Elasmobranchs (sharks, skates, and rays) are an ancient and diverse group of cartilaginous fish that play essential ecologic roles in aquatic ecosystems globally. Over 200 stingray species are grouped in the order Myliobatiformes, suborder Myliobatoidei (1). Most stingrays are demersal and can be found in tropical to subtropical marine or freshwater habitats. Stingray species are increasingly listed as vulnerable or endangered on the International Union for Conservation of Nature Red List of Threatened Species, an unfortunate trend shared by all elasmobranchs, with almost 1/3 being threatened with extinction (2). Consequently, stingrays are common in public aquariums due to their ecological importance and increasing need for conservation.

One unique physiologic feature of elasmobranchs is their preferential use of ketone bodies—especially β-hydroxybutyrate (β-HB)—rather than fatty acids for aerobic metabolism under normal conditions, predominately in skeletal and cardiac muscle (3–5). Besides providing buoyancy, elasmobranchs’ large, lipid-dense livers serve as ketogenic powerhouses (4). Within the liver, acetyl coenzyme A is formed from long-chain fatty acids by mitochondrial β-oxidation and utilized in numerous metabolic reactions, including generating ketone bodies. Ketone bodies are highly water-soluble, energy-rich fuels exported to extra-hepatic tissues without the need for protein-binding (3). Furthermore, ketone bodies likely play a significant role in supporting metabolism during periods of fasting in elasmobranchs (3–6). While the concentration of ketone bodies in fed elasmobranchs is similar to that observed in fasted (0–3 mmol/L) or ketoacidotic (> 3 mmol/L) mammals, marked ketosis has been found during starvation or prolonged fasting (4).

Glucose is potentially of less metabolic importance in elasmobranchs, although differences between tissues, species, and metabolic states exist (3–5, 7, 8). Plasma glucose concentrations are often low relative to teleosts and terrestrial vertebrates and appear relatively unchanged after feeding or during long-term fasting (3). Some elasmobranch species reportedly tolerate wide ranges of plasma glucose concentrations with no observed clinical effects (3). Nevertheless, a stress-induced increase in catecholamines or corticosteroids (e.g., 1α-hydroxycorticosterone) has been speculated to possibly correlate with increased glucose mobilization, which may reflect increased demand as an oxidative fuel source or supply increased rates of anaerobic glycolysis (3–5, 9, 10). This is best highlighted by the hyperglycemia observed in elasmobranchs as a response to various stressors, including capture, handling, and transport (10–13).

Conventional measurement of glucose and β-HB concentrations in veterinary medicine requires serum or plasma to be submitted to a reference laboratory. As an alternative to laboratory-based assays in human and veterinary medicine, point-of-care (POC) meters typically provide rapid results, require smaller sample volumes, and require no centrifugation or cold chain before testing (14). Within veterinary medicine, the diagnostic accuracy of POC glucose and β-HB meters have been explored in multiple domestic (15–21) and non-domestic species (22–28). Point-of-care glucose and β-HB meters may aid veterinarians, researchers, and wildlife managers working with elasmobranchs, including stingrays, by providing almost instant insight into their physiologic stress or metabolic state.

This study evaluated the diagnostic agreement of three commercial POC glucose and β-HB meters with reference laboratory methods for glucose and β-HB concentrations in stingrays. We hypothesized that the results from the POC meters evaluated would have sufficient agreement with reference laboratory methods and precision for clinical and field research purposes for glucose and β-HB concentrations in stingrays.

Plasma glucose and β-HB concentrations, as determined by the reference laboratory, ranged from <20–63 mg/dL (median: 35 mg/dL) and 0.05–5.38 mmol/L (median: 0.22 mmol/L, Table 1), respectively. PCV, BC, and TS for all stingrays ranged from 11–32% (median: 24%), 0–4% (median: 0%), and 4.4–8.1 g/dL (median: 7.0 g/dL). Glucose concentration ranges as determined by the reference laboratory were similar across species: < 20–53 mg/dL (median: 26 mg/dL) for Atlantic stingrays, 32–59 mg/dL (median: 48 mg/dL) for cownose rays, 23–39 mg/dL (median: 36 mg/dL) for southern stingrays, and < 20–63 mg/dL (median: 30 mg/dL) for yellow stingrays. Similarly, β-HB concentration ranges as determined by the reference laboratory were similar across species: 0.1–0.4 mmol/L (median: 0.22 mmol/L) for Atlantic stingrays, 0.08–0.69 mmol/L (median: 0.31 mmol/L) for cownose rays, 0.14–0.28 mmol/L (median: 0.23 mmol/L) for southern stingrays, and 0.05–5.38 mmol/L (median: 0.22 mmol/L) for yellow stingrays with the maximum of the yellow stingray range being skewed by one individual (5.38 mmol/L). With the one apparent yellow stingray outlier removed, the β-HB concentration range was 0.05–0.52 mmol/L for conspecifics included in this study.

The CV or imprecision of the three POC meters for glucose was less when using plasma (maximum: 0–4.9%) than whole blood (maximum: 11%); similarly, the CV of the three POC meters for β-HB was less when using plasma (maximum: 0%) than whole blood (maximum: 12.5%). The CV was 0% for all three POC meters when measuring β-HB using plasma (Table 2). All three POC meters had less TE_obs for both whole blood and plasma when measuring β-HB than glucose (Table 2). The proportions of results with percent relative error < 12, 15, and 20% were tabulated for each POC meter for glucose and β-HB using whole blood and plasma (Table 2).

Bland–Altman methods found the BHBCheck meter had significant positive bias (i.e., overestimated) for both glucose and β-HB regardless of the matrix (Table 3; Figures 1, 2). The Precision Xtra meter had a significant positive bias for glucose using plasma and β-HB using whole blood (Table 3; Figures 1, 2). The Nova Vet™ had significant negative bias (i.e., underestimated) for glucose regardless of the matrix (Table 3; Figures 1, 2). For glucose, the mean bias ranged from −18 to 48 mg/dL using whole blood and from −14 to 69 mg/dL using plasma across the three POC meters (Table 3). For β-HB, the mean bias ranged from −0.1 to 0.2 mmol/L using whole blood and from 0 to 0.1 mmol/L using plasma across the three POC meters (Table 3).

Passing-Bablok regression analyses found that the BHBCheck meter had a significant constant bias for glucose regardless of matrix and β-HB using plasma and a significant proportional bias for glucose using plasma (Table 3; Figures 3, 4). The Precision Xtra meter had a significant constant bias for glucose using whole blood and β-HB using plasma, as well as a significant proportional bias for both glucose and β-HB regardless of the matrix (Table 3; Figures 3, 4). The Nova Vet™ meter had a significant constant bias for β-HB using plasma and a significant proportional bias for glucose using whole blood and β-HB using plasma (Table 3; Figures 3, 4).

Linear mixed-effect models found sex, weight, PCV, and BC had no significant effect on differences between the individual POC meters and the reference laboratory for glucose or β-HB. Matrix had no significant effect on differences between the BHBCheck or Nova Vet™ meters and the reference laboratory methods for glucose or β-HB. However, matrix did have a significant effect on β-HB, but not glucose, for the Precision Xtra meter: differences in β-HB concentrations were significantly lower using whole blood than plasma (matrix estimate ± standard error: −2.3 ± 0.71 mmol/L, p value = 0.002).

This study evaluated the diagnostic agreement (i.e., accuracy) between three commercial POC glucose and β-HB meters to reference laboratory methods using whole blood and plasma from four stingray species commonly found in managed care environments. Plasma glucose concentrations determined by a reference laboratory method were similar to previous reports in cownose rays and southern stingrays (40–42); likewise, β-HB concentrations determined by a reference laboratory method fell within previously reported ranges in elasmobranchs (4). Of the three POC meters evaluated, the Precision Xtra—a human-licensed glucometer—showed the greatest agreement with the reference laboratory methods when measuring glucose from whole blood (mean bias: 0 mg/dL) and measuring β-HB from plasma (mean bias: 0.1 mmol/L) in stingrays. Stingray sex, weight, PCV, and BC did not significantly affect differences between the POC meters and the reference laboratory results. All three POC meters had better precision (i.e., intra-assay variability or repeatability) when using plasma than whole blood to measure glucose. Regarding the study’s hypothesis, the mean bias and CV for plasma β-HB concentrations was relatively small (min-max: 0–0.1 mmol/L and 0%, respectively) across all three POC meters, suggesting POC β-HB measurement may have sufficient agreement with reference laboratory methods and precision for clinical and field research purposes.

The American Society for Veterinary Clinical Pathology (ASVCP) has developed guidelines for quality assurance and control of POC glucose measurement in light of the rising number of commercially available POC glucometers (43). Point-of-care meter error can be categorized as pre-analytical, analytical, and post-analytical (43). Pre-analytical error was minimized in the present study by adequate observer training and using similar sampling techniques and venipuncture sites. However, species- or individual-specific matrix properties (e.g., blood viscosity, rheology, mean corpuscular volume, etc.) may have contributed to error between the three POC meters and the reference laboratory analyzer, particularly for glucose concentrations. The distribution of glucose between RBCs and plasma is an important matrix-related variable across species (43). Glucose is distributed equally between RBCs and plasma in humans, whereas in canine, feline, rabbit, and avian patients, glucose in circulation is found predominantly in plasma (43). To the authors’ knowledge, the distribution of glucose between RBCs and plasma in the four stingray species sampled is unknown and remains a potential avenue of future physiologic research. Analytical error was minimized during data collection by using test strips before they expired, routine calibration using manufacturer-provided control materials, and not operating the POC meters under extreme environmental conditions. Post-analytical error was avoided by using consistent units of measurement (mg/dL for glucose, mmol/L for β-HB) and recording results immediately after being displayed by the POC meters.

In human and veterinary patients, anemia or hemoconcentration can falsely increase or decrease POC meter glucose or β-HB measurements (15, 18, 21, 43, 44). In stingrays, linear mixed-effect models found PCV had no significant effect on differences between the three POC meters and the reference laboratory analyzer, despite a subset of sampled stingrays with PCV values lower than the recommended ranges for the meter. Possible reasons PCV influences POC meter glucose or β-HB measurements include built-in conversion factors or increased diffusion rate in test strips with lower PCV (43). This study did not assess whether analytical interference occurred due to other blood gas or biochemical parameters such as pH, pO₂, cholesterol, bilirubin, or biliverdin (43). All sampled stingrays were anesthetized with MS-222 to ensure safe and efficient handling, which could have resulted in analytical interference secondary to perianesthetic physiologic changes (e.g., acid–base status, body temperature, ventilation, vasoconstriction, etc.). Although veterinary studies are lacking, POC glucometers are less accurate intraoperatively in human patients under anesthesia (45, 46).

The ASVCP guidelines advise TE_obs for laboratory instruments, including POC glucose and β-HB meters, to be <10% for values below the reference interval or < 20% for values within and above the reference interval (43, 47). All three POC meters for both glucose and β-HB using whole blood and plasma had TE_obs > 20%, indicative of undesirable analytical performance. Furthermore, the US Food and Drug Administration requires 95 and 98% of measurements to be within 12 and 15% of reference concentrations for POC glucometers intended for professional healthcare use and 96 and 100% of measurements to be within 15 and 20% of reference concentrations for POC glucometers intended for over-the-counter use (26). All three POC meters evaluated did not meet US Food and Drug Administration relative error guidelines for human glucometers when measuring whole blood or plasma glucose concentrations in stingrays (27). To the authors’ knowledge, neither the US Food and Drug Administration nor ASVCP has produced guidelines for POC β-HB meters despite their frequent use in monitoring diabetes mellitus in humans and domestic small animals.

Regardless of specific statistical analyses employed to compare new and established diagnostic methods, it is essential to appraise the clinical relevance of their analytic agreement. Clarke or Parkes Error Grid Analysis has historically been used in human and veterinary POC glucometer research to understand better potential clinical outcomes associated with POC meter accuracy (48, 49). This study implemented neither method due to the wide range of glucose and β-HB concentrations observed in elasmobranchs and little consensus on pathophysiologic thresholds for either analyte (e.g., hypoglycemic, ketonemia). However, the BHBCheck meter could potentially lead to highly inaccurate glucose interpretations due to the magnitude of its mean bias using whole blood (48 mg/dL) or plasma (69 mg/dL) and the narrow range of glucose concentrations observed by the reference laboratory (< 20–63 mg/dL). Passing-Bablok regression analyses found that all three POC meters mainly exhibited significant positive proportional bias (i.e., 95% CI > 1) measuring glucose and β-HB, signifying greater error at higher concentrations. Consequently, future research should assess agreement with a reference laboratory under potential pathophysiologic states or physiologic extremes (e.g., hyperglycemia secondary to long-line capture, ketonemia due to prolonged fasting, etc.).

Limitations of the present study include limited sample sizes per species (n < 10) and variable fasting durations before sampling, given differences in husbandry routines at the separate facilities. While imputation by substituting values below the LLOQ with a constant was employed in the present study, multiple imputation, maximum likelihood estimation, or kernel density estimation may be better alternative approaches. Although variation in venipuncture sites can influence PCV in sharks (50) and downstream biochemical analyses in other taxa, no statistically significant differences between venipuncture sites were found in southern stingrays (51). Although plasma was separated from RBCs promptly via centrifugation, glucose and β-HB concentrations could have decreased slightly between venipuncture and processing due to metabolism by cellular constituents.

Rapid, reliable detection of hyperglycemia due to physiologic stress could help veterinarians and researchers working with elasmobranchs more objectively monitor their welfare during capture, handling, and transport. Similarly, accurate POC measurement of β-HB could provide almost instant, cost-effective insight into the metabolic state of the elasmobranch patient vs. the turnaround time and expense of sending a sample to a reference laboratory. Further research should investigate POC meter clinical utility under different metabolic states in elasmobranchs. Using POC meters to measure glucose and β-HB concentrations in stingrays may be possible when reference methods are inaccessible. β-HB measurement using plasma will likely have sufficient accuracy and precision for clinical and field research purposes for the three POC meters evaluated in this study, especially the human-licensed glucometer, Precision Xtra.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was approved by North Carolina State University Institutional Animal Care and Use Committee. The study was conducted in accordance with the local legislation and institutional requirements.

ND: Conceptualization, Investigation, Data Curation, Formal analysis, Writing – original draft, Writing – review & editing. CC: Writing – review & editing. LW: Writing – review & editing. EC: Writing – review & editing.