Evaluation of clinical chemistry using diluted sera in mice and rats Authors: Kil-Woong Ha, Jeonghee Cho, Kyung-Tai Kim, Jeongah Song Published: 2026 ABSTRACT AbstractBackgroundIn preclinical toxicity and efficacy studies, clinical pathology tests play a critical role in detecting pathological alterations, assessing drug effectiveness, and monitoring animal conditions. However, their application is frequently restricted by the limited sample volumes that can be collected from laboratory animals such as mice and young animals. To overcoming this limitation, dilution of samples may be utilized. In this study, we investigated the impact of pre-dilution by using a Toshiba analyzer and saline as the diluent, comparing analytical results from diluted and undiluted specimens in both mice and rats. Seventeen biochemical analytes were assessed in serum samples diluted 2.5-, 5-, 7.5-, and 10-fold, and these results were compared with those of undiluted samples.ResultsWe evaluated mean values, bias, two standard deviations (2SD), and the standard deviation index (SDI). To ensure data obtained from diluted samples are valid and dependable, we applied allowable total error (TEa), defined as the maximum acceptable error for a test result to remain reliable and clinically relevant. The mean values from 2.5-fold diluted samples exhibited no significant difference compared to those from undiluted samples in mice and rats, except Cl, Na, ALB, and TBIL. Furthermore, findings from bias, 2SD, SDI, and TEa analyses indicated that a 2.5-fold sample dilution is suitable for clinical chemistry analysis of multiple analytes, except for Cl, Na, and TBIL in both species.ConclusionsCollectively, these results may serve as a reference for the suitability of diluted serum samples in clinical chemistry analysis. Further studies utilizing various analytical instruments, diluents, or animal species are recommended.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12917-026-05448-1.Keywords: Clinical chemistry analysis, Diluted serum, Bias, Allowable total error Background Clinical and toxicological pathology have undergone significant development for decades, driven by advances in the pharmaceutical and biotechnology industries [29]. In preclinical toxicity and efficacy research, clinical pathology testing provides important means for detecting pathological changes and disease progression, evaluating therapeutic responses to administered drugs, or monitoring the health status of animals [22]. This encompasses hematology, blood coagulation tests, clinical chemistry, and urinalysis performed on blood, urine, and other body fluids from laboratory animals [1, 28]. Alterations in these test results are associated with abnormal physiological conditions in organs such as liver, kidney, brain, or heart. Therefore, these assessments act as relatively non-invasive tools for evaluating toxicity or drug efficacy over time. Clinical pathology testing is constrained by the limited sample volume that can be obtained from small laboratory animals such as mice and juvenile animals. This issue becomes especially pronounced during interim blood sampling, as the insufficient blood volume often hinder adequate clinical pathology testing. For instance, the weights of C57BL/6J mice from 3 to 24 weeks of age range from 10.1 g to 35.3 g [15]. Similarly, the weights of Sprague Dawley (SD) rats at postnatal days 7, 21, and 35 are approximately 14 g, 45 g, and 115 g, respectively [10, 18]. The NC3Rs recommend that no more than 10% of the total blood volume should be collected from mice and rats on single occasion, particularly using non-surgical or non-terminal method [19]. The total blood volumes in mice and rats are 58.5 mL and 64 mL per kg of body weight, respectively. Thus, a mouse weighing 25 g has an estimated total blood volume of 1.46 mL, allowing for the collection of up to 0.14 mL of blood. For rats at postnatal days 7 and 21, 35, the corresponding total blood volumes are 0.90, 2.88, and 7.36 mL, enabling the collection of 0.09 mL, 0.288 mL, and 0.736 mL, respectively. Clinical pathology testing relies on specialized analytical instruments and standardized methods, each of which necessitate a specific minimum sample volume [12]. Hematology tests, which evaluate blood cell counts and cellular composition, usually require between 0.15 and 0.30 mL of sample, although this volume can vary depending on the equipment. For clinical chemistry tests, the sample requirement depends on both the analytical equipment and the number of analytes to be assessed; however, regulatory and international guidelines [20] typically mandate at least 0.2 mL of serum for analysis. Collecting 0.2 mL of serum usually necessitates 0.4–0.5 mL of whole blood. The restricted blood volume available from laboratory animals presents a substantial barrier to conducting comprehensive clinical biochemical analyses [16]. The challenge of limited sample volume is especially prevalent in rodents, which are the primary model organism for toxicity and efficacy studies [16, 23]. Nevertheless, this obstacle also affects studies involving other species. For example, researchers encounter similar limitations when collecting blood from fetuses and neonates of non-rodent species, such as minipigs used in developmental or juvenile toxicology studies. In numerous mouse studies, animals are divided into separate groups for hematology and clinical biochemistry analyses, resulting in a higher total number of animals used and precluding the generation of comprehensive hematological and biochemical profile for each individual mouse [12]. Procuring both hematology and clinical chemistry data from the same subjects provides statistical advantages and enable more robust data interpretation compared to using additional animals for separate analyses. Certain researchers have employed pooled samples from multiple animals. However, these methods necessitate greater animal use overall, which is inconsistent with the 3R principle [25]. One strategy to address the challenge of insufficient sample volume is to dilute the sample prior to analysis [16]. When analyte values are above the upper limit of detection, sample dilution is advised by the equipment manufacturer. Nevertheless, predicting the impact of pre-dilution on the initial analytical results is challenging, as analytical methods and reagents may differ between equipment manufacturers. Additionally, potential biases may result from sample-diluent interactions, alterations in enzyme activity, and other variables [24]. Manual pipetting inaccuracies during the dilution process can also introduce errors. This study aimed to assess the impact of dilution in clinical chemistry analysis. Although some studies have addressed this topic, various analytical platform and diluents were used [12, 16, 27]. Here, we examined the effect of pre-dilution using a Toshiba analyzer with saline as the diluent, comparing the results from diluted samples to those from undiluted samples in both mice and rats. Methods Experimental animals Rats and mice were obtained from Orient Bio Inc. (Seongnam, Korea). Rat sera were obtained through a research project. Mice were bred at the Korea Institute of Toxicology. Both species were housed under controlled conditions: temperature of 22 ± 3 °C, relative humidity 30–70%, 12 h light/dark cycle, and air exchange at a rate of 10–20 times per hour. Animals were provided unrestricted access to sterile pelleted food and UV-irradiated, filtered tap water. All experimental procedures received approval from the Institutional Animal Care and Use Committee of the Korea Institute of Toxicology (Approval no.: IAC-23-01-0026-0030, 2211-0012) and were carried out in a compliance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International. Experimental design Seven‒eight-week-old C57BL/6 Mice were separate into three groups, with each group containing 10 males and 10 females. Because of restricted sample volume, specific analytes were assigned to each group (Table 1). Eleven-week-old male SD rats provided enough blood volume for comprehensive clinical chemistry measurement, so 17 analytes were assessed per animal. Our preliminary analysis in SD rats found that GGT levels in undiluted serum were either outside the quantitation range or undetectable, in agreement with previous reports [16]; thus, GGT was omitted. Table >1Grouping of clinical chemistry analytes in miceGroupSexNumber of AnimalsClinical chemistry analytesG1Male10ALB, ALT, AST, BUN, CREA, TPFemale10G2Male10ALP, CK, GLU, TCHO, TG,Female10G3Male10Ca, Cl, IP, K, Na, TBILFemale10 Grouping of clinical chemistry analytes in mice For euthanasia, animals were placed in an acrylic chamber and anesthetized via inhalation of 5% isoflurane. After confirming the absence of reflex response, blood was collected from the posterior vena cava. Collected blood was transferred into serum separation tube (BD) and left at room temperature for a minimum 30 min, followed by centrifugation at 3000 rpm for 10 min. The resulting serum was stored frozen (-70 °C) until analyzed. The samples were thawed only once before analysis. Analyses were completed within 30 days following sample collection. Clinical chemistry analysis All clinical chemistry analyses were conducted using an automatic clinical chemistry analyzer, TBA 120 FR (TOSHIBA, Japan), which utilized BioAnalyzer software (version V400*R001, TOSHIBA, Japan). The reagents, quality control (QC) materials, and calibrators were sourced from FUJIFILM Wako Pure Chemical Co. (Osaka, Japan), Canon Medical System Co. (Otawara, Japan), and Autochem (SHINHWA MEDI Co., Ltd. Hanam, Korea). Detailed descriptions of methodology and the analytical measurement range (AMR) for each clinical chemistry analytes can be found in Supplementary Table 1. To assess the analytical performance of the instrument, calibration and QC procedure for each analyte were conducted and confirmed before sample analysis. Manual sample dilutions were performed using Eppendorf pipettes (Reference®2, Eppendorf, Germany); routine inspection and annual calibration were performed in accordance with the manufacturer’s recommendations. To reduce potential operator-related bias, a single operator was responsible for diluting serum samples with sterile physiological saline (Daehan Pharmaceutical Industries Co., Ltd. Korea), as per equipment manufacturer recommendations, as well as carrying out the analyses. Serum samples underwent dilution as factors of 2.5, 5, 7.5, and 10. Diluted and undiluted samples were analyzed in the same analytical run to minimize inter-run variability. Statistical analysis For each analyte, the measured values obtained from diluted samples were multiplied by the respective dilution factor to generate corrected values, which were then compared to results from undiluted serum. Calculations for each analyte, including mean, standard deviation (SD), 2SD, standard deviation index (SDI), and bias (inaccuracy), were performed using Excel (Microsoft Office 2016, Redmond, WA, USA). Linear regression analysis was performed using GraphPad Prism (GraphPad Software, Ver 8.0.1, San Diego, USA), with the dilution factor as the independent variable and the corrected values as the dependent variable. Accuracy is expressed as percent recovery which is calculated relative to the values measured in undiluted samples. The acceptable limit for the accuracy analysis was ≤ 20% [7, 9]. For Bland–Altman analysis, the mean difference and 95% limits of agreement between the values of the dilute and undilute samples were calculated according to the method proposed by Bland et al. [2]. Statistical analyses consisted of a one-way analysis of variance, followed by either the Bonferroni test or the Dunnett T3 test (SPSS, Ver12.0.0, SPSS Inc., USA). Statistical significance was determined at a p-value < 0.05. Results Sample selection To determine how sample dilution impact clinical chemistry analysis, serum samples were diluted at factors of 2.5, 5, 7.5, and 10. Initially, we evaluated whether results from both diluted and undiluted samples remained within the AMR. One analytical value from an undiluted serum sample and between one and three analytical values from diluted samples were outside the AMR, leading to their exclusion from statistical analysis (Supplementary Table 2). Diluted samples containing over 9 our-of-range values were nonetheless included in the analysis. Specifically, CREA levels for 5-, 7.5-, and 10-fold diluted samples from both mice and rats were below the lower AMR limit, while a majority of Cl values in 7.5- and 10-fold diluted mouse samples, and in 5- and 10-fold diluted rat samples, were above the upper AMR limit. All such values were included in the analysis. Comparison of mean analytical value between diluted and undiluted samples First, we compared the mean vales of diluted samples with those of undiluted samples. The mean values and standard deviations for both undiluted and diluted samples are summarized in Table 2 and 3 and Supplementary Table 3. In male mice, statistically significant differences from the undiluted samples were observed in 10 analytes at a 5-fold dilution (the highest number), 5 analytes at a 7.5-fold dilution, and 4 analytes at a 10-fold dilution, whereas ALB, Cl, and Na displayed significant differences at a 2.5-fold dilution. In female mice, significant differences were identified in 12 analytes at a 5-fold dilution (the highest number), 5 analytes at a 7.5-fold dilution, and 5 analytes at a 10-fold dilution, while Cl, and Na, and TBIL exhibited differences at a 2.5-fold dilution. In rats, the number of analytes with significant differences increased with higher dilution factors, totaling 3, 9, 10, and 15 analytes at 2.5-fold, 5-fold, 7.5-fold, and 10-fold dilutions, respectively. Table 2Clinical chemistry analysis of diluted and undiluted serum in male mice Clinical chemistry analysis of diluted and undiluted serum in male mice Table 3Biochemical analysis of diluted and undiluted serum in ratsYellow-highlighted figures indicate that the measured values for all diluted samples were outside of the AMRRed figures indicate that the measured values of at least one diluted sample was outside of the AMRAbbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total proteina Values are presented as mean ± standard deviation (n = 10). Values are significantly different from the undiluted samples: * p < 0.05, **p < 0.01 Biochemical analysis of diluted and undiluted serum in rats Yellow-highlighted figures indicate that the measured values for all diluted samples were outside of the AMR Red figures indicate that the measured values of at least one diluted sample was outside of the AMR Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total protein a Values are presented as mean ± standard deviation (n = 10). Values are significantly different from the undiluted samples: * p < 0.05, **p < 0.01 Evaluation of dilution linearity and analytical performance To validate the dilution effect, we performed linear regression, recovery, and Bland–Altman analysis. Slope and y-intercept data from linear regression analysis were shown in Tables 4 and 5 and Supplementary Table 4. Based on testing of slopes significantly different from zero, proportional bias associated with increasing dilution factors was observed in 14 analytes in rats (excluding BUN, K, and TCHO) and in 4 analytes in male and female mice (Cl, CREA, Na, and TBIL). To evaluate the dilution accuracy, recovery percentage was calculated. Cl, Na, and TBIL values of diluted samples fell all out of acceptable limit in both mice and rats (Figs. 1 and 2 and Supplementary Fig. 1). Male and female CREA also exceeded the acceptable limit at 7.5- and 10-fold dilution. Bland–Altman analysis was used to assess the agreement between measurements obtained from undiluted samples and those of diluted samples. X axis shows the average of the two values of undiluted and diluted samples at each dilution factor and Y axis presents the difference between these two values. As shown in Figs. 3 and 4 and Supplementary Fig. 2, the values of the 2.5-fold diluted samples were mostly clustered around zero, indicating a little difference from the undiluted sample, while those of 10-fold diluted samples were distributed far from zero, suggesting a significant difference from undiluted samples. Although the limits of agreement are commonly used to evaluate agreement between two measurements, we additionally adopted TEa values because it was difficult to determine whether agreement interval was appropriate or excessively wide. The differences were outside of the TEa values at all dilution factors for Cl, Na, and TBIL. Table 4Slope and y-intercept data from linear regression analysis of clinical chemistry analytes in male miceAnalyteSlope ± SEy-intercept ± SESlope significantly non-zero aALB-0.02312±0.053073.617±0.3250.6925ALP-4.275±7.988566.8±48.920.6297ALT-0.122±0.428731.32±2.6250.7945AST-0.084±0.65453.43±4.0050.9059BUN-0.01902±0.300928.86±1.8430.9536Ca-0.05316±0.166110.74±1.0170.7699CK-1.768±1.855143.3±11.360.4108Cl146.3±7.53346.28±46.130.0003CREA-0.02849±0.0041510.3277±0.025420.0063GLU-0.5324±3.094236±18.950.8743IP0.05136±0.158511.19±0.97030.7671K-0.00364±0.1119.049±0.67980.9759Na147.7±7.76284.48±47.530.0003TBIL0.09644±0.0058420.1559±0.035770.0005TCHO-0.242±1.12285.44±6.8730.8431TG-0.8516±0.637553.87±3.9040.2739TP-0.01084±0.077365.688±0.47370.8974Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, NS not significant, SE standard error, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total proteina A p values < 0.05 for the test of slope ≠ 0 indicates that the regression slop is significantly different from zero. This suggests that the values of the diluted samples change across dilution levels and implies the presence of proportionate bias as the dilution factor increases Slope and y-intercept data from linear regression analysis of clinical chemistry analytes in male mice Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, NS not significant, SE standard error, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total protein a A p values < 0.05 for the test of slope ≠ 0 indicates that the regression slop is significantly different from zero. This suggests that the values of the diluted samples change across dilution levels and implies the presence of proportionate bias as the dilution factor increases Table 5Slope and y-intercept data from linear regression analysis of clinical chemistry analytes in ratsAnalyteSlope ± SEy-intercept ± SESlope significantly non-zero aALB0.1009±0.0057124.749±0.034980.0004ALP30.48±3.1331141±19.180.0023ALT1.363±0.0994955.51±0.60930.0008AST2.892±0.2718118.3±1.6650.0018BUN0.251±0.102822.29±0.62940.0924Ca0.2707±0.0377112.62±0.23090.0056CK3.656±0.5255282.3±3.2180.0061Cl142.4±6.92551.29±42.40.0003CREA-0.00995±0.001110.4285±0.0067980.0029GLU3.895±0.416173.9±2.5470.0026IP0.2064±0.0280111.45±0.17150.0052K0.4616±0.27439.06±1.680.191Na147.6±7.19397.36±44.050.0003TBIL0.09791±0.0065380.0252±0.040040.0006TCHO-0.252±1.31181.34±8.0310.8599TG0.9844±0.169374.28±1.0370.0101TP0.1682±0.014027.029±0.085880.0012Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, NS not significant, SE standard error, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total proteina A p values < 0.05 for the test of slope ≠ 0 indicates that the regression slop is significantly different from zero. This suggests that the values of the diluted samples change across dilution levels and implies the presence of proportionate bias as the dilution factor increases Slope and y-intercept data from linear regression analysis of clinical chemistry analytes in rats Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, NS not significant, SE standard error, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total protein a A p values < 0.05 for the test of slope ≠ 0 indicates that the regression slop is significantly different from zero. This suggests that the values of the diluted samples change across dilution levels and implies the presence of proportionate bias as the dilution factor increases Recovery percentage for specific analytes in male mice at dilution factors of 2.5 ‒ 10. The dashed line indicates the acceptable limit (80% ‒ 120%) (n = 10) Recovery percentage for specific analytes in rats at dilution factors of 2.5 ‒ 10. The dashed line indicates the acceptable limit (80% ‒ 120%) (n = 20) Plot of differences between diluted samples and undiluted samples versus the average of these two values in male mice. The average (X axis) represents the mean of the two values obtained from the diluted and undiluted samples at each dilution factor. The difference (Y axis) represents the difference between these two values. The dashed line indicates the mean difference, the short-to-long line indicates the limits of agreement (mean difference ± 1.96 standard deviation of difference), and the red line indicates the allowable total error (TEa) (n = 10)Plot of differences between diluted samples and undiluted samples Plot of differences between diluted samples and undiluted samples versus the average of these two values in rats. The average (X axis) represents the mean of the two values obtained from the diluted and undiluted samples at each dilution factor. The difference (Y axis) represents the difference between these two values. The dashed line indicates the mean difference, the short-to-long line indicates the limits of agreement (mean difference ± 1.96 standard deviation of difference), and the red line indicates the allowable total error (TEa) (n = 10) Comparative analysis of bias and allowable error limits in diluted versus undiluted samples Next, to assess the equivalence of diluted sample values to the undiluted samples, bias, 2SD, and SDI were evaluated. Bias was defined as the difference between measured mean and target mean, calculated as ((Meanmeausred – Meantarget) / Meantarget) x 100 [13, 17]. As shown in Tables 5 and 6, TBIL, Na, and Cl demonstrated substantial bias at all dilution factors. For the 2SD evaluation, values for 2.5-fold diluted samples generally remained within the 2SD range, except for Cl, Na, and T-BIL (Figs. 5 and 6 and Supplementary Fig. 3). These results suggest that the diluted values of TBIL, Na, and Cl deviated from undilute values due to dilution, indicating significant measurement errors. Table 6Analysis of mean bias (%) in mice and ratsAnalyteMale miceFemale miceRats2.5-fold5-fold7.5-fold10-fold2.5-fold5-fold7.5-fold10-fold2.5-fold5-fold7.5-fold10-foldALB9.9424.263.34-5.479.8930.055.42-7.016.8712.7316.7621.95ALP4.5821.45-0.06-7.754.5921.36-1.732.024.5812.9315.4522.02ALT2.5021.30-3.34-3.080.2816.04-2.188.063.6811.8816.1023.72AST3.5620.38-0.32-0.234.3219.021.472.333.8513.2015.9724.31BUN1.8916.27-2.66-1.111.0614.73-2.80-2.75-3.550.921.107.39Ca-0.4010.75-18.922.29-0.9910.81-14.83-0.962.419.1011.6921.81CK0.7015.99-6.05-14.973.4816.85-7.60-7.424.248.558.9714.30Cl213.99564.35926.141285.31209.52548.28901.831250.61211.83597.45920.731285.40CREA-11.46-26.02-73.38-87.40-10.55-16.38-61.59-88.24-2.24-11.94-14.49-23.44GLU2.9021.440.98-2.463.6121.37-1.58-4.293.3311.7813.8622.71IP5.8321.58-4.9812.374.9120.630.229.432.339.6610.6318.44K2.6218.32-0.971.093.3018.52-7.863.544.4612.8963.4828.83Na156.95418.54685.21954.19157.56418.73685.01950.59151.30427.52660.31920.44TBIL89.09251.25398.91580.86126.98368.42622.41882.98166.13513.95845.191199.88TCHO5.3722.302.31-2.653.7625.71-5.830.433.1110.4413.8823.56TG1.8212.68-9.83-14.620.6013.21-14.17-21.59-0.511.725.1111.83TP4.4523.051.20-0.845.5721.50-2.89-14.112.6211.8516.4022.76Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total protein Analysis of mean bias (%) in mice and rats Abbreviations: ALB albumin, ALP alkaline phosphatase, ALT alanine aminotransferase, AST aspartate aminotransferase, BUN blood urea nitrogen, Ca calcium, CK creatine phosphokinase, Cl chloride, CREA creatinine, GLU glucose, IP inorganic phosphorus, K potassium, Na sodium, TBIL total bilirubin, TCHO Total cholesterol, TG Triglyceride, TP total protein Allowable total error (TEa) and 2 standard deviation (2SD) for specific analytes in male mice at dilution factors of 2.5 ‒ 10. TEa values were derived from the ASVCP and CLIA guidelines. The red line indicates the TEa, the dotted black line represents 2SD, the hollow circles denote the mean, and the solid I-shaped bars show standard deviations (n = 10) Allowable total error (TEa) and 2 standard deviation (2SD) for specific analytes in rats at dilution factors of 2.5 ‒ 10. TEa values were derived from the ASVCP and CLIA guidelines. The red line indicates the TEa, the dotted black line represents 2SD, the hollow circles denote the mean, and the solid I-shaped bars show standard deviations (n = 20) Allowable total error (TEa) defines clinically acceptable error thresholds, with limits derived from the American Society for Veterinary Clinical Pathology (ASVCP) guidelines and Clinical Laboratory Improvement Amendments (CLIA) requirements (Supplementary Table 5) [13, 30]. TEa limits were typically comparable to or narrower than the 2SD ranges illustrated in Figs. 5 and 6 and Supplementary Fig. 3. Analytes with mean bias within TEa limits are listed in Table 7. Table 7Analytes within allowable total error at each dilution factorDilution FactorSpeciesNo. of AnalytesBiochemistry Analytes2.5Male mouse13ALP, ALT, AST, BUN, Ca, CK,a CREA, GLU, IP, K, TCHO, TG, TPFemale mouse14ALB, ALP, ALT, AST, BUN, Ca, CK, CREA, GLU, IP, K, TCHO, TG, TPRat14ALB, ALP, ALT, AST, BUN, Ca, CK, CREA, GLU, IP, K, TCHO, TG, TP5Male mouse5ALP, ALT, AST, CK, TGFemale mouse6ALP, ALT, AST, CK, CREA, TGRat11ALP, ALT, AST, BUN, Ca, CK, CREA, GLU, IP, TCHO, TG7.5Male mouse12 ALB, ALP, ALT, AST, BUN, CK, GLU, IP, K, TCHO, TG, TP Female mouse12 ALB, ALP, ALT, AST, BUN, CK, GLU, IP, K, TCHO, TG, TP Rat9ALP, ALT, AST, CK, CREA, GLU, IP, TCHO, TG10Male mouse13ALB, ALP, ALT, AST, BUN, Ca, CK, GLU, IP, K, TCHO, TG, TPFemale mouse11ALB, ALP, ALT, AST, BUN, Ca, CK, GLU, IP, K, TCHORat6ALP, ALT, AST, CK, IP, TG,a Underlined analytes indicate that the absolute value of SDI is less than 1 Analytes within allowable total error at each dilution factor a Underlined analytes indicate that the absolute value of SDI is less than 1 SDI is an additional performance metric, calculated as (lab mean ‒ group mean)/group SD [3]. In this study, the mean of diluted samples was used as lab mean and the mean of undiluted samples as group mean. Among 17 analytes, 11, 12, and 13 analytes in male mice, female mice, and rats, respectively, showed SDI values less than 1 (Table 7). Discussion Non-clinical trials conducted during the development of new drugs adhere to guidelines established by national regulatory authorities and international organizations, including the Organisation for Economic Co-operation and Development (OECD), European Medicines Agency (EMA), and United States Food and Drug Administration (FDA) [20, 26]. These guidelines outline essential information and requirements for non-clinical studies, such as test methodologies, study durations, sequential review processes, animal selection standards, dosing protocols, clinical pathology and histopathology evaluation, and toxicokinetic assessment. Owing to the rigorous standards in regulatory studies, sample “dilution” is strictly prohibited [24]. In contrast, dilution may be implemented in exploratory research during early-stage drug discovery, where the primary aim is to assess drug efficacy and potential toxic effects in target organs [16]. Diluted samples are often utilized when working with small sample volumes, particularly in small-sized animals like mice or fetuses. In these scenarios, increasing the number of animals and dividing them into subgroups within the same experimental group enable the performance of both hematologic and biochemical analyses. When costly pharmaceuticals are involved, using a large cohort of animals presents additional challenges. Sample dilution has been proposed as a practical solution to address these limitations [12, 16]. Choosing an appropriate diluent is pivotal for accurate analysis of diluted serum samples. In this investigation, sterile saline solution was selected as the diluent, in accordance with recommendations from the respective analysis equipment manufacturers. However, certain researchers advise against the use of sterile saline solution as a diluent, citing potential measurement biases in electrolyte analysis [27]. In previous investigations involving diluted plasma pooled from mice, sterile water for injection was utilized [12]. According to Goyal et al., sterile water for injection, which is highly pure, lack of bacteria, and contains minimal minerals, does not interfere with reagents-analyte reactions [12]. Additionally, research on dilution-induced bias in rodent serum analyses used enzyme reagent and ultra-pure water for enzymes and all other analytes, respectively [16]. Comprehensive comparison warrant conducting experiments with various diluents using identical samples and the same analytical system to assess potential differences. In the present study, sterile saline solution as the diluent led to the expectation of significant changes in sodium (Na) and chloride (Cl) concentrations, as both ions are major constituents of saline. Indeed, the concentration values of these electrolytes increased proportionally with higher dilution factor, supporting the absence of manual pipetting errors during sample preparation. However, the selection of an appropriate alternative diluent for electrolytes is not straightforward, as there are limited studies addressing the effects of dilution on clinical chemistry analysis. Goyal et al. [12] evaluated multiple diluents, including normal saline, calibration solution, plasma, and water for injection, for mouse plasma but reported unsatisfactory results for electrolyte measurements. Jonhs et al. [16] diluted serum using ultrawater and encounter assay range error for electrolytes. Furthermore, manufacturers of analytical instruments do not recommend diluting samples for electrolyte analysis [14]. These findings reveal that selecting a suitable alternative diluent for electrolytes is challenging. Dilution effects were validated with linear regression, accuracy, and Bland-Altman analysis. Linear regression analysis showed that the slopes for 13 analytes in mice and 3 analytes in rats did not significantly differ from zero, indicating that a significant proportional bias with increasing dilution factor was not observed in mice rather than in rats. To assess the agreement between two measurement methods, specifically whether dilution method can be used for clinical chemistry analysis, Bland-Altman analysis was used [2]. The limits of agreement were calculated using the mean difference and standard deviations of the differences between two measurements. However, an additional criterion is required to define whether the agreement interval is acceptably narrow. Giavarina, [11] proposed defining the limits of maximum acceptable differences based on biologically and analytically valid criteria. Therefore, TEa values were additionally applied in this study. For example, as shown in Figs. 3and 4, the differences at all dilution factors for all analytes, including Cl, Na, and TBIL, fell within the limits of agreement. However, based on TEa criteria, the differences of Cl, Na, and TBIL clearly exceeded the acceptable limits. This discrepancy is attributable to the fact that the limits of agreement are derived solely from the mean difference and the standard deviations of the differences. Taken together, Cl, Na, and TBIL values of diluted samples exhibited substantial bias across all dilution factors, as demonstrated by linear regression, accuracy, and Bland-Altman analysis. However, dilution at lower dilution factors did not have a significant effect. To verify the accuracy and reliability of data derived from diluted samples, statistical assessments were conducted and multiple acceptance criteria including bias, SDI, and TEa were evaluated, confirming the validity of results generated from diluted samples. TEa is a widely used quality concept that defines the acceptable analytical performance specifications for clinical purpose. The TEa limit represents the maximum permissible error for a test result to remain reliable and suitable for clinical application [13, 21]. TEa values were derived from the CLIA and the ASVCP guidelines [4, 13, 30] (Table 6). CLIA is a regulatory framework that applies to all facilities or sites in US handling human specimens for health assessment, diagnosis, prevention, or treatment of disease [5, 6]. It sets nationwide standards for clinical laboratory testing, aiming to guarantee the accuracy, reliability, and promptness of patient test results. CLIA further provides defined performance requirements for specific analytes, which are monitored through interlaboratory proficiency assessments or proficiency surveys. The ASVCP established a Quality Assurance and Laboratory Standards (QAS) Committee in 1996 and introduced performance guidelines for analytical methods in animal testing, building on CLIA standards [3, 8]. In this study, TEa criteria were selected in accordance with both CLIA and ASVCP. For analytes that were classified as not clinically relevant (NCR) at low concentrations by ASVCP criteria, CLIA standards were employed. In our study, mean bias met acceptable criteria at lower dilution factors; specifically, 14 analytes in 2.5-fold diluted samples fell within the TEa in both mice and rats. Due to issues related to the diluent, Na and Cl measurements in all diluted samples exceeded the TEa thresholds. CREA, results from 5-, 7.5-, and 10-fold diluted samples were below the minimum TEa threshold, while TBIL values in all diluted samples surpassed the upper TEa threshold. ALP, ALT, AST, CK, and TG (except for female mice at 10-fold dilution) results were consistently within TEa limit across mice and rats. calculation of the SDI represents another approach to assess the comparability of test results. The majority of analytes that adhered to TEa also exhibited an SDI value less than 1. Male and female mice and rats were used to evaluate the impact of dilution in clinical chemistry analysis. No significant differences were observed between male and female mice across multiple analytical assessments, including mean analytical value between diluted and undiluted samples, linear regression assay, Bland–Altman analysis, mean bias, and number of analytes within the TEa criteria. These results suggest that sex-specific variability in dilution is minimal in mice under the present experimental conditions. Compared to mice, rats exhibited a more significant proportional bias with increasing dilution factor. As a result, most analytical measurements and the number of analytes meeting the TEa criteria were affected by the dilution factor. This species-specific variability may be attributed to matrix effects, such as differences in unintended analytes or interfering substances in the sample [9]. Although the use of diluted samples for analysis remains a subject of debate, it may be warranted in unavoidable situations. For instance, White et al., investigated hematological composition by analyzing 10-fold diluted blood from newborn and juvenile mice up to day 35, demonstrating that WBC differential counts underwent dynamical changes during early development. Such hematological data are crucial for elucidating the pathogenesis of various diseases and may also serve as valuable screening tools [31]. In studies assessing toxicity or efficacy in fetuses, neonates, and juvenile animals, clinical biochemistry analysis employing diluted samples is frequently unavoidable. Conclusions In the present study, clinical chemical analyses were conducted using 2.5-, 5-, 7.5-, and 10-fold-diluted serum and compared to results from undiluted serum samples. The data revealed that values from 2.5-fold diluted samples were largely consistent with those of undiluted serum, thereby affirming their reliability. Statistical analysis indicated that no significant differences between the means of the 2.5-fold diluted and undiluted samples, except for Cl, Na, ALB, and TBIL. Evaluations of bias, 2SD, SDI, and TEa suggested that a 2.5-fold dilution is generally applicable for clinical chemistry analysis for several analytes excluding TBIL, Na, and Cl. In addition, this dilution factor demonstrated consistent recovery and overall performance across analytes, supporting its suitability as an optimal dilution factor. Nevertheless, the present study assessed the effects of serum dilution solely in rodent samples, utilizing a single analytical instrument and software, and employing only one type of diluent (saline solution). These limitations may restrict the generalizability of the findings to other analytical platforms, diluents, or species. However, this study may provide preliminary insights for studies involving rodent models or small-volume samples such as fetuses or young animals. Research on the effects of dilution in clinical chemistry, particularly in samples from laboratory animals, remains limited. To better understand dilution-related variability, further studies using various analytical platforms, alternate diluent, or different laboratory animal species are required. Supplementary Information Supplementary Material 1. Acknowledgements Not applicable. Abbreviations Albumin Alkaline phosphatase Alanine aminotransferase Analytical measurement range Aspartate aminotransferase Blood urea nitrogen Creatine phosphokinase Creatinine Glucose Inorganic phosphorus Not clinically relevant Quality control Sprague Dawley Standard deviations Standard deviation index Total bilirubin Total cholesterol Allowable total error Triglyceride Total protein Authors’ contributions K. Ha: Conceptualization, Data curation, Formal analysis, Writing – original draft. J. Cho: Data curation, Formal analysis, Writing – original draft. K. Kim: Funding acquisition, Writing – review & editing. J. Song: Data curation, Formal analysis, Funding acquisition, Writing – original draft, Writing – review & editing. Funding This study was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education of Korea (NRF-2021R1F1A1061858) and the Korea Institute of Toxicology grant funded by the Ministry of science and ICT (Project number: KK-2507-01). Data availability Not applicable. Declarations Ethics approval and consent to participate All experimental procedures involving mice and rats received approval from the Institutional Animal Care and Use Committee of the Korea Institute of Toxicology (Approval no.: IAC-23-01-0026-0030, 2211-0012) and were carried out in a compliance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Kil-Woong Ha and Jeonghee Cho contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement Not applicable.