Can IDEXX Angio DetectTM accurately detect canine angiostrongylosis?

s 123 40 18 11 27 7 PubMed 69 15 23 8 17 6 Total relevant papers when duplicates removed 8 CONFLICT OF INTEREST The authors declare no conflict of interest. This Knowledge Summary was originally prepared, by the first author, as a graded assessment for the Evidence Based Veterinary Medicine Module as part of the University of Liverpool’s RCVS accredited CertAVP postgraduate qualification. The first author wishes to thank the assessor for their feedback which has been extremely beneficial while preparing the revised Knowledge Summary for publication. Thanks also to Clare Boulton, Head of Library and Knowledge Services at RCVS Knowledge for assistance with the literature search during Knowledge Summary preparation.


The evidence
Three cross-sectional diagnostic test accuracy studies were found, of which two used case-control selection (Schnyder et al., 2014;and Canonne et al., 2018). The other study used cohort selection for the majority of the study population, but also included a group of known A. vasorum-positive cases and a control group of animals with other confirmed nematode infections (Liu et al., 2017). These studies were considered to represent the most appropriate study designs; however, they are subject to biases and limitations that limit the strength of evidence provided.
Two cross-sectional studies included some information regarding the accuracy of Angio Detect™ (Olivieri et al., 2017;and Lempereur et al., 2020), one of which also included a retrospective case series (Olivieri et al., 2017). A small cohort study at a single kennel tested dogs with Angio Detect™ and Baermann coprology on three occasions in a 6 week period (Di Cesare et al., 2014). A case-control study evaluated Angio Detect™ and Baermann coprology in dogs presenting with clinical signs suggestive of A. vasorum infection and a control group of dogs presented to participating practices for routine preventive health care or unrelated conditions (Lempereur et al., 2016). One case series utilised Angio Detect™ and Baermann coprology in addition to thoracic ultrasonography and radiography in dogs presenting with respiratory distress to participating practices (Venco et al., 2021).
When not included within the publication, the sensitivity and specificity (with 95% confidence intervals [CI]) of Angio Detect™ in comparison to Baermann coprology were calculated (Sergeant, 2018), where possible, and are presented in italic font within the summary of evidence tables.

Summary of the evidence 1. Canonne et al. (2018)
Population: • Client-owned dogs presented to the University Veterinary Small Animal Teaching Hospital of Liège, Belgium, between March 2013-April 2017 were retrospectively selected. • Dogs were of various breeds, aged from 9 months-10 years. • Inclusion criteria were respiratory clinical signs (cough, respiratory distress or exercise intolerance) and recorded diagnostic results for natural A. vasorum infection (Baermann coprology, Angio Detect™, and enzyme linked immunosorbent assays [ELISA] both for antigens and specific antibodies). • A. vasorum infection was suspected based on compatible respiratory signs (as above), radiographic findings, bronchoscopy, and bronchoalveolar lavage (BAL) cytology examination. • Owner-reported duration of respiratory clinical signs varied from 2 weeks-2 months. • Angiostrongylosis was confirmed in all seven dogs by: o positive quantitative (q) PCR from BAL samples and clinical recovery after administration of either 50 mg/kg q24h fenbendazole PO for 3 week treatment duration or 2.5 mg/kg moxidectin repeated after 2-4 weeks.
Sample size: Seven dogs (two male, five female).

Intervention details:
• At the time of diagnosis one or two serum samples were obtained and frozen prior to analysis.

Limitations:
• Volunteer bias could have been introduced due to the selection process of small animal practices. There was variability in faecal sample collection and the frequency with <3 consecutive daily samples collected was not reported. Both of these factors introduce measurement bias to Baermann coprology. • It is unclear what time period was considered recent for faecal collection. The method of faecal storage prior to Baermann coprology was also not explained. There may have been inconsistencies and variation, introducing Baermann coprology measurement bias. • None of the dogs that tested positive by either test method had signs of gastrointestinal parasitism.

Limitations:
• Volunteer bias could have been introduced due to the selection process of small animal practices. spectrum and/or referral bias as the spectrum of clinical signs and disease severity could vary in different clinical settings. • The method(s) used for faecal sample transport was not described and it is unclear if they were consistent. If faecal storage was not adequate it could have resulted in falsenegative results obtained by Baermann coprology. • The process of thawing the serum/plasma samples and duration of sample freezing was not described and it is therefore uncertain if these could have altered Angio Detect™ testing and introduced measurement bias. • It is unknown how many faecal samples were obtained over multiple days for a single patient compared to single faecal samples submitted, which could introduce measurement bias for Baermann coprology.

Olivieri et al. (2017)
Population: Kennel A. vasorum cross-sectional study: • Dogs that resided at a kennel in Italy.

Limitations:
• The method of selecting the eligible population is not reported and there is the potential for selection bias. • 25/26 dogs were reported by their owners to be receiving monthly ivermectin-based chemoprophylaxis at the time of presentation, but time since last administration was not reported. • Single, rather than three consecutive day, faecal samples were obtained for Baermann coprology, potentially reducing Baermann sensitivity and introducing measurement bias.

Appraisal, application and reflection
Canine angiostrongylosis varies from acute presentation of cardiovascular signs, neurological deficits, haematological abnormalities to subclinical infection, making diagnosis challenging to the clinician (Di Cesare  et   due to the small sample size in these studies, the confidence intervals for calculated sensitivity results were wide. Unanimous results were not obtained using both diagnostic methods with conflicting results occurring within the literature. Antigen detection inhibited by antigen-antibody complexes were considered the reason for negative results obtained with Angio Detect™ where Baermann coprology was positive (Schnyder et  Cross-sectional studies are better suited to estimate prevalence rather than evaluating diagnostic test accuracy, therefore the evidence obtained is of low strength, but still contributes to the PICO question (Olivieri et al., 2017;and Lempereur et al., 2020). The population size in Oliveri et al. (2017) was small. Although origin of the individual was recorded the relevance of this information in relation to interpretation was not explained. The sample size of the other cross-sectional study was larger overall; however, the prevalence of A. vasorum within the study population was low (Lempereur et al., 2020). Details on eligibility criteria were minimal and further information on why the dog presented to the clinic along with any clinical signs would have been useful due to the broad spectrum of clinical presentations A. vasorum-positive dogs can present with. It was uncertain if and when reported administration of anthelmintic treatment was based on owner declaration (which may be subject to recall bias) or clinical records (misclassification bias is possible if notes are inaccurate or if the anthelmintic was obtained from another source). Due to the practices entering the study voluntarily this could introduce volunteer bias. The reason that Baermann coprology and Angio Detect™ tests were not done on all dogs was not explained and could introduce measurement bias.
The retrospective case series is low in the hierarchy of evidence for evaluation of diagnostic accuracy (Olivieri et al., 2017). Naturally infected case inclusion criteria were limited. The description of methods for archival retrieval was not particularly detailed, which may introduce misclassification bias if author-led selection occurred. Missing data could potentially rule additional A. vasorum cases out of inclusion. It is unknown how many clinicians treated these cases with potentially no standardisation in clinical record keeping. Presumably differing clinician opinion caused inconsistency in the diagnostic approach to each patient.
The case-control study (Lempereur et  The population size was small for the cohort study (Di Cesare et al., 2014). Population demographic information was not recorded, and the publication also lacked any statistical analysis of obtained results. The aim of the study was not clearly defined nor was the process of kennel selection, with only a single kennel purposively selected. Detailed information of the kennel's previous A. vasorum history was undisclosed.
The population included in the case series was restricted to dogs under 2 years of age presenting to one of three institutions (Venco et al., 2021). Insufficient information was provided to ascertain if referral bias existed as the level(s) of veterinary care provided by these institutions (i.e. first opinion or referral) was not reported. There was inconsistency of faecal collection duration in the PICO-relevant literature. Only one study reported desiccation of faecal samples as a complication (Lempereur et al., 2016), with the other studies not mentioning if this also happened. Two studies did not describe methods for transporting faecal samples for submission (Liu et al., 2017;and Lempereur et al., 2020). The Baermann technique relies on active parasite larvae to obtain results therefore fresh faeces are preferable. If delay in testing is unavoidable, faecal samples should be stored at 2-8°C and transported on ice within a 24-hour period (Broussard, 2003). Delaying Baermann testing and/or the use of inappropriate transport media could result in false-negative results. Analysing the available evidence, there is weak evidence indicating that Angio Detect™ is a highly specific and moderately sensitive test for diagnosing angiostrongylosis, in comparison to Baermann coprology. Therefore, in the clinical scenario, the application of the quick, in-house Angio Detect™ test will be an aid for the clinician to rule out angiostrongylosis as a differential diagnosis for the canine patient. However, it should be appreciated that the sample sizes of the studies are small and the comparator test (Baermann coprology) is a debatable reference standard with questionable sensitivity due to intermittent shedding and pre-patent period. As Angio Detect™ has moderate sensitivity, if a negative result were obtained in the clinical scenario (i.e. a patient with high clinical suspicion of angiostrongylosis) additional diagnostic procedures, such as BAL, would be advised. Total relevant papers when duplicates removed 8