Annals of Clinical Biochemistry, 1977, 14, 275-278

Automated fluoride ion determination Determination of serum fluoride ion levels DAVID C. COWELL From the Department of Chemical Pathology, Liverpool Area Health Authority (Teaching), Ashton Street, Liverpool L3 5RT

A method is described, using standard continuous flow techniques, for the automated determination of serum fluoride ion concentration using a fluoride ion selective electrode. It is shown that the kinetics of the electrode response to changes in fluoride ion can be used for the accurate measurement of fluoride ion concentration in serum, and that equilibration of the electrode response is not a prerequisite for the measurement of fluoride ion. Recovery experiments are in the range 86,6-98,5 %, in-batch precision is 1,0-6,1 %, between-batch precision 5·5 %, with carryover of

SUMMARY

-2·99~-;;.

All these procedures were manual, requiring Sodium fluoride has been used in the treatment of osteoporosis and osteitis deformans (Paget's disease operator intervention for various periods of time. of the bone) for a number of years. The levels of Using the kinetics response of the fluoride electrode fluoride ion in both the serum and urine of these to changes in fluoride ion concentration, an autotreated patients has recently been shown to reflect the mated method for urinary fluoride ion concentration daily oral intake, with the consequent possible use of was achieved (Cowell, 1977) and this procedure can monitoring therapy (Cowell, 1975). be modified satisfactorily to determine serum levels. The sample throughput of published methods for the determination of serum fluoride ion levels has Materials and method varied greatly with electrode equilibration times ranging from a few minutes to a number of hours. APPARATUS Barnes and Runcie (1968) gave no indication of The apparatus used is the same as that described for electrode equilibration times. Fry and Taves (1970) the automated urine method (Cowell, 1977) except took potential measurements every minute until a that the curve regenerator is not required. constant reading was attained; this usually took between two and five minutes giving a throughput of REAGENTS about 12 samples an hour. Venkateswarlu et al. Stock fluoride standard. 5 mmol F)I: 210 mg of (1971) adsorbed the fluoride ion on to calcium sodium fluoride (Analar BDH), dried by desiccation, phosphate and then quantitated the adsorbed dissolved in deionised water, and the solution diluted fluoride, using an electrode equilibration time of to one litre. The standard is stored in a polyethylene two hours. Hall et al. (1972) used an equilibration bottle. Total ionic strength adjustment buffer (TISAB), time of 40 minutes, but as a standard addition method was used, a total analysis time of 80 minutes prepared as described by Cowell (1975). Working TISAB: stock fluoride ion standard is was required. In 1973, Singer and Armstrong (1973) used an electrode equilibration time of between 30 added to the TISAB to give a final concentration of and 60 minutes, while Jardillier and Desmet (1973) 2'5 flomol F-f1. Rennin (Koch-Light Laboratories) did not specify an electrode equilibration time. Using 1 g and 0·5 ml of a 35 % Brij solution are also added an electrode bank of six electrodes, Fuchs et al. (1975) per litre of TISAB. Working fluoride standards: increasing amounts of achieved a sample throughput of only 18 samples a day.Venkateswarlu (1975) using a Su.lsample required the stock fluoride ion standard are added to pooled a two-hour electrode equilibration time, while Cowell human serum to give levels approximating to 1, 4, 8, (1975) found a 15-minute equilibration time to be 13, and 18 flomol F- /1. These are then standardised acceptable. against aqueous fluoride ion standards using a manual 275

Downloaded from acb.sagepub.com by guest on June 4, 2016

276

David C. Cowell 17·25

,~ I

I

I' III! I

'

, "

,I

13·0

I' ! I

I

~I, "I' , , Ii" I

I, II !

13·0

I

II

1,-'_,.......--'

Fig. 1 Analytical run showing carryover experiments. (Chart speed 2 minutes/em). Trough to peak height, 'x', is measured (millilitres) for each peak.

electrode technique (Cowell, 1975). The pooled human serum is first filtered through bacteriological filters to sterilise it. When standardised, the working standards are aliquoted and stored frozen at - 20 e in plastic tubes. Stability of serum fluoride at - 20 e has been described (Cowell, 1975; Fuchs et at., 1975).

5-4

_________________ ---------------+2SD

D

),2

D

i-o

PROCEDURE

The flow diagram is identical with that used for the automated urine method with the exception of the curve regenerator. The sample and wash times are 30 seconds and 210 seconds respectively, giving a wash ratio of 7: 1 at a rate of 15 samples an hour. Drift standards are run every 10 samples. The standards and any samples after a water wash are preceded by an aliquot of serum in order to resaturate the system. The distance' x ' in millimetres (Fig. 1) is measured for each peak. A calibration curve is constructed by concentration versus' x ., and the unknowns are read from this curve.

';'=-

u,

~ '-8

-=-

.

I-H---f---->,E---Jbf'----t---t--t-----,-r-''W M eon

"0

-~

4-b

:>

:;::

E

----- -I SD

~ 4-4 en 4·2

---

---------0------ ------------

-2 SD

4·0 L-~..-,.....,.......,.......,.......,...._r_r___r___.,._.,-.._~ o 2 4 6 8 10 12 14 16 18 2022 24 26 28 30

Results

Days

Within-batch precision was determined by replicate assays at three serum fluoride ion concentrations (Table 1). Between-batch precision was determined during a three-month period using aliquots of calf serum stored at - 20 a coefficient of variation of 5'5% was obtained (Fig. 2).

Fig. 2 Between-hatch variation over a three-month period (30 analytical runs). Coefficient oj variation 5-5 %.

DC;

Table 2 Accuracy analysis-srecovery experiments umol F-jl

Table 1 Precision analysis-within-hatch

Recovery (%)

No. of determinations (n)

umol F"]l Mean (x)

Standard deviation (S D)

(CV)~-;;

21 21 21

0-78 6-34 13-85

0-048 0-064 0-152

6-1 1-0

Coefficient of variation

I-I

Added

Observed

Recovered

0 2-5 5-0 7'5 10-0 12-5 15-0

0-75 3-15 5-30 7-25 10-60 13-25 15-20

2-40 4-55 6-5 9-85 12'5 14-45

Downloaded from acb.sagepub.com by guest on June 4, 2016

96-0 91-0 86-6 98-5 100-0 96-3

277

Automated fluoride ion determination

The constants 'a' and 'b' described by Walker et al. (1970) have been calculated for the total system in the same manner as described by Cowell (1977). The

Table 3 Sample to sample carryover Carryover (%)

Value of hiRh concentration

Value of successive low concentration

17-25 IHXl 8·00 4-00

0·70 0-70 0-70 0'70

constants are 2·5 seconds and 8·0 seconds respectively for the rise curve and 2·5 seconds and 10'5 seconds for the fall curve when quantitated using a serum with a fluoride ion level of 1·\ jJomol F-/1.

-0'63 -0-84 -1-38 -2-99

Discussion Accuracy was determined by adding known amounts of fluoride ion to a specimen of human serum. A mean recovery of 94·7% was obtained (Table 2). Sample to sample carryover was assessed by the method of Broughton et al. (1974) and was found to have a maximum of - 2·99 % (Table 3 and Fig. 1). Comparison with the manual technique of Cowell, using an equilibration time of ] 5 minutes, is shown in Fig. 3. Statistical analysis of these data is given in Table 4. 12 n

= 54

y= 0-9475.+0-0726

ro .0,'1675

•

•

10

• _ 6

r~

o ~

c

l.

4

01lL-_~-~--~-~--..--~

o

2

4

Automated ,umol

nl

6

10

12

Fig. 3 Comparison of automated and manual techniques for the determination offluoride ion concentration in serum.

Table 4

Comparison with the manual technique of Cowell (/975). Statistical analysis of the data

No. of pairs

54

Regression coefficient b Regression coefficient a Standard error of the estimate (Sy) Student's. t test for b = 0 (for 52 degrees of freedom)

i 0-9475 ! 0-0728 0-766 !"mo! F- /I 26'7407 (p < 0-(01)

Correlation coefficient r

!0'9655 26-741

Student's t test for r = 0 (for 52 degree, of freedom)

(p < 0-(01)

Student's I test on the paired data (for 53 degrees of freedom) Standard deviation of the duplicates (SD II )

1-3037 (p < 0-2) 0-775 urnol F-/I

As fluoride ion levels in serum are approximately 10 times lower than the levels found in urine, an increase in sensitivity is necessary to enable measurement of serum fluoride ion levels by an automated technique without loss of precision and speed of analysis. By decreasing the background fluoride ion concentration in the working TISAB an increase in sensitivity was achieved. A further increase in sensitivity was achieved by lengthening the sample time to 30 seconds, but this requires a correspondingly longer wash time to give an adequate fall curve to the peak, and consequently reduces the throughput to ]5 samples an hour. The use of serum standards is necessary because of the effect of a protein solution on the kinetics of the fluoride electrode response. The exponential slope of the rise curve is steeper for a protein solution than an aqueous solution, which consequently gives greater peak heights for the same fluoride ion concentration. The constant 'b' for the rise curve, calculated on an aqueous sample of 1·5 jJomol F-/I, was found to be 12·5 seconds, which is 4· 5 seconds slower than on the serum sample of a similar fluoride ion concentration. Only the constant 'b' appears to be affected by the increase in sample time, lower background fluoride ion concentration, and lack of regeneration in comparison with the automated urine method of Cowell (1977). The inclusion of a serum sample before the standards and after a water wash was found necessary because of a small but significant rise in the baseline after a protein solution had been assayed. Baseline drift, arising from the build-up of protein on the electrode, has been described with the calcium ion selective electrode and was eliminated by the addition of the proteolytic enzyme, trypsin, to the buffer (Hartner et al., 1970). To avoid this difficulty with the fluoride electrode a proteolytic enzyme with activity at pH 5'2, the pH of the TlSAB was required. Rennin will coagulate milk at pH 6·8 although its pH optimum for proteolytic activity is pH 3·7. With serum as substrate it exhibits approximately 20 % of its optimal activity at pH 5·2 (Taylor, 1956). At the level of enzyme included in the TISAB, rennin appears to be effective in removing any protein buiId-

Downloaded from acb.sagepub.com by guest on June 4, 2016

278

David C. Cowell

up on the electrode surface and eliminating baseline drift. Measurement of the distance' x', trough to peak height, was necessary to obtain acceptable carryover when going from a high fluoride ion concentration to a low fluoride ion concentration (Fig. 1). If peak height only is measured the carryover and reproducibility are poor. This is because of the slower response of the electrode on the fall curve of the peak. By increasing the wash cycle the carryover can be reduced, but the throughput rate of samples would be reduced further. The within-batch and betweenbatch data presented indicate that this type of peak height measurement is reliable. Statistical analysis of the regression line data (Table 4) shows that the standard error of the estimate (Sy), calculated from the regression equation, and the standard deviation of the duplicates (SDd), calculated from the Student's t test, gave similar values. This similarity, in conjunction with the regression coefficient 'b' confirms that the proportional and random error between the manual and automated methods is small. The Student's t test on the paired regression line data indicates that there is no significant difference between the two sets of data. References Barnes, F. W., and Runcie, J. (1968). Potentiometric method for the determination of inorganic fluoride in biological material. Journal of Clinical Pathology, 21, 668-670. Broughton, P. M. G., Gowenlock, A. H., McCormack, J. J., and Neill, D. W. (1974). A revised scheme for the evaluation of automatic instruments for use in clinical chemistry. Annals of Clinical Biochemistry, 11, 207-218. Cowell, D. C. (1975). The determination of fluoride ion concentration in biological fluids and in the serum and urine of fluoride-treated patients with Paget's disease and osteoporosis. Medical Laboratory Technology, 32, 73-89.

Cowell, D. C. (1977). Automated fluoride ion determination. Determination of urine fluoride ion levels. Annals of Clinical Biochemistry, 14, 269-274. Fry, B. W., and Taves, D. R. (1970). Serum fluoride ion analysis with the fluoride electrode. Journal of Laboratory and Clinical Medicine, 75, 1020-1025. Fuchs, C., Dorn, D., Fuchs, C. A., Henning, H. Y., McIntosh, C., Scheler, F., and Stennert, M. (1975). Fluoride determination in plasma by ion-selective electrodes: A simplified method for the clinical laboratory. Clinica chimica acta, 60, 157-167. Hall, L. L., Smith, F. A., De Lopez, O. H., and Gardner, D. E. (1972). Direct potentiometric determination of total ionic fluoride in biological fluids. Clinical Chemistry; 18, 1455-1458. Hattner, R. S., Johnson, J. W., Bernstein, D. S., Wachman, A., and Brackman, J. (1970). Electrochemical determination of apparent ionized serum calcium using a calciumselective electrode: The method and values in normal humans and a comparison to total serum calcium. Clinica chimica acta, 28, 67-75. Jardillier, J. c., and Desmet, G. (1973). A study of serum fluoride ion levels using a specific electrode technique. Clinica chimica acta, 47, 357-363. Singer, L., and Armstrong, W. D. (1973). Determination of fluoride in ultrafiltrates of serum. Biochemical Medicine, 8,415-422. Taylor, W. H. (1956). Aspects of the proteolytic activity of the stomach and gastric juices in health and disease. DM Thesis (Oxford). Yenkateswarlu, P. (1975). A micro method for direct determination of ionic fluoride in body fluids with the hanging drop fluoride electrode. Clinica chimica acta, 59, 277-282. Yenkateswarlu, P., Singer, L., and Armstrong, W. D. (1971). Determination of ionic (plus ionizable) fluoride in biological fluids: Procedure based on adsorption of fluoride ion on calcium phosphate. Analytical Biochemistry, 42, 350-359. Walker, W. H. C., Pennock, C. A., and McGowan, G. K. (1970). Practical considerations in kinetics of continuous flow analysis. Clinica chimica acta, 27, 421-435.

Accepted for publication 17 June 1977.

Downloaded from acb.sagepub.com by guest on June 4, 2016