Hidden Biomass of Intact Malaria Parasites in the Human Spleen

Introduction While P. falciparum accounts for a majority of severe and fatal malaria cases worldwide, P. vivax is a major cause of morbidity outside of Africa, causing an estimated 70–390 million malaria cases per year [1]. Although previously considered benign, P. vivax is now recognized as capable of causing severe and fatal disease [2], [3], [4], [5], [6], [7], [8], [9], [10]. Despite this, little is known about the pathogenesis of severe disease in vivax malaria. In falciparum malaria, severe and fatal disease is characterised by cytoadherence of parasitized red blood cells (RBCs) to activated and dysfunctional endothelium, leading to parasite sequestration with microvascular obstruction [11], [12], [13]. As a result of sequestration the mature schizont stages of P. falciparum are rarely seen in peripheral blood [14]. Total parasite biomass is underestimated by circulating parasitemia quantitated by microscopy of peripheral blood, and is more accurately quantitated by plasma P. falciparum histidine rich protein-2 (HRP2). Pathogenesis and disease severity in falciparum malaria are both biomass-related: in contrast to peripheral parasitemia [15], [16], [17], plasma HRP2 is strongly and independently correlated with disease severity and mortality among both children [17], [18], [19], [20] and adults [21], [22]. Cytoadherence of infected RBCs to endothelial cells is 10-fold less in P. vivax infection than in falciparum malaria [23], and autopsy evidence for sequestration within the endothelium-lined microvasculature of vital-organs in vivax malaria is minimal [2], [4], [24]. This, together with the lower parasitemias resulting from preferential invasion of reticulocytes, is thought to account for the lower lethality of P. vivax [2]. The paucity of apparent endothelial microvascular sequestration in vivax malaria has led to the assumption that peripheral parasitemia reflects total parasite biomass. However, this assumption has been questioned [3], [25], [26], with more adhesive schizont-stages of P. vivax known to be under-represented in peripheral blood [26], [27]. Accumulation of P. vivax-infected RBCs in the spleen or bone marrow has been hypothesised [3], [25], [26] but not yet systematically investigated. In contrast to falciparum malaria, no study has evaluated the relationships between total parasite biomass, disease severity and systemic inflammation in vivax malaria. We propose that in vivax malaria, parasite lactate dehydrogenase (pLDH) and P. vivax-pLDH (PvLDH), produced by viable or recently killed parasites, may be used to estimate total parasite biomass. While plasma pLDH has been shown to demonstrate only moderate correlation with peripheral parasitemia in vivax malaria [28], a pLDH antigen capture enzyme-linked immunosorbent assay demonstrated a direct relationship between parasite production of pLDH and total P. vivax parasite concentration ex vivo, including all parasite stages, suggesting that pLDH reflects total P. vivax parasite biomass [29]. As with HRP2 in falciparum malaria [30], pLDH is produced to a greater extent by P. vivax schizonts and trophozoites than by ring-form parasites [31], and given the under-representation of mature P. vivax stages in peripheral blood [26], [27], pLDH may be a better marker of total parasite biomass and a better prognostic indicator than peripheral parasitemia. In falciparum malaria, clinical severity and mortality are also independently associated with impaired microvascular function [13] and with increased angiopoietin-2 (Ang-2) [32], a key product of endothelial Weibel-Palade Bodies (WPB) and an autocrine mediator of endothelial activation [33]. Despite the apparent paucity of endothelial microvascular sequestration in vital organs, P. vivax has been associated with greater endothelial activation than P. falciparum, with Ang-2 concentrations higher in non-severe vivax compared to non-severe falciparum malaria [34]. However the relationships between clinical severity, microvascular dysfunction and endothelial WPB exocytosis have not been assessed in vivax malaria. We tested the hypotheses that in vivax malaria, peripheral parasitemia would underestimate total parasite biomass, and that markers of total parasite biomass would be related to systemic inflammation and clinical severity. We also determined the relationship between vivax disease severity and endothelial activation and microvascular function, with patients with non-severe and severe falciparum malaria included for comparison. We found that total vivax biomass was underestimated by peripheral parasitemia, and was associated with both systemic inflammation and disease severity in vivax malaria. Severe vivax malaria was associated with increased endothelial activation and WPB exocytosis and with impaired microvascular function, comparable to that seen in severe falciparum malaria. Results Patients A total of 192 malaria patients and 74 healthy controls were enrolled. Malaria patients included 62 with vivax malaria (53 non-severe, 9 severe) and 130 with falciparum malaria (109 non-severe, 21 severe). The clinical and epidemiological features of 43 patients with vivax malaria (including 7 with severe disease) and 122 patients with falciparum malaria (including 13 with severe disease) have been previously reported [35]. Baseline characteristics are shown in Table 1. Among the 9 patients with severe vivax malaria, severity criteria included hypotension (n = 6), respiratory distress (n = 1), jaundice (n = 2), metabolic acidosis (n = 1), abnormal bleeding (n = 1) and multiple convulsions (n = 1). Three patients had 2 severity criteria and 6 had one. Pre-antibiotic blood cultures were negative in 4 patients, positive for Streptococcus pneumoniae in one [35], and not done in 4. No deaths occurred from either species. 10.1371/journal.ppat.1004558.t001 Table 1 Baseline characteristics of vivax and falciparum malaria patients and healthy controls. P. vivax P. falciparum Controls (n = 74*) Non-severe (n = 53) Severe (n = 9) P value (sev vs. ns) Non-severe (n = 109) Severe (n = 21) P value (sev vs. ns) Age, years median (IQR) 35 (23–44) 24 (18–39) 39 (30–52) 0.168 25 (17–39) 33 (19–45) 0.148 range 14–69 13–61 14–79 13–78 13–60 Males, n (%) 52 (70) 39 (74) 9 (100) 0.105 79 (72) 15 (71) 0.922 Weight, mean (sd); kg 58 (10) 56 (12) 58 (8) 0.584 56 (13) 66 (17) 0.002 Current smoker, n (%) 30 (41) 26 (49) 1 (11) 0.034 37 (34) 10 (48) 0.232 Fever duration, days; median (IQR) 5 (3–7) 4 (3–4) 0.135 5 (3–7) 7 (5–7) 0.014 Systolic blood pressure, mean (sd); mmHg 123 (15) 115 (16) 119 (18) 0.451 116 (16) 114 (17) 0.756 Pulse rate, mean (sd); beats/min 70 (12) 87 (19) 97 (14) 0.172 92 (18) 98 (16) 0.123 Respiratory rate, mean (sd); breaths/min 20 (3) 24 (5) 26 (4) 0.181 27 (6) 31 (7) 0.006 Temperature, mean (sd); °C 36.5 (0.4) 37.4 (1.2) 36.9 (0.5) 0.224 37.9 (1.1) 38.1 (1.4) 0.407 Time from malaria treatment to enrolment, hours; median (IQR) NA 5 (0 –– 11) 10 (6–15) 0.126 5 (0–12) 4 (0–10) 0.541 *includes all healthy controls who had Near Infrared Spectroscopy and/or blood analysis performed Peripheral parasitemia, schizontemia, markers of total parasite biomass and disease severity In patients with vivax malaria, the median peripheral parasitemia was 2.4-fold higher in severe compared to non-severe disease (10,243 parasites/µL vs. 4,209 parasites/µL; P = 0.021), while the median PvLDH was 3.7-fold higher (96.6 ng/ml vs. 26.2 ng/ml; P = 0.021) and pLDH 6.9-fold higher (308 ng/ml vs. 44.6 ng/ml; P = 0.015) (Table 2 and Fig. 1). After removing the patient with severe vivax malaria and concurrent bacteremia from the analysis, the median peripheral parasitemia was 1.8-fold higher in severe compared to non-severe disease (7,865 parasites/µL vs. 4,209 parasites/µL; P = 0.047), while the median PvLDH was 4.8-fold higher (125 ng/ml vs. 26.2 ng/ml; P = 0.002) and pLDH 7.3-fold higher (326 ng/ml vs. 44.6 ng/ml; P = 0.002). In falciparum malaria, median peripheral parasitemia and plasma HRP2 were 4.2-fold and 7.4-fold higher, respectively, in severe compared to non-severe disease (Table 2). 10.1371/journal.ppat.1004558.g001 Figure 1 Parasitemia (A) and total parasite biomass [pLDH (B), PvLDH (C), and HRP2 (D)] among patients with severe and non-severe vivax and falciparum malaria. 10.1371/journal.ppat.1004558.t002 Table 2 Parasite count and parasite biomass among patients with P. vivax and P. falciparum malaria. P. vivax P. falciparum Non-severe (n = 53) Severe (n = 9) P value (sev vs. ns Pv) Non-severe (n = 109) Severe (n = 21) P value (sev vs. ns Pf) Parasite count (parasites/µL) 4,209 (1700–10,165) 10,243 (4387–19,520) 0.021 10,500 (4014–32,267) 44,332 (9079–273, 909) 0.002 PvLDH (ng/ml) 26.2 (10.7–57.3) 96.6 (85.6–154) 0.012 PvLDH/Log parasitemia 3.06 (1.33–7.73) 10.42 (9.27–18.60) 0.027 pLDH (ng/ml) 44.6 (7.8–142) 308 (258–619) 0.015 50.23 (22.9–121) 86.79 (39.3–452) 0.016 pLDH/Log parasitemia 6.46 (1.60–20.1) 36.7 (27.9–58.3) 0.021 5.54 (2.33–13.13) 7.58 (3.51–65.65) 0.058 HRP2 (ng/ml) 72.8 (32.8–260) 541 (125–1590) 7.0 g/dL, and had not been previously enrolled in the study. Clinical details of patients enrolled from September 2010 – October 2011, in addition to details regarding excluded patients, have been previously reported [35]. Severe malaria was defined as the presence of ≥1 of: unrousable coma (Glasgow Coma Scale score 2) convulsions; respiratory distress (respiratory rate >30 breaths per minute and oxygen saturation 43 µmol/L plus parasitemia >20 000/µL [P. vivax] or >100,000 [P. falciparum] and/or creatinine >132 µmol/L); significant abnormal bleeding; hypoglycaemia (blood glucose 4 mmol/L); acute kidney injury (AKI; creatinine >265 µmol/L); hyperparasitemia (P. falciparum parasitemia >10%). Healthy controls were visitors or relatives of malaria patients, with no history of fever in the past 48 hours and with blood film negative for malaria parasites. Standardized history and physical examination were documented. Haematology, biochemistry, acid-base parameters, and lactate (by bedside blood analysis; iSTAT system) were obtained on admission. Parasite counts were determined by microscopy, and parasite species were identified by PCR [68], [69]. Patients were treated according to hospital guidelines, as previously described [35]. Endothelial activation, WPB release, ADAMTS13 and cytokines Venous blood collection in lithium heparin was centrifuged within 30 minutes of collection and plasma was stored at −70°C. Plasma concentrations of the endothelial activation markers ICAM-1, E-selectin, and Ang-2 were measured using ELISA (R&D Systems), and IL-6 and IL-10 were measured by flow cytometry (BD Cytometric Bead Array). Antigen concentrations of vWF and the vWF-cleaving enzyme, ADAMTS13, were measured in platelet poor plasma using ELISA (America Diagnostica), and ADAMTS 13 activity by fluorescence resonance energy transfer (FRET), as previously described [70]. Parasitemia and markers of total parasite biomass Peripheral blood parasitemia was quantitated per 200 white blood cells by microscopy and expressed as parasites/µL based on automated white cell count. Parasite stage distribution was quantitated on microscopy of peripheral blood. Plasma HRP2 (for P. falciparum) [32], genus-specific pLDH (for P. falciparum and P. vivax) and PvLDH (for P. vivax) were measured by ELISA [28] as proxies for total parasite biomass. Ratios of plasma pLDH and PvLDH to peripheral parasite density were expressed as ngs/1000 peripheral parasites, with the denominator log-transformed. PvLDH production in vitro To confirm evidence of PvLDH secretion across the parasite life-cycle, four cryopreserved P. vivax isolates with a high proportion of ring-stages were thawed and cultured ex-vivo as previously described [71], over 48–54 hours. Starting parasitemias were 3623/µL (98% rings), 10,605/µL (98% rings), 19,342/µL (65% rings), and 24,680/µL (98% rings). Thick and thin films were prepared at serial time points for stage differential and culture supernatant sampled for concentration of PvLDH. A progressive increase in concentration of PvLDH in culture supernatant across the parasite life-cycle was demonstrated ( Text S1 ), indicating secretion of PvLDH by all parasite stages. In vitro parasitemia remained stable throughout the culture duration. Microvascular function Microvascular function was assessed using Near Infrared Spectroscopy (InSpectra 650, Hutchinson Technology, Hutchinson, MN) which uses a probe applied to the thenar eminence to noninvasively measure microcirculatory oxygenation (tissue oxygen saturation; StO2) before and after an ischaemic stress, as previously described [13]. To induce an ischaemic stress, a vascular cuff was inflated to 200 mm Hg for 5 minutes and then rapidly deflated. Microvascular function was defined as the gradient of the StO2 recovery slope from release of the vascular cuff until StO2 had reached 85% of the baseline value. Statistical methods Statistical analysis was performed with STATA software (version 10.1; Statacorp, College Station, TX, USA). For continuous variables intergroup differences were initially compared using analysis of variance or Kruskal-Wallis tests depending on distribution. Student's T-test or Mann-Whitney tests were used for post-hoc pair-wise comparisons. Categorical variables were compared using χ2 or Fisher's exact test. Associations between continuous variables were assessed using Spearman's (ρ) or Peason's (r) correlation coefficients, depending on distribution. Ethics statement The study was approved by the Ethics Committees of the Malaysian Ministry of Health and Menzies School of Health Research. Informed written consent was provided by all participating adults, and by the parent or guardian of any participant aged <18 years. Supporting Information S1 Text In vitro production of PvLDH (S1A Fig.) and percentage of schizonts (S1B Fig.), and percentage of rings, trophozoites and schizonts (S1 Table). (DOCX) Click here for additional data file.

In acute malaria, red blood cells (RBCs) that have been parasitized, but no longer contain a malaria parasite, are found in the circulation (ring-infected erythrocyte surface antigen [RESA]-RBCs). These are thought to arise by splenic removal of dead or damaged intraerythrocytic parasites and return of the intact RBCs to the circulation. In a study of 5 patients with acute falciparum malaria who had previously undergone splenectomy, it was found that none of these 5 patients had any circulating RESA-RBCs, in contrast to the uniform finding of RESA-RBCs in all patients with acute malaria and intact spleens. Parasite clearance after artesunate treatment was markedly prolonged, although the parasites appeared to be dead and could not be cultured ex vivo. These observations confirm the central role of the spleen in the clearance of parasitized RBCs after antimalarial treatment with an artemisinin derivative. Current criteria for high-grade antimalarial drug resistance that are based on changes in parasitemia are not appropriate for asplenic patients.

Introduction Histidine-rich protein 2 (PfHRP2) is a 30-kDa protein produced by Plasmodium falciparum [1,2]. PfHRP2 is found in the parasite cytoplasm or parasite food vacuole, where it may play a role in the polymerization of heme to hemozoin [3,4]. It is also found in the host erythrocyte cytoplasm and red cell membrane [5]. PfHRP2 is released from infected erythrocytes as a water-soluble protein both in vivo and in vitro [6,7]. Sequestration of parasitised erythrocytes in the second half of the asexual life cycle (mature trophozoite and schizont stages) in the microvasculature of vital organs compromises the microcirculation, and is a central feature in the pathogenesis of falciparum malaria. Peripheral blood parasitaemia is very widely used to assess disease severity in malaria, but it is only a weak predictor of mortality in falciparum malaria, as the less pathogenic circulating stages can be counted whereas the more pathogenic sequestered mature parasitised erythrocytes are not seen and therefore not counted by the microscopist. These sequestered parasites secrete PfHRP2 into the plasma, and PfHRP2 is liberated at schizont rupture. The plasma concentration of this protein might therefore provide a better estimate for the patient's total parasite biomass, and hence be an accurate prognostic indicator. In a recent study we measured PfHRP2 quantitatively in synchronized P. falciparum cultures, and showed that approximately 89% of PfHRP2 is liberated at schizont rupture and that the variation in the amount released is limited [8]. In the current study we quantified plasma PfHRP2 concentrations in patients with falciparum malaria and, using a simple mathematical model, applied this approach to estimate the total body parasite biomass. We then related this to parameters of disease severity and outcome. Since the developmental stage distribution of circulating parasites also provides information on the sequestered parasites [9], this was also evaluated in relation to plasma PfHRP2 levels in these patients. Methods Patients In total 170 patients with uncomplicated malaria and 167 patients with severe malaria who participated in antimalarial treatment studies in Sangklaburi Hospital in western Thailand and in Mae Sot Hospital in northwestern Thailand were included in the study. These are low, seasonal transmission areas. Severe disease is seen in patients of all ages. Disease severity was classified using a modification of definitions employed by Hien et al. [10]. Patients were classified as having severe malaria if they met one of the following criteria: (a) Glasgow Coma Score (GCS) 42.8 μmol/l and with parasite count > 100,000/μl, (d) serum creatinine > 265 μmol/l with urine output 10%, (h) peripheral venous lactate > 4 mmol/l, or (i) peripheral venous bicarbonate 42.8 μmol/l), and two had renal failure (serum creatinine > 265 μmol/l). Plasma PfHRP2 concentrations were measured on admission, and then on days 1, 3, 7, 14, 21, and 21 after admission. The data were analysed using the WinNonlin statistical package (Pharsight, Mountain View, California, United States). There was no evidence of concentration dependence, and a single exponent term gave the best fit to the decline in plasma PfHRP2 concentrations with time (i.e., elimination of PfHRP2 was a first-order process). The mean (standard deviation) plasma elimination half-life (t½) of PfHRP2 was 3.67 (1.47) d, or 1.84 erythrocytic cycles, and was not correlated with measures of disease severity or biochemical evidence of renal or hepatic dysfunction. This information was used to construct a model to derive parasite numbers from the plasma PfHRP2 concentrations (Figure 1). It was assumed that patients were infected with a single relatively synchronous unimodal infection of P. falciparum [9]. Immediately following schizont rupture, released merozoites will infect new erythrocytes, expanding the infection with a multiplication factor M. If the number of parasites in the generation present at the moment of blood sampling is P, the number of parasites at the moment of the previous schizont rupture will be P/M. These parasites will release an amount of PfHRP2 (A, in grams) of 4.6 × 10−15 P/M into the circulation, giving a concentration of A/Vd, where Vd represents the volume of distribution of PfHRP2. As PfHRP2 is a relatively large molecule (30 kDa), it was assumed that Vd equalled the plasma volume, which is (100 − Hct) × total blood volume. The total blood volume was estimated to be 80 ml/kg body weight. As assessed in the pilot study described above, PfHRP2 elimination followed a first-order process, with a half-life (t½) of 3.67 d, or, for a 2-d cycle, 1.84 erythrocytic cycles, corresponding to an elimination constant k = 0.693/t½ = 0.377/asexual cycle. The concentration of PfHRP2 (in grams/litre) released at the moment of schizont rupture (CR) will then be given by the formula In the second half of the subsequent cycle, during the trophozoite and schizont stages, a total amount of 0.6 × 10−15 g of PfHRP2 per parasite will be released into the circulation by the newly infected erythrocytes. This production was assumed to be at a constant rate (1.2 × 10−15 g per parasite per cycle time) in the second half of the cycle. With time expressed as units of asexual cycle length and the elimination constant k = 0.377/cycle, the concentration CTS of PfHRP2 derived from these mature parasites during the second half of the erythocytic cycle will be given by the formula where R is the rate at which PfHRP2 is secreted into the plasma, equalling 1.2 × 10−15 P g per cycle, so that If C 0 is the concentration of PfHRP2 derived from the current cycle, C 0 = CR for the first half of the cycle, derived from schizont rupture of the previous cycle, and C 0 = CR + CTS during the second half of the cycle, derived from the residual amount released at schizont rupture and from release by trophozoites and schizonts from the present cycle. Since all blood samples were taken on admission, even if the patient was admitted during the night, it was assumed that blood samples were drawn randomly during the erythrocytic cycle of the parasite. The average concentration of PfHRP2 (C̄ 0 ) will thus be the area under the concentration–time curve divided by the time span of the erythrocytic cycle of the parasite generation present at the moment of blood sampling. For one cycle from t = 0 until t = 1, this will equal Substituting equations 1 and 3 into this equation, this can be solved to However, PfHRP2 accumulated from the erythrocytic cycles before the cycle during which the blood sample was taken will also contribute to the observed concentration of PfHRP2, because the elimination half-life exceeds the asexual cycle length. If C̄ 1 is the concentration of PfHRP2 derived from the cycle previous to the cycle at the moment of blood sampling, the amount of PfHRP2 (C̄ 1 ) produced in the previous cycle will be C̄ 0/M , and because of the exponential decline described by k, the average amount left in the next cycle will be e −k (C̄ 0/M) . For the n th previous cycle, the formula to calculate the contribution to the observed concentration of PfHRP2 can be generalized by induction to The total expected PfHRP2 concentration observed in a peripheral blood sample (C̄ obs ) can thus be expressed as which can be solved to Substituting equation 8 into equation 5 provides an explicit expression for the overall concentration as a function of the parameters: Figure 1 gives a graphic summary of the model. Rearrangement of equation 9 gives an explicit expression for the total parasite load Ptot as a function of the measured concentration: For the calculations of total body parasite biomass, an in vivo multiplication rate of eight was assumed. This is the average derived multiplication rate observed at detectable parasitaemias in patients with acute malaria in experimental infections before treatment [14]. Substituting these values into equation 10, and with the elimination constant k = 0.377, the total body number of parasites (Ptot) can now be calculated for a measured plasma PfHRP2 concentration (Cobs , in grams/litre): Vd (in litres) was estimated as (1 − Hct) × 80 × 10−3 × body weight (in kilograms), equalling the estimated plasma volume, so that The total circulating number of parasites (Pcir) was calculated from the observed peripheral blood parasitaemia (percent) in the peripheral blood thin smear multiplied by the estimated total body red blood cell mass. Assuming a total blood volume of 80 ml/kg body weight, total body red blood cell mass was estimated as where MRCV is the mean red blood cell volume. The estimated number of sequestered parasites was calculated as the difference between the estimated total parasite biomass and the total circulating parasite biomass. The model was subjected to uncertainty and sensitivity analysis with the following parameters as variables: (a) multiplication rate (M), the distribution of which was chosen to be uniform between one and ten; (b) PfHRP2 half-life (t½), normally distributed with a mean of 3.7 d and a standard deviation of 1.47 d, derived from the pilot study described above; (c) the amount of PfHRP2 released per cycle, normally distributed with a mean of 5.2 × 10−15 g and a standard deviation of 2.0 × 10−15 g (from [8]). The fixed parameters included: Hct = 0.35% and body weight = 50 kg, giving a Vd of 2.6 litres and an observed plasma PfHRP2 concentration of 1,000 × 10−6 g/l. Using these parameters 10,000 simulations were performed using the S-PLUS mathematical program (Insightful, Seattle, Washington, United States), with the predicted total parasite biomass as outcome. This showed that the predicted parasite biomass was log-normally distributed with a geometric mean of 2.9 × 1012. The coefficient of variation of the log-transformed outcome was 3%, indicating a high level of consistency. The partial rank correlation coefficients between each parameter and the outcome of the model were 0.64, −0.10, and −0.48 for multiplication rate, PfHRP2 half-life, and the amount of PfHRP2 secreted per cycle, respectively. This indicates that multiplication rate was the most influential factor affecting the total parasite biomass estimate, followed by the amount of PfHRP2 secreted per cycle. The variation in PfHRP2 half-life had only a small effect on the calculations. To illustrate the impact of the multiplication rate on the model, Figure 2 shows the total body number of parasites for a patient with a plasma PfHRP2 concentration of 1 × 10−3 g/l, Hct of 0.35%, and a body weight of 50 kg at different parasite multiplication rates. Statistical Analysis Data were analysed using SPSS for Windows release 10.05 (SPSS, Chicago, Illinois, United States). Continuous variables that were not normally distributed were log-transformed. The laboratory findings in the uncomplicated and severe groups were assessed using Student's t-tests for normal distributed parameters. Correlation was assessed by the methods of Pearson or Spearman as appropriate. Logistic regression was used to determine the relationship between mortality and plasma PfHRP2 or other prognostic factors including plasma haemoglobin, creatinine, lactate, bilirubin, and glucose concentrations. Results PfHRP2 Detection The lowest limit of detection of purified PfHRP2 was 1.5 × 10−6 g/l. The titration curves were similar when normal human plasma or PBS-TWEEN was used as the diluent. The mean (95% CI) plasma PfHRP2 concentration in 337 patients with falciparum malaria was 8.4 × 10−4 g/l (5.7 × 10−4 to 11.1 × 10−4 g/l). Patient Characteristics A total of 337 adult patients with falciparum malaria were enrolled over a period of 10 y. Patient characteristics are summarised in Table 1. According to the modified World Health Organization criteria adapted from Hien et al. [10], 167 patients had severe disease, of whom 28 (17%) died. Of the patients with severe disease 107 (64%) had cerebral malaria, defined as a GCS below 11. Twenty-two (13%) had renal failure (plasma creatinine > 265 μmol/l and diuresis 42.8 μmol/l), and 73 (44%) had a plasma lactate above 5 mmol/l. Eight (5%) patients were shocked on admission (systolic blood pressure below 80 mm Hg, with cold extremities). No patient had an admission Hct below 20%. Seventy-two patients (43%) were hyperparasitaemic (>10% of red cells parasitised) as assessed in the peripheral blood slide on admission. Pre-treatment before admission with antimalarial drugs was infrequent and did not differ significantly between patients with severe versus uncomplicated disease. In the group with uncomplicated malaria (n = 170), three patients were treated with chloroquine (which is completely ineffective in this area), two received one dose of an artemisinin-containing drug, two received a single dose of quinine, and in two patients an unknown antimalarial was taken. For the patients with severe disease (n = 167), the corresponding numbers were as follows: four patients were pre-treated with chloroquine, one with sulphadoxine-pyrimethamine (also ineffective), two with a single dose of quinine, and in another six patients an unknown antimalarial was taken. So only seven out of 337 patients were documented to be treated with an effective antimalarial drug, and in none was the dose curative. Estimated Total Body Parasite Biomass Compared to Peripheral Blood Parasitaemia and Parasite Developmental Stage Applying the model described above, the peripheral blood plasma concentration of PfHRP2 was used to estimate the total body parasite burden in 337 adult patients with falciparum malaria, assuming a uniform in vivo multiplication rate of eight (Figure 2). The geometric mean of the total parasite load for all patients was 7.0 × 1011 (95% CI 5.8 × 1011 to 8.5 × 1011) parasites/body (Figure 3). When the total parasite load was related to the estimated total red blood cell mass, the geometric mean whole body parasitaemia was 46 per 1,000 (95% CI 38 to 57 per 1,000) erythrocytes. The geometric mean of the simultaneously assessed peripheral blood parasitaemia for all patients was 28 per 1,000 (95% CI 25 to 33 per 1,000) erythrocytes. When the numbers of sequestered parasites were calculated (by subtraction) a negative value was obtained in 22% of the patients; in these cases patients were assumed to have very few or no sequestered parasites. The median number of sequestered parasites obtained in this way for all malaria cases was 2.5 × 1011 (interquartile range 4.7 × 109 to 1.5 × 1012) parasites/body. The number of sequestered parasites was also assessed in relation to the parasite stage in the peripheral blood smear. We have reported previously that in patients with severe malaria a predominance of mature stages (trophozoites and schizonts) in the peripheral blood (>104/μl) is associated with a fatal outcome, presumably because this reflects a stage distribution of more mature parasites composing the total parasite burden and thus a greater sequestered parasite biomass [9,11]. Using the previous cut-off value of >104/μl mature-stage parasites in the peripheral blood, the number of sequestered parasites in patients with severe malaria (median [interquartile range]) was four times higher if mature-stage parastites predominated (n = 79); 1.6 × 1012 (2.5 × 1011 to 5.4 × 1012) parasites compared with 4.0 × 1011 (6.7 × 109 to 2.1 × 1012) parasites if mature-stage parasites did not predominate (p = 0.002 by Mann-Whitney U test) (Figure S1). Overall there was a positive correlation between the proportion of late-stage parasites in the peripheral blood slide and the calculated proportion of sequestered parasites from the model (Spearman correlation, rs = 0.41, p < 0.001). Estimated Total Parasite Biomass and Severity of Disease The estimated total parasite biomass increased in proportion to severity of the disease.The geometric mean (95% CI) parasite load was six times lower in patients hospitalised without evidence of severe malaria than in patients with severe malaria: 2.8 × 1011 (2.3 × 1011 to 3.5 × 1011) versus 1.7 × 1012 (1.3 × 1012 to 2.3 × 1012) parasites/body (p < 0.001). Within the group of patients with severe malaria, those who died (n = 28) had over twice the geometric mean parasite load compared with patients who survived (n = 139): 3.4 × 1012 (95% CI 1.9 × 1012 to 6.3 × 1012) versus 1.5 × 1012 (95% CI 1.2 × 1012 to 2.0 × 1012) parasites/body (p = 0.03 with Bonferroni correction for multiple comparisons between the three groups; Figure 4). The calculated number of sequestered parasites also increased with severity of the disease. In “uncomplicated” malaria (n = 170) the median (interquartile range) number of sequestered parasites was ten times lower than in patients with severe malaria (n = 167): 9.0 × 1010 (6.7 × 109 to 4.1 × 1011) compared to 9.1 × 1011 (6.1 × 1010 to 3.2 × 1012) parasites/body (Mann-Whitney U test, p < 0.001). Within the patients with severe disease, the median (interquartile range) calculated number of sequestered parasites was twice as high in the fatal cases than in the patients who recovered: 1.4 × 1012 (4.3 × 1011 to 1.4 × 1013) versus 7.6 × 1011 (1.2 × 1010 to 2.7 × 1012), but this difference was of borderline statistical significance (Mann-Whitney U test, p = 0.055). Estimated Total Parasite Biomass in Relation to Other Markers of Severity Table 2 shows the associations between the estimated total parasite biomass and markers of disease severity. For comparison, the associations based on the calculated total circulating parasitaemia (from peripheral blood parasitaemia) with the same markers are shown. Within the 167 malaria patients with severe disease, 43 patients had cerebral malaria (defined as a GCS < 11 [10]). These patients had a higher estimated total parasite biomass than patients without coma, but also a higher peripheral blood parasitaemia. Plasma lactate concentrations can be regarded as a crude measure of the microcirculatory status and have a strong prognostic significance for mortality in severe malaria [15]. An admission plasma lactate concentration above 5 mmol/l indicates a poor prognosis. Patients with plasma lactate greater than 5 mmol/l (n = 73) had a significantly higher calculated total body parasite load than patients with a lower lactate concentration (n = 94): geometric mean (95% CI) 2.8 × 1012 (2.0 × 1012 to 4.1 × 1012) versus 1.3 × 1012 (0.9 × 1012 to 1.8 × 1012) parasites/body (Student's t-test, p = 0.002). The calculated sequestered parasite biomass was also significantly higher in patients with high plasma lactate concentrations (median 1.6 × 1012 parasites/body, interquartile range 9.4 × 1010 to 5.6 × 1012) than in those with an admission lactate concentration below 5 mmol/l (median 5.1 × 1011 parasites/body, interquartile range 9.1 × 109 to 2.0 × 1012, Mann-Whitney U test, p = 0.012) (Figure S2). There was a positive correlation between the calculated sequestered biomass and the plasma lactate concentration on admission (Spearman correlation, rs = 0.44, p < 0.001). There was no association between the circulating parasitaemia (calculated from the observed peripheral parasitaemia) and plasma lactate concentration. Of the severe malaria patients, 22 had renal failure, defined as a plasma creatinine above 265 μmol/l and a diuresis of less than 400 ml/24 h [10]. These patients had a significantly higher estimated geometric mean total parasite load than the patients without renal failure. It should be noted that the plasma PfHRP2 half-life was not affected by renal failure in our pilot study. There was no association between peripheral blood parasitaemia and renal failure. Jaundiced patients (n = 87), defined as those having a total plasma bilirubin above 42.8 μmol/l [10], had significantly higher estimated total numbers of parasites, but also higher peripheral blood parasitaemias, than those without jaundice (n = 80). Finally, a slight but significant negative correlation existed between admission Hct and the calculated log10 total parasite biomass (Pearson's method, n = 167, r = −0.38, p < 0.001), but there was also a weak, though significant, correlation between Hct and the log10 total circulating number of parasites (n = 167, r = −0.20, p = 0.008). The group of severe anaemic patients as a whole did not have a significantly higher number of either circulating or total parasites in their blood (Table 2). A logistic regression model was constructed with survival as the dependent variable and with the estimated total parasite biomass and previously described prognostic criteria in severe disease as independent variables [10]. Admission plasma lactate, plasma creatinine, and GCS contributed significantly in the model, but estimated total parasite biomass did not. This is in contrast to the univariate analysis, in which the estimated total parasite biomass and the sequestered parasite biomass (but not peripheral blood parasitaemia) correlated significantly with fatal outcome. Estimating the Time of Sequestration It is known from laboratory studies that cytoadherence via the main adhesion molecule PfEMP1 begins at approximately 12 h of parasite development under febrile conditions, and 50% of the maximum effect is obtained at approximately 14–16 h of development [16]. This concurs with the stage distributions of parasites observed in the blood smears of patients with malaria. If the time of admission to hospital and stage of parasite development (in hours) are unrelated, then the overall ratio of circulating parasites (Pcir) to total parasites (Ptot) in a population of patients gives the proportion of the asexual life cycle (assumed to be 48 h) for which parasites are circulating (Tcir, in hours) at an overall parasite multiplication rate of one (i.e., steady state). That is, But the parasite numbers are usually expanding, particularly in uncomplicated malaria before treatment. In the present model we have assumed a multiplication rate of eight per cycle for preceding cycles. If the parasite age distribution is normally distributed with a standard deviation of 4 h, as estimated previously [9], then this provides 27% more circulating rings overall in the patient's parasite population than if the multiplication rate is one per cycle. That is, If applied to the data from the patients in this series and this model, then with the assumptions above, the population mean parasite age at sequestration (Tcir) in uncomplicated malaria becomes and for severe malaria is The estimate in severe malaria is very close to the value observed in the laboratory. But in severe malaria the next parasite cycle may not sustain a multiplication rate of eight. For a multiplication rate of one this value becomes 17 h. Values between 14 and 17 h are compatible with laboratory observations, but the estimate of 23 h for patients hospitalised with uncomplicated malaria is significantly different from the value observed in ex vivo studies. As it is very unlikely that the age of parasites when they sequester is truly different between the two groups, this apparent excess of circulating parasites in uncomplicated malaria suggests that the assumption of an admission to hospital unrelated to stage of parasite development is incorrect in this group. This suggests that schizogony and the consequent pathological reaction to it are important in stimulating patients with uncomplicated malaria to come to hospital, whereas referral of patients with severe malaria is unrelated to the stage of development of their infecting parasite population. Discussion In this study we propose a model to estimate total parasite biomass using quantitation of a parasite-derived product (PfHRP2) that is predictably released into peripheral blood plasma in acute falciparum malaria. The estimated total parasite biomass was clearly associated with disease severity and outcome in univariate analyses. In a logistic regression analysis with outcome as dependent variable, the parasite load estimates did not predict disease severity better than conventional measures:GCS, plasma lactate, and serum creatinine outweighed the contribution of parasite load to the regression model, suggesting a close causal relationship between these variables. The estimated total parasite biomass was also associated with other well-established markers of severity. In contrast with this, peripheral blood parasitaemia and the derived total number of circulating parasites were not associated with disease outcome, nor with other important measures of severity such as admission plasma lactate concentrations. It should be noted that some of the criteria defining severe disease do include the level of the peripheral blood parasitaemia, which is variably related to total biomass, although in the model peripheral parasitaemia was not used in the calculations of parasite biomass. Pre-treatment with antimalarial drugs affecting PfHRP2 production cannot explain the differences in the estimated total parasite biomass between patients with severe versus uncomplicated malaria: documented pre-treatment was relatively rare (only seven out of 337 patients were pre-treated with an effective antimalarial drug), and this did not differ between the groups. The positive association between sequestered parasite biomass and disease severity fits with the widely accepted view that the sequestration of erythrocytes containing the mature forms of the parasite in the microvasculature of vital organs is the central pathological process in falciparum malaria. The greater the number of sequestered parasites, the more severe is the disease. The cytoadherence of parasitised erythrocytes to the endothelial lining of the microcirculation is very efficient. Mature forms of the parasite are rarely seen in the peripheral blood smear used to assess the patient, whereas they are central to pathology. Overrepresentation of late stages in the peripheral blood smear is a prognostic factor for fatal outcome, because it represents a larger sequestered parasite biomass [9,11]. Marchiafava and Bignami were the first to observe in their pathological studies of falciparum malaria that “malignancy coincides with an exceptionally abundant quantity of parasitic forms, a quantity much more abundant—where the cases terminate fatally—in the blood of the viscera than in the blood of the finger” [17]. Recent autopsy studies comparing parasite numbers in brain vessels with peripheral blood parasitaemia showed a median sequestration index (the ratio of parasitaemia in the capillaries and venules compared to that in the peripheral blood) of 40 (range 1.8–1,500) [18], and a recent electron microscopic study by Pongponratn et al. in patients with severe malaria showed a sequestration index of 50 in brain capillaries, but lower indices in other vital organs [19]. In our study the geometric mean ratio between the total body parasitaemia and the peripheral blood parasitaemia in fatal cases was 5.3 (95% CI 2.11 to 13.4) (data not shown). As this ratio represents the summed estimate of sequestration in the whole body, it supports the pathological observations of differential sequestration in the brain and other vital organs. The calculation of the sequestered biomass also revealed some limitations of the model we used, since in 22% of patients a negative value was calculated. This is not unexpected given that estimates for total parasite biomass and circulating parasite biomass derive from completely different parameters, and also that infections can be very synchronous [9]. Several factors may contribute to inaccuracies in the model. The amount of PfHRP2 secreted per parasite varies between different parasite strains; in our in vitro study the amount of PfHRP2 secreted per parasite per erythrocytic life cycle varied between 1.0 × 10−15 and 11.6 × 10−15 g. The antigenicity of PfHRP2 is not constant among parasite isolates because of variation in the expression of antigenic motifs. The calculations also assume a parasite multiplication factor of eight, which was derived from observations in early studies with experimental infection of humans with P. falciparum as a treatment for syphilis [14,20,21]. The multiplication factor has a significant impact on the model, as shown in the sensitivity analysis (see Figure 2). With lower multiplication factors, as are likely at high parasite densities, the contribution of the previous cycle to the total amount of PfHRP2 will increase, and a similar plasma concentration of PfHRP2 will represent a lower number of parasites. Indeed, in our model using a multiplication factor of eight, a small number of patients (21 out of 337 patients) had a calculated total number of parasites exceeding the calculated total number of erythrocytes. However, if a multiplication factor of one is used in these patients, no patient has a total body parasitaemia exceeding 100% of the calculated total number of parasites. Variation in another empirically derived estimate, the plasma half-life of PfHRP2, also has an impact on the model, but in the sensitivity analysis the effects were limited. In our study on plasma PfHRP2 kinetics in 13 patients with falciparum malaria, the longest plasma half-life was 6.6 d. Doubling the plasma PfHRP2 half-life in our model with an assumed multiplication factor of eight reduces the calculated total parasite load by a factor of 0.55. In high transmission areas, where partial immunity against the disease develops, clearance of PfHRP2 might be increased in the presence of antibodies against the protein. The model would thus underestimate the parasite burden and might need to be adapted for use in countries in sub-Saharan Africa. Another assumption made is that the volume of distribution (Vd) of the plasma PfHRP2 equals the plasma volume, since the size of the molecule will likely confine its presence to the intravascular compartment. However, binding to other proteins or cells could increase the apparent Vd, whereas dehydration could reduce Vd slightly. An increase or decrease in Vd will cause the model to under- or overestimate, respectively, the calculated total parasite load. Another source of variation is the PfHRP2 quantification itself. Higher concentrations of PfHRP2 require larger dilutions, with greater consequent errors. This may have contributed to the anomaly in those patients for whom the calculated total body parasitaemia as a percentage of circulating erythrocytes exceeded 100%. Finally, the time of blood-taking relative to the parasite developmental stage will influence the calculated parasite biomass. For example, assuming a multiplication factor of eight, an equal number of parasites will be associated with almost double the PfHRP2 concentration if the sample is drawn just before schizont rupture compared with halfway through the cycle, and the calculated parasite biomass will be affected by the same factor. P. falciparum infections do tend to be synchronous [9], and variations in PfHRP2 concentrations, and thus in the estimated biomass, related to the timing of the sample in relation to the stage of parasite development would be more substantial in those patients with synchronous infections. But the model predictions also provide an answer to this. By comparing the model-predicted parasite age at sequestration with the laboratory-observed values for cytoadherence, it was evident that in severe malaria the model gave a very good prediction compared with laboratory findings (14–17 h versus 14–16 h). This suggested that admission to hospital was unrelated to the timing of schizogony in severe malaria. But in patients with uncomplicated malaria, circulating stages were overrepresented, which suggests that schizogony, and the host reaction to it, probably do precipitate hospital admission. This is also supported by the predominance of parasites of the young ring stage observed commonly in admission hospital blood smears. Despite all these potential shortcomings, estimates of the total parasite load derived from the plasma PfHRP2 concentrations were within the range expected and correlated well with disease severity. In clinical studies of severe malaria, measurement of plasma PfHRP2 concentrations may be useful as a research tool to stratify patients by their parasite load. In summary, this study shows that quantitative measurements of plasma PfHRP2 in patients with falciparum malaria can be used to estimate the total parasite biomass, a parameter pivotal in the pathophysiology of the disease. This calculated total parasite biomass is associated with clinical measures of the severity of the disease. Supporting Information Figure S1 Correlation between Sequestered Parasite Biomass and Percentage of Late Stages Scatterplot showing the correlation between the sequestered parasite biomass (as a percentage of the total parasite biomass) and the percentage of late stages (trophozoites and schizonts) in an admission peripheral blood slide in 337 patients with falciparum malaria. The sequestered and total parasite biomass were derived from plasma PfHRP2 concentrations as discussed in the text (Spearman correlation, rs = 0.41, p < 0.001). (25 KB DOC). Click here for additional data file. Figure S2 Correlation between PfHRP2-Derived Parasite Biomass and Plasma Lactate Concentrations Scatterplot showing the correlation between the PfHRP2-derived sequestered parasite biomass on admission and admission plasma lactate concentrations in 337 patients with falciparum malaria (Spearman correlation, rs = 0.44, p < 0.001). (28 KB DOC). Click here for additional data file. Patient Summary Background Malaria is caused by parasites of the genus Plasmodium; P. falciparum causes the most severe disease. After a person is bitten by an infected mosquito, the parasites enter the person's red blood cells (erythrocytes), and then these infected red blood cells get stuck in small blood vessels, blocking the blood flow and causing many of the problems associated with malaria. The parasites multiply inside the red cells, which eventually split and release more parasites. Why Was This Study Done? It is possible under the microscope to see the number of parasites in the circulating red blood cells but hard to measure how many are inside red blood cells stuck in the small blood vessels at any one time. Knowing the number of parasites blocking up the blood vessels might give a better idea of how severe the disease a patient has. An indirect measure of the number of unseen parasites in the body is a protein, PfHRP2, which the parasites produce and which is released into the blood when the red blood cells split. What Did the Researchers Do and Find? They measured the amount of PfHRP2 in the blood of 337 adult patients with P. falciparum malaria and looked mathematically to see whether the PfHRP2 measurements could be linked to the total number of parasites each person had in their body. They found that people with severe malaria had high total numbers of parasites hidden in their red blood cells, as measured by the PfHRP2 levels. What Do These Findings Mean? Further work will need to be done to reproduce these findings in other groups of people with malaria. However, estimating the total number of parasites in this way may make it possible to predict more accurately the likely severity of a person's disease. Where Can I Get More Information Online? The World Health Organization has a set of Web pages on malaria: http://www.who.int/topics/malaria/en/ MedlinePlus has an interactive tutorial on malaria: http://www.nlm.nih.gov/medlineplus/tutorials/malaria/htm/index.htm