A method for detecting microbial contamination of cells and gene therapy products
By combining microfluidic chips and Raman spectroscopy with optical tweezers, rapid and accurate detection of microbial contamination in cell and gene therapy products has been achieved, solving the problems of long detection time and inaccurate detection in existing technologies, and providing a rapid and accurate detection method.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QINGDAO INST OF BIOENERGY & BIOPROCESS TECH CHINESE ACADEMY OF SCI
- Filing Date
- 2026-01-29
- Publication Date
- 2026-06-19
Smart Images
Figure CN122238296A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of microbial detection technology, and in particular relates to a method for detecting microbial contamination in cell and gene therapy products. Background Technology
[0002] Cell and gene therapy products are biological products that utilize live cells derived from humans or animals, which, after in vitro manipulation (such as isolation, culture, expansion, and gene modification), are reinfused or transplanted into patients for disease treatment, tissue repair, or functional regulation. These products primarily include immunotherapy (such as CAR-T), stem cell therapy, and gene-modified cell therapy. These products are highly personalized, possess live cell characteristics, and involve complex manufacturing processes, making them one of the most promising cutting-edge treatment methods in the current biomedical field.
[0003] However, the safety of these products is extremely critical, and the detection of microbial contaminants is a crucial step in ensuring their safety. Currently, for the detection of microorganisms in cell and gene therapy products, pharmacopoeia methods rely on traditional culture techniques, which take at least 14 days, becoming one of the key technical challenges restricting the industry's development. Other detection methods still face problems such as missed detections, false detections, long processing times, and insufficient detection limits due to interference from matrix components, the presence of slow-growing bacteria, and unknown microorganisms.
[0004] Therefore, developing rapid detection technologies for exogenous microbial contamination in cell and gene therapy products is crucial for ensuring product safety. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the technical problem to be solved by this invention is that current methods for detecting microbial contamination in cell and gene therapy products suffer from problems such as missed detections, false detections, long processing times, and insufficient detection limits. This invention proposes a method for detecting microbial contamination in cell and gene therapy products that is time-efficient and highly accurate.
[0006] To solve the aforementioned technical problem, the technical solution adopted by the present invention is as follows: This invention provides a method for detecting microbial contamination in cell and gene therapy products, comprising the following steps: S1: Using microfluidic chips to achieve the separation and enrichment of microorganisms; S2: Raman spectral data acquisition is performed using a 532nm laser combined with optical tweezers technology to obtain raw Raman spectral data; S3: Perform spectral quality control and preprocessing on the Raman spectral data obtained in step S2, and import the preprocessed Raman spectral data into the classification model for training to obtain a microbial and cell debris identification model.
[0007] Preferably, the cell and gene therapy product is an immune cell therapy product, and its components include T cells, T cell fragments, and microorganisms.
[0008] Preferably, in step S3, the spectral quality control includes: selecting a 2936 cm⁻¹ spectral depth. -1 Peak intensity is between 0-4000 and 3300 cm⁻¹ -1 Spectra with peak intensities ranging from -200 to 1000; abnormal spectra were eliminated through cosmic ray correction.
[0009] Preferably, in step S3, the pretreatment includes: cutting 400-3010cm... -1 The spectral range is defined; Gaussian smoothing is applied with a σ value of 1; baseline correction is performed using asymmetric reweighted penalized least squares; and the spectrum is normalized using maximum-minimum normalization.
[0010] Preferably, in step S3, the classification model is a Kan+LSTM hybrid prediction model.
[0011] Preferably, the microfluidic chip includes a substrate layer, a cover layer, and a fluid channel layer, wherein the cover layer is disposed on the side of the fluid channel layer away from the substrate layer, and the cover layer, the fluid channel layer, and the substrate layer are connected together.
[0012] Preferably, the cover layer is made of glass, and the substrate layer is made of ITO glass.
[0013] Preferably, the fluid channel layer is made of a perforated thin film, and the perforated portion of the perforated thin film forms a microfluidic channel.
[0014] Preferably, the fluid channel layer includes: A square annular channel, wherein a first sample inlet is provided at one end of the square annular channel; The first channel is configured as a straight line, and the first channel is connected to the end of the square annular channel away from the first inlet. A second inlet is also provided at the connection point, and the second inlet is located on the center line of the first channel. The first inlet, the second inlet, and the first channel are located in the same horizontal direction. The first inlet and the second inlet introduce different samples. The first channel is connected to a sample outlet channel at the end away from the square annular channel. The sample outlet channel includes a capture branch, a first waste liquid branch, and a second waste liquid branch. The inlet of the capture branch is on the same horizontal line as the first channel. The inlets of the first waste liquid branch and the second waste liquid branch are perpendicularly arranged on both sides of the inlet of the capture branch. The flow resistance ratio of the capture branch, the first waste liquid branch, and the second waste liquid branch is 5:1:1. The width of the first channel is 2.5 mm, the width of the capture branch inlet is 150 μm, and the width of the capture branch in the middle is 1 cm.
[0015] Preferably, the substrate layer has through holes that are respectively connected to the first inlet, the second inlet, the outlet of the capture branch, the outlet of the first waste liquid branch, and the outlet of the second waste liquid branch for fluid transport. An array of sharp-angled interdigitated electrodes is etched on the substrate layer below the first channel, and an array of vertical electrodes is etched on the substrate layer below the capture branch.
[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for detecting microbial contamination in cell and gene therapy products. It utilizes a microfluidic chip to enrich microorganisms, adjusting flow resistance and increasing the width of the capture branch to achieve high-speed enrichment at the front end and low-speed capture at the back end. However, because cell debris and microorganisms in the sample are highly similar in size and dielectric properties, cell debris is also enriched along with microorganisms during the enrichment process. Therefore, the Raman spectral fingerprint property is used to model the Raman spectra of cell debris and bacteria, enabling the differentiation of bacteria and cell debris through Raman spectroscopy. This eliminates the interference of cell debris on bacterial detection, eliminates the need for time-consuming culture steps, and completes separation in just 30 minutes from sample injection. Furthermore, it is universally applicable to various types of microbial contaminants, including Gram-positive / negative bacteria and fungi, avoiding the problems of microbial activity loss and abundance distortion during culture. This provides a universal platform for the rapid preprocessing of clinical and environmental samples. Attached Figure Description
[0017] Figure 1 This is a schematic diagram of the microfluidic chip provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the fluid channel layer provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of cell capture in the detection method provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the accuracy and recall of the classification model provided in this embodiment of the invention for identifying bacteria and cell debris. Explanation of reference numerals in the attached figures: 100. Fluid channel layer; 101. Square annular channel; 102. First injection port; 103. First channel; 104. Second injection port; 105. Capture branch; 106. First waste liquid branch; 107. Second waste liquid branch; 200. Cover sheet layer; 300. Substrate layer; 301. Sharp-angled interdigitated electrode array; 302. Vertical interdigitated electrode array. Detailed Implementation
[0018] The technical solutions in specific embodiments of the present invention will now be described in detail and completely with reference to the accompanying drawings. Obviously, the described embodiments are only some specific implementations of the overall technical solution of the present invention, and not all implementations. Based on the overall concept of the present invention, all other embodiments obtained by those skilled in the art fall within the protection scope of the present invention.
[0019] This invention provides a method for detecting microbial contamination in cell and gene therapy products, comprising the following steps: S1: Using microfluidic chips to achieve the separation and enrichment of microorganisms; S2: Raman spectral data acquisition is performed using a 532nm laser combined with optical tweezers technology to obtain raw Raman spectral data; S3: Perform spectral quality control and preprocessing on the Raman spectral data obtained in step S2, and import the preprocessed Raman spectral data into the classification model for training to obtain a microbial and cell debris identification model.
[0020] It should be noted that, currently, the core technical challenge in the field of microbial detection for cell and gene therapy products lies in how to achieve rapid enrichment of microorganisms from complex samples. Current pharmacopoeia methods rely on traditional culture techniques, which are time-consuming (at least 14 days) and inefficient. Current rapid detection methods, such as nucleic acid amplification and bioluminescence methods, are severely affected by the sample's cell matrix, leading to inaccurate detection. Solid-phase imaging methods use filtration to remove the cell matrix for bacterial enrichment, but this suffers from filter clogging and insufficient detection limits.
[0021] To address the aforementioned issues, microbial enrichment was first achieved using microfluidic chips. However, since cell debris and microorganisms in the sample are highly similar in size and dielectric properties, cell debris is also enriched along with microorganisms during the microbial enrichment process. This debris interference is a serious obstacle to the accurate detection of cell therapy products. Therefore, the fingerprint properties of Raman spectroscopy are utilized to perform spectral modeling on the Raman spectra of cell debris and bacteria, enabling the differentiation between bacteria and cell debris through Raman spectroscopy.
[0022] In one embodiment, the cell and gene therapy product is an immune cell therapy product, the components of which include T cells, T cell fragments and microorganisms.
[0023] In one embodiment, step S3, spectral quality control includes: selecting a 2936 cm⁻¹ spectral depth. -1 Peak intensity is between 0-4000 and 3300 cm⁻¹ -1 Spectra with peak intensities ranging from -200 to 1000; abnormal spectra were eliminated through cosmic ray correction.
[0024] In one embodiment, step S3, the preprocessing includes: cutting 400-3010cm... -1 The spectral range is defined; Gaussian smoothing is applied with a σ value of 1; baseline correction is performed using asymmetric reweighted penalized least squares; and the spectrum is normalized using maximum-minimum normalization.
[0025] Through spectral quality control and preprocessing, the purity, stability, and consistency of spectral data were significantly improved. Interference factors such as noise, baseline drift, and systematic errors were effectively removed, and the characteristic information of the target components was fully preserved, laying a solid foundation for subsequent efficient and accurate qualitative and quantitative analysis.
[0026] In one embodiment, in step S3, the classification model is a Kan+LSTM hybrid prediction model.
[0027] In one embodiment, the microfluidic chip includes a substrate layer, a cover layer, and a fluid channel layer, with the cover layer disposed on the side of the fluid channel layer away from the substrate layer, and the cover layer, the fluid channel layer, and the substrate layer connected together.
[0028] In one embodiment, the cover layer is made of glass, and the substrate layer is made of ITO glass.
[0029] In one embodiment, the fluid channel layer is made of a perforated film, and the perforated portion in the perforated film forms a microfluidic channel.
[0030] In one embodiment, the fluid channel layer includes: A square annular channel, with a first sample inlet at one end; The first channel is designed as a straight line and is connected to the end of the square annular channel away from the first inlet. A second inlet is also provided at the connection point, and the second inlet is located on the center line of the first channel. The first inlet, the second inlet, and the first channel are located in the same horizontal direction. The first inlet and the second inlet introduce different samples. The end of the first channel away from the square annular channel is connected to the sample outlet channel. The sample outlet channel includes a capture branch, a first waste liquid branch, and a second waste liquid branch. The inlet of the capture branch is on the same horizontal line as the first channel. The inlets of the first waste liquid branch and the second waste liquid branch are perpendicularly arranged on both sides of the inlet of the capture branch. The flow resistance ratio of the capture branch, the first waste liquid branch, and the second waste liquid branch is 5:1:1. The width of the first channel is 2.5 mm, the width of the capture branch inlet is 150 μm, and the width of the capture branch in the middle is 1 cm.
[0031] By adopting the above technical solution, the 150μm capture branch inlet can, on the one hand, utilize the covering effect of the buffer phase to prevent cells not dielectrically captured from entering the capture branch; on the other hand, the three bifurcations at the outlet of the first channel serve to divert the flow, allowing a small amount of fluid enriched with bacteria and debris to enter the capture branch. This low flow rate and widened capture branch channel achieve a low flow velocity, thus enabling stable dielectric capture. The inlet size is determined based on a 5:1:1 flow resistance ratio and channel width calculation. The width in the middle of the capture branch is set to 1cm. The higher the width, the lower the flow velocity and the smaller the fluid force. Cells are subjected to dielectric force and the opposite fluid force, making them easier to capture. In addition, based on the principle of squeeze flow separation, bacteria and debris, dielectrically enriched, enter the 1cm channel from the 150μm channel. Bacteria and larger debris move forward under the action of fluid inertial force, while very small, low-density debris is deflected. At both ends of the flow channel, a large number of small debris are further removed.
[0032] In one embodiment, the substrate layer has through holes that are respectively connected to the first inlet, the second inlet, the outlet of the capture branch, the outlet of the first waste liquid branch, and the outlet of the second waste liquid branch for fluid transport. An array of pointed interdigital electrodes is etched on the substrate layer below the first channel, and a vertical electrode array is etched on the substrate layer below the capture branch.
[0033] By adopting the above technical solution, the pointed interdigitated electrode array pointing to the capture branch below the first channel, by adjusting the dielectric parameters, makes the dielectric force generated by it greater than the hydrodynamic force of the target particles, thereby realizing the directional screening and guidance of target particles from complex mixed systems, greatly improving the targeting of particles entering the capture branch, and effectively reducing the interference of non-target particles; on the other hand, the vertical interdigitated electrode array below the capture branch can form a uniform and stable confinement electric field. When the target particles enter the capture branch, this electric field can provide a continuous dielectric binding force, reducing the escape risk caused by fluid disturbance, ensuring that the particles are efficiently enriched in the branch and are not easily detached from the capture area.
[0034] To provide a clearer and more detailed description of the method for detecting microbial contamination in cell and gene therapy products provided by the embodiments of the present invention, specific embodiments will be described below.
[0035] Example 1: Method for detecting microbial contamination in cell and gene therapy products Step 1: Use microfluidic chips to separate and enrich microorganisms; 1.1 The structure of the microfluidic chip is as follows: like Figure 1 , 2 The microfluidic chip comprises a fluid channel layer 100, a cover layer 200, and a substrate layer 300. The fluid channel layer 100 is fabricated using a perforated thin film double-sided adhesive, with the perforated portions forming microfluidic channels. The cover layer 200 is made of quartz glass, and the substrate layer 300 is made of ITO quartz glass. The cover layer 200 and the substrate layer 300 are disposed on opposite sides of the fluid channel layer 100, and the three are bonded together using double-sided adhesive of the same material as the substrate layer 300.
[0036] The fluid channel layer 100 includes a square annular channel 101, a first channel 103, and a sample outlet channel. One end of the annular channel 101 is provided with a first sample inlet 102; The first channel 103 is configured as a straight line, which is connected to the end of the square annular channel 101 away from the first injection port 102, and a second injection port 104 is provided at the connection point. The second injection port 104 is located on the center line of the first channel 103. The first injection port 102, the second injection port 104 and the first channel 103 are in the same horizontal direction, and are used to introduce different samples; The sample outlet channel includes a capture branch 105, a first waste liquid branch 106, and a second waste liquid branch 107. The inlet of the capture branch 105 is at the same horizontal line as the first channel 103. The inlets of the first waste liquid branch 106 and the second waste liquid branch 107 are perpendicularly arranged on both sides of the inlet of the capture branch 105. The outlets of the capture branch 105, the first waste liquid branch 106, and the second waste liquid branch 107 are arranged in parallel.
[0037] In this embodiment, the flow resistance ratio of the capture branch 105, the first waste liquid branch 106, and the second waste liquid branch 107 is 5:1:1, the width of the first channel 103 is 2.5 mm, the width of the inlet of the capture branch 105 is 150 μm, and the width of the middle of the capture branch 105 is 1 cm.
[0038] The substrate layer 300 has through-holes that connect to the first sample inlet 102, the second sample inlet 104, the outlet of the capture branch 105, the outlet of the first waste liquid branch 106, and the outlet of the second waste liquid branch 107, respectively, for fluid transport. An electrode array is etched on the substrate layer 300, comprising a pointed interdigitated electrode array 301 and a vertical interdigitated electrode array 302. The pointed interdigitated electrode array 301 is disposed below the first channel 103 and points towards the capture branch 105; The vertical interdigitated electrode array 302 is disposed below the capture branch 105.
[0039] 1.2 The specific operational steps for separating and enriching microorganisms in T-cell therapy product samples using microfluidic chips are as follows: The T-cell therapy product sample was first centrifuged at 8000 rpm for 5 min and resuspended three times, then resuspended in 10 ml of loading buffer (DEP buffer). Using a syringe, the T-cell therapy product sample was injected through the first inlet, and the buffer solution through the second inlet. The flow rate ratio of the T-cell therapy product sample to the buffer solution was 400:100 μL / min, with the buffer solution injected first to prevent the T-cell therapy product sample from directly entering the capture branch. The dielectric parameters of the first channel electrode were adjusted to 1 MHz, 20 Vpp, and a duty cycle of 5 Hz, 98.5%. At these flow rates and dielectric parameters, bacteria entered the buffer phase from the sample phase, arranging themselves at the tips of the pointed interdigitated electrode array, and then entered the capture branch with the fluid. The dielectric parameters of the capture branch were set to 1 MHz, 20 Vpp, and normally open. Due to the reduced flow rate, the width of the capture branch channel increased, the fluid velocity entering the capture branch decreased, and the bacteria were captured at the electrode edges. The above steps involve precisely configuring the structure of the fluid channel layer. Within the first channel, the liquids from the two inlets are in a laminar flow state. The two phases of fluid from the first inlet flow to the waste liquid branches on both sides, while the fluid from the second inlet is confined in the middle and covers the inlet of the capture branch, acting as a physical barrier to prevent the sample from the first inlet from entering the capture branch. This achieves path isolation of the liquids introduced from the two inlets, thereby enabling the separation and purification of bacteria in the sample. By adjusting the dielectric parameters, T cells in the T-cell therapy product experience a much greater fluid force than dielectric force, thus flowing with the sample phase fluid to the outlets of the waste liquid branches on both sides. Meanwhile, microorganisms and bacteria in the cell and gene therapy product are enriched in the buffer phase because the dielectric force they experience is still greater than the fluid force, thereby achieving the separation of microorganisms and cells.
[0040] Step 2: Raman spectral data acquisition was performed using a 532nm laser combined with optical tweezers to obtain the raw Raman spectral data: After 30 minutes of sample loading, the sample loading was stopped, the dielectric signal of the first channel was turned off, the dielectric of the capture branch was kept open, and the flow rate of the capture branch was reduced to 0 after 1 minute. The dielectric parameter was adjusted to 2Vpp to capture bacteria and cell debris on the electrode. Image recognition technology was used to accurately identify and locate the particles on the electrode. Raman spectroscopy data was acquired using a 532 laser combined with optical tweezers. The energy was set to 100mw and the acquisition time was 1s. Raman acquisition was performed on the particles in the capture branch. Step 3: Perform spectral quality control and preprocessing on the Raman spectral data obtained in step S2: The Raman spectral data obtained in step S2 are subjected to spectral quality control and preprocessing. Spectral quality control includes: ① Selecting a 2936 cm⁻¹ spectral depth. -1Peak intensity is between 0-4000 and 3300 cm⁻¹ -1 ① Spectra with peak intensities between -200 and 1000; ② Cosmic ray correction (removed as anomalies); Pretreatment includes: ① Cutting 400-3010cm -1 ① Spectral range; ② Gaussian smoothing with σ = 1; ③ Baseline correction using asymmetric weighted least squares method; ④ Normalization of the spectrum using maximum-minimum normalization. Step 4: Import the preprocessed Raman spectroscopy data into the classification model for training to obtain a model for identifying microorganisms and cell debris. Using the above process, we first collected modeling data. A total of 5224 T-cell fragment spectra and 4357 bacterial spectra were collected using pure samples for modeling (according to an 8:2 ratio of training set to test set). The model was built using Kan+LSTM (Recurrent Time Network) developed by Raman AI.
[0041] Predicting 1534 cell debris spectra and 1854 bacterial spectra (bacterial types not included in the training set), the results are as follows: the accuracy for both bacteria and cell debris is above 95%. Figure 4 ).
Claims
1. A method for detecting microbial contamination in cell and gene therapy products, characterized in that, Includes the following steps: S1: Using microfluidic chips to achieve the separation and enrichment of microorganisms; S2: Raman spectral data acquisition is performed using a 532nm laser combined with optical tweezers technology to obtain raw Raman spectral data; S3: Perform spectral quality control and preprocessing on the Raman spectral data obtained in step S2, and import the preprocessed Raman spectral data into the classification model for training to obtain a microbial and cell debris identification model.
2. The method for detecting microbial contamination in cell and gene therapy products according to claim 1, characterized in that, The cell and gene therapy product is an immune cell therapy product, and its components include T cells, T cell fragments, and microorganisms.
3. The method for detecting microbial contamination in cell and gene therapy products according to claim 2, characterized in that, In step S3, the spectral quality control includes: selecting a 2936 cm⁻¹ spectral depth. -1 Peak intensity is between 0-4000 and 3300 cm⁻¹ -1 Spectra with peak intensities ranging from -200 to 1000; abnormal spectra were eliminated through cosmic ray correction.
4. The method for detecting microbial contamination in cell and gene therapy products according to claim 2, characterized in that, In step S3, the preprocessing includes: cutting 400-3010cm... -1 The spectral range is defined; Gaussian smoothing is applied with a σ value of 1; baseline correction is performed using asymmetric reweighted penalized least squares; and the spectrum is normalized using maximum-minimum normalization.
5. The method for detecting microbial contamination in cell and gene therapy products according to claim 2, characterized in that, In step S3, the classification model is a Kan+LSTM hybrid prediction model.
6. The method for detecting microbial contamination in cell and gene therapy products according to claim 2, characterized in that, The microfluidic chip includes a substrate layer, a cover layer, and a fluid channel layer. The cover layer is disposed on the side of the fluid channel layer away from the substrate layer, and the cover layer, the fluid channel layer, and the substrate layer are connected together.
7. The method for detecting microbial contamination in cell and gene therapy products according to claim 6, characterized in that, The cover layer is made of glass, and the substrate layer is made of ITO glass.
8. The method for detecting microbial contamination in cell and gene therapy products according to claim 6, characterized in that, The fluid channel layer is made of a perforated thin film, and the perforated portion of the perforated thin film forms a microfluidic channel.
9. The method for detecting microbial contamination in cell and gene therapy products according to claim 8, characterized in that, The fluid channel layer includes: A square annular channel, wherein a first sample inlet is provided at one end of the square annular channel; The first channel is configured as a straight line, and the first channel is connected to the end of the square annular channel away from the first inlet. A second inlet is also provided at the connection point, and the second inlet is located on the center line of the first channel. The first inlet, the second inlet, and the first channel are located in the same horizontal direction. The first inlet and the second inlet introduce different samples. The first channel is connected to a sample outlet channel at the end away from the square annular channel. The sample outlet channel includes a capture branch, a first waste liquid branch, and a second waste liquid branch. The inlet of the capture branch is on the same horizontal line as the first channel. The inlets of the first waste liquid branch and the second waste liquid branch are perpendicularly arranged on both sides of the inlet of the capture branch. The flow resistance ratio of the capture branch, the first waste liquid branch, and the second waste liquid branch is 5:1:
1. The width of the first channel is 2.5 mm, the width of the capture branch inlet is 150 μm, and the width of the capture branch in the middle is 1 cm.
10. The method for detecting microbial contamination in cell and gene therapy products according to claim 9, characterized in that, The substrate layer has through holes that are respectively connected to the first inlet, the second inlet, the outlet of the capture branch, the outlet of the first waste liquid branch, and the outlet of the second waste liquid branch for fluid transport. An array of sharp-angled interdigitated electrodes is etched on the substrate layer below the first channel, and an array of vertical electrodes is etched on the substrate layer below the capture branch.