RECRUDESCENCE IN ARTESUNATE-TREATED PATIENTS WITH FALCIPARUM MALARIA IS DEPENDENT ON PARASITE BURDEN NOT ON PARASITE FACTORS

Patient recruitment, treatment, and evaluation.

Of 320 patients with falciparum malaria admitted during the period April 1998–March 1999 to the Bangkok Hospital for Tropical Diseases, Thailand, 150 patients were treated with artesunate tablets (600 mg total dose in 3 days; Guilin No. 1 Factory, Guangxi, China, registered for use in Thailand). Blood was collected routinely from all malaria patients. Inclusion criteria were as follows: had a case of acute uncomplicated falciparum malaria, were >18 years old, had no treatment with artemisinin derivatives during the previous 2 weeks, and gave informed consent. There were 104 patients who met the inclusion criteria for this study. Patients remained in the hospital for 28 days and had daily blood smears. Blood samples were taken before treatment on the day of admission and at the time of recrudescence and were stored in liquid nitrogen. Routine blood chemistry and hematologic parameters were determined automatically using a Boehringer Mannheim Hitachi 917 (Boehringer Mannheim, Indianapolis, IN) and Gen S (Coulter, Hialeah, FL).

The freezing solution for cryopreservation of parasites was prepared by first filtering a solution of either sorbitol or glycerol (4.2%) in 0.9% sodium chloride, through a 0.45-μm filter (New Bedford, MA). This solution was mixed with 99% glycerol to make a solution containing 3% sorbitol or glycerol, 0.65% sodium chloride, and 28% glycerol. Before cryopreservation, ring-stage patient blood samples were centrifuged at 400×g for 7 minutes at 4°C. The plasma was discarded, and the cell pellet was mixed with an equal volume of freezing solution. The mixture was divided into 0.5-ml aliquots, frozen, and stored in liquid nitrogen.

Parasite density was obtained from thin blood smears: 1,000 erythrocytes were examined, and the number of these infected with plasmodia yielded the percent parasitemia. Parasitemias were calculated as parasites per microliter of blood (multiplying percent parasitemia and red blood cell [RBC] counts).

Parasite drug sensitivity tests.

Blood samples were thawed, and the in vitro culture of the blood stage malaria parasite was performed by candle jar method as described previously using human AB serum.¹² Aliquots were removed and thawed underwater at room temperature. An equal volume of 3.5% sodium chloride was added, and the mixture was centrifuged at 400×g for 7 minutes at 4°C. The pellet was washed 3 times with stock RPMI 1640 medium and resuspended in RPMI. Parasites were cultured in a candle jar at 37°C until parasitemia reached 0.4% to 0.6% and were used for drug sensitivity assays. Individual isolates were not cultured for equal duration because culture time depended on starting parasitemia level; generally, no cultures were >1 month. Sensitivities to dihydroartemisinin were determined by the microdilution method using tritiated hypoxanthine incorporation.¹³ A 2-fold serial dilution of dihydroartemisinin was prepared in a 96-well plate to which a parasitized RBC suspension, mostly at the ring stage, was added. The plate was incubated at 37°C for 24 hours in a candle jar. Tritiated hypoxanthine (Amersham, Arlington Heights, IL) was added, after which the plate was incubated for another 18–20 hours. The infected cell suspension was harvested (Nunc Cell Harvester 8; Nalge Nunc International, Rochester, NY) and hypoxanthine incorporation was determined by a scintillation counter (Tricarb 2700 TR; Packard Company).

Hemoglobin typing and measurement of glucose-6-phosphate dehydrogenase.

Hemoglobin type was analyzed by cellulose acetate electrophoresis according to the method described by the manufacturer (Helena Laboratories, Beaumont, TX). Hemoglobins are separated at a pH of 8.6 on a cellulose acetate plate, then stained with Ponceau S. For some patients, the chromatogram could not distinguish conclusively between hemoglobin A₂ and hemoglobin E. In those cases, identification was based on the result of anion exchange chromatography (see later).

The quantitation of hemoglobin A₂ by anion exchange chromatography was performed according to the method described by the manufacturer (Helena Laboratories, Beaumont, TX). Under these conditions, hemoglobin A₂ was eluted while normal, and other abnormal hemoglobins were retained by the resin. The hemoglobin A₂ fraction was compared with a total hemoglobin fraction by determining the absorbance and calculating for the percentage of hemoglobin A₂. If the hemoglobin A₂ was <3.5%, then the inconclusive band on cellulose acetate was considered to be hemoglobin E.

Glucose-6-phosphate dehydrogenase (G6PD) deficiency was assessed by fluorescent qualitative spot test.¹⁴ The reaction mixture containing glucose-6-phosphate (G-6-P), tri-phosphopyridine nucleotide (TPN), saponin, and oxidized glutathione (GSSG) in potassium phosphate buffer was added to the blood samples. After incubation at 37°C for 3 minutes, the mixture was spotted onto a Whatman no. 1 filter paper. In the presence of G6PD, the substrate, G-6-P, was oxidized to 6-phosphogluconate, while TPN was reduced (TPNH), resulting in a fluorescent spot under long UV light.

Analysis of strains for prmdr1 polymorphisms.

Parasitized RBCs were harvested from culture when >80% of the parasites were trophozoites and stored at −20°C. DNA was extracted from the pellets using QIAamp DNA minikits (Qiagen, Valencia, CA). Four regions of the pfmdr1 gene were polymerase chain reaction (PCR) amplified by modifications of previously published methods.^15,16 Basically, we performed single-round PCR using the flanking primers described in these reports. For all 4 reactions, we used the PCR conditions described as PCR 184. We did not perform the second-round, however, allele-specific reaction. Instead, amplified DNA was purified and sequenced.¹⁷

Statistical analysis.

SAS version 8.2 (Cary, NC) was used to construct logistic regression models, with recrudescence as the dependent variable. Variables representing patient characteristics and laboratory test results were analyzed individually first, then multivariate models were constructed to control for important covariates. For simplicity, the continuous variable representing number of parasites per microliter of blood was categorized into a dichotomous variable. The cut-point for the dichotomous high/low parasitemia was set at 10,000 parasites/μl blood, which has been used in the literature.¹⁸ SAS also was used to conduct survival analysis for modeling time to recrudescence.

Recrudescent rate after artesunate administration.

A total of 104 patients with falciparum malaria were admitted to the malaria ward of the Hospital for Tropical Diseases and treated with 600 mg of artesunate for 3 days. All patients remained in the hospital for 28 days, and 32 patients recrudesced during their hospital stays (29%).

Characteristics of recrudescent and nonrecrudescent parasite isolates.

Of all the available blood samples, in vitro parasite cultures were obtained successfully from 6 nonrecrudescent patients and 16 recrudescent patients (on admission and during the episode of recrudescence) (Figure 1). Because artesunate is rapidly converted to dihydroartemisinin in vivo, 50% inhibitory concentrations (IC₅₀) for dihydroartemisinin were determined for each parasite isolate. A Wilcoxon signed rank test (for recrudescent versus nonrecrudescent patients) and a paired t-test (for admission versus recrudescent isolates) were used to evaluate differences in IC₅₀. There were no significant differences in IC₅₀ for parasites from recrudescent patients on admission (0.57 ± 0.44 ng/ml; range, 0.17–1.56 ng/ml) or recrudescence (0.615 ± 0.534 ng/ml; range, 0.12–1.75) or from nonrecrudescent patients (0.45 ± 0.22 ng/ml; range, 0.2–0.86 ng/ml).

Because all samples were frozen in liquid nitrogen before IC₅₀ determination, the effects of freezing and thawing on IC₅₀ were tested using the laboratory strain (K1). K1 was frozen and thawed under similar conditions (n = 5). The IC₅₀ was 0.38 ± 0.09 ng/ml before and 0.30 ± 0.07 ng/ml after freeze-thawing. There was no statistically significant difference using a paired t-test (P = 0.87), indicating that freezing, thawing, and reculturing did not affect the determination of IC₅₀.

In previous reports, pfmdr1 mutations have been associated with altered in vitro sensitivity to artemisinins.^19–21 Sequence polymorphisms in pfmdr1 that may decrease sensitivity have been reported at several codon positions: N86Y, Y184F, S1034C, N1042D, and D1246Y. For genotyping experiments, new cultures had to be established from cryopreserved stabilates. Cultures were obtained successfully from 10 recrudescent patients (from isolates obtained on admission and during their recrudescent episodes). We PCR amplified and sequenced the pfmdr1 genes from the admission and recrudescent cultures from these 10 patients. Initial and recrudescent cultures from 7 of 10 patients had wild-type pfmdr1 genotypes. Isolates from only 3 patients had mutations. In 2 of these 3 patients, isolates showed the same mutations before and after treatment (one had the N1042D mutation and the other was Y184F). In 1 of the 3 patients, the pretreatment isolate had a Y184F mutation that was not found in the post-treatment strain, suggesting that the original infection may have been polyclonal and that the nonmutant strain was responsible for the recrudescence.

Characteristics of recrudescent and nonrecrudescent patients.

The characteristics of the 32 recrudescent patients and 72 nonrecrudescent patients are shown in Table 1. The ages of patients with and without recrudescence were similar (25.9 ± 6.8 years and 24.9 ± 8.0 years). Both groups were predominantly male (69% of recrudescent patients and 63% of non-recrudescent patients). None of the patients were homozygous for thalassemia. Of 103 patients, 18 (17.3%) were heterozygous for hemoglobin E (Hb EA), a type of β-thalassemia, and of 104 patients, 9 (8.7%) had G6PD deficiency. No other abnormal hemoglobin bands were detected. There were no significant differences between the 2 groups in admission blood chemistries, mean corpuscular volume, mean corpuscular hemoglobin concentration, or platelet counts or in the percentage of patients with G6PD deficiencies or thalassemic trait.

There were several significant differences between patients who had recrudescence and patients who did not. First, the recrudescent patients had higher percent parasitemias on admission (2.65% versus 0.45% in nonrecrudescent patients). Second, recrudescent patients had significantly higher RBC counts, hemoglobin levels, and hematocrits (Table 1). Together, these data suggest that patients with higher parasite burdens were more likely to have recrudescence. When the number of parasites per microliter of blood was calculated, the difference was highly significant (P = 0.0035). The only other variable that was significantly different between recrudescent and nonrecrudescent patients was the white blood cell (WBC) count; recrudescent patients had significantly higher WBC counts than nonrecrudescent patients (6.6 versus 5.7). The odds ratio (OR) for recrudescence for patients having WBC counts >5.85 (the median) was 3.08 (95% confidence interval [CI], 1.27–7.44).

Logistic regression was done to confirm the association between recrudescence and factors found to be significant in univariate analyses, including RBC count, hematocrit, hemoglobin level, and percent parasitemia and parasitemia (number of parasites/μl) (Table 1). The level of parasitemia was represented by a dichotomous variable for high/low level of parasitemia, with 10,000 parasites/μl as the cut-point.¹⁸ (Similar results were found when the median, 12,150, was used as the cut-point.) Many of these factors are either closely related (RBC count, hemoglobin, and hematocrit) or indirectly related (i.e., RBC count × percent parasitemia = parasitemia). The strongest univariate association was with parasitemia, with an OR of 9.80 for the dichotomous variable (95% CI, 3.11–30.88).

The association of WBC count with recrudescence also was confirmed with logistic regression (OR, 1.24; 95% CI, 1.00–1.53). The OR indicates the increase in odds of recrudescence for each increase of 1,000/μL in WBC count. When controlling for parasitemia, the OR for recrudescence with increasing WBC count remained approximately the same but lost significance (OR, 1.20; 95% CI, 0.94–1.53). To explore the issue of host WBC count and recrudescent malaria, the types and amounts of WBC were determined. WBC types measured included eosinophils, basophils, monocytes, lymphocytes, and banded and segmented neutrophils. The measurements of each of these WBCs were analyzed as continuous variables to test for an association with recrudescence (trend) using logistic regression.

In univariate regression of WBC types, eosinophil, basophil, monocyte, and lymphocyte levels had no association with recrudescence (data not shown). Band neutrophils had an OR of 1.01 (95% CI, 1.00–1.02), and segmented neutrophils had an OR of 1.00 (95% CI, 1.00–1.01). When controlling for parasitemia, segmented neutrophils had an OR of 1.00 (95% CI, 1.00–1.01). The WBC types individually did not have a meaningful association with recrudescence.

We then evaluated whether time to recrudescence was associated with any particular host factor. Survival analyses were performed for all of the variables in Table 1 versus time to recrudescence. Significant associations were found for many of the same variables as in bivariate analysis (parasitemia, RBC count, hemoglobin level), which suggest that time to recrudescence depends on the parasite burden. We also found that time to recrudescence depended on blood urea nitrogen (BUN) level. Figure 2 shows survival curves of recrudescent patients categorized into normal (≤20 mg/dl) and high (>20 mg/dl) categories. The hazard ratio decreased significantly with the higher level of BUN (P = 0.028). Patients with high BUN levels (and poorer renal function) had recrudescence later than patients with low BUN levels.