Discriminating Reactive from Neoplastic Mandibular Lymphadenopathy in Dogs with Previously Treated Appendicular Osteosarcoma

A prospective, multi-institutional diagnostic-accuracy and competing-risks outcome study, with an embedded single-subject index case

Study type: Prospective observational cohort with nested diagnostic-accuracy evaluation (STARD-conformant) and an embedded N-of-1 index case

Referencing: Harvard (Cite Them Right)

Status: Study protocol — no results reported

Abstract

Background. Dogs that survive the perioperative period after amputation for appendicular osteosarcoma enter a surveillance phase in which any new mass is presumptively attributed to metastasis. In the mandibular region this attribution is anatomically questionable: the mandibular lymph nodes drain the head and oral cavity, not the appendicular skeleton, and the mandibular salivary gland is superimposed on them. Regional nodal metastasis was identified in only 4.4% of 228 dogs with appendicular osteosarcoma (Hillers et al., 2005), while only 17.0% of palpably enlarged mandibular nodes were metastatic on histology even in animals with a confirmed oral malignancy (Herring, Smith and Robertson, 2002).

Objectives. To estimate the diagnostic distribution of new mandibular-region masses in dogs with previously treated appendicular osteosarcoma; to quantify the incremental diagnostic accuracy of cytology, PCR for antigen receptor rearrangement (PARR) and flow cytometry against a histopathological reference standard; and to determine whether the mandibular diagnosis or the underlying osteosarcoma dominates subsequent mortality under a competing-risks framework.

Methods. Prospective multi-institutional enrolment of 136 dogs, each undergoing a standardised index-test panel and a histopathological reference standard, with survival follow-up to death or 730 days. Diagnostic accuracy is reported per STARD 2015 (Bossuyt et al., 2015). Survival is modelled with cause-specific hazards and Fine–Gray subdistribution models to accommodate the competing risk of osteosarcoma death.

Significance. The cohort has never been assembled. Clinicians currently reason about these masses from base rates borrowed from populations that do not resemble the patient in front of them, and owners are counselled toward palliative decisions on a presumption of metastasis that the available anatomical and epidemiological evidence does not support.

1. Introduction

1.1 The clinical problem

Canine appendicular osteosarcoma is the most common primary bone tumour of dogs and is micrometastatic in the majority of patients at the time of diagnosis (Bryan, 2024). Amputation resolves the primary lesion and the pain associated with it, but median survival after amputation alone is approximately 175 days, extending to roughly 279–317 days when adjuvant platinum-based chemotherapy is administered (Mauldin et al., 1988; McMahon et al., 2011; Frimberger, Chan and Moore, 2016). Between 29.5% and 43.2% of treated dogs are alive at one year, and 11.3% to 26.3% at two (McMahon et al., 2011; Frimberger, Chan and Moore, 2016).

Dogs in this surveillance phase are therefore under active suspicion of metastatic disease, and any new palpable mass is, in practice, interpreted through that lens. When the mass occupies the mandibular region, three distinct interpretations compete, and they carry radically different prognoses: metastasis of the original osteosarcoma; a second, unrelated malignancy; or a benign reactive, inflammatory or salivary process.

1.2 Why the default interpretation is probably wrong

Two lines of evidence argue against reflexive attribution to osteosarcoma. First, osteosarcoma disseminates predominantly by the haematogenous route to the lungs; regional lymph node metastasis was documented in only 10 of 228 dogs (4.4%) in the largest series addressing the question (Hillers et al., 2005). Second, the mandibular lymph nodes are not the regional basin for an appendicular tumour — they drain the head, oral cavity and pharynx.

Conversely, mandibular node enlargement is a common and usually benign finding. Herring, Smith and Robertson (2002) examined regional nodes histologically in 31 animals with oral or maxillofacial neoplasia and found that only 17.0% of palpably enlarged mandibular nodes contained metastatic disease; moreover, only 54.5% of animals that did have nodal metastasis had involvement of the mandibular node itself. Williams et al. (2024) reported that just 1 of 154 mandibular and superficial cervical nodes sampled during thyroid carcinoma staging was definitively metastatic (0.6%, or ≤1.9% including probable cases), concluding that routine cytology of these stations is low-yield. Consistent with this, sentinel node mapping studies in dogs with oral neoplasia demonstrate that the true sentinel node is frequently not the mandibular node (Bae et al., 2025).

An important caveat attaches to the primary base rate. The 4.4% figure of Hillers et al. (2005) is a proportion among necropsy-assessed dogs; animals that did not undergo complete nodal survey could carry undetected nodal metastasis, so the population-level rate is uncertain but the anatomical argument — that the mandibular node lies outside the drainage basin of an appendicular primary — is independent of that uncertainty. The present study is designed precisely to replace the transferred estimate with a measured one.

1.3 The competing differential

Multicentric lymphoma is the single most common cytological diagnosis obtained from canine lymph node aspirates, accounting for 351 of 1,274 canine samples (27.5%) in a large diagnostic-laboratory series (Amores-Fuster et al., 2015); the same series found 27.2% of canine samples non-diagnostic, a figure with direct operational consequences for any protocol relying on cytology. Canine lymphoma incidence is estimated at 20–100 cases per 100,000 dogs (Zandvliet, 2016).

Mast cell tumour is a relevant alternative in bully-breed dogs, and nodal involvement materially alters its prognosis: median survival fell from 1,722 days to 551 days in dogs with nodal metastasis in a series of 45 subcutaneous tumours (Cherzan et al., 2023), and stage I versus stage II disease corresponded to 6.2 years versus 0.8 years in an earlier cytological staging study (Krick et al., 2009). Oral mucosal mast cell tumours present with a high rate of nodal involvement (Elliott et al., 2016).

Metastasis from an occult oral primary is the possibility that most depends on examination quality. Oral melanoma with nodal metastasis carried a median survival of 406 days versus 534 days without, in 77 surgically treated dogs (Giacobino et al., 2025); stage-stratified survival ranges from 17–18 months for stage I to approximately 3 months for stage III (Bergman, 2007). Tonsillar carcinoma is the most aggressive of these, with an overall median survival of 126 days across 123 cases, rising to 381 days in non-metastatic disease (Treggiari et al., 2023), and 243 days with 40% one-year survival in a separate cohort (Grant and North, 2016).

Finally, the mandibular salivary gland occupies the same anatomical space. Salivary neoplasia is rare, at 15.3 cases per 100,000 dogs (Cray, Selmic and Ruple, 2020), but surgically excised salivary gland carcinoma has the longest median survival of any malignancy in this differential at 1,886 days, collapsing to 248 days where nodal metastasis is present (Bush et al., 2023). Non-neoplastic sialocele is surgically curable with recurrence rates of 0–7.1% depending on approach (Cinti et al., 2021; De Lorenzi et al., 2018), and is strongly associated with hypercortisolism (OR 15.56) and long-term glucocorticoid administration (OR 7.78) (Bae et al., 2023).

1.4 Knowledge gap

No study has reported the diagnostic distribution of new mandibular-region masses in dogs with previously treated appendicular osteosarcoma. Clinicians must therefore transport base rates from populations that differ from this one in age structure, prior treatment, immune status and surveillance intensity. The gap is not merely academic: in the absence of a defensible prior, an owner presented with a new neck mass in a dog with a cancer history is frequently counselled as though metastasis were established.

1.5 Aim

To generate the missing prior, to quantify how much each diagnostic test moves it, and to establish whether the new diagnosis or the pre-existing osteosarcoma governs survival.

2. Theoretical framework

The study is framed as a Bayesian diagnostic problem embedded in a competing-risks survival structure. The pre-test probability of each diagnosis is the quantity Section 1 shows to be unknown; each index test supplies a likelihood ratio; and the post-test probability determines both treatment and prognosis. Survival is then not a single-cause process — a dog may die of the mandibular diagnosis, of osteosarcoma, or of an unrelated cause — so naive Kaplan–Meier estimation of cause-specific mortality would be biased upward. The analysis plan in Section 5.5 addresses this explicitly.

3. Hypotheses

Five hypotheses are specified in advance, each with a null form and a pre-declared falsification criterion.

H1 (primary). H1: In dogs with previously treated appendicular osteosarcoma presenting with a new solitary mandibular-region mass, osteosarcoma metastasis accounts for ≤5% of final diagnoses. H0: The proportion attributable to osteosarcoma metastasis exceeds 5%. Falsification: H1 is rejected if the lower bound of the 95% Wilson confidence interval for the observed proportion exceeds 0.05. Rationale: 4.4% nodal metastasis across all stations in appendicular osteosarcoma (Hillers et al., 2005), with the mandibular node lying outside the drainage basin.

H2. H1: Cytology alone discriminates lymphoid neoplasia from reactive hyperplasia with accuracy ≥80%, but discriminates lymph node from salivary gland origin with significantly lower accuracy. H0: Cytological accuracy does not differ between the two discrimination tasks. Falsification: rejected if the difference in accuracy between tasks has a 95% CI excluding zero. Rationale: classification accuracy for lymphoma exceeded 80% for all six examiners across 161 nodal samples, while subtype accuracy fell as low as 20% (Martini et al., 2022).

H3. H1: Adding PARR to cytology increases diagnostic accuracy for lymphoid neoplasia relative to cytology alone. H0: PARR adds no incremental accuracy over cytology alone. Falsification: rejected if the paired difference in accuracy (McNemar) yields p ≥ 0.05. Rationale: PARR classified 162/163 samples (99%) as clonal and 56/67 (84%) as polyclonal concordantly with cytological or histological diagnosis in 271 patients (Waugh et al., 2016); sensitivity is assay- and tumour-dependent (Rout et al., 2025).

H4. H1: Among dogs whose mandibular mass is not osteosarcoma, time elapsed since amputation predicts overall survival more strongly than the mandibular diagnosis itself. H0: The mandibular diagnosis is the stronger predictor, or the two are equivalent. Falsification: rejected if the cause-specific hazard ratio for time-since-amputation does not exceed that for diagnosis category, by likelihood-ratio test on nested Cox models. Rationale: osteosarcoma mortality is concentrated in the first year post-amputation (McMahon et al., 2011; Frimberger, Chan and Moore, 2016).

H5. H1: The number of enlarged peripheral lymph node stations at presentation is monotonically associated with a final diagnosis of lymphoma. H0: Node-station count is independent of lymphoma diagnosis. Falsification: rejected if the Cochran–Armitage trend test yields p ≥ 0.05. Rationale: multicentric lymphoma is by definition multi-station, and is the most common nodal cytological diagnosis in dogs (Amores-Fuster et al., 2015).

4. Variables

4.1 Independent (predictor) variables

4.2 Dependent (outcome) variables

4.3 Covariates, confounders and effect modifiers

4.4 Derived analytic quantities

Sensitivity, specificity, predictive values and likelihood ratios are derived for each index test and target condition with 95% Wilson intervals. Post-test probabilities are computed by applying derived likelihood ratios to the pre-test distribution estimated under H1 — the study's principal translational product: a clinician holding a cytology result should be able to read the revised probability directly.

5. Methods

5.1 Design

Prospective, multi-institutional, observational cohort with a nested paired diagnostic-accuracy comparison in which every enrolled dog receives every index test and the same reference standard. Reporting follows STARD 2015 (Bossuyt et al., 2015).

5.2 Population

Inclusion. Dogs with histologically confirmed appendicular osteosarcoma treated by amputation, presenting with a new mandibular-region mass at least 30 days after surgery, whose owners consent to the full index-test panel and to follow-up.

Exclusion. Concurrent second malignancy diagnosed before day 0; mandibular mass present before amputation; systemic glucocorticoid within 14 days of aspiration (recorded and deferred rather than excluded where clinically feasible); owner unwilling to permit follow-up contact.

5.3 Index tests and reference standard

All index tests are performed on day 0 from the same lesion, using separate passes, interpreted independently and blind to one another and to outcome. The reference standard is histopathology with immunohistochemistry. Where surgical sampling is declined, a pre-specified composite reference standard of concordant dual-pathologist cytology plus documented clinical course to 180 days is applied; every case resolved this way is flagged and analysed separately in sensitivity analysis. This is a known source of partial verification bias and is declared rather than concealed.

5.4 Sample size

For H1, estimating a proportion of approximately 0.05 with a 95% confidence half-width of 0.04 requires 115 dogs; allowing 15% attrition gives a target enrolment of 136. Verified decision thresholds at n=136 (Wilson): 0 observed metastatic cases yields a 95% CI of 0–2.7% (H1 supported); 1 case yields 0.1–4.0% (supported); 2 cases yields 0.4–5.2% (indeterminate); 9 cases yields 3.5–12.1% (rejected). The study is therefore decisive if the true proportion is near 0–1% and ambiguous if it is near 4–5%; the protocol specifies pooling across institutions and accrual-window extension where necessary.

For H2/H3, estimating a sensitivity of 0.80 with a half-width of 0.10 requires 62 dogs per target condition; at anticipated lymphoma prevalence this exceeds the primary target, so these are powered as secondary objectives with confidence intervals reported in preference to hypothesis tests. For H4, detecting a hazard ratio of 2.0 at 80% power and α = 0.05 requires 66 deaths (Schoenfeld); a hazard ratio of 1.5 would require 191, which this cohort will not reach — a declared power limitation.

5.5 Statistical analysis plan

- H1. Proportion of each final diagnosis with 95% Wilson intervals; primary test is whether the interval for osteosarcoma metastasis excludes 0.05.

- H2, H3. Paired accuracy comparison by McNemar's test; sensitivity, specificity and likelihood ratios with Wilson intervals; inter-observer agreement by Cohen's κ.

- H4. Cause-specific Cox models and Fine–Gray subdistribution models with cumulative incidence functions by cause; nested models compared by likelihood-ratio test; proportional-hazards assumption examined by scaled Schoenfeld residuals, with time-varying coefficients where violated.

- H5. Cochran–Armitage trend test across ordered node-station categories, with multinomial logistic regression adjusted for Section 4.3 covariates.

- Multiplicity. H1 is the single primary comparison, tested at α = 0.05. H2–H5 are secondary, reported with Benjamini–Hochberg FDR control at 0.05.

- Missing data. Multiple imputation by chained equations for covariates missing at random; non-diagnostic index tests are retained as an observed category, never imputed.

5.6 Embedded index case

A single-subject prospective case is embedded within the protocol and follows the identical variable schedule. The index case is an 11-year-old female pit bull-type dog with appendicular osteosarcoma treated by amputation, in whom a solitary mandibular-region mass was first documented photographically on 4 September 2026 and was demonstrably absent from an image captured 34 days earlier on 1 August 2026. Retrospective onset bounding was performed by extracting DateTimeOriginal EXIF fields from nine owner photographs captured on a single device, establishing a defensible 34-day appearance window without recourse to owner recall. This illustrates the X4 measurement procedure and is proposed as a generalisable, zero-cost method for dating lesion onset in companion-animal oncology, where owner-reported timelines are otherwise the only available anchor and are known to be unreliable. The index case is reported descriptively, excluded from all pooled estimates, and no inference about the population is drawn from it.

6. Expected findings and their interpretation

The following are pre-registered expectations, not results.

If H1 is supported, the practical consequence is a change in counselling: a new mandibular mass in this population should not be presented to owners as presumptive metastasis, and definitive diagnosis should be pursued before prognostic statements are made. If H1 is rejected, the anatomical reasoning in Section 1.2 is wrong, and mandibular palpation should be added to routine osteosarcoma restaging. If H2 is supported, ultrasound should precede aspiration as a matter of routine, because the node-versus-salivary-gland distinction is the one cytology handles least well and imaging handles cheaply. If H4 is supported, prognostic conversations should lead with time since amputation rather than with the new diagnosis. A null result on H5 would indicate that single-station mandibular enlargement carries no protective information against lymphoma and that staging should be equally aggressive regardless.

7. Limitations

Partial verification. Owners of dogs with a terminal primary diagnosis frequently decline surgical biopsy. The composite reference standard mitigates but does not eliminate the resulting bias, which is reported transparently per STARD (Bossuyt et al., 2015).

Referral population. Enrolment through specialist oncology services over-samples owners who pursue treatment; survival estimates will exceed those of the general population and should not be quoted to primary-care clients without adjustment.

Power for secondary hypotheses. H2–H4 are underpowered at the primary target enrolment (Section 5.4); this is declared in advance.

Competing risk dominates follow-up. With a substantial fraction of dogs expected to die of osteosarcoma within the observation window, precision for mandibular-diagnosis-specific survival will be limited regardless of enrolment; mandibular-specific hazards will additionally be estimated on a survivor-selected (immune-selected) subset, a second selection bias declared here.

Uncertain reference base rate. The 4.4% of Hillers et al. (2005) derives from necropsy-selected dogs and may understate the population rate; this motivates direct measurement rather than limiting inference.

Breed heterogeneity. Somatic mutation burden and driver frequency vary by breed background in canine osteosarcoma (Sakthikumar et al., 2018; Megquier et al., 2022), and emerging multi-omic risk stratification (Garg et al., 2026) may render crude clinical stratification obsolete during the study's lifetime.

Cytology is imperfect by construction. Classification accuracy for lymphoma exceeded 80% but subtype accuracy fell as low as 20% in a six-examiner study (Martini et al., 2022); the protocol inherits this ceiling.

8. Ethical considerations and animal welfare

The protocol requires institutional animal care and use committee approval at every participating site and written owner consent. It is observational: no treatment is withheld, randomised or mandated, and every diagnostic procedure specified is one already indicated in the standard workup of a new mass. The only additional burden over standard care is the parallel aspirate passes required for PARR and flow cytometry, obtained during a single restraint episode. Quality of life (Y6) is measured rather than assumed, and a declining HRQoL trajectory triggers a clinician-initiated conversation about goals of care irrespective of study status. Follow-up is by owner contact and imposes no procedure on the animal.

9. Conclusion

A new mandibular mass in a dog that has survived amputation for appendicular osteosarcoma is routinely read as metastasis, and the available anatomical and epidemiological evidence suggests that reading is usually wrong — 4.4% nodal metastasis in appendicular osteosarcoma overall (Hillers et al., 2005), and 17.0% metastatic yield from palpably enlarged mandibular nodes even in dogs with confirmed oral cancer (Herring, Smith and Robertson, 2002). The cohort needed to replace that inference with a measured prior has never been assembled. This protocol specifies how to assemble it, what to measure, and how to analyse it under the competing risk that defines the population.

Treatment implications per diagnosis (for clinical context at presentation): reactive hyperplasia and sialocele are curable (dental therapy; gland ablation, 0–7.1% recurrence); multicentric lymphoma is managed with CHOP-based chemotherapy ± rabacfosadine alternation (Thamm et al., 2017) with medians in the 12–15 month range in responsive dogs (Kim et al., 2024; Lautscham et al., 2017); nodal-stage mast cell tumour requires excision with nodal dissection and kinase inhibition, with survival falling from 1,722 to 551 days on nodal involvement (Cherzan et al., 2023); oral melanoma is managed surgically with radiation and anti-CSPG4 vaccination (node-positive median 406 d; Giacobino et al., 2025); occult oral/tonsillar carcinoma carries the worst median survival (126 d; Treggiari et al., 2023); salivary carcinoma the best (1,886 d node-negative; Bush et al., 2023). Osteosarcoma metastasis itself is palliative-only — which is precisely the diagnosis the protocol argues should not be assumed.

References

Harvard (Cite Them Right) style. All entries resolved against the CrossRef REST API by DOI; the Williams entry has no registered DOI and is resolved against PubMed. No entry carried a retraction notice at the time of retrieval.

Amores-Fuster, I. et al. (2015) 'The diagnostic utility of lymph node cytology samples in dogs and cats', Journal of Small Animal Practice, 56(2), pp. 125–129. doi: 10.1111/jsap.12303.

Bae, J.Y. et al. (2023) 'Sialocele and Its Association with Hypercortisolism and Long-Term Glucocorticoid Treatment in Dogs: Retrospective Case–Control Study', Animals, 14(1), p. 120. doi: 10.3390/ani14010120.

Bae, S. et al. (2025) 'Sentinel lymph node mapping with computed tomography lymphangiography and intraoperative methylene blue peritumoral injection has a high detection rate with moderate agreement in dogs with oral neoplasms', Journal of the American Veterinary Medical Association, 263(12), pp. 1–9. doi: 10.2460/javma.24.11.0758.

Bergman, P.J. (2007) 'Canine Oral Melanoma', Clinical Techniques in Small Animal Practice, 22(2), pp. 55–60. doi: 10.1053/j.ctsap.2007.03.004.

Bossuyt, P.M. et al. (2015) 'STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies', BMJ, h5527. doi: 10.1136/bmj.h5527.

Bryan, J.N. (2024) 'Updates in Osteosarcoma', Veterinary Clinics of North America: Small Animal Practice, 54(3), pp. 523–539. doi: 10.1016/j.cvsm.2023.12.007.

Bush, K.M. et al. (2023) 'Outcomes and clinical features associated with surgically excised canine salivary gland carcinoma: A multi-institutional, retrospective, Veterinary Society of Surgical Oncology study', Veterinary Surgery, 52(3), pp. 370–378. doi: 10.1111/vsu.13928.

Cherzan, N.L. et al. (2023) 'Factors affecting prognosis in canine subcutaneous mast cell tumors: 45 cases', Veterinary Surgery, 52(4), pp. 531–537. doi: 10.1111/vsu.13944.

Cinti, F. et al. (2021) 'Complications between ventral and lateral approach for mandibular and sublingual sialoadenectomy in dogs with sialocele', Veterinary Surgery, 50(3), pp. 579–587. doi: 10.1111/vsu.13601.

Cray, M., Selmic, L.E. and Ruple, A. (2020) 'Salivary neoplasia in dogs and cats: 1996–2017', Veterinary Medicine and Science, 6(3), pp. 259–264. doi: 10.1002/vms3.228.

De Lorenzi, D. et al. (2018) 'Nasopharyngeal sialoceles in 11 brachycephalic dogs', Veterinary Surgery, 47(3), pp. 431–438. doi: 10.1111/vsu.12771.

Elliott, J.W. et al. (2016) 'Canine oral mucosal mast cell tumours', Veterinary and Comparative Oncology, 14(1), pp. 101–111. doi: 10.1111/vco.12071.

Frimberger, A.E., Chan, C.M. and Moore, A.S. (2016) 'Canine Osteosarcoma Treated by Post-Amputation Sequential Accelerated Doxorubicin and Carboplatin Chemotherapy: 38 Cases', Journal of the American Animal Hospital Association, 52(3), pp. 149–156. doi: 10.5326/JAAHA-MS-6315.

Garg, A. et al. (2026) 'Integrated Multiomic Profiling Enhances Risk Stratification and Prognostication in Canine Osteosarcoma', Clinical Cancer Research, 32(16), pp. 3668–3682. doi: 10.1158/1078-0432.CCR-25-4593.

Giacobino, D. et al. (2025) 'Influence of the extent of cervical lymph node dissection and lymph nodes metastases on prognosis in a cohort of dogs with oral malignant melanoma treated by surgical resection and adjuvant anti-CSPG4 electrovaccination: a retrospective study on 77 cases', Frontiers in Veterinary Science, 12. doi: 10.3389/fvets.2025.1616419.

Grant, J. and North, S. (2016) 'Evaluation of the factors contributing to long-term survival in canine tonsillar squamous cell carcinoma', Australian Veterinary Journal, 94(6), pp. 197–202. doi: 10.1111/avj.12444.

Herring, E.S., Smith, M.M. and Robertson, J.L. (2002) 'Lymph Node Staging of Oral and Maxillofacial Neoplasms in 31 Dogs and Cats', Journal of Veterinary Dentistry, 19(3), pp. 122–126. doi: 10.1177/089875640201900301.

Hillers, K.R. et al. (2005) 'Incidence and prognostic importance of lymph node metastases in dogs with appendicular osteosarcoma: 228 cases (1986–2003)', Journal of the American Veterinary Medical Association, 226(8), pp. 1364–1367. doi: 10.2460/javma.2005.226.1364.

Kim, T.H. et al. (2024) 'Clinical Outcome of Multicentric Lymphoma Treated with Cyclophosphamide, Doxorubicin, Vincristine, and Prednisolone (CHOP) in Small Breed Dogs', Animals, 14(20), p. 2994. doi: 10.3390/ani14202994.

Krick, E.L. et al. (2009) 'Cytological lymph node evaluation in dogs with mast cell tumours: association with grade and survival', Veterinary and Comparative Oncology, 7(2), pp. 130–138. doi: 10.1111/j.1476-5829.2009.00185.x.

Lautscham, E.M. et al. (2017) 'Comparison of a CHOP-LAsp-based protocol with and without maintenance for canine multicentric lymphoma', Veterinary Record, 180(12), p. 303. doi: 10.1136/vr.104077.

Martini, V. et al. (2022) 'Performance of lymph node cytopathology in diagnosis and characterization of lymphoma in dogs', Journal of Veterinary Internal Medicine, 36(1), pp. 204–214. doi: 10.1111/jvim.16326.

Mauldin, G.N. et al. (1988) 'Canine Osteosarcoma', Journal of Veterinary Internal Medicine, 2(4), pp. 177–180. doi: 10.1111/j.1939-1676.1988.tb00313.x.

McMahon, M. et al. (2011) 'Adjuvant Carboplatin and Gemcitabine Combination Chemotherapy Postamputation in Canine Appendicular Osteosarcoma', Journal of Veterinary Internal Medicine, 25(3), pp. 511–517. doi: 10.1111/j.1939-1676.2011.0697.x.

Megquier, K. et al. (2022) 'The genomic landscape of canine osteosarcoma cell lines reveals conserved structural complexity and pathway alterations', PLOS ONE, 17(9), e0274383. doi: 10.1371/journal.pone.0274383.

Meichner, K. et al. (2020) 'Multicenter flow cytometry proficiency testing of canine blood and lymph node samples', Veterinary Clinical Pathology, 49(2), pp. 249–257. doi: 10.1111/vcp.12843.

Rout, E.D. et al. (2025) 'Diagnostic Sensitivity of the PCR for Antigen Receptor Rearrangement (PARR) Assay for Canine Plasma Cell Tumors', Veterinary Clinical Pathology, 54(3), pp. 281–291. doi: 10.1111/vcp.70036.

Sakthikumar, S. et al. (2018) 'SETD2 Is Recurrently Mutated in Whole-Exome Sequenced Canine Osteosarcoma', Cancer Research, 78(13), pp. 3421–3431. doi: 10.1158/0008-5472.CAN-17-3558.

Thamm, D. et al. (2017) 'Alternating Rabacfosadine/Doxorubicin: Efficacy and Tolerability in Naïve Canine Multicentric Lymphoma', Journal of Veterinary Internal Medicine, 31(3), pp. 872–878. doi: 10.1111/jvim.14700.

Treggiari, E. et al. (2023) 'Tonsillar carcinoma in dogs: Treatment outcome and potential prognostic factors in 123 cases', Journal of Veterinary Internal Medicine, 37(1), pp. 247–257. doi: 10.1111/jvim.16623.

Waugh, E.M. et al. (2016) 'Optimisation and validation of a PCR for antigen receptor rearrangement (PARR) assay to detect clonality in canine lymphoid malignancies', Veterinary Immunology and Immunopathology, 182, pp. 115–124. doi: 10.1016/j.vetimm.2016.10.008.

Williams, C. et al. (2024) 'Cytologic findings in mandibular and superficial cervical lymph nodes of dogs with thyroid carcinoma', The Canadian Veterinary Journal, 65(9).

Zandvliet, M. (2016) 'Canine lymphoma: a review', Veterinary Quarterly, 36(2), pp. 76–104. doi: 10.1080/01652176.2016.1152633.