Fatal Plasmodium falciparum Malaria Causes Specific Patterns of Splenic Architectural Disorganization

Tissue samples.Spleen samples were taken postmortem from Vietnamese adults who died from P. falciparum malaria. These patients were studied as part of a double-blind control drug trial for the treatment of severe malaria with quinine or artemether in a special research ward at the Centre for Tropical Diseases in Ho Chi Minh City, Vietnam (19). Patients were recruited into the drug trial study on admission to the hospital when they gave informed consent, had asexual P. falciparum blood stages as determined by a peripheral blood smear, and had one or more of the following complications: a Glasgow coma score of less than 11 (cerebral malaria), anemia (hematocrit, <20%), jaundice (serum billirubin concentration, >2.5 mg/dl), renal impairment (urine output of <400 ml in 24 h and serum creatinine concentration of >3 mg/dl), hypoglycemia (blood glucose concentration, <40 mg/dl), hyperparasitemia (>10% parasitemia), and systolic blood pressure below 80 mm of Hg with cool extremities. Patients received either arthemether (50 mg/ml) or quinine dihydrochloride (250 mg/ml) for a minimum of 72 h. All patients underwent full clinical examination on admission to the hospital, and the level of peripheral blood parasitemia was determined every 4 h for the first 24 h and then every 6 h until three consecutive blood smears were negative or the patient died. For the patients who recovered from P. falciparum infection, the median time to total parasite clearance was 72 h (interquartile range [IQR], 54 to 102 h) for treatment with arthemether and 90 h (IQR, 66 to 108 h) for treatment with quinine. The median times to discharge of patients from the hospital were 288 h (IQR, 216 to 432 h) and 240 h (IQR, 192 to 336 h) for treatment with artmether and quinine, respectively (19). A detailed description of the clinical management of patients has been published previously (19).

Some patients died more than 100 h after admission to the hospital (19). These deaths were due to late clinical complications and irreparable organ damage, such as acute renal failure, intractable shock, respiratory arrest with continued pulse, and metabolic acidosis as a direct result of P. falciparum infection. A full autopsy was performed on each of the patients who died if informed consent was obtained from relatives. Autopsies were performed shortly after death (median time, 7 h; IQR, 3 to 12 h) (33).

We performed a histological and immunohistochemical study of the spleen for samples from 15 adults who died with severe falciparum malaria (Table 1). Eight of these patients had cerebral malaria, and seven patients suffered from at least one other complication of malarial disease, such as hyperparasitemia (four patients), anemia (three patients), and acute respiratory failure (four patients). When all malaria cases were considered, the average maximum level of parasitemia ranged from 480 to 1,644,104 iRBC per μl of blood, the median level of parasitemia was 112,537 iRBC per μl of blood, and the median admission level of parasitemia was 92,442 iRBC per μl of blood (IQR, 38,057 to 217,037 iRBC per μl of blood). The median level of parasitemia before death was 20 iRBC per μl of blood (IQR, 10 to 3,401 iRBC per μl of blood).

In order to detect features in the spleens from malaria patients which were specific for infection with P. falciparum rather than signs of systemic infectious disease, we also analyzed spleen sections from 12 United Kingdom patients dying from primary or secondary disseminated bacterial sepsis proven by premortem blood culture (Table 1). The normal controls were tissue samples from 16 trauma patients who had their spleens removed during surgery (control) but had normal splenic histology because such spleen specimens were not available from Vietnamese adults. These patients were older than the malaria patients or the control group of trauma patients; the median ages of the malaria, control, and sepsis patients were 39 years (IQR, 24 to 52 years), 38 years (IQR, 28 to 69 years), and 69 years (IQR, 62 to 82 years), respectively. Collection of postmortem tissues and the protocols used for this study were approved by the Central Scientific and Ethics Committee of the Centre for Tropical Diseases in Vietnam and Oxford. Autopsies were performed only after the patients' families gave informed consent.

Histology.Following fixation in 10% phosphate-buffered formalin, tissue blocks were dehydrated in graded alcohol by using standard histological techniques, embedded in paraffin, and sectioned with a microtome. Sections were cut for immunohistochemistry onto Snowcoat Xtra slides (Surgipath, Petersborough, United Kingdom). The histological appearance was observed following counterstaining with hematoxylin and eosin. In some cases silver staining was used to reveal reticulin fibers of the connective tissue framework of the spleen.

Immunohistochemistry.Unless stated otherwise, reagents were purchased from DAKO (Cambridge, United Kingdom). An immunohistochemistry analysis was performed with a panel of antibodies defining common leukocyte subgroups normally found in the spleen. These antibodies included CD20 (clone L26) defining B cells, CD3 (rabbit polyclonal antiserum or clone F7.2.38) defining T cells, CD68 (clone PGM-1) defining monocytes/macrophages, neutrophil elastase (clone NP54) defining neutrophils, and prion protein (3F4) recognizing myeloid dendritic cells. Staining for major histocompatibility complex (MHC) class II molecules (clone CR3/43) was also performed to examine patterns of immune activation among splenic leukocytes.

Tissue sections were dewaxed and rehydrated by using standard procedures. For antigen retrieval, sections were heated in either 20 mM Tris-5 mM EDTA (pH 9) (CD20, MHC class II, monoclonal antibody against CD3), 10 mM citric acid (pH 6) (polyclonal antibody against CD3, CD68), or 2 mM HCl (normal prion protein) for 4 min and then washed in 20 mM Tris-HCl (pH 7.5)-100 mM NaCl. The sections were stained with antibodies as indicated above by using the indirect alkaline phosphatase method (APAAP method) and were developed by using New Fuchsin Red (14). The sections were counterstained with hematoxylin and mounted.

Electron microscopy.The tissues were fixed in 2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) and processed as described previously (33). Thin sections were stained with uranyl acetate and lead citrate prior to examination with a Jeol 1200 EX II transmission electron microscope.

Quantitative analysis.For quantitative ultrastructural analysis of spleen sections, the squares of a copper grid were used as the reference area. A minimum of 100 infected and noninfected red blood cells were counted in 10 grid squares in three different areas, and the percentage of iRBC and the percentage of knob-negative iRBC were determined.

For quantitative immunohistochemical analysis, at least three independent areas were counted for each parameter. Blind counting of spleen sections was not possible because of the even macroscopic dark brown appearance of spleen sections from malaria cases. Different regions within the white pulp were distinguished by a combination of morphological features and reactivity with antibodies. Thus, primary B-cell follicles (or tangentially sectioned corona) were densely packed with cells. In the marginal zone surrounding the follicles, the cells were less densely packed and consisted mainly of B cells, as well as various amounts of macrophages, dendritic cells, and T cells (37). Cell numbers were determined by using a graticule, divided into 100 squares, and expressed as the number of cells in 0.1 mm² or 0.01 mm² (in the case of dendritic cells and B cells in the marginal zone) of the section. B cells, T cells, macrophages, neutrophils, and prion protein (PrP)-positive dendritic cells were counted at a magnification of ×400. The ratio of reactive lymphoid follicles to all lymphoid follicles and the areas covered by red pulp and white pulp (1 mm²) in the section examined were assessed by using a magnification of ×40. A statistical analysis was performed with SPSS (SPSS Inc., Chicago, Ill.) by using the Mann-Whitney U test, the chi-square test, or Spearman's correlation.

Gross splenic architecture.Splenomegaly is usual in malaria and indeed has been used historically as an epidemiological index for the intensity of malaria transmission in areas where malaria is endemic. We first compared splenic weight and gross architecture for all groups. The weight of the spleen was significantly increased in malaria cases (P < 0.001), whereas the weight of the spleen in sepsis cases was slightly but not significantly increased compared to controls (Table 2). The red pulp was frequently congested with erythrocytes both in cases of malaria and in cases of sepsis. The ratios of the area covered by the red pulp to the area covered by the white pulp were similar for malaria cases and controls and for sepsis cases and controls (Table 2). We therefore assumed that the increase in weight in malaria cases was due to expansion of both compartments, the red pulp and the white pulp, as has been reported previously for rodent malaria (44). Additional silver staining of extracellular fibers revealed that the red pulp and the white pulp remained clearly separated (Fig. 1a), implying that the overall architecture of the spleen was retained despite distortion and changes in the different cellular compartments within the white pulp, as described below.

• Open in new tab
• Download powerpoint

FIG. 1.

Histology and immunostaining of spleen sections. Representative sections of spleens from patients who had their spleens removed during surgery (normal) or who died with P. falciparum malaria (malaria) or with disseminated sepsis (sepsis) are shown. Sections were stained with a standard silver stain (a) to reveal reticular fibers or were immunostained with anti-CD20 (b) or anti-CD3 (c) to demonstrated segregation of B-cell and T-cell compartments in the white pulp. In normal and sepsis cases, T- and B-cell areas are clearly distinguished (dark staining) from the marginal zone (light staining in panel b) surrounding B-cell follicles and the periarteriolar lymphatic sheath. In malaria cases, the B-cell follicle (dark staining in panel b) remains clearly separated, but T cells (dark staining in panel c) appear to be scattered around arterioles in the B-cell follicle and throughout the white pulp. In malaria cases, the marginal zone has a ragged appearance and is almost completely devoid of B cells (light blue staining in panel b surrounding B-cell follicle). (d) The anti-CD68 antibody identifies cordal macrophages in the red pulp cords and tingable body macrophages in the germinal center. Again, in normal and sepsis cases, B-cell follicles (dark blue areas containing germinal centers in the section from a normal case) can be easily distinguished from the surrounding marginal zone (light blue), whereas in malaria cases the marginal zone is apparently absent. (e) HLA-DR-positive cells were scattered through the red pulp and in the marginal zone. In malaria cases, the white pulp and the marginal zone contain only a few HLA-DR-positive cells. (f) Strongly staining cells are present in the white pulp and near central arterioles. These cells have large cell bodies and cytoplasmic protrusions between neighboring cells typical of interdigitating dendritic cells. Note the strong expression of HLA-DR on sinusoidal lining cells (d) but the almost complete absence of strongly HLA-DR-positive cells in the white pulp (f) in spleen sections from malaria patients. (g and h) Immunostaining for neutrophils (elastase) (g) and PrP-positive myeloid dendritic cells (h). Neutrophils are enriched in the perifollicular zone just adjacent to the white pulp, as well as scattered throughout the red pulp. The staining pattern for PrP-positve myeloid dendritic cells is clearly distinct from that for neutrophils and macrophages. All sections were counterstained with hematoxylin. (a, b, and d to h) Magnification, ×90. (c) Magnification, ×360.

Changes in the distribution of B and T lymphocytes.In the spleen, B cells are located in B-cell follicles in the white pulp and in the marginal zone and are scattered throughout the red pulp. While primary B-cell follicles contain naïve B cells, the marginal zone contains memory B cells, as well as marginal zone B cells. Both B-cell subsets respond rapidly to invading pathogens by secretion of antibodies. Interestingly, we observed a profound depletion of B cells from the marginal zone surrounding B-cell follicles only in malaria cases (Fig. 1b and 2 and Table 1). This depletion did not result from migration of activated B cells into the red pulp to form B-cell foci or into B-cell follicles to induce germinal center reactions. Indeed, the frequency of germinal centers within lymphoid follicles was significantly reduced in malaria cases compared with controls (Table 2), indicating that differentiation of B cells to plasma cells and memory cells was compromised in these patients. A similar trend was observed for sepsis cases. Activated B cells and plasma cells can migrate into the red pulp to form B-cell foci and undergo differentiation and to secrete antibodies, respectively. However, the number of B cells in the red pulp was decreased in both malaria and sepsis cases compared to controls, indicating that the depletion of B cells from the marginal zone in malaria cases was not due to migration into the red pulp. Because in malaria cases splenic weight was increased significantly, we corrected the number of B cells found in the red pulp for variation in the splenic weight in the disease groups, assuming that the splenic volume was proportional to the splenic weight. Although only a rough estimate, this correction could give an indication of whether the expansion of the red pulp and the white pulp in the spleens of malaria cases was due to changes in the number of cells of a particular leukocyte subset. After correction for splenic weight, in malaria cases the reduction in the number of B cells per unit of area in the red pulp that we observed was no longer significant. In contrast, the overall number of B cells in sepsis cases remained significantly reduced after correction for splenic weight (Table 2).

• Open in new tab
• Download powerpoint

FIG. 2.

Immunostaining of spleen sections for B cells. Three representative spleen sections from P. falciparum malaria cases and from control cases (normal) are shown. In normal cases, B cells are densely packed in B-cell follicles. B cells are also abundant in the surrounding marginal zone but are more loosely packed. In malaria cases, B cells could be identified in B-cell follicles, but the number of cells was significantly reduced (P < 0.001, as determined by Mann-Whitney U test). All sections were counterstained with hematoxylin. Magnification, ×100.

The number of CD3-expressing T cells in the red pulp varied widely in both disease groups (Table 2 and Fig. 1c). When we corrected the number of T cells for splenic weight as described above, the overall number of T cells in the red pulp was significantly increased in malaria cases.

In addition, we observed in malaria cases but not in sepsis cases that T cells were not segregated into separate areas but were scattered diffusely in the white pulp and in B-cell follicles (Yates-corrected χ² for malaria cases versus controls, 9.31 [P < 0.01]; Yates-corrected χ² for sepsis cases versus controls, 0.01 [P = 0.93]) (Fig. 1b and c).

Changes in the distribution and phenotype of myeloid cells.Macrophages in the cords of the red pulp are important for the clearance of iRBC and parasite products. CD68-positive cordal macrophages in the red pulp were loaded with hemozoin, the polymerized form of heme resulting from digestion of hemoglobin by the parasite (Fig. 1d). This phenomenon accounted for the even macroscopic dark brown appearance of the spleens from malaria cases. The number of macrophages in the red pulp tended to be lower in malaria and sepsis cases than in controls, but the differences were not significant (Table 2). When we corrected for splenic weight as described above, the relative number of macrophages in the red pulp was significantly increased in malaria cases (Table 2). Expression of the MHC class II molecule on cordal macrophages was reduced in malaria cases compared to controls or sepsis cases (Fig. 1e). A similar phenomenon was observed for interdigitating dendritic cells in the white pulp, which were defined by their morphology and high levels of expression of MHC class II molecules (Fig. 1f). In contrast to the weak staining of MHC class II molecules on macrophages and interdigitating dendritic cells, we observed high levels of expression on sinusoidal lining cells in the red pulp in almost all spleen sections from malaria cases (χ² for malaria cases versus controls, 4.01 [P < 0.05]; χ² for sepsis cases versus controls, 0.06 [P = 0.8]) (Fig. 1e). These results indicate that ongoing inflammatory processes within the spleens of malaria patients were sufficient to induce upregulation of MHC class II molecules on sinusoidal lining cells but not on macrophages and interdigitating dendritic cells.

The perifollicular zone and to a certain extent the marginal zone contain numerous antigen-presenting and phagocytic cells, including macrophages, myeloid dendritic cells, and neutrophils. Neutrophils were enriched in the red pulp adjacent to white pulp, presumably the perifollicular zone, as has been described previously (37) (Fig. 1g). The number of neutrophils in this area was reduced in malaria and sepsis cases compared to controls (Table 2). When we corrected the number of neutrophils in the red pulp for splenic weight, the difference was no longer significant in malaria cases.

The normal form of PrP is expressed on myeloid dendritic cells in the marginal zone and in the red pulp, but interdigitating dendritic cells in the white pulp express this protein only weakly (9). The overall number of PrP-expressing myeloid dendritic cells was increased in malaria cases compared to control or sepsis cases. PrP-expressing myeloid dendritic cells were scattered throughout the red pulp, and in one-half of the control cases they accumulated immediately adjacent to the white pulp, presumably in the perifollicular zone (Fig. 1 h). In malaria cases, PrP-positive myeloid dendritic cells were significantly enriched in the red pulp and marginal zone (Fig. 3). The number of PrP-positive myeloid dendritic cells in the red pulp was correlated with the number in the marginal zone (Spearman's ρ, 0.815; P < 0.001) but was inversely correlated with maximum peripheral blood parasitemia (Spearman's ρ, −0.579; P < 0.05). Staining of consecutive section confirmed that PrP-expressing myeloid dendritic cells were distinct from neutrophils or macrophages (data not shown).

• Open in new tab
• Download powerpoint

FIG. 3.

Box plots of frequencies of myeloid dendritic cells in different spleen compartments. PrP-positive myeloid dendritic cells (DC) were counted in the marginal zone (open bars) and red pulp (shaded bars) in control, malaria, and sepsis spleen sections. The boxes indicate the median and 25th and 75th percentiles. Outliers are indicated by circles. There is a highly significant increase in the number of dendritic cells in both compartments in malaria cases compared to sepsis and control cases (P < 0.001, as determined by Mann-Whitney U test).

Mature parasite-infected red blood cells in the spleen.Using transmission electron microscopy, we analyzed the quality and quantity of intact iRBC in spleen sections of six patients who died with malaria. We identified infected cells using ultrastructural evidence of parasitic infection within the RBC and expression of electron-dense knob proteins on the erythrocyte surface.

The percentages of iRBC reflected the percentages seen in the brain and peripheral blood irrespective of whether the patients suffered from cerebral malaria or other severe complications (Table 3). In most patients the percentage of iRBC in the spleen was lower than the percentage in the brain but higher than the percentage in the peripheral circulation just before death. Many of the iRBC contained trophozoites or mature schizonts and were knob positive (K⁺), indicating that there was expression of variant surface antigens mediating cytoadherence (Fig. 4a and b). In one case, a high proportion of the iRBC were knob negative (K⁻), but these iRBC contained large parasites consistent with gametocyte formation. It was possible to identify iRBC pushing through fenestration between endothelial cells (Fig. 4a and c). In addition, a number of iRBC exhibited constrictions that appeared to isolate part of the cytoplasm containing the parasite (Fig. 4d). A number of macrophages were observed to contain iRBC, pigment crystals, and iRBC ghosts (Fig. 4d).

• Open in new tab
• Download powerpoint

FIG. 4.

Electron micrographs of the spleens of patients dying of malaria. (a) Low-power micrograph showing part of a sinus within the spleen containing a number iRBC (IR), some of which appear to be passing through fenestration between endothelial cells (En). Bar = 5 μm. (b) Detail of one iRBC containing a mature schizont with merozoites (M). Note the electron-dense knobs (K) at the surface of the red cell. Bar = 1 μm. (c) Higher-power micrograph showing three infected red blood cells (arrowheads) passing through small gaps between endothelial cells (En). P, parasite. Bar = 5 μm. (d) Infected knob-positive red blood cell (K) showing constriction of the cell cytoplasm (arrow), which appears to isolate the portion of the cell containing the parasite (P). Bar = 1 μm. (e) Section through a macrophage which has phagosomes containing infected red blood cells (IR), ghosts of infected cells (G), and pigment granules (Pi). Bar = 5 μm.