AN EVALUATION OF THE NUMERICAL CRITERION FOR "POSITIVE" WITH SKIN TESTS USING MYCOBACTERIUM LEPRAE ANTIGENS

'Abstract: There has been disagreement regarding the reaction sizes which could demarcate "positive" and "negative" reactions with Mycobacterium leprae antigens. In this investigation, positives and negatives were differentiated using statistical methods. The results of this analysis were in agreement with the commonly used reaction size of 3mm and larger being "positive" far Mitsuda reactivity. With Fernandez reactivity there is disagreement with the earlier recommended lOmm reaction size, the conclusion of this study being 3mm or more be considered positive,. With reactivity to Soluble Protein Antigen, there is no earlier data and it is suggested that reactions of 3mm or less be considered "negative" and 9mm or more "positive".


Introduction
In clinical or epidemiological pra-ctice, it is common to label a skin test as "positi~e" or "negative". What is meant here, is that either, there is 'specific' immunological reactivity or nonreactivity, respectively, or that an individual is infected or not infected, (i.e. with an infectious agent, e.g. with a given species of Mycobacterium). The demarcation of "positive" (or "reactor" or "infected"), and "negative" (or "nonreactor" or "not infected"), is often not clear and bedevilled by other factors which interfere in the interpretation. In the evaluation of mycobacterial infections there would intrude cross sensitisation by shared or similar antigens, and perhaps reactivities to multiple infections. Therefore in the clinical situation, the decision of the numerical criterion for separation of "positive" and "negative" is made on the basis of which value would give the least error in interpretation. An example in which such a criterion is commonly used, is in the tuberculin test, where in tropical countries, such as Sri Lanka, the cross reaction and multiple infections probably occur commonly.
In skin tests with Mycobacterium leprae antigens, different workers have used different numerical criteria of "positivity". Further, the qualitative criteria used, (erythema, swelling and induration), also have been different. Many authorities have not defined the rationale for the use of the criteria they have employed. Latterly, there has been more uniformity, with the World Health Organisation's Expert committee2 I * and International Leprosy 7 9 setting criteria for the evaluation and interpretation of lepromin tests. However, there has been no attempt t o scientifically demarcate between "positive" and "negative" reaction sizes with skin reactions t o Mycobacterium leprae antigens.
-In this report we have attemptend t o evaluate what reaction size should be used to distinguish between the positive (reactor) aild negatve (non P reactor) with skin tests with M. leprae antigens.

Materials and Methods
The results evaluated here were from skin test surveys using Lepromin A (bacillary content 3 or 4xl0'/ml.) and a soluble protein antigen (SPA) of M. leprae (protein content 10pg/ml. The results evaluated here are of adults (persons 1 2 years of age and above) in four different localities in Sri Lanka.
The analysis described here was based on the following observations and assumptions: (1) .The observation of reaction patterns with the three types of reactivity, showed that,, in general, the patterns were bimodal -' .with modes of reactors and non-reactors.

(3)
The assumption that the distribution of reactors was a normal distribution.
(4) The assumption that the point of intersection of the distributions of reactors and non-reactors,' if could be separately drawn, would be the best point of separation of the two (viz. "positive" and "negative"), in the whale unseparated, naturally seen distributions.
i Thus the problem here would be to separate out the distributions of reactors and non-reactors from a mixed distribution as occurs in general c population groups.

'Yositive " Reaction Size t o M. leprae Antigens
In examining the distributions of reactions of the different types it is seen that there is an area where there is an obvious intermingling of both reactor and non-reactor distributions, while elsewhere in the curve, it is likely that the reactor distribution includes only individuals of that one group.
Thus based on the assumption (3) above, and using reactions in the segment of the curve which probably belongs to the reactor distribution it would be possible to compute, mathematically the hypothetical best fit normal distribution of reactors, for any given type of reactivity using the formula: . , From the latter then it would be possible to deduce the probable distribution ofmon-reactors. Then, it is possible to obtain in most instances a separation of the cohorts of reactors and non-reactors and evaluate by simple visual inspection which the best point of separation of the two groups would be. In most instances, this would be the point of intersection of the cohorts of reactors (computed) and non reactors (deduced).
I t was possible to separate out those tested into several cohorts on the basis of geographical area of residence and BCG vaccination statusboth of which may be possible vazfiables affecting the frequency distributions of reaction^.^ v 2

3, Results
Probable normal distributions -of reactors, were computed for eight cohorts each of Fernandez and Mitsuda reactivity and four cohorts of SPA reactivity. From the latter, the probable distribution of non-reactors was deduced. In  Table are presented the possible cut off reaction sizes obtained from a visual inspection of the computed reactor, deduced non-reactor, and natural total distribution. The point of intersection of the former two distributions could be considered to be the satisfactory reaction size for discriminating between reactors and non-reactors. However, it should be noted that while with some (e.g. Mitsuda reactivity) w! the separation of reactors and non-reactors, both in the natural distributions, and computed and deduced distributions is clear with others, such demarcations are not as clear.

Discussion
In his orjginal paper, ~i t s u d a~~ considered a reaction of less than 10 rnm "weak", 10-20 mm "moderate" and larger than 20 mm "strong". The criteria used for "positivity" deterinination has since undergone extensive change. Thus with Mitsuda reactivity recent recommendations are of a smaller reaction size for the demarcation of "positivity". With 19 papers reviewed in this respect, nine used 5 mm 1 5 -7 p 9 -(over half the number of the latter reports were by the same researeh p u p ) , three used 4 mm1913714, dnd seven used 3 mm.29831611s;19*219 Thus among research workers 3 mm seems to be a more favoured reaction size used for demarcation). Recent workers have no doubt also been influenced by the decisions of International Leprosy ~o n g r e s s e s ,~~*~~ the most recent WHO Expert committee3 to examine this problem, and of an influential text book4 to recommend this reaction. size. On the other hand with Fernandez reactivity, though the available Literature is lesser the majority of workers have considered 10 rnm and above as "positive" (n -7)195*8-';14v2 while one paper considered 5 m m The two international Congresses2 9 ' and the WHO Expert Committee2 which examined this problem, all recommended 10 mm.
Evaluating our data. it is seen that with Mitsuda-reactivity, almost any degree of reactivity, if present, is a "positive". Thus a level of 3 mm or more as recommended, is an acceptable level at which to demarcate a ("reactor" or) "positive". With Fernandez reactivity, the conclusions of this analysis . are totally in disagreement with that of previous recommenda'tions. Of eight cohorts six show a clear demarcation at less than 3 mm. Thus 3 mm would seem a satisfactory point of demarcation.
With SPA reactivity, only. four cohorts were available for analysis; From these, the definition of the discriminating reaction size of "positive" is not.clear. I t is seen that the reactors form a distinctive sub-distribution within the total. The demarcation seems to be likely to be at between 3 and 9 mm. Thus the compromise reaction size of 5 mm as the cut off point was considered as the best, being in between. However, another and possibly more valid interpretation is that' those with reactions of 3 mm or less be .. defined as definite "negative" or "non reactor", A d with 9 mm or more, as definite "positive" or "reactor", and those of 4 to 8 mm, of doubtful status.
The analysis here has been made on the assumption that each whole population group distribution is of only two components, viz reactors and non-reactors. I t should be noted, that this, may not necessarily be soand that the whole distribution may contain different unrecognised cohorts. For example, in mycobacterial infections, skin testing may evoke a response of cross sensitisation rather than of specific infection. At present it is not possible t o identify these with a single skin test. Cross reactivity with the Mitsuda reactionmay lead to definite positivity, and not cause intermediate size reactions which interfere with interpretation. On the other hand cross reactions which interfere with interpretation, producing intermediate sized reactions, may be a possibility with Femandez and SPA reactivity (considering the nature of these reactions), though probably to a lesser degree than with tuberculin reactivity. The cross reaction and other possible factors which may have influenced reactivity have been ignored in this analysis.
This investigation was supported by grants from the World Health Organisation, Leprosy Section, Geneva, and the University of Peradeniya The antigens used in this study were supplied by Dr. R.C. Hastings