Alpha-1 acid glycoprotein reduction differentiates recovery from remission in feline infectious peritonitis treatment


 Feline infectious peritonitis (FIP) is a systemic immune-mediated inflammatory perivasculitis which occurs in a minority of cats infected with feline coronavirus (FCoV). A variety of therapies have been employed to treat this condition, which previously was usually fatal, but no parameters for differentiating FIP recovery from remission have been defined to enable clinicians to decide when it is safe to discontinue treatment. This observational study shows that a consistent reduction of the acute phase protein alpha-1 acid glycoprotein (AGP) to within normal limits (WNL i.e. 500μg/ml or below), as opposed to duration of survival, distinguishes recovery from remission. AGP consistently reduced to WNL in 26 recovered cats but remained elevated in 16 cats in remission, dipping to normal once in two of the latter. Anaemia was present in 58% (23/40) of the cats, and resolved more quickly than AGP in six recovered cats. Lymphopenia was observed in 43% (16/37) of the cats and reversed in nine recovered cats, but none of the remission group. Hyperglobulinaemia was slower than AGP to return to WNL in the recovered cats. FCoV antibody titre was high in all 42 cats at the outset and decreased significantly in 7 recovered cats, but too slowly to be a useful parameter to determine discontinuation of anti-viral treatments. Conclusion: a sustained return to normal levels of AGP was the most rapid and consistent indicator for differentiating recovery from remission following treatment for FIP. This study provides a useful model for differentiating acute from chronic coronavirus disease using acute phase protein monitoring.


Introduction
Feline coronavirus (FCoV) is a positive strand RNA virus belonging to the order Nidovirales, genus Coronavirus, and family Coronaviridae, subfamily alphacoronavirus. The FCoV species is further divided into two types: I and II, the rst type being wholly feline, type II FCoV arising from recombination events with canine coronavirus 1, 2 .
Feline infectious peritonitis (FIP) is an immune-mediated perivascular pyogranulomatous 3 . disease which affects 5-10% of cats infected with FCoV 4 .. FIP can present acutely, with effusions in one or more body cavities, or as a disease of chronic in ammation, with cachexia and variable organ damage depending on the sites of pyogranuloma formation.
The initial target for FCoV is the epithelial cells of the small intestine; from there macrophages engulf the virus and transport it to the mesenteric lymph nodes (MLN) and a brief systemic phase of infection follows when the target cell is the monocyte.
TNF-α is a major contributor to the in ammatory response and pathogenesis of FIP. FCoV-infected cells release a substance which causes apoptosis in nearby lymphocytes 8 : the mystery substance is probably TNF-α 5 : around 50% of cats with FIP are lymphopenic 9,10,11 . IL-6 stimulates hepatocytes to release acute phase proteins 7 such as alpha-1 acid glycoprotein (AGP) and serum amyloid A. AGP was rst reported to be elevated in FIP cases in 1997 12 . Raised AGP has been shown to be superior to histopathology of biopsy in differentiating FIP from similarly-presenting cases 13 . Monitoring AGP is an accurate predictor of survival in humans with sepsis 14 .
Many treatments, ranging from prednisolone, through feline interferon omega (rFeIFN omega) 15 , polyprenyl immunostimulant 16 , meloxicam 17 , to, most recently, speci c antivirals 18, 19 have been used. Those treatments involved in the present study are detailed in Tables 1 and 2.    Veterinary surgeons seek a criterion by which to be assured that the cat has recovered in order to know when to cease treatment. Markers for poor prognosis in FIP have been well de ned 10 , yet no guidelines have been established to de ne a recovery from FIP, probably because, until recently, it was widely assumed that recovery from FIP could not occur.
No survival time point after which a cat can be said to have recovered from FIP has been established. In their report of their seminal study of successful treatment of FIP with an antiviral drug, Pedersen et al wrote: "It raises the question of how long remission must be sustained to declare the disease cured, rather than in a sustained remission." 18 The purpose of this paper was to attempt to answer that question; but rather than duration of survival, we found that a return of elevated AGP to normal levels is a rapid and de nitive marker for recovery from FIP.

Materials And Methods
Selection of cases. This study retrospectively evaluated the medical records of 42 cats diagnosed with FIP between January 2004 and March 2021. Cats were selected for this study if sequential laboratory test results were available, including AGP measurements, a reasonable certainty that the FIP diagnosis was correct, and whose survival outcomes were known.
FIP diagnosis. The data presented here were collected over almost two decades. As new tests became available they were applied to the cases; consequently, details of FIP diagnosis of individual cats were not uniform. While demonstration of FCoV in biopsy or post mortem material by immunohistochemistry is deemed the gold standard for FIP diagnosis, for practical purposes it is generally accepted that a positive FCoV RNA 20 As shown in Tables 1 and 2, diagnosis of FIP was by histopathology (n= 13), of which two cases had positive immunohistochemistry (IHC). However, histopathology of biopsies was inconclusive in ve cases and was reported nonspeci cally as "pyogranulomatous in ammation." FIP diagnosis was by detection of FCoV RNA as demonstrated by RT-PCR in effusions (n = 10), or mesenteric lymph node (MLN) FNA 28 (n = 10); messenger RNA (mRNA) 26,27 or 3'UTR 20,21 in pbmc (n = 4). One of two cats in which the mutation tests 29,30 were performed was positive for the M1058L mutation, and neither cat was positive for the S1060A mutation.
Some cats ful lled multiple diagnostic criteria for FIP: seven cats had both positive FCoV RT-PCR tests and histopathology. All 42 cats were positive for at least some components of an FIP pro le consisting of typical clinical signs, FCoV antibody titre, raised AGP, hyperglobulinaemia, anaemia, lymphopenia (Tables 3 and 4). In 11 cats the FIP pro le was the sole evidence for its diagnosis, although a positive response to an anti-coronavirus drug was thought to corroborate the FIP diagnosis. AGP was raised in all cats except two where a pre-treatment test was not done. It has been documented that raised AGP was more sensitive than histopathology for diagnosing FIP in challenging cases 13 .  recommended stopping (or not commencing) the use of corticosteroids and to use meloxicam instead, (after a suitable wash out period and provided blood pressure and kidney function were normal).
Treatment protocols were variable, and are described to the best of our ability for each cat in Tables 1 and 2. They most commonly included recombinant feline interferon omega (rFeIFN-ω, Virbac, France) by subcutaneous injection (for effusive FIP) or diluted and administered per os daily (for non-effusive FIP) (n = 20 recovered and 13 remission cats). Oral rFeIFN-ω was also used as follow up to other therapies. Twelve cats were treated with a 5% oral adenosine nucleoside analogue (Mutian Xraphconn, Nantong Mutian Biotechnology Co. Ltd. China) 33 .
Supportive treatment included meloxicam per os daily; vitamin B12 (cobalamin) either as weekly injections or daily pills (Cobalaplex, Protexin Veterinary, UK); and Pro-kolin enterogenic probiotics (Protexin Veterinary, UK). Clients whose cats were treated with adenosine nucleoside analogue drugs were recommended to supplement with S-adenosyl-L-methionine (SAMe, Denamarin, Nutramax Laboratories, Inc, USA) for liver support.
Outcomes: death, recovery or remission. There are three possible outcomes following the diagnosis of FIP and treatment: death due to FIP-related causes; total recovery; or remission, de ned as an intermediate stage between cure and death, carrying the spectre of relapse. This latter state is a source of considerable stress for cat guardians, so that it would bring considerable reassurance to people to know that their cats are cured of FIP, rather than being in remission. Death due to a non-FIP reason occurred in two cases (Edward and Bea) who were lost to follow-up due to a road accident and cancer.
Case 1 (Basil 1) set the benchmark for de ning recovery from FIP against which other cases were compared. This cat was treated and was followed up for eleven years without relapsing. The criteria for declaring a cat recovered from FIP included the following: The cat returned to clinical normality, speci cally resolution of the clinical signs of FIP.
Globulin levels reduced to normal (≤ 45 g/l) or at least signi cantly reduced (often by over 15g/l as shown in Table   3).
Resolution of lymphopenia, where present, (normal de ned as greater than 1.5 x 10 9 /l ).
Haematocrit level increased to normal (de ned as 30% or over), with reversal of non-regenerative anaemia, where present.
At least three-fold reduction in FCoV antibody titre.
Clinical scores are shown in Tables 3 and 4. Since the purpose of this study was to assess whether or not AGP reduction was an indicator of clinical recovery, AGP was not included in the cat's score for de nition of a particular case as recovered or only in remission.
Cats were deemed to be in remission if only some of the criteria listed above were met: for example, cat 15 (Ragamu n) survived over 5 years but she was never clinically well during that time, requiring continuous treatment, and unfortunately she was lost to follow up.
AGP. AGP was measured by enzyme-linked immunosorbent assay (ELISA) (Avacta Animal Health, Wetherby, Yorkshire, UK) at the University of Glasgow Veterinary Diagnostic Services (VDS) in all cases except seven: the samples of Basil 1; Amy; Brook; Bugsy; Daisy; Roxanne and Pip were measured by radial immunodiffusion as previously described 12 .
The maximum value given on the y axis of Figures 1 and 2 was set at 5000 µg/ml in order to maintain legibility of the graphs. The minimum ELISA cut-off was 300 µg/ml and results reported as below 300 were plotted as 300.
Statistics. Due to small sample size, the Fisher's exact test was used with signi cance set at <0.05.
Ethical approval. All applicable international, national and institutional guidelines for the care of animals were followed.
All procedures performed in studies involving animals were in accordance with the ethical standards of the governing bodies of the veterinary surgeons of the countries in which the studies were conducted.
The reporting in this manuscript follows the recommendations in the ARRIVE guidelines.
Informed consent was obtained from the cat guardians for publication of their pet's data.

Results
The laboratory records of effusive (n = 17) and non-effusive (n= 25) FIP cases were reviewed.
Recovery, remission, and death. As shown in Tables 1 and 2, 26 cats recovered from FIP, with follow up periods of up to 11 years. Of the recovered cases, 11/26 (42%) had effusive FIP and 14 /26 (54%) were non-effusive. One cat (Nelson) with non-effusive FIP developed an effusion following biopsy; another cat (Boris) had non-effusive FIP, initially thought to be effusive, but later established to be a cardiogenic effusion (FCoV RT-PCR on his effusion was negative). Two of the noneffusive FIP cases were the colonic presentation.
One cat (Kitten 2) in the recovered group experienced a relapse: details are given below.
Sixteen cats experienced remissions of 1.5 to over 60 months. Median remission was 4.5 months (not including the outlier of Ragamu n who was lost to follow up after 5 years because her result would skew the median). Twelve cats died or were euthanased and 4 were lost to follow up.
AGP levels. AGP levels of recovered cats and those who experienced remission are shown in Figures 1 and 2 respectively. AGP levels were elevated (i.e., above 500μg/ml) in all FIP cases except in the cases of Wish and Mike, where the rst AGP test was taken after treatment had started, so the pre-treatment AGP results were unknown.
The time for AGP levels to return to normal in recovered cats varied from a minimum of less than 13, 20 and 22 days (Wish, Lyra and Basil 1) to over 16 months from onset of treatment, probably depending on the treatment being used ( Table 1). The AGP levels of one recovered cat (Chynah) did not return to normal and with the bene t of hindsight, this cat should perhaps be in the remission group although her veterinary surgeon reported that she was clinically recovered: unfortunately, she was lost to follow up.
Two of the cats (Holly and Daisy) who did not recover had AGP levels which reduced to under 500μg/ml on one occasion each and AGP levels reduced to under 1000μg/ml in two more cats. Therefore, it is important that two consecutive normal AGP results at least one week apart be obtained to ensure that recovery from FIP has occurred.
Of particular interest was the AGP level of Kitten 2 ( Figure 3) who was diagnosed with effusive FIP in October 2019. Her guardian stopped the oral adenosine nucleoside analogue (Mutian X) treatment on January 5 th 2020 following an AGP result of 709μg/ml. This young cat presented with an FIP relapse in February 2020 manifesting as hyperesthesia of the tail, which progressed to ataxia, then seizures. A second course was administered, this time at a double dose (one Mutian X 200 per kg) which is a level that enables su cient anti-viral drug to cross the blood brain barrier (Tony Xue, personal communication). The cat improved within 24 hours and the second course of Mutian X was continued until AGP reached below 500μg/ml: she is alive and well 17 months later. After this event, FIP cat guardians were recommended to give a double dose of oral (not injectable) Mutian for 7-10 days to clear virus from the brain to prevent neurological relapse.
Haematocrit. A lower cut-off of 30% was our de nition of normal below which a cat was said to be anaemic. This is the value determined by University of Glasgow Veterinary Diagnostic Services Laboratory.
Haematocrit was available for 25 of 26 recovered cats and 13 of 16 cats who went into remission. (Tables 3 and 4). One recovered cat (Chynah) and 3 remission cats (Yrael, Claude and Al e) had no haematology results and in ve recovered cats (Dante, Molly, Nelson, Edward, Kitten 2) records were not available until after treatment began -too late to determine whether or not they had been anaemic because by that time they were not anaemic. In two cats Hct was only marginally below normal (Tabitha 29.4% and Chester: 28.9%) on one occasion only, therefore they were not seriously anaemic and were discounted from analyses. Basil 2's anaemia was complicated by concurrent haemoplasmosis infection (Figure 4). This left usable haematocrit records for 15 recovered cats and 13 cats who went into remission: 6 of the 15 recovered cats had effusive FIP and 9 had non-effusive FIP; 2 remission cats had effusive FIP and 11 remission cats had non-effusive FIP. Recovered non-effusive 5 5 10 Remission effusive (but 1 became non-effusive) 2 0 2 Remission non-effusive (but one became effusive) 10 1 11 Eight of 9 (89%) cats with effusive FIP and 15 of 21 (71%) cats with non-effusive FIP were anaemic (Tables 3, 4 and 5). Whether FIP was effusive or non-effusive made no statistical difference to the likelihood of the cat being anaemic (p = 0.70).
Eleven of 17 (65%) cured and 12 of 13 (92%) cats in remission were anaemic: the presence of anaemia did not affect the cat's chances of recovery (p = 0.10).
Two remission cats became anaemic after FIP diagnosis, and one cat's anaemia resolved. Anaemia resolved in all of the cats who recovered, but Basil 2 required a one-month course of doxycycline for concurrent haemotropic mycoplasma infection for his anaemia to resolve. The haematocrit of Basil 2 is shown in Figure 4: it illustrates the typical wave pattern due to cyclical parasitemia with a periodicity of 7-10 days, resolving when doxycycline was begun.
While the average time for resolution of purely FIP related anaemia was 30.4 days in 10 cats (range 12 to 57 days: Table   3), from initiation of treatment, the average time for AGP to return to WNL was 45 days (range 20-117 days, not counting the two cats for whom there was no pre-treatment AGP). The outlier cat, Mike, was not counted in this analysis because there was no initial AGP for comparison: his anaemia resolved 154 days after treatment began.
Lymphopenia. A lymphocyte count of 1.5 x 10 9 /l was considered the lowest level at which lymphocyte count could be considered normal: this is the cut-off set by the VDS laboratory. Sixteen of 37 (43%) cats with FIP were lymphopenic on the rst available lymphocyte count closest to FIP diagnosis. The prevalence of lymphopenia was 38.5% (5/13) in effusive cases and 45.8% (11/24) in non-effusive cases.
Sequential lymphocyte counts were available for 25 of 26 recovered cats and 13 of 16 remission cats, but the rst count was two months after treatment began for one recovered cat (Edward). Although the median lymphocyte count was slightly higher for recovered cats (1.833 vs 1.366 x 10 9 /l), the difference was not signi cantly different between the recovered or remission groups (p=0.47).
It appeared that fewer recovered cats (9/24: 37%) than remission cats (7/13: 54%) were lymphopenic, but the difference was not statistically different (p = 0.5). Four recovered cats had very low lymphocyte counts but within the normal range (1.5 to 7.0 x 10 9 /l): one cat (Buddie) was lymphopenic on 2 of 7 samples and another cat (Molly) had low lymphocyte counts (below 2 x 10 9 /l, but above 1.5 x 10 9 /l which is considered the lower cut-off).
Lymphopenia resolved in all recovered cats. In the remission group, three cats became lymphopenic and the status of the other cats remained the same (Table 4).
Hyperglobulinaemia. Globulin levels were available for all the 26 recovered cats and for 13/16 remission cats, but there was no pre-treatment sample for one of the recovered cats and by the time he was tested his globulin levels were normal. Twenty-three of the remaining 25 recovered cats were hyperglobulinemic: globulin levels reduced in all 23 cats, to WNL in 16 cats, but were still elevated in 7 cats.
Eleven of 13 remission cats were hyperglobulinemic and globulin levels reduced in 4 cats.
AGP was a more accurate prognostic indicator than globulin reduction because globulins were slower to reduce than AGP level in recovered cats (7 of 23 cats still had elevated globulin levels once recovered) and they reduced in 4 of 12 remission cases.
FCoV antibody titre. As shown in Tables 3 and 4, all of the cats had very high FCoV antibody titres, being on or above the upper cut off point for the laboratory to which samples had been sent. Two exceptions were Nelson, who recovered, and the remission cat Pharaoh, whose FCoV antibody titre was only moderately high at 640 at time of diagnosis, but subsequently became very high. FCoV antibody titres remained high in almost all recovered cats for a very long period, even years (Table 3): the earliest that a signi cant reduction in FCoV antibody titre was seen was at 4 months postdiagnosis, and one cat became seronegative at 8 months.
FCoV antibody titres in the remission group did not decrease, with the exception of Rowley, whose titre decreased from >1280 to 640, likely due to immunosuppressive amounts of prednisolone with which he was being treated.

Discussion
Recovery implies that the disease is nished once and for all, whereas remission is de ned as the reduction or diminution of clinical signs, with the implication that the disease could reappear. The state of remission in our series usually culminated in death, mainly by euthanasia due to a prolonged cachexic state, or sometimes acutely, presenting as acute haemolytic anaemia, collapse, or neurological signs. While parameters and a score system for predicting imminent death have been well documented 10 , no such system has been previously de ned to differentiate recovery from remission. We present the rst documentation of sequential AGP testing of cats being treated for FIP and present evidence that a consistent reduction in AGP to normal levels is the most useful marker for differentiating recovery from remission, and is a clear indicator that it is safe to stop administering adenosine nucleoside analogue anti-viral drugs.
Raised AGP was shown to be 100% sensitive in FIP diagnosis 12,13 but is not speci c: it rises in both transient FCoV infection 34 and in other infections (e.g. bacterial peritonitis or pleurisy; Mycoplasma haemofelis infection 35 ). However, the sensitivity of AGP measurement in cats with FIP only in the brain has not been established: Rissi et al, 2018 36 reported that ve of 22 cats with neurological FIP had no FIP lesions in other organs, and it is possible that in those cats AGP measurement would not have been raised. Raised AGP was more sensitive than histopathology for diagnosing FIP in challenging cases 13 and this was also true in our series, where histopathology-especially of the MLN-was frequently non-speci c and reported as simply pyogranulomatous in ammation.
AGP was the most useful parameter for assessing recovery from FIP because it was raised in all our cases, whereas anaemia, lymphopenia and hyperglobulinaemia were not consistently found. Anaemia resolved more quickly than did AGP in ve of six of the nine anaemic cats in which the time to return to normal was not the same for both AGP and Hct (in three cats the identical intervals presumably simply re ected the intervals between blood samplings). Consequently, where AGP measurement is not possible and if the cat is anaemic purely due to FIP (e.g. uncomplicated by concurrent haemoplasmosis), then return of Hct to over 30% appears to be a good indicator of recovery, although our numbers were small. However, the anaemia of one of the cats in the remission group also resolved, so resolution of anaemia alone is not a guarantee of recovery.
The exact interval from FIP diagnosis to normal AGP could not be accurately determined because it depended on the frequency of the cat being blood tested, which was at the discretion of the cat's guardian: we were only able to determine that it was less than however many days from the start of treatment to the rst AGP test WNL. However, the intervals appeared to be noticeably shorter after 2019, when adenosine nucleoside analogue drugs were introduced 19 . The earliest that AGP returned to normal was around two to three weeks from the start of treatment.
The prevalence of lymphopenia in this study (43%) was similar to that of other studies: 49.5% of 184 cats with FIP 9 ; 50% of 106 cats 11. and 64% of 45 cats 10 . In the study of Riemer et al 2016 9 , lymphopenia was observed signi cantly more often in 139 cats with effusion and was documented in only 26.8% of 41 cats without effusion. Our results differed: 38.5% of effusive cases were lymphopenic compared with 48.8% of non-effusive cases, but our cohort of cats was much smaller than that of Riemer et al, 2016 9 .
In the study of Tsai et al (2011) 10 the prevalence of lymphopenia was 64% at initial presentation and increased to 91.7% zero to three days before death. In our study, lymphopenia reversed in the lymphopenic cats who recovered, and two cats in the remission group became lymphopenic. Accordingly, the development of lymphopenia in a cat with FIP who had not previously been lymphopenic is a poor prognostic sign. Our study could not assess the utility of documenting reversal of lymphopenia as a marker for recovery from FIP, because unfortunately during most of this study prednisolone was given to treat FIP, which would have interfered with (i.e. decreased or suppressed) the lymphocyte count. Further studies will be required to assess the utility of lymphopenia reversal as an indicator of FIP recovery, although given that around half of cats with FIP are not lymphopenic, it will clearly be less useful than AGP, which is consistently elevated in cats with FIP.
The de nitive marker for recovery from FCoV infection is the reduction of FCoV antibody titre to undetectable levels, signifying that viral antigen no longer remains in the body to stimulate an immune response; but this takes many months to achieve, consequently it is not a useful marker to determine when to stop treatment. In FIP recovered cats, signi cant reduction of FCoV antibody titre takes so long to achieve (usually over one year) that it would be inadvisable to sustain anti-viral treatment long term due to the risk of toxic adverse effects to the kidneys and liver. In contrast, antibody levels in FCoV infected cats without FIP decline within a few months of virus being eliminated from the intestine and faeces (data not shown).
We observed two relapsed cases who presented with a painful tail: Kitten 2, described in this paper, and another cat for whom we had no AGP results so was not documented here. The occurrence of idiopathic painful tail syndrome has been reported previously 37 and André et al, 2019 38 , described a neurological FIP case which began as plegia of the tail. FIP is the major cause of hydrocephalus in young cats 36,39,40 . Cerebrospinal uid (CSF) drains from the thecal sac at the level of the second sacral vertebra and it is likely that build up of CSF in this area pressed on the nerves of the cauda equina, causing an initial presentation of sensitivity and pain in the tail. As CSF built up in the ventricular system of the central nervous system the two cats progressed to show hind limb ataxia, then seizures. Fortunately, prompt administration of antiviral pills to Kitten 2 saved her life; the other cat was treated with an injectable nucleoside analogue (Mutian II), but died. Two other cases (outwith this study) treated with injectable nucleoside analogues have presented with neurological relapses, but recovered after a course of double dose pills: it seems counterintuitive for an oral formulation to cross the blood brain barrier more effectively than the injectable formulations, but that is our experience. Following the Kitten 2 relapse, clients were recommended to give a 7-10-day double dose of Mutian X to clear the brain of virus.
It might be wondered why there was only one relapse amongst the 17 cats (6%) which were treated with nucleoside analogue drugs when in the rst study where nucleoside analogues were used there were eight relapses in 26 cats (31%) 19 .
We believe that one of the most important reasons was that 12 cats were treated with an oral adenosine nucleoside analogue, which accesses the site of viral replication in the intestine and stopped virus shedding in faeces in all cases. Other factors which contributed to a successful outcome included being able to differentiate accurately between recovery and remission using AGP levels, therefore knowing when it was safe to discontinue the antiviral drug; secondly anti-viral treatment was followed up with long term oral feline interferon which has anti-viral and immunomodulatory activity: and thirdly, in-contact cats were tested for FCoV shedding and if positive, they were treated to stop virus shedding 33 thus reinfection was prevented (presumably re-infection could appear as a relapse). The approach of using anti-virals in subclinically infected cats has a parallel in SARS-CoV2 infection where two doses of prophylactic ivermectin reduced the risk of COVID-19 infection by 83% in the following month in health care workers in India 43 .
It should always be considered that the appearance of relapse or failure to respond to treatment may, in reality, be due to secondary conditions, for example the anaemia of Basil 2 was due to concurrent haemoplasmosis but initially appeared to be a lack of response to FIP treatment.
It is likely that immunosuppression due to FIP-induced lymphopenia leads to secondary infections in FIP. Thus, in addition to infectious anaemia, we have observed toxoplasmosis and clinical signs attributable to recrudescent latent feline herpesvirus infection, such as epiphora and sneezing. One relatively frequent curious condition that occurred was trichobezoar which is not explicable by immunosuppression. In most cases this was easily dealt with using a psylliumcontaining dry cat food (Royal Canin Aging 12+) but in one cat, who was not included in this series due to lack of AGP results, surgery was required to remove a large furball.
A failing of our study was the inability to compare weight gain or loss in the two groups. In a previous study, weight gain was an important measurement of treatment success 19 . Unfortunately weight records were only available for the most recent recovered cases and none in the remission group.
This study might be criticised for a lack of histopathological con rmation of FIP in many of the cases. Since the cats were eld cases, it would not have been ethical to ask for biopsies to be performed for the sake of this study, and most guardians did not elect to have histopathological con rmation of diagnosis in those cats who died. Many of the cats had positive FCoV RT-PCR tests of effusion or MLN FNA and such evidence has been accepted as diagnostic of FIP in other therapeutic studies 18, 19 . Demonstration of replicating FCoV by messenger RNA RT-PCR 26 in peripheral blood mononuclear cells (pbmc) was considered diagnostic of FIP, but it was borne in mind that FCoV mRNA had been detected in 5% of cats without FIP 26, 27 and during this study, we found that the primers cross-reacted with human DNA, giving false positive results on some samples (therefore positive pbmc mRNA results were con rmed by a 3' UTR RT-PCR).
There remained some cats who were diagnosed only by circumstantial evidence, but in all cases elevated FCoV antibody titres and AGP levels were documented, and response to anti-coronavirus drugs was further corroboration of a correct FIP diagnosis.

Conclusion
As new FIP treatments become available we have shown that AGP measurement will be a useful parameter to assess their e cacy. Increasing AGP can indicate that a cat's condition is deteriorating but a return to 500 μg/l or less sustained for at least one week indicated recovery from FIP and differentiated recovered cats from those in remission. It is inadvisable to reduce the dosage or frequency of treatment, even if clinical signs appear to have resolved, until AGP levels are once again normal.
Declarations Figure 1 Sequential AGP results of 25 cats who recovered from FIP This graph shows the time in months for AGP levels to return to normal, with Day 0 being the rst day of a speci c FIP treatment (where known: if unknown then Day 0 was the rst AGP test). The normal cut-off for AGP is 500 μg/ml. The y axis was cut-off at 5000. No cat had sequential AGP tests prior to treatment, which is why it appears as if AGP was decreasing prior to treatment -in reality, it would have been increasing. AGP of the 26 th recovered cat is shown in Figure 3.

Figure 2
Sequential AGP results of 16 cats who experienced remission from FIP This graph shows the AGP levels of cats in remission and illustrates that cats who did not fully recover from FIP retained high AGP levels: in Ragamu n's case, very high levels. The median time for survival was 4.5 months. The AGP levels in these cats was often suppressed by prednisolone treatment. These cats are the feline equivalent of humans with long COVID.