A t the tim e o f the co -tw in interview s, none presented w ith any sym ptom s th at a fte r fu rth e r investigation were considered to be a feature o f M ND, In addition, none o f the co-tw ins who died from other disorders prior to the study period, had undisclosed M N D according to the relatives interview ed. T herefore among this tw in sam ple, 4 MZ probands from 2 pairs were id entified as concordant fo r M ND , but 0 DZ probands and co -tw in s were found to be concordant.
Before a concordance rate for M ND among M Z twins was calculated, two o ther factors had to be considered: 1) W hether any o f the concordant probands had a fam ily history o f FM ND; and 2) how many o f the M Z pairs, w here the c o -tw in had also died, should be considered discordant for MND.
Proportion of Tw ins from FM ND Pedigrees: Am ong the seventy two probands, w here a fam ily m edical history was taken, 5 probands (3 M Z and 2 DZ) from 4 fam ilies were identified with a positive history o f FM ND. In addition there was no know n fam ily history of FM N D reported for 10 probands whose co -tw in died as an infant. T h erefo re, the proportion o f probands found to have a positive history o f FM N D was approxim ately 6.0% . In two fam ilies, a parent had died from MND; in one fam ily a parent, and probably a sibling had su ffered from MND; and in the fo u rth fam ily, a m aternal aunt had probable M ND, and the daughter of the proband had diagnosed MND.
Only two probands from 1 M Z pair, among the fo u r fam ilies id en tified , were concordant for MND. The other M Z co -tw in had rem ained discordant, since the onset o f his tw in’s illness for 6.5 years when interview ed, and will be followed up in the fu tu re. One like-sexed DZ co twin had been discordant for 7.5 years, and one unlike-sexed co -tw in died from acute myeloid leukaem ia 5 years a fte r the onset o f the proband’s illness.
D iscordant A dult D eath C o-tw in Pairs: When considering discordance fo r M ND among probands and their co-tw ins, tem poral spread o f the disease may be an im portant factor. No figures are presently available, though the variation in age o f FM N D su fferers, would indicate that M N D has a long latency, and possible large tem poral spread. T herefore follow ing epidem iological advice, it was decided that when the co -tw in died before the proband (3 M Z pairs and 11 DZ pairs), these should not be considered discordant fo r the disease when calculating the concordance rate.
Increased Risk for C o-tw ins: A very simple and crude estim ation o f the increased risk for the co-tw ins o f M ND affected twins can be obtained using the raw data. O bviously, an
unknow n proportion o f this risk will be the result o f genetic factors, and shared environm ental exposures.
The risk o f developing M ND among the general population o f E ngland and Wales is approxim ately 2:100,000 (see table 1), assum ing M ND is a random sporadic disease. H ow ever, among the non-F M N D twin sam ple, who reached the d efined risk period, this chance was 2:65 over 11 years; a risk 140 times greater than that for the general population. Am ong the M Z twins alone, the chance is 2:21 over 11 years; a risk 433 tim es greater than that fo r the general population. T herefore, this suggests that there is probably a significant genetic influence in the aetiology o f M ND.
M Z Proband C oncordance Rate: A crude M Z proband concordance rate for this twin sample was calculated using the basic M Z proband num bers obtained [15.4% (S ta n d a rd E rror - SE: 7.1% )]. This rate increased to 17.4% (SE: 7.9% ), when the M Z probands w ith a co -tw in dying before them selves were excluded from the calculation. When estim ating the genetic co ntribution to an apparently sporadic disease, any probands w ith a fam ily history that strongly suggests dom inant inheritance should be excluded (Em ery 1986; A lberm an and M acdonald personal com m unications). Two M Z probands m et this criteria, which yielded an am ended M Z proband concordance rate o f 10.0% (SE: 6.5% ). In turn this produced the follow ing estim ate for the M Z "correlation of liability" (Smith 1974) ’r’ = 0.717 (SE: 0.130). These figures are based on one pair, but confidence that this figure truly reflects the genetic influence in M ND is supported by the pair being fem ale, therefore th eir disorder cannot have been K en n ed y ’s Syndrom e, and that there was no suspicion o f any other affected fam ily m em bers. The tw ins died from a rapidly progressing M ND w ithin 20 m onths o f each other aged 57 and 59 years respectively.
DZ Proband C oncordance Rate: It was not possible to calculate a proband concordance rate for the D Z twins as no concordant pairs were found. This is probably an artifa c t o f small num bers, w hich has arisen despite collecting all the tw in pairs from the M N D population iden tified , because M ND is relatively rare. T herefore, unless M ND was highly genetic, it would be very unlikely that a concordant DZ pair would be found. This situation has occurred in previous tw in studies o f rare diseases (R u tter et al 1990). Two logical alternatives can be applied to provide estim ates o f the DZ concordance rate so h eritability calculations can be made. These are:
1) U sing the sibling rate for the disease calculated from all the siblings who lived to adulthood (> 20 years) among the twins fam ilies.
2) C alculating the DZ proband concordance rate for the tw in sample obtained, assum ing the next DZ pair identified would be concordant fo r the disease.
These two solutions can be used to provide a m inim um and m axim um boundary for the DZ concordance rate. Among the sibling data collected fo r the study tw ins, only 1 in 399 had probable M ND , giving a m inim um estim ate for DZ concordance o f 0.25% (SE: 0.25% ). If it is assum ed that the next DZ pair identified would be concordant fo r M ND , a m axim um estim ate for DZ concordance w ould be 2.22% (SE: 2.20% ). U sing these figures to calculate the lim its o f DZ "correlation o f liability", the true figure m ust fall betw een ’r ’= 0.291 (SE 0.004) (w hen utilising the m inim um DZ concordance rate), and 0.525 (SE 0.101) (when utilising the m axim um rate).
G enetic C ontribution to Sporadic MND: U nfo rtu n ately , w ithout a "correlation o f liability" estim ation for the DZ tw ins, calculating the h eritability of M ND would be prohibited. H ow ever, by substituting in the m axim um and m inim um DZ "correlation of liability" estim ates above, the range in which the true coefficient of genetic determ ination (’O ’) can be calculated. U sing ’G ’ = 2(rMz - r^^) (Smith 1974), ’G ’ ranges betw een 0.383, when substituting the m axim um DZ ’r ’, to 0.852, when substituting the m inim um DZ ’r’. T herefore the true co efficien t o f genetic determ ination for M ND m ust lie som ew here betw een these two extrem es.
A calculation com m only given to support the findings o f twin studies is to com pare the M Z and DZ pairw ise concordance rates. In this case such a com parison is rather meaningless because o f the small num bers o f twins and 0 values involved, due to M ND rarity , and not surprisingly the d ifferen ce betw een the M Z and DZ pairw ise concordance rate is not statistically significant (z = 0.41; p = 0.68). N either is it when the two given alternative DZ rates are substituted. An alternative m ethod o f displaying the d ifferen ce betw een M Z and DZ pairw ise concordance rates as a ratio (G ottesm an & Shields 1982) is also im possible with a DZ rate o f 0/44. How ever, when using the two alternative DZ rates above, m eaningful figures are obtained. If the m axim um DZ concordance rate is substituted the M Z:DZ ratio = 2.25, and taking the m inim um DZ rate the MZrDZ ratio = 19.95.