• Open Access

The Relationship of Serum Cobalamin to Methylmalonic Acid Concentrations and Clinical Variables in Cats


Corresponding author: K.W. Simpson, College of Veterinary Medicine, Cornell University, VMC 2011, Ithaca, NY 14853; e-mail: kws5@cornell.edu.



Serum cobalamin concentration [CBL] suggests CBL deficiency in cats but serum methylmalonic acid concentration [MMA] more accurately indicates CBL deficiency.


To examine the ability of [CBL] to predict CBL deficiency defined by increased [MMA], and relationships of [CBL] and [MMA] with select clinical and clinicopathological variables.


One hundred sixty-three client-owned cats with [CBL] measurements, 114 cats with simultaneous [MMA] measurements; 88 cats with medical information.


Prospectively collected [CBL] and [MMA] were compared using scatter plots, receiver operating characteristic and correlative analyses with historical [CBL] thresholds and those identified in the study. [CBL] and [MMA] were compared retrospectively to specific clinical and clinicopathological variables.


[CBL] correlated negatively with [MMA] (τ = −0.334, P < .0001). [MMA] ≥ 1,343 nmol/L identified CBL deficiency. [CBL] = 209 pg/mL optimized sensitivity (0.51), specificity (0.96), PPV (0.89), and NPV (0.74) for detecting [MMA] ≥ 1,343 nmol/L. Prevalence of CBL deficiency was 42% (48/114) when defined by [MMA] ≥ 1,343 nmol/L versus 23% (27/114) by [CBL] ≤ 209 pg/mL. Unexpectedly, 23 and 45% of 48 cats with [MMA] ≥ 1,343 nmol/L had [CBL] > 900 pg/mL and 290 pg/mL (historical thresholds). [CBL] correlated with mean corpuscular volume (τ = −0.199, P = .013) and [MMA] with hematocrit (τ = −0.28, P = .006).

Conclusions and Clinical Importance

Cobalamin deficiency ([MMA] ≥ 1,343 nmol/L) occurred in 42% of cats and is predicted with high specificity by [CBL] ≤ 209 pg/mL. CBL status correlates with microcytosis and anemia. Discordance between [CBL] and [MMA] cautions against relying on any single marker for determining CBL status.


serum cobalamin concentration


Cornell University Diagnostic Laboratory


Cornell University Hospital for Animals




inflammatory bowel disease


mean corpuscular volume


serum methylmalonic acid concentration




receiver-operating characteristic

Cobalamin (vitamin B12) is a water-soluble vitamin involved in neuronal function, hematopoiesis, DNA and fatty acid synthesis, and energy production.[1-3] CBL absorption requires binding proteins and specific receptors along various parts of the gastrointestinal tract.[4-6] Consequently, gastrointestinal disease can lead to CBL deficiency. Subnormal serum cobalamin concentrations [CBL] most frequently have been reported in cats with suspected gastrointestinal disease, and clinical findings of weight loss, diarrhea, vomiting, anorexia, lethargy, or thickened intestines.[5, 7] Low [CBL] also has been linked to increased mean corpuscular volume (MCV) and decreased serum phosphorous concentration in some studies.[5, 7] The types of gastrointestinal disease associated with subnormal [CBL] include inflammatory bowel disease (IBD), intestinal lymphoma, cholangiohepatitis or cholangitis, pancreatitis and exocrine pancreatic insufficiency,[5, 7-10] reflecting the involvement of these organs in feline CBL homeostasis.[5]

The reported prevalence of subnormal [CBL] in cats with gastrointestinal disease ranges from 0.1 to 78%.[5, 7-9, 11] This variation potentially is a consequence of the assay used, reference intervals, patient population, geographical location, or some combination of these factors.[5, 7-9, 11]

A potential solution to this problem is to determine the [CBL] that correlates with decreased activity of 2 CBL-dependent enzymes: methylmalonyl-CoA mutase and methionine synthase. In people, decreased activity of these enzymes causes increased serum and urine concentrations of methylmalonic acid [MMA] and homocysteine, respectively.[12] Cats with undetectable or subnormal [CBL] have significantly increased [MMA], but not increased homocysteine concentrations.[5, 6, 13] Cats with increased [MMA] also had significantly increased serum concentrations of methionine and significantly decreased serum concentrations of cystathionine and cysteine.[13] Furthermore, treating cats that had high [MMA] with parenteral CBL decreased the [MMA] and facilitated weight gain.[14]

These findings suggest that measurement of [CBL] is a useful marker of gastrointestinal disease, and that treating cats diagnosed with CBL deficiency on the basis of an increased [MMA] confers a therapeutic benefit. Because MMA is not routinely measured in most clinical laboratories, it would be valuable to know the [CBL] at which CBL deficiency develops (ie, the point at which cats would benefit from supplementation). Initial studies conducted in cats with subnormal [CBL] (defined a priori as < 290 pg/mL) showed that a [CBL] ≤ 160 pg/L had a 74% sensitivity and 80% specificity for detecting [MMA] > 867 nmol/L.[6] However, [MMA] has not been correlated with [CBL] > 290 pg/mL. Given that other laboratories have found substantially higher [CBL] in their healthy cat populations (290–900 pg/mL), it is possible that CBL deficiency, as defined by an increased [MMA], exists in cats with [CBL] > 290 pg/mL. Additionally, no data exist correlating gastrointestinal disease with CBL deficiency, or with the impact of CBL deficiency on clinicopathological variables.

Consequently, we sought to evaluate the ability of [CBL] to predict high [MMA], and the relationships of CBL and MMA to select clinical and clinicopathological variables.


Patients and Samples

Residual feline sera from cases (n = 163) that had been submitted for serum CBL analysis to the Cornell University Diagnostic Laboratory (CUDL) were prospectively archived from 2000 to 2002; 88 samples were from cats presented to the Cornell University Hospital for Animals (CUHA) and 75 were submitted directly to the CUDL by referring veterinarians. All sera were stored at −30°C and submitted as a batch for MMA analysis as detailed below. Information regarding age, sex, and breed, as well as presumed or known diagnoses, was recorded for cats presented to CUHA. Medical records for the 88 cats presenting to CUHA were examined for evidence of CBL administration before blood sampling and any cat that had received CBL was excluded. The 75 samples submitted directly to the CUDL lacked signalment and clinical information.

Presumptive diagnoses, originally ascribed to each of the 88 cats by the attending internist at CUHA, were collectively confirmed by 4 of the authors (KWS, MR, PW, OT) from examination of the medical record and based on history, physical examination findings and results of diagnostic testing (CBC, serum biochemistry, urinalysis, and abdominal ultrasound examination). Definitive diagnoses were based on the previously mentioned criteria together with histopathology. Cats were categorized as having disease of the alimentary system, including the gastrointestinal tract, pancreas, liver, or some combination of these, or having diseases unrelated to the alimentary system. To be included in the alimentary disease category, cats had to have ≥ 1 of the following clinical signs: weight loss, inappetence, vomiting, diarrhea, or failure to thrive and no evidence of another underlying disease process that could explain the clinical signs according to previously established criteria.[13]

Clinicopathological Analyses

CBC and serum biochemistry profiles were performed at the Clinical Pathology Laboratory at Cornell University using a hematology analyzer1 and a chemistry analyzer.2

Serum cobalamin concentrations were measured using an automated chemiluminescence assay.3 The CBL assay for the Immulite 1000 performs similarly to the previously described radioimmunoassay (RIA), (r2 = 0.9910, P < .0001 for CBL).[5] The interassay coefficient of variation for [CBL] in feline serum samples determined by RIA and Immulite 1000 assays over 39 assays was 8% for 386 pg/mL and 3% for 1280 pg/mL.

Methylmalonic acid, homocysteine, methionine, and cysteine were measured at the Gastrointestinal Laboratory at Texas A&M University using stable isotope dilution gas chromatography/mass spectrometry.4 The stable isotope is a deuterated version of the actual analyte of interest (ie, trideuterated MMA). Addition of a known amount of this isotope allows for calculation of a ratio of the area under the peak for both the endogenous analyte and the isotope. This information then can be used to calculate the concentration of the analyte in the sample.[13, 15]

Statistical Analysis

Because data were not normally distributed, we examined the relationship between [CBL] and [MMA] with Spearman rank correlation. We then created a scatter plot to visually determine the optimal cutoff points for our patient cohort in correctly identifying animals with CBL deficiency, as defined as the inflection point at which [MMA] increased consistently. Using the [MMA] identified by examining the scatter plot, and a historical [MMA] of 867 nmol/L, we performed a receiver operating characteristic (ROC) analysis to determine the optimal [CBL] to predict increased [MMA]. For this primary aim, we set significance at P < .05.

Having established independent cutoff points that optimally identified patients as being CBL-deficient (low [CBL] or high [MMA]), we performed a χ2 test or a Fisher's exact test to examine whether a [CBL] < 209 pg/mL or [MMA] ≥ 1343 nmol/L was associated with the presence of gastrointestinal disease. We also compared [CBL] and [MMA] in cats with histologically confirmed lymphoma and histologically confirmed IBD using a Mann–Whitney U-test. For these analyses, to compensate for experiment-wise error, we set significance at P < .013.

We tested biochemical variables for normality using the Shapiro–Wilk test and elected to use nonparametric statistics to examine the association of these variables with CBL or MMA. Because CBL deficiency is associated with hypochromic macrocytic anemias, we performed both Kendall's tau rank correlations and rank sum tests to examine the association of [CBL] < 209 pg/mL or [MMA] ≥ 1343 nmol/L and hematocrit (HCT), MCV, and serum phosphorus concentration. To preserve our experiment-wise error rate at .05, we set significance at P < .017 for these 3 comparisons. Because discordant [MMA] and [CBL] results have been associated with renal failure in humans, we performed Kendall's tau rank correlations to examine the association of MMA and serum creatinine concentration.

Finally, we examined the relationship between [CBL] or [MMA] and methionine or cysteine using Kendall's tau rank correlations. For these 4 analyses, we set significance at P < .0125 to preserve the experiment-wise error rate at .05. All statistical analyses were performed using commercial statistical software.5


Cohort Demographics

A complete signalment was available for 78/88 cats that presented to CUHA for evaluation; in 10 cats date of birth was not recorded. The median age was 12 years (range, 1–20 years). Forty-seven cats were male (45 neutered, 2 intact) and 41 cats were female (37 spayed, 4 intact). The breeds represented consisted of domestic shorthair (n = 69), Siamese (n = 8), domestic longhair (n = 6), Burmese (n = 2), Himalayan (n = 1), Persian (n = 1), and Russian Blue (n = 1).

Specific diagnoses in cats in the alimentary disease group included IBD, intestinal lymphoma, cholangitis, cholangiohepatitis, cholecystitis, adenocarcinoma of the colon, Helicobacter infection, hepatitis, hepatocellular carcinoma, hepatoma, pancreatitis, hepatic lipidosis, and coccidiosis. Specific diagnoses in cats in the nonalimentary disease group included chronic kidney disease, diabetic ketoacidosis, diabetes mellitus, feline immunodeficiency virus infection, hyperthyroidism, neurologic disease, squamous cell carcinoma, cardiac disease, and lymphoma of the palate.

Relationship of CBL and MMA

Simultaneous [CBL] and [MMA] measurements were obtained from 114 of 163 cats. Medical records were available for 55 of 114 cats. CBL showed a significant, but modest, negative correlation with [MMA] (τ = −0.334, P < .0001). To address the possibility that the external submissions and CUHA samples represented different populations, we compared the median [CBL] and [MMA] in these 2 populations. These populations did not differ in either MMA (CUHA: median, 740; range, 92–68,242; external median, 1,011; range, 184–75,580; P > .05) or CBL (CUHA: median, 815; range, 50–3,000; external median, 515; range, 50–3,000; P > .05).

Scatter Plot to Determine Breakpoint

We constructed a scatter plot of [CBL] versus [MMA] to examine whether an obvious inflection point could be identified independently of any preconceived values for either variable (Fig 1). This inflection point would help determine the [CBL] and [MMA] that would best predict [MMA] from known [CBL]. The graph demonstrated that a [CBL] of < 189 pg/mL identified cats with increased [MMA], with no false positive results below this [CBL]. The scatter plot also indicated that [MMA] ≥ 1343 nmol/L was a better indicator of CBL deficiency in this study population than the historical [MMA] of > 867 nmol/L because it improved sensitivity without decreasing specificity.

Figure 1.

Scatter plot of cobalamin concentration [CBL] versus methylmalonic acid concentration [MMA] in 116 cats. Vertical dashed lines indicate historical threshold [CBL] defining normal and subnormal [CBL] (290 and 900 pg/mL). Vertical solid line indicates the statistically optimal [CBL] for discriminating normal and subnormal [CBL] determined in this study (189 pg/mL). Horizontal dashed line indicates historical threshold MMA concentrations (867 nmol/L). Horizontal solid line indicates the optimal MMA concentration for discriminating normal and subnormal MMA concentrations determined in this study (1,343 nmol/L). Open circles—cats presenting to CUHA; closed circles—cats with samples submitted to the AHDL.

Similarly, ROC analysis of [MMA] versus [CBL] resulted in an area under the curve of 0.72 (95% CI, 0.63–0.80) using a cutoff of 867 nmol/L and 0.78 (95% CI, 0.7–0.86) using a cutoff of 1343 nmol/L. The Youden index suggested the optimal cutoff that maximized sensitivity and specificity was a [CBL] of 209 pg/ml, regardless of the [MMA] cutoff value (867 versus 1343 nmol/L) selected (Fig 2).

Figure 2.

Receiver operating characteristic curve for predicting cobalamin deficiency from methylmalonic acid [MMA] concentrations at both 867 and 1,343 nmol/L.

Prevalence of CBL Deficiency

Using this higher [MMA] (≥ 1,343 nmol/L) to define CBL deficiency in this study, we examined the CBL status of the 114 cats for which [MMA] was available. Based on our laboratory reference threshold concentration for normal serum [CBL] (> 900 pg/mL), 65 of 114 (57%) cats had subnormal [CBL]. Using the historical [CBL] of 290 pg/mL to define the threshold concentration for normal [CBL],[6] 34 of 114 (30%) cats had subnormal [CBL]. Using a [CBL] cutoff of 209 pg/mL, 26 of 114 (23%) cats were deficient. Conversely, using [MMA] ≥ 1,343 nmol/L to define CBL deficiency, 46 of 114 (41%) cats were deficient.

Subsequently, we examined the sensitivity and specificity of various [CBL] in identifying normal (< 1,343nmol/L) or abnormal (≥ 1,343 nmol/L) [MMA]. We used the threshold concentration provided by our 2 institutional laboratories (900 pg/mL for Cornell and 290 pg/mL for Texas A&M), the concentration identified by ROC analysis (209 pg/mL) and the concentration identified by visual inspection of the scatter plot (189 pg/mL). The results are shown in Table 1.

Table 1. Sensitivity and specificity of using cobalamin concentrations [CBL] to predict methylmalonic acid concentrations [MMA] < 1,343 nmol/L
[CBL] pg/mLPrevalence (48/114)SensitivitySpecificityPPVNPV
  1. NPV, negative predictive values; PPV, positive predictive values.

  2. 95% confidence intervals for the point estimates of test performance are provided in parentheses next to the point estimate.

9000.41 (0.32–0.5)0.77 (0.62–0.87)0.56 0.43–0.68)0.55 (0.42–0.67)0.78 (0.63–0.88)
2900.410.55 (0.4–0.7)0.88 (0.78–0.94)0.76 (0.58–0.89)0.74 (0.63–0.83)
2090.410.51 (0.36–0.66)0.96 (0.87–0.99)0.89 (0.7–0.97)0.74 (0.63–0.82)
1890.410.43 (0.29–0.58)1 (0.93–1.0)1 (0.8–1.0)0.72 (0.61–0.8)

We then examined the predictive values using an [MMA] < 1,343 nmol/L to indicate normal CBL status at various prevalences (Table 2). At each prevalence, positive predictive value increased dramatically as the [CBL] cutoff became more stringent, but negative predictive value decreased minimally.

Table 2. Predictive values of various cobalamin concentration [CBL] cutoff values at various prevalences as determined by methylmalonic acid concentration [MMA] < 1,343 nmol/L
[CBL] pg/mLPrevalenceSensitivitySpecificityPPVNPV
  1. NPV, negative predictive values; PPV, positive predictive values.


Discordant MMA and CBL Results

Discordant MMA and CBL results were seen in a high proportion of the cats (Table 3). Of the 46 cats with CBL deficiency defined as [MMA] ≥ 1,343 nmol/L, 11 (23%), 20 (46%), and 23 (50%) had [CBL] within the reference interval for healthy cats when applying the ≥ 900 pg/mL, ≥ 290 pg/mL, and ≥ 209 pg/mL cutoff values used at Cornell University, Texas A&M University and identified in this study. Of the 23 cats with [CBL] ≥ 209 pg/mL and [MMA] ≥ 1343 nmol/L, 13 were from the CUHA cohort (n = 55) and 10 were from the AHDL cohort (n = 60, P = .4; Fig 1).

Table 3. Discordant and concordant methylmalonic acid [MMA] and cobalamin [CBL] concentrations used to define subnormal and normal CBL status in 114 cats
  Serum [CBL] (pg/mL)
≥ 900< 900Total≥ 290< 290Total≥ 209< 209Total
  1. Bold numbers highlight discordant results.

  2. CBL values represent the reference threshold concentrations for cobalamin at Cornell University (900 pg/mL), Texas A&M University (290 pg/mL) and that observed in this study (209 pg/mL).

Serum [MMA] (nmol/L)≥ 1,343 11 3546 20 2646 23 2346
< 1,34338 30 6860 8 6865 3 68

Of the 65 cats with subnormal [CBL] (defined as [CBL] < 900 pg/mL), 30 (45%) had normal [MMA] (< 1,343 nmol/L). Using the more stringent cutoff of [CBL] < 290 pg/mL, 8 of 34 (23%) cats had [MMA]< 1,343 nmol/L.

Similar to the overall sample population, discordant [MMA] and [CBL] also were identified in the subset of 55 cats undergoing clinical evaluation at CUHA (Table 4). Of these cats, 20 had [MMA] ≥ 1,343 nmol/L; 7 of 20 (35%) had [CBL] ≥ 900 pg/mL and 14 of 20 (65%) had [CBL] ≥ 209 pg/mL. Definitive diagnoses in these cats included hepatobiliary disease (n = 8), IBD (n = 6), undefined small intestinal disease (n = 1), diabetic ketoacidosis (n = 1), and severe gingivitis (n = 1). Concurrent intestinal and hepatobiliary disease was present in 4 cats, concurrent IBD and iatrogenic hypothyroidism in 1 cat, concurrent chronic kidney disease and iatrogenic hypothyroidism in 1 cat, and concurrent cardiac, renal, and neurological disease in another. Although decreased renal function is associated with similar discordance in people,[12] we found only 2 cats with discordant results that were marginally azotemic (serum creatinine concentrations, 2.3 and 2.4 mg/dL).

Table 4. Discordant and concordant methylmalonic acid [MMA] and cobalamin [CBL] concentrations used to define subnormal and normal cobalamin status in 55 cats presenting for evaluation of disease
  Serum [CBL] (pg/mL)
≥ 900 pg/mL< 900 pg/mLTotal≥ 209 pg/mL< 209 pg/mLTotal
  1. Bold numbers highlight discordant results.

  2. CBL values represent the reference threshold concentrations for cobalamin at Cornell University (900 pg/mL) and that observed in this study (209 pg/mL).

[MMA] (nmol/L)≥ 1,343 nmol/L 7 1320 14 620
< 1,343 nmol/L21 14 3535 0 35

Conversely, 14/35 cats with [MMA] < 1,343 nmol/L had [CBL] < 900 pg/mL, but no cats had [CBL]< 209 pg/mL.

Relationship of Serum CBL or MMA Concentrations and HCT, MCV, Phosphorus, or Creatinine

Data from 72 cats were available to examine the relationships of CBL with HCT, MCV, and serum phosphorus concentration. Serum [CBL] weakly correlated negatively with MCV (τ = −0.199, P = .013) but not with HCT (P = .12) or serum phosphorus concentration (P = .7). Cats with low [CBL] (< 209 pg/mL) had a higher MCV (median, 50 fL; range, 46–57 fL) than cats without CBL deficiency (median, 47 fL; range, 39–58 fL; P = .015), but had similar HCT and serum phosphorus concentrations (P = .07 and .32, respectively).

Data from 45 cats were available to examine the relationships of MMA, HCT, MCV, phosphorus, and creatinine. [MMA] correlated with HCT (τ = −0.28, P = .006), but not MCV (P = .62) or serum phosphorus concentrations (P = .13). Cats with high [MMA] had a lower HCT (median, 31%; range, 23–43%) than cats with a normal [MMA] (median, 39%; range, 21–51%; P = .007), but had a similar MCV and serum phosphorus concentrations (P = .7 and .07, respectively). We found no correlation between serum creatinine concentration and MMA (τ = 0.119, P = .22).

Relationship of Serum CBL or MMA Concentrations to Methionine and Cysteine

Methionine or cysteine concentrations did not correlate with [CBL] (n = 127, P = .4, and P = .42, respectively) or [MMA] (n = 114, P = .05, and P = .07, respectively). Methionine, cysteine, and homocysteine concentrations in cats in this study were 103 μmol/L (range, 17–2,154), 174 μmol/L (range, 18–1,537) and 11.8 μmol/L (range, 4.6–66.4), respectively.

The relationships of clinical disease findings to CBL status, and specifically, of gastrointestinal disease to CBL status are presented as supplemental online data.


Our study provides novel information about CBL status in cats as determined by measurement of [CBL] and [MMA] in a relatively large cohort of cats, especially those with suspected or confirmed gastrointestinal disease. Our finding that [CBL] showed a significant but modest negative correlation with [MMA] supports the relationship previously observed between CBL and MMA in cats.[6] However, our study calls into question the use of [CBL] as a sole indicator of CBL deficiency in sick cats, and highlights a group of cats with discordant results, in which [MMA] is high (supporting CBL deficiency), but [CBL] is above the lower limit of the reference interval (indicating adequate CBL status). Clear explanations for this discordance were not apparent, but could reflect changes in the availability and cellular uptake of CBL associated with discordance in people.[12, 16, 17] Additional studies are needed to study this relationship in more detail in cats. Given that treatment of CBL deficiency can provide measurable therapeutic benefits to cats with gastrointestinal disease,[14] our study illustrates the complicated nature of assessing CBL status, where multiple methodologies and reference intervals exist, and do not necessarily agree. Clinicians should consider measuring MMA and CBL concurrently to better assess CBL status in cats, similar to recommendations in people.[18]

We found that a [CBL] of < 189 pg/mL could reliably identify cats with increased [MMA], with no false positive diagnoses below this [CBL], and that an [MMA] ≥ 1,343 nmol/L was a better indicator of CBL deficiency in the study population than the historical [MMA] of > 867 nmol/L,[6] because it provided fewer false negative results without increasing the rate of false positive results. Using ROC analysis, we found that a [CBL] of 209 pg/mL most accurately discriminated cats with high and low [MMA], when measured on the Immulite 1000 system. However, we recognize that ROC values typically are optimized on the basis of combined sensitivity and specificity, although exceptions exist in specific circumstances. Because CBL therapy is considered to have minimal risk, it could be argued that increasing the threshold for [CBL] above 290 pg/mL could provide clinical benefit to a greater number of cats. Nevertheless, when we examined the predictive values of various [CBL] at various disease prevalences with a [MMA] < 1,343 nmol/L indicating normal CBL status, we found that increasing the [CBL] cutoff to 900 pg/mL lowered the positive predictive value considerably, without impacting the negative predictive value (Table 2). The specific threshold values (189, 209, and 1,343 nmol/L) we identified in this study are based on the results obtained from our sample population. Whether decreasing the precision of our estimates, by rounding the threshold values to the nearest multiple of ten or hundred would produce similar outcomes remains to be determined.

Therefore, on the basis of our results, we formulated a strategy for evaluating cats with suspected CBL deficiency (Fig 3). In this strategy, cats with [CBL] ≤ 209 pg/mL would receive parenteral CBL supplementation and subsequent monitoring of response. Cats with [CBL] > 209 pg/mL would need to have [MMA] measured. Those with [MMA] < 1,343 nmol/L would be monitored, with periodic assessment of [CBL], whereas those with [MMA] ≥ 1,343 nmol/L would receive parenteral CBL supplementation and monitoring of clinical response and [MMA].

Figure 3.

Treatment decision strategy based on serum cobalamin [CBL] and serum methylmalonic acid [MMA] concentrations.

Similar to previous studies,[13] we observed a high prevalence of increased [MMA] or low [CBL] in cats presenting with signs of gastrointestinal disease (eg, vomiting, diarrhea, weight loss, anorexia, icterus, palpably thickened intestines). Of the cats in which a diagnosis was ultimately made, those with alimentary lymphoma had lower [CBL] than cats with IBD or other diseases. This observation agrees with previous studies of alimentary lymphoma in cats,[8] in which investigators also observed an association between CBL status and clinical outcomes. However, whether CBL supplementation of cats with alimentary lymphoma would alter outcome in these patients remains unknown.

Of particular interest were cats with discordant [MMA] and [CBL] results, particularly those cats with high [MMA] and normal [CBL]. Of the 23 cats identified with discordant results, clinical records were available for 13 cats. In the 10 cats without clinical records, we cannot speculate about the potential explanations for the discordance. In the 13 cats with medical records, possible explanations for such discordance included hepatobiliary disease (8 cats), diabetes mellitus (1 cat), and renal disease (2 cats), which all have been identified as factors in human patients.[12, 16, 17] Two discordant cats had a history of hyperthyroidism, which recently has been associated with low [CBL] in cats[19]; both of these cats had been treated before measuring [CBL] and [MMA], 1 had iatrogenic hypothyroidism after radioiodine treatment and 1 was euthyroid on methimazole. Because subnormal [CBL] persists in people after treatment of hyperthyroidism, the relationship of hyperthyroidism to CBL status is unclear.[20] The presence of concurrent diseases in many cats with discordant results made it difficult to determine the mechanisms underlying discordance (eg, concurrent intestinal and hepatobiliary disease, concurrent chronic kidney disease, iatrogenic hypothyroidism). The inability of [CBL] to reliably predict increased [MMA] in sick cats with [CBL] > 209 pg/mL coupled with the limited availability of MMA assays, suggest that routine administration of parenteral CBL to all cats considered at risk for CBL deficiency is the optimal solution. However, given that in humans, mechanisms associated with discordant [MMA] are attributable to alterations in CBL distribution and decreased cellular CBL uptake, rather than malabsorption, parenteral administration might fail to overcome this problem.[12, 16, 17] Monitoring [MMA] and clinical response after CBL supplementation of cats with discordant results might help resolve this dilemma.[18]

We found significant correlations of [CBL] with MCV and [MMA] with HCT that parallel findings in people in whom macrocytosis correlates with [CBL] and is considered the forerunner of the anemia of clinical CBL deficiency.[12]

In contrast to a previous study, we found no relationship of [MMA] or [CBL] with serum phosphorus concentration.[7] However, in that study, most cats had both low folate concentrations and low [CBL], whereas in our study only 5 cats met these criteria. Of these 5 cats, only 1 cat had serum phosphorus concentration measured and it was within reference interval for our laboratory. Additionally, in contrast to a previous study, we also found no relationship of cysteine or methionine with [CBL].[13] The reason for these differences in the 2 studies remains unknown.

In the first study to examine CBL deficiency in cats, which used an RIA that correlated with the bioassay, 49/80 (61%) of cats had a [CBL] below the lower limit of the reference interval (< 900 pg/mL).[5] Conversely, in a recent British study, using an automated chemiluminescence assay, only 11/39 cats (28.2%) had [CBL] below the lower limit of the reference interval (< 290 pg/mL).[9] Applying the 290 pg/mL cutoff to the original study yielded a prevalence of 27% (21/78), suggesting that differences in reference intervals among laboratories impact the identification of cats with subnormal [CBL] and CBL deficiency, similar to observations in human patients.[12, 21] In our study, we used the Immulite 1000 system to measure [CBL] because the RIA methodology used for previous studies is no longer available. Our clinical laboratory validated the assay by a direct comparison with the RIA and found that the Immulite 1000 system showed high correlation with RIA. This allowed us a direct comparison of data from several previous studies that also used the Immulite 1000 system to examine [CBL] in cats. The initial report of subnormal [CBL] in 80 cats, when measured by the RIA, reported a prevalence of subnormal [CBL] of 60% with the 900 pg/mL cutoff and 27% with the 290 pg/mL cut-off.[5] This concurs with our findings in this study using the Immulite 1000 assay that yielded prevalences of 57% and 29% using the same cutoff values, respectively, and a prevalence of 28.2% in 39 British cats using the 290 pg/mL cutoff and Immulite 1000 assay.[9] These observations infer that differences in reference intervals, rather than assay performance, underlie the variability in prevalence of CBL deficiency reported in cats with gastrointestinal disease.[5, 7, 9] In 50 healthy cats in Australia [CBL] determined with the Immulite 1000 system assay ranged from 345 to 3,668 pg/mL with median values of 1,609 pg/mL in cats aged 0–5 years and 1,200 pg/mL in cats older than 5 years (P = .39). Notably, 75% of cats had [CBL] > 1,000 pg/mL.[22] These ranges are substantially higher than those reported by Ruaux et al using the Immulite 1000 assay in 24 control cats that ranged from 600 to 1100 pg/mL.[6] It is not clear why there is such a difference in the reference ranges among different studies that employ the same CBL analysis method. It could reflect geographic variability or factors such as diet, presence of subclinical disease, and potentially the age of reference population. Similar issues of variable reference intervals have been the focus of much discussion in humans, where the reference intervals can dramatically affect perceived prevalence rates of CBL deficiency.[12, 21]


Our study has several limitations inherent with observational studies. Not all patients had complete clinical and biochemical data available; however, we evaluated a relatively large cohort of cats for each statistical association. We cannot discount the possibility of unidentified CBL supplementation before sampling in the samples sent in by first opinion clinicians. However, the populations were similar with regard to [CBL] and [MMA]. In some cats, no definitive diagnosis (as defined by histopathology) was obtained; however, these cats were presumptively diagnosed and treated based on the clinician's best evidence and judgment. Cats without a diagnosis were coded “open.” We did not evaluate therapeutic interventions.

We elected to use [MMA] as the reference standard to define CBL status in our cats, based on human medical and veterinary historical precedents. Our data suggest that [MMA] might not entirely accurately reflect CBL status in all cats (similar to observations in humans). However, this does not invalidate the use of [MMA] as the reference standard, until a better indicator of CBL status emerges.


Our study demonstrates the difficulties of accurately assessing the CBL status of sick cats. We observed increased [MMA] more frequently than subnormal [CBL] and found that in our population [MMA] ≥ 1,343 nmol/L and [CBL] < 209 pg/mL most accurately predicted CBL deficiency. We found correlations between [CBL] and MCV and [MMA] and anemia that might help identify clinical deficiency, and observed that cats with alimentary lymphoma had lower [CBL] than cats with IBD. However, there was a considerable degree of discordance between [CBL] and [MMA], suggesting that concurrent measurement of [MMA] and [CBL] will be more informative in identifying patients with potential CBL deficiency. Our data suggest that all sick cats with [CBL] < 209 pg/mL would benefit from CBL supplementation. However, in sick cats with [CBL] > 209 pg/mL, MMA assessment might help identify cats that could benefit from CBL supplementation. Although CBL supplementation of all cats with gastrointestinal diseases is inexpensive and safe, whether it would alter CBL status in all cats with apparent CBL deficiency (as assessed by [MMA]) is unknown.


The authors thank Steve Lamb and the staff of the Section of Endocrinology at the Cornell University Diagnostic Laboratory for assistance with assay validation, and Hollis Erb for statistical advice.

Grant: Supported by a grant from the Cornell Feline Health Center.

Conflict of Interest Declaration: Authors disclose no conflict of interest.


  1. 1

    ADVIA 120 Hematology Analyzer; Bayer HealthCare Diagnostics Division, Tarrytown, NY

  2. 2

    Roche Hitachi 917 Chemistry Analyzer; Roche Diagnostics, Indianapolis, IN

  3. 3

    Immulite 1000 Immunoassay System; Siemens Healthcare Diagnostics Deerfield, IL

  4. 4

    Thermo Finnigan GC/MS; Thermo Finnigan LLC, Austin, TX

  5. 5

    MedCalc ver.; MedCalc Software bvba, Mariakerke, Belgium