Archs oral 8~1. Vol 22, up. 83 to 86 Pergamon
Press 1977 Prmted in Great Britam.
HUMAN PAROTID SALIVA UREA IN RENAL FAILURE AND DURING DIALYSIS I. L. SHANNON,R. P. FELLER,G. EKNOYANand R. P. SUDDICK* Oral Disease Research Laboratory, VA Hospital, Baylor College of Medicine, University of Texas Dental Branch, Houston, Texas 77211, U.S.A.
Summary-Parotid saliva and plasma urea nitrogen concentrations were compared by saliva:plasma ratios and correlation coefficients under widely differing plasma urea levels, in patients with renal failure and in normal subjects. A near-perfect correlation (r = 0.97) was found in 56 pairs of samples from uraemic patients, with the ratio remaining constant at widely varying plasma urea levels. In unstimulated saliva of normal subjects, urea nitrogen levels were significantly higher than in plasma (ratio = 1.3); the reverse was true in stimulated saliva. The average ratio for all subjects at an average flow rate of 0.5 ml/gland/min was 0.66. The highest correlation coefficients were at the highest stimulated flow rates.
INTRODUffION Urea was first detected in human saliva more than 130 years ago (Wright, 1841). About 50 years ago, salivary urea was found to be increased during impaired renal function (Schmitz, 1922), and a salivary urea index was proposed as an indicator of renal function (Hench and Aldrich, 1922). Although several investigators (Morris and Way, 1924; Corkhill, 1925; Nikiforuk et al., 1956; Forland, Shannon and Katz, 1964) have reported that the salivary urea level varies in proportion to the blood level, others (Stealy, 1928; Ferris, Smith and Graves, 1923; Ferris, 1920) have contested the presence of a consistent correlation. Stealy (1928) reported a very poor correlation between salivary urea and blood urea in disease states. Ferris et al. (1923) found urea in the saliva of only 3 of 10 patients, and Ferris (1920) concluded that the presence of urea in saliva indicated kidney disease. Our study was designed to explore the potential of salivary urea measurements in predicting blood urea concentration in patients with impaired renal function. Parotid saliva urea and creatinine, as well as the major electrolytes, were examined, before and after dialysis, in six subjects by Dahlberg, Sreebny and King (19671. MATERIALSAND METHODS Parotid saliva was collected with a cup held over the parotid duct opening by vacuum (Shannon and Chauncey, 1967’1.Samples were collected either under resting (unstimulated) conditions or during gustatory stimulation with peppermint-, cherry- and grape-flavoured hard candy which, in that order, produce an ascending pattern of flow rates. Saliva was collected in graduated tubes and volume was read to the nearest 0.05 ml. Urea nitrogen (UN) was measured in both saliva and serum by the diacetyl monoxime method
(Ormsby, 1942; Rosenthal, 1955) as adapted for automatic processing (Shannon and Prigmore, 1960). In the study of impaired renal function, both venous blood and grape candy-stimulated parotid saliva samples were collected from 12 patients with terminal stage kidney disease on maintenance haemodialysis. Samples were collected prior to, at the midpoint and immediately upon completion of haemodialysis. The average dialysis time was about 5 h. Similar samples were obtained from 20 patients with renal failure not undergoing haemodialysis. These individuals presented various levels of impaired kidney function. The single criterion for inclusion in the study was a plasma UN level exceeding 18.0 mg/decilitre (dl). The relationship between plasma and saliva urea, and the effect of parotid saliva flow rate on urea concentration were also studied in 56 healthy subjects. Each participant provided a single 90-min unstimulated parotid saliva sample, followed by a series of three lo-min stimulated (peppermint, cherry, grape candy) samples. Each test sample in this stimulated series was preceded by a 5-min accommodation period to allow the gland to adjust to the new stimulus. A single venous blood sample was drawn after saliva sampling. The reproducibility of saliva urea levels in successive daily grape candy-stimulated parotid saliva samples was tested in nine healthy subjects. On five successive days, each subject provided a series of eight 5-min saliva samples. A separate test of the effects of extended grape-stimulated secretion on salivary urea levels was accomplished by taking a series of nine consecutive 20-min parotid saliva samples elicited by grape candy, the highest mode of reflex stimulation. RESULTS The mean ( f S.D.) serum urea nitrogen level for the 12 patients undergoing dialysis was 81.2 k 30.9 mg/dl before dialysis, decreasing significantly (p < 0.01) to 58.3 f 23.3 mg/dl at the midpoint of dialysis, and to 44.4 +_23.5 mg/dl upon completion.
* Medical University of South Carolina, Charleston, South Carolina; present address: Department of Oral Biology, University Iof Louisville School of Dentistry, Louisville, Kentucky, 1J.S.A. 83
I. L. Shannon, R. P. Feller, G. Eknoyan and R. P. Suddick
S
c
0100 0 20 40 60 80 PAROTIO SALIVA UREA NITROGEN [mg/lWml)
Fig.
2. Urea
Fig. 1. Serum and stimulated parotid saliva urea nitrogen levels in patients undergoing haemodialysis.
Comparable means (+S.D.) for parotid saliva UN were 59.8 f 21.6 mg/dl, 42.6 f 18.5 mg/dl and 31.8 f 16.0 mg/dl; each difference was significant at the 1 per cent level. Parotid saliva flow rates did not differ significantly at the 3 sampling intervals; the mean values (*SD.) were 0.80 f 0.37 ml/min, 0.70 k 0.76 ml/min and 0.76 f 0.52 ml/min. For all samples from all haemodialysis patients, the mean (*SD.) serum UN was 61.3 f 29.7 mg/dl and that for parotid saliva was 44.7 f 21.6 mg/dl. The correlation coefficient between these two fluids was 0.953 (p < 0.001) (Fig. 1). In 20 uraemic inpatients not undergoing haemodialysis, the individual serum UN levels ranged from 19.6 mg/dl to 222.1 mg/dl with a mean of 67.3 mg/dl (S.D. = 49.0). Parotid saliva UN concentrations ranged from 15.5 mg/dl to 155.8 mg/dl and the mean was 48.2 mg/dl (S.D. = 34.3). The mean parotid flow rate for all subjects was 0.74 ml/min (SD. = 0.52). There was an extremely high correlation (r = 0.99) between serum and parotid UN levels (Fig. 2). These findings were in accord with the high correlation between serum and saliva urea concentrations found in the dialysis patients. The data from this experiment were pooled with those from the haemodialysis study, thus providing a total of 56 paired parotid saliva and serum UN measurements for patients with kidney disorders. The correlation coefficient between serum and parotid saliva UN concentrations was again very high
0 40 00 120 160 200 PAROTID SALIVA UREA NllR06EN [mg/lOO ml] nitrogen in serum and stimulated saliva in 20 uraemic patients.
(r = 0.97) indicating a near perfect relationship between these two variables. The urea saliva : plasma ratio for all uraemic patients averaged 0.72. In 56 normal subjects, the mean serum UN was 15.0 mg/dl (S.D. = 3.0). Changing exogenous stimulants induced significant increases in flow rates and a trend towards a decrease in saliva UN values (Table 1). The mean unstimulated parotid saliva UN was 19.4 mg/dl (SD. = 5.0), resulting in a saliva : plasma urea ratio of 1.3. In all subjects, saliva UN was higher than serum UN. Thus, the unstimulated saliva UN not only exceeded the serum UN level but was also significantly higher than that found in saliva at any of the three levels of exogenous stimulation. In both unstimulated and stimulated secretions, there was a highly significant (p < 0.001) positive correlation between the concentrations of urea in parotid saliva and serum. The individual subject correlation coefficients between the unstimulated, peppermint-, cherry- and grape-flavour stimulated saliva UN and the serum UN were 0.75, 0.74, 0.86 and 0.86, respectively. Thus, higher correlations between urea levels in saliva and serum were found at the higher rates of stimulated flow. However, the correlations were not as high as in the uraemic patients; the reasons for this appear to be both methodological and statistical. In the uraemic patients, a very wide range of plasma (and thus salivary) urea concentrations was available for sampling. Furthermore, in the haemodialysis study, blood samples were taken before, during and at the end of dialysis. In the normal subjects,
Table 1. Effect of flow rate on parotid saliva UN concentrations (56 subjects); UN for these subjects = 15.0 mg/dl (SD. = 3.0)
Exogenous None Peppermint* Cherry* Grape?
stimulant
Parotid Mean
flow rate (ml/min) S.D.
0.031 0.420 0.672 0.915
0.018 0.250 0.351 0.415
* Life Savers, Inc., New York t Albert & Son, Inc., New York
parotid
Urea nitrogen Mean 19.4 10.1 8.9 9.0
mean serum
(mg/dl) SD. 5.02 2.64 2.15 2.07
Human saliva urea in renal failure one blood sample was taken before saliva sampling was begun, and this serum UN was used for comparison to the unstimulated, and all stimulated saliva samples. Lines of best tit were calculated for the prediction of serum urea from parotid urea values (Table 2). These results indicate that the collection method of choice for such a prediction is by means of a stimulant which produces a flow rate greater than 0.5 ml/min. Based upon the equations for the two highest levels of stimulation, 95 per cent confidence intervals for predicting serum urea concentrations were computed. For example, if a saliva collection was made from a patient with cherry candy and the resulting parotid saliva UN levels was 7.0 mg/dl, it could be predicted with 95 per cent confidence that the serum UN value would fall in the range of 9.6 and to 15.9 mg/dl. The average urea saliva:plasma ratio in the normal subjects in the cherry and grape-stimulated secretions was approximately 0.60. Over five consecutive days, each of 9 normal subjects provided 8 consecutive 5 min grape-stimulated parotid saliva samples each day (total period of stimulation = 40 min) (Table 3). The overall mean flow rate for the 360 samples was 0.54 ml/min. The saliva UN for all subjects was 9.9 mg/dl and the individual subject means ranged from 9.4 to 10.2 mg/dl. Within subjects, there was no statistically significant difference for either flow rates or saliva UN concentrations between these samples. In testing the effects of extended high level stimulation with grape candy, the mean flow rate for the 9 subjects for all samples over the 3 h period was 0.93 ml/min (SD. = 0.32) and the mean saliva UN level was 7.5 mgidl (S.D. = 3.0). The regression equations for flow rate and UN were calculated and tested for significance. There was no significant increase or decrease in either flow rate or UN concentration associated with prolonged stimulation (Table 3). DISCUSSION Although it was reported long ago that salivary urea levels can be used clinically in the diagnosis of uraemia (Morris and Way, 1924; Corkhill, 1925; Nikiforuk et al., 1956; Forland et al., 1964), this method has not gained wide acceptance as a routine clinical procedure. One Idifficulty of certain earlier studies was that whole saliva was employed. Bacterial contamiTable 3. Effects of prolonged
stimulation
85
Table 2. Estimation of serum urea nitrogen level (y) from parotid saliva concentration at different levels of stimulation Hard sour candy stimulant
Line of best fit y y y y
None Peppermint Cherry Grape
= = = =
6.34 6.53 4.37 3.90
+ + + +
0.448 (x) 0.843 (x) 1.190(x) 1.240(x)
Residual error 3.96 4.07 2.37 2.39
nation and resultant urease activity produced confusing results unless corrections were made by measuring ammonia nitrogen. Use of pure parotid saliva obviates these problems. It can be incubated at 37°C for 24 h without loss of urea (Albrectsen and Thaysen, 1955). In that study, a saliva:plasma ratio (0.65) which seemed to be relatively independent of large variations in plasma urea levels was calculated for 4 uraemic patients. Our study corroborates and provides additional information to the earlier dialysis study of Dahlberg et al. (1967). These investigators sampled parotid saliva only under stimulated conditions, and reported saliva:plasma urea ratios of 0.75 before dialysis and 0.70 after dialysis. Thus their values were very comparable to the overall average ratio of 0.72 in the stimulated secretions of all of our uraemic patients. The ratio for the normal subjects at parotid how rates greater than 0.5 ml/min was 0.60. For diagnostic purposes, we suggest that the average of these two ratios (0.66) be taken as the normal reference, and that the reciprocal of this ratio can be safely used, as a simple method, to estimate serum UN from parotid values (parotid UN x 1.5 = approximately serum UN), provided parotid flow rates are maintained at 0.5 ml/min or more during sampling. We believe there is now sufficient evidence to justify clinical use of parotid saliva UN concentrations for monitoring blood UN levels, for example, during haemodialysis. In addition, the data indicate that alternative interpretations should be offered for the cellular mechanisms by which urea is thought to enter saliva. In unstimulated parotid saliva, the saliva:plasma urea ratio was 1.3; yet urea is not known to be actively transported by the salivary glands and is (grape-flavoured
hard sour candy)
Number of subjects*
Duration of stimulation
Overall mean flow rate
Saliva urea nitrogen (UN)
Flow rate and UN : statistical analyses within subjects
9
40 min
0.54 ml/min (S.D. = 0.18)t
9.9 mg/dl (S.D. = 3.8)
No significant differences
9
3h
0.93 ml/min (S.D. = 0.32)$
7.5 mg/dl (S.D. = 2.3)
No significant differences
* Subjects were studied over 5 consecutive days. t 8 consecutive 5 min samples each day; N = 360. $9 consecutive 20 min samples each day; N = 405.
86
I. L. Shannon, R. P. Feller, G. Eknoyan and R. P. Suddick
believed to be managed passively in mammalian cells in general. The higher urea concentration in the unstimulated saliva could be simply explained on the basis of urea diffusion into saliva in the acini or proximal regions of the ducts, together with sodium and water but not urea reabsorption in the distal regions of the ducts. This would be consistent with the evidence for ductal sodium and water reabsorption offered by Shannon, Suddick and Chauncey (1969). It is interesting that the urea saliva:plasma ratio decreased while the saliva-plasma correlation coefficient increased with higher rates of flow. This is evident in comparing the results of the daily sampling reproducibility tests and the effects of extended stimulation. In the studies of urea behaviour during extended stimulation, at the mean stimulated flowrate of 0.54 ml/min, the saliva UN was 9.9 mg/dl, while at the higher rate of 0.9 ml/min, the UN was significantly lower at 7.5 mg/dl (Table 3). In the studies on 56 normal subjects, the highest salivaplasma correlation coefficients (0.86) were at the two highest flow rates, and the saliva UN values were lowest at these flow rates (Table 1). These findings are not consistent with the notion that urea enters the ducts by simple diffusion alone which is the mechanism offered by Burgen (1967) to account for its transfer into the ducts during active secretion. If diffusion is the sole mechanism by which urea enters the ducts, the saliva-plasma correlation coefficients should be independent of the salivary flow rate. Our results support the concept that urea, during physiological stimulation of secretion, may be moving in bulk flow with water through channels or pores of a diameter which would be partially restrictive to urea compared to water. The fact that the saliva urea concentration is lower at the highest rates of flow, and yet is more predictive of the plasma concentration, indicates that the plasma to saliva pathway for urea is consistently more direct at the highest flow rates and thus that entry by bulk flow becomes more predominant as flow rate increases. This may help explain the observation that the urea clearance appears to remain at a constant value in the stimulated secretions while plasma urea concentration changes. This is evident from two facts: (1) the saliva :plasma urea ratios remained constant in the patients undergoing haemodialysis (while the plasma urea was constantly decreasing), (2) the stimulated parotid flow rates of these patients remained unchanged throughout the dialysis procedure. Altogether, these findings are very consistent with the concepts of fluid secretion proposed by Suddick (1973); Shannon, Suddick and Dowd (1974). We do not suggest, however, that the described urea behaviour during stimulated secretion could not be explained by application of the more widely accepted concepts of fluid secretion reviewed by Schneyer, Young and Schneyer (1972).
REFERENCES
Albrectsen S. T. and Thaysen J. H. 1955. The excretion of parotid gland. Stand. f. &I. Lab. Invesr. 7, 231-238. Burgen A. S. V. 1967. Secretory processes in salivary glands. In: Handbook of Physiology, Alimentary Canal, Volume II. Secretion. (Edited by Code C. F.) pp. 561-579. Waverly Press, Baltimore. Corkhill A. B. 1925. The estimation of salivary urea as an index to renal prognosis, Med. J. Aust. 10, 236238. Dahlberg W. H., Sreebny L. M. and King B. 1967. Studies of parotid saliva and blood in hemodialysis patients, J. ~cppl. Physiol. 23, IO&108 Ferris H. C. 1920. Mouth hygiene, controlled by diet through salivary analysis. Dent. Cosmos 62, 453464. Ferris H. C., Smith E. D. and Graves E. V. 1923. Physiological and pathological variations in the composition of human saliva. J. Am. Dent. Ass. 10. 19-52. Forland M., Shannon I. L. and Katz F. H. 1964. Parotid fluid urea nitrogen for monitoring hemodialysis. New Enyl. .I. Med. 271, 27-38. Hench P. S. and Aldrich M. 1922. A salivary index to renal function. J. Am. Med. Ass. 81, 1977-2003. Morris J. L. and Way C. T. 1924. Further observations on chemical constituents of saliva. J. biol. Chem. 59, xxvi-xxvii. Nikiforuk G., Jackson S. J., Cox M. A. and Grainger R. M. 1956. Some blood and salivary nonprotein nitrogen constituents in children, and dental caries. J. Pediat. 49, 425431. Ormsby A. S. 1942. A direct calorimetric method for the determination of urea in blood and urine. J. biol. Chem. 246, 5955604. Rosenthal H. L. 1955. Determination of urea in blood and urine with diacetyl monoxime. Amdyr. Chem. 27, 198@1982. Schmitz H. S. 1922. Comparative concentration of urea in blood and saliva in a series of pathologic cases. J. Lab. c/in. Med. 8, 78 82. Schneyer L. H., Young J. A. and Schneyer C. A. 1972. Salivary secretion of electrolytes. Physiol. Rev. 52. 720-777. Shannon I. L. and Chauncey H. H. 1967. A parotid fluid collection device with improved stability characteristics. J. Oral Ther. Pharm. 4. 93-97. Shannon I. L., Suddick R. P. and Chauncey H. H. 1969. Etrect of atropine-induced flow rate depression on the composition of unstimulated parotid fluids, Archs orul Biol. 14. 761-770. Shannon I. L. and Prigmore J. R. 1960. The use of the Auto Analyzer in studies of the physiological chemistry of the parotid gland. Ann. N.Y. Acad. Sci. 87. 745-763. Shannon I. L., Suddick R. P. and Dowd F. J. 1974. Saliva: composition and secretion. In: Monographs in Oral Science, (Edited by Myers H. M.), Kargkr, Basef. Stealv C. L.. 1928. The clinical value of the salivarv urea index. J. iab. clin. Med. 14, 162-165. Suddick R. P., 1973. Does transepithelial hydrostatic pressure provide energy for fluid secretion? In: Mechanisms of’E.yocrine Secretion. (Edited by Han S. S., Sreebny L. and Suddick R. P.) pp. 4460. University of Michigan Press. Wright S. 1841-1842. Case of ascites in which, during spontaneous ptyalism that occurred after tapping, urea was detected in saliva. Lancet 1, 753-758.
Press 1977 Prmted in Great Britam.
HUMAN PAROTID SALIVA UREA IN RENAL FAILURE AND DURING DIALYSIS I. L. SHANNON,R. P. FELLER,G. EKNOYANand R. P. SUDDICK* Oral Disease Research Laboratory, VA Hospital, Baylor College of Medicine, University of Texas Dental Branch, Houston, Texas 77211, U.S.A.
Summary-Parotid saliva and plasma urea nitrogen concentrations were compared by saliva:plasma ratios and correlation coefficients under widely differing plasma urea levels, in patients with renal failure and in normal subjects. A near-perfect correlation (r = 0.97) was found in 56 pairs of samples from uraemic patients, with the ratio remaining constant at widely varying plasma urea levels. In unstimulated saliva of normal subjects, urea nitrogen levels were significantly higher than in plasma (ratio = 1.3); the reverse was true in stimulated saliva. The average ratio for all subjects at an average flow rate of 0.5 ml/gland/min was 0.66. The highest correlation coefficients were at the highest stimulated flow rates.
INTRODUffION Urea was first detected in human saliva more than 130 years ago (Wright, 1841). About 50 years ago, salivary urea was found to be increased during impaired renal function (Schmitz, 1922), and a salivary urea index was proposed as an indicator of renal function (Hench and Aldrich, 1922). Although several investigators (Morris and Way, 1924; Corkhill, 1925; Nikiforuk et al., 1956; Forland, Shannon and Katz, 1964) have reported that the salivary urea level varies in proportion to the blood level, others (Stealy, 1928; Ferris, Smith and Graves, 1923; Ferris, 1920) have contested the presence of a consistent correlation. Stealy (1928) reported a very poor correlation between salivary urea and blood urea in disease states. Ferris et al. (1923) found urea in the saliva of only 3 of 10 patients, and Ferris (1920) concluded that the presence of urea in saliva indicated kidney disease. Our study was designed to explore the potential of salivary urea measurements in predicting blood urea concentration in patients with impaired renal function. Parotid saliva urea and creatinine, as well as the major electrolytes, were examined, before and after dialysis, in six subjects by Dahlberg, Sreebny and King (19671. MATERIALSAND METHODS Parotid saliva was collected with a cup held over the parotid duct opening by vacuum (Shannon and Chauncey, 1967’1.Samples were collected either under resting (unstimulated) conditions or during gustatory stimulation with peppermint-, cherry- and grape-flavoured hard candy which, in that order, produce an ascending pattern of flow rates. Saliva was collected in graduated tubes and volume was read to the nearest 0.05 ml. Urea nitrogen (UN) was measured in both saliva and serum by the diacetyl monoxime method
(Ormsby, 1942; Rosenthal, 1955) as adapted for automatic processing (Shannon and Prigmore, 1960). In the study of impaired renal function, both venous blood and grape candy-stimulated parotid saliva samples were collected from 12 patients with terminal stage kidney disease on maintenance haemodialysis. Samples were collected prior to, at the midpoint and immediately upon completion of haemodialysis. The average dialysis time was about 5 h. Similar samples were obtained from 20 patients with renal failure not undergoing haemodialysis. These individuals presented various levels of impaired kidney function. The single criterion for inclusion in the study was a plasma UN level exceeding 18.0 mg/decilitre (dl). The relationship between plasma and saliva urea, and the effect of parotid saliva flow rate on urea concentration were also studied in 56 healthy subjects. Each participant provided a single 90-min unstimulated parotid saliva sample, followed by a series of three lo-min stimulated (peppermint, cherry, grape candy) samples. Each test sample in this stimulated series was preceded by a 5-min accommodation period to allow the gland to adjust to the new stimulus. A single venous blood sample was drawn after saliva sampling. The reproducibility of saliva urea levels in successive daily grape candy-stimulated parotid saliva samples was tested in nine healthy subjects. On five successive days, each subject provided a series of eight 5-min saliva samples. A separate test of the effects of extended grape-stimulated secretion on salivary urea levels was accomplished by taking a series of nine consecutive 20-min parotid saliva samples elicited by grape candy, the highest mode of reflex stimulation. RESULTS The mean ( f S.D.) serum urea nitrogen level for the 12 patients undergoing dialysis was 81.2 k 30.9 mg/dl before dialysis, decreasing significantly (p < 0.01) to 58.3 f 23.3 mg/dl at the midpoint of dialysis, and to 44.4 +_23.5 mg/dl upon completion.
* Medical University of South Carolina, Charleston, South Carolina; present address: Department of Oral Biology, University Iof Louisville School of Dentistry, Louisville, Kentucky, 1J.S.A. 83
I. L. Shannon, R. P. Feller, G. Eknoyan and R. P. Suddick
S
c
0100 0 20 40 60 80 PAROTIO SALIVA UREA NITROGEN [mg/lWml)
Fig.
2. Urea
Fig. 1. Serum and stimulated parotid saliva urea nitrogen levels in patients undergoing haemodialysis.
Comparable means (+S.D.) for parotid saliva UN were 59.8 f 21.6 mg/dl, 42.6 f 18.5 mg/dl and 31.8 f 16.0 mg/dl; each difference was significant at the 1 per cent level. Parotid saliva flow rates did not differ significantly at the 3 sampling intervals; the mean values (*SD.) were 0.80 f 0.37 ml/min, 0.70 k 0.76 ml/min and 0.76 f 0.52 ml/min. For all samples from all haemodialysis patients, the mean (*SD.) serum UN was 61.3 f 29.7 mg/dl and that for parotid saliva was 44.7 f 21.6 mg/dl. The correlation coefficient between these two fluids was 0.953 (p < 0.001) (Fig. 1). In 20 uraemic inpatients not undergoing haemodialysis, the individual serum UN levels ranged from 19.6 mg/dl to 222.1 mg/dl with a mean of 67.3 mg/dl (S.D. = 49.0). Parotid saliva UN concentrations ranged from 15.5 mg/dl to 155.8 mg/dl and the mean was 48.2 mg/dl (S.D. = 34.3). The mean parotid flow rate for all subjects was 0.74 ml/min (SD. = 0.52). There was an extremely high correlation (r = 0.99) between serum and parotid UN levels (Fig. 2). These findings were in accord with the high correlation between serum and saliva urea concentrations found in the dialysis patients. The data from this experiment were pooled with those from the haemodialysis study, thus providing a total of 56 paired parotid saliva and serum UN measurements for patients with kidney disorders. The correlation coefficient between serum and parotid saliva UN concentrations was again very high
0 40 00 120 160 200 PAROTID SALIVA UREA NllR06EN [mg/lOO ml] nitrogen in serum and stimulated saliva in 20 uraemic patients.
(r = 0.97) indicating a near perfect relationship between these two variables. The urea saliva : plasma ratio for all uraemic patients averaged 0.72. In 56 normal subjects, the mean serum UN was 15.0 mg/dl (S.D. = 3.0). Changing exogenous stimulants induced significant increases in flow rates and a trend towards a decrease in saliva UN values (Table 1). The mean unstimulated parotid saliva UN was 19.4 mg/dl (SD. = 5.0), resulting in a saliva : plasma urea ratio of 1.3. In all subjects, saliva UN was higher than serum UN. Thus, the unstimulated saliva UN not only exceeded the serum UN level but was also significantly higher than that found in saliva at any of the three levels of exogenous stimulation. In both unstimulated and stimulated secretions, there was a highly significant (p < 0.001) positive correlation between the concentrations of urea in parotid saliva and serum. The individual subject correlation coefficients between the unstimulated, peppermint-, cherry- and grape-flavour stimulated saliva UN and the serum UN were 0.75, 0.74, 0.86 and 0.86, respectively. Thus, higher correlations between urea levels in saliva and serum were found at the higher rates of stimulated flow. However, the correlations were not as high as in the uraemic patients; the reasons for this appear to be both methodological and statistical. In the uraemic patients, a very wide range of plasma (and thus salivary) urea concentrations was available for sampling. Furthermore, in the haemodialysis study, blood samples were taken before, during and at the end of dialysis. In the normal subjects,
Table 1. Effect of flow rate on parotid saliva UN concentrations (56 subjects); UN for these subjects = 15.0 mg/dl (SD. = 3.0)
Exogenous None Peppermint* Cherry* Grape?
stimulant
Parotid Mean
flow rate (ml/min) S.D.
0.031 0.420 0.672 0.915
0.018 0.250 0.351 0.415
* Life Savers, Inc., New York t Albert & Son, Inc., New York
parotid
Urea nitrogen Mean 19.4 10.1 8.9 9.0
mean serum
(mg/dl) SD. 5.02 2.64 2.15 2.07
Human saliva urea in renal failure one blood sample was taken before saliva sampling was begun, and this serum UN was used for comparison to the unstimulated, and all stimulated saliva samples. Lines of best tit were calculated for the prediction of serum urea from parotid urea values (Table 2). These results indicate that the collection method of choice for such a prediction is by means of a stimulant which produces a flow rate greater than 0.5 ml/min. Based upon the equations for the two highest levels of stimulation, 95 per cent confidence intervals for predicting serum urea concentrations were computed. For example, if a saliva collection was made from a patient with cherry candy and the resulting parotid saliva UN levels was 7.0 mg/dl, it could be predicted with 95 per cent confidence that the serum UN value would fall in the range of 9.6 and to 15.9 mg/dl. The average urea saliva:plasma ratio in the normal subjects in the cherry and grape-stimulated secretions was approximately 0.60. Over five consecutive days, each of 9 normal subjects provided 8 consecutive 5 min grape-stimulated parotid saliva samples each day (total period of stimulation = 40 min) (Table 3). The overall mean flow rate for the 360 samples was 0.54 ml/min. The saliva UN for all subjects was 9.9 mg/dl and the individual subject means ranged from 9.4 to 10.2 mg/dl. Within subjects, there was no statistically significant difference for either flow rates or saliva UN concentrations between these samples. In testing the effects of extended high level stimulation with grape candy, the mean flow rate for the 9 subjects for all samples over the 3 h period was 0.93 ml/min (SD. = 0.32) and the mean saliva UN level was 7.5 mgidl (S.D. = 3.0). The regression equations for flow rate and UN were calculated and tested for significance. There was no significant increase or decrease in either flow rate or UN concentration associated with prolonged stimulation (Table 3). DISCUSSION Although it was reported long ago that salivary urea levels can be used clinically in the diagnosis of uraemia (Morris and Way, 1924; Corkhill, 1925; Nikiforuk et al., 1956; Forland et al., 1964), this method has not gained wide acceptance as a routine clinical procedure. One Idifficulty of certain earlier studies was that whole saliva was employed. Bacterial contamiTable 3. Effects of prolonged
stimulation
85
Table 2. Estimation of serum urea nitrogen level (y) from parotid saliva concentration at different levels of stimulation Hard sour candy stimulant
Line of best fit y y y y
None Peppermint Cherry Grape
= = = =
6.34 6.53 4.37 3.90
+ + + +
0.448 (x) 0.843 (x) 1.190(x) 1.240(x)
Residual error 3.96 4.07 2.37 2.39
nation and resultant urease activity produced confusing results unless corrections were made by measuring ammonia nitrogen. Use of pure parotid saliva obviates these problems. It can be incubated at 37°C for 24 h without loss of urea (Albrectsen and Thaysen, 1955). In that study, a saliva:plasma ratio (0.65) which seemed to be relatively independent of large variations in plasma urea levels was calculated for 4 uraemic patients. Our study corroborates and provides additional information to the earlier dialysis study of Dahlberg et al. (1967). These investigators sampled parotid saliva only under stimulated conditions, and reported saliva:plasma urea ratios of 0.75 before dialysis and 0.70 after dialysis. Thus their values were very comparable to the overall average ratio of 0.72 in the stimulated secretions of all of our uraemic patients. The ratio for the normal subjects at parotid how rates greater than 0.5 ml/min was 0.60. For diagnostic purposes, we suggest that the average of these two ratios (0.66) be taken as the normal reference, and that the reciprocal of this ratio can be safely used, as a simple method, to estimate serum UN from parotid values (parotid UN x 1.5 = approximately serum UN), provided parotid flow rates are maintained at 0.5 ml/min or more during sampling. We believe there is now sufficient evidence to justify clinical use of parotid saliva UN concentrations for monitoring blood UN levels, for example, during haemodialysis. In addition, the data indicate that alternative interpretations should be offered for the cellular mechanisms by which urea is thought to enter saliva. In unstimulated parotid saliva, the saliva:plasma urea ratio was 1.3; yet urea is not known to be actively transported by the salivary glands and is (grape-flavoured
hard sour candy)
Number of subjects*
Duration of stimulation
Overall mean flow rate
Saliva urea nitrogen (UN)
Flow rate and UN : statistical analyses within subjects
9
40 min
0.54 ml/min (S.D. = 0.18)t
9.9 mg/dl (S.D. = 3.8)
No significant differences
9
3h
0.93 ml/min (S.D. = 0.32)$
7.5 mg/dl (S.D. = 2.3)
No significant differences
* Subjects were studied over 5 consecutive days. t 8 consecutive 5 min samples each day; N = 360. $9 consecutive 20 min samples each day; N = 405.
86
I. L. Shannon, R. P. Feller, G. Eknoyan and R. P. Suddick
believed to be managed passively in mammalian cells in general. The higher urea concentration in the unstimulated saliva could be simply explained on the basis of urea diffusion into saliva in the acini or proximal regions of the ducts, together with sodium and water but not urea reabsorption in the distal regions of the ducts. This would be consistent with the evidence for ductal sodium and water reabsorption offered by Shannon, Suddick and Chauncey (1969). It is interesting that the urea saliva:plasma ratio decreased while the saliva-plasma correlation coefficient increased with higher rates of flow. This is evident in comparing the results of the daily sampling reproducibility tests and the effects of extended stimulation. In the studies of urea behaviour during extended stimulation, at the mean stimulated flowrate of 0.54 ml/min, the saliva UN was 9.9 mg/dl, while at the higher rate of 0.9 ml/min, the UN was significantly lower at 7.5 mg/dl (Table 3). In the studies on 56 normal subjects, the highest salivaplasma correlation coefficients (0.86) were at the two highest flow rates, and the saliva UN values were lowest at these flow rates (Table 1). These findings are not consistent with the notion that urea enters the ducts by simple diffusion alone which is the mechanism offered by Burgen (1967) to account for its transfer into the ducts during active secretion. If diffusion is the sole mechanism by which urea enters the ducts, the saliva-plasma correlation coefficients should be independent of the salivary flow rate. Our results support the concept that urea, during physiological stimulation of secretion, may be moving in bulk flow with water through channels or pores of a diameter which would be partially restrictive to urea compared to water. The fact that the saliva urea concentration is lower at the highest rates of flow, and yet is more predictive of the plasma concentration, indicates that the plasma to saliva pathway for urea is consistently more direct at the highest flow rates and thus that entry by bulk flow becomes more predominant as flow rate increases. This may help explain the observation that the urea clearance appears to remain at a constant value in the stimulated secretions while plasma urea concentration changes. This is evident from two facts: (1) the saliva :plasma urea ratios remained constant in the patients undergoing haemodialysis (while the plasma urea was constantly decreasing), (2) the stimulated parotid flow rates of these patients remained unchanged throughout the dialysis procedure. Altogether, these findings are very consistent with the concepts of fluid secretion proposed by Suddick (1973); Shannon, Suddick and Dowd (1974). We do not suggest, however, that the described urea behaviour during stimulated secretion could not be explained by application of the more widely accepted concepts of fluid secretion reviewed by Schneyer, Young and Schneyer (1972).
REFERENCES
Albrectsen S. T. and Thaysen J. H. 1955. The excretion of parotid gland. Stand. f. &I. Lab. Invesr. 7, 231-238. Burgen A. S. V. 1967. Secretory processes in salivary glands. In: Handbook of Physiology, Alimentary Canal, Volume II. Secretion. (Edited by Code C. F.) pp. 561-579. Waverly Press, Baltimore. Corkhill A. B. 1925. The estimation of salivary urea as an index to renal prognosis, Med. J. Aust. 10, 236238. Dahlberg W. H., Sreebny L. M. and King B. 1967. Studies of parotid saliva and blood in hemodialysis patients, J. ~cppl. Physiol. 23, IO&108 Ferris H. C. 1920. Mouth hygiene, controlled by diet through salivary analysis. Dent. Cosmos 62, 453464. Ferris H. C., Smith E. D. and Graves E. V. 1923. Physiological and pathological variations in the composition of human saliva. J. Am. Dent. Ass. 10. 19-52. Forland M., Shannon I. L. and Katz F. H. 1964. Parotid fluid urea nitrogen for monitoring hemodialysis. New Enyl. .I. Med. 271, 27-38. Hench P. S. and Aldrich M. 1922. A salivary index to renal function. J. Am. Med. Ass. 81, 1977-2003. Morris J. L. and Way C. T. 1924. Further observations on chemical constituents of saliva. J. biol. Chem. 59, xxvi-xxvii. Nikiforuk G., Jackson S. J., Cox M. A. and Grainger R. M. 1956. Some blood and salivary nonprotein nitrogen constituents in children, and dental caries. J. Pediat. 49, 425431. Ormsby A. S. 1942. A direct calorimetric method for the determination of urea in blood and urine. J. biol. Chem. 246, 5955604. Rosenthal H. L. 1955. Determination of urea in blood and urine with diacetyl monoxime. Amdyr. Chem. 27, 198@1982. Schmitz H. S. 1922. Comparative concentration of urea in blood and saliva in a series of pathologic cases. J. Lab. c/in. Med. 8, 78 82. Schneyer L. H., Young J. A. and Schneyer C. A. 1972. Salivary secretion of electrolytes. Physiol. Rev. 52. 720-777. Shannon I. L. and Chauncey H. H. 1967. A parotid fluid collection device with improved stability characteristics. J. Oral Ther. Pharm. 4. 93-97. Shannon I. L., Suddick R. P. and Chauncey H. H. 1969. Etrect of atropine-induced flow rate depression on the composition of unstimulated parotid fluids, Archs orul Biol. 14. 761-770. Shannon I. L. and Prigmore J. R. 1960. The use of the Auto Analyzer in studies of the physiological chemistry of the parotid gland. Ann. N.Y. Acad. Sci. 87. 745-763. Shannon I. L., Suddick R. P. and Dowd F. J. 1974. Saliva: composition and secretion. In: Monographs in Oral Science, (Edited by Myers H. M.), Kargkr, Basef. Stealv C. L.. 1928. The clinical value of the salivarv urea index. J. iab. clin. Med. 14, 162-165. Suddick R. P., 1973. Does transepithelial hydrostatic pressure provide energy for fluid secretion? In: Mechanisms of’E.yocrine Secretion. (Edited by Han S. S., Sreebny L. and Suddick R. P.) pp. 4460. University of Michigan Press. Wright S. 1841-1842. Case of ascites in which, during spontaneous ptyalism that occurred after tapping, urea was detected in saliva. Lancet 1, 753-758.
