How quickly amyotrophic lateral sclerosis progresses may depend on the immune system, according to a study published March 16 in Acta Neuropathologica Communications. Taking an unbiased look at data garnered from the circulating immune cells of ALS patients, researchers led by Pamela Shaw and Winston Hide at the University of Sheffield in England found that the expression of three immune genes in the blood predicted the pace of deterioration. They claim this troika could serve as a useful prognostic biomarker. The researchers also reported that cerebrospinal fluid (CSF) levels of soluble TREM2—an indicator of immune cell activity in the brain—shot up early in disease, and declined later on, much as it does in people with Alzheimer’s.

Together, the findings implicate the immune response to diseased motor neurons in the rate of the neurons’ demise, commented Stanley Appel of Houston Methodist Neurological Institute. What’s more, the authors provide support for a disease-specific inflammatory response outside of the central nervous system, he added.

ALS progression varies widely, even among patients who harbor the same genetic mutations. While some imaging, electrophysiological, and even fluid biomarkers track or predict progression, it is unclear if they are specific to ALS (see Dec 2013 news; Jun 2015 newsMar 2017 news). This dearth of specific, predictive markers makes accurate prognosis difficult, and also complicates patient selection for, and analysis of, clinical trials.

Modular Mining. Starting by linking transcripts to disease pathology (a), researchers assembled modules of genes (b) that correlated with disease phenotypes (c) and ultimately identified blood and CSF expression markers of ALS progression (d). [Courtesy of Acta Neuropathologica Communications, 2017.]

To find better prognostic markers, first author Johnathan Cooper-Knock and colleagues took an unbiased, data-driven approach. In a nutshell, the researchers first looked for motor neuron gene expression patterns that correlated with ALS pathology in a small number of patients. Then, they used those genes to identify wider gene networks, or modules, whose activities correlated with disease progression in a larger patient sample set. Finally, they tested the most promising genes for biomarker potential, by measuring their expression in the blood lymphocytes of ALS patients and controls and in CSF (see image above).

To generate their initial list of genes, the researchers used laser capture microscopy to pluck at least 800 individual motor neurons from each of 11 postmortem spinal tissue samples, and then quantified their entire transcriptomes. Four patients had sporadic ALS, and seven were C9ORF72 hexanucleotide expansion carriers. In parallel, the researchers counted the number of p62 inclusions in nearby neurons from each sample. Correlating the transcriptomes with the protein pathology, they identified 83 transcripts that seemed specifically up- or down-regulated in ALS. Casting a wider net, Cooper-Knock used these 83 transcripts to create a larger network of genes known to be co-expressed. The researchers then subdivided that network into 82 modules of tightly co-expressed genes. Each module—essentially a clique of transcripts that share the same expression pattern—contained anywhere from 35 to 515 transcripts, and 45 modules contained at least one of the original 83 pathology-related transcripts.

To find the genes most tightly associated with the rate of ALS decline, the researchers next cross-referenced these 45 modules to three independent data sets. The first included 1,705 transcripts, from laser-captured motor neurons, which correlated with disease duration in 14 ALS patients, including the 11 patients used in the original p62 correlation study. The second comprised 4,070 transcripts, from blood-derived lymphoblastoid cell lines, that were differentially expressed between eight C9ORF72 carriers with slowly progressing disease and 18 carriers who deteriorated rapidly. Those who progressed rapidly had a disease duration of less than two years, while slowly progressing carriers survived longer than four years following diagnosis. Finally, the researchers checked if any of the modules contained any of the 62 hits from prior ALS genome-wide association studies. They then ranked the 45 modules in order of how enriched they were with genes from the three data sets. The two most enriched modules contained genes from all three. The top one was chock-full of immune genes, while the runner-up contained genes related to synaptic transmission.

The researchers focused on the immune module, reasoning that it had the best chance of containing genes that could serve as biomarkers in accessible tissues, including blood. Oddly enough for a module generated from laser capture microdissection of motor neurons, more than half of its 65 members, including TREM2, are known macrophage and microglial genes. The authors think that by comingling with damaged motor neurons, microglia may have hitched a ride during the laser capture. Despite this accidental windfall, Cooper-Knock still views the results as valid. He added that these hitchhikers are likely to be the microglia most intimately responding to diseased neurons.

To test for potential biomarkers, the researchers correlated expression of the module genes in patient lymphoblastoid cell lines with disease duration. Sure enough, expression levels of 15 of the genes distinguished 18 C9ORF72 carriers with rapidly progressing ALS from eight with a more slowly developing disease. Expression of 20 genes distinguished 10 rapidly progressing sporadic ALS patients from 10 people with slowly progressing disease. There were three genes in common—LILRA2, ITGB2, and CEBPD. All are expressed in macrophage/microglial cells, and involved in stimulating inflammatory immune responses. LILRA2 is a cell surface receptor; ITGB2 an integrin involved in leukocyte adhesion and activation; and CEBPD is a transcription factor. Considering expression levels of all three at once correctly classified the speed of decline of 85 percent of C9ORF72 carriers, and 60 percent of people with sporadic disease.

Cooper-Knock speculated that the lower predictive power of the biomarkers for sporadic disease likely stems from the greater heterogeneity in those cases, and that better specificity and sensitivity would be needed in the clinic. However, he added that because of the small sample size, the numbers are not conclusive and that he plans to test the blood biomarkers in larger numbers of ALS patients. He added that studying the function of these genes, and determining how they are involved in pathology, could help the field better understand ALS as well. Researchers agreed that if these results hold up in further testing, the markers could prove useful in disease prognosis, clinical trial selection, and tracking of response to treatments.

“Overall, this study represents a first and important non-biased step in searching for disease markers in ALS that can be measured in accessible tissues, and thus can be used to predict disease progression, and eventually enable personalized medicine,” commented Michal Schwartz of the Weizmann Institute of Science in Rehovot, Israel.

The findings dovetail with a recent report from Appel’s lab that dysfunction of regulatory T cells correlated with more rapid disease progression (see Beers et al., 2017). Appel said activation of circulating monocytes does as well. One possible explanation for peripheral immune cell involvement in ALS is that motor neurons are not completely walled off behind the blood-brain barrier, Appel said. Rather, at neuromuscular junctions, their axons come into direct contact with the systemic immune system.

Soluble TREM2 a CSF Marker
TREM2 popped up in a screen for proteins known to interact with those translated from genes in the immune module. Given that genetic variants in TREM2 have been linked to ALS, and the soluble extracellular domain of the protein to neurodegeneration, the researchers tested if CSF sTREM2 correlated with ALS progression. The protein ticked higher in CSF samples from 46 sporadic ALS patients, compared to 20 controls. Furthermore, sTREM2 appeared to rise early in disease, and fall in the late stages, similar to what has been reported for AD (see Mar 2016 news; Dec 2016 news). Interestingly, superimposed on this inverted U pattern, higher sTREM2 levels correlated with slower deterioration in later disease, hinting at a potential protective effect of the protein.

“This interesting study highlights the importance of a TREM2-dependent, potentially protective, microglia function in the pathogenesis of ALS,” wrote Christian Haass and Marc Súarez-Calvet of the German Center for Neurodegenerative Diseases in Munich in a joint comment to Alzforum. “This also supports a general function of TREM2 and microglia as modifiers for several distinct neurodegenerative diseases.” They added that longitudinal studies in larger cohorts will be needed to confirm these effects.

Laura Piccio of Washington University in St. Louis, who has studied CSF sTREM2 in AD and multiple sclerosis, thought the parallels between sTREM2 expression in different neurodegenerative diseases was intriguing (see Jan 2016 newsPiccio et al., 2008). However, she urged caution in interpreting the results, pointing out that the difference between sTREM2 levels in early and late ALS disease appears to be driven by a single outlier.

Cooper-Knock acknowledged the need to confirm results in larger numbers of patients. He proposed that high levels of sTREM2 early in disease could reflect a neuroprotective, phagocytic microglial response. Differing genetics could then dictate how rapidly that beneficial response sours, transforming into a damaging, pro-inflammatory response as sTREM2 drops. However, he and other commentators added that the function of sTREM2 is far from clear. For example, a recent study found that soluble TREM2 triggers pro-inflammatory responses in microglia (see Feb 2017 news).—Jessica Shugart


  1. This paper by Cooper-Knock and colleagues describes an extremely important study in understanding ALS, and in searching for prognostic biomarkers of ALS progression. Overall, the work emphasizes that ALS, not only in animal models but also in patients, is not a cell (motor neuron) or tissue (CNS) autonomous disease, and involves the local microglia, and the systemic immune system. In addition, this is the first study to use a robust and non-biased genomic approach to search for biomarkers that correlate with disease severity/progression, despite the heterogeneity of the disease.

    The authors, by using a systematic and non-biased approach, identified gene clusters, modules which change in the diseased motor neurons during disease progression and that correlated with a separate module that characterizes the immune response in ALS patients. Subsequently, they found that the immunological module was specifically associated with changes in microglia. Among the most robust changes in microglia that were found to correlate with disease severity were expression of Trem2, LILRA2, ITGB2, and CEBPD. Finally, they found that TREM2 expression is reflected in the CSF, and the other three markers, in the circulating immune cells.

    Overall, this study represents a first and important non-biased step in searching for disease markers in ALS that can be measured in accessible tissues, and thus can be used to predict disease progression, and eventually enable personalized medicine. The finding that sTREM2 increased in the CSF at early disease stages is a sign of disease onset, and its rapid elevation thereafter is a prognostic sign of slow disease progression, which suggests its potential beneficial effect in coping with the disease. This is consistent with the reported positive role of TREM2 in Alzheimer’s disease. These results encourage the use of Cooper-Knock's approach to search for similar biomarkers in other neurodegenerative diseases whose etiology and course are still poorly understood, as it may lead not only to identification of additional biomarkers, but also provide therapeutic targets. 

  2. Overall this is an interesting article, and Cooper-Knock and colleagues use a powerful system biology approach to identify potential molecules that could serve as biomarkers for disease progression in ALS. However, I would be very cautious in the interpretation of some of the results.

    In the first part of the study, it is surprising that from transcriptome analysis of spinal cord motor neurons and of lymphoblastoid cell lines (immortalized B cells) from ALS patients they find a set of genes that are related to microglia and macrophages. This is a little bit puzzling to me. They explain this by saying that in the laser capture procedure they may have taken some microglial cells along with the motor neurons. I am not sure this is a really good explanation. In addition, they also used transcriptome data from lymphoblastoid cell lines, which are not completely comparable to microglia/macrophages and they do not express TREM2.

    The researchers go on to measure levels of soluble TREM2 (sTREM2) in the cerebrospinal fluid (CSF) of a cohort of ALS patients and control individuals with non-inflammatory neurological disease. Interestingly, levels of sTREM2 are elevated in the CSF of patients with ALS compared to controls. This is very likely reflecting microglia activation in ALS patients as a response to the ongoing neurodegenerative process. This also recapitulates what was observed in patients with Alzheimer’s disease (AD) by our and other groups. Next, the authors show that CSF sTREM2 levels are higher in early versus late stages in ALS (Figure 4b in the paper). I would interpret these results with caution because the difference between early versus late ALS seems to be driven by the presence of an outlier in the group of early ALS. These results would need to be confirmed in a larger group of patients. It is interesting that these data are paralleling a model proposed in AD by other groups with higher levels of CSF sTREM2 in early AD cases and a decline at later stages of the disease (Suarez-Calvet 2016 and 2016). In my opinion, these data also require further confirmation in larger and possibly longitudinal data sets from AD patients.

    Finally, Cooper-Knock and colleagues found that levels of sTREM2 are positively correlated with disease duration in a subset of patients with late ALS, as if microglia activation could have a neuroprotective effect and slow down disease progression. The proposed model of a “U-shaped curve” for sTREM2 CSF levels (high in early stages and then levels that gradually decline at later stages at a possible different pace) would need to be confirmed in additional studies with increased number of subjects.


    . Early changes in CSF sTREM2 in dominantly inherited Alzheimer's disease occur after amyloid deposition and neuronal injury. Sci Transl Med. 2016 Dec 14;8(369):369ra178. PubMed.

    . sTREM2 cerebrospinal fluid levels are a potential biomarker for microglia activity in early-stage Alzheimer's disease and associate with neuronal injury markers. EMBO Mol Med. 2016 May 2;8(5):466-76. PubMed.

  3. Laura Piccio makes some good points regarding our work and we thank her for her comments. However, I would like to clarify a point regarding an outlier in our CSF study. Dr. Piccio draws attention to the outlier in the measurement of CSF soluble TREM2 in early ALS; this was noted at the time of analysis and its effect was removed by using a rank-based analysis rather than absolute values of measured soluble TREM2. The statistically significant difference quoted in the manuscript was derived in this way without a significant effect from the outlier sample.

    It is agreed that our proposed biomarkers require confirmation in larger sample groups, however, we believe the central tenant of our paper is robust—genes expressed in proportion to motor neuron pathology identify a gene set that is associated with microglial function, and has prognostic potential.

  4. The methods and models used in this paper are very impressive. However, at the moment a clear-cut diagnostic use of these biomarker candidates is difficult to see. Possibly follow-up investigations of these candidates will prove a clinical application.

  5. The interesting study of Cooper-Knock et al. highlights the importance of a TREM2-dependent, potentially protective, microglia function in the pathogenesis of ALS. This also supports a general function of TREM2 and microglia as modifiers for several distinct neurodegenerative diseases.

    Strikingly, changes in the levels of CSF sTREM2 follow a similar pattern to that we had observed in sporadic Alzheimer's disease and within the DIAN cohort, that is, a slight, but significant, increase in early stages of the disease.

    Moreover, this paper also shows that in later stages of ALS, higher levels of CSF sTREM2 positively correlate with longer disease duration. Thus, TREM2 may indeed have a protective function as suggested by the loss-of-function character of its disease-associated mutations. Nevertheless, further work in larger and specifically in longitudinal cohorts will be required to finally prove the hypothesis that levels of CSF sTREM2 predict a better or worse outcome not only in ALS but also in AD and other neurodegenerative diseases.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Biomarker Panel Predicts Slow or Fast ALS Course
  2. Blood Marker May Predict ALS Progression
  3. Pee P75: A Potential Biomarker of ALS Progression?
  4. Microglial Marker TREM2 Rises in Early Alzheimer’s and on Western Diet
  5. Paper Alert: Slotting TREM2 into Alzheimer’s Pathogenesis
  6. TREM2 Goes Up in Spinal Fluid in Early Alzheimer’s
  7. Does Soluble TREM2 Rile Up Microglia?

Paper Citations

  1. . ALS patients' regulatory T lymphocytes are dysfunctional, and correlate with disease progression rate and severity. JCI Insight. 2017 Mar 9;2(5):e89530. PubMed.
  2. . Identification of soluble TREM-2 in the cerebrospinal fluid and its association with multiple sclerosis and CNS inflammation. Brain. 2008 Nov;131(Pt 11):3081-91. Epub 2008 Sep 12 PubMed.

Further Reading


  1. . Evaluating the levels of CSF and serum factors in ALS. Brain Behav. 2017 Mar;7(3):e00637. Epub 2017 Feb 19 PubMed.

Primary Papers

  1. . A data-driven approach links microglia to pathology and prognosis in amyotrophic lateral sclerosis. Acta Neuropathol Commun. 2017 Mar 16;5(1):23. PubMed.