**6. Summary and conclusions**

*Biomarkers for PTSD.* At this time, there are a number of biomarkers that are associated with PTSD risk, symptoms, and symptom progression. Despite this association, due to the common comorbidity with both other psychiatric conditions and general health status, there is currently little chance of using any single marker as a diagnostic characterization. Future studies must do a more thorough examination of biological and psychological states within PTSD to be able to characterize a combination of biomarkers that may cluster around symptoms and symptom progression in a meaningful way. One way that this may be accomplished is through the use of biomarkers to identify features associated with PTSD, rather than with markers that are consistent with the DSM criteria [166]. For example, it may be that reduced hippocampal volume is associated both with PTSD and comorbid depressive state and can serve as a biomarker of the cluster of symptoms associated with both. This approach would necessitate a panel of biomarkers to increase the specificity, sensitivity, and replicability of any proposed tool. In fact, such an approach utilizing signals from multiple biological domains totaling in excess of one million unique markers was used

to define 343 candidate biomarkers via a combination of data-driven and hypothesis driven approaches. These features were further reduced to 28 based on performance and ability to track phenotype, resulting in a final panel which obtained impressive levels of accuracy, sensitivity, and specificity (81, 85, and 77%, respectively) [167].

*Biomarkers for mTBI.* Currently, there is insufficient evidence to support a relationship between biomarkers of mTBI and clinical outcomes, though many offer promise of acting in this capacity. For this relationship to be drawn, it is imperative that future research includes clinical outcome measures and that a standardized study design is utilized. From the non-exhaustive work cited here, it is clear that differences in methodology, especially related to the timing of sample collection, the length of follow-up, the clinical measurements performed, and the clinical population studied all could be leading to the sometimes-conflicting results reported and the relatively small, unconvincing effect sizes. Further, it is also apparent that although many of the reported biomarkers are sensitive to the presence of head impact, unless the candidate biomarker scales with symptoms reported, it will be of little clinical utility. In fact, there is often little disagreement as to whether an impact to the head has occurred, but rather, the intent of the biomarker is to assess whether that impact is going to result (or is the cause) of symptoms being reported.
