Testing urine samples with rK39 strip as the simplest non-invasive field diagnosis for visceral leishmaniasis: An early report from eastern IndiaRP Goswami1, RP Goswami2, S Das1, Y Ray1, M Rahman1
1 Department of Tropical Medicine, School of Tropical Medicine, Kolkata, West Bengal, India
2 Department of Medicine, Institute of Post Graduate Medical Education and Research, Kolkata, West Bengal, India
Correspondence Address: Source of Support: InBios International Inc., Seattle, USA provided the rK39 rapid diagnostic kits for Kala Azar (Kalazar Detect™), Conflict of Interest: None DOI: 10.4103/0022-3859.101378
Source of Support: InBios International Inc., Seattle, USA provided the rK39 rapid diagnostic kits for Kala Azar (Kalazar Detect™), Conflict of Interest: None
Background: Diagnosis of visceral leishmaniasis (VL) is a major obstacle in the control of this disease. The rK39 strip-test using patient's blood is a breakthrough; however, it still requires a blood sample, which is a concern for safety in the field. We tried to simplify the test using the patient's urine instead of blood. Aims: To observe the sensitivity and specificity of the urine test in comparison with the blood test. Materials and Methods: We tested active and post-treatment VL patients, Post Kala azar dermal leishmaniasis (PKDL), VL/HIV and control subjects (healthy, disease suspects and diseased other than VL) with the rK39 strip-test using blood and urine samples. Statistical Analysis: The level of agreement between the urine and blood testing was calculated by inter-rater agreement (kappa) statistics. Results: Forty-two active VL, 40 treated VL, six PKDL, three VL/HIV and 139 controls (54 healthy, 21 disease suspects and 64 diseased other than VL) were tested. All VL-related cases showed positive results with urine as well as blood samples (100%). The urine testing was found to have 100% sensitivity and 86.33% specificity for the diagnosis of VL. Kappa statistic between the two methods was 0.916 (P<0.001). Urine testing had more false-positive results in comparison with blood testing (13.67% vs. 9.45%), but the test subjects were from VL-endemic areas and they might be exposed to Leishmania donovani infection. Conclusions: The present study has the potentiality of providing a new, yet simplest non-invasive screening tool for VL in remote rural areas.
Keywords: India, rK39 strip test, urine sample, visceral leishmaniasis
Visceral leishmaniasis (VL) is endemic in eastern India, and occurs in remote peripheral areas lacking communication and modern health care facilities. Diagnosis of VL is a major obstacle in the control of this disease, as demonstration of parasite by splenic or bone-marrow aspiration is almost impossible in the peripheral rural areas because of lack of laboratory and multi-disciplinary support.  The recent advancement in the development of the rK39 immunochromatographic strip test using one drop of patient's blood is a major breakthrough in this direction. In previous studies from India, the test was found to be about 100% sensitive and 98% specific for the diagnosis of VL. ,, This strip test identifies immunoglobulin G directed against the K39 antigen - 39 amino acid sequence from the kinesin gene conserved in all Leishmania species. 
A 2006 review by Chappuis et al. reported the combined sensitivity and specificity estimate of rK39 dipstick to be 93.9% (87.7-97.1%) and 90.6% (66.8-97.9%), respectively.  Sensitivity was higher and more homogenous in the studies carried out in South Asia. ,, Specificity estimates were influenced by the type of controls. 
In 2002, a urine-based enzyme-linked immunosorbent assay (ELISA) against acetone-treated Leishmaina donovani promastigote antigen was developed to detect anti-L. donovani immunoglobulin G (IgG), although the field applicability was not reported by the authors.  In 2008, another ELISA against rKRP42 was developed that could be tested in the patient's urine sample.  Recently, a low molecular weight, heat-stable and carbohydrate-based leishmanial antigen was identified in VL patients' urine that could be detected by a latex agglutination test (KAtex).  KAtex has been evaluated in different field studies. , Encouraged by these findings, we examined the sensitivity and specificity of the K39 strip assay using urine samples of VL patients as well as control subjects over the past 2 months in two tertiary care hospitals at Kolkata, India, to confirm the utility of this simple, non-invasive diagnostic method in eastern India where most of the world VL cases occur.
InBios International Inc., Seattle, USA, supplied the rK39 rapid diagnostic kits for Kala Azar (Kalazar Detect™). According to the manufacturers' guide, 20-30 μL of serum is required. We modified the test using only one drop of blood directly instead of serum; the outcome of result was the same. 
We preformed the test with the usual method using one drop of patient's blood and two drops of chase buffer solution and also by using two drops of urine and one drop of chase buffer solution. Fresh blood and urine samples were used for the tests. The test is positive when two red lines appear in the middle of the nitrocellulose membrane (test and control areas), negative when only one red line appears (control area) and invalid when no line appears or one red line appears in the test area only. The result was read within 10 min.
This prospective study, approved by the Institutional Ethics Committee, included consecutive cases comprising active VL, post-treatment VL, post-Kala azar dermal leishmaniasis (PKDL), VL/HIV co-infected cases and control subjects from February to March, 2011. All the subjects were from West Bengal and Bihar. Assuming a 95% sensitivity of a screening test with 0.8 as desired power of study, the desired sample size was found to be 101. Only patients with parasitologically proven VL either from splenic or from bone marrow aspiration were included in this study. Control subjects were selected randomly from the outpatients and inpatients departments. The controls included disease suspects (prolonged fever with hepatosplenomegaly without parasitological evidence on splenic or bone marrow aspirate), patients other than VL (malaria, pneumonia, sepsis, tuberculosis, systemic lupus erythematosus, etc.) and healthy controls (family members of patients and doctors). Every fifth disease suspect and every seventh patient with disease other than VL were included. For healthy controls, every second family member of patients and doctors were recruited. Written informed consent was obtained from each of the participants. Any disease suspect refusing splenic or bone marrow aspiration or any subject not willing to furnish written consent was excluded. Control subjects who could not provide proper history regarding past VL infection were excluded. VL cases were parasitologically proven with either splenic or bone-marrow aspiration. PKDL was diagnosed on the basis of slit-skin smear or skin biopsy. Different investigators performed parasitological diagnosis and rk39 strip tests; each one was unaware of the other's result. Urine and blood samples from a single patient bore the same code numbers, and the tests were done by two different persons, each unaware of the other's result, and were finally read by a third investigator.
Appropriate statistical tools were used to present and analyze the data. The level of agreement between the urine and blood testing was calculated by inter-rater agreement (kappa) statistics.
[Table 1] provides the demographic, hematological and biochemical parameters of cases and controls. A total of 91 cases comprising of 42 active VL cases, 40 post-treatment VL cases, six PKDL cases and three VL/HIV co-infection cases were recruited. There were 139 controls, including 54 healthy controls, 21 disease suspects and 64 patients suffering from diseases other than VL. The last group consisted of patients suffering from malaria (eight), pneumonia (seven), sepsis (seven), tuberculosis (five), systemic lupus erythematosus (six), rheumatoid arthritis (six), HIV/AIDS (four), malignanacy (three), stroke (ten), liver abscess (two) and one each of Legionnaire's disease, filariasis, leprosy, cor pulmonale, amyloidosis and pancreatitis. The groups are age and sex matched. The active VL and VL/HIV co-infection cases have lower hemoglobin level and cell counts compared with the treated VL cases, PKDL cases and controls. All the groups had similar liver (P for one-way ANOVA=0.15) and renal function tests (P for one-way ANOVA=0.14).
[Table 2] gives the details of the test results with VL-related cases. All 42 VL cases gave positive results with urine as well as with blood samples (100%). Forty other VL cases who had been previously (1-3 months back) treated with amphotericin B at our institution and were regularly attending for follow-up were tested using their urine and blood samples. All of them remained positive after treatment (100%), up to 3 months, till reporting of these data.
Similarly, six PKDL and three VL-HIV co-infected patients showed positive results with both blood and urine samples (100%).
[Table 3] gives the details of the test results with control subjects. Control subjects were selected from three categories of patients: healthy controls, disease suspects (patients with prolonged fever with hepatosplenomegaly and absence of Leishman-Donovan (LD) body in splenic or bone marrow aspiration) and subjects diseased other than VL.
Of the 54 healthy controls, nine were positive with urine samples (16.67%), but only five of them were positive with blood samples (9.26%) (P=0.252). Among the 21 disease suspects, only one (4.76%) was positive with urine sample and none with blood (P=0.311). Of the 64 persons admitted at our institutions for diseases other than VL, nine were positive with urine samples (14.06%) but only six of them were positive with blood samples (9.37%) (P=0.41). The false-positive results in the last groups did not show any predilection for any specific disease, and the false-positive results were randomly distributed among the different disease groups as evidenced by paired proportion test among the different groups (for example, three of 10 in stroke patients vs. one of eight in malaria patients, P value=0.375).
Control subjects showing positive results were all from districts endemic for VL, except one (resident of Kolkata).
On the basis of these results, the urine testing was found to have 100% (confidence interval (C) 94.95-100%) sensitivity and 86.33% (CI 79.23-91.36%) specificity for the diagnosis of VL considering both active and treated cases. Testing with blood has a sensitivity of 100% (CI 94.95-100%) and specificity of 92.08% (85.95-95.78%) for the diagnosis of VL [Table 4]. The urine test was found to be false negative in none of the 91 VL-related cases, but the test was false positive in 13.67% of the controls.
The inter-rater agreement (kappa statistic) between the two methods was 0.916 (P<0.001), indicating a high level of agreement.
Testing patient's blood with the rk39 strip test is an easy non-invasive way of rapid accurate diagnosis for Indian visceral leishmaniasis, and it has almost replaced invasive methods like bone marrow or splenic aspiration. The test can be performed by the health assistants in the peripheral health centers. ,, This advancement could be made simpler by using patient's urine instead of blood so as to avoid contact with more hazardous biological material for peripheral health workers. Actually, no further equipment is necessary for this testing except the rK39 rapid diagnostic kits for Kala Azar. High sensitivity (100%) along with more than 85% specificity of this method is not inferior to blood testing. The test can be adopted as the simplest screening test for an outbreak of VL; even the patient or relative or local village leader or village health assistant can perform the test and can report to the local health authority. Extensive field trials are necessary to establish our opinion that the test can be adopted as the most simple screening test for suspicion of VL (fever for more than 2 or more weeks and splenomegaly) and PKDL at the primary health care level and further reference to physicians at primary health centers or district hospitals.
The characteristics of the rK39 strip test with blood are well documented. Various studies reported a high sensitivity of this test for the diagnosis of VL and PKDL. Particularly encouraging reports are mostly from the Indian subcontinent with high sensitivity (89-100%) and specificity (89-98.18%). ,,,,,,
The present study demonstrated a high sensitivity (100%) with specificity of 86.33% for the urine rK39 strip test in diagnosing VL, PKDL and HIV/VL co-infection. The level of agreement between the urine and blood testing was very high (kappa > 0.9).
When we were collecting data for the present study, we were totally unaware of the fact that evaluation of the rK-39 strip test using urine for the diagnosis of VL in an endemic area has been done in Bangladesh by Khan et al., showing a sensitivity and specificity 95% and 93.3%, respectively, compared with the serum-based rK-39 test.  The authors suggested that the urine-based rK-39 test could be a practical and efficient tool for the diagnosis of VL patients in rural areas, particularly where resources are limited. This was closely followed by another study by Chakravarty et al., who reported a sensitivity of 96.4%. In their study, specificity was low, varying from 66.7% in endemic healthy controls, 77.08% in non-endemic healthy controls to 62.2% in different diseases. With serum, sensitivity was 100%, whereas the specificity was 100%, 92.4% and 95.55% for the respective control groups.  Both these studies failed to demonstrate the 100% sensitivity that we claim. It is important to note that none of the groups used the chase buffer in their experiments. We tested different combinations of urine and buffer volumes. It is our opinion that a combination of two drops of patients' urine and one drop of buffer produces the best results.
Chakravarty et al. suggested that the high degree of positivity in non-VL controls in urine samples could be due to the non-specific binding of urinary proteins with rK39 on the strips, and that urine could not be used for the diagnosis of VL in the present rK39 strip format until the problem of non-specific binding can be resolved.  We differ and propose that there might be some inhibitory substance in blood blocking K39 response when K39 antibody is present in smaller amounts in non-VL controls. The kidneys may filter out such inhibitory factors in the blood so that the rK39 antibody could be detected uninhibited in urine; as a result of this, more subclinical infections from endemic areas showed positive results. As a result, specificity apparently becomes lower with urine samples.
To summarize, adding buffer to the urine sample may dilute the inhibitory factor from blood and increase the sensitivity of the test. In the other way round, filtered small quantity of K39 antibody in non-VL controls may show positivity in urine samples and apparently decrease the specificity of the test. This is why our results are so different from those of Chakravarty et al.
The sensitivity of the urine rK-39 strip test observed in our study (100%) is higher than that observed with the results of the urine ELISA against acetone-treated L. donovani promastigote antigen (93.3%).  Our results are similar to the recently developed rKRP42 urine ELISA test, which has a sensitivity of 94% and specificity of 99.6%.  The limitation of the latter is again the dearth of results showing field applicability.
In 2003, Solan-Galego et al. collected 95 dog urine samples from animals suffering from a variety of diseases that were examined for the presence of anti-Leishmania antibodies by a protein A ELISA. Of the 95 urine samples tested, 35 were found to be positive, 57 were found to be negative and three were found to be inconclusive. For the 35 positive urine samples, 26 had confirmed leishmaniasis. Results showed that urine antibodies were found only in proteinuric dogs. 
The only other commercially available urine test (KAtex) for the detection of VL has shown varied results. A study from Nepal showed very low sensitivity (47.7%) but high specificity (98.7%) for KAtex.  Similar studies, both in the Indian subcontinent and in different African countries, have demonstrated a low sensitivity of KAtex (55.6-67%). , This is a major limitation of KAtex as a reliable tool, and highlights the importance of urine k39 as a reliable, field applicable test.
The apparent limitation of the study is that urine testing has more false-positive results in comparison with blood testing (13.67% for urine vs. 9.45% for blood). The apparent false-positive results may not be true false positive as the test subjects were from VL-endemic areas and could be exposed to L. donovani infection.
Although cellular immunity is depressed in VL, humoral immunity remains intact. As a result, high levels of long-lasting antibodies are produced against Leishmania, but these are not protective. Antibody production occurs in both clinical and sub-clinical infections. For about 100 sub-clinical infections, about one clinically overt VL case occurs. Therefore, the urine test in the present form can detect VL among disease suspects (fever >2 weeks and splenomegaly not responding to anti-malarials or antibiotics in endemic areas). Larger studies in different endemic regions in diverse geographic locations are required to test our hypothesis that the test may also be used to detect sub-clinical infections in epidemiological settings.
After compilation of data, recently, we found that a new HIV/VL coinfected patient showed a positive result with the urine test but a negative result with the blood sample. Thus, the urine test helped us to suspect VL in a HIV patient with pyrexia of unknown origin (PUO) tested to be non-reactive to the rK39 strip test with the blood sample. In the future, a larger number of patients should be evaluated with the present technique described, with special regard toward field applicability, cost-effectiveness and comparison with other available urine testing measures.
In conclusion, the present study has the potentiality of providing a new, yet simple screening tool for VL.
The authors are indebted to Syamal Roy Chowdhury, InBios International Inc., Seattle, USA for supplying the rK39 rapid diagnostic kits for Kala Azar (Kalazar Detect TM ). They also acknowledge the encouragement and help by the Director and staff of the School of Tropical Medicine, Kolkata, India. They also are thankful to the Director and faculties of the Department of Medicine, Institution of Post Graduate Medical Education and Research, for the encouragement and support.
[Table 1], [Table 2], [Table 3], [Table 4]