Abstract
Canine heartworm, Dirofilaria immitis, causes a potentially fatal, multisystemic disease in dogs. Diagnosis of heartworm disease relies on serologic antigen detection and microfilariae identification. Immune-complex dissociation (ICD) of serum or plasma by heat treatment increases detection. We assessed urine as a sample for heartworm antigen detection in dogs with known and unknown infection status using a commercial ELISA. Twenty-nine matching serum or plasma and urine samples were collected from 20 experimentally infected dogs to assess antigen detection in urine. Matching serum and urine samples of 267 dogs admitted to a spay/neuter clinic with unknown infection status were also screened. In experimentally infected dogs, antigenemia was detected in 86% of samples pre-ICD and 91% post-ICD; antigenuria was detected in 76% pre-ICD and 72% post-ICD; differences were not statistically significant. In clinic samples, antigenemia was detected in 7.9% of dogs pre-ICD and 12.4% post-ICD; antigenuria was detected in 11.6% of dogs pre-ICD and 6.4% post-ICD. In experimental infection samples, sensitivity (Se), specificity (Sp), and positive predictive value of antigenuria were high (85%, 100%, 100%, respectively). In clinic samples, Se and negative predictive value were high (95% and 92%, respectively). Our data confirm that heartworm antigen can be detected in the urine of dogs with both experimental heartworm infections and with unknown infection status. However, antigenuria had high Sp and low Se compared to matching serum of clinic dog samples and urine cannot be recommended for routine screening.
Infection with the canine heartworm, Dirofilaria immitis, causes a multisystemic and potentially fatal disease in dogs and wild canids, with some degree of infection seen in cats. 25 Infections with this mosquito-borne filarial nematode have been identified in dogs worldwide, especially in warmer, tropical regions. 26 In the United States, heartworm infection in dogs has been diagnosed in all 50 states, with an increased occurrence in the warmer southern and southeastern regions.25,26 Texas in particular is considered an endemic region, and several studies have investigated the prevalence of D. immitis in wild canids in addition to client-owned and shelter dogs throughout the region.15,30,32 Clinical signs in the canine definitive host are absent for months after infection. Chronically affected dogs may eventually develop cough, dyspnea, exercise intolerance, or weakness, with potential progression to signs of right-sided congestive heart failure, ascites, anorexia, or caval syndrome. 21 The multisystemic effects of heartworm infection are well documented, including pulmonary, cardiac, and renal effects.1,2,5,7,10,22,23,25,27,31
The current recommendation to detect canine heartworm infections is to perform a sensitive and specific serologic antigen test coupled with a microfilariae detection test. 2 Circulating antigen is primarily secreted by adult female worms 6–9 mo post-infection (mpi), and the earliest it is detected most commonly is 5–7 mpi. 25 Heartworm infections have been detected as early as 98 d post-infection (dpi) using immune-complex dissociation (ICD) methods. 9 Heat treatment of either serum or plasma for ICD frees bound heartworm antigen, decreasing the likelihood of false-negative results.6,20 However, heat treatment is not recommended for routine screening for heartworm disease by the American Heartworm Society (AHS), and should be performed only when a negative result is obtained with a persisting clinician suspicion of heartworm disease.20,25 Microfilariae can be detected as early as 6 mpi in whole blood but may take longer to appear in some cases. 25 The modified Knott test is more sensitive than other microfilariae detection methods as it is a concentration technique and can be used to quantify and identify species of microfilariae. 25 The AHS recommends screening dogs for antigen and microfilariae at 7 mo of age and annually thereafter, or 7 mo after a missed dose of preventive. 25
Antigenuria following antigenemia is a common result of many infectious processes in humans and animals.8,12 Therefore, urine has been used as a template for antigen detection in other infectious diseases in dogs and humans, including histoplasmosis and blastomycosis. 11 Parasitic nematode infections of the urinary tract can be detected by identifying eggs by sedimentation and urinalysis, but urine has not been extensively explored for detection of nematode infections present outside the urinary tract.3,13,17
Urine was studied for the detection of heartworm antigen in 12 experimentally infected dogs at a single time >1-y post-infection (microfilariae have been detected in urine of an infected dogs in a rare clinical case).16,22 Urine might be a useful alternative biofluid for heartworm antigen detection based on the filtration of blood by the kidneys, and could possibly inform earlier detection of infection. We assessed the reliability of heartworm antigen detection in urine in a clinic population with unknown infection status using a commercial microtiter plate antigen ELISA. Our objectives were: 1) to re-assess the feasibility of heartworm antigen detection in urine from experimentally infected dogs at different times post-infection, including early infection; and 2) to assess the use of heartworm antigen detection in urine versus serum from dogs with unknown infection status in Texas, USA. We hypothesized that due to the filtration by the kidneys and concentration in the urinary bladder and previous evidence showing heartworm antigen detection in urine, antigen will be reliably detected in the urine of dogs with unknown infection status that have detectable antigen in serum or plasma.
Materials and methods
Sample acquisition
The collection of samples for our study was approved by the Texas A&M University Institutional Animal Care and Use Committee (2019-0200 CA) and in accordance with the relevant guidelines and regulations. Additionally, owner permission was obtained to collect clinic specimens at Hill Country Animal League (Boerne, TX, USA).
Experimental samples
Matching samples of whole blood, serum, and urine were obtained from 13 dogs experimentally infected with third-stage D. immitis larvae using methods described previously. 9 Urine was collected via catheterization. Whole blood, serum, and urine were refrigerated for up to 2 d prior to processing. An additional 7 archived samples of paired serum and urine obtained postmortem from experimentally infected dogs 5 mpi were also included along with 2 archived samples of paired serum and urine from uninfected control dogs to serve as negative controls. Antigen testing was performed on serum or heparinized plasma (archived samples) and urine as described below. All archival samples were stored at −20°C until testing was performed. In total, we tested 29 samples from 20 experimentally infected dogs at 3–37 mpi, and 2 samples from uninfected negative control dogs.
Clinic samples
Matching samples of whole blood, serum, and urine were obtained from 267 dogs with unknown heartworm infection status admitted for sterilization surgery at a high-volume spay/neuter clinic located in Boerne, Kendall County, southwest Texas. Samples were acquired from any dog admitted to the clinic for sterilization surgery that was either stated by the owner or estimated by the attending veterinarian or veterinary technician to be ≥6-mo-old during the study period. Sample collection was done after induction of anesthesia and before the surgical procedure.
Serum, urine, and whole blood samples were refrigerated after collection for a maximum of 2 d before processing. Whole blood from each animal was collected in a K2-EDTA vacutainer tube (Becton-Dickinson [BD]) and a clot activator vacutainer tube (BD). The clotted blood was centrifuged at room temperature for 10 min at 1,700 × g, and serum was transferred to a second plain tube. For urine collection in this population, the bladder of anesthetized dogs was manually expressed, and mid-stream urine was collected in a sample collection cup.
Modified Knott test
Modified Knott testing was performed for the evaluation of microfilaremia in K2-EDTA blood as described previously. 28 Samples were examined under 100× magnification for quantification and 400× as needed to identify the morphology of microfilariae. The remaining whole blood was stored at −20°C.
Antigen testing
Serum, plasma, and urine were tested with an antigen ELISA (DiroCHEK canine heartworm antigen test kit; Zoetis). This ELISA was validated for detection of antigen in serum and plasma with a reported sensitivity (Se) of 99% and specificity (Sp) of 96%.5,18 Testing was first performed following the manufacturer’s protocols followed by the addition of heat treatment as described previously.29,34 The volume of sample used was 50 μL and was the same regardless of biofluid type: serum, plasma, or urine, per the test kit instructions. For heat treatment, both serum or plasma and urine samples were heated at 104°C for 10 min, centrifuged (16,000 × g, 10 min), then tested with the ELISA following the manufacturer’s protocols for non–heat-treated samples. Positive or negative results were determined visually per the manufacturer’s instructions. A color change from colorless to blue was considered positive. In addition, optical density (OD) was determined by spectrophotometry (BioTek Synergy H1 microplate reader; Agilent) at 590 nm 5 min after the addition of the final solution and following determination of color change. ODs ≥0.069 were considered positive.24,28,32 Cutoff values for positive and negative status were determined as described previously. 33 Remaining serum, plasma, and urine were stored at −20°C.
Urine specific gravity
Urine specific gravity (USG) of the samples from the experimentally infected dogs was obtained from the Texas A&M Veterinary Medical Diagnostic Laboratory (College Station, TX, USA) as part of a complete urinalysis. Due to sampling limitations in the clinic, USG was not obtained for the clinic population.
Statistical analysis
All data were compiled in spreadsheets, and test performance characteristics were calculated using Excel (Office 365 v.18.2104.12721.0; Microsoft). Se, Sp, positive predictive value (PPV), and negative predictive value (NPV) were calculated for the detection of antigen in urine in comparison to serum or plasma both pre- and post-ICD.
Cohen kappa (κ) was calculated to assess agreement between the following categories of results: pre-ICD serum or plasma (ICD-SP) and pre-ICD urine (ICD-U), pre-ICD-SP and post-ICD-U, post-ICD-SP and pre-ICD-U, and post-ICD-SP and post-ICD-U. Kappa agreement between serum or plasma and urine was considered fair if 0.21–0.40, moderate if 0.41–0.60, substantial if 0.61–0.80, and almost-perfect if 0.81–1.00. 19 The McNemar paired χ2 test was used to compare pre– and post–heat-treatment results for antigen detection in both blood and urine. The percent change in OD590 between pre- and post-ICD was calculated for each population for both serum or plasma and urine, and the average percent change was calculated. Mean and median USG were calculated for all samples in the experimental population. The mean USG for positive urine samples and negative urine samples in the experimental population was then compared using an unpaired t-test, with ≤0.05 as the threshold for significance.
Results
Experimental population
Antigen detection in blood and urine
Twenty-nine samples were available from the 20 experimentally infected dogs at different times post-infection (Table 1; Fig. 1). Pre-ICD, 25 of 29 (86%) samples were antigen positive in serum or plasma; 20 of 22 (91%) samples tested positive post-ICD. Two samples converted from negative to positive when heat treated. Due to inadequate sample volume, 7 plasma samples were unavailable for post-ICD testing. Excluding these 7 samples, 18 of 22 (82%) samples were positive pre-ICD, and 20 of 22 (91%) post-ICD. In urine, antigen was detected in 22 of 29 (76%) samples pre-ICD and 21 of 29 (72%) samples post-ICD. There was no significant difference in pre- and post-ICD results in either serum or plasma or urine (p = 0.480, χ2 = 0.500, df = 1; p = 1.000, χ2 = 0.000, df = 1). In samples that were antigen positive in serum or plasma pre-ICD, 21 of the 25 (84%) corresponding urine samples had antigenuria. Compared to the total number of serum or plasma samples positive for antigen pre- or post-ICD (27 of 29), antigen was detected in the urine in 22 of 29 (76%) matching samples. Heartworm antigen was detected in urine as early as 4.5 mpi in 1 dog but was not detected in a different dog at the same time post-infection (Table 1). Samples from dogs 5–7 mpi were all antigenemic in serum or plasma pre-ICD and post-ICD when samples were available, but samples in this timeframe post-infection had variable antigenuria results. All 12 samples ≥7 mpi were positive in all biofluids (Table 1; Fig. 1). The 2 uninfected negative control dogs tested negative for heartworm antigen pre- and post-ICD in the serum and urine, and no microfilariae were detected on the modified Knott test.
Dogs experimentally infected with Dirofilaria immitis, by time post-infection (PI), with modified Knott (MK) and heartworm antigen ELISA results pre– and post–immune-complex dissociation (ICD) by heat treatment of serum or plasma and urine.
NS = no sample.
Dirofilaria immitis antigen status of all samples (n = 29) from experimentally infected dogs (n = 20) versus months post-infection (PI). Positivity in either pre- or post-ICD samples was labeled positive. Samples were tested only once via all methods, but some dogs are represented more than once due to serial sampling.
Agreement between serum or plasma and urine positivity status in the experimental population was variable (Table 2). Agreement between serum or plasma and urine was fair-to-moderate when serum or plasma samples were not heat-treated, but agreement improved when serum or plasma was heat-treated regardless of urine treatment. The overall agreement of antigenuria and antigenemia regardless of heat-treatment status of all samples was fair (0.38). The Se, Sp, and PPV of urine for heartworm antigen detection were all high compared to serum or plasma detection, with 100% Sp and PPV regardless of heat treatment. However, NPV of urine was low, although the sensitivity and NPV of urine for heartworm antigen detection increased slightly compared to heat-treated serum or plasma (Table 3). Of the 12 samples that were collected 7 mpi or longer, all were positive for antigen in both biofluids, pre- and post-ICD. Se, Sp, PPV, and NPV were all 100%, and agreement was almost-perfect at 1.00. The percent change in OD of serum samples was positive after heat treatment (18.4%); in urine, the percent change between pre- and post-ICD samples was negative (−3.8; Fig. 2A). The x̄ USG in the experimental population was 1.033 (median: 1.034; range: 1.012 to >1.060). The x̄ USG of samples with antigenuria was 1.031, and those without antigenuria had a x̄ USG of 1.039. There was no significant difference between USG of antigen-positive and antigen-negative samples (p = 0.102, df = 18, t = 1.720).
Agreement (Cohen kappa [κ]) between serum or plasma and urine, pre– and post–immune-complex dissociation (ICD) via heat treatment, in dogs experimentally infected with Dirofilaria immitis.
Sensitivity (Se), specificity (Sp), positive predictive value (PPV), and negative predictive value (NPV) of antigenuria for detecting antigenemia in dogs experimentally infected with Dirofilaria immitis (n = 29). Comparisons were made of pre– and post–immune-complex dissociation (ICD) via heat treatment for both sample types.
Average (avg) percent change in optical density (OD590) using the DiroCHEK canine heartworm antigen test kit after immune-complex dissociation via heat treatment of urine and serum from dogs in
Impact of microfilaremia on antigen detection
Fifteen of 22 (68%) samples with adequate volume of whole blood were microfilaremic. All 15 samples were antigenemic pre- and post-ICD, and 14 of 15 were antigenuric pre- and post-ICD. Seven of 22 (32%) samples were amicrofilaremic. The 7 amicrofilaremic samples were 3–5 mpi (Table 1). Of the 7 amicrofilaremic samples, 2 were antigenuric and antigenemic pre- and post-ICD, 1 was antigenemic only post-ICD and antigenuric pre- and post-ICD, 2 were antigenemic only (1 pre- and post-ICD, 1 only post-ICD), and 2 were negative for antigen in both biofluids pre- and post-ICD. Seven samples did not have adequate whole blood volume for microfilariae testing to be performed.
Clinic population
Antigen detection in blood and urine
Heartworm antigen was detected in 21 of 267 (7.9%) serum samples pre-ICD and in 33 of 267 (12.4%) serum samples post-ICD. Antigen was detected in 31 of 267 (11.6%) urine samples pre-ICD, and 17 of 267 (6.4%) urine samples post-ICD. In the 33 serum samples that were positive either pre- or post-ICD, antigen was detected in the urine in 19 of 33 (57.6%) dogs for which serum was positive post-ICD. For the 230 serum samples that were negative for heartworm antigen detection both pre- and post-ICD, antigen was detected in the urine in 12 of 230 (5.2%) samples (Table 4).
Clinic population modified Knott (MK) and heartworm antigen test results pre– and post–immune-complex dissociation (ICD) via heat treatment in serum or plasma and urine.
Agreement between antigen positivity in serum and urine in the clinic population was variable (Table 5). Agreement between serum and urine was moderate to almost-perfect when serum was not heat-treated, and moderate to substantial when serum was heat-treated. Heat treatment of urine samples improved the agreement between pre-ICD serum and post-ICD urine from substantial to almost-perfect. Overall agreement between antigenuria and antigenemia regardless of heat-treatment status was moderate (0.50).
Agreement (Cohen kappa [κ]) between serum and urine for detecting heartworm antigen in a clinic population of dogs (n = 267). Comparisons were made of pre– and post–immune-complex dissociation (ICD) via heat treatment for both sample types.
Mean Sp and NPV of urine for antigen detection in the clinic population were high; mean Se and PPV were lower. When neither sample type was heat-treated, Sp and NPV of urine were high; Se was moderate and PPV was low. When only urine was heat-treated, Sp, PPV, and NPV of urine were high; Se was moderate. When only serum was heat-treated, Sp and NPV of urine were high; Se and PPV were low. When both serum and urine samples were heat-treated, Sp, PPV, and NPV of urine were high; Se was low. In the clinic population, trends in the percent change in OD590 post-ICD compared with pre-ICD were similar to the experimental population, with an average percent increase for serum (3.4%) and decrease for urine (–7.6%; Fig. 2B). Of 15 dogs that were antigenuric but not antigenemic pre-ICD, 3 tested antigenemic positive post-ICD.
Impact of microfilariae on antigen detection
Three samples (1.1%) were microfilaremic; the remaining 264 samples were amicrofilaremic. All 3 microfilaremic samples had antigen detected in serum and urine, both pre- and post-ICD. Of the amicrofilaremic samples, 218 had no antigen detected in serum and urine both pre- and post-ICD, 13 had antigen detected in serum and urine both pre- and post-ICD, and an additional 13 had antigen detected only in serum post-ICD (Table 4).
Discussion
To date, a few studies have investigated heartworm-associated antigenuria in dogs, but knowledge on the use of urine as an alternative template for routine heartworm screening is limited.16,23 In experimental heartworm studies, D. immitis antigenuria was seen in infected dogs by 398 dpi and in infected cats by 240 dpi.5,16 However, we detected antigenuria as early as 135 dpi. In experimental infections, antigen detection in urine was consistent with serum or plasma for infections ≥217 dpi (7 mpi). As a screening test, urine was specific with a high NPV and was positive in some cases in which antigenemia was not detected. In both the experimental and clinic populations, the percent change in OD after heat treatment was positive in serum or blood and negative in urine.
During the 5–6 mpi period, antigen detection in serum or plasma has been described as inconsistent.10,14 One explanation is that, during this time, any worms that are present have not yet matured into adults, hence the amount of antigen available for detection is low. 35 Given that the earliest antigen is commonly detected is 5–7 mpi, or as early as 3 mpi with ICD methods, discrepancies in results are possible earlier in infection.9,25 These discrepancies are potentially responsible for the moderate agreement that we found. In our study, samples of all biofluids from infections <4 mpi were antigen-negative, pre- and post-ICD. All experimental serum or plasma samples tested for antigen at 5 mpi and 3 tested at 6 mpi were positive for antigen pre- and post-ICD (when adequate volume allowed for post-ICD testing) regardless of antigenuria status. All serum or plasma and urine samples were positive from 7 mpi onward in experimentally infected dogs, and antigenuria was consistent with antigenemia in experimentally infected dogs for which serial samples were obtained. This combined with the recommended timing of testing per the AHS guidelines provides evidence for the use of urine for antigen detection in established heartworm infections. Data from this population also demonstrate that samples ≥5 mpi typically contained adequate antigen levels for detection in the urine.
In the clinic population, heartworm antigen was detected in urine with high Sp and low Se, which may limit its use as a clinical screening test but might still be useful in a research setting, as few false-positive results would be expected. Although the low Se and PPV in this population are due to discordant results between biofluids and make interpretation difficult in a diagnostic and clinical context, there might be plausible biological explanations for such results. If these results are true negatives, the urine positivity could be due to cross-reactivity of antigens associated with other parasites (Spirocerca lupi, Dirofilaria repens, etc.) that may have been altered during the heat-treatment protocol as seen in serum and plasma of dogs infected with other nematodes, even if matching sera tested negative.4,33 Alternatively, in the case that all or some of these urine tests were true positives, heartworm antigen present in urine could have been detectable because of an increased urine concentration due to time of collection or dehydration, while antigen levels in serum were still present at low concentrations. Adequate urine volume was not available to measure USG on all samples; however, in the experimental population, samples had an adequate sample volume, and no significant difference was detected in the urine of animals with positive or negative serum or plasma samples. According to the manufacturer, false negatives in serum and plasma may occur in animals infected with ≤3 adult females. This, coupled with the detection of heartworm antigenuria in experimental samples as early as 4 mpi could suggest that these are true-positive results.
We found that antigen detection in serum pre- and post-ICD was comparable to that found in previous studies in a similar population from the same geographic area and higher than the expected prevalence for client-owned dogs in Texas. 15 Although heat treatment of urine was not beneficial for antigen detection, our study further supports the value of heat treatment of serum or plasma samples from dogs in endemic areas, or with an unknown history of heartworm prevention. 20 However, it is possible that in a different biofluid, such as urine, heartworm antigens are less stable and more prone to degradation, rendering them undetectable after heat treatment. The change in urine positivity from positive to negative post–heat treatment was more evident in the clinic samples than in the experimental samples.
Regarding the agreement between antigen detection in the serum or plasma and the detection of antigen in the urine, the results were variable depending on the population, experimental or clinic, and whether urine was subjected to ICD via heat treatment. In both populations, the average OD of serum samples increased with heat treatment, indicating more antigen for detection, and decreased in urine after the addition of heat, indicating less antigen available for detection.
Footnotes
Acknowledgements
We thank the staff at Hill Country Animal League for their enthusiasm and willingness to assist with sample collection.
Availability of data and materials
Enquiries may be directed to the corresponding author.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
Our study was supported by the Verocai Parasitology Laboratory at Texas A&M. Andrew C. Brown was supported by a grant from the Texas Veterinary Medical Foundation (TVMF) and the Texas A&M College of Veterinary Medicine and Biomedical Science.
