Since P. knowlesi malaria causes severe disease in approximately one of 10 patients and can be potentially fatal
, the search for reliable and unsophisticated markers of disease severity that complement the diagnosis of malaria needs to be intensified. The objectives of the study presented here were to test for an association between pigment-containing neutrophil counts and risk of disease severity and to compare with parasitaemia and platelet counts as predictors of severe P. knowlesi malaria. Pigment-containing monocytes are also a marker of severe P. falciparum malaria
. However, due to the low number of monocytes found in thin blood films, a count of 100 monocytes is highly time consuming and was considered as an impractical procedure for routine use and was therefore not performed here.
A strong association between PCN count and risk of disease severity was observed. The method involved counting 100 neutrophils, took 15–20 minutes and requires skilled personal to identify the correct cell morphology. It is both labour-intensive and time consuming in the routine malaria diagnostic setting. Therefore, the PCN count should have a superior prognostic ability over parasitaemia and platelet counts if it is to be recommended for routine use. Even though the AUC estimate for PCN was the highest among all parameter estimates, PCN counts did not have a higher diagnostic precision when compared with parasite and platelet counts. The lack of statistical significance might be explained by the low sample size. On the other hand, linear regression analysis revealed that the total variation in the amount of pigment-containing neutrophils can be explained by the changes in peripheral parasitaemia to an extent of up to 75%, leaving only a minor proportion in the total variation that might represent sequestration and would give additional prognostic information in P. knowlesi malaria. In any case, the observed differences in the AUC estimates are rather small and cannot justify a recommendation of the PCN counts as a routine test at the moment. Flow cytometry can achieve a fast and effective detection of pigment-containing WBCs
[15, 21] but remains a future approach in most settings.
In contrast, the parasite count on thick and thin blood films is a simpler method and is usually available in resource-poor settings. The peripheral blood parasitaemia on thick blood films has been shown to be a potential marker of severity in P. knowlesi malaria
. The authors of the same study revealed an inverse correlation between parasitaemia and platelet count and suggested that the latter could also be a predictive factor. The number of severe cases in that study was small and the findings were confirmed here, even if the platelet count seems to have a lower accuracy in predicting severe disease when compared with PCN counts and parasitaemia.
Assessing cut-offs with a more predictive than reflective capability was based on an appropriate relationship between positive and negative predictive values. Recommendation of a specific cut-off required a negative predictive value of at least 95% in the study. This would mean that one in 20 patients with a test result indicating non-severe disease or a low risk of developing complications would have severe disease at the time of measurement. None of the severity markers could reach high positive predictive values in the study, assuming a prevalence of 10% severe cases among hospitalized patients with P. knowlesi malaria. A positive predictive value of about 33% as calculated for the recommended cut-off of 35,000 parasites/μl means that two of three patients with a positive test indicating a severe disease or a high risk of developing complications would not suffer from severe malaria at the time of measurement. Managing them according to the WHO guidelines for severe malaria might appear as an overtreatment and potential waste of resources in an often resource-limited setting. However, when using the recommended cut-offs in multivariate logistic regression analysis any tested severity marker is independently associated with high risk of severity (OR 5.27 – 9.93). Therefore, having a positive test result in accord with the recommended cut-offs categorizes a patient to be at high risk of severity regardless of the actual clinical status. Management for severe malaria should consecutively be accomplished to prevent a potential development of severe disease in patients within this high-risk group even if they do not suffer from complications at the time of measurement. Consequently, the results of the study presented here indicate that any adult patient diagnosed with P. knowlesi malaria and a parasite count ≥35,000/μl or ≥1% or a platelet count ≤45,000/μl should be regarded at risk and should subsequently be treated as for severe malaria according to current WHO guidelines.
Using these cut-offs in areas with prevalence of both P. knowlesi and Plasmodium malariae may be an issue. Plasmodium knowlesi is often confused with P. malariae when making the diagnosis by light microscopy
. Plasmodium malariae infection causes a median parasitaemia of 8,875 parasites/μl with maximum of about 50,000 parasites/μl
, but rarely causes severe disease. P. knowlesi may be wrongly diagnosed in the presence of a P. malariae infection. In such cases, a parasite count ≥35,000/μl alone may lead to incorrect risk estimation although coupled with low platelet counts may be more specific. In Sarawak and Sabah, P. malariae infection is rare
[2, 24] and the benefit of properly diagnosing P. knowlesi would override misdiagnosing P. malariae in a small number of cases.
One major limitation of this study is the assumption that severe cases of P. knowlesi malaria occur with a prevalence of 10% among hospitalized patients. The prevalence and characteristics of P. knowlesi malaria should be measured within the framework of prospective studies in all locations in Southeast Asia where P. knowlesi malaria occurs
[3–6] and where validation of the findings presented here would be informative. Cut-offs might be adjusted according to the results gained in such studies. The analyses conducted here identify parasitaemia and platelet counts as measures of risk assessment for the development of severe disease in P. knowlesi infections. However, additional simple diagnostic tools to distinguish P. knowlesi from P. malariae infections in settings with limited resources are urgently needed to identify patients at risk of severe P. knowlesi malaria.