Dynamics of asymptomatic infections and symptomatic malaria show different temporal profiles among infants living in a high transmission area of Ghana

Background Although infants are vulnerable to malaria, the characteristics of their patterns of infections are not well described. This study aimed to examine the longitudinal profiles of asymptomatic infections and symptomatic malaria in the first year of life. Methods A birth cohort in Kintampo, Ghana (N = 1855) was followed actively with monthly blood sampling and passively for any febrile illness between 2008 and 2011. Malaria parasites were detected by light microscopy and infants who were infected or uninfected were identified. Infections were classified as either symptomatic or asymptomatic using fever and temperature readings over twelve months of follow up. The longitudinal infection profiles in all infants were then compared. Results: Asymptomatic infections and symptomatic malaria were observed at all ages but were rare the first months of life and the proportion of symptomatic malaria increased after six months. Among 1264 infants having microscopy data for at least eight monthly visits, four patterns were observed: parasite negative at all visits (36%), always asymptomatic (7%), always symptomatic (35%) and alternating between asymptomatic infections and symptomatic malaria (22%). The cumulative incidence of infection was highest in the alternating group, and many different profiles (87 different combinations) of asymptomatic infections and symptomatic malaria were observed in this group . Parasite densities were significantly low for the always asymptomatic group and highest for always symptomatic group. Conclusion Infants in malaria endemic areas experience highly different infection profiles over the first year of life despite living in the same area. In-depth investigations of why some infants are parasite free and others have repeated symptomatic malaria or maintain asymptomatic infections or alternate between asymptomatic infections and symptomatic malaria can contribute to understanding malaria susceptibility during infancy.

reservoirs for transmission can impede efforts to control and eliminate malaria (6,7).
Infants experience both asymptomatic infections and symptomatic malaria (8,9). However, during the first six months of life, infections are reported to be mainly asymptomatic, while between six and twelve months of age the incidence of both asymptomatic and symptomatic malaria infections increase (8,(10)(11)(12)(13). The low incidence of symptomatic malaria below six months of age has been attributed to presence of fetal hemoglobin (14,15) and passively acquired maternal IgG (16). Also, malaria-specific antibodies at birth (in maternal and/or cord blood) have been associated with protection against some malaria parasite antigens but not others (16)(17)(18)(19)(20)(21).
Studying the temporal patterns of asymptomatic infections and symptomatic malaria in infants is important for understanding relationships between susceptibility and immunity, and to guide control interventions such as presumptive treatment and vaccines. In addition, the challenge of identifying correlates of protection (22) can be overcome by studying large cohorts of exposed and non-exposed infants who undergo frequent sampling over longer periods. Nonetheless, previous studies assessing malaria morbidity in the first year of life are either based on large cohorts with few cross-sectional samplings (9,23), or with frequent sampling in small cohorts (16,19). Consequently, the longitudinal profiles of malaria outcomes and age-specific patterns of infections have not been studied on a large scale within the first year of life (16). Hence the characteristics and determinants of continuous infections in infancy are not well described, there is lack of clarity on the correlates of protection during infancy and the transition from asymptomatic infection to symptomatic malaria or the reverse, through the first year of life is poorly understood.
We conducted a study in a birth cohort living in a high transmission area of Ghana where blood sampling of infants for malaria parasitaemia was performed monthly and at passive/sick visits (24).
This study provided a platform to identify exposed and unexposed infants and to examine agespecific infection patterns among infants exhibiting similar longitudinal profiles of asymptomatic infection and symptomatic malaria through their first year of life.

Study location and population
Kintampo North Municipality (KNM) and Kintampo South District (KSD) consist together of 156 communities and covers an area of 7 162 km² in Ghana. Malaria transmission in this area is high and perennial, peaking between April and October (25,26).
In October 2008, a birth cohort study enrolled pregnant women from the study area (24). Infants were included in the study at birth between November 2008 and January 2011 and followed up actively every month until 12 months of age or to the end of the study in May 2011 or exit due to death, migration or loss to follow up (24).

Collection of samples and identification of parasites
Each child was visited monthly at home (active visits), by trained field staff who collected blood samples and completed standard questionnaires. In-between active visits, care-givers were encouraged to seek care from a study clinician, and for blood sampling if their infant was unwell (passive visits). Infants with fever and/or parasites were treated with artesunate/amodiaquine according to the national health policies. All infants were given free health insurance, paid for by the study.
Blood samples (500µl -1ml) were drawn on active and passive visits. Blood smears were examined for malaria parasites by light microscopy. Parasite densities were estimated per microlitre (µl) of blood, by assuming 8000 leukocytes per µl of blood, counting the number of parasites per 200 leukocytes and multiplying counts by 40 for the fraction of leukocytes examined. Parasite counts per 500 leukocytes and multiplied by 16 were performed if parasites were less than 100/µl. Microscopy test was not carried out on the passive visit if the infant had no fever in the past 48 hours or axillary temperature below 37.5°C. Details including enrollment and follow up visits have been previously described (24).

Definitions of symptomatic malaria and asymptomatic infection
Symptomatic malaria was defined as parasite positive by microscopy, a history of fever, illnesses or axillary temperature ≥ 37.5°C (elevated temperature) in the past 48 hours or subsequent seven days at either the active or passive visit. Asymptomatic infections were defined as parasite positive by microscopy at the monthly active visits with temperature < 37.5 °C and neither fever, illnesses nor passive visits in the preceding 48 hours or subsequent seven days.

Statistical analyses and profiling of symptomatic malaria and asymptomatic infections
Data were analyzed using Stata Version 14 (STATA Corporation, College Station, TX) and JMP statistical software. The number of infants with microscopy results from birth to one year of age was determined in all visits. Active and passive visits for which variables including temperature, fever or parasitaemia were missing were excluded as these were necessary for classification.
For the temporal profiling of asymptomatic infections and symptomatic malaria, infants having eight or more active and passive visits were included. Supplementary Figures 1a and 1b shows this in more detail.
The frequency distribution and mean of infections, cumulative incidence of infections, age at infection and medians of parasite densities were described for infants in the entire cohort as well as the groups with longitudinal asymptomatic infections or symptomatic malaria or mixed profiles. Infants from the different profiled groups having two or more infections were identified and the time (in days) intervals between any two infections were examined.
The Kruskal Wallis test was used to assess differences in age (in months), infections and parasite densities between groups, while Wilcoxon signed-rank tests were performed for within group analyses. Estimates with p-values ≤ 0.05 were considered statistically significant.

Characteristics of the cohort
The birth cohort included 1855 infants and the second of 36 twin births were excluded to prevent clustering effects during analyses ( Figure 1).  Figure 1). The infants had a mean follow up time of 6 (SD: ± 3) visits.
Malaria parasites were detected among 58% (971/1674) of infants at least once during the follow up period. In total, 55% (919/1674) of infants had malaria parasites and with corresponding temperature test results and/or history of fever, and these were used to define symptomatic or asymptomatic outcomes.

Malaria parasites, symptomatic malaria and asymptomatic infections in the cohort
Out of a total of 12926 malaria microcopy tests performed during the follow up for 1674 infants, 19% Prevalence of infections (symptomatic or asymptomatic) increased from birth to twelve months of age and was highest (20%, 261 infants infected/1314 total number of infants) at seven months of age ( Figure 2c). From birth to five months of age, infections were predominantly asymptomatic and from six to twelve months of age both symptomatic malaria and asymptomatic infections became more frequent among the infants (Figure 2c). This trend was also reflected in the cumulative incidence of symptomatic malaria which was similar to asymptomatic infections until five months of age, after which most infections were symptomatic ( Figure 2d).
For both symptomatic malaria episodes and asymptomatic infections, the median P. falciparum parasite density was significantly lower from birth to five months of age compared to six months of age and above (Supplementary Figure 2).

Profiles of longitudinal symptomatic malaria and asymptomatic infections
For detailed description of infection dynamics throughout the first year of life, infants (70%, 1264/1819) who had microscopy test results and recorded temperature and/or fever results on eight or more active visits and any passive visits were analyzed (Supplementary Figure 1a and 1b), to identify the temporal occurrence of asymptomatic infections and symptomatic malaria. In infants with malaria parasites, infections occurred intermittently throughout the year on both passive visits (where microscopy was performed on the same day and infants were treated) and active visits (where infants were 'healthy', and microscopy was performed at a later date and no efforts were made to treat) ( Figure 3). Infections among infants in the always asymptomatic group were identified during active visits only, unlike the other groups whose infections were identified during both active and passive visits ( Figure 3).

Infection profiles among groups of infants
Malaria parasites were detected among 64% (805/1264) of infants in this sub-cohort at least once during the active and passive visits.  (Table 2). Of these, 54% (274/506) had both symptomatic and asymptomatic infections i.e. alternating group, 39% (198/506) were in the always symptomatic group and 7% (34/506) in the always asymptomatic group ( Table 2).
The cumulative incidence of infections from birth to twelve months of age was highest in the alternating group, followed by always asymptomatic group -where it was similar to the always symptomatic group from eight months of age onwards (Figure 5a). Analysis of the number as well as the sequence of infections in the alternating group showed 87 different combinations of asymptomatic infections and symptomatic malaria (Table 3). Half (57%; 50/87) of these combinations of alternating asymptomatic infections and symptomatic malaria were unique to individuals (Table 3). Assessing the number of times asymptomatic infections and symptomatic malaria alternated in any of the 87 combinations showed three alternating infections [observed in six different sequences among 25% (68/274) of infants] were frequent, while ten alternations at the most occurred ( Table 3). The most common alternating sequence was to first have asymptomatic infection(s) followed by a symptomatic episode(s) (46%; 125/274). The second most common sequence was first symptomatic episode(s) followed by asymptomatic infection(s) (19%; 51/274) ( Table 3).
The time span between any two infections peaked between 27 and 38 days for all groups. Infections which were six months (180 days) or more apart were rarely observed in all groups (Supplementary Figure 3).

Age at infections among groups of infants
First infections were detected from one to twelve months of age [median = 6 months, (IQR: 5 -8)]. Whereas the first symptomatic malaria episodes begun from two to twelve months of age for infants in the always symptomatic group (Figure 5b), first infections were detected from one to eleven months of age for infants in the alternating group (Figure 5c). For infants in the always asymptomatic group the first infections occurred between one and twelve months of age (Figure 5d).
Although infections were detected at all ages, only few infants (5%; 36/805) had infections at two months of age, and they were mainly from the alternating group [69% (25/36)]. For all groups with malaria parasites, the number of times infants were infected (on separate occasions) increased as they aged (Figure 5b, 5c and 5d), but the proportion of infants infected decreased with age ( Table 2).
The median age of infection was seven months (IQR: 6 -10) for the alternating group and eight months for both always asymptomatic (IQR: 5 -9) and always symptomatic groups (IQR: 6 -10).

Parasite densities among groups of infants
The range of P. falciparum asexual parasite densities overall was 2 -974759 parasite/ul. Parasite densities increased with age in all groups (Figure 6), and the median parasite density at the first infection was highest in the always symptomatic group, followed by the alternating group and lowest in the always asymptomatic group (P < 0.001) ( Table 4). Compared to other groups, the always symptomatic group had the highest median parasite density from one to six months of age (P < 0.001) as well as from seven to twelve months of age (P < 0.001) ( Table 4).
In all groups, the parasite densities were lower the first six months of age compared to seven months of age and above (all P < 0.001).

Discussion
The longitudinal profiles and age-specific patterns of asymptomatic infections and symptomatic malaria were examined in a birth cohort from a high malaria transmission area of Ghana. Despite living in the same area, infants were found to experience highly diverse infection patterns the first year of life. Based on microscopy defined asymptomatic infections and symptomatic malaria, four main groups of infants were identified. A third of infants studied were parasite negative throughout, and of those having parasites, half had only symptomatic malaria, less than a quarter remained asymptomatic while a little over a quarter alternated between asymptomatic infections and symptomatic malaria at any time infections were detected. In total, 87 different patterns of alternating asymptomatic infection and symptomatic malaria (over half unique to individuals) were detected among 805 infants with P. falciparum parasites, suggesting that although there may be factors facilitating the transitioning from asymptomatic infections to symptomatic malaria, there is no uniform sequence of going between being either symptomatic or asymptomatic.
Given the development of naturally acquired immunity against malaria at intermittent time periods (27,28)  When comparing asymptomatic infections to symptomatic malaria in the entire cohort, the cumulative incidence of symptomatic malaria was higher from six to twelve months of age. However, when comparing the groups of infants, the cumulative incidence of infection was lowest in always symptomatic group and highest in the alternating group, indicating relatively poor protection against asymptomatic infections and symptomatic malaria among infants in alternating group. Importantly, given the higher number/cumulative incidence of infection, the alternating pattern may be observed among infants more frequently, compared to the other infection patterns. For instance, asymptomatic infections were frequently followed with symptomatic malaria among children aged six months to five years of age in urban Uganda (29). The frequent observation of infants with alternating infection patterns can diminish the detection of other infection patterns and their contribution to malaria the first year of life. Nevertheless, the findings in this study provide new approaches for analyzing data from cohort studies to yield a better understanding of immunity and susceptibility to malaria in infancy. For example, comparing individuals who are repeatedly infected and asymptomatic versus those symptomatic, or studying the same individuals when they are asymptomatic versus when they are symptomatic.
Assessing all infections from the three groups, two or more discrete infections (mostly detected 27 to 38 days apart) were more frequent than one infection. A previous study conducted in 2003/2004 showed children below three years of age and living in Kintampo had as many as seven malaria attacks/child/year (30). In this current study conducted in 2009/2010, the maximum mean number of infections was four, observed within the alternating group, indicating that although two or more infections were frequent than one, there has been a considerable drop in malaria among infants in Kintampo between 2003 and 2010, a reflection of global trends (1).
The age at first infection was lower for the alternating group (seven months) than for both always asymptomatic and always symptomatic groups (eight months). On the contrary, Wagner and others (16) reported a median age at first infection in infants at ten months in coastal Ghana although ITN usage was absent in the era of their research. However, their sample size was small (71 infants) and their definitions for malaria were based on microscopy/PCR and temperature ≥ 37.5 °C at the time of sample collection (16). Clearly, differences in sample size, definitions and study design/location may have contributed to the dis-similar age at first infection we observed.
Having asymptomatic infections first have been shown to reduce the risk of subsequent symptomatic malaria among two to ten year old children in rural Mali (31) but increase the risk among children who are six months to five years of age living in urban Uganda (29). In this study, infants in the alternating group whose first infections were asymptomatic were more compared to infants whose first infections were symptomatic. Also, in the always symptomatic group, malaria was observed from two months of age onwards. Although this study did not aim to model for risk of symptomatic malaria, the results indicate that after birth, asymptomatic infections occurred first before symptomatic malaria, possibly due to the influence of host genetics and/or immunity.
Consistent with the classical age distribution of infections in infants (12,32,33), we observed frequent asymptomatic infections between birth and five months of age and increased asymptomatic infections and symptomatic malaria from six to twelve months of age in the entire cohort and in the groups of infants studied. As infants aged, the number of times they were infected increased in all groups. However, the proportion of infants accumulating multiple infections as they aged decreased in all groups. These observations support the notion that there are mechanisms which reduce risk of symptoms but not parasite infections within the first six months of life (16,19). Nevertheless, the significantly lower age at first infection but higher cumulative incidence of infection for alternating group compared to other groups, suggest possible differences between the groups of infants regarding mechanisms (that may include host genetics, bed-net usage, exposure to maternal antibodies or parasite parameters) which may reduce the risk of symptoms.
Similar to observations in a hyperendemic area of Burkina Faso (9) and a meso-endemic area of Uganda (29), in this study, parasite densities increased with age and was highest in always symptomatic group. Generally, studies in malaria endemic areas show low parasite density in both asymptomatic infections and symptomatic malaria for infants below six months of age (9,13). High parasite density during symptomatic malaria in infants below six months of age is reported to be influenced by placental malaria (34). Whether low parasite densities lead to protection against symptoms is however unclear. Nevertheless, this study showed significantly lower parasite densities for below six months of age (where infections were predominantly asymptomatic) than above when the longitudinal patterns of infections was either considered or not. In addition, significantly lower parasite densities during the asymptomatic periods compared to the symptomatic periods were observed among infants in alternating group, suggesting that protection against symptoms could be influenced by low parasite densities. However, the significantly lower parasite densities throughout the entire year for always asymptomatic group compared to the other groups suggest that other mechanisms in addition to parasite densities may influence the manifestation of symptoms through the first year of life.
The proportions of asexual stages Plasmodium falciparum (96%) and Plasmodium malariae (2%) detected were similar to observations made in other malaria endemic areas of Africa (35). Sexual stages of the malaria parasites were of relatively low frequency and this observation is also consistent with that made in other endemic areas (36,37).
There may be underlying factors that influenced assigning infants to the always asymptomatic or always symptomatic group and the use of microscopy for parasite detection could be one major factor, therefore it is important to restrict interpretations to microscopy. All infants could have submicroscopic infections (that may be asymptomatic or symptomatic) detectable with methods which are more sensitive than microscopy e.g. polymerase chain reaction (PCR) (16,38). In this regard, carriage of submicroscopic infections would be important for examining low-grade parasitaemia among infants identified as parasite negative, an analyses beyond the detection limit of this study.
Moreover, a universally accepted or precise definition for asymptomatic malaria is lacking currently, and this may have influenced the assigning of an infection a symptomatic or an asymptomatic status.
While seasonality and parasite density have been considered in some studies (39-41), antimalarial usage, microscopy or polymerase chain reaction detection methods and varying duration between initial detection of parasites and onset of symptoms ranging from three to twenty-one days, have also been used to define malaria in other studies (5,21,42). Consequently, the profiles obtained for infants in this study could change if our definitions followed any of the above. However upon infection, infants show symptoms, compared to remaining asymptomatic (8,11,29,43), and this informed our expanded search for passive/hospital visits, elevated temperature, illnesses or fever, 48 hours before and seven days after an infection was detected. Further, the susceptibility of infants to

Ethical approval and consent to participate
This study was granted ethical approval from the Kintampo Health Research Centre (KHRC) Institutional Ethics Committee (Office for Human Research Protections federal wide assurance number is 00011103 and Institutional Research Board registration number is 0004854). Additional approval for this study (Dnr 2018(Dnr /1967 was from the Regional Ethics Board of Stockholm. Ethical approvals for the original birth cohort study were from KHRC, Ghana Health Service and London School of Hygiene and Tropical Medicine. Mothers gave consent for themselves and their infants to participate in the birth cohort study.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
All authors declare no competing interest

Authors' Contributions
AKB contributed to the design, data analyses and writing of this manuscript, SOA contributed to the design and writing, MA contributed to data analyses and writing, UH contributed to the data analyses, FBO contributed to the data analyses and writing, SG contributed to data analyses, DD contributed to the field work and laboratory analyses, GJ contributed to the laboratory analyses and writing, MFK contributed to the data analyses, EBK contributed to the field work, FO contributed to the design and writing, AF contributed to the design, data analyses and writing, and KPA contributed to the design, data analyses and writing of this manuscript.
All authors read and approved the final manuscript polymerase chain reaction as a routine method for the diagnosis of imported malaria.