The Norwegian Mother and Child Cohort Study (MoBa) is a prospective study with the purpose to find causes of diseases among mothers and children. Possible causal factors will be linked to information obtained from questionnaires, blood samples from mother, father and child, urine sample from mother and medical registries [4, 5]. We plan to collect and store biological material from 100,000 pregnancies. Pregnant women are recruited from all parts of Norway; as a result, specimens are collected at a number of hospitals and are stored in a central repository. Most of the biological material in MoBa, including EDTA-plasma, is stored in 96-well plates sealed with heat sealing foil. The 96-well format was selected to allow for rapid, semi-automatic handling for specimen storage [4]. This means that each sample stored on a plate undergoes a freeze-thaw cycle whenever a sample on this plate is retrieved. Given these logistics, we wanted to assess the impact both of multiple freeze-thaw cycles and longterm freezer storage on common analytes as surrogates for broad classes of agents.

For our quality assurance specimens, we created an EDTA plasma pool by collecting blood from 40 healthy adult donors in Oslo. Both males and non-pregnant females were included. The blood was collected venously in EDTA tubes (BD Vacutainer). Our goal was to obtain sufficient sample volume to allow repeated measures over time without the need for a new quality assurance pool. To simulate methods used in the MoBa study, the blood stood on the bench for up to 2 hours before plasma was centrifuged at 1800g for 10 minutes at room temperature, and plasma separated. The samples were pooled and stored at 4 °C overnight before aliquoting onto microtitre plates (AB Gene, Epsom, UK), 300 μl per well and sealed with Easy pierce heat sealing foils (AB gene, Epsom, UK). After aliquoting, all plasma samples were stored in the same manner as other MoBa samples, in electric freezers at −80 °C.

To evaluate the impact of freeze-thaw events on specimen quality, we thawed plates weekly for up to 100 weeks. For each time point, three samples, representing three different plates, were sent to the laboratory and were analyzed as three independent samples. Figure 1 shows the schedule of analyte testing following multiple freeze-thaw cycles over time. The EDTA plasma was thawed once a week. The thawing was done at room temperature (1 hour). The samples were retrieved using a pipette to pierce through the foil. When the retrieval of all selected wells on a plate was completed, the holes in the foil were re-sealed with small pieces of a re-sealing heat foil (AB gene, Epsom, UK). For freeze-thaw cycles where no retrievals were done, the samples were homogenized and left on the bench for another 1 hour before freezing again to mimic the approximate retrieval time. We drew an aliquot for chemical analysis after 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 30, 40, 50, 75 and 100 freeze-thaw cycles; t₁,t₂,…,t₁₀₀. A time zero sample (t₀) was analyzed on fresh plasma 24 hours after collection and simultaneously with freezing the rest of the plasma.

To assess that possible change in marker concentrations were the result of the freeze-thaw process and not storage time in the freezer, we also sent samples that had never been thawed before to analysis after 12 and 24 months, t_12m, t_24m.

The EDTA plasma was analysed for sodium, cholesterol, triglycerides, free fatty acids, vitamin E and AST and are presented in Table 1. We chose this suite of analytes to represent different aspects of sample integrity and as surrogates for other chemical agents which might behave similarly. Sodium was chosen as a marker of volume; if the sample was evaporating, the sodium concentration would be anticipated to increase [6]. We chose cholesterol and triglycerides as measures of lipid degradation as other investigators have reported changes over time in these analytes [7–10]. For sample analysis, we chose laboratories where the analyses were conducted on routine basis. When possible we also chose clinical laboratories where the analyses were ISO 17025 accredited. The samples for sodium, cholesterol, triglycerides, vitamin E and AST were sent at room temperature and analyzed the same day. Samples for free fatty acids, were sent frozen (−80 °C) to analysis and have therefore undergone an additional freeze-thaw cycle.

Results for the freeze-thaw cycle analyses are presented in Table 2. A majority of the analytes changed between collection and 1 freeze-thaw, but only sodium, cholesterol and vitamin E were significantly different. Because the design of the MoBa study and other biobanks require a minimum of 1 freeze-thaw cycle, we did all further comparisons to t₁. All analytes, except free fatty acids, were stable for at least the first 10 freeze-thaw cycles. The triglyceride and cholesterol concentrations dropped after 10 and we saw a significant change after 20 cycles. These changes corresponded to a reduction of 17% for triglycerides and 6% for cholesterol. Vitamin E was stable over the first 20 cycles and after 50 cycles the decrease was statistically significant. Free fatty acids were the most unstable and increased with each freeze-thaw cycle (Figure 2). After 10 cycles they had a significant change corresponding to an 11% increase and they increased by 31 % after 20 freeze-thaw cycles. The AST concentration had, as seen in figure 2, a different type of curve from the other analytes. The increase in concentration was statistically significant after 30 and 50 cycles, before it returned to the level near t₁ after 75 cycles. A similar change was seen after 20 cycles, but this change was not statistically significant.

Cholesterol and triglycerides were stable for approximately 10 freeze-thaw cycles, after that the levels started to decline. We saw no change in cholesterol level during storage without thawing after 1 and 2 years; hence we believe the change was due primarily to the freeze-thaw cycles alone. For triglycerides, we saw a significant decrease (6%) after 12 months without thawing. This suggests that the decrease we observed after 50 freeze-thaw cycles mainly was a result of the freeze-thaw process but also that storage time may influence concentration. Other investigators have explored the impact of plasma cholesterol and triglycerides over time [7–10]. Using 10 samples at each time point, Comstock and colleagues showed a small change for cholesterol concentration of −0.4 % per freeze-thaw cycles during three cycles in plasma stored at −70 °C [12]; we would be unable to detect such a small change in concentration. Other studies including our own have shown stability during 3–10 cycles in human plasma stored at −80 °C and for 10 cycles in baboon serum stored at −20 °C [10, 13]. For triglycerides it has been reported stability for 3 cycles in plasma stored at −80 °C [10].

Our results showed that free fatty acid levels increased with the number of freeze-thaw cycles. Already after 10 cycles we saw an 11% increase and after 20 cycles it was 31%. At the same time as the free fatty acids increased, we also saw, as already discussed, that cholesterol and triglycerides decreased; this may explain the increase in free fatty acids concentration. The 12 and 24 month samples showed that the increase in free fatty acids was not associated with storage time but the freeze-thaw cycles alone. Hirsch and colleagues reported overall free fatty acids in plasma were stable when stored at −20 °C for 12 months. Additionally, they indicated that some individual fatty acids in plasma changed over the course of one year, but that this was not due to lipid peroxidation [14]. Taken together these analyses suggest that free fatty acids should not be analysed after repeated freeze-thaw cycles.

The AST level was stable within the first 10–15 freeze-thaw cycles. Earlier the enzyme was shown to be relatively stable through six cycles in serum stored at −70 °C to −76 °C [15], and the same study found only small changes during long-term storage over 10 years. The changes observed were statistically significant, but the authors suggest that the changes were not clinically relevant. We will continue to analyze AST and the other components with yearly intervals to be able to say more about the behaviour after long-term storage.

In our study the vitamin E concentration was relatively stable during the first 20–30 cycles. Gunter et al has found decreasing concentrations (−13% and −22%) after five freeze-thaw cycles in two serum pools stored at −20 °C for 7 to 13 months, suggesting that our colder storage temperature may minimize degradation [16]. Further work by the NHANES laboratory and others have indicated that vitamin E is stable for up to 5 years at −70 °C [17,18]

Our study represents just one aspect of a quality control program for stored specimens. For this quality control study, we chose to use one plasma pool with an average concentration, thus we are limited in our ability to evaluate what would happen at the different analyte levels. We chose to use a balanced design that had the same number of replicates at each time period. In retrospect, the statistical analysis would have been strengthened by having more replicates at t₁. While we selected a wide range of analytes, clearly, they do not represent the full spectrum of analytes which may be considered in future investigations.

The laboratory accuracy of the analyses was very good. The coefficients of analytical variations were 1.0% for sodium, 3.1% for AST, 3.4% for triglycerides, 4.0% for cholesterol, 4.8% for Vitamin E and and 5.0% for free fatty acids. Because the analyses were run over a two-year period, we have to consider that small changes in the laboratories over this period could contribute to the observed changes after multiple freeze-thaw cycles. However, given that we did not observe any differences with samples stored for 12 and 24 months, it is unlikely that laboratory variability explained our results. Because we had such good analytical capabilities, small changes in concentration may have been statistically significant and yet may not represent meaningful changes in concentration compared to the population variability.