I'm really bummed Sidra wasn't able to make this. This was like, this is actually like my favorite talk that one of my mentees was going to give. And it's biomarker based risk stratification and pediatric sepsis from a low middle income country. It was difficult to get her out of Karachi and her hospital at a time of like a multiple, like, you know, COVID and then triple virus waves. And then she had a baby and then like it just, it all just kind of went to pot after the baby as it does. So sepsis is heterogeneous, multiple inciting etiologies. Most of the biomarker work has not occurred in the places where the mortality and the morbidity from sepsis is the worst. And so like Southeast Asia, Sub-Saharan Africa, Oceania, massively over-represented for both sepsis prevalence as well as mortality, but massively underrepresented in mechanistic studies. So one of the more famous biomarker-based risk stratifications that we have, Persevere 2, uses five biomarkers and platelets collected within 24 hours of PICU admission. You know CIDRA made this talk because it uses this god-awful SCCM default background. And within 24 hours of PICU admission, you collect those biomarkers, you have a reasonably good risk stratification tool with areas under the curve of greater than 0.8 in North American PICUs. So we wanted to assess whether or not this would perform in septic children in Pakistan, which is classified in the World Health Organization as a low-middle-income country. We also, since we had it, wanted to assess the utility of additional biomarkers, which may also have added prognostic utility. And so we hypothesized that Persevere 2 would actually discriminate mortality with an area under the curve of at least 0.8. So between, for the last two years, from November 20th to February 22 at Aga Khan, we enrolled septic children with a modified definition of sepsis using a PSOFA score of at least two and a lactate of at least two. And so there were some exclusion criteria of primarily related to size and limitations of care or expected survival. After informed consent, and this was the tricky part of the study, this took about a year to operationalize legally. After informed consent, plasma was collected in situated tubes at Aga Khan and then mailed on dry ice to either CHOP or to CINCI, where Persevere and additional biomarkers were measured. These were the median transit time of seven to 10 days on dry ice. And along the way, they would replenish the dry ice. And this took us a while to work out. And the first time the statements came, we're like totally convinced that it wasn't gonna work. It totally worked. So it absolutely 100% worked with no evident drop off of signal. So we enrolled 86 subjects as per our initial power calculation. The most common site of infection was lung followed by abdomen. The most subjects were bacterially infected. Then it was followed by culture negative and virus and reasonably high PILOD and PSOFA scores. And 88% of them were on vasopressors within 24 hours of PICU admission. 42 of them met PARDS criteria within 96 hours. And most of the time, it was like concurrent. It was within four hours of meeting sepsis criteria. And Persevere 2 applied to the Aga Khan cohort, predict discriminated mortality with an error under the curve of 0.83 and modest sensitivity and specificity with a better negative predictive than positive predictive value. And keeping in mind that the mortality in this cohort was 23%, and the mortality in most of the cohorts where Persevere has been tested to date is closer to like 12%. So in a cohort with double the mortality rate, this still worked pretty well. The low, medium, and high risk strata had the appropriate low, medium, and high risk of mortality that you would have expected. We also measured a few additional biomarkers. And so we measured some specific biomarkers of endotheliopathy and coagulation and lung injury. And we found that there were higher levels of angiopoietin 2, soluble thrombomodulin, and PI-1, or endotheliopathy and coagulopathy, primarily, in the non-survivors. In the subjects meeting ARDS criteria, we found elevated levels of soluble RAGE and surfactant protein D, or markers of epithelial dysfunction. And like surfactant protein D is like pretty good for type 2 epithelial, but RAGE, while it's been postulated that's primarily like, and it is at highest levels in type 1 epithelial cells, it's a fairly ubiquitous molecule. Probably it's in endothelium and like some innate immune cells as well. So for the first time we validated Persevere 2, which is well established in North American PICUs, in a low middle income country, with twice the mortality of comparable cohorts from higher income countries. And this kind of suggests to us that the molecular subphenotyping may actually work. And so a lot of the biomarker based prognostic and predictive enrichment stuff that we're doing in higher income countries may actually translate reasonably well to a lower middle income country. So the advantages of the study was that like it was prospectively enrolled. We like went out to explicitly test a established risk prediction tool. And blood collection was pretty seamless. The biomarkers were measured and analyzed by institutions that have been doing this. Like if you guys can't tell from the last two hours, this is like, yeah. Like watch football and measure biomarkers. That's literally all I do. Okay, and I'm all out of biomarkers today. The downsides of this study are that this was only recruited from a single center in Karachi. And it's a fairly well resourced center in Pakistan. So for a low middle income country. But a lot of low middle income countries are characterized by like great disparities in like having really well resourced as well as like lower resourced. Kind of like America. But the overall sample size was also small. It is however the largest biomarker cohort in pediatric sepsis report to date from Pakistan. And we used established definitions for sepsis and PARDS. There is some risk of misclassification like any of these studies. Because these are syndromic definitions. So we conclude that these biomarker based risk stratification strategies may actually work in children from low middle income countries. And that Persevere 2 seems to work reasonably well in this particular cohort. And that endotheliopathy and coagulopathy biomarkers may further improve this risk stratification. I'm not showing the data here. But if you were to add these biomarkers to Persevere 2, your area under the curve goes from .83 to .91.

biomarker-based risk stratification

pediatric sepsis

low-middle income countries

Persevere 2

mortality rate