BIOPRESERVATION AND BIOBANKING Volume 11, Number 6, 2013 ª Mary Ann Liebert, Inc. DOI: 10.1089/bio.2013.0050
A Novel Multiplex–Protein Array for Serum Diagnostics of Colorectal Cancer: Impact of Pre-analytical Storage Conditions Stefanie Bu¨nger,1 Katja Klempt-Giessing,1 Vicki Toner,2 Maria Kelly,2 Stephen P FitzGerald,2 Hermann Brenner,3 Ferdinand von Eggeling,4 and Jens K Habermann1
Introduction: Biomarker discovery studies seldom report on pre-analytical effects. We used a novel multiplex protein biochip for colorectal cancer screening to investigate effects of different storage temperatures and repeated freeze-thaw cycles. Methods: This biochip, composed of CEA, IL-8, VEGF, M-CSF, S100A11, C3adesArg, CD26, and CRP, was applied to twenty highly standardized preserved serum samples. Results: Aliquot comparison of long-term storage at - 80C (n = 20) versus - 170C (n = 20) did not show significant differences for any of the eight markers. In contrast, three freeze-thaw cycles (3 · 20 aliquots) detected changes in the serum level for all markers ( p < 0.05) but S100A11 and CD26: levels of CEA, IL-8, C3adesArg, and CRP increased, while VEGF and M-CSF levels decreased. However, applying diagnostic thresholds for CEA, IL8, and CRP revealed that freeze-thaw cycles did not affect diagnostic performance. In contrast, analysis of samples stored at - 80C compared to - 170C failed to detect one out of three detectable malignancies. Conclusion: We conclude that three freeze-thaw cycles modulated serum marker levels significantly, but do not compromise biochip diagnostic performance. For our marker panel, serum preservation at - 80C seems comparable to - 170C; however, storage at - 80C could lead to misdiagnosis. Our findings emphasize the need for standardized sample collection, processing, storage, and reporting.
Introduction
T
he proteome of body fluids is highly dynamic and varies according to the pathophysiological state of the individual.1 Changes in serum protein composition can reflect distinct disease stages, disease progression, and response to therapy.2,3 Quantitative and qualitative serum protein profiling has therefore become a perceived goal of clinical proteomics, in order to identify reliable diagnostic and prognostic biomarkers in clinical oncology. While biomarker discovery has been the main focus for the last decade, less attention has been paid to how sample handling can impact the serum proteome and hence influence biomarker discovery. Identifying biomarkers derived from human serum is particularly appealing because blood can be obtained with a high level of patient compliance and acceptability. Thus, blood-based analyses have become a paradigm for biomarker discovery. Notably, 40% to 90% of all errors within the entire diagnostic process occur in the pre-
analytical process due to the lack of standardized procedures for patient preparation, specimen acquisition, handling, and storage.4–6 Therefore, one of the main issues is the establishment of standard operating procedures (SOPs) for collection, preparation, and storage in order to control preanalytic variation.7,8 This is even more important since a caveat of biomarker discovery relies on the in vivo preservation of proteins within the sample and the consistency of protein changes due to a certain disease in between samples. Many pre-analytical factors have been identified as potential sources of bias for subsequent biomarker discovery, which has generated various proposals for standardized reporting, for example, according to the MIAPE (Minimum Information About a Proteomics Experiment), REMARK (Reporting recommendations for tumour MARKer prognostic studies), BRISQ (Biospecimen Reporting for Improved Study Quality), and HUPO plasma proteome project specimen collection and handling guidelines.9–12 However, there is still no consensus on storage temperatures and the number of freeze-thaw
1
Section for Translational Surgical Oncology and Biobanking, Department of Surgery, University of Lu¨beck, Lu¨beck, Germany. Randox Laboratories GmbH, Wu¨lfrath, Germany. 3 Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ), Heidelberg, Germany. 4 Core Unit Chip Application, Institute of Human Genetics, Jena University Hospital, Jena, Germany. On behalf of the BMBF-Consortium ‘‘Colorectal Cancer Screening Chip.’’ 2
379
¨ NGER ET AL. BU
380 cycles for samples used for biomarker studies. Hence, the value of biomarker studies without reporting on preanalytical conditions remains questionable. Our consortium previously generated a novel multiplex– protein biochip array for serum diagnostics of colon cancer. Application of this biochip showed significant differences between colon cancer patients and healthy controls for serum levels of carcinoembryonic antigen (CEA), interleukin 8 (IL8), vascular endothelial growth factor (VEGF), calgizzarin (S100A11), macrophage colony-stimulating factor (M-CSF), complement component 3a (C3adesArg), protein 26 (CD26), and C-reactive protein (CRP). The most promising marker combinations were CEA + IL-8 reaching 47% sensitivity at 86% specificity and CEA + CRP with 39% sensitivity at 86% specificity.13 These results were obtained using a collection of serum samples collected and processed under highly standardized conditions; for instance, storage within 30 minutes after phlebotomy in vapor phase nitrogen (below - 170C) until testing to provide optimal preservation of the in vivo serum proteome composition. Since clinical practice seldom allows for immediate nitrogen preservation of serum samples, we investigated two main aspects of potential sampling bias: effects of repeated freeze-thaw cycles and effects of long-term-storage of serum at two different temperatures (- 80C versus - 170C).
Materials and Methods Study group and sampling This study comprised 40 samples that were randomly selected from 3500 serum specimens collected at the University Medical Center Schleswig-Holstein, Campus Lu¨beck, Germany, between 2007 and 2010. This serum collection is part of the centralized biobank Interdisciplinary Center for Biobanking—Lu¨beck (ICB-L), the Surgical Center for Translational Oncology—Lu¨beck (SCTO-L), University of Lu¨beck, and the German Cancer Aid (DKH e.V.) funded network North German Tumorbank of Colorectal Cancer (ColoNet, #108446). Serum samples were collected adhering to the guidelines of the local ethical review board (Medical University of Lu¨beck, #07-124) and according to strict standard operation procedures (see below). Serum samples were obtained after oncologic resection. The study group comprised 20 patients with histologically confirmed and successfully resected colorectal cancer (16 men and 4 women, mean age 69 years). Detailed clinical data from the study group are summarized in Table 1. All venous blood samples were obtained using serum gel-monovettes (#01.1602, Sarstedt AG & Co, Nu¨mbrecht, Germany) and centrifuged at 1550 g for 10 min at 4C to separate the serum. Two aliquots from the same patient serum were obtained, resulting in a total sample number of 40 serum aliquots. Aliquots of the same sample were both stored at - 80C (Forma - 86C Freezer #8605, Thermo Fisher Scientific, Schwerte, Germany) and - 170C (N2) (Espace 331, 330L, Air Liquide, Du¨sseldorf, Germany), respectively, within 30 min after venous puncture. Since the freezer’s internal temperature might be heterogeneous or inconsistent as described by Su et al.,14 we validated the performance using freezer mapping to detect critical temperature gradients. Accordingly, samples were stored in the stable temperature zone (central to lower back side) of the freezer. Cryo-samples were stored in the nitrogen gas phase to avoid
Table 1. Clinical Data of Colorectal Cancer Patients from Whom Postoperative Serum Samples Were Analyzed Parameter Sex Age (years) UICC stage
T status (tumor size) N status (nodal status) M status (distant metastasis)
Value
Colorectal cancer patients (n = 20)
Female Male Range Mean 1 2 3 4 1 2 3 4 0 1 2 0 1
4 16 59–83 69 4 3 10 3 1 2 14 3 8 11 1 17 3
minimal temperature variances in sample storage due to varying filling levels or contamination among samples. Furthermore, an access authorization system and a freezer/ cryo SOP assured a stable temperature and filling level, which was cross-checked every day manually by trained personnel for temperature recording. Additionally, freezer and cryotank were connected to a systematically controlled alarm system.
Storage condition experiments Two aliquots from the same patient serum were obtained from 20 patients and stored as described above (- 80C and - 170C). Mean storage duration of the samples was 33.65 months (median: 29.1 months, range: 22.2–55.7 months). First, both aliquots were thawed on ice (at 4C for 5 min) and analyzed in parallel in order to compare effects of storage temperatures [- 80C (FT0) versus -170C (NT0)] on biomarker serum levels. Second, aliquots stored in nitrogen (-170C) were re-frozen at - 170C until thawing and multiplex analysis 24 h later. This freeze-thaw cycle was repeated once more after another 24 h, so that three analyses for each aliquot stored in nitrogen (NT0, NT1, NT2) could be performed. In summary, four measurements were available for each patient serum (FT0, NT0, NT1, NT2, Fig. 1).
Determination of biomarker serum levels An established multiplex biochip platform designed for the simultaneous quantitative detection of the serum markers IL-8, CEA, VEGF, M-CSF, S100A11, C3adesArg, CD26, and CRP was manufactured as previously described.15,16 The biochips were based on simultaneous chemiluminescent sandwich immunoassays using the Evidence Investigator analyser (Randox Laboratories Ltd., Crumlin, UK). This system allows handling of up to 54 samples (6 · 9 wells) in one biochip carrier. Samples were thawed on ice before multiplex assessment. The biochip was manufactured and
ARRAY AND PRE-ANALYTICAL CONDITIONS IN CRC
381
FIG. 1. Study design workflow. Two serum sample aliquots from the same peripheral blood tube per patient were obtained, resulting in a total number of 40 serum aliquots. Aliquots of the same serum samples were stored in the freezer at - 80C (F) and in nitrogen at - 170C (N), respectively, within 30 min after venous puncture. In the first study, both aliquots were thawed on ice and analyzed at time point zero (T0) in parallel in order to compare the effects on biomarker serum levels between - 80C (FT0) and N2 storage (NT0). Second, aliquots from N2 were re-frozen at - 170C until thawing and multiplex-analysis 24 h later. The freezethaw cycle was repeated on two additional occasions, generating three freeze-thaw cycles (NT0, NT1, NT2).
validated according to Randox Laboratories’ manufacturing guidelines and procedures.17,18 The biochip undergoes 100% quality control assessment after manufacture, avoiding the need for technical replicates for patient samples. The validation followed a series of approved standard operating procedures, including determination of sensitivity, accuracy, and precision of each assay as described before.13 The inclusion of calibrators and controls provides confidence in the inter-assay results reported. The analyte concentration present in the sample was calculated automatically using calibration curves (Evidence Investigator Software version 1.4). The analyzer routinely assesses the quality of assay performance as described by FitzGerald et al.17 The detailed protocol for chip incubation and processing as well as assay ranges and accuracy were described previously.13
Statistical analysis Using SPSS Statistics version 19 (IBM Corporation, Somers, NY), the serum levels of the individual markers in samples from different storage conditions were evaluated with respect to median levels and percentiles. Additionally, for the comparison of serum levels after freeze-thaw cycles, the changes were given as percentages, each with respect to the level of the unthawed aliquot. Nonparametric tests (Wilcoxon test and Friedmans analysis) were applied to compare median serum levels between paired-sample aliquots at different storage conditions. First, serum levels of aliquots stored at - 80C freezer (FT0) and in nitrogen at - 170C (NT0) were compared to examine effects of general long-term storage temperature. Second, effects of freezethaw cycles on serum levels were compared in aliquots without any prior thawing (NT0,) after one (NT1) and after two (NT2) freeze-thaw cycles.
Results Here we report on the effect of pre-analytical storage conditions on serum protein levels by applying a multiplex biochip assay comprised of CEA, IL-8, VEGF, M-CSF,
S100A11, C3adesArg, CD26, and CRP to 20 highly standardized preserved serum samples (Fig. 2).
Long-term storage at - 80 C versus - 170 C Serum aliquots from 20 colorectal cancer patients were stored long-term either at - 80C or at - 170C. Without any previous thawing cycle, these samples were analyzed for eight serum markers simultaneously (Table 2a). Serum levels for CD26 and CRP were lower in samples stored at - 80C compared to those stored at - 170C (CD26: median 595 vs. 629 ng/mL, CRP: median 1849 vs. 1893 ng/mL). In contrast, IL-8 (median 11.6 vs. 10.8 pg/mL), VEGF (median 111.2 vs. 106.7 pg/mL), S100A11 (median 19.0 vs. 18.3 ng/mL), and C3adesArg (716 vs. 517 ng/mL) showed higher levels in samples stored at - 80C compared to those stored at - 170C. However, none of these changes reached statistical significance. CEA (mean 0.87 vs. 0.86 ng/mL) and M-CSF (mean 6.4 vs. 6.4 pg/mL) showed almost identical serum levels for both storage temperatures.
Effects of freeze-thaw cycles In a second step, effects of freeze-thaw cycles on serum levels were evaluated. Serum levels of samples stored at - 170C (N2) without any previous thawing (NT0) were compared to levels of the same samples after one (NT1) and two (NT2) freeze-thaw cycles. In this analysis, median serum levels were significantly different between freeze-thaw cycles for six of the eight markers. The following markers showed an increase over the freeze-thaw cycles: CEA from 0.86 ng/ mL at NT0 to 1.24 ng/mL at NT2 ( p < 0.001), IL-8 from 10.8 to 13.4 pg/mL ( p < 0.001), C3adesArg from 517 to 731 ng/mL ( p < 0.001), and CRP from 1893 to 2041 ng/mL ( p = 0.043). In contrast, serum levels of VEGF decreased with increased freeze-thaw cycles from 106.7 to 97.7 pg/mL ( p < 0.0001). Also M-CSF showed significant differences in serum levels between all three groups (6.4, 5.5, and 6.3 pg/mL, p = 0.002), however, with a less pronounced decrease (Table 2b). Changes in serum levels for S100A11 (from 18.3 to 18.5 ng/
382
¨ NGER ET AL. BU
FIG. 2. Box plots for the eight serum biomarkers subjected to different storage parameters. Serum levels were measured after each storage condition in 20 postoperative serum samples from colorectal cancer patients. Samples from longterm storage at - 80C (FT0) vs. samples stored at - 170C (NT0) without thaw cycles or after first thaw cycle (NT1) and second thaw cycle (NT2).
mL, p = 0.116) and CD26 (629 to 594 ng/mL, p = 0.819) did not show statistical significance.
Effects of storage temperature and freeze-thaw cycles on diagnostic performance Our consortium recently developed a test algorithm using our biochips for early detection of colon cancer by CEA, IL-8, and CRP in a pilot study of 317 serum samples.13 A CEA threshold was established at 3.2 ng/mL, for IL-8 at 39.5 pg/ mL, and for CRP at 14,600 ng/mL. Based on these threshold levels, CEA showed a sensitivity of 26%, IL-8 of 22%, and CRP of 17%, all at 90% specificity.13 In order to evaluate the effects of storage temperatures and freeze-thaw cycles on the diagnostic performance of these three markers, we applied the previously determined thresholds to classify all samples as diseased (D) or healthy (H), depending on the preanalytical conditions (Table 3).
Comparison of samples stored at - 80C versus - 170C without thawing detected two patients’ CEA serum levels above the threshold in the - 80C group. In the - 170C group, one additional sample (total n = 3) was determined to be malignant by the CEA threshold. IL-8 and CRP threshold levels both detected the same patient sample as malignant independent of the storage temperature. Thus, despite no significant changes in the overall serum levels between aliquots stored at - 80C and - 170C, one of three patients (33%) would not have been detected as malignant using CEA levels, had serum been stored only at - 80C instead of storage at - 170C. In addition, the established thresholds were applied to the 20 patient samples that had undergone two freeze-thaw cycles after storage at - 170C. Interestingly, and in contrast to significant changes in the overall detection levels of CEA, IL8, and CRP, the modulations did not affect the diagnostic performance of the three markers. All patients positively
383
0.87 (0.5–2.2) 0.86 (0.4–2.2) = 0.601
- 80C (FT0) - 170C (NT0) Trend P value
0.86 (0.4–2.2) 1.14 (0.5–2.4) + 32.6 1.24 (0.6–2.5) + 44.2 > < 0.0001
- 170C (NT0) - 170C 1st cycle (NT1) Change (%) compared to NT0 - 170C 2nd cycle (NT2) Change (%) compared to NT0 Trend P value 10.8 (7.8–13.6) 11.8 (8.7–16.9) + 9.2 13.4 (10–16.5) + 24.1 > < 0.0001
IL-8 pg/mL 106.7 (77.6–163.2) 99.7 (66.8–139.6) - 6.6 97.7 (63.5–129.5) - 8.4 < < 0.0001
6.4 (5.1–7.7) 5.5 (4.4–7.1) - 14.1 6.3 (4.9–7.8) - 1.6 * 0.002
18.3 (12.8–23.5) 16.8 (12.1–23.3) - 8.2 18.5 (12.4–25.1) + 1.1 * 0.116
S100A11 ng/mL
Median serum level
19.0 (13.4–24.2) 18.3 (12.8–23.5) < 1.0
S100A11 ng/mL
M-CSF pg/mL
6.4 (4.3–7.7) 6.4 (5.1–7.7) = 0.765
M-CSF pg/mL
VEGF pg/mL
111.2 (64.7–175.8) 106.7 (77.6–163.2) < 0.823
VEGF pg/mL
Median serum level
517 (397–924) 591 (457–1067) + 14.3 731 (588–1437) + 41.4 > < 0.0001
C3adesArg ng/mL
716 (391–798) 517 (397–924) < 0.825
C3adesArg ng/mL
629 (496–682) 612 (510–699) - 2.7 594 (510–669) - 5.6 < 0.819
CD26 ng/mL
595 (506–665) 629 (496–682) > 0.086
CD26 ng/mL
1893 (1281–3355) 1916 (1394–3764) + 1.2 2041 (1304.5–3822) + 7.8 > 0,043
CRP ng/mL
1849 (1271–3518) 1893 (1281–3355) > 0.232
CRP ng/mL
a) Serum marker levels detected in samples (no thaw cycle) stored long-term at - 80C (FT0) vs. - 170C (NT0) and b) samples stored at - 170C (NT0) vs. same sample after 1st (NT1) vs. 2nd (NT2) thaw cycle. Trend over different storage conditions: > (increased serum level), < (decreased serum level), = (approx. constant serum level), * (no clear trend).
CEA ng/mL
11.6 (6.4–14.7) 10.8 (7.8–13.6) < 0.247
IL-8 pg/mL
Storage condition
b)
CEA ng/mL
Storage condition
a)
Table 2. Median Serum Level with 25th—75th Percentile of Eight Biomarkers in Postoperative Serum Samples of 20 Colorectal Cancer Patients at Different Storage Temperatures
¨ NGER ET AL. BU
384 Table 3. Number of Patients Identified by Defined Thresholds for CEA (3.2 ng/mL), IL8 (39.5 pg/mL), and CRP (14,600 ng/mL) as Diseased (D) or Healthy (H) Compared at Different Storage Temperatures Number of patients identified CEA
IL-8
CRP
Storage condition
D
H
D
H
D
H
FT0 (- 80C) NT0 (- 170C) NT1 (- 170C, 1st cycle) NT2 (- 170C, 2nd cycle)
2 3 3 3
18 17 17 17
1 1 1 1
19 19 19 19
1 1 1 1
19 19 19 19
Aliquots of long-term storage in the freezer at - 80C (FT0) vs. aliquots stored in N2 without freeze-thaw cycles (NT0) and after first thawing cycle (NT1) and second thawing cycle (NT2).
diagnosed in un-thawed samples were also detected after each freeze-thaw cycle (three patients were detected by CEA, one patient by IL-8, and one by CRP). Thus, the patient detected by IL-8 was also detected by CRP and CEA. Two patients were detected only by increased CEA levels of serum samples stored at - 170C with and without freeze-thaw cycles (Table 3). The low number of patients predicted to be malignant may be explained by the fact that the serum from the twenty cancer patients had been obtained post-operatively during follow-up, whereas in the original study the samples were pre-operative collections. Thus, serum levels of initial cancer markers may have decreased during therapy and recovery of the patients.
Discussion Many biomarker studies appear to be highly heterogeneous regarding sample quality management including collection, processing, and storage conditions. Awareness of the importance of pre-analytical factors is increasing and thorough knowledge of pre-analytical variables is essential to control their impact.8,15,19 This is the first study to report the effect of - 80C versus - 170C serum preservation and the effect of repeated freeze-thaw cycles on eight distinct serum biomarker levels assessed by a multiplex biochip array. Overall, long-term storage over a mean duration of 33.65 months at - 80C compared to - 170C did not influence serum levels of the eight biomarkers significantly. This contrasts with previous reports that demonstrated a clear advantage of plasma-based biomarker preservation at lower temperatures.12 However, marker stability during long-term storage is likely to be dependent on standard operating procedures (SOPs) such as rapid processing of serum samples before freezing. Our serum sample collection SOP from phlebotomy to freezing requires completion within 30 minutes. Prolonged periods before processing and freezing can impact the stability of markers at different storage temperatures.16,20 While low temperature freezing generally slows protein degradation, freeze-thaw cycles may lead to protein denaturation, aggregation, and other conformational changes that can cause modulation (e.g., resulting in a decreased or increased biomarker level). Importantly, degradation of biomarkers can be rapid and can vary between different
markers.21 The results of our study indicate that some proteins are more sensitive to freeze-thaw cycles than others. Serum levels of CEA, IL-8, VEGF, M-CSF, C3adesArg, and CRP appeared to be influenced by repeated freeze-thawing, whereas S100A11 and CD26 were not affected. Our results also showed that serum levels of different markers changed in different directions: CEA, IL-8, C3adesArg, and CRP increased with repeated freeze-thaw cycles, while VEGF decreased. In line with our results, Kisand et al.22 found that VEGF concentrations decreased after one freeze-thaw cycle from analysis of 36 serum samples. In contrast, Hetland et al.23 reported that VEGF serum levels, in contrast to plasma levels, were not significantly affected by up to nine freeze-thaw cycles ( - 80C). For CRP, Aziz et al.24 showed that serum levels were also unaffected by up to seven freezethaw cycles (- 70C). The differences among these studies, as well as our results, might be due to different antibodies, assays (enzyme immunoassay vs. multiplex array), storage temperatures (- 80C vs. - 170C), and/or other preanalytical sample handling factors applied. For CEA, IL-8, M-CSF, S100A11, CD26, and C3adesArg, no comparable marker stability studies are available that focus on freezethaw cycles. Based on our results, freeze-thaw cycles of serum samples can change distinct protein levels significantly. Thus, potential differences in serum levels between disease stages may not be evident if marker levels are substantially degraded by repeated freeze-thaw cycles, particularly if sample groups to be compared are not exposed to identical conditions. Furthermore, even if mean serum levels of distinct biomarkers are not significantly affected (e.g., in our study comparing - 80C and - 170C storage), individual class prediction can still be hampered leading to false diagnosis. Thus, sample quality is as important as quality of diagnostic performance itself for diagnostic studies. In clinical routine, repeated freeze/thaw cycles are common and thus, apparent biomarker changes might rely more on physical/ chemical changes in the sample than on physiological changes in the patients. This might hamper the reliability of established clinical tests, as well as newly implemented biomarker tests, for which marker stability has not yet been explored. While our data suggest that samples stored at - 80C should be readily diagnosable with the biochip, our findings also have highlighted the validity of routine sample storage at - 170C to avoid misdiagnosis of individual cases. In the event that sample freeze-thawing has occurred, a complete record of freeze-thaw cycles should be maintained and taken into consideration for downstream applications. Furthermore, in the future it would be desirable for a biochip design to include, for example, degradation markers that indicate low quality of the samples tested. The risk-management of the potential inaccuracy of diagnostic tests might be optimized by adding such degradation or quality marker proteins to each diagnostic assay in order to trace patient samples’ quality. Unfortunately, such markers are not yet implemented in clinical routine and require more research in order to be validated. However, SOPs for the pre-analytical phase should be applied permanently on all samples and not only for one special test. Given these results, we would like to emphasize the importance of detailed SOPs for sample quality management in biomarker discovery studies, as this will directly impact the
ARRAY AND PRE-ANALYTICAL CONDITIONS IN CRC biomarkers’ validity and clinical applicability. Several guidelines exist that can improve the development of these SOPs. The International Society for Biological and Environmental Repositories’ (ISBER) Biospecimen Science Working Group developed the SPREC reporting system (Standard PREanalytical Code), detailing the main pre-analytical factors affecting clinical fluid and solid biospecimens.25,26 Not only serum biomarkers are affected by pre-analytical bias, but also markers in plasma samples27 or viability of human cells for clinical application, for example, cell therapy.28 The ‘‘Biospecimen Reporting for Improved Study Quality’’ scheme (BRISQ) could provide a standardized approach for reporting such information.9 Additionally, several guidelines for publishing results in different areas already exist [STARD (diagnostics), REMARK (prognostics), CONSORT (clinical trials), STROBE (epidemiology), MIAME (microarray), and MIAPE (proteomics)].10,11,29–32 Besides storage temperature, storage duration should be taken into account at different storage temperatures as shown in several studies. Pieragostino et al.33 detected critical conditions for serum protein profiles depending on storage times and temperatures and demonstrated that upon a - 20C short term storage, characteristic degradation profiles occur. Furthermore, Schultz et al.34 demonstrated that urinary proteins are significantly underestimated after storage at - 20C compared to - 70C. Our current data are not influenced by storage duration, since samples at different storage temperatures (- 80C and - 170C) have undergone the same storage durations, as they were stored simultaneously. Our study has shown that three freeze-thaw cycles change serum marker levels significantly. Analyses of the eight serum markers evaluated in this study further suggest that serum preservation at - 80C is comparable to that at - 170C. However, storage at - 80C can lead to misdiagnosis, thus limiting diagnostic performance. Our findings clearly emphasize the importance of standardized sample collection, processing, storage, and reporting to enable the completion of essential large-scale, multicenter trials employing high-quality specimen cohorts.
385
2.
3.
4.
5. 6. 7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
Acknowledgments We would like to thank Elke Gheribi and Regina Kaatz for clinical sampling.
17.
Author Disclosure Statement
18.
The authors declare that they have no competing interests. The consortium ‘‘Colon Cancer Screening Chip’’ was generously supported by the German Federal Ministry of Education and Research (BMBF) within the Molecular Diagnostics funding scheme (Grants 01ES0720, 01ES0721, 01ES0722, and 01ES0723). The study was performed in collaboration with the ‘‘Surgical Center for Translational Oncology–Lu¨beck’’ (SCTO-L) and the ‘‘North German Tumorbank of Colorectal Cancer’’ network, the latter being generously supported by the German Cancer Aid Foundation (Dt. Krebshilfe e. V., Grant #108446).
19.
20.
21.
22.
References 1. Wilkins MR, Pasquali C, Appel RD, et al. From proteins to proteomes: Large scale protein identification by two-dimen-
23.
sional electrophoresis and amino acid analysis. Biotechnology (NY) 1996;14:61–65. Smith L, Lind MJ, Welham KJ, Cawkwell L. Cancer proteomics and its application to discovery of therapy response markers in human cancer. Cancer 2006;107:232–241. Kulasingam V, Diamandis EP. Strategies for discovering novel cancer biomarkers through utilization of emerging technologies. Nature Clin Pract Oncol 2008;5:588–599. Lippi G, Salvagno GL, Montagnana M, Guidi GC. Reliability of the thrombin-generation assay in frozen-thawed platelet-rich plasma. Clin Chem 2006;52:1827–1828. Carraro P, Plebani M. Errors in a stat laboratory: types and frequencies 10 years later. Clin Chem. Jul 2007;53(7):1338–1342. Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. Clin Chem 2002;48:691–698. Conrads TP, Hood BL, Veenstra TD. Sampling and analytical strategies for biomarker discovery using mass spectrometry. BioTech 2006;40:799–805. Ferguson RE, Hochstrasser DF, Banks RE. Impact of preanalytical variables on the analysis of biological fluids in proteomic studies. Proteomics Clin Applic 2007;1:739–746. Moore HM, Kelly AB, Jewell SD, et al. Biospecimen reporting for improved study quality (BRISQ). Cancer Cytopathol 2011;119: 92–101. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM. REporting recommendations for tumour MARKer prognostic studies (REMARK). Br J Cancer 2005;93:387–391. Taylor CF, Paton NW, Lilley KS, et al. The minimum information about a proteomics experiment (MIAPE). Nat Biotechnol 2007; 25:887–893. Rai AJ, Gelfand CA, Haywood BC, et al. HUPO Plasma Proteome Project specimen collection and handling: Towards the standardization of parameters for plasma proteome samples. Proteomics 2005;5:3262–3277. Bu¨nger S, Haug U, Kelly M, et al. A novel multiplex-protein array for serum diagnostics of colon cancer: A case-control study. BMC Cancer 2012;12:393. Su SC, Garbers S, Rieper TD, Toniolo P. Temperature variations in upright mechanical freezers. Cancer Epidemiol Biomarkers Prev 1996;5:139–140. Lombardi G, Lanteri P, Colombini A, Banfi G. Blood biochemical markers of bone turnover: Pre-analytical and technical aspects of sample collection and handling. Clin Chem Lab Med 2012;50:771–789. Albrethsen J, Bogebo R, Olsen J, Raskov H, Gammeltoft S. Preanalytical and analytical variation of surface-enhanced laser desorption-ionization time-of-flight mass spectrometry of human serum. Clin Chem Lab Med CCLM/FESCC 2006;44:1243–1252. Fitzgerald SP, McConnell RI, Huxley A. Simultaneous analysis of circulating human cytokines using a high-sensitivity cytokine biochip array. J Proteome Res 2008;7:450–455. Molloy RM, Mc Connell RI, Lamont JV, FitzGerald SP. Automation of biochip array technology for quality results. Clin Chem Lab Med 2005;43:1303–1313. Rai AJ, Vitzthum F. Effects of preanalytical variables on peptide and protein measurements in human serum and plasma: Implications for clinical proteomics. Expert Rev Proteomics 2006; 3:409–426. Brinc D, Chan MK, Venner AA, et al. Long-term stability of biochemical markers in pediatric serum specimens stored at - 80 degrees C: A CALIPER Substudy. Clin Bioch 2012;45:816–826. Goodwin RJ, Dungworth JC, Cobb SR, Pitt AR. Time-dependent evolution of tissue markers by MALDI-MS imaging. Proteomics 2008;8:3801–3808. Kisand K, Kerna I, Kumm J, Jonsson H, Tamm A. Impact of cryopreservation on serum concentration of matrix metalloproteinases (MMP)-7, TIMP-1, vascular growth factors (VEGF) and VEGF-R2 in Biobank samples. Clin Chem Lab Med 2011;49:229–235. Hetland ML, Christensen IJ, Lottenburger T, et al. Circulating VEGF as a biological marker in patients with rheumatoid ar-
386
24.
25.
26.
27.
28.
29. 30.
31.
thritis? Preanalytical and biological variability in healthy persons and in patients. Dis Markers 2008;24:1–10. Aziz N, Fahey JL, Detels R, Butch AW. Analytical performance of a highly sensitive C-reactive protein-based immunoassay and the effects of laboratory variables on levels of protein in blood. Clin Diag Lab Immunol 2003;10:652–657. Betsou F, Lehmann S, Ashton G, et al. Standard preanalytical coding for biospecimens: Defining the sample PREanalytical code. Cancer Epidemiol Biomarkers Prev 2010;19:1004–1011. Lehmann SG, Moore F, Ashton H, et al. Standard preanalytical coding for biospecimens: Review and implementation of the Sample PREanalytical Code (SPREC) Biopreserv Biobank 2012; 10:366–375. Paltiel L, Ronningen KS, Meltzer HM, Baker SV, Hoppin JA. Evaluation of freeze thaw cycles on stored plasma in the biobank of the Norwegian Mother and Child Cohort Study. Cell Preserv Technol 2008;6:223–230. Massie I, Selden C, Hodgson H, Fuller B. Storage temperatures for cold-chain delivery in cell therapy: A study of alginateencapsulated liver cell spheroids stored at - 80C or - 170C for up to 1 year. Tissue Engineer Part C, Methods. 2013;19:189–195. Bossuyt PM, Reitsma JB. The STARD initiative. Lancet 2003; 361:71. Brazma A, Hingamp P, Quackenbush J, et al. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001;29:365–371. Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: Updated guidelines for reporting parallel group randomised trials. J Clin Epidemiol 2010;63:834–840.
¨ NGER ET AL. BU 32. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344–349. 33. Pieragostino D, Petrucci F, Del Boccio P, et al. Pre-analytical factors in clinical proteomics investigations: Impact of ex vivo protein modifications for multiple sclerosis biomarker discovery. J Proteomics 2010;73:579–592. 34. Schultz CJ, Dalton RN, Turner C, Neil HA, Dunger DB. Freezing method affects the concentration and variability of urine proteins and the interpretation of data on microalbuminuria. The Oxford Regional Prospective Study Group. Diabetic Med J Br Diabetic Assoc 2000;17:7–14.
Address correspondence to: Jens K. Habermann, MD, PhD Section for Translational Surgical Oncology and Biobanking Department of Surgery University of Lu¨beck and University Medical Center Schleswig-Holstein Campus Lu¨beck Ratzeburger Allee 160 D-23538 Lu¨beck Germany E-mail: [email protected]
A Novel Multiplex–Protein Array for Serum Diagnostics of Colorectal Cancer: Impact of Pre-analytical Storage Conditions Stefanie Bu¨nger,1 Katja Klempt-Giessing,1 Vicki Toner,2 Maria Kelly,2 Stephen P FitzGerald,2 Hermann Brenner,3 Ferdinand von Eggeling,4 and Jens K Habermann1
Introduction: Biomarker discovery studies seldom report on pre-analytical effects. We used a novel multiplex protein biochip for colorectal cancer screening to investigate effects of different storage temperatures and repeated freeze-thaw cycles. Methods: This biochip, composed of CEA, IL-8, VEGF, M-CSF, S100A11, C3adesArg, CD26, and CRP, was applied to twenty highly standardized preserved serum samples. Results: Aliquot comparison of long-term storage at - 80C (n = 20) versus - 170C (n = 20) did not show significant differences for any of the eight markers. In contrast, three freeze-thaw cycles (3 · 20 aliquots) detected changes in the serum level for all markers ( p < 0.05) but S100A11 and CD26: levels of CEA, IL-8, C3adesArg, and CRP increased, while VEGF and M-CSF levels decreased. However, applying diagnostic thresholds for CEA, IL8, and CRP revealed that freeze-thaw cycles did not affect diagnostic performance. In contrast, analysis of samples stored at - 80C compared to - 170C failed to detect one out of three detectable malignancies. Conclusion: We conclude that three freeze-thaw cycles modulated serum marker levels significantly, but do not compromise biochip diagnostic performance. For our marker panel, serum preservation at - 80C seems comparable to - 170C; however, storage at - 80C could lead to misdiagnosis. Our findings emphasize the need for standardized sample collection, processing, storage, and reporting.
Introduction
T
he proteome of body fluids is highly dynamic and varies according to the pathophysiological state of the individual.1 Changes in serum protein composition can reflect distinct disease stages, disease progression, and response to therapy.2,3 Quantitative and qualitative serum protein profiling has therefore become a perceived goal of clinical proteomics, in order to identify reliable diagnostic and prognostic biomarkers in clinical oncology. While biomarker discovery has been the main focus for the last decade, less attention has been paid to how sample handling can impact the serum proteome and hence influence biomarker discovery. Identifying biomarkers derived from human serum is particularly appealing because blood can be obtained with a high level of patient compliance and acceptability. Thus, blood-based analyses have become a paradigm for biomarker discovery. Notably, 40% to 90% of all errors within the entire diagnostic process occur in the pre-
analytical process due to the lack of standardized procedures for patient preparation, specimen acquisition, handling, and storage.4–6 Therefore, one of the main issues is the establishment of standard operating procedures (SOPs) for collection, preparation, and storage in order to control preanalytic variation.7,8 This is even more important since a caveat of biomarker discovery relies on the in vivo preservation of proteins within the sample and the consistency of protein changes due to a certain disease in between samples. Many pre-analytical factors have been identified as potential sources of bias for subsequent biomarker discovery, which has generated various proposals for standardized reporting, for example, according to the MIAPE (Minimum Information About a Proteomics Experiment), REMARK (Reporting recommendations for tumour MARKer prognostic studies), BRISQ (Biospecimen Reporting for Improved Study Quality), and HUPO plasma proteome project specimen collection and handling guidelines.9–12 However, there is still no consensus on storage temperatures and the number of freeze-thaw
1
Section for Translational Surgical Oncology and Biobanking, Department of Surgery, University of Lu¨beck, Lu¨beck, Germany. Randox Laboratories GmbH, Wu¨lfrath, Germany. 3 Division of Clinical Epidemiology and Aging Research, German Cancer Research Center (DKFZ), Heidelberg, Germany. 4 Core Unit Chip Application, Institute of Human Genetics, Jena University Hospital, Jena, Germany. On behalf of the BMBF-Consortium ‘‘Colorectal Cancer Screening Chip.’’ 2
379
¨ NGER ET AL. BU
380 cycles for samples used for biomarker studies. Hence, the value of biomarker studies without reporting on preanalytical conditions remains questionable. Our consortium previously generated a novel multiplex– protein biochip array for serum diagnostics of colon cancer. Application of this biochip showed significant differences between colon cancer patients and healthy controls for serum levels of carcinoembryonic antigen (CEA), interleukin 8 (IL8), vascular endothelial growth factor (VEGF), calgizzarin (S100A11), macrophage colony-stimulating factor (M-CSF), complement component 3a (C3adesArg), protein 26 (CD26), and C-reactive protein (CRP). The most promising marker combinations were CEA + IL-8 reaching 47% sensitivity at 86% specificity and CEA + CRP with 39% sensitivity at 86% specificity.13 These results were obtained using a collection of serum samples collected and processed under highly standardized conditions; for instance, storage within 30 minutes after phlebotomy in vapor phase nitrogen (below - 170C) until testing to provide optimal preservation of the in vivo serum proteome composition. Since clinical practice seldom allows for immediate nitrogen preservation of serum samples, we investigated two main aspects of potential sampling bias: effects of repeated freeze-thaw cycles and effects of long-term-storage of serum at two different temperatures (- 80C versus - 170C).
Materials and Methods Study group and sampling This study comprised 40 samples that were randomly selected from 3500 serum specimens collected at the University Medical Center Schleswig-Holstein, Campus Lu¨beck, Germany, between 2007 and 2010. This serum collection is part of the centralized biobank Interdisciplinary Center for Biobanking—Lu¨beck (ICB-L), the Surgical Center for Translational Oncology—Lu¨beck (SCTO-L), University of Lu¨beck, and the German Cancer Aid (DKH e.V.) funded network North German Tumorbank of Colorectal Cancer (ColoNet, #108446). Serum samples were collected adhering to the guidelines of the local ethical review board (Medical University of Lu¨beck, #07-124) and according to strict standard operation procedures (see below). Serum samples were obtained after oncologic resection. The study group comprised 20 patients with histologically confirmed and successfully resected colorectal cancer (16 men and 4 women, mean age 69 years). Detailed clinical data from the study group are summarized in Table 1. All venous blood samples were obtained using serum gel-monovettes (#01.1602, Sarstedt AG & Co, Nu¨mbrecht, Germany) and centrifuged at 1550 g for 10 min at 4C to separate the serum. Two aliquots from the same patient serum were obtained, resulting in a total sample number of 40 serum aliquots. Aliquots of the same sample were both stored at - 80C (Forma - 86C Freezer #8605, Thermo Fisher Scientific, Schwerte, Germany) and - 170C (N2) (Espace 331, 330L, Air Liquide, Du¨sseldorf, Germany), respectively, within 30 min after venous puncture. Since the freezer’s internal temperature might be heterogeneous or inconsistent as described by Su et al.,14 we validated the performance using freezer mapping to detect critical temperature gradients. Accordingly, samples were stored in the stable temperature zone (central to lower back side) of the freezer. Cryo-samples were stored in the nitrogen gas phase to avoid
Table 1. Clinical Data of Colorectal Cancer Patients from Whom Postoperative Serum Samples Were Analyzed Parameter Sex Age (years) UICC stage
T status (tumor size) N status (nodal status) M status (distant metastasis)
Value
Colorectal cancer patients (n = 20)
Female Male Range Mean 1 2 3 4 1 2 3 4 0 1 2 0 1
4 16 59–83 69 4 3 10 3 1 2 14 3 8 11 1 17 3
minimal temperature variances in sample storage due to varying filling levels or contamination among samples. Furthermore, an access authorization system and a freezer/ cryo SOP assured a stable temperature and filling level, which was cross-checked every day manually by trained personnel for temperature recording. Additionally, freezer and cryotank were connected to a systematically controlled alarm system.
Storage condition experiments Two aliquots from the same patient serum were obtained from 20 patients and stored as described above (- 80C and - 170C). Mean storage duration of the samples was 33.65 months (median: 29.1 months, range: 22.2–55.7 months). First, both aliquots were thawed on ice (at 4C for 5 min) and analyzed in parallel in order to compare effects of storage temperatures [- 80C (FT0) versus -170C (NT0)] on biomarker serum levels. Second, aliquots stored in nitrogen (-170C) were re-frozen at - 170C until thawing and multiplex analysis 24 h later. This freeze-thaw cycle was repeated once more after another 24 h, so that three analyses for each aliquot stored in nitrogen (NT0, NT1, NT2) could be performed. In summary, four measurements were available for each patient serum (FT0, NT0, NT1, NT2, Fig. 1).
Determination of biomarker serum levels An established multiplex biochip platform designed for the simultaneous quantitative detection of the serum markers IL-8, CEA, VEGF, M-CSF, S100A11, C3adesArg, CD26, and CRP was manufactured as previously described.15,16 The biochips were based on simultaneous chemiluminescent sandwich immunoassays using the Evidence Investigator analyser (Randox Laboratories Ltd., Crumlin, UK). This system allows handling of up to 54 samples (6 · 9 wells) in one biochip carrier. Samples were thawed on ice before multiplex assessment. The biochip was manufactured and
ARRAY AND PRE-ANALYTICAL CONDITIONS IN CRC
381
FIG. 1. Study design workflow. Two serum sample aliquots from the same peripheral blood tube per patient were obtained, resulting in a total number of 40 serum aliquots. Aliquots of the same serum samples were stored in the freezer at - 80C (F) and in nitrogen at - 170C (N), respectively, within 30 min after venous puncture. In the first study, both aliquots were thawed on ice and analyzed at time point zero (T0) in parallel in order to compare the effects on biomarker serum levels between - 80C (FT0) and N2 storage (NT0). Second, aliquots from N2 were re-frozen at - 170C until thawing and multiplex-analysis 24 h later. The freezethaw cycle was repeated on two additional occasions, generating three freeze-thaw cycles (NT0, NT1, NT2).
validated according to Randox Laboratories’ manufacturing guidelines and procedures.17,18 The biochip undergoes 100% quality control assessment after manufacture, avoiding the need for technical replicates for patient samples. The validation followed a series of approved standard operating procedures, including determination of sensitivity, accuracy, and precision of each assay as described before.13 The inclusion of calibrators and controls provides confidence in the inter-assay results reported. The analyte concentration present in the sample was calculated automatically using calibration curves (Evidence Investigator Software version 1.4). The analyzer routinely assesses the quality of assay performance as described by FitzGerald et al.17 The detailed protocol for chip incubation and processing as well as assay ranges and accuracy were described previously.13
Statistical analysis Using SPSS Statistics version 19 (IBM Corporation, Somers, NY), the serum levels of the individual markers in samples from different storage conditions were evaluated with respect to median levels and percentiles. Additionally, for the comparison of serum levels after freeze-thaw cycles, the changes were given as percentages, each with respect to the level of the unthawed aliquot. Nonparametric tests (Wilcoxon test and Friedmans analysis) were applied to compare median serum levels between paired-sample aliquots at different storage conditions. First, serum levels of aliquots stored at - 80C freezer (FT0) and in nitrogen at - 170C (NT0) were compared to examine effects of general long-term storage temperature. Second, effects of freezethaw cycles on serum levels were compared in aliquots without any prior thawing (NT0,) after one (NT1) and after two (NT2) freeze-thaw cycles.
Results Here we report on the effect of pre-analytical storage conditions on serum protein levels by applying a multiplex biochip assay comprised of CEA, IL-8, VEGF, M-CSF,
S100A11, C3adesArg, CD26, and CRP to 20 highly standardized preserved serum samples (Fig. 2).
Long-term storage at - 80 C versus - 170 C Serum aliquots from 20 colorectal cancer patients were stored long-term either at - 80C or at - 170C. Without any previous thawing cycle, these samples were analyzed for eight serum markers simultaneously (Table 2a). Serum levels for CD26 and CRP were lower in samples stored at - 80C compared to those stored at - 170C (CD26: median 595 vs. 629 ng/mL, CRP: median 1849 vs. 1893 ng/mL). In contrast, IL-8 (median 11.6 vs. 10.8 pg/mL), VEGF (median 111.2 vs. 106.7 pg/mL), S100A11 (median 19.0 vs. 18.3 ng/mL), and C3adesArg (716 vs. 517 ng/mL) showed higher levels in samples stored at - 80C compared to those stored at - 170C. However, none of these changes reached statistical significance. CEA (mean 0.87 vs. 0.86 ng/mL) and M-CSF (mean 6.4 vs. 6.4 pg/mL) showed almost identical serum levels for both storage temperatures.
Effects of freeze-thaw cycles In a second step, effects of freeze-thaw cycles on serum levels were evaluated. Serum levels of samples stored at - 170C (N2) without any previous thawing (NT0) were compared to levels of the same samples after one (NT1) and two (NT2) freeze-thaw cycles. In this analysis, median serum levels were significantly different between freeze-thaw cycles for six of the eight markers. The following markers showed an increase over the freeze-thaw cycles: CEA from 0.86 ng/ mL at NT0 to 1.24 ng/mL at NT2 ( p < 0.001), IL-8 from 10.8 to 13.4 pg/mL ( p < 0.001), C3adesArg from 517 to 731 ng/mL ( p < 0.001), and CRP from 1893 to 2041 ng/mL ( p = 0.043). In contrast, serum levels of VEGF decreased with increased freeze-thaw cycles from 106.7 to 97.7 pg/mL ( p < 0.0001). Also M-CSF showed significant differences in serum levels between all three groups (6.4, 5.5, and 6.3 pg/mL, p = 0.002), however, with a less pronounced decrease (Table 2b). Changes in serum levels for S100A11 (from 18.3 to 18.5 ng/
382
¨ NGER ET AL. BU
FIG. 2. Box plots for the eight serum biomarkers subjected to different storage parameters. Serum levels were measured after each storage condition in 20 postoperative serum samples from colorectal cancer patients. Samples from longterm storage at - 80C (FT0) vs. samples stored at - 170C (NT0) without thaw cycles or after first thaw cycle (NT1) and second thaw cycle (NT2).
mL, p = 0.116) and CD26 (629 to 594 ng/mL, p = 0.819) did not show statistical significance.
Effects of storage temperature and freeze-thaw cycles on diagnostic performance Our consortium recently developed a test algorithm using our biochips for early detection of colon cancer by CEA, IL-8, and CRP in a pilot study of 317 serum samples.13 A CEA threshold was established at 3.2 ng/mL, for IL-8 at 39.5 pg/ mL, and for CRP at 14,600 ng/mL. Based on these threshold levels, CEA showed a sensitivity of 26%, IL-8 of 22%, and CRP of 17%, all at 90% specificity.13 In order to evaluate the effects of storage temperatures and freeze-thaw cycles on the diagnostic performance of these three markers, we applied the previously determined thresholds to classify all samples as diseased (D) or healthy (H), depending on the preanalytical conditions (Table 3).
Comparison of samples stored at - 80C versus - 170C without thawing detected two patients’ CEA serum levels above the threshold in the - 80C group. In the - 170C group, one additional sample (total n = 3) was determined to be malignant by the CEA threshold. IL-8 and CRP threshold levels both detected the same patient sample as malignant independent of the storage temperature. Thus, despite no significant changes in the overall serum levels between aliquots stored at - 80C and - 170C, one of three patients (33%) would not have been detected as malignant using CEA levels, had serum been stored only at - 80C instead of storage at - 170C. In addition, the established thresholds were applied to the 20 patient samples that had undergone two freeze-thaw cycles after storage at - 170C. Interestingly, and in contrast to significant changes in the overall detection levels of CEA, IL8, and CRP, the modulations did not affect the diagnostic performance of the three markers. All patients positively
383
0.87 (0.5–2.2) 0.86 (0.4–2.2) = 0.601
- 80C (FT0) - 170C (NT0) Trend P value
0.86 (0.4–2.2) 1.14 (0.5–2.4) + 32.6 1.24 (0.6–2.5) + 44.2 > < 0.0001
- 170C (NT0) - 170C 1st cycle (NT1) Change (%) compared to NT0 - 170C 2nd cycle (NT2) Change (%) compared to NT0 Trend P value 10.8 (7.8–13.6) 11.8 (8.7–16.9) + 9.2 13.4 (10–16.5) + 24.1 > < 0.0001
IL-8 pg/mL 106.7 (77.6–163.2) 99.7 (66.8–139.6) - 6.6 97.7 (63.5–129.5) - 8.4 < < 0.0001
6.4 (5.1–7.7) 5.5 (4.4–7.1) - 14.1 6.3 (4.9–7.8) - 1.6 * 0.002
18.3 (12.8–23.5) 16.8 (12.1–23.3) - 8.2 18.5 (12.4–25.1) + 1.1 * 0.116
S100A11 ng/mL
Median serum level
19.0 (13.4–24.2) 18.3 (12.8–23.5) < 1.0
S100A11 ng/mL
M-CSF pg/mL
6.4 (4.3–7.7) 6.4 (5.1–7.7) = 0.765
M-CSF pg/mL
VEGF pg/mL
111.2 (64.7–175.8) 106.7 (77.6–163.2) < 0.823
VEGF pg/mL
Median serum level
517 (397–924) 591 (457–1067) + 14.3 731 (588–1437) + 41.4 > < 0.0001
C3adesArg ng/mL
716 (391–798) 517 (397–924) < 0.825
C3adesArg ng/mL
629 (496–682) 612 (510–699) - 2.7 594 (510–669) - 5.6 < 0.819
CD26 ng/mL
595 (506–665) 629 (496–682) > 0.086
CD26 ng/mL
1893 (1281–3355) 1916 (1394–3764) + 1.2 2041 (1304.5–3822) + 7.8 > 0,043
CRP ng/mL
1849 (1271–3518) 1893 (1281–3355) > 0.232
CRP ng/mL
a) Serum marker levels detected in samples (no thaw cycle) stored long-term at - 80C (FT0) vs. - 170C (NT0) and b) samples stored at - 170C (NT0) vs. same sample after 1st (NT1) vs. 2nd (NT2) thaw cycle. Trend over different storage conditions: > (increased serum level), < (decreased serum level), = (approx. constant serum level), * (no clear trend).
CEA ng/mL
11.6 (6.4–14.7) 10.8 (7.8–13.6) < 0.247
IL-8 pg/mL
Storage condition
b)
CEA ng/mL
Storage condition
a)
Table 2. Median Serum Level with 25th—75th Percentile of Eight Biomarkers in Postoperative Serum Samples of 20 Colorectal Cancer Patients at Different Storage Temperatures
¨ NGER ET AL. BU
384 Table 3. Number of Patients Identified by Defined Thresholds for CEA (3.2 ng/mL), IL8 (39.5 pg/mL), and CRP (14,600 ng/mL) as Diseased (D) or Healthy (H) Compared at Different Storage Temperatures Number of patients identified CEA
IL-8
CRP
Storage condition
D
H
D
H
D
H
FT0 (- 80C) NT0 (- 170C) NT1 (- 170C, 1st cycle) NT2 (- 170C, 2nd cycle)
2 3 3 3
18 17 17 17
1 1 1 1
19 19 19 19
1 1 1 1
19 19 19 19
Aliquots of long-term storage in the freezer at - 80C (FT0) vs. aliquots stored in N2 without freeze-thaw cycles (NT0) and after first thawing cycle (NT1) and second thawing cycle (NT2).
diagnosed in un-thawed samples were also detected after each freeze-thaw cycle (three patients were detected by CEA, one patient by IL-8, and one by CRP). Thus, the patient detected by IL-8 was also detected by CRP and CEA. Two patients were detected only by increased CEA levels of serum samples stored at - 170C with and without freeze-thaw cycles (Table 3). The low number of patients predicted to be malignant may be explained by the fact that the serum from the twenty cancer patients had been obtained post-operatively during follow-up, whereas in the original study the samples were pre-operative collections. Thus, serum levels of initial cancer markers may have decreased during therapy and recovery of the patients.
Discussion Many biomarker studies appear to be highly heterogeneous regarding sample quality management including collection, processing, and storage conditions. Awareness of the importance of pre-analytical factors is increasing and thorough knowledge of pre-analytical variables is essential to control their impact.8,15,19 This is the first study to report the effect of - 80C versus - 170C serum preservation and the effect of repeated freeze-thaw cycles on eight distinct serum biomarker levels assessed by a multiplex biochip array. Overall, long-term storage over a mean duration of 33.65 months at - 80C compared to - 170C did not influence serum levels of the eight biomarkers significantly. This contrasts with previous reports that demonstrated a clear advantage of plasma-based biomarker preservation at lower temperatures.12 However, marker stability during long-term storage is likely to be dependent on standard operating procedures (SOPs) such as rapid processing of serum samples before freezing. Our serum sample collection SOP from phlebotomy to freezing requires completion within 30 minutes. Prolonged periods before processing and freezing can impact the stability of markers at different storage temperatures.16,20 While low temperature freezing generally slows protein degradation, freeze-thaw cycles may lead to protein denaturation, aggregation, and other conformational changes that can cause modulation (e.g., resulting in a decreased or increased biomarker level). Importantly, degradation of biomarkers can be rapid and can vary between different
markers.21 The results of our study indicate that some proteins are more sensitive to freeze-thaw cycles than others. Serum levels of CEA, IL-8, VEGF, M-CSF, C3adesArg, and CRP appeared to be influenced by repeated freeze-thawing, whereas S100A11 and CD26 were not affected. Our results also showed that serum levels of different markers changed in different directions: CEA, IL-8, C3adesArg, and CRP increased with repeated freeze-thaw cycles, while VEGF decreased. In line with our results, Kisand et al.22 found that VEGF concentrations decreased after one freeze-thaw cycle from analysis of 36 serum samples. In contrast, Hetland et al.23 reported that VEGF serum levels, in contrast to plasma levels, were not significantly affected by up to nine freeze-thaw cycles ( - 80C). For CRP, Aziz et al.24 showed that serum levels were also unaffected by up to seven freezethaw cycles (- 70C). The differences among these studies, as well as our results, might be due to different antibodies, assays (enzyme immunoassay vs. multiplex array), storage temperatures (- 80C vs. - 170C), and/or other preanalytical sample handling factors applied. For CEA, IL-8, M-CSF, S100A11, CD26, and C3adesArg, no comparable marker stability studies are available that focus on freezethaw cycles. Based on our results, freeze-thaw cycles of serum samples can change distinct protein levels significantly. Thus, potential differences in serum levels between disease stages may not be evident if marker levels are substantially degraded by repeated freeze-thaw cycles, particularly if sample groups to be compared are not exposed to identical conditions. Furthermore, even if mean serum levels of distinct biomarkers are not significantly affected (e.g., in our study comparing - 80C and - 170C storage), individual class prediction can still be hampered leading to false diagnosis. Thus, sample quality is as important as quality of diagnostic performance itself for diagnostic studies. In clinical routine, repeated freeze/thaw cycles are common and thus, apparent biomarker changes might rely more on physical/ chemical changes in the sample than on physiological changes in the patients. This might hamper the reliability of established clinical tests, as well as newly implemented biomarker tests, for which marker stability has not yet been explored. While our data suggest that samples stored at - 80C should be readily diagnosable with the biochip, our findings also have highlighted the validity of routine sample storage at - 170C to avoid misdiagnosis of individual cases. In the event that sample freeze-thawing has occurred, a complete record of freeze-thaw cycles should be maintained and taken into consideration for downstream applications. Furthermore, in the future it would be desirable for a biochip design to include, for example, degradation markers that indicate low quality of the samples tested. The risk-management of the potential inaccuracy of diagnostic tests might be optimized by adding such degradation or quality marker proteins to each diagnostic assay in order to trace patient samples’ quality. Unfortunately, such markers are not yet implemented in clinical routine and require more research in order to be validated. However, SOPs for the pre-analytical phase should be applied permanently on all samples and not only for one special test. Given these results, we would like to emphasize the importance of detailed SOPs for sample quality management in biomarker discovery studies, as this will directly impact the
ARRAY AND PRE-ANALYTICAL CONDITIONS IN CRC biomarkers’ validity and clinical applicability. Several guidelines exist that can improve the development of these SOPs. The International Society for Biological and Environmental Repositories’ (ISBER) Biospecimen Science Working Group developed the SPREC reporting system (Standard PREanalytical Code), detailing the main pre-analytical factors affecting clinical fluid and solid biospecimens.25,26 Not only serum biomarkers are affected by pre-analytical bias, but also markers in plasma samples27 or viability of human cells for clinical application, for example, cell therapy.28 The ‘‘Biospecimen Reporting for Improved Study Quality’’ scheme (BRISQ) could provide a standardized approach for reporting such information.9 Additionally, several guidelines for publishing results in different areas already exist [STARD (diagnostics), REMARK (prognostics), CONSORT (clinical trials), STROBE (epidemiology), MIAME (microarray), and MIAPE (proteomics)].10,11,29–32 Besides storage temperature, storage duration should be taken into account at different storage temperatures as shown in several studies. Pieragostino et al.33 detected critical conditions for serum protein profiles depending on storage times and temperatures and demonstrated that upon a - 20C short term storage, characteristic degradation profiles occur. Furthermore, Schultz et al.34 demonstrated that urinary proteins are significantly underestimated after storage at - 20C compared to - 70C. Our current data are not influenced by storage duration, since samples at different storage temperatures (- 80C and - 170C) have undergone the same storage durations, as they were stored simultaneously. Our study has shown that three freeze-thaw cycles change serum marker levels significantly. Analyses of the eight serum markers evaluated in this study further suggest that serum preservation at - 80C is comparable to that at - 170C. However, storage at - 80C can lead to misdiagnosis, thus limiting diagnostic performance. Our findings clearly emphasize the importance of standardized sample collection, processing, storage, and reporting to enable the completion of essential large-scale, multicenter trials employing high-quality specimen cohorts.
385
2.
3.
4.
5. 6. 7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
Acknowledgments We would like to thank Elke Gheribi and Regina Kaatz for clinical sampling.
17.
Author Disclosure Statement
18.
The authors declare that they have no competing interests. The consortium ‘‘Colon Cancer Screening Chip’’ was generously supported by the German Federal Ministry of Education and Research (BMBF) within the Molecular Diagnostics funding scheme (Grants 01ES0720, 01ES0721, 01ES0722, and 01ES0723). The study was performed in collaboration with the ‘‘Surgical Center for Translational Oncology–Lu¨beck’’ (SCTO-L) and the ‘‘North German Tumorbank of Colorectal Cancer’’ network, the latter being generously supported by the German Cancer Aid Foundation (Dt. Krebshilfe e. V., Grant #108446).
19.
20.
21.
22.
References 1. Wilkins MR, Pasquali C, Appel RD, et al. From proteins to proteomes: Large scale protein identification by two-dimen-
23.
sional electrophoresis and amino acid analysis. Biotechnology (NY) 1996;14:61–65. Smith L, Lind MJ, Welham KJ, Cawkwell L. Cancer proteomics and its application to discovery of therapy response markers in human cancer. Cancer 2006;107:232–241. Kulasingam V, Diamandis EP. Strategies for discovering novel cancer biomarkers through utilization of emerging technologies. Nature Clin Pract Oncol 2008;5:588–599. Lippi G, Salvagno GL, Montagnana M, Guidi GC. Reliability of the thrombin-generation assay in frozen-thawed platelet-rich plasma. Clin Chem 2006;52:1827–1828. Carraro P, Plebani M. Errors in a stat laboratory: types and frequencies 10 years later. Clin Chem. Jul 2007;53(7):1338–1342. Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. Clin Chem 2002;48:691–698. Conrads TP, Hood BL, Veenstra TD. Sampling and analytical strategies for biomarker discovery using mass spectrometry. BioTech 2006;40:799–805. Ferguson RE, Hochstrasser DF, Banks RE. Impact of preanalytical variables on the analysis of biological fluids in proteomic studies. Proteomics Clin Applic 2007;1:739–746. Moore HM, Kelly AB, Jewell SD, et al. Biospecimen reporting for improved study quality (BRISQ). Cancer Cytopathol 2011;119: 92–101. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, Clark GM. REporting recommendations for tumour MARKer prognostic studies (REMARK). Br J Cancer 2005;93:387–391. Taylor CF, Paton NW, Lilley KS, et al. The minimum information about a proteomics experiment (MIAPE). Nat Biotechnol 2007; 25:887–893. Rai AJ, Gelfand CA, Haywood BC, et al. HUPO Plasma Proteome Project specimen collection and handling: Towards the standardization of parameters for plasma proteome samples. Proteomics 2005;5:3262–3277. Bu¨nger S, Haug U, Kelly M, et al. A novel multiplex-protein array for serum diagnostics of colon cancer: A case-control study. BMC Cancer 2012;12:393. Su SC, Garbers S, Rieper TD, Toniolo P. Temperature variations in upright mechanical freezers. Cancer Epidemiol Biomarkers Prev 1996;5:139–140. Lombardi G, Lanteri P, Colombini A, Banfi G. Blood biochemical markers of bone turnover: Pre-analytical and technical aspects of sample collection and handling. Clin Chem Lab Med 2012;50:771–789. Albrethsen J, Bogebo R, Olsen J, Raskov H, Gammeltoft S. Preanalytical and analytical variation of surface-enhanced laser desorption-ionization time-of-flight mass spectrometry of human serum. Clin Chem Lab Med CCLM/FESCC 2006;44:1243–1252. Fitzgerald SP, McConnell RI, Huxley A. Simultaneous analysis of circulating human cytokines using a high-sensitivity cytokine biochip array. J Proteome Res 2008;7:450–455. Molloy RM, Mc Connell RI, Lamont JV, FitzGerald SP. Automation of biochip array technology for quality results. Clin Chem Lab Med 2005;43:1303–1313. Rai AJ, Vitzthum F. Effects of preanalytical variables on peptide and protein measurements in human serum and plasma: Implications for clinical proteomics. Expert Rev Proteomics 2006; 3:409–426. Brinc D, Chan MK, Venner AA, et al. Long-term stability of biochemical markers in pediatric serum specimens stored at - 80 degrees C: A CALIPER Substudy. Clin Bioch 2012;45:816–826. Goodwin RJ, Dungworth JC, Cobb SR, Pitt AR. Time-dependent evolution of tissue markers by MALDI-MS imaging. Proteomics 2008;8:3801–3808. Kisand K, Kerna I, Kumm J, Jonsson H, Tamm A. Impact of cryopreservation on serum concentration of matrix metalloproteinases (MMP)-7, TIMP-1, vascular growth factors (VEGF) and VEGF-R2 in Biobank samples. Clin Chem Lab Med 2011;49:229–235. Hetland ML, Christensen IJ, Lottenburger T, et al. Circulating VEGF as a biological marker in patients with rheumatoid ar-
386
24.
25.
26.
27.
28.
29. 30.
31.
thritis? Preanalytical and biological variability in healthy persons and in patients. Dis Markers 2008;24:1–10. Aziz N, Fahey JL, Detels R, Butch AW. Analytical performance of a highly sensitive C-reactive protein-based immunoassay and the effects of laboratory variables on levels of protein in blood. Clin Diag Lab Immunol 2003;10:652–657. Betsou F, Lehmann S, Ashton G, et al. Standard preanalytical coding for biospecimens: Defining the sample PREanalytical code. Cancer Epidemiol Biomarkers Prev 2010;19:1004–1011. Lehmann SG, Moore F, Ashton H, et al. Standard preanalytical coding for biospecimens: Review and implementation of the Sample PREanalytical Code (SPREC) Biopreserv Biobank 2012; 10:366–375. Paltiel L, Ronningen KS, Meltzer HM, Baker SV, Hoppin JA. Evaluation of freeze thaw cycles on stored plasma in the biobank of the Norwegian Mother and Child Cohort Study. Cell Preserv Technol 2008;6:223–230. Massie I, Selden C, Hodgson H, Fuller B. Storage temperatures for cold-chain delivery in cell therapy: A study of alginateencapsulated liver cell spheroids stored at - 80C or - 170C for up to 1 year. Tissue Engineer Part C, Methods. 2013;19:189–195. Bossuyt PM, Reitsma JB. The STARD initiative. Lancet 2003; 361:71. Brazma A, Hingamp P, Quackenbush J, et al. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat Genet 2001;29:365–371. Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: Updated guidelines for reporting parallel group randomised trials. J Clin Epidemiol 2010;63:834–840.
¨ NGER ET AL. BU 32. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344–349. 33. Pieragostino D, Petrucci F, Del Boccio P, et al. Pre-analytical factors in clinical proteomics investigations: Impact of ex vivo protein modifications for multiple sclerosis biomarker discovery. J Proteomics 2010;73:579–592. 34. Schultz CJ, Dalton RN, Turner C, Neil HA, Dunger DB. Freezing method affects the concentration and variability of urine proteins and the interpretation of data on microalbuminuria. The Oxford Regional Prospective Study Group. Diabetic Med J Br Diabetic Assoc 2000;17:7–14.
Address correspondence to: Jens K. Habermann, MD, PhD Section for Translational Surgical Oncology and Biobanking Department of Surgery University of Lu¨beck and University Medical Center Schleswig-Holstein Campus Lu¨beck Ratzeburger Allee 160 D-23538 Lu¨beck Germany E-mail: [email protected]
