Systematic review and meta-analysis: rapid diagnostic tests versus placental histology, microscopy and PCR for malaria in pregnant women

Malaria infection during pregnancy is a major public health problem in subtropical regions throughout the world. An estimated 125.2 million pregnancies occurred in areas with Plasmodium falciparum and/or Plasmodium vivax transmission in 2007, of which approximately 30.3 million occurred in Africa [1, 2]. Of the five human malaria species, P. falciparum causes the most severe effects during pregnancy, and P. vivax is associated with maternal anaemia and low birth weight [3, 4]. The effects of Plasmodium malariae, Plasmodium ovale and Plasmodium knowlesi infections in pregnancy are not well studied. P. falciparum infection during pregnancy is estimated to cause 10,000 maternal deaths each year and annually an estimated 75,000-200,000 infant deaths are linked to malaria in pregnancy (MiP) [1, 2, 5, 6]. In low-transmission areas, P. falciparum infection during pregnancy usually presents as a symptomatic, severe disease that can result in death of mother and foetus. In high-transmission areas few P. falciparum infections result in fever and maternal death, but the newborn infant can be severely affected by intrauterine growth retardation and pre-term delivery [2]. Infants born after a pregnancy affected by malaria often suffer from anaemia and have an increased risk of contracting malaria themselves [2]. Furthermore, P. falciparum infection during pregnancy increases the risk of stillbirth [2]. The severe effects of P. falciparum and P. vivax malaria on pregnant women and their (unborn) infants make early detection and subsequent treatment of great importance. Even though there are control measures to prevent malaria infection during pregnancy, such as intermittent preventive treatment (IPTp) and bed nets (ITNs), diagnosis is essential in areas where there is anti-malarial or insecticide resistance. IPTp greatly reduces prevalence of malaria and severe consequences, but women are not protected throughout the entire pregnancy and can still become infected between doses or after the final dose, especially when other protective measures such as ITNs are not being used, or parasites are resistant to sulphadoxine-pyrimethamine. IPTp is mostly applied in areas where there is high malaria transmission, where women have acquired immunity and infection during the pregnancy is often asymptomatic, but not without consequences. Therefore, accurate diagnostic tools are necessary to confirm infection. Additionally, screening and subsequent treatment of women during pregnancy might be more effective than a preventive approach in areas with low levels of transmission or highly seasonal transmission [7, 8]. Using IPTp in low transmission areas might result in a large proportion of pregnant women receiving sulphadoxine-pyrimethamine unnecessarily and a strategy has been proposed to screen pregnant women at antenatal care (ANC) visits with a rapid diagnostic test (RDT) and treat women who are positive with an effective anti-malarial [8]. An essential element of this strategy is good accuracy of the test and moreover it requires an affordable and quick diagnostic tool, such as an RDT.

P. falciparum infection in pregnant women presents differently than in non-pregnant women, where parasites are found in the circulation and can sequester to endothelial cells [9]. In pregnant women, P. falciparum malaria parasites express a different antigen variant (VAR2CSA) than in non-pregnant women, allowing them to sequester in the placenta and this is known as placental malaria [10]. Hence, in pregnant women peripheral parasitaemia can be absent or below the detection limit of microscopy [11, 12].

While microscopic examination of blood slides is considered the 'gold standard' for diagnosis in non-pregnancy related malaria, accurate detection of parasite infection in the placenta requires examination of histological sections of fixed placental tissue [13‐15]. An alternative is to examine placental blood with microscopy [11, 12, 15]. Placental histology and microscopic examination of placental blood can only be performed after delivery, when the placenta is available for examination. Since the detection of malaria parasites in the placenta is not possible during pregnancy, there is at this moment no other alternative than to detect the infection in peripheral blood.

There are alternatives to microscopic examination to test the peripheral blood though, such as rapid diagnostic tests (RDTs), which have the advantage of being quick and easy in remote settings. Depending on the manufacturer, the quality of the RDT in terms of accuracy and stability can be high [16, 17]. RDTs are based on the detection of parasite antigens in the blood by specific monoclonal antibodies. RDTs for malaria detect one or more of the following antigens: Histidine Rich Protein 2 (HRP2), Plasmodium Lactate Dehydrogenase (pLDH) or Aldolase. In a recent systematic review comparing diagnostic accuracy of RDTs for uncomplicated P. falciparum infection, it was reported that HRP2-based RDTs have better sensitivity than pLDH-based tests, although specificity is better for pLDH-based tests [18]. In general, RDTs detecting HRP2 are most commonly used, because they are less expensive, more stable across a wider temperature range and have a lower detection threshold than pLDH-based tests [19, 20]. HRP2-based tests, however, detect only P. falciparum, and antigenic variation of this antigen may cause false negative results [21]. The HRP2 antigen is excreted from the infected red blood cell, which can be beneficial for the diagnosis of placental malaria, as the antigen can be detected in the circulation even when the parasite is sequestered in the placenta [22].

Other alternatives for malaria diagnosis are DNA/RNA-based detection techniques, of which the polymerase chain reaction (PCR) is the most widely used [23, 24]. PCR is considered to have the most sensitive detection level of parasites (for both regular peripheral malaria and placental malaria), but requires highly trained staff and specialized equipment, which are not always available in resource-poor settings [11, 12, 15].

Both PCR and RDTs are reported to have a higher sensitivity for placental malaria in peripheral and placental blood than microscopy, but are considered not to be as accurate as placental histopathology, however, evidence for this conclusion has not been summarized [11]. Besides accuracy there are many other reasons for choosing to use a certain type of diagnostic test, such as affordability, number of tests to be performed in a certain time, equipment, trained staff, etc., that depend on the setting and location in which the test will be used. Without a sufficient level of accuracy, however, there is no justification of using a certain test, even if it is practical and affordable and perhaps the only possibility in a certain situation. Therefore, the aim of the present study is to investigate the published diagnostic accuracy of RDTs and PCR for the diagnosis of malaria infection in pregnant women compared to a reference standard. These tests should at least have a better sensitivity and specificity than peripheral microscopy. Different consequences of the results will be discussed.

To estimate the accuracy of RDTs for diagnosing malaria infection in pregnant women, the results of 49 studies were analysed. Few studies exist that fully evaluate microscopy, RDTs and PCR against the gold standard, placental histology, and each other. This makes it difficult to directly compare the accuracies of the different tests, and therefore currently no reliable data about the accuracy of RDTs and PCR for the detection of placental malaria is available. The present study shows that microscopy of both placental and peripheral blood do not detect many placental infections that are identified by histology (sensitivities of 54% and 44%) and cannot reliably replace histology as a reference standard for placental P. falciparum infection. Despite these limitations RDTs, especially HRP2-based tests, appear to have reasonable accuracy compared to microscopy. The World Health Organization (WHO) together with the Foundation for Innovative New Diagnostics (FIND) have performed extensive testing of many RDT devices and there is a great need to compare the best tests from that evaluation with histology, microscopy and PCR in pregnant women [20].

When using placental microscopy as a reference test, PCR has the best sensitivity, followed by RDT and both have higher sensitivity than peripheral microscopy. Peripheral microscopy, however, has the highest specificity, followed closely by RDT. Based on these results, RDTs seem a good alternative for diagnosis of placental malaria. For the determination of the accuracy of RDT and PCR compared to placental blood microscopy, however, tests performed on both peripheral and placental blood were pooled. This might have introduced bias, since the starting material is not the same, and therefore, the accuracy might be different. There were, however, too few studies performed on peripheral blood alone to perform meta-analysis. The pooled summary sensitivities and specificities are not that different from the summary values of these tests using placental blood, but this may be due to the fact that these tests are a large proportion of the pooled analysis.

Compared to the imperfect reference standard peripheral microscopy, the proportion of microscopy positives detected by any RDT ("sensitivity") was 81% [51-95 CI]. The proportion microscopy negatives, with a negative RDT ("specificity"), was 94% [76-99 CI]. As the RDTs seem to miss patients that are positive in microscopy, this is not very reassuring for the value of RDTs for the diagnosis of malaria in pregnant women. Nevertheless, HRP2-based RDTs might have adequate sensitivity (94% [91-96 CI]).

The results in this study suggest that the proportion of microscopy positives detected by HRP2-based RDTs compared to peripheral microscopy is higher than for pLDH-based RDTs. The proportion microscopy negatives with a negative RDT, however, is lower for HRP2 RDTs compared to pLDH RDTs. This pattern is similar to the results of a meta-analysis of RDTs for uncomplicated malaria, and the sensitivity of HRP2-based RDTs for pregnant women (94% [91-96 CI]) is very similar to that for uncomplicated malaria (95% [93-96 CI]) [18]. The specificity compared to peripheral microscopy for HRP2-based RDTs for pregnant women (81% [71-88 CI]), however, is much lower than for uncomplicated malaria (95% [93-99 CI]). A possible explanation for this observation is that peripheral microscopy is not a suitable reference test for placental malaria and does not detect all placental infections, whereas the HRP2 RDTs might be able to detect these infections, resulting in a lower specificity when compared to peripheral microscopy. This is strengthened by the specificity of HRP2 RDTs (90% [84-95 CI]) compared to placental blood microscopy, which detects more placental infections than peripheral microscopy. Direct comparisons of HRP2 RDTs with histology are needed to confirm this.

The sensitivity and specificity of tests vary with parasite density, and many included studies report lower sensitivities at lower parasite densities (two studies for PCR and RDT; seven for RDTs; four for microscopy). This factor is a particular challenge for malaria infections during pregnancy in both high and low transmission settings, especially for multigravid women who are often reported to have lower parasite densities. Of the included studies, 13 report higher parasite densities in peripheral and/or placental blood for primi- and/or secundigravidae. In high transmission areas, women have acquired immunity during their life, and although they might have substantial placental sequestration (especially in primi- and secundigravidae), they can have a lower amount of parasites in the circulation by clearance of infected red blood cells by the immune system. In low transmission areas, women have low or no immunity and women can get sick at initially low parasite densities. Therefore it is very important for the diagnosis of malaria in pregnant women that the test has sufficient accuracy at lower parasite densities, and low parasite density might partially explain the poor performance of pLDH RDTs.

In a previous report, prevalence of placental malaria estimated by different tests was compared within each study [11]. However, even if the prevalence detected by one test is higher than the prevalence detected with the other test, it does not necessarily mean that the same cases are detected. Additionally, in this way it is not clear whether the positives found by each test are true positives or false positives. Nevertheless, the conclusions of that report and this review are broadly similar. Histology found a higher prevalence than placental blood microscopy and placental blood microscopy in turn found a higher prevalence than peripheral microscopy in four out of six studies. In both peripheral and placental blood, in nearly all studies, higher prevalence was found using PCR than using RDT and higher prevalence was found using RDT than using microscopy. No studies compared prevalence estimated by RDT and PCR to prevalence found by histology.

The results presented in this review are mainly applicable to sub-Saharan Africa, as 85% of the included studies were conducted in that area. It is surprising that only a few studies were available evaluating diagnostics in the Asian-Pacific region, considering that most pregnancies at risk in the world are located in this region [6]. In line with this observation, most studies described P. falciparum infections, and only six studies reported P. vivax infections. Meta-analysis was based on studies of which the majority described P. falciparum infections and studies with P. vivax infections had little influence on the outcome. Therefore, conclusions in this review are mainly applicable to P. falciparum infections. Due to the potential pathological differences between P. falciparum and P. vivax infection in pregnancy a difference in diagnostic accuracy of the various tests is expected between the species and more studies should aim at evaluating diagnostics in P. vivax-infected pregnant women in the future. Too few studies with P. vivax infections were included in the meta-analysis to be able to determine if there is a difference in accuracy.

The use of PCR for diagnosis of (placental) malaria remains a matter of discussion, especially in the submicroscopic cases, because it is difficult to determine what the PCR is detecting. It is very sensitive in detecting parasite nucleic acids, but it is unclear if this is a residual from a non-viable sequestered parasite, or a viable parasite or gametocyte. Additionally, there is some discussion whether these low parasite levels detected by PCR are clinically relevant. Several studies have tried to address this issue by researching the association of a positive (submicroscopic) PCR result and outcome measures such as anaemia, low birth weight and premature delivery. A systematic review has been conducted to summarize this data and concluded that the frequency of anaemia is significantly lower in uninfected women compared to women with a submicroscopic infection, although the risk is lower than with microscopic infections, and a similar pattern is found for low birth weight [95]. The review on the effects of submicroscopic infections did not include three studies that have been described in the current study [60, 90, 93]. One of these three studies described a significantly increased risk of anaemia with submicroscopic infections compared to PCR- and microscopy negative women, as well [60]. One of the other studies describes that submicroscopic infection was predictive of low birth weight in HIV positive, but not HIV negative women [90]. More comparisons with histology might shed more light on this issue and show that a significant proportion of the peripheral submicroscopic infections, are in fact placental infections. As mentioned before, pregnant women often have low parasite densities, and tests should have good sensitivity at these low densities; PCR techniques often have better sensitivities at low parasite densities than RDTs and microscopy.