- Email: [email protected]
Studies on the Chick Assay for Vitamin D
Studies on the Chick Assay for Vitamin D II. A Comparison of Four Criteria of Calcification* J. A. CAMPBELL, B. B. MIGICOVSKY, 1 AND A. R. G. EMSLIE
Vitamin Assay Laboratory, Experimental Farm, Ottawa (Received for publication June 12, 1944)
T T WAS shown in a previous communi•*• cation, Campbell et al. (1944), that as a criterion of calcification iff chick assays of vitamin D the ash percent of the middle toe was no more variable than the ash percent of the tibia and its determination entailed much less labor. In that investigation the results of seven assays were presented. While the data were adequate to show the similarity in precision, it was felt desirable to obtain further comparisons of the estimate of potency by the two methods. At the same time a study has been made of further modifications of the method with a view to effecting still more economies in manipulation and labor. These modifications included elimination of extraction of the toes and limiting the feeding period to two weeks. The comparisons of estimates of potency were made on the basis of group ashing, with the exception of a few assays which were used to compare the precision of the tibia ash and radiographic technics. The Olsson radiographic technic (1936, 1939) is used by the British Standards Institution (1940) as an alternative to the tibia ash method. Apart from the initial cost of equipment, this method offers many advantages over the tibia method and permits of strictly comparable assays being made from the same chicks at two and three weeks of feeding, together with calculations of the precision. As the * Scientific Contribution No. I l l from the Division of Chemistry, Science Service. 1 Now on active service,
A.O.A.C. (1940) system of feeding has been used throughout, the x-ray data reported herein do not necessarily apply to the original procedure. Briefly, this paper presents the results of a series of assays calculated from the following criteria of calcification: ash percentage of tibiae, extracted toes and unextracted or "fresh" toes, and the "tarsometatarsal" distance as developed by Olsson. A comparison of the two and three week feeding periods is also given. EXPERIMENTAL
For control purposes it was necessary that the potency of the oils reported in this study be determined by the A.O.A.C. (1940) procedure using tibia ash as the criterion of calcification. At the same time, the center toes of both feet were removed as described previously (Campbell et al., 1944) and composited into 2 separate series of groups. One series was wrapped in filter papers for extraction, drying and ashing by the A.O.A.Cmethod, while the other series was placed in weighing bottles immediately after dissection. When the "fresh" weight had been obtained, the toes were transferred to crucibles and ashed without further treatment. The ash percentage was thus calculated on the fresh basis. Fifteen to 20 chicks were started in each group. If ashing could not be carried out on the toes immediately after weighing, they were all put in an oven to dry at 100°C. In most of the tests the Canadian Standard Reference Oil No. 1 was fed to 3 groups of
/
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
chicks at each of 3 levels while the unknown oil was fed at only 2 levels. The levels were spaced at equal logarithmic intervals to permit coding in the calculations. This investigation involved a series of 28 assays carried out over a period of approximately one year and required the v feeding of about 8,000 chicks. The vitamin D supplements assayed were of many types, representative of those most commonly used in Canadian poultry feeding, and varied in potency from 100 to 4,000 A.O.A.C. units per gram. The method of Knudsen and Tolle (1942) was used for calculating potency from composite data. For individual chick data the B.S.I. (1940) method of analysis was used to calculate the estimate of potency and its fiducial limits. This was supplemented by the X test of Bliss and Cat. tell (1943) to study the precision of the assays. In the B.S.I, method where Q 2 /P {i-e. regression function) exceeds 60 times s2 {i.e. variance for error) the calculation of potency is similar to the Knudsen-Tolle method, and the "approximate" formula is used for the calculation of fiducial limits. However, where this criterion is not satisfied, i.e., Q 2 /P<60 s2, a correction is necessary for both the estimate of potency and fiducial limits. Under these circumstances, as Irwin (1943) has pointed out, the approximate formula not only underestimates the fiducial range but also biases its position. In the comparison of the four methods in this paper, as most of the assays for tibia and toes were calculated from composite data where no correction was possible, it was felt that the use of the approximate formula for the radiographic data was justified even though the statistical criterion was not always satisfied. The significance of this problem will be discussed in a subsequent paper.
73
RADIOGRAPHIC TECHNIC
X-ray photographs of the right tibiotarso-metatarsal joint were taken of all chicks on the day before killing at 3 weeks of age. The radiographs were made on Dupont 5"X7", no screen, fine grain film with a Victor portable model x-ray apparatus. The cone of the instrument was fitted with a lead sleeve which reached almost to the film and protected the operator from excessive radiation. An exposure of one second at 12 inches, 15 m.a. and K..V. position 2 was found to be most satisfactory. The radiographs were taken through a one inch square hole cut in a lead mask under which the film holder was manipulated by means of a special frame. With this equipment it was possible to x-ray up to 300 chicks an hour. The films were examined under a microscope with a 12 X magnification (lower magnification would give equally good results) using a calibrated ocular scale with an extended zero line. This zero mark was lined up across the proximal ends of the two metatarsal bones and the distance of the tarsalbone read on the scale to the nearest 0.06 mm. Logarithms of the t.m.t. distance were used for all calculations and to avoid the difficulty of negative entries logarithms were taken of the values multiplied by 10. With practice it was found possible to read 800 radiographs a day with little effort. The suggestion of marking the end of the metatarsal bones with a scalpel under naked eye inspection was not found practicable as it could often be seen under the microscope that the line was not in the correct place. It was found in preliminary tests that the t.m.t. measurements made by this method were quite reproducible. The results of some of these tests are reported in Table 1. I t is evident from this table that, considering data from inexperienced read-
74
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
ers, there is a maximum difference of about 10 percent between the results of different readers. There is a difference of 6 percent between readings of the same person before and after becoming experienced. These differences would be considerably reduced with experience. All radiographic data reported in the body of this paper were read by the same person. RESULTS The estimates of potency of the series of 28 assays calculated from the four criteria
not exceed its standard error and therefore cannot be considered significantly different from zero. From the standard deviations of the differences, it may be expected that, for two in every three assays, the estimates of potency by any of the proposed methods will fall within approximately 20 percent of the estimate found with tibia or that in 95 percent of the assays the estimates will be within 40 percent of that by tibia ash. From assays in which the error has been calculated from individual data
TABLE 1.—Reproducibility of results by the radiographic technic (Assay 104) Inexperienced Assay Period (weeks)
2 2 3
Group
1-20 21-40 1-20
Observer R Potency
Limits*
117 88 US
Percent 79-127 78-128 83-120
Experienced Observer C
Potency
Limits*
106
Percent 80-125
105
82-122
Observer R Potency
Limits*
113 94 112
Percent 80-125 75-133 83-121
* Approximate limits at P =0.05.
of calcification are presented in Table 2. The estimates for individual assays are quite variable. From a control laboratory standpoint, however, in every case where the tibia method shows an oil to have reached its expected potency the other methods give the same interpretation within the usual limits of error. The differences between the estimate of potency determined by the tibia ash method and by each of the other three methods were calculated as percentages of the tibia ash values and are also shown in Table 2. From the means of these values and the distribution of the differences, it is evident that on the average there is no significant difference between the estimates by the four methods. In the case of the radiographic data the mean difference, although tending to be rather large, does
it is known that the estimates with tibia reported here are reproducible to approximately + 20 percent in 95 percent of the cases. As the other methods have similar precision, this accounts for the other 20 percent variation. When the estimates differ considerably, it is difficult to determine which method more nearly approximates the true potency. As the mean differences approximate zero, there is obviously on the whole no difference between the methods, the chief variation being in the estimates of the individual oils. If a mean of all four assays can be considered the most probable estimate of the true potency of an oil, then the most accurate method is the one which will most closely approximate that value. In order to compare the methods in this manner, means of the estimates of
75
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
potency by all methods for each oil and the standard deviations of the differences of the four methods from these means "were calculated. The standard deviations were 15.0, 11.4, 10.6, and 10.0 percent for tibia, extracted toes, fresh toes and t.m.t., respectively, while the mean differences were + 2 . 4 , + 0 . 2 , + 0 . 7 and - 3 . 3 . There seems no doubt from these figures that the toe ash and x-ray methods yield results
42„
15 TIBIAE (%ASH)
40L
5
b
c=l271
15
259
combined reference and unknown data and the combined decoded slopes are entered in the charts. It is evident that the response curves of all four methods are quite similar and although in this particular assay some of the points may not be strictly linear, yet relatively the methods show good agreement. Part of the series of the assays, namely, assays 87 to 95, which involved 1693, REFERENCE- ° UNKNOWN * 0-90,. RADIOGRAPHIC l + L0G.(TM.T) «
04.9 EXTRACTED TOES - (%ASH) bc = 6-83.
15
25~~9"
fe
259
i5
25
DOSE IN UNITS PER 100 GM. FEED (LOGARITHMIC SCALE)
FIG. 1. Response of chicks as measured by four criteria of calcification.
equally as reliable as those by the tibia ash method. The estimates of potency by the tibia ash method, if anything, are more variable than the estimates by the other three methods. In order to illustrate the ranges of response of the four criteria of calcification arid the similarity of their log. (dose)-response curves,.data from assay 104 were plotted on semi-log. paper and reproduced in Fig. 1. Approximately 45 chicks (3 groups of 15 each) were used to determine each of the points on the graph. To facilitate comparison the scale for the radiographic method has been inverted. The regression lines were calculated from the
chicks, was carried out using individual tibia ash, thus furnishing the basis for a comparison of the precision of the tibia ash, and radiographic methods. As in the earlier paper, the data used for this comparison were strictly comparable, each value for tibia ash having a corresponding value for the radiographic method. In Table 3 the fiducial limits in percent, the standard deviations (s) and the decoded slopes (b) of the log. (dose)-response curves have been tabulated and from these the X = (s/b) ratio for each assay has been calculated. I t can be seen that, although X for the radiographic method is greater in every case than that for the
76
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
tibia ash method, yet only in the case of assay 94 does the difference approach significance at the 5 percent point. By combining the 9 comparisons by the x2 method as outlined by Fisher (1934) an overall
This slight but constant difference is also reflected in the fiducial limits. I t is pointed out again that the radiographic technic of Olsson as used in this paper involved only the actual principle of x-raying the chicks
TABLE 2.—A comparison of four criteria of calcification Potency (A.O.A.C. units) as calculated from Assay
No. 84 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 111 112 114 115
Expected potency
Difference as percent of tibia
VtUUCb
Toe ash Tibia
ash
Extracted
Fresh
Log. t.m.t.
1200 2000
1835 3432
1957 4056
2066 4280
2689 3772
200 200 200 400 400 400 400 400 400 400 400
258 294 380 498 422 434 475 519 378 248 324
281 379 424 551 481 507 478 — — 265 244
— — — 540 566 536 — — 248 232
267 289 383 560 408 456 561 516 372 242 252
4000
3664
2962
2727
3228
200 200 200 200 200 100 400 200 400 200
377 233 214 238 401 102 538 447 864 321
377 211 210 226 361 108 536 225 446 246
324 219 189 211 318 110 582 342 587 308
360 199 202 244 386 112 421 269 522 223
2000 1000
3122 1380
2828 1758
2874 1443
3090 1157
400 200
373 224
570 280
566 272
444 247
•
—
.
Toe ash Extracted
Fresh
+7
+13"
+18
+25
+9
—.
+29 +12 +10 +13
— —
+8
Log. t.m.t.
+47 +10 + 3 - 2
+ 1 +12 - 3
+17 +1
+34 +24 —
— '
—
+7
0 -28 -26 -14
- 9 - 2 - 5
- 6
-15
-12 -11 -21
+3
-25 -19 0
-10
+6
0 -50 -48 -23 - 9 +27 +53 +25
.—.
+8 +8 -23 -32
- 4 - 8
+5 +18
- 1 - 2 - 2
-22 -12 - 5 - 6
-4 +10 -22 -40 -40 -31 - 1
+5
+52 +21
-16 +19 +10
+ 1.3
+ 0.5
- 3.1
S.D.
22.6
21.7
18.2
S.E.
4.4
4.6
3.4
Means
measure of the difference was obtained. X2 was found to be 26.036 which with n = 18 yielded a P value of approximately 0.09. Although approaching the 5 percent point, the difference cannot be considered significant from these data. This conclusion substantiates the work of Baker and Wright (1940) although in their data the bone ash tends to be the more variable.
and was not used as a complete method. As it was possible to x-ray the same chicks at both two and three weeks, the radiographic procedure offered the most valid basis for a comparison of the two feeding periods. Accordingly the potency, approximate fiducial limits and X ratios were calculated from a series of 8 assays for both feeding periods. The results of
STUDIES ON THE CHICK ASSAY FOR VITAMIN D.
these comparisons are presented in Table 4. I t should be pointed out that for this particular comparison of X's the data were not strictly paired, that is, if a chick died after being x-rayed at 2 weeks, its 2 week value was not removed. However, as mortality was low in this period (less than 3 per 100), it may be safely assumed that the data are comparable.
77
ference between the estimates of potency determined at 2 and at 3 weeks. Considering the precision of the method, it is evident from the X ratios that in every case the 2 week assay period yields more variable results than does the 3 week period. The differences, although large, are not significant when each assay is viewed by itself or when combined, as the standard
TABLE 3.—Comparison of the precision of the tibia ash and radiographic procedures Radiographic method (log t.m.t.)
Tibia ash method Assay No. 87 88 89 90 91 92 93 94 95 Means
Fiducial* Standard Decoded Fiducial* Standard limits deviation slope X = (s/b)±S.E. limits deviation Percent Percent s s b
Decoded slope b
X = (s/b)±S.E.
Difference tm.t. X tibia X
8S-118 85-118 84-119 87-115 85-118 86-117 86-116 88-114 87-115
3.0382 2.9260 2.8962 3.5029 3.6033 3.6302 4.0631 3.7767 3.9353
9.8075 10.1166 10.6268 13.8683 13.2644 14.1148 13.7714 15.5736 16.2649
0.310+0.031 0.289±0.028 0.273±0.025 0.253±0.023 0.272±0.028 0.257±0.025 0.295±0.029 0.243±0.020 0.242±0.021
83-121 82-123 82-121 85-118 84-119 85-118 82-122 84-119 85-118
0.1854 0.1774 0.1775 0.1982 0.2066 0.2093 0.2479 0.2339 0.2200
-0.5223 -0.5034 -0.5584 -0.6963 -0.7156 -0.7643 -0.6719 -0.7161 -0.7089
0.355±0.041 0.352±0.041 0.318±0.033 0.285±0.028 0.289±0.031 0.274±0.028 0.369±0.044 0.327±0.035 0.310±0.034
+0.045 +0.063 +0.045 +0.032 +0.017 +0.017 +0.074 +0.084 +0.068
86-117
3.4858
13.0454
0.270± 0.023
84-120
0.2062
-0.6508
0.320±0.033
+0.050
Number of chicks
606 509 578
' Approximate fiducial limits at P =0.05. According to the criterion of the B.S.L method, approximate limits are valid for the assays reported in this table.
From Table 4 it can be seen that in all but 2 of the assays the 2 week estimate is higher than that at 3 weeks. The mean difference between estimates of potency at 2 and 3 weeks is 11 percent. The combined probability of the significance of differences between estimates, using the X2 method of Fisher (1934) was found to be greater than 0.80. I t may therefore be concluded that there is no significant dif-
errors of the 2 week data are also relatively large. However, when the probabilities were recalculated, using as standard error of the difference between X's an average value of 0.06 obtained from 3 week data, it was found that all but 2 values were significant at the 1 percent point. The 2 week data may therefore be considered significantly less precise than the 3 week data when judged by the 3
TABLE 4.—Comparison of two and three week data Criterion Assay of Po- Limits* Calcifi- No. tency Percent cation Radiographs
Means
104 105 .106 107 108 111 112 114
113 396 275 584 249 4432 1533 402
Two Week Data
Three Week Data
Standard Deviation s
Slope b
X = (s/b)
80-125 72-139 64-157 66-152 69-146 76-132 59-171 63-158
0.1819 0.1586 0.1653 0.1613 0.1624 0.1463 0.1385 0.1367
0.4561 0.3028 0.2726 0.3015 0.3042 0.2402 0.2285 0.1825
0.399±0.057 0.524± 0.110 0.606±0.146 0.535±0.114 0.534±0.114 0.610±0.102 0.606±0.168 0.749±0.222
69-148
0.1564
0.2861
0.570
' ApproximatelimitsatP=0.05.
Standard Po- Limits* Deviatency Percent tion s 112 421 269 522 223 3090 1124 444
83-121 79-127 73-138 76-132 74-135 83-121 80-126 70-144
0.1905 0.1981 0.2025 0.1920 0.2050 0.1679 0.1492 0.1959
ence between X's
Slope b
X = (s/b)
0.5507 0.5390 0.4669 0.5020 0.4678 0.4375 0.4777 0.3538
0.346±0.044 0.368±0.056 0.434± 0.077 0.382±0.060 0.438±0.079 0.384±0.042 0.312±0.047 0.554±0.125
0.053 0.156 0.172 0.153 0.096 0.226 0.294 0.195
0.402
0.168
77-131 0.1876 0.4744
78
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
week standard of precision. The chief reason for this seems to be that the lower slope (b) of the log. (dose)-response curves at 2 weeks is not sufficiently balanced by a lesser variability (s). Substantially similar observations were made by De Witt et al. (1942). Although for some of the assays reported in Table 4 the approximate fiducial limits, particularly at 2 weeks, underestimate the true width of the fiducial range, yet they show the same
ferent toes on the extracted basis was also investigated at this time and the results of 5 assays are presented in Table 5. In some cases the 3 toes vary somewhat in the interpretation of potency in the various assays. The ash percentages indicate that there are constant differences between the ash content of the 3 toes. The data as a whole indicate that approximately the same variation may be expected between different toes as between
TABLE 5.—Comparison of the ash content and potency calculated from extracted ash data of different toe Toe ash (Percent) Assay No. 93 111 122A* 112B* 114
Inside toes
Middle toes
Outside toes
No. of groups averaged
12.58 13.90 14.06 13.57 14.07
13.39 14.56 14.72 14.40 14.78
13.74 15.04 15.01 14.97 15.41
20 38 15 16 16
Estimation of potency (A.O.A.C. units) Inside toes
Middle toes
Outside toes
483 2868 1390 1191 476
478 2828 1972 1544 570
481 3888 1888 1719 508
* A = A.O.A.C. method. B = B.S.I, method.
trend as the X ratios. Whatever limits are calculated, of course, do not affect the comparison by the X method. In order to furnish some idea as to how the other 3 criteria would react, one of the assays (No. 104) was carried out entirely in duplicate under the same conditions and at the same time. One set of chicks was killed at two weeks and the other was killed at 3 weeks. The estimates of potency were as follows: Radiographic Extracted toe Fresh toe Tibia
At 2 weeks At 3 weeks 112 94 108 87 110 104 98 102
The results of these comparisons, when considered in conjunction with those of Table 4, substantiate the conclusion that there is no significant difference between estimates of potency determined at 2 and at 3 weeks. The reproducibility of results using dif-
the other methods, and that, if toes are used for assay, the same toe must be used throughout any one assay. The use of 2 toes in an assay readily furnishes duplicate results, an average of which yields a considerably more reliable estimate of potency than either one alone. DISCUSSION
I t should be pointed out that, while all available data were included in this series of assays, not all of them would satisfy the criterion of the B.S.I, method for a valid assay. However, as on the average four to five times more groups per assay were employed than is ordinarily considered necessary, the results outlined above may be considered representative of what is to be expected from the four critieria of calcification under average conditions. If the data be viewed from this assumption, they suggest that although the methods are valid for sustain-
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
ing or rejecting a guarantee there is still much room for improvement in the technic as a whole. Certain observations may be made from the comparisons of the four methods presented in this paper, from the point of view both of the results obtained and of the labor involved in the various methods.. All methods show considerable and approximately equal variability in their estimate of the potency of the oils. For an estimate of the precision of an assay involving individual data, the radiographic technic has unquestioned advantages in manipulation and practically equal precision to other methods. If x-ray equipment is not available, the use of toe ash on the extracted basis offers considerable simplification and equal precision to the official tibia ash procedure as shown previously. If a purely routine procedure is required, the use of toes on the fresh basis undoubtedly furnishes the most rapid procedure. The 2 week feeding period seems of particular value for use with fresh toes although it also is applicable to the other procedures for estimates of potency involving composite data. Its application, however, may be limited for use with individual data as the precision is relatively low. In view of the fact that all 4 methods have proved of approximately equal value in this laboratory for assessing calcification in the vitamin D chick assay and that the suggested modifications offer improvement in the official method, it is recommended that more extensive studies be initiated on a collaborative basis to test more thoroughly these criteria of calcification with a view to their ultimate acceptance as alternative official procedures. SUMMARY
A series of 28 assays involving approxi-
79
mately 8,000 chicks has been carried out on various vitamin D feeding oils to compare four criteria of calcification, namely, ash percent of tibiae, fresh toes and extracted toes, and the Olsson radiographic technic. A comparison was also made of the two and three week feeding periods. The data show that the four criteria of response give on the average approximately the same estimate of potency. Different toes are equally useful but are not interchangeable in any one assay. The Olsson radiographic technic appears to be very slightly but not significantly more variable than tibia ash as a criterion of calcification. The two week feeding period is comparable to the three week period as far as an estimate of potency is concerned, but the precision is significantly lower. Considering the labor involved in these procedures, it is therefore concluded that the radiographic technic is unquestionably superior when individual data are desired. For routine assays involving composite data, toe ash on the fresh basis requires the least labor and seems equally reliable. ACKNOWLEDGMENT
The authors are indebted to the Production Service for the loan of the services of Miss V. Reynolds for reading the radiographs. The technical assistance of Mr. P. M. Richard is also acknowledged. REFERENCES
Assoc, of Official Agr. Chem., 1940. Methods of Analysis, 5th Edition, p. 371-372. Baker, A. Z., and M. D. Wright, 1940. Biological Assay of Vitamin D 3 .1. Assay methods at present in use, with particular reference to Olsson's radiographic technic. Analyst 65:326-335. Bliss, C. I., and McK. Cattell, 1943. Biological assay. Ann. Rev. Physiol. 5:479-539. British Standards Institution, 1940. British Standard Method for the Biological Assay of Vitamin D s by the Chick Method. Specification No. 911.
80
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
Campbell, J. A., B. B. Migicovsky and A. R. G. Emslie, 1944. Studies on the chick assay for vitamin D. I. Precision of tibia and toe ash as criteria of response. Poultry Sci. (in press). DeWitt, J. B., C. D. ToUe, H. W. Loy, Jr. and L. F. Knudsen, 1942. Observations on the chick method for the assay of vitamin D. III. Effect of decreasing length of assay period. J. Assoc. Off. Agr. Chem. 15:213-219. Fisher, R. A., 1934. Statistical methods for research workers. 5th Edition, Sec. 21.1. Oliver and Boyd, London. Irwin, J. O., 1943. On the calculation of the error of
biological assays. J. Hyg. 43:121-128. Knudsen, L. F., and C. D. ToUe, 1940. A statistical analysis of the A.O.A.C. collaborative study on assaying vitamin D by the chick method. J. Assoc. Off. Agr. Chem. 23:665-675. Olsson, N., 1936. Untersuchungen fiber die moglichkeit vitamin D—praparate mit wachsenden kiiken zu standardisieren. Archiv fiir GeflUgelkunde 10:423-436. Olsson, N., 1939. A method for quantitative estimation of vitamin D with growing chicks. Seventh World's Poultry Congress. Proc, p. 129-132.
Vitamin Assay Laboratory, Experimental Farm, Ottawa (Received for publication June 12, 1944)
T T WAS shown in a previous communi•*• cation, Campbell et al. (1944), that as a criterion of calcification iff chick assays of vitamin D the ash percent of the middle toe was no more variable than the ash percent of the tibia and its determination entailed much less labor. In that investigation the results of seven assays were presented. While the data were adequate to show the similarity in precision, it was felt desirable to obtain further comparisons of the estimate of potency by the two methods. At the same time a study has been made of further modifications of the method with a view to effecting still more economies in manipulation and labor. These modifications included elimination of extraction of the toes and limiting the feeding period to two weeks. The comparisons of estimates of potency were made on the basis of group ashing, with the exception of a few assays which were used to compare the precision of the tibia ash and radiographic technics. The Olsson radiographic technic (1936, 1939) is used by the British Standards Institution (1940) as an alternative to the tibia ash method. Apart from the initial cost of equipment, this method offers many advantages over the tibia method and permits of strictly comparable assays being made from the same chicks at two and three weeks of feeding, together with calculations of the precision. As the * Scientific Contribution No. I l l from the Division of Chemistry, Science Service. 1 Now on active service,
A.O.A.C. (1940) system of feeding has been used throughout, the x-ray data reported herein do not necessarily apply to the original procedure. Briefly, this paper presents the results of a series of assays calculated from the following criteria of calcification: ash percentage of tibiae, extracted toes and unextracted or "fresh" toes, and the "tarsometatarsal" distance as developed by Olsson. A comparison of the two and three week feeding periods is also given. EXPERIMENTAL
For control purposes it was necessary that the potency of the oils reported in this study be determined by the A.O.A.C. (1940) procedure using tibia ash as the criterion of calcification. At the same time, the center toes of both feet were removed as described previously (Campbell et al., 1944) and composited into 2 separate series of groups. One series was wrapped in filter papers for extraction, drying and ashing by the A.O.A.Cmethod, while the other series was placed in weighing bottles immediately after dissection. When the "fresh" weight had been obtained, the toes were transferred to crucibles and ashed without further treatment. The ash percentage was thus calculated on the fresh basis. Fifteen to 20 chicks were started in each group. If ashing could not be carried out on the toes immediately after weighing, they were all put in an oven to dry at 100°C. In most of the tests the Canadian Standard Reference Oil No. 1 was fed to 3 groups of
/
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
chicks at each of 3 levels while the unknown oil was fed at only 2 levels. The levels were spaced at equal logarithmic intervals to permit coding in the calculations. This investigation involved a series of 28 assays carried out over a period of approximately one year and required the v feeding of about 8,000 chicks. The vitamin D supplements assayed were of many types, representative of those most commonly used in Canadian poultry feeding, and varied in potency from 100 to 4,000 A.O.A.C. units per gram. The method of Knudsen and Tolle (1942) was used for calculating potency from composite data. For individual chick data the B.S.I. (1940) method of analysis was used to calculate the estimate of potency and its fiducial limits. This was supplemented by the X test of Bliss and Cat. tell (1943) to study the precision of the assays. In the B.S.I, method where Q 2 /P {i-e. regression function) exceeds 60 times s2 {i.e. variance for error) the calculation of potency is similar to the Knudsen-Tolle method, and the "approximate" formula is used for the calculation of fiducial limits. However, where this criterion is not satisfied, i.e., Q 2 /P<60 s2, a correction is necessary for both the estimate of potency and fiducial limits. Under these circumstances, as Irwin (1943) has pointed out, the approximate formula not only underestimates the fiducial range but also biases its position. In the comparison of the four methods in this paper, as most of the assays for tibia and toes were calculated from composite data where no correction was possible, it was felt that the use of the approximate formula for the radiographic data was justified even though the statistical criterion was not always satisfied. The significance of this problem will be discussed in a subsequent paper.
73
RADIOGRAPHIC TECHNIC
X-ray photographs of the right tibiotarso-metatarsal joint were taken of all chicks on the day before killing at 3 weeks of age. The radiographs were made on Dupont 5"X7", no screen, fine grain film with a Victor portable model x-ray apparatus. The cone of the instrument was fitted with a lead sleeve which reached almost to the film and protected the operator from excessive radiation. An exposure of one second at 12 inches, 15 m.a. and K..V. position 2 was found to be most satisfactory. The radiographs were taken through a one inch square hole cut in a lead mask under which the film holder was manipulated by means of a special frame. With this equipment it was possible to x-ray up to 300 chicks an hour. The films were examined under a microscope with a 12 X magnification (lower magnification would give equally good results) using a calibrated ocular scale with an extended zero line. This zero mark was lined up across the proximal ends of the two metatarsal bones and the distance of the tarsalbone read on the scale to the nearest 0.06 mm. Logarithms of the t.m.t. distance were used for all calculations and to avoid the difficulty of negative entries logarithms were taken of the values multiplied by 10. With practice it was found possible to read 800 radiographs a day with little effort. The suggestion of marking the end of the metatarsal bones with a scalpel under naked eye inspection was not found practicable as it could often be seen under the microscope that the line was not in the correct place. It was found in preliminary tests that the t.m.t. measurements made by this method were quite reproducible. The results of some of these tests are reported in Table 1. I t is evident from this table that, considering data from inexperienced read-
74
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
ers, there is a maximum difference of about 10 percent between the results of different readers. There is a difference of 6 percent between readings of the same person before and after becoming experienced. These differences would be considerably reduced with experience. All radiographic data reported in the body of this paper were read by the same person. RESULTS The estimates of potency of the series of 28 assays calculated from the four criteria
not exceed its standard error and therefore cannot be considered significantly different from zero. From the standard deviations of the differences, it may be expected that, for two in every three assays, the estimates of potency by any of the proposed methods will fall within approximately 20 percent of the estimate found with tibia or that in 95 percent of the assays the estimates will be within 40 percent of that by tibia ash. From assays in which the error has been calculated from individual data
TABLE 1.—Reproducibility of results by the radiographic technic (Assay 104) Inexperienced Assay Period (weeks)
2 2 3
Group
1-20 21-40 1-20
Observer R Potency
Limits*
117 88 US
Percent 79-127 78-128 83-120
Experienced Observer C
Potency
Limits*
106
Percent 80-125
105
82-122
Observer R Potency
Limits*
113 94 112
Percent 80-125 75-133 83-121
* Approximate limits at P =0.05.
of calcification are presented in Table 2. The estimates for individual assays are quite variable. From a control laboratory standpoint, however, in every case where the tibia method shows an oil to have reached its expected potency the other methods give the same interpretation within the usual limits of error. The differences between the estimate of potency determined by the tibia ash method and by each of the other three methods were calculated as percentages of the tibia ash values and are also shown in Table 2. From the means of these values and the distribution of the differences, it is evident that on the average there is no significant difference between the estimates by the four methods. In the case of the radiographic data the mean difference, although tending to be rather large, does
it is known that the estimates with tibia reported here are reproducible to approximately + 20 percent in 95 percent of the cases. As the other methods have similar precision, this accounts for the other 20 percent variation. When the estimates differ considerably, it is difficult to determine which method more nearly approximates the true potency. As the mean differences approximate zero, there is obviously on the whole no difference between the methods, the chief variation being in the estimates of the individual oils. If a mean of all four assays can be considered the most probable estimate of the true potency of an oil, then the most accurate method is the one which will most closely approximate that value. In order to compare the methods in this manner, means of the estimates of
75
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
potency by all methods for each oil and the standard deviations of the differences of the four methods from these means "were calculated. The standard deviations were 15.0, 11.4, 10.6, and 10.0 percent for tibia, extracted toes, fresh toes and t.m.t., respectively, while the mean differences were + 2 . 4 , + 0 . 2 , + 0 . 7 and - 3 . 3 . There seems no doubt from these figures that the toe ash and x-ray methods yield results
42„
15 TIBIAE (%ASH)
40L
5
b
c=l271
15
259
combined reference and unknown data and the combined decoded slopes are entered in the charts. It is evident that the response curves of all four methods are quite similar and although in this particular assay some of the points may not be strictly linear, yet relatively the methods show good agreement. Part of the series of the assays, namely, assays 87 to 95, which involved 1693, REFERENCE- ° UNKNOWN * 0-90,. RADIOGRAPHIC l + L0G.(TM.T) «
04.9 EXTRACTED TOES - (%ASH) bc = 6-83.
15
25~~9"
fe
259
i5
25
DOSE IN UNITS PER 100 GM. FEED (LOGARITHMIC SCALE)
FIG. 1. Response of chicks as measured by four criteria of calcification.
equally as reliable as those by the tibia ash method. The estimates of potency by the tibia ash method, if anything, are more variable than the estimates by the other three methods. In order to illustrate the ranges of response of the four criteria of calcification arid the similarity of their log. (dose)-response curves,.data from assay 104 were plotted on semi-log. paper and reproduced in Fig. 1. Approximately 45 chicks (3 groups of 15 each) were used to determine each of the points on the graph. To facilitate comparison the scale for the radiographic method has been inverted. The regression lines were calculated from the
chicks, was carried out using individual tibia ash, thus furnishing the basis for a comparison of the precision of the tibia ash, and radiographic methods. As in the earlier paper, the data used for this comparison were strictly comparable, each value for tibia ash having a corresponding value for the radiographic method. In Table 3 the fiducial limits in percent, the standard deviations (s) and the decoded slopes (b) of the log. (dose)-response curves have been tabulated and from these the X = (s/b) ratio for each assay has been calculated. I t can be seen that, although X for the radiographic method is greater in every case than that for the
76
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
tibia ash method, yet only in the case of assay 94 does the difference approach significance at the 5 percent point. By combining the 9 comparisons by the x2 method as outlined by Fisher (1934) an overall
This slight but constant difference is also reflected in the fiducial limits. I t is pointed out again that the radiographic technic of Olsson as used in this paper involved only the actual principle of x-raying the chicks
TABLE 2.—A comparison of four criteria of calcification Potency (A.O.A.C. units) as calculated from Assay
No. 84 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 111 112 114 115
Expected potency
Difference as percent of tibia
VtUUCb
Toe ash Tibia
ash
Extracted
Fresh
Log. t.m.t.
1200 2000
1835 3432
1957 4056
2066 4280
2689 3772
200 200 200 400 400 400 400 400 400 400 400
258 294 380 498 422 434 475 519 378 248 324
281 379 424 551 481 507 478 — — 265 244
— — — 540 566 536 — — 248 232
267 289 383 560 408 456 561 516 372 242 252
4000
3664
2962
2727
3228
200 200 200 200 200 100 400 200 400 200
377 233 214 238 401 102 538 447 864 321
377 211 210 226 361 108 536 225 446 246
324 219 189 211 318 110 582 342 587 308
360 199 202 244 386 112 421 269 522 223
2000 1000
3122 1380
2828 1758
2874 1443
3090 1157
400 200
373 224
570 280
566 272
444 247
•
—
.
Toe ash Extracted
Fresh
+7
+13"
+18
+25
+9
—.
+29 +12 +10 +13
— —
+8
Log. t.m.t.
+47 +10 + 3 - 2
+ 1 +12 - 3
+17 +1
+34 +24 —
— '
—
+7
0 -28 -26 -14
- 9 - 2 - 5
- 6
-15
-12 -11 -21
+3
-25 -19 0
-10
+6
0 -50 -48 -23 - 9 +27 +53 +25
.—.
+8 +8 -23 -32
- 4 - 8
+5 +18
- 1 - 2 - 2
-22 -12 - 5 - 6
-4 +10 -22 -40 -40 -31 - 1
+5
+52 +21
-16 +19 +10
+ 1.3
+ 0.5
- 3.1
S.D.
22.6
21.7
18.2
S.E.
4.4
4.6
3.4
Means
measure of the difference was obtained. X2 was found to be 26.036 which with n = 18 yielded a P value of approximately 0.09. Although approaching the 5 percent point, the difference cannot be considered significant from these data. This conclusion substantiates the work of Baker and Wright (1940) although in their data the bone ash tends to be the more variable.
and was not used as a complete method. As it was possible to x-ray the same chicks at both two and three weeks, the radiographic procedure offered the most valid basis for a comparison of the two feeding periods. Accordingly the potency, approximate fiducial limits and X ratios were calculated from a series of 8 assays for both feeding periods. The results of
STUDIES ON THE CHICK ASSAY FOR VITAMIN D.
these comparisons are presented in Table 4. I t should be pointed out that for this particular comparison of X's the data were not strictly paired, that is, if a chick died after being x-rayed at 2 weeks, its 2 week value was not removed. However, as mortality was low in this period (less than 3 per 100), it may be safely assumed that the data are comparable.
77
ference between the estimates of potency determined at 2 and at 3 weeks. Considering the precision of the method, it is evident from the X ratios that in every case the 2 week assay period yields more variable results than does the 3 week period. The differences, although large, are not significant when each assay is viewed by itself or when combined, as the standard
TABLE 3.—Comparison of the precision of the tibia ash and radiographic procedures Radiographic method (log t.m.t.)
Tibia ash method Assay No. 87 88 89 90 91 92 93 94 95 Means
Fiducial* Standard Decoded Fiducial* Standard limits deviation slope X = (s/b)±S.E. limits deviation Percent Percent s s b
Decoded slope b
X = (s/b)±S.E.
Difference tm.t. X tibia X
8S-118 85-118 84-119 87-115 85-118 86-117 86-116 88-114 87-115
3.0382 2.9260 2.8962 3.5029 3.6033 3.6302 4.0631 3.7767 3.9353
9.8075 10.1166 10.6268 13.8683 13.2644 14.1148 13.7714 15.5736 16.2649
0.310+0.031 0.289±0.028 0.273±0.025 0.253±0.023 0.272±0.028 0.257±0.025 0.295±0.029 0.243±0.020 0.242±0.021
83-121 82-123 82-121 85-118 84-119 85-118 82-122 84-119 85-118
0.1854 0.1774 0.1775 0.1982 0.2066 0.2093 0.2479 0.2339 0.2200
-0.5223 -0.5034 -0.5584 -0.6963 -0.7156 -0.7643 -0.6719 -0.7161 -0.7089
0.355±0.041 0.352±0.041 0.318±0.033 0.285±0.028 0.289±0.031 0.274±0.028 0.369±0.044 0.327±0.035 0.310±0.034
+0.045 +0.063 +0.045 +0.032 +0.017 +0.017 +0.074 +0.084 +0.068
86-117
3.4858
13.0454
0.270± 0.023
84-120
0.2062
-0.6508
0.320±0.033
+0.050
Number of chicks
606 509 578
' Approximate fiducial limits at P =0.05. According to the criterion of the B.S.L method, approximate limits are valid for the assays reported in this table.
From Table 4 it can be seen that in all but 2 of the assays the 2 week estimate is higher than that at 3 weeks. The mean difference between estimates of potency at 2 and 3 weeks is 11 percent. The combined probability of the significance of differences between estimates, using the X2 method of Fisher (1934) was found to be greater than 0.80. I t may therefore be concluded that there is no significant dif-
errors of the 2 week data are also relatively large. However, when the probabilities were recalculated, using as standard error of the difference between X's an average value of 0.06 obtained from 3 week data, it was found that all but 2 values were significant at the 1 percent point. The 2 week data may therefore be considered significantly less precise than the 3 week data when judged by the 3
TABLE 4.—Comparison of two and three week data Criterion Assay of Po- Limits* Calcifi- No. tency Percent cation Radiographs
Means
104 105 .106 107 108 111 112 114
113 396 275 584 249 4432 1533 402
Two Week Data
Three Week Data
Standard Deviation s
Slope b
X = (s/b)
80-125 72-139 64-157 66-152 69-146 76-132 59-171 63-158
0.1819 0.1586 0.1653 0.1613 0.1624 0.1463 0.1385 0.1367
0.4561 0.3028 0.2726 0.3015 0.3042 0.2402 0.2285 0.1825
0.399±0.057 0.524± 0.110 0.606±0.146 0.535±0.114 0.534±0.114 0.610±0.102 0.606±0.168 0.749±0.222
69-148
0.1564
0.2861
0.570
' ApproximatelimitsatP=0.05.
Standard Po- Limits* Deviatency Percent tion s 112 421 269 522 223 3090 1124 444
83-121 79-127 73-138 76-132 74-135 83-121 80-126 70-144
0.1905 0.1981 0.2025 0.1920 0.2050 0.1679 0.1492 0.1959
ence between X's
Slope b
X = (s/b)
0.5507 0.5390 0.4669 0.5020 0.4678 0.4375 0.4777 0.3538
0.346±0.044 0.368±0.056 0.434± 0.077 0.382±0.060 0.438±0.079 0.384±0.042 0.312±0.047 0.554±0.125
0.053 0.156 0.172 0.153 0.096 0.226 0.294 0.195
0.402
0.168
77-131 0.1876 0.4744
78
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
week standard of precision. The chief reason for this seems to be that the lower slope (b) of the log. (dose)-response curves at 2 weeks is not sufficiently balanced by a lesser variability (s). Substantially similar observations were made by De Witt et al. (1942). Although for some of the assays reported in Table 4 the approximate fiducial limits, particularly at 2 weeks, underestimate the true width of the fiducial range, yet they show the same
ferent toes on the extracted basis was also investigated at this time and the results of 5 assays are presented in Table 5. In some cases the 3 toes vary somewhat in the interpretation of potency in the various assays. The ash percentages indicate that there are constant differences between the ash content of the 3 toes. The data as a whole indicate that approximately the same variation may be expected between different toes as between
TABLE 5.—Comparison of the ash content and potency calculated from extracted ash data of different toe Toe ash (Percent) Assay No. 93 111 122A* 112B* 114
Inside toes
Middle toes
Outside toes
No. of groups averaged
12.58 13.90 14.06 13.57 14.07
13.39 14.56 14.72 14.40 14.78
13.74 15.04 15.01 14.97 15.41
20 38 15 16 16
Estimation of potency (A.O.A.C. units) Inside toes
Middle toes
Outside toes
483 2868 1390 1191 476
478 2828 1972 1544 570
481 3888 1888 1719 508
* A = A.O.A.C. method. B = B.S.I, method.
trend as the X ratios. Whatever limits are calculated, of course, do not affect the comparison by the X method. In order to furnish some idea as to how the other 3 criteria would react, one of the assays (No. 104) was carried out entirely in duplicate under the same conditions and at the same time. One set of chicks was killed at two weeks and the other was killed at 3 weeks. The estimates of potency were as follows: Radiographic Extracted toe Fresh toe Tibia
At 2 weeks At 3 weeks 112 94 108 87 110 104 98 102
The results of these comparisons, when considered in conjunction with those of Table 4, substantiate the conclusion that there is no significant difference between estimates of potency determined at 2 and at 3 weeks. The reproducibility of results using dif-
the other methods, and that, if toes are used for assay, the same toe must be used throughout any one assay. The use of 2 toes in an assay readily furnishes duplicate results, an average of which yields a considerably more reliable estimate of potency than either one alone. DISCUSSION
I t should be pointed out that, while all available data were included in this series of assays, not all of them would satisfy the criterion of the B.S.I, method for a valid assay. However, as on the average four to five times more groups per assay were employed than is ordinarily considered necessary, the results outlined above may be considered representative of what is to be expected from the four critieria of calcification under average conditions. If the data be viewed from this assumption, they suggest that although the methods are valid for sustain-
STUDIES ON THE CHICK ASSAY FOR VITAMIN D
ing or rejecting a guarantee there is still much room for improvement in the technic as a whole. Certain observations may be made from the comparisons of the four methods presented in this paper, from the point of view both of the results obtained and of the labor involved in the various methods.. All methods show considerable and approximately equal variability in their estimate of the potency of the oils. For an estimate of the precision of an assay involving individual data, the radiographic technic has unquestioned advantages in manipulation and practically equal precision to other methods. If x-ray equipment is not available, the use of toe ash on the extracted basis offers considerable simplification and equal precision to the official tibia ash procedure as shown previously. If a purely routine procedure is required, the use of toes on the fresh basis undoubtedly furnishes the most rapid procedure. The 2 week feeding period seems of particular value for use with fresh toes although it also is applicable to the other procedures for estimates of potency involving composite data. Its application, however, may be limited for use with individual data as the precision is relatively low. In view of the fact that all 4 methods have proved of approximately equal value in this laboratory for assessing calcification in the vitamin D chick assay and that the suggested modifications offer improvement in the official method, it is recommended that more extensive studies be initiated on a collaborative basis to test more thoroughly these criteria of calcification with a view to their ultimate acceptance as alternative official procedures. SUMMARY
A series of 28 assays involving approxi-
79
mately 8,000 chicks has been carried out on various vitamin D feeding oils to compare four criteria of calcification, namely, ash percent of tibiae, fresh toes and extracted toes, and the Olsson radiographic technic. A comparison was also made of the two and three week feeding periods. The data show that the four criteria of response give on the average approximately the same estimate of potency. Different toes are equally useful but are not interchangeable in any one assay. The Olsson radiographic technic appears to be very slightly but not significantly more variable than tibia ash as a criterion of calcification. The two week feeding period is comparable to the three week period as far as an estimate of potency is concerned, but the precision is significantly lower. Considering the labor involved in these procedures, it is therefore concluded that the radiographic technic is unquestionably superior when individual data are desired. For routine assays involving composite data, toe ash on the fresh basis requires the least labor and seems equally reliable. ACKNOWLEDGMENT
The authors are indebted to the Production Service for the loan of the services of Miss V. Reynolds for reading the radiographs. The technical assistance of Mr. P. M. Richard is also acknowledged. REFERENCES
Assoc, of Official Agr. Chem., 1940. Methods of Analysis, 5th Edition, p. 371-372. Baker, A. Z., and M. D. Wright, 1940. Biological Assay of Vitamin D 3 .1. Assay methods at present in use, with particular reference to Olsson's radiographic technic. Analyst 65:326-335. Bliss, C. I., and McK. Cattell, 1943. Biological assay. Ann. Rev. Physiol. 5:479-539. British Standards Institution, 1940. British Standard Method for the Biological Assay of Vitamin D s by the Chick Method. Specification No. 911.
80
J. A. CAMPBELL, B. B. MIGICOVSKY, AND A. R. G. EMSLIE
Campbell, J. A., B. B. Migicovsky and A. R. G. Emslie, 1944. Studies on the chick assay for vitamin D. I. Precision of tibia and toe ash as criteria of response. Poultry Sci. (in press). DeWitt, J. B., C. D. ToUe, H. W. Loy, Jr. and L. F. Knudsen, 1942. Observations on the chick method for the assay of vitamin D. III. Effect of decreasing length of assay period. J. Assoc. Off. Agr. Chem. 15:213-219. Fisher, R. A., 1934. Statistical methods for research workers. 5th Edition, Sec. 21.1. Oliver and Boyd, London. Irwin, J. O., 1943. On the calculation of the error of
biological assays. J. Hyg. 43:121-128. Knudsen, L. F., and C. D. ToUe, 1940. A statistical analysis of the A.O.A.C. collaborative study on assaying vitamin D by the chick method. J. Assoc. Off. Agr. Chem. 23:665-675. Olsson, N., 1936. Untersuchungen fiber die moglichkeit vitamin D—praparate mit wachsenden kiiken zu standardisieren. Archiv fiir GeflUgelkunde 10:423-436. Olsson, N., 1939. A method for quantitative estimation of vitamin D with growing chicks. Seventh World's Poultry Congress. Proc, p. 129-132.