This is the first molecular epidemiological study of leishmaniasis in the Jericho area. Molecular epidemiology is the application of molecular biology techniques to the study of the frequency, distribution and determinants of infectious and non-communicable diseases. For the first time, molecular diagnostic tools have been applied that lend themselves to the identification of different species of Leishmania as well as to the differentiation of strains of L. major of different origin.
It is in this study that L. tropica was first recognised as causative agent in Jericho, a classical L. major focus, as early as in 1998. The other striking finding came from the microsatellite analysis which showed that two genetically distinct populations of L. major are present in the Middle East. The Jordan Valley and the Arava/Negev desert populations genotypes are found in geographically close areas, less than 100 km apart from each other.
Molecular epidemiological studies will impact the look on leishmaniasis in Jericho district, as it showed differences in epidemiology patterns for both L. major and L. tropica species most likely reflecting different life cycles. This finding will, if taken into consideration by the authorities in Palestine, lead to a revision of control measures and even treatment regimens.
It has always been a dilemma to compare a newly proposed diagnostic method that is expected to be superior especially with regard to sensitivity, as in the case of ITS1-PCR, with other methods in the absence of a valid (highly sensitive and specific) or even acceptable gold standard. All conventional methods employed for the diagnosis of cutaneous leishmaniasis have modest to low rates of positivity, such as microscopy 42% by Aviles et al. (1999); 46.7% by Weigle et al. (2002); 48% by Andresen et al. (1996); 66.7% by Rodrigues et al. (2002); and 72% by Rodriguez et al. (2002), histopathology 33% by Aviles et al. (1999); 76% by Andresen et al. (1996) and 66.2% by Rodrigues et al. (2002) and culture 46.5% by Rodrigues et al. (2002) . Part of this broad range may be explained by the differing parasite densities in various types of materials. Thus, Andresen et al. (1996) had the best results with histological sections (76%) but only 48% in smears from ulcerations.
In the case of ulcerations, more invasive specimen sampling (biopsies versus scrapings) will yield better results. To overcome this problem, a broad case definition for CL was adopted, which depended on the presence of skin lesion(s), epidemiology of the area and compatible clinical history of the patient. This way we obtained the case group of Palestinians, living in an endemic area with high probability to become infected, all presenting typical lesions.
Skin-scrapings in comparison to other materials showed heterogeneity with regard to the number of parasites in a smear, as the parasites are not equally distributed in the tissue and the number of cells fixed on the slide may differ too.
In addition, in all cases of leishmaniasis, the host defense influences the density of parasites. In leishmaniasis diagnostics, the real problem is the cases with only few parasites in the material. To diagnose these cases correctly, one needs a very sensitive method. The method we have used was ITS1-PCR which has been proven as sensitive and specific method for direct detection and identification of Leishmania parasites in different clinical specimens (Schönian et al. 2003). The fact, that there were no positive PCR reactions in the negative group and that all negative controls were negative, confirmed that we obtained real positive results.
Whereas diseases that mimic CL such as leprosy, Blastomycosis, Lupus vulgaris, Yaws, tropical ulcers or skin cancers (Singh and Sivakomar, 2003; MOH, 2004) have not been reported in Jericho, skin disease like Streptococcus pyogenes (Streptococcus group A) infection, benign tumors, and infected wounds stings by insects and flies have been reported (Ministry of Health-Palestine, 2004).
ITS1-PCR proved to be more sensitive (87 %; 52/60) in comparison to graded microscopy regardless whether all samples (37%; 22/60), or only those that were stained correctly (49%; 22/45) were used.
In general, the sensitivity of ITS1-PCR was consistent with results of other studies that evaluated microscopy and PCR-based methods and revealed sensitivity ranging from 75.7% to 100% depending on the targeted DNA sequence amplified, i. e kinetoplast DNA, ribosomal RNA genes, minicircle DNA etc. the type of primers, i.e. genus-specific, sub-genus specific or species- specific and the nature of clinical sample (Piarroux et al., 1994; Andersen et al., 1996; Belli et al., 1998; Aviles et al., 1999; Weigle et al., 2002; Rodrigues et al., 2002; Motazedian et al., 2002).
However, the peculiarity of this study in comparison to the others cited above lies in the standardized procedure by testing the same area and quantity of Giemsa-stained skin scrapings by both graded microscopy and ITS1-PCR and subsequently statistically analyzing the results in matched pairs, which was meant to strengthen and validate the comparison process.
ITS1-PCR as a diagnostic test on skin scraping is not only advantageous over graded microscopy in its increased sensitivity, but also to the following aspects: (a) darkly- stained areas which are not favorable and in most cases difficult to screen by microscopy do not form any barrier against testing with ITS1-PCR. (b) Most of the clinically positive cases that were graded as negative by microscopy were positive by ITS1-PCR, which makes grading, particularly (-) and (+1) that formed most of the results, not a valid diagnostic tool. ITS1-PCR was not only able to diagnose the minimum number of amastigotes, down to one amastigote in 3 of the cases, but also it was sensitive enough to detect traces or sheds of leishmanial DNA which are impossible to detect by grade microscopy. (c) Procedural mishaps such as poor staining hindered or prevented diagnosis using graded microscopy in 3 squares, while it did not prevent diagnosis by ITS1-PCR which saves re-sampling and consequently time, effort and materials. There were two squares which were microscopy positive, grade (+1), and ITS1-PCR negative. One was proved to be PCR-false-negative as a result of DNA extraction failure, which may happen in some cases. In the other case microscopy showed only one amastigote in the whole square (>625 OIF). Larger series have to prove whether there may be a risk of false positive results in microscopy too.
There remain more critical points concerning our evaluation study. First, the predictive values depend on the incidence of a given event (disease). As we selected 20 cases with 3 test areas each and 15 controls, we set an incidence, which is far from real. Thus, it is only allowed to compare the negative predictive values within this study which speaks even more for the PCR. The second point is more serious: our clinical and the control group came from an endemic and a non-endemic area, respectively. That means that they belonged to two different populations but were treated here as samples of one population. We think this is justified because the control group thus comes from an area where the clinical diagnosis was valid. This could be not the case for many endemic regions where there will be always people showing no signs of the disease but bearing the parasite which might be detectable with a very sensitive method such as the PCR. That means if one needs a truly negative group, it can only be found in a non-endemic area.
If we arbitrarily consider the demonstration of amastigotes in a stained smear to be the ‘golden standard’ and compare it with the PCR, then the sensitivity of ITS1-PCR is 87% (52/60) (Table 3.1), compared with 37% in case of microscopy counting the badly stained smears as negative. Of the 45 correctly stained squares tested by graded microscopy, 22 were smear-positive, giving a rate of positivity of 49%. This rate of positivity can be considered as the sensitivity for graded microscopy when assuming the broad CL case-definition based on clinical picture. It is probably time to substitute microscopy as “Gold Standard” of leishmaniasis diagnostics by PCR.
For epidemiological, therapeutic and control reasons PCR is recommended. However, this does not abolish microscopy and in-vitro culture.
The genus-specific PCR primer pair used in this study which was described by El Tai et al. (2000) (see Material and methods) made it possible not only to diagnose leishmaniasis, but to differentiate all relevant old world Leishmania spp. including L. major and L. tropica, the predominating species in the district of Jericho-Palestine (Al-Jawabreh et al., 2004). However, a battery of controls including distilled water and L. turanica as negative and positive controls, respectively, inhibition and DNA extraction controls should be included to ensure reliability and validity of the run.
The results showed that ITS1-PCR-based techniques offer a very sensitive diagnostic method for Giemsa-stained smears collected from patients as compared to graded microscopy similar to that adopted by the WHO for VL (WHO, 1990), also indicating that Giemsa-stained smears used for microscopy can be re-used for ITS1-PCR and that oil, stain, debris and possible inhibitors do not form any obstacle for PCR diagnosis. Yet, making ITS1-PCR and PCR-based methods, in general, more economically and technically attractive in regions of high endemicity and bringing them from research institutions into daily clinical use remains a challenge.
The problem of gold standard is faced by almost all studies that compare different approaches for the diagnosis of Leishmania (Table 1.1). Of the 64 patients tested by all four methods, the sensitivity of the methods using WHO gold standard (positive by either microscopy or in-vitro culture or both) and the combined gold standard (at least one method out of four was positive) was comparable (Table 3.4). However, the gold standard based on clinical and epidemiological features gave low sensitivity for all methods used, either by the 64 patients as the denominator or the entire lot tested by that particular method such as 943 for microscopy or 270 for in-vitro culture. This shows that using the entire database as the denominator for the calculation of sensitivity leads to lowered values due to the sneaking of considerable number of false positive cases into the definition (gold standard) causing inflation of the denominator. The sensitivity will be even lower under the clinical gold standard if there is another disease in the area that mimics CL like Impetigo caused by Streptococcus pyogens which formed 10.8% of the skin diseases reported in Palestine (Ministry of Health-Palestine, 2004).
In all three gold standards, US as a sampling method was shown to give a higher sensitivity than using FP (Table 3.4). The reason for this is that the US has more clinical material which was completely used for the test, compared to two disks taken from FP. Therefore, taking more disks i.e. 4, may increase the sensitivity of PCR-ITS1. Despite the fact that the FPs and USs were stored under the same conditions, collected by the same technician and usually at the same time and that all Giemsa-stained slides that contained numerous amastigotes were always positive when tested by PCR, striking outcomes were noticed. Amongst the 64 cases tested by the 4 methods, 3 Leishmania cases were culture and microscopy positive but PCR negative by FP and US and 10 cases were positive by one PCR method and negatives by the other. This discrepancy causing false negative by PCR may be due to many factors such as the stage of the lesion (active, healing, new or chronic) which affects the number of parasites as the older the lesion the less is the probability of recovering amastigotes (Weigele et al., 1987; Weigle et al., 2002), the uneven distribution or unequal apportionment of amastigotes in the material sampled for different diagnostic approaches (Mathis A & Deplazes, 1995), lack of experience and severe secondary infection with enormous pus cells which may come on the expense of amastigotes, thus impairing diagnostic capability, uneven sampling from one part of the lesion and inhibition, which was minimal (<1%) in this study.
Sampling is a very critical issue, therefore, it is recommended to sample from more than one site in the lesion and to test different spots of the filter paper. Tissue containing the amastigotes should be included, while the oozing blood should be minimized, as it may have PCR inhibitors such as haemoglobin. Whenever clinical samples are involved, it is very important to have a stringent control panel to avoid discrepancies as false positives or false negatives. Furthermore, a defined diagnostic scheme like the one we propose in this study (Figure 3.14) which depends mainly on one primer pair and RFLP should be set and explained to the operating staff. Another scheme was suggested by Schonian et al. (2003) which depends on two primer pairs targeting two different DNA sequences for the sake of increasing sensitivity.
Unlike the USs, FPs have been used or proposed long ago as a sampling method for the diagnosis of different diseases such as trypanosomiasis (Katakura et al., 1997), leprosy (Tomimori-Yamashita et al., 1999), HIV (Nielsen et al., 1987), hepatitis B (Farzadgan et al., 1978) and the famous Guthrie’s test for phenylketonuria (PKU) (Guthrie and Susi, 1963). For leishmaniasis, it was introduced in1997 for diagnosing VL (Osman et al., 1997) and PKDL (Osman et al., 1998), and from 1997 on, we started to collect FPs for diagnostic purposes (Schonian et al., 2003, Al-Jawabreh et al., 2004).
An optimal FP sample (Figure 3.3) is no less than 3 mm thick (Whitman No. 3), sterilized, and properly labeled. Each filter paper close to the standard size of a glass slide (67x26 mm) is preferred to have 3-5 blood/tissue drops that are equal to the diameter of a euro cent, fully immersed in blood and tissue that can be seen from both sides of the FP. On the other hand, an optimal US sample (Figure 3.3) must be properly labeled, clean and have enough material. Both FP and US should be air dried, wrapped in aluminum foil and kept at room temperature till use.
There are several advantages to using FP; first they are light which is very important for international transport and sample exchange. Secondly, they have a long shelf-life at room temperature in which the earliest FP (+) in this study dates back to June 1997. The aluminum foil-wrapped FPs were left on the shelves at temperatures ranging from 12oC in winter to 37oC in summer. Campino et al., (2000) recommended freezing filter paper samples in humid climates. FPs can be used more than once if enough drops have been sampled. Finally, they can be easily processed, as they need only punching of disks. But on the other side, one would note at least one disadvantage for FPs as they require sterilization (UV or steam autoclaving) before use. Although USs also have a long-shelf life, they are heavy, need larger storing area, are difficult to be used more than once, are fragile and scraping material off the slides is time-consuming. The length of storage period, time elapsing from collection to testing, for FPs and USs, depends on the nature of the target molecule. For example, hot and humid storage conditions for 20 weeks caused progressive decline in HIV-1 antibodies (Behets et al., 1992), while in our case, samples were stored for 7 years at a temperature range of 12-37°C as the target molecule was DNA. Therefore, low-temperature storage at temperatures between 4oC and -20°C is recommended when testing for antibodies (Tomimori-Yamashita et al., 1999), while DNA can be stored for a long time at room temperature.
Filter papers (FPs) and unstained smears (USs) are secure tools for an optimized and well-controlled molecular-based assay for direct clinical diagnosis of leishmaniases as well as for strain genotyping in terms of long shelf-life, relative non-invasiveness compared to culture and histopathology. PCR diagnosis using these specimens is simple and relatively quick, not requiring prior cultivation of the parasites. In-vitro culture is prone to contamination (4.8%) and can be higher for field work (El-Tai et al., 2001) as well as the need for different fastidious ingredients depending on the species as some Leishmania species grow faster than others (Schuster & Sullivan 2002 and references therein).
The sensitivity of the conventional diagnostic methods, microscopy and in-vitro culture is lower (Table 3.4) than that of PCR techniques, 78% for FP 86% for US and 87% for stained smears. This runs parallel with other studies (Table 1.1). On the other hand, some studies produced conflicting results as they showed higher sensitivity of smear than PCR. This discrepancy (Table 1.1) is due to different gold standards being used to define a case of CL. In a study by Romero et al (2001), in-vitro culture was the gold standard. Such a poor gold standard produced a smear sensitivity of 95%. Belli et al (1998) used microscopy as a gold standard to obtain a sensitivity and specificity of 100%. Due to the lack of a good gold standard, Weigle et al (2002) used laboratory and clinical criteria to define a case of CL, which proved the higher sensitivity of PCR diagnosis. Another source of discrepancy is the high prevalence of CL in the study area, which may result in an increase of the sensitivity of all types of tests (Sharquie, et al., 2002). To prevent this, control groups of patients with non-leishmania lesions according to the gold standard used, which unfortunately most studies lack, should be involved. In this part of our study, the control group (15 Germans) was dropped as the ratio of case-to-control was less than 1 (Gordis, 1996) which leads to lowering others statistics like specificity and predictive values. The sampling method plays also a role in the comparisons by impacting overall sensitivity.
The comparison between traditional diagnostic methods and PCR with its various types was the object of several studies in the last decade (Hernandez-Montes et al., 1998; Katakura et al., 1998;; Breniere et al., 1999; Pirmez et al., 1999). In these studies, many types of specimens, sampling techniques, DNA targets and primers were proposed as candidates for diagnosing leishmaniasis. The clinical specimens and sampling methods used were: skin/lesion biopsies and syringe-sucked lesion aspirates for culture (Mathis & Deplazes, 1995, Matsumoto et al., 1999), blood and tissue spotted on filter papers (Osman et al., 1997; Harris et al ., 1998; Färnert et al ., 1999; Campino et al., 2000; Schonian et al., 2003; Al-Jawabreh et al., 2004), cotton swabs (Mimori et al., 2002), wooden toothpick (Belli et al., 1998), serum (Fissore et al., 2004), formalin-fixed biopsies obtained from u1cer lesions (Mimori et al., 1998), Giemsa-stained smears (Motazedian et al., 2002), formalin-fixed and paraffin-embedded tissue specimens (Laskay et al., 1995; Momeni et al., 1996) and even conjuctival samples from animals for PCR diagnosis of VL were proposed (Strauss-Ayali et al., 2004). The DNA targets ranged from conserved and variable regions of major classes of kinetoplast minicircle DNA (kDNA) (Bhattacharyya et al., 1993; Lachaud et al., 2002), sequenced whole kinetoplast DNA minicircles (de Bruijn and Barker, 1992; de Bruijn et al., 1993; Lopez et al., 1993, Anders et al., 2002); partial sequence, representing the most variable part of the small subunit ribosomal RNA (SSU rRNA gene (Van Eys et al., 1992), mini-exon gene (Brecelj et al., 2000), and ribosomal internal transcribed spacer 1 (ITS1) separating the genes coding for ssu rRNA and 5.8S rRNA (El Tai et al., 2000; Schonian et al., 2003). This wide range of alternatives may reflect enormous activity in scientific research for improving diagnostic capabilities; yet, this has brought us into such a puzzling situation that one would not be able to tell which is the correct or the reference method for diagnosing different types of leishmaniasis. In a sense and in addition to sensitivity and specificity, an ideal PCR setup for an endemic area should be reproducible, a parameter rarely tackled by studies. The reproducibility should be with-in run, between-runs and between laboratories. One-step PCR for differentiation of the prevalent species in areas such L. tropica and L. major in Jericho would be useful with the possibility of further extending characterization i.e by RFLP when needed. A screening field-PCR that can be applied outside the laboratory during field visits for rapid diagnosis and differentiation with minimum use of instrumentation and procedures is greatly desirable. The aim of such rapid PCR is to minimize the number of samples for further investigation and to give crucial decisions related to treatment.
It is very unfortunate that the researchers world-wide are still unable to introduce PCR as part of the operational case definition of leishmaniasis. For this reason, microscopy and culture, despite their low sensitivities, are still the reference methods for diagnosis of CL, and serology for MCL and VL (WHO Recommended Surveillance Standards, 1997).
PCR for clinical diagnostics has proven more sensitive than the other methods, is well-controlled (Table 3.5), and applicable for direct clinical diagnosis as well as genotyping of strains. Yet precautions should be taken as Reithinger et al (2003a) and others concluded that ELISA (81%) is more sensitive than PCR (31%) for active canine leishmaniasis. Calls by Schallig and Oskam (2002) and others to standardize diagnosis for leishmaniasis should be heeded as they come in parallel to the idea of incorporating PCR in the operational case definition for leishmaniasis in the future. The adaptation of PCR in the WHO operational case definition of leishmaniasis, exactly as in the case of typhus, smallpox, SARS or Anthrax (WHO, 1999), will free us from adopting the low sensitive smear and/or culture methods or clinical manifestation as 'gold standard' for method comparison, treatment clinical trials, control assessment studies, etc. Some researchers also favour approaching the gold standard by PCR (Vega-Lopez, 2003) whereas others argue against PCR being the only diagnostic 'gold standard' (Reithinger and Davies, 2002).
Restriction analysis of the ITS1-PCR positive samples revealed the co-appearance of L. major and L. tropica in Jericho and its immediate vicinity (Figure 3.2). The existence of L. major as causative agent of zoonotic CL in the lower Jordan Valley, including the margins of Jericho, has been well-documented (Schlein et al. 1982 and 1984; Al-Jawabreh et al. 2004). The presence of L. tropica in the human population of the Jericho area is a novel finding. This species has been, however, recorded as causing human cases of CL in a focus at a higher altitude halfway between Jerusalem (Al-Quds) and Jericho (A’riha) (Klaus et al. 1994; Jawabreh et al., 2001). There is clear overlap of the two species in almost all the populated areas included in this study, particularly in Jericho and the closely adjacent areas, i.e., the refugee camps. This change in the distribution of leishmanial species in the study area probably started more than a decade ago after the withdrawal of the Israeli Army when there was an extensive movement of the Palestinian population to and from the Jericho area. This included security forces, internal tourists, workers, farmers and Bedouin shepherds from all the other Palestinian districts, including the Gaza Strip. Although 92% of the individuals with L. tropica claimed that they did not leave Jericho during 3 months prior to the appearance of their lesion; for the present, one has to assume that the cases of CL caused by L. tropica are imported. Because of the significant difference in climatic conditions between the Jordan Valley and the mountains to the west of the Jordan Valley in both summer and winter, many of the people live in Jericho area during the winter and spend the summer in the hills surrounding Ramallah, Nablus, Jenin and Hebron (Al-khalil). Normally, they are in the hills when there are sand flies and transmission is occurring in both the mountains and the Jordan Valley. They arrive in the Jericho area only after the transmission period has ended. L. major does not circulate in the hills, but in the Jordan Valley where sand rats, the animal reservoir are found. Infections occurring in the migrant population are therefore most probably incurred in the hills, and caused by L. tropica. The parasites are then brought by the infected people to Jericho. With the absence of sand flies during the winter, these patients are, however, unlikely to be a source of CL caused by L. tropica in the Jericho area. To prove whether at least some of these cases were autochthonous, it should be confirmed that these people had not travelled out of the area for at least one sand fly season. Also one has to consider that the incubation period of CL caused by L. tropica seems to be much longer than that caused by L. major. It has to be emphasized that people always travelled between the hills and the Jordan Valley, yet in the past cases of CL, caused by L. tropica were extremely rare throughout the whole District of Jericho, including the mountainous regions. Now, more and more cases caused by L. tropica are being seen in many foci of this part of the Eastern Mediterranean region (Klaus et al. 1994; Anis et al. 2001; Nimri et al. 2002; Jacobson et al. 2003; Schnur et al. 2004). This indicates that there has definitely been a change in the epidemiology of CL in this area which appears to have started in the hills and spread down to the Jordan Valley.
Alternatively, one should search in the Jericho area for the presence of sand fly vectors, such as Ph. sergenti, which was found to transmit L. tropica in a focus at higher altitude just east of Jerusalem (Schnur et al. 2004) and Ph. arabicus which was found to harbour L. tropica in Tiberias (Jacobson et al., 2003). Detection of sand flies infected by L. tropica parasites and, if there is one, also animal reservoir hosts would provide evidence for endemicity of the parasite in Jericho area. A pilot study in September 2004 revealed the presence of Ph. sergenti in Jericho City and outside it (unpublished data), supporting the idea of autochthonous L. tropica cases. Schlein et al. (1984) also found this vector at Ein-Gedi, 40-50 km south of Jericho, on the western shore of the Dead Sea and in the Arava, which is even further south. The presence and increasing numbers of human cases in Jericho area caused by L. tropica are a definite change compared with the past when all CL cases from this area, from which parasites were isolated and identified, were shown to be due to L. major only.
Although the determinants of Leishmania infection such as the parasite, the vector, the reservoir and environment, are multi-factorial, the two demographic variables age and sex have been reported to be a source of variation in exposure to CL. Host behavioural/cultural factors were thought to make males more exposed than females, and lack of immunity to make children more exposed than adults (Arda and Kamal; 1983; Greenblatt et al., 1985; Al-Jawabreh et al., 2003). This gender-age pattern was noticed in CL in other parts of the world like Brazil (Jones et al., 1987) and, moreover, also in VL (Abdeen et al., 2002; Shiddo et al., 1995). In the current study, where causative agents of the disease were identified at species level by molecular techniques, it was attempted to find out if the patterns of sex-age distribution were different between CL cases caused by either L. major or L. tropica. The total incidence of CL was higher in males than in females and this tendency was even more pronounced in CL caused by L. major (M: F ~2:1) (Figure 3.7) rather than those caused by L. tropica (M: F 1:1). In one study conducted by Reithinger et al. (2003b) in Kabul-Afghanistan, L. tropica cases were more frequent in females and with increasing age. Although it has been reported that boys often developed VL threefold more than girls (Shiddo et al., 1995); in our study no differences were observed between male and female children (<14). In this age group both genders had the same risks of infection. The gender-age difference was clearer in adults between 20-29 years. These results favour the explanation that differences in exposure to the disease is based mainly on host behaviour for adults in their 20s. For children, it may be due to naivety of the immune system and to genetic factors which have been suggested to explain the sex-age variance in other parasitic infections such as malaria and schistomiasis (Cooke et al., 2003; Henri et al., 2002). Immune regulatory mediators have been shown to develop with age (Sack et al., 1998; Tsaknaridis et al., 2003). Using hamsters infected in the laboratory it was found that the burden of Leishmania infection was more on males than females and this was attributed to different levels of sex hormones and of cytokines known to promote experimental leishmaniasis: interleukin 4 and transforming growth factor (TGF) (Travi et al., 2002).
The seasonal distribution of CL from October to April mentioned earlier by Al-Jawabreh et al (2003) has been confirmed in this study (Figure 3. 9). Another study by Anis et al 2001 on patients from 1971-2000 showed some difference in seasonality, in which cases start to peak from June till December. This study does not, however, mention the geographical origin of the cases, making it difficult to explain the discrepancy.
The reasons for seasonality are, in general, the activity of sand fly vectors and the incubation period of the infection. L. major has an incubation period that ranges from a few weeks to a few months, averaging from 1 week to 3 months. It is longer for L. tropica which could range from 2 -24 months (Harrison’s Principles of Internal Medicine, 1991). Ph. papatasi, vector of L. major, has its activity and abundance peaks in the spring (April) and the autumn (October), and was found to be not active in mid-summer (July) (Wasserberg et al., 2002). Nevertheless, the seasonality pattern in our study suggests a continuous sand fly activity from April to October. Using molecular methods that allowed identification of Leishmania parasites into L. major and L. tropica did not add much to the overall picture of seasonality patterns (Figure 3.9).
Over the ten-year-study period, three peaks were recorded (Figure 3. 8): 1995, 2001 and 2004. Several explanations can be put forward to explain surges of CL (Neoumine, 1996). In 1994-1995, non-immune population was introduced into an active focus like the deployment of Palestinian soldiers. Urbanization process, which is believed to put the host closer to the reservoir that had taken place in the late 1990s, would be a plausible explanation for the 2001 peak. The dramatically increased prevalence of iron deficiency anaemia among Palestinian children and women (Abdeen et al., 2002; Al-Rai, 2005) and changing in farming patterns related to decreased water resources (Neoumine, 1996) can be presented as other possible explanations in addition to urbanization.
Climatic changes and global warming is another important factor that plays a role in the sand fly infections (Kuhn, 1999) which needs to be further studied in the district of Jericho.
In this study, the correlation between the number of cases and annual rainfall was very weak, yet Anis et al (2001) concluded the opposite.
PCR/RFLP has improved the epidemiological understanding by showing L. tropica peak taking place in 2001 while 2004 for L. major (Figure 3.8); thus, once again, proving the necessity of using PCR/RFLP or similar methods for species diagnosis.
Remarkably, both purely spatial and space-time statistics, proved the necessity of molecular diagnosis in the epidemiology of leishmaniasis as a parasitic infection. The spatial and /or space-time scan statistics have been successfully applied to both retrospective and prospective surveillance of various diseases including breast cancer (Gregorio et al., 2001; Sheehan et al., 2004), Creutzfeldt-Jakob disease (d’Aignaux et al., 2002), and systemic sclerosis (15), sudden infant death syndrome (George et al., 2001), West Nile virus (WNV) (Mostashari et al., 2003), bovine spongiform encephalopathy in France (Abrial et al., 2003) and prostate cancer (Klassen et al., 2005). In the current study, SaTScan statistic is used for the first time for spatial and spatial-temporal clustering of CL and, facilitated by the use of molecular methods, PCR and RFLP.
Each of the two leishmanial species had different clustering patterns. In purely spatial statistic, the most likely clusters for L. major were in Fasayil and A’uja villages (RR 4.607 and 2.438, respectively) while that of L. tropica was in Zubaidat village (RR 14.866). Also, the clustering of total CL, (CL= genotyped and non-genotyped cases), revealed more and different secondary clusters (4) than for L. major (2) and L. tropica (2) alone (Table 3.8, Figure 3.11). However, purely spatial analysis has the drawback that the power of detecting recently emerging clusters is effected by the length of time period analyzed (Kulldorff et al., 2001).The space-time statistic was put into effect to overcome this dilemma.
The space-time statistic also presented clearer differences in distribution between the two Leishmania species. The most likely cluster for L. major were Fasayil and A’uja villages in the years 2003-2004 (RR 4.587) which differed from L. tropica alone and from total CL (Table 3.9, Figure 3.12). Zubaidat village was the most likely cluster for both L. tropica and total CL. The time of appearance was, however, different; it was the period 2001-2004 for total CL in contrast to 2001-2002 for L. tropica alone. The temporal length of the Nabi-Musa cluster for CL and L. tropica was relatively long (2000-2004) and the same applied for the L. major clusters in Jericho city and the adjacent village of Nuaimeh (Figure 3.12). The plausible explanation for this is the endemicity of CL and L. major in these areas and that ideal determinants existing there support the flourishing of leishmaniasis for long periods. The following factors have been reported in the literature: i) climatic and topographical factors like surface temperature, soil type, vegetation and rainfall, although the latter lacks high correlation; ii) human activity and behaviour such as farming (Schlein et al., 1984; Neoumine, 1996; Ashford, 1996; Wasserberg et al., 2003); iii) the people’s utilization of available health resource which insures good reporting of cases; and iv) demographic factors like age and sex. The latter two variables are, however, most unlikely to have a great input in our study as they are even and uniform all over the district of Jericho (Booth and Dunne, 2004). Another factor is genetic predisposition leading to different host immune responses to parasitic infection, (Blackwell, 1996; Chang and McGwire, 2002; Desjeux, 2004), malaria (Stirnadel et al., 2000; Aucan et al., 2003; Cooke et al., 2003) and schistosomiasis (Bethony et al., 1999; Henri et al., 2002; Rodrigues et al., 1996). The long-term clusters identified in our study are, however, of less importance than the emerging or disappearing cluster like the L. tropica cluster in Aqbat-Jaber refugee camp in 2001-2002 (Fig 3.6-c) because new foci need new, prompt and timely action plan for therapy, control and health education.
It is worth noting that 1995 witnessed a surge in cases as many Palestinian soldiers and their families returned from the Diaspora or came down to Jericho from other Palestinian districts following the 1994 Oslo agreement (Al-Jawabreh et al., 2003). The protective relative risk of 0.686 was, however, non-significant (p=0.7620), as only 200 cases were observed instead of the expected 245 cases. The reason for this was that not all cases were reported in the relatively large city of Jericho as the surveillance program had been just newly initiated. Also, in Jericho City the L. major clustering showed significant protectiveness (RR 0.867, P=0.0140). We have to bear in mind that Jericho has the largest population in the Jordan Valley compared to the surrounding villages, refugee camps and Bedouin encampments, and SaTScan considers population size (40000) as background in its calculation.
The two spatial and space-time statistics show that L. tropica clusters are in the far north and far south of Jericho district. The Zubaidat village lies on rocky foot hills with sporadic caves ideal for sand fly vectors. Also in 2001, packs of rocky hyraxes were seen around the village and sometimes invading it. Nabi-Musa in the south, the other L. tropica focus, has the same topography mixed with semi-aridity as it is mid way between Jericho and Jerusalem. On the other hand both statistics indicate that L. major clusters occur in Fasayil and A’uja villages, and areas in and around the city of Jericho. These foci are characterized by soft alluvial soil and agricultural land suitable for the reservoir Psammomys obesus. An interesting secondary cluster in the L. tropica space-time statistic is the Aqbat-Jaber refugee camp which was recorded only during the period 2001-2002. The relative risk is not high and close to 1 (RR 1.029, p=0.015), but the refugee camp is adjacent to Jericho city, meaning that L. tropica is closing to the city, known as a classical focus of L. major. Early signs of the presence of L. tropica adjacent to the city gained support later as we have been able to catch Ph. sergenti, vector for L. tropica, in Wadi-Al-qelt (unpublished data). This wadi is a steep valley full of small caves and a relatively high plant coverage providing the city with water coming from the hills above. This valley which is very close to Aqbat-Jaber refugee camp could be the platform for L. tropica to enter the city, particularly as L. tropica has been isolated near the springs of the wadi 10 km away (Schnur et al., 2004). On the other hand, it is most unlikely that the northern focus of L. tropica at Zubaidat Village will affect the city in the near future because there are 60 km of wilderness between them.
One of the limitations of the current study is related to a village in the northern part of Jericho called in Jiftlik with a population of 4127 (PCBS, 2005) spread over a relatively large area. The people in this village seek their health service in a city other than Jericho causing under-reporting of cases of CL in our database. This explains why this village never appeared as a cluster and puts forward that aggressively diligent surveillance and good diagnostic facilities are pre-requisite to successful cluster analysis. The other limitation is the southern focus of L. tropica in Nabi Musa and Khan-al-ahmar area between Jericho and Jerusalem which is a semi-arid and desert place with few sporadic nomadic Bedouin encampments scattered over a relatively wide area. The population of these encampments is very small and this lead to considerably higher number of observed cases (17) than expected (0.73) unlike the ‘dilution effect’ seen in the City.
PCR as a molecular–based diagnostic method have had an eye-apparent effect on the epidemiological picture of leishmaniasis in Jericho. The classical epidemiology represented by the scan statistics of CL cases diagnosed by non-molecular (conventional) methods like microscopy and/or culture, that cannot distinguish between the two species, lead to different clustering patterns compared to the cases where molecular species identification by PCR was employed. This difference in epidemiology proves that genotyping of leishmaniasis in the district of Jericho is crucial, as we are dealing with two aetiologies having probably different life cycles and may need also different control measures.
The present study emphasizes the importance of utilizing SaTScan together with molecular epidemiology of leishmaniasis. It can be run annually as in this study or monthly and even weekly depending on the efflux of cases to monitor the development of CL over small periods of time. And furthermore, cluster analysis parameters can be adopted and calibrated depending on the situation, total population and disease prevalence like for example the meningococcal study conducted by Vogel and Elias at the University of Wuerzburg-Germany (personal communication) in which the maximum cluster size was 10% and the maximum temporal cluster size is 30 days, rather than 50% and one year used in this study, respectively. SaTScan is recommended as an early warning tool for systematic and periodic geographical disease surveillance (Kulldorff et al., 2001) for the following reasons:
The other method for public health surveillance, Shewhart’s Chart, has proved to be a simple and quick method for detecting trends, shifts and outbreaks in CL. On the other hand, this method has a few limitations when used for public health and epidemiological purposes. A major limitation is the need for relatively large numbers of “historical’’ data to develop a reliable target value, or centre line, and subsequently the in-control limits. A minimum of 20-30 numbers is required (Handbook of Statistical Methods; Vermaat et al., 2003). This limitation can cause false alarms and results different from what is assumed (Handbook of Statistical Methods; Quesenberry, 1993). To minimize this limitation, seasons instead of years can be used as units of time. Another problem is the failure to predict future behaviour of CL. The plotted chart would provide information only for the present situation depending on ‘good’ historical data. Shewhart’s Chart is a simple informative tool that can be readily applied in the field of molecular epidemiology for evaluating the historical and current status of CL in terms of trends, shifts and outbreaks. The same argument applies to the moving average technique. SaTScan, Schewhart’s Chart and moving average are useful early warning systems.
Ten microsatellite markerslocated on five chromosomes were used to analyse genetic variation in 106 strains of L. major from 19 countries in Asia and Africa. Distance-based method such as DPS and Ddm using NJ and UPGMA and model-based methods such as Structure which is based on an admixture model were used for the analyses.
Two problems related to the study setup itself have to be mentioned: sample size and number and nature of microsatellite markers in-use.
Population size is critical in population genetic studies; whether it is either total samples size or sample size from each geographical area. In studies on chicken and humans, 600 individuals were tested with 27 markers and over 1000 individuals from 52 populations by 377 microsatellite loci, respectively (Rosenberg et al., 2001, Rosenberg et al., 2002). It was found that using 10 individuals instead of 5 per population improved clustering by more than 90% and that accuracy of clustering decreases with decreasing number of samples (Rosenberg et al., 2001 and 2002). The question of how many individuals should be analysed per population and how many markers should be used is always raised in population studies. The presence of genetically atypical isolates or drastic genetic drifts or recombination will prevent having a 100% accurate clustering no matter how many markers or individuals are used. In general, the more samples from more areas collected and analyzed, the more accurate clustering will become. Yet, collecting samples of Leishmania is not an easy task.
In our study the number of collected isolates of L. major from Asia and Africa was 106. However Strains from Middle East (Southwest Asia) and Central Asia formed, 65% (69/106) of the total sample, and the rest are strains from 16 countries with some area having one isolate only. Pritchard et al (2000a) hints that, unlike Dps-based phenetic analysis (NJ, UPGMA) which is not affected by the population size (Mountain et al., 1997), Structure improves with larger samples sizes in each population. In our study the analysis by Structure was consistent and in agreement with both NJ and UPGMA bootstrap-supported Dps-based phenetic analysis, meaning that latter supported the results of the former. However, Ddm-based tree did not match Structure analysis which may be due to robustness of Ddm to population size fluctuations (Takezaki and Nei, 1996) and its weakness in cases of closely related populations (Goldstein et al., 1995b).
The proper selection of the right type and number of microsatellite marker plays a crucial role for their use in genetically structuring populations. As a rule of thumb, the higher the cross-population the less within-population variability (heterozygous) and the greater within-population homogeneity and across-population polymorphism is for genetic markers the more accurate and powerful genetic population structuring becomes (Reed 1973; Shriver et al. 1997; Rosenberg et al., 2001). Rosenberg et al. (2001) found that clustering was more accurate when 12-15 highly variable and informative markers were used rather than 6-7 markers. Therefore, it becomes challenging for such genetic studies to find the smallest optimal amount of genetic microsatellite markers for differentiating populations. To infer the genetic diversity of CL caused by L. major, a set of 10 independent microsatellite markers was developed which represented 5 chromosomes and proved to be sufficient for population analysis.
The main criteria for establishing the optimal clustering potential of the markers were: 1) highest number of alleles provided, 2) highest expected heterozygosity, and 3) highest Fst, which quantifies the between-population component of genetic variation. However, the former two were superior to the third. In general, if extremely informative markers (highly heterozygous) are available, a minimized number of loci will be needed (Rosenberg et al., 2001).
If we are able, using the above mentioned criteria, to utilize the most variable markers, this will minimize the laboratory work, in terms of resources, human and financial, thus increasing efficiency and cost-effectiveness. Yet over-using markers with maximum variability may lead to inflation of divergence times estimated by using genetic distance (Goldstein et al., 1995) and will bias population growth statistics (e.g., Zhivotovsky et al., 2000) and, hence, should be used cautiously. Using few numbers of markers will maximize the effect of homoplasy. This means that optimal number of samples and markers will always be an active debate.
Both distance-based methods using Dps and model-based methods using admixture model represented by the trees (Figure 3.18) and the plots of Q by Structure (Figure 3.16), respectively, gave comparable results proving that clustering did not follow geographical pattern. This was confirmed by the great genetic isolation between the populations identified. The clearest evidence of this, are the two genetically distinct and geographically proximal clusters, Jordan Valley (ME1) and Negev (ME2) that are separated by less than 100 km.
A peculiar case is the Central Asian cluster (n=39) which represented a very homogenous subgroup that remained solid after several runs and began to split off only at K=6 and re-joined again at K=7 in the Structure analysis. In addition, it had the lowest mean number of alleles (Figure 3.19). Many strains share identical genotypes. Eleven strains from Termiz foci in UZ isolated at the same time (year) shared the same number of alleles. Nine isolates from PS and IL also shared the same genotype, but the time elapsing between isolation and analysis ranged from 1 year to 36 years. The most plausible explanation for this is that this population emerged only recently and its evolutionary process is short. The isolation of Central Asian samples might be due to the fact that deserts create a barrier to genetic exchange by reducing the probability of contacts with other subgroups leading to a reproductive isolation, thus, gene flow is restricted to within-population leaving heterozygosity extremely low. Also, large population size may minimize the effect of genetic drift.
A probable explanation to the peculiarity of the central Asian foci lies in more than just genetic factors, but rather in the mechanical system supporting L. major. The life cycle is different from those in Middle East and African. The Reservoir in Central Asia is Rhombomys opimus while in the other two areas it is mainly P. obesus. In almost all cases, R. opimus was found to host, two types of leishmania species i.e. L. turanica and L. major, with the former being non-pathogenic for humans while the latter is. Even three leishmania species including L. gerbelli can co-exist the R. opimus (Strelkova et al., 1997). This phenomenon of mixed infections in the R. opimus is not seen in P. obesus. Furthermore, the vector in the Middle East is P. papatasi while its existence is not definite in Central Asia.
The African strains are the most genetically variable, as shown by the highest number of alleles found for these strains (Figure 3.19). This supports the theory of an African origin for L. major (see below: Origin of L. major and bottleneck theory). Two Turkish, one Iraqi and one Iranian strain grouped with African isolates which was displayed both in predefined (K=5) and non-predefined (K=7) genetic Structure analyses. The African clusters were more inter-mixed compared to the large representative samples such as those from Middle East and Central Asia. Thus, the difficulty of clustering in Africa may have been exacerbated by small sample sizes and the fact that little information can be drawn from a few isolates. Nonetheless, if it is agreed that phenetic analyses, in general, are less prone to the under-sampling effect, this would support the Structure results as they agree with cladograms.
Using Structure, it was possible to determine the membership of strains of unknown origin. Using population identification by relying on geographical origin of sampling which were 5 areas (K=5), Structure efficiently assigned the unknown strain of L. major MHOM/WA?/87/NEL2 isolated from a traveller to east rather than West Africa. This proves that predefined labels such as geographical origin of sample were highly informative about membership in genetic clusters (Figure 3.17).
The decreasing pattern of genetic diversity of strains of Leishmania across Africa to Middle East and, finally, Central Asia can be explained by the occurrence of a population bottleneck (or genetic bottleneck) which is the reduction of a population by more than 50% due to an evolutionary event. In this case, a population suffers from immediate and transient increase in heterozygosity as a result of rapidly loosing rare alleles (Cornuet & Luikart, 1996; Hoezel, 2002). In the long run, after generations of bottlenecking, the genetic diversity or polymorphism which will deprive this particular population of facing environmental changes is extremely reduced, this leads to extermination (Hoelzel, 1999). The bottleneck phenomenon is bound by the size and growth rate of the population which are inversely proportional to the bottleneck. The time it took for change to take place, usually very long, is another factor impacting a bottleneck. Based on this, the high degree of heterogeneity may indicate that Africa is the origin of L. major. The highest number of alleles amongst African genotypes may indicate that L. major stemmed from Africa and poured, geographically, into the Middle East, which has a lower number of alleles, until it reached Central Asia, with the lowest number of alleles. This process had an effect on a certain population of L. major in time and place dimensions. Hence, the low genetic variation in Central Asia may be due to i) an ancient genetic bottleneck event indicating long-term effective population size. ii). Demographic decline process a long time after a bottleneck event had taken place, followed by inbreeding in an isolated environment, and/or iii) several bottlenecks that exterminated the original L. major population leaving a small pocket that survived and continued its in-breeding with a certain degree of isolation (O’Brien et al.,1987; Gottelli et al., 1994).
Finally, the results of the studies on genetic variation can be viewed in a molecular epidemiological and public health sense. Information on the distribution of L. major genotypes, their movement patterns and their origin could influence treatment and chemically-based control methods. The short term challenges are to correlate differences in genotypes with different issues like the appearance and persistence of variants escaping immunity or the emergence of drug-resistance, pathogencity and susceptibility to CL. Although microsatellite analysis has proved to be a superior tool for genetic diversity studies, it did not reduce cost or labour even when it became fully automated, and this is another challenge.
The success of this study, and any study, is measured by its concrete commitment to the goals and objectives originally set. The conclusions and recommendations in the light of the study objectives are as follows:
Molecular clinical diagnosis
Molecular-based techniques, especially ITS1-PCR and RFLP, are essential tools for clinical diagnosis and strain genotyping of leishmaniasis. Basically, knowledge of the infecting species will provide a guide to appropriate treatment and improve control measures.
Molecular strain typing and genetic diversity
For research purposes it is advised to carry out multilocus microsatellite analysis for studying genetic heterogeneity in strains of both L. major, as mentioned in this study, and L. tropica. However, for this to succeed, mutual cooperation between countries of the area should exist. The current political turmoil in the Middle East makes the interference of a third party to act as a facilitator in such studies something unpreventable.
|© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.|
|DiML DTD Version 4.0||Zertifizierter Dokumentenserver|
der Humboldt-Universität zu Berlin