Blood Biomarkers of Chronic Inflammation in Gulf War Illness

[Kyla]

Blood Biomarkers of Chronic Inflammation in Gulf War Illness

• Gerhard J. Johnson ,
• Billie C. S. Slater,
• Linda A. Leis,
• Thomas S. Rector,
• Ronald R. Bach

• http://dx.doi.org/10.1371/journal.pone.0157855
• Published: June 28, 2016

Abstract
Background
More than twenty years following the end of the 1990–1991 Gulf War it is estimated that approximately 300,000 veterans of this conflict suffer from an unexplained chronic, multi-system disorder known as Gulf War Illness (GWI). The etiology of GWI may be exposure to chemical toxins, but it remains only partially defined, and its case definition is based only on symptoms. Objective criteria for the diagnosis of GWI are urgently needed for diagnosis and therapeutic research.

Objective
This study was designed to determine if blood biomarkers could provide objective criteria to assist diagnosis of GWI.

Design
A surveillance study of 85 Gulf War Veteran volunteers identified from the Department of Veterans Affairs Minnesota Gulf War registry was performed. All subjects were deployed to the Gulf War. Fifty seven subjects had GWI defined by CDC criteria, and 28 did not have symptomatic criteria for a diagnosis of GWI. Statistical analyses were performed on peripheral blood counts and assays of 61 plasma proteins using the Mann-Whitney rank sum test to compare biomarker distributions and stepwise logistic regression to formulate a diagnostic model.

Results
Lymphocyte, monocyte, neutrophil, and platelet counts were higher in GWI subjects. Six serum proteins associated with inflammation were significantly different in GWI subjects. A diagnostic model of three biomarkers—lymphocytes, monocytes, and C reactive protein—had a predicted probability of 90% (CI 76–90%) for diagnosing GWI when the probability of having GWI was above 70%.

Significance
The results of the current study indicate that inflammation is a component of the pathobiology of GWI. Analysis of the data resulted in a model utilizing three readily measurable biomarkers that appears to significantly augment the symptom-based case definition of GWI. These new observations are highly relevant to the diagnosis of GWI, and to therapeutic trials.

 

[Bob]

A diagnostic model of three biomarkers—lymphocytes, monocytes, and C reactive protein—had a predicted probability of 90% (CI 76–90%) for diagnosing GWI when the probability of having GWI was above 70%.

Can anyone work out what they mean by the bolded text? I suppose we'd have to read the full text to find out.

Edit: I'm not sure if I'm any clearer after reading this...

Diagnostic Model
Stepwise multivariable logistic regression led to a diagnostic model comprised of three biomarkers—lymphocytes, monocytes, and CRP. We estimated the probability of having GWI using this fitted model. A positive diagnosis of GWI was defined as a model-estimated probability exceeding 0.70 (70%). For subjects with a model-predicted probability of >70% [a criterion met by 40/80 (50%) of the sample] the positive predictive value of the confirmatory diagnostic model was 90% (95% confidence interval 76–97%) for diagnosing GWI. The corresponding negative predictive value for ruling out GWI was 50% (95% confidence interval 34–66%; Fig 1). The model c-statistic was 0.77 (95% confidence interval 0.67–0.88; p = 0.05).

 

[Bob]

Interestingly, the controls were "deployed Gulf War veterans". The study compared a GWI+ cohort with a CWI- cohort, whereby the GWI+ cohort were patients who had all the symptoms necessary for a GWI diagnosis, and the GWI- cohort did not meet the diagnostic criteria for GWI.

 

[Bob]

Hmm, it's similar to the recent Hanson paper whereby the diagnostic accuracy only applies to the specific participants in the study. The likelihood of these specific biomarkers being able to diagnose the wider patient population seems diminishing small. It's an interesting start, but there's a lot more work to be done.