Relationship between radiation dose and cerebral microbleed formation in dogs with intracranial tumors

Abstract

Background

Cerebral microbleeds (CMBs) are a possible sequela in human brain tumor patients treated with radiation therapy (RT). No such association is reported in dogs.

Objectives

To investigate whether CMBs occur in dogs after radiotherapy, and if there is an association between number and dose, and an increase over time.

Animals

Thirty‐four client‐owned dogs irradiated for primary intracranial neoplasia. ≥2 magnetic resonance imaging (MRI) scans including susceptibility‐weighted imaging (SWI) were required.

Methods

Retrospective, observational, single‐center study. Cerebral microbleeds identified on 3 T SWI were counted within the entire brain, and within low‐ (<20 Gy), intermediate‐ (20‐30 Gy), and high‐ (>30 Gy) dose regions. A generalized linear mixed‐effects model was used to analyze the relationship between the CMBs count and the predictor variables (irradiation dose, time after treatment).

Results

Median follow‐up time was 12.6 months (range, 1.8‐37.6 months). Eighty‐three MR scans were performed. In 4/15 dogs (27%, 95% CI, 10%‐52%) CMBs were present at baseline. ≥1 CMBs after RT were identified in 21/34 dogs (62%, 95% CI, 45%‐77%). With each month, the number of CMBs increased by 14% (95% CI, 11%‐16%; P < .001). The odds of developing CMBs in the high‐dose region are 4.7 times (95% CI, 3.9‐5.6; P < .001) greater compared with the low‐dose region.

Conclusion and Clinical Importance

RT is 1 possible cause of CMBs formation in dogs. Cerebral microbleeds are most likely to occur in the peritumoral high‐dose volume, to be chronic, and to increase in number over time. Their clinical relevance remains unknown.

Abbreviations

AIC

Akaike Information Criterion

CI

confidence interval

CMBs

cerebral microbleeds

CT

computed tomography

GLMM

generalized linear mixed‐effects model

GTV

gross tumor volume

IDL

isodose line

MRI

magnetic resonance imaging

RT

radiation therapy

SDMA

symmetric dimethylarginine

STROBE‐Vet

Strengthening the Reporting of Observational Studies in Epidemiology—Veterinary

SWI

susceptibility‐weighted imaging

TE

echo time

TR

repetition time

1. INTRODUCTION

Radiation therapy (RT) plays an essential role in the management of primary intracranial tumors in dogs. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ In the case of large lesions or high radiation dose, substantial proportions of normal brain tissue receive relevant amounts of collateral dose. Consequently, adverse effects might develop over time. These adverse effects, also termed “normal tissue toxicity”, are at least partly caused by vascular injury. ⁹ In human brain tumor patients, vascular injury has been associated with neuropsychological sequelae such as cognitive decline ¹⁰ , ¹¹ , ¹² and is accompanied by a deterioration in quality of life. ¹³ Risk and severity of cognitive impairment increase if certain areas such as the hippocampus are irradiated, and with higher RT doses and larger irradiated volumes. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Microscopically, the tissue response to radiation starts with characteristic changes of the vessels, and of the white matter, with demyelination. ⁹ In the case of RT‐induced vascular injury, cerebral microbleeds (CMBs) develop, with hemosiderin as a degradation product of hemoglobin accumulating in the brain. These CMBs can often histologically be identified as perivascular clusters of hemosiderin‐containing macrophages. ¹⁹

Clinically, cerebrovascular manifestations of late effects such as CMBs are increasingly recognized in vivo with magnetic resonance imaging (MRI), particularly with gradient‐echo sequences such as T2*‐ or susceptibility‐weighted imaging (SWI). ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ These sequences are sensitive to the distortion of the local magnetic field caused by paramagnetic (eg, hemosiderin; contains atoms with unpaired electrons) or diamagnetic (eg, calcifications; contains atoms with paired electrons) compounds. ²⁶ , ²⁷ , ²⁸ Given careful image interpretation and consideration of mimics, CMBs are well‐defined lesions detectable by noninvasive MRI techniques with high sensitivity and specificity as well as high reliability. ²⁸ , ²⁹ , ³⁰ In irradiated human brain tumor patients, CMBs first appear in the high‐dose brain areas and increase in number over time, especially with higher RT doses delivered to larger brain volumes. ²⁰ , ²¹ , ²² , ²³ Patients with the same disease (glioma) that were not irradiated, did not develop CMBs. ²⁰

If and to what extent CMBs occur after RT in dogs has not been investigated so far. In a large series of dogs with various diseases, 9.3% had CMBs. ³¹ The imaging changes were analogous to those described in humans. ³¹ , ³² In dogs, the prevalence of CMBs increases markedly in those >10 years of age, is more common in small dogs and in dogs with kidney disease. ³¹

The purpose of this retrospective study was to evaluate the occurrence of CMBs in dogs after RT for intracranial tumors, using SWI. It addresses the following questions: If dogs treated with RT for intracranial neoplasia develop CMBs, is there (1) an association between the number of CMBs and radiation dose, and (2) is there an increase of CMBs over time?

2. MATERIALS AND METHODS

2.1. Study design

Ours was a retrospective, observational, single‐center study conducted at the Division of Radiation Oncology and the Clinic for Diagnostic Imaging of the University Animal Hospital, Vetsuisse Faculty, University of Zurich, Zurich, Switzerland. Our study was reported according to STROBE‐Vet (Strengthening the Reporting of Observational Studies in Epidemiology—Veterinary) guidelines for observational studies. ³³

2.2. Case selection and image acquisition

Medical records of dogs with an imaging diagnosis of intracranial neoplasia, ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ treated with definitive‐intent RT (10 × 4 Gy, 10 × 4 Gy and additional 11% limited to the gross tumor volume (GTV) [“boost”], or 10 × 4 Gy with deliberate heterogeneous radiation dose delivery within the tumor) between 2016 and 2022 served as the basis of the investigation. Dogs that underwent a 2nd course of RT were included only until the start of reirradiation.

For study inclusion all dogs had to have at least 2 MRIs (1 before treatment and at least 1 follow‐up MRI, or ≥2 follow‐up MRIs). In addition to a minimum set of MRI sequences, ⁴² each MRI needed to include SWI.

Magnetic resonance imaging was acquired as previously described, ⁴³ with a 3 T magnetic field scanner (right‐handed) using a 15‐channel dS HeadSpine coil, or an 8‐channel dS SmallExtremity coil (Philips Ingenia, Philips AG, Zurich, Switzerland). For the 3‐dimensional (3D) SWI, repetition time (TR) was 31 ms, echo time (TE) 1st 7.2 ms, 1TE = 6.2 ms, 4 echoes, flip angle 17°, and slice thickness 2 mm. The sequence was acquired in the transverse plane. The CT examinations for radiation treatment planning were performed with a 16‐slice scanner (Philips Brilliance 16, Philips AG, Zurich, Switzerland).

2.3. Radiation isodose levels

Three isodose volumes were created as structures in the Eclipse planning system (version 15.1 Varian Oncology Systems, Palo Alto, California) to classify CMBs depending on their location in different dose regions. The isodose volumes of interest (low‐, intermediate‐, and high‐dose level) were chosen according to Wahl et al, ²³ but adjusted to isoeffective levels considering our fraction size of 4 Gy. For isoeffectiveness of the fraction size used in dogs ⁴⁴ (biologically effective dose, BED), we assumed an (α/β)‐ratio of 2 for brain late toxicity: The low‐dose level (<30 Gy in Wahl et al ²³ ) corresponds to a <20 Gy isodose line (IDL; <50% IDL of the used 10 × 4 Gy protocol), and the high‐dose level (>45 Gy in Wahl et al ²³ ) corresponds to >30 Gy IDL (>75% IDL of the used 10 × 4 Gy protocol).

An example with these specified volumes is shown in Figure 1.

FIGURE 1.

Sagittal reconstruction of the planning computed tomography (CT) acquired in the equilibrium phase after intravenous injection of contrast medium of a dog treated for a (presumed) meningioma along the falx cerebri in the right frontal region. The 3 isodose volumes (green contour: Low, <20 Gy; yellow contour: Intermediate, 20‐30 Gy; orange contour: High, >30 Gy) as well as the gross tumor volume (GTV, pink contour) are depicted.

2.4. Microbleed definition and count

Microbleeds were identified on SWI as small (<10 mm), ²⁶ discrete, extratumoral foci of susceptibility (signal voids) that did not correspond to vessels on consecutive transverse slices. Calcifications were excluded by using phase images included in the SWI sequence and comparing the MR to the available CT scans (strongly hyperattenuating on the latter). Computed tomography scans were available in all dogs at least at the time of planning CT before RT.

Identification and counting of signal voids consistent with microbleeds was done by 2 examiners (M. Dennler, board‐certified radiologist and C. Staudinger, resident in radiology). To minimize user error from counting, microbleeds were labeled and counted multiple times, until the same number of counts was obtained from 2 consecutive trials.

The number of presumed microbleeds was counted within the entire brain volume as contoured for RT and within the abovementioned regions of specific isodose volumes, by comparing the location of the CMBs in the MR scans to the volume in the treatment planning system.

The primary outcome was the occurrence of CMBs over time, stratified by different dose volumes.

2.5. Collected variables, concurrent health condition, and neuropathologic examination

Demographic data, including age, sex, weight, breed, head conformation (based on breed; for mixed breed dogs or equivocal cases, facial and nasal bones as well as soft tissue structures such as the soft palate and tongue were subjectively assessed on the planning CT scan), presumed tumor type, tumor location, GTV, brain volume was recorded.

Medical records were searched for concurrent diseases, with a focus on renal, cardiovascular, and endocrinologic conditions.

If available, neuropathologic examination results were reviewed. In addition, the (board‐certified) pathologists who had performed the examination were contacted and asked to reevaluate the slides regarding CMBs. For the examination, brains were removed and immersion‐fixed in 10% formalin for at least 7 days. Brains were cut and selected brain regions including tumor region, brainstem, cerebellum, midbrain, thalamus, caudate nuclei, and cortex were embedded in paraffin, cut at 4 μm thickness and stained with hematoxylin and eosin. Tissue slides were evaluated for tumor diagnosis and secondary changes to irradiation.

2.6. Statistical analysis

The data were collected using Microsoft Excel for Mac (version 16.80).

Descriptive statistics were performed of the whole study cohort and split by 2 groups: “CMBs+” (if the dog ever developed ≥1 CMB[s] after RT) versus “CMBs−” (no CMB after RT). Median and range were calculated for continuous variables (age, weight, GTV, brain volume, low−/intermediate−/high‐dose volume), whereas the absolute and relative counts were reported for categorical variables (sex, breed, head conformation, presumed tumor type, tumor location).

Cerebral microbleeds were reported as an absolute count together with the corresponding 95% confidence intervals (CI).

The maximum follow‐up time per dog was defined as the time of the latest included MR study; median and range were indicated.

Descriptive data analysis and the creation of plots were performed using the following packages: “tidyr”, “dplyr”, “ggplot2”, and “ggbreak” in R version 4.0.5 and 4.3.0. ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ “survival” and “survminer” were used for the creation of a Kaplan‐Meier plot to illustrate the earliest appearance of a CMB in each dog, if applicable, and to calculate the median time to event. ⁵⁰ , ⁵¹

Confidence intervals for binomial proportions were calculated with a “DescTools” package, using Jeffreys intervals. ⁵² , ⁵³

For repeated measurements, a generalized linear mixed‐effects model (GLMM) was used to analyze the relationship between the CMBs count and the predictor variables (irradiation dose and time after treatment). The GLMM was fitted using the “glmer” function from the “lme4” package. ⁵⁴ The random intercept term (1|ID) was added to account for repeated measurements within an animal. The selection of the best model was based on the Akaike Information Criterion (AIC), a commonly used criterion for model selection. Lower AIC values indicate a better trade‐off between model fit and complexity. The GLMM with the lowest AIC value was chosen as the preferred model for analysis. Two models were considered (with and without time variable).

3. RESULTS

3.1. Dog and tumor characteristics

Thirty‐four dogs fulfilled the inclusion criteria. Table 1 summarizes the demographic and tumor‐specific data for all dogs, and stratified by whether they did or did not develop CMBs after RT. Among the purebred dogs, the following breeds were represented: French bulldog (n = 6), boxer (n = 3), Labrador (n = 3), Boston terrier (n = 2), Yorkshire terrier (n = 2); American bulldog, Australian shepherd, bullmastiff, chihuahua, flat coated retriever, golden retriever, Irish soft coated wheaten terrier, Jack Russel terrier, miniature schnauzer, Norwich terrier, spitz, white Swiss shepherd dog (n = 1 each).

TABLE 1.

Signalment and tumor characteristics of the dogs that developed, that did not develop CMBs, and of the whole study cohort, respectively.

	CMBs+ (n = 21)	CMBs− (n = 13)	Total (N = 34)
Age (months) a	110 (38‐152)	100 (41‐136)	106 (38‐152)
Weight (kg) a	13.7 (2.5‐37.4)	14.4 (6.3‐31.5)	13.7 (2.5‐37.4)
GTV (cm3) a	2.3 (.6‐7.80)	3.8 (1.4‐5.1)	2.9 (.6‐7.8)
Brain volume (cm3) a	91.9 (57.0‐114.0)	85.7 (62.5‐133.0)	87.6 (57.0‐133.0)
Low‐dose volume (cm3) a	54.1 (30.0‐91.0)	61.1 (28.0‐97.6)	57.6 (28.0‐97.6)
Intermediate‐dose volume (cm3) a	8.7 (2.1‐15.5)	10.8 (3.7‐16.2)	9.0 (2.1‐16.2)
High‐dose volume (cm3) a	14.2 (6.6‐40.9)	17.5 (7.8‐70.8)	15.3 (6.6‐70.8)

	n (%)	n (%)	n (%)
Sex b
Female, intact	4 (67%)	2 (33%)	6 (18%)
Female, spayed	8 (53%)	7 (47%)	15 (44%)
Male, intact	3 (50%)	3 (50%)	6 (18%)
Male, castrated	6 (86%)	1 (14%)	7 (21%)
Breed and head conformation b
Purebred	17 (61%)	11 (39%)	28 (82%)
Mixed breed	4 (67%)	2 (33%)	6 (18%)
Brachycephalic	9 (69%)	4 (31%)	13 (38%)
Nonbrachycephalic	12 (57%)	9 (43%)	21 (62%)
Tumor type and neuraxis b
Extra‐axial
Meningioma	9 (56%)	7 (44%)	16 (47%)
Pituitary tumor	3 (75%)	1 (25%)	4 (12%)
Intraventricular tumor	1 (50%)	1 (50%)	2 (6%)
Intra‐axial
Glioma	8 (67%)	4 (33%)	12 (35%)
Tumor location b
Forebrain	15 (65%)	8 (35%)	23 (68%)
Cerebellum	2 (100%)	0 (0%)	2 (6%)
Brainstem	1 (20%)	4 (80%)	5 (15%)
Pituitary	3 (75%)	1 (25%)	4 (12%)

Abbreviations: CMBs+/CMBs−, dogs that developed/did not develop cerebral microbleeds after radiation therapy, respectively; GTV, gross tumor volume.

^a Expressed as median (range).

^b Expressed as absolute number (percentage).

All included dogs were part of studies conducted at the same institution; if necessary for the initial study, dogs were treated under approval by the Animal Ethics Council of the Canton of Zurich, Switzerland (Permit Numbers: ZH075/17, ZH021/19). Data of a subset of these dogs (15/31) were already published in earlier studies but did not include information on CMBs. ³ , ⁸ , ⁵⁵

3.2. Concurrent health condition

Before treatment or within the follow‐up period, 2 dogs (6%) showed a mild increase in creatinine concentration; 1 of these dogs had a further work‐up, revealing a normal urine protein : creatinine ratio and a normal symmetric dimethylarginine (SDMA) value. In total, SDMA concentration was measured in 11 dogs (32%), all of which were within the reference range. Urinalysis was performed in 12 dogs (35%), revealing a urine specific gravity <1.030 in 9/12 (75%), with values between 1.015 and 1.029. In these dogs, dipstick semiquantitative protein concentration was 0 in 3/9 (33%), + in 5/9 (56%), ++ in 1/9 (11%). Of these 9 dogs, 6/9 (67%) were under treatment with prednisolone at variable dosages once daily, 1/9 (11%) suffered from pituitary‐dependent hyperadrenocorticism, 6/9 (67%) received phenobarbital as an antiepileptic drug, and in 1 dog (11%), cystocentesis was performed after IV infusion with Ringer's acetate solution.

Four dogs (13%) had a diagnosis of mitral valve disease, all of which were classified as ACVIM stage B1.

Two dogs (6%) were diagnosed with pituitary‐dependent hyperadrenocorticism.

Before treatment, 1 dog (3%) had an episode of hypertension (average systolic blood pressure of 190 mmHg, measured by the referring veterinarian), which never reoccurred later.

3.3. RT and follow‐up MR scans

Nineteen (56%) dogs were treated with 10 × 4 Gy, 7 (21%) dogs with 10 × 4 Gy including a “boost” of 11% physical dose increase limited to the GTV. Eight (24%) dogs were treated with a new protocol of 10 × 4 Gy with deliberate heterogeneous radiation dose delivery within the tumor.

In all but 1 of the dogs, the low‐dose volume was the largest, followed by the high‐dose volume; the intermediate‐dose volume was usually much smaller (Table 1). In 1 dog, the intermediate‐dose volume was slightly larger than the high‐dose volume.

The median follow‐up time was 12.6 months, range 1.8 to 37.6 months.

In total, 83 MR scans were performed, with a median of 2 scans per dog: Twenty‐two (65%) dogs had 2, 9 (26%) dogs had 3, and 3 (9%) dogs had 4 MR scans. Among these, 15/83 (18%) were baseline MR scans, and 68/83 (82%) scans were performed after completed RT (45/68 (66%) in the 1st, 20/68 (29%) in the 2nd, 2/68 (3%) in the 3rd, and 1/68 (1%) in the 4th year after treatment).

The scans after completed RT were performed in 56/68 dogs (82%) for monitoring purposes, without the clinical suspicion of progressive disease. In 4/68 dogs (6%), there was a high clinical suspicion of progressive disease, which was then confirmed by imaging, whereas in 8/68 dogs (12%) clinical signs were equivocal and imaging findings did not support tumor progression.

3.4. Cerebral microbleeds

In 4/15 dogs (27%, 95% CI [10%, 52%]) CMBs were present at baseline. Three of these 4 dogs developed further CMBs after treatment.

One or more CMBs at any time after completed RT were identified in 21/34 dogs (62%, 95% CI [45%, 77%]). Figure 2 illustrates when and how many CMBs were detected in each of the 21 dogs. Among these 21 dogs, the median time to the 1st detected CMB was 230 days (95% CI, 191‐440). Considering the whole cohort, the median time was 372 days (95% CI, 230‐619; Figure 3).

FIGURE 2.

Change in the number of cerebral microbleeds (CMBs) over time. Each line corresponds to a single dog that underwent a follow‐up scan and developed CMBs. Note that some dogs already had microbleeds at the initial time point (as indicated with a nonzero 1st CMB‐count). The continuous lines are for better visualization and do not reflect a linear increase in CMBs. The y‐axis is discontinuous to include the outlier, without distorting the remaining graph. mo, months.

FIGURE 3.

Temporal representation of when a cerebral microbleed (CMB) was 1st detected, with each vertical line (step) corresponding to 1 of the 21 dogs that developed CMBs after radiotherapy. The tick marks represent censored cases, that is, the 13 dogs that did not develop CMBs. The gray area indicates the 95% confidence interval, the dashed lines parallel to the y‐axis represent 1 and 2 year(s) after radiotherapy, respectively. The x‐axis indicates time in days.

In the entire sample, 160 CMBs were detected after RT (sum of highest CMB count per dog, developed after RT; baseline count of CMBs excluded). The cumulative number and increase in microbleeds in the cohort over time, split by the 3 different dose levels (low, intermediate, and high), is illustrated in Figure 4. New microbleeds were predominantly seen at a late time point (>6 months after RT) and in the high‐dose volume (Figure 5).

FIGURE 4.

Cumulative number of cerebral microbleeds (CMBs) detected in the study cohort after radiotherapy, stratified by dose levels and period. mo, months.

FIGURE 5.

Boxplots for the comparison of the number of new cerebral microbleeds (CMBs) (only dogs that developed CMBs after treatment), split by dose volume and time after radiotherapy. Early includes time points ≤6 months, late >6 months after completed therapy. At any given point in time, the high‐dose volume has 4.7 times higher odds (95% CI, 3.9‐5.6; P < .001) of developing CMBs compared with the low‐dose volume. CI, confidence interval.

Cerebral microbleeds were primarily identified in the forebrain, coinciding with the predominant tumor location. The cerebellum and brainstem were affected only in individual cases.

Figure 6 shows as an example the SWI sequences of a single dog at 4 different time points after treatment.

FIGURE 6.

Example of an increasing number of cerebral microbleeds (CMBs) after radiation therapy (RT) in a single dog (same dog as in Figure 1). As can be seen on the susceptibility‐weighted imaging (SWI) sequences 2, 8, 14, and 18 months after RT, the number of signal voids/CMBs increased over time. The dog also showed an increase in CMBs in more caudal parts of the brain. Some of the signal voids are intratumoral, these were not counted as CMBs.

The detected CMBs were nonregressive in nature, except for 6 signal voids in 4 dogs: In the 1st dog, 1 signal void in the low‐dose volume, last detected 2 months after RT, was not visible anymore at the 38 months follow‐up MR scan. In the 2nd dog, 2 signal voids in the high‐dose volume detectable 7 months after RT, could not be recognized at the 10 months follow‐up MR scan. In the 3rd dog, 2 signal voids in the high‐dose volume, noticed at the 11 months recheck, were not visible 13 months later. In the 4th dog, the only signal void that was found at all, in the high‐dose volume, disappeared between the 9 and 21 months follow‐up MR scans.

3.5. Mixed‐effects model

The mixed model analysis revealed that with each month, the number of CMBs increased by 14% (95% CI [11%‐16%]; P < .001).

At any given point in time, the high‐dose volume has 4.7 times higher odds (95% CI, 3.9‐5.6; P < .001) of developing CMBs compared with the low‐dose volume; the intermediate‐dose volume has 10 times lower odds (95% CI, 0.06‐0.17, P < .001) of developing CMBs compared with the low‐dose volume.

No statistically significant association between demographic variables and the development of CMBs was found.

3.6. Neuropathological examination and correlation with MRI findings

In 3 dogs, the presumed location in the neuraxis and the tumor diagnosis were confirmed by neuropathology (2 gliomas, 1 pituitary tumor).

The 1st dog (French bulldog, 6 years old, female spayed, oligodendroglioma in the left frontal region) was euthanized 10 months after RT directly after the MRI scan, which had been performed because of a marked clinical deterioration caused by tumor progression and metastases. In the neuropathological examination, hemorrhage and hemosiderin‐containing macrophages within the tumor were detected. In 1 extratumoral localization, at the level of the basal nuclei, a focal, small bleeding consisting of extravasated erythrocytes was described.

Based on the SWI sequence, 8 extratumoral microbleeds were counted in the high‐dose volume of this dog, which also involved the left basal nuclei.

The 2nd dog (miniature schnauzer, 8 years old, male, suspicion of astrocytoma) was euthanized 2 months after RT, 2 days after the MRI scan, which had been performed because of marked clinical deterioration because of tumor progression. The neuropathologic examination revealed intratumoral hemorrhage with deposition of hemosiderin and fibrinoid necrosis of the blood vessels within and at the periphery of the mass. No extratumoral CMBs were seen in the MR scan.

The 3rd dog (Yorkshire terrier, 11 years old, male, pituitary carcinoma) was euthanized 31 months after RT, after the MRI scan, which had been performed because of marked clinical deterioration. Neuropathologic examination revealed a multifocal oligodendroglioma involving the ventricular system in addition to the previously treated pituitary tumor. In addition, diffuse leukoencephalopathy with marked myelin loss and numerous hemorrhages associated with vascular changes were noted. Smaller hemorrhages were present in the cortex; larger hemorrhages (up to .3 cm in diameter) were noted in the left caudate nucleus, in the left parietal cortex, and in the left hippocampus and were accompanied by the presence of hemosiderin‐laden macrophages. In total, 44 CMBs were counted on the SWI‐sequence, with 33 in the high‐dose, 2 in the intermediate‐dose, and 9 in the low‐dose volume.

4. DISCUSSION

The aims of our study were to elucidate whether dogs irradiated for intracranial tumors develop CMBs, if their manifestation and distribution vary depending on dose levels, and if these CMBs increase in number over time. Susceptibility‐weighted imaging was used as a bleeding‐sensitive sequence to detect CMBs. This investigation confirms that CMBs are a possible implication after irradiation of intracranial neoplasia in dogs. Most dogs (21/34, 62%) developed CMBs after treatment. The CMBs appeared predominantly in the high‐dose volume, most typically within the 1st year after RT (Figure 3) and increased in number over time. This is consistent with what is reported in irradiated human brain tumor patients: In humans, CMBs appear 1st in the high‐dose brain areas and increase in number over time, especially with higher radiation doses delivered to larger brain volumes. ²⁰ , ²¹ , ²² , ²³ Depending on imaging sensitivity (eg, strength of the magnet), 11% to 100% of microbleeds were found already at 4 months to 1 year, whereas no CMBs were detected in nonirradiated control patients. ⁵⁶ , ⁵⁷

Regardless of the cause, bleeding involves extravasation of blood components. In the surrounding tissue, hemoglobin is degraded subsequently. In humans, hemosiderin‐laden macrophages have been described to appear approximately 3 days after erythrocyte extravasation. ⁵⁸ In the case of RT, extravasation of blood components occurs after damage to endothelium, 1 form of late adverse effects (several months to years after treatment, given the slow rate of turnover of vascular endothelial cells). The relative location between the tumor and surrounding vessels is determinant in the topography of radiation‐induced vasculopathy. ⁵⁹

Radiation‐caused injury in human brain tumor patients has been associated with highly relevant neuropsychological sequelae such as cognitive decline ¹⁰ , ¹¹ , ¹² and is accompanied by a deterioration in quality of life. ¹³ Cognitive dysfunction is a well‐recognized disease entity in older dogs that shares features with human cognitive dysfunction syndromes. ⁶⁰ Inherently, the assessment of mild or gradually progressing cognitive deficits in dogs is difficult. Given the retrospective nature of the present study and the primary goal to identify CMBs after RT, cognitive status of the included dogs was not assessed. In follow‐up MR examinations, CMBs are either seen in dogs with controlled disease, in dogs with progressive disease, or no CMBs are seen at all. Our impression is that CMBs in animals with controlled disease are usually an incidental finding (“clinically silent”). No animal was presented with clinical deterioration where only new CMBs were found; deterioration is usually caused by tumor progression.

Radiation therapy is only 1 of many possible causes of CMBs, they are not pathognomonic. In human medicine, at least 30 different conditions have been described that converge on the same manifestation on susceptibility‐weighted sequences, ⁶¹ besides the fact that CMBs have also been found in healthy people, with the number increasing in the aging population. ²⁴ , ⁵⁶ , ⁶² , ⁶³ In the veterinary literature, CMBs have been reported in the context of chronic kidney failure, cardiovascular diseases, endocrinopathies, intracranial masses, proteinuria, traumatic brain injury, ³¹ , ³² , ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ , ⁶⁸ but not in the context of RT. It is reasonable to assume that dogs can also develop CMBs attributable to various disease entities or as part of the aging process. In a veterinary study that examined 582 dogs that had undergone brain MRI, microbleeds were presumed in 54 dogs (9.3%). ³¹ Among these 54 dogs, 14 dogs had an intracranial mass and ≥ 1 microbleed, whereas 22 dogs had an intracranial mass and no microbleed; no association between putative microbleeds and space‐occupying intracranial disease was found. Dogs with presumed microbleeds had a shorter median survival time than age‐ and breed‐matched controls. ³¹ Direct comparison to our study might be limited as T2* and not SWI was used for the detection of CMBs, in addition to the varying dog samples. There are both human and veterinary studies that show an improved detection of CMBs using SWI: on SWI, more CMBs/areas of signal voids are detected, and they appear more conspicuous. ⁴³ , ⁶⁴ , ⁶⁷ , ⁶⁹ , ⁷⁰ Another advantage of this sequence are the included phase images which allow to differentiate paramagnetic from diamagnetic property of a signal void.

In this cohort, the high‐dose volume was found to have 4.7 times higher odds of developing CMBs compared with the low‐dose volume. At 1st glance, the finding that the intermediate‐dose volume had 10 times lower odds of developing CMBs appears counterintuitive. This is most likely attributable to the size and proportions of the 3 dose volumes: in all but 1 of the dogs, the intermediate‐dose volume represented the smallest, whereas the low‐dose volume was the largest part of the brain. Thus, there is a higher chance of detecting CMBs of any cause in the low‐dose versus the intermediate‐dose volume.

In the present study, CMBs were detected before treatment in 4/15 dogs (27%). Considering the usually advanced age of dogs with intracranial neoplasia, 1 explanation for baseline CMBs is vasculopathy as part of the aging process.

In 4 dogs (12% of all dogs, 19% of dogs with CMBs after RT), signal voids previously labeled as CMBs could not be identified in a follow‐up scan. This is in line with studies in human medicine, where an increase of CMBs over time is seen in the majority of dogs after RT. ⁵⁶ One study described the decrease in size of individual CMBs after RT, with the total number and volume increasing. The decrease in size could be the result of blood product breakdown over time. ⁵⁶

In other studies investigating causes other than RT for CMBs, there have been few reports of individual CMBs that could not be detected on the follow‐up scan. ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ Specific explanations have not been provided yet. The signal void‐causing iron might be redistributed or reused, conceivably mediated by mononuclear phagocytes.

A strength of our study is the number of dogs that were irradiated and followed up with advanced imaging. Follow‐up imaging with reproducible imaging protocols at similar intervals after RT enables the investigation of treatment‐associated changes with longitudinal character.

The limitations of the present study are mostly because of its retrospective nature. Neuropathologic examination was available only in 3 dogs and had not been performed to detect or describe CMBs in particular, but to determine tumor histology and to identify potential radiation injury in the peritumoral tissue. However, we consider the examined specimens as valuable given that neuropathological examination confirmed in all 3 dogs, that microbleeds and vascular changes can occur in irradiated dogs.

Moreover, only 15 dogs (44%) had a baseline MR scan including SWI. Even though dogs are regularly referred to our clinic by large centers equipped with (low‐ or high‐field) MRI, none of these referred dogs had a baseline or recheck MR scan with an SWI sequence. Ideally, a scan with the same protocol as intended to monitor changes over time is obtained in all dogs before treatment, as a baseline for later comparison.

Another aspect is that SWI sequences could not be imported into the treatment planning system caused by the double number of slices (magnitude and phase images joined in series) which precluded coregistration of the CT scan and the SWI sequence. Thus, the volumes could not be displayed directly on the MR scans. Misclassification of borderline CMBs is possible, but a relevant distortion is considered unlikely.

Furthermore, no control dogs were included and examined. In theory, an adequate control group (dogs with brain tumors, but not treated with RT) is desirable, but poses an ethical dilemma, or is not even possible because the dogs do not survive long enough.

It is possible that the included dogs developed CMBs because of aging or 1 of the reasons already reported in the veterinary literature, such as renal disease. Nine dogs had a urine specific gravity <1.030, even though none of these showed an increase in creatinine concentration except for a mild increase in 2 dogs. In most of these dogs, there were additional reasons why the urine specific gravity could be low, such as prednisolone treatment, hyperadrenocorticism, infusion treatment. A standardized work‐up for renal disease and proteinuria was not available given the study design. Thus, not all counted signal voids in the treated dogs might be attributable to RT. However, a substantial distortion is implausible given the convincingly divergent amount of microbleeds depending on the dose volume, with a marked predominance of CMBs in the high‐dose volume. This pattern is not compatible with more randomly distributed bleedings of aging dogs or dogs with respective comorbidities.

In conclusion, RT can lead to CMBs in dogs. Radiation therapy‐induced CMBs are most likely to appear in the peritumoral high‐dose volume and to be permanent.

CONFLICT OF INTEREST DECLARATION

Authors declare no conflict of interest.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Staudinger C, Dennler M, Körner M, et al. Relationship between radiation dose and cerebral microbleed formation in dogs with intracranial tumors. J Vet Intern Med. 2024;38(6):3182‐3192. doi: 10.1111/jvim.17213