Alves G, Bonanni L.
Neurofilament light: A heavyweight diagnostic biomarker in neurodegenerative parkinsonism?.
Neurology. 2017 Mar 7;88(10):922-923. Epub 2017 Feb 8
PubMed.
It is an interesting study, describing the first identification of a biochemical marker in blood for a neurological disorder (other than autoimmune or genetic disorders), with a sufficiently high diagnostic accuracy (sensitivity and specificity of blood NfL each more than 80 percent). This is a major step forward in the eventual application of such biomarkers in clinical practice. The nice thing about this study is the major effort in collecting three different cohorts from two different countries. Besides, the findings remain significant in a cohort with a relatively short disease duration. However, it should be noted that both sensitivity and specificity drop to lower levels in this cohort (80 percent or lower), which would argue against the use of this test in diagnosis of early cases (because of too low a diagnostic accuracy).
However, we have to keep in mind that—as in many similar diagnostic studies—these patients groups are probably biased toward a selection with a clear diagnosis that was made after a number of years of follow-up. In future studies it should be investigated what the prognostic value is for these biomarkers at a time when the clinical diagnosis is still very uncertain. And, how much will this test add to the diagnostic work-up (clinical and neurological investigations, response to medication, MRI) that will be done anyway in such patients; in other words, does the implementation of the analysis of NfL in blood change the decisions of the treating clinician? These are questions that should be answered in follow-up studies.
A more technical aspect of the study is that quantification of NfL in blood now relies on the availability of specific equipment (Simoa, Quanterix). More widespread implementation of this biomarker assay will be possible when this biomarker test can be developed using other, more frequently used platforms. For the moment, the use of this test will be limited to specialty labs. Moreover, it would be nice to see if the same results can be replicated by others, thereby demonstrating the robustness of the assay across different laboratories.
On a slightly different topic, it is interesting to see that the PD groups within the London and Lund cohorts (cohorts nos. 1 and 2 in the study) not only differ in disease severity (by 0.6 points on the H&Y score) and four years in disease duration, but most strikingly in the levels of cerebrospinal fluid NfL, which, in the Lund group, are exactly the same in the PD and control groups, but in the London cohort are three times higher (and close to the levels in the atypical parkinsonism groups) than in controls. This makes you wonder, if CSF and blood levels are so well correlated, why in the London group the blood NfL levels are still relatively comparable to the controls. And what might be the reason for this large difference in the two PD groups? It would be interesting to further delineate the characteristics of the PD groups in these two cohorts.
I fully agree that further work is needed, such as evaluating whether the test adds to the diagnostic work-up of uncertain cases of parkinsonism, especially in a setting where the patients are examined by medical doctors who are not specialists in movement disorders. I also agree that it will take time to establish a method of measuring NfL in blood that can be used in a clinical routine setting, and I believe that technical platforms other than the Simoa could also be used in this context. However, the differences between the Lund and the London cohorts when it comes to CSF NfL levels should be interpreted with caution. In the London cohort the NfL levels in CSF (not blood) were only available in five cases with PD (as written in the manuscript, e.g. in the legend of Figure 1 and in the legend of Table 2). Of these five patients with PD, four had completely normal CSF NfL levels (in the same range as the controls), but there was one outlier with a value of more than 7,000 ng/L (which resulted in a huge effect on the mean value of this parameter presented in Table 2). This is the reason we write in the result section about CSF NfL in the London cohort that: "The number of PD cases with available CSF NfL measurements (n = 5) was too low to conduct statistical analysis."
Comments
Radboud University Nijmgen Medical Centre
It is an interesting study, describing the first identification of a biochemical marker in blood for a neurological disorder (other than autoimmune or genetic disorders), with a sufficiently high diagnostic accuracy (sensitivity and specificity of blood NfL each more than 80 percent). This is a major step forward in the eventual application of such biomarkers in clinical practice. The nice thing about this study is the major effort in collecting three different cohorts from two different countries. Besides, the findings remain significant in a cohort with a relatively short disease duration. However, it should be noted that both sensitivity and specificity drop to lower levels in this cohort (80 percent or lower), which would argue against the use of this test in diagnosis of early cases (because of too low a diagnostic accuracy).
However, we have to keep in mind that—as in many similar diagnostic studies—these patients groups are probably biased toward a selection with a clear diagnosis that was made after a number of years of follow-up. In future studies it should be investigated what the prognostic value is for these biomarkers at a time when the clinical diagnosis is still very uncertain. And, how much will this test add to the diagnostic work-up (clinical and neurological investigations, response to medication, MRI) that will be done anyway in such patients; in other words, does the implementation of the analysis of NfL in blood change the decisions of the treating clinician? These are questions that should be answered in follow-up studies.
A more technical aspect of the study is that quantification of NfL in blood now relies on the availability of specific equipment (Simoa, Quanterix). More widespread implementation of this biomarker assay will be possible when this biomarker test can be developed using other, more frequently used platforms. For the moment, the use of this test will be limited to specialty labs. Moreover, it would be nice to see if the same results can be replicated by others, thereby demonstrating the robustness of the assay across different laboratories.
On a slightly different topic, it is interesting to see that the PD groups within the London and Lund cohorts (cohorts nos. 1 and 2 in the study) not only differ in disease severity (by 0.6 points on the H&Y score) and four years in disease duration, but most strikingly in the levels of cerebrospinal fluid NfL, which, in the Lund group, are exactly the same in the PD and control groups, but in the London cohort are three times higher (and close to the levels in the atypical parkinsonism groups) than in controls. This makes you wonder, if CSF and blood levels are so well correlated, why in the London group the blood NfL levels are still relatively comparable to the controls. And what might be the reason for this large difference in the two PD groups? It would be interesting to further delineate the characteristics of the PD groups in these two cohorts.
View all comments by Marcel VerbeekLund University
I fully agree that further work is needed, such as evaluating whether the test adds to the diagnostic work-up of uncertain cases of parkinsonism, especially in a setting where the patients are examined by medical doctors who are not specialists in movement disorders. I also agree that it will take time to establish a method of measuring NfL in blood that can be used in a clinical routine setting, and I believe that technical platforms other than the Simoa could also be used in this context. However, the differences between the Lund and the London cohorts when it comes to CSF NfL levels should be interpreted with caution. In the London cohort the NfL levels in CSF (not blood) were only available in five cases with PD (as written in the manuscript, e.g. in the legend of Figure 1 and in the legend of Table 2). Of these five patients with PD, four had completely normal CSF NfL levels (in the same range as the controls), but there was one outlier with a value of more than 7,000 ng/L (which resulted in a huge effect on the mean value of this parameter presented in Table 2). This is the reason we write in the result section about CSF NfL in the London cohort that: "The number of PD cases with available CSF NfL measurements (n = 5) was too low to conduct statistical analysis."
View all comments by Oskar HanssonMake a Comment
To make a comment you must login or register.