Assessing the feasibility of applying machine learning to diagnosing non-effusive feline infectious peritonitis.

Feline infectious peritonitis (FIP) is a severe feline coronavirus-associated syndrome in cats, which is invariably fatal without anti-viral treatment. In the majority of non-effusive FIP cases encountered in practice, confirmatory diagnostic testing is not undertaken and reliance is given to the interpretation of valuable, but essentially non-specific, clinical signs and laboratory markers. We hypothesised that it may be feasible to develop a machine learning (ML) approach which may be applied to the analysis of clinical data to aid in the diagnosis of disease. A dataset encompassing 1939 suspected FIP cases was scored for clinical suspicion of FIP on the basis of history, signalment, clinical signs and laboratory results, using published guidelines, comprising 683 FIP (35.2%), and 1256 non-FIP (64.8%) cases. This dataset was used to train, validate and evaluate two diagnostic machine learning ensemble models. These models, which analysed signalment and laboratory data alone, allowed the accurate discrimination of FIP and non-FIP cases in line with expert opinion. To evaluate whether these models may have value as a diagnostic tool, they were applied to a collection of 80 cases for which the FIP status had been confirmed (FIP: n = 58 (72.5%), non-FIP: n = 22 (27.5%)). Both ensemble models detected FIP with an accuracy of 97.5%, an area under the curve (AUC) of 0.969, sensitivity of 95.45% and specificity of 98.28%. This work demonstrates that, in principle, ML can be usefully applied to the diagnosis of non-effusive FIP. Further work is required before ML may be deployed in the laboratory as a diagnostic tool, such as training models on datasets of confirmed cases and accounting for inter-laboratory variation. Nevertheless, these results illustrate the potential benefit of applying ML to standardising and accelerating the interpretation of clinical pathology data, thereby improving the diagnostic utility of existing laboratory tests.

Validation of an automated immunoturbidimetric assay for feline serum amyloid A.

BACKGROUND: Serum Amyloid A (SAA) is a major acute phase protein in cats, increasing rapidly in response to various inflammatory diseases. An automated latex-enhanced immunoturbidimetric assay for human SAA (LZ-SAA, Eiken), previously validated for use in cats, has had further major modification (VET-SAA, Eiken) for specific use in veterinary diagnostic laboratories but has yet to be validated in cats. RESULTS: Intra-assay and inter-assay CVs for the VET-SAA assay ranged from 1.88-3.57% and 3.98-6.74%, respectively. Linearity under dilution was acceptable with no prozone effect observed. Limit of detection was 1.65 mg/L and limit of quantification was 6 mg/L. Haemoglobin and triglyceride showed no adverse interference, but bilirubin produced positive bias in samples with low SAA. Comparison with the LZ-SAA assay showed significant correlation with proportional bias increasing as SAA concentration increased, likely related to differing calibration standards. SAA was significantly higher in patients with inflammatory disease compared with non-inflammatory disease, and in patients with moderate to highly elevated α1-AGP compared with patients with normal α1-AGP. Improvement of the assay range may be required to fully evaluate differences between disease groups at low SAA levels. Based on ROC curve analysis, at a cut-off point of 20.1 mg/L the VET-SAA assay discriminated between inflammatory and non-inflammatory disease with sensitivity of 0.93 and specificity of 0.99. CONCLUSIONS: The automated VET-SAA assay is a robust, precise, and accurate method for measurement of feline SAA which can clearly identify patients with inflammatory disease. It should be a valuable biomarker for use in feline medicine.

Disseminated Mycobacterium avium infection in a cat on long-term ciclosporin therapy and potential latent infection of an in-contact cat.

Case summary: An 8-year-old domestic shorthair cat receiving long-term ciclosporin treatment was evaluated for a history of weight loss and hyporexia. The main clinical finding was a cluster of enlarged mesenteric lymph nodes. Cytological examination of fine-needle aspirates showed granulomatous inflammation with abundant acid-fast bacilli. A diagnosis of Mycobacterium avium complex (MAC) infection was confirmed by PCR. The cat's clinical condition deteriorated rapidly despite appropriate antibiotic treatment and it was euthanased 2 weeks after initial presentation due to development of severe paraparesis and ataxia. Post-mortem examination revealed granulomatous inflammation affecting multiple lymph nodes and other organs with intrahistiocytic acid-fast bacilli consistent with mycobacteria when stained using Ziehl-Neelsen stain. Another cat in the same household was screened for infection using the interferon gamma release assay (IGRA), with the result being consistent with infection by non-tuberculous mycobacteria (NTM), which includes MAC; however, it had no grossly detectable disease. Relevance and novel information: This case report is an unusual presentation of disseminated MAC infection in a cat, which remains a rare diagnosis. Clinicians should be aware of unusual and rare presentations of this infection. The clinical findings, progression of disease and histopathology results add to the current clinical database for feline disseminated MAC infections. Another cat in the same household tested positive for NTM by IGRA without any gross disease. This was suggestive of latent MAC infection which, to our knowledge, has not been previously reported in an in-contact cat.

Primary multifocal muscular T-cell lymphoma with cutaneous involvement in a dog: A case report and review of the literature.

Canine lymphoma represents a heterogeneous group of lymphoid neoplasms, with multicentric nodal lymphoma being the most common presentation. Musculoskeletal involvement is uncommon, and primary muscular lymphoma is a very rare presentation. Only a few cases have been described in dogs, which were of variable classification and immunophenotype. Here, we report the case of a 5-year-old female neutered Beagle that presented with an intramuscular mass on the right shoulder and associated lameness and lethargy. One month after initial presentation, multiple cutaneous nodules appeared on the head, and staging with advanced imaging revealed additional masses affecting other muscles. Cytology, histopathology, immunohistochemistry, and PCR for antigen receptor rearrangements of one of the muscle masses and skin lesions supported a diagnosis of peripheral T-cell lymphoma with large granular lymphocytes at both sites. The dog was euthanized after diagnosis due to the poor prognosis. This is the first report of primary muscular peripheral T-cell lymphoma with large granular lymphocytes and cutaneous involvement in the dog. Despite being a rare presentation, lymphoma must be considered a differential in dogs presenting with a discrete, intramuscular, soft tissue mass.

Use of hyaluronidase in the comparison between manual and automated hematology analysis with the ADVIA 120 to improve analysis of feline body cavity effusions.

Classification of body cavity effusions is an important step in the investigation and diagnosis of disease in cats. Feline inflammatory effusions are often highly proteinaceous and viscous, which can cause clumping of white cells and subsequently inaccurate nucleated cell counts (NCCs) using automated and manual methods. Microscopic assessment of cellularity can also be difficult given the variable thickness of smears and cell clumping, which skews white cell distribution. The ADVIA 120 uses 2 white cell-counting channels, the basophil/lobularity (WBC/baso) and differential/peroxidase channels, which can provide quite different results in highly viscous feline samples and often disagree with smear assessment of cellularity. We investigated the effects of pre-incubation of feline effusion samples with hyaluronidase and its effects on NCCs and cellularity assessment. NCCs were obtained by automated analysis using the ADVIA 120 and by manual counting methods. Agreement was assessed using a Bland-Altman chart. Pretreatment of samples with hyaluronidase resulted in good agreement between the ADVIA basophil channel and manual counting methods in all samples in the study. However, improvements in NCCs after hyaluronidase treatment were significantly greater in clumped samples, and cell distribution of these samples on direct smears was also improved. Therefore, when nucleated cell clumping is observed on a direct smear, pretreatment of the sample with hyaluronidase prior to analysis on an automated analyzer is advised, with the WBC/baso channel displaying the most accurate NCC.

Optimisation and validation of a PCR for antigen receptor rearrangement (PARR) assay to detect clonality in canine lymphoid malignancies.

PCR for antigen receptor gene rearrangements (PARR) analysis is being increasingly used to assist diagnosis of canine lymphoma. In this study, PARR was carried out on consecutive samples received as part of routine diagnostic practice from 271 patients: 195 with lymphoid malignancies, 53 with reactive conditions and 23 with other neoplasms. Initially, published primer sets were used but later minor primer modifications were introduced and primers were rationalised to give a PARR panel that provides a good compromise between sensitivity and cost. Results were compared to diagnoses made by histology or cytology, coupled with immunophenotyping by flow cytometry or immunohistochemistry where possible. After exclusion of 11 poor quality samples, 230/260 (88%) gave a clear result with 162/163 (99%) of samples classified as clonal and 56/67 (84%) classified as polyclonal giving results concordant with the cytological/histological diagnosis. Among 30 samples with equivocal results, 21 had clonal peaks in a polyclonal background and nine showed little amplification. These were from patients with a range of neoplastic and non-neoplastic conditions emphasising the need to interpret such results carefully in concert with other diagnostic tests. The combination of primer sets used in this study resulted in a robust, highly specific and sensitive assay for detecting clonality.

Early post parturient changes in milk acute phase proteins.

The periparturient period is one of the most critical periods in the productive life of a dairy cow, and is the period when dairy cows are most susceptible to developing new intramammary infections (IMI) leading to mastitis. Acute phase proteins (APP) such as haptoglobin (Hp), mammary associated serum amyloid A3 (M-SAA3) and C-reactive protein (CRP) have been detected in milk during mastitis but their presence in colostrum and milk in the immediate postpartum period has had limited investigation. The hypothesis was tested that APP are a constituent of colostrum and milk during this period. Enzyme linked immunosorbent assays (ELISAs) were used to determine each APP's concentration in colostrum and milk collected daily from the first to tenth day following calving in 22 Holstein-Friesian dairy cows. Haptoglobin was assessed in individual quarters and composite milk samples while M-SAA3 and CRP concentration were determined in composite milk samples. Change in Hp in relation to the high abundance proteins during the transition from colostrum to milk were evaluated by 1 and 2 dimension electrophoresis and western blot. In 80% of the cows all APPs were detected in colostrum on the first day following parturition at moderately high levels but gradually decreased to minimal values in the milk by the 6th day after calving. The remaining cows (20%) showed different patterns in the daily milk APP concentrations and when an elevated level is detected could reflect the presence of IMI. Demonstration that APP are present in colostrum and milk following parturition but fall to low levels within 4 days means that elevated APP after this time could be biomarkers of post parturient mastitis allowing early intervention to reduce disease on dairy farms.

Effect of pre-analytical treatments on bovine milk acute phase proteins.

BACKGROUND: Samples for diagnostic procedures often require some form of pre-analytical preparation for preservation or safe handling during transportation prior to analysis in the laboratory. This is particularly important for milk samples which frequently need preservatives to retain milk composition as close to that found in freshly collected samples as possible. METHODS: Milk samples were treated by heating at 56 °C for 30 min or preserved by addition of either potassium dichromate or bronopol respectively. Haptoglobin (Hp), mammary associated serum amyloid A3 (M-SAA3) and C-reactive protein (CRP) were measured in the various treatment groups and in control samples which were not treated, using enzyme linked immunoassays. The concentrations of each APP were compared between treated and non-treated groups using the Wilcoxon signed ranks tests. RESULTS: Heat treatment of samples was found to have a significant lowering effect on milk M-SAA3 and CRP but not Hp. The use of potassium dichromate and bronopol as preservatives in milk had no significant effects on milk Hp and M-SAA3 concentration but lowered milk CRP values compared to controls. CONCLUSIONS: The observed effects of heating and preservative use on milk APP should be taken into consideration when assaying samples which have undergone heat treatment as a result of international transfer regulations involving biological samples or samples needing chemical preservation prior to transport to laboratory.

The major acute phase proteins of bovine milk in a commercial dairy herd.

BACKGROUND: Milk acute phase proteins (APP) have been identified and show promise as biomarkers of mastitis. However analysis of their profile in dairy cows from a production herd is necessary in order to confirm their benefits in mastitis diagnosis. The profiles of milk haptoglobin (Hp), mammary associated serum amyloid A3 (M-SAA3) and C-reactive protein (CRP) were determined in 54 composite milk (milk from all functional quarters of a cow's udder collected in a common receptacle) samples (CMS) from a commercial dairy farm. Milk Hp was also determined in individual quarter milk (milk from a single udder quarter) samples (QMS) (n = 149) of the cows. An ELISA was developed and validated for the determination of milk Hp while commercial kits were used for M-SAA3 and CRP assay respectively. Composite milk APP results were compared with cow factors including parity, stage of lactation, percentage protein and fat as well as somatic cell counts (SCC). RESULTS: Composite milk Hp ranged from <0.4-55 µg/ml with a median of 3.5 µg/ml; composite milk M-SAA3 ranged from <0.6-50 µg/ml and had a median of 1.2 µg/ml, while CRP ranged from <1.80-173 ng/ml and had a median of 24.6 ng/ml. Significant correlations were found between composite SCC and Hp (P-value <0.009) as well as parity and Hp (P < 0.009), but not between M-SAA3 and SCC, M-SAA3 and Hp, M-SAA3 and CRP or M-SAA3 and parity. Milk CRP was correlated with % fat (P = 0.002) and % protein (P = 0.001) of the milk samples. The lack of correlation of SCC with the M-SAA3 and CRP could result from these APP being more sensitive to intra-mammary infection than SCC. Quarter milk Hp had a range of <0.4-420 µg/ml with a median value of 3.6 µg/ml, with 92 % of samples below 20 µg/ml. CONCLUSION: Baseline values of Hp, M-SAA3 and CRP were established in composite milk from cows with normal SCC on the dairy farm. Parity was recognized as a possible confounding factor when diagnosing mastitis using Hp. The value of the APP, Hp, M-SAA3 and CRP as substitutes or to complement SCC in indicating udder inflammation, was demonstrated.

A case report of Mycoplasma wenyonii associated immune-mediated haemolytic anaemia in a dairy cow.

BACKGROUND AND CASE PRESENTATION: A three year old, second lactation Holstein dairy cow presented to the Scottish Centre for Production Animal Health and Food Safety, Glasgow University Veterinary School in November 2014 with a history of post-calving vulval/vaginal bleeding nine days prior to presentation, followed by a sudden reduction in milk yield. Subsequent investigations resulted in a diagnosis of immune-mediated haemolytic anaemia secondary to infection with Mycoplasma wenyonii. CONCLUSION: This report of a novel presentation of Mycoplasma wenyonii in a dairy cow illustrates the need to consider M.wenyonii as a potential differential diagnosis when a cow presents with anaemia and will discuss the potential implications of the condition at herd-level.

Use of CD134 as a primary receptor by the feline immunodeficiency virus.

Feline immunodeficiency virus (FIV) induces a disease similar to acquired immunodeficiency syndrome (AIDS) in cats, yet in contrast to human immunodeficiency virus (HIV), CD4 is not the viral receptor. We identified a primary receptor for FIV as CD134 (OX40), a T cell activation antigen and costimulatory molecule. CD134 expression promotes viral binding and renders cells permissive for viral entry, productive infection, and syncytium formation. Infection is CXCR4-dependent, analogous to infection with X4 strains of HIV. Thus, despite the evolutionary divergence of the feline and human lentiviruses, both viruses use receptors that target the virus to a subset of cells that are pivotal to the acquired immune response.