Do BRCA1/2 mutations and low FMR1 alleles interact or not?

Ricci et al¹ report in this issue that, contrary to prior suggestions,² FMR1 mutations may not be predictive of BRCA mutations status. Collaborating with Austrian colleagues, we in 2011/2012, almost accidentally, stumbled over unexpected results when looking at CGG repeats (CGGn) on the FMR1 gene in carriers of BRCA mutations (BRCA1 and BRCA2).² The idea for this study came from reports that selected FMR1^{2, 3} and BRCA1⁴ mutations, independently, appeared associated with premature ovarian senescence. Austrian colleagues, who had access to a range of BRCA1/2 mutation samples (64 different mutations) from a local Austrian population, and ability to utilize a well-established FMR1 assay, graciously offered their cooperation.²

BRCA1/2 and FMR1 results from investigated patients were, independently, transmitted from two University centers in Graz and Vienna to our center in New York, where associations were determined. Mutations of FMR1 were defined based on a newly defined ‘normal’ CGGn of 26–34.³ Based on this range, patients are defined as normal (norm, with both alleles in range), as heterozygous (het, with one allele in and one outside range) and as homozygous (hom, with both alleles outside range). Het and hom patients are then further sub-categorized as low (CGGn<26) or high (CGGn>34).

Utilizing these newly defined mutations, we reported significant associations with ovarian aging patterns,^{2, 3} ethnicities/races,^{5, 6} evidence for immune system activation² and, most recently, methylation patterns.⁷ Most striking variations were seen among het mutations, with low alleles associated with the most severe adverse outcomes, like lowered pregnancy chances with in vitro fertilization (IVF),^{2, 6} increased risk toward immune system activation^{2, 6} and abnormal methylation.⁷

In different female populations norm represented uniformly approximately 55–60% of women, het 40–45% and hom approximately 6–8%, a distribution pattern also confirmed in Austrian non-BRCA carrier women.² Austrian BRCA1/2 mutation carriers, however, differed to surprising degrees: they demonstrated almost complete absence of all constitutional FMR1 genotypes except for those with low (CGGn<26) alleles.²

In their here presented report Ricci et al¹ repeated this study, though with greatly differing results. Largest differences were seen in norm and het-norm/low women: while Austrian BRCA1/2 mutation carriers, comparatively, demonstrated significantly reduced numbers of norm FMR1 genotypes (6.1 vs 58.0%) and significantly increased numbers of het-norm/low (78.8 vs 20.5%), Italian BRCA1/2 carriers demonstrated practically identical numbers for norm FMR1 genotypes as non-carriers (51.2 vs 51.0%), although still somewhat elevated numbers for het-norm/low (32.6 vs 23.1%). Het-norm/high and hom patients (combined for all sub-genotypes) in both studies represented small numbers, and, therefore, appear insignificant for this discussion.

Observed differences between the data reported by Ricci et al¹ and ours,² therefore, raise three important questions; the first one being why are there such differences between studies performed in neighboring countries?

Explanations can, most likely, be found in two significant differences between the two studies: both studies are small but the Italian study is even smaller (43 vs 99 patients), increasing risks for patient selection biases. At 51.0%, for example, the norm FMR1 population appears significantly underrepresented in comparison to average populations in Austria² and the United States.^{2, 8} Likely more important is, however, that none of the 32 BRCA1/2 mutations in the Italian study was also represented among 64 different mutations observed in Austrian patients. In other words, Austrian and Italian women represented completely different BRCA1/2 mutations.

This may surprise, especially in neighboring countries. Austria and Italy, however, are known to demonstrate highly varying BRCA mutations (Table 1),⁹ and even in very small regions within Europe, and in relatively small populations considerable variations in BRCA mutations have been reported.^{9, 10}

Table 1 Most common BRCA1/2 mutations in Austrian and Italian women^a

If one assumes that differences in BRCA mutations in both studies are responsible for variations in the distribution of observed FMR1 mutation, the next question to be resolved is why different BRCA mutations may affect FMR1 mutation distribution differently?

Observing in Austrian BRCA mutation carriers, with relatively few exceptions, only low FMR1 alleles, we concluded that BRCA mutations, almost universally, had to be embryo-lethal, unless rescued by a low FMR1 allele.² Ricci et al¹ now, rightly, questions our conclusions, and suggest that low FMR1 alleles, therefore, cannot be used as a surrogate for BRCA1/2 screening, as we had suggested. They may be correct if the small study size of our data set resulted in an unfortunate statistical coincidence. Another possible explanation for observed discrepancy in findings, however, is that not all but only selected BRCA1/2 mutations are embryo-lethal. Interesting to note in such a context is that even Ricci’s Italian BRCA mutation carriers, comparatively, demonstrate nominally more het-low FMR1 alleles than their own non-carriers (32.6 vs 23.1%) and than reported by us in general populations.^{2, 8}

Rodent models of BRCA support potential embryo lethality of the BRCA gene, though degrees of lethality vary between models,² suggesting that embryo lethality, indeed, may vary between BRCA mutations.

We hypothesized, based on noted Austrian patient data, that rescue of BRCA mutation-carrying embryos by a low FMR1 allele may explain the so-called ‘BRCA paradox,’ defined by antiproliferative effects of BRCA in embryonic tissue (ie, potential embryo lethality) and proliferative activity in malignancies.² We furthermore hypothesized that low FMR1 alleles, in rescuing BRCA mutation carrying embryos, block antiproliferative effects of BRCA1/2. Surviving embryos, therefore, lack these antiproliferative characteristics of their BRCA1/2 genes, resulting in surviving adults being exposed to proliferative effects and BRCA1/2-associated cancer risks.

The study of Ricci et al¹ allows for a mild adjustment of this hypothesis: since now not all BRCA1/2 mutations appear equally embryo-lethal, not all BRCA1/2 mutations require rescue from embryo lethality by a low FMR1 allele. Only embryo-lethal BRCA1/2 mutations will require such rescue. Only adults with potentially embryo-lethal BRCA1/2 mutations, therefore, will experience the low FMR1 allele-induced effects of blocked antiproliferative BRCA1/2 mutations. A so expanded BRCA/FMR1 hypothesis, therefore, suggests that primarily embryo-lethal BRCA1/2 mutations are also oncogenic.

This modified hypothesis of potential BRCA/FMR1 interactions, paradoxically, is actually easier to confirm (or refute) than the original hypothesis: if correct, low FMR1 alleles can be expected to be disproportionally more frequent in breast and ovarian cancer-stricken BRCA1/2 mutation carriers in comparison to patients with the same cancers lacking BRCA mutations. We are, indeed, currently investigating this question in a North American patient population, and are looking forward to similar investigations from Ricci et al¹ and colleagues in other countries.

As proposed in our initial publication,² the interplay between BRCA and FMR1 genes, as of this moment, remains a hypothesis in need of confirmation. Sometime in the future the paper by Ricci et al¹ may be viewed as a first step in its confirmation or repudiation. Either way, their work deserves appreciation