Can the FMR1 Gene Predict Early Ovarian Aging?

“…the evaluation of triple-CGG counts on the FMR1 gene in young women may offer a partial view into their reproductive future.”

Since it has been known for decades that the so-called premutation range (55–200) CGG repeats on the FMR1 (fragile X) gene predispose towards premature ovarian failure (POF) [1], it is actually remarkable that, until relatively recently, nobody had explored the gene in terms of its relationship to ovarian function. This is even more notable since, in 1991, Fu et al. reported a significant distribution peak of 29–30 repeats in the general population, with relatively minor ‘noise’ at lower and higher counts (Figure 1) [2].

Figure 1.

Distribution of CGG triple-repeat counts on the FMR1 gene in a general population.

This distribution peak has no apparent relevance to the known physiologic functions or to neuropsychiatric risks, which are the principal reason as to why CGG counts on the FMR1 gene are currently clinically evaluated [1]. More importantly, current definitions of what represents normal counts or ranges that denote risk (common or normal: <40/45; intermediate or gray-zone: 40/45-55; premutation: 56-200; full mutation: >200), are practically exclusively determined by the risk for neuropsychiatric conditions [1]. Therefore, the reason that nobody has yet explored a possible separate function for Fu's distribution peak of 29–30 CGG repeats is somewhat puzzling.

We were initially attracted to the FMR1 gene because of its known association with POF risk at premutation range counts [1]. However, we thought it somewhat surprising that POF would be the only non-neuropsychiatric risk associated with the gene. Indeed, the risk for POF in women with premutation range repeats appeared to us to be completely isolated from all other pathologies associated with the gene. This raised in our minds the question of whether the risks for neuropsychiatric conditions and POF may not be independent in their respective associations with FMR1.

Under such a hypothesis, neuropsychiatric risks would be associated with CGG repeat ranges, currently well described in the literature (as noted above) [1], but the risk for premature ovarian senescence (POF only being its most severe expression) would be associated with a CGG range that is yet to be established. The current association of POF risk with premutation range CGG repeats, under such a concept, could be viewed as simply coincidental.

Fu's paper supported this hypothesis [2]. The reported distribution peak in the general population, of 30–31 CGG repeats, suggested a ‘normal’ range for ovarian function, since Chen et al. had reported 30 CGG repeats to represent the switching point between the positive and negative effects of the gene's translation [3]. The hypothesis was further supported by recent reports suggesting that high normal (<40/45) and intermediate/gray-zone (40/45–55) range repeats can also been associated with the risk for milder forms of premature ovarian senescence and POF [4,5].

“…repeats on the FMR1 gene could be associated with a risk for premature ovarian senescence, independent of the current classification of what represents normal and abnormal CGG counts.”

Intrigued, we began to explore whether CGG repeats on the FMR1 gene could be associated with a risk for premature ovarian senescence, independent of the current classification of what represents normal and abnormal CGG counts [1]. Under such a concept, POF (now often termed primary ovarian insufficiency [6]) was associated with premutation range repeats only because it represents the end-stage of ovarian senescence and, therefore, its most severe form. Therefore, milder forms of ovarian senescence, characterized by milder FSH elevations and/or diminished anti-Müllerian hormone (AMH) levels (now often termed occult ovarian insufficiency [6]), should be linearly associated with lower CGG repeats below the premutation range.

Confirming this concept step-by-step, we initially demonstrated that the number of CGG repeats to the right of the distribution peak of 29-30, and up to approximately 55 repeats, correlate with a risk for prematurely diminished ovarian reserve [7]. Moreover, this risk appears to be distinct from the risk for prematurely diminished ovarian reserve, associated with autoimmunity [8]. Autoimmunity, of course, represents a major risk factor for POF [9] and we recently reported that risk factors characterizing patients with occult forms of ovarian insufficiency demonstrate the same distribution pattern as patients with POF [10]. Therefore, various grades of premature ovarian senescence, from mild FSH elevations to outright POF, appear to be caused by the same underlying etiologies and pathophysiologies, with FMR1 and autoimmunity each representing approximately a third of the total risk [10].

If higher CGG repeat numbers denote an increased risk for early diminished ovarian reserve, then this should also be reflected in oocyte yields in IVF cycles. In a follow-up investigation, we demonstrated this to be the case [11]. Thus, determination of the number of CGG repeats suggested for the first time a potential clinical utility in infertility.

After establishing that higher CGG repeat numbers reflect the risk for premature decline in ovarian reserve, we wanted to make sure that infertile patient populations also reflected the distribution peak previously reported by Fu et al. [2]. This was important because, considering that women with premature ovarian senescence disproportionately accumulate in fertility centers [12], one could expect biases and possible right-shifts towards higher triple-CGG repeats. This was indeed demonstrated, although only to a very minor degree, thus providing significant confirmation of Fu's original work [13].

In confirming this distribution peak, the question now arose as to what the meaning of abnormally low triple-repeat counts may be? To our surprise, the risk for early-diminished ovarian reserve was almost the sme on both sides of the distribution peak of 29–30 repeats. Indeed, every five additional repeats beyond 30 on the upside denoted an approximately 40% increase in the risk for low AMH (an indicator of low ovarian reserve), while every five repeats on the downside reflected an additional risk of approximately 60% [14].

“…the distribution pattern of abnormal counts differs remarkably between different ethnicities, with Asians being the most and Caucasian being least the homogenous.”

This observation further strengthened the notion that the distribution peak, reported by Fu et al. [2], reflected a significant physiologic function of the FMR1 gene that relates to ovarian reserve. Therefore, we decided to investigate by standard statistical methodology, utilizing box and whisker blots, which of the triple-repeat numbers actually reflects ‘normal’, in terms of the function of the FMR1 gene with regard to ovarian reserve.

We first achieved this in an ethnically mixed population and repeated the analysis later in ethnically defined women of Caucasian, Asian (mostly Chinese) and African descent. Uniformly, box and whisker plots determined 26–32 CGG repeats as a normal range, with 30 repeats (the previously reported switching point between the positive and negative effects of gene translation [3]) as the median [15]. As Figure 1 demonstrates, this range also perfectly reflects the distribution peak reported by Fu et al. [2].

Thus, while all ethnicities appear to share a normal range of CGG repeats, the distribution pattern of abnormal counts differs remarkably between different ethnicities, with Asians being the most and Caucasian being least the homogenous [15].

The definition of a normal CGG count range allowed for the classification of patients with regard to their FMR1 gene status as normal (both alleles in range), heterozygous (one allele outside range) or homozygous (both alleles outside range). When this was done, distinct ovarian reserve patterns became apparent among women in different age groups [16]. During very young ages, normal women demonstrate the best ovarian reserve, followed by heterozygous women and with homozygous women demonstrating the lowest reserve. However, normal women deplete their ovarian reserve very quickly, and by age 33–34 years, heterozygous women, who deplete much more slowly, therefore start exceeding the ovarian reserve of normal women. Homozygous women at all stages demonstrate the lowest ovarian reserve until they are aged in their late 40s, when they also cross the reserve curve of normal women close to menopause.

We interpreted these findings as strongly suggestive of a regulatory role of the FMR1 gene on the recruitment process of follicles from storage at the primordial stage [16]. In normal women, recruitment at young ages takes place quickly and resolutely, leading to a relatively quick depletion of the ovarian reserve pool of primordial follicles by age 33–34 years. An overwhelming majority of conceptions, even today, still occurs at these young ages, and active follicular recruitment during those years, therefore, makes physiological sense.

However, from an evolutionary viewpoint, it also makes sense that nature protects the species by providing a reserve pool of potential mothers who preserve fertility into older age, should a natural disaster afflict the standard pool of women who conceive. Those can be found among heterozygous and, to a lesser degree, homozygous abnormal females who, under control of the FMR1 gene, recruit slower and, therefore, preserve ovarian reserves into older ages.

Therefore, we speculated that the preservation of devastating neuropsychiatric conditions, associated with premutation- and full mutation-range CGG expansions [1], can be seen as the evolutionary price for the previously noted benefits in favor of species preservation [16]. However, the gene can also be seen as utilitarian in daily clinical practice. Two primary clinical applications appear obvious: in infertile women, determination of abnormal CGG repeat counts can immediately help in solidifying a tentative diagnosis of diminished ovarian reserve. This is especially the case in younger women, where such a diagnosis, at times, can be difficult [12]. CGG counts in active patients, as previously noted, also correlate with expected oocyte yields with IVF [11].

However, probably the most important application of FMR1 testing lies in its potential as a screening tool. Approximately 10% of all women are believed to suffer from premature ovarian senescence. Among those, only 1% reach outright POF. This means that 90% of affected women will express milder forms of early ovarian aging and these women, even in competent fertility centers, often go undetected since abnormalities in ovarian reserve parameters will only be subtle (occult) [12].

“…the most important application of FMRI testing lies in its potential as a screening tool.”

Evidence suggests that approximately a third of the 90% of affected women demonstrate a FMR1-associated pathophysiology [10]. Such women can easily be identified as ‘at-risk’ at young ages and then can be carefully monitored with annual ovarian reserve assessments. Since heterozygous and homozygous women demonstrate comparatively low ovarian reserves at young ages [16], this will mostly affect women in these two groups. Once such assessments suggest that if a female is deviating from age-specific normal assessment parameters, she will probably still have a sufficient ovarian reserve to be able to conceive with appropriate treatment, will have time to reconsider her reproductive planning and/or consider fertility preserving steps, such as oocyte or embryo cryopreservation. In recent years, AMH has been proven as a superior ovarian reserve parameter to FSH [17] and has been suggested as a potentially valuable tool in prospectively screening women who are considered at risk [18].

However, one should not forget the normal woman with two normal count alleles. As noted previously, this woman will quite rapidly deplete her ovarian reserve and, by age 33–34 years, will fall below ovarian reserve levels of heterozygous women [16]. It is this patient population who will actually be at the highest infertility risk should they defer pregnancy into older age. Therefore, they may benefit most from fertility preservation at young ages, when oocytes have the highest reproductive potential.

Thus, the evaluation of triple-CGG counts on the FMR1 gene in young women may offer a partial view into their reproductive future. Since CGG counts can, at relatively low cost, be assessed via a routinely available commercial laboratory assay, such evaluations can be initiated immediately, once other investigators confirm our data regarding the predictive nature of CGG counts. Small-scale studies have already been published [19,20]. However, larger ones will have to be performed before wide-scale testing can be routinely recommended in order to screen for risk.

It is important to remember that abnormal CGG counts denote risk only, and do not per se reflect a diagnosis. Once risk is established, a smaller potential patient population then lends itself to closer prospective, and therefore more cost-effective, monitoring with AMH [17,18]. In addition, it is important to remember that the FMR1 gene represents only one among a number of potential etiologies leading to a risk for premature ovarian insufficiency and senescence. After genetic causes, autoimmunity represents the second most common cause [10]. Therefore, prospective screening for the risk for premature ovarian aging may, in the future, become a more universal concept, involving multiple etiologies and pathophysiologies.