Retrospective Study Evaluating Outcomes Following Palliative Radiotherapy With or Without Chemotherapy for Dogs With Presumed Cardiac Hemangiosarcoma

ABSTRACT

Canine cardiac hemangiosarcoma (cHSA) represents a complex clinical challenge in that those afflicted have an acute risk of death due to cardiac tamponade and high morbidity and mortality given the frequency of metastasis. Previous studies show that radiation therapy (RT) can decrease the risk of tamponade; however, an optimal approach has yet to be determined. The objective of this study was to evaluate the outcomes of dogs with presumptive cHSA treated with varied RT protocols and modalities, and to contrast findings with previously published literature. Secondary objectives were to assess differences in outcomes between those that received chemotherapy post‐RT or did not, single fraction versus multiple‐fraction RT protocols, and CT‐guided versus manually calculated treatment plans. Twenty‐seven dogs with echocardiographic evidence of an atrial or auricular cardiac mass that received RT were included. The frequency of pericardiocentesis before and post‐RT were compared. Overall survival time was determined, along with survival time specific to those that received chemotherapy, were treated with CT‐based radiation plans, and were prescribed a single fraction versus multiple fractions. Pericardiocentesis was performed an average of 1.1 times per week before RT, and an average of 0.18 times per week after RT (p = 0.01). Median overall survival time was 137 days. Plans made without CT guidance were associated with more adverse radiation events, but all were minimally impactful on quality of life. Most dogs died or were euthanized due to metastatic disease. This study shows similar benefits to previously published data in a larger cohort of dogs using a less‐conformal radiation modality. As well, it highlights future directions to identify optimal systemic therapies to delay the onset of metastasis.

1. Introduction

Canine hemangiosarcoma (HSA) is a tumour composed of malignant bone marrow precursor cells or endothelial cells, with primary sites most commonly including the spleen, right atrium, subcutaneous or cutaneous sites [1, 2]. Prognosis varies from very good with surgical excision of a cutaneous phenotype to poor or grave in the case of splenic or cardiac presentation [1, 2, 3]. Ultimately, the aggressive metastatic nature of visceral hemangiosarcoma leads to early death despite varied available interventions [1, 2]. It has been reported that around 29% of cardiac presentations have splenic involvement, and 8%–24% of splenic HSA also have cardiac masses, with the higher end of this estimate based on post‐mortem necropsy rather than at the time of initial diagnosis [1, 2]. Hemangiosarcoma is 10 times more common than other cardiac tumour types; other common sites of metastasis include lungs, liver, peritoneum and kidneys [1, 2, 3]. Cardiac hemangiosarcoma (cHSA) is most frequently found in the right atrium or right auricle, with additional reports describing presence or extension into the right atrioventricular junction or tricuspid valve [3, 4, 5, 6]. Ultrasonographic sensitivity for imaging diagnosis of a mass at either of these two most common sites is 95% for right atrial masses and only 60% for right auricular masses [7].

Correlates to canine hemangiosarcoma are familial and sporadic types of angiosarcoma (AS) in humans, which has shown to have genetic transcriptional programmes shared between tumours in dogs and people with significant overlap in the patterns of mutations detected in each species [8, 9, 10]. Similar to dogs, human AS can present as radiation‐ or solar‐induced cutaneous lesions though this subtype is genetically distinguishable from the more aggressive spontaneous form in both species [10]. In contrast to the canine counterparts, human cardiac AS is incredibly rare, with a reported incidence of around 4.7% of all AS patients [11]. Similar to cHSA, there are no successful interventions in most cases for people as the tumour is small and often not amenable to surgical resection with overall median survival time (MST) around 4 months for the familial version and a 5‐year survival of 14% for the sporadic type [9].

The cardiac presentation of canine hemangiosarcoma is one of particular interest due to the high risk for acute mortality given the propensity of the mass to cause pericardial effusions leading to tamponade and resultant compression to the cardiac chambers; without treatment, survival times are around 7–12 days [7, 12]. Surgery, when possible, has a reported MST of 4 months [5]. Another study describes surgery versus surgery and chemotherapy, which achieves around 42 days survival without chemotherapy or 175 days with adjuvant chemotherapy [13]. Monotherapy with doxorubicin alone has shown MST of around 116–140 days [12, 14]. Two smaller studies have shown that the use of ionising radiation to right auricular masses significantly reduces the frequency of pericardial effusions, thus prolonging survival [15, 16].

The goal of this retrospective study was to evaluate and describe outcomes of dogs with presumptive cHSA treated with varied RT protocols and modalities. Secondary objectives were to assess differences in outcome between the groups that received chemotherapy post‐RT or did not, fraction number (single vs. multiple) and CT‐guided versus manually calculated treatment plans.

2. Methods

2.1. Case Selection and Record Review

Records were reviewed for dogs presented to a single institution between July 2012 and February 2023 with suspected cHSA on echocardiogram that underwent RT. Echocardiography by a clinician on the Cardiology service was pursued in each case to further characterise the mass and approximate dimensions before treatment. Data were recorded from electronic medical records including patient signalment, current medications, concurrent conditions, previous pericardiocentesis events and status of pericardial effusion, radiation protocol administered, follow‐up imaging and overall survival time. Continuous 24‐h Holter monitor or continuous telemetry data before or after radiation was recorded. Dogs were not excluded if there was confirmed or highly suspected evidence of metastasis or concurrent disease. Any dog with lesions in the spleen, liver, lungs or lymph nodes were assumed to have metastatic hemangiosarcoma unless aspirate or biopsy supported a benign process or concurrent neoplasia. Dogs were excluded if the heart base mass was not located in the right atrium, right auricle, right AV groove or adjacent to these regions as these were assumed to be non‐hemangiosarcoma tumours. If the mass was diagnosed several months or more before treatment and documented to be slow‐growing, these cases were also excluded due to the lessened likelihood of hemangiosarcoma. Cases were censored if alive at the time of data collection or at the time of last follow‐up if there were no available records of subsequent visits to the referring veterinarians or reporting institution. Finally, necropsy reports were reviewed and dogs with non‐hemangiosarcoma tumours at the heart base were excluded.

All dogs were stable at the time of initial consult with the Oncology service, with those that presented in tamponade receiving pericardiocentesis by their primary vet or an emergency service before referral. Families were given the option to pursue computed tomography (CT) or thoracic radiography for radiation planning before RT. The dose was recorded for the lung and heart volumes in each case with a CT‐based plan. Concurrent or adjuvant chemotherapy was also recorded if administered. Evidence of radiographic radiation‐induced pneumonitis, pulmonary fibrosis or cardiac arrhythmias after RT was also recorded. Any adverse radiation events (ARE) that developed within the region of the treated field were considered acute radiation toxicity if within 3 months of treatment, and any developments outside of that time frame were considered late effects per the Veterinary Radiation Therapy Oncology Group 2023 criteria (VRTOG v2.0) [17]. Subgroups of those who died due to primary versus metastatic disease were evaluated. Death or euthanasia due to metastasis was presumed when records noted progressive pulmonary nodules, hemoabdomen, progressive neurologic signs or combinations of findings that would support a disseminated metastatic process before death or euthanasia; however, survival times were only calculated for those with confirmed dates of death and euthanasia, and those lost to follow‐up were censored. Any dog euthanized due to recurrent pericardial effusion was categorised as death due to primary tumour. Where available, tumour response was assessed using ultrasonographic or CT imaging and evaluated according to Response Evaluation Criteria in Solid Tumours [18].

2.2. CT

In the subset of dogs that received CT scans, each underwent general anaesthesia. Anaesthetic protocols varied based on the patient but generally included pre‐medication with a benzodiazepine and opioid, induction with propofol and maintained on gas inhalant. All dogs were positioned in sternal recumbency using an RT immobilisation setup via a uniquely conformed Vac‐Lok bag (Civco Systems, IA, United States). CT was acquired using a Philips Gemini TF Big Bore 16‐slice scanner (Philips Medical Systems, Nederland, B.V.) or a Siemens Somatom Force 256‐slice scanner (Siemens Healthineers, Erlangen, Germany). Pre‐contrast helical CT images were acquired, then 2 mL/kg of iohexol contrast agent (Omnipaque 350, GE Healthcare, Princeton, NJ, USA) was injected intravenously before postcontrast images were acquired. All image sets were reconstructed at 2.0 mm slices with a 512‐matrix using the smooth algorithm. Most dogs were manually hyperventilated to induce transient apnea before CT images acquired within the expiratory phase of the breathing cycle, a technique referred to as ‘manual breath hold’, to eliminate motion artefact from normal respiratory movement. This was a standardised approach for thoracic CT studies to enable accurate radiologic interpretation of thoracic organs.

2.3. Radiation Planning—Computer‐Based

Pre‐ and post‐contrast CT image sets were imported into the treatment planning software (Varian Medical Systems Inc., Palo Alto, California ¹ ). Gross tumour volume (GTV) was contoured to delineate mass volume from the adjacent normal cardiac tissue and cardiologists or cardiology residents verified tumour margins based on echocardiogram imaging with more challenging or small tumours. Organs at risk (OAR) included the heart, lungs, trachea, oesophagus and spinal cord; these were also contoured for use in normal tissue constraints. A clinical target volume (CTV) of 0–5 mm isotropic expansion was used based on clinician judgement, which was designed to include presumed microscopic extension of the tumour. The tumour volume was then expanded by 3–10 mm beyond the GTV or CTV to form the planning target volume (PTV), which accounted for interfraction and intrafraction movement as well as minor setup discrepancies. In some patients treated with a stereotactic protocol, the PTV was cropped to exclude sections of overlap with any of the sensitive OARs (including trachea and oesophagus) based on clinician assessment on a case‐by‐case basis. The three‐dimensional conformal radiation therapy (3D‐CRT) radiation plans were designed with two to four 6 MV coplanar static beams that utilised multi‐leaf collimators (MLCs) to conform to the tumour volume while minimising dose to the adjacent structures at risk (Figure 1a). No field‐in‐field or mounted wedge accessories were used to block the field in any case. Some dogs were treated with VMAT plans, which did not include CTV margination and included one to two partial or full rotational arcs (Figure 1b). Dose protocols ranged from one to five fractions with a total dose of 12–30 Gy. Planning objectives included 99% of the GTV (D99) and 95% of the PTV (D95) receiving 100% of the prescribed total dose. Plans were considered acceptable when the dose constraints to the OARs were met along with achieving the target dose objectives. All VMAT plans passed quality assurance (QA) check overseen by a medical physicist. QA was performed by gamma analysis using the Varian portal dosimetry system on VMAT arcs. A passing QA score was required with a minimum of 95% gamma for a 3 mm distance to agreement and a 3% absolute dose difference. Radiation data were collected in line with previous recommendations and is provided in the Supporting Information [19].

FIGURE 1.

(a) Left, a typical dose colour wash for a 3D‐CRT plan with four static beams. Prescribed dose was 12 Gy × 1 = 12 Gy. Orange to red hues represent 10–13 Gy, green represents 4–8 Gy and blue indicates < 3.50 Gy. PTV margin (outer magenta contour) is 3 mm symmetric expansion from the GTV (inner red contour). (b) Dose colour wash for a Volumetric Modulated Arc Therapy (VMAT) plan with two 360° coplanar arcs (fields not shown). Prescribed dose was 6 Gy × 5 = 30 Gy on consecutive days. Orange to red hues represent 30.5–33.2 Gy, green represents 13–30.5 Gy and blue indicates < 13 Gy. PTV margin (outer magenta contour) is 5 mm symmetric expansion from the GTV (inner red contour).

2.4. Radiation Planning—Manually Calculated Plans

Patients without CT scans before RT had plans calculated using lateral thoracic radiographs. Moderate collimation via the jaws of the gantry head achieved a rectangular target field at the base of the heart without use of MLCs, ensuring tumour coverage by centring the target over the approximate region of the right atrium. Patient positioning and final collimation adjustments were based on single‐view megavoltage port films before treatment. The treatment field was determined by ensuring inclusion of the cranial and cranioventral region of the heart as it appeared on port films, with the intention of complete dose coverage of the potential mass region in the right atrium, right auricle or adjacent sites (Figure 2). Monitor unit calculations were performed by two independent clinicians including one American College of Veterinary Radiology board‐certified Radiation Oncologist and verified using computer‐based calculation models (Microsoft Excel, Varian Eclipse). Dose was delivered in two coplanar beams at 0° and 180° with a 600 cGy/min dose rate and 6 MV photon energy with the patient positioned in lateral recumbency.

FIGURE 2.

An MV port film used for manually calculated treatment. Yellow box indicates the treatment field determined by the cardiac silhouette and adjacent landmarks.

2.5. Radiation Treatment—VMAT and 3DCRT Plans

All dogs underwent general anaesthesia for RT using a standardised protocol of induction using a benzodiazepine or opioid with propofol titrated to effect, then maintained on isoflurane inhalant anaesthetic. Monitoring equipment included indirect oscillometric blood pressure, end‐tidal capnography, EKG and pulse oximetry. Patient positioning was verified daily immediately before treatment using on‐board imaging (OBI) kilovoltage cone beam computed tomography (CBCT; Trilogy, Varian Systems). No respiratory motion control was used for radiation treatment, as the tumour target at the heart base moves minimally relative to the respiratory cycle and any intrafractional movement was addressed by the symmetric PTV expansion.

2.6. Statistical Analysis

Overall MST was calculated and comparative survival curves were generated using Kaplan Meier log‐rank (Mantel‐Cox) analysis. Survival was defined as the time between the start of RT to death or humane euthanasia from any cause. Dogs lost to follow up were censored at the time of the last recorded recheck. Frequency of pericardiocentesis events before and following RT were assessed using Wilcoxon signed rank testing. Univariate analysis between subgroups was evaluated using Fischer's exact testing and Gehan–Breslow–Wilcoxon testing. Categorical variables included computer‐based plans versus manually calculated, chemotherapy and single versus multiple fraction protocols. All analysis was conducted using GraphPad Prism 9.5.1 (GraphPad, San Diego, CA). A p‐value of 0.05 was used to determine significance.

3. Results

Twenty‐seven dogs met the inclusion criteria with no dogs excluded. Breeds included three Labradors, three Golden Retrievers (GRs), two German Shepherds, seven mixed breed dogs and one of each of the following: Australian shepherd, Pitbull, Bernese Mountain Dog, French Bulldog, Mastiff, Husky, Australian Terrier, Chihuahua, Bichon Frise, Miniature Schnauzer and Portuguese Water Dog. The average age was 10 years (range 4–15.4 years). There were 10 spayed females, 14 castrated males, one intact female and two intact males with a median weight of 30.7 kg (range, 2.7–65). All dogs received an echocardiogram before treatment that identified a mass near the right atrium, auricle or atrioventricular groove. Twenty‐six of the 27 (96%) dogs had pericardial effusion before their presentation for radiation and one dog had an incidentally found right‐atrioventricular groove mass during an echocardiogram to assess a low‐grade heart murmur. Twenty‐two of the 27 (81%) dogs had undergone pericardiocentesis due to moderate effusion or tamponade and 20/27 (74%) had evidence of pericardial and/or pleural effusion at the time of treatment. No dog had ante‐mortem diagnosis of hemangiosarcoma before treatment; all dogs were presumptively diagnosed via echocardiogram. Five dogs had 24‐h Holter monitoring during their course of treatment: two before RT establishing a normal baseline, and three after RT to recheck previously diagnosed arrhythmias (ventricular premature complexes in two, ventricular tachycardia in one) or to establish a new baseline in anticipation of possible additional RT. One dog was on anti‐arrhythmic medication (mexiletine; 4 mg/kg q12h) before RT and was continued on the medication following Holter monitor assessment 1 week post‐RT due to persistent ventricular ectopy and potential risk of fatal ventricular tachycardia.

Before RT, 25 dogs (92.5%) had thoracic and abdominal imaging performed, and two dogs were not fully staged. Twenty of the 26 dogs (76%) had confirmed or suspected metastasis of the spleen, liver, lymph nodes or lungs on either thoracic radiographs, CT or abdominal ultrasound before treatment. Of the 20 dogs with reported lesions, five had splenic lesions or nodules, six had pulmonary changes detected on thoracic radiography, eight had pulmonary lesions noted on CT, three had liver lesions and four dogs had lymphadenomegaly. Only one dog had pulmonary masses that were aspirated as carcinoma and this dog had a concurrent history of hepatocellular carcinoma, which was suspected to be the primary source. Three of four dogs had lymphadenomegaly of the sternal, tracheobronchial or cranial mediastinal lymph nodes, with one of the three dogs having a previous history of coccidioidomycosis. The fourth dog had all three enlarged lymph node sites as well as hepatic lymph node enlargement. One dog had a sternal lymph node sampled and additional stains were most consistent with mesothelioma versus atypical HSA, so the dog was included to track outcome. Another only had mild sternal and cranial mediastinal lymphadenomegaly, and aspirates did not show evidence of neoplastic cells, so this dog was excluded from the metastasis subgroup.

Of the 27 dogs, 11 were treated with 3D‐CRT, 4 with VMAT and 12 received a manually calculated plan. Twenty of 27 dogs (74%) received a 12 Gy × 1 protocol with either CT‐guided or manual planning; the other seven dogs received 15–30 Gy in one to five fractions. Two dogs did not complete the intended multiple‐fraction protocol due to acute decompensation during treatment. Information regarding dosimetry data is summarised in Table 1.

TABLE 1.

Dose information for CT‐guided plans, median listed with range.

Total dose	12 Gy	15 Gy	30 Gy	30 Gy
Protocol	12 Gy × 1	15 Gy × 1	10 Gy × 3	6 Gy × 5
Number of patients	10	1	2	2
Volume GTV (cm3)	8.6 (0.3–168)	5.5	7.05 (5.7–8.4)	67.15 (53.2–81.1)
Volume PTV (cm3)	60.35 (4.9–441.7)	17.1	15.2 (13.8–16.6)	130.45 (89–171.9)
D98% GTV (Gy)	12.14 (12.02–12.59)	14.98	29.33 (27.56–31.09)	30.72 (30.57–30.87)
D50% GTV (Gy)	12.51 (12.15–12.73)	15.29	31.93 (31.78–32.09)	31.49 (31.3–31.69)
D2% GTV (Gy)	12.76 (12.26–13.04)	15.4	32.43 (32.05–32.81)	32.33 (31.87–32.79)
D95% PTV (Gy)	11.89 (11.57–12.28)	14.72	28.76 (27.4–30.09)	29.89 (28.98–30.80)
D50% PTV (Gy)	12.33 (2.6–12.73)	15.19	31.34 (31.24–31.43)	31.44 (31.32–31.56)
D2% PTV (Gy)	12.9 (12.28–13.05)	15.4	32.37 (32.00–32.74)	32.43 (32.14–32.71)
Mean heart dose (Gy)	2.07 (0.96–6.35)	2.59	12.15 (11.67–12.62)	9.69 (8.36–11.01)
Mean heart‐PTV dose (Gy)	1.59 (0.85–5.39)	2.38	21.09 (10.9–31.25)	7.01 (6.69–7.33)
Mean lung dose (Gy)	0.638 (0.25–1.19)	0.43	5.77 (4.06–7.49)	3.78 (3.2–4.37)
V20Gy lung (%)	0	0	6.6 (4.2–9.0)	4.3 (2–6.6)
D2% trachea (Gy)	6.13 (2.76–12.01)	0.34	20.27 (11.6–28.94)	13.33 (11.34–15.33)
D2% oesophagus (Gy)	5.39 (0.55–12.36)	8.85	15.23 (8.23–22.22)	16.52 (14.2–18.83)
Conformity index	0.56 (0.19–0.89)	0.5995	0.75 (0.63–0.88)	0.83(0.81–0.86)
Homogeneity index	0.08 (0.05–0.12)	0.0546	0.12 (0.06–0.17)	0.08 (0.04–0.12)

Abbreviations: D2%: dose (Gy) to 2% of organ volume; GTV: gross tumour volume; Gy: grey; PTV: planning target volume.

Pre‐treatment mass sizes were measured via either echocardiogram or CT in 24 dogs (89%) and the average size was 9.26 cm² (range 0.7–46.2, median 5.9 cm²). Three‐dimensional CT measurements were standardised to two‐dimensional approximations to match with echocardiogram measurements by using the two longest dimensions in any one viewing plane to avoid underestimating mass size. Of the 24 dogs with pre‐RT measurements, nine dogs had at least one and up to five follow‐up echocardiograms; the post‐RT measurements of this subgroup averaged 5.47 cm² (range 0–18.49, median 3.57 cm²) and was significantly decreased from pre‐RT measurements (p = 0.016). Two dogs had a complete response (CR) at 28 and 100 days post‐RT; another had a partial response (PR) at 56 and 197 days post‐RT. Two others had stable disease (SD) at 25 and 28 days post‐RT. One dog had SD on 2‐ to‐3‐month serial recheck echocardiograms until progressive disease (PD) was noted at 334 days post‐RT. Out of the nine dogs with follow‐up imaging based on best response, 22% had a CR, 33% had a PR, 33% had an SD and 11% had aPD.

Seven (26%) dogs experienced recurrent pericardial effusion necessitating pericardiocentesis following RT. Six of the seven (88%) occurred within the first 30 days post‐RT and the other (14%) was at 51 days post‐treatment. Four of these dogs (57%) required one pericardiocentesis and did not have additional episodes of effusion; two (29%) were ultimately euthanized due to immediate re‐effusion following pericardiocentesis and one dog (14%) was euthanized at the time of diagnosis of recurrent pericardial effusion. Four dogs had non‐pericardial effusion events in the post‐treatment period in the pleural (2), peritoneal (1) and both peritoneal and pleural (1) spaces.

AREs within this cohort included six dogs with follow‐up imaging that detected changes within the heart or lungs. All dogs with suspected AREs were diagnosed within the first 3 months after RT and were classified as acute toxicity. Three dogs (50%) had grade one pulmonary toxicity, all of which were subclinical. Two dogs (33%) had Grade I cardiac toxicity and one (17%) had potential Grade III cardiac toxicity. Two dogs that each had suspected Grade I cardiac toxicity (supraventricular tachycardia [SVT] and ventricular ectopy) had CT‐based plans with a mean dose to the heart of 1.96 and 2.36 Gy (see Table 2), and both received doxorubicin post‐RT. The dog with ventricular ectopy also had a history of ventricular tachyarrhythmia before RT, so there is potential that this was not radiation‐induced. As well, the dog who had Grade III cardiotoxicity characterised by SVT and dilated cardiomyopathy (DCM) also had received doxorubicin and was a large breed dog (mastiff) treated with manually calculated RT. Four of six dogs (66%) had manually calculated plans without dose data to the OARs. No dogs exhibited Grade III toxicity of the lungs.

TABLE 2.

Mean doses to heart and lung volumes from the CT‐guided plans.

Plan	Cardiac dose (mean, cGy)	Pulmonary dose (mean, cGy)	VRTOG score
3DCRT	617.8	118.7
3DCRT	635	30.6
3DCRT	189.2	62.1
3DCRT	236.6	46.2	Grade 1 cardiac
3DCRT	196.3	67	Grade 1 cardiac
3DCRT	224.5	84.7
3DCRT	312	65.5
3DCRT	96.6	25.5
3DCRT	99.6	41.6
3DCRT	160.6	55.1
VMAT (SBRT)	835.8	319.7
VMAT (SBRT)	1101.6	437.3
VMAT (IMRT)	221.5	99.7
VMAT (SBRT)	1167.5	406.1
VMAT (SBRT)	1244.1	881.9
Median dose	236.6	67

Note: Of the two dogs that developed Grade I cardiac acute toxicity, both received mean total doses that fell in the median range of all dogs treated.

Abbreviations: 3DCRT: three‐dimensional conformal radiation therapy; cGy: centrigray; IMRT: intensity‐modulated radiation therapy; SBRT: stereotactic body radiation therapy; VMAT: volumetric modulated arc therapy; VRTOG: Veterinary Radiation Therapy Oncology Group.

The overall MST was 137 days (Figure 3; range: 2–652 days). Median follow‐up time in the lost‐to‐follow‐up group (8 dogs) was 160 days (2–652 days). Two dogs were alive at the time of data collection. Sixteen dogs received chemotherapy following RT and four received additional rescue protocols after PD with the first‐line chemotherapeutic (summarised in Table 3). MST was 141 days for dogs that received any chemotherapy versus 36 days without chemotherapy (p = 0.91).

FIGURE 3.

Overall MST of all dogs was 137 days (2–652 days). Time in days was calculated from the last radiation treatment to death or euthanasia. Tick marks indicate the time point at which patient was censored.

TABLE 3.

Chemotherapy agents given as first‐line treatment post‐RT and additional agents elected at the time of progressive disease (PD).

Post‐RT chemotherapy protocol	First line (# dogs)	Second line (# dogs)	Third line (# dogs)	Fourth line (#dogs)
Vinblastine ± propranolol	7
Carboplatin ± propranolol	5	1
Doxorubicin	3
Toceranib phosphate	1
L‐MTP‐PE		1		1 a
Mitoxantrone		1 a
Cyclophosphamide		1	1 a

Abbreviation: L‐MTP‐PE: muramyl tripeptide‐phosphatidyl ethanolamine.

^a Same patient.

Pericardiocentesis events before RT and post‐RT were evaluated on an event‐per‐week approach and the mean pre‐treatment value was 1.1 events per week (95% confidence interval: 1–1.5) and the mean post‐treatment value was 0.18 events per week (95% CI: 0–0.25). MST for the subgroup that had recurrent pericardial effusion compared to the group that did not re‐effuse varied widely with 75‐day median for the dogs that re‐effused versus 262 days for the dogs that did not (p = 0.04). MST between groups that received a single fraction treatment versus multiple was not found to be significant (137 vs. 37 days, p = 0.83) nor was CT‐guided RT versus manual calculations (137 vs. 133 days, p = 0.9). Within the CT‐guided RT plans, those treated with VMAT had an MST of 29 days, whereas the dogs who were treated with 3DCRT had a MST of 316 days (p = 0.2).

MST of the group with confirmed or suspected metastatic lesions before RT (18/26, 69%) was 127 days compared to 142 days among the eight dogs (31%) without suspected hemangiosarcoma metastasis (p = 0.56). Eighteen of 27 dogs (66%) were euthanized or deceased due to suspected metastasis.

4. Discussion

This group of 27 dogs with suspected cHSA were treated with radiation in a variety of protocols, the most common being a single‐fraction 12 Gy protocol. Consistent with previous publications, there was a significant reduction in the need for pericardiocentesis following treatment with radiation in any protocol [15, 16]. While in this cohort there was a 26% incidence of re‐effusion post‐RT, previous studies evaluating monotherapy doxorubicin have also shown similar decreases in re‐effusion post‐treatment (25%–32% re‐effusion incidences) [12, 14]. This suggests future prospective, randomised comparisons are necessary to conclude that RT alone confers this benefit. Twenty‐six dogs did not experience any periprocedural complications and the radiation and anaesthesia protocols were well‐tolerated. Two of six dogs with imaging diagnoses of AREs were treated with a CT‐guided plan and four had manually calculated plans. Of the two that had documented cardiac AREs with a CT‐guided plan, the dose to the heart was within cardiac tolerance and they received doxorubicin at eight and 16 days post‐RT (Table 2) [20].

AREs were assessed against dose conformity of the RT delivered and found the majority of patients that developed radiation‐induced injury were treated with the less conformal, manually‐calculated plan (66% vs. 33%). The two groups were not found to be statistically different; however, four of the 12 dogs treated with manually calculated parallel‐opposed fields showed evidence of Grade I lung ARE in the post‐RT period (33%). Ultimately, none of the dogs developed severe enough ARE to have affected quality of life and none required medical intervention. With the ease of access and reduced expense for a manually calculated treatment plan over a CT‐guided plan, there may be benefit in reducing dose to the adjacent lung fields via use of field blocking with MLCs in an irregularly calculated field size for the risk‐averse owner. None of the dogs treated here with a manually calculated plan had blocked fields. The risk of missing these small lesions with a blocked field likely outweighs the benefit in blocking OARs that are unlikely to impact outcome. The GTV/PTV targets can vary widely based on patient conformation, so it is important to account for this when planning the treatment field (Figure 4). The convention used for the majority of dogs within this study was to include the region ventral to the trachea and extending down to where the cardiac silhouette meets the sternum.

FIGURE 4.

Examples of patients with CT‐based plans that show variance in position of GTV/PTV target. Sagittal reconstructions with the dose colour wash shows 95% of the prescription dose (D95%) or higher covering the target. PTV (magenta) contour is a symmetric expansion of the GTV (red) ranging from 3 to 10 mm.

Though it was not statistically significant, dogs that were treated with 3DCRT lived longer than those with a more conformal VMAT plan, which suggests the possibility that less conformal plans inadvertently treat the extension of microscopic disease, leading to a survival benefit compared to more conformal approaches. Given that the PTV expansion ranged from 3 to 10 mm, it is possible that margins on the lower end of this range are too narrow and chance a geographic miss of the tumour cells, despite the right side of the heart having limited motion. This discrepancy could be addressed by including a symmetrically expanded CTV margin in the VMAT planning process or considering a wider, more judicious PTV margin.

Adjuvant chemotherapy protocols used within this patient cohort varied widely given the longevity of the retrospective inclusion window. There were not enough patients in each subgroup of chemotherapy treatment to effectively compare for efficacy or outcome. However, there is enough information within this cohort to say that metastasis was the most common life‐limiting factor; only four out of 27 dogs (14.8%) were euthanized due to recurrent, refractory pericardial effusion. Further investigation into the most effective chemotherapeutic remains subject to ongoing research; recent evidence supporting the use of vinblastine and propranolol post‐RT has achieved encouraging results in a small cohort that did not have evidence of pulmonary metastasis before treatment [15]. The addition of propranolol is gaining enthusiasm due to the pre‐clinical success with antagonism of the overexpressed beta‐2 receptor signalling in neoplastic endothelial cells, which has also been demonstrated clinically in a variety of human cancers [21]. The data in veterinary hemangiosarcoma is limited, and clinical trials are underway to assess the clinical benefit of propranolol; a small study of five dogs with hemangiosarcoma (cardiac not included) showed that the combination of an anthracycline (generally doxorubicin) and propranolol was well‐tolerated and achieved a clinical benefit of 80% [22].

The MST between the group with metastatic lesions versus those without confirmed or overt metastatic lesions was overall similar (123 vs. 142 days). In necropsies of dogs with metastatic cHSA in this study and previous studies, there were no reported metastatic thoracic lymph nodes noted, but one case in 51 had sternal lymph node metastasis seen on cytology [7, 15, 16]. Because lymph node metastasis appears to be so infrequent, lymphadenopathy noted on staging diagnostics should not be a contraindication for aggressive treatment. Given the insignificant difference between dogs with metastatic disease and those without, the presence of pulmonary lesions detected on CT at the time of RT may not correlate with a shorter survival. Ultimately, these observations are based on a very small sample size and should be validated with larger future cohorts.

Genomic research on canine visceral HSA has shown that while there are DNA copy number aberration (CNA) profiles that are conserved within the subgroups of breed‐specific visceral hemangiosarcoma patients, there were no highly recurrent somatic CNA signatures that could be considered hallmarks of this neoplasm. Interestingly, there were substantial differences found when comparing GR visceral HSA to that of other dogs. Australian Shepherds had increased evidence of ANGPT4, a pro‐angiogenic factor, compared to other breeds in this study (70% vs. mean of 31% in all others). As well, GR visceral HSA has been shown to have decreased incidence of VEGFA gain compared to the other breeds (22% vs. 40%) [23]. Additional studies have demonstrated the wide and extensive intra‐tumoral heterogeneity among visceral HSA cases, as well as the strong possibility that there are distinct subtypes within this neoplasm [24]. Although there has not been extensive study in the genomic signatures of cardiac HSA, if the same characteristics are shared by cHSA it may be reasonable to extrapolate that the varied outcomes seen clinically are due in part to the existence of more or less aggressive subtypes. Furthermore, next generation sequencing assays performed on canine visceral HSA found that, though human AS is mutationally complex compared to HSA, some subtypes of HSA resemble spontaneous AS and suggest potential for a canine comparative oncology model [10].

The use of doxorubicin in cHSA is controversial when combined with radiation. Doxorubicin has a known cardiotoxic effect when an accumulated lifetime dose is reached, and clinical manifestation of this toxicity can include arrhythmias in the acute setting and decreased fractional shortening progressing to DCM [25]. Additionally, tumours that invade the cardiac tissue can also disrupt the precise and coordinated signalling required to maintain typical cardiac functionality, leading to syncope, premature ventricular complexes, ventricular tachyarrhythmias, third‐degree AV‐block and supraventricular tachyarrhythmias [6, 26, 27, 28]. This characteristic of cardiac tumours makes interpretation of subsequent cardiotoxicity, whether secondary to the tumour, radiation‐ or chemotherapy‐induced, indubitably challenging. Dogs with newly developed SVT post‐RT were classified under the VRTOG cardiac grading, whereas large breed dogs that developed DCM post‐RT and post‐doxorubicin were included in VCOG cardiac grading [29].

Due to the retrospective nature of this study, limitations inherently included low case numbers and wide variation in radiation protocols and chemotherapy agents. Survival time and the number of post‐RT re‐effusion events were based on time to euthanasia or natural death. The decision to euthanize versus proceeding with additional pericardiocenteses was a subjective determination by the families rather than a biologically determined end point, which could have affected the conclusions on outcomes. Additionally, follow‐up imaging was limited and mainly included echocardiograms, as few families are willing to pursue recheck CT scans for their dogs with such a guarded prognosis. Additionally, measurements from echocardiography were approximate and images obtained were sonographer‐dependent. The goal in including all dogs treated with a suspected cHSA was to evaluate outcome and efficacy, as well as the tolerance of side effects in a less conformal plan compared to CT‐guided. Despite a seemingly inclusive cohort, there were not enough dogs in each subgroup for statistically significant relationships. As with previous publications, another weakness lies in the lack of confirmatory diagnosis and the decision to include versus exclude dogs that have cardiac masses that did not behave in the typical aggressive metastatic pattern. Whereas the authors have been careful to exclude any dog that had a histologic diagnosis of anything other than hemangiosarcoma, there is undoubtedly risk that some of the dogs included in this study may have evaded exclusion due to the challenge of obtaining cellular or tissue samples from a mass at the right heart base.

5. Conclusions

This study shows that RT in a variety of protocols and intensities was able to improve the recurrence of pericardial effusion and reduce the need for urgent pericardiocentesis for suspected cHSA. Additionally, there is a risk of re‐effusion post‐RT especially in the first month for approximately 26% of dogs, and it would be prudent to monitor clinical signs at home as well as serially repeat cardiac point‐of‐care ultrasonography on a 1–2 week basis post‐RT for that first 1–2 months to ensure no further effusions arise. Though this data does not show a benefit with chemotherapy, metastatic spread was the cause of death or euthanasia in most dogs. This provides ample evidence that RT has a role in local disease management for cHSA, is well tolerated using manually calculated plans, and supports future endeavours to optimise chemotherapeutic protocols to control systemic disease.

Ethics Statement

The dogs reported in this study were client‐owned pets with naturally occurring cancer. In all cases, informed consent was obtained from the clients to treat their dogs with the reported protocol. All dogs were treated according to the principles and practices of specialty medicine prevailing at the time of their treatment.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Wakamatsu C. N., Potter B. M., Leary D., Boss M.‐K., and Martin T. W., “Retrospective Study Evaluating Outcomes Following Palliative Radiotherapy With or Without Chemotherapy for Dogs With Presumed Cardiac Hemangiosarcoma,” Veterinary and Comparative Oncology 23, no. 3 (2025): 432–441, 10.1111/vco.13068.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.