Evaluation of calcium to phosphorus ratio in spot urine samples as a practical method to monitor phosphorus intake adequacy in sows

Abstract

The objective of this study was to evaluate the reliability of using Ca to P ratio measured in spot urine samples to assess P intake adequacy in gestating and lactating sows. A total of 36 sows were fed one of six concentrations of dietary total P (0.40%, 0.48%, 0.56%, 0.64%, 0.72%, and 0.80%) from day 7.5 ± 1 after breeding until the end of lactation (day 26.6 ± 1). Dietary Ca to P ratio was maintained constant across treatments at 1.25:1. Total 24-h urine samples were collected in mid- and late gestation (days 77.1 ± 2 and 112.4 ± 1), and early and late lactation (days 4.5 ± 1 and 18.2 ± 1). In parallel to 24-h collections, spot urine samples were collected at three different times (early morning, late morning, and late afternoon) in late gestation and late lactation. Urine Ca and P concentrations were measured and Ca to P ratio was calculated. Sows were classified as P-adequate or P-deficient according to dietary P intake. Urine Ca to P ratio was greater in sows fed P-deficient diets than sows fed P-adequate diets (P < 0.001). Receiver operating characteristic (ROC) curves were used to determine the cutoff values for urine Ca to P ratio to predict P intake adequacy. Three different categories of P intake were defined according to urine Ca to P ratio: deficient, adequate, and excessive. The area under the ROC for Ca to P ratio was 0.88 (95% CI 0.81 to 0.95). Best cutoff value of urine Ca to P ratio was 1.5 (sensitivity 94% and specificity 68%) to identify sows fed P-deficient diets and 0.5 for P-excessive diets (sensitivity 82% and specificity 82%). A strong relationship between Ca to P ratio in 24-h and spot urine samples was determined (r = 0.93, P < 0.01), independent of physiological state and collection time of spot samples (adjusted-R² = 0.86, P < 0.01). The degree of agreement between spot and 24-h urine for P intake adequacy, assessed by Cohen’s weighted kappa analysis, was substantial (0.78, 95% CI 0.69 to 0.88). We conclude that urinary Ca to P ratio provides a reliable prediction of the adequacy of P intake in reproducing sows. Urinary Ca to P ratio measurements in random spot urinary offers a practical method to determine dietary P adequacy.

Introduction

Kidneys provide a major role in the maintenance of systemic P homeostasis as the main excretory route of absorbed P is via urine. Under normal conditions, renal P reabsorption is regulated in response to dietary P absorption and physiological demand. Therefore, concentrations of P in urine have been used to evaluate the relationship between P intake and P nutritional status in human and animal studies (Caple et al., 1982; Rodehutscord, 1999; Hagemoser et al., 2000; St-Jules et al., 2017). Total 24-h urinary P excretion provided a reliable method to estimate P requirements in sows (Grez-Capdeville and Crenshaw, 2021). These results underline the value of urinalysis as a screening test and the plausibility of using urinary P measurements as a practical method to assess P status in reproducing sows.

The collection of urine over a 24-h period is frequently used for analysis of renal solute excretion in swine (Darriet et al., 2017; Penniston et al., 2017; Lee et al., 2019). The 24-h collection negates fluctuations in urine concentration and diurnal solute excretion known to occur throughout a 24-h cycle. However, collections of 24-h urine samples from pigs require the use of metabolic crates or indwelling bladder catheterization procedures, and thus limit the number of observations feasible in large-scale studies. Alternatively, single-void urine specimens (spot samples) represent a convenient and practical method to collect urine. Ideally, spot urine samples can be collected at any time of the day from spontaneous urination and corrected for urine dilutions using endogenous or exogenous markers. Therefore, urine collections based on spot samples, instead of 24-h collections, would facilitate the use of urinary P measurements as a practical application for monitoring P status in sow herds.

Creatinine is one of the most common markers used to adjust spot urine samples for volume dilutions. However, studies with sows reported that urinary excretion of creatinine was affected by body weight loss, physiological state, and phase of lactation (Bate and Hacker, 1981; Strathe et al., 2017). Therefore, these variables must be considered when using creatinine to normalize spot samples, which requires the collection of additional information.

In our previous work (Grez-Capdeville and Crenshaw, 2021), an inverse relationship was observed between 24-h urinary excretions of P and Ca in response to P intake. Phosphorus excretion increased as sows consumed more P. In contrast, urinary Ca decreased with increasing dietary P, and remained relatively low and constant in sows fed at or above the estimated P requirements. Similar results have been previously reported for humans (Hegsted et al., 1981; Senterre and Salle, 1988), pigs (Vipperman et al., 1974; Hagemoser et al., 2000; Gutierrez et al., 2015), and rodents (Anderson and Draper, 1972; Wood et al., 1988). This inverse relationship in urinary Ca and P excretions reflects adaptations to intake, absorption, and postabsorptive utilization of these minerals. Therefore, Ca to P ratio in urine may provide a method to determine P intake adequacy without a need to adjust urine for variable dilutions. The objective of this study was to evaluate the use of urinary Ca to P ratio measurements in spot urine samples as a method to assess adequacy of P intake in gestating and lactating sows.

Materials and Methods

The study was conducted at the University of Wisconsin-Madison Swine Research and Teaching Center. All animal procedures were approved by the University of Wisconsin-Madison Institutional Animal Care and Use Committee (protocol A005638).

Animals and experimental diets

This study was based on samples collected in our previous study designed to determine P requirements in reproducing sows using 24-h urinary P excretion. General management procedures were detailed previously (Grez-Capdeville and Crenshaw, 2021). Briefly, a total of 36 crossbred Camborough PIC sows (parity 3 to 7) were fed corn-soybean meal-based diets formulated to provide total P concentrations of 0.40%, 0.48%, 0.56%, 0.64%, 0.72%, or 0.80% (n = 6 sows per diet). Calcium to total P ratio was maintained constant across diets at 1.25:1. Detailed information on ingredients and nutrient composition of the experimental diets is presented in Supplementary Tables 1 and 2. The use of standardized total tract digestible P (STTD P) has been recommended in order to formulate diets that meet nutritional requirements and to promote a more efficient use of P (NRC, 2012). However, previous studies with sows have reported that digestibility of Ca and P is associated with specific physiological stages (Giesemann et al., 1998; Lee et al., 2019). As values for digestibility are variable, and total P concentrations in experimental diets can be determined by laboratory analyses, we decided to design this study based on concentrations of total P in the diets rather than STTD P. Concentrations of STTD P in the diets were calculated from values for feed ingredients in the NRC (2012). These values were used to estimate urinary P excretion as a percentage of STTD P intake. Dietary treatments were fed from day 7.5 ± 1 after breeding until the end of the lactation period (day 26.6 ± 1). Sows were fed 2.0 kg diet per day during gestation and had free access to feed from the day of farrowing until weaning. Feed intake was measured at the following intervals during lactation, day 0 to 5, 5 to 12, 12 to 19, and 19 to weaning. Intervals from day 0 to 5 and day 12 to 19 were used to calculate average daily feed intake in early and late lactation, respectively. Sows had continuous access to fresh water throughout the experiment.

Urine sample collections and analyses

Total 24-h urine samples (n = 133) were collected from all sows on gestation days 77.1 ± 2 and 112.4 ± 1, and lactation days 4.5 ± 1 and 18.2 ± 1. Spot urine samples (n = 108) were collected from 18 sows in parallel with the 24-h collections on gestation day 112.4 ± 1 and lactation day 18.2 ± 1 at three different time points: early morning (collection between 7300 and 9300 h), late morning (collection between 1030 and 1130 h), and late afternoon (collection between 1600 and 1700 h). Selected sows for collection of spot samples were fed dietary treatments with total P levels of 0.40%, 0.56%, and 0.80%.

The 24-h urine samples were collected using Foley indwelling bladder catheters. Procedures for catheter insertion and 24-h urine collections were previously described (Grez-Capdeville and Crenshaw, 2021). Spot urine samples were collected directly from the catheter tube into conical centrifuge tubes (50 mL). Samples were stored at 4 °C until analysis for concentrations of Ca and P (mg/g urine). Urine samples were digested with a nitric-perchloric acid mixture (3:1 v/v). Concentrations of P in the urine digests were measured by the spectrophotometric molybdovanadophosphate method (AOAC, 1980; method 2.022) using the Gilford Spectrophotometer 260 (Gilford Instrument Laboratories, Inc., Oberlin, OH). Concentrations of Ca were determined by flame atomic absorption spectrometry (Perkin-Elmer AAnalyst 400, Perkin-Elmer Corporation, Norwalk, CT). All samples were analyzed in duplicates. Spike-recovery tests were used to evaluate accuracy of analytical methods. The average recovery was 98% ± 9% for P and 105% ± 8% for Ca analyses.

Statistical analysis

All statistical analyses were performed using SAS (version 9.4; SAS institute Inc., Cary, NC). The adequacy of P intake was based on values for P requirements determined in our previous study (Grez-Capdeville and Crenshaw, 2021). Briefly, requirements were estimated by mathematical modeling of the P intake and urinary P excretion and curves. Different regression models were tested, and the final model was selected based on statistical fit to the observed data and appropriate description of the biological response. The estimated dietary total P concentrations were 0.51% (10.2 g/d) in mid-gestation, 0.52% (10.4 g/d) in late gestation, 0.51% (31.1 g/d) in early lactation, and 0.53% (40.3 g/d) in late lactation. Based on these estimates, sows were binary classified based on the concentration of P in the assigned dietary treatment as fed below (P-deficient) or above (P-adequate) the estimated requirement. A Mann–Whitney U-test was used to assess differences in 24-h urine Ca to P ratio measurements between P-deficient and P-adequate groups. Differences were considered statistically significant at P < 0.05.

The reliability of Ca to P ratio in 24-h urine samples to predict adequacy of P intake was assessed by a receiver operating characteristic (ROC) curve analysis. Sensitivity, or true positive rate, was defined as the proportion of actual positive (P-deficient sows) that were correctly identified by urine Ca to P ratio. Specificity, or true negative rate, was defined as actual negative (P-adequate sows) that were correctly identified by urine Ca to P ratio. Sensitivity [true positive/(true positive + false negative)] was plotted versus 1 − specificity [false positive/(false positive + true negative)], and the area under the curve was determined to evaluate the overall performance of urine Ca to P ratio as a classifier. Optimal cutoff values of urinary Ca to P ratio for predicting adequacy of P intake were selected as the values that maximized the distance from the uninformative diagonal (Youden’s index; sensitivity + specificity − 1) and the correct classification rate [(true positives + true negatives)/total observations].

Linear regression analysis and Pearson’s correlation coefficients were used to determine the relationship and strength of association between Ca to P ratio in 24-h and spot urine samples. Multiple linear regression models with stepwise backward elimination were used to determine the most important variables associated with the 24-h urinary Ca to P ratio. Variables included in the initial model involved Ca to P ratio in spot urine samples, time of spot sample collection (early morning, late morning, and late afternoon collections), physiological stage of sows (gestation and lactation), and all possible two- and three-way interactions. The effect of dietary treatment (concentration of total P in the diet) was not included in the initial model because this variable was correlated with urinary Ca to P excretion in spot urine samples and inclusion of dietary treatment only resulted in minor goodness-of-fit improvements (data not shown). The best-fit model was selected based on Akaike Information Criteria (AIC) and adjusted coefficient of determination values (adjusted-R²). Natural log-transformation was applied to Ca to P ratio in 24-h and spot urine samples to fulfill normal distribution assumptions. The degree of agreement between the two methods (spot and 24-h urine collections) to categorize urine Ca to P ratio and, therefore P intake adequacy in sows, was assessed by Cohen’s weighted kappa analysis. A kappa value of 0 to 0.20 indicates none to slight agreement; 0.21 to 0.40 fair; 0.41 to 0.60 moderate; 0.61 to 0.80 substantial; 0.81 to 1.00 almost perfect to perfect agreement (Landis and Koch, 1977).

Results

Calcium to P ratio in 24-h urine samples and selected Ca to P ratio cutoff values

The distribution of Ca to P ratio in 24-h urine samples of gestating and lactating sows fed different concentrations of dietary P is shown in Figure 1. Regardless of the physiological phase of sows, urine Ca to P ratio decreased as the concentration of total P in the diet increased. Sows fed total P concentrations below the estimated requirements had a higher (P < 0.001) urine Ca to P ratio (gestation = 3 ± 1; lactation = 88 ± 13) than sows fed above the requirements (gestation = 0.16 ± 0.02; lactation = 1.4 ± 0.8).

Figure 1.

Box and whisker plots of 24-h urinary Ca to P ratio of gestating (n = 68 samples from 34 sows) and lactating (n = 65 samples from 34 sows) sows fed different concentrations of dietary total P. Boxes represent the interquartile range and whiskers indicate extreme values. Horizontal lines and black dots represent the median and mean Ca to P ratio, respectively, for each level of total P. Open circles indicate potential outliers. Shaded areas represent urinary Ca to P ratios for sows fed concentrations of total P in the diet below the estimated requirements (~0.52%).

Based on ROC analysis (Figure 2), selected cutoff values to categorize urinary Ca to P ratio into groups were 1.56 (best correct classification rate; sensitivity = 96%, specificity = 68%) and 0.449 (best Youden’s index; sensitivity = 82%, specificity = 84%). Cutoff values were rounded up to 1.5 (sensitivity = 94%, specificity = 68%) and 0.5 (sensitivity = 82%, specificity = 82%), which were used as classification thresholds. A cutoff value of 1.5 indicates that sows with urine Ca to P ratio greater than 1.5 were fed P-deficient diets, and sows with urine Ca to P ratio less than 1.5 were fed P-adequate diets. P-adequate diets were defined as diets that provided sufficient amounts of P to meet physiological requirements. In order to optimize P use efficiency, the second cutoff value for urine Ca to P ratio of 0.5 was used to identify sows with excessive amounts of dietary P intake. Sows with urine Ca to P ratio values within the range of 1.5 and 0.5 were considered to be fed P-adequate diets. Therefore, three different levels of P intake were defined based on urine Ca to P ratio categories: deficient P intake with urine Ca to P ratio > 1.5, adequate P intake with urine Ca to P ratio within the range of 0.5 to 1.5, and excessive P intake with urine Ca to P ratio < 0.5.

Figure 2.

Receiver operating characteristic curve (black line) for Ca to P ratio in 24-h urine samples in predicting adequacy of P intake. Optimal urinary Ca to P ratio cutoff values based on Youden’s index and correct classification rate were 0.5 and 1.5, respectively. The diagonal (gray line) represents classification due to chance. AUC, area under the curve.

The Ca to P ratio in 24-h urine samples and daily P excretion of sows in each Ca to P ratio category are summarized in Table 1. As expected, daily excretion of urinary P was inversely associated with urine Ca to P ratio. Gestating and lactating sows with a urinary Ca to P ratio > 1.5 excreted less than 1% of total P intake (0.8% to 1.6% of STTD P intake). However, sows with a urine Ca to P ratio < 0.5 excreted approximately 14% and 8% of total P intake (22% and 12% of STTD P) in urine during gestation and lactation phases, respectively.

Table 1.

Mean (± SE) 24-h urine Ca to P ratio and urinary P excretion as a percentage of total and STTD P intake for each urine Ca to P ratio category

Urine Ca to P category	P intake	Urine samples, n1	24-h urine Ca to P ratio	Urine P excretion
				% total P intake2	% STTD P intake2
Gestation
>1.5	Deficient	8	8 ± 3	0.8 ± 0.2	1.6 ± 0.4
0.5 to 1.5	Adequate	7	0.69 ± 0.08	11 ± 2	20 ± 4
<0.5	Excessive	53	0.16 ± 0.01	13.6 ± 0.8	22 ± 1
Lactation
>1.5	Deficient	27	74 ± 12	0.4 ± 0.1	0.8 ± 0.2
0.5 to 1.5	Adequate	10	0.75 ± 0.08	2.7 ± 0.7	4 ± 1
<0.5	Excessive	28	0.22 ± 0.02	7.6 ± 0.6	12 ± 1

¹Total 24-h urine samples collected from 34 sows in gestation days 77.1 ± 2 and 112.4 ± 1, and lactation days 4.5 ± 1 and 18.2 ± 1.

²Total P and standardized total tract digestible (STTD) P intakes were calculated using feed intake measurements, formulated concentrations of total P in the diet, and calculated concentrations of STTD P in the diet.

Association between Ca to P ratio in 24-h and spot urine samples

The 24-h urine Ca to P ratio was highly correlated with the spot urine Ca to P ratio (r = 0.93, P < 0.001). Time of day for collection of the spot sample did not account for a significant amount of variation in the 24-h urinary Ca to P ratio. Thus, time of day for spot sample collection was removed from the final model during stepwise variable selection. The model including Ca to P ratio in spot samples and physiological stage of sows provided the best fit for 24-h urine Ca to P ratio (AIC = 131.6, adjusted-R² = 0.87, P < 0.001). However, physiological stage of sows explained only 1% of the variation in 24-h urine Ca to P ratio (based on adjusted-R²). Therefore, the selected final model (Figure 3) only included urine Ca to P ratio in spot samples (AIC = 139.3, adjusted-R² = 0.86, P < 0.001).

Figure 3.

Regression plot showing the relationship between natural log-transformed (Ln) 24-h urine Ca to P ratio (n = 36 samples from 18 sows) and Ln spot urine Ca to P ratio (n = 108 from 18 sows). R²_adj, adjusted coefficient of determination; r, correlation coefficient.

Similar percentages of urine samples were classified into the different Ca to P ratio categories using 24-h or spot urine samples (Table 2). The Cohen’s weighted kappa value was 0.78 (95% CI = 0.69 to 0.88), which indicates substantial agreement between the two methods.

Table 2.

Percentage of urine samples classified into the different Ca to P ratio categories using 24-h and spot samples, mean (± SE) of urine Ca to P ratio in 24-h and spot samples, and weighted kappa

		24-h1		Spot1
Urine Ca to P category	P intake	Urine samples, %	Ca to P ratio	Urine samples, %	Ca to P ratio	Kappa
>1.5	Deficient	33	85 ± 23	31	64 ± 16	0.78 (0.69 to 0.88)2
0.5 to 1.5	Adequate	11	0.68 ± 0.07	15	0.98 ± 0.08
<0.5	Excessive	56	0.10 ± 0.02	54	0.13 ± 0.02

¹Total 24-h (n = 36) and spot urine samples (n = 108) collected from 18 sows in gestation day 112.4 ± 1 and lactation day 18.2 ± 1. Spot urine samples were collected at three different time points within each 24-h collection period.

²Cohen’s weighted kappa value for overall agreement between the two methods. Values in parenthesis represent 95% confidence interval.

Discussion

Results from our previous study (Grez-Capdeville and Crenshaw, 2021), conducted to determine P requirements in reproducing sows, indicated that urinary P and Ca excretion exhibit an inverse response to dietary P intake. The use of urinary Ca to P ratio was proposed as a plausible indicator of dietary P intake adequacy in sows during gestation and lactation. In the current study, the use of urinary Ca to P ratio measurements in spot urine samples proved to be a valid method to classify sows into deficient, adequate, or excessive P intake categories across production phases, time of day that samples were collected, and across a range of dietary P concentrations.

The measurement of Ca to P ratio in urine samples has earlier been proposed as an indicator to evaluate dietary mineral imbalances in growing pigs (Hagemoser et al., 2000) and humans (Senterre and Salle, 1988). In these studies, researchers concluded that a value of urinary Ca to P ratio greater than 1 was associated with P deficiency. In line with this, we observed that sows fed below the estimated P requirements had a higher urinary Ca to P ratio compared with sows fed adequate or excessive dietary P. Several previous studies with growing pigs reported that P-deficient diets resulted in increased amounts of Ca excreted in urine (8% to 25% of Ca intake), independently of dietary Ca levels (Vipperman et al., 1974; Pointillart et al., 1986; Fernández, 1995). In the current study, experimental diets were formulated based on total P concentrations instead of using STTD P. One assumption might be that observed responses of elevated urinary Ca excretion were a consequence of dietary Ca to STTD P ratio imbalance. However, in a previous study with growing pigs (Gutierrez et al., 2015), the same pattern for urinary Ca and P excretion was observed when pigs were fed incremental concentrations of dietary STTD P with constant Ca to STTD P across diets.

The physiological mechanisms involved in the calciuric effect of P-deficient diets are not well defined. Possible regulations of intestinal absorption, renal reabsorption, and bone resorption have been described in dogs (Coburn and Massry, 1970). However, apparent Ca absorption was unaltered in pigs fed P-deficient diets (Pointillart et al., 1987), which infers that the increase in urinary Ca excretion is likely associated with impaired ability to retain the absorbed Ca in the body. Despite the specific mechanisms involved in this physiological response, the excess of circulating Ca is excreted in urine. Because P is absorbed and retained more efficiently at lower P intakes (Vipperman et al., 1974; Saddoris et al., 2010), a condition of P depletion would be expected to result in a high urinary Ca to P ratio. In the current study, a cutoff value for urinary Ca to P of 1.5 for P-deficient diets (sensitivity 94% and specificity 68%) was determined. In this study, data analysis using previously suggested cutoff value of 1 yielded a slightly lower sensitivity (93%) and greater specificity (70%). In terms of sow health and performance, a greater sensitivity is desirable to minimize false-negative results that can lead to P deficiencies.

On the other hand, we found that urine Ca to P ratios less than 0.5 (sensitivity 82% and specificity 82%) were associated with excessive dietary P intake. Previous studies have shown that urinary Ca excretion is marginal in pigs fed adequate or high-P diets (Miller et al., 1964; Vipperman et al., 1974; Fernández, 1995; Sørensen et al., 2018). In agreement with our findings, calculated urine Ca to P ratios from reported measurements of urinary Ca and P in those studies were low (0.01 to 0.3). Hagemoser et al. (2000) hypothesized that a low urine Ca to P ratio may indicate excessive P intake. However, they also postulated that an extremely low urine Ca to P ratio (<0.05) can be associated with Ca-restricted diets. A low excretion of Ca in urine was observed in sows fed a Ca-free diet (Lee et al., 2019), resulting in a low calculated urinary Ca to P ratio associated with levels of Ca intake rather than excessive P intake. However, sources of Ca used in swine diets are relatively inexpensive and have high Ca bioavailability (Ross et al., 1984). Thus, in practical applications, Ca over-supplementation is more likely to occur rather than Ca under-supplementation.

Our estimated cutoff values for deficiencies, adequacies, and excesses are consistent with results reported in a balance study with gestating sows (Everts et al., 1998a). Calculations of the urine Ca to P ratio in sows fed the control diet averaged 0.45 compared with 1.84 in sows fed P-deficient diets. Unfortunately, only a few Ca and P balance studies have been conducted in reproducing sows, and measurements of urinary Ca and P concentrations are generally not reported (Harmon et al., 1975; Everts et al., 1998b; Tan et al., 2016). This limits the validation of our findings with prior studies.

In our experiments the total 24-h urine collections were assumed to provide the standard for urine analysis. However, complete 24-h urine specimens are impractical to collect in large-scale studies. Our results showed that urine Ca to P ratios in total 24-h samples were highly correlated with Ca to P ratios measured in spot urine samples. Furthermore, this association was not influenced by physiological state and collection time of spot urine samples. The agreement between spot and 24-h urine Ca to P ratio analysis in categorizing P intake adequacy was substantial. Altogether these results provide evidence that Ca to P ratio measured in random spot urine samples adequately reflect Ca to P ratio in 24-h urine collections. Therefore, outcomes in our current study demonstrate the reliability of using urine Ca to P ratio as a practical method to assess P intake adequacy in sows.

Conclusions

Measurements of Ca to P ratio in urine samples represent a reliable indicator of P intake adequacy in sows. Urinary Ca to P ratio > 1.5 is associated with P-deficient diets, whereas a urine Ca to P ratio < 0.5 reflects excessive P intake. The agreement between Ca to P ratio measured in 24-h and spot urine samples was not affected by the physiological phase and time of the day of spot urine samples were collected. Therefore, evaluation of Ca to P ratio in random spot urine samples may be considered as a practical method for mass screening of P intake adequacy in sow herds. Future studies are needed to provide additional support for the proposed cutoff values for urinary Ca to P ratio, and to verify that relationships are consistent across herds with varied health status and genetic profiles, or in sows fed diets composed with different ingredients. The current results provide a basis for future large-scale screening of P intake adequacy and future approaches for development of precision feeding strategies applied to individual sows.

Glossary

Abbreviations

AIC

Akaike Information Criteria

ROC

receiver operating characteristic

STTD P

standardized total tract digestible P

Funding

This project was financially supported by the University of Wisconsin Hatch fund (142-AAB4963) and unrestricted research funds (233-Q893) provided by DSM Nutritional Products North America and Genus PIC.

Conflict of interest statement

The authors declare no real or perceived conflicts of interest.