Abstract
Enzyme-linked immunosorbent assay (ELISA) has high sensitivity and specificity for detection of feline immunodeficiency virus (FIV) antibodies, but does not distinguish between infection and vaccination. Real-time quantitative PCR (qPCR) assays may distinguish infected from vaccinated cats. Performance of a commercial qPCR assay was assessed with blood samples from 29 FIV-infected non-vaccinated, 26 FIV-uninfected vaccinated and 35 FIV-uninfected non-vaccinated cats. FIV infection status of cats was assigned based on a combination of vaccination and medical history, prevention of contact with potentially infected cats and two FIV antibody ELISA results. Test sensitivity and specificity were determined against this gold standard. The qPCR test yielded positive results in samples from 23/29 FIV-infected non-vaccinated, 2/26 FIV-uninfected vaccinated and 0/35 FIV-uninfected non-vaccinated cats. It was concluded that the qPCR test was moderately sensitive and highly specific for the diagnosis of FIV infection, and that it may be suitable for ruling out FIV infection in cats with a positive antibody ELISA result and unknown vaccination history.
Short Communication
Infection of cats with the feline immunodeficiency virus (FIV) is a worldwide problem with prevalence of 2.5% in the USA, 4.3% in Canada and 23.2% in Japan.1 –3 FIV targets mononuclear leukocytes, particularly T-lymphocytes, 4 and induces cell dysfunction or cell death, and cats may develop immunodeficiency in the late stages of the infection.5,6 Infection is for life and induces antibodies, which are detected as the mainstay of diagnosis. 7 The most widely used antibody ELISA (Idexx) in North America has a reported sensitivity and specificity of 98% and 93%, respectively. 8 Within days of infection viral RNA and proteins appear in host plasma, and seroconversion occurs after 3–4 weeks. Negative serology results in FIV-infected cats may be due to recent pre-seroconversion infection, while false-positive results may be due to appropriate detection of maternal or vaccine-induced antibodies, or due to sample mix-up or factors inherent to test technique, such as inaccurate timing or assay mishandling. 7 In 2003 an FIV vaccine consisting of inactivated whole virus subtype A and D (Fel-O-Vax; Fort Dodge Animal Health) was introduced to the Canadian market. The vaccine elicits anti-FIV antibodies that are difficult to distinguish from those induced by natural infection.9 –11
During FIV replication, viral genomic RNA enters cells, is reverse transcribed into DNA and then inserted into the host genome as proviral DNA prior to transcription of virus genes. Therefore, proviral DNA should be detectable by PCR in infected cats, but not in vaccinated cats, 12 and detection could be used in cats whose vaccination status is unknown. However, marked viral sequence variation, low blood lymphocyte proviral load during a prolonged clinically silent period of infection, combined with the need for special facilities, technical skill and strict quality control, are a challenge for polymerase chain reaction (PCR)-based assays. Conventional PCR yielding qualitative results is offered as a commercial test for the diagnosis for FIV infection by multiple laboratories; however, in one study up to 50% of PCR-based diagnoses from commercial laboratories disagreed with the FIV serological status of cats. 13 In addition, conventional PCR assays discriminated poorly between infected and vaccinated cats, though the reasons for this discrepancy were unclear. 14
Recently, quantitative real-time PCR (qPCR) assays have been developed for the detection of FIV. This type of PCR assay is qualitative and quantitative because newly synthesized PCR product is measured continuously, and a software algorithm is applied to determine the number of template copies accumulating relative to control samples. Primer and template sequence divergence, and low template copies affect qPCR just like conventional PCR; however, automated measurement of amplification products reduces the subjectivity of test result interpretation encountered in gel-based PCR formats, and should, therefore, lead to fewer false- positive and negative results. Several commercial qPCR assays to detect FIV genomic DNA in blood samples became available in the last few years. Assessment of one such qPCR assay showed overall lower sensitivity and specificity for diagnosing FIV infection than serologic assays. 15
Assessment of a new diagnostic technique should entail comparison to a gold standard. However, when there is a lack of a perfect gold standard, which, by definition, should be 100% specific and sensitive, the best available test or combination of tests or criteria should be used. 16 In the case of FIV there is no agreement on what the gold standard for determining the FIV infection status should be. Virus culture might be considered a gold standard in a viral disease, and has been suggested for FIV, but is best suited for viruses that consistently induce cytopathic changes in indicator cell lines. 14 As FIV is non-cytopathic, detection of viral components in culture by immunologic or functional assays is necessary, and adds complexity and additional technical requirements. While protocols are in place for the detection of HIV-infected cells, tests to quantify FIV have yet to be developed and standardized. Hence, the value of virus culture as a method for FIV diagnosis is equivocal. Other gold standards used in previous studies were results of serial serologic testing and Western blotting. 13 However, Western blotting is not widely available, and interpretation may be subjective. 17 More recently, sensitivity and specificity of serologic and PCR-based assays individually or in combination were assessed using Bayesian latent class models to obtain posterior probability distributions of tests used. This model relied on prior knowledge of the performance of the tests, and other parameters, such as the prevalence of FIV infection in different cat populations. 15
In this study, we aimed to assess the performance of a qPCR assay in cats of various FIV vaccination and infection statuses. We hypothesized that qPCR would be highly sensitive and specific.
The study protocol was approved by the Institutional Animal Care Committee at the University of Guelph in accordance with guidelines provided by the Canadian Council for Animal Care. Convenience samples from 90 owned domestic cats of various breeds and sex were available for analysis. Cats enrolled in this study came from 10 veterinary clinics across Ontario, Canada, and one veterinary clinic in Nova Scotia, Canada. Criteria for study enrollment and group assignment were based on complete medical and vaccination history, and results of two FIV antibody ELISA (Idexx) tests one within 2 years of enrollment and one at the time of enrollment. FIV infected non-vaccinated cats had two or more positive antibody ELISA test results, conditions associated with chronic FIV infection, such as stomatitis, lymphopenia and/or diarrhea, and no history of vaccination. FIV-uninfected vaccinated cats were vaccinated against FIV (Fel-O-Vax) between 1 and 3 years prior to study, had two negative antibody ELISA test results, including one immediately prior to vaccination, and were housed strictly indoors without contact with cats of unknown FIV status (as reported by owners). FIV-uninfected non-vaccinated cats had two or more negative antibody ELISA test results at least 6 weeks apart.
Blood was collected by venepuncture and approximately 1 ml each was placed into an ethylenediamine tetra-acetic acid tube, packaged on ice and sent to a commercial laboratory for qPCR. The identity and group assignment of the cats was concealed until all analyses were complete.
All outcome variables were tested for normality using Shapiro–Wilk tests and quantile plots. The diagnostic accuracy of qPCR was assessed by computing specificity (number of true negatives/total number of negatives) and sensitivity (number of true positives/total number of positives). Ninety-five percent binomial confidence intervals (CIs) for sensitivity and specificity were computed by the Pearson–Klopper method using R (‘library binom’). Analyses were performed using R software v. 2.12.1 (http://www.r-project.org/).
Twenty-three (79.3%) of 29 and two (7.7%) of 26 samples from FIV-infected cats and FIV uninfected vaccinated cats, respectively, yielded positive qPCR results (Table 1). Thirty-five (100%) of 35 samples from uninfected non-vaccinated cats were negative by qPCR.
Table 1.
Results of feline immunodeficiency virus (FIV) quantitative real-time polymerase chain reaction (qPCR) for samples from FIV-infected, FIV-uninfected non-vaccinated and FIV-uninfected vaccinated cats
| Group | qPCR | Number of samples (%) |
|---|---|---|
| Infected (n = 29) | + | 23 (79.3) |
| − | 6 (20.7) | |
| Uninfected vaccinated (n = 26) | − | 24 (92.3) |
| + | 2 (7.7) * | |
| Uninfected non-vaccinated (n = 35) | − | 35 (100) |
Samples from two uninfected vaccinated cats were re-tested by qPCR approximately 3 months after the initial test and negative results were obtained
Using the combination of antibody ELISA test result, vaccination, medical and housing history as the gold standard, the sensitivity and specificity of qPCR were calculated. The sensitivity and specificity of qPCR for samples from FIV non-vaccinated cats was 79.3% (95% binomial CI 60.3–92%) and 100% (95% binomial CI 90–100%), respectively. For samples from FIV-uninfected vaccinated cats, sensitivity could not be calculated, but specificity was 92.3% (95% binomial CI 74.9–99.1%).
Diagnosis of FIV infection in cats was based traditionally on the detection of serum antibodies, which persist for life in infected cats. 7 This approach is not suitable for testing cats with an unknown vaccination history, as vaccination against FIV induces antibodies indistinguishable by available ELISA from those of natural infection.9 –11 Distinction of vaccinated from infected cats may be accomplished by qPCR; therefore, the purpose of this study was to evaluate the performance of a commercially-available qPCR assay in samples from three different groups of cats of defined FIV status.
On the basis of the gold standard definition used in this study, the specificity of the qPCR assay was found to be high with 100% of samples from FIV non-vaccinated cats and 92.3% of samples from vaccinated cats categorized accurately. The sensitivity of the qPCR assay was lower at 79.3%. Ideally, samples from FIV infected vaccinated cats should also have been analyzed, but such samples could not be obtained. Hence, sensitivity of qPCR could not be calculated for vaccinated cats. One of the limitations of the study was the relatively small sample size, which did not allow estimates of sensitivity and specificity with a high precision. This is reflected in the width of the 95% CI around the point estimates. Further studies with larger sample sizes are recommended to increase the precision of these sensitivity and specificity estimates. For example, to obtain narrower 95% CIs around a point estimate for a sensitivity of 79.3% with a margin of error of ±5%, approximately 250 cats would have to be enrolled in each group.
In medicine and statistics, the ideal gold standard is a diagnostic test that is 100% sensitive and specific; however, as this can rarely be achieved, gold standards are often the best test or combination of tests available under reasonable conditions. 16 Group assignment of cats based on vaccination history provided by the veterinarian and two sequentially obtained ELISA test results was considered to be a robust method available for setting a gold standard for FIV vaccination and infection status. Vaccination history allowed classification into vaccinated and unvaccinated groups, and repeat ELISA in combination with history was chosen as the method to assign FIV status. Although Western blotting is the method most commonly recommended for confirmation of a positive ELISA test result in otherwise healthy cats, cats in the FIV-positive group in this study were tested by ELISA as indicated within 2 years prior to study and at enrollment into the study. 17 FIV-infected cats had also been tested at additional times earlier than 2 years prior to study and/or had FIV-associated conditions such as stomatitis, lymphopenia and chronic diarrhea. Therefore, the positive predictive value (PPV) of two positive ELISA results was very high, and Western blotting was not performed. However, owner ability to keep cats separate or indoors may not have been absolute, and a slight chance of false-positive or negative ELISA results cannot be excluded. Therefore, the possibility that cats may have been assigned erroneously to groups could not be ruled out entirely, but was considered to be small.
qPCR depends on a software algorithm set to identify the number of template copies that distinguishes non-specific amplification from amplification of true template, and is usually applied to enumerate template copy number rather than to provide a positive versus negative result. 18 It appears as if the qPCR software threshold in this assay for distinction between positive and negative results has been set conservatively, yielding high specificity, which is more suitable for a confirmatory test. Hence, if the FIV qPCR test is applied to verify a positive FIV antibody ELISA test result in a cat of unknown vaccination history or lacking clinical features suggestive of FIV infection, the high specificity will be the most valuable test characteristic.
Reasons why some uninfected vaccinated cats test positive with the FIV qPCR assay are unknown. Similar findings have been reported with conventional FIV PCR assays, where the number of false positive results was significantly greater for FIV-uninfected vaccinated cats than other groups of cats. 14 Possible reasons are the retention of viral nucleic acids as contaminants in the vaccine inducing false-positive results with such highly sensitive assays as qPCR, recent infection of vaccinated cats, or a higher likelihood of non-specific template amplification in vaccinated cats due to non-FIV components in the vaccine. The most likely cause of this aspect of PCR assays for FIV remains unresolved at this time.
A sensitivity of 79.3% for detecting FIV infection in cats suggests that the qPCR assay may be of limited value as a primary screening test. Possible reasons for the relatively low sensitivity are conservatively set software algorithm thresholds, limited primer homology with the range of field isolates of FIV, and possible template deterioration in the time between sample collection and analysis. All of these factors are considered contributory to rendering PCR an unreliable first-line test for detecting HIV infection, and similar principles should be applied for the diagnosis of FIV infection. 19 The primers used, and therefore the viral gene targeted for amplification by the commercial assay, was proprietary information unavailable to the authors.
A PPV value depends on the specificity and sensitivity of the test, and on the prevalence of infection in the population tested. As some vaccinated cats had positive qPCR test results, the PPV of qPCR also depends on the percentage of the cat population that is vaccinated. The number of North American cats that have been vaccinated against FIV is unknown, but appears to be low (unpublished observations). Furthermore, the likelihood of FIV infection depends on geographic location and risk factors such as cat age, sex and lifestyle.3,20 –22 Therefore, if qPCR were used as a screening test on healthy cats with limited risk factors, the PPV would be expected to be low. If, however, qPCR was used as a confirmatory test following a positive antibody ELISA result in a cat with unknown risk factors and vaccination history, PPV would likely be higher.
Conclusions
In conclusion, from the limited samples analyzed in this study, the qPCR test for FIV infection had high specificity, but moderate sensitivity. Additional studies with larger sample size are needed to confirm the preliminary findings. Results suggest that antibody ELISA should remain the mainstay of FIV testing and that qPCR has favorable characteristics as a confirmatory test in cats with a positive antibody test result.
Acknowledgments
We thank participating veterinary clinics (Atlantic Cat Hospital, Bytown Cat Hospital, Carling Animal Hospital, Centretown Veterinary Hospital, Charing Cross Cat Clinic, Colonial Veterinary Hospital, March Road Veterinary Hospital, Merivale Cat Hospital, Osgoode Veterinary Services, Prince of Wales Animal Hospital and The Cat Clinic) for providing historical and medical information, and blood samples from cats. We thank Hakimeh Mohammadi for technical support, and Hugues Beaufrère and Andrew Peregrine for assistance with statistical analyses and for review.
Footnotes
Funding: This work was performed at the University of Guelph. Funding was provided by the Pet Trust Foundation at the University of Guelph, and the Natural Science and Engineering Council of Canada. The sponsors were not involved in study design, data analysis and interpretation or in manuscript submission.
The authors do not have any potential conflicts of interest to declare.
Accepted: 19 December 2012
References
- 1. Levy JK, Scott HM, Lachtara JL, Crawford PC. Seroprevalence of feline leukemia virus and feline immunodeficiency virus infection among cats in North America and risk factors for seropositivity. J Am Vet Med Assoc 2006; 228: 371–376. [DOI] [PubMed] [Google Scholar]
- 2. Nakamura Y, Nakamura Y, Ura A, Hirata M, Sakuma M, Sakata Y, et al. An updated nation-wide epidemiological survey of feline immunodeficiency virus (FIV) infection in Japan. J Vet Med Sci 2010; 72: 1051–1056. [DOI] [PubMed] [Google Scholar]
- 3. Little S, Sears W, Lachtara J, Bienzle D. Seroprevalence of feline leukemia virus and feline immunodeficiency virus infection among cats in Canada. Can Vet J 2009; 50: 644–648. [PMC free article] [PubMed] [Google Scholar]
- 4. Dean GA, Reubel GH, Moore PF, Pedersen NC. Proviral burden and infection kinetics of feline immunodeficiency virus in lymphocyte subsets of blood and lymph node. J Virol 1996; 70: 5165–5169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Torten M, Franchini M, Barlough JE, George JW, Mozes E, Lutz H, et al. Progressive immune dysfunction in cats experimentally infected with feline immunodeficiency virus. J Virol 1991; 65: 2225–2230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bishop SA, Williams NA, Gruffydd-Jones TJ, Harbour DA, Stokes CR. Impaired T-cell priming and proliferation in cats infected with feline immunodeficiency virus. AIDS 1992; 6: 287–293. [DOI] [PubMed] [Google Scholar]
- 7. O’Connor TP, Tanguay S, Steinman R, Smith R, Barr MC, Yamamoto JK, et al. Development and evaluation of immunoassay for detection of antibodies to the feline T-lymphotropic lentivirus (feline immunodeficiency virus). J Clin Microbiol 1989; 27: 474–479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Pinches MD, Diesel G, Helps CR, Tasker S, Egan K, Gruffydd-Jones TJ. An update on FIV and FeLV test performance using a Bayesian statistical approach. Vet Clin Pathol 2007; 36: 141–147. [DOI] [PubMed] [Google Scholar]
- 9. Connell S. Manufacturer addresses concerns about FIV vaccine. J Am Vet Med Assoc 2003; 222: 149. [PubMed] [Google Scholar]
- 10. Levy JK, Crawford PC, Slater MR. Effect of vaccination against feline immunodeficiency virus on results of serologic testing in cats. J Am Vet Med Assoc 2004; 225: 1558–1561. [DOI] [PubMed] [Google Scholar]
- 11. Andersen PR, Tyrrell P. Feline immunodeficiency virus diagnosis after vaccination. Anim Health Res Rev 2004; 5: 327–330. [DOI] [PubMed] [Google Scholar]
- 12. Uhl EW, Heaton-Jones TG, Pu R, Yamamoto JK. FIV vaccine development and its importance to veterinary and human medicine: a review: FIV vaccine 2002 update and review. Vet Immunol Immunopathol 2002; 90: 113–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bienzle D, Reggeti F, Wen X, Little S, Hobson J, Kruth S. The variability of serological and molecular diagnosis of feline immunodeficiency virus infection. Can Vet J 2004; 45: 753–757. [PMC free article] [PubMed] [Google Scholar]
- 14. Crawford PC, Slater MR, Levy JK. Accuracy of polymerase chain reaction assays for diagnosis of feline immunodeficiency virus infection in cats. J Am Vet Med Assoc 2005; 226: 1503–1507. [DOI] [PubMed] [Google Scholar]
- 15. Morton JM, McCoy RJ, Kann RK, Gardner IA, Meers J. Validation of real-time polymerase chain reaction tests for diagnosing feline immunodeficiency virus infection in domestic cats using Bayesian latent class models. Prev Vet Med 2012; 104: 136–148. [DOI] [PubMed] [Google Scholar]
- 16. Alonzo TA, Pepe MS. Using a combination of reference tests to assess the accuracy of a new diagnostic test. Statist Med 1999; 18: 2987–3003. [DOI] [PubMed] [Google Scholar]
- 17. Crawford PC, Levy JK. New challenges for the diagnosis of feline immunodeficiency virus infection. Vet Clin North Am Small Anim Pract 2007; 37: 335–350. [DOI] [PubMed] [Google Scholar]
- 18. D’haene B, Vandesompele J, Hellemans J. Accurate and objective copy number profiling using real-time quantitative PCR. Methods 2010; 50: 262–270. [DOI] [PubMed] [Google Scholar]
- 19. Daskalakis D. HIV diagnostic testing: evolving technology and testing strategies. Top Antivir Med 2011; 19: 18–22. [PMC free article] [PubMed] [Google Scholar]
- 20. Ishida T, Washizu T, Toriyabe K, Motoyoshi S, Tomoda I, Pedersen NC. Feline immunodeficiency virus infection in cats of Japan. J Am Vet Med Assoc 1989; 194: 221–225. [PubMed] [Google Scholar]
- 21. Lee IT, Levy JK, Gorman SP, Crawford PC, Slater MR. Prevalence of feline leukemia virus infection and serum antibodies against feline immunodeficiency virus in unowned free-roaming cats. J Am Vet Med Assoc 2002; 220: 620–622. [DOI] [PubMed] [Google Scholar]
- 22. Norris JM, Bell ET, Hales L, Toribio JA, White JD, Wigney DI, et al. Prevalence of feline immunodeficiency virus infection in domesticated and feral cats in eastern Australia. J Feline Med Surg 2007; 9: 300–308. [DOI] [PMC free article] [PubMed] [Google Scholar]