Discussion

Plasma NGS demonstrated limited benefit in our patient population. It is difficult to ascertain which patients would benefit most consistently from NGS, as there was no apparent pattern to the patients who had positive results, nor did the result of NGS change therapy in most cases. However, there does appear to be a lower threshold to change therapy based on NGS in immunocompromised patients. Testing at will with no additional criteria is likely not appropriate and should not replace a thorough history and physical by the clinician. ^{Reference Davis4}

Thirty-seven percent of tests had antimicrobial changes within 2 days, with only a third of these being documented due to NGS. Although the impact of NGS varies significantly in current literature (7% to 56% ^{Reference Morales5} ), our findings appear to be inline with others’ experiences. Determination of NGS impact varies greatly among studies, mostly being based on clinical judgment. NGS uses a shotgun approach for detecting organisms, making it hard to determine commensal organisms versus pathogenic. It is up to the treating physician to determine if therapy should be changed based off of the NGS results and patient presentation. Recent literature has made an attempt to standardize criteria for clinical impact of NGS ^{Reference Hogan, Yang and Garner6} but still relies heavily on clinical judgment.

There was little correlation found between NGS and routine culture concordance or discordance with antimicrobial changes. Fifty percent of antimicrobial changes due to NGS were when we already had positive cultures of these, 3 NGS results were completely discordant with routine testing. These results offer little insight into understanding the number of organisms which appear in the NGS versus regular culture. Presumably, molecular testing is more sensitive; however, with different results in culture and NGS, the question becomes whether the NGS testing is sensitive enough to determine infectious causes. Furthermore, when a possible pathogen is identified it can be hard to direct therapy given no susceptibilities are available. In one case, NGS identified Pseudomonas aeruginosa and the patient was originally started on meropenem; however, once the cultures grew, the susceptibilities showed resistance to meropenem. The most recent panel released by NGS will now provide some antimicrobial resistance data, potentially reducing inappropriate therapy in the future for select organisms.

Clouding the picture further, a negative NGS result does not guarantee the absence of infection. In this review, 17 patients (12%) had pathogens on routine testing which were not identified by the NGS (Table 3), suggesting that NGS testing may not be a suitable method to “rule out” infectious causes. This was not isolated to our review. Similarly, Bergin et al. found that the addition of NGS in immunocompromised patients with pneumonia identified a potential pathogen in 21 more patients compared to usual care but failed to identify similar pathogens compared to usual care in 25 patients. ^{Reference Bergin, Chemaly and Dadwal7} A point to remember is that NGS results may have a number of organisms that will appear in the analytical results, though only the organisms recorded above a proprietary threshold will be reported in the patient result. In cases where the physician is concerned about an organism, a call to the company can lead to a conversation about other organisms present which did not exceed the infection threshold. In addition, the list of organisms that NGS is currently capable of detecting is not exhaustive. One deceased patient had a blood sample taken for NGS testing 2 days prior to death. The autopsy revealed that the immediate cause of death was aortic root rupture secondary to aortic valve replacement with disseminated fungal disease. The Formalin-Fixed Paraffin-Embedded blocks from autopsy were sent for sequencing and revealed that the organism causing the infection was an Exerohilum spp., which this NGS currently does not identify.

Although there are short comings to NGS testing, there are instances in which NGS test resulted in an organism that was not and likely would not have been identified or would have resulted in delayed treatment with current available testing. In one case, NGS testing revealed a Legionella longbeachae which did cause a change in therapy. This organism is not easily culturable and would not have been identified in other routine studies. In another case, the identification of a Mycobacterium abscessus from a pediatric patient with a tooth abscess was detected 4 days prior to AFB gram stain. Torque teno virus, which has no commercial diagnostic test, has popped up on the radar in the clinical laboratory with NGS. It has been associated with liver disease, though it has also been found in healthy individuals. ^{Reference Webb, Rakibuzzaman and Ramamoorthy8}

Although our study attempted to be as objective as possible, by only looking at the antimicrobial changes and discordance between routine tests, there are some limitations to this. First, we are unable to capture any possible potential advantages and disadvantages to using NGS testing. For example, avoidance of and/or unnecessary further diagnostic evaluation prompted by NGS. Second, we only looked at cultures collected within a 7-day window before or after the NGS test. Although this gives us an accurate representation of potential clinical status around the time of NGS, there were likely organisms that were present when NGS was collected that were not identified by routine tests during this time frame. Lastly, this review was limited greatly by chart documentation and ordering of NGS via a downtime form. Documentation surrounding reason for ordering and result utilization are limited, and ordering criteria was loosely followed, creating a more heterogeneous patient population outside of the institutional criteria.

There is no doubt that novel results from case studies are impressive. ^{Reference Caldararo, Algazaq and Schmidt9–Reference Kalyatanda, Rand and Lindner11} However, it is difficult to ascertain when NGS testing should be utilized, and it is unrealistic to utilize on all patients when the culprit of infection is unidentifiable by routine methods due to cost. As NGS becomes prominent and referenced as a potential diagnosis option in guidelines, a consensus-driven guidance would be helpful. ^{Reference Fowler, Durack and Selton-Suty12} However, based on the multiple published multi-center accounts, there does not appear to be a clear-cut population that would benefit from routine NGS testing. ^{Reference Hogan, Yang and Garner6,Reference Bell13–Reference Wilson, Sample and Zorn15} A continuous diagnostic stewardship program is prudent as the cost of NGS is high and does not often provide conclusive diagnostic results.