Skip secondary menu

Spinal Muscular Atrophy (SMA) Biomarker Qualification Workshop Summary


Spinal Muscular Atrophy, which is associated with a mutation in chromosome 5, is thought to be the second most common autosomal recessive disease in the US. SMA type 1 is one of the most common monogenetic causes of infantile death.   The disease, which manifests with muscle weakness and atrophy, is caused by loss of the survival of motor neuron 1 gene (SMN1) which codes for full-length protein. Advances in preclinical research have uncovered potential drug targets, however clinical development of treatments for SMA has been hampered by multiple factors including: 1) the paucity of natural history studies, 2) the absence of qualified biomarkers.

Biomarkers are defined as cellular, biochemical, and molecular (including genetic and epigenetic) characteristics measurable in biological materials by which normal and/or abnormal processes can be recognized and/or monitored. Biomarkers have multiple potential applications, including but not limited to, the ability to monitor disease progression, the ability to demonstrate response to an investigational agent, and the ability to assist in delineation of study populations based upon predicted response. In addition to the scientific knowledge to be gained by study of SMA-related biomarkers, it is hoped that biomarkers will have future regulatory use. 

The National Institute of Neurological Disorders and Stroke (NINDS) and the Food and Drug Administration (FDA) convened this workshop to discuss the state of the science for putative SMA biomarkers. Registration for the workshop was open; in attendance were academics, industry representatives, NIH/FDA employees and members of the public. The proceedings below summarize the overall content of the workshop's presentations and discussions. 

Introductory Comments

The SMA Biomarker workshop opened with a brief discussion of the NINDS initiative NeuroNEXT ( NeuroNEXT will provide an infrastructure for Phase II clinical trials in neurology and will encourage public-private partnerships with industry, academics and advocacy groups. The planned pilot project for this infrastructure is a SMA biomarker study (

Dr. Stephen Kolb set the stage for the day by providing an overview of SMA. In his remarks, he emphasized the importance of considering the relationship between biomarkers and clinical outcomes as well as the relationship between biomarkers and disease pathogenesis. He challenged the community to consider the possible explanations for changes in biomarkers which were not correlated with clinical improvements and the potential utility for such types of biomarkers.

Drs. Pariser and Walton (slide link) then spoke about regulatory considerations related to biomarker development in general as well as in relation to SMA and other rare diseases. The concept of ‘context-of-use’ was discussed, i.e. a given biomarker may have utility in a specific disease (or disease subset) for a specific purpose. Some of the important components in determination of ‘context-of-use’ include when the biomarker sampling is  done, how the sampling  is done, how the samples were analyzed, and how the data are to be  analyzed and interpreted. FDA now has a biomarker qualification program that will evaluate the utility of biomarkers outside of any specific drug development program.  Once a biomarker has been qualified for a specific ‘context–of-use,’ the biomarker may be used in drug development without the need to re-demonstrate the same utility. Context-of-use is not inflexible since science changes and additional utility may become apparent. It was emphasized that while biomarkers for use as surrogate endpoints such as glycosylated hemoglobin and blood pressure are the type most frequently discussed, other types of biomarkers have definite utility in the drug development process, are more commonly used, and may facilitate and improve drug development at all phases.

There are three general categories of biomarkers (individual biomarkers may have utility in more than one category):

  • Prognostic: Indicates future clinical course of the patient with respect to   a specified clinical outcome  in the absence of any additional therapeutic intervention
  • Predictive: Measured prior to an intervention; identifies patients who are relatively susceptible to a particular drug effect vs. less susceptible patients
  • Pharmacodynamic: response-indicator, post-treatment measurement/marker that reveals whether, or how large, a biological response has occurred in a particular patient; biomarkers suitable for use as a surrogate endpoint will be a small subset of pharmacodynamic biomarkers.

Biomarkers could be used to answer questions related to selection of patient populations, dose selection and regimen, adequacy of drug delivery to the expected site of action, proof-of-concept and pharmacodynamic assessment during early phase research. Biomarkers may give indications of toxicity as well.   Prognostic biomarkers may be used to help stratify study populations and allow for more meaningful comparisons between treatment groups; for example, studies may enroll by genotype/phenotype to determine whether those characteristics are associated with response. Predictive biomarkers may be used to select study populations with expected response; for example, evaluation of an agent in patients with a specific genetically-defined mutation in accordance with the putative mechanism of action.  [The FDA Biomarker Qualification team is available for advice and consultation; information may be found at]

Morning Session: Molecular Biomarkers

Dr F. Danilo Tiziano discussed SMN transcript levels (both isoforms, full length and lacking exon 7) using data from ongoing and recently completed cross-sectional and intervention trials in SMA patients. SMN transcript levels are apparently stable over a 12 month interval. At this time, it appears that SMN full length transcript levels seen in type 1 SMA patients are significantly lower than those seen in Type 2 and 3 individuals.  While there is some overlapping, the SMN full length transcript levels in each group of patients are lower than those found in healthy control individuals. While SMN full length transcript levels are not predictive or prognostic of motor performance as measured by the Modified Hammersmith scale, they are related to the traditional description of “non-sitters”, “sitter/standers” and “walkers “, that is to say the highest level of gross motor function achieved. However, he did note that motor performance, as measured by functional scales, does appear to be related to both number of years elapsed from diagnosis and lean body mass. . While SMN transcript levels are not associated with SMN2 copy number, his published data from an open label trial of salbutamol indicate that the SMN2 copy number may be related to better chance of molecular response to interventions that engage SMN2 as the therapeutic target. He noted that there needs to be an assessment of the correlation between SMN transcript level and clinical outcome in response to intervention treatments. Additionally, the correlation of SMN transcript level in blood with SMN those found in skeletal muscle and in motor neurons has yet to be determined in pre-clinical models of SMA.

Dr W. Ludo van der Pol discussed SMN protein levels as a potential biomarker. He summarized methodology to quantify SMN protein and reported developing an ELISA using recombinant SMN as reference which appeared to be able to distinguish between SMA types. Other assays, i.e. Western Blot and cell immunoassay, showed that SMN protein levels in SMA type 1 patients, but not types 2 or 3, are generally lower than in carriers and healthy controls. Two pilot studies suggest that both Western Blot and ELISA can detect changes in SMN concentrations in leukocytes from carriers and patients during treatment with valproic acid. He noted that there needs to be an assessment of the correlation between SMN copy number, SMN protein and clinical outcome in larger numbers of patients to corroborate these findings. Prospective evaluation of SMN protein in patient-derived cell lines was suggested as a way to evaluate SMN protein as a potential predictive marker. He stated that SMN protein levels as prognostic markers would need to be evaluated in randomized control trials since to date there has been no correlation of changes in SMN levels (as measured in blood) with changes in clinical outcome.

Dr. Dione Kobayashi reported data from the BforSMA study of 130 SMA Type 1, 2 and 3 patients as well as age-matched controls, aged 2-12 years old. This multicenter single-visit prospective study obtained urine and blood biospecimens as well as clinical outcome measure data including the Modified Hammersmith Functional Motor Scale, pulmonary function, and 10 Meter Walk Test. The specimens were analyzed on several platforms, including proteomic, metabolomic, exon array, and SMN copy number, protein and transcripts were also quantified. Novel technologies including a commercially available SMN ELISA kit, as well as an absolute transcript quantification were employed for these analyses. The assays performed in blood and urine found several protein and metabolite analytes which correlated strongly with measured motor function in the study participants. While SMN copy number, protein and transcript levels were altered according to SMA type, these metrics did not correlate robustly to SMA motor function, and generally did not correlate between each other. She suggested that post- transcriptional modifications may play a role in the lack of relationship between SMN RNA and protein levels, although these relationships should be explored in other tissues more pertinent to SMA than blood. Study findings have been submitted for publication, and a panel of immunoassays for the top plasma protein analytes is expected to be complete and available for community use by Fall 2011.

Afternoon session: Electrophysiological and other assessments

Dr Kathryn Swoboda reviewed CMAP and MUNE data accrued from population studies of SMA patients prospectively followed at the University of Utah.  She reviewed both published and unpublished data demonstrating the correlation between gross motor function, SMA type and CMAP and MUNE data. She also reviewed recently published multisite clinical trials data from the Project Cure SMA Investigator’s Network demonstrating a correlation of maximum ulnar CMAP with scores on the MHFMS-Extend. She showed unpublished longitudinal prospectively collected CMAP data from newborns enrolled in the STOP SMA study, who were known to have SMA on the basis of SMN genotype and an older sibling with either SMA type I or type II.  She showed the progressive decline in CMAP and MUNE values, and noted that in her experience electrophysiologic changes often precede and hence predict clinical progression of disease. She postulated that the early decline of CMAP and/or MUNE in conjunction with an SMN genotype associated with 2 copies of SMN2 would have prognostic value for predicting SMA phenotype and severity. However, she opined that the window of opportunity is limited since the greatest motor neuron loss in SMA type 1 patients occurs prior to 6 months; thus, newborn screening would be needed in order for early identification.  In SMA type 2 subjects, the onset and timing of greatest loss of CMAP amplitude and MUNE occurs later, and is most rapid between 12-30 months of age; the large majority of subjects in this group have reached a plateau in motor neuron loss and function by 30 months of age. Thus, in either SMA type I or II subjects, she hypothesizes that CMAP values have potential prognostic value in both demonstrating those in which disease progression is active, as well as in predicting those most likely to have a response to a therapy targeted to increasing SMN levels. Subjects with SMA type 3 have much more variable age of onset and progression and may prove more challenging as a group; further study is needed in this cohort.

Dr Seward Rutkove spoke on electrical impedance myography (EIM) and ultrasound. In comparison to MUNE, EIM is a simpler technique, is faster and can be done in almost any muscle. In ALS, EIM has been shown to correlate with MUNE and survival. While most of the work thus far has been in patients with ALS, he presented preliminary data showing that the technique might distinguish between SMA types.  He also has found evidence that EIM can demonstrate return of muscle quality after traumatic injury/disuse. Studies in ultrasound are more limited. There is evidence that it can distinguish between SMA type 2 and type 3 patients. However, the procedure is operator- and technique- dependent.

Dr Thomas Crawford discussed strength as a biomarker in SMA, noting that it represents a manifestation of the disease that has been used as an outcome measure in other neuromuscular diseases. He has developed an easy measurement tool for upper body strength that has been reliable in single investigator experience. Preliminary studies suggest some decline in this measure over one year time. Further study is required in order to demonstrate reproducibility across centers. Measurement of strength as a clinical tool is a potential disease marker, or, if properly evaluated, could become a qualified outcome measure in itself.  Needed for the latter is much more work to relate values across their range to items of functional import to the disease as it is experienced by patients.

Dr. Douglas Sproule discussed the use of dual-energy X-ray absorption (DEXA) to measure lean mass and MRI to assess muscle mass.  Muscle atrophy is presumed to closely associate with motor function in SMA.  DEXA scanning is feasible across the spectrum of SMA patients but its application requires assumptions regarding water and lipid content of lean tissues, and the proportion of lean tissue comprised by muscle in each limb, that have not as yet, been studied or validated in SMA patients. There are technical limitations related to differences between machines and need for daily calibration of each machine that must be carefully addressed before implementing the method in multi-site trials.  MRI has shown measurable differences in muscle volumes between SMA phenotypes; this modality has the additional potential advantage of allowing qualitative assessment such as fatty infiltration of muscle and measurement of individual muscle group volumes.  There is the potential that these additional details may improve the sensitivity of muscle imaging using MRI to predict clinical course, although available data is too limited at this point to assess the potential utility of this modality to clinical research studies in SMA.  Both methods have the potential advantage of central standardization, and the opportunity to use a single reader blinded to patient state or treatment status.

Concluding Session: Next Steps

To close the workshop, each of the speakers was invited to give a summary of the potential utility of the specific biomarkers which he/she had presented as well as a brief discussion of the perceived gaps in current scientific knowledge.

The biomarkers discussed may have predictive, prognostic, and/or pharmacodynamic utility; while ‘surrogate’ biomarkers are the type most frequently discussed, other types of biomarkers have definite utility in the drug development process and should not be given short shrift. The community will need to determine which of the current candidates are at a stage of readiness for further evaluation. At this point it is essential that the community prioritize the studies that will be needed to evaluate different types of biomarkers.  There are multiple possible markers with potential utility, therefore it is unlikely that a single investigational approach will suffice. It would be beneficial to have studies that would inform aspects of ‘context-of-use’ for biomarkers in SMA; those studies may be concurrent or sequential depending on design.

Last updated March 20, 2013