Between Prometheus and Thanatos: Reflections on Intracardiac Injection Models for Metastasis in Mice

Intraosseous (intratibial) injection.

This method involves directly injecting tumor cells into the bone (usually the tibia, though sometimes the femur) under general anesthesia. It guarantees a localized bone tumor at the injection site in the majority of animals. For example, injecting 10⁵ breast cancer cells into the mouse tibia leads to a readily measurable osteolytic tumor within 1 to 3 weeks. The major advantage is control and consistency, in that only the targeted bone develops a tumor, allowing studies of bone-tumor interactions without the noise of cancer spread elsewhere.⁷ From a welfare perspective, intratibial models avoid the acute complications of intracardiac injection: there is no risk of stroke or widespread tumor embolism. Early hindlimb paralysis is also avoided since tumor cells are confined to the bone rather than the spinal vasculature. However, intratibial injection is an invasive orthopedic procedure. It requires anesthesia and penetration into the bone, which causes local tissue damage and inflammation. Postsurgery, mice experience pain from the bone injury and expanding intramedullary tumor.^17,29 Studies⁷ note that intratibial injection “results in considerable damage to the bone cortex and marrow, promoting cytokine release and bone turnover during healing.” This local trauma can itself induce pain and potentially confound certain readouts (for example, bone remodeling) if not properly controlled. The direct manipulation of bone tissue and the subsequent tumor growth within the confined space of the bone marrow can lead to pain, which is often assessed through behavioral assays that measure changes in limb use and sensitivity to stimuli. The ability of the animal to bear weight on the injected limb may be compromised, and overall mobility can be affected. If the injection site is not adequately sealed, local tissue invasion by tumor cells and the formation of an extraskeletal tumor mass can occur, potentially causing additional pain and morbidity.³⁰ Analgesia is strongly recommended for intratibial models; guidelines equate the pain to that of a bone biopsy in humans and mandate at least 24 to 48 hours of analgesic coverage.¹⁸ Indeed, providing a nonsteroidal anti-inflammatory drug (NSAID) or opioid analgesic around the time of intratibial tumor cell implantation did not interfere with bone tumor growth in controlled experiments, but it did improve humane endpoints (treated mice had no worse tumor burden but presumably experienced less pain-related distress).¹⁷ Typically, intratibial models are terminated when the localized tumor causes severe bone destruction or impairment of the limb (for example, tumor-induced fracture or inability to use the leg) or if the tumor escapes to vital organs. Many intratibial studies report endpoint at approximately 3 to 4 weeks postinjection, when the tumor-bearing leg shows significant swelling and lameness. Compared with intracardiac models, systemic effects like body weight loss are milder in intratibial models (at least until very late stages or if metastasis occurs), because the tumor burden is localized.^1,30 Mice remain active and maintain weight longer, with pain largely manifesting as localized orthopedic discomfort (limping, reduced loading of the affected limb) rather than generalized cachexia. This model can, thus, be considered to result in moderately severe pain and distress, as it causes significant pain in the affected limb (which must be alleviated with analgesics), but it spares the animal from multiorgan failure. Importantly, intratibial models bypass the natural metastasis process; tumor cells are artificially placed in bone, so steps like intravasation, circulation, and homing are omitted.^7,29 While this limits the study of metastasis biology, it can be a refinement in terms of animal welfare for experiments focused solely on tumor–bone interactions or bone pain, since the animal is not subjected to the full burden of disseminated disease.

Intravenous or intra-arterial injection (tail vein or caudal artery).

Another alternative approach is to inject tumor cells into the circulation, aiming to seed bone metastases, without intracardiac injection and avoiding or minimizing acute trauma. Tail vein (intravenous) injection of cancer cells is relatively easy to perform and has a high survival rate through the procedure. However, in standard practice tail vein injection predominantly causes lung metastases, as the lungs are the first capillary bed encountered. Bone metastases from tail vein injection are rare with most cell lines, because few cells survive the lung passage to then reach bone. Researchers have developed bone-tropic variants of tumor cells to address this. For instance, repeated in vivo passaging of MDA-MB-231 breast cancer cells through bone lesions yielded sublines that produce bone metastases even after tail vein injection.³¹ One report⁷ notes that injecting such a clone via the lateral tail vein resulted in bone metastasis (primarily in hind limbs) in approximately 80% to 90% of mice, significantly improving reliability. Crucially, these intravenous/intra-arterial methods are not associated with the stroke or early paralysis risk seen in intracardiac injection.⁷ In other words, since the injection is performed in a peripheral vessel, the chance of accidentally flooding the brain with tumor clumps is minimized, and any initial misdistribution of cells tends to lodge in the lungs (which, while not the intended target, is less immediately fatal than a brain embolism).³¹ In terms of welfare, mice undergoing tail vein injection experience minimal procedure-related distress, typically just a brief needle stick under light restraint or anesthesia. Survival after injection is high, and there is no surgical recovery. The welfare burden is shifted from the procedure to the ensuing metastases: if bone lesions successfully form, the mice will exhibit similar skeletal pain and morbidity as in other models but often with a more restricted pattern.^7,11

The intracaudal arterial (tail artery) injection is a refinement of this approach: by injecting cells into the arterial circulation (via the caudal artery) rather than a vein, one can circumvent immediate entrapment in the lungs. Tumor cells enter the arterial system and circulate to peripheral tissues including bone. A 2018 study³² demonstrated that caudal artery injections produce robust bone metastasis in mice with higher efficiency and fewer off-target tumor formations. Notably, tail artery injections often favor metastases in hindlimb bones,^32,33 which can simplify monitoring (for example focusing on lameness in the rear legs) and potentially reduce the number of painful sites per animal. A key welfare benefit of these methods is that they often produce bone-only (or bone-predominant) metastasis without large primary tumors or extensive visceral metastases.⁷ This is supported by comparative data showing that caudal artery injection results in high rates of metastasis to bone, particularly the legs, lower spine, and pelvis, while substantially limiting spread to vital organs such as the lungs, liver, and brain. In contrast, intracardiac and intravenous models show widespread dissemination, with high lung involvement and high rates of brain, liver, and kidney metastases. Moreover, the caudal artery model is associated with markedly lower 2-week mortality (1% to 3%) compared with intravenous and intracardiac injection, both of which can reach 100% mortality within the same timeframe according to some studies.¹³ Depending on the tropism and biologic characteristics of the injected cell lines, disseminated tumor cells may colonize organs that share vascular features or anatomic proximity to the injection route. For instance, in experiments involving the breast cancer 4T1 cell line, there was a consistent tendency for tumor formation in the mammary fat pad region, most likely due to its innate tropism of the cell line for mammary tissue.¹³ This can reduce the overall health deterioration; for example, mice do not develop lung cell tumor seeding with resultant dyspnea, thereby avoiding one axis of distress. Nevertheless, once bone tumors grow, the severity converges with that of other bone metastasis models; and bone pain, risk of fracture, and necessity to euthanize when tumors compromise mobility or cause weight loss remain present. The time frame in successful intraarterial bone metastasis models is comparable to intracardiac models (bone lesions detectable in 2 to 4 weeks depending on the cell line used).^33,34 One consideration is that not all cell lines will form bone metastases via this route, and higher cell numbers or special cell variants are often needed. From the 3Rs (Replacement, Reduction, and Refinement) perspective, if achievable, intraarterial models are a refinement over intracardiac in that they reduce technical trauma and acute risk and warrant for a better tumor localization.¹³

From the author’s personal experience with the intra-arterial (caudal artery) model, several important practical and welfare considerations should be noted. Although technically similar to intravenous injection into the caudal vein, accessing the artery is more challenging due to its deeper location. As such, extensive hands-on training is essential to avoid misinjection, which can lead to unintended tumor growth in the tail. Because arteries bleed more than veins, hemostasis requires additional time and care. Even with fine needles (29 to 30 gauge), bleeding was considerable in our experience. Tumor localization was not always limited to hindlimb bones; in many cases, especially with triple-negative breast and prostate cancer cell lines, metastases also appeared in the pelvis and sacrum, increasing the potential for pain. Appropriate analgesia and clear humane endpoints are crucial. Primary lesions often developed at the epiphyseal and metaphyseal region of long bones of the hindlimb and progressed rapidly, sometimes causing spontaneous fractures. Monitoring for pain-related behavior (for example, grimace scale, gait abnormalities, grip strength) is advised. In advanced disease stages, even mild handling during intravenous injection can result in fractures due to compromised cortical bone. Isoflurane anesthesia may reduce this risk but makes venous access more technically demanding. In humanized NOD-SCID-Gamma mice with injected in the caudal artery with leukemia or lymphoma cell lines, pallor due to bone marrow infiltration and subsequent bone marrow failure was a reliable indicator of clinical decline and impending death, warranting timely euthanasia.

Orthotopic primary tumor with spontaneous metastasis.

Another approach is to implant tumor cells in their tissue of origin (orthotopic site) and allow metastasis to occur naturally. For breast cancer, this means injecting cells into the mammary fat pad of mice; for prostate cancer, injecting into the mouse prostate gland. Orthotopic models recapitulate the full metastatic cascade (primary tumor growth, intravasation, dissemination, and colonization) and have been reported in immunocompetent hosts (using murine tumor lines or grafts) to incorporate immune responses.³⁵ In terms of animal welfare, orthotopic implantation is a surgical or injection procedure but generally less invasive than intracardiac or intratibial routes. For example, a small incision to inject cells into the mammary fat pad is performed under anesthesia and heals quickly. Alternatively, the tumor cells can be injected directly percutaneously into the mammary fat pad.^13,34 The major welfare issue arises from the primary tumor burden. Orthotopic tumors will grow at the site of injection; in breast models, a mammary tumor can become quite large or ulcerate if not resected, causing pain and necessitating euthanasia per size limits before metastases have time to develop.³⁶ Indeed, a limitation of orthotopic breast cancer models is that bone metastases occur at low frequency and often after a long latency, because the primary tumor tends to cause the death of the animal first (or requires euthanasia when the size exceeds the humane endpoint).¹³ Standard breast cancer cell lines rarely give rise to spontaneous bone metastases from orthotopic mammary fat pad tumors in mice. The development of the 4T1.2 model provided one of the first demonstrations of reliable bone metastasis from a primary mammary tumor site.³⁵ Thus, many more animals are needed to observe a few bone metastases, and those that do get bone lesions also carry a huge primary tumor load. This increases overall morbidity: animals may experience pain from the primary tumor in addition to any metastasis-related pain.¹³ Routinely, primary mammary fat pad tumor resection is performed to allow metastases to emerge. While this can free the animal from the primary tumor burden, it entails an additional major surgery and recovery, and metastases may still take time to grow.³⁷ A comparative study¹³ reflected that fat pad inoculation without tumor resection, while yielding 100% tumor take, primarily resulted in tumor growth localized to the mammary fat pad region, with minimal spread to bone (16.67%) and virtually no involvement of the spine or pelvis. Metastases to vital organs are rare or absent, and bone lesions, when they occur, are typically limited to the leg. The 2-week mortality rate in this model was 0%, indicating a low overall disease burden and a more gradual progression. As such, the fat pad method represents a moderate severity model in terms of animal welfare. Further, this method tends to result in additive stress (large tumor + potential metastatic disease) and its utility in studying bone metastasis is limited due to low metastatic efficiency to the skeleton.¹³ Because of this, it is usually not the method of choice when bone metastasis is the experimental focus, and thus it is less frequently used in recent literature for bone metastasis specifically. When it is used, strict endpoint criteria (tumor size limits, body condition, and weight loss) should be applied to mitigate suffering, and those endpoints are usually reached before significant bone pain would ever occur.³⁶ For prostate cancer, spontaneous bone metastasis from an orthotopic prostate tumor in mice is likewise inefficient: transgenic models like TRAMP develop primary prostate tumors and lymph node/lung metastasis, but bone metastasis is very rare.¹ Exceptions include certain human prostate xenografts (for example, LAPC-9, LuCaP series) in SCID mice that have been reported to metastasize to bone in some instances, but, again, primary tumor morbidity is significant.²⁰ Orthotopic models are thus less useful for bone metastasis studies and can be more severe on animals unless the primary tumor is controlled. Other orthotopic models, such as the intraductal injection of cancer cells into the mouse mammary gland, have been developed to study breast cancer progression. However, these models often fail to fully recapitulate the metastatic process, particularly in terms of metastasis to bone. For instance, while intraductal injections can lead to tumor development within the mammary gland, they do not consistently result in bone metastases, limiting their utility in studying the full spectrum of metastatic disease.³⁸

Intracarotid (carotid artery) injection.

In this technique, tumor cells are injected into the internal carotid artery, which directly supplies the brain. The rationale is to deliver cells preferentially to the cerebral circulation, increasing the yield of brain metastases while avoiding initial passage through the lungs. Intracarotid injection is more invasive than the intracardiac method; it often involves exposing the carotid artery and, sometimes, temporarily ligating branches (like the external carotid) to direct flow to the brain.²² The procedure requires microsurgical skill but can be done with mice under anesthesia. In terms of welfare, intracarotid injection can be considered a refinement over intracardiac for brain-targeted metastasis due to several advantages.^22,39 First, it significantly reduces nonbrain metastases: since the majority of cells enter the brain circulation, fewer cells reach other organs. A 2023 study²⁶ used common carotid injection of 4T1 breast cancer cells to achieve tumor seeding in the brain (hippocampus region) as early as 7 days, with essentially no metastases in the liver or kidneys and only modest lung involvement. This suggests a more brain-specific disease, sparing the animal from simultaneous heavy tumor burden in multiple organs. In contrast, intracardiac injection can result in some mice experiencing stroke-like symptoms within hours if cells lodge in brain microvasculature improperly. The absence of early neurologic signs indicates that, when done correctly, intracarotid injection does not cause the kind of acute embolic brain injury that intracardiac injection sometimes does. In addition, mice only had a slight, transient weight loss from the surgery, which they recovered from in days,^26,39 showing that the surgical stress was manageable. However, as metastases grow in the brain over time, the welfare profile converges with that of other brain tumor models; ultimately, tumors in the brain will cause neurologic impairment and weight loss. The difference is that intracarotid models may allow a more controlled onset. For example, the study²⁶ above terminated at 14 days for analysis, with mice still neurologically intact. In one study,²³ mice injected via the internal carotid artery with MDA-MB-231 BR1-BR3 brain-seeking sublines became moribund between 41 and 59 days postinjection. Thus, intracarotid injection delays the emergence of symptoms to the time when the actual brain seeding with tumor cells forms, rather than causing any artificial early damage. This can be seen as a refinement, as animals might enjoy a longer symptom-free interval postinjection. Importantly, intracarotid models still necessitate close monitoring for neurologic deficits (when tumors eventually manifest) and carry the burden of a surgical procedure (incision in the neck, arterial manipulation). Proper perioperative analgesia and care of the surgical site are needed. The technical success rate might also be an issue; mishandling the carotid can lead to thrombus formation and stroke or fatal hemorrhage. Although the literature⁴⁰ suggests high brain-metastasis rates with intracarotid injection for breast cancer lines, one downside is potential variability in distribution, since some cells could still escape to systemic circulation; for instance, a proportion of mice developed lung metastases after carotid injection of 4T1 cells.²⁶ Intracarotid injection could be considered as a medium-refinement option, in that it reduces the initial procedure-related risk and nonbrain disease but still leads to a lethal brain disease that must be managed humanely. It is particularly advantageous for brain-tropic cell lines or patient-derived xenografts when precise and efficient seeding of the brain is required.⁴⁰

Intracranial (direct brain) injection.

This method involves directly implanting tumor cells into the brain parenchyma (for example, injection into the mouse striatum or cortex through a small burr hole in the skull). It is essentially creating a localized “metastatic” brain tumor without the actual metastatic process. Intracranial injection is widely used in neuro-oncology to model brain tumors (including metastases) because it guarantees that all animals have a tumor in the brain. From a welfare standpoint, intracranial implantation is an acute surgical intervention: under anesthesia, a tiny hole is drilled in the skull and a fine needle delivers the cell suspension into the brain. The surgery is typically quick (several minutes) and done aseptically; mice usually recover well from the procedure itself with adequate postoperative care. The presence of a growing tumor in the brain, however, will typically lead to neurologic deficits and morbidity as in other models.⁴¹ In direct intracranial injection models, the timeline of tumor growth can vary depending on the number of cells injected and their aggressiveness. In one study,⁴² stereotactic injections of 4T1 cells into the mouse brain produced substantial tumor burden within 10 days. The authors⁴² also showed that reducing the inoculum by 10-fold did not change the overall pattern of tumor distribution, although detailed effects on tumor progression or survival were not fully explored. The welfare profile of intracranial tumor cell injection models is characterized by localized brain effects. Because the tumor is confined to the brain, animals do not suffer from peripheral tumor burdens (no lung, liver, or bone lesions). This means weight loss may be less pronounced until the late stage, and issues like anemia or organ failure are minimal. The sole focus is neurologic function. Mice will be monitored for signs like impaired motor coordination, seizures, lethargy, head pressing, or paralysis.^36,41 When the tumor is small, mice appear normal; as it enlarges, they will start showing body weight loss, will start losing balance, or will show focal deficits (for example, one-sided weakness if the tumor affects one hemisphere’s motor cortex). Seizures can sometimes occur in intracranial injection models, depending on tumor location and phenotype and are a clear sign of distress requiring intervention (euthanasia if severe/frequent).⁴³ One benefit is that intracranial injection models often produce a single tumor mass rather than many micrometastases, so the progression might be more gradual and detectable (for example, a steady decline in motor function rather than sudden multiple impairments). On the other hand, an intracranial tumor can cause acute increases in intracranial pressure or hemorrhage if it grows rapidly, potentially leading to abrupt clinical downturns. Generally, the humane endpoint in intracranial models is reached when the mouse exhibits persistent neurologic symptoms, such as circling, ataxia, inability to right itself, along with other indicators like severe weight loss due to neurologic impairment, abnormal posture, seizures, or significant behavioral changes including reduced activity or grooming.^44,45 The severity of intracranial models ranges from moderate to high, depending on how aggressive the tumor is. If a very fast-growing tumor is implanted, the animal might deteriorate quickly (high severity, short survival). If a more slowly growing or smaller tumor is used, the animal might have a relatively longer period of mild symptoms before progression (moderate severity). Importantly, the initial surgical insult is considered moderate: a small cranial surgery is painful but can be managed with proper analgesia, and recovery is usually rapid if no complications occur. Analgesic coverage should be ensured for at least 48 hours postcraniotomy to manage pain associated with scalp incision and skull drilling.^18,41 Patient-derived xenograft (PDX) models of brain metastasis have also been developed, in which fragments of human metastatic brain tumors are implanted directly into the mouse brain or beneath the calvaria. These models represent a form of intracranial graft and involve similar procedural considerations, including stereotactic neurosurgery for implantation and subsequent monitoring for neurologic deficits or disease progression. In some cases, preconditioning is needed for PDX engraftment, which could add systemic side effects (like weight loss or susceptibility to infections) as a new welfare factor.⁴⁶

The animal welfare parameters associated with bone and brain metastasis models are depicted in Table 1.

Procedure-related trauma.

Intracardiac injection involves a blind or guided cardiac puncture, which, even in skilled hands, carries a notable risk of acute complications (approximately 10% mice suffer stroke or death).⁷ In contrast, peripheral injection techniques such as caudal tail vein or artery administration are associated with no procedural mortality. Intratibial and intracranial surgeries, while inherently more invasive, are generally well tolerated when performed under aseptic conditions with appropriate surgical technique, with complications being infrequent and typically minor. Intracarotid injection represents an intermediate level of invasiveness; although technically demanding, it can also be performed with minimal acute mortality when executed with precision.²⁶

Distribution of disease.

Intracardiac injection floods the circulation with tumor cells, often leading to widespread metastases (bones, lungs, liver, adrenal, and occasionally brain).^10,13 This multiorgan involvement means the animal may experience multiple concurrent health issues, such as bone pain, respiratory compromise, and possible neurologic issues. Indeed, one study²² noted metastases in bone, lung, liver, ovaries, and adrenal glands altogether in mice after intracardiac injection of a brain-seeking line. By contrast, as shown above, refined models are generally designed to limit metastases to one or 2 organ systems. Intratibial is localized to bone only; intracarotid largely confines disease to the brain (with a potential minor lung seeding); tail vein injection typically confines tumors to the lung (not desired for bone studies but relevant to lung metastasis studies); tail artery injection typically confines tumors to the hindlegs and lower spine; and intracranial injection confines tumors to the brain. Importantly, a focused metastatic pattern reduces the overall systemic burden on the animal. For example, a mouse bearing bone metastases following intracardiac injection may experience both substantial skeletal pain and respiratory distress due to concurrent pulmonary metastases. In contrast, a mouse subjected to intratibial injection typically presents with localized bone pain while maintaining normal respiratory and systemic function. Similarly, mice being used as intracardiac brain metastasis models may harbor occult visceral metastases, such as in the liver, potentially leading to weight loss and acute hepatic dysfunction, clinical features that are less likely to occur in models using intracarotid injection, which more selectively target the brain. Therefore, some methods represent alternatives that present a lower frequency of simultaneous severe symptoms, the severity being more compartmentalized, an important refinement.^17,22,26,29

Time to endpoint (survival).

Intracardiac metastasis models are highly aggressive and are often associated with the shortest time to humane endpoints due to rapid and widespread disease progression.^7,16 Some refined models can extend the time course. For instance, intracranial injection can be tuned to yield a longer survival if needed by adjusting cell numbers, and intracarotid injection might offer a slightly prolonged presymptomatic phase.^48,49 However, intratibial models can sometimes progress faster in the injected bone.⁵ In one sense, a shorter time to endpoint could be regarded as reducing pain and distress, but it can also mean a faster onset of intense symptoms without much transitional period. Models that extend survival slightly may allow more gradual disease progression and possibly more time to intervene or treat symptoms. Nevertheless, prolonged survival is only a welfare benefit if the animal’s quality of life is maintained; otherwise, it may prolong suffering unnecessarily. Several models allow a relatively good quality of life for a longer fraction of the study. For example, in the intracarotid 4T1 model, mice showed no symptoms for at least 2 weeks²⁶ and in other studies²³ became moribund with 41 to 59 days after implantation; whereas in some intracardiac brain models, clinical signs appeared as early as 2 to 3 weeks.^11,34 Thus, the alternatives to the intracardiac injection model can delay the onset of severity, giving researchers an opportunity to apply humane endpoints at the first sign of distress.

Pain and distress.

Bone metastasis models inherently involve pain, as bone lesions activate pain pathways. It should be noted that, in terms of observable animal welfare parameters such as pain behavior, signs of distress, and endpoint criteria, xenograft and syngeneic models often yield comparable profiles. Differences typically arise from the biologic behavior of the tumor in an immunocompetent or immunocompromised host, such as time to onset, metastatic spread, and aggressiveness. For instance, both immunocompetent and immunodeficient mice exhibit similar pain-related behaviors in bone metastasis models when the tumor burden is equivalent. Similarly, neurologic signs in brain metastasis models appear to correlate more with tumor localization and size than with host immune background. Therefore, while the immunologic context is critical for interpreting scientific outcomes, its impact on direct welfare assessment may be limited.^1,5,13,30,43 The intensity and manageability of pain can differ.⁵ Intracardiac bone models produce multiple lesions, including in weight-bearing bones and the spine, causing severe pain that is diffuse and harder to fully alleviate.⁵⁰ Intratibial models concentrate pain in one limb, and this can make it easier to recognize (for example, limping on one leg) and potentially easier to alleviate with analgesic strategies or to decide on euthanasia when that limb’s use is lost. Some refinement in bone metastasis models is seen in practice. For example, some protocols will prescore limb use and decide that if a mouse completely loses leg function, it is euthanized even if the mouse is otherwise healthy, thus preventing further suffering in a timely manner.^14,17 In intratibial models, use of analgesics is a measure that, as shown, does not impede tumor growth but does improve welfare.¹⁷ In intracardiac models, a mouse might still be using all limbs but has moderate generalized pain, which is more challenging to assess quantitatively and manage.^11,36 Neurologic distress in brain tumor models may result from progressive motor deficits, particularly when tumors are implanted in or near the motor cortex. In addition, impaired cerebrospinal fluid drainage due to tumor-associated remodeling of meningeal lymphatic vessels can contribute to increased intracranial pressure, potentially leading to symptoms such as cephalalgia and neurologic decline, as observed in some murine glioma models.^27,44 Some refinement in intracranial tumor models can be achieved through environmental enrichment strategies, such as providing enriched housing conditions, including enhanced social interaction, nesting material, and sensory stimulation, as this has been shown to mitigate neurologic decline and may even influence tumor progression and survival in glioma-bearing mice.^51,52 Such interventions can reduce stress-induced exacerbation of symptoms and support overall animal wellbeing throughout the disease course. However, while enrichment represents a refinement, the use of analgesics in response to pain should not be viewed as optional but rather as a mandatory obligation; animals exhibiting signs of pain or distress must be treated promptly and appropriately.¹⁸ In general, the frequency of severe pain or distress-related behaviors is highest in intracardiac injection models, as mice typically progress to a moribund state due to the widespread dissemination of tumor cells and resulting multiorgan involvement. In contrast, the presented refined models tend to induce a more localized disease burden (for example, isolated bone pain or isolated neurologic impairment), resulting in a narrower and more defined spectrum of distress behaviors.

Refinement and reduction potential.

Refined models can also be considered an application of the 3R principles by enabling more controlled experimental conditions and potentially reducing the number of animals required. For instance, intracranial injection typically results in a high and consistent tumor take rate within the brain, minimizing variability between animals. Intracranial models allow for more predictable outcomes, thereby potentially reducing the number of animals needed to achieve statistical significance.⁵³ Similarly, intratibial injection results in a highly consistent establishment of bone tumors across the majority of animals, owing to the direct delivery of tumor cells into the bone microenvironment. In contrast, intracardiac injection can lead to greater variability in metastatic patterns, with some mice failing to develop bone lesions due to preferential seeding of tumor cells in nonskeletal tissues.^7,13 Using models with more predictable outcomes means fewer animals used and less repetition thus allowing overall reduction in animal use. In terms of refinement, techniques such as ultrasound-guided intracardiac injection represent an advancement for intracardiac injection, as they enhance procedural accuracy and significantly reduce the incidence of misinjection-related complications, thereby improving overall animal welfare. Imaging guidance ensures the needle is in the ventricle and not invading other structures, thereby cutting down the likelihood of stroke and hemothorax.⁸ In addition, adjunctive techniques such as in vivo bioluminescence imaging or other imaging techniques allow earlier detection of metastases,¹⁰ thereby allowing endpoints to be implemented at the first sign of significant tumor burden rather than waiting for overt clinical decline. Early imaging can thus refine endpoints (for example, if imaging shows extensive spine metastases, one might euthanize before paralysis occurs, preemptively).

Imaging for severity assessment and longitudinal monitoring.

Beyond procedural refinements, advances in noninvasive imaging technologies offer additional opportunities to both refine welfare through enhanced monitoring and thereby reduce animal use by enabling detailed, longitudinal assessment of disease severity within the same cohort of animals. Besides the well established in vivo bioluminescence imaging,¹⁰ other minimally invasive imaging modalities offer validated tools to monitor tumor progression in both brain and bone metastasis models while enhancing animal welfare by reducing the need for experimental endpoint sacrifices and, thus, the number of animals needed. For example, in vivo micro-CT technologies have been reported to enable longitudinal quantification of osteolytic bone lesions in breast cancer intratibial models, with detectable changes as early as one week postinoculation and without altering disease progression,⁵⁴ while in vivo micro-CT combined with matched postmortem histopathology provides a refined, precise method for tracking lesion development over time.⁵⁵ MRI, especially T2-weighted or diffusion-weighted sequences, allows repeated noninvasive assessment of brain tumors and metastases, offering anatomic details and accurate monitoring of tumor volumes and vascular invasion.⁵⁶ Advanced MRI protocols have been successfully applied to noninvasively detect and monitor tumor burden in the brain but also the metastatic burden in visceral organs in mouse models.⁸ In comparative studies⁵⁷ of metastasis across organs, MRI showed greater sensitivity than micro-CT for small lung lesions (approximately 0.6 mm compared with 1 mm), and outperformed CT in detecting metastases in kidneys, bones, ovaries, and adrenal glands. Combining micro-CT with MRI offers a complementary approach; that is, while micro-CT provides high-resolution bone and lung imaging, MRI enhances detection sensitivity across soft tissues, enabling comprehensive whole-body monitoring of metastatic spread.