Effect of Multiple Freeze–Thaw Cycles on Coagulation Testing

A recent publication in this journal identified numerous potential sources of variables that may alter the results of coagulation testing.[1] Storage of blood samples is one identified area where potential sources of test bias may occur, especially when storage is compromised or not optimal. For freshly collected samples that cannot be tested within the recommended time frame for measurand stability, plasma is separated by centrifugation (ideally with double centrifugation to ensure depletion of platelets[2]), stored in capped polypropylene storage vials, and maintained frozen, preferably at −70°C or colder.

Historically, it has been suggested in recent publications[2] [3] and reagent manufacturer package inserts that plasma samples for coagulation should avoid multiple freeze–thaw processes, as sample integrity may be compromised. Despite the long historical laboratory lore about avoiding multiple freeze–thaw cycles for coagulation testing, there is actually very little published evidence to support this restriction. There have been a couple of recent freeze–thaw publications on a limited number of coagulation tests. For example, in a study related to antiphospholipid antibody testing, Maelegheer and Devreese demonstrated that five freeze–thaw cycles did not cause any significant difference or impact on final result interpretation of anticardiolipin or anti-Beta2 glycoprotein I antibody testing.[4] In another study, Zhao and colleagues indicated that factor (F) XII can undergo only a single freeze–thaw cycle before test accuracy is compromised (albeit FXII is not considered clinically significant); FII, FV, and FIX can undergo only two freeze–thaw cycles; and FXI can undergo three freeze–thaw cycles.[5] The authors used a 10% change threshold, but their study design introduced a significant bias in that the testing was not performed concurrently, and thus day-to-day imprecision potentially affected study findings. What is encouraging is the increasing number of publications that are describing freeze–thaw effects (e.g., of direct oral anticoagulant levels), with typically no adverse effect of up to three freeze–thaw cycles being reported.[6] What should be considered when performing any freeze–thaw study is the continuity of the thawing and testing process, as standardization will help avoid biases, particularly with fibrinogen and FVIII testing.[7]

Critically, the ability to utilize previously thawed samples may alleviate some challenges with algorithmic testing, as well as with preservation for biorepository purposes. An additional benefit could be sample use for longitudinal quality assurance (QA; e.g., lot-to-lot performance verification) or optimizing (maximum usage) the use of calibrators and controls. The purpose of this correspondence is to report on our experience, in which we evaluated effects of multiple (seven) freeze–thaw cycles on routine and specialized hemostasis testing. Anonymized shed platelet-poor blood plasmas from patients collected as part of their routine care were used for the study ([Table 1]). Although most samples represent individual patient samples, in some instances, similar plasma sample types (e.g., normal, unfractionated heparin treated, Coumadin [Bristol-Myers Squibb Pharma Company] treated, and liver disease) were pooled to assure adequate testing volume for each aliquot. As the primary purpose of the study was QA for internal laboratory practice, and as the samples were anonymized, without requirement of patient demographics, history, or diagnosis, no additional patient consent was required.

Table 1 Coagulation Measurands, sample size (N), testing method and range of results

Measurand, unit	N	Method	Once thawed result range	Within-thawed precision range (CV %)	Within-run precision, N = 7 (CV %)
PT, s	27	Clot-based	8.1–16.9	0.5–3.4	0.8
INR	27	Clot-based	0.90–1.90	0.0–4.9	1.0
APTT, s	26	Clot-based	24.5–52.2	1.2–3.7	2.3
Fibrinogen, g/L	20	Clot-based	2.4–5.3	1.0–4.0	0.9
D-dimer, µg/L	18	LIA	0.21–6.24	1.0–7.7	5.3
Factor II activity, IU/mL	10	Clot-based	0.47–1.22	1.5–5.3	2.1
Factor V activity, IU/mL	15	Clot-based	0.68–1.47	2.2–7.1	0.9
Factor VII activity, IU/mL	15	Clot-based	0.24–1.49	1.2–7.3	2.8
Factor VIII activity, IU/mL	15	Clot-based	0.60–2.86	5.2–13.1	5.2
Factor IX activity, IU/mL	15	Clot-based	0.59–1.91	2.1–4.9	2.2
Factor X activity, IU/mL	15	Clot-based	0.23–1.25	0.0–4.6	1.7
Factor XI activity, IU/mL	13	Clot-based	0.23–1.39	1.9–5.5	2.6
Factor XII activity, IU/mL	13	Clot-based	0.48–1.63	1.9–8.7	7.4
Factor XIII activity, IU/mL	5	Chromogenic	0.80–0.96	1.3–2.8	4.9
VWF activity, IU/mL	10	Agglutination	0.43–3.85	2.7–9.2	5.5
VWF antigen, IU/mL	5	LIA	0.63–1.10	2.2–7.2	8.0
Antithrombin activity, IU/mL	17	Chromogenic	0.76–1.14	0.0–5.0	2.3
Protein C activity, IU/mL	13	Chromogenic	0.30–1.23	1.4–7.4	1.5
Protein S activity, IU/mL	10	Clot-based	0.25–0.89	2.0–6.9	1.8
Plasminogen activity IU/mL	16	Chromogenic	0.92–1.14	0.8–4.8	1.0
DRVVT—screen, s	10	Clot-based	34.6–65.4	2.8–4.3	1.5
DRVVT—confirm, s	10	Clot-based	30.0–44.8	1.3–3.3	1.8
DRVVT—ratio	10	Clot-based	1.10–1.43	0.9–2.3	0.7
Anti-Xa (LMWH), U/mL	9	Chromogenic	0.08–0.66	2.2–10.7	5.9

Abbreviations: APTT, activated partial thromboplastin time; DRVVT, dilute Russell's viper venom time; INR, international normalized ratio; LIA, latex-immunoassay; LMWH, low-molecular-weight heparin; PT, prothrombin time; VWF, von Willebrand factor.

Notes: PT and INR using Innovin (Siemens Healthcare Diagnostics, Marburg, Germany) on ACL TOP700 (Instrumentation Laboratory [IL], Bedford, MA); APTT using SynthASil (IL) on ACL TOP700; Fibrinogen: Fibrinogen C XL (IL) on ACL TOP700; D-dimer: Innovance D-dimer (Siemens) on BCSXP (Siemens); F2, F5, F7, and F10: Siemens factor deficient plasma, Innovin reagent, on BCSXP; factors VIII, IX, XI, and XII: Siemens factor deficient plasma, Actin FS APTT reagent (Siemens), on BCSXP; factor XIII (Berichrom, Siemens), VWF activity (BC von Willebrand reagent, Siemens), protein C (Berichrom), plasminogen (Berichrom), DRVVT screen (LA1, Siemens), confirm (LA2, Siemens) with ratio on BCSXP; VWF antigen (HemosIL, IL) on BCSXP; protein S activity (STACLOT Protein S; Diagnostica Stago, Parsippancy, NJ) on BCSXP; anti-Xa: Chromogenix COAMATIC anti-Xa (IL) on BCSXP.

Seven aliquots of each sample were prepared and labeled with a unique study number, and frozen at −70°C or colder for at least 1 week. To minimize between-run bias or imprecision (usually reflected in Quality Control (QC)), each sample was prepared, thawed, and tested as follows: For Day 1 thaw, sample 7 was removed from the freezer and placed in 37°C waterbath for 5 minutes, gently mixed by inversion, and refrozen. On Day 2, samples 6 and 7 were thawed in 37°C waterbath for 5 minutes, mixed, and refrozen. The process was repeated in the same manner as noted in [Table 1], until Day 7. On Day 7, all samples were thawed and tested concurrently. As such, the only bias introduced would be the within-run imprecision. Thus, results of samples exceeding the within-run replicate coefficient of variations (CVs) would seem to be affected by multiple freeze–thaw cycles. We also used Pearson's correlation and Student's paired t-test to compare pooled groups of tests, with p < 0.05 indicating significant differences between results. A difference of > 15% was also taken into consideration for data analysis, as this acceptance threshold is commonly used based on recommended accuracy evaluation criteria.[8]

The test methods included clot based, latex-immunoassay (LIA), agglutination, and chromogenic methods. The effect of multiple freeze–thaw cycles demonstrated three distinct patterns of results: increased results, decreased results, or no significant difference in results. Measurands included prothrombin time (PT)/international normalized ratio (INR), activated partial thromboplastin time (APTT), fibrinogen, D-dimer, factor assays (FII, FV, etc.), von Willebrand factor (VWF) activity (by ristocetin cofactor [RCo]) and antigen, plasminogen, protein C (PC), protein S (PS), antithrombin (AT) activity, lupus anticoagulant testing using dilute Russell viper venom time (DRVVT) method, and low-molecular-weight heparin (LMWH) levels ([Table 1]).

PT/INR, APTT, fibrinogen, FII, FX, DRVVT screen, DRVVT confirm, and DRVVT ratio all demonstrated increased results as compared with once thawed samples. However, the median % change reflected relatively low or modest changes (thus, not considered clinically significant), with PT ranging from 1.4% (third thaw cycle) to 3.1% (seventh thaw cycle) ([Table 2]).

Table 2 Median % change between thaw cycles for all samples within a measurand

Measurand	2 cycles	3 cycles	4 cycles	5 cycles	6 cycles	7 cycles
PT	0.4	1.4[a]	2.1[a]	3.2[a]	3.0[a]	3.1[a]
INR	0.3	1.1	1.3	2.1[a]	2.1[a]	1.8[a]
APTT	1.2[a]	2.1[a]	4.0[a]	3.7[a]	5.4[a]	5.9[a]
Fibrinogen	2.4	2.5	3.1[a]	3.2[a]	3.2[a]	1.9
D-dimer	2.1	4.0	4.4	2.5	3.4	4.0
Factor II activity	2.3[a]	3.1[a]	2.9[a]	6.1[a]	4.6[a]	6.2[a]
Factor V activity	3.0	1.2	-2.2	-1.7	−2.3	−1.9
Factor VII activity	2.3[a]	1.3	2.2	0.2	1.9	0.1
Factor VIII activity	−1.3	−6.0[b]	−9.4[b]	−13.4[b]	−16.0[b]	−17.7[b]
Factor IX activity	−0.1	−0.8	−1.1	−1.4	−0.6	−0.8
Factor X activity	3.1[a]	4.3[a]	3.7[a]	4.0[a]	4.9[a]	5.2[a]
Factor XI activity	1.1	3.3	1.2	0.9	−0.7	0.5
Factor XII activity	3.0	2.7	2.0	1.5	1.3	−0.8
Factor XIII activity	3.4	1.7	4.0	3.7	3.7	3.9
von Willebrand factor activity	−2.8	−4.2	0.1	−2.2	0.9	−0.8
von Willebrand factor antigen	−4.4	−5.9	−5.7	−7.0	−9.6[b]	−8.7[b]
Antithrombin activity	0.0	0.3	−0.6	0.7	0.3	0.4
Protein C activity	−0.9	1.7	0.6	1.0	0.8	0.5
Protein S activity	−3.5	−2.2	−0.5	0.5	2.3	2.7
Plasminogen activity	−0.4	0.1	0.7	−0.1	0.3	−0.6
DRVVT—Screen	2.0[a]	4.0[a]	6.1[a]	7.8[a]	9.1[a]	9.9[a]
DRVVT—Confirm	1.3[a]	2.3[a]	3.3[a]	4.2[a]	5.0[a]	6.2[a]
DRVVT—Ratio	0.5	1.5[a]	2.0[a]	3.4[a]	4.0[a]	3.5[a]
Anti-Xa (LMWH)	1.8	1.8	2.5	1.7	−1.4	0.4

Abbreviations: APTT, activated partial thromboplastin time; DRVVT, dilute Russell's viper venom time; INR, international normalized ratio; LMWH, low-molecular-weight heparin; PT, prothrombin time.

^a Increased results as compared with once thawed sample, p < 0.05.

^b Decreased results as compared with once thawed samples, p < 0.05.

Only FVIII and surprisingly also VWF antigen (LIA based) demonstrated lower results for multiple freeze–thaw cycles as compared with once thawed samples ([Table 2]). Interestingly, VWF:RCo (using platelet agglutination method) did not demonstrate any significant difference between multiple freeze–thaw cycles, suggesting a potential method bias of multiple freeze–thaw cycles.

Four measurands (AT, FV, FXIII, and plasminogen) demonstrated regression statistics < 0.90 ([Table 3]), the others being ≥ 0.90. However, using Student's paired t-test, there was no statistical differences noted between any thaw cycles. No significant difference in freeze–thaw cycles was noted for FV, FIX, FXI, FXII, or FXIII activities as well as AT activity, PC activity, PS activity, plasminogen activity, and heparin levels (anti-Xa testing).

Table 3 Regression (R ²) statistics for select Measurands (those with R ² < 0.90)

Measurand	Thaw 2	Thaw 3	Thaw 4	Thaw 5	Thaw 6	Thaw 7
Antithrombin	0.956	0.887	0.897	0.883	0.889	0.898
Factor V	0.862	0.903	0.849	0.898	0.872	0.942
Factor XIII	0.847	0.798	0.776	0.858	0.677	0.747
Plasminogen	0.839	0.828	0.776	0.720	0.720	0.679

Note: For each measurand, there were no statistical differences between first and any thaw cycles, nor between any thaw cycles (e.g., thaw cycle 2 vs. thaw cycle 3; thaw cycle 3 vs. thaw cycle 4).

There have been several published studies on the effect of single or multiple freeze–thaw cycles on viral antibody testing,[9] [10] immunoglobulin testing,[11] and various chemistry studies including routine chemistry,[12] complement studies,[13] vitamin D studies,[14] subclinical markers for pregnancy outcomes,[15] specific microbiological testing methods,[16] and differences in storage conditions.[17] Some measurands (viral or syphilis studies,[10] [11] direct bilirubin,[12] cholesterol,[12] triglycerides[12]) were not affected by 10 freeze–thaw cycles, whereas total bilirubin was affected by more than one freeze–thaw cycle and blood urea nitrogen and calcium could undergo three freeze–thaw cycles.[12] In addition to storage conditions, differences in thawing methods (capped vs. uncapped, slow ice bath thaw vs. warmed water bath thaw) may contribute to bias associated with single or multiple freeze–thaw cycles[11] [12] [17]; so, standardization of thawing methods is recommended.[7]

Early studies on the effect of freezing and thawing were primarily performed using blood collected from normal donors, which was processed to create platelet-poor plasma units that are stored either frozen (fresh frozen plasma [FFP]) or in a liquid state. Prior to infusion, FFP is thawed in either a water bath or microwave, with the expectation that the product will be immediately used. It has been demonstrated that long-term storage of thawed FFP results in altered coagulation testing. Specifically, there is marked reduction in FV and FVIII, with associated increase in PT and APTT with long-term storage (5–7 days).[18] A more modest reduction (<15%) was seen with fibrinogen, FX, FXI, and AT.[18] In liquid state plasma (as processed from whole blood donors but never frozen), differences were noted in FV, FVII, FVIII, VWF, PS activity, and endogenous thrombin potential at day 15 when maintained at room temperature.[19]

A recent study estimated that nearly 4 million plasma units designated for use in the operating room were returned to the blood bank.[20] As the stability of thawed FFP is limited, a consideration for refreezing FFP might seem like a viable means of salvaging otherwise discarded unused thawed FFP, although these thawed units may instead be sold to pharmaceutical companies for fractionation and manufacturing of factor products. For freeze–thaw effects on FFP, Dzik and colleagues demonstrated that a slight prolongation was noted for twice thawed for APTT (< 1 s), PT (< 0.3 s), and FV (< 0.1 U/mL).[21] However, FVIII reduction was more pronounced after the second thaw cycle, with ∼ 0.10 IU/mL reduction.[21] It should be noted that the once thawed FFP was maintained at refrigerated temperature (1–6°C) for 2 to 4 hours to mimic clinical practice of thawed FFP storage, which may impact the significance of their findings. In 2003, Ben-Tal and colleagues demonstrated no significant difference between twice thawed FFP for fibrinogen, FII, FVI, or FIX, even when maintained at room temperature prior to second refreezing.[22]

The limitation of these blood bank/FFP findings for clinical laboratories are twofold: (1) the plasma used for blood banking purposes is collected into different anticoagulant (citrate-acid dextrose) and at a higher citrate concentration than is typically used for coagulation testing and (2) the blood bank units are sealed to prevent pH changes and microbial contamination, which may not reflect the relatively open (even capped) coagulation sample vials that may be exposed to air or other conditions.

It should be noted that some coagulation test results obtained from testing once frozen–thawed plasma will be statistical significantly different compared with results generated from fresh plasma samples. Specifically, PT, APTT, FV, FVIII, and DRVVT ratios are reduced, whereas fibrinogen and AT are increased after freezing.[23] No changes were noted with D-dimer, FII, FVII, FIX, or unfractionated heparin/LMWH anti-Xa testing. These observations should be taken into consideration if multiple freeze–thaw studies are compared with freshly tested samples.

A limitation of our study is the use of a single reagent source for a given measurand. It has been demonstrated that screening tests (PT and APTT) for reagent sensitivity to factor levels are different between manufacturers,[24] [25] [26] and therefore different observations may be seen with separate reagent systems, different methods (e.g., clot-based vs. chromogenic), or other coagulation measurands. Thus, it would be prudent for laboratories to assess local findings, especially if the reference interval determination and test sample matrix is discordant (e.g., reference interval determined on frozen samples but fresh samples are primarily tested and vice versa).

In regard to our observations for FV, FXIII, AT, and plasminogen, all were performed using chromogenic methods except for FV (clot based). As there were other clot based and chromogenic methods used, and the within-run precision for these tests are < 2.3%, this observation does not appear to reflect an association with a particular testing method, but may instead be related to either the specific measurand and/or specific reagents used in this study.

In conclusion, there is mounting evidence that multiple freeze–thaw cycles will minimally impact coagulation results. These observations will thus aid both clinicians and laboratories to clear the path to performing additional studies on samples already thawed and refrozen, without requiring additional phlebotomy. This would be especially relevant in neonatal and pediatric patients, or to discharged patients, given blood collection limits and/or blood acquisition challenges. As most coagulation reagent manufacturers discourage repeated thawed samples for testing, each laboratory should assess their own reagent systems, using our suggested protocol, to demonstrate local effects of multiple freeze–thaw cycles on their testing. With the exception of lupus anticoagulant testing using DRVVT methods, and possibly factor VIII, most measurands can endure ≥3 freeze–thaw cycles. Nevertheless, it remains unclear whether other lupus anticoagulant testing methods (e.g., hexagonal phase, platelet neutralization) or other coagulation reagent platforms will produce the same robustness or restrictions.

• References

• 1 Gosselin RC, Marlar RA. Preanalytical variables in coagulation testing: setting the stage for accurate results. Semin Thromb Hemost 2019; 45 (05) 433-448
• 2 Adcock Funk DM, Lippi G, Favaloro EJ. Quality standards for sample processing, transportation, and storage in hemostasis testing. Semin Thromb Hemost 2012; 38 (06) 576-585
• 3 Duncan EM, Rodgers SE, McRae SJ. Diagnostic testing for mild hemophilia a in patients with discrepant one-stage, two-stage, and chromogenic factor VIII:C assays. Semin Thromb Hemost 2013; 39 (03) 272-282
• 4 Maelegheer K, Devreese KMJ. The impact of repeated freeze-thaw cycles on antiphospholipid antibody titer. Res Pract Thromb Haemost 2018; 2 (02) 366-369
• 5 Zhao Y, Feng G, Feng L. Effects of pre-analytical storage time, temperature, and freeze-thaw times on coagulation factors activities in citrate-anticoagulated plasma. Ann Transl Med 2018; 6 (23) 456
• 6 Gosselin RC, Adcock DM, Douxfils J. An update on laboratory assessment for direct oral anticoagulants (DOACs). Int J Lab Hematol 2019; 41 (Suppl. 01) 33-39
• 7 Lima-Oliveira G, Adcock DM, Salvagno GL, Favaloro EJ, Lippi G. Mixing of thawed coagulation samples prior to testing: Is any technique better than another?. Clin Biochem 2016; 49 (18) 1399-1401
• 8 International Council for Harmonisation Bioanalytical Method Validation M10. Draft version 5.1; Endorsed June 7, 2018. Available at: https://www.fda.gov/media/128343/download . Accessed September 17, 2019
• 9 Krajden M, Minor JM, Rifkin O, Comanor L. Effect of multiple freeze-thaw cycles on hepatitis B virus DNA and hepatitis C virus RNA quantification as measured with branched-DNA technology. J Clin Microbiol 1999; 37 (06) 1683-1686
• 10 Pinsky NA, Huddleston JM, Jacobson RM, Wollan PC, Poland GA. Effect of multiple freeze-thaw cycles on detection of measles, mumps, and rubella virus antibodies. Clin Diagn Lab Immunol 2003; 10 (01) 19-21
• 11 Castro AR, Jost HA. Effect of multiple freeze and thaw cycles on the sensitivity of IgG and IgM immunoglobulins in the sera of patients with syphilis. Sex Transm Dis 2013; 40 (11) 870-871
• 12 Cuhadar S, Koseoglu M, Atay A, Dirican A. The effect of storage time and freeze-thaw cycles on the stability of serum samples. Biochem Med (Zagreb) 2013; 23 (01) 70-77
• 13 Yang S, McGookey M, Wang Y, Cataland SR, Wu HM. Effect of blood sampling, processing, and storage on the measurement of complement activation biomarkers. Am J Clin Pathol 2015; 143 (04) 558-565
• 14 Antoniucci DM, Black DM, Sellmeyer DE. Serum 25-hydroxyvitamin D is unaffected by multiple freeze-thaw cycles. Clin Chem 2005; 51 (01) 258-261
• 15 Rebholz SL, Melchior JT, Welge JA, Remaley AT, Davidson WS, Woollett LA. Effects of multiple freeze/thaw cycles on measurements of potential novel biomarkers associated with adverse pregnancy outcomes. J Clin Lab Med 2017 2. (01)
• 16 Gosselin RC, Honeychurch K, Kang HJ, Dwyre DM. Effects of storage and thawing conditions on coagulation testing. Int J Lab Hematol 2015; 37 (04) 551-559
• 17 Mitchell BL, Yasui Y, Li CI, Fitzpatrick AL, Lampe PD. Impact of freeze-thaw cycles and storage time on plasma samples used in mass spectrometry based biomarker discovery projects. Cancer Inform 2005; 1: 98-104
• 18 Alhumaidan H, Cheves T, Holme S, Sweeney J. Stability of coagulation factors in plasma prepared after a 24-hour room temperature hold. Transfusion 2010; 50 (09) 1934-1942
• 19 Gosselin RC, Marshall C, Dwyre DM. , et al. Coagulation profile of liquid-state plasma. Transfusion 2013; 53 (03) 579-590
• 20 Meyer MJ, Dzik WH, Levine WC. Reduction in operating room plasma waste after evidence-based quality improvement initiative. Anesth Analg 2018; 126 (05) 1662-1665
• 21 Dzik WH, Riibner MA, Linehan SK. Refreezing previously thawed fresh-frozen plasma. Stability of coagulation factors V and VIII:C. Transfusion 1989; 29 (07) 600-604
• 22 Ben-Tal O, Zwang E, Eichel R, Badalbev T, Hareuveni M. Vitamin K-dependent coagulation factors and fibrinogen levels in FFP remain stable upon repeated freezing and thawing. Transfusion 2003; 43 (07) 873-877
• 23 Gosselin RC, Dwyre DW. Determining the effect of freezing on coagulation testing: comparison of results between fresh and once frozen-thawed plasma. Blood Coagul Fibrinolysis 2015; 26 (01) 69-74
• 24 Bowyer A, Kitchen S, Makris M. The responsiveness of different APTT reagents to mild factor VIII, IX and XI deficiencies. Int J Lab Hematol 2011; 33 (02) 154-158
• 25 Toulon P, Eloit Y, Smahi M. , et al. In vitro sensitivity of different activated partial thromboplastin time reagents to mild clotting factor deficiencies. Int J Lab Hematol 2016; 38 (04) 389-396
• 26 Lawrie AS, Kitchen S, Efthymiou M, Mackie IJ, Machin SJ. Determination of APTT factor sensitivity--the misguiding guideline. Int J Lab Hematol 2013; 35 (06) 652-657