In the present study, we assembled data on 97 patients with GRN mutations from a 46-site international collaboration. This is the largest collection to date of GRN mutation cases and allows estimations of relative frequencies of GRN mutations as well as comparisons between GRN+ FTLD-TDP and GRN− FTLD-TDP.
We found 50 unique GRN
mutations in our 97-individual cohort. Of these, the most common was c.1477C>T (p.R493X), found in 18.6% of our GRN
cases. Our result accords with previous studies showing c.1477C>T (p.R493X) to be the most common GRN
mutation in one large US series16
and several smaller US series,38,39
as well as one of the most common GRN
mutations in a large UK series.40
We did not find any meaningful phenotypic differences between the c.1477C>T (p.R493X) mutation carriers and patients with other GRN
mutations, although our clinical data were limited to demographic details and diagnosis. Thus, we suspect that the preponderance of men observed in our c.1477C>T (p.R493X) carriers may be due to chance, as it has not been previously described and cannot be easily understood from a mechanistic perspective.
The second most common GRN
mutation in our study was c.26C>A (p.A9D), found in 6.2% of cases, supporting prior studies that showed this mutation to be relatively common.16,41
Interestingly, despite limitations in interpreting data from only 6 individuals with the c.26C>A (p.A9D) mutation, there appeared to be a younger age at onset, younger age at death, and higher preponderance of diagnoses within the Parkinson/Parkinson-plus spectrum for this particular GRN
mutation. Indeed, unlike most other GRN
mutations, c.26C>A (p.A9D) has been shown to not result in mRNA haploinsufficiency, although there may be protein haploinsufficiency because of an inability to secrete functional progranulin protein.20,21
In light of these differences in pathogenic mechanism, it would not be surprising if carriers of this particular GRN
mutation exhibited some differences in phenotype; this question could be addressed in studies with more c.26C>A (p.A9D) mutation carriers.
Accurate estimates of relative frequencies of each GRN mutation may be affected by a number of factors. First, inclusion of individuals who are closely related may over represent the frequency of a particular mutation. However, we minimized this possibility by entry criteriaand identity-by-state analysis. Second, sampling from different geographic regions/ethnic groups to different extents may affect the results. Our study included only individuals of white ancestry from Europe, North America, and Australia. Accordingly, our results may not be representative of other ethnicities and regions of the world not participating in our collaboration.
Included in our study were 2 GRN mutations that, to our knowledge, have not been described previously: c.139delG (p.D47TfsX7) and c.378C>A (p.C126X). We considered them to be pathogenic mutations because they both result in premature termination of transcript either by nonsense or frameshift mechanisms. Clinically and pathologically, neither show particularly exceptional features.
Because data on FTLD-TDP cases were collected by our collaborative group, with a special emphasis on carriers of GRN mutations, we could not assess the frequency of individual GRN mutations or GRN mutations as a whole in FTLD-TDP. However, the inclusion of 97 individuals with GRN mutations in one study allowed us to analyze in a statistically meaningful way whether there are group differences between GRN+ FTLD-TDP and GRN− FTLD-TDP. A subset of patients with GRN mutations included in this study has been reported in smaller studies from individual sites; however, the objective of the present study was to provide an overview of GRN+ FTLD-TDP, justifying our inclusion of these patients in our analyses.
We have provided clear evidence that people with FTLD-TDP who have GRN
mutations are younger at the onset of the disease and at death and have a lower incidence of concomitant MND compared with FTLD-TDP patients without GRN
mutations. Prior studies16,42
have shown younger age at onset for particular mutations in particular families, but, to our knowledge, an effect across GRN
mutations has not been shown. The relatively low frequency of concomitant MND in GRN
+ FTLD-TDP accords with prior studies.37
Although information is sparse on some patients noted to have MND, it does appear that MND is a bona fide feature of some cases of GRN
+ FTLD-TDP. In addition, our data support suggestions from prior studies6,28,37
+ FTLD-TDP may be associated with parkinsonian features more frequently than GRN
; this effect may be driven by the c.26C>A (p.A9D) mutations. It has been shown43
+ FTLD-TDP has a global mRNA expression signature that is distinct from that of GRN
− FTLD-TDP, so differences in clinical phenotype between these 2 groups are not altogether surprising.
In interpreting our data, we considered the possibility that 1 or 2 very common mutations with younger age at onset and less MND might be responsible for our findings. However, reanalysis of our data excluding either or both of the 2 most frequent mutations—c.1477C>T (p.R493X) and c.26C>A (p.A9D)—did not significantly change these results. We also considered the possibility that erroneous or missing information, always a factor in retrospective multicenter studies, or heterogeneity in ascertainment of disease and assignment of clinical diagnosis could account for our results. However, several lines of evidence make this possibility less likely. First, the omission of entire groups sharing a mutation (up to 24 GRN+ FTLD-TDP cases) did not change our main findings. Second, demographic and clinical details, along with tissue or DNA, were collected from the contributing site for the majority of patients with FTLD-TDP (294 of 550) without knowing a priori whether the individual harbored a GRN mutation; mutation status was determined by sequencing at the University of Pennsylvania. This makes it less likely that systematic reporting or ascertainment bias could create the observed differences. Indeed, we secondarily analyzed the finding that people with GRN mutations had both a younger age at onset and at death compared with those with no GRN mutations in the subgroups of patients with GRN status determined by the University of Pennsylvania vs by the contributing site and found no substantive differences in our results.
The 2006 discovery5,8
that mutations in GRN
are a cause of FTLD-TDP was a landmark research finding for a group of heterogeneous neurodegenerative dementias (ie, FTLD) that remain enigmatic diseases with respect to many key issues, including their prevalence and their underlying mechanisms of disease. In the present study, we have assembled the largest collection to date of GRN
mutation cases, evaluated the relative frequencies of individual GRN
mutations, and demonstrated that GRN
+ FTLD-TDP differs from GRN
− FTLD-TDP in key clinical features such as age at onset, age at death, and presence of MND. Studies on the specific pathways underlying these GRN
- associated differences may lead to insights on pathogenesis and possibilities for therapy for an otherwise fatal disease.