Microfluidic Sensor for Phagocytosis Quantification from Whole Blood Samples

Immune system activates in response to pathogenic infections in human bodies. One of the critical processes of the inflammatory pathway is the killing of pathogens/ microbes by specific white blood cells, called as Phagocytes. Neutrophils represent majority of phagocytes and recognize the immunoglobulin (IgG) bound bacteria with subsequent binding, internalization and finally killing. Quantifying the efficacy of the phagocytosis process will enable the personalized monitoring of patients’ immune response. This is critical for diagnostics of high-risk patients with infectious diseases in particular sepsis. Sepsis is a global health concern with major clinical diagnostics bottlenecks impacting the patients’ outcomes. In United States, more than a million patients are diagnosed with severe sepsis, out of which more than 30% die. Identifying high risk septic patients is critical to save their lives. To address this unmet need, here, we have developed a microfluidic biosensor to quantify the phagocytic activity of neutrophils using blood samples collected from hospital patients. The biosensor is composed of impedance spectroscopy based electronic sensor with integrated magnetic module. The microfluidic impedance sensor is built by fabricating co-planar gold microelectrodes on glass substrate and bonding it with PDMS based microfluidic channel. The impedance spectroscopy will generate the voltage pulses as each blood cell will flow through the sensing zone over microelectrodes. The sensing zone is coupled with the quadrupole magnetic configuration which allows generating high concentrated magnetic field lines at the start and end of the sensing zone. Further, to mimic the pathogens, IgG coated magnetic particles of similar size were used. As the blood cells will pass through the sensing zone, the magnetic particles internalized by neutrophils will experience a speed differential by magnetic modulation. The resulting electrical pulses will produce signature differences in its profile when a neutrophil with internalized magnetic particles will flow compared to non-phagocytes. We obtained multivariate data from each pulse which included pulse amplitude, width, rise/ fall times of the pulses. Further we developed a machine learning model based on artificial neural network (ANN) and fed each pulse data to classify a phagocytosis event. We collected 17 patient samples from Robert Wood Johnson Medical Hospital and ran on our microfluidic sensor. These samples were equally divided into control (with no phagocytosis) and positive (with phagocytosis) cells. Tens of thousands of blood cells from each sample were ran through our biosensor and collected data was fed to ANN model for classification. Contingency tables and receiver operating curves (ROC) were developed. ANN model was able to predict phagocytosis samples with 88% accuracy and AUC of 0.92. Similarly, lactate level measurements (provided by the hospital) were used to classify high-risk and low-risk patients (threshold of 2mm/L was used). Employing collected electronic data, our ANN model was also able to classify the patients to the appropriate risk groups with 88% accuracy and AUC of 0.85. In conclusion, we have developed microfabricated microfluidic biosensor with impedance spectroscopy and magnetic modulation. Sensor validation was done using real patient samples from the hospital. This sensor can be utilized for diagnostics applications for infectious diseases in different healthcare settings. Reference: C. Norton, U. Hassan , “Bioelectronic Sensor with Magnetic Modulation to Quantify Phagocytic Activity of Blood Cells Employing Machine Learning,” ACS Sensors , 7, 7, 1936–1945, 2022.