Risk of expansion to a full mutation in maternal premutation alleles
It has been demonstrated that the AGG interruption pattern in
FMR1 intermediate and premutation alleles can refine the risk of instability during parental transmission and it should be validated in genetic counseling prediction models [
10,
11]. To increase the utility of AGG interruption information to better predict the risk of expansion to a full mutation when counseling a premutation mother, we have updated our previous study that investigated the role of AGG interruptions in the stability of
FMR1 during maternal transmission.
A total of 710 transmissions from 525 premutation carrier mothers were considered in this study, approximately doubling the transmissions used in a previous study [
10]. This collection includes transmissions from the UC Davis MIND Institute (135 from 102 mothers), the Hospital Sant Joan de Déu (105 from 84 mothers), the University of Padova (16 from 14 mothers), Rush University (86 from 61 mothers), and our previously published study (368 from 264 mothers). Thus, the updated cohort is both more extensive and more genetically diverse than our previous study, which included only participants who were recruited through the UC Davis MIND Institute.
Initially, we recalculated the risk of an allele expanding to a full mutation during maternal transmission using total CGG repeat length and number of AGG interruptions of the maternal premutation allele. In agreement with our previous report [
10] alleles that had a total length of approximately 75 CGG repeats were determined to have the greatest difference in the frequency of expansion based on the number of AGG interruptions present. The model predicts that 83% (CI: 72.0%, 89.7%) of alleles with 75 total CGG repeats would expand to a full mutation when no AGG interruptions are present (75 CGG repeat pure stretch), but only 17% (CI: 8.2%, 31.7%) would expand to a full mutation when 2 or 3 AGG interruptions are present. All predicted risks are within the confidence intervals estimated in the previously published model. Alleles with 0, 1, 2 or 3 AGG interruptions are all estimated to expand to a full mutation in more than 90% of transmissions when the total length exceeds 95 CGG repeats.
Three of the 525 maternal premutation alleles had 3 AGG interruptions, we combined this group with alleles that were interrupted with 2 AGGs. It is likely that analyzing more transmissions of alleles with 3 AGG interruptions will show differences in the frequency of full mutation expansions when compared to alleles with 2 AGG interruptions. However, it has been shown in several studies that alleles with greater than 2 AGG interruptions are rare compared to the commonly observed patterns with 0, 1 or 2 AGG interruptions [
4,
11].
An increase in the odds of expansion to a full mutation was found to exist for each additional CGG repeat, but the odds of expansion decreased by 2.4- or 23-fold (243% or 2,434%) with 1, or 2 or 3 AGG interruptions, respectively. Additionally, a 16% increase in odds of expansion to a full mutation (defined as the probability of a full mutation child divided by the probability of a premutation child) occurred for every year of the mother’s age at the time of childbirth. While changes in the total length of the CGG repeat locus has the greatest affect on the risk of the allele expanding to a full mutation, our findings indicate an additive effect of AGG interruptions and maternal age, which appear to substantially influence the frequency with which expansions are expected.
The observation that maternal age is a risk factor for expansion to a full mutation during maternal transmission is an intriguing finding from this study. Previous studies have also examined maternal age as a variable associated with instability of the CGG repeat locus. Two studies did detect a correlation between age of the mother and the size of the CGG repeat. A sibling study found an increase in the mean size of the full mutation allele in the younger sibling [
24], and a study of premutation carrier offspring observed an association between maternal age and mutation status (premutation or full mutation) of the offspring, however when samples retrospectively collected were analyzed separately from samples collected prospectively, the maternal age effect was only present for the retrospective study, suggesting that recruitment bias rather than a true age effect could be responsible for the findings [
25]. No statistically significant association of maternal age upon the risk of expansion to a full mutation was found in two more recent studies that incorporated AGG interruptions into the analysis although the sample size included was much smaller than the one presented here [
10,
11]. As a result, additional data are warranted to further explore the risk associated with maternal age. If supported by further studies, then maternal age would be the first risk factor for CGG repeat expansion that an individual can use to lower their risk of having a child with FXS. This knowledge would also provide an increased incentive for individuals to be screened for a premutation allele at an early age.
The observed maternal age effect in our cohort could be explained by ascertainment bias; having a child with a full mutation may influence reproductive behavior, causing the women in our sample to delay or continue child-bearing until older ages. The age of mothers with premutation children was very similar to mothers with full mutation children, however further analyses should be performed on prospective cohorts in order to verify the role played by maternal age in risk of expansion to a full mutation.
The observed data for three separated collection sites (UC Davis, Rush University, and Hospital Sant Joan de Dèu), which included 135, 86, and 105 samples, showed a greater fluctuation in the frequency of alleles expanding to a full mutation due to a small sample size. However, the same key features were present for each cohort; the bivariable risk model (total CGG length and number of AGG interruptions) consistently predicted an increase in risk of expansion of the allele to a full mutation when total length increased in size. The observed similarities of predicted risks between cohorts suggest that the usage of a single model in a genetic counseling setting will be beneficial across various ethnicities; however more studies are warranted to further confirm these findings.
A model that calculates risk of expansion to a full mutation using pure CGG stretch, number of AGG interruptions, and maternal age produced the lowest AIC score of 235.6. Indeed, when pure CGG stretch is replaced with total CGG length the model has a higher AIC score of 241.6. However, while the model that incorporates pure CGG stretch is nominally the best fit model using the AIC, it predicts the highest risk of expansion to occur if an allele has 1 AGG interruption compared to having 0, 2, or 3 AGG interruptions, an inconsistency that is likely the result of an insufficient number of observations in the group with 2 or 3 AGG interruptions. A revised risk model calculated using logistic regression of the 710 maternal transmissions, considered the total length of the CGG repeat allele, the number of AGG interruptions, and the age of the mother at childbirth. This model was determined to be more suitable than that which considered pure CGG stretch instead of total length, confirming the previous data on a smaller sample [
10].
Previous studies have shown significant differences in the frequency of instabilities between male and female parents [
11]; however we did not observe a significant difference likely due to the small sample size of paternal transmissions included.
The odds of instability increased significantly with CGG repeat length (
P <0.001) and decreased significantly with number of AGG interruptions (
P = 0.002) in the parental allele, consistent with previous reports [
11].
Magnitude of expansion was significantly correlated with total CGG repeat length in our maternal cohort and was significantly reduced when more AGG interruptions were present in the maternal allele, these findings are consistent with Nolin et al. [
11] where the presence of AGG interruptions was shown to decrease the magnitude of instability. Inclusion of paternal transmissions had only a minimal effect on the outcomes of the above results.
When parental age was tested to determine if it could significantly predict either the frequency of instability (excluding expansions to full mutations) or the magnitude of instability, no effect was found for either maternal (P = 0.919, instability; P = 0.774, magnitude) or maternal and paternal transmissions (P = 0.892, instability; P = 0.907, magnitude).