Association of sub-microscopic malaria parasite carriage with transmission intensity in north-eastern Tanzania

The study area, study design, and sampling have been described elsewhere with parasitological, haematological, serological and entomological measures all demonstrating malaria transmission intensity decreases with altitude [9‐11]. Briefly, the study area is based in north-eastern Tanzania and runs from Kilimanjaro, through the Eastern Arc Pare and Usambara Mountains to Tanga on the coastal plain. Malaria prevalence has been shown to decrease with altitude and with distance from the coast, which is linked to average annual precipitation. In 2001, cross-sectional surveys were conducted during the short rainy season (October to November 2001) and finger-prick blood samples were collected into EDTA-coated tubes from an age-stratified sample of approximately 250 people from each of 13 villages (Figure 1) from three different age groups: 0-4 (n = 80), 5-14 (n = 80) and 15-45(n = 90) years old. The villages were situated at differing altitudes and in transects selected to be as similar as possible in terms of ethnic group, socio-economic status and accessibility to health care. Full details of demographic and clinical procedures can be found in Drakeley et al [9].

For PCR analysis, samples were chosen at random from the 13 villages to include a minimum of 60 samples from each village, with 20 from each of three age groups: 0-4, 5-14 and 15-45 years old. A total of 1,121 samples were analysed for the carriage of Plasmodium falciparum using nested PCR parasite. Prevalence by microscopy had been determined previously and results published in Drakeley et al [9].

DNA was extracted from the archived, pelleted samples, using Nucleon kit protocol for DNA extraction according to manufacturer instructions http://​www.​tepnel.​com. A nested PCR was used to amplify species-specific sequences of the small sub-unit ribosomal ribonucleic acid (18S SSU rRNA) genes of P. falciparum as described [12, 13]. DNA from the culture of P. falciparum (3D7 strain) as a positive control, DNA from blood samples of an individual never exposed to malaria as a negative control, and no template as a negative control, were included in each set of PCR for quality control. The PCR was carried out in a tetrad thermo-cycler (PTC-0240, The DNA engine Tetrad2^® Thermal Cycler, Bio-Rad, Hercules, California, USA), followed by gel electrophoresis to assign individuals as either parasite positive or negative.

Nested PCR was carried out to determine the numbers of msp2 (FC27 and IC/3D7) clones. To amplify DNA, a primer pair corresponding to the outer conserved region of the polymorphic repetitive block 3 of msp2 was used. Then the PCR product was re-amplified in a pair of reactions using primers specific for FC27 and IC/3D7 allelic types of msp2 as described by Snounou et al [14]. Number of products, corresponding to number of infecting FC27 and IC/3D7 clones, was counted after visualization on ethidium bromide (EtBr) stained 2.5 and 1.5% agarose gel, respectively.

The difference in PCR-based parasite prevalence in different villages was assessed by using chi-square and logistic regression to determine the association of parasite prevalence by PCR with different predictors as independent variables, such as age, distance from the coast (a crude proxy for rainfall) and altitude. PCR prevalence and MOI were examined for correlation with altitude, sero-conversion rate (SCR) for age related acquisition of antibodies to apical membrane antigen-1(AMA-1)[15] and measured and extrapolated entomological inoculation rate (EIR) [15] that had been collected previously. Regression analysis was conducted in Stata using the 'svy' function to allow for stratified analysis with village as the primary sampling unit.

Ethical clearance was obtained from the London School of Hygiene & Tropical Medicine (LSHTM), Kilimanjaro Christian Medical Centre, Tanzania (KCMC) and the Tanzanian National Medical Research Institute (NIMR).

The comparison between the 1,116 samples with both PCR and microscopy results is shown in Table 1. Of these, 62.8% were negative and 9.2% positive by both methods, however, PCR identified more parasite-positive individuals than microscopy: 34.4 vs 12.0% respectively. Thirty-one of 1,116 (2.7%) samples were positive by microscopy but negative by PCR; the majority of these samples (22) had parasite densities ≤ 80 parasites/μl. The samples that were microscopy positive but P. falciparum negative by PCR were analysed for other species. Two of the samples were Plasmodium malariae positive.

Using PCR as a gold-standard technique for malaria parasite detection, the sensitivity and specificity of microscopy were 26.8% (103/384) and 95.8% (701/732), respectively. The positive and negative predictive values were 78.4% (105/134) and 71.2% (699/982).

The 2.5-fold difference in the prevalence of parasitaemia by PCR compared to microscopy is in line with other recent observations. The difference was observed in all villages at different altitudes (Table 2 and Figure 2) and was relatively consistent at the different altitude bands despite a marked decrease in parasite prevalence with altitude: < 600 m 70.9 vs 28.6, 600-1,200 m 35.5 vs 9.9, > 1,200 m 15.8 vs 5.9. The difference between parasite prevalence by PCR was 3.2 in individuals aged between 15 and 45 years (34.5 vs 10.9) compared with 2.5 in those aged 1-5 (34.0 vs 13.5), though this was not statistically significant. The highest prevalence was observed in the middle (5-14 years) age group and was lowest in adults (15-45 years) (Table 3).

This was assessed in samples that were microscopy negative but P. falciparum PCR positive, and microscopy positive but PCR P. falciparum negative. A total of 117 samples were analysed for mixed infections by PCR, 12% of individuals had co-infections of P. falciparum and either P. malariae or P. ovale, and all were from lowlands. None of the samples analysed was Plasmodium vivax. The majority of samples (85.7%) with mixed infection were of P. falciparum and P. malariae. Six samples had Plasmodium ovale infection and out of these, four samples had mixed infection of P. falciparum, P. malariae and P. ovale.

After excluding individuals who were microscopically parasite positive, a total of 982 individuals were further involved in the analysis using logistic regression. In uni-variate logistic regression (Table 4), residents in the Tanga region were five-fold more likely to carry sub-microscopic parasites (OR, 5.00 [95%CI, 3.72-6.73]; p < 0.0001). This was also apparent within the Tanga region with both West Usambara and Tanga coast (Mgome village) transects, which were associated with increased carriage of sub-microscopic parasites; the magnitude being approximately four-fold (OR, 4.46 [95% CI, 0.88-22.49]; p = 0.067) and more than ten-fold (OR, 23.86 [95% CI, 7.00-81.29]; p < 0.0001) respectively. Analysis based on altitude band, regardless of region, indicated that individuals living in areas where the altitude was 600 m-1,200 m (OR, 0.28 [95%CI, 0.08-1.01]; p = 0.052) and above 1,200 m (OR, 0.10 [95%CI, 0.02-0.43]; p = 0.005) were less likely to carry sub-microscopic parasites as compared to individuals living in the area of < 600 m. Use of bednets, sex, fever and treatment were not associated with sub-microscopic parasite carriage in individuals (p > 0.05).

Individuals with sub-microscopic parasites had significantly lower mean haemoglobin levels (11.4 vs 12.1 g/dl respectively, F = 19.2 p < 0.0001), suggestive of longer term or chronic infections. However, an interaction with of other factors (such as iron deficiency anaemia, helminth infections, HIV, red blood cell polymorphisms) that are associated with decreased levels of haemoglobin cannot be ruled out. No significant difference in prevalence of sub-microscopic parasites was seen between the sexes or in those who reported fever or recent anti-malarial treatment.

The results showed that the majority of PCR positive subjects carried multiple strains; 76.2% (189/248). The MOI in PCR positive samples was significantly higher in high transmission/low altitude area (P = 0.025). There was no significant association between age and MOI (p > 0.05).

Comparison of the different measures of transmission intensity (Table 5, Figure 2) indicates that all measures were strongly correlated. Relevant to this study, highly significant correlation co-efficients were observed between altitude and parasite rate by PCR and by microscopy, predicted EIR and sero-conversion rate.