Comparative Evaluation of the Effects of Serum Separator Tubes and Transport Systems on Hormone and Tumor Marker Measurements.

This study included 57 participants (31 men and 26 women) with a median age of 44 yr (range, 22–68 yr). Among them, 10 had hyperlipidemia, nine had hypertension, two had hypothyroidism, two were receiving levothyroxine following total thyroidectomy for thyroid carcinoma, one had benign prostatic hyperplasia, and one was 19-weeks pregnant. LH and FSH levels in the pregnant participant were below the lower limit of quantification (LLoQ: both 0.3 mIU/mL) across all six tubes (A–F) and were, therefore, excluded from analysis. Estradiol levels in four postmenopausal women were undetectable (LLoQ: 5.0 pg/mL) and, therefore, excluded. In the pregnant participant, the estradiol level exceeded the upper limit of quantification (ULoQ: 3,000 pg/mL) and was excluded because it did not represent the general distribution. Hence, 52 samples were analyzed for estradiol. Total and free PSA were undetectable (LLoQ: both 0.01 ng/mL) in the 26 female participants; hence, these were analyzed only in the 31 male samples. CA 19-9 was undetectable (LLoQ: 0.2 U/mL) in five participants, leaving 52 samples for analysis. In the comparisons of PTS-transported samples (A–C), only estradiol exhibited clinically important differences among the 20 measurands. Estradiol showed clinically important differences between Ampulab and Vacutainer (A vs. B; average % bias of 5.43%) and between Vacutainer and Vacuette (B vs. C; –5.23%) ( Table 2 ), but not between Ampulab and Vacuette (A vs. C). In the comparisons of conveyor-transported samples (D–F), only estradiol exhibited clinically significant and important differences among the 20 measurands. Estradiol showed clinically important differences between Ampulab and Vacutainer (D vs. E; 5.22%) and between Ampulab and Vacuette (D vs. F; –4.30%), and a clinically significant difference between Vacutainer and Vacuette (E vs. F; –8.21%) ( Table 3 ). Across all tube comparisons, P-B Rs ranged from 0.936 to 1.000, revealing strong correlations among the three tubes. The 95% CIs of P-B slopes and intercepts included 1 and 0, respectively, for most measurands, except for estradiol, insulin, PTH, and CA 15-3 ( Supplemental Data Tables S1 and S2 ). Bland–Altman analysis confirmed that most differences were homogeneously distributed around the mean within the 95% CI, indicating high comparability among the three tubes ( Supplemental Data Fig. S1 ). In comparisons between PTS and conveyor transport, no clinically significant differences were observed for Ampulab (A vs. D) and Vacutainer (B vs. E) tubes. For Vacuette (C vs. F), only insulin showed a clinically important difference (–6.83%) ( Table 4 ). P-B Rs ranged from 0.929 to 1.000, revealing strong correlations between PTS and conveyor transport. The 95% CIs of P-B slopes and intercepts included 1 and 0, respectively, for most measurands, except for insulin and PTH ( Supplemental Data Table S3 ). Bland–Altman analysis confirmed that most differences were evenly distributed around the mean within 95% CI, indicating high comparability between the two transport methods ( Supplemental Data Fig. S2 ). Gross hemolysis was observed in 28.1% (16/57) of Ampulab tubes, 7.0% (4/57) of Vacutainer tubes, and 59.6% (34/57) of Vacuette tubes. The median values of HI, AST, K, and LDH were the lowest in Vacutainer tubes ( Fig. 1 ). Vacutainer HI and K values were significantly lower than those in Ampulab tubes, and HI, AST, and LDH values were significantly lower than those in Vacuette tubes. Finally, we assessed the proportions of samples exceeding the HI thresholds ( Fig. 2 ). The proportions exceeding the HI threshold of AST did not significantly differ between Ampulab and Vacutainer (A vs. B, 8.8% vs. 3.5%, P =0.375). However, the proportions exceeding the HI thresholds of K and LDH were significantly higher in Ampulab than in Vacutainer (21.1% vs. 8.8%, P =0.039; and 35.1% vs. 10.5%, P<0.001, respectively). Between Ampulab and Vacuette (A vs. C) and Vacutainer and Vacuette (B vs. C), the proportions exceeding the HI thresholds of AST, K, and LDH were significantly higher in Vacuette.

From January to February 2025, blood samples were collected from 57 adult volunteers (aged >19 yr) recruited through the hospital notice board of Samsung Medical Center, Seoul, Korea. Both healthy individuals and those with medical histories were included. To eliminate postural effects, all participants rested in a seated position for 30 min before phlebotomy. An experienced phlebotomist collected blood via venipuncture in the antecubital fossa using the BD Vacutainer blood collection system ( BD , Sunnyvale, CA, USA). From each participant, 30 mL whole blood was collected and distributed equally into six tubes: two Ampulab SSTs (Ampulab; Soyagreentec, Seoul, Korea), two Vacutainer SST II Advance Tubes (Vacutainer; BD), and two VACUETTE CAT SSTs (Vacuette; Greiner Bio-One, Kremsmünster, Austria). The order of blood draw was randomized among the three SST types. The characteristics of the SSTs are summarized in Table 1 . All tubes had the same size (13×100 mm) and sample volume (5 mL), but a different separator gel composition. All samples were obtained through a single venipuncture to minimize biological and sampling variability and to allow paired assessments of tube- and transport-specific effects. This study was approved by the Institutional Review Board of our hospital (approval No. 2024-11-068), and written informed consent was obtained from all participants. Pairs of identical tubes were randomly assigned to either PTS or conveyor transport, resulting in six sample categories: 1) collected in an Ampulab tube and transported via the PTS (A), 2) collected in a Vacutainer tube and transported via the PTS (B), 3) collected in a Vacuette tube and transported via the PTS (C), 4) collected in an Ampulab tube and transported via the conveyor system (D); 5) collected in a Vacutainer tube and transported via the conveyor system (E), and 6) collected in a Vacuette tube and transported via the conveyor system (F). For carrier-free PTS transport, we used a high-speed AirLab system (PazKorea, Hanam-si, Korea) that connects the outpatient phlebotomy clinic (1st floor) to the main laboratory (basement level 2). With a 70-m transport path and an average speed of 10 m/s, samples arrive in approximately 7 sec. The transport tube has an outer diameter of 25 mm and an inner diameter of 20 mm, making it suitable for direct SST transfer. For conveyor transport, we used a conveyor system (Hyundai Movex, Seoul, Korea) that transfers containers from the urine collection unit, adjacent to the phlebotomy clinic, to the main laboratory via conveyor belts. Although slower than the AirLab system (average speed, 0.5 m/s), the conveyor can accommodate large-volume samples and heavy items weighing up to 50 kg. Upon arrival in the main laboratory, samples were allowed to clot for 30 min and then centrifuged using a Kubota 5930 instrument (Kubota, Tokyo, Japan) at 25°C and 2,260× g for 6 min. These conditions slightly differed from the manufacturer’s recommendations due to technical issues, but the same conditions were applied to all samples ( Table 1 ). Twenty laboratory analytes, including 12 hormones and eight tumor markers, were measured. Seventeen analytes, including total triiodothyronine (T3), total thyroxine (T4), free T4, thyroid-stimulating hormone (TSH), luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, estradiol, insulin, C-peptide, alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), total prostate specific antigen (PSA), free PSA, human epididymis protein 4 (HE4), CA 125, and CA 19-9, were measured using a Cobas 8000 e801 analyzer (Roche Diagnostics, Basel, Switzerland). Testosterone, parathyroid hormone (PTH), and CA 15-3 were measured using a UniCel DXI 800 Access Immunoassay System (Beckman Coulter, Brea, CA, USA). To minimize analytical variation, all analytes were measured using identical reagent and calibrator lots. Comparative evaluations were performed according to the CLSI guidelines EP09c and GP34-A [ 19 , 20 ]. Differential effects of the three types of tubes were assessed by comparing results from A, B, and C (PTS-transported samples) with those from D, E, and F (conveyor-transported samples). Differential effects of the two transport methods were evaluated by comparing results from A and D (samples collected in Ampulab tubes) with those from B and E (samples collected in Vacutainer tubes) and C and F (samples collected in Vacuette tubes). Hemolysis was evaluated exclusively in PTS-transported samples (A–C), as none of the conveyor-transported samples (D–F) showed gross hemolysis. The conveyor system is the routine low-impact transport system used in our hospital. The AirLab system, which was under validation, is more likely to induce hemolysis due to higher mechanical stress. The hemolysis degree was assessed by comparing the rate of gross hemolysis, hemolytic index (HI), and levels of three hemolysis-sensitive analytes (AST, potassium [K], and lactate dehydrogenase [LDH]) among the three tubes. HI, AST, K, and LDH levels were measured using a Cobas 8000 c702 analyzer (Roche Diagnostics). Because hemolysis-sensitive analytes may exhibit clinically significant changes only when the HI exceeds specific thresholds, an additional probabilistic analysis of tube hemolysis was performed. The proportions of samples exceeding the manufacturer-established HI thresholds for these analytes (AST, 40 mg Hb/dL; K, 20 mg Hb/dL; and LDH, 15 mg Hb/dL, expressed as free Hb concentrations) were compared among the three tubes. The Shapiro–Wilk test and normal probability plots were used to assess the normality of differences in test results. Normally distributed data were analyzed using the paired Student’s t -test, whereas non-normal data were analyzed using the Wilcoxon signed-rank test. Passing–Bablok (P-B) regression was performed for each measurand comparison, except for the HI and hemolysis-sensitive measurands. P-B slopes and intercepts, along with their 95% confidence intervals (CIs) and P-B correlation coefficients (Rs), were calculated. In pairwise comparisons, the latter group served as the reference. For example, in A vs. B, average percentage (%) bias was calculated as (A–B)/B×100 (%), and P-B regression was fitted with B on the x-axis and A on the y-axis. Proportions of samples exceeding HI thresholds were compared using McNemar’s test. Statistical significance was defined as P<0.05. All statistical analyses were performed using R software v.4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). Clinical significance was determined using the optimal and desirable bias criteria from the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) biological variation database [ 21 ]. For analytes lacking EFLM data, the Westgard biological variation database was used [ 22 ]. An analyte was considered clinically significant when its average % bias exceeded the desirable bias, and clinically important when it exceeded the optimal bias but not the desirable bias. Bland–Altman plots were used to evaluate differences in the 20 analytes among the three types of tubes and between the two transport methods.

We, to our knowledge, for the first time, evaluated the combined effects of two major pre-analytical factors, blood collection tubes and sample transport methods, on hormone and tumor marker assays. Blood collection tube components, including the tube walls, rubber stoppers, lubricants, surfactants, separator gels, and clot activators, can interact with analytes and affect laboratory test results [ 23 – 24 ]. Therefore, the performance of novel blood collection tubes is validated through comparison with FDA-approved predicate devices, often BD Vacutainer and Greiner Bio-One Vacuette, as these are well-established brands commonly used in clinical laboratories [ 7 , 25 – 29 ]. We utilized these tubes as comparators to evaluate the performance of the Ampulab tube, recently introduced to the Korean domestic market. As no validation data exist for Ampulab, our study provides essential data for laboratory directors and quality assurance personnel considering its implementation. In tube comparisons, estradiol levels were higher in Ampulab than in Vacutainer for both PTS- and conveyor-transported samples (A vs. B and D vs. E). Similarly, estradiol levels were higher in Vacuette than in Vacutainer for both transport methods (C vs. B and F vs. E). These findings suggest that Ampulab and Vacuette may yield elevated estradiol results compared with those yielded by Vacutainer, potentially misdiagnosing hyper-estrogenic conditions such as endometriosis, polycystic ovarian syndrome, or ovarian cancer [ 30 ]. However, this interpretation warrants caution, as BD Vacutainer—although designated as the reference because it is routinely used in our hospital and widely adopted in validation studies—may not necessarily represent the true concentration; a potential negative bias in Vacutainer cannot be excluded. Therefore, laboratories that use SSTs from different brands should verify tube-specific estradiol differences and communicate them to clinicians. The clinically significant and important differences in estradiol concentrations among the tubes are likely attributable to interactions with the separator gels. Hydrophobic molecules partition into the polymer-based separator gels of SSTs; steroid hormones, with their hydrophobic carbon ring-based structures and low serum concentrations, are especially susceptible to such partitioning. In particular, SSTs can artificially reduce estradiol levels via gel adsorption when compared with plain tubes [ 13 , 26 , 31 ]. Differences in gel composition among SST brands likely influence the extent of adsorption, explaining the tube-to-tube variability of estradiol levels observed in our study. Yan et al . [ 29 ] reported clinically significant estradiol differences between BD Vacutainer rapid serum tubes and SSTs on the Abbott Architect and Siemens Centaur platforms (average % biases of 6.5% and 8.0%, respectively), whereas a clinically important estradiol difference (5.2%) was observed on the Roche Modular platform. Borai et al . [ 28 ] found clinically significant estradiol differences between V-PRO and BD Vacutainer SSTs on the Abbott Architect i 2000 analyzer (11.7%). Our transport-stratified results further suggest that gel adsorption effect may vary according to transport dynamics. PTS transport, with shorter transit times and higher acceleration, may reduce effective gel–serum contact and attenuate adsorption-related estradiol bias (e.g., Vacutainer–Vacuette bias: –5.23% with PTS vs. –8.21% with conveyor; Ampulab–Vacuette bias: –0.6% with PTS vs. –4.30% with conveyor). In contrast, Ampulab–Vacutainer comparisons showed similar estradiol bias between PTS and conveyor transport (5.43% vs. 5.22%), suggesting that transport-related modulation of gel adsorption may be more pronounced in Vacuette. Considering that tube-intrinsic gel chemistry and the transport method may jointly influence estradiol results, laboratory-specific validations are necessary. PTSs have improved pre-analytical workflow efficiency by reducing manual workload and turnaround time [ 32 ]. The AirLab system, a high-speed, carrier-free PTS implemented in our hospital, represents a significant advancement in laboratory automation. Its carrier-free design enables direct sample loading and streamlining transportation through immediate tube input. The system offers excellent scalability through integration with blood collection units for auto-loading and connectivity to automatic tube sorters for auto-unloading. Its flexible and lightweight transport tubes enable convenient setup and seamless adaptation to existing infrastructure. Despite these advantages, the risk of in-vitro hemolysis in PTS-transported samples persists. The hemolysis-sensitive parameters such as AST, K, and LDH are elevated in PTS-transported samples [ 33 – 35 ]. Therefore, laboratories must evaluate the impact of PTS on sample integrity before installation. We evaluated the hemolysis effect of the AirLab system. The gross hemolysis rates in PTS-transported samples based on visual inspection were unexpectedly high (28.1%, 7.0%, and 59.6% for Ampulab, Vacutainer, and Vacuette, respectively). These rates were comparable to the proportions of samples exceeding the HI threshold of K (21.1%, 8.8%, and 56.1%, respectively), supporting the validity of our visual assessment as hemolysis becomes visually detectable at an HI value of approximately 20 mg Hb/dL [ 36 ]. While hemolysis rates reported in previous PTS studies substantially vary depending on the HI threshold criteria and PTS type [ 33 , 37 ], one study using an HI threshold of 30 mg Hb/dL reported a hemolysis rate (8.1%) similar to that observed in Vacutainer samples in our study (7.0%) [ 33 ]. We also evaluated the influence of AirLab transport on the 20 analytes. Insulin levels were consistently lower in PTS-transported samples across all three tubes. A clinically important insulin bias was observed in Vacuette (C vs. F, –6.83%), whereas differences were smaller in Ampulab (A vs. D, –3.25%) and Vacutainer (B vs. E, –2.47%). These differences are likely attributable to PTS-induced hemolysis, which interferes with insulin levels through the release of insulin-degrading enzymes from erythrocytes [ 38 – 40 ]. This interpretation is supported by our hemolysis assessment, which revealed the highest degree of hemolysis in Vacuette, and the lowest in Vacutainer, consistent with the observed gradient in insulin bias. Karin et al . [ 38 ] reported clinically significant differences in insulin levels when serum free Hb exceeded 1.0 g/L, but not below 0.6 g/L. Applying these thresholds, we found that two samples with HI >100 mg Hb/dL (one Ampulab and one Vacuette) showed clinically significant biases (–47.3% and –27.7%), and among eight samples with HI 60–100 mg Hb/dL (one Ampulab and seven Vacuette), seven showed clinically significant biases (–34.4% to –10.7%) (insulin panels in Supplemental Data Fig. S2 ). These findings support the relevance of the proposed hemolysis thresholds for insulin and highlight the need for monitoring HI values when analyzing insulin in hemolysis-prone samples. Next, we examined the effects of PTS-induced hemolysis on other analytes. Among the aforementioned 10 PTS-transported samples with HI values >60 mg Hb/dL (two Ampulab and eight Vacuette), clinically significant biases were observed for total T3 (four samples), total T4 (three samples), free T4 (four samples), estradiol (one sample), PTH (two samples), and C-peptide (three samples), when compared with their paired conveyor-transported samples. However, these deviations were inconsistent in both magnitude and direction, unlike the uniformly negative insulin biases. These findings suggest that insulin is uniquely vulnerable to hemolysis-induced degradation, whereas the biases in other analytes likely reflect sporadic variability rather than systematic hemolysis effects. Several limitations should be acknowledged. First, we evaluated only SSTs, not other commonly used tube types such as plain serum or heparinized plasma tubes. Comparing tubes with fundamentally different compositions may have yielded more diverse and clinically meaningful insights into tube-specific effects. Second, total and free PSA did not meet the minimum 40-sample requirement for method comparison defined in the CLSI EP09c guideline [ 20 ]. Both PSA analytes were below the LLoQ in the 26 female participants, leaving only 31 male samples for analysis, limiting their statistical robustness. Third, our centrifugation protocol did not fully conform to the manufacturer’s recommendations. This minor difference may have affected the results. However, because the same protocol was applied to all samples, our study still provides a reliable analysis of the combined effects of two preanalytical factors. Lastly, only a few samples exceeded the medical decision limits. Despite efforts to include patients with hormonal abnormalities or malignancies, most analyte concentrations were within the reference ranges. This limited concentration coverage may have underestimated bias at high medical decision points; therefore, future studies including large numbers of high-concentration patient samples are warranted. In conclusion, the Ampulab tube showed equivalent performance to that of two comparator tubes for 12 hormones and eight tumor markers across the evaluated concentration ranges, with no clinically significant differences. However, estradiol showed clinically important differences in Ampulab versus the other tubes, as well as clinically significant differences between the two comparator tubes, likely attributable to differences in gel adsorption. The AirLab system performed similarly to the conveyor transport system, with no clinically significant differences. Insulin exhibited a clinically important negative bias in AirLab-transported Vacuette tubes, due to PTS-induced hemolysis. This study highlights the influence of these two major pre-analytical factors on hormone and tumor marker testing. The use of validated tubes and transport methods minimizes pre-analytical variability and maintains the clinical utility of laboratory test results.

Clinical laboratory tests involve complex, multi-step processes from sample collection to result reporting [ 1 ]. These processes can be divided into pre-analytical, analytical, and post-analytical phases, with the pre-analytical phase accounting for the majority of total laboratory errors [ 2 – 4 ]. Among various pre-analytical variables, blood collection tubes and sample transport methods influence sample integrity and analyte stability, contributing to variability in laboratory results. While the effects of these two variables on routine chemistry and hematology tests have been investigated [ 5 – 9 ], few studies have focused on hormones and tumor markers. Given their essential role in diagnosing and monitoring endocrine and oncological diseases [ 10 , 11 ], controlling pre-analytical conditions ensures analytical accuracy and clinical reliability. Blood collection tubes have evolved considerably, with the incorporation of various additives such as anticoagulants, clot activators, and polymeric separator gels to support accurate and efficient laboratory testing [ 12 ]. Serum separator tubes (SSTs) use thixotropic polymer gels to separate serum from cellular components upon centrifugation. These gels can interact with and adsorb hydrophobic analytes, such as steroid hormones, reducing serum concentrations [ 13 ]. Tube additives may leach into samples and interfere with immunoassays [ 14 , 15 ]. Therefore, blood collection tubes must be validated when measuring sensitive analytes such as hormones and tumor markers. Sample transport is another determinant of pre-analytical quality. Manual transport minimizes mechanical stress but is labor-intensive. Conveyor systems offer a low-impact alternative; however, their implementation may be constrained by existing infrastructure [ 16 ]. Pneumatic tube systems (PTSs) enable rapid, automated delivery but are associated with hemolysis [ 17 ], resulting in the release of proteolytic enzymes from erythrocytes, which can affect peptide hormone quantification [ 18 ]. Hence, transport systems must be carefully validated, considering the biochemical properties of the target analytes. The combined effects of blood collection tube type and transport method on hormones and tumor marker measurements have not been investigated, although these pre-analytical combinations vary across laboratories. Therefore, we evaluated the combined effects of using different SSTs from three manufacturers (Soyagreentec, Becton Dickinson (BD), and Greiner Bio-One) and two different transport systems, PTS and conveyor system, on 20 hormone and tumor marker analytes.