![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London WC1E 7HT,1 Ecology and Epidemiology Group, Department of Biological Sciences, University of Warwick, Coventry, United Kingdom,3 Dirección Regional de Salud Huánuco, Huánuco, Peru2
Received 15 August 2002/ Returned for modification 7 November 2002/ Accepted 15 December 2002
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
When carrying out epidemiological surveys to consider putative canine leishmaniasis control strategies, sensitive and specific tests for identifying ACL-infected dogs are paramount. Serological tests are the standard tools for identifying Leishmania-infected dogs during epidemiological mass-screening surveys because clinical and parasitological diagnoses (e.g., biopsy smears, parasite culture) are typically insensitive and because ACL infections in dogs are frequently asymptomatic (30). However, difficulties associated with the interpretation of cross-sectional seroprevalence data are that (i) serology is prone to nonspecific cross-reactions (false positives, e.g., to Trypanosoma cruzi or Leishmania infantum infection) (37), (ii) there may be a delay between infection and seroconversion (11, 28), (iii) a fraction of infected dogs may never seroconvert (e.g., due to innate resistance) (11), and (iv) seroconversion may not be permanent (e.g., due to development of humoral or cell-mediated immune response) (11, 28). Thus, sensitivity and specificity of serological tests can vary considerably and may underestimate the true prevalence and incidence of disease and, hence, the scale of the control problem.
It has been suggested that PCR-based methods for Leishmania diagnosis may provide the "gold standard" for determining the presence and identity of leishmanial infections not only for diagnosing active cases but also for monitoring parasitological cure of patients after chemotherapy and as a mass-screening tool to detect Leishmania infections in vertebrate hosts or sandfly vectors (35). Various PCR protocols have been developed for the detection of ACL-causing Leishmania in humans by using either purified DNA (from cultured parasites) (9) or clinical specimens (including lesion and scar biopsy specimens [4, 6, 9, 17, 21, 27, 29, 33, 34, 40] or blood [10, 12, 14, 15, 26]) and have consistently been proven to be more sensitive than other diagnostic methods, including in vitro culture of biopsy specimens, biopsy smears, and hamster inoculation. However, at present only one large (>50 samples tested) study has been reported where PCR was used as a diagnostic mass-screening tool to detect ACL in putative animal reservoirs (20). Leishmania (Viannia) parasites were detected in skin aspirates or biopsy specimens from 15 of 276 (5.4%) dogs, 4 of 153 (2.6%) Akodon spp., 2 of 72 (2.8%) Didelphis albiventris, 1 of 499 (0.2%) Phyllotis andinum, 0 of 178 Mus musculus, 0 of 8 Oryzymus spp., and 0 of 8 Rattus rattus. A second diagnostic test for comparative data on infection rates was not carried out, and hence, no conclusions about the PCR assay's sensitivity and specificity can be made.
In the work presented here the utility of PCR as a diagnostic mass-screening tool in epidemiological studies was evaluated. Specifically, we assessed the reproducibility of established PCR protocols on canine blood and bone marrow and compared their sensitivity and specificity to an enzyme-linked immunosorbent assay (ELISA) for detecting Leishmania (Viannia) spp. infection in dogs.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dog population and sampling. Between 1997 and 1999, dogs were surveyed by house-to-house visits or in the village square during rabies vaccination campaigns carried out by the local Ministry of Health. None of the dogs belonged to a recognizable breed and all were guard or hunting dogs. Dogs were examined for clinical signs of ACL, i.e., cutaneous lesions or scars. Impression smears were made of dermal scrapings or lesion biopsy samples from dogs with active cutaneous lesions, Giemsa stained, and examined microscopically (light microscope, oil immersion, x100 objective) for Leishmania amastigotes. Blood (2 to 10 ml) was taken from a total of 1,104 (553 in 1997, 289 in 1998, and 262 in 1999) dogs by venipuncture and processed as described previously, i.e., as buffy coat samples (BCS) and/or guanidine blood lysates (GBL); bone marrow samples (BMS) were taken from a subset of dogs (n = 42) (31). Some dogs (n = 138) were sampled in consecutive years; the epidemiological analysis of Leishmania (Viannia) spp. transmission dynamics in dogs is, however, beyond the scope of data presented here and will be reported elsewhere.
PCR.
DNA was phenol-chloroform extracted from BCS, GBL, and BMS as described previously and amplified using B1 (5'-GGG GTT GGT GTA ATA TAG TGG-3') and B2 (5'-CTA ATT GTG CAC GGG GGA GG-3') Leishmania (Viannia)-specific kinetoplast DNA primers. The PCR conditions and assay setup have been described previously (31). Amplification products were analyzed by electrophoresis on 1.5% agarose gels in 1x TAE (40 mM Tris-HCl, 40 mM acetic acid, 1 mM EDTA [pH 8.3]) buffer before being hybridized to a Leishmania (Viannia)-specific, [
-32P]ATP-labeled B3 (5'-TTG AAC GGG GTT TCT GTA TG-3') probe (31). Using this protocol, we were able to amplify one parasite in 400 µl of blood.
To assess the reproducibility of the used PCR-hybridization (HYB) protocol, all field samples were tested twice; when diagnostic outcomes differed, the samples were tested a third time, with the outcome being definitive. The diagnostic outcome was defined as samples being either positive by PCR and/or HYB or negative by both PCR and HYB. GBL were classed into four different groups: (i) PCR positive (PCR+) if an amplification product could be seen on the agarose gel, (ii) PCR negative (PCR-) if an amplification product could not be seen on the gel; (iii) HYB positive (HYB+) if an amplification product hybridized to the B3 probe; and (iv) HYB negative (HYB-) if an amplification product did not hybridize to the B3 probe. BCS and BMS were classed as either positive (PCR-HYB+) or negative (PCR-HYB-) by PCR-HYB, as the prior distinction as done for GBL could not be made due to the frequency of PCR artifacts (nonspecific bands) on the agarose gel.
ELISA. The ELISA protocol used has been described in detail previously (32), with 105 log-phase L. braziliensis promastigotes (MHOM/BR/76/M2903)/microtiter well used as the antigen. Serum samples were added at dilutions of 1/100, 1/400, and 1/800, and affinity-purified rabbit anti-dog immunoglobulin G (Sigma, Poole, United Kingdom) was added at a dilution of 1/1,500. The ELISA protocol was standardized as described before, with sample readings expressed as anti-Leishmania antibody units (LAU)/milliliter (32).
Negative and positive controls. Three groups of uninfected dog sera were used as negative controls for all diagnostic tests. The sera came from (i) dogs of various ages and breeds which had attended a veterinary clinic in Lima, Peru (n = 18); (ii) mongrel dogs from Belém, Brazil (n = 13); and (iii) dogs of various ages and breeds which had attended a veterinary clinic in Cambridge, United Kingdom (n = 13). Though Leishmania (Viannia) has been isolated from dogs both in Peru and Brazil, Lima and Belém are not areas of Leishmania (Viannia) endemicity. The positive-standard control serum as well as 15 other positive-control sera came from lesion and biopsy smear-positive dogs surveyed in the study.
Data analysis. Sensitivity, specificity, and positive (PPV) and negative predictive values for each diagnostic test were calculated as described previously (32). Analysis of ELISA data and all statistical analyses were carried out in Microsoft Excel 2000.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Field samples: clinical diagnosis. Of surveyed dogs, 21 of 1,104 (1.9%) had active cutaneous lesions and a further 21 of 1,104 (1.9%) had scars. Seventeen of 20 (85%) dogs with active lesions were biopsy smear positive.
Field samples: reproducibility of PCR assay. As described in Materials and Methods, all samples were tested by using a PCR-HYB protocol, with PCR amplification products being visualized by gel electrophoresis (PCR) prior to hybridization to a Leishmania (Viannia)-specific probe (HYB).
GBL. A total of 368 GBL samples were tested at least twice (Table 1). Three hundred ten (84%) of the samples had the same diagnostic outcomes in both replicates, while the remaining 58 samples had to be tested a third time to determine the final diagnosis (Table 1). Hence, a total of 794 (368 plus 368 plus 58) PCR-HYB assays were carried out. For all 22 samples positive by PCR, HYB with the B3 probe detected the B1 and B2 amplification products visible by gel electrophoresis. HYB was positive for a further 90 of 772 (12%) GBL sample replicates where B1 and B2 amplification products were not visible by gel electrophoresis (Table 1), i.e., HYB increased the frequency of positive assays by fivefold.
|
These results demonstrate the essential requirement for assaying all samples at least twice to minimize the possibility of false-positive HYB results or false-negative PCR results.
BCS. A total of 769 BCS were tested at least twice by PCR-HYB. Of the BCS, 701 of 769 (91%) had the same results in the first two PCR replicates, with 39 of 701 (6%) testing PCR-HYB+ and 662 of 701 (94%) BCS testing PCR-HYB-. Of the 68 samples that had to be tested a third time, 26 (38%) tested PCR-HYB+ and 42 (62%) tested PCR-HYB-.
BMS. A total of 42 BMS were tested at least twice by PCR-HYB. Of 42 BMS, 37 (88%) had the same results in both replicates, with 5 of 38 (16%) and 32 of 38 (84%) testing PCR-HYB+ and PCR-HYB-, respectively. Of the remainder, 2 (40%) tested PCR-HYB+ and 3 (60%) tested PCR-HYB- in the third replicate.
Comparison of the reproducibility and sensitivity of PCR for GBL, BCS, and BMS.
No difference between the reproducibility (i.e., in the proportion of samples that were consistent) of the PCR-HYB protocol on GBL, BCS, or BMS was observed (chi-square test, Yates-corrected, 
2 = 0.72, df = 2, P = 0.7). Samples were considered PCR-HYB+ if out of a maximum of three replicates, two replicates were PCR+ and HYB+ or PCR-and HYB+ (for GBL samples only), or HYB+ (for BCS and BMS). Additionally, GBL samples were considered positive if one replicate was PCR+ and HYB+ and one replicate was PCR-and HYB+. Samples were considered PCR-HYB- if out of a maximum of three replicates, two were negative after HYB. Using these definitions, 32 of 368 (8.7%), 65 of 769 (8.5%), and 7 of 42 (16.7%) were PCR-HYB+ on GBL, BCS, and BMS, respectively (Table 2); this observed difference in the proportion positive for GBL, BCS, and BMS was not significant (chi-square test, Yates-corrected, 2 = 3.35, df = 2, P = 0.19). When more than one sample was assayed from dogs, there were highly significant associations between the results: for example, among those 42 dogs with bone marrow samples, all four positive by GBL (and 4 of 4 positive by BCS) were also positive by BMS.
|
2 = 26.8, P < 0.001). ELISA. The mean antibody levels for the three groups of negative controls were 5,584 (standard deviation [SD] = 3,150) (Lima), 2,602 (SD = 1,707) (Belém), and 4,803 (SD = 4,728) (Cambridge) U/ml (Fig. 1). Whereas the mean log anti-LAU/milliliter of Lima and Belém (t test, t = 4.12, df = 29, P < 0.001) and Cambridge and Belém (t test, t = 2.21, df = 24, P = 0.04) dogs were significantly different, the mean LAU/milliliter of Lima and Cambridge dogs were not (t test, t = 1.26, df = 29, P = 0.22) (Fig. 1, inset).
|
A total of 1,059 field dogs were tested by ELISA, the frequency distribution of which is represented in Fig. 1. Two different methods were used to analyze the data. First, the standard definition of positive cutoff (CO), i.e., the mean level of antibody units/milliliter of negative controls plus 3 SD was used (11, 28). The mean level of antibody units/milliliter of all negative controls was 4,472 (SD = 3,299) U/ml; hence, the positivity CO was 14,369 U/ml (i.e., 4.16 LAU/ml). Using this CO, 221 of 1,059 (20.9%) dogs tested positive, 25 of 40 (62.5%) dogs were symptomatic, and 196 of 1,019 (19.2%) dogs were asymptomatic (Yates-corrected chi-square test: 2 = 36.1, P < 0.001) (Table 2). However, this CO is conservative and it is likely that it will underestimate the true proportion of infected dogs (11, 28). Not identifying all Leishmania (Viannia)-infected (and infectious) dogs could have a potentially great impact on the effectiveness of implemented control strategies (e.g., dog culling) should these dogs elude control coverage (11, 28, 32). Hence, a second method to determine the positive CO was used, which took into account that the frequency distribution of log antibody units/milliliter is bimodal (11). Using maximum likelihood, two log-lognormal distributions were fitted to the observed data, varying the mean and SD (Fig. 2). Both distributions yield cumulative probability functions, where the positive CO is the point at which the distributions intersect (Fig. 2), i.e., at 4.1 log antibody U/ml. Using this CO, 230 of 1,059 (21.7%) dogs were positive, 25 of 40 (62.5%) dogs were symptomatic, and 205 of 1,019 (20.1%) dogs were asymptomatic. This proportion of infected dogs is close to the best estimate of 249 (23.5%) dogs predicted by the fit of the log-lognormal distributions (i.e., the area under the log-lognormal distribution fitted to the right-hand tail of the ELISA LAU/milliliter distribution).
|
|
|
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
PCR as a diagnostic mass-screening tool. The classic diagnostic techniques for ACL diagnosis have a number of limitations. Microscopic examination of skin scrapings or lesion biopsy sample touch imprints, though rapid and low cost, has limited sensitivity, particularly in chronic lesions (e.g., mucocutaneous leishmaniasis) (39, 40). While in vitro culture techniques are slightly more sensitive than microscopic examination of lesion biopsy smears or histological samples, they are labor intensive and costly and are susceptible to microbiological contamination (16). They also are hampered by the particular growth requirements of different Leishmania strains, and as some strains grow better than others in vitro, the dominant strains can be inadvertently selected when culturing mixed infections (2). MST detects specific cutaneous delayed-type hypersensitivity but cannot distinguish between active and past infections (39). Also, logistical problems would be associated with its application in epidemiological dog surveys (e.g., dogs would have to be kenneled or monitored for 48 h prior reading the MST induration size). PCR has been shown to be particularly useful for the diagnosis of Leishmania (Viannia) infection, as parasite numbers in clinical samples are typically sparse (4, 6, 9, 17, 21, 27, 29, 33, 34, 40, 41).
Although a positive PCR was associated with higher LAU/milliliter (Fig. 3), the ELISA was much more sensitive than PCR in detecting Leishmania infection in symptomatic (chi-square test, Yates-corrected, 2 = 6.5, P = 0.1) and asymptomatic (chi-square test, Yates-corrected, 2 = 58.9, P < 0.001) field dogs. Considering the 100% specificity of the ELISACO4.16 test and the absence of any potential cross-reacting parasites in the study area, it is very likely that the 163 ELISA-positive but PCR-dogs were true positives (see below). Based on negative- and positive-control samples, the PPV was 100% for PCR and ELISACO4.16 and 81% for ELISACO4.1. The negative predictive value was 80% for PCR and 94% for ELISA. If one estimates the PPV from the PCR GBL replicate data, then the PPV for PCR+ and HYB+ is similar, i.e., 95% (of the 22 PCR assays that were PCR+ and HYB+ at least once, 21 were diagnosed as positive) (Table 1). However, the PPV for PCR-and HYB+ is only 31% (of 80 samples that were PCR-and HYB+ at least once, 25 were diagnosed as positive). This discrepancy is most probably due to samples having parasite numbers equal to or lower than the PCR positivity CO (i.e., 1 parasite per 400 µl of blood) (31). Whether future PCR protocols and new PCR primers will be able to reduce that false-negative rate remains to be established. Replicate results of established PCR protocols are rarely reported (18, 19, 28) but are of importance when interpreting epidemiological data because they show that the diagnostic test used is either not reliable (i.e., there is contamination or lack of reproducibility of the protocol) or not performing at its highest level of sensitivity. Considering that, in our case, a positive PCR result means that a dog is potentially infectious, the lack of replicates may have a significant impact on the outcome on dog intervention (e.g., culling) campaigns.
A PCR-based diagnostic test on blood is advantageous because samples can be obtained less invasively from the patient (human or dog) and are relatively easy to process. We also show that the choice of biopsy material does not appear to be as critical as for studies with PCR for diagnosing human ACL patients (e.g., skin snips and lesion aspirates), as the sensitivities of the PCR on GBL, BCS, or BMS were not significantly different. Studies which used PCR to detect L. infantum in dogs reported similar observations to the ones made here, showing that sensitivity was higher in clinically symptomatic dogs than in asymptomatic dogs (3, 5, 22, 32, 36). Only in two studies did the sensitivity of PCR vary according to biopsy origin (22, 36). The reason we could not demonstrate such an association is probably due to the pathogenesis of canine Leishmania (Viannia) infections, where parasites remain localized at the site of infection (38).
Only one large (>50 tested samples) study has used PCR as a mass-screening tool in epidemiological studies of ACL and reported lower ACL infection rates in dogs than the ones reported here (20). Similarly, other small studies (<50 tested samples) which used PCR to detect ACL in putative animal reservoirs failed to observe high ACL infection rates. Incidentally, in all studies, a second, comparative diagnostic test such as serology was not carried out.
Epidemiological considerations. The high prevalence shown in both asymptomatic and symptomatic dogs provides further evidence for their suspected role as (peridomestic) reservoir hosts of ACL (30). The detection of Leishmania (Viannia) DNA in canine blood implies that infected dogs should be infectious to blood-feeding sandfly vectors, but xenodiagnosis studies will be required to prove this. Although Leishmania (Viannia) DNA was detected in the blood and bone marrow of 8.4% (90 of 1,066) of the dogs tested, indicating that metastasis by hematogenous dissemination may be a more common phenomenon than previously acknowledged (38), blood samples from the majority of dogs (68% [28 of 41]) with active (and biopsy smear positive) lesions were PCR-. This is probably because, following inoculation by sandfly bite, Leishmania (Viannia) parasites are first localized in the dermis, with hematogenous dissemination occurring after an undefined interval (38). An interesting observation is the 52 PCR+ but ELISACO4.16-negative asymptomatic field dogs. Because the PCR was 100% specific, it is unlikely that they were false positives. They may have been false ELISA negatives, but considering that their mean level of LAU/milliliter (i.e., 3.61) was much lower than the 4.16 LAU/ml CO, this seems improbable. One explanation is that, although infected, the dogs had yet to develop a immune response to infection (i.e., prepatent period); 3 of 14 (21%) PCR+ but ELISA-negative dogs that were followed up became ELISA positive. Another explanation is that dogs have developed cellular immunity after self-cure (e.g., studies of L. infantum in dogs have shown that up to 50% of asymptomatic dogs have an anti-Leishmania cellular immune response) (1, 8, 25, 36), as it now becomes increasingly evident that persistence of parasites is required to maintain cell-mediated immunity in the long term (38). Of note is that despite the high prevalence of infection only 4% of surveyed dogs had clinical symptoms of ACL, which is considerably lower than the figures observed in cross-sectional studies of dogs in areas of ACL endemicity (30). This observation could be explained if the circulating Leishmania (Viannia) parasites are of low virulence (i.e., a small proportion of infection causes clinical disease) but high pathogenicity (i.e., a high proportion of parasite inoculations cause infections), which is characteristic of L. peruviana (38). Unfortunately, currently no rapid, easy-to-use diagnostic test exists to differentiate the closely related Leishmania (Viannia) spp., L. braziliensis, and L. peruviana. Surprisingly little known is known about the clinical and parasitological course of Leishmania (Viannia) in dogs (as in humans and other mammalian hosts) and the role of cellular immunity during infection. It also remains to be established whether Leishmania (Viannia)-resistant dogs could have an important role in ACL disease transmission.
This is the first large-scale study to test the feasibility of the use of PCR to detect Leishmania (Viannia) spp. parasites in host blood and to test whether PCR could be used as a diagnostic mass-screening tool in epidemiological studies. As is the case for zoonotic visceral leishmaniasis (3, 5, 22, 32, 36), PCR on blood alone does not appear to be the elusive gold standard for diagnosing ACL infections in dogs (or humans). Unless a more sensitive PCR protocol (e.g., PCR-ELISA or different PCR primers) is developed to detect asymptomatic ACL infections, the use of PCR should be restricted to the diagnosis of active cases (e.g., hospitals in Europe, the United States, and elsewhere increasingly rely on PCR for Leishmania diagnosis in human patients) (6, 41) and, in particular, MCL cases where common diagnostic tests (e.g., parasite culture, biopsy smears, and histopathology) are less sensitive. Nonetheless, the use of PCR on blood will have an important epidemiological application in studies monitoring the clinical and chemotherapeutic follow-up of ACL patients (10, 15, 34). Detection of disseminating Leishmania parasites in patient blood would indicate that they are at risk of developing mucocutaneous lesions, the treatment of which is more complicated than the treatment of the single lesions characteristic of ACL (38). Also, PCR combined with specific DNA probing and sequencing should help to identify and characterize those strains that are drug resistant and that cause the different clinical pathologies associated with ACL.
Thus, although there are concerns about the specificity of serological tests, they will remain the main diagnostic tool for epidemiological Leishmania mass-screening surveys. The main advantages of serological tests are that large numbers of samples can be processed readily and inexpensively, with comparatively low technical expertise required. Significantly, the recent identification of specific recombinant Leishmania antigens (e.g., the Leishmania donovani- and L. infantum-specific rK39) (7) suggests that serological tests may become more specific in the future. Several studies have shown that serological tests based on recombinant antigens (e.g., rK39-ELISA) are able to distinguish not only between present and past infections but between symptomatic and asymptomatic infections as well. Surprisingly, serological tests based on known Leishmania (Viannia) spp.-specific recombinant antigens (23) have yet to be tested as a diagnostic tool either for active cases or in epidemiological surveys.
|
ACKNOWLEDGMENTS |
|---|
This study was funded by the Sir Halley Stewart Trust.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Antimicrob. Agents Chemother. | Clin. Microbiol. Rev. |
|---|---|
| Clin. Vaccine Immunol. | ALL ASM JOURNALS |
|---|
