Multiplex fluorescent immunocytochemistry is a complementary test for the diagnosis of feline infectious peritonitis: a diagnostic clinical trial

Abstract

Objectives

Feline infectious peritonitis (FIP) is a common and treatable viral disease of cats for which there is no single, reliable ante-mortem diagnostic test. The goals of this study were to evaluate the diagnostic accuracy (sensitivity, specificity and positive and negative predictive values) of multiplex fluorescent immunocytochemistry (MF-ICC) for the diagnosis of FIP in clinical patients under field conditions, and to directly compare that accuracy to the diagnostic performance of other commonly used assays.

Methods

Cats presenting with clinical signs compatible with FIP were enrolled at two academic veterinary teaching hospitals and tested using a combination of complete blood count, biochemistry, fluid analysis, cytology, MF-ICC, serology and RT-PCR. Antibodies against both feline coronavirus and vimentin, as well as a nuclear stain, were used for the MF-ICC assay. Outcomes were determined by necropsy with histopathology and immunohistochemistry, response to antiviral therapy and/or clinical follow-up.

Results

A total of 84 cases comprising 58 cats with FIP and 26 without FIP (control cases) were included in this study. Using a threshold of one dual-expressing mononuclear cell or more, MF-ICC was found to have 77% sensitivity, 81% specificity, 92% positive predictive value and 53% negative predictive value in this cohort. Using a threshold of two or more dual-expressing mononuclear cells improved specificity but reduced both sensitivity and overall diagnostic accuracy. MF-ICC was more sensitive, but less specific, in effusion fluids over tissues. MF-ICC resulted in the highest overall diagnostic accuracy (78%) when compared with serology (75%), RT-PCR (69%) and serum albumin:globulin ratio (76%).

Conclusions and relevance

MF-ICC is an imperfect test but is considered complementary to other commonly used ante-mortem diagnostic assays for FIP. It can be used alongside RT-PCR and other minimally invasive tests to build a case for FIP in an individual patient.

Keywords

Feline infectious peritonitis coronavirus diagnostic accuracy multiplex fluorescent immunocytochemistry macrophage

Introduction

Feline infectious peritonitis (FIP) is a common coronaviral disease of great clinical importance to feline medicine. FIP affects cats and some non-domestic cats worldwide, with the highest incidence among young cats. It has a complex and still poorly understood pathogenesis but is widely accepted to be caused by transformation of the clinically mild feline enteric coronavirus (FECV) into the lethal, systemic feline infectious peritonitis virus (FIPV). FECV and FIPV are considered two different pathotypes of feline coronavirus (FCoV).¹ The main cellular tropism of FECV is colonic enterocytes, while FIPV can replicate efficiently within monocytes and macrophages, allowing for systemic spread.² If untreated, FIP is uniformly fatal; however, recent advances in antiviral therapy have opened the possibility of achieving clinical remission.^3,4

One of the most frustrating aspects of FIP for veterinary professionals and cat caregivers is the difficulty in achieving a diagnosis. Because there is no single, definitive, minimally invasive, ante-mortem diagnostic test for FIP, a presumptive diagnosis is typically achieved through a preponderance of clinicopathologic evidence, history and signalment. Virus-specific assays like RT-PCR, antigen detection and serology are complicated by their inability to distinguish between FECV and FIPV infections. The benefits, drawbacks and diagnostic accuracy of the various commercially available tests for FIP have been reviewed in detail.^5,6 Despite the development of novel tests such as spike gene mutation-based assays, which do not appear to outperform traditional RT-PCR,⁷ FIP remains a diagnosis of assumption for most patients. Particularly in light of newly available options for therapy of FIP,^3,4,8,9 an accurate, minimally invasive diagnostic test is urgently needed.

Our group recently developed and optimized a novel multiplex fluorescent immunocytochemistry (MF-ICC) assay to detect FCoV within macrophages using markers for both cellular and viral antigens.¹⁰ When applied to an FIP diagnosis, this assay is designed to capitalize on the characteristic gain-of-tropism for – and efficient replication of FIPV within – macrophages, which is typically not observed at high levels in patients with FECV alone. The aim of this study was to evaluate the diagnostic accuracy (sensitivity, specificity and positive and negative predictive values [PPV/NPV]) of MF-ICC for the diagnosis of FIP in clinical patients under field conditions. We hypothesized that MF-ICC would be more specific than published reports of standard immunocytochemistry (ICC) and would offer an improved method of achieving a definitive diagnosis over currently available tests (in particular, RT-PCR, serology and serum albumin:globulin [A:G] ratio).

Materials and methods

Animals

Patients were recruited from the Colorado State University (CSU) Veterinary Teaching Hospital and the Ohio State University (OSU) Veterinary Medical Center from September 2017 through July 2023. Cases were solicited from clinicians working in the small animal internal medicine, small animal emergency and critical care, and community practice services. Occasional cases were also recruited from neurology, ophthalmology or other specialty services, or from local participating animal shelters and referring veterinarians. The caregiver (or shelter veterinarian in the case of shelter cats) of every cat enrolled in this study provided informed consent, and the study received ethical approval at each institution (CSU Veterinary Teaching Hospital Clinical Review Board protocol #2017-111 and the OSU Institutional Animal Care and Use Committee protocol #2020A00000079).

Inclusion criteria were cats presenting with clinical signs of possible FIP: lethargy; hyporexia; fever; weight loss; hyperproteinemia with hyperglobulinemia; abdominal, thoracic or pericardial effusion; ocular or neurologic signs; and/or abnormal serum or fluid A:G ratio. Any animal with a clinical suspicion of FIP (‘wet’/effusive or ‘dry’/non-effusive form) could be included in this study. Exclusion criteria were cats presenting with a definitive diagnosis of a health condition (eg, neoplasia, chronic kidney disease, diabetes mellitus, toxoplasmosis) that could be causing similar signs.

Signalment and clinicopathologic data were recorded for each case. Cats were considered ‘effusive’ if any cavity effusion was detected by imaging, even if it was too little effusion to sample. As part of the study, the following diagnostic testing was performed: complete blood count (CBC), biochemistry panel, feline immunodeficiency virus (FIV)/feline leukemia virus (FeLV) point-of-care testing, urinalysis, fluid analysis with A:G ratio (if effusive), cytology of any abnormal organ (if non-effusive), effusion fluid or solid organ aspirate FCoV quantitative RT-PCR and FCoV serology. Given the nature of a clinical trial on critically ill patients, not all assays could be collected for each patient.

After enrollment, cats were ultimately included in this data set only if there was a known outcome regarding FIP. Cases were considered to have FIP if they met one of two criteria: (1) the animal died or was euthanized and underwent a necropsy with histopathologic confirmation of characteristic lesions and positive immunostaining by FCoV immunohistochemistry (IHC); or (2) the animal received a presumptive (based on PCR or other clinical diagnostics) clinical diagnosis of FIP and subsequently responded conclusively (ie, dramatic clinical improvement) to unlicensed GS-441524 antiviral therapy. Only unlicensed and unregulated (black market) antiviral therapy was available in the USA at the time this study was performed, which has been shown to be highly effective for at-home use despite a lack of veterinary oversight in many cases.¹¹ At the time, caregivers were sourcing these medications without direct veterinary involvement. The authors did not supply or administer any unlicensed products to patients included in this study. Cases were considered to not have FIP if they met one of three criteria: (1) the animal died or was euthanized and underwent a necropsy that ruled out FIP-related pathology; (2) the animal received a definitive diagnosis of non-FIP disease that explained the presenting clinical signs; or (3) the patient completely recovered from clinical signs despite no therapy for FIP. Cases determined to not have FIP were utilized as FIP-negative (control) cases.

Routine clinical diagnostics

CBC, biochemistry panel, urinalysis, cytology and fluid analysis were performed at the CSU or OSU reference laboratories. CBCs were performed on Advia 120 (CSU) or Advia 2120 (OSU) hematology analyzers, respectively (Siemens). Biochemistry panels and fluid A:G ratios were performed on Cobas c501 instruments (Roche). All analyzers are maintained according to typical veterinary quality assurance standards, including daily quality control protocols. FIV/FeLV point-of-care tests (SNAP Combo Test; IDEXX Laboratories) were performed either by the investigators, the attending clinicians or the CSU or OSU reference laboratory, depending on the circumstances of the case. Cytology and fluid analysis were performed according to routine protocols at each reference laboratory (CSU or OSU); microscopic evaluation was performed by a board-certified clinical pathologist and/or clinical pathology resident.

FCoV RT-PCR results were included in this study only if the attending veterinarian elected to submit this test (paid for by the caregiver rather than the study). These samples were all submitted as routine diagnostic submissions to the CSU Veterinary Diagnostic Laboratory, and each sample site was included independently (as with MF-ICC), such that some cats contributed more than one RT-PCR sample. Anti-FCoV serology testing was performed on a batched subset of this cohort on either previously frozen serum or plasma samples (when available) using a commercial ELISA kit (Feline Coronavirus Antibody Test Kit; IVD Technologies) and following the manufacturer’s instructions.

Multiplex fluorescent immunocytochemistry

MF-ICC was performed as previously described.¹⁰ Briefly, cavity fluid cells were washed once or twice with phosphate buffered saline (PBS; Fisher Scientific) depending on the fluid’s refractometric protein content. One wash was used for fluids with <5 mg/dl of protein; two washes were used for fluids with ⩾5 mg/dl of protein. Most clinical fluid samples were submitted in EDTA (lavender-top) tubes, although occasional no-additive (red-top) tubes were received by the reference laboratories for use with fluid analysis and/or MF-ICC. Concentrated cells were then spun onto glass slides with a double-funnel cytocentrifuge and fixed with 4% paraformaldehyde (PFA; Alfa Aesar). Washing and fixation occurred as soon as possible after fluid collection and no more than 48 h later. If fluids could not be processed immediately, they were stored at 4°C until washing and slide fixation. For non-effusive cases, unstained cytology slides were divided into two ‘wells’ using a Super PAP pen (Research Products International) and then fixed with 4% PFA.

After fixation, slides were stored at 4°C until the MF-ICC assay could be performed in batches (mostly within 1 week, but occasionally up to 12 weeks after fixation). Cells were then permeabilized with 0.25% Tween-20 (Sigma-Aldrich) and blocked with Background Sniper (Biocare Medical). The primary antibodies used were rabbit anti-vimentin (Biocare Medical) and mouse anti-feline coronavirus (Bio-Rad). An isotype control sample with Universal Negative Control Serum (BioCare Medical) was performed using the second cytospin circle for each slide. The secondary antibodies, goat anti-mouse IgG H&L conjugated to Dylight 488 (Novus) and goat anti-rabbit IgG H&L conjugated to Dylight 594 (Abcam), were applied to both the isotype control and the test circle. Lastly, 4′,6-diamidino-2-phenylindole (DAPI) Fluoromount-G (Electron Microscopy Sciences) was applied before cover-slipping. Slides were then reviewed under a microscope (Leica) with a direct fluorescence mount PhotoFluor LM-75 (89 North). The isotype control sample circle of each slide was used to set the acquisition settings of the microscope to minimize background autofluorescence. A sample was considered positive for FIP if co-expression of FCoV and vimentin was observed within at least one or two (separate analyses) intact mononuclear (macrophage-like) cells. Of note, the use of vimentin provided adequate size and structure to cells such that, along with nuclear staining, the identity of each cell was overt. Samples were considered non-diagnostic if fewer than five intact vimentin-positive cells were observed. Samples were reviewed and recorded at the time of analysis by the performing technician or research associate, and then later verified by investigators via review of composite still images.

Data analysis

Descriptive statistics were performed using Google Sheets (version 2024). P value determination on numerical clinicopathologic data was performed via Student’s t-test (Google Sheets). The sensitivity, specificity, PPV and NPV were calculated using standard formulae.¹² Cases were classified as follows: ‘true positive’ if MF-ICC results were positive and the case was FIP positive (according to criteria listed above); ‘true negative’ if MF-ICC results were negative and the case was FIP negative (according to criteria listed above); ‘false positive’ if MF-ICC results were positive but the case was FIP negative; ‘false negative’ if MF-ICC results were negative, but the case was FIP positive. Clopper–Pearson confidence intervals for diagnostic accuracy metrics were calculated where applicable. Finally, binary logistic regression for serum A:G ratio (as a continuous variable) and MF-ICC results was performed using SPSS version 31 (IBM).

Results

In total, 134 cases met the initial inclusion criteria and were enrolled in this study. A total of 16 cats died without a necropsy, 13 cases did not have a sample collected for MF-ICC, and an additional 21 cats did not have a known outcome regarding FIP as described above or were otherwise lost to follow-up. As a result, a total of 50 cases were excluded and 84 cases were included in this data set. Of these 84 included cases, 58 cats had FIP (n = 35 diagnosed via histopathology/IHC and n = 23 diagnosed via clinical data + positive response to antiviral therapy) and 26 did not have FIP (control cases), according to the criteria described above. For the 23 cats diagnosed via clinical data and positive response to GS-441524 antiviral therapy, very little information on the dosage, frequency, duration or route of administration of GS-441524 is available, as this occurred in caregivers’ homes outside of the control of this study. Of these 23 patients, the dosage (5–7 mg/kg) and duration (8–12 weeks) are known for only one animal. The route of administration is known to be subcutaneous injections for five animals, but unknown for the other 18 animals. However, a presumptive diagnosis was based on clinical data (signalment, history, physical examination, radiographs, ultrasound, CBC, biochemistry panel, fluid analysis, cytology and/or other infectious disease testing to rule out other differentials) and robust response to at-home GS-441524 administration. Although there is no way to clearly define a clinical presumptive diagnosis for FIP, very strong clinical suspicion was determined by a small animal internist who is also a certified feline specialist and very experienced with FIP (PC). Similarly, subsequent observation of dramatic clinical response to owner-driven antiviral therapy within 7 days and long-term clinical follow-up evaluation was conducted by the same clinician and author of this study (PC).

From these 84 cats, a total of 101 MF-ICC samples were analyzed, with multiple samples derived from some cats (eg, abdominal fluid and liver aspirate samples from the same cat). Of these 101 MF-ICC samples, results were unfortunately not recorded for three samples, and an additional 13 (13%) were considered non-diagnostic (ie, too few intact cells for interpretation). Thus, a total of 85 MF-ICC samples representing 69 cats (50 FIP cats and 19 control cats; 16 repeat samples) were available for analysis. The remaining 15 cats (for which the 16 MF-ICC results were unavailable) were still retained in this study as they contributed data for other comparative assays (eg, serology, RT-PCR). A schematic of patient enrollment is shown in Figure 1.

Figure 1

Schematic of patient enrollment totals. FIP = feline infectious peritonitis; ICC = immunocytochemistry; MF-ICC = multiplex fluorescent immunocytochemistry

Signalment and clinical attributes (Table 1) as well as clinicopathologic data (Table 2) for FIP-positive cases and non-FIP controls are presented for the 84 cats included in this study. Raw data for the diagnostic results from all 84 cats are provided in Table S1 in the supplementary material.

Table 1

Signalment and clinical attributes of feline infectious peritonitis (FIP)-positive and non-FIP (control) cases

Parameter	FIP	Non-FIP
Number	58	26
Age (years)	2.1 ± 2.7	5.9 ± 5.3
Sex/neuter status
Female intact	6	1
Female spayed	12	16
Male intact	4	0
Male castrated	31	9
Unknown	5	0
Effusion status
Non-effusive	11	16
Effusive	44	10
Unknown	3	0
Death/necropsy status
Alive at discharge	23	19
Died/necropsy performed	35	7

Data are n or mean ± SD

Table 2

Clinicopathologic data for feline infectious peritonitis (FIP)-positive cases and non-FIP controls

Parameter	FIP (n = 58)	Non-FIP (n = 26)	P value
Selected CBC values
HCT (%)	27.6 ± 7.6 (45)	31.1 ± 8.5 (24)	0.086
WBC (×10³/µl)	15.6 ± 7.4 (45)	23.8 ± 18.7 (18)	0.011*
Neutrophils (×10³/µl)	13.8 ± 7.5 (43)	18.6 ± 12.4 (23)	0.053
Lymphocytes (×10³/µl)	1.3 ± 1.1 (44)	2.3 ± 1.6 (24)	0.004*
Monocytes (×10³/µl)	0.4 ± 0.3 (38)	0.8 ± 0.5 (24)	0.001*
PLT (×10³/µl)	212.1 ± 158.5 (45)	322.1 ± 181.1 (24)	0.011*
MPV (fl)	20.5 ± 6.4 (41)	20.5 ± 6.4 (24)	0.004*
Plasma protein (g/dl)	8.5 ± 1.8 (33)	7.6 ± 1.2 (21)	0.046*
Selected serum biochemistry values
Total protein (g/dl)	8.4 ± 2.3 (43)	7.0 ± 1.3 (24)	0.011*
Albumin (g/dl)	2.2 ± 0.5 (44)	2.8 ± 0.6 (24)	<0.001*
Globulins (g/dl)	6.1 ± 2.4 (44)	4.2 ± 1.2 (24)	0.001*
A:G ratio	0.4 ± 0.2 (44)	0.7 ± 0.3 (24)	<0.001*
Total bilirubin (mg/dl)	1.0 ± 1.3 (44)	1.0 ± 3.6 (24)	0.923
Creatinine (mg/dl)	0.8 ± 0.4 (45)	1.7 ± 2.1 (24)	0.008*
Cholesterol (mg/dl)	146.1 ± 42.6 (44)	166.1 ± 72.7 (24)	0.158
ALP (IU/l)	26.5 ± 30.1 (44)	44.2 ± 66.7 (23)	0.138
GGT (IU/l)	0.6 ± 1.3 (40)	4.2 ± 16.0 (24)	0.161
AST (IU/l)	112.9 ± 149.0 (41)	67.7 ± 69.1 (22)	0.184
ALT (IU/l)	57.9 ± 58.7 (44)	81.9 ± 100.8 (24)	0.218
BUN (mg/dl)	19.0 ± 11.6 (44)	37.7 ± 39.1 (24)	0.004*
Calcium (mg/dl)	8.6 ± 0.8 (44)	9.6 ± 1.2 (24)	<0.001*
Sodium (mEQ/l)	146.0 ± 5.1(45)	150.0 ± 6.2 (24)	0.006*
Osmolality (mOsm/kg)	291.6 ± 11.0 (39)	308.8 ± 17.4 (24)	<0.001*
Glucose (mg/dl)	109.5 ± 35.1 (45)	146.8 ± 104.4 (24)	0.032*
Cavity fluid analysis
TNCC (#/µl)	5582 ± 5964 (33)	13,557 ± 27,520 (7)	0.125
Total protein (g/dl)	5.6 ± 1.4 (32)	4.0 ± 2.1 (8)	0.005*
A:G ratio	0.5 ± 0.2 (16)	1.6 (1)	N/A
% Neutrophils	63.1 ± 21.2 (31)	76.8 ± 17.2 (7)	0.120
% Macrophages (large mononuclear cells)	24.9 ± 20.2 (31)	11.1 ± 10.2 (9)	0.089
Cytology results
Inflammation	95 (20)	38 (10)	–
Neoplasia	0 (0)	27 (7)	–
Other	5 (1)	35 (9)	–
FCoV serology
Positive	94 (17)	50 (5)	–
Borderline	0	10 (1)	–
Negative	6 (1)	40 (4)	–
FCoV RT-PCR
Positive	58 (14)	8 (1)	–
Negative	42 (10)	92 (11)	–

Data are n (%) or mean ± SD (number of cats with data available for that parameter)

Statistically significant difference (P value <0.05)

A:G = albumin:globulin; ALP = alkaline phosphatase; ALT = alanine aminotransferase; AST = aspartate aminotransferase; BUN = blood urea nitrogen; FCoV = feline coronavirus; GGT = gamma-glutamyl transferase; HCT = hematocrit; MPV = mean platelet volume; PLT = platelet; TNCC = total nucleated cell count; WBC = white blood cell

Data were analyzed using two different thresholds for counting a case as positive for FIP via MF-ICC: first with the threshold of two or more intact mononuclear (macrophage-like) cells that display clear co-expression of FCoV and vimentin, and second with the threshold of one or more such cells. Examples of a positive MF-ICC result (Figure 2), negative MF-ICC result (Figure 3) and aberrant staining of granulocytes (counted as a negative result) (Figure 4) are presented. Diagnostic performance of the MF-ICC assay is presented in Tables 3 (⩾2 positive cells) and 4 (⩾1 positive cell).

Figure 2

Example of a positive multiplex fluorescent immunocytochemistry assay result on an abdominal effusion fluid sample from a cat later determined to have feline infectious peritonitis through necropsy with histopathology and immunohistochemistry. (a) 4′,6-diamidino-2-phenylindole, (b) feline coronavirus, (c) vimentin and (d) overlay

Figure 3

Example of a negative multiplex fluorescent immunocytochemistry assay result on a jejunal lymph node aspirate from a cat later determined to have non-feline infectious peritonitis disease (food-responsive inflammatory bowel disease) through clinical management with a hydrolyzed diet (and no antiviral therapy). (a) 4′,6-diamidino-2-phenylindole, (b) feline coronavirus, (c) vimentin and (d) overlay

Figure 4

Example of aberrant or artifactual feline coronavirus-positive staining of presumed neutrophils (cells with clearly segmented nuclei) within a multiplex fluorescent immunocytochemistry assay result on a liver aspirate from a cat later determined to have non-feline infectious peritonitis disease (sterile pyogranulomatous hepatitis) through histopathology and immunohistochemistry (negative for FCoV). (a) 4′,6-diamidino-2-phenylindole, (b) feline coronavirus, (c) vimentin and (d) overlay

Table 3

Diagnostic performance of the multiplex fluorescent immunocytochemistry (MF-ICC) assay at a threshold of ⩾2 positive cells (n = 85 samples)

Parameter	FIP	Non-FIP
MF-ICC positive	41	3
MF-ICC negative	23	18
Diagnostic accuracy (%)	69.4 (58.5–79.0)
Sensitivity (%)	64.1 (51.1–75.7)
Specificity (%)	85.7 (63.7–96.9)
PPV (%)	93.2 (82.5–97.5)
NPV (%)	43.9 (35.1–53.1)

Data in parentheses are 95% confidence intervals

FIP = feline infectious peritonitis; NPV = negative predictive value; PPV = positive predictive value

Table 4

Diagnostic performance of the multiplex fluorescent immunocytochemistry (MF-ICC) assay at a threshold of ⩾1 positive cell (n = 85 samples)

Parameter	FIP	Non-FIP
MF-ICC positive	49	4
MF-ICC negative	15	17
Diagnostic accuracy (%)	77.7 (67.3–85.9)
Sensitivity (%)	76.6 (64.3–86.3)
Specificity (%)	81.0 (58.1–94.6)
PPV (%)	92.4 (83.4–96.8)
NPV (%)	53.1 (41.0–64.9)

Data in parentheses are 95% confidence intervals

FIP = feline infectious peritonitis; NPV = negative predictive value; PPV = positive predictive value

Because the overall diagnostic performance was improved using the lower threshold of a single intact dual-expressing mononuclear cell, this threshold was used for all subsequent analyses. A total of 51 fluid (body cavity effusion) samples for MF-ICC, representing 46 cats, were included in this study. The diagnostic performance of MF-ICC in fluid/effusion samples is presented in Table 5. A total of 34 tissue samples for MF-ICC, representing 23 cats, were included in this study. Tissue samples included aspirates of spleen, liver, lymph nodes (peripheral and intrabdominal), gastrointestinal tract (colon, pyloric mass), cerebrum and aqueous humor. The diagnostic performance of MF-ICC in tissue samples is presented in Table 6.

Table 5

Diagnostic performance of the multiplex fluorescent immunocytochemistry (MF-ICC) assay at a threshold of ⩾1 positive cell in fluid (body cavity effusion) samples (n = 51)

Parameter	FIP	Non-FIP
MF-ICC positive	35	3
MF-ICC negative	8	5
Diagnostic accuracy (%)	78.4 (64.7–88.7)
Sensitivity (%)	81.4 (66.6–91.6)
Specificity (%)	62.5 (24.5–94.5)
PPV (%)	92.1 (82.5–96.7)
NPV (%)	38.5 (21.5–88.7)

Data in parentheses are 95% confidence intervals

FIP = feline infectious peritonitis; NPV = negative predictive value; PPV = positive predictive value

Table 6

Diagnostic performance of the multiplex fluorescent immunocytochemistry (MF-ICC) assay at a threshold of ⩾1 positive cell in tissue samples (n = 34)

Parameter	FIP	Non-FIP
MF-ICC positive	14	1
MF-ICC negative	7	12
Diagnostic accuracy (%)	76.5 (58.8–89.3)
Sensitivity (%)	66.7 (43.0–85.4)
Specificity (%)	92.3 (64.0–99.8)
PPV (%)	93.3 (67.5–98.9)
NPV (%)	63.2 (47.9–76.2)

Data in parentheses are 95% confidence intervals

FIP = feline infectious peritonitis; NPV = negative predictive value; PPV = positive predictive value

MF-ICC diagnostic accuracy results on a per-cat level (rather than a per-test level) are presented in Table 7. In this case, a positive result on any one sample (eg, fluid, spleen, liver, lymph node) from a cat with multiple samples tested was considered a positive result overall for FIP. Likewise, the cat was considered negative for FIP only if all MF-ICC samples from that cat were negative.

Table 7

Diagnostic performance of the multiplex fluorescent immunocytochemistry (MF-ICC) assay at a threshold of ⩾1 positive cell reported on a per-cat basis (n = 69 cats)

Parameter	FIP	Non-FIP
MF-ICC positive	40	3
MF-ICC negative	10	16
Diagnostic accuracy (%)	81.1 (69.9–89.6)
Sensitivity (%)	80.0 (66.3–90.0)
Specificity (%)	84.2 (60.4–96.6)
PPV (%)	93.0 (80.9–98.5)
NPV (%)	61.5 (40.6–79.8)

Data in parentheses are 95% confidence intervals. If any MF-ICC result was positive, the cat was considered positive for FIP

FIP = feline infectious peritonitis; NPV = negative predictive value; PPV = positive predictive value

The diagnostic accuracy parameters for common diagnostic modalities for FIP performed in this study cohort are compared directly to each other in Table 8. A serum A:G ratio less than 0.6 was used for this analysis.¹³

Table 8

Comparison of diagnostic accuracy parameters across the major feline infectious peritonitis diagnostic tests used in this study

	Test
	MF-ICC (n = 85)	Serology (n = 28)	RT-PCR (n = 34)	Serum A:G <0.6 (n = 68)
Diagnostic accuracy (%)	77.7 (67.3–85.9)	75.0 (55.1–89.3)	69.4 (51.9–83.7)	76.5 (64.6–85.9)
Sensitivity (%)	76.6 (64.3–86.3)	94.4 (72.7–99.9)	58.3 (36.6–77.9)	79.6 (64.7–90.2)
Specificity (%)	81.0 (58.1–94.6)	40.0 (12.2–73.8)	91.7 (61.5–99.8)	71 (64.7–90.2)
PPV (%)	92.4 (83.4–96.8)	73.9 (62.8–82.6)	93.3 (67.5–99.0)	83.3 (72.5–90.5)
NPV (%)	53.1 (41.0–64.9)	80.0 (34.0–96.9)	52.4 (39.9–64.5)	65.4 (52.2–75.9)

Data in parentheses are 95% confidence intervals. Values in bold denote the highest value for each parameter

A:G = albumin:globulin; MF-ICC = multiplex fluorescent immunocytochemistry; NPV = negative predictive value; PPV = positive predictive value

Finally, using binary logistic regression for serum A:G ratio (as a continuous variable) and MF-ICC results when both were available (n = 56 cats), the model was significant (P = 0.008) with an area under the curve (AUC) of 0.9495.

Discussion

Here we have determined the diagnostic accuracy of MF-ICC for the diagnosis of FIP in ante-mortem samples from clinical patients suspected of this disease and compared it head-to-head with other commonly used ante-mortem diagnostic testing modalities for FIP. Although these results demonstrate only moderate sensitivity (77%) and specificity (81%) at a threshold of one or more positive, dual-expressing mononuclear (macrophage-like) cells, the overall diagnostic accuracy (78%) was the highest across all modalities compared in this study (Table 8), and the PPV was reliable at 92%. The overall diagnostic accuracy was improved by lowering the previously held threshold of two or more positive cells down to visualization of a single positive cell co-expressing FCoV and vimentin. Although specificity was unsurprisingly decreased at this lower threshold, the sensitivity was improved and the PPV remained very similar (93% vs 92%). The PPV was also similar between fluid and tissue samples, but notably, the NPV was much lower for fluid samples. This is due to both the poor specificity in fluid samples and lower prevalence of FIP among cats with tissue samples provided for this study. Importantly, the prevalence of FIP in this study population was quite high, which automatically increases the PPV and decreases the NPV; however, this may be the appropriate population of interest for cats with a strong clinical suspicion of FIP. The specificity of the assay was high (92%) in tissue samples, indicating that a positive MF-ICC result on tissue can be a helpful way to rule in FIP.

The hypothesis of this work was that MF-ICC would increase diagnostic accuracy over published reports of standard chromogenic or single-color fluorescent ICC. In our hands, this assay has improved specificity (81%) over some previous reports (72.4% in an indirect chromogenic assay on 56 cats¹⁴ and 71.4% in a multiplexed MHC-II and FCoV direct immunofluorescence [DIF] assay on 17 cats¹⁵). However, an older study using DIF and a smaller number of cats (n = 32) with effusion reported a specificity of 100% (and sensitivity of 95%).¹⁶ The same investigators, using the same technique, reported a similarly high specificity of 100% in a larger cohort of 110 cats with effusion suspected to have FIP.¹⁷ Surprisingly, the latter two studies utilized a polyclonal antiserum rather than a monoclonal antibody, but still reported very high specificity. More recently, a similar DIF assay using polyclonal antiserum was reported to be 100% specific and 75% sensitive, with 40 FIP cases and 32 controls (data not shown in the cited study).¹⁸ The cause for variability in the rate of false positives among these studies, including ours, is unclear, though the use of a secondary antibody (indirect immunodetection) may be partially responsible for increased false positives as the only studies reporting high (100%) specificity utilized a DIF technique. That said, other factors in methodology, sample collection or patient population are likely at play as at least one study using DIF also had poor specificity.¹⁵ In addition, there have been a variety of antibodies generated against different strains of FCoV, and the results of ICC could depend on the choice of antibody used.^19,20 Further refinement and evaluation of FCoV antigen detection methods in ante-mortem samples from larger cohorts of cats is warranted.

False positives detected by MF-ICC included four patients at a threshold of one or more positive mononuclear cells (lowered to three false positives at a threshold of two or more positive mononuclear cells). Interestingly, three of the four cats were some of the oldest cats in our study (aged 12, 13 and 21.2 years) and were all spayed females. The fourth cat was a young (2-year-old) castrated male. These animals were ultimately diagnosed with the following (non-FIP) diseases: meloxicam overdose leading to acute kidney injury and perforated gastric ulcer; thymoma; liver cysts; and chronic lymphoid leukemia. Of these four false-positive MF-ICC results, three were run on abdominal fluid; the fourth was run on a splenic aspirate. All of them also had positive or equivocally positive granulocytes within the same sample. Taken together, this suggests that a positive MF-ICC result should perhaps be questioned more in older cats (where the suspicion for FIP is generally low), in fluid samples, or if there are concurrently positive neutrophil-like cells within the same sample, although greater numbers of MF-ICC samples must be evaluated to confirm each of these suggestions.

Several MF-ICC samples (from both FIP cases and non-FIP controls) in this study demonstrated positive fluorescence for FCoV within polymorphonuclear cells (granulocytes, presumably mostly neutrophils) (Figure 4). This has been observed in other laboratories performing DIF assays for FIP in cats (Dr Sally Coggins at the University of Sydney, personal communication). Based on what is known about the tropism of FIPV, our criteria for calling a sample positive included only evaluation of mononuclear cell positivity.^21,22 It is unclear whether this positive staining within granulocytes represents aberrant/non-specific (eg, Fc-receptor) binding of the primary or secondary antibody, phagocytosis of infected cell debris or could possibly indicate infection within neutrophils. To our knowledge, infection of neutrophils by FCoV has not been reported. Human neutrophils are known to express aminopeptidase N (APN, CD13).²³ Although this expression has not been similarly confirmed in feline neutrophils, feline APN receptor (fAPN) derived from feline Fcwf cells has been shown to be utilized as the viral receptor by serotype II strains of FCoV.²⁴ In addition, viremia has been reported in clinically healthy (non-FIP) cats as evidenced by positive FCoV RT-PCR in peripheral blood, even though FECV does not replicate efficiently within monocytes/macrophages.²¹ Thus it is conceivable that neutrophils could be truly positive for FCoV antigen in either FIP or control cats that were subclinically infected with the FECV pathotype of FCoV. However, non-specific binding or phagocytosis of cellular debris is considered more likely in our opinion. Regardless, it is important to evaluate only mononuclear cells (as evidenced by nuclear shape observed with DAPI staining) when performing this assay.

We had the unique opportunity to compare multiple diagnostic testing modalities directly in the same cohort of suspected FIP patients in this study, and hypothesized that MF-ICC would offer diagnostic advantages over other commonly available tests. The A:G ratio is frequently employed in FIP diagnostics, with values below 0.6 or 0.8 often suggested as the cutoff for FIP.¹³ In our study, the A:G ratio was significantly lower in FIP cases (mean 0.4 ± 0.2) compared with non-FIP cases (mean 0.7 ± 0.3). However, the diagnostic performance of the A:G ratio as a standalone diagnostic tool is limited, as lower ratios are also observed in other inflammatory or infectious conditions.

Serological assays detect FCoV antibodies, thereby serving as a measure of FCoV exposure. However, serology cannot distinguish between the pathogenic FIPV and the more common FECV pathotypes. This limitation is particularly problematic in high-prevalence environments, where exposure to FCoV is common and often unrelated to FIP disease manifestation. In this study, serology was more sensitive than MF-ICC, but far less specific. These results underscore serology’s vulnerability to false positives due to incidental FCoV exposure.

Finally, RT-PCR is frequently employed for detecting FCoV RNA in blood, effusions or tissue samples, offering an analytically sensitive means of identifying viral presence. Nevertheless, PCR shares serology’s limitation in distinguishing between FIP and FECV.²¹ In this study, RT-PCR (which should be performed on fluids or tissue, not the much less FIP-specific samples of blood or feces) was the most diagnostically specific, but the least sensitive, of the assays we compared. This finding reinforces what is considered in the field to be the main limitation of RT-PCR on fluid or tissue: that multiple samples (and sometimes multiple time points) are necessary to increase its diagnostic sensitivity into an acceptable range.⁵ Of note, we found one false-positive RT-PCR result (out of 12 control cat samples run) in this study, which highlights that although PCR is highly specific, false positives can still occur. Importantly, out of the 11 FIP cats with false-negative RT-PCR results in this study, MF-ICC was positive (⩾1 positive mononuclear cell) in eight (73%) of those cases. Likewise, among six cats with false-negative MF-ICC results, RT-PCR was positive in three (50%) of those cases. Thus, RT-PCR and MF-ICC may truly be considered complementary (rather than redundant) assays for FIP. It is likely that these four tests (MF-ICC, serology, RT-PCR and A:G ratio) would be most accurate when used in combination, as part of an algorithmic approach to FIP diagnosis, as none of them appears sufficient alone. Although we lacked sufficient cases with all four major test results (MF-ICC, serology, RT-PCR and serology) available in this data set to perform statistical combinations of all four tests, we did combine A:G ratio with MF-ICC results statistically (using binary logistic regression) as proof of this principle. Indeed, with these two parameters alone, the model had quite strong performance (AUC = 0.9495, n = 56 cats), suggesting that mathematical combinations of test results could be diagnostically beneficial. It is possible that these test results could even be added to machine-learning algorithms in future iterations of models being developed by our group and others.

The lower than ideal sensitivity of the MF-ICC assay is likely at least partially attributable to the relatively small number of cells that can be visualized manually through fluorescent microscopy. To address this issue, future work by our group is focused on converting the MF-ICC assay described here to a flow cytometry-based assay using similar principles. We anticipate that flow cytometry will be of increased sensitivity by analyzing tens or hundreds of thousands of cells with highly sensitive fluorescence detectors. However, it remains possible that the lower sensitivity observed here is more related to a true lack of FCoV-infected cells within effusion fluids or the tissues sampled by cytology. Future studies should aim to validate a flow cytometry-based assay for FIP and investigate which anatomic locations and sample types are most sensitive for detecting virus-infected cells in this type of assay.

This study evaluated the results of 85 MF-ICC samples from 69 individual cats, with 16 MF-ICC samples from repeat sites. There was redundant MF-ICC sampling on 12 cats (eight with two sites per cat and four with three sites per cat). Of these 12 cats, and using a threshold of one or more positive cells, four had discrepant results among different samples. All four of these cats were ultimately diagnosed with FIP. Interestingly, in three of these four cats, the spleen sample was positive while another tissue sample tested (liver or lymph node) was negative. Although very low numbers of animals are available, this could suggest that spleen is one of the more sensitive tissue samples to use for MF-ICC. In the fourth cat, the discrepant results were from two serial abdominal fluid samples, with the later one testing positive, likely as a result of more progressive disease. If MF-ICC results are evaluated on a per-cat basis, where any one sample being positive is interpreted as a positive result for the whole cat, diagnostic accuracy was found to improve to 81.1% (Table 7). Thus, sampling more than one site improves the ability of MF-ICC to accurately detect FIP in cats.

Finally, it is worth noting that 13/101 (13%) MF-ICC tests analyzed as part of this study were considered non-diagnostic, meaning that they lacked sufficient intact mononuclear cells expressing vimentin for adequate interpretation under the fluorescent microscope. As with any test based on microscopy, this determination is somewhat subjective. Nevertheless, this 13% represents a substantial proportion of submitted tests with unreadable results, which could potentially be exacerbated when this test is applied to samples subjected to time delays, shipment and/or temperature variation. Thus, methods for improving cellularity and cellular integrity of these samples (such as provision of samples in flow cytometry storage media) should be pursued. The cause for proportionately more non-diagnostic results in non-FIP control cats (n = 6/29, 21%) vs FIP-positive cats (7/72, 9.7%) is unknown.

Limitations of this study include relatively low numbers of FIP-negative cases, which is likely a result of the clinical trial recruitment design. For a patient to be included, the attending clinician had to approach the investigators with a suspected case of FIP and initiate enrollment. These clinicians may not have considered enrolling the patient in an ‘FIP diagnostic trial’ unless they were fairly confident about FIP as a differential. As a result, our case population was likely biased toward FIP-positive cats. This high prevalence has somewhat increased the calculated PPV and lowered the NPV. In addition, the incidence of cats with FIP comorbid to another diagnosis is unknown, and these cases were excluded from this trial. There were also particularly low numbers of cats tested by both MF-ICC and serology, or by both MF-ICC and RT-PCR, which somewhat limits the comparison across these tests. It is worth noting that although there are directional trends for metrics presented in Table 8, there the 95% confidence intervals around these point estimates frequently overlap owing to the overall low numbers of cats in this prospective diagnostic trial.

Another limitation is the inability to use the reference method of histopathology with IHC to confirm FIP in the large subset of cases (n = 23) that were treated with unlicensed therapy (GS-441524). As these animals are no longer dying of FIP, necropsy is not possible; therefore, obtaining histopathologic confirmation is not practical in the vast majority of these cases. Further, limited information is available for these cases as far as the dosage, route and timing of antiviral therapy, given the unregulated and at-home nature of antiviral therapy for FIP in the USA at that time. However, as described in the methods section, a robust response to owner-driven therapy within the first 7 days of treatment, in highly suspect cases, as monitored by an experienced clinician, was considered sufficiently supportive of FIP to be included in this data set. This is supported by multiple studies providing evidence that at-home, unlicensed therapy is highly effective for this disease.¹¹ Further, although previous studies have demonstrated variation in the amount of GS-441524 in unlicensed products, all 127 products tested in one study contained the drug GS-441524 when tested via liquid chromatography mass spectrometry.²⁵ Thus, although it remains possible that some of the 23 cats included in that portion of this study may not have had FIP (which could affect the reported diagnostic accuracy data for MF-ICC), we consider that highly unlikely. Indeed, all clinical trial work on FIP moving forward will need to take this new reality of available antiviral therapy (and lack of necropsy confirmation) into account.

Conclusions

MF-ICC with vimentin and FCoV markers is a relatively accurate test compared with other commonly used tests for the diagnosis of FIP in feline patients suspected of having the disease; therefore, a positive result can provide clinicians with strong support for FIP. As such, this assay will be a valuable addition to the diagnostic arsenal against this deadly disease, with due consideration to the cost, invasiveness and turnaround time of the various tests available. We envision MF-ICC being used complementarily alongside other diagnostic tests to build a case for FIP in an individual patient.

Supplemental Material

Table S1

Raw data for the diagnostic results from all 84 cats.

Footnotes

Acknowledgements

The authors would like to thank the veterinary clinicians and technicians of both institutions (the Colorado State University Veterinary Teaching Hospital/Diagnostic Medical Center and the Ohio State University Veterinary Medical Center/Blue Buffalo Veterinary Clinical Trials Office) for their help with collecting and processing samples for this study. We also thank student research assistants Michelle Cornwall, Brittany Allen, Lei Zhang and Kelly Larson for their work on this project. Finally, we sincerely thank all of the patients and owners involved for their thoughtful contribution to this work.

Accepted: 31 March 2026

Supplementary material

The following file is available as supplementary material:

Table S1: Raw data for the diagnostic results from all 84 cats.

Conflict of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This project was funded through the Morris Animal Foundation (grant #D18FE-011).

Ethical approval

The work described in this manuscript involved the use of non-experimental (owned or unowned) animals and procedures that differed from established internationally recognized high standards (‘best practice’) of veterinary clinical care for the individual patient. The study therefore had prior ethical approval from an established (or ad hoc) committee as stated in the manuscript.

Informed consent

Informed consent (verbal or written) was obtained from the owner or legal custodian of all animal(s) described in this work (experimental or non-experimental animals, including cadavers, tissues or samples) for all procedure(s) undertaken (prospective or retrospective studies). No animals or people are identifiable within this publication, and therefore additional informed consent for publication was not required.

ORCID iD

Samantha JM Evans

Petra Cerna

Wendy M Novicoff

References

Gao

Y-Y

Wang

Liang

X-Y

, et al. An updated review of feline coronavirus: mind the two biotypes. Virus Res 2023; 326. DOI: 10.1016/j.virusres.2023.199059.

Pedersen

NC.

An update on feline infectious peritonitis: virology and immunopathogenesis.

Vet J 2014; 201: 123–132.

Delaplace

Huet

Gambino

, et al. Feline coronavirus antivirals: a review. Pathogens 2021; 10 . DOI: 10.3390/PATHOGENS10091150.

Tasker

Addie

Egberink

, et al. Feline infectious peritonitis: European Advisory Board on Cat Diseases guidelines. Viruses 2023; 15. DOI: 10.3390/v15091847.

Felten

Hartmann

Diagnosis of feline infectious peritonitis: a review of the current literature.

Viruses 2019; 11 . DOI: 10.3390/v11111068.

Tasker

Diagnosis of feline infectious peritonitis: update on evidence supporting available tests.

J Feline Med Surg 2018; 20: 228–243.

Barker

Stranieri

Helps

, et al. Limitations of using feline coronavirus spike protein gene mutations to diagnose feline infectious peritonitis. Vet Res 2017; 48: 60.

Taylor

Coggins

Barker

, et al. Retrospective study and outcome of 307 cats with feline infectious peritonitis treated with legally sourced veterinary compounded preparations of remdesivir and GS-441524 (2020–2022). J Feline Med Surg 2023; 25. DOI: 10.1177/1098612X231194460.

Sase

Iwami

Sasaki

, et al. GS-441524 and molnupiravir are similarly effective for the treatment of cats with feline infectious peritonitis. Front Vet Sci 2024; 11. DOI: 10.3389/fvets.2024.1422408.

10.

Howell

Evans

SJM

Cornwall

, et al. Multiplex fluorescent immunocytochemistry for the diagnosis of feline infectious peritonitis: determining optimal storage conditions. Vet Clin Pathol 2020; 49: 640–645.

11.

Jones

Novicoff

Nadeau

, et al. Unlicensed GS-441524-like antiviral therapy can be effective for at-home treatment of feline infectious peritonitis. Animals (Basel) 2021; 11. DOI: 10.3390/ani11082257.

12.

Stockham

Scott

. Introductory concepts. In: Fundamentals of veterinary clinical pathology. 2nd ed. Blackwell Publishing, 2008, pp 38–45.

13.

Jeffery

Deitz

Hostetter

Positive predictive value of albumin: globulin ratio for feline infectious peritonitis in a mid-western referral hospital population.

J Feline Med Surg 2012; 14: 903–905.

14.

Felten

Matiasek

Gruendl

, et al. Investigation into the utility of an immunocytochemical assay in body cavity effusions for diagnosis of feline infectious peritonitis. J Feline Med Surg 2017; 19: 410–418.

15.

Litster

Pogranichniy

Lin

T-L.

Diagnostic utility of a direct immunofluorescence test to detect feline coronavirus antigen in macrophages in effusive feline infectious peritonitis.

Vet J 2013; 198: 362–366.

16.

Parodi

Cammarata

Paltrinieri

, et al. Using direct immunofluorescence to detect coronaviruses in peritoneal and pleural effusions. J Small Anim Pract 1993; 34: 609–613.

17.

Paltrinieri

Parodi

Cammarata

In vivo diagnosis of feline infectious peritonitis by comparison of protein content, cytology, and direct immunofluorescence test on peritoneal and pleural effusions.

J Vet Diagn Invest 1999; 11: 358–361.

18.

Worthing

Wigney

Dhand

, et al. Risk factors for feline infectious peritonitis in Australian cats. J Feline Med Surg 2012; 14: 405–412.

19.

Hohdatsu

Okada

Koyama

Characterization of monoclonal antibodies against feline infectious peritonitis virus type II and antigenic relationship between feline, porcine, and canine coronaviruses.

Arch Virol 1991; 117: 85–95.

20.

Corapi

Olsen

Scott

FW.

Monoclonal antibody analysis of neutralization and antibody-dependent enhancement of feline infectious peritonitis virus.

J Virol 1992; 66: 6695–6705.

21.

Jaimes

Whittaker

GR.

Feline coronavirus: insights into viral pathogenesis based on the spike protein structure and function.

Virology 2018; 517: 108–121.

22.

Balakumar

Wanakumjorn

Kimura

, et al. Beyond macrophages: FIPV tropism includes T and B lymphocytes. Vet Microbiol 2026; 313. DOI: 10.1016/j.vetmic.2025.110864.

23.

Pérez-Figueroa

Álvarez-Carrasco

Ortega

Crosslinking of membrane CD13 in human neutrophils mediates phagocytosis and production of reactive oxygen species, neutrophil extracellular traps and proinflammatory cytokines.

Front Immunol 2022; 13. DOI: 10.3389/fimmu.2022.994496.

24.

Tresnan

Holmes

KV.

Feline aminopeptidase N is a receptor for all group I coronaviruses.

Adv Exp Med Biol 1998; 440: 69–75.

25.

Kent

Guan

Jacque

, et al. Unlicensed antiviral products used for the at-home treatment of feline infectious peritonitis contain GS-441524 at significantly different amounts than advertised. J Am Vet Med Assoc 2024; 262: 489–497.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.16 MB

0.00 MB