*5.3. Histology*

Nine studies evaluated histological findings [1,4,8,24,28,29,32,36,38]. Only three of these studies evaluated small vein sample routes [24,32,36]. There was significant heterogeneity in method of reporting which ranged from narrative summary [8,24,32,38], to incidence [1,4,29,36], to a semi-subjective scoring system [28]. There were further di fferences since lesions were observed in di fferent anatomic regions as would be expected based on sampling location. For example, ocular trauma was reported in retrobulbar sampling, ye<sup>t</sup> was not seen after facial sampling. Similarly, foreign body steatitis is typically caused by hair shaft penetrance of the area and is less likely to occur when sampling hairless regions. In assimilating findings a judgement call has to be made as to whether the increased incidence of a histological finding implies greater welfare impact, or whether this is constituted by greater severity of lesion, or combination thereof. For these reasons, it was considered that vote counting was inappropriate and results have been summarised narratively (Figure 3).

**Figure 3.** Histological characteristics observed in studies comparing large-volume sample sites. Size of slice represents relative incidence. Di fference in reporting prevented assimilation of all findings. Compiled from [1,4,29,32,38]. Note that animals may have demonstrated more than one histological finding. *n* represents the number of animals.

Retrobulbar sampling consistently led to microscopic evidence of haemorrhage [4,8,28,36] in structures around and within the orbit, including the muscle, retroorbital sinus, harderian gland and nasolacrimal duct. Inflammatory infiltrate with constitute cells changing over time post-sample was also a key feature [4,8,28,36]. Occasional broken hair shafts setting up a foreign body reaction were reported [28,29,36,38]. Optic nerve damage appears to be a rare finding [4]. Massive necrosis, mononuclear cell infiltration, and fibroplasia of the harderian gland was observed in 2/18 animals [36].

Jugular vein sampling led to histological changes such as haemorrhage, inflammatory infiltrate, degenerative change, and oedema in muscle, subcutaneous connective and adipose tissue [8].

In facial-punctured animals, macroscopic observations were characterized by subcutaneous haemorrhage and oedema with an acute inflammatory response [1,36]. Focal muscle necrosis was also observed [1]. Signs of trauma persisted for five days [1]. When compared together, sublingual sampling led to fewer traumatic lesions than facial [1]. Lesions in the former were characterized by minimal to slight haemorrhage, with minimal acute inflammation [1,36]. Scar formation and production of granulation tissue was occurring after 5 days [1]. Trichogranuloma has been observed with both of these methods [2,36].

Anaesthesia appears to impact on severity and type of lesions noted. Comparatively fewer indicators of histological change were observed in retrobulbar sampling with anaesthesia compared to the conscious method [8]. In facial vein-sampled animals a di ffuse acute neutrophilic/fibrinoid inflammatory response was noted after conscious sampling, whereas the inflammatory response was more chronic in anaesthetized animals [1].This di fference in response was likely related to the presence of hair fragments deep within the puncture site in a number of animals [1].

Histology findings following tail sampling were generally mild. Mild neutrophilic inflammation was a consistent finding in mice sampled by tail incision [24,36], which also extended to the dermis in some animals [32]. In a proportion (3/5) of animals, tail amputation resulted in transection of the last caudal vertebra [32]. Additionally all of these animals had neutrophilic inflammation (generally mild) and fibrin at the tail tip [32]. In contrast to tail tip amputation, tail incision led to a shorter period of epidermal oedema, tail muscle involvement with necrosis and inflammatory infiltrate, and earlier proliferation of fibroblasts [36]. The inflammatory infiltrate progressed from neutrophilic to mononuclear cell in both groups over time [36]. Lesions following tail vein sample were primarily in the subcutaneous tissue and adipose tissue [8].

The incidence of histological change was higher after saphenous sample than tail vein (29.4% vs. 13.97% after 1 h) and lesions in the muscle were reported [8]. Alternately, minimal histological change, characterised by minor inflammatory infiltrate and bleeding into the muscle, was observed after saphenous puncture in another study [36]. The authors commented that this finding may have resulted from imperfect tissue sampling.

### *5.4. Behavioural Tests of Anxiety*

### 5.4.1. Elevated Plus Maze (EPM)

Three studies evaluated EPM performance after blood sampling [2,12,32]. Harikrishnan et al. 2018 [12] investigated all three major blood sampling routes, as well as tail incision. Whilst Moore et al. 2017 [32] examined tail amputation and tail incision in comparison with facial vein sampling. Teilmann et al. 2014 [2] compared facial and tail vein sampling using an EPM as part of a triple test. A further consideration is that the study by Harikrishnan et al. 2018 [12] utilised anaesthesia for retrobulbar sampling, but all other routes were performed conscious in this study and in other studies. This factor may confound study interpretation in spite of the behavioural testing occurring 24 h after sampling. As a result of the di fferences in experimental design, vote counting was considered inappropriate and results have been summarised narratively. A summary findings table (Table 4) for both the EPM and OFT behavioural test is provided below.


**Table 4.** Summary of direction of effect for behavioural tests elevated plus maze (EPM) and open field test (OFT) when compared to sham/unmanipulated controls or baseline values.

In Harikrishnan et al. 2018 [12], the groups spent significantly different durations in the open arms (*p* = 0.03), ye<sup>t</sup> there were no differences in time spent in closed arms or the number of open-arm visits. The groups also differed in the number of centre visits (*p* = 0.002). Facially vein-punctured animals did not differ from controls in these parameters, whereas retrobubular sampling caused greatest deviation from controls with reduced centre visits, least time in the centre and reduced activity. Sublingual and tail incision led to intermediate deviations from control values, frequently exhibiting more anxiety behaviour than facially sampled animals, but not differing from each other.

Moore et al. 2017 examined facial, tail amputation and tail incision routes to find that phlebotomy group did not affect performance in the elevated plus maze [32].

### 5.4.2. Open Field Test (OFT)

Four studies evaluated OFT performance after blood sampling [2,8,12,32]. Harikrishnan et al. 2018 [12] investigated all three major blood sampling routes, as well as tail incision. Whilst Moore et al. 2017 [32] evaluated tail amputation and tail incision, in comparison with facial vein sampling. Teilmann et al. 2014 [2] compared facial and tail vein sampling using an OFT as part of a triple test. Tsai et al. 2015 [8] studied retrobulbar (with and without anaesthesia), facial, tail vein, and saphenous bleeding.

Tsai et al. 2015 [8] measured total distance travelled in the test which is an important measure of locomotor activity. Distance travelled di ffered between groups being longest in mice that underwent tail vein bleeding (2.183 cm) or saphenous (2.110 cm), followed by mice that underwent retrobulbar (1.699 cm) and facial vein bleeding (1.226 cm). This di fference was significant for facially vein-sampled animals, but non-significant for other pairwise comparisons. Facial vein-sampled animals also had a lower average speed. This study has not been included in the summary table because values were not compared to a sham/unmanipulated control to enable a determination on direction of e ffect.

The study by Harikrishnan et al. 2018 [12] utilised anaesthesia for retrobulbar sampling but all other routes were performed conscious in this and in other studies. This may be a consideration in interpretation, in spite of testing occurring 24 h after sampling.

Mice subjected to retrobulbar sinus puncture were significantly less active than control mice for all three OFT parameters in Harikrishan et al. 2018 [12]. This was not observed for mice that underwent tail incision or facial vein puncture. An e ffect of anaesthesia was apparent since mice subjected only to isoflurane anaesthesia showed greater activity, and a higher number of centre entries, than mice subject to retrobulbar puncture under isoflurane anaesthesia [12]. This contrasts with the Tsai et al. 2015 study [8], where mice that underwent retrobulbar bleeding with anaesthesia performed similarly in the OFT to mice sampled without anaesthesia.

There were no di fferences in display of anxiety behaviour between facial vein-sampled, tail incision, sublingual and retrobulbar groups, as indicated by visits and time spent in the centre of the field [12]. Alternately, mice sampled by facial vein generally avoided the open field in the triple test with the authors concluding that they expressed more anxious behaviour than tail vein-sampled animals [2].

In Moore et al. 2017 [32], the facial vein group exhibited a significantly lower average speed in the OFT compared to tail amputation and incision groups. No other between-group di fferences were observed.
