A morphometric and histological study of placental malaria shows significant changes to villous architecture in both Plasmodium falciparum and Plasmodium vivax infection

Abstract

Background: Malaria in pregnancy remains a major health problem. Placental malaria infection may cause pathophysiological changes in pregnancy and result in morphological changes to placental villi. Quantitative histomorphological image analysis of placental biopsies was performed to compare placental villous architecture between active or treated placental malaria cases and controls.

Methods: A total of 67 placentas were studied from three clinical groups: control patients who did not have malaria (n = 27), active (n = 14) and treated (n=26) malaria cases, including both Plasmodium falciparum and Plasmodium vivax infections. Image analysis of histological placental sections was performed using ImageJ software to measure the number and size (area) of terminal villi, perimeter measurement per villus and total perimeter per unit area, and number of capillaries per villus (vascularity). Histological features of placental malaria were scored and these results were correlated with malaria status and clinical outcomes.

Results: Villous size correlated with vascularity (p <0.0001) but was inversely correlated with observed villi per unit area, (p = 0.0001). Significantly greater villous area and vascularity was observed in UK controls. Indices of histological malaria infection were significantly greater in active versus treated malaria cases. Active placental malaria cases showed significantly smaller villous area (p <0.0084), vascularity (p <0.0139) and perimeter (p <0.0006) than treated malaria cases or controls, but significantly more villi per unit area (p <0.0001). Villous size in treated malaria cases was significantly larger than active placental malaria cases (p <0.001) and similar to controls. There was a significant relationship between villous number and anaemia at the time of infection (p <0.0034), but not placental weight, birth weight or gestational age at delivery. No differences were found between histology or villous morphology comparing infections with P. falciparum or P. vivax.

Conclusions: These results imply that villous size, perimeter and vascularity are acutely decreased during active placental malaria, decreasing the surface area available for gas exchange per villus. However the increased number of villi per unit area offsets this change and persists after treatment. Histopathological and villous architectural changes may be reversed by early detection and appropriate anti-malarial treatment.

Figure 1

Flow chart showing the type and number of the cases in active and treated malaria and control groups.

Figure 2

Image processing steps used to generate quantitative outputs from haematoxyling and eosin-stained placental sections. The various stages of processing are graphically illustrated in the following sequence: (A) original image (magnification x400); (B) preprocessing step following removal of partial outlines and non-terminal villi; (C) adjusted threshold to generate solid red pixelated areas; (D) computerized measurement of area of terminal villi using the resultant outlines.

Figure 3

Correlations between the villous area, number of villi per high-powered field and villous vascularity. 3 A – Dot plot showing the negative correlation between the villous area and observed number of villi per hpf in individual cases (Line = line of best fit, Spearman rank correlation coefficient rho = −0.425, p <0.0001). 3 B – Dot plot showing the positive correlation between the villous area and villous vascularity in individual cases (Line = line of best fit, Spearman rho = 0.668, p <0.0001).

Figure 4

Boxplot graph comparing villous area between clinical groups of controls and cases derived from different fixative methods or geographical origin. A boxplot showing the distribution of villous area in sq μm between groups including UK controls, UPS cases (Thai/Karen cases fixed with formalin, divided into control non-malaria infected and malaria -infected UPS cases), PCA cases (Thai/Karen cases fixed with Streck fixative), and active placental malaria cases (Thai/Karen cases also fixed with formalin). UK controls showed significantly higher villous area than other groups (p <0.0004). PCA cases showed a non-significantly higher villous area than UPS controls fixed in formalin (p <0.0842). Active placental malaria cases showed significantly lower villous areas than all other groups *. Box and whisker plots for this and subsequent figures include; line = median, upper and lower bar = 25 and 75^th centiles, respectively, with shaded box covering interquartile range; whiskers cover maximum and minimum values.

Figure 5

Boxplot graphs showing variation in placental villous morphological parameters between clinical groups. 5 A - Villous size. There was a significantly lower villus cross sectional area in active malaria cases compared to control or treated malaria (*p <0.0084, Fisher’s exact test). 5 B - Villous vascularity. There was a significantly lower number of capillaries per villus in active malaria cases compared to control or treated malaria (*p <0.0139, Fisher’s exact test). 5 C: Number of villi per high-powered field. There was a significantly higher number of villi seen per hpf in both active and treated malaria cases compared to controls (*p <0.0001, Fisher’s exact test). 5 D: Mean perimeter measurement per villus. There was a significantly lower perimeter measurement per individual villus seen in active malaria cases compared to treated malaria cases and controls (*p <0.0006, Fisher’s exact test). 5 E - Total perimeter measurement for all villi in 10 hpf. There was a significantly higher total perimeter measurement of all villi seen per 10 hpf in both active malaria and treated malaria cases compared to controls (*p <0.0004, Fisher’s exact test).

Figure 6

Boxplot graph of villous number per high-powered field compared to maternal anaemia at the time of infection. Boxplot showing the significantly higher number of villi seen per 10 hpf in cases with mild and severe anaemia compared to patients without anaemia (*p <0.0034, Fisher’s exact test).