Diagnosis, investigation and management of hereditary spastic paraplegias in the era of next-generation sequencing

In northern European populations, autosomal dominant inheritance accounts for the majority of patients from families with pure HSP [15]. Autosomal recessive inheritance accounts for most of the remainder, while clear-cut X-linked inheritance is infrequent. Rarely, gonadal mosaicism for a dominant gene has been described and this can be mistaken for autosomal recessive inheritance [16]. In addition, some patients present without a family history. Potential genetic explanations for this are varied and include singleton autosomal recessive cases, new autosomal dominant mutations, or inherited autosomal dominant mutations with non-penetrance in a parent. The estimated relative frequency of mutations in selected genes associated with pure HSP is given in Table 2. From this table, it can be seen that testing for three genes, SPG4/SPAST, SPG3A/Atlastin1 and SPG31/REEP1, will identify the responsible mutation in approximately 50 % of families with autosomal dominant pure HSP. For SPG4/SPAST, this testing should include methodologies to pick up whole exon deletions, as these are a common mutational class in this subtype of HSP (they have also been described in other forms including SPG3A/Atlastin1 and SPG31/REEP1-HSP) [14, 17‐19]. For families with an autosomal recessive inheritance pattern, screening SPG5 (CYP7B1) will pick up the causative mutations in approximately 5–10 % of cases. However, the yield from testing each individual additional gene beyond these commoner genes is low. Using traditional “one at a time” sequential gene testing to detect these rare mutations can be very expensive and may not be justifiable for publically funded health services. Next-generation sequencing approaches are currently revolutionising this situation (see below).

As mentioned above, the complex HSPs are a diverse set of conditions, each with distinctive clinical features and each caused by one or a small set of genes. Here, we will focus on two of the more common presentations, HSP with thin corpus callosum and HSP accompanied by cerebellar ataxia.

As specific investigations of a patient with suspected HSP depend on their clinical features and inheritance pattern, careful clinical phenotyping of index cases and potentially affected family members is crucial. A detailed medical history should include developmental milestones and a three generational family tree. Affected individuals and ideally, apparently unaffected family members should undergo a clinical neurological examination, to document specific neurological features and identify asymptomatic family members who may have subtle abnormal signs on examination. This is particularly important for apparently sporadic cases or isolated sibships, where positive examination findings in an asymptomatic parent will point towards the probability of autosomal dominant inheritance.

Even in individuals who have typical clinical features of HSP and a positive family history of spastic paraparesis, it is important to consider alternative causes and importantly exclude other treatable familial conditions. In his regard, MR imaging of the brain and spinal cord is crucial and should be undertaken in all sporadic cases and at least one family member of familial presentations. Imaging will not only identify many structural, inflammatory or metabolic abnormalities but is also useful to guide investigations into specific genetic causes (for example, by highlighting atrophy patterns). Note that imaging does have an appreciable false negative rate, for example, it can be normal in certain metabolic or inflammatory conditions, including adrenoleukodystrophy or primary progressive multiple sclerosis.

A baseline metabolic blood screen in the index case could include Vitamins B12 and E, creatine kinase, very long chain fatty acids, white cell enzymes, anti-nuclear antibody and copper levels. However, individuals with more complex clinical features or an unclear family history may require further baseline tests. For example, in presumed sporadic pure HSP, a set of more extensive investigations is appropriate, as an inherited basis for the disease has not been proven. This includes CSF examination (including oligoclonal bands, HTLV serology), Syphilis and HIV serology as well as vasculitis screen [36‐38]. An oral short trial of Levodopa will identify those individuals with dopa-responsive dystonia and neurophysiology can be useful if there is an associated neuropathy or suspicion of amyotrophic lateral sclerosis [39]. Table 3 summarises a list of conditions to be considered in the differential diagnosis and investigation of sporadic pure HSP.

For those individuals in whom the above baseline investigations are unremarkable and/or where the family history and clinical features point towards a specific genetic diagnosis, it is then appropriate to test for the suspected mutation.

There is considerable clinical utility for HSP families in detecting their causative mutation. It gives diagnostic certainty, prevents further unnecessary investigations and opens the possibility of predictive and prenatal testing. However, until very recently, in many healthcare systems, economic factors necessitated that genetic investigations were restricted to analysis of HSP genes in which there was the highest chance of detecting a mutation, as the expense of analysing rarely-causative genes was not justified by the low mutation detection rate per additional gene analysed. However, this has changed with the introduction of next-generation sequencing, which allows cost-effective analysis of many genes together, using approaches including (1) sequencing of large panels of genes, (2) sequencing of all coding exons in the genome, or (3) sequencing of entire genomes. At present, a common approach uses next-generation sequencing to screen exons of panels of genes causally associated with specific conditions or groups of conditions. These new approaches hold the prospect of substantially increasing the mutational pick-up rate for HSP patients. Once fully integrated into molecular diagnostic services, it is possible to envisage that genetic testing could be introduced at an early stage in patient work-up, perhaps avoiding the need for other costly or painful investigations if a definitive molecular genetic diagnosis is made.

Although these developments are undeniably positive, there are pitfalls and caveats associated with next-generation sequencing approaches. Sequence variation is very common in the genome (for example, many millions of single nucleotide variants have been identified) [40], with the vast majority of these sequence changes having a neutral effect with no clinical consequence. In addition, many of these sequence changes are individually rare. Thus, next-generation approaches typically detect numerous rare sequence variants; the more genes that are analysed, the larger this problem becomes. A major problem lies in deciding which, if any, of the detected sequence changes are pathogenic. In most cases, this can be resolved using bioinformatics and other approaches (e.g. identifying previously reported and validated mutations, excluding known rare population polymorphisms, determining the likely effect on the amino acid sequence of the encoded protein and how deleterious this effect is likely to be, excluding mutations that do not co-segregate with disease) [41]. However, inevitably in some cases, a single conclusively pathogenic mutation cannot be identified and one or more candidate pathogenic mutations remain. This is a particular issue with mutations of the missense class (in which only a single amino acid of the encoded protein is altered), since it can be difficult to predict whether the resulting effect on the protein will have functional consequences. Such sequence changes are often reported as variants of uncertain significance. Defining approaches to determining which of these mutations are pathogenic will be a major future challenge, as will be determining ethical approaches to whether and how the uncertainty associated with such sequence changes is reported back to patients.

In addition, exon-sequencing approaches do not detect all mutational classes. A particular problem arises in detecting large-scale deletions or duplications involving whole exons, or in detecting promoter or deep intronic mutations. This is an issue for at least some HSP genes, e.g. whole exon deletions are a common cause of SPG4/SPAST HSP (see above). We anticipate that in the relatively near future this issue will be resolved by whole-genome sequencing, which has the capacity to identify large deletions/duplications, as well as promoter and deep intronic mutations. In a research setting, genome sequencing may also allow identification of causative mutations in genes that have not previously been associated with the disease under consideration. In the meantime, additional testing for deletions/duplications should be considered in selected genes for appropriate patients, prioritised based on clinical features, in whom mutations have not been identified in exon-sequencing panels.

High-throughput sequencing approaches also may present ethical issues if so-called “incidental findings” (clinically relevant changes in genes unrelated to the condition being tested) are detected. These are particularly important when whole-exome or whole-genome approaches are taken, where, given sufficient numbers of tested subjects, it is inevitable that other potentially clinically significant abnormalities, e.g. mutations in cancer predisposing genes or detection of carrier status for autosomal recessive disease, will be detected. Whether and how such findings are reported back to patients requires careful consideration of their clinical validity and utility, and the ethical issues surrounding this are currently a topic of much debate [42].

Finally, while next-generation sequencing approaches have the potential to increase rates of molecular genetic diagnosis in HSP, they do not remove the need for careful phenotyping––this is still important, as it can help to focus bioinformatics analysis onto the most relevant genes. Careful phenotyping may also better define the clinical spectrum associated with pathogenic mutations in particular genes.