Introduction
Calcium exists in 3 fractions within the body: protein bound, complexed, and ionized.1 Ionized calcium is the biologically active form and accounts for approximately 50% of the total serum calcium.1,2 Calcium homeostasis is closely regulated along with phosphorus via parathyroid hormone (PTH), calcitriol (1,25-dihydroxyvitamin D3), and calcitonin.1–3 During periods of hypocalcemia, PTH secretion increases plasma calcium via mobilization of calcium and phosphorus from bone, increased renal absorption of calcium, and increased urinary excretion of phosphorus as well as increased calcitriol synthesis.1,3 Calcitriol has similar functions in addition to resulting in increased intestinal absorption of calcium and phosphorus.1,3 Increased calcitriol results in negative feedback to inhibit PTH synthesis.1,3 Hypercalcemia suppresses PTH and upregulates calcitonin secretion to inhibit osteoclastic bone resorption.1,3
Hyperparathyroidism is often characterized by hypercalcemia and hypophosphatemia or low normal phosphorus concentrations.3 Parathyroid hormone–related protein (PTHrP) acts in the same physiologic fashion as PTH; however, PTHrP is synthesized by neoplastic cells and suppresses PTH secretion.3,4 Increased PTH secretion can occur with chronic kidney disease (CKD) due to retention of phosphorus and a decreased renal ability to synthesize vitamin D.1,3
In dogs, as in humans, the most frequent cause of total hypercalcemia is neoplasia, with lymphoma being most common.4–6 Neoplasia, followed by renal failure, hyperparathyroidism, and hypoadrenocorticism were the most common causes of ionized hypercalcemia.5 Other neoplastic causes include apocrine gland anal sac adenocarcinoma, multiple myeloma, and thyroid carcinoma.2,6
Studies have sought to evaluate hypercalcemia and underlying causes as well as methods for prediction of total and ionized calcium. In a 2009 study, dogs with hypercalcemia caused by neoplasia had higher ionized calcium concentrations than dogs with other causes of hypercalcemia; however, the magnitude of hypercalcemia could not reliably predict the underlying cause.5 Another study2 in 2019 confirmed neoplasia as the most common cause of ionized hypercalcemia in dogs presenting to a tertiary facility and that mild calcium increases were more likely to reflect nonpathologic or transient conditions.
Determining the etiology of hypercalcemia can pose a diagnostic conundrum when azotemia is present but phosphorus is normal, especially when attempting to differentiate between neoplasia and other causes. In 1 study7 of 40 dogs, those with lymphoproliferative disease had higher mean calcium concentration and lower mean inorganic phosphate concentration and were significantly older than the dogs with hypoadrenocorticism. This study did not evaluate the relationships between these values and azotemia or phosphorus.7 As previously stated, PTH and PTHrP would be expected to cause diverging effects on calcium and phosphorus, resulting in hypophosphatemia in the face of hypercalcemia.1 This expected pattern of divergence helps to rule out differential diagnoses for hypercalcemia that result in hyperphosphatemia, such as renal azotemia. When hypercalcemia is present along with azotemia and a normal phosphorus, a possible explanation for the normal phosphorus could be the presence of 2 counterforces, one decreasing phosphorus (ie, increased PTH) and one increasing phosphorus (ie, reduced glomerular filtration rate), ultimately resulting in an overall normal phosphorus.8 Given this, before hypercalcemia is assumed to be due to renal dysfunction, investigation of an underlying neoplastic process is indicated.
Human research has evaluated mineral levels in people with neoplasia. In 1 study,9 patients with breast cancer had a Ca:P significantly higher than the normal control group, thought to be due to abnormal PTH regulation. In another human study,10 Ca:P was found to be an accurate tool for differentiating patients with primary hyperparathyroidism from healthy controls. This specific ratio has not been evaluated in dogs as a method to identify the divergence of calcium and phosphorus due to PTH or PTHrP’s influence on disease.
The objective of this retrospective study was to evaluate dogs with total hypercalcemia, azotemia, and normal serum phosphorus concentrations to determine whether this constellation of laboratory abnormalities could predict neoplasia. We hypothesized that dogs with these findings were more likely to have a neoplastic diagnosis. A secondary aim was to determine whether Ca:P and/or iCa:P could be utilized to predict hypercalcemia of malignancy. We hypothesized that azotemic, hypercalcemic dogs with neoplasia would have a higher Ca:P and iCa:P than azotemic, hypercalcemic dogs with nonneoplastic hypercalcemia.
Materials and Methods
Medical records between 2010 and 2020 were reviewed for canine patients with hypercalcemia (defined as total calcium > 11.3 mg/dL), azotemia (defined as creatinine > 1.6 mg/dL),11 and normal phosphorus concentrations (defined as phosphorus 2.2 to 6.4 mg/dL based on the laboratory reference interval). If multiple visits noted this pattern of findings, information from the first visit with this pattern was used for the study. Information obtained from the medical records included age, breed, sex, prescribed medications, and diet information when available. Serum creatinine, BUN, total calcium, and phosphorus concentrations were measured using standard laboratory methods (Dimension Xpand Plus integrated chemistry system; Siemens Medical Solutions USA Inc). An ionized calcium was also recorded, if available, and hypercalcemia was defined as a value > 1.4 mmol/L (Point-of-care i-STAT CG8+; Abbott). A diagnosis of hypercalcemia of malignancy required a cytologic or histopathologic diagnosis of neoplasia for inclusion. Dogs were excluded if any of the key laboratory abnormalities were not available for review. Patients were also excluded if they had received chemotherapy, furosemide, steroids, strict renal diet, bisphosphonates, a phosphate binder, and/or calcitriol within 30 days prior to evaluation.
Statistical analysis
Statistical analyses were performed using statistical software (SPSS version 28.0.0.0; IBM Corp). The Shapiro-Wilk test was used to assess normality with examination of frequency histograms where needed to assess distributions. Data throughout were nonnormally distributed, so nonparametric tests were used for analysis. Descriptive statistics were obtained using standard methods and are reported as median (IQR) when nonnormally distributed.
Diagnoses for all cases were coded as presumed either nonneoplastic based on diagnostic work-up and eventual diagnosis or known neoplasia.
Dichotomous groups were compared using the Kolmogorov-Smirnov test for nonnormal data and independent samples Mann-Whitney U test where applicable. A parametric independent samples Mann-Whitney U test was used for comparison of Ca:P for neoplastic versus nonneoplastic cases. Receiver operating characteristic (ROC) curves were used for significant variables, specifically Ca:P and iCa:P, if applicable. Optimal sensitivity and specificity cutoffs were determined using ROC curve analysis. Multivariate logistic regression analysis was used to evaluate whether total calcium, azotemia, and normal phosphorus were predictive of neoplasia. Standard methods for sensitivity and specificity calculation were employed. A priori α level was set at P < .05 for all tests.
Results
A total of 105 dogs were included in the study. Fifty breeds were represented, with the most common being mixed breed (n = 26 [24.8%]), followed by Labrador Retriever (11 [10.5%]), Australian Cattle Dog (5 [4.8%]), Golden Retriever (4 [3.8%]), and Dachshund (4 [3.8%]). The rest of the breeds each represented < 5% of the included population. The median age for all cases was 10 years (IQR, 6 years). Examination of the frequency histogram showed a bimodal distribution of age with modes at 7 and 13 years of age. For the nonneoplasia group, the median age was 10.5 years (IQR, 8 years), and in the neoplasia group, the median age was 10 years (IQR, 5 years).
Out of the 105 cases, 37% (n = 39) had known neoplasia and 63% (66) had no evidence of neoplasia. Of the patients with known neoplasia, 46% (n = 18) had either lymphoma or apocrine gland anal sac adenocarcinoma and 54% (21) had other neoplasia. Of the patients with presumed nonneoplastic disease, 61% (n = 40) had CKD, 36% (24) were unknown or had other disease and presumed unlikely to be neoplastic given the diagnostic work-up, and 3% (2) had hypoadrenocorticism.
The nonneoplasia group had a median total calcium of 11.9 mg/dL (IQR, 0.8); the neoplasia group had a median total calcium of 13.8 mg/dL (IQR, 4). The highest individual total calcium values in the nonneoplasia group were 20, 16.1, and 15.9 mg/dL, whereas they were 23.6, 19.8, and 19.2 mg/dL in the neoplasia group. The median total calcium concentration was significantly higher in the neoplasia group than the nonneoplasia group (P < .001).
Median creatinine concentration for the nonneoplasia group was 2.5 mg/dL (IQR, 1.2) and 2.3 mg/dL (IQR, 1.2) for the neoplasia group. There was no significant difference in creatinine values between the nonneoplasia and neoplasia groups (P = .408). Median BUN for the nonneoplasia group was 52.5 mg/dL (IQR, 28) and 47 mg/dL (IQR, 20) for the neoplasia group. There was no significant difference in BUN concentration between the nonneoplasia and neoplasia groups (P = .229).
Not all patients had ionized calcium available. In the nonneoplasia group, 13 patients had a reported ionized calcium, and in the neoplasia group, 13 patients had a reported ionized calcium. For the nonneoplasia group, the median ionized calcium was 1.4 mmol/L (IQR, 0.18) and the highest values were 1.57, 1.56, and 1.53 mmol/L. For the neoplasia group, the median ionized calcium value was 1.76 mmol/L (IQR, 0.49) and the highest values were 2.14, 2.13, and 2.01 mmol/L.
In the nonneoplasia group, the median phosphorus was 4.8 mg/dL (IQR, 1.5). The median phosphorus for the neoplasia group was 4.6 mg/dL (IQR, 1.6). This was an expected finding as normal phosphorus was an inclusion criterion for this study, but there was no difference in phosphorus concentration (P = .725) between groups.
The median creatinine-to-phosphorus ratio (Cr:P) for the nonneoplasia group was 0.574 (IQR, 0.36) and was 0.528 (IQR, 0.35) for the neoplasia group. There was no significant difference (P = .535) in Cr:P between the nonneoplasia and neoplasia groups.
A multivariate analysis was used to predict neoplasia on the basis of total calcium, phosphorus, and creatinine. These variables significantly predicted neoplasia (F[3, 104] = 8.168; P < .001; R = 0.442). However, only total calcium added statistical significance to the overall model (P < .001), meaning that the contributions of the other variables (creatinine and phosphorus) were not significant (P = .167 and .405, respectively).
Median Ca:P was 2.54 (IQR, 0.91) for the nonneoplasia group and 3.0 (IQR, 1.21) for the malignancy group. Ca:P was significantly higher (P = .009) in the neoplasia group compared to the nonneoplasia group. An ROC curve was obtained for the Ca:P in neoplasia versus nonneoplasia cases (Figure 1). The area under the concentration-versus-time curve (AUC) was 0.652 (P = .003). The optimal sensitivity and specificity cutoff for Ca:P based on ROC analysis was 2.5, which yields an 80% sensitivity but only 46% specificity.
Results of receiver operating characteristic (ROC) curve analysis to determine the predictive value of the serum calcium-to-phosphorus ratio (Ca:P) for neoplasia versus nonneoplasia in 105 dogs with hypercalcemia (defined as serum total calcium > 11.3 mg/dL), azotemia (defined as serum creatinine > 1.6 mg/dL), and serum phosphorus concentrations within reference range (2.2 to 6.4 mg/dL) between 2010 and 2020.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
Results of receiver operating characteristic (ROC) curve analysis to determine the predictive value of the serum calcium-to-phosphorus ratio (Ca:P) for neoplasia versus nonneoplasia in 105 dogs with hypercalcemia (defined as serum total calcium > 11.3 mg/dL), azotemia (defined as serum creatinine > 1.6 mg/dL), and serum phosphorus concentrations within reference range (2.2 to 6.4 mg/dL) between 2010 and 2020.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
Results of receiver operating characteristic (ROC) curve analysis to determine the predictive value of the serum calcium-to-phosphorus ratio (Ca:P) for neoplasia versus nonneoplasia in 105 dogs with hypercalcemia (defined as serum total calcium > 11.3 mg/dL), azotemia (defined as serum creatinine > 1.6 mg/dL), and serum phosphorus concentrations within reference range (2.2 to 6.4 mg/dL) between 2010 and 2020.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
The median iCa:P was 0.302 (IQR, 0.12) for the nonneoplasia group and 0.395 (IQR, 0.11) for the neoplasia group. There was a significant difference between neoplasia and nonneoplasia cases in median iCa:P (P = .004). The ROC analysis for the iCa:P showed an AUC of 0.834 (P = .004; Figure 2). The optimal cutoff value for the iCa:P was 0.33, which yields a 92% sensitivity and 77% specificity.
Results of ROC curve analysis to determine the predictive value of the serum ionized calcium-to-phosphorus ratio (iCa:P) for neoplasia versus nonneoplasia in the dogs described in Figure 1.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
Results of ROC curve analysis to determine the predictive value of the serum ionized calcium-to-phosphorus ratio (iCa:P) for neoplasia versus nonneoplasia in the dogs described in Figure 1.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
Results of ROC curve analysis to determine the predictive value of the serum ionized calcium-to-phosphorus ratio (iCa:P) for neoplasia versus nonneoplasia in the dogs described in Figure 1.
Citation: Journal of the American Veterinary Medical Association 261, 9; 10.2460/javma.23.01.0039
Discussion
This study found that in azotemic dogs with normal serum phosphorus, the Ca:P was significantly higher in dogs with neoplasia than in dogs with nonneoplastic causes of hypercalcemia, but the sensitivity and specificity of this ratio was too low to be clinically useful. However, the iCa:P ratio was statistically significant, with a 92% sensitivity and 77% specificity for prediction of neoplasia at a cutoff of 0.33. Given that there was no difference in the phosphorous concentrations between the 2 groups, the increased ratio was mostly likely a reflection of the higher calcium concentration in dogs with neoplasia. When evaluating the constellation of laboratory abnormalities in the prediction of neoplasia, only the total calcium was a significant contributor to the prediction model.
Hypercalcemia of malignancy is common and falls under the umbrella of paraneoplastic syndromes.2,4,5,12 There are multiple mechanisms behind the development of hypercalcemia of malignancy including but not limited to PTH-related peptide, osteolytic neoplasia, and primary hyperparathyroidism from parathyroid carcinoma.4 The degree of hypercalcemia has been evaluated previously in veterinary studies in an attempt to predict neoplasia. There have been conflicting results, but overall, the magnitude of ionized hypercalcemia does not reliably predict neoplasia.2,5 This study confirmed previous associations of total hypercalcemia with neoplasia, finding that dogs in the neoplasia group had a significantly higher total calcium than dogs in the nonneoplasia group. As total calcium is a readily available value in veterinary practices, this information can be used by a wide array of veterinarians.
The Ca:P has not been previously evaluated in veterinary medicine; however, it has been minimally evaluated in humans in correlation with primary hyperparathyroidism and breast cancer. The paper by Madeo et al10 evaluating primary hyperparathyroidism and hypophosphatemia from healthy subjects found a cutoff ≥ 3.56 as clinically significant for a diagnosis of primary hyperparathyroidism. For our study, an observational cutoff ≥ 2.5 was used for determining sensitivity and specificity between the neoplasia and nonneoplasia groups. The overall sensitivity and specificity was not high and less discriminating than what has been reported in humans when using it as a diagnostic tool for differentiating primary hyperparathyroidism.10 The sensitivity and specificity for the Ca:P reported for differentiating primary hyperparathyroidism from normal controls in humans is higher given that patients with hyperparathyroidism generally have a low to low-normal phosphorus in combination with a high total calcium, thus making a higher overall ratio. Further studies are warranted to determine whether the ratio could be more clinically useful in a larger group of dogs with a less-specific constellation of laboratory abnormalities.
Other ratios were also evaluated including iCa:P and Cr:P. The iCa:P was specifically calculated, as the underlying physiology of diverging calcium and phosphorus concentration in the face of an increasing creatinine could be expected and used in determining the cause of hypercalcemia. Not every dog in the study had an ionized calcium available; however, in the dogs that did have one performed, the iCa:P was significantly higher in those with neoplastic causes of hypercalcemia, and the sensitivity and specificity were reasonable for prediction of neoplasia. This ratio is a potential area for further investigation in a larger population of dogs with measurement of iCa concentration.
The Ca:P was examined for a reason similar to the iCa:P, that it could reflect diverging influences on creatinine and phosphorous concentrations. Although a normal phosphorous concentration was a requirement for inclusion in this study, it remained possible that phosphorous concentrations between the 2 populations could differ despite being within a reference range. There was no significant difference in the Cr:P between the dogs with neoplasia and nonneoplasia. Possible explanations for the lack of difference in Cr:P between the groups could have included differences in degree of azotemia between the groups, but serum creatinine concentration was not different. Fibroblast growth factor-23 (FGF-23), a bone-derived hormone, promotes renal phosphorus excretion and suppresses 1ɑ-hydroxylase activity, leading to decreased 1,25(OH)2D synthesis.13–17 A result of increased FGF-23 secretion is reduction of serum phosphorous concentrations. FGF-23 concentration increases early in dogs with CKD and is increased in dogs with CKD and normal phosphorous concentrations; phosphorous concentrations may not become abnormally high until later stages of CKD.10,17 Thus, the hyperphosphatemic effects of decreased renal phosphorous excretion associated with CKD may have been mitigated by FGF-23 activity. The role of FGF-23 in dogs with azotemia provoked by hypercalcemia of malignancy has not been investigated to our knowledge.
This study had several limitations including that it was a retrospective study and had tight inclusion criteria of laboratory abnormalities. We did not, for example, compare all cases of hypercalcemia, regardless of phosphorus concentration or azotemia, as this was considered beyond the scope of our specific study.
Another limitation was that duration of these laboratory abnormalities was unknown and not a consideration for the inclusion criteria. Dogs with CKD generally have a slower onset of clinical signs, and the development of laboratory abnormalities in comparison to the often acute nature in dogs with neoplasia. A future study could prospectively evaluate this aspect, as this could also be an important difference between the neoplasia and nonneoplasia groups.
Previous studies of hypercalcemia in dogs have documented neoplasia as the most common cause; however, to the best of our knowledge, no studies have examined hypercalcemia in relation to azotemia.5,6,12 The high number of patients with neoplasia in our study population would suggest that in a hypercalcemic, azotemic, and normophosphatemic dog, there should be an effort to rule out neoplasia prior to assuming that the hypercalcemia is secondary to renal dysfunction in these cases.
In conclusion, the total calcium and Ca:P were higher in azotemic, normophosphatemic dogs with neoplasia; however, the low sensitivity and specificity of this ratio made it an unreliable tool to predict neoplasia in this specific study population. Evaluation of iCa:P might be a useful tool in predicting neoplasia in patients with this constellation of clinical signs. This study suggests a need for further investigation of the utility of the Ca:P in a broader population of dogs with hypercalcemia for predicting an underlying neoplastic process. We also caution against assumption of a diagnosis based solely on biochemical values, and further investigation by the clinician for a definitive diagnosis is still warranted.
Acknowledgments
No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors have nothing to declare.
References
- 1.↑
Hall JE, ME Hall. Parathyroid hormone, calcitonin, calcium and phosphate metabolism, vitamin D, bone, and teeth. In: Guyton and Hall Textbook of Medical Physiology. 14th ed. Saunders Elsevier; 2021:991–1006.
- 2.↑
Coady M, Fletcher DJ, Goggs R. Severity of ionized hypercalcemia and hypocalcemia is associated with etiology in dogs and cats. Front Vet Sci. 2019;6:276. doi:10.3389/fvets.2019.00276
- 3.↑
Skelly BJ. Primary hyperparathyroidism. In: Ettinger SJ, Feldman EC, Cote E, eds. Textbook of Veterinary Internal Medicine. 8th ed. Elsevier; 2017:1715–1727.
- 4.↑
Bergman PJ. Paraneoplastic hypercalcemia. Top Companion Anim Med. 2012;27(4):156–158. doi:10.1053/j.tcam.2012.09.003
- 5.↑
Messinger JS, Windham WR, Ward CR. Ionized hypercalcemia in dogs: a retrospective study of 109 cases (1998-2003). J Vet Intern Med. 2009;23(3):514–519. doi:10.1111/j.1939-1676.2009.0288.x
- 6.↑
Groth EM, Chew DJ, Lulich JP, et al. Determination of a serum total calcium concentration threshold for accurate prediction of ionized hypercalcemia in dogs with and without hyperphosphatemia. J Vet Intern Med. 2020;34(1):74–82. doi:10.1111/jvim.15654
- 7.↑
Elliott J, Dobson JM, Dunn JK, Herrtage ME, Jackson KF. Hypercalcaemia in the dog: a study of 40 cases. J Small Anim Pract. 1991;32(11):564–571. doi:10.1111/j.1748-5827.1991.tb00887.x
- 8.↑
Kruger JM, Osborne CA, Nachreiner RF, Refsal KR. Hypercalcemia and renal failure. Etiology, pathophysiology, diagnosis, and treatment. Vet Clin North Am Small Anim Pract. 1996;26(6):1417–1445. doi:10.1016/s0195-5616(96)50135-x
- 9.↑
Abdelgawad IA, El-Mously RH, Saber MM, Mansour OA, Shouman SA. Significance of serum levels of vitamin D and some related minerals in breast cancer patients. Int J Clin Exp Pathol. 2015;8(4):4074–4082.
- 10.↑
Madeo B, De Vincentis S, Kara E, et al. Reliability of calcium-phosphorus (Ca/P) ratio as a new, accurate and inexpensive tool in the diagnosis of some Ca-P disorders. J Endocrinol Invest. 2019;42(9):1041–1049. doi:10.1007/s40618-019-01025-6
- 11.↑
IRIS staging of CKD. Internation Renal Interest Society. Accessed January 24, 2023. www.iris-kidney.com/guidelines/staging.html
- 12.↑
Sakals S, Peta HG, Fernandez NJ, Allen AL. Determining the cause of hypercalcemia in a dog. Can Vet J. 2006;47(8):819–821.
- 13.↑
Parker VJ, Harjes LM, Dembek K, Young GS, Chew DJ, Toribio RE. Association of vitamin D metabolites with parathyroid hormone, fibroblast growth factor-23, calcium, and phosphorus in dogs with various stages of chronic kidney disease. J Vet Intern Med. 2017;31(3):791–798. doi:10.1111/jvim.14653
- 14.
Wolf M. Update on fibroblast growth factor 23 in chronic kidney disease. Kidney Int. 2012;82(7):737–747. doi:10.1038/ki.2012.176
- 15.
Harjes LM, Parker VJ, Dembek K, et al. Fibroblast growth factor-23 concentration in dogs with chronic kidney disease. J Vet Intern Med. 2017;31(3):784–790. doi:10.1111/jvim.14707
- 16.
Quarles LD. Skeletal secretion of FGF-23 regulates phosphate and vitamin D metabolism. Nat Rev Endocrinol. 2012;8(5):276–286. doi:10.1038/nrendo.2011.218
- 17.↑
Miyakawa H, Hsu HH, Ogawa M, Akabane R, Miyagawa Y, Takemura N. Association between serum fibroblast growth factor-23 concentration and development of hyperphosphatemia in normophosphatemic dogs with chronic kidney disease. J Vet Intern Med. 2021;35(5):2296–2305. doi:10.1111/jvim.16237
