乌鸦传媒

Machine learning enabled researchers to identify a genetic biomarker for predicting perforated appendicitis in children presenting with suspected appendicitis, according to a single-center prospective exploratory diagnostic study.

Whole-blood transcriptomic analysis in 71 children revealed a four-gene signature that predicted perforated appendicitis over simple appendicitis with 85.7% accuracy (95% CI 72.8-94.1), reported Robert Hancock, PhD, of the University of British Columbia in Vancouver, and colleagues in .

A smaller testing set used for internal validation resulted in 72.7% accuracy, with high sensitivity -- all perforated appendicitis cases were identified as such -- but lower specificity.

Appendicitis is the most common pediatric disease requiring urgent surgery, but its severity can vary greatly, with perforated appendicitis resulting in longer hospital stays and a much higher risk of complications and morbidity, the authors noted. Although surgery is a standard treatment for both simple and perforated appendicitis, "important differences in management can include different antibiotic choices, surgical prioritization, and alternative strategies, including nonoperative approaches," they wrote.

Yet the symptoms of both are similar, and data gleaned from ultrasonography, lab tests, and clinical scoring systems, such as the Alvarado score and Pediatric Appendicitis Score, are often inadequate in distinguishing the conditions sufficiently to administer appropriate, timely treatment.

Other imaging options have their own problems, explained Maurizio Pacilli, MBBS, MD, of Monash University and Monash Children's Hospital in Melbourne, Australia, and Rishikesan Kamaleswaran, PhD, of the Emory University School of Medicine and Georgia Institute of Technology in Atlanta, in an .

CT has more variable accuracy in children compared with adults and involves radiation exposure. MRI has accuracy similar to CT but "is limited by cost, availability, and the need for sedation or general anesthesia in young children," Pacilli and Kamaleswaran wrote.

"Therefore, a test that allows an accurate preoperative diagnosis of acute appendicitis, and ideally differentiation between simple appendicitis and perforated appendicitis, is highly desirable, allowing improved counseling and a tailored approach to individual patient management," they added.

With that goal in mind, the study authors identified a four-gene expression signature in which ANXA3, S100A8, S100A12, and PLBD1 were all significantly upregulated in patients with perforated appendicitis. Prior work from the same group had found that the first three of these were also predictive for sepsis and specifically for the most severe sepsis types.

Although the patients with perforated appendicitis had different proportions of various immune cell types, as well as elevated white blood and neutrophil counts, those differences couldn't fully account for the changes in gene expression identified in these patients. Hancock and colleagues compared these gene expression signatures with their previously published dataset of sepsis signatures. They found that 11 of the 14 patients with perforated appendicitis had signatures consistent with high sepsis severity, while none of the 35 patients with simple appendicitis showed high sepsis severity.

Furthermore, nine of the 14 patients' signatures corresponded to the most severe sepsis phenotype (neutrophilic-suppressive endotype). The researchers concluded that the sepsis signatures of the patients with perforated appendicitis "may be distinct indicators of severity and may also be relevant for diagnosis of severe appendicitis."

Pacilli and Kamaleswaran noted that "the assumption underpinning those findings is that the intricate functional response of innate inflammation may result from cross-talk between other inflammation disorders such as sepsis, which may be similarly reflected in whole blood."

The study authors' findings "could potentially allow clinicians to stratify patients by severity of disease and recognize the onset of sepsis early (ideally, at the time of presentation)," they wrote, which could aid in medical management decisions and "drive evidence-based strategies for antibiotic stewardship."

But they also pointed out that the biomarker's low specificity is "not particularly useful for diagnostic purposes." Furthermore, "the authors have not yet assessed their feasibility as a diagnostic tool in a clinical scenario, in terms of analytical time and costs," they added, and larger studies need to test the genetic signature's accuracy, sensitivity, and specificity. Finally, they noted that the authors "limit their investigations to the inflammatory process, rather than potential confounders that may contribute to these factors, such as obesity."

For this study, Hancock and team used peripheral whole-blood transcriptomics to analyze blood drawn before surgery in 71 children (mean age 11.8, 67.6% boys) who presented to Alberta Children's Hospital's emergency department in Calgary with abdominal pain and suspected appendicitis from November 2016 to April 2017. Blood draws occurred prior to antibiotics in all but five children, whose blood was drawn 5 minutes to 11 hours after receiving antibiotics.

The researchers defined perforated appendicitis as "any perforation of the appendix according to a pathology, surgical, or imaging report (in that order)." Of the 71 participants, 35 had simple appendicitis, 14 had perforated appendicitis, and 22 controls had no confirmed appendicitis but had abdominal pain.

The genetic analysis of children with perforated appendicitis showed a shifted gene expression in the blood compared with those with simple appendicitis or no appendicitis. Those with simple appendicitis and abdominal pain without appendicitis had "virtually indistinguishable" gene expression, Hancock and team wrote, whereas those with perforated appendicitis "had 1,200 to 3,000 significant differentially expressed genes," including "overrepresentation of various immune system and signal transduction, tissue damage or repair, and homeostasis pathways."

Comment