Plasma amino acids and inflammatory bowel disease

My recent post on the application of metabolomics to food fingerprinting got me thinking about the examination of biological fluids and how our metabolome might have some interesting secrets to one day share. The 'promise' of the science of metabolomics is that one day, we should be able to look at various biological fluids across a range of conditions and based on the compounds excreted/detected determine diagnosis, disease progression and how well someone responds to intervention. I hasten to add that we are nowhere near that position yet.

In light of this, it was perhaps inevitable ('it is your destiny') that this paper by Hisamatsu and colleagues* (full-text) on possible plasma biomarkers for inflammatory bowel disease (IBD) would get some attention. I'm pretty sure that I don't have to explain this to viewers but the term inflammatory bowel disease covers quite a bit of diagnostic ground including Crohn's disease and ulcerative colitis. Diagnostic confirmation of these conditions is quite a complicated process normally involving a combination of peripheral measures (blood tests, stool analysis) coupled with more direct observation of the bowel.

Hisamatsu and colleagues reported that a previously trialed network analysis of plasma amino acid levels in patients diagnosed with IBD might provide novel, non-invasive and importantly, objective biomarkers potentially opening up some new research areas into the nature of IBDs.

The details:

• Fasting plasma 'aminograms' for a discovery group of 102 Japanese adult patients diagnosed with Crohn's disease (CD) and 102 patients with ulcerative colitis (UC) were initially compared against 102 healthy control participants. The majority of participants were male (70%) with disease duration ranging between an average of 7.8-11 years. A minority of participants were described as having 'active' disease (CD = 29/102; UC = 38/102).  Obtained aminogram results were also validated using a validation set of participants (CD: n=63; UC: n=120; controls: n=108).
• Serum albumin levels were reported as significantly lower in the IBD groups compared to controls.
• Various differences were noted between the groups in terms of amino acid levels. The authors seemed to have concentrated on histidine and tryptophan as primary examples with a view to disease activity and correlation with C-reactive protein levels (a marker for inflammation).
• Implementing their network analysis (MIAI), the authors came up with a formula based on 6 amino acids which discriminated CD and UC groups from controls with ROC values ranging from 0.894 to 0.955 depending on whether the discovery or validation group were used and comparing across the IBDs with controls. 
• Depending on whether disease was active or in remission, a formula incorporating data for 7 amino acids was suggested to have some power in discriminating disease activity with ROC values ranging from 0.894 (CD active vs. CD remission) and 0.849 (UC active vs. UC remission).

The data produced by this study is interesting. OK, it is not totally 'diagnostic' either in terms of classifying IBD from controls or looking at active vs. remissive symptom presentation but it's not a bad start at all. Indeed the use of discovery and training sets takes me back to a wonderful paper covered last year on schizophrenia, which did find a perfect classification based on a handful of serum and urinary markers. It makes me wonder if they were looking at serum and urine at the same time or maybe incorporating a few more well-known markers, whether those ROC values might further approach the magical number 1 (denoting a perfect classification).

The use of statistical models allied to biochemical data is also interesting. I note that similar linkages have been used in other areas of medicine, possibly even to recreate speech from brain activity...?

* Hisamatsu T. et al. 2012 Novel, objective, multivariate biomarkers composed of plasma amino acid profiles for the diagnosis and assessment of inflammatory bowel disease. PLoS ONE. January 2012
DOI: 10.1371/journal.pone.0031131