Introduction

Hypoadrenocorticism in dogs is an uncommon endocrinopathy, with population prevalences up to 0.09% (166/191,434)^3,4 and increasing up to 4% (6/151)⁵ in dogs with signs of chronic gastrointestinal disease. Three previous studies^1,2,6 have assessed the utility of measuring resting serum or plasma cortisol concentrations as a screening test for hypoadrenocorticism, showing that resting serum or plasma cortisol concentration of 2 μg/dL (55 nmol/L) demonstrates excellent sensitivity (99.4% to 100%) and thus can be used to exclude hypoadrenocorticism.

Resting serum or plasma cortisol concentrations in the aforementioned studies^1,2,6 were measured using either a commercially available radioimmunoassay (Coat-a-Count Cortisol; Siemens Medical Solutions Diagnostics) or chemiluminescent assays (Immulite 1000 or 2000; Siemens Healthcare Diagnostics Ltd). Those systems are designed for use in reference laboratories. Radioimmunoassays have now been largely replaced by nonisotopic immunoassays.⁷ The Immulite 1000 and 2000 cortisol immunoassays are solid-phase, competitive chemiluminescent immunoassays. The test unit is composed of a solid phase coated with polyclonal rabbit anti-cortisol antibody and a liquid phase that contains cortisol conjugated to alkaline phosphatase. Cortisol in the test sample competes with alkaline phosphatase-conjugated cortisol for antibody binding sites on the solid phase. A chemiluminescent substrate is then added to evaluate the concentration of cortisol in the test sample.

An alternative benchtop competitive fluorescent enzyme immunoassay analyzer (AIA-360; Tosoh Bioscience Inc) is available for in-house use and can measure cortisol concentrations using canine serum or plasma. Similarly to the described chemiluminescent immunoassay, cortisol in the test sample competes with enzyme-labeled cortisol for binding sites on cortisol-specific antibodies, which are immobilized on magnetic beads. The magnetic beads are then incubated with the fluorogenic substrate 4-methylumbelliferyl phosphate. The amount of bound enzyme-labeled cortisol is inversely proportional to the cortisol concentration within the test sample.

This benchtop immunoassay analyzer has been shown to have an intra- and interassay coefficient of variability of 2.6%⁸ to 3% (ST AIA PACK-CORT; Tosoh Bioscience Inc) and 4%,⁸ respectively, for measurement of cortisol concentration, reflecting that this immunoassay has good precision. Results for serum cortisol concentrations measured with the AIA-360 have also been shown to strongly correlate (Spearman rank correlation coefficient of 0.97) with those measured with the Immulite 1000 cortisol immunoassay.⁸ However, the presence of a negative bias indicated that cortisol concentrations determined by each analyzer could not be used interchangeably.⁸ Consequently, further research is warranted to assess whether the serum or plasma cortisol cutoff points derived from previous studies are applicable to other cortisol immunoassays such as that used by the AIA-360.

The objective of the present study was to assess the diagnostic performance of the AIA-360 cortisol immunoassay as a screening test for hypoadrenocorticism in dogs. A secondary objective was to identify an optimal serum cortisol cutoff point applicable to the AIA-360, which could be reliably used to exclude hypoadrenocorticism in dogs. The null hypothesis was that previously derived cortisol cutoff points^1,2,6 are applicable to the AIA-360 cortisol immunoassay and new cutoff points are not required. The alternative hypothesis was that analyzer-specific cortisol cutoff points are required when using the AIA-360 cortisol immunoassay.

Materials and Methods

Medical records of all dogs that underwent an ACTH stimulation test at 2 UK-based referral centers between June 2015 and October 2019 were reviewed retrospectively. Dogs were included in the study if an ACTH stimulation test was performed and serum cortisol was measured using the AIA-360 fluorescent enzyme immunoassay analyzer. The standard protocol at both centers when performing an ACTH stimulation test was to measure resting serum cortisol concentrations and administer tetracosactide (not available in the US; 5 μg/kg, IV or IM),⁹ followed by measuring serum cortisol concentrations 60 minutes after the injection. Venous blood samples were obtained and placed into plain tubes to allow serum to be obtained for analysis. All serum cortisol concentrations were analyzed at onsite laboratories. The time from onset of clinical signs to assessment of serum cortisol concentration was not retrievable from the clinical records.

A diagnosis of hypoadrenocorticism was made if either the resting serum cortisol concentration was < 1 μg/dL and increased by < 5% following ACTH stimulation or if both the pre– and post–ACTH-stimulated serum cortisol concentrations were < 1 μg/dL. A value of 5% was used because a previous study⁸ assessing the precision of the AIA-360 cortisol immunoassay identified an average analyzer coefficient of variability of < 5%. In 2021 following data collection, the ALIVE Project released proposed criteria for the diagnosis of primary hypoadrenocorticism.¹⁰ These criteria stipulated that the pre– and post–ACTH-stimulated cortisol concentrations needed to be within or less than the lower quartile of the reference interval for the diagnosis of primary hypoadrenocorticism.¹⁰ The reference interval for resting serum or plasma cortisol concentration using the AIA-360 cortisol immunoassay is 1.09 to 8.7 μg/dL (30 to 240 nmol/L). All sampled dogs met both the original and ALIVE criteria; subsequently, the ALIVE criteria were used for the diagnosis of hypoadrenocorticism. The lower limit of detection reported by the AIA-360 analyzer was 0.2 μg/dL (5.5 nmol/L). For the purpose of statistical analysis, a serum cortisol concentration reported to be < 0.2 μg/dL (5.5 nmol/L) was adjusted to 0 μg/dL (0 nmol/L).

Dogs were excluded if any of the following criteria were met: an ACTH stimulation test was performed on the basis of a suspicion of hypercortisolism or as part of investigations into the functionality of an adrenal nodule or mass; if any dogs were receiving treatment or had previously received treatment for hypercortisolism; if dogs had received medication known to affect the pituitary-adrenal axis in the 4 weeks^1,2 preceding ACTH stimulation testing, including azole antifungal medication, progesterone, or glucocorticoids via any route; and if the ACTH stimulation test was performed with the use of an analyzer other than an AIA-360. Although there are studies^11,12 reporting duration of adrenocortical suppression following treatment with varying formulations of corticosteroids, consensus guidelines regarding the time required for normal adrenal function to return following treatment with glucocorticoids are not available,¹³ and thus the time period of 4 weeks was used on the basis of previous studies^1,2 assessing the utility of resting serum or plasma cortisol concentrations in dogs with hypoadrenocorticism.

Data collected from the medical records in the 4 weeks prior to ACTH stimulation testing included signalment and clinical signs. The following signalment information was documented: age, breed, sex, neuter status, and body weight. Whether vomiting, diarrhea, signs of abdominal pain, collapse, anorexia, weakness, polyuria, or polydipsia, alone or in combination, was reported was also obtained from either the hospital records or the referring veterinarian. If these signs were not documented in the medical records, it was assumed that they were not a presenting complaint. Resting and post-ACTH stimulation serum cortisol concentrations (μg/dL) measured with the AIA-360 fluorescent enzyme immunoassay analyzer were also recorded. Serum or plasma sodium (Na⁺; mmol/L), potassium (K⁺; mmol/L), total protein (g/L), and albumin (g/L) concentrations were also recorded when available.

At the time of study design, dogs were classified as having typical hypoadrenocorticism if their serum or plasma sodium concentration was below the reference interval, serum or plasma potassium concentration was above the reference interval, or both. Dogs were classified as having atypical hypoadrenocorticism if they lacked the aforementioned electrolyte abnormalities or had hypokalemia or hypernatremia. Following the release of the ALIVE project definitions for hypoadrenocorticism¹⁴ during data analysis, dogs who were originally classified as having typical hypoadrenocorticism were re-classified as having hyponatremic and or hyperkalemic hypoadrenocorticism (HA_T) and dogs with atypical hypoadrenocorticism were classified as having eunatremic, eukalemic hypoadrenocorticism (HA_A).

Results

Study population

A total of 998 dogs underwent an ACTH stimulation test between June 2015 and October 2019. There were 173 dogs that met the inclusion criteria; 102 dogs at Wear Referrals Veterinary Hospital and 71 dogs at Pride Veterinary Hospital (Figure 1). Of these 173 dogs, 20 (12%) were diagnosed with hypoadrenocorticism and 153 (88%) were diagnosed with nonadrenal illness, giving a period prevalence of 12% (95% CI, 7% to 16%).

View Full Size

Flow diagram indicating the inclusion and further subgroupings of 173 client-owned dogs that underwent an ACTH stimulation test with the use of a benchtop fluorescent enzyme immunoassay analyser (AIA-360; Tosoh Bioscience Inc) between June 2015 and October 2019. HA = hypoadrenocorticism. HAA = eunatremic, eukalemic hypoadrenocorticism. HAT = hyponatremic and, or hyperkalemic hypoadrenocorticism.

Citation: Journal of the American Veterinary Medical Association 261, 1; 10.2460/javma.22.08.0346

• Download Figure
• Download figure as PowerPoint slide

Dogs in the hypoadrenocorticism group consisted of 7 crossbred dogs, 3 Jack Russell Terriers, 2 Labrador Retrievers, and 1 each of West Highland White Terrier, Boxer, Tibetan Terrier, Airedale Terrier, English Springer Spaniel, Cavalier King Charles Spaniel, Chihuahua, and Yorkshire Terrier. Breed data in the nonadrenal illness group were available for 152 of the 153 (99%) dogs, which were reported as crossbred dogs (n = 27), Labrador Retrievers (13), West Highland White Terriers (9), Shih Tzu (8), Golden Retrievers (8), German Shepherd Dogs (7), Border Collies (5), Border Terriers (5), Cocker Spaniels (4), Schnauzers (4), French Bulldogs (4), Lhasa Apso (3), Tibetan Terriers (3), Chihuahuas (3), Cavalier King Charles Spaniels (3), Dachshunds (3), Hungarian Vizslas (2), Staffordshire Bull Terriers (2), Poodles (2), Pugs (2), Jack Russell Terriers (2), Rough Collies (2), Beagles (2), Boxers (2), Doberman Pinschers (2), Dogue de Bordeaux (2), English Springer Spaniels (2), and 1 each of Basset Hound, Bedlington Terrier, Bichon Frise, British Bulldog, Dalmatian, English Pointer, Fox Terrier, Greyhound, Lurcher, Maltese, Papillon, Patterdale Terrier, Rottweiler, Samoyed, Shetland Sheepdog, Siberian Husky, Smooth Collie, Saint Bernard, Weimaraner, Whippet, and Yorkshire Terrier.

Group characteristics

Results for group characteristics, clinical signs, and serum or plasma electrolytes and protein concentrations were compared for dogs with hypoadrenocorticism versus nonadrenal illness. The median age and body weight did not differ substantially between dogs with hypoadrenocorticism (4.75 years [IQR, 3.13 to 8.02 years] and 14.2 kg [IQR, 8.05 to 25.62 kg]) versus nonadrenal illness (4.17 years [IQR, 1.83 to 7.5 years] and 14.8 kg [IQR, 9.00 to 25.80 kg]). Similarly, the groups had no meaningful difference on the basis of sex (Table 1).

Table 1

Numbers (percentages) of 173 client-owned dogs that underwent an ACTH stimulation test between June 2015 and October 2019 and had various characteristics, stratified on the basis of whether dogs had hypoadrenocorticism (HA; n = 20) versus nonadrenal illness (NAI; 153).

Variable	No. (%) of dogs with HA	No. (%) of dogs with NAI	P value
Female	11 (55)	71 (46)	.469
Male	9 (45)	82 (54)	.469
Vomiting	15 (75)	95 (62)	.328
Diarrhea	10 (50)	59 (39)	.326
Abdominal pain	8 (40)	39 (26)	.170
Anorexia	13 (65)	59 (39)	.024
Weakness and collapse	7 (35)	15 (9.8)	.001
Polyuria, polydipsia, or both	4 (20)	26 (17)	.755

Anorexia was reported for 13 of the 20 (65%) dogs with hypoadrenocorticism, compared to 59 of the 153 (39%) dogs with nonadrenal illness. Weakness and collapse were documented for 7 of the 20 (35%) dogs with hypoadrenocorticism and 15 of the 153 (9.8%) dogs with nonadrenal illness. Anorexia (P = .024) and weakness and collapse (P = .001) were significantly more common in dogs with hypoadrenocorticism compared to nonadrenal illness (Table 1). There was no statistical difference in the other documented clinical signs.

Serum or plasma sodium concentration was available for all 20 dogs with hypoadrenocorticism and for 131 of 153 (86%) dogs with nonadrenal illness. Median serum or plasma sodium concentration was significantly (P < .001) lower for dogs with hypoadrenocorticism (141 mmol/L; IQR, 132 to 146 mmol/L), compared to nonadrenal illness (147 mmol/L; IQR, 145 to 149 mmol/L). Serum or plasma potassium concentrations were available for all 20 dogs with hypoadrenocorticism and 131 of the 153 (86%) dogs with nonadrenal illness. Median serum or plasma potassium concentration was significantly (P < .001) higher for dogs with hypoadrenocorticism (5.5 mmol/L; IQR, 4.70 to 6.70 mmol/L), compared to nonadrenal illness (4.1 mmol/L; IQR, 3.90 to 4.50 mmol/L). Serum or plasma albumin concentration was available for 19 of the 20 (95%) dogs with hypoadrenocorticism and 134 of the 153 (88%) dogs with nonadrenal illness. Median serum or plasma albumin concentration was significantly (P = .015) lower for dogs with hypoadrenocorticism (24 g/L; IQR, 21 to 33 g/L), compared to nonadrenal illness (31 g/L; IQR, 28 to 34 g/L). Median serum or plasma total protein concentration did not meaningfully differ between dogs with hypoadrenocorticism (60 g/L; IQR, 47 to 67 g/L) versus nonadrenal illness (62 g/L; IQR, 56 to 68 g/L).

Nine of the 20 (45%) dogs with hypoadrenocorticism lacked either hyponatremia or hyperkalemia and thus were classified as having HA_A. Eleven of the 20 (55%) dogs with hypoadrenocorticism had hyponatremia, hyperkalemia, or both and were thus classified as having HA_T. There was no significant difference in the age, sex, or body weight of dogs diagnosed with HA_T versus HA_A.

Median resting serum cortisol concentration was significantly (P < .001) lower for dogs with hypoadrenocorticism (0.58 μg/dL [16 nmol/L]; IQR, 0 to 0.58 μg/dL [0 to 16 nmol/L]; range, 0 to 0.58 μg/dL [0 to 16 nmol/L]), compared to dogs with nonadrenal illness (2 μg/dL [55 nmol/L]; IQR, 1.27 to 3.59 μg/dL [35 to 99 nmol/L]; range, 0 to 27.4 μg/dL [0 to 757 nmol/L]). Similarly, the median post-ACTH stimulation serum cortisol concentration was significantly (P < .001) lower for dogs with hypoadrenocorticism (0.58 μg/dL [16 nmol/L]; IQR, 0 to 0.58 μg/dL [0 to 16 nmol/L]) versus dogs with nonadrenal illness (11.8 μg/dL [326 nmol/L]; IQR, 9.2 to 14.79 μg/dL [254 to 408 nmol/L]). The median resting serum cortisol concentration did not meaningfully (P = .152) differ for dogs with HA_T (0 μg/dL [0 nmol/L]; IQR, 0 to 0.58 μg/dL [0 to 16 nmol/L]) versus HA_A (0.58μg/dL [16 nmol/L]; IQR, 0.29 to 0.58 μg/dL [8 to 16 nmol/L]).

A resting serum cortisol concentration of 2μg/dL (55nmol/L; Table 2), measured using the AIA-360 fluorescent enzyme immunoassay analyser, had a sensitivity of 100% (95% CI, 80% to 100%) and a specificity of 50% (95% CI, 42% to 58%) when used as a screening test for hypoadrenocorticism in dogs. At a cortisol concentration of 0.58 μg/dL (16 nmol/L), the analyzer similarly had a sensitivity of 100% but a specificity of 97% (95% CI, 92% to 99%).

Table 2

Calculated sensitivities, specificities, and positive (LR [+]) and negative (LR [–]) likelihood ratios for various serum or plasma resting cortisol concentrations measurements obtained with a benchtop fluorescent enzyme immunoassay analyzer (AIA-360; Tosoh Bioscience Inc) for the dogs described in Table 1.

Resting cortisol concentration (μg/dL)	% Sensitivity (95% CI)	% Specificity (95% CI)	LR (+)	LR (–)
2 μg/dL	100 (80–100)	50 (42–58)	1.99 (1.7–2.3)	0.0 (0.0–0.0)
1 μg/dL	100 (80–100)	88 (81–92)	8.1 (5.3–12.3)	0.0 (0.0–0.0)
0.8 μg/dL	100 (80–100)	94 (88–97)	15.3 (8.4–27.9)	0.0 (0.0–0.0)
0.58 μg/dL	100 (80–100)	97 (92–99)	30.6 (12.9–72.5)	0.0 (0.0–0.0)

The LR (+) and LR (–) were calculated at various cutoff points. A resting serum cortisol concentration of 2 μg/dL (55 nmol/L) had an LR (+) of 1.99 (95% CI, 1.7 to 2.3) and LR (–) of 0.0 (95% CI, 0.0 to 0.0; Table 2). A resting serum cortisol concentration of 0.58 μg/dL (16 nmol/L) had an LR (+) of 30.6 (95% CI, 12.9 to 72.5) and LR (–) of 0.0 (95% CI, 0.0 to 0.0).

The high diagnostic performance of the AIA-360 cortisol immunoassay is represented in a ROC curve with an area under the ROC curve of 0.98 (Figure 2). On the basis of the sample population, a serum cortisol concentration of 0.58 μg/dL (16 nmol/L) was deemed the optimal cutoff point for the exclusion of hypoadrenocorticism when screening dogs for hypoadrenocorticism.

View Full Size

Receiver operating characteristic curve demonstrating the diagnostic performance of the AIA-360 cortisol immunoassay as a screening test for hypoadrenocorticism in the 173 dogs described in Figure 1, with no dogs censored. The area under the receiver operating characteristic curve was 0.98.

Citation: Journal of the American Veterinary Medical Association 261, 1; 10.2460/javma.22.08.0346

• Download Figure
• Download figure as PowerPoint slide

Five of the 173 (0.03%) dogs had a resting serum cortisol concentration reported to be “less than 16 nmol/L” in the medical records. Subgroup analysis was performed, and when these dogs in subgroup A were censored from evaluation, the optimal resting serum cortisol cutoff point was 0.58 μg/dL (16 nmol/L) with a sensitivity and specificity of 100% and 97%, respectively. Furthermore, 11 of the 173 (0.06%) dogs had resting serum cortisol concentrations reported to be either “less than 16 nmol/L” or “16 nmol/L” in the medical records. When all 11 of these dogs were censored from evaluation, the optimal resting serum cortisol cutoff point was 0 μg/dL (0 nmol/L) with a sensitivity of 100% (95% CI, 63% to 100%) and a specificity of 99% (95% CI, 95% to 100%), consequently, only marginally affecting the results of the study.

Discussion

The main aim of the present study was to assess the diagnostic performance of the AIA-360 cortisol immunoassay as a screening test for hypoadrenocorticism in dogs and to calculate a cortisol concentration cutoff point that would assess the overall likelihood of hypoadrenocorticism. Although a previous study⁸ showed that serum cortisol concentrations measured by the AIA-360 correlate well with the Immulite, results cannot be used interchangeably. Consequently, previously calculated cutoff points used to exclude hypoadrenocorticism in dogs needed reevaluation for use with the AIA-360 analyzer.

Our findings confirmed that the serum or plasma cortisol concentration cutoff point of 2 μg/dL (55 nmol/L) often used as a screening test for hypoadrenocorticism when using the Immulite chemiluminescent assays is also applicable when serum cortisol is measured using the AIA-360 cortisol immunoassay. In this study, excellent sensitivities were maintained at serum cortisol concentrations as low as 0.58 μg/dL (16 nmol/L). Thus, in a similar population, if using the AIA-360 cortisol immunoassay, an ACTH stimulation test may not need to be performed in dogs with a resting serum cortisol concentration > 0.58 μg/dL (16 nmol/L) to exclude hypoadrenocorticism. However, a small proportion (3/153) of dogs in the nonadrenal illness group had serum resting cortisol concentrations < 0.58 μg/dL (16 nmol/L), leading to an overlap in resting serum cortisol concentrations in dogs with hypoadrenocorticism versus nonadrenal illness. We therefore still recommend that an ACTH stimulation test be performed prior to confirmation of hypoadrenocorticism in dogs.

Both the sample prevalence (12% [20/173]) and 95% CI of the estimated population prevalence (7% to 16%) were much higher than the prevalence reported in dogs with chronic signs of gastrointestinal disease.^5,16 The greater prevalence of hypoadrenocorticism in this study was likely the result of our inclusion criteria positively selecting for dogs that underwent an ACTH stimulation test as opposed to including all dogs that presented to the aforementioned referral centers during the study period or only selecting a subpopulation of those that were investigated for gastrointestinal disease.

We included group characteristics in this study to allow readers to compare the study sample to their perceived population of dogs with hypoadrenocorticism, thus inferring the relevance of the results. Dogs with hypoadrenocorticism in the sample population had a significantly lower median serum albumin concentration compared to dogs with nonadrenal illness. In the sample population, 42% (8/19) of dogs with hypoadrenocorticism had hypoalbuminemia. A similar finding has been reported previously, with 38.6% (17/44) of dogs with hypoadrenocorticism having concurrent hypoalbuminemia in one study,¹⁷ alongside hypoalbuminemia being the most common biochemical abnormality in dogs with HA_A in another study.¹⁸ Nine of the 20 dogs (45%) in the sample population lacked the classic hallmarks of hypoadrenocorticism (hyperkalemia, hyponatremia, or both) and thus were classified as having HA_A. A previous study¹⁹ showed that 8 of 25 (32%) dogs with hypoadrenocorticism had electrolyte profiles within reference ranges. It was not clear why the prevalence of HA_A was greater for the dogs of the present study. However, the vast majority of dogs in the present study were from a referral population, and thus it was possible that the prevalence was greater because testing for HA_A is potentially performed more frequently in dogs with gastrointestinal signs as well as nonspecific signs such as lethargy and weakness in referral centers. It was also possible that HA_A was less well known to primary care clinicians, and thus affected animals may have been more likely to have been referred. Equally, it could not be confirmed that all of these dogs had primary hypoadrenocorticism, as endogenous ACTH concentrations were not routinely measured; thus, it was possible that a proportion of these dogs had naturally occurring, or iatrogenic, secondary hypoadrenocorticism. Alternatively, IV fluid therapy by the referring veterinarians prior to ACTH stimulation testing for some of these dogs may have altered their electrolyte concentrations, resulting in these dogs being falsely diagnosed with HA_A.

The main limitation of this study was that serum cortisol concentrations measured using the AIA-360 were not run concurrently on analyzers from which previous resting serum or plasma cortisol cutoff point concentrations were based,^1,2,6 and thus direct comparisons between values obtained from the different immunoassays could not be performed. However, this limitation was offset by the fact that a previous study⁸ showed excellent correlation between serum cortisol concentration measurements obtained from the AIA-360 cortisol immunoassay to the Immulite 1000 cortisol immunoassay. A negative bias at a variety of cortisol concentrations has been shown when comparing results from the AIA-360 cortisol immunoassay to the Immulite 1000 cortisol immunoassay, with cortisol concentration measurements being lower in 63%⁸ of samples. It could therefore be expected that the optimal cortisol cutoff point concentration obtained using the AIA-360 cortisol immunoassay is lower than that applicable to previously studied analyzers.^1,2,6 Furthermore, the lowest reportable value for the AIA-360 cortisol immunoassay is 0.2 μg/dL (5.5 nmol/L), compared to 1 μg/dL (2.76 nmol/L) for the Immulite 1000 and 2000. Therefore, it is conceivable that the resting serum cortisol concentrations and thus optimal cortisol cutoff point were lower in our sample population due to the fact that the AIA-360 cortisol immunoassay reports a wider range of cortisol concentrations.

Another limitation of the present study was the retrospective nature alongside the small hypoadrenocorticism sample size, which was reflected in the width of the 95% CIs. From a clinical perspective, although the sensitivity of the optimal serum cortisol cutoff point of 0.58 μg/dL was 100%, the 95% CI ranged from 80% to 100%; thus, readers must be cognizant that a dog with a resting serum cortisol concentration above this concentration could still have hypoadrenocorticism. Future prospective studies should be performed with calculated sample sizes, standardized history taking, ACTH stimulation testing, and sample collection and storage.

Another limitation of the present study was the presence of a subgroup of dogs (subgroup A) with resting serum cortisol concentrations documented as “less than 16 nmol/L” (“less than 0.58 μg/dL”) in the medical records from one of the referral centers. This was not the lower limit of detection of the analyzer, and thus it was not apparent why this was reported in some dogs while other dogs had exact numeric values. To further complicate matters, 6 of the 173 dogs of the present study had a resting serum cortisol concentration reported to be 0.58 μg/dL (16 nmol/L). It was not clear, given the retrospective nature of this study, whether these dogs truly had a resting serum cortisol concentration of 0.58 μg/dL (16 nmol/L) or whether they too had a resting serum cortisol concentration “less than 0.58 μg/dL” (“less than 16 nmol/L”) that was then documented in the medical records as 0.58 μg/dL (16 nmol/L). To try to compensate for this, we further stratified the population to allow subgroup analysis to be performed. When dogs in subgroup A were censored from analysis, the sensitivity, specificity, and optimal resting serum cortisol cutoff point remained the same. When all dogs with either a resting serum cortisol concentration of 0.58 μg/dL (16 nmol/L) or “less than 0.58 μg/dL” (“less than 16 nmol/L”) were censored from evaluation, the optimal cortisol cutoff point decreased to 0 μg/dL (0 nmol/L). Further studies are required with exact serum cortisol concentrations to assess whether the optimal cortisol cutoff point to exclude hypoadrenocorticism in dogs using the AIA-360 cortisol immunoassay may actually be lower than 0.58μg/dL (16nmol/L).

In conclusion, the AIA-360 cortisol immunoassay demonstrated excellent sensitivity for the exclusion of hypoadrenocorticism and the previously reported cutoff points^1,2,6 can be applied with high sensitivity. The results of the present study also suggested that an ACTH stimulation test need only be performed to assess the likelihood of hypoadrenocorticism if the resting serum cortisol concentration is ≤ 0.58 μg/dL (16 nmol/L) when using the AIA-360 cortisol immunoassay.