Comparison of trace element concentrations in paired formalin-fixed paraffin-embedded and frozen human placentae

Abstract

Introduction

There is increasing interest in measuring metals concentrations in human placentas to better understand physiology, disease, and toxic exposures. For these purposes, formalin-fixed paraffin embedded (FFPE) tissues represent a valuable potential store of well-characterized tissues for analysis. However, the limited literature comparing metal concentrations in FFPE tissue to recently collected frozen tissues paints a confusing picture, and there is no published data directly comparing frozen and FFPE placental tissues.

Methods

Paired samples of frozen and FFPE tissue from 22 rapidly processed human singleton placentae were weighed and digested using standard clean laboratory procedures and subsequently analyzed for a suite of 13 metals using a Perkin Elmer DRC II ICP-MS. The analytical results were compared using either a paired t-test or a sign test depending on data normality.

Results

Concentrations of metals (aluminum (Al), Arsenic (As), barium (Ba), cadmium (Cd), chromium (Cr), copper (Cu), iron (Fe), gadolinium (Gd), mercury (Hg), manganese (Mn), lead (Pb), strontium (Sr), and zinc (Zn)) measured in both types of tissue preparations (frozen and FFPE) displayed a consistent range with other studies and did not display significantly different values from each paired specimens for any of the 13 metals analyzed.

Discussion

Within placentae, metals concentrations of measured trace elements (Al, As, Ba, Cd, Cr, Cu, Fe, Gd, Hg, Mn, Pb, Sr, and Zn) were consistent between FFPE and fresh placental tissue, without indications of systematic element loss or bias. FFPE may offer an important and convenient option for measuring trace metals in human placental tissues.

1. Introduction

Evaluating the metal content of placental tissues can improve our understanding of the developmental origins of health and disease [1–4]. Metals can play varying and complex synergistic and antagonistic roles in perinatal development. For example, some metals, like zinc (Zn) or copper (Cu) are essential for healthy development [5,6], while others, like lead (Pb) or cadmium (Cd) have no known essential biological role in humans and are toxic to the developing fetus [1,7,8].Thus, the reliable measurement of metal concentrations in placentae, including placental tissues which are collected, preserved, and stored by varying techniques, is of increasing interest and importance. Most prior work on metal concentrations in placental tissue has used fresh or frozen tissue; however, this approach can be burdensome as it requires time-sensitive collection and processing, limiting its utility for environmental studies. Formalin-fixed paraffin-embedded (FFPE) placental tissue stores represent a potentially important source of existing and archived material that has not been widely used to assess environmental exposures. If metals concentrations in tissue are not altered by the process of formalin fixation and paraffin embedding, FFPE could offer a convenient alternative that could open new research directions through these clinical biospecimens. Thus, it is important to assess the effect, if any, that FFPE processing has on metals concentrations by comparing FFPE to frozen tissue samples.

Several prior studies have examined this question in other species and tissue types with varying results. A study of bovine livers which digested FFPE tissue in nitric and hydrochloric acid without paraffin removal and measured Cu, Fe (iron), and Zn, found consistently lower concentrations of all measured elements in the FFPE tissues than in fresh, frozen tissues (P= 0.0005, n=12) [9]. Another study, which looked at Fe, Zn, Se (selenium), and Cd in human prostate samples, and which first removed paraffin from the tissue using xylene or hexane prior to nitric acid digestion, found statistically indistinguishable differences between frozen and FFPE tissue for Fe and Se, but lower concentrations in FFPE tissue for Zn and Cd (n=15, average recoveries for Zn and Cd in FFPE were 24% and 59% of fresh tissue, respectively) [10]. A more comprehensive suite of elements was measured by Wróbel et al (2020) in swine livers, which removed paraffin with xylene and ethanol prior to nitric acid digestion [11]. While some elements were observed to be much higher in the frozen tissue than the FFPE tissue (potassium (K), rubidium (Rb), manganese (Mn), bromine (Br)), others were higher in the FFPE tissue (calcium (Ca), strontium (Sr), nickel (Ni)), and still others were similar between the two tissue types (titanium, chromium (Cr), Fe, Cu, Zn, Se). This study did not report the number of samples measured and only mean values and standard deviations were given for element concentration results.

These wide-ranging, and often conflicting, results suggest that several factors, including the decision to remove paraffin prior to digestion, may affect the measured trace element concentrations. It is important to note that each of the differences described above exceed the range of reported analytical precision and accuracy indicating measurable differences in select metal concentrations according to sample selection and preparation technique. To our knowledge, there have been no studies published which directly compare metals concentrations in FFPE and freshly collected and frozen human placental tissue. This comparison was essential to allowing this research team to evaluate pathology paraffin blocks of placentae used for pathology evaluations of patients exposed to Gadolinium [unpublished results] having an MRI with or without Gadolinium contrast population.

To evaluate whether the two preparation techniques produce statistically different measured metal concentrations in human placentae and to investigate whether FFPE influenced metal concentrations, we measured 13 metals (Al, As, Ba, Cd, Cr, Cu, Fe, Gd, Hg, Mn, Pb, Sr, and Zn) in aliquots of freshly prepared paired frozen and FFPE tissues from 22 human placentae. Metal concentrations were analyzed via ICP-MS, and results were then compared using either a paired t-test or a sign test depending on data normality to determine if measured concentrations of an element were significantly different between the two sample preparations.

2. Materials and methods

2.1. Human tissue collection

Twenty-two singleton human placentae were collected from women delivering at term at the University of Rochester Medical Center, Rochester, New York. These collections were part of the UPSIDE/ECHO cohort study and were approved by the University of Rochester Human Subjects Review Board. All participating mothers signed informed consent prior to engaging in study activities.

Immediately upon delivery, a placental collection team sampled approximately 5g of villus tissue from the maternal side of the placenta. The villus tissue was sampled by removing the maternal endometrial layer and excising the villus tissue without damaging the chorionic plate. The 5g samples were frozen in liquid nitrogen and then transported and stored in a −80oC freezer until shipping to the analytical site.

After the 5g of tissue was collected as described above, the remainder of the placenta was taken to Surgical Pathology, where it was evaluated and placed in formalin. Per routine clinical protocols, the formalin fixed placentae were kept a minimum of 3 days in formalin and then evaluated through traditional review protocols whereby the tissue was sampled, placed in cassettes, then processed and placed in paraffin blocks. Standard histology sectioning was performed, and additional sections were cut for metal analyses. The thick tissue sections (50 micron) were placed in a metal free plastic tube and shipped to the analytical site, the Department of Environmental, Earth, and Ocean Sciences at the University of Massachusetts at Boston (UMB).

2.2. Sample Receipt and Storage

Placental specimens were received intact and frozen on dry ice and had no visible signs of thawing or other damage. After samples were individually logged in by manually entering the blinded sample code into an encrypted Excel spreadsheet, samples were stored in a −80°C freezer until the time of sample preparation. To minimize variability that could be related to storage time, all samples were prepared and analyzed at the same time.

2.3. Sample Preparation

All sample preparation for metal analysis was conducted in UMB’s Class 1,000 trace metal free clean laboratory facility. Before analysis, all samples were thawed at room temperature for approximately 4 hours. An aliquot of ~1,000 milligrams (1.0 grams) of each sample specimen was then individually selected for ICP-MS analysis of metal concentration. Sample aliquots were selected by cutting the placental tissue with a ceramic knife that was cleaned between each sample and blank verified at least once per sample preparation day.

For metal concentrations, each sample was weighed using a calibrated mass balance and the wet tissue weight was recorded. Each sample was then dried for 48 hours (or until dry) at 45°C in a vacuum oven. After drying, samples were re-weighed to determine the dry weight. Individual samples were pulverized and ground with a trace metal free ceramic mortar and pestle, which was cleaned between each sample. After grinding and pulverization, sample weights were re-recorded to ±0.001mg, and the tissue was digested in individual Teflon^® vessels using concentrated (15.9 mol L⁻¹) ultra-pure nitric acid (HNO3) at 70°C for approximately 6 hours or until digestion was complete [13,14]. Samples were then digested under temperature- and pressure-controlled conditions using closed SCP Science Class A digestion tubes and a SCP Science 48-sample digestion block with DigiProbe temperature control. After cooling, The paraffin resolidified and the remaining solution of digested placenta was decanted and baked for approximately 8 hours (or until dry) at 50°C to decant excess nitric acid. Residual dried material was then re-digested with ~0.5mL of ultra-pure HNO₃.

After sample digestion was complete, internal standards consisting of known quantities of 0.1 mL of ultra-pure hydrochloric acid (HCl) (used to stabilize mercury in solution) and indium (In), bismuth (Bi), and iridium (Ir) (obtained from SCP Science) (used to correct for matrix interferences and instrumental drift), were added to the digested placental material. The solutions were diluted using water purified to 18.2 MΩ cm resistance (by a Milli-Q water purification system, Millipore, Bedford, Mass., USA) to result in a final concentration of 2% nitric acid and 0.4% hydrochloric (by volume), and 5 ng/g of each of the internal standards (In, Ir, Bi) [15,16]. All standards used during ICP-MS analysis, including aliquots of the calibration standards, standards used to monitor external precision (certified NIST 1974b SRM and NIST 1566b SRM) and procedural blanks were prepared by the same process.

2.4. ICP-MS analysis

Trace metal concentrations were measured using a Perkin Elmer DRC (Dynamic Reaction Cell) II ICP-MS at UMB. Solution based ICP-MS analyses were conducted with modifications to the EPA 6020A methodologies [17,18], the approved ICP-MS analytical procedure for inorganic trace elements in soils and sediments by the methods demonstrated previously [15,19,20].

To clean sample lines and reduce memory effects during analyses, sample lines were sequentially washed with 18.2 MΩ cm resistivity (by a Milli-Q^® water purification system, Millipore, Bedford, Mass., USA) water for 90 seconds and a 2% nitric acid solution for 120 seconds sequentially between analyses. Procedural blanks were analyzed within each block of 10 samples, to monitor and correct for instrumental and procedural backgrounds. Calibration standards used to determine metal concentrations in tissues from both types of tissue sample preparation method included aliquots of 18.2 MΩ cm resistivity H₂O, NIST 1566b SRM, and NIST 1974b SRM spiked with known quantities of each metal in a linear range from 0.025 to 10 ng/g.

Standards were prepared from 1000 mg/L single element standards obtained from SCP Science, USA. Unaltered aliquots, and aliquots spiked to known concentrations of NIST 1974b, were used for matrix matched external calibration standards [13,19]. Unaltered and spiked aliquots of NIST 1566b SRM were monitored within each block of 10 samples to determine external precision. For all samples, the standard deviation was less than 6% and typically <3.5%. Exceptions were low concentration metals that approached method detection limits. Isobaric corrections (e.g., ArCl interferences on As and BaO interferences on Gd) were performed on-line using ICP-MS software or through the use of reaction gasses (hydrogen). Five duplicate analyses (n=5) were performed for all analytes for each sample solution.

2.5. Statistics

For each metal, the difference in concentrations between paired frozen and FFPE samples was calculated and tested for normality using a Shapiro-Wilkes test. For metals that displayed a normal distribution of differences, a paired t-test was conducted. For metals with non-normal distributions, a sign test was performed. An alpha of 0.05 was used for all tests, and p-values were corrected for family-wise error using a step-down Bonferroni correction adjustment. All reported regressions are simple linear regressions with Pearson coefficients of determination. All statistical calculations were performed in Python 3.10.1 using the scipy and pingouin packages [21,22].

3. Results

Of the 13 elements, differences between frozen and formalin fixed paraffin embedded (FFPE) tissues samples (i.e. M_Frozen – M_FFPE, where M is the concentration of a given metal) were normal for all elements except Al, Cd, Mn, and Sr. For elements with normally distributed differences, a paired t-test was performed, for the other four elements, a sign test was performed. After adjustment for family-wise error, all p-values were greater than 0.05, indicating there was no statistically significant difference between concentrations in the frozen placental samples and the FFPE samples for any metal measured. (Table 1, Fig. 1).

Table 1:

P-values for paired t-tests and sign tests performed to compare elemental concentrations in FFPE and fresh tissue and R² values for the linear regression computed between the two sample types (n=22).

Paired T-test				Sign Test
Element	p-value	R2		Element	p-value	R2
As	0.10	0.89		Al	1	0.79
Ba	0.33	0.92		Cd	0.50	0.73
Cr	1	0.56		Mn	1	0.75
Cu	0.81	0.80		Sr	0.052	0.87
Fe	1	0.83
Gd	1	0.91
Hg	1	0.91
Pb	1	0.90
Zn	1	0.81

Fig 1:

Scatterplots showing measured concentrations (ppb) of 13 metals in FFPE and Frozen tissues. The black line represents perfect 1:1 correlation, and the dashed gray line shows the best fit line through the data.

4. Discussion

We observed no statistically significant differences in metal concentrations measured between samples that were prepared from formalin fixed paraffin embedded (FFPE) tissues and freshly collected, frozen tissue samples. These results indicate that the choice of sample preparation does not impart consistent bias in the placental concentrations of metals measured in this study. The results of this study contrast with the limited literature on this topic in other kinds of tissues; several studies have reported differences in the measured values of select metals, including Cu, Fe, Zn, Cd, Mn, and Sr, according to tissue type and sample preparation method [9–11]. Wróbel et al (2020) found only small differences between the two tissues types for Cr, Fe, Cu, and Zn, and Sarafanov et al (2008) observed no difference in Fe concentrations between FFPE and frozen tissue samples [10,11]. However, Wróbel et al (2020) reported higher Mn concentrations in frozen tissue and higher Sr concentrations in FFPE tissue contra our results, and Srafanov et al (2008) found higher concentrations of Zn and Cd in frozen tissue. Bischoff et al (2008) found concentrations of Cu, Fe, and Zn to all be higher in frozen tissues, which is also at odds with our results [9].

There are several possible explanations for these discrepancies. First, it is possible that different tissue types respond differently to the formalin fixation and paraffin embedding process, as suggested by Sarafanov et al (2008) [10]. This could be related to differences in how tissues themselves respond to the fixation and paraffin embedding process (with Sarafanov et al (2008) suggesting a difference between human prostate and liver tissues), or heterogeneities in the tissues which are not adequately accounted for during various comparisons. In the case of our study, 1 g of tissue was used for both tissue preparation methods, and the coefficients of determination indicated mostly good agreement between the paired samples. The lower coefficients of determination for elements like Cr and Cd in this study may indicate that the natural variation of these elements in the placental tissue is greater than that of other elements, and that sample masses greater than the 1 g used here would better account for local variations in tissue metal concentrations. Future studies which may focus on other regions of the placenta should determine if fresh and FFPE samples of chorionic plate, vascular tissue, and endometrium are comparable for elements of interest, how much tissue to collect to account for natural variations in tissue metal concentrations, and whether there are regions of the placenta which have more homogenous metals concentrations than others.

Second, the decision to remove paraffin chemically prior to tissue digestion may variably mobilize trace elements from placental tissues. Two of the other studies published on this topic removed paraffin from the FFPE samples prior to digestion using organic solvents [10,11]. As a result, these studies do not allow for a direct comparison between preparation techniques, but instead compare the resulting metal concentrations observed after paraffin has been removed using organic solvents. Thus, the analytical sample preparation approach may directly alter metal levels by dissolving metals into the organic solvent or even altering the structure of the tissue. In comparison, our approach dissolves both tissue and paraffin at temperatures above the melting point of paraffin (<70°C), and then separates the digestate containing digested tissue once the paraffin resolidifies. Our hypothesis is that digestion at these temperatures in an acidic solution may keep metals dissolved in the subsequent aqueous solution formed after the introduction of 18.2 MΩ cm resistance water and stabilizing acids. For comparison, the one published study which did not include deparaffinization found that metals concentrations in FFPE tissue had consistently lower metals concentrations compared to fresh tissue, which may be a result of not adequately accounting for the additional mass added to the sample by the paraffin [9]. We accounted for this additional mass in this study by directly measuring the difference in mass of the tubes prior to sample preparation and after replicate sample analysis after drying (assuming paraffin was not volatilized and recognizing this mass as a minimum). Notably, the mass difference attributed to paraffin accounted for 19–43% differences in the assumed mass of tissue determined by measuring the dry weights of the respective tissues.

Contamination of paraffin, solvents, or acids used during the paraffin embedding process, deparaffinization, and/or digestion could also lead to differences in the measured metals concentrations between frozen and FFPE tissues. In addition to contamination, there may also be problems arising from prolonged exposure of tissue to formalin prior to being embedded in paraffin. Multiple studies have shown that prolonged (>1 year) formalin exposure may leach trace elements from biological tissues, though Bischoff et al (2008) reported no difference in Cu, Fe, or Zn concentrations between fresh bovine liver tissue and samples preserved in formalin for between 1 and 45 days [9,23,24]. While our data suggest that FFPE placental tissue prepared under research conditions can be used for the study of the 13 trace elements studied here, laboratory settings can vary, and best practices should always be followed regarding laboratory QA/QC to ensure that reagents used to digest samples or other sample preparation procedures do not introduce in-laboratory contamination or analytical interferences. This includes the use of metal-free reagents, digestion vessels, method blanks, as well as metal clean laboratory procedures. Finally, new techniques for the digestion of FFPE tissue should be rigorously tested against fresh tissue to ensure that no part of the procedure induces metal losses or contamination, especially if deparaffinization is part of the procedure. This validation should also include confirmation that the amount and number of placentas studied is sufficient to account for natural variations in metals concentrations throughout the relevant tissue studied, and for the specific metals of interest.

In our study, a sign test was performed on the data from four elements due to the non-normally distributed difference between their FFPE data and their frozen data. While a sign test does not rely on the same assumptions that a paired t-test does, it is less powerful. Therefore, future work may want to verify the results found here, especially for Sr considering the lower p-value. Additionally, there were a small number of large outliers for some elements, especially at high concentrations. While this is likely due to natural tissue variation, the goals of this study were not to assess natural tissue variation in metals concentration, and future studies may want to focus on tissues from highly exposed populations to ensure that element losses do not occur at high concentrations as well investigating different types of placenta tissue as mentioned above (e.g. chorionic plate tissue).

These investigations may facilitate a wider range of possible uses for formalin fixed tissues in perinatal research. In many instances retrospective studies have identified subjects with particular health conditions (e.g., autism) or environmental exposures (e.g., cadmium or lead exposure) and the placentae have been collected for pathological evaluations, with samples banked. The opportunity to measure metal concentrations in such samples provides a new and novel opportunity to assess exposures at birth by assessing the placentae. This may be especially important in large clinical studies as originally explored in the NIH National Children’s Study and now the NIH ECHO Study.

Data availability:

The research data associated with this article is available via Mendeley Data.