Asymptomatic Plasmodium falciparum infections may not be shortened by acquired immunity

Abstract

Background: The duration of untreated Plasmodium falciparum infections is a defining characteristic of the parasite's biology. It is not clear whether naturally acquired immunity (NAI) can shorten infections, despite the potential implications for malaria control and elimination as well as for basic research.

Methods: Data on the presence of P. falciparum msp2 genotypes in six blood samples collected over one year was analysed, together with four samples collected over 1 week, from a cohort in Navrongo (Ghana). Mathematical models assuming either exponential, Weibull, gamma, or log-normal infection durations were estimated separately for six age-groups. The method allowed for varying clonal acquisition and detection rates.

Results: The best fitting (Weibull) mean durations were 124 days (children <5 years old), 179 days (5-9 years), and 70-90 days (>10 years). This non-monotonic age pattern is not suggestive of an infection-clearing effect of NAI since immunity increases with exposure, and thus, age. Age-related differences in innate immunity are a more plausible explanation. 21% of blood-stage infections terminated within 1 week, in stark contrast to months of persistence in infections induced in neuro-syphilis patients (malariatherapy data). Age independence in this percentage raises the possibility that this clearance may result from innate mechanisms or genetic incompatibility between hosts and parasites, rather than from NAI.

Conclusion: In all ages of hosts a substantial proportion of infections are cleared in the first days or weeks of appearance in the blood, while others persist for many months. Although cumulative exposure and NAI increase with age, this does apparently not translate into an increased rate of termination of infections.

Fig. 1

Seasonal Transmission. Each group of bars shows the FOI estimates for one 2-month season. Estimates differ considerably depending on the mathematical model for infection clearance, but agree with respect to the pattern of seasonality. The Weibull model clearly gave the best fit to the data, thus yielding the most reliable estimates.

Fig. 2

Effects of host age on the average infection duration. The average duration of clonal infections is shown against the midpoint of each age group and compared across clearance models. Lower AIC values indicate a better fit to the data. All models qualitatively agree with respect to the age pattern: an initial increase in duration during childhood, a sudden drop around the time of puberty, and no further decrease thereafter. The best-fitting Weibull model estimates that average duration remains constant in adults at ca. 80 days, even after decade-long exposure. Both the non-monotonic changes during childhood and adolescence as well as the absence of a change in adults are consistent with NAI not acting to shorten infections.

Fig. 3

Short infections are similarly common in all age groups. The distribution of infection durations in Navrongo as estimated by the best-fitting Weibull model is illustrated for all age groups separately. Short infections are similarly common in all age groups, as indicated by the left-skewed PDF. An estimate from non-immune adult malaria therapy patients, where short infections are rare, is shown for comparison [8]. Mean durations are indicated by circles on the abscissa. If infections were cleared shortly after inoculation because of host immunity against particular antigenic variants one would expect an increase in the proportion of short infections with host age, a proxy for cumulative exposure.

Fig. 4

Effects of host age on clonal detectability. Detectability estimates from the long-interval data are in good agreement across different mathematical models. Starting from ca. 40% in children, the detectability of a single clone by PCR decreases with age to ca. 20%. Lower detectability in older hosts is likely due to a parasite density-reducing effect of NAI.

Fig. 5

The probability of re-detecting clonal infections decreases with time. Shown is the probability that a genotype present in the first of two samples will also be detected in the second, together with a 95% confidence interval. This probability decreases with length of the time interval between surveys (p $<$ 0.001), at a rate corresponding to 3% of infections cleared per day, or 21% per week. This estimate was obtained from the short-interval data which permits tracking of clones at high time resolution, with four blood samples taken within 7 days.