Study Suggests that Resection of Duodenum Decreases Intestinal Alkaline Phosphatase and Increases Risk of Complications
Duan R, von Ehrlich-Treuenstatt VH, Kakoschke SC, et al. Effect of Surgery on Postoperative Levels of the Gut Homeostasis-Regulating Enzyme Intestinal Alkaline Phosphatase. J Am Coll Surg, 2023, in press.
Editorial: Alverdy JC. Scientific Rationale behind Intestinal Alkaline Phosphatase Administration to Enhance Recovery after Pancreaticoduodenectomy. J Am Coll Surg. 2023, in press.
This study investigated the role of intestinal alkaline phosphatase (IAP) in determining the risk for postoperative complications following intestinal resection. The IAP enzyme is an important regulator of gut barrier function and prevents endotoxemia by detoxifying lipopolysaccharides (LPS). IAP is secreted by intestinal cells in the duodenum.
The authors hypothesized that resection of the duodenum would lead to a decrease in IAP and increase the risk of postoperative complications; blood, stool, and intestinal tissue samples were collected from patients (n = 88) who underwent pancreaticoduodenectomy for pancreatic cancer.
The data analysis showed that IAP decreased significantly after duodenal resection and blood levels of LPS increased. Patients with the largest increases in LPS consistently had longer lengths of hospital stay, a surrogate marker for increased postoperative complications, compared with patients who had small increases in LPS. The authors suggested that supplementation with exogenous IAP could potentially reduce risks of complications after duodenal resection.
In the editorial that accompanied this article, John C. Alverdy, MD, FACS, emphasized that further research is needed to confirm a causal relationship between reduced levels of IAP and postoperative complications. If causality is confirmed, addition of IAP supplements to enhanced recovery protocols could improve intestinal function and reduce risk of complications.
AI-Human Cooperation Show Value for Improving Medical Science, but Investigators Must Provide Consistent Human Input
Hunter DJ, Holmes C. Where Medical Statistics Meets Artificial Intelligence. N Engl J Med. 2023; 389(13): 1211-1219.
This review article has significant value for clinicians but is challenging to read because of the large amount of terminology related to artificial intelligence (AI) and statistical science that may not be familiar for surgeons.
Authors David J. Hunter, MB, and Christopher Holmes, PhD, noted that increases in computer power and data availability have fueled the growth of AI, and the technology can assist—and possibly conduct—medical research. Using AI to analyze datasets and draw conclusions from these analyses has potential value but carries significant risk.
The article emphasized that statistical analysis plays a critical role in medical research; statistical techniques offer an opportunity to draw reasonable conclusions from incomplete information. Successful use of statistics in medical research requires meticulous experimental design, quantification of uncertainty regarding conclusions, and careful evaluation of inferential statements drawn from analyzed data.
The main advantage of AI is the ability of the technology to extract complex, task-oriented features from very large datasets that allow prediction of data characteristics, which makes AI well suited for identifying valuable characteristics of large datasets (for example: images, genomics, and data from electronic health records).
The authors stressed that medical research statisticians use factors that lie outside the dataset including careful experimental design, understanding of the research question, and accounting for bias. Statisticians must also have a healthy suspicion of results that look too good to be true.
For this reason, AI-produced datasets used for medical research must have consistent human input so that methodologic and theoretical gaps in the AI data can be identified and corrected. AI-human cooperation will have tremendous potential value for improving medical science, but realization of this benefit will require consistent use of proven statistical techniques by investigators.
