Advances in Concentration Gradient Generation Approaches in a Microfluidic Device for Toxicity Analysis

This systematic review aimed to analyze the development and functionality of microfluidic concentration gradient generators (CGGs) for toxicological evaluation of different biological organisms. We searched articles using the keywords: concentration gradient generator, toxicity, and microfluidic device. Only 33 of the 352 articles found were included and examined regarding the fabrication of the microdevices, the characteristics of the CGG, the biological model, and the desired results. The main fabrication method was soft lithography, using polydimethylsiloxane (PDMS) material (91%) and SU-8 as the mold (58.3%). New technologies were applied to minimize shear and bubble problems, reduce costs, and accelerate prototyping. The Christmas tree CGG design and its variations were the most reported in the studies, as well as the convective method of generation (61%). Biological models included bacteria and nematodes for antibiotic screening, microalgae for pollutant toxicity, tumor and normal cells for, primarily, chemotherapy screening, and Zebrafish embryos for drug and metal developmental toxicity. The toxic effects of each concentration generated were evaluated mostly with imaging and microscopy techniques. This study showed an advantage of CGGs over other techniques and their applicability for several biological models. Even with soft lithography, PDMS, and Christmas tree being more popular in their respective categories, current studies aim to apply new technologies and intricate architectures to improve testing effectiveness and reduce common microfluidics problems, allowing for high applicability of toxicity tests in different medical and environmental models.

New technologies analysis showed innovations in their fabrication or materials, such as the development of facilitators in relation to the main reported problems in microfluidics, such as shear and bubbles (27%) [16,17,21,27,35,37,38,42,44], cost reduction, and rapid prototyping (12%) [18,28,36,48], and integration of other systems (9%) [19,20,24], such as electrodes, for example. Another 24% [22,23,29,31,32,34,41,47] presented precise technologies for the analysis of the organism in question, enabling customized development. A total of 27% of articles did not report new technologies [25,26,30,33,39,43,45,46]. Finally, regarding the dimensions of microfluidic devices used in the selected studies, mainly in the culture chamber and channel parts, the measurements were very particular for the purpose of the study; the larger chambers used a higher concentration of microorganisms in cultures or 3D culture. The devices made for the microalgae model had measurements with less variation, and, for the zebrafish embryos model, the chamber height was much shorter than for the others. According to the design and the structures of the microdevice developed by each author, some of the CGG's particularities are exemplified in Figure 2. Schematic diagram of microfluidic devices with CGG system for toxicological analysis, representing some of the studies selected in this systematic review. (A) Representation of device layers, gradient generator structure, details of fluid mixing units and air bubble valves. Adapted with permission from [44], Biosensors and Bioelectronics. (B) Project showing a physical map of the CGG system and the photomicrograph of Pyramimonas sp. and Chlorella. Adapted with permission from [23], Sensors (Switzerland). (C) A schematic showing the design of a µFSCD with a concentration gradient generator. It exposed the structures, dimensions, and characteristics of the two layers, adapted with permission from [34], Molecules. (D) Construction of the Sphero Chip system proving the measurement principle of the experimental scheme and results of the computational modeling of a CGG structure. Adapted with permission from [41], Lab on a Chip. (E) The microdevice contains eight sets of C-Chambers, which can simultaneously enable eight sets of noninterfering ASTs with each other. Antibiotics can be preincorporated into the C chambers with a specific mass gradient. AST and MIC results can be obtained by comparing the fluorescence intensities between each set of C-Chambers. Adapted with permission from [16], Biosensors and Bioelectronics. (F) CGG microdevice used for toxicity tests based on marine phytoplankton motility containing four units connected to a central removable outlet. Shown is the enlarged image of the single-frame unit containing an upstream CGG and downstream diffusible cameras. Motility signals can be collected in real time. Adapted with permission from [25], Marine Pollution Bulletin. (G) Schematic design of the CGG microfluidic chip with cell chambers (top panel) and the chip manufactured with pumping machine (bottom panel). Chamber-diffused Rh-123 (green) and morphological characteristics of A549 cells with or without CAF matrix are shown. Adapted with permission from [37], PLoS ONE. (H) CGG containing four parallel operational modules including inputs CSE: 18 parallel cell chambers and 6 cell inputs. A CSE concentration gradient is shown from entry one to six, adapted with permission from [40], Journal of Thoracic Oncology. (I) Schematic overview of the microfluidic device with a CGG and chambers with passive hydrodynamic cell trap arrays. It shows details of branching and diffusional mixing of two fluorescent fluids with different concentrations and optical micrograph of cell traps in PDMS. Adapted with permission from [42], Lab on a Chip. (J) Schematic drawing of the CGG device, illustrating cross-section and theoretical profiles of Ciprofloxacin concentration in the observation channel. Antibiotic solutions with 3× MIC (blue curve) or 6× MIC (orange curve). Adapted with permission from [17], Frontiers in Microbiology. Abbreviations: CGG: concentration gradient generator; µFSCD: microfluidic spheroid culture device; ASTs: antibiotic susceptibility tests; Rh−123: Rhodamine; A549: adenocarcinoma human alveolar basal epithelial cell line; CAF: cancer-associated fibroblasts; CSE: cigarette smoke extracts; PDMS: polydimethylsiloxane; MIC: minimal inhibitory concentration.

Concentration Gradient Generator Characteristics of Microfluidic Device
For the development of microfluidic devices capable of generating concentration gradients, it is necessary to establish general and functional characteristics, such as, as analyzed in Table 2, the methods and types of systems used to generate gradients, number of concentrations generated, linearity of the gradient, variation in concentrations of the compounds evaluated, and time to achieve gradient stability, as well as the duration of stability, in addition to information on simulation methods, validation, and advantages. Most studies used the Christmas tree gradient generation system (Figure 2A-D,F,G,I) [21][22][23][24][25][26][27]29,30,[34][35][36][37][38][39][42][43][44][45][46][47][48], which uses the convective method, either associated or not associated with other systems or with certain modifications. Less frequently, other systems also used convective methods, such as T-shaped channels (6%) ( Figure 2G) [20,37], serpentine channels ( Figure 2A-D,F-I) [23,25,34,37,[40][41][42]44] cascade mixing ( Figure 2H) [40], and 3D microchannel networks [18] (3% each). Diffusion methods used to generate gradients were associated with Y-junction systems (6%) [21,28], the snake model [24], droplet generation ( Figure 2E) [16], static-pressure-driven CGG [31], and membrane systems ( Figure 2J) [17], with 3% each.  The snake geometry is created by interactively folding a channel in an equal interval until the design specification is met according to the rules: L = (N + 1)ws + Ls − 3w, where w is the channel width, ws is the snake width, Ls is the snake length, N is the number of snake's bends, and D is the snake density (Ls/N)      Interestingly, two of the studies, one from the microorganisms group and another from the tumor cells group, used a combination of convective and diffusive methods, such as serpentine/T-shaped channels [20], and Christmas tree/Y junction ( Figure 2G) [37]. Only three studies did not report the generation method used, and the systems used were centrifugal CGG [19] and circular concentration gradient [33], while the study by Qin, Y.X. et al. reported neither the method nor the system [32]. The CGG structure was designed according to the gradient system used and the generation method, to ensure efficiency in toxicity screening. These important aspects of the CGG structure are highlighted in Table 2.
An alternative way to evaluate the functionality of the structures and the efficiency of the generation of gradients is the use of software to simulate the flow. In this review, only 18% of studies used COMSOL (software for multiphysics simulation) [17,18,28,30,31,41], mainly those on tumor cells and other models and microorganisms. The study by Han, B. et al. (microalgae group) performed the simulation through computational fluid dynamics (CFD) [24]. In the studies with embryos, simulation was not reported [47,48].
For the creation of the perfect gradient, a certain amount of time is required, and only 30% of the studies reported these data, varying from 1 to 1800 s [16,[26][27][28]36,37,41,42,45,48]. The stability time, also an important parameter, was reported in only 12% of studies [25][26][27], varying between 20 and 50 min in the microalgae group, with one study from the tumor cells and others group reporting an indefinite time of maintenance [28], while the other groups did not report this parameter.
Regarding the concentrations of drugs or stimuli used in the study of toxicity, some of the more frequently used substances showed a similar pattern. For the antibiotic toxicity screening in microorganisms, the most frequently used antibiotics were Ampicillin (AMP) and Tetracycline (TAC), with concentrations ranging from 0 to 13.1 µg/mL [16,18,19], while the concentration of Ciprofloxacin (CIPRO) ranged from 0 to 96 µg/mL [17]. For the evaluation of the toxicity in microalgae, the range of concentrations of copper (II) sulfate varied from 0 to 4.375 µM, and for mercury (II) chloride, from 0 to 4 µM [24,25]. The concentration variation in the main chemotherapies applied in toxicity screening in the tumor cells group varied from 0 to 600 mg/mL for 5-Fluorouracil (5-FU) [41,43,45,46], from 0 to 400 mg/mL for Cisplatin (CDDP) [29,35,45], 0 to 3.4 mg/mL for Paclitaxel (PTX) [35,37], and from 0 to 0.01825 mg/mL for DOX [30,36]. For studies related to embryogenesis, a lower range was used (0-100 µg/mL) for Adriamycin (ADM), DOX, 5-FU, and CDDP [48].
Some advantages regarding the CGG system and microdevice structure were reported in 55% of the studies. Shear-free fluid flow was a concern considered by 32% of the works, where shear minimization was provided, mainly, by the shape of the mixing channels, difference in heights in relation to the culture chamber, and use of splitting-mixing systems associated with serpentine channels [18,19,28,30,36,38]. Automation was also considered by 26% of the studies, so vacuum pressure channels (Figure 2A), centrifugal force [19], snakechannel torque-operated valves [26], and centripetal geometry [47] were used to minimize handling and optimize the generation of gradients [19,24,26,44,47]. To guarantee the linearity of the gradients, 21% of the studies reported the optimization of the structures by modifying the length of the channels (Figure 2A,H) [40,44], using micropillars in the culture chambers [30], radial splitting-mixing integration with a serpentine channel [33,45], and cascaded mixing ( Figure 2H) [23,30,33,40]. In addition, 11% reported a concern regarding the high performance of these devices, the number of concentrations generated using radial splitting-mixing integration with a serpentine channel, and centripetal geometry together with the arrangement of concentric serpentine channels [45,47]. Only 5% of the studies reported a concern with mimicking the gradient in vivo [28] and reusing the developed devices [43]. Table 3 shows the details of the biological model used, the characteristics of the culture environment, and the toxicity conditions analyzed. The main microorganisms used as biological models were, firstly, bacteria (71.4%), with the great majority of the studies choosing different E. coli strains [16][17][18][19]-with exception of the study by DiCicco [21], in which a canine bacteria species (S. pseudintermedius) was used-and, secondly, Caenorhabditis elegans (C. elegans) nematode (28.6%) [20,22]. Both models were utilized for antibiotic toxicity screening-with exception of the study by Zhang B [20], which employed manganese chloride combined with vitamin E, resveratrol, and other substances. The most tested antibiotics were AMP [16,19], CIPRO [17,18], and TAC [16,18], followed by Kanamycin (KAN) [16], Amikacin (AMK) [18], Fosfomycin (FO) [21], and Amoxicillin (AMX) [22], with an incubation time between 4 and 72 h. The longer periods of incubation were associated with the evaluation of genetic mutation and antibiotic resistance. The drugs' flow rates were reported in less than half of the studies (42.9%) [20][21][22], with a range of 10 to 300 µL/h, and the organisms were mostly cultured intra-CGG (85.7%) [16,17,[19][20][21][22] and in 2D culture, with only two studies reporting the use of 3D culture (28.6%) [16,17], one being a co-culture. The average number of organisms for the studies that used bacteria was around 10 6 cfu/mL (10 8 for canine bacterium) and, for those based on nematodes, 1 worm/mL. Regarding the culture environment, the principal medium employed for bacteria culture was Luria-Bertani broth (for the E. coli strains), excluding the study based on S. pseudintermedius, which used Columbia agar associated with Tryptic soy broth plus glucose, and for C. elegans, a nematode growth medium was chosen. All bacteria were incubated with temperatures ranging from 30 to 37 • C, and the nematode studies applied lower temperatures around 20 to 25 • C.         Among the marine microalgae studied, 80% were Chlorophyta (green microalgae) [23][24][25][26][27], the most frequently seen species being P. subcordilformis (33.3%) [24][25][26][27], P. Helgolandica var. tsingtaoensis (25%) [25][26][27], and Chlorella sp. (16.7%) [23,26], all of which are from the previously cited phylum. This model was utilized for evaluating the toxicity of water pollutants, mainly metals and composts, most frequently copper (80%) [24][25][26][27], followed by mercury [24,25], cadmium [24,27], lead [25], and zinc [24], as well as other substances, such as sodium hypochlorite [23] and phenol [27]. The flow rates of the pollutant solutions and the exposure times in the toxicity evaluations varied greatly, from 0.1 to 50 µL/min and 1 to 72 h, respectively. These toxicity assays were mainly conducted intra-CGG (80%) [23][24][25]27], in an F/2 medium (80%) [24][25][26][27] (an enriched seawater medium was used in one study [23]), in a 2D arrangement, with an average amount of microalgae of 10 5 individuals or a range between 240 and 580 cells/µL, maintained mainly at 25 • C and in controlled light illumination of 60 µmol photon/m 2 /s.
Only two studies used the zebrafish embryos as the biological model [47,48], culturing them in 2D arrangement and intra-CGG, but with different purposes. One of the studies [47], which was performed with 10 to 12 eggs per chamber, focused on lead acetate and copper sulfate toxicity screening by exposing the embryos to these pollutants for 48 h, using flow rates from 5 to 30 µL/min and incubating them in an aerated ultrapure water medium supplemented with nitric acid and sodium hydroxide at 28.5 • C. The second study [48] was performed with one embryo/chamber for chemotherapy toxicity assessment, the drugs employed being ADM, DOX, 5-FU, and CDDP, as well as vitamin C, in different stages of embryo development (4 to 72 h post-fertilization), with the flow rate of 4 µL/min, with incubation in an E3 embryo medium at 26 • C, alternating between anoxia and normoxia. Table 4 highlights the main points of the proposal, evaluation, and outcome of the selected studies. The main proposal of studies on the microfluidic devices that used microorganisms (bacteria and nematodes) as a model was to perform an AST with single or combined (due to the antagonism or synergism effect) drugs with different exposure times, using the MIC value as a reference to compare the results with the gold-standard method, searching for the best efficiency while using the lowest amount of drug possible, as assessed by cell viability fluorescent techniques, as well as the influence of the drug's concentration on genetic alterations and mutations that lead to drug resistance, an extremely relevant issue nowadays due to widespread misusage of antibiotics. The nematodes were used for different purposes; one study [20] evaluated the behavioral response of the worms by fluorescence imaging in the face of manganese toxicity and the protective effect of natural antioxidants while the other [22] evaluated the effectiveness of certain antibiotics on the treatment of bacterial infection on nematodes, either associated or not associated with natural substances, showing these to be of value when treating the infection.

Toxicity Screening Evaluation and Outcome of the CGG Microfluidic Device
For marine microalgae, the main concern was the toxicity of chemicals linked to environmental contamination, either individually or in combination. These compounds were assessed using viability and motility techniques, which revealed varying sensitivities between different phytoplankton species. One study [23] concluded that Chrorella is more resistant than Pyraminmonas sp. to NaClO and the other microalgae, indicating the greater resistance of P. subcordiformis and P. helgolandica var. tsingtoaensis to all metals tested, especially CuSO 4 , which was shown to be the most toxic.
Most studies that proposed the screening of antitumor substances in cells from humans and animals assessed its efficiency via the use of different fluorescent dyes associated with cell viability evaluation (Calcein AM/Pi, Hoechst 33342, and Annexin-V-FITC), concluding that the drugs have a time/dose-dependent effect in almost all cases in which a drug was tested singly, and, also that the combination of drugs had better efficacy in lower dosages, with similar results seen for assays performed on Petri dish cultures. Only a few studies [28,29] assessed the effects of chemotherapy (or, in two studies [32,40], cigarette smoke extract) on normal cells, through the evaluation of apoptosis or oxidative stress via a reactive oxygen species (ROS) assay, showing that the toxicity and the malignant transformation of cells depend on the time of exposure. The study by Fernandes [42] was the exception in these approaches, evaluating the α-synuclein (aSyn) production and aggregation in Saccharomyces cerevisiae exposed to iron and ascorbic acid, due to the supposed protective effect of these substances.
Each of the zebrafish embryo investigations had a different objective. By using morphometric and behavioral analysis, one study [47] showed the damage effects of metal in different stages of embryo development. The second study [48] focused on the effects of chemotherapeutics normally used in embryo development and the ability of vitamin C to reduce harm. To study the emergence of resistant bacteria in spatial CIPRO gradients, then to perform the genomic sequencing to identify the key mutations that lead to antibiotic resistance Fluorescence time-lapse microscopy; genomic sequencing and biofilm assay (96 wells) Most genes affected in 48-h and 72-h were related to the bacterial envelope (rfaG, rfaE, rfaQ, and rfaC). There were similar mutations (in the marR and rfaG genes) and a 2-4× increase in MIC in cells, even without antibiotics and in antibiotic gradient for 48 h that can be explained by the environmental stress, and at 72 h the MIC was 1-30× higher Microfluidics mimic the complexity of natural microenvironments for bacterial resistance research, facilitating the evolution of resistance and promoting genetic diversity, even before the antibiotics administration Sweet, E. et al. [18] To identify optimal drug compositions through MIC values of an AST for the treatment of antibiotic-resistant E. coli bacteria and 3D µ-CGG to allow a symmetrical gradient of fluids combined more than 2 drugs solution at time Resazurin metabolic indicator and spectrophotometry (OD600) The bacterial growth response and the drug MIC values were the following~20% for TAC at~0.26 mg/L,~5% for CIPRO at~50 µg/L, and~30% for AMK at~11 mg/L.
Lower MIC values increased growth. With combined drugs, an antagonism effect between~0.34 mg/L of TAC and~28.8 µg/L of CIPRO occured, and a synergism effect with~7.68 mg/L of AMK and 48.8 µg/L of CIPRO was seen, and the value recommended was~6.08 mg/L of AMK and~65.3 µg/L of CIPRO in the infection treatment Provides higher throughput when compared to traditional assays testing multiple antibiotics. The devices are customizable and can be rapidly and cheaply produced for immediate application in medical routine Table 4. Cont.

Study Study Proposal Techniques for Evaluation Outcomes Microfluidics Advantages
Tang, M. et al. [19] To generate discrete concentration levels through mixing predefined volumes of sample and diluent at different proportions automatically Absorbance by spectrophotometry The number of E. coli bacteria did not increase after 3 h of exposure to ≥3.5 µg/mL of the MIC value of AMP MIC produces results much more rapidly than traditional methods automatically, saving labor time.
When integrated with optical detection units, it is more compact and cheaper than commercial spectrometer-based systems. Able to perform multiple processes simultaneously and has a fully customizable concentration gradient Zhang, B. et al. [20] To encapsulate a number of worms into the individual chamber and investigate the diverse behavioral responses to manganese toxicity

Fluorescence images by stereomicroscopy
The worm's motility impairment was dose-and time-dependent when exposed to manganese; high concentrations can cause effects of DAergic neurodegeneration and cell death in the worms, and the natural antioxidants can protect against manganese-induced toxicity Semi-automized processes. The microfluidic chamber design permits the formation of restricted habitats for the organisms, the administration of precise chemical stimuli, and their reaction assessment by conventional microscopy due to the optical transparency of the device's materials. Additionally, it has a low cost, good biocompatibility, and versatility of chip DiCicco, M. and Neethirajan S. [21] To evaluate the in vitro activity of fosfomycin against MRSP biofilms, to determine the MBEC SYTO 9 dye from a LIVE/DEAD ® BacLightTM bacterial viability kit The MBEC value was 8.6 ± 2.1 µg/mL of FO, and the concentration of FO needed to remediate biofilm-embedded cells of MRSP A12 is 8.1 ± 0.9 µg/mL Facilitates fast analysis of bacterial resistance, pointing to the correct therapeutic conduct  To assess metals' toxicity to microalgae (copper, mercury, zinc, and cadmium) alone or in a binary/ternary/quaternary mixture Brightfield microscope P. subcordiformis motility inhibition increased with exposure to the increasing concentration of single pollutants of Cu, Hg, Zn, and Cd for 1 h. Hg was the most toxic, followed by Cu and Cd, and Zn was the least toxic. After 1 h, the metal mixture of Hg, Cu, and Cd with Zn was more damaging than Cu with Zn, Cd, or Hg Offers higher-throughput alternative to conventional methods and might be employed for other types of assays Zheng, G. et al. [25] To assess the marine phytoplankton motility and investigate the pollutants' toxicity effect (Hg, Pb, Cu, and phenol) Simplifies toxicity assays. The device allows for easy customization of culturing conditions. Can also be rapidly fabricated Simplifies and accelerates toxicity assays

Tumor cells and other models
Chennampally, P. et al. [28] To evaluate the effectiveness of rapamycin in rescuing the MN of ALS Immunostaining for GFP and TDP-43; Western blot ALS-affected motor neuron survival can be increased by 40.44% in a rapamycin dosage range between 0.4 and 1.0 µM As it is compatible with traditional techniques, they can be combined to obtain the advantages of both. Enhances throughput and results in the entire assay on only one cell culture. Enables multiple simultaneous tests, and has the capacity to stimulate cells to adopt spatial distribution and morphology similar to those in vivo Yin, L. et al. [29] To predict the nephrotoxicity induced by CDDP, GM, and CsA in renal chip

Calcein-AM/PI and CCK-8 assay
Cell viability was higher in static than fluidic co-culture condition. The cell viability was dose-dependent for all drugs. Cim neutralized and reduced the toxicity of CDDP, thus improving the survival rate of renal cells Automation of multiple processes. Studies can be performed on models which reproduce key features of an organ's physiology. Microfluidic devices can bring standardization, automation, and a reduction in costs to drug assays. They can also accelerate the whole process and lessen the impact of human bias To generate a shear-free microenvironment for long-term cell culture and adaptive cytoprotection analysis with a pumpless hydrogen peroxide gradient generator

Apoptosis by Annexin-V-FITC and PI
More stable and precise biochemical gradient by static pressure. Pretreatment of low-dose H2O2 protected NIH 3T3 cells against cytotoxicity. An H2O2 lethal dose results in 27.72% of apoptosis. Pretreatment for 24 h with lethal hydrogen peroxide exposure arrested the apoptosis in a dose-dependent manner. Apoptosis ratio decreased to~27,~22, and~14% with 25, 75, and 175 µM, respectively Simple operation, without the need for external equipment and easy fabrication. A portable device which provides stable concentration gradients and is suitable for long-term cell culture, due to its low shearing effect Qin, Y. X. et al. [32] To detect the role of the HHS in CSE-induced malignant transformation of 16HBE Apoptosis by fluorescence (Hoechst 33342), Western blot 16HBE CSE-induced cell apoptosis was dose-dependent, high doses (≥19.86%) promoted cell apoptosis, low doses (≤12.28%) promoted less apoptosis and continued cell growth (>80% cell viability). The best concentration for CSE stimulation was 12.25%, and after 15 weeks, some cells displayed condensed nuclei and abnormal nuclear-to-cytoplasmic ratios, atypical mitoses, and later a loss of contact inhibition. These alterations were not apparent in the cells treated with cyclosporine Provided greater efficiency, accuracy, lower time, high throughput, and constant control of microenvironmental conditions via computer programs (automation), simple operation, and low costs of construction compared to traditional methods. Emulates the in vivo cell microenvironment and permits the dynamic observation of their growth Table 4. Cont.

Study Study Proposal Techniques for Evaluation Outcomes Microfluidics Advantages
Luo, Y. et al. [33] To drug screen for diabetes with glipizide in 3D INS-1 high-glucose cell model through the circular CGG MTT, Calcein-AM/PI, Ultrasensitive Rat Insulin ELISA Assay kit After 24 h addition of glipizide, the decrease rate of inhibition rate with glipizide concentration was 0.5916 and 0.3183 for 3D and 2D models, respectively, and after 48 h, it was 0.9133 and 0.4817 for 3D and 2D models, respectively. The 3D model was more sensitive than the 2D model and produced a greater insulin production response in diabetes drug screening High throughput. The use of the 3D cell model, facilitated by the device, was shown to produce better results than the traditional method. Multiple parallel assays can be conducted Lim, W. and S. Park [34] To develop a µFSCD with a CGG that enables cells to form spheroids and grow in the presence of cancer drug gradients

Live/dead
The HCT116 cells' viabilities are drug dose dependent, their viability decrease (63%) after 5 days of 5 µM irinotecan treatment (highest concentration), while the cell viability in the control was 98%.
The device facilitates homogenous spheroid generation. Allows for high-throughput and multiple parallel assays. Its CGG system makes the generated concentrations easily calculable. The materials used allow the gradients formed to be maintained for long periods and the observation to be made using a conventional optical microscope Table 4. Cont.

Study Study Proposal Techniques for Evaluation Outcomes Microfluidics Advantages
Jin, D. et al. [35] To assess drug sensitivity in spheroid head and neck perivascular tumor model and toxicity in endothelium Hoechst 33342, PI, and immunostaining The IC 50 values of PTX, CDDP, and 5-FU for 3D-UM-SCC6 were 0.54, 5.5, and 454 µg/mL, respectively, and for ACC-M, they were 0.45, 5.2, and 400 µg/mL, respectively, being higher than in 2D culture. Low concentrations of PTX or 5-FU combined with CDDP had similar effects to high concentrations of a single drug on tumor cells and low cytotoxicity to HUVEC, leading tõ 50% apoptosis of tumor cells, and already high concentrations of combinations were toxic to HUVEC cells. Different patients' tumor cells showed relatively high sensitivities to both combinations with~60% survival, while others showed low sensitivity with 80% cell survival Allows for in vivo administration of drugs to be emulated. Microfluidic devices are better suited for the culture of spheroids, providing better results than conventional 2D culture methods. High throughput, lower costs, maintenance of concentration gradients for long periods of time, and real time analysis are features provided by the microdevice. If needed, more than one drug could be loaded into the device for testing Hong, B. et al. [36] To drug screen with CGG on a paper-based device Live/dead (Calcein AM/PI; Prestoblue) After 8 h, the cell viability was >50% with 50 µg/mL DOX and 20% with 200 µg/mL DOX Allows for multiple simultaneous assays under different drug concentrations to be conducted, as well as automation and a reduction in costs and reagent volumes, increasing the overall efficiency Ying, L. et al. [37] To assess the impact of CAF or HGF on the Met/PI3K/AKT phosphorylation, GRP78 expression and PTX-induced apoptosis in A549 cells cultured in the 3D matrix Viability (Rhodamine-123); immunofluorescence; Western blot; immunohistochemistry; apoptosis assay (Hochest33342 e PI); ELISA HGF in the CAF matrix activated the Met/PI3K/AKT and up-regulated GRP78 expression, promoting chemoresistance to PTX-mediated apoptosis in A549 cells. PI3K and GRP78 inhibitors elevated PTX action in cell viability: 90%, 95%, and 100% at 1.28, 2.59, and 4 µM PTX, respectively High throughput, high sensitivity, reduced substance volumes and overall experiment time. Emulates natural cell microenvironments The device showed more sensitivity in toxicity tests than in the 96-well culture (IC 50 of 17.8 versus 22.8 mM, respectively), being 128% higher and >1800% less time-consuming due to the use of an automated LabVIEW system that refreshes APAP on the target cells every 4 h Compared to the 96-well culture system, cells showed higher sensitivity to the substance tested, leading to the conclusion that the microdevice produces more accurate results. Time spent, as well as reagent and sample consumption, are reduced. Provides high throughput, integration of several techniques in one assay, and automation Pasirayi, G. et al. [39] To chemotherapeutically screen for PCN, PTX, and aspirin singly and combined in two types of tumor cells

Calcein AM
Concentrations of PCN and PTX LC 50 on MCF-7 were~60 and 0.63 µM, higher than in traditional culture (~51 and 0.55 µM), respectively. HepG2 showed the same results with high resistence to PCN (100 µM) with 70% of viability. A total of 0.2 µM of PTX reduced cell viability to 83%, while 4 mM aspirin alone reduced cell viability to 84%. PTX plus aspirina had a higher effect on the loss in cell viability than PTX alone Cells cultured in the microdevice showed more growth after exposure to drugs, compared to those cultured in 96-well culture plates. Has a low cost and provides the possibility of testing drug combinations Li, E. et al. [40] To investigate the potential mechanisms underlying tumor-like transformation of continual exposure of primarily cultured human bronchial epithelial cells to CSE Hoechst33342 and propidium iodide (PI); ROS Assay kit; immunofluorescent assay (GRP78, NF-κB, and PI3K) (E-cadherin and Vimentin); Western blot Lower doses (2.37-12.28%) of CSE stimulated cell proliferation, but not cell apoptosis, and higher doses (19.86-91.88%) induced cell apoptosis. All analyses were one-way and dose-dependent, as well as the results for ROS Emulates heavy smoking in humans and the lung microenvironment, making the device ideal for experiments of this kind due to its dimensions, its material properties, and the steady flow of the medium, therefore generating more accurate results. Additionally, it allows for parallel assays with diverse conditions, minimizing possible errors To screen for anticancer drug and chemoresistance phenomena using the SpheroChip and assessing metabolic activity via dynamic changes in two types of tumor cells Live/dead (Calcein AM/PI); Fluorescent resorufin (metabolic activity) The growth of HepG2 spheroids was slightly higher than that of HT-29 inside a chip. HT-29 spheroid had normal metabolic activity until To study aSyn production and aggregation in Saccharomyces cerevisiae using an elastomeric microfluidic device exposed to iron and ascorbic acid Live-cell imaging; tracking the behavior of single cells by fluorescence image and α-synuclein (aSyn) production The proportion of single cells trapped was higher for more loosely packed traps (43% for x = y = 20 µm). FeCl 3 induced the formation of aSyn inclusions in a concentration-dependent manner, and ascorbic acid reduced the formation of aSyn inclusions in Y4791 yeast cells The device enables the creation of controllable microenvironments with precise conditions and, also, the use of a minute quantity of solutions for creation of the concentration gradient, as well as the tracing, over time, of individual cells' responses, unlike traditional methods. Compared to manual mixing of solutions, the CGG is prone to less mistakes, and is faster and less complicated Jastrzebska, E. et al. [43] To assess drug combinations of He and 5-FU anticancer on normal mouse embryo cells (Balb/c 3T3) and human lung carcinoma cells (A549) Live/dead-Calcein AM/PI Celbx and NSAID inhibited the growth of cancer cells and indicate anticancer properties. After cells' incubation with Celbx, the viability of A549 cells was lower than normal Balb/c 3T3 cells, and Celbx plus 5-FU enhanced antitumor activity The CGG made it possible to obtain multiple combinations of the tested substances automatically and simultaneously, while also improving the repeatability To assess HeLa apoptosis of the single and combined effects of two drugs through combinatorial, quantitative, and predictable concentration gradient by repeated splitting and mixing DAPI, annexin V-FITC/PI apoptosis detection kit Cellular morphological changes with the increase in drug concentration: cell shrinkage, increase in cell granularity and chromatin condensation, and the most apoptosis characteristics. The apoptosis effect induced by CDDP was more obvious with the increase in stimulation time and concentration The CGG produces an extremely wide range of stable, customizable, and repeatable concentrations. It also possesses a compact design and provides high throughput, while reducing time of analysis Jedrych, E. et al. [46] To assess the 5-FU cytotoxicity on two human cancer cell lines Calcein AM/PI After 24 h, cell death by 5-FU increased in a concentration/time-dependent manner, inhibiting the survival of both cell types; HT-29 cells were less sensitive than A549 cells. The strongest inhibition, approaching 80% after 48 h of incubation, was observed for A549 cells exposed to 300 µM 5-FU The device allows for the execution of different methods of toxicological evaluation, as well as automation of processes. It also permits the simultaneous cultivation of cells with different characteristics and the lowering of costs and time needed To describe a phenotype-based whole-organism model to assess the developmental toxicity and teratogenicity of anticancer drug-induced zebrafish embryos Stereomicroscopy ADM and CDDP had similar toxicity and teratogenicity in 4 hpf embryos, and 5-FU was halved under the same conditions. These effects vary according to developmental embryo stages, mainly for DOX, which exhibited obvious time/dose-dependent toxicity and LD 50 = 91.7 µg/mL. As for the advantages of utilizing microfluidic devices rather than traditional macroscale methods, the great majority of the studies reported similar benefits. Firstly, some of the studies reported that the results obtained with microdevices correlate very well with those obtained using conventional methods, sometimes even mimicking more accurately in vivo conditions, showing that the technology in question can be adequately applied when studying toxicity. With that in mind, one of the most important aspects reported is the possibility provided by microfluidic CGG's ability to generate a very sizeable number of different concentrations (up to 65 in the studies analyzed) in a single device and, consequently, the possibility of conducting multiple parallel assays, both of which, allied with the prospect of automation of processes (such as the generation of the concentration gradient and metabolite collection), can significantly diminish the time expended and make this a high throughput method for toxicity screening. Other very significant advantages brought about by this technology include the small size of the devices, which translates to less space occupied, making it possible to have multiple devices running multiple assays at the same time, further increasing the throughput, and also, the low quantity of reagents used, decreasing the cost of the tests. Besides that, the microdevices can be easily and rapidly fabricated, with different well-established fabrication processes and a variety of materials, and easily operated, as well as integrated with other traditional techniques, combining the advantages of both. Microdevices are also more preferable for 3D cultures than some of the more traditional methods, and they make it possible to create microenvironments that are more like those seen in nature, producing results that are more trustworthy. The analysis of microfluidic devices can also be carried out via a variety of methods, providing the researcher with a lot of design flexibility. The device's versatility, which allows it to be developed in an endless number of ways for various purposes, with various test chemicals and biological models in mind, is still another significant advantage that can be seen. All these parameters are described in Table 4.
According to their toxicity methodologies and the biological models employed for this testing, the studies' findings are described in Figure 3 in conjunction with the major features that are considered in this systematic review.

Discussion
Advances in microfluidic device development technology for toxicity screening have provided remarkable advantages over conventional two-dimensional cultures due to the reduction in the sample consumption, reaction time, and cost of the operation. This systematic review gave a broad overview of the main aspects and trends regarding the manufacture of microfluidic devices, the promotion of the CGG's development to boost the effectiveness of its chemical and drug toxicity screening, and the most tried-and-true biological models for addressing issues concerning environment and medical treatments.
Regarding microfluidic device fabrication, all were manufactured in-house, providing device customization for more efficient testing, which was specific to each biological model used. There is still a strong tendency to use materials and manufacturing techniques such as PDMS and soft lithography (91%), but recent articles search for more sophisticated technology, such as a 3D printing, silicon micromachining, and direct writing photolithography using glass. PDMS is the most commonly used material in microfluidics, because of its flexibility, biocompatibility, nontoxicity, good stability, and high transparency [49], even though earlier studies-some from more than a decade ago-brought up disadvantages, such as the absorption of small molecules [50], its incompatibility with organic solvents [51], and its vapor permeability [52], and more recent articles have questioned its practicality and widespread use, citing the difficulty of translating results obtained with it to other materials and its poor scalability for commercial purposes [53][54][55][56] as major concerns. The studies that did not use PDMS reported the use of materials such as silicon and glass which have, roughly, the same advantages as PDMS beside hydrophilic capabilities, reusability, and flexibility [45]. The biomedical field finds 3D printing to be a highly valuable technology for diagnostic and/or therapeutic purposes; its applications range from tissue engineering to microscale robotics and biosensors, besides rapid prototyping flexibility and a variety of forms and functions, having the advantages of precisely controlling the spatial distribution layer-by-layer, the generation of heterogeneous microorgans, and 3D cellular arrangement on a chip [57,58]. Only one study [39] utilized thermoplastics in some way, which is an interesting finding, given that, in recent years, materials such as polycarbonate (PC), poly (methyl methacrylate) (PMMA), and cyclic olefin copolymer (COC) have been gaining notoriety and have been widely used in industry when aiming for the fabrication of a product [53,54] The studies that used conventional manufacturing varied the type and number of molds. Photolithography was the most used method of fabrication (79%), mostly due to its high accuracy, despite its high cost [59]. A study on optimization of SU-8 microstructure in high-transparency masks, printed in a photomask, however, showed the possibility of their fabrication with a low-cost process and without the requirement of cleanroom facilities [60]. Laser cutting techniques, as well as CNC, when compared to traditional photolithography and etching methods, have the advantages of being a simple, fast, and direct-write process for the fabrication of different geometrical shapes. Both techniques provide complex geometries with different layers, normally more than one layer (72.3%), with a micrometer scale. The layers represented the different environments and testing functions of the microdevice, providing greater efficiency within its complexity.
The CGG system is a faster and more accurate method for drug and chemical pollutant toxicity analysis. It only needs a small amount of reagent for multiplex analysis, which lowers the cost. It is also capable of screening at the molecular and cellular levels and has multistep liquid-handling capabilities, which is especially useful for complicated screening procedures, in addition to its features of miniaturization, integration, and automation of analytical systems [36,61].
The method of gradient generation was based on two patterns, convective and diffusive. Most of the selected studies used the convective method for gradient generation (67%), which is a simpler and easier method for drawing and calculating. In convection-based gradient generators, the concentration gradient depends on the flow field, which can produce shear stress above the physiological limit endured by cells. The diffusion-based gradient generator, on the other hand, offers isolated chambers due to the interface, and the inside reagents are protected from the outside shearing [10]. The Christmas tree generation system was the most used (61%), associated and not associated with other systems, which indicates the frequency of the convective pattern, and its main advantages were its simple design and a well-defined concentration range, allowing isolated assessment of each concentration. However, this pattern can be integrated with other systems such as Y-junction systems or in two separate layers, one in which the convection pattern is evident (the CGG layer), forming the concentrations, and one containing the culture chamber, in which the different concentrations flow through diffusion. A few studies used similar systems to the Christmas tree, such as serpentine channels, cascading mixing, and T-shaped systems, which have certain advantages, for instance, fewer stages [62]. The studies that used the diffusion pattern applied a variety of gradient systems, such as Y-junction, membrane, and droplet generation. This last system shows difficulty in controlling flow and concentration while maintaining the droplet shape, two crucial parameters for toxicity assessment.
Gradient linearity is the expected behavior of CGG, being reported in 91% of studies due to the need to assess dose dependency on drugs and toxicity. The studies used two methods to analyze CGG linearity and performance: flow simulation and validation. The flow simulation occurs in a stage before the CGG manufacturing, allowing quick design adjustment, but only seven studies (21%) reported analysis using the COMSOL software, likely due to its high cost and requirement of an expert user, making access to it difficult. CGG validation is a different type of analysis that can only be performed once the microdevice is complete. The most employed substances for this method were fluorescent agents (55%) [18,[20][21][22][25][26][27][28]31,34,35,37,38,42,43,45,47] followed by dye solutions (12%), bringing a visual analysis of the flows of the channels and the concentrations generated [19,29,36,38,39]. Some studies also performed a quantitative analysis to be compared with the final concentrations.
Among the most tested antibiotics in the toxicity analysis, there was a slight predominance of Ampicillin and Ciprofloxacin, which are effective against a wide range of both Gram-positive and Gram-negative bacteria, while having distinct modes of action [63]. The concentration range of both drugs showed a similar pattern (from 2 to 16), showing MIC values consistent with the gold standard of conventional analysis, being more efficient in terms of analysis time and material consumption, and allowing combined-drug analysis for synergism and antagonism effects, using a drug exposure time from 4 to 72 h. The use of prolonged subtherapeutic levels is a concern regarding bacterial resistance, with microfluidic devices having been shown to be more efficient for this analysis than conventional techniques due to the possibility of mimicking the in vivo microenvironment, while guaranteeing high performance [64]. The main bacteria utilized as a model was the Gram-negative Escherichia coli (E. Coli) (for example, E. Coli k-12), which may cause severe food poisoning and is a global health problem due to the rise in antibiotic resistance. Due to its unrivaled fast growth kinetics, high-cell-density cultures, and quick and simple exogenous DNA transformation, this species of bacteria is the most popular for use in toxicity assays [65]. Almost all bacteria were cultured inside the CGG system in a 2D culture dispersed in a medium from 30 to 37 • C, with the exception of the study by Sweet [18], which cultured them in a separate system (extra CGG), and the studies by Zeng [16], which used 3D culture, and Nagy, who used 3D co-culture [17].
A few studies also used nematode C. elegans as a model for toxicity screening, assessing the influence of antibacterial activity with various rhubarb components [22], and dopaminergic neurotoxicity induced or not by manganese associated with antioxidant elements [20]. This is a strong model organism because of its small size, optical transparency, short life cycle, and genetic tractability, among other advantageous traits, such as the ability to be infected by a variety of human pathogens and low cost of maintenance [66]. This microorganism was also cultured inside the CGG system in 2D culture dispersed in the medium from 20 to 25 • C.
Through the evaluation of metals and contaminants, a number of research articles have addressed the problem of environmental toxicity. Global pollutants such as mercury and lead, for instance, have an impact on both human health and the ecology around the world [67]. Microalgae have reportedly been used for biological detoxification, effluent treatment, control of toxic metals in natural waters or effluents, and control of toxic metals in naturally or industrially contaminated waters [68], as well as to retain and immobilize some compounds. Thus, it is essential to create tools that can investigate and aid in the creation of new technologies that are beneficial for the environment and, by extension, for human health and quality of life. Although other metals, such as arsenic (As), cadmium (Cd), chromium (Cr), lead (Pb), and mercury (Hg), are poisonous to microalgae, they can ingest trace amounts of metals, including boron (B), cobalt (Co), copper (Cu), iron (Fe), molybdenum (Mo), manganese (Mn), and zinc (Zn). Low-hazardous metal and compost concentrations can promote the growth and metabolism of microalgae because of the hormesis phenomena [69].
The metals Cu [24][25][26][27], Cd [24,27], and Hg [20,24] were evaluated the most often, singly or combined with other chemical elements, likely due to the high plastic accumulation in oceans from unrecycled waste and its decomposition [70] or the increase in mining and industrial activity, leading to mercury deposit [70]. The concentration range of these metals was similar (from 0 to 4 µM), varying from five to eight different concentrations tested. Green microalgae (Chlorophyta), the main model used, are photosynthetic protists and one of the groups of algae most closely related to terrestrial plants, also being used as indicators of water quality and having significant ecological importance, as they are components of phytoplankton, one of the primary producers in the food chain [71]. The microalgae were cultured mainly inside the CGG system-with the exception of the study by Zheng [26], which cultured them outside the CGG system-in 2D culture dispersed in the F/2 medium at about 25 • C, in controlled light illumination of 60 µmol photon/m 2 /s, close to normal environmental conditions. Zebrafish embryos are frequently used in metal toxicity studies due to their ability to grow outward and having clear enough bodies to be examined under a standard optical microscope [72]. One study in this review evaluated the Pb and Cu toxicity, singly and combined, regarding its teratogenicity in different stages of embryo development, such as the larval, juvenile, and adult stages [47]. Another study used this model to analyze different types of chemotherapy drugs and the protective effect of vitamin C during treatment, evaluating their influence according to the developmental embryo stages [48], using drug doses significantly lower than those applied in the tumor cell and other models group. These embryos were cultured inside the CGG system in 2D culture dispersed in different medium conditions at about 27 • C, and the toxicity evaluation occurred from 1 to 72 h of exposure.
The vast majority of the tumor cell and other model group studies performed antitumor toxicity screening using various tumor cell types from human [26,[30][31][32][33][34][35][36][37][39][40][41][42] or animals [33] and normal human cells [29,32,35], evaluating, primarily, the effect of the drugs 5-FU [35,41,43,45,46], CDDP [29,35,44,45], PTX [35,37,39], DOX [30,36], isolated or combined with others. CDDP and 5-FU combined are considered the standard antitumor treatment, and PTX followed by CDDP showed greater antitumor activity [73]. The toxicity of isolated Doxorubicin occur via acting on DNA by slowing or stopping the proliferation of cancer cells by inhibiting an enzyme called topoisomerase 2, their cardiotoxicity is the main factor restricting its use, and the total cumulative dose is the only factor currently utilized to predict toxicity, with microfluidics providing a new form of assessment [74]. 5-FU and CDDP also have activity on DNA, inhibiting thymidylate synthase, and crosslinking with the urine bases on the DNA to form DNA adducts, preventing repair of the DNA, leading to DNA damage and subsequently inducing apoptosis within cancer cells, respectively. The dose of these drugs was significant compared to other drugs, ranging from 0 to 600 mg/mL for 5-FU and 0 to 400 mg/mL for CDDP. PTX has a different antitumor mechanism, promoting the assembly of tubulin into microtubules and preventing the dissociation of microtubules, blocking cell cycle progression, preventing mitosis, and inhibiting the growth of cancer cells, being used, in the studies, in lower doses, from 0 to 3.4 mg/mL [75].
These cells were mainly cultured inside the CGG system in 2D conventional culture dispersed in DMEM or RPMI media, being mostly supplemented with fetal bovine serum (FBS) and other supplements at 37 • C. A few studies also used 3D culture and co-culture with different tumor or normal cells, and some studies specified the use of spheroids, a variation of conventional 3D culture. 2D cell culture models have been used to assess the toxicity or efficacy of drug candidates due to the ability to anticipate drug responses, but they have been found to be comparatively weak in comparison to 3D cell cultures, which have better functional and phenotypic characteristics, as well as predictability of therapeutic effectiveness [34,35]. In vivo, cells are arranged spatially into three-dimensional (3D) patterns that are encircled by an extracellular matrix (ECM), which leads to cancer cells growing in 3D cultures; in comparison to 2D cultures, these cells are more resistant to cytotoxic drugs [76]. Spheroids are one of the most relevant and modern models for cancer research. Their morphology and physiology are similar to those of a tumor in vivo, showing a network of cell-cell interactions, a 3D structure, the presence of a natural extracellular matrix, and nutrients, metabolites, and oxygen gradients [77,78].
Other contexts also used normal cells without the influence of chemotherapeutics. Two studies evaluated the influence of dose and time of exposure to tobacco extract on the malignant transformation of normal bronchial cells. The tobacco epidemic is one of the biggest public health threats the world has ever faced; there is no safe level of exposure to tobacco, and chronic cigarette-smoke-induced time-dependent epigenetic alterations can sensitize human bronchial epithelial cells for transformation by a single oncogene [79]. The study by Fernandes [42] investigated the basic molecular effects of aSyn in the context of living cells, with human aSyn being expressed in yeast and found to induce dose-dependent cytotoxicity, while iron (III) chloride and ascorbic acid were shown to have a protective effect [42]. The molecular basis of various human diseases has been extensively researched using Saccharomyces cerevisiae as a model organism. It is most well-studied in eukaryotic cells, while also being the easiest organism to grow under controlled circumstances and to manipulate genetically [80]. The study by Luo [33] used the INS-1 cells, which are a widely used and well-established model for the study of diabetes and their property of glucose-stimulated insulin secretion [81].
A relevant aspect in studies with microfluidics devices evaluated in the biological model was the flow rate used to infuse the nutrients and components to be tested for toxicity. This condition is very important in the biological environment. In microfluidic devices, shear stress is created by fluid flow injection due to several important aspects, such as channel dimensions and geometry, cell concentration, cell line type, and the way the flow rate is delivered, among others. Microfluidics provides a good way to mimic flows found in veins and small arteries, where the flow is usually unidirectional and laminar [82]. Shear stress can influence cell attachment [83], pathological response [84], and developmental biology [85].
The studies' outcomes in toxicity screening using the CGG system in the microfluidics device showed comparable results to the conventional toxicity studies, and the efficiency evaluation techniques applied were mainly based on fluorescence signals, followed by spectrophotometry and brightfield microscopy, molecular methods, and other techniques (enzyme-linked immunosorbent assay-ELISA, 3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide-MTT, and Western blot), showing the achievement of high efficiency in a faster way and the possibility of automation.
One of the limitations of this review was the lack of a detailed comparison of the complexity of the structures and geometries presented by the microfluidic devices developed in the studies. This analysis could help us better understand the significance of the micro-CGG on a global scale as well as the role that device design plays in the generation of the gradient and in each biological model that was investigated, but it was challenging due to the wide variation in the geometric arrangement and size of the studied biological models. Another limitation was the time frame used, 10 years, it was not sufficient to confirm whether there was a trend of CGG devices in relation to manufacturing characteristics, and previous gradient generation systems.
This systematic review also identified some methodological problems and research gaps, such as the relationship between the material used to make devices and the biological model or substance tested for toxicity, taking into account the benefits and drawbacks of each material, the sparse use of simulation procedures prior to device fabrication, and also the methodological care with regard to the duration of stable concentrations obtained by the CGG, which may compromise the accuracy of toxicity evaluation.

Conclusions
This systematic review showed a variety of toxicity assessment applications in the environmental and medical approaches through concentration gradient generation systems in microfluidic devices. Current studies have adopted new technologies and complex structures to customize the device according to the biological model, to achieve the best testing efficiency and to minimize typical microfluidics issues such as bubbles and shearing. The microfluidic gold-standard technique, soft lithography, using the polymer PDMS, was still the most frequently used, and the Christmas tree shape was the most prevalent CGG design, but alternative techniques and designs were employed to produce a larger variety of concentrations and drug combinations more precisely and more outcomes at once. Thus, the CGG microdevice is an alternative to common pipetting techniques for the evaluation of drugs'/substances' toxicity in various biological organisms, bringing greater precision with a lower cost.