Cobalamin, folate and inorganic phosphate abnormalities in ill cats

Abstract

Hypocobalaminaemia in cats has previously been identified, but the incidence reported has varied, and the frequency of folate deficiency is unknown. The aims of this study were to evaluate the incidence of low cobalamin and folate levels in a population of cats that were suffering predominantly from diseases of the alimentary tract (including the liver and pancreas) and to ascertain whether severity of disease (as assessed by bodyweight and body condition score (BCS)) related to degree of deficiency. The study population comprised 103 cats, of which 16.5% had low cobalamin levels and 38.8% had low folate levels. A serendipitous finding was inorganic phosphate levels below the reference range in 48% of the cases. Significant associations were found between subnormal cobalamin levels and median BCS (P=0.049); combined low folate and low cobalamin and bodyweight (P=0.002), BCS (P=0.024) and inorganic phosphate levels (P=0.003). The finding of low levels of folate and cobalamin in clinical cases suggests that supplementation may be indicated more frequently than is currently recognised.

Cobalamin (vitamin B₁₂) is a water soluble vitamin that is necessary for the function of a number of enzymes (Allen et al 1993). Its absorption requires separation of the cobalamin from dietary protein in the stomach, and subsequent binding to R protein for transportation to the duodenum. When the pH rises in the duodenum, cobalamin is released from R protein by the action of trypsin and chymotrypsin (Suchodolski and Steiner 2003), and binds to intrinsic factor (IF) which, in the cat, is produced exclusively by the pancreas (Fyfe 1993). In the ileum, the cobalamin-IF complexes bind to cubulin receptors on the enterocytes. This results in uptake and then release of cobalamin into the circulation, where it is transported to the cells bound to transport proteins such as transcobalamins (Suchodolski and Steiner 2003). Although enterohepatic recirculation of cobalamin has been shown to occur in dogs (Willigan et al 1958) and humans (Herbert and Colman 1988), it has not been documented in cats. Cats lack transcobalamin 1 (Linnell et al 1979) and can, therefore, become depleted of cobalamin within a month in the presence of severe malabsorption (Hall and Simpson 2000). Cobalamin is essential for two important enzymes involved in methionine metabolism, namely methylmalonyl CoA mutase and methionine synthase. Deficiency has been shown to affect methionine metabolism, leading to increased levels of methylmalonic acid and methionine, and decreased levels of cystathionine and cysteine (Ruaux et al 2001). Cobalamin deficiency can also lead to a ‘functional’ deficiency of folate, as the enzyme converting methylfolate to tetrahydrofolate, the form required for DNA synthesis, requires cobalamin.

Folate is also a water soluble vitamin, which is normally present in the diet in a polyglutamate form. Its absorption requires removal of all but one of the glutamate residues by folate conjugase, a brush border enzyme in the jejunum. The folate monoglutamate produced is subsequently absorbed in the proximal small intestine, a process that is facilitated by folate carriers (Suchodolski and Steiner 2003). Folate coenzymes are involved in metabolic reactions involving the transfer of a one-carbon unit, hence deficiency can result in abnormalities in purine and pyrimidine synthesis (Herbert and Colman 1988). In addition, folate deficiency has been shown to result in an increase in urinary formaminoglutamic acid (FIGLU), following injection of l-histidine, as the lack of tetrahydrofolate prevents conversion of FIGLU to glutamate (Thenen and Rasmussen 1978, Yu and Morris 1998). Folate and cobalamin are both required for synthesis of DNA. Deficiencies of either of these vitamins can affect haematopoiesis, resulting in macrocytosis and hypersegmentation of neutrophils.

Measurement of cobalamin and folate levels formerly found favour in the assessment of small intestinal bacterial overgrowth (SIBO) in dogs (Batt and Morgan 1982), although this was subsequently shown to be of minimal diagnostic benefit (Walkley and Neiger 2000, German et al 2003). The measurement of cobalamin and folate to diagnose SIBO in cats has not generally been performed, as SIBO appears to be rare in this species (Johnston et al 2001), and idiopathic SIBO has never been documented. Instead, interest has been shown in evaluating cobalamin levels in certain diseases; for example, cats with exocrine pancreatic insufficiency may suffer from hypocobalaminaemia, as a result of inadequate production of IF, pancreatic proteases and bicarbonate affecting cobalamin absorption (Hall and Simpson 2000). In addition, subnormal concentrations of cobalamin in cats with gastrointestinal (GI) disease have been demonstrated previously (Simpson et al 2001, Ruaux et al 2005), although the incidence of this problem in the UK was recently questioned (Ibarrola et al 2005).

To the authors' knowledge, an assessment of folate deficiency in the cat has not been previously performed, although reduction in levels of this vitamin might be expected with GI disease affecting its absorption. The aims of this study were, therefore, to evaluate the incidence of cobalamin and folate deficiencies in a population of cats that were suffering predominantly from diseases of the alimentary tract (including liver and pancreas) and to ascertain whether severity of disease (as assessed by bodyweight and body condition score (BCS)) related to degree of deficiency.

Materials and methods

Study population

The database at the University of Edinburgh Hospital for Small Animals (UEHFSA) was searched for all cats that had assessment of folate and cobalamin levels performed over a 7-year period from 1 June 1999 to 31 May 2006. In addition, cases were required to have had routine haematology and serum biochemistry performed within 48 h of sampling for folate and cobalamin levels. Blood samples were obtained by jugular venepuncture, following an overnight fast. Previous history from the referring veterinary surgeons was reviewed for prior administration of folate or cobalamin. When cases had been sampled on more than one occasion (n=11), only the data from the first visit were included in the analysis. Case records were searched for age, sex, breed, BCS, weight, and duration of illness. BCS was assessed on a 9-point scale as previously described (Laflamme 1997). Duration of illness was classified as up to 1 week, 1 week–1 month, 1–6 months, 6–12 months, or >12 months. Folate and cobalamin levels were recorded, as were routine haematology and serum biochemistry results. Eventual diagnosis was recorded, and classified as GI, hepatic, pancreatic or other. The diagnosis of pancreatitis was based on clinical signs supported by elevated feline pancreatic lipase and ultrasonographic appearance of the pancreas. The UEHFSA database was also searched to establish the breed prevalence within the general hospital feline population.

Cobalamin and folate measurement

Serum collected from the study population was submitted to one of four laboratories (see Table 1). The majority of samples (88%) were tested at one laboratory. The measurement of cobalamin in cats by the chemiluminescent immunoassay technique has been validated (Ruaux et al 2001), and the measurement of cobalamin and folate by the dual isotope competitive radioimmunoassay has been validated for dogs (Batt et al 1991). Validation of the measurement of folate by the chemiluminescent method and folate and cobalamin by the dual isotope competitive radioimmunoassay in cats has been performed internally by the laboratories (K. Findlay, personal communication, H. Evans, Personal communication).

Table 1.

Methods of measurement of cobalamin and folate

Laboratory	Number of cases	Cobalamin assay	Reference range	Folate assay	Reference range
Lab 1 *	91	Competitive chemiluminescent enzyme immunoassay †	290–1499 ng/l	Competitive immunoassay †	9.7–21.6 μg/l
Lab 2 ‡	8	Dual isotope competitive radioimmunoassay §	>150 ng/l	Dual isotope competitive radioimmunoassay §	8.5–20 μg/l
Lab 3 ‖	3	Competitive chemiluminescent enzyme immunoassay †	200–1680 ng/l	Competitive immunoassay †	13.4–38 μg/l
Lab 4 ¶	1	Competitive chemiluminescent enzyme immunoassay †	>150 pmol/l (>202 ng/l)	Competitive immunoassay †	8.5–20 μg/l

^*Gastrointestinal Laboratory, Texas A & M University.

^†Immulite 200 (Diagnostic Products Corporation).

^‡Cambridge Specialist Laboratories, Cambridge.

^§SimulTRAC-SNB (MP Biomedicals).

^‖Royal Veterinary College, London.

^¶Capital Diagnostics, Roslin.

Routine haematology and serum biochemistry

In the majority of cases (n=94) routine haematology and serum biochemistry were performed by the University of Edinburgh Veterinary Pathology Unit (UEVPU). Haematology analysis was performed using a Pentra 60 haematology analyser (ABX Diagnostics, Montpellier, France). Biochemistry was performed on an opeRA autoanalyser (Bayer Diagnostics, Tarrytow, NY, USA). Electrolytes were measured using an AVL 9180 Electrolyte Analyser (AVL Scientific Corporation, Roswell, USA). Three cases had analysis performed on ‘in-house’ machines (QBC haematology and VetTest biochemistry analysers; Idexx Laboratories, ME, USA), and six samples were submitted to external laboratories.

Statistical analysis

Statistical analysis was performed using a computer software package (Minitab for Windows; Minitab, USA). The χ² test was used to compare proportions between groups. A Kruskal–Wallis test was used when comparing groups for data that was not normally distributed, or where the sample size was small. A Mann–Whitney test was used for post-hoc testing to compare between two groups. Statistical significance was set at P<0.05.

Results

Study population

The initial search of the database identified 110 cases that had been sampled for cobalamin and folate levels. Three cases were excluded because the case records could not be found, and a further four cases were excluded because concurrent haematology and serum biochemistry sampling had not been performed, leaving a total of 103 cases. None of these cases had received folate supplementation prior to sampling, but three had received cobalamin within the month prior to referral. The mean age was 7.1 years (SD 4.48); 59 cases were female (57 neutered; two entire), and 44 cases were male (43 neutered; one entire). The majority of cats (n=62; 60.2%) were crossbreed (domestic shorthair n=55; domestic longhair n=7), with a total of 12 pure breeds being represented. Siamese was the most popular pure breed (8.7%), followed by Persian (5.8%), Bengal (4.8%), Birman (3.9%) and Maine Coon (3.9%). The remaining seven breeds comprised less than 3% each of the study population. The percentage of purebreds within the study population (39.8%) was notably higher than that within the general hospital feline population of 15.7% (see Fig 1). This difference was significant (P<0.001). Bodyweight was available for 102 cases, with a mean of 4.16 kg (SD 1.23). Body condition was scored out of 9, and was available for 89 cases; the median BCS was 4 (range 2–9). The median duration of illness was 10 weeks (range 0.5–324 weeks), with 33% of cases presented from 1 to 6 months. Fig 2 shows the distribution of reported presenting problems, with the three most frequent being vomiting (n=41), weight loss (n=36) and anorexia (n=34).

Fig 1.

Comparison of percentage purebreed vs non-purebreed cats in hospital and study populations. Hospital population=UEHFSA feline population, study population=103 cats within study, Domestic=non-purebreed.

Fig 2.

Presenting problem. Abdom. Pain=abdominal pain; PD=polydipsia; PU=polyuria; DM=diabetes mellitus; n=number.

Table 2 presents a breakdown of the case distribution by system affected. Fifty-nine cases (57%) had evidence of GI disease. Thirty cases (29%) had pancreatitis, five of which had triaditis, five had concurrent hepatic disease and four had concurrent diabetes mellitus. Thirteen cases (12.6%) showed evidence of hepatic disease and 13 cases (12.6%) were diagnosed with diseases that did not relate to the alimentary tract. In three cases (2.9%) no diagnosis was made.

Table 2.

Case distribution by body system affected

GI, pancreas and liver, n=5	GI and pancreas, n=4	Pancreas and liver, n=5	GI, n=50	Pancreas, n=21	Liver, n=2	Other, n=16
Triaditis (n=5)	Pancreatitis and salmonella (n=1)	Pancreatitis and hepatic lipidosis (n=2)	Inflammatory (n=26)	Pancreatitis (n=12)	Hepatitis and portosytemic shunts (n=1)	Endocrine (n=5) DM (n=2) DM and acromegaly (n=1) Hyperthyroid (n=1) Hypothyroid (n=1)
	Pancreatitis and FB (n=1)	Pancreatitis and hepatitis/cholangiohepatitis (n=2)	Infectious (n=13) Helicobacter (5) Salmonella (3) Tritrichomonas (3) Giardia (2)	Neoplasia (n=2)	Cholangiohepatitis (n=1)	No diagnosis (n=3)
	Gastric ulcer and EPI (n=1)	Pancreatitis, hepatitis and CRF (n=1)	Infiltrative (n=8) Lymphoma (n=4) Amyloidosis (n=1) Histiocytic (n=1) Colonic mass (n=1) Hypereosinophilic syndrome (n=1)	Pancreatitis and DM (n=3)	Miscellaneous (n=2)
	Pancreatic neoplasia and IBD (n=1)		Motility disorder (n=3)	Pancreatitis, DM and HAC (n=1)		Haematological (n=2)
				Pancreatopathy and hyperthyroidism (n=1)		Renal (n=1)
				Pancreatitis and pancreatolithiasis (n=1)		Immunodeficiency (n=1)
				Pancreatitis and EPI (n=1)		Mesenteric lymphadenitis (n=1) Oesophagitis (n=1)

GI=gastrointestinal; IBD=inflammatory bowel disease; FB=foreign body; CRF=chronic renal failure; EPI=exocrine pancreatic insufficiency; HAC=hyperadrenocorticism; DM=diabetes mellitus.

Cobalamin and folate

The mean cobalamin measurement was 795.7 ng/l (SD 441.9); 16.5% of cases (n=17) had cobalamin levels below the lower end of the reference range for the laboratory used, of which seven had values <100 ng/l, three had values between 100–200 ng/l and seven had values between 200 and 290 ng/l. The median BCS for cats with low cobalamin (BCS 3; range 2–7), was significantly lower (P=0.049) than the median BCS for cats with normal cobalamin (BCS 4; range 2–9). The mean weight for cats with low cobalamin was 3.83 kg (SD 1.281), which was not significantly different compared to a mean weight of 4.24 kg (SD 1.218) for cats with normal cobalamin levels (P=0.22).

The mean serum folate measurement was 12.7 μg/l (SD 6.97); 38.8% of cases (n=40) had folate levels below the lower end of the reference range and 9.7% (n=10) had folate levels above the reference range for the laboratory used. The median BCS for cats with low folate (BCS 3; range 2–7) was not significantly different (P=0.153) when compared to the median BCS for cats with normal folate (BCS 4; range 2–9). The mean weight for cats with low folate was 3.90 kg (SD 1.026), compared to a mean weight of 4.33 kg (SD 1.330) for cats with normal folate. This difference was not significant (P=0.067).

Five cases (4.8%) had low levels of both cobalamin and folate; two of these cases had lymphoma, one had inflammatory bowel disease, and two had pancreatitis (one with cholelithiasis). Cats with both a low cobalamin and low folate had a median BCS of 2.5 (range 2–3) and a mean body weight of 3.11 kg (SD 0.51), which was significantly lower than the median BCS of 4 (range 2–9), and mean bodyweight of 4.39 kg (SD 1.31) for cats that had neither low cobalamin nor folate (BCS P=0.024; bodyweight P=0.002; see Fig 3).

Fig 3.

Boxplots of bodyweight and body condition score against cobalamin and folate status. Bodyweight and body condition score are plotted for four groups; grey=neither cobalamin nor folate were low; red=low folate, normal cobalamin; green=low cobalamin, normal folate; blue=low cobalamin and low folate. Box represents 25–75% interquartile range, and horizontal bars represent mean (bodyweight) and median (BCS) values. *Represents outliers. Significant difference when compared to cats without low cobalamin or low folate. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Haematology and serum biochemistry

Anaemia (red blood cell count <5.5×10¹²/l and packed cell volume <24%) was identified in six cases. Macrocytosis (mean corpuscular volume (MCV)>55 fl) was only identified in three cases; two of these cases had pancreatic neoplasia (MCV 67 fl and 61 fl), and the third had liver disease, with acquired portosystemic shunts (MCV 67 fl). None of these three cases had low cobalamin or folate levels. The mean MCV for cats without low folate or cobalamin was 46.0 fl (SD 5.44); the mean MCV for cats with low folate was 45.7 fl (SD 5.91), which was not statistically significant (P=0.63); the mean MCV for cats with low cobalamin was 47.2 fl (SD 3.12), which also was not statistically significant (P=0.18). Two cats with anaemia had low folate levels; one cat had myelodysplasia, and the second had Candidatus Mycoplasma haemominutium infection and chronic renal failure.

Inorganic phosphate measurement was available for 100 cases, of which 48 had values below the reference range for the laboratory used. More than 90% of samples were submitted to the UEVPU, which has a reference range for inorganic phosphate of 1.4–2.5 mmol/l. Of the 48 cases with hypophosphataemia, 25 had GI disease, 15 had pancreatitis, 11 had hypoalbuminaemia, and three had hypocalcaemia. The mean inorganic phosphate level for cats in four groups was assessed: cats without low folate or cobalamin; cats with normal cobalamin and low folate; cats with normal folate and low cobalamin; cats with both low folate and cobalamin (see Fig 4). Cats with low folate and low cobalamin had significantly lower inorganic phosphate levels than cats with normal folate and normal cobalamin (P<0.003; see Fig 4).

Fig 4.

Histogram of inorganic phosphate. Frequency of inorganic phosphate values are plotted for four groups of cats. normal cobalamin; normal folate. normal cobalamin; low folate (P=0.836; T=−0.21)*. low cobalamin; normal folate (P=0.205; T=−1.34)*. low cobalamin; low folate (P=0.003; T=4.43)*. *comparing mean values with normal cobalamin; normal folate group.

Twenty-seven cases (26%) had low albumin, of which 19 had GI disease. Of the remaining eight cases, six had pancreatitis (one with concurrent cholangiohepatitis and chronic renal failure, and one with concurrent hepatic lipidosis), one cat had advanced liver disease, and in the remaining case, no diagnosis was made. Cats with GI disease were significantly more likely to have low albumin levels than cats that did not have GI disease (P=0.039). There was no association between serum folate and cobalamin levels and either serum albumin levels or the presence of GI disease (see Table 3).

Table 3.

Associations between variables

	Folate	Cobalamin	Inorganic phosphate	Albumin	GI disease
Folate
Cobalamin	0.172
Inorganic phosphate	0.120	0.850
Albumin	0.822	0.474	0.487
GI disease	0.610	0.520	0.927	0.039 *
Low folate and cobalamin			<0.001 *	0.832	0.664

Results for the χ² test for two proportions performed to compare the association between variables (significance P<0.05).

^*Significant association.

Discussion

The over-representation of purebred cats in this study was an unexpected finding. Whilst it could be argued that owners of purebred cats may be more likely to pursue investigations of disease than owners of crossbred cats, this would be likely to be true for all the UEHFSA population, which is predominantly referral based. Siamese cats were the most common breed represented, but were also the most common breed in the UEHFSA population. Although Siamese cats have been identified to be at increased risk for lymphoma (Louwerens et al 2005), all of the cases of lymphoma in this study occurred in domestic shorthair (n=3), or domestic longhair (n=1) cats. No specific diagnosis appeared common to a particular breed.

Low cobalamin levels were less common than low folate levels and were identified in 17 cases (16.5%). This incidence is lower than the incidence of 61% obtained in a study conducted in the United States (Simpson et al 2001). However, it is higher than the incidence obtained in a recent study conducted in the United Kingdom (Ibarrola et al 2005), where a total of 814 cats had cobalamin measured and only one had a value <150 ng/l, the lower end of the reference range used by the laboratory in that study. However, 44 had values less than 290 ng/l, which is comparable to the lower end of the reference range used in the majority of cases in the present study. Using 290 ng/l as the lower limit increases the incidence of hypocobalaminaemia in that study to 5.4%, which is still substantially less than that identified in the current study. The high incidence obtained in the study by Simpson et al (2001) may possibly reflect the use of a different reference range based on young cats, as younger cats have significantly higher cobalamin levels than older cats (Ibarrola et al 2005). The differences obtained between the two United Kingdom studies may have a number of explanations. The current study population contained a higher number of cats with alimentary tract disease than non-alimentary tract disease, compared to the clinical population studied by Ibarrola et al (2005). In addition, although the samples submitted to the laboratory in the study by Ibarrola et al (2005) were suspected to have GI disease, the final diagnosis was not known. Administration of cobalamin to cases prior to submission to the laboratory may also have been difficult to establish, as it would rely on submitting vets having disclosed this on a submission form. Only three cases in the current study were known to have received prior treatment with cobalamin, all of which had cobalamin values >1000 ng/l. Another factor may be the different techniques involved in measurement of cobalamin. In the present study, all the low results were obtained when a chemiluminescent enzyme immunoassay was used, rather than a dual isotope competitive radioimmunoassay.

Seven cases had cobalamin levels <100 ng/l (below the lower limit of detection for the assay); such low levels have previously been shown to be associated with metabolic abnormalities in amino acid metabolism (Ruaux et al 2001, 2005), however, no assessment of metabolic abnormalities associated with hypocobalaminaemia was performed in the current study. Severely decreased cobalamin levels in association with exocrine pancreatic insufficiency have previously been identified (Steiner and Williams 2000), and both the cases of exocrine pancreatic insufficiency in this study had cobalamin levels <100 ng/l. This supports the previous findings and indicates the need to evaluate cobalamin levels when exocrine pancreatic insufficiency is diagnosed.

Low serum folate levels were common in this study population (39%). Of the 40 cases with low folate levels, 25 were identified with disease of the GI tract. Of the remaining 15 cases, eight suffered from pancreatitis or cholangiohepatitis, four had miscellaneous conditions, and in three no diagnosis was made. However, GI biopsies had not been obtained in these 15 cases, therefore, it is possible that GI disease was present but had not been identified. As inflammatory bowel disease is frequently seen in association with pancreatitis and/or cholangiohepatitis (Weiss et al 1996), this would certainly seem feasible in at least some of the cases. Whilst pancreatic biopsy is considered the ‘gold standard’ for diagnosing pancreatitis, the diagnosis is often based on non-invasive tests such as abdominal ultrasound and pancreatic lipase levels, as was the case in this study. Having made a presumptive diagnosis of pancreatitis by non-invasive means, owners may, therefore, be reluctant to pursue intestinal biopsy in order to confirm whether concurrent GI disease is present. Folate absorption should primarily be affected by disease of the proximal small intestine, therefore, low folate levels should alert the clinician that concurrent GI disease could be present.

High folate levels were identified in 10 cats, but the significance of this is unknown. A possible explanation for this is increased synthesis of folate by bacteria, as proposed for the condition of SIBO. However, the diagnosis of SIBO in dogs has been called into question (German et al 2003), and seems even less likely to occur in cats (Johnston et al 2001). Alternative explanations for increased serum folate levels include a low intestinal pH, which favours folate absorption from the intestine, or haemolysis where folate is released from erythrocytes (Suchodolski and Steiner 2003).

Cats with low cobalamin levels did not show a significant difference in bodyweight compared to cats with normal cobalamin levels, but there was a significant difference in BCS. Although cats with low serum folate showed a trend towards lower body weights and BCS, the results were not significantly different. Comparison of bodyweight has some limitations in assessment of severity of disease, due to breed variations, therefore, BCS may be a more reliable indicator of loss of weight and condition associated with severity of disease. These findings suggest that diseases affecting cobalamin absorption may be more likely to cause loss of condition than those affecting folate absorption.

Combined cobalamin and folate deficiency was associated with lower mean body weight and BCS. Low cobalamin and folate has previously been identified in association with intestinal lymphoma, and is believed to be supportive of the presence of diffuse intestinal disease (Simpson et al 2001), hence the finding of lower bodyweights and BCS is not unsurprising. In the current study, two of the cases of lymphoma had low folate and low cobalamin and a third case of lymphoma had low folate, but not cobalamin. This latter case had, however, received two injections of cobalamin 21 and 7 days before sampling, which would have elevated the cobalamin result. Assessment of these vitamins and their supplementation may, therefore, be indicated in cases of GI lymphoma. It is not known whether the half-life of exogenous cobalamin is reduced in cats with lymphoma, as it appears to be for cats with inflammatory bowel disease (Simpson et al 2001), but weekly injections would seem appropriate.

Macrocytosis was not identified in association with folate or cobalamin deficiency in the current study. Previous studies involving dietary deficiency of folate have shown that macrocytosis is unusual, but megaloblastic erythroblasts can be detected in the bone marrow, and urinary excretion of formiminoglutamatase (FIGLU) is increased (Thenen and Rasmussen 1978, Yu and Morris 1998). In the study by Simpson et al (2001), macrocytosis was reported in six of 22 cats with hypocobalaminaemia. In that study, macrocytosis was diagnosed when the MCV exceeded 52 fl, and no quantification of the degree of macrocytosis was given. In the current study, the upper limit of the laboratory reference range was 55 fl, and the three cases of macrocytosis all had MCV>60 fl, which may account partially for the differences in incidence between the two studies. Macrocytosis has previously been reported to be rare in animals in association with cobalamin deficiency (Watson and Canfield 2000), and it is a late occurrence in humans with cobalamin or folate deficiencies. Reduction in serum levels of folate and/or cobalamin, hypersegmentation of neutrophils, decreased erythrocyte concentration of cobalamin and folate and elevated urine FIGLU are seen prior to the observation of macrocytosis (Herbert and Colman 1988). Measurement of erythrocyte cobalamin or folate levels and assessment of urinary FIGLU may therefore give further support to evidence of these deficiencies in the absence of megaloblastic anaemia.

The clinical significance of low folate levels in this population is unknown, as assessment of metabolic consequences such as increased urinary FIGLU levels was not performed. In addition, serum levels may not reflect red blood cell levels, and the sensitivities and specificities for the different methods of analysis are not known. Despite using the same technique of analysis, laboratory 3 has quite a different reference range from laboratories 1 and 4, which is difficult to explain. Comparison of results obtained by different techniques was beyond the scope of this paper, and to the authors' knowledge has not been performed for either folate or cobalamin.

Hypophosphataemia was an unexpected finding in the study population, affecting 46% of cases. Most tests (>90%) were performed at the UEVPU, which has a validated reference range of 1.4–2.5 mmol/l. Although this is a high lower limit compared to some laboratories, it is not dissimilar to the reference range for cats quoted by other authors (Bush 2005). In addition, during the period of this study, re-evaluation of the reference range was performed with samples from 20 healthy cats, which supported this reference range as valid.

Pseudohypophosphataemia may arise from sampling artefact with regard to time of day obtained, post-prandial alkaline tide and venepuncture technique, however, as the cats in the study population were sampled in a similar fashion as those used for generating the reference range, it was felt unlikely that this would contribute to the difference seen. Pseudohypophosphataemia in association with haemolysis and icterus has been identified in dogs (Harkin et al 1998), when a colourimetric assays was used. In the present study, 12 cats had hyperbilirubinaemia, although only four were clinically icteric. Five of these 12 cats had low phosphate, one had high phosphate, and six had normal phosphate. The cats with the two highest bilirubin values had normal phosphate values, therefore, it would appear unlikely that hyperbilirubinaemia was resulting in artefactual hypophosphataemia, although evaluation of phosphate by determination of soluble serum elements in hypophosphataemic animals would confirm this (Harkin et al 1998).

The majority of cases with hypophosphataemia had GI disease (n=25), or pancreatitis (n=9). Previously, hypophosphataemia in cats has predominantly been recognised in association with diabetes mellitus, hepatic lipidosis and enteral alimentation (Forrester and Moreland 1989, Adams et al 1993, Justin and Hohenhaus 1995). Decreased intestinal absorption of phosphate can lead to hypophosphataemia in humans (Forrester and Moreland 1989, Willard and DiBartola 2000), but the significance in veterinary patients is unknown. As phosphate is absorbed from the small intestine, by both passive diffusion and active mucosal transport (Willard and DiBartola 2000) it would not be unreasonable to expect that animals affected with diffuse GI disease become hypophosphataemic. Although numbers were low, the lowest mean inorganic phosphate levels were seen in the cases with both low folate and cobalamin, supporting malabsorption of these vitamins and this mineral secondary to diffuse GI disease. As this was also the population with lowest mean bodyweights and BCS, this could be considered the population most likely to receive nutritional support. As enteral nutrition has been associated with marked hypophosphataemia (Justin and Hohenhaus 1995), caution should be exercised when this is undertaken in this type of patient, with frequent monitoring of phosphate levels. No cases had an inorganic phosphate level below 0.32 mmol/l, which is the level that has previously been associated with haemolysis (Willard and DiBartola 2000).

There were a number of limitations to this study. The retrospective nature of the study meant that not all of the data were present for all cases, particularly BCS. Although less prone to breed variation, BCS can be a subjective measurement, especially if a standard description was not referred to at the time of scoring. A prospective study could enable the use of a ‘severity score’ which may have identified associations not possible with a retrospective study. Although widespread use of folate supplementation prior to investigation seems unlikely, it is possible that more cats than were identified had received cobalamin supplementation, especially where full printed notes were not received from referring vets. Further limitations include the lack of diagnosis obtained in some cases, a possible failure to diagnose more than one concurrent conditions and failure to obtain pancreatic biopsies for the diagnosis of pancreatitis. These limitations were affected not only by the retrospective nature of the study, but also by the clients' wishes and financial constraints. In addition, the small number of cases in the low folate and low cobalamin group limited statistical analysis of this group.

In conclusion, subnormal serum folate appears to be more common than hypocobalaminaemia in cats suffering from alimentary tract disease. There did not, however, appear to be an association between degree of folate deficiency and severity of disease, as assessed by bodyweight and BCS, and macrocytosis was not identified. Further work is required to assess the clinical significance of folate deficiency, and the role of folate supplementation. Cobalamin deficiency was identified more commonly in this study than from previous work in the United Kingdom. Several cases were severely hypocobalaminaemic, at a level where cobalamin supplementation is required to prevent metabolic consequences from the deficiency. The high incidence of hypophosphataemia has not been noted previously. Although of minimal clinical significance in itself, clinicians should be aware of phosphorous levels in this population, especially if enteral nutrition is contemplated.