Centre [30]. In a series of 229 P. falciparum infections from 2006 to 2015 in Hospital Universitario Príncipe de Asturias (HUPA), 2.6% were diabetics (unpublished data). As diabetes is a known risk factor for malaria due to P. falciparum in endemic areas [31], the higher prevalence of this chronic disease in P. ovale studies might indicate that would also be a risk factor, and deserves further investigation to confirm this finding.
In endemic areas, carrying the sickle cell trait represents a risk factor for infection by P. ovale spp. [20] and a partial protective factor against severe P. falciparum infection. In this study, 2 out of 47 African patients (4.2%) had drepanocytosis. Moreover, in the previous comparative retrospective study, 8.3% (2 out of 24) showed drepanocytosis and up to 23% (3/13) in the series of 16 imported P. ovale spp. [22]. Gathering African patients from the three studies (7/84) and comparing to 1.35% of patients in 221 African people with imported P. falciparum in HUPA during the period 2006–2015 (unpublished data) significant difference can be found (p < 0.005) that would support the results of the study conducted in endemic area.
As in previous studies, no differences were detected in parasitaemias between both groups [6, 9, 32]. Most of them had low parasite counts clearly below 1%, making it even more difficult to find statistically real differences. Surprisingly, two patients reached 2% parasitaemia, a Caucasian traveller without any immunity to malaria and a child with drepanocytosis and intense anaemia that enhanced the proportion of infected red blood cells. Such high parasitaemia has been rarely described, sometimes linked to serious complications such as spleen rupture [33]. Finally, the absence of Schüffner’s stippling in blood smear, which has been described as a possible feature specific to P. o. wallikeri, was not studied.
There was also no difference in the sensitivity of RDTs, which was low as described in the literature [2] without exceeding 45% in either group. Almost 20% of the cases in both groups were diagnosed only by PCR, as RDT and/or thick blood smears were negative. In endemic areas this percentage is much higher, especially in asymptomatic patients [3]. Therefore, if the clinical and epidemiological suspicion is high, it would be convenient to repeat the diagnostic tests and request molecular PCR techniques not to miss P. ovale spp. diagnosis.
This study confirms that this infection can cause malaria with severity criteria or complications, in this case 7.6% of all patients with no differences between both species. In series of imported malaria by P. falciparum the rate of severe cases was around 2–17% [11, 34], but with a much higher incidence of life-threatening complications, such as ADRS, severe anaemia or cerebral malaria than described in P. ovale spp. infection [8]. A rare case of haemophagocytosis was included that had been described only in P. falciparum and P. vivax infection [35], but not in P. ovale spp. Analysing the six patients with severe malaria or complications, two had drepanocytosis (which also produces anaemia) and one was diabetic, which might indicate the possible importance of these risk factors in the severity of P. ovale spp. infection.
In cases of malaria with tBR > 3 mg/dL, the WHO severity criteria add parasitaemia levels above 100,000/µL in P. falciparum and > 20,000/µL in P. vivax and Plasmodium knowlesi, but is not known in the case of P. ovale spp. In this study, all four cases with tBR > 3 mg/dL had parasitaemias over 4000/µL. In addition, four patients received parenteral artesunate, two of them with tBR > 3 mg/dL and one with unusually high parasitaemias of 2%. It is described that host response may reach full strength at lower parasitaemia in P. vivax and P. ovale spp. infection than in P. falciparum [36] so, perhaps it would be also convenient to establish a threshold of parasitaemia with jaundice and hyperparasitaemia in P. ovale spp. in the face of treatment decisions and outcome.
The treatments were very varied, including artemisinins alone or in combination, atovaquone/proguanil, chloroquine or quinine-doxycycline, all with good initial clinical evolution, which is compatible with the sensitivity of P. ovale spp. described for multiple antimalarials [8]. Over 70% of them received primaquine with good tolerance, although there was no long-term follow-up recorded to assess episodes of relapse or late recrudescence.
The prospective study design has minimized some of the limitations of the previous retrospective study but some of them still remain. First, although it is one of the largest studies carried out worldwide, the number of patients may still lack sufficient statistical power to show other differences between infections with both P. ovale species. Second, only strains of P. ovale from Africa, and patients from Africa and Europe were analysed; a study of infections and patients from Asia, Oceania or other places might show different results. Third, the behaviour of these species in endemic countries may be different but it would be difficult to gather a significant number of patients with P. ovale spp. monoinfection due to the high prevalence of mixed Plasmodium infections and malaria reinfections. Fourth, the date of infection is a minimal approximate and it is virtually impossible to distinguish a primary infection from a relapse. Last, a mix of ethnicities, non-immune and semi-immune patients led to heterogeneous study groups, although the multivariate statistical study has reduced this bias.