A Comparison of Blood Collection Techniques in Mice and their Effects on Welfare

Abstract

Multiple methods are used to collect blood from mice; these methods have different effects on animal welfare. This study compared blood collection from facial, chin, and saphenous locations with regard to various parameters, including the time needed to collect blood, the number of attempts needed, success at completing the blood collection, volume of blood loss, weight changes in the mouse, presence of external lesions after blood collection and gross lesions at necropsy, physical signs during blood collection (vocalization, urination, and defecation), fecal corticosterone after blood collection, and blood chemistry values. While no one technique was clearly better for animal welfare, each technique had benefits and drawbacks.

Several studies have compared the welfare and blood parameters associated with many common bleeding locations in mice (Mus musculus), including the submandibular (facial), retro-orbital, saphenous, sublingual, and multiple tail bleeding techniques. A recent review of the published data concluded “that there is not enough high-quality evidence available to make a determination on optimal blood sampling route.”²⁹ A newer blood collection technique was described³² that used the submental site, also referred to as chin² site; in this paper, use of that site will be referred to as the “chin bleed.” To the author’s knowledge, no other studies have compared the chin site to other common bleeding sites. Another commonly used site is the retroorbital sinus, which is no longer recommended at some institutions^5,16,18,21 despite studies supporting its use for blood sampling.^8,15 The NC3Rs²² recommends that “retro-orbital bleeding should only be performed under terminal anesthesia.” Another source requires anesthesia, albeit not terminal.¹³ The NIH recommends general or at least topical ophthalmic anesthesia for this method.²³ Sublingual blood collection also requires anesthesia,^3,12 as does tail tip amputation.^13,23,25 Therefore, a limited number of sites are available for collection of larger quantities of blood (100+ µL) without the use of anesthesia. Some studies support¹⁷ or are less favorable toward^20,26 the use of facial vein blood collection due to issues with animal welfare.

According to NC3Rs,²² the retroorbital method can be expected to yield 0.2 (recovery) to 0.5 (terminal) mL of whole blood, saphenous up to 0.15 mL, and the tail vein 0.05 to 0.2 mL. Other reported volumes include 0.1 to 0.15 mL from the tail vein³ and 0.2 to 0.3 mL from the retroorbital sinus or saphenous vein.¹³ In addition, the NIH states that the tail produced small volumes from the vein, but medium to large volumes from the artery.²³

Researchers may want to avoid the use of anesthesia due to added variables in the study and prolonged time for animal handling as compared with collection methods that do not require anesthesia. In addition, anesthesia itself can cause stress.^11,27 Based on the ability to collect larger volumes without the use of anesthesia, saphenous, facial, and chin bleeding are viable options. However, little is known about the welfare impacts of chin bleeding, especially in comparison to the other 2 bleeding sites. A previous study²⁶ reported that this technique was easy to learn and could provide good quality samples.

Materials and Methods

The study was performed within Laboratory Animal Resources at Iowa State University, under an IACUC-approved protocol. All work was done in an AAALAC-accredited facility.

Animals.

CD-1 mice, 15 male and 15 female, age 8 wk on arrival, were procured from Charles River Laboratories (Wilmington, MA). Mice were group-housed by sex, 5 per cage for a total of 30 mice. Mice were housed in individually ventilated single filter cages (Innovive, San Diego, CA) prefilled by the vendor with corncob bedding. All mice were fed Engivo Teklad 2014 diet (West Lafayette, IN) and received water bottles that were prefilled with acidified water (Aquavive, Innovive). All cages received cotton squares or nesting enrichment sheets (Innorichment, Innovive) and loft/shelves (developed in-house) for enrichment. Cage bottoms were changed weekly. Room temperature was maintained between 70 to 72°F (21.1-22.2°C), and humidity between 16% to 41%. The room was maintained on a 12-h light:dark cycle.

Study design.

The study was designed as a crossover with each mouse being bled once at each site. Mice were randomized among 6 groups, based on bleeding technique and order of bleeding site. Mice were acclimated for 8 d before the first blood collection was performed. For each mouse, the time between removal from its cage and being returned to its cage after sampling was recorded.

For saphenous sampling, the time between removal from the cage and being ready for collection (that is, applying petroleum jelly) was also recorded to account for any delay due to the need to restrain the mouse in the conical tube and clip hair from the blood collection site. The need for restraint in a conical tube (rather than scruffing), clipping hair, and application of petroleum jelly were specific to saphenous blood collection. Any blood loss, either from loss of collection into the blood tube or continued hemorrhage after return to the cage, was measured by weighing a preweighed paper towel that was also used to wipe off any blood on the fur. The weight gain on the paper towel was then included in the estimated blood lost. No urine that could have contributed to the weight change was observed on the paper towel. Mice showed minimal grooming while on the paper towel, so blood that was lost due to grooming was assumed to be negligible. Digital pressure was not used to stop continued bleeding after blood collection at any of the 3 sites. Although digital pressure is used routinely to stop bleeding after saphenous vein blood collection,⁵ we opted to omit if for several reasons. Applying digital pressure would have added a time variable that could be inconsistent, as pressure might be applied longer than necessary. More importantly from a welfare perspective, little to no bleeding was seen at the site after release of the mouse, implying that digital pressure was not necessary; this held true for all 3 collection sites and is consistent with what has been reported previously for the facial¹² and chin²⁶ locations.

Each mouse was individually weighed after blood was collected and bleeding had stopped. Mice were again weighed at 24 and 48 h after blood collection. At each weighing, mice were also examined for any external evidence of trauma associated with blood collection. The number of attempts at blood collection per site were recorded, with a maximum of 3 attempts permitted before the collection was considered unsuccessful. Urine or feces voiding and vocalization during blood collection were recorded.

Each sampling attempted to collect 150 to 200 µL of blood in a lithium heparin microvette tube (Sarstedt, Nümbrecht, Germany). When adequate samples were collected, they were analyzed using the iStat CHEM8+ (Zoetis, Parsippany, NJ). For collection of feces, mice were housed individually for a period of up to 30 min at 10 to 12 h after blood collection. This timing was based on a previous report.²⁸ Feces were immediately frozen at -20° C (-4° F) for later measurement of corticosterone.

Blood was collected again on days 22 and 43 of the study. Euthanasia was performed on day 45. The most recent site of blood collection for each mouse was examined grossly for trauma. The gross lesions were scored as 1 (no evidence of trauma), 2 (slight bruising or tissue discoloration), or 3 (moderate to severe bruising or tissue discoloration). All blood collections were performed between the hours of 0730 and 1100 All blood collections for each technique were performed by one individual. All individuals were trained on all 3 types of blood collection and were assigned based on comfort level with each technique and self-reported proficiency in the technique they were assigned. The individuals had various levels of experience with bleeding mice. The individual using the chin technique had the least experience but was able to perform multiple successful collections in a row before the study. The individual performing the saphenous technique was the most experienced.

Saphenous vein blood collection.

Mice were placed in a 50 mL conical tube with ventilation holes added to the closed end. The hind leg was extended, the hair over the saphenous vein shaved, and a small amount of petroleum jelly placed over the vessel. A 25-gauge needle was used to puncture the vein. Blood was collected directly into a lithium heparin microvette tube. Once 150 to 200 µL of blood was collected, the mouse was placed on a paper towel in a cage and monitored for reoccurrence of bleeding. If no bleeding occurred, the mouse was returned to the home cage.

Facial vein bleed.

Mice were scruffed with the scruff held over the shoulders rather than over the ears to avoid distorting the vasculature.¹⁰ Scruffing the mouse in a manner that includes the ears may cause variability in the position of the vessel and increase difficulty in successfully puncturing the vessel. The right cheek was punctured with a 5 mm lancet (MEDIpoint, Mineola, NY) using previously described landmarks¹⁰ and holding the flat surface of the lancet parallel to the muscle fibers. Blood was collected directly into a lithium heparin microvette tube. Once 150 to 200 µL of blood was collected, the mouse was placed on a paper towel in a cage and monitored for reoccurrence of bleeding. As soon as no additional bleeding occurred, the mouse was returned to the home cage.

Chin bleed.

Mice were scruffed and held in dorsal recumbency to expose the chin. A puncture was made using previously described landmarks²⁶ and a 5 mm lancet, with the flat surface of the lancet held parallel to the muscle fibers. Blood was collected directly into a lithium heparin microvette tube. Once 150 to 200 µL of blood was collected, the mouse was placed on a paper towel in a cage and monitored for bleeding. If no bleeding occurred, the mouse was returned to the home cage.

Fecal samples.

Fecal samples were thawed and weighed. The first half of the samples (all samples from the mice in the first 3 boxes) were processed as recommended by Enzo Life Sciences (Farmingdale, NY). Individual samples were mixed with 5 mL of 80% EtOH in distilled water and caped. Tubes were mixed and vortexed to ensure samples were fully broken up. Tubes were mixed overnight on a rotator at 4°C (39.2° F), then centrifuged for 15 min at 1500 rpm. One mL of assay buffer was added to 1 mL of supernatant from each sample. The ELISA was then run with duplicate wells of each sample using the Enzo Life Sciences Corticosterone ELISA kit. However, due to the small amount of feces collected, corticosterone levels were below the level of detection. The second half of the samples (from the fourth through sixth box of mice) were diluted with 2 mL 80% EtOH (rather than 5 mL), for a 4-fold dilution rather than 10-fold dilution. Values from this dilution were in the range of the assay. These samples were run in duplicate wells. Samples with small volumes of feces collected were not assayed, so a total of 13 mice (each with 3 bleeding time points) were assayed in the second ELISA run.

Statistical analysis.

Several analyses were conducted for a variety of purposes, because of the characteristics of dependent variables (continuous and categorical) and the different research questions proposed. Several analyses were specific for the nature of the data and research interest. First, the effect of the blood collection site on number of attempts, defecation, vocalization, 24 h weight change, 48 h weight change, time to return to home cage, time to return minus restraint time, volume of blood collected (mL), urination, total blood lost, percent of total blood volume lost, and blood lost on paper towels (mL) were analyzed. The analysis was run on each response separately by fitting a linear mixed-effects model, (cage and treatment as fixed effects and sample is a random effect as 3 measurements were taken on each sample) and adjusting for multiple comparisons by Bonferroni correction.⁶

The second analysis evaluated the effect of the blood collection site on successful blood collections (greater than 0.095 mL) without excess blood loss (no more than 10% of blood volume). These dependent variables were given binary values (success or failure based on the goal). Therefore, a generalized linear mixed model (logistic regression) was fit, with the same fixed effects and random effect as previous.

Third, because the gross lesion score was taken only at necropsy (no repeated measures), a simple 2-way ANOVA was conducted with cage and treatment as fixed effects and gross lesion score as the dependent variable. To determine differences between the 3 sites, Tukey’s test was performed. The output gives the difference in means, confidence levels, and the adjusted P values for all possible pairs.

When evaluating successful blood collections (greater than 0.095 mL) without excess blood loss (less than 10% blood volume), a logistic regression with random effects (on each sample) was fit. The response is a binary variable, with 1 indicating greater than 0.095. The covariates are the blood collection sites.

For blood chemistries, the cage was not considered, and the analysis was conducted on all individual values, comparing the mean values of the 3 blood collection sites and adjusting for multiple comparisons by using the Bonferroni correction.⁶ Because chemistries were measured on 3 dates and some samples are not measured for all 3 dates, the sample is included as a random effect in the model ${\text{y}}_{\text{jk}}= \mu + {\tau }_{\text{j}}+ {\beta }_{\kappa }+ {\epsilon }_{\text{jk}}$, and β_k approximately $N\left(0,{\sigma }_t{}^2\right)$ independent of ε_jk approximately $\text{N} \left(0,{\sigma }^2{}_{\text{e}}\right)$, where j is the treatment and k is kth sample, τ_j is fixed treatment effect. In addition, pairwise comparisons were performed with P values adjusted due to multiple comparisons and the significant value at 0.05.

Externally visible lesions were evaluated using the proportion test at both 24 and 48 h after blood collection; H0: $\pi \text{c} = \pi \text{s} = \pi \text{f}$, where π_* represents the proportion of having lesions of 3 treatments. The P value was adjusted for multiple tests using the Bonferroni correction.⁶

Fecal corticosterone levels were evaluated using a linear mixed model with mouse, date (3 dates), and collection site as covariates. Mice are treated as random because of missing data. Dates were treated as fixed.

Results

Statistically, the 3 different methods of blood collection had no significant effects on vocalization, defecation, number of attempts needed to collect blood, weight change (reported as absolute change, not a percentage of body weight) at 24 h or 48 h, or fecal corticosterone levels (Tables 1 and 2). Cage (Table 3) and sex also had no effect.

Table 1.

Comparisons of dependent variables with cage and treatment as fixed effects with adjusted P values and contrasts.

Dependent variables	Contrast	Estimate of difference between means	P value
number of attempts	c - f	0.10	0.89
	c - s	−0.07	0.95
	f - s	−0.17	0.73
Defecation (yes or no)	c - f	0.00	1.00
	c - s	−0.17	0.20
	f - s	−0.17	0.20
Vocalization (yes or no)	c - f	0.10	0.55
	c - s	0.13	0.35
	f - s	0.03	0.94
24 h weight change (g)	c - f	0.10	0.92
	c - s	0.20	0.71
	f - s	0.10	0.91
48 h weight change (g)	c - f	−0.17	0.91
	c - s	0.19	0.88
	f - s	0.37	0.65

The first column lists the dependent variables.

This table contains no significant results. (n = 30 for each group, c, f, and s.)

Table 2.

Mean and standard deviation of variables

Variable	Bleeding technique	Mean	Standard deviation
Time to return (s)	c	55.4	30.3
	f	47.6	24.6
	s	202.4	63.2
Time to return-restraint time (s)	s	52.3	14.4
Number of attempts (1-3)	c	1.8	0.9
	f	1.7	0.8
	s	1.8	0.9
Blood collected (mL)	c	0.092	0.061
	f	0.13	0.044
	s	0.101	0.049
Vocalizations (yes or no)	c	0.2	0.4
	f	0.1	0.3
	s	0.1	0.3
Urination (yes or no)	c	0.2	0.4
	f	0.2	0.4
	s	0.6	0.5
Defecation (yes or no)	c	0.2	0.4
	f	0.2	0.4
	s	0.3	0.5
Blood lost on paper towel (mL)	c	0.060	0.056
	f	0.092	0.089
	s	0.037	0.120
Total blood lost (mL)	c	0.152	0.099
	f	0.222	0.117
	s	0.138	0.132
Percent of total blood volume lost	c	0.060	0.039
	f	0.090	0.051
	s	0.056	0.056
24-h weight change (g)	c	0.09	1.02
	f	−0.01	1.06
	s	−0.12	0.80
48-h weight change (g)	c	0.03	2.25
	f	0.20	1.04
	s	−0.17	1.03
Gross lesion score (1-3)	c	1.7	0.23
	f	2.5	0.23
	s	2.7	0.23

Table 3.

Tukey HSD for multiple comparisons.

	Df	Sum Sq	Mean Sq	F value	Pr(>F)	Treatment	Diff	lwr	upr	p adj
cage	5	1.90	0.38	0.77	0.5831	f-c	0.76	−0.03	1.55	0.06
treatment	2	5.51	2.75	5.56	0.0111*	s-c	0.98	0.19	1.77	0.01*
Residuals	22	10.89	0.50			s-f	0.22	−0.57	1.01	0.77

Left portion is the summary of the fitted model, Right portion evaluates the gross lesion score as dependent variable (n = 10 for each group, c, f, and s.)

For blood collection time (Tables 2 and 4), both facial and chin sites had faster return times than saphenous (P < 0.0001 for both), with no difference between facial and chin. This was true whether the time was taken from the moment the mouse was removed from the cage until it was returned to the cage, or, in the case of the saphenous bleed, if the time required to restrain the mouse in the conical tube was not considered. On average, collection from the saphenous vein took 147 s longer than from the chin and 155 s longer than from the facial vein.

Table 4.

Comparisons of dependent variables with cage and treatment as fixed effects, with adjusted P values and contrasts.

Dependent variable	Contrast	Estimate of difference between means	P value
time to return (seconds)	c - f	7.80	0.75
	c - s	−146.93	0.00*
	f - s	−154.73	0.00*
time to return – restraint time (seconds)	c - f	7.80	0.76
	c - s	−94.67	0.00*
	f - s	−102.47	0.00*
blood collected (ml)	c - f	−0.04	0.02*
	c - s	−0.01	0.76
	f - s	0.03	0.09
Urine (presence)	c - f	−0.03	0.95
	c - s	−0.40	0.00*
	f - s	−0.37	0.00*
total blood lost (ml)	c - f	−0.07	0.04*
	c - s	0.01	0.88
	f - s	0.08	0.01*
percent total blood volume lost	c - f	−0.03	0.31
	c - s	0.02	0.53
	f - s	0.05	0.04*
blood lost on paper towel (ml)	c - f	−0.03	0.31
	c - s	0.02	0.53
	f - s	0.06	0.04*

The first column lists the dependent variables.

This table contains significant results. (n = 30 for each group, c, f, and s.)

The incidence of urine production was higher during the saphenous method as compared with either the chin or facial methods (P = 0.00 for both) (Tables 2 and 4). Blood lost on paper towels (Table 2 and 4) was significantly different between facial and saphenous sites, with the facial site having greater amounts of blood lost on the paper towel (P = 0.04).

A blood collection volume of less than 95 µL was considered an inadequate collection, a volume of greater than 95 µL but less than 10% of the total blood volume was considered a successful collection, and a volume of greater than 10% of the total blood volume was considered an over collection. Total blood volume was assumed to be 72 ml/kg;⁵ mass was converted to volume using the formula 1.06 g blood = 1 ml blood.²² According to NC3Rs,²² for a single sample, 10% of the total blood volume is the maximal acceptable collection volume. Another source³ states that single samples of greater than 15% of the total blood volume result in hypovolemic shock if collected rapidly. The 3 sites were not significantly different with regard to the likelihood of a successful collection (Figure 1, Table 5). With regard to the number of adequate collections (i.e., > 95 µL), the facial site had significantly more than did the chin site (P = 0.0076) and marginally more than did the saphenous site (P = 0.05103). The saphenous and chin sites were not significantly different (P = 0.50992). With regard to the number of over-collections (> 10% of the blood volume), the facial vein had significantly more over-collection than did the chin site (P = 0.0036), or the saphenous (P < 0.001), and the saphenous had significantly fewer than did the chin site (P = 0.0233) (Table 5).

Figure 1.

Box and whisker plot showing the likelihood of collecting adequate amounts of blood (>95µl), of over-collecting blood, and of successful collections (>95 µl but <10% blood volume.

Table 5.

Estimate, standard error, and probability of over collection, adequate collection, and successful collection

Over collection (>10% blood volume)	Estimate of difference between means	Std. Error	P value
f-c	10.27	3.2	0.0036*
s-c	−6.33	2.4	0.0233*
s-f	−16.53	4.7	<0.001*
Adequate collection (>95 µl)
f-c	2.19	0.7	0.0076*
s-c	0.06	0.6	0.50992
s-f	−1.58	0.7	0.05103*
Successful collection (>95 µl and not >10% blood volume)
f-c	0.54	0.5	0.554
s-c	0.96	0.5	0.170
s-f	0.42	0.5	0.711

For the gross lesion score at necropsy (Tables 3 and 6), confidence levels and P values show a significant difference between the saphenous and chin sites, with the saphenous site having a higher gross lesion score than chin (P = 0.01). No significant differences were found between saphenous and facial or between chin and facial sites. For external visual lesions present after blood collection, at 24 h the facial site had significantly fewer lesions than did either the chin (P = 0.031) or saphenous (P = 0.0038) sites (Table 7). At 48 h, both the facial (P = 0.0038) and chin (P = 0.015) sites had significantly fewer lesions than did the saphenous site.

Table 6.

Gross lesion scores at necropsy (n = 10 for each group, c, f, and s.)

Blood collection site	1	2	3
	No evidence of trauma	Slight bruising or tissue discoloration	Moderate to severe bruising or tissue discoloration
Chin	5	3	2
Saphenous	1	1	8
Facial	0	5	5

Table 7.

Proportion test of visible external lesions by bleeding type at 24 and 48 h after blood collection (n = 30 for each group, c, f, and s.)

24 h after blood collection
Test	P value	95% confidence interval
πc = πs	0.55	(−0.35, 0.15)
πf = πc	0.031*	(−0.37, −0.023)
πf = πs	0.0038*	(−0.49, −0.10)
48 h after blood collection
Test	P value	95% confidence interval
πc = πs	0.015*	(−0.47, −0.057)
πf = πc	1	(−0.131, 0.064)
πf = πs	0.0038	(−0.49, −0.10)

Blood chemistries showed no significant differences in glucose, BUN, sodium, and the anion gap with regard to collection site, and no significant differences between the remaining values for the facial and chin sites (Table 8). Potassium, chloride, ionized calcium, hematocrit, TCO2, and hemoglobin values were significantly higher in the saphenous group than either the facial or chin group (Table 9). However, the number of samples that were analyzed varied between 36 and 47 samples for each analyte. Many samples were not analyzed due to insufficient sample volume, quality control failure, or insufficient sample drawn into the analyzer chamber. For the saphenous site, 20 samples were insufficient (either due to the amount of blood drawn or the amount deposited in the testing chamber) and only 10 samples were analyzed. For the facial site, 5 samples were insufficient, 4 had quality control errors, and 21 were analyzed. For the chin site, 14 samples were insufficient, 1 had a quality control error, and 15 were analyzed.

Table 8.

Multiple comparison of blood chemistry parameters with adjusted P values and contrasts.

Chemistries	Contrast	Estimate	P value
Anion Gap (n = 36)	c - f	−0.32	0.89
	c - s	−1.53	0.59
	f - s	−1.21	0.69
Na (n = 47)	c - f	−9.53	0.40
	c - s	−7.98	0.63
	f - s	1.56	0.98
Glu (n = 44)	c - f	7.63	0.57
	c - s	17.45	0.12
	f - s	9.83	0.41
BUN (n = 45)	c - f	0.39	0.94
	c - s	−0.75	0.86
	f - s	−1.15	0.69

The first column lists the dependent variables. These results are not significant.

The sample size for each test is listed individually.

Table 9.

Multiple comparison of blood chemistries with adjusted P values and contrasts.

Chemistries	Contrast	Estimate	P value
K (n = 37)	c - f	−0.35	0.37
	c - s	−1.91	0.02*
	f - s	−1.56	0.03*
Cl (n = 44)	c - f	−0.64	0.71
	c - s	−3.84	0.01*
	f - s	−3.20	0.02*
iCa (n = 46)	c - f	−0.01	0.84
	c - s	0.17	<0.001*
	f - s	0.18	<0.001*
TCO2 (n = 46)	c - f	1.66	0.48
	c - s	6.94	<0.001*
	f - s	5.28	0.02*
Hb (n = 45)	c - f	0.74	0.09
	c - s	−2.10	<0.001*
	f - s	−2.84	<0.001*
Hct%PCV (n = 45)	c - f	2.19	0.09
	c - s	−6.14	<0.001*
	f - s	−8.33	<0.001*

The first column lists the dependent variables.

This table contains parameters with significant results.

The sample size for each test is listed individually.

No significant differences were found in the fecal corticosterone levels across groups (means ± SD are facial vein, 240 ± 65 pg/ml; saphenous vein, 235 ± 91 pg/ml; and chin site, 210 ± 70 pg/ml). Fecal corticosterone was measured only in male mice due to dilution issues with samples from the females.

Discussion

This study evaluated which blood collection technique (facial, saphenous, or chin) was preferable based on its effects on animal welfare. The data produced mixed conclusions, depending on which dependent variable was examined, with some variables having no significant effects.

In this study, we considered the incidence of urination, defecation and vocalizations to be indicative of stress. While nonspecific, these signs are also noninvasive. Urination was more frequent during the saphenous collection as compared with the facial and chin collections, suggesting greater stress in these mice. Another indication that the saphenous site was more stressful than the other sites was the time required for sample collection. Even if the time to restrain was not considered, saphenous collection took significantly longer than did collection from the other 2 sites. However, as restraint is necessary for use of this site, consideration of this variable is likely warranted. Saphenous blood collection took on average almost 2.5 min longer to complete than did the facial or chin sites. A shorter period of mouse restraint would be expected to cause less stress and therefore be better for animal welfare. Based on urination and time of restraint, chin and facial sites are preferable to the saphenous site. The longer time necessary for saphenous blood collection was consistent with findings of other studies that compared saphenous and facial sites.^14,19

A significant difference in weight change was not detected between the 3 collection sites. For the facial site in particular, this could be interpreted as not having caused sufficient muscle damage to inhibit eating. Mice in the facial group consistently had higher weights after blood collection compared to the other two groups, but did not have a significant change in weight at 24 or 48 h after blood collection. This lack of difference in weight change between groups is similar to the findings of a study comparing facial, chin, and retroorbital sites.²⁶ Another study concluded that sites involving the head region affected body weight more than those involving the tail and leg.²⁷ However, that study examined weight loss at 6 and 10 h after blood collection, rather than at 24 h and 48 h as in the current study.

Successful blood collection is important for animal welfare, as over-collection can be stressful and result in hematologic changes,³ and under-collection results in handling mice without obtaining adequate volumes of blood. In addition to the blood collected in the tube, we measured blood collected on a paper towel, as has been done previously.²⁶ While we initially assumed that blood collected on the paper towel would be due to hemorrhage after completion of blood collection, much of it was instead related to the rapid blood flow during collection, especially from the facial vein puncture. Facial puncture requires rapid positioning of the collection tube and rapid release of the mouse after collection. In addition, when collecting volumes near the upper acceptable limit, 1 or 2 extra drops can result in over-collection. Collections performed by the investigators during an initial iteration of the trial resulted in over-bleeding in all 3 sites, but more frequently from the facial vein (this preliminary study used a different set of mice, and the data are not included in the analysis of the current study). Extra training is necessary for use of the facial site due to the risk of over-bleeding the mouse. Although the initial goal for collections was 200 µL, personnel were ultimately asked to collect between 150-200 µL due to the risk of over collection, especially with the facial site. The facial site was previously found more likely to result in greater blood loss than the chin site.²⁶ Rapid collection was also mentioned in a previous description of the facial method.¹⁰

On average, the chin site was successful approximately 40% of the time, the facial site was successful approximately 55% of the time, and the saphenous site was successful approximately 67% of the time. Personnel proficiency may influence success with blood collection from mice. All personnel self-reported success with their designated collection site and demonstrated proficiency prior to the start of the study. However, varying levels of proficiency certainly could skew the results. While this study assigned each technique to only one person; further studies could include crossover in the personnel performing the blood collections at the 3 sites to control for this factor. No significant differences were detected in the number of attempts between collection sites, as was reported previously,²⁶ and may imply proficiency in their assigned site by all participants.

The difference in gross lesion score indicates that more trauma is associated with the saphenous site as compared with the chin site. Both the facial and chin sites bleed freely after a good quality puncture. However, to collect this relatively large amount of blood from the saphenous vein requires manipulation of the leg. The bruising that was seen at the saphenous site frequently extended along the lateral portion of the thigh, rather than being localized to the puncture site. Subjectively, tissue trauma at the time of euthanasia after saphenous collection included the following: bruising, ulceration, lacerations on the scrotum, knuckling over of the leg that had been bled, and bruising at the collection site. On necropsy, muscle bruising was the most frequently identified lesion, although one of the chin sites showed subcutaneous edema. The prevalence of lesions on the leg has been reported previously.²⁷ The gross lesion score for the chin was not significantly lower than the facial vein score, in contrast to a previous report.²⁶ However, as the previous paper noted, the use of different lancet sizes between techniques may have impacted the tissue damage.²⁶ Based on the gross lesion score, facial and chin blood collection sites are superior to the saphenous site.

With regard to tissue trauma during blood collection, among 3 mice bled from the chin site, 2 blood in the mouth and 1 had blood in the mouth and/or nose. This was suspected to result from deep punctures, possibly associated with the mouse moving at the time of puncture, and could potentially be prevented by using a shorter lancet, as perhaps the chin vessel is superficial enough that a 4 mm lancet would be adequate in a larger mouse. Mice were observed for externally visible trauma at the site of blood collection at the 24 h and 48 h weighing periods (Table 10). At 24 h, 5 of the mice that had been bled from the chin had a scab, and 1 mouse had swelling (this was the same mouse that had subcutaneous edema at necropsy). At 48 h, one mouse that had been bled from the chin had bruising, but had not had a scab at 24 h. At 24 h, among mice that were bled from the saphenous site, 1 had scabbing, 5 had bruising, 1 had a flank laceration, and 1 had a scrotal laceration. In addition, at 24 h after blood collection, 2 mice were seen to be knuckling over on the leg that had been bled. At 48 h, 2 mice had bruising at the saphenous site, 1 had a flank laceration, 2 had scrotal lacerations or ulcerations, 1 had a scrotal bruise, and 4 showed knuckling over on the leg that had been bled. Significantly more lesions were seen after the saphenous sampling than after sampling from the other 2 sites. Factors that may have contributed to this finding are the longer restraint time needed for saphenous collection and the need for restraint in a conical tube. Although the conical tube had no sharp edges, it nonetheless presented a firm edge that could have contributed to tissue trauma. In addition, the leg was manipulated (intermittently squeezed or the foot pumped) to encourage blood flow; this could have increased pressure at the edge of the conical tube. Finally, at the final collection in particular, mice were large (up to 48 g), which could have contributed to trauma from the tube. Modifying the restraint might be beneficial for large mice when using the saphenous site. After facial sampling, 1 mouse had blood coming from the mouth at the time of sampling; no lesions were seen externally at 24 and 48 h after bleeding. With regard to the number of visible lesions after blood collection, the only significant difference was between facial and saphenous sites, with facial sites having significantly fewer visual lesions. However, this finding may be related to potential masking of lesions with hair.

Table 10.

Number of lesions noted at 24 and 48 h after blood collection.

Site of blood collection	Immediately after blood collection	24 h after blood collection	48 h after blood collection
Chin	3	6	1
Saphenous	0	9	9
Facial	1	0	0

See text for descriptions of lesions (n = 30 for each group, c, f, and s).

Other studies also reported some degree of tissue trauma. Although histologic analysis was not performed in our study, another study found more histologic changes at the saphenous site as compared with the facial site when evaluated at 1 h, 3 d, and 14 d.²⁹ Another study comparing facial (in that study referred to as submandibular) and sublingual sites found no external lesions, but did find subcutaneous hemorrhage and edema on day 1 that had not fully resolved at 5 d.¹² These findings are comparable to the lesions seen in our study. In a comparison of facial and chin sites (respectively called submandibular and submental in the paper), more gross lesions were seen at the facial site than at the chin on 2 d (3 of 7 mice in the chin group and 5 of 5 in the facial group had a trace mark or hematoma).²⁶

Despite significant differences in some of the blood chemistry values, interpretation is difficult. All of the significant differences occurred between the saphenous site and the other 2 locations, with no significant differences between facial and chin samples. Previous studies found differences in CBC values^1,4,24 and blood chemistries^7,8 based on the site of blood collection and on differences in method of bleeding (using techniques such as warming).¹⁴ One study reported a higher hematocrit in tail blood as compared with retroorbital and heart samples, but no significant differences in hemoglobin.²⁴ The explanation for this was based on the use of an automated counter, which was not used in the current study. In our study, hemoglobin and hematocrit were both higher in saphenous samples as compared with samples from the other 2 sites. A study in rats found significant differences in red blood cell counts, hemoglobin, hematocrit, (all higher for saphenous), pCO_2, and Na (higher in orbital puncture) and pH, K, actual base excess, and standard bicarbonate (lower in orbital puncture).³⁰ The authors hypothesized that higher K and lower Na could be associated with erythrocyte damage during collection, or that higher K could be due to muscle contractions. While we found no differences in Na, K was significantly higher in the saphenous blood, similar to what was reported in rats.³⁰ Those authors also speculated that pCO₂ could be low due to hyperventilation. In our study, the TCO₂ (which is calculated by the istat from pCO₂) was significantly lower in the saphenous blood. Blood chemistry values may indicate that saphenous blood collection is more stressful (more muscle contractions and hyperventilation) or potentially causes more erythrocyte damage than does collection from the other 2 sites. As another study noted,¹⁴ the site used for blood collection can influence data, and variation in blood collection sites should be avoided in studies that measure these parameters. Furthermore, in our study sample sizes were unequal between groups due to the problems in running tests. Samples may have had insufficient quantity, clotting (sometimes observable), or filling errors with the equipment. Having fewer usable samples from saphenous blood collection is concerning, although whether the bleeding technique resulted in usable samples is uncertain.

Fecal corticosterone was measured as another indicator of stress. However, this analysis had complications. The fecal corticosterone test was run as per vendor recommendations, but due to the small volume of feces produced per mouse, obtaining an adequate amount of sample was difficult. In our design, mice from all groups were housed together in a cage, which did not allow collection of feces over a 24-h period, as was done in other studies.^9,11 Thus, only half of our samples were usable, as the first set was too dilute. Fecal corticosterone was measured rather than serum corticosterone because repeat blood collection shortly after the location-specific bleed would potentially over-bleed the mice, due to the larger volume necessary for the study, and did not complement the rest of the study design. Fecal corticosterone showed no statistical differences among groups. Similarly, fecal corticosterone showed no differences among groups in other studies that compared sublingual, facial, and retro-bulbar sites,⁹ or tail vein, sublingual, and facial sites;¹¹ in both cases, feces were collected 24 h after blood collection. Another study compared fecal corticosterone measured before, 8 h after, and 1 d after blood collection from the tail vein, the retrobulbar sinus, and via blood-sucking bugs; all groups had higher levels than baseline, but the change was greater when conventional techniques were used as compared with the bug group.³¹

Although a clearly preferred site could not be identified based on the parameters we measured, each site has some benefits. The saphenous site was more consistently successful at providing an adequate amount of blood without over-collection. However, that site also required a longer time for blood collection than did either of the other 2 techniques and caused more tissue damage than did the chin site and more external lesions than did the facial site. The saphenous site may be best used when smaller blood volumes are required. The facial site was good for successful blood collection, but over-collection occurred frequently. The facial site required greater care to prevent over-collection, but even if personnel are aware of the possibility (as in this study), over-collection can still occur. The only significant advantage to the chin collection site was fewer externally visible lesions at 48 h and a lower gross lesion score at necropsy as compared with the saphenous site. Based on our findings, any of the sites can be selected for specific studies. However, our data should be considered with regard to mouse welfare when choosing blood collection sites.