Protocol Assessing Neuroinflammation in Gulf War Illness Using Magnetic Resonance Spectroscopy, Younger

[forestglip]

Assessing Neuroinflammation in GWI Using MRS

Sponsor: University of Alabama at Birmingham
Principal Investigator: Jarred Younger

Study Description
This proposal is for a Tier 1 (Discovery) study that uses a human, cross-sectional, observational neuroimaging approach to measure neuroinflammation in Gulf War Illness (GWI). A whole-brain magnetic resonance spectroscopic imaging (MRSI) scan will be used to detect multiple markers of neuroinflammation in 30 individuals with GWI and 30 healthy veteran controls. The investigators have fully tested the MRSI scan in several patient groups, and have found strong evidence of neuroinflammation in fibromyalgia and chronic fatigue syndrome - conditions the investigators believe involve abnormal central immune system processing. The investigators have not, however, performed the scan on individuals with GWI. The discovery study will allow investigators to transfer this existing technology to the GWI field. This project meets the core Special Interest of investigating dysregulation between the immune and neurological systems in the brain.

The investigators' central hypothesis is that GWI involves chronic neuroinflammation. The symptoms of GWI (e.g. fatigue, musculoskeletal pain, sleep disturbances, and cognitive dysfunction) overlap heavily with classic cytokine-induced sickness responses. In the case of GWI, microglial cells in the brain can be pushed into a hypersensitized state by toxins or abnormal immune challenges, leading to chronic overproduction of pro-inflammatory factors that result in the primary symptoms of GWI.

To test the central hypothesis, it is necessary to measure neuroinflammation in humans in vivo. However, most techniques are too invasive for use in living individuals. To address that problem, the investigators use an MRSI scan which provides metabolite concentrations in 4,000 separate voxels, giving whole-brain coverage. The scan yields measurements for: Myo-inositol (a marker of glial cell proliferation), lactate (a product of anaerobic metabolism), choline (a sign of cellular breakdown), and N-acetylaspartate (a marker of neuronal health). The scan also provides absolute brain temperature, which is shown to be elevated with severe neuroinflammation. The five main outcomes are as follows:

• Myo-inositol (MI): Higher values represent greater microglia proliferation or gliosis.
• Lactate (Lac): Higher values represent more severe inflammatory activity.
• Choline (Cho): Higher values indicate greater cell turnover (inflammation, gliosis, or demyelination).
• NAA: Lower values represent neurodegeneration.
• Temperature: Higher values represent greater neuroinflammation. Metabolite and water-reference data will be analyzed using the MIDAS package and thermometry extension41. Absolute brain temperatures (in °C) within each voxel will be calculated by measuring the distance of the temperature-invariant NAA peak from the temperature-variant water peak using the formula: Tbrain = -102.76 × Δwater-NAA + 310.5°C, which has been validated in our scanner to produce reliable readings. Temperature will be expressed on absolute values in °C.

ClinicalTrials.gov

 

[forestglip]

They've updated the registration to show the summary statistics for the outcomes. I did a Welch's test using the provided means, standard deviations, and sample sizes. Probably won't be as accurate as the tests they'll run using the raw data but should be in the ball park of the right significance:

Myo-inositol - lower in GWI
p = 0.412
BH-corrected p = 0.412

Lactate - higher in GWI
p = 0.106
BH-corrected p = 0.133

Choline - lower in GWI
p = 0.023
BH-corrected p = 0.114

N-acetylaspartate - lower in GWI
p = 0.085
BH-corrected p = 0.133

Brain temperature - higher in GWI
p = 0.059
BH-corrected p = 0.133

Here's the code I used to get these:

#######################
# mean1, std1, nobs1 is healthy controls
# mean2, std2, nobs2 is GWI
#######################

import scipy.stats as stats
ps = []

#####################

# Myo-inositol

mean1 = 0.001260
std1 = 0.0002362
nobs1 = 19

mean2 = 0.001205
std2 = 0.0001699
nobs2 = 20

res = stats.ttest_ind_from_stats(mean1, std1, nobs1, mean2, std2, nobs2, equal_var=False, alternative='two-sided')
ps.append(res[1])
# Ttest_indResult(statistic=np.float64(0.8310812993307201), pvalue=np.float64(0.4119762327054679))

#####################

# Lactate

mean1 = 0.0003886
std1 = 0.0001629
nobs1 = 19

mean2 = 0.0005388
std2 = 0.000365
nobs2 = 20

res = stats.ttest_ind_from_stats(mean1, std1, nobs1, mean2, std2, nobs2, equal_var=False, alternative='two-sided')
ps.append(res[1])
# Ttest_indResult(statistic=np.float64(-1.6732425917986282), pvalue=np.float64(0.10601481462343196))

#####################

# Choline

mean1 = 0.0003838
std1 = 0.000051446
nobs1 = 19

mean2 = 0.0003499
std2 = 0.000034982
nobs2 = 20

res = stats.ttest_ind_from_stats(mean1, std1, nobs1, mean2, std2, nobs2, equal_var=False, alternative='two-sided')
ps.append(res[1])
# Ttest_indResult(statistic=np.float64(2.394181605769833), pvalue=np.float64(0.022783882258803012))

#####################

# N-acetylaspartate

mean1 = 0.0023191
std1 = 0.0001821
nobs1 = 19

mean2 = 0.0022186
std2 = 0.0001721
nobs2 = 20

res = stats.ttest_ind_from_stats(mean1, std1, nobs1, mean2, std2, nobs2, equal_var=False, alternative='two-sided')
ps.append(res[1])
# Ttest_indResult(statistic=np.float64(1.7693757298362183), pvalue=np.float64(0.0851713988293998))

#####################

# Temperature

mean1 = 36.622
std1 = 0.322
nobs1 = 19

mean2 = 36.826
std2 = 0.331
nobs2 = 20

res = stats.ttest_ind_from_stats(mean1, std1, nobs1, mean2, std2, nobs2, equal_var=False, alternative='two-sided')
ps.append(res[1])
# Ttest_indResult(statistic=np.float64(-1.9508269010446142), pvalue=np.float64(0.05868683728861156))

#####################

stats.false_discovery_control(ps)
# array([0.41197623, 0.13251852, 0.11391941, 0.13251852, 0.13251852])

Edit: Added multiple test correction.