Use of Blood-soaked Cellulose Filter Paper for Measuring Carbon and Nitrogen Stable Isotopes

Preparing serum or plasma (centrifugation) or maintenance of fluid whole blood present logistical difficulties in field conditions. The use of blood-soaked filter paper and similar matrices is well known for the detection of chemical constituents in humans (e.g., neonates), and has been validated via many approaches (Mei et al. 2010), thus, their use in analytical chemistry is not novel. Hansen et al. (2014) and others indicated whole blood as a fluid is less commonly used for biomonitoring due to the relative difficulty with its collection, storage, and transport than samples such as hair. The use of filter paper and similar matrices to sample blood for a variety of measures that address animal and human health can be important for remote community-based efforts (Brook et al. 2009; Curry et al. 2011). Filter paper methods have been validated for a variety of pathogen-specific antibodies of exposed hosts and some chemical constituents of blood in wildlife (Curry et al. 2014; Hansen et al. 2014; Jara et al. 2015; Kamps et al. 2015).

Nitrogen stable isotope ratios (δ¹³N) are basic trophic level measures that allow for determining normalized biomagnification factors and food web biomagnification factors for contaminants (Hoekstra et al. 2002a, b, 2003). Carbon stable isotope ratios (δ¹³C) can be used for assessing geographic movement such as migration of some marine mammals between regions (Schell et al. 1989; Hoekstra et al. 2002a). Many studies have used δ¹³C and δ¹⁵N in combination with environmental contaminants measures to assess the role of feeding ecology and other drivers in determining contaminant concentrations in animals and humans (Rea et al. 2013; Bentzen et al. 2014; McGrew et al. 2014; McHuron et al. 2014).

Our hypothesis was that δ¹³C and δ¹⁵N measured in eluates of blood collected on filter paper would be highly correlated and predictive of values measured in matched samples of anticoagulated whole blood. We collected whole blood from three mammalian species expected to have a wide range of δ¹³C and δ¹⁵N based on differences in foraging ecology, including wild bottlenose dolphins (Tursiops truncatus) from Sarasota Bay, Florida (n=10), captive muskox (Ovibos moschatus) from the University of Alaska (UAF) Large Animal Research Station (n=8), and wild moose (Alces alces) from Alaska (n=22). Well-mixed whole blood was applied to, and subsequently eluted from Advantec Nobuto® filter paper (Toyo Roshi Kaisha, Tokyo, Japan) as described in Hansen et al. (2014). Each filter paper, designed to absorb approximately 100 μL of whole blood, was air dried overnight. We eluted each dried filter paper into 400 μL of phosphate buffered saline in 2 mL microcentrifuge tubes for 16 h at 4 C. After elution, we transferred 200–250 μL of eluate to a clean tube. Eluates (two per individual) and matched whole blood samples (one per individual; 200 μL) were freeze-dried for 48 h as in Hansen et al. (2014) and δ¹³C and δ¹⁵N were analyzed in triplicate 0.25–0.50 mg dry samples at the Alaska Stable Isotope Facility at UAF according to methods described by Rea et al. (2013). The exact volume of blood eluted was not critical because stable isotope measurements represent the ratio of heavier to lighter isotope for each element (¹³C/¹²C and ¹⁵N/¹⁴N) rather than the concentration of each isotope. We calculated mean percent relative SD from individual measurements for each element and species to assess precision. We tested δ¹³C and δ¹⁵N data collected from whole blood and filter paper eluate for normality within each species using the Shapiro-Wilk test. Both comparison of duplicate eluate preparations and comparisons of whole blood and filter paper eluate isotope ratios were assessed using paired t-tests using SigmaPlot 11.1™ (Systat Software Inc., San Jose, California, USA). There were no significant differences between duplicate filter paper eluate preparations for either δ¹³C (P=0.086) or δ¹⁵N (P=0.624), so we compared data from whole blood and filter paper eluates using the average δ value from duplicate eluates. Linear regression and 95% confidence intervals (CI) of the slope were calculated using NCSS 2007™ statistical software (NCSS LLC, Kaysville, Utah, USA). We used a t‐statistic to determine if the slope of the linear regression of filter paper eluate compared to whole blood values was different than 1 (H₀ Slope=1). Differences were considered significant at α=0.05.

The whole blood δ¹³C ranged from −27.31 to −13.85‰ and δ¹⁵N from 1.61 to 11.53‰. Mean percent relative SD for triplicate measurements of individual whole blood and filter paper eluate samples was <1% for δ¹³C and <4% for δ¹⁵N for each species. There was no significant difference in each species between duplicate filter paper eluate preparations for either δ¹³C (dolphin, P=0.507; muskox, P=0.095; moose, P=0.379) or δ¹⁵N (dolphin, P=0.714; muskox, P=0.783; moose, P=0.249). Except for muskox δ¹³C, which was significantly lower in filter paper eluates compared to whole blood (P=0.040), there were no statistical differences between whole blood and filter paper eluate values for δ¹³C or δ¹⁵N (Table 1). The mean absolute difference was only 0.21‰, and the mean percent difference between the two sample types was only 1±1% of the whole blood values. For δ¹³C, the maximum difference between whole blood and filter paper eluate (muskox) was 0.58‰; 2% of the matched whole blood value. For δ¹⁵N, the maximum difference between whole blood and filter paper eluate (dolphin), was 1.15‰; 11% of the matched whole blood value. However, the mean percent difference between whole blood and filter paper eluate δ¹⁵N was <7%. Individual animal δ¹³C relative to δ¹⁵N values for both whole blood and filter paper eluates (Fig. 1) illustrate that minor differences between whole blood and filter paper δ¹³C and δ¹⁵N do not prevent discrimination between groups of animals with different feeding ecology. Our results showed no statistically or biologically relevant differences in δ¹³C and δ¹⁵N between our gold standard (whole blood) and eluates of filter paper. We recognize δ¹³C was statistically different for muskox but the magnitude of the actual difference was small (≤2%) and is not biologically meaningful in the context used.

Our data supported the use of filter paper for assessing δ¹³C and δ¹⁵N to address feeding ecology because 1) both techniques showed each species studied had a unique δ¹³C and δ¹⁵N profile (Fig. 1); and 2) linear regression for all animals and within herbivores (moose) and piscivores were highly correlated (slope not different from 1.0, for δ¹³C: dolphin: P=0.572, moose: P=0.093, for δ¹⁵N: dolphin: P=0.085, moose: P=0.082; relatively high r² values; Fig. 2). The CI for the slopes of both δ¹³C and δ¹⁵N for moose and dolphins bound unity (1.0 is within CI). The r² values indicate filter paper δ¹³C and δ¹⁵N as moderately to strongly predictive of whole blood values using the criteria of O'Hara et al. (2008); having no predictive value if r²≤35%, weakly predictive if r² is 36% to 55%, moderately predictive if r² is 56% to 75%, and strongly predictive if R²>75%. We did not conduct linear regression analysis for muskox because the sample size was small and the data tightly clustered.

Field protocols use multiple filter papers per animal; we collected 15 per animal, assembled into multiple “combs.” We were able to assess chemical feeding ecology associated with mercury using only three to four filter paper strips. This illustrates the potential efficiency provided by this approach, allowing concurrent measures of stable isotopes, contaminants, nutrients, serology, and the archiving of specimens.

Our validation study showed that there was no significant influence of cellulose filter papers on the measures of δ¹³C and δ¹⁵N in whole blood. Thus, measurement of whole blood δ¹³C and δ¹⁵N can be reliably accomplished using samples collected on filter paper. Measurements of δ¹³C and δ¹⁵N are relatively inexpensive compared to other chemical measurements, broadening the scope and utility of filter paper collection of samples. Measurement of δ¹³C and δ¹⁵N in whole blood, packed blood cells, or in other compartments have been used for detailed assessments of human diets (Wilkinson et al. 2007; Nash et al. 2012; O'Brien 2015). Thus, there is potential for adding filter paper techniques for this type of effort in situations where collection and processing of whole blood might be challenging.

We thank Brian Balmer, Jennifer Balmer, Kristina Cammen (Chicago Zoological Society's Sarasota Dolphin Research Program), John Blake, Carla Willetto, and Chris Terzi (University of Alaska Fairbanks) for blood processing assistance during the dolphin and muskox health assessments, and John Harley and Gary Lose for assistance in the laboratory. Blood from moose was provided by the Alaska Department of Fish and Game and was collected for population health assessments (IACUC Protocol 2013-025, State of Alaska Department of Fish and Game, Division of Wildlife Conservation). Wild moose samples were provided by Alaska Department of Fish and Game biologists; in particular, we thank Torsten Bentzen (obtained IACUC 2013-025 approval). Dolphin (National Marine Fisheries Service Scientific Research Permit 15543, Institutional Animal Care and Use Committee 11-09-RW1, funded by Dolphin Quest and the Office of Naval Research) and muskox (standard veterinary care, UAF Large Animal Research Station) were sampled as part of routine health assessments, and we thank the Chicago Zoological Society's Sarasota Dolphin Research Program staff and volunteers and UAF Animal Resource Center staff for collecting whole blood and filter paper samples. Analytical work was funded by the Rural Alaska Monitoring Program funded via the Alaska Native Tribal Health Consortium from a grant from the US Environmental Protection STAR Grant 83370501, and a grant from the US Fish and Wildlife Service, Arctic Landscape Conservation Consortium.