2018-05-16 Meeting notes - Hungtington's TAUG MRS Example

• 5/16/2018: Huntington Examples Team notes
  It seems that the main reason the MRS data are currently represented using the NV domain because the test is performed on live patients. In this scenario, there is no specimen (SPEC) per se since the properties and concentrations of the brain metabolites are analyzed in vivo. Even though the values under NVTEST and NVTESTCD look very similar to what’s traditionally represented by the lab domain, since A) there is no specimen, B) in vivo testing is performed in this case, and C) the LB domain is often used to house data pertaining to in vitro assessments, therefore, the Huntington TA team thinks it is best to use the NV findings domain to represent such data.
  There are several issues with this argument. The lab team states the following:
  1. It is confirmed that the LB domain was not created under the premise that it should only be used to house data pertaining to in vitro tests.
  2. Making the argument that the LB domain should only be used for in vitro study data and the body-system domains are for data pertaining to in vivo testing is dangerous. There are already exceptions to this arugment. For kidney transplant TA, the data regarding to the size, diameter and volume of the transplanted kidney were represented using the UR domain.
  3. The concept Cramer-Rao Lower Bound is now modeled as a NVTEST, the lab team thinks that if the MRS data are modeled in LB, this should be a separate row of record where the --TEST = brain metabolite and the LBTSTDTL can be used to house the value “Cramer-Rao Lower Bound”, since this is technically a further detail or “property” of the testing metabolite. So basically there will be two rows for each metabolite, one row to show the concentration of the metabolite and the 2nd row showing the Cramer-Rao Lower Bound of the metabolite --- in both rows, --TEST = Metabolite, and TSTDTL = concentration (row 1) and Cramer-Rao Lower Bound (row 2).
  4. If the line is drawn that LB is for in vitro data and physiology domains are for in vivo data, in the MRS example, the NVTEST-CD variables are using controlled terminology that’s currently published as LBTEST-CD, this would pose a semantic problem. If the in vitro/in vivo distinction between domains are made, the Huntington team will have to create new terminology for NVTEST-CD variables to reflect that these metabolite assessments are measuring “area under the curve” produced by the MRS in vivo and not actually using a “biological specimen” in vitro (which is defined in every LBTEST-CD value). While changing NVTEST-CD terminology may be the fix here, the lab team thinks that in the end, what’s being measured is the concentration of the metabolites regardless of the methodology, and the form of the output by different devices. (Measuring metabolite concentration in the form of area under the curve using MRS, or chemical testing via microdialysis, you are still just measuring the concentration of an analyte, so the TEST should be the analyte, which is in line with how LBTEST-CDs are created).
  5. Jon made the point that the LB domain does not use the LOC variable, and the metabolite assessments are made at NVLOC = putamen. It seems inappropriate to use LBSPEC instead LOC if the data were to be represented by LB, since there is no specimen in this in vivo brain scan. To this point, even though the --LOC variable is not listed in the LB domain table, it actually is a permissible variable for use in this domain. Jon also mentioned that the data in the NV example are derived data, not direct measurements.

• MRS Huntington's Example 1
• Cramer-Rao was originally listed as separate rows in the dataset but was ultimately decided to be collected via two non-standard variables.
• Jon explains MRS measurement: https://www.semanticscholar.org/paper/Proton-Magnetic-Resonance-Spectroscopy-of-the-Kousi-Tsougos/076f78eb0baca8ea736c2bd1a31bfb6020563014/figure/0
• Resolution of certain analytes may be difficult to distinguish so in some cases you are measuring multiple things. E.g. Glu/Gln=Glu+Gln; tCho = multiple choline-containing compounds.
  • We need to discuss how to call these things to make sure we are clear with our terminology.
• Is the LB domain only to be used for biospecimens that are physically removed from the subject and analysed in a laboratory?
  
  • Phil reports that point of care devices are becoming more used (though still early days) and would still be considered lab tests.
  • Sue reports that interstitial glucose measurements with a probe that is inserted into the interstitium would be considered a lab test, even though the analyte is not removed.
• What is the point of a domain? High Level Modeling principles indicate that a test should be context- and Method- independent suggesting that a test should only be present (in an ideal world) in one domain.
• LOC is available for use in the LB domain.
• The SPEC variable is not in the example. The SMEs indicate that they try to pick a spot that mostly contains grey matter. If we move to LB domain, the SPEC variable is permissible though Jon has indicated that he probably won't add it since it doesn't seem to be of importance. The SPEC would probably map to GREY MATTER TISSUE (if SPEC was chosen to be used).
• Decision: Erin/Nik/Phil/Anna (pending actions below) think this belongs in LB. Ward believes LB is the best way to go.
• The use of analyte testing within body system domains may be something to consider in future.

Actions: The LB domain definitions needs to be updated to clarify the difference between LB domain and body system domains.

Actions: Make clear to GGG that laboratory specimen types are not focused solely on ex vivo type testing. The Lab team does agree that analyte analysis done in vivo (at bedside, POC, etc.) can still be considered for LB domain.

Actions: Jon N to update the example. He will not add SPEC.