Functional Biosensing for Early Detection of Anastomotic Complications: Results of a Phase II Clinical Trial

Anastomotic leak (AL) after colorectal surgery remains a major source of morbidity and mortality, and current detection methods often fail to identify leaks early enough for timely and effective rescue. We propose a drain based sensing device that continuously monitors tissue recovery parameters at the anastomotic site. We hypothesize that monitoring of tissue impedance and intrinsic bowel myoelectric signals will provide high sensitivity and specificity for early leak detection (POD2-3) compared with existing clinical modalities.

We conducted a multicenter clinical trial investigating the sensitivity and specificity of anastomotic leak detection device which monitored local electrophysiological signals from the bowel wall. A standard surgical drain with integrated microelectrodes was placed adjacent to the anastomosis until POD3. Electrophysiological parameters served as a proxy for the physiological processes underlying tissue healing and the development of surgical complications, reflecting ischemia, motility disruption, and local inflammation. The sensor wirelessly transmitted data to a cloud platform for automated analysis. Patient with colorectal anastomosis with no diversion were recruited. Both surgeons and patients were blinded to the results of the detection divice. Care did not deviate from standard of care (SOC).

We enrolled 222 patients across 12 international sites, of which 47.5% were male with an average age of 63.2+12.8 years. The average height of the anastomosis was 13.8+5.8cm from the anal verge. Technical success was 93.7% (95% CI 89.6%-96.5%). We had no device related adverse events. AL occurred in 6.3% (N=14) of which Grade C was 5.4% (N=12) and 2 patients were diagnosed with a grade B leaks. Diversion was performed in 66.7% (N=8) of Grace C leaks. Of the patients with a leak 29% were clinically diagnosed with AL on readmission. The mean overall length of stay (including re-admissions where applicable) was 16.4+11.5 for grade Cs and 7.5+0.7 for grade Bs. By POD3, the device demonstrated overall AL detection sensitivity of 100.0% (95% CI 71.5%-100%) with specificity of 88.3% (95% CI 83.0%-92.5) for Grade C leaks and sensitivity of 84.6% (95% CI 54.6%-98.1%) with specificity of 88.7% (95% CI 83.4%-92.8) for Grade B+C. Diagnostic accuracy was 87.5% (95% CI: 82.2-91.7%). The AUC of 0.97 (95% CI: 0.96-0.99). The mean lead time to diagnosis of AL was 3.9 ± 3.7 days (p < 0.002) compared to clinical team. Re-operative rate for AL was 75% (9 from 12), with time to intervention of 7+3.8 days.

The device showed excellent diagnostic performance, with AUC 0.97 and early detection of anastomotic leaks-identifying AL an average of 3.9 days sooner than the clinical team. By POD3, sensitivity and specificity were high for clinically significant leaks. These findings support the device as a reliable early‑detection tool with potential to improve timely intervention and postoperative outcomes.